PACIFIC LINGUISTICS

Series C - No. 85

VERBAL CONSTRUCTIONS AND

VERBAL CLASSIFICATION IN NATAO RAN-AMIS

by

Teresa M. Chen

Department of Linguistics

Research School of Pacific Studies

THE AUSTRALIAN NATIONAL UNIVERSITY

Chen, T.M. Verbal constructions and verbal classification in Nataoran-Amis. C-85, iv + 300 pages. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-C85.cover ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. PACIFIC LINGUISTICS is issued through the Linguistic Circle of Canberra and consists oT four series:

SERIES A - Occasional Papers SERIES B - Monographs SERIES C - Books SERIES D - Special Publications

EDITOR: S.A. Wurm

ASSOCIATE EDITORS: D.C. Laycock, C.L. Voorhoeve, D.T. Tryon, T.E. Dutton

EDITORIAL ADVISERS: B.W. Bender K.A. McElhanon University of Hawaii Summer Institute of Linguistics David Bradley H.P. McKaughan La Trobe University University of Hawaii A. Capell P. MUhlhausler University of Sydney Linacre College, Oxford Michael G. Clyne G.N. O'Grady Monash University University of Victoria, B.C. S.H. Elbert A.K. Pawley University of Hawaii University of Auckland K.J. Franklin K.L. Pike Summer Institute of Linguistics Summer Institute of Linguistics W. W. Glover E.C. Polome Summer Institute of Linguistics University of Texas G.W. Grace Malcolm Ross University of Hawaii Australian National University M.A.K. Halliday Gillian Sankoff University of Sydney University of Pennsylvania E. Haugen W.A.L. Stokhof Harvard University University of Leiden A. Healey B.K. T'sou Summer Institute of Linguistics City Polytechnic of Hong Kong L.A. Hercus E. M. Uhlenbeck Australian National University University of Leiden Nguy�n Dang Liem J. W.M. Verhaar University of Hawaii Divine Word Institute, John Lynch Madang University of Papua New Guinea

All correspondence concerning PACIFIC LINGUISTICS, including orders and subscriptions, should be addressed to: The Secretary PACIFIC LINGUISTICS Department of Linguistics Research School of Pacific Studies The Australian National University G.P.O. Box 4, Canberra, A.C.T. 2601 Australia Copyright © The Author First Published 1987 Typeset by Anne Rees Printed by A.N.U. Printing Service Bound by Adriatic Bookbinders Pty Ltd The editors are indebted to the Australian National University for assistance in the production of this series

This publication was made possible by an initial grant from the Hunter Douglas Fund

ISSN 0078-7558 ISBN 0 85883 331 X TABLE OF CONTENTS page CHAPTER 1· INTRODUCTION TO THE AMISLANGUAGE 1 1.1 General Information 1 1.2 Linguistic Affinity 6 1.41.3 OrthograReview ofphy Lit anderature a Brie f Phonological Description 1417 CHAPTER 2: FRAMEWORK OF DESCRIPTION 26 2.22.1 LeLexicasexicase -FeatPre-ureands and Po Rust-1979les Model 2630 2.42.3 BasicConstra Notionsints on of the Case Mode l 4138 CHAPTER 3: CASE RELATIONS IN AMIS 58 3.1 The Patient Case Relation [ +PAT ] 59 3.2 The Agent Case Relation [ +AGT] 70 3.3 The Instrument Case Relation [ +INS ] 78 3.4 The Locus Case Relation [ +LOC ] 93 3.5 The Place Case Relation [ +PLC] 102 3.73.6 TheOther Tim Case-le Caseike Re Notionslation [ +TIM] 107113 CHAPTER 4: AMIS CASE MARKING SYSTEM 123 4.24.1 CaseLocal Paradigism in msLe xiincase Am is 125123 4.4.34 SuLocativemmary of Re Caselator-Mark Nounsing System 160139 CHAPTER 5: CLASSIFICATION OF AMIS 163 5.0 Verbal Classification 163 5.25.1 ThePrimary Role of InClassesflectional - Subcateg Featuresori sation in terms 164 5.3 Subcategof Caseori Frsatioame nFeature in termss of Semantic Features 210l7l CHAPTER 6: SYNTACTIC DERIVATION OF AMIS VERBS 235 6.1 Inter-Sentence Relationship 235 6.36.2 Verbs Derived from NounsOther Verbs 250238 AUTHORBIBLIOGRAPHY INDEX 291274 TOPLANGUAGEIC INDEX INDEX 294293

iii Chen, T.M. Verbal constructions and verbal classification in Nataoran-Amis. C-85, iv + 300 pages. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-C85.cover ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. CHAPTER 1

INTRODUCTION TO THE AMIS LANGUAGE

1.1 General Information The1966) Ami ons thelanguag islae ndis of spoken Formosa. by 89,The802 peopleAmis, as(a ans of et hn1964ic ; groupsee We, iare and the Wang most numeroustotal Formos'a of alln abotheri aboginalriginal population groups. ofIn Formo termssa" ofco nstithe tutingsize of 38.the3% of the geograpgroups, hiccorningal areaaf ter they the inha Atayalbit, , thethe AmiBununs rank and fourth the Pai amongwan peo theple abo. rigThinalis can be exandpl haveained dis bypersed the fact settlements that the . otheTher threeAmis peopgroupsle , inhaon bitthe ottheher moun hand,tain inhabit region a sefairlyttlemen continuousts. area along the eastern coastal plain living in concentrated runningThe Amis between people occuthe pyCentral a long Range narr owand band the ofCoastal lowla ndRange, the of Tungtai Formosa, Rift extending Valley , Hengchfrom the'un, Ta kilreachingi Rive ther in southe the northernrnmost point part of 'PHual'ingtungien District. They tocal thel themcoastse lvesof the"Pangcaq" litera [ptureal)ts. a1hTheJ thoughname "Athemi" namayme come"Ami " fromor "Athmis"eir hasword beforcome the es tafirstblis pehedrson in plural exclusive pronoun: am i. Indeed it is highly probable that upon their themfirstselv encounteres and thus with given outsid theers imprtheyessi hadon usedthat thi it s waspronoun the name to reoffer the to tribe . Another(also known conjectu as Chrei-, mingfavoured Lee andespe Kiomincially Sabykata) my el, derlyis that informa the nament, Ki"Amis"lang Tacis io derived from their word for north or northerners: am is. Inhabitantsof the Talikitheir own Rive peopler basin. areAfter gene theral northernly called grou "Amip s camePangcaq" into orcontact simply wit "Ah mis"outsid byer s ofand thebe camecentral known and as southern"Amis", theregions name aswas we extendedll. Yuan to (1cover969) holdthe Pangcaqs a third groups view . He claims that the term amis meaning northerner was first used by the Puyumas Pangcaqto refe r groupto the inir Pina immedin orate T' northernaitung in ne turnighb ourpicked. Theup theborde teringrm to Southernrefer to Pangthemselvcaq esgroups. When as we thell. term was officially adopted, it was extended to other Leaving the credibility of these conjectures to anthropologists and specialists bothin folklor the peoplee, we chooseand th eirto uselanguage the name. Sinc"Amie s"the th laroughoutnguage dethscisribed work herto e,refer to Nataorapproprania, teis. a northern dialect, the term is not only convenient but also

1 Chen, T.M. Verbal constructions and verbal classification in Nataoran-Amis. C-85, iv + 300 pages. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-C85.1 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative. 2

121°E 122°E

25°N

24°N

o

23°N

o 25 50 KM.

22°N

Figure 1.1 Distri bution of formo san Nati ve Ethnic Groups (Adapted from We i and liang 1966 and Tsuchida 1976) LEGEND: A Atayal, B Seedeq, C Thao, D Bunun, E Amis, F Puyuma, G Paiwan, HM TsouRukai. , I Saisiyat, J. Kuvalan (extinct), K Pazeh (extinct) , L Yami, AMISGROU PS: (1) Nanshih, (2) Hsiukulan, (3) Coastal, (4) Pinan, (5) Hengch'un. [ Framed area is enlarged and shown in Figures 1.2 and 1.3.] 3 The Amis territory is divided into five regions by Japanese and Chinese plainanthrop, olothegists Amis . groupExtending consists from of norfiveth branch to southes: along(1) Na nshihthe easter Amis,n coastal The(2) poHsiupulatiokulann Am ofis, Nanshih (3) Coastal Amis is Amis 14,0, 12;(4) HsiuPinankulan Amis,, 29 and,8 65;(5) CoastHengchal,'u n19 Ami,849s.; Pinanand Wang, 28 have,056; notand givenHengch'un any re, 1,0ference20 (a nors of justifi1964; catisee Weon i forand thWangis subgrou1966). pingWei, asYuansuming (1969that ) does it mentionis common thatknowledge Japanese among scholars Formosa liken ethnoloIna (1916)gists had. Howevmade erthe, anddivis liionngui intostic fi diffve groupserences according (Yuan 196 9:8to geograp, 411). hicalLater distri Japabutionnese ethno, andlo gistscultural threelike Utsush groupsika, wanam ely,(193 4)the, AsaiNorth (1ern935), the, and Central Mabuchi and (1 953)the Southerndistinguish grouedps onl. yOn proposedthe basis a oftwo-way social divis structureion into and akins Northernhip orga andni sata Southernion, We i group(1961). has By Yuan's account (1969:8-9), the five groups of Amis are geographically distributed as follovls (see Figure 1.1) : 1. NANSHIH AMIS. The northernmost group, located in the neighbourhood of Kuvatodaylans's Hualien who moved City int, ohas Yi aslan its at neia lateghboursr date the and Seedeqs have been of Taroko absorbed and by the the Chcoastainesel sttheripre . fromInc Busurinluded inin th theis northgroup areto Ma20kutaai settlemen in thets , southlocated (see al ongFigure the 1. 2) . 2. HSIUKULAN AMIS. This group occupies the narrow strip of land irrigated by CentralHualien Rivander Coastal in the Mounnorthtain and Ranges Hsiu kulanof River. in Inctheluded sout h,in betweenthis grou thep are aresett Tsulementsngao-Atayal known asand Tavar Iwatanong, ofVataan Bunun, . Kivit, and Pairasun. Their neighbours 3. COASTAL AMIS. Separated from Hsiukulan Amis by the Coastal Mountain Range, fromthe Coas thetal oth Amierss. groupInclu inhadedbits in thithes eastergroup nare coas settletal mentsterrace known and asis Tingisolatedalao, Vakonnorth , toMakutaai south. , Tsavue, Tsiukangan, Pesielen, and Malaulaon, extending from 4. PINAN AMIS. Located in the neighbourhood of today's T'aitung City, Pinan Amtheis Pu isyuma also, Bunun referred, Pai wanto as, andT' aitRukaiung Amgrouisps. . ThThereis group are hasalso as coitslonies neig hboursof the se knogroupswn asins Valideanngao the Pina (Malan Amn)is, Toran area., TorekIncluded, Vavo kulin th, Pijois groupho, andare Kantsetslatlementsi. 5. HENGCH'UN AMIS. The continuous geographic distribution of Amis is Pininte anrrupted Amis isby cutthe offPaiwan from andthe Puyumarest andgroups forms south a pocket of T' inai tungHengch. 'u A n,subgrou towardp of Amtheis. southern tip of the island. This group is later referred to as Hengch'un Wetribei ands, mentionWang (196 that6:37), in, in the re gardsevent toeen migrationth centur y,hist theory Amiofs Fogroupsrmosan were abo rigforceinald theto movewest, toward and thethe Puyumaseaster n fromcoast the by souththe Seed. Wheneqs fromthe Hanthe noChrth,inese thearrived Bununs in fromthe eastreduced coast to ain narrthe owea bandrly nin ineteent Hualienh centuryand T'ai, thetung Am distriis terrictstory. Bywas that further time, PaiwaHengchns'u, n andAmi Rus kaiswas fothatrme d suascc essfa resultully ofcut frontthem ieroff movem froment thes byPina then AmisPuyum groupas, . 4

24°N

...J q Q; ,... � ;;e � UJ () ll.J () 0

() - ll., - () � Q..

23.5°N

Figure 1.2 Settlements of Nanshih-Ami s (Ami s names )

LEGEND: 1 Busurin, 2 Sumadas, 3 Varvaran, 4 Parek, 5 Tsikasowan, 6 Tsipaukan, 7 Pokpok, 8 Ridao, 9 Nataoran, 10 Vanaxo, 11 Manran, 12 Rinaxa, 13 Taokak, 14 Daurek, 15 Tsiwidian, 16 Tsirakaran, 17 Karoroan, 18 Patorongan, 19 Vakon, 20 Makutaai. RIVERS (Chind Ma-anese na River,mes): e a HuMukuaalien Riv River,er, b f ShoufungHungyeh River, gc Hs\>Janiukulanli River, Riv er. 5

24

<: oq: Lu (J 0

(J .... - -.... _---- lI.. H 1 - , } (J /� ,/ oq: ,;'" ""',, ' " • J Q.. -, ., ' 23.SON \ ...... 1 .. �!

Figure 1.3 Settlements of Nanshih and Hsiukulan (Chinese names) LEGEND: A Hsinch'eng, B , C Chi-an, D Shoufung, E Fungpin, KF HsiFunglulinin, LG WaKuanjungfung,, HM JuisuiChohsi,, NI NantYuli,'o Ju T'aiDistrictungt. District A, B, C, D, E = Nanshih Region; F, G, H, I, J = Hsiukulan Region NATAORAN VILLAGES: 1 Peich'ang, 2 Ich'ang, 3 Nanch'ang, 4 Taohsiang, 5 Yunghsing. 6 Nataoran-Amis, the northern dialect under investigation, is spoken in the Nanshihspeakers regionof this. dialAccordingect. Natato theoran-A cenmisus takensettle mentsin 1964 are, therelocated are in1, Chi-an813 County·inThe names ofthe th nearese visuburbllages of areHualien: Peic hCity'ang,, formingIch'ang,a Nanchcluster'ang, of Taofivehsia vinglla, gesand. Yunghsapproximatelying. Al fivl mye mi inlesforman fromts downtown came from Hualien the vi Citllagey (sofee Nanch Figure'a ng,1. 3, area C).

1.2 Linguistic Affi nity That belong to the Austronesian family has long since been esgeogratablisphicahedl (fpositionor details of , Formosee Dyensa in 19re71lab tionand Tsto uchidathe othe 1976r areas:1-2) . of TheSouthe ast Aslingiaui instic which po sitioAustrn oneofsia then Formolanguagessan languagesare spoken as isa wholeseen inin Fi thegure Austrone 1.4. siaThen Fofamirmolysa, n howevelanguagesr, is asnot an clear early. ofMostfshoot attempts or a ma atjor classi branchfica oftion the haAusvetr ploneacedsian Blustlanguag 19es77) (cf.. Haudricourt 1962, 1965; Dyen 1965a, 1971b; Dahl 1976; and

1.2.1 The Position of Formosan Aborigi nal Languages in the JI.ustronesian Language Fami ly The few diachronic studies to date, most of which are based on secondary afsourceter thes, suggecurrentlyst that accepted evidence reconstruction from Formosan oflanguages the ProtO-A couldustr sigonnifiesiancantly (PAN ) soundIndone siasystem.n group Cape of Aullstr (1962onesi) poan intslanguage out sthat conce an rnsou tstamorending detai problemled des witcriptionshin the of thesethe Fo larmonguagessan languag to thees , restthe irof comAustroparativenesia n.morpho logy, and the relationship Dyen's lexicostatistic study (1965a) of these languages indicate high internal ofdiv erlexsiticostay andtistics low inte is rnalto be and trusted external, Dyencognate 's obse percentagrvation escan. thenIf bethe conmethodstrued assettl foemellowsnts: . FormYetosa, mayin ahave paper been enti onetled of "Tthehe Posioldesttion areas of theof MalAustayoporonesialynen sian Languageslanguages of"probably Formosa" con (1sti963)tute, Dyena si nglehas anfami alterly" nativ(Dyene sug196ge3:st262)ion be thatcause Formosan muchindications more cl oselyfrom the re lavocatedbularies genetically are thanthat onethe mightFormosa thinkn languages. The evidare encin efact anditself homog isenei simplyty. notBy convi1971ncing Dyen enougseemedh tofor haveus tomoved take towarda stand the bet homweenogen diveeityrsity assu"likelymption to proveby conceding to be due that to theer rorhig hin div theersit procedurey among ... Form" (Dosayenn la19nguages71a:15) . is The Dyenlexicos thatta tisticone cannot proced heurelp wondehas lentring itse aboutlf toits so va manylidity diff. erent conclusions for According to Dyen (1965a), Formosan aboriginal languages form two separate grouclassipsfied: the asAta anyalic independent subfami lyprimary and the mem Eastber ofForm theosa Austn Hesroneionsia, withn lin Atakagyale apartic from(the otherMalayo-Poly one beingnesian Enggano) and as ouonetsid of e theof twothe Ocenon-r.1alayo-Polyanic area; andne siathen Eastgroups 135°E 1500E 0 Ryukyus CO NTINENT AL 00

Botel Tobago CHINA ; 0 . TAIWAN (FORMOSA) .Batan

o

PACIFIC OCEAN

15°N ",0 PHILIPPINES Guam o " o Q �VD�SJ 4\ .yap MICRONESIA

• o·

• .: danao Palau , Q 1= oCZJl/}... Peninsula t:> o o � o,1Moluccas MELANE SIA 0° a� � .. Cele es 9 0 0° c::> New o� 0 Ceram Britain \) Guinea � , 135°E C> <0 o 0

Figure 1.4 Map of Southwestern Pacific Area 8 distFormosainctn Hesubgroupssion pl acedof this under lin thekage Mal (sayoee -PolyneFigure sia1.5)n group. Aga asin one, Dyen of 'sthe pr oposevensal regardinglexicostatist theic external findings gen isetic sim plyrelatio notnship convi ofncing Formo. san languages based on Proto-Austronesian

Malayo-Polynesian Enggano AtayaSubfamilicly East FormoHesionsan

Figure 1.5 Dyen 's Proposal fo r the Position of Formo san Languages in (based on Dyen 1965a: 29, 39, and 53)

N.B.40 branches The num frberom ofProto-A branchesustr oneissia notn. nec essarily exact. Actually there are TsBlustouic ,(1 and977) Paiwapresentsnic branch a differingent off pictu as threere of indsubgroupingependent branches with Atayal of ic, AustroneFigure 1.6sia)n, . against a fourth branch labelled as Malayo-Polynesian (see Austronesian

(FAtaormoyalicsan) (FTsouicormosan) (FPaiwaormosnican ) Malayo-Polynesian weM-Pste rn CenM-Ptral EasterM-P n SH-NNG Oceanic ABBREVIATIONS: M-PSH-N: NG:Mal Southayo-Poly Hanesialmahenra-North New Guinea �

Figure 1.6 Bl ust 's Subgrouping of the Austronesian Language Fami ly (adapted from Bl ust 1977:2) 9 Except for the treatment of Atayalic, Tsouic, and Paiwanic as three distinct subgroushows theps , AuBlusstronet'ssia subgroun (AN)ping family is not at incthe ohignsiherstent lev witelh asDahl fol'slows (1973): which Austronesian

Western AN EasternAN

Figure 1.7 Dahl's Subgrouping of AN (Adapted from Dahl 1973:124 and Blust 1977:7)

AustBlustrone attemptssian fami to lyjusti by confy hissider picingtur thee ofevid geneencetic on re whlaictionshih PANps pronouns within theare morerecon primstructaryed. subgroups The bipar in Taiwantition intis basedo a Mala onyo 1)-Poly thene Sesiacondn su Aubgroustronep andsian one or PMP)Polit, enandess 2) Shift the fo(i.e.llowing, the paradig shift matof +-muic lev, butelli ngnot: +kamu from 2pl. to 2sg. in PAN PMP 1 (k)ami (k) ami we (excl. ) amen amen ours (excl. ) 2 (k) ita (k) ita we (incl. ) iten aten ours (incl.) In their introduction to the Myths and Tradi tions of the Formosan Na tive languagesTribes (1 93wit5)h, Ogawathe Ind andone Asaisian havelanguage mades. a comAccordingparison toof themthe Fo, rmothe sanlanguages aborigina ofl Thethe irForm listosa ofn ab cognateorigines words are closelylooks qui retela tedconvincing to the Indindoneeed,sia butn latheynguage do s.not Austindicaronete siahown lathisnguag rees.lation would stand against further comparison with other Sothe far,exte quarnalntitativ relatioe nsmethodship of havethe Formonot providedsan aboriginal us with la nguaga definites. e answeDyen r(1 as970) to, methodsamong oth mayer s,prove feels to thatbe usewhereful . quantitativ Detailede demethodsscriptiv havee st udfaiiesled of, qual Formoitsaativn e abstorigudiesinal and languagessubstanti woateuld any be clnecaimsess aryof intbeerforenal we and co exteuld makernal comrelaparationshipstive . Theto theso -csubgroualled pingquali oftativ Inedo-Eur methodopean as hasformu beenlated ap byplied Brugmann to the (1Austrone884) in siareln ation languages(1959) and by Pa wleya num ber(1966 of, li19ng67)uists have wi gainedth var muchying degreesinsight ofint osucc theess relatio. Gracenship 10 of(1973) Ocea, nic�'lalton languages (1977) by, Zorc applying (1974a, the 19qual74bit, ativ1977)e methodand Reid. (1Wo978alff , and1978 Wob,lf f1979) havesubgrou alsoping dem foronstrated Philippine the relanglevuagancees . of such a method to the problem of The qualitative method involves more than mere cognate recognition. It exforami exclusivnes the elyling shareduistic li stnguiructurestic offeatures the la nganduage common undeinnovatior considernsation. Ifto it se canarch besynt disaxcovered, and le thatxicon a issu bstaexclusivntialel ybody shared of featuresby some orin phoallnolog of they, morphoAustronelogsiay, n hypothelanguagessis in for Form theosase , languagwe are esthen. Suchin a posiqualittionativ toe stfoudirmueslate should a subgrou alsoping be exlinktendedage. to Moreovlanguageser, thereof the is Philippines Benedict's andAusto-Thai Indonesia hypothe to tesisst th(Benedicteir Form ·1966osan , 1967a, 1967b, 1975) to.be considered. Eventually explicit procedures would haveto be toAus betr outonelisianedn inand order test edto pinpointfor all languagesthe positio whichn of areForm knownosan orlanguag suspeescted whoamong wo thrk em.wit h �'1eMeanwsterhilen Aus, nottronehingsia nde andfinitiv Asiae ticcan languagesbe said. areHoweve payingr, morelingui andsts more(197 6)atten, Blusttion (1to977) Form, osaDahnl- languag(1976), esDyen, as (1evidenced965b, 197 in1, the1974) works, Pawley of Benediand Rectid Yamada(1976) , (1Re974)id ,(1 978and a,Zorc 197 9)(197, Sta7). nleyForm (1osa974)n , eviStardenceosta is (1 much974) , soTsughtuchida after (197 for6) , histsubgrouoricalping reconstruction hypotheses (c f.and Dahl for-the 1976 :ve124-12rifi6)cation. of various homeland and Asla nguagthe essubseq of uentthe Ph chaptilippineers mays. showEven, Amthougis his I grammaam notticallpursuing y very com siparativemilar to prestudsentedies in her theis may work prove, it tois hopedbe use thatful forthe liAmnguiis matesticrial cla ssicollecficatiotedn andof the Formoof evidsanence la nguageswhich Brasugmann well asreq proviuiresding for draforwing the Austronereliable siaconn clusfamiionsly theregarding kinds subgroupi(Brugmann ng,1884 nam:2el53)y,. "sound, flectional, syntactic, and lexical" evidence

1.2.2 The Position of Amis among Formosan Aboriginal Languages In 1936, Erin Asai in his article on the language of the Yami tribe on Botel Tobagoin Figure Is la1.8.nd, presented a classification of the Formosan languages as shown Accorof theding Amis to groupthis , cladistissinctfication from , BununAmis , stTsou-Paiwanands in isolatio, and othen as rsthe. onlyToge membtherer Saisthereiyat are; fiv(2) e thegrou Bununps: (1grou) thep; Northern(3) the Tsou-Paiwangroup incl udinggroup Ata incyalluding, Seed Tsoueq, , andSa aroa, Kanakanavugroup including, Paiwan Yami, Puofyuma Bote, andl Tobago Rukai; Is land(4) theand Am Batanis group; of the and Ph il(5)ip pinethe Batans (see eaFirlgureier 1.4) divi. sionBas ofica Formlly,osa Asain languages's divis ionint ocorresp twelveonds subgroups to Ogawa: Ata andyal Asai's, Seed eq, YamiSaisiyat. , Bunun, Tsou, Saaroa, Kanakanavu, Paiwan, Puyuma, Rukai, Amis, and Raleigh Ferrell (1966, 1969) has proposed a classification of the Formosan abodivisriginalions sugcultugestedres byinto Dyen three (1 9ma63jo, r19 65agroups, 196 which6). agreesAccording with tothe the ling uistic 11 Dyen-Fethe Tsouirrec,ll areclass lingifuiicationstically, two dist ofin thect threefrom eagrouch psothe, namr andely, from the otAtaheryalic Formo andsan grou"F3"ps (1. 963)Within, "East the Fo remrmoainingsan Hesion" third (1categ965a)ory, and, whi "Cchatenate Dyen haForms variosaousn" ly(1965c) called, butdivers whityich . hasThe beenir labelexicostalled astist "Picaiwa percentnic" byages Ferr showell that (1969) mem, betherers of is the much PaTobagoiwanic Isla groundps, are, wit linguih the stipocallyssible more exception closely of rethelated Yami to laonenguage another of Bote thanl to memthembeserslv es.of any Innon-For othermo wordssan groups, they assudespimete thethe laintnguagernales todiv beersitlongy togetheramong even characthough teveryrise lithettle grou evidencep. of common innovation has been provided to Subgrouping within the Paiwanic group is largely uncertain. On the basis of hisBunun lexi, Amcoisst, atiPuyusticma,al Pai stwanudy, Dyenand pos (1965a)sibly Kuvalansuggested and the Rukai grou pinginto ofone Thaolingui, stic hassubgrou madep, a andsubgrouping Pazeh and based Saisiya upont intothe anothereflectior. n Feof rrPrelloto-A, onustr theone otsiahern hand*t and, into*C in Pa theiwa Paniciwa I nicand languagPaiwanices . II, Bywit hish the te stfo, rmether Pasubgroupiwanic prgroupeser vingis subdi the vided distlanguagesinctio ninclu betweende Pa iwan*t and, Pu *Cyuma ahd, theRuk ailatter, Fav orlalosingng, Thaoit. , PaFerrzehel, l'sSais Paiyaiwat,nic I TaokasSiraya, and YamiHoanya.. ThougAmohng it the is Paiwa theorniceti callyII lang plauagesusible are forAmi Ferres, Bununll to, Kuvalanbase , nevhis eParthelessiwanic tooII subhastgroupiy forng him upon to a drasiwng concle commonlusions sound about change subgrou, itping seems from only paramone parameteeters tor. see Itif isth eirdesi resultsrable to convergeexplore . severTheal exact, if not gene alltic, possiblerelationship estain whichblished Amis. stands with the other Paiwanic languages remains to be providedFerrell also by archaattemptedeologi tocal subgroupfindings Form. Heosa nhas ab plaorigcedines ��i ons witculturalh Paiw evianicdenc IIe 1969l�guage:51) s thatwhose is speakers more si misharelar to the cultures so-called of PaLittcioralfic areasCulture outsid Complexe of Form(Ferosa.rell Matethatchedrial rocultofsure, astraits well assu chnon-mate as elevrialated culturehouses ontraits piles su withch as ba higmboohl ywal developedls and remcosminiscogonentic my ofths mo reand southernvast pantheons and east incluern dingareas poweof therful Pac spiifritsic. of the sea, are It is also pointed out that Amis oral tradition admits of arrival in Formosa over(Bat an)the, etcsea.. haTheve been name ments ofio ovnedersea, indis plcatingaces such stepping as Sa nasaistones, , Vaifsa noti, Vatathe n oriintrgodinucing, of mi hetgreratioogeneousns. Chanceelements migration into the might Ami s havestock ha. ppened again and again, Up to this date, the subgrouping of Formosan languages into Atayalic, Tsouic, andinto Paconsiwanicider atioseemsn geFenerallrre11' ys undisputedarguments for (cf. dividing Tsuchid aPa 1976iwanic:9-1 int5)o. PaiwaTakingnic I and Paiwanic II (Ferrell 1969:25) and for including Kanakanavu, Saaroa, and evensubgrou Rukpiaing inas Tsou shown (F errin ellFigure 1974 1.9.:8) I we have the composite picture of PaiwaThe internic nalserve res laastio merelynship ofa conv Paiwaenientnic lang labeluages for isFo unrmoclsanear lang. Asuage its nowthat staarends , deconsfiniidetionred neof itclahessifr Ataicayalictory norfeatures Tsoui c.such Reasli ab"verlebal clas infsiflection"ication a1flaias suggts estthe ed 12 Taikai-dialect Atayal Proper TaikBunsui-dialectO-dialect Proto Atayal< �Hakku-Bandaidialec-dialect t NortheGroup rn Seedeq______MuTaro sha-diko-dialecalectt < Saisiyat Northern-dialect BununGroup ---- Bunun�CentrSoutheal-rn-dialedialecct t

Tsou ______AriRuftu- san-didialectalect ProtoKanakanavu Saaroa- _-======KanakanavuSa aroa Tsou­ �Southern-dialect Paiwan proper�Central-dialect Group Proto Northern-dialect Rukai Proper Proto Rukai To romaki �MantauranTordukanu-Kongadavanu

______Northe rn-dialect AmiGroups ---- Amis -=:::::::::::::CentrSouther al-dn-dialectialect Batan Proper (Philippine Islands) Bat� GrouP \Ya mi

Figure 1.8 Asai 's Classification of Formosan Languages 13

tayal Atayalic ==== � A _-======�Seedeq _ Tsou Kanakanavu Tsouic � ======..-e §§ �� ======����Saaroa � Rukai Favorlang Saisiyat Pazeh Paiwanic I Thao Puyuma Paiwan Paiwanic Siraya Bunun 'Paiwanic II�------Amis Kuvalan Yami

Figure 1.9 Tree Diagra� of Formosan Languages (Adapted from Tsuchida 1976 :8. 15; Ferrel l 1969:69) byconstruction Ferrell (1 mar970)ker, ev sysidencetem (Lofi sound1973, change1978; Re(Fiderre 1978lla) 1969, and, Dyenthe use1971a) of , pronouns on(B lustthe ot197her7) . on Furtherthe one morehand, aand com aparative better knowledgestudy of ofAm isthe with Paiwa Phiniclippine language and s exIndoneternalsia nre lationshipslanguages shou of ldthe shed Pa iwafurthernic group light, if on it the is proa blegroupm ofat internalall. and Concerninganswered: (1the) whether subgrou pingthe so of-c Paiwaallednic Paiwa lanicnguag langes, uagetwo s qudoest formions a remaingroup andto be (2) if so, what is the membership of this group? 14 Amis has always been classified as Paiwanic, which was suggested by Dyen as beingPhilippine "perhas psand more Indon closelyesia, thanrela tedto Ata wi yal"th the (D yenlanguage 1965b:s of287) the. SouthBased, onin the regardingsecondary Amsourcis whiches, Is idcanore be Dyensumma (1rised965a) ashas fo madellows se: veral observations (1) HeAmisions belong of Malays too-Po a sublynegroupsia n ofla nguagthe Centrales. ThFormois san sub"Eagroupst Formo hassa nvari Hesioousn"ly , beenand labe"Catenatelled asForm "F3"osa, n" by TsouicDyen. subgrouIt is psdist. inct from the Atayalic and the (2) Amis stands in closer relationship to Thao, Bunun, Puyurnathan to, PaiwanPazeh and, and Sais possiyatible. Kuvalan and Rukai (3) proto-Amis proviphonemdeses the*Wl evidence, *W2, and for *N reconstructing for Proto-Austro thenesia n. In the main, Ferrell (1969) does not dispute Dyen's classification. However, hethan place withs AmiPaiwans in, Puycloseurnar , reRulakatioi, nshipFavor withlang, BununThao, KuvPazehala, n,Sa Siisraiyatya,, Taokasand Yami, and Hoanreflexyaes. ofAs Proto-Austrmentioned eaonerliersian , *tFerre andll *C based. his subclassification on the

1.3 Review of Literature Although records on the Formosan people and their language go back almost three hundredvocabul aryyears lists, ea, rlybr ieflingui collectistic workons ofis textslimited, and almost transl exclusationsive ofly theto Bible. Exnocept des cripfor tivthee analysshort isgrammat has everical beensketch pu blisof Amiheds ongiven Am is.in OgawaAmong and the Asa eai rliest(1935) , records(ca. 1630) is Reandv. ReGeorgv. Dae nielCandi Gravidiusus' Acc' traountnsl ofation the of Inhabi the Gospetants lsof of Formosa St. Matthew reviand st.sed byJohn Re v.into Wi Silliarayam Campin be166ll1. who Gravpubliiusshed' edi a tionnew Londonof St. editioMatthewn wasin 1888. CampDutchbe domill alsonation publis andhed miss sevionaryeral histinfluoriencecal. accounts of Formosa in the days of The earliest record of Formosan aborigines dates back to about 1250 A.D. Chaoof trade Ju-Kua and's contact Chu Fan betweenChih (A nnals"the Middleof Barbarian Kingdom" Peo andples) the is"ba chirbefarian"ly an peoaccountples outside of China. Most of the descriptions in Chu Fan Chih are exactly the same as Sui Shu Li u Ch 'iu Chuan which predates the former. So it seems that Histhe toryearliest of the records Sui Dyn ofasty) Formo instsanea abod, rigalthougines hare som founde scho inla rsSui debateShu (R ecagaiords/nst it. LinIn theCh' ien-seventeenthkuang lef cent ustury som Che inesetravel trave notesll ersconcer to ningFormosa the sucustomsch as of the Formosan people, the most extensive of which is the six-part Fan Su Li u K'ao Inv(Inquiryestig ationsinto thein Sathevage Taiwan Cus tomsSeas)) inof his 173 6T'a A.D.i Hai Shih Ch'a Lu (Record of During the Japanese occupation which lasted for about fifty years and ended wiethnologith the endcal andof World linguistic War II accounts, Japane seof scFormoholarssan produced aboriginal me tictribeulouss. 15 A Japanese-Amis Vocabulary was published in 1934, to be followed only a year lain terwhich by Ogawafolklore and was Asai re'scorded Myths inand tran Trascdiriptionstions of of the the Formo nativsane Nalanguagetive Tribess with Japasame neseyear tran witnslaessedtion theand pu brieblicf ationdescript of Utsurikaive accountswa, Miyaof motothe languag, and Mabuchies. The's volumgenealoges. ical In and1936 cla" Asaissific preatorysent edstudy his of fiv thee-g roupFormo classisan nativefica tiontribe ofs, Formosan in two aboriginal languag'es, placing Amis, in an isolated group., FormBothosa. the DutchThe Dutchand Japane traditionse traditions is preserved were byrevived missiona in poriesst Woinrld the War form II of vocabuldialectalary Amistudys-French and Bible and French-Amistranslation Di. ctiForonary the hasAmis been lang comuagpile, eda pan­for the dialectsof the Roman of JuiCathshuiolic, Yu Churchli, Kuang whicDfu , claFenglims ina , greatand Feng numberpin byof Reconvertsv. Antoine among Du risthe ,Amimanifs peopleested in. theIn worksJapan , ofa persChichiiistwenta (1intere960) , stKano in Form(1946osa, n195 et2)hnology, Kano andis Seanthropogawa (1logi956)cal, andst udMabuchiies of the(19 53,Amis 195 tribe4a, 19inc54blu, de19 60works, 196 by6) Chen. InChi Form-luosa and, Michae(1965)l, WangD. Coe Sung- (195hsing4), Chen (1961) Chi; -luWei (1Hw961ei-, lin1965 (1, 958,1968) 196, 5)Liu, WeiPin -hsHweiungi-li net andal WangTopics Je n-yingof their (1 96inve6)stig, Chaiation Chen-kang include (1material967), and cul Raletureig, h kinFerreshipll sys(1966tem , and196 9). socialIn the starearucture of archa, folkloreeological, art stu, dyand, immigrationportant histreseaoryrch ofhas the been Ami cars peoriedple out. by1966 Sung) and Wen-hsun Minze Stuivand erhis (Cteamhang (1and96 7)Stuiv ander by 196Chang6). Kwang-chih (1956, 1963, periodLinguistic. Other studies than of the Amis work alms ostby micamess ionato ria escom pletmentionede halt be infore the, posknownt-war pufour-pageblications text and and ma nusa vocabulcripts aryof byAmi Montgomerys linguistic (1 materi962) ofals an counidentinsist offied only a dialectam6ng twenty of Am Formois, Ferrsanel lal'snguages 336-word for comlistparativ (1969e) whichand classi includedficatory Malan-Amis purposes , aso mefew undatedpapers onou tlphoneines ticsand wordand phonemic list by patternsthe same byauthor Torjes. en There (196 4a,is also1964b) a and short304-word papers lis t (1by96 4a,Fey 1964b(1962)) onfrom morp a hologycentral. dialectTorjes en(Ch' doesengkung) not sp, pluecifs y twowh ich phodialectsnological of Ami featuress he record thated, it wasbut probabone canly tenotll Natafromoran the . vocabulary entries and Under the auspices of the National Taiwan University and the Institute of His196tory8-69 and Philologduring they, Academiasummerof Sinic 1971a, to the co llecpretsent dat authora on the was Am isin Taiwanlanguag e.in dialecFieldworkts, butmat erialsthe bulk were of colle materialscted onco Nallectaoranted was, Ki bitfrom (oNar taoran-AmiKiwit) ands. Malan I worprekedlimina withry reporttwo main and in aformants 75-pag e fortext over (C heneig ht19 69)mont werehs. produceAt thatd. time, only a The 1968 Aboriginal Languages Research Project of marked the beginningsince the ofdeath revived of Pro resefesarchsor Tungint erT'estung-ho and . activitiesThe Univers in Formoity ofsa Han wailinguii hasstic s threecontributed facult toy mem thebers recent, Dr deve. Fang-Kueilopment Liof , FormoDr. Stasan nllingey Stauisticsrosta in, andthat Dr, . at least AboLawrirenceginal Re idLanguages have done Re seafieldrchwork Proj ectin Taiwan have beenand sevdoctoraleral memstudentsbers of at the the 16 University. Dr. Paul Jen-kuei Li and Dr. Jeng Heng-hsiung have since their graduationand train newreturned students to . Taiwan A few to of conduct their morestudents resear havech onstudied Formo Amisan s liandngui atstic s theleastsis three (Wang term1976) papers have beenor reports produc ed.(Chu 1975a, 1975b; Wang 1975) and an M.A. -In 1972, when- Dr. Stanley Starosta of the University of Hawaii spent some time Bununin Tai, wanRukai, Saisiyat, Seedeq,-and Tsou. His work resulted in an insightful , he collected data on several Formosan languages including Amis, modepaperl on(Star causativeosta 1974 verbs) . Thein FormosanAmis dialec languagest that andStarosta refinement studied of is the also lex icase Nataoof theran-Ami Univers,sit they of diale Hawaict i deswascribed also in Taiwanthe pres forent a stusedy.cond timDre. Lawrencein 1972, hisReid prim1965ary, he incoterellestcted being word inlist Seedeqs in atsix that languages time. incDuringluding his Am isfirst. visit in Other works that refer to the Amis language include Ferrell (1971, 1972, 1974, 1965a,76) and1965b Stan, ley1971) (1, 974)Dahl de(1aling976) , withFerr morphologyell (1969) , andBlust syntax (197, 0,and 197 Dyen7), Foley(1963 , reco(197nstr6), Reuctionid (1 and978a su, bgrou1978b,pi ng.1979) , and Tsuchida (1976), dealing with historical In the Bibliography I have entered all the references mentioned above plus some syntacLiao 197tically4), Tsou re lev(Tungant workset al on19 64Atayal; Sta ro(Egstaer od1967 19, 6519, 69196; Li6a, 19721966b; Sta; Yamadanley 1974and , 1976(Sung; Tsu1966chida, 196 9)197, 6)Ruk, Saisai (Liyati 19 73(Tsuc, 1975a)hida 1964, Puyuma; Li 1975(Sprengb, er1978 19b)71, , Kanakanavu1972), Pa iwan Tang(Fer rell1958; 1970 Li , 19719725c,, 19197784)b), , BununSeedeq (W u(Asa 1969i 19; Jeng53), Pazeh1977) , (LThaoi 197 (L6,i, 1978Chenb), and Kavacumulantio (Ln i of19 kno78a,wledg 1978e b)on. FormIt osais nhoped syntax that to thedate pr onesent the onestu dyhand can andre flecthe t the othecurrentr. stage of development of the lexicase model (see Chapter 2) on the All significant papers and doctoral dissertations contributing to the devare eloworkspment by ofSt thearosta lexicase (1971 a,model 1971 b,are 19ente71cred, 19 in73 , the1974 Bibli, 19ogra75a,phy 1975b. , Inclu1976aded, Taylor1976b, 19(1971)76c, , 1976dLi (1, 97193)77,, Kul 1978,lavani 1979ajaya, 19(1974)79b, , 1979cLee ,(1 971982d4), , Cltoark ap pear(1978) b), and by Ikra(1979)nagara . (1980), DeGuzman (1978), Harmon (1977), Acson (1979), and Fagan Also cited are selected references to works of theoretical interest which incl1970,ude 19 Chomsky71, 1977) (1, 965Gruber, 19 70)(1965, Cook, 196 (17)971), Gunt, Diherxon (1(197979)5), , HalFillelmore (1973) (1968,, Hallid 1969ay, (1996467,, 1971970)5), HudKeenandlest andon Comrie(1970) , (1Hudson977), Cen(1976)a (1, 977Jacke, 19ndo78)ff, Hoppe(1972)r , andJespe Thomrsenpson (1980)974), KurodaMcCawley (1 965)(1968), La, koffO'Brien (1970) (1971), Leech, Pla tt(196 (19)971), Lyons, Plei (1ne968)s (1, 97Matthews6), and FormosanStockwell or et Ausal tr(197one3)sia, n evenlanguages though andthese li nguiworkssti maycs. not directly deal with For comparative purposes, I have also included some works on Philippine lingstudyuist ofics verbal, in partconstructionsicular, syntac and verbaltic stu diesclassif thatication are re. levIncantluded to the are pr: esent BloDeGuzmanomfield (1 970(1917), 197, Ca8)pe, Dyenll (1 964)(1970), Cena, Har mon(1977, (1 977)1978, )Hidalg , Constantinoo and Hid (1alg965o , (119970)71), 17 �ess (1967), McGinn (1970), McKaughan (1958, 1962, 1970, 1973), Mintz (1973), MiCenrikia (1tani979) ,(1 Re97id2) , (1Pa966wley, 19, Re78aid, 19and78 b,Sta 197rosta9) , (1Sc978)hachter, Ramos and (1otanes974), Ramos(1972) and, Stev1977)ens . (1969, 1973), Tchekhoff (1974), Walton (1979), and Zorc (1974a, 1974b, Since there is inevitably an intricate overlap and cross-referencing between Formolinguisticsan languages theory andand laothernguage Austron descriptionesian laonnguages the one on hand the , otheandr, between it se.ems almsectostions imp withoutossible runn to grouping into the problemsbibliog raphof icaldupli entriescation andin anyomissio meaningn. fulI have entthereforeries list simplyed in maintained the alphab eticala tradi ordertional of bibli the ograpauthorshical' lastfile nam whiches. has its

1.4 Orthography and a Bri ef Phonol ogical Descri ption 1.4.1 Consonant and Vowel Inventory Thesounds orthog whichraphy observes adopted the here biu niisqu basedeness on·criterion a phonemic of autonomousanalysis of pho Nanologtaoran-Amisy except in the case of (a) w, y, and q, and of (b) [?n], where maintaining this criterionof consonants would and result semi-v inowe a lsloss is preof sentgeneredalisati in Fionsgure (s1.10,ee be andlow ) an. invenAn intoryven toryof symvowelbols sounds in Fig isures given 1.1 0in and Fi gure1.1 1,1.1 phone1. ticCor resvaluesponding are togiven the inorthog IPA rainph brackic ets. t, k, b, d, s, h, 1, r, ng , w, Nataoran-Amis consonants are: p, q, c, m, n, and y. This system corresponds to that found in many manysuch Phas ilTagalippineog , laexceptnguages for, Nataoran-Amis the absence ofha 9s anda four-v the preowesencl syste emof c.conv entionLike ally i, a, [+ a], centrsymbolalised, re byla tivee,ly highand vowel 0, wi foundth the also e repre in Insentingdonesia nthe la pepetnguages . � a Stress falls on the ultimate syllable of a content word (, verb, adjective, etc.). Function words are often unstressed as in ato and, saan then , toay aLrautomeady,atically and no shifts of. toWhen the a lastsuffix syl islab addedle of tothe theaff stixemed form,. theFor st ressexam ple, tarakaw taLL plus -ay becomes tarakaway the taLL one while bot(ng fish plus Stre-an ssand inthe Amis redu ispli prcativeedicta preblefix. ba- becomes ba-bot i ng-an the fi shing pLace .

1.4.2 Syl lable Structure and Canonical Form of Disyl lables Syllable structure of Nataoran-Amis is (C)V(C). Most Amis roots are wordsdisylla witbich havingan initial the vowecanonicall are exformtrem (C)elyV (C)rare (C), V(andC) . phoneIttically must be thenoted initial that vosyllawelble is doealwayss not preceded begin wi byth a anglottal initial st op.vowel exIn cepta disyl whenla bicit iswo rdimm, ediatelythe second dependingfollowing ona (pthehonemi environmc) vowelent , ina theglid efirst or a syllaglottalble . stopIn is the always latt erin casertedse, betconsonaweennt. the adjacentExamples vow ofels Amis. disyllaHence, blesphone areticall giveny, allin Fi sylgurela bles1.1 2.be gin with a CVCIn my.V Cdata occur no. words with the form V.VC, V.CVC, VC.V, VC.VC, VC.CVC, CVC.V, or 18

Labial Dental-Palatal Velar Glottal stop vl p[ p,ph] t[ t,th] k[ k,kh ] q[ ? , ?11] b[ b] d[ d,0] vd v, �, h[ , h ,11 , ?11 ] s[ s , � ] x c[ t s , t � , t s h ] m[ m] n[ n ng[ Nasal ] '1 J Lateral 1 [ 1 ,1.] r[ r , ] Retroflex J Semi-vowel w[ w J y[ j ]

Figure 1.10 Nataoran-Ami s Consonants

Front Centre Back High i[ i ,I,e,e] o[ u,u,o,o] e[ Mid i, e] Low a[ a]

Figure 1.11 Nataoran-Ami s Vowels

V cv vc cvc V i a ama the fa ther toa sa b is b l d cv to o o a then two star moon ni vc oh they

cvc mi hca ci lmi n year sour

Figure 1.12 Nataoran-Ami s Di syl labl es 19

1.4.3 The Status of q, w, and y Words with initial smooth-onset vowels are extremely 'rare in Amis; and this fact is reflected by the gaps in row 1 and row 3 of Figure 1.12. Words like the following: [?owaj] rattan, [?ajamJ bird, and [?a?ht�id] sa lty I have analysed as having an initial phonemic q because even though they bear phonetic ama [?ama] [ama] ohn i [?o?hn i] similarity with words like � fa ther and � [?o?hni] they, the initial glottal in the formerset of words does not disappear with prefixation. Examples are mi-qoay [mi?owaj] to gather rattan and qoa-qoay-an [?owa ?owajan] place to gather rattan .. But in words like amastop, fa ther initandia ohl niglottal they which stop aredis apofpearsten pronounced when pref ixeswith or an other initial format glottalives are addthated, many as .iofn cithe amawords [ts wiijathma initial] fa ther glottalapd k-ohn stopsi [k o7hnhave i]hi thestorically . Ity hadis pos initialsible thesmooth CV-o(C)nset .sylla voweblels st butructur acquiredes. Thethe glottalfew words st opI ha byve aninalogical my data pr withess urephonemi from c smooth-onset initial vowels: i a the, ama fa ther, ina mo ther, ohn i they, and proceala getss ofare initialphonologically consonant excepti insertiononal , becauseand pos theysibly are escaped high- thefreque an'ncyalogical words . The semi-vowel w [w] occurs in all positions. Examples are wad is tooth, fang, wawa child, hawan sword, kawnay basket, lakaw· bamboo po le, and rad iw song . Wordsis pr obabbegilenning that with the wwordare' wawa rare has, and been may borrreflectowed afrom secondary Chinese deve, forlo pmexaentmple. , andIt devethe initiallopment wst emmingin wama from fa the ther, wincommona mothenoun r, marandker wa cofor dog topic is prandobab nominally a later predicate: o. The semi-vowel y [j] occurs in all but word-initial position. I cannot find any words in my Amis data that begin with y, except for a few Chinese loan words like yising physician, doctor and yong so ldier . Other occurrences of y include the following examples: qayam bird, kayakay bridge , kays ing bowl, baboy pig, and kamay hand. It must be noted that though we have included the phonemes w and y, their Thestatus few isca quiteses there tenuous are . are Bothseco wndaryand devey arelopm veryents rare or inloan initial words . posiAltionthough. phoalmneticalost alwaysly [wbe] prandedict [y] edshow as upglides a lot automa, in ticallymedial posi insertionted the by irrules occ urrencel(a) and can tol( b)the givense rules belo w.is quiteThe smaltotall, listwhile of there words arethat many I canwords find that that confor do notm toconfor the m insgenetanceralisas tioof n.inte rvocIn myalic or wthoandgra phyy, mar I kinghave onlychosen those to lethatave outare notall predictable by environment as in siwa nine, sawad abandon, qayam bird, dayom [oajom] easy , cawi 1 [tsawi 1.] earrings, and cowi q [tsuwe?h] nightingale. The generalisation which motivates this orthographic decision is one of wouautomatld liicke glottalto point stop out andthat glide in Amiinsers theti onmo. st poRepularferring forms back fotor disFigyllaure bles1.12, are I CVrespe.CV ctivandely CV.CVC,, by anwith epenthe CV.Vtic and proce CV.VCss conve whichrted inse phonerts atically glide toor athe glottal same formsstop beenvitweenronm entthe . adjacentThe proce vowessls can. be What ex pris essedinserted by the depends following on the rul phonoes: logical 20

l(a) -+- [wJ I t-:a ) (b) -+- [j] I i-mI j

a_ : (c) -+- [7J I l } I ° {�} (Except for a_e, there are no other instances of e occurring in a vowel seqthe uencefollowing in my envirArnisonm daentsta, the: e_arefore, i_ rulese, e_i l(a), o_, (b)e, , e_ando, (c)and do e note) . address Examples of automatic consonant insertation are: qoay [?owaj J ra ttan, soal [sowatJ speech, toa [towaJ then, ma-olah [mawolahJ like, raic [J__ajit shJ rope, ni-aroq [nijar�?nJ place, ci ama [t�ijamaJ fa ther, ia [ijaJ the, kio [kijoJ eggplant, ci ina [t�ijinaJ mother, saan [sa?an] then, taes [ta?as] hit, bois [�u?isJ star, and poot [pU?uth] knife . This insertion takes place both witnamehinly cithe amaword and and ci acrinossa show word the bound proceariesss ope. rationTwo of acrtheoss exam wordples boun givendaries above , whmail-olae theh, the othe prer exfixam plesma- isshow added operation to the witsternhinol ahthe toword form. anIn intra the nsitcase ivofe verb. Ainter simivocaliclar epenthe positiction proce. Thssis isis foundby no alsomeans in anmany unusual Philippine proce ssla andnguages can bein phoneonsettical. ly explained in terms of formant transition and the manner of vowel If one adheres strictly to the structuralist dictum: "once a phoneme, always a phcnemthe exae"mples, then cited even abovethe predic, whichtable would w, be y,, inand my q opinionshould be, losingspelled a outsigni infica allnt phogenenologicalralisations gene asral expreisationssed. in Hencerules, thel(a) de, ci(b)sion, and is (cmade) by toomi recogttingnis alle the violainstancteess the of predibiuniquctenessable w,cr y,iteri andon q. in the orthography even though this

1.4.4 p, t, k, q, c, and Aspiration The p, t, k, and q are aspirated in word-final position. In initial and medial positions they are unaspirated. Examples showing p, t, and q in non-final positions include: panay [panaj] sweet-rice, tipos [tipus] rice, pito [pituJ seven, kilang [ki la�] tree, sarakod [sarakod] heel, qenem [?anam] 21 six, toqas [to?as ] ancestor, liqel [li?a 1. ] neck, qakqak [?ak?akh ] crow, and siqnaw [se?naw] cold. Examples showing p, t, k, and q in final position include: qicep [?itsaph ] bete l-nut, qabet [?a�ath ] gunpowder, bonak [�unakh ] saThend aspiration, kokoq [kokj at?h the] foreot,lea monse ofiq [mtheun£ glottal?h] soft, stop and can qomaq be de[?scuma?ribedh ] fiaseld. a pharyngeal fricative [h]. Since at the time of release the oral cavity may I;)], a[ a ], i [e £], maintain the shape of the final vowel 0 or 'V Ogawa and Asai (1935) interpreted the aspiration as voiceless vowels and represented predictablethem as [=6], [g], lessor ['is] paragogicinst ead vo ofwel aspiration is consid. eredIn an my ac analyscompanyising, th isfeature of the final q. The affricate c is also aspirated in word-final position and unaspirated elsewhere. This patterning is parallel to that of the obstruents p, t, k, and q. Examples are: ciris [t�iris] stream, cacay [tsatsaj ] one, and temang ic [tima� itsh ] weep� cry . A final p, t, k, q or c loses its aspiration when a suffix beginning with a vowel is attached to the word. Examples are: satip [satiph] western and sat i p-an [satipan] the west ; sepat [sipath ] fo ur and sa-sepat-ay [sasipataj ] the fo ur ; qacek [?atsakh ] dirty and qacek-ay [?atsakaj ] the dirty one ; pas iq [pa�e?h ] embarrassing and saka-pasi q-an [sakapa�e?an ] the most embarrassing thing ; qipoc [?ipotsh ] short and qipoc-ay [?ipotsaj ] the short oneposit. ionIn. short, the Amis stops and affricate are aspirated only in word-final

1.4.5 Pal atal i sation of c and 5 When followed by the vowel i, 5, and c are palatalised as in si-kawas [�ikawas ] haunted, cir is [t�iJis] stream , bas is [�a�is] pork , heci [hat�i] meat , and saciq [sat�e?h] whip . Elsewhere 5 is [5] and c is [ts] in initial and medial positions and [tsh ] in word-final position.

1.4.6 band d The voiced consonants b and d are realised as voiced stops in word-final position. Examples are: taheb [tahab] cover, qanob [?anob ] covet, qahcid [?a?ht�id ?aht� id] salty , bo lad [pulad] moon , sawad [sawad] abandon . It is to be'V noted here that words ending in b areand ex tremely rare in Nataoran­ Amis. Intervocalically and initially b and d are realised as . Examples are: tebos [ta�os ] sugar cane, qabe l [?a�a 1. ] charcoal, bo is [�u ?is] star, bot ing [put i�] fi sh, qada [?aoa ] enemy, qoda l [?ooa1. ] honey , dipong [oipo� ] nest� hive , doka [ouka ] wound, and da-demak-an [oaoimakan ] working place� office . d mal bawa a In consonant clusters, b and are stops, as in [mal bawa ] protruding and baqdet [pa?dath ] hot (of water) . An epenthetic [ a ] is usually inserted between the two consonants of a consonant cluster if the first element is the voiced stop b as in qabsaq [?abasa?h ] bland� taste1.4.8.less . A parallel is found with r and 1 which are given below in section 22

1.4.7 h In syllable-initial position, before the vowel i, h is realised as [xJ as in babahi [�a�axe J woman and bihid [�ixidJ cheek . Elsewhere h is realised as [hJ. Examples are: habay [hafaj J mi llet, hemay [hemaj J cooked rice, pohong [puhu�J horn3 antler, waheng [wahe� J mo lar tooth, and kaqhong [ka?ho� J shoulder blade . In syllable-final position, h is often realised as the pharyngeal [nJ. Examples are: olah [olan J like3 love and kapah [kapan J good in word-final position, and mihca [mrntsa J year, mohting [mon t i�J black, and lahngang [lan�a�J red as the final consonant of an initial syllable, or, in other words, as the first member of a consonant' cluster. In the latter case, [nJ alternates freely with [?nJ as in mihca [mrntsaJ � [mr?ntsaJ and mohti'ng [mon t i�J [mo?nti�J. [nJ [?nJ h � This alternation between and makes overlap phonemically with q, which has,as its allophones [?J and [?nJ. Looking at the environment of occurrence, we find that the [?nJ that alternates with [ n J is tn complementary distribution with the [?nJ allophone of q, which occurs onlytogether in word- by comfinalplem enpotarysitio n.dist ributionThe alternative and pho neticsolution simi oflarity grou wopinguld them lead to Howevthe esterab, thlishmentis analys of isan turnsaddi tionalout to phoneme be unsatis, say,factory H, be onsid esthree q andgroun h.ds . First, it loses the generalisation that q, like p, t, k, and c is aspirated in word­ final position and unaspirated elsewhere. Also, the three phonemes H, q, and h will each show a defective distribution, as can be shown by Figure 1. 13. Finally, the H, q, and h alternative itself entails a phonemic overlap with [nJadopting showing thi ups threeas allo-phonemephone soluof bothtio n.H and h. Thus we gain nothing by

Initial Medial Final V-- V V-- C H I ?n �n ?n q ? ? ? h x,h x,h n

Figure 1.13 An Al ternative Solution with Defective Distribution for H, q, and h 23 The two-phoneme solution adopted here shows the following distribution instead:

Ini tial Medial Final V- V V- C q ? ? I ? ?h

h x,h x,h h 'V ?h h

Figure 1.14 Di stri bution of q and h voThicise anallessysis stop alloconsonantsws us to mentioned capture the above phono. logical generalisation about Amis

1.4.8 Liquids and Nasal s Thefina lal poteralsition 1 isand retr asof thelex edfirst [t] elinement syl lableof a -fconsonantinal posit clustionser, i.e.. Inin theword­se positions, 1 optionally takes on an epenthetic [ a ] as in the following potal [putat] [puta1.a ] cope l [tsop 1. ] [tsop 1.a] examples: 'V front yOX'd, a 'V a cilmin [t�i1.mi n] tt�itamin] ma l cad [ma .tsad] hot �epper, 'V sour, and 1 'V [ma 1. tsad] same . Elsewhere 1 has the quality of a dark [ 1 ] as in many Philippine languages. Examples of words with initial 1 include: 1 ial [1ij a1.] sea, 10maq [luma?h ] house� home, and 10tongay [lutuQaj] infant. Examples of words with 1 as the second element of a consonant cluster are: cinglaw [t�iQlaw] excited and qangl iw [?aQ l iw] stale . Examples of words with intervocalic 1 are: kilang [kilaQ] tree, bolad [�ulad] moon, ala [ala] get, and 50101 [50101] fo rgive . The retroflex r is a flap [J] in all but syllable-final positions where it is [r]. 1, r [ a]. a trilled Like syllable-finala may also take an epenthetic a katar [katar] 'V [katar ] land� shore, sapor [sapur] [sapu r ] Examples are: e 'V rice sprout, qone r [?oner] 'V [?oner ] snake, qarkaq [?arka?h ] 'V [?areka?h ] appalling, borsen [�ursin] [�uresin ] numb, torcak [turtsakh] e 'V and 'V [tur tsakh ] beak . Examples showing r in initial and other medial positions include: rarapa [JaJapa ] water buffalo, ciris [t�iJis] stream, tireng [tiJGQ] body, and kapray [kapJaj ] pimple . The nasals m, n, and ng occur in all positions. Examples are: mata eye, nanom water, and ngangan name initially; ama fa ther, qenem six, and rangat fence intervocalically; tamdaw person, tenpeq feather, langdaw blue� green, toqman dark, siqnaw co ld, and lahngang red in consonant clusters; and qayam bird, karin wrist, and kakang crab in word-final position. 24

1.4.9 Vowel s In my orthography, the four vowels of Nataoran-Amis are symbolised by i, 0, a, and e (see Figure 1.11). i [e €J [i rJ The front vowel is realised as � before q and as � elsewhere. In unstressed syllables, [r J is more common than [iJ. Examplesare : qadidiq [?adrde?n J � [?adrd€?nJ small, kot iq [kote?n J � [kot€?n J serve you right!, qicep [?its aph J � [?rtsaph J betel-nut, lisin [li�inJ � [lr�inJ fe stival, mihca [mrntsaJ year, and nani [nan i J cat. (Note: my phonetic transcriptions thanrepre onesent phone particulatic t£anr tokenscription pronu nciforations a given as wordI recorded appears themin my. notSomeseti, mesfor more whereexample the, qicalterep nationappears isin relev my noteants, bothonly oneas [? tokenrts a phis] andcited [?.)it s aph J. Except 0 [0 �J [u The back vowel is realised as � before q and as � o J elsewhere. In unstressed syllables, the most common form is [oJ . Examples of words with 0 lotok [ l t kh J � r lutukh J highland, tebos [tafos J � [ta�us J the vowel are: a o o a sugar cane, taor [tawo r J � [tawu r J a kind of tree, banoq [�ano?h J � [�an�?hJ boke loq [�uk lo?n J Tfok l�?h J fe ather, down , and a � a stone . The choice of the 0 [oJ andsymbol that ratherit is inthan the u middleis based of onthe therange fact of thatphone ticis va thelues more of thcommonis phoneme form . writeThe range them of as theif backthey vowelare of is the act sameual lyhe loweright would than bethat mi ofslea thedi ng.front The vowe chl;oic toe of 0 also reflects a tradition in the description of Philippine languages such as(B entonMaranao 19 71(M:1cKaughan2-14). 1958, 1963), Bikol (Mintz 1970:17-19) , and Pangasinan The symbol a stands for a low central vowel [aJ. Examples of words with the vowel a are: qatay [?ataj J liver, sema [sama J tongue, mata [mata J eye, boaq [�owa?hJ kidney, and kapah [kapah J good. [i J The symbol e stands for the pepet � a in Amis. Examples of words with the vowel e are: tebos [ta�os J sugarcane, tesek [tasakh J real, qicep [?itsaph J bete doeesl-nut, not dimoccurel bedimforea 1. aq.J nea Int, andan unstrema-becssede r [m CV-abitssylalarblea J sa precededtiated. by Thean vowel initialsequence glottal, e is oft stenop notq and pronounce followedd. by Thanyis consonant results in C, an i. e.app, arentin a qCqeC sequence as in the following examples: qenem [?nimJ six and qepah [?pan J wine .

1.4.10 Polysyl labi c Words In some of the disyllabic examples given above, such as qabsaq bland, tasteless, cilmin sour, ma lcad same , borsen numb, and torcak beak, which are in the CVCCVC form and have b, r, or 1 as the first element of the consonant cluster, an epenthechangingtic a [disyla J isla bicinse wordrted betweeninto a trtheisylla two bicconsona one. nts, thus phonetically Many words of the CVCCVC form show two identical syllablesas in the followin� examples: qakqak [?akh ?akh J crow, ciwc iw [t�iwt� iwJ chick, bilbil [bitabit J lips, ngisngis [�is�isJ beard, tektek [takhtakh J po le, bar, thepay payphono [pajlogicalpaj J kite, rules andwe havebay bay given [baj thusbaj J mosquitofar regarding net. consonantsThese words in disinitialregard theand meextentdial pothatsit asiopinsrated. The st opsfirst are sylla allowedble sim inply word-m copiesedial the posi secondtion sylas inlable to 25 qakqak [?akh?akh ] crow and that the initial consonant such as b in bilbi l [bitabita ] lips and paybay [bajbaj ] mosquito net which should normally be a fricative [f] is assimilated to the initial stop of the second syllable. Trisyllabic and polysyllabic,words are formed by such morphological processes as reduplication and affixation. Examples are: lia-l ial-an [1 ijalijalan] seashore , ka-s i qnaw-an [ka�i?nawan ] winter , ka-koqkoq [kako?hko?h] frog , tam-tamdaw [tamtamdaw ] everybody , remia-remiad [JamijaJamijad] dai ly , and na-tao r-an [natawoJan ] Nataoran . Note that the medial q in the reduplicated stem of ka-koqkoq frog is aspirated. Examples of other trisyllabic and polysyllabic words are: bal ocoq [falotso?h ] heart, tangila [ta�il a] ear , kacomol i [katsomol i] snail, po lalohoq [pulaloho?h ] termite , qalipawnay [?alipawnaj ] firefly , and qadi papang [?aoipapa� ] butterfly . CHAPTER 2

FRAMEWORK OF DESCRIPT ION

2.1 Lexi case - Pre- and Post-1979 Model The theoretical framework adopted in this study is called the "lexicase" model. This approach to grammatical description grew out of Chomsky's model in Aspec ts of the1968 Theory, some timesof Syn outtax of(1 dissa965; hencetisfactforthion Aspecwith bothts) and. TheFil lmorelexicase's case mo degral mmaris a ofform con ofstraints generative desig grammarned to whichmake theis narrowlygrammar falsirestrictedfiable becand ausehence it ve harifias a blenumber. Thestudents lexi caseand mocodelllea hasgues beenat the developed Universit by y Stofan Hawaiiley St arandosta expounded and some in of some his fifty previouspapers, ardoctoralticles, disand sertadissertiotations tons the. desThe criptivelexicase analmodeysisl has of been Japa neseapplied by in Har(197vey4) , TaKusayloreia (1n 971)by Keedong, Rukai byLee Paul (1974) Li , (1Vie973)tnam, Thaiese by by Mary Praneebeth Kulla Clarkva ni(1975jaya, puformablisllyhed pu inbli 1980)shed , inTagalog 1978) , byMe Videlayua BetaDeGwiuzman by Kay(1976 Ik, ranagaraformally (1pu975blis, hedforma inll y Acson1978) , (1Kagayanen979), and by in Caro a nul mbHarermon of pape(1977)rs , onClassical languages and inclu Modernding Greek Sora , byEng Velishneeta, and1979 b,Fo rmoandsa ton byappear Stanl b)ey . StaTherosta lexicase (197lb mo, de1971c,l has 1973been, sho1974wn, to19 76dbe , a 19via77ble, 1978 , tralteanrnatsformationalive to therul moesre. powerful grammatical models that utilise Prior(PSR's) to, the1979 lexic, the onlexi, andcase a phonmodeologl consistical edcomp ofon enta set. ofIn phrase Figure st2.1ructure an ou rulestline Starof theosta le xica(1976bse , mo19del78) pr. ior to 1979 is presented. For further details, see In the pre-1979 version of the lexicase model, as in many grammars written in Chomskygenerate's la Aspecbelledts modtreesel , andcontext-free indicate hie phraserarchic stalructure rela tionsrules between were used sent toence constithe constittuentsuent. Thestructure basic functionand word ordeof PSr R'ofs wassentences to make in genethe ralalinguagsatioe.ns aboutAs the plalexiyedcase by moPSRdel's evolv in maked,ing it stawastem suentsggest abouted by constructionStarosta (1976 com:2)posi thattio n theand role order Stcouldarosta be takendecla reovers the by endcontextual of PSR' s featuresin the foandllowing redundancy statemen rulest: . In 1979,

26 27

Lexicon Phrase Str(PuctureSR) Rules Lexical rules: Subcategorisation Rules (SR) DeRedundancyrivation Rules (R(DR) Lexical entries: Lexical category features OtherCase featuressyntactic features PhoSemnologicalantic feat uresrepre sentations

Lexical items Trees or bracketed strings

I Syntactic Repre� sentations

Phonologicalj Component

Phonological Representations

Figure 2.1 Pre-1979 Lexi case Model (adapted from DeGuzman, 1976:9) 28 SiPSR'nces, we we havecan decreasetwo ways ofthe doclassing theof possamesibl theing grammar, RR's and mechacompatniiblesms ; withand siournce theory Redundancy by outla Rulewings and one contextualof the two Phrasefeatures Structure are nece ssRulesary elthatsewhere have into thego (Stgrammaarostar, it1979a is :7the0) . Howshown redundancy later in rusectlesion and 2.2.2.2. contextual features can actually replace PSR's will be Witcomponentsh the el shownimination in Fi ofgure PSR' 2.2.s, pos t-1979It is understood lexicase mo thatdel islexical composed entr ofies the are indicastill tedcharact in erisedFigure by2.1 such. Inlexical fact, thecate gtraorynsit featurion esfrom, case pre- featurto posest-19, etc.79 moasdel intaleavesct . practically all the basic assumptions and claims of the lexicase grammar

DerivationRules (DR) Lexemes 1 1 RedundancyRules (RR)

SubcategRulesorisat (SR) ion Inflectional RulRedundancyes (IRR) I I L WordsDe scriandptions Syntac tic

Figure 2.2 Components of a Lexi case Grammar (Post-1979 model adapted from Starosta 1976b:154 and 1979b:5). 29 We specishallfica nowtion turnand torule a deformalscriptionism. of the lexicase model, including feature justifiably so, by l,ooking at FigureAs' 2.2 one that may a havelexi alrcasee aaygrammar suspectam oued,nts and to a thorough description of the lexicon, we must point out that, in ab lexicase modsentencel, es,syntax and is the conce ived as "the way in which words can be com ined into entries" (Starosta combinato1977b:150)rial. potentWords areials piv ofotal words in as the mar lexikedcase in thfraeirmework lexi cal Standaros constita putstute it:the subject of study for both morphology and syntax. As Moandrp holosyntaxgy is anthe accountstudy of of the the interdistnalribution structure of words of words in , thatsentenc canes be. madeGrammar about is the the internalset of all and generalexter nal st distatemenributionts wordsof words which in havesentenc beenes , standripped the oflex iallcon aspectsis a list of ofthe irlex emes, sound-mgeneralea runingles-. form(St correspondarosta 1978ences:3) which can be predicted by A lexicase grammarthus contains rule statements of all the possible ofgene wordsralisa, wittiohns lexaboutical themat ricesform, strconteeamlinednt, and accordingly syntactic di. stributionWhat we getof clais ass es non-prelist ofdicta lexemesble , whichirregular are re, siduand alidiosyncra, quintessticential feature formss, andmarked a set only of lexicalfor rulesare no st rulesating ou alltsid thee ofsig thenif lexiicantcon gene, excraleptisa tionsfor the about Phone thetic laInternguagepre. tationThere Rulexicaseles (PIR grammar) and Seproper mantic (sInteeer Fipretationgure 2.3 beloRulesw) .(SIR ) which are excluded from Prleaste-1979 developed lexicase com grammarponent withinha d a pho thenolog modelical. comWhatpon witenth. phonoThislog wasical prob ablythe therepre lexisentacontio ton handleprovided phono forlogical e�ch lexical changes entr ory alterand withnations morpho witphonemichin word rules in boundaralways minimies, theal. roleAs DeGuzmof the anphono putslo gicalit, the comp phonoonentlogi incal a compolexicasenent mo isde lli wasmited onlythe sentence" to operations (DeGuzman that are1977 "a:7cross) . Sucwordh operations boundaries havebut witnot hinbeen the de boundsscribed of at phonany ollengthogy (Lini the1973: liter12-3ature2), they, but seem to toju dgest ayby wit Pahinul Li the's desGenerativecription Phonology of Rukai forframework his own. desLi criptionhas adapted. distinctive features and morpheme structure rules ArniSinces orthophonologgraphyy isprim beyondarily the on scothe pebas ofis the of presentautonomous stu, dyphonemic, I have principlesdesigned an phono(see sectiologicaln repre1.4) sentand haveations treated associ myated ort withogh raphlexicalic repr entriesesentations. as the Ainter lexicapretationse mode lis does pla cednot ouincltsidudee thea separaboundaryte semaof nticthe mod componeel, alntthoug. hSema semanticntic reinflaormtionsation between has in the factse beenlexical provided items witby hinthe acontent given ofsentence lexical. itBeyondems and thi thes kindhandle of sema "intrinsntic ic"inter orpre intentationsional based sema on ntiextc ra-inlinformationguistic, inthefor momdelatio ndoes su chnot as "p(Sertarospectivsta 1976e" b:3,(Fillmore 29), or19 77"r:5eal9, Standar ostaimagined 1978 worlds:25-29)" , (D"ceGuzmanontext 197of 6:8situa) . tion"The onsentencely conte. xt of significance to the model is the linguistic context within the 30

(CoGRAMMmpetenceAR )

Lexicon SubcategRedundancyorisation Rules Rule (Rs R)(SR ) InflectionalDerivation Redundancy Rules Rul(DR)es (IRR)

Syntactic-RepreSemasentationtic-Phonn ological

(performance)

C ONTEXT Semantic Phonetic PERCEPTUAL 0S FITUATION -)- RuIntleserpretation (SIR) RulInteserpretatio (PIRj ) n +- CONTRAIN TS Perceived Phonetic Sitand uations1Ro les Representatio1 n

Figure 2.3 A lexicase Flow Chart (Adapted from Starosta 1978:3)

2.2 Lexical Features and Lexical Rules The lexicon in a lexicase grammar consists of three basic types of lexical subcaterules andgor iza ationlist of rules lexica (SR)l entries, redundancy. The rules three (R R)types , and of delexirivationcal rules rules are (D: R).

