"Arts, Faculty of"@en . "Linguistics, Department of"@en . "DSpace"@en . "UBCV"@en . "Lyon, John"@en . "2014-01-02T16:33:48Z"@en . "2013"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "This dissertation investigates the syntax and semantics of equative structures (i.e. DP-DP structures and clefts) in the little studied and highly endangered Upper Nicola dialect of Okanagan Salish (a.k.a. Nsy\u00C3\u00ADlxc\u00C9\u0099n), and represents the first detailed investigation of equatives in a Salish language. From the theoretical perspective, Okanagan is noteworthy since there is no evidence for a predicational copula (contra Baker 2003, Adger & Ramchand 2003) while there is evidence for a null equative copula (Heycock & Kroch 1999), thereby supporting theories which argue for a structural distinction between predication and equation.\n\nOkanagan does not have an overt copula (A. Mattina 2001), yet does have sentences consisting only of two determiner phrases (DPs) (''DP-DP structures''). These exhibit a word order restriction which is absent from predications involving other syntactic categories, such that in answer to a WH-question, a directly referential demonstrative or proper name must precede a DP headed by the determiner i\u00CA\u0094 (an ''i\u00CA\u0094 DP''). The implication is that specificational sentences (Higgins 1973) are not possible in Okanagan. Given that i\u00CA\u0094 DPs permit intensional readings, and that i\u00CA\u0094 DPs never denote sets (Longobardi 1994, Matthewson 1998), I claim that the Okanagan equative head maps the intension of an individual to its extension, and is of type <,> (Romero 2005, Comorovski 2007). Since there are no specificational sentences in Okanagan, and the equivalent of Higgins' identificational sentence class (e.g. 'That is John' in English) pattern with copula-less, direct predications in Okanagan, the data support reducing Higgins' taxonomy to only two types for Okanagan: predicational and equative (Heller 2005).\n\nI claim that Okanagan clefts are also equative structures, based on evidence that clefts consist of two DPs and carry an implicature of exhaustivity (Davis et. al. 2004). This implicature stems from the maximality implicature carried by the determiner i\u00CA\u0094 which introduces the second DP (i.e. the residue). My analysis runs parallel to theories of English clefts which align cleft semantics to the semantics of determiners (Percus 1997, Hedberg 2000)."@en . "https://circle.library.ubc.ca/rest/handle/2429/45684?expand=metadata"@en . "Predication and Equation in Okanagan Salish:The Syntax and Semantics of Determiner PhrasesbyJohn LyonB.A. German, Auburn University, 2000B.Sc. International Business, Auburn University, 2000M.A. Linguistics, University of Montana, 2005A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Linguistics)The University Of British Columbia(Vancouver)December 2013? John Lyon, 2013AbstractThis dissertation investigates the syntax and semantics of equative structures (i.e.DP-DP structures and clefts) in the little studied and highly endangered UpperNicola dialect of Okanagan Salish (a.k.a. Nsy?lxc@n), and represents the first de-tailed investigation of equatives in a Salish language. From the theoretical per-spective, Okanagan is noteworthy since there is no evidence for a predicationalcopula (contra Baker (2003), Adger and Ramchand (2003)) while there is evidencefor a null equative copula (Heycock and Kroch, 1999), thereby supporting theorieswhich argue for a structural distinction between predication and equation.Okanagan does not have an overt copula (A. Mattina 2001), yet does havesentences consisting only of two determiner phrases (DPs) (?DP-DP structures?).These exhibit a word order restriction which is absent from predications involvingother syntactic categories, such that in answer to a WH-question, a directly refer-ential demonstrative or proper name must precede a DP headed by the determineriP (an ?iP DP?). The implication is that specificational sentences (Higgins, 1973)are not possible in Okanagan. Given that iP DPs permit intensional readings, andthat iP DPs never denote sets (Longobardi, 1994; Matthewson, 1998), I claim thatthe Okanagan equative head maps the intension of an individual to its extension,and is of type <,> (Romero, 2005; Comorovski, 2007). Since there areno specificational sentences in Okanagan, and the equivalent of Higgins? identi-ficational sentence class (e.g. That is John in English) pattern with copula-less,direct predications in Okanagan, the data support reducing Higgins? taxonomy toonly two types for Okanagan: predicational and equative (Heller, 2005).I claim that Okanagan clefts are also equative structures, based on evidencethat clefts consist of two DPs and carry an implicature of exhaustivity (Davis et al.,ii2004). This implicature stems from the maximality implicature carried by the de-terminer iP which introduces the second DP (i.e. the residue). My analysis runsparallel to theories of English clefts which align cleft semantics to the semantics ofdeterminers (Percus, 1997; Hedberg, 2000).iiiPrefaceThis dissertation consists of original and independent work by the author, JohnLyon, and is based on fieldwork with fluent speakers of the Upper Nicola dialectof Okanagan Salish. This fieldwork is covered by UBC Ethics Certificate num-ber H08-01182 under the title ?The Representation of Focus in Languages of thePacific Northwest?, a grant from the Social Sciences and Humanities ResearchCouncil of Canada.A version of Chapter 5 of this dissertation, Semantics of Okanagan DeterminerPhrases, has been accepted for publication by the International Journal of Amer-ican Linguistics, under the title Okanagan Determiner Phrases and Domain Re-striction. It appeared earlier as a working paper under the title The semantics ofdeterminer phrases in Okanagan in the Precedings for the 46th Annual Interna-tional Conference on Salish and Neighbouring Languages (ICSNL), University ofBritish Columbia Working Papers in Linguistics volume 30, pages 194?266.Significant portions of chapter 6 of this dissertation, The Syntax of Nomi-nal Modification, appeared as a working paper as Nominal modification in UpperNicola Okanagan: A working paper in the Precedings for the 45th Annual Interna-tional Conference on Salish and Neighbouring Languages (ICSNL), University ofBritish Columbia Working Papers in Linguistics volume 27, pages 199?233.The map given as Figure 1.2, entitled Geographic Distribution of Salish Lan-guages, is reproduced from The Salish Language Family: Reconstructing Syntaxby Paul D. Kroeber (p. xxxi) by kind permission of the University of NebraskaPress, copyright 1999 by the University of Nebraska Press.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivAbbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Purpose of this Study . . . . . . . . . . . . . . . . . . . . . . . . 11.2 The Okanagan Language . . . . . . . . . . . . . . . . . . . . . . 71.3 Ethics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101.4 The Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.4.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . 111.4.2 Interlinear Glossing and Orthographic Conventions . . . . 121.5 Salish Literature Review . . . . . . . . . . . . . . . . . . . . . . 151.5.1 Okanagan . . . . . . . . . . . . . . . . . . . . . . . . . . 151.5.2 Southern Interior Salish . . . . . . . . . . . . . . . . . . 161.5.3 Northern Interior Salish (and other Salish languages) . . . 161.6 Outline of the Dissertation . . . . . . . . . . . . . . . . . . . . . 17v2 Predication and Equation: Theoretical Background . . . . . . . . . 202.1 Predication and Equation . . . . . . . . . . . . . . . . . . . . . . 212.1.1 Defining Predication: Semantic Issues . . . . . . . . . . . 212.1.2 Copular Clauses and Higgins? (1973) Taxonomy . . . . . 252.2 Specificational Copular Syntax/Semantics:Predicate Raising or an Equative Head? . . . . . . . . . . . . . . 342.2.1 Small Clauses . . . . . . . . . . . . . . . . . . . . . . . . 352.2.2 Predicate Raising . . . . . . . . . . . . . . . . . . . . . . 372.2.3 Specificationals as Equatives . . . . . . . . . . . . . . . . 392.2.4 Directly Referential versus Non-Rigid DPs:An Intensional Asymmetry . . . . . . . . . . . . . . . . . 432.2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 452.3 Focus and Information Structure . . . . . . . . . . . . . . . . . . 452.3.1 Focus Alternatives and F-marking . . . . . . . . . . . . . 462.3.2 Prosodic Alignment and Focused Constituents . . . . . . 482.4 Clefts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502.4.1 ?Extraposition from Subject? Analyses of Clefts . . . . . . 512.4.2 Cleft Semantics: Exhaustivity Entailments and ExistencePresuppositions . . . . . . . . . . . . . . . . . . . . . . . 522.4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 542.5 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 553 Background in Okanagan Grammar . . . . . . . . . . . . . . . . . . 563.1 Phonology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563.2 Inflection: Pronominal, Valency, and Tense-Aspect . . . . . . . . 583.3 Clausal Word Order . . . . . . . . . . . . . . . . . . . . . . . . . 644 The Structure and Distribution of NP and DP . . . . . . . . . . . . . 734.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734.2 Distinguishing Nouns as a Syntactic Category . . . . . . . . . . . 744.3 NP versus DP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794.4 The Distribution of the Determiner and Oblique Marker . . . . . . 824.4.1 Subject Arguments . . . . . . . . . . . . . . . . . . . . . 83vi4.4.2 Core Objects vs. Quasi-Objects . . . . . . . . . . . . . . 844.4.3 The Syntactic Status of iP and t . . . . . . . . . . . . . . 884.4.4 Arguments of Lexical Intransitives . . . . . . . . . . . . . 904.4.5 Ditransitives . . . . . . . . . . . . . . . . . . . . . . . . 924.4.6 Possessor Intransitives . . . . . . . . . . . . . . . . . . . 944.4.7 The Oblique Marker in Other Environments . . . . . . . . 954.4.8 This ?Predictability? does not hold across Salish . . . . . . 964.4.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 974.5 Internal Structure of DP(and other Nominal Projections) . . . . . . . . . . . . . . . . . . 984.5.1 iP Occurs in D Position . . . . . . . . . . . . . . . . . . . 994.5.2 Three Hypotheses . . . . . . . . . . . . . . . . . . . . . . 994.5.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 1044.6 Other DPs: Demonstratives and Proper Names . . . . . . . . . . . 1064.6.1 Demonstrative DPs . . . . . . . . . . . . . . . . . . . . . 1064.6.2 Proper Name DPs . . . . . . . . . . . . . . . . . . . . . . 1154.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1215 Semantics of Okanagan Determiner Phrases . . . . . . . . . . . . . 1225.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1235.2 The Semantics of the Determiner iP and Oblique Marker t . . . . 1255.2.1 iP and t are not Deictic . . . . . . . . . . . . . . . . . . . 1265.2.2 iP and t are Not Definite . . . . . . . . . . . . . . . . . . 1285.2.3 iP and t are not English-like Existential Indefinites . . . . 1375.2.4 iP is Different than a Lillooet Widest-scope Indefinite . . . 1405.2.5 iP Carries an Implicature of Uniqueness and Maximality (tdoes not) . . . . . . . . . . . . . . . . . . . . . . . . . . 1425.2.6 iP does not Create a Generalized Quantifier . . . . . . . . 1475.2.7 Existential Sentences and iP DPs . . . . . . . . . . . . . . 1495.2.8 Generic Readings of iP DPs . . . . . . . . . . . . . . . . 1515.2.9 Intensionality and iP DPs . . . . . . . . . . . . . . . . . . 1535.2.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 1555.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156vii5.3.1 Okanagan iP . . . . . . . . . . . . . . . . . . . . . . . . 1565.3.2 Okanagan t . . . . . . . . . . . . . . . . . . . . . . . . . 1615.3.3 Explaining the Implicature Carried by iP . . . . . . . . . 1645.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1666 The Syntax of Nominal Modification . . . . . . . . . . . . . . . . . . 1686.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1696.1.1 Morphological Patterns of Nominal Modification . . . . . 1696.1.2 Distinguishing Clausal Subordination from Nominal Mod-ification . . . . . . . . . . . . . . . . . . . . . . . . . . . 1716.2 Attributive Modification . . . . . . . . . . . . . . . . . . . . . . 1756.2.1 Review: Complex Nominal Predicates and Complex DPs . 1766.2.2 Aspectual Restrictions on Attributive Modification . . . . 1796.2.3 Syntax of Attributive Modification . . . . . . . . . . . . . 1856.2.4 Summary of Attributive Modification . . . . . . . . . . . 1876.3 Relative Clauses . . . . . . . . . . . . . . . . . . . . . . . . . . . 1876.3.1 Introducing Relative Clauses . . . . . . . . . . . . . . . . 1886.3.2 Other Characteristics of Okanagan Relatives . . . . . . . 1966.3.3 Relative Clause Formation by Movement . . . . . . . . . 2006.3.4 Problems with Extending the Movement Account . . . . . 2106.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2116.5 Chapter Addendum: Notes on the ?Matching Effect? . . . . . . . . 2127 Direct Predications and DP-DP Structures: Syntax, Semantics, andInformation Structure . . . . . . . . . . . . . . . . . . . . . . . . . . 2157.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2157.1.1 Main Claims of this Chapter . . . . . . . . . . . . . . . . 2167.1.2 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . 2177.2 Direct Predication . . . . . . . . . . . . . . . . . . . . . . . . . . 2197.2.1 Direct Predications and Word Order . . . . . . . . . . . . 2207.2.2 Direct Predications and the Ban on Predicate Raising . . . 2267.2.3 (Near) Obligatory Subject-Raising over T . . . . . . . . . 2297.2.4 A Structural Analysis of Direct Predication . . . . . . . . 232viii7.2.5 Summary of Direct Predications . . . . . . . . . . . . . . 2367.3 DP-DP Structures . . . . . . . . . . . . . . . . . . . . . . . . . . 2367.3.1 DP-DP Structures and the Word Order Restriction . . . . 2377.3.2 No ?Predicate? Raising in Okanagan DP-DP Structures . . 2407.3.3 Connectivity (and Other) Effects and Okanagan DP-DPStructures . . . . . . . . . . . . . . . . . . . . . . . . . . 2427.3.4 An Equative Head . . . . . . . . . . . . . . . . . . . . . 2477.4 Information Structure and DP-DP structures . . . . . . . . . . . . 2497.4.1 The Exhaustivity Implicature in DP-DP structures . . . . 2507.4.2 DP-DP Predications are Non-presuppositional . . . . . . . 2567.4.3 An F-marked Constituent Must Occur Initially in a DP-DPstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . 2577.4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 2617.5 Analysis of the Equative Head . . . . . . . . . . . . . . . . . . . 2627.5.1 A Semantic Asymmetry . . . . . . . . . . . . . . . . . . 2627.5.2 Focus Alignment . . . . . . . . . . . . . . . . . . . . . . 2667.5.3 Final Analysis . . . . . . . . . . . . . . . . . . . . . . . 2687.6 Other Predication Types Involving Demonstratives and Proper Names2707.6.1 The Case for Identificational Sentences . . . . . . . . . . 2717.6.2 Predications Involving Two Proper Names . . . . . . . . . 2767.6.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 2807.7 Problem Inversions . . . . . . . . . . . . . . . . . . . . . . . . . 2807.8 Summary and Implications . . . . . . . . . . . . . . . . . . . . . 2867.8.1 Summary of Major Points . . . . . . . . . . . . . . . . . 2867.8.2 Implications . . . . . . . . . . . . . . . . . . . . . . . . 2878 Okanagan Clefts as Equatives . . . . . . . . . . . . . . . . . . . . . 2918.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2928.1.1 Terminological Preliminaries . . . . . . . . . . . . . . . . 2928.1.2 The Main Claim . . . . . . . . . . . . . . . . . . . . . . 2938.1.3 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . 2938.2 Introducing Introduced Clefts in Okanagan and Across Salish . . . 2978.2.1 The Classical Analysis: A Tripartite Structure . . . . . . . 297ix8.2.2 Introduced Clefts in Okanagan . . . . . . . . . . . . . . . 2988.2.3 DP Types and DP Focus . . . . . . . . . . . . . . . . . . 2998.3 Information Structure and Clefts . . . . . . . . . . . . . . . . . . 3018.3.1 Structural Focus . . . . . . . . . . . . . . . . . . . . . . 3018.3.2 The Exhaustivity Implicature in Clefts and Contrastive Focus3028.3.3 Clefts are Non-presuppositional . . . . . . . . . . . . . . 3068.3.4 Focus Cannot Fall Finally . . . . . . . . . . . . . . . . . 3088.3.5 Contrasting Clefts with Nominal Predicate Constructions . 3098.3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 3158.4 An Equative Analysis of Clefts . . . . . . . . . . . . . . . . . . . 3158.4.1 A Structural and Interpretive Ambiguity . . . . . . . . . . 3168.4.2 Clefts as Equatives: The Basic Idea . . . . . . . . . . . . 3178.5 Morpho-syntactic Evidence for the Equative Analysis of Clefts andDP-DP Structures . . . . . . . . . . . . . . . . . . . . . . . . . . 3208.5.1 Clefts and the (Optional) Initial Demonstrative . . . . . . 3208.5.2 Demonstrative Proclisis in Equatives . . . . . . . . . . . . 3238.5.3 Null Foci . . . . . . . . . . . . . . . . . . . . . . . . . . 3318.5.4 Intensionality and Cleft Residues . . . . . . . . . . . . . 3348.6 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3368.6.1 Syntactic Derivation . . . . . . . . . . . . . . . . . . . . 3368.6.2 Semantic Derivation . . . . . . . . . . . . . . . . . . . . 3398.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3428.8 Chapter Addendum A: Future Clefts and the Case for Clausal NPs 3438.9 Chapter Addendum B: Adjunct Clefts . . . . . . . . . . . . . . . 3469 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3499.1 Summary of Findings . . . . . . . . . . . . . . . . . . . . . . . . 3499.2 Implications for Okanagan Grammar . . . . . . . . . . . . . . . . 3539.2.1 Empirical Contributions . . . . . . . . . . . . . . . . . . 3539.2.2 Further Questions . . . . . . . . . . . . . . . . . . . . . . 3559.3 Implications for Salish . . . . . . . . . . . . . . . . . . . . . . . 3589.3.1 Implications for Southern Interior Salish . . . . . . . . . . 3619.3.2 Implications for Northern Interior Salish . . . . . . . . . . 376x9.3.3 Implications for Other Salish Languages . . . . . . . . . . 3969.3.4 Summary of Implications for Salish . . . . . . . . . . . . 4049.4 Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . 4099.4.1 A Distinction between Predication and Equation . . . . . 4109.4.2 Higgins? Taxonomic Classification for Okanagan . . . . . 4119.4.3 Pragmatic Differences between Okanagan and English Equa-tives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4159.4.4 Information Structural Differences between Okanagan andEnglish Equatives . . . . . . . . . . . . . . . . . . . . . . 4159.4.5 Exhaustivity in Okanagan Equatives . . . . . . . . . . . . 4169.4.6 Interpretive Variability in Okanagan versus English Equatives4169.4.7 Fixed Information Structure and the Connection to Clefts . 4179.4.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . 419Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420xiList of TablesTable 2.1 Referentiality and Higgins? Taxonomy (adapted from Mikkelsen(2011, 1810)) . . . . . . . . . . . . . . . . . . . . . . . . . . 27Table 3.1 Okanagan Consonant Phonemes . . . . . . . . . . . . . . . . . 57Table 3.2 Okanagan Vowel Phonemes . . . . . . . . . . . . . . . . . . . 57Table 3.3 Intransitive Paradigms . . . . . . . . . . . . . . . . . . . . . . 58Table 3.4 Transitive Paradigms . . . . . . . . . . . . . . . . . . . . . . . 59Table 3.5 Independent Pronouns . . . . . . . . . . . . . . . . . . . . . . 59Table 3.6 Summary of Word Orders in Upper Nicola Okanagan . . . . . 64Table 4.1 Distribution of iP and t across Grammatical Categories . . . . 98Table 4.2 Demonstratives in Okanagan, A. Mattina (1973) . . . . . . . . 106Table 5.1 Semantic Properties of iP and t . . . . . . . . . . . . . . . . . 125Table 5.2 Set Intersection and Maximality with Okanagan iP . . . . . . . 161Table 6.1 NP-introducing Articles versus Clausal Subordinators . . . . . 175Table 6.2 Distribution of Predicate Types in Attributive Modifications . . 185Table 6.3 Relativization Strategies in Four Interior Salish Languages . . . 208Table 6.4 Surface Patterns Displayed by Head/Modifier Introductory Par-ticles in Okanagan Sentences Involving Nominal Modification . 213Table 7.1 Directly versus Non-directly Referential DPs in Equative DP-DP Structures . . . . . . . . . . . . . . . . . . . . . . . . . . 263xiiTable 7.2 Reducing Higgins? Taxonomy: Identificational Sentences in Okana-gan Compared to Mikkelsen?s (2005) Classification of English 276Table 7.3 Higgins? Taxonomy and Okanagan Non-Verbal Predications . . 288Table 8.1 Clefting Predicates and Demonstratives across Four Interior Sal-ish Languages . . . . . . . . . . . . . . . . . . . . . . . . . . 299Table 9.1 ?Canonical-Order? Predication in Okanagan and English . . . . 350Table 9.2 ?Inverse-Order? Predication in Okanagan and English . . . . . 350Table 9.3 DP-DP Structures as Evidence for Equation across Salish: Ar-ranged by Constituency Type from Strongest to Weakest Evidence360Table 9.4 Evidence for Equative versus Predicational Analyses of South-ern Interior Salish DP-DP Structures and Clefts . . . . . . . . 376Table 9.5 Evidence for Equative versus Predicational Analyses of North-ern Interior Salish DP-DP Structures and Clefts . . . . . . . . 396Table 9.6 Evidence for Equative versus Predicational Analyses of DP-DPStructures and Clefts across Select Salish Languages . . . . . . 405Table 9.7 Higgins? Taxomony and Okanagan Non-Verbal Predications . . 412Table 9.8 Distribution of Semantic Types across DPs in Okanagan Non-Verbal Predications . . . . . . . . . . . . . . . . . . . . . . . 413xiiiList of FiguresFigure 1.1 Function Application in an Okanagan Direct Predication . . . 4Figure 1.2 Geographic Distribution of Salish Languages . . . . . . . . . 8Figure 1.3 Okanagan Dialect Areas (from Doak (1983, 17)) . . . . . . . 9Figure 2.1 The Copula ?be? and a Null Pred-head . . . . . . . . . . . . . 23Figure 2.2 DP Subject Raising and the Asymmetrical (Pred) Account . . 38Figure 2.3 DP Predicate Raising and the Asymmetrical (Pred) Account . 38Figure 2.4 DP Subject Raising and the Symmetrical (bare) Account . . . 39Figure 2.5 DP Predicate Raising and the Symmetrical (bare) Account . . 39Figure 4.1 Okanagan iP DP . . . . . . . . . . . . . . . . . . . . . . . . 100Figure 4.2 The KP hypothesis . . . . . . . . . . . . . . . . . . . . . . . 100Figure 4.3 The PP Hypothesis and Prosodic Inversion: ?towards the chief? 101Figure 4.4 The Headless Relative Hypothesis . . . . . . . . . . . . . . . 102Figure 4.5 Core argument . . . . . . . . . . . . . . . . . . . . . . . . . 104Figure 4.6 Locative adjunct with D . . . . . . . . . . . . . . . . . . . . 104Figure 4.7 Locative adjunct w/o D . . . . . . . . . . . . . . . . . . . . . 104Figure 4.8 Passive agent or instrumental adjunct . . . . . . . . . . . . . 105Figure 4.9 Quasi-object . . . . . . . . . . . . . . . . . . . . . . . . . . 105Figure 5.1 Semantic Composition of an Okanagan iP DP, Example I . . . 157Figure 5.2 Domain Restriction in Definite Contexts (e.g. 36a) . . . . . . 158Figure 5.3 Domain Restriction in Non-Unique Contexts (e.g. 10) . . . . 159Figure 5.4 Domain Restriction in Indefinite and Existential Contexts (e.g.15 and 28) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160xivFigure 5.5 Generic Interpretations of Okanagan DPs (e.g. 47) . . . . . . 161Figure 5.6 Semantic Incorporation of an Oblique Quasi-Object . . . . . . 163Figure 6.1 Complex Nominal Predicate . . . . . . . . . . . . . . . . . . 186Figure 6.2 Complex Nominal Predicate: Modifier Stacking . . . . . . . . 186Figure 6.3 Okanagan Locative Relative Clause . . . . . . . . . . . . . . 205Figure 6.4 Head-initial (a.k.a. ?post-nominal?) Relative Clause . . . . . . 206Figure 6.5 Head-final (a.k.a. ?pre-posed?) Relative Clause . . . . . . . . 207Figure 6.6 Headless Relative Clause . . . . . . . . . . . . . . . . . . . . 209Figure 7.1 Freely Ordered Bare Small Clause Hypothesis . . . . . . . . . 232Figure 7.2 Null Pred-head Small Clause Hypothesis . . . . . . . . . . . 233Figure 7.3 No Predicate Raising . . . . . . . . . . . . . . . . . . . . . . 234Figure 7.4 Subject Raising . . . . . . . . . . . . . . . . . . . . . . . . . 234Figure 7.5 Bare Small Clauses As Simple Saturation Relations . . . . . . 235Figure 7.6 An Oversimplified Equative Analysis . . . . . . . . . . . . . 248Figure 7.7 Semantic Asymmetry, more-referential DP is in initial position 265Figure 7.8 Semantic Asymmetry, less-referential DP is in initial position 266Figure 7.9 Final Equative Analysis of Okanagan DP-DP Structures . . . 269xvAbbreviationsABS absolutive DIR directive transitivizerABST absent DIST distalACC accusative DITR ditransitive applicativeAPPL (possessional) applicative DUB dubitativeATTR attributive EMPH emphaticAUT autonomous EPIS epistemic modalAUX auxiliary ERG ergative caseBEN benefactive applicative EVID evidentialBOUL bouletic modal EXIS assertion-of-existenceCAUS causative transitivizer EXTR extractionCISL cislocative FEM feminineCJCT conjunctive FRED final reduplicationCOMP complementizer FOC focusCONJ conjunction FUT futureCOP copula GEN genitiveCUST customary/habitual HAVE ?to have?DEIC deictic IM immediateDEF definite IMPF imperfectiveDEM demonstrative INCEPT inceptiveDEON deontic modal INCH inchoativeDET determiner INDEP independentDEV developmental INSTR instrumentalDIM diminutive INTR intransitivizer/intransitivexviIRED initial reduplication PAST past tense adverbialIRL irrealis PERF perfectiveLNK link PL pluralLOC locative POSS possessiveMASC masculine PRES presentMID middle marker PROG progressiveMIN -min- pre-transitivizer PROX proximalMUT mutative QUOT quotativeNEG negative RED reduplicationNOM nominalizer REFLEX reflexiveNST non-subject-topic REP reportativeOBJ object marker SG singularOBL oblique marker TR transitivizerOCC occupation U.POSS k?- unrealized possessorPART particle YNQ yes/no questionPASS passivexviiAcknowledgementsThere are many individuals and organizations which I wish to thank.First, I wish to thank my Okanagan language teachers in the Upper NicolaBand, in Quilchena and Douglas Lake. My two primary consultants, Lottie Lindleyand Sarah McLeod, showed overwhelming patience in their willingness to workwith a linguist, and I thank them for being willing to share their language with me.I thank Lottie for her hospitality, and for offering me accomodation on more thanone cold winter night. I also wish to thank Hank Charters, Nancy Saddleman, RitaStewart, Theresa Tom, and Wilford Tom for working with me on several occasions.Without Sharon Lindley inviting me into the community to do language work,none of this would have been possible, and she deserves high praise for her rolein organizing the annual Upper Nicola Language and Culture Camps at GlimpseLake, and for being so very supportive of my work.Next, I wish to thank my committee chair, Dr. Henry Davis, for directingthis dissertation, for sharing his broad and deep knowledge of the Salish languagefamily, for his balanced and well-thought-out Northern Interior Salish perspectivewhich helped to contextualize and temper my own ideas about Okanagan syntax,for being willing to read and critique numerous drafts of this work at its variousstages through the years, for driving me to focus on what I knew instead of worry-ing about what I did not understand, and not least of all for his ongoing financialsupport.I also wish to thank Dr. Lisa Matthewson, committee member, who introducedme to Salish field linguistics during 2006-2007 with a year-long field methodscourse on Lillooet (St??t?imcets). This provided me with useful analytical experi-ence, and a chance to tune my ear to Salish phonology, both of which were crucialxviiiskills for my future work in Okanagan. Lisa?s knowledge of the semantics of Lil-looet, and particularly semantic theories of determiners, were extremely helpful tome in my own work on Okanagan.Thanks also go to Dr. Hotze Rullmann, committee member, who introduced meto formal semantics my first semester at UBC, and with that a new way of lookingat language. His unwaivering interest in First Nations languages, and insistence onclear and concise argumentation, has been extremely helpful to me, particularly inthe latter stages of writing this dissertation.I should also take this opportunity to thank Dr. Anthony Mattina and Dr. NancyMattina, for being champions of the Okanagan language, and for doing an enor-mous amount of linguistic groundwork for Okanagan and the Southern Interior.Without this, and without such tools as the Colville-Okanagan Dictionary, my taskwould have been nearly impossible. As a MA student, Anthony Mattina helped toprovide me with a background in Salish linguistics which enabled me to continuemy studies at UBC.I also wish to thank Laura Thevarge and Herman Dan, St??t?imcets teachers andlanguage consultants, for inspiring me to continue doing fieldwork on Okanagan.Thanks to Yvonne H?bert for her previous work, and for allowing me access to herUpper Nicola corpus, and to SP?mlaPxw (Dr. Michele K. Johnson) for inspiring meto push on through the last few months of the dissertation, and for looking for thelight at the end of the tunnel.Thanks go to the other members of my graduate student cohort, Joel Dunham,Jennifer Glougie, and Mark Scott, with whom I managed to survive the first twoyears of coursework. Thanks especially to Joel who accompanied me out to theUpper Nicola for the first year or so, and helped to edit the two collections ofOkanagan texts which I have published.Dr. Michael Rochemont deserves thanks for reading and commenting on ear-lier drafts of my work on predication, and for being willing to meet on a weeklybasis to discuss predication and related topics.An enormous thanks to my friends, family, and loved ones who have beensupportive of me through the duration of this program. I have spent far too littletime with all of you.Finally, my research would not have been possible without financial supportxixthrough grants from the Jacobs Research Fund in Bellingham WA, the Ameri-can Philosophical Society in Philadelphia, PA, and through Henry Davis?s SSHRCgrant #410-2008-2535.xxChapter 1IntroductionIn this introductory chapter, I first give a brief overview of the central problemswhich this dissertation seeks to address (1.1). I then describe the geographic distri-bution of the Okanagan language and its sister languages of the Southern Interior(1.2). Next, I discuss ethical considerations (1.3). I then discuss my methodologyfor data collection (1.4.1), and describe how I present my data in terms of inter-linear glossing as well as the orthographic conventions I assume (1.4.2). Next Idiscuss relevant previous linguistic work on Okanagan and Interior Salish (1.5).This chapter ends with an outline of the dissertation (1.6).1.1 Purpose of this StudySouthern Interior Salish languages such as Okanagan are well-known for theirmorpho-syntactically rich aspectual systems (N. Mattina 1996b), but little descrip-tive or theoretical work has focused on the nominal domain. In comparing thesyntax and semantics of NPs (noun phrases) and DPs (determiner phrases), thisdissertation makes a substantial empirical contribution to the field, and lays thegroundwork for further theoretical work in this area.The major goal of this dissertation is to characterize and analyze the distribu-tion and function of NPs and DPs in sentence types without a main-clause eventivepredicate. These sentence types are what I shall refer to as non-verbal predications,and are also commonly referred to as copular clauses or copular predications for1English. I now discuss two types of non-verbal predication for Okanagan: directpredications and DP-DP structures.First, consider that in English, main clause non-verbal predications must be ac-companied by some inflected form of the verb be, since adjectives (e.g. productivein 1a) and nominals (e.g. boss in 1b) in English cannot be licensed as syntacticpredicates in main clause environments without a copula.(1) a. John is productive.b. Mary is boss.Although Okanagan has no overt copula (A. Mattina (2001, fns 5,10) andN. Mattina (1996b, 30)), Okanagan non-verbal predications are interpretively sim-ilar to copular clauses in English (Higgins, 1973). The examples in (2) exemplifydirect predications, and show how APs, NPs, and PPs appear to pattern with simpleVPs in being able to select directly for their arguments, which are in these casesDPs headed by the determiner iP. A comparison between (2) and (3) shows that therelative ordering of predicate and argument is not strict for direct predications.(2) a. [Payx?wt AP]tirediPDETtk?milxw.womanThe woman is tired.b. [s-yxw?p-m@x NP]NOM-shuswap-personiPDETp@ptw?naxw.old womanThe old woman is Shuswap.c. [ ?klLOCs@n-lasy@?t-[t]@n PP]LOC-plate-INSTRiPDETlpot.cupThe cup is in the cupboard.d. [c-xwuy V P]CISL-goiPDETsxw-l ?k-?m.OCC-bound-MIDA policeman came.2(3) a. iPDETtk?milxwwoman[Payx?wt AP].tiredThe woman is tired.b. iPDETp@ptw?naxwold woman[s-yxw?p-m@x NP].NOM-shuswap-personThe old woman is Shuswap.c. iPDETlpotcup[ ?klLOCs@n-lasy@?t-[t]@n PP].LOC-plate-INSTRThe cup is in the cupboard.d. iPDETsxw-l ?k-?mOCC-bound-MID[c-xwuy V P].CISL-goA policeman came.These data show that adjectives and nouns have similar distributions to verbs inOkanagan, as in other Salish languages (cf. Kinkade (1983), Jelinek (1998), Davis(1999a), Kroeber (1999) and many others). As such, adjectives and nominals inOkanagan appear to be able to directly predicate themselves of their arguments.The implications of this are quite interesting with regards to theories of copularpredication. First of all, assuming that NPs, for example, may function as predi-cates in Salish (Kroeber, 1999; Davis et al., 2004), there may be no need for anycopula in non-verbal predications involving a main clause NP (contra Baker (2003)who assumes a predicational copula in these cases).Regarding the predicate-argument distinction in Okanagan, the argument statusof the iP DPs above is established by data like (4-5):(4) a. iPDETtk?milxwwomana/the woman*She is a woman.b. tk?milxwwoman\u000B.he/she/itShe is a woman.3(5) a. Payx?wttired\u000B.he/she/itHe/she/it is tired.b. Payx?wttirediPDETtk?milxw.womanThe woman is tired.The DP iP tk?milxw ?the woman? in (4a) is not a complete sentence since there isno main clause predicate in this form, only a saturated argument expression, andnull predicates are not possible. Okanagan, like other Salish languages, is a pro-drop language. This means that (4b) and (5a), unlike (4a), are interpretable ascomplete sentences given an appropriate context. In (5b), the subject is overtlyrealized as a DP consisting of the determiner iP and its NP complement. The gen-eralization is that Okanagan predicates do not require overt arguments in order tobe interpretable as complete sentences; however, Okanagan argument expressionsdo require an overt predicate. Furthermore, given that NPs like tk?milxw ?woman?can be predicates, the distinction in (4-5) is evidence that a determiner makes aconstituent non-predicative (Longobardi, 1994; Chierchia, 1998).Assuming that lexical categories are inherently predicative, and that iP DPsare individual-denoting argument expressions, a simplified semantic analysis of asentence like (2b) is given as Figure 1.1:Figure 1.1: Function Application in an Okanagan Direct PredicationSt[[Shuswap]]([[the old woman]])NPPred?x[Shuswap(x)]syxw?pm@xDPSub je[[the old woman]]iP p@ptw?naxwIn brief, there appears to be a semantic distinction between syntactic categories,4where NPs, APs, VPs, and PPs can be predicative, while DPs cannot.1 This fitswith theories that D rather than N is crucial for referentiality (Longobardi, 1994)and that the determiner is crucial for converting a nominal predicate into an argu-ment (Chierchia, 1998).There is a class of structures in Okanagan that raise some potentially seriouscomplications for this account, however. These are what I refer to as DP-DP struc-tures, examples of which are shown in (6-8). Though there is little mention ofDP-DP structures in the literature, N. Mattina (1996b, 30) notes that exampleslike (6-8) ?consist of two adjacent [DPs] standing in an equivalence relationshipinterpreted as [DP = DP]. Equational sentences have neither a lexical verb nor acopula.?2(6) [ix?PDP]DEM[iPDETp@ptw?naxw DP].old.ladyShe is the old lady.(7) [Spike DP]Spike[iPDETylm?xw@m DP].chiefSpike is the chief.(8) [iPDETsq@ltm?xw DP]man[iPDETs@xw-p?x?-@m DP].OCC-hunt-MIDThe man is/was a hunter.Assuming that all of the DP expressions in (6-8) are expressions of type e, a seman-tic derivation along the lines of Figure 1.1, without any functional intermediary, isnot possible; but there is further evidence against analyzing either of the two DPsin (6-8) as predicates. Unlike sentences involving lexical predicates, as in (2-3),constituent ordering is either not free (in the case of demonstratives and propernames, given in (9-10)) or leads to interpretive differences (in the case of iP DPs,given as 11):1PPs are only sometimes acceptable as predicates in Upper Nicola Okanagan, and are judgedgrammatical or ungrammatical seemingly at random. The reasons for this are unclear.2N. Mattina (1996b) uses ?NP? where I use ?DP?, hence the square brackets.5(9) *[iPDETp@ptw?naxw DP]old.lady[ix?PDP].DEMThe old lady is her.(10) *[iPDETylm?xw@m DP]chief[Spike DP].SpikeThe chief is Spike.(11) [iPDETs@xw-p?x?-@m DP]OCC-hunt-MID[iPDETsq@ltm?xw DP].man#The man is/was a hunter.The hunter is a man.What explains the word order restriction of DP-DP structures? This question isparticularly interesting in light of theoretical work on copular predication, and Hig-gins? (1973, 1979) taxonomy of these structures, which I discuss in some detail inchapter 2. I claim that the word order restriction shows that structures directly anal-ogous to specificational sentences (a.k.a ?inverse predications?) in English (Hig-gins, 1973; Moro, 1997) are ungrammatical in Okanagan.As a reasonable null hypothesis, we could guess that specific discourse con-ditions make the Okanagan examples in (9) ungrammatical, since in English atleast, specificational sentences are only felicitous in a subset of the contexts whichsupport predicational sentences (Higgins, 1973; Mikkelsen, 2005). For example,(12-13) show that the specificational sentence ?The winner is Sam? is only felici-tous if ?the winner? is a topical expression, not if it is in focus (Mikkelsen, 2005).3In Okanagan, however, the relative discourse status of the initial DP is irrelevant insuch question/answer contexts, and the inverse, specificational configuration willalways be ungrammatical.(12) a. Q: Who is the winner?b. A: Sam is the winner. (predicational)c. A: The winner is Sam. (specificational)3By ?topic?, I informally refer to old information, or information that is already established in thediscourse, while by ?focus?, I mean new information being introduced to the discourse (Rochemont,1986).6(13) a. Q: Who is Sam?b. A: Sam is the winner. (predicational)c. A: #The winner is Sam. (specificational)Elucidating the source of this word order restriction and of other differences be-tween direct and DP-DP structures, will be the focus of the dissertation.In addition, I will explore the structure and interpretation of Okanagan clefts,and will show that they exhibit the same word order restriction as DP-DP structures,and that they share other important information structural and morpho-syntacticparallels as well. As such, I claim that both simple DP-DP structures and cleftsderive from one underlying equative configuration.1.2 The Okanagan LanguageThe Southern Interior sub-branch of the Salish language family consists of Colville-Okanagan (Nsy?lxc@n), Moses-Columbian (NxaPamxc?n), Coeur d?Alene (Snchitsu?-umshtsn), and the dialect continuum known as Spokane-Kalispel-Flathead (Seli?).The geographic relation between the Southern Interior languages and other Salishlanguages is represented in Figure 1.2 below.7Figure 1.2: Geographic Distribution of Salish LanguagesMap reproduced from The Salish Language Family: Reconstructing Syntax byPaul D. Kroeber (p. xxxi) by permission of the University of Nebraska Press.Copyright 1999 by the University of Nebraska Press.Okanagan is spoken in South-central British Columbia and North-central Wash-ington. It is critically endangered, being spoken by only about 250 speakers inCanada (FPHLCC, 2010), and by fewer in the United States. Four major dialectareas are recognized for the Okanagan language. These are represented in Figure1.3 as the northern dialects of ?Okanagan? proper and ?Lakes?, and the southerndialects of ?Sanpoil? and ?Colville?.8Figure 1.3: Okanagan Dialect Areas (from Doak (1983, 17))There are finer-grained dialect distinctions to be made as well, however. Asub-dialect of Okanagan proper is spoken in the Upper Nicola River valley andaround Nicola Lake, in the extreme northwest periphery of the Okanagan languagearea. I refer to this dialect as the ?Upper Nicola? dialect, though it is sometimesalso referred to as the ?Douglas Lake? dialect. The majority of the data in thisdissertation come from the Upper Nicola dialect. Differences between Okanagandialects are primarily lexical in nature, though I have also found several grammati-cal differences between the Upper Nicola dialect and published data from the mainOkanagan dialect and from Colville.44The grammatical differences which I have found are primarily related to the distribution of the9The Upper Nicola Okanagan Band is centered around the Douglas Lake (Sp?x?-m@n) and Quilchena (N? ?q??m@lx) reserves, close to the town of Merritt, B.C. TheNicola Valley was originally inhabited by the Nicola Athapaskan people (cf. Fig-ure 1.2), who lived in the area until they were absorbed by Salish-speaking peoplesin the 19th century (Boas and Teit, 1930). The Upper Nicola dialect of Okanaganis spoken by perhaps as few as 12 speakers (Sharon Lindley, p.c.), all in their sev-enties and eighties. This dialect is interesting, in part, because of its divergencefrom the more commonly heard Okanagan Valley dialect. These divergences stemfrom several factors, including geographic isolation, a high degree of bilingual-ism (Thompson-Okanagan), but also influence from neighboring Thompson andShuswap groups, who contemporaneously with the Okanagans, used the NicolaValley as a summer hunting ground (Boas and Teit, 1930).1.3 EthicsThe scope of my work falls under Dr. Henry Davis? SSHRC grant #410-2008-2535, and proceeds under the ethical consent guidelines as outlined in the UBCBehavioural Research Ethics Board.The opportunity to conduct linguistic work with Upper Nicola speakers arosefrom a request by Sharon Lindley, to Henry Davis, that linguists should comeinto the community in order to document the language and to assist in producingcurriculum resources. Sharon Lindley is the former principal of Nk?wala schoolin Douglas Lake, the language representative of the Upper Nicola Band at theEn?owkin Centre in Penticton, and by community consensus, an authority figureand champion of the language. As part of the process of documenting the lan-guage, it was understood that I could collect materials which would enable me toproduce a dissertation.Prior to commencing work with speakers of the Upper Nicola dialect in De-cember 2008, Dr. Davis and I attended an Elders? meeting, at the behest of SharonLindley, and at which the majority of the remaining speakers in Douglas Lake andQuilchena were present. At this meeting, the Elders identified the ?most fluent?determiner and oblique marker before nouns, given a specific grammatical context. I discuss thesedifferences when relevant.10of these speakers as being Lottie Lindley, and recommended that she be our pri-mary language resource in the community. After personally contacting Lottie andother speakers who I thought might be interested in occasional or regular languagework, we set up an initial appointment. The speakers were then asked to sign anethical consent form, in accordance with the requirements of the UBC BehaviouralResearch Ethics Board.With permission of the speakers and community, my textual and sound dataare either currently, or will be, archived at the University of Washington SpecialCollections, Melville Jacobs archive, in Seattle WA; and at the American Philo-sophical Society?s archives in Philadelphia, PA. Additionally, I have given copiesof all my data to the community; both directly to the speakers with whom I work,as well as to Sharon Lindley. In building my relationship with the Upper Nicolacommunity, I have endeavored to create and share language resources which mightbe deemed useful in language preservation and education, such as several subtitledand dubbed Okanagan films, as well as two collections of Upper Nicola narrativesby Lottie Lindley (Lindley and Lyon, 2012, 2013).1.4 The DataThe data in this dissertation come primarily from two speakers of the Upper Nicoladialect, Lottie Lindley and Sarah McLeod. I have worked with these two Elders farmore than with any other speakers in the community. I have also conducted severalelicitation sessions with Hank Charters, Nancy Saddleman, Rita Stewart, WilfordTom, and Teresa Tom during the course of my work in the Upper Nicola valley.1.4.1 MethodologyThe bulk of the data I cite consists of elicited material. Canonically, I give thespeaker a sentence in English which may be paired with a context, and the speakertranslates the English sentence into the Okanagan equivalent. In other cases, Iconstruct a context, and give the speaker a question in Okanagan, and they providethe contextually appropriate Okanagan answer.Ideally, I choose data that is volunteered by speakers in response to a given En-glish sentence. In some cases, in order to show a contrast in terms of grammatical-11ity, I cite data which I have constructed that has been judged either grammatical orungrammatical by a speaker. Constructed data is based on a volunteered form, butminimally altered. In most cases, grammatical forms have been volunteered whichare directly parallel to constructed data which I cite. Unless otherwise noted, thedata patterns which I investigate have been found to be consistent across speakers.I also utilize data that does not come from an elicitation session. Other datasources include sentences which are gleaned from volunteered texts, either fromLottie Lindley (Lindley and Lyon, 2012) or Sarah McLeod, from Colville sourcessuch as The Golden Woman (A. Mattina 1985) or Dora DeSautel?s ?aP k?capt?kw?(A. Mattina and DeSautel 2002), or from Yvonne H?bert?s unpublished UpperNicola corpus. I note data which has been extracted from sources other than myown. I make every effort to cite Upper Nicola data where possible, since it is pos-sible that there is significant dialect variation related to one or more crucial pointsof grammar in this dissertation.1.4.2 Interlinear Glossing and Orthographic ConventionsI use a three-level representation when presenting interlinear data: A ?near-phonemic?representation, a morpheme gloss, and a translation. Grammatical data is un-marked, ungrammatical data are introduced by an asterisk (*), marginally accept-able data or data whose grammaticality status is unclear are introduced by a super-script question mark (?), and contextually infelicitous data are introduced by a hashmark (#). An example is given below:(14) ucDUBwik-ssee-[DIR]-3SG.ERGiPDETsq@ltm?xwmaniPDETx?xwt@m?little.girlDid the man see the little girl?I discuss each of these three levels separately in the following sub-sections.Near-Phonemic TranscriptionThe first level consists of an Okanagan transcription using a standard American-ist (a.k.a. Northwest) orthography. This orthography has traditionally been usedin Salish linguistics since the 1960?s. It is the dominant orthography for Okana-12gan, and has been used in such references as the Colville-Okanagan Dictionary(A. Mattina 1987).5 I refer the reader to H?bert (1982b) and H?bert (1982a) for adetailed discussion of Americanist orthographic conventions.Because the phonetics of Okanagan by-and-large correspond transparently tothe underlying phonemic representation, I have decided not to give an additionalphonetic line when presenting data, unlike in A. Mattina and DeSautel (2002), forexample. The symbols I use, and their approximate phonetic values, are given inthe consonant and vowel charts in section 3.1. My transcriptions are ?near phone-mic? rather than simply ?phonemic? because I use schwa [@], which is not a fullvowel (A. Mattina 1973, 10). I use schwa phonetically, as I hear it. Most often, aschwa functions to break up consonant clusters, and in my own opinion makes theOkanagan easier to read.There are exceptions to the generalization that Okanagan morpho-phonologyis transparent, including for example (i) null transitivizers in 1st and 3rd personergative constructions with inherently stressed (a.k.a. ?strong?) roots (see 14), (ii)absorption of the final nasal in 1st and 2nd person possessive in- and an- precedinga nominal beginning with s- and (iii) the reduction of the iP determiner before 1stand 2nd person possessive prefixes. For cases like (i), I indicate a null transitivizerwithin square brackets in the morpheme gloss line (e.g. [DIR] in 14). For caseslike (ii), I indicate the nasal in square brackets. For cases like (iii), I will give aniP determiner in square brackets. My use of parentheses is distinct from my use ofsquare brackets: parentheses indicate optional material.I mark primary stress at the word level by an acute accent. For mono-syllabicwords, or words with only one full vowel and no perceptible schwa, I do not markstress. I divide each Okanagan word into morphemes, using a hyphen (-). Each5At least two other orthographies exist for Okanagan. The first was originally developed byRandy Bouchard in the 1960?s and represents Okanagan phonemically using Latin characters. It isessentially equivalent to the practical orthography used to write St??t?imcets (the Lillooet language)(cf van Eijk (1997)), yet is no longer used for Okanagan as far as I am aware. The second wasdeveloped by Christopher Parkin and SQam?t?caP (Sarah Peterson) at the Paul Creek language school,and is currently used at the Salish School in Spokane, WA. It is gaining currency at centers forNsy?lxc@n language such as the En?owkin Centre in Penticton, B.C., perhaps due to the relativelylarge volume of curriculum materials available in this orthography, or perhaps because it is easier tolearn than the standard Americanist orthography. The major difference between the Americanist andPaul Creek orthographies is that schwa (@) is not used in the latter; instead, the schwa is signalled byplacing an apostrophe on the immediately following consonant.13hyphenated morpheme corresponds to a gloss in the morpheme gloss line.6 I donot generally indicate null 3SG.ABS pronouns in either intransitive or transitivecontexts.For cited data from other Salish languages, I use a standard Americanist or-thography in the transcription line. I change morpheme glosses in some cases toreflect my labelling of the equivalent Okanagan morphemes. If there is no equiv-alent morpheme in Okanagan, or I do not cite an equivalent Okanagan morphemein this dissertation, I retain the author?s original morpheme gloss.Morpheme GlossThe second line of data consists of a morpheme gloss. A given morpheme mayeither consist of grammatical information, in which case it is glossed in small capsusing one of the abbreviations given in the Abbreviations table (pp. xvi-xvii), orof lexical information, in which case it occurs in normal, Roman type. Covertmorphology is indicated either in the morpheme gloss by square brackets, or in theOkanagan transcription within square brackets.Glossing conventions, and my choice of abbreviations, are primarily those usedin (Matthewson, 2005) I Wan Kwikws and Lindley and Lyon (2012), with supple-mental glosses borrowed when needed from works such as A. Mattina and De-Sautel (2002) and other sources. I have endeavoured to use standard abbreviations(e.g. DET for ?determiner?) whenever possible.TranslationTranslations of Okanagan volunteered forms consist of the English sentence whichwas given as a prompt for the Okanagan form. For cases where an Okanaganform was volunteered in response to a contextual prompt, either the speaker?s ownEnglish translation is given, or a translation which reflects the Okanagan form asliterally as possible. For constructed data involving negative judgements, I providethe closest equivalent English translation.6Cf. A. Mattina (2008) for a useful discussion on how to parse some of the more problematicforms.141.5 Salish Literature ReviewThere is a substantial body of literature in Salish linguistics, without which thisdissertation would not have been possible. I summarize the most relevant worksfor this dissertation in this section, dividing my discussion of the literature intothree parts: Okanagan, Southern Interior Salish, and Northern Interior Salish.1.5.1 OkanaganLinguistic work on Okanagan may be said to have originated with James Teit (cf.Boas and Teit (1930)), but not until the late 1960?s did intensive work on the lan-guage begin. Early work includes Watkins (1970), a dissertation on phonology, andAnthony Mattina?s dissertation Colville Grammatical Structure (A. Mattina 1973)which focuses mainly on the phonology and morphology of the language. AmongMattina?s other works are The Golden Woman (A. Mattina 1985), an interlinearanalysis of a Colville narrative, and the invaluable Colville-Okanagan Dictionary(A. Mattina 1987). I found the IJAL paper The Colville-Okanagan Transitive Sys-tem (A. Mattina 1982) to be a useful reference for understanding the Okanaganpronominal system. Nancy Mattina?s Aspect and Category in Okanagan Word For-mation (N. Mattina 1996b) provides an extremely useful analysis of the Okanaganaspectual and tense systems.The Upper Nicola dialect itself has received comparatively little documenta-tion, with the exception of a phonological overview (Pattison, 1978), and a series ofpapers by Yvonne H?bert, including her dissertation Transitivity in (Nicola Lake)Okanagan (1982b), and a report to the Canadian Ethnological Service Clausalstructure in (Nicola Lake) Okanagan (H?bert, 1982a).A highly useful resource for Okanagan linguistics is The Kinkade Collection:the On-Line Archive of Papers for the International Conference on Salish (andNeighbo(u)ring) Languages. This on-line archive is the result of several years ofdigitizing and organizing ICSNL conference papers, some of which were otherwisevery difficult to find. Included in this collection are many important papers by An-thony Mattina on the morpho-phonology and morpho-syntax of Okanagan. ICSNLpapers which I have personally found very useful are Okanagan Aspect: A WorkingPaper (A. Mattina 1993a), a precursor to Nancy Mattina?s dissertation (N. Mattina151996b), and Okanagan sentence types: A preliminary working paper (A. Mattina2001), which makes brief mention of DP-DP structures and the absence of anycopula in the language.1.5.2 Southern Interior SalishLinguistic material and analyses from other Southern Interior Salish languages,particularly Moses-Columbian, have proved useful to me in terms of understandinghow Okanagan fits into the areal picture, and for establishing base-line hypothesesconcerning previously unresearched corners of Okanagan grammar.Most noteworthy is Nancy Mattina?s IJAL paper Determiner Phrases In Moses-Columbia Salish (N. Mattina 2006), in which I found an areal basis for many ofthe ideas which I develop concerning the semantics of the Okanagan iP determinerand how these DPs contrast with oblique-marked nominals. Another document onMoses-Columbian which I found useful was Marie Willett?s dissertation A Gram-matical Sketch of Nxa?amxcin (Moses-Columbia Salish) (Willett, 2003). Addition-ally, Ewa Czaykowska-Higgins has kindly made available some of her field notes.1.5.3 Northern Interior Salish (and other Salish languages)There has been much high-quality descriptive and theoretical work on the NorthernInterior Salish languages of Lillooet (St??t?imcets), Thompson (N?ePkepmxc?n),and Shuswap (Secwepemctsin). I mention some of the most relevant work here.Henry Davis?s and Lisa Matthewson?s work on Lillooet has proved indispensi-ble to me in establishing the basic syntactic and semantic premises upon which Ibuild my main arguments, specifically in three areas: relative clauses, determinersemantics, and clefts. Other scholars in the field, notably Dwight Gardiner, CarrieGillon, Karsten Koch, Paul Kroeber, and Jan van Eijk have also made importantcontributions in one (or more) of these three areas, which I discuss below.First, Davis (2002, 2004, 2010a), building on previous observations in Kroeber(1999) establishes the basic argument for Lillooet that relative clauses are formedby means of clause-internal movement of a DP. Koch (2006) shows that the samefacts hold for Thompson. These analyses provide a framework, and a point ofcomparison, for my analysis of Okanagan relative clauses, which I claim are also16formed by clause-internal movement.Matthewson (1998, 1999, 2001) provides a comprehensive analysis of the se-mantics of the determiner system in Lillooet, as well as a detailed explanation ofhow Salish determiners differ semantically from those in English. Carrie Gillon(Gillon, 2006, 2009a,b) provides a similarly detailed analysis of the Squamish de-terminer system. My own analysis of the semantics of Okanagan DPs rests heavilyon Matthewson?s and Gillon?s original work.Davis et al. (2004) investigates cleft structures in Lillooet and Northern StraitsSalish, and establishes that they imply without presupposing or entailing exhaus-tivity, and do not carry a presupposition of existence, unlike English clefts. Theseinformation structural properties also hold for Okanagan. Koch (2008a) presents adetailed analysis of focus and information structure in Thompson, and Koch (2009)provides an analysis of Thompson clefts. His argument that focus in ThompsonSalish is not realized by pitch-accent, but by linear alignment, is an important find-ing. I claim that linear alignment constraints also play a role in the informationstructure of Okanagan.Other works which have been helpful to me include Gardiner (1993), whichexamines the syntax of topicalization in Shuswap, and van Eijk (1997), which con-sists of a detailed grammar of Lillooet morpho-phonology.1.6 Outline of the DissertationThis dissertation is structured as follows:Chapter 2 presents theoretical background on issues involving copular predica-tion and clefts, and a discussion of the analytical tools which I use for my analysisof Okanagan predications and equatives.Chapter 3 presents some basic aspects of Okanagan grammar, including phonol-ogy, pronominal inflection, and brief notes on the transitivity and the tense andaspectual systems, followed by a more in-depth discussion of word order.Chapter 4 investigates the general distribution of DPs in non-predicative con-texts in Okanagan, and the internal structure of DPs, particularly those headed bythe determiner iP. The internal structure of DPs is particularly important with re-gards to the distribution of DP-internal ?prepositions?, or locative markers, which17are a general characteristic of the Southern Interior. These data are important forunderstanding the syntax of relativization, as presented in chapter 6.Chapter 5 consists of a detailed investigation of the semantics of the determineriP, and DPs headed by this determiner. I argue that iP is non-presuppositional andcontext-sensitive, similar to Squamish deictic determiners (Gillon, 2006, 2009a),but non-deictic. The semantics of iP DPs is crucial for understanding DP-DP struc-tures, as well as the non-presuppositional and non-exhaustive characteristics ofOkanagan DP-DP structures and clefts.Chapter 6 presents data on attributive modification and relative clause modi-fication. Diagnostics are developed for distinguishing these two types of nominalmodification, and a movement analysis based on locative and oblique-centered rel-ative clauses along the lines of Davis (2004, 2010a) and Koch (2006) is motivated.The data and analysis of relative clauses in this section are important for clarifyingthe stucture of cleft residues, which I claim to be categorially DPs in Okanagan.Chapter 7 introduces Okanagan DP-DP structures, and discusses how thesediffer from direct predications. Based on a word order restriction which I take tobe evidence that neither constituent in a DP-DP structure is predicative, I claimthat DP-DP structures involve a null equative copula (Heycock and Kroch, 1999;Romero, 2005). I then discuss information structural properties of DP-DP struc-tures. These include an exhaustivity implicature (Davis et al., 2004), an absence ofany presupposition, and a requirement that a referential, focused DP occur initially.This means that Okanagan does not have specificational sentences (a.k.a. ?inversecopular clauses?) (Higgins, 1973). Explaining the absence of specificational sen-tences presents a challenge: the observation is that in DP-DP structures involvingeither a proper name or a demonstrative and an iP DP, the iP DP must follow theproper name or demonstrative. This poses a problem for a simple equational anal-ysis since both DPs denote individuals, and neither the equative functional headnor focus can distinguish among different types of DPs. Intuitively, the distinc-tion between demonstratives and proper names on the one hand, and iP DPs on theother, is that the former are directly referential, whereas the latter are not. I sug-gest that the Okanagan equational head is of type <,> (Romero, 2005;Comorovski, 2007), and links an intensional individual (an iP DP) to its extension.The equative head assigns a feature ?F? to its second argument, and this feature is18interpretable as ?focus?. Focus-sensitive alignment constraints (Koch, 2008a) thenensure that the focused DP occurs left-most.Chapter 8 introduces Okanagan clefts, and discusses how they are similar to,and different than clefts in other Salish languages. I show that Okanagan cleftsconsist of two DPs, and have an information structure identical to that found inDP-DP structures, as discussed in chapter 7: they imply exhaustivity (Davis et al.,2004), do not carry any presupposition of existence, and require that the focusedDP precede the residue DP. I then discuss morphosyntactic evidence that cleftsare structurally equivalent to DP-DP structures, which implies that clefts, too, areequatives.Chapter 9 discusses typological and theoretical implications of my analysis,addresses some further questions, and concludes.19Chapter 2Predication and Equation:Theoretical BackgroundThis chapter presents theoretical background and tools relevant to my analysis ofOkanagan predications and equatives, and consists of four main sections: (i) pred-ication versus equation, and Higgins? (1973,1979) taxonomy; (ii) syntactic andsemantic theories of specificational copular sentences; (iii) focus and informationstructure; and (iv) clefts.1 A more detailed outline of this chapter follows.First (section 2.1), I present semantic background on predication and equation,and discuss the question of whether the English copula be is best analyzed as beingambiguous between a predicational and an equative copula, or is unambiguouslypredicational (Partee, 1986). I next discuss Higgins? (1973,1979) taxonomy ofcopular sentences, and summarize more recent efforts in the literature to simplifythis taxonomy (Mikkelsen, 2011).Second (section 2.2), I contrast two analyses of specificational copular clauses.One school of thought claims that all non-verbal predication is mediated via thesame functional head (Adger and Ramchand, 2003; den Dikken, 2006) and thatspecificational clauses are derived from predicational clauses by syntactic inver-sion (?predicate raising?). The other school claims that predicational and equative1Particularly useful to me in drafting this section were summary articles (or articles with goodsummaries) on copular predication (Mikkelsen, 2011), clefts (Reeve, 2007), and focus and informa-tion structure (Krifka, 2008).20copular clauses are structurally distinct, and that specificational sentences are atype of equative (Heycock and Kroch, 1999; Heycock, 2012). I discuss in somedetail the problem of how best to treat specificational sentences, as pragmaticallyasymmetrical, within an equational semantics. Of immediate relevance to Okana-gan are Romero (2005) and Comorovski (2007), who claim that the equative copulais sensitive to intensionality. I present a similar analysis of the Okanagan equativecopula in chapter 7.Third (section 2.3), I discuss the alternatives-based approach to focus repre-sentation (Rooth, 1985, 1992). Focus theory is an important component of myanalysis of Okanagan equatives, since I claim that the subject of an equative is al-ways a focused element. To close this section, I summarize the findings of Koch(2008a) and Koch and Zimmermann (2009) with regards to focus alignment inneighbouring Thompson River Salish, and the non-universality of the stress-focuscorrespondence.Fourth (section 2.4), I discuss theories of English clefts which analyze the cleft-ing pronoun as a discontinous definite description with the residue clause, andwhich link the semantic and pragmatic effects of English clefts to the semantics ofthe definite determiner (Percus, 1997; Hedberg, 2000). In chapter 8, I show howOkanagan clefts support these theories.2.1 Predication and Equation2.1.1 Defining Predication: Semantic IssuesThis section introduces some of the basic semantic concepts underlying theories ofpredication, including a brief discussion of some complications which arise frominterpretive ambiguities in English predication, and the importance of correctlyidentifying the ?locus? of predication. These foundational issues are important forunderstanding how Okanagan predication is both similar to and different from thatfound in English.21Predication Versus EquationThe English copula be mediates relations of predication and equation (a.k.a. iden-tity) between two words or phrases. Whether the relation happens to be one ofpredication or equation is partially dependent on the semantic type of the words orphrases in the relation.2 Consider the following two sentences:(1) a. Predication: Tully is a bank robber.b. Equation: Cicero is Tully.Properly speaking, a predication relation is one that holds between an indi-vidual and a property. In other words, an individual x is understood as having aproperty P, or x is a member of the set denoted by P. Thus in English, in order for(1a) to be true, Tully must belong to the set denoted by a bank robber.An equative relation is one that holds between two individuals. In English,if we say Cicero is Tully, the most straightforward interpretation is that we areasserting that the individual denoted by Cicero is identical to the individual denotedby Tully (1b). Note however that (1b) also has a predicational reading in the contextwhere, for example, Cicero is playing the part of Tully in a play. This serves toillustrate that the distinction between predication and equation cannot necessarilybe understood strictly in terms of inherent differences between noun classes, butinvolves referentiality more generally.The Locus of Predication and Equation: The Copula or a Pred-head?Under some theories, the copula be itself instantiates the predication and identityrelations (Partee, 1986); however, not all theories of English predication automat-ically assign the copula a predicative semantics. Moro (2000) and den Dikken(2006), for example, attribute the semantics of copular predication to an abstractand usually covert functional projection called the Pred-head (Bowers, 1993; Baker,2003)3, which links the subject to the predicate in a small clause configuration. For2Higher order predications and equational relations are also logically and linguistically possible(Partee, 1986; Heycock and Kroch, 1999).3den Dikken (2006) refers to the Pred-head as a Relator. The copula can, but does not have tobe, a Pred-head, but is in the sentence The earth might be round according to den Dikken (2006, 15),since T is filled by a modal.22these theories, the copula is often relegated to the role of tense-carrier, and so min-imally conveys the information that, for example, Tully is a bank robber, or thatCicero is Tully, at the present time. Under a Pred-head analysis, sentence (1a) maybe represented as follows:Figure 2.1: The Copula ?be? and a Null Pred-headTPDPi (Sub j)TullyT?TisPredPti Pred?Pred\u000BDP(Pred)a bank robberFor the structure underlying Figure 2.1, the copula selects for a PredP smallclause, headed by a null functional projection, the Pred-head. The Pred-head linksthe predicate complement DP a bank robber to the referential subject Tully, andthe subject raises over the copula. I discuss the motivation for this structure in latersections, but suffice it here to note that it is the Pred-head which functions as thepredicational intermediary in Figure 2.1, not the copula ?be?.Copular Complements and Interpretive AmbiguitiesComplements of copulas in English come from a range of syntactic categories, asshown in (2) below. They are only rarely NPs, as with boss in (2a). A locationmay be predicated of an individual John by means of a PP predicate (2b), and anattributive property by means of an AP predicate (2c). A property may also bepredicated of John by means of a definite DP (2d).44This is not necessarily always the case for (2d) since multiple readings are available. Higgins(1973) holds that the definite description is either predicated of the subject John, or else identifieswho John is. These two interpretations correspond to Higgins? predicational and identificationalclasses, respectively. There is a third, equative interpretation of (2d) as well. See section 2.1.2 for adiscussion of Higgins? taxonomy.23(2) a. John is [bossNP].b. John is [from HuntsvillePP].c. John is [tallAP].d. John is [the President of the United StatesDP].I assume that the semantic relation between an individual John and the pred-icates in (2a-c) may be captured assuming a formalism like (3) for the Englishcopula, where x = John and P = boss, from Huntsville, tall (Williams, 1983):(3) ?P?x.P(x)Insofar as the President of the United States in (2d) denotes an occupation, italso denotes a property of John, and the predication relation in (2d) may likewisebe represented by (3). But if we instead assume one fairly standard analysis ofthe definite determiner the (4) (Heim, 2011), the DP in (2d) will denote a maximalindividual, and we are faced with the problem of having two individual-denotingexpressions, but no predicate.5(4) [[the]]= ?P : ?x?y[P(y)? x = y].?x.P(x)There are two basic solutions to this problem, discussed at length by Partee (1986).One analysis, which may be referred to as ambiguous be, proposes that be is am-biguous between a denotation like (5a, cf 3) and one which equates two individuals,as in (5b):65The formula in (4) takes a nominal predicate P as its argument, and presupposes that there is anindividual x which has the property P, and that for all other individuals y, if y has the property P, itmust be identical to x. This is a presupposition of uniqueness. It then asserts (by means of the iotaoperator ?) that x is the only individual with the property P.6Under the ?ambiguous be? approach, two individuals may also be equated by means of the ?up?operator ? of Chierchia (1984), as in (i), which maps an individual onto the singleton set of allindividuals that are identical to it:(i) ?y?x[?y(x)]A copula like (i) will yield a proposition essentially equivalent to the result of (3), the only differencebeing the semantic type of the first argument. Partee (1986) refers to the ?up? operation as Pred, andnotes that (i) is in some ways conceptually preferable to (5b), since ambiguous be can then at least beunderstood as always linking a predicate expression, whether inherent or derived, with its argument.Locating a type shift in the English copula itself (i) does not remove the need for a predicationalcopula (5a) in cases where the complement expression is a property-denoting NP, PP or AP (cf. 2).24(5) a. ?P?x.P(x)b. ?y?x.[x = y]The other analysis, which Partee (1986) instead argues for, may be referredto as unambiguous be, and assumes a single copula (5a) by allowing definite DPsand other individual-denoting expressions to type shift into predicates before thecopula selects them as complements. This means that English DPs must be able tofreely type-raise into properties.There is much debate in the literature on whether a type-shifting approach orone involving a separate equative copula is preferable. Choosing one approach overthe other for any given language depends in part on whether there is independentevidence that expressions in that language may type-shift. After discussing Hig-gins? taxonomy in more detail in the next section, I will touch on some of the finergrained points of this debate. For now, it is sufficient to note that copular clauseslike (2d) John is the president of the United States exhibit both predicational andequative interpretations, depending on whether or not the definite DP is construedas referential in context, and that these different interpretations receive explanationunder both the ?ambiguous be? and ?unambiguous be? approaches.An important set of questions arises as to whether the semantics of predicationcan vary cross-linguistically. More specifically, do all languages display evidencefor an ambiguity either in the locus of predication (i.e. copula or Pred-head) or inthe semantics of DPs, as English does, or are we sometimes able to dispense withambiguities altogether?2.1.2 Copular Clauses and Higgins? (1973) TaxonomyThe previous section introduced some of the basic semantic concepts and issueswhich are important to any theory of predication. I now move on to a discussion ofHiggins? (1973) taxonomy of English copular clauses. This taxonomy has been animportant standard in the literature for motivating taxonomies of non-verbal pred-ications in other languages (e.g. Danish (Mikkelsen, 2005) and Hebrew (Heller,2005)), and will also be useful for comparing Okanagan with other languages.To begin with, as discussed in the previous section, English normally requiresa tensed form of the copula be in main clause predications. The post-copular pred-25icate complement may consist of any one of a range of syntactic categories, asshown in (6).(6) a. AP Predicate: John is busy.b. NP Predicate: Lucy is boss.c. PP Predicate: Nancy is from Douglas Lake.d. DP Predicate (indefinite): Sarah is a teacher.e. DP Predicate (definite): Obama is the president.In main clause contexts, a predicate complement cannot generally precede a refer-ential subject (7a-d), unless the predicate complement is a definite DP (7e).7(7) a. AP Predicate: *Busy is John.b. NP Predicate: *Boss is Lucy.c. PP Predicate: *From Douglas Lake is Nancy.d. DP Predicate (indefinite): *A teacher is Sarah.e. DP Predicate (definite): The president is Obama.There thus appears to be something special about copular clauses containing twoDPs.The syntactic and semantic relationship between sentences like (6e) and (7e)has been the focus of much debate in the literature. Narrowing our focus for amoment onto copular clauses involving two DPs, Higgins (1973) establishes a four-way taxonomy for English copular clauses. Examples of each class are given asfollows:(8) a. Predicational: Tully is a/the bank robber.b. Specificational: The bank robber is Tully.c. Equative: The morning star is the evening star.d. Identificational: That place is Vancouver.The taxonomy is based on whether a DP is interpreted referentially or not,given a discourse context and a specific syntactic position within the predication.7This is an oversimplification, since PP inversions like (7c) are sometimes possible as highlysylistic variants, and indefinite DPs can precede referential subjects if the DP contains a modifier(Mikkelsen, 2005), e.g. A good president if ever there was one is Obama.26Table 2.1 shows how each of the DPs in (8) pattern with regards to referentiality.8I now discuss each of Higgins? categories in more detail.Table 2.1: Referentiality and Higgins? Taxonomy (adapted from Mikkelsen(2011, 1810))Copular sentence type 1st DP 2nd DPpredicational referential non-referentialspecificational non-referential referentialequative referential referentialidentificational referential ?identificational?Predicational SentencesUnder Higgins? (1973) theory, the initial DP in a predicational clause is referential,and the post-copular complement denotes a property which is predicated of thesubject. Predicational sentences like (8a) consist of a subject (e.g. Tully) and apredicate (e.g. a bank robber), linked by a tensed copula. Examples (6a-6d) arealso predicational, since AP, NP, and PP predicates all denote properties. (6e) toohas a predicational reading since the president, as an occupation, is interpretable asa property of an individual under the assumption that the DP can raise to a propertytype in this environment (Partee, 1986).In terms of their distribution in discourse, predicational sentences are unre-stricted. This stems from the information structural properties of predicationalsentences: Initial referential DPs and non-referential predicative DPs in final po-sition can both represent either new or old information, though intonation patternswill differ. Thus, (10) can answer either (9a) or (9b).(9) a. Who is Tully?b. Who is a/the bank robber?8It is unclear what ?identificational? means in Higgins? use of term in describing the referentialproperties of the 2nd DP in an identificational sentence (cf. Table 2.1, bottom right). Intuitivelyat least, both expressions in an identificational sentence are referential, and as such, these might begrouped with the equatives. The difference between the two classes is one of pragmatic function:identificationals are used for identifying names of things, generally speaking.27(10) Tully is a/the bank robber.Subjects of predicational sentences may also contain a modifying clause (11a) orbe referential WH-clauses, as in the predicational pseudocleft (11b):(11) a. The card/present/thing I bought for Sue is expensive.b. What I bought for Sue is expensive.The predicative status of the complement in a predicational clause is confirmedby English small clauses, which normally occur in embedded contexts as com-plements of Exceptional Case Marking (ECM) verbs like ?consider? (12a). Smallclauses are truth conditionally equivalent to full CPs (12b), but optionally lack anycopula (12a). The generalization here is that referential expressions, like John in(12c), are not permitted as complements within a small clause unless an overt cop-ula is present.(12) a. I consider [John (to be) a dangerous driver SC].b. I consider [that John is a dangerous driver CP].c. I consider [a dangerous driver *(to be) John SC].Predicational sentences thus exhibit the canonical English subject-predicate order-ing (Moro, 1997) as required in bare small clauses. Inverse predicate-subject or-dering is marked in English, as indicated by (12c). This leads naturally into adiscussion of specificational sentences.Specificational SentencesIntuitively speaking, specificational sentences specify who or what something orsomeone is, rather than saying something about someone or something, as is thecase with predicational sentences (Mikkelsen, 2011, 1809). In English, specifica-tionals restrict the domain of a predicative, discourse-old initial DP by identifyinga specific individual from within that domain via the second DP (Higgins, 1973;Mikkelsen, 2011), or according to Akmajian (1979), the second, referential DPprovides a value for a variable introduced in the first, non-referential DP.Consider that specificational sentences (13c,14c) are only felicitous in a subsetof contexts for which their predicational variants are felicitous (13b,14b).28(13) a. Q: Who is the winner?b. A: Sam is the winner.c. A: The winner is Sam.(14) a. Q: Who is Sam?b. A: Sam is the winner.c. A: #The winner is Sam.The DP the winner represents old information (i.e. the ?topic?) in (13b,c), but newinformation (i.e. the ?focus?) in (14b,c). The pragmatic markedness of specifica-tional sentences may be traced to the requirement that the initial DP represent orcontain old information (Birner, 1996; Mikkelsen, 2005).9 There is thus an infor-mation structural condition on the use of specificational sentences which does notapply to predicational sentences.10Specificational sentences most commonly have a definite DP in initial posi-tion (Higgins, 1973; Birner, 1996; Moro, 1997). Simple indefinite DPs in initialposition are usually ungrammatical (15a), but are much improved when that DPcontains a modifier, as with (15b) (Mikkelsen, 2005).11(15) a. *A president is Obama.b. A president I hope to meet someday is Obama.A sub-type of specificational sentence is known in the literature as a specifi-cational pseudocleft. Two examples are shown as (16). Like in specificationalcopular sentences, the post-copular constituent is ?more referential? than the pre-copular pseudocleft clause. Specificational pseudoclefts have been important in theliterature on copular clauses since they show connectivity effects, which I brieflydiscuss in section 2.2.3.9Or under a theory like Akmajian (1979), the open variable expression denoted by the first DP ina specificational sentence must already be, in some sense, under discussion in order for the sentenceas a whole to be pragmatically felicitous.10The exact formulation of this information structural condition is unclear, since as noted inMikkelsen (2005, 160), an initial DP being discourse-old does not guarantee that a specificationalclause is possible.11Similar data lead some researchers to propose that there are pragmatic requirements on specifi-cational sentences, involving notions such as ?contextual anchoring? (Comorovski, 2007) or ?risingdiscriminability? (Heller, 2005).29(16) a. What John is is a doctor.b. What John is is honest.There are two main schools of thought concerning specificational copular sen-tences which I briefly contrast here, and discuss in more detail in later sections. Thefirst explains specificational sentences in terms of a semantic asymmetry, wherebythe initial DP (or WH-clause) is a non-referential type predicate while thefinal DP is a referential expression, of type e. This makes possible an analysisof specificational sentences as syntactic inversions of predicational sentences, de-rived by raising the predicate over the subject (Moro, 1997; Adger and Ramchand,2003; den Dikken, 2006).12,13 The second school of thought analyzes specifica-tional sentences as a type of equative, where both expressions are semanticallyreferential (Heycock and Kroch, 1999). The argument here is that the locus ofthe asymmetry is information structural rather than semantic: the initial DP (orWH-clause) consists of relatively ?old? information (i.e. ?ground? in Heycock &Kroch?s terminology), and the final DP is in focus. For these theories, then, thereis no derivational relation between a predicational sentence and its correspondingspecificational variant.The answer as to whether the asymmetry in specificational sentences is seman-tic or pragmatic in nature is not simple, especially in light of data like (17) whichmay be analyzed as specificational or predicational, depending on which DP is thefocus, and which DP contains old information.(17) The winner is the loser.Data like (17) underscore the fact that placing any given copular sentence intoone versus another of Higgins? classes often depends on the context in which thesentence is spoken, and so even if the asymmetry between the first and second DPsin specificational sentences is semantic in nature, there must be an informationstructural asymmetry which corresponds to the semantic asymmetry, and whichserves to limit the range of contexts in which specificationals are felicitous.Finally, although there seems to be general concensus that specificational sen-12This means that what Higgins terms the ?specificational predicate?, e.g. Tully in The bank robberis Tully, is rather an underlying subject.13den Dikken (2006) reduces both specificationals and equatives to a specificational class.30tences have a fixed information structure, unlike predicational sentences, this doesnot always mean that the old information must precede the new information: notethat specificational pseudoclefts like (18a) can be inverted, while retaining theirspecificational interpretation (18b) (den Dikken et al., 2000). This shows that aspecificational interpretation is not inherently dependent on the ?ground? preced-ing the ?focus?.(18) a. Otto Preminger was who I met.b. Who I met was Otto Preminger.This suggests that the notion of specification is best understood in terms of afixed information-structural asymmetry between two constituents, and not in termsof any linear requirement that a less-referential or discourse-old expression (i.e.?ground? or ?topic?) precede a more-referential or discourse-new expression.14Equative SentencesEquative (a.k.a. identity) sentences are most famously represented in the philo-sophical tradition by examples like Cicero is Tully or The morning star is theevening star. They assert that an identity relation holds between two referentialexpressions.15 Unlike specificational sentences, truly equative sentences cannot beanalyzed as syntactically inverted predications, because neither expression is func-tioning as a predicate. For example, in certain contexts when we say The morningstar is the evening star, we are really stating that there are two unique definitedescriptions which both point to the same referent.Insofar as DPs may type-raise to properties (Partee, 1987), the prediction is thata sentence like The morning star is the evening star will also have specificationaland predicational interpretations, depending on the context (cf. also discussionaround 17). Although both definite DPs make singular reference in this case andan equative interpretation is most forthcoming, consider that in answer to the ques-tion Which star is the morning star? the response may be analyzed as a specifica-14Percus (1997) discusses the ?specificational character? of clefts, and derives clefts from specifi-cational sentences.15Or more correctly, two expressions of the same type, since higher type equatives like Honest ishonest do exist (Heycock and Kroch, 1999).31tional sentence. This shows that singular reference does not obviate the possibilityof type-raising, and singleton sets, by extension, are not information-structurallyequivalent to singular referents, assuming a correspondence between type-shiftingand information structure.16 Allowing DPs to freely type-raise to properties simpli-fies the semantics of the copula, but it also means that the surface form of a copularsentence, by itself, does not necessarily determine which of Higgins? classes it fallsinto.Adger and Ramchand (2003) and Geist (2007) argue that Scottish Gaelic andRussian, respectively, do not have true equative sentences, and that sentences whichappear to be equative are actually predicational. Note that even English exampleslike Cicero is Tully have predicational interpretations, as in a context where Tullyis a character in a play, or where Tully refers to the property of being-named-Tullyrather than referring to the actual referent.Identificational SentencesHiggins distinguishes a fourth class of copular sentence, identificational sentences,which are typically used to identify names of people or things. These are usuallycharacterized by having a deictic demonstrative or demonstrative phrase in subjectposition. English examples include That place is Vancouver (8d), This basket is acedar-bark basket, or That is John. The first DP is referential, and the second DPis ?identificational?, according to Higgins (1973) (cf. Table 2.1 above).In English, many identificational sentences are surface-similar to specifica-tional sentences (e.g. 8b), except that the initial DP is introduced by a demon-strative determiner. The primary discourse function of identificational sentencesis to relate the names of people, places or things to their referents, rather than torestrict a contextually salient domain, as is the case with specificational sentences.There is an interpretive overlap between identificational sentences, and bothspecificational and predicational sentences. First, Higgins (1973) notes that spec-ificational sentences, as a rule, also have identificational readings. Consider thatin a context where we are identifying who the contextually salient president is,the president in the sentence The president is Obama is not first and foremost a16For example, the first DP in a specificational sentence may type shift to a property (Partee, 1987),and this correlates with its status as a discourse old, non-referential expression.32discourse-old property that is being predicated of Obama, but is instead a salientindividual who Obama is being equated with. This means that a demonstrativeor demonstrative phrase subject is not mandatory for an identificational reading.Specificational and identificational sentences may therefore be distinguished bythe fact that while subjects of specificational sentences are generally discourse-old(and non-referential according to Higgins), subjects of identificational sentencesare not necessarily discourse-old, but must be contextually salient, i.e. somethingthat a speaker can point to. As a second interpretive overlap, identificational sen-tences like This basket is a cedar-bark basket have predicational readings: In asorting context where we are ascribing the property of being a cedar-bark basketto a particular basket as opposed to say, the property of being a cedar-root basket,then we have a predicational reading. Other identificational sentences, like Thatplace is Vancouver, appear only to have an identificational reading.Much of the recent literature on identificational sentences has attempted to re-duce them to one or another of Higgins? classes. For example, Mikkelsen (2005)assimilates identificational sentences with simple demonstrative subjects to thespecificational class, while those with demonstrative phrase subjects are ?demon-strative equatives?. Heller (2005) claims that identificational sentences are a typeof predicational sentence, while Birner et al. (2007) analyze identificationals asequatives.SummaryThis section has reviewed some basic points concerning Higgins? taxonomy ofcopular clauses, a classification which is largely based on discourse-dependent, in-terpretative possibilities of DPs in subject versus complement position of a copularclause.It is by no means clear that Higgins? taxonomy of English copular clausescannot be further simplified. For instance, it has been claimed that specification-als may be reduced to inverted predicationals (Moro, 1997), or alternatively, toequatives (Heycock and Kroch, 1999); or that identificationals may be reduced tospecificationals and equatives (Mikkelsen, 2005), or alternatively to predicationals(Heller, 2005). Revising Higgins? taxonomy depends not only on one?s semantic33analysis of DPs, but also on how one characterizes information structure, and onthe relationship between information structure and the semantics/syntax interface,and by whether one argues that the asymmetry in specificational copular clauses,for example, is semantic or pragmatic in nature. These are all important factors toconsider when investigating these types of sentences in different languages.I now move on to a more detailed discussion of the syntax (and compositionalsemantics) of copular predication, focusing on (i) the relationship between smallclauses, predicational sentences and specificational sentences within frameworkswhich argue for syntactic inversion (Moro, 1997; den Dikken, 2006); and contrast-ing this with (ii) frameworks which argue against syntactic inversion and for anequative analysis of specificationals (Heycock and Kroch, 1999).2.2 Specificational Copular Syntax/Semantics:Predicate Raising or an Equative Head?The preceding discussion has focused on some basic issues concerning the seman-tics of predication and equation and the various types of copular clauses throughwhich predication and equation are realized, as well as some informal discussionon information structural constraints on the distribution of copular clauses. I alsoincluded a brief overview of the debate between those who argue that the asym-metry in specificational clauses is semantic in nature, and those that argue for apragmatic asymmetry. This section investigates this debate in more detail.First (2.2.1), I begin by discussing similarities between non-verbal small clausepredications and main clause predications in English, which have given rise to the-ories whereby main clause predications are derived from small clauses by raisingeither the subject or the predicate of the small clause over the copula (Moro, 1997).These small clauses are usually taken to be projections of a Pred-head (Bowers,1993; den Dikken, 2006) or other functional projection, and are syntactically asym-metrical (Kayne, 1994).Second (2.2.2), I link these theories of small clauses with theories of copularsyntax which assume that there is semantic asymmetry between the two DPs ina specificational copular clause, and that there is a derivational relation betweenpredicational and specificational sentence types such that specificational sentences34are derived by raising the predicative DP over the subject and copula (a.k.a. ?pred-icate raising? or ?syntactic inversion? (Moro, 1997)). These theories are attractivesince they offer an intuitive explanation for the semantic similarities between thesetwo types of sentences, and are economical since, for variants of these theorieswhich reduce equatives to predicationals (den Dikken, 2006), all predication canbe reduced to a single type of small clause, and a single Pred-head.Third (2.2.3), I discuss the theories which assume a pragmatic asymmetry be-tween the two DPs in a specificational sentence, but which do not assume predicateraising. Heycock and Kroch (1999) argue that specificationals are semanticallyequative, but pragmatically asymmetrical: the first DP must be a ?ground?, or givenin the discourse, roughly speaking, while the second DP must be a ?focus?.Fourth (2.2.4), I discuss more recent work by Romero (2005) and Comorovski(2007) who have argued that specificationals are equative in the sense that theyequate two individuals, but are nevertheless semantically asymmetrical in the sensethat the specificational subject must be intensional.This discussion is relevant to Okanagan for the following reasons: I will showthat Okanagan does not have predicate raising (7.2.2, 7.3.2), which renders theinversion analysis inapplicable (Moro, 1997; den Dikken, 2006), and favors anequative analysis (Heycock and Kroch, 1999). Okanagan does not show connec-tivity effects for independent reasons (7.3.3), but does have DP-DP sentences witha fixed information structure and a fixed word order. I argue that the fixed infor-mation structure relies on a distinction between intensional and non-intensionalDPs (Romero, 2005; Comorovski, 2007), along with linear alignment constraintson focus (7.5).2.2.1 Small ClausesThis section briefly discusses small clauses in English, as a necessary backgroundfor syntactic theories of predication and equation.English small clauses often occur in embedded contexts as complements ofExceptional Case Marking (ECM) verbs like ?consider? (19a). Many claim that thesmall clause subject John and the predicate a dangerous driver form a constituent35(Jespersen, 1940; Stowell, 1981; Moro, 1997).17(19) a. I consider [John a dangerous driverSC].b. I consider [that John is a dangerous driverCP].Rothstein (1995, 32) notes that (19a), with an embedded small clause complement,is truth-conditionally equivalent to (19b), with an embedded CP. This illustrates thesemantic connection between small clauses and copular sentences.Embedded small clauses in English do not allow an inverse word order, as canbe seen by comparing (20) and (21). Moro (1995, 112) takes this as evidence thatthere is a basic direction to predication in English: the subject precedes the predi-cate. Recall from the preceding discussion that this same, basic subject-predicatedirectionality is also evident in predicational copular clauses.