UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Control in Skwxwu7mesh Jacobs, Peter William 2011

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

Item Metadata

Download

Media
24-ubc_2011_spring_jacobs_peter.pdf [ 4.71MB ]
Metadata
JSON: 24-1.0071786.json
JSON-LD: 24-1.0071786-ld.json
RDF/XML (Pretty): 24-1.0071786-rdf.xml
RDF/JSON: 24-1.0071786-rdf.json
Turtle: 24-1.0071786-turtle.txt
N-Triples: 24-1.0071786-rdf-ntriples.txt
Original Record: 24-1.0071786-source.json
Full Text
24-1.0071786-fulltext.txt
Citation
24-1.0071786.ris

Full Text

CONTROL IN S WXWU7MESH by PETER WILLIAM JACOBS  B.Th., Summit Pacific College (WPBC) 1988 M.A., University of Oregon 1992  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Linguistics)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2011 © Peter William Jacobs, 2011  Abstract This dissertation is an examination of the phenomenon of control in S wxwu7mesh (a.k.a the Squamish language). The notion of control has been part of the Salishan linguistic tradition for more than 30 years and it has been described as the ‗degree of control an agent has over an event‘ (Thompson 1979). It has been described as having one of two values: in control or limited control. An agent who is in control, is understood to initiate an event on purpose, to have control over the process of the event and to bring the event to culmination. An agent who has limited control may unintentionally initiate an event, or have difficulty in the process of the event and thus only managed to bring the event to completion. In this dissertation I argue that control is properly understood as a construct. That is, it is not a part of the basic meaning of any one morpheme. Rather it is constructed from both real world knowledge about events and from the morphosyntax of the constructions that are used to encode these events. I argue that control constructions have an aspectual core meaning. A control predicate (or c-predicate) has event initiation as its core meaning. A limited control predicate (or lc-predicate) has event culmination as its core meaning (Ritter and Rosen 2000). They are telic. I argue that it is from these two meanings - event initiation and event culmination - that the other notions commonly associated with control are inferred (e.g. on purpose, accidentally, etc.). I propose a morpho-syntactic analysis for the core aspectual difference between the two types of predicates. In particular, I argue that they differ in the position of object agreement: object agreement of c-predicates is VPinternal, while object agreement of lc-predicates is associated with an aspectual node within the extended verbal projection. I explore the consequences of this proposal for the reconstruction of Proto-Salish in general, and for the historical development of S wxwu7mesh in particular.  ii  Table of Contents Abstract ............................................................................................................................... ii Table of Contents ............................................................................................................... iii List of Tables ................................................................................................................... viii List of Figures .................................................................................................................. xiii List of Abbreviations ....................................................................................................... xiv Acknowledgements ............................................................................................................. ii Dedication ........................................................................................................................... v Prologue ............................................................................................................................. vi 1  Territory and Language.............................................................................................. vi  2  Language revitalization ............................................................................................ viii  Chapter 1 – Introduction ................................................................................................. 1 1  Introduction ................................................................................................................. 1  2  The problem with CONTROL ........................................................................................ 5 2.1  Kuipers (1967): volition ....................................................................................... 5  2.2  Thompson (1979) on CONTROL ............................................................................ 9  2.3  The aspectual character of CONTROL .................................................................. 16  3  Where is the morphological marking for CONTROL? ................................................ 26  4  Outline of my proposal ............................................................................................. 29  Chapter  -  7mesh ................................................................. 32  1  Introduction ............................................................................................................... 32  2  Linguistic classification ............................................................................................ 32  3  Methodology ............................................................................................................. 34  4  Grammatical sketch .................................................................................................. 35 4.1  The verbal template ............................................................................................ 36  4.1.1  The verbal stem ........................................................................................... 37  4.1.1.1  Roots .................................................................................................... 37  4.1.1.2  Lexical suffixes.................................................................................... 42 iii  4.1.1.3 4.1.2  Verbal prefixes .................................................................................... 44 The transitivity system ................................................................................ 48  4.1.2.1  Core transitivizers ................................................................................ 51  4.1.2.2  Applicative transitivizers ..................................................................... 58  4.1.2.3  Intransitivizers ..................................................................................... 62  4.1.2.4  The (in)transitivizers and the verbal template ..................................... 67  4.1.3  Person marking ........................................................................................... 73  4.1.3.1  Object agreement ................................................................................. 73  4.1.3.2  Subject agreement................................................................................ 79  4.1.3.3  Possessive marking .............................................................................. 84  4.2  Verb phrase auxiliaries and particles ................................................................. 86  4.3  Determiner phrases ............................................................................................. 89  4.4  Case .................................................................................................................... 91  4.5  Word order ......................................................................................................... 94  Chapter 3: The core meaning of CONTROL ................................................................... 97 1  Introduction ............................................................................................................... 97  2  Background ............................................................................................................. 100  3  Core transitives and culmination ............................................................................ 106  4  3.1  The -t transitivizer and culmination ................................................................. 107  3.2  The -Vt transitivizer and culmination ............................................................... 111  3.3  The -Vn transitivizer and culmination .............................................................. 115  3.4  The causative -s transitivizer and culmination ................................................. 118  3.5  The limited control transitivizer -nexw and culmination ................................. 123  3.6  Summary .......................................................................................................... 129  Intransitives and culmination .................................................................................. 131 4.1  The control intransitives and culmination ........................................................ 132  4.1.1  The control unergative -im and culmination ............................................. 132  4.1.2  The control reflexive -sut and culmination ............................................... 136  4.2  The limited control intransitives and culmination............................................ 141  4.2.1  The limited control unergative -nalhn and culmination ........................... 141  4.2.2  The limited control reflexive -numut and culmination ............................. 145 iv  4.3 5  Summary .......................................................................................................... 151  Applicatives and culmination ................................................................................. 153 5.1  The relational applicative -nit and culmination................................................ 154  5.2  The redirective applicative -shit and culmination ............................................ 158  5.3  The causative applicative -min and culmination .............................................. 162  5.4  The benefactive applicative - h‟ w n and culmination ................................... 164  6  Culmination implicatures ........................................................................................ 167  7  Summary ................................................................................................................. 173  Chapter 4 - The context of use of CONTROL................................................................ 181 1  Introduction ............................................................................................................. 181  2  Background ............................................................................................................. 183  3  What can limited control predicates mean? ............................................................ 189  4  3.1  Limited control transitivizer ............................................................................. 191  3.2  Limited control intransitivizers ........................................................................ 196  3.2.1  The limited control reflexive .................................................................... 197  3.2.2  The limited control unergative .................................................................. 201  3.3  Other non-control meanings ............................................................................. 206  3.4  Summary .......................................................................................................... 218  How to derive CONTROL .......................................................................................... 219 4.1  The proposal in a nutshell ................................................................................ 220  4.2  The context of use for lc-predicates ................................................................. 223  4.2.1  The event culmination meaning ................................................................ 223  4.2.2  The normal course of events does not allow for accidents ....................... 227  4.2.3  The normal course of difficulty is failure ................................................. 230  4.2.4  The normal course of events requires ability for completion ................... 236  4.2.5  Summary of limited control interpretations .............................................. 243  4.3 5  The contexts of use for c-predicates ................................................................. 244  Conclusion .............................................................................................................. 254  Chapter 5: The morphosyntax of CONTROL ............................................................... 258 1  Introduction ............................................................................................................. 258 v  2  3  4  5  6  In search of the morphological marking of CONTROL ............................................. 260 2.1  The transitivizer -t ............................................................................................ 261  2.2  The disappearing -t ........................................................................................... 262  2.3  Reanalyzing the c-transitivizers -Vn, -V-t and -t .............................................. 266  2.4  Reanalysis of the morphology of limited control transitives ........................... 276  2.5  Summary .......................................................................................................... 282  The syntax of events: an overview.......................................................................... 286 3.1  Decomposing verbs: generative semantics and beyond ................................... 288  3.2  Expanding the verb phrase ............................................................................... 290  3.3  Calculating telicity: the syntax-semantics interface ......................................... 293  The morphosyntax of CONTROL in S wxwu7mesh................................................. 297 4.1  The linearization of complex predicates .......................................................... 301  4.2  The distribution of object agreement ............................................................... 307  4.2.1  A universal object to event mapping ........................................................ 307  4.2.2  Dependent marking and word order are irrelevant ................................... 309  4.2.3  When CONTROL determines who to agree with ......................................... 312  4.2.4  Morphological differences in object agreement ....................................... 318  Alternatives ............................................................................................................. 321 5.1  CONTROL  5.2  Alternative morpho-syntactic accounts ............................................................ 326  5.3  CONTROL  is not a contrast in modality ............................................................. 330  Extending the analysis ............................................................................................ 337 6.1  The causative .................................................................................................... 338  6.2  Intransitives ...................................................................................................... 355  6.2.1  Lc-intransitivizers ..................................................................................... 356  6.2.2  C-intransitivizers ....................................................................................... 360  6.3 7  is not a contrast in perfectivity ......................................................... 322  Implications for predicate classes .................................................................... 364  Conclusion .............................................................................................................. 370  Chapter 6: Implications ................................................................................................ 373 1  Introduction ............................................................................................................. 373  2  S wxwu7mesh and the Proto-Salish object sets ..................................................... 374 vi  2.1  Proto-Salish object sets .................................................................................... 374  2.2  The development of S wxwu7mesh object sets .............................................. 377  2.3  Comparing accounts ......................................................................................... 391  2.4  Asp-set reanalyzed as spec agreement ............................................................. 394  3  Occurrence in FP-delimit ........................................................................................ 399  4  Telicity without CONTROL ....................................................................................... 405  5  Other non-control constructions in Salish............................................................... 409  6  Conclusion .............................................................................................................. 414  Chapter 7: Conclusion .................................................................................................. 417 1  Summary ................................................................................................................. 417  2  Remaining questions ............................................................................................... 421  Bibliography ................................................................................................................... 423 Appendix A: Roots and their transitivizers ..................................................................... 437 1  Zero transitivizer roots: -Ø...................................................................................... 438  2  -V transitivizer roots ................................................................................................ 441  3  -n transitivizer roots ................................................................................................ 444  vii  List of Tables Table 1  Volitional and non-volitional transitives (Kuipers 1967:77-78) ...................... 7  Table 2  Non-control labels in Salish ........................................................................... 14  Table 3  C- and lc-predicates pairs with primarily aspectual differences ..................... 18  TABLE 4  The meaning of CONTROL in S wxwu7mesh (Jacobs 2011) .......................... 26  Table 5  Volitional and non-volitional transitives (Kuipers 1967:77-78) .................... 27  Table 6  C- and lc-(in)transitivizers (adapted from Kuipers 1967) .............................. 28  Table 7  Salish language family.................................................................................... 33  Table 8  Lexical suffixes in S wxwu7mesh ................................................................. 44  Table 9  ore transitivizers in S wxwu7mesh ............................................................. 51  Table 10  Applicative transitivizers in S wxwu7mesh .................................................. 59  Table 11  Transitivizers and applicatives with/without -t .............................................. 61  Table 12  ore intransitivizers in S wxwu7mesh .......................................................... 62  Table 13  Intransitivizers not investigated ...................................................................... 66  Table 14  Templatic distribution of transitivizers........................................................... 72  Table 15  Templatic distribution of intransitivizers ....................................................... 72  Table 16  S wxwu7mesh object agreement suffixes (Kuipers 1967:85) ....................... 73  Table 17  S wxwu7mesh object agreement suffixes - revised (Jacobs 2011) ............... 79  Table 18  Types of subject/object agreement in S wxwu7mesh.................................... 80  Table 19  Nominative subject clitics .............................................................................. 81  Table 20  onjunctive subject clitics in S wxwu7mesh ................................................ 83  Table 21  Possessive marking in S wxwu7mesh ........................................................... 84 viii  Table 22  The determiner system of S wxwu7mesh (adapted from Gillon 2009) ......... 90  Table 23  The demonstrative system of S wxwu7mesh (adapted from Gillon 2009) ... 90  Table 24  ase marking for Ps in S wxwu7mesh ....................................................... 91  Table 25  Culmination cancellation and event continuation (Bar-el 2005:135, ex. 137) .. ...................................................................................................................... 102  Table 26  Event Cancellation vs. Event Continuation (Bar-el 2005:136, ex. 138) ...... 103  Table 27  S wxw 7mesh predicates: initial and final points (Bar-el 2005:200) ......... 104  Table 28  ore transitivizers in S wxwu7mesh ........................................................... 106  Table 29  C-transitive (-t) and the culmination entailment tests ................................. 110  Table 30  C-transitive (-t) and the scopal tests ............................................................. 110  Table 31  C-transitive (-Vt) and the culmination entailment tests ................................ 114  Table 32  C-transitive (-Vt) and the scopal tests ........................................................... 114  Table 33  C-transitive (-Vn) and the culmination entailment tests .............................. 117  Table 34  C-transitive (-Vn) and the scopal tests .......................................................... 118  Table 35  C-transitive (-s) and culmination entailment tests ........................................ 122  Table 36  C-transitive (-s) and the scopal tests ............................................................. 122  Table 37  Lc-transitive (-nexw) and culmination entailment tests ............................... 128  Table 38  Lc-transitive (-nexw) and the scopal tests..................................................... 128  Table 39  C-transitives, lc-transitives and the culmination entailment tests ................ 129  Table 40  C-transitives, lc-transitives and the scopal tests ........................................... 129  Table 41  Distribution of culmination entailments and inherent final points ............... 130  Table 42  C- and lc-intransitivizers .............................................................................. 131  Table 43  C-unergative (-im) and the culmination entailment tests.............................. 135 ix  Table 44  C-unergative (-im) and the scopal tests ........................................................ 136  Table 45  C-reflexive (-sut) and the culmination entailment tests................................ 140  Table 46  C-reflexive (-sut) and the scopal tests .......................................................... 140  Table 47  Lc-unergative (-nalhn) and the culmination entailment tests ...................... 144  Table 48  Lc-unergative (-nalhn) and the scopal tests .................................................. 144  Table 49  Lc-reflexive (-numut) and the culmination entailment tests ......................... 149  Table 50  Lc-reflexive (-numut) and the scopal tests ................................................... 150  Table 51  Intransitivizers and the culmination entailment tests ................................... 151  Table 52  Intransitivizers and the scopal tests .............................................................. 151  Table 53  Intransitivizers, culmination entailments and inherent final points .............. 152  Table 54  C- and lc-intransitives and the tried to interpretation ................................... 152  Table 55  pplicative transitivizers in S wxwu7mesh ................................................ 153  Table 56  The relational applicative (-nit) and the culmination entailment tests ......... 157  Table 57  The relational applicative (-nit) and the scopal tests .................................... 157  Table 58  The redirective applicative (-shit) and the culmination entailment tests ...... 161  Table 59  The redirective applicative (-shit) and the scopal tests................................. 161  Table 60  The causative applicative (-min) and the culmination entailment tests ........ 163  Table 61  The causative applicative (-min) and the scopal tests................................... 164  Table 62  The benefactive applicative (- h‟ w n) and the culmination entailment tests ... ...................................................................................................................... 166  Table 63  The benefactive applicative (- h‟ w n) and the scopal tests ....................... 167  Table 64  C-predicates and culmination implicatures .................................................. 173  Table 65  Summary of culmination entailments and implicatures ............................... 174 x  Table 66  C- and lc-intransitives and the tried to interpretation ................................... 176  Table 67  I-languages and D-languages (Ritter and Rosen 2000) ................................ 177  Table 68  I-languages and -language behaviour in S wxwu7mesh........................... 177  Table 69  Compatibility of non-control interpretations ................................................ 190  Table 70  The interpretations for lc-transitives and their linguistic contexts ............... 196  Table 71  The interpretations of lc-reflexives and their linguistic contexts ................. 201  Table 72  The interpretations of lc-unergative and their linguistic contexts ................ 206  Table 73  Non-control interpretations: Lillooet and S wxwu7mesh............................ 218  Table 74  Lc-markers, their interpretations and restrictions in S wxwu7mesh ........... 219  Table 75  C- and lc-predicates and their context of use ............................................... 256  Table 76  Culmination entailments: -t or no -t ............................................................. 261  Table 77  Occurrences of limited control marking in S wxwu7mesh ......................... 277  Table 78  Occurrences of limited control marking in Sechelt (Beaumont 1985) ......... 279  Table 79  S wxwu7mesh core transitivizers: Kuipers (1967) ...................................... 283  Table 80  S wxwu7mesh applicative transitivizers: Kuipers (1967) ........................... 283  Table 81  S wxwu7mesh core transitivizers: reanalyzed (Jacobs 2011) ..................... 284  Table 82  S wxwu7mesh applicative transitivizers: reanalyzed (Jacobs 2011) ........... 284  Table 83  The revised morphological template for core transitivizers ......................... 298  Table 84  VP Object Agreement Set (VP-set) .............................................................. 319  Table 85  Aspect Object Agreement Set (Asp-set) ....................................................... 319  Table 86  Culmination entailments: predicates with -t or no -t .................................... 338  Table 87  C- and lc-intransitivizers (Jacobs 2007) ....................................................... 355  Table 88  C- and lc-intransitivizers - revised (Jacobs 2011) ........................................ 356 xi  Table 89  S wxw 7mesh predicates: their initial and final points (Bar-el 2005:200) . 365  Table 90  S wxw 7mesh predicate classes with thematic and aspectual roles ............ 366  Table 91  Comparing c-unergative and c-transitive (Jacobs 2011) .............................. 367  Table 92  Proto-Salish object sets ................................................................................. 375  Table 93  Proto-Salish object sets (revised) (Kroeber 1999:25)................................... 376  Table 94  Proto-Salish object sets (Jacobs) .................................................................. 378  Table 95  Three analyses of PS object sets ................................................................... 379  Table 96  S wxwu7mesh object sets and intransitivizers ............................................ 380  Table 97  Proto-Salish and S wxwu7mesh object and intransitivizers: Asp-set.......... 383  Table 98  Proto-Salish and S wxwu7mesh objects and intransitivizers: VP-set ......... 384  Table 99  PS and PSQ with schwa reduction ............................................................... 385  Table 100  PSQ and -w deletion ................................................................................. 386  Table 101  PSQ -w reanalyzed as part of object marker ............................................. 387  Table 102  PSQ second person plural form innovated for the Asp-set ....................... 388  Table 103  PSQ first plural, second singular and plural innovation for VP-set .......... 389  Table 104  Causative reflexive reanalyzed ................................................................. 390  Table 105  Stages with the Asp-set ............................................................................. 391  Table 106  Stages for c-transitive with the VP-set...................................................... 391  Table 107  Stages for the causative with the VP-set ................................................... 392  Table 108  Stages for the causative – Newman‘s account .......................................... 393  xii  List of Figures Figure 1  The normal course of events: past-perfective context .................................. 225  Figure 2  The normal course of events: future tense .................................................... 226  Figure 3  The final event was not intended .................................................................. 228  Figure 4  We can‘t predict accidents ............................................................................ 230  Figure 5  Difficulty normally results in failure ............................................................ 233  Figure 6  We can‘t predict unusual circumstances ...................................................... 234  Figure 7  Overcoming difficulty creates confidence .................................................... 234  Figure 8  Overcoming difficulty creates ability in the future - Context 1 ................... 238  Figure 9  Overcoming difficulty creates ability in the future - Context 2 ................... 239  Figure 10  Activities only have initiation ................................................................... 240  Figure 11  Confidence in the present tense ................................................................ 241  Figure 12  Negation cancels event culmination ......................................................... 242  Figure 14  C-predicates assert the initial event .......................................................... 247  Figure 15  Lc-predicates assert the final event ........................................................... 248  xiii  List of Abbreviations 7mesh data 1 2 3 APPL AUX BENAPPL  cCREFL CRECIP CUE CAUS CAUSAPPL CONJ DEM DET ERG  fe FUT IMPF  ie INDP INTR  lcLCTR LCRECIP LCREFL MID NEG NOM OBJ OBL PART PASS PAST PL POL POS RC RE RECIP RED  REDAPPL REFL RELAPPL RL S SBJ SUB TR UE  first person second person third person applicative auxiliary benefactive applicative control control reflexive control reciprocal control unergative causative transitivizer causative applicative conjunctive demonstrative determiner ergative final event future tense imperfective aspect initiating event independent pronoun intransitivizer limited control limited control transitivizer limited control reciprocal limited control reflexive middle negation nominalizer object oblique case particle passive past tense plural polarity question marker possessive relative clause marker CV- reduplicant reciprocal CeC- reduplicant  redirective applicative reflexive relational applicative realis singular subjunctive subject transitivizer unergative  Finnish data 1SG ACC PART PAST PL  first person singular subject accusative case partitive case past tense plural  Halkomelem data 1 3 AUX CAUS DET LCTR NEG REFL S SUB TR  first person third person auxiliary causative transitivizer determiner limited control transitivizer negation reflexive singular subject control transitivizer  Hindi data ACC ERG PERF  accusative ergative perfective  Lillooet data 1 3 ACT ADD CAUS  xiv  first person third person active intransitivizer additive causative transitivizer  CIRC COMP CONJ DET DIR ERG EXIS FOC IMPF IRR NEG NOM PL POSS SG SUBJ YNQ  circumstantial modality complementizer conjunctive subject determiner directive transitivizer ergative existential focus predicate imperfective aspect irrealis negation nominalizer plural possessive singular subject yes/no question marker  Saanich data 1 ACC AUX CTR D INF NEG NCTR NOM SG  Sechelt data TR 3OBJ  DET LC NEG PAST PRES  1 3 CNJ CTR ERG INDC  third person plural determiner lexical causative negation past tense present tense  LINK NEG NTR PAST SG  Lushootseed data 1S.OBJ LCTR PASS TR  transitivizer third person object  Sliammon data  Malagasy data 3P  first person accompanying auxiliary control transitive demonstrative/determiner informative negation non-control transitive nominalizer singular  first person singular object limited control transitivizer passive transitivizer  first person third person conjunctive subject control transitivizer ergative subject indicative subject link vowel negative non-control transitive past tense singular  Thompson data 3OBJ DIR TR  ii  third person object directive transitivizer transitivizer  Acknowledgements  n stl‘i7 kwins kw‘enmantan i7xw kwetsiwit na ch‘awats tin sts‘its‘ p‘. Texwlam eskw‘ y kwins p‘ 7nexwan ta est txw sn chim tich m an. U haw k‘ap ch nchensts na eskw‘ y kwins es p‘nexwan. hen kw‘enmantumiyap i7xw ta newyap. I entered the PhD program at UBC almost by accident. One day my good friend, Mandy Jimmie and I were having lunch at large meeting out at Chilliwack when she said, ―I‘m thinking of going to do my Ph ‖ to which I responded ―that sounds like a good idea!‖ nd then I looked around to see who has just said that not believing that it was me. Rose-Marie chaine had already been trying every year to get my to go back and do a Ph . Thank you Mandy for ―tricking‖ me and thank you Rose-Marie for never letting up. It paid off. I also became interested in returning to school, in particular UBC, because of the professional relationship that I have had with a number of people for the last fifteen plus years. At one point Leora Bar-el, Carrie Gillon, Linda Watt, Martina Wiltschko and I were are all co-authoring an article. I had very limited exposure to Chomskian linguistics to this point and so I did not understand much of the discussion around the formulations of theory. I remember, though, Martina, saying to me that even if the current theory is proven wrong (which I will be), it still made us ask whole lot of new and interesting questions about language. This statement helped in getting me back to do a PhD. I would like to begin by thanking my committee members: Supervisor, Martina Wiltschko, and committee members, Henry Davis and Lisa Matthewson. As many asked me ―what‘s in it for your committee (devoting so much time to your dissertation)?