Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Verbal plurality and adverbial quantification : a case study of Skwxú7mesh (Squamish Salish) Bar-El, Leora Anne 1998

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_1998-0349.pdf [ 3.74MB ]
Metadata
JSON: 1.0088526.json
JSON-LD: 1.0088526+ld.json
RDF/XML (Pretty): 1.0088526.xml
RDF/JSON: 1.0088526+rdf.json
Turtle: 1.0088526+rdf-turtle.txt
N-Triples: 1.0088526+rdf-ntriples.txt
Original Record: 1.0088526 +original-record.json
Full Text
1.0088526.txt
Citation
1.0088526.ris

Full Text

VERBAL A  PLURALITY  AND ADVERBIAL  QUANTIFICATION:  C A S E S T U D Y O F SKWX.U7MESH (SQUAMISH  SALISH)  by LEORA ANNE BAR-EL B . A . (Hons), University of Western Ontario, 1996  A THESIS SUBMITTED IN P A R T I A L F U L L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Linguistics)  W e accept this thesis as conforming to the required standard  T H E UNIVERSITY OF BRITISH C O L U M B I A October 1998  © Leora Anne Bar-el, 1998  In  presenting this  degree at the  thesis  University  in  partial  fulfilment  of  the  requirements  of British  Columbia,  I agree that the  for  an  advanced  Library shall make it  freely available for reference and study. I further agree that permission for extensive copying of  this thesis for scholarly purposes may be  granted by  the  department  or  understood  that  by  his  or  her  representatives.  It  is  publication of this thesis for financial gain shall not be allowed without permission.  Department The University of British Columbia Vancouver, Canada  DE-6 (2788)  head of my copying  or  my written  ABSTRACT  The goal of this thesis is to present an analysis of verbal plurality and adverbial quantification in Skwxwu7mesh (Squamish Salish). This thesis provides a detailed analysis of a phenomenon in S k w x w u 7 m e s h that has never been explored: the effect of the auxiliary wa on predicates from various aspectual classes in both non-quantified and quantified sentences,  wa has been described as a morpheme  referring to a process that has duration either i n the form of a single act or the regular performance of it (Kuipers 1967). T w o central questions w i l l be addressed in this thesis. Firstly, what is the function of the auxiliary wa in Skwxwu7mesh?. In other words, why is wa obligatorily present for certain interpretations of predicates and obligatorily absent for others; furthermore, what does wa do to a predicate to yield the various readings? Secondly, why is wa obligatory with adverbs of quantification? T o answer these questions, this thesis proposes that wa is a pluractional marker that pluralizes the head of a predicate's event structure or the event type denoted by the predicate. A s s u m i n g Pustejovsky's (1991, 1995) event structure m o d e l representing the distinction between three primitive event types (states, processes, transitions), four aspectual classes are analyzed (activities , accomplishments , achievements and states) i n both English and Skwxwu7mesh. This thesis argues that Skwxwu7mesh provides crucial evidence that all bare predicates (that is, predicates without wa) are telic, with the exception of individual-level predicates, wa causes a predicate to be atelic v i a pluralization; this atelicity is marked by continuous and/or habitual readings for the predicates of the various classes. A s a consequence of these claims, this analysis suggests that activities and stage-level states are not primitives universally. This thesis argues that Kratzer's (1995) analysis of adverbs of quantification as unselective binders cannot account for Skwxwu7mesh; thus, adopting D e Swart's (1993,  ii  1995) event based approach to analyzing adverbial quantification, this thesis claims that S k w x w u 7 m e s h provides crucial evidence that Q-adverbs quantify over events only. The evidence derives from the fact that the pluractional marker wa is obligatory with both stagelevel stative predicates and individual-level predicates when they combine with a Q-adverb. The analysis presented i n this thesis claims that wa is the source of the plurality of events over which a Q-adverb quantifies.  iii  TABLE  OF CONTENTS  Abstract Table  ii  of  Contents  iv  Acknowledgement  yi  Dedication 1. 2.  INTRODUCTION  1  SOME BASIC M O R P H O L O G Y  2.1. 2.2.  2.3. 2.4. 3.  vii  THE  AND SYNTAX  Radical Headmarking The Verbal Complex 2.2.1. Tense 2.2.2. A u x i l i a r i e s 2.2.2.1. wa. 2.2.2.2. I hit' Overt D P s W o r d Order  SK_WX_WU7MESH  FACTS OF S&WX,WU7MESH  , .' •  PARADOX  3.1. S t a g e - L e v e l Stative Predicates 3.2. I n d i v i d u a l - L e v e l Predicates 3.3. Q u a n t i f i e d Sentences 3.4. T w o P r o b l e m s  4.  PLURACTIONAL  5.  SOLVING  M  ARKERS  4 . 1 . D e f i n i n g Pluractional M a r k e r s 4.2. Event/Phase L e v e l Repetition THE  FIRST  PROBLEM  5  .5 7 7 9 10 11 1.1 14 ...15  15 ..16 18 .....19 21  21 22  .25  5.1. T e r m i n o l o g y : T e l i c / A t e l i c 5.2. Theoretical M o d e l : Pustejovsky (1991, 1995) 5.2.1. E v e n t Structure 5.2.2. E v e n t Headedness 5.3. The M o d e l A p p l i e d : E n g l i s h : 5.3.1. A c h i e v e m e n t s 5.3.2. A c t i v i t i e s 5.3.3. A c c o m p l i s h m e n t s ; 5.3.4. States 5.4. The M o d e l A p p l i e d : S k w x w u 7 m e s h 5.4.1. A c h i e v e m e n t s ". 5.4.2. A c t i v i t i e s 5.4.2.1. A Comparison with French 5.4.2.2. The Event Structures of A c t i v i t i e s 5.4.3. A c c o m p l i s h m e n t s 5.4.4. States 5.4.4.1. Stage-Level Stative Predicates 5.4.4.2. I n d i v i d u a l - L e v e l Predicates 5.5. S u m m a r y .  iv  25 ..26 26 .......27 28 28 ....30 .......31 32 34 34 ..38 40 45 50 55 55 59 ..61  6.  SOLVING  THE  SECOND  P R O B L E M  ...64  6.1. Quantification over Events/Individuals: Kratzer (1995) 6.1.1. Temporal and Atemporal Interpretations 6.1.2. The Stage-Level/Individual-Level Distinction 6.1.3. When-Clauses 6.1.4. Simplex Clauses 6.2. The Event-Based Approach: De Swart (1993, 1996) 6.2.1. The Analysis of Q-adverbs in English 6.2.1.1. Stage-Level Stative Predicates 6.2.1.2. Individual-Level Predicates 6.2.2. The Analysis of Q-adverbs in Skwxwu7mesh 6.2.2.1. Stage-Level Stative Predicates 6.2.2.2. Individual-Level Predicates  7. T H E O R E T I C A L  IMPLICATIONS  AND  FURTHER  .*.  ;  ISSUES  65 ....65 66 68 70 72 73 73 74 75 75 76 78  7.1. Cross-Linguistic Implications for Quantification: On the Availability of Unselective Binding in Salish 79 7.2. Cross-Linguistic Implications for Aspectual Classes: Primitives in Event Structure Models 79 7.3. Issues for Further Research in Skwxwu7mesh 80 7.3.1. The Role of Verbal Reduplication 80 7.3.2. Types of Quantification 82  References  85  Appendix 1: Key to Skwxwu7mesh Orthography Appendix 2: Abbreviations  '.  88 89  v  ACKNOWLEDGEMENTS  This thesis would not have been possible without the help of many people in last two years. I would like to extend a big thank you to the Skwxwu7mesh community for the opportunity to work with their language. In particular, I would like to thank the elders who have given up their time to provide me with their patience and invaluable insights; each meeting was an educational experience (not simply linguistically). Thank you to Doris for filling every session with laughter. Thank you to Eva for hosting many meetings in her home, and for breaking it to me gently whenever I'm wrong. Thank you to Lawrence for sharing all those stories and for being a real sport when I ask all those odd sentences. And thank you to Yvonne for her warmth and kindness. I look forward to continuing to work with all of you. I also owe a great deal to the Skwxwu7mesh Nation Education Department for making this research possible and making our visits a lot of fun; thanks Valerie, Valerie, and Vanessa. In particular, I would like to thank Peter Jacobs for arranging our weekly meetings, sharing his intuitions, and all those last minute data checks. I owe so much to my committee, not only for making a linguist out of me, but for being such great friends (they all deserve medals for putting up with me). Henry Davis has played a large role in making this research possible; thanks for always having faith in me. Hamida Demirdache is responsible for getting me hooked on Skwxwu7mesh and cats; I hope France appreciates her as much as I do. I was fortunate to have Peter Jacobs, not only as a research associate, but as a committee member as well; thank you for always making sure there was a familiar face in the audience. Martina Wiltschko has taught me how to make a linguistic argument; thank you for providing me with a second home. The members of the UBC linguistics department have had an impact on my life here; in particular, I would like to thank Strang Burton for his semantic insights and for being a great dance partner. Rose-Marie Dechaine for her help with earlier versions of this analysis and for knowing that Friday night is not for work. Laura Downing for sharing my interest in spoken word events. Doug Pulleyblank for always having the answers, or knowing where to find them. Sue Urbanczyk (the urban-chick) for always being enthusiastic about my work. Many thanks to all those fellow students who made my transition to UBC (and Vancouver) an easy and enjoyable one and who have been very supportive (both academically and otherwise): in particular, Eleanor Blain, Lisa Chang, Tomio Hirose, Nicole Horseherder, Sandra Lai, Simo von Wolff, and especially Elizabeth Currie, who not only paved the way for this research, but who was always there for me. I was fortunate to meet the slew of students that arrived in my second year; in particular, Suzanne Gessner. Thank you to those who put up with me during WCCFL preparations (and lived to tell the story). A very special thank you goes to my friend Linda T. Watt; in addition to being a great linguist, she's been a central part of my life outside the department. Thanks for being there for both extremes). A big thanks to Carmen de Silva whose been a saint since day one. I'm glad you stuck around for one more year (well, did you really have a choice?) Thank you to my friends who simply love to be "linguist affiliates". To Shana Myara for being a great friend (your enthusiastic support has been invaluable). To Greg Pinel for being a big part of these last few months. I would like to acknowledge the support of my friends "back east": Leslie Young, Sarit Batner, the Batner family and the new Batner family (Yoram and Mamie). You guys were always just a phone call (or a 5 hour plane ride) away. Finally, to my family. Thank you to my brother Dan Bar-el for being my touchstone; your love and encouragement knows no bounds. To Dominique for her words of wisdom and unlimited generosity. This thesis could not have been completed without the support (financially and otherwise) of my parents Shirley and Avihu Bar-el. Your attempts to understand what exactly a linguist does are deeply appreciated. Thank you for always being proud of me, no matter what I chose to do. Fieldwork for this research was funded by SSHRCC grant #410-95-1519. Any errors in this thesis are my own.  vi  This thesis is dedicated to my brother Joe  vii  1. I N T R O D U C T I O N  The goal of this thesis is to present an analysis of verbal plurality and adverbial quantification in S k w x w u 7 m e s h . This thesis contributes to both the theoretical literature and the Salish literature. Skwxwu7mesh (Squamish) is a Coast Salish language spoken i n the Burrard Inlet and Howe Sound area around Vancouver,. British Columbia. There are no more than twenty native speakers left, the youngest in his late sixties. This thesis provides a detailed analysis of a phenomenon i n Skwxwu7mesh that has never been explored: the effect of the auxiliary wa on predicates from various aspectual classes in both non-quantified and quantified sentences, wa has been described as "a continuous-iterative clitic [that] refers to a process occupying a stretch of time, as having a duration. This duration may concern either a single act or the regular (iterated) performance of it" (Kuipers 1967:159) . Crucial to an understanding of this 1  phenomenon are the grammatical intuitions of fluent native speakers that are not always available from textual materials; this research thus contributes to the literature by documenting these judgments that may not be available in the near future. T w o central questions w i l l be addressed in this thesis. Firstly, what is the function o f the auxiliary wa in Skwxwu7mesh?. In other words, why is wa obligatorily present for certain interpretations of predicates and obligatorily absent for others; furthermore, what does wa do to a predicate to yield the various readings? Secondly, why is wa obligatory with adverbs of quantification? The answers to these questions have theoretical implications, implications for Skwxwu7mesh, as well as cross-linguistic implications for Salish. Pustejovsky (1991, 1995) uses event structures to represent the distinction between three primitive event types: states, processes, transitions (Bach 1986, Dowty 1979, Vendler 1967). The four-way aspectual class distinction can be analyzed i n Pustejovsky's model, where activities are analyzed as processes, accomplishments and achievements are analyzed as transitions and states are analyzed as states. O f these four classes, activities and states are, by  ^he status of Kuipers's use of the term clitic is not clear.  1  definition, atelic since they involve no culmination point or anticipated result. This thesis argues, however, that Skwxwu7mesh provides crucial evidence that all bare predicates (that is, predicates without wa) are telic, including activities and stage-level statives, but with the exception of individual-level predicates. What wa does to a predicate is cause it to become atelic; this atelicity is marked by continuous and/or habitual readings for the predicates o f the various classes. The analysis presented i n this thesis claims that wa targets either the head of a predicate's event structure (a subevent), or the event type denoted by the predicate for pluralization, leading to an atelic reading of the predicate. A s a consequence o f these claims, this analysis suggests that activities and stage-level states are not primitives universally. Adverbs of quantification (Q-adverbs), such as always, often, and sometimes, have been analyzed by Kratzer (1995) as unselective binders; that is, either spatiotemporal ('event') variables or individual variables can be bound by a Q-adverb. De Swart (1993, 1995), on the other hand, argues that Q-adverbs strictly quantify over events. It is argued i n this thesis that S k w x w u 7 m e s h provides crucial evidence that Q-adverbs quantify over events only. The evidence derives from the fact that the pluractional marker wa is obligatory with both stagelevel stative predicates and individual-level predicates when they combine with a Q-adverb. The analysis presented in this thesis claims that wa is the source of the plurality of events over which a Q-adverb quantifies. This thesis is organized as follows; Chapter 2 briefly outlines the basic morphology and syntax facts of Skwxwu7mesh that are relevant to an understanding of the data to be presented in the remainder of this thesis. Skwxwu7mesh is identified as a radical headmarking language with supporting data provided. The verbal complex is explored, with particular attention to the sources of temporal reference in Skwxwu7mesh as well as two auxiliaries that are central to the data throughout this thesis. A n outline of the distribution of overt D P s is provided, i n addition to a discussion o f what is, and what is not, encoded i n Salish D P s . Closing the chapter is a brief look at the basic Skwxwu7mesh word order pattern.  2  Chapter 3 outlines the questions to be addressed i n the remainder o f this thesis. This chapter presents data showing the distribution of wa among stage-level stative predicates and individual-level predicates. Differences in interpretations of predicates with and without wa are observed, leading to the first question: what is the function of wa i n Skwxwu7mesh? The relevant data illustrating the distribution o f wa i n quantified sentences (sentences with Q adverbs) is then presented, leading to the second question: why is wa obligatory with Q adverbs? This chapter concludes with the proposal that wa is a pluractional marker; the ways in which this proposal solves both questions is briefly outlined, along with the assumptions underlying the analysis to be presented. G i v e n the proposal mat wa is a pluractional marker, Chapter 4 then presents Lasersohn's (1995) analysis of plufattionaT markers. Pluractional markers are defined and their formal representation is given. Following Cusic (1981), Lasersohn incorporates the event level/phase level repetition parameter into his definition o f pluractional markers.  