THE BLACKFOOT CONFIGURATIONALITY CONSPIRACY: PARALLELS AND DIFFERENCES IN CLAUSAL AND NOMINAL STRUCTURES by HEATHER ANNE BLISS B.A. Honours, University of Calgary, 2003 M.A., University of Calgary, 2005 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Linguistics) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2013 ? Heather Anne Bliss, 2013 ii ABSTRACT This dissertation explores the argument-typing system of Blackfoot, a Plains Algonquian language spoken in Southern Alberta and Northwestern Montana. It develops a classification of the phrases, words, and morphemes in Blackfoot that are associated with arguments of the predicate (nominal expressions and argument-indexing verbal morphology) according to their internal and external syntax. The analysis sheds light on how and why Blackfoot displays properties of a non-configurational language. The main thesis is that non-configurationality in Blackfoot is a conspiracy resulting from properties of Blackfoot?s argument-typing system, and in particular the PROXIMATE/OBVIATIVE contrast, a type of reference-tracking morphology that disambiguates between multiple 3rd persons in a clause. The dissertation begins with a discussion of the theoretical assumptions, methodology, and the main proposal (Chapter 1) as well as a background on the relevant properties of Blackfoot morphosyntax (Chapter 2). Following that is a detailed discussion of the internal and external syntax of inflected nouns (Chapter 3), demonstratives (Chapter 4), person prefixes (Chapter 5) and number suffixes (6). Chapter 7 discusses the implications of Blackfoot?s argument-typing system for non-configurationality. Blackfoot is shown to be a partially non-configurational language, in which proximate nominal expressions are not subject to the same distributional constraints as obviative ones (i.e., proximate nominal expressions display non-configurational properties such as free word order and extensive use of null anaphora). Finally, Chapter 8 considers the proximate/obviative contrast in a broader cross-Algonquian context. The data and generalizations presented in this dissertation are largely from the author?s own fieldwork with two native speakers over a ten year period, and these are supplemented with data from text materials glossed and annotated by the author. As such, a key contribution of this research is empirical; it contributes to the documentation and analysis of this endangered First Nations language. iii PREFACE With the exception of some independent and collaborative projects that have been presented at conferences and/or submitted for publication, this dissertation is original, unpublished, independent work by the author, Heather Bliss. In Chapter 2, my discussion of the cline of audibility builds on the findings of Bliss and Glougie (2010) and Gick et al. (2012), both of which I am an author. The analysis of the direct/inverse as associating with an Aspect head builds on my M.A thesis (Bliss 2005), as well as Bliss et al. (2010a, b). In Chapter 5, the analysis of the long and short form person prefixes is taken from Bliss and Gruber (2011a, b). The discussion of the prefix na- is influenced by Bliss and Ritter (2007, 2009), although the analysis is different. The fieldwork undertaken for this dissertation is covered under ethics approval for the project ?Interface Syntax? (H10-02768) granted to the dissertation supervisor, Dr. Martina Wiltschko. iv TABLE OF CONTENTS Abstract ????????????.?????????????????????????..ii Preface??????????????????????????????????????. iii Table of Contents ??????????????????????????????????iv List of Tables ???????????????????????????????????.xii List of Figures ?????.?????????????????????????..????xv List of Abbreviations ????????????????????????????????xvi Acknowledgements ????????????????????????????????..xvii 1 INTRODUCTION ??????..????????????????????????1 1.1 Introduction ?????????????????????????????????..1 1.2 Methodology for Data Collection ?????????????????????????.2 1.3 Theoretical Assumptions ????????????????????????????.4 1.3.1 Syntactic Categories: The Universal Spine Hypothesis ?????????????.4 1.3.2 Syntactic Dependencies ???????????????.??????????.8 1.3.3 The Morphology-Syntax Interface ?????????????????????11 18.104.22.168 Linearization of Suffixes ?????????????????????.12 22.214.171.124 Linearization of Prefixes ?????????????????????.12 126.96.36.199 Linearization of Clitics ??????????????????????14 188.8.131.52 Post-Syntactic Spell-Out Rules ???????????????????15 1.3.4 Mapping Morphemes onto the Syntactic Spine ????????????????.18 1.4 Proposal ??????????????????????????????????.18 1.5 Outline of Dissertation..???.?????????????????????????.24 2 BACKGROUND ON BLACKFOOT ??????????????????????28 2.1 Introduction ?????????????????????????????????28 v 2.2 Language Profile ???????????????????????????????28 2.3 Nominal Expressions ?????????????????????????????.29 2.4 Verbal Complex ???????????????????????????????.33 2.4.1 Person Prefixes ????????????????????????????.34 2.4.2 Verbal Prefixes ????????????????????????????.37 184.108.40.206 Tense/Aspect/Modality ?????????????????????....38 220.127.116.11 Quantifiers and Other Scope-Taking Prefixes ?????????????.40 18.104.22.168 Adpositions ??????????????????????????..41 22.214.171.124 Other Verbal Prefixes ??????????????????????..42 2.4.3 Verb Stems ??????????????????????????????42 126.96.36.199 Roots ?????????????????????????????43 188.8.131.52 Incorporated Nouns ???????????????????????.43 184.108.40.206 Finals ?????????????????????????????44 2.4.4 Direct/Inverse Marking ?????????????????????????.45 2.4.5 Clause-Typing Suffixes ?????????????????????????48 2.4.6 Person/Number Suffixes ?????????????????????????50 2.4.7 Number Suffixes ????????????????????????????51 2.4.8 Enclitic Pronouns ??????????????????????????..?53 2.5 The Phonetic Realization of Proximate and Obviative Suffixes ?????????????56 2.5.1 Voiceless and Soundless Suffixes ?????????????????????57 2.5.2 Ghost Suffixes ????????????????????????????..59 2.5.3 Cline of Audibility ???????????????????????????61 2.6 Grammatical Functions ????????????????????????????.61 2.6.1 Subject and Object ???????????????????????????62 2.6.2 PoV Holder .. ?????????????????????????????69 2.6.3 Spec, IP and Spec, CP ??????????????????????????71 vi 2.6.4 Summary ???????????????????????????????74 2.7 Conclusion ?????????????????????????????????.74 3 MAPPING NOMINAL EXPRESSIONS ONTO THE SYNTACTIC SPINE ??????75 3.1 Introduction ?????????????????????????????????75 3.2 The Syntax of Proximate Singular (Nominal) Expressions ...??????????????76 3.2.1 Proximate Expressions are LINKPs???.. ?....???????????????.77 220.127.116.11 Proximate ?wa Associates with LINK?.?.??????????????78 18.104.22.168 Proximate ?wa does not Associate with ? ...??????????????88 3.2.2 Proximate Expressions are Adjuncts ????????????????????90 3.2.3 Summary ???????????????..???????????????96 3.3 The Syntax of Obviative Singular Nominal Expressions ?????.??????????.96 3.3.1 Obviative Singular Nominal Expressions are KPs ??.????.????????.97 3.3.2 Obviative Singular ?yi is a Generalized Case Marker ?????????????.98 3.3.3 Summary: Proximate versus Obviative Nominal Expressions ??????????103 3.4 The Syntax of Bare Nouns ???????????????????????????104 3.4.1 Diagnosing Pseudo-Incorporation ??????????????..??????105 3.4.2 The External Syntax of Pseudo-Incorporated Objects ?????????????107 3.4.3 Semantic Properties of Pseudo-Incorporated Objects ?????????????112 3.4.4 The Internal Syntax of Pseudo-Incorporated Objects ?????????????.115 3.4.5 Summary ??????????????????????????????120 3.5 The Syntax of Plural Nominal Expressions ???????????????????.?120 3.5.1 Plural Marking Associates with ? ..????????????????????.121 3.5.2 Plural Nouns can be Pseudo-Incorporated ?????????????????..125 3.5.3 Plural Nominal Expressions can be Arguments ???????????????.127 3.6 Conclusion ?????????????????????????????????131 vii 4 MAPPING DEMONSTRATIVES ONTO THE SYNTACTIC SPINE ????????133 4.1 Introduction ????????????????????????????????..133 4.2 Demonstratives are not D Heads ????????????????????????..134 4.2.1 Demonstratives are Morphosyntactically Complex ??????????????137 22.214.171.124 Demonstrative Stems ??????????????????????.138 126.96.36.199 Demonstrative Inflection ?????????????????????141 188.8.131.52 Demonstrative Post-Inflectional and Verbalizing Suffixes ????????145 4.2.2 Demonstratives Form a Constituent that Excludes the Noun ??????????.149 4.2.3 Demonstratives are not D: Cross-Algonquian Support ????????????..153 4.3 Demonstratives are Anchoring Arguments in Spec, DP ???????????????..156 4.3.1 The Syntactic Position of Demonstratives ?????????????????..156 4.3.2 The Syntactic Function of Demonstratives ?????????????????158 4.4 Conclusion ?????????????????????????????????163 5 MAPPING PERSON PREFIXES ONTO THE SYNTACTIC SPINE ????????.164 5.1 Introduction ???????????????????????????????.....164 5.2 Spec, IP in Blackfoot: Proposal and Predictions ??????????????????.166 5.2.1 Blackfoot INFL is Person-Based ?????????????????????167 5.2.2 A Feature-Geometric Model for Person-Based INFL ?????????????.173 5.3 Person Prefixes in Spec, IP are Sensitive to the Ir/realis Contrast ???????????..176 5.4 Person Prefixes in Spec, IP have Temporal Content ?????????????????183 5.4.1 Only the Long Form Prefixes are Temporally Restricted ???????????...183 184.108.40.206 Long and Short Form Prefixes Differ in their Categorical Status .?????184 220.127.116.11 Long Form Prefixes Refer to a Stage of an Individual ?????????.188 5.4.2 Only the Long Form Prefixes are in Spec, IP ????????????????.194 viii 5.5 Spec, IP has Deictic Content???????????????????? ??????198 5.5.1 na- Occupies the Same Syntactic Position as the Person Prefixes ????????199 5.5.2 na- is an Evidential Marker ???????????????????????202 5.6 Conclusion ?????????????????????????????????208 6 MAPPING NUMBER SUFFIXES ONTO THE SYNTACTIC SPINE ????????210 6.1 Introduction ????????????????????????????????..210 6.2 The Syntactic Position of Number Suffixes ????????????????????210 6.2.1 Number Suffixes in Matrix Declarative AI and TA Clauses?????????..?.211 6.2.2 Number Suffixes are not in vP ??????????????????????217 6.2.3 Number Suffixes are not in AspP ?????????????????????220 6.2.4 Number Suffixes are not in IP ??????????????????????.222 6.2.5 Number Suffixes are in CP ???????????????????????.224 18.104.22.168 Number Suffixes are Sensitive to Illocutionary Force ?????????..225 22.214.171.124 Number Suffixes are Restricted to Matrix Clauses ???????????234 6.2.6 Summary: Number Suffixes are in CP ???????????????????237 6.3 The Syntactic Function of Number Suffixes ???????????????????.239 6.3.1 Number Suffixes are Heads ???????????????????????239 6.3.2 Proximate versus Obviative and Plural ??????????????????..241 126.96.36.199 Obviative and Plural Clauses have Number Agreement in C ???????243 188.8.131.52 Proximate Clauses are LINKPS ??????????????????..245 6.3.3 Summary ??????????????????????????????.247 6.4 Conclusion ?????????????????????????????????247 7 IMPLICATIONS FOR (NON-)CONFIGURATIONALITY ????????????.250 7.1 Introduction ????????????????????????????????..250 ix 7.2 Non-Configurationality, Take 1: Hale?s (1983) Diagnostics ?????????????...251 7.2.1 Null Anaphora ????????????????????????????252 184.108.40.206 Null Anaphora with Local Persons ?????????????????.252 220.127.116.11.1 Local Subject and Object ????????????????252 18.104.22.168.2 Local Unindexed Object ????????????????254 22.214.171.124.3 Local Oblique ????????????????????.255 126.96.36.199 Null Anaphora with Proximate Singular 3rd Persons ??????????257 188.8.131.52.1 Proximate Singular Subject and Object ??????????257 184.108.40.206.2 Proximate Singular Unindexed Object ?????.?????.258 220.127.116.11.3 Proximate Singular Oblique ??????.????????.258 18.104.22.168 Null Anaphora with Obviative Singular and Plural 3rd Persons ???..??..259 22.214.171.124.1 Obviative Singular and Plural Subject and Object ???.??...260 126.96.36.199.2 Obviative Singular and Plural Unindexed Object ???.???262 188.8.131.52.3 Obviative Singular and Plural Oblique ?????.?????.263 184.108.40.206 Summary??????????????????????????.?264 7.2.2 Word Order?????????????????????????????. 264 220.127.116.11 Word Order and the Proximate/Obviative Contrast ??????????..265 18.104.22.168 Word Order is Focus-Sensitive ??????????????????.269 22.214.171.124 Summary ???????????????????????????272 7.2.3 Discontinuous Expressions ???????????????????????.272 7.2.4 Summary ??????????????????????????????.276 7.3 Non-Configurationality, Take 2: Pronominal Argument Hypothesis ??????????..277 7.3.1 Overview of the PAH ?????????????????????????.278 7.3.2 Clitics and Agreement ?????????????????????????279 7.3.3 Agreement and Syntactic Restrictions ???????????????????282 7.3.4 Summary ??????????????????????????????.286 x 7.4 Non-Configurationality, Take 3: Lack of Structural Asymmetries ???????????.287 7.4.1 Variable Binding Tests ?????????????????????????288 126.96.36.199 Variable Binding with Nominalizations ???????????????.289 188.8.131.52 Variable Binding with Possessed Nouns ???????????????292 7.4.2 Other Tests for C-Command Relations ??????????????????..294 184.108.40.206 Condition A Tests ???????????????????????..295 220.127.116.11 Superiority Tests ????????????????????????296 18.104.22.168 Quantifier Scope Tests ?????????????????????...297 22.214.171.124 Condition C Tests ???????????????????????..299 7.5 Conclusion ?????????????????????????????????303 8 BEYOND BLACKFOOT: CROSS-ALGONQUIAN COMPARISONS ???????..305 8.1 Introduction ????????????????????????????????..305 8.2 Obviation and In/dependence in Blackfoot: A Summary ???????????????306 8.3 The CP Layer Across Algonquian ????????????????????????309 8.3.1 CP in other Algonquian Languages ????????????????????.309 126.96.36.199 Clause-Typing Morphology and Initial Change in C ??????????.309 188.8.131.52 C Outside the Verbal Complex ??????????????????.315 184.108.40.206 Summary: CP across Algonquian ?????????????????..319 8.3.2 Encoding the Independent/Dependent Contrast in the CP Layer ?????????319 8.4 Obviation across Algonquian ????????????????????????...?320 8.4.1 (Some) Discourse Functions of Algonquian Obviation ????????????..322 8.4.2 Discourse Functions of Blackfoot Obviation ????????????????..324 8.4.3 Towards a Comparative Syntax of Algonquian Obviation ???????????.326 REFERENCES ??????????????????????????????????329 xi APPENDIX A: A SURVEY OF NOMINAL EXPRESSIONS ??????????????..355 A.1 Nominal Expressions that can Function as the Subject???????????????.. 355 A.2 Nominal Expressions that can Function as the Indexed Object?????????????362 A.3 Nominal Expressions that can Function as the AI Object ???????????????365 A.4 Nominal Expressions that can Function as the Unindexed Object of a Ditransitive Verb ??..369 A.5 Nominal Expressions that can Function as an Oblique ????????????????372 A.6 Summary ?????????????????????????????????..375 xii LIST OF TABLES Table 1.1. Grammatical Features???????????..??????????????.. 10 Table 2.1. Nominal Inflection??????????????????????????? 30 Table 2.2. Person Prefixes ????????????????????????????36 Table 2.3. Verb Finals ??????????????????????????????44 Table 2.4. Direct/Inverse Marking in Matrix Clauses ??????????????????46 Table 2.5. Clause Types ?????????????????????????????.48 Table 2.6. Person/Number Suffixes ?????????????????????????50 Table 2.7. Number Suffixes ????????????????????????????51 Table 2.8. Nominal and Verbal Number Suffixes Compared ???????????????.52 Table 2.9. Enclitic Pronouns ???????????????????????????..53 Table 2.10. Affixation versus Encliticization ?????????????????????..53 Table 2.11. Stimuli for Production Experiment (adapted from Gick et al. 2012, p. 52) ?????.57 Table 2.12. Objects of Morphologically Intransitive and Transitive Verbs ??????????.64 Table 2.13. Structural Correlates of Grammatical Functions????????????????74 Table 3.1. Four Types of Nominal Expressions ???????????????????..76 Table 3.2. Number/Obviation Morphology ?????????????????????..87 Table 3.3. Case Systems ????????????????????????????100 Table 3.4. Proximate versus Obviative Nominal Expressions ??????????????103 Table 3.5. Diagnostics for Pseudo-Incorporation ??????????????????..106 Table 3.6. Semantic Properties of Pseudo-Incorporation ????????????????115 Table 3.7. Diagnostics for Plural Marking as a Functional Head ?????????????123 Table 3.8. Structural Ambiguity with Plural Nouns ??????????????????128 xiii Table 4.1. Demonstrative Roots ?????????????????????????.139 Table 4.2. Nominal Inflection ??????????????????????????140 Table 4.3. Post-Inflectional Demonstrative Suffixes ?????????????????.145 Table 4.4. Montagnais Demonstratives (adapted from Cyr 1993: 199) ??????????..153 Table 4.5. Person Prefixes and Demonstrative Roots Compared ????????????...159 Table 4.6. Distribution of Person Prefixes and Demonstratives across Grammatical Functions ?.162 Table 5.1. Person Prefixes ???????????????????????????.164 Table 5.2. Clause Types ????????????????????????????.176 Table 5.3. Distribution of Person Prefixes in Conjunct Clauses ?????????????.180 Table 5.4. Distribution of Person Prefixes according to Clause Type ???????????182 Table 5.5. Long and Short Form Prefixes ?????????????????????..185 Table 5.6. Complementarity of na- and Person Prefixes ????????????????200 Table 5.7. Distribution of Person Prefixes and na- across Clause Types ??????????202 Table 6.1. Number Suffixes ?????????????????????????..?210 Table 6.2. Number Suffixes in Matrix Declarative Clauses ???????????????217 Table 6.3. Verb Finals ?????????????????????????????218 Table 6.4. Distribution of ?wa in Matrix Declarative Clauses ??????????????220 Table 6.5. Distribution of Number Suffixes According to Illocutionary Force ???????..234 Table 6.6. Distribution of Number Suffixes According to Clause Type and Illocutionary Force ?237 Table 6.7. Differences between ?wa versus ?yini and ?yi ..............................................................243 Table 6.8. Two Ways of Organizing the Number Suffixes ???????????????.246 Table 6.9. Argument-Indexing Morphology.?????????????????????248 Table 7.1. 3rd Person Argument Indexing Morphology ????????????????..250 xiv Table 7.2. Null Anaphora with Local Persons ????????????????????256 Table 7.3. Null Anaphora with Proximate Singular 3rd Persons ?????????????259 Table 7.4. Null Anaphora with Obviative Singular and Plural 3rd Persons ????..????..265 Table 7.5. Null Anaphora with all Persons ?????????????????????.265 Table 7.6. Word Order Patterns ?????????????????????????.269 Table 7.7. Partial Non-Configurationality: Hale?s Diagnostics ?????????????.277 Table 7.8. Predictions of the PAH ????????????????????????.286 Table 7.9. Variable Binding Patterns ??????????????????????.....289 Table 7.10. Partial Non-Configurationality: Summary ?????????????????303 Table 8.1. Proximate versus Obviative Nominal Expressions ??????????????308 Table A.1. Types of Nominal Expressions Functioning as the Subject ???????????361 Table A.2. Types of Nominal Expressions Functioning as the Indexed Object ????????364 Table A.3. Types of Nominal Expressions Functioning as the AI Object ??????????369 Table A.4. Types of Nominal Expressions Functioning as the Unindexed Object of a Ditransitive Verb ????????????????????????????????.372 Table A.5. Types of Nominal Expressions Functioning as an Oblique ???????????375 Table A.6. Summary of Types of Nominal Expressions ????????????????..376 xv LIST OF FIGURES Figure 2.1. Morphological Template for Verbal Complex ????????????????.34 Figure 2.2. Variable Realizations of the Proximate -wa Suffix ??????????????.61 Figure 4.1. Morphological Template for the Demonstratives ???????????????138 Figure 5.1. Morphological Template for Verbal Complex ????????????????164 Figure 5.2 Cowper?s (2005) Feature Geometric Representation of INFL ??????????173 Figure 5.3. Waldie?s (2013) Evidential Relations ???????????????????.203 xvi LIST OF ABBREVIATIONS 1,2,3 = 1st, 2nd, 3rd, 4th person ABL = ability modal ABS = absolutive ACCOMP = accompaniment AI = Animate Intransitive ANIM = animate ASSOC = associative BEN = benefactive CAUS = causative CL = classifier CN = conjunct nominal CONJ = conjunct order CONT = content DEM = demonstrative DIM = diminutive DIR = direct DIST = distributive ERG = ergative EVID = evidential FUT = future IC = initial change II = Inanimate Intransitive IMP = imperative IMPF = imperfective INAN = inanimate INCL = inclusive INSTR = instrument INTERR = interrogative INTNS = intensifier INV = inverse INVIS = invisible LOC = locative LOCAL = local person (1st or 2nd) MOD = modal (epistemic) MVG = moving NEG = negative NOM = nominalizer NONAFF = non-affirmative NONFUT = non-future NONLOC = non-local person (3rd) NONPART = non-particular OBV = obviative OT = other time PERF = perfect PL = plural POSS = possessive PRN = pronoun PROX = proximate PURP = purpose RECPR = reciprocal REFL = reflexive REP = reportative RESTR = restricted SBJN = subjunctive SG = singular SPKR = speaker STAT = stationary TA = Transitive Animate TI = Transitive Inanimate UNREAL = unreal order UNSPEC = unspecified person xvii ACKNOWLEDGEMENTS With everything I read I always look at the acknowledgements. I?m fascinated by the way an author?s vision can be shaped by the people around them. And how, really, a short statement of acknowledgement is just a small clue to the complex picture underneath. Here I have the chance to reflect on how my own vision has been shaped, and to try to paint a picture that expresses my gratitude. It?s but a drop in the bucket. First and indeed foremost, my thanks go to my Blackfoot teachers Rachel Ermineskin and Beatrice Bullshields. Rachel, the first time I ever heard Blackfoot it was spoken from your lips over ten years ago -- and over this time, you have shown so much patience and support as I stumble along. Beatrice, your commitment to your students and to your language is inspiring. Your rigour has made me a better language learner and a better linguist. To both of you, this dissertation is as much yours as mine. Nitsik?htasi?taki. I don?t even know where to begin in thanking my advisor, Martina Wiltschko. You have been my mentor and my friend, you have taught me more than I expected to learn, and you have pushed and challenged and inspired me ? not only with this dissertation, and not only in linguistics, but in so many aspects of my life. My other committee members, Rose-Marie D?chaine and Hotze Rullmann, have been instrumental in shaping this dissertation. Rose-Marie, my work has benefited from your ability to balance the ?bigger picture? with the intricate details. Hotze, the role you?ve played in my career stretches far beyond this dissertation. From supervising my honours thesis to encouraging me to do my Master?s, to working through this final stage with me, you have always supported me and helped me to think about things in ways I otherwise wouldn?t. To the other members of my defense committee, Marcel den Dikken, Henry Davis, and Barbara Dancygier, thank you for the care and attention you have paid to my dissertation. Your thoughtful questions helped make made my defense fun and exciting, have improved the final version of the dissertation, and have given me avenues to pursue in my future research. xviii I have been so fortunate to have had Betsy Ritter as a mentor, a friend, and a collaborator for many years. Betsy, you gave the tools to think the way I do about language, and even after I came to UBC, you continued to work with me to help me sharpen those tools. Bettina Gruber, our collaboration was so unexpected and intense and inspiring. I love the way we came from our different perspectives, learned how to learn together, and then converged. I can?t wait for more. This time in my life will always be tied up with fond memories of my cohort -- Elizabeth Ferch, Raphael Girard, Murray Schellenberg, Beth Stelle, Anita Szakay, Carmela Toews ? and of course our honourary cohort member Dennis Storoshenko. Our first two years together at UBC established a tight bond, and that?s carried me through the later years of grad school, well past the time when we were together more than we were not. I consider myself so lucky for having shared this experience with you guys, and I?ll always remember that closeness, despite the distance that separates us all now. My fellow Thesis Anonymous members ? Solveiga Armoskaite, Christiana Christodoulou, Atsushi Fujimori, Peter Jacobs, Sonja Thoma, and James Thompson - our meetings were so fruitful. Whether we were brainstorming, or motivating, or critiquing, or commiserating, or just laughing and eating and drinking, having your support elevated me. My understanding of Blackfoot has been broadened through my involvement with the UBC Blackfoot Research Group. Solveiga Armoskaite, Mike Barrie, Joel Dunham, Jennifer Glougie, Meagan Louie, Am?lia Reis Silva, it has been so much fun to uncover puzzles together. Of course, we all have had a huge leg up thanks to the scholarship of Don Frantz. Don, I appreciate so much you taking the time to answer my questions over the years, and to give me frank and thoughtful feedback on my work. There are other faculty in the Department of Linguistics who have taught me useful things that have impacted this dissertation. Thank you in particular to Bryan Gick, Gunnar Hansson, Lisa Matthewson, and Doug Pulleyblank. xix Of course, a dissertation isn?t written in a bubble, and life continued to go on all around me, even when I was wrapped up in writing. Carmelle Lamb, you pulled me out of my fog in all the right ways. I love all of our adventures, and I can?t help but think that during this crazy time in my life, you helped me maintain not only balance but, really, sanity. Meghan Haberl, Wayne Adams, and Tammy Oliver, you too have provided necessary distractions and unwavering support. I would be remiss if I didn?t mention Josh Bliss. Despite everything, you have always seen the value in me chasing after this, and you?ve graciously allowed yourself to be caught up in the chase too. The right ?ingredients? needed to fall into place for this dissertation to get written. I am so fortunate to have had the chance to escape on some wilderness writing retreats at exactly the times when I needed them. But when I couldn?t escape, O?Hare?s Pub proved to be a place with the right ambience ? and the right beer (Driftwood Fat Tug, which will now forever remind me of writing my dissertation). And Maki Performance Training provided the right outlet for all my frustrations and gave me the strength and stamina for the final leg of this journey. Steve, I don?t think either of us expected to find each other when we did ? and I don?t think we always knew how best to support each other through the challenges we?ve faced in the past few years. But despite all my fumblings, and your occasional bewilderment about this process, you have believed in and encouraged me. And as I learn about your lens for seeing the world, you give me a fresh perspective. Finally, to my daughters Beth Bliss and Anna Bliss, you didn?t choose to be along for this journey, but you have decided to be my fans nevertheless. Words cannot express how I feel about your role in this work. I?ve been driven by wanting to show you how being tenacious and persistent can bring you to your goals, and that sometimes it?s terribly hard ? but still you can persevere and achieve. But that?s not the whole story. You have grounded me ? and you?ve never let me escape reality, no matter which deadline is lurking. My achievements are the richer for this. 1 CHAPTER 1 INTRODUCTION 1.1. Introduction What does it mean for a language to be non-configurational? In a literal sense, a non-configurational language is a language lacking structure, a language without hierarchical relations between the pieces that comprise a sentence. However, the most basic assumption of the generativist enterprise is that, in all languages, there are building blocks which combine with each other to form consecutively larger building blocks, resulting in a hierarchical syntax. I refer to these building blocks as LINGUISTIC OBJECTS. Under this assumption, ?non-configurationality? becomes a cover term for a clustering of properties that obscure the evidence for structural relations between linguistic objects. The task for a linguist looking at the syntax of a non-configurational language is to uncover the hierarchical structure lurking beneath the mask of non-configurationality. In this dissertation, I take on this task for Blackfoot, a Plains Algonquian language spoken in Southern Alberta and Northwestern Montana. Blackfoot displays the hallmark properties of a non-configurational language: it has extensive null anaphora, variable word order, and discontinuous expressions (cf. Hale 1983). However, it shows evidence of hierarchical structure in the form of c-command relations between constituents both within and across clauses. I propose that Blackfoot?s mask of non-configurationality is the result of its ARGUMENT-TYPING SYSTEM, i.e., the classification of linguistic objects associated with the arguments of the predicate. Specifically, in this dissertation I analyse the internal and external syntax of Blackfoot?s argument expressions (inflected nouns and demonstratives) and argument-indexing verbal morphology (person prefixes and number suffixes). An emergent theme is the important role that PROXIMATE and OBVIATIVE morphology plays in obscuring Blackfoot?s hierarchical syntax. 