f rv s

BEYOND TTIECORE:TYPOLOGICAL VARIATION IN THE IDENTIFICATION OF PARTICIPANTS MenlewNnMrruuN

SANTABARBARA UNNERSITY CALIFoRNIA. oF Languagesyary not only in their argumentcategoriesbut also in the relative burden borne by core arguments,on the ono hand, and obliques or adjuncts,on the other, in conveying information- Here it is shown that the kinds of participantsidentified in many languageswith oblique nominals or adjuncts, such as companions,recipients, benenciaries,instruments, sources, and goals, are usually identified by alternative structuresin Mohawk. The only referentsintroducedby nominal adjuncts are places and times, entities that can characteize whole situations,The roles of all participants are specifiedin or inferable from the verb: in pronominal prefixes,applicativesuffrxes, incorporatednouns,directional prefixes,and the semanticsof the verb stem.It rrill be seen that such head-markingstructureis more than a simple formal vatiable. It can bave impodant implications for the lexicon and for issuesin languagedocumentation. structure,obliques, adjuncts,Mohawk, Iroquoian] [KEywoRDs:zugument

Basic sentencestructureis generally assumedto consist of a predicateand one or more constituentsidentifying participants. The participants are often classified into two groups, though the labels for the groups and the criteria for classification vary acrossthe literature. Common labels for tle two are core/oblique, primary/secondary, nucleus/satellite, argument/adjunct, and complement/adjunctamong others.In some traditions, the primary criterion for the distinction is formal marking: core argumentsmay be unmarked, for example, while obliques are distinguished by case suffixes, enclitics, determiners, or adpositions.In other traditions, the primary criterion is subcategorization: core arguments are those that are required by the verb, while adjuncts are optional. Our focus here is on neither terminology nor definitions but rather on systematicpatternsof cross-linguisticvariation. We know that languagesdiffer in the formal robustnessof their reflection of the distinction. Here we consider another kind of difference: the relative burden borne by the two kinds of constituentsin conveying information. Although we know that categorieslike subject and object are not purely semantic, there is a general assumption that semantic agents and patients tend to be representedas core arguments,while companions,beneficiaries, instruments, sources,goals, locations, and times tend to be representedas non-core. Of course,languagesoften provide devices for altering argument structure for syntactic and discoursepurposes,such as passive, antipassive,

[UAt, vot. 7r, do. 4, october 2005,pp.445-?2] @ 2005 by Th Univaity of chicaso, All rishts reserved002L70? 1/2005/7104-0003$t 0.OO

445

446

INTERNATIONAL JOURNAL OF AMERICAN LINCUISTICS

speechin variand applicativeconstructions.But if we comparespontaneous ous languages,we find substantial crossJinguistic differences in the distribution of information over core and non-core constituents.Certain kinds of participantsthat are usually categorizedas core argumentsin English, for example, are obligatorily categorizedas obliques in Eskimo-Aleut languages, becauseunidentifiable (indefinite) semanticpatients of transitive events are never permitted core status.In the Yup'ik sentencein (1), the motor is necessarily oblique, and the verb is grammatically intransitive, (l) Central Alaskan Yup'ik: Elizabeth Ali, speaker(personal communication) Kiputellinilia levaa-mek levaa-mek kipute-Ilini-Iria buy-apparently-INTRANsrrrvE.pARTrcIpIAL.3.sc motor-ABLATTvE he apparentlybought motor 'He apparently bought a motor'. By contrast,in a numtrerof other languages,many of them in the Americas, spontaneousspeech shows a much smaller proportion of obliques or adjuncts than might be expected on the basis of languageslike English. 1. The Iroquoian languages. In the Iroquoian languages of eastern North America, core arguments are easy to distinguish on morphological grounds. They are identified by pronominal prefixes in every verb. The pronominal prefixes are presentwhether coreferentialnominals are also present in the clause or not. In (2) below, the intransitive verb 'go' contains pronominal reference to its single argument 'she', while the transitive verb 'mind, look after' containspronominal referenceto its two arguments'I' and 'him'. Examples cited here are drawn from Mohawk, spoken primarily in present-dayQuebec,Ontario, and New York State,but the patternsare typical of all of the languages the family.l in
1I am grateful to the Mohawk speakersfrom Kahnawh:ke,Kanehsata:ke, Aikwesdhsne, Ohsw:ken,and Thaientand:ken who have generouslycontributed their time and expertise. Three anonymous reviewersalso provided useful discussion. of the material cited here that All came from unscripted,connectedspeech,pdmarily is attributed to specific Mohawk speakers conyersation, udless otherwise noted. I have benefited especially from discussion with Kaia'titdhkhe' Jacobs,of Kahnaw):ke, Quebec.The Mohawk is presented eheorthography in approvedby a conso ium of all of the communitiesand curently in use in schools.It is es' sentially phonemic.Obstruentsare automaticallyvoiced before voiced segments. Glottal stop ('). is represented with an apostrophe The two nasalizedvowels ale represented with digraphs: er for a [asalized carot a\d on for a high, back nasalized[u]. The other vowels i, e, a, o have their approximateIPA values. A colon (:) marks vowel length. An acuteaccent(.i) represents stresswith high tone (on short vowels) or rising tone (olr long vowels). A graveaccent(A) replines,epenthetic vowelsarein small caps:1,,{,E. resents stress with fallirg tone.In the segmented

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

447

(2) Mohawk pronominal prefixes: Cecelia Peters,speaker(p.c.) 6nska Kb:iahre' enska k-ehiahr-n' l.sc.Acr-remember-srATTvE one one I remember

waLiaten 'nikbn:raren' wa-hii-aten-'nikonhr-A-ren-'

ne ne

the FAcr- 1.sc/M.sc-MTDDlE-mind-JR-set.on-p the I minded him i'kdn:'a. ri-'ken'=a 1.sc/u.sc-be.older-sib=orlra mv younger brother Idh iah
NOt

tewakatuien:ture' te-wak-ate-rien't-A-r-E'
NEGATIVE.l.SC.PAT-MIDDLE. WATENESS.JR-bC-iN-STATIVE

not ne ne the the

do I know istdn:'a isten'=a mother=DM my mother ka' ka' where where

niiakawenbn:ne'. ni-iakaw-e-n-on-hne'
PARTTTM-FI.SG.PAT-go-DIR.APPLIC-STATM-PAST that she had gone to 'I remember once I minded my younger brother. I don't know where my mother had gone'.