‖ I know that it‘s in part because we share the same love of language and care for the communities that speak them. Martina always knew to let me go where I was going in my thoughts and then bring me back to the hard work of writing those thoughts down. I‘ve learned a new language to talk about linguistics in and it‘s opened up a whole ‘nuther world of cool questions. Henry brought a wide knowledge of the Salish family and, although we ended up disagreeing on some of my analysis, it was his questions that got me to rethink Skwxwu7mesh morphology. Lisa taught me that it‘s possible and necessary, for linguists to communicate their ideas to those different schools of thought. It‘s worth the effort and it increases the quality of the work. hen wa k‘ay‘ chten ta newyap n-nexw7uts ylh (I lift my hand up to you, my teachers). I would also like to thank my defence examiners: Rose-Marie chaine, Donna Gerdts, and Pat Moore. The time you took with your comments both before and after the defence helped me greatly in clarifying my work. I would like to acknowledge all elders with whom I‘ve work with both before and after starting my Phd. The following elders had already left us by the time I had started my dissertation: Lena Jacobs, my grandmother, who taught me how to converse like our ancestors; the prayer group: Auntie Nora Desmond, having lived away from the community for decades, taught me that we are all responsible to maintain our language; ii  Auntie Eva Lewis taught me that a teacher is always learning; Auntie Doris Williams taught me that linguistic knowledge is part of our cultural heritage; Auntie Yvonne Joseph taught me that a good teacher listens; Uncle Lawrence Baker taught me humour in our language, and spent countless hours answering my questions; Auntie Tina Cole taught me that our ancestors knew that it is our duty to teach all of our communities children our language; Uncle Ernie Harry taught me that humour is at the core of our language and who we are; it cannot be lost!; Uncle Frank Guerrero, as he spoke our language, showed me how much our language is a window to our own history; Uncle Alvie Andrews taught me that history starts with knowing the place names of where you come from; Uncle Frank Miranda was patient enough to teach me the many ways to say the same thing; besides teaching us all the language, they showed me how to always be grateful and let others know that you were. The elders whom I worked on this dissertation with include; Auntie Margaret Locke who taught me that an educated person is one who knows their language well; Auntie Addie Kermeen who still teaches me about our family and how they spoke; Uncle Alec Williams, who teaches me about parts of our Skwxwu7mesh history that I know little of; Auntie Stella who, through her story telling and her humour, introduced me to many of my great aunts and uncles whom I never saw on this side; we were never done our elicitation sessions until we had laughed. I would also like to thank my co-workers. First, thank you to Deborah Jacobs, my boss for supporting my education all these years, both in administrative role and as a fellow academic, and as a cousin and friend; to Rosalind Williams for the financial support throughout this long haul (the laptop came just in time!). To my fellow language teachers: an chap yew n. Vanessa Campbell was like an unofficial defence reader for my dissertation since I ran everything by her that I was thinking about. Alroy Baker and Val Moody, for getting me interested in reviving our language in the first place. Kathy and Chantel, for asking me the hard questions that forced me to thinker harder about how to explain things more clearly; Becky Cambpell, Tsetsiy xemts, for loving the learning of our language. Ray Natrarro, for taking everything you learn and offering it back to the people. Kirsten, Kwi n-Sna, Baker-Williams, fellow academic, friend, and being the person to get lost in Chicago Ohare with. To Valtenat, for leading by example and reclaiming your language. To Norm Guerrero, Jr., Sxwman, you were our student, and now you‘re our co-worker, yewan ha7lh. To cousin Angie Dawson, Kim Seward, Tracy Williams for giving me time to decompress and underscoring the real importance of our children learning their language. From the UBC side of things, I first of all have to thank all of you friends from Thesis nonymous. I‘m cured! Our discussions made me look deeper into the issues and come out with a better understanding of them. Thank you to Solveiga Armoskaite (brothers in arms), Heather Bliss (fellow hiker), Christie Christodoulou (encourages ourselves back to the bus stop), Atsushi Fujimori (another co-finisher), Olga Steriopolo (who was the first to show us it can be done), Sonja Thoma (foreign student, mother of twins, and a PhD candidate!), James Thompson (fellow Coast Salishanist, coinvestigator, and music lover). Edna Dharmaratne, you made being in school a safe and fun place from day one. iii  To all my fellow-Skwxwu7mesh-anists, upon whose work I have built, (in order of your appearance), thank you for your friendship and for being the excellent examples of academic collegiality: Hamida Demirdache (for having joy in Salish linguistics), Elizabeth Currie (for providing an excellent start to our joint collaboration between Skwxwu7mesh and UBC), Lalitadevi (Linda Watt, for helping us get out of the box in our elicitations), Leora Bar-el (for the best - and only ones I‘ve been to - Hannukah parties, and also for providing a wider foundation for all of us on Salish aspect), Carrie/Terrie Gillon (for showing up back in Vancouver just at the right time to let me know that it is not necessary to enjoy writing a dissertation at all times but that it‘s totally doable) and James Thompson (for being a real gentleman and letting me finish ahead of him in our race to finish). Also thank you to Strang Burton for being an unofficial committee member, for keeping my research grounded in the language revitalization goals that we all share. To the many other people I have met and got to know at UBC linguistics: Jennifer Able, Alexis Black, Jason Brown, Marion Caldecott, Mario Chavez-Peon, Yunhee Chung, Clare Cook, James Crippen, Donald Derrick, Joel Dunham, Hudu Fusheini, Jen Glougie, Hannah Greene, Analia Gutierrez, Yoko Ikegama, Kristin Johannsdottir, Karsten Koch Patrick Littell (ok‘ala ika walhdamus) John Lyon Stacey Menzies Jeff Muehlbauer Masaki Noguchi Martin Oberg Tyler Peterson ominique Quis eth Rogers Mark Scott m lia Silva, Anita Szakay, Audra Vincent, Ryan Waldie, Rachel Wojdak, Noriko Yamane-Tanaka. To my non-linguists friends who supported me, often listening to problems they had no idea about chen kw‘enmantumiyap: Gloria Tina Gungun erik Warren Walt Tony, Dave, Neil, Greg, Jay, Hash (Sasquatch and Yeti unite!), Mike and Jacqueline, and Donavin (who listened untiringly for all those years). Dr. McWhinney for a listening ear. To all my aunts and uncles and cousins, too many to list in one dissertation, many of whom have committed themselves to a lifetime of education. I love you all and ―huy chap‖. The defence luncheon was amazing because of all of you. To my brothers and sisters, George and Vanessa, Carla and Brian, you always provided a safe place to just be me. That kept me grounded. To my nieces and nephews, Katherine, Kyle, Brian and Shana, your unconditional love makes me feel so very light. To my parents T‘naxwtn iy Hamdzidi who wouldn‘t let me settle for anything less than the best that I can do, and supported my many, many, many years of school. Dad put into my mind the question ―how does this help the people?‖ and mom gave me a love of language.  .  iv  Dedication I dedicate this dissertation to my grandparents:  Adagilakw, a natural linguist and teacher. like his grandfather, George Hunt, before him, Granny Lakwa, who h d “univ rsity d gr in th Indi n l ngu g ”, Grampa Alfred, who taught us all to work hard and enjoy our family, Granny Lena, I wish you were on this side to see me finish. You taught me how to converse like our ancestors in the language, joined me in learning more about the language, and showed me that ta chim is all of our inheritance.  v  Prologue  I begin here with an introduction to the context in which this dissertation was written. First, I provide a S wxwu7mesh (a.k.a. Squamish) viewpoint of the S wxwu7mesh Snichim (a.k.a. the Squamish language). Second, I provide a brief description of the community of S wxwu7mesh people and our language revitalization efforts for the S wxwu7mesh language.  1  Territory and Language  Ta S wxw 7mesh snichim (or just S wxwu7mesh) is the ancestral language spoken by the S wxw 7mesh people, whose traditional territory extends west from Stelk ya (Roberts reek) south down to lksen (Point Grey) up through Sel lwetulh (the head of Indian Arm), and then again north up through tl‘a 7tsem (Howe Sound), up the headwaters of the Skwxw 7mesh Stakw (Squamish River valley) and also up to the headwaters of the h‘iy mesh Stakw (Cheakamus River) (Squamish Nation Dictionary Project 2011). This territory is situated in present day southwestern British Columbia, Canada. S wxw 7mesh is most closely related to other Salish languages spoken on the southern coast of British Columbia and around Puget Sound, Washington. The following  vi  is a S wxwu7mesh-centric viewpoint on the origin of these relationships.1 In the S wxwu7mesh syets2 („story‟) of the lood one group of S wxw 7mesh people is said to have been stranded on the top of  h‟ y (Mt. Garibaldi) during the Great Flood and  survived to be the ancestors of the present day S wxw 7mesh people. Dr. Louis Miranda, a S wxwu7mesh elder and speaker, tells in his version of this syets that during the lood another group split off from the group on Nch‘ ay. This other group landed on ws  (Mt. Baker, in present day Washington State). It is from these people that the  present day Xwsa7 people (Nooksack) have descended. In another account of the Flood story, told by S wxw 7mesh elder Dominic Charlie, he states that other groups also broke off from the Nch‘ ay group. These groups become the other groups of Salish speaking peoples living around Puget Sound (i.e., the Lushootseed speaking peoples, the Twana and the Klallam). Besides the account of the Flood stories, other S wxwu7mesh stories tell of how other Coast Salish speaking peoples, such as the speakers of Coast Salish languages spoken on southern Vancouver Island, are related to the S wxwu7mesh (i.e., the ‘ew chen dialect of S ‘em nem, or the Cowichan dialect of Halkomelem). One such story is recorded in Kuipers (1967) as the legend of a migration, and it was told to him by Dr. Miranda. In this account one family unit moved from h‘ w‘elhp (Gibson‘s Landing, said to be the first settlement of the S wxwu7mesh) across the Strait of Georgia to various places on southern Vancouver  1  A linguistic viewpoint of the Salish language family will be given in Chapter 2, §2. Syets is the S wxwu7mesh word for ‗historical story‘. This contrasts with s story from the myth time. 2  vii  , the word for a  Island. These people are the ancestors of the Coast Salish peoples living on southern Vancouver Island.  2  Language revitalization  The research, documentation and analysis of S wxwu7mesh reported in this dissertation is part of the larger effort in language revitalization by the S wxwu7mesh Nation. There are less than 10 first language speakers of S wxwu7mesh and the S wxwu7mesh Nation has been working to revive the use of the language in the community. Baker-Williams (2006) provides an examination of the history of this language revival, as well as some context for the loss of the language in everyday use in the community (e.g. residential school, public school policies, economic and religious factors, etc.). This research is also situated in the greater movement of language revitalization for First Nations languages in British Columbia and the rest of Canada. The first efforts at language revitalization began with a number of S wxwu7mesh language elders, in the 1960‘s, and for over forty years now the language has been taught in the community. S wxwu7mesh elder Dominic Charlie began teaching in the village of St‘ 7mes (a.k.a. Stawamus), then later Dr. Louis Miranda began teaching in the 1970s and 1980s. Uncle Louis (as he was affectionately known in the S wxw 7mesh community) also tirelessly hand-wrote hundreds of pages of word lists, stories, legends, language lessons and personal history. These writings were in S wxwu7mesh and often glossed interlinearly in English. They remain as unpublished manuscripts in the possession of the Department of Education. Dr. Miranda received an honorary doctorate viii  at Simon Fraser University for this work. Much of the linguistic documentation of the S wxwu7mesh language by Kuipers (1967, 1969) was done with Dr. Miranda. The S wxwu7mesh revitalization at present involves S wxwu7mesh language classes being taught at the local public high school in North Vancouver, BC (Carson Graham High School) and in a number of elementary schools in North Vancouver and in the district of Squamish. The S wxwu7mesh Nation also has its own school for 3-5 year olds called Xwmelch‘sten tsimxw‘awtxw (Capilano Littlest Ones School), where a bilingual-bicultural program provides S wxwu7mesh language curriculum. The group of people with whom I conducted fieldwork for the research of this dissertation resides within the traditional territory of the S wxw 7mesh Nation. They are called  , which is translated as Teachings for Your  Grandchildren. As the name implies, the teaching of language is intimately tied to the passing on of traditional S wxwu7mesh  (‗the values and teachings‘). This  group was formed in 1993 as a consultation group for the S wxw 7mesh Language Program, a program under the Department of Education of the Squamish Nation. This group is now the official language authority for the S wxw 7mesh language. Since its inception, a number of other language speakers have joined. Since this group began, we have together worked towards publishing a dictionary (Squamish Nation Dictionary Project 2011) and developing pedagogical materials. This group has also participated as consultants in hundreds of hours of elicitation sessions with graduate students from UBC and other universities. One of the main goals of this group has been to be as inclusive as possible of all S wxwu7mesh language speakers in the S wxwu7mesh Nation. For those who chose not to join this group directly, we would ix  instead make home visits with them to learn from them. My role in these language revitalization efforts is as a member of the S wxwu7mesh Language Program. My job has included linguistic research, teaching and program development. The linguistic research serves to provide a stronger linguistic basis for these efforts (cf. Miller 2005 for issues around developing standards and evaluation for First Nations language programs). Present efforts, besides the in-school programming, include the development of a teacher training program to train more S wxwu7mesh language teachers. One of the primary goals for this dissertation, then, is, as a S wxwu7mesh st lmexw (Squamish human being) to strengthen these efforts (Smith 1999, Wilson 2008).  x  Chapter 1 – Introduction  1  Introduction  The original question for this dissertation was as follows. S wxw 7mesh (like other Salish languages) is known for the pervasive marking of control. In particular, a transitive predicate is marked differently depending on the degree of control the agent has over the event (Thompson 1979, see also Bar-el 2005, Jacobs 2007).3 Consider the examples in (1).  (1)  a.  chen kw‘lh-at-Ø ta tiy 1S.SUB pour-TR-3OBJ DET tea ‗I poured the tea.‘ (on purpose)  b.  chen kw‘ lh-nexw-Ø ta tiy 1S.SUB spill-LCTR-3OBJ DET tea ‗I spilt the tea.‘ (accidentally)  Both verbs are complex in that they consist of the verbal root (kw‟ lh) and a transitivizer (-at and -nexw, respectively).4 The difference in the two transitivizers is typically described as marking a difference in degree of control. In (1)a, -at triggers an  3  In this dissertation I assume that an agent is the participant who causes an event or change of state in another participant (Dowty 1991:572). 4 The schwa of this root is reduced to zero when it occurs pretonically as in (1)a.  1  interpretation where the agent had a normal level of control over the event. Thus -at is sometimes referred to as the control transitivizer (henceforth c-transitivizer). In the English translation, predicates with a c-transitivizer (henceforth c-predicates) can be explicitly marked by an adverb such as on purpose. I will refer to this as the on-purpose interpretation. In contrast -nexw in (1)b triggers an interpretation where the agent has less than normal control over the event. Thus, -nexw is sometimes referred to as a limited control transitivizer (henceforth lc-transitivizer). Again, in the English translation, predicates with the lc-transitivizer (henceforth lc-predicates) can be explicitly marked by an adverb such as accidentally. I will refer to this as the accidental interpretation. This description of c- and lc-predicates is deceptively simple. That is, in light of the data in (1), one might hypothesize that c- and lc-transitivizers introduce the agent argument and simultaneously mark the agent‘s volition. This analysis is supported by the fact that the simplex verb kw‟ lh lacks an agent argument, as illustrated in (2).  (2)  kw‘elh ta tiy RL spill DET tea ‗The tea spilt.‘ na  While this analysis in terms of agent volition may capture the contrast in (1), it cannot account for the full range of data. In particular, both c-predicates and lc-predicates may be used for an agent which lacks volition. For c-predicates this is shown in (3)a where it is the wind is the cause of the door shutting, but it is not a volitional causer. Thus there is  2  no volitional agent involved in the first place. The same event may also be expressed with an lc-predicate as in (3)b.  (3)  a.  b.  p‘- -Ø-as RL shut-TR-3OBJ-3SUB ‗The wind shut the door.‘ na  ta DET  sp h m ta shew lh wind DET door  p‘-nexw-Ø-as ta sp h m ta shew lh RL shut-LCTR-3OBJ-3SUB DET wind DET door ‗The wind shut the door.‘ na  Furthermore, lc-predicates can be used even if the agent appears to be fully volitional, as in (4). Here the agent is fully intending to shoot the bottle and there is nothing accidental about the coming about of the event.  (4)  chen kw lash-nexw-Ø ta nexwl may 1S.SUB shoot-LCTR-3OBJ DET bottle ‗I managed to shoot the bottle.‘ Context: the subject is practice-shooting bottles  This much establishes that the semantic contrast between c- and lc-predicates is not as straightforward as it may seem at first sight. In fact, as I will show immediately below in §2, despite a few attempts to come to terms with it, the notion of control remains elusive. This then defines the central question to be addressed in this dissertation: how do we account for the interpretative distinction associated with the notion of control in S wxwu7mesh? 3  The goal of this dissertation is to argue that the notion of control is a construct rather than a primitive notion. By construct I mean that no particular morpheme in the control constructions explicitly encodes control/limited control, but rather, control interpretations are constructed from the interaction of the morphological marking on the predicate and pragmatic inferences based on i) what we know about the world, and ii) what speakers of S wxwu7mesh know about the morphological contrasts available in the language. Thus I argue that control/limited control is not a necessary property of the agent argument of either a c-predicate or lc-predicate. More precisely, I argue that the contrast between c- and lc-predicates is essentially aspectual in nature. The core thesis here is that lc-predicates encode that the natural endpoint encoded in the verb is in fact the actual endpoint of the event that occurred (or will occur in the future), but c-predicates do not. Furthermore, a c-predicate minimally encodes that its event has been initiated. Before I turn to a discussion of previous attempts to understand the notion of control, a note on terminology is in order. Despite the fact that I understand control to be a construct, I will nevertheless use it as a descriptive term because it has been so widely assumed in the previous literature. In particular, I will use the term CONTROL in small caps to refer to the general contrast which appears to pertain to the degree of control the agent has over the event. I will use the prefixes c- and lc- to refer to control and limited control, the two possible values associated with the notion of CONTROL.  4  2  The problem with CONTROL  The purpose of this section is to examine the literature on CONTROL in S wxwu7mesh as well as in Salish more generally. At the same time, this review will establish i) that CONTROL  cannot be equated with volitionality; ii) that control is essentially aspectual in  that it is concerned with whether a predicate‘s natural endpoint coincides with the actual endpoint of the given event — i.e., CONTROL is about event (non)culmination; iii) that the impression that we are dealing with degrees of control an agent has over an event arises because of what we know about the usual course of events.  2.1  Kuipers (1967): volition  As indicated above, CONTROL-marking appears to be tied to transitivizers in S wxwu7mesh. The first systematic description of transitivizers in S wxwu7mesh is found in Kuipers (1967). While he recognizes a systematic contrast between two sets of transitivizers, he does not couch his description in terms of CONTROL, but rather in terms of volitional vs. non-volitional. Although Kuipers does not provide a definition for volitional, for purposes of the present discussion I will assume that volition is ‗the act of using the will‘ (Neufelt 1997:1496). We have already seen, however, that CONTROL cannot reduce to volitionality. On the one hand, c-predicates can be used in the absence of a volitional agent, as in (5) where we have the wind as the agent of the event. On the other hand, lc-predicates can be used in the presence of a volitional agent as in (6)a-c. In (6)a the agent chose to shoot the bottle and did. In (6)b the agent chose to take all of the children home and he did. In (6)c the man chose to kill the bear and he did. 5  (5)  (6)  p‘- -Ø-as RL shut-TR-3OBJ-3SUB ‗The wind shut the door.‘ na  ta DET  spah m ta shew lh wind DET door  a.  chen kw lash-nexw-Ø ta nexwl may 1S.SUB shoot-LCTR-3OBJ DET bottle ‗I managed to shoot the bottle.‘ Context: the subject is practice-shooting bottles  b.  nam hiy m-nexw-Ø-as 7xwaxw go get.home-LCTR-3OBJ-3SUB all ‗He [T‘it‘k 7tstn] went and took all of them [the children] home.‘ (Kuiper 1967:221)  c.  na  w‘ y-nexw-Ø-as ta sw 7 a ta m xalh kill-LCTR-3OBJ-3SUB DET man DET bear ‗The man (has) killed the bear.‘ (Kuipers 1967:169) RL  The examples in (5) and (6) demonstrate that the difference between c- and lc-predicates does not reduce to a difference of volitionality. Interestingly, even the definitions that Kuipers provides to illustrate the lc-predicates (his non-volitional predicates) do not always appear to indicate that the agent is nonvolitional (or that the agent had no or only limited control over the event). Consider the examples listed in Table 1. We observe that the same root (which may or may not be used in isolation, as shown in the first column) can be used with a c-transitivizer (as shown in the second column) or with an lc-transitivizer (as shown in the third column). Note that Kuipers‘ translation of the lc-predicates is ―have X-ed‖. There does not appear to be any kind of non-volitionality or limited control involved. If anything, the interpretation appears to be perfect(ive) aspect.  6  (7)  root --  c-transitive p‘i7-t ‗to take, to grab‘  lc-transitive p‘ 7-nexw ‗to have, to hold‘  (8)  --  ch‘ w-at ‗to help‘  ch‘ w-nexw ‗to have helped‘  (9)  sum ‗to smell, to s m‗to smell, to give off odor‘ sniff‘5 lhaw ‗to run away‘ lhaw-s ‗to elope with‘  (10) Table 1  s m-nexw ‗to smell‘  lh w-nexw ‗to have eloped with‘ Volitional and non-volitional transitives (Kuipers 1967:77-78)  Further note that Kuipers describes four transitivizers as being c-transitivizers: -t, -Vt, -un, and –s, in examples 7-10 respectively. The first three c-transitivizers, in (7)-(9) he simply labels as transitivizers, while the fourth transitivizer -s, in (10) he describes as the causative. He does not provide sentence examples for these c- and lc-transitivizers in his section on transitivizers, but the following sentence examples are from elsewhere in the grammar and also from his dictionary section.  or comparison‘s sake I also provide both  the c-predicate and lc-predicate version of the same root. Note that the majority of Kuipers‘ (1967) sentence examples do not have the ‗have X-ed‘ translation but instead have the ‗accidentally‘ translation. I provide my own example for the causative version (14)a of the root hiy m ‗to get home‘ since Kuipers only has the lc-predicate version of this root.  5  This form is found in Kuipers (1969:62).  7  (11) a.  b.  (12) a.  b.  (13) a.  b.  (14) a.  -t transitivizer na p‘i7-t-s-as RL take-TR-1S.OBJ-3SUB ‗He grabbed me.‘ (Kuipers 1967:253) -nexw transitivizer chen p‘ 7-nexw-Ø 1S.SUB get-LCTR-3OBJ ‗I am holding it, I have caught it, I got it (note that the seizure was an act of the will the resulting holding or having is not).‘ (Kuipers 1967:69) -Vt transitivizer chen wa ch‘aw-at-umi 1S.SUB IMPF help-TR-2S.OBJ ‗I am helping you.‘ (Kuipers 1967:321) -nexw transitivizer n-u chen wa ch‘aw-n-umi RL-POL 1S.SUB IMPF help-LCTR-2S.OBJ ‗ m I of assistance to you?‘ (Kuipers 1967:320) -Vn transitivizer s m- -Ø-ka smell-TR-3OBJ-IMPER ‗Smell it!‘ (Kuipers 1969:62) -nexw transitivizer chen s m-nexw-Ø 1S.SUB smell-LCTR-3OBJ ‗I smell it.‘ (Kuipers 1967:304) -s causative chen hiy m-s-Ø 1S.SUB get.home-CAUS-3OBJ ‗I brought him home.‘  8  b.  -nexw transitivizer nam hiy m-nexw-Ø-as 7xwaxw go home-LCTR-3OBJ-3SUB all ‗He went and took all of them [the children] home.‘ (Kuipers 1967:221)  While Kuipers presents corresponding c-transitivizer and lc-transitivizer forms for each root (or in his terms: volitional and non-volitional transitives), the lc-transitive forms (12)b, (13)b, and (14)b do not have non-volitional translations. Only in example (11)b does Kuipers attempt to provide a possible context for the lc-predicate to indicate a nonvolitional meaning. He offers that while the choice to seize was volitional, the resulting state of holding is not volitional. Accepting such a contextualization of events, though, should allow any volitional event which has a resulting state to potentially be compatible with either an lc-predicate or a c-predicate, since all that matters is the non-volitionality of the resulting state. In sum, Kuipers‘ (1967) description of volitional/non-volitional transitivizers does not properly characterize the full range of meanings that he describes as being associated with CONTROL-marking.  2.2  Thompson (1979) on CONTROL  Thompson, in his seminal (1979) article, was the first to discuss the relevance of the notion of control for Salish. Even though I will argue that CONTROL is a construct and no particular morpheme encodes control as such, the definition of control and non-control provided by Thompson and Thompson (1992:52) is still vital for the understanding of the use of CONTROL-marking. These definitions are given in (15) and (16) below. 9  (15) Control ― ontrolled situations are those in which the agent functions with usual average capacities in keeping things under control.‖ (16) Non-control i) can be ―events which are natural spontaneous-happening without the intervention of any agent ‖ or ii) can be events which are ―unintentional accidental acts ‖ or iii) can be ―limited control [which is] intentional premeditated [events] which are carried out to excess, or are accomplished only with difficulty, or by means of much time special effort and/or patience and perhaps a little luck.‖  These definitions capture the examples we have already seen as follows (relevant examples are repeated below). In example (17)a the c-predicate appears to indicate that the agent is functioning ―with usual average capacities‖ and as such falls under the definition of control in (15). In contrast, the lc-predicate in (17)b seems to indicate that the agent performed the event accidentally and therefore was not in control of the event. This is captured by definition (16)(ii).  (17) a.  b.  chen kw lash-t-Ø ta nkw‘ekw‘ch stn 1S.SUB shoot-TR-3OBJ DET window ‗I shot the window.‘ (on purpose) chen kw lash-nexw-Ø ta nkw‘ekw‘ch stn 1S.SUB shoot-LCTR-3OBJ DET window ‗I shot the window.‘ (accidentally)  Finally, in (18) the lc-predicate is used despite the fact that the agent is fully volitional. However it appears that the agent ―accomplished [the event] only with difficulty‖ and therefore was not really in control over the outcome of the event. This is captured by definition (16)(iii). 10  (18) chen kw lash-nexw-Ø ta nexwl may 1S.SUB shoot-LCTR-3OBJ DET bottle ‗I managed to shoot the bottle.‘ Context: the subject is practice-shooting bottles  Thus, on the assumption that difficulty in accomplishing the desired result indicates lack of full control, the use of lc-predicates in this context is accounted for. As such, Thompson and Thompson‘s (1992) definition captures a wider range of data than the assumption that the relevant notion that c-marking vs. lc-marking encodes is volition vs. non-volition, respectively. As for the definition of non-control in (16)(i), it is not clear how such a reading is different from that of a bare unaccusative root. That is, many unaccusative roots can used on their own, without lc-marking, and still encode events that are: i) natural, ii) spontaneous, iii) without the intervention of an agent. For example, the clause in (19) with the unaccusative root yulh ‗to burn‘, can be used to describe a situation where the event occurs naturally, spontaneously and without the intervention of an agent, such as a forest fire. This clause fits all three conditions for the definition of non-control in (16)(i), yet there is no lc-marking present.  yulh ta sts tse burn DET trees ‗The forest is burning.‘  (19) na  RL  Such data with bare root unaccusatives shows that the definition of non-control may be too broad. In Chapter 4, §3.3 I examine the types of interpretations that are obtained in 11  S wxwu7mesh when unaccusatives are combined with lc-marking. In Chapter 6, §5 I provide an account for why some unaccusatives have lc-interpretations without any overt lc-(in)transitivizers present. The Thompson (1979) analysis has been very influential in Salish linguistics. It captures well the fact that control cannot be reduced to volitionality. It is, however, also not a necessary and sufficient description of control in S wxwu7mesh. In particular, it does not capture what appears to be a purely aspectual use of lc-marking. Consider for example the pair of sentences with a c-predicate (20)a and an lc-predicate (20)b.6  (20) a.  b.  chen ch y-n-t-umi 1S.SUB chase-TR-TR-2S.OBJ ‗I chased you.‘ chen ch y-n-umi 1S.SUB chase-LCTR-2S.OBJ ‗I caught up to you.‘  None of the conditions of non-control identified in (16) is met in example (20)b, despite the presence of the lc-transitivizer. This sentence is compatible with a situation where the agent has full control over the event, and had no difficulty completing it. Of course, this might reflect a difference between languages. Lc-marking in other Salish languages may not be compatible with the purely aspectual interpretation we observe in (20) in S wxwu7mesh.  urthermore since Thompson and Thompson did not  6  Note that sometimes a S wxwu7mesh predicate has two transitivizers as in (20)a. I examine the S wxwu7mesh transitivity system in detail in Chapter 2, §4.1.2.  12  investigate S wxwu7mesh in particular, their definition may be adequate for the Salish languages that they did investigate. Davis et al. (2009) have demonstrated for Lillooet that certain non-CONTROL interpretations in Lillooet (which overlap significantly with S wxwu7mesh limited control interpretations) have a modal base, and not an aspectual base. Such variation strongly indicates that CONTROL may not be a unified phenomenon across the Salish languages. While such variation would not necessarily be expected if indeed transitivizers directly encode the notion of CONTROL, on the present assumption that CONTROL is a construct, this type of variation may would be expected. That is, CONTROL  is expected to be constructed by different grammatical elements in different  languages in the same language family (e.g. transitivizers and modals in Lillooet vs. (in)transitivizers in S wxwu7mesh). To date, no thorough survey of CONTROL marking exists of all the different CONTROL phenomena in Salish. However, it is interesting to note that different researchers have used different labels for non-control constructions, including limited control, out of control, non-control, etc. It is not always clear if these labels are meant to indicate a difference in analysis or if they are different labels for the same phenomenon. I will informally use the term non-control as a cover term for all these constructions, but reserve the term limited control when referring to non-control in Coast Salish. Table 2 provides an overview of non-control terminology throughout the Salish family and the morphemes that are associated with these non-control meanings.  