This  distinction is discussed, with appropriate examples provided. In Chapter 5 the first problem (the function of wa) is solved. The chapter begins with an outline of the terminology to be used throughout the chapter; in particular, the way in which the terms telic and atelic are used throughout the analysis are discussed. The chapter then explores the theoretical model that is adopted for the analysis; Pustejovsky's (1991, 1995) event structure model (state, process, transition) is presented, in addition to a discussion of the notion of event headedness. This model is then applied to four aspectual classes i n English: achievements, activities, accomplishments and states (Vendler 1967, D o w t y 1979). Next, the model is applied to the same four aspectual classes in Skwxu7mesh; for each of the classes, the event structure of the bare predicate and the event structure representing the pluralized form of the predicate are given. A s well, the type of repetition that arises from pluralization is noted for each of the classes. In Chapter 6, the second question (why wa is obligatory with Q-adverbs) is solved. The chapter begins with a discussion of Kratzer's analysis of Q-adverbs as unselective binders;  3  her view of the temporal/atemporal distinction and the stage-level/individual-level distinction is briefly outlined. After presenting both complex and simplex sentences within her framework, it is noted that Kratzer's analysis cannot account for the Skwxwu7mesh data. De Swarts's (1993, 1995) analysis of Q-adverbs as quantifiers stricdy over events is then discussed and exemplified with English data. The chapter ends with an application of the analysis to Skwxwu7mesh, incorporating also the proposal that wa is a pluractional marker. Chapter 7 concludes this thesis by summarizing the findings of the analysis.  This  chapter also outlines two theoretical and cross-linguistic implications that arise from this analysis: the implications for quantification and the implications for aspectual classes.  The  first section further discusses the notion of unselective binding in Salish; the second section discusses the notion of primitives in Pustejovsky's model of event structures. The chapter ends by addressing the issues that require further research in Skwxwu7mesh.  4  2.  SOME BASIC M O R P H O L O G Y  AND SYNTAX FACTS OF S K W X W U 7 M E S H  The purpose of this chapter is to outline some of the basic properties of the S k w x w u 7 m e s h language so that are relevant to an understanding of the Skwxwu7mesh data presented i n the remainder of this thesis . 2  The properties reviewed in this chapter are, for the most part,  properties o f Salish languages in general. Section 2.1. identifies Skwxwu7mesh as a radical headmarking language whereby pronominal marking on the head is obligatory and overt lexical D P s are optional. Section 2.2. explores the verbal complex, with particular attention to tense and auxiliaries i n Skwxwu7mesh; the subsection on tense identifies S k w x w u 7 m e s h as a language lacking obligatory morphologically encoded tense and outlines the sources of temporal reference available in the language. The subsection on auxiliaries focuses on the distribution of those that are relevant to the data presented in this thesis. Section 2.3. examines the syntactic distribution of overt D P s and the semantic information that they encode. Finally, Section 2.4. provides an outline of the major word order pattern of Skwxwu7mesh.  2.1. Radical Headmarking Skwxwu7mesh is a radical headmarking language; consequendy, pronominal marking on the head is obligatory and overt D P arguments are optional. Accordingly, on the one hand, the sentence in ( l a ) contains both pronominal marking on the head and lexical D P s and is wellformed; on the other hand, the sentence in (lb) contains no pronominal marking on the head, but retains the lexical DPs, and is ill-formed. Furthermore, the sentence in ( l c ) is well-formed, though the lexical D P s kwelhi slhdnay' 'the lady' and ta mixfllh 'the bear' are omitted. The fact that pronominal markings, such as subject and object agreement markers, are obligatory in Skwxwu7mesh is illustrated in ( l ) : 3  S e e Currie (1997), Demirdache etal. (1994), Jacobs (1992), Kuipers (1967) for further discussion of morphology, syntax and semantics of Skwxwu7mesh. 2  Unless otherwise noted, all data presented in this thesis is taken from (Bar-el 1997, 1998); this stems from original fieldwork with fluent native speakers of Skwxwu7mesh. 3  5  (la)  kw'ach-nexw-0-as  kw61hi slhanay ta 1  see-TRANS-30BJsg-3SUBsg DEM.F lady  mixalh  DETbear  'the lady saw the bear' (lb)  *kw'ach-nexw  kwdlhi slhanay' ta  see-TRANS  DEM.F lady  mixalh  DETbear  'the lady saw the bear' (lc)  kw'ach-nexw-0-as see-TRANS-30BJsg-3SUBsg 'he/she/it saw him/her/it.  Notice that, along with much of the Salish literature, I assume a 0-marked third person object marker that follows the transitivizer and precedes the third person subject agreement marker; this follows the usual order of pronominal affixes, as shown i n the template i n (2).  This  template provides the basic structure of the morphological word (excluding clitics):  (2)  PREHX-ROOT-ASPECT-LEXICAL.SUm (Davis 1997a)  In Skwxwu7mesh, third person pronominal markers are suffixes that attach to the end of the stem while first and second person pronorninals that precede the predicate complex and can take temporal clitics, as in (3b) and (3c) . The sentences in (3) provide examples of the 4  first and second person clitics:  (3a)  chen  lhki7slha  Valerie  l S U B s g k n o w DET.F Valerie 'I know Valerie'  Currie (1997) notes that when the first and second person pronorninals follow the predicate, they are interpreted as future with certain predicates.  4  6  (3b)  chexw-kw  flhen  2SUBsg-already eat 'You already ate' (Jacobs 1992:10) (3c)  chen-t  xaam  lSUBsg-PAST cry 'I cried'  2.2. The Verbal Complex The purpose of this section is to identify two issues associated with the verbal complex; in the first subsection, I address the topic of tense and provide examples of the sources of temporal reference in Skwxwu7mesh. In the second subsection, I discuss two particular auxiliaries that are of relevance to this thesis.  2.2.1. Tense A property of Skwxwu7mesh, and Salish languages in general, is the lack of obligatory morphologically encoded tense (see Matthewson 1996). Instead, temporal reference is derived from three sources; these sources are listed in (4):  (4)  a. Temporal adverbs, auxiliaries, clitics b. The aspectual class of the predicate c. Determiners  A temporal adverb, such as kwi chel'aklh 'yesterday' can cause a predicate to be translated in the past; this is shown in (5):  7  (5)  Temporal Adverbs chen flhen kwi chel'aklh lSUBsgeat  DET yesterday  'I ate yesterday' (Currie 1997:22)  Clitics, such as the past tense marker -t or the future marker  can cause a predicate to be  translated in the past or future, respectively. This is illustrated by the two examples in (6):  (6a)  Clitics chen-t  kwach-nexw ta push  lSUBsg-PAST  see-TPvANS  DET cat  'I saw the cat'  (6b)  chen-ek'  flhen  lSUBsg-FUTeat Tm going to eat' (Currie 1997:28)  An auxiliary, such as the local-directional clitic mi 'come/become', can cause a predicate to be translated in the past This is demonstrated in (7):  (7)  Auxiliaries chen  mi  nach-i  lSUBsg come change-WTRANS 'My expression changed' (Kuipers 1967:162)  8  A predicate from the aspectual class of achievements, such as tl'ik. 'arrive/got here', will be translated in the past, since it identifies an immediate change of state (see Chapter 5 for further discussion). This is illustrated in (8):  (8)  Aspectual Class of the Predicate  chen  1  tl'ik  lSUBsg arrive 'I arrived/got here'  Determiners have not fully been explored in Skwxwu7mesh; however, rjtemirdache (to appear) shows that an absent determiner in a closely related language, St'aYimcets (Lillooet Salish), can restrict the predication time of the matrix predicate and the noun in one sentence. This is illustrated in (9):  (9)  Determiners  secsec [ni  kel7£qsten-s-a  ti  US-a]  Strong D E T . A B S E N T chief-3sg.POSS-DET D E T US-DET  'The (present, not visible) chief of the US is a fool' 'The (past, not visible) chief of the US was a fool' *'The (past, not visible) chief of the US is a fool' (St'dt'imcets; Demirdache to appear)  2.2.2. Auxiliaries  The set of auxiliaries in Skwxwu7mesh, and Salish languages in general, include elements with aspectual, adverbial and quantificational force. With respect to their distribution, they precede the main predicate with which they form a monoclausal unit which takes a single set of  9  pronominal markers. Little is known about the syntax of auxiliaries i n Skwxwu7mesh; i n this subsection, I focus on two auxiliaries that are relevant to an understanding of the data presented in this thesis: wa, "indicating continuity of an action or process" (Kuipers 1967:377) , and 5  Ihik! 'always'.  2.2.2.1. wa When it can appear in a clause, this auxiliary precedes the predicate; the data collected thus far suggests that wa itself cannot bear tense markings, such as the past tense marker -t. Instead, na is often inserted and the past tense marker is suffixed to i t . This is demonstrated by the 6  sentences in (10) below : 7  (10a)  wa payim lha PA rest  slhanay  DET.F lady  'the lady is resting' (10b)  *wa-t  payim lha  PA-PAST rest (10c)  na-t  DET.F lady  wa payim lha  RL-PAST PA rest  slhanay  slhanay  DET.F lady  'the lady was resting'  Notice that wa is glossed as a pluractional marker (PA) i n each o f the above examples; this issue is central to this thesis and is discussed in further detail in Chapter 4 and onwards.  Note that the definition of this auxiliary will be refined in Chapter 4 of this thesis and onwards. This indicates that wa must occupy a separate positionfromna and the first/second person pronominlas since they can each bear the past tense marker, the future marker and other clitics. A n examination of the morpheme na is beyond the scope of this thesis; for further discussion, see Kuipers (1967).  5  6  7  10  2.2.2.2. lhik' The second auxiliary that is relevant to the issues presented in this thesis is the quantificational adverb Ihik.' 'always'. This adverb is consistently positioned at the left edge of the clause, before wa ; this is illustrated in (11): 8  (11)  lhik'  wa 7i7tut ta mixalh  always PA sleep DETbear 'the bear is always sleeping'  lhik! in Skwxwu7mesh can bear tense marking ; the past tense marker -t can be suffixed to the 9  adverb, as shown in (12b): (12a)  lhik'  wa p£yim  always PA rest 'she's resting all the time'  (12b) lhik'-t  wapayim  always-PAST PA rest 'she used to rest all the time'  2.3. Overt DPs Overt DPs in Skwxwu7mesh obligatorily take one of the proclitic determiners from the table in (13):  There is evidence that lhik' can behave as a predicate: (i) lhik' kwi-n-s wa ts'fts'ap' always DET-1POSS-NOM PA woik 'I am always working' lit: 'my working is all the time/always' However, further research is necessary to confirm the category and position of lhik'- (H. Davis (p.c.) notes that this is also the case in St'aYimcets (Lilloet Salish)). B y the logic of the suggestion in Footnote 5, lhik! must be occupying the same position as na and the first/second person pronominals, but a position other than the one occupied by wa. 8  9  11  (13) DEFINITE  NON-PRESENT  PRESENT  WEAK  INDEFTMTE  STRONG  PROXIMAL  DISTAL  WEAK  STRONG  MASCULINE  ta  ti  tay'  kwa  kwetsi  kwi  FEMININE  lha  tsi  alhi  kwelha  kwelhi  kwes  (Kuipers 1967:137)  10  G i v e n recent work on D P s i n other Salish languages (Matthewson 1996, Demirdache to appear, 1997 for St'aTimcets), the classification of S k w x w u 7 m e s h D P s may need to be revised . 11  When a sentence contains both an overt subject D P and an overt object D P , the subject D P precedes the object D P . This is illustrated in (14):  (14)  kw'ach-nexw-0-as  kw61hi slhanay' ta  see-TRANS-30BJsg-3SUBsgDEM.F lady  mixalh  DETbear  'the lady saw the bear'  W h e n a sentence contains only one overt D P , but the sentence contains a transitive predicate, that overt D P is interpreted as the object; this is known as the "one-nominal interpretation" (Gerdts 1988, for Halkomelem). Gerdts's generalization, using the notion absolutive, is given in (15):  1 0  T h e term 'masculine' is used in the chart in (13) to indicate what Kuipers refers to as 'plain'.  ^See Currie (1997) for further discussion on DPs in Skwxwu7mesh.  12  (15)  One-nominal interpretation In the absence of marking for other persons, a single third person nominal i$ interpreted as the absolutive. (Gerdts 1988:59)  This generalization holds for Skwxwu7mesh as well and is illustrated in (16):  (16)  na kwach-nexw-0-as  ta  slhen-slhanay'  RL see-TRANS-30BJsg-3SUBsg DET RED-lady 'he met some women' *'the women met him'  Matthewson (1996) proposes that Salish determiners exhibit four properties; these are listed i n (17):  (17)  a. Salish determiners do not encode definiteness. b. Salish determiners do not encode specificity. c. There are no quanuficational determiners in Salish. d. Salish determiners encode 'assertion of existence'. (Matthewson 1996:20)  Although the spatiotemporal status of D P s i n S k w x w u 7 m e s h is not completely understood, Demirdache (1997) has argued that, i n St'aTimcets (Lillooet Salish), determiners may encode spatiotemporal distinctions; furthermore, she states that the temporal interpretation of an N P can determine the temporal interpretation of the main predicate of a sentence (see (9) above).  13  2.4. Word Order S k w x w u 7 m e s h word order, like that of other Salish languages, is predominantly predicate i n i t i a l . The sentences in (18) below demonstrate that, regardless of its class, the 12  predicate w i l l occur clause initially: (18a)  flhenta eat  mixalh  D E T bear  'the bear is eating' (18b)  hiyita  mixalh  big D E T bear 'the bear is big' (18c)  mixalh kweci bear  DEM.M  'that is a bear/there is a bear'  O t h e r word orders have been documented; however, the issue of word order is not directly related to the discussion at hand. For further discussion, see Currie (1997), Jacobs (1992) and Kuipers (1967). 12  14  3. T H E S K W X , W U 7 M E S H  PARADOX  The goal o f this chapter is to present three sets of S k w x w u 7 m e s h data w h i c h , upon comparison, raise two theoretical questions relating to predicate classes and adverbs o f quantification. Recall that wa has previously been described as "a continuous-iterative clitic [that] refers to a process occupying a stretch of time, as having a duration. This duration may concern either a single act or the regular (iterated) performance of it". (Kuipers 1967:159). In this chapter, I present preliminary data exemplifying the effect of wa on stage-level stative predicates and individual-level predicates. To begin, Section 3.1. looks at the behaviour of wa with stage-level stative predicates; next, the behaviour of wa with individual-level predicates is investigated i n Section 3.2.. Finally, the behaviour of wa i n quantified sentences containing both stage-level stative predicates and individual-level predicates is explored i n Section 3.3.. This chapter concludes with a summary o f the two problems that arise from the S^wxwu7mesh data, as well as providing a brief overview of the proposal that solves both problems.  3.1. Stage-level Stative Predicates A stage-level predicate expresses a transitory property; it holds true o f an individual (or set of individuals) at a particular time and/or place. A stative predicate is described as a predicate that persists for a duration of time, but is not, itself, an action (Mourelatos 1978, Vendler 1967); Thus, a stage-level stative predicate denotes a transitory property that involves no dynamics  (i.e. hungry, tired, angry). In Skwxwu7mesh, a stage-level stative predicate is always introduced by the auxiliary wa ; 13  W h e n prompted with the English sentence 'the bear is hungry', a speaker volunteers  the stage-level stative predicate ('hungry') with wa; on the other hand, when prompted with the same sentence i n Skwxwu7mesh, kw'ay' ta mixjalh (excuding wa), a speaker suggests that the sentence is ill-formed under the stage-level stative reading of the predicate 'hungry'. This is illustrated in (19):  1 3  F o r further discussion on wa in Skwxwu7mesh and other Salish languages, see Davis (1996,1997).  15  (19a)  wa kw'ay' ta  mixalh  PA hungry DET bear 'the bear is hungry'  (19b)  kw'ay' ta  mixalh  hungry DET bear *'the bear is hungry' / ' t h e bear got hungry'  Under the stage-level stative interpretation, a speaker may accept the sentence i n (19b), but w i l l acknowledge the fact that it sounds as though there is something missing; when asked to repeat the sentence, speakers consistently insert wa. Note that the sentence i n (19b) is well-formed with a change-of-state reading; however, once the sentence is given the  change-of-state  reading, the predicate is clearly no longer being interpreted as a stative predicate. This contrast is explored in further detail in Chapter 5.  3.2. Individual-level predicates A n individual-level predicate (i.e. big, strong, tall) denotes a permanent property; it is attributed to an individual only once, but holds of that individual permanently.  