Most examples appear in a multi-line format. The second line of these exarnples shows a morphological segmentation of the material; the third gives the gloss of each morpheme; the fourth provides a word-by-word fanslation; and the final line is a free translation. Abbreviations for glosses are as follows: AGT (grammatical) AGENT;AppLIc AppLIcATIvE: AUc AUGMENIATIVE; BEN BENEFACTIVE; CIS CISLOCATIVE;DIM DIMINUTIVE; DIR DIRECTIONAL; DIST DISTRIBUTTVE;DU DUAL; DV DUPLICATIVE; EXCL EXCLUSIVE; FACT FACTU{; FI FEMININE IN. DEFINITE GENDER: FT]T FUTURE TENSE; HAB HABTTUAL ASPECT: TNCLINCLUSIVE; INSTR INSTRUMENTAL; JR STEM JOINER; M MASCULINE GENDER; MID MIDDLE; N NEUTER GEI..'DER;NMZR NOMINALIZER:Ns NouN SUFFIX; P pERFEcTIvEASPECT(traditionally termed the Punctual by Iroquoianists); pAT (grammatical) PATTENT;PL PLITRAL; pRT pARTrrrvE; prjRp puRposrvE; REp sG REPETTTTVE; srNcuLAR; sr srATIvE; TRL TRANsLocATrvg. Affixes hyphens () and enclitics with equal signs (=). are linked to stems with

448

INTERNATIoNAL

JoURNAL OF AMERICAN LINGUISTICS

There are no casemarkers on nouns or noun phraseshere, and no real counterparts to the adpositions of languageslike English. Word order does not distinguish syntacticrelations or semanticroles. To identify the kinds of participants typically identified in oblique nominals or adjuncts in other languages,speakershave a variety of structural alternativesat their disposal. 2. Accompanying persons and objects. In many languages,companions and associatedobjects are typically identified in oblique constituents formed with adpositionsmeaning 'with', oblique determiners,or comitative or associativecase markers. In Mohawk, accompanyingpersons or objects are often introduced instead in clausesof their own. (3) Mohawk companion:Watshenni:ne' Sawyer, speaker(p.c.) 6:nen at.the.time akhs6tha ak-hsot=ha FI.sc/l.sc-be.grandparent.to=DlM mv grandmother th6:ne tho=hne there-that then 6:nen, onen now now tsik tsi=k as=only just entierd'then', en-t-ie-ra'then-' FUT-cIs-FI.sc.Acr-climb-p she will climb up here

eniatid:rente'. en-iaki-ahrent-E' FUT-l.EXcL.DU.AGT-sleep.tog.-p we two wiII sleep together

'Then my grandmother would come upstairs and sleep with me'. (4) Mohawk accompanyingobject: JosephineKaierithon Horne, speaker(p.c.) T6kani' oskeruSn:ton' enid:rihte', o-skenonton' en-ie-ri-ht-x' tdka=hni' maybe=too NEUTER-deer FUT.FI.sc.Acr-be.cooked-c.nusnnvE-p perhaps she will fry deer ne: teniale,v atatidhsten' ne: t-en-iakttva-atat-iehst-en-'
thAt DI,'PLTCATIVE-FUT-1.PL.EXCL.AGT-REFLEXI!'E-MiX-BEN.APPLIC.P

that we will combine for ourselves ne ne the the kana'tarokhin:we, ka-na'tar-o-k=onwe l.IElrER-bread-be.in.water-coNTrNuATn/E=senuine real bread

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

449

tanon' tanon' and and

kdn:ie ka-iie N-grease grease

entionkwd:ti'. en-t-ionlaua-ati-' FUT-cIsLocATIvE-l.PL.PAT-throw-P we will pour (it) there

'We would serve the cornbread with venison and gravy'. Of course, languageslike English allow speakersto introduce accompanying persons or objects in separateclausesas well- My grandmother would come upstairs, and then we would sleep together.This possibility can make the absenceof oblique comitative and other associative constructions in languageslike Mohawk that much easier to overlook. Mohawk speakerscan also introduce companionsas componentsof dual or plural core arguments. The companion is identified in an independent noun, but the pronominal prefix refers to the whole group. (5) Mohawk core companion: John Maracle, speaker kanien'kehd:ka Tensenihthaninnion' ka-nien='ke=haka t-en-seni-hthar-onnion-' N-flint=place=resident Dv-Fur-2.pu-talk-DlsT-p you two will talk Mohawk person tsio'k tsi=o'k 56=just all sorts of nahd:ten' nahd:ten' something things

ensenihninnion'. en-seni-hron-nion-' FI-rT-2.DU-say-DIST-P you two will say 'When you talk with a Mohawk, you can say all sorts of things', 3. Recipients and beneficiaries. Recipients and beneficiaries are identified in many languagesin oblique phraseswith dative casemarkers or adpositions meaning 'to' or 'for': 'I gave it to the mailman', 'I bought it for my mother', In Mohawk, such personsare typically cast as core arguments.No more than two argumentscan be specified within the Mohawk pronominal prefix complex- In situations involving three sets of participants, semantic agents and recipients/beneficiaries,which are usually human or at least animate, take precedenceover semantic patients. The system is thus akin to what has been termed a primary object language (Dryer 1986), though the Iroquoian pronominal prefx system shows a grammatical Agent/Patient pattern rather than a Subject/Object pattern (Mithun 1991).

450

INTERNATIoNAL

JoURNAL OF AMERICAN LINGUISTICS

Some verb roots, such as 'give', 'lend', and 'feed', have semanticrecipients or beneficiaries as core arguments. (6) Mohawk recipient as core: Rita Konwatsi'tsai6n:ni Phillips, speaker(p.c.) Kati' ndn:wa' alg!;fujatem'swri:wi. kati' n=onhwa' aa-ietshii-ate-ra'sw-awi just the=maybe oprATT!'E-FI.sc/2.PL-MTDDLE-luck-give.sTATTVE just maybe she might have luck given vou 'She might have given you good luck'. (7) Mohawk recipient as core: Karihw6nhawe' Lazore, speaker(p.c.) Ko n'serehtanihb : ne'. n kon-'sare-ht-t- n ih-en -hne' l. sc/2.sc-drag-utrlzR-JR-lend-srATwE-PAsr I car lent vou 'I lent you the car'. (8) Mohawk recipient as core: Watshenni:neSawyer, speaker(p.c.) enkhehnekanontin:ra' Wd:kehre' en-khe-hnek-t-nonten-hra-' wa'-k-ehr-' l.sc/3.pl-liquid-rn-feed-eunr-r FAcr- 1.sc.AcT-intend-p FUTI thought ! will go liquid feed them ki:ken. ki:ken this this 'I thought I would give them some water'. Additional verb stemswith semanticrecipients or beneficiariescan be created with robust derivational morphology. Added to the verb stem -kaia'k 'pay (a certain price)', for example, a benefactive applicative suffix forms a Added to -hnekar-'serve (a drink)', new stem -karia'k-s-'pay (someone)'. it yields -hnekar-hahs- 'serve (someone) a drink'. Added to -nhotonkw'open the door', it yields -nhotonkw-ahsi 'open the door for (someone)'. (9) Mohawk benefactiveapplicative: JosephineKaierithon Home, speaker(p.c.) nd:ne Ronateihwahser6n:ni nD:ne ron-ate-rihw-,n-hseronni
M.PL.AGT-MIDDLE-matteT.JR.maKe.STATIVE thg one

they have agreed

the

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

45I

Hudson Bal, Company Hudson Bay Company Hudson Bay Company Hudson Bay Company

tsi /si that that

rotinidkhon roti-niak-h-on 3.PL.PAT-marry-DIsrR-srATIvE their wives

. enkonwati luir i a' ks_e' en -konwati -kar-i a' k-s-t' FUT-3.Pr3.PL-price-hit-B they will pay to them

