Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

On the classification of predicates in Nłe?képmx (Thompson River Salish) Howett, Catherine Dawn 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-0087.pdf [ 1.37MB ]
JSON: 831-1.0087429.json
JSON-LD: 831-1.0087429-ld.json
RDF/XML (Pretty): 831-1.0087429-rdf.xml
RDF/JSON: 831-1.0087429-rdf.json
Turtle: 831-1.0087429-turtle.txt
N-Triples: 831-1.0087429-rdf-ntriples.txt
Original Record: 831-1.0087429-source.json
Full Text

Full Text

On The Classification of Predicates inN+e’?képmx(Thompson River Salish)byCatherine Dawn HowettB.A., The University of British Columbia, 1984A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF ARTSinTHE FACULTY OF GRADUATE STUDIES(Department of Linguistics)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIADecember 1993© Catherine Dawn Howett, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department ofThe University of British ColumbiaVancouver, CanadaDate -Z3I773DE-6 (2/88)ABSTRACTIn this thesis I discuss the semantic basis of the morphological form of predicatesin N+e?képmx, a Northern Interior Salish language. Intransitive and transitive use ofroots in Nle7képmx is morphologically marked; intransitives use a set of primary affixesand transitives use a set of transitivizers. I document the behavior of these morphosyntactic affixes with a subset of the predicates of Me?képmx to determine what isoptional, what is obligatory and what is blocked. I link this to an analysis of argumentstructure of predicates and subsequently create a classification of predicate types.I present an overview of the intransitive and transitive morphology of Meképmxin Chapter One. In Chapter Two I discuss current literature regarding the syntactic andsemantic diagnostics of unaccusative and unergative verbs. I create a semanticclassification of the set of roots, and discuss the behavior of roots with morpho-syntacticaffixes to determine the diagnostic potential of the affixes. In Chapter Three I discuss thepotential of an intransitive-transitive classification of roots.The data show that there is an unergative and unaccusative distinction in thelanguage, specific aspectual morpho-syntactic diagnostics distinguish unaccusatives andcausative and desiderative distinguish unergatives. The traditional analyses of Salishlanguages as having a majority of unaccusative roots and no underlying transitives isconfirmed.— 11 —AbstractTABLE OF CONTENTS11Table of Contents 111List of Tables . VAcknowledgements viDedication viiIntroduction .1Chapter One Overview of N+e?képmx MorphologyThe Pronominal SystemSimple Intransitive PredicatesPrimary AffixesTransitive PredicatesTransitivizing SuffixesComplex Intransitives: Reflexives and ReciprocalsRe-transitivized PredicatesDiscussion468101616192021Free FormsBound FormsMorpho-Syntactic DiagnosticsMiddleAutonomousInchoativeImmediateStativeOut-of-ControlDesiderativeCausativeDiscussion2323242730303234343839414244465054Chapter Two Intransitive Forms of N4-e’?képmx RootsIntroductionUnaccusative and Unergative IntransitivesSemantic Features of N+e’?képmx Rootsand the Notion of ControlPreliminary Semantic Classificationof Intransitive Roots— in —Chapter Three Transitive Forms of RootsIntroductionA Preliminary Semantic Grouping for aTransitive/Intransitive DistinctionMorpho-Syntactic DiagnosticsDirectiveIndirectiveCausativeRelationalDiscussion626263656570747476Chapter FourBibliographyConclusions and Future Work 8185Free Form RootsControl RootsAppendix 1Appendix 28992- iv -LIST OF TABLES1: Personal Pronominal Markers2: Template for Simple Intransitive Predicates3: Template for Transitive Predicates4: Template for Complex Intransitives5: Template for Re-transitivized Forms6: Preliminary Semantic Classification of Free Roots7: Preliminary Semantic Classification of Bound Roots8: Morpho-Syntactic Diagnostics ofUnaccusativity/lJnergativity .9: Semantic Classification of Roots10: Preliminary Semantic Classification ofTransitive/Intransitive: Free Roots11: Morpho-syntactic Behavior of Unergative Roots12: Morpho-syntactic Behavior of Unaccusative RootsTable 13: Morpho-syntactic Behavior of Miscellaneous RootsTableTableTableTableTableTableTableTableTableTableTableTable91619203133565964787980-vACKNOWLEDGEMENTSI would like to express my heartfelt appreciation to my language consultantDorothy Ursaki for her time, expertise and friendship.I wish to thank my thesis advisor M. Dale Kinkade, from whom I have learned agreat deal over the years; and my committee member Henry Davis whose input,encouragement and perserverence are much appreciated.During the time it took to complete this work I have had the pleasure of workingfor and learning from Ewa Czaykowska-Higgins. I have benefited a great deal from herknowledge of Columbian Salish and from her support.I wish also to acknowledge an exceptional group of family, friends and caregivers,without whom this thesis could not have been completed. Most of all, I thank myhusband Don and my daughter Hana who are my joy and inspiration.Last but not least, thanks to that Bear of Very Little Brain, who articulated thelot of all graduate students; that when “you Think of Things, you find sometimes that aThing which seemed very Thingish inside you is quite different when it gets out in theopen and has other people looking at it.” (A.A. Mime).-vi-DEDICATIONFor Dorothy Ursaki,Laurence C. Thompson and M. Terry Thompson,and the brown-eyed girlsHana and Amy- vii -INTRODUCTIONN4-e?képmx (also known as Thompson River Salish, or Thompson in the linguisticliterature) is a Northern Interior Salish Language spoken in an area of British Columbiabounded by the communities of Spuzzum, Lytton, Merritt and Kamloops.Comprehensive documentation and initial theoretical work has been done onMe’?képmx by Laurence C. and M. Terry Thompson working primarily in conjunctionwith the late Annie York of Spuzzum (Thompson and Thompson 1980, Thompson 1985,Thompson and Thompson 1992). Other published work includes an analysis ofnarratives and other work on discourse by Steven M. Egesdal (Egesdal 1992), acomparative study of subordination in Salish that includes N4-e7képmx data, and otherwork by Paul D. Kroeber (Kroeber 1991). Continuing work on prosodic phenomena isbeing done by Mandy Jimmie (Jimmie, to appear).At the local level, the language is being taught in an adult Basic Education classin Lytton by Mamie Henry (Mestanta Technical Institute), and in Merritt by MandyJimmie (Secwépcmc Centre). A book of stories and biographical information has beencompiled by Darwin Hanna (Hanna to appear).This work is based on data from a number of sources produced by Thompsonand Thompson: a grammar ‘The Thompson Language’ (1992), drafts of a ThompsonEnglish dictionary (1979 and 1990) and an English-Thompson list (1987), and a text“Push-Back-Sides-of-His-Hair” (1990). These materials were supplemented byconsultation with Dorothy Ursaki (originally of Spences Bridge and now of NewWestminister- henceforth DU). The database under discussion is therefore characteristic-1-largely of the Spuzzum and Spences Bridge dialects. This work has benefitted fromadditional discussions with Mandy Jimmie and Nora Jimmie of Merritt (Field Methods,University of British Columbia (1987-88)), and Beatrice Hanna of Langley (originally ofLytton).The focus of this thesis is a discussion of the semantic basis of transitivityalternations in N4-e?képmx. Intransitive and transitive use of roots in N4’e?képmx ismorphologically marked; most intransitives use a set of primary affixes, and virtually alltransitives use a set of transitive extensions in combination with a transitive morpheme.Roots do not take all possible intransitive and transitive derivational affixation. Thepurpose of this work is to document which derivational affixes are allowed with a set ofpredicates to determine what is optional, what is obligatory, and what is blocked. It willthen be determined whether this can be linked in a principled way to an analysis of theargument structure of predicates. This will lead to a classification of predicate types.In Chapter One, I present an overview of the intransitive and transitivemorphology of N+e?képmx. Chapter Two contains a discussion of current literatureregarding syntactic and semantic diagnostics of unaccusative and unergative verbs. Iapply the semantic diagnostics to a set of N4-e?képmx roots, and discuss the behavior ofroots with a set of morpho-syntactic diagnostics to determine if they can elucidateunderlying structure. N+e?képmx signals changes in event structure by addingmorphology. I assume that argument structure can be derived from an interactionbetween the Thematic Hierarchy and an Aspectual/Causal Hierarchy. The thematicstructure of a root is invariant, but there can be multiple event-structure possibilities for-2-a given thematic structure. The argument structure determines which argument fills deepstructure object and deep structure subject. In Chapter Three I discuss the potential ofan intransitive/transitive classification of roots in N+e?képmx. Chapter Four containsconclusions and recommendations for future work.My primary set of data is the subset of roots in N+e’?képmx that can appear asfree form intransitives in a main sentence. This group of roots cross-cuts the broadsemantic categories found in the literature. I document the morpho-syntactic processesavailable to this subset of roots and compare the behavior of a set of bound roots.-3-CHAPTER ONE1.0 Overview of N{e’?képmx MorphologyN+e?képmx’ (henceforth N+) words are created using a lexicon of roots, lexicalsuffixes and a finite set of derivational affixes and pronominal affixes.Thompson and Thompson (1990, 1992)2 have documented aproximately 2000 ofthe roots in Nl. Roots are the content morphemes of the language expressing concepts,processes, events, and states. They are comparable to English nouns and verbs. However,it is important to note that in Salish languages there is no clear consensus of a noun-verbdistinction corresponding to the distinction found in English (Kuipers 1968, Kinkade1983, Demers and Jelinek 1984, Hess and van Eijk 1985, Thompson and Thompson1. N/+e?kép=mx is the native term for the language in use in the Lytton-Spences Bridge area (DarwinHanna pc). Thompson and Thompson (1992) note that there is no single term that encompasses the entirespeech community, “although ni-e?képmx is sometimes extended for that purpose”. The term “designates thepeople of the central part of the territory” (Thompson and Thompson 1992:1). The prefix n- is a locative,=mx is a lexical suffix meaning ‘people’. The stem is “etymologically obscure” (Thompson & Thompson1992:1)2. Because much of the data I am using is from these existing sources, for the sake of uniformity I use thephonemic transcription, parsing strategies and set of morphological markings laid out in Thompson andThompson (1992). Language materials in use at band schools in Lytton and a forthcoming collection ofstories use a phonemically based spelling system created by Randy Bouchard. Thompson and Thompson(1992:197-198) contains a table comparing the two spelling systems.For reference, the core set of the Thompson’s morphological markings is reproduced here.In a complex form in N4-e?kepmx,Marker denotes a(n)[...] infixunderlying root/ surface stem of a root- non-reduplicating suffixreduplicating suffix= lexical suffix (lexical forms incorporated into the stein/base)A double set of markers (-= or --) signifies secondary derivation.Thompson and Thompson (1990, 1992) use [...J in the gloss to mark explanatory material such ascontext, and (...) to denote optional material, usually an implied object.-4-1992). A more neutral term for the discussion of lexical forms in N+ is predicate.All roots3 can be used to create predicates. The majority of roots expressactions, events or states. With the exception of the small set of seventy-four free formroots4 (see Appendix 1) which have a main intransitive form without affixation (butallow stative and out of control affixes), roots require primary affixation to create simpleintransitives. All roots combine with the set of derivational transitivizers to createtransitive predicates.5 Reciprocal and reflexive forms are complex intransitives that are3. Throughout this work I will use /CVC- to denote a bound root. Free forms and surface derived formswill have no marking on them.The following Thompson and Thompson (1992) set of abbreviations for morpheme glosses is used:MDL middle suffix RLT relationalINC inchoative ST stativeAUT autonomous LCL locativeDRV directive NOM nominalizerCAU causativeWhere the underlying form of transitivizers and pronominal affixes is masked by phonological processes andthird person is a zero morph (Section 1.2) the surface breakdown of forms does not always conform to amorpheme by morpheme gloss. Rather than use zero morphemes in the transcription I employ thecombinations 3\3, 3\lsg. lsg\2p1 etc. to designate what pronominal object and subject are.4. There is a set of forms that I consider more nominal, which brings the number of free forms up toaproximately one hundred of the two thousand items documented in the database. I will deal with a subsetof free forms. This set is comprised of the forms that were known to Dorothy Ursaki and accepted as freeform intransitives.5. There are phonological processes that mask the underlying forms of affixes. This is especially relevantin the case of transitive predicates which contain concatenations of the transitivizers (directive -n-t-,indirective -xi-t-, causative -s-t-, and relational -min-t-) and pronominal affixes.In these combinations, the loss of vowels in unstressed syllables leads to the loss of the transitive morpheme/-t! or to the simplification of resultant clusters, and to effects on the I-ni of the directive and relationalsuffixes.For the purposes of this discussion it is important to note that in general,t --> 0 n,n’,f _n,s,x”ts --> ccs --> ctt -->and continued-5-formed on transitive stems.There are roots that have limited productivity. For example, forms such as qwzem‘moss’ and tirix ‘sinew’ allow only a minimal number of derived (nominal) forms. Thereis no consistent morphological means of distinguishing between nominals and predicatein NI.6 These more nominal forms will not be discussed in detail in this work.1.1 The Pronominal SystemSalish languages can be described as pronominal argument languages (Jelinek1984, 1985). Predicates are complex forms with obligatory affixal or clitic marking ofverbal arguments. Nominals corresponding to the verbal argument suffixes are optional.Adjoining nominals can be marked as direct, oblique or locative.8 Although there ismuch variation, preferred word order is VOS (Kinkade 1990: 341, Matthewson 1993: 2-3).In NI, intransitive forms mark a single argument, although often a second isimplied. A nominal corresponding to subject is marked with the direct marker Ie/ or /1/,an implied object (unmarked on the predicate) is marked with the oblique marker /te/n --> 0 m,m’_ s,txn --> e before a homorganic obstruent--> ne(Thompson & Thompson 1992:35-43)See Thompson & Thompson (1992) for a more complete overview of phonological processes.6. One means of distinguishing nominals from predicates is that all nominals require a middle affix to forman active predicate, whereas not all active and stative roots do.7. See Mattina (1993) for another point of view.8. Locatives will not be discussed in this work.