3rd person argument expressions in Blackfoot are marked as either singular or plural, and singular argument expressions are coded as either proximate or obviative. 2 Clauses are similarly coded as to whether the arguments in the clause are proximate singular, obviative singular, or plural. I argue that the proximate marking signals that a phrase is independent; it cannot be structurally dependent on another constituent. As such, clauses coded as proximate are necessarily matrix clauses1, and argument expressions coded as proximate cannot occupy argument positions. Rather, they are adjoined outside the clause and bind a null pro argument inside the clause (cf. Baker 1991, 1996). As for obviative marking, I decompose it into two different types of dependency-marking. I analyse obviative marking on the clause as singular number agreement in C. Obviative marking on argument expressions, on the other hand, I analyse as a generalized case marker; it appears on all argument expressions in the clause, but does not co-vary with grammatical function (e.g., subject/object). Under this analysis, the mask of non-configurationality in Blackfoot is a conspiracy, resulting predominantly from two independent factors: the requirement that one class of 3rd argument expressions be syntactically independent, and the fact that case-marking is indiscriminate, and does not code grammatical functions. This chapter provides the relevant background for the dissertation and summarizes my main claims. The chapter proceeds as follows. In ?1.2, I discuss my methodology, and in ?1.3, I lay out my theoretical assumptions. In ?1.4, I walk through the main proposal of the dissertation, and in ?1.5, I outline the contributions of each subsequent chapter. 1.2. Methodology for Data Collection In this section, I discuss the methodology I employed for collecting the Blackfoot data presented in this dissertation. The primary source of data is my own fieldwork over the past ten years (2003-present) with Rachel Ermineskin, a speaker of the Siksik? dialect, and Beatrice Bullshields, a speaker of the Kain?? dialect. My fieldwork combines a variety of methods, including elicitation, conversation practice, and story collecting. 1 Here and throughout the dissertation I use term matrix clause to refer to clauses that are not subordinate. I opt for this term over the term root clause to avoid confusion with lexical (i.e., N or V) roots. 3 Regarding elicitation methods, these involve asking questions of the consultants in an interview-like setting. The questions are designed to elicit sentences and/or short monologues, as well as grammaticality judgments; they involve translation tasks (both from English to Blackfoot, and Blackfoot to English), as well as the use of pictures, stories, and other props that prompt the consultant to provide sentences, or judge the grammaticality of sentences I provide. The use of pictures was particularly important for eliciting judgments about c-command relations (see Chapter 7); here, I had the consultant match pictures depicting scenarios with sentences describing the scenarios. In all elicitation settings, I attempt to establish a clear discourse context, either verbally or via pictures and props. Regarding conversation practice, I participated in a Blackfoot conversation group at UBC, in which linguistics graduate students and professors met regularly with a consultant (Beatrice Bullshields) and played language games designed to increase our vocabulary and conversational abilities. Word and phrase lists were compiled, as well as transcripts of our meetings, and these are included in the corpus I draw on for the data in this dissertation. Regarding story collecting, the consultants have occasionally told stories, either traditional folktales or personal narratives, that I have recorded and analysed, either on my own or in collaboration with other members of the Blackfoot research group at UBC (cf. Bullshields et al. 2008). The generalizations presented in this dissertation reflect the consultants? judgments based on my fieldwork and are not always convergent with those presented in Frantz?s (1991, 2009) Blackfoot Grammar. When there are discrepancies between my consultants? judgments and the generalizations presented in the grammar(s), these discrepancies are noted. In addition to my fieldwork data, I draw on material from texts. The texts are four traditional Blackfoot stories that are part of a larger collection of narrative texts recorded and made available through the Niitsitapiisini: Our Way of Life exhibit at the Glenbow Museum in Calgary, Alberta. Each story is 4 transcribed in Blackfoot, with English and French translations and an accompanying audio recording. In the dissertation, transcriptions are presented as in the original texts; the morphological analysis and glossing is my own. Drawing on data from elicitation, conversation practice, stories I have collected, as well as texts I have analysed, this dissertation makes a contribution to the collection of Blackfoot language materials. 1.3. Theoretical Assumptions In this section, I outline my theoretical assumptions. In ?1.3.1, I outline my assumptions regarding syntactic categories, and in ?1.3.2, I outline my assumptions regarding syntactic dependencies. In ?1.3.3, I discuss my assumptions regarding the morphology-syntax interface. In ?1.3.4, I show how these three sets of assumptions define a model for mapping linguistic objects onto syntactic structure that sets the course for this dissertation. 1.3.1. Syntactic Categories: The Universal Spine Hypothesis The model of syntactic categories that I adopt is based on the UNIVERSAL SPINE HYPOTHESIS (cf. Wiltschko to appear b, Wiltschko and D?chaine 2010) The main premise of this framework is that there is a fixed and universal ordering of functional categories that dominate lexical categories, comprising what is referred to as the UNIVERSAL SPINE. The universal spine has both verbal and nominal instantiations, as given in (1a) and (1b), respectively. (1) a. [CP [IP [AspP [vP [VP]]]]] b. [KP [DP [?P [nP [NP]]]]] (1a) is the verbal spine and (1b) is the nominal spine. All languages employ these structures; linguistic objects map onto positions in the universal spine, yielding hierarchical structures corresponding to clauses and nominal expressions. A linguistic object may map onto a head position in the spine, or it may modify a head position. Any given head (X) can combine with a phrase (YP), and the resulting complex phrase 5 can further combine with other elements (ZP) without changing the categorial identity of that complex expression. (2) XP 3 ZP 3 X YP Crucially, the Universal Spine Hypothesis distinguishes between CATEGORIES (which correspond to the heads of the structures in (1)) and WORD CLASSES (which can be thought of as language-particular instances of how categories are instantiated). Languages vary in their inventories of word classes, and in which word classes associate with which categories, but under this hypothesis, all languages associate linguistic objects with a fixed set of categories. To give an example, some languages have a word class known as articles, whereas others do not have articles but they have demonstratives, and still others have both.2 In some of these languages, both articles and demonstratives associate with the category of D (cf. Abney 1987). Thus, word classes cannot be equated with categories; different word classes can associate with a single category. Moreover, it is not universally true that articles and demonstratives associate with D in all languages. For example, in many languages, demonstratives co-occur with determiners (which arguably associate with D) and function as adnominal modifiers (e.g., Hungarian, Greek, and many others, cf. Alexiadou et al. 2007). Even in English, the word class of articles is argued to partition into those that associate with D (e.g., the) versus those that associate with Num (or ? in my model), (e.g., a), cf. Lyons 1999; Ghomeshi 2003). These examples show that we cannot map a given linguistic object onto the syntactic spine on the basis of its word class alone. In other words, the generalization is that categories cannot be defined by word class membership. How, then, can we define categories? According to the Universal Spine Hypothesis, each layer in the structures in (1) is associated with a dedicated syntactic function. These functions are integral to the 2 I revisit the relationship between articles, demonstratives and the syntactic category D in Chapter 4. 6 layer itself; each layer of both the verbal and nominal spines shares the same core function. In (3), I give the spines with their associated functions. A brief explanation of each layer and its associated function follows. (3) CP LINKING KP 6 6 IP ANCHORING DP 6 6 AAspP VIEWPOINT ?P 6 6 vP CLASSIFICATION nP 6 6 The CLASSIFICATION LAYER of the clause is associated with thematic role assignment and event structure; v provides a structure for the thematic arguments of the predicate, e.g., agent and patient (e.g., Hale and Keyser 1993; Chomsky 1995). The nominal equivalent of v is n. Like vP, nP provides an architecture for the arguments of the noun; it can introduce an external argument (a possessor) and can license an internal argument. The classification layer is often associated with Aktionsart, or inner aspect, which classifies verbs according to their event structure (cf., Macdonald 2008; Travis 2010) and its nominal equivalent, SEINSART, which classifies nouns according to countability (cf. Bach 1986, Rijkhoff 1991). According to the Universal Spine Hypothesis, classification is not universally tied to semantic notions such as, e.g., telicity or count/mass; we predict that languages will vary in how verbs and nouns are classified (cf. Lowenstamm 2007 for arguments that n can associate with gender). As for Blackfoot, classification is tied to grammatical animacy. This is observed in Chapter 2, in which I develop an analysis of Blackfoot?s vP and in Chapter 3, in which I develop an analysis of Blackfoot?s nP. The VIEWPOINT LAYER of the clause provides a connection between the event and a perspective on that event. In the verbal spine, the viewpoint layer has been often associated with outer aspect or viewpoint aspect (to be distinguished from inner aspect or Aktionsart), which provides a temporal perspective on the event (e.g., perfective aspect views the event in its entirety from the outside; 7 imperfective views an interval of the event from within, cf. Comrie 1976; Smith 1997). In Chapter 2, I present an analysis of AspP in Blackfoot, which I argue is associated with the direct and inverse suffixes. It has been argued that Number is the nominal equivalent of Aspect (Megerdoomian 2008, and references therein). Following D?chaine and Wiltschko (2002) and others, I adopt the label ?P instead of NumP to refer to this head, as a way of dissociating the function of this head from its semantic content.3 In Chapter 3, I argue that the nominal plural suffixes in Blackfoot map onto ?. In the verbal spine, the ANCHORING LAYER connects the event situation to the utterance situation, a composite of the speech act participants, the moment of speaking, and the location of the speech act (cf. Cowper 2005). Although traditionally associated with Tense, Ritter and Wiltschko (2005, 2009, to appear) have argued that the category of INFL4 can also be associated with person and location. The nominal equivalent of INFL is D; just as INFL connects the event to the utterance, D connects the individual to the utterance. I discuss D in Blackfoot in Chapter 4, and INFL in Chapter 5. The LINKING LAYER situates the phrase (i.e., the clause or nominal expression) in a superordinate structure and/or the larger discourse. In the verbal spine, linking is associated with C, which can signal subordination, or can connect the clause to the larger discourse (e.g., by signalling the type of speech act, or by encoding information structural properties of the clause, cf. Rizzi 1997). In the nominal domain, linking is associated with K (Case), which connects the nominal expression to the clause (cf. Ogawa 2001). The linking layer in Blackfoot is associated with the number suffixes that appear on nouns and verbs; this is discussed in Chapters 3 and 6. An important point to note is the parallelism between the verbal and nominal spines. This parallelism has been long observed in generative grammar; in his (1970) Remarks on Nominalization, Chomsky observed that clauses and arguments can project parallel syntactic structures. This has been 3 In the literature (and in this dissertation), ??? is used in two ways: (i) to refer to a nominal functional head, and (ii) to refer to nominal features, e.g., person, number, gender. 4 Since Chomsky (1995) (based on Pollock 1989) it has been customary to refer to the head of the clause as T(ense), rather than its earlier label INFL(ection). However, once we divorce the anchoring category from its temporal content, then the label INFL is more appropriate. 8 elaborated and debated in various directions (cf. Abney 1987, Grimshaw 1990, Szabolski 1983, Ogawa 2001, and many others). Abstracting away from the details, the main point to be gleaned from this is that there is a parallelism between verbal and nominal spines. Under the Universal Spine Hypothesis, this parallelism is taken as evidence that there are syntactic layers with dedicated functions, and these functions are not integrally related to semantic content. Moreover, the Universal Spine Hypothesis allows for the possibility that a given linguistic object could associate with a particular position in the spine (e.g., head of the linking layer), but be neutral with respect to the nominal/verbal distinction. Under this view, category-neutral roots (e.g., English run) associate with the lexical layer of the spine but are not inherently categorized as nominal or verbal.5 The prediction is that there should be instances of neutrality in the functional layers as well. In Chapters 3 and 6, I argue that Blackfoot possesses a linguistic object (the proximate suffix ?wa) that maps onto the head position of the linking layer of the spine, but is not categorized as nominal or verbal. I refer to this head as LINK. Notably, in postulating that UG makes available a fixed set of functional categories, the Universal Spine Hypothesis situates itself ?in between? two other generative models of syntax: Minimalism (Chomsky 1995) and Cartography (Cinque 2002; Rizzi 2004). In its strictest form, Minimalism does away with functional categories (as primitives) and as such cannot account for the parallelisms observed between verbal and nominal structures. Cartographic approaches, on the other hand, allow for a wide range of categories, and the sheer number of proposed categories in the cartographic tradition obscures the parallelism between verbal and nominal structures. 1.3.2. Syntactic Dependencies The ways in which linguistic objects combine with one another is constrained by their categorial identity; linguistic objects associate with positions in the spine according to their categorial identity and this 5 Under some hypotheses, all roots are inherently category-neutral, and categorization takes place in the syntax (cf. Marantz 1997; Borer 2004; Arad 2005). However, following Armoskaite (2011), I assume that in at least some languages (including Blackfoot), roots are inherently categorized. See Chapter 3 for discussion. 9 dictates that they combine in a fixed order. Once linguistic objects associate with the spine, they may subsequently undergo various derivations, constrained by a dependency relation referred to as AGREE, and defined here narrowly as a dependency of a head on its Specifier (i.e., the sister of the phrase consisting of the head and its complement, represented by ZP in (2)). Thus I am adopting a mixed model, drawing on both representational and derivational approaches to syntax. Admittedly, such a model places a heavy burden on UG; it stipulates that UG supplies both the universal spine and the Agree mechanism. However, in my approach, the spine and Agree are responsible for different things: the spine dictates the combination of functional layers with one another based on the properties of their heads, and Agree is responsible for the combination of heads with their arguments inside the functional layers. Whether and how these two approaches could be integrated into a single model (representational or derivational) is a matter for future research. Agree is a feature-matching mechanism: the features (F) of a head must agree with the features of its Specifier. Following Chomksy (1995), I assume that any given head can bear a combination of interpretable and uninterpretable features. Interpretable features contribute to the semantic interpretation of the sentence and they project from the head to the whole phrase, as schematized in (4). (4) XP[F] 3 3 X[F] An uninterpretable feature must be checked by a matching interpretable feature in a local (i.e., Specifier-head) configuration. The uninterpretable features of the head are checked by the interpretable features of the Specifier, establishing a dependency relation between the head and the Specifier. 10 Both uninterpretable and interpretable features can be either valued (V) or unvalued.6 This yields a four-way contrast in features, as summarized in Table 1.1 below. Table 1.1. Grammatical Features Interpretable Uninterpretable Valued [F:V] [uF:V] Unvalued [F] [uF] Unvalued features are those that range over an inflectional class (e.g, person or number), but do not have a particular value (e.g., 1st person or plural number) associated with them. Valued features, on the other hand, are those that have a particular value associated with them. Uninterpretable valued features must be checked by a matching interpretable feature with the same value. Uninterpretable unvalued features, on the other hand, are checked by a matching feature, and may or may not be concurrently valued by the interpretable feature. (Uninterpretable unvalued features that are not valued by the interpretable feature remain unvalued.) These three options are schematized below. (5) a. XP b. XP c. XP 3 3 3 YP[?V] 3 YP[F:V] 3 YP[F] 3 X[uF:?V] X[uF] X[uF] In (5a), the head X bears an uninterpretable valued feature [uF:?V]; it is checked by the matching interpretable feature [F:?V] on the Specifier YP. In (5b), the head X bears an uninterpretable unvalued feature [uF]; it is checked and valued by the feature [F:V] on the Specifier YP. Finally, in (5c), the head X bears an uninterpretable unvalued feature [uF] that is checked by a matching interpretable unvalued feature [F] on the Specifier YP. Thus, the four-way contrast in grammatical features observed in Table 1.1. yields three different Agree possibilities: checking a valued feature (5a), checking and valuing an 6 This is contra Chomsky (2001) who assumes a two-way contrast between uninterpretable unvalued features, and valued interpretable features. However, it is inspired by Pesetsky and Torrego?s (2004) proposal for a more nuanced approach to grammatical features; they suggest that features can vary according to un/interpretability as well as whether they have a semantic value or are purely formal. 11 unvalued feature (5b), or checking (but not valuing) an unvalued feature (5c). In Chapter 5, I make use of all three of these Agree relations in my analysis of the arguments in Spec, IP. 1.3.3. The Morphology-Syntax Interface To this point, we have seen that the structural organization of linguistic objects is constrained by their categorial identities (i.e., which position in the spine they map onto), as well as subsequent Agree relations. Additionally, they may be constrained by linearization principles related to their morphological form. Linearization of morphemes, like Agree, involves derivations that take place after the initial mapping of linguistic objects onto the spine. In this section, I introduce my assumptions regarding the morphology-syntax interface (i.e., the interface between linguistic objects and the spine). The model I adopt is influenced by Rice (2000) and Pittman and Compton (2010), both of whom argue for a syntactic view of word formation in polysynthetic languages (Athapaskan languages and Inuit, respectively). Abstracting away from variations in the implementation, these authors share the insights that, in these languages, a phonological word may correspond to a syntactic phrase, and that the linearization of morphemes inside the word is conditioned by syntactic principles. I propose that the same is true of Blackfoot. The model I adopt in this dissertation centers around three main ideas: (i) Morphemes are linguistic objects comprised of sound-meaning bundles, and structure-building is the process of associating these bundles with the syntactic spine.7 (ii) The linearization of bound morphemes reflects this association, but depending on whether the morpheme is lexicalized as a suffix, prefix, or clitic, it is subject to a different linearization algorithm. (iii) Spell-out operations can take place post-syntactically (i.e., after linguistic objects associate with the spine, and after any Agree operations) and can affect the linearization of morphemes and morpheme combinations. 7 In other words, I do not assume that the syntax manipulates feature bundles that are associated with a phonetic form post-syntactically, e.g., Distributed Morphology (Halle and Marantz 1993). 12 In what follows, I introduce my assumptions about the linearization of suffixes, prefixes, and clitics, as well as post-syntactic spell-out operations. 220.127.116.11. Linearization of Suffixes In Blackfoot, suffixes tend to associate with head positions. As such, they exhibit Head Movement Constraint effects (cf. Travis 1984); they undergo cyclic head movement starting with the lexical root up to the highest head in the spine, as schematized below. (6) a. 3 Suffix3 3 Suffix2 3 Suffix1 3 Root b. [ Root ? Suffix1 ? Suffix2 ? Suffix3 ] In (6a), there are four morphemes, a root and 3 suffixes, and they associate with head positions in the syntactic spine. Beginning with the root, these morphemes undergo cyclic head movement resulting in the order in (6b). As such, they conform to Baker?s (1985) Mirror Principle; the suffix furthest from the root is highest in the structure. 18.104.22.168. Linearization of Prefixes In Blackfoot, prefixes tend not to be heads but modifiers. As such, they do not show Head Movement Constraint effects, and their surface order reflects a different linearization algorithm than that of the suffixes. (This type of asymmetry between suffixes and prefixes is common crosslinguistically, cf. di Sciullo 2005). Here and throughout the dissertation, I remain largely agnostic about the precise mechanisms that determine the relative ordering of prefixes. (The one place where I discuss prefixes is Chapter 5, in which 13 I argue that the person prefixes8 are in Spec, IP.) However, abstracting away from the details, we can observe that, at least in some cases, the linear order of Blackfoot?s prefixes correlates with scope relations. An example illustrating the scope relations between prefixes is given below. (7) a. Na Doris k?ksinsskaka?pssiwa. ann-wa D ikak-inasskaka?pssi-wa DEM-PROX D only-be.tidy.AI-PROX ?Doris is only tidy.? (i.e., not friendly and clever) b. Na Doris m??tsikaksinsskaka?pssiwa. maat- > ikak- ann-wa D maat-ikak-insskaka?psii-wa DEM-PROX D NEG-even-be.tidy.AI-PROX ?Doris is not even tidy.? (i.e., or friendly or clever) c. Na Doris kaks??nskaka?pssiwa. ikak- > sa- an-wa D ikak-sa-insskak-a?pssi-wa DEM-PROX D only-NEG-tidy-be.AI-PROX ?Doris is only not tidy person.? (i.e., she?s friendly and clever; she?s only not tidy) In (7a), the preverb ikak- is interpreted as ?only,? and in (7b), ikak- is preceded by the negative prefix maat- and it is interpreted as ?even.? This is similar to languages such as Dutch, German, Spanish, Finnish, and Swedish in which a word used to express ?even? is only licensed under the scope of negation (Rullmann 1997, and references therein). By analogy we can conclude that the maat- prefix in (7b) takes scope over the ikak- prefix. In (7c), we see the opposite case: when ikak- is followed by a negative prefix (here sa-), it is not interpreted as ?even? but ?only.? If the ?even? reading only obtains when ikak- scopes under negation, then we can conclude that in this example sa- does not take scope over ikak-. The contrast between (7b) and (7c) tells us that the linearization of prefixes can inform us about their relative scope relations. This is particularly evident if we compare (7c) with (8) below. (8) Anna Carmelle ??nikksiistapoowa k?msaikaksaapi?si p??taa. ann-wa C ii-inikk-miistap-oo-wa kam-sa-ikak-yaapi-hsi piitaa DEM-PROX C IC-angry-away-go.AI-PROX if-NEG-even-see.AI-CONJ eagle ?Carmelle will leave angry if she doesn?t even see an eagle.? 8 In fact, it isn?t clear to me whether the person prefixes are best analysed as prefixes or as clitics. In Chapter 2, I develop language-specific diagnostics for distinguishing clitics from affixes, and according to these, the person prefixes pattern as prefixes. However, see also D?chaine (1999) and Stacy (2004), who argue that they are clitics. 