2.2.1 Lexical Features Each lexical entry has a matrix of features, specifically those that are not Eaassigch nableentry byals osubcate has a gophorinologicalsation rules repr noresent predictableation. by redundancy rules. 31 Lexical features include: a. lexical category features such as [ +N ] for Noun, [ +V ] for Verb, [ +Det] for Determiner; b. case relation (CR) features such as [ +PAT ] for Patient and [ +AGT ] for Agent; c. case form (CF) features such as [ +Nom ] for Nominative, and localistic case formfeatures such as [ +drcn ] for "direction", and [ +sorc] for "source"; d. contextual features, including case frame features, particular to verbs, such as [ +[ +PAT ]] meaning obligatory co-occurrence with the Patient case relation, or [ -[ +Nom,-PAT ]] meaning that thePatient verb's case co -occurringrelation; Nominative case form can only be in the e. semantic features associated with syntactic or morphological consequences, such as [ +top ] for "topic" and [ +def ] for "definite"; f. morphological features such as those associated with voice and tense/aspect inflections on the verb, e.g. [ +inch ] for inchoative and [ +pot ] for potential; and g. otheas dicr tionidiosyarync mearatningic features assigned inclu to dingthe it idiomem byati convec usntioage, n,as e.gwell. [smaZZ] for ad iding and [once-upon-a-time ] for ia-hanto. Any of these lexical features can be used for the subcategorisation of lexical entrlexicalies . rulesLexical. items become fully specified only after the application of

2.2.2 Lexical Ru les Theare threethe Subcateg types oforisa lextionical, rulesRedundan thatcy, reandlate Derivation lexemes toRule fullys. speSubcategor­cified words withinisation the Rul eslanguage (SR) characte. Redundancyrise the Ru polesssible (RR) stalexicalte the categ predicatableories or clfeaturesasses notRules mar (DR)ked areon theword- lexemformationes. Both rules SR'. s and RR's are obligatory. Derivation

2.2.2.1 Subcategori sation Rules Subcatecategoriesgori. sationBoth ruprlesimary (S R)and identi contextualfy features features that aresubcateg placedorise in brackets lexical pre([ sence]) and or mar absenceked by ofa plu cesrtain (+) orcontextual a minus feature(-) to respfor ectiva particularely indicate cate gorythe , envisay theronm Verbent. . For example, [-[ +LOC ]] means the absence of [+LOC ] in that Thethat sy thembol lexical "plus categor minorus"y characte(±) beforerised a byfeature the featurein the outputto the ofleft a ofSR meansthe arr ow itcans bepr esencsubcatege ororised absence by. theFor feature example to: the right of the arrow on the basis of 32

(2.1) SR-i [ +Det ] -+ [ ±dem ] states that the lexical category of determiners (Det) is subcategorised into demonfeaturest racantiv esalso, markedbe ne gativelyby [ +dem marked] , and non-das inemon stratives, [ -dem] . The input (2.2) SR-ii [ -dern] -+ [ ±pers ] which states that non-demonstratives can be either personal [ +pers ] or non-personal [ -pers ] articles. Even before the elimination of phrase structure rules from the system, co-occuthe formrrence of contextual restrict ionsfeatur ones lexical. Si nceitems head were categ markedories on are the su bclexicalategorised items inin theterms dual of co-occurrencerole of subcate witgorih thesatioirn sistander theheads stating, contextual of co-occurrence features may perform reutistlisedrictions in SRbetween's as we collns asti tuentsRR's. . Contextual features are therefore It is to be noted that, the arrow -+)in SR's does not indicate an equivalent PSR"rewrit. e"Instea relad,tio then betweenarrow isthe to inputbe read and: output"is subcateg as itorised does with as" . a conventional SRand's those are di thatvided des intcribeo two the su btinflecypes:tional lexi calprope SRrties's such of as various SR-i andcl assesSR-ii of above , wordsisatio. n RulThees sec (IondSR) . subtByype the ofse SR rules's is words called are the subcateg Inflectiorisedonal Subcategor­according to the bymorp IRhologR's) iassociacal and tedpho witnologicalh certain consequences inflectional (whic featurh arees subsequently. Examples spareec: ified (2.3) SR-iii [ +N ] -+ [ ±plur ] plur "plural" (2.4) SR-iv [ +V] -+ [ ±pastJ past "past tense" (2.5) SR-v [ -past ] -+ [ ±3sing] 3sing = "third person singular" wordsfor Engl thatish cantaken be reprefrom Stsentaredosta by (1the97 6bbranching:220) . diaSuchgra rumles shown de finein Fisugurebclasses 2.4. of An ISR differs from a lexical SR in that instead of simply stating a set of well-formedness abbconditionsreviated on. le xicallexical entry items in, itthe actualexicllyon , opereconsratetits onuting an theentr yfeature was orig matricinallyes ofabstr theact separateed. (Ibid words. :2 24)from. which the Unlike the lexical SR's, ISR's do perform an operation which takes each input infllexicalectional entry featureandre places speci fiitcation with . twoOne entries could which say, dithenffer, that only ISinR' s replace lexicnot oveal rtlyentries listed by parain thedig mslexicon. Consequ as theently exam, plesISR' sin actuaFigurelly 2.4 generate will showitems. Theassig morphoned bylogical ISR's areand phonostatedlogical as Infl consequectionaencesl Redundancy of the inflecRulestio (InalRR) . featuresIRR's mustarrow be in orderedan IRR afterare introduced the SR's beby causeISR' s.the input features to the left of the In recent lexicase analysis (Acson 1979, Starosta 1982d and to appear b) beenlexical found SR 'sto andseparate ISR's havethem . been grouped together since no formal reason has 33

[+N J [ +V] [ -plrl� ] [ +p lrl ] [ -past] [+past] � [ -3si�ng] [ +3sing ] cat cats walkI walksI walked hordogse hodogsrse s seelike seeslike s saliwked datfootum datafeet putshake putsshakes putshaked sheepchild sheepchildren amgo , are goises waswent, were person pepeoplersons Figure 2.4 Exampl es of Subcategori sation of words by ISR's (Ibid.:22 1)

2.2.2.2 Redundancy Rules thatRedundancy SR's have Rules ± on(R R)the resemble right sidSubcatege of theori saarrtionow wh Ruileles RR (S'sR) alinlow form pluses, except and/or minuses, but never ±. Moreover, in RR's the arrow symbol ( + ) is to be printedierpretedctable ondi fftheer entbaslyis. ofIt the im pliesfeature the or adding combinatio on ofn ofunmarked feature features given(s) as reinputceiv eto futhell RRfeature. By sptheecific appliationcation, which of redundancyis needed rulfores su, bsequentlexical semaitemsntic wi ll interpretation. An implied feature is now made explicit by a RR and given exaas mpitsle outp is ut,taken wit fromh eit DeGheruz mana "plus (197" 6:(+)5) : or a "minus" (-) before it. An (2.6) RR-i {++strictv } 7 [+[+LOC]] This means that all verbs marked [+strict] are also marked with the feature on thea [+ LOCright] acta handnt sidmuest of ove thertly ru co-occurle. In withthis thecase ve, therb. added feature states that pairsGreek oflette rulers sa, invo �, lvingetc. maythe besame used featur in esconju. nctA ionredundancy with + andrule -ofto theco llaformpse (2.7) RR-ii [aFi] + [aFj] states that when the input feature Fi' is marked by a plus, so is the output outputfeature feature, Fj' and. thatThe whentwo subrthe ulesinput of feature·2.7 can isact markedually bybe staa minusted as, sofol islows the: (2.7a) (2.7b) 34

An(Li exam1973ple :8 5)of : RR-ii is taken from Paul Li's description of Rukai structure } (2.8) {a-dempers � [adef] [+Whatde f]th whisil rulee non-pe says rsonalis that [- pepersrsonal] arti [+clespers are] ar alwaysticles areinde alwaysfinite [-dedefinitf].e Likewirules ofse , therule form 2.9 2.9 in a theand fo2.9llb.owing is a notational variant of the pair of (2.9) RR-iii [a'Fi] .� [aFj ] (2.9a) [+Fi] � [-Fj] (2.9b) [-Fd ..... [+Fj] Anfeature example: of RR-iii is the following one in Amis involving contextual } (2.10) {aprsn+N � [.:.[ -aprsn ] prsn = "person" Whatarticles this andrule that says non-pe is thatrsonal Ami nounss personal do not occu dor after not occur personal afte artr non-peiclesrso. nal Wherethe lexias conthe byro lepar oftit lexioningical SRlexical's in aca tegorgrammariesis into to bindescaryribe subsets the st, ructureRR's se ofrve tore lasimplitions fyam ongthe lexilexicalcon byfeatur statinges. geLexineralical saSRtions's do aboutnot actuaimplillycational operate on SR'anythings, as thealr ea.tedyrm inim plithees lex, suiconbcate, thoughgorise thelexi incalfle setsctional by onesform alisdo. ingLexi polarcal addchoices predi betweenctable unmark€dthe plus andfeatures minus and values do not of onechange or anymore featuresfeatures that. RR are's only unordealreadyred marked with inrespect the matrixto ea chto othwhicher, theyRR's are orderedapplying. according While toSR 'sincr areea sing degreerule) . of generality, the last and most general RR being the �-rule (Omega­ (2.11) Omega-rule: � 1-[ +N] 1-[ +v] 1-[ +P J 1-[ +Det] 1 1-[ +PAT] 1 1-[ +AGT] I 1-[ +COR] 1 1-[ +REF] 1 1-[ +LOC] 1 1-[+PLc]1 1-[ +INS] I What the Omega-rule states is simply that nothing co-occurs with anything. Ththeis given last wordand modoesst generalnot co-occur of the withRR's any re flecsisttser one cate-wordgory utte. Therance Oms egina-r whuleich 35 co-occurrencerepresents the remoststrict primitiveion, i.e. phrase, non-c stro-occurructureenc ande. the Almol stot hergene rulesral form of ofthe grammarrule. that refer to optional environment can be stated as exceptions to this phraseIt is necestructussaryre torul pointes, PS outR' s,that had inbeen earlier used asve rsionsalternativ of thees tolex aic spaseecial theor classy, of RR's·that refer to contextual features. With the'elimination of PSR's from onlythe system, by the sest RRatements's. What on word disti ordenguir sheands RRco-occurrence's of this classrestrictio is thatn arethe madeir input matsuchrice as s [+containN] or [+onlyV], wheone refeatureas the whichoutput reoffers the toru ales ma rejorfe rslexi onlycal tocateg contextualory, anyfeatures form wistatingth pos co-occurrencesible combinations conditions of cate ingo teriesrms inof allsiste conceivarheadsble. Trseqeesuences of andfeature hiesrarc inhic RRal's actarrang asements well- formedare potnes$entially conditions allow oned. theHowevese treesr, , contextualmarking as illfeature-for medis anyviolat coned.fig urationStaro staof lexicalis obviously items quiin tea pleasedtree if withany contextualthe outcome of asre assigninghe writes the, role formerly played by PSR's to RR's with contextual features Finallysyntax is, thijusst is a probablygeneral sta atementgood way of to the de regfineula a ritiesgrammar that; canlexical be founditems in. the(St arosco-occurrenceta 1976b:1 84restrict) ions of individual RR's are also divided into two subtypes: lexical RR's such as rules 2.8, 2.10, andISR' 2.1s those1 (the features Omega-r ulesuch) asgiven [+plrl above] "pl, andural those". Thethat laaddtt erto typethe outputare called of the dividedInflect ionalinto Redundancytwo parts: Rutheles first (IRR) part. Eachstating IRR, the in syntturnact, mayic consequencesbe further of therest prricesenctionse , ofand a thegiven se inflcondec parttional st atingfeature the inmo terphormsphonemi of co-occurrencec consequences of the givenfollowing infl ecexamtionalple takensubrule froms, whichStarost area (1quite976d :4)di fferent in form. Consider (2.12) [+plrlJ -+ [-[ -plrl ] plrl "plural" ] -+ z] /[ +plrl ] Thesyntacti firstc partconsequence of rule of2.1 the2 intrinflectoduceionals a contextualfeature [+ plrlfeature]. whichWhat itstate sayss theis thatattribute a pl uralin thecon sastimetuent consti headtuent, which. Inis othe a nounr words, cannot, th isfo rulellow maa kesnon-plural a suffixstatement /z/ on is numtobe ber agreemadded entto all. Thelexical second items part posi oftiv ruleely 2.1marked2 st atesfor thethat a befeature referred plural toit asy. MorpRuhologicalles as ex emRuplifiedles (MR) by(S tarthe ostasecond 1976b part:1 52of) and2.12 ascan also consequMorphophoencesnemi ofc Ru theles presenc (DeGuzmane of 1977 particular:7). Such inflect rulesional st atefeatur the espho. nological Lexical RR's apply before lexical SR's and IRR's apply after ISR's. Lexemes whichthe part areicular irreg ularinflectitiesional or exceptionsfeature concer to nethed priorinflect toional the rulesappli cationare marked of SR for's wisoll that be blockedsubsequent. applications of ISR's introducing a ± choice of that feature 36

2.2.2.3 Derivation Rules A lexicase grammarinc ludes Derivation Rules (DR) "in fulfillment of part of itsrelationships obligation between to account words for that the byfact most that ob jespeakersctive criteria of a la hanguagve eto recogbe nise diconffersiderented type distinc of t"lexical (Star ostarule . ibid.Lexi :186case) . DRForm'S, writtenally, DR wit's hcon thest ituteuse ofa verya fletched arrow ( >� ), relate one lexical entry to another lexical entry or a lexical category to another lexical category. By means. of DR'S, new lexical entriesform of acan feature be created matrix by which analog spey withcifies exi thest ingalre onesady .exisIntingput entrto ay DRor is in the Featurescategory . can Outputbe add ofed, a omiDRtted is also, or givencopied in by theOR 's.form Th.eof a convfeatureentio man trixis that. all reprefeaturessenta metionsntioned even in though the out theyput maymatrixor mustmay not appear have inbeen the inderived the lexical reprefail sentto apationspear inof the the outputoriginal matrix lexeme; are necallessa featuresrily omi in tttheed; inputand · featuresmatrix thatof thein theitems out putin thematrix input are class carried not overspeci intoficall they mederivedntioned itinems the. inputAs an matrix example or, cogerundivnsidere thenomin followingalisatio nsrules: given by Starosta (1976b:198) for English (2.13) (a) +v +N+mass 1 >� -[ +oet+rtcl]- rtcl "article" r ) ) (ClFi (b) >� if) ] Rule 2.13 states thatl, corresponding(�::ll to every verb in Jthe lexicon is a mass nounlacks with the exactlyfeature [+thevJ sa andme thelexi seleccal featuretional s,re exceptstriction that on (i)a grammthe deaticarivedl subject noun [+afterNM], an and ar ticl(ii) e;it and adds im pliesthe features the occurrence [+N] and of[+ massa dete]; rmiprohibitsner whi chthe must occ urrencesatisfy thecorresponding same selec tionalverb. reThest ricsymboltions ")as ,, theis readgramm aticalas "imp suliesbject" and of isthe used to exprin ungrammaess seleticalitctionaly. restrictions without treating their violations as resulting non-Ikranagaracomplet el(1y975 prod) mauctivkes ea derivationdistinction rulebetweens. Thethe comformerplet elshey productivecalls the and the prod(WFAuctiv) . eSim derivationilarly, Arono rulesff (P(1976OR) :and35-45) the haslatt alsoer, Wordrecog niFormsedation the Andistalogiincestion between100% produc productivetive, wit andh thenonpr meaningsoductive of Word the derivedFormation it emsRules com (WFpletR'els)y. predicPOR'tables are onPOR thefrom bas heris desof thecription old itofems Be. tawiIk ranagara(Ikranagara gives 19 75the:22) following : example of a (2.14) +v +v r >+ l:C[:;=:J) 1 :�l�g:rYJ 37 Thalwaysis rule a correstatessponding that, degivenrived a comstativparative intre ansstativitive e intverbrans init Beivtawie verb, there, sharing is alspel ciftheication same features that the butcom witparativh thee addiitemtional is derived meaning. of "more" and the beWFA'scoming are sonot. : com WFpletelyA's state productiv the analogie thougcalh patternssome may on ha vethe thebas potis entialof- whic hof one besetlow of (iitbiemsd. :1is23) re lated to another. An e'xample of WFA in Betawi is given (2.15) [ +V ] ::+personerv WFA's are written with-a double barred fletched[ arrow, 1 indicating that they are dirsomeec verbstional in, al Betathoughwi thnotere comareplet correspondingely productiv persone. Rulenouns 2.1, ea5 stch ateof s whichthat shforare s petherf orsammse anfeature actions as or the is chargivenacteri verb sedbut by re afer stas tote are personferred whoto by ha bittheually verb . In this study, the distinction between PDR's and WFA's is kept. Associated with DR's are specifications of phonological modifications, if any, reprethat sentdisting theuish boun thedari outputes of afrom word the. inpIn ut.the caseSquare of Englishbrackets gerundivare usede to nomrightinalisatio boundaryns of, the a wordrig htwhere brac keta suff (]ix) inis partto be (b) add ofed. rule Sim 2.1il3 arlyindicates, a left the bracketbe added, ([ as) inin the the input following represents example the: left boundary at which a prefix is to (2.16) >-+ [ mi Rule 2.16 is the phonological part of a DR which describes a pattern in Arnis byintra whnsiich tiva pree verbsfix /mfromi/ can nouns be thatadded functionto a word as sttheernir, as cognate in the obj formationects. Theof consonantfollowing exaC, mpinclle udingshows thehow zero-an infixinitia /em/l: can be inserted after the initial (2.17) [ d >-+ [ dem Reduplication of an initial CV sequence can be expressed as: (2.1S) >-+ It is to be noted that whether or not an overt phonological change marks the wordschange, "zin erleo xemederiv idation"entity isis alsnoto exprecrucialssed to by the DR formu's. laThetion pri ofmary DR 'crits. eriaIn ot foherr phothe nologicalidentifi cation(cf. Sta ofro destariv ed19 76bforms:1S5-6 are) . syntactic and semantic rather than DerivationInput to DR ru'sles come ares fromunorde lexemesred. witTheyh theappirly pr beedicforetable subcate featuresgorisat speionci rufiedles by. assuthe mpapplitioncation that ofderivations lexical redundancyapply to strulernses rather, in accordance than inflected with the forms tradi. tioThenal brokenthe exis litencene lea ofding som efrom pos sifullyble exceptionsspecified words to th isto DRgene'sra inlisa Fitiongure . 2.2 indicates In summary, Figure 2.2 and Figure 2.3 together give a representation of the moabovest re, namcentel y,conception Lexical Redundancyof the lexi Rulescase mod (Lel.RR or Thesimp lexily RR)cal, Subcaterules degoriscribedsatio n 38 Rulconessti tute(SR) , theIn flectional"components Redundancy" of the lexiRulescon (IandRR) hence, and Derivationlexicase grammar Rules prope (DR) r.

2.3 Constraints on the Model It was mentioned earlier that the transition from the pre- to post-1979 Thelex iconlyase modesignifl ileavescant· changethe basic in the assu modemptionsl is andthe clabandonmentaims of the of thphraseeory intastructurect. rulin esthe, whichlexicon in. effect reduces the grammarto a description of the regularities Thecounte lexiractcase what model St arofosta gramm perceaticalived de assc riptionthe burgeoning was conceived power of in transan effformortations to andgrammar the s.de clineTime ofand em piragainical Starosta verifia bireitlitery atedof tra thensfor originalmational Chomsky generative an exrequiplicremitent, in thatorde rling touistic restore de emscpiricalriptions ve berifia generativbility,e, andthat a numis, berformal of and excconstraintsessive power on lexicaseof trans grammaformationsrs were (s eeprop Staosedrosta in 1979c) order . to eliminate the Amfolonglowing the constraints(cf. Starosta that 1979a have:6 0-2)been : imposed on lexicase grammars are the 1. There is no distinction between deep and surface structures, hieand rarchythere ontoare no anothe grammatr; thatical ruleSis, there that aremap no one ru sequenceles that oradjoin , deanotleheter, pestrrmuteucture, or. copy parts of one structure to produce 2. A grammarmay contain no rule features. That is, it may contain nothe lexicalnumber orfeature "addre ofss" the or form a parti [+Ricula] orr [rule-Ri J inwhere the "Rgrammari" is and where [+R J indicates that the item on which it is marked must undergi o rule Ri in every derivation, while [ -Ri J indicates that its host item may never undergo rule Ri. 3. Grammarsoperate on words, not morphemes. .Syntactic structures orare dishierembodiedarchical morphemly structuredes. The sequences lexicon ofconsists words , ofnot words string ands Thereof st emsare (wordsno rules minus ou tsidinflece oftio thenal lexiaffixcones. ), not morphemes. 4. Featuresin a synta arecti markedcal repre onlysen ontation lexical. items, not at higher nodes 5. Every syntactic constituent must have at least one lexical headlabel wh toic beh bygiven the Xto convthe entioconstin tuent(Chomsky of which1970) determinesit is the headthe . 6. (The Sisterhead Hypothesis, cf. Starosta 1975:74-5) a. Grammatical relations obtain only between the syntactic heasisterd of con a constistrtuenuctionts. and the lexical heads of its b. All syntactic subcategorisation of words is indicated in thelexical lexic itonems by which means occurof contextual as heads offeatures their oncon thosestructio ns. 39 c. Contextual"sisterhead s"feature, i.e.s , referlexical only heads to inherentof sister features consti tuentsof . whichThe term are "sdiisterrectl yco domnstitinateduents by" canthe be sa mede finednode . as Atany thtwois poi(ornt mo, rethe) contermssti tuents "head", ""attribute", "endocentric construction", and "exocentric construction" inshould a lexicase be intr oducgrammed.ar . TheThe conceptshead of a"hea constructiond" and "attribute is the " obareli gatoryvery important representative of that constru�tion. Centricity of a construction is ondetely rminedone' he byad, theit numis berendocentri of headsc, andih the if aconstruction construction - hasif morea construction than one has obtheli headgatory is memthebe itr,em it thatis exocentappearsri withoutc. In parenthea conventiosesnal whereas phrase the str attributeucture rule is , anfol itlowiemng in exam parplesenthe: ses on the right side of the arrow, as shown in the (2.19) S + (NP) V (NP} (2.20) NP + (net) (Adj) N (2.21) PP + P Examples 2.19 and 2.20 are endocentric whereas example 2.21 is exocentric. InThe examheadple of 2.1an 9,endoc V isentr theic hea constructiond of S. Inmust ex ampbe lea 2.lexical20, N isitem the rathe headr ofthan NP a. colexicalnstruction item.. In examplean exocen2. tric21, itconstruct is P. ionAs, forat least the attribute one head , mustit may be bea eie.g.the NPr a inlexical 2.19. item, e.g. Det and Adj in 2.20, or a construction itself, With an understanding of these basic concepts, one can now see that the hieraSisterrchyhead amongHypothe thesis co nstiis intuents essence of a clsentenceaim about. theSinc enature heads of are the dependency subcatethem, thego risedSister withhead reHypothespect tosis thealso synt seracticves as attributes a governing that principle are dependent for on dedirectlimitingion of the modi domainfication of lexi. Allcal posubcategssibleoris sequencesation andof esconstittablishiuentsng arethe heaadeqd-attriuatelybute accounted relatio fonsri. so is such grammatical information as centricity and Byrepre resentaferringtion to forthe anyRR's,given it sentencis alsoe. po ssiTheble lexito constructcase convention a tree for heconsad tructingby written trees on reqa veuirrticales that line, fordi rendocenectly undertric theconstructions node which, domithe nateslexical the theconstruction right or aslef at, whbutole one, and step that be attrilow thebute levs beel writtenof the hea on d,sla e.ntedg.: lines to (2.22) NP

the yellow [ +Det] [ +Adj] 40 For exocentric constructions, the co-heads are both written on slanted lines, eqa wholeui-dist andant one to thestep right down , ande. g.lef: t of the node dominating the construction as (2.23) PP over NP [ +p ] [ +N ] the [ +Det ] beIt ais wordrequired, no more that andthe noend le ssof . everyThe br constructionanch in a lexicase types candepend be encidenty treeified must by the category features on their respective lexical heads, such as [ +Det ] , [ +N ] , laandbe[ls +p such] , instea as S,d PPof , asand nodes NP arein thethereby tree optionalrepresentation and form. allyAlthoug redundah nodent, since theyreada bilitare predictabley. from the category features, they may be retained for Bycontextual referring features to the Siofster wordshead provide Hypothe ansis alter, lexinativcale redundancymechanism rulesto account util isiforng ofcon hiestitrarcuencyhical and dom poainsssible of sequences subcatego risatof constituentsion. The withinrole pla sentencesyed by phrase in te rms takenstructure ove r rulesby contextual in making feature statementss and aboutredundancy constit ruluentes . structures can thus be Together the constraints in this section narrowly restrict the class of posincessible 19 57gra. mmTherears, ruareling other out coa nslargetraints numb oner oflexicase "gene rativgrammae" rsgrammars, such aswri thetten de"onealt per with Sent" in secoctionnstraint 2.4.5. or the "l/Sent principle" which will be separately Constraints 3-6 above are the ones most directly related to the abandonment of PSunexpeR's.cted Sta conseqrostauenc (197es"9a :5of9) impannounosingced the these endconstraints of PSR's on as lexica "one ofse thegramm ars. Theillu replacementstrated belo wof byPhrase showing Str howucture a repreRulessentati by lexive calset redundancy of PSR's (2rul.2es4-2 ca.2n7) becan bePSR st's:ated as RR's (2.28-2.31) instead (ibid. :20). We shall begin with four (2.24) S ->- (NP) (Adv) V (Comp) (2.25) Comp ->- pAS (2.26) PP ->- P'NP (2.27) NP ->- (Det) (Adv) (Adj) N (pp) 41 These PSR's can be replaced by the following set of RR's: (2.28) [+V ] + + ([ +N]) a. ++( [ +Adv([ +V]-) ]) b.c. + ([ +p]) d. e. -[ +N][( +N��J] f. -[ +Adv[+V ][ +Adv][ +V] h.g. L+A-[dv+p][][ +N+p] ] i.j. (2.29) "+ [ +N] + ([+V]) (2.30) + ( + ([_ +o ed) + ([ +Adj+Adv]) -[ +Adj+Det ][ +D+Adjet ] -[ +Adv+Adj ][ +Adv+Oet]] -[ +Adv+Adj ][ To+Advet]] + ([ +p]) + ([+N +N ]) -- [+-pp ][+p] - = [+N][ +N] (2.31) Omega-rule (See rule 2.11) These RR's show that what was expressed by PSR's can also be expressed by RR's. Fo28jr exguaranampletee, 28as that and they28b willstate not that be an in NP the and wrong an Adv or der,can precedeand 28f anda verb 28 gwh ile prevof aent class more may than be onefollo ofwed ea bych aform sentenc frome precomcedingplement the (a verbs sta. ted Thatin rulethe s main2.2 4 verb ofand a 2.2cla5)use finds may beits foequivalentllowed by expra siesstersion co nstiin 28ctuent which who sesta heate dthat is athe verb main. verbBy usingas its aheads double a -neV gationand a P;, 28ine otstherates wordsthat , thethat com itplement is an exocenconstitrictuent must have 2.2constr8-2uctio.30 aren. laAlterl enviexcluronmeded bynts thenot Ome speciga-rulefically. Inal lowedthis byway rules, all pesuchrmi asssi ble coforns bytructions referring as weto llthe as contextual nonpermissi featureble sequencess on the oflexi thecal lang headsuage. are accounted

2.4 Basic Notions of Case Theinf luenceuse of of case Fil relmorelations's case in grammarthe lexicase, which mode originallyl can be proposedattributed the to the incor1966) poration. Fillmore of the(1 968)notion int roducedof case caseinto trreanlatisfonsormational as cate grammargories domi (Fillnatingmore 42 NP's in the base component, but subsequently ran into a numberof problems, becamincludinge subject the fact to Chomsky that by's repre critisentingcism (1 965case:68 re-74)latio ofns grammars as nodeswhich , his conapproachfuse thecategorial years has and beco relametio increanal notiosinglyns . abstrIna addict, andtion th, Filis tendencylmorean casehas coincgrammaridedover witrealisedh increasing in various neglect lang uageof thes. question of the ways in which case relations are exaLexicamplese, hasthe ' elattempiminationted to of overco transfme orthmaesetio dins fficultiesavoids the inproblem various of ways excess. iveFor absolvstrae ctnethe problemss, and treraisedating by caseChom skyas features. of lexical items has helped to Lesentencxicasee assumstructurees that of bothall languagcase reeslations, and thusand disticase nguishesformsare betweenbasic totwo the types ofis casecompletely featur presedic: case'tab,l ree latiofromn thefeatures other . andHence case bothform typesfeatur esof . feaNeturesither are one basemsicantic. Whereasre lationship case betweenrelation a isnominal under stoandod itsas theco-occurring syntacti calpredicaly sigtenifi, casecant correform spondenceis the gr ammbetaticalween carease lisreationlations or and manife casestation forms ofis bycase no remeanslation bi-u. niqueThe . provideCase re lathetion context and case by whichform featuresverbs can marked be subcategori on heads ofsed nominal. The comconsbitructionsnation of casefeature rel: ationcase andframe casefeatur formesfeatures . Case framegives risefeatures to a arethircontextual d type of casefeatures markedrelations on theallo wedlexi calto occurhead ofon athe construction lexical heads to indiof itscate attri thebute case categorforms andies .

2.4.1 Case Relations It is widely recognised that the subcategorisation mechanism of the so-called "Standard Theory" that grew out of Chomsky's Aspects model cannot adequately selaccountectional for syntrestrictiacticons and bet semaweenntic categ structuresories are of conce a languagerned. , Fiesplecialmorelly's wherecase vetheoryrb in provided terms of a partly partial sema remedyntical byly statingdefined the "c aslinges"uistic like Agent envi ronm(A)ent , Ob ofje cta deca(0) , deIns oftrum popentula rit(I)y, , andcase Expe grariencermmar, li(E)ke . otherUn fortunatesemanticlyall, y afteror logi nearlycall y aba sed hasgrammars almostsuc comh pletelyas generative negle ctedsema nticsthe actual, has endedlingu isticup in stra bluctureind alleyof sentences because init thesentenc procees.ss ofTh istrying results to des in cribea schis them betweenexternal deep situations and su rfreacefe rredstructure to by , those powecorrerfulsponding mechanisms to that which between allo wsema arbitnticrarys and and synta unnaturalx, and mapin pingsthe int ofroduction situational of repre(cf. Stasentarostations 19 onto82d: 11-1ling7)uistic. representations known as "surface structure" Unlikerelatio nsother in vea rsionssentence of by case app grammalying preconceivedr, lexicase does sema nticnot ident definitify ionsthe tocase the esextertablisnalhed sit primuationsari lyrepre on thesent edbasis by theof syntactic sentences crit. erInsteaia, d,and case are givenrelatio ns are intenscharacteionalristics sema nticthey dearefini foundtions to inhave acco. rdanceWhile wit thereh the are ranges no dis ofcovery semantic 43 proceduresFillmore (s eefor Fil caselmore re la19tions71; Ple, a inescouple 1976 of:1 heuri8-32, stic95-104 principles) and re finedproposed by by presentedStarosta (1in978 sectio) cann be2.4.5 employe andd, another one of being them beinga requirement the �'l/Sent of co-reprincifeple"rent iality ofor thereferen sametia casel inclus relatiionon inin orde a sirngle for clathe useoccurrence to be pe rmof issimoreble than. one instance For the description of languages, lexicase posits a small universal set of case desrelacriptiotions. n ofFrom any thparticularis finite setlang, uagea.su. bset can be appropriated for the Athe com courseprehensiv of thee set deve oflo casepment re oflatio thens lex(seeicase Fig modeures l.2.5 Caseand ' 2.7relations) has evolved (as in opposed to case forms) are by convention given names not ending in -i.ve recent(cf. Ha rmonlist of1977) case and re arelations symbol is ispresen'ed byted three in Figcapitalure 2.5letters below.. The most AGI' Agent CORINS InsCorrespondenttrument (formerly "Dative,j. or "Experiencer") . LOC Locus (formerly "Inner Locative") NCRMNS MeansIncrement PLCPAT PaPltiaceent (f or(fmeormerlyrly "O uter"" Locative or "T")hem e") TIMREF TimReferencee (formerly "Benefactive") Figure 2.5 Inventory of Case Relations (Starosta 1982d :17) Other case relations proposed and used by various lexicase grammarians include DeGuzmanComitative 19 78)(Taylor, Directio 1971;n Ku(Tllaaylorvani 19jaya71 ; 19Ha74rmon; Cl 19ark77) 19, 78;Factive Ikra nagara(Lee 19198074),; Mann1980)er, Neutral(Starosta (L ee1971 19b 74)and, Reason1974; Taylor (DeGuzman 1971 19; Kulla78; Harvamonnija ya197 7)1974, and; Ikra Renagarasult (I(Tkayranagaralor 1971) 1980) . Alte, Obrnajecttiv ore Obnomjeenclactivetur fore inclPatienudest Theme(Taylor for 1971 Patient; Li 1973; KuLocationllavan ijaor yaLocative 1974; Leefor 1974Locus; Cl(Taarkylor 1978; 1971 DeGuzman; Li 197 193;78 Ku; llHiavanmonnija 19ya77) 1974 , ; ComitativeClark 1978 ; (HIkarmonranagara 1977) 1980, the; DeGutermzman Corre 1978,spondent Harmon (C OR)1977) recentl, Companiony propo forsed to replace BeneDativfite and or ExpeBeneriencefactiver (F(Faagangan 19197979; Starosta 1981982d)2d) ,R,andeferenc Concomitante (REF) to (MNS)(CON) andfor CoInmitativecrement (NCR)or Com arepanion intr oduced(Fagan by1979) Sta. rostaTwo (MnewS b) case . Forrela antions overview: Means Figof nomure encla2.7 beloturew. for case relations since the inception of lexicase, see The numbers 1-13 in Figures 2.6, 2.7, and 2.8 represent usage by different lexicaselanguages gr. amm(Sariee ansBibli at ogradiffphyer entfor peri detailedods for pu thebli cationdescript dataion.) of Figdiffureerent 2.6 idefirstntifies appe arancethe works, and cit(ced) la innguage Figures(s) de2.7sc andribed 2.8. by (a) reference, (b) date of 44 ITEM REFERENCE FIRST APPEARED LANGUAGES 21 TaylorStarosta 19 197171 b, 1976d 1971 SoraJapanese 43 Ku1Li 19731avanija ya 1974 19741973 ThaiRukai 65 StarostaLee 1974 1974 1974 KusaieanFormosan 78 ClarIkranagarak 1978 1980 19751975 MeVietnamlayu Betaese wi 109 HaDeGurmonzman 1977 1978 19761977 TagalogKagayanen 1211 StarostaFagan 1979 19 78 19197978 EngMonolis-Aluh, , GeSormanlomon, etcIs. 13 Starosta 1982d 1979 General Figure 2.6 List of Some Linguistic Descriptions in the Lexicase Model . (This list is the basis for referenci ng items 1-13 in Figures 2.7 and 2.8) CR Description Article/Thesis/Dissertation 1 2 3 4 5 6 7 8 9 10 11 12 13 [+AGT ] Agent/Agentive + + + + + + + + + + + + + [ +OBJ+THM] ObThemeject/O bjective + + + + + + + + + + [ +NC+PATR]] PaInctirementent + + + [ +DAT] Dative + + + + + + + + + [[+ +COREXP] ExpCorreeriencerspondent + + + + [ +DIR+LOC] DirectioLocation/Locative/Locusn + + + + + + + + + + + + + [ +T+PLCIM]] TiPlacme e + + + + + + + + + + + + [+CONINS] ComiInstrtativument/Inse/Compantrumeion/ntalConcomitant + + + + + + + + + + + + + [[+ +MANMNS] MannerMeans + + + + + + [ +BEN+REF] BeneReferentfit/Bene/Referencefactiv e + + + + + + + + + + + [+NEUTFAC] ] NeutralFactive + [ +RSN+RES] ReReasonsult + + + Number of CR's 9 10 7 9 * 9 9 8 8 10 7 8 10 Figure 2.7 Summary of Lexi case Nomencl ature and Use of Case Re lations 45 Siindicatednce Star ostas ait emdealt 5, withthe featuresa number markedof Formosan under lait nguagesem 5 repre insent his a1974 union paperof CR 's Tfoundhe total in thenumber descriptions'of CR's atof'the all bottom the Formosan of columnlang 5uages is there treatedfore indiin thatcated pa peonlyr. byis notan asteriskincluded (*in) . thisVene tabulationeta Acson beca's desusecription she does of not Greek disc caseuss marcaseking relatio (1979)ns , becauseand Timothy his Matrenleyatment's work is not on pureSre stlyructure lexica se(M. anley 1972) is also excluded o sectIt ision intere 3.1)sting , there to. exists note that a co, nsistgivenent the syntacti assumptioncally sif Patientgnificant primary semantic (see ' diffAgenter encande Insbettrumweenent two, forsubsets example of , caseis not re laati matonste. r ofThe animacy differ encor eint betenweention , as has been assumedin earlier work, in the case grammar framework, but rather one ofactant immediacy directlyof orrelation indirec totly the im piPatngeients on (cf. the StaPatientrosta characte1978) . riWhetherses a an distStarostainctio 1982dn bet:2ween0-22 members): of at least three pairs of case relations (cf. IMMEDIATE (INNER) INDIRECT (OUTER) INS AGT LOC PLC COR Figure 2.8 Pairs of Case Relations Distinguished by the Immediacy of their Rel ation to the Patient Whereas the Agent is defined as the non-immediate perceived causer of the efactionfect ivede sigcausenated of by the the act veionrb, orthe event Instr reumferredent is tode byfined the asmain the pr immediedicatateor . Therespe Agentctively and in the a chainInstru meof ntcausation represent. thePrecis firstely andthe thesam elast dis tinlinkctions of Wheimmediareas cythe ap pliesLocus tois dethefined semantic as the charac (abstteriracsat tioor nconcr of ete)Locus locationand Place of. the Patiedistinntction, Plac betweene desig Correspondentnates the general and setReferentingt ofis theexactly action analogous or state to. thatThe betweenin an imm Instediarumte entcorresp and Agentondence or re Locuslation and to Pla thece Patie. Whereasnt, the CorrespondentReferent is the st ands toframe theof Patient reference its elf.for theThe whole disc oveactionry of, thusthis having parall elisonlym anamo indireng thect caserelat roleion detakefini thetions Patient and functioto be nsthe, accordingfundamental to andSta rosindita,spensable is made casepossi roleble only(op.ci whent.:22). we

2.4.2 Case Forms and Localistic Features 2.4.2.1 Case Forms asIt hasthere been are observed language thats (Fill casemore rel 19ations68; McKaughan are manif 19est62ed; Constantino in as many overt1965; waysand preReiddic 1964table and from 1966) case, and re lathattions the alone overt because case formthe mappcategorying isis notnot bi ne-uniquecessarily, 46 i.e., there is no one-to-one correspondence betw�en case relations and case offorms subje. ct-predicateFor these reasons and of, anddirect because object it, indirectis desirable objecto t, separateetc. from the the notions notioncase form "case as werelalltions as case", lexicaserelation. finds The it em neplcesoymentsary toof marktwo setseach ofnoun case N for difeaturespenses withinwith tra a nsforsinglemat leveions l andof descriptionreferences toena thingsbles a lilexicaseke "covert gramm roarles to" (C(Fookillmore 1971 1968:15-1:36)-4). or mechanisms like "required co-reference deletion" As with case relations, the set of case forms is also posited to be universal. AsFrom Joe thl isL. setFagan, a su(1bset979:3 can1) aptlybe dra putswn toit: account for any particular language. Case forms in a lexicase grammarare not language specific syntac(Fillmoretic devi1968c:2es1), . as theyInstea ared theyin Fil arelm oreantaken cato sebe grammauniversars lly recoglanguagesnisable by differsyntacticent devcaseices marke whichrs. are expressed in different Case forms are features marked on lexical items including nouns (N) , prepmorposihologicaltions (P)desinences, and dete whichrmine markrs (D themet) , arebut ofthe courseparticula languagr lexicale-specif itemsic. or The numberof distinct case forms varies from language to language. It seems hasthat atthe lea lowerst a nomlimitinativ is eset case at twoform with NM (tStahe rosurfacesta cl aimingstructure that gramm "everaticaly language su1978bje:4)ct) . andThe an uppe accusr ativelimit reachedform AC by(s urlexicaface stse ructuregrammarians nonsubjso ect fars) is" (Seigtarostaht, with Nominative [ +NMJ , Accusative [ +AC J , Range [ +RJ , Instrumental [ +IJ , Benefactive [ +B J , Comitative [ +c J , Manner [ +MJ , and Locative [ +LJ lexica(Kullavaseni grammariansjaya 1974). for Figurethe des 2.cription9 presents of variousa summary languag of casees . forms used by As in Figure 2.7, the features marked under item 5 in Figure 2.9 represent a papeunionr. of CFThe's total found numinber all of the CF Formo's atsa then labotnguagestom of treated column 5in isStar thostaerefor'se 1974also ofindica the teddiscuss onlyion by , anthere aste risis nok (*indication) . As for as ittoem whether 11, due theto thelist gene of CFral's nat isure intended to be exhaustive (cf. starosta 1978:4-5). The total number of CF's for(Sta thisrosta it 1982d),em is casethere formsfore als�are notindica ditedscuss byed, an henceasteris itsk. omissInion item in 13Figure 2.9. Case("Gr ammformsatical are Renecelatiossnsary") toas characte"subject-of"rise suchand gr"direcammaticalt-object-of" functional a sentenc notioens ca(Chomskyse. According 1965:68-74 to) , Filnotionslmore which(1968: have6-8) , causedsuch confusionmuch con fusioarisesn infrom the thest udyfact of ofthat the, insente traditionalnce" was wronglystudies ofassu casemed toin beLatin st raigandht Greekforward, theand concept clear and"sub thusject clathe ssifnominativicatorye wascriteria large lyfor ne caseglect usesed. haFilve beenlmore confusedalso po intsin theout pa thatst, the esWhenpe ciallit comesy for tothe the use dess ofcription case which of no n-Ican ndo-Ebe clauropessifiedan la onnguages synt actisuchc grounas theds . Phitendencylippine to labreanguagk esaway, evenfrom more Indo-Eur confusopeanion traditionarises as anda result proli offerate a prevalent innovativ e terminothe termlog "ty op(cic"f. McKaughanfor gramm atical1973: 121-1subjec23)ts . byInflu Philiencedppin istby s thesuch common as McKaughanuse of 47 CF Descriptions Articles/Theses/Dissertations 1 2 3 4 5 6 7 8 9 10 11 12 [ +NMJ Nominative + + + + + + + + + + + + [ +LJ+ACJ LocatAccusativive e + + + + + + + + + + + + [[+ +CI J] ' ComiInstrumentaltative + + + + + + + + + + [ +G+BJ BeneGenitivfactive e + + + + + + + + [ +D+oJ ObDativejecti ve + + + [+[ +MRJJ MaRangenne r + + + [ +R+DRJ J ReasonDirecti onal + + [ +CN+ABJ ContrastiveAbsolutive + + Number of CF's 764 8 * 5 7 766 * 6 Figure 2.9 Summary of Lexicase Nomencl ature and Use of Case Forms (s(1ub962)jec, tivFilallmoreisation) made anda distinct "secondaryion betweentopicalisa "ptirimaryon" (Filtopicallmoreisa 19tion"68:57 -58). McKaughanHowever, this (1973) term hasinologi reconscaidel infredelic theity use ha ofs now the been term rendered "topic" byunnece manyss ary, since Pha comilipletelyppinists app andropriate concluded term that for th thisis term category wasmi . sleading and that "subject" is Proposing a set of case form features apart from case relation features in sulexicasebsets within helps tothe sortset outof formalcase relatuniversalions andgr amm caseatical forms fu nctas naturallyional notions. bounded passThus ivesuch vo notionsice, etc. as willsubj beect accounted, topic, dir forect with and re indirferenceect obto jecasect, activeforms, whandile exill-pldefinedained in notions terms ofsuch case as re"llaogicaltions . subject", "logical object", etc. will be

2.4.2.2 Localistic Features Case forms may also carry semantic information of the sort that indicate locatforms iosucn, h directioas Nominativen, or orientatio [NomJ, andn. posOthesibler Accusativethan the purely [Acc J syandntactic Genitiv casee [Ganden directioJ, case nfor compms onents.in a lexi caseFor examgrammarple, are Cl arkanalysable's analysis as ofcom plexeVietnames ofse location (1978:39-50) employsJ such localistic features as [±lcnJ (location), [±dir] (t(direrminusection)),, to[± srcsubcate(sourcgorisee) , the[± extlocatJ (exiveten caset) , [±formgol [+] LJ(goa whichl), andrealises[± ter ] the case reformslat ionsis there[+LOCJ,fore [+"lTIocalMJ, istic"[+DAT], (cf and. Ande[+ OBJrsonJ. 19The71; Acsonlexic ase19 79)approach. to case Whinfleile ctionsclassi calinto lo meacalisticningful thecompooriesnents were, inconcerned lexicase wi theth thelocal analysisticis ap proachof case is 48 extended to prepositions, relator nouns, and even verbs as well '(cf. Starosta use1978s :4-5)local. isticFor caseexam pleform, infeatures her st udysuch of as Tagalog[+loc], verbs [+dir ],(1978) and , [+Videagoal ] DeGuzman susupplembclassifentedying by Tagalog morphological verb st emsand voiandce-r analysingelated Tagalogfeatures voisuchce asinfle [±ergction] in. In Veneeta Acson's analysis of the case-marking systems of Classical and Modern areGreek analysed (Acson 1979into )local , the measemanningtics compof preposonentsitio inns terms as weof ll the as irof semacasentic infle ctions casimisela inflritiesectio andns d�ffcombineerencinto es. a unIn ifClaiedssi systcalem Greek of case, prep marositking.ions andThe nominal ofgramm theati meaningscally sig ofnif theicant case mea reninglations of aon prep theosi headtional noun phrase and theis thenlocalistic a functi on prepofeaturessiti onons thein prepClassicositialon and and Mode onrn theGreek case according-inflected to noun the. foAcsonllowing subdivides set of cilocalrcu mamsemanticbience, featursourcees, : intirelamatetion, alinterior, surface, prolat, associaiveti, on,termin veusrti, calpath, comitative, side, , assigandprox nedimate to prep (Aositicson ons1979 are:240) also. Loasscalisticigned to featuresinflected com nompatibleinals witto haccount those for the casefact thatrelati thonsese. inflOfec thetions fiv adde case infor inflemationctions to thein Cla meaningssical contributed Greek, namel byy, nominativnon-locale, cases vocative while, dativethe rema, accusiningativ threee, andare genitivelocal. , Eachthe offirst the twose threeare obas lishoquewn casesby the is folidentilowingfied case by ainfle specictionfic matrixtree (i bidconsist. :63)ing : of local features, +N ±plI

. ). ,

DAT ACC GEN NOM Figure 2.10 Basic Cl assical Greek Case Inflection Tree Acson has thus analysed all case form features (except [Nom]) in Classical Greekcorre spondintoences local istic(cf. op. features cit.:2 45)in accordance: with the following set of NOMINATIVE: [+NM] ACCUSATIVEDATIVE: : [-[+drcn],- sorc] drcnsorc "d"sourceirecti" on" GENITIVE: [ +drcn, +sorc] 49 Localistic features are treated as "sub-CF features" by Clark (1978:39) and characteappropriaterised in as the "l ocalsense semathatntic loc alisticfeatures" features by Acson are (1 979sema:47)ntic. featuresBoth te ofrms are intolocat meaionningful, direction compone, andnts orienta. tion that further analyse case formfeatures Having been analysed into meaning localistic components, conventional lexicase CF"Ab'slativ suche" as, "D[+ACativ] e"and, etc [+L. ] areas wein lla strictas corresponding sense no longe tradir tionalneeded labeas als formal such as part of the grammar. They have been supplantedin all. cases except [+NM] by cl"caseusters form" of alongsidelocalistic loc featuralistices. featuWhetheresr isor renotal lwey a shouldterrninologi retaincal the qu estioterm n. Howeveand probablyr it is unive resolvrsaled , setit , remainsallowing true the that theory these to accountfeatures fordo theform fact a co herent (1) that all languages can e�ress the same set of concepts, evcontributeden though mainlyparticular by case aspects infle ofctions the meaning . (Hungarian) may be, prepositions (English) , or verb classes (Tagalog) i and as(2) inthat coverb the derivalocus tiofon the (tsehe featuresdeTivatio cann ofswitch prepos cateitionsgori es, frominflect verbsions; Clfromark captured1978) and postp theosit formationions (S tarostaof case 1978:5). Therealis incorporationation beyond of nom localisinal conm stitintouents lexicase and alto lowsbette usr tointerpret extend thecase verbal case markIn mying analys or "fisocus of '" Natasystemoran inArni las,nguages I have ofalso the usedPhil loippinecalistic type featuressuch as Tagalogto . themcharacte the risemeaning and henceof direct subcategoriseion. some of the motion verbs that carry with In the present study, I have gone one step beyond Clark, DeGuzman, and Acson in replocalalisticcing all feature CF features on nous ns(wit, preph theosit exceptionions, and of de Nomteinativrminerse). byThat corresponding is, the 'Jcaseohn Andeformsrson of earlieras being lex arbitraicase ryan alyand/orsis , redundantwhich have (A beennderso propen 19rly77:6 4-criticis69), haveed by beenlocal replacedistic "case by relocallatioisticns" possemantitedic by com Andersonponents whichhimse lf.cl oselyThe aplexicaproximatese the elapproachements , inhoweve the lexicalr, still repredifferssentations from Ande ofrson words's fr(inamework surface in str locaucturting�) theratsehe r thisthan loin calisticextremel yanalys abstractis of synta Arnisctic- case semanformstic, see repre sectiosentation 4.1ns. . For details of manAs mentionedifestations above are,. caseCase forms forms are indi notca tinglanguage the prspeceseifnceic, of but certain their case relations are in turn given overt ·expression by the case marking system of the exprelanguagess case. The rel ationscase marking is lang mechanismuage particula used tor. expliDependingcitly mark on thecase la ngformsuage and, it mayprep beos itdoneions throughor pos tpnounositio orns ve, rbconstraints choice, aff onixation word ord ofer, the caseverb infle, use ctofion of nounsdevic esor. pronounsIn other and words articles, languages (and adjusee ctivspecifices), or lexical a combination, morphologic of theseal, and formsyntactic on a nomindevicesal toconst directlyituent . or indirectly indicate the presence of a case Camodise ficamarktioning hasof abeen noun undersstern,toodas is as foundan overt in case-inflemorphologicalcted afflanguagesixation suchor as 50 Latintraditional, Greek notion, and Te oflu gu.case infleThis ctiontype . of caseIn recent mark ingtre correspondsatments of caseto thegr ammar, howeveor morpr,hological the concept configuration of case mark thating has ha s thebeen sam extendede function to coveras the any case syntactic morpinflechologtionsica dol prope in Latinrties, incof theluding NP 'sbut and not pp li'smi whitedch tobea syntacr casetic re andlatio ns. In ordelexicar tose hascapture broadened cross-li thengui conceptstic andof dicaseachro marknicing gen toera includelisatio nsgramm, inatical fact , caandte verbalgories mainflectionifestedn by(DeG non-nominaluzman 1978) de. vices such as verbal subcategorisation

2.4.3 Patterns of Correspondence between CF and CR In the previous sections we have shown that case forms are grammatically asthesocia othetedr. withWh ilecase case rel markationsing on is the la nguageone hand specific and with, the overt associ caseation mark orings on maycorrespondences be of univers betweenal inter caseest . relations and case forms demonstrate patterns that It is important for the description of any language to give an account of the wayscase informs whichare caserealised relatio. ns Of correpartspondicular to intere case stfo rmsis theand tendencthe waysy orin prewhicferenceh whenfor certainsuch a casetendency rela tionsis manif toested be associa by a groupted with of certainlanguages case. Oneforms such, esp ecially genecorrespondenceralisation patternsthat can beis thedrawn "S ubthroughject Choice cross-li Hiengrarchy"uistic stforudy accus of ative betweenlanguages "d eep"(cf. Filcaselmore rel ations1968) . and In"su Fipelrfilmorecial"'s framgrammeworkatical, the su bjcorrespondenceects is treated variousas "sub jectmechaniivalsmsisa tiobyn" which. According"deep structures to him, sU...bjec aretival conveisartedtion intois onesu rfaceof the reprecase resenlatationstions, deepof sent stencructurees" (icategoriebid. :32) s,. ontoIt issu arface trans conformatfiguratioionalns mapp. ingThe of strrelaucturetion "su. bject-of" is transformationally introduced prior to the surface In lexicase the correspondence between CR's and CFts (including sub-CF theloc alisticsame feature feature matrixs) is exprein a ssedsingle by levejointll ofy reprspecifesyingentation the . CR Forand exathe mpleCF in, Nominativethe feature scase [N omfor,+PAT]m is markedassocia inted the wit hlexica the Patientl matrix caseof a relnounation mean. thatBy the ordeconveringntion has, the no CFtheor precedesetical thesigni CRfi incance the. matriLocalcesistic, although featur esof , coursebeing thsub-CFis [+feL,+atursresc,+, extalso,+LOC precede] and CR[+ L,features+ter,+D ATin ] thefrom li Clstingark . (1978Exam:4ples2,49) are. : Starosta (1978:6) suggests a rather neat way of characterising four of the majorConsidering syntactic case word to be classes an inherent in terms feature of the ofse nou CRns and, StarCF ostafeatur thinkes. s that the categorycase rela Nountion couand lda becase univ form"ersally as oppo definedsed to as Adve"tharbst cate, whichgory may which carry car onlyries a CR's, and Verbs and Prepositions, which may be marked for CF but. not CR. Theintere correst spondencein grammatica betweenl des CR'criptios andns the, si nominativence the case case form form [N omis] ofis parfrequentlyticular 51 equatedlexicase with fram eworkthe "s, ubtheject" assig in nmaent subj ofect [N-promedi] tocate a nounconstructionma kes that. nounWithin the the grthisamm [Natom]ical constisubjectuentt of ais sentthe enccasee. relationWe say chosenthat the or part seleicctedular asCR su marbjkedect .ofon theand thesentence requirement. Subj ecit tpla choicesce onor co-occurringsubject sele actantsction is, eadependentch of which on theis verb associasubjectted choice with, theFil lmoreverb in(1968 a part:33) icmaulak�sr casethe folrellowingationsh observationip. With : respect to For most combinations of cases there is a 'preferred' or choice'unmar ked- the' su bjecsubject tchoi is seunique; for lyso medetermin there edis. no Inactual general thefollo 'uwingnmarke ruled' : subject choice seems to follow the thereIf there is anis anI, A,it beit comesbecomes the thesu bjecsubject;t; otherwi otherwise, sethe, if subject is the o. Iffor we Engl appislyh Filat lealmorest,'s if hie anrarchy actant to is a markedlexical [+ analysAGr],is it, weshould would norma expectlly beth at, chosenthere isover no [+thoseAGT], marked then by[+ INS[+INS] is] orchosen [+PAT ove] ras [+suPATbjec] whent of boththe sentare encepresent. If. notionOtherwi se'sub, [+jecPATtival] beisacometions the' is suusebjecfult. onlyFil iflmore there further are sentences notes th atto , a "tlahenguage whviewich, offersubjec at choice providesof subje ct"a conv(op.enient cit. :5way8) . to Fromaccount Fil lmorefor typo's pointlogical of notiondistinctio of nscla. use Intypes case andgr ammar"provide terms the, thecategorial various arraysand con offig urationalcases define the inlangformationuages (i biford. deter:54-55)mining. Fil thelmore surfac ha�e distiredefinednctions the thattraditional are found" te rmsin "nominative" and "accusative", "ergative" and "nominative", "active"0 and tra"inactivnsitivee" , andby referringintransitiv toe sentencthe surfacees. manifestation of his A and in Ledeterminationxicase has furtherby defining developed Patient the asFil thelm oreancentral concept case ofrelation case formand by esSubjectablist hingchoice case and formstypolog as y a canseparate then beparameter formally forand gramma explicticitlyal descriptionexpressed in. ertegative-tyrms of thepe assoclanguagiationes, [Nof om[N] omis] withalways a mappedcertain ontoCR in[+ thePAT]. follo Inwing non- wayfocus: in lainnguages accordance of thewith accus Filatlmoreive-t'sype, hie rarchy[Nom] . is normally mapped onto a particular CR AsTraditiosociatednally wit, h voicesubjec referst choice to arethe mothedifica notionstio n ofof "vtheoic vee" rband stern "focuto s" in. dicate diffof traerennsitcesive in verbs subj ectin Englchoiishce. reqForuires example that , thethe ver chob icappee ofar [+ inPAT the] aspass subjecive t voas icethe. subjTheect verb instea is d,sa idnorm to allybe in in its accord activ wite hvoice the Filif solmoreanme other subj CRect is choicechosen hiesubjrarchyecthood. forIn laa nggivenuages ve whererb, the more two-way than distitwo CRnctio's cann between be candidates active andfor pasMarasinaove , voicefor instamay bencecome, some in adequateverbs have to are gisterconsiderable verbal amountCR concord of freedom. In to Thechoose subjec the t-msubjecarkingt from prepos amongitionthe so [+ isAG T],asso [+ciaPAtedT], wit [+hIN whicheverT], and [+ REFact ant] CR is's. 52 chosencat�gory to ofbe thethe NPsu bjecmarkedt, bywh ileso a(cf. verb McKaughanal affix indicates1962:47; Filthelmore case 1968rela:5tio5)n. VoicePhili ppineinflec andtion For associamosan tedlanguages with multiple is comm onlypos sibilitiesreferred toof assu bj"focuects" choice. in Maranao examples from McKaughan (1962:48; cited in Fillmore 1968:55) showing Agent-focus (AF) , Patient-(RFfocus) (OF; from "object-focus") , Instrument-focus given(IF) , andin lexi Referent-case notatiofocus n. areAlthoug givenh abelo thorow. ugh (Tlexihe featurecase an alysismatrices of Maranaoare [Lcouldcv.+P�T] mean reasfor kosig nmkaentraba oof insome 2.3 case2c, I relationshave deli, beratelysay, [L cvchosen,+LOC ] theinstead present of anfocusalys. is) to illustrate the supposed relationship between subject choice and (2.32a) AF: somombalj? so mama?· sa karabao +[+V +AGT] Nom+AGT Ace+PAT +[ T] ( J ( J - -AGTAcc -;:� -PAT the man butchers the carabao f{:: sombal j?jn mama? so karabao (2.32b) OF: f l 0 +V Gen Nom +[[ +AGT] (+AGT J (+PAT ) - -AGTGen the carabao is the thing that the man butchers jsombal j? mama? so lat (2.32c) IF: :f�;�:l 0 ge ko ka rabao +V Gen Nom +[ +AGT] (+AGT } (+INS } ���T +[ +I+PATNS] ( } - Nom-INS - Gen-AGT - Lcv-PAT it is with the knife that the man butchers the carabao 53 (2.32d) RF: somba! j?an 0 mama? so major sa karabao +V +[ +AGT] (Nom+REF } (ACC+PAT ) +[ +PAT]+REF] - Nom-REF - Gen-AGT - Acc-PAT it is fo r the mayor that the man butchers the carabao

In each of the sentence examplsombaes, ! thej ? subject marker so andthe focus or voice inflection on the verb stern to butcher are underlined. From these examples, we can also see that the correspondence between CR's and CFcan's be in matched such languages with any isone not of one-the to-onefour case. The re lanominativetions: [+ AG[NT],om], [+ PAforT], exam [+pleINS,], threeand [+ diffREF].er entThe case [+ PATforms] case: [A ccrelation], [Nom],, in and these [L cvexam]. ples, can be matched with

2.4.4 Case Frame Features Theframe array. Case of caseframe relationss are contextual that co-occur features with which each alvelowrb primaryconstitutes verb itssterns case to beabsence broadly of parsubcategticulaorir casesed inrelations terms of inthe the grammat verb'sica syntaclly alticlowab contextle presence. or Accordingset of ma jorto verbStarosta categorie (1978:3s.4) Every, "a lexi verbcase in thegrammar languageal lows must only fit a intovery onelimited of probaby derivationbly eight or or bo ninerrow categorieing must s,fit anditse everylf into new one ver bof thatthese enters same thecategori languagees". weUsing have the nine inherent possible case synt relationactic verbfeatures categor [+PAiesT], corning [+AGT],from [+ branchesCOR] and of[+LO theC], treereplaced in Fig byure COR 2.1 in 1 the(cf. current Starosta framework ibid.). LettingNote that X therepre formesentr anyEXP CRhas feature been , [-[+[+X]] means non-co-occurobligatory co-occrence urrencewith a witsisterh X. constiFor tuentexample marked, [- [+X,AG whT]]il meanse "d"mustoes occurnot occur with inan theAGT envi sistronmenter". of a sister marked AGT" and [+[+AGT]] means Contextual features may involve more than one feature. The one or more verbcontextual may occur features const whichitute thejOintly verb's speci "casefy thefram casee". -rela Theted case context frame inis which shown a in ana row unorde andred separated collection by commas of case or-r elastackedted contextual up in a column featur. esIn, ei extherample lined 2.3 3,up the in VI(aI) andlocation (b) forms verb aresuch equivalent as "put" (cf.repr eseFigntationsure 2.11) of. the case frame for a Class 54 (2.33) (a) [+[ +PAT J J ,+[ +AGT J ,-[ +CORJ ,+[ +LOC J J (b) +[ +AGT+PAT J (+[-[ +LOC+CORJ 1

I [-[ +AGT J J [ +[ +AGT J J [-[ +COR�J J [+[ +COR J J [ -[ +COR� J J [ +[ +COR J J [ -[ +LOC J J[ +[ +LOC J J [-[ +LOC J J[ +[ +LOC J J [ -[ + LOCGJ J[ +[ + LOC J J [-[ + LOC� J J[ +[ + LOC J J II �III IV� V VI VII VIII IX Figure 2.11 Subcategorisation of Verbs by Case Frame Features

Forlayers al l ofcontextual square brackets featur esare, inclurequirdinged , casethe innerframe bracketsfeatures markedenclosing on verthe bscase, two whofeaturele feature that sermatrixves asof the the contextlexical and item the and outer enclosing brackets in formationrepresentingon the co-occurInformationrence on (msequearkedntial by +)order or non-co-occ is given byurrence a li ne(marked before byor -) after in suchthe contexinner t. brackeconstittsuent to markedindicate by respectivelythe particula ther CRpo sitifeatureon preceding. If the or li nefollowing is omitt theed , it metheans consti that tuentthe co withnsti tuentthe sp withecified the CRCR. in Le questttingion X bemay anyeit hernon-contextual precede or follow [+feature_[+xJ],, we[+[+ mayxJ _Jhave, poandsitiv [+[+e xJJ,contextual meaning featuresobligatory of occuthe rrencfollowinge be foreforms, : occurrenceconstituent aftspeer,ci fiedand byco-oc [+xJcurrence (with withor wi thout(irrespectiv intervene ofing order) materia thel) , sistander the negativnon-co-occe counterpartsurrence before [--,- [+afterxJJ,, [-and[+ xJwith J,the and co [-nstitu[+xJJent meaning specif respeied byct [+ivexJ.ly Non-ofeaturebliga(s) toryin the co-occurrence inner brackets is indi; i.e.cat, edthe by featurethe use or of features parenthe thatses aroundserve asthe context. Whereas [+ [ +x J J reads "must occur before [+x J," a contextual Itsfeature negative of the counterpart form [+_([ is +x formJ) J allyreads supe"mayrfluousor may, notand occuris not beusedfore. [+xJ". Theverbs case shown fra mesin Figcharacteure 2.1rising1 are asthe folninelows po: ssible major syntactic categories of 55 CATEGORY CASE FRAME ]] I [-[ +PAT ] [ ] [ ] [ ]] II [+[ [ +PAT ] ,-[ +AGT ] ,-[ +COR] ,-[ +LOC ]] III +[ +PAT ] ,- +AGT ] ,-[ +COR] ,+[ +LOC ] ] IV [ + +PAT ] , -[ +AGT ] , +[ +COR] , -[ +LOC]] V [ +[[ +PAT ] , -[[ +AGT ] ,+[ +COR] ,+[ +LOC VI [ +[ + PAT ] , +[ + AGT ] , -[ +COR] , -[ +LOC ]]] ] PAT ] , ++[+AGT ] , -[ +COR] , + +LOC ]] VII [ +[ +PAT , +AGT ,+ +COR , -[ +LOC VIIIIX [+[+ +PAT ] ,+[ +AGT ] ,+[ +COR] ,+[ +LOC ]] Figure 2.12 Case Frame Features of Verb Categories

Actusinceally the, neonlygative the poonessit areive predictedfeatures would by the have Om ega-ruleto be li sted(cf. insect theion lexic2.2.2on.2), . Hence the singie feature [ +[ +PAT ]] is sufficient to identify verb category II, while [ +[ +PAT ] +[ +AGT ]] uniq�ely identifies category VI. Aagreement double negatio relatinons notation between istwo adopted constituent in lexicase heads . convAention contextual to indicate feature of the form [ -[ +X,-Y ]] marked on an item of class [ +Z ] is to be interpreted as an enviobligatoryronment paradig of Z. maticIt would solidarity allow betweenthe fol lowingthe two three features combinations X and Y toin bethe marX,-kedY. on Fora sisterexample of, Z:the (1)associa +X,+Y,tio n(2) of -X CR,-Y,'s wit andh CF(3)'s -X can,+Y, be butst atednot in(4) terms +of features such as -[Nom,-PAT] , which states that if a verb has a subject at alitsl, suitbj mayect , onlyi.e. accept, to be a assocnominaliated actant with withthe [ Nomthe ] Patientcase fo caserm. reThela tnotationion to be [simply -PAT ] . states that, if a subject occurs, it cannot simultaneously be marked as An implicational dependency can be expressed using the following lexicase convention. Two features in the same,a matrixy]]' canbe associated by the logical implication symbol "::>", as in [::>[ +X to mean that if X is present, aY semis expecanticteim d.plica Thtionsis notation which canis beused violated to repre witsenthout sel neecectionssa rilyrest riaffctionsecting, derivationalgrammaticalit y.rul esThe. notation(see Rule is2.1 used3a above primar) ily in the statement of

2.4.5 The l/Sent Constraint In section 2.3 .six contraints imposed on possible grammarsby the lexicase moassigdel nmenthave beenof case lis retedla. tionsTo whichthese wecan nowbe addstated a very as follocruciawsl: constraint"no case role on the appears more than once in a single claus[ e" (Starosta 1982] d:5) . This constraint, onrefe Filrredlmore to 'sby "o Starne-ostainsta ncas e-p"Theer-c l/Sentlause" principleoneiper-sentence for case constraint"assignment is based strong(cf. Filerlmore constraint 1971:3 8;than Pl eithenes Fil 197lmo6:2rean1,9 8-104)counter butpart, turns sinc oute it to cannot be a muchbe evaded throughit can througthe postulath the useion ofof trans extremformationsely abstract in Fildeeplmorean structures case grammain thers way. that 56 Filviewlmore of theuses great situational difficulties roles entailed as crite byria this for pracasectic reela tioin nattem assigptingnmen tot. makeIn Stageneralrosta st (1atements982d:9) feaboutels thathow itcase is rebelastti onsto demapfine onto case sur refacelati onscon figsyntauraticticallyons, , witclauseh the con l/stSentructio conn"st. raintIn seimporvingsing as th ais "urequiremniversalent me, tatheorlexicaseeti iscal abprinciplele to cut of throughsituational the tangleroles andof muavoidltiple any semadirnticect reintferenceerpretations to perspe associctivesated which with come intol/Sent pla principley in Fillmore, the 'schoice new autonomousand assig nmdeepent ofst ructurecase re levlationsel. canBy beup holdjustiingfied the prein aference systema. ticSta wayrosta, lea isving able less to roomshow forthat su, withinbjective the gues lexisworcasek or fra pemeworkrsonal whereand st syntaraightforcticward criteria system are of the verbal prim arycla ssifconsiideratiocation andns , derivthereation can , becase a simple transassignmformatent ionsand suorbject arbi chtrarilyoice wit incrhouteasing appe thealing case to otherwiserelation inventorunmotivatedy. In his paper entitled "The One Per Sent Solution" (1978), Starosta demonstrates howframework some of. theFor problems example , raisedproblems by Filrelalmoreted tocan the be assigsolvnmented wit hinof case the lexirelacasetions occurrenceto NP's wit hof part/w time holeand locatioor inclusn actantsion relatio, problemsns, as which well forcedas to the Fil mulmoreltiple to add adequaa numbertely of handlednew CR 'swithi suchn asthe Pathlexi, caseSource framework, etc. towit hishout inventory the addit, canion beof new theand seem probpiricallylems by de bilitallowingating for formal mult iplemechani occurrencesms. sInstea of CRd,'s lexi withcase shared hand les rel/Sentference con, strandaint new makes CR's. correctMore primpoedirtactionsntly , aboutStaro stthea prodemopertiesnstrates of that the morpnot hologicalexplained causativein other frameworksverbs, prope ofrties ling uisticwhich havedescri onptioly beenn. described but

2.4.6 Summary beenIn summary disting, uitheshed notions. Case of recaselatio renslatio arens synta, casecticall forms y andsig nificase cantmark ersemas haventic recasela tionsform betweenfeatures nomiare nalsfeatures and cointrnstituoducedent byhea dslexi. caseCase to forms account and for loc alistic tradicategotiriesonal suchgramm asatical nominative subject, accus and objativeect, , dativeas well, and as othercase inflecasesction, wh ich Paul morphoLi (197logical3:110) concallsfig theurations "surface wit h casesimilars", as fu wenctillons as . otheBothr synt caseacti rec laorti ons and macakesse formsuse ofare so mela nguageor all un ofiversal these . featurThe esdes. criptionCase marke of ars par aretic theular actual language overt langrealisauagetio spens cioffic case. Deprelaendingtions andon the forms la nguin agea sentence, verb class. esCase, prep markosiingti onsis or postpword osiordetionsr, or, dea tercombinaminerstion, reofla torthese nou mayns , benominal utili sedand/or for verbalcase mar inflekingction. , The(wit hl/Sent the probab Constraintle exception is intr oducedof [Nom]) in conneare notction subje witcth CRto assigthis nmentconstraint; CF's. .------_ .. -

57 Case frame features are provided as a device for the classification of verb betweenstems and features for stating in the co-oc samecurrence matrix inrestri a syntacticctions bet attributeween constit of aue ntshead, wordor . beIn presentChapter ed3,, eaandch the case re alisarelatiotionn usedof ea inch thecase present relati ondes bycription localistic of Ami cases will Chapterformfeatures 4. andcase markers in Amis will be described and illustrated in CHAPTER 3

CASE RELATIONS IN AMIS

(1974:301) AmisStarosta: gave the following list of case relations for Nataoran AGT, BEN, COM, OAT, INS, LOC, OBJ, TIM. with changes in labels to conform to current notational conventions, and correassigningsponds the to nam thee PLCfol lowingto the listformer: Outer Locative, Starosta's inventory AGT, REF, CON, COR, INS, LOC, PLC, PAT, TIM. semaIt isntic found characte in thisrisa studytion somthatetim alesthoug assoch soiatedme Am witis hno minalthe notions actants of meet RSN the (Reason), MNS (Means), and NCR (Increment)2.7 re alised as separate2.4.1) , case relations leastin some, theylanguages can be (cf.formally summaryas sigin ned Fig ureto one ofof the se ctionCR's listed abovein Ami. s at In the pretreatedsent analysas typesis , ofCON LOC, (Conco to bemitant) realised and byCOR distinc (Corretivspondee setsnt) , offor local insistictance , caseare genformer alfeaturisatioesns, thereby. reducing the CR inventory and capturing more Inlexic Amisase, dethes criptionssituational of otherrole associlanguagesated withis sig thenalled Manner exclu casesiv elyrelation by verbal in complementation, a device which is also used frequently in Amis as an plalaceternativ, ande timeto the. use of nominal actants to encode the notions of instrument,

In the present study, then, following the convention that all feature symbols- weare ha tove bethe enclosed following in bracketsinventory and of markedcase re eitlationher byfeatures a plu s marked(+) or on a minusAmis nouns( ),: [ +AGT ] Agent [ +PAT ] Patient [ +INS ] Instrument [ +PLC ] Place [ +LOC ] Locus [ +TIM] Time As we have mentioned in section 2.4.1, the lexicase model assumes that Patient ais cothensi stentfundam entalsynta cticallycase rela sigtionifin. cantFol semantlowing icth differenceis assumptio betn, weenStarosta two subsets finds of case relations, the inner and the outer CR's as defined in terms of the2.8 immediacyfrom sectionof 2.4.1their reabovelat ionshipfor easy with re ferencethe Patient. . Here I repeat Figure

58 59 IMMEDIATE (INNER) INDIRECT (OUTER) INS AGT LOC PLC COR REF Figure 2.8 Pairs of Case Relations Disti nguished by the Immediacy of their Re lation to the Patient ofIn REFArnis are, the subsumed inner COR under is treatedPLC. Nevert as a helesstype of, theLOC whdistileinction some of bet theween functions the otherinner problemsandouter involved case relation in treating still Agentholds . as anIt outernow appe casears rethatlatio theren, esp mayecia bell y with regard to the fact that Agent subcategorises verbs but thatp other outer cannotcase re bela tionsgone dointon't. in aHowev descer,riptiv theis work has broadsuch as theor thisetical. im lications which fromIn regard the Filto lmoreancase assig approachnment , wherethe lexicathe choicese approach of crit iseria sig nifiis concerncantlyed diff. erent Fila certainlmore operated given fu ncwithtio nthe with principle respect thatto a oncegiven an staentityte or isaction perce, ivedit isto have wasrequired obje ctivto haelyve thethe sasamme e externalgrammat icsital uationfunction. Howevin aller sentences, Starosta re (1lated982d :2to5-29 what) doefunctios notn toagree the wit NPh whose Fillmore referent about has always the asssameigning role inthe thesam esa megramma externaltical withsituation a change. Hein arguesFillmore that's the"pers grapemmatctive"ical. function of an entity does change In lexicase, case assignment may change in lexical derivation. with a change inro lesthe may case also fr amechang of e.ve rbs Th, usthe, inmap theping as betsigweennment caseof CRre's,lations one shouldand si tuationallook at propeThis rtis iesa fundamof theental sentence difference itself, betweennot only Starosta at the andsit uationFillmore it. correspondsThe to. regrammariansolution ofto thcaptureis di ffmoreerence syntact will icdepend gene ralon iswhichations approach. allows the This chapter will be devoted to a discussion of the individual case relations somein Arnis alternat, theirive re waysalisa tionsof expre, anssid, ngin sithemi laRamir notioficanstio. n section under each CR,

3.1 The Patient Case Rel ation [+PAT ] Thegramm Patientar. By case fundam relationental weis meanthe fundamthat entalif a verb case ha relas onlytion onein aco-occurring lexicase actant, it must be [+PAT], not[ counting] [of course] the non-inherent or peripheralany sentence case regar redlesslations of ofthe +PLCchoiceand of +TIMverb. thatIf cana verb occur has inonly practi one cally Patientco-occurring. It nomifollonalws actantthat allin theint ransNomitinaivetiv everbs case wit formh only, it oneis alwaysnuclea rthe actant hasubjve ectla Patientess) . subject if they have a subject at all (that is, if they are not 60

Theother Patient case rel caseations rel ationand compis alsolement sema typenticallys can be central characte torised a case in systemterms inof thatthe inneimmediacyr comple ofmethe ntsir are relationship essential withto the the su bclaPatiessifint. cationInner ofcase verbs rel andations are and, in thethat present the unexpre analysssedis , sutreatedbject ofas anob liginneratory com. plementAlso , shouldthere beis theco-r erequirementferential withCR's, the are suoptiobjectnal Patient. Moreov of er,the higtheher requirements verb. Outer that comtheplem subjentsect, oflike an outer embeddedof the higS herbe verunexpreb ,needssed not and app thatly toit outer should com beplem co-reentsferential. with the Patient

3.1.1 Characteri stics The Patient case relation has been previously labelled as OBJ (Object) or THM (Theme) in earlier lexicase descriptions. Starosta lat�r found these two oblabejelsct" towhich be pot, inentially its traditional confusing us sincagee, Obrefersject mayonly be to confusedan associa withtion "d irof ecthist caseIkranagara relation (1 980)with took the Accusativfrom GrUbere caseand Jackendoform, andff the who ter usedm Theme it to which refe r to this casein in reforlamationtion, stris useducture by to Prag reuefer Sc tohool the grobammjeariansct about andwhich the irnew fo lloinformatwers worionking, i.e. the theme, is predicated. In the present (1study968) , the term Patient(1970) is adoptedintended. and used in roughly the same way Lyons and Chafe have An entity marked by [+PAT ] is characterised by Starosta (1978:9) , following withGrUber which and Jackendit co-occursoff, as: one of the following, depending on the verb class (a) the verentityb, which is viewed as affected by the action of (b) thein spaceentity, which is viewed as moving or as being located or (c) whosethe entity state which is chang is vieweding. as existing in a state, or Fil"thelmore sema, nticin referringally most toneutral the Obcasejectiv, thee casecase reofla tionanything, also repre recogsentablenises itby to a be nounthe semwhoseant icrole inter inpretation the actio nof orthe st verbate iden itself"tified (F byil lmorethe ve1968rb is:2 5)ide. ntifiedIn by generalstate or, exiit stencerefers tois thatiden tielfiedement by which the verb.is aff ecItted is bydue the to actionthis wid, ore whoserange baskeof semat"ntic. Howev shadingser, becausethat Fil oflmore Fil lmorecalls's his re liaObncejec tivon e sitcaseuational relation criteria a "waste toand be sele fundamctionalental re striin thectio samnse, heway doesthat not it assuis inme lexihis caseObjec. tive case relation thatIn lex haivecase been, the treated Patient as casedisti relationnct case may re latsubsuionsme inseveral other case -ligrammke notionsars, incfor ludingexample Experiencer, these notions and Recansult be/Factitive treated aswith the intrainterpretationsnsitive verbs given. Into Sorathe , psych"neutroloal"gica Patientl and creativcase reela tioverbsn when (cf. itStarosta appears 1976respectively:1073-107 8)with . 61