13  Branch of Salish Bella Coola  Language Bella Coola  Morphologic al marking -aynix  Label limited control transitivizer; out of control intransitivizer non-control transitivizer non-control transitivizer  -aylayx Coast  Sliammon  -ng  Saanich  -nəxʷ  Halkomelem  -nəxʷ  -nexw  limited control transitivizer non-volitional transitivizer; limited control transitivizer non-purpose;  n/a  controlled subject; subject-out-ofcontrol not encoded  S wxwu7mesh -nexw  Sechelt  Tsamosan Interior  Upper Chehalis oeur d‘ lene  Lillooet  -C2 reduplication ; -nun ka- -a;  -sut; -s Thompson  -VC2 reduplicant -nw ɬn; -nw nt  Spokane  Table 2  -VC2 reduplication Non-control labels in Salish 14  Davis and Saunders (1997:69-70)  Watanabe (2003:202) Montler (1986:§2.5.2.2) Kiyota (2008:54) Gerdts (2008) Kuipers (1967:77); Bar-el (2005:366), Jacobs (2007) Kuipers et al. (1973:6) Beaumont (1977:12) Beaumont (1985) Kinkade (1991)  non-control resultative non-control (in)transitive resultative; out of control; circumstantial modal; out of control; causative, neutral control out of control;  Doak (1997:45)  noncontrol middle; noncontrol transitive out of control  Thompson (1992:106)  Van Eijk (1997:51); Demirdache (1997); Davis et al (2009); van Eijk (1997:103) van Eijk (1997:111) Thompson (1992:101)  Thompson (1992:107) Carlson and Thompson (1981)  For the Interior branch of Salish, Doak (1997) uses the label non-control (in)transitive for the suffix –nun in oeur d‘ lene. Van Eijk (1997) labels ka- -a in Lillooet as a resultative, while Demirdache (1997) labels it an out of control marker. Davis et al (2009), however, analyze ka- -a as a circumstantial modal. Both Demirdache (1997) and van Eijk (1997) describe the -s causative transitivizer as neutral control. For Thompson, Thompson and Thompson (1992:99, 106-7) label -V2 reduplication as out of control marking, the suffix -nweɬn as a non-control middle, and -nw n‟t as a non-control transitive. For Spokane and Thompson, Carlson and Thompson (1981) use the label out of control for -VC2 reduplication. For the Coast Salish branch Watanabe (2003) uses the label non-control for Sliammon as does Kiyota (2008) for Saanich. Gerdts (2008) uses the label limited control for Halkomelem, as does Bar-el (2005) and Jacobs (2007) for S wxwu7mesh. Kuipers et al. (1973) uses the term non-purpose for the transitivizer –nexw in Sechelt, while Beaumont (1977) describes it as a controlled subject construction, and Beaumont (1985) describes it as a subject-out-of-control construction. Davis and Saunders (1997) use the term limited control for the transitivizer -aynixw and out of control for intransitivizer -aylayx in Bella Coola. Apparently none of the languages of the Tsamosan branch encoded a control distinction (Kinkade 1991). While I will keep using the pre-theoretical label non-control, the question remains as to whether the differences in terms necessarily reflect differences in interpretations. Moreover, given the assumption that CONTROL is a construct, we might expect that CONTROL-marking  has different effects depending on the kind of morpheme that it is  associated with. In most Coast Salish languages CONTROL-marking is tied to (in)transitivizing suffixes, but this is not the case across all languages for non-control 15  marking. For example, in most Interior Salish languages c-predicates are marked by transitivizers on the verb as in S wxwu7mesh, but non-control is variously marked as just discussed. Kroeber (1999:29-30) analyzes the non-control/limited control marker in some languages as a pre-transitivizer because they appear to require the presence of a following transitivizer, such as -nt.  2.3  The aspectual character of CONTROL  As mentioned above, lc-predicates appear to be compatible with an interpretation which is neutral about the degree of control the agent has over the event. Nevertheless, even in contexts where CONTROL seems to play no role in the interpretation of the predicate, there still is a semantic difference between c-predicates and lc-predicates. In particular, in these cases, the difference between c- and lc-predicates appears to be primarily aspectual. In particular, the relevant aspectual difference appears to be whether the described event necessarily culminates or not. Lc-predicates are only compatible with a culminated event, while c-predicates are compatible with both a culminated event or a nonculminated event (Bar-el 2005 for S wxwu7mesh, Gerdts 2008 for Halkomelem, Kiyota 2008 and Turner 2010 for Saanich, Watanabe 2003 for Sliammon). Furthermore, such differences in culmination properties are often translated by completely different lexical items in English. Consider for example the sentences in (21). The c-predicate in (21)a encodes that an event of chasing occurred but not necessarily an event of catching up. In other words, the chasing event did not culminate in an event of catching up. The limited control version in (21)b, in contrast, indicates that both an event of pursuing and an event 16  of catching up occurred. That is, the chasing/pursuing event did culminate in a catching up event. (21) a.  b.  chen ch y-n-t-umi 1S.SUB chase-TR-TR-2S.OBJ ‗I chased/pursued you.‘ chen ch y-n-umi 1S.SUB chase-LCTR-2S.OBJ ‗I caught up to you.‘  The following table of examples shows more pairs of c-predicates and lc-predicates that appear to differ primarily, not in terms of CONTROL, but in what appears to an aspectual difference with regards to event culmination. These data also presents us with a nontrivial problem in analyzing CONTROL: why are some pairs of S wxwu7mesh c- and lcpredicates translated by completely different lexical items in English. I return to this issue in Chapter 6, §4.  17  C-predicate version  Lc-predicate  (22) a. kw‘ach-t  C-predicate translation ‗to look at it‘  (23) a. p‘i7-t  ‗to take/grab it‘  b. p‘ 7-nexw  b. kw‘ ch-nexw  Lc-predicate translation ‗to see it‘  ‗to have/hold/receive it‘ (24) a. yelx-t ‗to search for it‘ b. y lx-nexw ‗to have found it‘ (25) a. ta7l-t ‗to study it‘ b. t l-nexw ‗to have learnt it to realize it, to have found it out, to discover‘ (26) a. w‘ y-ut ‗to beat (a person)‘ b. w‘ y-nexw ‗to have beat (a ‗to kill (game)‘ person) up‘ ‗to have killed (game)‘ (27) a. h y-ut ‗to create it‘ b. h y-nexw ‗to finish it‘ (28) a. kw‘en‗to pour it‘ b. kw‘ n-nexw ‗to spill it; to have poured it‘ Table 3 C- and lc-predicates pairs with primarily aspectual differences  In (22)a the c-predicate version of the root kw‟ h refers to an event of looking while in (22)b the lc-predicate version of the same root refers to an event of seeing. From an aspectual viewpoint, seeing could be understood as the culmination of an event of looking. In (23)a the c-predicate version of the root p‟i refers to an event of grabbing/taking, while the lc-predicate version of the same root in (23)b refers to an event of having/holding/receiving. From an aspectual viewpoint, having/holding/receiving could be understood as the culmination of an event of grabbing/taking. In (24)a the c-predicate version of the root y lx refers to an event of searching, while the lc-predicate version in (24)b refers to an event of finding. From an aspectual viewpoint, finding could be understood to be the culmination of a searching event. The c-predicate version of the root tel/ta7l (25)a refers to an event of studying, while the lc-predicate version in (25)b refers to an event of finding out/learning/realizing. 18  From an aspectual viewpoint, events involving learning/realizing/finding out/realizing could be seen as types of culmination of an event of studying. The relationship between the c- and lc-predicate versions of the roots in (26) and (27) and the types of events that they refer to are more complex and less straightforward than the relationship of other c- and lc-predicate pairs we have seen so far. The relationship is even more idiosyncratic from the point of view of their English translations. The cpredicate version in (26)a refers to either beating (a person) or to killing (game), and the lc-predicate version in (26)b refers to either an event of having beat (a person) up or an event of killing (game) (to refer to an event of killing of a person, S wxwu7mesh has a separate lexical item kw yutsmixw ‗to murder‘). The complicating factor in understanding the relationship between the c- and lc-predicate versions here, is their relationship to the root w‟uy which means ‗to die‘ (whether the subject is human animal or plant). Thus, it is not possible to straightforwardly derive the meanings of the transitive versions of this root to the root itself, which does not simply mean ‗to cause to die‘. The c-predicate version of the root huy (27)a refers to an event of creating, while the lc-predicate version in (27)b refers to an event of finishing. An aspectual description of these two form is more difficult to make, unless we consider that the lc-predicate means something like ‗finish creating‘. Then we would have finish as the culmination of an event of creating, which does not seem to be the correct way to define these predicates. Finally, in (28) the c-predicate version of the root kw‟ lh refers to an event of pouring, while the lc-predicate version can have either an event of spilling, which could be construed as pouring unintentionally (a limited control translation) or as ‗to have poured‘ 19  which from an aspectual viewpoint could be understood as the culmination of a pouring event. On the basis of examples such as those in Table 3, we may conclude that CONTROL marking is not always about the degree of control the agent has over the event. There are some idiosyncrasies in the relationship between the c- and lc-predicates of some roots. If control were the only relevant distinction between c- and lc-predicates, we would expect a more regular correspondence between the c-predicate and lc-predicate versions of a root which represented this difference in CONTROL than we actually do find. We, in fact, obtain a richer description of event types than is predicted by a primarily CONTROL-based account. For example, the root in (25) t l/ta7l is not translated as ‗managed to study ‘ or ‗accidentally study‘ but rather it has a wide range of meanings such as ‗to have learnt/realize/found out/discovered‘. Such examples, again, point to a problem in the investigation of control in S wxwu7mesh – the problem of the differing patterns of lexicalization between S wxwu7mesh and English which I return to in Chapter 6, §4. Such examples, also indicate that at least in some contexts, the CONTROL meanings of CONTROL-marking  are neutralized. But even if we identify the contexts for the  neutralization of CONTROL, we would still have to determine how the difference in event culmination gets to be associated with CONTROL-marking. The central goal of this dissertation is to argue for an alternative hypothesis according to which the aspectual contrast in terms of event culmination constitutes the semantic core of CONTROL-predicates. I analyze all cases pertaining to the notions of control or volition as cases of pragmatic inferences (see Kiyota 2008:82, for a similar  20  claim for Saanich limited control). That is, the CONTROL meanings are inferred from the aspectual properties. The assumption that CONTROL in S wxwu7mesh is essentially about event completion leads to the question of whether this aspectual meaning is found in other Salish languages, and not just specific to S wxwu7mesh. This is indeed the case for Coast Salish. A number of other researchers who have investigated the aspectual quality of c- and lc-predicates in Coast Salish have come to similar conclusions. For example, according to Watanabe (2003:202-219) lc-predicates in Sliammon are compatible with the accidental and the managed to interpretation.7 However, they also have a clearly aspectual character which Watanabe explores with the following test. When a c-transitive verb (control is marked by the suffix -t, a cognate to the S wxwu7mesh control transitive suffix in (1)a) is in the perfective aspect, the result expressed by the verb can be explicitly denied without inducing a contradiction (29)a. But with the lcpredicate version, marked by the transitive suffix -əxʷ (a cognate to the S wxwu7mesh lc-transitivizer –nexw as in (1)b), the result cannot be cancelled without inducing a contradiction (29)b:  (29) Sliammon a. c-predicate k‘əp-t-uɬ čən Ɂiy xʷaɁ k‘əp-as cut-CTR-PAST 1SG.INDC and NEG cut-3CNJ ‗I cut it but it is not cut.‘ (Watanabe, 2003:205, ex. 18-41a)  7  Watanabe (2003) notes that for Sliammon, this property of CONTROL was also noted by Davis (1978).  21  b.  lc-predicate k‘əp-əxʷ-an #Ɂiy xʷaɁ k‘əp-as cut-NTR-1SG.ERG and NEG cut-3CNJ (‗I cut it but it is not cut.‘) (Watanabe 2003:205, ex.18-41b)  Because of this aspectually based difference between c- and lc-predicates in Sliammon, Watanabe (2003:204) considers the possibility that an event culmination reading — that is, an aspectual reading — is the basic meaning of limited control in Sliammon and that the other limited control interpretations, such as accidentally or managed to, are derived from this basic meaning. Watanabe does not further explore, though, how these interpretations might be derived from their aspectual meanings. S wxwu7mesh exhibits the same contrast regarding event culmination as Sliammon. Bar-el et al. (2005) were the first to note this for S wxwu7mesh c-predicates (and for Lillooet). In particular, they note that for a c-predicate in Skwxwu7mesh used in the perfective aspect, event culmination can be explicitly denied without inducing a contradiction (30)a. Thus, as noted in Bar-el (2005), the aspectual properties of cpredicates have important ramifications for the aspectual classification of predicates (that is, their inner aspect or Aktionsart properties). Both Bar-el et al. (2005) and Bar-el (2005) provide an analysis of the absence of event culmination in c-predicates. Moreover, both studies also note that lc-predicates in S wxwu7mesh behave differently in this respect, although they do not investigate their properties (n.b. Lillooet does not have lc-transitive predicates). In my research I have found that the S wxwu7mesh lcpredicates, like their Sliammon counterparts, can only be used if the event culminates. Lc-predicates (in this case marked with -nexw) require event culmination. As a result, the culmination may not be denied without inducing a contradiction (30)b: 22  (30) S wxwu7mesh a. c-predicate na p‘aya -en-t-Ø-as RL fix-TR-TR-3OBJ-3SUB ‗He (John) fixed his canoe  ta DET  John John  ta DET  snexw lh-s canoe-3POS  welh haw ‘-as i h y-nexw-Ø-as but NEG SBJ-3CONJ PRES finish-LCTR-3OBJ-3SUB but he didn‘t finish (fixing) it.‘ (Bar-el et al. 2005, ex. 12) b.  lc-predicate na p‘aya -nexw-Ø-as RL fix-LCTR-3OBJ-3SUB ‗He (John) fixed his canoe  ta DET  John John  ta DET  snexw lh-s canoe-3POS  #welh haw ‘-as i h y-nexw-Ø-as but NEG SBJ-3CONJ PRES finish-LCTR-3OBJ-3SUB but he didn‘t finish (fixing) it.‘  This is a striking result in light of the fact that these c-predicates otherwise look like what Vendler (1967) describes as accomplishment predicates in English. In English, accomplishment predicates do not allow for event culmination to be cancelled, as shown in (31) much like the lc-transitive in S wxwu7mesh in (30)b.  (31) John fixed the fence (# but he didn‘t finish fixing it).  The absence of a requirement for event culmination for c-predicates appears to be quite common throughout the Salish family (Gerdts 2008 for Halkomelem, Kiyota 2008 for Saanich, Matthewson 2004a for Lillooet, Turner 2010 for Saanich and Watanabe 2003 for Sliammon). The requirement of event completion for lc-predicates appears to be quite common throughout Coast Salish (Gerdts 2008 for Halkomelem, Kiyota 2008 for Saanich, Turner 2010 for Saanich and Watanabe 2003 for Sliammon). For example, 23  Kiyota (2008) demonstrates this contrast between c- and lc-predicates for Saanich, another Coast Salish language closely related to S wxwu7mesh. The result expressed by the c-predicate (marked by -t, cognate to the S wxwu7mesh -t) can be cancelled without inducing a contradiction (32)a. In contrast, the lc-predicate (marked with -nəxʷ, a cognate to the S wxwu7mesh -nexw) requires event culmination and consequently the culmination cannot be cancelled without inducing a contradiction (32)b.  (32) Saanich a. c-predicate ləɁə sən kʷəʔ l -t AUX 1.sg INF get.fixed-CTR ‗I fixed the table  tsə D  latəm table  ʔiʔ ʔawa sən šəq-naxʷ ACC NEG 1.SG complete-NCTR but I didn‘t finish it.‘ (Kiyota 2008:59, ex.42a) b.  lc-predicate tə sp ʔəs D bear  ləɁə qsən kʷəʔ qʷəy-nəxʷ AUX 1.SG INF die-NCTR ‗I (accidentally) killed the bear,  #ʔiʔ ʔawa s-qʷəy ACC NEG NOM-die but it didn‘t die.‘ Speaker‘s comment: Contradiction!  (Kiyota 2008:59, ex.43a)  Gerdts (2008) provides similar data from Halkomelem. C-predicates do not require event culmination while lc-predicates do. As a consequence the culmination of the event can be cancelled without a contradiction with c-predicates (marked by -t, cognate to  24  Skwxwu7mesh -t) as in (33)a. The cancellation of event culmination results in a contradiction with lc-predicates (marked by -nəxʷ), however (33)b.  (33) Halkomelem a. c-predicate niʔ cən qa:y-t tᶿə speʔəϴ AUX 1S.SUB kill-TR DET bear ʔiʔ ʔəwə niʔ -əs qay. and NEG AUX-3SSUB die ‗I killed the bear but it didn‘t die.‘ b.  lc-predicate #niʔ cən qəy-nə ʷ tᶿə speʔəϴ AUX 1S.SUB kill-LCTR DET bear ʔiʔ ʔəwə niʔ -əs qay. and NEG AUX-3SSUB die ‗I managed to kill the bear but it didn‘t die.‘  These data also support the claim that lc-marking has a core aspectual meaning (cf. Kiyota 2008 and Watanabe 2003), at least for Coast Salish. Kiyota (2008:82) states that the various interpretations associated with lc-marking in Saanich are pragmatic inferences, but he leaves the exact nature of these inferences to further research. The analytical challenge remains then. What is the relation between CONTROLmarking and event culmination? And how do these aspectual properties lead to meanings that appear to pertain to the notion of control? In other words, if control or volition is not part of the meaning of CONTROL marking then why do c-predicates appear to differ from lc-predicates in this respect, at least sometimes? The generalization to be accounted for is summarized in Table 4. 25  core meaning  inferred meaning  c-predicate  event culmination possible  control  lc-predicate  event culmination necessary  limited control  Table 4  The meaning of CONTROL in S wxwu7mesh (Jacobs 2011)  In sum, one of the goals of this dissertation is to establish that, indeed, the core meaning of control constructions is about event culmination rather than about degrees of control the agent has over the event. But the question is how? This raises another analytical challenge associated with control marking: identifying the particular morpheme which is responsible for encoding whether or not the event culminates. The answer to this question is not quite straightforward. I turn to this problem in the next section.  3  Where is the morphological marking for CONTROL?  While Kuipers‘ (1967) analysis does not adequately describe the meanings associated with CONTROL-marking, it nevertheless provides us with an important description of some of the morphological properties associated with it. Consider again the contrast between c- and lc-predicates illustrated in Table 1, repeated below as Table 5:  26  (34)  root --  c-transitive p‘i7-t take, grab  lc-transitive p‘ 7-nexw have, hold  (35)  --  ch‘ w-at help  ch‘ w-nexw have helped  (36)  sum to smell, to give off odor lhaw to run away  s m-  (37) Table 5  smell, sniff8  lhaw-s elope with  s m-nexw smell lh w-nexw have eloped with  Volitional and non-volitional transitives (Kuipers 1967:77-78)  Paying attention to the c-transitivizers, we observe that there are four different types of c-transitivizers and only one lc-transitivizer (-nexw): i) the root may be followed by the -t transitivizer as in as in (34), ii) the root may be followed by the transitivizer -Vt (where V is a copy vowel of the root) as in (35), iii) the root may be followed by -Vn (where V is sometimes a copy of the root vowel as in (36), and finally iv) the root may be followed by the causative transitivizer -s as in (37). Each of these different c-predicates has a corresponding lc-counterpart. This introduces another puzzle associated with the CONTROL marking that I set out to address in this dissertation. If CONTROL-marking is best analyzed as a morphologically marked semantic contrast, then why are there four c-transitivizers but only one lctransitivizer among the core transitivizers? Moreover, in addition to the core transitivizers, Kuipers (1967:95-97) notes that the contrast of volitionality also pertains to some of the intransitivizers: reflexives and reciprocals. In particular, he classifies the reflexive -sut and the reciprocal –way as  8  This form is found in Kuipers (1969:62).  27  encoding volitionality. They contrast with the non-volitional reflexive –numut and the non-volitional reciprocal -n w s, respectively. This is summarized in Table 6, where I have replaced Kuipers‘ label volitional/non-volitional with the terminology I am using: control (c)/limited control (lc).  Transitive Reflexive Reciprocal -t -sut -way -Vt -Vn -s limited control -nexw -numut -n w s Table 6 C- and lc-(in)transitivizers (adapted from Kuipers 1967) control  Note that the transitivizers are not transparently related to the reflexive and reciprocal counterparts. Thus, if indeed we are dealing with the same morphological contrast, then what is responsible for CONTROL marking in each case? In sum, while Kuipers‘ semantic characterization of CONTROL in terms of volitionality is not empirically adequate, his initial description still sets the agenda for our investigation. We can identify the following issues any analysis needs to address. The morphosyntax of CONTROL: A)  Why are there 4 control transitivizers as opposed to 1 limited control transitivizer?  B)  What determines the distribution of the different control transitivizers?  C)  Is this problem reducible to allophony or is there more to this problem?  D)  If the contrast in CONTROL extends to reflexives and reciprocals, 28  then what is the morpheme that encodes the contrast?  4  Outline of my proposal  The following is the outline of this dissertation. A) Background (Chapter 2) I present a linguistic viewpoint of the place of S wxwu7mesh in the Salish language family. I provide the background on the methodology I used to collected the data used for this dissertation. I then provide a grammatical sketch, outlining some of the morphological issues that I address in this dissertation.  B) Aspectual core for CONTROL (Chapter 3) I show that c- and lc-transitivizers mark an aspectual difference. In particular, lcpredicates assert that the natural endpoint encoded in the predicate is the actual endpoint of the event that occurred. In contrast, c-predicates do NOT assert that the natural endpoint encoded in the predicate is in fact the final point of the event. Rather, all that they assert is event initiation.  C) The context of use of CONTROL: Pragmatic inferences (Chapter 4) I show that the impression that c- vs. lc-transitives encode a degree of control comes about via pragmatic inferences. These inferences arise via a combination of linguistic knowledge on the one hand and world-knowledge on the other hand. In particular, if a speaker chooses to make an assertion about the culmination of the event, the listener can 29  infer that there must be a reason for this choice: why wasn‘t the initiation of the event not asserted? The listener may infer that something was unusual about the initiation of the event, e.g. the agent caused the event accidentally. As such, this type of inference is triggered by linguistic knowledge pertaining to the systematic morphological contrast available in the language. Moreover, the impression that c-predicates mark control is the result of worldknowledge. If a speaker indicates that an agent initiated an event, and does not state otherwise, the listener assumes that the agent was volitional and had full control over the process of event such that she brought the event to culmination.  D) The syntax of CONTROL (Chapter 5) Assuming that event culmination or the lack thereof is about telicity, I provide a morphosyntactic model for representing these two different aspectual differences assuming Travis‘ (2010) phrase structure. I propose that when object agreement is associated with FP-delimit, a telic reading is obtained: that is, our lc-predicates. But, when object agreement is associated with VP then the predicate is not telic: that is our cpredicates. Instead, only the thematic role of the participant is marked. I then provide a preliminary account of how this analysis can be extended to the causative and the intransitivizers. I separate the causative from the other c-transitivizers because it presents a more complicated picture regarding event structure than the other c-transitivizers.  30  E) Implications (Chapter 6) Assuming that the two different types of object agreement were present in Proto-Salish, I provide a very preliminary account for the development of the present day S wxwu7mesh object agreement/intransitive markers from the two Proto-Salish object sets (Newman 1979). I then provide a preliminary account of how my morphosyntactic model can be extended to other Coast Salish languages. I also examine briefly why limited control-like meanings are not associated with culmination in languages like English. I then discuss how other non-control constructions in S wxwu7mesh and other Salish languages might fit with my analysis of CONTROL presented in this dissertation.  31  Chapter 2 - Background on  1  7mesh  Introduction  In this chapter I first provide a linguistic background to the S wxw 7mesh language in §2. I then provide a description of the methodology used in collecting the data for this dissertation in §3. I then provide a grammatical sketch of the S wxwu7mesh language with a focus on the parts that are relevant to the construction of CONTROL in §4.  2  Linguistic classification  In the current Salishanist linguistic classification, S wxw 7mesh is a member of the Central Salish branch of the Salish language family. This term, however, is not used by S wxw 7mesh people themselves. The preferred term is ‗Coast Salish people‘. The following table is a classification based on Czaykowska-Higgins and Kinkade (1998a) but with the term ‗Coast Salish‘ instead of ‗Central Salish‘. I also provide the S wxwu7mesh names, where they exist, for other Salish languages or dialects. This classification is based on Czaykowska-Higgins and Kinkade (1998a).  32  Branch  Language  Bella Coola Coast  Bella Coola Comox  S wxwu7mesh Name P lxwela Sliy min Xwem lhkwu Tsalh lhtxw  Sechelt Pentlatch Squamish Halkomelem  Shish 7lh  Nooksack  Xwsa7  S wxwu7mesh S ‘em nem  Northern Straits  Tsamosan  Tillamook Interior  Table 7  Lushootseed Klallam Twana Upper Chehalis Cowlitz Lower Chehalis Quinault Tillamook Shuswap Thompson Lillooet  (Xwlesh)  Dialects  Sliammon, Halmalko, Island Comox Klahoose  Upriver, Downriver, Island  Semiahmoo, Saanich, Lummi, Songish, Samish, Sooke Northern, Southern  Satsop, Oakville, Tenino  Lhek pmexw Stl‘ lmexw L xwels  Okanagan Moses-Columbian Kalispel oeur d‘ lene Salish language family  Lillooet/Upper Lillooet Mt. Currie/Lower Lillooet Northern, Southern/Colville Kalispel, Spokane, Flathead  33  3  Methodology  The data in this dissertation comes from four sources. One source of data for this dissertation is ert Kuipers‘ two volume: The Squamish Language: Grammar Texts and ictionary (1967 1969). Kuipers‘ consultants were largely from a generation previous to the consultants that I worked with. His consultants were older family members of a number of the consultants that I worked with. escendents of Kuipers‘ consultants still recall stories of his visits to the S wxwu7mesh communities in the 1950‘s and 1960‘s. A second set of data I used was collected by employees of the S wxwu7mesh Nation Department of Education (SNED) in North Vancouver, Canada, and it includes: i)  words and sentences collected as part of the S wxwu7mesh-Xwel ten S exwts (the Squamish-English Dictionary Project),  ii)  texts from traditional stories,  iii)  texts from curriculum for the Squamish Language Program (elementary, high school and college level S wxwu7mesh language classes).  A third source of data I used was the data collected by researchers other than myself and it includes elicitation sessions from graduate students from the Department of Linguistics at UBC and post-doctoral researchers, also associated with UBC. As an employee of SNED I participated in most of these elicitation sessions. All these sessions were originally recorded on minidisc and have since been transferred to external hard drives.  34  The fourth source of data is my own elicitation sessions conducted as part of the research for this dissertation. The methodology that I used when eliciting data for this dissertation consisted primarily of four tasks: i)  translation of constructed S wxwu7mesh sentences into English,  ii)  translation of English sentences into S wxwu7mesh,  iii)  providing the speakers with S wxwu7mesh sentences and then asking them about the appropriateness of that sentence in various contexts,  iv)  providing the speakers with S wxwu7mesh sentences and then asking them for appropriate contexts for those sentence.  All my elicitation sessions were originally recorded on my laptop Dell XPS-M1530. The second, third and fourth sets of data I used are all housed with SNED and have been transferred to external hard drives.  4  Grammatical sketch  It is the main goal of this dissertation to show that CONTROL is constructed from a combination of morphological marking and pragmatic inference. As such, we need to get acquainted with those aspects of the S wxwu7mesh grammar that are relevant to the construction of CONTROL. Moreover, a general discussion of some of the key aspects of the grammar will facilitate the discussion of the data. Since CONTROL is intimately tied to the transitivizer system in the form of (in)transitive suffixes on the verb, we start with a discussion of the template for the complex verb in §4.1. We briefly examine the verbal stem in §4.1.1. We then examine in some detail the properties of the transitivizing 35  system in §4.1.2. Then we discuss person marking in §4.1.3. We then move on to discuss other components of the grammar that are not directly relevant for CONTROLmarking but that are still vital in the understanding of the S wxwu7mesh data: verb phrase auxiliaries and particles in §4.2, the determiner phrases in §4.3, case in §4.4 and linearization and word order in §0.  4.1  The verbal template  As mentioned above, CONTROL marking is intimately tied to the (in)transitivity system in S wxwu7mesh. In particular, the difference between c-predicates and lc-predicates lies in the choice of the (in)transitivizer. Thus, a useful starting point for our discussion is the morphological template for the S wxwu7mesh verb, given in (1). This template is based on the one given by Czaykowska-Higgins and Kinkade (1998a: 23) for the Salish word.  (1)  Template for the S wxwu7mesh verb  ASPect-REDuplication-ROOT-Lexical.Suffix-TRansitivizer/INtransitivizer-OBJectSUBject-NUMber  I discuss each of these morphological slots and the pieces that occupy them in turn. The depth of discussion will depend on the degree of relevance the pieces have for the discussion throughout the dissertation.  36  I will start with a discussion of the central part of the verbal template, the parts of the verbal stem: roots in §4.1.1.1 and lexical suffixes in §4.1.1.2. I then briefly discuss the prefixes in §4.1.1.3 before moving on to those pieces that play a key role in the construction of CONTROL, namely the (in)transitivizers, in §4.1.2. Finally I discuss person marking in §4.1.3.  4.1.1  The verbal stem  The verbal stem consists of the root (§4.1.1.1), lexical suffixes (§4.1.1.2), and the verbal prefixes (§4.1.1.3).  4.1.1.1  Roots  I start the discussion with a brief overview of the morpho-syntactic and semantic behaviour of roots. Roots are the core of the template:  (2)  Template for the S wxwu7mesh verb: the ROOT ASP-RED-ROOT-LS-TR/INT-OBJ-SUB-NUM  As illustrated in (3)a, many roots may appear in their bare form (i.e., when the agreement affixes are phonologically null). However, roots may be surrounded by both prefixes and suffixes as shown in (3)b, which is an example of a verb that has various morphological slots occupied. 37  (3)  a.  na  tsexw-Ø RL hit-3SUB ‗he got hit.‘ (by something thrown)  b.  na  es-ch‘us-um-nit-Ø-as-wit STAT-shun-MID-RELAPPL-TR-3OBJ-3SUB-PL ‗They shunned him.‘ RL  In S wxwu7mesh, all roots are morphologically intransitive. The majority of roots are inherently patient oriented: that is, their only argument is the participant that undergoes the event (rather than the agent). Since Perlmutter (1978), such verbs have been known as unaccusative.  (4)  a.  chen tsexw 1S.SUB hit ‗I got hit.‘ (by something thrown)  b.  na  c.  chet p‘ ya 1P.SUB recover ‗We recovered we got better (from being sick).‘  ynexw-Ø RL alive-3SUB ‗S/he is alive.‘  Also, as per Bar-el (2005), I assume that all unaccusatives have culmination entailments in the perfective aspect. Bar-el (2005:90-91, ex.53a-b) shows that it is not possible to question the culmination of an accusative root in the perfective without inducing a contradiction shown in (5) below:  38  w‘uy kwa John RL die DET John ‗John died.‘  (5) a.  na  na7-xw u wa es- w‘ y kwa John RL-still POL IMPF STAT-sick DET John ‗Is John still sick?‘ Speaker‘s comments: Not OK as a question after the previous statement: ―Why would you ask a question like that if he‘s already dead?‖  b.  Because of these facts, I assume that unaccusative roots, besides having a patient/undergoer argument, also have culmination as part of their inherent meaning. There are also roots whose only argument appears to be the agent of the event. As such they could be classified as unergative roots (Perlmutter 1978).  (6)  a.  chen lhich‘ 1S.SUB cut ‗I am cutting.‘  b.  na  lhen-Ø-wit eat-3SUB-PL ‗They ate.‘ RL  c.  chet ts‘its‘ p‘ 1P.SUB work ‗We work.‘  Bar-el (2005) demonstrates that unergatives (in her terms, activities), in contrast to unaccusatives, do not have culmination entailments as the following example  (7) a.  na mesh kwa John rl walk det John ‗John is walking 39  b.  i iw yti na7-xw wa mesh and maybe RL-still IMPF walk and maybe he‘s still walking.‘  This example shows that it is possible add a phrase that indicates that the event of the unergative root is still continuing. It does not have to culminate. Because of these facts, I assume that unergative roots have an agent argument and they do not have culmination as part of their inherent meaning. Davis (1997) argues that all Salish roots are underlyingly unaccusative. This is known as the deep unaccusativity hypothesis. For those roots that appear to be agentoriented he argues for the presence of a zero morpheme which makes the root unergative. However, for the purpose of this dissertation I assume that roots can be underlyingly unaccusative or unergative. This assumption is similar to Gerdts and Hukari (2006a), except that I do not assume that there are also inherently transitive roots in S wxwu7mesh. That is, I do not assume that there are any roots which have both an agent and a theme role associated with them. I argue in Chapter 5 that this assumption for S wxwu7mesh provides for a clearer basis for understanding the differences between the various transitivizers in S wxwu7mesh. I also assume that unaccusatives some in two subclasses: i) regular unaccusatives and verbs of motion (Gerdts 1991 for Halkomelem). Verbs of motion can be classified as unaccusative because their subject functions as a theme: it is the participant who undergoes the change in location. But these roots, in addition to having a them role are also lexically specified for the goal of their motion or the source from which this motion takes place. Take, for example, the verb of motion t‟ukw‟ ‗to come/go home‘ as in 40  examples (8)a-b. Not only does this verb indicate motion, but it also has as the goal of this motion: ‗home‘. The verb of motion wu w‟ has as its goal ‗downstream‘ (8)c and mich has as its goal ‗upstream‘ (8)d. Some verbs of motion, such as kwum in (8)e, indicate movement away from a source ‗the beach‘.  (8)  a.  chen nam ’ ’ 1S.SUB go go.home ‗I went home.‘  b.  nam chen go 1S.SUB ‗I will go home.  c.  na  ’ ’ go.home  mi ’-i7-Ø come go.downstream-INCH-3SUB ‗He got downstream here.‘ RL  d.  e.  na  wa  RL  IMPF  -Ø go.upstream-3SUB ‗He is going upstream.‘ chet nam kwum 1P.SUB go go.up.from.beach ‗We went up away from the beach.‘  Normally these roots occur with an auxiliary mi ‗to come‘ or n m ‗to go‘ which serve to indicate the location of the goal/source with regards to the speaker or to the narrative  41  context.9 For example, the auxiliary n m in (8)a-b indicates that the speaker is not physically located at ‗home‘ at the time of utterance. While I treat these verbs of motion as a type of unaccusative, it is ambiguous as to whether their argument, other than the goal/source argument, is a theme or agent. In Chapter 5 I will discuss some of their aspectual properties and their interaction with the causative to provide a clearer picture of the issues in describing the inherent thematic roles of verbs of motion. One other issue regarding roots in S wxwu7mesh, and in Salish in general, is the issue of whether nouns and verbs exist as distinct lexical categories (cf. CzaykowskaHiggins and Kinkade 1998a: 35-38 for a brief overview of the issues). For the purposes of this dissertation I assume that there is such a distinction, along with Demirdache and Matthewson (1995) and others.  