A n individual-level  predicate i n Skwxwu7mesh cannot be introduced by wa. When prompted with the English sentence 'the bear is big', a speaker volunteers the Skwxwu7mesh equivalent of the predicate, hiyi ('big') without wa; conversely, when prompted with the same Skwxwu7mesh sentence, (this time adding wa) wa hiyi ta mixalh, a speaker suggests that the sentence is ill-formed under the individual-level reading of the predicate hiyi ('big'). This generalization is illustrated by the examples in ( 2 0 ) : 14  Notice that the bare form of the individual-level predicate is missing the change-of-state reading that is given to stage-level statives when they lack wa. 14  16  (20a) hiyi ta  mixalh  big DET bear 'the bear is big'  (20b) ?wahiyita  mixalh  PA big DET bear *'the bear is big'  In the same way that stage-level predicates are given a different reading when they occur without wa, individual-level predicates are observed to yield stage-level interpretations when they occur with wa. This contrast in readings is illustrated by the sentences in (21):  (21a) chen  iy7im  lSUBsg strong 'I am strong' DW 30-07-98  (21b) chen  ;  waiy7im lh-7an  wa ts'ets'kw'a-t-sut  lSUBsg PA strong when-lSUBsg PA run-TRANS-REFL 'I am strong when I'm running' DW 30-07-98  Notice, however, that to obtain the stage-level interpretation of the individual-level predicate iy7im 'strong', context is required (i.e. the remainder of the sentence ...lh7an wa ts'ets'kw'atsut '...when I am running'); Again, the contrast in interpretations is examined further in Chapter 5 of this thesis.  17  3.3. Quantified Sentences The third set of data that raise questions about the function of wa are sentences containing the auxiliary Ihif 'always'; in Skwxwu7mesh, wa is obligatory when a Q-adverb is present, regardless of whether the predicate is a stage-level stative predicate or an individual-level predicate. Thus, when Ihif is added to a sentence containing the stage-level predicate kw'ay' 'hungry', wa is obligatorily present; furthermore, when Ihif is added to a sentence containing the individual-level predicate hiyi 'big', wa, again, is obligatorily present. This generalization is illustrated in (22) below:  (22a)  lhik'  *(wa) kw'ay'ta  always PA  mixalh  hungry DET bear  'the bear is always hungry'  (22b)  lhik'  *(wa) hiyi ta  always PA  mex-mixalh  big DET RED-bear  .  'these bears are always big'  Note that the speaker suggests that one might use the sentence i n (22b) i f you were talking about "a group of bears i n this area". Moreover, notice that the sentence i n (22b) has a plural D P as its subject; there is an additional requirement in quantified sentences with individual-level predicates that the subject be a plural D P . A quantified sentence containing a singular D P as its subject is ill-formed; this is illustrated in (23):  (23)  *lhik'  wa hiyi ta  1  mixalh  always PA big DETbear  18  The requirement for a plural D P w i l l be discussed in further detail in Chapter 6.  3.4. Two Problems G i v e n the S k w x w u 7 m e s h facts presented in the first three sections of this chapter, two problems arise; these problems have theoretical implications with respect to predicate classes and adverbial quantification. The first problem, in general terms, addresses the question of the function of wa in Skwxwu7mesh. This problem can be separated into two smaller issues; Firstly, it is necessary to explain why wa is obligatorily present with some predicates under certain readings and obligatorily absent with other predicates under other readings. Secondly, we need to address why wa yields different readings when it combines with different classes of predicates. The second problem raised by the data above is a paradox. It has been shown that wa is incompatible with an individual-level reading i n a non-quantified sentence (cf. (21)), obligatory with a stage-level stative reading in a non-quantified sentence (cf. (20)), and also obligatory when a Q-adverb combines with either a stage-level stative predicate or an individual-level predicate (cf. (22)). This generalization is summarized i n (24) below:  (24) STAGE-LEVEL  INDIVIDUAL-  STATTVE  LEVEL  READINGS  READINGS  *(wa)  *wa  *(wa)  *(wa)  NON-  QUAN1MED SENTENCES QUANTIFIED SENTENCES  Thus, the problem to address is why an individual-level predicate in a quantified sentence must be introduced by wa, which has been shown to yield ungrammatically when it combines with an individual-level predicate in a non-quantified sentence.  19  T o answer both of these questions, I propose that wa is a pluractional  marker  (Lasersohn 1995) that pluralizes the head o f a predicate's event structure or the event type denoted by the predicate. Since the event structures of predicates from different aspectual classes vary, it is expected that predicates introduced by wa w i l l vary in interpretation according to the type of event structure wa pluralizes. T o explain the second problem relating to Q-adverbs, I assume an analysis whereby quantification is strictly over events (De Swart 1993, 1996). Thus, the solution to the second problem (why wa is obligatory with Q-adverbs) is divided into three parts; firstly, I assume that Q-adverbs quantify over events. Following De Swart, for quantification to take place, a plurality of events is required. Finally, the proposal that wa is a pluractional marker explains why wa is obligatory with Q-adverbs because it is wa that gives rise to the plurality o f events over which the Q-adverb can quantify.  20  4. P L U R A C T I O N A L  MARKERS  The goal of this Chapter is to define pluractional markers (Lasersohn 1995) and outline the parameter that differentiates the readings that arise from the use of pluractional markers. Section 4.1. provides a basic representation of pluractional markers. Section 4.2. then outlines the distinction between event level repetition and phase level repetition and incorporates this distinction into the definition of pluractional markers.  4.1. Defining Pluractional Markers Pluractional markers are morphemes that "attach to the verb to indicate a multiplicity of actions, whether i n v o l v i n g multiple participants, times or locations" (Lasersohn  1995:240).  Pluractional markers "do not reflect the plurality of a verb's arguments so much as plurality of the verb itself:  the verb is understood to represent the occurrence of multiple events"  (Lasersohn 1995:241). A pluractional verb (a verb + pluractional marker) w i l l hold true of a group o f events i f and only i f its corresponding "singular" verb (its bare form) holds true of each individual event i n the group. This basic meaning of pluractional markers is given by the representation i n (25):  (25)  V-PA(X)  <=> VeeX[V(e)]  & card(X) > n (Lasersohn 1995:242)  where V=verb PA=pluractional marker X=ranges over sets of events e=event  Notice that since the number of events implied by a pluractional marker is pragmatically fixed, the condition on the cardinality of events is specified as n in the definition but must be stated  21  simply as no less than 2. Depending on the type of reading that arises from the attachment of a pluractional marker to a verb, further conditions may be added to the representation in (25).  4.2. Event/Phase Level Repetition In examining the readings exhibited cross-linguistically by pluractional markers, Cusic (1981) suggests that variation i n meaning results from a two-setting parameter between phase repetition and event/occasion repetition which gives rise to the difference between a "repetitive" action and a "repeated" action. For the remainder of this thesis, I refer to this distinction as the phase/event level repetition distinction. Phase repetition entails multiple events, which may be of a different type, which sum up to form a single token of the event type corresponding to the verb; thus i n (26) below, the nibbling event consists of multiple events of small biting and a small biting does not itself constitute a 'nibble':  (26)  The mouse nibbled the cheese. (Lasersohn 1995:243-4)  Notice that phases of an event may be of the same type as the event type of the predicate, only smaller; i n other words, the hopping event in (27) is made up of a series of hops, giving rise again to phase repetition where a single hop is distinguished from the larger hopping event. Consider the sentence in (27):  (27)  The kangaroo hopped across the field  Event repetition entails multiple events o f the type denoted by the verb; thus, the larger nibbling event i n (28) is itself repeated, not just the smaller phases internal to a single nibbling:  22  (28)  The mouse nibbled the cheese again and again on Thursday.  W i t h repsect to the sentence in (29), the larger hopping event (a series of hops across the field) is being repeated (not simply the smaller hops internal to a single hopping event), yielding event-level repetition. Consider the sentence in (29):  (29)  The kangaroo hopped across the field (over and over) on Friday.  To account for the difference between phase level and event level repetition, Lasersohn replaces V on the right side of the biconditional in the definition for pluractional markers by a free variable ranging over properties of events (P). This is illustrated in (30):  (30)  V-PA(X) <=> WeeX[P(e)] & card(X) > n (Lasersohn 1995:255) where V=verb PA=pluractional marker X=ranges over sets of events e=event P=free variable ranging over properties of events  In other words, a pluractional verb ( V - P A ) w i l l hold true of a group of events ( X ) i f and only i f all events (e) i n the set of events (X) holds true of a single event with a certain property (P) and there are no less than two events in the set of events ( X ) . H o w P is identified yields the distinction between phase level and event level repetition; thus, to obtain the event level repetition the free variable ranging over properties of events is equated with the verb (P = V). For phase level repetition, the identity of the free variable P is determined lexically; this is to  23  say that the phases that make up an event are different for each verb (i.e. 'hop' for the verb hopping' and 'small bite' for the verb 'nibbling'). The proposal of this thesis is that wa is a pluractional marker that pluralizes the head of an event structure.or the event type denoted by the predicate. Thus far, I have focused on stage-level stative predicates and individual-level predicates. In the next chapter, I examine the behaviour of wa with predicates from other aspectual classes; this, i n turn, w i l l lead to an explanation of why wa is necessary with Q-adverbs in Skwxwu7mesh, which I later provide in Chapter 6.  24  5. S O L V I N G T H E F I R S T  PROBLEM  This chapter addresses the first problem (the function of wa) under the proposal that wa is a pluractional marker.  I examine the distribution o f wa across four aspectual classes:  achievements, accomplishments, activities and states (Dowty 1979, Vendler 1967). I claim that bare predicates in Skwxwu7mesh are always telic, with the single exception o f individuallevel predicates; predicates are made atelic by the addition of the pluractional marker wa. While achievements and accomplishments are already understood as telic, this claim implies that stage-level stative predicates and activities are derived classes. Section 5.2. outlines the theoretical model that I adopt for this thesis.  I follow  Pustejovsky's (1991, 1995) analysis of event structures for three primitive aspectual classes: states, processes and transitions; I adopt his notion of event headedness to account for the target of pluralization. In Section 5.3.1 apply Pustejovsky's model to English data, which lays the ground for the analysis of Skwxwu7mesh; event structure representations are presented for each o f the four aspectual classes: achievements, activities, accomplishments, states. Section 5.4. then takes the Pustejovsky model and applies it to S k w x w u 7 m e s h , noting certain significant differences from English, and taking into particular consideration the event/phase level repetition parameter. Section 5.5. summarizes the findings of this chapter. I begin with an overview o f the terminology adopted for this analysis, focusing on the telic/atelic distinction.  5.1. Terminology: Telic/Ateiic D a h l (1981) notes that the intuitive semantic difference between a verb phrase that contains a reference to a 'terminal point' in which the action comes to an end and a verb phrase that does not contain a reference to such a point, has been given many labels in the literature . Drawing 15  on Garey's (1957) terms telic and atelic, for example, Brinton (1988) uses these terms to refer to "situations which include or do not include a goal" (1988:25). Noting that there are various  1 5  F o r a list of these terms, see Dahl (1980:80).  25  problems with the terminology, i n this thesis, I use the terms telic and atelic to refer to situations that have or lack a culmination point, respectively. W i t h respect to the translations offered for Skwxwu7mesh sentences, an atelic predicate can involve either a continuation of a single event or the iteration of events.  5.2. Theoretical Model: Pustejovsky (1991, 1995) The purpose of this section is to provide an overview of the theoretical model that I adopt to account for the Skwxwu7mesh data. Pustejovsky proposes a configurational theory of event structure that takes into consideration the aspectual properties of verbs; given that this thesis explores the distribution of wa across aspectual classes, Pustejovsky's model provides a structural means to distinguish between the aspectual classes. This section focuses on two central issues: the representation o f event structure and the notion o f event headedness within an event structure representation.  5.2.1. Event Structure Pustejovsky (1991) suggests that event types fall into three classes: states, processes and transitions; he defines a state (S) as a single event which is evaluated relative to no other event (i.e. be sick, love). The event structure of a state is given i n (31):  (31)  State  S e  A process (P) is defined as a sequence of events identifying the same semantic expression (i.e. run, push); this event structure is given in (32): (32)  Process  P  26  A transition (T) is defined as an event identifying a semantic expression which is evaluated relative to its opposition (i.e. open, build); this is shown in (33):  (33) Transition  T -.Ei  ' Ei  Transitions can have different subeventual structures; an accomplishment may consist of a process as a first subevent with a final resulting state. This is illustrated in (34): (34)  T  P ei  / \ e  S n  A n achievement may consist of an initial -<S and a final S; this is illustrated i n (35): (35)  5.2.2 Event  T  Headedness  For Pustejovsky, event headedness provides a way of indicating a type of foregrounding of event arguments; he claims that the event structure of a predicate provides a configuration where events are distinguished by relative prominence i n addition to temporal precedence. Although it is a property of all event types, Pustejovsky states that headedness "acts to distinguish the transitions, specifying what part of the matrix event is being focused by the lexical item in question" (1995:72). The head of the transition representing an accomplishment is the initial process ( E i ) ; this is illustrated in (36):  27  (36)  T Ei  E2 where Ei =event head  The event structure of an achievement differs from an accomplishment in that it is the final resulting state (E2) that is the event head; this is illustrated in (37): (37)  T Ei  E2 where E2=event head  Thus, the head is the "most prominent subevent in the event structure of a predicate which contributes to the 'focus' of the interpretation" (Pustejovsky 1995:72).  