Inintran Amisis,tive the Patientverbs alike case, inclurelationding is verbs indisp thatensable indicafor te existtransencitivee, andlocati "on, Aganden post, ansess Instrion. umentDepending, a Locus on, athe Refer verbent, a, Patientor with mayany occcombinationur alone , ofwit thhese an , providednot exceed that thr eethe. numberof nominal actants in the case frame of the verb does PatientCertain verbactants classes. For in examAmiples impose, meteoro specialogil calsel verbsectional imply impli thatcations their on su bjectheirts , sentencif any, esshould: be indicative of only time or place as in the following ma-kotem kina. remiad (3.1) cloudy this day [ +vJ ( cloudy today:� :T1 (3.2) caay he'nay ka [si-solda] ko Natao ran not ever have-snow Na tao ran ( � (+AdV ) [+p] ( � (Nom ·i:t neverint) snows+rnnnr in Nataoran�int ) +PAT) In these sentences, we can say that the notions of time or place are subsumed undersubject the of neutral verbs thatand baindisiccate Patient meteor caseologi relcalation phenom whenena. they occur as the In Amis, verbs derived from nominals that can be interpreted as being the sitrequationaluire abst ractinstr umentsubjec tsof thethat actionsare derived repre nousentedns. byHere the derivedis an example verbs : normally (3.3) kamay-en ko sa-pi-bacaq non iam do-by-hand means-of-laundering +v +N lexclGen +e�: } +abst (+LOC) abst "abstract" -( -PAT Nom+PAT we+inst launder by hand+inst Lit. our means of laundering is by hand In this sentence, the predicate is a verb and thus carries no case relation at alInstrl, whumeilnte thein accordancesubject nominal with theis asslexiignedcase therequirement Patient casethat CRrelation be as siginstead�ed of primarily with reference to syntactic rather3.3 than situational criteria. [The +PAT equ].ational Here iscounterpart a simila r ofexam exampleple: would also have its subject marked (3.4) o baqdet-ay a nanom ko sa-pi -bacaq non iam hot-object water means-of- laundering lexcl [ Neu] +abst+N Nom+PAT we launder with ho t water+inst Lit. our means of laundering is hot water 62

Theare seintroducedmantic dis intotinc thetive featurefeatures matrices [+inst ] of(i nstrnounsum entalsuch as) andsa-p [+i ab-bst]acaq (abybs tract)a imlexicalplica tionsderivational of denomi rulenal , verbsand thesesuch asfeatures kamay -ensatisf in exampley the sele3.3. ctional In this analysis, the subjects of non-verbal constr�ctions including equationalpredicate noun sentences carries wi noll CRalways at albel. as sigThatned isthe, the Patient Neutral case case re lationform ,(Neu) and thein thea CR; exa itmp leis onlyabove markedmarks bya nominala CF. predNoteica thatte. in Asthe a sepredicate examples itthe does Nominat not haveive +(Nom) or -., GenitInsteaive d,(G en)the , symandbols Neutral Nom , (Neu)Gen, Neucase, etcforms. areare usednot markedas conv witenienth eit coveherr labels for bundles of localistic 6,case form features (see section 2.4.2�2 and [±setrnsction] 4.1(tra) nsiti. Laterve) andin ofChapter [±er gv] howev(ergativer, e)the, arese covertreated lab elsas un, itlike features that of andcorrespondences marked by + betweenor - in contextual"double negation" features and of "horseshthe verbsoe". notations to indicate Centralfollowing to we the shall system disc ofuss case (1) antheal ysisbasic is lexithe casePatient cla imcase that rel PATation is . the In the CR.fundam ental CR and (2) the traditional role of "object" and the assignment of caseIn the relation lexicase. modeExceptl the for Patient impersonal case verrelbsation, Patient is regarded is required as the infundam everyental case Patientframe. whichIf a verbis nec hasessa onlyrily one realised obligatory in the no Nominativeminal acta ntcase, it form is nece withssa alrill y a non-syntacticimpersonal crit eriaverbs and. notSinc toe casesituational assignme rolntes in, manylexi caseof therefers Fill moreanprimarily case to relations have been reanalysed as being none other than the (3.5Patient-3.1 3)case assrelaignedtion . the ThePatient subject case nom reinalslatio nin eventhe thoughfollowing, fr omexa maples Fil lmorean vantageare all poiexamnt,ples the isir in roles aw aywould a neutr be varialisaedtion. Theor centralPatientisa casetion reofla ditionfferent in the se sitinsuationaltrument, experienroles, correcer, spondtime,ing and to pla thece Fil. lmoreanFillmore notions would probabof agently , assign diffdifferentence casein morprelhoationslogical to theor sesyntactic subjec tprop nomertiesinals even. though there is no (3.5) ma-cal ibad kako a:ngry. furious 15 [ +V] (Nom ) I am a:ngry +PAT (3.6) ma-olah kako itisoan love ls 25 [ +V] (Nom 1 (LCV 1 I love you +PAT +LOC (3.7) ma-talaw kako t ira tamdaw-an afraid-of ls that man [ +V] (Nom ) [ +Det] (Lev 1 lamafra id of+PAT that ma:n +LOC 63

(3.7a) ma-talaw kako itir a tamdaw-an afraid-of ls that man [ +vJ Nom ) [ +DetJ (+PATJ I am afraid of that man (���c) bearsIn example the Accusative3.7 tira tacamdasew-an. form isof an thealysed demonstr as Locativeative beca Locususe even(1) the though Accusative it normformis (s eesuspicious examplely3.7 sia) mi·andlar which .to the is Locat substiveitutable form itirherea withtamda w-anno change which inis the meformani ng,is awh kiilend theof a locativetransitio nounn form suffix between -an onthe tirAccusa taativemdaw-an and sugge Locatstsive that the markings, (2) statistically it is abnormalto have Accusative with ergative ma­ verbsone educated, and (3) in theformant use whoof thiknowss seeming English Accus veryative well Locus and tendsform tois conregularisefined to so theme Amitendencys structure to uses accusativein accordance demo withnstrativ theires English interchangea counterparbly withts. locaHe tivehas ona es, esthepe preciallsenty when analys theyis , areI amfolaccepting lowed by loaccusativecative nounsly-marked with theforms -anli kesuff ix. In tira tamdawan as variants of [Lcv,+LOCJ forms suchas itira tamdawan (for this repreinformantsentable only)as and[Acc not,+LO seCJ,tting to accup ounta separatefor th CF-CRis marginal corre spondphenomenceenon, . For the transi3.1.3.1tion and between derivational [ACC,+ PATJrule andDR-27 [L incv ,Chapter+LOCJ, re6.fer to Figure 3.1 in section Thesit uationalPatient surolesbjects and inhence the differfollowingent examcaseples rela tionscorrespond in a Filto differlmoreanent anal ysis. (3.8) ma-tawa cira laugh 3s [ +vJ Nom+PAT he is laughing( ) (3.9) ta-lomaq cira go-home 3s [ +vJ No+PATm he goes home( ) (3.10) ma-bekac kia wawa run child [ +vJ the child is runn(:��T)ing (3.11) siqnaw kina 1 ab i cold this evening [ ( +vJ Nom+PAT . it is chilly thisl evening) (3.12) no lomaq nomako kina sera family ls this land [ GenJ Nom+PAT my family owns this land ( ) Lit. this land is my fami ly 's 64

(3.13) kamay-en ko sapi bacaq non iam hand-ed means-of-laundering lexcl [ +V] (:��T1 we launder by hand Lit. our means of laundering is by hand Fillmore would probably have assigned an Experiencer case relation to cira "3s" in example 3.8, an Agent case relation to wawa child in example3.1 0, a Time case relation to labi evening in example 3.11, a Locative case relation to sera land in example 3.12, and an Instrument case relation to sapi bacaq means­ for-laundering in" example 3.13, because he had not considered the Patient case reasslaigtinedon cato sebe refundamlationsental in andaccordance indispensable with "ra inw" hissitu fr"ationalamework rol, andes . instead The lexicase way of assigning case relations in accordance with syntactic criteriacase relation is a . logicalThis consequenceclaim not only of thecapture claims a thatgene Patientralisatio isn theabout fundam structuralental resimiferencelarity forof idsentencesentifying such inne asr examand pleouters 3.8(o-3r, .1immediate3, it alsoand provides indirect a , pointsee of Figsubcureateg 2.8orise) case verbs re labytio thnseir. casewith frathmeeses assand ignaccounted case neatly relat forions thewe ircan basicderiv ationalclaim. interrelationships. The whole verbal system revolves around this

3.1.2 Real i sation The Patient case relation can be manifested by the Nominative (Nom) or AmisAccus areative given (Acc belo) casew. forms. Examples of the various realisations of PAT in

3.1.2.1 Nominative Patient coIfnstr [+PATuctio] occursn, or byin anyitself, ergative or wit coh nsnotructin-inherenton, it caseis alrewayslations realised in a verbalin the Nominativfollowinge: case form (Nom) . Examples showing [Nom,+PAT] include the (3.14) si-nan i kia wawa have cat child [ +V] (:��T the child has a cat J (3.15) ma-rarid ko tangila n ira become-infected ear 3s [+V] (:��T his ear is infected J 65

(3.16) taes-en c ira beaten[ +vJ 3sNom (+PAT he Was beaten up) (3.17) misa-labi kia babah i eook-dinner woman [ +vJ Nom+PAT the woman is eooking( dinner (3.18) ma-bekac kia wawa run ehild J [ +VJ Nom+PAT the ehild is runn( ing The following examples show [Nom,+PATJ J co-occurring with other actants: (3.19) ma-bahbah nako kia waco no lakaw drive-aJ.Jay dogNom stGeniek :�rgV (+PAT (+INS I( driveJ the dog aJ.Jay with a stiek J Lit. the dog drives-aJ.Jay by me withJ a stiek (3.20) ma-tawa no taw kia apa laugh-at p'eople foNomol :�rgv �:�T (+PAT people( ) laugh at( the ) fo ol ) Lit. the fo ol laughs-at by the people (3.21) ma-qonqon ni Kolas kako hurry .. urge-on Kalas "Is :�rgv ��T (:�:T Kalas( hurried) me ( ) ) Lit. I hurried by Kalas Itfor must Patient be pointed to appe outar inthat the therenominative is a un caivseersal while tendency Instrume innts erg andative Agents languages aplapearnguage in, showsthe ge nitthisive patter casen (cf.of correspFillmoreondence 1968 :1with4). erAmgativis, e beingverbs a. mixedThe classselectioificn ationof one of realverbsisa. tionIn ofAm ais CR, the ove accusativr the othere su dependsbclass onof verbsthe choose the196ir8:33) subjec, whetsreas according the categ toory Fill ofmore ergative's "Sub verbsject Choicealways Hi chooseerarchy" Pa tient(Fillmore as the ir alwaysgrammatical real issuedbje byct the. ForGenit erivegativ casee verbsform., an Agent which is not the subject is 66

3.1.2.2 Accusative Patient A non-pronominal Patient actant can also take the Accusative (Acc) case form. The Accusative case marker to has an indefinite reading, while the other haveaccus anative Ac cusativDeterminerse case are form all in de thefinitir e.infl ecNotetional that paradig persom,nal andpronouns so cannot do notbe marked as Accusative Patients. Examples with non-pronominal [Acc,+PAT] are: (3.22) mi-qosaw to haw kamo to hemay n i Apoy? save a LPead QM rice rns +AdV+spct +AGT ACC+PAT Gen+LOC :�-ergv +prft 1 have you saved some] rice fo r Apoy? ) ) Lit. have[ youl saved[ some of Apoyr;m's rice ? ( r (3.23) pa-t i reng kam i to cacodadan saka binaolan nia pala build lexcl school fo r inhabitant ViLLa rns Nom+AGT ACC+PAT n LOC+LOC +LOCGen :�-ergv J J ;: J ( J r we bui1ld a schoo( l for the( people of( the vil lage ( (3.24) pa[ -taini cira to baro saka itakoan bring 3s flower fo r ls rns T T n LCV+LOC he:� brought) (:�� some flowers�;� fo r) me;: ) ( ) Whendefinite Patient readin is g.(re alWhenised Patientin theJ Nomi is nativreali( esed case( in theform Accusative, then, it , alwayshoweve hasr, ita is demoto benst interrativpree, tedas asin examindeplesfinit e3.2 un5 lessand it3.2 6is belo otherw: wise specified by a (3.25) pabe l i c ira tia codad itakoan 3sNom bookACC lsLCV -ergvs +AGT +PAT +LOC she gave the( book) to me( ) ( ) (3.26) lmi-:�:pinlar o kam i a ma l kaka tina ka ri reng load lexcl be-with-brothers this cart( rns T int Ace+PAT we:� brothers) :��loadJ this ��cart l J deIn finiteorder tointer traprensl( taatetio nthe for (English the ob sentencesje(cts, weJ incan examsubstiplestute 3.2 a2-3 Demon.24 witstrath aive for the article to of to hemay (some ) rice , to cacodadan a school, and to baro conveys(some) flow deersfiniteness. The closestbut is neutral sUbstitute with would respe bect totia proximi which, tyli. ke English the , In section 3.4.3.5 below, we will see that, in order to express a definite ofsubjectten resorts and a todefinite anti-pas notionalsive, ze"dirro-deectrived object" intra innsit theive sa meverbs sentence with a, AmLOCis case perersolationanl namecorrespondings and pronouns to the do PAT not ofhave their Accusative transitiv formse sourc, theyes. cannotSince occur as 67 direct objects of accusative transitive verbs. When the notional object is a pederivedrsonal intraname nsitor iva epronoun verb. , it can only be expressed as the Locus with a

3.1.3 Rami fications 3.1.3.1 Non-specific Object and Accusative Patient In"dir traectditi obonalject" te. rms,In Amnon-suis, bjsuchect notionalPatients dire are ctof obtenje equatedcts have with two diffthe notionalerent refolalisalowingtio nsexam, namelyples: , as Accusative Patient or Locative Locus. Consider the (3.27) mi-kilimi kako to bot ing lsNom fiACCsh -ergv +AGT. +PAT I am Zoo king( fo r J some/a( fi shJ or, IamZoo king fo r the fi sh (3.28) mi-t:�::kilrim l kako tia bot ing lsNom fiACshC -ergv +AGT +PAT I am Zoo king( fo r J the fi sh( J (3.29) mi-kiliml kako tia tamdaw t:�:lr lsNom the manACC -ergv +AGT +PAT I am Zoo king( fo r ) the man( ) (3.30) mi-:�:kil im2 kako tia tamdaw-an Zook-ft :or lsNom the manLCV? ��rns +PAT +LOC I( am Zoo king( fo r ) the man( ) (3.31) mi-kil)im 2 kako it ia tamdaw-an Zook-for lsNom the manLCV ��rns +PAT +LOC Iam( Zoo) king( fo r ) the man( (3.32) mi-kil im2 kako i c i Panay-an) Zook-for lsNom PanayLCV ��rns +PAT +LOC I( am Zoo) king( fo r ) Panay ( ) 68

(3.33) mi-ki l im2 kako icira-an look-for ls 3s rns � T C I(� am� looking} (: fo�r ) him/her (��� ) Coulddepends we onhave the anassig accusativenment of traCRnsit's. iveIf verb the incase all frame these cons exaistsmples? of a PATIt re andall yan AGT3.2., 3.3and andif thesec tionAGT 5.1is the below) subj ectan accusat, then weive havetra nsitby deivefinit verbion. (cIff. thesec tion notthenional we would ·direct have ob jecan t intris treatedansitive asverb a LOC. realised in the Locative case form, notiOn exaonalmining objec thet isnomi non�pronominalnal actants in and the non-pe examplrsonales, we, itfind is thatrealised when inthe the Accunamesa, tivit woulde case nece form.ss arilyWhen be there alisednotional in obtheject Locative is a pronoun case form or , a sipencerson�l Ami s pronounsthe clea n-andcut pe carsonalses are name exams plesdo not 3.3 ha0 veand Accusative 3.31 where forms the . definBordeite ringnon­ between pronominor the Locativeal, non-pe casersonal form notional. In the obseject borde canrline be markedcases, by whether eithe rthe the deter Accusatmineiver is in the Acc or Lcv case form, the N's invariably have an -an suffix that formally places them in the category of location nouns. As with itia Locativetamdaw-an Locusin exam evenple though3.31, thetia determinetamdaw-an r inbears exam plethe Accusativ3.30 is ane alysedcase form as . Thechange gradual in so meshift distinc fromtive Accus semanativetic to featurLocativees, markespeingcially seems the tofeatures correspond of to a dereprefinitsentedeness by and the sp eciffeatureicity matrixes. Schema intically Figure , 3.1the begradationlow: can be Acc+PAT +PATAcc +LOCLcv ? +LOCLcv Lcv+LOC Lcv+LOC -prsn-prnn -prnn-prsn -prsn-prnn -prsn-prnn +prsn-prnn +prsn+prnn -df-lcnttn +dfnt-lctn +dfnt+lctn +lctn +lctn +lctn I +dfnt +dfnt +dfnt to bot ing (3.27) tia(3. I 28)bot ing tia tamdaw (3.29) tia tamdawan (3.30) itia tamdawan (3.31) ici Panayan (3.32) ic(3.3i raan3) Figure 3.1 Feature Gradation from Accusative PAT to Locative . LOC Thedirec gradualtion of shift greater of measpecificitning takesy, with place sp oneecific featureity forma at llya tim dee,fined in theby the shcomiftbinat fromion Accusat of the ivefeatures Patient [+ prnnto Locative], [+prsn Locus], [+ thenlctn], corresp and [+ondsdfnt to]. a Thegrowth in 69 specificity for the notional object. My use of the term "specific" thus withdiffer res spefromct toits thetraditional speaker's us asagesumption which oppoof theses hear the ernotion's knowledg of "defe ofinit theene ss" Hoppereferernt and. Thom It psoncorre sponds(1980:2 most53, 287)closel. y Beingto the pe larsonalbei "iandndividuate definited" would used bybe more specific than being non-personal and/or indefinite. An Accusative Patiespecifntic, obthenjec, t.denotes a non-specific object while a Locative Locus denotes a We now returnto the crucial question of CR-assignment which will determine our Accusativanalysis eof CFthe, we ve�b assi asgn we thell. CR· Toof thePAT . notionalTo the obnotionalject re aliobsedject in re theali sed in theaccus Locativeative tra CFnsit, weiv eas verbsign . the InCR the of LOC.latter caseIn the, we former have yetcase to, we ass haveign anthe fundamis assigentalned LOCCR, , i.thee. ,'notional PAT, to actorsome actant is assig. nedThus PAT, wh. ileThe the re notionalsultant caseobjec t frameintransindicates itive verb an wi intrath ansitiv LOC casee ve relrb, ationpossi blycorresponding an anti-pa toss ivethe, PATzero-derived of its transitive Source. Thus, mi-kil iml in examples 3.27-3.29 is transitive while mi-specifickil imations2. in exam areples given 3.3 be0-3low.33: is intransitive. Their respective feature mi-kil iml mi-kilim2 +v +v +trns-ergv -trns +[ +AGT+PAT] +[ +LOC+PAT] -(Nom-AGT ) -(Nom-PAT ) -(ACC -PAT ) -(LCV -LOC ) Their derivational relation is recapitulated in section 6.2. This analysis possibly invites the criticism that the use of two different CF's tooand twoext ravagadifferntent. CR'Whys toca n'tdisti wenguish keep the just CR two invariant? verbs (o r Wetwo agreeverb thatclasses) the is marcanking easily of beAc cusativtaken eca Patientre of by and the Locaredundancytive Locus rule iss beloindeedw: redundant, but that RR-la [ +PAT] -+ [Acc] / [+[ +Nom]] RR-lb [ACC] -+ [ +PAT] / [-INS] RR-2a [ +LOC] -+ [Lcv] / [+[ +Nom]] RR-2b [LcvJ -+ [ +LOC] / (-PLC] -TIM , Givendepending the properon whethe envir weronmen havet, a we CR canor pra edCF ictas theinput CR to from the theredundancy CF, or vice rule ve. rsa, AccusatIn Arnis,ive no n-sucasebje formct PATrealises is re eitalisedher onlya PAT inor thean AcINScusativ. Non-sue casebj ectfo rmLOC wh isile the realisalisede LOC, only PLC in , theor TIMLocativ. e case form though the Locative case form may 70 diIfff weer entchose CF to's, adopt we would the haalterve nativto inte roducesolution i�to to our keep sys thetem sam eiether CR afor Locative the two notPatient need edor elsean Accusatwhere. ive ForLocus tqose, ei thatther maycase findbeing the a CF-CRconceptual associ juationmp from which PAT is acceptto LOC , allfrom at AGT once to, wePAT must, and poifromnt outtransiti onceve more to thatintra, nsitin ivlexicae veserb , tooCR 'sha arerd to invariaassignednt inabstr accordanceact situational with syn notiontactic critor roleeria. and When not anwith intr a ansitsuppoivsedle vey rb is 'derivfollowsed naturallyfrom a tra nsitifrom vethe vedefinirb, ortion vi ceof ve thersase, theverb recategorie-assignments. of CR's

3.2 The Agent Case Relation [+AGT J 3.2.1 Characteri stics Fil"typlmoreically (1 animate968:24) instlooksigator upon theof the re ferentaction of ident an Agentified byacta thent veas rb"the. This viewrede finedwas undisputed the Agent byin alexicase lexicas egrammarians grammaras unt "theil non-Starosimmtaedi (1ate978 perce:12) ived practcauseric eof st theems actionpartly of from the the ve rb"realisa. Thtionis departure that "intent fromion previous cannot bele xicasepart of theStarosta defini 1978tion: 12-1of Agent3). inSince a caFilse lmoregrammar"'s ter (cm f."i Plnstigeinesator" 19 76seems:105- to146 implyand intentis alsoion prom, itpted can bybe themi sleaconsiderationding. The thatrede thefini ti"typon icallyof the animaAgentte" case part relatio of n Filsemalmorentic 'scharacte definitristicion is isalso not mireslstricteading.ed to FirstAgent ofactant all, s.anima Forten essinstance as a , Patiencan alsot, beas theanimate entit. y whichSecondly is , viewedthe Agent as mo needving notor beingalways locatedbe animate in space. , Starosta (1978:16-20) has provided convincing arguments to show that mach ine theand Agenttsunam case i in relthe ationfollo: wing examples, though inanimate, should be assigned (3.34) Th is mach ine pol ishes the floor with a rotary brush. [+AGT J [+PAT J (Platt 1971 :2[2)+I NS J (3.35) The tsunami inundated Kap iol ani Park wi th a towering [+AGT J [+PAT J wa ll of water. [+INS J (Starosta 1978:18) Thean animateanalys isAgent of 3.3: 4 and 3.35 is parallel to that of 3.36 below which contains (3.36) Mildred swayed her aud ience with logical argumen ts. [ +AGI'J [ +PAT J (Starosta[ +I 1978NS J :16) Hencestructural, Staros simitala'srities redefinit betweenion of sentences Agent captures with animate general veisarsustions inanimate about diffsubjeerctsent, genecase ralrelaisationstions towhich the aresubj ectsothe rwiseof syntactlost throughically par theallel assig sentencnment esof . 71 Thecontrol pre sentas discussed semantic dein finiPleitionnes of(197 Agent6:83-87) is basedand Sta on rothesta notions (1978:1 of3) . cause In aand chainthe Agent of causation[+AGT], and, the the pe percerceivedived pr terimaryminal cause cause rel relationation is normalas the lyInstr codedument as [+INS]. Most languages allow the grammatical encoding of a .maximum of two linknks may in apthepear chain as , Meaonens AGT. and one INS, although an occasional intermediate In establishing the presence of an actant in the Agent case relation, several poAgentints nev shoulder occurs be kept alone in . mind..It mustFirst always of alco-occurl, in the with lexicase a Patie systemnt, the, an thatfundam eventalery sentenc case ree la(extioncept. forThis so meis theimpe naturalrsonal or consequence meteorolo giofcal the sentenc assumptiones) must contain at .least a Patient. Secondly, besides the Patient, an Agent alwayschain ofim pliescausation an Instr. Thirdlyument, , whichthe Agent is the must imm edactiate on effsomeectivthinge percecause ivined the as disaaffectssedoci byated th efrom action itself. is the ThPatientis hold, s andtrue al eventhough for the re Patientflexives may. notThe beentity phgrammatysicalicalyl lyseparate separate from from the ea Agentch oth, er.the two actants must be conceptually and

3.2.2 Real i sation TheGenitiv Agente (Gcaseen) recasela tionforms can in be Am realisedis. Exa minples eit herare thegiven Nomi belonatw.ive (Nom) or the

3.2.2.1 Nomi native Agent Agenttermino actalogynts, a typitransitcallyive occur verb with is saidtransit to beive inver thebs . "activIn etr voiadicetional" if the AgThentis active-pas is the subjesivect di, andchoto inmy thehas "passbeeniv disputede voice" forif Pha non-agentilippine-ty ispe thelang suuagebjects . Stsucharosta as Tagalog (1974:2 98)(cf. findsCena the1978; dis DeGuzmantinction 1978useful:106) but. re Forst rictsFormosan it onlyLanguag to esthe, Filacculmoresativ'se subclass"Subject ofCho verbsice Hie whichrarchy" choose (cf. theFilirlmore subjec 196ts8 :3according3) . with to the nomiaccusativnativee tra(Nom)nsi tcaseive subcformlass as inof the ver bsfol, lowingthe Agent examp isles normally: re alised in the (3.37) mi-radom kia babah i to nanom tebom woman wa ter weU Nom ) (ACC ) [LCV ) -ergv (+AGT +PAT +LOC the woman draws water from the we U (3_ 38) mi-ala kia wama tia nan i - [:r�rue fa ther oat Nom ) ACC ) -ergv (+AGT (+PAT Fthe� �farther resoues the oat 72

(3.39) mi-ki lim kako to bada l to saka it i a babah ian seek ls bernes fo r the girl, woman ( Nom ACC ( LCV I':m� rnslooking1 (+AGT fo r) berries(+PAT 1fo r the girl;:n) (+LOC) bot ing (3.40) mi-ai c ira itakoan to 3s fish om ls (N+AGT ) (+LOCLCV ) (ACC+PAT ) +sorc-ergv P:::Ihe asked me fo r the fi sh 3.2.2.2 Genitive Agent An Agent which is not subject is always5.1.3.2 real ised in the Genitive (Gen) case andform accusa. As tivewill labenguage shown, inturns sect oution to have twobelo suw,bclasses Amis, being of verbs a mixed that ercangative be called ergative transitivma-e which take genitive agent and nominativ3.4e patient1-3.43) . The verb-en can be either a ergative transitive (as in examples or the3.4 4-3.45)"p ass. ive voice" of an accusative transitive verb (as in examples withfollowing ergativ exame plesverbs show, Patient genitive is alwaysAgent chosenand Patient as the su bjecgrammatt withical thesubjec ma­t. The hisergativ iteme numbertransitivegiven -ve rinbs parenthe. Inclusesded. are sentences from starosta's Amis data, (3.41) ma-radom ko nanom tebom n ira babah i d rom water lL'eII that woman Nom LCV Gen :�:: (+PAT ) (+LOC ) (+AGT) that[+erg womanvl draws the water fr om a we ll Lit. the water draws from a we ll by that woman (3.42) ma-ala kia nan i nia wama rescue the cat the fa ther Nom Gen "\ :� s (+PAT) (+AGTJ the+ergv =falther rescued the cat Lit. the[ cat rescued by the fa ther (3.43) ma-melaw nomako kiso nasan i (Ml71) 2s just-a-while-ago s:� t� LCV rns · ��T) (:�:T) (+TIM) I +ejustrgv saw you Lit. you[ jusl t saw by me 73 The -en verbs are also syntactically ergative. Thus they also· require their Agentsexample tos: appear in the genitive case form in Amis, as in the following (3.44) sawad-en ira ko hal i-qepah-ay a kakawaw abandon aLtopead n drinking-much habit +AdV 3s Nom :� sl (+spctj (��TJ (+PAT J he[+ergv has= given up drinking Lit. the drinking habit has (been) abandoned by him (3.45) me l aw-en wawa k tomay (M126) -Of no child o bear (Gen (Nom +ergv +AGT +PAT ] the:r �chiZd took care ofJ the ,bear Lit. the[ bearr took care of by the chiZd While these verbs allow [+3.2.3.3AGT] in their case frame, they are by definition matra-nsitverbsive and(see the gectio -en nverbs are belotransitivew) . Atve therb classessame time which, si ncerequire both PAT the su(cf.bjec Starts ostarather 19 than74:2 97)AGT. subjects, both classes are by definition ergative

3.2.3 Rami fications Related to the Agent case relation, there are three interesting issues that posdesersessve orfurther. The discfactuss thation . the Firsttwo are, the siremi laris the in issformue has of ofgenitten iveraised agent doub andts Vabout the status of the construction head, whether it should be analysed as a prob(witlemh ofgenit addingive age an nt)Agent or toas athe N case(with fraa pome ssessof derivedor) . Secausativcondlye, vertherebs . is theIf withoutthe source violating verb alre theady l/Sent has anconstraint? Agent, how canThird anotherly, there Agent is bethe introduced issue of traof nsitivdirectity obje. ctsTraditional. In the lymost, transi recenttive conception verbs are ofidenti the fiedlexicase by themod preel, sence traframens itofivity the vecorrerb. spondsWe shallto the explor presence thee seof Agentissues and/orin the Instrfollowingument sein ctiothe nsca. s'e

3.2.3.1 Genitive Agent versus Possessor WhenA genit theyive areagent confu andsed a , pothessess corrector, being identi simificationlar in formof the, areconstruction easily confu headsed is. NPalso with affect a nominaled. Ifhead a form. Ifis a identiform isfied ide asntified a possess as ora ge, nitthenive it agent is part, it of is an Asthen a partsister of co thenstitu caseent fra, methe offor them playsverb whicha role isin the su bclaheadssif ofying the verconstrucbs. tion. Hencesentence, the, bothconfusion syntact canically make anda big semanti differcallyence. in our understanding of the 74 ArnisIn the can follo helpwing determine we will whethershow how thea reGenfere isnc ae potossess wordor order of the and su bjpronounect ors notin . Therise po tosi amtibigon uitof y.a non-suExambpleject 3.4Ag6bent and, whether 3.46c eaorch not have it twois a readingspronoun , becamay usegive bothform non-suin Arnisbjec. t ThAgentsis phenomenon and nomi nalis saposidses to sorsbe aare common marked feature by the of sam eregativ Genie tive andsyntax Reid, shared1978:5 by6) . Arnisand many other Philippine-type languages (cf. Pawley taes-en nia tamdaw wawa (3.46a) beat the man ko child Gen Nom (+AGT} (+PAT} the+erg manv l beat the child ta[:�=ses-en wawa nia tamdaw (3.46b) beat ko child (Nom :+ergv�rns +PAT } the( manl beat the child the man 's child got beaten or, taes-en ko wawa nira (3.46c) b ns ��T/+LOcl he+e beatrgv the child his:�: chlild got(:;�:l beaten (; or, [ Having experimented with various word sequences, we find that ambiquity arises ongenitly whenive NP a . non-g No enitambiquityive non-pronominalwould arise NPif theis immnon-genitiveediately fo NPllo wedimmedia by telya preno nominalceding apo genitivessessor NPmay isoccur a pronoun within, duethe tosame the consti co-occurrencetuent afte rre sta pronounriction . that Exasingmpleular 3. pronoun46b would cir noa forlonger ko wawa be ambtheiguous child if as wein suexbstiampletute 3.4 the6d bethirdlow : person (3.46d) taes-en cira nia tamdaw beat 3s the man the: man beat(:��T} him (��T) : Sentenceinterpretation 3.46d hasof( �theonly ) genitive one inte formrpre tationis ruled. outThe by nominal the co-occurrence possessor restriction which can� be stated by the following redundancy rule: (3.47) [ +prnn] [- [+N] ] NPThis constitu rule saentys . thatThere a pronounfore, a ispo neverssessor fo whichllowed isby codedanother as [GN withinen,+LOC the] in sathisme wouldstudy havenever to occurs be analysed immediate as ly[G enafter,+AGT a ] pronounas in exam. ple A Gen 3.4 form6d. in that position 75 wordAnothe orderr way of to the disa twombi nominalguate sentence constituens likets . 3.4When6b and the 3.4 ge6cnitiv is eto NP rev, ersewhether the or witnoth itthe is nominal a pronoun possess, comeor imm andedia retelyceives after a unique the verbinte, rpretatioit. is no n longeras a non­ confused unasubjembictguous Agent patt. ernExample: 3.46e, as well as 3.46a above, shows the (3.46e) taes-en n ira ko wawa beat 3s chiZd (Gen (Nom :+ergv�rns +AGT J +PAT J he[ beatl the chiZd However, due to the similarity in morphological shape, derivationally related di-en-stsuffiinguisxedh (cf.nouns Starand ost verbsa, Pa wleyin Austrone, and Resiaidn 1982languages) . When are th ofeseten formsdifficult are to posfollosesswedor by or a anGen Ag, entit . is Insom etimsuches cases difficult, the pre to sencetell whetherof a Gen theform Gen is form of nois usea prein detesentrm analyiningsis the of nominalAmis, I orhave verbal relied sta ontus theof prea precedingsence or abformsence. Inof thea Deverbhoodtermine ofr ona constituentthe predicate. word to determine the respective nounhood or As3.4 6afor andthe 3.que46est areion posof whethersessors , orth noten, the prdee-sucisionbjec dependst Gen formson the in grexamammplesatical categorisation of the head of the· predicate constituent. If it is a Verb, then the Gen form is unmista(0kably ci) not a possessor. (i)If it is a concrete Noun, markedGen form by eithatther imm ediaa Neutraltely followsor it oris undoua Locativbtedlye an caseattribute marke r,of thethen the nominal predicate. The statussa- piof-angang a Gen meansform isfo rstill caZZ ingin doubt peopZe when ,the head sectis a iondeverbal 3.3.2.1 Noun or , thesuch Engli as sh gerund like shoot ing and cook ing. citedI would in be inclined to label the Gen formas a possessor [+LOC ] if the derived N has wordsalready like acquired mean in ag, somewhat lan ding, concrete coverin g,and bi ndlexicaliseding and fr meaninosting.g, asOtherwi with Englse, ifish theregularly derivation predicta stillble seems, the CRfresh of withthe Gen the formverbal corresponds meaning promi to thenent su bjectand of theDepending source onve rbthe fromverb which, the itschoice nominal of CR constituent to match the head Gen has form been in derivsuch ed. "embedded" clauses follows Fillmore] 's Subject Choice Hierarchy. If the source verbthe deverbal is transit nounive is, with, at [+leAGTast inin thitsis caseanalys frameis, treated, then the as Gena genit formive follo agentwing andsentence not a. possessor on the derived noun, be it subject or predicate of the

3.2.3.2 Agent in Causative Constructions A derived causative verb calls for an additional act ant in its case frame. TheStevens additional (1973) actantand othe isr somcaseet grammaimes markedrians . withIn a lexi"caucaseser", thcaseis adrelditioationnal by Agentactant (cf.is none Starosta other 1978 than:1 the2-17 perce, 40-4ived2). primary causal relation, that is the 76 The lexicase claim that the "causer" in a causative1974) construction is an Agent is constructionssupported by Formoconformsan languagesgenerally to(cf. the Starosta pattern in Figwhoseure 3.2indir fore ctAustro causationnesian 'ilanguagen the Ges.nit iveIn Arniscase', formthe , Agentand soof doesan er thegative "cau traser"ns ofiti vean veerrbgativ alwayse causat appearsive verb"causer". Theas asi mplgrammatest exicaplal Agenationnt. for this observation is to analyse the Howevederivedr, froma gramm a traaticalnsitiv cone veflictrb wh ariichses req whenuires a morpan Agenthological actant causative, and when ve thrb isis Agenderivt.ed causaHow,tive then ve, canrb reqa uircasees framean additional accommodate "cau thser"is, adactaditiontnal that Agent is alsoNP an witconfhoutlic t violatingby reassig thening One the-per-Sent "original constraint?" 'Agent of theStarosta non-causative resolve s sourcethis verb generalto another pattern case presentedrelation, byfo rmeStevensrly called for Austrone the Experiencesian "indirer. Toct useaction the Starostacausative's" caseconstruction assignment, we incan Fig illureust 3.2rate be thelow : difference between Stevens' and Stevens [causer] [agent] [ object] [other case relation] Starosta [ +AGT] [+EXP] [ +PAT] [other case relation] Figure 3.2 Case Assignment for Indirect Causation Construction (cf. Starosta 1974 :285)

Stverbalarosta deriv's analyationsis and al thelows mappthe ingcapturing of CR's of onto important CF's. geneIn theralisa pretiosentns staboutudy of Arni5.1s. 3.such5), structures have been analysed somewhat differently (see section still simplebut andthe straigaddinghtforward of an Agent. in the derivation of causative verbs is

3.2.3.3 Transitivity While Fillmore does not subclassify verbs in terms of transitives or verbsintrans. (1itiv974)es recog, Praneenises Kul transitivelavanijaya verbs, in heras constitutingstudy of Thai anaccusative explicit lexitransicaltiv e reqcateuiresgory within its two ful idlyenti spefyingcified case matrix featur thees features. A tra [+(nsi[+tiveNM ])verb] and in Thai [-req[uir+NM,ed+OB toJ] haveJ. potentialThe first feforature a nom meansinative that actant a transitive, that is ve, arb grammatin Thaiic al"i s trasubject,nsiti vein veitsrb doesenvi ronmnot ent"allow (i bian d.OBJ:106 su)bj . ectThe, that second is , featurethe co-occurrence means that aof an actastudynt) . in nominative objective case (i.e., Nominative Patient in the present

Kullavanijaya's definition for transitive verbs__ has gone a step beyond Fillmore's specification1968:27) by the case frame: +[ 0(1) (A) ] for an English verb associalike opention (F witillmoreh case relationin featuresthat lexica are sebrought case forintom featuresplay as weandll the in irthe thecharacte definitiorisationn more of prveeciserb classes than . traditionalThis reference definitio to ncase of traformnsit featuresive verbs makes in

�, 77 withoutterms of any the inte prervsenceening of prep an osiobjectiont, a(J nounespersen occurring 1969:1 imm16) ediaandtely even afteChomsr kythe's verb definieattion, with the Strict Subcategorisation feature: [+ NP] for the·English andverbs CF 's ofgr theow , NPand in beadliditioeve n(C homskyto requi 19ring65:9 4)its, presincesence itr. estricts the CR's whichWe should has beenrecall assumed that accordingby all subsequent to Kullava lexicasenijaya's des decriptiofinitionns of, altransl intran­itivity, siverbstiv e haveverbs non-Patient have Patient subjec subjects. t Howeve(if therer, Kuislla ava sunijabjecyat)'s wh deilfinie trationnsitiv is e wouldapplic beab leintra onlynsit toi veaccusa, sincetive thela nglexicauages.se definitiBy her deonfinit requiresion, aller gativeergative verbs verbs to hanotionve Patient of tra nsitsubjecivityts (Sistarosta relevant 197 4to:2 97)ergative. Howeve languagesr, it nowas weappellars as accusthat theative lioneske to(cf. prop Hoppeose r thusand Thompextendsson ove 19r80) erg. ativeThe delangfinituagiones ofas wetrallnsiti as vityaccusat I wouldive one s. In the present study, transitive verbs in}, Arnis are defined in terms of the case areframe de featurfined wites: h [+the[ +PA featurT]] withes [-[I [{+[ NS + INS]] and+[ [-+AGT[+AG]}T]]. ], wh ileWhat intra thisnsi meansti ve verbsis that traintransitnsitiive veverbs verbs require exclu dethe the co-occurrenceir co-occurr enceof [+. INSThis], [+deAGfiniT],tion or bothapplies, wh iltoe bothspecia erl gativecase liandmited accusat to accusative verbsive andlanguage Kullas.va nijaInya Figure's definit 5.2 ofion sectfollowsion 5.1.2as a below, transitive verbs are cross-classified with the feature of ergativity, yieldtwo suingbclasses three ofsubclasses accusative of traergativensitive tra verbsnsitive (I IIverbsa and (IVa)II b,for IVb Arnis, and. Vb) and

This treatment disallows the option__ ality of "inner" cases that Fillmore permits in his case frames, such as +[ 0 (I) (A) ] mentioned above for open. Within the lexicase framework, a change in basic case openframe featureHe openeds entai thels do theor. deriva­ tion into another verbal categoryThe, so doo thatr open edin. belongs diffto a erseparateence in classcase framefrom . openInst inea d of deriving traonnsit accountive and of intra theirnsit ive open1970 :3from87, Cooka deep 1971 str:1ucture6-24) , containingthe relationship an iden betticalween verb the entryse two (chomopf. Fihonousllmore verbs2.2.2.3 is) . accountedThe de finitiofor inn lexigivencase in by th meansis study of Delearivds ationto the Rul recognitiones (see sect ofion homopand intrahonousnsit verbsive veasrbs be, longingcharacte torised different by the verbarray cla ofs sesobli, amonggatory themCR's tra in nsitivthe e excaseplicitly frame andrejec byts su thebjec assut choicemption fromof the among "unit thyese of CR the's. word" Thus which lexi undecase rlies grammarthe treatmentframeworks of such as wewordsll asin generativ, for exame plesemant, Filicslmore. 's and Anderson's case In some cases, then, derivationally related verbs involving apparently similar havesituational quite dirofferlesent but caseimplying frame diffs ander entsynta perspcticec prtivopeesrties of the. Thatsame is,situation based mayon untheli kesyntactic Fillmorea criterian case modaffordedels, may by theass igndi fferenceone combinatio in realn isaof tiocasen, lexicafeaturesse , to Ruleone vewillrb andbe useda diff toer entaccount combination for the tore latethe dnessother of rel theated verb verb classes. A,Deriv involvedation theand sethe ver reasbs. sigThnmisent lic ofense case to relareasstioignns andcase surebjlaecttions cho inice the in thelexi case modelframes of provideverbs. s Fornew detai insightls ofinto syntactic the re laderivationtionship between of Arnis transve rbsitiv, seee andChap intranster 6.itiv e 78

3.3 The Instrument Case Rel ation [+INS ] 3.3.1 Characteristics TheIns trumInstentrument, Mea nscase, andre laMannetion r.here Itis isextended broadened to tocover mean ·the any notions immedi ofate and intercausemediate as sugg escauseted byof anSta actionrosta (1inst978ead and of to ju apstpear the a)most. immediate effective thereIt is pois inteda general out by tendency Starosta in innatural The Fa ceslang ofuages Case for (S taroscertainta case1973 :3re) lathattions or casetogether-like of notions these caseto be rerelaalisedtions byis notthe tosame be casedismi formsssed andas merely that the accid falenlingtal . patternsIn fact, thecros mapps-linginguisticalls of casey thatrelatio itns is onto not caseall tooforms hast exyhi tobit cla suimch thatsimilar they anarealys naturalis, the to natural the point grou ofping being of laInsnguagetrument univ, Meaernssal, . andIn Manner the pr ises entexpl ained as . are laconsequencetion. of the fact that all three notions reflect a single common case Now let us characterise the notions of Instrument, Means, and Manner before we moreshow thanin the one next of thesectionse three how notionsthe Amis exprelanguagessed nevat oneer crtimeowds. theWhat case th isframe means with this esthate notions there underis no dangerthe same of labeviolatingl. the l/Sent constraint even if we put According to Starosta, the pre syntactic notion of Instrument designates the entitreferredy pe rceivto byed the as vetherb , immediatewhereas Agentef fe ctiveis perceived cause of asthe an actionindirect or eventcause in the rechainferred of tocaus asation the . MeansOccasional. In telyrms, anof inter the mediatecause and cause control is involvedrelations and, we it canis orsay a thatMeans an, butAgent not nece vicess arilyversa impl (cf.ies St artheosta conceptual 1978:13-14) presence . of an Instrument In a lexicase analysis, an Instrument may be animate or inanimate, concrete or animateabstractne. ss Theis inva prioriolved inassu themption diff erentiof mostation ve rsofion thes of Agent case fromgrammar the that 18-Instr32ume; Stntarosta has been 1978 found:7,13-14) to be. Thereforesyntactically, Fil unmlmoreotivated's orig inal(cf. Pleinesdefinition 1976 of: haInstrs beenument extended as "the inimmediate lexicaseph toysical cover causeanimate of andan eveabnt"stract (Fill instrmoreum ents1970 :1as16) wethisll chap(Staterosr,ta the 1978 tradi:487)tio. nalBe notionsides, ofas shinstrown uminent the can previous sometimes two be se ctionsassigned of thesynt [+acticPAT ] patteor [+rnAGT of] casethe sentencerelation. insteaThatd isof to [+ saINy,S], a depending notional onins thetrument is notdecis alwaysion of matchedcase as upsig witnmenth theis madeInstr primumentarily case onre thelation basis. Theof syntacultimatetic semacritnticseria. toOur int practiceerfere with here syntact is simpicly co onensideratio that doesns. not allow situational Thethe preaction syntactic impinging notion on theof MeansPAT" is(S tardeostafined MS by c:lSta)rosta . In as the the chain "traj ofec tory of causatioInstrumentn, . the CoMeansnsid erde thesig natesfollowing an inter Englishmedia exate mcauseples,: between Agent and 79

John loaded the hay onto the truck. (3.48a) AGT PAT LaC (3.48b) John loaded the truck with hay. AGTBees were swarminPATg in the gaMNS rden . (3.49a) PAT The garden was swa rming withLaC bees . (3.49b) PAT MNS (3.50) John loaded the truck with hay with a pitch fork. AGTHe opened the dooPATr wit h a keyMNS wit h hisINS le ft hand. (3.5.1) AGT PAT INS MNS In dealing with related sentences such as 3.48a and 3.48b, and 3.49a and 3.49b, EnglishStarosta at propo leastsed, athe separate addition MNS of casethe notionrelatio nof is Means necess toary the inmetatheor additiony. to ForINS cotonsti accounttuent s,for say sentences, key and such his as le exampleft hands, 3.5are 0 notand co- 3.5re1.ferential Since , thewe wouldtwo have violatedsame case there laljSenttion. constraint The subtle had dist we inctiondecided bet toween analyse the immethem diateas having and the the systemintermediate of every cause language of an . action, however, may not be reflected in the case Theperfo prerme syntad" (Sctictarosta notion to apof pearManner a:l) de. signatesIn Engl is"theh, manthisner notion in which can bean eitactionher is wiexpthres assed in by the manner following adverb examialspl ores: by prepositional phrases with the preposition (3.52a) He packed the sui tcase careful ly. (+AdV+mnnr He packed the sui tcase with caJ re . (3.52b) MAN (3.53a) He responded enthus iastical ly. (+AdV ) He responded wit+mnnrh enthus iasm. (3.53b) MAN Agisain expr, notessed all throug langhuages verbal beh deviceave like inst Englisead h.of by Inthe Amis case, the system notion. Inof Manotherner worthe dscase, Man framener is withnever Instr reumalisedent or as Means a nominal in Am constituentis. Its re andalisa hencetion byneve a r classcrowds beloof manw. ner verbs and verbal complements will be presented in section 3.7.1 We have seen that in English the realisations of Instrument, with.Means , and Manner are similar, viz., they can all be marked by the preposition The Fa cesSo isof the Carealse isation of the notion of Concomitant (or Comitative) . In (StarostaMAN 197 3:3-4), Starosta maintains that the possibility of realising COM,atte stedINS , thanand the bylanguages the instr coveumentalred in casehis studyform is - Engactuallylish, muchGerman more, and widely Sora - 80

MAN, woulda natural suggest class. byWhat vi rtuethis of means the iris sithatmil arityCOM, inINS surf, aceand manifes posstationibly MNS as, wefOITll . pointas in semafor nticsyntac intertic pretatpatterniionng.. ThWhisile as susynchrmptiononicall offeyrs ita mayvery -leadgood to vantage an .woptimalith the caseassum systemption withof natural the least classes numb erof ofrela casetional re latinotionsons requimay alsored, provideoperating neware maninsifightested into by languageother sets typology of languag and essub. grouping when alternative patterns

3.3.2 Real isation Likerealised the Agentin the caseNominativ relatione (No, m)the orIns thetrument Genitiveca se (Green)la ticaseon inform Ami. s canIt becan also beExam replesalised are ingiven the beloAccusatw. ive (Acc) case form with a "Means" reading.

3.3.2.1 Nomi native Instrument When the Instrument is the subject of a sentence, co-occurring [+AGT J i·s to bf� recasealised form. in theExamples Genitivare e :(G en) case formand [+PATJ in the Accusative (Acc) (3.54) sa-pa-ahcid ko c i 1 aq to tood used-for-making-something-salty salt things [ +vJ (ACC salt is used fo r making things salty(:�: S ) +PAT ) or, salt is used fo r salting things (3.55) mami-asik c ira to adawang no cacodadan assigned-to-sweep 3s front school [ +vJ he is assigned to sweep(:�:s1 the fro(���T)nt of the school (3.56) sa-pi-angang nomako kina piqio to tamdaw used-far-calling this whistle people [ +vJ lsGen Nom ACC use this whistle (to+AGT caU ) people( +INS 1 (+PAT I ) 3.54-3.56 Althoughbe somewh theat mi verbssleading in exam to plelabes l them "Insthaverument instrFoc umus"ental in thesubj Phectsilippine, it would sense because they do not form a parasa-pdiga-m, and sa-pare i-not marked by any consistent IF affix. The verb forms with andsa- prefixesmam arei- secondary forms basedma- onm thei- reanalmaN-pysisi-, of deverbal nouns, while possibly results choicefrom betweenor analysing alsomami- secondaryas ma-mi deve- orlo aspm maNents.-pi beloWe willw. Morphodiscusslogi ourca ll� speaa cokingnsist, entthese IF formsform. areThese too irregforms ularwould and ha vesporadic to be toindividually justify the marked positing in Jf hithestori lexiccalon accidentas verbs orthat as thetake spearheadNominative of Insan trumenincipientt, eit syntacher astic the change result. of 81 To avoid setting up a special class of Nominative Instrument verbs, one might propo3.54-3se.5 6as an alternative analysis that the PAT's marked by to in examples could be some other case relatio'n, sayma, mi-REF, instead of3.5 PAT5. , wouldespecially then witmakeh theit poerssigativeble -lforook theing su verbbjects wit hto be markedin exam asple Patient of This ordinary e�gative verbal constructions. (1) mami- I have decided against thema- REFanal ysis because these ma-mi-verbs are not memmaN-berspi- of the ergative (2) class (cf. the discussion 'of a versus analysis below) ; REF case relation saiska oti,herwise subsumed under the PLCconsist caseing rela oftio then prepand ositiois alwaysn saka manif andes tedthe byloca tive determinera composite i, marandker not by the accusative marker to; (3) the to forms in these examplesare all asinde desfinitcribede, which in sec istion in accord3.1.3.2; wi andth. the (4) characteREF is anristics optional of Accusativ outer CRe butPatient the actants introduced bysa-p to i-,are sa-paobligatory ma miand- must be included in the case frame . for subcategorising and verbs. The alternative analysis ofsubcate thesegori actantssation as. REFThere's wouldfore, makethe actants them periphemark edral by andto henceare ass notigned rel evant to Accusative Patientmami- case featuresma-m as i-indicatedmaN-p ini-, the examples. As to whether(1) comes from or my considerationsmi-, are as mi-folloasikws: 'to sweepsince (something) there are accusative transitive verbs marked by such as mami in Amis, they provide the derivationalmami-as ik sources for thederived fromverbs mi- whichasik byare addingalso accusa the ertivegativ. e Ama- verbpre likefix. The probwouldlem here then is be shouldtwofold an. erFirstgative, neitpreherfix appethe arsource in th notis context?the derived Informs other are words ergative, the , derivedwhy byverb ma -is prenotfix a ationon-intentionaln. Secondly er, gativeif we analyse, as it mi-shouldas ik be as ifa itfinite �ad beeninfle derivedcted form, it cannot be rederived because derivation almost always appliesmi- to stems, not inflected formsm-p. i- And our present conspi-ideration is to treat the verbs mi-as morpasikho logicallypi- asik, verbs, with basic stems. The stem formof is then and it is the pi-latter form that should parmi-ticipate in further derivation. Since it is the form*ma corresponding-pi-as ik, toma- miverbs-as ik. that should(2) appear in derivedma mi-verbsasik, we anticipate maN-pi-asik not N-p If we suggest that resultsm, from instead, with the sequence regularly realised'as we may havepi- side-stepped the issue of erinflgativityected mi- and,form at theto be same furthertim e, derivalloweded. theHoweve r,form we instearun intod of other the probN-pi lemsas we withll asth m-pis analysi- whenis we. makeFirst the, we cl areaim thatassuming pi- nasalis the assibasicmi lafortionm, withbut nasal assimilation1.4) . is not an active morphophonemic process in this language su(cf.pported section by synchronicIn phonoother logywords. , Sethecondly nasal, asscomparingimilati onAmi anals withysis other is not AustroneandIlo kasiano,n welanguage find thats, es pethecially propo languagesed maN-s formsin the do Phnotil ippicorrenesspond such inas meanTagaloging, specifically in their instrumental usage, with simi5.1,lar forms found in other languagesmi. - Thirpi-dly, as will be shown in section the correspondence between and forms mi-can be accounted for by derivation instead of by ainflect ma-mi-ionana, thuslysis al polowingssible . forms to be rederived; that is to say, making 82

ma-mi- maN-pi- Neither the nor the solution seem to be totally satismafactorymi- . Sosingle for derithe vationaltime being unit we willas in leexampleave the 3.5is5sue above. open and simply cite as a 3.54, sa-pa-ahcid, The verb in example pa-ahcid to make (something)is derived safromlty a . derived causative traahcnsidit ive-verb stem The basic form to be salty is a stative intransitive verb that takes6. a [+PAT] su3.5bject4, . witSuchhout deriv theational problemat -relaictions presence are accountedof a non-sub for jeinct Cha Agentpter as in Examexampplele 3.56, lends itssa-elf readily to supporting a verbal instead of a nominal interpretati�n ofto thetood thingsforms, in these examples, since the position of the Patient actant, to the right of the subject would make it impossisa-pblea-ahc to asidsi gn a binomnarinaly equationalised form insteaconstid.tuent structure if we were to analyse as 3.56, sa-pi-angang, The verb in example pi-angang belongs to a class of verbs derived from transitive verbs like to call (someone) with AGT, PAT, and INS asin ththeeir su bjcaecse t framof esthe. sentenceIn this isexam a pleviolation, the appe of Filarancelmore of's INS Subj insteaect Choiced of AGT propHierarchyosed to for avo accusaid thtivis econsequence languages., wouldThe bealter tonativ treate analsa-pysisi -angang, which he remig asht abe derivednomako reanalysed nominal pr edicaas a pote sseof ssiva non-vere pronounbal , marequationalked as [G enconstructio,+LOC]. n,Th witis musth be ruled out, however, in view of (1) the absence(2) of a Determiner such as 0 before the allegtoed ta mdanominalw , ised predicateto toformod , and the3.5 po4,si tion of the Patient actantto assig n the binarywhich consti, liketuent structurein example which is reqwoulduired makeif the it sentenceimpossible is to be analysed as equational. Though a homophonoussa-p nomi -anganginalised form does exist(cf. sect for ioneach 3.3. mem3.2)ber , ofex theamp leve rb3.5 6class is formato whllicyh a verbal constructionbelongs . It is to be emphasised here that both the mam i- and sa-pi- forms cited in these examples can be used as Nouns as wellma mi-as Verbssa-p, thei- two categories being derivationallysyntactically andre lasematedntic. Moreoveally in r,many waysand. Their verbsdistributio are sin,mi howevlar er, is complementary with respect to thesa-p featuresi- of animateness and humanness marked oninanimate the [+INS) ins] actatrumentnt. subjeWhilcte , a a mami- verb takesinvariably a non-hu takesman an(mostly animate (mostly human)cira onemami-. as I ikha ve assigned 3.5an 5,INS case relation to the animate becasubjuectse , firstof of all, the instructural example parallelsrather between than anma AGTmi- andcase sa-prelai­tion constructions shouldma mi-not be igsa-pnoredi- lest we would lose some generalisations. Sethatcond thely , Instrboth umentthe is habituallyand usedver bsfor, theas citstaedted in purp theosese exa. mplThirdlyes, indica, AGT te withneed noina nimlongeraten beess associa(cf. setedction exclu 3.2sive abovlye) wit. h animateness and INS exclusively The same complementarity exists with mami- and sa-pi- verbs in other syntactic erclassgativese. transitiveIn the fol malowingmi- and exam sa-pplesi- whichverbs , showthe slidenticalight overla casep inframes the for selectional restriwacoctions dog for these verbs makes it possible for a non-human but folanimatelowing noUn exam lipleke s will showto. co-occur with both types of verbs as the 83

sa-pi -adop kina koang no lomaq ako (3.57a) used-for-hunting this famil [ +V] m Gen Is +PAT-anmt (+AGTj my fa mi ly uses this gun fo-huronr hu nting this gun is used fo r hunting by my fami ly Lit. sa-pi-adop (3.57b) kina FWaCO 1 no lomaq ako used-for-hunting this fam il [ +V] m Is +a+PATnmt (��Tj my fami ly uses this dog fo-huron�r hunting Lit. this dog is used fo r hunting by my fami ly mami-adop kina ftamdaw1 no lomaq ako (3.58a) assigned-to-hunt this man famil [ +V] om Gen Is +a+PATnmt (+AGTj family assigns this manr+huron to hunt my this man is assigned to hunt by my family Lit. mam i -adop kina waco no lomaq ako (3.58b) assigned-to-hunt this fami l¥ [ +V] m Gen Is +anmt+PAT (+AGT J my family assigns this dog-huron� to hunt Lit. this dog is assigned to hunt by my family The selectional re�lirement of these verbsf appa1 rently carries over from their derivational sourcessa as-pi deverbal-V nouns in equational sentences (see sectionpi -V) and6.2.7 a ) ;ma mithat-v nounis, a is human noun(one whois lexicapi-v'llys) . non-humanWhen a non(the-human thing sa-p usedi-V tonoun is the predicate of anmam equationali-V sentence, it is equated with a non-human Whensubj ectthe, seand predi a humancate nominalsnoun are isreanalysed similarly as equated verbs with(cf. Stara humanost a,su bjecPawleyt. , deriand vationReid 198. 2), these selectional restrictions on subjects carry over in

3.3.2.2 Genitive Instrument When [+INS] is not the subject of a sentence, it is usually realised in the Starostagenitive 's(G en)data case, his formitem. numExamberples again be indilow cainclutedde in sentenceparenth esesM147: from 84

(3.59) ma-asek no bet iIi c ira strike lightning 3s :�rgv :�:T he( was ) struck by lightning( ) Lit. he struck by li(��s)ghtning (3.60) ma-nanoang ko k i 1 ang no ba 1 i (M147) move3 wave treeNom windGen :�rgv +PAT . +INS the( tree) is moving( in ) the wind( ) Lit. the tree is moving by the wind (3.61) ma-bahbah kia waco no lakaw nia tamdaw drive-away the dog stick man :�rgv :�:T the( man) drove the( dog) away with a stick or, the man 's stick drove the dog away Lit. the dog drove away by the man(��s) with a stick(��) Bothcrite [+rionINS of] actants "independent in exam action"ples 3.5 (cf.9 and Plei 3.6ne0 s wo197uld6,:9 by8) , Pleibe ntreatedes' de fiasning active "Forces" and hence Agents. Our treatment of 1 ightn ing, wind, earthquake , criterialra in, tsuna "tmihe, etimmc.edia incy th ofis contextthe effectiv differse cause from rather his in thanthat itwes indepregardendence as " exam(Staplesrosta are 1978 Instr:17)um. ents Thus, sibothnce eabecht il of i li themghtning acts and imm edibal atei windly on in the th esePatient in bringinggrammatically about Insthetrum ·effecentst ofinst theead action of Agents. Theis bothanalysis compat ofib Forcele wits has Fil lmore's lexitreatcasement interof naturalpretation for cesof Ag (cf.ent Filas beinglmore viewed1971:4 4)as andremote consist froment an with action the and howeveactingr, indirectly only sema nticthrough reaso somens . interWithoutmediary syntactic (see sect reasonsion 3. 2)to. supportThese areit, , rethela argtiveum toent thehere sit woulduation be, itsomewhat is hard weak to drawand ar abitr cleaaryr li. ne Sibencetween imm edAgentiacy andis Instrfoundum toent be when irrel onlyevant one to actant such conis siprdeesentratio, nses. peciallyFortunate whenly animatene, in sentencss esis like exenaamblpleing 3.6us 1 toboth cl aiman theimme diexistate encande aof non- a Geimmediatenitive Inscause trumen aret. present, thus Sentence 3.61 is ambiguous. Due to the rather free ordering of nominal constituents in Amis verbal constructions, nia tamdaw can be interpreted either as a Genitive Agent or as the possessor of lakaw. Sentence 3.61a is the unverbamb andiguous marked coun aste rpartGenitiv of e 3.Ag61ent, wit. h nia tamdaw coming immediatelyaf ter the (3.61a) ma-bahbah nia tamdaw kia waco no lakaw drive-away man dog stick :�rgv +AGTGen :��T the( manJ dro ve the (dog away with( a stickJ It has been pointed out by Pawley andJ Reid (1978:56) that(��s PhilJ ippine-type ordelanguagringes ofare sentence prone tocon thstitis kiuentsnd of, the ambiguity head-attbeca ributeuse orderof their in NPrather's, andfree the 85 avoidsharing such of theambig geuitynitive, most form Phil by ipbothpine a-t noypen-su labjecnguagest Agent resort and toa possea favouredssor. wordTo theorder verb which, thus, as mak in ingexample it less 3.6 1a,likely places to be the inte Genitiverpreted Age asnt a posimmedisessateor.ly after Howevanotherer, ki placingnd of pro a blem,genitive one formwhich imm suggediaestelyts the after rein theterpre verbtation may givofe the"ri severb to consformtructions as a noun that. Inresemble sectio n equational3.3.2.1 above constructio we have nsalr. eaThedy enc simiounteredlarity betverbalween , sa-pi- sa-pi- d sa(1982y, ) as theverbs result and of the ren0int4llSerpretation is explaine of bynomin Stalisedarosta, prePawldieycates, and of Re id Phclefiltippi sentencesne-type aslanguag verbses in. the course of the evolution of focus systems in Although "focused" verbs were derived from Proto-Austronesian nominals with variousas well , deresultingrivational in affverbixe sforms in suchwhich a wayare , homophthe nominalonous wit sourcesh and ofareten retained notindisti arisengui inshable Amis , fromhowev theerir, si sourcence in noumy nsanalys. Thisis nom partinalicula predr ambicateiguits arey does neceare neverssarily preceded preceded by by dete Neutralrmine rsor. Locative determiners while verbal predicates

3.3.2.3 Accusative Instrument AmTheis pre. sySintacticnce what notion can be of characte Means isrised real asised Means in thenev Accusativer co-occurse case with form a in wenotional take adInstrvantagumee ntof asth nomiis comnalplem constentarityituents and together extend withinour INS thecase sa mere lasenttionenc toe, coverInstrum Meaentns as. inAs thea reexamsultples of belothis�. reanal ysis, Means is marked as Accusative (3.62) tomes-en non iam a [mi-pinaro] kina kari reng to rarami make-full lexcl fi ll this cart straw +V] " Nom [ T} �int +PAT 1 we fi lled the( �cart with(� straw} ( (���s1 (3.63) tomes-en no�mako [ ko ni-pinaro t ina sinatowik] [to sima 1 ]] make-full lexcl fU lin [ this bottle oil [ +V ] +AGTGen Nom+PAT ACC+PAT ACC+INS ( } [+past ( ( } I fi lled the bottle with oilr Examples 3.62 and 3.63 are different in structure, but bothJ of them require an AccusativThe seconde occurrenceInstrument ofto PATindi incate exam theple inter 3.6mediate3 is the causeresult of of the deri fillivationng act, with. the deverbal noun nipinaro fi lling carrying with it a PAT and an INS from the verbalfilling source is com. pletedThe ininitial the past n- pre. fix in nipinaro signals that the act of The following example shows a related structure in which the notion of Means centis notral exprisatiessedon, i.e. as Ac, cusativreanalyse isIns oftrumen Meanst, aspos a siblgrammy aticalas a re Patientsult of . 86

(3.64) tomes-en non iam [ko ni-pinaro [to ra rami ] [i tina ke lakelaq]] make-full lexcl f in straw this ox-cart [+V] Gen ��� ACC LCV (+AGT ] [+p+PATast (+PAT ] (+LOC ] we Zoad the straw onto this r ox-cart Thethan ,derivation a reanalysis proce of sscase of centrrelatioalinssation. The, asconst seenituent in th isassig examnedple the, doePATs morecase re"transportedlation has becoobject"me the, the target entity of moved an action through, being space aff, ecratherted by than the siactionmply as a partiChaptercip 6.ating as an indirect cause of it. We shall return to this point in Note also that in these ex�ples, the head of the construction is a resultative verb tomes-en make-fulZ indicating the result of filling. Our accomanalysisplishm alsoent holds· is expr trueess edfor, assentences in the folin whichlowing no example result: or extent of (3.65) pinaro- i to bingkes kina koakoq nom i so fi ll tobaccoACC this p'ipeNom 2s prt +INS] +PAT put(:� some] tobacco( in your pipe!( ] undeIn thisrstood sentence because the the extent verb ofis impefillirativeng is . not Otherexpres examsedples. Theof AccAgentusa tivsubjee ct is aboveInstrum -ents3.66 givenand 3.6be7low with can Aceasilcusativy bee accountedInstrument forlike by examthe plesanal ysis3.62 preandsented 3.65, exaandm ples3.68 and3.63 3.6and9 with3.64 -a derivedand they noun are andinclu henceded onlytwo asapparent a treat PA T'sfor data-pas in hiles. (3.66) tomes-en nomako kina saqtian to coco no ra rapa make-fuZZ lexcl this cup milk cow [+V] Gen Nom (+AGT ) (+PAT ) ( S ) I fi ZZed this cup with miZk ��� (3.67) a [tomes-en non iam kina ko iti to ra raq ] make-fuZZ lexcl this sackNom beansACC ( utr) int) ( T) +PAT ) +INS ) we:� are goingr:� to fi ZZ ��this� sack with( beans ( (3.68) tomes-en nomako. a [mi-pinaro] to s imal kina sin atowik make-full lexcl fi ll, t-in oil this bottZe [ +V ] +V ACC Nom T (-fint (+INS (+PAT we fi lled this(�� bottleJ with oirl J J (3.69) hacoa ko ni-pinaro nom i so to nanom? how-much fi Llin 2s water [ +V ] Nom Gen ACC [+past+PAT ! (+AGT ) (+INS) how much water did you put in? 87

) hacoa how much� to what extent. In example 3.69, the main verb3.q is8, This sentenceIncrement, whichas we llwill as beexam treatedple in expresectionsses 3.7.2. the notion For theof "extimetent" ·bei ng,or these examcan beple res seralisveed to. demonstrate the different ways in which the notion of Means

3.3.3 Ramifi cations thatIn Am theis, CRlang'suage marked can on employ nominal to expractantsess aresuch on notionsly one kiasnd "i nofstrument" syntactic and device expre"placess".io n Thusin verbal case-l constructionsike notions suchor comasplex "man ner"verb informs Ami s(s eefind sect theirion 3.7.1) CRrather's in thanthe aslang caseuage-mars, kedthere NP are's. ofEventen alterfor nativethose notions ways of thatexpre hassveion matching using verbalsame. devices though the meaning or perspective will not always be exactly the In the constructions described in sections 3.3.3.1 and 3.3.3.2 below, the markednotion onof anyinstr ofume thent nomiis beingnal actants expressed. , but no explicit InstrumentCR is

3.3.3.1 Deverbal Nouns with Impl ied Instrument

One way to express the notion of instrument is througsa- , h nominalisation of action ve"therbs means. When of perfan actionorming vethe rb stactionem is indicatedprefixed by the sourcea noun ver isb" created. Th isto mean nominalderived prediN cancatebe usedwhich as re sufersbject to ofthe an concreteequational ob jesentencect that , is"e quateemployedd" to as a an beins interpretedtrument of the as eitactionher deimpliedfinite byor theinde vefiniterb stem. . Here The arenomi anal few pr examedicapleste : can (3.70) kamay noniam ko sa-pinaro tina karireng-an o hand lexcl means-of-loading this cart / ? [Neu] �. m I� � rACC LCV wethe loadmeans this of cartloading with(: thisT) our cart hands is our handsl+PAT ) (+LOC) Lit. nanom no tebom kd sa-pi bacaq non iam (3.71) a water well Lmeans-of-laundering J Nom '" lexcl [Neu] , we launder with well (water+PATJ. )", . (�:�T) our meansof laundering is well water Or T I Lit. 88 (3.72) . o needZerimom anda'to threadkel iw ko measa-pns-ofi-tab-mendingid non iam [Neu] lexclGen : �T +AGT to bodoy ni ina ako � J J cZothes mother Is ( ( � we mend�� T my mother's cZothes with needZe and thread Lit. our (meansJ of mending my mother's cZothes is needZe and thread nominalSubjects pr ofed equationalicates in thesentseences examples are markedare marked as Nominative by the Neutral Patien (Neu)ts. caseThe fonn. Prtheedisecate constructios are notns mar. ked for case relations, so no [+INS] actant appears in Case3.70-3 fr.7ame2 s are carried over in the process of nominalisation. In examples all nominal constituents that follow the deverbalsa- sa-noun are to be V.analysed as subordinate sister act ants governed by the N derived from the case Theforms case. relationsIn these nominaof thelisesed co-occurring constructio nsactants, the areGenitive realised case inform obl mark.:;ique theFil lmoreCR in's the Su bjcaseect Choiceframe ofHie therarchy corresponding. verb which is highest in It turns out3.7 0 that th3.7is2, anal ysis allows for the tinpresencea kari renganof two Patients in3.7 0 exais mtreatedples as [+andPAT ] withif respect the nominal to sa-p actanti -nar o, while sa-pi -nainro exitsampleelf is [+alsoPAT marked] actants [+ PATare] asnot thesis tersubjes ctof ofthe thesa meequational lexical constructionsentence. headSince, ththeeir occurrence in the same tinsentencea kari rengandoes not violate the l/Sent constraint. Of course, if we analyse as3.7 Locative0 Locus instead, the issue d02s pointnot ex: ist. A tree diagram of example is given below to illustrate this (3.70a) S NP Det NP .hkamayand � o [Neu] non iam means-sa-pofi -l-naroadingo lexcl ko Nom+PAT cartkari reng-an tin/1a we load this cart with our hands( ) this ACC+PAT Lit. the means of Zoading this cart is our hands ( J 89 In the next section we will show that a derived sa- nouncan also f�nction as thederived subjec byt addingof a derived the suffix verb -enwith to imthepli Ned donewhose instrumenby re handferentt. This isthebe class handinst ed)ofrument verbs of is thekamay action [+N] handconcerne (cf.d, se asction in kam3.3.3ay-en.2 .1[+)V] . (lit. from A possible alternative analysis of noniam in examples3.7 1 and 3.72 is to treat Howeit asv er,a truethis po anssessor,alysis isi.e. probab, a Genitively more apLocuspropriate, instea tod deverbalof a Genitive nouns Agesuchnt as. duck-cfurtheral deler,rive mixd er,into and the ble classnder, of to concrete cite En nouglishns . examples, which. have been

3.3.3.2 Derived Verbs with Impl ied Instruments Therereferents are twoare sitsubclassesuational of ins verbstruments in Arnis of thethat actions are de rivindiedcated from by nouns these whose verbs . VerbsIn other belo wordsnginq, tothe thinstese rumentsubcla sse�for thedo notact ionallo isw [+impliedINS] in in ththeeir ve caserb sternframe. .