4.1.1.2  Lexical suffixes  Beside the root, there is another slot in the verbal template which hosts morphemes with lexical content. That is, roots can combine with suffixes which in the Salishanist literature are known as lexical suffixes (Czaykowska-Higgins and Kinkade 1998a, Gerdts 2003, Hinkson 1999, Kinkade 1998, Wiltschko 2009). In particular, these suffixes  9  For reasons which are not clear, for some of these verbs of motion (e.g. t‟ukw‟ ‗go home‘) the presence of these auxiliaries appears to be obligatorily. This requires futher research.  42  (mostly) attach directly to the root as shown in the template repeated below for convenience and illustrated in the data below.  (9)  Template for the S wxwu7mesh verb: Lexical suffixes ASP-RED-ROOT-Lexical.Suffixes-TR/INT-OBJ-SUB-NUM  The following are examples containing lexical suffixes. In (10)a, the lexical suffix is yus ‗eye‘ and in (10)a it is - w ‗head‘. Note that the transitivizers are attached outside of the lexical suffixes.  na chemx-en-t-Ø-as ta st 7uxwlh pitch-eye -TR-TR-3OBJ-3SUB DET children ‗She put pitch on the children‘s eyes.‘  (10)  a.  RL  b.  es-h m- -s-t-em-Ø STAT-covered-head-CAUS-TR-PASS-3SUB ‗They had her head covered.‘ na  RL  S wxwu7mesh (like the other Salish languages) has over a hundred such suffixes. They differ from roots in that they cannot stand on their own. Moreover, the lexical suffix is often not transparently phonologically related to the corresponding independent noun. In a number of cases in S wxwu7mesh there is some obvious relation with the independent noun, but the relation is never fully transparent. The following table provides examples of some of the lexical suffixes. The independent full word is listed in the first column, the corresponding lexical suffix in the second column and then an example word with the lexical suffix in the third column. 43  bed  independent word slhaw n  lexical suffix  example  -a7lh  p‘aya -a7lh-m ‗fix one‘s own bed‘ fix-bed-MID xw-hiy‗have big eyes‘ LOC-big-eye hiy‗big fire‘ big-fire hiy -shen ‗have big feet‘ big-foot si7- - m ‗wipe one‘s own nose‘ wipe-nose-MID tsi ‗stabbed in the tongue‘ stabbed-tongue  eye  el m  fire  y yulh  -ikwup  foot, leg  sxen  -shen  nose  m sen  - s  tongue  me lxwtsalh  -alxwtsalh  Table 8  4.1.1.3  -ayus  Lexical suffixes in S wxwu7mesh  Verbal prefixes  The number of prefixes in S wxwu7mesh is considerably less than the number of suffixes. Nevertheless, the prefixes that do occur are productive. While the template has the ASPectual and REDuplicant prefixes, semantically the REDuplicants can also have aspectual meanings. Furthermore, the ASPectual slot also includes the nominalizer. Nevertheless, I will still use these two labels as pretheoretical labels. There are two types of prefixal reduplication: CVC- and CV-. These occur directly preceding the root as indicated in the template: (11)  Template for the S wxwu7mesh verb: Reduplicants ASP-RED-ROOT-LS-TR/INT-OBJ-SUB-NUM  44  CVC reduplication can occur with both verbs and nouns. With nouns it indicates plurality.  (12) a.  push ‗cat‘  b.  pesh-p sh RED-cat ‗cats‘  c.  s- wem y NOM-dog ‗dog‘  d.  s- wem- wem y NOM-RED-dog ‗dogs‘  With verbal roots it can indicate plurality for the internal argument as in reading (i) of example (13) or it can indicate repetitive aspect (i.e. plurality of the event) as in reading (ii) of example (13).  kwel-kw lash-t-Ø-as ta m xalh RED-shoot-TR-3OBJ-3SUB DET bear i) ‗He shot a number of bears.‘ (at the same time) ii) ‗He shot the bear repeatedly.‘  (13) na  RL  Bar-el (2005) analyzes the CV- reduplicant as the progressive aspect. It only occurs with verbs. Compare the verbal root without CV- (14)a and then with it (14)b.  45  (14) a.  chen nam t‘ukw‘. 1S.SUB go go.home ‗I went home.‘  b.  t‘ -t‘ukw‘ chen. RE-go.home 1S.SUB ‗I‘m on my way home.‘  The next prefix out from the REDuplicants is the stative es-, which I place in the ASPect slot, as indicated in the template:  Template for the S wxwu7mesh verb – Aspectual prefix  (15)  ASP-RED-ROOT-LS-TR/INT-OBJ-SUB-NUM The aspectual prefix es- ‗stative‘ derives a predicate denoting a resulting stative.  (16) a.  na  es-ch‘ich‘-Ø STAT-twisted-3SUB ‗It is twisted.‘ RL  b.  ch‘ich‘-Ø twisted-3SUB ‗It got twisted.‘ na  RL  I include the nominalizer s-, in the same slot as es- in ASPect slot, even though its function appears to be more syntactic than aspectual. It can derive nouns from verbs as in (17). It also occurs to indicate a nominalized (dependent) clause, wherein it looks more like a clausal proclitic than a prefix. The example in (18) has the nominalized clause functioning as a complement clause. The s- can also indicate certain types of  46  extraction. In (19)a the s- indicates that a formerly oblique argument, t lh h‟tn (19)b, has been extracted.  (17) s-ta w NOM-drink ‗water‘ t l-nexw-Ø-as RL know-LCTR-3OBJ-3SUB ‗He knows that I worked.‘  (18) na  (19) a.  na RL  ts‘its‘ p‘ work  lhach‘tn (na n-s-7 xwa7-t-umi) knife RL 1S.POS-NOM-give-TR-2S.OBJ ‗This is the knife that I gave you.‘ nilh ti FOC  b.  kwi-n-s DET-1S.POS-NOM  DET  chen xwa7-t-umi 1S.SUB give-TR-2S.OBJ ‗I gave you this knife.‘  t-ti OBJ-DET  lhach‘tn knife  Since neither lexical suffixes nor prefixes are the focus of our investigation, we can revise the template to reflect only the pieces relevant for the present discussion. I will refer to the root in combination with prefixes and lexical suffixes as the stem. This leaves us with the following template which isolates all and only the pieces relevant for the present discussion.  (20)  Template for S wxwu7mesh verb: stem [ASP-RED-ROOT-LS]STEM-TR/INT-OBJ-SUB-NUM  47  4.1.2  The transitivity system  As noted above, all S wxwu7mesh roots are inherently intransitive and consequently may only co-occur with one argument, which is realized as the subject. Thematically, this argument may be an agent (21)a or a patient (21)b.  (21) a.  b.  chet ts‘its‘ p‘ 1P.SUB work ‗We work.‘ na  tsexw-Ø hit-3SUB ‗He got hit.‘ (by something thrown) RL  Bare roots may never be transitive and as such cannot directly combine with object agreement. This is illustrated in (22).  (22) *chen tsexw-umi 1S.SUB hit-2S.OBJ (intended meaning: I hit you)  Rather, in order for any root to co-occur with two arguments a transitivizer needs to be added to allow for object agreement (23)a, or one of the arguments must occur as on oblique argument (23)b.  48  (23) a.  b.  chen tsexw-n-umi 1S.SUB hit-TR-2S.OBJ ‗I hit you.‘ chet xw-im t-kwi sts‘ 7 in 1P.SUB collect-CUE OBL-DET bullrush ‗We collected bullrushes.‘  Within the morphological template, the transitivizer occupies the position immediately following the stem and preceding object markers as illustrated in (24).10  (24)  Template for S wxwu7mesh verb: transitivizer STEM-TR/INT-OBJ-SUB-NUM  There are a number of transitivizers and intransitivizers in S wxwu7mesh that appear to occupy this morphological slot, many of which have been described as encoding either control or limited control (Jacobs 2007). While the main thesis I put forth in this dissertation is that CONTROL is a construct and therefore not directly encoded in any morpheme (including the (in)transitivizers), morphological marking still plays a role in constructing the relevant interpretations. Crucially, the locus of c- and lc-marking can be identified as TR/INT. This can be seen on the basis of the contrast introduced in Chapter 1 and repeated below:  10  None of the morphemes that occur inside the stem has an effect on the arguments that the root can cooccur with.  49  (25) a.  chen kw‘lh-at-Ø ta tiy 1S.SUB pour-TR-3OBJ DET tea ‗I poured the tea.‘ (on purpose)  b.  chen kw‘ lh-nexw-Ø ta tiy 1S.SUB spill-LCTR-3OBJ DET tea ‗I spilt the tea.‘ (accidentally)  In (25)a the stem is immediately followed by the transitivizer -at and the result is a cpredicate: the agent is interpreted as having control over the event. In contrast, in (25)b the stem is immediately followed by the transitivizer -nexw and the result is an lcpredicate: the event is interpreted as coming about accidentally or with great difficulty. Since the transitive system plays a crucial role in CONTROL-marking, I discuss it in some detail. For now, the discussion is based on the only previous description of the transitivizers in S wxwu7mesh, namely that of Kuipers (1967). It will be clear from this discussion that CONTROL-marking in the form of transitivizers not only presents us with questions regarding the semantics associated with it (as discussed in chapter 1) but it also presents us with some non-trivial morphological problems. I first discuss the core transitivizers (§4.1.2.1), and then the applicative transitivizers that add a different type of argument (e.g. a benefactive argument) to the predicate (§4.1.2.2). Finally I show that there are also a number of intransitivizers that appear to occupy the same morphological slot (§4.1.2.3).  50  4.1.2.1  Core transitivizers  The transitivizers have already been introduced in Chapter 1 and appear to be responsible for c- and lc-marking, I now call them core transitivizers, in contrast to the applicative transitivizers. They are repeated below.  Core Transitivizers control -t -Vt -Vn -s limited control -nexw Table 9 Core transitivizers in S wxwu7mesh  The V preceding t and n represents an abstract vowel. The quality of this vowel depends on various phonological factors. It can appear as a copy of the root vowel and henceforth, I refer to it as the copy-vowel.11 In my own database, I have 59 roots (listed in Appendix A, §1) that occur in the –t transitive construction, 68 roots (listed in Appendix A, §2) that occur in the –Vt construction and 233 CVC roots (listed in Appendix A §3) that occur in the –Vn construction. The number of roots that can occur with the causative –s appears to be open ended.  11  See Dyck (2004) for a phonological account of the copy vowels. I provide an alternative account in Appendix B §2.  51  Each of these transitivizers is illustrated in examples (26)-(29) below. For each transitivizer I provide an example with the bare root first (the a examples) and then an example with the same root with the relevant transitivizer (the b examples).  (26) -t a.  na  lixw-Ø ta smant put.down-3SUB DET rock ‗The stone has been laid down (e.g. as a marker) RL  b.  na  lixw-t-Ø-as ta smant put.down-TR-3OBJ-3SUB DET rock ‗He put the rock down.‘ RL  (27) -Vt a.  b.  chen lhich‘ 1S.SUB cut ‗I cut.‘ chen lh ch‘-it-Ø 1S.SUB cut-TR-3OBJ ‗I cut it.‘  (28) -Vn a. na  yulh-Ø ta y yulh burn-3SUB DET fire ‗The fire is burning.‘ RL  b.  (29) -s a.  b.  chen y lhta 1s.sub burn-tr det ‗I burned the food.‘  s7 lhen food  chen lhen 1S.SUB eat ‗I ate.‘ ilhen-s-t- mulh-as eat-CAUS-TR-1PL.OBJ-3SUB ‗She fed us.‘ na  RL  52  (30) -nexw a. chen kw lash 1s.sub shoot ‗I shot.‘ b.  chen kw lash-nexw-Ø 1s.sub shoot-lctr-3obj i) ‗I managed to/got to shoot it.‘ ii) ‗I accidentally shot it.‘  We also noted in Chapter 1 two morphological puzzles that arise in this context: why are there four c-transitivizers as opposed to one lc-transitivizer? And what determines the distribution of the different c-transitivizers? According to Kuipers, the distribution of three of the transitivizers is partly lexically conditioned. In particular, the use of –t and -Vt is restricted to a closed class of roots. In contrast, –Vn appears to be a productive c-transitivizer, as is the causative -s. One reason to consider the –Vn transitivizer as a productive transitivizer come from certain combinations of root and lexical suffixes combinations. For roots that normally take the –t or –Vt transitivizers, when a lexical suffix is present, they instead take the –Vn transitivizer. For example, take the bare root in (31)a. It takes the transitivizer -t in (31)b. But, when a lexical suffix is present it takes the transitivizer -Vn in (31)c. The same holds for the root from (27), repeated here again as (32)a, which takes the transitivizer –Vt¸ when the root has no lexical suffixes (32)b. However, when a lexical suffix is present, the stem takes the –Vn transitivizer (32)c.  53  (31) -t  -Vn a. na xwil-Ø te-n yen s RL come.off-3SUB DET-1S.POS teeth ‗My teeth come about (by themselves).‘ (Kuipers 1967:372) b.  chen xwi7l-t-Ø te-n kap 1S.SUB take.off-TR-3OBJ DET-1S.POS coat ‗I took off my coat.‘  b.  chen xwil-ts-en-Ø 1S.SUB take.off-mouth-TR-3OBJ ‗I opened the door.‘  ta DET  shew lh door  (32) -Vt  -Vn a. chen lhich‘ 1S.SUB cut ‗I cut.‘ b.  chen lh ch‘-it-Ø 1S.SUB cut-TR-3OBJ ‗I cut it.‘  c.  chen lhich‘- ch-n-Ø 1S.SUB cut-hand-TR-3OBJ ‗I cut his hand.‘  There is an wrinkle with the productive -Vn transitivizer and the causative –s when the object is third person. Both the -Vn transitivizer and the causative -s are sometimes followed by -t (which is homophonous with the simple –t transitivizer), and sometimes not. Thus, consider the two sets of examples in (33) and in (34). In (33)a, -Vn is followed by -t. In contrast, in (33)b which contains the same root, -Vn is not followed by -t. In (34)a, -s occurs followed by -t. In contrast, in (34)b which contains the same root, -s is not followed by -t.  54  (33) a.  3sg. subject  3sg. object  -Vn-t na  lh w‘-an-t-Ø-as slap-TR-TR-3OBJ-3SUB ‗He slapped him.‘ RL  b.  (34) a.  -Vn-[ ] chen lh w‘-an-[ ]-Ø 1S.SUB slap-TR-?-3OBJ ‗I slapped him.‘  1sg. subject  3sg. object  -s-t na  3sg. subject  3sg. object ta kw‘ xwa7 DET box  t 7-s-t-Ø-as RL do-CAUS-TR-3OBJ-3SUB ‗He made a box.‘ b.  -s-[ ] chen ta7-s-[ ]-Ø 1S.SUB do-CAUS-?-3OBJ ‗I made a box.‘  ta DET  1 sg. subject  3sg. subject kw‘ xwa7 box  The two sets of examples minimally differ in that the subject of the (a) examples is realized by the third person agreement suffix –as on the verb, while in the (b) examples it is realized by a preverbal clitic for first person singular subject. The contrast in (33) and (34) raises two questions. First, if -t and -Vn are both transitivizers, then why can they co-occur? And, if -t and -s are both transitivizers, then why can they co-occur? In other words, why does S wxwu7mesh have –t as an apparently redundant transitivizer? Or, does -t have a separate function? Kuipers‘ (1967) answer to these questions was that -t is only a true transitivizer if it functions as the sole transitivizer. If, however, it co-occurs with –Vn or –s, then –t appears to be redundant and in such cases Kuipers (1967:68) calls it an ―extender‖ for a following suffix (i.e. object or subject suffixes, or the passive suffix). But he also describes it as the same morpheme as the -t transitivizer (Kuipers 1967:259). This 55  suggests that the status of -t is unclear in such cases. In Appendix B §1, I provide a preliminary phonological account according to which –t is present morphologically in cases like (33)a and (34)a, and that it is simply deleted word finally due to phonotactic constraints. In anticipation of this analysis, then, I mark the presence of the second transitivizer with square brackets when I understand it to be present morphologically, but deleted phonologically. This is illustrated in (35) and (36):  (35) a.  lh w‘-an-t-Ø-as slap-TR-TR-3OBJ-3SUB ‗He slapped him.‘ na  RL  b.  (36) a.  b.  chen lh w‘-an-[ ]-Ø 1S.SUB slap-TR-TR-3OBJ ‗I slapped him.‘ t 7-s-t-Ø-as do-CAUS-TR-3OBJ-3SUB ‗He made a box.‘ na  ta  RL  DET  chen ta7-s-[ ]-Ø 1S.SUB do-CAUS-TR-3OBJ ‗I made a box.‘  ta  kw‘ xwa7 box  kw‘ xwa7 DET box  Another related puzzle regarding the putative extender function of -t is its absence with the lc-intransitivizer. That is, when the lc-transitivizer is present, the -t never occurs.  kw‘ach-nexw-(*t)-Ø-as see-LCTR-TR-3OBJ-3SUB ‗He saw it.‘  (37) na  RL  56  I will return to this issue in chapter 5, where I will argue for a unified analysis of –t. In particular, I will argue that –t plays a crucial role in encoding the event-structure associated with c-predicates. The limited control transitive predicates present us with another interesting feature. Kuipers (1967:68) describes the lc-transitive as having two allomorphs. When an overt object suffix (38)a-d or the passive (38)e follows it, then it surfaces just as -n. In any other context (i.e. when the object is third person), it surfaces as -nexw, whether an overt subject suffix follows (39)a or not (39)b.  (38) a.  m y-n-emsh-as forget-LCTR-1S.OBJ-3SUB ‗He forgot me.‘ na  RL  b.  m y-n-umulh-as forget-LCTR-1P.OBJ-3SUB ‗He forgot us.‘ na  RL  c.  chen kw‘ach-n-umi 1S.SUB see-LCTR-2S.OBJ ‗I saw you.‘  d.  chen kw‘ach-n-umi-yap 1S.SUB see-LCTR-2S.OBJ-PL ‗I saw you all.‘  b.  chen kw‘ach-n-m 1S.SUB see-LCTR-PASS ‗I was seen.‘  (39) a.  kw‘ ch-nexw-Ø-as see-LCTR-3OBJ-3SUB ‗He saw him.‘ na  RL  b.  chen kw‘ ch-nexw-Ø 1S.SUB see-LCTR-3OBJ ‗I saw him.‘ 57  This distribution raises the question of what phonological environment subsumes the overt object suffixes and the passive, yet excludes the subject suffixes. Does this merely have to be stipulated or can it be derived?  4.1.2.2  Applicative transitivizers  In addition to the core transitivizers, whose function appears to be grammatical transitivization, there are a number of transitivizers that add an extra argument (beyond agent and patient). These transitivizers are typically described as applicative transitivizers or simply as applicatives in the Salishanist literature (cf. Kiyosawa 2006, and Kiyosawa and Gerdts 2010, for a broad overview and analysis of Salish applicatives). I have adopted some of Kiyosawa‘s (2006:109 for -ni and 146 for -shi and ) labels for these applicatives, since Kuipers (1967:78-9) only describes all the applicatives as complex transitivizers. As per Kiyosawa, I label -nit as a relational applicative and -shit as a redirective applicative. I provide my own term for -min as a causative applicative, partly to differentiate it from -nit, since Kiyosawa only describes them both as relational applicatives. In S wxwu7mesh, -min also has a different semantics from –nit, although I do not explore this issue in this dissertation. The benefactive applicative - h‟ w n has not been described by Kuipers (1967) or others as such. I propose that it is also an applicative since it also allows for object agreement on the verb. The applicatives are summarized in Table 10 below.  58  Form Label -nit relational applicative (RELAPPL) -shit redirective applicative (REDAPPL) -min causative applicative (CAUSAPPL) -ch‘ewan benefactive applicative (BENAPPL) Table 10 Applicative transitivizers in S wxwu7mesh  Each of these applicatives is illustrated in examples (26)-(29) below. For each applicative I provide an example first with the bare root (the (a) examples) and then an example of the same root with the relevant applicative transitivizer (the (b) examples).  (40) a. -nit chen yew nts 1S.SUB understand ‗I understand.‘ b.  chen yew nts-nit-Ø 1S.SUB understand-RELAPPL-3OBJ ‗I understand him.‘  (41) a. -shit chen ts‘its‘ p‘ 1.SUB work ‗I worked.‘ b.  chen ts‘its‘ p‘-shit-Ø 1.SUB work-REDAPPL-3OBJ ‗I worked for him.‘  (42) a. -min chen tkwaya7n 1S.SUB listen ‗I listened/heard.‘ b.  chen tkwaya7n-Ø 1S.SUB listen-CAUSAPPL-OBJ ‗I listened to/heard him.‘ 59  (43) a. - h‟ w n chen l lum 1S.SUB sing ‗I sang.‘ b.  chen l lum- ’ -Ø 1S.SUB sing-BENAPPL-3OBJ ‗I sang for him.‘  As with any transitive clause in S wxwu7mesh, the clause with an applicative has only two positions for grammatical roles: subject and object. If there is another argument present, it must be realized as an oblique (cf. §4.4 on DP case). In the following example, the predicate has the redirective applicative -shit, and the object agreement is with the second person goal/recipient. The theme/patient, the car, is realized as an oblique argument. There is no agreement marking on the verb for this patient argument.  (44) chen sat-shit-umi t-ta 1S.SUB hand.over-REDAPPL-2OBJ OBL-DET ‗I gave the car to you.‘  t txwem car  While the relational and redirective applicatives contain the segment -t (45)a-b as part of their lexical entry, the causative and benefactive applicatives do not contain –t (45)c-d as a part of their lexical entry. Note for the causative transitivizer in (42)c and the benefactive transitivizer in (42)d, that they do not have the –t transitivizer following. Now compare those examples to those in (45)c-d below where these transitivizers do have the transitivizer –t following.  60  yew nts-nit-Ø-as RL understand-RELAPPL-3OBJ-3SUB ‗He understands him.‘  (45) a.  na  b.  na  ts‘its‘ p‘-shit-Ø-as work-REDAPPL-3OBJ-3SUB ‗He worked for him.‘ RL  c.  na  tkwaya7n-t-Ø-as listen-CAUSAPPL-TR-3OBJ-3SUB ‗He listened to him.‘ RL  d.  l lum- ’ -t-Ø-as sing-BENAPPL-TR-3OBJ-3SUB ‗He sang for him.‘ na  RL  This pattern for -t with the applicatives is reminiscent of the pattern we observed with the -t with the core transitivizers. Some of the applicatives have -t as part of their lexical entry and the other applicatives only have -t in certain contexts.  Core transitivizers Applicative transitivizers -t -nit -Vt -shit -Vn-[t] -min-[t] -s-[t] - h w n-[t] limited control -nexw Table 11 Transitivizers and applicatives with/without -t control  For some transitivizers the presence of -t appears to be obligatory. According to Kuipers (1967:78-9) the presence of -t is part of the lexical entry of -nit and -shit. That is, they are no longer synchronically analyzable as morphologically complex. Moreover, for those transitivizers where -t is only optionally present, there are two questions we need to consider: i) What is the function of -t and ii) what determines its distribution? I address 61  the first question in Chapter 5 and the second question in Appendix B, §2. In Chapter 5 I analyze it as the same -t that is present with the core transitivizers. In Appendix B, §1. I provide a preliminary phonological explanation for the distribution of –t, wherein –t is present underlyingly in cases like (42)b and (43)b but gets deleted due to phonotactic constraints.  4.1.2.3  Intransitivizers  Just like there are transitivizers in S wxwu7mesh which appear to be responsible for cand lc-marking, there are also intransitivizers that appear to be responsible for the same contrast in CONTROL. They are presented in Table 12.  Unergative (UE) -im  Reflexive (REFL) Control (C) -sut -n mut Limited control (LC) -nalhn -numut Table 12 Core intransitivizers in S wxwu7mesh  Reciprocal (RECIP) -way -n w s  First, note that the c-intransitivizers and the lc-intransitivizers are not transparently related to one another. The intransitivizers have their own morphological complexities. I discuss the lc-intransitivizers first and then the c-intransitivizers. The following are examples of the lc-intransitivizers.  62  (46) a.  b.  chen kwelash1S.SUB shoot-LCUE ‗I managed to shoot (it).‘ na RL  i) ii) c.  kwelash-Ø shoot-LCREFL-3SUB ‗He shot himself accidentally.‘ ‗He got to shoot.‘  chet kw‘achas 1P.SUB see-LCRECIP ‗We got to see each other.‘  The lc-intransitivizers appear to attach directly to the verbal stem. Each of the lcintransitivizers has an initial element n just like the lc-transitivizer -nexw. This leads to the question of whether the lc-transitivizer and lc-intransitivizers are further decomposable such that both sets would contain -n as a morpheme. And if so, what is its function? The description of the control intransitivizers is more complex. I present them in order from simpler to more complex. The control unergative –im attaches directly to the verbal stem.  (47) root + c-unergative a. nam chet exwt-kwi go 1P.SUB gather-CUE OBL-DET ‗We are going to collect bullrushes.‘ b.  sts‘ 7 in bullrush  pe-peh- -Ø ta spah m ti stsi7s RL IMPF RE-blow-CUE-3SUB DET wind DET today ‗The wind is blowing today.‘ na  wa  The c-reflexive -sut and the c-reciprocal -way do not attach directly to the verbal stem. They always attach to a c-predicate. The following are examples of the c63  reciprocal -way attached to a stem with the transitivizers -t (48)a, -Vt (48)b, -Vn (48)c and -s (48)d.  (48) transitive + c-reciprocal a. na wa kw‘awch-t-way-wit RL IMPF staring-TR-RECIP-PL ‗They are staring at one another.‘ b.  chet wa ch‘ w-at-way 1PL.SUB IMPF help-TR-CRECIP ‗We help one another.‘  c.  na  d.  lhi ‘-t na wa always-PAST RL IMPF  chema7-n-t-way-Ø-wit RL IMPF back.carry-TR-TR-CRECIP-3SUB-PL ‗They were piggy-backing each other.‘ wa  ch‘aw-ch‘ w-s-t-way RED-help-CAUS-TR-RECIP  kwekw n st mexw old people ‗The old people used to always help one another.‘ ta  DET  The reciprocal with the -Vn and -s also presents us with the same wrinkle as with the object suffixes. The -t occurs following the transitivizer. This raises the same question as we have with the object suffixes. Why do these transitivizers occur with a seemingly redundant transitivizer -t when the reciprocal follows? As for reflexives, there are two different control reflexives. The c-reflexive -sut occurs with a c-predicate constructed with one of the control transitivizers -t, -Vt or -Vn (49)a-c. The c-reflexive -n mut only occurs with a c-predicate with the causative transitivizer -s (50). I follow Dyck (2004:325-327) and Kuipers (1967:95, §137) in assuming that the schwa preceding the c-reflexive -n mut in (50) is epenthetic. Note that 64  the c-reflexive -n mut is similar to the lc-reflexive -numut as in (46)b. Historically they may have come from the same source, but synchronically they are different.  (49) transitivizer + c-reflexive a. na w 7-t-sut-Ø RL be.with-TR-CREFL-3SUB ‗He joined in.‘ b.  ch‘ w-at-sut chexw help-TR-CREFL 2S.SUB ‗Help yourself!‘  c.  na  xim-in-t-sut-Ø RL pull.hair-TR-TR-CREFL-3SUB ‗She grabbed her own hair.‘  (50) causative + c-reflexive wa chexw yuu-s-tIMFP 2S.SUB careful-CAUS-TR-REFL ‗Take care of yourself!‘  When the -Vn transitivizer and the causative -s are followed by the respective reflexive marker, the -t appears, with -Vn in (49)c and with -s in (50). This is reminiscent of the distribution of -t preceding object suffixes. This raises the question of why these two transitivizers require -t when they take the reflexive suffix. For completeness‘ sake I present the remaining intransitivizers, which I do not investigate in this dissertation.  65  Intransitivizers -m passive (passive) - m middle (MID1) -Vm middle (MID2) Table 13  Intransitivizers not investigated  It is in fact not clear if the first three intransitivizers are allomorphs of the same intransitivizer -m. Except for the passive –m, the control status of these intransitivizers has not been ascertained. The passive simply has the same control interpretation as the transitive base that it is attached to. The passivized c-predicate has a control interpretation in (51)a while the passivized lc-predicate has a limited control interpretation in (51)b.  (51) a.  b.  n-u chexw kw‘ach-t-m RL-POL 2S.SUB look-TR-PASS ‗ id he look at you?‘ n-u chexw RL-POL 2S.SUB ‗ id he see you?‘  kw‘ach-n-m look-LCTR-PASS  Below are examples of the two middle intransitivizers: the MID1 (52)a and the MID2 (52)b.  (52) a.  mikw‘-shnchexw clean-feet-MID 2S.SUB ‗ lean your feet!‘  66  b.  chen sh kw‘-um 1s.sub bathe-MID2 ‗I had a bath‘  4.1.2.4  The (in)transitivizers and the verbal template  In this section, I describe the (in)transitivizing system in S wxwu7mesh in terms of the verbal template. I identify some analytical issues that arise in this context. First, if -t, -Vn and -s are all transitivizers, then why can -Vn and -t co-occur, and why can -s and -t co-occur? In other words, what is the function of -t in (33)b and (34)b (repeated here as (53)a and (53)b respectively)?  (53) a.  -Vn-t na  lh w‘-an-t-Ø-as slap-TR-TR-3OBJ-3SUB ‗He slapped him.‘ RL  b.  -s-t na  ta  RL  DET  ta7-s-t-Ø-as do-CAUS-TR-3OBJ-3SUB ‗He made a box.‘  kw‘ xwa7 box  The possibility for two transitivizers to co-occur suggests that we need to revise the template of the S wxwu7mesh verb to accommodate for the distribution of -t. For the time being I will label both slots in the template as TR, indicating that both have to do with transitivity. The precise role of each slot will be the subject of investigation in chapter 5, where I argue that –t contributes to the aspectual reading of the predicate.  67  Revised template for the S wxwu7mesh verb: 2 transitivizers  (54)  STEM-TR1-TR2-OBJ-SUBJ-NUM -Vn -t -s -t  Since the other two c-transitivizers, -t and -Vt contain -t, I will associate them with the TR2 slot for the time being. This raises the question of why these two transitivizers do not have a transitivizer in the TR1 slot. In Chapter 5, §2.3, I propose that both of these constructions are two transitivizer constructions. For the constructions with the transitivizer -t I propose that they are composed of a zero transitivizer -Ø followed by the transitivizer -t; for the constructions with the transitivizers -Vt, I propose that they are composed of a transitivizer -V followed by the transitivizer -t. Revised template for the S wxwu7mesh verb: transitivizers only in TR2  (55)  STEM-TR1-TR2-OBJ-SUBJ-NUM -? -t -? -Vt  The intransitivizers present some further challenges with the revised template. The relevant examples of the lc-intransitivizers, from (46)a-c, are repeated here as (56)a-c.  (56) a.  b.  chen kwelash1S.SUB shoot-LCUE ‗I managed to shoot (it).‘ na  kwelash-Ø RL shoot-LCREFL-3SUB ‗He shot himself accidentally.‘ ‗He got to shoot.‘  68  c.  chet kw‘ach1P.SUB see-LCRECIP ‗We got to see each other.‘  The lc-intransitivizers appear to occur directly after the stem, in the same slot where the transitivizers occur. This is illustrated in (57):  (57)  Revised template for the S wxwu7mesh verb: lc-intransitivizers (in TR1 slot) STEM-TR1 -TR2-OBJ-SUBJ-NUM -LCINTR-  Recall that the c-unergative also attaches directly to the stem as shown in (47)a, repeated below as (58).  (58)  nam chet exwt-kwi go 1P.SUB gather-CUE OBL-DET ‗We are going to collect bullrushes.‘  sts‘ 7 in bullrush  The c-unergative also seems to occur right after the stem just as the transitivizers and lcintransitivizers do, suggesting that the c-unergative occurs in the same slot TR1. This is illustrated in (59):  (59)  Template for S wxwu7mesh verb: c-unergative (in TR slot) STEM-TR1 -TR2-OBJ-SUBJ-NUM -CUE-  Note that the appearance of the c-unergative precludes the presence of object agreement. 69  The c-reflexive -sut and the c-reciprocal -way, however, present a more complicated picture. Consider the c-reflexive -sut (48)c and the c-reciprocal -way (49)c, repeated here as (60)a and (60)b, respectively.  xim-in-t-sut-Ø RL pull.hair-TR-TR-CREFL-3SUB ‗She grabbed her own hair.‘  (60) a.  na  b.  na  chema7-n-t-way-Ø-wit RL IMPF back.carry-TR-TR-CRECIP-3SUB-PL ‗They were piggy-backing each other.‘ wa  They are attached after the TR2 rather than after the stem, suggesting that they occur in the same slot as the object agreement: (61)  Template for S wxwu7mesh verb: c-intransitivizer (in OBJect slot) STEM-TR1-TR2-OBJ -SUBJ-NUM -Vn -t -CREFL -Vn -t -CRECIP-  This raises the question as to why the c-reflexive and the c-reciprocal occur in the object slot while the lc-reflexive and the lc-reflexive attach directly after the stem in the same slot as the transitivizers? Suppose we assume that the lc-intransitivizers are associated with the object slot, just like the c-reciprocal and c-reflexives. If so, then the lcintransitivizers would not occur with an overt transitivizer. This is illustrated in (62): (62) Revised template for the S wxwu7mesh verb: lc-intransitivizers (in OBJect slot) STEM-TR1-TR2-OBJ -SUBJ-NUM -? -? -LCINTR70  This would leave us with the question of why the c-reciprocal and c-reflexive require an overt transitivizer while their counterparts, the lc-reciprocal and the lc-reflexive, do not? In Chapter 5, §6.2.1, I propose that the lc-intransitives are decomposable into a transitivizer -n followed by their respective intransitivizers. Another question that the template approach poses concerns the lc-transitivizer. Consider again (37)b repeated below as (63).  kw‘ach-nexw-Ø-as see-LCTR-3OBJ-3SUB ‗He saw it.‘  (63) na  RL  Again, this transitivizer does not occur with a second transitivizer. This could suggest that it occurs in the same slot as the other transitivizers. This is shown in (64):  (64) Template for S wxwu7mesh verb: lc-transitivizer STEM-TR1 -TR2-OBJ-SUBJ-NUM -LCTR-  A question then arises as to why the lc-transitivizer is the only transitivizer to not have -t, either as part of its lexical representation or as a second transitivizer. This may suggest that -nexw and -t occupy the same position (TR2) and are thus in complementary distribution. In Chapter 5, §4, I propose that -nexw is decomposable into the transitivizer -n followed by an overt agreement marker -exw for third person. I also provide a morphosyntactic account where object agreement is associated with the same syntactic position as the transitivizer -t. 71  Table 14 summarizes the templatic distribution of the transitivizers and Table 15 summarizes the templatic distribution of the intransitivizers we have developed so far. Transitivizer stem TR1 TR2 -t stem -t -Vt stem -Vt -nit stem -nit -shit stem -shit -Vn stem -Vn -t -s stem -s -t -min stem -min -t -ch‘ewan stem - h‟ w n -t -nexw stem -nexw Table 14 Templatic distribution of transitivizers  Object O O O O O O O O O  Suffix S S S S S S S S S  Intransitivizer lc-unergative lc-reflexive lc-reciprocal c-unergative c-reflexive  Object  Subject  stem stem stem stem stem stem  TR1 -nalhn -numut -n w s -im  TR2  -t -Vt stem -Vn -t c-reciprocal stem -t -Vt stem -Vn -t Table 15 Templatic distribution of intransitivizers  -sut -sut -way -way  This leaves us with those slots of the template which host person marking: object, subject and number marking. I will discuss these immediately below.  72  4.1.3  Person marking  In this section I provide a description of person marking in S wxwu7mesh. I begin with object agreement followed by subject agreement and then possessive marking, and finally the independent pronouns.  4.1.3.1  Object agreement  As described by Kuipers (1967:85-87), S wxwu7mesh has a set of object agreement suffixes with two distinct forms for first person singular. This is illustrated in Table 16.  1 2 3 Table 16  Singular Plural -ts -emsh -umulh -umi -umi-yap; -umi-wit -Ø -Ø(-wit) S wxwu7mesh object agreement suffixes (Kuipers 1967:85)  Object agreement is associated with the following slot in our template:  (65) Template for S wxwu7mesh verb: object agreement STEM-TR1-TR2-OBJect-SUBJ-NUM  The distribution of the two first singular agreement suffixes is determined by the preceding transitivizer that they occur with. The first singular object agreement -emsh is  73  only used with the limited control transitivizer -nexw (in its reduced form -n) and the suffix -ts is used with all other transitivizers. The following are some examples of these object suffixes: the first singular -ts (66)a, the first singular -emsh (66)b, the first plural -umulh (66)c, the second singular -umi (67)a, the second plural -umi-yap in both (67)b and (67)c, the third singular -Ø (68)a and the third plural -Ø-wit (66)b. Plurality is optionally marked for third person.  (66) a.  chexw kw‘ach-t-ts 2S.SUB look.at-TR-1S.OBJ ‗You looked at me.‘  b.  chexw kw‘ach-n-emsh 2S.SUB see-LCTR-1S.OBJ ‗You saw me.‘  c.  chexw ch‘aw-at-umulh 2S.SUB help-TR-1PL.OBJ ‗You helped us.‘  (67) a.  chen ch‘aw-at-umi 1S.SUB help-TR-2S.OBJ ‗I helped you.‘  b.  chen ch‘aw-at-umi-yap 1S.SUB help-TR-2S.OBJ-2PL ‗I helped you (pl.).‘  c.  chen ch‘aw-at-umi-wit 1S.SUB help-TR-2S.OBJ-PL ‗I helped you (pl.).‘  (68) a.  b.  chen ch‘aw-at-Ø 1S.SUB help-TR-3OBJ ‗I helped him/her.‘ chen ch‘aw-at-Ø(-wit) 1S.SUB help-TR-3OBJ-PL ‗I helped them.