14 In (8), the same negative prefix as in (7c), sa-, appears on the subordinate verb, and here it precedes ikak-. In this case, the ?even? reading? of ikak- obtains, suggesting that here sa- scopes over ikak-. The generalization is as follows: When sa- precedes ikak-, it scopes over it; when sa- follows ikak-, it does not. (See Bliss 2010b for an analysis of ikak-). In short, the linear order of prefixes correlates with scope relations. If we assume that scope relations are structurally determined (e.g., Aoun and Li 1993), then this suggests that (at least some of) the prefixes are linearized according to their relative height in the structure, as schematized below. (9) a. XP 3 Prefix1 3 YP 3 Prefix2 3 ZP 3 Prefix3 3 Root b. [Prefix1 ? Prefix2 ? Prefix3 ? Root] In (9a), the three prefixes are in specifier/adjunct positions, and their relative syntactic positions directly determine their linear position in (9b). 22.214.171.124. Linearization of Clitics Clitics are an ?in-between? class: they are neither affixes nor words (cf. Zwicky 1977) and are not subject to the same linearization algorithms as affixes. In particular, the linearization of clitics is determined by phonological factors, and not syntactic ones such as c-command. In Chapter 2, I discuss some Blackfoot-specific diagnostics for distinguishing between clitics and affixes, and I demonstrate that Blackfoot?s clitics attach outside the prosodic domain of affixes, as schematized below. (10) [[Prefix(es) ? Root - Suffix(es)] ? Clitic(s)] 15 The position(s) of Blackfoot?s clitics is dictated by the phonology, and does not inform us about the clitic?s syntactic position. This becomes relevant in Chapters 6 and 7, in which I argue that (some) enclitic pronouns occupy argument positions in the clause. Thus, depending on the grammatical function the enclitic fills, it has a different syntactic position. Nevertheless, it always attaches in the same place in the surface string. 126.96.36.199. Post-Syntactic Spell-out Rules Following the mapping of linguistic objects to the syntactic spine, and any subsequent derivations, there may be mismatches between the surface string and the syntactic structure. Specifically, morphemes that map onto the spine may not appear in the surface string, and conversely, morphemes that are not mapped on to the spine may appear in the string. I assume that these mismatches reflect post-syntactic spell-out restrictions. These spell-out restrictions may (i) constrain the co-occurrence of morphemes in the syntax, or (ii) require the spell-out of an extra-syntactic morpheme. Regarding the first possibility, consider the structure in (11a), in which three morphemes ? 3A, 3B, and 3C - are each associated with different syntactic positions. However, as shown in (11b), only one of these can be spelled out in the surface form. A spell-out restriction blocks the three from co-occurring. (11) a. 3 4 3 3A 3 3B 3 3C 3 2 3 1 b. [ 1 ? 2 ? 3 ? 4 ] 8 A B C One-to-many mappings of this sort have been proposed for Blackfoot, as well as other Algonquian languages. For example, Bliss et al. (2010a,b) proposed that the direct and inverse suffixes that all appear in the same morphological position on verbs in both Blackfoot and Nishnaabemwin are split across three 16 different syntactic positions. The details of the analysis are not relevant here, but what is relevant is that the mapping between morphology and syntax is not isomorphic; one morphological position maps on to many syntactic positions. Bliss et al. propose a series of spell-out restrictions to account for this mapping. Furthermore, the direct/inverse morphology has been proposed to be split across multiple syntactic positions in Western Naskapi (Brittain 1999) and Plains Cree (D?chaine and Reinholtz 1997, 2008), as well as in Proto-Algonquian (Oxford 2012). Additionally, Slavin (2012) observes that verb class finals can map onto different syntactic positions in Oji-Cree. This suggests that spell-out restrictions that block the co-occurrence of multiple syntactic heads in the surface morphology are pervasive in Algonquian languages. In Chapter 3, I appeal to this mechanism in my account of plural nominal expressions. Regarding the second possibility (i.e., that an extra-syntactic morpheme be spelled-out), there are (at least) two types of spell-out rules to consider, illustrated in (12) and (15) below. (12) a. 3 4 3 3 3 2 3 1 b. [ 1 ? 2 ? 3 ? 4 ? 5] In (12), there are four morphemes associated with the syntactic spine (a), but five morphemes in the surface string (b). In this case, morpheme 5 is spelled out post-syntactically. An example that may qualify as an extra-syntactic element like morpheme 5 is the so-called ?connective I? in Blackfoot (Frantz 2009, p. 77). This is a morpheme that appears between certain prefix-verb concatenations without any apparent phonological or syntactic motivation. An example is given below. (13) a. Passk??t! passkaa-t dance-imp ?Dance!? 17 b. ??passkaayaawa. a-i-passkaa-yi-aawa IMPF-i-dance-PL-3PL.PRN ?They are dancing.? (Frantz 2009: 77) (14) Aapoy??naattsiwa. aapoyiinaattsi-wa be.brown.II-PROX ?It is/was brown.? (Frantz and Russell 1995: 11) In (13a), the verb stem is consonant-initial, and in (13b), it is prefixed with the imperfective marker a-. In this context, a morpheme ?i- intervenes between the prefix and the stem. Note that this morpheme is not the result of epenthesis; (14) shows that /a+p/ sequences are not prohibited by the phonology. However, it has no identifiable semantic or syntactic function. The second type of many-to-one relation between morphemes and syntactic positions is schematized in (15). (15) a. 3 4 3 3 3 2 3 1 b. [ 1 ? 2A ? 3 ? 4 ? 2B ] In (15), morpheme 2 is associated with a single syntactic position but is spelled out twice in the surface string (represented as 2A and 2B). I refer to this as COPYING. The notion of copying can account for concordial agreement systems, the insight being that if an identical morpheme appears on multiple elements, then one instantiation reflects its ?true? syntactic position, and the others are copies, spelling out an agreement relation (cf. D?chaine et al. 2013; Wiltschko 2009).9 In Chapter 4 I return to the issue of copying, as it pertains to the nominal inflection that appears on demonstratives. 9 These analyses draw on Wiltschko and D?chaine?s (2010) model of Interface Syntax, which allows the phonetic form of a morpheme to associate with the spine either early or late in the derivation. An early association results in agreement with copying, whereas a late association results in agreement without copying. My analysis differs, insofar as I assume that morphemes always associate early, and thus copying reflects a post-syntactic operation. 18 In summary, morphemes can map onto syntactic positions in various ways, depending on whether they are suffixes, prefixes, or clitics. Moreover, post-syntactic spell-out restrictions can result in one-to-many or many-to-one relations between the morphemes and syntactic positions. 1.3.4. Mapping Morphemes onto the Syntactic Spine The Universal Spine Hypothesis defines a methodological paradigm for a researcher looking at the syntax of a given language (cf. Wiltschko 2011). For any given language, we can ask the question: Which linguistic objects associate with which functional categories? The framework I am adopting gives us two families of diagnostics for determining this association: diagnostics based on syntactic position and diagnostics based on syntactic function. Regarding tests for syntactic position, tests for structural asymmetries can establish dependency relations within a clause, and the linear order of words and morphemes can inform syntactic structure. Notably, these tests allow us to establish relative but not absolute structural positions. Under the framework adopted here, tests for syntactic function allow us to establish absolute positions. A linguistic object can be located in a particular layer in the spine if it fulfills the core function associated with the syntactic category that instantiates that layer. Throughout this dissertation, I combine tests for syntactic position and syntactic function in order to map Blackfoot?s linguistic objects onto the syntactic spine. 1.4. Proposal I began my dissertation research wanting to know how to represent the structure of a sentence like that in (16). (16) Na Beth ??kohkottsikooni?pa anni ksikkok??wayi. ann-wa Beth yaak-ohkott-ikooni-?p -wa ann-yi ksikkok??wa-yi DEM-PROX Beth FUT-ABL-take.down.TI-1:INAN-PROX DEM-INAN tent-INAN ?Beth can take down the tent.? The standard assumption is that in the English translation of (16), the ARGUMENT EXPRESSIONS ? ?Beth? and ?the tent? ? map onto ARGUMENT POSITIONS, as in (17). 19 (17) IP 3 DPi 3 5 I vP Beth can 3 <DPi> 3 v VP 3 V DP take down 5 the tent In Blackfoot, as in numerous other languages, the mapping from argument expressions to argument positions is less transparent. Unlike in English, argument expressions in Blackfoot can be moved around (18a) split apart (18b) or omitted (18c), obscuring the mapping between the argument expressions and the positions they are associated with.10 (18) a. ??kohkottsikooni?pa anni ksikkok??wayi na Beth. yaak-ohkott-ikooni-?p-wa ann-yi ksikkok??wa-yi ann-wa Beth FUT-ABL-take.down.TI-1:INAN-PROX DEM-INAN tent-INAN DEM-PROX Beth ?Beth can take down the tent.? b. Na ??kohkottsikooni?pa Beth anni ksikkok??wayi. ann-wa yaak-ohkott-ikooni-?p-wa Beth ann-yi ksikkok??wa-yi DEM-PROX FUT-ABL-take.down.TI-1:INAN-PROX Beth DEM-INAN tent-INAN ?Beth can take down the tent.? c. ??kohkottsikooni?pa anni ksikkok??wayi. yaak-ohkott-ikooni-?p-wa ann-yi ksikkok??wa-yi FUT-ABL-take.down.TI-1:INAN-PROX DEM-INAN tent-INAN ?She can take down the tent.? Languages with this clustering of properties (i.e., languages that exhibit patterns exemplified by (18)) are labelled as non-configurational (cf. Hale 1983), but non-configurationality has long been understood to be a cover term for languages with a surface appearance that obscures hierarchical relations between constituents, rather than a literal label meant to suggest that any language in fact lacks hierarchical structure. 10 The examples in (18) are a subset of the possible ways in which argument expressions can be moved or split apart. See Chapter 7 for a more detailed discussion. 20 The goal of this dissertation is to uncover the hierarchical relations in Blackfoot that are obscured by the non-configurational properties of the language, and to determine the reason(s) why Blackfoot appears to be non-configurational. My main thesis is that non-configurationality in Blackfoot is a conspiracy resulting from properties of Blackfoot?s argument-typing system, and in particular the proximate/obviative contrast, referred to as OBVIATION. In Algonquian languages, obviation is a type of reference-tracking morphology, disambiguating between multiple 3rd persons in a clause. At most one 3rd person referent can be marked proximate in a clause; all others are marked obviative. Often obviation is correlated with notions of discourse saliency; proximate 3rd persons are typically more salient than obviative ones.11 In Blackfoot, the morphological reflexes of obviation surface on argument expressions and on clauses. Examples are given below. (19) a. ?yissksimmaawa oma imit??wa. a-yissksimaa-wa om-wa imitaa-wa IMPF-carry.load.AI-PROX DEM-PROX dog-PROX ?That dog (PROX) is a pack dog.? (lit: it carries loads) b. ?yissksimmaayini omi imit??yi. a-yissksimaa-yini om-yi imitaa-yi IMPF-carry.load.AI-OBV DEM-OBV dog-OBV ?That dog (OBV) is a pack dog.? (lit: it carries loads) In (19a), both the verb and the argument expression (the noun and the demonstrative) are marked proximate by virtue of the suffix ?wa. In (19b), the argument expression (the noun and the demonstrative) is marked obviative by virtue of the suffix ?yi, and the verb is marked obviative by virtue of the suffix ?yini. In this dissertation, I argue that obviation is a grammatical construct rather than a primitive. In particular, it is constructed from three ingredients, corresponding with the three obviation suffixes in (19): -wa, -yi, and ?yini. First, regarding the proximate suffix ?wa, I argue that, regardless of whether it appears on clauses or argument expressions, it associates with the highest functional head in the spine. I refer to its syntactic 11 For example, proximate marking has been said to code the perspective-holder (cf. Dahlstrom 1991; Russell 1991; M?hlbauer 2008) and/or the discourse topic (cf. Russell 1996; Junker 2004; Genee 2009; Goddard 1984, 1990). The discourse functions associated with obviation vary across Algonquian languages. I discuss this in Chapter 8. 21 category as LINK to reflect the fact that it is in the linking layer of the spine, but is neutral with respect to which spine it appears in (nominal or verbal). I demonstrate that any expression that is marked with ?wa can be construed as either a predicate or an argument, as shown below. (20) a. (Anna) ??sttokimaawa. (ann-wa) a-isttokimaa-wa (DEM-PROX) IMPF-drum.AI-PROX ?S/he is drumming.? b. Anna ??sttokimaawa ?kaomatap??wa. ann-wa a-isttokimaa-wa akaa-omatap-oo-wa DEM-PROX IMPF-drum.AI-PROX PERF-begin-go.AI-PROX ?The drummer has just left.? OR ?S/he is drumming (and) she has just left.? (21) a. (Oma) p??sa. (om-wa) poos-wa (DEM-PROX) cat-PROX ?That is a cat.? b. Oma p??sa ?yo?kaawa. om-wa poos-wa a-yo?kaa-wa DEM-PROX cat-PROX IMPF-sleep.AI-PROX ?That cat is sleeping.? OR ?That is a cat (and) it is sleeping.? In (20a), the verb is suffixed with ?wa and it is interpreted as a predicate; in (20b), the same verb can be interpreted either as an argument or as a predicate. Similarly in (21a), the noun is suffixed with ?wa and interpreted as a predicate; in (21b), the same noun can be interpreted either as an argument or as a predicate. I propose that the predicate/argument flexibility observed in examples like (20) and (21) reflects a structural ambiguity observed with LINKPS: LINK may or may not license null pro in its Specifier. In the former case, the resulting structure is interpreted as a saturated predicate, i.e., a complete proposition. In the latter case, the resulting structure is interpreted as an argument. 22 (22) a. LINKP b. LINKP 3 3 pro 3 LINK DP LINK DP -wa 6 -wa 6 p??s p??s ?S/he is a cat.? ?cat? Although they can be interpreted semantically as arguments of the predicate, LINKPs are not arguments in the syntactic sense, i.e., they do not occupy argument positions but are instead adjoined to the clause. Turning now to the obviative suffix ?yi, I propose that it associates with the functional head K in the nominal spine and its function is to link the argument expression to the clause. In other words, -yi encodes the dependency relation between the argument expression and the clause. Being categorized as K, -yi is conceived of in terms of a case marker. Just as nominative and accusative case in an Indo-European language code KPs as appearing in subject and object positions, -yi also codes the nominal expressions as appearing in argument positions. However, whereas K in a nominative/accusative system co-varies with grammatical functions, K in Blackfoot is generalized and does not co-vary. As such, the same instantiation of K appears on every argument expression in the clause. This is schematized below. (23) Indo- European Blackfoot KNOM = subject K (-yi) = subject KACC = object K (-yi) = object The other type of morphology subsumed under the umbrella of obviation is the number morphology that appears on the right edge of the verbal complex. Examples are given below. (24) a. ?yissksimmaawa oma imit??wa. a-yissksimaa-wa om-wa imitaa-wa IMPF-carry.load.AI-PROX DEM-PROX dog-PROX ?That dog (PROX) is a pack dog.? (lit: it carries loads) b. ?yissksimmaayini omi imit??yi. a-yissksimaa-yini om-yi imitaa-yi IMPF-carry.load.AI-OBV DEM-OBV dog-OBV ?That dog (OBV) is a pack dog.? (lit: it carries loads) 23 c. ?yissksimmaayi omiksi imit??ks. a-yissksimaa-yi om-iksi imitaa-iksi IMPF-carry.load.AI-PL DEM-PL dog-PL ?Those dogs are pack dogs.? (lit: they carry loads) (24a) and (24b) are repeated from (19) above, and in (24c), we see that the plural suffix ?yi contrasts with proximate ?wa and obviative ?yini. As discussed above, the proximate suffix ?wa maps onto the functional head LINK, and just as proximate argument expressions are syntactically independent, so are proximate clauses. They are necessarily matrix clauses, and they do not require an argument expression to saturate the predicate. As for obviative ?yini and plural ?yi, I demonstrate that these map onto C, the highest functional head in the verbal spine. Further, I argue that they are number agreement markers; -yini is singular agreement in C and ?yi is plural agreement in C. Whereas proximate ?wa does not require an argument in its Specifier, obviative ?yini (and plural ?yi) do. (25) a. LINKP b. CP c. CP 3 3 3 3 KP[SG] 3 KP[PL] 3 -wa -yini -yi [uSG] [uPL] To summarize, in this dissertation I decompose Blackfoot?s obviation system into three different pieces: (i) proximate marking, the morphological encoding of a requirement that a phrase be syntactically independent, (ii) obviative marking on nouns, a generalized case marker, signalling that a nominal expression associates with an argument position, and (iii) number agreement in C. Each of these contributes independently to Blackfoot?s appearance of non-configurationality; proximate argument expressions are not mapped onto argument positions, obviative argument expressions are not coded according to their different argument positions, and number agreement in C creates an ?extra? argument position not cued to grammatical function. Thus, my dissertation contributes to the growing body of literature showing ways in which different languages can conspire to be non-configurational (e.g., Legate 24 2002; Pensalfini 2004), and provides further evidence against the view of non-configurationality as a macro-parameter (contra Chomsky 1981, Hale 1983; Baker 1996). Moreover, I show that, once obviation is taken out of the picture, Blackfoot bears robust similarities to other (configurational) languages. In particular, I look at nominal expressions that are neither proximate nor obviative, and I show that they pattern like other bare nouns in a variety of unrelated languages. For example, similar to NPs in Hungarian (Farkas and de Swart (2003), Niuean (Massam 2001), Chamorro (Chung and Ladusaw 2004) and many other languages, bare nouns in Blackfoot have a close syntactic relation with V, and are necessarily narrow-scoping and non-specific. The fact that, without obviation, Blackfoot nominal expressions bear strong similarities to nominal expressions in other languages provides support for the claim that obviation is the main contributor to Blackfoot?s non-configurational profile. 1.5. Outline of Dissertation Following this chapter, the dissertation has seven additional chapters. This section gives a brief synopsis of each chapter. In Chapter 2, I provide background information on the Blackfoot language, focusing on the morphosyntactic properties of nominal expressions and the verbal complex. I also discuss variability in the phonetic realization of the proximate and obviative suffixes, and I argue that, despite the fact that, for some speakers, proximate and obviative suffixes are completely inaudible, they are nevertheless active in the grammar. Additionally, I develop an analysis of the syntactic correlates of the grammatical functions in the language. I argue that the grammatical functions of subject and object in Blackfoot are both associated with external argument positions of v heads, and I introduce a grammatical function that is associated with the Spec, AspP position and that I refer to as the Point-of-View (PoV) holder. The PoV holder is determined by the direct/inverse system: if the verb is coded as direct the subject is the PoV holder, and if the verb is inverse the object is the PoV holder. 25 In Chapter 3, I explore the syntax of nominal expressions. I argue that, depending on whether a nominal expression is inflected as proximate singular, obviative singular, plural, or is uninflected (i.e., bare), it has different internal and external syntax. First, regarding proximate singular nominal expressions, I argue that the proximate suffix ?wa maps onto the highest functional head in the spine, but is neutral with respect to whether it associates with a nominal or verbal spine. I adopt the label LINK to refer to this head, and I propose that LINKPS are adjoined to the clause; they do not occupy argument positions. As for obviative singular nominal expressions, I argue that obviative ?yi also maps onto the highest head in the spine, but it is restricted to nominal phrases. Hence, it is K. I propose that K is a generalized case marker; it is required on all argument expressions in case positions, but does not co-vary with grammatical function. As such, every argument expression inside the clause is a KP. Regarding bare nouns, I argue that these are pseudo-incorporated: they as phrasal complements to V and they are not assigned case. I discuss the syntactic and semantic properties of pseudo-incorporated nominal expressions and I show that they pattern like pseudo-incorporated nominal expressions in a variety of languages: they are restricted to an immediately post-verbal position, their internal syntax is restricted (they are nPs), and they are narrow-scoping and non-specific. Finally, I argue that plural nouns are structurally ambiguous; they can be pseudo-incorporated in Comp, VP, in which case they are ?Ps. Alternatively, they can occupy argument positions, in which case they are KPs. In the latter case, I propose a spell-out restriction that blocks the realization of the case suffix ?yi on plural nouns. In Chapter 4, I consider the syntax of demonstratives. They are required with the subject and object argument expressions, and as such have a distribution expected of D, which in many languages is required to turn a nominal predicate into an argument. I argue that Blackfoot demonstratives are not D heads but phrases that map onto the Spec, DP position. I discuss the various morphemes that comprise the demonstratives, and although I do not present an analysis of their internal structure, I suggest that their morphological composition is amenable to a syntactic treatment. As for the syntactic function of the demonstratives, I propose that they serve the anchoring function associated with the D layer of the spine, connecting the nominal expression to the utterance situation. 26 In Chapter 5, I look at the person prefixes that appear at the left edge of the verbal complex. I claim that these appear in Spec, IP and, like the demonstratives in Spec, DP, they serve an anchoring function, connecting the clause to the utterance situation. Adapting Ritter and Wiltschko?s (to appear) analysis of Blackfoot INFL as person-based, I propose that INFL bears an uninterpretable person feature and I claim that this restricts the inventory of items that can appear in Spec, IP. Specifically, I argue that Spec, IP arguments only appear in realis (i.e., real-world) contexts, and I show that the person prefixes are restricted to realis clauses in Blackfoot (see also D?chaine and Wiltschko 2010). Moreover, I argue that arguments in Spec, IP necessarily have a temporal component, even though INFL itself does not have temporal (i.e., tense) features. Finally, I show that, in addition to the person prefixes, Spec, IP can be occupied by an evidential prefix na-. Like an expletive subject (e.g., English it), the evidential prefix does not refer to an event participant, but unlike expletive subjects, na- is not contentless; it has deictic content. I argue that this follow from the fact that the uninterpretable person features on INFL require that arguments in Spec, IP have deictic content. In Chapter 6, I look at the number suffixes that appear at the right edge of the verbal complex. I argue that these map onto the highest functional head in the verbal spine. I develop and apply language-specific diagnostics for each head in the verbal spine, and I show that the number/obviation suffixes do not meet the diagnostics for v, Asp, or INFL. However, they do meet the diagnostics for C: they are sensitive to clause-typing and illocutionary force. As argued in Chapter 3, proximate ?wa associates with LINK, a head that is neutral with respect to whether it heads a nominal or verbal spine, and allows for predicate/argument flexibility. The other two number suffixes, obviative ?yini and plural ?yi associate with C and have number agreement features that are checked by an argument in Spec, CP. In Chapter 7, I consider from various angles the question of whether Blackfoot can be considered a non-configurational language. First, I take Hale?s (1983) diagnostics for non-configurationality ? extensive null anaphora, free word order, and discontinuous expressions ? and I apply them to Blackfoot. I show that Blackfoot exhibits an asymmetry; whereas proximate expressions can be freely omitted and moved around, obviative and plural ones cannot. Second, I consider whether Blackfoot 27 meets the criteria for a Pronominal Argument (PA) language (cf. Jelinek 1984, Baker 1991, 1996). Here again, I observe an asymmetry: proximate argument expressions meet the criteria, but obviative and plural ones do not. Finally, I look at the hierarchical organization of the arguments in the clause, and using various c-command tests I demonstrate proximate arguments asymmetrically c-command obviative ones. This is consistent with the claim that proximate argument expressions are clause-external adjuncts. Finally, in Chapter 8, I situate Blackfoot within a broader cross-Algonquian context. I point out that, whereas in Blackfoot, proximate and obviative morphology maps onto the highest functional layer in the spine, this is not uniformly the case across Algonquian. I look at the highest functional layer of the clause, CP, in various other Algonquian languages, and. I demonstrate that the functional material associated with the CP layer in the other Algonquian languages is located lower in the clause in Blackfoot. Nevertheless, the core function of this layer ? linking ? is maintained across languages. Additionally, I compare the obviation system of Blackfoot with that of other Algonquian languages. I discuss the various discourse functions that have been associated with obviation, and I propose that they share a common thread: they encode discourse in/dependence. I suggest that discourse in/dependence in Blackfoot?s obviation system is not lexically encoded, but arises by virtue of the syntactic properties associated with the proximate and obviative markers. I conclude by speculating that this may also be true in other Algonquian languages. 28 CHAPTER 2 BACKGROUND ON BLACKFOOT 2.1. Introduction The goal of this chapter is to provide an overview of the properties of the Blackfoot grammar that are relevant to this dissertation. The description draws heavily on Frantz?s (1991, 2009) reference grammar. Not intended to be a complete description (or analysis) of Blackfoot grammar, the grammatical sketch presented includes references to research on the same phenomena, in Blackfoot and/or related Algonquian languages. This section proceeds as follows: in ?2.2 is a profile of the language, and in ?2.3 is an introduction to the morphosyntactic properties of nominal expressions. In ?2.4 is an overview of the morphosyntactic properties of the verbal complex, and in ?2.5, I discuss the phonetic realization of the proximate and obviative suffixes, which appear on both nouns and verbs and are discussed in detail in Chapters 3 and 6. In ?2.6, I present a structural analysis of Blackfoot?s grammatical functions. Finally, in ?2.7 I conclude. 