-P

'The Hudson Bay Company agreedto pay their wives'. (10) Mohawk benefactive applicative: Kaia'tit6hkhe' Jacobs, speaker(p.c.) ken Enwd:ton' ken en-w-al-on-'
FUT-NEUTER.AGT.MIDDLE-bE.DOSSibIC.P INTERROGATTVE INTERROGATIVE

it will be possible

enkatathnekdrhahse' en-katat-hne t- r-hahs-s' kFUT1.sc.AGT-REFLEXIvE-liquid-JR-set-BEN.APPUC-P I will liquid.set.formvself 'May I serve myself liquid = May I have a cup of coffee?' (11) Mohawk benefactiveapplicative:Mae Niioronhi:'a Montour, speaker(p.c.) Wah nwanhot 6nkwahse'. o w a -honwa-nho-t on-kw-e- hs-n'
FACT-FI. SG/M. SG.dOOT.COVET-REVERSIVE.JR.BEN.APPLIC-P

she door uncovered.for !!3q 'She opened the door for him'. 4. Instruments. In many languages, semantic instruments are usually introduced as obliques: 'I patched it up with tape'. Their oblique status is signaled by adpositions ('with'), oblique determiners,or instrumental case markers. In Mohawk, instruments are usually introduced as a core argument of an additional clause. (12) Mohawk separateclause: JosephineKaierfthon Horne, speaker(p.c.) Ne sha'kannhsta ne sha'-ka-nenhst-at the coINcIDENT-NEUTER-corn-be.one the samg corn

452

INTERNATIONAL JOURNAL OF AMER]CAN LINGUISTICS

iah,vtitstha' iakw-at-hst-ha' 1-ExcL.PL.AGT-MIDDLE-use-HAB we use it n6:nen ne=6:nen the=now when eniakwatkdtston'. en-iakwa-at-katston-' FUT-l.ExcL.PL.Act-vID-make.soup-r we will make soup

'We make soup with the same kind of corn'. The absenceof instrumental oblique nominals in languageslike Mohawk can easily go unnoticed, since the multi-clause structure above is also possible in English; We use the sanxecorn when making soup. At a certain point, however, wondering whether the lack of instrumental nominals in my experiencewith Mohawk was simply an accident, I asked a skilled speaker how shewould translate 'He eatshis peaswith a spoon'. This speakerhas an especially fine senseof what is said in the languageand what is not. Her answer was revealing. She replied, "Well, you know we generally use the verb satshd:ri just for soup." She saw my request as a question about verbal semantics. (The stem -atshori is indeed used for slurping liquids like soup.) I amendedmy request to 'He eats his peas with a knife'. Her quick translation broke the model into two clauses. (13) Mohawk elicited translation:Kaia'tit6hke' Jacobs,speaker(p.c.) wA:ratste' A:share' wa-hra-at-hst-E' a'shar-'
KNife-N.S FACT-M.SG.AGT-MIDDLE.USe-P

knife tsi tsi AS as

he used wa'thatskd:hon'. wa'-t-ha-at-ska'hon-' FACT-DUPLICATIVE-M.SG.AGT-MIDDLE-diNE.P he ate

'He ate his peas with a knife'. Semanticinstrumentscan also be introduced in a secondkind of construction in Mohawk: as a core argument of a derived applicative verb. Mohawk contains severalapplicative suffixes that add an instrument to the set of core afsuments.

VARIAIION IN THE IDENTIFICATION OF PARTICIPANTS

453

(14) Mohawk instrumental applicative: Cecelia Peters, speaker (p.c.) Thd n6n: tho n=o-onwe there the=tt-place place that nihontekh:thahlcwe'.. . . ni -hon-ate-ka-'t -ha' -hkwe'
PARTITIVE.M, PL.AGT-MIDDLE-bUTN-CAUS-HAB-PAST

that they used to make a fire Tanon' ne: tdnon' ne: and it is ni' ohni' too

i alat,atenonhsa' tariha' tdhkhwa'. iakwa-ate-nonhs a' tarih-A-'t -A-hkw -ha' gxcr,.pr.lct-rraIDDLE-house-hot-JR-cAUsATtvE-JR-INSTR. AP 1. 4-HAB we caused our house to be hot wit! it 'There they used to build a fire. And we heated the house with it too'. (15) Mohawk instrumental applicative: JosephineKaierfthon Horne, speaker(p.c.) khok rononnhdhhuen. 6niehte' khok ron-onnhe-khw-en o-nieht-' M.pL.pAT-live-rNsrR.Appuc-srATIvE NEUTER-snow-Ns only they were living on it snow only 'They were surviving on snow'. (16) Mohawk instrumental applicative: John Maracle, speaker akw6: ne onhuehonwehndha' Ethd:ne onkwe=honwe=hneha' e-thohne akw6:kon ne person=genuine=style the tlere all Indian language the ratiwennotdhkwen. rati-wenn-ot-A-!fu-en
M.PL-WOTd-StANd.JR-INSTR.APPLIC-STATIVE

they word stood with it 'They spokein Indian'.

454

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

overtly in Mohawk pronominal preNeuter argumentsare not represented other argument present.In an intransitive verb with fixes unless there is no just a neuter agent or patient, the single argument is identified by a pronominal prefix: ka-hbn:tsi 'it is black', i6lken'it is rotten'. In a transitive verb, however, sluchas rati-wennotdhkwen '@9y word.stood.with it' above ('they spoke.in it'), the pronominal prefix rati-'theylil' has the same form as that used in intransitives srch as rati-ttikhe 'thev are running'. There is no extra morpheme corresponding to 'it'. Since semantic instruments are typically neuter, we might wonder whether they are in fact core arguments of the instrumental applicative verbs, since they are not mentioned in the pronorninal prefix. Evidence of their core status comes from the coining of lexical nominals.2Terms for objects are often formed in Iroquoian languages directly from morphological verbs, without nominalizing morphology. Such deverbal referring expressionscan designatethe activity predicated by the verb, or they can designatea core argument of the verb, as in (17). (17) Mohawk lexical ra'swdhtha' waterenn6tha' ieid:tare' nominals with verbal morphology 'fireman' 'he extinguishes = 'radio' 'it standsup songs' = 'photograph' is bodily in it' 'one