3.3.3.2.1 Denominal N-en Verbs One subclass of Arnisderived verbs with implied Instruments is characterised by donethe "paswithsiv hande" derivational suffix -en. 3.2.3.1,In addition to the example kamay-en from my data: already cited in section we can list the following [w] [+N] lamal-en donewith be fi fired)re lamal fire done(lit . with knife knife hawan-en (lit. be knifed) hawan banoh-en done bewith fefe athered)ather banoh feather (lit. pole, stick lakaw-en donewith be po poleled) lakaw (lit. wooden basin paqtang-en done bewith basined)basin paqtang don(lite . with soap soap sabon-en (lit. be soaped) sabon (Taiwanese loan) Thesecase framederived. AsN-en mem verbsbers withof the im plideriveded ins -entrum verbents classdo not (s peree mitsection [+INS 6.2.3] in ) the, ir thconcreteese verbs, i. e.,requir refee Patientrring to su concretebjects. phTheysical Patient things su bjeaffectedct can by be theei thactioner , oras suabstrabjectct. inc ludeThe somefollowing sentences examples from ofStaros derivedta's -en da taverbs. with concrete nouns 90 (3.73)

Lit. (3.74) filarema l -en ni Panay a [pcaua-kse-to-akomokoqboil] (M43.1) [ +V] Panayboi led the:� waterT with�� intfire Lit. the water was fi( redJ by Panay( toJ make it boil (3.75) knifhawane -en a [msli a-pughteracok ] no olmatoasayd man ko pigbaboy (M57.1) [ +V] the old mansl a��ughteredint the pig:� withT a knif:��e T Lit. the pig was knif( ed inJ slaughter( ingJ by the old( manJ Note that the Agent actant and the verbal complement introduced by a can freely exchais identicalnge the irin pos meaningitions to. examThusple, example 3.74. 3.74a, given in a tree diagram below, (3.7(M434a).2) S v PP P/� S VI I lama l -en a pa-kakomokoq i Panay ko nanom f cause-to-boil n Panay water [: s +AGTGen Nom+PAT Panay+e��rgv boiledl the water with fire( J ( J Normaconstructiolly thens Patient. The NPsubj'sec int occursa verbal in constructionsentence-final can po switchsition pos in itisuonsch rather Gefrnieelytive without NP fol lorewsSUlting the Nominative in change NPof meaandning an am, biexceptguity, ofaris coures. se, Mywhen in forthema nts, howevethat ther, seemedcontext unconcernedwould prevent about any thmisunderstandis ambiguitying., tak ing for granted perhaps Desectrivedion N-en3.3.3.1 verbs wh icalsoh act co-o as ccurthe irwit Pah tientthe derived subjec tssa-. nounsWe may pre resentedcall thatin a sa- noun refers to the means of performing the action indicated by the source verbsubject.. The Thesentence following thus examconsistsple shows of a thedenominal combination V predicateof a series and a ofdeverbal such N denominalcompound sentenceV's as construction: heads and a deverbal subject nominal in a 91 (3.76) dopaqtangne-with--enbasin, dokamne-with-ay-en , hand donesabon-wi-en th-so ap. [ +v J [ +vJ [ +vJ ko measa-pins-ofbacoq-laund ering noniam lexclGen we :��ourT laundry using a wood+AGTen basin� hands� and soap Lit. ourdo means( ) of laundering is (by a ) wooden basin� hands� and soap Threesentence things 3.76. ar� worthFi·rst, men sentencetioning 3.7 in 6 conbearsjunctio parntial with. si milconstructionsarity to con listkeructions Secondlylike exam, plesthough 3. 70-3sentence.72 due 3.7 to6 hasthe useas itsof headsa deverbal verb formssa- noun with as thitse -ensubjec suffixt. , comit plemis unlentsike. examThpleirdlys 3., it73- should3.75 in benot pointed requiring out thatco-occ whenurring a derived verbal -en V is additioncoupled with to ita s sa-indeN,finit the e imgenepliedric insmeanitrumeng.nt has a habitual reading in Noteformer the is .ap anparent equational similarit senty encebetween whil e3.7 the7 a andlatt er,3. 77 b.a verbal In myconstruction analysis, the. kamay ko sa-pi bacoq non iam (3.77a) 0 hand implement-for-laundering lexcl [NeuJ Nom Gen we launder by hand+PAT +AGT (3.77b) be-hkamay-enande d ko imsa-p( plemi )bac ent-foq or-laundering nolexcl( niam) [ +vJ Nom Gen z"elaunder by hand+PAT +AGT Thoughexamples kamay-en 3.73-3 .7in5 on3. 77accountb is probab (of th)ly eir di difffferenterence from in thocasese( fr-en)ame verbs, the inana lysis ofrequired it as ina verb this isstu judystifi thated all by N'sthe exceptfollowing pronouns conside areratio precededns: (1 ) bywh ail eDe t,it is kamay-en here occurs alone in sentence initial position and (2) while kamay , like any other N, can be used as subject of a sentence, kamay-en never appears as such in my data. Example 3.78 below shows sa-pibacoq further derived as a verb with kamay serving as its subject. (3.78) usa-pi bacoqor-laund ering ko handkamay non iam Nom lexclGen ::� +PAT +AGT 1 we[:�+e launderrgvl by hand It is suggested by Starosta, Pawley, and (Reid) (1(982), in accordance with the nom"frominal *fisryed-an con PANstr uctionsto i-fry came" hypothe firstsis and that that in si Proto-Amilar usvetronerbal consianstr theuctions and languagfocus systes. em areIf thehist "*foricallyry-an PAN more" hypotrecenthesis devel is opmecorrentsct in, thenPhili Amisppine is-t probablyype equationalcaught in thesen midstencetrs eaonm withthe oneproduct hand iveand derivations further derived yielding V's onderived the oth N'ser for. 92

The derivation of Instrument-subject sa- verbs such as sa-pi -angang use-for­ calling sa-pi-adop use-far-hunting and 3.57b)(f romis examaccountedple 3.5 for6) inand section 6.2.7. (from examples 3.57a

3.3.3.2.2 Denominal si-N Verbs Another class of Amis derived verbs with implied notional instrument is a subset of a class of Amisverbs characterised by the derivational prefix si-. Whareil estativ "passe iveintra" N-ennsi tverbsive ve arerbs . ergativeAs intra transinsittiveive verbsaction they ve rbsdo , notsi- Nco-occ verbsur with Agent or Instrument; the latter being redundantly excluded because si-N verbs do not permit co-occurring [ +INS J . Members of the derived si-N verb class with implied Instrument include: [ +vJ [+N J si-bakic with-a-bucket bakic bucket (English loan?) si-tokar with-a-ladder toka r ladder si-solo with-a-pestZe solo pestle s i - raq i c with-a-rope raq ic rope si-polong with-an-axe pol ong axe The gloss of these si-N verbs could have been have-a-N to generalise to other si-N verbs such as si-kawa s have-a-ghost� haunted, si-pida have-money� rich, si-ngangan have-a-name� famous. si-N withand implied instruments behave more liHowevke maer,nner syntact verbs ically(see secti the on 3.7.1verbs) in thatin exa theymples can 3.7 function9 and 3.8 as0 andthe mainare thereverb' forewith glossedobligatory as with-a verbal-N complem. ent, as (3.79) with-a-busi-bakic cket kia womanbabah i a [mi-draw-fradromom to waternanom wetebollm J [+vJ ���T ��int ���T the woman used a bucket( ) to draw ( water) from ( the ) well (���c) (3.80) didna with-a-[si -tokalar dder thekis fawamather a [mi-rescue�ala get thet i a catnan i J J �ast ��int ���T ��int ���T the(: fa)ther ( used) a ladder to ( resuce) the( cat ) ( ) The derived verb with implied instrument is intransitive, and always takes a Patient subject. The verbal complement is introduced by the preposition a whichand has is anco-re impliedferent suialbjec witht, usual the Nominly theat Agentive Patient of an ofembedded the main tra verbnsitive. verbThe , desynsctacticribed behaviourin section of 3.7.1. si-N verbs is thus similar to that of the manner verbs Unlike manner verbs, however, si-N verbs can also be used as verbal complements withembedded transi verbtiv ein main examp verlebs 3.7. 9, Rewever getsing the the fo orderllowing of we thell- mainformed verb Ami ands sentenc the e: 93 (3.79a) drawmi-r-fadromom kia womanbabah i to waternanom wetebllom a with-[ si-baa-buckkic] et �rns Nom+AGT ACC+PAT LCV+LOC [+V] the(: woman) draws( water) from( the) weZZ( with) a bucket NominaThe unexpretive Agssedent su ofbj ecthet mainof the verb embedded. "Subverb ject is ch oice"co-re ferentfor embeddedial with verbsthe is intrasimilarnsi totive that ver bsof , thethe main Patient verbs; is thatalways is the to ssayubject,, for butergative for traverbsnsi tivande Suaccusativbject Choe vericebs Hie, rarchythe cho. ice of the unexpressed subject follows Fillmore's The use of si-N forms to express the notion of instrument in such examples may beFoc us"the source(Dahl 1973of the:118 belief, 128; Ferrethat Amisll 19 has71:4 a- 5)si-. preHowevefix markingr, as il lust"Instrumeratednt by the examleast)ple sactually in this has sect a ionrather, the diffderivatioerent nalsyntactic si- pre andfix sema(in nticNata oranfunction Amis. atAs pohassi beenting noteda consi abovestent in "I nssecttrumionent 3.3.2.1, Focus" Iaffix have innot Am foundis, alt suffihoughcient, as basis for ilInslustrumtrateentd. by Itexam muplesst be 3.5re-4-3emp.5hasised6, there that are these verbs verbthat for takems Nomiare nativsporadice seco"Insndarytrument deve Fo"cus"lopments. and it would be misleading to labelthem as

3.4 The Locus Case Relation [+LOC] 3.4.1 Characteri st1cs In Aspects of the Theory of Syntax, Chomsky makes a distinction between two Titypesme conofstit locuationalent, "h ascon nosti parttuentsicula, namr elconny,ec thetion Pl withace cothensti verb"tuent (Cwhichhomsky, like the Chomsky1965:101) re, fersand tothose the locadistintionalction co nstitas oneuents of degr whichees subcategof "coheorisesion" thebetween verb . the veambrbiguous and itssentence accompa: nying nominal actants. He illustrates his point with an (3.81) He deci ded on the boat. whichboat." may mean "He -chose the boat." or "He made the decision while on the Halldistingidayuis mahikesng bea twecorrespondingen an "inner" divi andsio nan in"o hisute r"ci plrcumsace tantfunctionial ro les(Hall, iday 1970:149) . In the lexicase framework, Starosta also posits an inner locative which is termed Locus [+LOC ] and an outer locative which is termed Place [to+PLC Sta]r. ostaWh, ile"may Pla recef erge neralto spatly ialrefers, concept to spatualial, proxim locational, or, Locus even , legaccordinally g deco-occursfined locatio" (Starnsosta, depending 1978:20) on. the semantic class of the verb with which it reIn lathetion lexica, a Locusse framework specifies, given only thethat lothecation Patient of the is Patiethe fundamnt, notental that case of the Agenta whole or is other left actanto thets . PlaceThe case sett reling ationof the (c scf. enesectio forn the3.5 action) . or state as 94 Asver bsan , innesuchr caseas verbs relati ofon locati, Locuson orgene locomrallyotion occurs. Withwith ath eserestricted verbs, therange of preverbssence. Stof aros[+LOCta ] furtheris obli pointsg�tory andout thatits pre, sence in turn characterises these Patient and Locus can be said to identify each other: onlyif a oneloca oftional the sentenceconstituent's actants singles andout letheave locathe,ti poonsi oftion , oflocational the othe rco partnstituenticipants is anin the(inne actionr) Locus open, and, the the actant whose position it specifies is the Patient (op. cit. :20) . The so-called "indirect object" of ditransitive verbs in Tagalog has been re-analysed by DeGuzman (1978:51-54) as having the [ +LOC ] case relation instead of the Dative [ +DAT ] (relabelled in some lexicase studies as Experiencer [ +EXP ] onor Correspondentthe basis of Fil [+CORlmorean] which case was anal assysisigned. to it in earlier lexicase grammars

3.4.2 Real i sation The Locus case relation can be realised only in the Locative [-Neu,+lctn,-sorc, -goal] case form (abbreviated as Lcv) . The presyntactic notion of location, like that of instrument, can also be expresectionssed 3.4.3.2 via nominal below) pre. dicates or nominalisation in NP-NP constructions (see With only one exception from my data, the Locus case relation is never reali�.ed conin thestruction Nominativ containinge case form a Nomin in Amativis.e LocusThe resemfact blesthat athe nom headinalised of the for m stagsuggestse in Amthatis, we re mightflecting have the caught situation the formingin Proto-A of ustLocus-ronefocussian beat foreits theinci pient deveand seclopmtienton of3.4.2 Phil.2)ippine. focus systems (cf. Starosta, Pawley, and Reid 1982;

3.4.2.1 Locative Locus The [+LOC ] actant usually occurs in a post-verbal position, and mostly immediately after the [ +PAT ] actant. The following examples show Locative Locusthe classi with filocationcation of, locoAmismotion verbs) and : transportation verbs (see section 5.2 f01' (3.82) miZea-cangn-agairaynst c3s ir a waUcabang [ +V ] Nom LCV LCV he is Zeaning agai(+PATnst } (+Det the } waZZ [,+LOC } (3.83) go-up-tota-ngasa c3s ir a' gatepanan toa [mridei -d akaw ] [ +V ] Nom LCV Lev [ +V] he rode up (to+PAT the/a } (+Det gate} (+LOC} ,------_.._------_._------

95 (3.84) rornall-k o roko roo nowan ini thekia stbekoneelo q [tgo-toaraJward [ +v J (��:TJ [ +vJ botsasatom ofno hilolltok the(��: tJstone (��;cJ is ro lling downhill (3.85) stt-andern- i renq nowan ini c3s ir a sidtepea r ofno posalst iI i [ +vJ Nom LCV LCV he is standing by the/(+PAa Tpo J (st+Det J (+LOC J (3.86) itilat -en ko tobaccotabako 2siso baskekopkopt +ergv (Nom+PAT J [Lcv J (+LOCLCV J put+mptv your tobacco in the/a basket heaIn exad N mwitplesh a3.8 pof4 ssessandr ive3.85, attribute the loc ativeN. expreThe headssion N isin than eseNP NPco nsisconstrtinguctions of a nounspeci fies(see thesection spat ial3.4. orientat3.4) . Locativeion and isre laretorferred nouns to behas avea loca syntativecti recallylator li ke commonnouns and co-occur with the locative determiner i. In examples inde3.82-3finit.86,e. the Iflo decativefinit edeterminerness is to i becan sp beecif intieder, preoneted nor asm allyeith useser definit one ofe orthe locclassat ivofe demlocationonstrativ nounses -whoitseia, mem itbersina, are or usuallyitira - markedtogethe byr with the dea nounrivatio fromnal the suffix -an . An example with locative demonstratives is given below: (3.87) thereira are ko chadildiwawaren itia yardpotal {an} (:�ctnJ (Nom+PAT J +DetLcv (�LOCJ +dmns-prxm there are children in the yard-rmte The -an suffix marks location nouns such as potalan yard that are synchronically derived as well as those that have a petrified -an such as kakarayan heaven. For a noun like potal yard or panan gate that already designates a location, the adding of a locational suffix -an is optional. From section 3.1.3.2 we have seen in examples 3.30-3.33 that notional objects can"spe becific realis objeedct ass" Locativeof intra nsitLocusive. veTheserbs. actantsSee sectio aren to3.4. be 3.4unde belorstoodw. as 96

3.4.2.2 Locus-focus in Ami s? WithAmis dothees exception not have Nominativeof only one Locusform forin msmy . data, cited as example 3.91 below, ratherIn Tagalog, common according. When theto De inneGuzmanr loca's destivecri caseption re la(1tio978n: 88-9is 0)realised, [ +NM ,+LOCin the] seems nominative case form, the verb is always marked by the affix -an . Among herDe-G uzmanpage and's exam sentenceples arenum be(herr givenCF's andin parenthe CR's areses ci tedafter without each Tagalog modifica exatiomple)n an 1 (3.88) apppapahly i ran heniy a ng bavarnishrn i s ang metablesa (89:#74.b) +AC +AC +NM he will app(+AGTly varnish ) (+OBJon/to ) the tab(+LOCle) (3.89) launderpinag l abhan sheniy a ng dredamitss na soilednaput i kan +AC +AC (+AGT ) (+OBJ) ang basinpa l anggana (89:#75.b) the basin(:��) was used by her to launder the soiled dress in (3.90) givebinig yan n i Cl ara ng giftrega lo s i Rosa (77.b) +AC +AC +NM Rosa was given(+AGT a gift ) by(+OBJ Cl ara) (+LOC ) To me, these Tagalog sentences look like equational sentences with deverbal witlocah tiothemn nouns the originalserving ascase nominal frame profedi thecat verbes, thestemsse . derivedHoweve nounsr, si ncecarrying the or der ofa nominal the [ +OBJ anal] andysis theuntenable [ +LOC ] canat least excha ngefor therather subjec freelt-inty inernal Tagalog exam, plesthus, makI ing agree with DeGuzman that these are truly verbal constructions with [ +NM,+LOC ] moreand wou advaldnced be wi stalligeng than to admitit is that for Amtheis focus. system in Tagalog is at a much" NominativeThere is only Locus one inex Amampisle, thein myexa datample thatbeing may the sugge folstlow theing : presence of (3.91) storesa-s i ngaq-an ko granaryal il i to graintipos grain is stored in the granary ( ��T Lit. the granary is the (:��c)storing place� for) grains This is not an NP-NP construction because such an analysis would call for a Nominative Patient; and if al il i is the Patient subject, what can tipos be? In other words, to analyse sa-s i ngaq-an as a Noun would leave tipos unaccounted for. I have not considered the alternative to analyse to tipos as a complement of ko al il i with the following structural analysis because nominal 97 complements in Arnisare normally in the Genitive case form or introduced by the ligature a (see section 4.2.2) . The following analysis is rejected here. (3.92) NP

Det NP al iIi koI t i pos ��toI Totreating analyse both sa -sNPi ng'saq-a as nits as co-occurringa Verb, on the actants other. handThe, vecanrb solvforme theadmi problemttedly has by reduits pliorigincation in ofa derived the first loca syltionlab nounle and. thesa -saddingi ngaq-an of theresults loca tifromonal partial suffix -an to mean the place where a certain action is habitually performed. Examples include: ba-bot ing-an fishing-ground, qa-qadop-an hunting-ground, ta-taeq-an anus, in addition to sa-s i ngaq-an storage. Locus-It shouldfocus be formnoted- eventhat thoug the h verbthe suformbjec oft exampleis identi3. fied91 do ases Nom notinativ suggee stLocus a .

3.4.3 Ramifications 3.4.3.1 Motion Verbs Thereof loca istio a n diisstin exprctess classed by of the sit uationaPatient llcasey motion relatio verbsn inst withead. which These the notverbsion Examare plesergat areive : verbs with [+PAT] subject but no [+LOC] in their case frame. (3.93) clima-dakmb-overaw thenia chiwawald ko ferangadnce Nom =s T +PAT +e�rgv ( ) the[: boyl climed (over)(�� )the fence (3.94) ma-hatebo n3s ir a ko c i ri s river T +PAT +ergv ( ) : 3 het:: jumrped (over)�� the stream[�� i 98

(3.95) sma-dancgoyros s n ira ko sc i riam 5 � river 3s ;�: � Gen+AGT +PAT he[:+ergv :swam::l (across() the) river ( ) (3.96) easB�bma-taq l ib aklso ko houselomaq nam2plo whenio goingtara a workdqdema k +v Nom [ +V] [ +V] +trns T +PAT 1 [+ergpassedv (by)( �your house( on way to work I my For these ergative transitive� ve) rbs, Amis does not have intransitive Whatevercounterparts degrees like of those transJ suggeitivitysted might by the be Englsuggeishsted tra bynsla thetion pairs in parenthe of Englishses . formssituation (cf.ally Hoppe mortion and verbs Thompson are unique1980) isin sithemply sense not anthat is suethey her aree. markedThese for Am the:.s Seelocal secistiontic featu5.2.2re bes thatlow. are otherwise marked on a co-occurring [+LOC ] actan1:.

3.4.3.2 Locative Predicates AsIn Amshownis, manywith thesitu notionational of roles inst findrument the (cf.ir expresectiossnions 3.2.3 through) , synt non-CRactic meansdevices. suchthe caseas comsystplementationem in Amis. or nominalisation are commonly used as alternatives 1:0 Thecon strfoluctlowingion can exam alsople s expreshow sshow the a notionnominal ofpredicate location in: a descriptive NP-NP (3.97) kawlefti 1 i iso thekia doorpanan [Lcv] 2s Nom the door is to your lef(+PATt ) (3.98) houselomaq thekia chiadiwldawaren [Lcv] Nom the children are( +PATin )the house Noteare markedthat by by lexicaseCF. The convention Locative , CFnominal, as shown predicates by exam pleare s not3. 97marked and 3.98,for CR, is but Theusually rela tionrealised between as i.su bjeAmctis and doe predicates not have isa exprecopulassed to instlink eadthe by two word NP 's.orde:c , ini. e.th, ispre casedica, te[L precv] cedingfor the su predicatebject, and asthe oppo disedffer enceto [N omin ] casefor theforms su, bjecwhicht. l�:; , Whenlocativ a elocativ elemente nominal carries predicate with it theis notionfollowed of by lo acation verbal in comreplelatiomenn t,to the the geactionneral orsetting state inindica spaceted where by the an veevent rb. takesIn other place words and , henceit indicates the notion the of 99 exprePlace ssor thethe notionouter oflocative Place . in Arnis It turns (see outsectio thisn is3.5 the) . mostExam ples.common areway : to (3.99) homelomaq kakls o a eat-dma- lainnerbi [Lcv] (:��T) [��int} I eat dinner at home (3.100) Nataoran c ira a ma-aroq Nataoran 3s live [Lcv] he lives in Nataoran(:��T) (��int) (3.101) na [ herit i en i ko frcabienday aklso a [grOwtataang-u ]] (:�ast) [Lcv] (Nom+PAT ) (+v-fint my friend grew up here j (3.102) theritilae -an a [Rutpat el i ] ko bookcodad nom i so tabsapatle [Lcv] Nom 2s LCV put your books(��int) down the (table+PAT ) (+LOC) on (3.103) granaryal iIi-an niam ko graintipos a [ mi-storsie ngaq ] [Lcv] lexcl Nom we store (aU) the grain(+PA in Tour) granary(��int) (3.104) hallike-to i [ i granaryal iIi ko ratqedo a [bumiildsa-d-ni estpong ]] [ +V] [Lcv] Nom rats like to build their(+PAT nestsJ in ( the��intJ granary expreIt sufficesssed as to a CRsay infor Arnis thethough time being it can that be exthepre notionssed asof such Place. is A notdiscuss normalion lyof Plgivenace inand seanction account 3.5 beforlo w.some of these examplesof locative predicates will be

3.4.3.3 Spatial Orientation Where English employs a battery of prepositions such as at, on , in, by, befo re , after, from , to, for, between , above , below, under, and through to specify equivalentorientation locativein space reandlator time, nouns Arnis instwou ealdd. use A aslist thei ofr thetrans Arnislatiolocative n semantically reclosseslator nounsand corresponding appearing in prep my ositidata onsis given enclosed belo win with parenth approximateeses. English 100

lalabo inside (in� inside)

1 ikot outside (outside) tepar side� proximity (by� near) kodokodol top� space above (on) sasa bottom� space below (under) qaqayaway front (before) 1 al i kol back (behind� after) terong middle� midst (between) Formconstallyructio, locativns ande earechla oftor them nouns req functionuires the as co-occurrencethe syntactic of hea a dsge nitiof theve irNP , seritsve obs liasgatory the point attribu ofte re, ferenceto indi catefor itsthe spatconcreteial orientatio physicaln. locat Theion head whic Nh is always preceded by the locative Det i. Therelat followingor nouns . are sentence examples showing locative NP's containing locative (3.105) froqaqayant way no stpaationngi roan kia road lalan [Lcv J [Gen J the road lies before the station (:�:T} (3.106) back1 al ik ol no treeki la ng a [mi-lhide imakJ thekia chadildiwawaren [Lcv J [ GenJ Nom the children hid behind the (��int}trees (+PAT } (3.107) locateira ko persontamdaw outside1 i kot no windowsas inar an [ +vJ Nom LCV [ GenJ someone is outside(+PAT } of the(+LOC window J (3.108) locateira ko beartomay inslalaidebo thisnina houselomaq [ +vJ Nom LCV [ GenJ the bear is inside(+PAT J this (house+LOC J (3.109) [5usei -t-laokadder r thekia fawamather a [mi-get-fala]rom r :�intJ r:�:TJ (��intJ thetia catnan i abovekodok odol no ktree i 1 ang ] ( �T J [ Gen] above/the fa��upther in usedthe (a tree� ��cla dde1 r and rescued the cat from 101 (3.110) comekati [ sit-dma-a roqoum middZeterong nolsmako andato n i Moinather 2si so J [ +vJ [ +vJ LCV come sit between me and(+LOC your} mother

The locative NP in example 3.110 is rather complex in structure. An analysis i terong nomako ato ni ina iso in between me and your mother offol lowing tree diagram: is given in the (3.110a) NP Det NP A PP middZeterong LCV �P� NP (+LOC ) nomako ls a [ GenJ and� withL Det A� NP / inI a Mother N n i [ GenJ I iso 2sI A localistic analysis of locative relator nouns will be given in section 4.4.

3.4.3.4 Specifi c-Object Intransitives Asaccordance we have witseenh degreesin sectio ofn sp3.1.3.1,ecificit therey. Spiseci a gradualficity isshif nott ofa feature features in in increaitselfses. , Inthe Fig semanticure 3.1 wefeatures have seenof [+ thatdfnt asJ (dtheef initdegreee), [+ofprsn specifJ (piercitsonay l), and(non-p[+ prnnersonalJ are) andadded [-prnn on oneJ (n on-proat a timenomi. nal)Non-sp. Itecif doesic thennot havemeans to [be-prsn ] indefinitefinite. readingThe Det. to in example 3.27 can have either a definite or It is interesting to note that in Tagalog (cf. DeGuzman 1978:36-38), the shift defromfiniten Accusativess. e Into AmisLocative, howeve markr,ing the ta kesnotion pl aceof spalongecificity the param as desetercri ofbed in thesectio Accusn 3.1.3.1ative Pati comeents intowhereas pla y.the Onsp ecithefic non-spe end iscific occu piedend of by the the speLocativectrum is Locan usabrupt. Thoughchange thefrom ve rbthe formtransit remainsive tounchanged the int ransthroughoutitive verb the categorygradual shiftis , 102 effLocativeected witLocush a arechange to be in unders CR-assigtoodnme asnt no. tioBothnal accus"objecativts",e Patientthough the and latter isas governedthe specific by an ob intraject ofnsit intransive .verbitiv. e Examverbsple. s 3.30-3.33 show Locative Locus

3.5 The Pl ace Case Relation [+PLC] 3.5.1 Characteri stics Inthere the isunive an outersal r setlocativ of casee case re laretiolationns prop, Plosedace by[+PL StaC],rosta corresponding (see sectio ton the3.1 ) , inner Locus [+LOC]. As we have already pointed out in section 3.4.1, the Plsenseace thatconst ituentit can hasoccur no with par ticulaany conrlystr closeuction connection, verbal or wit non-veh the rbavel,rb , andin theany vetransportationrb class without as thebeing Locus restrict case edrela totio verbsn is . of locatioInsteadn, of lospcomotecifyingion, theand loactioncation or of st theate . Patient, a Place constituent gives the general setting of an Being an outer case relation, PLC does not serve to subcategorise verbs. Whconilestit LOCuent obse mayrve occurs the instrict sentenc post-e-initVerbial and po post-Patientsition preceding word theord verber, a, PLC esa psieciamilallr y Locativewhen there case is foalrrm.eady a LOC constituent in the sentence marked by Inlang Arniuages, itfavours is almost a differ unneentce sssynarytacti to havec de vicea separate to indi catePlace theCR begeneralcause settingthe ofcombination an event within time verbal and spacecomplem, entsi.e.., theThe use nomi ofnal nomi prenadicatesl predicates are unmarkedin for CRloca bytio lexicasen of an eventconventio is n.synta Actic nomiallnaly thepredi headcate of that an NP-NP indica cotenss tructiothe genen. ral Locativesentence-i nonitiaminall pospredicaitiontes and are require signalled the byco-occurrence the locative of mar verbalker i comin plements tocom plemindientacate tiothen stractionsateg ywhose is comm generalonly usedsettings in Arnis are for being expre predissingcated attributes. This or beloadverw.bia ls. We shall return to this point in section 3.7 and section 4.3 Howevesentencesr, a givendistinct as exam Placeples case 3.1 rel11 andation 3.1 must12 be belo w.set upTh onese the two basis examples of happen nominalto be the predi onlycates one s areI canabunda findnt in. my data, although examples of locative

3.5.2 Realisation The Place case relation in Arnisis 'realised only by the Locative case form, tosimi thelar l/Sentto the coLocusnstr aintcase provided relation that. Like the sequencethe Locus ofCR, locativ Place e isnominal not su bject themactantsselv markedes. by [+PLC] hold co-referential inclusion relationship among 103

3.5.2.1 Locati ve Place The following examples show PLC realised in the Locative case form. Examples Locative3.111 and NP 3.'s112 in showthe satheme co-occurrencesentence, a sioftuation two non-co-refer that motivatesential the no n-incsettinglus upive conof astraint separate is notPLC violcase atedrelation. in addition to LOC in Amis so that the l/Sent (3.111) donaka 't hang-piqoc an-to i 1 ici Moinaanther schooLcacodadan QMhaw? [+V] +-fv int don't hang( onJ to Mother(���c at schooJ L,(���J OK? (3.112) wia LL findmakatepa 2skis o 3sici raan somewhereicoacoa markepacakt-payayLace [+v] ��int � T you'LL( findJ him(: somew: J her(�e��c inJ the(���cJ market-p Lace(���J noThen-inclus l/Sent iveco nslocatrainttive requiresconstitu entsthat weas havingtreat the two nodiffn-cero-refenter CRential's.. The two henceLocative permi Plttedace con by stitthe uel/Sennts tin conexamstraiplent. 3. 112 are co-referential inclusive and It should be re-emphasised that these two examples are the only instances in my datalocal thatNP's ca avllaila forble inner in theand syst outeemr , localthere CRis's now in the quesamest ionsentence of which. With one totwo assign when there is only one loca" tive NP constituent in a sentence. Since the Amisthe ou laternguag caese favoursrelation the., I useam otinclnominedinal to pre interpretdicates anto expreonly ssloca thetiv notioe CR nas ofthe oneinner locat Locusive instCR wouldead. haveIf itsuffic had noted forbeen the for who exalem pleslanguage 3.111. andFortunate 3.112, lyju, stthe accuracyseeming lossof des in criptioneconomy andin the our Amis extensCR-system ion of th isis compsparensselyated used for CR by tothe cover the benotionlow) . of Reference, with the help of localistic features (see section 3.5.3.2

3.5.3 Ramifi cations 3.5.3.1 Locative Predicates We have already dealt briefly with nominal predicates as an alternative sy(Antacticgain, it deviceis not to clea exprer whetherss the notionthe notion of lobeingcation expre in sessedction as the3.4. predicate3.2 above . is anOnce in neragain or I anam outerin favour loca tivofe theor notioneven whether of Locus th isbeca isuse a mean I feelingful a clqueoserstion . decohescrsiipton ivree lacontionstructio betweenns . the Sppredicateecifically and, it the is Patient the location subjec oft in the such Patient suconventionbject that no is CR sp needecif beied marked by the on lo nomcativeinal predicate predicates. , EvI wouen thougld sahy byth atlexic the ase locatthan withive predicatePLC.) constituent is probably more closely associated with LOC 104 Apre stdicarikingtes whichsimil aritwilly beexists des cribedbetween in Arnis sect locativeion 3.6.3. pr edicaThiste showss and thethat tempothe ral exprnotionsessed of by tim sie milarand loca syntactiction fo derm vicesa natural whether semantic or not clathess CR-system. They are is Ramiinvolvficaedti. on Forsect examplesion of theof Locuslocativ casee pr edicarelationtes . see section 3.4.3.2 under thE'

3.5.3.2 Referent as Place The notion of Referent (REF) is also known in the earlier lexicase literature as1979 Bene:150)fit andor Benefor thefactive purpose (BEN of) . suggTheesting name changea broader is rangefairly ofrecent rela tio(cf.nsh Faganips.' Accor"targetding or to·evaluative Starosta re(197ference6a:23) point, the Reof ferentthe action case orrela stationte asident a whoifiesle ".the APatie REFnt is. anIt oute is rthe case action rela ortion st beatecause as a itwhole does thatnot spis ecievaluatedfically evaluatefor its the peporsonsitiv ore (othr ingbene. factivThate) certain or negat personive (oror thingdetrim entalis the) pointeffect ofon re a ferencecertain. Theclass occ. urrence of a REF is optional and should be independent of the verb InsteaIn Arnisd , ofthe having notion a ofseparate Referent CR callcan beed Resubsufermedent , underwe may the extend Place thecase notio relan tionof . Plaevaluatce toion incl. ude A conReferentstituent as carrying an entit ythe which notion ser veofs Referentas a point can ofbe remarkedference for [act+PLCant,+goal types] . are Thinis com anplealysmentaryis is possi distrbleibution because. , Thataccording is, the to Arnis my datalang , uagethese does not have sentences with a [ +PLC,+goal ] co-occurring with another non­ conco-restitferenuenttia neverl [ +PLC co-occurs] . In withother another words, whatPlace mightconsti betuent called and a soReferent the l/Sent constraint is not viol, ated by treating them as the same CR. As was mentioned ofin outersection locativ 3.5.2e.1 is above more, Lofrequcativentlye Pl aceexpre ocssedcurs by only the veryuse rarelyof nomi naland prtheedic notiateson . Thidentiis leafyingves the Plevaluaace CRtive rather reference free pointto take of on the th actionis addi ortional state function. of Thecase notion relatio ofn Referefurthernt charac, repretesentedrised byin ththeis locstudyalistic as an dir insectiotancenal of feat theure Plac ofe [ +goal ] , is realised by a prepositional phrase with the preposition saka offotenllowed pronounced by a locativ withe theNP markedi-glide by inser a located,tive thus determiner giving rise like to i the(s akare -analysedand i sakay) [ ] form . Except for the preceding preposition and the feat'ure +goal 0:1 thecon stheaitdue N,nts thein examreferentples NP3.1 11is siandmi la3.1r 12in. formHowev to er,the itLocative is like Plly acethat a relocaferenttion . is Examoftenples animate are givenwhile beanlow oute. r locative always refers to an inanima'te (3.113) cookmisa-k-dinneralabi c i Mointhera fosarka lexclitam ia n rns Nom+PAT frt Lev+PLC Mot( ��her ) prepares dinn( er ) for(:: us ) [+goal 1 105 (3.114) buildpa-ti reng kam i to schcacodadanooL fosakar {y) p'eopLebinawlan nia vipaLlaLage lexclNom :�rns +AGT ���T :!frt +PLCLCV we-e burgviLt] a schoo] L for( the peopLe( f the vi( Uage] [ ( ] a ) (3.115) makemisa-k-soupabi nowan InI kakIs o fosakar {y) flal ima a peopLetamdaw ��rns (:��T :!frt �+PLC�� ( making soup for fi] ve peopLe] [+goal ] I am ) ( (3.116) si-niala kia mantamda w to bagkara wad fosakr a {y) frpaiendl iw n ira Nom 3s s] (+AGT (ACC+PAT :!frt (LCV+PLC the-e rgvman carried the) bag for his) rfriend) ) Even with this [extended:: use of the Place case relation, the Amis language does examnot plesseem to3.1 have11 and any 3. co-occurrence112 given above of. LOCSequence and PLC ofin Locativ the samee NPsente's seemsnce beyond to be lilocatmitedive moCRstly, that to is,co-re byferenti eitheral LOC inclus or PLCive conand stnotituents by bot markedh CR's. by the same The notion of referent can also be expressed by verbal means in Amis with the verb patado heLp followed by a verbal complement as in example 3.117 below. (3.117) hpataP do kako a wrimi-pteacodad-Letter tia blimabotek-aynd-one a womanbabah i IsNom �+trnst +AGT ��int ACC+PAT ) [-ewrorgvte1 a Letter for the blind woman ( I Lit. I heLped the( bLi) nd woman( write) a Letter

3.5.3.3 Habitual Place The following examples are NP-NP constructions which feature nominalised forms pecarryingrformed the. meaning of a location where a certain action is habitually (3.118) bel iq no lotok ko qoaqoay-an non iam next-beyond hiLL rattan-area lexcl [ Lcv] we gather our rattan on the next :��ThiL)L Lit. our rattan area is on the next hir LL (3.119) o hunting-gqaqadop-anround nonlexcliam thiskina fiomeLaqd [ Neu] Nom this fieLd is our hunting ground +PAT ( ) 106 In each of the nominalised forms, the root is reduplicated to indicate halocabittualion nounaction. whTheile habitual the derivational place nouns su ffixare -anthus indobtainedicates thatthroug ith deis rivationa .

3.5.3.4 Co-referential ity and the l/Sent Constraint Even though the metatheory allows for the co-occurrence of inner and outer caselocativ fraesme, . the ThAmisis isla nguagedone by showsemploying a str ongdiff preerentference syntactic to avoid devic crowdies, oneng witthe hin whichand one show wi thoutboth the CR-innesystr andem . outeTher caseonly re instlationances appe I havearing in in my the Ami sas medat a againsentence to havecheck been co- regivenferential in secitytion of the3.5.2.1 locative. In elethmentsis section. , they are ci tE�d (3.120) donaka 't hang-piqocon-to i 1 ic i Mointheraan schoocacodadanl QMhaw? [ +v ] ��intJ (LCV+LOC (LCV+PLC don't hang( on to Mother at schooJ l� OK? (3.121) wia ll mafind-ka tepa 2skis o 3s i c i raan somewhereicoacoJa markpacaket-ayaypla ce [ +V] Nom you'll (fi ��intJnd him somewhere(+PAT (LCVin+LOC the J ma(LCVrket-p+PLC ) lace (LCV+PLC ) Wethat ha vetwo alreaof thedy pointedlocative outNP 'sin incon eajuchnctioJ of nth eseto thesentense examcesples are notin sect co-reionfer 3.5.ential,2.1 norsuch isci onercums inclustancesive that of wethe assothigner . themThe two l/Sent differ coentnstraint locative would CR 's.require und,=r Now I would like to point out that the two PLC actants in example 3.121, icoacoa somewhere and i pacakayay in the market-place, however, hold a co-rrepreesentaferentiontial ofand sentenceinclusive 3.1 rel21ationship is given bebetweenlow: themselves. The tree (3.l21a) wi.. �all S NP NP NP I I fimakatepand N N N k2s i so 3si c i raan somewhereicoacoa 0/1marpackeakt-ayayp lace ( ��T (LCV+toc +PLC +PLC : J J (LCV ) in/at (LCV ) 107

3.6 The Time Case Relation [+TIMJ 3.6.1 Characteristics standThe Time in caan seimmediate relationcorre is, spondencelike Place to, an the outer Patient case. reRathelationr, which Time givesdoes not the Timetempo deralsig franatesme ofthe re generalference settingfor the wholeof an eventaction in or te strmsate of. itsIn othelocatiqnr words, , verborien typtationes. , and duration in time. Like the Place CR, Time can occur with all Whileplace, recog we arenisi requng theired par byallels the l/Sent in the constraint semantic tochara setcte upri twosation CR 'ofs, Platimece andand Timereferents, to accommodateare never in th aeir co-re co-occurrenceferential ininclus theion sam ree lationshipsentence, . since their

3.6.2 Real isation Time (TIM) does not have a unique case form associated with it. It is· arerealised given inbe lowthe . Locative (Lcv) and the Accusative (Acc) case forms. Examples

3.6.2.1 Locative Time The following examples show TIM realised by the Locative (Lcv) case form. (3.122) didna [0 whatmaan ko dedemaedkan ni AdopAdop yesnacterday i 1 a a afkalternoonahokan J ? : ast [NeuJ :�:T +TIMLCV what( ; did) Adopdo ye ster( day) afternoon? ( ) (3.123) cornersalopiko na did [0 stpacaorekayay befna-oreaya w J :;ast :��pc [ Lcv J at (���c)the corner3 ( there) used( to) be a store before (3.124) niklifeaqortimeip ako 0 Amnopangcaqis-language ko soaeechl nomako Is � Is have���M spoken Amis all::�dC my life :�:T I ( ) ( ) l ) Lit. alr my life3 my speech has been Amis (of PangcaqJ (3.125) bedkakaktimeema tan thel e-angeno 'f to nako � Is LCV+TIM +trns� +AGTGen ( ) [+ergvr ( ) 108 ko clbodoothesy ako ; lateritoa ppacelol to pajamassakabot iq ls - (:� T} t +PACCAT } at bedtime,� take off clot[:-ergv::he]s and then( put on the pajamas I my (I) Note that in examples 3.124 and 3.125, TIM occurs before the construction hea.d, whetherThis pos iittion is prea verbceding or thea nominal construction predicate head, juindicatesst like PLCthat in TIMexample and PLC 3.1 are23. outer CR' s. An inner CR such as LOC would not be allowed before the Patien-t. Example 3.123 shows a topicalised NP-NP construction embeddedunder the verb na. The Locative Time constituent i na-ayaw before is analysed as the prSinceedica CRte's ofare the not NP-NP marked construction on nominal thoughpredica itt escome, thes at-notional the end Tim of e thecon sentencestituent . in example-3.1 23 is simply marked by the -case form [Lcv].

3.6.2.2 Accusative Time The following examples show TIM realised by the Accusative (Acc) case form. to, It is not restricted to the Accusative form of the non-personal article 0, ofthough Accusa totiv ise theTime most are frequentlygiven belo w.en countered form in this context. Example�; (3.126) takeala- en thiskina bookcodad a [pbring-hoatalomaqme] thistina evening1 ab i Nom ACC +ergv:�rns (+PAT} ( ��int} (+TIM} take[ +impr this l book home tonight (3.127) doncaay't ka [olikelah] kakls o a [ {lO-outtahka 1 to evening1 ab i ] : gtv � int (:��T) [+p] ( ��int} (��;M} I �don 't like to� go out at night (3.128) o( whatmaan ) ko fosaodka( -ranaorm)- br eakfast no2smis o to tomorrowcila a morningdabak? [Neu] f what will/y(:ou� �ThaveJ for breakfast tomorrow(��� moMJrning? (3.129) kauickl amkam sa kako a [sidre-bssodoy ] to evdabaery-morning-dabak 1 ls l��rnsget dressed(:��rJ quick (:�ly�T everJ y morning��int (���MJ I ) ( J 109 (3.130) ma-ngtirede roq sa kakls o a [tgo-a-homeIomaq ] to tina-medade-after-workmak � +AdV � ACC ( alway"srns come (+rnnr homeJ tired(:�:T} after (work �int ] (+TIM] I � J As we already know, to is unmarked for definiteness and the �IM expressions with to can be interpreted as being either definite or indefinite.

3.6.3 Ramifications 3.6.3.1 Temporal Predicates The presyntactic notion of Time, like that of Place or Instrument, can be expreexpressssedions by may nominal be used predicates as nominal in NP-NPtime-word con stpreructiodicatesns. inLocative descriptive tempo NP-NPral weconstr simplyuctio�s mark. theSi ncehead pr N edicwithates the areLcv not case marked form . forExam CR plesin a arelexi: case analysis, (3.131) sevenpi toay sa kakls o a [Ige-et-um-op ad ] [Lcv] +AdV+aspt (Nom+PAT [ +p ] (��intJ get up at[-drtv seven1 J I (3.132) sevenpitoay a timetoki a [mhavea-Iabi-supper] kakols [LCV] (Japanese loan) (��intJ (Nom+PAT I eat supper at seven o'clock J (3.133) middlete rong exhanact t no sevenpitoay a o'tacltokockia n a 01 [Lcv] (+AdV+rnnnr j / [ setm i-I i neng ] nomako ko timetok i (Japanese loan) +v lsGen Nom +trns (+AGT } (+PAT ] +e-fintrgv I set the time at exactly seven o'clock Thecomplem locativentatione Time of[ pr aedi finite1 cates verbin examof plesaction 3.. 131-3.133 all require the Locative relator nouns such as terong middle are used as head nouns of temporal expressions. As with locati�e expressions, these relator nouns indicate orientation. By referring to certain points on the time axis like qaqayaw before, te ong middle , toa after, the event r(see ' section and4.4) . we can indicate the time in relation to 110

3.6.3.2 Tel ling Time in Amis Stathan.rosta one ins(1978tance:496 pe, r506) sentence has indi providedcated thatthat, thefor sePLC nominal and Tim co, nstitthereuen cants be ar:.ngmore thedomai san.me CRThus are theco-re follferentialowing sentenc and incluse fromive Amis of , onethough another with within at least the twosame independent [ +TIM nominals, does not violate the l/Sent constraint: (3.134) ] yesnaciterladay [i fosepataur y ko timetok i] to afternoonnamalahok [Lcv] (ACC :��T) \+TIM) maan(���M) ko demakan nomiso? ( o what doing 2s [ Neu ] :��T yestyou doierdayng? at fo(ur o') clock in the afternoon3what were One might have already noticed the Nominative marker ko in the time expression i sepatay ko toki at four o'clock and wondered about its syntactic analysis. The expression i sepatay ko tok i is actually a complete sentence with a phraselocative wit dehoutscript anyiv estr predicateuctural mo, embeddeddification and. used In othe in plr wordsace of, aexample single 3.[ +TIM134 ] takes a sentence with a temporal predicate and uses it, subject and all, as a Titheme hourphrase and. theIn ha examlf hourples 3.1respe35activ andely 3.. 135b we shall show how Amis expresse�; (3.135a) eiwaghtlo-ay ko titokmei nomako a [mdeparta-l in gad ] Nom ls [Lcv] (+PAT ) [+p] (+v-fint ) I leave (the house) at eight (3.135b) eightwa lo-ay ko timetoki [ ira ko halfpangk iw] Lcv ] Nom [ +v] Nom [ (+PAT ) (+PAT ) a [ arrivetangasa working-pkadademakanlace] [ +V] (���) I get to the office at half past eight whichThe half means hour ther is e expreis onessed half, by anused embedded as a comexistentialplement of sentence the locat, iveira dekoscr pangkiptiiw,ve predicate nominal, i wa l o-ay at eight o'clock. Both 3.135a and 3.135b have an abstract quality noun toki time as subject and a time expression, simple or beingcomplex sp, ecias fiedthe islocative given predicateas the verbal. The com plementaction whose of the tem descriporalptive setting NP-NP is The half hour, ira ko pangk iw, timeconstructio in examn.ple 3.134, is an unmodified embeddedlikesentence the inclu. Treeded portion diagram of schematiccorresponding repre tosent exaationmples. 3.134 and 3.135b are given below for a clearer 111

(3.134a) s

D Det S NP maan Det NP nad 1 a N ;r-0 �demakan �NP DetIt ko nomr ; '0 De/1t nama 1 ahok sepatay DeAt to toki

ko yesteryou doingday?at four o'clock in the afternoon� what were (3.135c) S PP PA S wa 1 oay Det / ; i ItoUi ca a NP ko �Det tanga,a� Ipangki iw Det� ko kadademakan

I get to the office at half past eight 112 Thefol lowingAmis way form oful coua: nting a numoerbetween ten andtwenty is based on the [ ira ko n] n (3.136) sabaw 1, 9 n-ttwentyeen exist also involving an embedded S. The number11 is therefore sabaw ira ko cacay and 12 , sabaw ira ko tosa. Twelve o'clock is then i sabaw-ay ko tok i ira ko tosa as in the following example: (3.137) aboutkia probabhaw ly [ i twentysabaw-a y ko timetok i [ existira ko twotosa ] [ +v] +Adv] [Lcv] Nom [ +V ] Nom [ (+PAT)' (+PAT ) nomals ko a [ gota-out 1 apota 1 a [meat-lualahoknch] ] I'll probably( ��intgo out) for lunch( ��int at ) about twelve r TheDuration syntactic, o a deviceperiod used of tiforme , theis expressssedion as of theduration extent isof thatan actionused foorr state. exprinvolvingessing des thecriptive notion profedica Incremteents and de verbalscribed com inplem sectioentationn 3.7.2., si milaIt r isto a thatdevice ofused diff forer encethe exprbetweenession the of expre Timessi. onThere of Duration are, howev ander Time, three: (1) intere with stDurationing poin, ts the descriptive predicate is not marked by the locative marker i, (2) the subject is not tok i time, an abstract quality noun, and (3) while the head of heada Time of predicate a Duration is predicate a derived isN, an morp underivedhologically numeral marked. byHer ethe is suffixan example -ay , : the (3.138) twentysabaw [ existira ko sevenpito ] a yearmihca c3s ir a +N [+V] Nom Nom (+nmrl ) (+PAT) (+PAT) NataoranNataoran a [mlivea-ar oq J LCV he (+LOClived ) in Nataoran(�� intfor) seventeen years Without a preceding determiner and a nominal suffix, the status of the head of sabaw sabaw ira ko pito a mihca seventeen years ais Duration unclear. phraseSyntacti such callyas , it inis more like a Noun because it takes an notattributiv analysede nominal as an adcomjeplemctiveent because, for examit pleis not, a mispecihca fiedin sentence by degree 3. li138ke. the It is deresescmriptivbles ea verbadjec tivin esthat (cf. it Sccanhachte be mordi MSfied:19) by. anIn aspectualsome respe advercts bsaba suchw as henay already, as in the following example. 113 (3.138a) twentysabaw athenaypead [ exirista ko sevenpi to] a yearmihca mrl +AdV+prft (:� ) [+cont} 'c3s ir a NataoranNataoran a [mlivea-ar oq ] Nom+PAT LCV+LOC �int he( has) aZready( )li ved in Nataoran(� ) for seventeen years The adverbial element henay can be regarded as a sentence adverbial that is pointattached out tothat the, heain Amisd of theat leaconstructionst, it is not. Peunusualrhaps thisfor nominalis a good ele plmentsace to be modified by tense or aspect, as witnessed by the terms na-s i-pida-ay the one who used to be rich and Na-taor-an Nataoran3 the pZace that used to have the taor trees, na the numerals herebeing as knoNouwnns , tomar beking the them indi witcatorh the of pastspecia tensel contextual. I have feature analysed [ -[ +Det]] to ,indicate that members of this subclass of Nouns do not co-occur witprovidh Determineed. rs. In section 5.3.1 below, an alternative verbal analysis is

3.7 Other Case-l ike Notions Thefro m una ivmemersalbership set of casesix - reAgentivelations , (A)since, Instr itsum inceptionental (I) in, Dativ 1968,e has(D) , grown orFact moreitive as (Fsho) , wnLo cativin manye (L)Fil, lmoreanand Ob jecasectiv egr amm(0) ars(Fil. lmoreFor exam1968ple:24-25), Nil toson a dozen Body(1973 Pa:1rt03, For130-132)ce, Mat haserial lis, tedExp erfiiencerfteen "D(Deepativ Cae)ses, Ob",je namctivele,y AgManneent, r,Instr Extentumen, t, ReIn asonhis , studyLocative of Bunun, Temp, oralJeng, Source(1977:62-, Pat63)h, hasand alsoGoal come(inclu up dingwith Benea lisfactivt ofe) . havefifteen st ayedcases within. Lexi thecase range gramma of riasevenns , toas tenFig ureCR's. 2.7 inIn sectlexicaionse 2.4.1, a consta showsnt, efattemfortpt beginning has been madewith toStar curostab any's reunnjeecessction aryof increasea [cause r]in casethe CRrela inventorytion to be, the recentadded on list behal off CRof's de proporivedsed caus byativ Staero verbssta (1 982(Stad:ro17)sta, 1974given:28 in5-292) Figure . 2.5The ofmo st REsectF, ionand 2.4.1TIM. , consists of ten members: AGT, COR, INS, LOC, MNS, NCR, PAT, PLC, In the sections above, we have already presented PAT, AGT, INS (including MNS), LOC,inter mePLCdia (itenclu causeding of REF) an , actionand TIM.. As INSsho wnis broadenedin the disc touss meionan anyin sectimmediateion 3.2and .3 underand me AGTans andcan actuin secallytion be 3.3.3redist underributed INS in, theth esefunction two case of notionalrelations instr. ument I would like to point out again that, though the notions of Patient, Agent, Locusexclus, ivInstrely byument case, etc syst. areems , uni.e.iver bysal marking, not all the languages nominal expreactantsss thofese the notions verb withthan CRone's ki andnd CFof 's.syntac Intic Am deviceis, many wh ilcasee oth-likeers notions, such ascan Manne be exprer, mayssed bypas by s morethe 114 case system and use verbal complementation as the only means of surface maPlanifceest asation the oute. r+n locatsectionive is3.5 expre I havessi blealso almostshown exclusthat inive Amisly as the nomi notionnal of predisyntaccatetics deviin NP-NPces by constructiowhich the notionsns, also ofwith Ins verbaltrument complemen, Locus, Platationce, . andThe Time carealn beisa altertion witnativelyhin the expre casess edsyst areem . given in their respective sections on caIn se-lithe kesec notiotionsns 3.7.1,, namely 3.7.2, Man andne r,3. 7.3Increm beentlow, and I shall Conco mitadealnt with, and three show o1:her how ofthe Correse notionsspondent are isexpre treatedssed syntacin secticallytion 3.7.4. in Am is. The presyntactic. notion

3.7.1 Manner The presyn tactic notion of Manner designates the manner in which an action is perMannerform edis . expreIn ssEngled byish a andnominal Thai ac(Kultantla varealnijaisedya 1974 in the:53-5 Instr4), umtheental notion case of fo :rrn. In(Kulla the vafonijllowingaya·o s originalis an Engl sentishence examp numberle to begiven follo inwed parenthe by twoses) Th ai: examples (3.139) He packed the su i tcase with care . (3.140) Dangdaeaeng kikhaall mankhon dietaay withdooy recklessnesskhwaampra?maat (48) (::J Dang killed a man recklessly. (3.141) Dangdaeaeng seehen hekhaw thuuk shootying [:�lkap etaa (49) Dang saw him shot with his own(:�l eyes �. � In Amis, however, the notion of Manner is not expressedl through the case verbalsystem. complemRatheentr, , itas isin expre the ssedfollowing through exam theples use: of a stative verb with a (3.142) ma-seng ib c ira a [mi-l ias] sad 3s leave Nom+PAT [ +V] he( ��rnsJ went away( sadlyJ (3.143) hastyma-kala h c3s ir a a [mi-lleaveia s] rns ( � T [ +V] he(� �leftJ in �a :hurryJ 115 (3.144) angryma-ca l ibad cir3s a a [mikick-pa rokod to doorpanan J (: � [+VJ ACC he(� �kickedrns) at the·� T) door in anger (+PAT ) (3.145) si-Iosa cira a [mi-si-olah to wawa n ira J with-tears 3s plead child 3s NOm [ +vJ ACC in( � �tearsrnsJ she (ple+PATadedJ . to her child (+PAT J 3.142-3.145 The main verbs3.142 -3.in144 exam ples ma - are all stative verbs. While those3.145 in, examsi-Ioplessa, is derived frombelong lo sato tear the withcla thess derivational, the main ve prerb infix ex si-amplewhich is used to derive verbs from nouns that are the implied3.2 instr.3.2.um2)ents, of the actions ·designated by the verbal complements (see section marking another podesigssiblnatee general�anneris isation similar that to in that nat uralfor desiglanguagesnating the inst syntactrumentic. device used to (3.146) edeng han n ira a [mi -bawah J kia pan an fc ce t 3s break door �� :�:v (Gen ( (:��T ) he[+trns+erg brokev 1 +mnnrthe+rslt door +AGTby fo ) rce ��int) Lit. he forced[ the door1 to break (3.147) cakewit han nira a [mi -bawa kina beke loq h ith-ease lift J this rock :� ::W t:�� [;� [ +vJ (:��T) he[+ergv :liftedl this rock[+rs withltrl ease�T) han Theergative use oftra nsittoiv ede versigbsnate. man Thenerse verbsis common are withresul tativea specia verbsl class whose of efunderivedfect is indicated3.147 by their verbal complements. Verbs such as those in examples 3.146 areand often manarener di ffadvericultbials to glossand notbecause transit theirive verclosestbs. equivalents in English Example 3.148 below has an instrument reading. It is parallel in structure withCR's markedthe Ami ons examnominalples constituentsabove in that but the by ca verbalse-like de notionsvices instare eanotd. expressed by (3.148) with-a-bucketsi-bakic kia womanbabah i a [mdraw-fromi -radom to waternanom i wetebllom J Nom } ACC LCV the(�� rnswomanJ drew water(+PAT from the( ��intwell1 with a (bucket+PAT } (+LOC 1 116

3.7.2 Increment The notion of Increment (NCR) has been introduced to the universal set of ca!;e relanesstions and caseby Starostaassignment (to inapp pairsear a)of to En dealglish with sentences issues liofke st theructural followi relang: ted­ Mary handed the book to John . (3.149a) AGT PAT LOC (3.149b) Ma ry handed John the book . AGTMa ry gave thPATe pig to NCRJohn in a blanket. (3.150a) AGT PAT LOC MNS Mary gave John the p·i g in a blanket . (3.150b) AGT PAT NCR MNS The correspondence between the (a) and (b) sentences has been handled by a number(cf. Jackendoffof linguis 1972ts with:156-178 a trans; Akmformationajian and ruleHeny cal1975led: 178-17"Dative9,1 Movem83-186)ent. " sentencesIn a lexi caselike analys3.151cis point, the tol/Sent the needconstraint for est andab litheshing saturation a new CR of. CR's in Mary towed John three mi les in a dinghy. (3.151a) AGT PAT ? MNS (3.151b) Ma ry towed John three mi les in a dinghy. AGTMa ry towe d PATJohn th ree mi ?le s in a LOCdinghy with he r motor launch. (3.151c) AGT PAT three mi les ? LOC INS ThePAT , constia LOCtuent, or an INS due to inthe these l/Sent exam coplesnst raicannotnt� posConceptusibly beall y,an AGT,the notiona of Increment (NCR) expresses the "extentI to which the state or effect applie!; innerto the casePAT " re(Slationtarosta, ifto theapp earnotion a: l)is. expresuppossedse asthe a CRIncrement on nominal would co benstit anuents , because of its close relatiowenshipigh with the PAT and its obligatory co-occurrence with certain verbsThe elsuchephant as we ighsin twoex amtoplens. 3.152. (3.152) PAT NCR? HoweveCR's, namr, ely,there theis Correstill spondeone innernt (C OR)case, whrelaichtion might in beStarosta extended's universal to cover theset of notiondistribution of Incr inement the sense(NCR) thatsince theythe dotwo notseem seem to beto inco-occur complementary in the same sentenceinvestigatio withn. the same verb. I shall leave this topic for future stIn ateAm isor, efthefect pre apsyntacticplies to notionthe Patien of Int"crem, isent not, i.e.expre, thessed "easxtent a CR to. whichRather thE!, it is expressed as the predicate ofbaqk aet des weigcriptivht, etenes cons duratructiontion , whose tasuraqyaqbject is Zengthan abstract, or, quait maylity be noun expr suchess edas as a predicate with a verbal complemand ent, in andconstructions Place when comtheypa rabarele not to thoreprese sentedused to by expressCR's. the notions of Manner, Time, 117 Examples 3.153-3.161 below show the notion of increment expre3.1ssed60 as the descriptive predicate of abstract quality nouns. �ay Example has nominal whpredicatesile the othersclearly have marked nominal by thepredicates nominal thatsuffix rese mbandle adthejec presencetival predicate of Det s. Thethat heads do not of require these constructionsa preceding Debet.long The to heada special N in exaclassmple of 3.1nume59ral is nouns especially interesting because it has the nominal affixrp but not the Det. In whichthese exa amp"stateles", theapplies notion to ofthe incrPatientement. is to be inte reted as the extent to (3.153) how-much,hacoa to-what-extent ko weightbaqket n i ra pigbaboy QMsaw? Nom how(:� mrlmuch) does that pig weigh?( +PAT ) (3.154) thisina smada"l"l-iding-oneay a pigbaboy , fosepaur t a tenpol oq +Neu+topc mrl [ ira ko )im a] a dodoy kia ko baqket ( fi1 ve pound abo( :�ut ) weight Nom as for the sma"l"lpig, it weighs about (thirty+PAT ) -five pounds (3.155) how-muchhacoa ko widthdadaha l thisnina windowsas i nga ran? how( :�mrl wid) e is this(��: window?T) (3.156) how-muchhacoa QMsaw ko temperaturetohem ninthisa dayremia d? what( :�mrl is) the temperature(:�:T ) today? a po loq a dad i dekan ko taraqyaq nia kayakay (3.157) fi1 imvea ten metre "length bridge the mrlbridge is metres "long T ( :� ) 50 (:�: ) (3.158) hacoa [tadah nomako itisoan ]? how-much ko debt Is 2s howmrl much do owe youT ? how( :� much) is I debt(:�: )to you? Lit. my (3.159) what-extenthacoa-ay ko sdadahal no wearingsasabodoyan nomi so? ��: 2s what(�:u ) size do you (wear?+PAT ) 118

terong-ay ko tataang no kaway nomako (3.160) o medium-one bigness package ls my package(�:u} is medium-(:�:siTzed} (3.161) how-mUchhacoa ko duratenestion no codadletter a [tarrivea-ngasa TaipeiTa i hok o J? +N Nom (Japanese loan) how(+nmrl long.} does (it+PAT take } a letter to get(�� toint Tai} pei? Examples 3.162-3.164 below show an alternative way for expressing the notion of unexpreincrementssed, that subj ectis, asof thean embedded predicate verbal of an constructionanimate subj. ect Inthat these is examalsopl thees , theapplied notion to theof incPatientrement. is to be interpreted as the extent to which an "effect" (3.162) fi1 imvea alhenayread 3 still a miyearhca kako a [ hereit in i a [mlia-ave roq J J +N +AdV ls I'(+nmrlve alread J (+sptly lij ved here for five(:�: yearsT} ( ��int} (3.163) manypina atreadhenay a yearmihca kohn i na [mmara-riedramod J m 1 they(:�mrl have) (:: been:�l j married for quite��+PAT a few(:� astyears} (��int} (3.164) paskinact-onceacay sa aboutkia (kako a [mi-see araw +AdV lsNom (:�ast} [+sptl-cont1 (+PAT } ( ��int} to movieega J to onecacay-ay a weeklal ip ayan (Japanese loan) ACC I used to go to the movies (about+TIMJ once a week

3.7.3 Concomi tant Fagantradit io(1nally979:151) kn ownhas as used the theComi tertativm Concomitante case, de finedcase byrel Tayloration to (1 denote971:42) whatas anis acactivitytant which or stateis associa descritedbed in(s eea par alsoallel Starosta way with 1976d anothe:108l)r ac. tant The inchange the verofbal lainbel -iv eis forin keepincase fog rmswith. Harmon's convention (1977) of reserving terms ending verbsIn Am iswith, the ma nol-tiopren fixedof concomitant to the Noun can which be ma isnif a estpartnered by ora sp companecial ionclass in ofthe activitycomplement indica and tedalways by theindi maincate s verbinde. finiteThe derivedco-parti verbcipa ntsis used. Exam onlyples as area verb: al 119

misa-kero kami a ma l-kaka (3.165) dance lexcl be-with-eZder-sibZing ( ( ( I3I:�int) datogethernced with with:�:T) my myeZder eZder ��int)brothers brothers and andsisters sisters3 danced Lit. misa-toron kami a ma l-wina (3.166) make-rice-cake lexcl be-with-Mother Mo(:�int)ther and I made(:�:T)some rice-cake(��int) Lit. I3 together with Mother3 made some rice-cake mi-tapid kami a ma l-wina to bodoy (3.167) mend lexcl be-with-Mother clothes ( ( ( (ACC MoI3: �intthertogether) and :��TI with mend) Mo thether3 ��intcZ othemendJ s the clothes+PAT ) Lit. mi-pinaro kam i a ma l -kaka tina karireng (3.168) fiZZ lexcl be-with-eZder-sibZing this cart ( ( ( (ACC I3this: �int)together cart with:��TJ my eZder��int brothersJ and sisters3 fiZZ+PATed J pinaro-en non iam a ma l -kaka kina ka ri reng (3.169) fiZZed lexcl be-with-eZder-sibZing this cart I3( �inttogetherJ (with:��T J eZd(er��int) brothers and sisters3 fiZZ(Nom+PATed this: cart my J Lit. this cart fiZZed by me and my eZder brothers and sisters Note that in all these examples the first person exclusive plural form is used evencomplem thougenth, theintroduced English by gloss a, always is in thecomes si immngularedia. telyAl so,after the the verbal first person exclusa ive pronoun before other nominal actants in the same sentence. Since maalsol -kaka introduces and ma l-win a nominala as derived complem nouentns., Thone mayis analysbe temispted is ruto ledinterpret out beca use, doas nounotns take, they nominal should com beplem precededents. byBesid Det esand, therethere isis innone fact. anAls altero, pronounsnative manas ifines theta tiofoln loforwing the exam notioplesn : of concomitant in Arnis using nominal complements, (3.166a) mamike-rice-csa-toron ake kam i a ci mointhera ako lexclNom [ Corn] ls I(: and�int myJ mother made(+PATsome) rice-cake 120 (3.170) bring-mah-kolotogetherm kam i a to younger-saba sibling ako lexcl [ Com ] ls :�int , :�:T a( [m'l i-lveias thistin(a p.ba) laen) area :�J ��� i and -fintyounger brothers+PAT J andsisters band together to I'leave ( mythis } area ( The tree-diagram for example 3.170 is given below: (3.170a) S NP PP mah-kolom

s- I G kami PP PI� S �P� ; NP a!\ V NP mi- ias l I � DetI saba NPI DI bal an to� l tina I NI / ako to saba ako If we adopt this analysis andkami treat, as a nominal complement or a mocasedi fierrelation of the. headIn fanounct, to sabawe akdoo, not not rebeingally needa sist aer separatehead of Concomithe maintant ve r:o, does n�t take part in the case frame that classia fies the verb and therefore to need not be marked for CR. Thea to prep osition and the accusative Ifcase form form a compositesaba case, marker for the Comitativekam casei form'.sab a we must assigkamn i a CR to it should bear the same CR as since is include1 sabin a and is therefore co-re[ferent]ial with it. In example 3.170, then, kam i andshould be causealso bethe markedactant by is' not+PAT in. theBecause verb's of domain its co-reof subcategoriferentialitysation with, the l/Sent constrainta a ciis not violated. Parallel to to is 4.3.5for personalato and nam, es� . They are both composite case markers. See a sectioto. n for conjunction derived through a reanalysis of\ Toin Amsumis up, there, while is thnoer neede is toa disetstinct up a caseconcomitant marker forcase the re comilationtative for casethe form thereanalysforeis ofhas th veris yla nguagemarginal. useThe incate theg orysyst willem. not serve to classify verbs and 121

3.7.4 Correspondent Inset thofis case sec tionrelations, I venture. In to view suggest of its the marginal elimination use and of theCOR factfrom thethat univeits rsal shouldfunctions be canchalleng be reed.as signed to other CR's, the status of COR in the system As Fillmore's Dative (OAT) case is broadened to cover such relations as poscarrsessedied- by noun the , suinbj ectaddition of a psychologicalto its traditional verb andfunction the pos as sessor"indirect of a ob ject", init Lehasxicase been rela(cf.belled Fagan as1979 Ex:1pe50riencer and Fig (EureXP) 2.7, and in lasectter ionas Correspondent2.4.1) . (COR) As a CR, the Correspondent by its various names has always had some peripheral fuinstancenctions, whichthe Correspondent I suspect could case be re laastiosignedn has to been other used case to re accountlations for. Forthe verbssubjec oft co exinstistencetuents, locatioof the n,class and poofssessi psychologicalon. Th isve rfunctionbs, inclu canding now perha be ps thatreassig Patientned to is the the Patient fundamental case recase latio reln ationin accordance. with the lexicase claim possessedAlso, the nounsCorre spondent(cf. Eagan case 1979 relatio:54-5n6) has. been I am notused su tore mark that theth ispo ssessis necoressa ofry becausea verb need, by tolexi becase marked convention for CR., onlyWh ilethose predi headcate nouns nom inalsthat are notsiste sistrheadserhea ofds ofto thea verb nom, inalthey attributes are not required within NPto beconstru markedctions for . CR. EvenThe if sam thee reshouldlation apply betweento be the a headCorre andsponde itsnt attribute. In fact is, toit -bcane exbepre anyssed reasonable by a CR, CRit witdoeshout not having have to worryassign aboutit the the local l/Sentistic co nsfeaturestraint . [-nmtvIn the,+drcn present,+sorc anal] whichysis corre, I haspondve chosen to LOCthe Genis aitiv poess iblecase analform.ys isNo. case relation is assigned to the possessor, though The notion of Correspondent has been regarded as particularly useful for excomepres ssingto "m orsomepho logical"indire ctcausative object" veofr bsditra" whichnsiti veresult verbs from, es pecithe causally atwhenivis ationit causativeof transitive verbs verbs are de(cf.rived Starosta in languages 1978:541-542 such as,547-55 Tagalog0). (cf.When DeGuzman morpho logical1978: space341-36 9)to , whichit is thenecessa AGT ryor toanot haveher CRa dummy can be category demoted to or de reassignatesigned the when perceptualja new AGTCorre issponde introducednt. More by causativspecifice allderivationy, the do. wngradedThis dummy AGT is cate regoryinter prehas tedbeen as the Correspondentrequired this , laandtter if re thela tiocasen toofra meto bealre readyinte containsrpreted asa Corresponda Locus. entDeGuzman, l/Sent constructions(ibid.:�43-349) in has which provided such renumeinterrouspretatio ex�plesn of ofCR tra's nsittakesive pl causativeace. HoweveCorresponr, dentI am. notConceptual convincedly that, a downthe downgradedgraded AGT haAGTs behascome to bean reinterpretedintermedia�e or as evenappro priaan immediatetely identi causefied ofas a apot meansentia orl anact instion rumentand would. probaJudgingbly by be DeGumorezman " s exampleappear onces, th isin eitis hera very the plasourceusible or alternativderived casee si ncframe [+esINS. ] does not ever 122 Thus, the following Tagalog examples from DeGuzman (cited by Starosta in exaStarmostaples 19corr�sponding78:550-551) tocan DeGuzmanbe reinter's pretedexample ass 71a,bfollo wsand (t74hea, b)two : pairs of (3.171a) giveibibi gay n i Ruby ang BibleIi ya kay Lynn the Bible will (be+AGTGen given ) to (LynnNom+PAT ) by Ruby (LCV+LOC ) . (3.171b) caipuse-apabto-giveigay n i Jim kay Ruby ang BibleBibIi ya kay Lynn Jim will have Ruby( +AGTGengive ) the Bible(LCV+INS ) to Lynn (Nom+PAT ) (LCV+ LaC ) (3.172a) reiblateinal ita sheniya sa mothernanay ang hapnangyapeningri the happening(+AGTGen was ) related(LCV+LOC to ) mother by(Nom+PAT her ) (3.172b) cauipinse-abato-relatel ita n i Pepe sa herkan iya sa nanay ang happenir.nangyarig Pepe asked her to relate(+AGTGen the) happening(LCV+INS ) to �other(+LOCLCV ) (Nom+PAT )

The strongest objection to this analysis wouldng , be a formalsa. one, that Tagalog non-nominative INS's are always marked by not by Butsa note sathat the notpamamionag ofi tan means ng NP is by ex meanspressed of by and a therela twotor notionsnoun introduced can perhaps by bein assoc iated. Where case assignment is concerned, as long as the l/Sent constraint is not violatconstiedtuents by thiswithin reas thesig nmentsame cla, thatuse , isthere, by endingis suffi upcie withnt syntatwo [+cticINS and] semantic Correspondentsimilari ty to warrantunder INS the. subsuming A saturated o'f thcase'is partframeic willular thenfunction contain of the case relations PAT, AGT, LaC, and INS for DeGuzman's "Cipausativeinabal it Ditraa causensit-to-relateive Inin formsentenceation Ver3.172bs"b above(ibid. :346) as exemplified by Another objection may come from the inertia bestowed by tradition to cling to the notionidea that of the"sub "iject"ndirect, "indirect object" obhasjec tot" beis ato separatebe thought case of reaslation a syncretisrr. Like orlea neutrst, LaCali, sationINS, and of perhapsa number PAT of . caseAlso rela, eventions thougwhichh thereinclude, are inenough Amis atcase relations to go around, the absenceof ankno wExp eriencer is badly missed when it comeslove . to Someepishowtem, ologicaleven though verbs we suchcan assigas n AGTand in psych the ologicalplace of verbs COR (orsuch EXP as, or baseDAT) , ourthe dese finitverbsion do ofnot transitivity seem to be onnotiona syntacticlly tra criternsitiveia,. theseHowev verbser, sibencelong we tofram thees . class of transitive verbs with both an AGT and a PAT in their case CHAPTER 4

AMIS CASE MARKING SYSTEM

As we have already mentioned in section 2.4.2, case relations are realised by ofcase case forms represen that aretation drawn, mediating from a univ betweenersal andset yet and distinct forma sepa fromrate the param univeterersal onset the of othecaser. relations on the one hand and the language-specific case markers equivalentThe case-mark syntacticing system or morphoof a languagelogical consconfiistsgurations of cla sthatses ofser vegr ammto atsigicnalally the presencethe case offorms part. icSiularnce caseeac h relationslanguage . employsThese diff classeserent of syn configurationstactic and are morphospecific.logical configurations to realise their CR's, case-marking is language­ It has also been indicated in section 2.4.2 that the pairing of CR's and CF's ais num notber biu ofniq caseue andmarke thatrs . eachIn pair this of chapte CR-CFr, correlationswe will show can how be CF ma'snif andest caseed by paradig4.2) ms can be. analysed in terms of loca1istic features (sections 4.1 and (sectioandn pre4.3sent) . theIn seCF-CRction asso4.4,cia wetions will inapply Arnis a and local theisirtic ma nifanalestysisations to locaanddirti veec tionrelator. nouns that are used in Arnis to indic�te spatial orientation

4.1 Localism in Lexi case

Sinccase e inJohn a generativeM. Anderson grammar's proposain Thel to Grammar incorporate of Case loc alistic(1971), severalinterpretation doctoral of disserfor furthertations charac in theteri Lexisationcase ofmo delcase ha formsve ut associilisedated localistic mainly withsema nticthe features 197realisa3:12tion1-123; of the Locus case1974 rela:55-57tion ).or its termino(1978)logica l equivalent (Li "locative sub-caseKulla vaformsnija" ya- location, directionClark, extent .. hassource extended, goal the, and use of terminus - to the analysis(197 of8) locative relator nouns as well as prepositions. Bothfeatur Clesark for and the DeGuzm subcanlassific ationhav e ofef verfectivelybs. Acson utilis (197ed 9)local appliesistic semlocalantisticic nominalfeatures inflto ethection anal, whichysis of together both Cla cossicnstalitu andte theModern cas e-Greekmark ingprep systosiemtions. Alland withthese the lexi lexicasecase stu diesframework have . demonstrated that Anderson's proposal is compatible

123 124 In this dissertation case forms are represented by localistic semantic features on Determiners, Prepositions, and Pronouns. Some case forms are realised. by a pronounscombination are ofnot prepos overtlyition in andflec teddete rminefor caser. , Thoughthey are Amis mar nounskeo.for other both thanCF and CR onfor thetheir subcateg sisterheadsorisa. tionSinc of e predicatescase forms which are furtheris dependent speci fiedon the by caselocal isticfeatun�s mofeaturstlyes dir, etheyctional are andin factlocational bundles· instea of distid ofnctiv beinge sema the ntibasicc features feature whichs are themseforms suchlves . as NomThis(i nativis thee) reason, Gen(i whytive), in, and this Acc study(usativ, abbe) reviationsare not marked of case by a abb"+" reviationsor a "-" sigaren, usedthe indicain placetor of of thebasicir corresponding features. Neve featurerthe lessbundles, these and are marked on nouns, determiners, and prepositions. Localistic features' on verbs, howevbundleser, may are benot present replaceable. This by treCF atmentabbreviations.is consist evenent thoug withh thethe convcorrespondingention once bysugge bothsted CF by and Sta CR)rosta, Prep to ositdistinguishions (marked the bysyntactic CF alon e)categories, and Verbs of Nouns(marked (m byarked nespecitherifications a CF nor. a CR) by the different requirements on case-feature noThen- Nominlocalatisticive casefeature for [m+nmtv is dis].tinguished Other no n-fromlocal theistic obliq featuresue case usedforms inby thea [±anacnlysistr] (cofontrastiv Amis casee)-mar. ki[+prdcng system] marks inclu the denom [±inalprdc pr] edi(predicatescativ, bothe), [+andnmtv ] inand senten[-nmtvce-in] onesitial, of posNP-NPitions constructio and has nsa constrast. [+cntrive] marksmeaning topics (see which section occur 4.3.7:section Neutral 4.3: Su mmaryCase Formof Amis) . s Case-MA featurearking tree System for beAmilows case. forms is given in caseBy Acson forms's accougrows nt,out theof incora revivedporation inte ofre stlocal in theistic works features of Maximus in an Plaanalysnudesis , of Maximusa thirteent assigh nedcentury to the grammarian three Greek (cf. obAcsonlique 1979cases:16) - . the Accordingaccusative to, dativeR.H. Ro, binc.nds, geni... sotive that - "ainn entiits more stsema banticsic disti field,nction nam elyof approachrelative ingloc, ationstatic and po movemsitionent, and, separationsemantic field it isof exhaustive relative lyloca dividedtion and between movement them" is by(R obinsno means 1972 restricted:108) . Thi�: to variousthe characte case-rismaratiokingn deviof caseces infleare usedction to onexpre nounsss . the Dependingnotions imonplied the byla nguagethe , casethe affixrelatioationns . or Assu ppletioFillmoren ofhas nouns pointed or pronouns out, case (F markillmoreing 1968is not:21) li. mited to If the application of localistic �emantic features to case-marking systems i�; theoranalyseticis toally such sig nisupplementaryficant, we shouldor alter attemptnative todevices extend asloc thealisic use offeature relator ponoussinsble, prep toosit speciionsfy , certainpostposi clationssses, orof otherverbs parsuchticles as the. locatioIt may n,even mo tiobe n, �Id iftransp suchortatio infornmation verbs iswit noth diralreectadyional carrie, oried byntation one alof , theor co-occurringlocational featur nomiesna:." moactantstion verbs. See in, Amisfor examwhicpleh would, sectio requin 3.4.3.1re such forin formationa class ofto sitbe uatiomarkednall ony the verbsaccordingly for subsequent in section sema 5.2.3.2ntic inte and rpe5.tatio3.3.2.n. These verbs are subcategorised 125

4.2 Case Paradigm in Amis ' Except for pronounsp, Amis nouns do not overtly manifest case inflection .. reInsflecttead , theit caseis the·mark repihosig. tionsIn theand presentdeterminers study that, therefore precede, locthe alisticnouns that feature Deteranalysismine rsis esapplitabledished to Pr as6nou a nsseparate, Prepos cateitionsgory, distinctand Determine fromrs Prep, withositio ns on the one hand and from Nouns on the other .(see discussion in section 4.2.1 beloreferredw) . toAmi ass nounssis·terheads do not forover thetly subcategori reflect casesa tioninflec of tiothen, V'sbut andsi ncP'e s,they those are toofeatures. marked on Det's and Pronouns must be redundantly marked on the Nouns inflIt wasec tedmentioned for case that. , Howevexcepter, forthe pronoutheoryns req, Amiuirs esnouns that arecase- notform overtly features be maron kedthe oncase all features nouns, siof ncetheir the sissubclaterhe.assifdsi. cationInfor andmation select carriedion of by ve .rbsthe dependhead N ofon anits NPco-occurring constructio n ' orshould Det. be coIn nsitheste casent, ofif notlocativ identicae relatorl, with nouns that (s markedee section 4.4) and derivedp -an nouns, the head N actually carries more orin formationless neutra+ aboutised rel locatativeive location marker , i.dire ction, and orientation than the more