-6-or /Ic’e/, as in 1-2:1. qWc.c e tmIxWfmoveOC 3 DIR earthIt is an earthquake2. k’á+-t kn t c’Ik’fsticky-IM lsg OBL pitchI got stuck up with pitchTransitives mark two arguments, which surface as syntactic subject and object. Both aremarked with a direct marker, as can be seen in examples 3 and 4.3. q’áy’-e-s e slnük”e?-sshoot-DRV-O-3\3 DIR NOM!friend-3psvhe shot his friend4. n/kn’-cém-s e qWu?LOC-fpoison-DRV\lob-3 DIR waterthe water poisoned meDitransitive predicates are associated with two objects, one of which is not overtlymarked on the predicate. Direct marks what in English glosses would correspond to theindirect object, as in examples 5 through 7:5. q”ec’-ne t s/+a?x=ánsffi11-3\1sg OBL NOM-/eatI filled him up with food6. u?ex pték”’+-xo-ne t s/pték+-sAUX ..fstory-IND-3\lsg OBL NOM/story-3psv.I am telling it for him, his story (347c)7. má-x-t-x” e smü+ec t s-/zélt-s%fsmash-IND-2sg. DIR woman OBL NOM-/ smashed the woman’s dishThis is a characteristic of all Salish languages (Mattina 1993).The N4- system of pronominal affixation can express a maximum of two argumentson the predicate. Intransitives in simple (main) clauses are marked by the use ofindicative enclitics to mark person and number of the single argument (Example 2).-7-Transitives and ditransitives utilize sets of distinct subject and object suffixes in VOSorder (Examples 3 through 7). There is a separate set of suffixes that mark indefinitesubjects, in what have been termed ‘passive’ or ‘indefinite’ forms, used in conjunctionwith the regular set of object affixes. There is also a set of genitive/possessive pronominalmarkers used with nominals and to express arguments of predicates in nominalizedsubordinate clauses. The pronominal paradigms relevant to the discussion aresummarized in Table 1 below.Table 1: Personal Pronominal Markers(Thompson 1985:407)1.2 Simple Intransitive PredicatesThe majority of roots in NI- are bound forms (Examples 1-4, 5, and 7) that neversurface in unsuffixed form. Only a minimal set of the 1800-2000 roots documented byIntransitive Transitive SuffixesClitics(Indicative) Object Subject Indefinite SubjectSg.1. kn -sém -(é)ne/-n -séy-me-sey! -Si -si-me2. Ic” -s(I) (e)xwP1.1. kt -éy! -i -em -éy-t-(e)t -i-t2. kp -Cym/-im -(e)p -Cym-et,-im-etGen3. 0 0 -(e)s, -0-(é)m-es-8-Thompson and Thompson (1990, 1992) are free form intransitives.The basic template for intransitive bases and primary/secondary affixation9in N4is presented in the following table, along with relevant examples. Material in parenthesesis optional.Table 2: Template for Simple Intransitive Predicates(PREFIX)-[fROOT( REDUP.)= (LEXTCALSUFFIX)]stem(1/2 AFFIXATION)?es- VC-op/[-?-]’° -mémnstative out-of-control inchoative (INC) desiderative(ST) (OC) -t (DESID)immediate (TM)-yxautonomous (AUT)-omemiddle (MDL)Examples of Simple Intransitive PredicatesSTATIVE8. ?es-péw-tSTsJwell.up-IMswollenOUT OF CONTROL9. ?e qwaz.zbANT %fsweatOC 1sI might sweat (342C)hlINCHOATIVE10. n-pfshiver-INCshiver, feel cold, chilly9. I discuss only a subset of affixes that create intransitive forms in Ni-. These are the most relevant to thediscussion because their use is regular and they are common in the data. There are a number of otheraspectual affixes but examples of their use is limited, although they are productive. See Thompson andThompson (1992) for a complete overview.10. The alternates for Inchoative distinguish between strong (stressed) roots I-?-! and weak (unstressed)roots I-op/. Because glottalization in Ni- can come from a number of sources, my discussion of inchoativewill focus primarily on forms created from weak roots with the suffix /-op/, although see Section 2.53 forsome examples with I-?-!.11. Examples that have numbers in parentheses are elicitation forms from DU. The number refers to theirlisting in my field notes. Examples from other consultants have initials. ‘Sf stands for suggested form.-9-IMMEDIATE11. pdw-tJswell.up-IMswell up, swollenAUTONOMOUS12. pdw-ix%fswell.up-AUT[of snake, toad] puff up, swell upMIDDLE13. ?á?e-me..Jlullaby-MDLsing lullabyFREE FORM14. xVesit knfwalk lsg.clI walked (221)DESIDERATIVE15. x”’esit-mémn icuIwalk-DESID lsgI want to walk1.2.1 Primary AffixesThis section deals with the most productive and common subset of the aspectualand other affixes that create intransitive forms in NI. These are primary affixes, becausethey affix directly to the root or to stems (created by roots and lexical suffixes)’2 tocreate intransitives.Many bound roots potentially take all primary affixes and transitivizers. Some mayhave a second set of transitive derivations based on a single primary affix, as illustratedbelow:16.Intransitive form16.a. k+-pfseparate-INC 3it came apart12. I will not discuss the nature of lexical suffixes in this work. See Thompson and Thompson (1992) foran overview of these forms.- 10 -Transitive forms16.b. k+-st-éss/separate-CAU-3\3someone separates things16.c. k+-p-st-éssfseparate-INC-CAU-3\3someone manages to detach somethingThere is a small set of roots, as in 17, that allow only one primary affix; all otherderivation is based on this base rather than on the root.17.Intransitive forms17.a. ?eI?üy-m’AFFMaugh-MDLlaugh, smile17.b. ?eI?uy-m-ü4’AFF/..Jlaugh-MDL-habitualalways laughingTransitive form17.c. ?eI?üy-m-s-cAFF/sflaugh-MDL-CAU-3/3make someone laughThompson and Thompson (1992) divides intransitive affixes into two groupsreflecting aspectual function and ‘voice’ (defined as “specifying varying relationshipsbetween states and actions and their subjects (1992:99)). These can be regarded asdistinguishing affixes that mark agentivity. They classify primary affixes as plus, minus,or unmarked for the feature of agent control depending upon whether they affect theagency of a root’s arguments (See Chapter Two for further discussion).Aspectual affixes are discussed as referring to static (henceforth stative) ordynamic events. Stative aspects have an emphasis on a resultant state rather than on achange of state.- 11 -Immediate I-ti, as shown in 18 through 20, designates states and actions that havejust gone into effect, and impending states or action. It can also designate generalcharacteristics of things.18. a. b. c.7lq’-t ?Iq’ Jscrape-IM sfscrape-MDLscraped off to scrape hair off deerhide[hair of buckskin]19. .[nIk’- nIk’-t nIk’-mJut Jut-IM Jèut-MDLget cut cut s.t.20. ?ix- ?Ix-tfscratch/stripe fcratch-IM fscratch -MDLbe scratched! make a stroke!make scratch noise stripeStative I?es-I designates completed actions and states where the eventprecipitating the change of state is more remote, as shown in 21 through 23. Statives arealso used for descriptives, including permanent qualities of objects and for statingpossession.21. a. b. c...fpuys- ?es!püys PÜYStXW.fkilI ST-Jkill 3 Jkill-TR?-3!2aghe’s been killed you kill him22. Jk+- ?es!k+%fseparate ST-..fseparatedetached23. fk’ük” ?es/k’ük”Jburst ST—Jburstalready burstMany free form roots can also be found with alternate intransitive forms created withstative /‘?es-/.- 12 -24. a. b.?uiq”e? ?es/?uq”e?sfdrink 3 ST-fdrink 3he is drinking he is drunk25. a. b. c.?écq” kn wi? ?ecqvt ?es-I?ecq”sJbake lsg EMP Jbake-IM ST-..JbakeI baked (just) baked (already) baked (410a1)It is possible to get combinations of the stative aspects.26. a. b.?es/?Iyx”-t%fpit.steam ST-sfpit.steam-IMit is already steamedInchoative I-?-! or /-p/ is one of the dynamic aspects. The two allomorphs areconditioned by root strength; I-?-! is infixed into strong roots (underlying stress) and !-p/is suffixed to weak (no underlying stress) roots. This affix marks a developing orchanging event that occurs without the intervention of an agent. Noises, naturalphenomena, and some involuntary bodily processes are often marked by inchoative.27. a. b. c.Jld-- k+-p ?es!k+Jeparate ..fseparate-INC 3 ST-Jeparateit comes apart apart, off, detached28. a. b. c...[péw- p[?]ew péw-ixswel1 /swel1-INC swe11-AUTbegin to swell [of toad] swell, puff upFree form intransitives are not found with inchoative alternates.Middle /-Gme/ marks states and activities in which there is an agent subject.Middle forms tend to be continuing or progressive. It is the most common primary affix,and is used to create an intransitive from what would be considered nominal forms (seeAppendix 1). With many middle intransitives there appears to be an implied non-specific- 13 -object. In other cases there is a reflexive reading. Many characteristic functions andactivities of animates and objects are middle intransitives. Some examples of derivationswith the middle affix are:29. fk+=ékst-m[separate=hand-MDL 3he lets go (of something)30. Jc’k-mIpush-MDL 3he pushes/propels himself31. sfçnom.fring-MDL 3[bellJ rings, [clock] strikesFree form roots (with minor exceptions) do not allow middle affixation.Autonomous /-vx/ references actions controlled by a specific agent (usually+ human). It also marks lexical items related to posture and movement, and to someinanimate states (dynamic).32. a. b. c.[qWec q”c-iyx q”cc e tmIx”..fmove move-AUT 3 Jmove-OC DIR earthmove about, earthquakeset out [person]33. a. b.cáq’”-ixfied/brown .fred/brown-AUT 3[berries] turn red, ripenFree form roots do not take autonomous affixation.Out of Control /-VC-/ indicates that an event or a state has developed without anobvious cause, spontaneously or with the intervention of an agent that is not ‘in-focus’in the discourse. It can also have an ‘exceptional volition’ reading, where the act hastaken persistence or requires a special skill on the part of the protagonist. This- 14 -reduplicative affix is quite productive in Salish languages (Carison and Thompson 1982).In N4 it adds to both agentive and patient-oriented roots. When it is added to stronglyagentive roots, it usually results in an intransitive with an ‘exceptional volition’ agentreading. Thompson and Thompson (1992) note the significance of out-of-control creatinga non-volitional patient form with some agentive roots (1992:57). I will return to thispoint in Chapter Two. Some forms that exemplify the split are:34. 4a?x=án[n]s knfeat[0C] lsg (+ctl)I manage to eat (agent oriented out-of-control form)35. téwukn/ lsg (+ctl)I get sold (something), someone sells me something (patient oriented)The out of control affix is quite productive and is used with both bound and free formroots.A single secondary affix, desiderative /-mémn/ is presented in this work. This suffixattaches directly to roots, is not found in combination with the primary affixes, and isnever used as a base for lexical suffixation. Therefore I designate this a secondary affix.As can be seen in the examples below, desiderative marking primarily creates intransitiveforms with the semantics of ‘want X’.36. xVesitmémn kn I want to walk (235sf)Iwalk-DESID lsg37. nes-mémn lcn I want to gofgo-DESID lsg38. naq”1-memn kn I want to steal (265 sf)Jsteal-DESID lsgThese forms can be transitivized by adding the transitive I-ti or by adding causative orindirective suffixes. There are no data on relational use (see Section 2.5.7).- 15 -1.3 Transitive PredicatesTransitive predicates are created by the addition of an extension (-n- directive ,-scausative, -mm- relational, or -xi- indirective) in sequence with the transitive affix (-t- TR)to a stem.’3 In this work these combinations are referred to as transitivizers and can beconsidered to be unitary forms (i.e. (-n-t-) = directive). Some combinations of extensionsare possible; these are min-xit, min-st, xi-st. The two argument positions are marked bypronominal suffixes following the transitivizer. The transitivizers are listed below inTable Transitivizing SuffixesTable 3: Template for Transitive Predicates[BASE]-(EXTENSIONS)-TRANSITIVE-OBJECT-SUBJECT’4-n -t- DRy15-s -t- CAU-xi -t- IND-mm -t- RLT-mm-s -t- RLT CAU-xi-s -t- IND CAU-min-xi -t- RLT INDThe formation of a directive transitive (DRV) with /-n-t-/ denotes a transitive witha subject (Agent) and a direct object (Patient/Theme). These forms usually indicateintentional punctual actions. This transitivizer is barred from attaching to some roots (forexample, roots like q?ém ‘[of baby] nurse’). With other roots it forces a change insemantics (for example, zoq” ‘die’, analyzed as a patient-oriented root by Thompson and13. There is a closed set of roots that may occur with just the transitivizer /-t-/. However, reflexive andreciprocal forms based on directive transitives of these roots often have the full extension !-n-t-I.14. See Table 1.15. Abbreviations will be defined below.- 16 -Thompson (1992), becomes zoqves ‘kill something’. Some examples of roots withdirective are:39. ?üq’’e?-ne I drank it (MJ24b)’6fdrink-DR-3/lsg40. kWises drop something, let fall intentionally, throw down%ffall.drop-DRV-3/341. c’q”’-t-es he writes it [a letter]..fwrite-TR-3/3Causative transitives (CAU) created with /-s-t-/ are the most common of thetransitive predicates. In general, in the case of roots with a clear actor in thecorresponding intransitive form, the causative transitive reads that there is someoneaffecting that actor. In forms where there is a non-agent in the corresponding intransitiveform, the causative has an accidental causation reading, as can be seen in the formsbelow. Some forms allow both readings, as can be seen in examples 44a and 44b.42. ?üq”e?-s-t-p you people give him something to drink (cause him to%[drink-CAU-O-3/2p1 drink), give/serve drinks to someone43. nesstx’” you were able to take him along (MJ26a)Igo-CAU-3/2sg44.a. k’’i(s)-s-c drop something accidentally..ffall.drop-CAU-3/344.b. k”ls-s-cm-s she caused me to fallJfall.drop-CAU- lobj-3/3The indirective transitivizer (IND) of the form /-xi-t-/ has also been referred toin comparative Salish literature as a benefactive or applicative (Mattina 1993, Carison1980, Kinkade 1980, Thompson and Thompson 1980), and can be compared to the16. MJ = Mandy Jimmie.- 17 -English preposition ‘for’. It marks ditransitive predicates where the object pronoun refersto an indirect/oblique object and there is an unrealized implied object.45. ?