Nominals are coined especially frequently from instrumental applicative verbs, becausespeakersoften name introduced items in terms of their functions. The fact that thesewords are usedto refer to instrumentsindicates that the instruments are indeed core arsumentsof the verb. (18) Mohawk instrumental verbs as lexicalized nominals 'one writes with it' - 'pen, pencil' iehiat6nhkhwa' iehio'thii6nhkhwa''o n e s h a rp e n s - 'pencil sharpener' with it' 'one dish-wipes ieksokewdhtha' = 'dish towel' with it' iontko'tsonnid:tha''on e h a irs t v le -ma k e s = 'hairbrush' with it'
2Th term "noun" is reservedherefor words with speciflcinternal morphologicaistructu!9, The like 6niehte' '$iow' (o-nieht-E' NEUTER-snow-NouN.suFFIx). term "nominal" is used for Nominals may tre whatevertheir internal structure. words that areused asreferdng expressions, morphologicalnounslike 'snow'; they may be forms overtly derivedfrom morphologicalYerbs with nominalizing morphology like kaia'takenheidnhtsheru''paralysis' (ka-ia't-A-kenhei-onliterally 'that which has htshEr-a' NEvTER-body-rR-die-srATrvE-NoMINALIZER-NouN,stlFFIx, physically died'); or they may be morphological verbs without additional morphology, like literally 'he M.sc.AGENr-go.out-cAUS$rvE-HABITUAL, ra'swdhtha' 'fiteman' (ra-a'swa-r,t-rra' extinguishes').

VAR1ATION IN THE IDENTIF1CAIION OF PARTICIPANTS

455

io nhtsohare'tdhkhwa " iontenakto r6kstha'

one hand-washes with it' 'one covers one's bed with it'

= ' sink' = 'bedspread'

Semantic instruments can be identified in a third way in Mohawk: with an incorporatednoun. As in other compounding, the semanticrole of the incorporated noun stem is not specified: the noun simply indicates the involvement of a kind of entity. Incorporated nouns in Mohawk most often representsemanticpatients, becausethey can narrow the semantic scope of the verb in useful ways, but they can also evoke kinds of instruments and locations. In some casestheir semantic role may be inferred from context, but often there is little basis for seeing one role over another.The verb -rh-, for example, incorporatesnouns referring to a substancewith which something is coated, such as paint, butter, or even plastic (expressedwith the noun root -hneht-, the term for 'pitch'). One might thus interpret the incorporated noun as an instrument. (19) Mohawk incorporation of instrument (l9a) wakdrhon wake-rh-on
STATTVE 1.SG.PAT-COAt'I have coated it'.

(l9b) wakftg tshglirh on n' wak-ko tsher-e-rh-on n'

sTATM l,sc.PAT-paint-coatI have pqilll-coated it 'I have paintedit'. (l9c) wakewistohserdrhon ne ne wake-wistohsrr-,c-rh-on the 1.sc.PAT-butter-JR-coat-STATIVE the I have butter coated it 'I have buttered the bread'. The verb root -rh- could, however, just as easily be translated 'smear', in which case the incorporated noun might be argued to representa semantic patient. The Mohawk structure does not specify the semantic role. Further examplesof incorporatednouns that could be interpreted as instruments are below, though there is no formal distinction among semanlic patients, instruments.means.and locations. kand:taro ka-na'tl,r-o NEUTER-bread-be.in.water bread

456

INTERNATIoNAL

JOURNAL OF AMERICAN LINGUISTICS

(20) Mohawk incorporated instrument: Watshenni:ne Sawyer, speaker(p.c.) kahseriie\dneren' . ka-hseiie't-ansren-' NEuten-cord-tie.up-srATlvE it is cord-tied 'It [a bundle] was tied up with a cord'. (21) Mohawk incorporatedinstrument:CaroleeKonwatibn:se'Jacobs, speaker(p.c.) tenkoniahsrawbn:'eke'. i ahsi r-,q. en'ek-E' t -en-kon i -w 1.SG/2.SC-blanket-JR-wrap-P DUPLTCATTVE-FUTI will blanket-wrapyou 'I will wrap you up with/in a blanket'. (22) Mohawk incorporatedinstrument:Kaia'titehkhe' Jacobs, speaker(p.c.) Tanon' kaia'tdkerahs tanon' ka-ia't-akr-ahs NEUTER.AcT-body-stink-HAB and goat and rahshari:ne' ra-hshar-in-e'
vE M.sc.AcT-leash-lead-sTATl

(it) he was leash-leading 'And he was leading a goat with/on a leash'. 5. Place: location, source, goal. In many languages,locations, sources, and goals are identified in obliques marked with case suffixes, enclitics, determiners, or adpositions. In Mohawk, they can be identified in several ways. They are often introduced as a core argument in one clause,then carried over into subsequentdiscourse with deictic particles like eh or tho 'there' or verbal prefixes like the Translocatl\r away, over there or the Cise here, there. To render the equivalent of 'He saw some fish in a log', locative in the speaker (23) usedthreeclauses. (23) Mohawk separate clauses:Sonny Edwards,speaker(p.c,) ki:-. Wahatkdhtho' kf:ken wa-ha-at-kahtho-'
FACT-M.sC.ACT-M I DDLE-see-P this

he saw

this

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

45'1

ioronteniin:'en, . . . io-ront-enien'-en NeutEn. pAT-tree-fall -over-sTATM fallen ftee Khiahatkdhtho' th - i- a-ha-at -kahtho-'
CONTRASTIVE-TRANSLOCATIVE-FACT-M. SG.AGT-MIDDLE-SCE.P

suddenly he saw over there

rafiid:ti. En k6ntsion'. rati-ia't-i eh k-itsi-on' NEtrfER-fish-Nstherg Ivr,n-body-be.in.srATrvE

fish there they are bodilY inside 'He saw some fish in a log'. Like the comitative and instrumental constructions described eadier, these multi-clause constructions have counterparts in English, which can make their systematic use easy to overlook. The two directional prefixes, the Translocative'away, over there' and the Cislocative 'hither, there', arenot argumentsbut rather adverbials.The two argumentsof the verb 'see' in (23) above are 'he' and 'the fish'. The single argument of the verb 'climb', seen earlier in (4) 'My grandmother would come upstairs', is the grandmother. The Mohawk verb -ra'then'climb' is always intransitive, unlike its English counlerpan. A place can also be introduced in Mohawk as an incorporatednoun in one discourseby deictic particles and/or clause,then carried over into subsequent affixes. (24) Mohawk place incorporated in separateclause: Minnie Hill, speaker Eatons Ronnonha thi: that Eatons they Ah 6h there there Enska inska one once tuhontenonhsdn:ni'. td-hon-ate-nonhs-onni-' FACT.CISLOCATTVE-M.PL.AGT-MrOnr-r-hOUSe-make-r they their building built