4.2.1 Determi ners Determiner (Det) is a constituent of NP. In Amis, a Determiner always precedesPP. every non-pronominal N and is never followed by a P within the same Determiners' can be distinguished from Prepositions on four counts. First, a non-pronomico-occur innal the concrete same prep Nounosit ionis alwaysphrase preceded, it is expected by a Det that. Ifthe a P alwaysand a Det precedesare derivationally the Det. relatedsecohd lyto, nominalsome De terformsmine thatrs, spcaneci serficavelly as theheads Demon of strativnominales , relatedconstruc totio nsverbs. Prep(cf. osiCltionsark 1978, on: 26the7-28 other7). handSuch, ditendfferences to be derivationallyin derivational potentproductialive keepfor thethese two closed categories categorie aparts. evenThirdly thoug, h Determinersthe processes are areinflec not tedver y whereasfor case Prepin Amisosit ionsand showshow onlyan infle one ctionaluninflect paraded igformm forfor mem eachbers lexical of th isentr classy , beloand ngingInstrum toent th areis clamarkssed. onlyLas tlyby , Deterwhereasmine mostrs, therecase arerela tionsat lea suchst two as casePatient are Detlations if the that head can N be is renotal isea pronound in the. formThese of CRa PP's, aremarked Locus � ywith a P thefol lowed by Comia pronountative, PCF immandedia Place/Referenttely precedes wit theh Nthe wi Benethoutfactiv an intee CFrven. ingWhen De thet. head N is The distinction between Prepositions and Determiners is obscured by the fact thatsuch somas ettheimes Comi ittativ takese ora Pthe and Be nea Defact tivto e joinintly Amis ma. niOnefest isa temcertainpted cato setreat form , thedecide com biwhatned it forms syntac as a ticsin gleclass lexical is. entrIt isy, notbut a thenDet becauseit is di itfficult does notto thfollois walter an natinfliveectional, the ob paradligatorigm;y norDet beforeis it a aP non-pronobecause weminal wil N.l not There havefore, by , 126 we have accepted the solution that it is' a combinatorial case marker, with twc separateas the sig lexicnal alfor items the presencefrom two ofdiff a erentsingle syntacticcase rela categtion, oriesor rat joiherntly, a siactingngle CF-CRof Cla associassical tionand Mode(cf.rn Acson Greek) 1979 . for the combinatorial case-marking systems The inflectional paradigms of Determiners 'parallel that of personal pronouns in Arnisinflecti (seeonal se ctionprefixes 4.2.3 on be Nounslow) . ,IWheren a sensesupporting, Determiners phonetic look evidence like caseis notlacking without, the' some dist degreeinction of is arbi indeedtrarine tenuousss that and we it set must up be the conceded grammat thatical it is category of Determiners apart from Nouns. The decision ca� perhaps beia the, juinastified this, by their a derivationalthat relation that exists betinweena-an formsthis plalikece i ra-ay thatand one. and nominalised forms i a,such in a,as ira and As inflectional affixes, and would not have intrcl, e thisrequired independent status to engage in derivationsini. hereAlso , historically a may have been developed from the relaa tor noun plus the ligature while there has been no evidence that would beini attracted ina to a following nounsynchro to nicformally a lexin Arnisical , unitas the. Formsfollowing parallel pair ofto examplesand willcoexi show:st (4.1a) i tinia n i a roq at this place (4.1b) itina niaroq at this place andit is easy to see how the second could have develop���from the first. Even though such derivationaliti napot ential or history does1conclusively determinE' the status of a form like as a Determiner itasin oppoa sedni arto oqa nominal prefi>:, it does show that, at a certain point in time, and a could be pretreatedceding as entrtwo y separaterather thanlexical the entrfollowingies, with one . the Thligeseature forms attractedcould very towe thell be one step behind the pronouns which have already acquired case itiinflesoanctional to you� withparadig youms andby noinmakocorpo ofrating me� bycase me infle. ctional prefixes, as in [±ArnisdrnnDeterminerss] (demonstrativ can bee), subcateg [±prxm] orised(proximate) by the andfeatures [±rmot [±] prsn(remo] te)(persona as shownl) , in inthefle fectionalature treeparadigm in Fig ature the 4.1, bot withtom ofeac theh Dettr eefit. ting into the case noIn needArnis , forthe a Geseparatenitive caseInstr formumental also casemarks form non-su. bject Instruments, so there is Petherersonal is pronounsa gap in theand [+nounsDet,+are prsn never] col markedumn in byFig urethe Accusativ4.1, comparae caseble toform that, so in the paradigm for personal pronouns (see Figureci 4.2 in sei,ction 4.2.3) . Note homophonyalso that witthe h tothepic locativform ofe [+formDet of,+prsn [+Det] is,-prsn and,-drnn nots]. perhaps to avoid 127

+prsn

.�+pr= -rrnot7\ +rrnot

+Det +Det +Det +Det (+Det ) [-drnns-prsn1 +dmns-prsn +drnns-prsn +drnns-prsn +prsn -rmot-prxm +rrnot-prxm [+prxm1

i a ira ina Topic 0 ci Nominative ko kia kira kina c i Genitive no nia n i ra nina n i Accusative to tia t i ra tina iti a iti ra itina ici Locative '" G� " I" ( " ..... Figure 4.1 Feature Tree and Infl ectional Paradigm of Ami s Determiners 128

4.2.2 Prepositions and Combinatorial Case Markers

Amis has a very small inventory of prepositions. The meanings of locationin,, onorientation, above , be, andlow, direbeforctione, aft as er,desig benateds ide, bybe twesuchen , English prepositions as expressed by prepositions in Amis. Some of this etcfunctional. are in loadmost iscases shi ftednot to relatspatialor orientanouns whichtion . are Theused use extensively of relator innouns Ami s hasto exprecompensass tedtemporal for theand lack ofinventory overt case. Locativinflectie onrel onator nouns nouns and in theAmi smals arel presentedsize of the in prep sectionositio 4.4n belo�l. The remainder of the locational load is shifted to the verb system, which can directioalso be n,characte sourcerised, and by go allocal, likeistic those sema nticon Tagalog features verbs suc hdescr as ibedlocation in , DeatteGuzmansted 'seven anal inysis a la(1nguage978:43 as, 181-18rich in7) . prepositionsThis kind ofas Engdivilissih,on whereof labor the is � � illufunctionstrated of bya diexamecplestional 4. prep2a andosi 4.2tionb belophrasew, cantaken shif fromt toBennett the verb (1975, as:20) : (4.2a) The brideg room has gone into the church . (4.2b) The brideg room ente red the church. The "goal" element which is realised in example 4.2a by ena dirterec tional PP is exprethessed chu inrch exam-, ple 4.2b partly by the transitive verb and partly by the NP, and no direenctionalter PP is needed. Some Amis locational verbs resemlocatbleive NPthe actantEnglish to ver repreb sent in"goa thatl", they"sourc doe" not, or require "path" . a co-occurring These insi tuatiosectionnally 3.4.3.1 motion as verbswell areas seanalysedction 5.2.3 as tra.2 nsitand 5.3ive .3.2.verbs and are presented beBenne shttifted (op. to citthe. :2verb5) also, thus gives neutr exalisiamplesng thein which locative the PPdire wherectional directio elementnalit cany is concerned. TheHer ebri aredeg hisroom exhasampl comees: to the church . (4.3a) The brideg room has arrived at the church . (4.3b) to at While the preposition carries the "goal" reading, the preparrivositione does notIn Am. isThe, the "g oalloca" tivelee mei ntdoes is notactually give expreinformationssed by onthe dir verecbtiona lity inseitteadher,. so the functional load is on the motion verbs. Examplesi of Amis motion verbs with5.2.2 loca undertive verbal actants cla markssifedica bytion a merein Chapter locativ 5.e Det are given in section Prepositions in Amis do not form an inflectional paradigm. As a first apprcas-likeoxim ationnotions, th oseof gothatal, occursource in, et constructionc. are given within theNP 'sfollowing and expr essilistng with the correfunctiospondingn and dist Engributionlish gloss toes deter. Wemine shall whethe examiner or eacnoth theyof them all belongin terms to ofthe Preposition classa . with (4.4) ato Comitative with� and namaka Comitative from saka Source for the sake of tangasa BeneGoal facti ve up to� until 129 In Amis, as in other languages, prepositional phrases are exocentric constructionsturn can be rea witlish edob asli gatorya verbal P andor aS, nomi ·or withnal claP useand NPconstruc. Thetio embeddedn. In Sthe in constructionformer case, thea verbal head comof plemthe constructionent. The verbal is a comverbplement and we cocalnsistsl the ofem beddedthe head Vthat which none is ofnon- themfinit cane beand expre its ssedco-occurring in the Nom nominativeinal act case ants form., wit h .theIn theconstraint latter ·beingcase, theco-re embeddedferential S witis han that NP-NP of constructio �he highern. verb The, is su usualbjeclyt of unexpr the embeddedessed, thusS, . giving rise to the· nominal complement which is the NP after P in a preposition clphrausasees. , with In bot theh recasferencees, we mayof there fermissi tong the com embeddedplement goveS's asrned infinitival by the next higPrepheositior clausen can. beSc hematiexpressedcall y,as the twofollowing types oftwo sentencetree di agracomms:plements after a (4.Sa) "Verbal" Complen:tent P � (4.Sb) "Nominal" Complement

P� S I NP Det (NP) (NP) The embedded S in schema 4.Sa is a sentence complement with a non-finite head Vhas which a nominal may or head may , notwhi havch eis co-occurringa predicate nomiNP's.nativ Thee, butembedded no su bjecS int scNPhema. By4.S b referring to the X convention (cf. section 2.1) which imposes a constraint on lexicallexicase head grammars, we mustthat reviseevery node4.Sb ofas afollo treews must: directly dominate at least one (4.Sc) "Nominal" Complement PP P�� NP Det 130 While this is consistent with the. model and its constraints, an analysis like 4.Ssentencec has lostcomplem theents gene. ralAlisasotio, itn mustthat , bein pointedArnis, all out comhereplements that inare the in analfactysis oftime an being NP-NP thereconstruction is no ea thesy waysame to constraintcircumvent has this also incon beensist vioencylated in andthe for the constrlexicauctionsse framewo as rkwe. ll, UnI tilam thatgoing ti tome sticwe kcan to solvthee strtheuctural problem analys for NP-NPis of 4.Sb cominstplemeaden of t. 4.Sc because it can capture the generalisation about sentence Arnisprep ositions that a can introduce a sententialka complement which is a verbalna constr(indicatinguction sourceinclu de:or earlier(positi eventve li) gatur. Ine)teresting, (negativ as theye li gaturare ine) the, andir own exclurightde, dthey from are sys nottemati relatedc disc toussio the nsma niin fethstaistion section of case. forms in Arnisand thus prepIn dealositingions with that the in troduceArnis case sentence-marking comsystemplements, we thatare only have intnomiernalested pre indi thosecates , in other words., those that fit into schema 4.Sc. Structurally, the PP has the 4.Sconstituc aboveent representstructure theof noma P inalfol lowedactants by anin NPthe. originalThe optional case frameNP's ofin theschema deverbalsource V N,if thea co-occurring head N is derived NP can befrom the a posV formsess. or ofIf the head N.N is not a with this initial understanding of the internal structures of PP and NP, in attributeconjunctios,n withwe shall a bas nowic undereturnrstanding to our oftenta sentenctive elist com plofem Arnisentsprep andositio possessns ive associif its atedmembers .with are the indeedcase -li prepke notionsositions of. goal, source, etc. given above to see, There are four different a forms in Arnis. One is the a form that introduces airreal sentencis, e potencompltialement, andwit hfuture a no n-tifinitme ase verbopposed and to gives na wh itic ha hareads inga com ofpletiv e Exreadingamples. areIt: fits into schema 4.Sa and is to be treated as a preposition. (4.6) ma-ngalike� wantlay (a) [ eatk-em-aen to candypi ang ] ko wawachild [ +V] [+p] �� ACC Nom the child wants to( eatint candy ) (+PAT ) (+PAT ) (4.7) malik-nge a lay QMhaw k i so a [ eatk-em-a en to fibotshin g]? [ 2s +V] Nom [+p] �� ACC do you like to eat(+PAT fish? ) ( int) (+PAT ) (4.8) ohah , canden gay ki50 a [ use-5 i -maamaansomething ] a [mien-b ohat tia doorpanan :I [ 2s +V] Nom [+p] �� [+p] �� ACC oh� you can open(+PAT )the door (withint something) � int) (+PAT)1 131 (4.9) donaka 't (a) [phiti -t aes thetia chiwawald] [ ] [ ] [ ] don+V't hit+p the chi+V ld ' (.+ACCPAT ) a Therevowel oris asemiv tendencyowel. to Thedrop use ofespe a ciallyto link when the themain pre vercedingb or predicateword ends into a sentence complement is quite common in Amis, as attested by the expressions 3.3, involving3.5, 3.6, theand notions3,7.1 respectiv. of Instrely)ument. , Place, Time, and Manner (see sections Another a form in Amis is a main verb. It belongs to the subclass of Aufut'urexilia timry Verbse. Exam becauseples incit alwayslude: requires a sentence complement. It implies (4.10) a [ i eoa kako a [ma-lal itemoq ieiraan ]]? shall where meet 3s [ ] ls [ ] +v Lcv Nom ) +p ) (���c) +fint (+PAT (��int where[+f+xlryutr shouZd 'I meet him? (4.11) wia ll [ do-noteaay ka [ rainoradan ] ] +v +v [ +p ] �� +fint +fint ( int) it+f+xlry isnutr 't Uke+ngtv+xlryZy to rain In Amis, tense is expressed by the tense verbs na (past tense) an'd a (future . tense) . Parallel to manner verbs, tense verbs are treateda as heads of aconstructions result of elis witonh verbalwith the com previousplemertts vowe. Thel. liUnlikegature otheris miauxiliassingry, probably verbs, as however, the tense verbs take finite instead of5.1. infinit e verbal complements. We shall returna to the tense verbs in section TheN-a- N.third Exam formples isinclu the delig: ature element in nominal compounds of the form (4.12) havesi-s-a-cooaqpai qay-Z-breezea-ba l i dusksansa n deb [ ] [ there+V is a cooZ breeze atl+TIMLC dusV k) (4.13) oZdma-t-peopoas-ayZe a persontamdaw [the +N ] eZderZy person[ +N ] (4.14) thisan ini a dayremi ad this day�[ +p] today [ Neu ] 132 Intermediate between the compound form in 4.12 and a head-attribute NP construction with the4.1 head3 N 4.1linked4, to the attributive N by the preposition a wia pothoutsses asor De, t,as liinke exampleand 4. 15,. is the head-attribute construction involvin9 (4.15) wawachild no KilanKi lang [ Kilang's+D chiet]ld ( +LOCGen j TheIt fitsfourth into a formschema is 4.s thec prepand ositthe prepion ositia whichon bearsis fol thelowed Comi by tativean NP markedcase form by . th: N.Accusa Thistive prepDet osito tionfor a andnon-personal the deter miheadner N toor or by' c cii constituteif the head the is coma personalposite caseget inmar Amiskers a fornew the form Comi atotativ whice h casereq uiresform . the Throughco-occurrence a reanalys of anotis ofhe ar Detot, wein thelin kedfoll byowing ato bearNP and identical is thus CFtreated and CR, as athis si nglenew lexicalformis untreatedit. Sias ncea the NP's co-ordinatingillustrated by cothenju nctionfollowing. exam The plescombinatorial: case marker a to or a ci can be (4.16) mowinather a to childwawa [ +Det] [ Com] mother and/wi(�:m th child (4.17) kam i a J c i Moinather lexcl [ +Det] [ Com] Mother and�:m ( J I Lit. we with Mother (4.18) ci Ki lang a ci Adop [ +Det] [ Com] �: Kilang and (AdopmJ a a ci 4.16-4.18 doRegard not lesschange of the irCR formof the. precedingIn the fol NPlowing, toexam andple , howevein exr,am plesthe Det after prepato agrosieestion with a, isthat a co-ordinatingof the preceding con juNPnction. This. shows that ato, unlike the (4.19) n i Ki 1 ang andato ni Adop (Gen [+conj] by Kil+AGTlang and Aabp 4.19 J Theare two NPco-ordinated's in pocanssess alsoors be of analysed the head as N. [G en,+LOC]. In that case they We shall have more examples of nominal com3.7.3plements in section 4.3.7.3 below and the readercase-l ikeshould notion re ferof backConcomitant to section or Comitativefor . the Amis way of expressing 133

a, To sum[ up, ] there are two prepositions ato only one of which is marked by a case reformanalys, Coniis of(C theomi tativconibinatoe). rialThe caforsem markeris nota to a. preposition. It is the The formtangasa V is a verb in Amis. V According to my data, tangasa can be used as a maintangasa as well ,as a non-fiVnit e in an enibeddedS. V.,E xamples given below showform probablyused lends as itsa elffinit moree readias welyll toas a prepnon-osifinittione analTheysis no becausen-fini teof Engitslis wordh, orderits Eng thatlish correspondstranslatio n,to and that the of absence an expre ofss 'iona co-occurring involving a P PPbe infore the enibeddedS. If we look at theta-ngasa class of motion verbs in Amis (see seta-ctions motion5.1 and verb 5:2)s , andwe wouldthat, whenfind mothattion verbs arehas usedthe ascharact non-eristicfinite heads prefix of of enibeddedta-ngasa co nstructions, they are not preceded by prepositions. Therefore, ta-ngasa usedis a bothverb andas a notmain a prepverbositi and ason. a non-The finit'followinge verb exainm plesa sentence show complement. ta-ngasa c ira panan a [mi -dakaw ] (4.23a) go-up-to 3s gate ride ] Nom (LCV ] [ +p ( ] he: �introde up to+PAT the gate+LOC ] ��int (4.23b) miride( -dak aw c( ira ] [ go-up-tota-ngasa ] gatepanan 3sNom ( (LCV ] he: �introde up +PATto the gate��int +LOC (4.24) howhac( oa much ] ko( lengthtenes] no lettcodad] er a [tarrivea-ngasa Taipeiihoku ]? [ ] Neu Nom+PAT ( ��int (L(JapaneCV ] se loan) how long does(it take] a letter to get to ] Taipei?+LOC namaka Thereanalysed form as a mainis als verbo ass naociated and a withverbal verb pre fixforms. maka- In, which fact , derivesit should verbs be withnon-fi annit ime pliedverb . sourceThe frfolomlowing nouns exthatamples re fershow to maka-N location verbs or plusedace both, on theas finite verb or as non-fmakainit-eraay verb of anc ireniOeddeda a S: [ r-em-akat ] (4.25) -afar ls walk Nom+PAT +P ��int he+sorc� walked:r (here) (from ] afar ( ) (4.26) flmaFy- rebahoy ko birdqayam na [mO-faka-rom-roofbadahong ] [tgoar-toa--mountain]otok ] Nom +P the:�int bird flew from+PAT the roofl��int+sorc to the mountain ��int+goal ((StarostaJ 1974(:308, ] M121i my reanall ysis) [ l 134 maka-N Asauxili non-aryfinite verb verbna whichin an indi embeddedcates pastS, time. verbsWhen are the usual exprlyessi preonceded of notional by the source or origin follows another verb, the subordinate clause shows na maka-naN isin reinterits predicatepreted . as a In preposthe pritiesonent that system introduces of analys theis , sentencethe auxilia complery mentverb with maka-Nma ka-veNrb as its head. However, I do not rule*namaka out frtheom po ssibility of na and developing into a unit preposition with plus a N, similar to the way a3.3. and3.2 si-.2)may. suggest a unit preposition *asi plus a N (cf. reintersectionpretation becauseAt the Npresent following stage a , prepthe osilanguagetion is isexpe notcted ready to befor this Vieprecededtnamese by has a Dereceivedt. The a relatiothoroughnship treat betweenment incoverbs Clark and(197 prep8:267-287)ositions and in th is topic certainly merits further4.4 stu dy with Formosan languages. saka. The last candidate from list above i to be considered a preposition is Siisnce tem ptingthis form to take combines *sak aywith as athe si ngleDet unito t.mark Butthe thBeneis isfactive not thecase sol formution, it pronomiadoptednal here N becauseafter *sak I amay doesunwil notling req touir makee a theco-occurring concession Dethatt, nor a non­ am I ready to call *sakay a Det 4.2.1itself because it does not fit into the paradigms presented in sectionsaka above. i, I have instead analysed it into two units, allthe prepconcreteositio non-pronomn andinal the nounsDet in toArnis be areconsist to beent preceded with our by as suma Detption. that From the tentative list given earlier in this section, we have eliminated all but two as members of the Preposition category, which area toalso ainv ciolv ed in the Benecase-factivemarking saksystema whic ofh Arnico-occurss. They wi�h are an theNP Comiin thetative Locativ e orcase formand. the

4.2.3 Pronouns In Arnis, within the Noun category, only Pronouns are inflected for case, with "case" figuratively re197ferring9:15) to. those forms which "fall away" from the Nomithe nativ"obliquee (cf. cas es"Acson, and they are Non-Nominativeanalysable in tercasesms ofare di alsorectional referred and to a� locationalsemantical categfeaturesories within that exa haustivlocalisticely dividframewe orkthe. sema Thenti syntc fieldactical of andrel ative location[ andJ movement among [the fivJ e "oblique" [ cases J for Arnis personal[ pronour,sJ are: ±drcn (directional), ±sorc (source), ±goal4.2(g oal) and ±assn compo(associatiosite treen) , givingin Figure us 4.the6 offe aturesectio treen 4.3 in beloFigw.ure which is part of the Ofrealised the five by a"o blPrepositionique" case andforms an , infleCornand cted Ben pronoun are com, orbinatorial a non-pronomi case nalforms N with anthe inflecprepfoostedrition me De. t.sak a Withfol lowedpersonal by apronou pronounns , inthe the Ben Lcv case case form form, is realisuch sedas sakaby itakoan a As for the Corn case form, there are no instances of the prepsystematicosition gap immratediahertely than fola datalowe d gapby an(s eeinfle sectioctedn pronoun4.3.5) . . This can be a 135 CF

-drcn -sorc� +sorc �+goa1 -ass�n +assn Lcv AccI ComI Ben Gen Nom Figure 4.2 Local istic Analysis of Amis Case Forms Amis Personal Pronouns can be subcategorised by the features [±plrlJ (plural) , [±FigspurekrJ 4.3 (spe withaker) ea, chand personal [±addr Jpronoun (addressee) in the as systshownem infit theting featureinto the tree case in inflealso incluctionalded paradig in them paradig at them. bottom In ofthe thetree tree diagr. am The, someNom andof the 'Top + formsand -are thirdfeatures persons are revecomingrsed out in orderin the torig yieldht seq auence matrix. with the first, second, and :+prsn:=n -plrl [ ] +plrl � +spkr �-spkr /--...+sP� �Pkr +addrI -addrI +addrI -addrI +addrI -addrI ls 2s 3s lincl lexcl 2pl 3pl ako iso c ira ita ami amo ohni Top kako kiso c ira kita kami kamo kohni Nom nako niso n ira mi ta niam namo nohni Gen nomako nomi so nomita niniam ako i so ita amo ohni Acc Lcv itakoan itisoan iciraan itamian ititaan itamoan itohnian Figure 4.3 Amis Personal Pronouns 136

k- Nominativthe nominativee personal Det's pronouns such as koin andArnis ki havena. a As Topformativeic or aswh nomich inalis alspredicateo found of.Ln an equational senten¢-ce, the pronoun sternis used with no affixation. We can singsay ularthat whereit has c air a doesformative n·ot sho. w Thea k- onformatively exception for Nomlies nor in isthe itthird diff personeren­ tiated for Nom and Top. Historically, the third person pronouns in cira Ausis troformnesiaallyn relalanguages·ted to arethe deriveddemonstr atfromive demonira.strativ e pronouns. In Arnis, Genitive personal pronouns show three differentn- patterns of realisation. In general, we can say that the presence of an formative manifests ·the Genitnina. ive case form. This is also true3.2.3 for Genitive Determinerssuc h as no and Wefor may non-su recallbject from Agent se ctionor Ins trument.that as inwe Arnisll as thefor Genitive the poss escasesor formof a is pos usedsessed N. Possessive pronouns serving as attribute to a headn- N which is concrete have the option of- taakkingo, istheo, shorterita, amo forms, withohn i the formative left out. Theconsid shorterered clitic forms pronouns because, even andthough the(notir attaa comchmentplete paradigto the m)N doesare betweennot result the inhead a shiftN and in the stress short po forsitmion of , anot poshingsess iveis alprlowedonoun to.. inteA Genitrveneive Agentfullest is andnever clitiq real isedforms inis thea setshort of Geclnitivitic eform pronoun. Int formsermediate that becantween be usedthe as either a possessor or a Genitive Agent. mitTheya, are shorter thann- "full" and, exceptIt must forbe pointedthe first out person that inclusivthough thesee for minte rmediatestill formskeep theform a comformativeplete . paradigAgents. m, Whtheyenever are notthe asful polest·pular fo rmsas thesuch fulas lernomako forms and when no mitit a comesare avtoaila Geblenitiv, e it is considered "good stylenak" o to usemi thta em. The use of a corresponding myintermediate informants formthoug suchh they as use themand oftenwould enoug beh inco nstheideredir own "sspeechloppy . speecForh" by possesinterchangeasive pronoubly. ns, all variant forms, up to as many as three, can be used In Fig(1ure978 4.3:11), I have, for comparative purposes, included the five points ReTopicid (Top) , Nominativeuses to characte(Nom) , riseGeni tivedeterm (Gen)iner , andAccusative pronoun (Asystecc) , ms,and nam Locatively, e (Lcv) , but it must be pointed out that3.1.3, in Arnis, personal pronouns do not have Accusativea notional casePatient fo rmsis . alwaysIn secreinterpretedtion above as a , Locuswe have if alrits eadyreferent mentioned is a that pronounshall retur, thusn tointra thisnsit pointivising in sect a traionnsit 4.3iv.6e veandrb 6.3wit.2h beAccuslowative. Patient. We Since Arnis personal pronouns do not have Accusative case forms, translational equivalentsthe Locativ e ofcase English form. "direct" and "indirect" objects can both be realised in Arnii- s Locative personal-an pronouns havet- three formatives added to the stern: an prefix, an suffix, and a formativetina which corresponds to that found in Accusative determiners such as to and (1that972 :1co-occur38) with common nouns. It is pointed out by Schachter and otanes (1) that the EnglishI hur t"s myselfelf,/ selves"(2) formations have twoI wdintfferent mysel f.us es: reflexive, as in andconstructions intensive to expre, as inss the ereflexive andLike the Tagaintenslog,ive Arni. s uses different 137

Theattributiv Amisbod equivalentey. NP construction of Englis, hthe re headflexive of pronounsthe construction takes the be ingform the of Nan tireng meaning pi-Thesin attributeanot-i tois ati Gerengnit ivnoe mpronoun i so . Examples include: (4.27a) care-for. body 2s [take +vJ care of yourself(ACC+PAT ) Lit. take care of your body/health (4.27b) wama-ngant lay kako a pa- kakapah to tireng ako ls [ make-good body ls J [ +vJ Nom [+P J �� Ace I want to make(+PAT my ) self healthy( int J (+PAT ) Lit. I want to make my body healthy The Amis equivalent of English intensive pronouns also takes the form of set idioligaturems with to' , theas pronounin the fostemllowing link exedam toples 'its: corresponding Genitive form by a (4.28a) ohn i· to nohn i they themselves (4.28b) amo to namo you (pl.) yourselves (4.28c) ako to nomako I myself These attributive constructions are used for intensifying and emphatic precedipgpurposes. theFor verbal this prreasonedica, tethey. Examare plesusually are found: in the topic position (4.29) indeedso 1 i nay , ohnthey-i -to-nohnthemselvesi, macontented-tama to ko heartbalocoq [+Adv J (Top ) [+vJ +Adv (Nom ) indeed� they+emph (their heart) will be conl( +�nc+tented�pcth 1 +PAT (4.30) I-mako-to-yselfnoma ko , notcaay everaca ka [ take�k-em-aen eat to wineqepah J as(TOP+emph fo r ) me myself[ �+v IJ don[+Adv't drink J [+P J (��int) (ACC+PAT )

Interrogative personal pronouns in Amis also form a partialwhe recase, when paradig, how m.many , Imhowpersonal much , wh onesat, correspondingwh ich to English question words listed in Figure and4.4 beloware. not overtly inflected for case. They are also 138 ENGLISH GLOSSARY CASE FORM AMISEX PRESSION who� whom ....:l Nom/Neu cima � whose� by whom � Gen nima gJ to whom� with whom � Lcv icimaan p.. Ben saka icimaanan for whom Neu icoa where Neu ' hacoa how much� 'how many � how many z Neu pina 0 Ul ( ) (at) what time� when p:; Lcv i hacoa ko toke � (at) what time� when � Lcv ( i ) pinaay ko toke H Neu o maan what Neu 0 icoaan which one Figure 4.4 Ami s Interrogative Pronouns Thethat firstof the four other inte personalrrogativ pronounse pronouns in fallAmis into (see a Figparadiure 4.3gm corresponding) . to As(at a) whamattert ti meof orfact when, the are inte notrrogativ pronouens expr. essiIn theons exprecorrespondingssions giv ento theabove English, quhacaloaity and expre pinaayssed areby nomithe nalsubje prctedi NPcate, whichs indicating, in this quantity case, is ortok extente time. of theBoth hacoa and pinaay require a sentence complement with a head V which indicates em actsectioionn or3. a7.2) proce. ss for which a time referent is overtly expressed (see Indefinite pronouns in Amis are derived from interrogative pronouns through reduindeplifinitecatio pronounsn or attr areibut givenion. inCorre the spondfollowehcesing: between interrogative and INTEANDRROGAT EXPRIVEESSIO PRONOUNSNS INDEFINITEAND EXPRESSIO PRONOUNSNS cima who cimacima someone� anyone o maan what o maamaan something icoa where icoacoa somewhere ana icoacoa whenever o icoaan a toke sometime o icoaan a remiad some day maan sa how 0 maamaan ( sa) somehow Figure 4.5 Correspondence between Interrogative and Indefinite Pronouns 139

4.3 Summary of Amis Case-Marking System The system for marking case relations in Arnisinvo lves the use of case-marking deterindividualminers caseand prepformsosit. ioAsns sta. tedPronouns previous arely pe, casermanently forms markedare repre forsented their as Noubundlesns, Prep of ositdistinctivions, ande lexical Determine featurrs. es, Itmo stlyis to local be undeisticrstood semant thatic infe aturethe s on prethatsent corre studyspond case to sevformser alare features not un itthat featur jointlyes; rath characer, terisethey area part covericula labr elscase Accusatform. iveThere (Acc) are, Locativesix such case(Lcv) forms, Comi: tativethe Nom (Cinativorn) , eand (Nom) Bene, factivGenitivee (B en)(Gen). , twoThe byfirst prep fourositi caseons [+formspJ inare con rejualisnctedion by wi dethter a minersdeterm ine[+Oer. tJ and the remaining carryingThe cornbina. a torialcase form case (i. rr.are.ke, ar, bundlewith aof prep locosialistiction and featur a dees)ter minerin a prep eachosi tional phraseconstitue dontes . notThe pose case any fo dirmff oficult a prepy inosi identitionalfyi phraseng the iscase the form virtual of the matrix entire comis biningthe case the re lationfeature ofs of the the head prep N ositof ioitsn andsister the NP Oe. t, whiPrepleositi the oncases are relatio n unmaused rkbyed Veneeta for case Acson relat (197ion9). . TheMy com CRbinatorial assignment CF, howeve approachr, diffis erssimilar from to that previousinstead oflexicase creating works new onesuchs asto DecorrespondGuzman (1 with978) becausethe Comi tativI use e estaor blisthe hed CR's aenefactivefollowing example CF's. : The marking of CF and CR on a PP can be illustrated by the (4.31) PP P------NP � I saka 0/ic i Panayan Ben Lcv LCV (+LOC } Virtual Matrix: Ben Lcv+LOC for the sake of Panay[l 1 treeThe casein Figure-marking 4. 6,system with of the Arnis featucan res be apsuplimmarilycable prtoes bothented personal in the andcompo non­site personal(nominative nouns), [± inclucntrdingJ (contras pronoutivnse). , The[±drc featurn} (desirect usedional are),: [±nmtvsorc}] (source), [±Stgorictlyal} (gspeaoal)king, and, [± [±prdcassn} } is(a notssocia a CFtio n)feature, and . [±prdcIt is} (pincredicativluded ine) the. feature treeCF's becauseand the irit correspondingcan mark both pr semantiedicativc ande use syntactic. differences between some 140 CF +nmtv +drcn -prdc . +prdc -prdc +prdc -,o rc -cntr �+cntr 7\ �o+goa1 -prdc +prdc A .� -assn� +ass n A Lcv (Lcv) AccI Com I Ben Gen (Gen) Nom Top Neu Figure 4.6 Composite Feature Tree for Ami s Case Forms markerOf these co nsicasesting forms of, onlya prepos Com itionand Ben and are a determiner realised by or a a c?mbinatpronounorial. The case othen; areforms all them markedselves by. only a determiner with non-pronominal nouns or by the pronoun tradiFor convtionalenienc labee lsand forreada casebil ityforms, andare toused facilit in theate pla unceiversal of the comp featureariso nsmat, soricesme expandedgrammar litfromeratu�e Figure. 4.2.With theIn exceptionfact, all butof Top Neu(i arec) and traditional Neu(tral) labe, allls the in ca�;e namelexicases end. in I -icouldve , a have convention followed Ha Harmonrmon sets (1 977up :67)for innaming referring case formsto the in topic as theis aContrastive widely recog casenised form and but est abI havelished chosen term Topicin the inst literatureead, because. As thefor latter Neutral , Hait rmoncould's haveconvention been called, but thenPredicative it might orbe Desmicriptivesleading becauseto be co nsitherestent are wittheh prepipredicatescativ e aslocative well. and predicative genitive forms that are descriptive In the following are correspondences between the convenient cover labels and newthe namesfeature for repre the sentationspredicative of Locativethe case andforms Genitive. Note forms that. I have not devi�ed Neutral (Neu) : [ +nmt v, +prdc ] Topic (TOp) : [ +nmtv,-prdc,+cntr ] Nominative (Nom) : [ +nmtv,-prdc,-cntr ] Genitive (Gen) : [ -nmtv,+drcn,+sorc] Locative (Lcv) : [ -nmtv, -drcn ] Accusative (Acc) : [ -nmtv,+drcn,-sorc,-goal,-assn] Comitative . (Com) : [ -nmtv,+drcn,-sorc,-goal,+assn] Benefactive (Ben) : [ -nmtv,+drcn,-sorc,+goal ] The features are marked on Prepositions, Determiners, or Nouns. Specifically, orBen onand both Com the are Demarkedt and onthe P's,non-pronominal while all others head H areof markedan NP. on either pronouns 141 Correfollowsponding encestable : between CF's and CR's in Amiscan be summarised in the CF PAT AGT INS LOC PLC TIM Topic + + �Nominative + + + ? Genitive + + (+) Accusative + (+) + + Locative + + + Comitative (+) (+) (+) Benefactive + ?

Figure 4.7 Summary of Ami s CF-CR Correspondences Genitive Locus is placed in parentheses because it is used only to mark the poparentssessheorses of beca a posusesess ited is noun only. usedAlso jo, intlyAccusative with theAgent Comitative is placed case in form. exampleThe status (see of sectio Nominativn 3.4.2e Locus.2). cannot be decided on the basis of only one In section 3.7.3 and section 4.2.2, we presented the Comitative case formas a composite form consisting of the preposition a and an inflected determiner, ci with personal nouns and to with non-personal nouns. The case form assigned Asto thforese CR nouns assig nmeis ntCorn,, one though may witnoteh thatto one in may all add the Accexam toples the givenCF assig in nmensectionst. ci case3.7.3 relatio and 4.2.2ns. theAs nouns the detai followingled ana a lysisto or of a sentencehave not3. 170been shows assigned, the NPany a ' afo sillowingster consti the preptuentosit ofion the mainis a vernomib,nal and com isplem theentre foreof thenot hea requiredd noun ,tohence be not markedCR and , forif weCR . thinkIf wethat insist attributive that all nouns nouns are are co-re to befer markedential witforh both the irCF head and nouandns 3., 170then, the we CRlook of to the the hea headd noun nouns is PAT.for ourwith solutio examn.ples In 4.1 exam6-4.1ples8, the3.166a most CR'likes lycan candi be daterealis ised PAT in , thebut Nomi AGT nativand eINS case are formalso and pos aresible com bepatiblecause allwith the these Accuin sesactiontive ca4.se3.3.4 form be malownif. ested by the Det to. See also the summary statement Thefol lowedTopic by fo discrm wiussllions be discon ussednominal in prsectioedicatesn 4.3 (s.7.1ectio belon 4.w,3. to7.2) be andimm edianominaltely comin plementssix ensuing (sectio section ns4.3.7.3. The) , seafter CF's we in deal fact wit haveh the already six dis beentinct introduced case forms in prethe sentatvariousio n sectiohere nsis meantdealing to with be a thereca repitulatioalisationsn fromof CR a 's,diff soer enttheir perspe ctive. 142

4.3.1 The Nominative Case Form (Nom) In general the Nominative case form (Nom) corresponds to the ?ubject of the amsentbientence., andExcept impersonal with vea smallrbs, evernumbey sir mpleof verbs sent whenceich hasinc onelude, meteorand onlyolog oneic alCR, realistatedsed as inthe the fol nomilowingnativ redundancye case form rule. Thinvolvingis co-occurrence the feature restr [+nmtvictionJ: can be (4.32) [+nmtv J [-[ +nmtv J J J Thgeneis ralruleise ut overili sesthe theTopic feature form which[+nmtv, ifinst it eadcorre ofsponds the Nominative to the su bjCFect so asof tothe asent subjectence, , isbut redundantly not when bothmarked are [+ markednmtvJ. by Athe sentence feature may [+nmtv containJ. Una Topilike c theand Topicobliq uecase case form forms is notcharacte furtherrised ana bylysed the byfeature loca list[-nmtvic featurJ, a Nominativees. or a The Nominative case form in Amis is marked by Determiners with a k- element, i.e., ko , kia, kira, and kina, if the actant is a commonor an abstract noun; by ci if the actant is a single personal name; and by ca if the personal name has a plural meaning, like the Smiths in English. Except for the third person singular cira, pronouns in the nominative case formalso show a k- element. HerThe e NomiI wouldnative only pronouns mention are in: pakakssio , ngkis thato, cithera, th kairmid , personkita, sikamongula, andr pe kohnrsonali. pronoun cira might have developed from *ci ira meaning the one over there. Theindicator presence of theof k-Nominativein most of CF the, but pronouns since ourand grammardeterminers deals iswith clea lexicalrly an entries or words, no more and no less, morphological pieces like k- are of lexicallittle interestentries tofrom the one synta syntacx excepttic su bclasswhen we to deal anot withher . the derivation of thougIt is hpointed CR's are out in by many DeGuzm instanan ces(1978 neutra:72-74)lised that by, thein TagalogCF's, the at identityleast, even of the veCRrb of. theThe su bjeaffixct markNP isings always on therecoverable verbs that through prese rvethe inaffixformation mark ingabout of the CRfocus's neutr systemalised. While by the Tagalog Nominativ hase a casefull formverb areinfle rectferredion paradig to as mvoice assoc oriated the withonly Nominativthe Patiente Agentand the, Patient Agent , toInstr participateument, etc. actively, the Ami in s thela nguageNominativ permitse case . I haveform . yet Instato findnces anyof Nominstinativances e ofInstr Nomiumenativnt ande Time Nominative or Place Locusin my areAmi srare da ta.and

4.3.1.1 Nomi native Patient Beto ingbe the onlyfundamental CR in thecase case rel ationframe , ofthe a vePatientrb and casein there laNomtioninative is of tencase foundform . In section 3.1.1, we have pointed out that all intransitive verbs have a Patientsubclass su ofbj veectrbs, if, it they is havealways a suthebje Patientct at allthat. isWith realised the er gativein the Nominativtransitive casesubc lassesform, ofco-occurring verbs such with as thethe subcuninflectlass ofed verbs verb withstem . impliedThere instrare umealsont s, i.e.lexicase, the conventsi-N verbsion, andthe thesubj N-enect ofver anbs NP-NP, which construction require Patient always su bjecbearsts the. By 143 Patient case relation, thus contributing generously to the pool of Nominative Patien3.1.2.1ts and. Exam3.1.2.2ples when are weabundant presented in my the data Patient and many case arerel ationcited . in sections

4.3.1.2 Nominative Agent co-ocIn seccurtion wit 3.2.2h tra wensit haveive alverreadybs, and pointed that without that the Agentaccusativ .actantse tra nsitypitivecally subclass of verbs, the Agent is normallyu realised in the Nominative case form. ThoughIn fact notio, Agentnally is thereat the cantop be of animate the acc andsati inave niSumatebje ctAg entsChoice (s Hieee discrarchyuss. ion in animasectionte . 3.2.1)', all instances of Nominative Agent that I have in my data are An Agent always co-occurs with a Patient. If the Agent is realised in the Nomspecifinativeic notio, thenal Patient object is. reExamalisedples inca nthe be Accusfoundativ in e sectcaseions form 3.2.2.1 if it isand a 3 • 1. 3 . 1 above.

4.3.1.3 Nominative Ins trument In section 3.3.2.1, we have mentioned that verbs with Instrument subjects are secocharactendaryrise forms this. subTheclreass is. noAs co nsistexamplesent ver3.5bal6 and affix 3.57a that of wesec cantion use 3.3.2.1 to mayreferring show, theto thecase aff reixla marktion ingof onthe thesubjec verbst NP. cannotIn the bese examuniqueplesly , deterbothmined verbs by otherare marked has Nom by inativthe pree Pafixtient sa-pi-, instea butd. one Alhasl we Nominative can say isInst thatrument the whaffixile theforms presa-psencea- , ma ofmi-, an Acandcusativ sa-pi-e Patientare indi, cattheive Nominative of the notion NP cqp of be inst assigrumenednt . the In the basInstruic mecasent recaselat relionation, whic. h isOth theerwise Patien, thet. Nominative NP would be assigned the Verbssubjects with. implied instruments take Patient subjects and not Instrument

4.3.1.4 Nominative Locus In section 3.4.2.2, we have given the only example we have in my Amis data that showsinstance a Locuswhether subje thect aff. ixIt mark is diingffic onult the to verb te llis on charact the baseristicis of ofa si angle subcla ss of verbs that take Locus subject, even though the prefix sa- suggests the meaning of "purpose" and the suffix -an indicates a location.

4.3. 1.5 Affix Marking on Verbs Even though the Amis language does not show a verb inflectional paradigm and hassubject in the in Ammaisjor isity stillof cases recoverable only Patient from orthe Agentverb suformbjec. tsThe, the verbal CR of affixesthe 144 andcharacte the caserised frames by its are case in fraa wayme andmut uallyimplicational definable rel. ationsSince bet a weenverb CRclass's andis CFthat's, may giv enget a camcharaouflactegedristic by a verbpar ticularform, we CF should, expeciall be abley the to Nomi recovenativer those. ForCR 's example, given the characteristic verbal affix ma- , we know that it either marks intransitive verbs that have only orieU ob ligatory co-occurring actant whichverbs witis hnece nomssinaarilytive thePatient Patient and and genitiv the es bjAgectent . or Withsignals the er charactegative traristicnsitive verbal affix -en , we know that it marks ergative "passive" transitsive verbs and thatactants the, onesubjec markedt is bynec essaNomirilnativy ea Patientand the otheif ther, caseGeni tifrve.ame coWit�nsi thets of two characteristic verbal affix mi-, on the other hand, we can assume that it .marksnot a Patientaccusative. Whtrailensi tivthe e suverbsbjec t andcan thatbe ei thether su anbjec Agentt of orthe anunmark Instrumed entform, isthe nonPatient-specif is icre al(seeised se inction the 3.1.3.1Accusativ) . e case form and is to be interpreted as As for those verbs that take an Instrumentor Locus subject, there are also exmoramphopleslogical in secti cuesons lea 3.3.2.1ding to andthe 3.4.identi2.2fi maycation show of, the the affixsubject mark CRing. onAs theour unamverbsbig togetheruously deter withmine the presencthe CR one or the absence Nominat ofiv ean conAccusativestituent .case· Si nceform therecan is nosubj characteect constiristictuent mark ining the onse casthe esve, rbwe tomust uniquely say that ide Arnintifs y doesthe notCR of have the a fully Tagaldevelopedog or voice Maranao inflec may tionhave or. focus system as a Philippine-type language like

4.3.2 The Geniti ve Case Form (Gen) TyAgentpically or Ins, antr erumegativnt ree alitranssedit inive the verb Ge nihatives a Patientcase form su. bject and a co-occurring A localistic feature analysis yields the following minimal characterisation for wethe may Ge niaddtiv ethe case ad ditioformn: al[+ drcnfeatures,+sorc [+].prdc If,- idntit functions] to disti ngasuis a nomih itnal from pr edicathe te, non-pr(identedicativificationale use) onon thethe othoneer. ha nd, and from the Neu form that is [+idnt] The Genitive case form in Arnisis marked by Determiners with an n- element, i.e., no, nia, nira, and nina if the actant is a cornmon noun, by ni if the actant is a singular personal name, and by na if the personal name has a plural meaning, like of/by the Sm iths in English. Pronouns in the Genitive case for.n also show the n- element, as in nomako , nomiso, nira, non iam, nomita, and nohn i. We can say that the presence of n- manifests the Genitive case form. posThe sessiGenitivon. e CFBy allexicaso marksse convdescrientptivion, e nominominalnal pr predicatesedicates arethat not indi marcateked by CR's. Only CF's are required in the feature matrices of nominal predicates. l Inthe Figurlonge fo4.3rms, wecan havebe usedlist edas nomithreenal alter predicatesnate set sand of Geni Genitivetive Agentspronou, nswh. ile Onity inis pethermi capacityssible forof pomemssbeessrs orsfrom. allGenitiv threee nomisetsnal to appeattributesar as nomiare nalmarked attributes by the reprefeaturessents [G ena Locus,+LOC]: because conceptually the source implicit in the Genitive C� 145

4.3.2.1 Genitive Agent Typically, Genitive Agents co-occur with ergative transitive verbs marked by the suffix -en or by the prefix ma-, as shown by the examples given in section po3.2.2.2ssessive above attribute. Since, therea Ge nitisive st ructuralAgent is am markedbiguit yin whenthe sama gee niwaytive as NPa follows anotherfor free NPordering in a verbal of nomi connalstr actantsuction with, it anis posergativsiblee headfor a. GeSinincetive AmAgentis al tolows ofcome some after co-oc an currenceNP with whichrestri ctionsit has benot weensister pronouns relatio andn. non-pronounsDue to the existwithinence a have�omplex only NP one, certain reading com, witbinationsh the head-attribu of pronoun teand, non-pronomior possessednal-pos consesstisortuents, reading may poruledint outrathe. r 'Examcleaplesrly. 3.46a-3.46e in section 3.2.3.1 serve to illustrate.this Thedetermine similarit the y syntbetweenactic a categGeniorytive ofAgent some andderived a po ssesnounssor an madkes sec ondait dirilffyic ultderiv toed verbs. Given a form like sa-pi -angang , followed by the genitive form nomako , it is not easy to tell whether to interpret sa-pi -angang as a possessed noun or theas anst erringentgative covensrbtr withaint athat co-occurring all common Ge andnitiv abest AGenractt. nouns Si, ncederiv I haveed or set up notunderiv precededed, would by a have Det wouldto be precededbe analysed by asa Detverbs, forms. Exceptlike sa-pfor i the-angang case thatof are numeralnouns ands, vethisrbs assu. mption seems to work very well in distinguishing between

4.3.2.2 Geniti ve Instrument Typically Genitive Instruments co-occur with ergative transitive verbs. occReferringurrence toof seGectionnitiv e3.3.2.2, Instrum weent not is iclie mithatted , tolike only Nomi a nativefew ex amInstrples.um entWh, ilethe examconstipletus entslike marked3.61 andas Ge3.6nitiv2 eseta blisInstrh umethent need in exam forples this 3.5 cate9 andgory 3.6, the0 can very weofll "i bemme direanalysedacy" in the as chainGenitiv ofe Agentcause hadand contrStarostaol. notUnfort insiunatested only , thethere crit iser noion informal Amis synt to acticstreng distincthen ourtion po sibetiontween. a Genitive Agent and a Genitive Instrument

4.3.2.3 Genitive Locus Genitive Locus is the characterisation we can give to an attributive possessor. previousAn attributive lexicase pos studsessories doe, howevs noter really, a Correspondent need to be marCR hakeds beenby a assigCR. nedIn to the poposssesssessorion sois as an toex parperienceallelr, the or anal a correysisspondent that the (csuf.bject Fagan of 1979 a ve:rb150) of . In Figurecorrespondence 4.6, Genit is ivenot Locus a part is of placed the case in par frameenthe thatses becan cause serve th isto CF-CRsubcate gorise verbsis neces andsa ryI am. st ill not convinced that the assignment of a CR to the possessor 146

4.3.3 The Accusative Case Form (Acc) The Accusative case form in Amis can realise the Patient, the Instrument, and the Time case relations. When following a Comitative preposition a, the AccusaInstrumtivente case formrelation can . be matched with the Patient, the Agent, or the A witThe h ccusaa t- tiveformative case inform the is dete typicallyrminers ma fornif esnon-peted byrsona thel' nac.ounscusative. Personal case mark namir..ges, oblikejects pe rsonalare m anpronouifestedns , indo the not Locative have Accusative case form. case forms. Notional direct

4.3.3.1 Accusative Patient As has been discussed in sections 3.1.3.1 and 3.4.3.4, the Patient case notionalrelation obrealisedject marked in the by Accusative the features case [ -prformnn ,has-prsn a ].re ferentIf the wh notionalich is a object. is [+prnn] or [+prsn], [then it is manifest[ed as ]a Locative Locus instead. If' theflu ctuatesnotional between object the is Accus-prnnative,-prsn and] but Locative +lctn. , Ththenis itsful l maspectrumnifestationof Normpossallyible onecombinations would expect in te a rmslocation of features noun to and be forms preced ised very by a intlocativeeresting. Alsdeteormine, whenr andcomparing not by Ami an s accusativewith othe r determwesterinn er,Austrone but thesia patten languagesrn is theresuch . as Tagalthe indeog finit(cf. eDeGuzman Patient 19 wh78il:e36-38) definit, onee Patientswould expect are leftthe Accusfor theative Locative to mark case only otheformr. AccusaAs Figuretive Dete3.1rm showsiners, thalsois markis not de thefinit casee Patien. Exceptts. for the Det to, all Examples of Accusative Patients can be found in section 3.1.2 2 under Accusativ3.3.2.1 undere Patient Nominative, in se ctionInstrum 3.2.2.1ent, in under section Nominative 3.4.2.1 Agein ntcon, juinnctio secntion wit h the discussobject iontransi on tiveLocus ver-fbsocus. in Amis, and in section 3.1.3.1 on non-specific

4.3.3.2 Accusative Instrument InIns trumentsection case3.3.2.1 relatio we haven to demoncoverstrated the notion the feasiof meansbilit yin of Am extendingis. What thecan be notion(Starostaally tointerpreted appear a: l)as or a an"t raje"intermctoryedi ofate the" causeaction is imrealpingingised onsynta thectically PAT" andis not morpho foundlogic to allybe co as-occ theurring Accusative with Nominative Instrument Ins. trumentSince Accusative or Genit iveInstr ument Inconstrastrumentint. wi thin the same clause, there is no danger of violating the l/Sent

4.3.3.3 Accusative Time In section 3.6.2.2, we have given many examples of the Time case relation wordrealis ordered in andthe mayAc cusappeatiarve becasefore formthe . head Wh ofile a Locativeverbal orTime NP-NP may constructionenjoy a freer, .------

147 AccusativeAs the exam plesTime alwaysalso show occurs, the post-vmarkingerball of y,Accus comingative evenTime afteris not the li mitedsubjec tto NP the. Det to , though it is the most popular form.

4.3.3.4 Accusative Agent Thesister Comi NPta tivin ethe ca Accusativese form in Amicases asform marked. As on the the argument Preposi tioin nsectio a reqnsui 3.7.3,res a of4.2.2, the constructioand 4.3 goesn , andthe notentire as subordinatedPP is to be treaundetedr the as heada si Nster of ofan NPthe. head V Evidencethe Accusative in suppo consrttitu of entthis. analysis comes chiefly from positional var.iation of Where CR assignment is concerned, the NP which is a sister of the Cbmitative preposition a normally agrees in CR with the Nominative NP. Depending on the evenverb acla Locusss, theCR, Nominativeand the Accusativ NP may ecarry NP thata Patie is nta part, an Agof entthe; Comian Instativtrumee ntPP , or would agree with it in t�rms of the CR. If this assumption is correct, then weanalys wouldis haveis not a defloasilingrable CF. because that canwe wouldbe matched like to accounta number for of CRthese's. syntactiThis c patternscorresponden withoutces forhaving describing to unne thecessa caserily-mark expanding systthe numem berof the of CF-CRlanguag �. We could, of course, choose to ignore this agreement in CR between the Comiinsteatativd. e NPAgain and ththeis Nomisolutionativn eis NP unde andsi esrabletablis behcause a new we CRwould, say have, Concomi to tant Acintrcusativoduce e anca seadditio formnal with CR th asis we newll CRas . the consequential mapping of the In want of a better solution, I have chosen the first not-so-desirable Nomisolutionativn eof NP mark, soing that the theCRir of agreementthe Comi tativcan bee NP called in accordance upon to prepare to the theCR ofway the bothfor the CR andreanaly CF. sisThoug of a h toI expectas a si nglea host unit of newthat CF-CR associa assocites NPations's wh asich a agree in Accconsequenceusative Ageofnt this wh icsoluh co-occurstion, what on lyI actu withally accus addativ toe thetra nssystitiveme isve rjubsst. the Accusativmy data. e Instrument and Accusative Locus, though expected, are not found in

4.3.4 The Locati ve Case Form (Lcv) ThePatie Locativent, Loc uscase, Pla formce, andcan Timehost . a numTheber locativ of casee formrela tionsalso teamsin Am withis, na theme ly, preposition saka to mark the Benefactive case form (see section 4.3.6). The determiners i, itia, itina, and itira mark the locative case form of commonnou ns. Before singular personal names, the determineris ici. The expected plural form is ica, but I do not have it in my data. The set of pronounsabove which marked may beby therealised Locative in theCF Locancativ alsoe beCF. assigned any of the CR's listed 148

The locative marker i co-occurs with location nouns and locative relator nouns, as in expressions like i lomaq at home , natao ran in Nataoran , and i lal ikol no kilaki langan behind the bushes . The locative Det i is unique in that it is syntactically different from other locative Det's, i.e., itia, itira, itina, and ici on two counts. First, in Amis, +dmns] and [ +prsn ] locative Det's normally co-occur with location nouns characte[ rised by the derivational suffix -an as in itia. lomaq-an in the house and ici ina-an with Mo ther .· The [ -prsn,-dmns ] locative marker i, however, does not require the co-occurrence of -an-suffixed location nouns. In expressions like i kakarayan in Heaven , i lomaq at home , .i lalan on road, roandot i. nataoran in Nataoran , the suffix -an is either absent or petrified in the nouSecondns, lyinc, ludingi is the those sole thatdeterminer refer to that po sicantion be andused orientatio with locativen in timrele.ator All locative determinersand pronouns contain an i- formative. Examples are i c i, iti na , i takoan , it i soan , i c i raan , and i tohn ian.

4.3.4.1 Locative Locus Locus, being an inner CR, serves to subcategorise verbs. Some subclasses of verbslocomotio forn, which ditr theansitiv preesence verbs of, aand Locus psych isological required verbs inclu whichde verbs have ofa targetlocation or, eaan chob classject towardis given which belo pew.rception or feeling is directed. One example from (4.33) ira ko tamdaw potal (-an) locate people yard, outside :�ctn) :��T) (���C) some/the people are outside (4.34) a( [ tara ( kako da-demak-an ] will go ls working-place, office +f+vutr ] +V-f int) Nom+PAT ) (���C) I am going to the office (4.35) pabe( l i kak( o tia( codad iciraan give ls book 3s [ +v] Nom ACC LCV +AGT +PAT ] +LOC } I give the book to him (4.36) ma-olah ( kako] itis( oan ( like ls 2s :;SCh Nom+PAT +LOCLCV I love you Allforma locativtive. e markeFor( morers, } examincluding(ples], determinerssee( sectJ ion and3.4.2.1. pronou nsThere, co nsistare more of anco­ i­ occurrencesubclasses. re strictSee sectioions betweenn 5.3.2 ain Locativethe next Locus chapte NPr. and the various verb 149

4.3.4.2 Locative Place Place, being an outer case relation, is not used in the subcategorisation of verbs and is optional. It also enjoys a greaterr freedom in word ordering in Plthatace, whcanile occur Locative in eit Locusher sentencenormally-initial occurs orafte in post-verbalthe predicate po, sia tioLocativen. Locative Locus and Locative Place share the same locative marker i. Locative relator nouns can be marked ·as Locative Place if its co-occurrence withA sentence the verb may iscontain optiona twol. or Othmoreerwise NP's, markedit is to as beLocative mar�ed Plaas Locativece provided Loc usthe. otherlocations within re fetherred sam toe spaare tialone domadj·acentain. of the other or one inclusive of the

4.3.4.3 Locati ve Time The Time case relation can be realised by either the Locative or the Accusative casefreedom form in. wordLocative order. TimeAccusative resembles Tim Locativee, on thePl aceother in hathatnd , itwould enjo beys a greater restricted to a post-verbal position. The marker for Locative Time, like that of Locative Place, is the Det i. There are also locative irelator nquns that containcan be marked two or Locative more NP 'sTime marked. As as with Locative Locativ Timee Locusprovided or Pthatace , thea sentencetemporal may locationsinclusive ofreferred the oth toer. by these NP's are either one adjacent of the other or one

4.3.4.4 Locative Predicate Locative predicate is a kind of nominal predicate which is the head of a predicatedescriptive, orNP-NP, for constructionthat matter,. anyAs nomiheadnal of predithe constructioncate need not, bea locativmarkede for [LCR.cv ] becaA locausetive it prediis identicalcate is markedin form only to locativefor CF, NPwhich's inis othergiven envitheronm labelents . ExamPredicaplestes of and loca setivection predicates 3.6.3.1 undecan rbe Tempo foundral inPredi sectioncates 3.4.3.2. under Locative

4.3.5 The Comi tative Case Form (Com) marksIn Arni thes, the Comi preptativoseit caseion a formin con. juThenction CR of wit theh anaccus accusativative NPe determinerafter the to preposition a is either Patient or Agent. The composite case marker has the form of a to or a ci , meaning in the company of. The form ato, derived through a reanalysis of a to, precedes an NP that shares theare : same CF and CR with the NP immediately before the preposition. Examples (4.37al nomako · ato n i ina ako ls and Mo ther ls

by(: :�1me and my Mo ther(�:�1 150

(4.37b) itakoan ato i c i Panayan ls and Panay l���cJ +LOC to me and Panay (LCV (4.37c) terong nomako ato Jni ina iso middle ls and Mo ther) 2s Gen+LOC +LOCGen between me and( your Mo ther ( J ato J Actualthe agreemently, ofcould CF beand intCRer bepretedtween theas a twoco-ordin NP's. atingThis co njutreatnctiomentn wouldbecause lea ofve the preposition a as the sole bearer of the Comitative case form in Arnis. Verbsinclude that those are "mutsubcategual" ororised "reci byproca the l"pre verbssence thatof theare Comimorphotativlogie callycase form marked by the prefix mal- such as ma l-taes hit each other3 fight and mal -kaka be (with) elder brothers and/or sisters .

4.3.6 The Benefacti ve Case Form (Ben) Thein conBenejuncfactivetion with case a formLocativ is markede form. by Wethe canprep saositiy thaton sakaBen has fo r a thecompo sakesi te of case marker. The locative form that follows saka can be any of the locative determiners i, ici, itia, itina, or itira, or any one of the locative-marked pronouns. The head of the NP which follows saka is a location noun typically marked by the suffix -an . By rule l(b) in section 1.4.3, a [j] glide is inserted between saka and the unitial vowel i of the locative form that follows, thus giving rise to the pronunciation [sakaj ] and the possible reanalysis of the Benefactive preposition as sakay.

4.3.7 Neutral Case Forms FigIn ureaddition 4.6 twoto addithe tisixona casel forms for msthat, I haveI have inclu labededlled in To thepic tree (Top) dia andgram Neutral in co(Neu)ntrastiv . Bothe use share in sentencethe same-initial morphological position form as , webutll becauseas differ ofent the semair ntic prefundictiocatnsiv, e theyand havefunctions to be as kept the ashead two ofdistin constructioct formsns. , TopWh ileis Neunon-pr isedic ative.

4.3.7.1 Topic Topic is distinctly marked by the features [ -prdc,-nmtv,+cntr] and is used in a pospre-verbalition preceding position the as nominalin examples predicate 4.40 andas 4.in 41,exa morples in 4.a 4sentence2-4.44. -initiaTherel is a topic marker iri or a pause to separate the topic from the predicate that immediately follows. The topic marker belongs to a unit set which has iri as 151 itsfrom only the membeother r.CF 'sA" in topic that isit so hasmewhat the extrasimilar function to the Nomof inativshowinge andcontrast differ witenth therespe Nominativect to an exinternal that thesituation topic utior lisesthe di asc pooursisetional. Yet device it is asdiff weerllent as afrom distidiffernctlyent overt marked mar andking. fit into It isthe treated "case inflec as a tioCF beng.lca�se" paradi thegm topic (see Figformsure are1.1 Pronouin sectins)on . 4.2.1 under Determiners and Figure 4.3 in section 4.2.3 under A topic marked by ci, ia, ina, or ira, or any of their morphological and structural" equivalents in the pronoun category, i.e., ako, iso, cira, am i, ita, amo , and ohni, corresponds to the subject of the sentence. Its position belooksfore beyond the verb the orco nfa nominalines of predicatethe sentence sig". nalsCarol a cOQ Harmontrast or(1977 emph :67)asis, forwh ich "exam(1976ple: 110), has . calledShe has th isill usformtratedthe Contrastivethis notion [+of CNco] ntrcaseas t formby the after fol Jenglowing examgivenples in parenthefrom Kagayanenses) : (Harmon's original page number and sentence number are (4.38) mari an pa-t i n�ga-en-d in batag an (p.67,#3.52) Mary+CN make[ +v] -sleep she ch+NMild +AGT J (:�TJ (+OBJ J as( fo r Mary� she will put the baby to sleep (4.39) batag an utu r-en pa l ka rn i an (p.68,#3.53) child+CN cut[ +V ] Paul meat+NM +BEN (:�TJ (+OBJ J \ as( fo r] the child� Paul will cut the meat fo r him theNote verbthat formin the, agre firsteing Kagayanen in CR with exam theple topic, there (m arkedis a byclitic the pronouncase form attached feature to Am[+CisN], )whateve . In ther the se CRcond the exam toplepic , mayhowev haveer,, nothere clitic is no pronoun clitic is pronoun present bearing. In theKagayanen same CR. exam pleAmis instea topicd constructioof the first.ns , therefore, resemble the second ' " Notcorres allponds CF' s toand the CR Patient' s can be su bject"topica ofli sea d"verbal. Inor Am non-verbalis, in mo stconstruction cases, a to. pic Acontrastive topic, like func thetio sun,bject and , thisis nece is whyssa riwely finddefinit thate moifst it topiis tocs haveare markeda by definite determiners like ia and ira. Examples 4.40 and 4.41 below show the topic occurringin a pre-v beerbalfore po nomsitioninal , predicatas opposedes. to examples 4.42-4.44 that show the (4.40) ia tamdaw ( i r i ) na taes-en haw nomiso]? man TM , did [ hit QM 2s [ TOp] ast �int (::�T (:� �+ergv J as fo r the man� did you hitJ him[ ? l 152

(4.41) ira babahi (iri), si-bakic a [mi-radom woman TM with-buoket draw-from [ TOp] [ +V] (��int] to nanom tebom ] water well (���T] (���cJ as fo r that woman3 she uses a buoket to draw water from the well (4.42) ia tebom ( i r i ) , o ra radom-an to nanom we ll TM plaoe-to-draw-from water [Top] [ Neu] (���T J nira babahi that woman (�:�T] as fo r the we ll3 it is the plaoe from whioh that woman draws wa ter (4.43) ia lomaq ( i r i) , o ni-pa -ini-an n ira babahi fami ly TM reason-far-bringing that woman [ TOp] [ Neu] (Gen+AGT ] a [ m i -radom to nanom tebom ] draw-from waterAcc LCV (��intJ rl+P AT J (+LOC J as for the familY3 it is the reason fo r whioh that woman is drawing water from the we ll i a toka r ( i r i ) , sa-pa-ala nia wama (4.44) 0 ladderTop TM meaNeuns-f ar-getting fa ther (:�TJ tia nani oat (���TJ as fo r the ladder3 it is what the fa ther used to resoue/ge t the oat with In my Amis data, the [Top]-[Neu] configuration shown in examples 4.42-4.44 4.seems41. to Inbe amore great frequent number thanof cas thees , [Ttheop ]nom-[+V]inal pattern predicate of exammarkedples as4.4 [N0 eu]and is willa noun be demorerived di scussfrom aions verb on stnominalem. Itpre seemsdicates abstract in sect andion inde4.3finite.7.2 be. lowThe. re 153

4.3.7.2 Nominal Predicates In Amis, there is a case form I would call Neutral (Neu) that is marked on a nominalsubject . pre-Bydicate lex icaseand provide conventios newn, nominformationinal pre ditocate identis arefy mark or edchar byacte CF risebut notthe by CR. _ In this respect, a nominal predicate differs from a topic. There are [+threenmtv pr] oneedic, atii.e. ve , casethe Prformsedi catein Am Nomisinativ, but e.I haveAs onlyhas beengiven shown a new inla belthe- to the discompotinguisite shedfeature by thetreeir abovefeature (Fig speureci fi4.6cations) , Nom which, Top, isand reca Neupitula can beted below: Nom: [-prdc,+nmtv,-cntr] [-prdc,+nmtv,+cntr] Neu: [ +prdc , +nmtv ] In a strict sense, [±prdc] (predicative) is not a sub-CF feature and the CF lafeaturbelses. nor mallyIn the cover present only thestudy feature, the feature[+nmtv] Neu (nomi reprenativsentse) andthe thefeature loca listicmatrix thedistin featurection .-[+prdcPredi,+nmtvcate ] Locativewhile othe andr Predicatelabels do notGenitive norma woullyld make have that to be markedGenitiv efor forms the featurearesimi [+larlyprdc marke]. Sid, nceI have the predicativesimply extended and non-prethe use dicativeof one predicativelabel (Gen) Locativeto cover withboth . the Thelabel same Lcv ap. pliesExam toples the arepredi givencative be lowand. non­ (4.45) lomaq kia nan i houseLCV cat (+prdc ) (:��TJ the cat is in the house a (4.46) pitoay sa kako [ I -em-oad ] seven ls get-up LCV +AdV ] ] (+prdc ] +aspt-drtv 1 (:��T (��int at (seven I ge t up (4.47) itakoan kia nan i with-me cat (� dC (:��TJ the�� cat] is with me (4.48) nomako kina alobo mine this bag dC] (:��T) this(::� bag is mine As we were saying, the new label Neu will be used to mark those nominal withpredicates [Lcv] andthat [G areen ].identical Examples in 4.4form9a with to 4. the51b Tobepiclow andcorre notspond those to idtheential Whtopicileal examisedples NP-NP 4.4 2-con4st.44ructions do not 4.4require2-4.4 4a sugivenbjec tin NP the, exampreviousples 4.4se9-4ction.51. do. 154 Thenominalse corresponding constituents sentencesbut they invcarryolv ediffer the entsame insi formationtuation contextcontent andby visimirtuelar of theforir di stfferructuralent sema dinticffer enintercespretations. Moreov er,because the (a)the andNP 's(b) chosen sentence as ssu belowbject incall markedthese sentenceon a diff pairserent areNP . diff Coernsientder and the hence following the feature examples of : definiteness is (4.49a) , ora radom-an to nanom nira babah i kia tebom plaoe-to-draw-from water that woman well [Neu] (ACC Nom +PAT J (:�TJ [+dfnt+PAT the well is where that woman draws water from or, the well is that woman 's plaoe to draw water from ] (4.49b) ia tebom ko raradom-an to nanom n ira babahi we[ Nellu] p.���e-to-draw-from waterACC that womanGen [+df+PATnt ] (+PAT 1 (+AGT J the plaoe from whioh that woman draws water is the we ll (4.50a) o ni-pay- ini-an nira babah i to nanom kia lomaq reason-far-bringing that woman water [Neu] (+AGTGen J ( ���TJ +dfnt the fami ly is why that woman gets water or, the family is that woman 's reason fo r getting waterf<[:f:Y ] (4.50b) ia lomaq ko ni-pay- ini-an nira babahi to nanom fa[ Neumily] reason-far-bringing that woman waterACC [:�:T+dfnt ( :�T J (+PAT J the reason why that 1woman gets water is the family (4.51a) o sa-pay-ala J nia wama tia nan i kia toka r mea[Neuns-f] or-getting faGenther oatACC ladder (+AGT J (+PAT J +dfnt the ladder is what the fa ther used to get/resoue the oat or, the ladder is the fa ther 's means fo r ge tting/re[:�:Tsouing ] the oat (4.51b) ia toka r ko sa-pay-ala nia wama tia nan i [Nladdereu] means-for-getting fa(Gethern oatACC :�:T+dfnt l+AGT ) (+PAT ) what the fa ther[ used to get/resoue the oat is the ladder NomiIn thenativese pairs NP isof alwayseiampl deesfinite, three andthingsJ the nominalare worth predicate mentioning., where Firnewst, the theinform choicatione ofis .de providterminersed, can, a [Nbe euei] thnomier naldefinite predicate or inde whichfinite is . a commonDependingnoun on 155 can either be definite or indefinite. If it is an abstract noun, derived from anda ve perb,rsona it lis namealwayss are inde alwaysfinite de andfinite marked. optionally by the Det o. Pronouns Secondly, while nominal predicates marked by [Gen] or [Lcv] are descriptive, ththoisse dis markedtinct byion [N iseu not] can re beflect eiedthe byr de a scridifferptivence eor inid moentifrphoiclogicalational shape. In. Arnis, conRathestrr,uction it is. reCoflecnsidteder in the the fol sulowingbject examchoiceples between: the two NP's in an NP-NP (4.52) ci Ki 1 ang kako Kilang ls [Neu] (Nom+PAT ) I am Kilang (4.53) 0 wawa ni Ki 1 ang kako child Kilang ls [Neu] (Nom+PAT I am Kilang ' s child J (4.54) 0 wawa no Pangcaq kako child Amis ls [Neu] (Nom+PAT ) I am a member of the (Am is) Pangcaq tribe (4.55) tada-mi -adop-ay c i Adop person-who-hunt-by-profession Adop [ Neu] (Nom+PAT ) Adop is a hunter by profession (4.56) misa-lomaq-ay ci Bot ing carpenter BOtin [Neu] (:�:T Boting is a carpenter r (4.57) cima ko kil im-en nomi so? who[Neu ] seNomarch 2sGen (+PAT J (+AGT ) who are you looking fo r? Lit. who is the one sought by you? (4.58) i a misa-lomaq-ay i r i , ci Bot ing carpenterTOP TM Boting[Neu] (+PAT J the carpenter is Boting ia tamdaw i i , 0 ma-apa-ay (4.59) r man TM fo ol (TOP+PAT ) [Neu] the man is a fool 156

(4.60) ia tamdaw i ri , caay ka [0 wama nomako ] man TM not fa ther ls +PAT [+V] [+p] [Neu] the �) man is not my fa ther (4.61a) cira iri, 0 si-kawas-ay no niaroq niam 3s TM witch-doctor place lexcl Top+PAT [Neu] he( is ) the witch-doctor of our place (4.61b) o si-kawas-ay no niaroq niam cira witch-doctor place 3s [Neu] N+PAT� he is the witch-doctor of our pl{ ace ) In Amis, the distinction between description and identification, if it has to beNo r expreis dessedfinit ate nealssl, a isre notliable re flecuected sinc bye botha di fftheer encindee infinit morphoe Detlogi 0 andcal theshape . exdemamonstraples giventive Oet above ia , areI would found ventur in thee seto examsayples that. theOn nomithe nalba sispredi of catethe alone condoesju nctnot ioncarry with an other identif syntayingctic function and/or inmorpho Amislogi. calIf has devic to esbe. doneIn exin amples predicate4.55-4.59 poandsi 4.tion61a which-4.61b siitgn isals the the derived identifi N cationalin eith erfu nctiothe, sunbj ofect the or nominal prediway ofcate contrast. In examin additionples 4. 58to-4 the.59 deverbaland 4.61a [N theeu ] topicnominal further predicate ident thatifies by traalreansldyation iden oftifi theesse. examBothples the sestrrveuctural to illu destrscriateption this and point the. English Thconirstraintdly, though is not sentences violated 4. because49a-4.5 one1b showPAT twois thePAT suconbjecsttit ofuen thets, NP-NPthe l/Sent overconstruction from the whi casele theframe other of the PAT sourceis domi V.nated La bystly the, thederived Genitive N which const carriesituent :_t nounfollo wingor a theGenitive derived Agent N can if bethe interpretedsource V be aslongs eit herto the a po erssgativeessor oftrans theitiv derivede diclaffssicult. Lucy inkil they, undethe strstaructuralnding ofamb thiguityese sentenc does notes, possseemibly to createbecause any only , a colincrmitedete sunounbset claof ss,the aldelowrived the nounspossessor, those inter whichpretation have been. derived into the

4.3.7.3 Non-verbal Complement In section 4.2.2, in conjunction with the discussion on prepositions, we have already introduced some complements that can corneafter the prepositions a, Allka , andthree na prep, whichosi tionsare prep canosi intionstroduce that both can verbal introduce and non-verbala sentential com plecommpleentsment. . Whecomplre emencasets ma. rking is concerned, we pay attention only to the non-verbal Otherare not nominal introduced complements by prep ositiare theons po. ssessExamorsples of ofpo NPssess's wited headh pos nousessnsors, but marked th�y 157 asgoing Genitiv to inclue Locusde themhave he beenre. presented before in section 3.2.3.1 and we are not formsIn sec, tionone 4.2.2,is a prep we osihavetion alr thateady introducesmentioned thata sentential, of the threecomple mendifft.erent It a takes bothfound verbalin se ctioandn non-verbal4.2.2. Ifconstruction the embedded asS itsis ancom pleNP-NPment construction. Examples, witcanh bethe intrsubjectoduced NP byunde anyrst oodDet., theIt nomi seemsnal thatpredicate the only bears likely the Neunominal case comformple mentsand is not after the preposition a are attributive in nature. These nominal complements fofolllolowwing a head example N ands: are to be translated as: "the N which is ..." as in the (4.62) wawa a babahi child fema le girl� daughter or, child who is fe ma le (4.63) saqpiq-ay a bal i cool-one wind cool breeze or, cool one which is a breeze (4.64) adada-ay a kamay sore-one hand� arm sore arm or, sore-one which is the arm The following examples show the head-attribute nominal forms used in sentences: (4.65) ira ko saqpiq-ay a bal i sansandeb exist cool-one wind dusk [o+V] (Nom+PAT (LCV+TIM ] there is a cool ] breeze at dusk (4.66) si-nan i kia wawa a babahi nira have-cat child fema le 3s [ +V] Nom+PAT his daughter has( a ] cat (4.67) icoa-an ko adada-ay a kamay? which-one sore-one hand3 arm [ Neu] Nom+PAT which of your arms( is] sore ? In these examples, the expected Det preceding the attributive N (of the nominal predicate of the embedded S) has disappeared after the preposition a, thusrevise creating schema 4.5an cinter as mediatefollows : mon(4str.5d)os ity"Nom ofinal the" Comformplem NaNent: , which forces us to