(20) a. DP Predicate: I consider [John a dangerous driverSC].b. AP Predicate: I consider [John boringSC].c. DP Predicate: I consider [John the cause of the riotSC].d. DP Predicate: I consider [these the best pictures of MarySC].(21) a. DP Predicate: *I consider [a dangerous driver John SC].b. AP Predicate: *I consider [boring John SC].c. DP Predicate: *I consider [the cause of the riot John SC].d. DP Predicate: *I consider [the best pictures of Mary theseSC].Note that a non-copular particle as may optionally occur between the small clausesubject and predicate in (20), with no change in meaning.18 Moro (1995) and DenDikken (2006) claim that as is an optional spell-out of a Pred-head, whose functionis to ?link? the subject and predicate.A non-finite copula may also occur between the subject and predicate (22).1917There are alternative analyses of examples like (19a). It has been argued, for example, that Johnand a dangerous driver do not form a constituent, but are rather separate arguments of the main clauseverb consider (e.g. Williams (1983)), or that John is the argument of a complex predicate consistingof consider and a dangerous driver (Chomsky, 1975). I do not further discuss these theories of smallclauses, since they are dependent on small clauses being embedded structures. Okanagan smallclauses do not need to be embedded (cf. section 7.2).18Though some English speakers find (20b) ungrammatical with ?as?.19In this case, we no longer have a ?small clause? but instead a ?projection of Infl? (Rothstein,36For cases involving the copula, unlike those with as or no linking element, predi-cate inversion is licensed if a definite DP is in predicate position (23c-d). This issimilar to the pattern seen with non-embedded finite specificational copular clauses(cf. 7).(22) a. DP Predicate: I consider [John to be a dangerous driver IP].b. AP Predicate: I consider [John to be boring IP].c. DP Predicate: I consider [John to be the cause of the riot IP].d. DP Predicate: I consider [these to be the best pictures of MaryIP].(23) a. DP Predicate: *I consider [a dangerous driver to be John].b. AP Predicate: *I consider [boring to be John].c. DP Predicate: I consider [the cause of the riot to be John].d. DP Predicate: I consider [the best pictures of Mary to be these].2.2.2 Predicate RaisingStowell (1981), Pereltsvaig (2001) and others assume that the copula provides thenecessary structure (i.e. T(ense) head) for the subject to raise out of its base-generated initial position within the small clause (24a). Moro (1997) extends theanalysis to include DP-predicate raising as well (24b).(24) a. I consider [[Johni] to be [ti [the cause of the riot DP] SC] IP].(canonical ordering)b. I consider [[the cause of the rioti] to be [[John DP] ti SC] IP].(inverse ordering)Under this analysis, DPs are unique in their ability to raise to specifier of T (23),and (21c,d) are ungrammatical because there is no copula, and thus no landing sitefor a raised DP predicate.1995). Chomsky (1981) also states that small clauses differ from other clause types in that thepredicate is not linked to INFL. Thus, while a small clause consists of DP XP, for example, a copularclause XP will be linked to INFL, i.e. DP INFL XP. Functional heads which are claimed to playa major role in copular predication in other languages do not do so in Okanagan, and there is noevidence that the subject is linked to the predicate via an AGR(eement) node, at least for the non-verbal predications investigated in this dissertation.37Both symmetrical (Stowell, 1981; Pereltsvaig, 2001) and asymmetrical (Moro,1995; den Dikken, 2006) analyses of the base-generated structural relation betweenthe small clause subject and predicate exist. Under the asymmetrical account, thesmall clause is projected by a functional Pred-head (Figures 2.2-2.3). Under thesymmetrical account, there is no functional head intervening between subject andpredicate (Figures 2.4-2.5).20 Under either account, a tensed copula selects for asmall clause complement. This general picture derives both predicational and spec-ificational copular clauses from an underlying subject-initial small clause throughsyntactic raising of either the subject (yielding a predicational configuration) or thepredicate (yielding a specificational configuration).Figure 2.2: DP Subject Raising andthe Asymmetrical (Pred) Ac-countTPDPi (Sub j)JohnT?TisPredPti Pred?Pred\u000BDP(Pred)the cause of the riotFigure 2.3: DP Predicate Raisingand the Asymmetrical (Pred)AccountTPDPi (Pred)the cause of the riotT?TisPredPDP(Sub j)JohnPred?Pred\u000Bti20See den Dikken (2006, ch.3) for arguments against a bare, symmetric analysis of small clauses.38Figure 2.4: DP Subject Raising andthe Symmetrical (bare) Ac-countTPDPi (Sub j)JohnT?TisSCti DP(Pred)the cause of the riotFigure 2.5: DP Predicate Raisingand the Symmetrical (bare)AccountTPDPi (Pred)the cause of the riotT?TisSCDP(Sub j)JohntiEnglish small clauses without a copula may not normally occur in non-embeddedcontexts (25), presumably due to a tense-anchoring requirement.21(25) a. *John a dangerous driver.b. John is a dangerous driver.Okanagan does not allow predicate-raising (see chapter 7) in either direct predi-cations or DP-DP structures, as evidenced by the fact that no functional head mayintervene between an initial predicate and a final subject in a direct predication, anda non-subject DP can normally never precede a subject DP in a DP-DP structure.2.2.3 Specificationals as EquativesIn this section, I summarize portions of Heycock and Kroch (1999), who arguefor the existence of an equative head, and for a fundamental distinction betweenequative and predicative small clauses. They claim that the English copula is not21Although English does not normally allow non-embedded small clauses (ia,b), they can some-times occur in ?informal contexts? (Moro, 1995, 113) (ic).(i) a. *[John a dangerous driver SC].b. *[Lucy the boss SC].c. [John the cause of the riot SC]? I can?t believe it!d. *[The cause of the riot JohnSC]? I can?t believe it!Inversions of non-embedded small clauses are always ungrammatical (id), presumably because thepredicate must raise out of its base-generated position, but there is no landing site.39ambiguous (Partee, 1987), but is always semantically vacuous, and may select foreither type of small clause. They also claim that specificational sentences are notinverse predicational sentences, but are equatives with an additional pragmatic re-quirement that the initial DP represent old information. Their claims are directlyrelevant to my analysis of Okanagan: I claim that Okanagan equatives are projec-tions of a null head, and that equative and predicative small clauses are distinct.Their argument is based on the existence of several kinds of data which pose prob-lems for the inversion account, primarily involving specificational pseudoclefts. Ipresent some key aspects of their analysis below.First of all, there is a set of phenomena known in the literature on specifi-cational pseudocleft clauses as connectivity effects (Higgins, 1973; Heycock andKroch, 1999; Mikkelsen, 2005), so-called because a constituent in a higher struc-tural position behaves as if it were in a lower position with regards to standardstructural diagnostics such as binding.22 I briefly illustrate connectivity effectswith an example involving Condition A of binding theory, An anaphor must bebound within its governing category (cf. Chomsky (1981), data from Mikkelsen(2011)).Example (26a) shows a specificational pseudocleft, in which the antecedent R-expression Harvey does not c-command, yet appears to bind, the anaphor himself.Example (26b) is a non-copular sentences in which the antecedent R-expressionHarvey does c-command the anaphor himself and binds it, as expected underBinding Condition A. Example (26c) shows a topicalized phrase, in which theR-expression Harvey does not c-command and cannot bind the anaphor himself.(26) Principle Aa. What Harveyi did next was wash himselfi thoroughly.b. Harveyi washed himselfi thoroughly.22There are four types of connectivity effects. These are: (i) binding connectivity (ii) bound vari-able connectivity, (iii) negative polarity item (NPI) connectivity, and (iv) opacity connectivity. Thereare three broad approaches to explaining connectivity effects: The ellipsis approach (den Dikkenet al., 2000; Schlenker, 2003) is compatible with predicate raising analyses, while the logical form(Heycock and Kroch, 1999) and semantic approaches (Jacobson, 1994; Sharvit, 1999; Heller, 2002)do not require syntactic movement. Since Okanagan does not straightforwardly show connectivityeffects, a detailed explanation of these approaches goes beyond the scope of this thesis, though I referthe reader to Mikkelsen (2011) for a concise summary of the argumentation behind these approaches.40c. *Before Harveyi left, Miriam washed himselfi thoroughly.The issue here is that if the antecedent in specificational pseudocleft does not c-command an expression which it binds (26a), we expect these examples to patternlike non-pseudocleft data involving non-c-commanding antecedents (26c); how-ever the initial element behaves as if it did stand in a c-command relation (26b).Specificational sentences with non-pseudocleft subjects (27a) pattern with parallelsentences involving pseudocleft subjects (27b) in terms of connectivity effects, asshown for example with the following binding condition C data from Heycock andKroch (1999).(27) a. *Hisi claim was that Johni was innocent.b. *What hei claimed was that Johni was innocent.Heycock and Kroch (1999) take the existence of connectivity effects to be prob-lematic for an analysis of specificational pseudoclefts as inverted predications: ifsyntactic inversion were involved in cases like (26a), then the prediction is thatthey would not behave exactly like their simple sentence paraphrases (26b), butsimilarly to other cases in which movement has clearly occurred (26c).23Next, Heycock & Kroch view data such as (28) to be prima facie evidence fora separate class of equatives. In (28a) for example, there is no sense in which oneinstance of honest is ?more-predicative? than the other, and neither expression isreferential. In (28b) it seems clear that two referential attitudes are being equatedwith one another. Heycock and Kroch (1999) use these data in support of equationas a more general semantic phenomenon by which two expressions of the same23Despite having structures which might be argued to be equivalent to English inverse pseudo-clefts, connectivity cannot be tested in Okanagan because of several language-specific properties.These are as follows (cf. section 7.3.3):a. Specificational sentences are not permitted.b. There is no overt copula.c. 3rd person pronouns are normally null, and reflexivization is an operation on the predicate,not on an argument.d. Okanagan like other Salish languages (Davis, 2006, 2009) regularly violates condition C.e. Okanagan (and the rest of Salish) lacks WH-relative clauses, and so Okanagan has nothingcomparable to WH-pseudoclefts in English.41type are linked together.(28) a. Honest is honest.b. Your attitude towards Jones is my attitude towards Davies.A further argument that they cite in favor of a separate class of equatives comesfrom data pairs like (29a,b). For (29a), the final constituent honest is specifyingthe variable in the initial pseudocleft clause what John is. Assuming honest istype , this means that the pseudocleft clause must be type <,t> underan inversion analysis, where the predicate constituent is of a higher type than thesubject. But it is not necessarily the case that a pseudocleft clause must be ofthis type, since in (29b), what John is must be of type . If however (29a) isanalyzed as an equative sentence where both constituents are of type , and(29b) as a predicational sentence with a subject of type e, then the pseudo-cleftclause can always be of type .(29) a. What John is is honest.b. I am what John is.Next, consider that (30-32) are grammatical as equations between two expres-sions of type , according to Heycock and Kroch (1999). Examples (33a, 34a)are grammatical as predications, while (33b, 34b) are not since ?it is not possible totreat any constituent appearing in [subject] position as predicated of a postcopularargument.? (Heycock and Kroch, 1999, 380). Basically, an inversion analysis mustexplain why inversion is possible for (30b-32b), but not (33b, 34b).(30) a. Proud of his daughters is what he isb. What he is is proud of his daughters.(31) a. Honest is the one thing that I have always wanted a man to be.b. The one thing that I have always wanted a man to be is honest.(32) a. Honest is what I want a man to be.b. What I want a man to be is honest.42(33) a. John is the one thing that I have always wanted a man to be. (that is, he?shonest)b. *The one thing that I have always wanted a man to be is John.(34) a. John is what I want a man to be.b. *What I want a man to be is John.According to Heycock and Kroch (1999), the fact that (33b, 34b) are ungram-matical fits in with a more general picture that predicates cannot normally precedetheir subjects (35a-c). (35d) is grammatical precisely because the initial definiteDP is not a predicate, but is referential.(35) a. *A doctor is John.b. *Boss is Mary.c. *Proud of his daughters is John.d. The best candidate for the job is John.Given these empirical facts, Heycock and Kroch (1999, 382) argue for two separatetypes of small clauses, both involving the same semantically vacuous copula. Thereis no null predicational head for predicational small clauses, while there is a nullequative head for the equative cases. I advance a similar analysis for Okanagan inchapter 7.To conclude this section, Heycock and Kroch (1999) explain that there is apragmatic condition on specificational sentences (i.e. they have a fixed informationstructure (Prince, 1978)), such that the initial DP or pseudocleft clause must forma ?[back]ground? and the final DP must form a ?focus? (Vallduv?, 1992), essentiallya version of the structured meaning approach to focus (von Stechow, 1990; Krifka,1991). Okanagan DP-DP structures also show a fixed information structure, similarto inverse specificational pseudoclefts in English (e.g. 18a), where the focusedconstituent always precedes the non-focused constituent.2.2.4 Directly Referential versus Non-Rigid DPs:An Intensional AsymmetryThere have been alternative, semantic approaches towards explaining the informa-tion structural asymmetry in specificational sentences which trace the asymmetry43to intensionality rather than predicativity, most notably Romero (2005) and Co-morovski (2007). The general idea rests on the following distinction.Demonstratives (Kaplan, 1977) and proper names (Kripke, 1982) are directlyreferential; they denote entities, and their intensions are rigid individual concepts(i.e. they are constant functions, and denote the same individual in every world).As such, and assuming that type-shifting applies only as a last resort (Partee, 1987),demonstratives and proper names are best treated as expressions of type e. In con-trast, definite DPs denote non-rigid individual concepts (i.e. they may denote dif-ferent individuals in different worlds). In order to capture this non-rigidity, definiteDPs may be understood as optionally type-shifting to type . Romero (2005)and Comorovski (2007) argue for an intensional type analysis of the firstDP in a specificational copular clause (linearly speaking), and for a type e analysisof the second DP, with Romero?s claim resting on an analysis of the first DP as aconcealed question. Strictly speaking, specificationals are semantically asymmet-rical equatives under this analysis, with the equative head mapping an individual?sintension to its extension. Romero (2005) gives the following semantics for thespecificational (equative) copula:24(36) [[be]] = ?xe?y?ws.y(w) = xThe copula in (36) takes an extensional individual x (the second DP in a specifi-cational sentence) and an intensional individual y (the first DP) as arguments, andyields a proposition that is true in a world w if and only if y applied to w is identicalto x. This approach has the benefit of not positing multiple levels of LF, which isnecessary under Heycock and Kroch?s analysis of connectivity effects, but on theother hand, does not appear to reflect the fixed information structure of specifica-tionals.25I claim that a semantics similar to (36) comprises a part of the Okanagan equa-tive copula, with the exception that the arguments are reversed: the intensional DPforms the first argument of the copula (i.e. the second DP, linearly speaking), and24The underlining on the y argument indicates intensionality.25Neither does it account for connectivity effects, but since Okanagan does not show connectivityeffects, I do not concern myself further with this issue.44the extensional DP the second argument (i.e. the first DP) (cf. section 7.5).26,272.2.5 SummaryThis section has focused on theories of specificational copular sentences. Thereis a debate in the field over whether there is a derivational, syntactic relation-ship between semantically-asymmetrical predicational and specificational clauses(Moro, 1997; den Dikken, 2006), or whether specificationals are pragmatically-asymmetrical equative structures (Heycock and Kroch, 1999; Romero, 2005). Muchof the debate centers around specificational pseudocleft data, though there are alsoquestions as to how best to account for the fixed information structure of specifica-tional copular sentences.I now give a general discussion of my assumptions concerning focus and infor-mation structure, followed by a brief discussion of theories of clefts, which I argueto be equative in Okanagan.2.3 Focus and Information StructureThis section lays out aspects of theories of focus and information structure whichare crucial for my analysis of Okanagan equatives. The discussion is not intendedto be a comprehensive introduction to information theory or its application toOkanagan: this is an area for future work.In section 2.3.1, I briefly explain the alternatives approach to focus (Rooth,1985, 1992), and the syntactic realization of focus as F-marking (Jackendoff, 1972;Selkirk, 1995). In section 2.3.2, I discuss how a focused constituent aligns with theedge of a prosodic phrase in Thompson River Salish (Koch, 2008a), a NorthernInterior Salish language spoken immediately adjacent to the Okanagan languagearea. The relevance of this discussion becomes clear in chapter 7, where I claim26Similar to ?inverse? specificational pseudoclefts in English (den Dikken et al., 2000). See dis-cussion in section 2.1.2.27The base semantic type of Okanagan DPs which are headed by the determiner iP is of type, yet they are of type e when the world variable is existentially bound (cf. section 5.3). Thebase semantic type of Okanagan proper names is a type , and these lower to type e in argumentpositions and in equative contexts (either as a result of a null referential determiner or N-to-D raising(Longobardi, 1994)) (cf. section 4.6.2). Simple demonstratives are uniformly of type e, yet allowintensional readings when they are adjoined to an iP DP.45that focus is integral to the meaning of the equative copula in Okanagan, and thatan F-marked DP in an equative aligns to the left edge.2.3.1 Focus Alternatives and F-markingI now briefly introduce the alternatives-based approach (Rooth, 1985, 1992) whichI adopt for this dissertation. This approach assumes that expressions have twodifferent denotations: an ordinary semantic interpretation and a focus semanticinterpretation. In answer to a question such as (37), the constituent John in (38a) issyntactically marked as a focus by a feature ?F? (Jackendoff, 1972; Selkirk, 1995).(38b) shows the ordinary semantic meaning of the sentence, and (38c) shows thefocus semantic meaning under an alternatives representation. By focusing Johnin (38a), one invokes all of the contextually relevant alternatives to John (i.e. the?contrast set?), as represented in (38c). The focused constituent John in (38a) isphonologically realized with a prominent pitch accent in English.(37) Which one of these people passed the exam?(38) a. [John]F passed the exam.b. Meaning: PASSED THE EXAM(JOHN)c. Alternatives: {PASSED THE EXAM(x)|x ? De}= {PASSED THE EXAM(JOHN),PASSED THE EXAM(PETE),PASSED THE EXAM(SAM), ...}The set of alternatives forms a partially ordered scale of propositions, wherestronger propositions (i.e. those more likely to be true) are ranked higher. Alterna-tive propositions that are not asserted are generally ruled out by scalar implicature(Rooth, 1992). In the case of (38a), when the speaker asserts that John passed theexam (38b) with a pitch accent on John, all of the non-asserted alternatives to Johnpassed the exam (e.g. Pete passed the exam, Sam passed the exam, etc.) are thennormally ruled out by scalar implicature.2828The implicature can be cancelled. In answer to Which of these people passed the exam?, one cananswer [John]F passed the exam, and [Pete]F passed the exam too.46In English, the uniqueness/exhaustivity implicature associated with focus canbe strengthened in several ways, such as by the addition of a focus-sensitive exclu-sive particle like only (39a), or by means of a cleft (39b). In (39a), the addition ofonly entails that the non-asserted alternative propositions in (38c) are false. In cleftstructures like (39b), the exhaustivity may arguably be derived from the semanticsof a covert definite determiner (cf. Percus (1997), section 2.4 below).(39) a. Only John passed the exam (# and Pete did too).b. It is John who passed the exam (# and Pete did too).As I show in chapter 7, Okanagan equatives have a fixed information struc-ture, similar in some ways to English clefts. I claim that the null equative head inOkanagan lexically assigns the syntactic feature ?F? to its second (leftmost) argu-ment, which is interpretable as ?focus? at the interfaces (Jackendoff, 1972; Selkirk,1995). This is illustrated in (40a).(40) a. [ix?PDP]FDEM==[iPDETp@ptw?naxw DP].old.ladySHE is the old lady.b. *[iPDETp@ptw?naxw DP]old.lady==[ix?PDP]F .DEMThe old lady is HER.At present, there is no general algorithm for assigning F in contexts without anequative head, either by movement to a prosodic edge or by assignment of pitchaccent. However, when F is assigned by the equative head to its second argument,a prosodic-alignment constraint (which I discuss in the next section) assures focusoccurs leftmost. Together with selectional restrictions on the equative head, thesederive the absence of specificational sentences in Okanagan (40b). The initial F-marked constituent in an Okanagan equative may, but does not necessarily, receivean exhaustive interpretation (cf. section 7.4.1) due to a maximality implicatureassociated with the determiner iP in the second DP of an equative DP-DP struc-ture (cf. section 5.3.3). This essentially follows the analysis of Percus (1997) forEnglish clefts.47Note that while much of the literature on information structure makes a distinc-tion between two types of focus, contrastive and (new-)information focus (cf. forexample Szabolcsi (1981), Rochemont (1986), Kiss (1998), Selkirk (2007)), I setthis distinction aside, since it plays no part in the analysis of equatives given here.29Such a distinction will likely be relevant for a more comprehensive account of in-formation structure in Okanagan, but this work remains to be done. Here, I utilizeonly those aspects of the theory which are necessary to account for the informationstructure of equative sentences in Okanagan.2.3.2 Prosodic Alignment and Focused ConstituentsIn this section, I outline a prosodic-alignment theory of focus realization (Koch,2008a; F?ry, 2013) which generalizes the prosodic realization of focus to languageswhich lack a stress-focus correspondence (e.g. Thompson Salish). Previous workon English and other largely European languages had assumed the universality ofa stress-focus correspondence, given here in the form assumed by Reinhart (1995):(41) Stress-Focus Correspondence Principle:The focus of a clause is a(ny) constituent containing the main stress of theintonational phrase, as determined by the stress rule (Reinhart, 1995, 62).However, more recent cross-linguistic work, including work on Thompson RiverSalish (Koch, 2008a), has undermined the universality of the stress-focus corre-spondence principle. I briefly demonstrate how stress and focus correspond inEnglish, and summarize Koch?s claim that in Thompson, focus aligns to the edgeof a prosodic phrase (cf. F?ry (2013)).In English and other stress languages, focus aligns with stress, as a prosodichead. In a case involving default CP focus (e.g. 42, where the entire sentenceis a new-information), the subject is parsed into one prosodic phrase, and the VPinto a separate prosodic phrase (Chomsky, 1971; Jackendoff, 1972; Selkirk, 1995;Selkirk and Kratzer, 2007). Each prosodic phrase has a pitch accent (marked by?X?), and in English, the right-most prosodic head in a prosodic phrase bears a pitchaccent (hence the verb ?saw? does not bear a pitch accent). The intonational phrase29Koch (2008) similarly ignores the distinction between new information and contrastive focus inhis analysis of information structure in Thompson Salish.48carries the nuclear pitch accent, which by default in English, is right-headed. Thegeneralization is that by default, the rightmost lexical stress is prominent.(42) ( X )(X )( X )intonation-phrase (nuclear pitch-accent)prosodic-phrase (pitch accent)[J?hn s?w Mon?que FOC] Koch (2008a, 120, ex.4)In English clefts, however, the leftmost lexical stress is most prominent. Givenmaterial (indicated by G) is not parsed in a prosodic phrase (Selkirk and Kratzer,2007) during an initial step 1, but is parsed recursively into a prosodic phrase dur-ing step 2, under the assumption that all material must be parsed before prosodicphrases are parsed into an intonation phrase. Since the leftmost lexical stress is theonly pitch accent in the intonation phrase, it is also the most prominent. The ob-servation here is that through a process of destressing given material, nuclear pitchaccent can associate with a prosodic head which is not rightmost, and that in sen-tences which involve narrow focus (e.g. clefts), nuclear-pitch accent will associatewith the narrowly focused constituent.(43) ( X )( ( X ) )( X )intonation-phraseprosodic-phrase, step 2prosodic-phrase, step 1It was [Mon?que FOC] [that J?hn s?w G]. Koch (2008a, 120, ex.5)Koch (2008a) claims that for languages like Thompson River Salish, focusaligns with prosodic edges, rather than prosodic heads. Since cleft foci are not in-tonationally prominent in Thompson, listeners must rely on other means to recoverfocus, and as a general rule, the focused element occurs left-most in Thompson(more specifically, focus associates with the leftmost lexical item, excluding anyfunctional heads). This is shown to be the case for both nominal predicate con-structions (NPCs) and clefts. In Thompson, nuclear stress falls on the right-mostpitch accent, but focus is aligned to the left (44). Koch captures this generalizationwith (45).49(44) (X)(X)T?Pe.NEG(( X[qw?PFOC]water)??uPjust(eDETX )X )s-P?qweP-kt.NOM-drink-1PL.POSSintonation-phraseprosodic-phraseNo, we?ll just drink [water FOC]. (Thompson, Koch (2008a, 251, ex.15))(45) FOCUS LEFT: Align the left edge of the focus-marked p(rosodic)-phrasewith the left edge of an intonational phrase.In contrast to Thompson, which exhibits left-edge alignment in both NPCs (44) andother predicative contexts, as well as in clefts, Okanagan only shows such effectsfor clefts and other simple equatives (section 7.5.2). This implies that in Thomp-son, the feature F is assigned freely, whereas in Okanagan it is restricted to equativecontexts. The reasons for this difference are unclear, and await more detailed in-vestigation of information structure, and its prosodic reflexes, in Okanagan.302.4 CleftsThis section discusses several relevant syntactic and semantic aspects of theoriesof English clefts which directly inform my analysis of Okanagan clefts as equativestructures (chapter 8).31 An example of an English cleft, from Reeve (2007), isgiven in (46):(46) It was the snake that the mongoose caught.This sentence may be informally characterized as consisting of three parts (except-ing the copula): an initial clefting pronoun it; a DP in focus; and a residue CP(a.k.a. remnant), as represented in (47):30There has been little intonational work done on Okanagan, excepting Barthmaier (2004) whoconducts an acoustic study of several Okanagan narratives (A. Mattina and DeSautel 2002). Hismain finding is that intonation phrases do exist in Okanagan, and that they correspond to syntacticphrasing.31The relationship between ?clefts?-proper (e.g. 46) and ?pseudoclefts?, examples of which werediscussed in section 2.2.3, is still a matter of some contention. The two types of structures are notnecessarily derivationally related, although Percus (1997) does effectively derive a cleft from a th-pseudocleft (e.g. ?The one that the mongoose caught was the snake?).50(47) [It cle f t?pronoun] was [the snake DP? f ocus] [that the mongoose caught residue].First, I give a brief outline of the ?extraposition-from-subject? theory of cleftswhich I adopt in this dissertation, focusing on the versions proposed by Percus(1997) and Hedberg (2000), who treat the introductory clefting pronoun (i.e. it inEnglish) as forming an underlying constituent with the residue clause.32 Second, Idiscuss the semantics and pragmatics of English clefts, focusing on two properties:the presupposition of existence carried by the residue clause, and the presupposi-tion that the DP in focus position be interpreted exhaustively.2.4.1 ?Extraposition from Subject? Analyses of CleftsThere are two main versions of the extraposition from subject analysis of Englishclefts (Akmajian, 1970; Schachter, 1973; Emonds, 1976; Gundel, 1977; Wirth,1978; Percus, 1997), both of which are traceable to Jespersen (1927). For the firstversion (48), the cleft clause originates in the subject position of a WH-pseudocleft,then the CP is extraposed and it is inserted. For the second version (49), the cleftclause originates as part of a definite description in the subject position of a spec-ificational copular clause (a.k.a. ?th-pseudocleft?) (Percus, 1997). The CP is thenextraposed and the definite description remnant the one is spelled-out as it.(48) a. [What the mongoose caught CP] was [the snake DP f ocus]. (Base structure)b. was [the snake DP f ocus] [that the mongoose caught CP]. (Extraposition)c. It was [the snake DP f ocus] [that the mongoose caught CP]. (?It? insertion)(49) Percus (1997)a. [The \u000B [that you saw CPi]][is the deer]. (Base structure)b. [The \u000B ti][is the deer][that you saw i]. (Extraposition)32There are many other analyses of English clefts available, for example the so-called ?expletive?analysis, which has its roots in Jespersen (1937). Here, the initial it is neither a semantically inter-preted pronoun nor the head of a definite description. The focused DP is base generated in its surfaceposition, and the cleft clause is a complement of the focused DP (Chomsky, 1977; Halvorsen, 1978;Delahunty, 1982; Rochemont, 1986; Heggie, 1993; Kiss, 1998, 1999). Because I do not adopt any ofthese theories for Okanagan, I do not discuss them further. See Reeve (2007) for a concise summaryof several of the more major theoretical camps.51c. It is the deer that you saw. (Spellout of the as it)A treatment similar to Percus (1997) (50) is advanced by Hedberg (2000), dif-fering primarily in the sense that the definite determiner selects directly for a CP,rather than a null NP (49), and the CP is ?lowered? to adjoin to the focus DP.(50) Hedberg (2000)a. [The [that you saw i]][is the [deer NP]]. (Base structure)b. [The ti][is the [deer [that you saw i]NP]. (CP Lowering)c. It is the deer that you saw. (Spellout of the as it)Crucially, for both Percus (1997) and Hedberg (2000) cleft residues are discon-tinuous definite descriptions. Under these analyses, the semantics of clefts followsdirectly from the semantics of the definite determiner.2.4.2 Cleft Semantics: Exhaustivity Entailments and ExistencePresuppositionsEnglish clefts presuppose exhaustivity of the DP in focus position (Percus, 1997),and the cleft residue carries a presupposition of existence (Percus, 1997; Kiss,1998; Hedberg, 2000). Take again our cleft example (46), shown below as (51a),with paraphrases of the exhaustivity presupposition (51b) and existence presuppo-sition (51c).(51) a. It was the snake that the mongoose caught.b. Exhaustivity Presupposition: The mongoose caught only one thing.c. Existence Presupposition: The mongoose caught something.The fact that exhaustivity is presupposed comes from general agreement amongEnglish speakers that sentences like (52a,b) are unacceptable. If it were not apresupposition of the cleft in (51a) that the mongoose caught only one thing, thenit should be possible to assert that the mongoose also caught other individuals,but this is not the case. Exhaustivity appears to be a presupposition, rather than52entailment, since (51b) survives under negation, as illustrated by (52c).33(52) a. #It was the snake that the mongoose caught, and it was the rabbit that themongoose caught too.b. ?It was the snake that the mongoose caught, and the mongoose caught arabbit too.c. It wasn?t the snake that the mongoose caught.The presupposition of existence carried by the residue clause is clearly observableby the fact that in an out-of-the-blue context (53a), clefts are unacceptable in En-glish (53b), as well as by the fact that under negation (54), the presupposition thatthe mongoose caught something survives.(53) a. Context: The speaker walks into the room and tells the addressee what hesaw at the zoo today.b. #It was the snake that the mongoose caught.(54) It wasn?t the snake that the mongoose caught.Percus (1997, 339-340) formalizes the exhaustivity and existence presupposi-tions similarly to (55):(55) a. Exhaustivity PresuppositionIt is [?]FOC that has the property ?entails ?x[?(x)? x = ?] (only ? has the property ?)33Other data seem to indicate that exhaustivity is an entailment rather than a presupposition inEnglish clefts (Lisa Matthewson, p.c.). In (i) below (especially (ib)), any presupposition that themongoose caught only one thing does not survive negation:(i) a. It wasn?t the snake that the mongoose caught, it was the rabbits.b. It wasn?t the snake that the mongoose caught, it was a rabbit and a mouse.The presupposition account could nevertheless be salvaged for (51) by instead assuming that thepresupposition is that the mongoose caught only one maximal singular or plural individual: for (ia),the rabbits denotes the plural sum of all the contexually salient rabbits, and for (ib) a rabbit anda mouse denotes the sum of some pair consisting of a rabbit and a mouse. In any case, it is notcrucial for my analysis of Okanagan whether exhaustivity in English clefts is a presupposition or anentailment, so I retain Percus? original presupposition analysis.53b. Existence PresuppositionIn a cleft of the form It is [?]FOC that has the property ?,there is a presupposition that ?x?(x)(there exists some individual that has the property ?).By treating the residue clause as a discontinuous definite description (49-50),Percus and Hedberg are able to align the semantics of the definite determiner withthe semantics of clefts. In other words, the English determiner the is commonlyassumed to presuppose the existence of a referent, and presuppose the uniquenessand/or maximality of that referent (Heim, 2011). The maximality presuppositionof a definite DP in an specificational/equative environment, such as in Percus? basestructure (56, cf. 49a) will lead to an exhaustivity presupposition for the focusedDP.34(56) [The \u000B [that you saw CPi]][is the deer]. (Base Structure)2.4.3 SummaryFor English, the evidence in favor of any particular analysis of clefts is subtle. Forthe purposes of this thesis, I will argue that clefts in Okanagan (and DP-DP struc-tures), broadly support theories that (i) analyze the cleft residue as a discontinuousconstituent with the cleft pronoun, and (ii) align the semantics of clefts with thesemantics of determiners (Percus, 1997; Hedberg, 2000).Point (i) is supported by evidence that the Okanagan clefting demonstrative ix?Pforms an underlying constituent with the residue clause (8.5.2), and point (ii) bythe fact that Okanagan clefts lack any presupposition of existence or exhaustivity,though they do carry an exhaustivity implicature (8.6.2) which I claim is linked tothe maximality implicature of the determiner iP (5.3).The possibility of extending a Percus/Hedberg type analysis of clefts to Salishlanguages is not without precedent: Shank (2003) discusses the option in somedetail with regards to clefts in Northern Straits Salish, and Koch (2008a, 2009)34Whether the initial DP in (56) is a semantic predicate, or semantically referential (Heycock andKroch, 1999) (cf. section 2.2.3) does not affect the basic point that the focused DP will receive anexhaustive interpetation since as a predicate, a definite DP will denote a singleton set.54for Thompson River Salish. Both end up rejecting this analysis, due to the factthat residues appear to be bare CPs in these languages, rather than DPs, and assuch it is not straightforwardly possible to link the semantics of clefts with de-terminers.35 For Okanagan, residues may be analyzed as DPs (cf. section 8.4),and so a Percus-style analysis is applicable. More specifically, since Okanaganresidues may contain overt NP heads, Okanagan supports Percus (1997) over Hed-berg (2000) (compare 49a and 50a), who in principle allows for this possibility.2.5 Chapter SummaryThis chapter has focused on four major theoretical areas: the semantic distinc-tion between predication and equation and Higgins? (1973) taxonomy of copularclauses (2.1); the syntactic and semantic relationship between predicational andspecificational sentences and evidence for a separate class of equatives (2.2); focusand information structure (2.3); and syntactic and semantic theories of clefts (2.4).Each of these areas provide useful tools for understanding the Okanagan data, tobe discussed in the following chapters.35Though see discussion in chapter 9 where I suggest that for Salish languages with CP cleftresidues, the copula is the spell-out of a determiner which selects only for a CP (Hedberg, 2000).Under this analysis, all Salish clefts may potentially be analyzable as equative.55Chapter 3Background in OkanaganGrammarThis chapter presents some basic aspects of Okanagan grammar, including phonol-ogy, pronominal inflection, and brief notes on transitivity and the tense and as-pectual systems. Since this chapter is primarily meant as a terminological andparadigmatic reference tool for the reader, I limit the amount of data I give duringmy brief discussion on pronominal inflection, transitivity and aspect. I do howeverdedicate relatively more space in this chapter to discussing word order, since thisis particularly important for an understanding of subsequent chapters, and appearsto exhibit some dialectal differences.3.1 PhonologyThe following tables represent the consonant (Table 3.1) and vowel (Table 3.2)inventories of Okanagan. The phonemic symbols are written in a standard Ameri-canist orthography.56Table 3.1: Okanagan Consonant Phonemeslabialalveolaralveo-palatallateralpalatalvelarlabio-velaruvularlabio-uvularpharyngeallabio-pharyngealglottalstop/affricate p t c? k kw q qwejective ?p ?t ?c ?? ?k ?kw ?q ?qwfricative s ? x xw x? x?wresonant m n r l y G w Q Qw hglottalized ?m ?n ?r ?l ?y ?w ?Q ?Qw PTable 3.2: Okanagan Vowel Phonemesfront central backhigh i umid (@)low aConcerning vowels, schwa is not a full vowel (i.e. it cannot carry stress exceptfor in a few scattered loan words), and so I include it in parentheses. Also, surface[e] and [o] are possible in the context of a post-velar consonant, as documentedin A. Mattina (1973, 10-11). These are underlyingly /i/ and /u/, respectively. It isalso worth mentioning that the Okanagan orthography deviates from the standardAmericanist writing system in the following way: since there is no glottalized / ??c/phoneme in Okanagan, and neither is there an alveo-palatal unglottalized /c/, theOkanagan orthography uses ?c? for /c?/.For an in-depth discussion of general phonology and (morpho-)phonologicalprocesses in Okanagan, I direct the reader to Watkins (1970) (Northern Okana-gan) and A. Mattina (1973) (for Colville). Additionally, there are studies focus-57ing on pharyngeal movement (A. Mattina 1979), and sandhi effects and morpho-phonemics (H?bert (1978) and A. Mattina (2000)). For discussion of Upper Nicolaphonology, see Pattison (1978). In this dissertation, I add footnotes concerningspecific morpho-phonological processes when relevant to the discussion, but oth-erwise have little to say about the phonology.3.2 Inflection: Pronominal, Valency, and Tense-AspectHere I present some important aspects of Okanagan inflectional morphology, in-cluding pronominal paradigms, morphemes related to (in)transitivity, and a fewbrief notes on tense and aspect. The purpose here is to provide the reader withsome basic background in these areas, which will aid in comprehending the overallstructure of the examples I cite in this dissertation. Since all three of these areashave received attention in the literature, I will for the most part direct the interestedreader to other sources for more information.First, I will give a brief overview of the pronominal system (cf. A. Mattina(1982) and N. Mattina (1996b, 36) for more detailed descriptions). I give theparadigms in the following tables, followed by some discussion and data, withrelevant morphemes highlighted in bold type.Table 3.3: Intransitive ParadigmsParadigm 1 Paradigm 2ABSOLUTIVE POSSESSIVE1SG kn i(n)-2SG kw a(n)-3SG \u000B -s1PL kwu -tt2PL p -mp3PL -lx -slx58Table 3.4: Transitive ParadigmsParadigm 1 Paradigm 2ABSOLUTIVE POSSESSOR ACCUSATIVE ERGATIVEOBJECT SUBJECT OBJECT SUBJECT1SG kwu i(n)- kwu -(i)n2SG kw a(n)- -s,-m -(i)xw3SG \u000B -s -\u000B -(i)s1PL kwu -tt kwu ... -m -(i)m,-t2PL p -mp -?(ul)m -(i)p3PL \u000B -slx \u000B ... -lx -(i)slxTable 3.5: Independent PronounsINDEPENDENT1SG inc?2SG anw?3SG cni?c1PL mn?m?t@t2PL mn?m?t@mp3PL mn?m?ts@lxOkanagan may be characterized, roughly, as a ?split-ergative? language in termsof its pronominal system: there is a partial paradigmatic overlap between abso-lutive subjects in the intransitive paradigms (Paradigm 1, Table 3.3) and absolu-tive objects in nominalized possessor structures, which are syntactically transitive(Paradigm 1, Table 3.4). Thus the second person singular proclitic kw functions as asubject marker in intransitive contexts (1a), and as an object marker in syntacticallytransitive structures with possessor subjects (1b).59(1) a. kw2SG.ABSQaP-nc?t.laugh-REFLEXYou laughed.b. kw2SG.ABSi[n]-s- ?k?q?xw-@m.1SG.POSS-NOM-protect-MIDI am protecting you.Primarily accusative object suffixes are used in the ergative paradigm (2a)(Paradigm 2, Table 3.4), with 1st person objects being a notable exception (2b).(2) a. NormanNormanc@ ?mEPISp?l-st-@m-s.beat-CAUS-2SG.ACC-3SG.ERGNorman will punish you.b. kwu1SG.ABSc?n-[n]t-xwtell-DIR-2SG.ERG?aPCOMPnisleaveIvan.IvanYou told me when Ivan left.The distribution of pronominal morphology rests on a distinction between ?for-mal? (i.e. morphological) transitivity, semantic transitivity, and syntactic transitiv-ity. Constructions which use the possessor subject pronominal paradigm (e.g. 1b,cf. Paradigm 1, Table 3.4) are formally intransitive but semantically and syntac-tically transitive, while those with ergative morphology (e.g. 2, cf. Paradigm 2,Table 3.4) are formally, semantically, and syntactically transitive. Constructionswhich involve predicates that are lexical, underived intransitives (e.g. adjectivalpredicates as in 3a) are formally, semantically, and syntactically intransitive. Pred-icates which are formally intransitivized by reflexive morphology (e.g. 1a) or bythe middle suffix -@m (e.g. 3b) (and do not involve possessor subjects), or in-transitivizer -(aP)x (e.g. 3c) are both formally and syntactically intransitive, butsemantically transitive.11For formal intransitives like (3b,c), the predicates may select for oblique-marked quasi-objects,hence they are semantically transitive. I discuss these at length below for example 6 and in chapter4.60(3) a. kn1SG.ABSPilxwt.hungryI am hungry.b. kn1SG.ABSPaws-p?x?-@m.go-hunt-MIDI went hunting.c. kn1SG.ABSs-c- ?kwu?l-x.NOM-CUST-work-INTRI am working.Formal transitives take ergative subjects, and contain one of several transitiviz-ers, including -nt- ?directive? (4a), -st- ?causative? (4b), -?t- ?possessional applica-tive? (4c), and -x(i)t- ?benefactive applicative? (4d) (A. Mattina (1982) and N. Mat-tina (1996b)).2 The transitivizer morphemes in (4) are highlighted in bold type.(4) a. iPDET?kwu?l-nc?t-[t]nmake-REFLEX-INSTRc-n-qw ?n-mi[n]-nt-s.CUST-n-pity-MIN-DIR-3SG.ERGGod bless you. (said after one sneezes)Literally: The creator take pity on you.b. ?-xwuy-st-sreturn-go-CAUS-3SG.ERGiPDETtu ?m-s.mother-3SG.POSSShe took her mother homec. kwu1SG.ABSc-xwi ?c-?t-xwCUST-give-APPL-2SG.ERGiPDETlpot.cupPass me the cup.d. ?kwu?l-xt-nmake-BEN-1SG.ERGtOBLy?mx?waP.cedar.root.basketI made someone a cedar root basket.Syntactically transitive predicates (more specifically, those with possessor sub-2Okanagan also has transitivizers -t??t-, -n?nt-, -n??t- and -n?st- (A. Mattina (1982) and N. Mat-tina (1996b)).61jects) do not contain transitivizers (5), yet may select for subject and object ar-guments.3(5) i[n]-s- ?c? ?qw-@m1SG.POSS-NOM-point-MIDiPDETpus.catI am pointing at the cat.Formally intransitive predicates take absolutive subjects and are marked by -@m?middle? or active intransitivizer -x/-aPx morphology, and do not select for ob-jects, but may occur with an oblique-marked ?quasi-object? (N. Mattina (1993b)and Davis and Matthewson (2003)).(6) a. kn1SG.ABSkaPk?c-@mfind-MIDtOBLsp@pl?naP.rabbitI found a rabbit.b. kn1SG.ABSks-n-P?ys-aPxFUT-n-buy-INTRtOBLi-k?-k@w?p.1SG.POSS-U.POSS-horseI am going to buy a horse.Passive predicates, characterized as having a transitivizer plus the ?passive? suffix-@m, may also occur with an oblique-marked agent (7) though in these cases, an iPdeterminer often co-occurs with the oblique-marked nominal.(7) k?l-@nt-@mchase-DIR-PASSiPDETtOBLskmx?st.bearHe was chased by the bear.Independent pronouns are primarily used for emphatic purposes (Table 3.5), andnormally co-occur with and co-refer with a pronoun from one of the other sets.43N. Mattina (1996b, 39) notes that nominalized irrealis predicates (i.e. those prefixed by ks-)take possessor subjects in the singular, but ergative subjects in the plural. This reflects a historicalprocess whereby nominalized intransitive predicates are gradually being reanalyzed as transitives(Henry Davis, p.c.).4I have data showing that in contexts where an independent pronoun and a demonstrative arebeing equated, e.g. ix?P inc? ?That?s me?, co-referring absolutive morphology is not necessary.62(8) a. inc?1SG.INDEPkn1SG.ABSm@ ?q-?nk.full-stomachMe, I?m full.b. mn?m?t@mp2PL.INDEPtOBLsqilxw,native.peopletaPl?Pveryp2PL.ABSx?ast.goodYou people, you are good people.The Okanagan aspectual system is based on a set of morphemes which areprimarily prefixes and which attach to verbal stems. These include most notablyprospective/future ks- (9a), and customary/habitual c- (9a,b).5(9) a. wa ?yyesc-my-st-inCUST-know-CAUS-1SG.ERGiPDETsq@ltm?xwmaniPDETks-Paws-?@??t-m?xaPx.FUT-go-line.fish-INTRYes, I know the man who is going to go fishing.b. pintkalwaysc- ?kw??l-@mCUST-make-MIDtOBLy?mx?waP.cedar.bark.basketShe?s always basket-making.The transitivity and aspect systems interact in numerous ways to yield sen-tences with specific aspectual and temporal interpretations. I refer the reader toA. Mattina (1993a) and N. Mattina (1996b, section 2.1.1), who provide detaileddescriptions and analyses of the Okanagan aspect and transitivity systems, as doesH?bert (1982b), albeit within a different theoretical framework.For the purposes of this dissertation, it is important to note that there is a syn-tactic and semantic distinction in the nominal domain between full arguments andquasi-arguments (cf. chapter 4), and that this distinction correlates not only withdifferences in nominal morphosyntax, but also with differences in transitivity andaspect. Because I correlate full argumenthood with nominal morphosyntax, ratherthan directly with transitivity and aspect, I do not further address the transitivity5Additionally, ks- and c- may combine to form a perfect aspect, and c- may combine with thenominalizer s- to form sc-, yielding an imperfective aspect with formal intransitives, and a perfectiveaspect with nominalized possessor forms (A. Mattina 1993a).63and aspectual system except to establish its correlation with syntactic argument-hood in chapter 4.63.3 Clausal Word OrderThis section presents data on clausal word order for Okanagan, with some notes ondialectal differences between Upper Nicola and other dialects. The following tablegives a summary of permissible and non-permissible word orders in Okanagan. Idiscuss each of these word orders in turn.Table 3.6: Summary of Word Orders in Upper Nicola Okanaganword order X/* notesSV X unmarkedSVO X unmarkedSOV * ungrammaticalVS X unmarkedVO X unmarkedVSO X unmarkedVOS X unmarked (in non-ambiguous contexts)OV X marked, object topic/focus structureOSV (X) marked, object topic/focus structureOVS (X) marked, object topic/focus structureOkanagan, like other Salish languages, has been argued to be a fundamentallypredicate-initial language, although it has also been noted that word order is flexi-ble (N. Mattina 1996b).For simple intransitives with DP subject arguments, a VS ordering (10) or6I argue in that full argumenthood correlates with nominal morphosyntax, rather than transitivityand aspect, because some quantifiers are restricted to co-occuring with the determiner iP (e.g. yaQy?Q?all?). Since quantifiers are not part of the transitivity or aspectual systems, I claim that quantifierslike yaQy?Q ?all? can only select for a full DP argument. Thus, argumenthood is independent of thetransitivity and aspectual systems.64an SV-ordering (11) is equally acceptable (N. Mattina (1994), Baptiste (2001)).7Within running discourse, pre-verbal subjects are normally ?topical?, by which Imean an element that is discourse-old, informally speaking. In elicitation contextshowever, the two word orders are interchangeable.(10) a. xwaP-xw?stmany-walkiPDETtk?milxw.womanThe woman started walking.b. ??axwtdeadiPDETq?qxw@lxfishThe fish are dead.(11) a. iPDETtk?milxwwomanxwaP-xw?st.many-walkThe woman started walking.b. iPDETq?qxw@lxfish??axwt.deadThe fish are dead.For transitive sentences with an object DP as the single overt argument, bothVO (12) and OV (13) are acceptable orders. OV order may be used to signal atopical object,8,9 however an initial object is not necessarily topical: sentenceslike (13a) are judged felicitous in out-of-the-blue circumstances as well, hence theindefinite DP in the English translation.10(12) a. w?k-@nsee-[DIR]-1SG.ERGiPDETs?maP.white.personI saw a white person.7Not all Salish languages permit pre-verbal subjects, for example Northern Straits (Montler,1993) and Upper Lillooet (Davis, 1999b) do not.8See Gardiner (1993) for discussion of pre-verbal topical objects in Shuswap.9The iP determiner in (12b) and (13b) is underlyingly present but regularly reduces before 1stperson possessive prefix in- and 2nd person possessive prefix an- (A. Mattina 2000, 157)10Darnell (1995, 99) found in his textual study of Colville-Okanagan (A. Mattina 1985) that non-contrastive topics could not be pre-posed. Non-contrastive non-topics may, however, be pre-posed, afinding which seems to support out-of-the-blue uses of data like (13a).65b. n-P?ys-@nn-buy-[DIR]-1SG.ERG[iP][DET]an- ?q@ ?y-m?n.2SG.POSS-write-INSTRI bought your book.(13) a. iPDETs?maPwhite.personw?k-@n.see-[DIR]-1SG.ERGI saw a white person.b. [iP][DET]an- ?q@ ?y-m?n2SG.POSS-write-INSTRn-P?ys-@n.n-buy-[DIR]-1SG.ERGI bought your book.For transitive sentences with two overt nominal DP arguments, SVO is thepreferred and most common word order in elicitation contexts (14).11 In texts andconversations, however, transitive sentences involving two overt DPs are extremelyrare (A. Mattina 2001), since anaphoric DPs are normally null.(14) a. iPDETs?maPwhite.personwik-ssee-[DIR]-3SG.ERGiPDETx?xwt@m.little.girlThe white person saw the little girl.b. iPDETylm?xw@mchiefx?mink-swant-[DIR]-3SG.ERGiPDETautomobile.automobileThe chief wants the car.c. iPDETx?xwt@mlittle.girlwik-ssee-[DIR]-3SG.ERGix?P.DEMThe little girl saw that.d. JohnJohnwik-ssee-[DIR]-3SG.ERGMary.MaryJohn saw Mary.11H?bert (1982b, 47) analyzes SVO sentences as topicalization of a subject. N. Mattina (1994, 95)states that ?nominals in preverbal positions appear to have a focus semantics?, but it is unclear fromthese accounts what exactly is meant by a subject being ?topical? or in ?focus?.66e. haYNQiPDETs@nk? ?caPsq?x?aPhorsewik-ssee-[DIR]-3SG.ERGiPDETx?xwt@m?little.girlDid the horse see the girl?Verb-initial transitive sentences are also commonplace. In the Upper Nicoladialect, the first DP following the verb is nearly always interpreted as the subject,yielding a surface VSO ordering (15a-c) (Baptiste, 2001; H?bert, 1982a,b).12,13(15) a. wik-ssee-[DIR]-3SG.ERGiPDETx?xwt@mlittle.girliPDETsam?P.white.personThe little girl saw the white person.b. x?mink-slike-[DIR]-3SG.ERGiPDETsk@mx?stbeariPDETs?yaP.saskatoon.berryBears like saskatoon berries.c. wik-ssee-[DIR]-3SG.ERGiPDETx?xwt@mlittle.girlix?P.DEMThe little girl saw it.*It saw the little girl.d. n? ?qw-@m-ssteal-MID-[DIR]-3SG.ERGBenBeniPDETs@nk? ?caPsq?x?aP.horseBen stole the horse.VOS interpretations are possible in cases where the subject is animate and theobject inanimate (A. Mattina 2004), as in (16a,b), though these are not consistently12VSO is not consistently judged grammatical if the two post-predicative DPs are proper names(i). N. Mattina (1994, 96) finds these cases to be ungrammatical.(i) ?wik-ssee-[DIR]-3SG.ERGJohnJohnMaryMaryJohn saw MaryLL: John wiks Mary would be better. Doesn?t sound right.13The interpretive restriction in (15c) patterns opposite to what is found in Northern Interior andCentral Salish, where the One Nominal Interpretation constraint would force the determiner-headedDP to be interpreted as an object, and a single demonstrative as a subject (cf. Gerdts and Hukari(2004).67judged grammatical in the Upper Nicola dialect (16c), and are usually corrected toan SVO order.14(16) a. x?mink-slike-[DIR]-3SG.ERGiPDETs?yaPsaskatoon.berryiPDETsk@mx?st.bearBears like saskatoon berries.#Saskatoon berries like bears.b. cmayEPIS?tQap-nt-?sshoot-DIR-3SG.ERGiPDETs ??aPc?n@mdeerJohn.JohnMaybe John shot the deer.#Maybe the deer shot John.c. *x?mink-swant-[DIR]-3SG.ERGiPDETautomobileautomobileiPDETylm?xw@m.chiefThe chief wants the car.#The car wants the chief.For the Colville dialect, N. Mattina (1994) states that VSO and VOS are both ac-ceptable, so long as there is no ambiguity,15 though it seems clear for the UpperNicola dialect that VSO is strongly preferred.Word order in subordinate clauses follows the same pattern as that found inmain clauses, allowing for either subject-initial (17a) or verb-initial (17b-c) order-ing. The difference in translations between (17b) and (17c) exemplifies the prefer-ence for a VSO interpretation over VOS in contexts for which the animacy of thetwo arguments is equivalent.14This suggests that VOS interpretations of the examples in (15) should also be possible given asuitable context (for 15a at least).15See Davis (2005) for similar findings in Lillooet, though VOS is unmarked in the Upper dialect,while VSO is unmarked in the Lower dialect.68(17) a. kn1SG.ABSn-stilsn-thinkiPDET?qwQay-lqsblack-robecaP-nt-?shit-DIR-3SG.ERGiPDETylm?xw@m.chiefI think that the priest hit the chief.b. kn1SG.ABSn-stilsn-thinkcaP-nt-?shit-DIR-3SG.ERGiPDET?qwQay-lqsblack-robeiPDETylm?xw@m.chiefI think that the priest hit the chief.c. kn1SG.ABSn-stilsn-thinkcaP-nt-?shit-DIR-3SG.ERGiPDETylm?xw@mchiefiPDET?qwQay-lqs.black-robeI think that the chief hit the priest.For V-initial sentences involving phonologically heavy DP arguments, a strongpreference for an object-reading of the heavy DP surfaces (cf. Davis (2005) forequivalent data in Lillooet):16(18) a. ??aP ??aP-nt-?slook.for-DIR-3SG.ERGiPDETs?lxwaPbigiPDETpiqwhiteiPDETk@kw?pdogGertie.GertieVOS: Gertie was looking for the big white dog.*VSO: The big white dog was looking for Gertie.b. ??aP ??aP-nt-?slook.for-DIR-3SG.ERGGertieGertieiPDETs?lxwaPbigiPDETpiqwhiteiPDETk@kw?p.dog?VOS: The big white dog was looking for GertieVSO: Gertie was looking for the big white dog.For the Colville and Upper Nicola dialects, both VSO and VOS are possible;however VSO is strongly preferred in the Upper Nicola (and Lakes) dialect, whileVOS is preferred in Colville. This most likely represents a dialectal difference(Baptiste, 2001, 21), with more southerly dialects allowing freer post-predicativeordering of subject and object than the more northerly dialects.SOV is not a possible order (19). Baptiste (2001, 19) describes this as a restric-tion against more than one DP occurring pre-predicatively.16Though ideally, animacy should be controlled for in (18) by either making both referents humanor both non-human.69(19) a. *iPDETylm?xw@mchiefiPDET?qQway-lqsblack-robecaP-nt-?s.hit-DIR-3SG.ERG*The chief hit the priest.b. *JohnJohnMaryMarywik-s.see-[DIR]-3SG.ERG*John saw Mary.Mary saw John.c. *kn1SG.ABSn-stilsn-thinkJohnJohnMaryMarywik-s.see-[DIR]-3SG.ERGI think that John saw Mary.Nevertheless, in contexts involving an object with a special discourse status (e.g.focused or topicalized), a surface OSV ordering is possible, as shown in the subor-dinate clause of (20, cf. 17 above), and also illustrated in (21), where the frontedobject is modified by a relative clause.17(20) kn1SG.ABSn-stilsn-thinkiPDET?qwQay-lqsblack-robeiPDETylm?xw@mchiefcaP-nt-?s.hit-DIR-3SG.ERGI think that it was the priest that the chief hit.(21) iPDETsk@mx?stbeariPDET?tQap-nt-?sshoot-DIR-3SG.ERGJohnJohnkaPk?c-iP-s.find-MID-3SG.POSSThe bear hei shot, Johni found.OVS interpretations are not possible in unmarked contexts.18 Thus, the initialclause in (22) is normally only interpretable as John saw Mary, and not Mary17For (21), the middle suffix -m becomes -iP before a 3rd person possessive morpheme (A. Mattina1993a, 251).18N. Mattina (1994, 96) indicates that OVS is made possible by pairing an inanimate object withan animate subject (i), similarly to the VOS data given as (16), at least for 2 out of 4 of her speakers.(i) iPDETsqla ?wmoneywikw-shide-[DIR]-3SG.ERGin-t? ?m.1SG.POSS-motherMy mother hid the money. (N. Mattina 1994)It is unclear whether (i) was elicited with focus on iP sqla ?w ?the money?, however. Baptiste (2001)found data like (i) to be ungrammatical with the speakers she worked with, and I have found OVSsentences to be grammatical only in contexts involving object focus (24).70saw John.19 Comparing (23a) with (23b), we see that the latter is pragmaticallyinfelicitous since the pre-predicative DP is interpreted as a subject.(22) JohnJohnwik-ssee-[DIR]-3SG.ERGMary,MarylutNEGAlice.AliceJohn saw Mary, not Alice.(23) a. iPDETs- ?kw- ?kw?y-m-@ltNOM-IRED-small-m-childPi?-seat-[DIR]-3SG.ERGiPDETs-?tx-a?qNOM-sweet-fruittOBLspiPs ?c??t.yesterdayThe child ate the fruit yesterday.b. #iPDETs-?tx-a?qNOM-sweet-fruitPi?-seat-[DIR]-3SG.ERGiPDETs- ?kw- ?kw?y-m-@ltNOM-IRED-small-m-childtOBLspiPs ?c??t.yesterday#The fruit ate the child yesterday(!!)Nevertheless, OVS is possible in contexts involving corrective focus, indicated bybold type (24b).20 Clefting of the object is also a possibility in these contexts(24c).2119Similar facts obtain for Lower Lillooet, as documented in (Davis, 2007).20These examples illustrate stripping of SV in the second conjunct, which suggests that an objectcan move out of VP.21Clefting of a DP constituent is indicated by the pre-predicative determiner iP, and is discussedin some detail in chapters 7 and 8. It is an interesting fact that unmarked object fronting, as in (24b),can be used to signal corrective focus. Unmarked fronting can also signal a contrastive topic in somecases in Okanagan (cf. Gardiner (1993) for unmarked fronting and contrastive topics in Shuswap).For reasons of space, I do not discuss unmarked fronting in detail in this dissertation; however it isimportant to note that corrective focus is not limited only to clefting or equative environments, whichare discussed in chapters 7 and 8.71(24) a. Q:Q:ucDUBJohnJohnwik-ssee-[DIR]-3SG.ERGAlice?AliceQ: Did John see Alice?b. A:A:MaryMarywik-ssee-[DIR]-3SG.ERGJohn,JohnlutNEGAlice.AliceA: John saw Mary, not Alice.c. A:A:MaryMaryiPDETwik-ssee-[DIR]-3SG.ERGJohn,JohnlutNEGAlice.AliceA: John saw Mary, not Alice.Literally: It was Mary that John saw, not Alice.The answer in (24b) is also felicitous as a response to the question Did Alice seeJohn?, with corrective focus on Mary as the subject in this case. This is as expected,given that SVO word order is unmarked.In sum, it seems plausible that O-initial structures involving focus or topical-ization are derived by a leftward movement of a DP out of a base V-initial structurein Okanagan. While pre-predicative focus in Northern Interior Salish and Cen-tral Salish is overtly marked by A?-extraction morphology, such morphology hasbeen lost in Southern Interior Salish (Kroeber, 1999). It is therefore not possibleto tell from surface morphology whether O-initial structures are derived by A orA? movement. Regarding S-initial structures, it is less clear that these are derived,since initial subjects do not necessarily receive a focused interpretation, and the*SOV/OSV asymmetry suggests a different derivation for preverbal subjects ver-sus preverbal objects. Clearly, more work needs to be done on word order andconfigurationality in Okanagan.The next chapter consists of a closer inspection of the syntactic distributionof argument expressions in the language, specifically the distribution of the deter-miner iP and oblique marker t in their role of introducing core and oblique argu-ments.72Chapter 4The Structure and Distribution ofNP and DPThe overall goal of this chapter is to give the reader a broad overview of Okanagansyntax, with specific focus on the distribution and form of NPs and DPs within thesentence, and the internal structure of DPs.4.1 IntroductionIn this chapter, I first delimit the concept of ?noun? for Okanagan, and the syntacticcategories for which noun-hood is integral, specifically NP and DP. Both of thesecategories can be defined and distinguished in terms of their distributions. Section4.2 presents evidence that nouns form a non-derived, inherently predicative lexicalclass in Okanagan. DPs are a derived class, and may be formed by prefacing a(sometimes covert) NP with the determiner iP.1 DPs, unlike NPs, do not normallyfunction as syntactic predicates (section 4.3).2Section 4.4 focuses on the distribution of the determiner iP and the obliquemarker t, as the two primary markers of arguments in the language. Although thetwo particles themselves can co-occur, the data show that the grammatical envi-ronment will reliably predict whether a specific argument type will be introduced1Proper nouns are lexical NPs which may also be converted into DPs. See section 4.6.2.2Except in predications consisting of two DPs. This is the major focus of chapter 7.73by iP, t or both.3 iP usually marks a direct grammatical relation (i.e. subject orobject), whereas t marks an indirect, or oblique, grammatical relation (e.g. passiveagent, instrument, temporal adjunct, oblique argument) (N. Mattina 1996b, 45-50).In section 4.5, I present my syntactic analysis of Okanagan iP DPs, and thendiscuss the distribution of demonstrative DPs and proper name DPs in 4.6. I sum-marize and conclude this chapter in section 4.7.4.2 Distinguishing Nouns as a Syntactic CategoryOkanagan nouns generally denote ?persons, places, or things?. Some examples areillustrated in (1):(1) a. ?ti ?kwt ?lake?b. q?qxw@lx ?fish?c. x?xwt@m ?little girl?d. st@mt?maP ?grandmother?e. sp?x?m@n ?scraper, tool (Douglas Lake)?f. s ?t@x???q ?blueberries, sweet berries?g. s@nl ?km?n ?jail?Morphologically, all nouns consist minimally of a root, and may be simplex (e.g.1a), or may have an analyzable nominalizer prefix s- (e.g. 1d-f), contain a lo-cational circumfix s@n...min (e.g. 1g) or one of several instrumental suffixes like-m@n (e.g. 1e). In addition, nouns may be formed by synchronic or diachronicprocesses of reduplication (1b,c,d), and may also contain lexical suffixes like -a?q?round object? (e.g. 1f). All these morphological operations are derivational, andaffect neither the ability of the resulting word to function as a predicate, nor asthe complement to a determiner, given a larger syntactic context. For this reason,I largely ignore morphological differences between nouns.4 I refer the reader to3Broadly speaking, the determiner iP is used in both referential and non-referential contexts,as might be expected for a language which utilizes just one determiner. There is nevertheless atendency for Okanagan speakers to use formally intransitive verbs, and thus introduce nominals withthe oblique marker t, in non-referential contexts. I defer discussion of the semantics of iP and t untilchapter 5.4N. Mattina (1996b, 25) also notes that the internal structure of bases (e.g. the nouns in 1) are notrelevant to syntax.74N. Mattina (1996b) for a discussion of noun-related morphological derivations.5It has often been remarked that lexical items corresponding to nouns (2a), ad-jectives (2b), and verbs (2c) in English may all function as main clause predi-cates in Salish languages (Kinkade, 1983; Jelinek, 1998; Davis, 1999b), includingOkanagan (N. Mattina 1996b). Such data has been used to motivate claims thatSalish languages in fact lack lexical categorial distinctions, and so we must lookelsewhere for evidence.(2) a. [p@ptw?naxw NP]old.womaniPDETs@xw- ?maP ?m?yaP-m.OCC-teach-MIDThe teacher is an old woman.b. [p@x?p?x?t AP]smartiPDETs@xw- ?maP ?m?yaP-m.OCC-teach-MIDThe teacher is smart.c. [n-ya ?kw-m?(n)-nt-xw V P]n-cross.over-MIN-DIR-2SG.ERGiPDETt@mxw?laPxw.landYou crossed over the land.Syntactic evidence for distinguishing N, A, and V as lexical classes comes fromdata involving complex nominal predicates (CNPs) (Demirdache and Matthewson,1995; Davis et al., 1997; Koch, 2004). CNPs consist of a NP projection of a nomi-nal head which is attributively modified by either another NP, or an AP (cf. section6.2 for structural analysis). In (3a) below, the noun tk?milxw ?woman? is beingmodified by the adjective x?ast ?good?, and the entire modified complex is the mainclause predicate, taking the DP iP ylm?xw@mt@t ?our chief? as an argument. Cru-cially, the modifying constituent must precede the head noun (3b,4b), and be linkedto the head noun by the oblique marker t (3c,4c).65There are other morphological tests for noun-hood discussed in H?bert (1982a, 49): e.g. theresulting category of an element prefixed by s@xw- ?habitual agent?, or suffixed by -tn or -mn/-m?n?instrumental?, is a noun.6See chapter 6 for tests which help to distinguish attributive from relative clause modification,and discussion of an additional requirement that a modifying adjectival constituent be either anindividual-level predicate, or if not, be prefixed by stative/customary ac-.75(3) a. [x?astgoodtATTRtk?milxw CNP]womaniPDETylm?xw@m-t@t.chief-1PL.POSSOur chief is a good woman.b. *[tk?milxwwomantATTRx?ast CNP]goodiPDETylm?xw@m-t@t.chief-1PL.POSSOur chief is a good woman.c. *[x?astgoodtk?milxw CNP]womaniPDETylm?xw@m-t@t.chief-1PL.POSSOur chief is a good woman.(4) a. kn1SG.ABS[s?lxwaPbigtATTRsq@ltm?xw CNP].manI am a big man.b. *kn1SG.ABS[sq@ltm?xwmantATTRs?lxwaPCNP].bigI am a big man.c. *kn1SG.ABS[s?lxwaPbigsq@ltm?xw CNP].manI am a big man.An NP can also modify another NP. The linear order between attributive NP com-binations appears to be free in certain cases (5), while there are restrictions in othercases (6-7):(5) a. [s@n- ?maP ?m?yaP-t@nLOC-teach-INSTRtATTR?q@ ?y-m?n CNP].write-INSTRThat?s a school book.b. [ ?q@ ?y-m?nwrite-INSTRtATTRs@n- ?maP ?m?yaP-t@n CNP].LOC-teach-INSTRThat?s a school book.76(6) a. ix?PDEM[ ??@x?- ??x??pRED-growntATTR?qwQay-lqs CNP].black-robeThat?s an old-man priest.b. *ix?PDEM[ ?qwQay-lqsblack-robetATTR??@x?- ??x??p CNP].RED-grown?That?s a priest old-man.(7) a. ix?PDEM[sqilxwnative.persontATTRt@tw?t CNP].boyThat?s a native boy.b. *ix?PDEM[t@tw?tboytATTRsqilxw CNP].native.person?That?s a boy native.The ungrammaticality of (6b) and (7b) could be argued to stem from the fact that??@x? ??x??p ?old man? and sqilxw ?native person? are in fact adjectives, and can there-fore not occur in final position of a CNP, however unlike the adjectival modifiers in(3-4), lexical items such as sqilxw may occur in final position of a CNP when themodifier is clearly adjectival (8).(8) ix?PDEM[x?astgoodtATTRsqilxw CNP].native.personThose people are good Native people.The generalization therefore seems to be that some nouns (e.g. sqilxw ?native per-son?) may function as NP heads (8) or as NP modifiers (7a), while other nouns (e.g.t@tw?t ?boy?) may only function as NP heads (7). Adjectives (e.g. x?ast ?good?), bycontrast, can only ever function as modifiers within a CNP (3a,8), never as heads(3b).To summarize the data and generalizations so far: first, an attributive modifiermust precede the constituent it is modifying; second, an NP can function as eithera modifier, or a modifiee; third, an AP may not function as a modifiee (cf. 3b-4b).We thus have syntactic evidence for a categorial distinction between AP and NP,and by assumption, also A and N.77Verbs and adjectives may be distinguished by means of complex DPs, whichare argument expressions consisting of a CNP complement to an iP determiner.Examples are shown below in (9).7 While an AP may function as an attributivemodifier (9a-b), a VP cannot (9c-d):8(9) a. w?k-@nsee-[DIR]-1SG.ERGiPDET[ ??laldeadtATTRk@kw?p CNP].dogI saw a dead dog.b. w?k-@nsee-[DIR]-1SG.ERGiPDET[qwQaybluetATTRs ?wa?r? ?kx@n CNP].frogI saw a blue frog.c. *w?k-@nsee-[DIR]-1SG.ERGiPDET[?txwt?lxflying.aroundtATTRsk@kQ?kaPCNP].birdsI saw the flying birds.d. *w?k-@nsee-[DIR]-1SG.ERGaPDET[c-?@?tp-m@-nc?tCUST-jump-MIN-REFLEXtATTRx?xwt@m CNP].little.girlI saw the jumping girl.The generalization may be that adjectives like in (9a,b) do not project aspectualclausal structure, but that unergative verbs do (9c,d), and that aspectual structurecannot occur in syntactic positions reserved for attributive modifiers (Koch, 2006;Davis, 2011), but instead must assume the form of a relative clause.9In sum, the syntactic category NP in Okanagan can be defined distributionallyas that class of items which can be both attributively and clausally modified. Anin-depth discussion of attributive and relative clause modification may be found7Modifiers within a complex DP structure are subject to slightly less stringent conditions thanthose in predicative CNPs: stage-level, but non-eventive, modifiers are permitted (Lyon, 2010a;Davis, 2011). See also chapter 6.8The iP determiner becomes aP in certain contexts, including before c- ?customary/habitual? asin (9d, cf. 74c), ?- ?back, again? and k?- ?have? (A. Mattina 2000, 151).9Transitive predicates are also ungrammatical as attributive modifiers. See chapter 6.78in chapter 6, the purpose here being only to convince the reader that there is asyntactically distinguishable noun category in the language.4.3 NP versus DPThere is an important distinction to be made in this dissertation between NPs andDPs: NPs are predicates; noun complements of the determiner iP (i.e. iP DPs) arenot predicates, since the argument position of the noun is saturated by the deter-miner. This predicts that NPs and DPs should display different syntactic behaviour,and this is indeed the case.Given an appropriate context, a bare noun is interpretable as a predicate takinga null 3rd person pronoun as an argument (10a); a complete sentence in otherwords. The argument may be overtly realized as a demonstrative (10b), a propername (10c), or an iP DP (10d). A bare NP cannot function as an argument (10d).(10) a. s@xw-p?x?-@mOCC-hunt-MID\u000B.pro(He/she) is a hunter.b. s@xw-p?x?-@mOCC-hunt-MIDix?P.DEMHe/she is a hunter.c. s@xw-p?x?-@mOCC-hunt-MIDSpike.SpikeSpike is a hunter.d. s@xw-p?x?-@mOCC-hunt-MID*(iP)*(DET)??@x?- ??x??p.RED-grownThe old man is a hunter.An isolated iP DP (11), by contrast, is only interpretable as a fragment. Speakersnever translate isolated iP DPs as complete sentences, unlike isolated NPs (10a).(11) iPDETs@xw-p?x?-@mOCC-hunt-MID(*\u000B).(*pro)a/the hunter79Thus, the categorial distinction between NP and DP corresponds to a predicate/ar-gument distinction, and the determiner iP functions to convert a predicate into anargument expression (Longobardi, 1994; Chierchia, 1998). In other words, bareNPs are always predicates, but NP complements to an iP determiner constituteargument expressions.10It is important to note that D selects only for NP in Okanagan, and not forany other category. This is not immediately apparent, since nominal predicateslike p@ptw?naxw ?old woman? (12a), adjectival predicates such as px?p?x?t ?smart?(12b), and verbal predicates like nya ?kwm?ntxw ?you crossed over it? (12c) may allbe preceded by a determiner.(12) a. iPDET[p@ptw?naxw NP]old.womanthe old womanb. iPDET[p@x?p?x?t AP]smartthe (one who is) smartc. iPDET[n-ya ?kw-m?n-[n]t-xw V P]n-cross.over-MIN-DIR-2SG.ERGthe (thing that) you crossed overFor cases involving AP (12b) and VP (12c), there is evidence for a null NPhead (Davis, 2011), and that these cases involve ?headless? relative clauses (Kroe-ber, 1997). In other words, the AP and VP are modifying a null NP head, asschematized in (13) for (12c). The particular analysis given as (13) is justified insome detail during my discussion of relative clauses in section 6.3.3.(13) [iP [\u000B j [[iP [\u000B NP j] DPi] [nya ?kwm?ntxw ti V P] CP] NP] DP]In (13), an initial determiner iP selects for a null NP (subscript j) which is modifiedby a relative clause CP containing a DP which has moved from a post-verbal posi-10H?bert (1982a, 35) states that ?it is only nominal arguments, and not predicates, which aremarked with a ?determiner??. While true that the data in (12) are all arguments, their status as argu-ments is the result of the determiner iP, and not the lexical category of the determiner?s complement.80tion within the clause to the left-edge of the relative clause CP (cf. (Davis, 2004,2010a) for Lillooet, and Kroeber (1997, 1999) and Koch (2006) for Thompson).The second, clause-introducing determiner deletes due to a filter on sequences ofidentical determiners, as discussed in chapter 6 (cf. Davis (2010a) for Lillooet).The head NP may also be overt (14, ?the land you crossed over?), in which case aniP determiner surfaces before the head as well as the clause.(14) [iP [t@mxw?laPxw j [[iP [\u000B NP j] DPi] [nya ?kwm?ntxw ti V P] CP] NP] DP]Clear evidence for the existence of relative clauses in Okanagan comes fromdata involving long range extraction. In (15), there is a gap following the finalintransitive predicate xwuy ?go?. The determiner iP which precedes the transitivepredicate wiks ?she saw her? forms a DP constituent with a null NP, and this DPhas raised from the gap site, in a manner analogous to the structure represented in(14).11(15) kn-x?t-@nhelp-BEN-1SG.ERGiPDETtk?milxwwoman?kl- ?klaxwRED-eveningiPDETwik-ssee-[DIR]-3SG.ERGSarahSarah?aPCOMPxwuy.goI helped the woman who Sarah saw leave last night.Only NPs may head relative clauses (16a). (16b) shows that an adjective can-not function as a relative clause head, and is only marginally acceptable under aninterpretation where the adjective is modifying a null NP head (Demirdache andMatthewson, 1995; Davis et al., 1997; Davis, 2011).12,13 A verb also cannot func-tion as the head of a relative clause (16c).11I have been unable to elicit headless examples of long range extraction, although these are pos-sible in Lillooet (Davis, 2010a, 12, ex.22).12These facts are different than those documented for Straits Salish in Montler (1993), where it isshown that adjectives may occur in these positions.13The determiner iP lowers to aP before the customary prefix c-, as illustrated in (16a,b) (A. Mat-tina 2000).81(16) a. c-my-st-inCUST-know-CAUS-1SG.ERGix?PDEMiPDET[tk?milxw NP]womanaPDETc- ?cu ?m-qs-[s]t-s.CUST-suck-nose-CAUS-3SG.ERGI know the lady that he kissed.b. #c-my-st-inCUST-know-CAUS-1SG.ERGix?PDEMiPDET[x@x??saPt AP]beautifulaPDETc- ?cu ?m-qs-[s]t-s.CUST-suck-nose-CAUS-3SG.ERG#I know the beautiful he kissed.SM: You didn?t say what, a pretty something was kissed?c. *c-my-st-inCUST-know-CAUS-1SG.ERGiPDET[s-c-Pitx V P]NOM-CUST-sleepiPDET?cu ?m-qs-[s]t-s.suck-nose-CAUS-3SG.ERG*I know the sleeping he kissed.Given that headless relatives exist in Okanagan, and that non-NP categoriesmay not head a relative clause (16), the most economical theory is one where Donly selects for NPs, and that apparent cases of direct selection (cf. 12b,c) actuallyinvolve modification of a null NP.To conclude, this section has argued for the following points:a. There is a syntactic distinction between NPs and DPs: NPs are predicativeexpressions, while DPs are not.b. Determiners do not select for categories other than NP.4.4 The Distribution of the Determiner and ObliqueMarkerIn this section, I focus on the syntactic distribution of the determiner, obliquemarker, and other morphemes which associate with nouns in Okanagan. Syntactic82arguments which are not proper names or demonstratives are obligatorily markedby either the determiner iP, the oblique marker t, or both. Locative adjuncts areintroduced by one of several locative particles which are in complementary distri-bution with t.4.4.1 Subject ArgumentsThe determiner iP must introduce a non-proper noun or non-demonstrative subjectargument. Oblique marked nominals are categorically banned in subject positionsfor both transitive (17a,b) as well as morphologically intransitive (18a,b) and lexi-cally intransitive (18c) predicates.14(17) a. iP/*tDET/*OBLsq@ltm?xwmanwik-ssee-[DIR]-3SG.ERGiPDETtk?milxw.womanThe man saw the woman.b. iP/*tDET/*OBLk@kw?pdogtal?Preallyx?mink-slike-[DIR]-3SG.ERGiPDETs- ?cim.NOM-boneThe dog really likes the bone.(18) a. iP/*tDET/*OBLsxw-l ?k-amOCC-bound-MIDcmayEPISc-kic-x.CISL-arrive-INTRA policeman might come.b. iP/*tDET/*OBLsqilxwnative.peopleac-t?rq-?m.CUST-kick-MIDThe native people are dancing.c. iP/*tDET/*OBLtk?milxwwomanPayx?wt.tiredThe woman is tired.14Although N. Mattina (1996b, 41) has noticed that ergative subjects allow optional oblique-marking in the Okanagan Valley dialect, and Kroeber (1999) makes the same observation for otherlanguages of the Southern Interior, speakers of the Upper Nicola dialect do not use oblique mark-ing on ergative subjects, instead relying on word order to disambiguate a DP?s grammatical status.Ergative subjects in H?bert?s Upper Nicola corpus are not marked as oblique.834.4.2 Core Objects vs. Quasi-ObjectsThe distribution of iP versus t in their roles of introducing objects is syntacti-cally predictable (N. Mattina 1996b, 45),15 as illustrated by (19) and (20) below.The determiner iP introduces objects of formally transitive predicates, as in (19a).Oblique arguments (a.k.a quasi-objects) of morphologically intransitive predicates(20a) will always be introduced by the oblique marker.16(19) a. ?kw??l-@nmake-[DIR]-1SG.ERGiPDETy?mx?waP.cedar.bark.basketI made the basket.b. * ?kw??l-@nmake-[DIR]-1SG.ERGy?mx?waP.cedar.bark.basketI made the basket.c. * ?kw??l-@nmake-[DIR]-1SG.ERGtOBLy?mx?waP.cedar.bark.basketI made the basket.d. * ?kw??l-@nmake-[DIR]-1SG.ERGiPDETtOBLy?mx?waP.cedar.bark.basketI made the basket.(20) a. kn1SG.ABSs-c- ?kw??l-@mNOM-CUST-make-MIDtOBLlat?p.tableI?m making a table.b. *kn1SG.ABSs-c- ?kw??l-@mNOM-CUST-make-MIDlat?p.tableI?m making a table.15N. Mattina (1996b, 46) notes for the Okanagan Valley dialect that ?case marking is not a com-pletely reliable means of identifying the grammatical relation of an NP?. It does however seem to be amore reliable means in the Upper Nicola dialect since (i) oblique quasi-objects cannot be introducedby iP, and (ii) ergative subjects cannot be marked oblique by t, unless they are passives (cf. 31b, forexample).16N. Mattina (1996b, 46) gives data showing that iP and t may co-occur in introducing a quasi-object. This may represent a dialect variation, since the Upper Nicola speakers I have worked withdo not allow this.84c. *kn1SG.ABSs-c- ?kw??l-@mNOM-CUST-make-MIDiPDETlat?p.tableI?m making a table.d. *kn1SG.ABSs-c- ?kw??l-@mNOM-CUST-make-MIDiPDETtOBLlat?p.tableI?m making a table.Bare nominals (excluding proper names) are ungrammatical in non-predicativepositions (19b,20b).17 The oblique marker t may not introduce the object of aformally transitive predicate (19c), and the determiner iP may not introduce theoblique argument of a morphologically intransitive predicate (20c). Finally, iP andt together cannot mark an absolutive argument (19d) or oblique argument (20d).Oblique arguments of intransitive predicates are quasi-objects18, meaning thatthey are semantically entailed by the predicate, but not registered by agreementmorphology (N. Mattina 1996b, 45). In section 5.3.2, I analyze quasi-objects as se-mantically incorporated nouns (Van Geenhoven, 1998). When there is no obliqueargument, then absolutive-subject intransitive predicates inflected with middle suf-fix -m (21) or intransitivizers -(m?x)aPx/-x (22) may indicate an activity in progress;however in actuality, many of these predicates are infelicitous without objects (e.g.23):(21) a. kn1SG.ABS?p ?yq-am.cook-MIDI?m baking.b. kn1SG.ABSx?m?nk-@m.want-MIDI want some.17A. Mattina (1973, 112) discusses an indefinite/definite contrast between bare nominal comple-ments (e.g. w?k@n sqilxw ?I saw some people?) and DP complements (e.g. w?k@n iP sqilxw ?I sawthe/those people?). Indeed, this is possible in other languages of the Southern Interior (cf. chapter9), but my own research with the Upper Nicola dialect suggests that bare nominal complements arealways ungrammatical in this dialect, perhaps under influence from Thompson.18This term comes from Davis and Matthewson (2003). N. Mattina (1996b, 42) refers to these asgeneric objects.85(22) a. kn1SG.ABSs-c- ?kwu?l-x.NOM-CUST-make-INTRI?m working.b. kn1SG.ABSs-c- ?p ?yq-mix.NOM-CUST-cook-INTRI?m cooking.(23) a. #kn1SG.ABSw?k-@m.see-MIDI?m seeing. (Consultant: You have to say what you see.)b. kn1SG.ABSw?k-@msee-MIDtOBLsp@pl?naP.rabbitI saw a rabbit.Morphologically intransitive constructions (21-23) do not permit anaphoric refer-ence to a previously introduced discourse referent (cf. chapter 5), but for transitiveconstructions, even in cases where a DP is not overt, a null pronoun is present (24)which takes a discourse-salient overt DP, or else a contextually salient referent, asan antecedent (cf. Davis and Matthewson (2003) for Lillooet, Gerdts and Hukari(2003) for Halkomelem).(24) a. n-??pt-@m-@nn-forget-MIN[?]-[DIR]-1SG.ERG[\u000BDP].I forgot it.b. n? ?k-@nt-xwcut-DIR-2SG.ERG[\u000BDP].You cut it.Now consider the following pair (25a,b) which on the surface seem quite sim-ilar, both involving an unergative predicate ks ??xw?paPx ?x will win?, but whichactually denote two different propositions. If the nominal s@nk? ?caPsq?x?aP ?horse?is introduced by the determiner iP, it is interpretable only as the subject (25a). If itis introduced by the oblique marker t, it is interpretable only as an oblique, quasi-86object (25b).(25) a. ks- ??xw?p-aPxFUT-win-INCEPTiPDETs@nk? ?caPsq?x?aP.horseThe/(That) horse is going to win.*He?s going to win the horse.b. ks- ??xw?p-aPxFUT-win-INCEPTtOBLs@nk? ?caPsq?x?aP.horseHe?s going to win a horse.*The/(that) horse is going to win.It is also worth noting that morphologically similar predicates may show differentselectional properties. Consider ks ?kw??l@?laPx ?will be born? (26) and ks ?kw??laPx?will make? (27). The former is unaccusative, and the latter is unergative, as ev-idenced by the distribution of iP and t.19 Thus, ?will be born? may only take anexperiencer subject DP as an argument (26), since it is not semantically transitive,whereas ?will make? may take an iP DP as a subject argument (27c), but not as anobject (27b).(26) a. ks- ?kw??l-@?l-aPxFUT-make-FRED-INCEPTiPDETs- ?kw- ?kw?y-m-@lt.NOM-IRED-small-m-childThe baby?s gonna be born.b. *ks- ?kw??l-@?l-aPxFUT-make-FRED-INCEPTtOBLs- ?kw- ?kw?y-m-@lt.NOM-IRED-small-m-childThe baby?s gonna be born.c. *iPDETtk?milxwwomanks- ?kw??l-@?l-aPxFUT-make-FRED-INCEPTiPDETs- ?kw- ?kw?y-m-@lt.NOM-IRED-small-m-child*The woman will borned the child.19See Davis (1997) for arguments that Salish roots are uniformly associated with a single internalargument, hence unaccusative. Though his arguments presumably apply to Okanagan as well, I usethe terms ?unaccusative? and ?unergative? descriptively to distinguish intransitive predicates whichtake experiencer DP arguments from those which take agentive DP arguments, without making anydeeper syntactic claims concerning unaccusativity in Okanagan.87(27) a. ks- ?kw??l-aPxFUT-make-INCEPTtOBLpwmin.drumHe?s gonna make a drum.b. *ks- ?kw??l-aPxFUT-make-INCEPTiPDETpwmin.drumHe?s gonna make a drum.c. iPDETsq@ltm?xwmanks- ?kw??l-aPxFUT-make-INCEPTtOBLpwmin.drumThe man will make a drum.4.4.3 The Syntactic Status of iP and tThe absence of bare nominal arguments in Okanagan suggests that iP is necessaryfor converting a predicate nominal into an argument (28-30). This is claimed to bea core property of the D position (Longobardi, 1994). In other words, bare NPs cannever be arguments in Okanagan, even in generic contexts (30).20(28) a. *[p@ptw?naxw NP]old.woman[s@xw- ?maP ?m?yaP-m NP].OCC-teach-MIDThe teacher is an old woman.b. [p@ptw?naxw NP]old.woman[iPDETs@xw- ?maP ?m?yaP-m DP].OCC-teach-MIDThe teacher is an old woman.(29) a. *[w?k-@n V P]see-[DIR]-1SG.ERG[sqilxw NP].native.peopleI saw native people.b. [w?k-@n V P]see-[DIR]-1SG.ERG[iPDETsqilxw DP].native.peopleI saw native people.20See section 5.2.8 for a semantic analysis of generic interpretations of Okanagan iP DPs.88(30) a. *[sk@mx?st NP]bear[x?mink-s V P]like-[DIR]-3SG.ERG[iPDETs?yaPDP].saskatoon.berryBears like saskatoon berries.b. [iPDETsk@mx?st DP]bear[x?mink-s V P]like-[DIR]-3SG.ERG[iPDETs?yaPDP].saskatoon.berryBears like saskatoon berries.The oblique marker t and the preposition-like locative particles l, tl, ?kl may co-occur with the determiner iP in certain contexts, as shown for example in (31)and (32) below. The oblique marker t co-occurs with the determiner iP in specificgrammatical environments: i.e. when marking the agent of a passive (31a,b)21 oran instrument (31c). The locative particles in (32) are in complementary distri-bution with the oblique marker t, suggesting that they occur in the same syntacticposition.22,23(31) a. iPDETylm?xw@mchiefiPDETk@w?p-shorse-3SG.POSSxw?y-?t-@mgo-APPL-PASS[iPDETtOBLsqwsiP-s].son-3SG.POSSThe chief?s horse was taken by his son.b. k?l-nt-@mchase-DIR-PASS[iPDETtOBLsk@mx?st].bearHe was chased by the bear.c. ?tQap-nt-?sshoot-DIR-3SG.ERG[iPDETtOBLs-wlwlm-ink].NOM-iron-weaponHe shot it with a gun.21I use the term ?passive? as a purely descriptive term. See N. Mattina (1996b, 40-41) for argu-ments that these constructions may not be syntactically intransitive.22In Northern Interior Salish languages, equivalents of the Okanagan locative particles l, tl, ?kl andthe oblique marker t always precede the determiners. This makes them straightforwardly analyzableas prepositions (P) which select for DP complements. In Southern Interior Salish, however, theseparticles always follow determiners (Kroeber, 1999).23A. Mattina (1973) refers to these locative markers as follows: l ?locational? meaning ?point oftime or place at which...? (p. 116); tl ?ablative? indicating ?motion from? (p. 119); and ?kl ?allative?meaning ?motion to/into? (p.120).89(32) a. miy@s-?tiqw@lqwmore-tall[iPDETtlLOC?c@c?p-s].little.sister-3SG.POSSShe is taller than her little sister.b. Pak[?]-s-x?w?s@mHAVE-NOM-soap.berries[iPDET?klLOC?ti ?kwt].lakeThere are some soap-berries next to the lake.c. x?wayqnpileiPDETsmi ?kwtsnow[iPDETlLOCn- ?km-qn-i?xw].n- ?km-head-houseThe snow piled on the roof.Data like (31) are strong evidence that the oblique marker t is not a determiner. Inchapter 5, I discuss a semantic restriction on specific readings of oblique argumentsof formally intransitive predicates, and conclude that t is semantically vacuous, andthat quasi-objects are semantically incorporated.244.4.4 Arguments of Lexical IntransitivesAdjectival and nominal predicates, although lexically intransitive, are not markedas such by overt morphology.25 Like other predicates, these predicates also selectfor iP DP subject arguments (33a), and may not take a bare nominal as a subject(33b). iP and t may not co-occur in this context (33c).(33) a. ?t?qw@lqwtalliPDETx?xwt@mlittle.girlThe little girl is tall.b. *?t?qw@lqwtallx?xwt@mlittle.girlThe little girl is tall.24I claim that intransitivizers (e.g. -@m) encode semantic incorporation in Okanagan (Van Geen-hoven, 1998). See section 5.3.2.25N. Mattina (1996b) refers to these as simple intransitives.90c. *?t?qw@lqwtalliPDETtOBLx?xwt@mlittle.girlThe little girl is tall.At first glance, (34a) below seems to show that adjectives may also take subjectarguments introduced by the oblique marker t, but this is not the case. (34a) isan attributively-modified CNP (see section 4.2), with a null pronominal argument(Davis et al., 1997). The fact that a demonstrative can occur as an argument ofthe CNP (34b) is evidence for a null pronoun in (34a) (as well as the fact thatthese are often used in contexts in which a referent can be pointed out). (34c) isungrammatical because the iP DP and demonstrative are construed as two separatearguments, yet the predicate is intransitive.26(34) a. [?t?qw@lqwtalltATTRx?xwt@m CNP]girl[\u000B DP]That?s a tall girl.b. [?t?qw@lqwtalltATTRx?xwt@m CNP]girlix?P.DEMThat?s a tall girl.c. *?t?qw@lqwtalliPDETx?xwt@mgirlix?P.DEMThat?s a tall girl.Similar data is shown below in (35). The CNP in (35a) can select a nominaliP DP as an argument. (35b) is ungrammatical because the adjectival predicatex?ast ?good? is saturated by the first DP iP sq@ltm?xw ?the man?, and the second DPiP ylm?xw@m ?the chief? cannot function as an adjunct.27 If we made the parallel26The demonstrative in (34c) cannot form a constitutent with the iP DP to its left, and so (34c)is also ungrammatical under an interpretation equivalent to ?That girl is tall?. See section 4.6.1 fordiscussion of demonstrative-associated DP arguments.27(35b) should theoretically be interpretable as a relative clause, i.e. ?The man who is a chief isgood?, and data in chapter 6, involving main clause transitive predicates, support this as a possibility.In other words, there is evidence that NPs project covert clausal structure in certain cases. (35b) maybe ungrammatical because there is a dispreference for interpreting modifying NPs as clausal in thecontext of a main-clause intransitive.91assumption that the oblique-marked nominal t sq@ltm?xw in (35a) was an argumentof x?ast ?good?, we would incorrectly predict that this sentence too should be un-grammatical.(35) a. [x?astgoodtATTRsq@ltm?xw CNP]maniPDETylm?xw@m.chiefThe chief is a good man.b. *x?astgoodiPDETsq@ltm?xwmaniPDETylm?xw@m.chiefThe chief is a good man.Simple nominal predicates (36) also fall under the classification of ?lexical intran-sitive?, and their selectional restrictions are the same as the adjectives in (33).(36) a. ?qwQay-lqsblack-robeiPDET??@x?- ??x??p.RED-grownThe old man is a priest.b. * ?qwQaylqsblack-robe??@-x? ??x??p.RED-grownThe old man is a priest.c. * ?qwQay-lqsblack-robeiPDETtOBL??@x?- ??x??p.RED-grownThe old man is a priest.4.4.5 DitransitivesThemes of benefactive (a.k.a. ?dative? or ?transitive?) applicative sentences, whichare characterized by the -xt- morpheme (Gerdts and Kiyosawa, 2010), are intro-duced by the oblique marker (37a,b) (N. Mattina (1996b), A. Mattina (2001),Barthmaier (2002)). The determiner is not grammatical in this position.28 Agent28Although cf. N. Mattina (1993b) for data showing that iP may sometimes co-occur with t whenintroducing an applicative theme. This would make them similar to locative adjuncts, which oftenbut not always have co-occurring overt determiners. Upper Nicola speakers do not allow iP in this92and goal arguments are both introduced by iP (37c).(37) a. ?kwu?l-xt-nmake-BEN-1SG.ERGiPDETylm?xw@mchief*iP/t*DET/OBLy?mx?waPcedar.bark.basketI made the chief a basket.b. c-P?kw-xt-m-nCISL-bring-BEN-2SG.ABS-1SG.ERG*iP/t*DET/OBLPaP?saP.eggI brought you an egg.c. iPDETtk?milxwwomanxwi ?c-xt-sgive-BEN-3SG.ERGiPDETsq@ltm?xwmantOBLy?mx?waP.cedar.bark.basketThe woman gave the man a basket.For possessional (a.k.a. ditransitive) applicatives, characterized by the -?t- mor-pheme, the theme must be introduced by an iP determiner, and not an obliquemarker (N. Mattina 1996b, 47).(38) a. n- ?c? ?w-?t-@n-l@xn-wash-APPL-1SG.ERG-3PL.ABSiP/*tDET/*OBLlasy@?t-s@lx.dish-3PL.POSSI washed their dishes.b. kwu1SG.ABSc-xwi ?c-?t-xwCUST-give-APPL-2SG.ERGiP/*tDET/*OBLlpot.cupPass me the cup.Barthmaier (2002, 4-5) states that ?consistently in texts we find -xt- predicatesselected when a speaker chooses to focus on the recipient... predicates with -?t-allow speakers to include the patient [i.e. theme] in the core, in addition to therecipient, to signify its worthiness of attention.?syntactic context.934.4.6 Possessor IntransitivesMain clause predicates can sometimes be inflected with possessor subject mor-phology, rather than absolutive or ergative morphology (cf. section 3.2).29 Theyare formally intransitive, lacking any transitivizer, but are semantically and syn-tactically transitive, and introduce their patient arguments with an iP determiner.30Examples of main clause possessor intransitives are given below in (39).