5.3. The Model Applied: English The goal of this section is to show how Pustejovsky's model applies to the four principal aspectual classes in English: achievements, activities, accomplishments, states. For each aspectual class, a brief description of the properties associated with a predicate from that class is provided, in addition to some examples. The event structures for each of the classes will act as a means of comparison in the upcoming section focusing on Skwxwu7mesh.  5.3.1. Achievements Achievement predicates (i.e. win, arrive, find) "capture the inception or the the climax of an act...they can be indefinitely placed within a temporal stretch, but they cannot in themselves occur over or throughout a temporal stretch" (Mourelatos 1978:416). In English, achievement predicates are uttered in the past; this is illustrated in (38):  28  (38a)  I won ten dollars  (38b)  I arrived  (38c)  I found a book  Bennett and Partee (1978) suggest that the reason for the past tense reading o f achievement predicates is due to the fact that achievements are only true at moments; this suggests that an achievement cannot be truthfully uttered in the present tense because once an achievement i s . true, it must already be in the past. Extending Bennet and Partee's account for past tense readings, we can explain why achievement predicates are odd as present progressives; this is to say that an immediate change of state is difficult to stretch in time. This oddity is illustrated by the sentences i n (39):  (39a)  ?I am winning ten dollars  (39b)  ?I am arriving  (39c)  ?I am finding a book  Note that the sentence / am arriving in (39b) is well-formed under a future interpretation; the sentence is odd, however, under the present interpretation. Contrary to the view that the progressive can only apply to predicates true of events which take time, Kearns (1991) claims that the progressive can be applied to predicates such as the ones i n the sentences in (39) and are well-formed i f uttered with exact timing. This is to say that i f you say "I am winning" at the exact moment that you are undergoing the change-of state, the sentence is well-formed; thus, she states that although it is difficult to utter truly, the progressive form of an achievement predicate is not ill-formed. In Pustejovsky's model, an achievement predicate such as arrive would be represented as a transition from - i E to E (where E is a variable for any event type); this is illustrated in (40):  29  (40)  T -IE  E  5.3.2. Activitites A n activity predicate is homogeneous in that "any part of the process is of the same nature as the whole" (Vendler 1967:101); i n Mourelatos's terms, "the time stretch o f activities is inherently indefinite in that they involve no culmination or anticipated result" (1978:416). In English, activity predicates are often given i n the progressive; this is illustrated i n (41):  (41a)  I am running  (41b)  I am singing  (41c)  I am working  Thus, at any stretch of time within the period of my running, singing or working, it is true that / am running, singing or working; this is what Dowty (1986) refers to as the "subinterval property" o f events. The same facts are observed when these predicates are translated into the past; activity sentences translated in the past progressive are provided in (42):  (42a)  I was running  (42b)  I was singing  (42c)  I was working  Pustejovsky represents a predicate from this class as a process (note that we have seen this structure as the initial subevent of an accomplishment predicate; here, however, the process occurs without a culminating state or transition). Consider the representation i n (43):  30  (43)  P ei  e  n  5.3.3. Accomplishments A n accomplishment predicate is not homogenous in that we can refer to the whole of a time segment without referring to a single moment; they "have duration intrinsically" (Mourelatos 1978:416) . 16  The sentences i n (44) provide some examples of accomplishment predicates i n English given i n the past : 17  (44a)  I built a house  (44b)  I ate an apple  (44c)  I tore the cloth  Thus, i n (44a) the activity of building resulted i n the the state of a house that is built, i n (44b) the activity of eating resulted i n an eaten apple and in (44c) the activity of tearing resulted i n a torn cloth. In their past tense forms, an accomplishment predicate is telic. In its present-progressive form, an accomplishment predicate yields a reading whereby the resulting state is anticipated, but not yet completed. The sentences i n (45) are examples of accomplishment predicates in English given i n the present-progressive:  N o t e that it has been claimed that accomplishment predicates and achivement predicates are not distinguished by event structure (that is they both consist of an initial process and a reulting state); instead, they are distinguished by the duration of the transition: the transition for achievement predicates occurs instantaniously while the transition for accomplishment predicates can take time. N o t e that there is no simple present in English; consequently, the sentences are uttered in the past (see Chapter 5 for further discussion). 16  17  31  (45a)  I am building a house  (45b)  I am eating an apple  (45c)  I am tearing the cloth  In Pustejovsky's model, an accomplishment predicate is a transition, T, consisting of an initial process, P, and final resulting state, S or a transition, T. This is illustrated in (46): (46)  T  P  y\  / \  ei  S/T  ,.e  n  Notice that whether the initial subevent in the structure of an accomplishment is a state or a transition is not crucial to the analysis presented in this thesis; thus, I note this issue, but ignore it for the remainder of the analysis.  5.3.4. States State predicates, i n English, are those which "may endure or persist over stretches o f time...they cannot be qualified as actions...they do not have progressive forms...they involve no dynamics...Though it may arise or be acquired as a result of change, the state itself does not constitute a change" (Mourelatos 1978:416). State predicates can be divided into two classes: stage-level states and individual-level states. R e c a l l that an individual-level predicate denotes a permanent property; English' examples are illustrated i n (47):  (47 a)  I am big  (47b)  I am tall  (47c)  I am strong  32  A stage-level predicate, on the other hand, denotes a transitory property; the sentences i n (48) each contain a stage-level predicate of the aspectual class "state":  (48a)  I am angry  (48b)  I am hungry  (48c)  I am tired  Pustejovsky (1995) claims that the distinction between a stage-level state and an individual-level state is the fact that a stage-level state has "an inherent reference to that factor that brings this state about...the 'coming into being' factor, the agentive role" (1995:225). Thus, an individual-level predicate is represented as a single event; this is shown in (49):  (49)  tall  S  I  e  The event structure of a stage-level stative predicate, on the other hand, must make reference to a default intial process subevent that is there only when the sentence overtly marks that it is there; i n other words, in the sentence John is angry from reading the newspaper, the factor that brought the state of being angry about is overtly specified (reading the newspaper).  This  would be represented as (50):  (50)  T P  S  However, for the simpler sentence John is angry, Pustejovsky notes that this initial process is a default event. This problem w i l l be further discussed in Section 5.4.4.  33  5.4. The Model Applied: Sj£W2Lwu7mesh N o w that Pustejovsky's model has been exemplified i n English, the goal of this section is to take this event structure model and apply it to S k w x w u 7 m e s h . The section proceeds as follows; i n each subsection, I provide an event structure representation for an aspectual class and show how its interpretation is modified by wa. I examine the phase level/event level distinction as it is applied to each aspectual class. Underlying my analysis is the proposal that wa is a pluractional marker; I show how this claim leads to a unified explanation for its effect on the V e n d l e r - D o w t y aspectual classes.  I argue that bare stage-level predicates i n  Skwxwu7mesh are telic and that predicates are made atelic by the addition o f wa. The first subsection looks at predicates from the aspectual class of achievements.  5.4.1. Achievements In Skwxwu7mesh, bare achievement predicates are translated i n the past, yielding the same reading as achievement predicates in English. This is illustrated i n (51):  (51a)  tl'ik  ta  John  arrive D E T John 'John got here'/'John arrived' LB 21-07-98  (51b)  chen  tl'exwenk  lSUBsg w i n 'I win/I w o n ' 1 8  1 8  T h e status of the gloss "I win" requires further investigation.  34  (51c)  chen  wf/xwem  lSUBsg fall 'I fell [from above]' LB 25-08-98  The speaker suggests that the event in (51a) must be completed in that it "can't mean he hasn't arrived yet". Recall that Bennet and Partee (1978) account for the past tense reading of an achievement by claiming that it can only true at a moment and once it is true, it is in the past; this description of an achievement predicate holds for Skwxwu7mesh as well. A bare achievement predicate is telic; it is translated into the English simple past in order to convey the fact that a cuhnination point has been reached. I assume an event structure for the bare form of an achievement predicate in Skwxwu7mesh that parallels English; an achievement predicate is represented by a transition in which an event type (E) is evaluated relative to its opposition. This event structure is given in (52): (52)  tl'ik. arrive  T -,E  E  When introduced by wa, an achievement predicate is translated as a repeated or a habitual action. This is illustrated by the data in (53):  (53a)  ?chen wa tl'ik lSUBsg PA arrive.  35  (53b)  chen  wa tl'exwenk  lSUBsgPA win 7 am a winner/Twin all the time'  (53c)  chen  wa wfTxwem  lSUBsg PA fall '[I'm] making a habit of [falling]' L B 25-08-98  A n achievement predicate with wa in Skwxwu7mesh is atelic since there is no culmination point specified for the habitual actions. The habitual reading follows from the fact that an achievement predicate cannot be made atelic by stretching the time of a single transition; instead, wa causes the predicate to become atelic by increasing the number of times that the transition occurs without specifying a culmination point. Note that the speaker's comments in reference to these sentences further emphasize the atelic interpretations of the predicates. Probably due to pragmatic reasons, the speaker neither confirms nor denies the grammaticality of (53a); however, his suggestion that the sentence "would have to mean 'more than once'" indicates that the habitual reading is the expected one with achievement predicates introduced by wa as opposed to the interpretation where a single event of arriving is not yet culminated. The gloss given for (53b), 'I am a winneryT win all the time', also suggests a habitutal action in that the event of winning is being repeated. For (53c), the speaker states that the sentence means "you're making a habit of [falling]".  19  Since a habitual reading arises when wa is added to an achievement, I claim that wa pluralizes the event type denoted by the predicate, the transition (T); thus the representation of a  N o t e that this motivates a distinction between achievements and accomplishments in Skwxwu7mesh; accomplishments allow pluractional modification of a subevent (see Section 5.4.3. for discussion) whereas achievements do not. 19  36  pluralized achievement predicate consists of an unspecified number of culminated transitions. This event structure is provided in (54): (54)  wa tl'ik PA arrive  T-PA T„ E -iE  E  The representation i n (54) denotes both telic and atelic events; this is to say that although the pluralized T indicates that the transition is atelic (here, a habitual event), this atelic event is comprised of a number of telic transitions. Thus, wa does not necessarily cause each portion of an event to be atelic; wa causes the target of pluralization to be atelic. When introduced by wa, an achievement denotes multiple events of the type denoted by the predicate itself; the plurality comes from the transition (T) from - i E to E being repeated over and over. W i t h respect to Cusic's parameter, a pluralized achievement predicate yields event level repetition as opposed to phase level repetition. Recall that event repetition is represented as (55), where the free variable ranging over properties of events (P) is identical to the verb itself (V):  (55)  V-PA(X) <=> VeeX[P(e)]  & card(X) > n  where P=V  Thus, equating P with the class of achievement predicates demonstrates that an achievement yields event level repetition when it is pluralized. Predicates from other aspectual classes allow for both event level repetition and phase level repetition; this is shown i n the following subsections.  37  5.4.2. Activities In Skwxwu7mesh an activity predicate is translated in the present tense and given a continuous reading for its form with wa and without wa; thus, the translation 'the woman is singing', which suggests an incomplete event, is provided for both the sentence with wa in (56a) and without wa in (56b:  (56a) wa lulum ta PA sing  slhanay'  DET woman  'the woman is singing'  (56b) lulum ta sing  slhanay'  DET woman  'the woman is singing'  However, the sentences in (56a) and (56b) are not equivalent. We can distinguish activity predicates with and withut wa by the addition of the past tense enclitic, -t; thus, a bare activity predicate with -t is translated in the simple past. On the other hand, an activity predicate with -t introduced by wa is given a past continuous reading. This is illustrated in (57) where the activity predicate x_aam 'cry' in (57b), with a past tense marker but without wa is translated as 'I cried' and not 'I was crying':  (57a) chen-t  wa xaam  lSUBsg-PAST PA cry 'I was crying' LB 25-08-98  38  (57b)  chen-t  xaam  lSUBsg-PASTcry "I cried' L B 25-08-98  Before presenting the event structure of activity predicates, it is first necessary to understand how the interpretations of these two forms differ. The reason for the apparent equivalence of the two forms can be attributed to the fact that there is no "real" present i n English; this is to say that English only has the presentprogressive, as exemplified in (58a) and the generic present, as exemplified i n (58b):  (58a)  I am singing  (58b)  I sing  Thus, sing i n (58a) is an action occuring now (at utterance time) that has not yet culminated; i n (58b), on the other hand, sing is understood as a habitual action, not (necessarily) occurring at utterance time. Consequently, the fact that the Skwxwu7mesh sentences i n (56a) and (56b) above are often glossed the same way may be attributed to the fact that the only possible English translation for either sentence is the present progressive. I propose that the distinction between the two forms of the activity predicate is related to inherent telicity; that is, the fundamental difference between the two is that the bare form (the predicate without wa) is telic; the addition of wa causes the predicate to become atelic. This distinction does not come out in the present in English because the event time, now, overlaps with utterance time; this is to say that the only way in English to express the event time now is in the present progressive.  39  I have shown that the two forms o f activity predicates i n S k w x w u 7 m e s h can be distinguished with the addition of the past tense marker; in the remainder o f this section I use felicity tests to further demonstrate this distinction. I draw on data from French as a parallel.  5.4.2.1. A Comparison with French French utilizes both a simple, morphologically encoded, present and a complex, paraphrastic, present marked by the expression en train de; it is the complex present that parallels what I am referring to as the "real" present that is lacking i n English.  T h e complex present is  characterized by the fact that the event it refers to must be occurring now, and cannot be stretched into the past. The sentences in (59a) and (59b) are examples o f the simple present and complex present, respectively:  (59a)  L a femme chante 'The woman is singing'  (59b)  L a femme est en train de chanter 'The woman is singing (as we speak)'  Note that the French speaker is faced with the same difficulty as the Skw&wu7mesh speaker in providing distinct English glosses for the two sentences in their respective languages. Notice that, i n French, adding the durational temporal modifier depuis deux heures 'since two o'clock' to the predicate in the simple present yields a grammatical sentence, while adding it to the predicate i n the complex present yields an ungrammatical sentence; this is illustrated in (60):  (60a)  Je chante depuis deux heures I've been singing since two o'clock'  40  (60b)  * Je suis en train de chanter depuis deux heures  Thus, while the predicate in the simple present can be stretched in time, the predicate in the complex present cannot. Turning to the data from Skwxwu7mesh, the proposal that wa causes a predicate to be atelic predicts that the addition of a durational temporal modifier to an activity predicate introduced by wa will yield a grammatical sentence; the sentence in (61) demonstrates that this is the case : 20  (61)  chen(-t)  wa lulumti-na7kwi chelaklh  lSUBsg(-PAST) PA sing  from DET yesterday  I've been singing since yesterday' LB 21-07-98  Note that the speaker specifies that at speechtime"you're still singing". Compare this with the simple present form of 'sing' in French Je chante depuis deux heures. We expect, then, that adding the durational temporal modifier to a sentence containing an activity predicate without wa, will yield an ill-formed sentence since it is not possible to stretch the event time of the telic predicate to indicate that the event began in the past; this is illustrated by the ungrammatical sentence in (62):  (62)  *chen lulumti-na7kwi chelaklh lSUBsg sing  from DET yesterday  LB 21-07-98  2 0  T h e glosses of the two morphemes in ti-na7 is not certain and has thus been glossed simply as 'from'.  