üq”e?-x-c drink a beverage belonging to someone, somebody drank someIdrink-IND-3/3 drink of yours (MJ24e)46. kWisxne I dropped it to him (intentionally)fdrop.fall-IND-3/147. c’q”’-xi-t-es he writes (a letter] to/for someonefwrite-IND-3/3Forms created with the relational transitivizer (REL) /-min-t-/ have semantics thatare very hard to pin down. Thompson and Thompson describe it as refering to ‘objectstoward which the subject is moving, or in relation to which the action is accomplished’(1992:73). It provides transitives for many stems that cannot otherwise be transitivized.The meaning of predicates created with /-min-t/ are often abstract.48. cq’-mmn-s he ran into him (NJ531)Ihit/throw-REL-3/349. ce?x”’-mIn-s be happy to see someone, appreciate! be grateful for somethingfhappy-REL-3/3 (p36D)Some combinations of transitivizers are possible. Causative occurs with indirectiveand relational. Relational and indirective can combine as ‘min-xit’ but not as ‘xit-min’.Combinations with directive are not allowed.50. nés-min-s-t-x” you were able to bring it there (MJ26b)fgo-REL-CAU-TR-3/2sg51. n6s-xi-t-s-t-x” you brought it there for him (MJ26d)’7fgo-IND-CAU-3/2sg52. ce?x”-mI-x-c congratulate someone (about something), appreciate! befhappy-RLT-IND-TR-3!3 grateful for something17. This form is problematic because it contains two instances of the transitivizer t and no reduction of thevowel in -xi-.- 18 -1.4 Complex Intransitives: Reflexives and ReciprocalsReciprocals and reflexives in N+ have complex internal structure. They are lexicalforms, as they are based on transitive affixes, but take the intransitive clitics to mark the(linked) single argument. The template for the structure of complex intransitives is givenbelow in Table 4 with accompanying examples.Table 4: Template for Complex Intransitives[BASE]-(EXTENSION(S))-TRANS.-RECIP./REFL. CLITIC_waxw/_sut-mm-xi- -t-min-s-tExamples of Reciprocal forms:’853. be jealous of each otherjjealous-DRV-TR-RECIP 354. OkStWáXW know each other..fknow-CAU-TR-RECIP 355. zu-min-t-wáx” help each other (NJ21)’9ftake care of-RLT-TR-RECIP 356. xc-xit-wáw’x” kt we bet for one another, placedJbet-IND-RECIP ipi bets for one another (14a)57. ?es-zu-min-s-t-wáx” take care of/guide one another.ftake care of-RLT-CAU-TR-RECIP 358. témn kas +uk”’-min-s-t-wáw’x” kp don’t you remember each other?./hook-RLT-CAU-TR-RECIP 2pl (18d, MJ15)18. All my examples are of reciprocals for ease of comparison. Reflexive forms are very common. SeeThompson & Thompson (1992).19. NJ = Nora Jimmie.- 19 -1.4.1 Re-transitivized PredicatesReflexive and reciprocal intransitives can serve as the base for the addition of therelational and causative transitivizers. As the examples below show, causative is the mostcommonly used re-transitivization. There do not appear to be any re-transitivizationsbased on causative reflexives or reciprocals in the data base, nor could this combinationbe elicited. Below in Table 5 is the template of the structure of re-transitivized formswith accompanying examples.Table 5: Template for Re-transitivized Forms[BASE]-EXTENSION-TR.-RECIP. (IREF).-EXTENSION-TRANS.-OBJECT-SUBJECT-n -t _waxw -s -t-mm -t wáx” -s -t-xi -t wáxw -s -t-n -t wáx -m -tExamples of Re-transitivized Reciprocals:59. )c-e-t-wáx’’-s-t-x” you bet with him.,Jbet-DRV-TR-RECIP-CAU-TR-3/2sg60.a to fight each other (19a)%[fight-DRV-TR-RECIP 360.b k’q”’-e-t-wáw’x”-s-cm-s he made us fight each other, it was his ideaJfight-DRV-TR-RECIP-CAU-TR-1sg/3 that I fought this person (19b)61. nk’-n-t-wáw’x’’-s-cm-s that person caused me to change something,fexchange-DRV-TR-RECIP-CAU-TR-1sgJ3 with him (MJ22)62. ?+k”-min-t-wá’x’’-s-cm-s someone caused us to remember each,Jhook-RLT-TR-RECIP-CAU-TR- lsg/3 other (MJ16)63. he made us bet against each otherJbet-IND-TR-RECIP-CAU-TR-lpl/3 (14d)64. k’q”-e-t-wáw’x”-m-t-lyxs two people fighting over one person%ffight-DRV-TR-RECIP-RLT-TR-3/3pLemph (19c)- 20 -65. ?nk’-n-t-wáw’x”-min-s someone was able to trade/exchange afexchange-DRV-TR-RECIP-RLT-(TR)?-3/3 particular thing (MJ23)There was one questionable form elicited that re-transitivized with the indirective,but this form also requires a causative. It was problematic in a number of ways: in termsof stress, in the vowel in the indirective not being reduced, and because there was nocoalescence of It-si. I could not duplicate this form in elicitation with my primaryconsultant.66. nak’-n-t-wáw’x”’-xit-s-cm-s he caused me to change something for (MJ22a)Re-transitivization is a productive process, but not a commonly used one.Consultants prefer to give nominalized bi-clausal forms of the type:67. ?exe john p+u an-s/náq” my stealing was John’s work (267a)fwork PSV-NOM-tea1The pronominal arguments marked on the re-transitivized forms often differ from whatis expected. For example, in 60.b and 62, one might expect the object to be plural. Withrespect to the semantics of retransitivized forms, semantic judgements by consultants arenot as clear as for plain reflexives and reciprocals. Note also that in examples 56 versus63, reference of the indirective changes from benefactive to malefactive.1.5 DiscussionThis overview of the morphology of NI- predicates raises a number of points. Inthe creation of intransitives there is a distinction between bound roots that requireprimary affixation and roots that surface as free form intransitives. The latter (with minorexceptions) do not take the majority of primary affixes (immediate, autonomous, middle,and inchoative) but specific roots do allow the stative /?es-/ and out of control affixation.- 21 -These points will be discussed in Chapter Two.Looking at the productivity of transitivizers it can be seen that causative andrelational are different in nature from directive and indirective. Thompson andThompson (1992) records causative as the most commonly used transitivizer. Intransitivestems (root plus primary affix) are usually transitivized with causative or relational;directive transitivization of these forms is exceptionally rare, indirective transitivizationsare only slightly less so. The data from re-transitivized forms in the examples above alsoshow causative to be generally productive. Relational has a different nature from thatof causative. It is not used as regularly, although it has a greater range than directive andindirective. These points will be addressed in Chapter Three.- 22 -CHAPTER TWO2.0 Intransitive Forms of N+e’?képmx Roots2.1 IntroductionIn theory, the difference in productivity of the intransitive affixes in N4 can belinked to differences in underlying argument structure of lexical items. Indeed, there hasbeen a great deal of cross-linguistic work (Perlmutter 1978, Perimutter and Rosen 1984,Grimshaw 1987, Levin and Rappaport 1989) documenting and discussing what is termedthe split-intransitive hypothesis, a syntactic distinction between two classes of intransitiveforms. Originally articulated as the Unaccusative Hypothesis (Perlmutter 1978) the splitdistinguishes between unaccusative and unergative classes of lexical items. This distinctionis stated in terms of argument structure, as an unergative has a single agent argument,and an unaccusative a single patient argument. This translates into a configurationaldifference (unergatives have a subject generated at deep structure, unaccusatives havea deep structure direct object) that has ramifications for the syntactic behavior of theseforms at the clause and sentence level. There are language-specific syntactic processesserving as diagnostics to distinguish between the two classes of intransitives.It has been noted across languages that lexical items that are recognized as unaccusativeand unergative tend to separate into regular semantic classes.In this chapter I discuss the literature on semantic and syntactic diagnostics forunaccusativity with particular attention to Gerdts’ (1991) discussion of unaccusativity inHalkomelem Salish. I apply the semantic diagnostics to the set of NI- free forms todistinguish two general classes of roots that are potentially unergative and unaccusative.- 23 -I discuss the intransitive and transitive behavior of these roots to determine thediagnostic potential of the transitive affixes and to determine if semantic classes matchwith morphological marking.2.2 Unaccusative and Unergative IntransitivesSemantic diagnostics of unaccusative versus unergative predicates distinguisheither agentivity (for example, actor volitionality) or inherent lexical aspect (for example,the nature of and limits on the event specified by the predicate). Languages can varywith respect to which parameter governs the split. Across languages, unergativesgenerally refer to lexical items of volitional action, manner of motion with protagonistcontrol (for example ‘run’), manner of speaking and involuntary bodily processes, oratelic activities (events with no clear endpoint). The semantic classes that are associatedwith unaccusative are non-volitional actions, states (existing or happening), motion verbswith no inherent protagonist control (for example ‘roll’), and psychological predicates.Telic activities (events with defineable endpoints) such as verbs of motion with inherentlyspecified direction (for example, ‘arrive’) and predicates of duration also tend to beunaccusative. (Grimshaw 1987, Levin and Rappaport 1989, Grimshaw 1990, Gerdts1991:230)).Syntactic diagnostics of unaccusativity are processes that distinguish betweenunaccusatives and unergatives on the basis of their configuration. Passive is a diagnosticin some languages, as it may apply to unergatives but not to unaccusatives. Othersyntactic diagnostics, such as auxiliary selection in Romance languages and there-insertionin English, are considered to mark unaccusativity.- 24 -There is however, evidence that the syntactic-semantic link is not a simple one.There is the potential for unaccusative mismatches. A syntactic mismatch means thatsyntactic diagnostics are actually sensitive to a specific semantic feature rather than anargument configuration based on the correspondence between a set of features. Levinand Rappaport (1989) refer to Zaenen’s (1988) discussion of Dutch, revealing thatprenominal perfect participles in Dutch modify telic intransitives rather than allunaccusatives as a class. Zaenen also shows that impersonal passivization in Dutchfunctions to distinguish verbs with protagonist control rather than the class of unergatives.This type of result has led some researchers to consider that unaccusativity is not aunified phenomenon and therefore there is no need for a syntactic distinction betweenintransitives.Semantic mismatches have been discussed by Rosen (1984), who notes that acrossand within languages lexical items of the same semantic class can vary with respect totheir unaccusative and unergative properties. For example, she finds that the verbtranslated as ‘die’ in Italian is unaccusative, but in Choctaw it is unergative. Within theclass of bodily process roots in Italian, some exhibit unergative behavior and othersunaccusative behavior.These conflicting data show that syntactic configuration cannot be universallypredicted on the basis of general semantic class, and that syntactic diagnostics must becarefully chosen to ensure that they indicate the entire class of unaccusatives rather thana single semantic feature. Yet, I follow Gerdts (1991) who claims that ‘lexical semanticscan serve to suggest the class of a verb, with syntactic diagnostics verifying its assignment- 25 -to a particular class’.Gerdts discusses Halkomelem (Coast Salish) (henceforth Hk) data with respectto two syntactic tests, desiderative and causative, that show there is some basis forproposing that the unergative/unaccusative distinction is relevant to the grammar of thelanguage. Gerdts considers the morphological processes of causativization to be anindicator of Causative Clause Union, a highly constrained rule in Hk requiring thedownstairs clause to be unergative. According to her analysis, desiderative morphologymarks multi-predicate clauses, where the suffix marks the addition of a predicate thatinherits the argument structure from the main predicate. This rule requires that the‘cognizer’ referent of the desiderative is the subject of the main predicate. Her resultsshow that unergatives allow both causative and desiderative constructions, whileunaccusatives split into three subclasses: a group of process verbs that allow desiderative,with a shift in semantics to a ‘future’ meaning, a group of state verbs that allowcausatives, with a shift in semantics to a ‘resultative’ meaning, and a group of mixedprocess and state verbs that allow neither of the two constructions.In the second part of the paper Gerdts presents evidence from derived statives(addition of stative morphology) and derived processes (addition of inchoativemorphology). This evidence confirms her conclusions that causative and desiderativetogether provide a syntactic test for distinguishing unergatives and unaccusatives and thatevent structure plays a role in the classification of intransitive predicates in Hk.Gerdts cites Hukari (1976) as noting that ‘the overwhelming majority ofHalkomelem roots are unaccusative (1991:237). Thus her data are composed of roots- 26 -with primary affixation that appear as surface unergatives and unaccusatives.One of the purposes of this paper is to observe the workings of causative anddesiderative affixation in N4 to see if the processes are comparable to Hk. It isnoteworthy that in N+ the desiderative marker is a secondary affix that adds directly tothe root or to stems formed with lexical affixation. I do not consider desiderative aprimary marker, because although it is in complementary distribution with the primaryaffixes, it does not appear inside lexical suffixes, as they do. These forms are intransitiveand can be transitivized with the set of transitive affixes. In Hk, the desiderative affixforming intransitive forms is attached outside the transitive affix /-t-/.2.3 Semantic Features of N4e?képmx Roots and the Notion of ControlThompson and Thompson (1992) consider roots in N+ to be underlyinglyintransitive, with the majority being unaccusative or patient-oriented in nature. Ratherthan selecting a set of ‘unsystematically selected intransitive verbs that [potentially]involve suffixation’ (Gerdts 1991: 237) I have chosen as my database for this study thesubset of roots that can stand as free form intransitives. I chose this set on theassumption that primary affixes might signal a change in argument structure, andtherefore confuse the issue of root argument structure.This set of forms can be split into general classes according to the semanticdiagnostics cited above. This analysis can be motivated using evidence from theThompson and Thompson (1992) and Thompson (1985) analysis of the feature of controlin N+. Thompson and Thompson discuss ‘patient-oriented’ and ‘agent-oriented’ roots inN using the term control to characterize roots and some derivational affixes. The- 27 -distinction indicates whether the subject of a predicate is in volitional control of theaction expressed by the root/stem.’A brief summary of Thompson and Thompson (1992) and Thompson (1985)follows. The semantics of control correspond to a subject’s degree of control over anaction or an event. Out-of-control entails an action that is accidental, spontaneous,accomplished after much effort, or as a result of luck. Control marking means that thesubject is agent and acting with purpose or volition. Variations can specify that thesubject is acting with ‘self-interest’, or have other idiomatic meanings. All roots, lexicalsuffixes, and grammatical affixes are either control, non-control or unmarked. Pronominalsuffixes and clitics are never marked. This is therefore a lexical feature of roots andgrammatical suffixes. The NI predicate with its combination of control-marked roots andaffixes is marked control or non-control according to a dominance hierarchy whichspecifies the relative strength of the constituents. Intransitive and transitive predicatesdepending on the combined control status of their constituents, are either unmarked,marked for control, or marked out-of-control. The majority of roots in NI- are assumedto be non-control. Thompson and Thompson (1992) give a number of fifty control rootsin NI. However, perusal of all sources found thirty one bound and free forms (seeAppendix 2). The analysis of control distinguishes the workings of the agentivecomponent (but not the aspectual element) discussed in the literature on theunaccusative/unergative distinction.In NI intransitives there is no pronominal distinction made between lexical items1. This use of the term control is not the configurational control of Government & Binding Theory.