458

INTERNATIoNAL

JoURNAL OF AMERICAN LINGUISTICS

6h 6h
thCTE

iewakataweia'tbn:ne' !g-wak-at-aweia't-on-hne'
TRL-1.SG.PAT-MID.ENtET-ST.PAST

there

over there I have entered

'Ididgo@ once'. Placescan also be introduced as the core argument of an applicative verb. Some old directional applicatives are very common in Mohawk, because they occur with extremely frequent motion verbs like 'go', though they are no longer productive. (In some of the other Iroquoian languages,phonological changeshave obliterated most traces of the directional applicatives.) These applicativesadd a sourceor goal to the set of core argumentsof verbs describing directed motion. (25) Mohawk directional applicative with source (25a) I:ien'. Lie-e-' pRoTHETIC-Fr.SG. going-STATIvE AGT-be. 'She is/was going'. (25b\ Kordhne kora=hne government=place Canada nitiakaw6:non. ni-t-iakaw-e-n-on
PARTITIVE-CISLOCATIVE-FI.SG.PAT- gO.DIR. APPLIC- STAT1VE

there she has come from it 'She came here from Canada'. (26) Mohawk directional applicative with goal (26a\ Wd:re'. wa-hr-e-' going-SterrVe FACT-M.SG.AGT-be. 'He is/was going'. (26b) Tiohtid:ke Tiohtid:ke
Montfeal

wd:rehte'. wa'-hra-e-ht-s'
FACT-M.SG.AGT-gO-DIR.APPLIC-P

Montreal

he went to it

'He went to Monffeal'.

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

459

containing verbs with directional applicativesmust contain referSentences ence to a specific place, like the words for 'Canada' and 'Montreal' in (25) and (26). Places can also be identified in Mohawk by incorporated nouns. Again, the incorporatednoun is not a syntactic argument;it simply indicates the involvement of a kind of entity. Its semanticrole remains unspecified,though it may be inferred from the meaning of tle verb and the situation. The role of the noun meaning 'liquid' in (27), for example, can be interpreted as the source of motion from the meaning of the verb 'emerge'. (27) Mohawk incorporatedsource:Sonny Edwards,speaker(p.c.) Tonsahahnefuitka'we'. w t- onsa-ha-hnek-otkq' -e' DUPLICATIVE-FAcT.REP-M.sG.AGT-liquid-emerge-r 'He came back up out of the water'. In many cases,however, t}te meaning of the verb does not lead to an interpretation of a specific semanticrole. The incorporated'palm' in (28) could be labeleda sourceor a goal, but no formal distinction is marked in the Mohawk. (28) Mohawk incorporated source or path: Joe Tiorhakw6n:te' Dove, speaker(p.c.) Idh tewati4:sen iah te+r-at-ies-en
not NECATTVE-NEUTER-MIDDLE-be.easy-STATIVI

not

is it easy

taionkwahsi6hsere'a-hsi -ohser-E' t-aa-ionkw 1.PL.PAT-palm-slip-P DUPLTCATTVE-OPTATMwould it hand-slip us 'It's not easy for it to slip from/through our hands = for us to lose our grip'. (29) Mohawk incorporated goal: fosephine Kaierfthon Horne, speaker(p.c.) ne Tekanwf:ta Wahshako' nikonhnita' ne Tekanwi:ta wa-hshako'nikonhr-l-t'a-' the Tekanawi:ta FAcT-M,sc/M.PL-mind-JR-insert-P the Deeanawida he mind inserted them tdnon' AiD:wate' tdnon' Aii:wate' and Hiawatha Hiawatha and

460

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

nahianitiohh5n:ni' ne aa-hi-an-itiohk-onni-' OPTATIVE-M.DU.AGT-MIDDLE-group-make-P the for them to group make ne ne

the

Kaianere'kri:wa. ka-ianEr-e'=kowa
NEUTER-IAW-NS=AUG

the

great law

'He inspired Deganawidaand Hiawatha to organizethe League'. Placesmay also be evoked as semanticfeaturesof verb roots. Some verbs contain an indication of location or direction as an inherent part of their meaning.There is, for example, a verb root -o 'be in liquid', which forms the basis of several derived stems. (30) Mohawk (3Oa) -o-o-hw-o-kw verb stems involving water 'be in liquid' 'put in liquid' (with cAUs$NE-r]t), 'immerse' 'fish out' 'take out of liquid' (with REvERsIVE-kli'),

(3Ob) Wa'd:kghwe' wa'-iak-o-hw-n' liquid-CaUSATM-P FACT-FLSG,AGT-be.in. = she immersed (it) she put it in liquid 'She put it in water'. Another verb root -nhont- means 'having in the mouth'. (31) Mohawk verb stemsinvolving the mouth (3la) -nhont'have somethingin mouth' -nhonta' 'put somethinginto mouth' 31b) Wakdnhonte'. wake-nhont-r' l. sc.plr-have.in.mouth-sTATrvE 'I have it in mv mouth' (cigarette, pencil sticking out that one is chewing on). (3lc\ Wa'khenh6nta'. wa'-khe-nhont-'a-' FAcr- 1.sc/3-have.in.mouth-cAUsATrvE-P 'I put it into her mouth'. In modern Mohawk, the stem-nhont- 'have in mouth' is no longer segmentable. There is evidence,however, that it was originally formed from the verb

VARIATION IN THE IDENTIFICAIION

OF PARTICIPANTS

461

root -ont- 'stick out, be attached' with an incorporatednoun root -nho-- This noun root is not used in the modern languagewith the meaning 'mou(h', but it does resemble the noun rool -nhoh- that serves as the basis of the noun ka-nh6h-a' 'door' (Neuren-door-NouN.suFFlx).It also matches an element of the verb se-4fu!:-ton 'close the 5!gq1',which appearsto have been formed by incorporation of a noun root -nho- into the verb -ton 'cover': 2'SG'AGTopening-cover.The modern verb root -nhont- can now incorporate another noun. (32) Mohawk 'mouth' verb with incorporated noun: Joe Awenhrdthen Deer tehatikwdnie's Idh wi' te-hati-kweni-e's i6h wahi'
NOt TAG NEGATIVE-M.PL.AGT-bE.AbIC-HAB

not

you see

were theY able

akonwaristanh6nta' a : -konwa-fi911- o nt-' a-' nh SG-metal-have.in.mouth-cAUSATM-P 3,PLIZOrC. OPTATTVEthey would metal-insert.into.mouth her ki: ki:ken this this akohsd:tenhs. ako-hsaten-hs FI.Pet-carrY.on.back-HeB horse

'They weren't able to get the bit into the horse'smouth, you see'. 6. Nominal adjuncts. So far, we have seenthat the participantstypically identified in oblique or adjunct nominals in other languages-namely, companions, associates,recipients, beneficiaries, instruments, sources, goals, and locations-are expressedin alternative structuresin Mohawk: as core argumentsof simplex verbs, as core argumentsof applicative verbs, as components of core arguments,as incorporated nouns, as directional affixes, or as semantic elements of verb roots. Independentnominals that are neither core argumentsnor incorporated do occur, however, to identify places. (33) Mohawk source nominal: Wilfred Jaimison, speaker tahahtdn:ti'. Michigan ta-ha-ahtenti'' Michigan place-name cIsLocATrvE-M.sG.AGT-depart-P he departedthis way Michigan 'He came from Michiean'.