2.2. Language Profile Blackfoot consists of four mutually intelligible dialects, spoken on three reserves in Southern Alberta and one reservation in Northwestern Montana. In Alberta, the three dialects are Siksik? (aka Blackfoot), Kain?? (aka Blood), and Piikani (aka Peigan), and in Montana, the dialect is Blackfeet. The population of Blackfoot speakers is less than 10,000; it includes very few (if any) first language learners, and few monolingual speakers (Russell and Genee 2006). Frantz (2009) reports that the past twenty years have seen the language being used with less frequency, resulting in an increase of sub-dialects and idiolects. Moreover, speakers often report a distinction between ?old Blackfoot? (spoken by people in their seventies and upwards) and ?new Blackfoot? (spoken by people in their forties to sixties). To date, documentation on these two varieties has focused on phonetic differences (cf. Bortolin and 29 McLennan 1995; Kaneko 1999), and it has been observed that the old/new characterization reflects a continuum of language change, as opposed to two distinct language varieties (cf. Van Der Mark 2003). Blackfoot is a member of the Algonquian language family, of which there are three geographically-defined sub-groupings: Central Algonquian (including the Cree and Ojibwe dialects), Eastern Algonquian (including Micmac and Passamaquoddy), and Plains Algonquian. Blackfoot, along with Arapaho and Cheyenne, is part of the Plains Algonquian sub-group (Lewis 2009). Blackfoot is thought to be the most divergent of the Algonquian languages (cf. Goddard 1974), having separated from Proto-Algonquian earlier than other Algonquian languages (cf. Proulx 1989). Its genetic affiliation within the Algonquian language family is yet unclear. Like other Algonquian languages, Blackfoot can be defined typologically as a polysynthetic head-marking language. A clause consists minimally of a VERBAL COMPLEX, i.e., an inflected verb stem and may or may not contain nominal expressions representing arguments and/or adjuncts to the clause. As will be discussed in detail in Chapter 7, overt nominal expressions are in most cases optional, are relatively unrestricted in their linear order, and may be discontinuous. In what follows, I briefly discuss the form and distribution of nominal expressions, and then I give an overview of the morphology that comprises the verbal complex. 2.3. Nominal Expressions This section gives an introduction to Blackfoot?s nominal expressions. The syntax of nominal expressions is discussed in Chapter 3, and a detailed survey of the types of nominal expressions that can fulfill the various grammatical functions is presented in Appendix A. Nominal expressions consist minimally of an independent pronoun or noun, and the latter may be modified by a demonstrative or numeral or both. Other types of modifiers, such as adjectives or quantifiers, are expressed as nominal prefixes or as verbal predicates. Nouns can also be marked for possession; the person prefixes that function as possessors are the same as the prefixes that appear on verbs and they are discussed in ?2.4.1 below. 30 Nouns are lexically specified as either animate or inanimate, a grammatical distinction that does not necessarily reflect ontological distinctions of animacy.12 Nouns can be inflected for number (singular versus plural), and animate singular nouns can also be inflected as either proximate or obviative. Inanimate singular nouns are inflected as obviative; they cannot be inflected as proximate. The proximate/obviative distinction is neutralized in the plural. The inflectional suffixes that appear on nouns are presented in Table 2.1. Table 2.1. Nominal Inflection Animate Inanimate Singular Proximate -wa -- Obviative -yi Plural -iksi -istsi As shown in Table 2.1., only grammatically animate nouns may be marked with the proximate singular suffix ?wa, but both grammatically animate and inanimate nouns may be marked with the obviative singular suffix ?yi. Animate and inanimate nouns are morphologically distinguished in the plural. Examples illustrating these different inflectional suffixes are given in (1) and (2) below. As shown in (1) and (2), the nominal inflectional suffixes also appear on demonstratives. (The syntax of demonstratives is discussed in Chapter 4.) (1) a. Oma s??hkomaapiwa ??ksspitaawa. om-wa saahkomaapi-wa iik-sspitaa-wa DEM-PROX boy-PROX INTNS-be.tall.AI-PROX ?That boy (PROX) is tall.? b. Omi s??hkomaapiyi ??ksspitaayin?yi. om-yi saahkomaapi-yi iik-sspitaa-yini-ayi DEM-OBV boy-OBV INTNS-be.tall.AI-OBV-3SG.PRN ?That boy (OBV) is tall.? 12 Whereas grammatically inanimate nouns in Blackfoot are always ontologically inanimate, grammatically animate nouns may be ontologically animate (?sentient?) or inanimate (?non-sentient?), cf. Bliss (2005a). Examples of non-sentient animate nouns include istt?an ?knife? and po?taatsis ?stove.? Only sentient animate nouns may function as the subject in a transitive clause. While animacy is a common feature of Algonquian, the deployment of animacy contrasts across the Algonquian languages is not uniform (cf. M?hlbauer 2008). 31 c. Omiksi s??hkomaapiks ??ksspitaayaawa. om-iksi saahkomaapi-iksi iik-sspitaa-yi-aawa DEM-PL boy-PL INTNS-be.tall.AI-PL-3PL.PRN ?Those boys are tall.? (2) a. Omi n??pioyisi ??ksspiiwa. om-yi naapioyis-yi iik-sspii-wa DEM-INAN house-INAN INTNS-be.tall.AI-PROX ?That house is tall.? b. Omistsi n??pioyists ??ksspiiyaawa. om-istsi naapioyis-istsi iik-sspii-yi-aawa DEM-PL house-PL INTNS-be.tall.AI-PL-3PL.PRN ?Those houses are tall.? In (1), the grammatically animate noun s??hkomaapi ?boy? is inflected with the proximate singular suffix ?wa (a), the obviative singular suffix ?yi (b), and the plural suffix ?iksi (c). In (2), the grammatically inanimate noun n??pioyis ?house? is inflected with the suffix ?yi (glossed as INAN, but formally identical to the obviative suffix that appears on animate singular nouns) and the plural suffix ?istsi. Regarding proximate and obviative marking, this serves a reference-tracking function, disambiguating between multiple 3rd persons in a clause. At most one 3rd person referent can be marked proximate in a clause; all other singular nouns are marked obviative, as illustrated below. (3) a. Mat?nni na Leo ??hpokinihkimiiwa ni nit?ni. matonni ann-wa Leo ii-ohpok-inihki-m-yii ann-yi n-itan-yi yesterday DEM-PROX Leo IC-ACCOMP-sing.AI-TA-3-4-PROX DEM-OBV 1-daughter-OBV ?<esterday /eo sang with my daughter.? b. * Mat?nni na Leo ??hpokinihkimiiwa na nit?na. matonni ann-wa Leo ii-ohpok-inihki-m-yii ann-wa n-itan-wa yesterday DEM-PROX Leo IC-ACCOMP-sing.AI-TA-3-4-PROX DEM-PROX 1-daughter-PROX intended: ?<esterday /eo sang with my daughter.? (4) a. Anna Beatrice ??staawa annisk Irvine ? ann-wa B a-isstaa-wa ann-yi-hk I DEM-PROX B IMPF-want.AI-PROX DEM-OBV-INVIS I ? om??hkitsspiyissi omi p??sskaan. om-aahk-it-ihpiyi-hs-yi om-yi paisskaa-n-yi 3-MOD-LOC-dance.AI-CONJ-OBV DEM-INAN dance-NOM-INAN ?Beatrice wants Irvine to dance at the dance.? 32 b. * Anna Beatrice ??staawa annahk Irvine ? ann-wa B a-isstaa-wa ann-wa-hk I DEM-PROX B IMPF-want.AI-PROX DEM-PROX-INVIS I ? om??hkitsspiyissi omi p??sskaan. om-aahk-it-ihpiyi-hs-yi om-yi paisskaa-n-yi 3-MOD-LOC-dance.AI-CONJ-OBV DEM-INAN dance-NOM-INAN intended: ?Beatrice wants Irvine to dance at the dance.? In (3a), the subject, na Leo, is proximate (as evidenced by the proximate marking on the demonstrative), and the object ni nit?ni ?my daughter? is obviative. (3b) shows that it is ungrammatical for both the subject and object to be marked proximate. In (4), the subject of the matrix clause, anna Beatrice, is proximate and the subject of the subordinate clause, annisk Irvine, is obviative. (4b) shows that it is ungrammatical for both to be marked proximate. The constraint against multiple proximate 3rd persons extends to possessed nouns as well. Singular nouns possessed by a 3rd person possessor are obligatorily obviative, regardless of whether the possessor is proximate or obviative, as shown below. (5) a. Ama nit?mitaama iy??stapoksk??siwa. am-wa nit-omitaa-m-wa i-yiistap-okska?si-wa DEM-PROX 1-dog-POSS-PROX IC-run.away.AI-PROX ?My dog ran away.? b. Anna kit?mitaama iy??stapoksk??siwa. ann-wa kit-omitaa-m-wa i-yiistap-okska?si-wa DEM-PROX 2-dog-POSS-PROX IC-run.away.AI-PROX ?<our dog ran away.? c. *Anna ot?mitaama iy??stapoksk??siwa. ann-wa ot-omitaa-m-wa i-yiistap-okska?si-wa DEM-PROX 3-dog-POSS-PROX IC-run.away.AI-PROX intended: ?Her dog ran away.? d. Anni ot?mitaami iy??stapoksk??siyin?yi. ann-yi ot-omitaa-m-yi i-yiistap-okska?si-yini-ayi DEM-OBV 3-dog-POSS-OBV IC-run.away.AI-OBV-3PRN ?Her dog ran away.? In (5a) and (5b), the possessor is a local person (1st and 2nd, respectively), and the possessed noun is marked as proximate. In (5c), the possessor is 3rd person, and it is ungrammatical for the possessed noun 33 to be marked as proximate. (5d) is the grammatical alternative to (5c), in which the possessed noun is marked as obviative. As for the interpretational difference between proximate and obviative nominal expressions, in the Algonquianist literature this is often discussed in the context of discourse functions. Although there is considerable variation across Algonquian, in all the languages proximate nominal expressions are thought to be more discourse-salient than obviative ones in some sense (e.g., the proximate nominal expression is the perspective-holder and/or discourse topic, cf. Dahlstrom 1991; Genee 2009; Goddard 1984, 1990; Junker 2004; M?hlbauer 2008; Russell 1991, 1996). Specifically regarding the discourse functions of the proximate/obviative contrast in Blackfoot, Genee (2009) claims that proximate marking appears on discourse topics, i.e., those nominal expressions that refer to what the sentence or larger discourse is about. Throughout this dissertation I focus largely on the syntactic properties of proximate and obviative morphology, but I also discuss their discourse functions in Chapter 8. Nominal expressions can fulfill a variety of different grammatical functions in the clause. In this dissertation, I make reference to the SUBJECT and OBJECT, which in ?2.6 I correlate with vP-internal positions.13 I distinguish between INDEXED and UNINDEXED objects; the former show agreement on the verb whereas the latter do not.14 Additionally, I make reference to OBLIQUES , which I define as nominal expressions that are introduced by an adpositional prefix (discussed in ?188.8.131.52 below). 2.4. Verbal Complex In morphological terms, the verbal complex can be described in terms of a template, with designated slots for each type of affix. The template is given in Figure 2.1, and is followed by a brief discussion of each slot. 13 In the Algonquianist literature, the subject and object are often referred to as the ACTOR and GOAL, respectively (cf. Bloomfield 1962; Hockett 1966; Wolfart 1973). 14 In the Algonquianist literature, indexed and unindexed objects are often referred to as the PRIMARY OBJECT and SECONDARY OBJECT, respectively (cf. Rhodes 1990). 34 Figure 2.1. Morphological Template for Verbal Complex Person?(Prefixe(s))?Root?(Noun)?Final?Direct/Inverse?Clause Type-Person/Number?Number?Clitic(s) Stem Suffixes 2.4.1. Person Prefixes The leftmost position in the verbal complex is where the person prefixes appear. Examples are given in (6). (6) a. Nits??maahkiaaki. nit-yaamaahki-aaki 1-sweep-AI ?I swept.? b. Kits??maahkiaaki. kit-yaamaahki-aaki 2-sweep-AI ?<ou swept.? c. ?ots??maahkiaakissi. ot-yaamaahki-aaki-hs-yi 3-sweep-AI-CONJ-OBV ?(when) s/he swept.? Regarding 3rd person ot-, it is restricted in distribution in ways that 1st person nit- and 2nd person kit- are not. Whereas nit- and kit- appear whenever required in both matrix clauses and subordinate conjunct clauses (see ?2.4.5), ot- is restricted to conjunct clauses, as well as matrix transitive clauses in which an obviative 3rd person acts on a proximate one. In environments in which the 3rd person prefix does not appear, the prefix slot is empty, or may be occupied by a prefix na- (discussed below.) As in other Algonquian languages, there is only one morphological slot for the person prefixes. The 2nd person prefix kit- is used whenever there is a 2nd person argument in the clause (7), and the 1st person prefix nit- is used whenever there is a 1st person argument, but no 2nd person (8). The 3rd person ot- is used only if neither of the arguments is 1st or 2nd person (and if the distributional criteria outlined in the preceding paragraph are met), (9). 35 (7) a. Kits??htooki. kit-yooht-o-oki 2-hear-TA-2:1 ?<ou heard me.? b. Kits??htoo. kit-yooht-o-o 2-hear-TA-1:2 ?I heard you.? c. Kits??htowawa. kit-yooht-o-a-wa 2-hear-TA-DIR-PROX ?<ou heard him/her.? d. Kits??htooka kit-yooht-o-ok-wa 2-hear-TA-INV-PROX ?S/he heard you.? (8) a. Nits??htowawa. nit-yooht-o-a-wa 1-hear-TA-DIR-PROX ?I heard him/her.? b. Nits??htooka nit-yooht-o-ok-wa 1-hear-TA-INV-PROX ?S/he heard me.? (9) a. ?yoohtoyiiwa ii-yooht-o-yii-wa IC-hear-TA-3:4-PROX ?HePROX heard herOBV.? b. Ots??htooka. ot-yooht-o-ok-wa 3-hear-TA-INV-PROX ?SheOBV heard himPROX.? The person prefixes also appear on nouns to mark the possessor. The primary difference between the verbal and nominal prefixes lies in the encoding of the inclusive. On verbs, the inclusive is marked by the 36 absence of a prefix (along with a suffix ?o?p, see ?2.4.6); on nouns, the inclusive is marked with a second person prefix (along with a suffix ?(i)nnoon). 15 This is shown below. (10) a. ??maahkiaakio?p. yaamaahki-aaki-o?p sweep-AI-INCL ?We (INCL) swept.? b. kitsaam??hkimaa'tsinnoon kit-yaamaahkimaa?tsis-innoon 2-broom-INCL ?our (INCL) broom? The forms of the prefixes in (6)-(10) are the long form prefixes; there are also short forms, as shown in Table 2.2. Table 2.2. Person Prefixes Long forms Short forms 1st nit- n- 2nd kit- k- 3rd ot- w- In Chapter 5, I discuss the distribution and interpretation of the long and short form prefixes (see also Bliss and Gruber 2011a, b). Here I present some examples of the short form prefixes; in particular, they are required in the context of the perfect (11), certain modals (12), and inalienable possession (13). (11) a. Nik??yo?kaa. n-ikaa-yo?kaa 1-PERF-sleep.AI ?I have slept.? b. * Nitsik??yo?kaa. nit-ikaa-yo?kaa 1-PERF-sleep.AI intended: ?I have slept.? 15 The ?o?p form on verbs is also used in impersonal constructions, whereas the kit-?-(i)nnoon form on nouns is used only with inclusive reference. 37 (12) a. K??hksikkamihpiyi. k-aahk-ikkam-ihpiyi 2-MOD-if-dance ?You might dance.? b. Kit??hksikkamihpiyi ?kikat???ssiksinaasstopi. kit-aahk-ikkam-ihpiyi (?kit-kata?-ssiksinaasi-htopi) 2-MOD-if-dance ( 2-NEG-break.leg.AI-UNREAL) ?You would dance (?if you hadn?t broken your leg).? (13) a. oks?ssts w-iksisst-yi 3-mother-OBV ?his/her mother? b. * ots?ksissts ot-iksisst-yi 3-mother-OBV intended: ?his/her mother? In the Siksik? dialect only (cf. Frantz 1991, 2009), the leftmost slot in the verbal complex can also be occupied by the prefix na-. In Chapter 5, I develop an analysis of na- as an evidential marker (see also Bliss and Ritter 2007, 2009). Illustrative examples of na- are given below. (14) 1?tssks?ni?p anna imit??wa n??siksipiiw?yi ni John. nit-ssksini-?p ann-wa imitaa-wa na-siksip-yii-wa-ayi ann-yi J 1-know.TI-1:INAN DEM-PROX dog-PROX EVID-bite.TA-3:4-PROX-3PRN DEM-OBV J ?I know the dog bit -ohn.? (15) N??sootaawa. na-i-sootaa-wa EVID-rain.II-PROX ?It rained.? (16) Kiist?wa ki niist?wa n???wato'p anni nap?yini. kiistowa ki niistowa na-oowato-?p ann-yi napayin-yi 2SG.PRN and 1SG.PRN EVID-eat.TI-1:INAN DEM-INAN bread-OBV ?<ou and I ate the bread.? 2.4.2. Verbal Prefixes Following the person prefixes, there may appear a range of different verbal prefixes that encode temporal, aspectual, modal, quantificational, adpositional, and adverbial meanings or functions. (These are referred 38 to as PREVERBS in the Algonquianist literature, cf. Bloomfield 1927). Multiple prefixes can occur in a single verbal complex, and there are various restrictions on the co-occurrence and ordering of verbal prefixes. A discussion of these restrictions is beyond the scope of this thesis. However, to facilitate readability of the data, in what follows, I give an overview of the types of verbal prefixes in Blackfoot. I do not discuss prefixes in combination with each other, and the list of prefixes discussed here is not exhaustive. 184.108.40.206. Tense/Aspect/Modality Verbs that do not have a prefix that signals a temporal/aspectual/modal meaning are typically interpreted with past time reference16 and perfective aspect (Armoskaite 2008). An example is given below. (17) 1itoksk??si. nit-okska?si 1-run.AI ?I ran.? The addition of a verbal prefix can disambiguate temporal and/or aspectual reference. For example, the prefix ?- signals imperfective aspect (cf. Dunham 2007, 2008), the prefix y??k- signals future time reference, and the prefix ikaa- signals the perfect. Examples are given below. (18) Nit?okska?si. nit-a-okska?si 1-IMPF-run.AI ?I am running.? (19) Nit??kokska?si. nit-yaak-okska?si 1-FUT-run.AI ?I will run.? 16 Frantz (2009: 16) claims that unmarked verbs always receive a past time interpretation; Ritter and Wiltschko (2004) report that unmarked verbs are ambiguous between past and present time readings. In my own fieldwork, unmarked verbs are typically interpreted as having past time reference; however, a present time reading can be elicited as well. 39 (20) Nik?okska?si. n-ikaa-okska?si 1-PERF-run ?I have run.? Regarding the perfect marker ikaa- (and its word-initial allomorph akaa-), Frantz (2009: 34) identifies this morpheme as a perfective marker, but I analyse it as a perfect marker, the difference being that perfective aspect involves event completion and/or the perspective on the event as a whole, whereas the perfect involves temporal anteriority and current relevance (cf. Portner 2003 and references therein). I refer to the following generalizations to diagnose ikaa- as a perfect marker: (i) ikaa- can co-occur with imperfective ?- (if it were a perfective marker, it would be incompatible with the imperfective); (ii) ikaa- is used to express the meaning of ?already,? which has a similar semantic function to the perfect (cf. Mittwoch 1993), and (iii) like the English perfect, ikaa- is incompatible with the adverb meaning ?yesterday? (Iatridou et al. 2002). These three points are illustrated in (21)-(23) below. (21) Amo n?naawa ?kaa?paistotakiwa n??pioyii. amo ninaa-wa akaa-a?p-a-istotaki-wa naapioyis-i DEM man-PROX PERF-around-IMPF-work.AI-PROX house-NONPART ?This man has built a house.? (Frantz 2009: 35 (m)) (22) Na Myaani ak??kamotaa. ann-wa M akaa-ikamotaa-wa DEM-PROX M PERF-give.birth.AI-PROX ?Mary has already had her baby.? (23) (*Matonni) na Myaani ?kaihkitaawa sitokihkiitaan. matonni ann-wa myaani akaa-ihkitaa-wa sitokihkiitaan yesterday DEM-PROX Mary PERF-bake.AI-PROX pie ?(<esterday) Mary has baked a pie.? In addition to the imperfective, future, and perfect prefixes, a prefix ??-17 can also signal past time reference. An example is given below. 17 The ??- prefix is different from other prefixes in that it substitutes for the leftmost vowel of the verb complex (not including the person prefix). As such, it is related to INITIAL CHANGE, a morphophonological phenomenon found in many Algonquian languages that affects vowel quality at the left edge of the verb complex (cf. Taylor 1967; Proulx 2005). I discuss initial change in more detail in Chapter 8. 40 (24) Nits??kska?si. nit-ii-okska?si 1-IC-run.AI ?I ran.? The two prefixes aahk-18 and ohkott- signal epistemic and deontic modality, respectively. The epistemic modal is ambiguous between necessity and possibility readings. A detailed discussion of Blackfoot?s modal system can be found in Louie (in prep), and Reis Silva (in prep). (25) N??hkokska?si. n-aahk-okska?si 1-MOD-run.AI ?I might/must run.? (26) Nitohk?ttokska?si. nit-ohkott-okska?si 1-ABL-run.AI ?I can run.? 220.127.116.11. Quantifiers and Other Scope-Taking Prefixes The universal quantifier ohkana- is a verbal prefix that can associate with either the subject or the object. If both arguments are plural, then the quantifier is ambiguous between subject- and object-associated interpretations. (27) Nitohkan??hpommatoo?pinnaaniaawa. nit-ohkana-ohpommatoo-?p-innaan-yi-aawa 1-all-buy.TI-1:INAN-1PL-3PL-3PL.PRN ?We bought all of them.? OR ?We all bought them.? (Frantz 1991: 88) Other scope-taking prefixes include the negation prefix m??t-19 as well as various focus-sensitive operators such as ikak- ?only?20 and matt- ?also.? (xamples are given below. Detailed descriptions of these (and other) scope-taking prefixes can be found in Bliss (2010b), and Louie (2008, 2011a, b). 18 In addition to aahk-, Frantz and Russell (1995) list the forms aahkaPa?p and aahksikkaPa?p. These are phonological variants of a morphologically complex form, aahk + ikkam- ?if? a?p-. 19 The verbal prefix m??t- marks clausal negation. A second negation prefix sa- appears closer to the verb stem and marks predicate negation (cf. Louie 2008). 41 (28) Nim??tokska?si. ni-maat-okska?si 1-NEG-run.AI ?I didn?t run.? (29) Nik?kokska?si. n-ikak-okska?si 1-only-run.AI ?I only ran.? (30) Nimattoksk??si. nit-matt-okska?si 1-also-dance.AI ?I also danced.? 18.104.22.168. Adpositions Adpositions are verbal prefixes that introduce oblique nominal expressions in the clause. Examples are given below. (31) Nits?tsooyi anni it??sooyo?pi. nit-it-ioyi ann-yi itaisooyo?p-yi 1-LOC-eat DEM-INAN table-INAN ?I ate at the table.? (32) Iihtsip?kihkiniiw anni Leo otoht??piksspi. ii-oht-ipakihkin-yii-wa ann-yi L ot-iiht??piksspi -yi IC-INSTR-strike.on.head.TA-3:4-PROX DEM-OBV L 3-hammer-OBV ?She hit /eo over the head with her hammer.? (33) Napay?ni nomohpiywatoo?pa ?mihka i?ksisakoyihka. napayin-i n-omohp-iowatoo-?p-wa om-yi-hka i?ksisako-yi-hka bread-NONPART 1-ASSOC-eat.TI-1:INAN-PROX DEM-INAN-INVS meat-INAN-INVS ?I ate the meat with bread.? (Frantz 2009: 92) In (31), the spatiotemporal adposition it- introduces a location, in (32) oht- introduces an instrument, and in (33), omohp- introduces an associate. In each of the examples, the nominal expression introduced by the adposition does not control agreement on the verb. Throughout this dissertation I use the term OBLIQUE to refer to nominal expressions that are introduced by an adposition and do not control agreement on the verb. (These differ from nominal expressions that are introduced by an applicative verb 20 In the context of negation, ikak- is interpreted as ?even,? not ?only,? (cf. Bliss 2010b). 42 final such as benefactive ?omo, which function as direct objects and do control agreement on the verb, cf. Bliss 2007, 2010a). Adpositions are referred to as LINKERS by Frantz (1991, 2009) and are cognate with what are referred to as RELATIVE ROOTS in other Algonquian languages (e.g., Rhodes 2010). Detailed discussions of Blackfoot adpositions can be found in Bliss (2011, 2012a), Bliss et al. (2013), Hanson et al. (2010), and Louie (2009). 22.214.171.124. Other Verbal Prefixes Various other meanings are expressed via prefixes. Included in this list are adverbial prefixes as well as prefixes that encode meanings typically associated with RESTRUCTURING PREDICATES (infinitival verbs such as ?try? or ?start? that lack certain clausal properties, cf. Wurmbrand 2001). Illustrative examples are given below. (34) Nitsikkamoksk??si. nit-ikkam-okska?si 1-fast-run.AI ?I ran Tuickly.? (35) Nitss?akokska?si. nit-ssaak-okska?si 1-try-run.AI ?I tried to run.? (36) Nitsstsimoksk??si. nit-sstsim-okska?si 1-reluctant-run.AI ?I reluctantly ran.? 2.4.3. Verb Stems Verb stems are composed of a root and a stem-forming suffix referred to as a FINAL, plus an optional incorporated noun.21 Each is discussed in turn below. 21 In the Algonquianist literature, the verb stem is described as consisting of an INITIAL, MEDIAL, and FINAL (cf. Bloomfield 1946), corresponding to what I am here calling the root, incorporated noun, and final, respectively. In Blackfoot, incorporated nouns are not necessarily in medial position in the verb stem; some precede the root, as shown in (38) and (40). 43 126.96.36.199. Roots The root contributes the main lexical content of the verb complex. Some examples are given below. (37) a. N?tsstsisttaki. nit-ihtsistt-aki 1-swallow-AI ?I swallowed (something).? b. Nit??psstaki. nit-a?psst-aki 1-wave-AI ?I waved.? c. N?tsskssaki. nit-ihkss-aki 1-dry-AI ?I dried something.? For a detailed discussion of Blackfoot roots, see Armoskaite (2011). 188.8.131.52. Incorporated Nouns Incorporated nouns are optional in the verb stem. A detailed investigation of Blackfoot noun incorporation is pending, but as a starting point it can be observed that some incorporated nouns precede and some follow the verb root, and they encode body parts as well as other entities (cf. Barrie and Dunham 2008; Dunham 2009). Examples are given below. (38) Nits?ssapaapino?toka. nit-sap-aapin-o?t-o-ok-wa 1-in-eye-grasp-TA-INV-PROX ?He poked me in the eye.? (Dunham 2009: 4) (39) Nits??paksik?niooka. nit-ii-ipak-ika-ini-ok-wa 1-IC-hit-foot-TA-INV-PROX ?She struck me on the legs.? (Frantz and Russell 1995: 55) (40) Anna Leo ??nnokaikskimaa annohk. ann-wa L a-innoka-ikskim-aa-wa annohk DEM-PROX L IMPF-elk-hunt-AI-PROX now ?/eo is elk-hunting today.? 44 184.108.40.206. Finals Verb stems obligatorily include a final, which encodes transitivity and animacy. In the Algonquianist tradition, finals are classified as in Table 2.3. Table 2.3. Verb Finals II Inanimate Intransitive Subject = Inanimate AI Animate Intransitive Subject = Animate TI Transitive Inanimate Object = Inanimate TA Transitive Animate Object = Animate Throughout the dissertation, I refer to the first two types of finals (II and AI) as forming MORPHOLOGICALLY INTRANSITIVE verbs, and the latter two (TI and TA) as forming MORPHOLOGICALLY TRANSITIVE verbs. In the examples, verb finals are glossed according to the labels in Table 2.3 (e.g., II, AI, TI, TA). In Blackfoot, there are restrictions on which roots can combine with which finals. Few (if any) roots can be used with all four classes of finals, but some can be used with three of the four. An example is given below. (See Armoskaite 2011 for a more detailed discussion of Blackfoot stem formation.) (41) a. N??hkiitaawa. na-ihkiit-aa-wa EVID-bake-AI-PROX ?S/he baked (something).? b. N??hkiitatsiiwa omi pi?kss??. na-ihkiit-at-yii-wa om-yi pi?kssii-yi EVID-bake-TA-3:4-PROX DEM-OBV chicken-OBV ?S/he baked that chicken.? c. N??hkiitatooma omi napay?ni. na-ihkiit-atoo-m-wa om-yi napayin-yi EVID-bake-TI-3:INAN-PROX DEM-INAN bread-INAN ?S/he baked that bread.? The forms of the finals are dependent on the root to which they attach. For example, whereas the root iihkiit ?bake? selects the finals -aa (AI), at (TA), and -atoo (TI), the root ikooki't ?regret the loss? selects a different set of finals: -aki (AI), -mm (TA), and ?i (TI). 