- 28 -classified by Thompson and Thompson (1992) and Thompson (1985) as patient-orientedand agent-oriented, as shown below:PATIENT1. qlIlkn...%fangry lsg (patient)I was rather angry2. ...Iyq’ knffaint lsg (patient)...I fainted [yesterday\other context] (389a)3. wüxWtJnow 3 (non-agent)it is snowing (257a,370a)AGENT4. pték”+ kn u?ex.fnarrate lsg (agent) AUXI am telling a story (347)5. x”esit kts/walk ipl (agent)we walked/travelled6. skew knfyawn lsg (non-patient)I yawn (280a, 333)To a large extent, I believe the distinction between unergative and unaccusativeintransitive predicates can be associated with the distinction made between control andnon-control roots, as shown in examples 1 through 6. This gives partial support to theanalysis of an unaccusative/unergative distinction; what is needed is evidence that theaspectual part of the equation is also relevant in NI. I propose that the control featuresof roots and derivational affixes as documented in Thompson and Thompson (1992) andThompson (1985) should be linked to agentivity, and may be subsumed into a distinctionbetween unaccusative and unergative use of predicates. Although the two systems do notmatch precisely, for example involuntary bodily process roots are often non-control, and- 29 -there is a single weather root that is marked control, it will be seen that these particularroot classes do not pattern with the agentive and stative forms which make up the bulkof the data set.In the following sections I propose to use the semantic categories from thecrosslinguistic literature on unaccusatives and unergatives, making special reference toThompson and Thompson’s notion of control to classify a representative subset of theroots of N4-.2.4 Preliminary Semantic Classification of Intransitive RootsThe following tables present preliminary semantic groupings of free forms and aselected set of bound forms. The semantic criteria for this classification are 1) agentivityas determined by the Thompson and Thompson (1992) analysis of control, 2) obviousstates and the telic/atelic distinction. The distinction will be motivated in the followingsections by applying a variety of morpho-syntactic tests (Sections 2.5.1-2.5.8).Table 6 and table 7 group the free and bound forms in two sets under the majorheadings of Unergatives and Unaccusatives according to the semantic criteria givenabove, with sub-headings for subsets. These tables will be revised on the basis of theevidence from the morpho-syntactic tests and a final table will he presented in theDiscussion.2.4.1 Free FormsIn Table 6 I give an initial listing of potential unaccusative and unergative forms.- 30 -Table 6: Preliminary Semantic Classification of Free RootsA. UNERGATIVESAgentive Actions (atelic, + control)1 2 3 4 5 6x”esit ?üq”’e? nés náq”’ taxwalk drink go steal run.animal paddle7 8 9 10 11 12ml+t q?ém p’nt’ k”üce k”üme qáytvisit [infant] nurse return descend ascend reach,topto water to water13 14q’amIn knthrow helpAgentive Actions (atelic, no control marking cited)15 16 17 18 19 20fwóyt páq” twép m’on’ k’éy ?ecq”sleep watch back up give stop bake21 22 23kéze? pték”4tell a lie tell a story, get dressedbe a lie be a storytellerInvoluntary Bodily Processes (telic, -control, actor)24 25 26 27wék’k’ ? sxe shéw méc’xvomit sneeze yawn blink(-ctl) (-ctl)B. UNACCUSATIVESState or Change of State (telic)1 2 3 4 5 6zoqw k”is iyq’ pz’é+ y’e qnoxvdie fall faint worthless feel good sick7 8 9 10 11 12xáni +ai 4yük” máq’ xiyhurt cold bumped into satiated go ashore finished- 31 -13 14 15 16 17 184á ptük” k’y’áq2 Ic’áq’ c’dk”’wound.heals3 ooze break tumble.down shine bang/noise19 20 21 22C’ló’ pl’üx” c’k’ lpüx”hot weather! punctured exhausted have a hole throughfoodWeather (atelic, inanimate state)23 24 25 26wüx’”t x’”ák’’ tek4’ k’aüssnow frost rain hail(+ctl)Psychological Predicates (states)27 28 29paqwu? wyjxm qlIlafraid jealous angry2.4.2 Bound FormsFor comparative purposes, the set of bound form roots listed in Table 7 has beenchosen. This set can be broken down into unergatives and unaccusatives, but there is athird set of forms. These are unanalyzed at this point because they appear to have anactive reading. They are not however listed as control roots.2. The dictionary form of this root is k’m’áq.3. DU allows this root only with an inanimate referent. In Thompson & Thompson (1990) the root can referto a healer.- 32 -Table 7: Preliminary Semantic Classification of Bound RootsA.UNERGATIVEAgentive Action (bound control roots)1 2 3 4 5 6sf?üs- Jk”én- Ik’éx- Jpile- .fpiIysdiscard grasp joke give inform/tell kill7 8 9 10 11 12sfpuyt fql- fcü(n)- .f+Im- Jéwlie down bite cheer say cut brush sell to13pointBodily Processes14 15 16 17 18 19Jikéy- Jsüp’- Ic’ék- fx”Ic’- /qázeject/vomit urinate breathe itch defecate sweatB. UNACCUSATIVESState or Change of State (no control marking)1 2 3 4 5 6[pew- /q”ec- fk’ok’”swell move startle burst escape rest7 8 9 10 11 12Jiey- .fk’dmcrazy lazy thirsty hungry stealthy separate(unmarked for ctl)13 14curve brown/redPsychological Predicates15 16[c’éx- Jzdw’shy/shame annoy (-dl)- 33 -C. UNANALYZED FORMS1 2 3 4 5 6%fJf- Iciq- .fnIk’hide dig sweep put on top cut/slice scratch7 8 9Icéwextinguish sit (plural) reachIn the remainder of the Chapter I discuss the behavior of these roots with the setof primary affixes, the causative and the desiderative. The results from this analysis willcorroborate the semantic classification of bound and free roots. It will also show that theunanalyzed set (Table 7, Set C) which at first would appear to be active-agentive, in factpattern with the unaccusatives.2.5 Morpho-Syntactic DiagnosticsI begin with the primary affixes and their behavior with free roots and boundroots. The first of these is the middle suffix.2.5.1 MiddleIn Capter One, it was shown that middle marks events in which there is an agentsubject. It is the most common primary affix, and many middle intransitives appear toimply a non-specific object. In other cases there is a reflexive reading. Middle is markedfor control. Given this characterization, I would assume the middle affix to be markingagency. Free forms differ primarily from bound forms in that (with a few exceptions)they do not take the middle affix.Grammaticality judgements by DU show that of the free forms listed in- 34 -Table 6 only three take a middle form.4 All these forms have a semantic changeassociated with them, as can he seen in the examples below.7. c’ek”-m wi? she lit her lamp.[shine-MDL 3 PCL8. x”ák”-m.ffrost-MDL 39. nés-am knfgo-MDL lsgOf the bound roots listedmiddle suffix.510. pI1a-m..finforin-MDL11. cnn-rnfsay-MDL12. téw-m lsg13. cü+-m knJpoint-MDL lsgTwo middle forms were14. +im-om knfcut.brush-MDL lsg15. +im-p-m6..[cut.brush-INC-MDL4. The dictionary listing for two forms allowed middles, méc’x ‘blink’ (to wink at someone), and +á ‘heal’(shaman specializing in marital relations) but these were not accepted as grammatical by DU.5. These include forms that are listed in the dictionary as free forms, but which were not accepted by DUwithout primary affixation.6. This form is the only example of a predicate with two primary affixes on it, and as such is an exceptionto the rule of mutual exclusivity for all primary affixes except stative and immediate, which can occurtogether.- 35 -make something frosty [object or glass]I brought some with mein Table 7, bound control ditransitives accepted theinform (someone about something)speak for someone (to somone for a purpose)! choose a spouseI am selling (something to someone)(404b)I pointed, indicated (something to someone)grammatical for the control root /4-Im-, as can be seen below.I cut brush (230)to cut brush (275)These results, and those in 10 and 13 are expected given the agentive-active nature ofthese roots.Of the bound involuntary bodily process roots the majority take middles with thecharacteristic agentive reading.16. x”üc-om kn I threw up (247b)fvomit-MDL lsg17. tkéy’-m to urinate.furinate-MDLSome bound involuntary bodily process roots, however, were ungrammatical in themiddle form. These forms allow only an inchoative form or an out-of-control-form.18. *cIcW..m make something bleed (*326b)7%Jbleed-MDL 319. u?ex kn c[.?]Ic” I am bleeding (326a)lsg %Jbleed-INC20. *qázm sweatfsweat-MDL21. ?e qáz z kn I might sweat (342c)weat-OCThe unanalyzed items listed in Table 7 easily took the middle form, as exemplified below.This appears, at this stage, to show that these forms are indeed active-agentives.22. yfW hide (something)hide-MDL23. ciq-m dig (somthing).fdig-.MDL24. ?á’-m sweep (something)weep MDL7. This form is listed as grammatical in the dictionary.- 36 -25. +ép-m extinguish [fire/light].fextinguish-MDLOf the forms listed as bound state and psychological predicates, few in the database hadmiddle alternates except the following four forms. Note that the middle of only two ofthese forms has an implied object, c’éx-m ‘shame’ and áw’-m ‘thirsty’ do not have thischaracteristic reading.26. k+-m subtractfseparate-MDL27. caq”’-m paint (something) red.Jbrown/red-MDL28. c’éx-m kn I am ashamed/shy (96)fshame-MDL lsg29. áw’-m thirsty.fthirsty-MDLThe rest of the bound states had no elicited data or were ungrammatical.30. *qzinJtartle-MDL31. *tey.m (327b)Jhungiy-MDLThe results of this section are inconclusive. The middle affix applies to activeagentives (the bound ditransitive control forms), involuntary bodily processes, and theunanalyzed set. There are also examples of middles with the bound states andpsychological predicates. I conclude that the small number of middles in this set of boundforms is due to an elicitation gap, because middle is documented as a productive formin the language. Therefore, the middle affix does not provide conclusive evidence for anunergative/unaccusative split in N+.- 37 -2.5.2 AutonomousThe autonomous suffix references actions that are controlled by a specific agent,although it also attaches to some inanimate states (dynamic). Autonomous is consideredto be marked non-control.Data on autonomous forms were not extensive, but free forms from a range ofsemantic classes were deemed ungrammatical by the consultant, as the following show.32.fwalk-AUT33yawn-AUT34. * +yük”-ixJbumped-AUT35. pl’ñx”-ixJpunctured-AUTBound forms often take an autonomous affix, as can be seen from the range of formsfrom the active, stative, and unanalyzed sets.36. k’é-ix joke, jest.J]oke-AUT37. x+-ix to cheer oneself upfcheer-AUT38. +qWlyx mount a horsefput on top-AUT39. hide oneself.fhide-AUT40. péw-ix [of snake, toad] puff up, swell upwe1l.up-AUT41. k+-lyx get away from a situationIeparate-AUT- 38 -42. k”’olc’-Iyx [road] curves sharplyfcurve-AUT43 qW4 to turn red of own accord.Jbrown/red-AUT44. k’ém-ix sneak alongfstealthy-AUT45. mél-ix take a restfrest-AUTNo involuntary bodily process roots or psychological predicates were elicited with theautonomous affix. Autonomous appears to apply across the board, thereforeit is not auseful diagnostic of an unaccusative/unergative distinction.2.5.3 InchoativeInchoative marks a developing action that occurs without the intervention of anagent. Inchoative is considered to have non-control marking. Given its nature as anaspectual marker, and its lack of effect on agency, the inchoative has potential to be anaspectual diagnostic of an unaccusative/unergative distinction.Free form intransitives are not found with inchoative alternates, with theexception of the one stative form listed below.46. he c’e[?]k” it might shine (409)..Jhine.[INC] 3There are no data regarding the use of inchoative with bound control (agentive) roots.Bound involuntary bodily processes that take inchoative are:47. u?ex kn c[ ?]Ic’’ I am bleeding (326a)lsg Jb1eed-INC48. n-p shiver, feel coldJhiver-INC- 39 -It can be seen from the following forms that inchoative can be used as a primary affixwith the set of questionable roots:49. yVWp disappear%/hide-INC50. 4qWp landed therefput on top-INCIt is common with stative bound forms.51. k4’-p it came apart%fseparate-INC52. ci[ redJbrown/red-INC53. qz-op to be startledfstartle-INC54. k’k”-p burstJburst-INCThe use of inchoative presents the first set of data that the aspectual elementdistinguishes between roots in N+. The examples in 47 through 54, from the unanalyzedset and the bound statives, show that: (1) The unanalyzed forms are patterning with thestates (Examples 49-54), and, (2) Involuntary bodily processes do not pattern with theother agentives. The inchoative creates a change of semantics to non-control in examples57 and 58. It is also of note that there is no inchoative use possible with the agentive freeforms, and that there is a complete gap of data for the agentive bound forms. It appears,therefore, that inchoative applies only to unaccusative roots, which is evidence that thequestionable set of forms are unaccusatives. As this affix is aspectual in nature, this isevidence that the aspectual element is important to a distinction between roots in N+.