462

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

(34) Mohawk goal nominal: Joe Tiorahkwdn:te' Dove, speaker kt:ken, Tseid:ta ki:ken ts-ie-ia't-at RIP-Fr.AcT-body-be.one.sTATIve this this one woman Sharhd:'on Sharhd:'on place-name Chateauguay iahshakoti'teron'. i-a-hshakot'i'tnron'' TRANsLocATTvE-FAcr-3.PIIF'sc-take-P they took her awaY

'They took t}is one woman 191Q!3193ggggy" (35) Mohawk location Thds tho sa's there formerlY there formerlY tieruikere' t-ie-naktr-e'
CISLOCATIVE-FI. SG.AGT-TCSidE-STATIVE

nominal: Joe Awenhrdthen Deer, speaker nonkwd: n=onlcwd:ti the=side the side Kan6n:no. New York CitY

over there she resides 'She used to live over there in New York City'. These place nominals are the closest structuresin Mohawk to obliques in The sentences (33), (34), and (35) would be acceptablewithout them: 'He came', 'They took her away', 'She used to live there'. Many of thesewords show one of a small set of endings (Mithun 2001). (36) Recurring endings on place nominals =okon =akon =(d:')ke =hne =akta' ={hen =kehson =konhson etc. =( a\ti At first glance, theseendings might be taken for locative casemarkers' Several facts, however, indicate that they do not represent inflectional case. Apart from the pair =(a)'ke/=hne, the various endings listed in (36) are not phonologically or lexically conditioned altemants: different endings can be attachedto the same nominal. The different endings do not distinguish the roles of the nominals in the predication, that is, the relation between the place and the event or state.Nominals carrying the ending ='ke, for example, can designatesources,goals, or locations' (When a stressedsyllable ends in glottal stop, the tone is dragged down and the glottal disappears,leaving vowel length.)

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

463

(37) Inflectional case? Karonhid:ke nithawi:non. Awin:ke ienid:kohwe'. Ahskwd:kenitskote'.

'He came down from heaven'. 'She will put it into the creek'. 'They're sitting on the porch'.

The endings are derivational nominalizers that are added to nominals (whatever their internal morphological structure) to create terms for places. The various endings create terms for different points or areasin relation to the base.The most general is =(a)'ke/=hne 'place'. Others are =akta'place neat', -(a)kon'place inside', =okon 'place wder', =ti 'placebeyond', =ihen 'place in the middle of' , =kehson 'place through or around', and =konhson 'place among'. These endings are an integral part of many place-names. (38) Some place-nameswith place nominalizers ('original ax-makerplace') :k onwb e ' F ra n c e ' O'seronni' ('old country place') Onhontsiakaibn : n e ' E u ro p e ' Ranatakaridhsne'Washington' ('towndestroyerplace') They are word-formation devices rather than obligatory ilflectional case markers. Many place-namesdo not contain any of them. (39) Some place-nameswithout place nominalizers ('1og in water') 'Toronto' Aterdnto ('pail in water') 'Ottawa' Kand:tso ('beautifullake') Kaniatari:io'LakeOntario' The derivational nominalizers remain in words for places even when these nominals are functioning as core arguments.The fact that the markers do not below. The specify syntacticfunction can be seenby comparing the sentences derived place nominal aw?n:ke,literally 'body.of.water=place',functions as a core argumentin (40a). It could be the answerto a questionlike Nahd:ten' enhsdten'nik)nhrdtsheke' 'What will you guard?' (with appropriateshift in word order for pragmatic purposes).The same word awDn:kelocates a state in (40b), specifies a source in (40c), and identifies a goal in (40d), with no changein form. (The term awin:ke is usedfor bodies of water like lakes, rivers, and ponds, rather than for water coming out of a well ol faucet, or water in a cup. The speakerwas referring to the Saint Lawrence River.) Mohawk awdn:ke 'place at a body of water': Sonny Edwards, speaker(p.c.) (40a) Core argument i:'i n2:'e enkaten'nikonhrartitsheke' i:'i ni:'e en-k-aten-'nikonhr-,t-rat-hs-ek-a' I itis Fur-l.sc.MIDDLE-mind-JR-put.on-HAB-coNT-P I it is I will out mv mind on it continuallv (40)

464

INTERNATToNAL JOURNAL oF AMERICAN LINGUISTICS

nawan:ke.. ne awen='ke the=water=place the water place 'I will continueto guard the water'. (406) Location ne N6: id:ken' it is one says the awin:ke awen='ke
.\1,.21s1=place

tsi ndn:we' to ever

enhen'terbn:take'. en-ha-i'terion't-ak-t'
FUT-M.SG.AGT-reSide-CONTINUATIVE.P

water place

he will live

'And so, they say,he will live in the water forever'. (40c) Source Enskatkonhsotdhsi' en-s -k-at -konhs-otah si -' nawin:ke; ne awen='ke

FUT-REP-1.sc.AcT-MIDDLE-face-appear-Pthe water=place the water place I will show my face again

'I will reveal my face again from the water'. (40d) Goal AwDn:ke awen='ke water=place water place ni:' ne i:'i the=myself I myself

nienhdnske'. n-i-enhens-k-e-'
PARTITTVE-TRANSLOCATM-FUT-REP- I . SG.AGT- go-P

I will go back there 'I will return to the water'.3 Further examples of place nominals functioning as core argumentscan be seenin the excerpts from conversationbelow.
3In part because of the absence of case marking, the precise syntactic alrd semantic relations of some of these forms could be intelpteted differently. One could argue, for example, that if the criterion for core argument status is obligaroriless, the location in (4Or) 'He will Iive in the water forever' is a core argument, since the vertt 'reside, dwell. live' normally occurs with a specification of place. One could also argue that the semantic role of the water in (40c) 'I will reveal my face again fiom the water' is vague, indetermi[ate between location and source. It is listed here as source simply on the basis of the translation given by the lipeaker.

VAR1ATION IN THE IDENTIFICATION OF PARTICIPANTS

465

(41) Place nominals as core arguments: Nicholas and S. Phillips, K. (p.c.) speakers Nicholas Kanerahtenh6:wi wd:kehre' Ne: ki' wa'-k-ehr-n' ne: ki' l it.is in.fact FACT- .sc.Acr-want-P I wanted it is in fact akon ri' w andn: tonhse', a : -kon- ri' w anonton-hs E' OPTATIVE- 1.SG/2.SG.ASK-EEN.EPPLIC-P

I would ask you Kanehsatd:ke ka-nehsat=a'ke 11-y=place (place-name) iehsient4:ri ie-hs-ienteri TRL-2.sG.AGT-know you know !1 there se' se' indeed indeed wahi' .. . wahi' TAG right?