P�NP

(net) (NF) (NP) 158

NaNcan be derived into a nominal compciund of the formN-a- N , arso mentione,d in section 4.2.2 in conjunction with the various functions of a. The compound 6.2.1noun can) as bein furtherthe fol delowingrived exa intmople a : stative verb of the form 5 i -N (see sectior( (4.68) si-saqpi qay-a-bal i sandandeb have-a-cool-breeze dusk [+v ] LCV+TIM J there is a cool breeze at( dusk The derived verb si-saqpiqay-a-bal i depicts a meteorological phenomenon in a way similar to other si-verbs of a simpler form such as si-kodem cloudy and si-orad rainy . Another a form is the preposition a which bears the Comitative case form. It is followed by an NP marked by the Accusative case form. The preposition a and the accusative determiner to or ci constitute the composite case marker for the Comitative case form. Through a reanalysis of a to , we get in Arnis a new form ato which requires the co-occurrence of another Det in the following NP and is thus treated as a single lexical unit. Since the NP's linked by ato bear identical CF and CR, this new form is a co-ordinating conjunction. ThE! AccusativComitativee case formform,, cani.e. cover, the onlypreposit the ionPAT a andin AGTcon jucasencti onrel withations the on non-pronouthe two linsnked. NPThe's. rest is handled by ato �ith whatever CF and CR shared by Wheimplreiesas thesom ethingpreposi thattion isa, or when wi llintroducing be, i.e., soma non-verbalething that sent existsence incom plenon-pas1:ment (' na time, the preposition implies past time, but I am not sure that there is it preposition na co-occurring with a non-verbal complement. The following exaformmples is better are the analysed only one ass aI canverb findthan inas my a prepdataosi butti, onin. this context, the na (4.69) salopiko na [0 pacakayay] na-ayaw C01'ner past store before [ +V] [ Neu] LCV (+TIMJ there(� ;�cJused to be a store at the C01'ner (4.70) caay ka [ na [,0 paemotay ] ] ci Looh not was fi sherman Looh +v [+p] +v [ Neu] Nom +fint -fint (+PAT +xlr+ngtvy +xlry+past J Looh [definitely ] was not a fi sherman before As for the preposition ka , it always follows the negative verb caay or the prohibitive verb aka . In the former case, it may be followed by either a averbal non-verbal or a non-verbalcomplement cominclplemudeent the. folExamlowplesing : of the preposition ka introducJ_ng (4.71) caay ka [0 wama nomako ] c ira not fa ther ls 3s [ +V] [+P J [ Neu] Nom (+PAT ) he is not my fa ther ------

159

(4.72) o misa-lomaq-ay ci Dihang , caay kaw [ci Adop ] carpenter Dihang not Adop � T [Neu] [ Neu] it (�is iDihang that is the carpenter� not Adop Lit. the carpenter is Dihang� not Adop In example 4.72, the first) English translation can be misleading. The known Adinfopormation. Sinceis the the Nomcarpenterinative, andAccusative the new , inTopformationic, and Neutralis Diha ngcase, contrasted forms of with personal nouns are identical in shape, marked by the Det ci for singular and ca for plural, it is difficult to tell them apart. The fact that ci Adop is a comsubjectplement and helps it he tolps determinethe analy itssis stof atusthe asfirst a [N parteu] prediof thecate sentenc NP inste. eadNote of a that the preposition ka of example 4.71 is phonetically combined with the [kaw ]. following Det 0, yielding the pronunciation The combinatorial form has been reanalysed as one unit kaw when it is applied to personal nouns and pronouns. The following examples show kaw before pronouns whereas example 4.72 shows kaw before a personal name: (4.73) caay kaw nomako kin i not mine this [ +V] [+p] [Gen ] Nom +PAT J this is not mine ( (4.74) caay kaw [ i takoan ] ko nan i n ira no t with-me cat 3s [ +V] [+p] [Lcv] Nom (l+PAT J I don 't have her cat with me Beforethe fol verballowing excomamplementsple: , we find only the ka form and not the kaw form as in (4.75) caay ka-banaq kako to so 1 i nayay a baqket not know ls exact real weight [ +v] �int (Nom (ACC Idonot � know thel+ exactAGT J weight+PAT In view of verb forms (prefixedJ by ma - and pa-, wel consider ka also as a verbal preexamfixple. 4.7Th6 isbelow analysis, accusa is tivadoptede transitiv becausee verbs, as canafter be theseen negative with pi -sverbsolol doin not require the preposition ka which is used in the language with nominal prepcomplementsosition . andWe a verbalcan say pre thatfix the. negative ligature ka has diversified into a caay pi-so kia babah i , nika caay [ka-demec] ko (4.76) [ I 0 I] not peY'l7lit woman� wife but not yield [ +v] ��int Nom+PAT [ +V] ��int babaq inay n ira a [mi -sawad tia wawaan ] man ( 3sJ abandon( J child ( J Nom+PAT (��int (LCV the wife did not permi t it� but the husband+LOC J insisted on aband( oningJ the child J 160

4.4 Locati ve Relator Nouns spatial orientation is expressed by Itthe wasuse mentioned of locative in relatsectioron nouns3.4.3.3 in thatArnis . In this section we shall explore esthepecially syntactic in andterms seman of localtic characteistic featurristicses . of these locative relator nouns, Locativesyntactic re.headslator of nouns locative are aNP set constructions of locative-t toype spe nounscify whichspatial fun�tion orientation as the. An example of such a locative NP is: tepa r no a 1 (4.77) 1 an side road by the side of the road Locative-type nouns like tepar side are used to supplement the very small inventory of prepositions in a language like Arnis, Sora, or Tagalog and to . suchcompe nsaas Chte inesefor theand la Vietnack of mesecase. inflections on nouns in many analytic l.anguages Thompson (1965:200-202) and Clark (1978:53-61) have labelled nouns of this type in1971 Vietn:195-200amese) has"locativ oncee calledrelator th noueirns" counterparts while Star ostain Sora (196 "noun7:169-179 auxi liaandrie s". The"noun ter auxim "lliaocativeries" mayrelat sugor genoustns" that was they subsequ are entlyless than adopte fudll- becausefledged the nou nsterm. Thelex iclacon ssby ofthe locative following relator set of nouns featur (Nr)es : in Arnis can be characterised in the (4.78) :�+pctssnd Nr's are different[ fro�l other location nouns such as Nataoran , lomaq house , lalan road, and locative pronouns like icoa where , itini here and iti la there , attributein that an. Nr Inis th obislig senseatoril, Nry po'sssess are edsynt [ +psacticallysd] and bound requires. Nr a 'spo inssess Arnisive, as in saSorati sf(Sty artheosta sele 19ctional71:195) requirementsand Vietname ofse certain(Clark 1978verbs:5 9)that, are carry necess infaryormati to or.. on ofspati motional orie, andntation verbs. of Thtransportationis is shown in. Chapter 5 with verbs of location, verbs witArnish Nrbi nary's given deictic in the dis listtinct ivein sectifeatureson 3. 4.3indi catingabove can orienta be furthertion in characte space. rised The features used are: [ ±ncls] (enclosure), [ ±ntrr] (interior) , [ ±hrzn] prominent(horizontal and), [su±lpetrlrior] (lofa teralcomple),mentary and [ ±yang bipola] (fr romqual Chities)inese . yang A featurefor the moretree for the Nr's is given in Figure 4.8. -':}'"'''' '1" The localistic features of orientation are mostly self-explanatory. For example, the feature [ +ncls ] refers to an orientation with which an enclosur·e is relevant as a point of reference. With [ +hrzn ] , it is the horizontal plaor nethe. "pAnositiv orientatione" end of markedbipola [r +yangqualities] refers. to the more prominent, superior, 161 Nr +ncls -ntrr +ntrr -ntrr� +ntrr -hrzn� +hrzn - �g +yan g �+ltr l Yl" � -ya ang terong 1 i lalabo middle outskotide inside sasa tepar space spacekodoko dol side below ahove lal qaqayaway backik ol front

Figure 4.8 Semantic Feature Tree of Locative Relator Nouns Thecharacte followingrise the are cltheass Subcategorisationand members of loca Rultivees andrel atorRedundancy nouns Ruin lesAm isneeded. to RR-l [+lctn] -prsn-prnn -drcn-nmtv +PLC+LOC RR-2 +N r +lctn -+ l+pssd J [��:�vl __ :���� [ + -drcn +sorc SR-l [ +N] -+ ±prnn] SR-2 [ -prnn] -+ [±[ prsn] -+ [ SR-3 [-prsn] [ ±lctn] SR-4 [ +lctn] -+ ±pssd] SR-5 (+l+pssdctn -+ (±nclsJ±ntrr J -+ SR-6 (-ncls-ntrr J [±hrzn] [+hrzn] -+ [±ltrl ] SR-7 [ SR-8 ([ -lt-hrznrl]J] -+ [±yang] 162 Whatit is RR-l the resayssult is ofthat sub-categ locationor isanounstion are. ne It ithealsor pronounssays that norlocation personal nouns nouns are; markedPlace caseby there laLocativetion. Thecase contextualform [ -nmtv feature,-drcn ] in, and RR-2 by saeitysher that a Locusa posses of seda precedinglocation noun locative, i.e. determiner, a locative and re alator follo nounwing, NPmust mar alwaysked by co-occurthe Genit withive acase spformecif [ ically-nmtv,+drcn stated,+sorc for ]the. NrNo 's.Inflec tional Redundancy Rules need to be A fully specified feature matrix of an Nr will contain at least the followin9: (4.78a) +N +p+lctnssd -drcn-nmtv (:���)-Det � mtv + � plus whatever locative(=��� �] features [:=:�: it takes to uniquely identify each lexical Nr. Thus, for terong middle --we need two additional features: [ -ncls,+ntrr] while qaqayaway front needs four: [ -ncls,-ntrr,+hrzn,+yang] . CHAPTER 5

CLASSIFICATION OF AMIS VERBS

5.0 Verbal Classification Leximorphocallogical inform ationfeatures on verbsthat diinclustinguishdes those one syntactsubclassic of, semaverbsntic fr, omand anothe r. weAs weget idean ntifincreasinglyy in our ana finerlysis sub�ateg more syntorisaacticaltion ofly verelevrbsant. Ultidistinctivmatelye, lexifeaturcales , contrastsinformation with is meanta minimal to uniquelyset of di idstinctiventify eache featu lexicalres. entryThis bywould way provide of the badetaisisls for as phono the difflogicalerent re shadesalisation of colourand semantic or the interdiffpretationerent degrees. Suchof minute beemotional, in the intens first ityplace are, bestgood grammarleft to ialexins. cograpDiffherserences and semain meaningnticists that who aresh ould wouldnon-di thstinctivereforee bein excthe ludedsense fromthat grammar they doprope not producer. For syntac grammticatical consequ descencesriptio ns, geneit isra lisasufficienttions about for a syntagrammarianctic strto ucture take s onand the distr tasksibutional of capturing patte rns, and of represubcategsentingorisa tionthese, derivationgeneralisations, and redundancyformally and ru exlesplicitl. y in terms of lexical Ever since Fillmore proposed to "clear the stage in order to examine some of semathe waysntic charactein which risticscase concepts of certain can beEng calledlish ve onrbs" to (Fdeillscrmoreibe the196 8b:syntac383)tic, the and descriconceptsption of ofcase predica relationste and and verb case typ fraesme. haveAs Fill beenmore indi hasspensable pointed inout the, The abstract study of predicates allows us, first of all, argumto desentscribe the eachre arepredicate associ aaccordingted with it.to theSuch num bera descri of ption maypredicate be thought-words ofin as natural analogous lang uageto a claccordingassification to the of num ber of(ibid nouns. :3 73)they . require in a syntactically complete expression Toco-occur define provides verb classes us with according a fundam toent theal , casesthough with not which the only the wayverbs of canverbal or must desubcategfine verbsorisa intion te. rms Ramoof thes (1ir973 inherent:110), forsema insntictance featur, findses. itCl necessark (1ary978 to:4 3) Vietnauses bothmese semaverbsntic. andDeGu casezman fr(1ame978 features:166-258) as, in her her bas detaiis ofled classific study ofation Tagalog of

163 164 verbs, has approached the issue of verbal classification from three different peinherrspeentctiv semaes, nticnamely feature, froms. the angles of case fr�mes, verbal inflection, and The following procedure is followed in the present study. First, a primary wouldclassif lookication into is(1) obtained the implica by retionalferring re tolation the betweencase fra meCF featurand CRes, .(2) Ththeen , we ordeinherentr to semantrefineic the featur initiales, andcla ssif(3) ictheation morp. hologiBy crcaloss-cla shapessifs ofying the theverbs prima�' in verbthe initiaclassesl classi with tqefica threetion. types of features mentioned above, we can refine An attempt is made here to show that, to a large extent, verb classes defined bythe casemorpho fralogicalme features form andclasses the ofcorre verbalsponden affixceses between, as we CFll andas a CR cla indeedssification match basedand sema onnti inherentc features sema nticdo converge feature ons. a crosIn others-cla wordsssific, ationcase , ofmorpho verbslogical. ,

5.1 The Rol e of Infl ectional Features Siinflence ctionDeGuzm andan derivation(1978:129-13 within5) has the taken lexi greatcase painsmodel, toI wouldclari fybegin the withnotions a brie of f notexamination making "aof cleaherr-cut ideas distinc. DeGuzmantion betweencriticises infle previousctional worksand derivational on Tagalog for becacateguseories, in ordo proceing sosses", she (icanbid. st :1rip29) off. whatShe findsshe identit useifiesful toas inflectseparateional the tHO heraffixes des criptionand deal of with verbal the remaindesubcategorir, whichsation sheand calls derivation the prim. aryThe verbcrit erstemia , shein paradiguses tom disting for a uissingleh infle lexicalction entry and derivation, (b) predicta involvbileity (a of) theform exi andstence meani ofng, a andof case(c) changeframe. inActual semanticly, contentcriterion and/or (b) folsynlowstact icfrom featur criteesrion, parti (a)cularly . that It is important to note here that DeGuzman's criterion (c) frees derivation withinfrom the the con classfines ofof ve therbs traditional. By crite Partsrion (c)of Speecha change and in alcaselows fraforme derivationsfeatures changesuch as inthe verbal addi tionsubcate, degoryletion and, oris rehenceinter pretationa derivation of . case relations means a In the present study we have proposed to incorporate CF-CR mappings as extended caseCF-CR frmapameping featurs inclues. de Thsubjeis hasct choiceserious. theorIf aet changeical iminplic extendedations becauscase fre amethe prefeaturesent s lexalsicaseo means mod el,a change voic e inis verbalnecessa subcaterily a goryderivational, it follo wsfeature that., within"Voic e"the notinv olvinflees bothctional a changeby DeGuzman in sema'sntic third content crite rionand . syntactic features, hence it i!; Asdeveloped the second enough crit forer ionDeGuzm goesan, perhato clpsaim thefu llTagalog predicta focusbil itsysty ofem the is werelall­ted producforms, tivyete oneand shouldalso fully bear predictable in mind that in there terms areof derivationsmeaning and thatform are(cf. 100% conIkranagaracerned, the198 0:Ami22s, data51; Ar alsoonoff su ppo1976rt: 35-a 45)derivational . Where analysfull prisedi. ctaEvenbilit thoughy is 165 there are what one may call "instrumental focus" and "locus focus" ableconstr (suctee ionssections in the 3.3.2.1 language and, 3.4.2their .2)occurrence. is very limited and unpredict­ Where aspect is concerned, Amis expresses what would notionally correspond to .DeGuzman's inflectional aspect features: [ ±fin ] (finite), [ ±beg] (begun), [ ±comp ] (completed), and [ ±imp ] (imperative) with a variety of syntactic deviin Amcesis . is notFor insreflectanceted, bythe adiff differerenceence betweenin morp hologicalfinite and mar non-king;finit rathee verbsr, it isare thea fewabsenc exame plesof a: nominative actant th�t signals a non-finite verb. Here (5.1a) si-tokar kia wama a [mi- a 1 a tia nani use-ladder fa ther rescue cat (ACCT :�int J (���TJ ��int l+PA kodokodo l no k i 1 ang ] J ( top� above tree ( J Lev+LOC the fa ther uses a ladder to rescue the cat from the tree wama (5.1b) na ( [] 5 i -toka r kia a [mi-ala did use-ladder fa ther rescue int int ( int +past:� (:� �� tia[ nanl i Jkodokodo l no kilang]] ] cat top� above tree (���T) (���C) the father used a ladder to rescue the cat from the tree (5.1c) ma-ala kia nan i nia wama a [si -tokar] rescue cat fa ther use-ladder (���TJ :�int ��int kodokodo l no kilang (��) ( top�J above tree ( ) (���C] the fa ther rescued the cat with a ladder from the tree Lit. the cat rescued by the father who used a ladder from the tree In example 5.1a, si-takor use-a-ladder is a finite verb while in 5.1b and 5.1c si-takor is non-finite. Likewise, in example 5.2a below, mi-parokod kick is finite while in example 5.2b mi-parokod is non-finite. mi-parokod ira to panan (5.2a) c kick 3s door int (���T] ACC+PAT he: �kicks at the door ( J ( ) 166

(5.2b) ma-calibad cira a [mi -parokodJ to panan angry 3s kick doACCor �intJ ��int +PAT J he(: is kicking(:�� at Jthe door( angri) ly ( Therela tionfinite an d,and inno n-fAmiinits at e leadissttincti, a dionffer signencalse ina dicasefferenc framee featurein domi nancewhere .the Nominativeconstructio nsCF withis concer tensened ver. bsExcept, all seforntences quotativ in eAm andis, compoundsimple ·or sentencescomplex, andare astakesumed finite to have com plementsonly one thatfinite are verbident. ifAsiable wi llby be the shown pres beloencew, of tense a Nominative verbs NPcom. plementsNon-f initwhiche verbs do not occur have in su bjeccomplext NP sentences' s. As suasch, heads they ofneve sentencer prec ede thE! mainredundancy verb. rulThesis: co-occurrence restriction can be expressed by the followinq (5.3) [ +vJ + ([+vJ) -=[+fint] J There are subclasses of verbs that (take an obligatory sentence complement. . Ifis the headembedded of thesentence construction is a verbal can thenconstruction be referred, the to non-fin for theite cla verbssif whicationich Negof ationthe main Verb ve. rb, which may be a Manner Verb, an Auxiliary Verb, or a Thus, all verbs can initially be classified into two categories, namely, tho�;e dothatn't have. Am toong take those sentenc� that �o com notple mentstake sentwithence non com-finiteplements verbs wit andh no thn-finiteose that heads are the main verbs of simple sentences! as well as the Tense verbs: a for arefuture fo llotensewed byor irfiniterealis verbs and thatna for are past accom tensepanied or byrea alis No. minativSincee NPa and, the nair existence violates the co-occurrence restriction expressed in rule 5.3 that it veinstearb cannotd of adverbs be follo orwed pre byfixe anothers, we wouldfinite haveverb to. explicitlyIf we recog marknise them them as as verbs exceptioin Englisnsh., as we must with grammatically analogous verbs such as know or say After a predicate indicating time, place, or instrument, a non-finite action vetherb nousuallyn-finit efol comlowsplement. In offa anotct, hermost ve verbsrb. canIn parothertici wordspate , rathermost verbs freel havey as a obtainedfinite and with a no then- finfollowingite members subcateghip. oriThesesation two rule subclasses: can be initially (5.4) [ +vJ [±_[ -fint] J Ofsm allthe groupsubc lassof tenseof verbs verbs that that do takenot takefinit noe n-comfiniteplements comp. lemSeeents exam, thereples 3.8is the0,. 5.1foll2aowing and subcategori5.12b belowsation. Th isru furtherle: subcategorisation can be stated by the (5.5) [-_[ -findJ+ [ ±_[ +fint J J with verbs that require a sentence complement, the verbal sisterhead can be featurecalled upons [±_[ to -fcharacteintJJ andrise [± the [+ mainfint JJverb supplement. We can case say frthatame thefeatures contextual in the subcategorisation of verbs. -The [+_[ -fint J J subclass incluQes Manner verbs, 167 aInstr umentalna, verbs , Aspectual verbs, andRR Ne gation verbs. The two Tense verbs, __and which are exceptions to the 5.3, can be characterised as [+Cha pter[+fint 3) ]].wit h Theverbal following complements are sentence: examples (the first four from

ma-seng ib c ira a [mi-l ias] (3.142) sad leave 3sNom [+p] [:�int] (+PAT ] ( ��int] he+rnn wentnr away sadly si-losa c ira a [mi-si-olah to wawa n ira] (3.145) with-tears plead child 3s ) 3s +V r Nom [+p] +v (ACC [+fint ] l+PAT J (-fintJ l+PAT ] in+rnn tearsnr she pleaded with her child si-bakic kia babahi a [mi-radom to nanom tebom ] (3.79) with-a-bucket woman draw-from water we ll LCV the(:�i womannt] used a bucket(���T to] draw water(��int] from the (we ���T]ll (+LOC ) na [ s i -toka r kia wama a [mi-ala tia nan i ] ] (3.80) did with-a-ladder fa ther rescue get cat Nom [+p] +v 1 Ace the( :�int) fa ther (: used�int) a ladder to rescue(+PAT ) the cat(-f int (+PAT ] ma-banaq to haw kiso a [mi -bot ing]? (5.6) know alrea QM catch- Sh +AdV 2sNom [+p] +v do(:�int) you know(+sptl how1 to fish?(+PAT ) (-fintr ma-ngalay haw kiso a [mi-icep ]? (5.7) like QM chew-betel-nut 2sNom [+p ] do(:�int) you like to chew(+PAT bete ) l-nuts ?( ��int ) sa-terep sa han a [orad-an ] ko remiad (5.8) stop just already rain weather +AdV +AdV [+pj Nom (:�int ) [+sptl 1 [+sptl 1 (��int) (+PAT ) it has just -drtvstopped +prft raining (5.9 caay kako ka [ nga 1 ay a [tara]] ) do-not Is want, like go +v 1 (Nom ] [+p ] [+fintj l+PAT ( :�gtv) (��int) (��int] I +ngtvdon 't want to go 168

' (5.10) aka ka [ t em ang i c ] don+v 't cry+V- - ) +fint :�gtv -fintJ +mprt+ngtv [ ' ( 1 ( don 't cry ! (5.11) caay ka [01ah ] kako a [ tahka 1 to 1 ab i 1 ] do-not like ls (JP-out night int Nom +P ACC+TIM :�+nintgtv :�gtv �� ] +PAT ��int ] I don 't like to go out at night Note that except( whenI it (comes ] after( a negation( ] verb, a (non-finit] e verb( is always introduced by the preposition a or na. After a negation verb caay or aka, the preposition ka is used. Depending on the verb, the non-finite form after ka may be just the root form, stripped of all affixation, as olah and nga lay in examples 5.11 and 5.9. Or, it can be an affixed form such as requiredt-em-ang icand in, exasam hasple been 5.10. mentioned After aabo tenseve, theverb verb na orfol a,lowing no prep a tenseositi onver bis is finite, such as si-tokar in example 3.80. Perhaps historically there was once vowa prepel. osition a, now assimilated to the previous word that ends in the same Most Amis verbs can appear as the head of a sentence complement. Even the [ + [ -fint ]] verbs themselves can in turn be the head of a sentence complemen1:. Example 3.80 above shows an instrumental verb si-toka r governed by the tense verb na while it in turn takes mi-ala as its own verbal complement. Example 5.9 shows an affective verb ngal ay after the negation 'verb caay, which in turn thetakes su bjecttara as NP itscoming complem afterent the. Exseampcondle V 5.1instea1 showsd of a thesi mifirstlar construction. The followin9 with paiversr a,of with exam plesa di ffershowence a tensein mea verbning. com plementing a negation verb and vice (5.12a) na [ caay ka [0 pacemotay ] ci Looh ] (M76.1) was not fisherman Looh +v 1 +v Nom Hint +fint :�gtvj ::uj +PAT j [+tnse+xlry ] +ngtv+xlry ( ( ( not[ a fisherman Looh was (5.12b) caay ka [ na [0 pacemotay ] ci Looh ] (M76. 2) not was fisherman Looh • +v +v Nom +fint gtv -fint ::u +PAT +xl+ngtvry :� +tnse+xlry Looh [defi(nitely) ] was[ not a fishe( rman] before ( ] In most cases, a non-finite verb form is homophonous with its finite tocounter be mempabersrt. ofWe a corensgularider paradigthe finitm becausee [ +fint both] and thethe verbnon- finiteforms af[ te-frint the] formE: positive ligature a and those after the negative ligature ka are predictable from the source. Even though an aspectual verb such as sa-te rep in example 169 predi5.12 ctisable neve byr foundits morpho to be logicalused as shaa penon- andfinite meani verbng. , theThese non-occurrence exceptions canis be indicatin termsed of in itsthe fu lexill produccon andtivit latery and handled predi bycta bilredundancyity, we rulcan essa. y thatIn generalthe , verbsfinite/non in Am-finitis. e dichotomy is an inflectional feature that covers all the Asp(comectpleted) features are marcorrekedspon on dingadve torbs DeGuzmaninstead 'sof [ verbs±beg] in(begun Amis.) andThe [ ±c poompst] -verbal partinot shiftcles theare stanalysedress on asthe adve ultrbsimate instea syllad bleof verbalof the sufverfixesbs. becauseI have chosenthey do to characterise Amis aspectual [ +sptl ] adverbs with the following features; ASPECTUAL [ +sptl ] - This positively marked feature is used to othedistingr adveruishbs the, sasety, ofthe aspectualmanner adver advebsrbs. from INCHOATIVE [ ±inch ] - The positively marked feature means that the beeventgun toor takeaction pl indicatedace. It bycorresp the verondsb hasto DeGuzman's aspect feature [ ±beg] . PERFECTIVE [ ±prft ] - The positively marked feature means that the eventcomplet ored action. It indicacorrespondsted by tothe DeGuzm verban has's been aspect feature [ ±comp ] . DURATIVE [ ±drtv] - Theevent po orsit iveactionly markedindicated feature by themeans ve rbthat lasts the markfor eda duration feature meansin time the, wh actionile the is ne punctuagativelyl. PROGRESSIVE - This is a cover label for three features; [or+inc actionh,-pr indift,+drtvcated] . by theIt means verb thatis on-goi the ng.event PRESENT [ +pres ] - This is a tense feature that is sometimes attached to an aspect feature. An example is prograniniessi now vewhich. Th isis marked feature as is present needed to distinguish an ini from to which is also marked [ +inch,-prft,+drtv ] for progressive but unmarked for the feature [ ±pres ] . In the following are a few examples with aspectual adverbs. Examples 5.13 and 5.14 show the difference between an ini and to. In examples 5.15 and 5.16 we havespecifi sequencescation ofof theadve memberbs whors. se meaning is the sum total of the feature (5.13) mi-bacaq an ini cira c i r is wash-clothes now 3s river [ +V] +Adv Nom LCV +sptl (+PAT ) (+PLC ) +inc-prfth +drtv+pres she is washing clothes by the river 170

(5.14) ma-korokoro tOI kia bekeloq (a) [ tara sasa ro ll stone (lo-toward bottom Nom r+v � (Lcv int +Adv+spt l +PAT J -fint +LOC J ( :� J +inch ( l J l +drtv-prft no 1 otok J hill the stone is/was ro lling downhill Lit. the stone is/was ro lling to go toward the bottom of the hill (5.15) sa-terep sa han ko orad stop just a!read rain [ +vJ +AdV+sptl +AdV+sptl Nom+PAT -drtv +prft! ( ) the rain has[ just st[opped (5.16) kaen sa hanI sapaqyo ci kaka ako, eat J· alr'ead¥ medicine older-sib ling ls [ J [ ( +v +sptlv +AdV+sptl +Det J ACC+PAT T -drtv +prft l ) (:�� ) i ri [���kemat[ sa t03 then sleepI justJ ·alread¥ [ +Conj J [ +vJ l l ::-drtv:� :+lnch;:� my elder brother [has justl taken[ theJ medicine and gone to sleep Notein ex amthatple there5.16 areis a threedeter minerdifferent in the to accusativforms in ethese case examformpl es(cf.. FigFirsture , 4.1t02 in section 4.2.1) . The other two forms, tOI and t03 , seen respectively in examples 5.14 and 5.16, are aspectual adverbs. tOI is a progressive aspect marker while t03 marks the beginning of a state or action. With t03 , the feature [ -drtvJ of the sa additive is neutralised so that the combined form sa to may mean that a certain state of being or action has just begun and is astill certain going action on. hasThe beencomb cominedplet formed. sa han in examples 5.15 and 5.16 means that Astive for verb impe formsrative isver quitebs, siregnceular the and corre the spondencemeaning alwaysbetween pr edictathem andble non-im, theypera are ­ prqualiesentfied st udyto be, howevtreateder, asI haveinfle chosenctional to by handle DeGu zmanthese's recritlatederion verb (a) forms. In with the dersomeivation derivational in acco procerdancesses with may her als crito enjoeriony full(c) , predictawith thebil undeity rstandingof form and that meansyntacingtic (c ritfeatureriones , (b)particularly) while ef fectingthat of changes the case in frasemame ntic(crit contenterion (cand/or)). The6.3.2.5 derivation) . of imperative verbs will be presented in Chapter 6 (section In summary, both types of DeGuzman's inflectional features - voice and Phaspectualilippine-t -ype are notlanguage viables cantoo lsspeak for ofthe vo anaice lysparadigis of msAm isand. focus Whil esyst em for the 171 veandrbs com, plemAmis entationresorts , toto di hiffgherligentht syntactic certain casdeve-licesike, notionsspecifically. We nomhaveinal alsoisation shown(completed) that DedoGuzman not have's inflect ready ionalcounterparts aspect featurein Amis [±. beg]Rat he(ber,gun aspe) andct [±iscomp ] manifested by aspectual verbs such as sa-terep in examples 5.8 and 5.15, which takeverbs verbalare also com plehanmdledents, by and derivation by the usein Amiof saspectual because , adveras willbs. be Imseenperative in secformstion invo 5.2.5.1lves changes and secti inon synt 6.3.2.5,actic features their corre and spondecategorynce . with non-imperative

5.2 Primary Verb Classes - Subcategorisation in terms of Case Frame Features The approach of determining verb classes on the basis of co-occurring CR's and CFgrammar's has of been Japane adopsete d(1 in971) lexica, Li sein hissince grammar its inceptionof Rukai. (1973)Taylor, Kullin hisavan caseijaya inDeGuzman her study in her of Thaianaly verbssis of (1Tagalog974), Clark verbs in(1 97her8) standudy Fagan of Vietnamese in his des (19cription78) , deteof Monormine-Alu ve ofrb theclasses Solo monand Islaconstructionnds (1979) typ alles . utilise case frame features to provideFigure 2.1a br1 oadin sectsubcategion 2.oris4.4ation exemplifies of verbs how. caseIn ourframe lex ifeaturescase mode canl, webe usedhave to [+[+extendedX]], wherecase frXame repre featuressents a fromCR, to(1) indithosecate contextual the co-occurrence features of thecase form [+rel[+ationX],+[+ S Y]],with where the verbX and to Y incluare bothde (2) CR those's, to contextualindicate thefeatures co-occurrence of the formof X and Y with the verb and (3) contextual features of the form [+[+X],-[+y]] to indithosecate contextual non-occurrence features of of Y thein theform ca [-se[z frame,-Y]] of where the veZ rbis wita CFh X,and and Y, a(4) CR, to repreclasssent. Not thee mappthe conventioning of CF andthat CR a which CF is isnot characte marked risticby a "p oflu as" particularor "minus " verb becausebundle of, asfeatur has beenes, but shown not infeatures section the4.1ms, elcaseves . forms For areins tancecover , lathebe lssy mbforol a [N(seeom] Figstureands 4.for6 in[+ nmtvsection,-prdc 4.3,-)cn . tr]How andever [G, enli] kestands othe r forcover [-nmtv labe,+lsdrcn such,+ sorcas ] [±-, trnsespe] ciall(transiy intiv doube) andle neg [±erationgv] notatio(ergativnse). , CF When la beappels canaring also unmar be kedmarked, they + orare assudescmeribedd to inbe sepoctionsitive 2.. 4.4.The double negation convention has already been Classification of Amis verbs by case frame features can be shown by the tree diagindicatingram in theFigure oc currence5.1a and or 5.1 non-occurrenceb. At each node of a isparticul a contextualar CR infeature the case frameclasses of will the beve rbsgiven. Adunderditional the recasespectiv framee subsefeaturesctions for be lowindividual. verb Figsubjureectless 5.1b sentshowsence twos. categClassories VI Iof isimpe uniqrsonalue in ve therbs sense. Both that occur it req in uires the co-occurrenceAgent for ergative of a Geverbsnitiv ore aac potantssess. orSi nceof a a nomgenitinaliv eform NP , sigI nalshave eichosenther anto treclassat whichrakat- ancan wa beLk des as criptivelya V. My lachoicebelled does as imperesultrsonal in theer gativecreatio tran nsiof ative verb makeverb . greThateris maygener bealis quiteations a terminowith thelogical features mon strofos trityansit, butivit ity alandlows er gativityme to . 172 [ +V]

J Figure J S.lb -[ +AGT] +[ +AGT] -[ +LOC]� +[ +LOC ] -[+�LOCJ -[ +INS�] +[ +IN S] I II IIII IVI V

e.g. , ma-orip ta-ngasa taes-en tomes-en pabe l i aliveI arriveI hi.t I fiZZI giveI

adada ma-olah mi -k i 1 im mi-pal ita hurt love seek ask

Figure 5.la Classification of Amis Verbs in terms of Case Frame Features; with examples

(Figure S. 1al - - . - (�[Nom]J -[ +AGT]� +[ +AGT ] VII VIII

e.g. , siqnaw rakat-an ooldI walkI

orad-an rain

Figure 5.lb Classification of Ami s Impersonal Verbs in terms of Case Frame Features; with examples 173 Talabeylorls for(1971 the:1 74)characte uses whatristic he case calls frames "pseu dofor-f etheature primarys" as vederbscr claiptivesses cover. His pra(1978ctic:4e4) hasand been DeGu zmanfollo wed(1978 by:1 other73) in lexicasetheir de grammariansscription of such verbal as Clclassark es. whTheseile , desatcript the ivsamee latimebels, briallowdging for theeasier gap betreferenceween a tofeature the featureanalysis matri andces the moreto groups tradi oftional verb liclassesnguistic by taxonomythe name. of I thehaveir commonfollowed case DeGuzman relation in . referringFor examcanplebe re, siferrednce classes to as a IIIgrou, pIV, as V,agentive and VI I versharebs. theCorrespond AGT case inglyrelat, ionthose, they withoutThis latter an Agent group ininc thelude cases verbs frame from can clabe ssreferes I,red II, to and as VInon-agentive. verbs. In the following are listed my choice of cover labels and their corresponding Ficasegure frames 5.1b: for the seven primary verb classes presented in Figure 5.1a and I Simple non-agentive [ +[ +PAT],-[ +AGT] , -[ +LOC]] II Intransitive locative [ +[ +PAT],-[ +AGT] , +[ +LOC]] III Simple transitive [+[ +PAT],+[ +AGT],-[ +LOC],-[ +INS]] IV Transitive instrumental [+[ +PAT] ,+[ +AGT ],-[ +LOC] ,+[ +INS]] V Transitive locative [+[ +PAT] ,+[ +AGT] ,+[ +LOC]] VI Impersonal intransitive [ -[ Nom] , -[ + AGT ] ] VII Impersonal transitive [-[ Nom] ,+[ +AGT] J Each of these classes can be further characterised and subcategorised in terms ofintrodu semanticedc. andFor morp inshologicaltance, with featur an esadditional. Then othersemantic conven featureient , lasabey,ls [+ maylc tn]be spforeci "lfic-ocatioobjecn"t, weintra cannsi separatetive verbs the andloc ationthe psyc andho logicallocomotion verbs verbs that from also the ha ve [+"p[sychologi+LOC]] incal" the ircan casesort framoutes the. psychoFurthermlogicalore, verbsthe feature from the[+ pscsph]ecif foric-o bject verbsintransitiv in a esformal. Thewayse, becauseinherent theysema nticalso bearfeatures morp hologialso secalrve andto susyntbclaacticssif y ofconsequence the verbss. will Classi be preficationsented ofin verbssection in 5.3. terms of inherent semantic features Verb"impe classesrsonal" VIif weand are VI I payingcan be attentionreferred to theas bein non-occurrenceg "subjectl ess"of theor Nominative CFframe. . It Thefollows sequence that ofthe reasoningfundamental can casebe outlined relation asPAT fol islows absent: in the case 1. These verbs are ergative. 2.3. ErThesegativ verbse verbs do nottake requireonly PAT the sU bjco-occurrenceects. of the 4. NoThereminativforee, casethere form.is no PAT in their case frame. The traditional notions of intransitive, transitive, and ditransitive verbs verbsare clo aresely characte tied inrised to our by theunde prerstandingsence of of the ca seAGT fra caseme featurrelationes. . TraThereforensitive , classes III, IV, V, and VII can be collectively referred to as transitive / reverferredbs. Correto as spondintrainglynsitiv, etho versebs wi. thoutTh isan laAGTtter in groupthe caseincludes frame verbscan be fr om 174 classes I, II, and VI. The notion of ditransitive verbs implies the presence oflocativ a "deir .verbsect ob areject" the and only an candidates"indirect ob forjec t"the. ditraIn Arninsitivs, eclass verb V categortransiy.tive In this study, I have extended the set of case frame features topatterns'include ofth o�eCE" s contextual features which reprrb esorent verb the clacharactess. Theseristic contextual mapping features eiandthe CR'r takes for ona partthe iculaform [+[+r veX],+[+Y]] where X and Y are both CR's, such as the form [-[ Z,-y ]], where Z is a CF and Y, a CR. AsPatient has been and Agdesentcribed, or takein sec ontion 2.4.4 above, the double negation notation means that if the CF Z is present, it is necessarily with the CR Y. Specifically, when Z equals the Nominative case form, this notationnative, thatwould is, imply chothatsen onl asy thesubjec CRt. indicated It is bypossible Y can beto respealisedcify mo inre thethan Nomi one such contextual feature so that the mapping of CF and CR in two or moreof nom certaininal actants verbs or can verb be clajointlysses . referred to for the explicit characterisation These extended case frame features formally characterise the notions of oftra nsithe tivitco-occurrencey and ergativity of two . CR' Tras, nsitPAT andivity, AGT in, and th isper studyhaps , PATis anddefined INS . in ThE!telns definispecificallytion of, erthegativity association is rel ofated the toNom theinativ associae CF tionwith between the Patient CF's CRand inCR' s, traInsnsittrumentive . verbsThe, andco-occurrence the associa retionstrict ofion the betweenGenitive the CF Nomi withnativ the e AgentCF and or all aav Paailatientble CRsubject.'s is also known as "subject choice". Ergative verbs always have If(tra wensitiv redee)fine and the the case CF-CR feature associa assoctioniat [-ion[N om[+,[+PAT-PAT]]], +[in+AGT the ]]pre assence [+tr nsof )a perspectiveGenitive Agent of suas bjec[+ert gv]choi (ecergativ, splittinge), we thecan subcategotransitiverise ve rbverbs classes from IIIthe, IV,. Figureand Ve ach5.2. into non-ergative (accusative) and ergative subclasses as shown in The(Tay featurelor 1971 [±:174)trns be] andcause [± ertheygv] areare juwhatst convTaylorenient would cover call la be"pseuls dothat-fea stturandes" for acontextual collection feature of featur [+[ es+AGT. ]]In wh thileis [- sttrudyns], thecovers feature the non-agenti[+trns] im veplies verbs the. ��he [-feature[Gen, -AG[+T]],ergv ]wh repreile [-sentsergv the] covers implica thotionalse tran resiltiveations verbs [-[ Nomthat,-PAT do ]]not and al low Paand/ortient [- su[bject.Gen,-AG T]],If wethen extend all intr the anmeaningsitive verbsof [+ erwouldgv] to be meanergative [-[Nom and,-PA bothT]] traimpensirsonaltive becauseverb classes of the VI presenceand VII wouldof [+[+AG alsoT]], be erandgative class. VIClass is intran VII issit ive. Arnisis a mixed ergative or a split ergative language in the sense that the nolan-ergnguageative has traor nsitaccusativeive verbs type be. longingFor exam to plethe , ertheregative are type er gatas ivwee llverbs as theli ke methel aw-enco-occurring watch and PAT ma is-mela alwaysw see inwho these Noco-occurringminative. AGTAt theis insame the timeGenitive, the wh:.le lalangnguuageage s.has verbsThat isthat to sayobey, Fillmthere oreare's verbs subjec thatt choice would hie alwaysrarchy take for AGTaccusa subjEt'.v!cet ifINS anis AGT pr esentis pre. sentIf, neotheritherwise AGT theynor wouldINS is takepres entINS , subjecthen, t PATprovided gets tothat be anthe 175

-[ +LOC] +[ +LOC]

-[ +INS] +[ +INS] -ergv +ergv � I -e gv +e gv -e gv +e gv � � � I IlIa IIIb IVa IVb Va Vb r r r rI I mi-kiI l im taes-enI mi-pinarI o tomes-en pabeI 1 i pa-pi-angaI ng-en seek hit fi U fi U-up give have X caU Y ? mi-banaq ma-ca 1 iw sa-pi-angang pa-tara pa-pi-ngiciq-en notify borrow use-fo r-caUing send Zet X share Y ? mi-pal ita ask

mi-cal iw borrow

pa-sebanaq-i te U

pa-totod i-i pass

Figure 5.2 Cross-cl assification of Transitive Verbs with the Feature of Ergativity; with examples intrasubject.nsitiv Bute and in thethe notionlatter of case er gativitywhen no AGTis nonorn-dis INStinctiv is presente; that, theis , veforrb bothis eralwaysgative PAT and. accusative languages, the subject of intransitive sentences is The Amis language seems especially sensitive about crowding of its case frames, posimpssilebly sentence because, theof themaximum very numlimitedber of inv co-occurringentory of distinct actants caseis thr formsee. . VerbIn a anotherclasses verbwhich classalready by haveadding three another CR's CR in toth eirtheir case case frames fram escannot. For derive example into, bycausati the vissyntationactic ofmeans the traof nsitsentenceive locativecomplementation verbs in ratheverb r classthan byV ismorp thushologi donecal meansframe whichof class involv V ises alreaverbaldy affisaturatedxation wi andth thethree addition CR's, name of aly CR, PAT, si, nceAGT the, and case verbsLOC. inThe class same IV,applie whoses tocharacte the causativisristic caseation fra ofme theis traalsons satitivuratede instr withumental three 176

CR'ands. the Anabsence explanation of COR infor thethis case beha inventorviour foy llofows Arni froms, thoughthe l/Sent that stillconstraint does notin Am exisplai. n the absence of a class of [+[+PAT],+[+AGT],+[+LOC],+[+INS]] verb!; Agentive verbs normally implies the co-occurrence of an optional INS. But in notionalthe case ofinstr veumentrb cla sshas V,to therebe· inc isorporated no room infor a theve rboptional form which expa nsholdsion , a so the relcompleationshipmentising goe relas int ionthe withother the dire agentivection: theve rbhig. herPos verbssibly are the avaicauslableal to anexpre INSss on the the notion verb . of Heinstrre umis enta typi, so calit exampleneed not: be expressed (aniliiguously) as (5.17) si-bakic kia babah i a [mi-radom to nanom tebom ] use-bucket woman draw-from water well [+V] (:�:TJ (��int] �� TJ (���cJ the woman draws water from the well with( a �bucket The incorporated notional instrumentin this exampleis indefinite. If the notionalemployed . insContrumsientder has the a foldelowingfinite exreampferencele: , a different syntactic device i!; (5.18) na o maan saw ko sa-pi-bohat nom i so tia panan? did(+V what[ Neu] QM m S-fo r-opening 2s doorACC l+past J [;::+PATj (:�T] (+PAT J with what did you open the door? Lit. what was your means fo r opening the door? na itini-an a cokcok ko sa-pi -bohat (5.19) 0 one key m s-f r-Opening l a't [:!�J [;::+PAT j � (ak:�o tia pan an lsGen doACCor ) (+AGT J (+PATJ I opened the door with this key Examare plesexpressed 5.18 byand NP-NP 5.19 constructionsshow that notional with ains nomtruminalisedents wit formh de whichfinite imrepliesferenb; the notionsubject of NP instr. ument, as in sa-pi -bohat means-for-opening, acting as the Both the complementation and the nominalisation devices are syntactic rather Whthanile morpthe hologicalsituation mechanisms expressed byfor theadding sentence case-li mayke inclunotionsde motore a thansent encethree. ca!;e­ liinke the notio casens fra, onmely of a maximuma single of head three ve rb.may beThe expr remainderessed as of overt case- caselike relanotionstions wouthe ldRami havefica totion be expresectionsssed of by eachother case syntactic relation dev, icesI have. givenIn Cha manypter exam 3, plesunder J.n Arnisoutershowing case re thatlatio nsmany, succahse-li as plke acenotions, time, , esmanpeciallyner, and those refe associarent (orted bene withfacti.ve) the 177 find their expressions as predicates of verbal constructions with an obligatory nounssentence are com derivedplement forms, or thatas pre imdicaplytes the of case NP-NP-like constructions notions concer inned whic. h the head potentAmis verbsial ofwi takth twoing oron lessan optional CR's, i.e.CR such, with as unsPLCaturated or TIM. caseTh isfram addedes, have CR thenthe becomesV (transitiv parte oflocative the case) verbsframe wou. ldClass have IVto (tberans expliciitive tlyins markedtrumental [-[) +PLCand ]]cl andass [-opt[+ionalTIM]] oute in orderr CR's. to blocIn kaddition the appli to cationthis, classof the IVRR verbs which would introduce have s toth bees e n�gativeexplicitlyly marked [-CR[ may+LOC not]] andbe classintroduced V verbs to marthekedir alr [-[ea + dyINS sa]]turated �o that casethe framfeaturese. isOne the syntactic necessity consequence of resorting of to this a verbalnegative comlyplementation marked contextual strate gy if, wordssay, a, notionalthe negative inst markrumenting is forces to be the expre verbssed to with use thea class complementation V verb. In devi otherce . Thframeis chprovideoice betweens another ver balmeans com toplem crosentations-classif andy Amian .addis vetiorbsnal. CR to the case Generally speaking, an outer CR like PLC· or TIM is not used for the primary doesubcategs provideorisa tiseoncondary of ver, bsif . notThe redundant presence in orformation absence forof ansubcateg outer orisCR, ationhoweve. r, andFor timeinstance as pre, whdicailetes agentive of NP-NP verbs constructions use the ·syntacti augmentedc devices by sentence that expre compless mplentsace , expreimpersonalss the verbsnotions do notof locationas a rule in utspaceilize and verbal time com. plem Impeentationrsonal verbs device expre to ss placTIM erespective and time ly.within the case system, using the outer case relation PLC and The primary verb classes and their immediate subclasses as shown in Figure S.lrulesa, : S.lb, and S.2 can be arrived at by the following set of subcategorisation SR-l [ +V] + (±f±_[int -f int] ] SR-2 [-_[ +fint]] + (±_(�:u)) SR-3 (:�int) + [ ±[Nom]] SR-4 (:�int) (±±[[ +PAT+AGT] ) SR-S +PAT]) + [±[ +LOC]] SR-6 (:[' + [±[ +INS]] -[ +LOC+PAT ] SR-7 [:(+AGT l])] + [±ergv] These SR's are to(:[' be preceded by the following redundancy rules: 178 RR-l +V+[ +PAT ] [-[ +INS]] +[ +LOC+AGT]] RR-2 + [+trtls] (�+[[ +I+AGTNS ] ] RR-3 [+ergv ] Nom l 1 -(-PAT1 Gen -(-AGT 1 RR-4 +ergv [-[ +PAT]] (-[Nom] ) SR-l gives us the sets of finite and non-finite verbs, each with or without verbalcomplem entscomplem. entSR-3s. takesSR-2 thedist siingmpuisle hesfinit thee verbsverbs andwit hsubcateg or withoutorises nomi themnal verbsaccording are subjto theectle presssence or impeor absencersonal verof bsthe. NomiSR-4nativ takese casethe siformmple. finit-[Nome ] verbs and±[+AG subcateTJ. Verbsgorises characte them accordingrised by tothe thecase contextual frame of [+features[+PAT],- ±[[++PATAGT]]] and ofconstit [+[+PATute ],the+[+AG non-agentiveT]] constit verbute theclasses agentive I 'a ndve rbII- . classesVerbs IIIwith, IVthe, andcase V. frame Verbs with [-[ Nom] ,-[ +AGT]] or redundantly the case frame of [-[ +Nom] ,.+[ +AGT � ] subcor [-ateg[+PATorises],+[+AG theT]] non-agentiv belong to e classverbs VIIinto impe subclassesrsonal ve Ir bsand. IISR-5, and the agentive verbs into subclass V as opposed to subclasses III and IV. SR-6 breaaccordingks down to thethe groucontextualp of non-locative feature [± [agentive+INS]]. verbsFinal intoly, SR-7subclas cross-clases IIIssifies and IV allIlIa agentive and IIIb verbs for thewi thsi mpthele featuretransit [±iveer gv]ver bs(er, gativsubclae)sses, giving IVa and us suIVbbcla forss esthe tralocanstivitive eve rinstrbs. umental verbs, and subclasses Va and Vb for the transitive Thesaturated redundancy case rulframees re(RRite-l)rate , and the the non-admittance definitions ofof tra a nsfourthitiv ityCR in(RR- the2) and notergativity have a subj(RR-3)ect. , itRR-4 does fo notllows have from a co-occurringRR-3 that if PATan ergativeeither. verb does

5.2.1 Simple Non-Agentive [+[ +PAT ],-[ +AGT],-[ +LOC ]] This class of verbs show only one positively marked CR, namely, PAT, in their caseframe fra, theme.se non-agentiveSince there verbsis room can for be more subcateg CR'sorised in the byir unsaturatedthe contextual case feature [±under[+IN S]].class IaIf in no FigureINS is 5.2add belowed, we. haveIf thean INSsimple is addintraed, nsitivwe haeve verbsby de groupedfinition belowof tra, nsnon-agentiveitivity a group simp ofle non-agentivtransitive verbse tra nswillitive be verbsfurther. subcategIn sectionorised 5.2.3 by erthegative feature (cl [±asers gv]IIId) (e rgativsubclasse) , ofresult verbsing. in an accusative (class IIIc) and an 179 The following feature tree will show how simple intransitive verbs and · non-agentiveno�-agentive verbssimple by tra SR-4nsitive: verbs are parcelled out from the class of simple (Figure 5.la) - - +V . +[ +PATJ -[ +AGT J -[ +L-OCJ - -[ +INS J [ - +[ +INS J IaI lIIc,I IIld (see Figure 5.4) SimpleVerbs Intra nsitive Non-agentiveTransitive Verbs Simp le Figure 5.3 Subcategorisation of Class I Simple Non-Agentive Verbs Classesverbs are IIIc agentive and Il. Id Wecl earlshally illureturnstrate to the sepoint verbsthat in notsect ionall 5.2.3.transit ive intraIn thisnsitiv secetion ver bswe . willIn limitterms ourof morphopresenlogicaltation toshape cla ssand la, intrin the sicsimp semanticle togetherfeatures , isthis the iris commona highly charac heteterirogeneousstic casegroup fr. ameWhat, which bin dsis the members [+intra[+PATnsitJ,-ive[+AG verbsTJ,-[ +LOareCJ,- the[+ foINllSJJowing. groupsIncluded of in ver thbsis: category of simple (1) Examsimpleples action inclu de:verbs characterised by the infix -ern- . t-ern- i reng stand t-ern-angic weep3 cry r-ern-adiw sing r-ern-akat walk k-ern-aen eat l-ern-oad rise3 get up (2) involuntary action verbs marked by rna- such as: rna-tepo fa ll rna-korokor roll rna-kelol abate (of the wind) (3) Exameteormplesological include verbs: also marked by rna- . rna-orad rainy rna-bal i windy rna-kotern cloudy

I . 180

(4) phenomenal or meteorological verbs marked by the suffix -an : orad-an rain ba I i-an blowing (of the wind) bal ios-an typhoon, gale betili-an thunder-stoT'm lelen-an earthquake lelesi-an flood (5) stindicatingative verbs psyc markedhologi bycal theor prementalfix ma-state. incluExamplesde: of verbs ma-olah pleased, happy ma -cekok amazed, surprised ma -cal ibad angry ma -canaran annoyed ma-ta law frightened ma-q inal envious ma -ngodo enWarrassed Examfolloplewins g:of verbs indicating physical state include the ma-ori p alive ma-patay dead ma-toas grown up ma -soso fa t ma-kopit skinny ma -botek. blind ma -latoq fragile ma -kedal dry ma-d i cem short ma-talem sharp ma-lahol ro tten ma-lepot broken ma -ngangiq ajar ma-kadep united (6) amatmbientosphe verbsric envi thatronme indicatent, espe thecially state in ofregard the enc toomp classiimateng and verbsdarkne apsspear or liinght the withir rootreference form. to Itthe is ti intereme of stingthe day. to note Th esethat Examtheseples unaff incluixedde verbs: all show a eve eve syllable structure. toqman dark tanglal light, daylight siqnaw cold cal des WaT'm soqmed humid (7) descriptive verbs expressing physical quality. Members of this grousize,p alsoshape appe, colourar in, weigtheirht root, etc. form of . physicalThese obverbsjec tsindi. cateMany the of Examtheseples verbs inclu alsode: show the eVeeVe syllable structure. 181

ad idiq smaU tataang large arawraw round baqkec heavy tarakaw taU lahkak red langdaw blue� green� azure bohcal white coplak sour ahc id salty asoq delicious kapah good bangca l beautiful (8) nounsdescriptive are carried verbs derivover edto thefrom derived nouns. verQubsalities. The ofcharacte the sourceristic preexistfixence is orsi-, in which the pos issessi usedon also of theto indicatePatient ansubj obecje�t. t in Examples of si-N descriptive verbs include: si-pida rich� lit. have-money si-ngangan famous� lit. have-name si-kawas haunted� lit. have-ghost si-kotem cloudy � lit. have-cloud si-wawa child-ed� lit. have-child si-on ing fi lthy� lit. have-dirt The prefix si- combines rather freely with any concrete noun to giveFurthe a r straigexamplesht- forwardare: meaning of possession. si-nan i have-cat si-lomaq have-house si-alopa l have-persimmon The subject of a si-N verb can also refer to a place. In that exiscasetence, instead. Comp of area pos thesess followingive read ingpai, r itof resumes example thes: meaning of (5.20a) si-nan i kako have-cat ls ��rns �� T I have a cat : (5.20b) si-( l okedJ aw ( koJ niaroq no Ka lenko exist-wildcat place of Hualien ��rns Nom+PAT J (Japanese loan) there are wildcats in the neighbourhood of Hualien verbsRelated which to ( thehave si- J anN exclusverbs ivein th(exiisstential group are read theing su. bjecFortle examss si-pleN , si-kawas means God exists or ghosts exist/there are ghosts . Thesedescribed subj ecintless section verbs 5.2. be6 longbelo w.to class VI impersonal verbs 182

s i- verbs are (9) usualN verbsly fo llthatowed imply by a veinstrrbalument comple orment tool ind. icTheseating the action Examfor whichples incthel,ude implied: instrument is used. si-bakic use-a-bucket si-l akaw use-a-stick 5 i -toka r use-a-Zadder si-raic use-a-rope si-losa use-tears (10) verbsof the thatPatient indicate subject. exi stenceThis or group non-exis has onlytence twoof themem bersrefe, rentnam ely, era exist� there is awaay exist-not� there is not (11) The existential verbs era there is and awaay there is not , when usedor pronoun with a, Patientalso ser vesubj toect sig thatnal pois ssessmodifiedion. byThe a genitivepossessed noun entityfollowing is mapairnif esof tedexam asples the: Patient suqject. Compare the (5.2la) si-nan i kia babah i have-cat woman ' +v-trns ' ] Nom+PAT ,the woman has a cat (5.2lb) er( a ko nani( ]n i a baba'h i cat of woman s T the woman has a:;: cat . Lit. a cat of the woman 's exists (12) verbs that (��:indicatel num(ber or] quantity. Included in this set areredu pltheica cardinalted. Exam numbeplesrs , arein : their root form or partially (5.22a) tosa ko ayam two bird ��rns Nom+PAT there are two birds Lit. the( birds) are [two J(i n number) (5.22b) ta-tosa ko ayam two bird ��rns Nom+PAT there are two (birds Lit. the( birds) are two )(i n number) (5.23) cacay ko kawas one God ��rns Nom+PAT there is (onZy)( one God Lit. God( is) One ) 183 Thebe lookedexisten upontial syntverbsactically era there and is sema andntically awaay there as sp isecial not casescan of number verbs with awaay indicating the number zero and era indicating an unspecified number n, where n � 1. (13) verbs derived from nouns that are affected by the action. Thethesese obintrajectsnsi aretive affecverbste haved is notimpl preieddi obctajecblets. ExactlySome imply how Examacquiplessiti onare, : others, the disposal or consumption of the object. m -kasoy gather-firewood m -tipos harvest, lit. gather-rice-grains m -bot ing catch-fish m -ieep chew-bete l-nut m -tabako smoke-tobacco m -poe iq peeL-skin (of fruit) m -eoqeoq suck-at-nippLe m -nanom drink-water m -kabi drink-soup m -panaq shoot-arrow (14) verbs derived from nouns that are the product of creative Examactiviplesties are. : These verbs also have implied objects. misa-qepah make-wine misa-to ron make-ric.e-cake m i sa-kab i make-soup misa-kalabi make-dinner misa-lomaq buiZd-house misa-ribeng buiLd-embankment misa-os iq count-numbers misa-banaq make-belief, deceive misa-be ra-berang put-on-airs, arrogant misa-kero ' dance misa-tamdaw make-fri ends (15) Examlocomplesotion inclu verbsde : with implied direction or goal. ta-ini come, arrive ta-ra « ta- i ra) go (there ) ta-l omaq go home (16) verbs indicating mutual or reciprocal activities. These verbs are marked· either by masa- or mal-. masa-soa l converse with each other masa-eepeep kiss each other masa-skong fight each other masa-so rah id fi re at each other mal-kal iay reconciLe mal -qapa put arms around each other 's shouLder ma I-a lip intercourse 184

The prefix rnal- also marks verbs indicating activity Examparticplesipated are : in by two or more people. rnal -taes hit/fight one another mal -qepah drink together Verbs with the prefix rnal - can also be derived from·nouns whose Examreferpleentss areare: the group participants. rnal- kaka be with elder sibling rnaI- wi na be with mo ther rna l -hetay (Japanese loan) be with the army TheTh�yse behave verbs withlike mannimpliered verbs concomitants in that requirethey take pl urala non- sufinitbjecets . asaction the nonverb-f initeas th eircom plobementligatory of ancom plactionemen t,verb or., they can serve In section 5.3 below, we shall subcategorise class la, as well as other heterSubcategogeneousorisation verb will classes be carried, in terms on ofunt inherentil variations seman tic.in syntac featuretics..a nd morphoaccountedlogical for. patterns within each major category are formally and adequately Listedclass Ia.be low are sentence examples for the simple intransitive verbs of (3.85) t-ern- i reng an ini c ira tepa r no sa I iIi stand now 3s side of post ( rns [+Adv] Nom+PAT ] (LCV+LOC ] he�� is st) anding by the/a( post (5.24) rna-ko rokoro an ini kia bekeloq [ta-ra sasa no lotok ] ro( ll now the stone ) go-toward bottom of hill +v-t rns [+Adv] Nom+PAT int (L+I.OCCV 1 thel stone) is ro lling down( hillJ ( �� J (5.25) orad-an kina rern iad rain this day rns T it( �� is raining/ra(:�:iny today) Lit. today is] raining/rainy (5.26) rna-cal ibad kia rnatoas-ay angry the old-one rns Nom+PAT the(�� old) man is upset( ) (5.27) rna-patay to kia rarapa dead already the water-buffa lo ( rns [+Adv] ( T] �he�� wa)ter buffa lo is dead:�: 185

(5.28) ma-ngangiq ko sas i nga ran window -trns (Nom+PAT] the window is ajar (5.29) (ar��awraw 1 ko bolad na-cila round moon !/-esterday l-t(+vrns ] (Nom+PAT ] (LCV+TIM ] the moon was fu ll last night (5.30) si-kawa s kina lomaq haunted this house (��rnsJ rNol+PATm . this house is hauntedJ (5.31) era ko kawas exist God� host ( ��rns] "(Nom+PAT God exists r or, there are ghosts (5.32) awaay ko kawas exist-not . God� host ( ��rns] (:�:T God does not exist r or, there are no ghosts (5.33) na [mi -kasoy kami ( i) na-c i 1 a did ga ther-firewood lexcl yesLCVterday (:�int] ( :�int) (:�:T) (+TIMJ be 1 i q nina 1 otok ] beyond this hill . (���cJ we gathered firewooq yesterday beyond the back of this hill (5.34) mi -bot ing kami tarawadaw catCh- iSh lexcl river (��rns (:�:T (���cJ f J river we go fi shing in the (5.35) misa-ribeng kami it in i build-embankment lexcl here (��rnsJ (:�:TJ (���cJ we are building an embankment here 186

(5.36) masa-so rahid kia tatosa-ay a tamdaw fire-at-each-other two man ��rns :��T the two men fired at each other (5.37) ma( l-taesJ kia ( adiwJ awa itia potal-an fight/hit-one-another chiZdren yard ��rns :��T (���c1 the chiZdren are fighting (together)] in the yard (5.38) ma( l qepahJ amin ko bi(nawlan taloq-an drink-together aZZ ViZZa ers community-house ��rns +Adv :��T (���c] aZZ the viZ Zagers drink togetherj in the community house To top the list (of simple] intransitive ver(bs, we have the interrogative verbs ma-maan and masa (which probably came from ma-saan) . These verbs can roughly be translated into English as how-is-it-going? and they are used to seek examinforplesmation are about: the state of being the Patient subject is in. Sentence (5.39) ma-maan saw kiso? how-do QM 2s ��rns Nom+PAT how are you doing/feeli] ng? or, what( 's ] happened( to you? (5.40) ma-sa saw kiso? how-do QM 2s ��rns Nom+PAT how are you doing/feeling? or, what 's happened to you? The wide range of( possi) ble answers( ) to these two questions reflects the wide coveragepsychological of ma st- verbsate, a inchange Arnis. of Thesta teanswer, or amay simp focusle mo ontion a . physical state, a

5.2.2 Int. ansiti ve Locati ve [+[ +PAT ],-[ +AGT],+[ +LOC ]] In our Arnisda ta, there are no instances of LOC co-occurring with INS in the caclassse frIIame show of agentivan unsaturatede and non-agentiv case framee withverbs onallyike two. CREven's, clathoughss IImem isber nots of matterfurther of subcate fact, gotorised class in IIte rmsintra ofnsiti the vecontextual locative vefeaturerbs, the[ ± [ +IcontextualNS ]]. Asfeature a -that[ +INS the] istwo to CR be' s,redundantly INS and LOC adde, ared. mutuTheally following exclusiv twoe inredundancy the same caserules frshowame . RR-Sa [ +[ +LOC ]] [ -[ + INS ] ] RR-Sb [ +[ +INS ]] [-[ +LOC ]] 187 As has been mentioned before, a negatively marked contextual case relation parwouldticula implyr case that notion the lang neededuage wouldto be exresortplicitly to a exprcomplementationessed. device if that Like class Ia simple intransitive verbs, class II is also a neterogeneous Whatgroup binds. There the memis bersno unique together charact is eristictheir common affix charanor ctersemanticistic inter case pretationframe, . whichintransitiv is [ +[e +PATlocat] ,-ive[ +AGT verbs] ,+[ are+LOC the] ,- [ folINSlowing]]. Incluthreeded grou inps th: is category of (1) lofocomotionllowing exverbsampl escharact: erised by the prefix ta- as in the ta-ra « ta-i ra) go (there) ta-ini come, arrive ta-ngasa go-up-to ta-tihi go-against, oppose ta-ra -pota I go-outside ta-todo go-in-between ta-Iomaq go home (2) verbs indicating existence and/or number in a specified universe claor locationss Ia. . TheyThe incluse verbsde the are two re exilatedstential to the verbs existential as well verbs as the in carincdinalludes numthe bersfollowing with or ver withoutbs: reduplication. This group era there is, there are some awaay there is not tosa , ta-tosa there are two lima , I a-I ima there are five (3) specific-object intransitive verbs. In sections 3.1.3.1 and re3.4.alis3.4ed we as haveAccus shownative thatPatient in Ami whsile non-sp specieciffic icob objecjetscts are are realised asThe Locativespecific-o Locusbject. intraIn thensi tivelatt erverbs case inclu, the deve actionrb is de- verbstra nsasit weivislled . asrespective psychologily. calExam verbsples. incluDerivationde: prefixes are mi- and ma­ mi -k i lim seek, look-for mi-alod throw-at mi-angang ca ll ma-olah like, love ma -ca l ibad angry-at ma-canaren annoyed ma-tal aw afraid-of psychoThe ma-logverbsical instative this groupverbs arein classderivatio Ia whichnally arere laalsoted markedto the by verbsthe ma with- pre classfix. IaThe si mpderivationalle intransit reivelat verbsion of onthis the grooneup hand of classand II withshown classin Chapter IlIa si 6 mp(sleectio trannsitiv 6.3.e2.3 verbs) . on the other will be 188

Listclassed IIbe. low are sentence examples for the intransitive locative verbs of (3.83) ta-ngasa c ira panan a [mi -dakaw ] 3 gate ride Noms LCV (+PAT J (+LOC J ( ��intJ he-trns rode up to the gate (5.41) F�::na :r [ta-ra c ira itia pacakay-ay] did 3s market Nom (Lcv J [:�intl +fint [+PATJ l+LOC she+xl wentry to-trns the marke t (5.42) era ko tamdawI 1 ikot no sas inaran there-is l�V person outside of window J J (���C there(��rns is someone(:��T outs ide of theJ window or, someone is outside of the window (5.43) awaay ko oner Hawaii there-is-not snake Hawaii (���cJ there( ��rnsJ are no snakes(:��TJ in Hawaii (5.44) mi-alod c ira itakoan throw-at 3s ls rNom (LCV he( � �isrns thrJ owingl+PAT J(t hings+LOC) J at me (5.45) mi-angang to haw kiso iciraan? ca U already QM 2s 3s [+Adv] Nom ) (LCV have(��rns you J caUed him? (+PAT +LOC ) (5.46) ma -olah kako itisoan like, love ls 2s ] Nom ) (LCV ) I( ��loverns you (+PAT +LOC (5.47) ma-ca 1 i bad kia matoas-ay ic i Panay-an angry-at old-one Panay Nom LCV the( �� rnolds] man is angry(+PAT at ] Panay (+LOC ] (5.48) ma -ta law kako ira tamdaw-an afraid-of tthat man lsNom LCV (� rnsafra ] id (of+PAT that ] ( I am� man +LOC ] 189

5.2.3 Simple Transitive [+[ +PAT ],+[ +AGT],-[+LOC],-[+INS]] Transitive verbs can be subcategorised in termsof the extended case frame thefeature perma nownent la, beassocilledation as [�e ofrgv the] (eNominativergative), withCF with er gativitythe Pati entforma CRlly and de thefined as classiassociationfication of ofthe ArnisAgenttra ornsit Insivetrument verbs withwith the the Genfeatureitive ofCF .,er gativityCross­ has given(cf. Fig usure accus 5.2)ative. andIn sectiergationve,5. sUbc2.3.2lasses below for, we verbwill cla pressessent lIt class, IV IlIa,, and Vthe sithemp silemp accusle accusativeative tra nsittransitivive verbse verbs wit hwith Agent Instr subjecumentt together subject. wit hIn cla sessction IIIc, Geni5.2.3.2,tive Agentclass andIIIb Genitiveand class Instr IIId,ume ntthe respesimpctivle erelygativ, wille tra bensit presenive yerbsted. with FigureClasses 5.3 III cin andsec tIIIiond 5.2are .1dis) . tingUished by the featu,re +[ +INS J (see SR-6 and

5.2.3. 1 Accusative Simpl e Transitive verClassbs . IlIaAccu fromsative Figure tra nsi5.2tive are verbsthe non-ergative in Arnishave or theaccus characteative risticsimple pretransfixitiv e mi-.corresponding with these to thever. bstraditional, PAT is tynotionpicall yof re"dalisedirect obin jecthet" Ac. cusativAs hase beencase form, non-spdiscussedecif icin obsectject.ion �.1.3.1,Examples Accus areative: 'Patient in Arnis is associated with a (5.49) mi-qosaw to haw kamo to hemay n i Apoy? save alP+AdVead QM m riceACC of Apoy +sptl ��+AGT +PAT } :-ergv [+cmpll J have�rn you saved some rice[ fo r Apoy ? ( or, have[ yousl saved some of Apoy 's rice ? (5.50) mi-ala kia wama tia nan i (J escue fa ther cat (Nom � ACC n +AGT +PAT -ergv1 l J ( thet fa:ther rescues the cat ) (5.51) mi-[: kil im kako to bada l , l Or Is berries Nom J ACC J -ergv +AGT (+PAT I looking fo r berries am�: : ( (3.27) mi-[: ki lil m kako to bot ing l Or Is fish Nom+AGT ACC+PAT J -ergv J ( I looking( fo r a/some fish [:am�:: :l 190 can see that Accusative Patient is associaIf we retedfer wit backh theto sectioless spn ecif3.1.3.1,ic re ferentswe in the feature gradation presented in Figure 3.1. In.It faisct, to inbe standardnoted that ter minologspecificity, ythey is notare mututo beally equated exclus witiveh detefinitrms, witheness "s. pecific" meaning that the referent is clear in the minds of b01:h theasser speakerts that andthe there ferenhearert isand clear "def onlyinite "in meaning the mind that of the speakespeakerr. assumesMy use and oE thefeaturese ter [+msdf ntdeviate] (defs initsomewhate) to befrom re intethe rstandardpreted asus spageecif, inic that when I theal lowfeature the s [+featurelctn] matr(locatioix. n)The, [+ prsndegree] (pofer spsoecificn), andity [+ prnnincreases]. (pronoun) as these are featureaddec;1tos are the addedsection on 3.1.one3.1, at a my ti .deme finiin the tion order of sp ecificitcited. yAs come hass beenclosest ment toioned what ea Hopperlie.r i:1a:1d Thomhave psonclearly would il calllustrated "individuated" this point ,(1 we�8 0:2are53, not 287) going. Sitonce repeat exam plesthem her3.2e.7- 3.30 It is perhaps important to point out, however, that the gradual shift from non-sptransitivecifice to to intra specnsitificive and and the from corresponding Accusative diPatientrectio� to ofLocative change Locusfrom in Amiinstances will, wouldupset thinksome ling thatuists the . more Hopper specif andic theThom psonobje ct(i bid.is, the:253) "m, orefor" Locativetransitiv Locuse the , verbwith shouldapprop beriate. Aladjustso, mentwh�n Accusin CR-asativesig nmPatientent, the changes verb beintocom2s saintratisfnsy itivisedthe definitio in then profoce an ssan andti-pass takesive aderivation Patient su procbjectess. , Thatalthough would an ti­ passAccordingives usuall to Stary ostahave less(Per sonalspecific commun rathericatio thann), morethis kispndecif ofic ¢-marobjeckedts . anti­ passaccurateive isand unheard I have ofno inbetter his experien way to ceaccount. To for the theextent pattern that exthehibited Amis datathere are, theI wo notionuld have of tosp ecificconcludeity that. this is an anti-passive process which involves Reclassferring I non-a backgentive to Fig ureverbs 5.3 with in thesect contextualion 5.2.1, wefeature find [±that[+I NifS]], we wesubcate will gogeriset [+an[ +PAintT]ran,-sit[+AGiveT],- [+[+LO+PAT],-C],+[+I+AGTNS]]],- [su+LObclasC],-s.[+ INTheS]] lasubcttelassr subclass and a tracannsiti be ve desubcascrtegiptiveorisaly tionlabe oflled the these non-agentivsimple non-agentive transie tivtrae nsverbsitive by verbs the . featureA further [±er gv] latt(ergativeer, er) gativegives . us Thsubclassesis can be IIIcform ulaand tedIIId, into the SR-8 fo rmeberlo w:accusa tive and the (5.52) SR-8 + [±ergv ] l+[�� :���+INS ] and shown by Figure 5.4 (or n page jj19 1). Class IIIc are the non-agentive accusative transitive verbs which obey the Filis chosenlmorean assu bjecsubject.t cho iceThe hie verbsrarchy in. exaInmple thes absence3.54-3.5 6of inan seAgenctiont, Instr3.3.2ument.1 belong-[+LOC],+ to[+ thisINS ],cla-ergssv. ]. TheI willcharacte repeatristic only case exa mpleframe 3.is55 [+ to[ +PAilT],-lustrate[+AGT], the IIIcfeature can spbeecif foundication in sectof theion verb3.3.2.1 and under its contextNominative. MoreIns trumexamentples. of class 191

(3.55) mami-a sik c ira to adawang no cacodadan assig[ +v] ned-to-sweep 3s fpont school (:�:S � T he is assigned to sweep ) the fpo(�nt� of the school J (Figure 5.3) +v - -- +[-[ +AGT]+PAT] +[-[ +I+LOCNS] +e gv IIIr d e.g. sa-pi-angang ma-asek used-fop-cal ling­ stpikeI someone sa-pa-ahc id ma-noang used-fop-making­ mo ve, wave something-salty mam i -as i k ma-apa assigned-to-sweep dumbfound

Figure 5.4 Subcategori sation of Non-Agentive Transitive Verbs wi th the Feature of Ergativity; with exampl es

5.2.3.2 Ergative Simpl e Transitive Byco -occudefinitrringion Ag, entan er orgative Instru tramensitint in vethe verb Genitive has a Patientform. Thesubj ectergative and a simple traIn nsitsectiveion verbs3.2.2.2 with we haveGenit givenive Agent exam bepleslong of to class class III IIIb bergative from Figure transit 5.2.ive verbsgiven bewithlow the: prefix ma- (see examples 3.42-3.43). A few more examples are (5.53) rna-ala nom i so ko bodoy ako t (bY mistake} 2s cloNomthes Is s T (+PAT 1 :�::+ergv l �� you[ have taken my clot( hes l (by mistake) Lit. my clothes have taken by you (by mistake) 192

(5.54) ma-adop t03 haw namo kia koyo? hunted-down az+AdVpead QM 2pl foNomx :�r", +sptl ( T1 (+PAT1 +ergv +prft1 :� have you hunted down the fox? Lit. has[ the] fo x al[ready hunted down by you? (5.55) ma-asik t03 nomako ko karimocoq no lomaq swept alread ls corners house :�rns :::�l ( 1 (:�:T1 +ergv +prft1 ��� I've swept every corner of the house Lit. the[ cornersl [ of the house already swept by me (5.56) ma-angang t03 haw nom i so c ira? called alread QM 2sGen 3sNom :�rns+ergv :::�l+prft1 (+AGT 1 (+PAT 1 have you called him already ? Lit. has( he ] alr l eady ca lled by you? (5.57) ma-oay t03 non iam kia ni-osaw-an a oa-oay-an gathered-rattan alread lexcl reserved-placeNom rattan-area :�rns+ergv :::�l+prft (���T1 (+PAT J we[ havel gathered ( all the rattan (there is) in the rattan reservation Lit. the reserved rattan areaJ has already gathered/exhausted by us These ergative verbs can be interpreted as the "passive voice" of corresponding witrathnsitiv aner e gativverbse orverb intra the nsitivPatiente verbs subj ectwi thhas im plieda definit objecte readins, markedg. byIn mi-.the folformslow ingin class, the verbsIIIa orfr omclass exam I.ples This 5.53-5 deriv.57 ationalare li stedrelat withion willtheir be related accounted mi­ for in Chapter 6 (section 6.3.2.3) . Class IIIb Class IIIa or Class I ma-ala taken (by mistake) mi-ala take ma-adop hunted-down mi-adop hunt ma-saik swept mi-asik sweep ma-angang called mi-angang call ma-oay gathered-rattan mi -oay gather-rattan ma- Syntactically similar to the ergatives but much rarer in class IIIb is a group of verbs characterised by the derivational suffix -en (see examples 3.44 and 3.45) . A few more sentence examples are given below: 193

(5.58) taes-en n i ra kia wawa hit 3s child - +V Gen Nom +trns+ergv (+AGT ) (+PAT ) he hit the chiZd Lit. the[ chi1 Zd hit by him (5.59) cebar-an nomako ko sikal a [ma-bot iq] s -out lsGen matNom sZeep [:�:::+ergvl (+AGT ) (+PAT ) (��int) I spread out the ma t to sZeep Lit. the mat spread/ra ZZed out by me to sZeep It is difficult, if not impossible, to predict whether the "passive" form of a mi- transitive verb is marked by ma- or -en . Examples 3.43 and 3.45, taken from"pass Starostaive" affix'ses field, innote Amiss atand leastlisted, are again derivational below, support and not our infle analysctionalis that. Since the meaning of the two ergative forms, ma-me law see and me l aw-en watch, care-fdistriorbution are , notthey predictable have to be fromderiv morphoed. logical shape nor by syntactic (3.43) ma-melaw nomako kiso na-san i (M171) see ls 2s � i ago +v 1 Nom �: [+erg+trnsv (��T) (+PAT ) (+p+TIastM I just saw you a whiZe ago Lit. you jusJ t saw by me a whiZe ago (3.45) me l aw-en no wawa koT tomay (M126) ;Oh" take-oape-of chGenild bearNom �+e+trnsrgv (+AGT ) (+PAT 1 the chiZd took care of the bear Lit. the bearJ took care of by the chiZd Withineven though class no III LOCb isis a reqsubclassuired inof thema- caseverbs fr thatame. implyThe motioncharacte orristic locomotion prefix ofis exprethis sssubced lassas the of Nomisituationativenally Patient motion. verbsExampl ises ma- are, : and the notional locus (5. 60) ma-dakaw nia wawa ko rangad cZimb-over child feNomnce :�rns (��T) l+PAT( 1 +motn+ergv )j the chiZd cZimbed ( over) the fe nce Lit. the fence cZimbed (over) by the chiZd 194

(5.61) ma-dangoy n ira ko c i r i 5 s� m-across 3s river +trns� (+AGTGen ] (Nom+PAT ] +ergv+motn l he swam (across) the river Lit. the river swam (across) by him The additional semantic feature [+motn] (motion) does not have any immediate syntactransitictive consverbsequence. It which is added would to diff idenertifentiy atethis thsubcis classlass offrom verbs othe andr er togative provideexamples more can beinfor foundmation in sectfor ionsubsequent 3.4.3.1 semantic. interpretation. More previousHere again sectio the ntradit we haveional ju notionst seen of that tra Arnisnsitivityintra nsiis challengtive verbsed. with In Locativethe Locustheir actutranallysitiv havee counterpart notional s,ob majenifestcts thated areas Acmocusativre specifice Patient than. theThe ob Arnisjects of Englishergative trtraannsislationstive motion either, verbs but dothey not are sound consist veryentl "tyra nstransitivite"ive with where th eiformalr criteris the ia,pr imaryi.e. , conscaseide frrationame and. morphological features are concerned, and that Not included in class IIIb are members of class IIId which are ergative verbstransitiv indicae verbste action with Georni sttiveate Instrimmediatelyument instea effecdted of byGe nitivnaturale Ag forcesent. suchThe seas seclighttionin ng,3.3.2.2 wind , willstorms show, and: earthquakes, as examples 3.59 and 3.60 from (3.59) ma-asek no bet iIi c ira strike lightning 3sNom +e:�rnrgvs (��S) (+PAT ) he( was struck by lightning Lit. hel struck] by lightning (3.60) ma-nanoang ko kilang no bal i (M147) move� wave tree wind ( :�rns) (:�:T) (+IGenNS the tree is moving in the wind ] Lit. the tree is moving by the wind ClassPlease IIIredfer can also be characto the terisedfeature by tree SR-8 in (5Figure.52) in5. 4the there previous. subsection. Therest Gericnittedive to Insnaturaltrument forcof es class, as theIIId fo erlgativlowinge trasentnsiencetive ex verbsamples iswi notll show': (5.62) ma-apa kako no bangcal n ira dumbfound lsNom beautGen 3s :�rns+ergv) (+PAT 1 (+INSl I[ am dumbfl ounded by her beauty 195

(5.63) ma-cocaq kako no anaboq irritated (in the eye) ls dust rns (Nom+PAT S +erg:� v l �� my( eyes are irritated by the dust Lit. Il irritatedl (in the eye) by the] dus t( ] (5.64) ma-palawad ko ba l ocoq nomako nina rad iw moved heart ls this song rns Nom+PAT S :�+ergv ( :�� my[ heartl is moved/touched] by this song ( ] Lit. my heart moved/touched by this song (5.65) ma-canar kako no se laq n i ama ako disturbed lsNom snoringGen fa ther ls rns +PAT +INS ] :�+ergv J I[ was disl turbed( by my fa(ther 'S snoring Nor does the instrument have to be inanimate [ -anmt ] . The following two examples show ergative verbs with animate [ +anmt ] Genitive Instruments: (5.66) ma-cekok kako nom i so startLed lsNom 2sGen +ergrnsv +PAT +INS you:� startLed me ! J Lit. I[ startl led (by you] ( (5.67) ma -rahteng toay nomako kina demak recalled alread ls this incidentNom +ergvrns +prftl �� +PAT I':�ve recalled:::� this inc(idenstJ J Lit. this[ incidl ent [ has already recalled by( me One may argue that if we change ourI analysis of 5.62-5.67 as well as 3.59-3.60 tostill have be a traGenitivensitive Agent and er insteagatived . of aIf Ge wenitiv did e thatInstr, umverbsent , inthe th verbsese ex wouldamples would belonot beng necespto classary III. bAlong and the the instrsame umveinental, a categsimilarory , ari.e.gument class can IIId,be undeadvancedrstanding in favour that bothof grouping AGT and class INS canIII bec witheith erIlIa. animate In viewor ina ofnim ourate , this printeesentrchang. e Inseems the plausiabsenceble of es valpeciallyid syntac whentic only disti onencti ofon theforse th twoese eR two's typesis ofsi tuatioconstructional criterins, wea. find ourselves in danger of assigning case relations by Starosta (1982d:23-25) recognises the existence of an inner-outer distinction betweenconsiderat Agentions andsuch Inst as rumenanimacyt. orInst voealitiod ofn, reStarlyingost ona proposed external to sit disuationaltinguis h 196 the Instwo trumentcase re wolationsuld be inass teignedrms of to thetheir external relation intofluence the ob whichligatory was Paviewetientd : e.S woumoreld imm be ediathe teactantly impinging referring on theto the Patient entit y(cf. viewed section as the 3.3.1 ulti) , mateand theforce Agent of t.he actioand IIIn asc ve a rbswho, leand (c off. classsectio nII Ib3.2.1 and) . IIIThed verbs amb igisuous an anunfortunatealyses of classresult IIIe of . thInsistru apprmentoach are. preThsentis am inbiguit the. ysa mecan casebe reframesolved.. onlyIn other when wordsboth an, we 'Agent would and have en clto assre ferIII to si mpclassle tra IVnsit transitivive veer bsinstr. umental verbs to resolve this problem with First, let us consider Starosta's English example which has to be given ambiguous analyses:. (5.68a) The A�stor m dest royed the treePM house . (5.68b) The storINS m destroyed the tree PAT house . corresponding to two possible perspectives, according to which the storm is desviewedtructio as eitn. herTh theis amultbigimateuity would(AGT) beor retheso immlvededi ifate both (INS) an efAgentfector and of an the SuInsbjecttrument Cho wereice Hieprerarchysent, : since Agent would outrank Instrument in Fillmore's (5.68c) The AGTstor m destroyed the treePAT hou se wi th a powe rful gustINS of wind. For the time being, I am leaving the ambiguity involving class .111 verbs unclassresolv IVed. tra nsitivWe shale instrl returnumenta tol vethisrbs problem. in section 5.2.4 when we come to verbsIn summary as we, llcla asss natural IIId, asforce shown action by the ver exambs. plesThe, co charactensists ofristic psyc hologicalcase frame for class IIId verbs is [+[+PAT ] ,-[+AGT],-[+LOC],+[+INS],+ergv].