- 40 -2.5.4 ImmediateImmediate designates states and actions that have just gone into effect, andimpending states or action. It is unmarked for the feature of control. As such, it is asecond potential aspectual diagnostic.There was a single immediate free form accepted as grammatical from a rangeelicited of DU.55. x’’ak’’-t it is frosty.Jfrost-IMAs can be seen, with the bound roots immediate is a common alternate for theunanalyzed set:56. nik’-t get cutIcut/slice-IM57. ?á-t just now sweptq’weep-IMand stative roots:58. péw-t swell up, swollen..Iwell.up-IM59. zéw’-t tired of somethingfannoy-IM60. téy-t kn I am hungry (327a)..Jhungry-IM lsgThe majority of these forms also have alternate intransitives with other primary affixes,as will he seen in the discussion. The use of the immediate affix provides corroborativeevidence for an unaccusative/unergative split. All forms have a patient reading; therefore,like inchoative it is diagnostic for unaccusativity.- 41 -2.5.5 StativeStative designates accomplished actions and states with a remote cause. It isconsidered to be unmarked for control (Thompson and Thompson 1992).Statives elicited with free form roots showed that grammaticality of forms variedwithin each semantic grouping:61. *?es/xWlt (385b sf)ST-s/walk62. *?es/nés (405b sf)ST-s/go63. ?es/?üq”e? drunkST-s/drink 364. ?e(s)/sliéw yawningST-s/yawn 365. *?es/dyq (389f sOST-faint66. * ?es/xáni (384f sf)ST-/hurt67. ?es/zoq” already deadST-/die 368. ?es/q”nox’” be sickST—/sick 369. ?es/ptük”’ (390h Sf)ST-s/ooze70. *?esIclOxw (419h Sf) ?es/k’yáq”’ already broken [rope, string] (262 sI)ST-5jbreak 372. ?es/plüx” it is punctured (383c sf)ST-s/punctureAll psychological predicates and weather roots tested as ungrammatical in the stative- 42 -form. Bound control roots regularly take a stative alternate with a corresponding shiftto a patient reading:73. ?es/k”’én it has been takenST-./grasp74. ?es/puiys already killedST-5/kill75. ?es/püyt lying downST-sJlie.down76. ?es/+lm [of brush] already trimmedST-s/cut.brush77. ?es/cü+ it is pointed outST-s/pointThe questionable set (78-79) and statives (80-83) show a corresponding shift:78. ?es!ciq already dugST-5[dig79. ?es/+ép [of fire] extinguished/outST-5/extinguish80. ?es/k+ detachedST-sJeparate81. ?es/caq’’ redST-sjbrown/red82. ?es/zéw already tiredST-s/annoy83. 7es/k’uik” already burstST-5jburstStative appears with causatives and other derived forms, so there is no evidence for theorder of attachment of this form. It may be a late process, as Gerdts (1991) suggests.I have no explanation for why stative does not appear with some free forms, but doesappear with others. It is of note that with the bound control forms stative creates a- 43 -patient form (recall there was no evidence for inchoative or immediate use with theseforms). Stative does not appear to be a diagnostic in the same way as inchoative andimmediate, because it applies both to agentive and non-agentive forms. However, its usewith the bound control forms may be analyzed as evidence that at some level these formsare unaccusative as well.2.5.6 Out-of-ControlThis reduplicative affix indicates that an event or a state has developed withoutan obvious cause. It is quite productive and adds to both agent- and patient-orientedroots. Recall that it is marked non-control, because when it is added to agentive roots,it gives an exceptional volition agent reading in some cases and a non-volitional patientreading in other cases.Out-of-control forms were limited in the data. However, agentive forms that tookout-of-control had the characteristic exceptional energy required or non-volitionalreading, depending on the semantic requirements of the form, as in:84. x”’eslt t manage to walkfwa1k-OC85. ?es!?iiq’’[ oq”] e7 get a drink somehowST-Jdrink-OC86. néss manage! are enabled to gofgo-OC87. mlc[c]e?q manage to get up.,fsit-OC88. nóx ox”’ animal is forced to run%[run-OC89. twep-op move backwards inadvertently, get moved backJback up-OC- 44 -90. 1’éyi be stopped, come to a halttop-OCThe state roots pattern in the same way.91. zóq’’ . oq’’ murderedIdie-OC92. cWyjX[ x]m-me-t-m have someone get a little jealous of onejealous-OC-RLT-3/indef sub.Two forms appeared to alternate between the agentive and non-agentive readings.93 manage to go to sleep/get put to sleep%fsleep-OC94. cük” uk” it got finished (with difficulty)/someone finished itsJfinish-OCThere were exceptional control and stative free forms that were ungrammatical with anout-of-control affix. These were k”’ñce ‘descend to the water’(412i), kwice ‘ascend to thewater (413i), and pz’éI ‘worthless’ (398n).The bound control roots,8 show the same pattern, as the agentives and states, ascan be seen in:95. ?fls os be discarded (out-of-control)fdiscard-OC96. k”'énn taken by someone (uncontrolled).fgrasp-OC97. püy[ i]s get killed or beaten up (out-of-control)Jkill-OC98. cün n be talked about (out-of-control)ay-OC99. téwu kn I have been sold something (despite resistance) (out-of-control)[sell-OC lsg8. This includes forms that Thompson & Thompson (1992) designate as + control, and the + control freeforms in their data that DU would not accept as grammatical without the middle primary affix.- 45 -So did the rest of the bound roots of the unanalyzed set, listed below. Note that one ofthese forms (example 100) has the exceptional volition reading more characteristic ofagentive forms.100. +qw.oq manage to mountIput on top101. +épop someone else has put it [a fire] outIextinguish102. k++ removed by someonefseparate-OC103. cdqw.oq come to be red mé+ put to pasture, laid off..frest-OC105. cIqq dug up (by someone)..fdig-OCGiven the fact that the out-of-control affix applies to free and bound forms alike,it is not a diagnostic of unaccusativity versus unergativity. The distinction that Thompsonand Thompson (1992) note between exceptional volitional and patient readings foragentive free form roots appears to be merely a factor of the root semantics pushing aparticular reading, as the bound agentives take an out-of-control reading where anexceptional volition reading would be expected. Out-of-control has no effect on theargument structure of a root.2.5.7 DesiderativeThe next set of morpho-syntactic processes I compare are desiderative andcausative. These two diagnostics were chosen on the basis of available comparative workin Hk (Gerdts 1991). Gerdts shows that these two tests serve as a diagnostic for an- 46 -unaccuisative/unergative distinction in Hk. I apply them to the data to determine if thesame results apply in N+. First I will consider the desiderative data. Recall that thedesiderative is a secondary affix that attaches directly to roots. It is unmarked for control.Desiderative is a productive process in N4-, but is not one of the more commonformations (like causative, directive, or middle). There is a tendency for the consultantto use a clausal formation in the same form of the English gloss.To recapitulate Gerdts (1991), unergatives in Hk allow desiderative with thetypical gloss of ‘want X’. Some unaccusative process verbs allow desiderative, with a shiftin semantics to a future meaning, and some process verbs disallow desiderative. Stateverbs never take the construction.Desiderative forms were easily elicited, as were grammaticality judgements. Amajority of free form agentives were grammatical, as can be seen in:106. x”esit-méinn kn I want to walk (235sf)fwalk-DESID lsg107. nes-mémn kn I want to go[go-DESID lsg108. naq”-mémn kn I want to steal (265 sf).[steal-DESID lsg109. nox’’-mémn [bird, horse] wants to run.,frun-DESID 3110. tax-mdmn kn I want to paddle (242 sI)fpaddle-DESID lsg111. *7uq’e?memn kn I want to drink (271 sf) (DU: would somehow mean liquour).fdrink-DESID lsgInvoluntary bodily process root ranged in grammaticality according to their semantics,as 112 through 115 show:- 47 -112. wek’k’-mémn kn I want to throw up (248a)Ivomit-DESID lsg113. shew-mémn feel like yawning, begin to yawn.fyawn-DESID 3114. ?*?.sxemémn kn I want to sneeze (246 sf)Jneeze-DESID lsg115. ‘mec’x-mémn kn I want to blink (298b sf)Jb1ink-DESID lsgThere were two exceptional free form states that took the desiderative. These are listedin 116 and 117:116. zoq’’-mémn%fdie-DESID 3117. u?ex xan’i-mdmn.Jhurt-DESID 3The rest of the states were118. *kWismémnffall-DESID 3119. ?iyq’-memnJfaint-DESID 3120. *qwnoxwmemnJick-DESID 3121. *maqmemnfsatiated-DESID 3122. *+Wmémnfheal-DESID 3123. *kyaqmémn.Jbreak-DESID 3124. *c1ox’mémnJhot-DESID 3125. *SluxWmémnfnoise-DESID 3he is near death, wants to die (310 a,b)it is dangerous, he is asking to get hurt (348c Sf)ungrammatical in the desiderative, as can be seen in:to want to fall(313b sf)to want to faint (389 sf)to want to be sick (350b Sf)to want to be satiated (3971 sf)wound wants to heal (351e sf)string wants to break (262a sf)weather wants to be hot (419e sf)noise wants to be made (416e)- 48 -126. *cok’mémn want to shine (409e)fshine-DESID 3127. *plux’mémn want to be punctured (383)fpuncture-DESID 3No data were elicited for psychological items. It is noteworthy that the only evidence ofthe second future reading found in Hk was with the weather roots. Three of the weatherforms had an inceptive/future reading to the desiderative form. The others wereungrammatical.128. wuxWtmémn it looks like snow (257b sf)fsnow-DESID129. tek+-mémn it looks like rain (253 sf)[rain-DESID130. cap-memn becoming dusk (372)fdusk-DESIDTransitive desideratives are created by adding transitive /-t-/ directly to thedesiderative affix. Indirective and causative are also possible transitivizers. There wereno data regarding relational use.131. x’”esit-mémn-ne + sqaqya? I want to walk the dog (239)[walk-DESID-TR-3/1 DIR fdog132. ta-mémn-ne + sëq’áw+ I want to paddle the boat (243 sf)%Ipaddle-DESID-TR-3/l DIR ..Jboat133. x”esit-mérnn-s-t-m we want him to walk (236sf)%[walk-DESID-CAU-3/lpl134. x”esit-mémn-xo-c-n te sqáqca? I want to walk your dog for you (237)walk-DESID-IND-3/1sg OBL .fdogSome transitivizations of acceptable desiderative intransitive forms were ungrammatical.135. *naqvmemnsne I want him to steal (266a sf)steal-DESID-CAU-3/1It is of interest that while desiderative constructions in Hk and NI- are similar in that in- 49 -both languages desideratives can be both transitive and intransitive, they differ in thatin Hk the desiderative morphology is added to a base which includes the transitivizer,whereas in N the desiderative suffix becomes part of the base that transitivizers areadded to.The desiderative appears to distinguish actor-events. Except for the examples ofzóq’’ ‘die’ and áni ‘hurt’, which are grouped semantically and morpho-syntactically aspatient-oriented forms, desiderative is only grammatical with agentive roots and some ofthe bodily process forms. Only some bound involuntary bodily process roots areungrammatical. Once again, involuntary bodily processes are not patterning withagentives as a unitary group. It is of particular note that, apart from the two exceptionsof ‘die’ and ‘hurt’ listed above, desiderative never suffixes to states. Weather roots (oneof which is documented by Thompson and Thompson to be control) are an anomalousset as they form desideratives with future semantics. Desiderative morphology is then adiagnostic of unergativity in N4.2.5.8 CausativeIn NI-, causative is the most productive of the transitivizers. As was seen in theintroduction, it is the transitivizer most commonly used to transitivize primary affixedroots and to re-transitivize complex intransitives. The results of elicitation show that itapplies nearly across the board. There are some exceptions to this productivity.Causative does not occur with ?écq” ‘bake’, naqw ‘steal’(267 sf), m’n ‘give’, twep ‘movebackwards’, zác ‘get dressed’, c1uxv ‘noise’, %fk’z- ‘lazy’, máq’ ‘satiated’9(397i sf) and the9. The root máq’ does not take any transitivizers.- 50 -weather roots xwák ‘frost’, tékl ‘rain’, káxvus ‘hail’, and cap ‘dusk’. There is, however,a causative form based on the weather’° root ‘snow’ in the database (see example 136)that shows that if an appropriate context and protagonist can be found a causative is notblocked.136. wux”t-s-t-s [Coyote] makes it snow (this form was rejected by%Inow-CAU-3/3 by DU (257c sf))The set of items to which causative does not apply is not a unitary one, and the formsshould be considered idiosyncratic.Causative, therefore, is not selective in what semantic classes (and, by logicalextension, syntactic classes) it applies to. Therefore it cannot be used in conjunction withdesiderative to distinguish unaccusatives in the same way as in Hk.However, the interpretation of causative transitivization in N+ with agent-orientedand patient-oriented roots can be shown to distinguish semantic classes.With actor-activity roots causative has a straight causal reading, as in:137. x”esIt-s-c help someone to walk, take someone for a walk/wa1k-CAU-3/3138. no-s-c force an animal to runfrun-CAU-3/3139. tá-m-s-cm-s he made me paddle (270)%fpaddle-RLT-CAU- lsg-3Two involuntary bodily processes, in 140 and 141, take causative.140. wék’k’-s-cm-s he caused me to vomit/vomit-CAU- lsg-310. The weather roots do not take directive or indirective forms. They often have relational forms withindirect subject pronominals.- 51 -141. shew-s-t-és they made us yawn (and they kept talking) (281 sf)s[yawn-CAU- lpl/3The bound form control roots also appear to have this causal reading:142. püyt-s-c put someone to bed%/lie down-CAU-3/3143. you cheered me upST-QLT-fcheer-CAU- lsg-2sgHowever, with the majority of state roots, as Thompson and Thompson (1992) havedocumented, there is an accidental reading to causative activity.144. zóq”-s-c kill someone accidentally/die-CAU-3/3145. do something that indirectly results in making someone sickfsick-CAU-3/3146. án’i-s-cm-s x? he hurt me accidentally!hurt-CAU.2sg-3147. +yük”’-sm-s he accidentally bumped me (381b)Jbump.into-CAU- lsg-3148. ptük’’-st-ne I (accidentally) caused it (my sore) to ooze (390d)..fooze-CAU-3/lsg149. k’yoq”-st-éne I accidentally broke the rope (379b)Jbreak-CAU-3/1sg150. zk’áq-s-c he made it fall down (accidentally)ftumble-CAU-3/3151. I punctured it accidentally (383a).fpuncture-CAU-3/lsg152. (‘le) dyq’-s-cm-s x?o thats what made me faint (389c2)Jfaint-CAU- lsg-3 (no animate cause allowed)153. c’lOx’’-s-cm-s it makes me hotIhot-CAU- lsg-311. The form listed in the dictionary is q+ ‘cheer’. DU is unfamiliar with this form.