(39) a. i-ks-Paws-p@qw?l@x-@m1SG.POSS-FUT-go-visit-MIDiPDETsq@ltm?xwmaniPDETmutlive?klLOC?ti ?kwt.lakeI?m going to visit the man that lives by the lake.b. i-ks-x?l?t-@m1SG.POSS-FUT-invite-MIDiPDETs-?l@x?-?l?x?tNOM-IRED-friendu?CONJnixwalsoi(n)-?q?qcaP.1SG.POSS-older brotherI am going to invite my friends, and my older brother will too.c. in-x?m?nk1SG.POSS-wantiPDETn? ?k-m@n.cut-INSTRI want the knife.In certain contexts, oblique arguments and iP DP patients are semantically in-distinguishable, though note that subject agreement morphology on the main pred-icate must vary appropriately (40a,b).(40) a. kn1SG.ABSx?m?nk-@mwant-MIDtOBLsp@pl?naPrabbitiPDET?k@- ?k?pIRED-softiPDETs?piP-s.skin-3SG.POSSI want a rabbit?s soft fur.29N. Mattina (1996b, 56, section 2.2.1.1) distinguishes between possessor morphology, whichattaches only to nouns, and genitive morphology which attaches only to verbs. The two paradigmsare identical in form, as shown in Tables 3.3 and 3.4, and so I label them all as POSS.30N. Mattina (1996b, 39) shows that for a subset of possessor intransitive predicates, namely thoseinflected for future/irrealis by the prefix ks-, a transitivizer -nt- is present for cases with plural subjectsbut absent with singular subjects.94b. in-x?m?nk1SG.POSS-wantiPDETsp@pl?naPrabbitiPDET?k@- ?k?pIRED-softiPDETs?piP-s.skin-3SG.POSSI want a rabbit?s soft fur.4.4.7 The Oblique Marker in Other EnvironmentsNon-locative adverbs are often introduced by the oblique marker t. These are un-grammatical both with co-occuring determiners, as well as a determiner in lieu ofthe oblique marker (41):(41) a. (*iP)(*DET)tOBLspiPs ?c??tyesterdaykiPCOMPni ?k-scut-[DIR]-3SG.ERGiPDETs ?p? ?c@n.ropeIt was yesterday that he cut the rope.b. kn1SG.ABSPaws-p?x?-@mgo-hunt-MID(*iP)(*DET)tOBLspiPs ?c??t.yesterdayI went hunting yesterday.c. cmayEPISx?astgoodi-ks-c-P?tx1SG.POSS-FUT-CUST-sleepQapn?Pnow(*iP)(*DET)tOBL?klaxw.eveningMaybe I will sleep well tonight.d. (*iP)(*DET)tOBLs-Pistk,NOM-wintertaPl?Pverykn1SG.ABS?kw??l-@m.work-MIDLast winter, I worked a lot.Unlike oblique arguments of intransitive predicates, the ungrammaticality of the iPdeterminer here cannot be attributed to selectional restrictions on the main pred-icate, since the adjuncts illustrated in (41) are more on par syntactically with thelocative adjuncts which do allow iP determiners to co-occur with a locative marker(cf. 32).954.4.8 This ?Predictability? does not hold across SalishThe syntactic predictability of oblique marking in Okanagan contrasts sharply withthe facts in related languages like Lillooet (Northern Interior Salish), where deter-miner choice does not automatically co-vary with the transitivity of the main pred-icate. (42) shows that in Lillooet, unlike Okanagan (43), different determiners maybe used in the same syntactic context.31(42) a. x???t-mi ?n-aswant-MIN.TR-3SG.ERGtiDETxw? ?k-t@n-a.cut-INSTR-EXISHe wants the/a knife. (Lillooet, Henry Davis, p.c.)b. x???t-mi ?n-aswant-MIN.TR-3SG.ERGkuDETxw? ?k-t@n.cut-INSTRHe wants a knife. (Lillooet, Henry Davis, p.c.)(43) a. *ix?PDEMx?m?nk-@mwant-MIDiPDETn? ?k-m@n.cut-INSTRHe wants the knife.b. ix?PDEMx?m?nk-@mwant-MIDtOBLn? ?k-m@n.cut-INSTRHe wants a knife.c. (ix?P)DEMx?mink-swant-[DIR]-3SG.ERGiPDETn? ?k-m@n.cut-INSTRHe likes the knife.d. *(ix?P)DEMx?mink-swant-[DIR]-3SG.ERGtOBLn? ?k-m@n.cut-INSTRHe likes that knife.31Both assertion-of-existence ti...a and non-assertion-of-existence ku may be used in the samesyntactic context, but only if the context is intensional (Matthewson, 1998). Squamish (a.k.a.Skwxw?7mesh) patterns with Lillooet in allowing both deictic and non-deictic determiners in thesame syntactic context (Gillon, 2006).96e. x?mink-swant-[DIR]-3SG.ERGix?PDEMtOBLn? ?k-m@n.cut-INSTRHe likes that knife.The apparent availability of both iP and t for Okanagan (43c,d) is not an exceptionto this rule, since these two examples involve different syntactic structures. For(43c), the demonstrative denotes the 3rd person subject and the iP DP is the objectargument of the transitive predicate. For (43d), the obligatory demonstrative mustdenote the knife, and forms a discontinuous DP constituent with the final oblique-marked nominal t n? ?km@n. The demonstrative and oblique-marked nominal trans-parently form a constituent in (43e).32 Evidence for this analysis comes from thefact that without the initial demonstrative in (43d), the sentence is ungrammatical,while the demonstrative in (43c) is optional, given an anaphoric subject. (43c-43e)therefore have DP object arguments.4.4.9 SummaryThe main factors determining whether a nominal is introduced by the determineriP, the oblique marker t, or both, are the selectional properties of the main predi-cate.33 The somewhat simplified picture is that iP introduces subjects of both tran-sitives and intransitives, and transitive objects, while t introduces passive agents,intransitive quasi-objects, and other temporal adjuncts (N. Mattina 1996b). Theco-occurrence of iP and t is only possible in a restricted set of grammatical con-texts (cf. 32). The basic distribution of the determiner iP and the oblique markert across the major grammatical and thematic relations is shown below in Table 1,with cross-referencing to relevant examples.34It is important to keep in mind that the oblique marker t appears to have (at32The exact difference between a demonstrative-associated oblique NP and a demonstrative-associated iP DP is elusive, but becomes important in the chapter 8. I assume that the demonstrativein (43d) is undergoing proclisis, similar to the enclisis process documented for Lillooet demonstra-tives (Davis, 2010c).33N. Mattina (2002, 20) makes the same point for Moses-Columbian, stating that ?determinerchoice is dictated by the clause head?.34The determiner iP is represented in parentheses in table 4.1 since it is not present for propername passive agents, or locative adjuncts. It seems to be optional in some cases for instrumentaladjuncts for reasons I cannot yet determine.97Table 4.1: Distribution of iP and t across Grammatical Categories1 Subjects iP - ex. 17-182 Core objects iP - ex. 19-203 Quasi-objects - t ex. 19-204 Benefactive Applicative Theme - t ex. 375 Possessional Applicative Theme iP - ex. 386 Possessor Patients iP - ex. 397 Passive Agents (iP) t ex. 31a,b8 Instrumental adjuncts (iP) t ex. 31c9 Temporal adjuncts - t ex. 4110 Locative adjuncts (iP) { ?kl, l, tl} ex. 32least) two functions: (i) it case-marks a nominal as an oblique argument (e.g. 20a);(ii) it links a nominal head to an attributive modifier (e.g. 3a).I now discuss the internal syntactic structure of Okanagan DPs.4.5 Internal Structure of DP(and other Nominal Projections)In this section I weigh evidence for three possible structural analyses of OkanaganDPs, and associated super-structure. The major challenge here is to account forthe determiner-oblique/locative ordering characteristic of Southern Interior Salish.The three hypotheses are as follows:a. The KP hypothesis holds that oblique and locative markers are case-marking(K) heads which are base generated internal to DP, i.e. in their surface po-sition.b. The PP hypothesis holds that oblique and locative markers are prepo-sitional heads, and undergo a surface-level prosodic inversion (Halpern,1995) with D.c. The headless relative hypothesis holds that the determiner forms a con-98stituent with a null NP, and that this constituent is an underlying argumentof a PP or KP predicate.Deciding which of these possibilities accurately refects the syntactic structureof Okanagan DPs is a complicated issue. I discuss some of the problems below,before settling on an analysis which posits (a) for certain Okanagan nominal pro-jections, and (b) for others.4.5.1 iP Occurs in D PositionAll three hypotheses depend on an important point: I claim that iP belongs to thefunctional category D. This claim is supported by the following points.First of all, iP is probably historically cognate with the referential determinerGe in Shuswap (Northern Interior Salish) (Henry Davis, p.c. 2012), which has beenanalyzed as a determiner that occurs in D position (Gardiner, 1996).Secondly, the semantic behavior of iP resembles more closely the so-called?strong? determiners of better-studied languages like English and Italian (cf. forexample Zamparelli (1995)) than any other element in Okanagan, and also includesas a subset of its interpretive possibilities the wide-scope readings characteristic ofassertion-of-existence DPs in Lillooet (Matthewson, 1998, 2001) (see chapter 5),both of which support the conclusion that if Okanagan has a D determiner at all, itmust be iP.Thirdly, as data in this chapter have shown, iP is necessary for converting apredicate nominal into an argument (Longobardi, 1994).4.5.2 Three HypothesesUnder the simplest analysis, iP is a D-head which selects for an NP complement(Figure 4.1). But recall that oblique and locative markers occur between the deter-miner and its NP complement, which may be evidence for a DP-internal functionalprojection for Okanagan (and other languages of the Southern Interior), a ?Casephrase? of sorts (Kroeber, 1986).35 I refer to this hypothesis as the case phrase KP35Bittner and Hale (1996) posit a case phrase (KP) as the nominal equivalent of CP in the verbaldomain. They assume that K selects a DP for an argument, similar to a preposition, rather than theother way around, which must be the case for Okanagan under this analysis.99hypothesis (Figure 4.2).Figure 4.1: Okanagan iP DPDPDiPNPNylm?xw@m?the chief?Figure 4.2: The KP hypothesisDPDiPKPKtNPNylm?xw@m?by the chief?The oblique marked DP structure in Figure 4.2 represents a passive agent orinstrumental adjunct under the KP analysis. If we remove the D shell in Figure4.2, what remains is a KP, which is the category of an oblique marked quasi-objectof a formally intransitive predicate. Analyzing quasi-objects as structurally lesscomplex than DP is consistent with their non-referential semantics (Gillon, 2009b).A theory-internal problem concerning case-assignment arises from this analysis,however: Normally, a case-marker will assign case to the head of its complementphrase, but in Figure 4.2, case-assignment must occur in an ?upwards? fashion.For cases where a locative marker occurs to the right of a determiner (cf. Figure4.2), an alternative analysis may be motivated: a late-stage, prosodic inversion ofthe two particles (Halpern, 1995).36 This second hypothesis assumes a PP structure(PP hypothesis), as in Figure 4.3, and has the benefit of resolving the problem ofcase-assignment which the KP hypothesis faces.36It is important to note that the proto-Salish ordering was almost certainly preposition-initial(Kroeber, 1999), which implies that the Southern Interior innovated the modern surface order. Su-perficially similar observations may be made for the ordering of absolutive pronominal pro-cliticswith respect to specific complementizers. E.g. the 1st singular absolutive pro-clitic kn occurs be-fore the temporal complementizer ?(aP), but after other complementizers like kwaP ?because? or mi?future.? Whether or not this case can also be explained by a late-stage inversion is unclear.100Figure 4.3: The PP Hypothesis and Prosodic Inversion: ?towards the chief?[ ?kl [iP ylm?xw@m DP] PP] =? [iP ?kl-[ylm?xw@m DP]PP]LOC DET chief DET LOC chiefThe prosodic inversion represented in Figure 4.3 assumes that the oblique andlocative particles in Okanagan are syntactically Ps, but that Okanagan Ps are pro-clitics which require a prosodic word as a host, and that nouns, but not iP deter-miners, are prosodic words. A preposition like ?kl will therefore move to the rightof iP in order to attach to an NP host.37Support for the PP hypothesis comes from the fact that these locative struc-tures may function syntactically as predicates (44) (Kroeber, 1999, 61), and thatsemantically they denote properties of individuals.38 In (44a) for example, iP ?kls@n ?kQ?wm@n ?at the church? may be analyzed as denoting the set of individualsthat stand in a particular spatial relation to the church, and in (44b), iP l n ?kmqni?xwmay be analyzed as denoting the set of individuals ?on the roof?, which serves torestrict the main clause predicate x?wayqn ?to pile?.(44) a. JohnJohn(iP)(DET)?klLOCs@n- ?kQ?w-m@n.LOC-pray-INSTRJohn is at the church.b. x?wayqnpileiPDETsmi ?kwtsnowiPDETlLOCn- ?km-qn-i?xw.n- ?km-head-houseThe snow piled on the roof.37This would then technically be a case of ?host-splitting?, where iP is separating the prepositionfrom its host, and assumes that /iP NP/ does not form a prosodic word.38Kroeber (1999, 62) notes that PPs can be predicates in only some Salish languages, and thattheir distribution is different than that of nominal and verbal predicates. In Okanagan, too, PPs arenot always acceptable as main clause predicates: examples like (44a) below are not consistentlyjudged grammatical, regardless of the presence or absence of a determiner. It remains unclear whythis should be the case, or why there should be speaker variation with regards to the acceptabilityof locative phrase predicates. Under Baker?s (2003) analysis, Ps are fundamentally functional ratherthan lexical categories, but ?intrude? into the lexical category domain in some languages. The Salishpattern would certainly be consistent with his view. In any case, locative phrases are marginallyacceptable as predicates in Okanagan.101Note that the determiner in locative phrases is not always present (cf. 44a). Assuch, it is worthwhile considering whether or not the apparent optionality of thedeterminer is syntactically significant: i.e. the possibility that with the determiner,a locative phrase is a DP, while without, it is a PP where P selects directly for an NP.This brings us to the third hypothesis, which saves the interpretation of the locativephrase under a KP analysis, and potentially, the KP analysis itself. The analysisinvolves treating locative phrases with overt determiners as headless relative clauseDPs (cf. section 4.3 above, and chapter 6). This possibility is represented below asFigure (4.4).Figure 4.4: The Headless Relative HypothesisDPDiPNPNPN\u000BCPSpecDPiDiPNPpro jC?C\u000BPredPDPtiPred?Pred\u000BKPK?klNPNs@n ?kQ?wm@n.?the (one who is) at the church?In Figure 4.4, the initial iP determiner is introducing a null NP, which is itselfmodified by a non-verbal predicational relative clause. A double-determiner filter(Davis, 2010a) ensures that only one of the determiners is actually pronounced.The headless relative analysis potentially explains (44a) under an equative inter-102pretation (i.e. John is the one who is at the church),39 but does not explain (44b),since the locative phrase in this case cannot stand in an equative relation with anyother constituent. Thus, the headless relative hypothesis cannot account for all oc-currences of locative phrases with overt determiners, and it is simpler to assume thePP hypothesis for locative phrases, regardless of whether the determiner is present.One potential drawback of the PP hypothesis is as follows: if we assume thatthe case-marker t always occurs in P position, and that P always selects for aDP complement, then there must be a null determiner introducing quasi-objects.However, there is little evidence for a null determiner, and in fact, there is cross-linguistic evidence from Lillooet Salish against positing a null determiner in thecontext of a quasi-object. The semantic interpretations of Okanagan quasi-objectsindicate that if there were a null determiner present, it would be semantically vac-uous, similar to the Lillooet non-assertion-of-existence determiner kwu (Matthew-son, 1998) (cf. section 5.3.2). But Lillooet kwu DPs can occur in core argumentpositions under the scope of a transitive intensional verb (45a), while Okanaganquasi-objects cannot (45b).40(45) a. x?? ??-mi ?n-aswant-MIN.TR-3SG.ERG[kwuDETxw? ?kt@nDP].knifeHe wants a knife. (Lillooet, Henry Davis, p.c.)b. *x?mink-swant-[DIR]-3SG.ERGtOBLn? ?km@n.knifeHe wants a knife. (Okanagan)If there is a requirement in Salish that only DPs may occur in core argument posi-tions, then the discrepancy between (45a) and (45b) is at once explained by assum-ing that there is no null determiner in (45b).39If this is the case, the prediction is that with the determiner, (44) should carry an implicature ofexhaustivity, such that John is the only one who is at the church. See chapter 7.40Okanagan iP DPs subsume all the interpretations of determinerless quasi-objects, given an ap-propriate context, whereas the interpretations allowed by Lillooet non-assertion of existence deter-miner kwu and the assertion of existence determiner ti...a are mutually exclusive.1034.5.3 AnalysisI will assume the PP hypothesis for passive agents and locative obliques, withprosodic inversion of the preposition and determiner. This approach best explainsthe predicative semantics of locative-marked DPs, the selectional restrictions onpassivized predicates, and fits nicely with the description of a semantically mean-ingful locative particle t meaning ?source? (A. Mattina 1973): t is a preposition inthese cases. For core iP DP arguments, I do not assume that there is a null-casemarking preposition selecting for the DP. For locative obliques without overt deter-miners, I do not assume that there is a null determiner, but rather that P can selectdirectly for NP.For quasi-objects of intransitive predicates, I assume the KP analysis. Thismeans that quasi-objects are structurally less complex than full DPs. The obliquemarker is a K-head for these cases.41 This analysis is consistent with a noun-incorporation analysis of quasi-objects, which I present in chapter 5.The structures I assume are as follows:Figure 4.5: Core ar-gumentDPDiPNPNylm?xw@m?the chief?Figure 4.6: Locativeadjunct with DPPP\u000BiDPDiPNPN?kli-ylm?xw@m?towards the chief?Figure 4.7: Locativeadjunct w/o DPPP?klNPNylm?xw@m?towards the chief?41In Moses-Columbian, though not in Okanagan, absolutive DPs are optionally introduced by aparticle wa (Willett, 2003). As in Okanagan however, locative markers occur to the right of determin-ers. If locative markers in Moses-Columbian undergo prosodic inversion with determiners (the PPhypothesis), but the particle wa does not (the KP hypothesis), Moses-Columbian may be argued tohave two separate case-marking positions, which lends indirect support to the argument I am makingfor Okanagan, namely, separating P from K.104Figure 4.8: Passive agent or instru-mental adjunctPPP\u000BiDPDiPNPNti-ylm?xw@m?by the chief?Figure 4.9: Quasi-objectKPKtNPNq?qxw@lx?some fish?In sum, the implication is that while locative markers are always prepositions,the oblique marker t is only a preposition when it marks a passive agent or in-strument (i.e. contexts in which an iP determiner co-occurs). When it introducesa quasi-object, it is a K-head (i.e. contexts in which an iP determiner may notco-occur).Proper names support this view of DP structure, since they occur in all envi-ronments in which an iP DP may occur and, like iP DPs, are marked oblique whenthey are passive agents (46a), yet proper names occur in none of the environmentsin which an oblique quasi-object may occur (46b). This makes sense if propernames can be prepositional objects, like other DPs, but cannot be non-referentialquasi-objects.42(46) a. TinaTinaw?k-@nt-@msee-DIR-PASStOBLIvan.IvanTina was seen by Ivan.b. *kn1SG.ABSw?k-@msee-MIDtOBLJohn.John*I saw a John.42Okanagan prepositions can select for either NP (44a, without the determiner) or DP (46a).Proper names can be predicates in certain contexts (section 4.6.2), which raises the question as towhy (46b) is ungrammatical. I suggest that while semantic incorporation requires that the incor-porated constituent be of type (cf. section 5.3.2), proper names are normally understood asbeing directly referential, or if they are predicates, as denoting singleton sets, which conflicts with arequirement that the incorporated noun be non-specific.105I now move on to a more detailed discussion of two other types of OkanaganDPs which are important to this dissertation: demonstratives and proper names.4.6 Other DPs: Demonstratives and Proper NamesThe distribution of iP DPs outlined in previous sections is similar to two other typesof DPs in Okanagan, demonstratives and proper names. The following subsectionsdiscuss each of these in turn.4.6.1 Demonstrative DPsThe Okanagan demonstrative system encodes spatial deictic distinctions as well asmovement of a referent relative to the speaker (Table 4.2). Demonstratives can bedivided into ?simple? individual-denoting demonstratives, and predicative demon-strative adverbials.43 I will limit my discussion largely to the simple demonstra-tives, since these will be analyzed as DPs.Table 4.2: Demonstratives in Okanagan, A. Mattina (1973)SimpleDemonstrativesDemonstrative AdverbsLocation Source DirectionFromDirectionToProximal ax?P al?P at?P atl?P a ?kl?PDistal ix?P il?P it?P itl?P i ?kl?PThe demonstrative adverbs are transparently related to the prepositions l ?at?, t?source?, tl ?from?, and ?kl ?to/towards? (A. Mattina 1973).44The Okanagan simple demonstratives ax?P ?this? and ix?P ?that? are not lim-ited to referring to inanimate, or non-human objects. They can easily refer to asentient, contextually relevant human subject.45 Simple demonstratives function43It is worthwhile to note that while i ?kl?P is classified here as an adverbial demonstrative, it issometimes used as a simple demonstrative by speakers of the Upper Nicola dialect.44There is no particle x corresponding to the simple demonstratives.45Unlike the case for Thompson xeP ?that? (Koch, 2008a, 273).106as arguments for a wide range of predicates, including adjectival (47a,b), locative(47c), verbal intransitive (47d), transitive (47e,f), and nominal (47g) predicates.Argument demonstratives may either follow (47) or precede (48) their predicates.46The distribution of these demonstratives is the same as that of iP DPs, as alreadydiscussed.(47) a. p@x?p?x?tsmartix?P.DEMThat one is smart.b. ?t@xtsweetix?P.DEMIt is sweet.c. ?klLOCs@n- ?kQ?w-m@nLOC-pray-INSTRix?P.DEMHe is at church.d. c-q?c@lxCUST-runix?P.DEMHe is running.e. P??-@neat-[DIR]-1SG.ERGix?P.DEMI ate that.f. paP-nt-?sfold-DIR-3SG.ERGix?P.DEMHe/she folded it.g. ?q@ ?y-m?nwrite-INSTRax?PDEMThis is a book.46In Kalispel, the demonstrative i?e (cognate with Okanagan ix?P) cannot follow a lexical pred-icate, as in Okanagan (47) below. Sally Thomason (p.c.) indicates that i?e seems to function as adiscourse particle, and so this might explain its restriction to pre-predicative positions.107(48) a. ix?PDEMp@x?p?x?t.smartThat one is smart.b. ix?PDEM?t@xt.sweetIt is sweet.c. ix?PDEM?klLOCs@n- ?kQ?w-m@n.LOC-pray-INSTRHe is at church.d. ix?PDEMc-q?c@lx.CUST-runHe runs.e. ix?PDEMP??@n.eat(INTR)He?s eating.f. ix?PDEMpaP-nt-?s.fold-DIR-3SG.ERGHe/she folded it.g. ax?PDEM?q@ ?y-m?n.write-INSTRThis is a book.Examples (49-52) below serve to emphasize the similar distribution of simpledemonstratives and nominal iP DPs, in both object (49-51) and subject (52) po-sitions.(49) a. iPDETx?xwt@mlittle.girl?c? ?m-qs-@ssuck.nose-[DIR]-3SG.ERGix?P.DEMThe girl kissed him.b. iPDETx?xwt@mlittle.girl?c? ?m-qs-@ssuck-nose-[DIR]-3SG.ERGiPDETt@tw?t.boyThe girl kissed the boy.108(50) a. paP-nt-?sfold-DIR-3SG.ERGix?P.DEMHe folded it.b. paP-nt-?sfold-DIR-3SG.ERGiPDET?q@ ?y-m?n.write-INSTRHe folded the paper.(51) a. s? ?q-@nsplit-[DIR]-1SG.ERGix?P.DEMI split this.b. s? ?q-@nsplit-[DIR]-1SG.ERGiPDETsli ?p.firewoodI split the firewood.(52) a. ix?PDEMs?ws(t)-@sdrink-[DIR]-3SG.ERGiPDETsiw?kw.waterHe is drinking water.b. iPDETsq@ltm?xwmans?ws(t)-@sdrink-[DIR]-3SG.ERGiPDETsiw?kw.waterThe man is drinking the water.Simple demonstratives, like iP DPs, function as direct arguments of adverbialdemonstratives like i ?kliP ?over there? (53,54):(53) a. i ?kl?PDEMix?P.DEMIt (e.g. the cup) is over there.b. ix?PDEMi ?kl?P.DEMIt (e.g. the cup) is over there.(54) a. i ?kl?PDEMiPDETsqwsiP-s.son-3SG.POSSHis son is over there.109b. iPDETsqwsiP-sson-3SG.POSSi ?kl?P.DEMHis son is over there.Since demonstratives pattern with iP DPs, I assume that they are a type of DP,but it is important to note that demonstratives may also associate with an iP DP,as in other Interior Salish languages (Matthewson and Davis (1995), Matthewson(1998), Kroeber (1999), N. Mattina (2006)), and form DP constituents with theirassociated iP DPs (55).47,48(55) a. kaPk?c-@nfind-[DIR]-1SG.ERG[ax?PDEMiPDETsqla ?w DP].moneyI found this money.b. w?k-@nsee-[DIR]-1SG.ERG[ix?PDEMiPDETylm?xw@m DP].chiefI saw that chief.The demonstratives in (55) are analogous to English uses of demonstratives asdeictic determiners, except that in Okanagan, the demonstrative cannot be analyzedas a determiner because it co-occurs with the determiner iP, and demonstrativescannot replace determiners (56a-57a).49(56) a. *kw?n-(n)ttake-DIRax?PDEM?p?kwlaP.ballTake this ball.47Demonstrative adverbs may also associate with an iP DP.48N. Mattina (2006, 105) claims that Moses-Columbian DP-adjacent demonstratives are best ana-lyzed as intransitive predicate demonstratives. She states that because they do not occur within a DP(i.e. inside of a determiner), they are not attached to a DP, but does not discuss the possibility thatthey might occur in a specifier position, or possibly adjoined to DP (see discussion below).49Although in fast speech, the iP determiner is sometimes difficult to hear after a demonstrativebecause of the segmental identity between the second syllable of a demonstrative and the determineritself, but the fact that they can co-occur is any case sufficient evidence for a non-D analysis of sim-ple demonstratives in Okanagan. Determiners are much more regularly dropped in other languagesof the Southern Interior, and in Moses-Columbian, for example, N. Mattina (2006) rules out mor-phophonological reasons for missing determiners, and so it is less clear what the syntactic status ofdemonstratives is for Moses-Columbian in cases where a determiner is not apparent.110b. kw?n-(n)ttake-DIRax?PDEMiPDET?p?kwlaP.ballTake this ball.(57) a. *ax?PDEM?ti ?kwtlakex?astgoodtATTRs@n-caQ-cQ?-lx-t@n.LOC-bathe-RED-body-INSTRThis lake is a good place to swim.b. ax?PDEMiPDET?ti ?kwtlakex?astgoodtATTRs@n-caQ-cQ?-lx-t@n.LOC-bathe-RED-body-INSTRThis lake is a good place to swim.The distributional evidence thus suggests that demonstratives cannot be analyzedas occurring in D position, unlike iP.It is also important to note that a demonstrative cannot associate with a quasi-object of a morphologically intransitive predicate (58).50(58) a. *kn1SG.ABSx?m?nk-@mwant-MID(*ix?P)DEMtOBLn? ?k-m@n.cut-INSTRI want that knife.b. *kn1SG.ABSw?k-@msee-MID(*ix?P)DEMtOBLsq@ltm?xw.manI saw that man.The reason for this restriction is that middle intransitive predicates such as those in(58) may only select for a quasi-object, not a full DP. The data in (59) show thatdemonstratives can associate with oblique-marked NPs, creating constituents thatbehave syntactically like core argument DPs.50The same restriction applies to themes of transitive -x(i)t- applicatives, however interestingly abare demonstrative can function as an applicative theme, e.g. kwu xwi ?cxts ax?P ?He gave me this?.Davis and Matthewson (2003) note that bare demonstratives can function as applicative themes inLillooet, but not as intransitive quasi-objects. I currently am lacking data showing whether or not abare demonstrative can function as a quasi-object in Okanagan; however I predict that this should notbe possible. The acceptability of bare demonstratives, but not demonstrative-associated oblique NPs,as ditransitive themes is interesting and may support positing distinct semantic analyses for these twocases. In chapter 8, I claim on independent grounds that demonstrative-associated oblique NPs areof type , while bare demonstratives are uniformly of type e.111(59) a. c-n-kwn?-st-@nCUST-n-sing-CAUS-1SG.ERGix?PDEMtOBLqw?l@m.songI sang that song.b. w?k-@nsee-[DIR]-1SG.ERGix?PDEMtOBLsq@ltm?xw.manI?ve seen that man (before).In contexts similar to (59), the oblique marker alternates more or less freely withiP, with no apparent semantic effect (60).51(60) a. kwin-[n]ttake-DIRax?PDEMtOBLp? ?kwlaP.ballTake this ball.b. kwin-[n]ttake-DIRax?PDEMiPDETp? ?kwlaP.ballTake this ball.This is directly parallel to an alternation in Lillooet (Matthewson and Davis, 1995;Matthewson, 1998) between DPs containing a demonstrative and an assertion ofexistence determiner ti...a (61a) and DPs containing a demonstrative and a non-assertion of existence determiner kwu (61b), with the exception that in Okanagan, tis not a determiner.(61) a. t@xwp-m?n-?kanbuy-APPL-1SG.ERGtiPDEMtiDETk??h-a.car-EXISI bought that car. (Lillooet, Matthewson (1998, 217, ex.81a))b. t@xwp-m?n-?kanbuy-APPL-1SG.ERGtiPDEMkwuDETka?h.carI bought that car. (Lillooet, Matthewson (1998, 216, ex.80a))51In Okanagan, the alternation between iP and t also occurs in the context of demonstrative asso-ciated DPs in equative contexts, and establishes an important morphosyntactic basis for analyzingOkanagan clefts as equatives (cf. section 8.5.2). There is also an information structural differencewhich surfaces between DPs with demonstratives adjoined to iP DPs, and those with demonstrativesadjoined to t NPs, such that in the former case, the entire constituent may be in focus, whereas in thelatter case, only the demonstrative may be in focus. See section 8.5.112A demonstrative-associated oblique marker in Okanagan (62, cf. section 8.5.2),and a demonstrative-associated kwu DP in Lillooet (63) may only select for an NP.(62) ix?PDEMiP/*tDET/*OBLpaP-nt-?s.fold-DIR-3SG.ERGThat?s what he folded.(63) a. t@xwp-m?n-?kanbuy-APPL-1SG.ERGtiPDEMtiDET???k-ago-EXISkn?tiP.DEICI bought that one that?s going by there.(Lillooet, Matthewson (1998, 217, ex.81c))b. *t@xwp-m?n-?kanbuy-APPL-1SG.ERGtiPDEMkwuDET???kgokn?tiP.DEICI bought that one that?s going by there.(Lillooet, Matthewson (1998, 216, ex.80a))Baker (2003, 182) understands this pattern as resulting from the fact that onlya subset of determiners in a language license null nouns (cf. English I admirethe/*a/?*that rich.). Thus, while Okanagan iP and Lillooet ti...a may both licensenull NPs (cf. discussion in 4.3 on null NPs), the oblique marker t and non-assertionof existence determiner kwu must select for a lexical NP. The reason behind this isunclear.In summary, demonstratives can function as stand-alone arguments (64a,65a),exactly like iP DPs (64b,65b), or may associate with an iP DP or oblique markedNP (64c,65c).(64) a. s? ?q-@nsplit-[DIR]-1SG.ERGix?P.DEMI split that.b. s? ?q-@nsplit-[DIR]-1SG.ERGiPDETsli ?p.firewoodI split the firewood.113c. s? ?q-@nsplit-[DIR]-1SG.ERGix?PDEMt/iPOBL/DETsli ?p.firewoodI split that firewood.(65) a. ix?PDEMs?ws(t)-@sdrink-[DIR]-3SG.ERGiPDETsiw?kw.waterHe is drinking water.b. iPDETsq@ltm?xwmans?ws(t)-@sdrink-[DIR]-3SG.ERGiPDETsiw?kw.waterThe man is drinking (the) water.c. ix?PDEMt/iPDETsq@ltm?xwmans?ws(t)-@sdrink-[DIR]-3SG.ERGiPOBL/DETsiw?kw.waterThat man is drinking the water.Okanagan demonstratives present several major analytical difficulties. These areas follows:(66) a. The Projection Problem: If simple demonstratives are not determiners(56-57), but their external syntax is identical to iP DPs, are simple demon-stratives categorially DPs, and if so, where is the projecting head?b. Internal Constituency: A demonstrative licenses an oblique-marked NPto function as a DP argument (59), but where is the determiner?c. Compositionality: If demonstratives are type e expressions, like iP DPs,how can a demonstrative compose with an iP DP of the same type?There is no easy solution to these problems. Matthewson and Davis (1995) andMatthewson (1998) discuss similar issues concerning Lillooet demonstratives anddemonstrative-associated DPs (e.g. 61 and 63). Davis (2006, 2009) shows thatdemonstrative-associated DPs in Lillooet permit condition C violations, whereassimple demonstratives do not, a finding which may indicate that there is a structuraland semantic difference between demonstratives, depending on whether they occurin isolation or adjoined to a DP. Clearly this problem requires further research, bothin Okanagan and in other Salish languages.114For the purposes of this dissertation, I assume based on their external distribu-tion that simple demonstratives, demonstrative-associated iP DPs, and demonstrative-associated t NPs are all categorially DPs (67), and thus abstract away from theproblems in (66).(67) a. s? ?q-@nsplit-[DIR]-1SG.ERG[ix?PDP].DEMI split that.b. s? ?q-@nsplit-[DIR]-1SG.ERG[ix?PDEMt/iPOBL/DETsli ?p DP].firewoodI split that firewood.4.6.2 Proper Name DPsProper names pattern with simple demonstratives and nominal iP DPs in beingarguments of adjectival, locative, intransitive, transitive, and nominal predicates.As with nominal iP DPs and demonstratives, the linear order of a predicate and aproper name argument is free. (68) shows the proper name argument preceding thepredicate52, and (69) shows the proper name argument following the predicate:(68) a. IvanIvan?Qa?t.wetIvan is wet.b. MikeMike?klLOCs@n- ?maP ?m?yaP-t@n.LOC-teach-INSTRMike is at school.52For (69c), but not (68e), the proper name is interpretable as an object. The availability of thesubject reading in either case, and the fact that it is required in (68e), is illustrative of how theOne Nominal Interpretation Effect (ONI) (Gerdts, 1988) is sometimes inoperative in Okanagan. Allother things being equal, a pre-predicative transitive argument is preferably interpreted as a subject,whereas a post-predicative transitive argument is preferably interpreted as an object. The equivalentof (68e) is ungrammatical in (Lower) Lillooet since a pre-predicative DP must be interpreted as asubject, but this can only result in an ONI violation (Davis, 1999b).115c. JohnJohnnis.leaveJohn left.d. TinaTina?cqw-aqw-m?st.cry-FRED-INTR.REFLEXTina cried (to herself).e. PetePetewik-s.see-[DIR]-3SG.ERGPete saw him./*He saw Pete.f. NormanNormankwu1SG.ABSn-Qay-Qay-?nk-s.n-tickle-RED-stomach-[DIR]-3SG.ERGNorman tickled me.g. JohnJohns@xw-mr?m-@m.OCC-medicine-MIDJohn is a doctor.(69) a. ?klLOCs@n- ?kQ?w-m@nLOC-pray-INSTRJohn.JohnJohn is at the church.b. qas-nc?tscratch-REFLEXIvan.IvanIvan scratched himself.c. wik-ssee-[DIR]-3SG.ERGPete.PetePete saw him./He saw Pete.d. kwu1SG.ABStaq-swave-[DIR]-3SG.ERGIvan.IvanIvan waved at me.e. s@xw-mr?m-@mOCC-medicine-MIDJohn.JohnJohn is a doctor.116Assuming proper names are a type of argument DP, we predict that proper namesmay substitute for both argument demonstratives and nominal DPs, which is indeedthe case (70-71):53(70) a. NancyNancys@xw- ?maP ?m?yaP-m.OCC-teach-MIDNancy is a teacher.b. ix?PDEMs@xw- ?maP ?m?yaP-m.OCC-teach-MIDShe/That is a teacher.c. [iP][DET]is ?kw?y1SG.POSS-mothers@xw- ?maP ?m?yaP-m.OCC-teach-MIDMy mother is a teacher.(71) a. SpikeSpikePayx?wt.tiredSpike is tired.b. ix?PDEMPayx?wt.tiredHe?s tired.c. iPDETtk?mi?xwwomanPayx?wt.tiredThe woman is tired.Proper names also function as arguments for adverbial demonstratives (72), sim-ilarly to nominal iP DPs and simple demonstratives, which were discussed in theprevious section.(72) a. i ?kl?PDEMJohn.JohnJohn is over there. (answer to ?Where is John??)53As indicated for (70c), an iP determiner predictably reduces before first i(n)- and second a(n)-person possessive morphology (A. Mattina 2000). It is always present in similar contexts involvingthird person possessive agreement.117b. JohnJohni ?kl?P.DEMJohn is over there. (answer to ?Where is John??)Despite the preceding data showing that proper names can clearly pattern withother DP arguments, I assume that Okanagan proper names are best analyzed aslexical NPs, which either undergo N-to-D raising in argument contexts (Longob-ardi, 1994), or else are complements to a null referential determiner. The choicebetween these two analyses is not important for the purposes of this dissertation.54To begin with, proper names do not normally occur with an iP determiner inargument DP contexts in Okanagan, as illustrated by (73):(73) a. JohnJohnwik-ssee-[DIR]-3SG.ERGMaryMaryJohn saw Mary.b. #JohnJohnwik-ssee-[DIR]-3SG.ERGiPDETMaryMaryJohn saw Mary.The reason for this is not because the determiner is ungrammatical before propernames, however, but rather that the determiner implies that a proper name, suchas Mary, is a common noun of sorts whose predicate domain is a non-singletonset. Comparing (73) above with (74) below, we see that the determiner can occurbefore a proper name argument in marked contexts.54Proper names occur with non-deictic determiner kwi in Squamish (Gillon, 2006), and with theproper name determiner kw in Lillooet (van Eijk, 1997; Matthewson, 1998), and so a hypothesiswhereby Okanagan has a null, referential determiner that only occurs with proper names receivessome cross-linguistic support.118(74) a. yaQy?QtalliPDETMaryMary?tw?st-l@x.stand-3PL.ABSAll the Marys stood up.b. iPDET??@x?- ??x??pRED-growniPDETJohnJohniPDETc-n-Pu?xw.CISL-n-enterThe old John came in (i.e. not the young John).c. iPDETD?ny@lDanielaPDETc-my-st-in,CUST-know-CAUS-1SG.ERGx?astgoodiPDET?klLOCsqilxw.native.peopleThe Daniel that I know is good to the people.In (74a), Mary is a common noun of sorts, since it denotes a class of individualswith the property of being Mary, or having the name Mary. The iP determiner ismandatory here, required by yaQy?Qt ?all? as a quantifier in argument position. For(74b), the ?old John? is contrasted to a contextually salient ?young John?, whereJohn denotes a set of individuals with that name. In (74c), iP Daniel denotes anindividual which is contrasted with other individuals named Daniel which are notgood to the people.A second piece of evidence that proper names are lexical NPs comes fromdata showing that they may be affixed by possessive pronouns (75a), similarly tocommon nouns (75b).(75) a. inc?1SG.INDEPin-M?ry1SG.POSS-Marymys-x?astmore-goodtlLOCanw?2SG.INDEPtlLOCan-M?ry.2SG.POSS-MaryMy Mary is nicer than your Mary.Context: You and a friend both have daughters named Mary, and arearguing over which one is nicer.119b. inc?1SG.INDEPi(n)-s?t@mkP?lt1SG.POSS-daughtermys-x?astmore-goodtlLOCanw?2SG.INDEPtlLOCa(n)-s?t@mkP?lt.2SG.POSS-daughterMy daughter is nicer than your daughter.Proper nouns may also take absolutive pronominal morphology (76a) similarly tocommon nouns (76b), but unlike demonstratives (76c).(76) a. kn1SG.ABSJohn.JohnI?m John.b. kn1SG.ABSs@xw- ?q@ ?y-?m.OCC-write-MIDI?m a writer.c. *kwu1PL.ABSax?P.DEMThis is us.What the data in (73-76) seem to show is that Okanagan proper names caneither themselves be DP arguments (73), NPs which combine with iP determinersto form DPs (74, 75), or bare NP predicates (76a). For Longobardi (1994), propernames are inserted into the syntax as NPs, but raise to an empty D position atlogical form, through which they receive their directly-referential interpretations.A similar analysis of Okanagan proper names seems possible: if as NPs, they donot undergo raising to D to achieve DP-status (74), then they can still become DPswhen they are complements of an iP determiner.55 Syntactic evidence for N-to-Draising is non-existent, however, and so as an alternative, it may be preferable toassume a null referential determiner that selects only for proper names, similar toLillooet kw (van Eijk, 1997; Matthewson, 1998).An alternative analysis which is consistent with the data in this section is toassume that proper names are lexically DPs, but that they may be coerced into NPs55Though their interpretations will be different because of implicatures associated with iP. Seechapter 5.120in specific syntactic and pragmatic contexts. At present, there is little evidencefor or against such an analysis, but it is worth mentioning that (i) a coercion-to-NP analysis of proper names might predict that demonstratives could be coercedinto NPs as well, but this is clearly not the case, and (ii) there is cross-Salishanevidence that proper names may co-occur with referential determiners in unmarkedcontexts.564.7 SummaryThis chapter first introduced the basic concept of noun-hood in Okanagan Salish(section 4.2), and then presented syntactic tests for distinguishing nouns from othercategories (section 4.3). Next, the general distribution of predicates and argumentswas discussed, and the distribution of the determiner iP and oblique marker t wasshown to be grammatically predictable (section 4.4). Then, I presented my syn-tactic analysis of iP DPs and oblique-marked quasi-objects (section 4.5), beforediscussing demonstratives and proper names as two other types of DPs (4.6).Chapter 5 discusses the semantics of core iP DP arguments, and how they differsemantically from oblique-marked quasi-objects.56In chapter 7, I argue that proper names must be analyzed as DPs in equative contexts, but may bepredicative NPs in identificational contexts. Unfortunately, these data do not seem to clarify whethercoercion-to-NP analysis or a null referential determiner (or N-to-D raising) analysis is correct forOkanagan proper names since in both cases, the distribution of NP and DP proper names is limited toonly specific grammatical environments, and in both cases, null derivations/morphemes are involved.121Chapter 5Semantics of OkanaganDeterminer PhrasesThe semantics of determiners and determiner phrases (DPs) in Okanagan Salishhas not previously been investigated in any depth, and apart from N. Mattina?s(2006) study of Moses-Columbian determiners, there are no other systematic inves-tigations into the semantics of determiners for Southern Interior Salish languages.This chapter analyzes the Okanagan determiner iP as involving a choice functionover a contextually restricted domain of individuals (Reinhart, 1997; Kratzer, 1998;Matthewson, 1999). The referential strength of iP is determined by the number ofindividuals within a restricted context set and by whether the choice function se-lects a singular, a plural, or a maximal individual from that context set. The Okana-gan oblique marker t is not a determiner, but a semantically vacuous morphologi-cal reflex of semantic incorporation (Van Geenhoven, 1998; Chung and Ladusaw,2004; Farkas and de Swart, 2004; Carlson, 2006), which is a property of certainintransitivizing morphemes in Okanagan, such as middle -m.The semantics of iP is important for an understanding of the semantics of DP-DP structures, as discussed in chapter 7.1225.1 IntroductionThe determiner systems of Okanagan and the other languages of the SouthernInterior differ in many ways from those of Northern Interior and Central Salishlanguages. In particular, the Upper Nicola dialect of Okanagan makes use ofonly one determiner, iP.1 This contrasts starkly with Northern Interior languagessuch as Lillooet, for example, which has an elaborate determiner system (vanEijk, 1997; Matthewson, 1998), and somewhat less starkly with Thompson andShuswap, which each have two referential determiners, and one non-referentialdeterminer (Kroeber, 1999, 70).Based on a range of tests, I claim that the referential strength of an iP DP iscontextually determined, and sensitive to domain restriction. Technically speaking,a choice function f (Reinhart, 1997) selects one singular or plural individual fromthe intersection of the nominal and contextual (C) domains (Gillon, 2006). BecauseiP DPs allow individual-concept readings, unlike Lillooet assertion-of-existenceDPs (Demirdache, 1996), I claim that iP DPs may denote intensional individuals. Iassume the following semantics for iP (1a), and claim that it carries the maximalityimplicature given in (1b):(1) a. [[iP]] = ?P?w[ f (?x[P(x)(w)?C(x)(w)])]b. Maximality implicature:f = MAXThe formula says that the determiner iP takes as its first argument a (nominal) pred-icate P, and as its second argument a world w, and asserts that a free variable overchoice functions f selects an individual x from the intersection of P and the contextset C (also a free variable) in that world.2 The maximality implicature states thatthe choice function is equivalent to the maximal individual in the intersection of P1Colville-Okanagan utilizes a second determiner ?@P (A. Mattina 1973). Upper Nicola speakersdo not allow ?@P to take nominal complements except in identificational copular sentences, in whichsomething is being given a name. Given this very limited distribution before nominals, and the factthat ?@P/?aP is used as a clausal subordinator in other contexts, I analyze it as a complementizer forthe Upper Nicola dialect.2Concerning the semantic type of the variables, P and C are of type <>, and f is of type<,e>.123and the context set C.3 (1b) represents the default setting for iP, however in caseswhere f is existentially bound, or selects one individual from a set of contextuallysalient individuals, this default is overridden.Oblique quasi-objects introduced by t differ both syntactically and semanticallyfrom iP DP objects. I claim that the case-marker t is semantically vacuous, and sim-ply serves to indicate that the following nominal is semantically incorporated (VanGeenhoven, 1998; Chung and Ladusaw, 2004; Farkas and de Swart, 2004; Carlson,2006). Quasi-objects in Okanagan exhibit many of the cross-linguistic hallmarksof semantically incorporated nouns (Carlson, 2006). I suggest that the intransitiviz-ing morphology (e.g. ?middle? suffix -@m) which licenses oblique quasi-objects en-codes semantic incorporation (see section 5.3.2). Most importantly for this thesis,t does not employ domain restriction, since it does not access the context.As discussed in chapter 4, the distribution of iP and t is syntactically pre-dictable, and the possible semantic interpretations of a formally transitive predicatewith an iP DP object overlap with the interpretations of a formally intransitive pred-icate with a quasi-object introduced by t, specifically for cases where the defaultfor iP, f = MAX, is overridden. A speaker?s manipulation of the morpho-syntax oftransitivity (and the accompanying use of iP versus t in argument contexts) leadsto a strong conversational implicature, given as (2):(2) Conversational implicature:Because the default for iP is f = MAX, a speaker?s use of an intransitiveconstruction with a quasi-object introduced by t carries a conversationalimplicature of non-maximality.The chapter is outlined as follows: Section 5.2 investigates the semantics of iPand t. First I show that iP and t are not deictic (5.2.1), not definite (5.2.2), not likeEnglish indefinites (5.2.3), and not like Lillooet assertion-of-existence wide-scopeindefinites (Matthewson, 1998, 1999) (5.2.4). Next, I show that t is felicitous insome environments where iP is not, and vice-versa, and conclude that iP carries animplicature of uniqueness and maximality of a referent, while t does not (5.2.5).3Rullmann (1995, 143) defines the maximal individual as ?the unique element of the set of whichall other elements of the set are parts.? Applied to a set A, the formal definition is MAX(A) = ?x[x ?A??x? ? A[x?? x]].124In 5.2.6, I give additional data which suggest that iP restricts the domain of indi-viduals, rather than creating a generalized quantifier (Barwise and Cooper, 1981).Sections 5.2.7 and 5.2.8 present additional data showing existential and genericuses of iP, which are important in showing that domain restriction may be vacuousin certain cases. Section 5.2.9 discusses in more detail intensional readings of iPDPs, and section 5.2.10 summarizes section 5.2. Section 5.3 presents my analysis,and section 5.4 concludes.5.2 The Semantics of the Determiner iP and ObliqueMarker tThis section introduces data relevant to determining the semantics of iP and t. I testfor deixis, definiteness and specificity, English-like existential indefiniteness, andwidest-scope indefiniteness. Many of these tests were first utilized by Matthewson(1998) and Gillon (2006) in their studies of Lillooet and Squamish determiners,respectively. Table 4.1 below lists relevant semantic properties:Table 5.1: Semantic Properties of iP and tSection Property iP t5.2.1 Deictic x x5.2.2 Definite x x-presuppose existence x x-assert uniqueness/maximality x x-specificity x x5.2.3 English-like existential indefinite x x5.2.4 Wide-scope indefinite (cf. Lillooet and Squamish) x x-can take wide-scope w.r.t. modals and negation X x-must take wide-scope w.r.t. modals and negation x x5.2.5 Carries an implicature of uniqueness/maximality X xFollowing these tests, I present some additional data suggesting that iP DPs donot create generalized quantifiers (5.2.6), and allow for both existential (5.2.7) and125generic readings (5.2.8), as well as intensional readings (5.2.9). I then summarizemy findings.5.2.1 iP and t are not DeicticFirstly, the determiner iP does not encode spatial deixis or a visible/invisible dis-tinction. It may introduce a nominal like sq@ltm?xw ?man? whose referent is eitherproximal and visible to the speaker (3a) or distal and invisible (3b) to the speaker.(3a) is likewise felicitous if the speaker cannot see the man, and (3b) if the speakerwere watching a moonwalk on television.(3) a. a ?kl?PDEMc-xwistCUST-walkiPDETsq@ltm?xw.manA man is walking over here.b. iPDETsq@ltm?xwmanc-xwistCUST-walkiPDETlLOCx?y??n@x?w.moonA man is walking on the moon.Speaker knowledge of the location of a referent is irrelevant in determiningwhether or not an iP DP may be used. In (4), for example, the speaker is assertingthat John is looking for a book, but may then overtly cancel any implicature that heor she knows where that book is. In other words, if iP entailed knowledge of thelocation of a referent, then it should not be possible to utter the second conjunct of(4).(4) JohnJohnc- ??aP ??aP-nt-?sCUST-look.for-DIR-3SG.ERGiPDETn- ?q@ ?y-?s-t@n,n-write-eye-INSTRu?CONJlutNEG?taEMPHc-my-st-?nCUST-know-DIR-1SG.ERG?kaPk?n.whereJohn is looking for a book, but I don?t know where it is.It is important to consider whether iP might be specified as having a ?neutral?deictic feature, similar to Squamish ta (Gillon, 2006), rather than being unspecifiedfor deixis. The Squamish neutral determiner can be used ?for referents which canbe located or were locatable at some point by the speaker? (Gillon, 2006, 46). A126felicitous use of iP, however, seems unrelated to whether or not a referent is evenin principle locatable:(5) lutNEG?t@EMPHc-w?k-st-nCUST-see-CAUS-1SG.ERGiPDETsq@ltm?xwmank-c-n ?c? ?w-@m-s.k-CUST-wash.dishes-MID-3SG.POSSI?ve never seen a man washing dishes.Although a ?neutral? deictic feature implies that there are no deictic restrictionson the use of an item, it also implies that there are other available ?non-neutral?deictic values which might associate with other determiners. As there are no otherdeterminers in Okanagan, and deixis is largely determined by demonstratives inthe language, I take this as evidence that iP has no deictic specification.4As further evidence that Okanagan iP is non-deictic, consider Matthewson(2008), who notes that Lillooet assertion-of-existence DPs do not allow bound vari-able, E-type (Cooper, 1979; Evans, 1980; Elbourne, 2001), or other anaphoric read-ings. Instead, null pronominals must be used to achieve these readings. She claimsthat ?deictic features of the overt DPs force reference to the discourse situation,and this prevents binding or variation across situations? (p. 543). Although boundvariable readings for Okanagan DPs are restricted,5 E-type readings are easily ob-tainable. This essentially means that the denotation of an iP DP can vary acrosssituations, and does not entail the existence of a particular individual at a singlelocation. In (6), the final DP iP s ??aPc?n@m denotes some non-specific individualwhich Norman says he will shoot.4In this respect then, Okanagan iP is similar to the Moses-Columbian ?non-demonstrative, gen-eral? neutral determiner Pani (N. Mattina 2002).5Bound variable readings of iP DPs in distributive contexts are not possible without overt posses-sor morphology on the nominal, implying that while iP DPs may be bound across worlds and times,they may not be distributed over. I discuss some data involving distributive readings in section 5.2.6,but largely sidestep these issues for reasons of space.127(6) NormanNormanks-p?x?-aPxFUT-hunt-INCEPT?COMP?tQap-?mshoot-MIDtOBLs ??aPc?n@m,deerx?mink-swant-[DIR]-3SG.ERGswitwhoks- ?c?qw-iP-sFUT-skin-MID-3SG.POSSiPDETs ??aPc?n@m.deerNorman said he?s going to kill a deer and he wants someone to skin thedeer. (adapted from Matthewson (2008, ex.45))Okanagan iP DPs also allow co-varying interpretations in cases similar to thosewhich involve bridging in English. The DP iP ylm?xw@m ?the chief? in (7) denotesa different individual for every reserve that the speaker visits.(7) kn1SG.ABS?aPCOMPc-xwuyCUST-go?klLOCsqlxw-?laPxw,native.person-landpintkalwaysiPDETylm?xw@mchiefiPDETcxPitfirstac-qw@lqw?l-st-n.CUST-talk-CAUS-1SG.ERGEvery time I visit a reserve, I talk to the chief. (adapted from Matthewson(2008, ex.51))The availability of E-type readings for Okanagan iP DPs sets them apart from Lil-looet assertion-of-existence DPs. This is consistent with a claim that OkanaganiP has no deictic features to force reference to the discourse situation, or preventbinding across situations. The case-marker t also cannot be analyzed as having de-ictic features: it allows only narrow-scope, non-specific interpretations, and is bestanalyzed as semantically vacuous, as I will show in following sub-sections.5.2.2 iP and t are Not DefiniteiP and t do Not Encode Familiarity or Presuppose ExistenceThere is no familiarity requirement associated with iP. That is, iP does not pre-suppose the existence of a unique or non-unique referent which satisfies the NPrestrictor, and places no constraints on the common ground of discourse. Evidence128for this claim comes from the beginning of narratives, where a new discourse ref-erent may be introduced by iP (cf. similar data for Lillooet (Matthewson, 1999),for Squamish (Gillon, 2006), and for Moses-Columbian (N. Mattina 2006)):(8) ?qs?pilong.agokwukwREPiPDETt@tw?t.boyx?w?l-st-s@lx.abandon-CAUS-3PL.ERGA long time ago, there was a/*the boy. They abandoned him.(Lindley and Lyon, 2012, stz. 139)The above use of iP patterns with the English indefinite determiner a. Unlike anEnglish indefinite, however, once a discourse referent is established, an iP DP eas-ily allows a co-referential reading. In other words, iP is felicitous in both definiteand indefinite contexts. To illustrate, (9a) shows a new referent iP x ??ut ?a rock?being introduced at the beginning of a narrative. At a later point in the story, thereferent is once again invoked using an iP DP (9b).(9) a. lLOC?qwumqn-?tkwhead-waterk-s?lxwaPHAVE-bigiPDETx ??utrockil?PDEMswitwhoxiPw??lxpass.byu?CONJc- ?kQ?w-@m.CUST-pray-MIDAt Chapperon Lake there is a big rock where people who pass by pray at.(Lindley and Lyon, 2012, stz. 116)b. u?CONJks-knx?t-m-sFUT-help-2SG.ACC-3SG.ERGiPDETx ??ut.rockIt will help you, the rock.(Lindley and Lyon, 2012, adapted from stz. 122)iP and t do not Assert or Presuppose Uniqueness or MaximalityUniqueness assertions, and maximality assertions for plural DPs (Link, 1983), aresometimes considered properties of the English definite determiner (Heim, 2011).It quickly becomes apparent that iP and t do not assert or presuppose the uniqueness129of a referent.6 First of all, iP is felicitous in situations where it is an implicit fact thatthere is more than one contextually-salient element satisfying the nominal property.Consider (10) below:(10) Context: There are two cups on a table, equidistant from the speaker.kwu1SG.ABSc-kwi[n]-?tCISL-take.something.for.someone-APPLiPDETlpot.cupBring me a cup. (adapted from Gillon (2006, 88)])Consultant?s comment: Then I?d pass you one of the cups.Given that neither one of the two cups in (10) is specifically under discussion, thecontext set must include both cups. Any assertion or presupposition of uniquenessis therefore incompatible with this context. Similarly for mass nouns, iP does notassert maximality:(11) P??-@neat-[DIR]-1SG.ERGiPDETs-?tx-itkw,NOM-sweet-watern?x?@m?CONJil?PDEM?win-xt-m-nleave-BEN-2SG.ACC-1SG.ERGmiFUTnixwalsoanw?2SG.INDEPkw2SG.ABSks-P??@n-aPxFUT-eat-INTRtOBLs-?tx-itkw.soupI ate some soup, but I saved you some so you can eat too.Data also show that neither iP nor t encode a presupposition of uniqueness (12).(12) also exemplifies the pragmatic overlap of quasi-objects and iP DPs.(12) Context: I enter a room and tell you what happened to me today.a. t-kic-nt-meet-[DIR]-1SG.ERGiPDETsq@ltm?xwmanQapn?Pnowsx?@lx?Q?lt.todayI met a man today.6This makes iP similar to deictic determiners in Squamish (Gillon, 2006, 88). Matthewson (2008,15) argues that Lillooet assertion-of-existence determiners presuppose uniqueness relative to a situ-ation, which allows cancellation of their maximality effects.130b. kn1SG.ABSt-k?c-@mt-meet-MIDtOBLsq@ltm?xwmanQapn?Pnowsx?@lx?Q?lt.todayI met a man today.It is not feasible for the speaker to expect the hearer to be familiar with theparticular man to which the speaker refers in this context. If iP or t encoded apresupposition of uniqueness, we might expect presupposition failure in contextsfor which the hearer does not have in mind the same unique referent as the speaker,yet there is no presupposition failure for (12).7In question-and-answer contexts (e.g. 13-14), iP may be used to establish a newdiscourse referent (13b), or answer a question related to a previously establisheddiscourse referent (14b). A construction involving a quasi-object cannot be usedto answer a question related to a previously established discourse referent (14c).It can instead only be construed as establishing a new discourse referent, and istherefore infelicitous in the context given in (14a).8(13) a. Context: Questioner has no idea what the addressee might have thrown.sti ?mwhatiPDETc ?qmin-[n]t-xw?throw-DIR-2SG.ERGWhat did you throw?b. c ?qmi[n]-nthrow-[DIR]-1SG.ERGiPDET?p?kwlaP.ballI threw a/the ball.7It is possible that the reason why there is no presupposition failure in cases like (12) is due to thefact that Okanagan speakers more easily accomodate presuppositions than do speakers in English. Inlight of the fact that parameterizing accomodation is no easy task, I do not follow this general lineof reasoning, but instead argue that iP is non-presuppositional, following Matthewson (2006a) forLillooet. See also discussion in section 8.3.3.8The syntax of WH-questions in Okanagan requires further work. The constituent introduced byiP in (13a) is a DP containing a headless relative relative clause (cf. chapter 6), and while WH itemshave traditionally been analyzed as predicates in the Salish literature, since they occur in predicateposition, there is some doubt as to whether this is the correct analysis or not. Baptiste (2001, sec-tion 3.3) weighs three separate possible analyses of WH-questions in Okanagan: WH in-situ, WHmovement, and clefting. Under the WH in-situ analysis in particular, a WH item may be analyzableas a DP, however since they do not have the same distribution has other DPs (e.g. they cannot occurpost-verbally, generally), Baptiste rejects this hypothesis.131c. kn1SG.ABSc ?qm?n-@mthrow-MIDtOBL?p?kwlaP.ballI threw a ball.(14) a. Context: Questioner is specifically asking what happened to a definite ball.xPk?n@mwhereiPDET?p?kwlaP?ballWhere is the ball?b. c ?qmi[n]-nthrow-[DIR]-1SG.ERGiPDET?p?kwlaP.ballI threw a/the ball.c. #kn1SG.ABSc ?qm?n-@mthrow-MIDtOBL?p?kwlaP.ballI threw a ball.If iP or t presupposed uniqueness, we might expect (i) iP to be infelicitous in anindefinite context, which it is not (cf. 13b as an answer to 13a); and (ii) t to befelicitous in a definite context, which it is not (cf. 14c as an answer to 14a). Sinceeither may freely be used in indefinite contexts, neither can be analyzed as encodinga presupposition of uniqueness.9Finally, data showing that iP DPs may occur in existential sentences providefurther evidence that there is no maximality assertion associated with iP:(15) xwPitmanyiPDETsiw?kwwater?klLOC?kaP??s.over the hillThere is a lot of water over that hill.That t also does not assert uniqueness or maximality follows from the factthat data like (14c) cannot be used in contexts involving a previous establishedunique referent (14a), but there is evidence that t cannot be used in any contextsinvolving a unique referent. In (16), x?y??n@x?w ?sun? denotes a singleton set, a factwhich is implicitly part of the interlocutors? common ground. Since t cannot make9(14c) is not possible as an answer to (14a) because t does not reference the context, and so cannotco-refer with a previously introduced discourse referent.132anaphoric reference to the context set, (16b) is only interpretable as introducinga new discourse referent to the common ground, resulting in an implicature thatthere is ?more than one? sun.10(16) Context: It?s been cloudy for several days now. I come over to your houseand without any prior context ask you:a. ucDUBw?k-@nt-xwsee-DIR-2SG.ERGiPDETx?y??n@x?wsunQapn?P?todayDid you see the sun today?b. #ucDUBkw2SG.ABSw?k-@msee-MIDtOBLx?y??n@x?wsunQapn?P?todayDid you see a sun today?I conclude that neither the determiner iP nor the oblique marker t assert or presup-pose uniqueness or maximality of a referent, however while iP can make referenceto the context set and select a maximal individual, t cannot, by implicature.iP and t do not Encode SpecificityOkanagan iP can be felicitously used for both specific and non-specific referents,and as such does not encode specificity, or reflect any specific/non-specific distinc-tion. The case-marker t, by contrast, is consistently non-specific (cf. N. Mattina(2006) on Moses-Columbian).Diesing (1992) and En? (1991) define specificity as involving a non-empty andcontextually salient set P, where a DP denotes a sub-part of P. For Ludlow andNeale (1991), a contextually salient set P need not be discourse-old in order touse a specific determiner; the only necessary presupposition is that the set P isnon-empty. Under either definition, Okanagan iP does not encode specificity.The following two English sentences, taken from Matthewson (1998, 95-97),illustrate how the English indefinite determiner a permits a specific reading:10The same issue arises with proper names in oblique contexts.133(17) a. Sophie didn?t buy a book I recommended.b. Every boy in Mary?s class fancies a girl who Mary doesn?t know.The indefinite DP in (17a) refers to a specific book. Likewise in (17b), on thereading where ?a girl? takes scope over the distributive operator, ?a girl? is alsointerpreted specifically. In both cases, the DP is discourse-new, and in both cases,there is a presupposition that the set P is non-empty. English a does not encode anypresupposition that the set P is non-empty, since as the following sentence shows(also taken from Matthewson, 1998), a is also felicitous in contexts where there isno presupposition that P is non-empty (e.g. since there are no unicorns).(18) Sophie didn?t buy a unicorn.Similarly in Okanagan, the felicitous use of iP and t does not depend on any pre-supposition that a set P is non-empty. (19c,d), for example, are fine in both worldswhere sasquatches exist (Context A) and do not (Context B) exist.(19) a. Context A: There are such things as sasquatches. Some of your friendshave seen one, but you never have.b. Context B: There are no such things as sasquatches.c. lutNEGny ?Qipalways?t@EMPHc-w?k-st-@nCUST-see-CAUS-1SG.ERGiPDET?cwan?ytmx.sasquatchI?ve never seen a/the sasquatch.Consultant: I?ve never seen a sasquatch in my whole life.d. lutNEG?t@EMPHkn1SG.ABSc-w?k-@mCUST-see-MIDtOBL?cwan?ytmx.sasquatchI?ve never seen any sasquatch.?Consultant: I?ve never seen no sasquatch.The DP iP ?cwan?ytmx ?a sasquatch? in (19c) may have either a specific or a non-specific reading in worlds where sasquatches exist (depending on whether it scopesabove or below negation), and a non-specific reading in worlds where sasquatches134do not exist.11I assume that specificity is not a property of a DP whose referent varies acrossworlds and times. In (20) below, the DP iP sq@ltmixw ?a man? permits both aspecific interpretation (i.e. the same man sits at the same table every day, andJohn always argues with that particular man), and a non-specific interpretation (i.e.John argues with whichever man happens to be sitting at a particular table).(20) JohnJohnpintkalwayskaPCOMPc-qwal-st-w?xw-sCUST-argue-CAUS-RECIPR-3SG.ERGiPDETsq@ltm?xwmaniPDETmutsitlLOClat?p.tableJohn always argues with the man who sits at the table.Temporal adverbials also appear to force non-specific, narrow scope readings ofiP DPs.12 In (21) below, the iP DP cannot be interpreted specifically, since thesame letter cannot arrive every day, but instead denotes a different non-specificindividual for every afternoon. Okanagan DPs, unlike those in Lillooet, need notbe bound to the utterance situation, since otherwise we predict only an infelicitousspecific reading for the DPs in (21).(21) yaQy?QtalliPDETs@nya ?kwq?nafternoonac-kic-xCUST-arrive-INTRiPDET?q@ ?y-m?n.write-INSTREvery afternoon a letter arrives.(22) shows that Okanagan iP DPs also permit individual concept readings. Anindividual concept is of type , a function from worlds/times to individuals.This is unlike the case for Lillooet assertion-of-existence determiners (Demirdache,1996):11The consultant prefers a specific reading of the DP in (19c), and will often retranslate suchcases using an English definite determiner. I suggest that this is due to a pragmatic implicature (cf.section 5.3.3) which is based on the fact that while both (19c) and (19d) are available in non-specificcontexts, the quasi-object in (19d) has only a non-specific reading.12This may involve binding of implicit world and time arguments associated with the nominal. Cfalso the E-type and bridging cases discussed in section 5.2.1.135(22) cmayEPISiPDETylm?xw@mchieflaPk?nwheniPDETtk?milxw.womanMaybe someday the chief will be a woman.For (22), the referent of iP ylm?xw@m ?the chief? is any non-specific future indi-vidual who happens to be the chief at that future time. Data like these support anintensional analysis of iP DPs (cf. section 5.2.9).Okanagan t, unlike iP, may only ever be used non-specifically, which is con-sistent with an analysis whereby quasi-objects are semantically incorporated (VanGeenhoven, 1998; Chung and Ladusaw, 2004; Carlson, 2006). By way of exam-ple, many realis relative clauses with quasi-object nominal heads (e.g. 23a) areungrammatical, since the relative clause head must be interpreted specifically. Inother words, t sq@ltm?xw ?a man? in (23a) must scope under negation since it canonly be interpreted non-specifically, but t is inconsistent with the selectional re-strictions of the relative clause restrictor, which as a formally transitive predicaterequires that the head be introduced by iP, thus implying that a specific man wasseen (Cf. chapter 6 for further discussion of relative clauses). (23a) is correctedto (23b), where the DP iP sq@ltmixw allows a specific interpretation, similar to theEnglish indefinite in (17a).(23) a. *lutNEGkn1SG.ABS?t@EMPHkaPk?c-@mfind-MIDtOBLsq@ltm?xwmaniPDETw?k-@nsee-[DIR]-1SG.ERGtOBLspiPs ?c??t.yesterday# (?) I didn?t find a/any man that I saw yesterday.b. lutNEG?t@EMPHkaPk?c-nfind-[DIR]-1SG.ERGiPDETsq@ltm?xwmaniPDETw?k-@nsee-[DIR]-1SG.ERGtOBLspiPs ?c??t.yesterdayI didn?t find the man that I saw yesterday.Having established that Okanagan iP and t do not presuppose existence, pre-suppose or assert uniqueness or maximality, or encode specificity, I now move on136to further data showing how iP differs from both the English indefinite determinera and Lillooet widest-scope assertion-of-existence determiners.5.2.3 iP and t are not English-like Existential IndefinitesOkanagan iP is clearly not a definite determiner, but it is possible that it couldbe similar to a Russellian interpretation of the English indefinite determiner a, asrepresented by (24):(24) [[a]]= ?P.?Q.?x[P(x)?Q(x)]Assuming (24) for iP entails that iP asserts the existence of some individual thatsatisfies the nominal property. Since there is no presupposition or assertion ofuniqueness associated with either a or iP, as we have seen, (24) might be taken asa plausible candidate for iP. Note that (24) allows the referent to be any individualwhich satisfies the nominal property. While DP co-reference is possible with (24),the availability of English definite the pragmatically blocks indefinite a DPs frombeing used co-referentially in most contexts (Heim, 2011). Given that Okanagandoes not have a contrasting definite determiner, assuming (24) for iP predicts thatthere should be no pragmatic restrictions on iP DP co-reference.Okanagan iP allows co-referential readings more easily than English a, whichis as predicted given that iP may be used in definite contexts, and that there is nodedicated definite determiner to block co-referential interpretations. (25a) showsthat in Okanagan, an iP DP does not force a co-referential reading, similarly toEnglish indefinite a. (25b) shows that an iP DP may just as easily allow a co-referential reading as a non-co-referential one, however.1313Within the same sentence, and even across sentences, null pronominals are often preferred overovert DPs as a topic maintenance strategy (cf. for example Davis (1994) for relevant data in Lillooetand Gerdts and Hukari (2003) for Halkomelem). For example, a co-referential reading of (25a) ispossible if there is a null pronominal in the second conjunct, rather than an overt DP (i):(i) w?k-@nsee-[DIR]-1SG.ERGix?PDEM(iPDETx?w ??iP)mountain.goatlLOCn? ?q??m@lx,Quilchenau?CONJw?k-@nsee-[DIR]-1SG.ERGlLOCsp?x?m@n.(Sp?x?m@n) Douglas.LakeI saw a mountain goat in Quilchena, and I saw (the same one) in Sp?x?m@n (DouglasLake). (adapted from Matthewson (1999, ex.56)).137(25) a. w?k-@nsee-[DIR]-1SG.ERGiPDETx?w ??iPmountain.goatlLOCn? ?q??m@lx,Quilchenau?CONJw?k-@nsee-[DIR]-1SG.ERGiPDETx?w ??iPmountain.goatlLOCsp?x?m@n.Sp?x?m@n (Douglas.Lake)I saw a mountain goat in Quilchena, and I saw a mountain goat inSp?x?m@n (Douglas Lake) (adapted from Matthewson (1999, ex.56)).Consultant?s Comment: Good, 2 different goats.b. JohnJohnn-Pul ?q?s-sn-lift-[DIR]-3SG.ERGiPDETlLOCnx?lsQ??xwt@n,windowu?CONJMaryMaryn-xn?s-sn-close-[DIR]-3SG.ERG(iPDETlLOCnx?lsQ??xwt@n).windowJohn opened a window, and Mary closed the window.Consultant?s Comment: Same window, just to make it clear, you say ?win-dow?.Consider that in a context like (25a), it is implausible that the same mountain goatcould be at Quilchena and Douglas Lake, since these two reserves are 15 kilometersapart, while in (25b), it is entirely plausible that Mary closed the same windowwhich John opened. Note that in English, it is usually infelicitous to say Johnopened a window, and Mary closed a window, if in fact the intended referent is thesame window.More importantly, co-reference between two identical iP DPs is preferred,whenever possible. In (26), given that the same chief cannot be born in two sepa-rate places, the prediction is that if iP was the same as an English indefinite, the iPDP in the second conjunct could be used without knaqs ?another?, but this is not thecase. In other words, knaqs prevents co-reference between two occurrences of iPylm?xw@m ?the chief? in a context which requires two separate referents. Note thatthe context in (25a) strongly favors, but does not absolutely require, two separate138referents, which may also explain why knaqs is not required in (25a).14,15(26) ?qs?pilong.agoiPDETylm?xw@mchief?kw??l-@?lmake-FREDlLOCQuilchenaQuilchenauPCONJ*(iPDETknaqs)anotheriPDETylm?xw@mchief?kw??l-@?lmake-FREDlLOCVancouver.VancouverLong ago, a chief was born in Quilchena and a/another chief was born inVancouver.Gillon (2006, 108) claims that in Squamish, co-reference between two deicticDPs with the same noun is expected, but that since maximality is only implicatedand not asserted, pragmatics can overrule a co-referential reading. I claim thatthe same facts hold for the Okanagan determiner iP: the co-referential reading ispragmatically overruled in (25a), and overruled with some additional help fromknaqs ?another? in (26).Quasi-objects in Okanagan are consistently indefinite and non-specific, simi-larly to those in Moses-Columbian (N. Mattina 2006), and appear to pattern moreclosely to English indefinites than to iP DPs. Once a discourse referent has beenintroduced, it is infelicitous to refer back to the same referent with a quasi-object,as illustrated by (27a). The relevant contrasting nominal expressions are bolded.(27) a. #kn1SG.ABSPaws-p?x?-@mgo-hunt-MIDtOBLspiPs ?c??t.yesterdaykn1SG.ABSw?k-@msee-MIDtOBLsk@mx?st,bearu?CONJkn1SG.ABS?tQap-?mshoot-MIDtOBLsk@mx?st.bear.I went hunting yesterday, and I saw a bear, and I shot the bear.14There seems to be some speaker variation with regards to whether or not knaqs ?another? isneeded to block co-reference for cases like (25-26).15Example (26) contains the sequence iP knaqs iP ylm?xw@m ?another chief?. According to thecriteria developed in chapter 6, this should be a DP constituent containing a pre-posed relative clause,literally ?the chief who is another?. The correctness of this hypothesis is unclear, however, sinceelements such as knaqs ?one (HUMAN)?, which are ostensibly quantifiers, might not be able to projectclauses. It is telling, however, (i) that simple nouns also appear as modifiers in this configuration, as(95) shows in chapter 8, and (ii) weak quantifiers like xwPit ?many? appear in this configuration, andfunction as main clause predicates. While more research needs to be done on ?determiner doubling?in examples such as (26, cf. 40 below), I tentatively assume that relative clause modification isinvolved in these cases.139b. kn1SG.ABSPaws-p?x?-@mgo-hunt-MIDtOBLspiPs ?c??t.yesterdaykn1SG.ABSw?k-@msee-MIDtOBLsk@mx?st,bearu?CONJ?tQap-nt-?nshoot-DIR-1SG.ERGiPDETsk@mxist.bear.I went hunting yesterday, and I saw a bear, and I shot the bear.This pattern is consistent with an analysis whereby iP can reference a contextuallyrestricted domain, which in the case of (27b) is a singleton set. t cannot reference acontextually restricted domain, and so the implicature in (27a) is that there are twodifferent bears under discussion. Oblique t, however, also cannot be analyzed asan English-like indefinite since it does not license specific or other types of wide-scope readings which a exhibits.5.2.4 iP is Different than a Lillooet Widest-scope IndefiniteMatthewson (1999) analyzes the Lillooet assertion-of-existence determiner ti...a asa widest-scope indefinite determiner, utilizing a choice-function analysis adaptedfrom Reinhart (1997). She assumes that the choice function is existentially closedat the highest level.16 This analysis correctly blocks narrow-scope, E-type, andbound variable readings.The strongest piece of evidence against analyzing Okanagan iP as a widest-scope indefinite comes from data suggesting that iP is possible in contexts whichdo not assert the existence of any individual. This is clearly shown to be the casewith data where iP scopes under negation (28, cf. 19a).(28) Context: ?Do you know any chiefs??lutNEG?taEMPHc-my-st-inCUST-know-CAUS-1SG.ERGiPDETyl-ylm?xw@mRED-chiefI don?t know any chiefs.For Okanagan, an unambiguous narrow scope reading of an object nominalmay be achieved by using a quasi-object and an intransitive form, rather than aniP DP. Compare transitive (29a-30a) which have iP DP objects (highlighted), with(29b-30b) with quasi-objects (highlighted).16Matthewson (2001) analyzes the choice function as a free variable (Kratzer, 1998).140(29) a. lutNEG?t@EMPHc-wik-st-nCUST-see-CAUS-1SG.ERGiPDETsq@ltm?xwmank-c-n ?c? ?w-@m-s.k-CUST-wash.dishes-MID-3SG.POSSI?ve never seen a/the man washing dishes.b. lutNEG?t@EMPHkn1SG.ABSc-w?k-@mCUST-see-MIDtOBLsq@ltm?xwmantOBLk-c-n ?c? ?w-@m-s.k-CUST-wash.dishes-MID-3SG.POSSI?ve never seen a man wash dishes.(30) a. ?tiEMPHiPDETsqilxwnative.peopleiPDETw?k-@n,see-[DIR]-1SG.ERGlutNEGw?k-@nsee-[DIR]-1SG.ERGiPDETsm-s?maP.RED-white.peopleI only see the Indian people, I didn?t see any/the white people.b. ?tiEMPHiPDETsqilxwnative.peopleiPDETw?k-@n,see-[DIR]-1SG.ERGlutNEGkn1SG.ABS?t@EMPHw?k-@msee-MIDtOBLsm-s?maP.RED-white.peopleI only see the Indian people, I didn?t see any white people.Under their narrow scope readings, the two forms in each pair essentially conveythe same meaning, however (29a-30a) have an additional wide-scope reading ofthe object nominal which (29b-30b) do not.Okanagan iP DPs may scope under a modal, a fact which sets Okanagan iPapart from both Lillooet assertion-of-existence and Squamish deictic determiners(31). Additionally, section 5.2.1 showed that iP DPs permit both E-type and in-dividual concept readings, neither of which are possible for Lillooet assertion-of-existence DPs.(31) Context: Set in a strange land, the speaker has no idea if any horses existhere.141iPDETsq@ltm?xwmancmayEPISkaPk?c-iP-sfind-MID-[DIR]-3SG.ERGiPDETs@nk? ?caPsq?x?aP,horse?COMPi ?kl?P-@lx.DEM-3PL.ABSThe man might find a horse, if there are any out there.Finally, (31-32) both demonstrate that an iP DP may be used in a context wherethe existence of a referent, iP s@nk? ?caPsq?x?aP ?the horse? or iP sqwsiP ?his son?, isnot entailed.(32) cakwBOULSpikeSpike?aPCOMPk[?]-sqwsiP,HAVE-soncmayEPISix?PDEMiPDETsqwsiP-sson-3SG.POSSiPDETk?-ylm?xw@m.FUT-chiefIf Spike had a son, I guess his son would be the chief.The Lillooet equivalents of the examples in (28-32) all require the non-assertion-of-existence determiner kwu. Although neither Okanagan iP nor Lillooet assertion-of-existence determiners encode definiteness, we have seen that Okanagan iP per-mits a wider range of readings than Lillooet assertion-of-existence determiners,including crucially, narrow-scope readings.5.2.5 iP Carries an Implicature of Uniqueness and Maximality (tdoes not)Data suggest that iP neither asserts nor presupposes uniqueness in the case of sin-gular referents (e.g. 10), and neither asserts nor presupposes maximality in thecase of plural or mass referents (e.g. 11), but there is nevertheless evidence that iPimplies both of these qualities. In argument contexts, where iP DPs contrast withoblique quasi-objects, I claim that iP carries an implicature of maximality (33a, cf.section 5.3.3) and gives rise to a conversational implicature (33b):1717The maximality implicature carried by iP gives rise to a separate conversational implicature inequative contexts. I discuss this in section 7.4.1.142(33) a. Maximality implicature:f = MAXb. Conversational implicature:Because the default for iP is f = MAX, a speaker?s use of an intransitiveconstruction with a quasi-object introduced by t carries a conversationalimplicature of non-maximality.An implicature of maximality is evident from comparing (34a) with (34b). For(34a), the implication is that all the berries were eaten, but this implicature is can-cellable (34b).(34) Context: There was a bowl of berries on the table, but now it is gone. I ask?What happened to the berries?? You reply:a. Pi?-@neat-[DIR]-1SG.ERGiPDETs- ?p ?yq-a?q.NOM-ripe-fruitI ate (all) the berries.b. Pi?-@neat-[DIR]-1SG.ERGiPDETs- ?p ?yq-a?q,NOM-ripe-fruitn?x?@m?CONJil?PDEM?kim-xt-m-nexcept-BEN-2SG.OBJ-1SG.ERGiPDETs- ?p ?yq-a?q.NOM-ripe-fruitI ate some/#the berries, but I saved you some.An implicature of uniqueness is apparent in many question-and-answer con-texts involving iP DPs. Consider (35) uttered in a context where my friend and Iare tossing around a ball, and my friend throws it to me while I am not looking.The DP iP ?p?kwlaP denotes a definite ball in (35), and it is only felicitous to answer(35) using an iP DP (36a).(35) xPk?n@mwhereiPDET?p?kwlaP?ballWhere is the ball?143(36) a. c ?qmi[n]-nthrow-[DIR]-1SG.ERGiPDET?p?kwlaP.ballI threw the/#a ball.b. #kn1SG.ABSc ?qm?n-@mthrow-MIDtOBL?p?kwlaP.ballI threw the/#a ball.Superficially, this contrast resembles the definite/indefinite contrast seen in En-glish. The DP iP ?p?kwlaP in (36a) certainly denotes the unique ball in the discoursecontext, but as I have shown, the fact that only (36a) is felicitous cannot be due toany presupposition or assertion of uniqueness associated with iP. Instead, this fol-lows simply from the fact that iP is sensitive to the context. Responses involvingquasi-objects (36b) are infelicitous in these contexts because t is not contextuallysensitive, and as such can only be interpreted as infelicitously introducing a newdiscourse referent.The opposite pattern obtains in question-and-answer contexts where the ques-tion includes a quasi-object. Consider (37) uttered in a context where two friendsare discussing their ravaged garden, and are wondering who or what could possiblyhave been the culprit. A felicitous answer must include an oblique marked nominal(38a).(37) ucDUBkw2SG.ABSw?k-@msee-MIDtOBLsp@pl?naP?rabbitDid you see a/any/some rabbit(s)?(38) a. wa ?yyeskn1SG.ABSw?k-@msee-MIDtOBLsp@pl?naP.rabbitYes, I saw a/some rabbit(s).b. #wa ?yyesw?k-@nsee-[DIR]-1SG.ERGiPDETsp@pl?naP.rabbitYes, I saw a/some rabbit(s).Since t is always non-maximal, the questioner in (37) is not referencing anymaximal set of rabbits, yet (38b) answers (37) as if the questioner had been imply-144ing reference to a maximal set of rabbits.The implicature of uniqueness and maximality associated with iP is reinforcedby the scalar opposition which holds between iP and t, as exemplified by (39).(39) a. kn1SG.ABSPaws-p?x?-@mgo-hunt-MIDu?CONJkn1SG.ABS?tQap-?mshoot-MIDtOBLs ??aPc?n@m,deeru?CONJi-ks- ?p ?yq-?m1SG.POSS-FUT-cook-MIDiPDETs ??aPc?n@m.deerI went hunting and I shot a deer, and I?m gonna cook the (entire) deer.b. kn1SG.ABSPaws-p?x?-@mgo-hunt-MIDu?CONJkn1SG.ABS?tQap-?mshoot-MIDtOBLs ??aPc?n@m,deeru?CONJkn1SG.ABSks- ?kw?lcnc?t-aPxFUT-cook-INTRtOBLs ??aPc?n@m.deerI went hunting and I shot a deer, and I?m gonna bake some deer formyself.?Consultant?s Comment: It?s the same deer. The deer that you shot, you?regonna cook some of it, you?re not gonna cook the whole thing.For (39a), a speaker implies that the maximal sub-part of iP s ??aPc?n@m ?the deer?is being cooked (in this case a contextually salient individual), while for the quasi-object in (39b), maximality is not implied. The scalar relation between iP and tgives rise to a non-maximal, partitive-like reading. I claim that this cannot be anactual partitive reading, however, but involves only pragmatic inferencing: Withregards to (39b), if you shoot some deer, the deer you cook is likely to involvewhatever deer it was that you shot. It is important to show that true partitive read-ings are not possible with t, since I am claiming that t is semantically vacuous anddoes not reference the context.(40) below supports the hypothesis that true partitive readings are not possiblewith t. In (40a), the iP DP in the second conjunct refers to one of the four blanketsin the first conjunct. In other words, the context set in this case includes fourblankets, and the iP DP in the second conjunct singles out one of these. In (40b)the quasi-object cannot refer to one of the blankets in the first conjunct, since tdoes not reference the context. Pragmatic inferencing is not a complicating factor145in (40), as it may be in (39), since ?making a blanket? does not follow from ?puttinga blanket down in a trunk?, especially since the blankets in the trunk are alreadymade. In sum, if true partitive readings were possible with t, then the prediction isthat (40b) should have a partitive reading available, but this is not the case.(40) a. musfourtATTRs? ?c@mblanketsac-n-qm?[n]-nCUST-LOC-lay.down-[DIR]-1SG.ERGiPDETlLOCtrunk,trunkkw??l-@nmake-[DIR]-1SG.ERGiPDETnaqsoneiPDETsi ?c@m.blanketThere are four blankets that I put in the trunk, I made one of them.b. musfourtATTRs? ?c@mblanketsac-n-qm?[n]-nCUST-LOC-lay.down-[DIR]-1SG.ERGiPDETlLOCtrunk,trunku?CONJkn1SG.ABSkw??l-@mmake-MIDtOBLnaqsonetATTRsi ?c@m.blanketThere are four blankets that I put in the trunk, and I?m making one.Consultant?s Comment: That would be the fifth one.By using iP then, it is understood that the speaker is referencing a contextually-salient, and possibly restricted set. For context sets involving single referents, thereferent of an iP DP will be interpreted as unique in its context, so long as there isno overt cancellation of the uniqueness implicature (33a). For context sets involv-ing multiple referents, the referent of an iP DP will be interpreted as maximal inits context, unless a singular noun is used to denote one individual from within thecontext set, or there is an overt cancellation of the maximality implicature (33a).18By using oblique t, however, it is understood that the speaker is not referencing acontextually-salient set (33b).It is infelicitous to use an iP DP if it is explicit from the context that the set isempty. Consider the following exchange. In (41), speaker A establishes a referentfor a particular chief, and then speaker B asserts that they do not have a chief,thereby negating the existence of any referent for the DP iP ylm?xw@m.1918Plurality is not obligatorily marked on Okanagan nouns. Some nouns mark plurality suppletivelyor by reduplication, while for others, the singular and plural forms are identical.19Note that (41) is similar to examples like (6) in section 5.2.1 showing that iP DPs may have146(41) A: kn1SG.ABSn?x@?lhearx?astgood?aPCOMPc-qw@lqw?ltCUST-speakiPDETylm?xw@m.chiefI heard that the (your) chief is a good speaker.B: lutNEGkwu1PL.ABS?t@EMPHk?-ylm?xw@m,HAVE-chiefu?CONJsxPkinxhowmiFUTx?astgood?aPCOMPc-qw@lqw?ltCUST-speak(#iP(DETylm?xw@m).chief)We don?t have a chief, so how can the chief be a good speaker?By using an iP DP, speaker B is referring to the same chief that speaker A mis-takenly assumes to exist. The implicature of uniqueness arising from B?s use of iPresults in infelicity, because B?s initial clause asserts that the contextually-relevantset of chiefs is empty.In sum, the data show that iP allows both specific and non-specific readings,and is felicitous in both definite and indefinite contexts. The case-marker t allowsonly non-specific readings and cannot be used as a definite. I claim that iP carriesan implicature of uniqueness/maximality (33), while t does not since it does notaccess the context set.5.2.6 iP does not Create a Generalized QuantifierThe distribution of Okanagan iP DPs provides some evidence that they cannot begeneralized quantifiers of type <,t> (Barwise and Cooper, 1981), but insteadalways denote individuals of type e (Matthewson, 1998) or . While empiricalevidence against generalized quantifiers in Okanagan remains sketchy at this point,support for this idea nevertheless comes from data showing that the universal quan-tifier yaQy?Qt ?all?, which itself never occurs in D position, can only associate withan iP DP (42), and never with a quasi-object (43):E-type readings. The difference with (6) is that, although the domain may only include individualsin future or counterfactual worlds, the domain is nevertheless non-empty in these cases. See alsocases involving iP DPs scoping under if-clauses. pro, on the other hand, may reference an emptyset (41), which follows if there is no maximality or uniqueness implicature associated with pro (cf.Matthewson (2008) for a discussion of related differences between full DPs and pro in Lillooet).147(42) P??-@neat-[DIR]-1SG.ERG(yaQy?Qt)(all)iPDETq?qxw@lx.fishI ate (all) the fish.(43) kn1SG.ABSPi?neat(INTR)(*yaQy?Qt)(*all)tOBLq?qxw@lx.fishI ate all the fish.These data constitute evidence that the determiner iP is not a semantically vacuousagreement or case marker, but is instead a necessary step for deriving a stronglyquantified DP (Matthewson, 1998, 2001). I suggest that iP provides the neces-sary domain for the strong quantifier, a domain which neither a quasi-object nor abare nominal NP can provide. In other words, DP-adjoined strong quantifiers likeyaQy?Qt ?all? in Okanagan require arguments of type e (cf. Matthewson (1998) forLillooet), but oblique quasi-objects and bare NPs both denote sets of type .Under the assumption that a strong quantifier cannot select for a generalized quan-tifier argument of type <,t> (contra Giannakidou (2004)), these data also pro-vide evidence that iP does not create a generalized quantifier.Further evidence against a GQ-forming analysis of iP may come from restric-tions on distributive readings. Okanagan iP DPs allow less-than-widest-scopereadings, as shown by their ability to scope under negation and modals (section5.2.4). They nevertheless seem to disallow distributive readings, similar to Lillooet(Matthewson, 1999; Davis, 2010b), which is unexpected under a GQ analysis.20(44a) shows that a DP including the noun tk?milxw ?woman? cannot be interpretedas scoping under the quantified subject, and (44b) shows the same thing for theDP iP s ?kw ?kw?ym@lt ?the child?. In other words, (44a) cannot be interpreted as everyman being loved by a different woman, and (44b) cannot mean that each womankissed a different child.20Davis (2010b) analyzes the equivalent of strong quantifiers in Lillooet as domain-adjusting op-erators over the denotations of plural DPs, following Brisson (1998).148(44) a. #yaQy?QtalliPDETs-q@l-q@ltm?xwNOM-RED-manx?m?nk-nt-@mlike-DIR-PASSiPDETtOBLtk?milxw.womanTarget: Every man has a woman who love(s) him. (adapted fromMatthewson (2008, ex.54))Literally: Every man is loved by a woman.Consultant?s Comment: All the men were loved by this one woman.b. #yaQy?QtalliPDETs-maP-mP?mNOM-RED-women?c@ ?m- ?c? ?m-qs-@sRED-suck-nose-[DIR]-3SG.ERGiPDETs- ?kw- ?kwiy-m-@lt.NOM-small-m-childTarget: All the women kiss a child (different children.)Consultant?s Comment: You need to say sc@cm?laP ?children?.In contexts where more than one woman or child is under discussion, the sup-pletive plural forms smaPmP?m ?women? and sc@cm?laP ?children? must be used,but plural objects in these contexts do not have clear distributive readings, but arerather consistent with cumulative interpretations (e.g. All the men are loved by thewomen) (Davis, 2010b).21A cornerstone of Matthewson?s (1999) argument that DPs in Lillooet are notgeneralized quantifiers comes from data showing that they are scopally inert, andtherefore always take widest-scope. Although Okanagan iP DPs do not necessarilytake widest-scope, as we have seen, they apparently do scope over distributiveoperators.22 This receives explanation under the assumption that iP does not createa generalized quantifier. This problem requires further research, however.5.2.7 Existential Sentences and iP DPsFor Okanagan, iP DPs are commonly volunteered as subjects of sentences denot-ing existential propositions (45). Matthewson (1999) cites similar data for Lillooet21Alternatively, (44) may involve an English-like dependent plurality (Hotze Rullmann, p.c.),where the plurality of the object must match that of the subject. This problem requires further work.22The choice function seems to scope over the distributive operator. That is, at most one singularor plural individual can be denoted by an iP DP, and a narrow scope interpretation does not entailthat a distributive interpretation also be available. This issue requires further investigation.149assertion-of-existence determiners, and Gillon (2006, 87) for Squamish deictic de-terminers. In many existential contexts, forms like (45a) are interchangeable withforms like (45b), which do not have iP DP subjects, but rather prefix a nominalpredicate with (Pa)k?- ?have?.23(45) a. i ?kl?PDEMiPDETs ??aPc?n@mdeeriPDET?klLOCwist.highThere?s deer up in the hills.b. i ?kl?PDEMk?-s ??aPc?n@mHAVE-deeriPDET?klLOCwist.up.highThere?s deer up in the hills.Below, (46a) shows that iP is compatible with the non-proportional weak quantifierxwPit ?many? in these contexts,24 but not the strong quantifier yaQy?Qt ?all? (46b).Under the assumption that English and Okanagan existential sentences should pat-tern similarly, this contrast provides supporting evidence that these are indeed ex-istential sentences (Milsark, 1977).(46) a. xwPitmanyiPDETsiw?kwwater?klLOC?kaP??s.over.the.hillThere is a lot of water over that hill.b. *yaQy?QtalliPDETsiw?kwwater?klLOC?kaP??s.over.the.hillThere is all the water over that hill.23These are not always interchangeable, however. There is an implicature of uniqueness/maximal-ity associated with using the iP forms which surfaces in certain contexts, and which renders theminfelicitous as existentials.24The quantifier is the syntactic predicate in these sentences. Strong quantifiers cannot function aspredicates in Salish (Matthewson, 1998, 278).1505.2.8 Generic Readings of iP DPsGeneric interpretations of nouns require the iP determiner in Okanagan (47a,b).25I claim that generic iP DPs denote intensional maximal pluralities of type (Chierchia, 1998), and that appealing to kinds (Carlson, 1977) as a distinct typeof individual is unnecessary for Okanagan. Intensionality is a necessary compo-nent to this analysis for two reasons: (i) certain predicates (e.g. unicorns, griffins,sasquatches, etc.) do not have instantiations in the actual world, but do in certainpossible worlds; and (ii) exceptions can be made to a generic statement, e.g. Dogslike to run, but my dog doesn?t, which under a purely extensional analysis wouldbe contradictory.An analysis like that of Chierchia (1998) is supported for Okanagan by thefact that any sentence containing a DP with a generic interpretation also has anon-generic interpretation available. For example, the sentences in (47a,b) allowgeneric interpretations of the subject iP DPs given an appropriate context, yet alsohave non-generic interpretations available, equivalent to The bear(s) like(s) thesaskatoons and The dog likes to run.(47) a. iPDETs-(km)-k@mx?stbear(s)x?mink-slike-[CAUS]-3SG.ERGiPDETs?yaP.saskatoon.berry(All) bears like saskatoon berries.b. iPDETk@kw?pdogt?PliPveryx?mink-slike-[CAUS]-3SG.ERGiPDETsnaxwt.run(ANIMAL)Dogs really like to run.Consider that some predicates in English do not distribute to atomic individu-als, but only apply to kinds (Carlson, 1977) (48):(48) Bears get bigger as you go north.The predication ?get bigger as you go north? is not true of any atomic individualbear, but only the kind ?bear?. Independent evidence for kind-denoting nominalsis not forthcoming for Okanagan, however. (49) was elicited as a translation of the25Similar data exist for Shuswap (Gardiner, 1993), and Lillooet (Matthewson, 1998). Determinersare apparently optional in generic contexts in Moses-Columbian N. Mattina (2006, 127).151generic sentence (48), but was later translated back into English as an existentialsentence:(49) ?klLOC?ca?tcoldiPDETt@mxw?laPxw,groundtaPl?PverykwukwREPp?s ??atlarge.PLiPDETkiPl?wnagrizzly.bearnaP?CONJsk@mx?st.black.bearTarget: Grizzlies and black bears get bigger as you go north.Volunteered Gloss: Where there?s a cold country, there?s lots of biggrizzly bears.Under an existential interpretation of (49), the complex DP iP kiPl?wnaP naP?sk@mx?st denotes non-specific individual grizzlies and black bears, not their cor-responding kinds. The generalization seems to be that if an Okanagan sentence hasa generic interpretation, it also has either an episodic (47) or an existential inter-pretation (49). I take this as evidence that the iP DPs within these sentences denoteindividuals rather than kinds.26By assuming that kinds reduce to contextually unrestricted, intensional max-imal pluralities in Okanagan, generic readings fall out independently. In otherwords, a generic versus non-generic interpretation of an Okanagan iP DP dependson whether the DP denotes all individuals in some world which satisfy a property(i.e. an intensional, contextually unrestricted maximal plural individual), or a sub-set of individuals which satisfy a property (i.e. a non-maximal singular or pluralindividual).Consider that although plural definites do not have generic readings in English(only bare plurals and singular definites), other languages such as Spanish may use26Demonstratives, too, can denote maximal pluralities. In (i) below (cf. 21b above, a minimalpair sentence without the demonstrative), the iP DP is interpreted as a non-specific singular or pluralindividual under the scope of the temporal adverbial, while the demonstrative denotes the maximalplurality which instantiates the non-specific individual.