45 (42) a. 1??kooki?takiwa. na-ikooki?t-aki-wa EVID-regret.loss-AI-PROX ?She regretted the loss (of something)? b. N??kookimmiiwa anni oks?ssts. na-ikooki?t-mm-yii-wa ann-yi w-iksisst-yi EVID-regret.loss-TA-3:4-PROX DEM-OBV 3-mother-OBV ?She regretted the loss of her mother.? c. 1??kooki?tsima anni ook??wayi. na-ikooki?t-i-m-wa ann-yi w-ookoowa-yi EVID-regret.loss-TI-3:INAN-PROX DEM-INAN 3-home-INAN ?She regretted the loss of her home.? 2.4.4. Direct/Inverse Marking Direct/inverse marking is sometimes referred to as THEME MARKING in the Algonquianist tradition (Frantz 2009). It is obligatory for morphologically transitive verbs, and signals the grammatical relations of the participants in the clause. Descriptively, direct/inverse marking is determined on the basis of the person hierarchy in (43). (43) 1st > 2nd > 3rd proximate > 3rd obviative When a higher-ranking participant is the subject, direct morphology appears on the verb, and when a higher-ranking participant is the object, inverse morphology appears on the verb. For example, the (a) and (b) examples in (44) differ formally only in the direct/inverse marking; the direct suffix -a in (44a) signals that a higher-ranking 1st person is the subject and the lower-ranking 3rd person is the object, and the inverse suffix ?ok in (44b) signals that the 1st person is the object and 3rd person is the subject. (44) a. Nitsik?motsiipiawa. nit-ikamotsiip-i-a-wa 1-rescue-TA-DIR-PROX ?I rescued him/her.? b. Nitsik?motsiipioka. nit-ikamotsiip-i-ok-wa 1-rescue-TA-INV-PROX ?S/he rescued me.? 46 The form of the direct/inverse markers varies depending on the person features of the participants. Table 2.4 gives the direct/inverse markers for matrix clauses.22 Table 2.4. Direct/Inverse Marking in Matrix Clauses Direct Inverse TA Local -o -oki Mixed -a -ok Non-Local -yii -ok TI Local -?p -- Non-Local -m -- As is customary in the Algonquianist tradition, I have referred to the Transitive Animate markers as LOCAL, MIXED, and NON-LOCAL. Local direct/inverse markers are used for 1st and 2nd person only, mixed direct/inverse markers are used for local and non-local persons, and non-local direct/inverse markers are used for 3rd persons only.23 Examples of each are given in (45). (45) 1st > 2nd a. Kits??hkssammo. b. Kits??hkssammoki. kit-ii-ohksssa-mm-o kit-ii-ohksssa-mm-oki 2-IC-pity-TA-1:2 2-IC-pity-TA-2:1 ?I pitied you.? ?<ou pitied me.? 1st > 3rd c. Nits??hkssammawa. d. Nits??hkssammoka. nit-ii-ohksssa-mm-a-wa nit-ii-ohksssa-mm-ok-wa 1-IC-pity-TA-DIR-PROX 1-IC-pity-TA-INV-PROX ?I pitied him/her.? ?S/he pitied me.? 2nd > 3rd e. Kits??hkssammawa. f. Kits??hkssammoka. kit-ii-ohksssa-mm-a-wa kit-ii-ohksssa-mm-ok-wa 2-IC-pity-TA-DIR-PROX 2-IC-pity-TA-INV-PROX ?<ou pitied him/her.? ?S/he pitied you.? 22 Direct/inverse markers also vary according to clause type cf. Bliss et al. (2010a, b). A discussion of the different clause types is in ?2.4.5. 23 Following Bliss et al. (2010a, b), I assume that the mixed series instantiates the ?core? direct/inverse system and the local and non-local series are in some sense peripheral. My glossing conventions reflect this assumption: the ?a and ?ok morphemes are glossed as ?DIR(ect)? and ?INV(erse)? respectively; others are glossed according to the persons involved in the clause (e.g., -o is glossed as ?1:2? to indicate that it is used in clauses with a 1st person acting on a 2nd person). 47 3rd PROX > 3rd OBV g. ??hkssammiiwa. h. Ots??hkssammoka. ii-ohksssa-mm-yii-wa ot-ii-ohksssa-mm-ok-wa IC-pity-TA-3:4-PROX 3-IC-pity-TA-INV-PROX ?ShePROX pitied himOBV.? ?HeOBV pitied herPROX.? The local suffixes are shown in (45a/b); when there is a 1st person subject and 2nd person object, the direct suffix ?o is used, and when there is a 2nd person subject and 1st person object, the inverse suffix ?oki is used. The sentences in (45c-f) provide additional examples of the mixed direct/inverse markers, direct ?a for a 1st or 2nd person subject and 3rd person object, and inverse ?ok for a 3rd person subject and a 1st or 2nd person object. The non-local direct/inverse markers are exemplified in (45g/h). When a proximate 3rd person is the subject, and an obviative 3rd person is the object, a direct suffix ?yii is used, and when an obviative 3rd person is the subject and a proximate 3rd person is the object, the inverse suffix ?ok is used. Plural 3rd persons are not coded as proximate or obviative, and a transitive clause containing two plural 3rd persons can be marked as either direct or inverse (depending on discourse conditions), as shown in (46) below. (46) a. Omiksi imit??ks iihkan??ksisaisskoyiiyaaw omiksi p??siks. om-iksi imitaa-iksi ii-ohkana-oksisaissk-o-yii-yi-aawa om-iksi poos-iksi DEM-PL dog-PL IC-all-chase-TA-3:4-PL-3PL.PRN DEM-PL cat-PL ?The dogs chased all the cats.? b. Omiksi imit??ks otohkan??ksisaisskookiyaaw omiksi p??siks. om-iksi imitaa-iksi ot-ohkana-oksisaissk-o-ok-yi-aawa om-iksi poos-iksi DEM-PL dog-PL 3-all-chase-TA-INV-PL-3PL.PRN DEM-PL cat-PL ?The dogs chased all the cats.? Regarding the Transitive Inanimate series, there are dedicated morphemes for indicating a local person subject acting on an inanimate object, and a non-local person subject acting on an inanimate object. Examples are given in (47). (47) a. N?tssimatoo?pa. nit-ssim-atoo-?p-wa 1-smell-TI-1:INAN-PROX ?I smelled it.? 48 b. K?tssimatoo?pa. kit-ssim-atoo-?p-wa 2-smell-TI-2:INAN-PROX ?<ou smelled it.? c. ??ssimatooma. ii-ssim-atoo-m-wa IC-smell-TI-3:INAN-PROX ?S/he smelled it.? There are no inverse suffixes in the TI series because Blackfoot does not permit non-sentient (and hence inanimate) subjects in transitive clauses. In order to express an event in which a non-sentient referent acts upon a sentient one, a construction is used in which an unspecified (formally local) person functions as the subject, and the non-sentient referent is introduced by the means/instrument adposition oht-. (48) a. * Anna pok?na ??sitoyiiwa anni pook??yi. ann-wa pokon-wa yiisit-o-yii-wa ann-yi pookaa-yi DEM-PROX ball-PROX hit-TA-3:4-PROX DEM-OBV child-OBV intended: ?The ball hit the child.? b. Anna pook??wa iihts??sitowawa anni pok?ni. ann-wa pookaa-wa ii-oht-yiisit-o-a-wa ann-yi pokon-yi DEM-PROX child-PROX IC-INSTR-hit-TA-DIR-PROX DEM-OBV ball-OBV ?(Someone) hit the child with the ball.? A syntactic analysis of the direct/inverse is presented in ?2.6.2. 2.4.5. Clause-Typing Suffixes Following the direct/inverse suffixes are clause-typing suffixes. A summary of Blackfoot?s clause types is given in Table 2.5 below. Table 2.5. Clause Types Clause type Suffix Distribution Matrix Imperative -t commands Independent -- elsewhere Subordinate Subjunctive -iniki conditional/hypothetical Unreal -htopi past counterfactual Conjunct -hs-yi elsewhere 49 As shown in Table 2.5, there are two clause types that are used with matrix clauses: imperative and independent. Imperative clauses are marked with a suffix ?t, and independent clauses are not overtly marked. Examples are given below. (49) Soks?nihkit! sok-inihki-t good-sing.AI-IMP ?Sing well!? (50) Kitsoks?nihki. kit-sok-inihki 2-good-sing.AI ?<ou sang well.? Regarding subordinate clauses, there are three types. Subjunctive clauses are used for expressing conditional and/or hypothetical situations and are marked with the suffix ?iniki (or a variant, depending on the person specification of the arguments). Unreal clauses are used in counterfactual contexts with a past time orientation, and are marked with the suffix ?htopi. Finally, the conjunct is the elsewhere clause type, used for subordinate clauses that are neither subjunctive nor unreal. Conjunct clauses are marked with two suffixes ?hs and ?yi; (in Chapter 3 I develop an analysis of ?yi as the obviative suffix; accordingly, it is glossed as OBV.) Examples of the subordinate clause types are given below. (51) ,kkaP?yo?kainoainiki, nit?akahkayi. ikkam-a-yo?kaa-inoa-iniki nit-yaak-waahkayi if-IMPF-sleep.AI-2PL-SBJN 1-FUT-go.home.AI ?If you (PL) are sleeping, I?ll go home.? (Frantz 2009: 110, (l)) (52) Nits?tss?yoyihtopi, nit?aksoyi ?nnohka. nit-it-say-ioyi-htopi nit-yaak-ioyi annohka 1-LOC-NEG-eat.AI-UNREAL 1-FUT-eat.AI now ?If I hadn?t eaten then, I?d eat now.? (Frantz 2009: 113 (x)) (53) Nits?kstaataa anna John nin??hkohkookssi omi isttoani. nit-ik-staa-t-a-wa ann-wa J nit-aahk-ohkot-ok-hs-yi omi isttoan-yi 1-INTNS-want-TA-DIR-PROX DEM-PROX J 1-MOD-give.TA-INV-CONJ -OBV DEM knife-OBV ?I want -ohn to give me that knife? 50 2.4.6. Person/Number Suffixes In the verbal complex, there are two suffix positions for number marking. The first marks plural number for local persons (1st and 2nd) and the second marks number (and the proximate/obviative contrast) for non-local (3rd) persons. This subsection focuses on the former, which I henceforth refer to as the person/number suffixes. A summary of the inventory of person/number suffixes is given in Table 2.6 and examples are given below. Table 2.6. Person/Number Suffixes Form Meaning Gloss -(hp)innaan24 1st person plural 1PL -(hp)oaa 2nd person plural 2PL -o?p Inclusive (/impersonal) INCL (54) a. Nit??ks??yihpinnaan pis?tsskitaan. nit-yaak-ioyi-hpinnaan pis?tsskitaan 1-FUT-eat.AI-1PL cake ?We (EXCL) will eat cake.? b. Kit??ks??yihpoaa pis?tsskitaan. kit-yaak-ioyi-hpoaa pis?tsskitaan 2-FUT-eat.AI-2PL cake ?<ou (PL) will eat cake.? c. ??ks??yo?p pis?tsskitaan. yaak-ioyi-o?p pis?tsskitaan FUT-eat.AI-INCL cake ?We (INCL) will eat cake.? OR ?Someone will eat cake.? As there is only one morphological slot for person/number suffixes, in local transitive clauses with two plural participants, only the 1st person suffix appears, as shown below. (55) Kitsin??hpinnaan. kit-inoo-o-hpinnaan 2-see.TA-1:2-1PL ?We (EXCL) saw you (SG/PL).? 24 -hp only appears when all participants in the clause are local (1st and 2nd) persons. The syntax of ?hp is discussed in Chapter 5 (see also Ritter and Wiltschko, to appear). 51 2.4.7. Number Suffixes Following the person/number suffixes (which mark plural 1st and 2nd persons) are the number suffixes, which are used exclusively with 3rd persons. There are three such suffixes, as summarized in Table 2.7 and exemplified below. Table 2.7. Number Suffixes Form Meaning Gloss -wa 3rd person (proximate singular) PROX -yi 3rd person plural PL -yini 3rd person obviative singular OBV (56) a. Anna Rosie ??ks??yiwa pis?tsskitaan. ann-wa R yaak-ioyi-wa pis?tsskitaan DEM-PROX R FUT-eat.AI-PROX cake ?Rosie (PROX) will eat cake.? b. Omiksi aak??koaiks ??ks??yiyaawa pis?tsskitaan. om-iksi aakiikoan-iksi yaak-ioyi-yi-aawa pis?tsskitaan DEM-PL girl-PL FUT-eat.AI-3PL-3PL.PRN cake ?Those girls will eat cake.? c. Anna Rosie oks?ssts ??ks??yiyin?yi pis?tsskitaan. ann-wa R w-iksisst-yi yaak-ioyi-yini-ayi pis?tsskitaan DEM-PROX R 3-mother-OBV FUT-eat.AI-OBV-3PRN cake ?Rosie?s mother (OBV) will eat cake.? There is only one morphological slot for the number suffixes. The obviative suffix ?yini has the most limited distribution; it appears only in matrix declarative clauses in which all of the arguments are obviative and animate, as shown in (57) below. The plural suffix ?yi appears in matrix declarative clauses in which there is at least one plural argument but no proximate argument, as shown in (58). Finally, the proximate suffix ?wa is the default suffix, appearing elsewhere (including, for example, matrix declarative clauses with a single 3rd person argument). This is shown in (59). (57) 2tsski?tsokini anni oks?ssts anni ot?mitaami. ot-sski?t-i-ok-yini ann-yi w-iksisst-yi ann-yi ot-imitaa-m-yi 3-frighten-TA-INV-OBV DEM-OBV 3-mother-OBV DEM-OBV 3-dog-POSS-OBV ?Heri dog frightened heri mother.? 52 (58) N??noyiiyi omiksi s??hkomapiks anni ot?mitaami. na-iin-o-yii-yi om-iksi saahkomapi-iksi ann-yi ot-imitaa-m-yi EVID -see-TA-3:4-PL DEM-PL boy-PL DEM-OBV 3-dog-POSS-OBV ?The boys saw her dog.? (59) Nits??noawa oma imit??. nit-iin-o-a-wa om-wa imitaa-wa 1-see-TA-DIR-PROX DEM-PROX dog-PROX ?I saw that dog.? Notably, the inflectional contrasts expressed by the number suffixes on verbs differs from those expressed by the number suffixes on nouns. As discussed in ?2.3 above, in the nominal paradigm the proximate suffix ?wa is restricted to animate singular nouns, but in the verbal paradigm -wa has a wider distribution, functioning as the elsewhere suffix and appearing in contexts where number and obviation contrasts are neutralized. Moreover, whereas the nominal obviative suffix ?yi appears on both animate and inanimate nouns, the verbal obviative suffix ?yini is restricted to animate singular reference only. Finally, whereas the nominal plural suffixes encode animacy distinctions, animacy is neutralized with the verbal plural suffixes. These differences in the inflectional contrasts of the nominal and verbal paradigms are summarized in Table 2.8. Table 2.8. Nominal and Verbal Number Suffixes Compared Nominal paradigm Verbal paradigm Proximate animate singular (-wa) elsewhere (-wa) Obviative singular (-yi) animate singular (-yini) Plural animate plural (-iksi) or inanimate plural (-istsi) plural (-yi) The syntax of the nominal number suffixes is discussed in Chapter 3, and the syntax of the verbal number suffixes is discussed in Chapter 6. 53 2.4.8. Enclitic Pronouns Enclitic pronouns can appear at the right edge of the verbal complex to reference plural and/or singular obviative/inanimate arguments. There are four enclitic pronouns, given in Table 2.9 below.25 Table 2.9. Enclitic Pronouns Form Meaning -?yi Singular -aiksi plural animate -aistsi plural inanimate -aawa plural Enclitics can be distinguished from affixes according to Blackfoot-specific diagnostics. These are summarized in Table 2.10 (see also Fox and Frantz 1979; Frantz 2009, chapter 9). Table 2.10. Affixation versus Encliticization Affixes Enclitics Sensitive to distribution of nominal expression ? ? Stackable ? ? Pitch accent is additive ? ? Regarding the first diagnostic in Table 2.10, agreement affixes appear on the verb regardless of whether a nominal expression is present or not, and regardless of its position with respect to the verb. For example, in the preceding section, we saw that the number suffixes ?wa, -yi, and ?yini index proximate, plural, and obviative arguments, respectively. Regardless of whether the argument expression is preverbal, postverbal, or null, the number suffix appears on the verb. This is illustrated with ?wa in (60). (60) a. Oma saahk?maapiiwa kita'p??sstooka. om-wa saahkomaapii-wa kit-a?p-a-issto-ok-wa DEM-PROX boy-PROX 2-around-IMPF-wave.AI-INV-PROX ?That boy is waving at you.? 25 Frantz (2009: 49) refers to the first three clitics in Table 2.9 as DTP (?Distinct Third Person?) pronouns. He notes that they are used when then there is another 3rd person in the clause, and that they don?t refer to ?major? (i.e., proximate singular) 3rd persons. The fourth clitic, -aawa, similarly cannot refer to proximate singular 3rd persons, but may be used if there are no other 3rd persons in the clause. 54 b. Kita'p??sstooka oma saahk?maapiiwa. kit-a?p-a-issto-ok-wa om-wa saahkomaapii-wa 2-around-IMPF-wave.AI-INV-PROX DEM-PROX boy-PROX ?That boy is waving at you.? c. Kita'p??sstooka. kit-a?p-a-issto-ok-wa 2-around-IMPF-wave.AI-INV-PROX ?S/he is waving at you.? Unlike agreement affixes, enclitics are sensitive to the distribution of the nominal expression. In particular, they are used under two conditions: (i) if the nominal expression is null and (ii) if the nominal expression is preverbal. An example is given in below. (61) a. 0??no?tooyi n?hpapiiyihpiksi. maan-o?too-yi n-ohpapiiyihp-iksi just-arrive.AI-PL 1-relative-PL ?My relatives just arrived.? b. N?hpapiiyihpiksi m??no?tooyaaw. n-ohpapiiyihp-iksi maan-o?too-yi-aawa 1-relative-PL just-arrive.AI-PL-3PL.PRN ?My relatives just arrived.? c. M??no?tooyaaw. maan-o?too-yi-aawa just-arrive.AI-PL-3PL.PRN ?They just arrived.? In short, whereas affixes are not sensitive to the distribution of nominal expressions, enclitics are. Regarding the second diagnostic, affixes are not stackable. For example, there is only one morphological position in the verbal complex for the number suffixes; two or three suffixes cannot be concurrently realized on a single form. Conversely, enclitics are able to stack, with more than one enclitic appearing on a single verb. The precise conditions under which a single verb can host multiple enclitics 55 are yet unclear26, but below are some examples of verbs with multiple enclitics (referencing multiple arguments). (62) Anniksisk nit?kkaiks nit?hkokiyaawaists. ann-iksi-hk n-itakkaa-iksi nit-ohkot-ok-yi-aawa-aistsi DEM-PL-INVIS 1-friend-PL 1-give.TA-INV-PL-3PL.PRN-3PL.PRN ?My friends gave them to me.? (63) Aakaitapi matapiiksi aisaakiohtayissitapiiyaawaiksi. waaka-itapi matapi-iksi a-isaaki-oht-ayissitap-yii-yi-aawa-aiksi many-person person-PL IMPF-still-CONT-keep.sacred-3:4-PL-3PL.PRN-3PL.PRN ?Many people still keep them as sacred bundles.? (Innisskimm, Line 14) Regarding the third diagnostic, both affixes and enclitics may or may not be lexically specified with a pitch accent. However, in cases where they do have an inherent pitch accent, affixes and enclitics differ with respect to the realization of the pitch accent in the verbal complex. With affixes, pitch accent is transferable: if an affix with an inherent pitch accent is affixed to a word, the main pitch accent of the word transfers to the affix. With enclitics, on the other hand, pitch accent is additive: if an enclitic with inherent pitch accent is cliticized to a word, the main pitch accent of the word is maintained, and the enclitic adds a second pitch accent. This is illustrated below. (64) a. ,iksk??siwa. ii-okska?si-wa IC-run.AI-PROX ?S/he ran.? b. ??kokska?siwa. yaak-okska?si-wa FUT-run.AI-PROX ?S/he will run.? c. ?ykska?siwa. a-okska?si-wa IMPF-run.AI-PROX ?S/he is running.? 26 See Fox and Frantz (1979), and Frantz (2009) for discussion. According to these sources, a maximum of two enclitics can attach to the verb, and the leftmost enclitic must index either the subject or the object. I have yet to replicate these findings in my own fieldwork. 56 (65) a. Iiksk??siiyin?yi. ii-okska?si-yini-ayi IC-run.AI-OBV-3PRN ?S/he (OBV) ran.? b. ??kokska?siyin?yi. yaak-okska?si-yini-ayi FUT-run.AI-OBV-3PRN ?S/he (OBV) will run.? c. ?ykska?siyin?yi. a-okska?si-yini-ayi IMPF-run.AI-OBV-3PRN ?S/he (OBV) is running.? In (64a), the pitch accent falls on the second syllable, and in (64b) and (64c), we see that if the future prefix y??k- or the imperfective prefix ?- is affixed to the verb, the pitch accent transfers to the prefix. The same pattern is observed in (65). However, in these examples, an enclitic ??yi also appears on the verb, and it contributes a second pitch accent. In short, whereas pitch accent is transferable with affixes, it is additive with enclitics. This suggests that affixes are prosodified as part of the verbal complex, but enclitics constitute their own prosodic domain. The distinction between agreement affixes and enclitics is revisited in Chapter 7. 2.5. The Phonetic Realization of Proximate and Obviative Suffixes In this section, I discuss the phonetic realization of the proximate and obviative suffixes that appear on both nouns and verbs. These were discussed in ?2.3 and ?2.4.7 above; additional examples are given below. (66) a. ?yo?kaawa oma imit??w. a-yo?kaa-wa om-wa imitaa-wa IMPF-sleep.AI-PROX DEM-PROX dog-PROX ?That dog (prox) is sleeping.? b. ?yo?kaayini omi imit??yi. a-yo?kaa-yini om-yi imitaa-yi IMPF-sleep-OBV DEM-OBV dog-OBV ?That dog (obv) is sleeping.? 57 The proximate (-wa) and obviative (-yi, -yini) suffixes play an important role in the analysis developed in this dissertation. The reason why a discussion of their phonetic realization is relevant is that, as will become apparent, there is variation across speakers in the pronunciation of these suffixes, and not all speakers have an audible suffix in their grammar. Nevertheless, I argue that, despite a cline in audibility across speakers, the proximate and obviative suffixes are indeed active in the grammar. In ?2.5.1, I summarize the findings of Gick et al. (2012), who demonstrate that, for some speakers, proximate and obviative suffixes are soundless (i.e., articulated but not acoustically realized). In ?2.5.2, I summarize the findings of Bliss and Glougie (2010), who demonstrate that, for some other speakers, the suffixes are not phonetically realized but are nevertheless phonologically active. In ?2.5.3, I pull these findings together and propose that, across speakers, there is a cline of audibility in the production of proximate and obviative suffixes. 2.5.1. Voiceless and Soundless Suffixes Frantz (2009) claims that vowels are voiceless word-finally (p. 5), and that the number suffix ?wa is ?rarely audible? (p. 8). *ick et al. (2012) show that, at least for some speakers, word-final vowels are not simply voiceless but SOUNDLESS. This finding is based on two experiments, a production experiment and a perception experiment. For the production experiment, we collected a combination of ultrasound, video, and acoustic recordings of a single speaker producing minimal pairs that differ only in the final vowel of the number suffix. The target forms are given in Table 2.11 (soundless vowels are represented with underlining). Table 2.11. Stimuli for Production Experiment (adapted from Gick et al. 2012, p. 52) Proximate ( -a) forms Obviative ( -i) forms English translation si?k??na si?k??ni ?blanket? kis?mma kis?mmi ?moon? Each form was produced in a carrier phrase ten times, yielding forty tokens in total. The carrier phrases for the proximate and obviative conditions are given in (67a) and (67b) respectively. 58 (67) a. Nits?kssta n??hksinowahsi oma __. nit-ik-sstaa n-aahk-in-o-a-hs-yi om-wa __ 1-INTNS-want.AI 1-MOD-see-TA-DIR-CONJ-OBV DEM-PROX __ ?I want to see that BBB.? b. Nits?kssta m??hksinowahsi omi __. nit-ik-sstaa m-aahk-in-o-a-hs-yi om-yi __ 1-INTNS-want.AI 3-MOD-see-TA-DIR-CONJ-OBV DEM-OBV __ ?I want him to see that BB.? Gick et al. found that the forms did not show statistically significant differences in acoustic measures; there was no audible (or visible from the spectrogram) vowel following the nasal, and the vowel preceding the nasal showed no differences in F1, F2, or F3 values. However, the forms did show significant differences in articulatory measures; at the temporal midpoint of the production of the final (soundless) vowel, the ?a forms showed statistically significant differences from the?i forms in both lip aperture and tongue height. In essence, the final vowels (which instantiate the proximate and obviative suffixes) were shown to be SOUNDLESS: articulated but not acoustically realized. The perception experiment by Gick et al. strengthened the conclusion that these vowels are indeed soundless. Given that we had only negative acoustic evidence in the production experiment, we wanted to address the possibility that there could be audible distinctions between the word-final vowels that escape our acoustic measures. The results of the perception experiment indicate that this is indeed not the case. Using a similar set of stimuli as in the production experiment27, we had a second Blackfoot speaker participate in a forced-choice task in which she identified whether the forms she heard belonged to the proximate (-a) or obviative (-i) context. In the control condition, the forms were presented with the demonstrative determiner oma or omi and in the experimental condition, the forms were presented in isolation without a disambiguating determiner. We found that, in the control condition, the listener could accurately identify whether the form was proximate or obviative, but in the experimental condition, she 27 The same speaker as in the production experiment recorded the forms for the perception experiment, in the same carrier phrases as in (66). However, to lessen the chance that coarticulation with the final vowel of the demonstrative determiner, the adjectival prefix omahk- ?big? was added to the noun. 59 could not. These results confirm that, for at least some Blackfoot speakers, word-final vowels are soundless; they are not acoustically distinct. 2.5.2. Ghost Suffixes Bliss and Glougie (2010) report on the productions of another speaker, who has no phonetic realization of the proximate and obviative suffixes in word final position, but the suffixes behave phonologically as though they are present. I refer to these as ?ghost suffixes,? by analogy with ?ghost segments? discussed in the phonological literature (e.g., Szpyra 1992; Zoll 1996). To confirm that the proximate and obviative suffixes are not articulatorily realized, we ran a pilot experiment in which we attempted to replicate the lip aperture results from the study by Gick et al. We found no significant difference in the ?a versus ?i tokens, and for both ?a and ?i tokens, there was considerable variability; sometimes there was complete closure of the lips and sometimes the lips were spread. We take this as preliminary evidence that, for this particular speaker, there is no articulatory realization of the word final vowels. Despite not being acoustically or articulatorily realized, the proximate and obviative suffixes in word-final position are phonologically active. It is in this sense that they are ?ghosts;? although not present phonetically the suffixes can either block or trigger phonological processes. For example, as shown in the paradigm in (68), the regular process of word-final devoicing affects the stem-final vowel of 1st and 2nd person inflected verbs, but not 3rd person forms. We take this as evidence that there is an unrealized ?wa suffix, and that this suffix blocks word-final devoicing. (68) a. [nite t?u:xwtsim ] b. [kite t?u:xwtsim ] c. [e t?u:xwtsimi] nit-aiitoohtsimi kit-aiitoohtsimi aiitoohtsimi-wa 1-understand.BF.AI 2-understand.BF.AI understand.BF.AI-PROX ?I understand Blackfoot? ?<ou understand Blackfoot? ?S/he understands Blackfoot? 60 Other phonological processes that target the right boundary of the word are also blocked in 3rd person forms. For example, verb stems ending in an underlying /?m:/ elide the ?mm word-finally.28 However, as shown in (69), it is retained in 3rd person forms: (69) a. [nite k?o?po?] b. [kite k?o?po?] c. [e k?o?pomm] nit-a-iko?pomm kit-a-iko?pomm a-iko?pomm-wa 1-IMPF-be.afraid.AI 2-IMPF-be.afraid.AI IMPF-be.afraid.AI-PROX ?I am afraid? ?<ou are afraid? ?S/he is afraid? Not only are the number suffixes phonologically active even when not phonetically realized, they are distinguished from one another. The contrast between proximate ?(w)a and obviative ?(y)i can be seen with possessed nouns, which are obligatorily obviative if the possessor is 3rd person. According to a regular phonological rule in the grammar, sequences of /t+i/ surface as [tsi]. Nouns stems with a final /-t/ surface with final [-ts] in obviative but not proximate contexts. This is shown in (70). (70) a. [niksis:t] b. [niksis:t] c. [oksis:ts] n-iksisst-wa k-iksisst-wa w-iksisst-yi 1-mother-PROX 2-mother-PROX 3-mother-OBV ?I am afraid? ?<ou are afraid? ?S/he is afraid? The data in (70) shows that, although not phonetically realized as ?yi, the obviative suffix is nevertheless phonologically active, and triggers assibilation of the final ?t. The final thing to note is that, in certain environments, the proximate and obviative suffixes are fully realized. Due to a minimal word requirement in Blackfoot (Kaneko 1999; Derrick 2007), there is no word-final voicing with demonstratives. Further, when followed by enclitics, proximate?wa and obviative?yini are fully voiced. Examples are given in (71) below. (71) a. Iisstsim??hkatsiiw?yi. ii-sstsimaahk-at-yii-wa-ayi IC-hire-TA-3:4-PROX-3SG.PRN ?He hired her.? 28 Frantz and Russell (1995) use the notation ?3mm? to identify these forms. 