41  Note that the speaker emphasizes the fact that the event time of the predicate is now when he states that the sentence in (62) is a "contradiction" and that you seem to be saying that "you're singing now, since yesterday". Next let us consider what happens in French when a temporal modifier specifying the exact point at which the activity is taking place is added to sentences containing the simple present and complex present forms of the predicate. Introducing the phrase a deux heures 'at 2 o'clock' to a sentence containing the simple present form of an activity predicate yields a wellformed sentence translated in the future; this is illustrated in (53):  (63)  Je chante a deux heures 'I'm singing at 2 o'clock'/! w i l l be singing at 2 o'clock'  Thus, the sentence is understood to mean that the singing event is not taking place at speech time, but w i l l be taking place at some specified time, two o'clock, in the future. A d d i n g the same temporal modifier to the complex present form of the predicate is i l l formed, since the predicate can only be interpreted as occurring at speech time; this is illustrated by the sentence i n (64):  (64)  * Je suis en train de chanter a deux heures  In addition to lacking an interpretation in the future, unlike its simple present counterpart, the sentence in (64) cannot be interpreted in the past either; this is to say that 'I sang at 2 o'clock' is not a possible translation, further emphasizing that the event time of a complex present predicate is now. Continuing with the predictions for Skwxwu7mesh, and using the data from French as a means of comparison, we expect that the addition of a temporal modifier indicating a specific point i n time to an activity predicate introduced by wa w i l l yield a well-formed sentence; this  42  prediction is borne out and can be seen with the addition of the phrase na7 t-kwi an7us-t 'at two o'clock'. The translation of the sentence in (65) demonstrates that the singing event will occur in the immediate future and cannot mean that the singing took place at some point in the past:  (65)  chen  wa lulum na7 t-kwi  lSUBsg PA sing  an7us-k  LOC OBL-DET two-IRR  'I am singing at two o'clock' LB 21-07-98  Unlike in French, when wa is omitted from the sentence, the sentence remains well-formed, under a different interpretation. Recall that in French, neither the future nor the past readings were available for the sentence containing a predicate in the complex present as well as a temporal modifier referring to a particular point in time. In Skwxwu7mesh, on the other hand, the past interpretation is available; this is illustrated by the sentence in (66):  (66)  chen  lulum na7 t-kwi  an7us-k  lSUBsg sing LOC OBL-DET two-IRR 'I sang at two o'clock' LB 21-07-98  The fact that the sentence in (66) is translated in the past follows from the fact that a Skwxwu7mesh predicate lacking wa is telic; this is to say that since in English the two present forms of an activity predicate (progressive and generic) are atelic, the only way to indicate the telic form of an activity predicate is to translate it into the simple past. Lastly, the telicity distinction is further reinforced with the addition of a subordinate clause containing a telic predicate kwi ses Wit ta John 'when John got here'; this clause is  43  compatible with both the telic and atelic forms of the matrix predicate. Beginning with the atelic form o f the activity predicate (i.e. the predicate introduced by wa), we observe that i n (67), adding the subordinate clause yields an overlapping interpretation where the event of the matrix predicate is already underway when the event of the subordinate predicate occurs:  (67)  chen  wa lulum k w i s-es  lSUBsg PA sing  tl'ik'  ta  John  D E T NOM-3POSS arrive D E T John  'I was singing when John got here' D W 30-07-98  G i v e n the proposal that predicates lacking wa are telic, we do not expect an overlapping interpretation to be possible; instead, we find that adding the subordinate clause to a sentence in which the matrix predicate is not introduced by wa yields an consecutive interpretation in which the singing started at the time that John got here. This is shown in (68):  (68)  chen  lulum k w i s-es  lSUBsg sing  tl'ik' ta  John  D E T NOM-3POSS arrive D E T John  'I sang when John got here' D W 30-07-98  The fact that the sentence is translated in the past, though no overt past tense marker is used, again supports the claim that the predicate is telic . 21  The French data crucially differs from Skwxwu7mesh with respect to this test. Recall that, unlike Skwxwu7mesh, French marks tense overtly; thus, the sentences in (67) and (68) cannot be reproduced in French with the simple/complex present. Instead, the sentences are offered with the simple past and the imperfect. They yield a consecutive interpretation and an overlapping interpretation, respectively: 2  1  (ii)  J'ai chante' quand Jean est arriv6 'I sang when John arrived' (iii) J'6tais en train de chanter quand Jean est arriv6 'I was singing when John arrived' This is the exact opposite pattern that we observe in Skwxwu7mesh.  44  The chart in (69) summarizes the similarities and differences between French and Skwxwu7mesh.  (69) SINCE/DEPUIS  AT/A  X  X  X  X  E N TRAIN D E FRENCH SIMPLE  •  •  (future)  PRESENT -WA  X  SKWXWU7MESH +WA  •  •  •  (past)  (future)  5.4.2.2. The Event Structures of Activities The data above entails that the representation of an activity predicate i n S k w x w u 7 m e s h is different from that i n E n g l i s h or French; however, when an activity predicate i n Skwxwu7mesh is introduced by wa, it is translated as a continuous event, much like those i n English. Consequently, for an activity predicate introduced by wa, I w i l l assume the event structure that Pustejovsky (1991) suggests for a process predicate; recall that Pustejovsky represents process predicates as a sequence of events identifying the same semantic expression. Thus, the event structure of a pluralized activity predicate is represented as in (70):  (70)  wa lulum PA sing  e-PA /  ei  / \ ^ e  n  Consequently, this proposal for Skwxwu7mesh crucially claims that a process is derived, as opposed to being a primitive; this is to say that a sequence of events only arises with the  45  addition of wa. The next issue to consider, then, is the event structure of the bare form of an activity predicate; in other words, we must ask 'from what is this process derived?'. The Skwxwu7mesh sentences above, all of which contain the bare form of the activity predicate 'sing', are consistently translated in the simple past; this is accounted for by the proposal that bare predicates in Skwxwu7mesh denote telic events. Since an atelic activity predicate is represented by a sequence of events, I propose that the telic form of an activity predicate is represented by a single event, e, with no further structure; the representation for activity predicates lacking wa is given in (71): (71)  lulum sing  e  Although it may be difficult to conceive of a singing event being divided into a sequence of single events of sing, there are other activity predicates in which the distinction between a single event and a sequence of events is much clearer. Notice that when the predicate xwitem 'jump' is introduced by wa, the present-progressive reading is given; the atelic form of the predicate is given in (72):  (72)  chen  wa xwitem  lSUBsg PA jump 'I am jumping  1  However, the predicate in its bare form is not only translated in the past, but is interpreted as a single event, having occurred one time; this is illustrated in (73):  (73)  chen  xwitem  lSUBsg jump 'I jumped (once)'  46  Thus, we can answer the question above ('from what are complex activity predicates derived?') by stating that activity predicates are derived from single events; pluralizing e yields an atelic process in the form of a continuous reading. Thus, when introduced by wa, an activity denotes multiple singular, telic events which sum up to form an atelic token of the predicate. With respect to Cusic's parameter, a pluralized activity predicate yields phase level repetition. Recall that phase level repetition is represented as i n (74), where the free variable ranging over properties of events (P) is fixed lexically:  (74)  V-PA(X) <=> VeeX[P(e)]  & card(X) > n where P=lexically fixed  P, in this case, would be a single event of the predicate 'sing'. The continuous event reading, however, is not the only possible reading available for an activity predicate introduced by wa; a habitual reading is also available. Consequently, pluralizing an activity predicate can also yield event level repetition. M u c h like in English and French, a Skwxwu7mesh activity predicate can y i e l d a habitual reading. Firstly, notice that in English, the habitual reading arises when an activity predicate is given in the present (without the progressive markings); this is illustrated by the sentences i n (75):  (75a)  I run  ('I am a runner')  (75b)  I sing  ( ' l a m a singer')  (75c)  I work  ('I am a worker')  In French, we observe that the habitual reading is available for an activity predicate i n its simple present form, but not in its complex present form; thus, the sentence i n (76a) can  47  have a habitual interpretation in addition to its continuous interpretation. In contrast, the sentence in (76b) does not yield a habitual reading. This contrast is illustrated in (76):  (76a)  Je chante /'I sing [habitually]' /'I am singing'  (76b) Je suis en train de chanter *'I sing [habitually]' /'I am singing (as we speak)'  In Skwxwu7rnesh, the habitual reading is available for an activity predicate only if it is introduced by wa, given an appropriate contextual set-up; thus, as an answer to a question about one's daily morning actions, an activity predicate must appear in its pluralized form to indicate that the event is a habitual event. This pattern is illustrated by the sentences in (77); the two sentences in (77b) and (77c) can each be matched with the sentence in (77a) to form question-answer pairs:  (77a)  chexw wa chanem i7xw natlh 2SUBsg PA do.what all  k'-axw  wa umsem  morning IRR-2SUBsg PA wake.up  'What do you do every morning when you get up?' (lit: 'you do what every morning when you wake up') LB 25-08-98  48  (77b)  chen  wa lulum  lSUBsg PA sing 'I sing' LB 25-08-98  (77c)  chen  wa ts'its'ap'  lSUBsgPA work 'I work [every morning]' LB 25-08-98  The habitual reading follows from the proposal that wa causes a predicate to be atelic. Recall that an achievement predicate yields a habitual reading when pluralized; thus, in that same way that an achievement predicate becomes atelic by the addition of wa, so too can an activity predicate become atelic and yield a habitual reading when introduced by wa. We must then explain how to represent a pluralized activity predicate in Skwxwu7mesh that yields a habitual reading. Given the fact that a bare activity predicate is represented as a single event, e, with no further structure, the event structure of a pluralized activity predicate yielding a habitual reading is identical to the event structure of an activity predicate yielding a continuous reading. This structure is shown in (78): (78)  wa lulum PA sing  e-PA  Since the reading represented by the structure in (78) is a habitual one, the type of repetition that arises from this pluralization of an activity predicate is event level, where the pluractional marker yields multiple events of the type denoted by the predicate; that is pluralized e gives rise  49  to a number of singing events that are understood to occur on separate occasions. Thus, the definition for the pluractional marker for an activity predicate that yields a habitual reading w i l l be identical to one proposed for an achievement predicate, where the free variable ranging over a set of properties (P) is equated with the verb itself (V); in this case, the verb is a member of the aspectual class o f activity predicates. This representation is shown i n (79):  (79)  V-PA(X) <=> VeeX[P(e)]  & card(X) > n  where P=V  Consequently, an activity predicate differs from an achievement predicate i n that a pluralized achievement predicate only yields the habitual reading, and hence event level repetition; a pluralized activity predicate can yield either a habitual reading or a continuous reading, and hence event level repetition or phase level repetition. The i m p l i c a t i o n of this analysis of an activity predicate is that, at least i n S k w x w u 7 m e s h , it w i l l be impossible to distinguish between a habitual reading and a continuous reading in the event structure of a pluralized activity predicate. This is not true of predicates from all aspectual classes, however, since a pluralized accomplishment predicate yields both types o f repetition that are marked differently i n the event structure. This issue is explored i n the following subsection.  5.4.3.  Accomplishments  A n accomplishment predicate is made up of an activity and a culmination point; recall that Pustejovsky's representation for an accomplishment predicate is a transition (T) consisting of an initital subevent (P) and a final resulting state (S) (or a final transition (T)). representation is shown again in (80):  50  This  (80)  T P ei  S/T e  n  In S k w x w u 7 m e s h , however, I have claimed that the bare form of an activity predicate is represented by e; thus, we expect the bare form of an accomplishment predicate to be a transition consisting of an initial subevent e and a final resulting state. This representation is provided in (81): (81)  T e  S  T o verify that this representation is the correct one, it is necessary to examine the appropriate data. When it occurs without wa, an accomplishment predicate yields a reading whereby the event has culminated; the sentence is, consequently, translated in the past. This is illustrated by the sentences i n (82):  (82a)  chen  ti-lam7  lSUBsg make-house 'I made a house' D W 30-07-98  51  (82b)  chen  sikw'-it  t-en  yekway  lSUBsg cut-TRANS OBL-lPOSS dress 'I tore my dress' E J 27-08-98 228  For the sentence i n (82a), the speaker states that it "sounds like [the house is finished]"; thus, it is not surprising that the sentence is translated in the past even though no overt past tense marker is present . W e can take the speaker's translation and comment to reinforce the fact 22  that a bare accomplishment predicate in Skwxwu7mesh denotes a completed event. Thus, the data in (82) and the speaker's comments support the representation in (81). Recall Pustejovsky's claim that the event head (i.e. the portion of the matrix event that is in focus) of an accomplishment predicate is the initial process subevent. G i v e n that the event structure of a Skwxwu7mesh accomplishment predicate consists o f an initial subevent e, the proposal that a pluractional marker pluralizes the head of an event structure predicts that e is the target of pluralization. Since a pluralized e can yield a continuous reading of an activity, we expect that a pluralized accomplishment predicate yields a continuous reading of its initial subevent. This prediction is confirmed by the data in (83); when an accomplishment predicate is introduced by wa, a present continuous reading arises. This is illustrated by the sentences i n (83):  (83a)  chen  wa  ti-lam7  lSUBsg PA make-house 'I am making a house' D W 30-07-98  Recall from Chapter 2 that the aspectual class of the predicate may define the temporal interpretation of a sentence in the absence of overt tense marking.  22  52  (83b)  chen  w a sikw'-it  ten  yekway  lSUBsg PA cut-TRANS lPOSS dress 'I was cutting my dress' E J 27-08-98  The speaker emphasizes the fact that the predicate with wa is not yet culminated when she states that the house is "not finished"; this again demonstrated that the pluractional marker, wa, causes the event to be atelic. I propose that an accomplishment predicate introduced by wa is represented by a transition consisting of an initial pluralized subevent e and a resulting state. This structure is given i n (84):  (84)  wa ti-lam7  T  PA make-house  S  e-PA  This representation raises a question about the status of phase level repetition; according to Cusic (1981), phase repetition entails multiple events which sum up to form a single token of the type denoted by the verb. Though it is clear that e is being pluralized, it is not clear whether it is e in the aspectual representation of a process that is being pluralized, or e that dominates a sequence of such primitive subevents (i.e. the e that is a phase of the transition). This is an analogue ot the problem alluded to above with respect to an activity predicate where, recall, it was impossible to distinguish between phase-level repetition and event-level repetition. For an accomplishment predicate, the distinction between event level and phase level repetition is found i n the event structure since the event structure of an accomplishment is complex; that is, there is further structure in the representation of an accomplishment predicate, namely, a higher transition (T).  53  Recall that an achievement predicate is represented by a transition; when pluralized, the T in the event structure of the achievement predicate is targeted, yielding a habitual reading. We already know that it is possible for one predicate to yield two different readings when pluralized; furthermore, we know that transitions can be targets for pluralization. Thus, we predict that an accomplishment predicate can also yield a habitual reading, in which case, the transition denoted by the predicate will be pluralized (as opposed to the initial subevent). Before examining this representation, let us confirm that this prediction holds true in the data. Given the appropriate context, an accomplishment predicate in Skwxwu7mesh can indeed yield a habitual reading; thus, specifying a person's present occupation should yield a statement in which a particular event occurs multiple times without indicating a culmination point. Since this type of statement denotes an atelic event, we expect that in Skwxwu7mesh, an accomplishment predicate with a habitual reading should be pluralized. The fact that the predicate is introduced by wa demonstrates that this is exactly the case; this is illustrated in (85):  (85)  Peter na wa tehim ta  lam7 nilh s-ts'its'ap'-s  Peter R L PA make DET house F O C NOM-work-3POSS 'Peter builds houses, that's his job'  A pluralized accomplishment predicate that yields a habitual reading is represented as a pluralized T consisting of an unspecified number of Ts. This representation is given in (86): (86)  wa ti-lam7  T-PA  54  W i t h this representation, the pluractional marker wa is not pluralizing the event head, but the event type denoted by the accomplishment predicate . 