(i) P??-@neat-[DIR]-1SG.ERGix?PDEMiPDETq?qxw@lxfishyaQy?Qtallsx?@lx?Q?lt.dayI eat that fish every day.SM: Yeah, you?re talking about whatever kind of fish, ling-cod, kokanee, salmon.152definite plurals in generic contexts (Chierchia, 1998). Okanagan is therefore nottypologically unusual in allowing generic interpretations of plural DPs,27 and sothere is precedent for reducing kind readings to intensional maximal pluralities forsome languages. Under this analysis, the absence of contextual restriction is cru-cial, since if a nominal with individuals in its extension is contextually restricted, ageneric reading will not be possible.Intensionality, as a necessary component of a Chierchia-style analysis of genericreadings, is independently motivated for Okanagan iP DPs in non-generic contexts.I turn now to this data.5.2.9 Intensionality and iP DPsData like (50-51) show that the determiner iP is compatible with non-deictic, non-specific interpretations. They also show that iP DPs permit individual conceptreadings (En?, 1981; Demirdache, 1996), and are therefore of type . In otherwords, the DP iP ylm?xw@m ?the chief? in (50-51) does not necessarily pick out asingle individual, but whoever happens to be chief at some particular time.(50) cmayEPISiPDETylm?xw@mchieflaPk?nwheniPDETtk?milxw.womanSomeday the chief will be a woman.(51) iPDETylm?xw@mchiefny ?Qipalwaysks- ??aP- ??aP-?s-aPxFUT-RED-look.for-eye-INCEPTtOBLpx?p?x?tsmarttOBLtk?milxwwomanmiFUTsicbeforeylm?xw@m.chiefA chief always has to look for a smart woman in order to get elected.For Okanagan, there is an ambiguity in whether an iP DP like iP ylm?xw@m ?thechief? has an extensional, or an intensional reading. Under an extensional reading,iP ylm?xw@m ?the chief? has a specific or non-specific interpretation, depending onthe context. For (50-51), iP ylm?xw@m clearly does not denote a specific chief.For (50), it denotes an individual in some possible world that has the property of27Recall that bare plurals are independently ruled out in Okanagan, since bare nominals are alwaysungrammatical in non-predicative positions (cf. Matthewson (2001, 185) for Lillooet).153being a chief in that world, and for (51), it denotes whoever happens to be thechief in every possible world. These are intensional readings of iP DPs, where thereferent is determined by the accessible worlds of an epistemic modal like cmay?might/maybe? in (50) (Menzies, 2012), or ?future? ks- in (51).There is further evidence for an intensional analysis of iP DPs, involving anominal prefix in Okanagan, k?- ?to be? which usually (52a) but not always (52b)co-occurs with a possessive affix (A. Mattina 1996a):28(52) a. i-k?-c?txw1SG.POSS-TO.BE-housemy house-to-be (A. Mattina 1996a, 239)b. kw2SG.ABSk?-ylm?xw@m.TO.BE-chiefYou will be a chief.You are a chief-to-be. (A. Mattina 1996a, 239)I suggest that k?- ?to be? may be analyzed semantically as follows (cf. Matthewson(2006b) for a similar analysis of Lillooet kelh).(53) [[k?-]] = ?P?x? t?w?t??w?[P(x)(t?)(w?)?R(w,w?)? t < t?]The formula in (53) takes an NP predicate as an argument, and converts it intoan intensional predicate, of type >>.29 The proposition is true if thepredicate is true of the subject argument in some world w? which stands in an ac-cessibility relation R to the evaluation world w, and at some time t? which followsthe utterance time t. Sentence (52b) may be represented as in (54), meaning es-sentially ?You are the chief at some future time in some close possible world.? Theidea is that (52b) is false if ?you? are ?a chief? in the utterence world at the presenttime, but true if ?you? are ?a chief? in some closely accessible world at a futuretime.(54) ? t?w?t??w?[chie f (you)(t?)(w?)?R(w,w?)? t < t?]28A. Mattina (1996a) claims that the nominal prefix k?- is distinct from the verbal future prefix ks-,since there is complementary distribution of these two prefixes across the two word classes.29Type i refers to time intervals.154The crucial point here is that these intensional predicates can be complements ofan iP determiner, as shown below in (55a,b) iP k?ylm?xw@m ?the chief-to-be?.(55) a. cakwBOULSpikeSpike?aPCOMPk[?]-sqwsiP,HAVE-soncmayEPISix?PDEMiPDETk?-ylm?xw@m.TO.BE-chiefIf Spike had a son, I guess that would be the chief.b. inc?1SG.INDEPiPDETk?-ylm?xw@m.TO.BE-chiefI?m gonna be the chief.Similar to the case of iP ylm?xw@m ?the chief? in (50), the referent of iP k?-ylm?xw@m ?the chief-to-be? in (55a,b) cannot be found by choosing a member ofthe set of actual world chiefs. I suggest that iP k?-ylm?xw@m ?the chief-to-be? in(55a,b) denotes an intensional individual, of type .30This implies that an iP determiner can select for either an extensional predicateof type , or an intensional predicate of type >. Given that ?exten-sional? iP DPs also allow intensional readings, I suggest that an implicit worldvariable is always present as part of domain restriction, and that an iP DP mayalways denote an intensional individual, of type . This is an important com-ponent in my analysis of DP-DP structures, and has played a role in some analysesof equative copular clauses in English, where one of the arguments denotes anintensional individual (Romero, 2005).5.2.10 SummaryOkanagan iP closely resembles Squamish deictic determiners in being non-definiteand contextually sensitive, but does not have any obvious deictic features, in con-trast to both Lillooet and Squamish. The absence of deictic features (and the avail-ability of existential closure of a choice function variable f at any level, cf. sec-tion 5.3) plausibly explains the fact that iP permits non-specific and other narrowscope interpretations. iP nevertheless carries a uniqueness/maximality implicature30I am abstracting away from time intervals (type i) by making this statement. For the purposes ofthis thesis, the s type may be understood as representing world/time pairs: With k?- prefixed nouns,these must be future world/time pairs; for bare nouns (which also allow intensional readings), thereis no temporal restriction.155which favors co-referential interpretations for DPs, context permitting. This dis-tinguishes iP from the English indefinite determiner. Maximal interpretations of iPDPs include definite and generic-like interpretations. Non-maximal interpretationsinclude existential, non-specific, and other indefinite uses.In the next section (5.3), I analyze iP as being sensitive to contextual domainrestriction, since (i) the choice function can pick out any singular or plural indi-vidual within the intersection of the nominal and contextual domains, and (ii) thenominal domain can be a subset of the contextual domain (i.e. domain restrictionis vacuous). The case-marker t is not a determiner, and always exhibits narrow-scope, non-specific readings. As such, I analyze t as indicating semantic incor-poration of the noun (Van Geenhoven, 1998; Chung and Ladusaw, 2004; Carlson,2006). Though iP can also be used in narrow-scope, non-specific contexts, it con-versationally implicates uniqueness/maximality from the fact that iP, unlike t, canaccess a contextually restricted domain.5.3 Analysis5.3.1 Okanagan iPThe semantic analysis I propose for the Okanagan determiner iP is the same as thatargued for by Gillon (2006, 10) for Squamish deictic determiners (56a), but withintensionality built in, as in (56b).31 I assume that intensionality ultimately derivesfrom nouns (as represented below in 5.1).(56) a. [[ta]] = ?P[ f (?x[P(x)?C(x)])] (Squamish, Gillon (2006, 10))b. [[iP]] = ?P?w[ f (?x[P(x)(w)?C(x)(w)])] (Okanagan)In (56b), a variable over choice functions, f, selects one singular or plural individualfrom the intersection of the context set C and the nominal property P. The worldvariable w allows the referent of the individual selected by the choice function tovary across worlds, and w is existentially bound by either the utterance world or by31The difference between Squamish and Okanagan, as represented in (56) is probably not substan-tive, but rather Gillon (2006) is abstracting away from intensionality.156a higher modal operator. Under this analysis, Okanagan iP creates an expressionof type .I assume that the world variable may either be bound by a modal or otheroperator, resulting in an intensional reading, or at the level of discourse by beingidentified with the actual world, creating an essentially referential expression. Aderivation of a basic iP DP iP ylm?xw@m ?the chief? is shown in Figure 5.1.32Figure 5.1: Semantic Composition of an Okanagan iP DP, Example IDP?w[ f (?x[chie f (x)(w)?C(x)(w)])]D<>,>?P?w[ f (?x[P(x)(w)?C(x)(w)])]iPNP>?x?w.chie f (x)(w)Nylm?xw@m?the chief?There are four important factors which give rise to the range of readings seenwith Okanagan iP DPs:i. The level at which existential closure of the choice function variable occursii. Whether a higher intensional operator (e.g. a modal) shifts the interpreta-tion world, or whether the absence of such an operator allows the intensionto be applied to the actual worldiii. The number of individuals in the intersection between P and Civ. Whether a maximal or non-maximal individual is selected by the choice32Semantic composition of an iP DP containing a nominal prefixed by k?- ?to be? (cf. 55) isnecessarily more complex: since time variables are involved with such nouns, the denotation of iPmust also be modified to include times. Since this is a more general problem relating to the semanticsof determiners, cross-linguistically speaking, I abstract away from times in my definition of iP.157functionConcerning factor (i), I assume that existential closure of the choice functionvariable may occur at any level (Reinhart, 1997). This correctly derives both wide-and narrow-scope readings of iP DPs, and crucially differs from Matthewson?s(1999) analysis of Lillooet assertion-of-existence DPs, which must be existentiallyclosed at the highest level, i.e. the speaker?s discourse situation.Concerning factor (ii), if there is a higher intensional operator such as a modalwhich shifts the interpretation world, an intensional, non-referential interpretationof the iP DP will result. If there is no such higher operator, then the intension ofthe sentence is applied to the actual world, and a referential reading of the iP DPwill result. For iP DPs which involve a k?- prefixed NP (cf. 53), the intension ofthe individual denoted by the iP DP is applied to some possible world, regardlessof the presence or absence of a higher modal operator.Concerning factors (iii) and (iv), a set of schematic representations may behelpful (I abstract away from intensionality, but provide cross-referencing to rel-evant Okanagan examples next to the figure captions). The following diagramsrepresent discourse states: C stands for the set of contextually salient individuals,and P stands for a nominal property. In definite contexts, where a discourse ref-erent has already been established, the intersection of P and C includes only onesingular or plural individual x. The choice function associated with the determinermust select that maximal individual, as represented in Figure 5.2. In this case, thedefault maximality implicature, f = MAX, is satisfied.Figure 5.2: Domain Restriction in Definite Contexts (e.g. 36a)For restricted contexts involving multiple possible referents, if the determiner158does not select the entire set as a maximal plural individual, as in Figure 5.2, thenit must select one individual from the intersection of C and P, as in Figure 5.3. Insuch a context, a speaker may achieve either a specific interpretation, for instanceby pointing or using some other demonstrative gesture, or a non-specific interpre-tation, in which case the choice function selects any individual, and a narrow scopeinterpretation results. In these cases, the default maximality implicature, f = MAX,is cancelled.Figure 5.3: Domain Restriction in Non-Unique Contexts (e.g. 10)In some indefinite contexts, such as at the beginning of a text or in an existentialsentence, there is no contextual restriction of P, and P is a subset of C.33 Sincethere is no contextual restriction, the choice function may potentially select anyindividual in the domain of P. Once again, in these cases the default maximalityimplicature, f = MAX, is cancelled.33In other words, as part of the interlocutors? shared beliefs about individuals which inhabit theactual and possible worlds, C is equal to De in out-of-the-blue cases.159Figure 5.4: Domain Restriction in Indefinite and Existential Contexts (e.g.15 and 28)We see that Figures 5.3 and 5.4 are equivalent except that there is no domainrestriction in Figure 5.4. Figure 5.4 is a pre-requisite discourse state for a non-specific, existential interpretation of an iP DP. In existential sentences, the entiredomain of P is contextually salient (P ? C), a non-maximal singular or plural indi-vidual selected by a choice function may have an existential interpretation. Read-ings of iP DPs in existential sentences are equivalent to other non-specific indef-inite readings, except that the absence of domain restriction results in their beinginterpreted as referencing non-specific instantiations of a contextually unrestrictedset.Generic interpretations of iP DPs differ from other indefinite uses of iP DPsonly in the sense that an intensional maximal plural individual is chosen from acontextually unrestricted domain. Because iP may independently denote contex-tually unrestricted non-maximal pluralities (cf. section 3.2.7), and does not pre-suppose or assert maximality (cf. section 3.2), but allows maximal readings indefinite contexts (cf. section 3.2.5), it is logical that iP should also allow maximalreadings in contextually unrestricted contexts. In other words, since existentialquantification over plural individuals is necessary for plural existential readings,generic readings might arise from universal quantification over pluralities, or un-der a choice function analysis, selection of the maximal plural individual whichsatisfies a contextually unrestricted predicate (Figure 5.5). The default maximalityimplicature, f = MAX, is satisfied here.160Figure 5.5: Generic Interpretations of Okanagan DPs (e.g. 47)The interaction between domain restriction and whether the choice functionselects a maximal or non-maximal individual (factors iii/iv) yields a four-way split,shown as Table 5.2.Table 5.2: Set Intersection and Maximality with Okanagan iPmaximal non-maximalP ? C generic existential/non-restricted indefinite(Figure 5.5, ex.47) (Figure 5.4, ex.45)(P ? C) ? P definite restricted indefinite(Figure 5.2, ex.35-36) (Figure 5.3, ex.10)In sum, the domain restriction analysis of iP given as (56b) explains the ab-sence of any deictic features, the absence of any presupposition or assertion ofuniqueness/maximality, the availability of narrow-scope readings, and predicts theavailability of existential and generic readings.5.3.2 Okanagan tN. Mattina (2006, 126-128) states that Moses-Columbian obliques are ?semanti-cally oblique in the sense that they consistently show the partiality of reference ofEnglish some...? and are ?consistently nonspecific in their interpretations.? Mat-tina?s observation accurately describes quasi-objects in Okanagan as well.The oblique marker t does not employ domain restriction, which is predicted161under an analysis like Gillon (2009), since t is not a category D item (cf. chapter4). Since there is little evidence for a null determiner, I claim that the obliquemarker t, when it introduces a quasi-object of a formally intransitive predicate, isa semantically vacuous indicator that the NP is undergoing semantic incorporation(Van Geenhoven, 1998; Chung and Ladusaw, 2004; Farkas and de Swart, 2004;Carlson, 2006).34For Van Geenhoven (1998), semantic incorporation involves a nominal pred-icate of type (or > in an intensional setting) that functions as anargument of an incorporating verb. The incorporating verb introduces an exis-tentially bound, entity-denoting variable which is asserted to have the propertydenoted by the incorporated noun. For Okanagan, the semantics of incorporationcan be written directly into the denotation of the intransitivizer. For example, theintransitivizer suffix -@m may be represented as follows (minus world variables):35(57) [[-@m]] = ?P?Q?x?y[P(y)(x)?Q(y)]For (57), the intransitivizer -@m takes a transitive root P as its first argument,36 anda set-denoting quasi-object as its second argument Q, and then asserts that thereis some individual y that satisfies the property Q, and that this individual y corre-sponds to the thematic object of the predicate P. The property Q is thus a restrictivemodifier of the predicate P. The identity of the individual y is crucially not depen-dent on the context.37 A semantic derivation of the quasi-object containing VP34Themes of transitive applicatives (i.e. predicates which are ditransitivized by -x(i)t-) are alsointroduced by t. N. Mattina (1996b, 49) notes that both quasi-objects of middle intransitives (a.k.a.generic objects) and benefactive (a.k.a. dative) themes are ?generic?, or non-referential, and so it maybe possible to analyze these as incorporated objects.35The morpheme -@m also occurs with syntactically transitive predicates with possessor subjects.These predicates select for patient arguments introduced by iP, which means that the semantics givenin (57) cannot be extended to all occurrences of -@m. This issue could potentially be solved bypositing a homophonous suffix -@m for syntactically transitive predicates, though I remain agnosticon this issue here.36The root in Figure 5.6, ?kwu ?l must be represented as transitive, given (57). An analysis of someSalish roots as being transitive follows claims made by Gerdts (2006) and Gerdts and Hukari (2012),but contravenes Thompson and Thompson (1992) and others who claim that roots are fundamentallyintransitive, and Davis and Matthewson (2009), and references therein, for arguments that all Salishverb roots are unaccusative (Davis, 1997). There are several possible modifications which could bemade to (57) in order to make it consistent with the intransitive or unaccusative root hypotheses, butI abstract away from these issues here.37Readings resembling distributive readings are available for oblique quasi-objects in cases in-162?kw??l@m t pwm?n ?make a drum? is shown as figure 5.6:Figure 5.6: Semantic Incorporation of an Oblique Quasi-ObjectVP?x?y[made(y)(x)?drum(y)]VP?Q?x?y[made(y)(x)?Q(y)]??x?y[made(x)(y)]?kw??lINTR?P?Q?x?y[P(y)(x)?Q(y)]-@mKP?x[drum(x)]KtNP?x[drum(x)]pwm?nAssociating semantic incorporation with the intransitivizer (Van Geenhoven,1998) is preferred over analyzing t as having RESTRICT semantics (Chung andLadusaw, 2004) for the following reason: If t marked a RESTRICT operation, thenthe distribution of quasi-objects would not be expected to be limited to contexts in-volving formally intransitive predicates. In other words, the distribution of obliquequasi-objects/themes is dependent on the selectional restrictions of specific mor-phemes (cf. section 4.4.2). Okanagan t is thus different than the Squamish deter-miner kwi, which Gillon (2006, 10) analyzes as composing with a predicate viaRESTRICT (Chung and Ladusaw, 2004).Analyzing t as semantically vacuous has the added benefit of potentially al-lowing a unified analysis of other occurrences of t, specifically as a marker ofattributive modification (cf. section 6.2), where it is also plausibly semanticallyvacuous. The Lillooet non-assertion of existence determiner kwu has a similar dis-tribution to Okanagan t. Werle (2000) unifies the attributive and non-referentialDP object uses of kwu by analyzing DPs headed by kwu as of type <,>,i.e. they are predicate modifiers (Heim and Kratzer, 1998). Lillooet differs fromOkanagan, however, by the fact that kwu DPs can function as core objects of transi-volving a quantified subject DP. These cases can be reduced to distributivity over events associatedwith noun-incorporated predicates.163tive predicates in polarity or intensional contexts (cf. section 4.4.8), which meansthat the non-attributive distribution of kwu DPs cannot be captured simply by writ-ing incorporation semantics into the meaning of intransitivizers, as is the case forOkanagan.I suggest that there may be a split in Salish languages, whereby those languageswhich introduce quasi-objects with a determiner (e.g. Lillooet and Squamish) uti-lize semantically contentful determiners and predicate modification/RESTRICT se-mantics in such contexts (Werle, 2000; Gillon, 2006), while languages which in-troduce quasi-objects with an oblique marker (e.g. Okanagan) utilize instead asemantically vacuous oblique-marker and semantic incorporation. This view accu-rately reflects the narrower distribution of Okanagan t in comparison with Lillooetkwu.38 Finally, by not adopting Werle (2000) for Okanagan t, we are not forcedto analyze NPs as attributive modifiers of demonstratives in DP-DP structures andclefts (cf. chapters 7 and 8).5.3.3 Explaining the Implicature Carried by iPThe Okanagan determiner iP carries the maximality implicature given as (58b), andin argument contexts, where iP contrasts with t, it gives rise to the conversationalimplicature given as (58c):39(58) a. [[iP]] = ?P?w[ f (?x[P(x)(w)?C(x)(w)])]b. Maximality implicature:f = MAXc. Conversational implicature:Because the default for iP is f = MAX, a speaker?s use of an intransitiveconstruction with a quasi-object introduced by t carries a conversationalimplicature of non-maximality.38Maintaining an analysis of t as semantically vacuous furthermore explains its use as case mark-ing a passive agent (cf. section 4.4.3), and it is interesting to note that in Lillooet, an oblique markerP@ also marks a passive agent, rather than kwu, which supports analyzing kwu as encoding RESTRICTand P@ as semantically vacuous.39The determiner iP contributes a different conversational implicature in equative contexts, whereequatives stand in opposition to direct predications. See section 7.4.1.164The implicature in (58b) arises from an implicit assumption which speakershave that a context set C is non-trivial, and that there is some choice functionf which picks out some salient individual. In other words, since the pragmaticfunction of a context set C is to allow for contextual restriction of a domain ofindividuals P, the implicit assumption is that the contextual domain is restricted tothe maximum extent possible. Likewise, since choice functions pick out individu-als, the implicit assumption is that there is some individual within the contextuallyrestricted domain which the choice function picks out. In cases where C is equiv-alent to the universe and P is a subset of C, C will be trivial, and so the implicitassumption of non-triviality is overruled.In addition to the assumption of non-triviality, there is an assumption whichaccompanies the use of an iP DP such that the contextual domain is restricted to themaximum extent possible: i.e. the assumption is that no contextually non-salientindividuals are included in the intersection of P and C. The default f = MAX meansthat an iP DP will, by implicature, reference only individuals in the intersectionof P and C, and all individuals in the intersection of P and C. This in turn meansthat the size and set membership of the intersection of P and C will preferentiallyremain constant across uses of any particular iP DP in a given discourse context.If f 6= MAX, then the assumption that no contextually non-salient individuals areincluded in the intersection of P and C is overruled, as depicted in Figure 5.3.The size of the intersection is then decreased to include just that singular or pluralindividual selected by f, and at the next reference of this individual using an iP DP,the default f = MAX is restored. The default of f = MAX accounts for the uses of iPas a definite, however unlike in English, there is no presupposition involved sincein contexts where more than one salient individual is present in the context set (e.g.Figure 5.3), f = MAX cannot hold.Crucially, the iP determiner is sensitive to the context, as indicated by its se-mantic denotation (58a, cf. 56b), whereas the oblique marker t is semanticallyvacuous, and so cannot be sensitive to the context. Since the use of an obliquemarked quasi-object carries with it a conversational implicature of non-maximality(58c), the default expectation is that in cases for which a context set C will be triv-ial, the speaker will use an oblique marked quasi-object. In Gricean terms, this isa Quantity implicature, following the sub-maxim make your contribution as infor-165mative as is required. Non-maximal uses of iP in contexts for which C is trivial isless informative, in other words.The tight correlation in Okanagan between aspect, predicate transitivity, andthe selectional restrictions of a predicate mean that if it were the case that iP al-lowed only maximal readings, that formally transitive predicates would be categor-ically incompatible with non-maximal readings of nominal expressions. Allowingcancellation of the implicature in (58b) circumvents this problem, and the range ofpredicates which permit non-maximal readings of nominal expressions is therebyincreased. From the Salish perspective, languages with single-determiner systemslike Okanagan offer an interesting point of comparison, since allowing cancellationof the maximality implicature carried by iP essentially takes the place of there be-ing a secondary non-referential determiner, present in most other Salish languages.5.4 SummaryIn this chapter, I have investigated the semantic distribution of the determiner iPand the oblique marker t in Okanagan, and concluded that iP is best analyzed asa non-deictic pragmatically conditioned determiner (Gillon, 2006; D?chaine andTremblay, 2011), and that t is a semantically vacuous morphological reflex of se-mantic incorporation (Van Geenhoven, 1998).Okanagan iP shares some similarities with deictic determiners in Squamish(Gillon, 2006) and assertion-of-existence determiners in Lillooet, but allows for awider range of readings, including narrow-scope readings usually associated withnon-deictic and non-assertion-of-existence determiners (Matthewson, 1998). Thedeterminer iP does not presuppose or assert uniqueness or maximality or speci-ficity, and does not require a widest-scope interpretation. I have shown that iP iscontextually sensitive and permits co-referential readings, while t is not contextu-ally sensitive, and does not allow co-reference. I have claimed that iP carries animplicature of uniqueness/maximality via the fact that it can reference a contex-tually restricted domain, whereas t cannot. iP also permits intensional readings.The semantics of iP are given as (59a), the maximality implicature carried by iP isgiven as (59b), and the conversational implicature associated with oblique-markedquasi-objects is given as (59c).166(59) a. [[iP]] = ?P?w[ f (?x[P(x)(w)?C(x)(w)])]b. Maximality implicature:f = MAXc. Conversational implicature:Because the default for iP is f = MAX, a speaker?s use of an intransitiveconstruction with a quasi-object introduced by t carries a conversationalimplicature of non-maximality.The analysis in (59) is important for my analysis of DP-DP structures andclefts, both of which I claim are equative. These are discussed in chapters 7 and8, respectively. Before investigating equatives, I discuss nominal modification, in-cluding relative clauses, which are important for an understanding of clefts.167Chapter 6The Syntax of NominalModificationThis chapter discusses two types of nominal modification for Okanagan: attribu-tive modification, and relative clauses. Nominal modification has been examinedin other Salish languages: Straits Salish (Montler, 1993), Lillooet (Davis et al.,1997; Davis, 2002, 2004, 2010a, 2011), Shuswap (Davis et al., 1997), and Thomp-son (Koch, 2004, 2006). Other than Lyon (2010a), there has been no systematicinvestigations of nominal modification in Okanagan to date, although N. Mattina(1994) presents some relevant data on attributive modification, and H?bert (1982b,45,122) presents several examples of relative clauses.I show that attributive modifiers must be non-verbal predicates (i.e. adjectivalor nominal predicates) which modify a head noun, while relative clauses may beformed from both verbal and non-verbal modifiers. In this respect, Okanagan pat-terns similarly to the Northern Interior Salish languages Lillooet and Thompson.The distribution of iP and t in nominal modification contexts in Okanagan is in-dicative of relative clause versus attributive modification, and the presence versusabsence of clausal structure associated with the modifier.The structures motivated by these data are important for an understanding ofthe arguments employed in chapter 8, specifically: (i) the syntax of cleft residueclauses and the relation between relative clauses and cleft residues; and (ii) the dis-tinct behavior of oblique t in cleft environments, which must be kept separate from168its use as an attributive marker or case-marker introducing an oblique argument.After presenting the problem of distinguishing different types of nominal mod-ification in terms of the morphosyntactic distribution of iP and t (section 6.1.1),I go on to distinguish nominal modification from clausal subordination (section6.1.2). I then discuss attributive modification in some detail (section 6.2), beforeturning to relative clauses (section 6.3).6.1 Preliminaries6.1.1 Morphological Patterns of Nominal ModificationConsider the following data:(1) a. w?k-@nsee-[DIR]-1SG.ERGiPDET?kw@c ?kw?ctstrongtATTRylm?xw@m.chiefI saw the strong chief.b. w?k-@nsee-[DIR]-1SG.ERGiPDET?kw@c ?kw?ctstrongiPDETylm?xw@m.chiefI saw the strong chief.The data pair in (1) seem to be semantically equivalent, differing only in whetherthe oblique marker t or the iP determiner introduces the head noun.1 The questionI pose is whether or not the morphosyntactic difference illustrated here also indi-cates a deeper syntactic distinction. I argue that it does, and will present evidencethat (1a) involves attributive modification (i.e. not-necessarily-intersective modifi-cation) on analogy with similar constructions in the Northern Interior, while (1b)involves relative clause modification.21I gloss the oblique marker t as ATTR ?attributive? in (1a), rather than simply OBL ?oblique?because its function here is to mark attributive modification, and the use of t in this environmentshould be kept separate from its use in introducing quasi-objects. The question of whether these twoseparate uses of t correspond to two distinct but homophonous morphemes, or whether these twoseparate uses may receive a unified analysis, has yet to be determined.2There are six logically possible surface patterns involving determiner iP and oblique markert in their capacity of introducing heads and modifiers in nominal modification structures, with anadditional dimension of variation being whether the head precedes the modifier or vice versa. Atable showing the surface patterns of iP and t in modification contexts is given in section 6.5.169There are other modification patterns worth examining as well: an alternationsimilar to that shown for (1) is not allowed for quasi-objects of intransitive predi-cates (2). This is surprising given that in neither transitive (1b) nor intransitive (2b)is there a direct selectional relation between the transitivity of the main predicateand the determiner which introduces the head noun of its complement.(2) a. kn1SG.ABSw?k-@msee-MIDtOBL?kw@c ?kw?ctstrongtATTRylm?xw@m.chiefI saw a strong chief.b. *kn1SG.ABSw?k-@msee-MIDtOBL?kw@c ?kw?ctstrongiPDETylm?xw@m.chiefI saw a strong chief.Additionally, data similar to (3a, cf. 1a) show that a nominal head may precedeor follow the modifier, showing that pre-nominal and post-nominal modificationare both possibilities. In (3b), an adjectival modifier follows a head noun.(3) a. w?k-@nsee-[DIR]-1SG.ERGiPDET?kw@c ?kwactstrongiPDETtk?milxw.womanI saw the strong woman.b. w?k-@nsee-[DIR]-1SG.ERGiPDETtk?milxwwomaniPDET?kw@c ?kwact.strongI saw the woman that is strong.Before discussing data like these in detail (relative clauses in particular), it isimportant to distinguish nominal modification from clausal subordination, sincesuperficially at least, both consist of predicative material introduced by functionalparticles. For Okanagan, I show that clausal subordination utilizes a different setof particles than nominal modification, and that on this basis alone, the two classesmay be distinguished.1706.1.2 Distinguishing Clausal Subordination from NominalModificationSubordination in Okanagan takes a variety of forms, only some of which I discusshere. My goal is simply to show that the determiner iP and oblique marker t neverplay the role of complementizer. This fact contrasts with neighboring Thompson,where the present e and irrealis k determiners, as well as the oblique marker t, mayall occur in complementizer positions (Kroeber, 1999, 207-211).Complementizers in Okanagan are largely optional, unlike in Northern InteriorSalish languages (Kroeber, 1999). One sometimes finds ? or ?aP, which A. Mattina(1973, 114) describes as being ?sequential complement particles?. (4) was volun-teered without a complementizer, but was judged good with a complementizer.(4) n-??pt-@m-@nforget-[MIN-DIR]-1SG.ERG(?)(COMP)i-ks-kwan-?m1SG.POSS-FUT-take-MIDin-qw?cqn.1SG.POSS-hatI forgot to get my hat.Factive complements also allow but do not require a ? complementizer, as shownin (5a) and (5b). An iP determiner is not allowed in this position.3(5) a. x?astgoodny ?Qipalwayskwu1PL.ABS(?)/*iP(COMP)/*DETk[s]-swit-m?stFUT-try-INTR.REFLEXiP(DET)lLOCn-qw@lqw?l-t@n-t@t.n-speak-INSTR-1PL.POSSIt is good that we?re trying to save our language.b. x?astgoodkwu1PL.ABS(?)/*iP(COMP)/*DETc-paP-paPs-?lxCUST-RED-think-DEViPDET?klLOCks-c-xw?y-t@t.FUT-CUST-go-1PL.POSSIt is smart to think about the future.3A. Mattina (1973, 114) states that when the complementizer intervenes between a proclitic andits host predicate, the proclitic subject apparently has a focused reading. However, I have not detectedany obvious focus-sensitive interpretation.171I assume a null complementizer for cases similar to (4-5) where a ? or ?aP comple-mentizer is possible, but not overt. This null complementizer cannot automaticallybe assumed to be a null version of ? or ?aP, however. This is because comple-ments of propositional attitude predicates disallow the complementizer ? (6), yetalso presumably involve clausal subordination.(6) a. kn1SG.ABSn-st?ls-@mn-think-MID*?/*t/*iP*COMP/*OBL/*DETx?m?nk-@mwant-MIDtOBLk?-citxw-s.U.POSS-house-3SG.POSSI thought he wanted a houseb. kn1SG.ABSn-stilsn-think*?/*t/*iP*COMP/*OBL/*DETks-m@qw-qw-m?xaPx.FUT-snow-FRED-INCEPTI think it?s going to snow.The data in (6) also illustrate how the oblique particle t does not function as acomplementizer in Upper Nicola Okanagan propositional attitude complements,unlike in Thompson or other dialects of Okanagan (Kroeber, 1999, 233).Causal (7) and conditional clauses (8) are optionally introduced by the loca-tive particle tl ?from? (Kroeber, 1999, 236). Unlike with prepositional phrases (seechapter 4), a determiner may not precede the locative marker in subordination con-texts (7).4,5(7) kn1SG.ABS?t@ ?kw-nc?tlay.down-REFLEX(*iP)(*DET)tlLOCi-s-P?yx?wt.1SG.POSS-NOM-tiredI laid down because I was tired.(8) c@ ?mEPIS??awtgo.outiPDETs-c-wa?rNOM-CUST-fire(tl)(LOC)lutNEGkw2SG.ABS?aPCOMPw?r?s@m.build.the.fireThe fire will go out if you don?t put on more wood.4Thompson introduces causal complements with oblique t and determiner e (Kroeber 1999, 210).5It is unclear whether tl is syntactically a complementizer in these cases, or a preposition.172Locative tl also functions as a factive complementizer (9).6(9) a. x?astgoodtlLOC?-xwuy-st-sreturn-go-CAUS-3SG.ERGiPDETtu ?m-s.mother-3SG.POSSIt?s good that she took her mother home.b. kn1SG.ABSqw@l-c[n]-nc?tspeak-mouth-REFLEXtlLOCn-??pt-@m-@nn-forget-[MIN-DIR]-1SG.ERGin-lakl?.1SG.POSS-keyI was angry at myself because I forgot my keys.In addition to tl (7-8), Upper Nicola speakers also use a form kwaP as a causalcomplementizer (cf Kroeber (1999, 354) and A. Mattina (1985, sz.421)).7(10) a. kn1SG.ABS?t@ ?kw-nc?tlay.down-REFLEXkwaPCOMPkn1SG.ABSs-Payx?wt.NOM-tiredI laid down because I was tired.b. ?t@ ?kw-nc?tlay.down-REFLEXkwaPCOMPPi?-seat-[DIR]-3SG.ERGyaQy?QtalliPDETlas?p.soupShe laid down because she ate all the soup.The ? complementizer is required in Upper Nicola Okanagan for interrogative com-plements (11).(11) a. kwu1SG.ABSs?w-@ntask-DIR?COMPi-ks-xwuy.1SG.POSS-FUT-goHe asked me if I was going to go.6In Lillooet, factive complements are marked by the determiner ti...a. (Henry Davis, p.c.)7kwaP and tl seem interchangeable in some sentences, but not in others. It is unclear to me howthe two differ semantically.173b. s?w-enask-[DIR]-1SG.ERGi[n]-s-n-ylm?xw@m1SG.POSS-NOM-n-chief?COMPi-k[s]-s@xw- ?maP ?may?P-m.1SG.POSS-U.POSS-OCC-teach-MIDI asked my boss if I could be the teacher.The future marker mi also appears to function as a complementizer (12), oftenin conjunction with the adverbial sic ?new? (12b), yielding a meaning equivalent to?before? in English:8(12) a. kn1SG.ABSks-l?mt-aPxFUT-glad-INCEPTmiFUTSarahSarahm@ ?q-?nk.full-stomachI will be glad when Sarah is full.b. lutNEGny ?Qip!always??lapstop?aPCOMPc-?tQap-?mCUST-shoot-MIDmiFUTsicnewxwuy-st-xw!go-CAUS-2SG.ERGNever! The shooting has got to stop before you can take her!Another complementizer that should be mentioned is kiP, which is used only inadjunct-focused clefts and WH-questions (Baptiste, 2001, 16-17). Neither iP nor tmay substitute for kiP in these environments.9(13) a. tlLOC?k?Palqwacross.the.linekiP/*iP/*tCOMP/*DET/*OBLkn1SG.ABSs-c-xwuy-x.NOM-CISL-go-INTRIt?s from the U.S.A. that I came.b. ?kaPkinwherekiP/*iP/*tCOMP/*DET/*OBLw?k-@nt-xwsee-DIR-2SG.ERGiPDETsk@kQ?kaP?birdsWhere are the birds you saw?8In chapter 7, I present the hypothesis that future mi may be a tense head in some instances, butmay raise to C. mi also introduces residue clauses in clefts.9Clefts involving kiP are discussed in detail in chapter 8, section 8.9. For the moment, it isimportant to note that kiP, unlike iP or t, cannot introduce simple NPs, which disqualifies it frombeing a determiner.174In sum, the determiner iP and oblique marker t are notably absent from the in-ventory of particles introducing subordinate clauses in Okanagan, though both areused in nominal modification contexts.10 Because the complementizers discussedin this section cannot introduce NPs in argument position, we can use the com-plementary distribution of clausal subordinators and NP-introducing articles as adiagnostic to distinguish relative clauses and attributive modification from otherstructures involving clausal subordination in the Upper Nicola dialect (Table 6.1).Table 6.1: NP-introducing Articles versus Clausal SubordinatorsIntroduces ClausalNP in argu-ment positionSubordinatoriP X *t X *?(aP) * Xtl * XkwaP * Xmi * XkiP * XI now move onto a discussion of attributive modification.6.2 Attributive ModificationAttributive modification in Okanagan is used in two related syntactic construc-tions. In predicate position, attributively modified nouns form complex nominalpredicates (CNPs) (Davis et al., 1997) (see relevant discussion in section 4.2), andin argument position, a CNP may form a constituent with an introductory deter-miner iP or oblique marker t, thereby forming what I refer to as complex DPs andcomplex obliques, respectively.10Montler (1993, 253) also notes that clausal attributives (a.k.a. relative clauses) in Saanich areclearly distinguishable from other forms of clausal subordination, however there is a different prob-lem in Saanich: Since a determiner does not introduce the modifying clause, it becomes impossibleto tell whether the modifying clause is in fact a non-subordinated, separate sentence.175Attributive modifiers cannot project tense/aspectual structure, and there are se-mantic restrictions such that individual-level predicates (i.e. permanent propertiesof individuals (Carlson, 1977)), and predicates which are derived by the custom-ary/stative prefix ac- can be modifiers, whereas stage-level predicates (i.e. prop-erties true of an individual for a temporal stage (Carlson, 1977)), eventive unac-cusatives, unergatives, and transitive predicates may not.11,12 This is one distin-guishing factor between attributive and relative clause modification.Before beginning, it is important to make clear that the marker t which oc-curs in contexts involving attributive modification is an element distinct from themarker t which introduces oblique arguments of formal intransitives, as discussedin chapter 4. The attributive marker t does not reflect the syntactic or semantic re-strictions of any selecting predicate, and never co-occurs adjacent to iP. To reflectthis distinction, I gloss attributive t as ?ATTR? rather than ?OBL?. Like oblique t,however, it seems clear that attributive t is semantically vacuous.136.2.1 Review: Complex Nominal Predicates and Complex DPsAs mentioned in section 4.2, attributive t links an adjectival (or nominal) modi-fier to a head nominal in complex nominal predicate (CNP) structures (14) andattributively-modified, complex DPs (15) (Davis et al., 1997; Lyon, 2010a). Struc-tures analogous to (14-17) are also found in Shuswap and Lillooet (Davis et al.,1997), as well as in Thompson (Koch, 2006).14 The oblique marker is obligatory11As mentioned in a footnote in chapter 4, I use the terms ?unaccusative? and ?unergative? descrip-tively to distinguish intransitive predicates which take experiencer DP arguments from those whichtake agentive DP arguments, without making any deeper syntactic claims concerning unaccusativity.12It may be the case that customary/stative ac- is an event-variable saturator, which correlatessyntactically with the absence of tense/aspectual structure for ac- prefixed states; however I do notinclude the details of such an analysis here (cf Koch (2006)).13See section 5.3.2 for analysis of quasi-object introducing t as semantically vacuous, and discus-sion of how all uses of t may be analyzed as semantically vacuous. It is tempting to analyze t asalways indicative of predicate modification, however t also intervenes between demonstratives andnominals in equative contexts (cf. chapter 8), and I argue that demonstratives are not predicative.14In Thompson, the attributive marker t may co-occur with an irrealis determiner k in attributivemodification contexts (Koch, 2006). Lillooet contrasts with Shuswap, Thompson, and Okanagan byhaving no oblique marker in this environment, but only an optional, irrealis determiner ku. Assumingthat the Thompson pattern reflects an earlier stage of Interior Salish, this implies both that oblique thas been lost in this environment in Lillooet (Henry Davis, p.c.), and that the k-determiner has beenlost from this particular environment in Shuswap. Okanagan, perhaps taking a similar, localized176in both CNPs and complex DPs:15(14) a. [s?lxwaPbigtATTRsq@ltm?xw CNP]man[iPDETylm?xw@m. DP]chiefThe chief is a big man.b. *[s?lxwaPbigsq@ltm?xw CNP]man[iPDETylm?xw@m.DP]chiefThe chief is a big man.(15) a. [w?k-@n V P]see-[DIR]-1SG.ERG[iPDETs?lxwaPbigtATTRylm?xw@m. DP]chiefI saw the big chief.b. *[w?k-@n V P]see-[DIR]-1SG.ERG[iPDETs?lxwaPbigylm?xw@m. DP]chiefI saw the big chief.The attributive modifier must precede the nominal head (16-17):(16) a. [x?astgoodtATTRtk?milxw CNP]womaniPDETylmixw@m-t@t.chief-1PL.POSSOur chief is a good woman.b. *[tk?milxwwomantATTRx?ast CNP]goodiPDETylm?xw@m-t@t.chief-1PL.POSSOur chief is a good woman.(17) a. [c@c? ?maPtsmall(PL)tATTRsk@kQ?kaPCNP]birdsi-s-c-w?k.1SG.POSS-NOM-CUST-seeThe ones I saw were small birds.b. *[sk@kQ?kaPbirdstATTRc@c? ?maPt CNP]small(PL)i-s-c-w?k.1SG.POSS-NOM-CUST-seeThe ones I saw were small birds.reduction to its logical conclusion, now has no k-type determiner anywhere in its grammar.15According to (N. Mattina 1994, 5), some dialects of Okanagan allow attributive t (or the deter-miner) to be absent in these contexts. The Upper Nicola dialect does not allow this.177Modifiers within CNPs and complex DPs may also be nominal, rather thanadjectival:(18) a. [[s@xw- ?maP ?m?yaP-m NP]OCC-teach-MIDtATTRp@ptw?naxw CNP].old.womanThe teacher is an old lady.Literally: She is a teacher old lady.b. w?k-@nsee-[DIR]-1SG.ERGiPDET[[tk?milxw NP]womantATTR?qwQay-lqs CNP].black-robeI saw a woman minister.Also recall from chapter 4 that the determiner iP predictably introduces a coreargument of a transitive main predicate. As such, complex iP DPs cannot serve asquasi-objects of formally intransitive predicates (19a), or as theme arguments ofditransitives (20a). Speakers will instead either correct these to complex obliqueforms (19b,20b), or change the transitivity of the main predicate to agree with theiP determiner (19c,20c).(19) a. *kn1SG.ABSw?k-@msee-MIDiPDETqwQaybluetATTRs ?wa?r? ?kxn.frog.I saw a blue frog.b. kn1SG.ABSw?k-@msee-MIDtOBLqwQaybluetATTRs ?wa?r? ?kxn.frog.I saw a blue frog.c. w?k-@nsee-[DIR]-1SG.ERGiPDETqwQaybluetATTRs ?wa?r? ?kxn.frog.I saw a blue frog.178(20) a. *xw? ?c-xt-@m-@ngive-BEN-2SG.ACC-1SG.ERGiPDET?c ?w- ?ca ?wtRED-cleantATTRlasm?st.shirtI gave you a clean shirt.b. xw? ?c-xt-@m-@ngive-BEN-2SG.ACC-1SG.ERGtOBL?c ?w- ?ca ?wtRED-cleantATTRlasm?st.shirtI gave you a clean shirt.c. xw? ?c-?t-@m-@ngive-APPL-2SG.ACC-1SG.ERGiPOBL?c ?w- ?ca ?wtRED-cleantATTRlasm?st.shirtI gave you a clean shirt.In other words, the initial particle of an attributively modified argument expressionreflects the selectional restrictions of a higher predicate.6.2.2 Aspectual Restrictions on Attributive ModificationDemirdache and Matthewson (1995) and Davis et al. (1997) show that modifiersin CNPs must be individual-level predicates in Lillooet and Shuswap. OkanaganCNPs (21-22) pattern similarly to these Northern Interior Salish languages; how-ever, complex DPs are less stringent, since they do allow stage-level, unaccusativepredicates as modifiers (23) (cf Davis (2011) for a similar finding in Lillooet):(21) a. t@?-t??tRED-straighttATTRylm?xw@m.chiefHe?s a straightforward chief.b. iPDETt@?-t??tRED-straighttATTRylm?xw@mchiefk- ?mylt-@m-sk-visit-MIN-[DIR]-3SG.ERGiPDETp@ptw?naxw.old.womanThe straightforward chief visited the old woman.(22) a. ?t?qw@lqwtalltATTRx?xwt@mlittle.girlSusy.SusySuzy is a tall girl.179b. iPDET?t?qw@lqwtalltATTRsq@ltm?xwmanQac?m.tie-MIDThe tall man is tying (things).(23) a. *qw? ?m-@ ?mfrightened-FREDtATTRsq@ltm?xw.manHe is a frightened man.b. iPDETqw? ?m-@ ?mfrightened-FREDtATTRsq@ltm?xwmank- ?mylt-@m-sk-visit-MIN-[DIR]-3SG.ERGiPDETp@ptw?naxw.old.womanThe frightened man visited the old woman.Comparing the ungrammatical (a) cases with grammatical (b) cases below, wesee that by prefixing customary/stative ac- to a stage-level modifier, the sentencebecomes grammatical.16,17 (26c) confirms that stage-level modifiers without ac-are acceptable as modifiers within a complex DP structure (cf 23b).(24) a. *tal?PverypaP-paPs-?lxRED-feel.bad-DEVtATTRsq@ltm?xw.manHe?s a worried man.b. tal?Pveryc-paP-paPs-?lxCUST-RED-feel.bad-DEVtATTRsq@ltm?xw.manHe?s a worried man.16See A. Mattina (1993a) and N. Mattina (1996b) for discussion of the functions of (a)c- as a?customary/habitual? marker, and as a ?stative? marker. The ?customary/habitual? intepretation of(a)c-prefixed predicates is possible with unergatives (cf. 32 below) and other eventive predicates, butnot a stative interpretation, which is dependent on the predicate being a non-eventive unaccusative.It is unclear whether the stative and customary/habitual uses of (a)c- follow from these uses corre-sponding to two separate yet homophonous morphemes, or whether these two uses might be unifiedsemantically.17The stage versus individual-level status of the predicate modifiers in this section is confirmed bytheir ability/inability to occur with ?short-time-span? versus ?longer-time-span? adverbials, respec-tively. For reasons of space, I do not include these data here.180(25) a. *n- ?kw-p-ilsn-gone-MUT-thoughts(ix?P)(DEM)tATTRsq@ltm?xw.manThat?s a lonely man.b. c-n- ?kw-p-ilsCUST-n-gone-MUT-thoughts(ix?P)(DEM)tATTRsq@ltm?xw.manThat?s a lonely man.(26) a. *kn1SG.ABSQimtangrytATTRsq@ltm?xwmanI?m an angry man.b. tal?Pverykn1SG.ABSc-QimtCUST-angrytATTRsq@ltm?xw.manI?m an angry man.c. n?x?l-m-@nhear-MIN-[DIR]-1SG.ERGiPDETQimtangrytATTRsq@ltmixwmanI heard the angry man.Adding ac- to an individual-level predicate is ungrammatical (27-28):(27) a. cax?redtATTRlasm?stshirtiPDETxwi ?c-xt-xw.give-BEN-2SG.ERGWhat you gave him was a red shirt.b. ?t?qw@lqwtalltATTRtk?milxwwomaniPDETs-c-wik-s.NOM-CUST-see-3SG.POSSHe saw a tall woman.c. n-x?@l-x?l-??n-scared-RED-verytATTRsq@ltm?xw.manThat?s a (characteristically) scared man.(28) a. *ac-c?x?CUST-redtATTRlasm?stshirtiPDETxwi ?c-xt-xw.give-BEN-2SG.ERGWhat you gave him was a red shirt.181b. *ac-?t?qw@lqwCUST-talltATTRtk?milxwwomaniPDETs-c-wik-s.NOM-CUST-see-3SG.POSSHe saw a tall woman.c. *ac-n-x?@l-x?l-??CUST-n-scared-RED-verytATTRsq@ltm?xw.manThat?s a (characteristically) scared man.In sum, adding customary/stative ac- to a stage-level unaccusative predicate seemsto allow that predicate to function as an individual-level predicate, which in turnallows it to modify a nominal predicate within a CNP structure. If we analyze ac-as converting a stage-level predicate to an individual-level predicate, or somethingsemantically similar (cf. Koch (2006) for Thompson), then Davis et al?s (1997)analysis of CNP modifiers in Lillooet and Shuswap may for all intents and purposesbe extended to Okanagan.18Unergatives and eventive unaccusatives are disallowed as attributive modifiers.Koch (2006, 149) discusses a similar finding for Thompson. In these cases, speak-ers will usually correct to a pattern where iP introduces both the head and the mod-ifier. Variable head-modifier ordering is allowed here, which I take to be evidencethat these are relative clause modifications. This is justified in section 6.3.(29) a. *kn-x?t-@nhelp-BEN-1SG.ERGiPDET?t ?kw-a ?kwfell-FREDtATTRt@tw?t.boyI helped the boy who fell down.b. kn-x?t-@nhelp-BEN-1SG.ERGiPDET?t ?kw-a ?kwfell-FREDiPDETt@tw?t.boyI helped the boy who fell down.c. kn-x?t-@nhelp-[DIR]-1SG.ERGiPDETt@tw?tboyiPDET?t ?kw-a ?kw.fell-FREDI helped the boy who fell down.18The difference being that the notion of ?intersective? vs. ?non-intersective? modifiers does notseem to be relevant to an Okanagan speaker?s judgements of attributive modifications.182(30) a. *kn-x?t-@nhelp-BEN-1SG.ERGiPDETx?nnumthurttATTRx?xwt@m.little.girlI helped the girl who got hurt.b. kn-x?t-@nhelp-BEN-1SG.ERGiPDETx?nnumthurtiPDETx?xwt@m.little.girlI helped the girl who got hurt.c. kn-x?t-@nhelp-BEN-1SG.ERGiPDETx?xwt@mlittle girliPDETx?nnumt.hurtI helped the girl who got hurt.(31) a. *w?k-@nsee-[DIR]-1SG.ERGiPDET?txwtilxflytATTRsk@kQ?kaP.birdsI saw the flying birds.b. w?k-@nsee-[DIR]-1SG.ERGiPDET?txwtilxflyiPDETsk@kQ?kaP.birdsI saw the flying birds.c. w?k-@nsee-[DIR]-1SG.ERGiPDETsk@kQ?kaPbirdsiPDET?txwtilx.flyI saw the flying birds.(32) a. *w?k-@nsee-[DIR]-1SG.ERGaPDETc-?@?tp-m@-nc?tCUST-jump-MIN-REFLEXtATTRx?xwt@m.little.girlI saw the little girl who jumped.b. w?k-@nsee-[DIR]-1SG.ERGaPDETc-?@?tp-m@-nc?tCUST-jump-MIN-REFLEXiPDETx?xwt@m.little.girlI saw the little girl who jumped.c. w?k-@nsee-[DIR]-1SG.ERGiPDETx?xwt@mlittle.girlaPDETc-?@?tp-m@-nc?t.CUST-jump-MIN-REFLEXI saw a little girl who jumped.Semantically transitive possessor predicates (33) and formally transitive pred-183icates (34) are also uniformly ungrammatical as attributive modifiers, as shown bythe (a) examples below. Speakers correct examples of this type to what I claim is ahead-final relative clause configuration (b cases).(33) a. *Pi?-@neat-[DIR]-1SG.ERGiPDETs-c-x?? ?w-s@lxNOM-CUST-dry-3PL.POSStATTRq?qxw@lx.fishI ate the fish that are drying.b. Pi?-@neat-[DIR]-1SG.ERGiPDETs-c-x?? ?w-s@lxNOM-CUST-dry-3PL.POSSiPDETq?qxw@lx.fishI ate the fish that are drying.(34) a. *w?k-@nsee-[DIR]-1SG.ERGiPDET?tQap-@nt-?s@lxshoot-DIR-3PL.ERGtATTRsk@kQ?kaP.birdsI saw some birds that they shot.b. i-s-c- ??aP ??aP-?m1SG.POSS-NOM-CUST-look.for-MIDiPDET?tQap-@nt-?s@lxshoot-DIR-3PL.ERGiPDETsk@kQ?kaP.birdsI?m looking for some birds that they shot.In sum, attributive modification may be defined as a strictly head-final config-uration, where the modifier is a non-clausal, non-eventive adjectival or nominalelement, and the nominal head is introduced by ?attributive? t.The distribution of modifiers in CNPs and complex DPs provides a language-internal diagnostic for distinguishing: (i) individual-level versus stage-level states,and (ii) states from unergatives and more complex, eventive types of predicates.The pattern is summarized in Table 6.2.184Table 6.2: Distribution of Predicate Types in Attributive ModificationsPredicate Class CNPmodiferw/o ac-CNPmodifierw/ ac-complexDPmodifiersemantic effect ofadding ac-I-level states X * X -S-level states * X X I-level stateother predicates * * * customary/habitualAlthough eventive unaccusatives, unergatives, and thematically/formally tran-sitive predicates cannot occur as attributive modifiers, the data in this section haveshown that they may occur as modifiers if the head is introduced by the determineriP, rather than attributive t. Additionally, cases where the head is introduced by iPwere shown to exhibit variable head-modifier ordering. I suggest that these twocharacteristics, either separately or in tandem, are indicative of a relative clausestructure, and will have more to say on this in following sections, but first, I dis-cuss the syntax of attributive modification.6.2.3 Syntax of Attributive ModificationFollowing Koch (2006) for Thompson, I suggest that attributive t may be under-stood as an overt indicator that predicate modification (Heim and Kratzer, 1998)is occurring; a morphological reflex of sorts, devoid of semantic content.19 Just asin English, predicate modification is a covert operation in Okanagan, but unlike inEnglish, an overt indication of the operation (i.e. t) is required. A compositionalrepresentation of the complex nominal predicate in (14a) is shown below as Figure6.1:19This is similar to my analysis of quasi-object t as a reflex of semantic incorporation in chapter 5.As mentioned in a previous footnote in this chapter, it is tempting to analyze t as always indicative ofpredicate modification, however t also intervenes between demonstratives and nominals in equativecontexts (cf. chapter 8), and I argue that demonstratives are not predicative.185Figure 6.1: Complex Nominal PredicateNP?x.[big(x)?man(x)]AP?x.big(x)As?lxwaPtNP?x.man(x)Nsq@ltm?xwIt is relatively easy to show that attributive t associates with the modifying AP,rather than the head NP.20 Evidence comes from data involving modifier stacking,such as (35). The structure I propose for (35) is shown as (6.2).(35) n- ??x?-cinn-loud-mouthtATTR?kw- ?kwy?maPIRED-littletATTRautomobil.automobile(That?s a) loud little car.Figure 6.2: Complex Nominal Predicate: Modifier StackingNPAPAn ??x?cintNPAPA?kwy ?kwy?maPtNPNautomobileAn alternative to Figure 6.2 associates t with the head nominal. For (35), thisalternative implies that a predicate adjective might select for either an AP (e.g.20There is no evidence that t is a constituent of a DP structure in these cases, since an (overt atleast) determiner cannot co-occur with t in this environment, unlike the case for Thompson whichallows oblique t and determiner k sequences to occur between an attributive modifier and head.186?kw ?kwy?maP ?little?) or an NP complement (e.g. automobil ?car?), so long as thecomplement is introduced by t, but we then lose the generalization that the finalelement of a CNP must always be a noun. In other words, the alternative predictsthat n ??x?cin t ?kw ?kwy?maP ?that?s a loud little? should be grammatical, but it is not.In any case, attributive t is semantically vacuous: It neither changes a predicateNP into an argument, nor licenses an NP to function as a predicate.6.2.4 Summary of Attributive ModificationThis section has shown that there are semantic constraints on the modifying con-stituent in an attributive modification: Only predicates that either are individual-level predicates already (e.g. t@?t??t ?straight/true?), or have been coerced into suchpredicates by adding customary/stative ac- (e.g. qw?m@ ?m ?frightened?) may occurin this position. I cautiously suggest that the category of adjectives in Okanagancomprises just that class of basic and derived lexical items that may occur as mod-ifiers, but not heads, in complex DP structures.21The ungrammaticality of eventive unaccusatives, unergatives, and transitives asattributive modifiers supports an analysis whereby constructions for which they canmodify a nominal are structurally distinct from attributive modifications. I suggestthat this distinction corresponds to a structural distinction between attributive andrelative clause modification, to which I now turn.6.3 Relative ClausesNominal modifications involving predicates other than non-eventive unaccusativesfollow morphosyntactic patterns distinct from those outlined in the previous sec-tion for attributive modification. I claim that these involve relative clause modifica-tion. I begin first with some basic properties and some surface-level characteristicsof relative clauses, before presenting evidence that Okanagan relative clauses areformed by movement of a DP to the left periphery of the relative clause (cf. Davis(2004, 2010a) for Lillooet and Koch (2006) for Thompson.)21See Davis (2011) and Koch (2006) for discussion of this issue in Lillooet and Thompson, re-spectively.1876.3.1 Introducing Relative ClausesI adopt the syntactic definition of a relative clause given in Kroeber (1999, 252):?Relative clauses are clauses that contain gaps that are co-referent with an NP ina higher clause, and the relative clause serves to restrict the type of entity denotedby the matrix NP.? By way of example, the bracketed, ergative-inflected relativeclause iP w?k@n ?that I saw? restricts the bird under discussion in (36a), and thenominalized relative clause [iP] isc ?kw??l ?the (one) that I made? restricts the type ofshirt under discussion in (36b).(36) a. ?tQ?p-nt-?nshoot-DIR-1SG.ERGiPDETsk@kQ?kaPbird[iPDETw?k-@n].see-[DIR]-1SG.ERGI shot the bird that I saw.b. iPDETlasm?stshirt[[iP][DET]i-s-c- ?kw??l]1SG.POSS-NOM-CUST-makes@?l-m?-n.lose-MIN-[DIR]-1SG.ERGThe shirt which I made, I lost.I follow Heim and Kratzer (1998, 65) for English (and Koch (2006) for Thomp-son) in assuming the following semantic distinction between relative clauses andattributive modification: A relative clause by definition utilizes a rule of predi-cate modification in addition to lambda-abstraction of the variable correspondingto a WH-gap, whereas attributive modification simply involves a rule of predicatemodification. Syntactically, lambda-abstraction is induced by A? extraction of aconstituent, leaving a gap. Attributive modification is analogous to adjectival mod-ification in English.Keenan (1985) states that an important feature of a true relative clause is thatit has to be a unique grammatical construction. For Okanagan (and the rest of theSouthern Interior), a relative clause is not identifiable by special inflectional mor-phology on the clausal modifier (Kroeber, 1999, 272), but instead by the followingcriteria:188a. In the case of a thematically transitive modifier, a relative clause structureis identifiable by an iP determiner and/or t oblique marker which precedesthe modifier, and by the availability of (and preference for) head-initialmodification.b. In the case of an intransitive modifier, a relative clause structure is identifi-able by an iP determiner or t oblique marker which precedes the head, andby the availability of a head-initial structure.Under these criteria, the following are all examples of relative clauses. (37a) and(37d) have semantically transitive modifying clauses, while (37b,c) both have in-transitive modifying clauses.(37) a. wa ?yyesc-my-st-inCUST-know-CAUS-1SG.ERGiPDETsq@ltm?xwmaniPDETks- ?kwu?l-?xw-m-sFUT-make-house-MID-3SG.POSStOBLcitxw.houseI know a man who can build you a house.b. wa ?yyeskaPk?c-@nfind-[DIR]-1SG.ERGiPDETs@nk? ?caPsq?x?aPhorseaPDETc-yalt.CUST-run.awayI found the horse that ran away.c. kn1SG.ABSw?k-@msee-MIDtOBLsk@kQ?kaPbirdstOBLc@-c? ?maPt.RED-small(PL)I saw some birds that were small.d. kn1SG.ABSs-c- ?kwu?l-xNOM-CUST-make-INTRtOBLy?mx?waPcedar.bark.baskettOBLks-yaP-y?Px?aP-s@lx.FUT-IRED-show-3PL.POSSI?m making a basket that they will show.None of the examples in (37) can be analyzed as attributive modifications, becausethey are not head-final modifications.2222Head-final variants of all four of these sentences are possible however. A head-final version189Unlike attributive modification, relative clause modification is prototypicallyhead-initial (38a), although head-final relative clauses (38b) are possible and arefreely volunteered. (38c) confirms that the head-final ordering cannot be a case ofattributive modification, since attributive t is not permitted in this context.(38) a. ?tQ?p-nt-?nshoot-DIR-1SG.ERGiPDETsk@kQ?kaPbirdiPDETw?k-@n.see-[DIR]-1SG.ERGI shot the bird that I saw.b. ?tQ?p-nt-?nshoot-DIR-1SG.ERGiPDETw?k-@nsee-[DIR]-1SG.ERGiPDETsk@kQ?kaP.birdI shot the bird that I saw.c. *?tQ?p-nt-?nshoot-DIR-1SG.ERGiPDETw?k-@nsee-[DIR]-1SG.ERGtOBLsk@kQ?kaP.birdI shot the bird that I saw.For Okanagan, either iP or t must introduce both the head and clausal portion of arelative.23(39) a. *wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDET?p ?yq-nt-iscook-DIR-3SG.ERGq?qxw@lx.fishYes, I ate the fish that he cooked.b. wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDET?p ?yq-nt-iscook-DIR-3SG.ERGiPDETq?qxw@lx.fishYes, I ate the fish that he cooked.of (37c) is ambiguously a case of attributive modification. The structure of post-nominal adjectivalmodification, especially those introduced by the oblique marker such as (37c), is unclear and needsfurther work.23This essentially means that pre-nominal (39a) and post-posed (40a) relatives are ungrammaticalin Okanagan. I introduce this terminology, stemming from Davis (2010a), below in section 6.3.3.190(40) a. *wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDETq?qxw@lxfish?p ?yq-nt-is.cook-DIR-3SG.ERGYes, I ate the fish that he cooked.b. wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDETq?qxw@lxfishiPDET?p ?yq-nt-is.cook-DIR-3SG.ERGYes, I ate the fish that he cooked.The most straightforward relativization patterns involve a ?matching effect? be-tween the particle which introduces the head nominal and the particle which intro-duces the modifier, as in the head-initial (41) and head-final examples (42) below.The initial particle follows predictably from the selectional properties of the mainclause predicate. As such, one might guess that the second particle is simply acopy of the initial particle, but I will show in the next section that things are not sosimple.24(41) a. wa ?yyescaP-nt-?spunch-DIR-3SG.ERGiPDETsq@ltm?xwmaniPDETwik-s.see-[DIR]-3SG.ERGHe hit the man he saw.b. i-s-c- ?kw??l-@m1SG.POSS-NOM-CUST-fix-MIDiPDETlpotcupiPDETm?Q-@n.break-[DIR]-1SG.ERGI am fixing the cup that I broke.c. JohnJohn?kw??l-@mmake-MIDtOBLyamx?waPbaskettOBLk?-s-n- ?qw??-t@n-s.U.POSS-NOM-n-pack-INSTR-3SG.POSSJohn made the basket he was going to carry.d. kn1SG.ABSs-c- ?kwu?l-xNOM-CUST-make-INTRtOBLy?mx?waPcedar.bark.baskettOBLks-yaP-y?Px?aP-s@lx.FUT-RED-show-3PL.POSSI?m making a basket that they will show.24The second particle(s) is determined by the selectional properties of the relative clause predicate,at least for relative clause types where clause-internal movement of a DP can be demonstrated. Seesection 6.3.3. See also 6.5 for a more technical discussion of the matching effect.191(42) a. wa ?yyescaP-nt-?spunch-DIR-3SG.ERGiPDETwik-ssee-[DIR]-3SG.ERGiPDETsq@ltm?xw.manHe hit the man he saw.b. i-s-c- ?kw??l-@m1SG.POSS-NOM-CUST-fix-MIDiPDETm?Q-@nbreak-[DIR]-1SG.ERGiPDETlpot.cupI am fixing the cup that I broke.c. JohnJohn?kw??l-@mmake-MIDtOBLk?-s-n- ?qw??-t@n-sU.POSS-NOM-n-pack-INSTR-3SG.POSStOBLyamx?waP.basketJohn made the basket he was going to carry.d. kn1SG.ABSs-c- ?kwu?l-xIMPF-make-IMPFtOBLks-yaP-y?Px?aP-s@lxFUT-RED-show-3PL.POSStOBLy?mx?waP.cedar.bark.basketI?m making a basket that they will show.The modifiers in (41-42) are all thematically transitive, with either ergativeor possessive subject morphology. Likewise non-eventive unaccusatives (43-45),eventive unaccusative modifiers (46) and unergative modifiers (47-48) are alsogrammatical within these ?matching? patterns, with variable ordering between thehead and modifier:(43) a. w?k-@nsee-[DIR]-1SG.ERGiPDETsk@kQ?kaPbirdsiPDETc@-c?PmaPt.RED-small(PL)I saw the small birds.b. w?k-@nsee-[DIR]-1SG.ERGiPDETc@-c? ?maPtRED-small(PL)iPDETsk@kQ?kaP.birdsI saw the small birds.192(44) a. kn-x?t-@nhelp-BEN-1SG.ERGiPDETtk?milxwwomaniPDETpaP-paPs-?lx.RED-sad-DEVI helped the sad woman.b. kn-x?t-@nhelp-BEN-1SG.ERGiPDETpaP-paPs-?lxRED-sad-DEViPDETtk?milxw.womanI helped the sad woman.(45) a. ix?PDETPamn-(n)t-?sfeed-DIR-3SG.ERGiPDETpuscatiPDETPilxwt.hungryHe fed the cat that was hungry.b. Pamn-(n)t-?s@lxfeed-DIR-3PL.ERGiPDETPilxwthungryiPDETpupQas.kittenThey fed the hungry kitten.(46) a. kn-x?t-@nhelp-BEN-1SG.ERGiPDETt@tw?tboyiPDET?t ?kw-a ?kw.fall.down-REDI helped the boy who fell down.b. kn-x?t-@nhelp-BEN-1SG.ERGiPDET?t ?kw-a ?kwfall.down-REDiPDETt@tw?t.boyI helped the boy who fell down.(47) a. w?k-@nsee-[DIR]-1SG.ERGiPDETx?xwt@mlittle.girlaPDETc-?@?tp-m@-nc?t.CUST-jump-MIN-REFLEXI saw a little girl that jumped.b. w?k-@nsee-[DIR]-1SG.ERGaPDETc-?@?tp-m@-nc?tCUST-jump-MIN-REFLEXiPDETx?xwt@m.little.girlI saw the boy who jumped.193(48) a. w?k-@nsee-[DIR]-1SG.ERGiPDETsk@kQ?kaPbirdsiPDET?txwtilx.flyI saw the birds that were flying.b. w?k-@nsee-[DIR]-1SG.ERGiPDET?txwtilxflyiPDETsk@kQ?kaP.birdsI saw the flying birds.Unaccusative (49-51) and unergative predicates (52-53) alike may also modify anoblique argument within a matching t pattern:25(49) a. kn1SG.ABSw?k-@msee-MIDtOBLs ?wa?r? ?kxnfrogtOBLqwQay.blueI see a frog that is blue.b. kn1SG.ABSw?k-@msee-MIDtOBLqwQaybluetOBLs ?wa?r? ?kxn.frogI see a frog that is blue.(50) a. kn1SG.ABSPiys-@mbuy-MIDtOBLlasm?stshirttOBL?c ?w- ?ca ?wt.RED-cleanI bought a clean shirt.b. kn1SG.ABSPiys-@mbuy-MIDtOBL?c ?w- ?ca ?wtRED-cleantOBLlasm?st.shirtI bought a clean shirt.25Though it should be said that the head-initial versions of (52-53) are more marginal than thehead-final versions. This could be taken as evidence that the ?double t? pattern is more closely relatedto attributive than to relative clause modification, although it cannot be understood as attributiveclause modification, since the modifiers in the examples cannot occur in CNPs. See related discussionin section 6.5.194(51) a. kn1SG.ABSw?k-@msee-MIDtOBLsk@kQ?kaPbirdstDETc@-c?PmaPt.RED-small(PL)I saw some small birdsb. kn1SG.ABSw?k-@msee-MIDtDETc@-c? ?maPtRED-small(PL)tOBLsk@kQ?kaP.birdsI saw some small birds.(52) a. kn1SG.ABSw?k-@msee-MIDtOBLsp@pl?naPrabbittOBLc-?@?tp-m@-nc?t.CUST-jump-MIN-REFLEXI saw the rabbit who jumped.b. kn1SG.ABSw?k-@msee-MIDtOBLc-?@?tp-m@-nc?tCUST-jump-MIN-REFLEXtOBLsp@pl?naP.rabbitI saw the rabbit who jumped.(53) a. kn1SG.ABSw?k-@msee-MIDtOBLtuP-t@tw?tRED-boytOBLc-caP-cQ?lx.CUST-RED-batheI saw some swimming boys.b. kn1SG.ABSw?k-@msee-MIDtOBLc-caP-cQ?lxCUST-RED-bathetOBLtuP-t@tw?t.RED-boyI saw some swimming boys.The head-final versions of (49-51) may be analyzed, ambiguously, as attributivelymodified complex obliques; however, the head-initial versions of these examplescannot be. Furthermore, given the distinctive morpho-syntactic pattern of attribu-tive modification, neither the head-initial nor head-final versions of (43-48) areanalyzable as attributive modifications, at least not on par with the CNPs and com-plex obliques discussed in the previous section. This is because iP is a determiner,but attributive t is not, which crucially implies a structural distinction.Nouns can also occur as modifiers within a relative clause modification pattern.Note that (54) below is translated as a head-initial modification, and not as a head-final (e.g. ?lady boss?) attributive.195(54) w?k-@nsee-[DIR]-1SG.ERGiPDETtk?milxwwomaniPDETylm?xw@m.chiefI met the lady who is chief/boss.If (54) does in fact involve a relative clause, the significant implication is that nounscan project clausal structure.26,276.3.2 Other Characteristics of Okanagan RelativesThere are other noteworthy characteristics of Okanagan relative clauses which Iwill briefly touch upon in this section. These are as follows:a. No dedicated relative clause inflectional pattern (Kroeber, 1999)b. No WH-relative pronouns (Davis, 2010a)c. Headless relatives are possible, and quite common (cf. section 4.3)d. A wide range of grammatical roles can be relativized, oblique argumentsbeing an exceptione. Long-distance relativization is possible(a) Unlike many other Salish languages, Okanagan relative clauses do notexhibit any special inflectional pattern. In other words, pronominal morphologyfound on relative clauses may also generally be found on main clause predicates(Kroeber, 1999, 272,304). Thus, the ergative relative clause predicate in (55a) andthe nominalized relative in (55b) can both function as main clause predicates inappropriate contexts.2826This possibility becomes important to consider especially with regards to DP-DP predicationand cleft data. I discuss this in more detail in section 8.8.27Nouns prefixed by Pak?- ?to have? are also able to function as relative clause predicates:(i) a. w?k-@nsee-[DIR]-1SG.ERGiPDETsq@ltm?xwman[iP][DET]Pak?-c?txwHAVE-house(iP)(DET)?klLOC?ti ?kwt.lakeI?ve seen a man that had a house by the lake.b. w?k-@nsee-[DIR]-1SG.ERG[iP][DET]Pak?-c?txwHAVE-houseiPDETsq@ltm?xw.manI?ve seen a man that has a house.28The exact semantic difference between (55a) and (55b), if there actually is one, remains unclear.196(55) a. ?tQap-nt-?nshoot-DIR-1SG.ERGiPDETw?k-@nsee-[DIR]-1SG.ERGiPDETsk@kQ?kaP.birdI shot the bird that I?ve seen.b. ?tQap-nt-?nshoot-DIR-1SG.ERG[iP][DET]i-s-c-w?k1SG.POSS-NOM-CUST-seeiPDETsk@kQ?kaP.birdI shot the bird that I?ve seen.(b) As in the rest of Salish (Davis, 2010a), there are no relative pronouns, WHor otherwise, in Okanagan (56):(56) a. wa ?yyescaP-nt-?spunch-DIR-3SG.ERGiPDETsq@ltm?xwman(*swit)whoiPDETwik-s.see-[DIR]-3SG.ERGHe hit the man who he saw.b. JohnJohn?kwu?l-@mmake-MIDtOBLy?mx?waPbasket(*sti ?m)whattOBLk?-s-n- ?qw??-t@n-s.U.POSS-NOM-n-pack-INSTR-3SG.POSSJohn made the basket which he was going to carry.(c) Both subject and object-centered ?headless? relatives are common in Okana-gan (57). I assume that these are a special type of head-initial relative, where thehead noun, and its selecting determiner, are both null (see previous discussion insection 4.3).(57) a. ?q@ ?y-nt-?xwwrite-DIR-2SG.ERGiPDETqw@l-qw?l-st-m-@n.RED-speak-CAUS-2SG.ACC-1SG.ERGWrite down what I?m telling you.Speakers indicate that nominalized forms like (55b) are past-tense completive, while ergative formslike (55a) are present-tense completive, but my research suggests that there is no clear demarcationbetween the two, and that both can be uttered felicitously within an identical discourse situation.197b. kaPk?c-@nfind-[DIR]-1SG.ERGaPDETc-s?l-m?-st-@n.CUST-lose-MIN-CAUS-1SG.ERGI found the one I was looking for.c. TinaTinawik-ssee-[DIR]-3SG.ERGiPDETxwi ?c-xt-sgive-BEN-3SG.ERGtOBLq?qxw@lx.fishTina saw the one she handed the fish to.Demonstratives appear to function as relative clause heads (58), but since demon-stratives often associate with a constituent DP (cf. section 4.6), (58) may alsobe analyzed as a headless relative under the assumption that the demonstrative isassociated with a null DP.(58) w?k-@nsee-[DIR]-1SG.ERGix?PDEMiPDETks-kn-xit-m-s.FUT-help-BEN-2SG.ABS-3SG.ERGI saw the one who will help you.(d) Many grammatical roles may be relativized in Okanagan. In addition to therelativized transitive objects and subjects, themes of benefactive applicatives mayalso be relativized (59):(59) a. kwin-[n]ttake-DIRiPDETq?qxw@lxfishiPDETxwi ?c-xt-m-@n.give-BEN-2SG.ACC-1SG.ERGTake the fish that I?m giving you.b. tal?Pveryin-x??st1SG.POSS-goodiPDETy?mx?waPbasketiPDETkwu1SG.ABS?kwu?l-xt-xw.make-BEN-2SG.ERGI like the basket that you made me.Rather than directly extract a subject of a transitive predicate, speakers often preferto passivize the relative clause predicate as part of a topic maintenance operation,and extract the agent. In (60), the clausal remnant is inflected as passive by thesuffix -m, and is introduced by the sequence iP t, which together indicate that the198passive agent has been extracted.29(60) s-c- ??aP- ??aP-?m-sNOM-CUST-RED-look.for-MID-3SG.POSSiPDETs@xw- ?maP- ?m?yaP-m-sOCC-RED-teach-MID-3SG.POSSiPDETtOBLkn-x?t-@mhelp-DIR-PASSiPDETlLOCs@n- ?q@ ?y-m?n-t@n.LOC-write-INSTR-INSTRHe?s looking for the teacher that helped him at school.Oblique arguments of formally intransitive predicates cannot generally be rela-tivized (61a). In these cases, speakers will normally correct the relativized predi-cate to a transitive form (61b,c) (Montler, 1993), though (61b) shows that it is in-sufficient to change the relativized predicate to a transitive form in the context of amorphologically intransitive main clause predicate. There are apparent exceptionsto the generalization that oblique arguments of formally intransitive predicates can-not be relativized (62).(61) a. *kn1SG.ABSks-kaPk?c-aPxFUT-find-INTRtOBLautomobileautomobilekn1SG.ABS(t)(OBL)s-c- ??aP- ??P-?s-x.NOM-CUST-RED-look.for-eye-INTRI?m gonna find the car I?m looking for.b. *kn1SG.ABSks-kaPk?c-aPxFUT-find-INTRtOBLautomobileautomobiletOBLc- ??aP- ??aP-st-?n.CUST-RED-look.for-CAUS-1SG.ERGI?m gonna find the car I?m looking for.c. i-ks-kaPk?c-@m1SG.POSS-FUT-find-MIDiPDETautomobileautomobileaPDETc- ??aP- ??aP-st-?n.CUST-RED-look.for-CAUS-1SG.ERGI?m going to find the car that I?m looking for.29Passive-agent relatives show evidence for formation through A? movement, as I discuss in thenext section. See specifically the discussion around example 67.199(62) kn1SG.ABSx?m?nk-@mwant-MIDtOBLsiw?kwwatertOBLks-s?wst-xFUT-drink-INTRi[n]-s?l?x?t.1SG.POSS-friendI want some water for my friend to drink.(e) Long-distance relativization is possible (63). For (63a), the nominal heads ??aPc?n@m ?deer? is an underlying object argument of the transitive imperativecp ?y ?qnt?kw ?cook it!?, while for (63b), the nominal head tk?m?lxw ?woman? is anunderlying subject argument of the intransitive predicate xwuy ?go.?(63) a. JohnJohn?tQap-nt-?sshoot-DIR-3SG.ERGiPDETs ??aPc?n@mdeeriPDETcu-ssay-[DIR]-3SG.ERGNormanNormayBennyBenny?c- ?p ?yq-nt-?kw!?.CUST-bake-DIR-IMPJohn shot the deer that Norman told Ben to cook.b. ucYNQwik-nt-xwsee-DIR-2SG.ERGiPDETtk?milxwwoman?kl- ?klaxwRED-eveningiPDETwik-ssee-[DIR]-3SG.ERGSarahSarah?aPCOMPxwuy?goDo you know the woman who Sarah saw leave (early last night)?6.3.3 Relative Clause Formation by MovementAs first noted by Kroeber (1997, 396) for Thompson, locative relatives seem toinvolve clause internal movement of a PP to the left periphery of a relative clause.Kroeber notes that in examples like (64), ?...the preposition codes the relation ofgap to relative clause predicate, not the relation of the whole relative clause to thematrix predicate.?200(64) (w)P?xPROGkn1SGxw?P-mlook.for-MIDteOBL.DETnp?ytn2bed[[n-ein-DET[\u000BNP2 ]PP1 ]xw? ?yFUTwn1SG.CJCTQw? ?ytsleept1CP]I?m looking for a bed where I?m gonna sleep. (Koch, 2006, 132)In other words, because the preposition n ?in? in (64) helps to specify the loca-tion of the sleeping event, and not the looking event (i.e. it fixes the location of ?thebed? in this case), the preposition may plausibly be analyzed as having moved froma position inside the relative clause, following the verb Qw? ?yt ?sleep?. Davis (2004)and Koch (2006) have shown for Lillooet and Thompson respectively that the de-terminer also moves, or rather, the clause internal DP ?pied-pipes? the prepositionto a clause-initial position. This is illustrated by the bracketing in (64).Since Lillooet and Thompson determiners vary with regards to their spatio-temporal properties, Davis (2004, 2010a) and Koch (2006) are able to show thatthe determiner introducing the relative clause reflects the spatio-temporal proper-ties of the relative clause predicate, rather than the main clause predicate, confirm-ing that movement also occurs in relatives which do not involve locative marking.For Okanagan, it is not possible to use different determiners as a diagnostic formovement, since there is only one determiner involved in relativization, iP. Never-theless, the oblique marker t as well as the other locative markers, help to confirmthat movement has occurred. I now discuss why.Recall that for Okanagan, the oblique marker t and locative markers ?kl, l and tlmay co-occur with iP. These particle sequences help provide evidence for clause-internal movement. In main clauses, the combination of iP and t introduces in-struments and passive agents, as in (65), and the combination of iP and a locativeparticle designates a DP as a locative adjunct, as in (66) iP tl sq@ltm?xw ?from theman?.201(65) a. ?tQap-nt-?sshoot-DIR-3SG.ERG[iPDETtOBLs-wlwlm-inkPP].NOM-iron-weaponHe shot it with a gun.b. MikeMike?c? ?m-qs-nt-@msuck-nose-DIR-PASS[iPDETtOBLtk?milxwPP].womanMike was kissed by the woman.(66) c-ylt-m?-st-s@lxCUST-run.away-MIN-CAUS-3PL.ERG[iPDETtlLOCsq@ltm?xwPP].manThey?re running away from the man.In support of a movement analysis for Okanagan relatives, consider that wheninstruments and passive agents like those in (65) are relativized, the relative clauseis introduced by both iP and t (67):30,31(67) a. kwu1SG.GEN??P ?qw-@m-?tshow-MID-APPLiPDETn? ?k-m@n2cut-INSTR[[iPDETtOBL[\u000BNP2]PP1]ni ?k-nt-xwcut-DIR-2SG.ERGt1CP].Show me the knife that you cut it with.30I leave off bracketing for the DP in (67) for the reader?s sake. Recall from chapter 4 that Ihave analyzed the bracketed PPs in (67) as being introduced by a determiner because of a prosodicinversion of D and P.31H?bert (1982b, 46, ex.46) argues that oblique extractions are ungrammatical, and shows anungrammatical case of an instrument extraction (as a type of oblique), but her particular example islikely ungrammatical since when overt, the sequence iP t is stranded at the end of the sentence, andwhen not overt, there is nothing to code the relation of ?the knife? to ?the man?:(i) *iPDETtt ?witboywik-ssee-[DIR]-3SG.ERGiPDETn? ?km@ncut-INSTRiPDETsq@ltm?xwmanpul-st-skill-CAUS-3SG.ERGiPDETx??Qx?aQcrow(iP(DETt).OBL)The boy saw the knife that the man killed the crow with.202b. MikeMikewik-ssee-[DIR]-3SG.ERGiPDETtk?milxw2woman[[iPDETtOBL[\u000BNP2]PP1]?c? ?m-qs-nt-@msuck-nose-DIR-PASSt1CP].Mike saw the woman he was kissed by.Note that iP and t normally only co-occur when introducing a passive agentor instrument, or before clauses from which these grammatical roles have beenextracted. In extraction contexts involving passive patients, for example, iP t maynot introduce the relative clause, only iP. (68) shows an example of an extractedpatient iP tk?milxw ?the woman?, where the clausal remnant is introduced by thedeterminer iP, and an in-situ clause-internal agent is introduced by iP t. Oblique tcannot introduce the relativized predicate, since it is the patient and not the agentthat has been extracted.(68) JohnJohns-c- ??aP- ??aP-?m-sNOM-CUST-RED-look.for-MID-3SG.POSSiPDETtk?milxwwomaniPDET(*t)(*OBL)kn-x?t-@mhelp-[DIR]-PASSiPDETtOBLsq@ltm?xw.manJohn is looking for the woman who was helped by the man.Given that the distribution of the sequence iP t is limited to the same grammat-ical subset in both extraction and non-extraction contexts, the sequence iP t in (67)constitutes evidence for clause-internal movement.Similarly, when a locative adjunct is extracted in Okanagan, the relative clauseis introduced by a determiner plus locative marker sequence, thus furnishing ev-idence parallel to Thompson (64) that clause-internal movement has indeed oc-curred. Compare (66) and (69a), in particular.3232Not all locative relatives in Okanagan follow the same relativization strategy as that exhibited in(69a-b). In (i) below, the head sq@ltmixw ?man? is coded as a direct patient argument of the possessorpredicate ikstwm ?I?m going to sell?, while the shirt is a theme of the lexical ditransitive tw ?tosell?. In (ii), the standard locative extraction strategy is utilized, however the semantics of ?from? isduplicated in a main clause demonstrative adverbial itliP as well as the locative marker tl, in additionto the clause-internal locative marker. The nature of this ?copying? effect is unclear to me, but (ii)may support the movement account if the first occurrence of tl is a copy of the second occurrence,which although normally deleted is not in this case for some reason. (Thanks to Henry Davis for203(69) a. wik-@nsee-[DIR]-1SG.ERGiPDETsq@ltm?xw2man[[iPDETtlLOC[\u000BNP2]PP1]c-ylt-m?-st-s@lxCUST-run.away-MIN-CAUS-3PL.ERGt1CP].I see the man that they?re running away from.b. ucYNQc-my-st-?xwCUST-know-CAUS-2SG.ERGiPDETsq@ltm?xw2man[[iPDET?klLOC[\u000BNP2]PP1] tw-m?-st-@m-@nsell-MIN-CAUS-MIN[?]-1SG.ERGiPDETlasm?stshirtt1CP].Do you know the man that I sold the shirt to?Following Koch (2006) and Davis (2010a), I claim that for Okanagan instru-mental and passive agent relatives (67) and locative relatives (69), a DP internalto the relative clause has raised to the left periphery of the relative clause CP. Thenoun in the moved DP then plausibly undergoes deletion through identity with theclause exterior head NP. The following structure is one possible representation ofthe relative clause in (69a):pointing this out to me.)(i) ucYNQc-my-st-ixwCUST-know-CAUS-2SG.ERGiPDETsq@ltm?xwman[iP][DET]i-ks-tw-m1SG.POSS-FUT-sell-MIDiPDETtOBLlasm?st?shirtDo you know the man I?m going to sell the shirt to?(ii) kw@n-?xwtake-[DIR]-2SG.ERGitl?PDEMtlLOCs?yaPsaskatoonsiPDETtlLOCc-x?@ ?w- ?w-x?t-m-@n.CUST-dry-FRED-BEN-2SG.ACC-1SG.ERGTake from these berries that I am drying for you.204Figure 6.3: Okanagan Locative Relative ClauseDPDiPNPNPNsq@ltm?xw jCPSpecPPiPtlDPDiPNPpro jC?C\u000BTP... VPcylt@m?stl@x PPti?the man that they?re running away from?Assuming that all relative clauses in Okanagan are similarly formed, the struc-ture in (6.3) implies that the sequence of particles introducing the clausal remnantshould always code the relation of the gap to the relative clause predicate. Considerthat subject and object extractions in Okanagan are characterized by having the de-terminer iP introduce both the head and the clausal remnant.33 Since transitivepredicates always select for iP DP objects in main clause contexts (70a), the pre-diction is that when an object is extracted, the clausal remnant will be introducedby only a determiner iP. This prediction is upheld (70b).34(70) a. wik-ssee-[DIR]-3SG.ERGiPmansq@ltm?xwHe saw the man.33At least when the main clause predicate is transitive. When the main clause predicate is intran-sitive, and the modifier is an irrealis, nominalized form (cf. 41c,d), t may introduce both head andmodifier. A coherent syntactic account of these double-oblique modifications has yet to be workedout, though see section 6.5 for some discussion.34These are consistent with the movement hypothesis, but do not constitute a particularly strongargument for it, since as Koch (2006) notes for similar cases in Thompson, the two determiners maysimply be copies of one another.205b. wa ?yyescaP-nt-?spunch-DIR-3SG.ERGiPDETsq@ltm?xwmaniPDET(*t)(*OBL)wik-s.see-[DIR]-3SG.ERGHe hit the man he saw.As a working hypothesis then, I assume that all Okanagan relatives are formedby clause-internal movement. I further assume that relative clauses are canoni-cally head-initial, and that head-final relatives are derived from head-initials by anadditional movement of the relative clause CP to a position preceding the DP con-taining the head, presumably Spec DP.35 Compare the head-initial relative clauseiP sq@ltm?xw iP kwu wiks ?the man who saw me? (6.4) with its equivalent head-finalversion iP kwu wiks iP sq@ltm?xw (6.5):Figure 6.4: Head-initial (a.k.a. ?post-nominal?) Relative ClauseDPDiPNPNPNsq@ltm?xw jCPSpecDPtiDiPNPNpro jC?C\u000BTP... VPkwu wiks DPi?The man who saw me.?35Or possibly adjoined to DP. Pre-posed (head-final) relatives in Okanagan (and Thompson) aregenerally more marked than post-nominal (head-initial) forms (cf. Koch (2006) for Thompson).206Figure 6.5: Head-final (a.k.a. ?pre-posed?) Relative ClauseDPSpecCPkSpecDPtiDiPNPNpro jC?C\u000BTP... VPkwu wiks DPiD?DiPNPNPNsq@ltm?xw jCPtk?The man who saw me.?Head-initial (6.4, cf. 72b below) and head-final (6.5, cf. 71b) relative clausesmay be referred to respectively as post-nominal, and pre-posed relatives follow-ing Davis (2002, 2004, 2010a), who develops a typology of Salish relative clausesbased not only on relative head-modifier ordering, but also on whether or not a par-ticle introduces both the head and modifier. Two other types of relatives in Davis?typology, pre-nominal (71a) and post-posed (72a) relatives are ungrammatical inOkanagan (cf. 39-40).36(71) a. *wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDET?p ?yq-nt-iscook-DIR-3SG.ERGq?qxw@lx.fishYes, I ate the fish that he cooked.b. wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDET?p ?yq-nt-iscook-DIR-3SG.ERGiPDETq?qxw@lx.fishYes, I ate the fish that he cooked.36Straits Salish (Montler, 1993) and Lillooet (Davis, 2010a) contrast with Okanagan since bothallow these types of relatives. Davis (2010a) claims that relatives in Lillooet are all derived from acommon pre-nominal structure. Okanagan, like Thompson, has marked pre-posed relatives, and soDavis?s analysis would require first extraposition, and then pre-posing.207(72) a. *wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDETq?qxw@lxfish?p ?yq-nt-is.cook-DIR-3SG.ERGYes, I ate the fish that he cooked.b. wa ?yyesPi?-@neat-[DIR]-1SG.ERGiPDETq?qxw@lxfishiPDET?p ?yq-nt-is.cook-DIR-3SG.ERGYes, I ate the fish that he cooked.The following table compares relativization possibilities in Okanagan withthree other Interior Salish languages, in light of Davis? typology.37 ?D1? and ?D2?refer to the linear order of determiners.Table 6.3: Relativization Strategies in Four Interior Salish LanguagesPre- Post- Post- Pre-nominal posed nominal posed[D1[clauseNP]][D1[NPclause]][D1[NP[D2clause]]][D1 clauseD2 NP]Lillooet X X X *Thompson * * X XMoses-Columbian X X (X) *Okanagan * * X XI assume that headless relatives in Okanagan (cf. 57) are a sub-type of post-nominal relative clause, where one of the determiners deletes as a result of adouble-determiner filter, formulated by Davis (2010a, 22) as consisting of two parts(73). I give the structure of the headless relative in (57) as Figure 6.6, where a de-terminer containing a head NP is phonologically adjacent to a determiner whichheads the moved DP, and so deletes.(73) a. Double Determiner Filter*[D1...D2] where no lexical head intervenes between D1 and D237In Moses-Columbian, post-nominal relatives are possible (N. Mattina 2006, 124), but the obliquemarker is becoming optional there (Willett, 2003, 109).208b. Determiner DeletionDelete one of two phonologically adjacent determiners.Figure 6.6: Headless Relative ClauseDPDiPNPNPN\u000B jCPSpecDPtiDiPNPNpro jC?C\u000BTP... VPqw@lqw?lstm@n DPi?the thing I am telling you.?A few short comments on the markedness of pre-posed relatives are in orderhere. Pre-posed relatives involving passive agent extractions are generally marginalto ungrammatical (74a). There are examples of pre-posed locative relatives volun-teered during elicitation sessions, as in (74b) below, however these are commonlyjudged ungrammatical when presented to a speaker.(74) a. *MikeMikewik-ssee-[DIR]-3SG.ERGiPDETtOBL?c? ?m-qs-nt-@msuck-nose-DIR-PASSiPDETtk?milxw.womanMike saw the woman he was kissed by.209b. iPDETtlLOCkwu1SG.ABScun-[n]t-xwsay-DIR-2SG.ERGiPDETs@n-tw-m?s[t]-t@n,LOC-sell-INTR.REFLEX-INSTRitl?PDEMax?PDEMkn1SG.ABSn-P?ys-@m.n-buy-MIDFrom the store you told me about, that is where I bought this.The fact that data like (74a,b) are marginal to ungrammatical can be explained bythe resulting linear clash between the selectional restrictions of the main clausepredicate, and the particle(s) that immediately follows the predicate. For example,although the transitive predicate in (74a) selects for an object introduced by iP, itdoes not select for an object introduced by iP t. Interestingly, left-dislocating theentire DP argument containing a pre-posed relative may circumvent this constraint(74b), but more work needs to be done here.6.3.4 Problems with Extending the Movement AccountExtending the movement account as discussed in this chapter to all Okanagan rel-atives encounters several problems, two of which are as follows.First, the distribution of t before a clause does not always code the relationof the gap to the relative clause predicate: Upper Nicola relative clauses inflectedwith ks- future may be preceded by iP and t, but a main clause argument of sucha predicate may not be (Lyon, 2011). To illustrate, a possessor intransitive likeksyaPy?Px?aPs@lx ?they will look at it? selects for a core, iP DP object, as in (75a),and the oblique marker is not possible here. Nevertheless, an oblique marker op-tionally surfaces for a relativized argument of a predicate inflected with future ks-,as in (75b). Because the sequence iP t does not reflect the selectional properties ofthe relative clause predicate, it is unclear how the movement account argued for inthis chapter applies to data like (75b).38(75) a. ks-yaP-y?Px?aP-s@lxFUT-show-[DIR]-3PL.POSSiPDET(*t)(*OBL)pwm?ndrumThey will show a drum.38Lyon (2011) analyzes this occurrence of t as a remnant of an earlier relativization strategy.210b. ?kw??l-@nmake-[DIR]-1SG.ERGiPDETy?mx?waPcedar.bark.basketiPDET(t)OBLks-yaP-y?Px?aP-s@lx.FUT-IRED-show-3PL.POSSI made a basket that they will show.Second, extractions of benefactive themes (76b) involve a relative clause predicateintroduced by iP, rather than t, which is unexpected given that benefactive themesare introduced by t in main clause contexts (76a) (cf. Davis and Matthewson (2003)and Gerdts and Kiyosawa (2010, 47-50) who note that ?oblique? objects are able toextract directly in -xit marked applicative predicates in Lillooet.)(76) a. xwi ?c-xt-m-ngive-BEN-2SG.ACC-1SG.ERGtOBLq?qxw@lx.fishI gave you a fish.b. kwin-ttake-DIRiPDETq?qxw@lxfishiPDETxwi ?c-xt-m-n.give-BEN-2SG.ACC-1SG.ERGTake the fish that I?m giving you.The crucial point, however, is that Okanagan locative and passive-agent rela-tive clauses show evidence for A? movement, similarly to Thompson and Lillooetrelatives, and that core-argument extractions are also consistent with the movementanalysis. Other cases of relativization such as (75-76) may ultimately be explain-able within this basic theory, with additional modifications.6.4 SummaryThis chapter has presented data relating to two types of nominal modification inOkanagan: attributive and relative clause modification. I have claimed that nominalmodification, broadly speaking, can be distinguished from clausal subordination bythe distribution of particles. The determiner iP and oblique marker t are not usedas clausal subordinators, but are used in structures involving nominal modification.Next, attributive modification can be distinguished from a relative clause by thefollowing:211(i) Attributive modification is strictly head final, relative clause modificationmay either be head initial or head final.(ii) A nominal head must be introduced by t in an attributive structure, but mayalso be introduced by iP in a relative clause structure.