5.2.4 Transitive Instrumental [+[+PAT J,+[ +AGTJ,+[ +INS]] [+The[ +PAcharacteT],+[+AGTristic],+[ +IcaseNS ]].frame Siofnce tran thesi casetive frameinstrumentalof Amis verbs verbs ofgets cla sassturated IV is withCR, saonlyy, [+threeLOC], case in threeirlations case , frameclass. IVTh verbsis li mitscan neve the r derivadd onational another poten innertial offeature this [-class[+LO C]]of ver canbs be, rulingredundantly out the add additioned. Should of a aCR notio. Thnalus theloca tiocontextualn need tocom plementationbe explicitly device express wouedld with be usa ed.class IV transitive verb, a syntactic As has already been discussed in section 3.3.2.3 under Accusative Instrument, whencase theform INS, it actantis to beof classinterpreted IV tra nsitivas an e"i verbsnterm ediisate realis" causeed inor the means accusa of tive meactionntioning indicated of the by means the byverb which. This an actionclass ofis verbsaccomplis allowshed . the explicit 197

5.2.4.1 Accusative Transitive Instrumenta l Cross-classifying class IV with the feature [ ±ergv ] (ergative) gives us, as Filaguretter, 5.2ergative shows,. subclasses IVa and IVb, the former being accusative and the Unfortunatethere is a gaply, wherein my dataaccus. ativeCorresponding transitive toinstr a umentalclass IVbverbs ergativ are e conceverb rned, pinaro-i fi ll (see example 5.72 in the next section) is the accusative case formit is mi- usedpin aras o afi nolln- givenfinite inverb exam. ples 5.69 and 5.70 below. In both instances (5.69) tomes-en noniam a [ mi-pinaro] kina kari reng to ra ram i fi lled-up lexcl fi ll this cart stACCraw :+e�rnsrgv ] ��int :��T ] +INS ] we fi lled the��T cart up (full) with straw Lit. the[ cartl fi( lled up (ful( l) by] me with (straw ( (5.70) tomes-en nomako a [ m i -pina ro ] to simal kina si-natoik fi lled-up ls fi ll oil this bottle �rns ��int ACC+INS Nom+PAT :+ergv [� l ] I fi lled the bot� tle up (full) with oil Lit. the[ bottll e fi lled up ( (full) with oil( by ) me ( ) There is no way to ascertain the complete case frame of the non-finite mi-follopinwingaro exceptexample indirectly: through that of a related form ni-pinaro in the (5.71) tomes-en nomako [ ko ni-pinaro tina si-natoik to simal ] f -up lsGen fiNomlling this bottle oil :�:+ergv: +AGT ] +PAT ���T ���S I fi:lled this bottle with oil Lit. the[ fi lll ing( of this bo(ttle ] with oil (is)( fi] lled ·up( (full)] by me accusIn exaativemple tran5.71,sit theive derivedverb and noun carries ni-pin witarho fiit lltheing orihasginal its casesource frame in anminus theverb Nominative tomes-en. AGTThough actant thus which shown is toco-re be ferena plaustialible wit hca ndidatethe AGT offor the class hig herIVa , mi-pindata gaparo isfi llfil willled with not morebe inc dirludedect evin idencethe inventory. of class IVa until our From section 3.3.2.1 of Nominative Instrument, we have one more candidate, sa-pi-angang used-for-calling-someone , for class IVa. This verb from example 3.5to parti6 has althely mecharacteet the risticdefini casetion offrame ergativity of class. IVThe and co-occurring it has a gen PAT,itive agent howevclarityer, ofis presentation realised in , theI amacc repeusatatingive case exam pleform. 3.5 6 Forhere ease. of reference and 198

(3.56) sa-pi -angang nomako kina piqio to tamdaw u or-calling ls this whistle p'eople Gen . Nom ACC [:?e�:::rgvl +AGT +INS +PAT I use this whistle (to ca) ll people( ) ( ) Lit. this whistle (is) used by me to call people Since sa�pi -angang violates the Fillmorean subject choice hierarchy for asaccusativ a truee insvetrumentalrbs, I have focus decided construction to exclude (s eeit sefromction class 3. 3.2IVa.1, )and . classify it It has been pointed out in section 3.3.2.1 that sa-pi -angang is presumably AGTderiv, edPAT from, and a INStra nsiin tiveits caseverb frapime-angang. In ca otherll (someone) words, thewh icexish shoultenced alsoof have sa-pi -angang implies the existence of its source, an accusative transitive againstinrum. ental verb belonging to class IVa. Regretably, we encounter a data gap Unless we accept sa-pi -angang used-for-calling-someone or mi-pinaro fi ll as a meman berempty of se classt. ByIVa all, we indicatiwould haveons , to,howev forer, the th timeis accusativ being, elea subcve classlass of IVa as categtransiorytive for instru possimentalble· future verbs usedoes. exist and we should at least establish this

5.2.4.2 Ergative Transitive Instrumental In this category we have ergative transitive verbs with accusative instrument Wethat als iso tohave be inint therispreted subc lassas an verbs "inte withrmediat g�nitivee" cause instr orume meansnt asof we anll actio as n. gerenilationtive canagent further. This subcateg differoriseence in cl CF-CRass IVb assoc ergativeiation trforansit theive INS ve caserbs. The stem pinaro fi ll seen in section 5.2.4.1 can also takeon an ergative form. Though I do not have a pinaro-en in my data, I do have the ergative imperative form pinaro-i in the following example: (5.72) pinaro- i to bingkos kina koakoq nomi so shredded-tobacco this p'ipe 2s +trns ACC+INS Nom+PAT +m+eprtrgv ( ) ( ) fi ll your pipe with some tobacco ! Acharacte genitivrise agenttic rzcase is undeframerst ofood class with IVbth iser gativimperae tivetra nsiverbtive, soinst pinrumearo-ntali has verb�; the . Another example of class IVb verbs is tomes-en fi lled up given in examples 5.69-5.70 with the verbal complement mi-pinaro. In example 5.73 below we will show tomes-en with the same case frame but without a verbal complement. 199 (5.73) a [ tomes-en noniam kina koiti to ra raq] will fi lled-up lexclGen this sackNom beanACC :�int+futr ( :�int) (+AGT ) (+PAT ) (+INS ) we are going to fi ll up this sack with beans Lit. this[ sackl is going to be fi lled up with beans by us Thetom es-mainen verbfrom examtomesple-en 5.7in 3,exa doesmple not 5.7 be1,long though to sharingclass IVb. an identIts icalcase formframe witis h noundiffer. ent from that of a class IV verb and its Patient subject is a derived Example 3.61 from section 3.3.2.2 shows a genitive instrumentco-occu rring with aINS genitive as two disagenttinct in casethe saremel ationssentence even. thoughThis deincisiv examelplesy est 5.ab62-5lishes.67 AGTand and 3.59-3.60, the analyses can be ambiguous. The genitive instrument lakaw stick doin theexam pleINS actants3.61 which in exais mrepeatedples 3.5 belo9-3.60.w, does Nor not does indicate it indicate natural an forces, as above"interm. edia It tede" sigcausenate ors themea ns"imm, ediatas do e"the eff INSec tivactantse cause in actingexamples on the5.6 9-5PAT..7 3 (3.61) ma-bahbah kia waco no lakaw nia tamdaw drive-away dog stick man " (:�rns) (:��T) (:��S) (:��T) the man drove the dog away with a stick Lit. the dog drove away by the man with a stick wouldThe fol expectlowing itsexam INSple tois be probl realisedematic in. theSi nceGenitive the verb case has form a ma-. prInstefeaixd,, onethe tyINSpical actant of ofthe ma-a-en loved rbsthr. ow-at is realised in the accusative case form, a form (5.74) ma-alod kako n ira to bekeloq throw-at lsNom 3s stACCone :�rns+ergv (+PAT ) (��) (+INS ) he threw at me (and hit me) with a stone I hit byl him throwing with a stone Lit. [A � CFThe rema-alisverationb in ofexa themp leINS 5.7 acta4 andnt . thatSinc ine examthereple is3.6 no1 diabovefference differ in only in the morphoup withlogical an explanation shape or . caseThe fr answerame features perhaps oth lieserwise in , awe di areffer hard-putence in sema to nticomec interpretation. It is inherent in the verb ma-alod throw-at to have a trajectory along which an �ntermediateljcause, in this case, bekeloq stone exerts its influence. With ma-bahbah drive-away , the instrument lakaw stick neveimmediater leave causes the andhand is ofthere the foreagent re toalis traveed asl a trajGenitiveectory Ins. trumenIt ist. always the As to why alod throw-at takes on a ma- prefix instead of the expected -en suffix as in tomes-en fi ll-up, it is perhaps another historical accident. As we have shown earlier in section 5.2.3.2, the root melaw see can derive into ma-melaw see as well as melaw-en watch . In this section we have also seen that, in our data, alod throw-at is matched only with ma-, tomes fi ll-up only 200 wibe thdi sm-eniss, edand aspi datanaro figapsll , withbut wene itdoher need ma- morenor sys-entemat. Thicis elic is tooita tionfrequent before to we canknow for sure which stems go with ma- and which with -en, and why. I have also been unable to add an AGT to the case frame of class IIIc or class 5.III6d8c instr. Theumental dist inctiverbson asbetween Starosta class was Il abIale and to doIII withc, between his English class exampleIIIb and sentencesIIId, and bet3.6ween1-3.6 AGT2 may and there INS forefor bethe justa�alysesified onlyof sentences on situational 5.62-5 .67ground and s, depenis undedingsirable on thein perspective lexicase practi in relceation, we have to theado ptedPatient the acta prentsent. anThoughalysis thi� becausethe basis (1) of the evidence distinctio like n exambetweenples AGT3.6 1 andand INS5.7 has4, and alrea (2)dy we been would est haveablis hedto on naturalsomehow seekforce semanverbstic and explanation the psychological if we are ve torbs event. Inually my opinion disting, uishto utilise the an maexistingking even distinc semantiction charactand itserisa definititionson. in the metatheory is a superior way for

5.2.5 Transitive Locative [+[+PAT J,+[ +AGTJ,+[ +LOC]] Class V verbs have the characteristic case frame [ +[ +PAT ] ,+[ +AGT ] ,+[ +LOC ] J". Withadditional already CR. three That CR 'sis into thesayir, evencase thoughframe, classthese Vverbs verbs are can agentive never takeverbs on an which would thus be expected to have a potentially co-occurring [ +INS ] , they catonnot be siaddmultaneou this CRsly to expretheirss edcase, would frame most. The li kelynotion take of theins trformument of , a ifverb it needsin a comredundantlyplementation added structure to the case. Hence frame theof thecontextualse verbs feature. [ -[ +INS ]] can be This class includes the transportation verbs and information verbs. Transportationlocation to anothe verbs+. invInolv a e strictthe transportation sense, information of certain words obarejec tsthe fromspecial one Thclassus, ofform transportationally in terms verbsof case which frame involve and sema thentically transportation in terms ofof inintriformatiollnsic . meaas niexng,amples transportation 5.75-5.81 wi verbsll show and, theyinformation are simi verbslarly marareked sim inila mor. st Furtheinstancrmorees. ,

5.2.5.1 Accusati ve Transitive Locative Cross-classifying class V verbs with the feature of ergativity should yield, as Figure 5.2 shows, an accusative subclass Va and an ergative subclass Vb. In Amis, ' the distribution of accusative and ergative transitive locative verbs corensisflectsts diffof declarativeerent synt acticforms functio, the ernsgative. Wh ilesubclass the accus containsative impesubclassrative fonns. locaIn thtiveis secverbstion which we will inclu givede exampletransportats of ionthe andde clinarativeformation accusative verbs. transitive 201

(5.75) pa-osa an ini kia babahi to ka-kaen-en itia ma-omah-ay de liver now woman food farmer [+Adv J :�rns (No+AGTm l+PAT(ACC (LCV+LOC -ergv J J J ·the womenl are bringing fo od to the farmers (5.76) .p[a-ta-ini c ira to baro saka itakoan 3s flower for Is Nom (ACC LCV } -ergvs (+AGT +PAT ;:n (+LOC he brought me someJ flowers J ( J Lit. he:�: broughtl some flowers fo r me (5.77) lpa-ta-ra cira to cacay a codad send 3s one book� letter +v 1 Nom (ACC +trns-ergv (+AGT } l+PAT } saba nira [ youngJ er-sibling 3s (���C} he sent a letter to his younger brother (5.78) pa-sebanaq ci kaka ako itakoan tia kongko teU elder-sibling Is Is story Nom } � (Japanese loan) :�rns-ergv (+AGT (�� C) (ACC+PAT } my[ elder brother told me the story (5.79) caay l [ pa-caqob c ira to [ni-pal ita-en ako not(+V answer 3s aSkinA Is l+fint (:��T} (+PATCC r +ngtv -ergv a demakI JJ questionl( :!��� he did not answer my question Verbs in these examples show the characteristic prefix pa- which is typically the marker of causative verbs. Strictly speaking, the verb pa-ta- ini bring Howeveis literallyr, there aau arese-t someo-be -hereverb formswhile inpa this-sebanaq class te llthat is seemliteral to bely basicaause-t ando-know . undesentencerived 5.. 81.Exam Maybeples arehi storicallypal ita ask theyfrom were sentence also deriv5.80 edand, butpabe synchrol i give nicallyfrom theyfollowing are not exa mfurtherples: analysa. ble into smaller meaningful units. Consider the 202

(5.80) mi-pal ita c ira itakoan to [ caay-kangaay ask 3s ls not-eas'!/-one Nom (LCV (ACC :-e�rnsrgv (+AGT +LOC +PAT a[ demak] ] J J J question he asked me a difficult question (5.81) pabe l i kako ici raan (-an) to [ cacay a codad] ls 3s . one book, letter Nom (LCV ACC -ergvs (+AGT +LOC (+PAT ) I gave one/a bookJ to himJ Note that the stl:��em pall ita ask is further derived and marked by the prefix mi­ whichcorre spondingalso marks "spe sicific-omple accusbject"ative intra transitnsitiveive verbsverbs (s(seeee sectiosectionn 5.2 5.2.3.1.2) . ) and Transportation and Informationverbs are traditionally known as ditransitive ver"indirebs, charact obcterisedject" of byditra a directnsitive obverbsject hasand beenan indirect reanalysed obje byct . DeGuzmanThe so-called (1978: 51-54)(relabelled as having in �he the stu Locusdy as [+theLOC] Correspondent case relation [+ CORinstea]) dcase of therelation Dative wh [+ichDAT was] assiganalynedsis . to it in earlier lexicase grammars onthe basis of Fillmorean case Amis also shows a structural parallel between the [+LOC] actant of locative usesverbs toand juthestify so-called her treat "imentndirect of theob je"ict"ndirect of ditran objesitct" iveas [+verbsLOC] whichinstea DeGuzmand of [+IIDAT "spe] ci(orfic- [+oCORbject"]). intrThe aLOCnsitiv actantes, isof always ditrans spiteciveific ve inrbs Am, is.like that of class Class Va accusative transitive verbs consist also of a group of imperative verbsabove . whichThese corre verbsspond al toso havethe transportation the characteristic and in caseformation frame [+verbs[+PAT presented],+[+AGT� , +[und+LOersC],-tood[+, INisS]] reali of classsed in V the verbs genitive but the cairse co-occurringform. Unlike AGT theactantir non­, if not ofimpe nomrativeinativ countee AGTrpart. Theys, classfall Vatogether imperative with verbsthe so-called have nomi "lnativeocus focLOCus" inst verbsec.d descriFillmobedrean insu bjsectioect ncho 3.4.2.2.ice hiera rchyThey. areThese accusative verbs areverbs characte becauserised they by obeythe t.he thecaus atcliivetic pre formfix ofpa the- and se condthe impepersonrative singular suffix Genitive-i. Th Agentis suffix iso which-i is probably otheimperwisrativee is. nevCoernside explr theici tlyfollowing express edexam withples a: transportation or information (5.82) pa-todod-i henaca kako to c i laq (haw) ? p. s please ls salt QM �� (Nom ACC +trns l+LOC } (+PAT ) [+mprt-ergv would you please pass the salt to me ? 203

(5.83) pa-melaw- i kako tia cokcok [sa-pi -bohat let+V ...see� show lsNom keyACC used-for-opening +trns (+LOC } (+PAT} +mp-ergvrt iso tia panan ] 2sGen door (+AGT ( T} show meJ the key��� with which you opened the door (5.84) pa-sebanaq-i tia ngangan nia tamdaw te+vll � let ...know name of person +trns (���T} +mprt-ergv [ka-kil im-en iso ] looked-for 2sGen (+AGT teU me the name ofJ the person whom you are looking fo r As these examples show, imperative verbs do not necessarily stay within the showsame categidenticalory ofcase the irframe source feature verbss and. Thougare inh theagreement source' withand therespe derivct toed the form Canfeature be repreof ersentgativedit asy, foltherelows is: in this case a change in CF-CR mapping which Nominative AGT >+ (Genitive AGT) Locative LOC >+ Nominative LOC Accusative PAT >+ Accusative PAT This change of features in the extended case frame puts the -i imperative verbs inclass a diVaffe. rentReferring category to from the thcriteirerion corresponding that derivation source tyverbspically in classchanges III a or aresyntac derivti�ed class. (cf. DeGuzman 1978:133 and section 5.1) , these imperative verbs

5.2.5.2 � rgative Transitive Locative Classimperative Vb co nsiformssts �f mainly indi rectof indirect causative causa verbstive which ver bscan. beIt rouginchludely tras thensla ted as -"have X do Y to Z" . These verbs have the characteristic case frame [acta+[ +PATnt, ]if,+[ +AGTnot unde],+[ +LOCrstood] ,-, [+INSis real]] ofised class in theV verbsgenitive and thecaseir fo co-occrm. urringThey show AGT the"passi characteve" formsristic. Herecausativ are e aprefix few ex amppa-leands: the -en suffix for ergative 204

(S.8S) pa-pi-ciciq-en ko ising tina poceq-an ha+�e ...operate (Tdoctoraiwanese loan) this boi(Lcvl +trns ) l+LOC1 +m+ergvprt (:�:T havel the doctor operate on this boil (S.86) pa-pi-angang-en ci Dihang ici Holam-an have ...ca ll D �n Holam :�rns (�+PAT: j (LCV+LOC ) +mprt+ergv have[ Dihangl call/summon Ho lam (S.87) pa-pi-qonqon-en ci Pohay ic i ama-an h . urge Po hay Father :�:�� (Nom+PAT ) (LCV+LOC 1 +m+ergprtv [k[ i a l ka l amkam a [ta-lomaq ]] so-that Uick go-home +mnn+v r ,. l-fr+vint 1 have Pohay lgo urgeJ Fa ther to go home quickly

The -en imperative verbs, which can be translated as "have X do y to Z" , differ from the -i imperative verbs on five counts: (1) when both groups share the charact(2) Thougheristic they prehavefix the pa - samfore casecausative rela tioverbsns, , thethey map haveping diff of erCRent's to suffix CF'ses is. Acdiffcuserativeent forPatient these wh twoil e grou-enps impe: -irative impe ratverbsive haveverbs Nominative have Nominative Patien Locust and and impeLocativerative Locus verbs. are(3) accusativePursuant fromwhil ethe the diff -ener enceimpe rativein subjs ectare erchoicegativ, e.the -i (4) The -i imperatives are transportation or information verbs; the -en interimperativespreted areas an not "i. ndir(S)ec t Theob ject"focus, whofil -ie verbsthe focus is theof LoCusthe -en whi verbsch may is bethe "direct object" manifested by the PAT case relation. Historically, the -i impe(Starosrativeta, s Pacommutewley, andwit hRe verbsid, 198 wit2:7h 0)locat whilivee thefocus -en orimpe rerativeferentsial commute focus wit h verbsThe fo withllowing goal pair focus of exam(ibiplesd.:8 S)ser. ve Thisto furtheragrees withillustr obseatervation this point (2) above: . (S.88a) pabe 1 i - i kia wawa tia codad e child book (Nom 1 ACC Hrn,-ergv l+LOC (+PAT ) ql +mprt give:� the1 child the book 205

(5.88b) pabe 1 i-en kia codad iti a wawa-an (l e book child Nom LCV } +e+trnsrgv (+PAT } (+LOC l+mprt give:� the book to the child While the secondr person AGT is understood in these causative-imperative exprimperativessionse verbal, the addingconstr uctioof an n AGTwould in theresult first in ora non-im thirdpe peratrsonive tost anatem -enent . The following example shows an -en verb with Genitive Agent expressed longerexplicitl [+mpy. rt] Si(incempe rativthe Agente). is not in the second person, the verb is no (5.89) pa-pi-ngiciq-en aka cira tina qepah-an let+v ...share lsGen 3sNom this wineLCV +trns (+AGT ] (+PAT (+LOC l-m+ergvprt J J I let him share this (bottle of) wine Assp eciin ficexam objecples t.5.8 5-Moreove5.87, r,the allLocative these Locus-en verbs in examhaveple the 5.8ir9 sourceindicates in thea despeciriva·fic-otionalbje procect intrassesnsitiv relatinge verbs the ofse veclassrb classesII. We in shalChapterl deal 6. wit h the

5.2.6 Impersonal Intransitive [-[ Nom],-[+AGT]] ,Impefundamrsonalental verbsand obareli gatorysystem aticcase excerelatiptionson and to that our clallaim sentenc that ePatient construc istio thens , verbalare subj orectle non-verbss, theyal, haveare neveone rtandhe lessonly consone ideredsubject intra. ns Thoughitive clverbsass VIbe verbscause [+they[+PAT do ],+not[ +AGhaveT]]. the Evencharacte if weristic are wicasell ingfra meto ofrede trafinensiti trveansi vertivbsit, ywh asic hsi mpis ly beingverbs . agentive [+[+AGT]], this class would still not qualify as transitive meClassteor VIological impersonal phenom inenatra. nsiThetivese verbs impe rsonalare phenome verbsnal do notverbs ex hibitthat india sicatengle precharactfix whicheristic is affix shared. bySome so meappe of arthe in clathess root I and form cla; ssso meII comeintra witnsith ivethe ver mabs­ ; some are marked by the suffix -an which is also shared by class VII impersonal verbs; and some show the prefix si-. Since this class included underived root stitsems de asfini wengll chaas ractverbseristic marked lies by notthe dein rivmorpatioholognal icalaff ixesshape ma, -,but -an in, orthe si-,case framefor "phe specinomenficatal".ion [-The[ Nocontextualm], -[ +AGT ]]feature and the [- [Ninherentom]] im pliessemantic [-[ feature+PAT]], [+theph en] non-occto charaurrencecterise of a therela PATted caseset ofrel intraationnsi. tivThee phenomenalfeature [+ phverbsen] witcan hbe the used case also frame [+[ +PAT], -[ +AGT]]. 206 Examples of class VI verbs include: (5.90a) orad-an na-ci 1 a rain+v yesLCVterday -[ Nom (+TIM1 -[ +PAT] ] +phen-[ +AGT ] it rained yesterday

(5.90b) rna -orad na-ci 1 a rain+v yesterLCV day Nom (+TIM1 -[ +PAT] ] +phen-[ +AGT ] it rained yesterday (5.9la) bal i-an ( tOl ) potal wind-bZ-ow+V +Adv outsidLCV e -[ Nom +sptl (+PLC1 -[ +PAT] ] +inch +phen-[ +AGT ] +drtv-prft the wind is bZ-owing outside (5.9lb) rna-bal i ( tOl) potal wind-bZ-ow+v +Adv outside -[ Nom +sptl ( ���C] -[ +PAT] ] +inch +phen-[ +AGT ] +drtv-prft the wind is bZ-owing outside (5.92a) si-lelesian aca tepa r no Ta rawadaw flood+v again+AdV neighbourhood Hual.ien River -[ Nom (+rnnnr1 (���C] -[ +PAT] ] +phen-[ +AGT ] it is flooding again near the Hua lien River (5.92b) lelesian aca tepar no Ta rawadaw flood+v againAdV neighbourhood Hual.ien River - Nom (+rnnnrJ (���cl -[[ +PAT] ] +phen-[ +AGT ] it is flooding again near the Hualien River 207

toqman parod (S.93) dark kitchen +v-[ Nom ] [Ll+PLCcv -[ +AGT+PAT] J it+phen is dark in the kitchen tanglal henay (S.94) light still +v +Adv -[ Nom] +sptl -[ +AGT+PAT]] +inc-prfth it+phen is still light (outside) siqnaw (i) na-sangdeb (S.9Sa) co ld last-night �[-[ Nom] (��;MJ -[ +AGT+PAT]] it+phen was cold last night na-sangdeh siqnaw (S .9Sb) last-night cold +AdV +v (+TIMJ -[ Nom] -[ +AGT+PAT]] it was cold last+phen night outerIt was casement reionedlation earlier in the in casesec tionframe S.2 of that a verb the, prthoughesenc note or es absentiasence l ofto anthe subcategorfor its characteisationri ofsatio then. verb,For doesinst provideance, agentive secondar verbsy or withredundant more saturatedinformation notionalcase frames ti meuse and the pl verbalace; hence comple thementation absence deviceof PLC andfor TIMthe exprin theessirion case of frames. Non-agentivenotions with verbsthe oute, incr ludingcase relatio classns VI; heimpencersonal the pr verbsesenc, e canof PLCaccommo anddate TIM thein se thePLC iror casea TIM fr amcasees . relatioIn alln isexc preptesent S.94. of the examples given above, either a Thederivational co-occurring potential Place ofor beingTime forrea nalysedclass VI asim personala Patient verbs CR, thushas derivatthe ionally refollalowingting a pairs subjec oftless exam plesverb : to one with a Patient subject. Consider the 208

(5.96a) toqman itina lomaq da+vrk this house -[ Nom] (LCV+PLC ] -[ +PAT] +phen-[ +AGT] it is dark in this house (5.96b) toqman kina lomaq dark+V this house(Nom ) +[ +PAT] +�AT \ -[ � J [+phen+AGT ] this house is dark (5.97a) siqnaw it in i a remiad cold this day +V-[Nom ] (LCV+TIM ] -[ +PAT] +phen-[ +AGT ] it is co ld today (5.97b) siqnaw kina remiad co+Vld this dayNom +[ +PAT] (+PA1' +phen-[ +AGT ] J this day is cold Inreinter lexiprecaseted, only. Si innernce PLC case and re laTIMt�ons in the thatse examsubcategplesorise underg verbso rei cannter prbeeta tion viacase defrriamevation of class, they VI should verbs bewould regarded have asto beinne mor dificaseed relatioas [-[Nnsom. ],-[+AGTThen, ],the bet(+[ween+PLC ])inne, (+[r +TIMand ])ou].ter plaWhilcee weor haveloca tiocon,ns ideredwe have in not the comodelnsidered the disthe tinctioninner­ outerinterests distinc. tion for time. This is worth looking into for theoretical

5.2.7 Impersonal Transitive [-[ Nom],+[ +AGT J] I am reluctant to label the only co-occurring actant in the case frame of class VIIthat verbsPatient as ais Patientthe fundam evenental though case I amre latknowinglyion. My viola reasonting isthe that lexicase, there clisaim casejust nomark estaingblis systhedem asCF-CR pre sentedcorrespondence in Chapter known 4. asTo Ge niacctivepte Patienta Genitive in thePatient Amis Genanalitysisive wouldAgent createanalysis ev thaten more is adoptedserious problemshere, we . haveIf, the instea advantaged, we acceptof the tracaptunsiringtivit greatery and er genegativitralisay tito oncover by being the set able of to impe extendrsonal the ve notionsrbs. of 209 with a Genitive Agent analysis, class VII verbs can be considered both traco-occurringnsitive and Agent ergative and itbecause is typical it is fortypic eralgat ivfore trasyntaxnsiti tove showverbs an toacta havent a markedan Agent in when the Genit itco-occursive case wit formh a. traThatnsit Geivenit veiverb. NP Henceis usual we lycan iden apptilyfied the as evendefining though characte no Patientristics su bjectof trans is itivpreesent and. erAnotgativeher verbs gene raltois verbatio nclass we canVI I make suisbje thatcts, impeif theyrsonal had ver subjecbs, tsboth at traalnsitl. ive and intransitive, would have Patient Thein thechar followingacteristic ex ampaffixles for: impersonal transitive verbs is the suffix -an as (5.98) rakat-an no matoas-ay wa lk old-one +v-[ Nom ] -[ +PAT] (�:�] +[+t +AGT]rns +dlbr+ergv the old man [ de liberately ] walked (5.99) rakat-an nomako [ ta-ra ca-codad-an ] wa+vlk ls go schooLCV l -[ Nom] ��int (+LOC } -[ +PAT] (�:�J ( ] +[+t +AGTrns ] +dlbr+ergv ) [deliberately ] wa lk(ed) to school In these examplesI the contextual features [-[Nom]] and [-[+PAT]] are co-occurredundanringtly marke PAT. d. AlsoA su, bjecthe contextualtless ergative feature verb [-by[ +PAdeT],+finiti[AonGT]] cannot and [+havetrns a] classare redundan of impetlyrsonal marked ver bsto . sigTherenal the is extended no need useto redeof trafinensi thetivit featurey with [+ thiserg v] whichbecause is the eq uivdataalent support to the the contextual original anfeatalyuresis [-. [Nom,-PAT],-[Gen,-AGT]] Class VII impersonal transitive verbs such as rakat-an walk are associated with action verb. Understanding from our gloss, rakat-an seems to indicate chodeliceiberate, the actioperforn meron thechose part to ofwalk the insteaperfordme of,r. sayIt, meanstakin g thata ride, given. a Correspondingintrinsic sema nticaction feature verbs [+suchdlbr as] t-(deliem-akatberate wa) lkis arethere non-defore lipropoberatesed. to The furtheris placed chara on cterisethe actio thisn its clasself, notof suonbj ectlessany nomi venalrbs focus. with. A[+ condlbrtrasting], emphas is examis givenple with belo w.a [-dlbr] (non-deliberate) verb sharing the same root rakat wa lk 210

(5.100) r-em-akat [ta-ra ca-codad-an] waZk go schooZ +v+[+PAT ] ��int L+LOC� -[+AGT] [ ) ( -dlbr-trns J I wa Zk(ed) to schooZ Agaimarkedn, thein excontextualample 5.1 00feature, thiss time[+[ +PAtoT],- contrast[+AGT]] with and the [- trprnsevious] are examredundantlyples.

5.3 Subcategori sation in terms of Semanti c Features caseIn sec frtioamen 5.2featur wees have. Crarrivedoss-cla atssif sevyingen primary them wit verbh the classes features on ofthe erbasgatiisvi ofty. wecl asshave V increasedtransitives the into inven ergatorytiv eto andten accusativeby splitt ingsubcl classasses III. , SR-6class andIV , SR-and7 take(the sithemp sile mpintrale non-agnsitiventives), esclass in classIIIc I(t heand accusativsubcategorisee non-agentive them int o trclassansitiv Iae insinsttrumrumenentaltals)s) ., andIn classthe end IIId we (thavehe ertwegativlve e venon-agentiverb classes. transitive In this section, these twelve verb classes will be further subcategorised in matermsjor catof egoryinherent are semadeqanuatetic lyfeature accounteds until fo morpr. hologicalTo supplement variations the minimal within each wein formationshall also req useuired local to istickeep thefeatures syntac toto-mo furtherrpho logicalanalyse catelocatgoriivese vedistirbs. nct,

5.3.1 Simple Intransiti ve Verbs WeIn havesection list 5.2.1ed there we have sixteen seen diff thater entclass subgroups Ia is a withighhly des heterogeneouscriptive labels group and. willexampl bees pre. sentThoseed labein thisls are sectio relatedn. to the intrinsic semantic features that Intrinsic semantic features that can be used to further subcategorise simple actintraionnsit verivebs , verbsthose incluthat dedisti thosenguish that between disting quantityuish between and qualstateit, y,proc thoseess , thatand verbsidentif wity diffh impliederent kigeneralnds of orpsyc meteorhologicalologic veral bsphenom, andena those, instr thatument charac or terise etcacces. soryThese, product feature ors acqare uischoitsenion becau, locasetio, onn orthe mo onetio n,ha ndexis, theytence repreor posentssess ion, morphoimportantlogica conceptsl characte in theristics sema nticin Am domais. in and, on the other, they have distinct First, we can separate those verbs that indicate ambient or meteorological phenomenaThe [+phen fromJ verbs the are re stto ofbe classinter preIa withted as the having feature to [±dophen withJ all(phenom encomenalpass).ing atmospstate hericof indiv enviidualronmen obt,jec weatherts. or climate instead of, say, the quality or 211 The class of [+phen] verbs is still a heterogeneous group, consisting of four witsubgrh oupsthe feature. We can[±mb isolant] te(amb theie nt)ambient. Wh verbsile [+ mbfromnt ] theverbs me teorologideal withcal the verbs weatheencompassir. ngIn enviAmisronm, reentference such toas darclimatekness and or tempelightrature, the [-suchmbnt as] verbshumidit dealy, witheath orform coolne, likess someis handled[-phen ] byqu [alit+mbynt verbs] verbs which. Amb we ientshall verbs see beappelowar. in their root Theand [-fogmb. nt] Theverbsse verbsdeal withdo not me teformorological a morphologi phenomenacally homogsuch eneousas rain subgrou, wind, p.stor m, Rather, they are variously marked by -an , ma-, and si-. A careful examination meof teorologicalthe examples pheno revealsmenon the, ea factch hig thhligat, htsthoug ah speallcial three aspect forms of refer it. toFirst a , the si- verbs highlight existence or occurrence of a particular meteorological byphenomenon the feature at a[+ certainexst] (etixistencme or e)in . a certainAlso be placlonginge. toThey the can[+exst be characte] group arerised the existential verbs era there is and awaay there is not . In this context, theformir, howevePatientr, suisbjec rarelyt must used re ferin myto dataa meteor. Non-occurrenceological phenom ofenon cer. tain The negative meverbalteor olcomogicalplement events, as inis theinstead foll owingexpres exsedampl withe: caay not to be followed by a (5.101) caay henay ka [si-solda] ko Nataoran not yet[+Adv] have-snow NataorNom an in� int (+PAT ) it(:� never1 snows in Nataoran( �� ) From the gloss of my Amis examples, it is not so easy to tell the meanings of ma- and -an meteorological verbs apart. However, by referring to their usage elsewhere in the language, we find that ma- is mostly used with stative verbs whOf ilecourse -an , isthe found notion to occurof del withiberate class act VIIion doessubjec nottle ssapp delly iberateto meteor acoltogicalion verbs . verbs, but this other usage of -an suggests that it is not a stative affix. Hencegroups, . we emItploy is inthe the feature very nature[±stt v]of (sa tativmeteore) oltoogical distingu phenomenonish between to be the at twoonce a state and a process. In takes the ma- verbs to highlight the stative aspect Chaandpte ther 6,-an theverbsse two to higsubgroupshlight thealso prohavecess di aspefferentct ofde rivit. ationalAs will poten be tialshown. in Wefo llocanwing sum subcateup the gorisatsubcateiongo risrulationes: of class Ia verbs so far with the SR-9 +[+V +PAT ] -[ +AGT] [ ±phen] phen "phenomenal" --[[ +I+LOCNS] SR-IO [ +phen] [ ±mbnt] mbnt "ambient" SR-11 [ -mbnt] [ ±exst] exst "existence" SR-12 [ -exst] [±sttv] sttv "stative" which can be represented by the following feature tree: 212 +V +-[[ +AGT]+PAT] -[ +LOC] -[ +INS]

-phenI · I (FigureI 5.5b) -mbnt +mbnt -exst +exst � -sttv� +sttv characteristic affix (c.a.) : -an ma- si-

orad-an ma -orad si-orad toqman rain rainy occur-snow dark bal ios-an ma-ba 1 i si-kotem tanglal typhoon windy occur-cloud "light lelen-an ma-kotem si-l elen soqmed earthquake cloudy occur- humid earthquake

Figure 5.5a Subcategori sation of Cl ass Ia Verbs in terms of Inherent Semantic Features ; with examples Thsubjeis ctlessfour-way phenom analysisenal verbsis repe ofated class in VIsectio. n 5.3.6 to subcategorise the heterNow weog eneousturn to, eventhe [- afphenter ] theverbs [+ phofen class] verbs Ia. have Thspislit group off . remVerbsains higinhly th is regroupflected cover in many at leastfeatures ten diffin theer entseman morptichological domain. formsThis of wide aff coverageixation, namis ely, witthe h rootma- , form-em-. , ta-, masa- , ma l-, mi-, misa-, si-, partial reduplication and First we can subcategorise the [-phen] verbs into [+sttv] (stative) and [-sttv] seman(non-stictative features) verbs (s, eewhich Figure are 5.5inb turnand 5.subca5c tegresporisedectively by ).a number of inherent Amor ongpossess the stionativ. e Alsoverbs, thereare verbsare verbsthat thatindi catedeal qualwithity psychol, quantogicality, exist and encothere, physical states. There are also the si-N verbs that are derived from nouns whichqualit ycan embodim be interent.pr eted as the implied objects of existence, possession, or 213 Wenu mbeuser verbsthe feature. This [± quansubgroup] (quant in ity)turn coto nsistsisolate of a verbs subgroup in the of irquan roottitativ forme andor verbs obtained through partial reduplication of the rootc form. The former qutypealit seemsy of theto hig numhligeralht. existWe encecan usewh ilethe thefeature latter [± exfostuses] (e onxi stencthe e)desc toriptive keep them distinct. Then the feature [±imob] (implied object) is' introduced to canseparate also bethe furthersi-N verbs analysed from withthe rest the offeature the [- [±quanexst] ] ver(ebsxistenc. Thee) , si-witN h verbs[+ex st] onverbs a detakingscriptive on anread existingential. Non-q oruant possessitativeive verbsreading without and [- eximpliedst] verbs objec tatsking include ma- verbs and some root forms. Both types are descriptive and both types depict physical states or qualities though the ma- verbs also depict psycand hothelog rootical formsstat esare. notIn wantfurther of analyseda good semant hereic. correlation, the ma- forms Thesemantic subc ategfeaturesorisa tioncan beof summthe [-arisedphen ] inst theative fo verbsllowing in subcateterms gofori inherentsation rules: SR-13 +v +[-[ +AGT+PAT] -[ +INS+LOC] -+ [ ±quan] quan "quantity" +s-phenttv SR-14 [ -quan] -+ [ ±imob] imob "implied object"

SR-15 [ +quan] -+ ([ +imob] ) [ ±exst] exst "existence" statesMoreover, with the the ma- featureverbs [±capschn be ] further(psycholog analysedical) . intoThe psychol [+imobogical,+exst and] si- phN ysicalverbs empcanhas be isesfurther exist anencealysed or occurrenceinto a subgroup. The that fo rmehigr hligsubgrouphts po co-occursssession andwit hone that [+subcateganmt] oris(animateation) sucanbject be stat and edthe in lathett er,fol pllowingace or rul locates: ion. This further SR-16 "ttv [-i-quanmob l -+ [±psch] psch "psychological" SR-17 +imob -+ ( -exst) [±pssn] pssn "possession" Thesynt acticlast two or morpsubcatehologorisgicalation conseq rulesuenc arees. included even though they do not have non-phenomThe followingenal featureverbs intree ter mssums of up inhe therent subcateg semanticori safeaturtion esof . class Ia 214 (Figure S.Sa) - -

- - ++v+PAT] -[ +AGT] -[ +LOC] -[-phen +INS ] -sttv ------+sttv I I I (FigurI e S.Sc) -quan -imob +imob -exst +exst �� A -xtrn +xtrn -exst +exst I I � -pAn +p n I c.a. : ma- si- si- si- r r e.g. : rna-orip langdaw si-ngangan si-I okedaw ta-tosa tosa alivI e blue� famousI exist­I two two green wildcat rna-olah kapah si-pida si-nan i I a-I irna lima happy good rich have-cat fi ve five rna-toas bangca l si-kawas si-wawa awaay grown-up pre tty haunted have-child none

Figure 5.5b Subcategorisation of Class Ia [-phen ] Verbs in terms of Inherent Semantic Features ; wi th exampl es We are considering the feature [±xtrn] (extrinsic) for distinguishing intrinsinsic . Whandile extri physicnsical qual qualitiesities. areFor extri instansnceic, , psychopossesslogicalion can qu beal itiesconsi dearered intrins extr ic. Still the semantic distinction between the root forms and the rna- stative verbs is not very clean-cut, because rna-orip alive can be interpreted as a physical qucontalitraryy whtoil ethe kapah feature good marking can be givenconsider ined Figure evaluative S.Sb. and hence psychological, Amongor dispthe osal no n-ofstat an imiveplied verbs ob areject verbs, verbs that that indica inditeca tethe group production activity, acq, verbsuisi tion thatinvoluntary represent act movemention. or motion toward an implied goal, and verbs of 215 First, we use the feature [±vlnt] (voluntary) to separate the involuntary obactjeionct) verbsis used from to thebrea rek stdown of the [+grouvlntp. ] verbsThen , intothe twofeature big [±cateimobgor] ies(imp: litheed [+consimob]umpt subgroupion, and whichassoci ationincludes or companverbs ofion produc; and thetion [-, imobacqui] sisubgroution, pdi whichsposal or andincludes motion verbs with ofim pliedgroup pagoalrtici. pation, mutual or reciprocal action, movement, The [+imob] verbs show three different affixes: mal-, mi-, and misa-. The mithesa- productN verbs imareplied verbs in thethat sourceindica N.te creativeThe mi- Nor verbsproductive are deriv activited fromies, N's wit hthat conare suafmedfected as a by result the activ of theities activit. They. impliThe edma obl-Nje verbscts can are be deacqurivedired from or N's Thethat suarebject partne of ars ma inl- Na vegrouprb isactivit alwaysy plurinvolalving. Thetwo orfeature more [±partprdticipa] (pntrods. uct) Thenis used the tofeature separate [± affthet] mi(asa-ffecN ted)verbs isfrom used the to othe keepr distincttwo [+imob] the sumi-bgrouN verbsps. fromfeature the [±maacqul-N ] ve(arbscqui. siThetion) mi- forN verbs the purpose can be offurther detailed subcateg semaorisedntic by the interinconsequpretatentiaionl., but, morphologically speaking, this subcategorisation is The [-imob] verbs have four different affixes: -em-, ta-, masa-, and mal-. Siindicatmilare togroup the maactivitiel-N verbss. in Whenthe [+thereimob are] subgroup only two, thepartic mal-ipaverbsnts inhere an also masactivita-. y whichIf we chooseinvolves not mutual to pay or attentionreciprocal to actithe onnu, mbether verbof parti is markcipantsed by and wherethe gre [+gagruprious] (gnatureroup) ofrepre thesent activities group s,parti we wouldcipation have and the [+ [-rcprgrup] ,-(recrcpripr] ocveal)rbs , affstandsixes for: ta-reciprandocal -em- act. ionSince. theThe [-ta-gruppre,-fixrcpr combin] verbses withshow atwo noun diff stemerent to formfeature a mo[±tiongoal verb] to withseparate implied the goalta-N , verbswe can fromuse thethe -em-locaverbslistic. semantic Wesubcategor can sum isupation our analysrules: is of class Ia [-phen,-sttv] verbs with the following SR-18 +v +[-[ +AGT+PAT] -[ +LOC+INS] -+ [±vlnt] vlnt "voluntary" -stt-phenv SR-19 [+vlnt] -+ [±imob] imob "implied object" SR-20 -+ [ ±prdt] prdt "product"

SR-2l [ -prdt] -+ [±afft] afft "affected"

SR-22 + lnt -+ (-l:'mob;] [ ±grup] grup "group" SR-23 [ -grup] -+ [ ±rcpr ] rcpr "reciprocal" SR-24 [ -rcpr ] -+ [±goal] 216 and two redundancy rules: RR-6 -prdc -afft ] [ +grup] ( +motn motn "motion" [+goal] (+drcn�J drcn "direction The two redundancy rules introduce redundant semantic features. In Amis, a [-neceprdcssarily,-afft a] diverbrectional is a group motion verb verb. . And, in general, a [+goal] verb is Thexamisples analys foris ea canch alsocateg beory shown. as a feature tree in Figure 5.5c below, with (Figure 5.5b) +v ------+[-[ +AGT]+PAT] -[ +I+LOC]NS] -st-phentv -vlnt +vlnt -imob +imob �--- -grup +grup -prdt� +prdt � -rcpr� +rcpr -afft +afft -goal +goal I� I c.a. : ma - -em- ta- masa- ma l- ma l- mi - mi sa- I I I I e.g.: ma-kelol ta-lomaq ma l-taes mi-kasoy abate (of go -home hit/fight­ fa ther­ wind) one­ fi rewood another t-em-ang ic masa-soa l ma l-kaka misa-kabi cry, weep converse be-with­ make-soup elder­ sibling ma- tepo ta-ini ma l -qepah mi-bot ing fa ll come drink­ catch-fish together k-em-aen masa-cepcep ma l-wina misa-tamdaw eat kiss be-with­ make -friends mo ther Figure 5.5c Subcategorisation of Class Ia -phen ,-sttv Verbs in terms of Inherent Semantic Features ; with examples 217

5.3.2 Intransitive Locative Verbs In section 5.2.2 we have shown that 'class II is also a heterogeneous group. It included locomotion verbs such as ta-ngasa go-up-to in example 5.40, verbs indicating existence, location, possession, or number such as era there is and awaay there is not in examples 5.42 and 5.43, and the specific�object inintr examansipletives 5.4 such8. Weas mi-canalo used thethrow-at inherent in exam semantple ic5. 44feature, ands ma[±-tallctnaw] af(lraocidation) of and [±motn] (motion) to set up a thr.ee-way distinction as follows: +v +[-[ +AGT+PAT] +[-[ +I+LacNS] -lctn +lctn -motn� +mo tn

e.g. : mi-alod ma-talaw era I ta-tosa ta-ngasa throw-at afraid-of there is two go-up-to mi -ki lim ma-olah awaay I a-I ima ta-ini look-for love there is not five come, arrive

Figure 5.6a Subcategori sation of Class II Verbs by Inherent Semantic Features; wi th exampl es The first' group, namely, the specific-object intransitive verbs characterised by [-lctn] include a subgroup of action verbs marked by mi- and a subgroup of [±psychopsch]logi (pcalsycho verbslogical marked), we by can ma dis-. tingUsinguish thebet weeninhe rentthese semantic two subgrou featureps. The second group, i.e., the [+lctn,-motn] verbs, includes era there is and awacan aybe there either is anot place in orthe air person exist entialand it andde lilocatimits vea unusiverseage. inThe which LaC actantthe entitiesexamples indicatedshow era used by the respe PATctively exist witor hdo a notpl aceex istand. withThe a folpersolowingn: pair of (5.102a) era ko adiwawa potal there+v are chilNom.dren yardLCV -trns (+PAT J (+LOC1) l+l-motnctn therer are some children in the yard 218

(5.102b) era ko adiwawa itakoan there(+V are children lsLCV ) +l-trctnns (:�:T1 (+LOCJ t -motn there are some children with me verbsIn the thatfol lowingindicate pair num ofber exam as plweesll, tosaas exi twost encande witta-thinosa a twoun iverseare inter definedpreted by as LOC. These verbs are also members of class II [ +lctn,-motn J verbs. (5.103a) tosa ko ayam dipong two+v biNomrd nestLCV -trns (+PAT J (+LOC J [+l-motnctn there are two birds in the nest (5.103b) ta-tosa ko ayam dipong two+v birdNom nestLCV -trns (+PAT (+LOC +lct-motnn J J there are two birds in the nest Since awaay there is none can be interpreted as indicating the number zero and era there are some as indicating an unspecified number n, they can be looked uponcan be as furtherspecia l charactnumber erisedverbs . by Conceptthe featureually of they exi formstence a : natural group which RR-8 +v -trns -+ [ +exst J exst "existence" +lct-motnn byHoweve a morphor, sincelogical the cardinalfeature whichnumbers I shacanll be ter redum [+plicatednmrlJ (nu, mtheyeral) should to set be themmarked apart from era and awaay . The first consonant of a [ +nmrl J verb can be reduplicated with a vowel a inserted after it. These [ +nmrl J verbs correspond to the [ +nmrl J nouns described in section 3.7.2 under Increment. The third group, i.e., the [ +lctn,+motn J verbs are locomotion verbs. These verbsfurther are subcat uniquelyegorise charac thisteri groupsed bymorph theolog preicalfix lyta. -, Howeveso therer, isthese no needlocom otiorto verbswhich cansupplement be, and the wi llminimal be further information characterised carried by by local the isticlocativ semantice determine featuresrs. Such verbal localistic features include [ ±lctn J (location), [ ±drcn J (direction), wit[ ±goalh this] (g oainl) se, ction[ ±sorc 5.3.8.] (sourc e), and [±assn] (association). We will deal Subcatesemanticgori featuresations canof class be sum IIma risedintransi in tivethe follocativelowing verbs subcateg in ortermsisa tioofn intrirulesn�: ic 219 SR-25 +[+v +PAT ] -[ +AGT] -+ [ ±lctn] +[ +LOC ] -[ +INS] SR-26 [ +lctn] -+ [±motn]

SR-27 +lctn -+ ( -motn ] [±nmrl] SR-28 [ -lctn] -+ [ ±psch] and represented by the following feature tree: +v +[ +PAT] -[ +AGT] +[ +LOC ] -[ +INS] -lctn +lctn -psch +psch -motn +motn � � -nmrl� +nmtl c.a. : mi - ma- Cla-I ta- e.g. : mi-alod ma-talaw era ta-tosa ta-ngasa throwI -at afraid-ofI there is two I go-uI p-to mi -ki 1 im ma-olah awaay la-l ima ta-ini look-for love there is not five come, arrive

Figure 5.6b Subcategorisation of Class II Verbs by Inherent Semantic Features ; wi th exampl es Please also refer to the feature tree in Figure 5.6a above.

5.3.3 Simple Transitive Verbs Classaccusat IIIive verbstransitiv have esal, readyand class been suIIIbcategb, theorised simp leint ero gaclasstive IlIa,transitiv the essi, mple [+according[+PAT],+ [+to INtheS]] feature to sat isf[±ergy thev]. de finitiSinceon we ofal tralownsi eittiveher [+ver[bs+PA, T],we+[ can+AGT si]]ngle or outclass the III insb totrum formental class transitiv IIId. es from class IlIa to form class IIIc, and from 220 In addition to syntactic consequences, especially that of subject choice, ersubjectgativity of anhas ersemagativntice ve imrbpl carriesications wit as hwe itll a. defi In nitpare ticulareadingr, , theexcept Patient when it Theis anAccusat existentialive Patient verb ofsuch an asaccusa era thtiveere verbare , someon thein otexamherple ha nd5.1, 02ais andto be 5. 10��b. intformuerprelatedted inas thebeing follo non-spwingecif redundancyic. Such rul essemantic: characterisations can be

RR-9 [ +dfnt ] I[ -exst] exst = "existence" RR-IO prnn "pronoun" [ +spfc ] prsn "person" splctfcn "s"lpeciocation"fic" RR-ll [ -spfc ] -prnn [-lc-prsntn ] RR 12 [ -spfc ] - - These rules also sum up our earlier( (discuss-� ] ions on Nominative Patient (section 3.1.2.1) , Accusative Patient and non-sp�ecific�) objects (section 3.1.2.2 and marked3.1.3.1 as) . AccusativThese rue lesPatient guarantee. that personal nouns and pronouns never get

5.3.3.1 Simple Accusative Transiti ve NomiBothnat classive AgentIlIa andwh ileclass class IIIc IIIc belong has Nomito thnativeis group Instr. umeClassnt. IlIa has By the examples given in section 5.2.3.1, class IlIa is a homogeneous group. Its members share the characteristic prefix mi- in addition to the cornmoncase frame [ +[ +PAT ] ,+[ +AGT ] ,-[ +LOC ] ,-[ +INS ] J. The Patient actant that is realised incas esthe, Accusativan indefine itcasee ob formject . is Sitonce be theunde subgrouprstood as has a non already-speci beenfic and unique, in lymo !;t characfurtherterised subcateg byori thesatio concurrencen is nece ofssary case. frame and morphological features, no As for class IIIc, which is characterised by the case frame [+[ +PAT ] , -[ +AGT ] " -[ +LOC ] ,+[ +INS ]], its members are shown to be variously marked by sa-pa-, markingssa- pi-, anddi ffermami only- in bythe viexamplrtue ofes derivatigiven in onalsection sourc 3.3.2.1es. In. otheThe r firstwords two, thE!y boil down to the prefix sa- indicating instrument used to perform the action implied in the source verb. The difference between sa- and the prefix mam i -­ diffcan beer enceaccounted can be for form byulat theed sema in nticthe fol featurelowing [ ±asubcatenmt ] g(aorinimatsatioe)n. ruleThis: SR- 29 +v +[ +PAT ] -[ +AGT ] -[ +LOC ] -+ [ ±anmt ] +-e[ +IrgvNS ] 221 Thefol lowingsubcateg featureorisat treeion of: class IlIa and class IIIc verbs can be shown by the

-[ +INS J +[ +INS J (+[ +AGT J ) (-[ +AGT J) -anmt� +anmt c.a. : mi- sa-I mamI i- e.g. : mi -a la sa-pi -angang mami-asik get�I rescue used-foI r-calling assignedI -to-sweep mi-ki 1 im sa-pa-ahcid look-for used-for-making- something-salty

Figure 5.7 Subeategorisation of Class IlIa and llle Simple Aeeus'ative Transitive Verbs ; wi th exampl es

5.3.3.2 Simple Ergative Transitive AgentsBoth class while III clab andss IIIclassd has III Gend beitivlonge Instr to thumentis groups. . Class IIIb has Genitive Class IIIb is not morphologically homogeneous. As has been presented in section 5.2.3.2, these verbs can be marked by either ma- or -en. While some verb stems combine with both ma- and -en to form new verbs, others show only one form, with either ma- or -en but not both, in the lexicon. There is no theapparent stem , indiwhycat it ionhap, penswhether this byway. morpho Onelogical plausible shape explanation or by seman ticI can content offer ofis that"pass theive"se andare anproducts "anti-pas ofsi twove" compederivtingation deriv, ifational indeed bothproce ssesthe -en, sa y,forms a and the ma- forms are derived. Amis, being a split ergative language, has both ergative verbs and accusative verbsThough inthe its pa sssystiveem verbs. Accusativ have fale lenverbs together have both with "atruectiv e"er gativeand "p verbsassiv e"in formsthat . arethey marked have indi commonfferentlyNomi fromnativ thee Patienttrue er andgativ Geesni. tiveIn Ag Amients,, pathessi passve verbsive verbs are markedthe two by groups -en wh couileld true be formergativeulateds byare themarked following by ma-. sUbcate The godistirisationnctio nrule between: 222 SR-30 +[+v +PAT ] +[-[ +LOC+AGT]] [±pssv] pssv "passive" +erg-[ +INSv ] The feature [±pssv] (passive)" is not really an inherent semantic feature becaunot addse , anythingother than to ouran indiundecationrstanding of the of vertheb's mean deingrivational of the verbhistory. We, itcan doe�: say erthatgativ thee [+vepssvrbs. ] verbsAs we are do notderived allow wh ilethe usethe [-cofpssv rule] verbsfeature ares inbasic lexi undecaseriv, theed indistingfeature [±uipssvshable] .shou exceptld be forexclu theded.ir diffThenerent the affix -en esverbs. and the ma- verbs are obThejec mat -whverbsich isof co class-refer IIentialIb consis witht ofthe a Patientsubgroup su bject.of verbs Examwithples an iminpliedclude ma-adop hunted-down from example 5.54 and ma-oay gathered-rattan from example 5.57. As has already been pointed out in section 5.2.3.2, these ma-N ergative verbstransit iniv eclass verbs IlIa. syst ematicallyIf we accept corre "ispondmplied to obtheject" mi- Nas acancusa inherenttive transi semantivetic thefeature follo ofwing the sesubcateg verbsori, wesation can setrule them: apart from the rest of class IIIb by SR-31 +v +[ +PAT+AGT] -[ +IN+LOC]S] [ ±imob] imob "implied object" +erg-pssvv Theor loc maomo- verbstion ofeven cla thoughss III bno also LOC incluis presentde a subgroupin their of case verbs fram thates. implyExam plemotions exainclumpdele ma-dak5.61. awWe cliwouldmb-over use fromthe semaexamnticple 5.feature60 and [+ mamotn-dang] oy(moti swim-acron) tooss set fr theom �e beverbs form apartulat edfrom in the refolstlo wingof class subcat IIIegb orisaergativetion ruletransit: ive verbs. This car SR-32 +v +[ +PAT+AGT] -[ +INS+LOC] -+ [±motn] +ergv-pssv The ma- motion verbs can be further characterised by localistic features to willprovide deal more wit hin thformationis in sect on ionthe 5.orientatio3.8. n and direction of the motion. We Class IIId ergative transitive verbs with Genitive Instrument are marked by the characteristic prefix ma-. Since this subgroup is uniquely characterised by necescase sarframey. as Itwe llis asintere morpstihologicalng to po featureint out , thnoat furtherthe Geni subcategtive Instorisatrumention canis be a natural force, such as betil i lightning from example 3.61, an abstract quality such as bangcal beauty from example 5.62, a physical object such as 223

anaboq dust from example 5.63, an action such as selaq snoring from example 5.65, or even a person such as nomiso "2s" in example 5.66. Except when the meaINS ningindicates. To aset nat theural st ageforce for, the subsequent ma- verbs sema haventic a psychinterprologicaletation or, the cogn itive inherent semantic feature [ ±psch ] (psychological) is used to differentiate the psychforceological. Hence reactthe foionl lowingto a st suimulusbcateg orisatfrom a ionphy sirulecal: response to a natural SR-33 +v +[ +PAT ] -[ ] +AGT -+ [ ] -[ +LOC ] ±psch +e+[ +INSrgv ] INS indicates a natural force if the verb is marked [ -psch ] , but if the verb is marked [ +psch ] , it does not co-occur with a natural force INS. Ourthe anafollowinglysis of tree class di agramIII si: mple ergative transitive verbs can be summarised in

-[ +INS ] +[ +INS ] (+[ +AGT ] ) (-[ +AGT ] )

-pssv +pssv (+f-pschorc) (-+pschforc ) -irno �mob � -motn �+motn � ma-me law ma-dakaw me l aw-en ma-apa see I climbI -over watch� care­ dumbfounded fo r ma-ala ma-adop ma-asek taken (by hun ted-down struck (by lightning) mi-dangay cebar-en ma-canar swum-across spread-out disturbed ma-as ik ma-oay ma -nanoang swept gathered waved (in the rattan wind)

Figure 5.8 Subcategorisation of Class IIIb and IIId Simple Ergati ve Transitive Verbs; with examples 224

5.3.4 Transitive Instrumental Verbs Class IV transitive instrumental verbs have the characteristic case frame [ea+[sy+PA dueT ] ,+to[ +AGT the ]scarcit,+[ +INSy]] of. examSubcategples. orisatWhateverion of weth haveis group shows of greatverbs is not heterdistinctogeneit functiony in morp to hologicaleach form . shapes. Here we try our best to assign a As shown in section 5.2.4, class IV transitive instrumental verbs can be suIVbbca, tegin orisedterms of int ero gativityan accusativ. e subclass, IVa, and an ergative subclass,

5.3.4.1 Accusative Transiti ve Instrumental In section 5.2.4.1 we have established class IVa as an empty set. Looking at the two candidates for class membership, namely, mi-pinaro fi ll and satwo-pi disti-angangnct usedsubcateg fo rories calli ngfor, weclass can IVanev. ert Wehe lessknow make by analogyan at�empt with tothe set up Nominativenominalised Agent form , nian-pi Acnarcusato fiivell ingIns trumthaten a t,cla andss anIVa Accusat mi- veriveb shouPatientld have (s eea Genexampitivlee 5.Agent71). , andThe ansa-p Accusativei- verb hasPatient instea (sd eea Nomexaminatple ive3.5 6)Ins. trumenMoreovet, a r, we know from the examples in section 3.3.2.1 that the prefix sa- is typical of deri(instrved um entainanimatel) to instrcross-umclaentalssif yusag thee.se twoWe verbscan use. theThe semadistinnticctio featuren between [ ±nst t.he.r ] mi-ruleand: the sa-pi- verbs can be formulated by the following subcategorisatior. SR-33a +V +[ +PAT ] [ +[ +AGT ] -+ ±nstr] +[ +INS ] t -ergv It is to be noted that in this context [ -nstr ] implies that the referent of the INS"imme actantdiate" iscause the meansof the orevent "inte indirmediatecated " bycause the verbrather. thanSinc ethe the inst starumetus ntof or the twosubca verbstegori insat reionlat anyion tofurth classer. IVa is uncertain, we will not pursue the

5.3.4.2 Ergative Transitive Instrumental In section 5.2.4.2 we have seen examples for two subcategories of class IVb: Gethenit -enive verbsInstrumen co-occt. urringBoth withtype Accusatives have Nom inativInstrumente Patient whil eand the Ge manitiv- verbse Agent show. meansSeman ticallyor "inter speamediateking, " thecause Accusat of anive eve Insnttrument, as oppo issed to tobe aninter instprerumetednt as wh theich is the "immediate" cause. We can use the feature [ ±nstr] (instrumental) introduced in the previous section to distinguish between the ma- and -en meaverbsns. [Perhaps-nstr ] animplies inner-outer that the disti co-occurringnction may INSprove is to be inta moreerpre getedneral as the approach. For the time being, the same distinction is marked as [ +nstr] and [ -nstr J. 225 Weverbs can of sum class up the IV b cawithse frame the foandllowing semantic subc diffategorierencessation bet ruleween: the ma- and -en SR-33b +V+[ +PAT ] +[+AGT] -+ [ ±nstrJ +ergv+[ +INS ] whichstatement can beabout comb insinedtrum withent s SRand-33a mea inns the: last section to make a more general SR-33 +[ +PAT+AGT]] [ ±nstr ] nstr "instrumental" l +[ +INS] Wethe canfeature sum up[± nsthetr r:[cr] oss-classifiin the followingcation feature of class tree IVa: and class IVb verbs with +[+V +PA T] +[ +I+AGT]NS]

-nstr� +nstr -nstr� +nstr c.a. : miI - sa-pI i- -enI ma-I e.g. : mi-pinaro sa-pi -angang tomes-en ma-bahbah fi ll used-for-calling fi ll-up drive-away ? ? Figure 5.9 Subcategorisation of Class IV Verbs in terms of Inherent Semantic Features ; wi th exampl es

5.3.5 Transiti ve Locative Verbs From section 5.2.5 we have seen that class V [+[+PAT],+[+AGT],+[+LOC],-[+INS]J verbsaccusativ are e crosand s-classian ergatfiedive witsubgrouph the . featWeure can of app ergativly locityalistic and yield semantic an spatfeaturesial oritoentat furtherion characteand direcrisetion th canese be verbs dis coveredso that to more supplem informationent the minion mal inffeaturesormation for carried locative by verbsthe ne willutral be locatspeciallive dey dealtterminers with. inLoc sectalisticion 5.3.8. semantic 226

5.3.5.1 Accusative Transiti ve Locative Classforms Vaof accusativtransportatione transit and ivein folocatrmationive verbsverbs . consiThesests verbs only ofare declmarkedarative by the prefix pa- in most cases, but as the examples in section 5.2.5.1 show, fossilised forms with pa- incorporated in the stem can either appear in the root form, such as pabe l i give or with the prefix mi-, such as mi-pal ita ask. Weverb can im pliessubcategorise the· moving class or Vatransportation with the feature of a ph[±motnysical] (motioobjectn) in. space A [+motn. A] [-ormotn transmi] verbssi onis anof ininforformationmation verbfrom personin this A contextto person. B.It impliesClass Vathe verbsimpa rtingare mostly marked by pa-, but there are also verbs that appear in their root form or with the prefix mi-. This is to be expected because a class normally contains some derived and some underived members. Some forms like pal ita ash and pabe l i give probably were derived forms which got reanalysed as a root in the coursefollowing of hirulestor: y. We can sum up the subcategorisation of class Va with SR-34 +V+[ +PAT ] +[+[ +LOC+AGT] [ ±motn] -e-[ +INSrgv ] and the following feature tree, with examples: +V+[+PAT ] +[ +AGT] +[-[ +I+LOCNS]

(+i-motnnfo) (-+motninfo) Transportation InformationVerbsI VerbsI

e.g. : pa-sebanaq pa-osa tell I de livI er pa-ceqob pa-ta-ra answer send mi-pal ita pabe 1 i ask give

Figure 5.10 Subcategori sation of Class Va Accusative Transitive Locative Verbs ; wi th exampl es 227 From section 5.2.5.1 we know that the imperative formsof class Va oftransportation accusative verbs and in haveformation similar verbs CR's also but bethelong impe torativ classes areVa. diffTheerseent two from sets thsuffixeir decl is arativeadded to sourcethe verb in thstemat theas inAgent the actafollowingnt is exaundersmplestood: while an -i Va DECLARATIVE Va IMPERATIVE pa-sebanaq pa-sebanaq-i tell. pabe 1 i pabe 1 i - i give pa-osa pa-osa-i deliver Thecan besuocateg statedorisation as follo wsof : class Va verbs with the feature [±mptvJ (imperative) SR-35a ++[V +PAT ] +[ +LOC+AGT]] [±mptv] -e-[ +IrgvNS ]

5.3.5.2 Ergative Transitive Locative Fromconsist secti maionnly 5.2.5.2 of indir we ectknow causativ that classes. VbVerbs ergative in th trais classnsitiv eare locativ uniquee lyverbs charactypicalteris of edall by erg theative caus verbsative (spreee fixFigure pa- plus5.2 inpi- sectioand nthe 5.2 -en) . suffix which is [±Likemp tvcla] ss(impe Va rativverbse), . classThis Vb can verbs be stacan tedalso as befol subcateglows: orised by ·the feature SR-35b +V+[ +PAT ] +[ +LOC+AGT]] [±mptv] +ergv-[ +INS ] With class Vb imperative verbs, the Agent actant is also understood. With Seetheir exam non-plesimpe 5.rative85-5.8 counte9 in secrparttions, 5.2.5.2. the Genitiv e Agent is overtly expressed. SR-general35b canstat beement combined about with class SR- V 35aloca intive the vepreviousrbs. section to yield a more SR-35 +[+V +PAT ] +[+AGT] -+ [±mptv] +[-[ +INS+LOC] Thatfollowing the Agent rule : actant is understood for imperative verbs can be stated in the RR-13 [+mptv] [¢[ +AGT]] 228

The ¢ is different from "_" because the latter would imply non-co-occurrence. The cross-classification of Class V verbs with the features [±e'rgv] (ergative) Figureand [±mptv 5.11] belo(impew.rative ) results in the four-way distinction as shown in +V +[ +PAT]+AGT] +[-[ +I+LOCNS]

------m tv +m tv -m tv +m tv r�r r�r c.a. : pa- pa- . . . - i pa-pi- ...-en pa-pi- ...-en

pabel i pabe 1 i - i pa-pi-ciciq -en pa-pi - cic iq- en e.g. : give give.' have X opeY'ate have X opeY'ate on Y on Y.' pa-sebanaq pa-sebanaq-i pa-pi-angang-en pa-pi -angang-en telZ. tell! have X calZ. Y have X calZ. Y! Figure 5.11 Cross-classification of Cl ass V Transiti ve Locative Verbs wi th the Feature [±mptv ]; with examples

Referring back to SR-34 in the previous section, we know that [-ergv] verbs Gan alsthe o twobe subcategfeaturesorised yield byfour the subcateg featureories [±motn] for class(motio Van) ., exemTogetherplified with by theSR- four35, examthe plesderivation in Fig ureof impe5.11.rative We forms.shall refer to Chapter 6 (section 6.3.2.5) for

5.3.6 Impersonal Intransitive Verbs verbsIn sect areion phenom5.2.6enal we have ver bspointed, i.e. , outverbs that that class indic VI ateimpe amrsonalbient orintran meteorsitiveological phenomfeatureena sp. ecificaThe tionfeature of [+thphenis class] (phenom by theen al)fo llocanwing be redundancyredundantly ruleadded: to the RR-14 + (�[+PAT-[ +AGT]l] [+phen] Morphologically speakingma, class- VI is a hete-an rogeneous groupsi-, with its member:; mayvario appeuslyar marked in the by root the form. prefix, the suffix, or the prefix; or they