23  5.4.4. States Recall that a state is a predicate that persists over a stretch of time and apparently involves no dynamics. There are two types of stative predicates: stage-level and individual-level. In the first subsection, I discuss stage-level stative predicates and i n the second subsection, I discuss individual-level predicates.  5.4.4.1. Stage-level Stative Predicates W e observe, once again, that i n Skwxwu7mesh a stage-level state is introduced by wa; this is illustrated in (87):  (87a)  wa kw'ay' ta  mixalh  P A hungry D E T bear 'the bear is hungry'  (87b)  chen  wa t'ayak na7  t-kwi.  an7us-k  lSUBsg P A angry LOC OBL-DET two-IRR 'I was mad at two o'clock  1  D W 30-07-98  ^^Preliminary data shows that reduplication of the predicate, as well as the addition of wa, yields a reading whereby the transitions are occurring continuously in time; this is apparent from the speaker's explanation that you are building houses "one after the other": (iv) chen wa te-thfm ta lam7 lSUBsg PA RED-make DET house T m making houses, one after the other' Note that the speaker also offers a sentence where the predicate is not reduplicated, but the object is and says that this is an alternative to the sentence where the predicate is reduplicated. I leave this issue to further research.  55  (87c) chen  wa lhchiws  lSUBsg PA tired 'I am tired' LB 25-08-98  However, stage-level state predicates are well-formed without wa, but yield a different interpretation; in their bare form, stage-level state predicates are translated with a change-ofstate reading. This is illustrated in (88):  (88a) chen  t'ayakna7 t-kwi  an7us-k  lSUBsg angry LOC OBL-DET two-IRR 'I got mad at two o'clock' DW 30-07-98  (88b) chen  kw'ay"  lSUBsg hungry 'I got hungry' LB 25-08-98  (88c) chen  kw'ay'kwi-n-s  na kwach-nexw-an  ta  sch'exwk  lSUBsg hungry DET-1P0SS-N0M RL see-TRANS-lSUBsg DET fried.food 'I got hungry when I saw the bannock' LB 25-08-98  56  (88d)  chen  lhchiws  lSUBsg tired 'I got tired' L B 25-08-98  Recall that Pustejovsky suggests that the feature that distinguishes between a stage-level predicate and an individual-level predicate is that a stage-level predicate has an "inherent reference to that factor that brings this state about" (1995:225). Thus, for Pustejovsky, the event structure of a stage-level predicate is represented by a transition with an initial default subevent process (P) and a final resulting state (S), where the resulting state is the event head. This was shown in (50) and is repeated in (89):  (89)  T  P  / \  S where S=event head  Pustejovsky must c l a i m that the initial subevent is a default because he does not want to represent a change-of-state reading; Pustejovsky's goal is to represent the reading of a stagelevel predicate.  The complex structure that Pustejovsky proposes is not a problem for  Skwxwu7mesh, however, since the change-of-state reading is exactly the reading that a bare stage-level predicate yields. Taking into consideration the representation proposed for a bare activity predicate (cf. (71)), I propose that a bare stage-level predicate is represented by a transition (T) with an initial subevent e and a final resulting state (S). This event structure is given in (90):  (90)  T  57  Assuming this structure for the bare form of the stage-level predicate (with the change-of-state reading), we must then explain what portion of this event structure is the target for plurahzation; i n other words, what is wa pluralizing in order to yield the stative reading? Since the resulting state of the transition is the head of the event structure, it should be the target for plurahzation; moreover, the fact that the stage-level stative reading arises when wa is added to the predicate also demonstrates that it is the final state that is being pluralized,. Thus, the transition representing a complex stage-level predicate consists of an initial subevent e and a final pluralized state (S-PA); this event structure is given in (91):  (91)  T  ei  e  n  Thus, I am claiming that a stage-level stative predicate is derived; however, my claim is that it is derived from a transition, not an individual-level predicate; moreover, notice that the pluralized state now resembles a derived process. W i t h respect to phase level repetition, the same issue that is raised with a pluralized accomplishment applies to complex stage-level stative predicates. The event structure o f a stage-level predicate, like that of an accomplishment predicate, is complex; pluralization targets a subevent portion of the event structure. Thus, it is unclear whether it is pluralized e that yields the phase level repetition or the fact that S is merely a phase of the transition representing a bare stage-level predicate that yields phase level repetition . 24  S i n c e the structure provided in (91) is that of a transition, we expect that a habitutal reading should be available (as with accomplishments and achievements). At this point, there is not enough data to conclude whether the habitual reading is simply missing from predicates of this class; I leave this issue to further research. 2 4  58  5.4.4.2. Individual-Level Predicates Since an individual-level predicate denotes a property that can only be attributed to an individual once and whose duration is not bounded, it is, by definition, atelic. Thus, contrary to predicates from all other aspectual classes, the bare form of an individual-level predicate in Skwxwu7mesh is expected to be atelic.  Consequently, we expect that individual-level  predicates are not compatible with wa since the pluractional marker would entail a plurality that is not available for predicates of this class; this, in fact, is the case. This is illustrated by the sentences in (92):  (92a)  hiyita. mixalh big DET bear 'the bear is big'  (92b)  wa hiyita  mixalh  PA big DETbear *'the bear is big'  (92c)  chen  iy7fm  lSUBsg strong 'I am strong' DW 30-07-98  Follow Pustejovsky, I assume the event structure of a state for a predicate from the class of individual-level predicates; this is illustrated in (93) : 25  F o r the analysis presented in this thesis, I distinguish this representation from the representation of a single event e with no further structure. 2 5  59  S  (93)  I  e  Notice, however, when a clause specifying a time frame in which the matrix predicate holds is added to the sentence, the predicate is introduced by wa. This is illustrated in (94), with the addition of the clause lh-7an wa ts'ets'kw'a-t-sut 'when I'm running':  (94)  wa iy7im lh-7an  chen  wa ts'ets'kw'a-t-sut  lSUBsg PA strong when-lSUBsg PA run-TRANS-REFL 'I am strong when I'm running'  DW 30-07-98  This would suggest that the individual-level predicate is being treated as a stage-level predicate in that the predicate becomes a temporary property that can be attributed to the individual more than once . What, then, is the event structure of a stage-level predicate that is derived from an 26  individual-level predicate? There are two ways to answer this question; firstly, we might represent a stage-level stative predicate derived from an individual-level predicate as a pluralized state yielding event level repetition. It is not clear how how this event structure should be represented; however, a first attempt might appear as the structure in (95):  (95)  S-PA  / \  Si  I  e  S  I  n  e  A s noted in Kratzer (1995), "if the distinction between stage-level and individual-level predicates is operative in natural language, it cannot be a distinction that is made in the lexicon of a language once and for all" (125-6). 2 6  60  Secondly, the representation of a stage-level predicate that is derived from an. individual-level predicate may appear exactly as a pluralized state. This structure is illustrated in (96): (96)  S-PA  / \ ei  e  n  This representation parallels the final subevent of a pluralized stage-level stative predicate that is derived from a transition (cf. (91) above). The next issue to address would then be whether the change-of-state reading is available for a stage-level stative predicate derived from an individual-level predicate; if the change-of-state reading is available, we would expect the event structure of a stage-level stative predicate that is derived from an individual-level predicate and one that is derived from a transition, to be the same. The next question would then be whether the change-of-state reading is available for an individual-level predicate; at this point in the analysis, this reading does not seem to be available. I leave this issue to further research.  5.4. Summary The chart in (97) provides a summary of the findings of this chapter; for each aspectual class, the type of repetition that arises from pluralization is given, along with a specification of what portion of the event structure is pluralized to yield that reading.  61  (97) EVENT-LEVEL  PHASE-LEVEL  X  ACHIEVEMENT  +WA  /CD  ACTIVITY  +WA  /(e)  /(e)  ACCOMPLISHMENT  +WA  /CD  /(e)  I-L  +WA  X  X  S-L  +WA  ?(T)  STATE  /  (S)  The X in both columns for individual-level predicates indicates that pluralization of an individual-level predicate is simply not possible; however, since individual-level predicates can yield stage-level stative readings when they are introduced by wa, it should be noted that the stage-level stative reading of the predicate is a pluralized form, though it remains unclear whether the type of repetition is phase level or event level. The two event structure models discussed in this chapter are summarized i n (98):  (98) Pustejovsky's  S 1  1  e  State  representations T  P  /  ei  S\  /\ / \  \ e  Process  -,E  1  Proposed representations for Skwxwu7mesh  1  S  |  1  E 1  n  Transition  1  e  1 i  / \ s\  —iE  e  State  T  Single Event  E  Transition  The event structures proposed for Skwxwu7mesh differ from those proposed by Pustejovsky in one crucial way; according to Pustejovsky's model, a process is a primitive. According to  62  the model presented in this thesis, a process is derived from a single event, represented by The issue of primitives is discussed further in Chapter 7.  63  6. S O L V I N G T H E S E C O N D  PROBLEM  A s outlined i n Chapter 3, the Skwxwu7mesh data brings about an interesting paradox with repsect to adverbs of quantification. Recall that wa is incompatible with an individual-level predicate reading in a non-quantified sentence, yet wa is obligatory with an individual-level predicate in a quantified sentence. This is repeated in (98):  (98a)  hiyita  mixalh  big DET bear 'the bear is big'  (98b)  ?wahiyita  mixalh  PA big DET bear *'the bear is big'  (98b)  lhik' always  *(wa) hiyi ta  mex-mixalh  PA big DETRED-bear  'these bears are always big'  In this chapter, I argue that these facts provide crucial evidence that Q-adverbs should be analyzed as quantifiers strictly over events (De Swart); furthermore, I use the proposal that wa is a pluractional marker to explain how quantification over events operates i n Skwxwu7mesh. Section 6.1. outlines the most widely known proposal for Q-adverbs (Kratzer 1995); Kratzer analyzes Q-adverbs as unselective binders in which a Q-adverb can either bind a spatiotemporal variable or an individual variable, yielding a temporal and an atemporal interpretation, respectively. Thus, to begin this section, I present Kratzer's view of the temporal/atemporal distinction together with her view of the stage-level/individual-level distinction (Subsection 6.1.2.). I show how Kratzer analyzes 'when-clause' data in her unselective binding approach  64  and then informally apply her analysis to simplex sentences. I conclude Section 6.1. by explaining how Kratzer's analysis of Q-adverbs cannot account for the Skwxwu7mesh data. Section 6.2. presents an analysis of Q-adverbs that does capture the Skwxwu7mesh facts; I claim that Skwxwu7mesh provides crucial evidence for De Swart's (1993, 1995) event-based approach to analyzing Q-adverbs whereby quantification is strictly over events. H e r analysis of sentences containing a stage-level predicate/individual-level predicate is presented and then applied to Skwxwu7mesh. Crucially, De Swart's analysis assumes that a plurality of events is required for quantification; consequently, while De Swart claims that the set of events is either encoded i n the predicate or arises from an (in)definite N P , I argue that S k w x w u 7 m e s h provides evidence that the set of events is always created overtly, by the pluractional marker  wa.  6.1. Quantification over Events/Individuals: Kratzer (1995) It is Kratzer's (1995) c l a i m that adverbs of quantification (Q-adverbs), such as always, usually, rarely...cm bind either spatiotemporal ("event") variables or individual variables. This section outlines this approach to analyzing Q-adverbs; I begin with a discussion o f the two types of interpretations that arise from quantification by Q-adverbs.  6.1.1. Temporal and Atemporal Interpretations There are two possible interpretations of sentences containing Q-adverbs: temporal and atemporal interpretations. The sentence in (99a) below is an example of a sentence that yields a temporal interpretation while the sentence in (99b) yields an atemporal interpretation:  (99a)  John is always hungry  (temporal)  (99b)  A bear is always big  (atemporal)  65  Informally, the interpretation in (99a) is described as temporal since the sentence seems to be providing information about the times that John is hungry; that is, the sentence i n (99a) can be paraphrased At all times, John is hungry. Although the same adverb, always, is used in (99b), the sentence is paraphrased All bears are big; this is to say that the interpretation is atemporal since the sentence does not provide any information about the times at which bears are big. In Kratzer's unselective binding approach to analyzing Q-adverbs, temporal and atemporal interpretations are understood to arise from different types of quantification; this is to say that the readings that arise are dependent on the variable that is quantified over by the Q adverb. Thus, when a Q-adverb quantifies over a variable that ranges over a spatio-temporal location, a temporal interpretation arises; on the other hand, an atemporal interpretation arises in the case where a Q-adverb binds an atemporal entity, or an individual variable. Before examining Kratzer's analysis of Q-adverbs, I provide an outline of her proposal on the distinction between stage-level and individual-level predicates.  6.1.2. The Stage-level/Individual-level  Distinction  Recall that a stage-level predicate expresses a transitory property while an individual-level predicate expresses a permanent property. These two types of predicates are exemplified again in (100);  (100a)  John is tired  (stage-level predicate)  (100b)  John is tall  (individual-level predicate)  The predicate tired i n (100a) is understood to describe a temporary property o f John; the predicate tall i n (100b), on the other hand, is understood to describe a property of John that holds at a l l times. Kratzer captures the distinction between stage-level and individual-level predicates i n terms of argument structure; she claims that stage-level predicates have an extra spatiotemporal argument that individual predicates lack. Thus the argument structure of the 1  66  stage-level predicate tired i n (100a) is represented as in (101a) while the argument structure of the individual-level predicate tall in (100b) is represented as in (101b) below:  (101a)  tired (x, 1)  (101b)  tall(x)  where/ = spatiotemporal argument  According to this proposal, a stage-level predicate w i l l be true of its subject at a certain time (or place), while an individual-level predicate cannot select a particular time (or place) in which it w i l l hold true of its subject; thus, the extra argument o f stage-level predicates specifies the time/location where the event takes place. This captures the fact that a stage-level predicate denotes a temporary property and an individual-level predicate denotes a permanent property. Consequently, the grammaticality o f (102a) and the ungrammatically o f (102b) below are accounted for:  (102a)  John is tired today  (102b)  *John is tall today  The interpretation of (102a) is that there is a specific time, namely today, when John has the property of being tired, accounted for by the fact that tired has an extra argument position that is reserved for that information. In contrast, specifying the time at which John has the property of being tall yields an ungrammatical sentence; Kratzer attributes the ungrammaticality o f (102b) to the fact that the predicate tall does not have a spatiotemporal argument that can be assigned to the temporal modifier today. Notice that (102b) can be judged grammatical in a particular context; for example, i f John is a short person but was seen walking on stilts on a particular day, the sentence John is tall today may not sound odd. Kratzer notes this problem and attributes it to the issue o f classification; she claims that the stage-level/individual-level distinction is not made i n the  67  lexicon "once and for all" but it varies in context because it is a distinction that is "operative in natural language". This is to say that i f John walked on stilts once a week, the property  of  being tall w o u l d be a stage-level property for John (that is, i f it assumed that the stilts themselves can actually vary John's "height").  