61 b. 2tsi?naks?pokaayi ?waasai?niin?yi. ot-i?nak-pokaa-yi a-waasai?ni-yini-ayi 3-small-child-OBV IMPF-cry.AI-OBV-3SG.PRN ?Her baby is crying.? The data in (71) show that, although the proximate and obviative suffixes are sometimes ?ghosts? in this particular speaker?s grammar, they are fully realized in other contexts. This further supports the claim that they play an active role in the grammar. 2.5.3. Cline of Audibility In sum, although productions of the proximate and obviative suffixes vary across contexts and across speakers, even in their ?weakest? phonetic realization (i.e., ?ghosts?), the suffixes are active in the grammar. Figure 2.2. below summarizes the variable realizations of the proximate and obviative suffixes, ranging from fully articulated and voiced in non-word-final contexts, to voiceless (as reported by Frantz 2009), to soundless (as reported by Gick et al. 2012) to unrealized ghosts (as reported by Bliss and Glougie 2010). Figure 2.2. Variable Realizations of the Proximate ?wa Suffix /-wa/ [-(w)a] [-(w)?] [-(w)a] [-(w)_] voiced voiceless soundless unrealized ghosts 2.6. Grammatical Functions I assume that the structural correlates of grammatical functions such as subject and object can vary cross-linguistically. As such, the phrase structure positions corresponding to the various grammatical functions Cline of Audibility 62 must be defined on a language by language basis. In this section, I map Blackfoot?s grammatical functions onto structural positions. The section proceeds as follows. In ?2.6.1, I identify vP-internal positions for the subject and object. In ?2.6.2, I discuss a grammatical function that I refer to as the POINT-OF-VIEW (POV) HOLDER. This grammatical function is associated with a vP-external argument position, Spec, AspP and is determined by the direct/inverse system. In ?2.6.3, I discuss the Spec, IP and Spec, CP positions and show that neither of these are correlated with the grammatical functions of subject or object. As such, subject and object can be defined as vP-internal in Blackfoot. 2.6.1. Subject and Object As noted in ?2.4.3, Blackfoot verb stems are morphologically complex, consisting minimally of a root plus a final. The four classes of finals were given in Table 2.3 above, repeated again below. Table 2.3. Verb Finals II Inanimate Intransitive Subject = Inanimate AI Animate Intransitive Subject = Animate TI Transitive Inanimate Object = Inanimate TA Transitive Animate Object = Animate As shown in Table 2.3, AI and II finals vary according to the animacy of the subject, and TA and TI finals vary according to the animacy of the object. Examples are given below. (72) a. Anna Leo iks?sto?siwa. ann-wa L iksisto-?s -wa DEM-PROX L be.hot.AI-PROX ?/eo has a fever? (lit: ?/eo is hot.?) b. Omi pak?yittsii iks?stoyiwa. om-yi pakoyittsi-yi iksisto-yi-wa DEM-INAN fire-INAN be.hot-II-PROX ?That fire is hot.? (73) a. Iihp?mmatsiiwa amoiksi si?k??niks. ii-ohpomm-at-yii-wa amo-iksi si?kaan-iksi PST-buy-TA-3:4-PROX DEM-ANIM.PL blanket-ANIM.PL ?She bought these blankets.? 63 b Iihp?mmatooma amostsi ks??stsimaanistsi ii-ohpomm-atoo-m-wa amo-istsi ksiistsimann-istsi PST-buy-TI-3:INAN-PROX DEM-INAN.PL bead-INAN.PL ?She bought these beads.? It has been widely argued that Algonquian finals are light verbs that associate with the functional head v (cf. Bruening and Rackowski 2000 for Passamaquoddy; Hirose 2003 for Plains Cree; Brittain 2003 for Western Naskapi; Quinn 2006 for Penobscot; Mathieu 2006 for Ojibwe; Ritter and Rosen 2010a for Blackfoot). As is characteristic of the category v across languages (e.g., Chomsky 1995), the finals can be thought to introduce an external argument and license an internal argument. There are (at least) two ways we can conceive of this. First, if we assume a model of argument structure in which structural positions correspond with theta roles (e.g., UTAH, Baker 1988), then this suggests that the agent is mapped onto the external argument position and the patient is mapped onto the internal argument position, as follows: (74) a. vP 3 Agent 3 v VP Final 3 V Patient Root In (74), the final introduces the agent (canonically the subject) and licenses the patient (canonically the object). The problem with this model is that it doesn?t capture the generalization that the four classes of finals determine the animacy of the absolutive argument (intransitive subject / transitive object). Under the model in (74), this would mean that, in the case of intransitive verbs, the final agrees in animacy with the external argument, but in the case of transitive verbs, the final agrees in animacy with the internal argument. It cannot simply be the case that the verb only agrees with the external argument in the case when there is no internal argument. The reason why not is that the distinction between morphologically transitive and intransitive stems is not straightforwardly a matter of transitivity. Both intransitive and 64 transitive verb stems can license an internal argument (an object), but with different syntactic properties. The object of a morphologically intransitive verb can be either animate or inanimate and can be a bare noun, but it must be immediately postverbal. Conversely, the object of a morphologically transitive verb must agree in animacy with the final, it cannot be a bare noun, and it doesn?t show strict ordering restrictions. This is summarized in Table 2.12, and examples illustrating the differences between morphologically intransitive and transitive verbs are given below. Table 2.12. Objects of Morphologically Intransitive and Transitive Verbs Morphologically Intransitive Morphologically Transitive Bare Noun ? ? Must be Immediately Postverbal ? ? (75) a. N?tsooyi ?passtaamiinaam. nit-ioyi apasstaamiinaam 1-eat.AI apple ?I ate an apple.? b. * ?passtaamiinaam n?tsooyi. apasstaamiinaam nit-ioyi apple 1-eat.AI intended: ?I ate an apple.? (76) a. Nits??watawa oma ?passtaamiinaama. nit-iowat-a-wa om-wa apasstaaminaam-wa 1-eat.TA-DIR-PROX DEM-PROX apple-PROX ?I ate that apple.? b. * Nits??watawa ?passtaamiinaam. nit-iowat-a-wa apasstaaminaam 1-eat.TA-DIR-PROX apple intended: ?I ate an apple.? c. Oma ?passtaamiinaama nits??watawa. om-wa apasstaaminaam-wa nit-iowat-a-wa DEM-PROX apple-PROX 1-eat.TA-DIR-PROX ?I ate that apple.? (77) a. N?tsooyi sit?kihkiitaan. nit-ioyi sitokihkiitaan 1-eat.AI pie ?I ate pie.? 65 b. * Sit?kihkiitaan n?tsooyi. sitokihkitaan nit-ioyi pie 1-eat.AI intended: ?I ate pie.? (78) a. Nits??watoohpa omi sit?kihkiitaan. nit-iowatoo-hp-wa om-yi sitokihkitaan-yi 1-eat.TI-1:INAN-PROX DEM-INAN pie-INAN ?I ate that pie.? b. * Nits??watoohpa sit?kihkiitaan. nit-iowatoo-hp-wa sitokihkitaan-yi 1-eat.TI-1:INAN-PROX pie-INAN intended: ?I ate pie.? c. Omi sit?kihkiitaan nits??watoohpa . om-yi sitokihkitaan-yi nit-iowatoo-hp-wa DEM-INAN pie-INAN 1-eat.TI-1:INAN-PROX ?I ate that pie.? In (75), the intransitive verb takes an animate object, ?passtaamiinaam ?apple,? which is a bare noun and cannot appear preverbally. In (76), the transitive verb takes the same animate noun as an object, but it cannot be bare and can be preverbal. This same pattern is seen with inanimate objects in (77) and (78); both the intransitive and transitive verbs can take an inanimate object, but with different properties. I propose an alternative to the structure in (74). In particular, I propose that finals introduce external arguments via an Agree relation. They bear an uninterpretable animacy feature ([uANIM(ate)] or [uINAN(imate)]) that is checked by a matching feature in Spec, vP. First consider the structure of morphologically intransitive verbs, i.e., those with an AI or II final. (79) a. vP b. vP 3 3 Subject 3 Subject 3 [ANIM] v VP [INAN] v VP AI 3 II 3 [uANIM] V (Object) [uINAN] V (Object) Root Root In (79a), the AI final has an uninterpretable [ANIM] feature that requires the external argument to be animate, and in (79b), the II final has an uninterpretable [INAN] feature that requires the external 66 argument to be inanimate. Although morphologically intransitive, these verbs can optionally take an internal argument as complement to V, akin to an incorporated noun.29 I refer to the objects of morphologically intransitive verbs as UNINDEXED OBJECTS to reflect the fact that they are not morphologically indexed on the verb. Now consider morphologically transitive verbs, I propose that they have a more complex structure than morphologically intransitive ones; they have a recursive vP structure, with the TA/TI finals appearing higher in the structure than the AI/II finals, and attracting the internal argument from VP to the Specifier of the higher vP. As such, the object position for morphologically transitive verbs is a derived position, as shown below.30 (80) a. vP b. vP 3 3 Object 3 Object 3 [ANIM] v vP [INAN] v vP TA 3 TI 3 [uANIM] Subject 3 [uINAN] Subject 3 v VP v VP 3 3 V <Object> V <Object> In (80a), the TA final combines with the vP and attracts an animate object to its Specifier via an Agree relation. Similarly in (80b), the TI final combines with the vP and attracts an inanimate object to its Specifier. Evidence in support of the claim that TA/TI finals are higher than AI/II finals comes from nominalization. Bliss et al. (2012) describe and categorize the various nominalization patterns in Blackfoot, and they observe that one particular nominalization pattern (referred to by Frantz (2009: 115) 29 In Chapter 3, I argue that unindexed objects of morphologically intransitive verbs are pseudo-incorporated. 30 Under this analysis, there is a derived position for objects but not for subjects. This is the opposite of, e.g., English, which has a derived subject position (Spec, IP) but is not typically thought to have a derived object position (English objects are introduced in and remain in Comp, VP, cf. Adger 2003). 67 as abstract nominalization) is formed by adding a nominalizing suffix ?(hsi)n31 to morphologically intransitive stems. While this pattern of nominalization requires an AI or II final, it cannot contain a TA or TI final. Examples are given below. (81) a. ??waahsini ik??kitapsoka?piiwa kiist?. a-owaa-hsin-yi ik-yaak-itap-sok-a?pii-wa kiisto IMPF-eat.AI-NOM-INAN INTNS-FUT-good-be-PROX 2SG.PRN ?(ating will be good for you.? b. *??waatoo(?p)ssini (nik??pis) ik??kitapsoka?piiwa kiist?. a-owa-atoo-(?p-)hsin-yi nit-koopis ik-yaak-itap-sok-a?pii-wa kiisto IMPF-eat-TI-(2 -INAN-) NOM-INAN 1-soup INTNS-FUT-good-be.II-PROX 2SG.PRN intended: ?(ating (my soup) will be good for you.? (82) a. Iksk??maani ??kohkotsiksst?natapiiwa. Ikskim-aa-n-yi yaak-ohkot-ik-sstonnat-a?pii-wa hunt-AI-NOM-INAN FUT-ABL-INTNS-dangerous-be.II-PROX ?Hunting can be dangerous.? b. * Iksk??maat(s)sini (annahkayi ponoka) ??kohkotsiksst?natapiiwa. ikskiim-aat-(yii -) hsin-yi annahkayi ponoka yaak-ohkot-ik-sstonnat-a?pii-wa hunt-TA-(3:4 -) NOM-INAN DEM elk FUT-ABLE-INTNS-dangerous-be.II -PROX intended: ?Hunting that elk can be dangerous.? Under the assumption that nominalizations target different levels of the extended verbal projection (cf. Abney 1987; Borsley and Kornfilt 2000; Schueler 2006; Kornfilt and Whitman 2011), the data in (81) and (82) can be accounted for under the analysis that TA/TI finals are higher in the structure than AI/II ones. In particular, if TA/TI finals are higher than AI/II finals, then they are outside the structure that is the target of nominalization (i.e., the vP headed by an AI/II final to which ?(hsi)n attaches).32 Moreover, the claim that the object position for morphologically transitive verbs is a derived position is consistent with the observation that the object of a morphologically transitive verb can bear a 31 The suffix ?(hsi)n has two phonologically-conditioned allomorphs: It surfaces as ?n following a-final verb stems, and ?hsin elsewhere. 32 Aside from the nominalization facts, another prediction of this analysis is that, if AI/II finals map onto the lower v head and TA/TI finals map onto the higher v head, the two should be able to co-occur. This is possible with certain derived stems (e.g., causative and accompaniment verbs, cf. Frantz 2009, chapter 18) but generally speaking, there is no morphological realization of the lower v head in the structures in (80). I speculate that this could be accounted for with appeal to post-syntactic spell-out restrictions, as discussed in Chapter 1. 68 variety of different thematic roles. In other words, it is not tied to a position that is projected as part of the verb?s argument structure. Rather, the object can bear a thematic role typically attributed to a direct object, such as patient (83) or theme (84), or it can bear a thematic role typically attributed to an indirect object, such as beneficiary (85) or source (86). In the examples below, the object is in boldface. (83) N?ksi n?naiks ??sskonakatsiiyaa ni ?aattsistaayi. ann-iksi ninaa-iksi a-isskonakat-yii-yi-aawa ann-yi aaattsistaa-yi DEM-PL man-PL IMPF-shoot.at.TA-3:4-3PL-3PL.PRN DEM-OBV rabbit-OBV ?The men shot at the rabbit.? (84) Niht??hkanayi amiksi si?k??niksi anni ?kssin. n-iht-aahkani-a-yi am-iksi s ?kaan-iksi ann-yi akssin-yi 1-PURP-sew.TA-DIR-PL DEM-PL blanket-PL DEM-INAN bed-INAN ?I sewed those blankets for the bed.? (85) Anna Rosie ??mmskatoomoyiiwa annisk ??mi. ann-wa R ii-mmsk-atoo-omo-yii-wa ann-yi-hk w-om-yi DEM-PROX R. IC-save.food-TI-TA.BEN-3:4-PROX DEM-OBV-NV 3-husband-OBV ?Rosie saved food for her husband.? (86) Anna Rosie nito?t?moka nitsinik??siPiks. ann-wa R nit-o?t-omo-ok-wa nit-inika?simiks-yi DEM-PROX R 1-take-TA.BEN-INV-PROX 1-car-OBV ?Rosie took my car from me.? The examples in (83)-(86) show that the object of a morphologically transitive verb can bear a variety of different thematic roles. This is consistent with the claim that the object position is a derived position; it is not projected as part of the argument structure inherent to the verb?s lexical entry. To summarize, I have adopted the assumption that verb finals associate with the category v, and they agree with an external argument; AI/II finals combine with VP and agree with the subject, and TA/TI finals combine with vP and agree with the object. As such, in Blackfoot, the grammatical functions of subject and object both map onto external argument positions inside the vP. Additionally, unindexed objects may appear in Comp, VP. This is schematized for morphologically intransitive and transitive verbs below. 69 (87) a. Intransitive b. Transitive vP 3 Object 3 vP v vP 3 TA/TI Final 3 Subject 3 Subject 3 v VP v VP AI/II Final 3 3 V (Unindexed Object) V <Object> Root Root 2.6.2. PoV Holder In this section, I argue that Asp in Blackfoot is associated with the direct/inverse suffixes, and introduces an external argument with a grammatical function I refer to as the Point-of-View (PoV) Holder. I present an analysis of the Blackfoot?s direct/inverse, adapted from Bliss (2005a) and Bliss et al. (2010a, b). Descriptively, direct/inverse marking is sensitive to the person hierarchy below. (88) Direct/Inverse Hierarchy 1st > 2nd > 3rd PROXIMATE > 3rd OBVIATIVE When a direct suffix appears on the verb, the higher-ranking participant is the subject, and when an inverse suffix appears on the verb, the higher-ranking participant is the object. Examples illustrating this are given below. (89) a. Nitsik?motsiipiawa. nit-ikamotsiip-i-a-wa 1-rescue-TA-DIR-PROX ?I rescued him/her.? b. Nitsik?motsiipioka. nit-ikamotsiip-i-ok-wa 1-rescue-TA-INV-PROX ?S/he rescued me.? 70 In (89a), a higher-ranking 1st person is the subject, a lower-ranking 3rd person is the object, and a direct suffix ?a appears on the verb. In (89b), the higher-ranking 1st person is the object, and the lower-ranking 3rd person is the subject, and an inverse suffix ?ok appears on the verb. Whether the direct/inverse in Algonquian is strictly a morphological phenomenon or involves syntactic inversion, akin to the passive in languages like English, has been a source of debate (e.g., Aissen 1997, McGinnis 1999 argue against a syntactic inversion account). Bruening (2001, 2009) makes a compelling case for a syntactic (A-movement) account of Passamaquoddy?s direct/inverse system, using quantifier scope and variable binding data to support his claim. Building on Bruening?s analysis, Bliss (2005a) develops a syntactic account of Blackfoot?s direct/inverse system, which has been subsequently refined by Bliss et al. (2010a, b). Here I present an amalgamation of these analyses. One of the main insights of Bliss (2005a) is that direct/inverse marking grammaticizes the relation between grammatical functions and point-of-view, by encoding whether the PoV holder is the subject or the object. For this reason, Bliss (2005a) argues the direct and inverse markers associate with a functional head Point-of-View (PoV). Bliss et al. (2010a, b) claim that the PoV head is located between IP and vP, and fulfills the same functional role as Viewpoint (i.e., Outer) Aspect in a Tense-based language like English. Following DeLancey (1981), we assume that Aspect is the temporal analog of Point-of-View; just as Aspect provides a temporal perspective on events (e.g., Smith 1997), Point-of-View provides a participant?s perspective. Put another way, Aspect locates event times relative to a reference time (or point-of-view time), and Point-of-View locates event participants relative to a point-of-view holder. In what follows, I refer to both Outer Aspect and Point-of-View by the same category label: Asp. The Asp head has different morphological realizations: the different direct and inverse suffixes.33 Bliss (2005a) employs a feature geometric approach to ? features (cf. Harley and Ritter 2002), analysing the various direct/inverse markers as being specified for different person features, but all sharing the 33 Bliss et al. (2010a, b) propose that the direct and inverse morphemes map onto three different structural positions: Asp plus two agreement positions. I abstract away from these details here. 71 feature [Sentient]. A desirable consequence of this analysis is that the ranking of arguments in the direct/inverse is derived from the featural content of the direct and inverse markers and the hierarchy itself is rendered epiphenomenal. Abstracting away from the details, I refer to the feature [uSent(ient]) as the primary ? feature on the Asp head. 34 When Asp is direct, the subject moves to Spec, AspP, and when Asp is inverse, the object moves to Spec, AspP. Both of these are shown below. (90) a Direct b. Inverse AspP AspP 3 3 Subject[Sent] Asp' Object[Sent] Asp' 3 3 Asp vP Asp vP DIR[uSent] 3 INV[uSent] 3 Object v' <Object> v' 3 3 v vP v vP 3 3 <Subject> 3 Subject 3 v VP v VP 4 4 In sum, Spec, AspP in Blackfoot can be identified with a grammatical function, the PoV holder. This is a derived position, occupied by the subject when Asp is direct, and the object when Asp is inverse. 2.6.3. Spec, IP and Spec, CP The two remaining positions ? Spec, IP and Spec, CP ? are discussed in detail in Chapters 5 and 6, respectively. Here, I present a brief summary of the properties of these two positions. As outlined in Chapter 1, I assume the IP domain is the anchoring domain; it connects the clause to the utterance situation. Ritter and Wiltschko (to appear) argue that, cross-linguistically, INFL can vary in its substantive content; INFL can be tense-based, relating event times to utterance times, location-based, relating event locations to utterance locations, or person-based, relating event participants to 34 The other features of the different direct and inverse markers ensure that the ?right? argument moves up to Spec, AspP. For instance, in (90a) the object fails to intervene because it does not meet the featural requirements of Asp. See Bliss (2005a) for details. 72 utterance participants. Ritter and Wiltschko argue that Blackfoot exemplifies the latter case: INFL in Blackfoot is person-based, and the person prefixes that appear at the left edge of the verbal complex appear in Spec, IP. Examples of the person prefixes are given below.35 (91) a. Nits??tsitapinihka?siPi Otsskapin??ki. nit-niit-itapi-inihka?simi otssk-apini-aakii 1-genuine-person-be.named.AI blue-eye-woman ?My Indian name is Otskapinaaki.? (lit. ?blue-eyed woman?) b. Kitsiitsitapinihka?siPi Otskapinaaki. kit-niit-itapi-inihka?simi otssk-apini-aakii 2-genuine-person-be.named.AI blue-eye-woman ?<our Indian name is Otskapinaaki.? (lit. ?blue-eyed woman?) In Chapter 5, I develop an analysis of the Spec, IP position. I argue that INFL bears uninterpretable [Pers(on)] features, and that the person prefixes in Spec, IP check these uninterpretable features. Furthermore, I argue that Spec, IP can be occupied not only by person prefixes but also by an evidential prefix that signals speaker certainty of a past time event. Regardless of which prefix occupies Spec, IP, it is subject to the featural requirements of INFL, as schematized below.36 (92) IP 3 XP[Pers] 3 I AspP [uPers] 6 Regarding the Spec, CP position, in Chapter 6, I demonstrate that the obviative (-yini) and plural (-yi) suffixes that appear at the right edge of the verbal complex associated with C.37 They bear 35 In addition to 1st person nit- and 2nd person kit-, there is a 3rd person prefix ot-, which is restricted to matrix TA inverse clauses, and conjunct clauses. 36 Moreover, unlike, e.g., English, Spec, IP need not be occupied in Blackfoot. As such, the status of the EPP (cf. Chomsky 1981) in this language is unclear. 37 The suffixes ?yini and ?yi alternate with proximate ?wa, which I argue in Chapter 6 occupies the same position as ?yini and ?yi but does not have uninterpretable ? features that need to be checked in an Agree relation. 73 uninterpretable Number features which are checked by a third person argument (KP) in Spec, CP. Examples of each, with their respective structures, are given below. (93) a. ?ykska?siyini [KP anni ossk?ni]. b. ?ykska?siyi [KP anniks ossk?niks]. a-okska?si-yini ann-yi w-isskan-yi a-okska?si-yi ann-iksi w-isskan-iksi IMPF-run.AI-OBV DEM-OBV 3-sister-OBV IMPF-run.AI-PL DEM-PL 3-sister-PL ?His sister is running.? ?His sisters are running.? CP CP 3 3 KP[SG] 3 KP[PL] 3 C IP C IP -yini[uSG] 6 -yi[uPL] 6 This proposal suggests that Spec, CP is an A-position in Blackfoot, and this raises questions about the availability of the Spec, CP position for A' -extraction. In fact, the relative paucity of A' -extraction phenomena in Blackfoot support this proposal. In particular, wh-questions in Blackfoot are formed of nominalizations (cf. Frantz 2009), and it is not clear that they involve wh-movement (see chapter 8, ?220.127.116.11). Similarly, Blackfoot does not have relative clauses, instead employing a nominalization strategy to fulfill the function of relative clauses (see chapter 7, ?18.104.22.168, also see Bliss, to appear). The one possible case of A' -extraction involves fronting of a subordinate constituent to the matrix clause, as shown in (94b) below. (94) a. Nits?kssta om??hksaowaat?hksaa omiksi imit???ks. nit-iksstaa om-aahk-saw-at-ohki-saa om-iksi imitaa-iksi 1-want.AI 3-MOD-NEG-again-bark.AI-NONAFF DEM-PL dog-PL ?I want those dogs to stop barking.? b. Omiksi imit???ks nits?kssta om??hksaowaat?hksaa. Whether fronting of this sort can be characterized as topicalization through an A' -position is yet unclear. In short, although a detailed investigation of Blackfoot?s A-extraction phenomena is pending, the data available suggest that Spec, CP may not be utilized for A' -extraction. This is predicted under the account that Spec, CP is an A-position. 74 2.6.4. Summary In summary, I have argued that the grammatical functions of subject and object map onto two different Spec, vP positions in Blackfoot, with the object appearing in a derived position above the subject. Moreover, I have proposed that an additional grammatical function ? PoV holder ? is operative in Blackfoot and maps onto the Spec, AspP position. As for the higher Specifier positions in the clausal spine, Spec, IP and Spec, CP, I have argued that these are associated with person and number agreement. The Specifier positions and their associated grammatical functions are given in Table 2.13 below. Table 2.13. Structural Correlates of Grammatical Functions Positions Grammatical Functions Spec, vP (intransitive) Subject Spec, vP (transitive) Object Spec, AspP PoV holder Spec, IP -- (locus of person prefixes) Spec, CP -- (locus of number agreement) 2.7. Conclusion This chapter has provided information on Blackfoot that is relevant for this dissertation. I have discussed morphosyntactic properties of nominal expressions, and given an overview of the morphemes that comprise the verbal complex. Additionally, I have provided evidence that, despite a cline in audibility, the proximate and obviative suffixes that appear on nouns and verbs are indeed active in the grammar. Finally, I have mapped grammatical functions onto structural positions, arguing that the subject and object are both external arguments in vP, and there is an additional grammatical function, PoV holder, that is associated with Spec, AspP and is determined by the direct/inverse. 75 CHAPTER 3 MAPPING NOMINAL EXPRESSIONS ONTO THE SYNTACTIC SPINE 3.1. Introduction Having established some background on Blackfoot morphosyntax, I now turn to the task of developing an analysis of Blackfoot?s argument-typing system. The overarching goal of this dissertation is to reveal the hierarchical relations in Blackfoot that are obscured by the non-configurational properties of the language, and the main thesis is that the ways in which the various linguistic objects associated with arguments map onto the syntactic spine conspire to give Blackfoot a non-configurational profile.I begin the analysis of Blackfoot?s argument-typing system with a discussion of the internal and external syntax of its nominal expressions. I look at four types of nominal expressions: proximate singular, obviative singular, plural, and uninflected (bare) nouns. Examples are given below. (1) a. Nitsik??stsimmaa oma aak??wa. nit-ik-a-istsimm-a-wa om-wa aakii-wa 1-INTNS-IMPF-respect-DIR-PROX DEM-PROX woman-PROX ?I respect that woman.? b. Anna Anna iik??stsimmiiwa omi aak??yi. ann-wa A ii-ik-a-istsimm-yii-wa om-yi aakii-yi DEM-PROX A IC-INTNS-IMPF-respect-3:4-PROX DEM-OBV woman-OBV ?Anna respects that woman.? c. Nitsik??stsimmayi omiksi aak??ks. nit-ik-a-istsimm-a-yi om-iksi aakii-iksi 1-INTNS-IMPF-respect-DIR-PL DEM-PL woman-PL ?I respect those women.? (2) a. Anna Beth ??yiwa immists??hkitaan. ann-wa B a-ooyi-wa immistsiihkitaan ann-wa B IMPF-eat.AI-PROX frybread ?Beth is eating frybread.? b. Anna Beth ??yiwa immists??hkitaanists. ann-wa B a-ooyi-wa immistsiihkitaan-istsi DEM-PROX B IMPF-eat.AI-PROX frybread-PL ?Beth is eating (pieces of) frybread.? 