‘ 74  A few further notes about plural marking and object agreement are in order. The main point of this examination is to motivate the analysis that plural agreement is separate from person agreement. There are two plural markers, -yap which only occurs with second person arguments and –wit which occurs with second and third person arguments. The first person plural object, however, does not co-occur with a separate plural marker. The plural -yap also occurs with the second person plural possessives (69)a and with the second person independent pronoun (69)b.  (69) a.  a-s wem y-yap 2.POS-dog-2.PL ‗your (pl) dog‘ ta  DET  b.  ta  new-yap 2S.INDP -2.PL ‗You all‘ DET  When -yap is used to agree with the object, it may either directly follow the second person agreement marker –umi (70)a or it can optionally appear after subject clitic as in (70)b.12  (70) a.  lh ‘i7-s-t-umi-yap chan know-CAUS-TR-2S.OBJ-2.PL 1S.SUB ‗I know you (pl).‘  12  Closely related Sechelt has a very similar pattern with its second person plural agreement marker -elap ( eaumont 1985:83-87 ) cognate to S wxwu7mesh -yap.  75  b.  lh ‘i7-s-t-umi chan-yap know-CAUS-TR-2S.OBJ 1S.SUB -2.PL ‗I know you (pl).‘  The plural marker –wit may either mark plurality of the object (71)a or the subject (71)b (cf. Bar-el, Jacobs and Wiltschko 2001 for a more detailed description).13  (71) a.  b.  chen s xwt-nexw-Ø-wit 1S.SUB recognize-LCTR-3OBJ-PL ‗I recognize them.‘ suxwt-n-emsh- s-wit RL recognize-LCTR-1S.OBJ-3SUB-PL ‗They recognize me.‘ na  Plural –wit does not normally occur when a plural DP is present, though (72).  suxwt-n-emsh-as-(*wit) t-en s iyay recognize-LCTR-1S.OBJ-3SUB-(PL) DET-1S.POS friends ‗My friends recognized me.‘  (72) na  RL  Furthermore, -wit can occur as an alternative to mark plural second person objects instead of -yap; compare (73)a to (73)b (repeated from (70)b). (73) a.  lh ‘i7-s-t-umi know-CAUS-TR-2S.OBJ ‗I know you (pl).‘  chan-wit 1S.SUB-PL  13  Roberts (1999) and Davis (2003) provide analysess of -wit in Lillooet, the cognate to the S wxwu7mesh -wit. The distribution of -wit in Lillooet appears to differ from -wit in S wxwu7mesh, and I leave a fuller comparision to future research.  76  b.  lh ‘i7-s-t-umi chan-yap know-CAUS-TR-2S.OBJ 1S.SUB -2.PL ‗I know you (pl).‘  Furthermore, the plural marker -wit does not have the same status as the other parts of the template, in that it can encliticize to a clitic which precedes the verb (compare (69)c-d to (74)a-b).  (74) a.  b.  chen-wit kw‘ach-nexw-Ø 1S.SUB-PL see-LCTR-3OBJ ‗I saw them.‘ chen kw‘ach-nexw-Ø-wit 1S.SUB see-LCTR-3OBJ-PL ‗I saw them.‘  In contrast, the second person plural -yap cannot encliticize to a clitic preceding the verb (75).  (75) chen-(*yap) kw‘ach-n-umi 1S.SUB-2PL see-LCTR-2S.OBJ ‗I saw you all.‘  When -wit occurs in a clause with two third person null arguments, it is ambiguous between a reading where it pluralizes a subject or an object (53).  77  lh ‘i7-s-t-Ø-as-wit RL know-CAUS-TR-3OBJ-3SUB-PL ‗They know him he knows them.‘  (76) na  As already mentioned, the first person plural suffix is not further segmentable into person and number components (compare (77)a and (77)b).14  (77) a.  b.  *kw‘ach-t-um look-TR-1PL.OBJ ‗Look at us!‘  chexw-ulh 2S.SUB-1PL.OBJ  kw‘ach-tlook-TR-1PL.OBJ ‗Look at us!‘  chexw 2S.SUB  Because of these facts on the separability of the plural markers -yap and -wit, I treat second and third person plural marking as separate from the object agreement and part of the NUM slot as proposed in the template for the S wxwu7mesh verb, represented such as:  14  Historically it may have contained an older form of the 1 st plural possessive, since the independent possessive has the –ulh part in common: nilh s7ulh lam FOC 1P.INDP.POSS house ‗That‘s our house‘  78  (78) Template for S wxwu7mesh verb: Plural marking STEM-TR1-TR2-OBJ-SUBJ-NUM -2PL -3PL  With plural second and third object agreement separated from the object agreement, we have the following revised object agreement paradigm.  1 2 3 Table 17  4.1.3.2  -ts  Singular -emsh  Plural -umulh  -umi -Ø S wxwu7mesh object agreement suffixes - revised (Jacobs 2011)  Subject agreement  S wxwu7mesh subject agreement is conditioned by clause typing (cf. Jacobs 1992): i) matrix clauses, ii) conjunctive clauses, iii) factive clauses, and iv) nominalized clauses (Kuipers 1967:85-93). Since most of the examples in this dissertation are matrix clauses and conjunctive clauses, I only focus on these two types of subject agreement. In the matrix clause, S wxwu7mesh has been described as having a split ergative system (Jacobs 1992). Third person is marked in an ergative/absolutive alignment by means of suffixes on the verb. First and second persons are marked in a nominative/accusative alignment by means of clitics. This is summarized in Table 18.  79  Person  Subject of Transitive Subject of Intransitive Object of Transitive (S) (A) (O) Subject clitics Object suffixes -as Ø Types of subject/object agreement in S wxwu7mesh  1st and 2nd 3rd Table 18  The third person subject of a transitive predicate is marked by the suffix -as on the verb, as shown in (79)a.15 The absolutive argument has the null agreement marker -Ø: that is, the third person object of a transitive, as in (79)b, and the third person subject of an intransitive, as in (79)c.  (79) a.  b.  c.  ch‘ w-at-s-as help-TR-1S.OBJ-3SUB ‗The woman helped me.‘ na  lha  RL  DET  chen kw‘ ch-nexw-Ø RL see-LCTR-3OBJ ‗I saw the bear.‘  ta  huy 7-Ø RL leave-3SUB ‗The bear left.‘  m xalh bear  na  ta DET  DET  slh nay woman  m xalh bear  From this data, we can conclude that at least third person is marked following object marking on the verb. This can be represented in our template as in (80):  15  I refrain from glossing –as as ergative, since my reanalysis in Chapter 5 ultimately has two different types of agreement for third person objects, making either a nominative/accusative or ergative/absolutive description problematic. I simply gloss –as as third person subject.  80  (80) Template for S wxwu7mesh verb: Plural marking STEM-TR1-TR2-OBJ-SUBJ-NUM -third  First and second person subjects are marked by nominative subject clitics, as shown in Table 19 below.16  Singular  Plural  1st person chen/chan chet/chat nd 2 person chexw/chaxw chap/chayap Table 19 Nominative subject clitics  For example, take the first and second persons singular. They are marked by nominative subject clitic both for the subject of the transitive, as in (81)a and (82)a and for the subject of an intransitive clause, as in (81)b and (82)b. Meanwhile the first person singular and second person object are marked by object agreement suffixes (as discussed in § 4.1.3.1) on the verb, as in (81)c and (82)c, respectively)  (81) a.  b.  16  chen ch‘ w-at-umi 1S.SUB help-TR-2S.OBJ ‗I helped you.‘ chen huy 7 1S.SUB leave ‗ I left.‘  In §4.5 I discuss the syntax of these subject clitics.  81  c.  (82) a.  kw‘ach-n-emsh-as RL see-LCTR-1S.OBJ-3SUB ‗He saw me.‘` na  chexw ch‘ w-at-Ø 2S.SUB help-TR-3OBJ ‗You helped him.‘  b.  chexw huy 7 2S.SUB leave ‗ You left.‘  c.  chen kw‘ach-n-umi 1S.SUB see-LCTR-2S.OBJ ‗I saw him.  Note that all the subject clitics begin with ch-. This suggests the possibility of decomposing these clitics: for example, Kuipers (1967:156, §223) analyzes ch- as a clitic which acts essentially as a kind of do-support, thus relating it to the verbal root cha- ‗to do to act to make‘. The remaining part of the subject clitics usually occurs in its reduced form (i.e., the vowel is a schwa) as in (83)a and (84)a. However, they can also occur in non-reduced form with the full vowel /a/. The full form is typically used for verum focus as in (83)b and (84)b. For the purpose of this dissertation I treat subject clitics as unanalyzable units since nothing bears on their decomposition.  (83) a.  b.  chen ts‘its‘ p‘ 1S.SUBJ work ‗I worked, I am working, I work.‘ chan ts‘its‘ p‘ 1S.SUB work ‗I did work.‘ 82  (84) a.  b.  chet ts‘its‘ p‘ 1P.SUBJ work ‗We worked we are working we work. chat ts‘its‘ p‘ 1PL.SUBJ work ‗We did work.‘  The second person plural, though, does not have a reduced form. The two forms differ in register. The longer form chayap is more likely to occur in formal settings, while the shorter form chap is used in both formal and informal settings. From these examples, we can conclude that these subject clitics are not part our verbal template. Conjunctive subject clitics are marked in a nominative/accusative alignment for all three persons as shown in the following table. In the data in this dissertation they are always attached to the subjunctive marker (which is further glottalized when the conjunctive subject clitic is attached). This is not always the case, but I do not explore this issue in this dissertation.  Singular st  1 person 2nd person 3rd person Table 20  Plural  ‟-an ‟-at ‟-axw ‟- p, ‟-ayap ‟-as ‟-as(-wit) onjunctive subject clitics in S wxwu7mesh  In the data in this dissertation, the conjunctive subject clitics are always used in the context of negation as in (85). They can, however, occur in a variety of other constructions which I do not explore here. 83  ‘-an NEG SBJ-1S.CONJ ‗I didn‘t work.‘  (85) a.  haw  b.  haw  ‘-as  i PRES  i  NEG SBJ-3CONJ PRES  ts‘its‘ p‘. work  lhen. eat  ‗S/he didn‘t eat.‘  4.1.3.3  Possessive marking  The possessive affixes are summarized in Table 21.17  Singular  Plural  1  n-  -chet  2  7a-  7a- -yap  3  -s  -s-wit  Table 21  Possessive marking in S wxwu7mesh  The possessives in S wxwu7mesh occur either preceding or following the nominal head, depending on the particular affix. The first and second person singular possessives precede the head. The first plural and third person are suffixed to the head. The second person plural possessive is comprised of the second person possessive a- preceding the  17  I include the initial glottal stop here for the second person possessives. Note, though, that in the practical orthography that it is normally not written.  84  head and the plural morpheme -yap following the head. This is illustrated in (86) with the noun h sh 7 (‗mother‘).  n-ch sha7 ‗my mother‘  d.  ch sha7-chet ‗our mother‘  b  7a-ch sha7 ‗your (sg.) mother‘  e.  7a-ch sha7-yap ‗your (pl.) mother‘  c.  ch sha7-s ‗his/her mother‘  f.  ch sha7-s-wit ‗their mother‘  (86) a.  While the first and second person singular possessives can occur directly attached to the head as in (87)a and (88)a, they often encliticized to the preceding determiner as in (87)b and (88)b:  (87) a.  ets m small ‗my small cat‘ ta  DET  b.  (88) a.  te-n ets m DET-1S.POS small ‗my small cat‘  n-push 1S.POS-cat  push cat  hiy 7a-sna big 2S.POS-name ‗your great name‘ ta  DET  b.  ta-7a hiy sna DET-2S.POS big name ‗your great name‘  85  4.2  Verb phrase auxiliaries and particles  S wxwu7mesh has approximately 20 auxiliaries, particles and clitics in the verb phrase. Kuipers (1967:155-164) describes them all as clitics. They include the person clitics described in the previous section. The distribution of the remaining particles has not been well researched. In this section I discuss two of the auxiliaries, and two particles which have variable syntax. Much of the data discussed in this dissertation have the auxiliaries na (89)a or i (89)b. In the S wxwu7mesh practical orthography, these auxiliaries are written as separate words. I follow this convention in this dissertation.  (89) a.  b.  ts‘its‘ p‘-Ø RL IMPF work-3SUB ‗He is/was working.‘ na  wa  ts‘its‘ p‘-Ø PRES IMPF work-3SUB ‗He is working.‘ i  wa  Kuipers (1967:155-158) describes na and i as deictic clitics because they can occur as full predicates on their own when they have a deictic meaning: na7 ‗be located over there‘ (90)a and i ‗be located here‘ (90)b.  (90) a.  chen wa na7 1S.SUB IMPF be.there ‗I was there.‘  t-kwetsi OBL-DEM  86  b.  chen wa i t-ti 1S.SUB IMPF be.here OBL-DET ‗I am here.‘  The semantics of the auxiliaries na and i is not well understood nor researched. Ritter and Wiltschko (2005) analyze their cognates in Halkomelem as locative auxiliaries which serve as the functional equivalent of tense. I simply gloss these clitics as realis (RL) and present (PRES) respectively. Two important particles are the two tense morphemes: -t past tense (91)a-b and future tense (92)a-c.  (91) a.  b.  (92) a.  chen-t ts‘its‘ p‘ 1S.SUB-PAST work ‗I worked.‘ chexw-t wa ncha? 2S.SUB-PAST IMPF where ‗Where have you been?‘ ts‘its‘ p‘ chen work 1S.SUB ‗I will work.‘  ’ FUT  b.  chen ’ ts‘its‘ p‘ 1S.SUB FUT work ‗I will work.‘  c.  silha7- n-[ ]-Ø u chexw buy-TR-TR-3OBJ POL 2S.SUB ‗Will you buy it?‘  ’? FUT  87  ‟  These tense morphemes are particles with variable syntax. When a nominative subject clitic is present, they are encliticized to them as in (91) and (92). In the absence of a subject clitic (i.e., when the subject is third person), the tense morphemes attach to the first word in the sentence. For the past tense -t, this is usually one of the auxiliaries na or i (93)a-b, while for the future tense this is usually the verb in, as in (94)a, or an adverb, as in (94)b.  (93) a.  b.  na7(-t) wa ts‘its‘ p‘-Ø RL-PAST IMPF work-3SUB ‗He was working he used to work.‘ es w‘ y-Ø sick ‗He has been sick.‘ i(-t)  wa  PRES-PAST IMPF  (94) a.  b.  xwekw-s-t-Ø-as *( ’) use-CAUS-TR-3OBJ-3SUB FUT ‗He will use it.‘ a7 s *( ’) soon FUT ‗It won‘t be long ‘ ‗He‘ll be right back.‘  The properties of the tense morphemes have not been well researched (see Bar-el et al. 2004 for an examination of word order and tense effects). Note that while the past tense morpheme often seems to be optional, unless required to disambiguate, the future tense morpheme seems to be almost obligatory. Thus while the above sentences (93)a-b are acceptable without the past tense morpheme, the two future tense sentences (94)a-b are not. 88  4.3  Determiner phrases  A determiner phrase (henceforth DP) minimally consists of a determiner and a noun phrase, as illustrated in (95). In S wxwu7mesh, all noun phrases are obligatorily preceded by an overt determiner as illustrated in (96)a-b.  (95) a.  sw 7 a man ‗the/a man‘ ta  DET  b.  sh nay woman ‗the/a woman‘ lha  DET  (96) a.  chen kw‘ ch-nexw-Ø 1S.SUB see-LCTR-3OBJ ‗I saw the/a man.‘  *(ta) sw 7 a DET man  b.  chen kw‘ ch-nexw-Ø 1S.SUB see-LCTR-3OBJ ‗I saw the/a woman.‘  *(lha) DET  slh nay woman  S wxwu7mesh determiners are marked for several features: case (direct vs. oblique),18 gender (feminine vs. gender-neutral), deixis (neutral, proximal, medial, distal vs. nondeictic) as in Table 22. The S wxwu7mesh demonstratives mark case (direct vs. oblique), gender (feminine vs. gender-neutral), number (singular, number-neutral vs. plural) and deixis (neutral, proximal, medial vs. distal) as in Table 23. For a fuller  18  See §4.4 following for a description of these terms.  89  discussion of the syntax and semantics of the S wxwu7mesh determiners and demonstratives see Gillon (2006, 2009).  Neutral  Deictic Proximal  Nondeictic  Distal, invisible gender-neutral ta ti kwa kwi feminine lha tsi kw lh kwes Table 22 The determiner system of S wxwu7mesh (adapted from Gillon 2009)  gender-neutral  Table 23  numberneutral plural  Neutral, invisible kw y  Proximal  Medial  t,tw  t y  Distal Unmarked Invisible kwetsi  kwiy wit iy wit) iytsi wit) kw tsiwit kw wit feminine kw s ts w lhi kw lhi The demonstrative system of S wxwu7mesh (adapted from Gillon 2009)  S wxwu7mesh is a pro-drop language. DPs themselves are not obligatory and third person arguments are often realized by what I assume is a null pronoun. In the following example, there are no overt DPs for the third person arguments. While I assume a null third person pronoun in these cases, I do not include them in the glosses.  (97) a.  s xwt-nexw-Ø-as recognize-LCTR-3OBJ-3SUB ‗S/he recognizes it/him/her.‘ na  RL  90  b.  chen s xwt-nexw-Ø 1S.SUB recognize-LCTR-3OBJ ‗I recognize it/him/her.‘  c.  na  suxwt-n-emsh-as recognize-LCTR-1S.OBJ-3SUB ‗S/he recognizes me.‘ RL  4.4  Case  Kuipers (1967:136) describes two cases for the DP in S wxwu7mesh: absolutive and relative case. Similar case types in other Coast Salish languages are sometimes labelled as direct case and oblique case (Kroeber 1999:37), and these are the terms that I will use for them in this dissertation. The corresponding markers for these cases are given in Table 24.  direct case DPs Ø oblique case toblique case with: tl‟ i) proper names ii) first and second person independent pronouns Table 24 Case marking for DPs in S wxwu7mesh  Direct case is not overtly marked. It is the case that occurs with both subject and direct object DPs. Take, for example, a transitive clause with two third person arguments in (98). Here both the subject DP and the object DP are in the direct case and hence not overtly marked for case.  91  kw‘ach-nexw-Ø-as RL see-LCTR-3OBJ-3SUB ‗The man saw a rabbit.‘  (98) na  ta DET  sw 7 a ta ns-h hupit man DET rabbit  The oblique case marker t- attaches only to a determiner preceding a common noun, while tl‟ only attaches to proper names or to first or second person independent pronouns. The distribution of oblique case t- is summarized and exemplified below in (99)-(104). It occurs with: i)  the agent of a passive: kw‘ach-n-em t-kwa men-s RL see-LCTR-PASS OBL-DET child-3POS ‗He was seen by his son ‘ ‗His son saw him.‘  (99) na  ii) the object of an unergative verb: (100) chen wa exw- m 1S.SUB IMPF gather-CUE ‗I‘m gathering bulrushes.‘  sts‘ 7 in. bullrush  t-kwi OBL-DET  iii) the second object of a ditransitive: (101) chen tsexw-shit-Ø ta sw wlus 1S.SUB get.hit-RDIR-3OBJ DET young.man ‗I threw the ball to the young man.‘  iv) the object of a prepositional verb: (102) chen wa na7 1S.SUB IMPF be.at ‗I was at the beach.‘  t-kwa OBL-DET  yalh w beach  92  t-kwetsi OBL-DEM  sch‘ w la ball  v) the object of a prepositional verb (txwnew) functioning like a preposition: (103) tsexw-s-[ ]-Ø chexw txw-new t-ta get.hit-CAUS-TR-3OBJ 2S.SUB OOC-inside OBL-DET ‗Throw it into the net.‘  switn net  vi) an instrument: w‘ w-ut-ts-as RL hit-TR-1SG.OBJ-3SUB ‗He hit me with an axe.‘  (104) na  t-kwetsi OBL-DEM  w‘ w men axe  The distribution of oblique marker tl‟ is summarized and exemplified below in . It occurs with: i)  the agent of a passive  (105) kw‘ach-n-em ’ Asxw RL see-LCTR-PASS OBL/DET Seal ‗He was seen by Seal ‘ ‗Seal saw him.‘  ii) the second object of a ditransitive: kw‘ach-mixw-s-t-s-as-wit RL see-person-CAUS-TR-1S.OBJ-3SUB-PL ‗They showed you to me.‘  (106) na  ’ OBL/DET  iii) the object of a prepositional verb  xwey kwelhi na7 tl‘a St‘ 7mes RL born dem at obl/det St‘a7mes ‗She was born at St‘a7mes (a village near Squamish).‘  (107) na  93  new 2S.INDP  The oblique marker tl‟ does not occur with object of an unergative, nor with instruments.  4.5  Word order  Word order in S wxwu7mesh differs in matrix clauses depending on whether the subject is a speech act participant (that is, first or second person) or not (that is, third person). As noted in §4.1.3, first and second person subjects are encoded with subject clitics and third person is marked by agreement on the verb. The word order for subject clitics is as follows: they either appear at the beginning of the clause (108)a or following the first word (108)b.  (108) a.  b.  chen ts‘its‘ap‘ 1S.SUB work ‗I work I am working I worked.‘ ts‘its‘ p‘ chen work 1S.SUB ‗I will work.‘  The word order possibilities for DPs in a transitive clause are as follows. A clause with VP DP DP can be interpreted as VSO (reading (i) of (109)a) or VOS (reading (ii) of (109)a). A clause with DP VP DP can only be interpreted as SVO (reading (i) of (109)b), but not OVS (reading (ii) of (109)b). The subject DP of an intransitive clause can appear either following the verb (110)a or preceding the verb (110)b:  94  (109) a.  b.  (110) a.  VSO or VOS na ch‘ w-at-Ø-as lha slh nay RL help-TR-3OBJ-3SUB DET woman i) ‗The woman helped the man ‘ ii) ‗The man helped the woman.‘ SVO but not OVS lha slh nay na ch‘ w-at-Ø-as DET woman RL help-TR-3OBJ-3SUB i) ‗The woman helped the man.‘ ii) *‗The man helped the woman.‘  ta DET  sw 7 a man  ta DET  sw 7 a man  VS na  ts‘its‘ p‘-Ø ta work-3SUB DET man ‗The man worked.‘ RL  b.  SV ta  na  man RL ‗The man worked.‘ DET  ts‘its‘ p‘-Ø work-3SUB  Certain tense effects have been noted for these different word orders with subject clitics (Bar-el et al. 2004, Currie 1997, Jacobs 1992, Kuipers 1967:172). When a subject clitic precedes the verb, a past or present tense reading is obtained (111)a-b. But, when the subject clitic follows the verb, a future reading is obtained (112)a-b.  (111) SV a.  b.  chen lhen 1S.SUB eat ‗I ate I am eating.‘ chet ts‘its‘ p‘ 1P.SUB work ‗We worked we are working.‘  95  (112) VS a.  b.  lhen chen eat 1S.SUB ‗I‘ll eat.‘ ts‘its‘ p‘ chet work 1P.SUB ‗We‘ll work.‘  Such tense effects are not obtained with overt subject DPs. The positioning of the DP does not have any tense effect. The tense of the clause is the same when the DP follows the verb (110)a as when it precedes the verb (110)b. (113) a.  VS na i) ii)  ts‘its‘ p‘-Ø ta work-3SUB DET man ‗The man worked.‘ *‗The man will work.‘  SV ta  na  RL  b.  man RL ‗The man worked.‘ DET  ts‘its‘ p‘-Ø work-3SUB  96  Chapter 3: The core meaning of CONTROL  1  Introduction  As discussed in Chapter 1, the central question for this dissertation is: how do we account for the semantic contrast associated with the marking of CONTROL by means of the transitive and intransitive markers? Consider again the examples in (1) which appear to indicate that CONTROL is about the degree of control an agent has over the event.  (1)  a.  chen kw‘lh-at-Ø ta sta w 1S.SUB pour-TR-3OBJ DET water ‗I poured the water.‘ (on purpose)  b.  chen kw‘ lh-nexw-Ø ta sta w 1S.SUB pour-LCTR-3OBJ DET water i) ‗I spilt the water.‘ (accidentally) ii) ‗I managed to pour the water.‘  As discussed in chapter 1, the degree of control does not always seem to be an inherent part of the meaning of such predicates. When speakers translate c- or lc-predicates, they often do not translate them into English with the adverbs on purpose or accidentally. This, of course, could mean that the phenomenon of CONTROL is not easily translatable into English. I take the fact that these adverbs are often not used to translate control constructions to at least suggest that these adverbial meanings (and other adverbial  97  meanings used to translate control constructions) are not part of the lexical entry of any (in)transitivizer. Moreover, there is another difference that correlates with c- and lc-marking which has nothing to do with the degree of control the agent has over the event. In particular, cpredicates, in past perfective contexts (as first noted by Bar-el et al. 2005 and Bar-el 2005), do not require event culmination but lc-predicates do. This is illustrated in (2).  (2)  a.  c-predicate does not require culmination chen kwélash-t-Ø ta míxalh, 1S.SUB shoot-TR-3OBJ DET bear ‗I shot the bear ‘ welh na t‘emt‘ m te-n skw lash but RL astray DET-1S.POS shot ‗but I missed (lit. my shot went astray).‘  b.  lc-predicate requires culmination chen kwélash-nexw-Ø ta míxalh, 1S.SUB shoot-LCTR-3OBJ DET bear ‗I shot the bear ‘ #welh na t‘emt‘ m but RL astray ‗but I missed.‘  te-n DET-1S.POS  skw lash shot  The example in (2)a shows that it is possible to use the c-predicate even if the event does not culminate as intended, i.e., if the bear did not actually get shot. In other words, the cpredicate in (2)a can be felicitously uttered if the natural endpoint of the event is not reached. In contrast, the example in (2)b with the lc-predicate shows that it is not possible to deny the culmination of the event without inducing a contradiction. The bear must be 98  shot. Thus, the lc-predicate can only be uttered felicitously if the natural endpoint of the event is reached. For the rest of this chapter I will use the term event culmination (or simply culmination) to refer to this type of meaning. The goal of this chapter, then, is to show that this difference in event culmination is one of the primary meaning contrasts between c- and lc-predicates. Thus, c-predicates do not require event culmination, whereas lc-predicates do. I examine all types of predicates that have been analyzed as being marked for either control or limited control by means of different transitivizers and intransitivizers. In particular, I investigate whether or not they require culmination. We will see that all of the lc-predicates require event culmination whereas none of the c-predicates do. I also show that all that is required to felicitously use a c-predicate is that the event was initiated. It does not commit the speaker to anything regarding the end of the event (whether it culminated or not). However, note that c-predicates can still be used if the event has culminated, and especially in out-ofthe-blue contexts culmination is implied. This is summarized in (3).  (3) Core contrast associated with CONTROL i) lc-predicate: event culmination necessary ii) c-predicate: event culmination possible  I begin, in §2, by reviewing Bar-el‘s (2005) investigation of culmination in S wxwu7mesh. I present the four diagnostics she uses to test if a verbal predicate entails culmination. I then present a summary of her findings. Crucially, however, Bar-el does not systematically investigate lc-predicates; nor does she systematically investigate each type of c-predicate. Furthermore, she does not investigate the properties of c-predicates 99  marked by the -Vt transitivizer. This present section fills these gaps. The systematic investigation of all CONTROL predicates will help provide the basis for another goal of this present study: to understand the relation between verbal morphology and CONTROL, a topic I return to in Chapter 5. Thus, in (§3) I systematically investigate all four predicates identified in chapter 2 as c-transitives as well as the lc-transitives. I do not only investigate the core transitivizers, but also the intransitives (§4) (with the exception of cand lc-reciprocals and a type of c-reflexive). I, then, examine in (§5) the applicatives with regards to culmination entailments. In (§6) I examine the c-predicates with regards to culmination implicatures (that is, where a c-predicates implies culmination or not). In §7 I provide a summary of our findings.  2  Background  Bar-el (2005) argues for the existence of four different predicate classes in S wxwu7mesh: activities, accomplishments, achievements and inchoative states. These predicate classes are based on the verb classes originally introduced by Vendler: activities, accomplishments, achievements, and states (Vendler 1967). Bar-el‘s study in part, tests the cross-linguistic validity of these predicate classes. As a result of her findings, Bar-el proposes a number of modifications to the standard definitions of these predicate classes in order to account for the properties of their S wxwu7mesh  100  equivalents.19 In particular, Bar-el‘s primary claim is that S wxwu7mesh predicate classes are defined by the absence or presence of initial and final event points in their semantic representation (Smith 1997 and Rothstein 2004). Initial and final points are defined as basic parts of the event structure of a verbal predicate.20 In order to determine whether a verb has inherent initial or final points, Bar-el (2005:67-75) develops various diagnostics. They are summarized in (4). The main focus of my study is on culmination and thus final points and I will only consider her diagnostics for final points. In the summary section of this chapter, §7, however, I return to the issue of inherent initial points and how this relates to c-predicates (which I argue are initiating predicates).  (4) Diagnosing final points in S wxwu7mesh (Bar-el 2005:64-74) i) The culmination cancellation test If a predicate encodes event culmination, then adding a sentence that indicates that culmination did not take place is infelicitous. ii)  The event continuation test If a predicate encodes event culmination, then adding a sentence that asserts that the event (may have) continued is infelicitous.  iii)  The scope of ilh ‗almost‘ If a predicate encodes event culmination then ilh may take scope over the final point alone, indicating that the event almost culminated (i.e., it started).  19  Bar-el does not explore the properties of homogeneous states in S wxwu7mesh. See Kiyota (2008) for an analysis of homogeneous states in Saanich. 20 A verb with an inherent initial point has an initial BECOME subevent as in (i), whereas a verb with an inherent final point has a final BECOME subevent in its denotation., as in (ii). (i)  Initial point:  (ii) Final point:  λe.Ǝe1Ǝe2[e=s(e1⊔e2) ˄ (BECOME(P))(e1) ˄ (DO(P)(e2)] λe.Ǝe1Ǝe2[e=s(e1⊔e2) ˄ (DO(P)(e1) ˄ (BECOME(Q))(e2)] (Bar-el 2005, ex. 9a-b, 8)  101  If a predicate does not encode culmination, then ilh takes scope over the whole event, indicating that the event almost started. iv)  The scope of negation: If a predicate encodes event culmination then negation may take scope over the final point alone, indicating that the event started but did not culminate. If a predicate does not encode culmination, then negation takes scope over the whole event, indicating that the event did not start.  Table 25 is a summary of Bar-el‘s results from the first two tests and Table 26 is a summary of the second two tests (the scopal tests). In this dissertation I also call the first two tests the culmination entailment tests since they test whether a predicate in the pastperfective has to culminate in the real world. I also call the second two tests the scopal tests since they use scopal operators to test for inherent points.  Test 1 Test 2 Final Point Culmination Cancellation Event Continuation Conjunctions Questions Conjunctions Questions Activity ✓ ✓ ✓ ✓ ✘ Accomplishment ✓ ✓ ✓ ✓ ✘ Achievement ✘ ✘ ✓ Inchoative State ✓ ✓ ✘ Table 25 Culmination cancellation and event continuation (Bar-el 2005:135, ex. 137) (✘=infelicitous; ✓=felicitous; - =data not yet tested)  102  Readings induced by scope tests Test 3: The Scope of ilh (‗almost‘) Test 4: The Scope of Negation Test Test Event Event NonEvent Event NonCancellation completion Cancellation completion (= almost started) (=almost (=did not (= did not culminated) start) culminate) Activities ✓ ✘ ✓ ✘ Accomplishments ✓ ✘ ✓ ✘ Achievements ✓ ✘ ✓ ✘ Inchoative ✓ ✘ ✓ ✘ States Table 26 Event Cancellation vs. Event Continuation (Bar-el 2005:136, ex. 138) (✘=infelicitous; ✓=felicitous)  As summarized in Table 25, the culmination cancellation tests shows that only achievements do not allow for cancellation of culmination (Bar-el did not test inchoative states). She concludes that achievements, and only achievements have inherent final points. Note that Bar-el for the Test 1 and Test 2 used both conjunctions (a conjoined sentence) and questions to test if it is possible to deny the culmination of a given predicate. Since both conjunctions and questions always have the same result, I only use the conjunction test for Test 1 and Test 2. As for the scopal tests in Table 26, the scope of ilh (‗almost‘) test obtained an almost started reading (i.e. the event cancellation reading) for all four predicate types and not an almost culminated reading (i.e. the event non-completion reading). The scope of negation test obtained a did not start reading (i.e. the event cancellation reading) for all four predicate types and not a did not culminate reading (i.e. the event non-completion reading). Bar-el concludes that ilh (‗almost‘) is taking scope over the whole event. Consequently - she argues - activities, accomplishments and inchoative states do not have 103  inherent final points. If they did, then these scopal tests should allow for a reading wherein the event started and it almost culminated. Bar-el argues that for achievement these scopal tests indicate that achievements only have one subevent, BECOME subevent, and this subevent tests as both an inherent initial point and an inherent final point. After also exploring tests for initial points, Bar-el (2005) argues for the following classification for activities, accomplishment, achievement and inchoative state predicates in S wxwu7mesh, with regards to both inherent initial and final points.  Initial point Final point   activities   accomplishments   achievements   inchoative states Table 27 S wxw 7mesh predicates: initial and final points (Bar-el 2005:200)  Again, I do not explore initial points in this dissertation, but I do claim that c-predicates are I(nitiating) predicates. I also claim that this initiating property is different from predicates with inherent initial points. This issue is discussed in the conclusion in (§7). Bar-el also provides a formal semantic analysis of the –Vt, and –Vn transitivizers, the transitivizers present in most of her accomplishment predicates, an analysis which will I discuss more fully in Chapter 5, (§5.3), after providing my own analysis. She demonstrates that unaccusative roots have culmination entailments. She also assumes that c-transitives are all are derived from unaccusative roots following Davis (1997). Following Matthewson (2004a), Bar-el then argues that c-marking in the form of -Vt and 104  -Vn transitivizers ―introduces the agent‘s control over the event (and possibly also the agent itself …) but must also be responsible for removing the culmination entailment the requirement that the event culminate in the actual world, which is part of the meaning of the achievement from which the accomplishment is derived‖ ( ar-el 2005:130f, also see Matthewson 2004). Note that Bar-el does not examine lc-predicates, and thus does not provide an account for them. For the other transitivizers and intransitivizers that occur in any of the predicates that she examines, she notes their existence and provides some preliminary observations about the possible semantic contribution they make. One of the goals of this dissertation is to establish the morphology-semantics mapping for both c-predicates and lc-predicates, to investigate whether there is a strict correspondence between the morphological make-up of a predicate and its semantic interpretation, in particular relative to culmination. To do this we need to investigate whether or not culmination is entailed for all types of predicates, although as stated in Chapter 2 (§2.1.2.3), I do not examine the properties of all types of intransitives. In what follows, I will present data on all the c- and lc-transitivizers (§3). In anticipation of my morphosyntactic analysis for CONTROL, I also examine the applicatives with regards to culmination entailments. In §4 then I examine the c- and lc-intransitivizers, with regards to culmination. From this examination we will be better able to describe the culmination properties of S wxwu7mesh predicates based on the type of (in)transitivizer they occur with.  105  3  Core transitives and culmination  In this section I will test whether each of the five different core transitivizers behaves the same with respect to culmination. The core transitivizers as discussed in Chapter 2 (that is, those transitivizers which appear to be responsible for control and limited control meanings) are presented here again: Core Transitivizers c-marking -t -Vt -Vn -s lc-marking -nexw Table 28 Core transitivizers in S wxwu7mesh21  We will see that predicates marked with c-transitivizers do not require event culmination. As such, event culmination does not appear to be part of the inherent meaning of cpredicates. I demonstrate, however, with the culmination cancellation test that a cpredicate minimally requires that its event has been initiated. I further demonstrate that predicates marked with the lc-transitivizer require culmination, and consequently I argue that culmination is part of the inherent meaning of lc-predicates. In the following sections I present data for each of the core transitivizers with Barel‘s four tests. Most of the examples of the c-transitivizers are from Bar-el (2005). She  21  See Appendix A, §1, for all the roots that occur in the –t transitivizer construction, Appendix A, §2, for all the roots that occur in the –Vt construction and Appendix A, §3, for all the CVC roots that occur in the – Vn construction. As mentioned in Chapter 2 §4.1.2.1, except for a handful of roots, all roots or stems longer than CVC occur in the –Vn construction.  