- 52 -One state root has two sets of causatives, one with the accidental reading and one withthe straight causal, as in:154. k’”i-st-one I dropped it accidentally (314)ffall-CAU-3!lsg155. kWisscms she caused me to fall.ffall-CAU-lsg-3The directive form of this root is:156. k”is-e-s drop something intentionally, let fall intentionallyffall-DR-3/3One patient-oriented root does not take the accidental reading.157. c’ék”-s-ne wi I shone a light on his face (intentionally) (409)Jlight-CAU-3/1sgPsychological predicates do not take the accidental reading. Some have two sets offorms.158. V”yIxm-s-cm-s u?ex he makes me jealous (358)Jiealous-CAU- lsg-3159. qlIl-s-ne I made him angry.fangry-CAU-3/lsg160. páq’’u?-s-cm-s he scared me (300) or he is afraid of me (302)fafraid-CAU-1sg/3161. páq”’u-s-ne I am afraid of him (303)Iafraid-CAU-3/lsgIn the following section, I tabulate these results and discuss them.- 53 -2.6 DiscussionThe data from the primary affixes inchoative, autonomous and middle show onlythat free forms do not take these affixes and bound forms do. It is important toremember that all bound forms do not take all primary affixes. This thesis makes noattempt to distinguish the reasons why bound roots differ in this respect.From their behavior with primary affixes, the fact that they appear to allow allprimary affixes, and the fact that the agency of the predicate changes with affixation, thefollowing forms that were the unanalyzed set should be grouped with the state predicates.I review the behavior of these roots with primary affixation below in example 162. Inexample 163 I show comparative affixation with a set of state roots.162. a b c d e[ciq- /?a- ..[+q”- fnIk’Jhide s/dig q’weep fput on top fcutMIDDLEii. Yçw1 ciq-m ?á-m nIk’-mJhide-MDL /dig-MDL sweep-MDL %fcut-MDLhide s.t. dig s.t. sweep s.t. cut s.t.AUTONOMOUSiii. yçW..jyX - - +qW4yx -..Jhide-AUT %[put on top-AUThide oneself mount a horseINCHOATIVEiv. yçWp - +q’’-p -Ihide-INC fput on top-INCdisapppear landed thereSTAT1VEV. - ?eslciq - - nIk’-tST-dig /cut-IMalready dug get cut- 54 -IMMEDIATEVi. - - ?áy-t -weep-IMjust been sweptOUT-OF-CONTROLVii. - - - 4qw.oqwfput on top-OCmanage to mount163. a b ci. [caq”- s[pdwJeparate %fwellMIDDLEii. k4-m caq’’-m -Jeparate-MDL paint (s.t.) redAUTONOMOUSiii. k+-Iyx qWjx pew-ixeparate-AUT we11.up-AUTget away turn red of [of snake, toad]from s.t. own accord puff up, swell upINCHOATIVEiv. k+-p ci[?]q’’ p[’?]dwJeparate-INC fswe11-INCit came apart red begin to swellIMMEDIATEV. - péw-tfwe1l.up-IMswollenSTATIVEvi. ?es/k+ ?es!caq” ?es/péw-tST-.fseparate ST-.fswell.up-IMdetached red swollenOUT-OF-CONTROLvii. k++ qW.oqW%Ieparate-OC by come to be redsomeone inappropriately- 55 -It can be seen that the inchoative and immediate of all these forms have patientreadings. It is also noteworthy that these roots appear to create alternate forms with allprimary affixes. The stative affix appears to behave differently from the other primaryaffixes. Unlike the other primary affixes listed above it can attach to some of the freeforms. It is productive with bound forms as well. There seems to be no obvious patternto which forms it will attach to and which forms it will not attach to.While the out-of-control affix has no diagnostic value, the desiderative andcausative appear to be relevant morpho-syntactic diagnostics of underlying root structure.The aspectual affixes are a primary set of diagnostics of unaccusativity, as these affixes(inchoative and immediate) attach only to patient-oriented roots. Therefore, the set ofquestionable roots, despite their vague reading as actions, are in fact unaccusatives. Theodd set is the agent-oriented/unergative roots which do not allow these affixes. Therefore,anything that the aspectual affixes attach to are unaccusatives. Table 8 summarizes thegeneral patterns of distinctions in semantics of the classes with these morpho-syntacticaffixes.Table 8: Morpho-Syntactic Diagnostics of Unaccusativity/UnergativitySemantic Class Desiderative Causative Immediate InchoativeAgentive-Action grammatical causal reading ungrammatical ungrammaticalBodily Proc. grammatical causal reading variable variableState ungrammatical accid. reading grammatical grammaticalPsychological no data causal reading no data no dataWeather future reading causal reading no data no dataor unknown- 56 -The distinction between bound and free forms is not made on the basis ofsemantic class or syntactic class. Free forms pattern both unaccusatively and unergatively,and they belong to a range of semantic classes.Desiderative forms with two exceptions distinguish agentive forms. Weather rootshave a future reading with this affix. The results from the analysis of the desiderative inN+ partially match those of Gerdts, in that the items that can be semantically classed aspotential unergatives allow desiderative formation. The potential unaccusatives split intostate/change of state roots that do not allow desiderative, and the anomalous set ofweather roots which form a future with this suffix. There are two exceptions, in that zóq”‘die’ and án’i ‘hurt’ pattern with the unergative forms.Causative distinguishes state roots with an accidental reading from agentive,psychological, and involuntary bodily process roots with a causative reading. The resultsof an analysis of the patterning of causative show that while causative is productive inapplying to all classes of roots, the unaccusative state roots are distinguished by havingan accidental reading.The primary aspectual affixes (immediate and inchoative) are strong diagnosticsof unaccusative, as can be seen from the fact that there is always a patient reading onthe root when they are attached. This means that the questionable forms listed in thepreliminary classification are in fact unaccusative roots.It is noteworthy that the bound control forms allow a middle form. It is also ofinterest that the involuntary bodily process roots, and psychological and weatherpredicates behave in a manner that is not as clearly characterizable as simply- 57 -unaccusative or unergative.These results show that the two elements of aspect and agentivity play a role inN; there is an unergative and unaccusative distinction in the language. The semanticclassification is therefore a useful tool for distinguishing between unaccusatives andunergatives, and the cross-linguistic generalizations hold. In table 9 I present a revisedclassification of the bound and free forms discussed in this Chapter. The data from themorpho-syntactic diagnostics show that there are: a set of agentive/unergative forms; anda set of state/unaccusative forms. The results also show that involuntary bodily processroots, weather roots, and psychological predicates must be considered seperate sets. Itis of note that involuntary bodily processes are not behaving as a unitary set. This resultmatches the patterning of these forms cross-lingusitically, as discussed in Rosen (1984).- 58 -Table 9: Classification of Free RootsA. UNERGATIVES1 2 3 4 5 6xwesIt ?üq’’e? nés naq’” IIOXW tawalk drink go steal run.animal paddle7 8 9 10 11 12mlFt q?ém p’nt’ kvüce k”üme qáytvisit [infant] nurse return descend ascend reach.top13 14 15 16 17 18q’amIn kii óyt páq’’ twép m’n’throw help sleep watch back up give19 20 21 22 23 24k’éy ?ecq” kéze? pték’’+ záx fUtsstop bake tell a lie, tell a story, get dressed fdiscardbe a lie be a storyteller25 26 27 28 29 30%fp1le- fpuys- Ipüytgrasp joke give inform/tell kill lie down31 32 33 34 35 36/téw- %fcu+bite cheer say cut brush sell to point- 59 -B. UNACCUSATIVES1 2 3 4lá ptük” k’y’áq’2 k’áq’wound.heals13 ooze break tumble.down5 6 7 8c’ék” ?lüx” c’lóx” plÜXwshine bang/noise hot weather/food punctured9 10 11 12c’k’ lpüx” zoqw k”isexhausted have a hole through die fall13 14 15 16Iyq’ kn pz’é+ y’e q”nó”faint worthless good sick17 18 19 20xáni +aci +yük” máq’hurt cold bumped into satiated21 22 23 24xIy cük” [qWec fqzgo ashore finished move startle25 26 27 28Jméi-- .JkWacburst escape rest crazy29 30 31 32cáw’- Jey- Jk’émlazy thirsty hungry stealthy33 34 35 36../k’lc’- fcéq”- .[yc”’- IciqIcurve brown/red hide dig37 38 39 40Inik’sweep put on top cut/slice scratch41 42 43f4’ép- .f4-éq- /céwextinguish sit (plural) reach12. The dictionary form of this root is k’m’áq.13. DU allows this root only with an inanimate referent. In Thompson & Thompson (1990) the root canrefer to a healer.- 60 -C. PSYCHOLOGICAL PREDICATES1 2 3 4páq”u?fWyj qlIlafraid jealous angry shy/shameannoy (-Ct!)D. WEATHER1 2 3 4wuix”t xákw tek+snow frost rain hail (+ctl)E. INVOLUNTARY BODILY PROCESSES1 2 3 4wék’k’ sxe shéw méc’xvomit (-ctl) sneeze (-Ct!) yawn blink5 6 7 8sftkéy- fsfip’eject/vomit urinate breathe itch9 10%[qázdefecate sweatIn Chapter Three I take the free forms discussed here and analyze them todetermine if there is evidence for underlying transitives in NL- 61 -CHAPTER THREE3.0 Transitive Forms of Roots3.1 IntroductionIn this chapter I discuss the potential for underlying transitive and intransitiveroots in N+. Opinion on this issue has been split. Gerdts (1989) hypothesizes underlyingtransitives and intransitives, as does Kuipers (1968). The opposite view that all roots areintransitive is the more traditional one, held by Thompson and Thompson (1992).Apart from causative, the transitivizing suffixes do not apply across the board.Directive, indirective and relational are blocked for different items. The focus of thischapter is to discuss the set of free roots with respect to their behavior with the set oftransitivizers. I bring in some comparative data from bound forms. The basis fordistinguishing between transitive and intransitive roots is the logical assumption that thetransitive use of predicates will be based on their thematic underpinnings. Therefore,directive should apply only to transitive roots, and never to a form that is classified asan unaccusative. Indirectives should apply to ditransitives.First I set up a preliminary semantic grouping of forms that could potentially betransitive versus those that could be intransitive. I base this classification on whether theroot appears to imply a patient as well as an agent. To aid in the classification ofambiguous forms, I bring in relevant data from Chapter Two. Some roots will potentiallyfit into both categories. In the sections following, I will present the use of thetransitivizers with each set of roots as morpho-syntactic tests of transitivity. In Section3.4 I present a discussion of the results and suggest what their implications are for the- 62 -argument structure of predicates.It is of note that a conclusive test for transitivity would be if a root surfaced onlyin a transitive form. There do not appear to be any roots of this nature in the database.A second test is if roots only surface as a middle (intransitive) or as a transitive form.There are a number of roots that surface only as middles, but their semantics wouldindicate that there is no patient possible (?üy-m ‘laugh’, p’áq’-m ‘bloom’, I’y-m ‘wadeinto water’) and their transitivization potential is limited to causatives. From a perusalof the database it appears that most roots allow at least two primary affix alternates.While it is questionable to use middle as a sole diagnostic of transitivity it may be usefulin conjunction with directive and indirective in a discussion of bound forms.3.2 Preliminary Semantic Grouping for a Transitive/Intransitive DistinctionFollowing the semantic grouping of roots into the two main classes of States andAgentives, the class of what could be termed variable roots, and the minor classes ofPsychological, Weather, and Involuntary Bodily Process Predicates, the semantic groupinginto potential transitive and intransitive sets is reasonably straightforward. Transitiveforms must be agentive and allow a patient. Intransitives can be either agentive or nonagentive forms. I present an initial grouping in Table 10 below.- 63 -Table 10: Preliminary Semantic Classification of Transitive/Intransitive: Free RootsA. TRANSITIVE ROOTS1 2 3 4 5náq” ta mIl’t pték”+drink steal paddle visit narrate6 7 8 9 10páq” q’amIn cük”’ knwatch bake throw finish helpDitransitive Roots11 12 13 14sfcü(n)- fcü+ /téwdonate say point sellB. INTRANSITIVE ROOTSAgentive Roots1 2 3 4 5xWesIt q?em óyt p’nt’ kwücewalk nurse sieep return descend6 7 8 9 10k”üme qáyt kdze? xIy zácascend lie go.ashore get dressedNon-Agentive Roots11 12 13 14 15 16xáni Iyq’ pz’é+ y’e qWnoxW maq’be hurt faint worthless feel good be sick satiated17 18 19 20 21 22+áxi +áx” ptük’’ k’áq’ rluxw c’1O”be cold wound heals ooze tumble.down bang/noise hot23 24 25 26 27 28lpüx’’ plüx” +yük’7’ k”+ k’y’áqexhausted have hole punctured bumped spill break29 30 31 32 33 34nds k”’is zóq” k’éy twép nógo fall/drop die stop back up animal run35 36 37 38 39 40wék’k’ ?ésxe shéw méc’x páq”u?shine vomit sneeze yawn blink afraid41 42 43 44 45 46fWyj) qlIl wxk’ x’’ak’’ téki k’aüsjealous angry snow frost rain hail-64-3.3 Morpho-Syntactic DiagnosticsIf the set of transitivizers were going to be diagnostic of underlying transitivity, itwould be logical to assume that use of the directive would mark an underlying transitiveform. The indirective should apply to transitive forms with underlying goal arguments.Neither of these transitivizers should ever apply to a non-agentive form. Causative, it hasbeen seen in Chapter Two, applies productively to all predicates. Relational use maymark a transitivization option for an underlying intransitive form.3.3.1 DirectiveDirective patterns differently from the the relational and causative because itdoes not, with minor exceptions, appear to transitivize forms with the primary affixes, nordoes it ever re-transitivize complex intransitives, or combine with other transitivizers. Theindirective appears to pattern with the directive, except that it is found in combinationwith other transitivizers.The following examples document the grammaticality judgements of DU withrespect to the use of directive with the forms in Table 10.A. Transitive Roots: Directive Use61. ?uqwe?ns drink something.,[drink-DRV-3/362. tá-e-s propel something [canoe,boat]Jpaddle-DRV-3/363. pték”’+-e-s tell someone a story[narrate-DRV-3/364. ?écq”-e-s bake something.Jbake-DRV-3/3- 65 -65. cük”-e-s complete somethingJfinished-DRV-3/366. *naqn.1. (268)Jteal-DRV67. *mi4,tnt (344c).[visit-DRVOf the roots that I would consider to be transitive, naqv and mIt are exceptionsin that they do not allow directive. Other forms in this classification pattern as expected.Comparative data from the bound agentive/control roots show that these forms whichsemantically would be considered transitive take the directive.68. ?