(iii) Attributive modifiers must be non-eventive, stage or individual-level un-accusative predicates. An eventive predicate can only modify a nominalthrough a relative clause structure.I have shown that a subset of Okanagan relative clauses show evidence for aclause internal, A? movement of a DP or PP to the left periphery of an embedded CP(Koch, 2006; Davis, 2010a). The movement account of relative clause formationoutlined in this chapter will be of particular importance during the discussion ofcleft clauses in later chapters.6.5 Chapter Addendum: Notes on the ?Matching Effect?This addendum consists of a technical discussion of the patterns which iP and tdisplay in their nominal modification roles. Specifically, I discuss the ?matchingeffect? seen with Okanagan relative clauses in more detail, as well as problematicpatterns in need of further work.There are six possible surface patterns involving determiner iP and obliquemarker t in their capacity of introducing heads and modifiers in nominal modi-fication structures. These six patterns are displayed in Table 6.4. Each patternis indicative of either attributive modification (?attr?), relative clause modification(?rel?) or in at least one case, ambiguously both.212Table 6.4: Surface Patterns Displayed by Head/Modifier Introductory Parti-cles in Okanagan Sentences Involving Nominal ModificationPattern head-initial head-finalbeforenominalbeforemodifierbeforemodifierbeforenominal1 ? iP t attr iP t2 rel iP iP rel iP iP3 rel t t rel/attr t t4 rel iP iP t rel(*) iP t iP5 rel t iP t rel(*) iP t iP6 * t iP * t iPNot all logically possible patternings of iP and t are grammatical, as shown forpattern 6 in the above table, showing that there is a ?matching effect? in Okanaganrelative clause modifications (patterns 2-5) whereby the particle that introduces thehead NP must also introduce the modifier, regardless of whether any additionalparticles may or may not introduce the modifier. There is no matching effect forpattern 1, which may be exclusively characteristic of attributive modification, al-though the status of head-initial pattern 1 modifications is unclear. The implicationis that a matching effect is diagnostic of a clausal modifier, given that only non-clausal, non-eventive modifiers can occur as modifiers with attributive pattern 1modifications.The movement account successfully captures patterns 2 and 4-5, with the ex-ception of those cases where iP t does not code a passive agent (cf. 75 above).Given that the relative clause-introducing particle(s) must match the selectionalrestrictions of the clausal predicate, the matching effect may be roughly character-ized as a requirement that the head of the relative clause be introduced by (at leastone of) the particles which introduce the relative clause. There is then the addi-tional requirement that the main clause predicate be able to select for the relativeclause head. For extracted ditransitive themes (cf. 76) above, however, the match-ing effect appears to stem from a requirement that the clause-introducing particle213match the head-introducing particle, as determined by the selectional properties ofthe main clause predicate. The ?direction? of the matching effect thus seems to bevariable, depending on the type of clausal modification.The ungrammaticality of head-final patterns 4 and 5 seems to be due to a linearrequirement that the particle(s) directly following an initial main clause predicatealso match that predicate?s selectional restrictions. In sum, there are three factorsinvolved in these matching effects:a. The selectional properties of the relative clause predicate.b. The selectional properties of the main clause predicate.c. A linear adjacency requirement between the main clause-predicate and theparticles which immediately follow, such that the particles be consistentwith the main clause-predicate?s selectional restriction.The status of pattern 3 remains unclear. While head-final pattern 3 modifica-tions involving non-eventive modifiers are straightforwardly analyzable as attribu-tive modifications, it is less clear what the status of head-initial pattern 3 non-eventive modifications is (cf. also head-initial pattern 1). For pattern 3 head-initialclausal modifications, the ?direction? of the matching effect is similar to that seemwith ditransitive theme extractions, i.e. the t introducing the clause usually matchesthe selectional properties of a main-clause intransitive predicate, rather than the se-lectional properties of the relative clause predicate (and quasi-objects are generallynot extractable). Furthermore, an optional pre-modifier determiner iP sometimessurfaces for head-initial pattern 3, yielding head-initial pattern 5. As such, evidencefor clause-internal movement is not forthcoming for pattern 3.Further work is required in elucidating and explaining the patterns shown inTable 6.4, but I hope that this chapter has made a significant contribution to ourunderstanding of nominal modification in Okanagan.214Chapter 7Direct Predications and DP-DPStructures: Syntax, Semantics,and Information Structure7.1 IntroductionAt this point, all of the necessary analytical ingredients are in place for addressingthe central question of this dissertation: namely, what is the structure and interpre-tation of DP-DP structures, and how do they differ from direct predications? Directpredications consist minimally of a lexical predicate (an NP, AP, or PP) and a DPargument, as in (1a). DP-DP structures consist of two DPs, as in (1b).(1) a. [ix?PDP]DEM[y?mx?waPNP] .cedar.bark.basketThat is a basket. Direct predicationb. [ix?PDP]DEM[iPDETy?mx?waPDP] .cedar.bark.basketThat is a/the basket. DP-DP structure2157.1.1 Main Claims of this ChapterThe main claims of this chapter are as follows:a. Okanagan DP-DP structures are syntactically and information-structurallydistinct from direct predications.b. Okanagan DP-DP structures are semantically equative structures (Heycockand Kroch, 1999), given that neither DP can be a predicate (Longobardi,1994; Matthewson, 1998).c. Okanagan DP-DP structures involve a null, equative copula, while directpredications do not involve any copula.d. The maximality implicature carried by the determiner iP (cf. chapter 5)derives an implicature of exhaustivity (Davis et al., 2004) for DP-DP struc-tures.e. Okanagan DP-DP structures display a fixed information structure. Theinitial, referential DP in an Okanagan DP-DP structure is always in focus(Rooth, 1992).f. Specificational DP-DP structures are not possible in Okanagan. This re-ceives an explanation whereby:(i) there is a ban on predicate raising (a.k.a. ?syntactic inversion?) (Moro,1997; den Dikken, 2006).(ii) the equational head selects only intensional () (Romero, 2005;Comorovski, 2007) iP DPs as a complement, not directly referential DPs.(iii) the equational head assigns a feature ?F? (i.e. focus) to its second ar-gument (i.e. the DP in specifier position).(iv) information-structural alignment constraints force the focus to occurleft-most (Koch, 2008a).g. Identificational sentences may be reduced to the predicational class insome cases (cf. Heller (2005) for English and Hebrew), and to the equativeclass in other cases.216h. Apparent cases of syntactic inversion involving DP-DP structures may beanalyzed either as identificational predications, or as involving topicaliza-tion of the second, iP DP in a DP-DP structure. These apparent cases ofinversion are possible within a running discourse, but not in answer to aWH-question.i. Since specificationals do not exist, Higgins? taxonomy may be reduced totwo types for Okanagan (cf. item g): predicational and equative (cf. Heller(2005) for English and Hebrew).7.1.2 Chapter OutlineThis chapter is summarized as follows.First, I discuss direct predications (7.2, cf. 1a). These are interpretively equiv-alent to either predicational or identificational copular clauses in English (Higgins,1973, 1979), depending on the context in which they are used, and whether or not ademonstrative functions as the subject (7.2.1). There is a syntactic ban on predicateraising (a.k.a. ?inversion?) for direct predications (7.2.2), though there is relativelyunconstrained word ordering of subject and predicate. I take this as evidence thatthere is no null copula for direct predications. This means that lexical projectionsmay be inherently predicative (Davis (1999a) for Salish and Stowell (1981), contraBaker (1996) and Adger and Ramchand (2003)), and makes possible an analysisof direct predications as bare small clauses (7.2.4).Next, I discuss DP-DP structures (7.3, cf.1b). Though they are structurally dis-tinct from direct predications, the interpretation of DP-DP structures overlaps withthat of both predicational and equative clauses in English (Higgins, 1973) (7.3.1).This interpretive variability is made possible by the fact that the iP determiner al-lows both maximal and non-maximal interpretations (cf. chapter 5). In answer to aWH-question, DP-DP structures require the more-referential DP to occur initially,unlike the case for direct predications, where subjects routinely occur finally inthese contexts.1 Like direct predications, DP-DP structures do not allow predicate1There is an analogy to be made between the more-referential DP in a DP-DP structure and thesubject of a direct predication, especially in cases where DP-DP structures and direct predicationsare interpretively equivalent. Nevertheless, I refrain from calling the more-referential DP in a DP-DP217raising (7.3.2). For independent reasons, it is not possible to test for connectivityeffects in Okanagan DP-DP structures (7.3.3). I claim that DP-DP structures arestructurally asymmetrical projections of an equative head (Heycock and Kroch,1999), which is compositionally required by the fact that neither DP can functionas a syntactic predicate. The equative head in effect licenses the second DP as asyntactic predicate (7.3.4).I then discuss information structural and pragmatic properties of DP-DP struc-tures which set them apart from direct predications (7.4). Direct predications donot imply exhaustivity, whereas DP-DP structures carry an exhaustivity implica-ture (7.4.1). In addition to the exhaustivity implicature, DP-DP structures sharetwo other properties with Okanagan clefts (discussed in chapter 8): DP-DP struc-tures do not carry any presupposition of existence (7.4.2), and the more-referentialDP in focus must always occur to the left. These parallels provide support for myargument that DP-DP structures and clefts both derive from an underlying equativeconfiguration. The ban against focus-final DP-DP structures essentially means thatthe analogue to specificational copular clauses in English is not possible in Okana-gan (7.4.3). This suggests that the equative head is sensitive to the type of DP whichit selects for, especially given the structurally independent ban on syntactic inver-sion in Okanagan (7.5). I argue for an intensionality-based semantic asymmetry inDP-DP structures (Romero, 2005; Comorovski, 2007; Heycock, 2012).I then present my semantic analysis of the equative head (7.5). The equativehead selects for an intensional DP (Romero, 2005; Comorovski, 2007), and mapsthe intension of the individual to its extension. It also assigns a feature ?F? to itssecond argument (i.e. the extensional DP in specifier position), which is interpretedby the pragmatics as focused (Rooth, 1992). In other words, all of the contextuallyrelevant alternatives to the referent of the specifier DP in an equative structure areinvoked (cf. section 2.3.1). An information-structural alignment constraint forcesthe focus to occur left-most (Koch, 2008a). The exhaustivity implicature carriedby an equative sentence is derived from the maximality implicature introduced bythe determiner iP: since non-coreference between two identical occurrences of iPDPs is independently possible, though not preferred (cf. chapter 5), an exhaus-tive reading of an equative holds only if the maximality implicature carried bystructure a ?subject?.218the iP DP is not cancelled. This analysis successfully explains the availability of?pseudo-predicational? readings in DP-DP equatives, by which I mean that DP-DPequatives are in certain cases interpretively indistinguishable from correspondingdirect predications.Next (7.6.1), I discuss data which seem to correspond to Higgins? identifi-cational class of copular clauses, and suggest that these might be reduced to apredicational class (Heller, 2005). These sentence-types involve demonstrativeor demonstrative-associated DPs as subjects, and proper names as predicates, al-though predications involving two proper names pattern similarly. The predicativestatus of proper names in these cases is supported by the relatively free orderingof demonstrative and proper name, but also receives independent morpho-syntacticsupport: proper names may function as hosts to absolutive subject proclitics, likeother predicates, and can in certain contexts be complements to an iP determiner(cf. section 4.6.2). In contrast, proper names cannot be predicates within an equa-tive structure, a fact that may be attributed to the requirement that the equative headselect for an intensional DP.Next (7.7), I discuss several examples of problematic ?inversion? data, whichdo not follow from the arguments made so far in this chapter. At first glance,they seem to be cases of specificational sentences; however importantly, they arenot possible as answers to WH-questions. I weigh two possible analyses of thesecases, as either identificational sentences with null demonstratives, or as equativesinvolving topicalization of the second DP.In closing, I summarize and discuss implications related to extending Higgins?taxonomy to Okanagan (7.8).7.2 Direct PredicationThis section introduces further examples of Okanagan direct predications (cf 1a).I introduce some basic direct predication data and show that neither syntactic em-beddedness nor prosodic heaviness affect the basic generalization that the subjectand predicate can occur in either order. Next, I show that predicate-initial directpredications cannot be derived by predicate raising, which in conjunction with con-trasting DP-DP structure data, I take to be evidence that direct predications are bare219small clauses and do not involve any copula or functional head serving as an in-termediary between the subject and predicate. Finally, I summarize the data andanalysis.7.2.1 Direct Predications and Word OrderOkanagan makes extensive use of non-verbal predication (A. Mattina and DeSautel2002), but does not have an overt copula (A. Mattina 2001). Examples of nominalpredications are shown in (2), where the nominal ?p?naP ?birch bark basket? func-tions as a predicate, and the demonstrative ax?P ?this? functions as the argument.An adjectival predication is shown in (3), where ?Qa ?t ?wet? functions as the pred-icate, and the proper name Ivan as the argument. The linear order of subject andpredicate is free with Okanagan direct predications (N. Mattina 1996b, 33-34).(2) a. ax?PDEM?p?naP.birch.bark.basketThis is a basket.b. ?p?naPbirch.bark.basketax?P.DEMThis is a basket. (A. Mattina 2001, fn11)(3) a. IvanIvan?Qa?t.wetIvan is wet.b. ?Qa?twetIvan.IvanIvan is wet.As mentioned in the introduction, I refer to examples like (2-3) as direct predica-tions. In terms of Higgins? taxonomy, the nominal predications in (2) may be eitheridentificational or predicational.An example of an identificational interpretation of a direct predication is givenin (4). In response to (4a), for example, a speaker may answer with either subject-initial (4b) or subject-final (4c). Here, the speaker may be teaching the questioner220the name by which the referent may be called, in which case the interpretation isidentificational (cf. section 2.1.2). Alternatively, the speaker may be identifyingan individual, denoted by the demonstrative, as belonging to the set of rabbits, inwhich case the interpretation is predicational (cf. section 2.1.2).(4) a. sti ?mwhatix?P?DEMWhat is that?b. ix?PDEMsp@pl?naP.rabbitThat?s a rabbit.c. sp@pl?naPrabbitix?P.DEMThat?s a rabbit.Another example of a predicational interpretation of a direct predication is givenbelow as (5). In response to (5a), a speaker may answer with either subject-initial(5b) or subject-final (5c). Here, the speaker is identifying John as belonging to theset of carpenters.(5) a. sti ?mwhatJohnJohniPDETs-c- ?kw??l-s?NOM-CUST-make-3SG.POSSWhat does John do (for work)?b. JohnJohns@xw- ?kw??l-?xw-@m.OCC-make-house-MIDJohn is a carpenter.c. s@xw- ?kw??l-?xw-@mOCC-make-house-MIDJohn.JohnJohn is a carpenter.A slightly more complex example is given below in (6). Here, the argument DP[is ?l?x?t iP sckw?ns DP], literally ?my friend?s taking?, contains a possessor and a221nominalized headless relative clause.2(6) Context: Your friend went shopping, you ask what they got.a. sti ?mwhat[iP]DETa[n]-s?l?x?t2SG.POSS-friendiPDETs-c-kw?n-s?NOM-CUST-take-3SG.POSSWhat did your friend get?b. s@n ?k?caPsq?x?aPhorse[iP]DET[i[n]-s?l?x?t1SG.POSS-friendiPDETs-c-kw?n-s DP].NOM-CUST-take-3SG.POSSMy friend got a horse.Literally: The thing my friend got is a horse.c. [iP]DET[i[n]-s?l?x?t1SG.POSS-friendiPDETs-c-kw?n-s DP]NOM-CUST-take-3SG.POSSs@n ?k?caPsq?x?aP.horseMy friend got a horse.Literally: The thing my friend got is a horse.Subject-initial and predicate-initial versions appear to be semantically and prag-matically equivalent. Under the assumption that a diagnostic for focus is the answerto a WH-question (Jackendoff, 1972; Selkirk, 1995), and that Okanagan is similarto neighboring Thompson River Salish (Koch, 2008a) in terms of aligning focus toprosodic edges, this means that focus alignment constraints do not apply to directpredications. I will have more to say on this issue in section 7.4.The data shown below as (7-8) show that verbal intransitives, as well as prepo-sitional phrases, have distributions identical to the nominal and adjectival predi-cates in (2-3).32In Okanagan, both possessor and possessum are introduced by iP determiners. Matthewson andDavis (1995, 19) analyze possessive structures in Lillooet as consisting of a possessed DP, whosehead noun is adjoined by the possessor DP. Cases where the possessor precedes the possessum, as in(6) involve possessor scrambling in Lillooet.3I make no claims here about the structure of verbal predications, since it is likely that additionalaspectual projections are involved in these cases.222(7) a. JohnJohn[ ?cqw-aqwV P].cry-REDJohn cried.b. JohnJohn[Payx?wtAP].tiredJohn is tired.c. iPDETs-qwsiP-sNOM-son-3SG.POSS[ ?klLOCMerrittPP].MerrittHis son is in Merritt.d. ix?PDEM[y?mx?waPNP].cedar.bark.basketThat is a basket.e. iPDET??@x?- ??x??pRED-grown[ ?qwQay-lqsNP].black-robeThe old man is a priest.(8) a. [ ?cqw-aqwV P]cry-REDJohn.JohnJohn cried.b. [Payx?wtAP]tiredJohn.JohnJohn is tired.c. [ ?klLOCMerrittPP]MerrittiPDETs-qwsiP-s.NOM-son-3SG.POSSHis son is in Merritt.d. [y?mx?waPNP]cedar.bark.basketix?P.DEMThat is a basket.e. [ ?qwQay-lqsNP]black-robeiPDET??@x?- ??x??p.RED-grownThe old man is a priest.223For Okanagan, both main (9) and subordinated clause (10) direct predicationsallow variable subject-predicate word ordering:4,5(9) a. [SarahSub jSarahm@ ?q-?nkPred].full-stomachSarah is full.b. [m@ ?q-?nkPredfull-stomachSarahSub j].SarahSarah is full.c. [MarySub jMaryx?@-x??saPtPred].RED-prettyMary is pretty.d. [x?@-x??saPtPredRED-prettyMarySub j].MaryMary is pretty.(10) a. kn1SG.ABSlimtglad\u000B[COMP][Sarah Sub jSarahm@ ?q-?nk Pred].full-stomachI?m glad Sarah is full.b. kn1SG.ABSlimtglad\u000B[COMP][m@ ?q-?nk Predfull-stomachSarah Sub j].SarahI?m glad Sarah is full.c. tal?Pverykn1SG.ABSn-stilsn-think\u000B[COMP][Mary Sub jMaryx?@-x??saPt Pred].RED-prettyI think Mary is really pretty.4Complementizers are largely optional in Okanagan (cf. Kroeber (1999) and discussion in section6.1.2), and so it is often difficult to tell whether a small clause is a CP constituent, or a direct com-plement of main clause verb. Just as in English, however, the absence of an overt complementizerdoes not necessarily mean that a functional CP structure is not present.5The Subj ?subject? and Pred ?predicate? labels in examples (9-14) are for expository purposesonly.224d. tal?Pverykn1SG.ABSn-stilsn-think\u000B[COMP][x?@-x??saPt PredRED-prettyMary Sub j].MaryI think Mary is really pretty.Finally, prosodic heaviness of a subject or predicate constituent does not de-termine its surface position in a direct predication. The data in (11-12) show thata prosodically heavier predicate may either follow or precede a relatively lightersubject6, and (13-14) show that a prosodically heavier subject may either follow orprecede a relatively lighter predicate.7,8(11) a. [iPDETylm?xw@m Sub j]chief[x?astgoodtATTRsq@ltm?xw Pred].manThe chief is a good man.b. [x?astgoodtATTRsq@ltm?xw Pred]man[iPDETylm?xw@m Sub j].chiefThe chief is a good man.(12) a. [ix?PSub j]DEM[ ?pis ??P-?xnbroad-shoulderedtATTRylm?xw@m Pred].chiefThats a broad-shouldered chief.b. [ ?pis ??P-?x?nbroad-shoulderedtATTRylm?xw@m Pred]chief[ix?PSub j].DEMThats a broad-shouldered chief.6The predicates in (11-12) are Complex Nominal Predicates, or in other words, attributively mod-ified NPs, which are themselves categorially NP predicates (cf. section 6.2).7The structure in (13) is a focus structure referred to as a Nominal Predicate Construction (NPC)in Davis et al. (2004) and as a ?bare? cleft in (Kroeber, 1999). The predicate nominal is in focus inthese cases, though no focus-movement is involved.8The variable ordering of subject and predicate in direct predications seems reminiscent of thepredicational/specificational alternation seen in English. Unlike English specificational sentences,however, Okanagan direct predications simply ascribe a property to the subject DP. Predicates inOkanagan direct predications are not DPs, and so a specificational analysis of direct predication isnot possible in any case.225(13) a. [syxw?p-m@x Pred]Shuswap-person[iPDET??@x?- ??x??pRED-growniPDETqw@l-qw?l-st-@n Sub j].RED-speak-CAUS-1SG.ERGThe old men that I talked to were Shuswaps.(adapted from Davis et al. (2004))b. [iPDET??@x?- ??x??pRED-grownaPDETc-qw@l-qw?l-st-@n Sub j]CUST-RED-speak-CAUS-1SG.ERG[syxw?p-m@x-@lx Pred].Shuswap-person-3PL.ABSThe old men that I talked to were Shuswaps.(14) a. [s@xw-knxt-?t??n Pred]OCC-help-person[ix?PDEMiPDETtk?milxw Sub j].womanThat woman is a helper.b. [ix?PDEMiPDETtk?milxw Sub j]woman[s@xw-knxt-?t??n Pred].OCC-help-personThat woman is a helper.I now provide some data which show that predicate-initial ordering of directpredications is not derived by predicate raising. These data are relevant becauseboth direct predications and DP-DP structures, where the second DP is a syntac-tically licensed predicate, disallow predicate raising. As such, a ban on syntacticinversion can be understood as a more general property of Okanagan grammar.7.2.2 Direct Predications and the Ban on Predicate RaisingIn this section, I take a look at direct predications in the context of various func-tional particles, which I assume correspond to functional heads in the syntax (Cinque,1999), and show that the predicate-initial ordering is not derived by predicate rais-ing (Moro, 1997).Pre-predicative particles encoding tense, modality, discourse deixis, and otherfunctions commonly introduce Okanagan sentences, including those containing di-rect predications. These particles have traditionally been analyzed as clitics in theSalish literature (Kroeber, 1999). In Okanagan, they attach to the left periphery of226the predicate complex. Since these particles may be analyzed as operators that ap-ply at the propositional level, it is reasonable to assume that they occupy positionshigher than the propositional core. Such particles include:(15) a. cmay- epistemic modalb. mat- epistemic modal9c. cakw- bouletic modald. ha- question markere. ?@m- past tensef. mi- futureFor the subject-initial (a) and predicate-initial (b) examples in (16-19) below,the pre-predicative particles precede the direct predication. The crucial facts tonotice here are that the subject DP can occur before the introductory particle, asshown by the (c) cases, but the predicate cannot, as shown by the (d) cases.(16) a. cmayEPISJohnJohn[ ?cqw-aqw V P].cry-REDJohn might cry.b. cmay [ ?cqwaqw V P] John.c. John cmay [ ?cqw?qw V P].d. *[ ?cqw?qw V P] cmay John.(17) a. cakwBOUL[sysyus AP]activeiPDETs@xw-p?x?-@m.OCC-hunt-MIDA/The hunter should be active.b. cakw iP s@xwpix?@m [siysiy?s AP].c. iP s@x?wp?x?@m cakw [siysiy?s AP].d. *[siysiy?s AP] cakw iP s@xwp?x?@m.9See Menzies (2012) for a semantic analysis of the Okanagan modal system.227(18) a. haYNQix?PDEM[y?mx?waPNP]?cedar.bark.basketIs that a basket?b. ha [y?mx?waPNP] ix?P?c. ix?P ha [y?mx?waPNP]?d. *[y?mx?waPNP] ha ix?P?(19) a. cakwBOULNormanNorman[s@xw- ?kw??l-?xw-@m NP].OCC-make-house-MIDNorman should be a carpenter.b. cakw [s@xw ?kw??l?xw@m NP] Norman.c. Norman cakw [s@xw ?kw??l?xw@m NP].d. *[s@xw ?kw??l?xw@m NP] cakw Norman.Put simply, if the predicate precedes the subject, nothing can intervene betweenthe two. Assuming that particles like those listed in (15) occupy a higher posi-tion in the clause than a base-generated subject and predicate, and in view of theungrammaticality of the (d) cases, it seems clear that the (a,b) cases represent thebase forms, and that the (c) cases are derived by raising the subject DP out of itsbase-generated position.10Gardiner (1993) argues for closely related Shuswap that anything occuring tothe left of the clitic string can be analyzed as either a base-generated external topic,or a movement-derived topic-denoting phrase. Assuming that the subjects in the(c) cases above are topicalized, the implication is that subjects but not predicatescan undergo topicalization movement.10Subject raising as a syntactic phenomenon may also underlie the variable SVO and VSO wordorders displayed in garden-variety transitive sentences. The functional motivation for subject move-ment, if any, remains unclear, and so for now, I treat it solely as a syntactic phenomenon.2287.2.3 (Near) Obligatory Subject-Raising over TTense marking is null for both present and past tenses in Okanagan, but I assumethat a T projection is nevertheless always present, and I argue in this section that Tselects a direct predication as a complement.At first glance, it appears that a direct predication such as (20) is ambiguousbetween a present and past tense interpretation, but there are default readings inSalish languages for eventive predicates depending on aspectual class, and a defaultpresent tense reading for stative predicates (cf. for example N. Mattina (1996b) forOkanagan, Bar-el (2006) for Squamish).11 Thus, the default reading for (20) ispresent tense, John is a teacher. Context will almost always disambiguate thetense in such cases (cf Matthewson (2006b) for Lillooet), and in other cases, thereis a range of temporal adverbials which serve to disambiguate tense.(20) s@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDJohn.JohnJohn is/was a teacher.For future tense, the morpheme mi is sometimes used.12 It occurs in a positionlower than the epistemic modal, as shown by the data in (21) which appear toinvolve subject raising.(21) a. JohnJohncmayEPISmiFUTs@xw-p?x?-@m.OCC-hunt-MIDJohn is gonna be a hunter.b. cmayEPISJohnJohnmiFUTs@xw-p?x?-@m.OCC-hunt-MIDJohn is gonna be a hunter.c. *JohnJohnmiFUTcmayEPISs@xw-p?x?-@m.OCC-hunt-MIDJohn is gonna be a hunter.11See also N. Mattina (1996b, 63-64) for a related discussion of ?neutral? aspect in Okanagan.12It is not obligatory for future interpretations. Verbal futures are more often marked as such by aprefix ks-, perhaps more accurately described as a ?prospective aspect? or a modal. Cf. my analysisof nominal irrealis k?- in chapter 5.229Subject raising over mi appears to be (near) obligatory, as shown in (22) whichdisplays the default pattern. Given that subject raising is not obligatory in thecontext of other pre-predicative particles (see previous section), (22) constitutesevidence that mi is in a relatively low position, perhaps T(ense). Assuming that miis a T head, evidence against an argument whereby a direct predication contains aT projection as part of its basic configuration (i.e. as a predicational small clause)comes from (22c): given that predicate raising is not a possibility, present and pasttense predicate-initial direct predications should also be ungrammatical, but theyare not. In other words, T can select for a direct predication, and it is probablythe case that a direct predication must form a constituent with T at some level,but a predicational small clause does not itself contain T. This view of the relationbetween T(ense) and a predicative small clause fits with generally accepted notionsof small clauses in English (cf. section 2.2.2).(22) a. JohnJohnmiFUTs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn is going to be a teacher.b. *miFUTs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDJohn.JohnJohn is going to be a teacher.c. *s@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDmiFUTJohn.JohnJohn is going to be a teacher.d. *miFUTJohnJohns@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn is going to be a teacher.There are, however, also data that appear to show a subject lower than mi, hencesubject raising is only ?near?-obligatory. (23a) shows that an independent pronounsubject can follow mi, and (23b) shows an independent pronoun subject precedingmi, though an interesting ?doubling? of future-marking optionally occurs in thesecases.230(23) a. miFUTmn?m?t@t1PL.INDEPkwu1PL.ABSxiPt-m?st.run.PL-INTR.REFLEXWe are going to run.b. (mi)FUTanw?2SG.INDEPmiFUTkw2SG.ABSxwuy.goYeah, you go.Taken together, (22-23) support a view whereby T can select for a direct predi-cation, and subject raising is near-obligatory. The ?doubling? effect in (23b) mayreflect an optional T-to-C movement of mi, with an overt trace left behind, and con-texts involving adjunct WH-questions provide independent evidence that mi mayin fact be a complementizer. The data in (24) show that mi is in complementarydistribution with the complementizer kiP, which introduces the residue clause in anadjunct cleft or WH-question. These data show that a raised subject can occupy aposition higher than C (perhaps Spec C).(24) a. *lutNEGc-my-st-inCUST-know-CAUS-1SG.ERGkaPk?nwherekiP / miCOMP / FUTJohnJohnxwuy.goI don?t know where John went/will go.b. lutNEGc-my-st-inCUST-know-CAUS-1SG.ERGkaPk?nwhereJohnJohnkiP / miCOMP / FUTxwuy.goI don?t know where John went/will go.A contrast surfaces between non-embedded subject-final direct predications(25a) and embedded subject-final direct predications (25b) in the context of mi,such that an embedded subject in final position does not obligatorily raise (compare25b with 24a,b).(25) a. *miFUTxwuygoJohn.JohnJohn will go.b. lutNEGc-my-st-inCUST-know-CAUS-1SG.ERGkaPk?nwherekiP / miCOMP / FUTxwuygoJohn.JohnI don?t know where John went/will go.231Speakers indicate that for the grammatical variant of (25a), John mi xwuy, one issingling out a particular individual, and so it could be considered to be a focusstructure, perhaps a kind of cleft. Unraised subjects may be ungrammatical inthis environment because the focus position is left empty. In WH-contexts how-ever (25b), the WH-item is presumably in focus, which may license the subject toremain in-situ. The contrast between (24a) and (25b), in particular, still needs ex-planation, but may provide evidence that the base ordering of an Okanagan directpredication is predicate-initial.7.2.4 A Structural Analysis of Direct PredicationThe ban on predicate-raising, as just discussed, does not itself decide between astructure whereby the predicate and subject DP form a freely ordered small clauseconstituent to the exclusion of any Pred-head (Figure 7.1 below, where F repre-sents one of the pre-predicative particles just discussed), and a structure wherebya null Pred-head selects for a lexical predicate (assuming that rightward subjectsare possible) (Figure 7.2 below). I refer to these two possibilities as the bare smallclause hypothesis and the Pred-head hypothesis, respectively.Figure 7.1: Freely Ordered Bare Small Clause HypothesisFPF?F SCNPPredy?mx?waPDPSub jix?PFPF?F SCDPSub jix?PNPPredy?mx?waP232Figure 7.2: Null Pred-head Small Clause HypothesisFPF?F PredPPred?Pred\u000BNPPredy?mx?waPDPSub jix?PFPF?F PredPDPSub jix?PPred?Pred\u000BNPPredy?mx?waPUnder the bare small clause hypothesis, direct predications are syntacticallysymmetrical (i.e. bare) small clauses (Williams, 1975; Stowell, 1981; Moro, 2000),consisting only of a DP subject and a semantically unsaturated XP predicate (Hig-ginbotham, 1985), where X ? {N,P,A,V}. This theory assumes that the lexical cat-egories themselves are predicative (Stowell, 1981), and that the small clause sub-ject is left-adjoined to the small clause predicate (Manzini, 1983; Heggie, 1988).Under the Pred-head analysis (Bowers, 1993; Baker, 2003; den Dikken, 2006), afunctional head Pred selects for a semantically predicative constituent and a DPargument. The Pred-head is semantically vacuous.There is good evidence that predicate-initial direct predications are not derivedby predicate raising under either analysis (Figure 7.3 below), while subject-raisingis permitted under either analysis (Figure 7.4 below).233Figure 7.3: No Predicate Raising*FPNPiPredy?mx?waPF?F SCDPSub jax?Pti*FPNPiPredy?mx?waPF?F PredPDPSub jix?PPred?Pred\u000BtiFigure 7.4: Subject RaisingFPDPiSub jix?PF?F SCNPPredy?mx?waPtiFPDPiSub jix?PF?F PredPPred?Pred\u000BNPPredy?mx?waPtiSubject raising as depicted in Figure 7.4 shows a subject raising out of a predicate-initial bare small clause or PredP. Under either analysis, this would yield a basepredicate-initial word order for Okanagan, which would coincide nicely with thebasic word order facts of Northern Interior Salish and Central Salish languages(Czaykowska-Higgins and Kinkade, 1998). Subject raising out of a predicate-finalbare small clause or PredP is nevertheless also a possibility.It is empirically unclear whether the bare small clause or PredP hypothesis forOkanagan direct predications is correct. Deciding between the two hypotheses may234reduce to arguments for theoretical economy. The bare clause hypothesis is attrac-tive since it straightforwardly reflects the semantic status of direct predications assimple saturation relations, as depicted below in Figure 7.5. On the other hand,for syntactic theories which adopt the thesis of antisymmetry (Kayne, 1994), thishypothesis faces major syntactic problems since if both subject and predicate aremaximal projections (Rothstein, 1995) and the subject is adjoined to the predicate(Manzini, 1983; Heggie, 1988), then there will be a violation of antisymmetry.13Figure 7.5: Bare Small Clauses As Simple Saturation RelationsSC[[basket]]([[this]])DPSub j[[this]]ax?PNPPred?x[basket(x)]y?mx?waPSC[[basket]]([[this]])NPPred?x[basket(x)]y?mx?waPDPSub j[[this]]ax?PThe Pred-head hypothesis is more in line with current theories of predication(Heycock and Kroch, 1999; Adger and Ramchand, 2003; Baker, 2003; Mikkelsen,2005; den Dikken, 2006), whereby all non-verbal categories must be licensed assyntactic predicates via a null functional Pred-head; however, for Okanagan thisapproach violates Occam?s razor, for two reasons:a. The Pred-head in a direct predication must be both semantically empty, asin the formulation ?P?x.[P(x)] (Partee, 1986), and phonologically null.14b. It requires postulating not one, but two distinct, phonologically null Pred-heads for Okanagan, one for direct predications and one for DP-DP struc-tures, as we shall see.13The structures in Figure 7.5 are not necessarily associated with any c-commanding functionalprojection, though they can always be embedded within a larger structure.14Assuming that linguistic objects are divided into three types of information: phonological, se-mantic, and syntactic information, Wiltschko (2005) argues that ?at least more than half of the infor-mation associated with any given linguistic object? must be fully interpreted, that is, non-expletive.A semantically vacuous, phonologically null copula will be unable to syntactically project underWiltschko?s framework, hence for Okanagan, the Pred-head analysis is ruled out. Although theOkanagan equative head is phonologically null, it has semantic content.235In any case, nothing crucial hinges on which hypothesis is correct since I willshow that direct predications, whatever their base-generated form, are still distinctfrom DP-DP structures.7.2.5 Summary of Direct PredicationsBy way of summary, Okanagan predication is clearly different than in English. En-glish NP, AP, and simple indefinite DP predicates cannot precede their subjects inEnglish copular and small clauses. Moro (1997) appeals to a ?basic directionality?to explain the word order facts for most English predications: Predicates cannotgenerally precede subjects because the basic direction of English predication issubject-initial.15In Okanagan, lexical predicates may easily precede their subjects. If predicate-raising is not a possibility in these cases, as I have argued based on data like (16-19) above, it seems that the predicate must occur in its base-generated position.Because a subject may freely occur before or after a lexical predicate, it seems thateither direct predications are bare small clauses without any basic directionality,or else there is a null Pred-head linking the predicate to the subject, and that thesubject may occur as either a rightward or leftward specifier. Under either analysis,Okanagan direct predications stand in contrast to predicational copular clauses inEnglish.The next section discusses DP-DP structures. I show that a less-referential iPDP cannot precede a more-referential DP, even in the absence of any pre-predicativeparticle. This constitutes a major difference between direct predications and DP-DP structures, and a strong piece of evidence in favor of analyzing DP-DP struc-tures as structurally distinct from direct predications.7.3 DP-DP StructuresAs the term suggests, DP-DP structures are sentences involving two DPs. Afterdiscussing basic DP-DP structure data, I discuss a word order restriction whichclearly sets DP-DP structures apart from the direct predications discussed imme-15Recall from 2.2 that this pattern cannot be due to the presence or absence of a copula, sincecopular clauses and small clauses display the same pattern in this respect.236diately above in section 7.2. I suggest that the word order restriction ultimatelyderives from the fact that DPs cannot be predicates (Longobardi, 1994; Matthew-son, 1998). From this, I build an argument that DP-DP structures are semanticallyequative (Heycock and Kroch, 1999), and that there is a null functional head whichencodes the equative relation (i.e. Id-head). I give a preliminary semantic analysisbefore moving on to important information structural properties of DP-DP struc-tures, which suggest a slightly more complex view of equation in Okanagan.7.3.1 DP-DP Structures and the Word Order RestrictionExamples of canonical Okanagan DP-DP structures are shown below in (26) (cf.also N. Mattina (1996b, 30)). In (26a), for example, the demonstrative ix?P ?that,he, she? is the first DP, and iP p@ptw?naxw ?the old woman? is the second DP. (26c-d) show that in addition to demonstratives, proper names and iP DPs may alsooccur initially in a DP-DP structure. As the translations suggest, DP-DP structuresencompass interpretations analogous to English predicational and equative copularsentences (Higgins, 1973, 1979).16(26) a. [ix?PDP]DEM[iPDETp@ptw?naxw DP].old.ladyShe is an/the old lady.b. [ax?PDP]DEM[iPDET?p?naPDP].birch.bark.basketThis is a/the basket.c. [John DP]John[iPDETs@xw-mr?m-@m DP].OCC-medicine-MIDJohn is a/the doctor.d. [iPDETsq@ltm?xw DP]man[iPDETs@xw-p?x?-@m DP].OCC-hunt-MIDThe man is a/the hunter.16(N. Mattina 1996b, 30) says that examples like 27-28 ?consist of two adjacent NPs [(DPs)]standing in an equivalence relationship interpreted as ?NP = NP? ([DP = DP]). Equational sentenceshave neither a lexical verb nor a copula.?237The data in (26) are problematic since there is no obvious predicate, and it is afundamental property of Salish that DPs cannot be predicates (Matthewson, 1998).Moreover, there is independent evidence that Okanagan DPs are not predicates: aword-order restriction surfaces in answers to WH-questions such that a demonstra-tive or proper name DP cannot follow an iP DP (27-28). Thus, only (27c) and (28c)are possible as answers, not (27d) or (28d).(27) a. switwhoix?P?DEMWho is she?b. ix?PDEMhaYNQtOBLp@ptw?naxw?old.ladyIs she the old lady?c. ix?PDEMiPDETp@ptw?naxw.old.ladyShe is the old lady.d. *iPDETp@ptw?naxwold.ladyix?P.DEMThe old lady is her.(28) a. switwhoix?PDEM?@COMPSpike?SpikeWho is Spike?b. switwhoiPDETylm?xw@m?chiefWho is the chief?c. SpikeSpikeiPDETylm?xw@m.chiefSpike is the chief.d. *iPDETylm?xw@mchiefSpike.SpikeThe chief is Spike.238Analyzing the demonstratives in (26a,b) and (27c), the proper name in (26c)and (28c), and the initial iP DP in (26d) as predicates in these contexts would bein keeping with the broad generalization that Salish languages are predicate-initial(cf. Kroeber (1991, 26) and Czaykowska-Higgins and Kinkade (1998, 37)). I claimthat these DPs are not predicates, however, based on the fact that while sentenceswith lexical predicates (i.e. direct predications) allow for variable word order of thepredicate and subject argument, DP-DP structures do not allow variable orderingof the two DPs in these contexts. This furthermore suggests that Okanagan DPs donot freely type shift into predicative functions (Partee, 1986).My reasoning against analyzing any DP in data such as (26-28) as a predicateis as follows: if the demonstrative or proper name were a predicate in (27c,28c),then the prediction is that an iP DP could function as an argument expression in(27d,28d) on analogy with the direct predication data: but this is not the case.Similarly, if the iP DP were a predicate in (27c,28c), then the prediction is thatthe demonstrative or proper name could function as an argument expression in(27d,28d), but this too is not possible.There are also restrictions on DP-DP structures involving two iP DPs, as in (29)below. The case of (29) is slightly more complex than that of (27-28), since whiletwo simple iP DPs may occur in either order, there is nevertheless an interpretiverestriction: the first DP must be more referential than the second DP. This meansthat DP-DP structures are clearly asymmetrical, unlike direct predications.(29) a. iPDETsq@ltm?xwmaniPDETs@xw-p?x?-@m.OCC-hunt-MIDThe man is/was a hunter.b. iPDETs@xw-p?x?-@mOCC-hunt-MIDiPDETsq@ltm?xw.man(i) The hunter is a man.(ii) #The hunter was a man.(iii) *The man is a hunter.(iv) *The man was a hunter.The fact that the initial DPs in (29) must be more referential than the second DPs239is indicated by the infelicity of a past-tense interpretation of (29b). To explain,consider that simple direct predications and DP-DP structures have both presentand past tense interpretations available.17 Under the past-tense interpretation of(29b,ii), the sentence is only interpretable under the somewhat unusual readingthat the referent of the DP iP s@xwp?x?@m ?the hunter? was, but is no longer ?a man?.This issue does not arise for a past tense interpretation of (29a), since it is perfectlyreasonable for a man to cease being a hunter. (29b,iii-iv) show that the final DPmay not be more referential than the first DP. By way of reminder, note that forthe direct predication corresponding to (29b), where the initial determiner preced-ing s@xwp?x?@m ?hunter? is absent, the final DP iP sq@ltm?xw ?the man? must be thereferential subject.In sum, DP-DP structures exhibit a word order restriction such that (i) the ini-tial DP must be more referential than the final DP; and (ii) a demonstrative orproper name, if present, must precede an iP DP. This generalization can be infor-mally described as a requirement that a directly referential DP (i.e. a proper nameor demonstrative), if present, must precede a non-directly referential DP (i.e. aniP DP), and that in cases involving two iP DPs, the first iP DP must be ?morereferential? than the second. Given that directly referential DPs are more refer-ential, in some sense, than iP DPs, the overall generalization is as follows: themore-referential DP must come initially.187.3.2 No ?Predicate? Raising in Okanagan DP-DP StructuresAn important, unifying feature of both direct predications and DP-DP structuresin Okanagan is the absence of predicate raising (Moro, 1997; den Dikken, 2006).For direct predications, the ban on predicate raising only becomes evident in thecontext of a pre-predicative particle (cf 16-19), while for DP-DP structures, it is17This is generally the case for Salish non-verbal predicates in the absence of overt tense or as-pectual morphology (Matthewson, 2006b). There are various strategies to disambiguate present andpast tense readings, when necessary. For Okanagan, these include the customary/habitual aspectualprefix ac-, which favors a present-tense interpretation, and past tense adverbials such as ??@m ?past?and ?qs?pi ?long ago?.18Heller (2005) couches similar observations in terms of discriminability, where English speci-ficational sentences exhibit a rising discriminability (i.e. the second DP is always more referentialthan the first). Okanagan DP-DP structures, in contrast, exhibit a ?falling discriminability? (i.e. thesecond DP is always less referential than the first).240immediately evident (28), since a less-referential iP DP can never precede themore-referential one. For the sake of completeness, I include examples of DP-DPstructures in the context of pre-predicative particles (30-33), these being roughlyparallel to the direct predications introduced above as (16-19).The (a) and (b) examples below show that DP-DP structures with an in-situ,more-referential DP (a cases) and those with a raised more-referential DP (b cases)are both possibilities. The (c) and (d) cases are both ungrammatical because a less-referential DP cannot precede the more-referential DP in a predicational context.Assuming that the less-referential iP DPs in these examples are somehow licensedas predicates by a Pred-head, then we can make the categorical claim that inverseordering derived by predicate-raising (Moro, 1997; den Dikken, 2006) is categori-cally banned in Okanagan:(30) a. wa ?yyesJohnJohn[iPDETs@xw- ?maP- ?m?yaP-m DP].OCC-RED-teach-MIDYes, John is the teacher.b. John wa ?y [iP s@xw ?maP ?m?yaPm DP].c. *[iP s@xw ?maP ?m?yaPm DP] wa ?y John.d. *wa ?y [iP s@xw ?maP ?m?yaPm DP] John.(31) a. matEPIScmayEPISJohnJohn[iPDETs@xw-p?x?-@m DP].OCC-hunt-MIDMaybe John is the hunter.b. mat John cmay [iP s@xwp?x?@mDP].c. John mat cmay [iP s@xwp?x?@mDP].d. *[iP s@xwp?x?@m DP] mat cmay John.(32) a. na ?kw@mEVIDJohnJohn[iPDETs@xw-p?x?-@m DP].OCC-hunt-MIDI guess John was the hunter.b. John na ?kw@m [iP s@xwp?x?@m DP].241c. *na ?kw@m [iP s@xwp?x?@m DP] John.d. *[iP s@xwp?x?@m DP] na ?kw@m John.(33) a. ??@mPASTJohnJohn[iPDETs@xw- ?maP- ?m?yaP-m DP].OCC-RED-teach-MIDJohn used to be the teacher.b. John ??@m [iP s@xw ?maP ?m?yaPm DP].c. * ??@m [iP s@xw ?maP ?m?yaPm DP] John.d. [*iP s@xw ?maP ?m?yaPm DP] ??@m John.In sum, the implication from the Okanagan direct predications and DP-DP structuredata is that it is not possible to move the predicate of a direct predication or a less-referential DP in a DP-DP structure out of a small clause. Direct predications andDP-DP structures nevertheless differ by the fact that a lexical predicate can precedea subject in a direct predication, but a less-referential DP cannot precede a more-referential DP in a DP-DP structure. Given that this asymmetry cannot be dueto predicate raising in direct predications, I take this as evidence for two distinctanalyses for direct predications and DP-DP structures.Before moving on to the issue of how DP-DP structures are semantically com-posed, I briefly discuss the issue of testing for connectivity effects in OkanaganDP-DP structures.7.3.3 Connectivity (and Other) Effects and Okanagan DP-DPStructuresConnectivity effects, as discussed in section 2.2.3, are a hallmark of specificationalpseudoclefts and specificational copular clauses, which under analyses such as thatof Heycock and Kroch (1999) are equative rather than inverted predicational cop-ular clauses. It is not straightforwardly possible to test for connectivity effects inOkanagan, or other distinguishing properties of specificational sentences such aspronominalization (Mikkelsen, 2005), for the following reasons:242(34) a. Okanagan (and the rest of Salish) lacks WH-relative clauses, and so Okana-gan has nothing comparable to WH-pseudoclefts in English.b. 3rd person pronouns are normally null, and reflexivization is an operationon the predicate, not on an argument.c. Okanagan like other Salish languages (Davis, 2006, 2009) regularly vio-lates condition C.d. There is no animate/inanimate or gender/non-gendered split in the pronom-inal system, so it is not possible to test for pronominalization. Tag ques-tions, which provide one test for pronominalization, are independently notpossible in Okanagan.Regarding (34a), Okanagan does not have WH-relative clauses, which for En-glish are important for showing connectivity effects. Nevertheless, recall from sec-tion 2.2.3 that English specificational th-pseudoclefts (e.g. The thing that Harveyidid next was wash himselfi thoroughly) also show connectivity effects. Okana-gan does employ headless relative clauses in argument positions, and so these DPscould be argued to correspond directly to th-pseudoclefts in English. Examples ofDP-DP structures involving headless relative clauses are given in (35); however(36) shows that a specificational word ordering is uniformly ungrammatical.19,20(35) a. ix?PDEMiPDETkwu1SG.ABSwik-s.see-(DIR)-3SG.ERGThat?s the one that saw me.b. JohnJohniPDETkwu1SG.ABSwik-s.see-(DIR)-3SG.ERGJohn is the one that saw me.19The data in (35) exemplify what I refer to as ?clefts?. I leave off an in-depth discussion of thesetypes of data until chapter 8.20Technically speaking, it should be possible for DP-DP structures consisting of two headless-relative clause DPs, each introduced by iP, to exist. These would be analogous to English ?The onewho hit me is the one who chased me? or ?What I don?t like is what John doesn?t like?. I have tried toelicit examples such as these as well as other types of higher order equatives in Okanagan, but noneof these appear to be possible.243c. iPDETsq@ltm?xwmaniPDETkwu1SG.ABSwik-s.see-(DIR)-3SG.ERGThe man is the one that saw me.(36) a. *iPDETkwu1SG.ABSwik-ssee-(DIR)-3SG.ERGix?P.DEMThe one that saw me is him.b. *iPDETkwu1SG.ABSwik-ssee-(DIR)-3SG.ERGJohn.JohnThe one that saw me is John.c. *iPDETkwu1SG.ABSwik-ssee-(DIR)-3SG.ERGiPDETsq@ltm?xw.manThe one that saw me is the man.Regarding (34b), it is not straightforwardly possible to test for Condition Aconnectivity effects in Okanagan because reflexives in Okanagan are verbal suf-fixes, not independent words, and because predications involving reflexivizationpattern like any other direct predication, in the sense that they allow the subject toprecede or follow the predicate.21(37) a. MaryMarys- ?k@[s]t-m-nc?t-x.NOM-blame-MIN-REFLEX-INTRMary is blaming herself.b. s- ?k@[s]t-m-nc?t-xNOM-blame-MIN-REFLEX-INTRMary.MaryMary is blaming herself.In contexts involving exhaustivity, a cleft may be used (38a), but the reflexive pred-icate (in this case, a headless relative clause predicate) may not precede the focusedDP (38b). While it is possible that (38b) is ungrammatical due to the fact that thereflexive is neither bound nor c-commanded by its antecedent ?John?, data such as(36) indicate that it is a more general property of Okanagan grammar that specifi-cational word order is not possible, and not contingent on any particular binding21There is no audible /s/ in the root ?k@st ?bad? in (37-38) below for reasons I cannot determine.244condition.(38) a. ix?PDEMJohnJohnaPDETc- ?k@[s]t-m-nc?t,CUST-blame-MIN-REFLEXlutNEGSue.SueIt?s John who is blaming himself, not Sue.b. *ix?PDEMaPDETc- ?k@[s]t-m-nc?tCUST-blame-MIN-REFLEXJohn,JohnlutNEGSue.SueThat one who is blaming himself is John, not Sue.It is worthwhile noting that attempting to elicit examples of specificational pseudo-clefts analogous to often cited English examples results in uniformly ungrammati-cal judgements:(39) a. *iPDETcawt-sdoing-3SG.POSSMaryMaryiPDETq?c@lx.runTarget: What Mary did was run.Literally: The doing of Mary was run.b. *iPDETcawt-sdoing-3SG.POSSJohnJohniPDETcaP-cQ?-lx-(aPx).RED-bathe-(INTR)Target: What John did was wash himself.Literally: The doing of John was washing.Regarding (34c), some speakers of Okanagan allow violations of Binding Con-dition C under certain circumstances, though this phenomenon is not well-studiedfor Okanagan. (See Davis (2006, 2009) for a discussion of Condition C in Lil-looet. The examples below were adapted from this work.) To illustrate, in (40b)and (41b), an embedded R-expression is co-valued with a c-commanding pronoun.This considerably complicates attempts to test for Condition C connectivity effects.(40) a. MaryMarys-cut-xNOM-say-INTRks-Paws-?@??t-m?xaPxFUT-go-fish-INCEPT?aPCOMPx?l?p.tomorrowMaryi said proi was going fishing tomorrow.Literally: Maryi said shei was going fishing tomorrow.245b. s-cut-xNOM-say-INTRks-Paws-?@??t-m?xaPxFUT-go-fish-INCEPTMaryMary?aPCOMPx?l?p.tomorrowproi said Maryi was going fishing tomorrow.Literally: Shei said Maryi was going fishing tomorrow.(41) a. BillBilln-stilsthinkcmayEPIS??xwupwiniPDETlLOCs- ?qw@- ?qw? ??aP-x@n.NOM-RED-race-footBilli wondered if proi would win the race.Literally: Billi wondered if hei would win the race.b. n-stilsn-thinkcmayEPIS??xwupwinBillBilliPDETlLOCs- ?qw@- ?qw? ??aP-x@n.NOM-RED-race-footproi wondered if Billi would win the race.Literally: Hei wondered if Billi would win the race.Regarding (34d), there are other diagnostics in the literature on copular clauseswhich are used to distinguish specificational from predicational and equative sen-tences in English. The pronominalization test (Mikkelsen, 2005), for example, in-volves a tag question which includes a subject-referring pronoun. For predicational(42a) and equative sentences (42b) in English, the pronoun in the tag question mustbe gendered. For specificationals (42c) however, the pronoun must be neuter. Theargument is that predicational and equative sentences allow [+human] subjects,whereas specificational sentences do not.(42) data from Comorovski (2007)a. Susan is a violinist, isn?t she / *it?b. Aurore Dupin is George Sand, isn?t she / *it?c. The (female) winner is Susan, isn?t it / ?she?For Okanagan, however, the third person absolutive pronominal agreement in Okana-gan is null, while the third person ergative occurs as -(i)s on a predicate inflectedas transitive (cf. section 3.2). Both absolutive and ergative agreement morphol-ogy may optionally be accompanied by an overt demonstrative ix?P, however theform of ix?P is invariant, and not affected by the animacy or gender of a particularreferent.246Tag questions are independently not possible in Okanagan. (43a) was volun-teered as equivalent to an example of an English tag question, however the fact thatit has two occurrences of the yes-no question particle ha indicates that it actuallyconsists of a series of two questions. Removing the first occurrence of ha (43b)is ungrammatical. In any case, the null third person pronominal argument of theintransitive predicate un?xw ?true? most likely refers to an implied proposition (theanswer to the question), rather than to Susan.(43) a. ix?PDEMhaYNQShush?n,SusanhaYNQwn?xw?trueTarget: That?s Susan, isn?t it? (adapted from Mikkelsen (2005, 121))Literally: Is that Susan? Is it true?b. *ix?PDEMShush?n,SusanhaYNQwnixw?trueTarget: That?s Susan, isn?t it? (adapted from Mikkelsen (2005, 121))In sum, although binding conditions are not well understood for Okanagan, itdoes not seem possible to test for connectivity effects in Okanagan (35-41), or todistinguish copular sentence types in Okanagan using tests such as pronominal-ization (43). As far as connectivity effects are concerned, the facts for Okanaganactually simplify an equative analysis, since there is no need, for example, to positmultiple levels of LF in order to reconstruct binding relations (Heycock and Kroch,1999).7.3.4 An Equative HeadThe evidence suggests that DP-DP structures like (27-29) cannot be analyzed asstructurally on a par with direct predications. Given that neither of the DPs ina DP-DP structure is a predicate, yet the structures are well-formed, I claim thatDP-DP structures are projections of a null functional head that equates the two DParguments (Heycock and Kroch, 1999). The basic idea (though oversimplified) isshown below in (7.6). A null equative head (Id for ?identity?) converts its first DPargument into an equational predicate before taking its second DP argument:247Figure 7.6: An Oversimplified Equative AnalysisIdP? f (?x[basket(x)?C(x)]) = thisDPSub jthisax?PId?? z[ f (?x[basket(x)?C(x)]) = z]Id?y? z[y = z]=DPPredf (?x[basket(x)?C(x)])D?P[ f (?x[P(x)?C(x)])]iPNP?x[basket(x)]y?mx?waP?This is a/the basket?Analyzing the Okanagan equative head simply as ?y? z[y = z] (Geist, 2007) isproblematic, however. First, it incorrectly predicts that any DP-DP structure shouldbe possible (e.g. 27b,28b), since any individual-denoting expression can functionas the complement to the equative head. Second, even under a (correct) stipulationthat only iP DPs can function as equative head complements, the other, more-referential DP might still linearly follow the predicate DP, allowing for rightwardspecifiers (cf Figure 7.2 above), but this cannot be a possibility. I therefore rejectthis analysis.To resolve these issues, I claim that the equative head must be sensitive towhether a DP is extensional-only (directly referential) or allows intensional read-ings (Romero, 2005; Comorovski, 2007). The extensional DP must occur left-mostdue to the fact that the equative head assigns a feature ?F? to its second argument(the extensional DP), and focus alignment constraints require that focus occurs tothe left (Koch, 2008a). I will discuss each of these points in a separate section.I first discuss information-structural properties of DP-DP structures and how they248differ from direct predications, before giving my final semantic analysis of theequative head.7.4 Information Structure and DP-DP structuresThis section investigates the following information structural properties of Okana-gan DP-DP structures:(44) DP-DP structuresa. The second iP DP in a DP-DP structure introduces an implicature of ex-haustivity (not an entailment) (cf. Davis et al. (2004) for clefts in Lillooetand Northern Straits).b. There is no presupposition associated with using a DP-DP structure.c. The initial DP in a DP-DP structure is interpreted as a focus, and a DP infocus cannot occur finally (as shown for both clefts and direct predicationsin Thompson River Salish by Koch (2008a)).It is worthwhile to point out that Okanagan clefts (and clefts in other Salish lan-guages) also show these properties. This reinforces a connection between DP-DPstructures and clefts, which I will discuss in chapter 8.The information structural properties of DP-DP structures differ in certain waysfrom those found in direct predications:(45) Direct predicationsa. Direct predications do not imply exhaustivity.b. Focus in a direct predication may occur either initially or finally (cf. sec-tion 7.2.1).I begin with a discussion of (44a).2497.4.1 The Exhaustivity Implicature in DP-DP structuresFor Okanagan DP-DP structures, the final iP DP introduces an implicature of ex-haustivity to a DP-DP structure. The implicature of exhaustivity stems from thefact that iP carries an implicature of maximality, given again below as (46) (cf.chapter 5).(46) Maximality implicature of iP:f = MAXBy way of illustrating, for (47a) the implicature is that iP p@ptw?naxw denotes themaximal, contextually salient ?old lady?. This implicature is missing from the cor-responding direct predication, as in (47b) where p@ptw?naxw is a nominal predicate.(47) a. [ix?PDP]DEM==[iPDETp@ptw?naxw DP].old.ladyShe is the old lady.b. [ix?PDP]DEM[p@ptw?naxw NP].old.ladyShe is an old lady.If an equative head intervenes between the two DPs in (47a), then the sentence as awhole will carry a secondary implicature that ix?P ?she? is the only individual equiv-alent to the referent of iP p@ptw?naxw, hence ix?P may be interpreted exhaustively.This derived implicature will be absent from (47b), since there is no iP determinerbefore the final NP, and thus no equative head. The exhaustivity implicature carriedby a DP-DP structure may be expressed as follows (48):(48) Exhaustivity Implicature:A sentence of the form [x DP] = [iP Y DP]a. Asserts:? f .x = f (Y )b. Via (46), this implicates:x = MAX(Y )250c. And assuming (48b) is satisfied, a DP-DP structure asserts:x is the only Y.Expanding upon (48c), the reasoning is that if iP Y denotes a maximal individual,then there are no other individuals denoted by iP Y with which some individualbesides x might be equated, hence x is the only Y. The exhaustivity expressed in(48c) is an entailment of (48b), however since (48b) is itself an implicature, and theentailment of (48c) is dependent on an implicature being satisfied, the exhaustivityexpressed in (48c) is ultimately an implicature of an equative sentence.The data support this general picture, since in contexts which require exhaustiv-ity, only DP-DP structures can be used. To illustrate, the question in (49a) requiresan exhaustive answer, since the question is asking which of a defined group of peo-ple are carpenters.22 In such contexts, a DP-DP structure (49b) is felicitous, whilea direct nominal predication (49c) is not.23(49) a. switwhoitl?PDEMtOBLs@xw- ?kw??l-?xw-@m?OCC-make-house-MIDWhich ones of those people are carpenters?b. BillBillnaP?CONJJohnJohnu?CONJSteveSteveiPDETs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDBill, John and Steve are the carpenters.c. #s@xw- ?kw??l-?xw-@mOCC-make-house-MIDBillBillnaP?CONJJohnJohnu?CONJSteve.SteveBill, John and Steve are carpenters.In (49b), ?Bill, John and Steve?, as a conjoined DP, is interpreted exhaustively inthat it identifies the exhaustive subset of individuals denoted by the DP iP s@xw ?kw??l-?xw@m ?the carpenters?. In other words, following (48), if iP s@xw ?kw??l?xw@m denotes22I abstract away from the semantics and pragmatics of questions, and whether or not answers toquestions are uniformly exhaustive (cf. Karttunen (1977)). There may be a formal distinction to bemade between exhaustivity as introduced by an iP DP, and pragmatic exhaustivity as found in Q/Acontexts, but answering this question goes beyond the scope of this thesis.23The alternation between iP and t in question contexts such as (49a) is not well understood,though there may be a subtle semantic difference, possibly related to the alternation between iP andt in cleft contexts (cf. section 8.5).251a maximal plural individual, then the sentence asserts that the sum of ?Bill, Johnand Steve? is identical to this maximal plural individual. The sentence entails thatBill, John, and Steve are carpenters, and assuming that the maximality is satisfied, itimplies that Bill, John and Steve are the only carpenters. In contrast, ?Bill, John andSteve? in (49c) is not interpreted exhaustively, since this sentence simply ascribesthe property of being a carpenter to three separate individuals.A similar example is given below as (50), where a forced-choice question re-quires an exhaustive answer (50b). While ?Wilford? answers the question given in(50c), it is not interpreted exhaustively as required by the context.24(50) a. switwhoiPDETs@xw- ?kw??l-?xw-@m?OCC-make-house-MIDhaYNQSpikeSpike?k@mCONJWilford?WilfordWho is the carpenter, Spike or Wilford?b. WilfordWilfordiPDETs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDWilford is the carpenter.c. #WilfordWilfords@xw- ?kw??l-?xw-@m.OCC-make-house-MIDWilford is a carpenter.Another example is given below as (51):(51) a. haPk?nwhichiPDETtlLOCp@t-p@ptw?naxwRED-old.womaniPDETqw@l-qw?l-st-@m-s?speak-CAUS-2SG.ABS-3SG.ERGWhich one of the old ladies talked to you?b. ix?PDEMiPDETp@ptw?naxw.old.womanThat?s the old lady.24It may still be true in the case of (50c) that Wilford just so happens to be the only individual withthe property of being a carpenter, but as a direct predication this is not implied by (50c).252c. #ix?PDEMp@ptw?naxw.old.womanThat?s an old lady.d. #p@ptw?naxwold.womanix?P.DEMThat?s an old lady.The subjects of direct predications like (49c), (50c) and (51c,d) are not interpretedexhaustively. This illustrates a semantic distinction between subject DPs in directpredications versus more-referential DPs in DP-DP structures.Notice that in answer to the equivalent questions in English, the answer doesnot have to assume a special morpho-syntactic form. Consider that in answer to En-glish (52a) below, both (52b) and (52c) are possible answers (intonational promi-nence is roughly indicated by bold type). The intonation which (52b) has in answerto (52a), however, will be different than its intonation in answer to a question like(53a):(52) a. Which ones of those people are carpenters?b. Bill, John and Steve are carpenters.c. Bill, John and Steve are the carpenters.(53) a. What do Bill, John and Steve do for work?b. Bill, John and Steve are carpenters.c. *Bill, John and Steve are carpenters.The point here is that in English, exhaustivity can be signaled via intonation. Ifintonation were also able to signal exhaustivity in Okanagan (which at least im-pressionistically speaking, does not seem to be the case), then the prediction is thatdirect predications should be felicitious in exhaustive contexts, but this is not thecase. Okanagan is different than English in the sense that a specific syntactic form(a DP-DP structure) must be used in these cases. In other words, in the absence of253intonation cues, exhaustivity must be signalled by the syntax.The exhaustivity implicature associated with DP-DP structures is easily can-cellable. The first DP-DP structure in (54a) carries an implicature that Bill is theonly doctor, but the second DP-DP structure cancels this implicature. In fact, anumerical adverb like knaqs ?one, another? can occur internal to the second DP(54b).25(54) a. BillBilliPDETs@xw-mr?m-@m,OCC-medicine-MIDu?CONJJohnJohnnixwalsoiPDETs@xw-mr?m-@m.OCC-medicine-MIDBill is a doctor, and John is a doctor too.b. BillBilliPDETs@xw-mr?m-@m,OCC-medicine-MIDu?CONJJohnJohniPDETknaqsanotheriPDETs@xw-mr?m-@m.OCC-medicine-MIDBill is a doctor, and John is another doctor.Cancellability of the exhaustivity implicature follows independently from the factthat two occurrences of identical iP DPs can be non-co-referent (e.g. in contextswhen f is existentially bound; cf. chapter 5). That is, in contexts where an iP DPdenotes a non-maximal individual, the implicature of maximality is cancelled, andnon-co-reference is possible.In answer to questions which do not involve subject DP focus (55a), either aDP-DP structure (55b) or a direct predication (55c,d) can be felicitously used asan answer. The exhaustivity implicature carried by (55b), that Mary is the onlyteacher, is cancelled since the second DP, iP s@xw ?maP ?m?yaPm, denotes a non-maximal individual in this context. (55b) is an example of what I refer to as apseudo-predicational reading of a DP-DP structure: when the exhaustivity impli-cature is cancelled, the DP-DP structure is pragmatically equivalent to a directpredication (55d,e), yet semantically equative.25I tentatively assume that knaqs in (54b) is a pre-posed relative clause modifier of s@xwmr?m@m?doctor?, based on arguments made in chapter 6.254(55) a. sti ?mwhatMaryMaryaPDETc- ?kw??l-st-s?CUST-make-CAUS-3SG.ERGWhat does Mary do for work?b. MaryMaryiPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDMary is a teacher.c. MaryMarys@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDMary is a teacher.d. s@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDMary.MaryMary is a teacher.A final piece of evidence that the initial DP in a DP-DP structure receives anexhaustive interpretation comes from data like (56). The context set up by thequestion in (56a) requires an exhaustive answer. The DP-DP structure in (56b)implies that Bill is the only teacher in the set which includes Bill and John, but thedirect predication in (56c) simply ascribes the property of being a teacher to Bill,without referencing the particular set of teachers currently under discussion.(56) a. haYNQBillBillnaP?CONJJohnJohniPDETs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDiPDETlLOCs@n- ?q@ ?y-m?n-t@n?LOC-write-INSTR-INSTRAre Bill and John the teachers at this school?b. lut,NEGBillBilliPDETs@xw- ?maP- ?m?yaP-m,OCC-RED-teach-MIDlutNEGJohn.JohnNo, Bill is the teacher, not John.c. #lut,NEGBillBills@xw- ?maP- ?m?yaP-m,OCC-RED-teach-MIDlutNEGJohn.John#No, Bill is a teacher, not John.255In sum, we have arrived at the following important generalization: DP-DPstructures must be used in contexts which require an exhaustive interpretation,while in contexts which do not require an exhaustive interpretation, both DP-DPstructures and direct predications may be used. Assuming that Okanagan lacks in-tonational cues to signal exhaustivity (cf. Koch (2008a) for Thompson), this is asexpected. The exhaustivity implicature is derivable from the maximality implica-ture carried by the determiner iP, and is cancellable. When satisfied, the initial DPis interpreted as the only individual equivalent to the referent of the second DP, andis therefore interpreted exhaustively.7.4.2 DP-DP Predications are Non-presuppositionalDP-DP structures in Okanagan do not carry a presupposition of existence. Thisis most clearly shown by the fact that DP-DP structures can be used in out-of-the-blue contexts (57). If (57-58) carried a presupposition of existence, i.e. Thereis a carpenter and There is a cleaner respectively, then the prediction is that theseshould be infelicitous in out-of-the-blue contexts, but they are perfectly acceptable.(57) Context: at the beginning of a story.?qs?pilong.agoSpikeSpikeiPDETs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDLiterally: Long ago, Spike was a carpenter.(In Context: Long ago, there was a carpenter named Spike.)(58) Context: I come over and walk in and am looking around, clearly amazedat how clean everything is. You say:JerryJerryiPDETs@xw-xw ?kw-?m.OCC-clean-MIDJerry is the cleaner.This property of DP-DP structures is not especially surprising, given that iP de-terminers do not carry a presupposition of existence (cf. section 5.2.2). (57-58)are also instances of pseudo-predicational interpretations of DP-DP structures, asdiscussed in the previous section. The exhaustivity implicature normally carriedby a DP-DP structure is cancelled via the fact that the second iP DP denotes a256non-maximal referent in these contexts.The absence of any presupposition of existence is also a property of Okanaganclefts, as will be discussed in chapter 8.7.4.3 An F-marked Constituent Must Occur Initially in a DP-DPstructureImportantly, in DP-DP structures, the more-referential DP must come initially. Toillustrate, in answer to (59), one could answer either (60a) or (61a), with ix?P ?that?or Spike receiving an exhaustive interpretation, but not (60b,61b) where the demon-strative or proper name is occurring in final position.(59) a. switwhoiPDETs@xw-p?x?-@m?OCC-hunt-MIDWho is the hunter?(60) a. ix?PDEMiPDETs@xw-p?x?-@m.OCC-hunt-MIDThat?s the hunter.b. *iPDETs@xw-p?x?-@mOCC-hunt-MIDix?P.DEMThat?s the hunter./The hunter is him.(61) a. SpikeSpikeiPDETs@xw-p?x?-@m.OCC-hunt-MIDSpike is a/the hunter.b. *iPDETs@xw-p?x?-@mOCC-hunt-MIDSpike.SpikeThe hunter is Spike.This finding is significant since it suggests that the information structural equiva-lents to English specificational copular sentences (Higgins, 1973, 1979), for exam-ple The hunter is Spike where the DP in final position may be intepreted exhaus-tively, are not possible in Okanagan. Recall that while the English specificationalcopular sentence in (62c,63c) is infelicitous as an answer to (63a), it is felicitous257as answer to (62a), unlike Okanagan (61b).(62) a. Who is the hunter?b. Spike is the hunter.c. The hunter is Spike.(63) a. Who is Spike?b. Spike is the hunter.c. #The hunter is Spike.For DP-DP structures containing a demonstrative (or proper name) and an iPDP, the former must always precede the latter, regardless of the WH-question.26 Toillustrate, the questions in (64a) and (65a) make the iP DP in the replies in (64b,c)and (65b,c) a focus, but the iP DP still cannot precede the demonstrative or propername.27(64) a. switwhoix?P?DEMWho is that?b. ix?PDEMiPDETs@xw-p?x?-@m.OCC-hunt-MIDThat?s the hunter.c. *iPDETs@xw-p?x?-@mOCC-hunt-MIDix?P.DEMThat?s the hunter./The hunter is him.26Though in the absence of any WH-question, apparent inversions of this pattern are possible. Ileave off discussion of these until 7.7, since there is reasonable doubt that these inversions involve anexhaustivity implicature, F-marking, or an equative head.27The complementizer ?@ in (65) is sometimes used in identificational sentences, preceding aproper name. Speakers often translate this use of ?@ as instantiating a relation such as ?x is called y?or ?x is named y?. The optional use of the complementizer in this environment supports an argumentwhereby proper names in these cases are predicates, ascribed to a contextually salient entity. Seesection 7.6.1 for further discussion of Okanagan identificational sentences.258(65) a. switwho?@COMPSpike?SpikeWho is Spike?b. SpikeSpikeiPDETs@xw-p?x?-@m.OCC-hunt-MIDSpike is a/the hunter.c. *iPDETs@xw-p?x?-@mOCC-hunt-MIDSpike.SpikeSpike is the hunter./The hunter is Spike.Recall that there is nothing inherently wrong with an initial iP DP functioning as afocus just in case the second DP is also introduced by iP, and is not a proper nameor demonstrative, as (66) shows.(66) Context: answer to ?Who is the hunter??iPDETylm?xw@mchiefiPDETs@xw-p?x?-@m.OCC-hunt-MIDThe chief is the hunter.These data suggest that the pragmatic notion of ?ground? (i.e. the background inthe discourse context), used by Heycock and Kroch (1999) for explaining the fixedinformation structure of specificational sentences in English, and the requirementthat the initial DP represent relatively old information, is not relevant in OkanaganDP-DP structures, since inverse structures like (64c) and (65c) are always ungram-matical as answers to WH-questions. As such, a purely pragmatic account of theword order restriction in Okanagan DP-DP structures, similar to that of Heycockand Kroch (1999), will fail.There is no general requirement in Okanagan that focus in a direct predicationoccur initially (cf. 55). In (67b,c) below, the NP s@xwmaPm?yaPm ?teacher? is afocus, but can either precede or follow the subject. (68a) uttered in a context wherethe speaker walks into a room, sees bandages lying on a table but has no idea whathappened or how many people got hurt or who they are, allows (68b) as a response,259where Mary is a focus.28(67) a. sti ?mwhatMaryMaryaPDETc- ?kw??l-st-s?CUST-make-CAUS-3SG.ERGWhat does Mary do for work?b. MaryMarys@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDMary is a teacher.c. s@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDMaryMaryMary is a teacher.(68) a. switwhowhoiPDETDETx@nnumt?get.hurthurtWho got hurt?b. MaryMaryx@nnumt.get.hurtMary got hurt.28There is, however, a preference. In (i), a subject initial form is infelicitous, though not ungram-matical.(i) a. sti ?mwhatmatEPISaPDETc- ?kw??l-st-sCUST-make-CAUS-3SG.ERGiPDET??@x?- ??x??p.RED-grownWhat does the old man do for work?b. #iPDET??@x?- ??x??pRED-grown?qwQay-lqs.black-robeThe old man is a priest.c. ?qwQay-lqsblack-robeiPDET??@x? ??x??p.RED-grownThe old man is a priest.The preference for predicate-initial ordering for direct predications which do not involve subjectDP-focus becomes stronger for nominal predicate constructions (NPCs, i.e. ?bare clefts?), where theargument DP is a headless relative clause. These are discussed in chapter 8.260c. x@nnumtget.hurtMary.MaryMary got hurt.I claim that the word order restriction apparent in DP-DP structures derivesfrom a lexical property of the equative head, such that it F-marks its second ar-gument (cf. sections 2.3.1 and 7.5). Linear-alignment constraints then force theF-marked constituent to occur left-most (Koch, 2008a). Foci in direct predica-tions, by contrast, do not appear to follow the same linear-alignment constraints.In concrete terms, although we can state as a pragmatic principle for Okanagan ?AnF-marked focus must occur initially in Okanagan DP-DP structures?, this principleby itself does not explain why an iP DP can be F-marked only if the second DP isan iP DP, and not a proper name or demonstrative. I therefore appeal to a semanticasymmetry between iP DPs on the one hand, and proper names and demonstrativeson the other (cf. section 7.5).7.4.4 SummaryTo close this section, we have seen that Okanagan DP-DP structures carry an im-plicature of exhaustivity. This is given as follows (cf. 48):(69) Exhaustivity Implicature:A sentence of the form [x DP] = [iP Y DP]a. Asserts:? f .x = f (Y )b. Via the maximality implicature of iP (cf. 46) this implicates:x = MAX(Y )c. And assuming (48b) is satisfied, a DP-DP structure asserts:x is the only YThis exhaustivity implicature carried by the sentence is cancellable, just in case thesecond iP DP denotes a non-maximal individual.We have also seen that DP-DP structures exhibit a strict word ordering, such261that a more-referential DP may not follow a less-referential iP DP in a question-answer context, regardless of the information-structural status of the two DPs. Theword order restriction is plausibly explained by the following set of principles:(70) Information Structural Principles of Okanagan DP-DP Structuresa. F-marking is a lexical property of a null equative copula.b. The initial DP of a DP-DP structure must be F-marked.c. An F-marked constituent aligns to the left (Koch, 2008a).These principles, by themselves, do not explain why an initial iP DP can be F-marked only if the second DP is an iP DP. The analysis of the equative head whichI present in the next section solves these issues.7.5 Analysis of the Equative Head7.5.1 A Semantic AsymmetryA simple equative analysis of Okanagan DP-DP structures does not explain whythe second DP cannot be directly referential. In order to derive the word order re-striction seen with Okanagan DP-DP structures, I appeal to a semantic asymmetrybased on intensionality: the equative head distinguishes between directly referen-tial, extensional DPs (proper names (Kripke, 1982) and demonstratives (Kaplan,1977)) and non-directly referential, intensional DPs (iP DPs) (Romero, 2005; Co-morovski, 2007), and selects only an intensional DP as a complement. The surfacedistribution of Okanagan DP types in equative structures is shown below in Table7.1.2929As Table 7.1 implies, equatives consisting of two proper names are not possible in Okanagan.In section 7.6.1, I claim that sentences consisting of two proper names form a subclass of directpredications.262Table 7.1: Directly versus Non-directly Referential DPs in Equative DP-DPStructuresOkanagan DPs Demonstratives Proper Names iP DPs-directly referential X X *-initial DP in a DP-DPstructureX X X-final DP in a DP-DPstructure* * XIntensionality is argued to be a defining characteristic of the initial DP in aspecificational sentence in (Romero, 2005), based on the fact that definite DPs aresometimes interpretable as concealed questions. The specificational copula is giventhe following denotation in (Romero, 2005, 715, ex.67a).30 Underlining on the yargument indicates intensionality in (71).(71) Romero (2005)[[be]] = ?x?y?ws.y(w) = xRecall that Okanagan iP DPs allow individual concept readings, as in (72),repeated from chapter 5, section 5.2.9. For (72), the referent of iP ylm?xw@m ?thechief? is any non-specific future individual which happens to be the chief at thatfuture time.31(72) cmayEPISiPDETylm?xw@mchieflaPk?nwheniPDETtk?milxw.womanMaybe someday the chief will be a woman.30Under such an analysis, the copula be is ambiguous. It either maps an intensional individual to anextensional one, or vice versa, depending on whether the sentence is predicational or specificational(cf also Partee (1986)).31Line Mikkelsen (p.c.) rightly points out that if (72) is a DP-DP structure, I do not straight-forwardly predict the initial DP iP ylm?xw@m ?the chief? to be interpreted intensionally, which itobviously is in this context. The presence of the WH-item laPk?n ?when? is unclear, however, and soit is not clear that (72) is in fact a DP-DP structure. More examples like this should be checked.263As such, I claim that the Okanagan equative head employs an asymmetry similarto (71), however with the order of the arguments reversed, so that the equationalhead?s complement is intensional (i.e. an iP DP), while the specifier argument isextensional (i.e. referential, either a directly referential expression or an iP DP).(73) Okanagan[[=]] = ?x?ye?ws[x(w) = y]This analysis assumes that directly referential expressions (i.e. demonstratives,proper names) may not be of an intensional type .32While this analysis captures the semantic asymmetry between the two DPs,by itself it incorrectly allows for both grammatical structures where the more-referential DP is in initial position (Figure 7.7) and ungrammatical structures wherethe less-referential DP is in initial position (Figure 7.8):32It may also be argued that all DPs may be of an intensional type , but that directly referentialDPs are constant functions, whereas non-directly referential DPs are not (Kripke, 1982). Under thisassumption, a filter could be written into the semantics of (73) such that constant functions fromworlds to individuals are not permitted for the first argument (i.e. there has to be at least one worldwhere the DP denotes an individual y, rather than an individual x). Since this is a more general issuefor copular analyses which rely on an intensional asymmetry (e.g. Romero (2005)), I assume for thesake of simplicity that directly referential DPs simply cannot be of type .264Figure 7.7: Semantic Asymmetry, more-referential DP is in initial positionIdP?w[[ f (?x[basket(x)(w)?C(x)(w)])] = this]DPSub jthisax?PId??ye?w[[ f (?x[basket(x)(w)?C(x)(w)])] = y]Id?x?ye?ws[x(w) = y]=DPPred?w[ f (?x[basket(x)(w)?C(x)(w)])]D?P?w[ f (?x[P(x)(w)?C(x)(w)])]iPNP?x?w[basket(x)(w)]y?mx?waP?This is a/the basket?265Figure 7.8: Semantic Asymmetry, less-referential DP is in initial positionIdP?w[[ f (?x[basket(x)(w)?C(x)(w)])] = this]Id??ye?w[[ f (?x[basket(x)(w)?C(x)(w)])] = y]Id?x?ye?ws[x(w) = y]=DPPred?w[ f (?x[basket(x)(w)?C(x)(w)])]D?P?w[ f (?x[P(x)(w)?C(x)(w)])]iPNP?x?w[basket(x)(w)]y?mx?waPDPSub jthisax?P?This is a/the basket?This is because while (73) reflects the semantic asymmetry in DP-DP struc-tures, it does not reflect the fixed information structure. I claim that focus align-ment constraints (cf. Koch (2008a) for Thompson) force the more-referential DP,as an exhaustive focus, to occur left-most.7.5.2 Focus AlignmentThe problem with assuming nothing beyond a semantic asymmetry for the Okana-gan equative head, is that there is nothing to rule out the structure given as Fig-ure 7.8. Focus alignment constraints have proved useful in analyzing closely-related Thompson Salish (Koch, 2008a), and their applicability in Okanagan ap-266pears promising as well.Koch (2008a) investigates the focus and intonation properties of ThompsonRiver Salish nominal predicate constructions and clefts. He observes for clefts andnominal predicate constructions (NPCs) that a focused constituent must occur tothe left of non-focused material. Hence, the NPC (74a) is grammatical but (74b) isnot.(74) a. T?Pe.NEG[qw?PFOC]water??uPjusteDETs-P?qweP-kt.NOM-drink-1PL.POSSNo, we?ll just drink [water FOC]. (Thompson, Koch (2008a, 251, ex.15))b. *T?Pe.NEGP?qweP-ktdrink-1PL.POSS??uPjusteDET[qw?PFOC].waterNo, we?ll just drink [water FOC]. (Thompson, Koch (2008a, 251, ex.17c))Koch (2008a, 251, ex.13) captures this generalization as follows:33(75) FOCUS LEFT: Align the left edge of the focus-marked p(rosodic)-phrasewith the left edge of an intonational phrase.Koch has strong acoustic evidence for analyzing the entire sentence in (74a), ex-cluding the initial negation, as an intonational phrase (Pierrehumbert, 1980). Hefound that nuclear pitch accent occurs on the right edge of an intonational phrase,as indicated the first line of (76) below. An intonational phrase may consist of oneor more prosodic phrases, each marked by a pitch accent, as in the second line of(76). What Koch shows for Thompson is that focus does not coincide with nu-clear pitch accent. In other words, ?narrowly focused constituents do not attractadditional prosodic prominence? (Koch, 2008a, 169, ex.20).(76) (X)(X)T?Pe.NEG(( X[qw?PFOC]water)??uPjust(eDETX )X )s-P?qweP-kt.NOM-drink-1PL.POSSintonation-phraseprosodic-phraseNo, we?ll just drink [water FOC]. (Thompson, Koch (2008a, 251, ex.15))33Koch (2008a) develops a fine-grained Optimality Theoretic analysis involving focus alignmentconstraints, which I abstract away from here.267There have to date been no systematic studies investigating acoustic correlatesof focus in Okanagan34; however, it is plausible that the FOCUS LEFT constraint(75) is also operative in Okanagan DP-DP structures, as in (77):(77) a. haYNQBillBillnaP?CONJJohnJohniPDETs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDiPDETlLOCs@n- ?q@ ?y-m?n-t@n?LOC-write-MIN-INSTRAre Bill and John the teachers at this school?b. lut,NEGBillBilliPDETs@xw- ?maP- ?m?yaP-m,OCC-RED-teach-MIDlutNEGJohn.JohnNo, Bill is the teacher, not John.c. *lut,NEGiPDETs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDBill,BilllutNEGJohn.JohnNo, the teacher is Bill, not John.I suggest that as a lexical property, the equative head assigns a syntactic fea-ture ?F? (i.e. F-marking) to its second argument, and this is interpretable to theinformation structural component of the grammar as ?focus?.357.5.3 Final AnalysisAssuming that FOCUS LEFT (75) requires an F-marked constituent to align to theleft, the structure in (Figure 7.8), where the less-referential DP precedes the more-referential DP in exhaustive focus, is correctly ruled out. The final analysis of theequative head is given as (78-79), and a sample derivation is given in Figure (7.9)(78) Okanagan, final analysis[[=]] = ?x?ye?ws[x(w) = y]34It is worthwhile to mention Barthmaier (2004), who establishes the existence of intonationalunits in Colville-Okanagan.35There is no evidence for focus-related movement (Ogihara, 1987) in Okanagan DP-DP struc-tures.268(79) The equative head assigns a syntactic feature ?F? to its second argument.Figure 7.9: Final Equative Analysis of Okanagan DP-DP StructuresIdP?w[[ f (?x[basket(x)(w)?C(x)(w)])] = this]DPSub jthis[ax?PF ]Id??ye?w[[ f (?x[basket(x)(w)?C(x)(w)])] = y]Id?x?ye?ws[x(w) = y]=DPPred?w[ f (?x[basket(x)(w)?C(x)(w)])]D?P?w[ f (?x[P(x)(w)?C(x)(w)])]iPNP?x?w[basket(x)(w)]y?mx?waP?This is a/the basket?The copula in Figure 7.9 (cf. 78) takes an intensional individual x (the secondDP) and an extensional individual y (the first DP) as arguments, and yields a propo-sition that is true in a world w if and only if x applied to w is identical to y. Giventhat iP DPs occur as type expressions when they saturate the initial argumentposition of the equative head, yet are type e expressions when they saturate thesecond argument position. I assume that iP DPs have the property of being able tofreely type lower to extensional expressions when necessary.3636See section 9.4.2 for a summary discussion of the semantic types of Okanagan DPs, and theirtype-shifting possibilities.269The exhaustivity implicature, as applied to the example in Figure (7.9), is de-rived as in (81), assuming the maximality implicature of the determiner as given in(80):(80) Maximality implicature of iP:f = MAX(81) Exhaustivity Implicature:The sentence [ax?PFocus DP] = [iP y?mx?waP DP]a. Asserts:? f .[this = ?w. f (?x[basket(x)(w)?C(x)(w)])]b. Via (80), this implicates:this = ?wMAX(?x[basket(x)(w)?C(x)(w)])]c. And assuming (81b) is satisfied, asserts:?this? is the only x equal to ?wMAX(?x[basket(x)(w)?C(x)(w)])]In contexts where the second iP DP is interpreted as non-maximal, for example in acontext where there is more than one basket in the context, only the assertion holds(81a).This analysis of the equative head predicts that DP-DP equatives consisting ofa demonstrative and a proper name (e.g. This is John) or two proper names (e.g.Cicero is Tully) should not be possible, since both expressions are presumablydirectly referential in these cases. These types of sentences are, in fact, possiblein Okanagan; however there is reason to believe that they are not equatives. Inthe next section, I discuss these types of data, and argue that these comprise anidentificational sentence sub-class (Higgins, 1973), which may be reducible to atype of direct predication.7.6 Other Predication Types Involving Demonstrativesand Proper NamesThis section investigates two other types of predication, the first involving a demon-strative or demonstrative-associated DP and a proper name, the second involving270two proper names. I claim that both may be reduced to sub-types of direct predi-cation (excepting cases which may be analyzed as truncated clefts), since a propername in these cases must be a predicate. These data have interesting implicationswith regards to extending Higgins? taxonomy to Okanagan. I provide a taxonomyof Okanagan non-verbal predications in Table 7.3 below.7.6.1 The Case for Identificational SentencesThere are data in Okanagan which appear to correspond to Higgins? identifica-tional class, as opposed to the direct predications or DP-DP equatives examined sofar. These data normally involve a simple demonstrative and a proper name, or ademonstrative-associated iP DP and a proper name. A proper name is a predicatein these cases, an ascription of sorts, and I suggest that identificational sentencesin Okanagan are a subtype of direct predication. Like other direct predications,these sentence types allow variable word order in WH-contexts, as shown for casesinvolving a demonstrative and a proper name (82-83).(82) a. ix?PDEMSpike.SpikeThat is Spike. (answer to ?Who is Spike??)b. ix?PDEMSarah.SarahThat is Sarah. (answer to ?Who is that woman??)(83) a. LottieLottieix?P.DEMThat?s Lottie. (answer to ?Who is SaQ?lqs??)b. SpikeSpikeix?P.DEMIt?s Spike. (answer to ?Who is knocking on the door??)Independent evidence that proper names can be predicates in certain contextscomes from subject procliticization data. Consider that like any lexical predicate(84a), a subject proclitic may attach to a proper name (84b). This is not the case271for iP DPs (85a) or demonstratives (85b).37(84) a. kn1SG.ABSs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDI?m a carpenter.b. kn1SG.ABSJohn.JohnI?m John.(85) a. *kn1SG.ABSiPDETs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDI?m a/the carpenter.b. *kn1SG.ABSix?PDEMThat is me.In equatives, a proper name cannot be a predicate, as evidenced by the fact that itcannot occur in final position like other predicates. It seems reasonable to proposethat sentences like (82-83) have a different structure than equatives, one which isconsistent with a predicative analysis of proper names. On this basis, I suggest thatOkanagan too has a class of identificational sentences, though these may be ana-lyzed as a subclass of direct predication in the sense that no functional intermediaryis necessarily involved.Data such as (82-83) raise at least two questions:a. What prevents proper names from functioning as predicates in equationalstructures?b. What prevents non-demonstrative-associated iP DPs from functioning asdirect arguments to proper names in an identificational sentence?Regarding question (a), the selectional restrictions of the copula require that itssecond DP argument must be of type e, which means that in equative structures, aproper name must undergo either N-to-D raising or be the complement of a null D37See N. Mattina (2006, 111) for similar observations in Moses-Columbian.272(cf. discussion in section 4.6.2). Since proper names do not denote non-rigid indi-vidual concepts of type , they cannot function as complements to the equativehead.Question (b) presents a more challenging issue, especially in light of the factthat identificationals similar to (82-83), but involving a demonstrative-associatediP DP (86) are also possible. The adjoined demonstrative minimally distinguishesan Okanagan identificational sentence (86), which allows variable word ordering,from an equative, which does not allow free word order (87).38(86) a. ix?PDEMiPDETs@xw- ?kw?l-m?n-@mOCC-work-MIN-MIDJohn.JohnThat hard worker is John.b. JohnJohnix?PDEMiPDETs@xw- ?kw?l-m?n-@m.OCC-work-MIN-MIDJohn is the one that is a hard worker.(87) a. *iPDETs@xw- ?kw?l-m?n-@mOCC-work-MIN-MIDJohn.JohnThe hard worker is John.b. JohnJohniPDETs@xw- ?kw?l-m?n-@m.OCC-work-MIN-MIDJohn is the hard worker.Based on the data seen so far, the generalization seems to be that an identificationalsentence requires the presence of a demonstrative, and under the assumption thatidentificational sentences are a sub-type of direct predication, the demonstrative in(82-83, 86) somehow licenses ascriptive readings of proper names. (This is notquite correct, however, since identificationals involving two proper names are alsopossible, as I show in section 7.6.2.)It is crucial to note that for cases where the proper name occurs in final position(cf. 86a), the proper name cannot be in focus: (88c, cf. 86a) below cannot be usedas answer to (88b). (88d) can be used as an answer to (88b) since it is ambiguously38(86b) is ambiguously an equative, though (86a) is not. I return to this issue below.273equative, and the focus occurs initially.(88) a. switwhoixiPDEM?@COMPJohn?JohnWho is John? (Question 1)b. switwhoix?PDEMtOBLs@xw- ?kw?l-m?n-@m?OCC-work-MIN-MIDWho is the hard worker? (Question 2)c. ix?PDEMiPDETs@xw- ?kw?l-m?n-@mOCC-work-MIN-MIDJohn.JohnThat hard worker is John. (answers 1, not 2)d. JohnJohnix?PDEMiPDETs@xw- ?kw?l-m?n-@m.OCC-work-MIN-MIDJohn is the one that is a hard worker. (answers 1 or 2)The data in (88) merit further discussion. In answer to (88a), (88c) does not havethe information structure of a specificational sentence, rather it appears to be at-tributing the name John to a contextually salient entity (i.e., it is an identificationalsentence). I would argue that an exhaustivity implicature is not involved in thecase of (88c), though it is in (88d), at least in answer to (88b). This means that(88a) does not require an exhaustive answer, and so (88c,d) are acceptable as directpredications, and (88d) as a DP-DP structure where the exhaustivity implicaturehas been cancelled. (88b), by contrast, does require an exhaustive answer, and soassuming that predicates cannot be exhaustive, the proper name must be interpretedas individual-denoting, and (88c) is only interpretable as an ungrammatical inverseequative, leaving (88d) as the only possible response.Identificational sentences involving proper names and simple demonstratives(cf. 82-83), where the proper name is in focus, may be analyzed as truncated clefts(Hedberg, 2000; Mikkelsen, 2005). Such an analysis is motivated by data like (89),where a contextually salient residue clause aP cnpuw?p@m ?who was knocking onthe door? is omissable.3939I investigate these facts in more detail in chapter 8. A demonstrative can also precede or follow274(89) Truncated Cleft analysisa. ix?PDEMSpikeSpike(aPDETc-n-pw-?p-@m).CUST-n-knock-INCH-MIDIt?s Spike (who is knocking on the door).b. SpikeSpikeix?PDEM(aPDETc-n-pw-?p-@m).CUST-n-knock-INCH-MIDSpike is the one (who is knocking on the door).I argue in chapter 8 that Okanagan clefts are equative structures, and so if the trun-cated cleft analysis of simple demonstrative identificationals is correct, then iden-tificational sentences cannot be reduced to direct predications. Not all identifica-tional sentences may be reduced to the truncated cleft analysis, however: cases in-volving demonstrative-associated iP-DPs such as (86a, 88c), where a proper nameoccurs finally, are not analyzable as clefts, truncated or otherwise, since propernames cannot function as cleft residues, just as they cannot function as the final DPin a DP-DP structure.In sum, identificationals involving simple demonstratives are surface ambigu-ous between predicational sentences and truncated clefts, the difference being whet-her or not the proper name is a predicate. Those involving demonstrative-associatedDPs with proper names in final position may only be analyzed as direct predica-tions, while those with proper names in initial position may be equative or pred-icational. For Danish and English, Mikkelsen (2005) analyzes identificationalsentences with simple demonstrative subjects as specificational, and those withdemonstrative DP subjects as equatives. The differences between this English clas-sification and what I propose for Okanagan are represented below in Table 7.2.an iP DP in a cleft context, but if the demonstrative follows the iP DP, then the residue clause must beovert (since otherwise we have something surface-identical to an ungrammatical inverse equative).275Table 7.2: Reducing Higgins? Taxonomy: Identificational Sentences inOkanagan Compared to Mikkelsen?s (2005) Classification of EnglishEnglish OkanaganDemonstrative subjects specificational predicational/equative (truncated cleft)Demonstrative DP subjects equative predicational/equative (truncated cleft)It is also worthwhile to note that Mikkelsen (2005) distinguishes between sen-tences like She is Susan which are analyzed as equatives, and sentences like Thatis Susan which are specificationals. For Okanagan, there is no distinction betweenthese two types, since the demonstrative ix?P may denote both human and non-human referents.7.6.2 Predications Involving Two Proper NamesOkanagan allows a predicative relation to exist between two proper names. Thesetypes of data are directly analogous to classical English examples such as Cicerois Tully, which are usually understood to be equative. In Okanagan, however, thereis good evidence that these cases are a sub-type of predicational clause, since likethe identificational sentences just discussed, a proper name must be predicative.In (90), for example, Spike is playing the part of the famous chief Chillhitzia ina play. In (91a), the addressee does not know who SaQ?lqs is, and so the speakeridentifies Lottie as being the bearer of the Indian name SaQ?lqs (91b).