106  has examples of c-predicates with the -t and with the -Vn transitivizer. She does not have any examples of c-predicates with the -Vt transitivizer. For this transitivizer, I provide data from my own field work. For the causative -s she has one example. I provide a few more examples from my own field work. Bar-el does not investigate lc-transitive predicates, except for one example. I provide data from my own field work for this transitivizer. These findings are new for S wxwu7mesh and they confirm a pattern for lc-transitives that various researchers have reported for other Coast Salish languages: that is, that lc-predicates entail culmination (Davis 1978 and Watanabe 2003 for Sliammon, Gerdts 2008 for Halkomelem, Kiyota 2008 for Saanich). Bar-el (2005:133) herself suggests this is the case for S wxwu7mesh but does not further investigate it.  3.1  The -t transitivizer and culmination  In this section I show that the c-predicates marked with the -t transitivizer do not entail culmination. Consider the example below. Culmination cancellation test: c-predicate with -t (5) chen lhen-t-Ø ta h mten 1S.SUB weave-TR-3OBJ DET blanket ‗I‘m making a blanket welh haw ‘-an i h y-nexw-Ø but NEG SBJ-1S.CONJ PRES finish-LCTR-3OBJ but I didn‘t finish it.‘ (Bar-el 2005:82, ex. 33f)  107  The test in (5) shows that it is possible to explicitly deny that the event culminated. This test show that these predicate do not have culmination entailments. The next test also shows that predicates marked by the -t transitivizer can felicitously be used if the event still continues. This also shows that these c-transitives with -t do not have culmination entailments. Example (6)c is from my own field work.  Event continuation test: c-predicates with –t (6) a. na xel-t-Ø-as ta sxwexwiy m lha Mary, RL write-TR-3OBJ-3SUBJ DET story DET Mary ‗Mary wrote a story. iw yti na7-xw wa x l-t-Ø-as maybe RL-still IMPF write-TR-3OBJ-3SUBJ Maybe she‘s still writing it.‘ (Bar-el 2005:83, ex. 37a) b.  chen w‘el-t-Ø ta smeyts ti natlh, 1S.SUB cook-TR-3OBJ DET meat DET morning ‗I cooked the meat this morning iw yti na7-xw wa w‘el ta smeyts maybe RL-still IMPF cook DET meat and (maybe) it‘s still cooking.‘ (Bar-el 2005:83, ex. 37a)  c.  ch‘em-t-Ø-as ta s wemay ten sxen, bite-TR-3OBJ-3SUB DET dog DET-1.POS leg ‗The dog jumped me on my ankle na  RL  i na7-xw wa ch‘em-t-Ø-as and RL-still IMPF bite-TR-3OBJ-3SUB and he‘s still holding me on my ankle.‘ Next we turn to the scope of ilh (‗almost‘). In (7), ilh (‗almost‘) takes scope over the whole event and obtains the almost started reading (i.e. the event cancellation reading) wherein Mary never even started the event of writing. It does not obtain the 108  almost culminated reading (i.e. the event non-completion reading), the reading wherein the event of writing started but never culminated. This test shows that c-predicates with -t do not have inherent final points. If they did, then we would expect ilh (‗almost‘) to take scope over this final subevent.  Scope of ilh (‗almost‘) test: c-predicate with -t (7) na x l-t-Ø-as ta sxwexwiy m lha Mary almost RL write-TR-3OBJ-3SUBJ DET story DET Mary ‗Mary almost wrote a story.‘ Speaker‘s comments: ‗She didn‘t even start.‘ (Bar-el 2005:106, ex. 83a)  The same effects can be observed with the -t-marked c-predicate under the scope of negation in (8) (the (b) example is from my own field work). Its sentence can only mean that the whole event did not take place (i.e. the did not start reading). It cannot receive the almost culminated reading wherein the event started but only the event culmination is negated (i.e. the event non-completion reading). In both the (a) and (b) examples the event did not take place at all. No writing took place in (a) and no shooting took place in (b). This test also shows that these predicates do not have inherent final points.  Scope of negation test: c-predicates with -t (8) a. haw -an xel-t-Ø ta sxwexwiy m NEG SBJ-1S.CONJ write-TR-3OBJ DET story ‗I‘m not writing a story.‘  Context: You started it but you are not doing it now Speaker‘s comments: ―I‘m not going to write a story.‖ (Bar-el 2005:117, ex.108a)  109  b.  ‘-an i kw lash-t-Ø ta m xalh NEG SBJ-1S.CONJ PRES shoot-TR-3OBJ DET bear ‗I didn‘t shoot the bear.‘ ✓Context: You didn‘t shoot at all. You didn‘t shoot. Speaker‘s comments: ―You were going to shoot it but then you didn‘t.‖ haw  The findings of this section are summarized in the two tables below.  -t Table 29  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ C-transitive (-t) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -t ✓ ✘ ✓ ✘ Table 30 C-transitive (-t) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The results of the tests in Table 29 show that c-transitives with -t, in the past-perfective, do not have culmination entailments. The results in the two tests in Table 30 show that these predicate do not have inherent final points. In summary, all four of Bar-el‘s tests show that c-predicates marked with the -t transitivizer do not have culmination entailments.  110  3.2  The -Vt transitivizer and culmination  Bar-el (2005) does not provide any examples c-predicates with the -Vt transitivizer. I, therefore, provide examples from my own field work. Consider the examples in (9) which contain c-predicates marked with the -Vt transitivizer in the culmination cancellation test.  Culmination cancellation test: c-predicates with -Vt (9) a. chen ch‘aw-at-Ø te-n siy y 1S.SUB help-TR-3OBJ DET-1S.POS friend ‗I (went to) help my friend welh haw  ‘-as  ya  CONJ NEG SBJ-3CONJ PRES  em t at.home  but he wasn‘t home chen melh huy 7 1S.SUB so leave so I left.‘ b.  chen lh ch‘-it-Ø 1S.SUB cut-TR-3OBJ ‗I tried to cut the bread,  ta DET  sepl n bread  welh es-kw‘ y. an tl‘exw-Ø but STAT-cannot too hard-3SUB but I couldn‘t. It was too hard.‘  It is possible to deny the culmination of a -Vt marked event without inducing a contradiction. In (9)a it is possible that no actual helping takes place; only an attempt to help took place. In (9)b, it is possible that the bread did not actually get cut. There was an attempt to cut the bread. I will call this the tried to interpretation. It occurs even though the lexical item t‟  tsut ‗to try‘ is not present. I argue that this interpretation 111  indicates that a c-predicate minimally indicates that an event has been initiated. This interpretation also occurs with c-intransitives, as is shown in (§4.1). The next test also shows that c-predicates marked by the -Vt transitivizer can felicitously be used if the event still continues.  Event continuation test: c-predicates with -Vt (10) a. chen ch‘aw-at-Ø te-n 1S.SUB help-TR-3OBJ DET-1S.POS ‗I helped my friend,  siy y friend  na7-xw chen wa ch‘ w-at-Ø. RL-still 1S.SUB IMPF help-TR-3OBJ (and we‘re still working on it.‘) b.  chen lh ch‘-it-Ø 1S.SUB cut-TR-3OBJ ‗I sliced the bread,  ta DET  sepl n bread  na7-xw chen wa lh ch‘-it-Ø RL-still 1S.SUB IMPF cut-TR-3OBJ and I‘m still slicing it.‘  The event continuation test thus shows that the c-predicate marked by the -Vt transitivizer does not entail event culmination. If these c-predicates entailed culmination, it should not be felicitous to say that the event continued after having already said that event culminated. Next we turn to the scope of ilh (‗almost‘).  112  Scope of ilh (‗almost‘) test: c-predicates with -Vt (11) a. chen i ch‘ w-at-Ø. almost 1S.SUB PRES help-TR-3OBJ ‗I almost helped her.‘ ontext and speaker‘s comments: ‗… but I decided not to, or I changed my mind.‘ Context: I helped her a bit, but decided to quit. b.  na lh ch‘-it-Ø-as ta sts‘ wi7 lha Kirsten almost RL cut-TR-3OBJ-3SUBJ DET fish DET Kirsten ‗Kirsten was going to slice the fish but never got around to it.‘  In (11)a, ilh (‗almost‘) is compatible with either the almost started reading or the almost culminated reading. In (11)b, ilh (‗almost‘) takes scope over the whole event and therefore asserts that Kirsten never even started the event of cutting the fish, that is, the almost started reading. The result of this test are less clear than what Bar-el (2005) found for c-transitives with -t, which only obtained the the almost started reading. C-predicates with -Vt under the scope of negation only obtain one reading, namely the did not start reading (an event cancellation reading). Scope of negation test: c-predicates with -Vt (12) a. haw -an ch‘aw-at-Ø lhe-n siy y NEG SBJ-1S.CONJ help-TR-3OBJ DET-1S.POS friend ‗I didn‘t help my friend.‘  Context: You started to but you are not doing it now. ✓Context: You were going to help but then you didn‘t. b.  ‘-an i lh ch‘-it-Ø ta sts‘ wi7 NEG SBJ-1S.CONJ PRES cut-TR-3OBJ DET fish ‗I didn‘t cut the fish.‘ ✘Context: You started to cut it but you didn‘t finish cutting it. ✓Context: You didn‘t cut it at all. You didn‘t cut. haw  113  The sentence can only mean that the whole event did not take place (it did not even start). It cannot receive the interpretation according to which the event started but only the culmination of the event is negated. Since negation cannot take scope over the final event, we can conclude that the c-predicates with the -Vt transitivizer are not associated with inherent final points.  -Vt Table 31  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ C-transitive (-Vt) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -Vt ✓ ✓ ✓ ✘ Table 32 C-transitive (-Vt) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  In sum, Bar-el‘s (2005) tests show that c-predicates marked with -Vt do not have culmination entailments. The second two scopal tests are less clear than what Bar-el (2005) found for the c-predicates with -t. Note, though that with the scope of negation test, the only reading is the did not start reading. After analyzing other predicates with the scope of ilh (‗almost‘) test I will conclude that it is a less reliable test for inherent points, especially with transitive predicates. Importantly, though, for our investigation 114  here, neither scope test allows for only the event non-completion reading. I take this as a strong indication that predicates with -Vt do not have inherent final points.  3.3  The -Vn transitivizer and culmination  In this section I provide data from Bar-el (2005) for c-predicates with the -Vn transitivizer. The following example is of a c-predicate with the -Vn transitivizer and the culmination cancellation test.  Culmination cancellation test: c-predicates with -Vn (13) na m kw‘-in-t-Ø-as ta lhxenptn RL clean-TR-TR-3OBJ-3SUB DET floor ‗Mary washed the floor ‘  lha DET  Mary Mary  welh haw ‘-as i h y-nexw-Ø-as but NEG SBJ-3CONJ PRES finish-LCTR-3OBJ-3SUB ‗but she didn‘t finish.‘ (Bar-el 2005:81, ex. 33c)  This example shows that it is possible to deny the culmination without inducing a contradiction. This shows that c-predicates with -Vn do not have culmination entailments. The next test also shows that c-predicates marked by the -Vn transitivizer can felicitously be used if the event still continues.  115  Event continuation test: c-predicate with -Vn (14) na m kw‘-in-t-Ø-as ta lhxenptn RL clean-TR-TR-3OBJ-3SUB DET floor ‗Mary washed the floor ‘  lha DET  Mary Mary  i na7-xw wa m kw‘-in-t-Ø-as and RL-still IMPF clean-TR-TR-3OBJ-3SUB ‗and she‘s still washing it.‘ (Bar-el 2005:84, ex. 37d)  The event continuation test thus shows that the c-predicate marked by the -Vn transitivizer does not entail event culmination. If these c-predicates entailed culmination, it should not be felicitous to say that the event continued after having already said that the event culminated. Next we turn to the scope of ilh (‗almost‘).  Scope of ilh (‗almost‘) test: c-predicate with -Vn (15) chen m kw‘-in-[ ]-Ø ta t txwem almost 1S.SUB help-TR-TR-3OBJ DET car ‗I nearly started to wash the car.‘ (Bar-el 2005:106, ex. 83b)  In (15) ilh (‗almost‘) takes scope over the whole event and obtains the almost started reading and therefore asserts that I never even started the event of washing the car. This suggests that c-predicates with the -Vn transitivizer do not encode inherent final points otherwise ilh ‗almost‘ should be able to take scope over the final event. Note under the scope of ilh (‗almost‘) the c-predicate can be translated as ‗nearly started‘.  116  The same effects can be observed with the c-predicate marked with -Vn under the scope of negation.  Scope of negation test: c-predicate with -Vn (16) haw -as i m kw‘-in-t-Ø-as NEG SBJ-3CONJ PRES wash-TR-TR-3OBJ-3SUB lhx npten kwa John DET floor DET John ‗John didn‘t wash the floor.‘ ✓Context: he didn‘t even start. (Bar-el 2005:117,ex.108b) ta  The sentence can only mean that the whole event did not take place (it did not even start). It does not receive the interpretation according to which the event started but only the event culmination is negated. Since negation does not take scope over the final event, cpredicates with the -Vn transitivizer, I argue, do not have inherent final points in their semantic representation. The following tables are a summary of all four of Bar-el‘s tests for the c-predicates marked with the –Vn transitivizer.  -Vn Table 33  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ C-transitive (-Vn) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  117  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -Vn ✓ ✘ ✓ ✘ Table 34 C-transitive (-Vn) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested) Bar-el‘s (2005) testing of c-transitives with -Vn shows that these predicates do not have culmination entailments, and they do not have inherent final points.  3.4  The causative -s transitivizer and culmination  Bar-el (2005) only provides one example of her tests with c-predicates with the –s causative, with the culmination cancellation test. I, therefore, provide data from my own field work  Culmination cancellation test: c-predicate with -s (17) a. na ch 7-s-t-Ø-as kwi kw‘ xwa7 lha Mary RL do-CAUS-TR-3OBJ-3SUB DET box DET Mary ‗Mary made a box.‘ b.  h y-nexw-Ø-as RL POL finish-LCTR-3OBJ-3SUB ‗ id she finish it?‘ (Bar-el 2005:82, ex.34a-b) na7 u  The fact that the question in (b) can felicitously uttered following (a) indicates that the cpredicate marked with the -s causative does not entail culmination. 118  In my own field work, I have tested other causative-marked predicates using the culmination cancellation test as shown in (18). The same result is obtained. It is possible to deny the culmination of the causative -s-marked predicate without inducing a contradiction.  Culmination cancellation test: c-predicates with -s (18) a. chen t 7-s-[ ]-Ø ta kw‘ xwa7 1S.SUB do-CAUS-TR-3OBJ DET box ‗I made a box welh haw ‘-an but NEG SBJ-1S.CONJ but I didn‘t finish it.‘ b.  i PRES  h y-nexw-Ø finish-LCTR-3OBJ  chen nam-s-[ ]-Ø tiwa s7ixwalh 1S.SUB go-CAUS-TR-3OBJ DEM child ‗I‘m bringing my child to the hospital.  (t)-ta OBL-DET  w‘uy wtxw, hospital  chiy lh i tsixw chet. soon PRES arrive 1P.SUB We‘re just about there.‘ Context: I‘m on the phone telling someone this. c.  chen nam-s-t-umi ta s7ilhenawtxw, welh es-kw‘ay. 1S.SUB go-CAUS-TR-2OBJ DET restaurant but STAT-cannot ‗I was going to invite you to the restaurant but I couldn‘t make it.‘ Speaker‘s comment: Something turned up (so that‘s why I didn‘t take you).‘ Speaker‘s comment: I didn‘t take you because w n t l (‗I didn‘t have any money‘).‘  All three c-predicates marked with the causative -s allow the cancellation of the culmination of the event without inducing a contradiction. This shows that c-predicates with the causative, like all the other c-predicates, do not entail culmination. Note that the same predicate n m-s ‗go-CAUS = to bring‘ is used in both (b) and (c) examples. In the 119  (b) example the event of bringing has started but has not culminated. In the (c) example, though, no event of bringing has actually taken place. Just the intention of bringing took place or possibly an invitation, but no actual bringing nevertheless. The next test also shows that c-predicates marked by the -s causative can felicitously be used if the event still continues.  Event continuation test: c-predicate with -Vs (19) na ta7-s-t-Ø-as ta kw‘ xwa7 kwa Ray RL make-TR-TR-3OBJ-3SUB DET box DET Ray i na7-xw wa ta7-s-t-Ø-as and RL-still IMPF make-TR-TR-3OBJ-3SUB ‗Ray‘s still carving it and he‘s carving it. He never got around to finishing it.‘  In this example, the speaker judges the sentence grammatical, but does not provide a literal clause by clause translation. Instead she indicates that the point of these clauses is to indicate that despite all this activity of making and making (more specifically ‗carving‘) a box, the event has not yet culminated in the box actually being finished. The event continuation test, then, shows that the c-predicate marked by the -s causative does not entail event culmination. If these c-predicates entailed culmination, it should not be felicitous to say that the event continued after having already said that event culminated. Next we turn to the scope of ilh (‗almost‘).  120  Scope of ilh (‗almost‘) test: c-predicate with –s causative (20) ilh na ta7-s-t-Ø-as ta kw‘ xwa7 kwa almost RL make-CAUS-TR-3OBJ DET box DET ‗He almost finished a box.  Ray Ray  In (20) ilh (‗almost‘) takes scope only over the final part of the event. This is the almost finished reading. This at least indicates that for c-predicates with the -s transitivizer, the final event scopes under ilh (‗almost‘). Note again that this test is a less reliable diagnostic for inherent points. That is, I argue that this test does not show that these ctransitives with -s have inherent final points. It may actually point to a more pragmatically based account for the scopal test of ilh (‗almost‘), especially when the results of the other three of Bar-el‘s tests are taken into account. The following examples test the c-predicate marked with -s under the scope of negation.  Scope of negation test: c-predicate with -s (21) haw -as i t 7-s-t-Ø-as ta kw‘ xwa7 kwa Ray NEG SBJ-3CONJ PRES make-CAUS-TR-3OBJ-3SUB DET box DET Ray ‗He didn‘t make a box.‘ Speaker‘s comments: ‗He never touched it. Period. He didn‘t do anything. Na mi umet (‗he got lazy‘).‘  Under the scope of negation the c-predicate with -s obtains the did not start reading, which the speaker‘s comments indicate very clearly. Negation here takes scope over the whole event. Possibly the only event that took place was the promise of making a box.  121  The results for the c-predicate with -s the causative are summarized in the following two tables.  -s Table 35  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ C-transitive (-s) and culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 The Scope of ilh (‗almost‘) Test Event Cancellation Event Non(= almost started) completion (=almost culminated) -s -✓ Table 36 C-transitive (-s) and the scopal tests  Test 4 The Scope of Negation Test Event Cancellation Event Non(=did not start) completion (= did not culminate) ✓ ✘  (✘=not obtained; ✓=obtained; -- data not yet tested)  Three out of four of Bar-el‘s tests show that c-predicates marked with the –s transitivizer, do not have culmination entailments. Only the scope of ilh (‗almost‘) test seems to contradict this generalization in that it appears to indicate that these predicates do have final points, although I have only one token and I did not check it for the event cancellation reading. We have to conclude, therefore, that this test is not a reliable diagnostic for inherent points. The data here with the causative further suggests that the scopal effects of ilh (‗almost‘) are pragmatically conditioned and that they do not necessarily pick out inherent initial or final points. Since the first two tests clearly 122  indicate that c-transitives with -s do not have culmination entailments and the scope of negation test clearly indicates that it can only take scope over the final event, I argue, that c-transitives with -s do not have inherent final points. That c-transitives with -s do not have inherent final points is consistent with the fact that many predicates formed with the causative –s have stative meanings, as shown in (22)a-b. Stative predicates are standardly assumed to not have any internal event structure, meaning they have neither inherent initial nor inherent final points. (22) a.  b.  3.5  es-lh ‘i7-s STAT-know-CAUS ‗to know (a person or thing)‘ es-li7-s STAT-store-CAUS ‗to keep (something) stored away.‘  The limited control transitivizer -nexw and culmination  In this section, I investigate lc-transitives. These are marked with the lc-transitivizer -nexw. Since lc-predicates were not systematically investigated in Bar-el (2005), I provide data from my own field work, using all four of her tests for culmination. They all converge on the same result: lc-transitives entail culmination. The data in (23) show that event culmination cannot be cancelled. Culmination cancellation test: lc-predicates with -nexw (23) a. chen kw lash-nexw-Ø ta m xalh 1S.SUB shoot-LCTR-3OBJ DET bear ‗I shot the bear ‘  123  #welh chen but 1S.SUB (‗but I missed.) b.  t‘emt‘ m astray  chen lh ch‘-nexw-Ø ta p‘ sxwem 1S.SUB cut-LCTR-3OBJ DET crusty ‗I cut the crusty bread…‘ #welh es-kw‘ y but STAT-cannot (‗but I couldn‘t.)  c.  d.  chen p‘ ya -nexw-Ø ta tetxwem 1S.SUB fix-LCTR-3OBJ DET car ‗I fixed the car ‘ #welh haw ‘-an but NEG SBJ-1S.CONJ (‗but I didn‘t finish.‘)  i  chen t 7-nexw-Ø 1S.SUB make-LCTR-3OBJ ‗I made a box ‘  ta  #welh haw ‘-an but NEG SBJ-1S.CONJ (‗but I didn‘t finish it.‘)  i  PRES  DET  PRES  h y-nexw-Ø finish-LCTR-3OBJ kw‘ xwa7 box h y-nexw-Ø finish-LCTR-3OBJ  In all examples in (23), it is impossible to deny culmination without inducing a contradiction. Thus, I conclude that lc-predicates entail culmination. For completeness, note that all the examples in (23) have a corresponding cpredicate. The c-predicate version of the root in (23)a is formed with the -t transitivizer (24)a. The c-predicate version of the root in (23)b is formed with the -Vt transitivizer (24)b. The c-predicate version of the root in (23)c is formed with the -Vn transitivizer (24)c. The c-predicate version of the root in (23)d is formed with the -s causative transitivizer (24)d. All of the c-predicate versions in (24)a-d are minimal pairs to the 124  sentences in (23)a-d. All of the c-predicate versions can have their culmination denied without inducing a contradiction.  Culmination cancellation test: all four c-transitives (24) a. -t chen kw lash-t-Ø ta m xalh 1S.SUB shoot-TR-3OBJ DET bear ‗I shot the bear ‘ welh na t‘emt‘ m but RL astray ‗but I missed.‘ b.  te-n DET-1S.POS  skw lash shot  -Vt chen lh ch‘-it-Ø ta p‘ sxwem 1S.SUB cut-TR-3OBJ DET crusty ‗I (tried to) cut the crusty bread ‘ welh es-kw‘ y but STAT-cannot ‗but I couldn‘t.‘  c.  -Vn chen p‘ ya -en-[ ]-Ø ta tetxwem 1S.SUB fix-TR-TR-3OBJ DET car ‗I fixed the car ‘ welh haw ‘-an i h y-nexw-Ø but NEG SBJ-1S.CONJ PRES finish-LCTR-3OBJ ‗but I didn‘t finish (fixing it).‘  d.  -s chen t 7-s-[ ]-Ø 1S.SUB make-TR-TR-3OBJ ‗I made a box ‘ welh haw ‘-an but NEG SBJ-1S.CONJ ‗but I didn‘t finish it.‘  ta DET  i PRES  125  kw‘ xwa7 box  huy-nexw-Ø finish-LCTR-3OBJ  As for the event continuation test with the lc-transitive, Bar-el (2005) provides the following example.  Event continuation test: lc-transitive (25) chen x wtl‘-nexw-Ø 1S.SUB break-LCTR-3OBJ ‗I broke it,‘ #i na7-xw chen wa xewtl‘- n-[ ]-Ø and RL-still 1S.SUB IMPF break-TR-TR-3OBJ (#‗and I‘m still breaking it.‘) Speaker‘s comments: ―You already broke it ...can‘t still be breaking it ... it‘s already broken.‖ (Bar-el, 2005:220, ex.35)  This example shows that it is impossible to assert that event continued when the predicate is marked as an lc-transitive. The following is an example of the lc-transitive with the scope of ilh (‗almost‘) test.  Scope of ilh test: lc-transitives (26) na i p‘ ya -nexw-as ten t txwem. almost RL PRES fix-LCTR-3OBJ DET-1S.POS car ‗They almost repaired my car but they never got around to it.‘ ✓Context: They started but never finished the car. ✘Context: They never even started fixing the car.  The scope of ilh ‗almost‘ test with the lc-transitive in (26) obtains the almost culminate reading (i.e. the event non-completion reading) but not the almost started reading.  ilh  (‗almost‘), then, is only taking scope over the final subevent. Importantly, speakers reject the almost started reading with the lc-transitive under the scope of ilh (‗almost‘). The following are examples of the lc-transitive with the scope of negation test. 126  Scope of negation test: lc-transitives (27) a. haw ‘-an i kw lash-nexw-Ø ta m xalh NEG SBJ-1.CONJ PRES shoot-LCTR-3OBJ DET bear ‗I didn‘t shoot the bear.‘ ✓Context: I shot, but I missed. ✘Context: I didn‘t shoot at all. Speaker‘s comment: No, you said you shot already. b.  ‘-an i p‘ ya -nexw-Ø te-n t txwem NEG SBJ-1.CONJ PRES fix-TR-3OBJ DET-1S.POS car ‗I didn‘t get my car fixed.‘ ✓Context: You started fixing it but didn‘t finish fixing it. ✘Context: You didn‘t start fixing it yet. haw  In both of these examples, the lc-transitive only obtains the did not culminate reading (i.e. the event non-completion reading). The speaker rejects the did not start reading (i.e. the event cancellation reading). In (27)a an event of shooting must have taken place. In (27)b an event of fixing must have taken place. The negation, then, simply cancels that either of these events culminated. I conclude that these examples indicate that there is only one subevent of the lc-predicate that ilh (‗almost‘) can take scope over: the final event of the lc-transitive. The following two table summarize our findings with Bar-el‘s four test applied to the lc-transitives. This is the first systematic examination of the culmination properties of lctransitives in S wxwu7mesh.  127  -nexw Table 37  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✘ -✘ -Lc-transitive (-nexw) and culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -nexw ✘ ✓ ✘ ✓ Table 38 Lc-transitive (-nexw) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The first two tests, in Table 37, show that it is not felicitous to deny the culmination of the lc-transitive without inducing a contradiction, nor is it felicitous to indicate that its event continued. I argue that these two test indicate that the lc-transitive predicates do have culmination entailments, in contrast to c-transitives, which do not have culmination entailments. The two scopal tests summarized in Table 38, show that ilh (‗almost‘) and negation can only take scope over the culmination of the event. I conclude that lc-transitive predicates are associated with inherent final points.  128  3.6  Summary  The following two tables summarize our findings for Bar-el‘s four tests for final points for all four c-transitives and for the lc-transitives.22  -t -Vt -Vn -s -nexw Table 39  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✘ ✘ C-transitives, lc-transitives and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -t ✓ ✘ ✓ ✘ -Vt ✓ ✓ ✓ ✘ -Vn ✓ ✘ ✓ ✘ -s -✓ ✓ ✘ -nexw ✘ ✓ ✘ ✓ Table 40 C-transitives, lc-transitives and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  22  Note that I did not use questions for the two culmination tests. The  129  One of the goals of this dissertation is to investigate the relation between a predicate‘s morphological make-up and its semantic properties. In this section, we have investigated predicates marked as c-transitives and lc-transitives, respectively. Using Bar-el‘s (2005) tests we have determined that none of the transitive c-predicates have culmination entailments, while the transitive lc-predicates do have culmination entailments, as indicate in Table 39. The results of the scopal tests, indicated in Table 40, show that they have inherent final points. This is, in fact, in line with the behaviour of lc-transitives in other Coast Salish languages (Gerdts 2008, Kiyota 2008,Watanabe 2003).  Culmination Entailments Inherent final points C-transitives -t ✘ ✘ -Vt ✘ ✘ -Vn ✘ ✘ -s ✘ ✘ Lc-transitives -nexw ✓ ✓ Table 41 Distribution of culmination entailments and inherent final points  One other interpretation that we found with the culmination cancellation test, with one of the c-transitives, -Vt, is the tried to interpretation. This interpretation will also be important in our understanding the essential contrast between c- and lc-predicates. In §4 when we examine both c- and lc-intransitives, we will find this same tried to interpretation is available with c-intransitives, similar to their c-transitive counterparts. In §7, I propose that this tried to interpretation indicates that c-predicates minimally indicate event initiation. I also argue that predicates with inherent initial points as per Bar-el (2005) are a different from predicates which minimally indicate event initiation. 130  Based on this distribution of culmination entailments, we may conclude that the S wxwu7mesh contrast in CONTROL-marking is fundamentally about event-culmination. If so, we might expect that culmination entailments will be associated with all and only lc-predicates, but never with c-predicates. We still need to investigate whether other morphologically complex predicates are associated with culmination entailments and whether they have inherent final points. We also still need to investigate whether other predicates occur with the tried to interpretation, as we have found with the c-transitives. In what follows I discuss the culmination entailments of intransitives, which have also previously been described as contrasting in CONTROL (Chapter 2, §4.1.2.3).  4  Intransitives and culmination  In Table 42, I show the set of S wxwu7mesh intransitivizers which are purported to derive c-intransitives and lc-transitive, respectively (Jacobs 2007).  c-intransitive  Unergative (UE) -im  lc-intransitives -nalhen Table 42 C- and lc-intransitivizers  Reflexive (REFL) -sut -namut -numut  Reciprocal (RECIP) -way -newas  The intransitivizers divide into suffixes which derive unergative predicates from roots and suffixes which derive reflexive and reciprocal predicates from predicates which are already transitivized. In what follows I investigate four out of the seven intransitivizers 131  in terms of their culmination entailments. For lack of data, however, I will have to leave the reflexive -n mut23 and both reciprocals (all of which are shaded in Table 42) for future research.24 In what follows, I show that the two c-intransitivizers, the c-unergative -im and the creflexive -sut do not have culmination entailments. I also show that they can occur with the tried to interpretation. Thus, they behave like the c-transitives in both respects. Moreover, I show that the two lc-intransitivizers (lc-unergative and lc-reflexive) are associated with culmination entailments, just like the lc-transitives are.  4.1  The control intransitives and culmination  In this section I present my results using Bar-el‘s four tests for inherent final points with the c-unergative -im and the c-reflexive -sut.  4.1.1  The control unergative -  and culmination  In this section I present my findings with the c-unergative -im. I demonstrate that the cunergative predicates do not have culmination entailments and that they do not inherent  23  Since this reflexive only occurs with the causative I will call it the caus-reflexive to distinguish it from the other c-reflexive –sut. 24 For Sliammon, Watanabe (2003) found that the event culmination entailments of reciprocals must take into consideration all the possible pairs. For example, an lc-reciprocal can be used in a situation where only one relevant pair completed the event. It does not require all the event pairs to have completed the event.  132  final points. The following are examples of the c-unergative with the culmination cancellation test (28)a-b.  Culmination cancellation test: c-unergatives with -im (28) a. chen suxwtwelh es-kw‘ay 1S.SUB recognize-CUE but STAT-cannot ‗I tried to recognize him but I couldn‘t.‘ b.  chen kw‘shwelh es-kw‘ y 1S.SUB count-CUE but STAT-cannot ‗I tried to count but I couldn‘t.‘  c.  chen kw‘sh(t)t-en tala 1S.SUB count-CUE OBL-DET-1S.POS money ‗I‘m counting my money, welh haw ‘-an i h y-nexw-Ø. but NEG SBJ-1S.CONJ PRES finish-LCTR-3OBJ but I never finished it. Speaker‘s comments: Maybe I got tired because I had too much money to count.  c.  ta  John na tahJohn RL do-CUE ‗John is building a house ‘ DET  (t)ta lam, OBL-DET house  welh xwew xw ‘-as i but not.yet SBJ-3CONJ PRES ‗but he hasn‘t finished it yet.‘  h y-nexw-Ø-as finish-LCTR-3OBJ-3SUB  As with the c-transitives, the c-unergative is compatible with the cancellation of event culmination. Also, as predicted from our examination of the c-transitives, the cunergative, can be translated into English with the tried to interpretation, an interpretation which we have already seen in §3.2 with the c-transitivizer, -Vt. Again, try does not have  133  a direct correspondent in the S wxwu7mesh data. Thus, the c-unergative does not entail culmination but instead minimally requires that its event be initiated. The following example has the c-unergative with the event continuation test.  Event continuation test: c-unergative with -im (29) na kw‘sh- -wit (t)ta tala-s-wit, RL count-CUE-PL OBL-DET money-3POS-PL ‗They‘re counting iw yti na7xw wa kw‘sh-im. maybe RL-still IMPF count-CUE maybe they‘re still counting their money.‘  Example (29) shows that it is possible to continue the event denoted by the c-unergative without inducing a contradiction. The following examples show the c-unergative with the scope of ilh (‗almost‘) test. Scope of ilh (‗almost‘) test: c-unergatives with -im (30) a. chen i kw‘sh(t)ten tala. almost 1S.SUB PRES count-CUE OBL-DET-1S.POS money ‗I was going to count my money.‘ Context: I had too much money so I didn‘t count it at all. Context: I started but I got tired because I had so much money. b.  chen i tah(t)ta lam almost 1S.SUB PRES make-CUE OBL-DET house ‗I was going to build a house (but I never got around to it yet).‘ ✓Context: You were just thinking about it. ✓Context: The house could almost be finished.  The example in (30)a obtains the almost started reading (an event cancellation reading) in the first context and the almost culminated reading (an event non-completion reading) in the second context. Example (30)b also obtains both readings. I note here that the 134  almost started reading is, more often than not, the first reading provided for these sentences in an out-of-the-blue context. But if questioned, speakers also readily accept a context for the almost culminated reading. The following example shows the c-unergative under the scope of negation.  Scope of negation test: c-unergative with -im (31) haw ‘-an i tah(t)-ta lam. NEG SBJ-1S.CONJ PRES make-CUE OBL-DET house ‗I never made a house.‘ ✓Context: I didn‘t start. ✓Context: I haven‘t finished it yet.  In (31) the c-unergative predicate, under the scope of negation, can obtain the did not start reading (i.e. the event cancellation reading). But, it is also compatible with the did not culminate reading (i.e. an event non-completion reading), a context wherein the event of building the house has started but is not yet finished. As with the scope of ilh (‗almost‘) test with the c-unergative, the scope of negation test with the c-unergative can obtain either the almost started or the almost culminated reading. The following table summarizes our findings for the c-unergative.  -im Table 43  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ C-unergative (-im) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  135  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -im ✓ ✓ ✓ ✓ Table 44 C-unergative (-im) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The results of the first two tests, given in Table 43, show that c-unergatives allow for the event cancellation reading. This, I argue, shows that c-unergatives do not have culmination entailments. The results of the second two tests, given in Table 44, show that c-unergatives are compatible with both the event cancellation reading and the event non-completion reading. The fact that c-unergatives do obtain the event cancellation reading shows that these predicates do not have inherent final points. If they did, these two tests should only allow for one reading, the event non-completion reading, as we saw for lc-transitives.  4.1.2  The control reflexive -sut and culmination  In this section I present my findings with the c-reflexive -sut with Bar-el‘s (2005) four tests for final points. I demonstrate that the c-reflexive predicates do not have culmination entailments and that they do not inherent final points. The following examples are of c-reflexives with the culmination cancellation test.  136  Culmination cancellation test: c-reflexives with -sut (32) a. na kw lash-t-sut-Ø welh es-kw‘ay RL shoot-TR-CREFL-3SUB but STAT-cannot ‗He tried to shoot himself but he couldn‘t.‘ b.  chen ch‘ w-at-sut 1S.SUB help-TR-CREFL ‗I tried to help myself ‘ welh es-kw‘ y. chen men el m but STAT-cannot 1S.SUB just weak ‗but I couldn‘t. I was weak.‘  The same results are obtained with the c-reflexive as with the c-unergative. The culmination of the event can be denied without inducing a contradiction. The event of shooting (32)a does not have to result in the subject actually shooting himself. The cpredicate can be translated with the tried to interpretation, an interpretation which we have seen with the c-transitives and the c-unergative. The translation, tried to, does not have a direct correspondent in the S wxwu7mesh data. The culmination of the event in (32)b can also be denied without inducing a contradiction. The agent in (32)b only need have ‗tried to help‘. The following are examples of the c-reflexive with the event continuation test. The (a) example has the c-reflexive with the -Vt transitivizer and the (b) example has the creflexive with the -Vn transitivizer.  Event continuation test: c-reflexives with -sut (33) a. na ch‘aw-at-sut-Ø-wit, RL help-TR-CREFL-3SUB-PL ‘They started helping them  137  i na7-xw wa ch‘aw-at-sut-Ø-wit. and RL-still IMPF help-TR-CREFL-3SUB-PL and they‘re still helping.‘ b.  chen ch‘it-en-t-sut, 1S.SUB near-TR-TR-CREFL ‗I‘m going near you i na7-xw chen wa ch‘it-en-t-sut. and RL-still 1S.SUB IMPF near-TR-TR-CREFL and I‘m still near you.‘  In both examples in (33) the event continuation test shows that is felicitous to add a sentence which indicates that the event of the c-reflexive is continuing. The (a) example describes a situation wherein the event of helping has not culminated.25 But the (b) example describes an event wherein the event has culminated, i.e. the subject has successfully brought himself close and the continuing clause asserts that the subject is maintaining this position. The event continuation test shows that for the c-reflexive, the continuation may be the event itself (the (a) example) or the continuation may be where the resulting state of the event continues (the (b) example). In neither case does the addition of this continuation clause create a contradiction. The following examples are of the c-reflexive with the scope of ilh (‗almost‘) test.  25  Note that often the c-reflexive is translated very similarily to its c-transitive counterpart, with an agent and a theme argument which are not co-referential. In this way the resemble the c-unergatives.  138  Scope of ilh (‗almost‘) test: c-reflexive with -sut (34) a. chen i lixw-t-sut. almost 1S.SUB PRES laid.down-TR-CREFL ‗I was just about to sit down and rest [i.e. lie down] (and then the phone rang).‘ b.  chen i ch‘ t-en-t-sut. almost 1S.SUB PRES near-TR-TR-CREFL ‗I was going to sit near you.‘ ✓Context: I was just thinking about it. ✓Context: I was already going towards you (to sit next to you), but someone come along and took that seat beside you.  Both examples of the scope of ilh (‗almost‘) test with the c-reflexive obtain the almost started reading. In (34)a the subject had the intention of lying down, but then was interrupted. Example (34)b, can be used to describe a situation wherein the subject had the intention of going closer, but they never did more than think about it, as described in the first context. It is also compatible with a situation wherein the subject did start to move closer but then was interrupted. This is the almost culminated reading (i.e. the event non-completion reading). Thus, in the scope of ilh, the c-reflexive can obtain either the almost started or almost culminated readings. The following are examples of the c-reflexive with the scope of negation test.  Scope of negation test: c-reflexives with -sut (35) a. haw ‘-an i lixw-t-sut. NEG SBJ-1S.CONJ PRES lay-TR-CREFL ‗I never got around to resting.‘ Context: I was too busy. Context: Just as I was sitting down, someone phoned me so I had to get up. b.  ‘-an i h l-it-sut. NEG SBJ-1S.CONJ PRES roll-TR-CREFL ‗I‘m not one for rolling.‘ haw  139  Example (35)a, obtains the didn‟t start reading wherein the subject did not even get a chance to try and rest, as in the first context. The second context describes the didn‟t culminate reading wherein the event started but never actually culminated. In (35)b, the c-reflexive has a reading wherein the subject never got around to even starting to roll since she was not in the habit of rolling in the first place. This, I argue, is an instance of the didn‟t start reading. The following two tables provide a summary of our findings for the c-reflexive -sut.  -sut Table 45  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ C-reflexive (-sut) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -sut ✓ ✓ ✓ ✓ Table 46 C-reflexive (-sut) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The results of the first two tests, given in Table 45, show that predicates marked with the c-reflexive -sut allow for the event cancellation reading. This, I argue, shows that creflexives do not have culmination entailments. I conclude that c-reflexives are not associated with inherent final points. If they were, we would expect that the first two 140  tests would only pick out the final point. The results of the second two tests, summarized in Table 46, show that c-reflexives are compatible with both the event cancellation reading and the event non-completion reading. Again this shows that c-reflexives are not associated with inherent final points. If they did, these two tests should only allow for one reading, the event non-completion reading.  4.2  The limited control intransitives and culmination  In this section I present my results of the lc-unergative -nalhn, and the lc-reflexive -numut with Bar-el‘s four tests for inherent final points. I demonstrate that predicates with these intransitivizers are associated with culmination entailments and argue that they do have inherent final points.  4.2.1  The limited control unergative -nalhn and culmination  In this section I examine the lc-unergative. This construction was not described as such by Kuipers (1967:133, §187:40). He gives only two lexical items with this morpheme.26  (36) a.  ‘aw-alhn paid-LCUE ‗to be punished‘  26  Note, that the -n part of the lc-unergative -nalhn appears to get elided when the final consonant of the root is a w.  141  b.  ‘aw-alhn-s paid-LCUE-CAUS ‗to punish (someone)‘  In my own research I have found that it is still productive to a limited degree, and only for some speakers. For those speakers for whom it is productive, it is used less so than the lc-transitive or the lc-reflexive constructions. This made it difficult at times to collect data. The following are some of the data I was able to obtain. Here are examples of lc-unergatives with the culmination cancellation test.  Culmination cancellation test: lc-unergatives with -nalhn (37) a. chen suxwt, #welh es-kw‘ y 1S.SUB recognize-LCUE but STAT-cannot i) ‗I recognized him, (#but I couldn‘t).‘ ii) *‗I tried to recognize him but I couldn‘t.‘ b.  chen kw‘ach, #welh es-kw‘ y 1S.SUB look-LCUE but STAT-cannot i) ‗I saw, (#but I couldn‘t).‘ ii) *‗I tried to see him but I couldn‘t.‘  The cancellation of the event culmination results in a contradiction in both examples in (37) . Furthermore, the lc-unergatives do not obtain the tried to interpretation, as the cunergatives do with this test. The next example is of the lc-unergative under the scope of ilh (‗almost‘).  142  Scope of ilh (‗almost‘) test: lc-unergative with -nalhn (38) chen i kwelashta sxwi7shn. almost 1S.SUB PRES shoot-LCUE DET deer ‗I almost shot the deer.‘ ✓Context: I tried to shoot it but I missed it. ✘Context: I didn‘t even try to shoot anything.  The lc-unergative under the scope of ilh (‗almost‘) obtains only the almost culminated (i.e. the event non-completion reading). That is, this sentence is only compatible with a context where I tried to shoot the deer. It does not obtain the almost started reading (i.e. the event cancellation reading), as can be seen on the basis of the fact that (38) is not compatible with the second context. The following is an example of the lc-unergative under the scope of negation. Scope of negation test: lc-unergative (39) haw ‘-an i kwelash-nalhn ta sxwi7shn. NEG SBJ-1S.CONJ PRES shoot-LCUE DET deer ‗I missed it.‘ ✓Context: I was trying to shoot the deer but I missed. ✘Context: I didn‘t even try to shoot it.  The lc-unergative under the scope of negation obtains only the almost culminated (i.e. the event non-completion reading). It does not obtain the almost started reading (i.e. the event cancellation reading). Thus (39) is only compatible with the first context but not the second context. The following two tables summarize our findings for the lc-unergative.  143  -nalhn Table 47  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✘ ✘ Lc-unergative (-nalhn) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -nalhn ✘ ✓ ✘ ✓ Table 48 Lc-unergative (-nalhn) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The results of the first two tests, given in Table 47, show that lc-unergative (-nalhn) predicate does not allow for the event cancellation reading. This, I argue, shows that lcunergatives do not have culmination entailments. The results of the second two tests, given in Table 48, show that lc-unergatives are only compatible with the event noncompletion reading and they do not obtain the event cancellation reading. I argue that these results show that the lc-unergatives predicates do have inherent final points. If they did not, then we would expect that these two tests should allow for both readings as the cunergative does.  144  4.2.2  The limited control reflexive -numut and culmination  In this section, I examine the lc-reflexive -numut with Bar-el‘s (2005) four tests for final points. The results of some of these tests at first appear to indicate that the lc-reflexives do not have final points. This contradicts our expectation for the lc-reflexives. I provide an account for the lc-reflexives wherein they do have culmination entailments and they do have final points. The following examples are of lc-reflexives with the culmination cancellation test.  Culmination cancellation test: the lc-reflexive (40) a. na kwelash-Ø #welh es-kw‘ y RL shoot-LCREFL-3SUB but STAT-cannot i) ‗He shot himself, (#but he couldn‘t).‘ ii) *‗He tried to shoot himself but he couldn‘t.‘ b.  chen ts‘its‘ p‘-numut, #welh es-kw‘ y 1S.SUB work-LCREFL but STAT-cannot i) ‗I managed to work, (#but I couldn‘t).‘ ii) *‗I tried to work but I couldn‘t.‘  c.  chen huy-n mut, #welh haw ‘-an 1S.SUB finish-LCREFL but NEG SBJ-1S.CONJ ‗I managed to finish (#but I didn‘t finish it).‘  i PRES  huy-nexw-Ø finish-LCTR-3OBJ  As with the lc-transitives and lc-unergative, it is not possible to cancel the event culmination without inducing a contradiction. Furthermore, the lc-reflexives do not obtain the tried to interpretation, unlike the c-reflexives. The following example is of the lc-reflexive with the event continuation test.  145  Event continuation test: the lc-reflexive (41) chen ilhen- mut 1S.SUB eat-LCREFL ‗I started eating ✓i na7xw chen wa ilhen. and RL-still 1S.SUB IMPF eat and I‘m still eating.‘ Speaker‘s comment: I never got around to finishing.  The data in (41) appears to contradict our expectations about that the lc-reflexive predicate, as an lc-predicate. Other lc-predicates do not allow for a following sentences which indicates that its event continues, without inducing a contradiction. Because of those results, I have analyzed those other lc-predicates as having inherent final points. The lc-reflexive in (41), however, can be used in a context where the event continued. That is, the lc-reflexive does not appear to require event culmination. I now provide an account wherein these lc-reflexives still do have culmination entailments. I propose that that the lack of culmination entailment is only apparent. In particular, I propose that the lc-reflexive (with an unergative root) encodes two events. The first event is an event of getting oneself to the point wherein one can start eating. For example, the agent overcame his nausea after being sick. The second event is an event of eating. A more literal translation that would help express this could be ‗I managed to get myself to the point where I could start eating.‘ The reflexive event, some unnamed event in which the agent got himself to the point of being able to start eating, is the event that culminated. The clause with the continuation phrase in (41) is not a continuation of this first culminated event, but rather a continuation of the second event - the event of eating which may or may not have culminated. The data in (41), taken this way, then, does not 146  contradict our expectations of the lc-reflexives as lc-predicates. The lc-reflexives do have culmination entailments. This analysis predicts that unaccusatives should test differently than the unergatives with the lc-reflexive with the event continuation test. In Bar-el‘s (2005) analysis of unergatives, their only inherent point is a DO event, and the inherent point of an unaccusative is a BECOME event. In my analysis for (41), the culmination of the lcreflexive, with the unergative root, is when the agent reaches the point where he can begin the eating event. I predict that unaccusative roots should not allow for an event continuation clause because its inherent point is its culmination. I leave this for future research. One other feature of these lc-reflexives with unergative roots is the appearance of an overt DP for the theme/patient. For example, the above example in (41) could also have an overt DP as in (42).  (42) chen ilhenkwi sh wi 1S.SUB eat-LCREFL DET carrot ‗I ate the carrot.‘ Speaker‘s comment: ‗(You ate) the whole thing.‘  Although, I have not tested this, I predict that the presence of the overt DP in examples with the lc-reflexive with an unergative root will force a specific type of culmination reading. I predict that the culmination coincides, not with the starting point of the unergative as we saw in (41), but rather with the culmination of the eating of the carrot. The following are examples of the lc-reflexive with the scope of negation test.  147  Scope of negation test: the lc-reflexive (43) a. haw ‘-an i ts‘its‘ap‘NEG SBJ-1S.CONJ PRES work-LCRELF i) ‗I didn‘t get a job.‘ ii) ‗I didn‘t manage to work. b.  nam. go-LCREFL ‗(Somehow) he couldn‘t go.‘ haw  ‘-as  .  i  NEG SBJ-3CONJ PRES  In both examples in (43), the lc-reflexives only appear to obtain the did not start reading. The first example (43)a has two translations. The first translation describes a situation where the agent did not start working because they could not even get a job. The second translation simply indicates that the agent was not able to overcome some impediment to starting to work. This is the reason why he did not work. Example (43)b indicates that the agent did not go because he was unable to go, for whatever reason. The lack of ability is what prevented him from starting to work. That the lc-reflexives only obtain the did not start reading is unexpected from the viewpoint of other lc-predicates such as the lc-transitives and the lc-unergatives. Both of these other lc-predicates only obtained the did not culminate reading, and they did not obtain the did not start reading. My account for this unexpected behaviour is very similar to my account just presented for the behaviour of the lc-reflexive with the event continuation test. I propose that the lc-reflexive here also makes reference to two different events. In (43)a for the first translation, the first event was an event of looking for a job. This event actually took place and it is this event which did not culminate in the successful finding of a job. This is the reason for the agent not working. Thus we have a covert did not culminate reading with the lc-reflexive. 148  The second translation for (43)a, I argue also describes two events. The first event is the attempt at overcoming some impediment to starting to work (e.g. fatigue, laziness, distractions). But, the agent was not successful in overcoming this impediment. This lack of success is the reason for the agent not working. This, too, I argue is another instance of a covert did not culminate reading. Example (43)b, I argue, also describes two events similarly to the second translation of (43)a. The first event is an attempt to overcome some unnamed impediment to going. The agent, though, was unsuccessful at overcoming this impediment. The result of this is that the agent did not go. Again, I argue, that that this represents a covert almost culminated reading. The agent started an event of trying to overcome the impediment but this event did not culminate in him overcoming the impediment. The following two tables summarize our findings for the lc-reflexives.  -numut Table 49  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✘ ✘ Lc-reflexive (-numut) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  149  Readings induced by scope tests Test 3 Test 4 The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -numut --✘ ✓ Table 50 Lc-reflexive (-numut) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The results of the first test, given in Table 47, show that lc-reflexive (-numut) predicate does not allow for the cancellation of its culmination without inducing a contradiction. The second test showed that it is possible to add a sentence which says that its event continues without inducing a contradiction. I argued, though, that there is actually an event present which has culminated. This is the event leading up to where the agent got himself to the point where he is able to perform the activity described by the root. Thus I argue that both tests show that lc-reflexives have culmination entailments. I do not have data for the third test. The results of the fourth test, though, at first appeared to contradict our expectations of lc-reflexives as lc-predicates. They only appear to obtain the did not start reading, an event cancellation reading. I provide a similar account to what I said for the event continuation test. I proposed that the lcreflexive under the scope of negation describes two separate events. The first event describes an event wherein the agent started an event with the intention of getting to the point of starting the event described by the root. This first event is the event that started but did not culminate.  150  This, I argue, shows that lc-reflexives do have culmination entailments and they do have inherent final points, just like other lc-predicates.  4.3  Summary  The following two tables provide a summary of my findings using Bar-els‘ (2005) four tests for inherent final points, on four of the intransitivizers previously described by Jacobs (2007) as c- and lc-intransitivizers.  -im -sut -nalhn -numut Table 51  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ ✓ ✓ ✘ ✘ ✘ ✘ Intransitivizers and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested) Readings induced by scope tests  Test 3 The Scope of ilh (‗almost‘) Test Event Cancellation Event Non(= almost started) completion (=almost culminated) -im ✓ ✓ -sut ✓ ✓ -nalhn ✘ ✓ -numut ✘ ✓ Table 52 Intransitivizers and the scopal tests  Test 4 The Scope of Negation Test Event Cancellation Event Non(=did not start) completion (= did not culminate) ✓ ✓ ✓ ✓ ✘ ✓ ✘ ✓  (✘=not obtained; ✓=obtained; -- data not yet tested)  151  From these four tests I concluded that the two c-intransitivizers constructions do not have inherent final points and that the lc-intransitivizers do have inherent final points. This is summarized in the following Table 53.  Culmination Entailments Inherent final points -im ✘ ✘ -sut ✘ ✘ Lc-intransitives -nalhn ✓ ✓ -numut ✓ ✓ Table 53 Intransitivizers, culmination entailments and inherent final points C-intransitives  Besides not having culmination entailments, we noted that with the culmination cancellation test, c-intransitives can also occur with the tried to interpretation, just as the c-transitives can. The lc-predicates, though, do not have this interpretation.  tried to c-intransitives  lc-intransitives  Table 54  c-unergative  -im  ✓  c-reflexive  -sut  ✓  lc-unergative  -nalhn  ✘  c-reflexive  -numut  ✘  C- and lc-intransitives and the tried to interpretation  152  5  Applicatives and culmination  In this section I investigate whether or not predicates marked with applicatives require culmination. The four S wxwu7mesh applicatives are repeated in Table 55.  Form Label -nit relational applicative (RELAPPL) -shit redirective applicative (REDAPPL) -min causative applicative (CAUSAPPL) - h‟ w n benefactive applicative (BENAPPL) Table 55 Applicative transitivizers in S wxwu7mesh  These four applicatives have not been described for S wxwu7mesh as having a particular CONTROL  value. In this dissertation, I only examine these predicates with regards to  culmination entailments and leave the matter of CONTROL interpretations for further research. I investigate the applicatives in this chapter, with regards to their culmination properties, in anticipation of my morphosyntactic analysis of CONTROL constructions. Recall that for the causative applicative and the benefactive applicative, that they are followed by -t when an object or subject suffix is present. I argue in Chapter 5, that applicative constructions and c-predicates share one thing morphologically in common, the element -t. I will argue that the presence of this -t correlates with predicates which lack culmination entailments. I examine all four of the applicative constructions: -nit, -shit, -min and - h‟ w n. I use all four of Bar-el‘s tests for final points when it was possible to elicit the data. I do not have extensive data for the causative applicative -min because of problems with  153  elicitation. It only appears with a few lexical items. Some examples recorded in Kuipers (1967), are not recognized by contemporary speakers. I argue that although the data is limited, its behaviour mirrors that of the c-predicates: applicative predicates do not have culmination entailments and they do not have inherent final points.  5.1  The relational applicative -nit and culmination  Consider first the behaviour of a predicate marked with the relational applicative -nit in the context of the culmination cancellation test a in (44).  Culmination cancellation test: applicative predicate with -nit (44) a. na he-mi-nit-umulh-as ta mixalh, RL PROG-come-RELAPPL-1P.OBJ-3SUB DET bear ‗The bear was coming towards us i haw ‘as i tl‘i and NEG SBJ-3CONJ PRES arrive but didn‘t reach us.‘ b.  chen tkwaya7n-nit-umi 1S.SUB listen/hear-RELAPPL-2OBJ ‗I listened to you welh haw n-s-tkwaya7n. an chexw s7alh but none 1S.POS-NOM-listen/hear very 2S.SUB quiet but I didn‘t hear a thing. You were too quiet.‘  In (44)a, the event of the bear coming towards us does not have to culminate in the bear actually reaching us. In (44)b, the event of listening does not have to culminate in actually hearing anything that you say since you were speaking too quietly. A problem 154  with testing the predicate tkw y 7n in this example is that all that might be necessary for it to culminate is an act of listening. That is, actually hearing what one is listening to may not be a requirement for culmination. This is at least a problem with the pair of predicates ‗listen‘ and ‗hear‘ in English. But at least these data suggest that the relational applicative predicates do not have culmination entailments. The next examples are with relational applicative predicates with the event continuation test.  Event continuation test: applicative predicate with -nit (45) a. na he-mi-ni-t-umulh-as ta mixalh, RL RE-come-RELAPPL-TR-1P.OBJ-3SUB DET bear ‗The bear‘s coming closer to us i na7xw wa hemi and RL-still IMPF come and it‘s coming (even) closer.‘ b.  chen tkwaya7n-nit-umi 1S.SUB listen/hear-RELAPPL-2OBJ ‗I listened to you i na7xw chen wa tkw ya7n and RL-still 1s.sub IMPF listen and I‘m still listening.  In (45)a it is felicitous to indicate that the event of the bear coming this way can still continue. That is, the bear has not reached us yet. In (45)b it is felicitous to indicate that the event of listening may still be continuing. What is not clear from this example is whether we have two events of listening or just one. That is, does this sentence mean something like ‗I listened to you (before) and I‘m still listening to you (now).‘ These  155  data are suggestive that the relational applicative predicates do not have culmination entailments. The following shows the relational applicative predicate with the scope of ilh (‗almost‘) test.  Scope of ilh (‗almost‘) test: applicative predicate with -nit (46) na he-mi-nit-umulh-as ta mixalh. almost RL RE-come-RELAPPL-1P.OBJ-3SUB DET bear ‗The bear almost reached us. It came to us but it didn‘t reach us.‘ In (46) the scope of ilh (‗almost‘) obtains the almost culminated reading (i.e. a event non-completion reading). This is the reading where the event begins, but does not culminate in the bear reaching us. I did not find the almost started reading (i.e. an event cancellation reading) wherein no event of the bear coming towards us happened at all. This needs to be checked. As we have seen in §3 and §4, this test is not always reliable in testing for a predicate‘s inherent points. The following example shows the relational applicative predicate with the scope of negation test.  Scope of negation test: applicative predicate with -nit (47) haw ‘as i he-mi-nit-umulh-as ta mixalh. NEG SBJ-3CONJ PRES RE-come-RELAPPL-1P.OBJ-3SUB DET bear ‗The bear didn‘t come towards us. It went the other way. It didn‘t come at us at all.‘  156  In (47) in the scope of negation, the relational applicative predicate obtains the event did not start reading (i.e. an event cancellation reading). I did not find the did not culminate reading here. I provide a summary of our finds with the relational applicative -nit with Bar-el‘s (2005) tests for final points, and then provide some discussion.  -nit Table 56  Test 1 Test 2 Culmination Cancellation Test Event Continuation Test ✓ ✓ The relational applicative (-nit) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -nit --✓ ✓ Table 57 The relational applicative (-nit) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  Table 56 shows that for the relational applicative predicate, it is felicitous to deny its culmination without inducing a contradiction. It is also felicitous to indicate that the event of the relational applicative predicates continues, without inducing a contradiction. I argue that these two test show that relational applicative predicates do not have culmination entailments.  157  The data in Table 57 is a bit more complicated. The scopal test with ilh (‗almost‘) obtains the almost culminated reading (i.e. an event non-completion reading). In Bar-el‘s (2005) analysis, this should indicate that these predicates have inherent final points. But as we have seen elsewhere with the c-predicates, if a predicate has neither inherent initial or final points, then the predicate may obtain either reading with the scope of ilh (‗almost‘) test. The scope of negation test, though, obtains the did not start reading (i.e. an event cancellation reading). I argue that the availability of this reading shows that relational applicative predicates do not have inherent final points. I conclude that these tests indicate that relational applicative predicates do not have inherent final points.  5.2  The redirective applicative -shit and culmination  I present my findings for the applicative predicates with redirective applicative -shit first and then I will provide some discussion about these data. The following examples show the behaviour of the redirective applicative predicates with the culmination cancellation test.  Culmination cancellation test: applicative predicate with -shit (48) a. chen p‘aya -shit- mi ta a t txwem, 1S.SUB repair-REDAPPL-2OBJ DET 2POS car ‗I‘ve repaired your car welh haw ‘-an i h y-nexw-Ø. but NEG SBJ-1S.CONJ PRES finish-LCTR-3OBJ but I didn‘t finish it.‘ Context: I only did some but not all of the repairs. 158  b.  chen n m-shit-Ø t-en siy y (t)-t-en 1S.SUB go-REDAPPL-3OBJ DET-1S.POS OBL-DET-1S.POS ‗I was going to lend my car (to my friend)  t txwem, car  welh es-kw‘ y. but STAT-cannot but I couldn‘t.‘ Speaker‘s comment: ‗The car still needs repairs.‘ c.  chen nam-shit-Ø t-en siyay (t)-t-en 1S.SUB go-REDAPPL-3OBJ DET-1S.POS OBL-DET-1S.POS ‗I was going to drive my car to my friends welh haw  ‘an  BUT NEG SBJ-1S.CONJ  i PRES  tetxwem, car  tsixw. arrive  but I didn‘t make it.‘ Context: I decided not to. Context: I was going there and then it broke down.  In (48)a, it is possible to deny the culmination of the redirective applicative predicate without inducing a contradiction. The car does not have to be completely fixed. In (48)b, the predicate n m-shit besides meaning ‗to lend‘ can also mean ‗to take/bring to (someone)‘. It is possible to deny that the event culminates without inducing a contradiction. The described event indicates that there was only an intention to lend the car. Likewise, as in (48)c, the same predicate n m-shit can be used in a context where only an intention to lend happened (as indicated by the second context), and this is the reason for not arriving at the friend‘s place. But, this predicate can also be used in a context where the event of taking the car had started but it did not culminate because the car broke down half way there (as indicate in context one). These data show that the redirective applicative predicates do not have culmination entailments.  159  The following example shows the redirective applicative predicate with the scope of ilh (‗almost‘) test.  The scope of ilh (‗almost‘) test: applicative predicate with -shit (49) chen i n m-shit-Ø almost 1S.SUB PRES go-REDAPPL-3OBJ t-en siy y (t)-t-en t txwem. DET-1S.POS friend OBL-DET-1S.POS car ‗I was thinking of driving my car to my friends.‘ Context: I only thought about it and that‘s as far as I got. Context: I got it part way and then the car died. In (49) the scope of ilh (‗almost‘) obtains the almost started reading (i.e. an event cancellation reading). This is the first context where all that happened was an intention to bring the car to the friend‘s place. But it can also obtain the almost culminated reading (i.e. the event non-completion reading). This is the second context, where the car was brought half way to the friend‘s place and then it broke down. Again these data show the unreliability of the scope of ilh (‗almost‘) test with regards to inherent points. The next example is of the redirective applicative predicate with the scope of negation test.  The scope of negation test: applicative predicates with -shit (50) haw ‘an i n m-shit-Ø NEG SBJ-1S.CONJ PRES go-REDAPPL-3OBJ t-en siy y (t)-t-en t txwem. DET-1S.POS friend OBL-DET-1S.POS car ‗I couldn‘t make it to my friends place (to bring them my car).‘  160  In (50), the scope of negation test with the redirective applicative predicate obtains the event did not start reading (i.e. the event cancellation reading). I have not checked for the event non-completion reading. I provide a summary of our findings with the redirective applicative predicates, formed with -shit and Bar-el‘s four tests for final points and then provide some discussion.  -shit Table 58  Culmination Cancellation Test Event Continuation Test ✓ -The redirective applicative (-shit) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  Readings induced by scope tests The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -shit -✓ ✓ ✓ Table 59 The redirective applicative (-shit) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  Table 58 shows that for the redirective applicative predicate with -shit, it is felicitous to deny its cancellation without inducing a contradiction. I conclude that redirective applicative predicates do not have culmination entailments. I do not have any data with the event continuation test. The scope of ilh (‗almost‘) test as indicate in Table 59 is a bit more complicated. The redirective applicative predicate obtains both the almost started and the almost 161  culminated reading, just as it did with the relational applicative predicates. I take these results as evidence that the scope of ilh (‗almost‘) test does not only pick out inherent points. The negation test obtains the did not start reading (i.e. the event cancellation reading). I conclude, from these data, that the redirective applicative predicates do not have inherent final points.  5.3  The causative applicative -  and culmination  I present my findings for the applicative predicates with the causative applicative -min first and then I will provide some discussion about these data. The following example has the causative applicative predicate with the culmination cancellation test.  Culmination cancellation test: applicative predicate with -min (51) chen ch‘ wi7-t-umi, 1S.SUB fed.up-CAUSAPPL-TR-2OBJ ‗I got tired of you welh haw ‘an i h y-nexw-Ø but NEG SBJ-1S.CONJ PRES finish-LCTR-3OBJ and I‘m not done (being tired of you).‘  It is not clear what is being tested here, since it is not clear what the culmination of an event of getting tired/fed up with someone is. But if culmination requires finishing, then we can at least say here that the causative applicative predicate does not require that its event has finished without inducing a contradiction. The next example has the causative applicative with the scope of ilh (‗almost‘) test. 162  The scope of ilh (‗almost‘): applicative predicate with -min (52) chen i ch‘ wi7-[ ]-Ø kwen almost 1S.SUB PRES bored-CAUSAPPL-TR-3OBJ DET ‗I was almost bored with my brother.‘  s a7 younger.brother  It is not clear which reading is obtained with this test. It appears to be getting the almost culminated reading. I argue here, as I did with the other instances of this test we have seen, that it is not accurate with regards to picking out inherent points. The next example has the causative applicative predicate with the scope of negation test.  The scope of negation test: applicative predicate with -min (53) haw ‘an i ch‘ wi7-[ ]-Ø kwen s a7 NEG SBJ-1S.CONJ PRES bored-CAUSAPPL-TR-3OBJ DET younger.brother I didn‘t get bored with him (my younger brother) (I continued talking with him).