üs-c discard something[discard-DRV-3/369. k’’én-s take something, grab..fgrasp-DRV-3/370. püys-c kill something! someoneJlci1l-DRV-3/3All forms that can be considered underlyingly ditransitive take the directive.Ditransitive Roots: Directive Use71. m’on’t-és give someone (something)!feed someoneIgive-DRV-3!372. cü4-e-s point out someone/thingIpoint-DRV-3/373. téw-e-s sell (something) to someoneJell-DRV-3/374. cu-n-ne I told him (something)ay-DRV- lsg!3B. Intransitive Roots: Directive UseAgentive Roots appear to take directive as in:- 66 -75. kéze?-s lie to someone, deceive someone,Rie-DRV-3/376. k”üce-(n)-s put something into the water[descend.water-DRV-3/377. k’me-n-s move something from the waterfascend.water-DRV-3/378. záx-e-s clothe! dress someonefdress-DRV-3/3Roots that were documented in Chapter Two as unaccusative also take the directive asas can be seen in:79. +yükWcms he bumped meJbumped.into-DRV-1sg-380. k”4-t-és pour, dump something outfspi11-DRV-3/381. kiqwtene I broke it on purpose (379a)Jbreak-DRV-1sg/382. c’ló-e-s make something hot.,Jbot-DRV-3/383. c’ok-t-és finish the last of somethingfexhausted-DRV-3/384. lpüx’’-e-s he made a hole in itJhave a hole-DRV-3/385. pz’é4--e-s not care about something (uncommon form)fworthless-DRV-3/386. qWnoxWes make someone feel bad [directly responsible]Jick-DRV-3/387. ptk”--t-éne I burst it on purpose (my sore) (390e)Jooze-DRV-3!lsgRoots from both the agentive and unaccusative sets were ungrammatical, as in:88. *xwesItnt (233a)fwalk-DRV- 67 -89. *cWoytnt (309c)fsleep-DRV90. *pntnt (406c)..freturn-DRV91. *qáytnt (388d) *xIyqnt (389)ffaint-DRV93. *q’.t. (397j)fsatiated-DRV94. ái-n-t- (355).[cold-DRV-95. ?‘xáni-n-t- (384e)sJbe hurt-DRV96. (377)umb1e.down-DRV-97 *jW (383)fpunctured-DRVThe forms that are classified as intransitive, therefore, are problematic in that alarge number of them take directives. This problem might be mitigated somewhat if itpertained to the agentive forms, for it can be seen that the directives for these formsreference locations and goals rather than theme arguments. These forms might beconsidered exceptional. It can be seen that some of the involuntary bodily process rootshave the same type of reading.98. wék’k’-e-s vomit on somethingfvomit-DRV-3/399. ?osxé-ne I sneezed on him (323c).Jneeze-DRV-3/3The problematic forms are the forms that are clearly unaccusative/state roots that- 68 -allow the directive (as in examples 79 to 87). These forms have the characteristicdirective semantics, and as such, cannot be considered exceptional.The forms that I would consider to have both a transitive and intransitiveargument structure on the basis of their change in semantics all take directives asexpected.100. nés nes-c convey, take someone somewherefgo/take Jgo-DRV-3/3101. kwIs k”Is-e-s drop something.ffall/drop %fdrop-DRV-3/3102. zoqw zóq”-e-s kill somethingfdie/ki1l fdie-DRV-3/3103. c’ék” c’ék”-e-s shine light on somthing, light the wayfiight!sMne .Jlight-DRV-3/3The inanimate agentive form below is anomalous, as the transitive form appears to havean intransitive reading with a change in root semantics to ‘crawl’, as opposed to a causalreading or a directive reading such as ‘to run a race’.104. [nó” nó-e-s [many-legged creature] following agentive forms are problematic if they are considered to be underlyinglyintransitive because they reference a theme argument, and could not be subsumed underthe rubric of exceptionality.105. k’éy-e-s stop someone (from doing something)Jtop-DRV-3!3106. twep-e-s cause/force something to move backwardsJback.up-DRV-3/3Psychological and Weather roots pattern as would be expected if they were intransitive.- 69 -107. (300a)5[afraid-DRV-108. *q0111.n..t.. (391b)s/angry-DRy-109. wÜ’-n-t/snow-DRV110.s/frostDRV111.s/rain-DRV112. *kaxWüsnt5/hail-DRV-On the basis of the examples (in 79 through 87) which show that roots that clearlypattern as unaccusative states (Chapter Two) take directive, it would appear that theassumption that directive is a diagnostic of underlying transitivity is flawed.3.3.2 IndirectiveIf the indirective transitivizer were a diagnostic of transitivity, then indirectiveforms would require underlying agent and goal arguments. The following are examplesof the behavior of the hypothesized transitive and intransitive classes with indirective.Note that these examples would have to be underlyingly ditransitive to take theindirective. This is clearly an non-intuitive and inelegant way of approaching this firstset of agentive forms.A. Transitive Roots: Indirective Use113. ?üq”e?-x-c drink a beverage belonging to someones/drink-IND-3/3114. náq’’-x-c steal something for someone..Jteal-IND-3/3- 70 -115. mI4’t-x-cm-s she will visit for me (344e).fvisit-IND-lsg/3116. ?u?ex ptókv4xnne te s/ptdk4-.s I am telling it for him (his story) (to people)((347c)fnarrate-IND-3/lsg OBL NOM-%fstory117. ?ecq”-x-ne I baked it for him in the pit oven (296d).Jbake-IND-3/lsg118. cük’’-x-ne I finished it for him (357d)sffinish-1ND3/lsgIf indirective was a diagnostic of underlying structure, it should not appear as frequentlyas it does.The ditransitive roots pattern as expected, as the examples below show.Ditransitive Roots119. cüt-x-c accept someone, speak for someone.Jay-IND-3/3120. cü4’-x-c point toward someone..[point-IND-3/3121. téw-m-x-c sell (something) belonging to someoneJell-RLT?-IND-3/3122. (289b sf)/give-INDIt can be seen from the results below that intransitive roots which should not takeindirective allow it with some forms.B. Intransitive Roots: Agentive123. kvücexncms wi? tén-s/qqn he is taking my horse down towards the river (412h)/descend.water-IND- lsg-3...124. k’iime-xa-cm-s wi? tén-s/qcq he is taking my horse away from the rivers edge (413h)[ascend.water-IND-1s-3...125. xIy-x-s-cm-s he might land it for me (417f)%fland-IND- lsg-3- 71 -126. p’n’t-x-c return (something) to someonefreturn-IND-3/3127. (233a)%fwalk-IND128. *cWoytxjt (309b)1eep-IND-129. *qáytxit (388f) Agentives are problematic because they have the regular semantics of anindirective form.B. Intransitive Roots: Non-AgentiveUnaccusative intransitives are inconclusive, but appear to pattern in the same wayas the agentives.130. pz’é+-xit-Iyxs x t? t x’pië they thought nothing of his property (398j)/worthless-IND-3/3p1131. 9-yük”-xi-t (381d)Jbumped.into-IND132. (383b)Jpunctured-INDThe forms with two potential argument structures, it can be seen, patterned asexpected with the exception of zóq” ‘die’. These forms would not normally be consideredditransitives.133. ns-x-ne te s!saq”-s I am taking along their papers (345a)fgo-IND-3/lsg OBL NOM-fpaper-3psv.134. JkwIs-x-ne I dropped it to him (315)[drop-IND-3!1sg135. k’éy-x-cm-s he stopped it for me (runaway horse) (414h)[stop-IND- lsg-3- 72 -136. twép-x-cm-s Wi? ten/ká he backed it up for me, my car (4131f)Jback.up-IND-1sg-3137. he shines the light for meIlight-IND-1sg-3138. *zoqWxit (311a)fdie-INDInvoluntary bodily process roots and psychological roots split into those that allowan indirective and those that do not. The expected result is that these forms would notaccept indirective.139. wék’k’-xt-om + her feet were vomited onsJiomit-IND-3/3indef. DIR NOM-Jfeet140. ?esx-éne I sneezed on him (323c)’Jneeze-IND-3/lsg141. *shéwxjt (320b)jawn-IND142. cWyj)qflX..1e te cIt”-s I am jealous of his house (358)Jjealous-IND-3/1sg OBL Jhouse-3psv143. (301a)fafraid-IND144. *qolIl..xi4 (306b)fangry-.INDThe results of this section show that indirective patterns much like directive, inthat it is not blocked on non-agentive forms. It also marks many roots that are notintuitively thought of as indirective. These data suggest, therefore, that becauseindirective must bring added argument structure to a root, and since it seems to applyacross the board, this transitivizer presents no evidence for an underlyingtransitive/intransitive distinction.1. This form is inconclusive, it may be a directive with a locative reading, or it may be indirective.- 73 -3.3.3 CausativeThe causative transitivizer is of a different nature from directive and indirective.Recall that it is very productive, applying to all semantic classes of roots. It is also usedexceptionally to transitivize complex intransitives, and is the most common transitivizerused with primary affixed stems. For examples of the use of causative with the free formset, see Chapter Two. This transitivizer is not a useful diagnostic for elucidating atransitive/intransitive split.3.3.4 RelationalThe relational transitivizer is not as clear-cut a diagnostic as directive andindirective are, nor is it obviously non-applicable as is the causative. Given its nature asthe transitivizer of last resort, I hypothesize that it may he a transitivizer of intransitiveroots. It is available to re-transitivize complex intransitives and can be used with primaryaffixed stems. The data following show, however, that the results are far from being ascomprehensive as the directive and indirective.Of the potential transitive forms, the available data show that it is possible to geta relational on the majority of the following forms.145. náq”-m-s steal something from someonefsteal-RLT-3/3146. ml4t’-m-s visit someone/visit-RLT-3/3147. (357)%ffinished-RLTDitransitive roots all take a relational, as the forms below show.148. cüt-m-s talk about someoneay-RLT-3!3- 74 -149. téw-m-s sell somethingsfsell-RLT-3/3150. m’n’-min-m-s give (something) away to someonefgive-RLT-?-3/3Intransitive roots of an agentive nature generally allow a relational, being blockedin a number of cases.151. xWesitmtis he walked to meet usfwalk-RLT- lpl/3152. p’én’t-m-s [of weather or season] come back return to someone..[return-RLT-3/3153. n/kwücemes go downriver after somethingLOC-.fdescend.water-RLT-3/3154. n/k”üme-me-s go upriver after (to get) somethingLOC-..fascend.water-RLT-3/3155. *qayt..min..t. (388g) *c”óytmint (321b)1eep-RLT-Of the non-agentive intransitive roots, which would all be expected to all take relational,it can be seen that some do not allow it.157. *pzé+mint (398i)Iworthless-RLT158. plüx”’-min-t- (383b)[punctured-RLT159. y’e-min-s he likes.... (394h)fgood-RLT-3/3160. (381d)Jbumped.into-RLTThe forms with two alternate argument structures take the relational affix except wherethe semantics would bar it.- 75 -161. nés-m-cm-s he took me with him (269 sf)Jgo-RLT-lsg-3162. twdp-m-cm-s he backed up on me (41311’)%Jback.up-RLT-1-3163. (310d)fdie-RLTThe majority of involuntary bodily process roots and psychological roots take therelational, as can be seen in the forms below.164. wék’k’-m-s vomit something up.fvomit-RLT-3/3165. méc’x-rn-s blink at someone%fblink-RLT-3/3166. shéw-mt-m someone makes one yawnfyawn-RLT-3/indef167. ?sxemin-t- (323e)neeze-RLT168. fwyIxmmes be jealous of someonefjealous-RLT-3/3169. qlil-m-ne I got mad at him (306)./angry-RLT-3/lsg170. (304a).fafraid-RLT3.4 DiscussionIf one starts from a conservative assumption that N4-, like English, has underlyingintransitive and transitive roots, and presupposes that directive and mdirective weremarkers of underlying transitives, we find some minor evidence that there are transitiveroots in N4-. From the discussion of the out-of-control data in Chapter 2 comes data thatthere are agentive roots with an exceptional volition reading in the out-of-control forms.If semantics of the effects of this affix are split up, these might appear to maintain their- 76 -agent participant. These forms do not take the directive suffix. One could hypothesizethat these forms were underlyingly intransitive. The second set of agentive roots takesthe characteristic non-volitional (patient) reading when the out-of-control affix is added.These forms allow the directive transitivizer and could be considered transitive. Thebound control forms that take the out-of-control reading would then be transtive by thisanalysis.However, the majority of evidence points to the fact that rather than there beingan intransitive/transitive split in NI, the morpho-syntactic diagnostics discussed andexemplified in this Chapter also distinguish agentive and non-agentive forms.In the following tables, I summarize the data from the diagnostics in Sections 3.3.1to 3.3.4 above. In these tables, an asterisk marks an ungrammatical form, and a plus signmarks a grammatical form. A dash signifies a gap in the elicited data. A combination ofplus and asterisk symbols means that the form was given in the dictionary, but wasconsidered ungrammatical by DU.- 77 -Table 11: Morpho-Syntactic Behavior of Unerative RootsRoot Gloss MID CAU DRV IND RLT1 ?uq”e drink * + + + *2 ptek”’4’ narrate/tell a story * + + + -3 tax paddle (a canoe) * + + + +4 keze? be/tell a lie * + + - -5 k’”uce descend-water */+ + + + +6 k”ume ascend-water * + + + +7 nes go + + + + +8 m’n’ give, donate * * + * +9 zax get dressed * * + - -10 ?ecq’’ bake */+ * + + *11 k’ey stop * + + + *12 cwoyt sleep * + * *13 cuk”’ finished/end * + + + *14 mi4-t visit * + * + +—miii—15 p’en’t return + + * + +16 qayt reach-top * + * * *17 x”’esit walk * + * + +18 micoq’ to sit * + - -19 q?em nurse * + * - -20 nox animal-run * + + - +21 naq”’ steal * * * + +- 78 -Table 12: Moroho-Svntactic Behavior of Unaccusative RootsRoot Gloss MID CAU DRV IND RLT1 xiy goashore + * +2 clOXV [weather] be hot * + + -3 c’ek”’ shine/give light + + + +4 an’i be/get hurt * + * -5 iyq’ faint * + * -6 laxi becold * + * * *7 pl’ux” punctured * + * * *8 zk’aq’ things piled up fall * + * -9 clUX”’ it is a noise/s.t. bangs * -10 maq’ satiatied * * -11 +a be healed/heal up *7+ + + +12 qwnoxw be sick/ill * + + - -13 zoq’’ die/be dead * + + * *14 k”is fall, be born * + + + *15 4-yuk” bang (into s.t.) * + + * *16 k”+ [liquid] flow * + + -run, spill17 Ic’m’aq / jyqW [cord] break, + + -pull apart18 ipux” [ice/glass] have + + -a hole19 ptuk” ooze out, spring * + +water source20 pz’e+ foolish * + + + *2. Dorothy Ursaki does not allow a middle form +á”'-m. However, the dictionary has the form:4áx”‘’-m tok ywin’ shaman specializing in marital realations- 79 -Table 13: Morpho-Syntactic Behavior of Miscellaneous RootsRoot Gloss MID CAU DRV IND RLT1 ?osxe sneeze * + + *2 wék’k’ vomit * + + +3 shew yawn * + - *4 paqwu fear/frighten * + *5 q1il angiy/mad * + * * +6 hail * * * - +7 tek4’ rain * * * - +8 wux’’t snow * *1+ * - ÷9 cap dusk * * - +- 80 -CHAPTER FOUR4.0 Conclusions and Future WorkThe focus of this thesis was a discussion of the semantic basis of transitivityalternations in NIe?képmx. Because intransitive and transitive use of roots in NIe?képmxis morphologically marked, the purpose of this work was to document what derivationalaffixes are allowed with a set of predicates. This led to a classification of predicate types.In Chapter One, I presented an overview of the morphology of N4-e?képmx. ChapterTwo