(90) SpikeSpike?C@lx? ?caPChillhitziatOBL?kl- ?kl?xw.RED-eveningSpike was Chillhitzia last night.(91) a. switwhoix?PDEM?@COMPSaQ?lqs?SaQ?lqsWho is SaQ?lqs?276b. kn1SG.ABSnstilsthinkLottieLottieSaQ?lqs.SaQ?lqsI think that Lottie is SaQ?lqs.Evidence that the second proper names in (90,91b) are predicative comes from datashowing that the predicative proper name may precede the referential one withoutany change in meaning (92, cf. 90), and that a subject proper name (e.g. ?Sarah?)may raise over negation (93b), but the predicative proper name may not (93c) (cf.section 7.2.2).40(92) ?C@lx? ?caPChillhitziaSpikeSpiketOBLs- ?kl- ?kl?xw.NOM-RED-eveningSpike was Chillhitzia last night.(93) a. lutNEGSarahSarah?tEMPHSaQ?lqs.SaQ?lqsSarah is not SaQ?lqs.b. SarahSarahlutNEG?tEMPHSaQ?lqs.SaQ?lqsSarah is not SaQ?lqs.c. *SaQ?lqsSaQ?lqslutNEG?tEMPHSarah.SarahSarah is not SaQ?lqs.Further evidence for analyzing one of the proper names as a predicate comesfrom the presence of the complementizer ?@ in these contexts: it optionally in-tervenes between subject and predicate in these contexts. Thus (94a) and (94b)are pragmatically equivalent in a context where someone is being ascribed with aname.4140The emphative marker ?t(i) regularly co-occurs with negation. I assume that it pro-cliticizes to apredicate.41It should be mentioned that there is no inherent difference between Christian and Indian names.In answer to ?Who is Lottie?? for example, (i) below is perfectly felicitous (cf. (94b)). In this case,?Lottie? is the predicate.277(94) a. LottieLottieSaQ?lqs.SaQ?lqs.Lottie SaQ?lqs.b. LottieLottie?@COMPSaQ?lqs.SaQ?lqs.Lottie is SaQ?lqs.The complementizer ?@ may also optionally be used in contexts for which someoneis teaching another the name of something. Either a direct predication (95a) or apredication with ?@ (95b) may be used in a context where a mother is teaching hertwo year old the names of objects.42(95) a. ax?PDEMlpot,cupax?PDEMnc@cq?pt@n,bowlax?PDEMlasy@t.plateThis is (called) a cup, this is (called) a bowl, this is (called) a plate.b. ax?PDEM?@COMPlpot,cupax?PDEM?@COMPnc@cq?pt@n,bowlax?PDEM?@COMPlasy@t.plateThis is (called) a cup, this is (called) a bowl, this is (called) a plate.The complementizer ?@ is only felicitous in naming contexts, hence (96b) soundsvery strange to a speaker.(96) a. JohnJohns@xw-k- ?cx?w?p@laP-m.OCC-k-judge-MIDJohn is a judge.b. ?JohnJohn?@COMPs@xw-k- ?cx?w?p@laP-m.OCC-k-judge-MID(?)John is called a judge.(i) SaPalqsSaPalqs?@COMPLottie.LottieSaPalqs is Lottie.42See discussion of Lillooet in section 9.3.2. The Lillooet auxiliary waP is required in identifica-tional contexts for some speakers (Davis, 2010c).278Finally, the predicative proper name may optionally be introduced by an iPdeterminer in these contexts (97). These are equative structures. From the syntacticperspective, the availability of an optional iP determiner is as expected given thatproper names may be analyzed as lexical NPs.43 From the pragmatic perspective,iP is predicted to be possible given that equatives are pragmatically equivalent todirect predications in cases where the exhaustivity implicature is cancelled (cf.section 7.4.1). Like with other equatives, the proper name introduced by iP cannotprecede the bare proper name (98).(97) a. LottieLottieiPDETSaPalqs.SaPalqsLottie is SaPalqs.b. SpikeSpikeiPDET?C@lx? ?caPChillhitziatOBL?kl- ?klaxw.RED-eveningSpike was Chillhitzia last night.(98) *iPDETSaQ?lqsSaPalqsLottie.LottieLottie is SaPalqs.This brings us to the question of what exactly it means for a proper name to bea predicate. In (97a), for example, it seems odd to argue that the exhaustivity im-plicature is cancelled, and that this is the reason that it is pragmatically equivalentto (94a), since the proper name predicate in (94a) presumably denotes a singletonset, and the proper name iP DP in (97a) a maximal individual.In light of the fact that DP-DP equatives permit ?pseudo-predicational? readingsin case the exhaustivity implicature is not satisfied, it seems more useful to say thatdirect predications involving either a demonstrative (or demonstrative-associatedDP) and a proper name (7.6.1), or else two proper names, permit ?pseudo-equative?readings since proper names denote singleton sets in these instances. We thereforepredict that true equatives (97a) and pseudo-equative direct predications (94a) willbe pragmatically equivalent.43Or if not lexical NPs, then arguably coercible into NPs.2797.6.3 SummaryThis section first presented Okanagan data involving a demonstrative (or demon-strative-associated DP) and a proper name, and has shown that these pattern morelike direct predications than equatives, both in terms of word order and in termsof information structure. I have claimed that these cases correspond to Higgins?identificational sentence class, and that for Okanagan they form a sub-type of di-rect predication where the proper name is a predicate. Identificational sentencesinvolving a simple demonstrative may in some cases be analyzed ambiguously astruncated clefts, a type of equative (cf. chapter 8). Next, I discussed sentencesinvolving two proper names and showed that one of the proper names must beanalyzed as a predicate, similar to the case for identificational sentences. Directpredications involving proper name predicates permit ?pseudo-equative? interpre-tations, since the proper name predicate denotes a singleton set.I now move on to a discussion of problematic cases of ?inversion?, where aless-referential iP DP precedes what appears to be a more-referential, focused DP.7.7 Problem InversionsThere are data which appear to show that Okanagan does in fact have specifica-tional sentences. Within a single discourse turn, a speaker can utter (99a), andthen follow up with (99b) or (99c), which appear to be inverse equatives. (99d)was actually judged ungrammatical in this context, which is surprising given thatit displays the expected ordering of an equative.(99) a. BillBilliPDETs@xw-mr?m-@m.OCC-medicine-MIDBill is a doctor.b. tkPkaPs?l@mtwo(HUMAN)iPDETs@xw-sxw-mr?m-@mOCC-RED-medicine-MIDJohnJohnu?CONJMary.MaryTwo (other) doctors are John and Mary.280c. iPDETknaqsone(HUMAN)iPDETs@xw-mr?m-@mOCC-medicine-MIDJohn.JohnAnother doctor is John.Consultant: If you?re telling a story about a doctor, introducing him.d. *JohnJohnu?CONJMaryMarytkPkaPs?l@mtwo(HUMAN)iPDETs@xw-sxw-mr?m-@m.OCC-RED-medicine-MIDJohn and Mary are two other doctors.Other, similar examples are given in (100):(100) a.txwaP-xwPitRED-many(HUMAN)iPDETs-maP-mP-?m,NOM-RED-womanu?CONJi ?kl?PDEMiPDETknaqsone(HUMAN)iPDET?kw@ ?kwy?maPsmallSue.SueThere are a lot of women, and the smallest woman there is Sue.b.Context: You are having an argument with your friend about who is theprime minister.iPDETprimeprimeministerministertlLOCCanadaCanadalutNEG?t@EMPHTrudeau,Trudeau,n?x?@m?CONJHarperHarperiPDETprimeprimeminister.ministerThe prime minister of Canada isn?t Trudeau, it?s Harper who is the primeminister.Consider also that at the beginning of a narrative, both canonical (101a) andinverse (101b) predications were judged grammatical:(101) a. ?qs?pilong.agoSpikeSpikeiPDETs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDLong ago, Spike was a carpenter.(Literally: Long ago, there was a carpenter named Spike.)281b.?qs?pilong.agoiPDETs@xw- ?kw??l-?xw-@mOCC-make-house-MIDSpike.SpikeLong ago, Spike was a carpenter.(Literally: Long ago, there was a carpenter named Spike.)There is one thing that seems to be clear about these problematic cases of in-version: They are not felicitous as answers to a WH-question, as (102) shows.(102) a.switwhoiPDETs@xw- ?kw??l-?xw-@mOCC-make-house-MID?qs?pi?long.agoWho was the carpenter long ago?b.?qs?pilong.agoSpikeSpikeiPDETs@xw- ?kw??l-?xw-@m.OCC-make-house-MIDLong ago, Spike was a carpenter.c.* ?qs?pilong.agoiPDETs@xw- ?kw??l-?xw-@mOCC-make-house-MIDSpike.SpikeLong ago, Spike was a carpenter.Compare also the inversion in (103c) which is similar to (100a), but is not possiblein answer to the question (103a).(103) a.switwhoiPDETmys-?t?qw@lqwmost-talltlLOCs-t@mxw?laPxw-mp?NOM-land-2PL.POSSWho is the tallest person in your community?b.SpikeSpikeiPDETmys-?t?qw@lqwmost-talltlLOCs-t@mxw?laPxw-t@t.NOM-land-1PL.POSSSpike is the tallest person in our community.c.*iPDETmys-?t?qw@lqwmost-talltlLOCs-t@mxw?laPxw-t@tNOM-land-1PL.POSSSpike.SpikeThe tallest person in our community is Spike.Also consider that (104, cf. 100b) appears to show inversion around negation, butrecall that predicate raising is never a possibility, including in contexts involving282negation (105c). This goes for DP-DP structures as well (106), in answer to aWH-question.(104) Context: You are having an argument with your friend about who is theprime minister.iPDETprimeprimeministerministertlLOCCanadaCanadalutNEG?t@EMPHTrudeau,Trudeau,n?x?@m?CONJHarperHarperiPDETprimeprimeminister.ministerThe prime minister of Canada isn?t Trudeau, it?s Harper who is the primeminister.(105) a.SarahSarahlutNEG?tEMPHsyxw?p-m@x.Shuswap-personSarah is not Shuswap, she?s Okanagan.b.lutNEGSarahSarah?tEMPHsyxw?p-m@x.Shuswap-personSarah is not Shuswap.c.*syxw?p-m@xShuswap-personlutNEG?tEMPHSarah.SarahSarah is not Shuswap.(106) a.haYNQJohnJohniPDETs@xw- ?maP- ?may?P-m?OCC-RED-teach-MIDIs John the teacher?b.lutNEGJohnJohn?t@EMPHiPDETs@xw- ?maP- ?may?P-m.OCC-RED-teach-MIDJohn is not the teacher.c.*iPDETs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDlutNEG?t@EMPHJohn.JohnThe teacher is not John.In sum, these problematic inversions are not felicious in answer to a WH-283question, which is unexpected if they were in fact true specificationals. Speakersindicate that these inversions are good in introductory contexts, when a referent isbeing introduced (cf. 99c), though the contexts in which the data in (100) werejudged indicate that a non-contrastive iP DP can also precede a proper name.There are three possible analyses of these inversion cases:a. As identificational sentences, where the final proper name is a predicate.b. As standard equatives with topicalization movement of a less-referentialiP DP to a position preceding the more-referential DP (Heggie, 1988;Mikkelsen, 2005; Partee, 2010).c. Inversion is in fact possible in DP-DP structures, but if they contain a fo-cused constituent it has to be in initial position.The first possibility is that these are identificational sentences, similar to thosediscussed in section 7.6.1 which also display variable word order, but minus ademonstrative (or possibly introduced by a null demonstrative). That means thatthe proper name is a predicate in these cases, being ascribed to a contextuallysalient entity, and is not interpreted exhaustively. This seems especially promisingfor the out-of-the-blue cases (cf. 101), where it seems likely that the proper nameis an ascription of sorts, however for inversions within a single discourse turn (cf.100), this analysis seems less clear, since these do appear to be specificational inthe sense that the proper name specifies who or what the initial iP DP is.The second possible analysis is that these involve topicalization, or left disloca-tion of an iP DP (cf. Mikkelsen (2005) for a discussion of predicate topicalization).That is, these are underlyingly equative, with focus on the proper name, and theiP DP is dislocated to adjoin to some high functional position. The negation datain this section actually indirectly support the topicalization hypothesis, since weotherwise expect identificational sentences to be good both as answers to WH-questions as well as in other contexts, but this is not the case (compare 100b and106c). This view additionally fits with the intuition that for ongoing topic data like(100), the proper name is a focus. We can then state that a F-marked DP in finalposition is underlyingly the initial, more-referential DP of an equative clause, andthat the surface-initial iP DP is left-dislocated, outside of the intonational phrase.284Acoustical evidence is needed to substantiate this hypothesis, however.The third possible analysis essentially reduces to an argument that it is not alexical property of the equative head to assign a feature ?F? to its second argument,interpretable by the information structural component as ?focus?. In section 7.5, Iclaimed that DP-DP equatives inherently involve focus. If one were to argue thatDP-DP structures may occur without focus, we lose the generalization that wordorder is rigid in WH-contexts, and so I do not further discuss this possibility.It is interesting that Koch (2012) notes that speakers of Thompson River Salishdo not obligatorily mark contrastive focus within their own discourse turn. Exam-ple (107) is a direct predication: the main predicate is Qw? ?yt ?sleep?, and the propername Sam in the second clause is constrasted with Bill in the first clause. For theequivalent sentence in English, Sam should be obligatorily marked as a focus. ForThompson, this means that Sam should occur at the left periphery of the secondclause, but it instead occurs in final position.(107) P?x-iP ??uP-xePIMPF-still-DEMQw? ?ytsleepeDETBill,BillPe?andP?x-iP ??uPIMPF-stillQw? ?ytsleepeDETS?m.SamBill is still sleeping, and Sam is still sleeping.(Thompson, Koch (2012, ex.39))Unlike Thompson, it is notable that the focused DP in an Okanagan direct predica-tion may occur in final position in answer to a WH-question. Since the problematiccases of Okanagan inversions discussed in this section pattern like Thompson (107)in terms of not being felicitous in answer to a WH-question, this may be evidencefor analyzing the Okanagan cases as identificational direct predications, assumingthat Thompson introduced clefts, which are the closest analogue to DP-DP struc-tures in Okanagan, cannot occur with a focused DP in final position. On the otherhand, the proper name in final position in (107) is clearly a DP, whereas under theidentificational predication analysis of the Okanagan inversion data in this section,the proper name should be an NP predicate. This means that it is also a possibilitythat while Thompson allows focus to occur finally for direct predications within adiscourse turn, Okanagan allows focus-final DP-DP structures within a discourseturn, assuming that the topicalization hypothesis is correct.In any case, it seems clear that for Okanagan, as with Thompson, there is a285difference between discourse strategies which are used across discourse turns (e.g.WH-question/answer pairs) versus within a discourse turn (cf. 100) or at the be-ginning of a narrative (cf. 101). Under a cross-linguistic assumption that specifica-tional sentences should be felicitous as answers to at least some WH-questions, wecan maintain the argument that Okanagan does not have specificational sentences,though it remains unclear exactly why an appropriate WH-question cannot yieldan information structural configuration which licenses topicalization, and a surfaceordering for DP-DP structures whereby the more-referential DP follows the less-referential DP. It will be a worthwhile endeavor to investigate other types of datainvolving multiple discourse turns, aside from question/answer contexts, in orderto see whether the distribution of these inverse structures is truly dependent on thestatus of an utterance with respect to interlocution.7.8 Summary and Implications7.8.1 Summary of Major PointsThis chapter has argued for the following points:a. Okanagan DP-DP predications are syntactically and information-structurallydistinct from direct predications, broadly supporting Heycock and Kroch(1999) who posit distinct predicational and equative small clauses for En-glish.b. Okanagan direct predications do not involve any copula.c. Okanagan DP-DP structures are equative (Heycock and Kroch, 1999), andare projections of a null equative copula.d. Okanagan DP-DP structures carry an implicature of exhaustivity, which istraceable to the maximality implicature of the iP determiner in the secondDP. The implicature is cancellable, leading to the availability of ?pseudo-predicational? readings for which a DP-DP structure is pragmatically equiv-alent to its corresponding direct predication.e. The initial DP of a DP-DP structure is interpreted as a focus.286f. Specificational DP-DP structures are not possible in Okanagan. This re-ceives an explanation whereby:(i) the equative head selects only intensional () iP DPs as a comple-ment (Romero, 2005), not directly referential DPs.(ii) the equative head assings a feature ?F? to its second argument, whichis interpretable as a focus (Rooth, 1992).(iii) alignment constraints force an F-marked constituent to occur left-most(Koch, 2008a).(iv) there is a ban on predicate raising (a.k.a. ?syntactic inversion?) (Moro,1997; den Dikken, 2006).g. Identificational sentences, and sentences involving two proper names in apredicative relation, may be reduced either to a sub-type of direct predi-cation (i.e. a proper name is a predicate), or in some cases to a truncatedclefts, which are a type of equative.h. Apparent cases of syntactic inversion involving DP-DP structures pluasi-bly involve topicalization of an iP DP. These structures are possible withina running discourse, but not in answer to a WH-question.7.8.2 ImplicationsThere are several interesting implications to this analysis of the Okanagan data,a few of which I will touch on here. I refer the reader to a more comprehensivediscussion of implications in chapter 9.A Higgins? Taxonomic Classification of Okanagan Non-Verbal PredicationsFirst of all, Higgins? taxonomy appears to be reducible to two types for Okana-gan, predicational and equative, as depicted below in the following chart, wheresubjects and more-referential DPs are indicated in bold italic type, predicates andless-referential DPs are in normal italic type, and cleft residues are in brackets.287Table 7.3: Higgins? Taxonomy and Okanagan Non-Verbal PredicationsCLAUSE TYPE EXAMPLE SUBJECT COMPLEMENTPredicational Wilford s@xw ?kw? ?l?xw@m e s@xw ?kw? ?l?xw@m Wilford(Wilford is a carpenter)(Identificational) Spike ?C@lx? ?caP?C@lx? ?caP Spike(Spike is Chillhitzia)John ix?P iP s@xw ?kw ?lm?n@mix?P iP s@xw ?kw ?lm?n@m John(That hard worker is John)ix?P JohnJohn ix?P(That?s John)Equative Wilford iP s@xw ?kw? ?l?xw@m e *iP s@xw ?kw? ?l?xw@m Wilford(Wilford is a/the carpenter)(Identificational ix?P John [.... (residue)](truncated cleft)) John ix?P [.... (residue)](It?s John)(Identificational John ix?P [iP s@xw ?kw ?lm?n@m](full cleft)) *ix?P [iP s@xw ?kw ?lm?n@m] John(It?s John that is the hard worker.)This chart shows that there are two clear classes of non-verbal predication in288Okanagan corresponding to Higgins? predicational and equative classes. Identifi-cational sentences may be analyzed as either predicational or equative, dependingon (i) whether a proper name is predicative or referential (which in turn depends inpart on whether the context requires exhaustivity); and (ii) information structure,specifically where focus falls.Note that under the truncated cleft hypothesis, a non-subject demonstrativecan only be an intensional under the assumption that it is adjoined to a nullresidue headed by the determiner iP. This means that when the demonstrative oc-curs initially, it forms a discontinuous constituent with the residue clause. I presentevidence for such an analysis in section 8.5.2. While there are unresolved questionsof compositionality here which I have not been able to solve (cf. section 4.6.1), no-tice that if we were to assume that simple demonstratives could also be non-rigidindividual concepts of type , then we predict that a specificational word ordershould be possible (e.g. iP s@xw ?kw? ?l?xw@m ix?P), (since iP DPs are type e whenthey are the second arguments of the equative copula) but this is not the case.44Implications for Reducing the Identificational ClassMy analysis of Okanagan identificationals offers interesting points of comparisonwith previous analyses of English identificationals. Here, I briefly discuss how myanalysis contrasts with analyses of English identificationals by Mikkelsen (2005)and Heller (2005).Mikkelsen (2005) assimilates English identificational sentences with simpledemonstrative subjects to the specificational class, while those with demonstra-tive phrase subjects are ?demonstrative equatives?. Okanagan identificational sen-tences involving simple demonstratives may in some cases be analyzed as trun-cated clefts (a type of equative), but in other cases are best analyzed as predica-tional. Okanagan identificational sentences with demonstrative-associated iP DPsmay only be analyzed as equatives if a non-predicative proper name occurs initially,otherwise they must be analyzed as predicational. Okanagan thus offers some sup-port for analyzing identificationals involving ?demonstrative phrase? subjects as44It is possible that further work on binding condition C in Okanagan could offer independentsupport for a semantic distinction between bare demonstratives and demonstratives adjoined to DPs(cf. Davis (2006, 2009) for Lillooet).289non-predicational, although an analysis of these as specificationals is of coursedependent on a language having specificationals in the first place.Heller (2005, 197) states that identificational sentences are ?predicational sen-tences in which the post-copular phrase is an essential property that is assumedto be new information, so the entity denoted by the pre-copular phrase has tobe picked out by an expression that does not presuppose this property.? Okana-gan predicational identificational sentences largely support this characterization:a proper name like John in these cases may be viewed as an essential property,and crucially, the demonstrative or demonstrative-associated DP is referential. Forequative identificationals in Okanagan, however, the proper name is a referentialexpression, and the demonstrative, which adjoins to an overt or truncated iP DP,denotes a non-rigid individual concept.45In sum, the Okanagan data broadly support Mikkelsen (2005) and Heller (2005)in attempting to reduce Higgins? identificational class to either a predicational orequative class.There are other implications to this classification worth mentioning, which Idiscuss in a section on theoretical implications (9.4). The next chapter investigatesthe syntax, semantics, and information structure of clefts in Okanagan. These willbe shown to be information structurally, and morpho-syntactically, equivalent tosimpler DP-DP structures. On the basis of their commonalities, I claim that bothderive from a common underlyingly equative structure. In this respect, the Okana-gan data support theories which derive clefts from copular clauses.45Heller and Wolter (2008) analyze identificational sentences with simple demonstratives as atype of predicational sentence. They argue that the initial demonstrative is an individual conceptof type , and that the post-copular complement is a function from worlds to sets of individualconcepts, of type ,t>> (i.e. ?a sort? (Gupta, 1980)). There is little evidence at this point forpositing such higher types for Okanagan, though this and similar analyses must remain as distinctpossibilities.290Chapter 8Okanagan Clefts as EquativesThis chapter investigates a class of sentences known as ?clefts? in the wider litera-ture on Salish languages (Kroeber, 1999; Davis et al., 2004; Koch, 2008a, 2009).A typical Okanagan cleft is shown in (1), with its three possible translations (focusindicated in the English translation by bold type):1(1) ix?PDEMiPDETs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDiPDETkwu1SG.ABSqw@l-qw?l-st-s.RED-speak-(CAUS)-3SG.ERGa. That?s the teacher that talked to me.b. That teacher is the one that talked to me.c. It?s the teacher who talked to me.Textual examples of similar clefts can be found in A. Mattina (1985):2(2) a. wa ?yyesix?PDEM[iP]DETi-s-@n-P@m-P?maP-t1SG.POSS-NOM-n-RED-grandchildy@PDETn-k@?t-k?t-?s-@s...n-cut.off-RED-head-3SG.POSS[?]It?s my grandchildren whose heads are cut off. (A. Mattina 1985, stz.100)1Unlike in English, focus is not necessarily signalled by pitch accenting in Okanagan.2The determiner y@P in (2a) is a variant of iP, found in Colville.291b. ...ax?PDEMiPDET?q@ ?y-m?nwrite-INSTRiPDETs-c- ?q? ?y-t@t.NOM-CUST-write-1PL.POSSThat?s the letter we wrote. (A. Mattina 1985, stz.181)8.1 IntroductionThis section first introduces some terminological conveniences which I use whendiscussing Okanagan clefts, and then presents the main claims of this chapter, fol-lowed by an outline of this chapter.8.1.1 Terminological PreliminariesIn the Salish literature, the term ?cleft? traditionally encompasses both ?bare clefts?,which are a form of direct predication, and ?introduced clefts? which are so namedbecause they are normally introduced by a clefting predicate. I use the term ?cleft?as a descriptive term to refer to Okanagan sentences which involve structural focus,lack a main clause lexical predicate (NP, VP, or AP), and are typically though notalways introduced by a distal demonstrative ix?P while ending with a non-focusedclausal constituent.I begin with an assumption that the discourse-related and interpretive parallelsbetween clefts in Okanagan and clefts in other (Salish and non-Salish) languagesalso imply certain syntactic and semantic parallels, and my discussion will reflectthis assumption. This assumption allows us the use of descriptive terms such as?focus position? and ?residue clause?, and is useful in comparing Okanagan cleftswith those in other Salish languages.The interpretive ambiguity in Okanagan clefts (cf. 1) necessitates some im-portant terminological clarifications: By focus and in focus, I refer to a linguisticexpression with the information-structural property of indicating alternatives rele-vant for the interpretation of that expression (Rooth, 1985, 1992) (cf. section 2.3.1)and/or the left peripheral position in an Okanagan equative which may be inter-preted exhaustively assuming that the exhaustivity implicature is satisfied (chapter7). In contrast, I use focus position as a descriptive term referring to the con-stituent which canonically follows either a clefting predicate or demonstrative (e.g.292the demonstrative ix?P in (1)) and which precedes the residue clause (e.g. iP kwuqw@lqw?lsts ?that talked to me? in (1)) (Kroeber, 1999; Davis et al., 2004; Koch,2009). As implied by Okanagan (1), interpretation (a), the constituent in the focusposition and the constituent in focus are not necessarily the same.38.1.2 The Main ClaimMy main claim is as follows:(3) Okanagan clefts and DP-DP structures are both equatives.This claim is based on information structural and morpho-syntactic parallelsbetween clefts and DP-DP structures: clefts imply exhaustivity, lack any presup-position of existence, and require focus to be aligned left (despite the appearancesof 1c), exactly as with DP-DP structures (cf. section 7.4). Morpho-syntactically,the only difference between DP-DP structures and clefts is that in the former class,the final DP is non-clausal, while in the latter class, the final DP is clausal. Okana-gan clefts, unlike clefts in many other Salish languages (e.g. Thompson (Koch,2009), Lillooet and Northern Straits (Davis et al., 2004)), allow the residue clauseto contain an overt NP head which provides strong evidence that the residue is a DP.I present a derivational analysis whereby the Okanagan clefting demonstrative ix?Pforms an underlying constituent with the equative complement. The discontinuousconstituency of the clefting demonstrative and residue clause in Okanagan cleftsmay therefore be understood as broadly supporting ?extraposition-from-subject?-style theories (Percus, 1997; Hedberg, 2000) (cf. section 2.4.1).8.1.3 Chapter OutlineThis chapter proceeds as follows:First, I introduce and discuss Okanagan clefts in the context of previous stud-ies of clefts in Salish (section 8.2). I describe clefts under a classical analysis astripartite structures consisting of an introductory particle, a ?focus position?, and aresidue clause (8.2.1). Any type of DP can occupy the ?focus position? (8.2.3).3This issue is related to the fact that an initial non-focused clefting demonstrative must be dis-tinguished from an initial focused demonstrative, which is the 2nd argument of the equative copula.This distinction must be kept in mind throughout the reading of this chapter.293Second, I show that Okanagan introduced clefts (i.e. clefts where a DP focusis introduced by a demonstrative) share the same information structural propertiesas DP-DP structures (8.3): i.e. they imply without entailing exhaustivity (8.3.2),they lack any presupposition (8.3.3), and they require that the focused DP precedethe residue clause (8.3.4). This constitutes information structural evidence for acommon analysis of clefts with DP-DP structures. I then briefly contrast nominalpredicate constructions (NPCs, a.k.a. ?bare? clefts) (Kroeber, 1999; Davis et al.,2004) with introduced clefts (8.3.5). I analyze these as direct predications, andthen set them aside.Third, I discuss a three-way interpretive ambiguity for clefts with an iP DP in?focus position? (8.4.1, cf. 1):a. Demonstrative Focus: The initial demonstrative can be in focus, to theexclusion of a following ?focus position? iP DP and residue clause.b. Demonstrative DP Focus: The demonstrative and a constituent iP DP (in?focus position?) are in focus, to the exclusion of the final clause.c. DP Focus: The ?focus position? iP DP can be in focus, to the exclusion ofthe initial demonstrative and residue clause.I suggest that interpretations (a) and (b) straightforwardly support an equativeanalysis (8.4.2). (Interpretation (c) does as well, though since this case is morecomplex I save discussion of this until section 8.5.) For interpretation (a), I claimthat a headed relative clause DP is the first argument of the equative head and thedemonstrative is the focused, second argument; for interpretation (b), a headlessrelative clause DP is the first argument of the equative head, and a demonstrative-associated DP is the focused, second argument. Okanagan clefts are straightfor-wardly amenable to the equative analysis because:a. Okanagan canonically introduces clefts with a demonstrative, rather thanthe dedicated clefting predicate characteristic of other Salish languages,such as Thompson (Kroeber, 1999; Koch, 2008a) or Lillooet (Davis et al.,2004).b. It is an independent fact about Okanagan grammar that demonstratives can294form constituents with iP DPs (chapter 4).c. Okanagan bare CPs are surface indistinguishable from headless relativeclause DPs, despite the fact that iP is not a complementizer (cf. the me-chanics of relative clause formation, chapter 6), but Okanagan cleft residuesallow overt NP heads, unlike in Thompson (Koch, 2009).As such, the only surface distinction between DP-DP structures and clefts is thatthe second DP in a cleft contains clausal material, and so the semantics of equationmay proceed straightforwardly.Fourth, I return to a discussion of the DP focus interpretation (c) which initiallyseems problematic for the equative analysis because the presence of the initial,non-focused and deictically weak demonstrative is unexplained both in terms of itsstructural role as well as its interpretive role in the equative structure (8.5).4 Thereare several important facts worth noting here:a. For the DP Focus (c) interpretation, the initial demonstrative is optional.b. For the DP Focus (c) interpretation, the initial demonstrative is generallynot a spatial deictic.c. For the DP Focus (c) interpretation, a non-spatially deictic demonstrativecan also follow the focused DP (but not generally occur in both positions).d. For the Demonstrative Focus (a) and Demonstrative iP DP focus (b) inter-pretations, an additional, non-spatially deictic demonstrative can alwaysoccur initially.e. For simple DP-DP structures, an additional, non-spatially deictic demon-strative can always occur initially.f. Clefts with a proper name in focus position only allow the DP focus in-terpretation (c), since demonstratives do not form constituents with propernames.I suggest that the availability and optionality of non-spatially deictic demonstra-4?Non-spatially deictic? demonstratives, or ?deictically-weak? demonstratives may be analyzableas discourse deictics in some instances, as I discuss in section 8.6.2.295tives under all three interpretations, as well as with DP-DP structures, supports aunified analysis of all three interpretations as underlyingly equative (8.5.1).I claim that the initial non-spatially deictic ix?P demonstrative originates as aconstituent with the equative complement, and procliticizes to the focused DP ininitial position (8.5.2).5 This claim is supported by a variant of DP-DP structuresand clefts whereby the oblique marker t rather than the determiner iP introduces thefinal DP. Proclisis is for stylistic effect, possibly prosodically motivated, and servesto highlight the constituent in contrastive focus. Clefts with initial demonstrativesand those with demonstratives in their base-generated position (DP-DP structures)are truth-conditionally and pragmatically equivalent. Demonstrative proclisis is awidespread phenomenon in Okanagan, and not limited to sentence-types whichinvolve structural focus. I then discuss clefts which appear to have bare demon-stratives in focus position, and weigh evidence as to whether these cases actuallyinvolve a null focus (8.5.3). Finally, I discuss data showing that demonstrative-associated DPs allow intensional readings support an equative analysis of Okana-gan clefts (8.5.4).Before closing this chapter, I present my analysis of Okanagan clefts with sev-eral sample derivations (8.6), and a discussion of how Okanagan offers syntacticand semantic support for cleft theories such as Percus (1997). Finally, I summarizethe findings of this chapter (8.7).I include two addenda to this chapter. First, I include a section which shows thatNP residues may be analyzed in some cases as projecting clausal structure (8.8).The idea is supported by future clefts, i.e. clefts for which a residue is introducedby the future marker mi rather than a determiner. This analysis potentially extendsto simple nominal iP DPs in residue position. Second, I present data on a type ofcleft involving an adjunct, rather than an argument, in focus position (8.9). Adjunctclefts are not straightforwardly amenable to the equative analysis.5Non-spatially deictic demonstratives may in fact be discourse deictics, which is a hypothesis Iadvance below in section 8.6.2.2968.2 Introducing Introduced Clefts in Okanagan andAcross SalishThe purpose of this section is to briefly review clefts in other Salish languages, toshow how Okanagan cleft structures compare, and to present some basic morpho-syntactic facts about Okanagan clefts.8.2.1 The Classical Analysis: A Tripartite Structure?Introduced? clefts (Kroeber, 1999) in Salish are generally structures where a DPoccurs in ?focus position?, and is introduced by a dedicated clefting predicate, acopula of sorts, as in the Lillooet example in (4a) and the Thompson example in(4b), below. A residue clause, analyzed as a CP by Davis et al. (2004) for Lillooetand Northern Straits and Koch (2008a) for Thompson, follows the DP constituentin focus position. Bold-type indicates focus. This tripartite structure is representedschematically for (4a,b) as (5a,b) along with the descriptive terminology that I willuse.6(4) a. ni?CLEFT[?-JohnDP]NOM-JohntaDETkwan-ta?l?-hatake-NST-EXIStaDETk??h-?-acar-3.POSS-EXIS?-Mary.NOM-MaryIt?s John that took Mary?s car. (Lillooet, Henry Davis, p.c.)b. ?ceCLEFT[?DETR?ssDP]RosseDETp?nt-@t-mus.paint-DIR-3.ERG.EXTRIt was Ross that painted it. (Thompson, Koch (2008b, 2))(5) a. [ni?]cle f ting predicate [?-John] f ocus position [ta kwan-ta?l?-ha ta k??h-?-a ?-Mary]residue.b. [ ?ce]cle f ting predicate [? R?ss] f ocus position [e p?nt-@t-mus]residue.6Henry Davis (p.c.) notes that example (4) is the corrected form of sentence (14) in Davis et al.(2004, 106).2978.2.2 Introduced Clefts in OkanaganOkanagan also exhibits a class of sentences which are directly analogous to Lil-looet and Thompson (4); however there is no dedicated clefting predicate in thelanguage. Instead, canonical Okanagan clefts are introduced by a distal demon-strative ix?P, which is not always pronounced, as represented in (6a). The basicstructure of (6a), under a tripartite analysis, is given as (6b).(6) a. Context: I saw a deer and a bear on a stroll through the woods today, youdid not see the bear but rather the deer.(ix?P)CLEFT[iPDETs ??aPc?n@mDP]deeriPDETwik-@nt-xw.see-DIR-2SG.ERGDP Focus: It?s the deer you saw.b. [(ix?P)]cle f ting demonstrative [iP s ??aPc?n@m] f ocus position [iP wik@ntxw]residue.There are questions which arise from data like (6):a. Is the demonstrative ix?P in (37a) a clefting predicate or copula, analogousto Thompson ?ce (4b) and Lillooet ni? (4a)?b. Given that the clefting demonstrative in (6a) is optional, can (6a) be ana-lyzed as a cleft?The answers to question (a) is ?no?.7 Okanagan clefting demonstrative ix?P is prob-ably cognate with the demonstrative xeP in Thompson, which often encliticizes tothe clefting particle ?ce (7a), and clearly cognate with the Shuswap demonstrativey@G?P which is optional (7b) in the same contexts as the Okanagan demonstrative(6a).87Proximal ax?P does not typically function as a ?clefting? demonstrative, but when it occurs inclefts, it is normally in focus position. A sequence of two occurrences of ix?P is theoretically possible,but presumably filtered out by a general anti-haplology filter banning adjacent instances of the samelexical item. Sequences of ix?P ax?P are possible, with ax?P invariably being in focus.8I discuss in more detail clefts in other Salish languages in the implications section of the conclu-sion chapter, 9.3.298(7) a. ?c?CLEFTxePDEM[eDETMoniqueDP]MoniqueeDETwik-t-ne.see-DIR-1SG.ERGIt was Monique that I saw. (Thompson, Koch (2008b, 7))b. (y@G?P)DEM[GDETsqE?lmxwDP]manGiPDEMGDETwik-t-s.see-DIR-3SG.ERGIt?s the man that she saw. (Shuswap, Gardiner (1993, 76-78))Since the Thompson pattern shows that a clefting predicate can co-occur with aclefting demonstrative, I suggest that for Shuswap and Okanagan, the clefting pred-icate itself is null. This is consistent with the hypothesis that there is a null equativehead involved in Okanagan clefts, and that this null equative head is the cleftingpredicate.9The distribution of clefting predicates and clefting demonstratives across thesefour Interior Salish languages is as follows:Table 8.1: Clefting Predicates and Demonstratives across Four Interior SalishLanguagesOkanagan Shuswap Thompson Lillooetclefting predicate \u000B \u000B ?ce nilhdemonstrative ix?P y@G?P xeP \u000BIn answer to the second question then, Okanagan sentences like (6) fit into awider typology of sentences which are known as ?clefts? in the Salish literature,and so it makes sense to refer to these sentences as ?clefts? as well.8.2.3 DP Types and DP FocusAny kind of DP can occur in the focus position of a cleft, including iP DPs (8),proper names (9) and demonstratives (10). The initial non-spatially deictic cleftingdemonstrative is always a possibility, but never a requirement.9This implies that the clefting predicates in Lillooet and Thompson are spellouts of a null equativehead. Koch (2008a) analyzes Thompson ?ce as essentially a predicational copula with an associatedimplicature of existence. See section 9.3.2 for discussion.299(8) a. (ix?P)DEMiPDETylm?xw@mchiefiPDETkwu1SG.ABSwik-s.see-[DIR]-3SG.ERGIt?s the chief who saw me.b. (ix?P)DEMiPDETsmaPm?mwomenaPDETc-qw@l-qw?l-st-@n.CUST-RED-speak-(CAUS)-1SG.ERGIt?s the women I was talking to.c. (ix?P)DEMiPDETsqilxwIndian personiPDETw?k-@n,see-[DIR]-1SG.ERGlutNEGiPDETs?maP.white personIt?s Indian people I saw, not white people.(9) a. (ix?P)DEMRonRoniPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGIt?s Ron that I talked to.b. (ixiP)DEMSpikeSpikeiPDETkwu1SG.ABSx?lit-sinvite-[DIR]-3SG.ERGi-ks-kPam?n[?]-i ?wt-@m1SG.POSS-FUT-stand.next.to-MID?aPCOMPqw@l-qw?lt.RED-speakIt?s Spike who invites me to stand beside him when he speaks.(10) a. (ix?P)DEMax?PDEMiPDETlpotcupaPDETc- ?kw??l-@m-st-@n.CUST-make-MIN-CAUS-1SG.ERGIt?s this cup that I was using.b. (ix?P)DEMax?PDEMiPDETs@nk? ?caPsq?xaPhorseiPDETks- ??w?p-aPx.FUT-win-INCEPTIt?s this horse that is going to win.c. wa ?yyes(ix?P)DEMax?PDEM(iP)(DET)i-ks-c- ?kw??l.1SG.POSS-FUT-CUST-makeThis is what I?m gonna do.Notice that without the initial demonstrative, these clefts clearly resemble equative300DP-DP structures, except for the fact that the second constituent contains or con-sists of clausal material. I now turn to the information structure of clefts, whichprovides evidence that they are equivalent to DP-DP structures.8.3 Information Structure and CleftsThis chapter discusses the information structural properties of clefts. I show thatthese properties are the same as those found in DP-DP structures, as discussed inthe previous chapter. This makes clefts information-structurally amenable to theequative analysis I outlined in the previous chapter.8.3.1 Structural FocusEnglish clefts are clearly connected to some broader notion of focus, as are cleftsin Okanagan and other Salish languages. The propositional content of an Englishcleft is equivalent to that found in a non-cleft version of the same sentence (Reeve,2011), however there is an information structural difference: a clefted constituentreceives contrastive (a.k.a. identificational) focus (Rochemont, 1986; Kiss, 1998).Similar facts hold for Okanagan clefts. The demonstrative-focused cleft in(11a) is truth-conditionally equivalent to the non-cleft form (11b), however thereis an information structural difference: the demonstrative ix?P ?that? is a focus in(11a), but not in (11b), which is a garden variety transitive sentence involving aninitial demonstrative-associated DP object (cf. 3.3).(11) a. Context: You saw a specific deer earlier and now you?re seeing it again.ix?PDEMiPDETs ??aPc?n@mdeeriPDETwik-nt-xw.see-DIR-2SG.ERGThat?s the deer you saw.b. ix?PDEMiPDETs ??aPc?n@mdeerwik-nt-xw.see-DIR-2SG.ERGYou saw that deer.Given the morphosyntactic and interpretive differences between (11a) and (11b), itis reasonable to argue that (11a) involves structural focus, whereas (11b) does not.301The morphosyntactic correlate of a structural focus position in the case of (11a)is the fact that an iP determiner precedes a clausal predicate.10 DP-DP structureshave exactly the same property, as described in chapter 7.I claim that clefts share the same information structure as DP-DP structures:(i) they carry an exhaustivity implicature which is cancellable (8.3.2, cf. 7.4.1);(ii) there is no presupposition of existence tied to an Okanagan cleft (8.3.3); and(iii) the DP in focus must precede the residue (8.3.4). I then contrast introducedclefts with nominal predicate constructions, and argue that these may be analyzedas direct predications (8.3.5).8.3.2 The Exhaustivity Implicature in Clefts and Contrastive FocusAs first observed by Davis et al. (2004), clefts in Northern Straits Salish and Lil-looet lack the exhaustivity presupposition which is characteristic of English clefts(12).11 In this regard, Okanagan patterns similarly to the languages investigated byDavis et al. (2004).(12) It is [?]FOC that has the property ?presupposes ?x[?(x)?? x = ?] (only ? has the property ?)For Okanagan, a cleft must be used in contexts which require an exhaustive in-terpretation. This is directly parallel to the distinction between DP-DP and subject-initial direct predications (cf. 7.4.1). The questions in (13-14) involve subjectfocus, and require an exhaustive answer. Because they are not clefts, (13b) wasjudged infelicitous in this context (either with or without an initial demonstrative),as was (14c).(13) Context: One person got up early to leave the meeting.Question: Is Bill the one who left?10This means that (11a) lacks any categorially predicative constituent in the main clause (exceptfor the null equative head): Transitive w?k@ntxw ?you saw it? is introduced by an iP determiner, whichindicates that the predicate has been relativized. (11b), by contrast, does not involve structural focus,and has a clear main-clause predicate w?k@ntxw ?you saw it?.11See section 2.4.2 where I discuss the issue of whether exhaustivity is an entailment or presuppo-sition of English clefts. Since the choice between these is not crucial for my analysis of Okanaganclefts, I retain Percus? (1997) original analysis whereby exhaustivity is a presupposition.302a. (ix?P)(DEM)JohnJohniPDETnis,left,lutNEG?tEMPHBill.BillJohn (is the one that) left, not Bill.b. #(ix?P)(DEM)JohnJohnnis,left,lutNEG?tEMPHBill.BillJohn left, not Bill.(14) Context: You heard about a meeting at the health office, only one personshowed up, and you ask if Norman was that person.a. haYNQNormanNormaniPDETc-kic-x?CISL-arrive-INTRIs Norman the one that got there?b. ki ?w,yesNormanNormaniPDETc-kic-x.CISL-arrive-INTRYes, Norman (is the one who) got there.c. #ki ?w,yesNormanNormanc-kic-x.CISL-arrive-INTRYes, Norman got there.The determiner iP induces an exhaustivity implicature in these cases, exactlylike the determiner which introduces the second DP in a DP-DP structure. Omittingthe determiner, as in (13b) and (14c), yields a type of direct, verbal predication. In(13b), for example, John is in focus as an answer to the question, but the sentenceas a whole does not carry an exhaustivity implicature, as required by the context.Notice that in English, it is not necessary to use a cleft in response to the ques-tions in (13-14). That is, the answer does not have to assume a special morpho-syntactic form. I speculated in section 7.4.1 that this requirement may be relatedto the fact that English can signal exhaustivity by pitch accenting within a sentencewhich does not have a dedicated structural focus position, while Okanagan doesnot have this strategy available, assuming that it is similar to Thompson (Koch,2008a) in not using pitch accenting for focus. This interesting difference betweenEnglish and Okanagan merits further study.303In answer to questions which do not involve subject DP focus (15), the iP deter-miner is possible, but not required. With an iP determiner, as in (15c) for example,the speaker is implying that John is the only one that went.(15) Context: There was a meeting up at the health office, and I?m wondering ifJohn went.a. haYNQJohnJohnxwuygo?k[l]LOCc-qwaP-qwP?l?CUST-RED-speakDid John go to the meeting?b. wa ?y,yesJohnJohnxwuy.goYes, John went.c. wa ?y,yes(ix?P)(DEM)JohnJohniPDETxwuy.goYes, John is the one who went.For Okanagan clefts, exhaustivity is not a presupposition (Percus, 1997; Hed-berg, 2000), but only an implicature (Davis et al., 2004). In (16c), the implicaturethat John is the only one who stole a cookie is cancelled by the following conjunct,in which it is asserted that Peter also stole a cookie. The equivalent of (16c) inEnglish is infelicitous.(16) a. switwhoiPDETn? ?qw-@m-ssteal-MID-3SG.POSSiPDET?t@xt.sweetWho stole the cookie?b. JohnJohnn? ?qw-@m-ssteal-MID-3SG.POSSiPDET?t@xtsweetu?CONJnixwalsoPeterPeterna ?qwsteal.INTRtOBL?t@xt.sweetJohn stole a cookie, and Peter too stole a cookie.(adapted from (Krifka, 2008, 253))304c. JohnJohniPDETn? ?qw-@m-ssteal-MID[?]-3SG.POSSiPDET?t@xtsweetu?CONJnixwalsoPeterPeteriPDETna ?qwsteal(INTR)tOBL?t@xt.sweet#It?s John who stole a cookie, and Peter too stole a cookie.(adapted from (Krifka, 2008, 253))Taken together, the data in (13-16) show that a cleft structure is required for anexhaustive interpretation (cf. Thoma (2007) for similar facts in Lillooet), but ex-haustivity is not presupposed or entailed. (17b) below confirms that Okanaganclefts do not presuppose or entail exhaustivity, but that exhaustivity can be entailedby adding kmax ?only? (17c).12(17) a. sw?twhoiPDETPilxwt?hungryWho is hungry?b. ix?PDEMiPDETs-maP-mP?mNOM-RED-womenaPDETc-Pal-Pilxwt,CUST-RED-hungryu?CONJnixwalsoiPDETs-q@l-q@ltm?xw.NOM-RED-manIt?s the women that are hungry, and also the men.(adapted from Davis et al. (2004, 109-110))12In at least three Interior Salish languages (Lillooet, Thompson, and Okanagan), the equivalent of?only? consists of a predicate and an emphatic clitic. It may be the case that in Okanagan, emphatic?ti contributes the exhaustivity entailment rather than kmix ?only?, as shown by the infelicity of (i)below, though the semantic and information structural contribution of each particle remains unclearat this point.(i) *?tiEMPHiPDETs-maP-mP?mNOM-RED-womenaPDETc-Pal-Pilxwt,CUST-RED-hungryu?CONJnixwalsoiPDETs-q@l-q@ltm?xw.NOM-RED-manIt?s only the women that are hungry, and also the men.In Lillooet, tsukw ?only? rather than emphatic t?u7 appears to be the main carrier of exhaustivity(Henry Davis, p.c.), similarly to the cognate particles in Thompson (cf. Koch and Zimmermann(2009)).305c. ?tiEMPHkmaxonlyiPDETs-maP-mP?mNOM-RED-womenaPDETc-Pal-Pilxwt.CUST-RED-hungryIt?s only the women that are hungry.(adapted from Davis et al. (2004, 109-110))d. #?tiEMPHkmaxonlyiPDETs-maP-mP?mNOM-RED-womenaPDETc-Pal-Pilxwt,CUST-RED-hungryu?CONJnixwalsoiPDETs-q@l-q@ltm?xw.NOM-RED-man#It?s only the women that are hungry, and also the men.(adapted from Davis et al. (2004, 109-110))The interaction between the focus-sensitive operator kmax ?only? and the con-stituent in focus is an interesting issue worth exploring for Okanagan, but goesbeyond the scope of this thesis.8.3.3 Clefts are Non-presuppositionalOkanagan clefts also lack the presupposition of existence usually associated withEnglish clefts (18). This has already been observed for Northern Straits and Lil-looet by Davis et al. (2004), and for Thompson in Koch (2008a).(18) (Percus, 1997, 339)In a cleft of the form It is [?]FOC that has the property ?,there is a presupposition that ?x?(x)(there exists some individual that has the property ?).The residue clause of (19b) n ?kP?mt?ws@m@n ?I rode pro? is not presupposed in-formation, considering the question under discussion. Note that the English equiv-alent to (19b) is unacceptable as an answer to (19a), because of a presuppositionthat ?I rode something? which is not satisfied in this particular context.(19) a. sti ?mwhatan-c?wt2SG.POSS-doingsQapn?Pnowsx?@lx?Q?lt?dayWhat did you do today?306b. ix?PDEMiPDETs@nk? ?caPsq?x?aPhorseiPDETn-k-P?mt-?ws-@m-@n.n-sit-middle-MIN-[DIR]-1SG.ERGI rode that horse.(Literally, ?That?s the horse that I rode? or ?It?s the horse that I rode?.)By way of another example, qwl? ?w@m ?berry-picking? is given information in thecontext set up by the question in (20a), but it is not presupposed that any actualberry-picking occurred.(20) a. haYNQkw2SG.ABS?qw?l?w-@mpick.berries-MIDQapn?Pnowsx?@lx?Q?lt?dayDid you pick berries today?b. ix?PDEMiPDETs- ?p ?yq-a?qNOM-ripe-fruit(iP)(DET)i-s- ?qw?liw.1SG.POSS-NOM-pick.berriesI picked those berries.(Literally: Those berries are what I picked, or, It?s those berries that Ipicked.)The data in (21) below show that clefting is optional in both non-presuppositional(21a) and presuppositional (21b) contexts. When there is no iP determiner in theresponse, the structure is not a cleft, but a regular predication, possibly involvingunmarked fronting of a DP argument.13(21) a. Context: Friend/Addressee is visibly upset.kw2SG.ABSsxPk?nx?what.happenedWhat happened to you?b. Context: Friend sees that the addressee?s garden has been eaten.13The fact that clefting is optional in answer to (21b) is similar to the data given as (15) in theprevious sub-section. Clefting is required for an exhaustive interpretation, but clefting does notentail exhaustivity. By clefting in answer to (21b), the speaker is implying that the deer is the onlyindividual that ate his plants, however since clefting is optional here, it can be inferred that the contextdoes not require exhaustivity.307sti ?mwhatiPDETPi?-(?)t-@m-seat-APPL[?]-MID-3SG.POSSa-s- ?kw?n-?q?2SG.POSS-NOM-grow-cropWhat ate your plants?c. iPDETs ??aPc?n@mdeer(iP)(DET)kwu1SG.ABSPi?-(?)t-seat-APPL[?]-MID-3SG.POSSi-s- ?kw?n-?q.1SG.POSS-NOM-plant-cropA deer ate my plants. (good as answer to 21a and 21b).It should be noted that it is within the realm of possibility that Okanaganclefts (and DP-DP structures) do in fact carry an existential presupposition, butthat Okanagan speakers accommodate this presupposition more easily than En-glish speakers do. Because I am aligning the semantics and pragmatics of clefts tothe semantics and pragmatics of the determiner iP, there is good evidence that iP isnon-presuppositional, and parameterizing accomodation seems unlikely, I do notpursue this hypothesis (cf. also Matthewson (2006a) on the absence of presuppo-sitions in Lillooet).8.3.4 Focus Cannot Fall FinallyJust as with DP-DP structures, the DP in focus in an Okanagan cleft cannot occurfinally:14(22) Answer to the question ?Who is the Shuswap you talked to??a. (ix?P)DEMRonRoniPDETsyxw?p-m@xShuswap-personiPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGIt?s Ron who is that Shuswap I talked to.b. *(ix?P)DEMiPDETsyxw?p-m@xShuswap-personiPDETqw@l-qw?l-st-@nRED-speak-CAUS-1SG.ERGRon.RonThat Shuswap I talked to is Ron.14See Koch (2008a, 225) who makes the same point for Thompson.308(23) a. (ix?P)DEMiPDETsqilxwnative.personiPDETw?k-@n,see-[DIR]-1SG.ERGlutNEGiPDETs?maP.white.personThe ones I saw were native people, not white people.b. *(ix?P)DEMiPDETw?k-@nsee-[DIR]-1SG.ERGiPDETsqilxw,native.personlutNEGiPDETs?maP.white.personThe ones I saw were native people, not white people.This pattern stands in contrast to nominal predicate constructions (NPCs), wherea predicative NP is typically though not always in focus, yet can occur in eitherinitial or final position. For Okanagan NPCs, because the ordering of predicateand argument constituents in contexts involving WH-questions is relatively uncon-strained (cf. section 7.3.1), I take this as evidence that these are direct predications.I now discuss NPCs in some detail.8.3.5 Contrasting Clefts with Nominal Predicate ConstructionsIn this section, I compare the information structural properties of introduced clefts,as just discussed, with nominal predicate constructions (a.k.a. ?bare clefts?), whichI consider to be a sub-type of direct predication, on the grounds that (i) either thepredicate or the argument can be a focus, (ii) the ordering of predicate and argumentconstituents in contexts involving WH-questions is relatively unconstrained, and(iii) there is no implicature of exhaustivity.The simplest NPCs involve a nominal predicate taking a DP as a subject argu-ment (24), which may either follow or precede the predicate (cf. section 7.3.1).309(24) a. ?qwQay-lqsblack-robeiPDET??@x?- ??x??p.RED-grownThe old man is a priest.b. iPDET??@x?- ??x??pRED-grown?qwQay-lqs.black-robeThe old man is a priest.Nominal predicates may also select headed post-nominal (25b) or pre-posed (25c)relative clauses as arguments, as well as headless relative clauses (26-27).15 Thesetypes of NPCs are sometimes known as ?bare clefts? in the Salish literature (Kroe-ber, 1999).16(25) a. sti ?mwhattOBLsk@kQ?kaPbirdaPDETc-wik-st-xw?CUST-see-CAUS-2SG.ERGWhat are those birds that you saw?b. p@qlq?neaglesiPDETsk@kQ?kaPbirdsiPDETw?k-@n.see-[DIR]-1SG.ERGThe birds that I saw are eagles.c. p@qlq?neagle[iP][DET]i-s-c-w?k1SG.POSS-NOM-CUST-seeiPDETsk@kQ?kaP.birdsThe birds that I saw are eagles.(26) a. sti ?mwhatiPDETw?k@ntxw?see-DIR-2SG.ERGWhat did you see?b. x?w ??iPmountain.goatiPDETw?k-@n.see-[DIR]-1SG.ERGA mountain goat is what I saw.15See sections 4.3 and 6.3 for discussion of headless relative clauses.16The alternation between iP and t in question contexts such as (49a) is not well understood,though there may be a subtle semantic difference, possibly related to the alternation between iP andt in cleft contexts.310(27) a. haP ?k?nwhichiPDET??@x?- ??x??pRED-growniPDETqw@l-qw?l-st-xw?RED-speak-CAUS-2SG.ERGWhich old man did you talk to?b. syxw?p-m@xShuswap-personiPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGA Shuswap was the one that I talked to.(25-27) are straightforward variants of the simpler NPCs in (24), the only differ-ence being that the nominal predicate is taking a relative clause (either headed orheadless) as a direct argument.A relative clause within an NPC may be inflected with either ergative sub-ject morphology (e.g. 25b) or nominalized, possessor subject morphology (e.g.25c). For NPCs involving headless relative clauses in argument position, however,ergative-marked transitive relative clauses are often judged ungrammatical, as theexamples in (28) show (but cf. grammatical 26a,b):17(28) a. *s ??aPc?n@mdeeriPDETwik-nt-xw.see-DIR-2SG.ERGA deer is what you saw.b. *s ?wa?r? ?kxnfrogiPDETw?k-@nsee-[DIR]-1SG.ERGA frog is what I saw.As predicted under a direct predication analysis, NPCs with relative clauses inargument position also permit variable subject/predicate ordering (30):(29) haP ?k?nwhichiPDET??@x?- ??x??pRED-growniPDETqw@l-qw?l-st-xw?RED-speak-CAUS-2SG.ERGWhich old man did you talk to?17This may represent a general dispreference for ergative-centered relative clauses (Kroeber, 1999,305).311(30) a. syxw?p-m@x-@lxShuswap-person-3PLiPDET??@x?-@x?- ??x??pRED-RED-growniPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGShuswaps were the old men that I talked to.b. iPDET??@x?-@x?- ??x??pRED-RED-grownaPDETc-qw@l-qw?l-st-@nRED-speak-CAUS-1SG.ERGsyxw?p-m@x-@lx.Shuswap-person-3PLThe old men that I talked to were Shuswaps.(31) a. syxw?p-m@xShuswap-personix?PDEMiPDET??@x?- ??x??pRED-growniPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGThose old men I talked to are Shuswaps.b. ix?PDEMiPDET??@x?- ??x??pRED-growniPDETqw@l-qw?l-st-@nRED-speak-CAUS-1SG.ERGsyxw?p-m@x.Shuswap-personThose old men I talked to were Shuswaps.Headless relative clause arguments of nominal predicates (cf. 26) are moremarginal than headed relative clause arguments in elicitation contexts; however,these are also possible and sometimes volunteered, as the following question/an-swer pairs show. This distribution helps confirm the status of the determiner-introduced clauses in NPC contexts as DP arguments.1818The argument-indicating, reduced determiner preceding the nominalized clause iscwik ?my see-ing? in the answers to (32) is inferable from the absence of a determiner preceding the nominaltk?m?lxw ?woman?. (Recall that iP regularly reduces before 1st and 2nd person possessive prefixes.)This is because sentences consisting of two predicative elements are ungrammatical in Okanagan,e.g.:(i) * ??@x? ??x??pold.man?qwQaylqs.priestThe old man is a priest.312(32) a. Q: switwho[iP][DET]a-s-c-w?k?2SG.POSS-NOM-CUST-seeWho did you see?b. A1: tk?milxwwoman[iP][DET]i-s-c-w?k.1SG.POSS-NOM-CUST-seeI saw a woman. (Literally: Woman is the thing that was my seeing.)c. A2: [iP][DET]i-s-c-w?k1SG.POSS-NOM-CUST-seetk?milxw.womanI saw a woman. (Literally: The thing that was my seeing is woman.)Under the assumption that focus may be identified as the answer to a WH-question (Jackendoff, 1972; Selkirk, 1995), the answers in (30-32) show that likedirect predications, focus in an NPC is not necessarily aligned to the left-edge ofthe sentence in Okanagan, in contrast to introduced clefts and DP-DP structures,and in contrast to Thompson NPCs and introduced clefts as elucidated by Koch(2008a).19Despite data like (32) however, focus-final structures involving headless rela-tive clause subjects are usually unacceptable (33c).(33) a. Q: sti ?mwhat(iP)(DET)a-s-c-w?k?2SG.POSS-NOM-CUST-seeWhat did you see?b. A1: x?w ??iPmountain.goat[iP][DET]i-s-c-w?k.1SG.POSS-NOM-CUST-seeI saw a mountain goat.c. A2: *[iP][DET]i-s-c-w?k1SG.POSS-NOM-CUST-seex?w ??iP.mountain.goatI saw a mountain goat.19Koch (2008a) found that fronted subjects in Thompson NPCs are in a separate intonation phrase.As such, a focused predicate in final position may still be left-most within its intonation phrase. Seesection 9.3.4 for further discussion.313Compare also (34a) which has an initial headed relative clause DP subject with(34b) which has an initial headless relative clause as a subject. The dispreferencefor (33c) and (34b) may perhaps be more succinctly captured by the generalizationthat headless relative clauses may not easily precede their predicates.(34) a. iPDET??@x?-@x?- ??x??pRED-RED-grownaPDETc-qw@l-qw?l-st-@nCUST-RED-speak-CAUS-1SG.ERGsyxw?p-m@x-@lx.Shuswap-person-3PLThe old men that I talked to were Shuswaps.b. *aPDETc-qw@l-qw?l-st-@nCUST-RED-speak-CAUS-1SG.ERGsyxw?p-m@x-@lx.Shuswap-person-3PLThe ones that I talked to were Shuswaps.The structures I assume for Okanagan NPCs are given in (35-36). Assumingthat headless relative clauses involve clause-internal movement of a DP consistingof a D plus an empty NP, the syntactic position of the overt determiner depends onwhich of the two determiners (the external one introducing the containing DP, orthe internal one at the left periphery of the clause) is pronounced. I assume thatthe Double Determiner Filter of Davis (2010a, 22) ensures that only one of twoadjacent Ds can be pronounced.20(35) p@qlq?neaglesiPDETsk@kQ?kaP jbirds[[iPDET[\u000B NP j] DPi] w?k-@n ti CP].see-[DIR]-1SG.ERGThe birds I saw were eagles.(36) a. syxw?p-m@xShuswap-people\u000B(DET)\u000B j(NP)[[iPDET[\u000B NP j] DPi] qw@l-qw?l-st-@n ti CP].speak-CAUS-1SG.ERGShuswaps were the ones that I talked to.b. syxw?p-m@xShuswap-peopleiPDET\u000B j(NP)[[\u000B(DET)[\u000B NP j] DPi] qw@l-qw?l-st-@n ti CP].speak-CAUS-1SG.ERGShuswaps were the ones that I talked to.20See discussion of the Double Determiner Filter in section 6.3.3, where I motivate my analysis ofrelative clauses.3148.3.6 SummaryThis section has shown that clefts, like DP-DP structures, lack an exhaustivityentailment or presupposition of existence, and that the constituent in focus cannotoccur in final position. I take this as evidence that these two classes share a commonunderlying configuration. Clefts contrast with NPCs, which allow variable wordorder, like simple direct predications. In light of information structural parallelsbetween DP-DP structures and clefts, I now discuss some syntactic parallels whichsupport the hypothesis that Okanagan clefts are equative.8.4 An Equative Analysis of CleftsGiven the information structural parallels between DP-DP structures and clefts,this section presents the basic analysis of clefts as equative. Okanagan clefts arestraightforwardly amenable to the equative analysis because:a. Okanagan canonically introduces clefts with a demonstrative, rather thanthe dedicated clefting predicate characteristic of other Salish languages,such as Thompson (Kroeber, 1999; Koch, 2008a) or Lillooet (Davis et al.,2004).b. It is an independent fact about Okanagan grammar (and those of otherSalish languages (Matthewson, 1998)) that demonstratives can form con-stituents with iP DPs (chapter 3). This means that the introductory demon-strative can be analyzed as a constituent with an iP DP in an equative.c. Although Okanagan residue CPs are surface indistinguishable from head-less relative clause DPs, since both can be introduced by an iP determiner(cf. the mechanics of relative clause formation in chapter 6), cleft residuesin Okanagan allow overt NP heads, unlike in Lillooet (Davis et al., 2004)or Thompson (Koch, 2009). This means that the residue clause can beanalyzed as a DP constituent.As such, the only surface distinction between DP-DP structures and clefts is thatthe second DP in a cleft contains a relative clause.21 After presenting the basic21This is not true for adjunct clefts, where the residue clause is not a DP, but I defer discussion of315analysis in this section, I discuss more complex data, and further evidence for anequative analysis of clefts in subsequent sections.I begin with a discussion of structural and interpretive ambiguities in Okanaganclefts for which an iP DP occurs in focus position.8.4.1 A Structural and Interpretive AmbiguityThe fact that a demonstrative introduces Okanagan clefts results in an interpretiveand structural ambiguity, as illustrated below in (37). In (37a-b), the demonstrativeis deictic, and may either itself constitute the focused DP (37a) to the exclusionof the following headed relative clause DP, or else adjoin to a DP which is alsoin focus (37b). This variability is predicted since both lone demonstratives anddemonstrative-associated DPs independently function as DPs in argument contexts.In (37c), the optional demonstrative is not a spatial deictic, but is similar to theEnglish clefting particle it.(37) a. Context 1: You saw one distinctive looking deer earlier and now you?reseeing it again.[ix?PDP]DEMiPDETs ??aPc?n@mdeeriPDETw?k-@nt-xw.see-DIR-2SG.ERGDemonstrative Focus: That?s the deer you saw.b. Context 1: You saw one distinctive looking deer earlier and now you?reseeing it again.[ix?PDEM[iPDETs ??aPc?n@mDP]DP]deeriPDETw?k-@nt-xw.see-DIR-2SG.ERGDemonstrative-DP Focus: That deer is the one you saw.c. Context 2: I saw a deer and a bear on a stroll through the woods today, youdid not see the bear but rather the deer.(ix?P)(DEM)[iPDETs ??aPc?n@mDP]deeriPDETw?k-@nt-xw.see-DIR-2SG.ERGDP Focus: It?s the deer you saw.these until the end of the chapter.316These three interpretations are defined and summarized as follows:(38) a. Demonstrative Focus: The initial demonstrative is in focus, to the exclu-sion of the following iP DP and residue clause.b. Demonstrative DP Focus: The demonstrative and a constituent iP DP (in?focus position?) are in focus, to the exclusion of the final clause.c. DP Focus: The iP DP is in focus, to the exclusion of the initial demonstra-tive and residue clause.Given information structural parallels between DP-DP structures and clefts, Isuggest that the ?demonstrative focus? (a) and ?demonstrative DP focus? (b) inter-pretations are straightforwardly amenable to an equative analysis, and I show howin the next sub-section. The ?DP Focus? interpretation (c) is slightly more compli-cated, so I set this aside until section 8.5.8.4.2 Clefts as Equatives: The Basic IdeaThis section presents my basic argument that the focused constituent of a cleftis equivalent to the initial focused DP in a simple equative sentence, and that theresidue clause is a DP which can contain either a headed or headless relative clause.I claim that for the ?demonstrative focus? interpretation (a), a headed rela-tive clause DP is the residue and the demonstrative is in focus (39a); and for the?demonstrative iP DP focus? interpretation (b), a headless relative clause DP is theresidue, and a demonstrative-associated DP is in focus (39b). My proposed distri-bution of the null equative head is as indicated:(39) a. Context 1: You saw one distinctive looking deer earlier and now you?reseeing it again.[ix?PDP]DEM= [iPDETs ??aPc?n@mdeeriPDETw?k-@nt-xwDP].see-DIR-2SG.ERGDemonstrative Focus: That?s the deer you saw.b. Context 1: You saw one distinctive looking deer earlier and now you?reseeing it again.317[ix?PDEM[iPDETs ??aPc?n@mDP]DP]deer= [iPDETw?k-@nt-xwDP].see-DIR-2SG.ERGDemonstrative-DP Focus: That deer is the one you saw.The status of Okanagan residue clauses such as (39b) as DPs is not immediatelyobvious, since for headed relative clauses, an iP determiner usually introduces boththe head and the clause (see chapter 6 for details, and preceding discussion on NPCstructure). On the one hand, if we assume that the Double Determiner Filter isoperative in cleft residues, a headless relative (cf. 39b) may be analyzed as a DP.Such an assumption must be qualified, however, especially in light of the fact thatKoch (2008a, 105) analyzes Thompson cleft residues as CPs.Koch?s (2008) argument against a DP analysis of Thompson residues rests par-tially on the fact that they may not contain overt NP heads. In contrast to the caseof Thompson, Lillooet, and Northern Straits Salish, however, an Okanagan residueclause can be overtly headed, as shown in (40b). (40) shows more clearly than (39)that the residue consists of a DP, since a proper name cannot form a constituentwith an iP DP. This constitutes evidence that the residue clause in (40a) contains anull head, and therefore that the residue clause is categorially a DP.(40) a. Answer to ?Who did you talk to??(ix?P)DEMRonRon= iPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGThat?s Ron who (is the one) I talked to.b. Answer to ?Who is the Shuswap you talked to??(ix?P)DEMRonRon= iPDETsyxw?p-m@xShuswap-personiPDETqw@l-qw?l-st-@n.RED-speak-CAUS-1SG.ERGThat?s Ron who is the Shuswap I talked to.Further evidence that the residue clause in an Okanagan cleft is a DP comes fromclefts involving demonstrative focus (41a,b), particularly for a subset of caseswhich involve pre-posed relatives in residue position (41a). Data like (41a) showthat analyzing the DP iP s@nx?w@x?w?yaPqn ?the tepee? in (41b) as a constituent withthe pre-equative head demonstrative, to the exclusion of the residue clause, is not318viable.22(41) Context: Pointing out a particular tepee from among a group of tepees.a. ix?PDEM= iPDETs-c- ?kwu?l-sNOM-CUST-make-3SG.POSSiPDETs@nx?w@x?w?yaPqn.tepeeThat?s the tepee he made.b. ix?PDEM= iPDETs@nx?w@x?w?yaPqntepeeiPDETs-c- ?kwu?l-s.NOM-CUST-make-3SG.POSSThat?s the tepee he made.The alternation between (41a) and (41b) is expected, given that pre-posed and post-nominal relative clause DPs are possible in non-equative contexts (cf. chapter 6). Itis worthwhile noting that examples like (41a) are sometimes volunteered, but rarelyjudged grammatical in elicitation contexts. This fits with the general observationthat pre-posed relative clauses are more marked than post-nominal relatives.Finally, headed cleft residues show evidence for clause-internal movement,suggesting that these are true instances of relative clauses (42). Recall from section6.3.3 that relative clauses from which a passive agent are extracted are introducedby the sequence iP t. This sequence codes the relation of the head NP to the relativeclause predicate.23(42) ix?PDEM= tOBLtk?milxwwomaniPDETtOBL?c? ?m-qs-nt-@m.suck-nose-DIR-PASSThat?s the lady he was kissed by.22It is important to note that there is no information structural ambiguity for clefts with pre-posedheaded residues: the demonstrative is always in focus. If we were to propose that the initial demon-strative in (41a), for example, formed a constituent with the following headless relative clause iPsc ?kwu?ls ?what he made?, the prediction is that this sentence would be interpreted as ?That thing hemade is a TEPEE?, similarly to a specificational pseudocleft, but as we?ve seen, focus cannot occurfinally in an Okanagan cleft, and so this interpretation is absent. (41a) is also not interpretable asmeaning ??THAT thing he made is a tepee? (as opposed to the other things he made, which are nottepees), since I have found that pre-posed clefts are generally only felicitous when the contrast setconsists of instantiations of the relative clause head, in this case, tepees. (41a) is however ambigu-ously interpretable as ?That?s the one who made the tepee?, in which case the DP relative in theresidue contains a null subject-centered head.23I discuss the fact that t rather than iP introduces the head of the residue in (42) in the next section,8.5.319I conclude that the residue of an Okanagan cleft is categorially a DP, and thatclefts with demonstrative focus and demonstrative iP DP focus interpretations maybe analyzed as equative structures on a par with simple DP-DP structures, as dis-cussed in chapter 7. I now discuss ?DP Focus interpretation? clefts, which typicallyinvolve an optional, non-spatially deictic demonstrative. The occurrence of simi-lar demonstratives in both DP-DP structures and clefts is evidence for a commonequative analysis.8.5 Morpho-syntactic Evidence for the Equative Analysisof Clefts and DP-DP StructuresThis section presents the problem of ?DP Focus interpretation? clefts, where anintroductory demonstrative is normally non-spatially deictic. I first show that thesedemonstratives occur in both DP-DP structures and clefts, and may either precedeor follow the DP in focus position. Next, I argue that the demonstrative is base-generated as forming a DP constituent with the residue, and optionally moves tothe front of the sentence through a process of stylistic proclisis.8.5.1 Clefts and the (Optional) Initial DemonstrativeThe ?DP Focus? interpretation, as defined above in (38c), initially seems to raisequestions for the equative analysis. To illustrate, if the equational functional headis placed as shown in (43a-c), then the presence of the initial, non-focused andnon-spatially deictic demonstrative is unexplained, both in terms of its syntacticand semantic role.(43) a. (ix?P)DEMiPDETylm?xw@mchief= iPDETkwu1SG.ABSwik-s.see-[DIR]-3SG.ERGDP Focus: It?s the chief who saw me.b. (ix?P)DEMax?PDEM= iPDETs@nk? ?caPsq?xaPhorseiPDETks- ??w?p-aPx.FUT-win-INCEPTThis is the horse that?s going to win.320c. (ix?P)DEMDannyDanny= iPDETxwi ?c-xt-sgive-BEN-3SG.ERGtOBLsqla ?wmoneyDion.DionDP Focus: It?s Danny who gave some money to Dion.(H?bert, 1980, recordings)There are two important facts worth noting here relating to the initial optionaldemonstrative, which I claim reinforce the equative analysis. First of all, (44)shows that both clefts (44a) and simple DP-DP structures (44b) may be introducedby non-spatially deictic, optional demonstratives:(44) a. (ix?P)DEMJohnJohn= iPDETPi?-seat-[DIR]-3SG.ERGiPDET?p@l.appleIt?s JOHN who ate the apple.b. (ix?P)DEMJohnJohn= iPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn is the teacher.Second, the non-spatially deictic demonstrative may either precede or followthe focused DP in a cleft (45) or a DP-DP structure (46) with no apparent semanticor pragmatic difference.(45) a. (ix?P)DEMMaryMary= iPDETkwu1SG.ABSxwi ?c-xt-sgive-BEN-3SG.ERGtOBLy?mx?waP.cedar.bark.basketIt?s Mary who gave me a basket.b. MaryMary= (ix?P)DEMiPDETkwu1SG.ABSxwi ?c-xt-sgive-BEN-3SG.ERGtOBLy?mx?waP.cedar.bark.basketMary is the one who gave me a basket.321(46) a. switwhoQ?pnaPnowiPDETylm?xw@m?chiefWho is the chief now?b. (ix?P)DEMSpikeSpike= iPDETylm?xw@m.chiefIt?s Spike who is the chief.c. SpikeSpike= (ix?P)DEMiPDETylm?xw@m.chiefSpike is the one who is the chief.I suggest that the availability and optionality of non-spatially deictic demonstra-tives under all three cleft interpretations, as well as DP-DP structures, supportsa unified analysis of all three interpretations as underlyingly equative. I claimthat the optional pre-focus demonstrative is equivalent to the optional post-focusdemonstrative in these examples, and that the demonstrative in these cases is base-generated in a post-focus position, as forming a DP constituent with the residue.The demonstrative moves to the front of the sentence via an optional, late-derivational,morpho-phonological, stylistic proclisis. The next section presents evidence forsuch an analysis.Before moving on, however, the natural question arises given data like (45-46)as to whether or not two non-spatially deictic demonstratives can occur in both po-sitions at once. The proclisis analysis predicts that this is not possible, and indeed,for cases involving two demonstratives, at least one of the demonstratives will usu-ally be interpreted as a spatial deictic. In example (47b), for example, the initialdemonstrative is a spatial deictic. Especially given the WH-question under discus-sion (47a), (47b) is most straightforwardly analyzed as involving a left-dislocateddemonstrative-associated DP plus an equative structure (47b).2424Although acoustic evidence would be useful in confirming this hypothesis.322(47) a. sti ?mwhatyaPx??sDEMiPDETk?-x?s-i ?ws?k?-good-middleWhat is that field over there?b. ixiPDEMiPDETk?-x?s-iwsk?-good-middleix?PDEMiPDETs@n- ?kw?n-?q-t@n.LOC-grow-crop-INSTRThat field is a garden.Literally: That field, it?s a garden.Other data such as (48) might in principle be analyzed as involving two non-spatially deictic demonstratives, but (48) is just as easily analyzable as a sequenceof two equatives: the first is a truncated cleft, and the second is a cleft with ademonstrative in focus position, and a headless relative clause DP in residue posi-tion.(48) ixiPDEMiPDETs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDix?PDEMiPDETkwu1SG.ABSqw@l-qw?l-st-s.RED-speak-CAUS-3SG.ERGIt?s the teacher that is the one who spoke to me.It?s the teacher. That/she is the one that spoke to me.In sum, I take data such as (47-48) to be indirect support for the proclisis analysis.8.5.2 Demonstrative Proclisis in EquativesThis section presents evidence for the idea that a non-spatially deictic cleftingdemonstrative originates in a position adjoined to the residue DP.First, consider that there is a variation on DP-DP structures (49a) whereby trather than iP introduces a nominal in final position, as in (49b,c) below. (Thesame variation is apparent in clefts, as we will see.)(49) a. ix?PDEMiPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThat?s the teacher.323b. ix?PDEMtOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThat?s the teacher.c. ix?PDEMmatEPIStOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThat might be the teacher.The oblique marker t is restricted to occurring just before nouns in these con-texts.25 This means that residue DPs without overt NP heads cannot be introducedby t (50-52).26 The distribution of t in this environment indicates that it must selectfor an overt NP, unlike iP which can license a null NP head. The demonstrativeix?P can freely select for either iP or t (49a,b; 52).27(50) a. *ax?PDEMtOBLks- ??w?p-aPx.FUT-win-INCEPTThat?s what will win.b. ax?PDEMtOBLs@nk? ?caPsq?x?aPhorseiPDETks- ??w?p-aPx.FUT-win-INCEPTThis is the horse who will win the race.(51) a. ix?PDEMiP/*tDET/*OBLpaP-nt-?s.fold-DIR-3SG.ERGThat?s what he folded.25In cases where t is associated with a constituent extracted from inside a relative, t may precedea verbal predicate (cf. chapter 6).26Adjectives, like verbs, may also not be introduced by t in these contexts:(i) a. ix?PDEMax?PDEMiPDETcax?.redThis is the one that is red.b. *ix?PDEMax?PDEMtOBLcax?.redThis is the one that is red.27The identical pattern holds DP-internally in Lillooet. A demonstrative tiP ?that? may select foreither an assertion-of-existence DP headed by the determiner ti...a, or a non-assertion-of-existenceDP headed by the determiner kwu. kwu requires an overt NP head, however ti...a freely allowsheadless relative clauses (Matthewson and Davis, 1995). See related discussion in section 4.6.1.324b. ax?PDEMiP/*tDET/*OBLxw?y-st-@m.go-CAUS-1PL.ERGThat?s what we will bring along.c. ix?PDEMiP/*tDET/*OBLyalt.run.away.That?s the one that ran away.(52) a. Context: Pointing out the deer that was just shot.ix?PDEMiP/tDET/OBLs ??aPc?n@m.deerThat?s the deer.b. Context: Picking out an old woman out of a police line-up.ax?PDEMiP/tDET/OBLp@ptw?naxw.old.womanThat?s the old lady.Next, recall from section 4.6.1 that an oblique marked NP is not itself a DP,as shown by its inability to occur in transitive argument contexts (53a) or in thefocus position of a cleft (53b). A demonstrative licenses an oblique marked NP tofunction as an argument in both contexts (54), however.(53) a. *c-n-kwn?-st-@nCUST-n-sing-CAUS-1SG.ERG[tOBLqw??l@m DP].songI sang that song.b. *[tOBLqw??l@m DP]songiPDETc-n-kwn?-st-@n.CUST-n-sing-CAUS-1SG.ERGIt?s the song that I sang.325(54) a. c-n-kwn?-st-@nCUST-n-sing-CAUS-1SG.ERG[ix?PDEMtOBLqw??l@m DP].songI sang that song.b. [ix?PDEMtOBLqw??l@m DP]songiPDETc-n-kwn?-st-@n.CUST-sing-CAUS-1SG.ERGTHAT?s the song that I sang.Also, note that in the context of a demonstrative, t is sometimes interchangeablewith the iP determiner in argument positions, with no apparent semantic distinc-tion.28(55) a. c-n-kwn?-st-@nCUST-n-sing-CAUS-1SG.ERG[ix?PDEMtOBLqw??l@m DP].songI sang that song.b. c-n-kwn?-st-@nCUST-n-sing-CAUS-1SG.ERG[ix?PDEMiPDETqw??l@m DP].songI sang that song.While the internal structure of a DP consisting of a demonstrative associated withan oblique t NP is unclear,29 it is sufficient for current purposes to note that anoblique-marked NP is not a DP, but when it forms a constituent with a demonstra-tive, the resulting constituent functions syntactically as a DP.With this background in mind, I now present the argument for proclisis. Acontrast surfaces between DP-DP structures with focused demonstratives on theone hand (56), and those with focused proper names (57) or iP DPs (58), such thatthe latter do not allow the t-variant.28It is only sometimes interchangeable in argument contexts, for reasons I cannot yet determine.29In chapter 4 I argued that there is no evidence for a null determiner for oblique quasi-objects,and since Okanagan demonstratives do not occur in D position, the most straightforward analysis ofthis particular occurrence of t is as an idiosyncratic spell-out of iP, but this too runs into problems,for two reasons. First, the alternation in clefts carries an information-structural function (discussedin the next section): t unambiguously signals that what follows it is not in focus, and demonstrativesare not optional for t clefts. Second, iP may select for a headless relative clause in this environment,whereas t may not (cf. section 8.5.2, and (Matthewson and Davis, 1995) and (Matthewson, 1998,section 3.2.4) for Lillooet). I leave the problem of the internal constituency of DPs consisting of ademonstrative-associated t NP to further work.326(56) a. ix?PDEMtOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThat?s the teacher.b. ix?PDEMiPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThat?s the teacher.(57) a. *JohnJohntOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn?s the teacher.b. JohnJohniPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn?s the teacher.(58) a. *iPDETp@ptw?naxwold.ladytOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThe old lady is the teacher.b. iPDETp@ptw?naxwold.ladyiPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDThe old lady is the teacher.Because a demonstrative is needed to syntactically license a t NP to functionas a DP, (57a) cannot be a DP-DP structure. Adding a demonstrative either beforeor after the proper name makes the structure licit, as in (59).(59) a. ix?PDEMJohnJohntOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn?s the teacher.b. JohnJohnix?PDEMtOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDJohn?s the teacher.I therefore claim that the demonstrative is base generated and interpreted as aDP constituent with the residue (60a), but undergoes optional proclisis to initial327position (60b). I also extend this analysis to canonical DP-DP structures, where iPintroduces the second DP (60c).(60) a. [John DP]John==[ix?PDEMtOBLs@xw- ?maP- ?m?yaP-m DP].OCC-RED-teach-MID(i) It?s John who is the teacher.(ii) John is that teacher.b. ix?P1DEM[John DP]John==[t1 tOBLs@xw- ?maP- ?m?yaP-m DP].OCC-RED-teach-MID(i) It?s John who is the teacher.(ii) John is that teacher.c. ix?P1DEM[John DP]John==[t1 iPDETs@xw- ?maP- ?m?yaP-m DP].OCC-RED-teach-MID(i) It?s John who is the teacher.(ii) John is that teacher.Notice that under one possible interpretation of (60), ?John is that teacher?, thedemonstrative is deictic and construed as a (sometimes discontinuous) constituentwith the oblique-marked NP. This is strong evidence that an initial demonstrative issemantically interpreted in its post-copular position.30 Under the other interpreta-tion of (60), the demonstrative does not appear to be a spatial deictic, but I suggestthat it may be analyzed as a discourse deictic (cf. section 8.6.2).Clefts with demonstratives in focus show the same variation between t and iP(61).31 The iP variant (61b) displays the 3-way interpretive ambiguity, however,while the t variant does not (61a), allowing only the demonstrative focus reading.This is somewhat unexpected given that the demonstrative-associated t NP shouldbe able to function as a focused DP constituent. The interpretive restriction on(61a) suggests that (62) is not a possible structure.30The iP-variant equivalent of (60) could be analyzed as fitting into Higgins? identificational sen-tence class, especially since the demonstrative appears to be deictic in these cases. But since theproper name must be in focus for (60), and focus cannot occur finally, I analyze these as equatives.31See H?bert (1982a, 355-356) for additional examples of t-variant cleft residues, involving bene-factive applicative residue predicates.328(61) a. ixiPDEMtOBLst??@mboatiPDETs-c- ?kwu?l-sNOM-CUST-make-3SG.POSSi-sl?x?t.1SG.POSS-friendThat?s the boat that my friend made (not this boat).*That boat is the one that my friend made (not this house).*It?s the boat that my friend made (not the house).b. ixiPDEMiPDETst??@mboatiPDETs-c- ?kwu?l-sNOM-CUST-make-3SG.POSSi-sl?x?t.1SG.POSS-friendThat?s the boat that my friend made (not this boat).That boat is the one that my friend made (not this house).It?s the boat that my friend made (not the house).(62) *[ixiP t st??@m DP] = [iP sc ?kwu?ls isl?x?tDP].In clefts (and DP-DP structures) any material following t is not in focus, so I ana-lyze all material following t as part of the residue.t-variant clefts involving pre-posed relative clause residues (63, cf. 41a) showthe same interpretive restriction as canonical cases (61): the demonstrative must bein focus to the exclusion of the remaining material.(63) Context: Couple of frogs were hopping around then disappeared, you sawone. When they re-appear, I asked you which one you saw.ix?PDEM= iPDETw?k-@nsee-[DIR]-1SG.ERGtOBLs ?wa?r? ?kxn.frogThat?s the frog that I saw.Demonstrative proclisis occurs not only in DP-DP structures, but also in clefts(64), though it is less clear whether proclisis occurs in clefts like (61-63), whereonly a demonstrative is in focus. I address this question in the next section.(64) a. NancyNancy= ix?PDEMtOBLs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDiPDETkwu1SG.ABSwik-s.see-[DIR]-1SG.ERGIt?s/That?s Nancy who is the teacher who saw me.329b. ix?PDEMNancyNancy= tOBLs@xw- ?maP- ?m?yaP-mOCC-RED-teach-MIDiPDETkwu1SG.ABSwik-s.see-[DIR]-1SG.ERGIt?s/That?s Nancy who is the teacher who saw me.Before closing, it is important to note that demonstrative proclisis is not lim-ited to equative environments. In transitive contexts, a demonstrative-associatedto a post-verbal object DP can move to the front of the sentence. This is mostclearly shown with proximal ax?P (65a,b), since an initial distal ix?P in this con-text can be construed as a non-constituent, discourse functor.32 Nevertheless, fordemonstrative-associated transitive object NPs introduced by t, an initial demon-strative can safely be construed as an underlying constituent with the t NP (66a,b).(65) a. kaPk?c-@nfind-[DIR]-1SG.ERGax?PDEMiPDETlpot.cupI found this cup.b. ax?PDEMkaPk?c-@nfind-[DIR]-1SG.ERGiPDETlpot.cupI found this cup.(66) a. s?w-@ndrink-[DIR]-1SG.ERGix?PDEMtOBLsiw?kw.waterI drank that water.b. ix?PDEMs?w-@ndrink-[DIR]-1SG.ERGtOBLsiw?kw.waterI drank that water.32The demonstrative ix?P has various poorly described discourse functions, among them signallingtemporal sequencing (Lyon, 2010b):(i) wa ?yyesix?PDEMlx?w ?p-?mrun.out-MIDax?PDEMiPDETs@xw- ?kw??l-@m.OCC-work-MIDThen he ran out, the working man. (Colville, A. Mattina (1985, stz.450))330Cross-Salishan evidence for demonstrative proclisis in Okanagan comes fromLillooet, where a demonstrative may undergo enclisis to second position, detachingfrom the DP with which it associates (Matthewson and Davis, 1995, 21):(67) ni?CLEFTtiPDEMk@?laPfirsts? ?qu?halflc?PahereLillooet-aLillooet-DETkwuDETsm??ac?.womanShe was the first half-breed woman in Lillooet.(Lillooet, van Eijk and Williams (1981, 70))I conclude that demonstrative proclisis in Okanagan serves a stylistic function,or possibly a prosodic function, yet to be determined.8.5.3 Null FociDemonstrative proclisis appears to have a stylistic or prosodic function: sentenceswith initial demonstratives are truth conditionally equivalent to sentences with non-initial demonstratives. The two variants also seem to be information structurallyequivalent in equatives: in t-variant clefts where there is a DP focus (e.g. Ron in(68a)), the demonstrative is invariably not in focus, while for t variant clefts whereonly a demonstrative precedes t, the demonstrative must be in focus (68b).(68) a. ix?PDEMRonRontOBLsyxw?p-m@xShuswap-personaPDETc-qw@l-qw?l-st-@n.CUST-RED-speak-CAUS-1SG.ERGThat?s Ron who is the Shuswap I was talking to.b. ix?PDEMtOBLsyxw?p-m@xShuswap-personaPDETc-qw@l-qw?l-st-@n.CUST-RED-speak-CAUS-1SG.ERGThat?s the Shuswap I was talking to.The question arises as to whether the demonstrative in (68b) can undergo proclisis,and if so, to what it attaches.At first glance, it appears that the demonstrative in (68b) has raised to a fo-cus position, but this creates an analytical inconsistency, since the demonstrativein (68a) is definitely not a constrastive focus. There are two potential explanationsfor the pattern in (68b), as shown below for simpler equative cases: (i) The demon-strative in focus position is null (i.e. ?a null focus?), and the demonstrative in the331residue DP either does not move (69a) or else procliticizes to the null subject (69b);(ii) The focus position is filled by a demonstrative DP (70a), while the demonstra-tive in residue position is not spelled-out, presumably due to some restriction ona sequence of two identical demonstratives. Proclisis in this case (70b) would bevacuous.(69) Null Focus Hypothesisa. [\u000BDP] ==[ix?PDEMtOBLs@xw- ?maP- ?m?yaP-mDP].OCC-RED-teach-MIDThat?s the teacher.b. ix?P1DEM[\u000BDP] ==[t1 tOBLs@xw- ?maP- ?m?yaP-mDP].OCC-RED-teach-MIDThat?s the teacher.(70) ?Double Demonstrative? Filter Hypothesisa. [ix?PDP]DEM==[(ix?P)DEMtOBLs@xw- ?maP- ?m?yaP-mDP].OCC-RED-teach-MIDThat?s the teacher.b. (ix?P)1DEM[ix?PDP] ==[t1 tOBLs@xw- ?maP- ?m?yaP-mDP].OCC-RED-teach-MIDThat?s the teacher.The same analytical ambiguity technically exists for iP-variant equatives with demon-strative focus interpretations, equivalent to (69-70), although since iP DPs do notneed demonstratives to license them, an initial focused demonstrative is most straigh-forwardly analyzed in base-generated position. In other words, no proclisis is in-volved in these cases.There is cross-linguistic evidence supporting the existence of null foci. In Lil-looet, for example, where clefts are introduced by an unambiguous clefting predi-cate nilh, a demonstrative focus is possible (71a) but not always overtly pronounced(71b). Shank (2003) also provides an example of a cleft from Northern Straitswith a null, focused 3rd person pronoun (72)33, and Koch (2008a) conjectures that33It is unclear in what position in (72) Shank considers the null pronoun to occur.332Thompson (73) involves a null focus on the basis that the demonstrative xeP is nottypically used to refer to people.(71) a. ni?CLEFTtiPDEMwaPIMPFl@x?-l?x?-?-anRED-remember-CAUS-1SG.ERGi-w-anwhen.PAST-IMPF-1SG.CJCTkwikw?.smallThat?s what I remember from when I was small.(Lillooet, Matthewson (2005, 404, ex.333))b. ni?CLEFT\u000B(DEM)waPIMPFl@x?-l?x?-?-anRED-remember-CAUS-1SG.ERGi-w?P-a?-tuPwhen.PAST-IMPF-3.CJCT-THEN???q-mi ?n-c?-asarrive-RED-1SG.OBJ-3.ERGkw-sDET-NOMPipay?n.PipayanI remember when Pipayan came to fetch me.(Lillooet, Matthewson (2005, 358, ex.31))(72) P@ ?wLNKhayonlyPa?ljust[?ts-@tbreak-TRkws@DET??c? ?wiPCP].plateHe?s the only one that broke a plate.(Northern Straits, Shank (2003, 232, f.n.20))(73) ?,oh?c?CLEFTxePDEM[?DETn-sk?xzePFOC]1SG.POSS-mothereCOMP?qwy-? ?w-mripe-harvest-MIDteOBLs ?qw?yt,fruitPe?and?c?CLEFTxePDEM[\u000B FOC]3SGeCOMPs-txw-?p-sNOM-buy-INCH-3SG.POSSteOBLm? ??qiymushroomtuxwfromeDETnt?wmn.storeIt was my mother that picked the fruit, and it was her that bought themushrooms at the store. (Thompson, Koch (2008a, 273, ex.51))333In light of these facts, the null focus hypothesis (69) seems plausible for Okana-gan, specifically (69a), since there is no morpho-phonological motivation to pro-cliticize to a null subject (69b). The null focus hypothesis does not reflect theassertive force of an equative, however. To explain, in (69) the assertion is ?Thatteacher is [\u000B FOC]?, but this does not intuitively seem to be correct. (70) straight-forwardly reflects the equative assertion, and is also more in line with my analysisof clefts involving proper names in focus position (cf. 64).This does not lessen the possibility that the null focus hypothesis is correct forLillooet, Northern Straits, and Thompson, however. But unlike Okanagan, theselanguages have overt clefting predicates, and so a null focus is recoverable fromthe syntactic context. In other words, unlike Okanagan ix?P, there is no sense inwhich a clefting predicate will be misconstrued as a DP constituent.8.5.4 Intensionality and Cleft ResiduesAssuming that the demonstrative proclisis analysis is correct, the equative analysisas applied to a cleft like (74) predicts that the residue DP has the semantic de-notation given as (75). In other words, demonstrative-associated DPs must allowintensional readings, just like regular iP DPs (cf. section 5.2.9).(74) ix?PDEMRonRontOBLsyxw?p-m@xShuswap-personaPDETc-qw@l-qw?l-st-@n.CUST-RED-speak-CAUS-1SG.ERGThat?s Ron who is the Shuswap I was talking to.(75) ?w[ f (?x[that Shuswap I was talking to(x)(w)?C(x)(w)])]This seems reasonable in light of the fact that the demonstrative in a cleft such as(74) is normally not a spatial deictic, similar to iP DPs, and so should be able tomake reference to an individual concept. Examples (76-77) are not clefts, but showdemonstrative-associated DPs denoting individual concepts.334(76) Context: A chief sent you some flowers, but you have no idea who thechief is.matEPISswitwho[ix?PDEMtOBLylm?xw@m DP]chiefmatEPISx?astgoodtOBLsqilxw.native.personWhoever that chief is, he must be a good person.(77) switwhomatEPIS[ix?PDEMtOBLsqilxwnative.personaPDETc- ?maP- ?m?yaP-mDP],CUST-RED-teach-MIDkmaxonly?tiEMPHsyilxOkanagan.personiPDETk(?)-s- ?maP- ?m?yaP-s.U.POSS-NOM-RED-teach-3SG.POSSSomeday, whoever is teacher will be teaching only in Okanagan.Demonstrative-associated iP DPs allow generic readings (78). I analyzed Okana-gan generics as intensional maximal pluralities, following Chierchia (1998), in sec-tion 5.2.8.(78) P??-@neat-[DIR]-1SG.ERGix?PDEMiPDETq?qxw@lxfishyaQy?Qtallsx?