Taking this into consideration, Kratzer  generalizes that a predicate's argument structure w i l l change depending on its classification as a stage-level or individual-level predicate. She couches the distinction between stage-level predicates and individual-level predicates in a discussion of adverbs of quantification i n order to derive the meaning effects of temporal and atemporal readings.  6.1.3.  When-clauses  Kratzer suggests that 'when-clause' data provide independent evidence for the extra argument position in stage-level predicates. Consider her examples in (103) below:  (103a)  When Mary speaks French, she speaks it well.  (103b)  *When Mary knows French, she knows it well. (Kratzer 1995:129)  Kratzer states that 'when-clause' sentences contain a null operator with the meaning of always; thus an informal representation of (103a) and (103b) above would appear as (104a) and (104b) below respectively:  68  (104a)  Alwaysj [speaks (Mary, French, 1)] [speaks-well (Mary, French, 1)] restrictive clause  (104b)  nuclear scope  *Always [knows (Mary, French)] [knows-well (Mary, French)] restrictive clause  nuclear scope (Kratzer. 1995:130)  In (104a), speaks introduces a spatio-temporal variable since it is a stage-level predicate; always  is co-indexed with the spatio-temporal variable / since / is the free variable i n the  restrictive clause. Consequendy, the quantifier binds all free occurences of the variable / in its entire scope. In (104b), on the other hand, the predicate know does not introduce a spatiotemporal variable since it is an individual-level predicate; thus, the sentence is ungrammatical because there are no free variables for the Q-adverb to bind. It is crucial to her analysis that Kratzer assumes the "Prohibition against Vacuous Quantification" which informally states that a quantifier must have a variable to bind and it must bind every occurence of that variable (i.e. in both the restrictive clause and the nuclear scope). Thus, the exclusion of (104b) follows from the fact that natural language does not permit vacuous quantification. F o l l o w i n g H e i m (1982), Kratzer assumes that indefinite N P s also introduce a free variable that can be bound by the Q-adverb "always"; this is a crucial assumption for Kratzer's analysis in that it helps to explain the grammaticality of the sentences in (105) below:  (105a)  When Mary knows a foreign language, she knows it well  (105b)  When a Moroccan knows French, she knows it well (Kratzer 1995:130)  Both the indefinite N P a foreign language in (105a) and a Moroccan in (105b) introduce a free individual variable (as opposed to the spatio-temporal variable, /, introduced by stage-level  69  predicates); furthermore, the pronoun it in the nuclear scope must also introduce an individual variable.  Consequently, the non-overt operator always w i l l bind these variables i n their  respective sentences, yielding the following analyses:  (106a) AlwaySx [foreign language (x) & knows (Mary, x)] [knows w e l l (Mary, x)] (106b) AlwaySx [Moroccan (x) & knows (x, French)] [knows-well (x, French)]  Thus i n Kratzer's analysis, both spatio-temporal variables and individual variables can be bound by a Q-adverb, yielding temporal and atemporal interpretations, respectively.  6.1.4. Simplex Clauses Although Kratzer does not address mono-clausal sentences i n her analysis, the same facts are observed for them. In the sentence John is always hungry, always binds the event variable introduced by the stage-level predicate hungry; this is illustrated i n (107a) (formal representations have not been shown): (107a) John is always hungry  Q  (l)  i  ^  Kratzer thus accounts for the temporal reading that arises. In the sentence A bear is always big, there is no stage-level predicate introducing an event variable; instead, there is an individual-level predicate, big, that does not introduce an event variable, nor does it introduce an individual variable. There is, however, an indefinite N P , a bear, that introduces a free individual variable; thus, always binds the variable introduced by the indefinite N P . This is shown in (108): (108)  A bear is always big  (x)  t  Q  )  70  Kratzer thus accounts for the atemporal reading that arises. Finally, Kratzer must account for the fact that John is always big is an ill-formed sentence; the ungrammaticality is explained by the fact that there is no free variable for the Q adverb to bind. The individual-level predicate big does not introduce an event variable and the proper name John does not introduce a free individual variable; this is illustrated in (109):  (109)  *John is always big Q  Recall that Kratzer assumes that natural language does not permit vacuous quantification; thus, the sentence in (109) cannot yield a temporal nor an atemporal interpretation. Kratzer's analysis demonstrates that temporal and atemporal readings result from two. types of quantification; quantification over events yields a temporal reading and quantification over individuals yields an atemporal reading. The issue is whether Kratzer's analysis of Q adverbs can account for the generalizations observed in Skwxwu7mesh. The analysis of the Skwxwu7mesh data presented i n this thesis argues that a plural D P is not sufficient to make a sentence containing a Q-adverb well-formed; instead, the claim is that a l l predicates must be pluralized when they occur in a sentence containing a Q-adverb. Thus, this thesis claims that the unselective binding approach to quantification cannot account for Skwxwu7mesh. Others have proposed that quantification is strictly over events and that quantification over individuals is not necessary to to account for atemporal readings (De Swart 1993,1995); it is this account of Q-adverbs that I w i l l adopt. The following section outlines De Swart's analysis of Q adverbs and demonstrates how the Skwxwu7mesh facts can be explained.  71  6.2. The Event-Based Apporoach:  De Swart (1993, 1996)  De Swart claims that Q-adverbs require a plurality of events (situations) on which to operate; this set arises i f one of the predicate's arguments is variable. Rather than distinguishing between stage-level and individual-level predicates by whether the predicate introduces an extra argument position for events or spatiotemporal locations (the Davidsonian argument), de Swart « suggests that a Davidsonian argument can be added to every predicate. She then introduces a uniqueness presupposition on the Davidsonian argument to account for the fact that individuallevel predicates and 'once-only' predicates do not combine with a Q-adverb:  (110) Uniqueness presupposition on the Davidsonian argument:  If not empty, the set of events that is in the denotation of a 'once-only' predicate is a singleton set for all models and each assignnment of individuals to the arguments of the predicate (de Swart 1996:179)  Although uniqueness blocks quantification, sentences containing an individual-level predicate or 'once-only' predicate can combine with Q-adverbs i f an N P in the sentence introduces a variable; to account for this, de Swart offers the following plurality condition on quantification: i  (111) Plurality condition on quantification:  A Q-adverb does not quantify over a set of situations i f it is known that this set has a cardinality less than two. A set of situations is known to be a singleton set if: 1) the predicate contained in the sentence satisfies the uniqueness presupposition on the Davidsonian argument, and 2) there is no (in)definite N P present in the sentence which allows indirect binding by means of quantification over assignments. (de Swart 1993:130)  Thus, the crucial difference between Kratzer's approach and D e Swart's approach is that Kratzer's approach allows for either spatiotemporal variables or individual variables to be bound by a Q-adverb. D e Swart's approach, on the other hand, does not allow for direct binding o f individual variables, but for indirect binding by means of quantification of assignments.  .  72  •  -  D e Swart claims that the temporal and atemporal interpretations do not result from two different types of quantification predicted by unselective binding; instead, she claims that the meaning effects result from pragmatics. In the event-based analysis, the mapping relationship between participants and events w i l l vary depending on the predicate. The next subsection applies D e S wart's analysis to English sentences containing Q-adverbs.  6.2.1. The Analysis of Q-adverbs in English 6.2.1.1. Stage-level Stative Predicates W i t h a stage-level predicate, a number of events can have the same participant; that is to say that there is no exact one-to-one mapping of participants and events. Consider the example i n (112) below:  (112)  John is always hungry  The event-based analysis assumes that John being hungry is an event; i n (112), there is a plurality o f events i n which John is the participant of every event of John being hungry. The sentence i n (112) denotes a set of participants and a set of events; there is only one participant, John, i n the set of participants, and there are a number of events i n the set of events. Crucially, the Q-adverb requires that John be associated with each and every one of the events i n the set; the Q-adverb then quantifies over the events of John being hungry. This is illustrated i n (113) below:  (113)  (^T^^/^T'X  l=John e i , e2, e3 = events of John being hungry  73  To explain how the set of events arises, De Swart would say that the fact that hungry is a stagelevel predicate that does not satisfy the uniquenes presupposition on its Davidsonian argument suggests that the set of situations has a cardinality of more than one.  6.2.1.2. Individual-Level Predicates A n individual-level predicate differs from a stage-level predicate with respect to the mapping relation that exists between the set of participants and the set of events; for a stage-level stative predicate, a number o f events can have the same participant. For an individual-level predicate, however, each event can have one and only one participant. Consider the sentence i n (114):  (114)  Bears are always big  Under D e Swart's analysis, every event of a bear that is big w i l l have its own bear; that set of events is i n a one-to-one correspondence with the set of bears that are big. Thus, no two events i n which there is a big bear w i l l involve the same bear; this corresponds to 'once-only' predicates (i.e. die) i n that the same action cannot be repeated with respect to the same individual. The mapping of events and participants for individual-level predicates is shown i n (115) below:  1,2,3 = bears e i , &i, e3 = events of bears that are big  The one-to-one mapping relationship between the set of participants and the set o f events can be explained by stating that the Q-adverb requires each event to have one and only one participant. D e Swart claims that the plurality i n the set of events is created by the generic subject; although, according to her plurality condition on quantification, an individual-level predicate (i.e. big) satisfies the uniqueness presupposition on the Davidsonian argument, there  74  is an N P (the generic subject, bears) present i n the sentence that allows for "indirect binding by means o f quantification over assignments" (De Swart 1993:130). G i v e n the Skwxwu7mesh facts, a different explanation is required. The next subsection applies D e Swart's analysis to Skwxwu7mesh data.  6.2.2. The Analysis of Q-adverbs in Skwxwu7mesh 6.2.2.1. Stage-level Stative Predicates Recall that in Skwxwu7mesh, wa is obligatory with a stage-level predicate when it combines with a Q-adverb; this is repeated in (116):  (116a)  lhik'  wa kw'ay' ta  mixalh  always P A hungry D E T bear 'the bear is always hungry'  (116b)  *lhik'  kw'ay' ta  mixalh  always hungry D E T bear  G i v e n the proposal that wa is a pluractional marker in Skwxwu7mesh, we can account for the temporal reading of the sentence i n (116a) results from the quantification over a plurality o f events, events of the bear being hungry, for one participant, the bear.  Thus, D e Swart's  analysis can account for the Skwxwu7mesh data i n that a plurality of events is required for quantification. However, the analysis for Skwxwu7mesh assumed i n this thesis differs from De Swart's analysis i n one crucial respect; while De Swart claims the plurality of events results from the the fact that a stage-level predicate does not satisfy the uniqueness presupposition, Skwxwu7mesh provides evidence that the plurality of events is marked overtly. Under the claim that wa is a pluractional marker, we account for the set of plural events by stating that wa pluralizes the predicate, yielding a set of events over which the Q-adverb can quantify.  75  6.2.2.2. Individual-Level Predicates Thus far, I have claimed that the plurality in the set of events in Skwxwu7mesh is created by a pluralized predicate (i.e. a predicate introduced by wa). Recall that an individual-level stative predicate in Skwxwu7mesh is not introduced by wa; in the cases where they are introduced by wa, they are glossed as stage-level predicates.  However, the translation for a quantified  sentence containing an individual-level predicate demonstrates that the predicate is not being used as a stage-level predicate; this is illustrated in (117):  (117)  lhik  1  w a h i y i ta  always P A big  mex-mixalh  DETRED-bear  'these bears are always big' speaker: "a group of bears in this area"  Since the predicate is not pluralized in its non-quantified form, I claim that the predicate is pluralized i n order to satisfy a plurality condition on quantification. This explains why an individual-level predicate is introduced by wa when it combines with a Q-adverb, but lacks wa in its non-quantified form. However, there is still a question remaining as to why a plural D P is required. Notice that in Skwxwu7mesh, a quantified sentence containing an individual-level predicate and wa must also have a plural D P as its subject; with a singular D P , the sentence is judged ungrammatical. This is illustrated in (118):  (118)  *lhik'  wa hiyita  mixalh  always P A big D E T bear  76  Recall Matthewson's (1996) claim that D P s in Salish do not encode a distinction between definiteness and indefiniteness; under this view, the judgment in (118) is expected since the D P selects one bear, not a group of them. Consequently, the D P in (118), unlike the generic subject i n E n g l i s h , does not denote any plurality.  In the event-based approach, the  ungrammaticality of (118) is explained by the fact that there is no plurality of events over which the Q-adverb can quantify; since there is only one event of a bear being big, the quantification is, according to de Swart, "in some sense trivial" (1996:179). However, according to the analysis proposed in this thesis, the source of the plurality of events is the pluralized predicate and not the D P ; thus, we would expect the sentence in (118) to be grammatical since it contains a pluralized predicate. W e can explain the requirement for a plural D P by stating that the oneto-one mapping relationship between the set of events and the set of participants would not hold i f the D P was singular, this is to say that with a singular D P there would be one bear in the set of participants and a plurality of events in the set of events. This mapping relationship is only available for stage-level predicates (cf. (113)); in order to satisfy the one-to-one mapping requirement for a Q-adverb in combination with an individual-level predicate, a plural D P must be introduced.  77  7.  THEORETICAL/CROSS-LINGUISTIC  IMPLICATIONS A N D F U R T H E R  ISSUES  This thesis has argued that wa is a pluractional marker that pluralizes the head of a predicate's event structure or the event type denoted by the predicate.  