‘  The scope of negation test, with the causative applicative predicate, obtains the did not start reading. The negation cancels the initial part of this event - the point of getting bored. I argue that this test at least shows that the causative applicative predicates do not have inherent final points. The following tables summarize our findings for the causative applicative predicates.  -min Table 60  Culmination Cancellation Test Event Continuation Test ✓ -The causative applicative (-min) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested) 163  Readings induced by scope tests The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Cancellation Event NonEvent Cancellation Event Non(= almost started) completion (=did not start) completion (=almost (= did not culminated) culminate) -min --✓ ✓ Table 61 The causative applicative (-min) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  The results of the first test show that causative applicative predicates do not have to finish. With the particular predicate used in this example there are problems in defining what a culminated event would look like. I argued that the fact that the event does not have to finish is consistent with a predicate that does not have culmination entailments. The results of the scope of ilh (‗almost‘) are ambiguous as I have argued they are with other types of predicates. The scope of negation test shows that these causative applicative predicates do not have inherent final points. I conclude that the causative applicative predicates do not have inherent final points.  5.4  The benefactive applicative -  and culmination  I present my findings for the applicative predicates with benefactive applicative - h‟ wan first and then I will provide some discussion about these data. The following examples have the benefactive applicative predicate with the culmination cancellation test.  164  Culmination cancellation test: applicative predicate with - h‟ w n (54) a. na lulum- ’ -t-umulh-as ta Ha7lh Skwayl Slulum. RL sing-BENAPPL-TR-1P.OBJ-3SUB DET good day song ‗They (the children) tried to help us sing (the Good ay Song) welh eskw‘ay but STAT-cannot but they couldn‘t do it. ‗ Speaker‘s comment: ‗They weren‘t familiar with the song.‘ b.  chen kw‘enmaylh- ’ -t-umi-yap ta Ha7lh Mali, 1S.SUB pray-BENAPPL-TR-2OBJ-2PL DET good Mary ‗I was going to say the Hail Mary for you, welh chen may-nexw-Ø but 1S.SUB forget-LCTR-3OBJ but I forgot the prayer.‘  In example (54)a the children never sang the song that they were going to help to sing. They only tried to help with singing it. In (54)b, the culmination of the event - the saying of the prayer - never actually took place, and thus the event never culminated. Both examples show that it is possible to deny the culmination of the benefactive applicative predicates without inducing a contradiction. The next examples are with benefactive applicative predicates with the event continuation test.  Event continuation test: applicative predicate with - h‟ w n (55) chen kw‘enmaylh- ’ -t-umi ta Ha7lh 1S.SUB pray-BENAPPL-TR-2S.OBJ DET good ‗I‘m saying the Hail Mary for you i na7-xw chen wa kw‘enmaylh. and RL-still 1S.SUB IMPF pray but I‘m still praying.‘  165  Mali, Mary  In (55) it is felicitous to indicate that the event of praying the Hail Mary is still continuing without inducing a contradiction. The following example is the scope of ilh (‗almost‘) with the benefactive applicative predicate.  Scope of ilh (‗almost‘): applicative predicate with - h‟ w n (56) chen lulum- ’ -t-umi ta ha7lh almost 1S.SUB sing-BENAPPL-TR-2S.OBJ DET good ‗I was going to sing to you (the Good ay Song)  skwayl slulum, day song  welh es-kw‘ay. chen es-7i7xi. chen may-nexw-Ø but STAT-cannot 1S.SUB STAT-shy 1S.SUB forget-LCTR-3OBJ but I couldn‘t make it (because) I was shy (or because) I forgot it.‘ The scope of ilh (‗almost‘) obtains the almost started reading (i.e. an event cancellation reading). This is the reading where the event does not begin. The following tables summarize our findings for the benefactive applicative (h‟ w n) and Bar-el‘s (2005) tests for inherent final points.  - h‟ w n Table 62  Culmination Cancellation Test Event Continuation Test ✓ ✓ The benefactive applicative (- h‟ w n) and the culmination entailment tests (✘=infelicitous; ✓=felicitous; -- data not yet tested)  166  - h‟ w n Table 63  Readings induced by scope tests The Scope of ilh (‗almost‘) Test The Scope of Negation Test Event Event NonEvent Event NonCancellation completion Cancellation completion (= almost started) (=almost (=did not start) (= did not culminated) culminate) ---✓ The benefactive applicative (- h‟ w n) and the scopal tests (✘=not obtained; ✓=obtained; -- data not yet tested)  As shown in Table 62 it is possible to deny the culmination of the benefactive applicative predicate without inducing a contradiction. This shows that these predicates do not have culmination entailments. I do not have any examples of the event continuation test. In Table 63 the results of the scope of ilh (‗almost‘) test is consistent with a predicate that lacks inherent final points (although as I have already argued, this test is not always reliable for picking out inherent points). I do not have any examples of the scope of negation test. I conclude from these findings that benefactive applicative predicates do not have inherent final points.  6  Culmination implicatures  Thus far I have shown that c-predicates do not entail culmination. However, as I now show, c-predicates are still compatible with event culmination, a property first described for c-transitives in S wxwu7mesh by Bar-el et al. (2005) and Bar-el (2005) (Bar-el et al. 2005 also show this for c-transitives in Lillooet). Both studies show that in out-of-theblue contexts c-transitives are most naturally interpreted as referring to a culminating 167  event. To put it another way, c-predicates in out-of-the-blue contexts imply that the event culminated. That is, c-predicates have culmination implicatures. Both Bar-el et al. (2005) and Bar-el (2005) provide a modal analysis to derive this implicature. I examine their analysis in Chapter 5,§5.3, after providing my own analysis in Chapters 4 and 5. In this section, I demonstrate that most, but not all, c-predicates in out-of-the-blue contexts have culmination implicatures. The data I draw on primarily is taken from Bar-el (2005) and Bar-el et al. (2005). I also include data from my own field work. Bar-el et al. (2005) note that in out-of-the-blue contexts, and without any explicit denial of culmination, accomplishment c-predicates in the perfective aspect are normally translated in the past tense and involve culmination. I present data for each of the ctransitives here. Since neither Bar-el et al. (2005) nor Bar-el (2005) have examples of a c-predicate with -Vt nor with the causative -s, I provide data from my own field work. They confirm the conclusion that S wxwu7mesh c-predicates are compatible with culmination and even imply culmination in out-of-the-blue contexts. Consider the following examples with the -t transitivizer (57), with the -Vt transitivizer (58), with the -Vn transitivizer (59) and with the causative transitivizer -s (60). Out-of-the-blue context with the -t transitivizer (57) na xel-t-Ø-as ta sxwexwiy m lha Mary RL write-TR-3OBJ-3SUB DET story DET Mary ‗Mary wrote a story.‘ Speaker‘s comments: ‗She wrote it ...she‘s finished.‘ (Bar-el et al 2005:6a)  168  Out-of-the-blue context with the -Vt transitivizer (58) a. chen lh ch‘-it-Ø ta sepl n 1S.SUB cut-TR-3OBJ DET bread ‗I cut the bread.‘ b.  a-stl‘ 7 u 2POS-want POL ‗ o you want some?  Out-of-the-blue context with the -Vn transitivizer (59) chen p‘ ya -en-[ ]-Ø ta t txwem 1S.SUB fix-TR-TR-3OBJ DET car ‗I fixed the car.‘ Speaker‘s comments: ‗You already fixed it.‘ (Bar-el et al 2005:6b) Out-of-the-blue context with the -s transitivizer (60) chen t 7-s-[ ]-Ø ta sitn 1S.SUB do-caus-TR-3OBJ DET basket ‗I made a basket.‘ Speaker‘s comments: ‗ hen melh h ynexw (‗Yes I finished it‘).‘  In all cases, c-predicates in out-of-the-blue contexts, without any denial of culmination, are translated in the past and are interpreted as reaching culmination. In examples (57), (59) and (60) the speaker indicates that the event is finished as intended. In (58) the speaker assumes that the bread has been cut and thus it is ok to question if someone wants some of this bread. The same facts obtain with the c-reflexive. The data presented here is from my own field work. Since the c-reflexive can occur with any of the c-transitivizers -t, -Vt or -Vn, I present data with each transitivizer plus reflexive.  169  Out-of-the-blue context with -t and c-reflexive (61) na kw lash-t-sut-Ø RL shoot-TR-CREFL-3SUB ‗He shot himself.‘ Speaker‘s comment: ‗He killed himself.‘ Out-of-the-blue context with -Vt and c-reflexive (62) a. chen ch‘ w-at-sut 1S.SUB help-TR-CREFL ‗I helped myself.‘ Speaker‘s comment: ‗You were successful.‘ b.  chen kw‘lh-at-sut 1S.SUB spill-TR-CREFL ‗(I was riding my canoe and) I tipped over (lit. I had a spill).‘ Interviewer: ‗It happened?‘ Speaker: ‗Yes. It really happened.‘  Out-of-the-blue context with the -Vn and c-reflexive (63) chen kw y-an-t-sut 1S.SUB hide-TR-TR-CREFL ‗I hid myself.‘ Speaker‘s comment: ‗You‘re hiding behind the house right now.‘  The c-reflexives in (61), (62) and (63) are all translated in the past, regardless of the ctransitivizer they are based on. In an out-of-the-blue context, the c-reflexive with -t in (61) indicates that the subject killed himself by shooting himself. In an out-of-the-blue context, in (62)a, the c-reflexive with -Vt, the subject was successful in helping herself. In an out-of-the-blue context, in (62)b, the c-reflexive with -Vt obtains a reading wherein the canoe actually tipped over. In (63), in an out-of-the-blue context, the subject was successful in hiding himself. I conclude that c-reflexives (just like c-transitives) are associated with culmination implicatures. C-transitives and c-reflexives (which are based on c-transitives) differ, however, from the c-unergatives with regards to culmination implicatures. The latter are often not 170  translated in the past tense and are often not translated with culmination. Bar-el et al. (2005) and Bar-el (2005) examine the translations of activity predicates in out-of-the-blue contexts, but they do not examine c-unergative predicates. I provide some examples here of the c-unergative from my own database.  Out of the blue context with the c-unergative (64) a. chen tsel w‘1S.SUB kick-CUE i) ‗I kick‘ ii) ‗I am kicking.‘ b.  na RL  i) ii)  kw‘sh-im-wit (t)ta t la count-CUE-PL OBL-DET money ‗They‘re counting their money.‘ ‗They counted their money.‘  s t- -Ø RL hand.out-CUE-3SUB ‗He was paying.‘  c.  na  d.  chen y lh1S.SUB burn-CUE ‗I did a burning I made a fire.‘  e.  chen taht-kwi 1S.SUB make-CUE OBL-DET ‗I‘m making a house.‘  lam house  f.  chen tah1S.SUB make-CUE ‗I built a house.‘  lam house  g.  chen p‘ ts‘1S.SUB sew-CUE ‗I knit a sweater.‘  t-ta OBL-DET  (t)-ta swita OBL-DET sweater  In examples (64)a-c of the c-unergative, the event can be interpreted as ongoing, in the present tense (64)a,b,e or in the past tense (64)b-d. It can have an habitual reading, as in 171  translation (i) for (64)a. Without an overt oblique DP for the patient, the patient is translated as non-specific in (64)d. With an overt DP (64)e with the non-deictic determiner kwi, the event is translated in the present tense without culmination, but with an overt DP with the deictic determiner ta, the event is translated in the past tense and with an apparent event culmination interpretation.27 From this set of meanings of cpredicates with the c-unergative, I conclude that they do not imply culmination, although they are not incompatible with culmination. I will thus describe them as unspecified for culmination. In this section I showed that c-transitives and c-reflexives have culmination implicatures in out-of-the-blue contexts. In contrast, I showed that c-unergatives do not have culmination implicatures in out-of-the-blue contexts. This is summarized in the table below.  27  The implicature or lack of implicature that the event is completed may actually be coming not from the predicate but from the determiner. The non-deictic determiner kwi could be said to be blocking any implicature of event culmination or the deictic determiner ta could be said to be giving rise to this implicature. This requires further research (Thank you to Carrie Gillon for discussion on the possible role of the determiners).  172  Culmination implicature c-transitives: ✓ -t c-transitivizer -Vt c-transitivizer -Vn c-transitivizer -s causative transitivizer C-intransitive: -sut c-reflexive C-intransitive ✘ -im c-unergative Table 64 C-predicates and culmination implicatures  7  Summary  In this chapter, we have seen that there is a strict correspondence between culmination and the morphological marking for CONTROL in S wxwu7mesh. Using Bar-el‘s (2005) four tests for inherent final points, we have found for all the lc-predicates I have investigated that they have culmination entailments. This includes the lc-transitivizer -nexw, the lc-unergative -nalhn and the lc-reflexive -numut. In contrast, these same tests have shown that none of the c-transitivizers or intransitivizers that I investigated have culmination entailments. My results with the scope of ilh (‗almost‘) were more variable than Bar-el‘s (2005) finding and I concluded that it is a less reliable test for inherent points. I also tested the four applicative predicates with these same four tests and found that none of them have culmination entailments. In the final section, however, I showed that all of c-predicates are still compatible with event culmination, i.e., they do not encode non-culmination. In fact, in out-of-the-blue contexts they are normally considered to encode culmination. The c-unergative has more variable behaviour with 173  regards to culmination. I concluded from this that they do not have culmination entailments nor do they have culmination implicatures. These findings for this chapter are summarized in Table 65. Note that if a predicate already has culmination entailments, it cannot have culmination implicatures also.  Culmination Entailment Implicature Core transitives -t ✓ ✘ -Vt ✓ ✘ -Vn ✓ ✘ -s ✓ ✘ -nexw ✓ Applicatives -nit -✘ -shit ✘ --min ✘ -- h‟ w n ✘ -Intransitives -im ✘ ✘ -sut ✓ ✘ -nalhn ✓ -numut ✓ Table 65 Summary of culmination entailments and implicatures  We now have a partial answer to the question regarding the semantic-morphology mapping: lc-predicates require culmination, c-predicates are compatible with but do not require culmination and finally c-unergatives do not have a culmination implicature. The fact that transitive c-predicates do not require culmination is remarkable from the point of view of English. As Bar-el (2005) notes, these predicates are akin to accomplishment predicates in English; but in English, event culmination with accomplishments in the perfective is required. Consequently, Bar-el (2005) argues that 174  the class of accomplishments differs cross-linguistically as to whether or not they contain a final point. Moreover, Kiyota (2008:82) argues that lc-transitives in Saanich, which do require culmination, are transitive achievements. Kiyota, assuming avis‘ (1997) Deep Unaccusativity Hypothesis, proposes that the function of the lc-transitivizer is simply to take an intransitive achievement and add an overt agent. Even more recently, Turner (2010) has shown, that c- and lc-reflexives in Saanich pattern like c- and lc-transitives. The c-reflexives have culmination implicatures but the lc-reflexives have culmination entailments. Non-culminating accomplishments have been observed to exist in a variety of languages. For instance, Kothari (2008) demonstrates this for Hindi. In the following example, it is felicitous to deny the culmination of the event of the Hindi predicate khaa ‗to eat‘ in the perfective aspect without inducing a contradiction.  (65) maayaa-ne apnaa sandwich khaa-yaa Maya-ERG her sandwich eat-PERF ‗Maya ate her sandwich ‘ par use khatam nahiin but it-ACC finish not ‗but did not finish it.‘  ki-yaa do-PERF  Other researchers have described non-culminating accomplishments for many other unrelated languages such as Koenig and Muansuwan (2000) for Thai, Ritter and Rosen (2000) for Central Pomo, Dyirbal, Icelandic, Irish, Southern Tiwa, Lakhota, and in some instances in Japanese, Smith (1997) for Chinese and Travis (2010) for Malagasy.  175  While exploring the culmination properties of c- and lc-predicates, we also noted that c-predicates, with the culmination cancellation test, obtain what I have described as the tried to interpretation. This interpretation is obtained even though the predicate t‟  tsut  ‗to try‘ is not present. In contrast, lc-predicate do not obtain this tried to interpretation. These findings are summarized in Table 66 following.  c-intransitives lc-intransitives  Table 66  c-unergative c-reflexive lc-unergative  -im -sut -nalhn  tried to ✓ ✓ ✘  c-reflexive -numut ✘ C- and lc-intransitives and the tried to interpretation  I argue now that there is another way to think about the connection between CONTROL-marking,  culmination and the tried to interpretation. According to Ritter and  Rosen (2000), languages divide into delimiting languages (D-languages) and initiating languages (I-languages). D-languages are characterized by accomplishments patterning with achievements: they require culmination. Accordingly, Chinese, English, Finnish and Haitian Creole belong to the class of D-languages. In contrast, I-languages are characterized by accomplishments patterning with activities: they do not require culmination. Accordingly, Icelandic, Irish and Japanese belong to the class of Ilanguages. Ritter and Rosen‘s (2000) - and I-languages are summarized in Table 67. The shaded cells indicate predicate types that have culmination entailments. The clear cells indicate predicate-types that do not have culmination entailments.  176  Activities (run, sing, walk)  Accomplishments Achievements (build a house, fix a car) (arrive, win a race)  I-languages (Icelandic, Irish, Japanese) D-languages (English, Finnish, Haitian Creole, Chinese) Table 67 I-languages and D-languages (Ritter and Rosen 2000)  S wxwu7mesh presents an interesting case in that it is a language where both patterns are observed. As far as culmination entailments are concerned, c-predicates pattern with accomplishments in I-languages and could thus be characterized as (minimally) initiating predicates, or I-predicates. We find the clearest evidence for this, I argue, with the tried to interpretation that occurs with c-predicates with the culmination cancellation test. This is shown in Table 68 with the clear cells. In contrast, lc-predicates pattern with achievements in that they have culmination entailments, and thus could be characterized as delimiting predicates (or D-predicates). This is shown in with the shaded cells in Table 68.  Activities  Accomplishments Achievements I-language C-predicates D-language Lc-predicates Table 68 I-languages and -language behaviour in S wxwu7mesh  177  An important point to make here is that I am not conflating initiating predicates with predicates that have inherent initial points. Bar-el‘s (2005) examination of accomplishments and inherent initial points already requires us to make such a distinction. Take for example, one of Bar-el‘s tests (2005:140) for initial points - the ―readings induced by punctual/adverbials test‖. She uses this test to see what reading(s) can occur with a predicate when a punctual temporal phrase is present. The assumption in this test is that if a predicate has an inherent point, then this punctual temporal phrase will pick out that point. Thus the punctual temporal phrase and the predicate‘s inherent initial point will coincide. For accomplishments, Bar-el found that they are quite variable regarding which subevent (i.e. initial, mid or final subevent) the punctual temporal phrase picked out. In the following example adapted from Bar-el (2005), the punctual temporal phrase n  tkwi nus ‗at two o‘clock‘ may coincide with the inception of the c-predicate event,  that is, with the starting point of the writing of the story, the mid point of the event of writing the story, or the final point of the event of writing the story, that is, when the story is finished being written.  (66) na RL  x l-t-Ø-as ta sxwexwiy m kwa John write-TR-3OBJ-3SUB DET story DET John  na7 t-kwi nusat OBL-DET two-o‘clock ‗John wrote the story at two o‘clock.‘ Context: Inceptive; he started writing at two o‘clock. Context: Medial; he was in the midst of writing at two o‘clock. Context: inal; he finished the story at two o‘clock. (Bar-el 2005:163, ex.46) 178  Since the punctual temporal phrase does not consistently pick out any one subevent, Barel concludes that c-transitives do not have inherent initial points, and argues that it also provides supporting evidence for her proposal that they do not have inherent final points. If they had either inherent initial or final points, then the punctual temporal phrase should pick out this point. Take the medial reading in (66). It indicates an event has been initiated but not yet finished. Even though the predicate is used to describe an initiated event, this does not force an alignment of the punctual temporal phrase with this initiation point. The temporal phrase may, in fact, still align with the medial part of the event. As per Bar-el (2005), I argue that this shows that event initiation and inherent initial points are not the same thing. A c-predicate can be used to describe an event which has been initiated but this does not mean that they have inherent initial points. Likewise, a c-predicate can describe an event which has culminated, but this also does not mean that they have inherent final points. A possible reason why c-predicates indicate event initiation is as follows. When a cpredicate indicates that its event has taken place (i.e. the c-predicate is in the pastperfective), then the minimal requirement for that to be true is that the agent initiated that event. Therefore, when the process and the culmination of the event are denied, all that is left to assert is the initiation of the event. In Chapter 5, §6.3, I return to the issue of how predicate classes are constructed in S wxwu7mesh and compare Bar-el‘s (2005) account to mine with regards to predicates with inherent initial points versus initiating predicates. As for the notion of CONTROL, if c-predicates are I-predicates and lc-predicates are D-predicates, the question is raised as to how event initiation and event culmination relate 179  to this notion of CONTROL? What determines whether or not c- and lc-marking is associated with a difference in the degree of control the agent has over the event? This is the question I take on in the next chapter.  180  Chapter 4 - The context of use of CONTROL  1  Introduction  The goal of this chapter is to investigate the context of use for c- and lc-predicates. We have seen in the last chapter that S wxwu7mesh has a productive contrast between predicates that entail culmination and those that do not, that is, our lc- and c-predicates, respectively. We also saw that c-predicates, with the culmination cancellation test, can occur with the tried to interpretation, but lc-predicates cannot. At the end of Chapter 3, I further proposed that another way to look at c- and lc-predicates is as initiating predicates (I-predicates) and delimiting predicates (D-predicates), respectively, following Ritter and Rosen‘s (2000) language typology. An initiating-predicate, I proposed, minimally requires event initiation but event culmination is not required. Similarly, I assumed that delimiting-predicates minimally require event culmination, i.e., that the end of the event is, in fact, demarcated by its natural endpoint. In (1)a, I informally represent cpredicates. The event is initiated (ie=initiating event) but nothing is said explicitly about the nature of the end of the event: it may be reaching its natural endpoint but it need not. This is indicated by the absence of fe (final event) at the right edge of the event.  181  (1)  a.  c-predicates = initiating predicates28 [ie --------------]event  b.  lc-predicates = delimiting predicates [-------------fe]event  I informally represent lc-predicates as in (1)b. The event culminates. This is indicated by the presence of fe. Nothing, however, is said explicitly about the nature of the beginning of the event. This is indicated by the absence of ie at the left edge of the event. In this chapter I argue that the so-called control and limited control interpretations emerge because of this contrast, but are not directly encoded by any morpheme in the cand lc-predicates themselves. That is, CONTROL-interpretations do not arise because they are part of the lexical entry of any of the (in)transitivizing suffixes. Rather, I argue, they arise via an interaction of the linguistic context (I- or D-predicate) and the discourse context. I call the combination of these two contexts – the context of use. In particular, c-predicates, since they only demarcate the initiation of the event, can be used in contexts where the natural endpoint of the event is not reached. Conversely, lc-predicates, since they only demarcate that there is a final event, but say nothing about how this final event comes about, can be used in contexts where the natural endpoint comes about in unusual ways.  28  I take this representation to be informal in nature, not to be confused with the formal representation of Bar-el (2005). Bar-el argues that initial points are not represented in S wxwu7mesh accomplishments (my c-predicates). I will have to leave the formal implementation of the generalization in (1) for another occasion.  182  As a point of departure, I start with an overview of the literature on the meanings associated with limited-control (§2). I then show which of these meanings are obtained by S wxwu7mesh lc-predicates (§3). Finally I show how these meanings are derived under the present proposal (§4).  2  Background  In this section I provide the background on the control interpretations that may be associated with c- and lc-predicates. In the literature on CONTROL, the discussion is mainly in terms of the readings associated with CONTROL. On my analysis, however, these meanings are not encoded in the predicates. Instead, the semantic contrast is that between an initiating and a delimiting predicate. All other meanings are not directly encoded in the predicates themselves. Rather, they reflect the context of use for these predicates.29 Some contexts of use associated with a given predicate are somewhat unexpected from the point of view of English, a purely delimiting language. As a consequence, translations will sometimes reflect this discrepancy between S wxwu7mesh and English predicates. As discussed in Chapter 1, Thompson (1979) is the first to identify CONTROL as the relevant contrast that divides Salish predicates into two classes. Since then, other  29  A similar distinction is made in Davis, Matthewson, and Rullmann 2009. They refer to meaningcomponents that are directly encoded as ―readings‖ whereas meaning-components which arise via inferences are referred to as ―interpretations‖. On the present view ―reading‖ corresponds to the core meaning of a predicate and ―interpretation‖ corresponds to the context of use.  183  researchers such as Demirdache (1997), Davis and Demirdache (2000) and Davis, Matthewson and Rullman (2009) have further refined our understanding of the interpretation of the non-control construction in Lillooet. Note that Lillooet differs from S wxwu7mesh in its morphological make-up. While both control and limited control marking are tied to the system of morphological transitivizers and intransitivizers in S wxwu7mesh, in Lillooet control marking is associated with transitive marking while non-control is associated variously. In Lillooet, as in S wxwu7mesh, a c-predicate can be derived from an intransitive root by a transitivizer -Vn (cognate to the S wxwu7mesh transitivizer -Vn). Compare the unaccusative root in (2)a with the same root transitivized with the -Vn transitivizer in (2)b.  Lillooet (2) a.  b.  k‘ c ti s-ts‘w n-a dry DET NOM-salmon-EXIS ‗The salmon dried the salmon is dry‘ (Demirdache 1997:ex. 1a) k‘ c- ’-as ti s-ts‘w n-a dry-DIR-3ERG DET NOM-salmon-EXIS ‗ ucky dried the salmon.‘ (Demirdache 1997:ex.52)  s-Bucky NOM-Bucky  Non-control interpretations can be associated with a dedicated circumfix ka- -a, the ‗circumstantial modal‘. It obtains the managed to interpretation, attached to an intransitive predicate in (3)a, or a predicate transitivized with the causative in (3)b.  184  (3) a.  b.  ….  t‘u7 ka-tsunam‘-cal=kan-a=t‘u7 but CIRC-teach-ACT=1SG.SUBJ-CIRC=ADD ‗(I was sick) but I still managed to teach.‘ ka-gw l-s=kan-a CIRC-burn-CAUS-1S.SUBJ-CIRC ‗I managed to get it lit.‘ (Davis et al. 2009:ex.6b)  The causative transitivizer has been described as a neutral control marker in Lillooet (Demirdache 1997, van Eijk 1997:111). For the causativized predicate in (4), van Eijk (1997:111) proposes that this predicate is neutral control because ‗the subject causes the object to carry out an action over which the object is in control (i.e., the object is not in full control, since it had to share control with the object).‘  (4)  tl‘i -s30 arrive-CAUS ‗to bring (here)‘ (= ‗to cause to arrive here‘)  As a consequence, non-control marking does not systematically contrast with its converse, control, in Lillooet. Davis et al. (2009) focuses their investigation on the interpretation of non-control but do not address the interpretations associated with control. For control they assume the analysis of Bar-el et al. (2005), wherein the transitivizer -Vn introduces control.  30  This predicate is a cognate to the S wxwu7mesh predicate tl‘i -s ‗to bring (him/it) here‘.  185  Another difference between non-control marking in Lillooet and lc-predicates in S wxwu7mesh is that non-control predicates in Lillooet are not associated with culmination entailments. I do not have data on the causative, though, so it is not clear if they also lack culmination entailments. Since the morphological systems associated with CONTROL  differ between the two languages, it is therefore expected that the context of use  for non-control marking differs from that of limited control marking in S wxwu7mesh. Nevertheless, I use the five interpretations that Davis et al. (2009) describe for ka- -a as the basis for my investigation into the context of use for S wxwu7mesh lc-predicates. This is because it is the most thorough description of the available meanings to date. Davis et al. (2009) identify the following five interpretations associate with noncontrol predicates. (5)  Interpretations of ka-…-a: a. ability b. manage-to c. accidentally d. suddenly/unexpectedly e. non-controllable  I illustrate each of these interpretations with one of the Lillooet examples provided by Davis et al. (2009). The ability interpretation covers typical ability attributions, which in English are expressed with can or be able to.  (6)  c y‘=lhkacw=ha ka-cwák-a lh=ma7g‘ lm‘ecw=as going.to=1SG.SUBJ=YNQ CIRC-wake-CIRC COMP=daybreak=3CONJ ‗ re you going to be able to wake up at dawn?‘ ( avis 2006)  186  The manage-to interpretation indicates that the event required an unusual amount of effort.  (7)  ka-gwél-s=kan-a CIRC-burn-CAUS=1SG.SUBJ-CIRC ‗I managed to get it lit.‘ (van ijk 1997:51)  The accidentally interpretation indicates that the action was not on purpose.  (8)  ka-gwél-s=kan-a ta=n-g y‘tten=a CIRC-burn-CAUS=1SG.SUBJ-CIRC DET=1SG.POSS-bed=EXIS ‗I accidentally set my bed on fire.‘ ( avis 2006)  The suddenly/unexpectedly interpretation indicates that the event happened suddenly or abruptly.  (9)  ka-q‘ek‘w-ts=kán-a CIRC-close-mouth=1SG.SUBJ-CIRC ‗My mouth got closed suddenly.‘ ( lexander et al. to appear)  Finally, the non-controllable interpretation arises when the event is not controllable by an animate agent.  (10) ka-lhéxw-a ta=snéqwem=a CIRC-come.up-CIRC DET=sun=EXIS ‗The sun came out.‘ ( avis 2006)  Davis et al. (2009) argue that the lexical meaning of ka- -a is as a circumstantial modal and not as a marker of the degree of control the agent has over the event. We get the 187  impression that we are dealing with control because of the interaction of circumstantial modality and the context of use. In this respect my analysis of S wxwu7mesh lcmarking converges with Davis et al. (2009) analysis of non-control marking. The difference in our analyses is based on the different behaviour of the respective constructions. Davis et al. argue that non-control interpretations are derived from the lexical meaning of ka- -a as a circumstantial modal, while in my analysis the limited control interpretations are derived from the aspectual nature of the CONTROL constructions. I argue that event culmination and event initiation – aspectual notions – underlie the contrast in CONTROL in S wxwu7mesh. Davis et al. (2009) argue that the core meaning of ka- -a is a modal meaning, since the ka- -a predicate in Lillooet does not have an actuality entailment, meaning the ka- -a marked event does not have to have taken place in the actual world as the following sets of data demonstrate.  (11) a.  qwenúxw=kan i=nátcw=as, sick=1SG.SUBJ when.PAST=day=3CONJ ‗I was sick yesterday t‘u7 ka-tsunam‘-cal=lhkán-a=t‘u7 but CIRC-teach-ACT=1S.SUBJ-CIRC=ADD but I still managed to teach.‘ ( avis 2006)  b.  qwenúxw=kan i=nátcw=as, sick=1SG.SUBJ when.PAST=day=3CONJ ‗I was sick yesterday. ka-tsunam‘-cal=lhkán-a=ka, t‘u7 cw7 oy=t‘u7 CIRC-teach-ACT=1S.SUBJ-CIRC=IRR but NEG=ADD I could have taught but I didn‘t.‘ ( avis 2006)  188  These two examples differ in that in (11)a the teaching event marked by ka- -a is asserted to have happened in the real world, whereas in (11)b the teaching event, also marked with ka- -a, is asserted to not have taken place in the real world. This suggest that predicates with ka- -a do not have actuality entailments. These predicates may imply that they took place in the real world but they do not entail it. Such a difference in the core meaning of a control construction is to be expected if CONTROL  is a construct in Salish. That is, we may expect CONTROL to be derived in  different ways, with different morphological, syntactic, semantic and pragmatic properties. As shown for Lillooet, some CONTROL meanings are in part associated with modality. The interpretations associated with this construction overlap with limited control interpretations in S wxwu7mesh, but they also diverge. In S wxwu7mesh CONTROL  is associated with an aspectual meaning. Possibly in languages where CONTROL  is associated with -C2 reduplication (e.g. Spokane), which is also described as having aspectual properties (Carlson 1996), it has different, but overlapping, properties with CONTROL  in S wxwu7mesh and Lillooet.  In the following sections I will investigate the meanings that occur with lc-predicates in S wxwu7mesh.  3  What can limited control predicates mean?  The goal of this section is to examine whether the five interpretations of ka- -a listed in (5) above are available for S wxwu7mesh lc-predicates, including those which contain the lc-transitivizer, the lc-unergative and the lc-reflexive. We will see that some, but not 189  all, interpretations are compatible with lc-predicates. Out of the five interpretations compatible with non-control predicates in Lillooet, only three are available for lcpredicates in S wxwu7mesh. The results of this investigation are summarized in Table 69 following.  Lillooet i) managed to ✓ ii) accidentally ✓ iii) able to ✓ iv) unex