@lx?Q?lt.dayI eat that kind of fish every day.Literally: I eat that fish every day.SM: Yeah, you?re talking about whatever kind of fish, ling-cod, kokanee,salmon.I do have several examples of clefts for which a demonstrative-associated cleftresidue allows an intensional reading. Example (79a) shows a demonstrative in itsbase generated position adjoined to the residue clause. The entire residue in thiscase minimally denotes ?that person who helped me?, but given the non-specificinterpretation of the DP in focus position, the residue clause may denote a maxi-mal plurality, i.e. ?that kind of individual that helped me?. In other words, sincethe alternatives to iP sq@ltm?xw ?a man? consist of ?a woman?, ?a child?, etc., thedemonstrative-associated residue has a generic reading. The initial modal mattakes scope over the entire cleft, and binds the world variable of the maximal plu-ral individual denoted by the residue clause. The residue clause in (79b) displays anon-generic intensional reading.335(79) a. matEPISiPDETsq@ltm?xwmanix?PDEMiPDETkwu1SG.ABSkn-xit-s.help-BEN-3SG.ERGMaybe a man was the one who helped me.b. ix?PEPISmatDETtmansq@ltm?xwDEMiPDETkwu1SG.ABSkn-xit-s.help-BEN-3SG.ERGThat might be the man who helped me.In sum, demonstrative-associated DPs are semantically similar to regular iPDPs in terms of allowing intensional readings. This is consistent with an equativeanalysis of clefts, whereby the first argument of the equative head is an intensionalindividual (7.5).8.6 Analysis8.6.1 Syntactic DerivationThis chapter has presented both information structural and morpho-syntactic argu-ments in favor of an equative analysis of Okanagan clefts. Sample cleft derivationsare shown below (overlooking the details of relative clause formation, cf. section6.3.3). A demonstrative focus cleft is shown in (80a), a demonstrative DP focuscleft in (80b), and an iP DP focus cleft in (80c). Optional proclisis is indicated witha subscript ?1?.(80) a. (ix?P1)(DEM)[ax?PDP]DEM[ [= Id]?BE?[t1 [t/iPOBL/DETs ??aPc?n@msee-DIR-2SG.ERG[iPOBL/DETw?k-@nt-xwCP]deerDP]DP]Id?].This?s the deer you saw (i.e. not that one).b. (ix?P1)(DEM)[ax?PDEM[t/iPOBL/DETs ??aPc?n@mDP]DP]deer[ [= Id]?BE?[t1 [(iP)(DET)\u000B[iPDETw?k-@nt-xwCP]see-DIR-2SG.ERGDP]DP]Id?].This deer is the one you saw (i.e. not that bear).336c. (ix?P1)(DEM)[iPDETs ??aPc?n@mDP]deer[ [= Id]?BE?[t1 [(iP)(DET)\u000B [iPDETw?k-@nt-xwCP]see-DIR-2SG.ERGDP]DP]Id?].It?s the deer that you saw (i.e. not the bear).Syntactically speaking, Okanagan supports theories of clefts such as Percus(1997) and Hedberg (2000), as discussed in section 2.4.34 These theories analyzethe cleft pronoun it as a discontinuous constituent with the residue clause, similarto my analysis of ix?P, and both English it and Okanagan ix?P introduce clefts intheir respective languages. I briefly recap their analyses here.Percus (1997) analyzes the cleft pronoun it as the spellout of a definite deter-miner the, which forms a DP constituent with a null NP, which is itself the head ofthe residue clause. A derivation is represented in (81).(81) Percus (1997)a. [The \u000B [that you saw i]][is the deer]. (Base structure)b. [The \u000B ti][is the deer][that you saw i]. (Extraposition)c. It is the deer that you saw. (Spellout of the as it)Hedberg (2000) presents a very similar analysis, except that the definite de-terminer selects for a bare CP, rather than a null NP. The CP lowers to a positionadjoined to the focus, as in (82):(82) Hedberg (2000)a. [The [that you saw i]][is the [deer NP]]. (Base structure)b. [The ti][is the [deer [that you saw i]NP]. (CP Lowering)c. It is the deer that you saw. (Spellout of the as it)34Okanagan does not support ?expletive? theories of clefts (cf. Reeve (2007) for an overview).This because clefting ix?P is not a syntactic subject, it is not inserted, and it is required in contextswhere the residue is introduced by t, as explained above.337Okanagan ix?P is similar to the English clefting pronoun it, under a Percusor Hedberg-style analysis, in the sense that it too forms an underlying constituentwith the residue clause. Unlike in English, however, ix?P is not the spellout ofa determiner, and it is not necessarily overt. Another point of difference betweenEnglish clefts and Okanagan clefts is that in the latter, the residue is already in finalposition, and so no extraposing is required. Instead, it is the demonstrative ix?P thatoptionally moves to initial position.The fact that Okanagan residues can be headed by overt NPs supports Percus(1997) over Hedberg (2000), in the sense that the null NP in (81) may, in princi-ple, have overt content. In actuality, English residues cannot be headed since theclefting pronoun it is a spell-out of the plus a null NP head (Percus, 1997), whicheffectively means that only a sub-class of specificational sentences may be derivedinto clefts. For Okanagan clefts, however, because there is no special spell-out pro-cedure involved, headed residues are perfectly acceptable. The English equivalentof the Okanagan structure using a Percus-style representation of a headless residueis given in (83a), with (83b) representing optional proclisis of the demonstrative.(84) shows the English equivalent of an Okanagan example with a headed residue.(83) English equivalent of Okanagan:?The deer is the one that you saw??It?s the deer that you saw?a. [The deer](=)[(that) the \u000B [that you saw]].b. (That)[the deer](=)[the \u000B [that you saw]].(84) English equivalent of Okanagan:?The deer is the animal that you saw??It?s the deer that is the animal you saw?a. [The deer](=)[(that) the animal [that you saw]].b. (That)[the deer](=)[the animal [that you saw]].One of the major goals of Percus (1997, 338) is to account for the ?specifica-tional character? of clefts. That is, clefts and specificational th-pseudoclefts (e.g.338The one that you saw is the deer) are constrained in their discourse functions in?precisely the same way?. By analyzing clefts as syntactically derived from aspecificational sentence, he is able to explain their common semantic and prag-matic properties. Okanagan does not have specificational sentences in the classicsense, but it does have DP-DP equatives with a fixed information structure resem-bling inverse specificational copular clauses in English, like the following examplefrom den Dikken et al. (2000), except that in English, where the initial DP is alwaysinterpreted exhaustively, it is only implied to be exhaustive in Okanagan.(85) Otto Preminger was who I met.Given that Okanagan clefts are information structurally equivalent to simple DP-DP equatives, and that this is directly parallel to Percus? observation that Englishclefts are pragmatically equivalent to specificational pseudoclefts, it makes senseto argue in favor of an equative analysis of Okanagan clefts.8.6.2 Semantic DerivationThe major benefit of Percus (1997) for English theories of clefts is a semantic one:he is able to link the presupposition of existence and exhaustivity entailment ofEnglish clefts to the semantics of the definite determiner (cf. section 2.4.2). ForOkanagan, the exhaustivity implicature carried by clefts is linked to the maximalityimplicature of the determiner iP (86, cf. section 7.4.1), and clefts carry the sameexhaustivity implicature as DP-DP structures, shown below again as (87).(86) Maximality implicature of iP:f = MAX(87) Exhaustivity Implicature:A sentence of the form [x DP] = [iP Y DP]a. Asserts:? f .x = f (Y )b. Via (86) this implicates:x = MAX(Y )339c. And assuming (87b) is satisfied, a DP-DP structure asserts:x is the only YFor the sake of concreteness, I will show how (86-87) applies to (88).(88) a. (ix?P)DEM[iPDETs ??aPc?n@mDP]deer==iPDETwik-@nt-xw.see-DIR-2SG.ERGDP Focus: It?s the deer you saw.b. Asserts:? f . f (?x[deer(x)(w)?C(x)(w)]) = ?w? f . f (?x[what you saw(x)(w)?C(x)(w)])c. Implies:MAX(?x[deer(x)(w)?C(x)(w)]) = ?wMAX(?x[what you saw(x)(w)?C(x)(w)])d. And assuming (88c) is satisfied, entailsMAX(?x[deer(x)(w)?C(x)(w)]) is the only x equal to?wMAX(?x[what you saw(x)(w)?C(x)(w)])In contexts for which (88c) is not satisfied, the exhaustivity implicature is can-celled, and only the assertion in (88b) holds. This makes data like (89, cf. 16)possible in Okanagan, unlike in English, where exhaustivity is presupposed.(89) JohnJohniPDETn? ?qw-@m-ssteal-MID[?]-3SG.POSSiPDET?t@xtsweetu?CONJnixwalsoPeterPeteriPDETna ?qwsteal(INTR)tOBL?t@xt.sweet#It?s John who stole a cookie, and Peter too stole a cookie.(adapted from (Krifka, 2008, 253))Assertions such as (88b) are also not reliant on any presupposition of existence:the equative in (88b) simply asserts the existence of two individuals, and equatesthem with one another. This means that an Okanagan cleft can be used withoutany prior context (90, cf. 19). In such a context, (90) essentially means ?There is ahorse and I rode that horse?.340(90) a. sti ?mwhatan-c?wt2SG.POSS-doingsQapn?Pnowsx?@lx?Q?lt?dayWhat did you do today?b. ix?PDEMiPDETs@nk? ?caPsq?x?aPhorseiPDETn-k-P?mt-?ws-@m-@n.n-sit-middle-MIN-[DIR]-1SG.ERGI rode that horse.(Literally, ?That?s the horse that I rode? or ?It?s the horse that I rode?.)The semantic role of the Okanagan ?clefting? demonstrative in equative struc-tures is less clear. It is not strictly necessary for an equative structure, and it seemsto be transparent to both maximal and non-maximal readings of its adjoined iP DPin the sense that these DPs permit both deictic and intensional readings (8.5.4).Also, recall that for cases like (91a,b), both spatially deictic and non-spatially de-ictic readings of the demonstrative are available.(91) a. ix?PDEMJohnJohntOBLs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MID(i) It?s John who is the teacher.(ii) John is that teacher.b. ix?PDEMRonRontOBLsyxw?p-m@xShuswap-personaPDETc-qw@l-qw?l-st-@n.CUST-RED-speak-CAUS-1SG.ERGIt?s/That?s Ron who is the Shuswap I was talking to.For now, I assume that the semantic contribution of the Okanagan cleftingdemonstrative ix?P to the residue DP, and to adjoined DPs in general (cf. section4.6.1) is one of deictic features. These deictic features may be spatial, in which casethe deictic force of the demonstrative is obvious, but I suggest that they may alsobe discourse deictics, in which case the deictic force of the demonstrative is muchless apparent, to the point of appearing to be deictically vacuous in some instances.For example, consider that (92b) directly follows (92a) as part of a much longerstory. The demonstrative in (92b) does not denote any of the discourse participantsin (92a), and there is no apparent discourse participant in (92b). The most plausi-ble explanation is that the demonstrative denotes the entire proposition(s) given as341(92a), as a form of discourse deixis.35(92) a. wa ?yalreadywiP-s- ?cx?w-@nt-?s@lx,finish-NOM-instruct-DIR-3PL.ERGwa ?yalready?k@?P@mc?n.agreeThey got done telling him what to do, he agreed.(Colville, A. Mattina (1985, stz.348))b. ix?PDEMu?CONJwa ?yalreadymP? ?n,...noonIt was past noon,... (Colville, A. Mattina (1985, stz.348))Since clefts are normally part of larger discourses, it is viable hypothesis that ap-parent cases of deictically weak demonstratives in Okanagan clefts are in fact dis-course deictics, similar to (92b). In these cases, their function is not to situate thereferent of the residue DP in spatial terms, but rather to situate the referent of theresidue DP in terms of the larger discourse. Given that a demonstrative-associatediP DP in an argument position can be used in a context where the referent is notspatially or temporally present, but may have a discourse antecedent, the hypoth-esis that clefting demonstratives may be discourse deictics does not necessarilyconflict with the proclisis hypothesis. A full description and analysis of discourseuses of Okanagan demonstratives goes beyond the scope of this thesis, however.8.7 SummaryThis chapter has focused on the syntax, semantics, and information structure ofclefts in Okanagan. I have claimed that Okanagan clefts are structurally equiva-lent to DP-DP structures, based on information-structural parallels, including anexhaustivity implicature, an absence of any presupposition of existence (cf. Daviset al. (2004); Koch (2008a)), and a requirement that a DP constituent in contrastivefocus be aligned left (cf. Koch (2008a) for Thompson). Okanagan clefts allowthe residue clause to contain an NP head, which I take as evidence that the residueis categorially a DP, unlike clefts in closely-related Thompson Salish. I discussmorpho-syntactic evidence that the clefting demonstrative ix?P forms an underly-35Examples like (92b) are known in the literature as cases of ?and fronting?. These are discussedat length by Kroeber (1999, 366).342ing constituent with the residue, and moves to the front of the sentence in an op-tional, stylistic proclisis. Given clear parallels with DP-DP structures, I claim thatOkanagan clefts involve a null equative head, linking a residue DP to a focused DP.The implications of this analysis are important for theories of clefts, particu-larly those that analyze the cleft pronoun it as a discontinous constituent with theresidue clause, and trace the semantic and pragmatic properties of clefts to the pres-ence of an underlying definite determiner (Percus, 1997; Hedberg, 2000). I analyzeOkanagan ?clefting? ix?P as an underlying constituent with the residue clause, andlink the exhaustivity implicature and the absence of any presupposition of existenceto the determiner iP.8.8 Chapter Addendum A: Future Clefts and the Casefor Clausal NPsThis addendum presents evidence for my claim that simple NPs may in some casesbe analyzed as projecting clausal structure. This claim is based on cleft data like(93a), which seem to show a DP iP s@xwp?x?@m ?the hunter? in the position of arelative clause, modifying an NP head ylm?xw@m ?chief?, analogous to a clear caseof relative clause modification (93b).(93) a. ix?PDEM= tOBLylm?xw@mchiefiPDETs@xw-p?x?-@m.OCC-hunt-MIDThat?s the chief that is the hunter.b. ix?PDEM==tOBLsyxw?p-m@xShuswap-personaPDETc-qw@l-qw?l-st-@n.CUST-RED-speak-CAUS-1SG.ERGThat?s the Shuswap I was talking to.The question is, what is the structural analysis of (93a)? Given that the modi-ficational pattern OBL-NP-DET-modifier does not exist in non-cleft relativizationcontexts (cf. section 6.5), it may seem at first glance equally likely that (93a) in-volves some marked type of attributive modification, found only in cleft contexts,rather than true relative clause modification.Evidence for clause-projecting NPs comes from future clefts. The future markermi may introduce either verbal (94a) or nominal (94b) cleft residues in future con-343texts, similarly to the determiner iP (95).36 Because future mi is not a determiner,but either a complementizer or a tense head, then assuming that complementizersand tense heads do not select directly for NPs, (94b) provides strong evidence thatNPs may project covert clausal structure.37(94) a. ix?PDEM(t(OBL?q@ ?y-m?n)write-INSTR)miFUTpaP-nt-?s.fold-DIR-3SG.ERGThat?s the paper he is going to fold.b. ix?PDEM(t(OBLt@tw?t)boy)miFUTylm?xw@m.chiefThat?s (the boy) who will be a chief.(95) a. ix?PDEM(t(OBL?q@ ?y-m?n)write-INSTR)iPDETpaP-nt-?s.fold-DIR-3SG.ERGThat?s the paper he folded.b. ix?PDEM(t(OBLsq@ltm?xw)man)iPDETylm?xw@m.chiefThat?s (the man) who is the chief.I conclude that nouns may project covert clausal structure in some cases. The im-plication is that a simple iP DP (96a) may be ambiguously construed as containing36Shuswap clefts can also introduce their residues with a future complementizer meP, as data fromKuipers (1974, 83) shows:(i) y?-@kweit-QUOTyG?yDEMmePFUTmlm?lqwns.IRED-paint-DIR?-3SG.ERG?This is the one he?s going to paint.37Interestingly, while future mi may introduce either a verb or a noun, it is apparently not able tointroduce an adjective.(i) *ix?PDEMax?PDEMmiOBLcax?.redThis is the one that will be red.In cleft contexts, the distribution of the morphemes t and mi provide us with a diagnostic for identi-fying the lexical category of the constituents they precede, since t may only introduce a noun, not averb or an adjective (cf. section 8.5.2).344covert clausal structure (96b) in modifier positions.38(96) a. Simple iP DP:[iP [s@xwp?x?@mNP]DP]b. Headless relative clause iP DP:[iP \u000B1 [[(iP) \u000BDP]2 s@xwp?x?@m t2 CP]DP]Headed future cleft residues pattern like other relative clauses in allowing bothpost-nominal (97a) and pre-posed (97b) ordering.(97) a. ax?PDEMtOBLy?mx?waPcedar.bark.basketmiFUT?kw??l-@n,make-[DIR]-1SG.ERGlutNEGyaPx??s.DEMThis is the basket I?m gonna make, not that one.b. ax?PDEMmiFUT?kw??l-@nmake-[DIR]-1SG.ERGtOBLy?mx?waP,cedar.bark.basketlutNEGyaPx??s.DEMThis is the basket I?m gonna make, not that one.As such, one expects evidence for movement in future clefts (cf. section 6.3.3),but this is not the case. Specifically, since y?mx?waP ?cedar bark basket? is anunderlying core object of the relative clause predicate mi ?kw? ?l@n ?I will make x?, theprediction is that an iP determiner should surface before mi, assuming that mi is ineither C or T position. The fact that iP does not surface strongly suggests that thereis a filter deleting the determiner in this environment, similar to the ?doubly-filledcomplementizer? effect in English (Henry Davis, p.c.).The future clefts discussed so far in this section are in principle amenable to theequative analysis. In contexts for which an adjunct such as a PP is clefted however(98a), future clefts are not staightforwardly amenable to an equative analysis, sincePPs are not referential expressions. This brings us to a more general problem,concerning adjunct clefts. Future mi alternates with the complementizer kiP innon-future adjunct clefts (98b), which I now discuss in further detail.38An iP DP in a modifier position may also be ambiguously analyzed as bare CP, e.g. [[iP \u000BDP]2s@xwp?x?@m t2 CP], though since I have already shown that cleft residues are categorially DPs inOkanagan, I do not further discuss this possibility.345(98) a. ?klLOCs@n- ?kQaw-m@nLOC-pray-INSTRmiFUTkwu1PL.ABSP??lus.gather.It?s at the church that we will gather.b. ?klLOCs@n- ?kQaw-m@nLOC-pray-INSTRkiPCOMPkwu1PL.ABSP??lus.gather.It?s at the church that we gathered.8.9 Chapter Addendum B: Adjunct CleftsI now move to a discussion of a different type of Okanagan cleft, which I referto as adjunct clefts. Adjunct clefts cannot be analyzed as DP equatives, sinceneither the focused constituent nor the residue clause can be analyzed as DPs. Theconstituent in focus position is an underlying adjunct of the residue clause, usuallya PP or other locative or temporal adverbial. The residue clause is introduced bythe complementizer kiP, rather than by an iP determiner.39In (99a) below, a PP iP ?kl nkw ?r?tkw ?at Glimpse Lake? occupies the focus po-sition, and a CP kiP kn ksp?lxaPx ?that/where I?ll be camping? occurs in residueposition. In (99b) an adverbial demonstrative i ?kliP ?over there? is in focus position.Simple demonstratives can occur in the focus position of an adjunct cleft if theydenote passive agents (99c), for example.39I have at least one example from Hebert?s recorded Upper Nicola corpus where a temporaladjunct occurs in focus position, while the residue clause is introduced by iP rather than kiP:(i) ix?PDETlLOC?kwu ?kwn?s ?q@tfew-daysiPDETxa?rnt?nsoak-DIR-1SG.ERGiPDETs ??aPcnm?? ?caP.deer-skinIt?s for a few days that I soaked the deer hide.These are judged ungrammatical when tested in an elicitation context, however.346(99) a. ix?PDEMiPDET?klLOCn-kw ?r-?tkwn-yellow-waterkiPFOCkn1SG.ABSks-p?lx-aPx.FUT-camp-INCEPTIt?s at Glimpse Lake that I?ll be camping.b. i ?kl?PDEMkiPFOC?k?- ?klaxw.LOC-eveningIt?s over there that he went out of sight.c. ix?PDEMkiPFOCw?k@nt@m.see-DIR-PASSHe?s the one that was seen.Adjunct clefts do not allow the pre-posed residue clauses (100), unlike other clefts,which suggests that the residues are not relative clauses, but bare CPs.40(100) *ix?PDEMkiPCOMPkn1SG.ABSks-p?lx-aPxFUT-camp-INCEPTiPDET?klLOCn-kw ?r-itkw.n-yellow-waterIt?s up at Glimpse Lake that I?ll be camping.As with other Okanagan clefts, adjunct clefts lack any presupposition of existence.To illustrate, consider that the adjunct cleft in (102) below is good as an answer toeither question in (101). As an answer to either question, the focused constituent in(102) ?kl nkw ?ritkw ?at Glimpse Lake? is not given or presupposed in any sense. Theresidue clause kn ksp?lxaPx ?I am going to camp? is given when it is included in ananswer to (101a), but it is not presupposed as an answer to either (101a) or(101b).(101) a.haYNQkw2SG.ABSPaws-p?lx-aPxgo-camp-INCEPTQ?pnaPnows ?klaxw?eveningAre you gonna go camping this evening?b.sti ?mwhatan-c?wt2SG.POSS-doingsQapn?Pnows ?klaxw?eveningWhat did you do today?40Also, similarly to Thompson Koch (2008a), locative adjunct cleft residues are not introduced byprepositions, as might be expected if they were formed via the same processes as locative relativeclauses. See section 9.3.2.347(102) ?klLOCn-kw ?r-itkwn-yellow-waterkiPCOMPkn1SG.ABSks-p?lx-aPx.FUT-camp-INCEPTI?m camping at Glimpse lake.(Literally, It?s at Glimpse Lake that I?m camping.)It is unclear whether adjunct clefts carry an implicature of exhaustivity, similar toother clefts in Okanagan. Data like (103a) are consistent with an analysis wherebyadjunct clefts have no exhaustivity effect, and I have so far been unable to deter-mine that non-cleft data involving locative adjuncts (e.g. 103b) are infelicitous incontexts for which adjunct clefts are not. If it turns out that adjunct clefts lack anyexhaustivity effect, this could be due to the fact that there is no iP determiner tocontribute an exhaustivity implicature.41(103) a. ?klLOCs@n- ?kQaw-mnLOC-pray-INSTRkiPCOMPkwu1PL.ABSyaQp,arrive(PL)u?CONJ(nixw)(also)?klLOCsn- ?maP- ?m?yaP-t@n.LOC-RED-teach-INSTRWe got to the church, and then we went to the classroom.b.kn1SG.ABSckicxarrivetlLOCs@n- ?kQaw-m@n.LOC-pray-INSTRI came from church.Though it may be possible to analyze adjunct clefts as equatives, how exactly thisanalysis might be spelled out is unclear, and something I leave for future work.41While it may be possible to deconstruct kiP into a sequence k ?complementizer? plus iP deter-miner, this is speculative.348Chapter 9ConclusionThis closing chapter first summarizes the main findings of this dissertation (section9.1), then discusses implications of my analysis for Okanagan grammar and futurework in the language (section 9.2). I then conduct a survey of DP-DP structuresin clefts across a handful of Salish languages and discuss whether data in theselanguages are amenable to an equative analysis (section 9.3). I end this chapter,and this dissertation, with some implications of my analysis for theories of non-verbal predication (9.4).9.1 Summary of FindingsThis dissertation has argued for a syntactic, semantic, and information structuraldistinction between direct predications in Okanagan, which involve a lexical pred-icate and a DP argument with no copula or other functional intermediary; and DP-DP structures, which are projections of a null equative head (Heycock and Kroch,1999; Romero, 2005). Okanagan presents an interesting case study for testing the-ories of copular predication: since there is no overt copula (N. Mattina 1996a),many of the standard diagnostics in the literature for identifying a particular cop-ular sentence type are inapplicable. Nevertheless, I use independent evidence forsyntactic constituency and semantic type in tandem with word order constraints toreach useful generalizations about how predication and equation is manifested inOkanagan.349English small clauses and copular clauses are canonically subject-initial (Moro,1997), and the range of predicates which can occur before their subjects is nor-mally restricted to DP predicates. For Okanagan, subject-predicate word order isremarkably free except in the case of predications involving two DPs. The patternof ?inverse? predication in Okanagan thus appears to be opposite to that found forEnglish. The full patterns are given below for both languages (cf. Table 9.7 for theOkanagan data).Table 9.1: ?Canonical-Order? Predication in Okanagan and EnglishOkanagan English ?Canonical?X X John is a doctorX X Mary is bossX X John is proud of his daughtersX X John is the best candidate for the job.Table 9.2: ?Inverse-Order? Predication in Okanagan and EnglishOkanagan English ?Inverse?X * A doctor is JohnX * Boss is MaryX * Proud of his daughters is John* X The best candidate for the job is John.The ?inverse? Okanagan pattern, as displayed in Table 9.2, suggests that Hig-gins? specificational type copular sentences, where a final referential DP is in focusand interpreted exhaustively (Percus, 1997), are not possible. This is not specifi-cally due to a ban on syntactic inversion for DPs, since none of the inverse patternsin Okanagan are derived by syntactic inversion (Mikkelsen, 2005; den Dikken,2006) (sections 7.2.2 and 7.3.2). What distinguishes DPs in Okanagan from lex-ical categories such as NP is that they are a referential type Longobardi (1994);350Matthewson (1998), and cannot type shift. I have claimed that the word-order re-quirement in Okanagan DP-DP structures derives from the fact that they are struc-turally distinct, projections of a null equative head (Heycock and Kroch, 1999).Explaining the absence of inverse DP-DP structures in Okanagan raises someinteresting questions regarding the nature of equation in Okanagan. The fixed in-formation structure cannot be explained on purely pragmatic grounds, since the un-grammaticality of the inverse ordering is not related to any old information require-ment on specificational subjects (Heycock and Kroch, 1999; Mikkelsen, 2005), butrather to the semantic type of its DP complement. Okanagan DP-DP structures al-ways involve a directly referential DP (a proper name or demonstrative) or a refer-ential iP DP, and a non-directly referential DP headed by the determiner iP (section7.5). iP DPs display independent evidence for allowing intensional readings, suchas individual concept readings (En?, 1981; Demirdache, 1996) and generic read-ings (Chierchia, 1998). I have claimed that the Okanagan equative head encodes asemantic asymmetry, and maps the intension of an individual to its extension (type<,>) (Romero, 2005; Comorovski, 2007). The semantics, by itself, doesnot derive the word order restriction of DP-DP structures, however. Informationstructure also plays a role.The determiner iP carries a maximality implicature (section 5.3.3). In an equa-tive context, the sentence as a whole will carry an exhaustivity implicature, suchthat the referent of the directly-referential DP will be interpreted as the only in-dividual equivalent to the referent of the intensional iP DP (section 7.4.1). Byassuming that the directly-referential DP is also assigned a syntactic feature ?F?(interpretable as focus) (Jackendoff, 1972) by the equative head, and that focus isaligned to the left edge of an intonational phrase (Koch, 2008a), the word orderrestriction of DP-DP structures, and the absence of specificational sentences fromthe language, is accounted for. Analyzing Okanagan DP-DP structures as havinga fixed information structure via a structural focus position receives addition sup-port from evidence that they are information-structurally and morpho-syntacticallyequivalent to Okanagan clefts.Since maximality is only an implicature for Okanagan iP DPs, and the exhaus-tivity effect in equatives stems from the determiner?s maximality implicature, theprediction is that exhaustivity too will be an implicature and not an entailment in351Okanagan. This means that although DP-DP structures do have readings that aredirectly analogous to inverse specificationals in English, where the initial DP isinterpreted exhaustively, this is not necessarily the case: in contexts for which thedeterminer?s maximality implicature is not satisfied, ?pseudo-predicational? read-ings of DP-DP structures are possible. This is most clearly demonstrated by thefact that in a non-maximal context, a DP-DP structure will be pragmatically equiv-alent to a direct predication. This is predicted by the domain restriction semanticswhich I motivate for iP in chapter 5.3.3.In section 7.6 I discuss predications involving either a demonstrative and aproper name, a demonstrative-associated DP and a proper name, or two propernames. These might in principle be analyzable as equatives, but I argue thatthey are not, based primarily on the fact that demonstratives are never predicates,proper names can be predicates under certain circumstances, and that a predicativeproper name can precede a subject demonstrative, demonstrative-associated DP, orproper name for these cases, similarly to other types of direct predication. Descrip-tively, these sentences fit Higgins? identificational class (at least the examples withdemonstratives), but in actuality, I argue that they are a form of direct predication,with the caveat that examples involving simple demonstratives and proper namesmay in some cases analyzable as truncated clefts, and hence equative. This analysiseffectively means that while proper names are not predicates in equative structures,due to the selectional restrictions of the equative head, they may be predicates ina direct predication. Assuming that identificational sentences may be reduced to aform of direct predication (Heller, 2005), Higgins? taxonomy may be reduced toonly two types for Okanagan: equative and predicational (cf. Table 7.3).There are apparent cases of ?inverse? DP-DP structures, however. In contextsfor which a DP-DP structure does not answer a WH-question (e.g. in an out-of-the-blue context, or within a monologue), an iP DP may sometimes precede adirectly referential DP. I suggest that these may be instances of either predicatetopicalization or left dislocation, rather than true specificationals. If a fronted iPDP is outside of the intonational phrase containing a focused subject, then focusalignment constraints still hold, although the acoustic work remains to be done.DP-DP structures are information-structurally and morpho-syntactically equiv-alent to clefts in Okanagan, the only difference between the two being that for a352cleft, the second DP consists of either a headed or headless relative clause (cf.chapter 6). Information structurally, clefts include (i) an exhaustivity implicaturerather than a presupposition or entailment (Davis et al., 2004); (ii) the absence ofany presupposition of existence (Davis et al., 2004); and (iii) a ban on focus oc-curring finally. In tandem with morpho-syntactic evidence, I claim that Okanaganclefts are also equative structures.An equative analysis of clefts is novel in the Salish literature, though Shank(2003) and Koch (2008a) consider this as a possibility for Northern Straits andThompson, respectively, before settling on a predicational analysis. An equativeanalysis is also not entirely without motivation in the theoretical literature. Percus(1997), for example, derives clefts from specificational pseudoclefts, which in turnHeycock and Kroch (1999) argue to be equative. I show that Okanagan supportstheories of English clefts such as Percus (1997) and Hedberg (2000) since (i) theOkanagan ?clefting? demonstrative ix?P was shown to be an underlying constituentwith the residue DP, and (ii) the semantics/pragmatics of Okanagan clefts may betraced to the semantics/pragmatics of the determiner iP. Since Okanagan clefts arederived from the equivalent of inverse specificational sentences, where the residueclause is generated in final position, extraposing is unnecessary.9.2 Implications for Okanagan GrammarThis section summarizes the major empirical contributions of this thesis beforediscussing further questions. These further questions may be thought of as settinga research agenda for future work in the area.9.2.1 Empirical ContributionsThis dissertation has made several important empirical contributions to our under-standing of Okanagan syntax and semantics. I will briefly discuss the more majorcontributions in the following areas:a. Determiner Semantics (chapter 5)b. Relative Clauses (chapter 6)c. Predication and Equation (chapter 7)353d. Clefts (chapter 8)Determiner SemanticsChapter 5 consists of a detailed investigation of the semantic properties of theOkanagan determiner iP and the oblique marker t in argument contexts. It wasshown that Okanagan shares many of the same semantic properties as Lillooetassertion-of-existence determiners (Matthewson, 1998, 1999) and Squamish deic-tic determiners (Gillon, 2006, 2009a), but crucially permits narrow scope readingsand intensional readings. Other than N. Mattina (2006) which investigates deter-miners in Moses-Columbian, this chapter represents the only thorough account ofdeterminer semantics in a Southern Interior Salish language. My analysis pro-vides crucial points of comparison with determiners in Northern Interior Salish,supports N. Mattina?s analysis of Moses-Columbian determiners with comparativeevidence, and provides a basis for understanding the semantics of Okanagan equa-tive DP-DP structures and clefts.Relative ClausesChapter 6 discusses nominal modification in Okanagan, and the difference betweenattributive modification and relative clauses. I show that Okanagan relative clausesprovide evidence for clause-internal movement of a DP to the left-periphery of CP,a pattern that has been established for the Northern Interior languages of Lillooet(Davis, 2004, 2010a) and Thompson (Kroeber, 1997; Koch, 2006). Like otherSalish languages, Okanagan allows both headed and headless relative clauses, aswell as both head-initial and head-final variants.Predication and EquationChapter 7 represents the first detailed investigation of predication and equation in aSalish language, and elucidates previous observations made by N. Mattina (1996b)and A. Mattina (2000), supporting their descriptive statements concerning the ex-istence of equative structures. They claim that there ?is no copula? in Okanaganequatives (Mattina, 1996b, 30). While this is true in the sense that there is no verb354?to be?, I have claimed that there must be a null functional head in equative struc-tures, since neither of the two DP constituents can be a predicate. The semanticsof the equative head fit with intensionality-based accounts of English equatives(Romero, 2005; Comorovski, 2007), and the Okanagan data support the possibilityof reducing Higgins? taxonomy to two types, predicational and equative, for at leastsome languages. The Okanagan equative data additionally support Koch?s (2008a)analysis of Thompson River Salish, where a focused constituent must align to theleft-edge of an intonation phrase; however, focus-alignment facts are less clear fordirect predications.CleftsChapter 8 shows that Okanagan clefts are information-structurally equivalent tosimpler DP-DP structures, supporting an analysis whereby both types are syntac-tically equivalent. The semantic and pragmatic properties of Okanagan clefts sup-port previous investigations of clefts in Thompson (Koch, 2008a, 2009; Koch andZimmermann, 2009), as well as Lillooet and Northern Straits (Davis et al., 2004),though they warrant a different syntactic analysis. Residues in Okanagan clefts areDPs, rather than bare CPs, as evidenced by the fact that the residues may containan overt NP head. This difference raises some interesting typological and historicalquestions, some of which I seek to address later in section 9.3.9.2.2 Further QuestionsThere are many unanswered questions, and much further work to be done forOkanagan in the areas of predication, focus, and clefts, as well as more gener-ally in the areas of clause-level syntax and semantics. I seek to address a few ofthese here in this section.Acoustic Evidence for FocusPerhaps the most pressing unanswered questions stem from the lack of acousticdata for Okanagan DP-DP structures and clefts. Barthmaier (2004) establishes theexistence of intonational phrases in Okanagan by acoustically analyzing severalnarrative passages from A. Mattina and DeSautel (2002); however no further work355has been done in this area. Information-structurally, it can be established that focusaligns to the left in Okanagan equatives; however a detailed investigation of theacoustic correlates of focus similar to that of Koch (2008a) for Thompson RiverSalish remains to be carried out. I have assumed, based on my own impressions,that there is no acoustic correlate to focus in Okanagan, similar to Thompson.A detailed acoustic analysis may potentially shed light on unexplained cases ofinversion (section 7.7), and confirm whether these may or may not be cases ofpredicate topicalization, for example.As a related issue, a focused constituent in an Okanagan direct predication doesnot appear to be strictly subject to a FOCUS-LEFT constraint (Koch, 2008a), unlikethe case for the focused DP in an equative structure. This is indicated by the avail-ability of flexible subject-predicate word ordering for direct predications in the con-text of a WH-question. While direct predications involving headless relative clausearguments (i.e. NPCs) seem to show a stronger preference for focus-predicate ini-tial ordering, the reasons for this variability, as well as the more general differencebetween focus alignment in direct predications versus equatives, remain unclear.Koch (2008a) found that fronted subjects in Thompson NPCs constitute a separateintonation phrase, and that as such, a focused predicate in final position may still beleft-most within its intonation phrase. I suggested that unexplained cases of inverseDP-DP structures might be explained similarly (section 7.7), though the acousticwork has yet to be undertaken.Semantics of DemonstrativesAnother area in need of further research concerns the semantics of demonstratives(cf. cursory remarks in section 8.6.2). In this thesis, I have assumed that simpledemonstratives are directly referential (Kaplan, 1977, 1989) argument expressionsbased on evidence that their distributions are equivalent to other DPs in argumentcontexts, and that they cannot function as syntactic predicates. Additionally, basedon distributional and pragmatic evidence I have made the blanket assumption thatdemonstratives adjoined to iP DPs and demonstratives adjoined to t NPs are bothcategorially DPs (cf. section 4.6.1, and Matthewson and Davis (1995) for Lil-looet). It is unclear however what the semantic contribution of demonstratives is in356contexts in which they adjoin to a DP, especially given that they appear to be trans-parent to intensional readings. It is also unclear what the internal syntax of thesestructures is. A successful answer to these questions may potentially have majorramifications for my analysis of Okanagan clefts. While I believe there is good ev-idence that demonstratives cannot be simple property denoting expressions, it maybe worthwhile investigating the possibility that demonstratives may be higher typepredicates (King, 2001). A related problem concerns so-called ?discourse? uses ofdemonstratives, and how these compare with demonstratives in argument contexts(Lyon (2010b), cf. section 8.6.2).Discourse Constraints on InversionIt is an interesting fact that WH-questions in Okanagan do not license specifica-tional sentences, even in cases where an initial iP DP is discourse old (Birner,1996; Mikkelsen, 2005), and no less interesting that sentences resembling specifi-cationals are possible in out-of-the-blue and ongoing topic contexts (section 7.7).Explaining this pattern from the larger perspective of whether there may or may notbe universal constraints on DP ?inversion? remains a major area of further study.While it seems clear that being discourse-old is not a universally sufficient condi-tion for a DP being able to invert around a subject (Mikkelsen, 2005), for Okanaganat least, it seems likely that ?focus? will ultimately be the more relevant factor inexplaining this pattern.Question/Answer CongruenceIn sections 7.4.1 and 8.3.2 it was shown that in question/answer contexts for whichan exhaustive answer is required, a DP-DP structure or cleft must be used, ratherthan a direct predication. I suggested that this may be due to a conversationalimplicature arising from the use of iP in answer contexts: since direct predicationsdo not imply exhaustivity, it is more informative for a speaker to use a DP-DPstructure or cleft in a context for which an exhaustive reading is required. This isreminiscent of the fact that in English, a speaker cannot felicitously use a markedrising-intonation pattern in a context for which an exhaustive answer is required,and the Okanagan pattern makes sense assuming that intonation does not signal357exhaustivity in the language. This needs to be confirmed by acoustic evidence,however.Focus and Focus-Sensitive OperatorsThe syntactic correlates of focus remain poorly understood in Okanagan, and littlework has been done in terms of investigating the role which focus-sensitive op-erators like kmix/kmax ?only? play in Okanagan. Other related topics in need offurther work include second-occurrence focus and focus projection.Nominal ModificationWhile the groundwork has been laid for a thorough analysis of nominal modifi-cation in Okanagan (chapter 6), there remain many unanswered questions. Forexample, head-final pattern 3, as discussed in Table 6.4 where the oblique markerprecedes both the head and the modifier, is consistent with both attributive andrelative clause modification. The question arises as to whether it is possible todisambiguate this pattern. Regarding head-initial pattern 3, which I suggest is in-dicative of relative clause modification, is there a way of establishing that thesecases involve clause-internal movement? As another issue for further research,Davis (2011) discusses post-nominal attributive modification in Lillooet, and it re-mains unclear whether this is possible in Okanagan. Perhaps most pressing is thequestion of why patterns of relative clause modification differ for cleft versus non-cleft contexts. For example, head-initial pattern 6 is ungrammatical in non-cleftcontexts, but grammatical for a cleft residue. While there is evidence for clause-internal movement of a DP in cleft residues, in keeping with relative clauses inother syntactic contexts, it is less forthcoming for clefts.9.3 Implications for SalishThe existence of DP-DP structures in Okanagan leads me to ask the followingquestions: do any other Salish languages have DP-DP stuctures? If so, are theysimilar to or the same as clefts in these languages? Is it possible for the equativeanalysis to be extended to DP-DP structures and clefts in other Salish languages?The answers to all these questions appear to be ?yes?. This section investigates data358from other Salish languages, the overall goal being to provide future researchersinterested in predication and equation across Salish with a data base and a set ofgeneralizations from which to begin.Table 9.3 below is intended to provide a key to understanding different struc-tures across Salish in terms of how strongly a given structure supports the equativehypothesis. The strongest evidence for equatives come from DP-DP structures withno overt copula, and where each DP is headed by a non-vacuous determiner thatselects for simple NPs (line 1). Cases where one of the constituents is a demonstra-tive or proper name provide less strong evidence, since these may be predicative insome Salish languages (line 2). Variants of lines 1-2 which include an overt copulaor non-constituent demonstrative (lines 3-4) constitute still weaker evidence, sincethese elements may also be used in clefts, which are not necessarily equative. Vari-ants of lines 1-2 which include a demonstrative (D) or complementizer (C) headedclause as one of the constituents (lines 5-6) constitute even weaker evidence, sincedepending on the language, the status of these constituents as DPs may be suspect.Finally, lines 7-8 represent canonical cases of clefting in Salish, and represent theweakest evidence for equatives.1 The analysis of clefts as equatives is contingenton both focus and residue being categorially DPs. A no less valuable perspectivemay be gleaned from turning Table 9.3 on its head; namely, whether it may be thecase that simple DP-DP structures in a particular Salish language may be analyzedas clefts.1I do not represent the possibility of null determiners in Table 9.3, although these are common inSouthern Interior Salish. Unless there is strong independent motivation for assigning a DP structurein these cases, structures involving null determiners do not offer particularly strong evidence forequation.359Table 9.3: DP-DP Structures as Evidence for Equation across Salish: Ar-ranged by Constituency Type from Strongest to Weakest EvidenceOvert Copula orDemonstrativeFirst DP?focus?Second DP?residue?1 * [D NP] [D NP]2 * [DEM / P.N.] [D NP]* [D NP] [DEM / P.N.]3 X [D NP] [D NP]4 X [DEM / P.N.] [D NP]X [D NP] [DEM / P.N.]5 * [D NP] [D/C clause]* [D/C clause] [D NP]6 * [DEM / P.N.] [D/C clause]* [D/C clause] [DEM / P.N.]7 X [D NP] [D/C clause]X [D/C clause] [D NP]8 X [DEM / P.N.] [D/C clause]X [D/C clause] [DEM / P.N.]What follows are fragmentary data sets from other Salish languages, exhibitingpatterns shown in Table 9.3. How well each different language supports the equa-tive hypothesis depends on whether data corresponding to lines 1-4 are possible ina particular language, and whether cleft residues may be safely analyzed as DPs.Some languages will rank more highly with regards to Table 9.3 than other lan-guages, as we shall see. I should reiterate that this is only a preliminary survey, andthat I make no definitive claims with regards to any other Salish languages besidesOkanagan. The data from each language investigated here merit further study.One limiting factor of this survey is the absence of negative data from the cor-pus for languages whose primary sources are texts or text-based grammars. Forexample, in most languages I have been unable to verify the extent of subject-predicate word order flexibility, or the presence or absence of specificational sen-360tences. Another major limiting factor for Southern Interior Salish is the fact thatdeterminers are commonly null in argument contexts, which means that distin-guishing an NP predicate from a DP is not easy.First, I survey data in other Southern Interior Salish languages (9.3.1): Kalispel(Montana Salish), Coeur d?Alene, and Moses-Columbian. Next, I move on toNorthern Interior Salish languages (9.3.2): Shuswap, Thompson, and Lillooet.Then, I briefly discuss data from Northern Straits and Bella Coola (9.3.3), beforesummarizing the section (9.3.4). There are data from other Salish languages whichare not discussed here. See Kroeber (1999, ch. 7) for an overview.9.3.1 Implications for Southern Interior SalishVarious complications arise when trying to establish the existence of DP-DP struc-tures in the Southern Interior. These include:a. the fact that argument introducing determiners are often optionalb. some question as to the distribution of determiners versus complementizersin the Southern Interiorc. establishing that the second DP in a putative DP-DP structure can containan overt NP headI summarize the data for Southern Interior Salish, and how the data inform us withregards to the existence of equative structures, at the end of this subsection.Kalispel (Montana Salish)Kalispel, like Okanagan, seems to permit both subject-initial (1) and subject-final(2) direct predications. Examples (1a, 2a) may be analyzable as a Higgins?-typeidentificational sentence, which I have characterized as a type of direct predicationin section 7.6.1. Also similarly to Okanagan, subject-initial predications do notnecessarily involve topicalization (Nico Baier, p.c. 2013).(1) a. i?eDEM.DISTpus.cat?That is a cat?. (Kalispel, Tachini Pete p.c. 2011)361b. i?eDEM.DISTayx?t.tired?He gets tired?. (Kalispel, Tachini Pete p.c. 2011)c. c?onJohnilm?xwm.chief?John is the chief? (Kalispel, Tachini Pete p.c. 2011)d. ?uDETsmPemwomansxwmimeyePm.teacherThe woman is a teacher. (Kalispel, Tachini Pete p.c. 2011)(2) a. puscat?iheP.DEM.PROX?This is a cat?. (Kalispel, Tachini Pete p.c. 2011)b. smPemwoman?uDETsxwmimeyePm.teacherThe teacher is a woman. (Kalispel, Tachini Pete p.c. 2011)Kalispel also exhibits what seem to be DP-DP structures (3), with an NP be-ing introduced by the determiner ?u(P). (3e) is particularly compelling evidence forequative structures in the language (cf. line 1 in Table 9.3). Furthermore, the in-definite translation of the second DP in (3e) indicates that like Okanagan, KalispelDP-DP structures permit pseudo-predicational readings.2(3) a. i?eDEM.DIST?uDETpus.cat?That is a cat?. (Kalispel, Tachini Pete p.c. 2011)b. ?e ?yDEM.DIST?uDETpus.cat?That is a cat?. (What I refer to is a cat.) (Kalispel, Tachini Pete p.c. 2011)2The difference between ?ey (e.g. 3b) and i?e (e.g. 3a) is unclear. The former i?e(P) is glossedas ?DEM; deictic? in Camp (2007, 109), while ?ey, or ??Pi, is glossed as ?that; ?ePi, ??P, ?eP inCamp (2007, 116). Based on translations, it seems that both might function as argument expressions,however there are distributional differences between the two, such that i?e(P) cannot occur finally(cf. 5-6).362c. i?eDEM.DIST?uDETilm?xwm.chief?That specific person is a/the chief.? (Kalispel, Tachini Pete p.c. 2011)d. c?onJohn?uDETilm?xwm.chief?John is the chief? (Kalispel, Tachini Pete p.c. 2011)e. ?uDETsmPemwoman?uDETsxwmimeyePm.teacherThe woman is a teacher. (Kalispel, Tachini Pete p.c. 2011)A complicating factor when studying DP-DP structures in Kalispel comes fromthe fact that bare-nominal arguments are allowed (this is a more general analyticalissue for Southern Interior Salish, outside of Okanagan). In (4a), the object argu-ment sq?lixw is not introduced by any determiner, unlike the case for Upper NicolaOkanagan. (4b) either shows that demonstratives can occur in D-position, or elsethat a demonstrative can adjoin to a null-headed DP.(4) a. hoythenc?n1SG.ABSn-Pu?xwn-go.inuandw?c?-(n)t-nsee-DIR-1SG.ERGsq?lixwpeopleiPARTt?P.crowdI went in and saw crowds of people. (Kalispel, Camp (2007, p.19, text III)b. hoythenxwistwalk?iP?DEM.PROX?tt ?wit.young.boyThen the young man walked away. (Kalispel, Camp (2007, p.28, text IV)Despite data showing that subject-final predications are possible (2), it seemsclear that the Kalispel demonstrative i?e cannot occur after the predicate. This is asexpected for DP-DP structures (5) under the equative analysis I have proposed forOkanagan, but also seems to hold for direct predications (6), which is unexpected,assuming that the Kalispel demonstrative i?e is cognate with Okanagan ix?P,3 andassuming that it is able to function as an argument DP. Sarah Thomason (p.c. 2013)notes that i?e may function as a discourse particle, though in this case its distribu-3In Kalispel and Coeur d?Alene, proto-Salish plain velar obstruents /k/ and /x/ are palatalized to/c?/ and /?/, respectively (Kuipers, 2002, 3).363tion is even more unexpected, given that discourse uses of Okanagan ix?P tend tooccur in initial position (cf. Lyon (2010b)).(5) *?uDETpuscati?e.DEM.DIST?That is the cat?. (Kalispel, Tachini Pete p.c. 2011)(6) a. *puscati?e.DEM.DISTThat is a cat. (Kalispel, Tachini Pete p.c. 2011)b. *ayx?ttiredi?e.DEM.DISTHe gets tired. (Kalispel, Tachini Pete p.c. 2011)c. *ilm?xwmchiefi?e.DEM.DISTThis is a chief. (Kalispel, Tachini Pete p.c. 2011)There are also unexplained Kalispel data which seem to show a demonstrative ?e ?yas an argument of a determiner (7a,b), which may be evidence that some demon-stratives are underlyingly predicative in Kalispel. The same is not true of i?e (7c),which given that it is cognate with Okanagan ix?P is unsurprising. The data in(7a,b) are consistent with identificational interpretations, where the initial con-stituent is an NP predicate, and not a DP.(7) a. puscat?uDET?e ?y.DEM.DIST?That is a cat?. (Kalispel, Tachini Pete p.c. 2011)b. c?onJohn?uDET?e ?y.DEM.DIST?That is John?. (Kalispel, Tachini Pete p.c. 2011)c. *ilm?xwmchief?uDETi?e.DEM.DIST?This is a chief?. (Kalispel, Tachini Pete p.c. 2011)364There are clearer cases of identificational uses of DP-DP structures involving thedeterminer ?u(P)in Kalispel (8a), given that ?u(P) can precede a proper name inan introductory context. This is directly parallel to Okanagan identificationals (cf.section 7.6.1), morphologically speaking (8b) since both languages utilize a variantof the particle ?. In Upper Nicola Okangan, however ?@ and ?aP are complemen-tizers and not determiners. This does not necessarily mean that Kalispel ?u(P) maynot function as both a complementizer and a determiner (cf. Thompson; Koch(2008a, 2009)). If ?u(P) is functioning as a complementizer in (8a), this poten-tially allows us to analyze the demonstratives in (7a,b) as referential expressionsembedded within a CP, rather than predicative.4(8) a. ?e ?yDEM.DIST?uDETc?on.John?That is John?. (Kalispel, Tachini Pete p.c. 2011)b. ix?PDEM.DIST?@COMPSpike.JohnThat is Spike. (Okanagan)Nevertheless, Camp (2007) glosses ?u(P) as an ?article?, and I take data like (3e)to be strong evidence that it is a determiner in at least some cases. Furthermore,from a cursory examination of the corpus, it has a distribution nearly identical toOkanagan iP. For example, it introduces passive agents:(9) kwu1SG.ABSsew-nt-mask-DIR-PASS?uPDETtOBLhin- ?px?w ?px?w?t.1SG.POSS-parentsMy parents asked me. (Kalispel, Camp (2007, p.24, text III)Textual data (Camp, 2007) show examples of DP-DP structures and clefts.5(10a,b) may be analyzed as equative under the assumption that the initial demon-strative is referential, while (10c) shows a clear example of a headless relative4Notice also that (8a) is essentially the reverse of (7b). This may suggest that ?u is being reana-lyzed as a copula, rather than a determiner. See Gillon (2006) who assigns the non-deictic determinerkwi the semantics of a predicational copula, although it does not have the distribution of a copula.The point is that a semantically weak (or vacuous) determiner may be reinterpreted as a copula, alinker of sorts, rather than being inherently associated with a following nominal.5??yu is glossed as a DEM (Camp, 2007, 116).365clause being introduced by a determiner.(10) a. ??yuDEM.DIST?uPDEThin-sqltmixw.1SG.POSS-man...That is my husband. (Kalispel, Camp (2007, 37))b. ??PiDEM.PROX?uPDETqwuyeP?1SG.INDEPhin- ?kw ?n-c?st-m?st-tn1SG.POSS-try-hand-INTR.REFLEX-INSTR?This is my fixing?. (Kalispel, Camp (2007, 53))c. ??PiDEM.PROX?uPDETqeP-e?t ?cl?lqw-i.FUT-play.stick.game.again-FUTThis is the one who now is going to play. (Kalispel, Camp (2007, 78))Kalispel clefts may be amenable to the equative analysis. In (11), an initial demon-strative may be analyzed as adjoined to a null-headed locative relative clause infocus position. The residue clause is an NP inflected with irrealis morphology (cf.section 5.2.9).(11) ??PiDEM.DIST?uPDETiLOCkw?l-lqsred-shirt==?uPDETi-qs-mP?m.1SG.POSS-U.POSS-wifeThat one in the red shirt will be my wife.(Kalispel, Kroeber (1995, II, 19))Like Okanagan, initial demonstratives are most likely optional in clefts and DP-DP structures, as indicated by data like (12a). This example also shows that cleftswith residues introduced by a future complementizer are possible in Kalispel, as inOkanagan (8.8) and Shuswap (9.3.2), where a doubly-filled complementizer filterprevents the future marker from co-occuring with a clause-introducing determiner.Though translations constitute weak evidence, the translation of (12a) suggeststhat there is a null focused demonstrative, and that the initial DP is topicalized, asrepresented in (12b). Though there is no overt mention of fire in the discourse, thespeaker earlier identifies himself as ?Sweat-Lodge?, and so assuming that there is anull focus in these cases, pragmatic inference may license topicalization.366(12) a. ?uPDEThin-s-w?l-?i1SG.POSS-firemFUThin-q?-x?a ?q-mn.1SG.POSS-pay-INSTRMy fire, that will be my payment. (Kalispel, Camp (2007, 85))b. ?uPDEThin-s-w?l-?i,1SG.POSS-fire\u000B = mFUThin-q?-x?a ?q-mn.1SG.POSS-pay-INSTRMy fire, that will be my payment. (Kalispel)Clefts may also involve c?@m? ?only? in Kalispel (13). The deictic c? is glossedas ?this; deictic? in Camp (2007, 104), and presumably forms a constituent withs@ ?lx?a ?l?x?ts ?his friends?, exemplifying the tendency for demonstratives to take theplace of determiners before nominals in argument positions (cf. 4b).(13) mFUTc?@m?onlyc?DEM.DISTs@-?lx?a-?l?x?t-sNOM-RED-friend-3SG.POSSmFUTq?P1PL.TRANSP??ntm.eat-TR-1PL.TRANS....and we will eat only his friends. (Kalispel, Camp (2007, 59))Finally, I found a textual example which exhibits a specificational ordering (14),similar to Okanagan data in section 7.7. This example may be analyzable as anidentificational sentence.(14) ??PiDEM?uPDETqeP-@?-t- ?cl-?lqw-iFUT-play.stick.game-FUT?uPDETs-kwest-sname-3SG.POSSs-x?wi-x?wi-?-t-t ?w?t.Sleepy.BoyThe name of this man who was now going to play, was Sleepy Boy.Literally: The one who was going to play the stick game, his name wasSleepy Boy. (Kalispel, Camp (2007, 79))Coeur d?AleneRelevant data from Coeur d?Alene is less forthcoming; however Doak (1997) andReichard (1947) include data which might be considered to exemplify DP-DP367structures and clefts. The elements xwE, cE, and ?E are listed as determiners inDoak (1997, 41), and the elements hi? and ? are listed as subordinators (i.e. com-plementizers) (Doak, 1997, 43), with ? possibly being a reduced form of hi?.In (15a), a determiner xwE introduces what is arguably an NP, though it isunclear whether the initial demonstrative ciP may be analyzable as a predicate.In (15b), the determiner introduces what is clearly an ergative-inflected relativeclause.6 Assuming that smax??Pc? ?n ?Grizzly Bear? is being used as a proper namein this context, it may also be analyzable as a DP. A similar example is shown as(15c), where a different determiner ?E introduces the residue. A third determiner cEintroduces an ergative-inflected relative clause in (15d), however the focused con-stituent p?pEPEt ?our father? is straightforwardly analyzable as an NP, hence (15d)may be considered to be an NPC.7 For lack of clearer evidence, all the examples in(15) may be analyzable as NPCs.(15) a. ciPDEM.DISTxwEDEThn-s-cEn-kw?nxw-cn.1SG.POSS-NOM-CUST-LOC-take-mouthThat was my answer. (Coeur d?Alene, Doak (1997, 272, ex. 525))b. kwu ?mthens-max??Pc? ?nGrizzly.bearxwEDET?iPt-Es-c?E(t)ta?lqinEP-nt-s.first-CONN-stomp.on-DIR-3SG.ERGGrizzly Bear was the first to stomp on him.(Coeur d?Alene, Reichard texts: Lynx, 037)c. yoINTERJ?pEP ??c ?nLynx?EDET??cumissinglutNEGhECOMPs-c?i-nP??xw-s.NOM-CISL-enter-3SG.POSSMy! It is Lynx that is the absent one...(Coeur d?Alene, Reichard texts: Lynx, 014d)d. Ekwnsay?tuPwellp?pEP-Etfather-1PL.POSScEDETpulu(t)-st-xw.kill-CAUS-2SG.ERGThey said ?It is our father that you killed.?(Coeur d?Alene, Reichard texts: Coyote Steals Son?s wife, 120b)6Shannon Bischoff has provided the glosses for the Reichard text examples. I leave the Coeurd?Alene in its original transcription rather than converting it to the Americanist orthography.7Though it is possible that p?pEP-Et ?our father? is a DP, albeit with a null determiner.368Example (16) comes from a text, and shows a topicalized DP consisting of ademonstrative and an NP (no determiner), followed by a focused DP (in bold typehere), and what appears to be a headless relative clause introduced by a comple-mentizer h:?, which is presumably the same complementizer hi? listed in (Doak,1997, 43), minus the vowel. These data suggest that Coeur d?Alene data with clearexamples of focused DPs may not be analyzable as equatives, and supports ananalysis of the data in (15) as examples of NPCs.(16) hOithenxwi ?ya?DEM.PROX?:tci?pbucket?a?DETtsa?r?wtsi?ntc ?ncoyote?s.sonh:?COMPsa?xwts.pack-DIR-3SG.ERGThen this bucket, Coyote?s youngest was the one who packed it.(Coeur d?Alene, Reichard texts: Badger and Coyote)Data involving focused independent pronouns show that residues are not neces-sarily introduced by either a determiner or a complementizer. In (17a), the residueis introduced by future c?E?, perhaps due to a double-filled complementizer filter(cf. Okanagan and Shuswap). In (17b), the residue is introduced by what Bischoffglosses as a ?connective? but which I assume is a complementizer, following Doak(1997, 43), rather than a determiner ?E. Assuming that ? in (17b) is indeed a comple-mentizer, and thus indicative a true cleft structure (cf. 16), the focused independentpronoun may be analyzable as a DP.8(17) a. nEPkwunthink?EDETsmyiwCoyotec?EPoughtc?n1SG.ABSPEngwt1SG.INDEPc?E?FUTc?nlikex?Eminc?COMPhESnipe?t?taqwi ?n.Coyote thought, ?I ought to be the one who is liked by Snipe...?(Coeur d?Alene, Reichard texts: Coyote Steals Son?s Wife, 004a)8Although the fact that absolutive morphology attaches to them rather seems to indicate thatindependent pronouns are predicates, at least from the Okanagan perspective.369b. hECOMPc?n1SG.ABSPEngwt1SG.INDEP?COMPcE-n-kwin-[n]t-s-n.CUST-n-take-DIR-2SG.ACC-1SG.ERGI am the one who saved you.(Coeur d?Alene, Reichard texts: Coyote Steals Son?s Wife, 132b)I have found several examples of specificational-like word orderings in the corpus.(18a), ironically, appears at first glance to be the strongest evidence thus far forequatives in Coeur d?Alene; however, the translation indicates that the proper nameis an appositive of sorts, and that it might be better analyzed as involving a nullfocus, as in (18b).(18) a. xwEDEThn-s-qwE?- sqwEs-EP1SG.POSS-NOM-sonxwEDETErnie...ErnieIt was my son, Ernie ... (Coeur d?Alene Doak (1997)))b. \u000B = xwEDEThn-s-qwE?- sqwEs-EP,1SG.POSS-NOMs-sonxwEDETErnie...ErnieIt was my son, Ernie ...Based on weak translation evidence, example (19) below may show a focusedproper name in final position, but tellingly, it is introduced by a complementizer,and not a determiner. Assuming that Coeur d?Alene follows the Okanagan pat-tern for these cases (and arguably Kalispel as well), (19) may be analyzed as anidentificational sentence.9(19) xwiPDEM.PROXxwEDETx?Eminc?likehECOMPqElpyE.Black.SwanBlack Swan was the one liked.Literally: That one who was liked is Black Swan.(Coeur d?Alene, Reichard texts: Coyote Steals Son?s Wife, 003a)The status of the initial particle hE in (19) and (17b) is unclear. N. Mattina (2006,102) lists it as an ?article?, on par with the other three determiners listed in Doak(1997, 41), and cognate with Okanagan iP. If this is indeed the case, then data like9Alternatively, (19) may be a cleft, equivalent to It is the one who was liked that is Black Swan,although the equivalent cleft in Okanagan is not possible.370(19) may in fact be analyzable as a specificational DP-DP structure.Moses-ColumbianMoses-Columbian exhibits relatively clear examples of DP-DP structures. Ana-lyzing the data is complicated by the fact that just as in Kalispel, determiners areoptional in Moses-Columbian (Willett (2003), N. Mattina (2006)). In (20), an overtdeterminer Pan? introduces an NP forming a DP constituent which might be ana-lyzed as standing in an equative relationship with an initial proximal demonstrativePax?P.10(20) a. Pax?PDEM.PROX[Pan?DETMaryMarylGENstxwul-s DP].house-3SG.POSSThis is Mary?s house. (Moses-Columbian, N. Mattina (2002, 264))b. Pax?PDEM.PROXwaABS[Pan?DETMaryMarylGENstxwul-s DP].house-3SG.POSSThis is Mary?s house. (Moses-Columbian, N. Mattina (2006, 111, ex. 35))Willett (2003, 87) notes that since the particle wa optionally marks absolutive argu-ments, as in (20b)11, this implies that the demonstratives in these cases are intran-sitive predicates. If the demonstratives are predicates, the prediction based on theOkanagan pattern is that they should also be able to follow their DP arguments, es-pecially given that while ?unmarked intransitive word order is VS? (Willett, 2003,95), SV is also possible as a case of ?unmarked fronting? (p.105). (21) is such anexample,12 but I have not been able to find any data involving Pax?P in final po-sition to substantiate the claim that these are predicates, nor do Moses-Columbiandemonstratives appear to function as complements to Pan?, which might be pre-10Willett (2003, 84) and N. Mattina (2006, 102) both analyze Pax?P, as well as Pac? and Pa??P asdeterminers on par with Pan?, rather than as demonstratives. It is unclear what the examples in (20)would mean, however, if both Pax?P and Pan? were determiners, and in any case, they should not beinterpretable as complete propositions. As such, I analyze Pax?P as a demonstrative, cognate withOkanagan proximal ax?P (cf. N. Mattina (2006, 102)).11The determiner Pan? is also optional in these cases.12?Grandfather? has already been introduced, but not mentioned in the immediately precedingcontext. ?Grandfather?s hands? were mentions 3 stanzas previous, however. These cases of unmarkedfronting are different from left-dislocated ?and-fronting? constructions, for which an initial DP isseparated from the rest of the sentence by kwaP ?and?.371dicted if demonstratives were predicative (cf. Kalispel 7a,b).(21) Pan?DET[?]s-p@-p?saPNOM-RED-grandfatherPac-??....q-@lx.CUST-sit(SG)-bodyGrandfather was sitting down.(Moses-Columbian, The Story of Crow?s Daughter, Davis (1990, 40))Clefts in Moses-Columbian may show evidence for being equative, though thisis unclear. For the cases in (22) below, a residue clause is introduced by a com-plementizer ?uP, which is not used as a determiner in Moses-Columbian (like inOkanagan, but unlike in Kalispel and Coeur d?Alene) and does not seem to occur inrelativization contexts either (Willett, 2003, section 5.4).13 Absolutive wa-markingin (22a,b) is interesting because it either marks the residue as an internal argumentof the focused constituent, or else introduces a null NP which is co-referent withthe focused absolutive (cf. Lyon (2011)).(22) a. s ??aPc?n@mdeer?uPCOMPwaABSPa ?wt?p-sfollow-[TR]-3.ERGt ?w?t.boyIt was the deer that the boy followed.(Moses-Columbian, Willett (2003, 109, ex.120))b. MaryMary?uPCOMPwaABSk??-xt-sgive-APPL-3.ERGtOBLy?mx?waPbasketJohn,JohnlutNEGwaABSIsabel.IsabelIt was Mary that John gave a basket to, not Isabel.(Moses-Columbian, Willett (2003, 109, ex.122))c. JohnJohn?uPCOMPk?- ?c@?m-u?s?-n.k?-suck-face-(dir)-1sg.ergJohn is the one I kissed.(Moses-Columbian, Ewa Czaykowska-Higgins, fieldnotes, 92.196)If, as seems likely, the second constituent in (22) is a predicative CP rather than13Willett (2003, 107-108) classifies these as ?quasi-clefts? (i.e. NPCs), where the initial NP func-tions as a predicate.372a DP, then following an analysis like Shank (2003) and Koch (2008a), the initialnominal should be a referential DP, i.e. these are true clefts. This seems like aplausible analysis for the data in (22), given that determiners are regularly null,given that ?deer? and ?boy? receive maximal interpretations despite the absence ofany overt determiner, and given that there are other similar data involving focusedNPs which are introduced by overt deictics (23):(23) Pac?DEM.DISTsmP?mmwoman?uPCOMPtOBLm?xwt.laughIt was that woman who laughed.(Moses-Columbian, Willett (2003, 109, ex.125))Other types of Moses-Columbian focus structures have residues that are intro-duced by determiners (24a) or other deictics (24b).14 As with the data involvingcomplementizer-introduced residues (22), bare nominals are possible in initial po-sition.(24) a. stxw?lhousePan?DETPi[n]-s-c-t?w.1SG.POSS-NOM-CUST-buyIt?s a house that I bought.(Moses-Columbian, Willett (2003, 113, ex.147))b. smP?mmwomanPac?DEM.DIST?t@ ?m-nc?t.cut-REFLEX.The woman cut herself. (Moses-Columbian, Willett (2003, 113, ex.150))While the non-maximal translation of ?house? in (24a) is consistent with an analysisof these cases as NPCs, assuming that Moses-Columbian determiners are similar toOkanagan iP in allowing non-maximal readings (cf. chapter 5), these could also beequative structures with null determiners preceding the initial nominal. The moststraightforward analysis is to analyze these cases as NPCs, however.As with Okanagan (cf. section 8.9), adjuncts in Moses-Columbian are clefted14Willett (2003) analyzes these as true ?clefts?. The difference between ?clefts? and ?quasi-clefts?for Willett is that the residue is introduced by a ?subordinator? in the latter, and by a determiner forthe former. I footnote her use of the terminology since this particular division seems questionable tome.373using a special particle ci, rather than the complementizer ?uP or a determinerPan?.15 In (25a), a passive agent is being clefted, and in (25b), an instrumentaladjunct is clefted.(25) a. tOBLJohnJohnciCOMPc@k-nt-mhit-DIR-PASSMary.MaryIt?s John who hit Mary.(Moses-Columbian, Ewa Czaykowska-Higgins, fieldnotes, 92.224)b. tOBL?n ?n? ?k ?mnknifeciCOMP?t@m-s.cut-[DIR]-3SG.ERGA knife is what he cut it with.(Moses-Columbian, Ewa Czaykowska-Higgins, fieldnotes, 92.206)As another point of interest, example (26) shows that null demonstrative fociare possible in identificational contexts (cf. section 8.5.3).(26) kwaPandwaABSs?w?kw.waterThat is water.(Moses-Columbian, The Story of Crow?s Daughter, Davis (1990, 42)))Finally, specificational word ordering is also found in identificational contextsin Moses-Columbian, just as with Okanagan, Kalispel, and Coeur d?Alene.(27) a. Pan?DETwaABSki ?Q?naPteenage.girllGENs- ?c ?qw-@nc?t-sNOM-name-REFLEX-3SG.POSSi:mli...EmilyThe girl?s name was Emily...(Moses-Columbian, The Story of Crow?s Daughter, Davis (1990, 2)))15Although it is tempting to analyze Moses-Columbian ci as cognate with Okanagan kiP, HenryDavis (p.c.) indicates that this is probably not the case, since there is no palatalization of velars inMoses-Columbian.374b. Pan?DETtt ?w?tboywaABS?w?taP.?witaP.The boy?s was W??taP.(Moses-Columbian, The Story of Crow?s Daughter, Davis (1990, 2)))SummaryFrom this brief overview of Southern Interior Salish, Kalispel (Montana Salish)shows the strongest possible evidence for equatives (3e), repeated below as (28).(28) ?uDETsmPemwoman?uDETsxwmimeyePm.teacherThe woman is a teacher. (Kalispel, Tachini Pete p.c. 2011)Since cleft residues in Kalispel are routinely introduced by the determiner ?u(P),they may be analyzable as DPs in Kalispel, though I was not able to find a clearexample of a headed relative in residue position. The status of demonstratives issomewhat unclear in the language since (i) there is conflicting evidence that theymay be predicative, and (ii) the demonstrative i?e patterns radically different fromOkanagan ix?P in not being able to follow a nominal predicate. The semantics ofthe determiner ?u(P) are unclear as well, but assuming that it is functionally equiva-lent to Okanagan iP, which seems likely given that they have more or less identicaldistributions, prospects for extending the equative analysis I have developed forOkanagan DP-DP structures and clefts to Kalispel seem very promising. The dis-tribution of ?u(P) in identificational contexts patterns with the use of Okanagan?@ in similar contexts, which raises the possibility that ?u(P) may have a limitedfunction as a complementizer in Kalispel.Coeur d?Alene, strikingly, does not show evidence for equatives. Rather, itfollows the pattern exhibited by Thompson (Koch, 2008a, 2009): NPCs introduceresidues with determiners, whereas cleft residues are introduced by complemen-tizers. This is unexpected, given that Kalispel and Okanagan show clear evidencefor equatives, but it is possible that this is simply due to a gap in the data. Coeurd?Alene identificational sentences utilize complementizers before proper names,similarly to Okanagan.375Moses-Columbian does not exhibit straightforward evidence for equative struc-tures. First of all, it is unclear whether demonstratives may be predicates, orwhether absolutive wa may help decide this issue. Secondly, it is often unclearwhat the categorial status of the clefted constituent is, given that determiners maybe null. Since ?uP is always a complementizer in Moses-Columbian, a subset ofcleft residues may be analyzed unambiguously as CPs. Assuming that a CP residuerequires a DP focus for Salish clefts, a bare nominal in focus position may be an-alyzed as a DP, following the pattern in Thompson. For other cases, where theresidue is introduced by a determiner, I have not been able to find any unambigu-ous examples of determiner-headed DPs in focus position. There is therefore noreason at the moment to reject the null hypothesis that these are in fact NPCs.In sum, Kalispel shows strong evidence for equative DP-DP structures andprobably clefts as well, similarly to Okanagan, while Coeur d?Alene and Moses-Columbian do not.Table 9.4: Evidence for Equative versus Predicational Analyses of SouthernInterior Salish DP-DP Structures and CleftsEquative PredicationalOkanagan X *Kalispel (Montana Salish) X *Coeur d?Alene * XMoses-Columbian * X9.3.2 Implications for Northern Interior SalishThis section discusses data from the Northern Interior Salish languages Shuswap,Thompson and Lillooet. Clefts have been studied much more intensively in theselanguages than in the Southern Interior languages just discussed. I begin with adiscussion of Shuswap data involving DP-DP structures and clefts. I then discussin some detail Thompson data from Koch (2008a), and in particular, his analysisof clefts in the language. Although he does not adopt an equative analysis of cleftsfor Thompson (cf. also Shank (2003) for Northern Straits in section 9.3.3), his376arguments are still directly relevant to this dissertation. Finally, I discuss cleft datafrom Lillooet (Davis et al., 2004). I then summarize the results of this survey.ShuswapShuswap appears to have DP-DP structures; however, their grammatical status isunclear. According to Gardiner (1996), (29a) below is translatable as a demonstrative-associated DP, but not as a complete sentence.16,17 However, Kuipers (1989) showsa clear case of a DP-DP structure where the second DP is introduced by the absentdeterminer l (29b). This difference may be attributable to a generational differencebetween speakers.(29) a. yeG?yDEM.DISTGeDETn ?cePsq?x?eP.horsethat horse*The horse is that. (Shuswap, Gardiner (1993, 181, ex. 61))b. yGiPDEM.DISTlDET.ABSn-xkw@tkw?t?stn.1sg.poss-eyesThose are my eyes. (Shuswap, Kuipers (1989, 48, ex. 33))Lai (1998) shows that a proper name can be in the focus position of a cleft(30a), but she analyzes the proper names in these cases as predicates, since a propername cannot be introduced by a determiner in this position, whereas proper namesin argument positions are normally introduced by determiners (unlike the case forOkanagan). As such, (30a) is equivalent to an NPC for Lai (1998). Independentpronouns can also occur in this position (30b).(30) a. JohnJohnG@DETw?kt-t-m-@s.see-TR-PASS-3SG.CJCTIt is John that saw him/her. (Shuswap, (Lai, 1998, 311, ex. 20b))16Data from Gardiner (1996) and Lai (1998) were originally given in a practical orthography,which I have standardized to an Americanist orthography for the sake of consistency with otherresources such as Kuipers (1974) and Gardiner (1993). Additionally, Kuipers (1974, 1989) writesdeterminers as prefixes, rather than as separate words. I write them as separate words for expositorypurposes.17Neither should (29a) be interpretable as ?That is the horse?, but this is less clear.377b. nwiPs3.EMPHGiPDEM.PRESG@DETx?w@nt.fastIt?s that one that is fast. (Shuswap, (Gardiner, 1996, 181, ex. 59))I was nevertheless able to find some clear-cut cases of clefts involving focusedindependent pronoun DPs (31), which may be analyzed as equative.18 In (31a), theabsent determiner l introduces a headless relative clause in final position, and in(31b), the default determiner G introduces a headless relative clause.(31) a. G@DETn-c?@?c?w@P1SG.POSS-EMPHlDET.ABSwx-st@-x.mention-TR-2SG.ERGI am the one you mentioned. (Shuswap, (Kuipers, 1974, 117, line 39))b. G@DETn-c?@?c?w@P1SG.POSS-EMPHG@DETw?kt-t-m-@s.see-TR-PASS-3SG.CJCTIt is I that saw him. (Shuswap, Lai (1998, 312, f.n. 14))Gardiner (1993) contains similar examples, with a residue introduced by thereduced demonstrative GiP (32) rather than the default G determiner (31b).19,20(32) GDETn-c?E?c?w@P1SG.POSS-EMPHGiPDEM.PRESGDETw?[w]k-t-sm-s.see-RED-TR-1SG.OBJ-3SG.ERGI?m the one that she saw.18Shuswap independent pronouns are never introduced by determiners except for the 1st personsingular in the Northern dialect (Kuipers, 1974), (cf. 31). Okanagan independent pronouns cannot bepreceded by determiners, and their distribution in equative and predicational structures is somewhatunclear, which is why I have abstracted away from similar data in Okanagan for this thesis.19Kuipers (1974, 57) describes the demonstratives GiP ?present? and luP ?non-present? as ?generaldeictics?. ?The former often has to remain untranslated... very often inserted in non-formal speech?20There is a section on ?clefts? in Lai (1998, section 2.1), but these data resemble cases of un-marked fronting, rather than clefts, since there is no determiner or other particle preceding the residue(i):(i) ScottScottx?w@.x?wist@?t-@n.like(RED)-1SG.SUBJIt is Scott that I like.Note that in Lillooet, a residue introducing particle is also not obligatory (cf. Davis et al. (2004)and section 9.3.2 below), however in Lillooet, an overt copula ni? unambiguously indicates a cleftstructure.378It is unclear to me whether (31b) and (32) actually have substantively differentstructures or not. Given (i) that demonstratives can adjoin to DPs in Shuswap(29,30b), and (ii) that GiP is a reduced demonstrative (Gardiner, 1993, 181)21, itseems plausible that (31b) and (32) have the same structure, and that GiP onlyoptionally introduces cleft residues in Shuswap (cf. 30), similarly to the case forOkanagan where optional ix?P does not procliticize to initial position. All examplesanalyzed as ?clefts? in (Gardiner, 1993) involve a reduced demonstrative introduc-ing the residue clause (33), either luP ?non-present? (33a) or GiP ?present? (33b,c)(Kuipers, 1974, 57).(33) a. GDETJohnJohnluPDEM.ABSTlDETm-wik-t-s.PERF-see-TR-3SG.SUBJIt?s John that he saw. (Shuswap, (Gardiner, 1993, 73, ex.4))b. GDETMaryMaryGiPDEM.PRESGDETJohnJohnlDETm- ?c? ?m-qs-n-s.PERF-lick-nose-TR-3SG.SUBJIt was Mary that John kissed. (Shuswap, (Gardiner, 1993, 76, f.n.3))c. MaryMaryGiPDEM.PRESGDETwik-t-m-@ssee-TR-PASS-3SG.CJCTMary is the one that saw him. (Shuswap, (Gardiner, 1993, 93, ex.71))Importantly for the sake of comparison with Okanagan, Gardiner (1993, 78)notes the presence of an optional initial demonstrative (34b), non-reduced y@G?P.Note the change in translation between (34a) without an initial demonstrative, and(34b) with an initial demonstrative. This suggests that it is the demonstrative whichis in focus in (34b), implying that headed residues may be possible in Shuswap, andthus that clefts are equative.2221Gardiner (1996, 181) analyzes the visible distal particle GiP (and invisible luP) as reduceddemonstratives which have been ?grammaticized as focus particles?. He glosses GiP as PART ?parti-cle?, but I gloss it as DEM ?demonstrative? in order to make my point.22If this is the right analysis, then it shows that the reduced demonstrative GiP can move to theleft periphery of a clausal CP, along with a determiner G, in Shuswap. Alternatively, if the DPG sqE?lmxw ?the man? is in focus position as a constituent with the initial demonstrative, then thereduced demonstrative GiP is directly equivalent to the clefting demonstrative ix?P in Okanagan forcases where it remains in-situ.379(34) a. GDETsqE?lmxwmanGiPDEM.PRESGDETwik-t-s.see-TR-3SG.ERGIt?s the man that she saw. (Shuswap, (Gardiner, 1993, 78, ex.14))b. y@G?PDEMGDETsqE?lmxwmanGiPDEM.PRESGDETwik-t-s.see-TR-3SG.ERGThat?s the man that she saw. (Shuswap, (Gardiner, 1993, 78, ex.17))For cases involving demonstrative-focus, Shuswap determiner G alternates withthe oblique-irrealis determiner ?tk, as shown in (35). This is directly analogous tothe alternation in Okanagan demonstrative-focus clefts between the determiner iPand the oblique marker t (cf. section 8.5.2).(35) a. yG?PDEMyP@?nDEM?tkOBL.IRRqlm?xwmanlDET.ABSTm-w?wk-c-m-s.PERF-see-DIR[?]-2SG.ACC-3SG.ERGThis is the man who saw me. (Shuswap, Kuipers (1974, 83))b. yG?PDEMyP@?nDEM?tkOBL.IRRqlm?xwmanlDET.ABSTm-tP@?Py-n.PERF-meet-DIR[?]-1SG.ERGThis is the man I met. (Shuswap, Kuipers (1974, 83))Notice that both examples in (35) begin with a sequence of two demonstratives,and both are translated as having focus on the second, proximal demonstrative.The first demonstrative is presumably a deictically weak clefting demonstrative,and its initial position is consistent with an argument whereby it has undergoneproclisis to initial position, while forming an underlying constituent with the nom-inal introduced by ?tk, exactly as in Okanagan.23One major problem with analyzing Shuswap clefting demonstratives as discon-tinuous constituents with a residue clause is that Shuswap, as a Northern Interiorlanguage, is a primarily encliticizing language, unlike Southern Interior Salish lan-guages which are procliticizing languages. Shuswap has both reduced demonstra-23Henry Davis (p.c. 2013) points out that if the initial demonstrative were underlyingly a con-stituent with the residue in (35a,b), it should match the absent deictic features of the residue-internaldeterminer. I am not sure if this should necessarily be the case, however, since in Thompson (46d)below, for example, a distal demonstrative is arguably being equated to a referent which is introducedby the remote determiner ?, and in Okanagan, a deictically weak occurrence of ix?P, as in a cleftingcontext, can easily refer to something spatially and/or temporally remote.380tives (e.g. GiP, similar to Thompson clefting xeP), and full variants (e.g. y@G?P)which carry stress. Shuswap reduced demonstratives do not introduce clefts in thedata that I have found, which makes sense if they are enclitics. It is possible thatShuswap clefting demonstrative y@G?P and Okanagan clefting demonstrative ix?Pare treated as full words by the phonology, rather than clitics, and that the processI have labelled ?demonstrative proclisis? for Okanagan might more accurately becalled ?demonstrative floating?.Also similar to Okanagan, the pattern involving the the oblique-irrealis deter-miner ?tk (sometimes tk (Kuipers, 1974, 57)) surfaces in non-cleft environments(36), which supports an argument whereby the demonstrative forms a constituentwith the oblique-marked nominal.(36) yG?PDEMtkOBL.IRRqlm?xwmanPexIMPFGDETs ?cn?qw-st-s@sfight-CAUS-3SG.ERGGDETknk?knm.bearThis man fought with a bear. (Shuswap, Kuipers (1989, 36, ex. 60))In the data below, an initial demonstrative is in focus, followed by a cleftingdemonstrative GiP which is a constituent with the residue, similar to Okanagan datawhere a non-focused demonstrative does not undergo proclisis. (37b) is interestingsince it contains a sequence of three demonstratives. The first demonstrative is infocus, the second is a clefting demonstrative and introduces the residue, while thethird refers to the residue-internal subject. (37c) shows that Shuswap cleft residuescan be introduced by a future particle meP, equivalent to Okanagan future mi, andKalispel m. I have added an ?=? below to indicate the position of the proposedequative head.(37) a. y@G?PDEM= GiPDEMGDETwik-t-m.see-TR-PASSThat?s the one that was seen. (Shuswap, (Gardiner, 1993, 89, ex.51))b. y@G?PDEM= GiPDEMy@GE?yDEMGDETwik-t-s.see-TR-3SG.SUBJThat?s the one that this one saw. (Shuswap, (Gardiner, 1993, 89, ex.51))381c. y?-@kweit-QUOTyG?yDEM= mePFUTml-m?lqw-nt-m-@sIRED-paint-DIR-AD.CONJ-3SG.ERGGDETcitxw.houseThis is the one that is going to paint the house.(Shuswap, Kroeber (1999, 390))(37c) is important for another reason. The initial particle y? is glossed inKuipers (1974, 267) as a ?deictic-anaphoric? stem. It is possible that y? is thespellout of an equative copula, and that Shuswap, like Thompson, utilizes bothovert copulas and demonstratives in cleft contexts.24 This is speculative, however.ThompsonKoch (2008a, ch.7), Koch (2009), and Koch and Zimmermann (2009) provide asyntactic and semantic analysis of clefts in Thompson River Salish which offersdirect points of comparison to my own proposed analysis of Okanagan clefts asequative. I first discuss relevant syntactic and semantic aspects of Thompson cleftsin light of Koch?s analysis, before raising some questions concerning a data setwhich seem to show that Thompson has DP-DP structures.Thompson focus structures involving a DP focus are introduced by the cleftingpredicate ?ce (38) (cf. discussion in section 8.2.1). Both DP focus (bracketed andbolded below) and residue clause are introduced by what appear to be determiners,however Koch (2008a, 2009) argues that the clause-introducing particle is actuallya complementizer, and Thompson residues are bare CPs (cf. also Davis et al. (2004)for Lillooet and Northern Straits Salish).(38) ?ceCLEFT[?DETR?ssDP]RosseDETp?nt-@-t-\u000B-mus.paint-DRV-TR-3.OBJ-SUBJ.EXTRIt was Ross that painted it. (Thompson, Koch (2008b, 2))24This also raises questions concerning the morphological structure of demonstratives in clefts,and whether an initial demonstrative y@G?P might be further analyzable as y@-G?P ?COP-DEM?. Thismight also imply that the ?reduced? post-focus demonstrative G?P is simply not occurring with thecopula.382This essentially precludes an equative analysis of Thompson clefts, though Kochhas good reasons for rejecting the equative analysis, based on the syntactic andsemantic properties of cleft residues in Thompson. His four syntactic arguments(2009, p.3) against analyzing cleft residues as DPs are as follows:(39) a. cleft structures do not bear transitive marking.b. clefts have rigid post-predicative word order.c. cleft residues are not introduced by the full range of determiners.d. cleft residues do not have overt NP heads.Among these arguments, (39c) and (39d) are the strongest, and the ones I willdiscuss here.Concerning the third argument (39c), Thompson cleft residues are introducedby the specific determiner e (40b) or the irrealis determiner k, and never by theremote determiner ? (40a) or by the oblique marker t. The determiners e and kare also used as complementizers in Thompson. The oblique marker t obligatorilyintroduces relative clauses in Thompson (Kroeber, 1997; Koch, 2006), and so itsabsence in (40b) is unexpected.(40) a. * ?ceCLEFTeDETMon?queMonique[?DETw?k-t-ne].see-TR-1SG.ERGIt was Monique that I saw. (Thompson, Koch (2009, ex.13a))b. ?ceCLEFTeDETMon?queMonique[eDETw?k-t-ne].see-TR-1SG.ERGIt was Monique that I saw. (Thompson, Koch (2009, ex.13a))As Koch (2008a, 226) states, the ?absence of ? here is especially telling, sinceit indicates that the determiner introducing residue clauses does not begin as anargument DP of the subordinated verb, as relative pronouns (t)-? do in relativeclauses.? In other words, cleft residues are not structurally equivalent to relativeclauses in Thompson.(41a) below shows an oblique marker-determiner sequence t-? introducing arelative clause, while (41b) and (41c) together show that no sequence of obliquemarker plus determiner can introduce an NPC residue or an introduced cleft residue383clause in Thompson.(41) a. Relative Clausecw-?mmake-MIDkn1SG.ABSxePDEMteOBLk?picoffeet-?OBL-COMPs-t?w-cn-me-sNOM-buy-mouth-MID-3SG.POSS?DETnsmP?m1SG.POSS-wife?DETspiPx?wt.dayI made the coffee that my wife bought yesterday.(Thompson, Koch (2008a, 211, ex.8))b. NPC[k?pi FOC]coffeexePDEM(*t)(*OBL)eDETn-s-cw-?m1SG.POSS-NOM-make-MID?COMPq??-tawake-IMwn1SG.CJCT?DETsnw?nwen.morningI made [coffee FOC] when I got up this morning.(Thompson, Koch (2008a, 211, ex.9))c. Cleft clauses are not introduced by a determiner from inside the residueclause [give the grammatical case (without oblique marking)]* ?ceCLEFT?DETRossRoss[t-eiOBL(/t-k/OBL-IRL/t-?)/OBL-DET[p?nt-e-t-mus ti]].paint-DRV-TR-SUBJ.EXTRIt?s Ross who painted it. (Thompson, Koch (2008a, 225, ex.36))Relating to this point, Koch (2008a) shows that while headless locative relativeclause DPs are introduced by a preposition and determiner which have moved froma position following the relative clause predicate (42a), residues in focus structuresare generally not (42b):25(42) a. Headless locative relative clauses are introduced by the clause-internalpreposition25The same is true for Okanagan, though Okanagan uses an entirely separate structure for cleftedprepositional phrases: adjunct clefts (cf. section 8.9).384cu-t-?sfix-TR-3SG.ERG[n-?iin-DETixw? ?yFUTun1SG.CJCTm?cePqsitti DP].She fixed what I was going to sit in. (Thompson, Kroeber (1997, 397))b. Locative residues in focus structures lack an initial preposition frominside the residue clauseWhich way did she sleep?[nein-DETsx?? ?c ?k ?n-s FOC]back-3SG.POSSekwuEVID??uPEMPH[(*n)(*in)kIRLQ? ?y-tsleep-IMus DP].3.CJCTShe slept [on her back FOC]. (Thompson, Koch (2008a, 226, ex.37b)Moving on to the fourth argument (39d), Koch (2008a, 232) shows cruciallythat Thompson cleft residues cannot contain an overt NP head (cf. (Davis et al.,2004) for Lillooet and Northern Straits Salish):(43) Thompson Cleft residue clauses lack an overt nominal head* ?ceCLEFTxePDEMeDEThel? ?weagle[eDETspzupz?Pbirdt-eiOBL-DETwP?xPROGn-xw?l-ixLOC-fly-AUTti].The birds that are flying are the eagles. (Thompson, Koch (2008a, 232))Taken together, the points given above as (39c,d) provide strong syntactic evidencethat the residue of a Thompson cleft is not a DP, but is instead a CP.Koch?s analysis of the cleft clause as a predicative CP leads to a semantic anal-ysis of the clefting predicate ?ce as in (44a) with the implicature in (44b) (Koch(2008a, 237), cf. Shank (2003) for Northern Straits):(44) a. [[ ?ce]] = ?xe?P.P(x)b. Implicature: there exists some x such that P(x) = 1The formula in (44) means that ?ce is essentially a predicational copula. Thisanalysis is supported by data showing that Thompson clefts lack exhaustivity ef-fects (45a) and any presupposition of existence (45b)(cf. Percus (1997) for English,and related data for Okanagan in section 2.4.2).385(45) a. Context: Peter went fishing, did anyone else go fishing??ceCLEFTekwuEVIDPe?and??uPevenxePDEMeDETJohn.JohnJohn did too. (??It was also John that went fishing.)(Thompson, Koch (2008a, 215, ex.18))b. Context: Addressee is being handed a bow and arrow, with no prior men-tion that something is to be used for hunting.?ceCLEFTx?PDEMeDETqwez-t-?suse-TR-3SG.ERGeDETxPROG?q? ?q ?y- ?m.shoot(DIM)-MIDThis here is to use for hunting.(Thompson, Koch (2008a, 220, ex.30))Thompson and Thompson (1992, 216, line 189))He explains that by analying the second semantic argument of ?ce (i.e. the residue)as a predicate, the absence of any exhaustivity entailment in Thompson clefts fol-lows (45a), since as a predicate, the residue may be true of more than one in-dividual. This analysis also explains why Thompson clefts can be used in non-presuppositional contexts (45b), since predicates are inherently non-maximal. Theimplicature in (44b) arises from an assumption that the set is non-empty.This analysis correctly derives differences between English clefts, which haveexhaustivity entailments and existential presuppositions, from clefts in Thompson,which do not. For theories of English clefts such as Percus (1997) and Hedberg(2000), the residue clause is the restriction of a concealed definite description, andis therefore a referential type e (cf. discussion of these theories in section 2.4). Theexhaustivity entailment and existential presupposition of English clefts can there-fore be linked directly with the definite determiner the. In contrast, Koch (2008a)analyzes residue-introducing determiners in Thompson clefts as complementizerswhich do not saturate the predicative CP, and so the semantics of clefts in Thomp-son are necessarily divorced from the semantics of the determiners.It is important to note however that Salish determiners are well known for lack-ing exhaustivity entailments and existential presuppositions, as shown for Lillooet(Matthewson, 1998), Squamish (Gillon, 2006), and Okanagan (cf. chapter 5). Fol-lowing Shank (2003, 225), Koch (2008a, 238) actually considers the possibility386that the residue-introducing determiner is not semantically vacuous, but instead in-troduces a choice function (Reinhart, 1997; Matthewson, 1999) which selects anentity from the set denoted by the relative clause, essentially a version of the equa-tive analysis I adopt in chapters 7 and 8. As an existentially quantified indefinite,the absence of any exhaustivity entailment or existential presupposition follows,and the semantics of clefts in Thompson could in principle be linked to the seman-tics of determiners (Percus, 1997; Hedberg, 2000).Though he ultimately does not adopt an equative analysis, Koch (2008a, 231-2)shows data which are directly analogous to simple DP-DP structures in Okanagan,except that they are introduced by either the clefting predicate ?ce (46b-c) or cukw??uP ?only? (46a). With regards to data of this type, Kroeber (1999, 370) statesthat the ?predicative particles [e.g. ni?] can be regarded as a sort of identificationalcopula, marking the nominal expression that immediately follows it as identical inreference to the nominal expression that is the subject of the whole construction(which may be a null pronominal).?(46) a. cukwonly??uPjust?DETAliceAliceeDETn-sn? ?kweP.1SG.POSS-friend.Only Alice is my friend.(Thompson, Koch (2008a, 232, ex.45a), Kroeber (1997, 389))b. ?c?CLEFT?nYNQxePDEM[kIRLeP-sn? ?kwePDP]2SG.POSS-friend[?DETKris?DP]?ChrisIs Chris your friend? (Thompson, Koch (2008a, 232, ex.45b))c. HeP?y,yes?c?CLEFTxePDEM[eDET?pls?DP].appleYes, that?s an apple. (Thompson, Koch (2008a, 232, ex.45c))d. ?c?CLEFTxePDEM[?DETn-sq?czeP, DP]1SG.POSS-fatherPe?and?c?CLEFTxePDEM[?DETn-s?nciPDP].1SG.POSS-brotherThis is my father, and this is my brother.(Thompson, Koch (2008a, 232, ex.45d))387These data yield several generalizations, and raise several important questions,each of which I will discuss in turn:(47) a. Thompson demonstrative xeP is plausibly analyzable as a focused DP(46c,d), analogous to Okanagan ix?P.b. Since a ? determiner, which is not used as a complementizer in Thompson,can introduce an NP in what appears to be a DP-DP structure (cf. 46b,d),how is it possible to reconcile the denotation of ?ce in (44) with the use of?ce in these contexts?c. The Thompson determiner ? has the same distribution as the Okanaganoblique marker t in focus structural contexts: it can precede an NP (46b,d)but not a clausal residue (cf. 40a). Both Thompson (t)-? and Okanagan tmay occur before headed relative clauses in non-cleft contexts (cf. Thomp-son 41a).Regarding (47a), Koch (2008a, 273) questions whether the unstressed demon-strative xeP can be a focus. Compare the identificational direct predication with astressed referential demonstrative in (48a) below with the identificational DP-DPstructure with an unstressed demonstrative enclitic (48b cf. 46c) and the identifica-tional DP-DP structure with a stressed demonstrative (48c). Neither the syntacticplacement of a demonstrative, nor whether it is stressed or not, appears to affectwhether or not the demonstrative can be referential, and the translations of (48b)and (48c) are at least consistent with an interpretation where both demonstrativesare in focus. As a step towards clarifying this issue, it might be useful to testwhether (48b) and (48c) both give rise to an exhaustivity implicature.