76 In (1), there are examples of proximate singular, obviative singular, and plural nominal expressions functioning as the indexed object of a morphologically transitive verb. In (2), there are examples of a bare noun and a plural noun functioning as the unindexed object of a morphologically intransitive verb. (For a survey of the types of nominal expressions that can be used with each grammatical function, see Appendix A.) I argue that the nominal expressions in (1) have more functional structure than the nominal expressions in (2). In particular, proximate singular nominal expressions like those in (1a) have a functional head I refer to as LINK; it is the highest functional head in the spine, but is neutral with respect to whether it appears in a nominal or verbal spine. The obviative singular and plural nominal expressions in (1b) and (1c) have the same amount of structure as the proximate one in (1a), but are categorized as KPs rather than LINKPS. This difference corresponds with a difference in their external syntax; LINKPS are adjoined to the clause but KPs appear in argument positions. As for the nominal expressions in (2), I propose that bare nouns are nPs and bare plurals are ?Ps; both are pseudo-incorporated, in the sense of Massam (2001). Under this analysis, plural nouns are structurally ambiguous: they can be KPs or ?Ps. A summary of my analytical claims about the four types is given in Table 3.1. Table 3.1. Four Types of Nominal Expressions Syntactic category Relation to Clause Syntactic Position Proximate Singular (-wa) LINKP Adjunct Adjoined to LINKP Obviative Singular (-yi) KP Argument A-position in the clause Bare (? ) nP Pseudo-incorporated VP complement Plural ( -iksi/ -istsi) ?P Pseudo-incorporated VP complement KP Argument A-position in the clause This chapter proceeds as follows. In ?3.2, I discuss proximate singular nominal expressions and in ?3.3 obviative singular nominal expressions. Bare nouns are discussed in ?3.4, and plural nouns in ?3.5. In ?3.6 I conclude. 3.2. The Syntax of Proximate Singular ( Nominal) Expressions In this section, I discuss the syntax of proximate singular nominal expressions, such as that in (3) below. 77 (3) Oma p??taawa ??paawaniwa. om-wa piitaa-wa a-ipaawani-wa DEM-PROX eagle-PROX IMPF-fly.AI-PROX ?That eagle is flying.? The starting point for my analysis of proximate singular nominal expressions is the observation that the same suffix, -wa, attaches to either noun stems or verb stems. In ?3.2.1, I argue that -wa maps onto the highest head in the spine, but that it is indiscriminate with respect to whether it associates with a nominal or a verbal spine. I adopt the label LINK to refer to this head. In ?3.2.2, I propose that proximate expressions do not appear in argument positions but are adjoined to the clause and bind null pro argument(s), as schematized below. (4) LINKP LINKPi LINKP 5 3 oma p??taawa proi 3 LINK IP -wa 3 I AspP 3 proi 3 Asp vP 3 <proi> 6 ??paawaani 3.2.1. Proximate Expressions are LINKPs In this section, I discuss the syntactic category of the proximate suffix ?wa. In ?22.214.171.124, I propose that -wa associates with a functional head LINK that is neutral with respect to whether it appears in a nominal or verbal spine. I demonstrate that expressions marked with ?wa (whether nominal or verbal) can be interpreted as predicates or arguments, and I take this as evidence for the neutral status of ?wa. In ?126.96.36.199, I discuss an alternative analysis, namely that ?wa associates with the functional head ?. The ?P analysis can capture predicate/argument flexibility (cf. D?chaine and Wiltschko 2002), but I demonstrate 78 that proximate expressions do not have the distribution of other ?Ps (namely plural nouns), and as such the ?P analysis is not tenable. 188.8.131.52. Proximate ?wa Associates with LINK In this section, I argue that ?wa is neutral: the -wa suffix that appears on proximate argument expressions is the same -wa suffix that appears on proximate-marked clauses. Both map onto the highest functional head in the spine, which as discussed in Chapter 1 I take to be the head of the ?linking? domain of the spine, i.e., the layer that is responsible for linking the phrase to a superordinate structure or the larger discourse. In the verbal spine, the linking head is C and in the nominal spine it is K, as shown in (5). (5) a. [CP [IP [AspP [vP [VP]]]]] b. [KP [DP [?P [nP [NP]]]]] I propose that ?wa associates with the head of the linking layer of the spine, but it is not categorized as either C or K. Rather, it is a neutral instantiation of the linking head, not specified as either verbal (C) or nominal (K). I refer to its category label as LINK.38 The LINK head can combine with either a verbal or nominal spine, as schematized in (6). (6) a. [LINKP -wa [IP [AspP [vP [VP]]]]] b. [LINKP -wa [DP [?P [nP [NP]]]]] What evidence is there that the ?wa suffix that appears in the nominal spine maps onto the same syntactifc category as the one that appears in the verbal spine? The first piece of evidence is that the two do not co-occur. Consider the data in (7). 38 Choosing the appropriate label for this head is difficult. With lexical categories, we use the term ROOT (or ?) to refer to lexical items that are neither N nor V but category-neutral (e.g., Marantz 1997; Borer 2004; Arad 2005). However, there is no equivalent terminology to refer to category-neutral functors. The motivation for the term LINK is to capture the observation that ?wa is neutral with respect to the nominal/verbal distinction but is still associated with the linking layer in the spine. 79 (7) a. Amokso?ki nit?mitaamiksi. am-o-iksi-o?k-yi nit-omitaa-m-iksi DEM-RESTR-PL-o?k-PL 1-dog-POSS-PL ?My dogs are these ones.? b. Amoo?ka nit?mitaama. (not *amo -wa-o?k-wa) am-o-o?k-wa nit-omitaa-m-wa DEM-RESTR-o?k-PROX 1-dog-POSS-PROX ?My dog is this one.? In (7), there are two examples of so-called ?verbalized? demonstratives (cf. Uhlenbeck 1938), i.e., demonstratives that have been modified with a suffix ?o?k that allows them to function predicatively. The syntax of demonstratives is discussed in Chapter 4, but for now, the observation to be gleaned from the data in (7) is that a demonstrative marked with ?o?k may take both nominal and verbal morphology. In (7a), the demonstrative is modified by both the nominal plural suffix ?iksi and the verbal plural suffix ?yi (separated by the suffix ?o?k.) However, in (7b) we see that the same is not possible with a proximate singular demonstrative; the ?wa suffix only appears in the position of the verbal number suffix, following ?o?k. Representations of the two demonstratives (without category labels) are given below. (8) a. aPokso?ki b. amoo?ka 3 3 -yi 3 -wa 3 -o?k 3 -o?k 3 -iksi 3 -o am -o am (8a) corresponds to (7a); there are two number suffixes: nominal ?iksi, and verbal ?yi. (8b) corresponds to (7b); here there is only one number suffix: -wa, in the higher position, corresponding to verbal -yi. There is no ?wa suffix that appears in the same position as nominal ?iksi, suggesting that ?wa necessarily maps onto the higher position. Although examples like this are rare (and perhaps impossible except with this small class of verbalized demonstratives), the fact that they exist at all supports the claim that there is only one position for ?wa. Furthermore, the examples in (7) support the view that ?wa is structurally higher than the 80 nominal plural suffixes; -iksi in (7a) attaches closer to the demonstrative root than does ?wa in (7b). I return to this point in ?3.5, when I consider the syntax of plural nouns. With respect to its syntactic position, in Chapter 6 I consider the syntax of ?wa on verbs and I demonstrate that it maps onto the highest functional head in the verbal spine. Part of the evidence I give is negative; I show that it does not occupy any of the lower functional heads in the verbal spine, namely v, Asp, or INFL. The positive evidence I provide is that the distribution of ?wa is sensitive to illocutionary force and clause type, both of which are associated with the linking domain of the clause (i.e., the CP layer). In short, -wa maps onto the highest functional head in the verbal spine. By extension, I propose that ?wa occupies this same position when it associates with a nominal spine: whether it appears with nouns or verbs, it maps onto LINK, the highest functional head in the spine. As for the claim that ?wa is a neutral head LINK and not categorized as C or K, the primary piece of evidence in support of this proposal is the fact that both nouns and verbs can be construed as either predicates or arguments when suffixed with ?wa. Consider the examples below.39 (9) a. (Anna) ??hpiyiwa. (ann-wa) a-ihpiyi-wa (DEM-PROX) IMPF-dance.AI-PROX ?S/he is dancing.? b. Anna ??hpiyiwa ?kaomatap??wa. ann-wa a-ihpiyi-wa akaa-omatap-oo-wa DEM-PROX IMPF-dance.AI-PROX PERF-begin-go.AI-PROX ?The one who dances has just left.? (10) a. Piit??wa. piitaa-wa eagle-PROX ?S/he is an eagle.? b. Oma piit??wa. om-wa piitaa-wa DEM-PROX eagle-PROX ?That is an eagle.? 39 My consultants typically translate proximate-marked verbs as relative clauses when they function as arguments, and I have maintained these translations in the examples presented here. A more literal translation of, e.g., (9b) would be ?The dancing one just left.? 81 c. Oma pitt??wa ??pottawa. om-wa piitaa-wa a-ipottaa-wa DEM-PROX eagle-PROX IMPF-fly.AI-PROX ?That eagle is flying.? In (9), the verb ihpiyi ?dance? is suffixed with ?wa and it can be interpreted as either a predicate (a) or an argument (b). Similarly in (10), the noun piitaa ?eagle? is suffixed with ?wa and it can be interpreted as either a predicate (a/b) or an argument (c). Note that even when the demonstrative is absent, the predicate construals of both the verb and the noun are available. Moreover, note that the predicate construals do not require an overt argument expression to form a complete proposition. Rather, they permit null arguments, which I take to be instantiated by pro. I propose that the predicate/argument flexibility reflects a structural ambiguity observed with LINKPS: LINK may or may not license null pro in its Specifier.40 In the former case, the resulting structure is interpreted as a saturated predicate, i.e., a complete proposition. In the latter case, the resulting structure is interpreted as an argument. 41 This is illustrated for the examples in (9) and (10) above in (11) and (12) below. (11) a. LINKP b. LINKP 3 3 pro 3 LINK IP LINK IP -wa 6 -wa 6 ??hpiyi ??hpiyi ?S/he is dancing? ?the one who dances? 40 The proposal that pro is licensed in Spec, LinkP suggests that Spec, LinkP is an A-position, and not an A' -position. This parallels the proposal introduced in Chapter 2 (and elaborated on in Chapter 6, ?184.108.40.206) that Spec, CP is also an A-position. See Chapter 2, ?2.6.3 for discussion. 41 Although taken here to reflect a structural ambiguity, the question remains whether the predicate/argument flexibility observed with proximate expressions is indeed a case of ambiguity at the level of semantics. In other words, are there indeed two construals (predicate and argument) for proximate expressions, or are they unspecified with respect to the predicate/argument distinction? It is unclear to me whether and how to test this empirically, as well as how to model predicate/argument flexibility if it were found not to be a case of ambiguity. 82 (12) a. LINKP b. LINKP 3 3 pro 3 LINK DP LINK DP -wa 6 -wa 6 piit?? piit?? ?S/he is an eagle? ?eagle? In (11a) and (12a), a null pro appears in the Specifier of LINK, and the LINKP is interpreted as a saturated predicate, i.e., a proposition. In (11b) and (12b), there is no pro in Spec, LINKP, and the LINKP is interpreted as an argument. This flexibility in interpretation is a property of ?wa, and not a property of verb and noun roots. Armoskaite (2011) demonstrates that noun and verb roots in Blackfoot are inherently categorized as nominal and verbal respectively. The evidence for this is that roots themselves are restricted in terms of what functional material they can combine with. Nominal roots can combine with nominal but not verbal inflection (e.g., verb class finals), and verbal roots cannot combine with verbal but not nominal inflection (e.g., plural marking). Examples are given below. (13) a. ??kapayinists o?k-napayin-istsi raw-bread-PL ?flours? (Frantz and Russell 1995, p. 119) b. * Anna Jane ??kapayinatsiiwa omi ni?taw??kii. ann-wa J o?kapayin-at-yii-wa om-yi ni?tawaakii-yi DEM-PROX J flour-TA-3:4-PROX DEM-OBV chicken-OBV intended: ?-ane breaded the chicken.? (adapted from Armoskaite 2011, p. 92) (14) a. ??kottakiwa. yaak-ottak-i-wa FUT-give.drink-AI-PROX ?S/he will serve drinks.? (Frantz and Russell 1995, p. 145) b. * ottakiks ottak-iksi give.drink-PL intended: ?bartenders? (Armoskaite 2011, p. 62) 83 In (13), we see that the noun root napayin can take nominal inflection (plural ?istsi) but not verbal inflection (TA final ?at). Conversely in (14), we see that the verb root ottak can take verbal inflection (AI final ?i) but not nominal inflection (plural ?iksi). These data show that the nominal and verbal roots are not category-neutral; they are inherently categorized. As such, the predicate/argument flexibility observed with proximate expressions does not reduce to a property of the roots. Furthermore, unless the form is marked with ?wa it is not flexible in its interpretation. For example, if a form is marked with obviative ?yi, it can only be interpreted as an argument, as shown below. (15) a. * (Anni) ??hpiyiyi. (ann-yi) a-ihpiyi-yi (DEM-OBV) IMPF-dance.AI-OBV intended: ?S/he is dancing.? b. Anni ??hpiyiyi ?kaomatap??yin?yi. ann-yi a-ihpiyi-yi akaa-omatap-oo-yini-ayi DEM-OBV IMPF-dance.AI-OBV PERF-begin-go.AI-OBV-3SG.PRN ?The one who dances has just left.? (16) a. * Piit??yi. piitaa-yi eagle-OBV intended: ?S/he is an eagle.? b. * Omi piit??yi. om-wa piitaa-yi DEM-PROX eagle-OBV intended ?That is an eagle.? b. Om-yi pitt??yi ??pottayin?yi. om-yi piitaa-yi a-ipottaa-yini-ayi DEM-OBV eagle-OBV IMPF-fly.AI-OBV-3SG.PRN ?That eagle is flying.? The examples in (15) and (16) are identical to those in (9) and (10), except that obviative ?yi is used in place of proximate ?wa. Unlike proximate expressions, which can be interpreted as predicates or arguments, obviative expressions can only function as arguments. Additionally, expressions that are unmarked for the proximate/obviative contrast do not exhibit the flexibility observed with proximate-84 marked expressions. For example, whereas bare noun stems can function as unindexed objects of morphologically intransitive verbs, verb stems cannot. (17) a. Nits??yaapi piit??. nit-ii-yaapi piitaa 1-IC-see.AI eagle ?I saw an eagle.? b. * Nits??yaapi ??hpiyi. nit-ii-yaapi a-ihpiyi 1-IC-see.AI IMPF-dance.AI intended: ?I saw a dancer.? In (17), the same stems that exhibited predicate/argument flexibility when suffixed with ?wa appear without ?wa and only the noun is grammatical. The fact that ??hpiyi ?dance? cannot be interpreted as an argument unless it is suffixed with ?wa indicates that the predicate/argument flexibility is a property of ?wa and not a property of the thing to which it attaches. Conversely, whereas verb stems are used without proximate/obviative suffixes when they form imperative clauses42, nouns cannot be used in this context, as shown below. (18) a. * Piit??t! piitaa-t eagle-IMP intended: ?Be an eagle!? b. Ihpiy ?t! ihpiyi-t dance-IMP ?Dance!? Again, for a nominal or verbal form to be flexible in its interpretation, it requires ?wa. Put another way, the predicate/argument flexibility is a property of ?wa. To this point, the examples illustrating predicate/argument flexibility have been limited to verb or noun stems suffixed with ?wa. However, any form with a noun or verb stem is ambiguous between clause and argument readings if it is suffixed with ?wa. Some additional examples are given below. 42 Proximate/obviative suffixes cannot be used in imperative clauses. This is discussed in detail in Chapter 6. 85 (19) a. Oma nin?? it??miihkaawa omi niy?tahtaani. om-wa ninaa-wa it-a-omii-hkaa-wa om-yi niyitahtaan-yi DEM-PROX man-PROX LOC-IMPF-fish-acquire.AI-PROX DEM-INAN river-INAN ?That man is fishing at the river.? b. Nitsik?yaahsimaa oma nin?? it??miihkaawa omi niy?tahtaani. nit-ik-a-yaahsimaa oma ninaa-wa it-a-omii-hkaa-wa omi niyitahtaan-yi 1-INTNS-IMPF-like.AI DEM man-PROX LOC-IMPF-fish-acquire.AI-PROX DEM river-INAN ?I like that man who is fishing at the river.? (20) a. Oma nit??hkiitooka napay?n. om-wa nit-a-ihkiit-o-ok-wa napayin DEM-PROX 1-IMPF-bake-TA.BEN-INV-PROX bread ?S/he bakes bread for me.? b. Oma nit??hkiitooka napay?n ?kao?toowa. om-wa nit-a-ihkiit-o-ok-wa napayin akaa-o?too-wa DEM-PROX 1-IMPF-bake-TA.BEN-INV-PROX bread PERF-arrive.AI-PROX ?The one who bakes bread for me is here.? (21) a. Oma i?nak??kiikoana. om-wa i?nak-aakii-koan-wa DEM-PROX small-woman-DIM-PROX ?That?s a small girl.? b. Oma i?nak??kiikoana ??ksiksspitaawa ??sopoksistawa?ssi. om-wa i?nak-aakiikoan-wa yaak-ik-sspiitaa-wa a-sopok-istawa?si-hs-yi DEM-PROX small-girl-PROX FUT-INTNS-be.tall.AI-PROX IMPF-finish-grow.AI-CONJ-OBV ?That small girl will be tall when she stops growing.? (22) a. Anna nitsikok??pomitaama. ann-wa nit-ik-oka?p-imitaa-m-wa DEM-PROX 1-INTNS-bad-dog-POSS-PROX ?He is my bad dog.? b. Anna nitsikok??poPitaaPa ??sikstakiwa. ann-wa nit-ik-oka?p-imitaa-m-wa a-sikstaki-wa DEM-PROX 3-INTNS-bad-dog-POSS-PROX IMPF-bite.AI-PROX ?My bad dog bites.? (19) and (20) are examples of verbal forms marked with ?wa. In (19a), a verb modified by an adpositional prefix it- that introduces a locative oblique is construed as a predicate, and in (19b) it is construed as the object. In (20a), a morphologically intransitive verb with an unindexed object is construed as a predicate, and in (20b) it is construed as the subject. (21) and (22) are examples of nominal forms marked with ?wa. In (21a), a noun with an adjectival prefix and diminutive suffix is construed as a predicate, and in (21b) it 86 is construed as the subject. In (22a), a possessed noun marked with an adjectival prefix and an intensifier prefix is construed as as a predicate, and in (22b) it is construed as the subject. In short, so long as an expression is marked with ?wa, it can be interpreted as either a predicate or an argument. Given that any expression formed from a noun or verb stem and suffixed with ?wa can be interpreted as either a predicate or an argument, it is clear that the flexibility is a property of ?wa. In comparision, neither plural nor obviative forms are flexible between predicate and argument readings, as shown below. (23) a. Omiksi ??kskimaayaawa. om-iksi a-ikskimaa-yi-aawa DEM-PL IMPF-hunt.AI-PL-3PL.PRN ?Those guys hunt.? b. * Omiksi ??kskimaayaawa ak??sko?tooyaawa. om-iksi a-ikskimaa-yi-aawa akaa-ssk-o?too-yi-aawa DEM-PL IMPF-hunt.AI-PL-3PL.PRN PERF-back-arrive.AI-PL-3PL.PRN intended: ?Those hunters have come back.? (24) a. * Omiksi ??kskimaiks. om-iksi a-ikskimaa-iksi DEM-PL IMPF-hunt.AI-PL intended: ?Those are hunters.? b. Omiksi ??kskimaiks ak??sko?tooyaawa. om-iksi a-ikskimaa-iksi akaa-ssk-o?too-yi-aawa DEM-PL IMPF-hunt.AI-PL PERF-back-arrive.AI-PL-3PL.PRN ?Those hunters have come back.? (25) a. Ni Myaani ikaisskahsa?p???takiyin?yi. ann-yi M ik-aisskahs-a?p-a-o?taki-yini-ayi DEM-OBV M INTNS-always-around-IMPF-work.AI-OBV-3SG.PRN ?Mary is always working.? b. * Ni ikaisskahsa?p???takiyini ?kao?tooyin?yi. ann-yi ik-aisskahs-a?p-a-o?taki-yini akaa-o?too-yini-ayi DEM-OBV INTNS-always-around-IMPF-work.AI-OBV PERF-arrive.AI-OBV.3SG.PRN intended: The one who is always working is here.? (26) a. * Annihk aahks??yiihk. ann-yi-hk aahksa-a-ooyi-yi-hk DEM-OBV-INVIS always-IMPF-eat.AI-OBV-INVIS intended: ?S/he is always eating.? 87 b. Annihk aahks??yiihk iy??sta?pooyin?yi. ann-yi-hk aahksa-a-ooyi-yi-hk ii-yiistap-oo-yini-ayi DEM-OBV-INVIS always-IMPF-eat.AI-OBV-INVIS IC-away-go.AI-OBV-3SG.PRN ?That one who is always eating went away.? In (23), we see that forms marked with the verbal plural marker ?yi are necessarily predicative; they cannot be used to form arguments. Conversely, in (24), we see that forms marked with the nominal plural marker ?iksi are necessarily arguments; they cannot be used predicatively. This follows if ?yi and ?iksi differ in terms of their syntactic categories; whereas ?yi maps onto a functional category in the verbal domain (namely C, as I will argue in Chapter 5), -iksi maps onto a functional category in the nominal domain. A similar contrast is observed in (25) and (26); the obviative marker ?yini is restricted to predicative contexts, whereas obviative ?yi is restricted to argument interpretations. Again, this follows if ?yini and ?yi differ with respect to their syntactic categories; whereas ?yini is verbal, -yi is nominal. This is summarized in Table 3.2 below. Table 3.2. Number/obviation morphology Verbal Nominal Proximate -wa ? ? Plural -iksi / -istsi ? ? -yi ? ? Obviative -yi ? ? -yini ? ? In comparison to the plural and obviative markers, ?wa can associate with the highest head position of either the verbal or nominal spine. As such, it is not inherently categorized as either verbal (C) or nominal (K). Rather, it is neutral, as schematized below. (27) a. [LINKP -wa [IP [AspP [vP [VP]]]]] b. [LINKP -wa [DP [?P [nP [NP]]]]] In (27), -wa maps onto the head LINK, and can combine with either verbal (IP) or nominal (DP) structures. This is qualitatively different from saying that, in the verbal spine ?wa maps onto C, but in the nominal spine it maps onto the nominal equivalent of C, namely K. Instead, by using the label LINK for 88 both structures, I am claiming that ?wa is not categorized as either nominal nor verbal, but can associate with both nominal and verbal structures. Empirically, the consequence of this is that ?wa-marked expressions themselves are ambiguous; they can have the distribution and interpretation of either CPs or KPs. Note that the alternative option of employing different category labels for ?wa depending on whether it is nominal or verbal (i.e., K or C) would suggest that -wa can function as a ?switch? head, i.e., something that can either nominalize something otherwise verbal, or ?verbalize? something otherwise nominal (see Panagiotidis and Grohmann 2009; Kornfilt and Whitman 2011 for a discussion of ?switch? heads in nominalizations). In other words, under this analysis, when ?wa associates with K it would nominalize, and when it associates C it would verbalize. However, this analysis fails to capture the insight that ?wa is not inherently categorized as either nominal or verbal; it maps onto the linking layer of the spine, but is neutral with respect to the K/C distinction. 220.127.116.11. Proximate ?wa does not Associate with ?In the preceding section, I argued that ?wa associates with the highest functional head in the spine, and that it is neutral with respect to the nominal/verbal distinction. I proposed the label LINK to refer to this neutral functional head, and I demonstrated that phrases suffixed with ?wa can be intepreted as predicates or arguments. In this section, I argue against an alternative analysis, namely that ?wa maps onto the functional head ?. The reason for entertaining this analysis is that, as argued by D?chaine and Wiltschko (2002), ?Ps can function either as predicates or arguments. Given the flexibility observed with proximate expressions (i.e., that they can be construed as predicates or arguments), it is worth considering whether ?wa associates with ?, rather than LINK. I have two arguments against the ?P analysis of proximate nominal expressions. First, as discussed in ?18.104.22.168 and elaborated in detail in Chapter 6, there is evidence to suggest that, in the verbal domain, ?wa is associated with the linking layer (i.e., that it maps onto the highest functional head in the spine). I do not have compelling reasons to think that this same morpheme sits lower in the nominal 89 spine, and in fact, as observed in (7) with the ?verbalized? demonstratives, the lower position that is occupied by nominal plural morphology is not occupied by ?wa, suggesting that ?wa is consistently high. Second, in ?3.5, I argue that the nominal plural suffixes map onto ?. Part of my evidence for this is that plural nouns can function as unindexed objects of morphologically intransitive verbs. In ?3.4 and ?3.5, I argue that unindexed objects of morphologically intransitive verbs are pseudo-incorporated, and that they restrict the predicate without saturating an argument position. An example is given below. (28) Nits??pi pi?kss??ks. nit-yaapi pi?kssi-iksi 1-see.AI bird-PL intended: ?I saw (some) birds.? In (28), the animate plural noun pi?kss??ks ?birds? functions as the unindexed object of a morphologically intransitive verb, a grammatical function I argue is associated with pseudo-incorporation. Moreover, I argue that pseudo-incorporation can maximally target ?Ps; nominal expressions with more functional structure cannot be pseudo-incorporated. Thus, if proximate ?wa mapped onto ?, we would predict that proximate nominal expressions could be pseudo-incorporated, i.e., they could function as unindexed objects of morphologically intransitive verbs. This prediction is not borne out, as shown below. (29) * Nit??kskimaa ponok?wa. nit-a-ikskimaa ponoka-wa 1-IMPF-hunt.AI elk-PROX intended: ?I am hunting an/the elk (proximate).? The fact that proximate nouns such as ponok?wa ?elk? cannot be pseudo-incorporated suggests that they are not ?Ps. In sum, because proximate ?wa maps onto the highest functional head in the verbal spine, and because proximate nominal expressions do not have the distribution of other ?Ps (i.e., plural nouns), I conclude that they are not ?Ps. 90 3.2.2. Proximate Expressions are Adjuncts What is the syntactic function of ?wa? I have argued that it associates with LINK, the highest head in the spine. As discussed in Chapter 1, the syntactic function of the linking layer is to connect the clause to a superordinate structure or the larger discourse. Cross-linguistically, we see these functions exemplified with, for example, complementizers (C heads) that connect a subordinate clause to a higher clause, or case marking (K heads) that connect an argument expression to the clause. How does ?wa fit into this class of linking heads? I propose that ?wa signals that the expression is not linked, i.e., that it is syntactically independent. On clauses, this is evidenced by the fact that ?wa is restricted to matrix clauses only; it does not appear on subordinate clauses, as shown below (and discussed in detail in Chapter 6). (30) Im ??taatsootaawa. ii-maat-matt-sootaa-wa IC-NEG-again-rain-PROX ?It?s not raining anymore.? (31) a. Nits?kssta m??hksawaatsootaahsi. nit-ik-sst-aa m-aahk-saw-matt-sootaa-hs-yi 1-INTNS-want-AI 3-MOD-NEG-again-rain.AI-CONJ-OBV ?I want it to stop raining.? (lit: ?I want it not to rain again.?) b. * Nits?kssta m??hksawaatsootaahsiwa. nit-ik-sst-aa m-aahk-saw-matt-sootaa-hs-yi-wa 1-INTNS-want-AI 3-MOD-NEG-again-rain.AI-CONJ-OBV-PROX intended: ?I want it to stop raining.? In (30), we see that ?wa is obligatory on the matrix clause, but in (31) we see that it is ungrammatical on the subordinate clause. As for proximate nominal expressions, I propose that they too cannot be syntactically dependent. In other words, although they can be interpreted as arguments, they do not have the syntactic distribution of arguments; they do not appear in argument positions. To formalize this insight, I suggest that they are subject to an anti-A-position condition that is part of the lexical entry for -wa, as stated in (32). 