As a mattef of fact, what I wanted to ask you was whether you know Kanehsath:ke'. Sha'tenkenhrltie'Phillips ni' N6: ki'k ne i:'i ne: ki'=k it.is in.fact=just the I it is in fact just myself nitient6:ri.... kand:takon ni-k-ienteri ka-nat=tkon 11-1su.'n=place.in PRT-l.SG.AGT-know I know it town 'I mvself iust know the village' Kanerahtenhi4:wiNicholas kahehtd:ke? Khne ok--ne ka-heht=a'ke and the t-field-place n.nd what about the countryside [the surrounding area]'!' (42) Place as core argument: Mary McDonald, speaker (p.c.) wahsekhrti:ri' ka' ndn: [.,.] tanon' tsi wa-hsek-hrori-' ka' n6n:we tanon' tsi how FAcr-2.sc/l.sc-tell-P what place and

466

INTERNATIoNAL JoURNAL oF AMERICAN LINGUISTICS

ohr6n:wakon o-hronw=Akon Nguren-ditch=place.inside ditch itken? Lt-ka-i

PROTHETIC-CISLOCATIVE-NEUTER.AGT.bE. thETE

it was there '[Rememberhow it was snowing] and how you told me where the ditch was?' A further indication of the derivational status of the endings is the fact that the bases of many place nominals no longer exist in the language on their own- The analysisof awbn:ke'water place' is clear,with its nominalizer ='ke, blut d:wen' is not a word in Mohawk or in its close relatives, though its cognatesstill serveas the regular word for 'water' in the more distantly related Tuscaroraand Cherokee. As the products of derivation, place nominals do not always have meanings that are the exact sums of their components.Like all results of wordformation processes,they are derived for specific purposes, as labels for particular concepts. As lexical items, they can undergo further semantic development on their own. (43) Some idiomatic terms for places (43a) ohson'karlle o-hson'kar=gik
*gg.1gp_!sa1d=place

lit., 'board place' > 'floor' (43b) aten'in:rqftgn aten'enhr-tkon fence=place.inside lit., 'place inside the fence' > 'yard' (43c) wastonhrondn:ke waston-hronon'= lk B oston=resident=place lit., 'Bostonianplace' > 'United States' (43d) ohonts6:kon o_onhontsi=9lg! NEUTER-earth=place. under lit.. 'place under the earth' > 'cellar'

VARIATIoN lN THE IDENTIFICATION OF PARTICIPANTS

467

Furthermore, nominals derived with the locative nominalizers can serve as the input to further derivation. (44) Place nominals as input to further derivation (44a) karonhig;ke ieronhia'kehr6:non' ie-ronhi=a'ke=hronon' ka-ronhi=a'ke Et-sky=place=resident wruten-sky-place 'sky place, the heavens' (44b) kand:tgkon ka-nat=Abg.L NEuTER-town=place.in 'residentof the heavens'> 'angel' kanatakonhr6n:non' ka-nat=Akon=hronon' NEUTER-town=Pl399j!=I9$C9!(

'town, village settlement"villagers' It appearsthat the locative nominalizers originated as verb roots that incorporatedthe noun stemsthey now appearwith. In two of them, =kihson'here and there' (as in owisa'=kdhson '[running] here and there over the ice') and =Mnhson 'here and there among' (as in karha=k6nhson 'frunningl through the forest, among the trees'), one can identify a distributive suffix that usually appearson verbs. At present,however, most cannotbe used on their own as predicates. As in many languages,the constructionsused to identify places are also used for times, whatever their syntactic role in the clause. (45) Mohawk core time nominal: JosephineKaierfthon Horne, speaker(p.c.) 6n:we' Kanenna'kd:ne t...1 wa'-w-e' ka-nenna'ke=b!9
NEUTER-AUIUMN=PIACE FACT-NEUTER.AGT-gO-P

'AC$!qq came [and it was time for the family to move on]'. (46) Mohawk adverbial time nominal: JosephineKaierfthon Horne, speaker(p.c.) Kenh,vitD:ne

ka-ikwite=fu9
N-spring=place tentitewakdha'. t-en-t-itewa-ko-ha-'
DUPLICATM-FUT-CTSLOCATMI.INCL.PL.AGT-get-PURP-P

'In the spring we will come back to get it'.

468

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

As noted, though independentnominals may designateplaces and times, there is no indication on the nominals themselvesof their syntactic or semantic roles. Their semantic roles can usually be inferred from material in the verb. In some contexts, the directional prefixes found on verbs provide disambiguation.With directed motion verbs containing a Translocative'thither, away', the place term is usually identified as the goal of motion by default, since the sourceis the deictic center,the presentlocation. (47) Mohawk Translocativemotion verb: Watshenni:neSawyer, speaker(p.c.) Sok n6n:wa' tsi iehiatonhseratah|wdhtha' sok ndn:wa' tsi ie-hiaton-hser-'c+'a-hkwaht-ha'
SO NOW At FI.AGT-WTiIE.NMZR-JR-iNSEIt-INSTR.APPLIC-HAB

so

no

at

one puts letters in with it

ionsaiakwakwdtho'. !- onsa -i akwa-kwatho-' l .EXCL.PL. ACT-stop.by-P TRANSLOCATTVE-REP.FACTwe went back awav 'So then we went back to the post office'. With motion verbs containing a Cislocative 'hither', place terms are conventionally identified as the source of the motion by default, since the goal is the present location. (48) Mohawk Cislocativemotion verb: CeceliaPeters,speaker(p.c.) Ahhttesdhsne [ewakahtdnkion. Ahhuesdhsne te-wak-ahtenti-on cIsLocATrvE-l.sc.PAT-go-STATIVE place-name place-name I have come lggg 'f 've come from Ahkwesdhsne'. These examples also illustrate the contribution of verbal semantics to inferencesabout the semanticrole of place and time nominals. Place words that occur with verbs like -en'n- 'happen' or -naker- 'reside' are likely to identify locations. Those that occur with verbs l1ke -otka'w- 'emerge' or -niiontah,v 'remove' are likely to designatesources.Those that occur with verbs like -ataweio't-'enter' or -ta' 'insert' are likely to name goals. Those that occur with verbs like -atawenhrat- 'put self over' (as in wa'thatenhrawen:rate' 'he fence put himself over' = 'he jumped over the fence') arelikely below. to name paths. The role of such inference can be seenin the passage There is no overt specification of the roles of the steeple or the beam, but they are easy to infer.