This proposal explains the  distribution of wa across four aspectual classes (achievements, activities, accomplishments and states). Furthermore, this proposal accounts for the different interpretations that arise when a predicate is introduced by wa. Whether a pluralized predicate yields event level repetition or phase level repetition can be identified in the event structure of the predicate (with the exception of activity predicates). This proposal leads to the claim that bare predicates (with the exception of individual-level predicates) in Skwxwu7mesh are telic; thus, what wa does to a predicate is cause it to become atelic. Atelicity is marked by either the continuation of a single event or the iteration of an event . This analysis of Skwxwu7mesh wa is a formalization o f Kuipers's 27  description of wa as a morpheme referring to a process that has duration either i n the form of a single act or the regular performance of it. The proposal of wa as a pluractional marker also accounts for the fact that wa is obligatory w i t h a stage-level stative predicate and an i n d i v i d u a l - l e v e l predicate i n S k w x w u 7 m e s h quantified sentences; this thesis argued that this obligatoriness provides evidence that Q-adverbs quantify strictly over events. The analysis presented i n this thesis has both cross-linguistic and theoretical implications.  This chapter outlines two major implications of this analysis; Section 7.1.  outlines the implications for aspectual classes with respect to Pustejovky's event structure representations and the event structures proposed for S k w x w u 7 m e s h . In particular, this section addresses the issue of cross-linguistic primitives. Section 7.2. outlines the implications of this analysis for quantification; in particular, the question of whether unselective binding is available i n Skwxwu7mesh, and other Salish languages, is revisited. This chapter concludes with a discussion of some issues related to this thesis that require further investigation; some  27  R e c a l l that the status of the pluralization of an individual-level predicate remains unclear.  78  preliminary data on the role of predicate reduplication and different types of quantification in Skwxwu7mesh is presented and briefly discussed.  7.1. Cross-Linguistic/Theoretical  Implications  for Aspectual Classes: On  Primitives in Event Structure models In Chapter 5 I presented Pustejovsky's event structure m o d e l and illustrated how S k w x w u 7 m e s h predicates are analyzed within his theory; I argued that a process i n S k w x w u 7 m e s h is derived.  Since Pustejovsky's model is intended to be a universal  representation of predicate classes, this claim has a larger theoretical implication; i n other words, i f a process is derived in at least one language, namely Skwxwu7mesh, Pustejovsky's model cannot be a universal representation of predicate classes. There are two approaches to setting out an explanation of these facts, both of which raise further questions. The first approach assumes that event structures are universal; this approach w i l l require further cross-linguistic research to verify i f this, in fact is the case. However, on a much simpler level, the question to address is whether English can be accounted for within a model that does not assume process as a primitive class. The second approach assumes that event structures are not universal; i f this is the correct approach, we are left with the task of finding a parameter.  This is to say that a  parameter must be established in order to account for each of the world's langauges; a common thread must be identified to determine for which language the parameter is on and for which it is off.  7.2. Cross-Linguistic Implications for Quantification: On the Availability of Unselective Binders in Salish In Chapter 6 I claimed that Kratzer's unselective binding approach to analyzing Q-adverbs cannot account for the fact that wa is obligatory in S k w x w u 7 m e s h quantified sentences. Consequently, I argued that quantification is stricdy over events (following De Swart). Crucial  79  to Kratzer's analysis is the fact that indefinite N P s introduce variables that can be bound by a Q-adverb; recall, however, that Matthewson (1996) claims that DPs in Salish languages do not encode a distinction between definiteness and indefiniteness. Thus, it may not be the case that Kratzer's analysis is, i n fact, wrong; instead, we might say that, due to the status of D P s i n Salish, Kratzer's account of Q-adverbs is blocked from applying. Consequently, D e Swart's claim that Q-adverbs only quantify over events may be the default analysis that applies to a language such as Skwxwu7mesh when quantification over individuals is prevented. Jelinek (1995) claims that Straits Salish has unselective adverbial quantification. She argues that Straits Salish lacks nouns and thus, D-quantificaiton is not available. Instead, she states that A-quantification is the only type of quantification available, and consequently, adverbial quantification is unselective. Since Salish languages are thought to exhibit many of the same properties, the analysis of adverbial quantification presented i n this thesis demonstrates that at least one Salish language does not have unselective quantification. Further research regarding this issue is required.  7.3. Issues for Further Research in SJs.wx.wu7mesh 7.3.1. The Role of Predicate Reduplication Reduplication is a related issue to a discussion of quantification. Demers and Jelinek point out that "reduplication is a process...[that] falls into the class of A-quantifiers" (1996:75). Furthermore, they state that "while particular reduplication patterns mark particular quantitative notions, these patterns are not confined to a particular root class, but produce a reading constrained by the lexical semantic properties of the root and other morphological material present" (1996:78). Kuipers (1967) classifies reduplication patterns in Skwxwu7mesh into two general classes: total reduplication (i.e. C V C ) and partial reduplication (i.e. C V ) .  He  states that "total reduplication serves to express plurality or collectiveness i n nouns, iteration, intensity or distributiveness i n verbs..."; partial reduplication, on the other hand is said to express "continuous i n verbs...diminutiveness i n nouns and verbs" (1967:98). Preliminary  80  research has shown that the stage-level/ individual-level distinction is of relevance to reduplication; stage-level predicates can be reduplicated but individual-level predicates cannot be reduplicated. The purpose of this section is to outline some prehrninary reduplication facts, and to show that reduplication of a stage-level predicate is not enough to license a Q-adverb. In non-quantified sentences containing stage-level stative predicates, some speakers allow wa to be dropped i f the predicate is reduplicated (recall that non-reduplicated stage-level predicates require wa); this is illustrated in (119):  (119)  kw'a-kw'ay' ta RED-hungry  mixalh  DET bear  'the bear's hungry', 'the bear's getting hungry'  Kuipers (1967:346) glosses this reduplicated form as 'be very hungry'; however, speakers do not tend to offer the intensifier gloss. In general, speakers seem to prefer both wa and the reduplicated predicate; this is shown in (120):  (120)  wa kw'a-kw'ay' ta PA RED-hungry  mixalh  DETbear  'the bear is hungry'  Although wa can sometimes be omitted from non-quantified sentences, quantified sentences containing stage-level reduplicated predicates require wa for a judgement of well-formedness. This is illustrated in (121):  (121)  lhik'  chen  wa  kw'a-kw'ay'  always lSUBsg PA RED-hungry 'I'm always hungry'  81  What is of major importance is discovering the differences, i f any, i n interpretations of a reduplicated predicate and a non-reduplicated predicate.  The second gloss given for the  sentence i n (119) above suggests that reduplication may yield an inchoative reading of the predicate. The question that arises from these facts is whether reduplication of a predicate can quantify directly over individuals, as is predicated by Kratzer's unselective binding approach, or must reduplication of a predicate also quantify only over events, as claimed by D e Swart's event-based approach? Further examination of reduplication patterns in Skwxwu7mesh w i l l confirm which analysis can account for the reduplication facts . 28  7.3.2. Types of Quantification Thus far, the focus of this research has been on the Q-adverb always.  Demirdache et.al.  (1994) investigate, and provide a case for, D-quantification i n three Salish languages, one o f which is Skwxwu7mesh; their focus was on the quantifier all. Further research is required in order to establish exactly how many types of quantification exist i n Skwxwu7mesh. Although little is known about the the behaviour of other Q-adverbs in Skwxwu7mesh some preliminary data sketches an oudine for further investigation. Skwxwu7mesh has an alternative construction for always, which is also translated as often or usually; it appears as though wa is obligatory with this construction as w e l l . A n example of a sentence containing this construction is given in (122):  (122)  men huy  kwi-s  wa-s  lulum ta  slhen-lhanay'  just be.finished D E T - N O M PA-3POSS sing D E T RED-woman 'the ladies are always singing', 'the ladies are usually singing'  28  Refer to Chapter 5 (footnote 21) for other evidence of predicate reduplication.  82  The other interpretations of always follow from the fact that generics are not universal; that is, there can be an exception, but the sentence can still be true. There appear to be two constructions available for a translation of sometimes; they contain morphemes whose glosses are not yet determined. These are shown i n (123):  (123)  na(7) tl'-(7)an  kw'iyflsh  LOC (t)lh-lSUBsg dance 'Sometimes I dance' lit: 'there are times when I dance'  (123)  yetl'-axw  kw'iyflsh  ye (t)lh-2SUBsg dance 'Sometimes you dance'  The construction for rarely is formed by the negation marker + always; the status of the obligatoriness of wa in this construction is undetermined. This is illustrated i n (124):  (124)  haw-k-elh  lhik' (wa) lulum kwelhi slh£nay'  NEG-IRR-(e)lh always PA  sing  DEM.F woman  'that woman rarely sings'  The position of lhik' in this type of construction appears to be fixed; this is illustrated by the i l l formedness of (125):  (125)  *lhik'  haw-k-elh  lulum  always NEG-IRR-(e)lh sing  83  Further research w i l l determine the status of these constructions and how they fit within the analysis presented in this thesis.  84  REFERENCES  Bach, E m m o n . 1986. "The Algebra of Events". Linguistics and Philosophy 9:5-16. Bar-el, Leora. 1997,1998. Fieldnotes. Bennett, M i c h a e l and Barbara Partee. 1978. Towards the Logic of Tense and Aspect in English. Bloomington, Indiana: Indiana University Linguistics Club. Brinton, Laurel J . 1988. The Development of English Aspectual Systems: Aspectualizers post-verbal particles. Cambridge: Cambridge University Press.  and  Carlson, Gregory N . 1977. References to Kinds in English. P h . D . dissertation, University of Massachusetts, Amherst. Published 1980 by Garland Press, N e w Y o r k . Currie, Elizabeth. 1996, 1997. Fieldnotes. Currie, Elizabeth. 1997. Topic Time: The Syntax and Semantics of SqwXwulmish Adverbs. M A Thesis, University of British Columbia.  Temporal  Cusic, David. 1981. Verbal Plurality and Aspect. Ph.D. dissertation, Stanford Unviversity. D a h l , Osten. 1981. " O n the Definition of the T e l i c - A t e l i c (Bounded-Nonbounded) Distinction". Tedeschi, Philip J. and Annie Zaenen, eds., Syntax and Semantics (14): Tense and Aspect. N e w York: Academic Press. Davis, Henry. 1997a. Class notes, U B C , Salish Syntax, January. Davis, Henry. 1997b. Wa7 and W a : St'aYimcets and SqwXu7mish. Paper presented at the University of British Columbia Salish Working Group, November. Davis, Henry. 1996. wa7 t'u7 wa7. Paper presented at the Victoria Salish Morpho-Syntax Workshop, January. Davis, Henry, I-Ju Sandra L a i and L i s a Matthewson. 1997. Cedar Roots and Singing Detectives: Attributive Modification in Salish and English. Papers for the 31st International Conference on Salish and Neighbouring Languages. Demers, Richard A . and Eloise Jelinek. 1996. "Reduplication, Quantification, and Aspect i n Straits Salish". Papers for the 31th International Conferrence on Salish and Neighbouring Languages, 75-79. Vancouver, British Columbia: University of British Columbia. Demirdache, Hamida. (to appear). "On the Temporal Location of Predication Times: The Role of Determiners in Lillooet Salish." Proceedings of the Sixteenth West Coast Conference on Formal Linguistics, University of Washington, Seattle, W A . Demirdache, Hamida. 1997. "Predication Times in St'dt'imcets (Lillooet Salish)". presented at the Texas Linguistics Society.  85  Paper  Demirdache, Hamida, Dwight Gardiner, Peter Jacobs and L i s a Matthewson. 1994. "The case for D-quantification in Salish: ' A l l ' in St'aYimcets, Squamish and Secwepemctsin". Papers for the 29th International Conference on Salish and Neighboring Languages, 145-203. Pablo, Montana: Salish Kootenai College. D e Swart, Henriette. 1993. Adverbs of Quantification: N e w Y o r k : Garland Publishing, Inc.  A Generalized  Quantifier  Approach.  D e Swart, Henriette. 1996. "(In)definites and genericity". Kanazawa, M . , Piii6n, C . and Henriette de Swart, eds., Quantifiers, Deduction and Context. Stanford: C S L I Publications. 171-194. Dowty, David. 1979. Word Meaning and Montague Grammar.  Dordrecht: Reidel.  Dowty, David. 1986. "The Effects of Apectual Class on the Temporal Structure of Discourse, Pragmatics or Semantics" Linguistics and Philosophy 9. Galton, Antony. 1984. The Logic of Aspect: An Axiomatic Approach. Press. Garey, Howard. 1957. "Verbal Aspect in French". Language Gerdts, Donna. 1988. Object and Absolutive in Halkomelem  33:91-110. Salish.  H e i m , I. 1982. The semantics of definite and indefinite NPs. Massachusetts, Amherst, Mass. Jacobs, Peter. 1992. Subordinate Clauses in Squamish: Thesis, University of Oregon.  Oxford: Claredon  Garland, N e w Y o r k .  P h . D . Thesis, University of  A Coast Salish Language.  MA  Jelinek, Eloise. 1995. "Quantification i n Straits Salish". B a c h , E m m o n , Eloise Jelinek, A n g e l i k a Kratzer and Barbara Partee, eds., Quantification in Natural Languages. Dordrecht: Kluwer Academic Publishers. 487-540. Kearns, K a t e . 1991. The Semantics of the E n g l i s h Progressive. Massachusetts Institute of Technology, Cambridge, Mass. Kuipers, Aert. 1967. The Squamish Language: Mouton & C o .  Grammar,  Texts, Dictionary.  P h . D . Thesis, The Hague:  Kratzer, Angelika. 1995. "Stage-Level and Individual-Level Predicates". Carlson, Gregory N . and Francis Jeffry Pelletier, eds., The Generic Book. Chicago: The University of Chicago Press. 125-175. Lasersohn, Peter. 1995. Plurality, Publishers.  Conjunction  and Events. Dordrecht: K l u w e r Academic  L e w i s , D a v i d . 1975. "Adverbs of Quantification". Keenan, E d w a r d L . , ed., Formal Semantics of Natural Language. Cambridge: Cambridge University Press. 3-15. Matthewson, Lisa. 1996. Determiner Systems and Quantificational Salish. P h . D . Thesis, University of British Columbia.  86  Strategies: Evidence from  Mourelatos, Alexander P . D . Philosophy 2:415-434.  1978.  Events, Processes, and States.  Linguistics  and  Pustejovsky, James. 1991. The syntax of event structure. Cognition 41:47-81. Pustejovsky, James. 1995. The Generative Lexicon.  Cambridge, M A : M I T Press.  van Hout, Angeliek. 1996. Event Semantics of Verb Frame Alternations: A Case Study of Dutch and its Acquisition. Tilburg: Tilburg Dissertation in Language Studies. Vendler, Zeno. 1957. "Verbs and Times". Philosophical  Review 66:143-160.  V e r k u y l , Henk. 1993. A Theory of Aspectuality: The Interaction between Temporal Atemporal Structure. Cambridge: Cambridge University Press.  87  and  APPENDIX  orthography  1:  K E Y TO SKWXWU7MESH  phonetic symbol  ORTHOGRAPHY  orthography  phonetic symbol  p  P  xw  p'  y P  k  q q  m  m  k'  m'  m  kw  t  t  kw'  q •>w q X  »  V  f  t  X  ts  c  xw  ts' s n  c s n  h w y  h w y  ch  V  y'  y  ch'  c » c  sh  s  i  i , e, e  lh  4  u  u,  a  a  7  ?  lh'  >  e  >  1  1 k  k  k'  i<  kw  k  kw'  k  w  w  88  0,  0  APPENDIX  2:  LIST OF ABBREVIATIONS  *  ill-formed sentence  /  (used to contrast with '*' when a different interpretation is available)  ?  well-formedness/ill-formedness of the sentence is unclear  lPOSS  first person possessive  3POSS  third person possessive  lSUBsg  first person singular subject  2SUBsg  second person singular subject  3SUBsg  third person singular subject  30BJsg  third person singular object  DEM.F  feminine demonstrative  DEM.M  masculine demonstrative  DET  determiner  DET.F  feminine determiner  (e)  epenthetic schwa  FOC  focus marker  FUT  future  INTRANS  intransitivizer  IRR  irreahs  LOC  locative  NEG  negative  NOM  norninalizer  OBL  oblique  PA  pluractional marker  PAST  past tense marker  RED  reduplicant  89  REFL  reflexive  RL  realis  TRANS  transitivizer  For St'dt'imcets 3sgP0SS  third person singular possessive  DET  determiner  DET.ABSENT absent determiner  90  

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
Ukraine 122 0
United States 50 4
China 25 0
Canada 18 2
Germany 7 75
United Kingdom 5 0
France 4 1
Philippines 4 0
Switzerland 2 0
Israel 2 0
Hong Kong 1 0
City Views Downloads
Unknown 135 75
Zhengzhou 10 0
Mountain View 9 0
San Francisco 8 0
New Westminster 8 0
Fort Worth 8 0
Beijing 7 0
Ashburn 6 0
Patchogue 6 0
Tübingen 5 1
Hangzhou 4 0
Shenzhen 4 0
Vancouver 4 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0088526/manifest

Comment

Related Items