91 (32) Anti-A-Position Condition -wa cannot head an XP in an A-Position. Whether the condition in (32) can be derived from other primitives in the grammar is yet unclear. Although it is a ?brute force? stipulation, the anti-A-position condition on ?wa nevertheless ensures that phrases headed by ?wa are syntactically independent. The consequence of this for proximate argument expressions is that, although interpreted as arguments, they do not function as syntactic arguments and they do not appear in argument positions inside the clause. Rather, I propose that they are adjoined outside the clause, and bind a null pro inside the clause. In essence, this is the same implementation as Baker?s (1991, 199) Pronominal Argument Hypothesis (PAH): the argument position with which a proximate argument expression is associated is occupied by a null pro. However, as discussed in 22.214.171.124, when LINKP is interpreted predicatively, it licenses a null pro in its Specifier. Under my proposal, the pro introduced in LINKP binds a null pro in argument position, as shown below. (33) LINKP LINKPi LINKP 5 3 oma p??taawa proi 3 LINK IP -wa 3 I AspP 3 proi 3 Asp vP 3 <proi> 6 ??paawaani ?The eagle is flying? In (33), the nominal expression oma p??taawa ?that eagle? is sister to the verbal expression ??paawaaniwa, and it binds a null pro in Spec, LINKP. This pro binds a pro in argument position, i.e., that maps onto Spec,vP and moves to Spec, AspP. (See Chapter 2 for a discussion of these argument positions. 92 In a nutshell, I assume that the subject is introduced in Spec, vP and the Point-of-View (PoV) holder occupies Spec, AspP; if the subject is the PoV holder, then it moves to Spec, AspP.) The proposal that ?wa can optionally license a null pro Specifier captures the empirical observation that proximate nominal expressions are ambiguous. They can be interpreted as arguments (as in (33)) or they can be interpreted as predicates, as shown below. (34) Oma aak??koana ??hpiyiwa. om-wa aakii-koan-wa a-ihpiyi-wa DEM-PROX woman-DIM-PROX IMPF-dance.AI-PROX ?She is a girl (and) she is dancing.? In (34), a proximate nominal expression oma aakiikoana ?that girl? combines with a proximate clause ??hpiyiwa. Indeed this can yield a subject-predicate structure (i.e., ?That girl is dancing?). However, it can be equally interpreted as in (34), as a conjunction of two clauses, both of which license a null pro, as in (35). (35) LINKP LINKP LINKP 3 3 pro 3 pro 3 LINK DP LINK IP -wa 6 -wa 6 oma aak??koan ??hpiyi Moreover, because LINKPS are ambiguous and can be formed from either nominal or verbal stems, either two proximate nouns or two proximate verbs can combine with each other, as shown below. (36) Oma nin??wa imit??wa ! om-wa ninaa-wa imitaa-wa DEM-PROX man-PROX dog-PROX ?That man is a dog!? (OR ?That dog is a man.?) 93 (37) Annahk n?ssi?sa ?stohkanainaawa ?tsskinaa?yiks. ann-wa-hk n-iss-i?s-wa isstohkana-ninaa-wa iitsskina?yiiksi DEM-PROX-INVIS 1-young-brother-PROX most-man-PROX Horns.Society ?My younger brother is head of the Horn Society.? (OR ?The head of the Horn Society is my brother.?) (38) Oma ??kskimaawa ?yo?kaawa. om-wa a-ikskimaa-wa a-yo?kaa-wa DEM-PROX IMPF-hunt.AI-PROX IMPF-sleep-PROX ?That hunter is sleeping.? (OR ?That sleeping one is hunting.?) In (36), there are two nouns in a predicative relationship, and in (37), two nouns are in equative relationship.43 In both cases, the nouns are both marked with ?wa. In (38) there are two proximate verbs. All three sentences are ambiguous as to to which proximate expression is interpreted as the argument and which one is interpreted as the predicate. In fact, clauses with two proximate expressions are many ways ambiguous with respect to what is interpreted as the predicate and what is interpreted as the argument. This is shown in (39) and (40). (39) Oma aak??koana ??kska?siwa. om-wa aakii-koan-wa a-okska?si-wa DEM-PROX woman-DIM-PROX IMPF-run-PROX ?That girl is running.? ? OR ?That girl is a runner.? ? OR ?That runner is a girl.? ? OR ?She is a girl (and) she is running.? ? OR ?She is a girl (and) she is a runner.? (40) Oma aak??koana ??kska?siwa P??no?toowa. om-wa aakii-koan-wa a-okska?si-wa maan-o?too-wa DEM-PROX woman-DIM-PROX IMPF-run.AI-PROX recent-arrive.AI-PROX ?The girl who is running just arrived.? OR ?The girl who is a runner just arrived.? The sentence in (39) has both a proximate noun (modified by a demonstrative) and a proximate verb. One interpretation of this sentence has the noun interpreted as the argument, and the verb interpreted as the predicate (?That girl is running / is a runner?). However, a number of alternative interpretations are 43 There is seemingly no empirical difference between predicative and equative sentences; under the right discourse conditions, (3) could be interpreted as eTuative (?That man is the dog?) and (37) could be interpreted as predicative. Whether the predicatives versus equative interpretations correspond to predicative versus equative structures is yet to be explored. 94 available. For example, the sentence can be interpreted in the opposite way, with the verb as the argument and the noun as the predicate (?That runner is a girl.?) Additionally, both the noun and the verb could be interpreted predicatively, with translations along the lines of ?She is a girl (and) she is running? or ?She is a girl (and) she is a runner.? Finally, both the noun and verb could be interpreted as arguments, as shown in (40).44 In short, whether ?wa combines with a nominal or verbal spine, it can yield either predicate or argument interpretations. Regardless of their interpretation, however, LINKPs do not appear in argument positions; they are adjoined outside the clause. As adjuncts, they are predicted to be optional and able to adjoin to the right or the left of the clause.45 This prediction is borne out, as shown below. (41) a. N?tsspommoawa anna n?nsstsinaana. nit-sspommo-a -wa ann-wa n-insst-innaan-wa 1-help.TA-DIR-PROX DEM-PROX 1-sister-1PL-PROX ?I helped our sister.? b. Anna n?nsstsinaana n?tsspommoawa. ann-wa n-insst-innaan-wa nit-sspommo-a -wa DEM-PROX 1-sister-1PL-PROX 1-help.TA-DIR-PROX ?I helped our sister.? c. N?tsspommoawa. nit-sspommo-a -wa 1-help.TA-DIR-PROX ?I helped her.? In (41a), the proximate argument expression anna n?nsstsinaana ?our sister? appears postverbally, and in (41b) it appears preverbally. In (41c), the argument expression is null. The flexibility and optionality of proximate expressions is predicted by the adjunction analysis.46 44 Regarding eliciting these interpretations in fieldwork contexts, it is not the case that my consultants supply all of the different interpretations for a given Blackfoot sentence, but rather that each of the English translations can be rendered with the same Blackfoot sentence. 45 The idea that there is both right- and left-adjunction is contra Kayne (1994), who claims that, universally, adjunction is strictly on the left. However, it is consistent with Baker?s (199, 200) claim that (many) polysynthetic languages permit both right- and left- adjunction. 46 Obviative and plural expressions do not show the same flexibility as proximate ones. Unlike proximate expressions, when obviative and plural argument expressions are preverbal or null, they are obligatorily indexed by an enclitic on the verb. This is discussed in detail in Chapter 7. 95 Furthermore, the adjunction analysis predicts that, if proximate expressions are adjuncts, they should be able to iterate, as schematized below.47 (42) LINKP LINKP LINKP 5 LINKP LINKP 5 LINKP LINKP 5 ? This prediction is borne out, as shown in (43) and (44). (43) Anna imitaawa kakokihka?siwa ??sikstakiwa ??hkiwa. ann-wa imitaa-wa kak-okihka?si-wa a-sikstaki-wa a-ohki-wa DEM-PROX dog-PROX only- INTNS-misbehave.AI-PROX IMPF-bite.AI-PROX IMPF-bark.AI-PROX ?That dog acts badly (and) he bites (and) he barks.? (44) Anna Rosie iiss?pihkitaawa ??ksisako iihpihk?taawa ? ann-wa R ii-ssap-ihkitaa-wa i?ksisako ii-ohp-ihkitaa-wa DEM-PROX R IC-inside-cook.AI-PROX meat IC-ACCOMP-cook.AI-PROX ? niistsik?pa?si ki pis?tsiinikim niistsika?a?si ki pisatsiinikimm carrot and onion ?Rosie roasted meat with carrots and onions.? /it: ?Rosie cooked meat inside and she cooked it with carrots and onions.? (43) and (44) are each judged as single sentences. In (43), a series of proximate expressions are in a single clause and all refer to the same individual. In (44) the first verb iiss?pihkitaawa ?roasted? has an unindexed object ??ksisako ?meat,? and the second verb iihpihk?taawa ?cooked with? has an obliTue argument niistsik?pa?si ki pis?tsiinikim ?carrots and onions,? introduced by the adposition ohp-. Under the analysis developed here, each proximate expression in these two clauses is a LINKP adjunct that binds a null pro in the LINKP following it, as shown in (45). 47 In fact, the adjunction analysis predicts that there could be many other tree structures as well. The implications of this are yet to be explored. 96 (45) LINKP LINKP i LINKP 5 that dog LINKP LINKP 5 ?proi acts badly ? LINKP LINKP 5 5 ?pUoi bites ? ?pUoi barks 3.2.3. Summary In summary, in this section, I have argued that proximate nominal expressions are LINKPs that adjoin outside the clause. They may be interpreted as predicates or arguments, but in either case they are subject to an anti-A-position condition on ?wa, ensuring that they do not appear in argument positions. As adjuncts, they can be omitted, can be freely ordered with respect to one another, and can iterate. This contrasts with obviative argument expressions, which I demonstrate in the following section are neither neutral with respect to the nominal/verbal distinction, nor adjoined outside the clause. 3.3. The Syntax of Obviative Singular Nominal Expressions In this section, I discuss the syntax of obviative singular nomina l expressions. I argue that they are headed by the obviative suffix ?yi, which maps onto a functional head K. In terms of its syntactic function, I propose that obviative ?yi links the argument expression to the clause. More specifically, I develop an analysis of ?yi as a generalized case marker; it is required on all argument expressions inside the clause, but it does not co-vary with grammatical function. Under this analysis, obviative nominal expressions are KPs that map onto argument positions inside the clause. This section proceeds as follows: In ?3.3.1, I argue for the categorial status of obviative nominal expressions as KPs, and in ?3.3.2, I develop an analysis of obviative ?yi as a case marker. 97 3.3.1. Obviative Singular Nominal Expressions are KPs In ?3.2, I argued that the proximate suffix ?wa maps onto the highest functional head in the spine. Here I propose that the same is true of the obviative suffix ?yi. The two suffixes, -wa and ?yi, are in complementary distribution, both occurring at the right edge of a nominal expression. Examples are given below. (46) a. si?k??na si?kaan-wa blanket-PROX b. si?k??ni si?kaan-yi blanket-OBV b. * si?k??n-wa-yi c. * si?k??n-yi-wa Given the complementarity of ?wa and ?yi, I assume that ?yi occupies the same syntactic position as ?wa. However, I propose that the two differ with respect to their categorial identity; whereas ?wa maps onto the head LINK, -yi maps onto K. This proposal is based on the observation that, whereas ?wa is neutral with respect to whether it can head a nominal or verbal structure, -yi is restricted to nominal contexts only. Evidence for this comes from the observation that ?yi cannot appear on clauses; it is restricted to argument expressions. Examples are given below. (47) a. Anni otakk??yi otohk?kin?yi omistsi pis?tssaisskists. ann-yi w-itakkaa-yi ot-ohkot-ok-yini-ayi om-istsi pisatssaisski-istsi DEM-OBV 3-friend-OBV 3-give.TA-INV-OBV-3SG.PRN DEM-PL flower-PL ?Her friend gave her those flowers.? b. * Anni otakk??yi. ann-yi w-itakkaa-yi DEM-OBV 3-friend-OBV intended: ?That is her friend.? c. Anni otakk??yin?yi. ann-yi w-itakkaa-yini-ayi DEM-OBV 3-friend-OBV-3SG.PRN ?That is her friend.? 98 (48) a. Annahk Beth ik??sstsi?iPii annisk ot??ssksiniPa?tsoki. annahk B ik-a-istsimm-yii-wa annisk ot-a-issksinima?tsi-ok-yi DEM B INTNS-IMPF-admire.TA-3:4-PROX DEM 3-IMPF-teach.TA-INV-OBV ?Beth admires her teacher.? (lit: the one who teaches her) b. * Annisk otaissksiniPa?tsoki. annisk ot-a-issksinima?tsi-ok-yi DEM 3-IMPF-teach.TA-INV-OBV Intended: ?S/he teaches him/her.? c. Annisk otaissksiniPa?tsoka. annisk ot-a-issksinima?tsi-ok-wa DEM 3-IMPF-teach.TA-INV-PROX ?S/he teaches him/her.? d. Annisk otaissksiniPa?tsokin?yi. annisk ot-a-issksinima?tsi-ok-yini-ayi DEM 3-IMPF-teach.TA-INV-OBV-3SG.PRN ?S/he teaches him/her (obv)? In (47a), the obviative noun otakk??yi functions as an argument (the subject), and in (47b) it functions as a clause and this is ungrammatical. For the noun to function predicatively, it must be marked with the obviative suffix from the verbal paradigm, -yini, as in (47c). Similarly, in (48a), the inflected verb form ot??ssksiniPa?tsok ?s/he teaches him/her? is marked with the obviative suffix ?yi, and as such, it can function as an argument (the object). However, this form cannot function as a clause, as shown in (48b). A predicative reading is available only if the inflected verb is suffixed with the ?wa (if the object is proximate, as in (48c)) or ?yini (if the object is obviative, as in (48d)). In short, obviative expressions marked with ?yi are necessarily argument expressions; they cannot be interpreted as predicates. This distinguishes ?yi from its proximate counterpart ?wa. In other words, whereas ?wa is neutral with respect to whether it can head nominal or verbal structures, -yi is inherently nominal. As such, I propose that it maps onto the functional head K. 3.3.2. Obviative Singular ?yi is a Generalized Case Marker What is the syntactic function of K? As discussed in Chapter 1, I assume that the core function of K is to link an argument expression to the superordinate structure, i.e., the clause. Moreover, I assume that, in at 99 least some languages, K is associated with case morphology. Thus, I claim that obviative ?yi is a case marker.48 In more familiar case systems (e.g., the nominative/accusative systems of Indo-European languages), case (K) features co-vary with grammatical function: nominative case is assigned to the subject, and accusative case to the direct object.49 This can be observed in the pronominal system of English, as shown below. (49) a. She walked the dog. b. The dog walked her. Not all case systems are organized in this way. Argument expressions in ergative/absolutive case systems also co-vary with grammatical functions, but along different lines: the subject of an intransitive clause and the object are marked with one case (absolutive), and the subject of a transitive clause with another (ergative). This is illustrated with data from Dyirbal below. (50) a. ?uma banaga -nyu. father.ABS return-NONFUT ?Father returned.? b. yabu ?uma-?gu bura -n. mother.ABS father-ERG see-NONFUT ?Mother saw father.? (Dixon 1994: 10) There are various formalizations of nominative/accusative and ergative/absolutive case systems in the literature. For instance, case alternations have been be formalized with appeal to Agree relations with case-assigning heads (e.g., finite T assigns nominative, v assigns accusative, e.g., Chomsky 1995) or with appeal to spell-out operations that are sensitive to the dependency of one case on another (e.g., 48 This analysis is contra Ritter and Rosen (2005), who claim that Algonquian languages lack case altogether. See also Bruening (2009) for arguments against Ritter and Rosen?s analysis. 49 This characterization simplifies the facts somewhat. For example, ECM subjects in English are not nominative but accusative case-marked (Chomsky 1981) and quirky subjects in Icelandic are not nominative but dative case-marked (Maling 1990). 100 accusative/ergative case is a dependent case, spelled out in the presence of a nominative/absolutive, e.g., Marantz 1991, McFadden 2004). Setting aside the question of which model best captures these types of case alternations, the empirical observation is that Blackfoot?s case system is markedly different: it does not co-vary with grammatical function. Rather, aside from at most one proximate expression adjoined to the clause, all argument expressions in a clause are marked with the K head ?yi, regardless of their grammatical function.50 This is summarized in Table 3.3 below. Table 3.3. Case Systems Grammatical function English Dyirbal Blackfoot Subject - transitive Nominative Ergative Obviative Subject - intransitive Absolutive Object - transitive Accusative In what sense is obviative -yi a case marker? I propose that ?yi simply signals that an argument expression is linked to the clause. Just as nominative case, for example, links an argument expression to the clause by signalling that it is in the subject position, obviative ?yi links an argument expression to the clause by signalling that it is in an argument position. As such, it can be thought of as a generalized case marker, not specified for any particular case position, but marking argument expressions as linked to the clause. Just as case is thought to be required on argument expressions in languages with case alternations (e.g., the Case Filter, Chomsky 1981), I propose that case is also required on argument expressions in Blackfoot. I formalize this via the following condition: (51) Linking Condition on Argument Expressions An argument expression can appear in an argument position inside the clause iff it is a KP. The insight that (51) is intended to capture is that, if the function of K is to link argument expressions to the clause, then an argument expression cannot be linked unless it is a KP. The empirical consequence of this is that, in Blackfoot, all argument expressions are KPs. 50 This is an overgeneralization; plural argument expressions are not marked with -yi. However, in ?3.5 below I argue that plural argument expressions are case marked, but the realization of ?yi in the surface string is blocked by the plural suffix. 101 The analysis of obviative ?yi as a K head that links the argument expression to the clause accounts for a number of generalizations about obviative nominal expressions. First, it correctly predicts that obviative nominal expressions cannot function independently as clauses; examples illustrating this were given in (47) and (48) above, and an additional example is given below. (52) a. Ann?sk onssts ook??wayi iisstsits??yin?yi. an-yi-hk w-insst-yi w-ookoowa-yi ii-sstsitsii-yini-ayi DEM-OBV-INVIS 3-sister-OBV 3-home-OBV IC-burn.II-OBV-3SG.PRN ?His sister?s home burnt down.? b. * Ann?sk onssts ook??wayi. an-yi-hk w-insst-yi w-ookoowa-yi DEM-OBV-INVIS 3-sister-OBV 3-home-OBV intended: ?That is his sister?s home.? In (52), we see that the obviative nominal expression ann?sk onssts ook??wayi ?his sister?s home? can function as argument (here the subject), but it cannot function independently as a clause. This is consistent with the analysis of obviative ?yi as a K head that signals dependency. Furthermore, if ?yi is a case marker that appears on all arguments, then nominal expressions marked with -yi should not be restricted in terms of which grammatical function they associate with. This prediction is borne out; as shown in Table 3.3, obviative nominal expressions can fulfill the grammatical roles associated with both nominative and accusative (or ergative and absolutive) cases. This is shown below. (53) ??soksstayini anni otssitsimaani. a-sok-sstaa-yini ann-yi ot-issitsimaan-yi IMPF-well-nurse.AI-OBV DEM-OBV 3-baby-OBV ?Her baby is nursing well.? (54) ??hpommatooma omi as?ka'simi. ii-ohpomm-atoo-m-wa om-yi asoka?sim-yi IC-buy-TI-3:INAN-PROX DEM-OBV dress-OBV ?She bought that dress.? In (53), the obviative nominative nominal expression funtions as the subject, and in (54) it functions as the indexed object. In addition, obviative nominal expressions can fulfill all other grammatical functions 102 as well (aside from unindexed objects of morphologically intransitive verbs, which I will argue in ?3.4.1 are pseudo-incorporated.) Examples are given below. (55) ??hpomm oyiiwa anni ot?ni amoyi as?ka'simi. ii-ohpomm-o-yii-wa ann-yi w-itan-yi amo-yi asoka?sim-yi IC-buy-TA-3:4-PROX DEM-OBV 3-daughter-OBV DEM-OBV dress-OBV ?She bought that dress for her daughter.? (56) Anna Beth ??kohtahtsaowaihkitaawa annisk Heather. ann-wa B yaak-ohtahtsaowa-ihkitaa-wa ann-yi-hk H DEM-PROX B FUT-instead.of-cook.AI-PROX DEM-OBV-INVIS H ?Beth will take Heather?s place in cooking.? In (55), the obviative nominal expression functions as the unindexed object of a ditransitive verb, and in (56) it functions as an oblique. In short, obviative nominal expressions are unrestricted with respect to grammatical function. This is consistent with the analysis of ?yi as a case marker that is required on every argument expression inside the clause. Finally, if ?yi simply spells out the dependency relation between a clause and its arguments, then we predict that it should not be restricted to nominal arguments. Rather, under the Linking Condition in (51), regardless of whether a linguistic object in an argument position refers to an individual or a proposition, it should be a KP. This prediction is borne out. Just as nominal argument expressions are marked with the suffix ?yi, so are clausal arguments. In particular, in addition to the conjunct suffix ?hs, subordinate conjunct clauses require a morpheme ?yi whose function has until now been unexplained (cf. Frantz 1991, 2009). Examples are given below. (57) Anna Rosie ??ssksinima nit??ksspommowahsi anni Leo. ann-wa R ii-ssksin-i-m-wa nit-aak-sspomm-o-a-hs-yi ann-yi L DEM-PROX R IC-know-TI-3:INAN-PROX 1-FUT-help-TA-DIR-CONJ -OBV DEM-OBV L ?Rosie knows that I?m going to help /eo.? (58) ?kssoka?piiwa kit?yiitsittsimaahsoaayi. ik-sok-a?pii-wa kit-a-yiitsittsimaa-hs-oaawa-yi INTNS-good-be.AI-PROX 2-IMPF-slice.meat.AI-CONJ -2PL-OBV ?It?s good that you (pl) are thinly slicing meat.? 103 In (57), the subordinate clause is marked with a conjunct suffix ?hs plus a suffix ?yi. In (58), these two suffixes are separated by the 2nd person plural suffix ?oaawa. As such, -yi occupies the same morphological position as the proximate and obviative suffixes that appear on clauses. Unlike -yi, proximate ?wa is not permitted on conjunct clauses (see (31) above) and neither is the obviative agreement suffix ?yini that appears on matrix clauses. (This is discussed in detail in Chapter 6.) However, the fact that ?yi appears on conjunct clauses is predicted by the analysis of ?yi as a case marker; as a complement of the matrix verb, the conjunct clause occupies an argument position.51 3.3.3. Summary: Proximate versus Obviative Nominal Expressions In this section, I have discussed the syntax of obviative nominal expressions. I have claimed they are headed by the obviative suffix ?yi which maps onto a functional head K. The obviative suffix is case marker; it functions to link the expression to the clause, and it is required on all expressions occupying argument positions. Compared with proximate ?wa, which I argued in ?3.2 marks the expression as independent, obviative ?yi marks the expression as dependent; by linking it to the clause, it signals a dependency relation between the argument expression and the clause. A summary of the differences between proximate and obviative nominal expressions is given in Table 3.4. Table 3.4. Proximate versus Obviative Nominal Expressions Proximate ( -wa) Obviative ( -yi ) Syntactic category LINKP KP Syntactic function Marks phrase as independent Marks phrase as dependent / linked Syntactic position Adjoined to LINKP Argument position inside the clause 51 The claim that clausal arguments in Blackfoot are case-marked goes against Stowell?s (1981) Case Resistance Principle (CRP), which states that clausal arguments resist case. However, the lack of a clear partition between clausal and nominal arguments appears to a more general property of Blackfoot. For instance, there are no relative clauses in Blackfoot; clausal nominalizations (which may also be case-marked with ?yi) are used instead (cf. Bliss, to appear). As such, there is ample evidence to suggest that the CRP does not hold up in Blackfoot. 104 3.4. The Syntax of Bare Nouns In this section, I discuss the syntax of bare nouns.52 Bare nouns are restricted in distribution; they cannot function as subjects or indexed objects, as shown below. (59) a. * Iss?tsimaan ?waasai?niwa. issitsimaan a-waasai?ni-wa baby IMPF-cry.AI-PROX intended: ?A baby was crying.? b. Oma iss?tsimaana ?waasai?niwa. om-wa issitsimaan-wa a-waasai?ni-wa DEM-PROX baby-PROX IMPF-cry.AI-PROX intended: ?The baby was crying.? (60) a. * Nitaahk?niay si?k??n. nit-waahkani-a-yi si?kaan 1-sew.TA-DIR-PL blanket intended: ?I sewed a blanket.? b. Nitaahk?niay amoksi si?k??niks. nit-waahkani-a-yi amo-iksi si?kaan-iksi 1-sew.TA-DIR-PL DEM-PL blanket-PL ?I sewed those blankets.? In (59a), the bare noun iss?tsimaan is ungrammatical as the subject, and similarly in (60a), the bare noun si?k??n is ungrammatical as the object. (59b) and (60b) demonstrate that subjects and indexed objects require number marking, as well as demonstratives (see Appendix A for a detailed survey). However, bare nouns can function as the unindexed objects of morphologically intransitive (AI) verbs, henceforth referred to as AI OBJECTS. An example is given below. (61) Nits??pi ??papomm. nit-yaapi aipapomm 1-see.AI lightning ?I saw (a flash of) lightning.? In this section, I propose that AI objects are pseudo-incorporated, in the sense of Massam (2001). In ?3.4.1, I discuss the properties of pseudo-incorporation cross-linguistically, and I establish a set of 52 Frantz (2009: 11-12) discusses a ?non-particular? suffix ?i that appears on otherwise bare nouns. This suffix is absent from the grammars of both of my consultants. 105 diagnostics for pseudo-incorporation in Blackfoot. In the remaining subsections, I demonstrate that AI objects meet these diagnostics; they have the external syntax (?3.4.2), semantic characteristics (?3.4.3), and internal symtax (?3.4.4) of pseudo-incorporated nominal expressions. 3.4.1. Diagnosing Pseudo-Incorporation Cross-linguistically, there are various constructions in which a nominal expression has a ?tighter-than-normal? relation to the verb. Put another way, in many languages, there are semantically and/or syntactically impoverished nominal expressions that may satisfy the thematic requirements of a verb, but fail to function as full-fledged arguments of the predicate. For example, Hungarian (Farkas and de Swart 2003), Turkish (Bliss 2003), Mandarin (Rullmann and You 2006), and many other languages contrast bare nouns (NPs) with DP arguments; the former show a tighter connection to the verb (e.g., they may be restricted to positions adjacent to the verb) and they are typically narrow-scoping, indefinite or non-specific, and are often number-neutral.53 Similarly, many languages (including Blackfoot)54 exhibit noun incorporation, a phenomenon in which a noun forms a morphological and/or syntactic unit with the verb, e.g., an N-V compound. While there is considerable variation in these types of constructions, the general observation is that many languages permit close relationships between a verb and a nominal expression, and in such cases the nominal expression is typically in some way deficient, i.e., not a full DP. Acknowledging the heterogeneity of such constructions, van Gee
UBC Theses and Dissertations
The Blackfoot configurationality conspiracy : parallels and differences in clausal and nominal structures Bliss, Heather Anne 2013
Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
- 24-ubc_2014_spring_bliss_heather.pdf [ 5.36MB ]
- JSON: 24-1.0165708.json
- JSON-LD: 24-1.0165708-ld.json
- RDF/XML (Pretty): 24-1.0165708-rdf.xml
- RDF/JSON: 24-1.0165708-rdf.json
- Turtle: 24-1.0165708-turtle.txt
- N-Triples: 24-1.0165708-rdf-ntriples.txt
- Original Record: 24-1.0165708-source.json
- Full Text