VARIATION IN THE IDENTIFICATION OF PARTICIPANTS

469

(49) Mohawk verbal semantics and inference: JosephineHorne, speakbr(p.c.) Wahathwi st anii ontd : ko' i w a-hati -hwi st -e-nii o nt- e-h,v -' g-JR-REvERsrvE-P FAcr-M.pL.Acr-metal-JR-han they metal unhung (it) ononhsatokenhti:ke o -nonhs-t-to kenhti=' ke NEUTER-house-Jn-be.holv=place Holy house place kahnia'sd:ke; ka-hnia's=a'ke
NEIJTER.INALIENABLE.PoSSESSoR-neck=place

its neck place karontd:ke ka-ront=a'ke NEUTER-log=place log place w ahatihwdnerenke' . wa-hati-hwanerenk-n' FACT-M.PL.AGT-tie-P they tied (it)

'The bell was taken (from) the church steeple and fastened (to) a beam'. Placesand times appearto be the only kinds of participants that are identified in nominal adjuncts in Mohawk. These are also the only participants that can characterizethe situation as a whole, locating not just the action or statein spaceand time but all of the participants as well. It should be noted, however, that the differencein semanticscopedoesnot correspondperfectly to differencesin formal expressiontype: within the verb versus outside the verb as adjuncts. 7. Implications. As we have seen,most of the pafiicipants that would be identified in obliques or adjuncts in many languagesare identified in other structuresin Mohawk. Associated persons or objects, recipients, beneficiaries, instruments, and some places are identified in several different ways: as core argumentsof simple verbs, as componentsof core arguments,as core argumentsof derived applicative verbs, as incorporated nouns, with the aid of directional prefixes on verbs, or aspart of the lexical semanticsof the verb stem. Their roles are specified entirely within the verb. These are of course prototypical head-marking structures.

470

INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

Structureslike these are not uncommon cross-linguistically, particularly in the Americas. At first glance, they might be dismissed as simply an arbitrary formal alternative,asreflectionsof a single typological parameterwithout further interest. A closer look suggeststhat the formal locus of marking can have deeperimplications. Such marking can affect the ratio of predicatesto nominals- When each participant is introduced as a core argumentof its own clause,the predicate/ argument ratio rises, along with the clause density. When participants are evoked by meansof incorporatednouns or semanticelementsof verb stems, ratio rises again. there are fewer argumentsand the predicate/argument The impact of such marking can affect the lexicon. In languageswith robust oblique constructions,the inflectional casemarkers and/or adpositions that identify them tend to be fully productive or nearly so. There is an expectation that every noun will have forms in all casesor can be combined with every semantically appropriate adposition. If there is a noun meaning 'styrofoam,' for example, and the languagehas inflectional instrumental and locative cases,we would predict that there are forms meaning 'with styrofoam' and 'in styrofoam'" There is also an expectationthat the combinations will, with perhapsonly a few exceptions,be semanticallytransparent.If we know the meaning of a noun and the meaning of the instrumental casesuffix, we can predict the meaning of the noun in the instrumental case.The properties of full productivity and semantic transparencythat we expect of inflection do not characterizemost of the alternativesto obliques that we find in languageslike Mohawk: the applicatives,noun incorporation, and lexical semantics. As we saw, Mohawk has three types of applicatives:benefactives,instrumentals, and directionals. The benefactive applicatives add an experiencer, recipient, or beneficiary to the set of core arguments.They occur with many different verb roots, but speakershave clear ideas of which benefactive applicative verbs exist and which could but do not. Based on the verb stem -atenhninon 'sell' there is a benefactive applicative srem -atenhnino-J'sell 'Sell it to me!'. Basedon the verb stem -koha'go to': Tah,vatenhnt:non-\ get' there is a benefactiveapplicative stem -koha-'s'go get for': Takk6ha-'s 'Go get it for me!'. Based on the verb -o'lcwat 'dig', however, there is no verb 'dig for', though the combination would make sense.Digging for someone else's benefit has simply not been a concept that speakershave thought name-worthy.The secondset of applicatives,the instrumental applicatives, are highly productive but primarily within one domain: the creation of new know the terms for items describedin terms of their functions. Still, speakers part of the landifference between those instrumental applicatives that are guage and those that could exist but are not in use. The third set, the direc-

VARIATION IN THE ]DENTIFICATION OF PARTICIPANTS

4',71

tional applicatives, are ubiquitous, becausethey occur on highly frequent motion verbs like 'go', 'run', 'fly', 'climb', 'chase','crawl', and 'carry', but they are no longer productive at all. The differencesin productivity between oblique casemarkers and applicativesreflect the fact that the recipients,beneficiaries, instruments,sources,and goals of some activities are highly relevant and frequently mentioned, while those of others are notNoun incorporation shows a similar range of productivity and transparency. Some verbs always incorporate,some usually do, some often do, some rarely do, and somenever do. Somenouns occur only incorporated,some are usually incorporated, some often are, and some never are. Speakersknow which noun-verb structuresexist and which could but do not. Stems containing incorporatednouns also show a continuum of semantictransparency. from fully transparentto highly idiomatic. The notion of giving someone not good luck, for example, is expressed, surprisingly, with the semantically transparent compound stem -ra'sw-awi- 'luck-give'. Being bored or frustrated, on the other hand, is expressedwith the less transparent stem -ate' : tekateien' talcdrtahs'l am ien' t-a-kari-'MIDDLE-awareness-JoINER-bite' bored'. The lexicon shows even less predictability. We saw that Mohawk contains verb roots -o 'be in liquid' and -nhont- 'have in mouth'. Though we may have some hunches, we cannot predict whether Mohawk will have other rools that include places as part of their meaning, nor, if they do, which places they will be. Such differenceshave implications for language documentation.Though comitative, benefactive,and instrumental nominal adjuncts do not generally occur in unscriptedMohawk speech,they have, on occasion,appearedwhen speakers were trying very hard to render close, literal translationsof English sentences. documentation consists principally of elicited translations of If from a contact languageby strong bilinguals, differences isolated sentences like those discussedhere can be obscured.Differences between the oblique structu(es in languages like English and the alternative structures of languageslike Mohawk go beyond the locus of formal marking. They point to pervasivedifferencesin the lexicon- The lexicon provides a repository of the conceptsthat have been expressedthe most often over the course of develthat opment of a language.If we attempt to describeform without substance, we will be missing is, structure without the lexical material that carries it, from part of the essence the language.Translationsof isolated sentences of a contact language can certainly be helpful in getting one's bearings at the outset, but they are usually lexically impoverished. The rarity of comitative, associative,benefactive, and instrumental adjuncts in spontaneous Mohawk speechsuggestsnot only that languagescan

472

INTERNATIoNAL

JoURNAL OF AMERICAN LINGUISTICS

differ in the relative burden carried by core arguments and other constituents, but also that simple differences in form can lead to important differencesin substance.
REFERENCES DR'YER, MATTI$W. 1986. Primary objects, secondary objects, and antidative. Language 62:808-45. MnHuN, MAFJANNB. 1991. Active/Agentive case marking and its motivations, Language 67:5lO-46. 2001. Actualization pattems in grammaticalization: From clause to locative morphology. Actualization: Linguistic Change in Progress, ed. Hendng Andersen, pp. 143-68. Amsterdam: Johl Beniamins.