comprised an application of the semantic diagnostics from the literature onunaccusative and unergative (Grimshaw 1987, Levin & Rappaport 1989, Grimshaw 1990,Gerdts 1991) to a set of NIe?képmx roots. A set of morpho-syntactic diagnostics, includingprimary affixes, and (following Gerdts 1991) desiderative and causative affixes, was analyzedto determine if they could elucidate underlying structure.The results of the data collection documented in Chapter Two show that unergativeand unaccusative is a relevant distinction in NI. There is a distinction based on agency thatmarks unergatives, as seen from the causative evidence and corroborated by the desiderativediagnostic. Unaccusatives are distinguished by the aspectual markers of inchoative andimmediate. It is apparent from the data that the psychological, involuntary and weatherpredicates do not pattern evenly with either the unergative or unaccusative groups. Thisevidence corroborates Rosen (1984) by showing that for some roots the semanticunderpinnings of argument structure vary from language to language. This can be seen mostclearly with the involuntary bodily process roots which do not behave as a unitary set, andwith the weather roots. This is the only set where the desiderative has a future reading. This- 81 -shows that the weather forms cannot be grouped with the state/unaccusative predicates. Itis possible that there may be some process roots that allow a future reading, and that thesewere missed in elicitation. Another possibility is that the full extent of the use of such formis being lost by my consultant. It is of note that during the elicitation of the weather formsDU attempted to create a second translation of these forms by setting them in a sentencesuch as ‘the earth wants to rain’.In Chapter Three I discussed the potential of an intransitive/transitive classificationof roots in N4-e?képmx. It can be seen from this discussion that there is little evidence fortransitive roots in NI. The morpho-syntactic tests in combination do show agency as arelevant semantic dimension, corroborating the evidence from Chapter Two. But there areforms that are clearly documented as unaccusative in Chapter Two that take directive andindirective forms. These data show that these transitivizers cannot be diagnostic of anunderlying transitive form.I conclude therefore, following Thompson and Thompson (1992), that roots in NI areunderlyingly intransitive. Given the proportion of agentive versus non-agentive roots, it isalso clear that the traditional view holds that in Salish most forms are patient-oriented.The implications of these data for cross-linguistic work are that transitive roots arenot universal. The data from Chapter Two show that the unaccusative and unergativedistinction is a viable one in NI. The behavior of the involuntary bodily process, weather andpsychological predicates tends to corroborate Rosen (1984), showing that for a limited setof forms in the data, the semantic underpinnings of argument structure can vary fromlanguage to language.- 82 -In retrospect, there are a number of methodological considerations that wouldcontribute to a clearer analysis of these data. This thesis is based on the extensive workdone by Thompson & Thompson with Annie York and other consultants, and on itemselicited from DU, a fluent and very linguistically astute consultant. However, thegrammaticality judgements come from a single source (DU), as the Thompson database doesnot distinguish between elicitation gaps and forms that are ungrammatical. An attempt wasmade to find a second consultant with whom to verify forms, but it was difficult to gainaccess to a speaker with the degree of fluency required.A second consideration is that the semantic classification in Chapter Three is largelyintuitive. A more conclusive, syntactic test of these judgements would be to elicit intransitiveforms with oblique object nominals.My primary set of data was the subset of roots in N4-e’?képmx that can appearas free form intransitives in a main sentence. This group of roots cross-cuts the broadsemantic categories found in the literature. However, it is important to note that whilechoosing the set of free forms gave a finite set of common roots with an identifiable subsetof agentive forms, the majority set (and potentially more interesting set) of roots is that ofthe bound forms. Further work on these forms would explain the distinction between boundand free forms. It is of note that the inchoative and immediate forms (the aspectual affixesdistinguished in the data) do not attach to the free forms. Thus the next step would be todetermine if the distinction is an aspectual one. Looking at the type of roots that surface asfree forms, one can see that these roots are very common. Their status as free forms maybe linked to this semantic familiarity.- 83 -Future work would be to map the semantic structure of bound roots in light of theconclusions reached in this thesis. An interesting issue is the event-structure/aspectual natureof the primary affixes and their patterning with roots. The results of Chapter Two show thatthe forms discussed as primary affixes are not a unitary set, as stative and out-of-controlbehave quite differently with respect to free forms from the other affixes.- 84 -BIBLIOGRAPHYCampbell, L. and M. Mithun (eds.) 1979. The Languages of Native America. Austin:University of Texas Press.Carison, B.F. 1980. Two-Goal Transitive Stems in Spokane Salish. InternationalJournal of American Linguistics 46:21-26.Carison, B.F. and L.C. Thompson. 1982. Out of Control in two (maybe more) SalishLanguages. Anthropological Linguistics 24:51-65.Comrie, B. 1981. Language Universals and Linguistic Typology. Chicago:University ofChicago Press.Demers, R. and E. Jelinek. 1984. Word-Building Rules and Grammatical Categoriesin Lummi. Papers of the Nineteenth International Conference on Salishan andNeighboring Languages. University of Victoria. Victoria, British Columbia.Dowty, D.R. 1979. Word Meaning and Montague Grammar. D. Reidel, Dordrecht.Egesdal, S.M. 1992. Styalized character’s speech in Thompson Salish narrative.University of Montana Occasional Papers in Linguistics No. 9.Gerdts, D. 1979. Out of Control in Ilokano. Proceedings of the Fifth Annual Meetingof the Berkeley Linguistics Society, pp. 81-93.Gerdts, D. 1984. A Relational Analysis of Halkomelem Causals. Syntax andSemantics; The Syntax of Native American Languages 16, ed. E.D. Cook andDonna Gerdts, pp. 169-203. Orlando: Academic Press.Gerdts, D. 1988. Object and Absolutive in Halkomelem Salish. New York:Garland.Gerdts, D. 1989. Relational Parameters of Reflexives: The Halkomelem Evidence.Theoretical Perspectives on Native American Languages. ed. Donna Gerdtsand Karin Michelson, pp. 259-80. Albany;State University of New York Press.Gerdts, D. 1991. Unaccusative Mismatches in Halkomelem Salish. InternationalJournal of American Linguistics 57:230-50.Gerdts, D. (to appear). Morphologically Mediated Relational Profiles. Proceedings ofthe Eighteenth Annual Meeting of the Berkeley Linguistics Society.Grimshaw, J. 1987. Unaccusatives: An Overview. NELS 17, pp. 244-258.- 85 -Grimshaw, J. 1990. Argument Structure. Cambridge: The MIT Press.Hanna, D. (to appear). Our Stories: Nie?képmx Oral Narratives.Hess, T. and J. van Eijk. 1985. Noun and Verb in Salishan. Twentieth InternationalConference on Salish and Neighbouring Languages. UBC. Vancouver, BritishColumbia.Hopper, P.J. and S.A. Thompson. 1980. Transitivity in grammar and Discourse.Language 56:251-299.Hukari, T.E. 1976. Transitivity in Halkomelem. Working Papers for the EleventhInternational Conference on Salishan Languages. Seattle, Washington.Jackendoff, R. 1990. Semantic Structures. Cambridge: The MIT Press.Jelinek, E. 1984. Empty categories, case, and configurationality. Natural Languageand Linguistic Theory 2:39-76.Jelenik, E. 1985. The projection principle and the argument type parameter. Paperpresented at the Sixtieth Annual Meeting of the Linguistic Society of America.Seattle, Washington.Jimmie, M. (to appear). A Prosodic Analysis of N+e’?képmx Reduplication. UBC.Vancouver, British Columbia.Kinkade, M.D. 1980. Columbian Salish -xi,4,-tuil-. International Journal of AmericanLinguistics 46:33-36.Kinkade, M.D. 1981. The Source of the Upper Chehalis Reflexive. InternationalJournal of American Linguistics 47:336-39.Kinkade, M.D. 1982. Transitive Inflection in (Moses) Columbian Salish. KansasWorking Papers in Lingistics; Studies in Native American Languages, Vol. 7.pp. 49-62.Kinkade, M.D. 1983. Salish Evidence against the Universality of Noun and Verb.Lingua 21:610-626.Kinkade, M.D. 1990. Sorting out Third Persons in Salishan Discourse. InternationalJournal of American Linguistics 56:341-60.Kroeber, P.D. 1991. Comparative Syntax of Subordinates in Salish. Phd. thesis.University of Chicago. Chicago, Illinois.- 86 -Kuipers, A.H. 1968. The categories verb-noun and transitive-intransitive in Englishand Squamish. Lingua 21:610-626. North-Holland Publishing Co., Amsterdam.Levin, B. and M. Rappaport. 1989. An Approach to Unaccusative Mismatches.NELS 19 pp. 3 14-329.Matthewson, L. 1993. A configurational pronominal argument language. ms. UBC.Vancouver, British Columbia.Mattina, A. 1982. The Colville-Okanagan Transitive System. International Journal ofAmerican Linguistics 48:421-35.Mattina, N. 1993. Some lexical properties of Colville-Okanagan ditransitives.Twenty-eighth International Conference on Salish and NeighbouringLanguages. University of Washington. Seattle, Washington.Perlmutter, D.M. 1978. Impersonal Passives and the Unaccusative Hypothesis, BLS 4.Berkeley, California.Perimutter, D.M. and C.G. Rosen. 1984. Studies in Relational Grammar 2.Chicago:University of Chicago Press. Chicago, Illinois.Rosen, C. 1984. The Interface between Semantic Roles and Initial GrammaticalRelations, in D.M. Perlmutter and C. Rosen eds., Studies in RelationalGrammar 2, University of Chicago Press. Chicago, Illinois. 38-77.Spencer, A. 1991. Morphological Theory; An Introduction to Word Structure inGenerative Grammar. Cambridge:Basil Blackwell.Tenny, C.L. 1987. Grammaticalizing Aspect and Affectedness. Ph.d. dissertation. MIT.Thompson, L.C. 1985. Control in Salish Grammar. Relational Typology, Trends inLinguistics (Studies and Monographs 28) ed. F. Plank, pp. 391-428. Mouton.Thompson, L.C. and M.T. Thompson. 1979. Thompson-English Dictionary. ms.Thompson, L.C. and M.T. Thompson. 1980. Thompson Salish //-xi//.International Journal of American Linguistics 46:27-32.Thompson, L.C. and M.T. Thompson. 1987. English-Thompson Finderlist. ms.Thompson, L.C. and M.T. Thompson. 1990. Appendices. ms.- 87 -Thompson, L.C. and M.T. Thompson. 1992. The Thompson Language. University ofMontana Occasional Papers in Linguistics. No. 8.Thompson, L.C. and M.T. Thompson, A. York, M. Joe, S. Egesdal. 1990.Push-Back-Sides-of-His-Hair: A Thompson Salish Myth. ms.Zaenen, A. 1988. Unaccusative Verbs in Dutch and the Syntax-Semantics Interface.unpublished ms. Xerox PARC. Palo Alto, California.- 88 -APPENDIX 1Free Form RootsAgentive Activity1. ?écq” bake in ashes2. ?üq’”e? drink3. k”üce descend-water4. k’’üme ascend-water5. ICéy stop6. m’ón/m’n’ give (s.t.), make a donation! s.t. given7. mIcq’ to sit8. mI+t (make a) visit9. naq”’ steal10. nés go to (place), go for (purpose)...11. nó animal-run12. pték”1 narrate, tell a story/narrator, story-teller13. p’én’t come back, go back, return, revive...14. tá to paddle (a canoe)15. twep go, move backwards16. q?em [of infant] nurse, suck at breast17. qáyt reach the top, summit of s.t.18. xiy go ashore19. xvesIt walk, go, take a trip, travel20. zác get dressed, put one’s clothes on...21. VwÔyt sleep22. kn help23. páq” watch24. téw sell25. cü+ point26. l’Im cut brush27. cün say28. kéze? be a lie/falsehood, tell a lie/falsehood29. cük’’ that’s all, as soon as, be completed, finishedNon-Agentive Forms1. k”is fall down, be born2. +yük” bang/bump (into something)3. pl’flx” puncture, touch (s.t.), finger goes through4. an’i be/get hurt5. Iyq’ faint, lose conciousness6. ky’üx” perforated7. kw61 [of liquid] flow, run, spill8. )c’m’áq [of cord] break, pull apart- 89 -9. ptfik’’ ooze out, water source, spring10. zI’aq’ (things piled up) come down, fall; (of house) cave in11. flüxv s.t. bangs, makes a racket12. c’k exhausted, finished13. laxi be cold14. +á be healed, heal15. Ic’z lazy16. máq’ be full, satiated17. pz’é+ free, easily done, worthless, reckless, foolish18. qwnoxw be sick, ill19. wméx live, be alive20. züm big, large, great21. zóq” [person, animal, plant] die, be deadStates: Inanimate1. ?éy present, here2. có burn/blacken, burn up, burn black3. c’l’ó wash Out [of road,hillj4. c’ló (be) hot, hot [weather, object]5. c’ma [of berries] mashed, [of clothes] worn out6. k+ detached, apart7. k4üxw severed, cut off8. lpüx’” [of ice/glass] have a hole through9. Ic’iy’ difficult (to do), hard, elevated [language]10. )‘méx braided11. néx’’ strongly, very, intensively12. q’lux” curled13. .ctáq [obj/substance] develops a hole14. zn6k coiled [of snake, rope]15. c’ek’’ shine, give light16. Ilüx” be a noiseInvoluntary bodily processes1. shéw yawn2. ?ésxe sneeze3. wék’k’ vomit- 90 -Psychological1. c’éx shame2. k”á crazy, insane3. paqwu? fear/scare/frighten4. q1il angry, get angry/mad5 jealous, enviousWeather/Nature1. 1c’aüs (to) hail2. ték4 to rain3. WÜX”t to snow4. lap it is dusk, twilight5. x”ák”’ get frosty, there is frostNominal Forms1. ?ék”’n bait/salmon roe2. ?éyk kinnickkinnick berry3. ?imc grandchild4. ?Iic white worms (in rotten wood)5. cItx” house6. c’y’é basket7. képu(w) coat8. kix elder sister/cousin9. kwülu?xr dog salmon10. Ilk” prayer beads11. 4-wéy’st autumn12. I’émn body hair/fur13. I’éms birch basket14. I’ix”e+ different/strange, wrong one15. k’ümk’ whistle/drinking tube16. mc’ü4’t pus17. qmüt hat18. qlex”' blanket19. qwtel grease20. qWu water/river21. q”zém moss22. séw’t slave23. só”'m’ edible sprouts, blackcaps, sunflower seeds24. tinx sinew25. tmix”' land, earth26. niix”' ladder27. zm’én’ bird nest/one’s origins- 91 -APPENDIX 2Control RootsFree Forms1. x’’esIt walk2. cü(n) say3. náq” steal4. tá paddle5. ?üqwe? drink6. nó animal run7. im cut brush8. cü+ point9. nés go10. k6n help11. mI+t visit12. mlá(m) heal13. p’nt return14. q?ém to nurse (breastfeed)15. qáyt reach top16. q’amin throw17. téw sell to18. k”’üceh descend to water19. kwümeh ascend from water20. m’n donate21. só’est descend (this form may be a re-analyzed reflexive)22. ték4 rainBound forms1. f7üs- discard2. fk”én- grasp3. ../k’éx- joke4. %fné- give5. fpIle- inform/tell6. %fpuys- kill7. fpüyt- lie down8. %fqól- bite9. fqa- cheer- 92 -


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items