(48) a. [n ?q?x?cetn FOC]keyx?Pe.DEMThat?s a [key FOC]. (Thompson, Koch (2008a, 45, ex.9) )b. HeP?y,yes?c?CLEFTxePDEM[eDET?pls?DP].appleYes, that?s an apple. (Thompson, Koch (2008a, 232, ex.45c))388c. ?ceCLEFT?nYNQx?PeDEMkDET.IRLePnt?ytn.teapotIs that your teapot? (Thompson, Kroeber (1999, 371))Two alternative analyses for data such as (48b) are as follows: (i) there is eithera null focus; or (ii) the bracketed DPs are in focus and there is a null residue, i.e.these are truncated clefts. As truncated clefts, it is unclear what the elided residueclause in (48b) could denote: one possibility is an empty predicate whose set isdetermined by the context; a second possibility, assuming that these are equatives,is that there is a null pronominal in residue position (Kroeber, 1999, 370).This brings us to a discussion of (47b). If the remote determiner ? in (46b,d)is indicative of a DP structure, then these examples may only be analyzed as equa-tives, and it follows that the copula ?ce must be lexically ambiguous between anequative type > and a predicational type ,t>>, as implied byKroeber (1999, ch.7).Concerning the absence of Thompson remote determiner ? before a clausalresidue (47c), I have claimed that when Okanagan t occurs in clefts, it introducesa necessarily overt head NP of a relative clause (cf. section 8.5.2). I suggest thatThompson ? does not introduce a clausal residue because like Okanagan t, it selectsfor only overt NPs in these contexts. I have also argued that pre-nominal t inOkanagan equatives must be licensed by a demonstrative ix?P (cf. section 8.5.2),and it is notable that all of the data in which ?ce and ? co-occur in (46) involve ademonstrative xeP. While this could just be a coincidental correspondence, it isworth noting. Assuming that Thompson DP-DP structures share similarities withthose in Okanagan, the prediction is that controlling for a non-human referent,(46d) should be possible with focus on the demonstrative, and a relative clausemodifying a head introduced by ?.Concerning the absence of the Thompson oblique marker t before a clausalresidue (47c), Koch (2006, 133) analyzes relative clause-introducing t as adjoinedto CP, as the head of some higher projection ?XP?. In headless relatives, the obliquemarker is obligatorily absent (cf. Koch (2008a, 45, ex.9)), possibly due to the factthat this particular projection is absent from headless varieties for some reason.Note that the presence or absence of XP does not affect clause-internal movement389to the left-periphery of CP. The absence of t before cleft residues could be due tothe fact that like headless relative clauses, this particular projection is absent.While it is the case that positing selectional restrictions for remote determiner ?in cleft contexts, and positing that the oblique-marker containing XP projection isabsent for cleft residue clauses, together plausibly explains the absence of both ofthese particles in cleft contexts, this story admittedly does not explain why an overtNP head cannot occur within a residue clause in Thompson. It is therefore moststraightforward to acknowledge that while Thompson has DP-DP structures thatare plausibly equative, clefts cannot be equative (cf. Coeur d?Alene and Moses-Columbian), and that there is a semantic ambiguity in the type of the copula ?ce.LillooetThe closest equivalents to DP-DP structures in Lillooet are preceded by the cleftingpredicate nilh (49). The first constituent (the ?focus?) is commonly a demonstrative,while the second constituent (the ?residue?) can be either a NP (49a) or a DP (49b),though not all speakers allow DPs in this position. Henry Davis (p.c.) mentionsthat the determiner which introduces the second constituent is optional except incases where a proper name is in focus (49c).(49) a. ni?COPcPaDEM?claP.basketThis is a basket. (Lillooet, Davis (2010c, ch. 6 ex.48))b. ?ni?COPcPaDEMtaDET?cl?P-a.basket-EXISThis is a basket. (Lillooet, Davis (2010c, ch. 6 ex.48?))c. ni?COPs-C?qw@m?aPNOM-C?qw@m?aPtiDETskw?cic-s-aname-3SG.POSS-DETtiDETk?kwpiP-a.chief-EXISC?qw@m?aP is the chief?s name. (Lillooet, Jan van Eijk, p.c.)Similar to the case of Kalispel i?e (cf. section 9.3.1), a Lillooet demonstrativecannot follow an NP or DP predicate in residue position (50a,b) (or alternatively,demonstratives cannot be residues). Similar to Okanagan DP-DP structures, a di-rectly referential expression cannot follow a determiner-headed DP in the context390of a the copula ni? (50c).26(50) a. *ni?COP?cl?PbasketcPa.DEMThis is a basket. (Lillooet, Jan van Eijk, p.c.)b. *ni?COPtaDET?cl?P-abasket-EXIScPa.DEMThis is a basket. (Lillooet, Jan van Eijk, p.c.)c. *ni?COPtiDETskw?cic-s-aname-3SG.POSS-EXIStiDETk?kwpiP-achief-DETs-C?qw@m?aP.NOM-C?qw@m?aPThe chief?s name is C?qw@m?aP. (Lillooet, Jan van Eijk, p.c.)In identificational contexts, auxiliary waP introduces a final NP (51a,b), simi-larly to Okanagan identificationals where the complementizer ? introduces a properor common noun NP. In Lillooet, however, waP is not optional in identificationalcontexts. Thus, (51c) cannot be used when teaching someone the name ?coffee?.27Okanagan and Lillooet thus differ, in the sense that the ? complementizer is notrequired for an identificational reading in Okanagan.(51) a. ni?COPtiPDEMwaPAUXT@QtQ?yna.T@QtQ?ynaThat was (who is called) T@QtQ?yna.(Lillooet, Davis (2010c, ch. 6 ex.55))b. ni?COPcPaDEMwaPAUXk??pi.coffeeThis is (what is called) coffee. (Lillooet, Davis (2010c, ch. 6 ex.53))26The assumption is that (50c) is not felicitous in contexts for which ?predicate topicalization? ispossible in Okanagan. ni? then always requires contrastive focus on the DP in focus position.27With regards to the observation that auxiliary waP normally indicates that a predicate is a tem-porary property of an individual, Davis (2010c, ch.6, 12) notes that ?traditional ucwalmc?wts namesare in some sense temporary possessions of the people who bear them: they are passed on fromgeneration to generation, like heirlooms.?391c. ni?COPcPaDEMk??pi.coffeeThis is coffee (here). (Lillooet, Davis (2010c, ch. 6 ex.52))Proper names, assertion-of-existence DPs, and headless relative clause DPsmay all occur in focus position as well.(52) a. ni?COPs-SpikeNOM-SpiketiDETwaPAUXk?kwpiP.chiefSpike is the chief. (Lillooet, Henry Davis, p.c. 2013)b. ni?COPtiDETk?kwpiP-achief-EXISwaPAUXs-C?qw@m?aP.NOM-C?qw@m?aPIt?s the chief who is called C?qw@m?aP (not someone else.)(Lillooet, Henry Davis, p.c. 2013)c. ni?COPtaDETs-Pal@lnaP?l-aNOM-youngest-EXISwaPAUXs-Kw?mcx@n.NOM-Kw?mcx@nIt?s the youngest who is called Kw?mcx@n (not the eldest).(Lillooet, Henry Davis, p.c. 2013)Davis et al. (2004) show that NPCs and introduced clefts in Lillooet (andNorthern Straits) lack any exhaustivity entailment or existential presupposition,similarly to Thompson and Okanagan. As such, I will not repeat their seman-tic arguments or associated data here, but will instead focus on morpho-syntacticproperties of Lillooet clefts.Davis et al. (2004, 102) find that ?the basic difference between NPCs and cleftsis that the residue in an NPC is a headed relative clause, whose head may be phono-logically null, whereas in a cleft the residue is a bare CP.? This notably contrastswith Thompson NPCs, which cannot have overt heads Koch (2008a, 233).Clear examples of Lillooet clefts are introduced by a clefting predicate ni?,followed by a DP in focus position, as shown in (53a). Like data involving an NPor DP in residue position (cf. 49), a determiner is only required before a clausalresidue if the focused DP is a proper name (compare 53a and 53b).392(53) a. ni?COP[?-JohnDP]NOM-JohntaDETkwan-ta?l?-hatake-NST-EXIStaDETk??h-?-acar-3.POSS-EXIS?-Mary.NOM-MaryIt?s John that took Mary?s car. (Lillooet, Henry Davis, p.c.)b. ni?COPtiDETkw?kwpiP-achief-EXISwaPAUX?kw@z?s-@m.work-MIDIt?s the chief who is working. (Lillooet, Jan van Eijk, p.c.)Concening cleft residues, Davis et al. (2004, 104, fn6) note that Lillooet in-troduced cleft residues do not permit overt NP heads unless the head follows theclausal portion of the residue. Thus, compare (54a,b) with (54c).(54) a. ni?CLEFTPiDET.PLq@?m@?m@ ?n-aold.person-DET?m@?m??ac?woman(PL)[n@?DET.PLqwa?lqw@?l@??lt-?-an-a.]speak-CAUS-1SG.ERG-DETIt was the old women that I spoke to.(Lillooet, Davis et al. (2004, 103, ex.8))b. *ni?CLEFTPiDET.PLq@?m@?m@ ?n-aold.person-DET?m@?m??ac?woman(PL)[n@?DET.PL?xw?pm@x-aShuswap-DETqwa?lqw@?l@??lt-?-an.]speak-CAUS-1SG.ERGIt was the old women who were the Shuswaps that I spoke to.(Lillooet, Davis et al. (2004, 104, ex.10))c. ni?CLEFTPiDET.PLq@?m@?m@ ?n-aold.person-DET?m@?m??ac?woman(PL)[n@?DET.PLqwa?lqw@?l@??lt-?-an-aspeak-CAUS-1SG.ERG-DET?xw?pm@x.]ShuswapIt was the old women who were the Shuswaps that I spoke to.(Lillooet, Davis et al. (2004, 104, fn6))393While the bare CP analysis of Lillooet cleft residues is supported by data like (53-54a,b), it is seemingly refuted by data like (54c).The headed relative in brackets in (54a) is termed a postposed relative clause(i.e DET [NP CLAUSE]) in Davis (2010a, 4), whereas the headed relative in (54b)is termed a prenominal relative clause (i.e DET [CLAUSE NP]). The main thesis ofDavis (2010a) argues that all types of relative clauses in Lillooet are derived froma common prenominal core. Specifically, ?Postposed relatives must be derivedfrom prenominal relatives via extraposition of a residue clause [a TP] containingthe trace of the relativized DP? (p.36, cf. derivation on page 24-25). It is possiblethat extraposition is banned in Lillooet cleft residues, though not in normal rela-tivization contexts, because clefts do not provide a necessary adjunction site for theextraposed clause.Although Okanagan clefts are introduced with demonstratives, and Lillooetclefts with clefting predicate ni?, it is worthwhile to note that in Lillooet clefts in-volving independent pronouns, the clefting predicate ni? is only optionally spelledout, as in (55a,b) (Thoma, 2007).28 This is reminiscent of the fact that Okana-gan ix?P is optionally present in equatives. Before an assertion-of-existence DPhowever, ni? is obligatory (55c).(55) a. ni?COPsn?wa2SG.INDEPtiDETnu ?kw-Pan-?n-a.help-DIR-1SG.ERG-EXISIt is you who I helped. (Lillooet, Thoma (2007, ex.6a))b. (ni?)COPsni?3SG.INDEPtiDETPa ?cx?@n-t?li-hasee-DIR-NTOP-DETkwDETs-John.NOM-JohnShe saw John. (Lit. The one who saw John is she.)(Lillooet, Thoma (2007, ex.7a))c. *(ni?)*(COP)tiDETp?P ?yaxw-amouse-EXIStiDET???k-ago-EXISk?tiP.DEICThe mouse is going along. (Lillooet, Thoma (2007, ex.7c))Lillooet clearly allows DPs in residue position for DP-DP structures, and overtly28Henry Davis (p.c., 2013) has informed me that clefted proper names also regularly occur withoutan introductory ni?.394headed residues in clefts, with restrictions. An equative analysis of these structuresseems possible for Lillooet, though given that assertion-of-existence DPs in Lil-looet do not permit individual concept readings (Demirdache, 1996; Matthewson,1998), unlike Okanagan iP, the prediction is that Lillooet clefts should not displayany semantic asymmetry. Then again, Lillooet assertion-of-existence DPs are usedin contexts for which generics are used in English (Lisa Matthewson, p.c.), andso under an analysis of generic interpretations of Salish DPs as involving maxi-mal intensional pluralities (Chierchia, 1998), it is possible that there is some othermechanism in Lillooet which blocks intensional readings in contextually restrictedor non-maximal contexts. An intensionally-based asymmetric account of DP-DPstructures in Lillooet, accompanied by focus-alignment constraints, could derivethe fact that there are word order restrictions in Lillooet DP-DP structures, (49,50),just as in Okanagan.As far as explaining the alternation between DP versus NP residues (e.g. 49a,b),Davis (2010c, ch.6, 10) seems to indicate that there is no semantic difference be-tween examples like (49a) and (49b), and that it may be simply a strong preferencefor speakers to drop the determiner in this environment. Hence, (49a) may beanalyzable as an equative, just as (49b). If not, then Lillooet ni? may be semanti-cally ambiguous between a predicational and equative copula, similar to the caseof Thompson ?ce.29SummaryShuswap is much more similar to Okanagan in terms of its DP-DP structuresand clefts than any other Salish language, even more so than any of the South-ern Interior languages. Although I have not been able to locate a clearly equativeDP-DP structure for Shuswap, consisting of two simple determiner headed-DPswith NP complements, given the morphosyntactic parallels between Okanagan andShuswap, it would be surprising if this were not possible. Since cleft residues areintroduced by determiners in Shuswap, and there is some data to indicate that thesemight possibly be headed residues, I suggest that Shuswap DP-DP structures andclefts may be analyzed as equatives, similarly to Okanagan. Shuswap requires29Henry Davis (p.c.) notes that the Lillooet imperfective auxiliary waP is gradually losing itsaspectual force and assuming the role of a copula.395further work in terms of ascertaining the level of predicate-argument word orderflexibility, and it is also unclear what the semantics of Shuswap determiners are(though cf. Gardiner (1996) for a preliminary analysis), and whether or not di-rect predications may be pragmatically equivalent to DP-DP structures in someinstances.Thompson appears to be quite strict in terms of disallowing headed residueclauses in clefts (Koch, 2008a, 2009), and given the absence of the oblique markert and remote determiner ? before cleft residue clauses, there is good evidence thatThompson residues are bare CPs. Nevertheless, Thompson also exhibits DP-DPstructures, introduced by the copula ?ce. Through the information structural andsyntactic status of these examples is unclear, there is reasonably good evidencethat these might be analyzed as equatives. If so, then ?ce may be semanticallyambiguous between an equative and a predicational copula.Lillooet is similar to Thompson in the sense that there is evidence that thecopula ni? is semantically ambiguous, however for Lillooet, it may be the DP-DPstructures which are ambiguously predicational, rather than the clefts, since it isunclear whether determiners are acceptable in identificational sentence residues.Assuming that data showing that cleft residues may be headed is robust, then forall intensive purposes, they may be analyzed as equatives.Table 9.5: Evidence for Equative versus Predicational Analyses of NorthernInterior Salish DP-DP Structures and CleftsEquative PredicationalShuswap X *Thompson (X) XLillooet X (X)9.3.3 Implications for Other Salish LanguagesThis section discusses data from two other Salish languages: Northern Straits, aCentral Salish language, and Bella Coola, which forms its own sub-branch of theSalish family. Ideally, this survey should include data from other Central Salish396languages, the Tsamosan branch, and Tillamook, however for reasons of space,this endeavor must wait. I begin with a discussion of Shank (2003) who developsan analysis of the copula ni? in Northern Straits Salish, before embarking on a moredata-centered discussion of Bella Coola.Northern StraitsShank (2003) develops a semantic analysis of the copula ni? in Northern StraitsSalish. His analysis treats the copula as of type <,>, essentially a predi-cate modifier. Syntactically, the copula takes a CP as its first argument, and a DPsubject as its second argument. This analysis is similar to Koch?s (2009) analysis ofThompson clefting predicate ?ce, except with the order of the semantic argumentsreversed.Interestingly, Shank considers an equative analysis (56) of the Northern Straitscopula as well, based on data like (57) where it seems that two DPs are beingequated with one another. The determiner which introduces the final DP is optional.(56) [[ni?]] = ?x?y.x = y(57) a. n??3SG.PREDkws@DETRichardRichard(kws@)(DET)l@pl?t..priestRichard is the priest. (Northern Straits, Shank (2003, p. 218, ex. 5a))b. n??3SG.PREDkws@DETJamesJames(kws@)(DET)Jimmy..JimmyJames is Jimmy. (Northern Straits, Shank (2003, p. 220, ex. 11b))Cleft residues in Northern Straits may also optionally be introduced by a deter-miner (58), which on the surface at least, makes them amenable to the equativeanalysis.(58) n??3SG.PREDkws@DETRichardRichard(kws@)DET?ts-@tbreak-TRkws@DETl?Psn.plateIt?s Richard that broke a plate.(Northern Straits, Shank (2003, p. 219, ex. 8a))397Since there is a preference for the residue, nominal or otherwise, to not be intro-duced by a determiner, Shank (2003) argues that residues are relative clause CPs,rather than DPs, and therefore that the equative analysis (56) is incorrect. Based oncomparative evidence with closely related Saanich, where the determiner kws@ isisomorphic with a complementizer and optional before relative clauses (Montler,1993), Shank argues that the optional determiners for similar cases in NorthernStraits are also complementizers.Davis et al. (2004, 103) provide one example showing that cleft residues inNorthern Straits may not contain an overt NP, which fits with an analysis of theclause introducing particles as being complementizers, similar to Thompson.(59) *ni?COPkws@DETxw@n?t@mwhite.personkws@DETs??niPwomanleN-n-@n.see-TR-1SG.ERGIt was a white person that was the girl that I saw.(Northern Straits, Davis et al. (2004, p.108, ex.9))Semantically speaking, exhaustivity is only an implicature for Northern Straitsclefts. Shank (2003, 227) incorporates exhaustivity as part of his final semanticanalysis of nilh (60). Applied to an example like (58), the sentence asserts that?Richard is a subpart of the maximal individual who broke a plate.? Because of ascalar implicature by which any stronger alternative that is not asserted is ruled outRooth (1992), speakers tend to interpret the individual as a non-proper subpart ofa maximal individual.(60) [[ni?]] = ?P?x.[x? y]? y = ?z.PzAlthough this analysis captures the exhaustivity effects, Shank notes that the iotaoperator in (60) predicts that there will be a presupposition of existence associatedwith Northern Straits clefts, but this does not seem to be the case, given that theycan be used in out-of-the-blue circumstances (cf. Davis et al. (2004)). Assumingthat determiners in Northern Straits are non-presuppositional, exhaustivity effectsmight instead derive from the determiner rather than from copula/clefting predi-cate, as I have claimed is the case for Okanagan, and as originally considered byShank (2003).398Northern Straits ni? also functions as an independent pronoun: in argumentpositions, it must be preceded by a determiner, while in non-argument positionsit is not. Shank states that ?Another benefit of this analysis which was lackingwith the identificational copula analysis is that the pronominal nature of the predi-cate is transparent. There is a free variable (the variable y in [60] which is free toreceive its reference from context, as pronouns are? (p. 228). Any benefit of hav-ing a contextually-sensitve variable as part of the meaning of the Northern Straitscopula might as well be achieved by analyzing the Northern Straits copula as equa-tive and its determiners as being contextually sensitive, as in Okanagan, thoughthis approach will not transparently capture the other use of ni? as an independentpronoun.Northern Straits ni? is different than Lillooet ni?, since the latter is not used asan independent pronoun (though cf. Thoma (2007) for discussion), but the two lan-guages are similar in the sense that residue determiners are optional in at least somecontexts. If the generalization that Northern Straits does not allow headed residuesis robust, clefts in Northern Straits cannot be equative, which may be another dif-ference between the two languages. Northern Straits does have relatively clearcases of DP-DP structures, and so it is possible that ni? is semantically ambiguous,as in Lillooet. Alternatively, if simple nouns can function as relative clauses incases like (57) (cf. discussion for Okanagan in section 8.8), then the copula maybe unambiguously predicational. This is within the realm of possibility, consider-ing that in Lillooet identificational contexts at least, nouns may be preceded by anauxiliary predicate waP, and (57b) does seem to be used as an identificational.Bella CoolaBella Coola (a.k.a. Nuxalk) at first glance exhibits strikingly clear examples ofDP-DP structures (61a).30(61) a. ti-PimlkPROX.MASC-manti-staltmx-tx.PROX.MASC-chief-DET.DEF.MASCThe chief is a man. (Bella Coola, Davis and Saunders (1997a, 111))30Glosses are as appear in Davis and Saunders (1997a), supplemented by Hank Nater (p.c.) andmyself.399b. Pimlk-\u000Bman-heti-nusPu?lX-tx.PROX.MASC-thief-DET.DEF.MASCThe thief is a man. (Bella Coola, Davis and Saunders (1997a, 27))Despite having a prefix ti, which appears to be a determiner similar in form toLillooet ti...a31, the initial constituent in (61a) has been identified by Hank Nater(p.c.) as a predicate, analogous to the initial constituent in (61b) which is moreclearly a case of direct predication. As such, the prefix ti may be analyzed as a non-saturating deictic prefix. The suffix -tx on the second constituent of (61a-b) marksa ?definite?-subject in intransitive contexts (Hank Nater, p.c.), and as a subject, ispresumably a DP. I gloss -tx as a determiner for the purposes of this discussion.(61a,b) are both analyzable as direct predications, although the predicate in (61a)has an additional deictic specification.Note that conferring subject DP status on both constituents by adding -tx, asin (62), is ungrammatical, which implies that DP-DP structures are not in factpossible.(62) *ti-Pimlk-txPROX.MASC-man-DET.DEF.MASCti-staltmx-tx.PROX.MASC-chief-DET.DEF.MASCThe chief is a man. (Bella Coola, Nater (p.c.))?Definite? subject marking is apparently not absolutely required in these con-texts, but is strongly preferred, as indicated by (63b,c) (Davis and Saunders, 1997b,226-227), discussed in terms of their appropriateness as answers to (63a). This in-dicates that overt determiners are strongly preferred for subject arguments.32 Thesecond translation in (63c) indicates that the initial subject DP is interpreted asa cleft focus, and the initial translation suggests that exhaustivity in Bella Coolaclefts is only an implicature.31Davis and Saunders (1997a) do not gloss the ti prefixes at all. In Davis and Saunders (1997b,226), they refer to these deictic prefixes as involving the ?semantics of Particularization?. As such,it is possible that they are ?determiners?. Bella Coola determiners encode both gender and deicticdistinctions (Davis and Saunders, 1997b, 226-227), which I render in the gloss line.32Or alternatively, if the deictic prefixes are in fact determiners, then the generalization may bethat ?predicate? DPs are only licensed in the context of a subject DP (i.e. an equative context, cf.61a), but there is no subject in (63b).400(63) a. wa-\u000B-kswho-he-INDIVIDUATIVEti- ?kx-ct.PROX.MASC-see-he/youWho saw you?b. #ci-xnasPROX.FEM-womanci- ?kx-cs.PROX.FEM-see-she/meA woman saw me.(Bella Coola, Davis and Saunders (1997b, 226, ex.18a))c. ci-xnas-cxPROX.FEM-woman-DET.DEF.FEMci- ?kx-cs.PROX.FEM-see-she/meThe woman saw me.The one who saw me was the woman.(Bella Coola, Davis and Saunders (1997b, 227, ex.18c))The difference between clefts (63c) and direct predications (61) is that the subjectDP can never precede the predicate (64a-b) in a direct predication, even in topical-ization contexts, since aside from clefts, Bella Coola is a strictly predicate-initiallanguage (Hank Nater, p.c.).(64) a. *ti-staltmx-txPROX.MASC-chief-DET.DEF.MASCti-Pimlk.PROX.MASC-manThe chief is a man. (Bella Coola, Nater (p.c.))b. *ti-nusPu?lX-txPROX.MASC-thief-DET.DEF.MASCPimlk-\u000B.man-heThe thief is a man. (Bella Coola, Nater (p.c.))Davis and Saunders (1997b, 228) provide examples that they suggest are in-terpretable both as isolated relative clauses, and as complete sentences. Note thatunlike unambiguous cases of clefting (63c), definite ?subject? marking occurs onthe final constituent, never on the initial NP (as far as I can tell).401(65) a. ti-wa ?cPROX.MASC-dogti- ??ikm-tx.PROX.MASC-run-DET.DEF.MASCThe one that?s running is a dog.the dog which is running(Bella Coola, Davis and Saunders (1997b, 225, ex.12))b. ti-PimlkPROX.MASC-manti-ksnmak-tx.DET.PROX.MASC-work-DET.DEF.MASCThe man is the one working.the man who is working(Bella Coola, Davis and Saunders (1997b, 228, ex.19a))Definite marking for unambiguous cases of relativization always occurs either onthe relative clause predicate itself (66a), or else on a clause-internal DP (subject,or object in transitive contexts) (66b). (66c) indicates that nominals may functionas relative clauses in Bella Coola, as in Okanagan.33(66) a. ya-\u000Bgood-heti-PimlkPROX.MASC-manti-ksnmak-tx.PROX.MASC-work-DET.DEF.MASCThe man who is working is good.(Bella Coola, Davis and Saunders (1997b, 228, ex.19b))b. ti-nusPu:lXPROX.MASC-thiefti- ?kx-isPROX.MASC-see-he/himti-Pa ?qwli:kw-tx.PROX.MASC-policeman-DET.DEF.MASCthe thief whom the policeman saw.(Bella Coola, Davis and Saunders (1997b, 229, ex.25))c. Pa?nap-i?know-we/himti-staltmxPROX.MASC-chiefti-Pimlk-tx.PROX.MASC-man-DET.DEF.MASCWe know the man who is chief.(Bella Coola, Davis and Saunders (1997a, 108))From these data, we can infer that an initial subject, marked by -tx (or other33Concerning (66c), Hank Nater (p.c.) notes that the verbal root nap means ?to know something?,while ?kyuk means ?to know somebody?, and that as such, the latter should replace the former in thisexample.402deictic suffix within the same paradigm), indicates a cleft structure, but that directpredications are strictly subject final. The complete sentence interpretations of (65)may be analyzed as direct predications, although it is unclear why a definite ?deter-miner? -tx does not occur on the clausal head for the relative clause interpretationsof (65), but does occur on the relative clause head for (66c), though it may be arestriction on definite-marked head-initial relative clauses, which might otherwisebe interpreted as clefts.Hank Nater (p.c.) notes that for data like (61a), there is a null allomorph of theintroductory particle tix, glossed as a male or neutral gender ?identifier?, a cleftingpredicate in other words (67a).34 One might think that Bella Coola tix is equivalentto Okanagan ix?P, especially on the basis of data like (67b), however Nater (p.c.)considers tix to be a predicate, and more crucially, tix is not a deictic particle.35(67) a. (tix)COPti-PimlkPROX.MASC-manti-staltmx-tx.PROX.MASC-chief-DET.DEF.MASCThe chief is a man. (Bella Coola, Nater, (p.c.))It is a man who is the chief.b. tix-\u000BCOP-heti-nusPu?lX-tx.PROX.MASC-thief-DET.DEF.MASCHe?s the thief. (Bella Coola, Davis and Saunders (1997a, 114))Comparing (68b) with (68c), both of which appear to be clefts, we see thatcopula tix is optional here as well.36,3734Other identifiers, or copulas, include female gender cix and mass wix.35Like Okanagan ix?P in DP-DP structures, tix cannot occur finally (i), but this may be due to thestrictly predicate-initial ordering:(i) *ti-nusPu?lX-txPROX.MASC-thief-DET.DEF.MASC(tix-\u000B).be he-heHe?s the thief. (Bella Coola, Nater, p.c.)36Davis and Saunders (1997a) do not include a -tx suffix on ti-mna in (68c). Nater (p.c.) statesthat the constitiuent ti-mna ?-son? in (68c) should be suffixed by -tx in order to make it definite, andso I add it here. This makes sense given that ?son? in (34) has the definite suffix, and implies thatcleft foci are ?definite? DPs in Bella Coola.37Nater (p.c.) also notes that the constituent ti-Paya? in (68a), possibly a headless relative DP,should include the future morpheme ka, i.e. ti-ka-Paya? given that the question is translated in the403(68) a. wa-\u000B-kswho-he-INDIVIDUATIVEti-Paya?PROX.MASC-walkPu?-cumu???LOC-Cumu??Who?s going to walk to Cumu???(Bella Coola, Davis and Saunders (1997a, 108))b. (tix)(COP)ti-yaPROX.MASC-goodti-mna-txPROX.MASC-son-DET.DEF.MASCti-Paya?.PROX.MASC-walkIt?s the good son who?s walking.(Bella Coola, Davis and Saunders (1997a, 113))c. tix-\u000B-kwCOP-he-QUOTti-yaPROX.MASC-goodti-mna-txPROX.MASC-son-DET.DEF.MASCti-Paya?.PROX.MASC-walkIt is, I?m told, the good son who is going.(Bella Coola, Davis and Saunders (1997a, 108))In sum, (67) and (68) together show that the same copula can be used with bothdirect predications and clefts, with the major difference between the two typesof structures being that clefts are subject-initial, whereas direct predications arestrictly subject-final. (67b) additionally suggests that the copula tix may select fora null pronoun as an argument. The data indicate that equatives may not be possiblewith Bella Coola, although more research is required to confirm this hypothesis.9.3.4 Summary of Implications for SalishAlthough there are crucial data missing for many of the languages surveyed in thissection, and a significant number of other Salish languages are not represented atall in this survey, there are nevertheless some interesting preliminary generaliza-tions to be made concerning DP-DP structures and clefts across Salish. There arealso implications for the historical development of DP-DP structures across thefamily, which raise a huge number of questions for further empirical work on thesefuture tense. Likewise for (68c). Sense tense is non-crucial to this investigation, I retain the originalforms and glosses in this case.404languages, especially in the Southern Interior.GeneralizationsI begin with a summary of the major patterns established by this survey (Table 9.6):Table 9.6: Evidence for Equative versus Predicational Analyses of DP-DPStructures and Clefts across Select Salish LanguagesEquative Predicational Copula ?CleftingDEM?Okanagan X * \u000B (ix?P)Kalispel (Montana Salish) X * \u000B (i?e / ?e ?y)Coeur d?Alene * X \u000B (ciP[?])Moses-Columbian * X \u000B (Pax?P)Shuswap X * \u000B (yGiP)Thompson (X) X ?ce xePLillooet X (X) ni? \u000BNorthern Straits (X) X ni? \u000BBella Coola * X (tix) \u000BThe first generalization is that the more western Northern Interior Salish lan-guages, as well as Northern Straits and Bella Coola use a copula (i.e. a cleftingpredicate) rather than a demonstrative to signal structural focus. Thompson usesboth a copula and a demonstrative, while the more eastern Northern Interior Sal-ish language Shuswap, and the entirety of the Southern Interior do not use overtcopulas, but optionally signal structural focus with a demonstrative.The second generalization is that languages which do not have overt copulasgenerally show clearer evidence in favor of either having equative or predicationalDP-DP structures and clefts (excepting possibly Bella Coola). Languages whichhave a dedicated, overt copula use them for both DP-DP structures and clefts,though clefts cannot in certain cases be reduced to an equative analysis.The third generalization is that there seems to be a slightly stronger tendencyfor languages which do not use overt copulas to have structures which may be405strongly argued to be equative (i.e. Okanagan, Kalispel, and Shuswap which usedemonstratives in DP-DP structures and clefts, versus Lillooet which uses an overtcopula).Concerning the Possible Historical Development of EquativesIn light of these generalizations, it seems historically plausible that equatives maybe an innovation of Southern Interior Salish (or possibly Shuswap), attributibleto the fact that these languages use optional demonstratives rather than dedicatedclefting predicates in clefts and DP-DP structures.To explain, demonstratives in Okanagan were shown to form constituents withiP DPs, and oblique-marked NPs were shown to function as argument expressionswhen they are associated with a demonstrative. Although the internal constituencyof demonstrative-associated oblique-marked NPs is unclear, the point is that thesyntactic status of a particle as being clearly a determiner is not necessarily rele-vant to its immediate constituent being able to function as a DP argument, whenthe immediate constituent is associated with a demonstrative. For languages wherea demonstrative may be argued to form a constituent with a residue clause (Okana-gan, Shuswap), regardless of whether a determiner or complementizer (or obliquemarker) follows that demonstrative, the entire constituent may have the distributionof an (internally-complex) DP argument.In this sense, demonstratives may have played a role in complementizers be-ing reanalyzed as determiners in languages like Kalispel, for instance. That is,since Kalispel does not have an overt copula, and demonstratives are used to signalstructural focus, assuming that they form constituents with residue clauses and/ormay adjoin to a headed or headless relative clause, a residue-introducing com-plementizer *?u(P) (cf. Moses-Columbian where it is a complementizer) may havebeen reanalyzed as a determiner, assuming that the entire demonstrative-introducedclause is interpretable as a headless relative clause argument. C-to-D reanalysis didnot occur for Coeur d?Alene and Moses-Columbian, and cleft residues remain un-ambiguously CPs. I tentatively suggest the following historical derivation:(69) Possible Development of Equative Clefts in Southern Interior Salish (andShuswap)406a. Stage 1: copula + DP focus + \u000B CP residue(Proto-Interior Salish #1)b. Stage 2: copula + DP focus + (DEM) CP residue(Proto-Interior Salish #2, Proto-Thompson, Proto-Lillooet)c. Stage 3: \u000B + DP focus + (DEM) CP residue(Proto-Interior Salish #3, except Proto-Thompson and Proto-Lillooet)d. Stage 4: \u000B + DP focus + (DEM) CP residue(Coeur d?Alene, Moses-Columbian: no C-to-D reanalysis)e. Stage 5: \u000B + DP focus + (DEM) DP residue(Okanagan, Kalispel, Shuswap: C-to-D reanalysis)For languages which use overt copulas and introduce their cleft residues withcomplementizers (e.g. Thompson), there is no possibility of analyzing the copulaas a DP-constituent with a residue (synchronically speaking, at least), and so notendency to reanalyze residue-introducing complementizers as determiners. Thisis likely the older pattern, which Coeur d?Alene and Moses-Columbian retained,despite the loss of an overt copula.38Possible Semantic Motivation for C-to-D ReanalysisThere may also have been a semantic motivation for reanalyzing complementizersas determiners in some languages. Consider again for a moment the predicationalanalysis of the Northern Straits copula developed in Shank (2003):(70) [[ni?]] = ?P?x.[x? y]? y = ?z.PzHere, exhaustivity arises from a scalar implicature by which any stronger alterna-tive that is not asserted is ruled out Rooth (1992), and so speakers tend to interpretthe individual as a non-proper subpart of a maximal individual. Given this seman-38It is unclear how languages which use overt copulas and introduce their cleft residues withdeterminers (e.g. Lillooet, possibly Northern Straits) fit into this overall picture, since there is noclear motivation for re-analyzing complementizers as determiners in these languages. Given thatLillooet ni? was borrowed from Central Salish, a detailed analysis of Central Salish clefts is necessarybefore anything less speculative can be said.407tic analysis, there is no need for a residue-introducing determiner to contribute anexhaustivity implicature, as under the Percus/Hedberg-style analysis of Okanagan.In Southern Interior Salish and Shuswap, however, there are no overt copulas, anda demonstrative cannot be assumed to carry exhaustivity as a conventional impli-cature, since it also occurs in non-cleft environments. Exhaustivity implicaturesin Southern Interior Salish and Shuswap might nevertheless be introduced by de-terminers, as they are in Okanagan. Two strong possibilities that emerge are that(i) complementizers were reanalyzed as determiners in Okanagan, Kalispel, andShuswap, because there was no overt copula, but nevertheless a need for an ex-haustivity implicature; or (ii) an overt copula was dropped because complementiz-ers were reanalyzed as determiners, and the exhaustivity implicature carried by thecopula became redundant (Lillooet is a potential candidate for such a process, andni? is already optional before independent pronouns and possibly proper names).39A Unified Analysis of Salish Clefts?The question arises as to whether it may be possible to find a unified semantic andsyntactic analysis of DP-DP structures and clefts across Salish. One possible an-swer comes to mind:40 If we were to assume that for Salish languages with overtcopulas, the copula is the spell-out of a determiner which selects directly for aCP (Hedberg, 2000), then we can explain the absence of (or difficulty in obtain-ing) overt NP heads in residue clauses in languages like Thompson and NorthernStraits, and analyze clefts in these languages as equative. Shank (2003) discussesthat ni? is also used as a third person pronominal in Northern Straits, and so by39On the subject of demonstratives as markers of structural focus, Diessel (1999, 148-149), citinga study by Luo (1997), discusses the fact that focus markers in many languages share the same mor-phological form as copulas and demonstratives. Diessel (1999, 148) states that ?it is conceivable thatfocus markers may also develop directly from identificational demonstratives in nonverbal clauses.?(i) [[DEM \u000B NP S] [REL CL S]S]? [[FOC NP NP] : S] (Diessel, 1999, 148)Diessel also states that ?.. the focal part of the cleft construction includes an identificational demon-strative in a nonverbal clause..... [this] may account for the development of focus markers in lan-guages such as Ambulas and Mokilese, where demonstratives and focus markers are morphologicallyrelated but show no obvious relationship to a copula.? It is therefore possible that demonstratives inthe Southern Interior are developing into dedicated clefting predicates.40Thanks to Henry Davis (p.c. 2013) for suggesting this as a possibility.408analyzing pronouns as determiners (Postal, 1966; Elbourne, 2001), the Hedberganalysis might receive independent support. Allowing for a determiner to selectfor an overt or covert NP head (Percus, 1997) in Salish languages without overtcopulas, then it may be possible to subsume predicational clefts as a type of equa-tive. The parameter of variation, then, reduces to the selectional restrictions of alanguage?s determiners.Intensionality in Salish EquativesAnother question worth considering is whether equative structures across Salish in-volve the same intensional asymmetry which I have claimed holds for Okanagan.Answering this question will involve not only conducting detailed investigations ofdeterminer semantics in languages which appear to have equatives (esp. Kalispeland Shuswap), but also comparing syntactic and information structural propertiesof equatives with direct predications in these languages. In Lillooet for example,as I briefly mentioned, assertion-of-existence DPs permit generic readings, but notindividual concept readings. There could be microvariation between languagesin terms of allowing contextually restricted individual concepts, but much morework needs to be done before anything can be said for certain. A detailed com-parative study of determiner semantics and the syntactic and information structuralproperties of equatives in Okanagan, Kalispel, Shuswap, and Lillooet, could domuch to clarify the nature of copular predication in Salish. The results could thenbe compared to a study of determiner semantics in languages without equatives(Coeur d?Alene and Moses-Columbian), in order to test whether there might beany necessary correlation between determiner properties such as intensionality ormaximality, and the existence of equatives. I leave this for future work.9.4 Theoretical ImplicationsThis section summarizes the major implications of my analysis for the theory. First,I discuss how Okanagan supports a fundamental distinction between predicationaland equative clauses. Next, I present my classification of Okanagan non-verbalpredications within Higgins? taxonomy.4099.4.1 A Distinction between Predication and EquationOkanagan supports (Heycock and Kroch, 1999) who argue for a distinction be-tween predicational and equative small clauses and copular clauses. DP-DP equa-tives are not reducible to direct predications in Okanagan, contra some theories ofEnglish copular clauses Moro (1997); Adger and Ramchand (2003); den Dikken(2006), since Okanagan DPs, like DPs in other Salish languages, cannot be pred-icative (Longobardi, 1994; Matthewson, 1998). The distinction between predica-tion and equation therefore corresponds to the syntactic and semantic distinctionsbetween NP and DP: Okanagan NPs may directly predicate themselves of a DPargument, whereas Okanagan DPs may not. Assuming that raising to type is not a possibility for Okanagan DPs (which seems apparent given the word orderrestriction), and via the principle of compositionality, a null equative copula mustexist for Okanagan DP-DP structures. I have argued for an intensionality-basedasymmetry between the two DPs in a DP-DP equative structure (Romero, 2005),based on independent evidence that Okanagan iP DPs may denote individual con-cepts. My semantic analysis of the copula is given as follows:(71) Okanagan equative copula, final analysis[[=]] = ?x?ye?ws[x(w) = y]The distinction between predicational and equative sentences is motivated bya word order restriction that is manifest for DP-DP structures in answer to WH-questions, which is not apparent for a corresponding direct predication, and by thefact that syntactic inversion is not a possibility (Moro, 1997; den Dikken, 2006)).This word order restriction results from the following lexical property of the equa-tive head:(72) The equative head assigns a syntactic feature ?F? to its second argument.This feature is interpretable as focus (Rooth, 1992), and alignment constraints forcethe DP specifier of the equative clause to occur to the left (Koch, 2008a). Thedistinction between predication and equation in Okanagan is therefore not only asemantic distinction, but also an information structural one.4109.4.2 Higgins? Taxonomic Classification for OkanaganFor Okanagan, since specificational sentences are not possible, and identification-als pattern like other instances of either direct predication or equation, Okanagansupports reducing Higgins? taxonomy to only two types, predicational and equative(Heller, 2005). The taxonomy I propose for Okanagan is shown below in Table 9.7(cf. Table 7.3).411Table 9.7: Higgins? Taxomony and Okanagan Non-Verbal PredicationsCLAUSE TYPE EXAMPLE SUBJECT COMPLEMENTPredicational Wilford s@xw ?kw? ?l?xw@m e s@xw ?kw? ?l?xw@m Wilford(Wilford is a carpenter)(Identificational) Spike ?C@lx? ?caP?C@lx? ?caP Spike(Spike is Chillhitzia)John ix?P iP s@xw ?kw ?lm?n@mix?P iP s@xw ?kw ?lm?n@m John(That hard worker is John)ix?P JohnJohn ix?P(That?s John)Equative Wilford iP s@xw ?kw? ?l?xw@m e *iP s@xw ?kw? ?l?xw@m Wilford(Wilford is a/the carpenter)(Identificational ix?P John [.... (residue)](truncated cleft)) John ix?P [.... (residue)](It?s John)(Identificational John ix?P [iP s@xw ?kw ?lm?n@m](full cleft)) *ix?P [iP s@xw ?kw ?lm?n@m] John(It?s John that is the hard worker.)The difference between Okanagan and languages like English with regards to412Higgins? taxonomy is in part due to differences in the semantic type of the DPsinvolved in non-verbal predications. The distribution of DPs according to seman-tic type in contexts involving non-verbal predication in Okanagan and English isshown below in Table 9.8.Table 9.8: Distribution of Semantic Types across DPs in Okanagan Non-Verbal PredicationsOkanagan e iP DPs X * Xbare demonstratives X * *demonstrative-associatedDPsX * Xproper names X X *English e DET DPs X X XRomero (2005)bare demonstratives X XMikkelsen (2005) XHeller & Wolter (2008)demonstrative phrases X * *proper names X XHeller (2005) *There are four major differences which I discuss here.First, there is good semantic and distributional evidence in Okanagan againstanalyzing demonstratives and iP DPs as predicates of type . The analogousexpressions in English are often assumed to be of type in certain copularenvironments: e.g. determiner-headed DPs may be analyzed as initial predicatesin specificational sentences (Moro, 1997; Mikkelsen, 2005; den Dikken, 2006),bare demonstratives may be analyzed as initial predicates in identificationals (asub-class of specificational for Mikkelsen (2005) e.g. That is Susan).Second, in non-verbal predication contexts, demonstrative-associated DPs inOkanagan are only of type e when they are the subject of an identificational predi-cation or equative, but they can be of type when they are in the complementposition of an equative (i.e. when they are cleft residues). In English, demon-413strative phrases have been argued to be of type e as subjects of identificationalequatives (Mikkelsen, 2005), as well as subjects of identificational predications(Heller, 2005). Adjoined demonstratives in Okanagan ?inherit? the intensionalityof the overt (or covert) iP DP to which they adjoin, as evidenced by data showingthat they allow generic and other intensional readings.41Third, there is good distributional evidence that proper names are not neces-sarily type e referential expressions in Okanagan. In equatives, they can only bethe second argument of the equative head, and must be of type e, however in pred-ications, they may either be type e subjects or predicative non-subject expressions.Since they pattern distributionally like other lexical predicates in identificationalcontexts, I suggest that they are of type , and are singleton-set-denoting prop-erties. For English, proper names are usually always analyzed as referential, al-though Heller (2005, 197) analyzes them as an ?essential property that is assumedto be new information? in the complement position of an identificational predica-tion.Fourth, neither bare demonstratives nor proper names can denote individualconcepts of type in Okanagan. Intensionality is a special property of iPDPs, although a demonstrative-associated iP DP may be of type when itis in the residue position of a cleft (i.e. functioning as the first argument of theequative copula), including cases where the demonstrative is adjoined to the nullresidue of a truncated cleft. Heller and Wolter (2008) allow proper names to be oftype ,t>>, that is, functions from worlds to sets of individual concepts.Since evidence for such a higher type is not immediately apparent for Okanagan, Iassume that proper names, like bare demonstratives, are directly referential, rigiddesignators (Kaplan, 1977, 1989; Kripke, 1982).In sum, from Table 9.8, we see that the most apparent semantic differencesbetween Okanagan and English DPs in copular environments are that Okanagan iPDPs cannot be of type , while English determiner-headed DPs can, and thatwhile English bare demonstratives have been analyzed as intensional (Heller andWolter, 2008), it is the Okanagan demonstratives in positions adjoined to an iP DPwhich I argue to be intensional, not bare demonstratives.41Though it remains an unsolved compositional problem how exactly this is achieved. See section4.6.1 and section 8.6.2 for discussion.4149.4.3 Pragmatic Differences between Okanagan and EnglishEquativesAnother major difference between Okanagan and English surfaces with regards tothe pragmatics of non-verbal predications. For English specificational sentences,there is commonly acknowledged to be an old information requirement on the ini-tial DP (Birner, 1996; Heycock and Kroch, 1999; Mikkelsen, 2005). Regardlessof whether one analyzes English specificationals as equative clauses or as invertedpredicational clauses, it is significant to note that there is no old information re-quirement on any DP in an Okanagan equative: the word order is strictly fixed,as required by the semantics of the equative copula (cf. section 7.5) coupled withfocus alignment constraints (Koch, 2008a). Although apparent cases of inversion(which I have suggested involve topicalization) in some cases involve an initialiP DP that constitutes old information (cf. section 7.7), this is not a requirement,as shown by the fact that these cases are also felicitous in non-presuppositional,out-of-the-blue contexts.9.4.4 Information Structural Differences between Okanagan andEnglish EquativesYet another major difference between Okanagan and English can be found with re-gards to information structure. English copular clauses are not hard-wired in termsof focus, by which I mean that it is not a lexical property of the copula be, or ofany Pred-head or Equative-head analyses of English, to assign a feature ?F? to oneof its arguments. In terms of focus-alignment facts in Okanagan, such an approachis motivated because DP-DP structures require the more-referential DP (with the?F? feature) to occur initially, and in terms of pragmatics, DP-DP structures arerequired in contrastive contexts (cf. section 7.4.1). In contrast, English specifica-tionals may be pragmatically and/or semantically asymmetrical (depending on thetheory one adopts), and the fixed information structure (whereby the final DP isinterpretable as a focus) is dependant on the satisfaction of a pragmatic conditionsuch that the initial DP represent old information. For Okanagan however, the fixedinformation structure is a direct result of the semantic and lexical properties of theequative head: that is, there is no pragmatic condition involved.4159.4.5 Exhaustivity in Okanagan EquativesOkanagan DP-DP equative structures carry an implicature of exhaustivity. Thefact that exhaustivity is only implied, and not entailed, means that the focused DPin an Okanagan DP-DP equative structure may be interpreted either exhaustivelyor non-exhaustively. The exhaustivity implicature is induced from a maximalityimplicature associated with the determiner iP in the equative head?s first argument.Both are given below:(73) Maximality implicature of iP:f = MAX(74) Exhaustivity Implicature:A sentence of the form [x DP] = [iP Y DP]a. Asserts:? f .x = f (Y )b. Via (73) this implicates:x = MAX(Y )c. And assuming (74b) is satisfied, a DP-DP structure asserts:x is the only YThat is, if all else is equal, an iP DP will be interpreted as denoting the maximalsingular or plural individual in the intersection of the context set and the NP set.In an equative sentence, the maximality implicature induces an exhaustivity impli-cature, such that the initial focused DP will be interpreted as the only individualequivalent to the maximal referent denoted by the iP DP.9.4.6 Interpretive Variability in Okanagan versus English EquativesSince exhaustivity in Okanagan is only an implicature, this leads to an interpretivevariability for DP-DP structures that is interesting from a cross-linguistic perspec-tive on copular predication: (pseudo-)predicational readings are made available viaan equative semantics. For cases where the exhaustivity implicature of a DP-DPstructure is not satisfied, the direct and DP-DP structure below will be functionally416equivalent:(75) a. MaryMarys@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDMary is a teacher.b. MaryMaryiPDETs@xw- ?maP- ?m?yaP-m.OCC-RED-teach-MIDMary is a teacher.The interpretive ambiguity in Okanagan equative DP-DP structures like (75b) de-pends not on semantic type, since the iP DP will always be of type , and theinitial subject of type e. It instead depends on whether or not an iP DP is interpretedas maximal or non-maximal.This relates to my earlier discussion pertaining to English sentences such asThe morning star is the evening star, which although canonically considered to beequative, may also have a predicational or specificational interpretation dependingon context. Under theories which analyze specificationals as inverted predication-als, the first or second DP may be of type , depending on the interpretation.For Heycock & Kroch, the specificational interpretation of The morning star isthe evening star must be semantically equative, with an additional ?old informa-tion? requirement on the initial DP. A non-specificational equative interpretationis presumably still available in contexts where the initial DP does not convey oldinformation. The point is that for both English specificationals, which are prag-matically (Heycock and Kroch, 1999) and information-structurally asymmetrical(Higgins, 1973), and Okanagan equatives, which are semantically and information-structurally asymmetrical, the interpretive variability of a given sentence may bederived without resorting to semantic type-shifting.9.4.7 Fixed Information Structure and the Connection to CleftsIt is instructive to compare Okanagan equatives to English inverse specificationalpseudoclefts such as (76a) (den Dikken et al., 2000) and clefts such as (76b).417(76) a. Otto Preminger was who I met.b. It was Otto Preminger who I met.Okanagan equatives have a fixed information structure similar to the sentences in(76) in terms of the fact that the initial DP is intepretable as a focus, but unlikeEnglish, there is no requirement that the the second constituent be old information,since there is no presupposition associated with Okanagan equatives. There is a re-quirement that the second constituent be intensional, however. As such, Okanagansentences involving a fixed information structure (i.e. equative DP-DP structuresand clefts) are defined as a class by the presence of an iP DP, and thus by inten-sionality. English sentences involving a fixed information structure such as (76)are defined as a class by an existential presupposition.These differences between English and Okanagan explain the relative prag-matic markedness of English sentences involving fixed information structure, andthe relative unmarkedness of the corresponding Okanagan structures.Okanagan clefts also carry an implicature of exhaustivity, exactly as do simplerOkanagan DP-DP equative structures. As such, and in tandem with independentmorphosyntactic evidence involving the oblique marker t (which I introduce inchapter 4), I claim that Okanagan clefts are also equative DP-DP structures, andpresent a derivational analysis of clefts as equative structures (cf. section 8.6).This analysis supports theories of English clefts which derive clefts from simplercopular predications (Percus, 1997), and relates the fixed information structure ofOkanagan DP-DP structures to the semantics of the iP determiner. Okanagan DP-DP structures may be viewed as a type of cleft, given their fixed information struc-ture.Percus (1997) derives English clefts from specificational copular clauses, not-ing that both structures are constrained by discourse in the same way. Under theargument that English specificationals are equative (Heycock and Kroch, 1999),and in light of the information structural parallels between Okanagan DP-DP equa-tives and clefts on the one hand and English specificational sentences and clefts onthe other (Percus, 1997), I suggest that clefts in both languages may be analyzedas equatives. Strictly speaking, Okanagan supports (Percus, 1997) over (Hedberg,2000), since the former in principle permits residue clauses to be headed. In ac-418tuality, headed residues in English should be ungrammatical under Percus? deriva-tion (cf. section 8.6), but this is due rather to the extraposition requirement onEnglish clefts, and because it is a spell-out of the plus a null head. Other Sal-ish languages, which introduce their cleft residues with complementizers and donot allow residue-internal NP heads, may ultimately support Hedberg (2000) (cf.section 9.3.4).9.4.8 ConclusionIn conclusion, my approach to Okanagan predication and equation supports a viewof Higgins? taxonomy as a taxonomy of ?syntax-semantics alignments which isfurther conditioned by information structure? (Mikkelsen, 2011, 1813), and shedsnew light on the debate between inverse predicational versus equative analysesof specificational sentences, how Higgins? taxonomy might be reducible, and thederivational relation between copular sentences with fixed information structureand clefts.419BibliographyAdger, D. and G. Ramchand (2003). Predication and Equation. LinguisticInquiry 34(3), 325?359. ? pages ii, 20, 30, 32, 217, 235, 410Akmajian, A. (1970). On deriving cleft sentences from pseudo-cleft sentences.Linguistic Inquiry 1, 149?168. ? pages 51Akmajian, A. (1979). Aspects of the Grammar of Focus in English. New York:Garland. ? pages 28, 29Baker, M. (2003). Lexical Categories: Verbs, Nouns, and Adjectives. New York:Cambridge University Press. ? pages ii, 3, 22, 113, 233, 235Baker, M. C. (1996). The Polysynthesis Parameter. Oxford Studies inComparative Syntax. New York: Oxford University Press. ? pages 217Baptiste, M. (2001). WH Questions in Okanagan. Master?s thesis, University ofBritish Columbia, Vancouver, BC. ? pages 65, 67, 69, 70, 131, 174Bar-el, L. A. (2006). Aspectual distinctions in Skwxw?7mesh. Ph. D. thesis,University of British Columbia. ? pages 229Barthmaier, P. (2002). Transitivity and lexical suffixes in Okanagan. In Papers forthe 37th Annual International Conference on Salish and NeighboringLanguages, Volume 9, pp. 1?17. UBCWPL. ? pages 92, 93Barthmaier, P. (2004). Intonation Units in Okanagan. In D. B. Gerdts andL. Matthewson (Eds.), Studies in Salish Linguistics: In Honor of M. DaleKinkade, Volume 17 of University of Montana Occasional Papers inLinguistics, pp. 30?42. ? pages 50, 268, 355Barwise, J. and R. Cooper (1981). Generalized quantifiers and natural language.Linguistics and Philosophy 4, 159?219. ? pages 125, 147420Birner, B. (1996). The Discourse Function of Inversion in English. New York:Garland. ? pages 29, 357, 415Birner, B., J. Kaplan, and G. Ward (2007). Functional compositionality and theinter- action of discourse constraints. Language 83, 323?349. ? pages 33Bittner, M. and K. Hale (1996). The structural determination of case andagreement. Linguistic Inquiry 27, 1?68. ? pages 99Boas, F. and J. Teit (1930). Coeur d?Alene, Flathead and Okanagon Indians.Washington, DC: United States Government Printing Office. ? pages 10, 15Bowers, J. (1993). The syntax of predication. Linguistic Inquiry 24, 591?656. ?pages 22, 34, 233Brisson, C. (1998). Distributivity, Maximality, and Floating Quantifiers. Ph. D.thesis, Rutgers University. ? pages 148Camp, K. M. (2007). An Interlinear Analysis of Seven Kalispel texts from HansVogt. Master?s thesis, University of Montana, Missoula, MT. ? pages 362,363, 365, 366, 367Carlson, G. (1977). Reference to Kinds in English. Ph. D. thesis, University ofMassachusetts, Amherst. ? pages 151, 176Carlson, G. (2006). The meaningful bounds of incorporation. In S. Vogeleer andL. Tasmowski (Eds.), Non-definiteness and plurality, pp. 35?50. JohnBenjamins. ? pages 122, 124, 136, 156, 162Chierchia, G. (1984). Topics in the Syntax and Semantics of Infinitives andGerunds. Ph. D. thesis, University of Massachusetts, Amherst. ? pages 24Chierchia, G. (1998). Reference to kinds across languages. Natural LanguageSemantics 6, 339?405. ? pages 4, 5, 80, 151, 153, 335, 351, 395Chomsky, N. (1971). Deep structure, surface structure, and semanticinterpretation. In D. Steinberg and L. Jakobovits (Eds.), Semantics: AnInterdisciplinary Reader in Philosophy, Linguistics, and Psychology, pp.183?216. New York: Cambridge University Press. ? pages 48Chomsky, N. (1975). The logical structure of linguistic theory. Chicago:University of Chicago Press. ? pages 36421Chomsky, N. (1977). On Wh-movement. In P. Culicover, T. Wasow, andA. Akmajian (Eds.), Formal Syntax, pp. 71?132. New York: Academic Press.? pages 51Chomsky, N. (1981). Lectures in Government and Binding, Volume 9 of Studiesin Generative Grammar. Dordrecht: Foris. ? pages 37, 40Chung, S. and W. Ladusaw (2004). Restriction and saturation. Cambridge, MA:MIT Press. ? pages 122, 124, 136, 156, 162, 163Cinque, G. (1999). Adverbs and functional heads: A cross-linguistic perspective.Oxford Studies in Comparative Syntax. New York: Oxford University Press. ?pages 226Comorovski, I. (2007). Constituent questions and the copula of specification. InI. Comorovski and K. von Heusinger (Eds.), Existence. Semantics and Syntax,pp. 49?77. Dordrecht: Springer. ? pages ii, 18, 21, 29, 35, 44, 216, 218, 246,248, 262, 351, 355Cooper, R. (1979). The Interpretation of Pronouns. In F. Heny and H. Schnelle(Eds.), Syntax and semantics, vol. 10. New York: Academic Press. ? pages127Czaykowska-Higgins, E. and M. D. Kinkade (1998). Salish Languages andLinguistics: Theoretical and Descriptive Perspectives. Trends in Linguistics:Studies and Monographs 107. Berlin: Morton de Gruyter. ? pages 234, 239Darnell, M. (1995). Preverbal nominals in Colville-Okanagan. In P. Downing andM. Noonan (Eds.), Word order in Discourse, pp. 85?104. Amsterdam andPhiladephia: John Benjamins. ? pages 65Davis, E. (1990). x??Qx?aQ s?t?mkaPs: The Story of Crow?s Daughter. manuscript.? pages 372, 374, 375Davis, H. (1994). Tali-Ho! In Papers for the 29th Annual InternationalConference on Salish and Neighboring Languages, pp. 117?144. ? pages 137Davis, H. (1997). Deep Unaccusativity and Zero Syntax in St??t?imcets. InA. Mendikoetxea and M. Uribe-Etxebarria (Eds.), Theoretical Issues at theMorphology-Syntax Interface (Supplements of the International Journal ofBasque Linguistics and Philology), pp. 55?96. Bilbao, Spain. ? pages 87, 162Davis, H. (1999a). On Nouns and Nominalization in Salish. manuscript. ? pages3, 217422Davis, H. (1999b). Word Order and Configurationality in St??t?imcets. In Papersfor the 34th Annual International Conference on Salish and NeighboringLanguages, pp. 61?83. ? pages 65, 75, 115Davis, H. (2002). Categorial restrictions on modification in St?at?imcets. InPapers for the 37th Annual International Conference on Salish andNeighboring Languages, Volume 7 of University of British Columbia WorkingPapers in Linguistics, pp. 61?75. UBCWPL. ? pages 16, 168, 207Davis, H. (2004). Locative relative clauses in St??t?imcets (Lillooet Salish). InJ. Brown and T. Peterson (Eds.), Papers for the 39th Annual InternationalConference on Salish and Neighboring Languages, Volume 14 of University ofBritish Columbia Working Papers in Linguistics, pp. 83?116. UBCWPL. ?pages 16, 18, 81, 168, 187, 201, 207, 354Davis, H. (2005). Constituency and Coordination in St??t?imcets (Lillooet Salish).In A. Carnie, S. A. Dooley, and H. Harley (Eds.), Verb First: on the Syntax ofVerb Initial Languages. Amsterdam: John Benjamins. ? pages 68, 69Davis, H. (2006). The Status of Condition C in St??t?imcets. In S. Bischoff,L. Butler, P. Norquest, and D. Siddiqi (Eds.), MIT Working Papers inEndangered Languages: Studies in Salishan, pp. 49?92. Cambridge, MA:MITWPL. ? pages 41, 114, 243, 245, 289Davis, H. (2007). VP ellipsis in St??t?imcets and its implications. manuscript. ?pages 71Davis, H. (2009). Cross-linguistic Variation in Anaphoric Dependencies:Evidence from the Pacific Northwest. Natural Language and LinguisticTheory 27(1), 1?43. ? pages 41, 114, 243, 245, 289Davis, H. (2010a). A Unified Analysis of Relative Clauses in St??t?imcets.Northwest Journal of Linguistics 4(1), 1?43. ? pages 16, 18, 81, 102, 168,187, 190, 196, 197, 201, 204, 207, 208, 212, 314, 354, 394Davis, H. (2010b). Salish Languages lack generalized quantifiers after all! InN. Li and D. Lutz (Eds.), Proceeding of Semantics and Linguistic Theory(SALT) 20. ? pages 148, 149Davis, H. (2010c). Teaching Grammar of St??t?imcets. manuscript. ? pages 97,278, 390, 391, 392, 395Davis, H. (2011). Stalking the Adjective in St??t?imcets. Northwest Journal ofLinguistics 5(2), 1?60. ? pages 78, 80, 81, 168, 179, 187, 358423Davis, H., S. Lai, and L. Matthewson (1997). Cedar Roots and SingingDetectives: Attributive Modification in Salish and English. In Papers for the32nd Annual International Conference on Salish and Neighboring Languages,pp. 291?316. ? pages 75, 81, 91, 168, 175, 176, 179Davis, H. and L. Matthewson (2003). Quasi-objects in St??t?imcets: on the(semi)independence of agreement and case. In A. Carnie, H. Harley, and M. A.Willie (Eds.), A Festschrift for Eloise Jelinek (Linguistik Aktuell 62), pp.80?106. Amsterdam: John Benjamins. ? pages 62, 85, 86, 111, 211Davis, H. and L. Matthewson (2009). Issues in Salish Syntax and Semantics.Language and Linguistic Compass 3/4, 1097?1166. ? pages 162Davis, H., L. Matthewson, and S. Shank (2004). Clefts vs. nominal predicates intwo Salish languages. In D. B. Gerdts and L. Matthewson (Eds.), Studies inSalish Linguistics in Honour of M. Dale Kinkade, pp. 100?117. University ofMontana Occasional Papers in Linguistics, vol. 10. ? pages ii, 3, 17, 18, 19,216, 225, 226, 249, 291, 293, 294, 297, 302, 304, 305, 306, 315, 342, 353, 355,377, 378, 382, 385, 392, 393, 398Davis, P. W. and R. Saunders (1997a). A Grammar of Bella Coola. Missoula, MT:University of Montana Occasional Papers in Linguistics, vol. 13. ? pages 399,400, 402, 403, 404Davis, P. W. and R. Saunders (1997b). The Bella Coola relative clause. InE. Czaykowska-Higgins and M. D. Kinkade (Eds.), Salish Languages andLinguistics: Theoretical and Descriptive Perspectives, Number 107 in Trendsin Linguistics: Studies and Monographs, pp. 219?234. Berlin: Mouton deGruyter. ? pages 400, 401, 402D?chaine, R.-M. and M. Tremblay (2011). Deriving Nominal Reference. InProceedings of WECOL 2011, Vancouver, BC. ? pages 166Delahunty, G. (1982). Topics in the syntax and semantics of English cleftsentences. Ph. D. thesis, University of California. ? pages 51Demirdache, H. (1996). The Chief of the United States Sentences in St??t?imcets(Lillooet Salish): A Cross-Linguistic Asymmetry in the TemporalInterpretation of Noun Phrases and Its Implications. In Papers for the 31stAnnual International Conference on Salish and Neighboring Languages, pp.79?100. ? pages 123, 135, 153, 351, 395424Demirdache, H. and L. Matthewson (1995). On the universality of syntacticcategories. In Proceedings of the 25th Annual Meeting of the North EasternLinguistic Society, pp. 79?94. Graduate Linguistic Student Association,University of Massachusetts, Amherst. ? pages 75, 81, 179den Dikken, M. (2006). Relators and Linkers: The Syntax of Predication,Predicate Inversion, and Copulas, Volume 47 of Linguistic InquiryMonographs. Cambridge, MA: MIT Press. ? pages 20, 22, 30, 34, 35, 38, 45,216, 233, 235, 240, 241, 287, 350, 410, 413den Dikken, M., A. Meinunger, and C. Wilder (2000). Pseudoclefts and ellipsis.Studia Linguistica 54, 41?89. ? pages 31, 40, 45, 339, 417Diesing, M. (1992). Indefinites. Cambridge, MA: MIT Press. ? pages 133Diessel, H. (1999). Demonstratives: Form, Function and Grammaticalization.Amsterdam: John Benjamins. ? pages 408Doak, I. G. (1983). The 1908 Okanagan word lists of James Teit. Missoula, MT:University of Montana Occasional Papers in Linguistics, vol. 3. ? pages xiv, 9Doak, I. G. (1997). Coeur d?Alene grammatical relations. Ph. D. thesis,University of Texas at Austin. ? pages 367, 368, 369, 370Elbourne, P. (2001). E-type Anaphora as NP-deletion. Natural LanguageSemantics 9, 241?288. ? pages 127, 409Emonds, J. (1976). A transformational approach to English syntax: root,structure- preserving, and local transformations. New York: Academic Press.? pages 51En?, M. (1981). Tense without Scope: An analysis of Nouns as Indexicals. Ph. D.thesis, University of Wisconsin, Madison. ? pages 153, 351En?, M. (1991). The semantics of specificity. Linguistic Inquiry 22, 1?25. ?pages 133Evans, G. (1980). Pronouns. Linguistic Inquiry 11, 337?362. ? pages 127Farkas, D. and H. de Swart (2004). The semantics of incorporation. Chicago:CSLI Publications. ? pages 122, 124, 162F?ry, C. (2013). Focus as prosodic alignment. Natural Language and LinguisticTheory 31, 683?734. ? pages 48425FPHLCC (2010). Report on the Status of B.C. First Nations Languages, 2010.Brentwood Bay, BC: First Peoples? Heritage, Language and Culture Council.? pages 8Gardiner, D. (1993). Structural Asymmetries and Preverbal Positions in Shuswap.Ph. D. thesis, Simon Fraser University, Burnaby, BC. ? pages 17, 65, 71, 151,228, 299, 377, 378, 379, 380, 381Gardiner, D. (1996). Determiner Phrases in Secwepmectsin (Shuswap). In Papersfor the 31st Annual International Conference on Salish and NeighboringLanguages, pp. 175?184. ? pages 99, 377, 378, 379, 396Geist, L. (2007). Predication and equation in copular sentences. Russian vs.English. In I. Comorovski and K. von Heusinger (Eds.), Existence. Syntax andSemantics, pp. 79?105. Dordrecht: Springer. ? pages 32, 248Gerdts, D. (1988). Object and Absolutive in Halkomelem Salish. New York:Garland. ? pages 115Gerdts, D. B. (2006). Argument Realization in Halkomelem: A Study in VerbClassification. In Proceedings of WSCLA XI, pp. 61?81. UBCWPL vol 19. ?pages 162Gerdts, D. B. and T. E. Hukari (2003). The expression of NPs in Halkomelemtexts. In Papers for the 38th Annual International Conference on Salish andNeighboring Languages, pp. 91?126. UBCWPL vol 11. ? pages 86, 137Gerdts, D. B. and T. E. Hukari (2004). Determiners and transitivity inHalkomelem texts. In D. B. Gerdts and L. Matthewson (Eds.), Studies in SalishLinguistics in Honour of M. Dale Kinkade, pp. 151?171. University of MontanaOccasional Papers in Linguistics, vol. 10. ? pages 67Gerdts, D. B. and T. E. Hukari (2012). A Closer Look at SalishIntransitive/Transitive Alternations. In Proceedings of the Thirty-SecondAnnual Meeting of the Berkeley Linguistics Society, pp. 503?514. University ofCalifornia, Berkeley. ? pages 162Gerdts, D. B. and K. Kiyosawa (2010). Salish Applicatives. Leiden: Brill. ?pages 92, 211Giannakidou, A. (2004). Domain restriction and the arguments of quantificationaldeterminers. In Proceedings of SALT XIV, Ithaca, NY, pp. 110?128. ? pages148426Gillon, C. (2006). The semantics of determiners: domain restriction inSkwx?7mesh. Ph. D. thesis, University of British Columbia, Vancouver, BC. ?pages 17, 18, 96, 118, 123, 125, 126, 129, 130, 139, 150, 156, 163, 164, 166,354, 365, 386Gillon, C. (2009a). Deictic features: evidence from Skwx?7mesh. InternationalJournal of American Linguistics 75, 1?27. ? pages 17, 18, 354Gillon, C. (2009b). The semantic core of determiners: Evidence fromSkwx?7mesh. In J. Ghomeshi, I. Paul, and M. Wiltschko (Eds.), Determiners:Universals and variation, pp. 177?214. Amsterdam: John Benjamins. ? pages17, 100Gundel, J. K. (1977). Where do cleft sentences come from? Language 53, 34?59.? pages 51Gupta, A. (1980). The Logic of Common Nouns: An Investigation in QuantifiedModal Logic. New Haven, CT: Yale University Press. ? pages 290Halpern, A. (1995). On the Placement and Morphology of Clitics. Dissertationsin Linguistics. Stanford: CSLI Publications. ? pages 98, 100Halvorsen, P.-K. (1978). The syntax and semantics of cleft constructions. Ph. D.thesis, University of Texas at Austin. ? pages 51H?bert, Y. (1978). Sandhi in a Salishan Language: Okanagan. In Papers for the13th Annual International Conference on Salishan Languages, pp. 26?56. ?pages 58H?bert, Y. (1978-1980). Yvonne H?bert?s Field Recordings. ? pages 321H?bert, Y. (1982a). Clausal structure in (Nicola Lake) Okanagan. In Final reportto Canadian Ethnology Service. National Museum of Man, Ottawa. ? pages13, 15, 67, 75, 80, 328H?bert, Y. (1982b). Transitivity in (Nicola Lake) Okanagan. Ph. D. thesis,University of British Columbia. ? pages 13, 15, 63, 66, 67, 168, 202Hedberg, N. (2000). The Referential Status of Clefts. Language 76(4), 891?920.? pages iii, 21, 51, 52, 54, 55, 274, 293, 304, 337, 338, 343, 353, 386, 387,408, 418, 419Heggie, L. (1988). The Syntax of Copular Structures. Ph. D. thesis, University ofSouthern California. ? pages 233, 235, 284427Heggie, L. (1993). The range of null operators: evidence from clefting. NaturalLanguage and Linguistic Theory 11, 45?84. ? pages 51Heim, I. (2011). Definiteness and Indefiniteness. In C. Maienborn, K. vonHeusinger, and P. Portner (Eds.), An International Handbook of NaturalLanguage Meaning, vol. 1. Berlin: Mouton de Gruyter. ? pages 24, 54, 129,137Heim, I. and A. Kratzer (1998). Semantics in Generative Grammar. Malden,Massachusetts: Blackwell Publishing. ? pages 163, 185, 188Heller, D. (2002). On the relation of connectivity and specificational pseudoclefts.Natural Language Semantics 10, 243?284. ? pages 40Heller, D. (2005). Identity and Information. Semantic and Pragmatic Aspects ofSpecificational Sentences. Ph. D. thesis, Rutgers University. ? pages ii, 25, 29,33, 216, 217, 219, 240, 289, 290, 352, 411, 413, 414Heller, D. and L. Wolter (2008). ?That is Rosa?: Identificational Sentences asIntensional Predication. Proceedings of Sinn und Bedeutung 12, pp. 226?240.University of Oslo. ? pages 290, 414Heycock, C. (2012). Specification, equation, and agreement in copular sentences.Canadian Journal of Linguistics 57(2), 209?240. ? pages 21, 218Heycock, C. and A. Kroch (1999). Pseudocleft Connectedness: Implications forthe LF Interface Level. Linguistic Inquiry 30(3), 365?397. ? pages ii, 18, 21,22, 30, 31, 33, 34, 35, 39, 40, 41, 42, 43, 45, 54, 216, 218, 235, 237, 242, 247,259, 286, 349, 351, 353, 410, 415, 417, 418Higginbotham, J. (1985). On semantics. Linguistic Inquiry 16, 547?593. ?pages 233Higgins, F. R. (1973). The Pseudo-cleft Construction In English. Ph. D. thesis,Massachusetts Institute of Technology, Cambridge, MA. ? pages ii, 2, 6, 18,23, 26, 28, 29, 32, 40, 217, 237, 257, 270, 417Jackendoff, R. (1972). Semantic Interpretation in Generative Grammar.Cambridge, MA: MIT Press. ? pages 45, 46, 47, 48, 222, 313, 351Jacobson, P. (1994). Binding connectivity in copular sentences. In M. Harvey andL. Santelmann (Eds.), Proceedings of Semantics and Linguistic Theory (SALT)IV, pp. 161?178. ? pages 40428Jelinek, E. (1998). Prepositions in Northern Straits Salish and the Noun/VerbQuestion. In E. Czaykowska-Higgins and M. D. Kinkade (Eds.), SalishLanguages and Linguistics: Theoretical and Descriptive Perspectives, pp.325?346. Berlin: Morton de Gruyter. ? pages 3, 75Jespersen, O. (1927). A modern English grammar on historical principles. PartIII: Syntax. Second Volume. Heidelberg: Carl Winters. ? pages 51Jespersen, O. (1937). Analytic syntax. New York: Holt, Rinehart, and Winston.? pages 51Jespersen, O. (1940). A modern English grammar on historical principles,Volume V:4 of Syntax. Copenhagen: Munksgaard. ? pages 36Kaplan, D. (1977). Demonstratives. In J. Almong, J. Perry, and H. Wettstein(Eds.), Themes from Kaplan (1989), pp. 481?563. Oxford, England: OxfordUniversity Press. ? pages 44, 262, 356, 414Kaplan, D. (1989). Afterthoughts. In J. Almong, J. Perry, and H. Wettstein (Eds.),Themes from Kaplan, pp. 565?614. Oxford, England: Oxford University Press.? pages 356, 414Karttunen, L. (1977). Syntax and Semantics of Questions. Linguistics andPhilosophy 1, 3?44. ? pages 251Kayne, R. (1994). The Antisymmetry of Syntax. Cambridge, MA: MIT Press. ?pages 34, 235Keenan, E. (1985). Relative Clauses. In T. Shopen (Ed.), Language Typology andSyntactic Description, vol. 2, pp. 141?170. New York: Cambridge UniversityPress. ? pages 188King, J. C. (2001). Complex demonstratives: a quantificational account.Cambridge, MA: MIT Press. ? pages 357Kinkade, M. D. (1983). Salish evidence against the universality of ?noun? and?verb?. Lingua 60, 25?40. ? pages 3, 75Kiss, K. ?. (1998). Identificational focus versus information focus. Language 74,245?273. ? pages 48, 51, 52, 301Kiss, K. ?. (1999). The English cleft construction as a focus phrase. In L. Mereu(Ed.), Boundaries of morphology and syntax, pp. 217?229. Amsterdam: JohnBenjamins. ? pages 51429Koch, K. (2004). On predicate modification in N?ePkepmxcin (Thompson RiverSalish). In J. Brown and T. Peterson (Eds.), Papers for the 39th AnnualInternational Conference on Salish and Neighboring Languages, Vancouver,BC, pp. 269?282. UBCWPL vol 14. ? pages 75, 168Koch, K. (2006). Nominal Modification in N?ePkepmxc?n (Thompson RiverSalish). In S. Bischoff, L. Butler, P. Norquest, and D. Siddiqi (Eds.), Studies inSalishan, Working Papers on Endangered and Less Familiar Languages 7,Cambridge, MA, pp. 127?158. MITWPL. ? pages 16, 18, 78, 81, 168, 176,182, 185, 187, 188, 201, 204, 205, 212, 354, 383, 389Koch, K. (2008a). Focus and Intonation in N?ePkepmxc?n (Thompson RiverSalish). Ph. D. thesis, University of British Columbia. ? pages 17, 19, 21, 45,48, 49, 50, 54, 106, 216, 218, 222, 248, 249, 256, 261, 262, 266, 267, 287, 291,294, 297, 299, 303, 306, 308, 313, 315, 318, 332, 333, 342, 347, 351, 353, 355,356, 365, 373, 375, 376, 382, 383, 384, 385, 386, 387, 388, 389, 392, 396, 410,415Koch, K. (2008b). Focus Projection in N?ePkepmxc?n (Thompson River Salish).In C. B. Chang and H. J. Haynie (Eds.), Proceedings of the 26th West CoastConference on Formal Linguistics, Somerville, MA, pp. 348?356. CascadillaProceedings Project. ? pages 297, 299, 382Koch, K. (2009). Some Issues in the Structure and Interpretation of Clefts inN?ePkepmxc?n (Thompson River Salish). In Papers for the 43rd AnnualInternational Conference on Salish and Neighboring Languages, pp. 97?120.UBCWPL vol 23. ? pages 17, 54, 291, 293, 295, 315, 355, 365, 375, 382,383, 396Koch, K. (2012). An Overview of some focus-related properties of ThompsonSalish. UBC Focus Workshop, May 10, 2012, Vancouver, BC. ? pages 285Koch, K. and M. Zimmermann (2009). Focus-sensitive operators inN?ePkepmxcin. In M. Prinzhorn, V. Schmitt, and S. Zobel (Eds.), Proceedingsof Sinn und Bedeutung 14, Vienna. ? pages 21, 305, 355, 382Kratzer, A. (1998). Scope or Pseudo-Scope? Are There Wide-Scope Indefinites?In S. Rothstein (Ed.), Events in Grammar. Dordrecht: Kluwer AcademicPublishers. ? pages 122, 140Krifka, M. (1991). A compositional semantics for multiple focus constructions.In J. Jacobs (Ed.), Informationsstruktur und Grammatik, Sonderheft der430Linguistischen Berichte, pp. 17?53. Opladen: Westdeutscher Verlag. ? pages43Krifka, M. (2008). Basic Notions of Information Structure. Acta LinguisticaHungarica 55(3-4), 243?276. ? pages 20, 304, 305, 340Kripke, S. A. (1982). Naming and Necessity. Cambridge, MA: HarvardUniversity Press. ? pages 44, 262, 264, 414Kroeber, P. (1986). Antipassives and the differentiation of progressive aspect inSouthern Interior Salish. CLS 22(part 1), 75?89. ? pages 99Kroeber, P. (1991). Comparative Syntax of Subordination in Salish. Ph. D. thesis,University of Chicago. ? pages 239Kroeber, P. (1995). Rhetorical structure of a Kalispel narrative. AnthropologicalLinguistics 37(2), 119?140. ? pages 366Kroeber, P. (1997). Relativization in Thompson River Salish. AnthropologicalLinguistics 39, 376?422. ? pages 80, 81, 200, 354, 383, 385, 387Kroeber, P. (1999). The Salish Language Family: Reconstructing Syntax. Lincoln,Nebraska: University of Nebraska Press. ? pages 3, 16, 72, 81, 83, 89, 100,101, 110, 123, 171, 172, 173, 188, 196, 225, 226, 291, 293, 294, 297, 310, 311,315, 342, 361, 382, 387, 389Kuipers, A. H. (1974). The Shuswap language. The Hague, Netherlands: Moutonde Gruyter. ? pages 344, 377, 378, 379, 380, 381, 382Kuipers, A. H. (1989). A Report on Shuswap (with a Squamish Lexical Appendix),Volume 2 of Langues Et Soci?t?s D?Am?rique Traditionnelle. Paris:Peeters-Selaf. ? pages 377, 381Kuipers, A. H. (2002). Salish Etymological Dictionary. Missoula, MT: Universityof Montana Occasional Papers in Linguistics vol. 16. ? pages 363Lai, S. (1998). Secwepemcts?n Independent Pronouns: Evidence forSubject-Object Asymmetry. In Papers for the 33rd Annual InternationalConference on Salish and Neighboring Languages, pp. 307?321. ? pages 377,378Lindley, L. and J. Lyon (2012). 12 Upper Nicola Okanagan Texts. In Papers forthe 47th Annual International Conference on Salish and NeighboringLanguages. UBCWPL vol 33. ? pages 11, 12, 14, 129431Lindley, L. and J. Lyon (2013). 12 more Upper Nicola Okanagan narratives. InPapers for the 48th Annual International Conference on Salish andNeighboring Languages, pp. 44?113. UBCWPL vol 35. ? pages 11Link, G. (1983). The logical analysis of plural and mass terms: Alattice-theoretical approach. In R. Bauerle, C. Schwarze, and A. von Stechow(Eds.), Meaning, use and interpretation of language, pp. 302?323. Berlin:Mouton de Gruyter. ? pages 129Longobardi, G. (1994). Proper names and the theory of N-movement in syntaxand logical form. Linguistic Inquiry 25, 609?655. ? pages ii, 4, 5, 45, 80, 88,99, 118, 120, 216, 237, 350, 410Ludlow, P. and S. Neale (1991). Indefinite descriptions: In defence of Russell.Linguistics and Philosophy 14, 171?202. ? pages 133Luo, C. (1997). Iconicity or Economy? Polysemy between demonstratives,copulas and contrastive focus markers. Chicago Linguistic Society 33,273?286. ? pages 408Lyon, J. (2010a). Nominal Modification in Upper Nicola Okanagan: A WorkingPaper. In Papers for the 45th Annual International Conference on Salish andNeighboring Languages, pp. 199?233. UBCWPL vol 27. ? pages 78, 168, 176Lyon, J. (2010b). Okanagan Demonstratives and their Uses in Story. In Papersfor the 44th Annual International Conference on Salish and NeighboringLanguages, pp. 26?45. UBCWPL vol 27. ? pages 330, 357, 364Lyon, J. (2011). Oblique Marked Relatives in Southern Interior Salish: HistoricalImplications for a Movement Analysis. In Proceedings of the University ofVictoria Workshop on the Syntax of Relative Clauses, Number 2, pp. 29?68.Working Papers of the Linguistics Circle of the University of Victoria vol 21.? pages 210, 372Manzini, R. (1983). On control and control theory. Linguistic Inquiry 14,421?446. ? pages 233, 235Matthewson, L. (1998). Determiner Systems and Quantificational Strategies:Evidence from Salish. The Hague: Holland Academic Graphics. ? pages ii,17, 96, 99, 103, 110, 112, 113, 114, 118, 120, 123, 124, 125, 133, 147, 148,150, 151, 216, 237, 238, 315, 326, 351, 354, 386, 395, 410432Matthewson, L. (1999). On the interpretation of wide scope indefinites. NaturalLanguage Semantics 7, 79?134. ? pages 17, 122, 124, 129, 137, 138, 140,148, 149, 354, 387Matthewson, L. (2001). Quantification and the Nature of Cross-LinguisticVariation. Natural Language Semantics 9, 145?189. ? pages 17, 99, 140, 148,153Matthewson, L. (2005). When I Was Small - I Wan Kwikws: A GrammaticalAnalysis of St??t?imcets Oral Narratives. Vancouver, BC: UBC Press. ? pages14, 333Matthewson, L. (2006a). Presuppositions and Cross-linguistic Variation.Proceedings of the North East Linguistic Society (NELS) 26. ? pages 131, 308Matthewson, L. (2006b). Temporal semantics in a superficially tenselesslanguage. Linguistics and Philosophy 29, 673?713. ? pages 154, 229, 240Matthewson, L. (2008). Pronouns, Presuppositions, and Semantic Variation. InT. Friedman and S. Ito (Eds.), Proceedings of SALT XVIII, Ithaca, NY. ? pages127, 128, 130, 147, 149Matthewson, L. and H. Davis (1995). The structure of DP in St??t?imcets(Lillooet Salish). In Papers for the 30th Annual International Conference onSalish and Neighboring Languages, pp. 54?68. ? pages 110, 112, 114, 222,324, 326, 331, 356Mattina, A. (1973). Colville Grammatical Structures. Ph. D. thesis, University ofHawaii. ? pages xii, 13, 15, 57, 85, 89, 104, 106, 123, 171Mattina, A. (1979). Pharyngeal Movement in Colville and Related Phenomena inthe Interior Salishan languages. International Journal of AmericanLinguistics 45, 17?24. ? pages 58Mattina, A. (1982). The Colville-Okanagan Transitive System. InternationalJournal of American Linguistics 48, 421?435. ? pages 15, 58, 61Mattina, A. (1985). The Golden Woman: The Colville Narrative of Peter J.Seymour. Tucson, Arizona: University of Arizona Press. ? pages 12, 15, 65,173, 291, 292, 330, 342Mattina, A. (1987). Colville-Okanagan Dictionary. University of MontanaOccasional Papers in Linguistics vol. 7. ? pages 13, 15433Mattina, A. (1993a). Okanagan Aspect: A Working Paper. In Papers for the 28thAnnual International Conference on Salish and Neighboring Languages, pp.233?263. ? pages 15, 63, 70, 180Mattina, A. (1996a). Interior Salish to-be and intentional forms - a working paper.In Papers for the 32nd Annual International Conference on Salish andNeighboring Languages, pp. 239?248. ? pages 154, 349Mattina, A. (2000). Okanagan sandhi & morphophonemics. In Papers for the35th Annual International Conference on Salish and Neighboring Languages,pp. 147?158. UBCWPL vol 3. ? pages 58, 65, 78, 81, 117, 354Mattina, A. (2001). Okanagan sentence types: A preliminary working paper. InL. Bar-el, L. T. Watt, and I. Wilson (Eds.), Papers for the 36th AnnualInternational Conference on Salish and Neighboring Languages, pp. 215?236.UBCWPL vol 6. ? pages ii, 2, 16, 66, 92, 220Mattina, A. (2004). The Okanagan Transitive Sentence Prototype. In Papers forthe 39th Annual International Conference on Salish and NeighboringLanguages. UBCWPL vol 14. ? pages 67Mattina, A. (2008). How I Interlinearize \"Invaluable Stories\" and Other Texts. InPapers for the 43rd Annual International Conference on Salish andNeighboring Languages, pp. 135?150. UBCWPL vol 23. ? pages 14Mattina, A. and M. DeSautel (2002). Dora Noyes DeSautel ?aP k?capt?kw?.Missoula, MT: University of Montana Occasional Papers in Linguistics vol. 15.? pages 12, 13, 14, 50, 220, 355Mattina, N. (1993b). Some Lexical Properties of Colville-Okanagan Ditransitives.In Papers for the 28th Annual International Conference on Salish andNeighboring Languages, pp. 265?284. ? pages 62, 92Mattina, N. (1994). Notes on Colville-Okanagan Word Order. In Proceedings ofthe 10th Annual Northwest Linguistic Conference, Burnaby, BC, pp. 93?102.Simon Fraser University. ? pages 65, 66, 67, 68, 70, 168, 177Mattina, N. (1996b). Aspect and Category in Okanagan Word Formation. Ph. D.thesis, Simon Frasier University, Burnaby, BC. ? pages 1, 2, 5, 15, 16, 58, 61,62, 63, 64, 74, 75, 83, 84, 85, 89, 90, 92, 93, 94, 97, 162, 180, 220, 229, 237,354434Mattina, N. (2002). Notes on determiner phrases in Moses-Columbia Salish. InPapers for the 37th Annual International Conference on Salish andNeighboring Languages, Volume 9 of University of British Columbia WorkingPapers in Linguistics, pp. 261?286. UBCWPL. ? pages 97, 127, 371Mattina, N. (2006). Determiner Phrases in Moses-Columbian Salish.International Journal of American Linguistics 72(1), 97?134. ? pages 16, 110,129, 133, 139, 151, 161, 208, 272, 354, 370, 371Menzies, S. (2012). The Okanagan Modal System. Master?s thesis, University ofBritish Columbia. ? pages 154, 227Mikkelsen, L. (2005). Copular Clauses: Specification, predication, and equation,Volume 85 of Linguistik Aktuell. Amsterdam: John Benjamins. ? pages 6, 25,26, 29, 33, 40, 235, 242, 246, 247, 274, 275, 276, 284, 289, 290, 350, 351, 357,413, 414, 415Mikkelsen, L. (2011). Copular Clauses (chapter 68). In K. von Heusinger,C. Maienborn, and P. Portner (Eds.), Semantics, pp. 1805?1829. Berlin:Mouton de Gruyter. ? pages xii, 20, 27, 28, 40, 419Milsark, G. (1977). Toward an Explanation of Certain Peculiarities of theExistential Construction in English. Linguistic Analysis 3, 1?29. ? pages 150Montler, T. (1993). Attributive Constructions in Saanich. In A. Mattina andT. Montler (Eds.), American Indian Linguistics and Ethnography in Honor ofLaurence C. Thompson, Missoula, MT. University of Montana OccasionalPapers in Linguistics vol. 10. ? pages 65, 81, 168, 175, 199, 207, 398Moro, A. (1995). Small Clauses with Predicative Nominals. In A. Cardinalettiand M. T. Guasti (Eds.), Syntax and semantics, vol. 28: Small Clauses, pp.109?132. New York: Academic Press. ? pages 36, 38, 39Moro, A. (1997). The raising of predicates: Predicative noun phrases and thetheory of clause structure. Cambridge: Cambridge University Press. ? pages6, 28, 29, 30, 33, 34, 35, 36, 37, 45, 216, 226, 236, 240, 241, 287, 350, 410, 413Moro, A. (2000). Dynamic Antisymmetry. Cambridge, MA: MIT Press. ? pages22, 233Ogihara, T. (1987). Obligatory Focus in Japanese and Type-Shifting Principles. InM. Crowhurst (Ed.), WCCFL 6, pp. 213?227. Stanford Linguistics Association.? pages 268435Partee, B. H. (1986). Ambiguous Pseudoclefts with Unambiguous Be. InS. Berman, J.-W. Choe, and J. McDonough (Eds.), Proceedings of NELS XVI,Amherst, MA. University of Massachusetts, Amherst. ? pages 20, 22, 24, 25,27, 235, 239, 263Partee, B. H. (1987). Noun Phrase Interpretation and Type-shifting Principles. InJ. Groenendijk, D. de Jongh, and M. Stokhof (Eds.), Studies in DiscourseRepresentation Theory and the Theory of Generalized Quantifiers, Dordrecht.Foris. ? pages 31, 32, 40, 44Partee, B. H. (2010). Specificational Copular Sentences in Russian and English.Oslo Studies in Language 2(1), 25?49. ? pages 284Pattison, L. (1978). Douglas Lake Okanagan: Phonology and Morphology.Master?s thesis, University of British Columbia, Vancouver, BC. ? pages 15,58Percus, O. (1997). Prying open the cleft. In Proceedings of the 27th AnnualMeeting of the North Eastern Linguistic Society, pp. 337?351. GraduateLinguistic Student Association, University of Massachusetts, Amherst. ?pages iii, 21, 31, 47, 50, 51, 52, 53, 54, 55, 293, 296, 304, 306, 337, 338, 339,343, 350, 353, 385, 386, 387, 409, 418Pereltsvaig, A. (2001). On the Nature of Intra-clausal Relations: A Study ofCopular Sentences in Russian and Italian. Ph. D. thesis, McGill University,Montreal. ? pages 37, 38Pierrehumbert, J. (1980). The Phonology and Phonetics of English Intonation.Ph. D. thesis, Massachusetts Institute of Technology. ? pages 267Postal, P. (1966). On So-Called ?Pronouns? in English. In F. Dinneen (Ed.),Report on the Seventeenth Annual Round Table Meeting on Linguistics andLanguages Studies, pp. 177?206. Washington, DC: Georgetown UniversityPress. ? pages 409Prince, E. (1978). A comparison of wh-clefts and it-clefts in discourse.Language 54, 883?906. ? pages 43Reeve, M. (2007). Relatives and pronouns in the English cleft construction. UCLWorking Papers in Linguistics 19, 157?182. ? pages 20, 50, 51, 337Reeve, M. (2011). The syntactic structure of English clefts. Lingua 121, 142?171.? pages 301436Reichard, G. A. (1947). An Analysis of Coeur d?Alene Indian Myths. TheAmerican Folklore Society Memoirs, vol 41. Philadelphia: American FolkloreSociety. ? pages 367Reinhart, T. (1995). Interface strategies. OTS Working Papers in Linguistics.Utrecht University. ? pages 48Reinhart, T. (1997). Quantifier Scope: How Labor is Divided Between QR andChoice Functions. Linguistics and Philosophy 20, 335?397. ? pages 122, 123,140, 158, 387Rochemont, M. (1986). Focus in generative grammar. Amsterdam: JohnBenjamins. ? pages 6, 48, 51, 301Romero, M. (2005). Concealed Questions and Specificational Subjects.Linguistics and Philosophy 28, 687?737. ? pages ii, 18, 21, 35, 44, 45, 155,216, 218, 248, 262, 263, 264, 287, 349, 351, 355, 410, 413Rooth, M. (1985). Association with focus. Ph. D. thesis, University ofMassachusetts, Amherst. ? pages 21, 45, 46, 292Rooth, M. (1992). A theory of focus interpretation. Natural LanguageSemantics 1, 75?116. ? pages 21, 45, 46, 216, 218, 287, 292, 398, 407, 410Rothstein, S. (1995). Small Clauses and Copular Constructions. In A. Cardinalettiand M. T. Guasti (Eds.), Syntax and Semantics, Volume 28, pp. 27?48. NewYork: Academic Press. ? pages 36, 235Rullmann, H. (1995). Maximality in the Semantics of Wh-Constructions. Ph. D.thesis, University of Massachusetts, Amherst. ? pages 124Schachter, P. (1973). Focus and relativization. Language 49, 19?46. ? pages 51Schlenker, P. (2003). Clausal equations (a note on the connectivity problem).Natural Language and Linguistic Theory 21, 157?214. ? pages 40Selkirk, E. (1995). Sentence prosody: Intonation, stress and phrasing. InJ. Goldsmith (Ed.), The Handbook of Phonological Theory, pp. 550?569.Cambridge, MA: Blackwell Publishing. ? pages 45, 46, 47, 48, 222, 313Selkirk, E. (2007). Contrastive focus, givenness, and the unmarked status of?discourse new?. In C. F?ry, G. Fanselow, and M. Krifka (Eds.), The Notions ofInformation Structure, Interdisciplinary Studies on Information Structure, pp.125?146. Potsdam: Universit?tsverlag Potsdam. ? pages 48437Selkirk, E. and A. Kratzer (2007). Phase theory and prosodie spellout: The caseof verbs. The Linguistic Review 2-3, 93?135. ? pages 48, 49Shank, S. (2003). A preliminary semantics for pronominal predicates. InJ. Brown and M. Kalmar (Eds.), Papers for the 38th Annual InternationalConference on Salish and Neighboring Languages, pp. 215?236. UBCWPL vol11. ? pages 54, 332, 333, 353, 373, 376, 385, 386, 397, 398, 407, 408Sharvit, Y. (1999). Connectivity in specificational sentences. Natural LanguageSemantics 7, 299?339. ? pages 40Stowell, T. (1981). Origins of Phrase Structure. Ph. D. thesis, MassachusettsInstitute of Technology. ? pages 36, 37, 38, 217, 233Szabolcsi, A. (1981). The Semantics of Topic-Focus Articulation. InJ. Groenendijk (Ed.), Formal methods in the study of language. University ofAmsterdam. ? pages 48Thoma, S. (2007). The Categorial Status of St??t?imcets Independent Pronouns.Qualifying Paper, University of British Columbia. ? pages 305, 394, 399Thompson, L. and T. Thompson (1992). The Thompson Language. Missoula,MT: University of Montana Occasional Papers in Linguistics vol. 8. ? pages162, 386Vallduv?, E. (1992). The informational component. New York: Garland. ? pages43van Eijk, J. P. (1997). The Lillooet Language: Phonology, Morphology, Syntax.Vancouver, BC: UBC Press. ? pages 13, 17, 118, 120, 123van Eijk, J. P. and L. Williams (1981). Lillooet Legends and Stories. MountCurrie, BC: Ts?zil Publishing House. ? pages 331Van Geenhoven, V. (1998). Semantic incorporation and indefinite descriptions.Stanford, CA: CSLI Publications. ? pages 85, 90, 122, 124, 136, 156, 162,163, 166von Stechow, A. (1990). Focusing and backgrounding operators. In W. Abraham(Ed.), Discourse Particles, pp. 37?84. Amsterdam: John Benjamins. ? pages43Watkins, D. (1970). A description of the phonemes and position classes in themorphology of Head of the Lake Okanagan (Salish). Ph. D. thesis, Universityof Alberta, Edmonton, AB. ? pages 15, 57438Werle, A. (2000). Semantic incorporation in Lillooet. In Papers for the 35thAnnual International Conference on Salish and Neighboring Languages, pp.219?226. UBCWPL vol 3. ? pages 163, 164Willett, M. L. (2003). A Grammatical Sketch of Nxa?amxcin (Moses-ColumbiaSalish). Ph. D. thesis, University of Victoria, Victoria, BC. ? pages 16, 104,208, 371, 372, 373Williams, E. (1975). Small clauses in English. In J. Kimball (Ed.), Syntax andSemantics, Volume 4, pp. 249?273. New York: Academic Press. ? pages 233Williams, E. (1983). Against Small Clauses. Linguistic Inquiry 14, 287?308. ?pages 24, 36Wiltschko, M. (2005). Expletive categorical information. A case study of numbermarking in Halkomelem. In L. Bateman and C. Ussery (Eds.), Proceedings ofthe North East Linguistic Society (NELS) 35, pp. 631?646. Amherst, MA:Graduate Linguistic Student Association, University of Massachusetts,Amherst. ? pages 235Wirth, J. (1978). The derivation of cleft sentences in English. Glossa 121, 58?81.? pages 51Zamparelli, R. (1995). Layers in the Determiner Phrase. Ph. D. thesis, Universityof Rochester, New York. ? pages 99439"@en . "Thesis/Dissertation"@en . "2014-05"@en . "10.14288/1.0165734"@en . "eng"@en . "Linguistics"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "Attribution-NonCommercial-NoDerivs 2.5 Canada"@en . "http://creativecommons.org/licenses/by-nc-nd/2.5/ca/"@en . "Graduate"@en . "Predication and equation in Okanagan Salish : the syntax and semantics of determiner phrases"@en . "Text"@en . "http://hdl.handle.net/2429/45684"@en .