-� 229 Class VI subjectless verbs have their parallels in class Ia simple intransitive veforrbs the, wherefour thegroups corre ofspond subjecingtless verbs phenom take enalon a verbsPatient is suthebject.refore siOurmilar analys to is thatsect ionfor 5.3the.1-corresponding) . intra- nsitive verbs in class Ia (see Figure 5.5a in First of-all, we can separate the affixed forms from the root forms with the envifeatureronm [±entmb ntsuch] (ambas darkient)ne. ss Aor [+ ligmbnhtt], orverb atmosp has herto doic wittemperatureh the en comofpass hqmiingdity . Examon theples other inclu hadend to, deqmanals dark with, weathsiqnawer. co ld, and soqmed humid. A [-mbnt] verb, Toverbs the. root A close orad rainscrutin , wey ofcan theadd exam ma-,ples -an in, orclass si- VIand conobtainfirms threeour earlier[-mbnt ] Aco ncldiffuserention made affi xin means sect iona di 5.3.1fferent that hi ghltheseight forms on theare meteor semanticallyological dis evetinctnt. . phenomenonThe prefix insi- a indicatesparticular the locale occurrence. It ofcarr theies im witpliedh it meteoro its typiclogialcal existential phmeaysicalning. orThe psych prefixological ma- stis atefound and inof class involuntary Ia and classaction II. asWith an indiclasscator III ofand clstateass reIV,sulti ma-ngis fromthe tya pical"passiv marke" eract ionfor whergativeich cannot verbs be. voThereluntari itly imcontrpliesolled a bythere thefore Patient consi sustentlybject . invTheolu ntasemary nticand/or inter stpretationative. Hence for ma, we- verbsuse theis feature [±VIs. ttv] (stative) here to separate the ma- verbs from the -an verbs in class Our analysis for class VI impersonal intransitive verbs can be summedup in the folsectiolowingn 5.3.1. tree diagram which is the product of SR-10, SR-ll and SR-12 from

-mbnt +mbny -exst +exst -sttv �+sttv � c.a. : -anI maI - si- e.g. : orad-an ma-orad si-orad toqman rain rainy have-rain dal'k bet iIi-an ma-kotem si-kotem siqnaw thunders torn cloudy have-cloud cold

Figure 5.12 Subcategori sation of Class VI Verbs in terms of Inherent Semantic Features ; wi th exampl es 230 Somintee rpreredundanttation . semanFortic ins featuretance, s [+exstcan also] (e bexist addedence ) toim pliesfacilit a atephy semasicalntic state of being· and is therefore [+sttv] (stative) as well. [-sttv] implies action 0:[ process. In the next section, we will see that the suffix -an indicates phenomdeliberateena. action, but that interpretation does not apply to meteorological

5.3.7 Impersonal Transitive Verbs marThisking group. Asis uniquehas been in discussed Amis in its in casesectio frn ame5.2.7, as we classll as VII in verbsits morpho seem logicto �l semanticindicate delfeatureiberate [+ dlbraction] (deli on betherate part) isof thereforethe perform suggeer. stedThe to intfurtherrinsic charactethe followriseing th redundancyis class ofrule subjec: tless verbs. This feature can be added by RR-15 NOm] [+dlbr] �+[[ +AGT] This feature makes[ it easy] ---for us to contrast a class VII deliberate action verb such as rakat-an [deliberately ] wa lk with a class Ia non-deliberate sectintraionns itiv5.2e.7) verb. such as r-em-aka t walk (see examples 5.99 and 5.100 in

5.3.8 Semantic Features of Locative Verbs Locative verbs are characterised by the presence of the contextual feature [+ve[rb+LO clC]]ass inII the, their intracase nsitframesive. locativeHence, verbslocative, and verbs from inc claludess V,mem thebers from trafurthernsitiv charactee locativerise ve therbs locative. The setverbs of aresema nticthe localfeatureistics wh semaich nticwill featuresbe used to [±(adrcnssocia] tion)(direc . tion)These, [± goalfeature] (gs oaarel) , si[±milasorcr ] to(s ourcthe e)local, andistic poss semaiblyntic [±assn featu] �es wethe havelocat usiveed toverbs analyse, because the thecasese formsverbs. carryWe useinfor themmation to furtheron location charact anderis� directiocarried byn asthe the neutralir intr loinsiccative features determine whichrs . supplements the minimal information Class II locative verbs can be cross-classified with the feature [±drcn] means(directio the n)transportation. The direct orional moveme featurent of, bythe DeGuzmco-occurringan's definitio PAT ton or(1 978from:182) a pl, ace dethosesig natedlocative by the verbs co-occurring that do not LOC expre. ss"Non-di thisrectio type nofal" movem wouldent then. Instreferead, to in co-occurringthis context PATnon-direct is locatedional or verbs station refeed.r to a particular place at which the Furthe(motion)rmore, wit, hdir "mecotiotionaln" mea veningrbs ca then bemovement subcate ofgori thesed ph byysical the featureobjects [± motn] goalcorre [+spondinggoal] or to non-goal the PAT [-actantgoal.] veMotionrbs. verbsIn Am iscan, non-goal be further verbs spe arecified source as veinvolverbs, tothe be actualredundantly movement marked of phy [+sicsorcal] objects(source). . TheyNon-motion can be verbsfurther do not 231 subcsubcategategorisedory of bypsych theolog semaicalntic verbs feature which [± pschexpre] ss(ps feeychlologicings thatal) areto yield originated the fromsubcateg a sourceory of orsp ecidirectficed-ob towardject in atra goalnsit orives targ thatet , doas no oppot expsedre toss psycthe hological feefurtherlings char. Allacteri locatsed iveby psychothe featurelogical [± goalverbs]. are directional. They can be locIn alistsummaryic , semanticthe subcate featuregorisas tioncan be of st classated byII theintra fonsillotiwingve locativ SR's: e verbs by SR-36 +v+[ +PAT ] -[ +AGT] -+ [±drcn] drcn "direction" +[-[ +I+LOCNS] SR-37 [+drcn] -+ [±motn] motn "motion" SR-38 +drcn ( +motn ] -+ [±goal] +drcn SR-39 (-motn ] -+ [±psch] psch "psychological SR-40 [+psch] -+ [±goal] The SR's can be supplemented by the following RR: RR-16 [agoal] [-asorc] sorc = "source" These rules produce the following feature tree for class II verbs: +v +[-[ +AGT+PAT] +[-[ +LOC+INS] -drcn +drcn

-psch +psch -goal +goal �,� � verbslocation obspjececific­I t verbs logicalpsycho­I verbs verbssourceI verbsgoalI e.g. : ma-aI roq mi -kiI lim ma-olaI h rnaka-I raay ta-ngasaI reside-at look-for love­ come-from­ go-up-to toward afar ira mi-banaq rna-cal ibad mi-cal iw ta-ra locate-at notify angry-at borrow-from go -to

Figure 5.13 Subcategorisation of Intransitive Locative Verbs by Local istic Semantic Features ; wi th examples 232 The following sentence examples illustrate the [-drcn] or location verbs of classand not II. mo viWithng in thspaceese verbs. , the position of the PAT referent is stationary (5.104) ira ko nan i lomaq Z-ocate-at cat house Nom 1 LCV ��rns-drcn (+PAT (+LOC } the[ catl is in ' the house (5.105) awaay ko nan i lomaq not-Z-ocate-at catNom houseLCV ��rns-drcn (+PAT 1 (+LOC 1 the[ catl is not in the house (3.85) t-em- i reng an ini c ira tepar no sa 1 iIi stand now 3s side post +AdV Nom 1 (LCV 1 ��rns-drcn (+sptl 1 (+APT) l +LOC) he[ is st1 anding by the post (5.106) mi-cangray cira cabang Lean-a+V ins, 3sNom waLCVH [ -drcn-trnsr (+PAT ] (+LOC ] he Z-eans against the waH (5.107) ma-aroq kako pat iamay reside ls town +v-tr ns1 Nom+PAT LCV+LOC -drcn J ( ] ( 1 I[ l.ive in town directClass ionalII [+drcn moti] onverbs verbs are, thesubcate PAT gorerisedferent by mo vesthe infeature space [± awaymotn fromJ . With(i.e. , [-actagoalnt.]) orHer die rearected a towardfew sentence (i.e. , exam[+goalples]): the location specified by the LOC (3.83) ta-ngasa c ira panan a [mi -dakaw ] go-u+v p-to 3sNom gateLCV ride +drcn-trns (+PAT 1 (+LOC 1 (��int 1 +g+motnoal he rode up to the gate 233

(5.108) maka-raay kako roma no pat iamay come-from-afar ls other toum +v (Nom ) (LCV 1 +drcn-trns +PAT +LOC +motn-goal I came from a toum fa r away Memis bershiratherp smalof classl. TheII directionalsmall inventory motion of verbsclass , IIinclu motionding verbs goal/source is compe vensarbsted, sectiofor byn a 5.3.1fair numandber sect ofion class 5.3.3 I ) andwhich class expre IIIss non-l situatocativeionall motiony simi laverbsr notions (see motionfrom a withoutdifferent having persp eca LOCtive actant. These in non-theirlocative case fram motiones. verbs can indicate Withthe verb a class, as inI non the-loc folativlowinge mo examtionples verb: , the notional goal is incorporated in ta-lomaq go-home ta-l ikol go-back ta-qeka 1 go-out ta-ra-potal go-outside, lit. go-to-yard ta-ra-l otok go-to-mountain The derivation that gives rise to these verbs is very productive. Any noun thatintra nsrepreitivesents non a-locat locaivetion mo cantion com verbbine. witClassh ta- I orno n-lota-ra-cativeto formmotion an verbs can be(dir furtherection) characte, [+motnri] sed(motio byn) the, andlo cal[+goistical] sema(goanticl). featureSince thes [+ drcnlast ] feature im5.3.1plies (s eethe Figotherure 5.5twoc), I. have chosen it as a distinctive feature in section Class IIIb non-locative motion verbs are marked by the feature [+motnJ. They arethe setran verbssiti veare vernotbs derived. Unli kefrom cla nounsss I whichintrans canit ivebe non-interlocativepreted asmotion an im pliedverbs , orientatiodifferent persn orpe dirctivecetion from of Class the mot II ionloca. tivClasse mo tiIIonIb motverbsion, verbssince , rewitflecth the a se vether bser, gativea noti onaltransitiv sourcee verb, goal. , or The pathfol lowingis manif examestpledes as, includingthe Patient the suverbsbject of verfrombs exam: ples 5.60 and 5.61, are members of class IIIb non-locative motion ma-dakaw climb-over ma-dangoy swim-across ma-hatebo jump-over ma-taql ib pass-by Clspecificass II -odirectionalbject intr ansnon-mitiveotsion marked verbs asin [-cludepsch]. psychol Theogical direc [+tiopscnalh] verbs and psychological verbs can be further characterised by the feature [±goal]. An example of [+goal] verb is ma-olah love-toward where the LOC actant is the ma-targetalt awtowardafra id-of,which onthe theemot otionher ishand directed, can be. lookedThe LOC upon actant as the of sourcea verb oflike fear, thus making ma-talaw a -goal verb. Since the [±goal] distinction does notsubcateg have orianysat morphoion furthelogicalr. consequencesBy the same hertoke,en we, wewill wou notld notpursue furthe ther 234 subcategorise the [ -psch ] verbs in class II. The specific-object locintraalisticnsitives features are si mplyare cotonce berned inter. preted as being directional where Allthe Classchain Vof causative causation verbs is syntact are traicallynsitive expr andes seddirecti. Ino nalother. wordsWith th, theese verbs, whetherdirection ove ofrtly in fluenceexpress edis regior notster, ed1:S bythe caseremote fra mecau sefeature. Thes. LaCThe actant AGT actais ntthe, notional "indirect object" of ditransitive verbs. with class Vb indirect whcausativile thee PATverbs is, the LaC"direct actant ob ject"is to imm be ediainterpretedtely affec asted an by int theer mediatecaused agactention:, . The sequence of influence proceeds from AGT to PAT through LaC. Class Vb verbs, therefore,' can be redundantly marked as [+goal ] . Wi th class Va indi"ditrcateansit onlive"y thetranspo direcrtationtion toward and intheformation LaC acta verntbs. , Thallese the verbs exam, plethers eseemfore to, can also be redundantly marked as [ +goal ] . Thlocalis isfurthertic semantic charac tefeaturerisation can of be class stated V traby nsitthe ivefo lllocatowingive redundancy verbs by rule: RR-17 +V +[ +PAT ] +[ +AGT ] [ +goal ] +[ +LOC ] -[ ] \. +INS CHAPTER 6

SYNTACTIC DERIVAT ION OF AMIS VERBS

6.1 Inter-Sentence Rel ationship One of the main concerns of generative grammarians is the relatedness of grammatical structures. In the Aspec ts model, as well as other versions of sutranspposedform toational be accounted grammar forinclu byding trans Filformationlmorean case rules gramm. ars, relatedness is This approach subsequently ran into many problems, among which were the problemstransformation of lexi rulescal insedo notrtion re andally the reveal lack how of cothenstra structureints. s Woarerse re stilatedll, tothe ea(structuralch other. changeFor insta) portionsnce, theof S.a D.tra nsf(structuralormation descrule ription)give us andlittle S.c. caseinformation relations on howand casethe co formsnstitu. entsBe sidofes a , sentenceas trans areformation realig nedrules in wi tethinrms ofthe any"sta ndardsemantic theory" shifts do ornot change allow infor pe rspa changeectiv e ineit meaherning. , they do not account for In this regard, the lexicase approach is superior because (1) it has rid itself ofrul esa too, (2) powe it rfhasul builtand hence in a arbitrarynumber of me constraintschanism, i. one. , governmthe transent formand atioselection n, among them the l/Sent constraint, (3) syntactic description in the lexicase problemmodel is ofha ndledlexical within insertio then lexivanconishes, and and with co-occurrence the abandonment restri ofctions PSR' betweens, the co(4)nsti reltuentsatedne ssare ofst atablesentence as coredundancynstructions rul caesn bewith sta contextualted as rel atedfeaturneesss , betweenand phoverbsnological or verb featuresclasses witstateh dre spinect the tode sharedrivation syntac rulestic (D, R'semantis), makc,ing and it unnecase ceassigssarynm entto invoke. transformations that ignore changes in perspective and According to Starosta (to appear a:l), one of the fundamental claims of the inglexicase to th frais meworkhypothe issis what, the may Patient be referred case re tolat asion "Patieis obntligator centralitily pry"esent. Accord­ in the Ancasederson frames (1977) of all and ver Gruberbs. Si(1mi965)lar, thoughclaims havetheir been approaches made by differlinguist in s such as markssignific theant perceptual ways from centre that ofof the a predi lexicatcaseion mode, andl. syntactiSemanticallcally,y, a thenum berPatient of ecreonomilationscallyhips, andamong insverbal ightfu clally usesin terms can beof accounted verbal derivation for explic, itlincly,uding

235 236 zeresortingro-derivation to ad hocrules and inv emolvpiricallying the" promvacuousotio ntrans or demformaotiontional of Parultienes. ts, without By stating such rules in terms of the limited set of case relations available consin a traintlexicase on thegrammar kinds (cf. of derivStarostationa 1982d) rules, wewhich pl aceare an po sslautomatble: icall em piricderivedal verbsset of mu postssi fitble into case the frame sets ofthat po ssicanble be verbconstr clauctedsses wit as h charactethese avriailasedble by cathese newrela derivedtions. verbs A coro willllary ente ofr tha isclass proce whichss is al thatready alm containsost without some exunderivedception, all memrembersains . essentiallyThat is, thecon stasetnt of, and basic each syntacti of the calbasicly de classesfinable con verbtains classes some undeclassries.ved charte'r members and some other members derived from items in other As6.1 an trans exampformle atioof nallythis appro, as achis comm, insonlytead done of deinriving Fillm oreana sentenc casee suchgrammar as 6.2 from 6.1(Fill inmore the 19lexic71:4on9),: the verb of sentence 6.2 can be derived from the verb of (6.1) John hit! his canePAT against the LOCfence . (6.2) John hit2 the PATfence with his caneMNS . That is, examples 6.1 and 6.2 involve pairs of distinct though homophonous wordsdiffering, iden intic synal tactiin "pc artsubclass of speec andh" pe, rootrspe ctivemeani. ng, and pronunciation, but Theas nulexicalmber 6.3: derivation process which produces hit2 from hit1 can be formalised (6.3) DR-l (+V +v +[ +LOC] + ([ +MNS]) +LOc >-+ +PAT 1 J{l aFi 1 J (aFi 1 +PAT +MNS J (�Fj 1 J (�Fj 1 yFk yFk nothatn- locativis, corree, whospondingse Pa tientto a correlocativspondse ve rbto, thethere Locus may beof anotherthe underived verb wh verbich , isand whoeffectse opt, theional Locus inner has Means been acta"promontted" corr estoponds Patient to , thebe comingunderived the percPatieneptualt. In pecentreripheral of the Me anssentence relation, and. the former Patient has been "demoted" to the more This process of derivation then imposes a new perspec tive on the event of hireasttingsignm byent reinter of a prPatienteting thecase sirelatiotuationaln to goalthe ofLocus the caseaction rel. ationThe in the fir�t undergoeconstructionr of obthelig actatorilyion, and down reintergrades prethets previous it for thePatie derivednt, wh verbich enascoded the maasni thefesti thingng thebe ingMeans moved case through relat ionspace. , Thusbut canein sentence in sentenc 6.2e it6.1 is ispe percerceiviveded as

L-______�______237 the means of hitting. Similarly, fence, which acted as the goal of the motion sentenceof the cane 6.2. in sentence 6.1, is viewed as the entity affected by hitting in Derivation Rules such as DR-l formalise the tendency in natural languages for classverbs . of Everyone class verb to st beem reintercorrespondspreted to analoga conceptuicallyali assation verbs ofof a somepar ticularother aexternal situation situation. Thus, byand deriv each ingcase a stfraemme fromrepre onesents case a particframe ulaclarss pe torsp anothectivere , on weperspective are putting. the situation represented by the verb stem into a new Threvisealing derivational way to theap proachdescription can be ofap manyplied ot inher a inter-sentformally siencmpele re andla tionshipsconceptu ally theinvol basving is disfor tinctia neaonst ty pologof casey of-mark suching re andlationships transiti vi. ty, and in ·fact provides frameworkIn conclu siofon gramma, lexicasetical, beingdescri ptiona non-trans, does fornotmat accountional generativefor inter-se lexicalistntential clarelatiousesnshi are psacc witountedh trans forform inat termsion rulesof verbal. Inst derivationead, relatio procnshiessesps invamongolving verbal the whichpromotio marksn or the dem otperceptualion of the centre case relof aations pred ictoatio andn. from the Patient position AmiIn ths.is Wechapte shallr, weli mitare ourselvesgoing to preto sentthose someDR's syn thattac ticaffect derivation the case rulfeatureses for deregirivsterationed procein thesses source that. (aThat) in corporateis to sa y,a wecase would notion only, and cons eitiderher those (b) delete toa CR, or subtrac(c) add tinga CR, fr omor the(d) totalreint ernumberpret theof CRcase's ofrol esthe. sourceThis Vpart wit houtof the adding study classiis impofirtantcation as bea causesequel, firstto a ofst udyall of, we verbal can demonconststructionsrate the andlexi verbalcase claim withoutthat inter- the sentenceuse of trarelnsfationshipsormation canrul esbe. adeqSeuatelycondly , accountedwe want tofor test with the DR 's, hypotderivationhesis of AmisPatient verbs central can shedity wit someh Amligisht data on . the Alsodeve, loapme stntudy of of thesyntact focusic lasystnguagem esin . Amis as compared with Proto-Austronesian and Philippine-type Now a few words about the form of derivation rules (DR's) are in order even Boththough the we inputhave alrand eathedy preoutputsented of Derivationa DR is given Ru lesin underthe fo sectiorm of na 2.2.feature2.3 beforematrix. The input and the output is related by a fletched arrow (>�) . A DR can changnew lexe, icaaddl orentries delet bye syntactianalogy c,wit semanh pretic-exi orsting phono onlogicales. New feature featuress to produceare added contextualto the output feat matrixure, marked only. byThey either can abe "p inlus the" (+) for orm ofa "msiinusmple" feature(-) sig n.or I havethis liis stedonly case a matter features of conveniencbefore semane. ticIt features is to be fornoted ea sethat of rein ferencethis chapte, but r, cospnsiecistentficall wity inh DRthe's, lexi casecase form conv laentbelsion are for expl expreicssitlying marked impli cationalby + or rela-, tot ionbe . Asfail for to deappeletioarn, in thethe conventionoutput matrix is thatare neceall featuressarily s omiin tttheed . inputOther matrix feature thats matriof theces input are carriedclass not over specific into allythe derivedmentioned cla ssin . the input and/or output 238 With derivation rules, correspondence between a set of input and output safeaturesme Greek can lett be erindicated to mark bythe usicorrespondingng, instead offeature the pluss on bothor minus sides si gn,of thethe wasarrow first, such introduced as yFk in inthe sectio inputn and2.2.2. output2 under matric Redundancyes of DR Rule-l. s.T, hisLa stconv, butent ion notfor thelea st,"p romforotio statingn" and the "d emcorrespotion" ondenceof case between relatio nscase, we feature use, sin which addit accountsion to featurethe Greek im pliesletters the, a selecti"horseonshoe of" another(J) to implyfeature that. theFor occexamurrenceple, 'the of notation one s whichof J[ +LOCsignal,y Fthei] andcorresp J[+PATondence,yFi ] betinween DR-l the re ferLOC toin selethe ctionalsource andimpl theica tiPATons in the output, with reference to the features' yF i. The "horseshoe" notation is diin ffsecterention from2.4.4, the in"d oublethat theneg ationformer" notationrefers to, whichselecti wasonal also impli firstcations presented a viol"horseationshoe of" notation which doe iss usednot ne onlycessa witrilyh de affrivationect grammati rules calitin thy.is disserThe tation. Changes in phono�ogical representation, if any, are expressed as a second part: whichof the isder repeatedivation belorulew. as inThe ex ampfeaturele 2.1s 3shown for Englin thisish gerundiveseemingly nomseparateinalisa parttion offirst the partrule. are actually part of the main input and output matrices in the (2.13) (a) +V ) +N

+Nom > .... -[ +rtcl] J [ ctFi j)l +Det J (ctFi ] i (b) >-+ - f) ] This separate listing of parts (a) and (b) is only a rule convention adopted to hibearghl inight mind the thatmorphoph the onemictwo parts changes of the in dea rivationmore tr aditiorule nalare wayactua. llyWe inseshouldparable andstatement should isbe unnviewedecessa asry a. single DR. For zero derivations, the morphophonemic

6.2 Verbs Deri ved from Noun� It has been repeatedly pointed out in Chapter 3 and Chapter 5 that, in Amis, manycases ca, se-lthe higikeher notions verb iscan deribe vedexpre fromssed aby noun verbal in whichcomple themen tatcaseionlike. notioIn mon stis embodied. For instance, an instrumental verb like si-tokar use-ladder has tokar ladder as its implied instrument; ma l -kaka be-with-elder-sibling has kaka elder sibling as its implied companion; and mi-bot ing catch-fish has bot ing fithesh characteri as its implsticied case obje ctfra. me Allof [+ th[+PAese T],-derived[+AGT],- verbs[+LO areC],- intr[+INaS]].nsit iveThey, wit hare alonethere forewithout members complem of entationclass Ia.. Sometimes, these derived verbs can be used Verbs that are derived from nouns are of two types: (1) those which incorporate ancase unde frarivedme of N theand source(2) those V) andwhich create take a anew deriv verbed Nfrom (a deverbalit. N with origina:, 239 Theand isfirst descri typebed inc inlude sects theions incorporation6.2.1 to 6.2.6 of belo notionalw. Thepatient case .orro leins assutrumenmed t,by theseparate deriv edCR V,in sathey, casethe notioframe n ofof theins newtrum verbent, . willThus not itoccur is theagain deri·avs eda verb carriedthat as sumesby the the source respo N.nsi bility of carrying the case role expected to be The second type obtains the complex verb formsthrough a secondary derivation nominalfrom deverbal sources source is de nouscrnsibed. inTh issect secondion 6.2.7 type beloof w.verbal derivation from

6.2.1 Verbs that Imply Phenomena A subclass of verbs in Amis that are derived from nouns that are semantically markedexistential as [+ phverbsenJ . (phenomenon) include meteorological verbs, ambient verbs, and Meteorological phenonmena such as orad rain, bal i wind, bal iws typhoon , kotem cloud, lelesi j1ood, lelen earthquake , bet il i thunderstorm, lightning , and ca l amay fo g are derived into verb forms by three possible processes: exist(1) byential adding [+ exthest pre] verfixb, si- to the [+phen,-mbnt] noun, thus making it an st(2)ativ by c adding[+sttv] the ver preb, fixor ma- to the [+phen,-mbntJ noun, thus making it a pro(3) cessby addingverb which the suffixis marked -an [-toex thest ,[+-sttphenv] ,by-mbnt the] prenounsent, thus anal makysising. it a Process (1) can be expressed by the following derivation rule: ( DR-2 +N +V[ +Phen >+ - +Nom] Il-mbnt + ([ +PLC]) 1 +phen+ ([ +TIM ]) +exst-mbnt exstmbnt "e"ambxistencient" e" -+ [ si-

Examples include: si - ko tem cloudy from kotem cloud and si-saqpi qay-a-bal i breezy from saqpiqay-a-bal i coo l-wind, breeze . Note that sawp iqay-a-ba l i is 4.2.2a compound unde r formPrep, ositthe iofunctionns. This of forma having is abeen compound previ ouslybecause discussed of its sema in nticsection unis ita y compoundas well andas the not tre a atmentsyntactic it receconstrivesuctio byn DR-2is furtheras a si nglesuppo unrtedit . by theThat factit that a form produced in analogy to the N-a-N pattern, say, *ma l bawaay-a-pising swdeolalinglen witfaceh hasnomi nalbeen dererivationjected by or DR-2com poundingas input into thisthe rulestudy. , I Siamnce incl I udingamnot thisin DR-2 discuss. ion here only to justify the characterisation of the input features Proceidenticalss (2) to canthat be ofsta DR-2ted : as the following derivation rule with input features 240 DR-3 hen >->- +-[v +No m] ::-mbnt + ([ +PLC]) 1 ( ] +phen+ ([ +T IMJ) +stt-mbntv sttv "stative" ->- [rna - Examples include: rna-orad raining from orad rain and rna-bal i (wind) blowing from ba 1 i wind.' onlyProcess a diff (3) ercanence be in desc tworibed sema nticby a derivationfeatures: rule similar to DR-2 or DR-3 with (+N +v DR-4 +Phen -[ +Nom] l-mbnt ] + ([([ +TIM+PLC]) +phen-mbnt -exst-sttv ->- -an ] Examples include: orad-an raining from orad rain , bal iws-an having a typhoon from bal iws typhoon . All three subtypes of these meteorological phenomenon verbs are subjectless, do notfoll oco-occwing exaur mwithples a: PAT, and may have a Time and/or Place actant, as in the (6.4) si-saqpiqay-a-ba l i sansandeb C��l-b�eezy dusk( lLcv 1 (+P-mbnthen +TIM it is breezyJ at dusk (6.5) rna -orad tO l potal rainin atrea still outside +v ( +Adv (LCV 1 [l+-mPbnthenl +inc+sptlh +PLC +drtv-prft it has been rainingTJ outside ThereDR, but is with ano therno changessubset ofin morphophenomenonlogical verbs sha whpeic. h Theare seobtained are the by amb a ientsimi [+lambr nt] verbs which include: siqnaw cold and I iqcec snow . The derived verbs are similar to other class I stative verbs such as arawraw round , coplak sour , adada sick , and lahkak red in not showing any additives to the basic forms. canThe bederi stavationted asru lefo llowsthat : derives the [+mbnt] stative verbs from nominal sources 241

DR-5 +N � +V +Phen >+ -[ +Nom] [+mbnt + ([+TIM+PLC ]) J +phen+ ([ +stt+mbntv Since this is a zero-derivation, no morphophonemic statement needs to be made. Suintrabjecnstleitssive phenomenon verbs by addingverbs acan su bjfurtherect PAT be toder theivedir caseinto thefram escategory. Nouns of thatsimp le mayrefer occur to weatheras head or of timethe iror suplbjaceect, NPas arethe refolstlowingricted exam to plesome will general show te: rms that (6.6) si-kotem kina remi ad this day�Nom weather Cl( :�:::-mbnt l (+PAT ] it is c�oudy today Lit. this day/weather is cloudy Thsectision der 6.3.2.1ivation befromlow. su bjectless verbs to intransitive verbs will be treated in

6.2.2 Verbs that Imply Measure Verbstempe ratureindicating, and phweysightical are sta detesrivatio of beingnally such rel atedas cotolo urabst, sizracte, qushapeality, nouns indicatingfollowing rule these: physical states. The relation can be expressed by the DR-6 +V+[ +PAT] -phen-[ +AGT ] +meas meas = "measure" No morphophonemic statement is made in DR-6 because the derivationally related measureto claim nouns that theand verbsdirectio sharen is thefrom sa meN tobasic V and form not. theIt othe is rin waya way around arbi becausetrary procthe measureess: nouns are abstract. We may also consider the following derivation DR-6a -[ +PAT+AGT] >+ fl:C+meas-phen VThe in fol(b)lowing: pair of examples show the related forms used as N in (a) and as 242

(6.7a) ma-banah haw kiso to baqkec nia aba? know QM 2s ht box rns Nom+AGT } +PAT :-erg� v l +meas do you l know the weight of the ��box? (6.7b) baqkec[ kia aba l h box f s � T +meas (� : ] the��:: boxl is heavy Other words that[ belong to this categdry include: tarakaw height/tall, tataang size/big , and taraqyaq length/long . [+RextrnferringJ (e backxtrin tosic) Fig quureality 5.5 b,ver webs , canwit placeh the thefol lowingderived redundancy measure verulerbs : with the ] RR-18 [ +meas ] + [ +xtrn

6.2.3 Verbs that Imply Instrument Asthere has are been two men subctionedlasses in ofsect verbsion 3.3.3in Am.2is unde thatr arethe derivedInstrument from case nouns re latwhoseion , Asre fershownents rearespe ctisituationalvely in secinstionstruments 3.3.3.2.1 of the andact ions3.3.3.2.2, indicated the bymorp thehologicalse verbs . shape of the derived form can either be modified by the suffix -en , making the resultant form an ergative verb, or by the prefix si-, making the resultant asform fol a lowsstative: verb. The derivation rule for the first type can be formulated DR-7 +v+[ +PATJ -[ +AGT+INS] +t+ernsrgv ertrnsgv "e"trgativeransitive" " +nstr-sttv nstr "instrument" ] -en ] The derivation rule for the second type can be stated as follows: DR-8 +v str >+ +[ +PAT] (:� ) -[ +AGT] -t-[ +IrnsNS ] +nstr+sttv + i - [ 5 243 These verbs are normally used with verbal complements which depict the activitiesrespective seforctio whnsic hmentioned the implied above instr. uments are used. For examples see the The prefix si- can also be attached to non-instrumental nouns to indicate that aintri certainnsic qualobjectiti esis . in Theseexist enceverbs, inalso one belong's possess to theion , classor po ofssesses simp lecertain reintraadilyns itivpredie vectablerbs. asNote those that with the im plmeaningsied inst ofrument these (cf.derived Figure verbs S.S bare in not as section S.3.1) . Examples include: si-pida rich from pida money , si-kawas haunderivtedation from rule kawa cans ghos be t,stated and assi -nganganfollows : famous from ngangan name . This DR-9 [+N] >-+ +[+v +PAT ] -[ +INS+AGT] +stt-phenv -+ [ si-

6.2.4 Verbs that Imply Product Ththeis su classbject of Patient verbs . impliesThe verb a product is a "maorking" resul vet rbfrom and the the acti producton pe rfisor momedstly by a physical object though it can be something more abstract like a dance as in misa-kero make a dance and numbers as in misa-os i make numbers, count . The tystpicatedal bypr eftheix fol to lowingthis class rule : is misa- and the derivational relation can be DR-10 [+N] >-+ +[+v +PAT ] +prdt-[ +AGT ] prdt "product" misa- -+ [ Theverb verbwhil eis theint notionalransitive agentbecause is therei ntenotionalrpreted patient as the issyntac incorporatedtic PAT. inExa themples include: misa-lomaq bui ld from lomaq house , misa-labi make supper from labi evening meal , misa-ke ro dance from kero dance , misa-os i count from os i number , and misa-toron make rice-cake from toron rice-cake . One example involving semantic shift is misa-tamdaw make friends from tamdaw person, people . non-phenomenonReferring back andto Fignon-stativure S.Sc,e weverbs know wit thath im pl[+prdtied ob] verbsject. are redundantly

6.2.5 Verbs that Imply Acquisition produceWhile thes theverbs ob jecint theexpl previousicitly mentionedsection involvein the verba creative form, activitythe verbs thatwe shall obdescriject.be inSi ncethis itsec istio then involvenotional acti patvitiesient thatby which is aff oneec tedacq byuir esthe theact ionimplied and 244 asis theincor syntporatedactic in PAT the and derived the verbs verb areform intrans, the notiitiveonal. Thagentis classgets reinteris preted characterised by the prefix mi- (see Figure 5.5c in section 5.3.1). ExampleE include: mi-qadop hun t from qadop game3 animal , mi-bot ing catch-fish from bot ing fish , mi-qoay gather rattan from qoay rattan , and mi-kasoy gather­ fiforewoodllows: from kasoy firewood . This derivation process can be stated as DR-ll >-+ ::ncr +PAT ( 1 +afft-[ +AGT ] afftcncr "aff"concreteected"" -+ [ m i-[:[ 1] Ainta semanke ortic consu shiftmp hastion takenof certain place withfood theitems fo insllowingtead wordsof the whiric acqh meansuisiti onthe. Examples include: mi-q icep chew betel-nuts from qicep bete l-nut and mi-qepah drink wine from qepah wine . The typical causative prefix pa- can also be used with concrete nouns to imply acquisition. If the subject is tireng no mako myself, Lit. my body , the utterance would have the idiomatic meaning of let me have something . Examples are: pa-titi let someone have some meat , pa-q icep let someone have some betel-nut , and pa-nanom let someone have some water . The derivational process morphopcan be honemiccovered bystatem DR-entlla : which is similar to DR-ll, with only a difference in DR-lla >-+ +PAT] +afft-[ +AGT ] [ pa- -+ [:[ ]

6.2.6 Verbs that Imply Companion In Amis, there is the productive process to take a human noun and make it an implied companion or concomitant in a verb marked by ma l-. These verbs take pluralderivation subj proceect andss whichare usually produces fo lltheowedse byverbs a verbal can be comstplematedent as . follTheows : DR-12 (+N +v l+humn ) +[-[ +PA+AGTT]] + [-fint] fint "finite" - +Nom-plrl humn "human" +cmpn plrcmpnl "p"companlural"ion "

-+ [mal- Examples include ma l-wina be-with-one 's-mo ther and ma l -kaka be-with-one 's­ elder-sibling . 245

6.2.7 Verbs Deri ved from Deri ved Nouns In Amis, there are verbs that are derived from derived nouns. Input to this typethe non-nominativeof derivation actaare ntsdeverbal carried nouns ove thatr from show the upir insou NP-NPrce ve rcobsns. tructionsSince theywith thereare seforecondari quitely derivcomplexed , inthe foserm. verbs Exam haveples two are la yersverbs ofthat affix areation con sianddered are sectinstrumionenta 3.4.2l focus.2) ve r(cf.bs. sections 3.3.2.1) and locative or locus focus (cf. An instrumental focus construction is a "passive", i.e., non-agent focus coconnsstrtructionsuction. inAccording Austronesi toan Stlaarngostauages, Pawl, espeeyciall, andy Reinid Ph (1il98ippine2:13) , lapassnguageive s, arean altgeneernativerally amenabverballe ap toproach an NP-NP. Itana islys onlyis wh inich the is resiglaniftiveilycan tlyinfrequent simpler than examcertaiplesnty where that wea binareary de alingIC analys withis re isal impverbossi focusble that. As we showncan sainy witsechti ons theore3.3.2.1tical and im3.3.3.1plication, Amis for is Proto-Aus a languagetrone blesissaned syntax with a isfew great such. examThepl es. The occurrenceStarosta, Pawofley such, andverbal Reid con (ibidstructions.) to advance in Tagalog the hypothe and Amisiss has that ena verbalbled focus conin Prstroto-Austrouctions werenesia lan terwas developedonly at an from in cinominalpient staisatige on.and that the "passive" ofFrom NP-NP sect constructioion 3.3.3.1ns we. have In examsectplesion 3.3.2.1showing we deverbal have exam nounsples usedwith assi misubjelar cts theforms fol aplowingpearing exa unquemplesstio: nably as the heads of verbal constructions. Consider (3.70) o kamay non iam ko sa-pinaro tina ka ri rengan hand lexcl means-of- loading this cart [ Neu] ACC Tl (+PAT 1 we loaded this cart with:;: our hands Lit. the means of loading this( cart is our hands (3.71) o nanom no tebom ko sa-pi bacaq non iam water we ll means-of- laundering lexcl ] ( ( [Neu Nom+PAT +AGTGen we launder with we ll wa terl 1 l ) Lit. our means of laundering is we ll water with these examples, the analysis is decidedly an NP-NP construction because detherived nominal nouns pr whichedicates se rveare ascl easubjrlyect marked are clearlyby the markedNeutral by determiner the Nominative 0 while the determiner ko . With the following examples, however, only a verbal analysis is possible. The absenceinitial posiof a tiondete andrminer the beinforetervention the de rivedof the forms subj whichect be appefore arthe in Accusatsentencive e whichPatient appe alaredl point in secto tiona verbal 3.3 .2analys.1. is. Here I am quoting the few examples 246

(3.54) sa-pa-ahcid ko c i 1 aq to tood used-for-making-something-salty salNomt thingsACC rns 1 +INS 1 +PAT 1 sa(:�lt is used fo r making things salty ( ( or, salt is used fo r salting things (3.56) sa-pi -angang nomako kina piqio to tamdaw used-far-calling lsGen this whistle peopACC le rns +AGT S +PAT 1 I( :�use this) whistle (to ca) ll people(� �: ) ( Intermediate between the clearly nominal and the clearly verbal constructions arecons tructionssome seemingly or as amNP-NPbiguous co nstrexamuctionsples which due tocan thebe viewedresemblance as "pass of theive " genitverbalivE' markermuch prob to lethem becaposusesess, ivein maran NP-NPker in constructionAmis. In our, the an derivedalysis, nounthis doesis other notwise cau� e othernever hafoundnd, altoways be usedappe inars thein predicatesentence-initial position pos. itTheion . verbalThat formthe , senton ence­the notinitial preceded form byis ato debeter anaminelysedr. asIf a siVmilar is als nominalo suppo prrtededi bycates the andfact verbal that it is prStaredostaicates, Pawley had once, and co-existed Reid, the incon Proto-Austrotrast has alnesreadyian , beenas hypot losthesised in Amis by. Typical of ergative syntax, the marking for the genitive case form and that of ama posbiguoussessor anal areysis ide: ntical in Amis. This gives the following example an (3.57b) sa-pi -adop kina waco no lomaq ako used-far-hunting this do g family ls [ ( +vJ Nom+PAT ? (i) this dog is used fo r huntingl by) my fami ly (ii) this dog of my family 's is used fo r hunting In this example, no lomaq ako can be. analysed as (i) the Nominative Age�t or asseco (indi) readingthe possess givenor inof the waco glo dogss., yieInlding either respe casectively, it shouthe ldfirst not andaffec thet diagra verbalams : analysis. This ambiguity can be illustrated by the following tree (i)

I NP �P �o t I I sa-pi-adop kina waco nia lomaq ak� o [ +V J Nom ) /'I (Gen ,;1+PAT, +AGT01" 1 my family uses this( dog j fo r huntingl 247 (ii)

��t �Det� NP sa-pi-adop kina waco niI a lomaq ak�o [+V] �T this day of my fami:ly� 's is used fo r hun ting verbsThe der canivation be st atedrelating by the the fodeverballlowing( nounsru] le: and their corresponding instrumental DR-l3 +N +v T 1 +[ +PAT] J :;: 1 +AGT +[ +AGT] J (�Fj ] >+ +[ +INS] +Gen +PAT - -AGT( ] J aFi +ACCJ +AGT - -PAT( J (�F. J +cncr( ] -(+-AGrG�n J -(+ACC-PAT ) -(-I+NomNS J +nstr( ] nstr = "instrumental" thereDR-13 is ontoa zer othe de derrivationived verb. It. coAspies was fromindi thecated source in sect nounions whatever 3.3.2.1 mark anding 3.3.3.1, if the notional instrumentis animate, the prefix mam i- is used. If whichit is theinan nounsimate , werewe havederiv theed . preThefix additionsa- added ofto anthe ob lisourcegatory verb Nomi stnatem ivfrome reInsintertrumpreenttation is indica of CRted's by is thereq uir"doubleed, theneg correation"spondence feature . betweenSince theno input AGT redundanand the outputtly st atedAGT andin DR-that13 betweenwith Greek the leinputtters PAT and andthe theim plioutputcatio nalPAT notatioare n matbecauserices I. wantBy toIe emxicasep has convenise thetion similarit, howevey r,between any features the input of theand outputsource not exprederivedssed form in , eitsoher DR- input13 can orbe outputrestated are ec automatonomicallyically as carriedfollows over: to the 248 DR-13a +v +[ +PAT] +[ +AGT ] +[ +INS] +Nom -(-INS } +nstr" Besides the ,"instrumental focus" verbs, we also have a few examples of conclude"locativ e thatfocus" Ami verbss focus in systAmiems. isThe at beexamstples in itares incipientso rare thatstage we, haveresemb tol"ing Likethe siDRtuation-13, the of pr Procotesso-Aus fortrone derivingsian (cf."loc Stativearos tafocus, Pawl" verbsey, andhas Refarid more1982 ) . nominalverbal categcandidateory. s Infor otherinput words than , thothoughse that the actualprocessly isget avderivedailable into, not themany insis thattrumental in Am isfocus nomin or allocisaativetion cofocusnstructions verbs are li kefound exam inples Am is3.7. 0 Ourand obse3.71 rvationcited notioabove nalare locusfar more than frequare ently"pass ive"used con tost higruchltioightns . a notional Even if instrthe nominumentalisa or ation devicescribed is notin sectused,ions there 3.3.3.2 is still and 3.4.3.2the com plemthatent areat commonlyion deviceused s which in Ami weres to expreunderdevelss notionalopment instrof theument Amis orfocus loc ussystem. Th. is probably accounts for the In the following, we will give an example of "locative focus" verbs derived DR-from14 deverbal. The examnounsple and is takenstate thefrom resectiolationshin 3.4.2.2p between under the Lo twocus- cafocusteg oriesin Am witis. h (3.91) sa-s i ngaq-an ko al i 1 i to tipos pZace-for-storing granary grain [+v] I\j "") [ Nom ACC l+LOC +PAT grain is stored in the granary) ( 1 Lit. the granary is the pZace fo r storing grain Thethe analysconstructiois of n examheadple to 3.9sugge1 isst clea thatrly they ver balshould. beTh ernomie isnal no codeterminernstituents be andfore thefor poan sitNP-NPion coofnstr the uctioAccusn.ativ eTh Patientis is the ma kesonly itexam imppleossible I have to inmake my andata IC showingcut betweena derived the " locativverb forme fo andcus " itsverb source with nouna deverbal can be sourcestated noun as fol. loThews : relatednes�: 249 DR-14 +v +PAT +[ +PAT] ) etFi +[ +AGT] ) +AGT +[ +LOC] P( �Fj ) +PAT +Gen >-+ ) ] etF�· -(-AGT) (+AGT) ) (�Fj ) +lctn - +Gen-AGT +hbtl (+ACC) ] -(-PAT ) (��� - +Nom-LOC +lct( n ) lctn "location" +hbtl hbtl "habitual" Whatcarried this over ru leto satheys deisrived that theverb casewith frtheame addifeaturestion ofof a theNominative source nounLoc usare. whichDR-14 isindica alsote as azer habitualo derivation location. The is Ccopiedl a - ... ontoan af thefixat derivion oned verbthe source. noun DR-14 contains many featu�es which remain unchanged but are redundantly listed bein bothsimpl theified input as foandllows the: ou.tput matrices. By lexicase convention it could DR-14a

Personally I prefer to have the input and output categories more fully idespecintifiedfied as. in DR-14 so that these categories can be immediately and uniquely 250

6.3 Verbs Deri ved from Other Verbs Amis verbs are grouped into seven primary verb classes according to their caSE! frafollmeow ingfeatur diesagram, as, weshown will in listFigures the 5.1sevena andprimary 5.1b verbin sec classestion 5.2. wit h theIn their matrixcharact eristicto indic caseate the fra mnumesber, wit ofh pothesi nutivelymber marked1, 2, orcase 3 atre thelations bot tomfor ofea eacch h arrclaowsss. howIn ththeese saAmmeis di verbagram classes, we will are alsorela tedshow toschem one aticallyanother witderivationh fletchedally . VI VII I II III IV V +[ +PATl +[ +PAT] ( -[ +AGT+Nom]l +[-[ +PAT+AGT] +AGT] +[ +AGT] +[ +AGrl+PAT]l +[ +AGT+PAT] -[ +LOC] -[ +LOC] -[ +Lac] +[ +LOC] (+[ +INS] ] -[ +LOC] +[ +Lac] -[ +INS] -[ +INS] -[ +INS] -[ +INS] -[ +LOC] +[ +INS] lr-[ +INS] [-[ 1 l 1 1 [-[ 2 [ 2 l 3 j 3 (a) > [j ::�;I ll (b) ) l (c) -+1> l (c') > l (d) > I (e) > l (f) ) l (g) � < (h) � <

Fi gure 6.1 Derivational Processes Rel ating Ami s Primary Verb Classes Classnot patient VI isless marked in witourh anthealy numesis. ral As1 becauseshown in itFigure is con 6.1,side moredst su flebjectlesstched arr butOW 5 powithint proce to thesses rig (b)ht , direction(c), (c' ),mean (d)ing, (e)that and mo st(f )of . theProce process sses(hY suaddbtracts a CR , aas CR . Alreinterl othepretatior procensses of doCR 'nots. add or delete CR's, even though they may involve the Claas wessll III. hasEven been though extended we have to incsomewludehat tra sinsimplitivefied verbs the charactewith [+[ristic+PAT], +case[+IN S]] itframes is toof be the unde non-insrstoodtrum thatental both and subc instrlassesumental are trrepanresentsitiveed verbs under incla Figssure III 6.1,. Wemem bersindicate of classthose IIIderiv as ationsprocess that (c) addand anthose AGT tothat class add anI verbsINS asto prmakeocess them (c '). Stated somewhat differently, these derivational processes relate verb class VI tofo lverblowing class simpl I,if verbied tabclassulation I to: verb class II, and so on, as shown in the 251

(a) VI >-+ I (b) I >-+ II (c) I >-+ IlIa, IIIb (c' ) I >-+ IIIc, IIId (d) III >-+ V (e) II >-+ V (f) III >-+ IV (g) III >-+ II (h) III >-+ I In section 6.3.1 below, we will present Amis verbal derivation processes belongingremainder toof typethe procs thatesses add liCRsted's. in InFig secture ion6.1 6.3.2,. we will deal with the

6.3.1 Deri vations that Add eR's In this section, we will present some Amis verbal derivational rules belonging Figto typesure 6.1 (b) to, be(c) proc, (cesses'), (d) that, (e) add, and a CR (f) to which the source are schema. tically shown in

6.3.1.1 Adding an AGT Referring to Figure 6.1, we see that both type (c) and type (e) derivation procetypicallysses addreferred an Agent to asto "tthera nsiticase visaframetio ofn" thethough source some verbDR's. re Typelating (c class) is I andas "cclassausativisa III jointion ".in withWh ileproce thessesse DR in's type involve (d) tothe formaddi ation group of anof AgeDR'snt kno, theywn dospeci notficati necessarilyons of the involve source any ve otherb asr changethese featuresin the caseare carriedframe feature over to the CRderived's even ve thoughrb. Ina othe"new"r wordsAgent , isthey adde mayd. not require a reinterpretation of the

6.3.1.1.1 Transiti vi sation According to our definition of transitivity, the adding of an AGT or an INS to theboth caseproce frssame (c) of andan proceintransitivss (c'e) verare b trawillnsitiv makeisati it ontra procensitivssese. . The Inre thisfore , AGTsubse toctiothe n, casewe willframes de alof only the witsourceh type ve rbs(c). derivation processes that add an derIn Amivedis, intoneo sigroupmple of tr anssimpleitive intr verbsansitiv of eclass verbs III from are classthe verbs I that that can indicatebe psychthe saologme morphoical statlogicales. formThese is verbscarried are ovetypicallyr with tramarkednsitivisa by thetio pren. fixEx rnaamples- and include rna-olah which means pleased, happy when it is intransitive and like, love when it is transitive; rna -talaw which means fpightened when intransitive and afpaid-of when transitive; and rna-cal ibad which means angpy when sectiointransitivns 5.2.1e and and angp 5.2.2.y-at when transitive. More examples can be found in 252 The reason I am positing a direction of derivation from intransitive to transitive instead of the other way round is that ma- is typically a prefix ::or intransitive stative verbs and that an accusative transitive verb with a ma­ proceinsteassd ofcan thebe statypicalted asmi- foprellowsfix: is highly marked. This transitivisation DR-15 (+V (+V +[ +PAT] +[ +PAT] -[ +AGT] +[ +AGT] +PAT :H- +AGT ) ( aFi ] ) ( aFi ] +Nom +PAT -( -PAT ] -(+spcf ) spcf "specific" +psch +psch psch "psychological" maThetrice eliminations may again of yieldfeatures a sisimplmilaerrly repre markedsentation in both here the. inputSince andthe theope outpu':ration ruleis si mpin lethis and ststudyraig, htfespeorwciallardy, whenI am notthe megoingrit ofto sirepempatlic itity foris evernot obvioy derivat.us. i.on Referring to SR-13 and SR-16, we know that class I psychological verbs are AGTsta tiisve realisedverbs. inBy theRR-2 Genitive, we know case that form an , agentivethen, by verbRR- 3,is thetra nsiverbtiv ise. ergatIfive ':he. onlyOther wisewith , non-spthe verbecific is non-eobjectsrgativ as ein orthe accus follativeowing. exaThemp lederiveds: forms are ·used (6.8) malike-ol ah kako to fibotshin g ls( ) rns Nom+AGT ) ACC+PAT :-e�likergv fishl (-spfc I( l J (6.9) �ma-tialawf ci3s ra to kawas {� Nom +trns (+AGT ) +PAT he-e isrgv afra id of ghosts l-spfc These derived verbsl may be further derivedF�� tSbyI process (g) into class II specific-object intraTnsitives, making it seem as though these verbs had dedirriectlyvation derived can be into sketched class as II fol throughlows: process (b). This sequence of (c) (g) (b) I ):-----+ III >------+ II I )>----+ II We6.3.2.3.1 will deal belo withw. process (b) in section 6.3.1.3 and process (g) in section 253

6.3.1.1.2 Causativisation In section 3.2.3.2 we have briefly reviewed the lexicase claim that the 1974"cause:285r" , of19 78:a deriv12-17ed,40-4 caus2)ative. Fig verbure 3.2is actu illustratesally a gramm Starostaatical's Agent counter­ (Starosta propo1973)sa. l Theto StAmievens data's case easily assignment support forth iscausative claim be cocausenstr, uctionsas we shal(Stevensl see beloevenw, req theuire introduction the reassig nmentof a "nofew" case Ag entrelatio in causatins. ve constructions does not Causative verbs in Amis are morphologically marked by the prefix pa-. AgentSyntac ticallyadded to, theytheir di caseffer framfromes the. irWh noiln-cae theusativ sourcee sourceverbs mayformsor in may having not have an Thea co-occurring "new" AGT introduced AGT, corresponding by the DR derived is referred causative to as verbsthe causeralways ofhave the oneact. ion. Withverbs the. presencDependinge of on anthe Agent case, allfeature causats ofive the verbs source are verbybs de, finitidifferenton tra typensitivs e classof caus IIIative verbs verbs can areall creatserveed. as inputIn Ami tos, caweusativisa find thattio n,class thus I, making class II, and vercausativbs. isation the most productive syntactic derivation process for Amis Corresponding to the broad application of causative derivations, the meaning andsource form verb of . a causativeIn general verb, a varicausativees according verb den tootes the an categ indirecorisatit actionon of wititsh the Agentthe verb causing base, bedirec it tinga change, or aloflowing state theor occurrencean action (DofeGu whatzman is1978 des:3ignated37). by Causativisation relating class I intransitive verbs to class III transitive converbssi deredis in causessenceativisa a trtionans beitivicausesa tiowen caprocen gainss . greate The r degenerivationralisa procetion ssin dois ing traso. nsitAlivisal causativtion procee verbsss des showcribed the characin sectteristicion 6.3.1.1.1 prefix does pa- whichnot prod theuce . Alinterso, pretheretatio isn thof issi mple"indi trarectnsitiv action"e verbs meaning. which is different from the Exfoamllplesowing of: causative verbs derived from class I intransitive verbs include the (6.10) pa-qahen to haw c ira to hemay? provided-fo r, already QM 3s rice ain (by begging) �� [+Adv ] Nom ACC +trns (+AGT ] +PAT-spfc rl+c-eausrgv [ 1 has he been provided fo r with some rice ? (6.11) aka [ pa-qon i ng to cabang ] do+v-not ��e-s me thing-dirty wa( Accll +xlr+ngtvy +f+trnsintl +PA-spfcT +mptv +caus-ergv l don 't make the waj lls dirty 254

(6.12) pa-ta-ini henaca to apol bring,+v cause-to-be-here please li(ACCme (stone) +trns-ergv +PAT-SPfC +mptv+caus please bring (me) some lime (stone) l I (6.13) pa-qisiq ko wawa ano ma-bot iq make-someone-urinate child if sleep +v � � [+v] +trns+ergv � T +caus+mptv ( ] have the child urinate before going to bed (6.14) pa-ka-mo taq-en ko ma-Iasang-ay , caused-to-vomit drunk-person +v (���T 1 +t+ergvrns +m+causptv kia ma-ngaay ko tireng nira so-that we ll body, health 3s [ +v] (���TJ induce the drunk man to vomit, so he 'll feel better (6.15) pa-pi-ol iq-en kam i n i ama niam lotok naci la to ld- to-go-gather- lexcl Fa ther lexcl hill yesterday a -grass Nom Gen :f � +PAT 1 +AGT 1 r+erg+caus: :v our fa ther sent us to( the mountains( to gather alfafa grass Even though all these derived causative verbs have their source in class I and end up with the characteristic causative prefix pa- in all the derived forms, the source verbs can be quite diverse in form. For example, pa-qon i ng make.­ something-dirty has as its source the intransitive verb si-qon ing dirty , pa-ta-ini bring is derived from ta-ini come , pa-qisiq make-someone-urinate from mi-qisiq urinate , and pa-pi-ol iq-en to ld-to-go-gather alfafa-grass from mi-ol iq gather-alfafa-grass . AGTIn theor PATimpe rativein the analysentencesiss,, isexam usuallyples 6.1 understo1 to 6.1od4,. theThe senegativecond person impe, rativebe it whsenticench ise, foi.e.llowed, ex byam plea finite6.11 hascom plemas itsent main. I verbhave thealso au xilconsidiaryered ver ban akaalysing do not the form aka into two constituents, namely, the irrealis auxiliary verb a and thenegative negat iveinfini preptiveosit formsion kato, butgo with then thewe wouldnegativ havee impe to rativehave a, diin ffadditionerent set to of 255 the set of infinitive forms after preposition a and that after the negation openverb caayfor futureand the inv negativeestigation prep. osition ka. I would like to leave this issue Asergative shown byand thethe exaacmcusaplestive, deri subcategved causatoriesive of verbs class areIII. found The in causativboth thee canderivation be sta tedre latingas fol classlows: I intransitive verbs and class IIIb ergative verbs DR-16 +v +v +[ +PAT] >-+ +[ +PAT] -[ +AGT] +[ +AGT] +INS] -[ +INS] +Nom - -( -PAT J [ Gen -(-AGT J +caus (+mptv) I[ k'- i - -+ [ pa- m i- [ 5 ] l[ -+ pa - The[+trns com] binatio(transnitiv of e)contextual and the comfebiaturenation[ s [+ [of+PA NominativeT],+[+AGT ]]Patient is equivalent and Geni tivto e Agentnotations means are [+ erusedgv] in(e thergativ sente)ence. Likeexam plesthe usethrough of CFout la. belsThough, the anshorter AGT is schematintroduceicalld, they represented CR's are not as reintefollowsrpre: ted. The correspondences can be Nominative PAT >-----+ Nominative PAT ¢ >-----+ Genitive AGT TheIlIa causat accusativive ederiv verbsat ioncan rulebe streatedlating as classfollows I : intransitive verbs and class DR-17 +v +PATl >. +[ +PAT] :�+AGT J +[ +AGT] r -[ +INS] -[ +INS J l-[ +ACC -( -PAT J +PAT -(+spfc (���;} J l+caus-(+mptv)

-+ [ pa - 256 [-Theer gv]com bin(non-eationrgative of Nominative or accusativ Agente) and. AccusIn bothativ DR-16e Patient and DR is-17 eq, uivalentan AGT is to asintr sooducedme othe intr ocase the reoutputlation matr evenix though. The, PATin thein thecase sourceof a derivedis not accusatreinterpret.edive causatinterpretedive verb as , beinit gis norealn-spisedecif inic. the accusative case form and is to be Theuni formcausat forivem eitimpeher.rative Some, forms such belo asnging exam plesto class 6.1 4 IIIand do 6.1 not5, seeshowm tothe have a characteristic imperative suffix -en for class Vb, while others just appear in thefeaturesir pa- alreaprefixeddy st atedstems in. DR-16Other and than DR the-17 , caseit isfra dimeffic andult morp tohological make any geneformsra. lisation about the exact morphological shapes of the input and output What we have done so far is to describe the derivational processes represented classas type III (c) derived in Fig urecausativ 6.1.e verbsThese by DR adding's relate an AGTclass to I theintra casensiti framesve verbs of theto sourcealso adds ve rbsa CR. . NowSi wence will the outputlook at of proce processss (d) (d) which are , derivedas shown ca usin ativeFigure verb!; 6.1,, wesup erfassuiciame llthaty, howevean Agentr, desthe igcasenatin framesg the indirectshow that causea LOC isis add addeded. instead. By onlylook ingrepre intosents the anway indirec the CRt 'sde rivare ationalreinte rprlietnk edbetween, we fi ndclass out IlIathat accusprocessativ e(d: verbs and class Vb ergative causative verbs. In other words, process (d) may resultadding offrom th esethe procesuccessssesive canappl beica sketiontched of proceas folsslows (g): and process (e). The (g) (e) (d) I I I >>-----+ II >>-----+ v I II )>-----r) V Our examples show that class Vb causative verbs always have specific objects thethat spareecific manifest objeedcts asof Locative their class Locus II insource the converbsstructio. Asn. a matTheyter corre of factspond, to CRtaking's, si thence path reinter frompr etaclasstio nII has to classalready Vb been makes taken it unnecarecess ofary by tothe reinterpre·t detraspecifnsiic-otivisabjectti onintra procensitivss es(g) to. beHowev deriveder, sifromnce claI cossns IlIidear byall proce classss II (g) , the Screalhema ultticallyimate , sourcesthe re interof thepr etcausativeation of CRderiv's canation be shouldrepresented still beas infol classlows: lIla. IlIa II Vb Nom AGT >>----+ Nom PAT >--­ Nom PAT Acc PAT >>----+ Loc LOC >>-----+ Loc LOC ¢ >----+ Gen AGT Theand classfollowing V. pairsThey areof exaaccountedmples clea forrly by showproce a ssre lati(e). onship between class II (6.16a) kia mi-qonqon I"to saw itakoan? how-come Y'Ush QM ls ] Lev :�lry :-tr�intns +spfc+LOC how( come you are Y'Ushing me ? [ [ l I 257

(6.16b) (pa-)pi-qonqon-en c i Pohang i c i ama-an have-someone-go-rush- Pohang Fa ther someone( ) -else +t+vrns Nom+PAT +LOCLCV +ergv ( ] (+spfcJ +caus+mptv have Pohang go rush/hurry Father (6.17a) mi-alod c ira itakoan thro( w-at 3s ) Is +V-tr ns Nom+PAT LCV+LOC l J J [+spfc 1 he is thr9wing( (something) at me (6.17b) pa-pi-alod-en haw c ira itisoan? allowed- to-throw- QM 3s 2s at-someone ( ) +v+trns Nom LCV +ergv l+PATJ [+s+LOCpfc 1 +m+causptv would you let him throw at you? exaNot mpallle 6.1class5, Vbin derivedthe foll ocausativwing, I e haveverbs exa aremples impera correspondingtive. In additionto 6.16b toand 6.17b to show the non-imperative counterparts of class Vb imperative causative veexamrbs.pl es. The non-second person Genitive Agent is overtly expressed in these (6.16c) (pa) pi-qonqon-en ako ci Pohay ici ama-an have-someone-rush- Is Pohay Fa ther someone-else ( +v+trns +AGTGen Nom+PAT Lcv+LOC +caus+ergv ( J ( ] +SPfC1 I have Pohay (go) rush/hurry Fa ther l (6.l7c) pa-pi-alod-en ni ina c ira tina ma 1 i-an allowed-to-throw Mo ther 3s this ball +v Nom LCV 1 +trns ��T +PAT +LOC +erg+causv ( ( ] [+SPfC} Mo ther allows him to throw this) ball Interestingly enough, class IIIb ergative transitive verbs do not serve as input to causative derivations in Amis even though all class Vb derived catousativ be dire everbsctly dearerived er gativefrom class. All II our sp ecifexamic-oplesbject show intraclassnsit Vbives causat whichive verbs correinsteadspond of toclass class IlIa IlIa as accusathe directtive inputtransi totive the verbs deriv. ationIf we rul cones sithatder classcreate II 258 class Vb causative verbs, the problem of changing from [-ergv] to [+ergv] ervangativitishes.y. Class II verbs are intransitive and are hence unmarked for The derivational processes which produce both the imperative and the non-generalimpe rulerativ eas causatfollowsive: verbs in class Vb can be formulated in one single DR-18 +V +V +[ +PAT] +[ +PAT] +[ +LOC] >+ +[ +LOC] -[ +AGT] < +[ +AGT]> +LOC Gen -(-spfc ] -(-AGT ] +caus < -mptv>

[ mi -+ [pa-pi-

-+ -en] DR-18 adds an AGT toa class II specific-object intransitive verb and create�; Gefromni tivite a caseclass form Vb. causativeThis is verbat once. Thea tranewlynsi tiviaddedsa tionAGT andis re causativialised insat theion procaction"ess . interThepr outputetation is. anThe er procegativess trans as staitivtede verbin DR- which18 is has simp anle "iandndirect addedstraig AGThtforward does not. Nocon reflicintert witpreh tationone which of CRwas's prisevious requiredly pre sincesent the. Thenewly characteristic prefix of class II specific-object intransitive verb mi- is morphopchangedhonemic into the subparts characte ofri DRstic-18 affix. pa-p i - ...-en of class Vb, as shown by t:he ThThereis proce is alsoss belo a causngs ativisto typeation (d) proceshownss inwhich Figure re la6.1tes andclass is notIlIa re tointe classrpretab Va. le theas theca sesum fra ofme proceof thess source(g) and verb process, we will(e). discussSince thit isin proce sectssion adds6.3.1.3. a LOC to Toframes sum upto , crproceeate ssnew (c) ve andrbs . proceWhenss co(e)nsid eaerch edadds together an AGT witto h theproce sourcess (g case) , th ese processes can account for the links between verb classes I, II, IlIa, and Vb. veTherb pathcan betaken schem by atiseda simple as infoltralowsnsitiv: e verb to become a class Vb causative (c) I ) IIIa ->-(b) . (g->) - (d) - .... (e) - .... II1 ) Vb Figure 6.2 The Paths Leading from Intransitive to Causative Wedisc wiusllsi ondeal of with proce prosscess (b) together(g) in se withction proce 6.3.2.3ss (d)and in return sectio ton 6.3a bri.1.3ef below. 259

6.3.1.2 Adding an INS Class I verbs can also add on an INS to become class III transitive verbs. ConThe siderivationder the fol procelowingsses exam wereples repre fromsented class asIIId proce dessscribed (c') in Figsectionure 6.1 5.2.3.2:. (5.63) ma-cocaq kako no anaboq irritated ls dust eye) Nom (Gen +ergv (+PAT ] l+INS ] I[:��:I irr itated (in the eye) am by the dust (5.64) ma-pa l awad ko ba l ocoq noma ko nina rad iw mo ved heartNom ls this songGen :�+ergvrns (+PAT (+INS ] myl heartl is moved by Jthis song These class IIId ma- verbs are clearly different from those psychological verbs derivand eredgativit by DRy.-15 . They differ in terms of case marking, case frame features, toIt hasan AGT, been an mentioned INS can alsoin sect beion added 6.3.1.1 to the unde caser Tra framensi tiviof ansatio intran thatnsitiv, pare verballel beto produceanalysed a astra havingnsitive a veGenirbtiv. e TakenAgent ininstea isolad tionof a, examGenitplesive 5.Instr63 umandent 5.6 si4 ncecan weBut do, ifnot we accepttake the animat wholeeness case as systa critem erintoion co fornsid diserationtinguis, wehing would AGT andfind INSthat. an agentiveare cited analys againis belo is w:imp ossible. Compare example 6.9 with example 5.63 which (6.9) ma-talaw c ira to kawas 3sNom tS -ergv (+AGT ) +P-spfAT c he is afraid of ghosts � (5.63) ma-cocaq kako nol ��anlab oq Frirritated�r ls dust (!� th eye) Nom (Gen +erg+trnsv (+PAT ) l+INS ) Il am irritated (in the eye) by the dust Thoughsubclasses superf by iciallvirtuey Iofthe thetwoir verbs diffe rentlook casesimila markr, ingsthey . belongCorrespondingly to two different they cannot have identical case frames. To derive a class IIId verb like ma-cocaq irrifollowingtated der(in ivationthe eye rule) from: its corresponding class I source, we have the 260 DR-19 +v (+V +[ +PAT] +[ +PAT] -[+AGT] >-+ -[ +AGT] +INS] +[ +INS] l-[ -(+Gen-INS } +psch Theis a detrarivednsitivisa verb tionis by proc definessiti. on transitive and ergative. DR-19, like DR-:S, Weve rbscan, asals othe derive foll owingclass exam IIIcple accuss willativ showe causativ: e verbs from class I quality (6.18) pa-ahcid ko c i 1 aq to kabi e-some thing-saltY salt ) +trns� (Nom+INS} l+PAT �+caus-ergv �-s��pfc 1 the salt makes the soup salty (6.19) (sa-)pa-ahcid ko c i 1 aq to tood used-for-making- salt things �ething-SOltY (Nom ) ACC +trns l +INS J +PAT r+caus-ergv [-spfc 1 salt is used fo r making anything sa lty (6.20) sa-pa-ka-asoq ko c i 1 aq to sinabel used-for-making- salt fo od S · ng-tasty t Nom (Acc (+INS ) l+PAT +caus-er:gv: j -SPfC 1 sa:�lt: is used fo r making the fo od tasty other than saying[ that the characteristic prefix pa- is present in all the dera preivedfixed causative stem, offorms a quality, we do statnot iveknow ve whenrb would and whybe used a root as formthe source, as oppo ofsed to derivation. For example, pa-qoning make-some thing-dirty from example 6.11 has as its source the intransitive verb si-qoning dirty , pa-ahcid make-something-· salty and sa-pa-ahcid used-for-making-something-salty have as their source the verb ahcid salty in its root form, and sa-pa-ka-asoq used-for-making-something­ betasty used use ons aits derived own as stemthe mainka-a soqverb inst ofead a sentenceof the root. Th formis may as oqseem tasty chaoti whichc, butcan itforms is maytypi havecal forcome derivations from differ toent demon sourcestrates by suchanalog sporady. icAnd behaviou, even thoughr. Si mithelar mechaniderivationalsms are proces thereses, ce, rtainthus leavocabuvingla gapsry itinems the wou leldxi notcon . undergo certain 261 Soverbs far through our exam theple addings of cla ofss anIII INS causative are all verbsaccusa derivedtive. Thfromeir class case Iframes qual ityare partypicalallel ofto classDR-17 IIasIc fol. lowsThis: causative derivation can be stated in a rule DR-20 +v (+V +[ +PAT] +[ +PAT] -[ +AGT] >+ -[ +AGT] l-[ +INS] +[ +INS] -(ACC-PAT ) -(+spfc+PAT ) -( +Nom-INS ) l+caus [ k� - +[(sa-) pa- ([ 5 I - ) +[(sa-) pa- Thederives optional an ins pretrumefixntal sa- actuverb allfromy sig a nalscausative anot herstern de. rivationalSince the proce process sswhich does notit here change. any case or morphological features, we are not going to deal with We can combine DR-17 and DR-20 to make a more general statement about derivgeneralationalised traprocensissestivisa thattio n addand ancaus AGTativisa or an tionINS . proce DR-2ss l thatcan beproduce viewed as a accusaare .comtivebin ed.causative verbs from class I. In it, process (c) and process (c') DR-2l (+v +v +[ +PAT J +[+PAT] -[ +AGT] >+ [+AGT] l-[ +INS] + [+INS] -(+ACC -PAT ) -(+spfc+PAT )

+caus

[ pa-

+ [pa- 262 theIn DR-2sourcel, an is AGT not or re aninte INSrpreted is introduced as some other into thecase output relation matrix even. thoughThe PAT, in in t·:le case ofform a derivedand is toaccus be ativeinter pretedcausat asive beingverb , non-spit isecific realised. in the accusative Anotheschematicallyr derivational in Figure pro 6.1,cess thatit takes adds a anclass INS beIIIlo ngstra nsitto typeive verb(f) . andAs adds sho wnan ThINSis tois its not casea very frame productive to make proit acess class in AmIV isinstr. ume Whatnt alit tradoesnsitiv is toe ovveertlrb. y expresectiossn a5.2. notional4 will meansshow: within the case system, as the following examples from (5.70) tomes-en nomako a [mi-pinaro] f up fi ll ( ls T ( ��intJ +ergv ( � J tol sima l �kina si-natoik :�:::-oill this bottle (���sJ (:��T] I fi lled the bottle up (full) with oil Lit. the bottle fi lled up (full) with oil by me (5.72) pinaro- i to bingkos kina koakoq nom i so fi ll shredded-tobacco this r.ipe 2s ACC ] (:��T] :�rns (+INS l+mptv+ergv fir ll your pipe with some tobacco! Asa mege nicasetive frameAGT is. understoodThe accusa tivein example INS is 5.7 to 2,be sointerpreted both of the as verbsthe means have orthe "intermediate cause" of the action. The derivational source of tomes-en fi lled-up is probably a corresponding form from class III; and the source of pinaro- i fi ll is most likely mi-pinaro fi ll , seen in example 5.70.

6.3.1.3 Addi ng a LOC It was mentioned in section 6.3.1.1.1 that process (b) which adds a LOC to classthe former I source tra nsiverbstivisi couldng classvery weI verbsll be theinto sumclass of proceIII verbsss (c) and and the proce latterss (f) , takingII verbs class in ex IIIamples verbs 5.4 and4-5 detra.48 innsit sectionivising 5.2.2 them are into sp eciclassfic- IIobject. Al l the class thatintra, nsforit ivesevery derived specific from-obj classect intraIII nsitisourcesve . verbWe in can cla sayss withII, we con havefidenc a e corresneed aponding separate class proce IIIss tra(b)ns. itive verb with non-specific object, so we do .not In section 6.3.1.1.2 above, we have encountered the derivational process (d) whexamich,ples li ke6. l6procea-6.lss6c (b)and, exarepremplessents 6. l7athe -6sum.l 7c,of procetwo procss esses(d) re. latesAs vershobwn clbyass III to verb class V in two steps, first by process (g) and then by process (e) . 263 Ththeis seeminglyindirect pathmore directgoing frompath IIIfrom to class II, and III fromto V IIwhich to V,will is moreinvolve natural extensiv thane causativereinterpretation verb is of de rivthe edcase from featur a classes. IlISpaecific accusaallytive, whentransit a classive source Vb er, gativeif we acceptreinter propretcess both (d) the as Agent the direct and the path Patient of derivation in the source, we wou in ldad havedition to to the Ifadding we accept of a new the Ageindirent. ctThe path switch of deriv fromation [-er, gvwe] wouldto [+ergv not ] haisve hitoghly re intmarkeerpred. t anyadde d.case features already assigned to class II. Only the genitive Agent is byClass proce Va sscausa (d) tivas e thever bsfoll, howeveowing pairr, are of directlyexamples derived would sugg fromes classt: IlIa verbs (6.21a) mi-ala kako to qe lon eh ls chair (Nom � ACC f�; I 1 +tr-ergvns l +AGT ) (+PAT ) Il fe tched a/the chair (6.21b) pa-pi-alaI cira itakoan to tope r nira have-+v someone-get-something 3sNom lsLCV hatACC 3s +trns (+AGT ) (+LOC ) (+PAT ) +caus-ergv he told me to fe tch his hat (for him) Thereq uircorreses theponde renceinter betprweenetatio mi-n alaof the get orandigina pa-pl AGTi-ala as h aLaC,ve- somthuseone lea-getving-so mroomethi ngfor unchangthe newed. AGT thatIt seemsis introduced that this by derivat the derionivat proioncess proc doeesss not. Thego througPAT remainsh class II asclass an Va,inter witmediateh the sourceinput and. theInst eadoutput, it agr goeseei dirng witectlyh eacfromh ot classher wit IlIah re tospe ct toform theulated feature as foofllo ergatws: ivity. Both are accusative. This process can be DR-22 (+V +v I +[+PAT] 1 +[ +PAT] I+[ +AGT] >+ +[+AGT] -[ +Lac] r+[ +LOC] +AGT +LOC J ( aFj ) J (aFj ) -ergv -ergv +caus [mi- -+ [pa-pi- The introduction of a "new" AGT into the case frame causes the original AGT to berepre reintersentedpreted as fo herellows as: LaC. The correspondences can be schematically 264

¢ >-+ Nominative AGT Nominative AGT >-+ Locative LOC Accusative PAT >-+ Accusative PAT Thaddis a wrapsCR to up the our ca dese scriframesption of ofthe so mesource Amis ver verbalbs. deInri vationalthe next procesectionsses, we that will dealsource with ca sesome frames verbal but dedorivational not add anyproce CRsses's. that reinterpret the CR's in the

6.3.2 Deri vations that Do Not Add CR's In the previous section, we have seen verbal DR's that add a CR to the source carelatiose framens, . i.e.Si, nceAGT, only LOC the, and ba INSsic bycase the re prelationsent, i.studye., , theare PAT, admi andtted innerinto casecase introducedframes for bythe a purposederivational of defining process the mu primaryst be one verb of classthe threees, a innecaser relcaseation relatiosectionsns 6.3. .1 We.1 have, 6.3.1 covered.2, and the 6.3.1.3 DR's that resp addectiv anel AGTy. , an INS, or a LOC in In this section, we are going to examine those DR's that do not add CR's, Theinclu DRding's we the are detra goingnsi totivisa see tionhere proce involvesses prim (g) arilyand (h)the shownreint erinpretation Figure 6.1. of CRreint's erpretedaccompan iedas bysome a otchanghere CRin, perspit isec saidtive to. beWhen "d ema otPATed" actantfrom the is central pothatsi tioitn. is "pWhenromoted" a no n-Patientto the cen actanttral po gesitstio reintn. erpretedThere are as ala soPatie examnt,ples we whe sayre 6.the21 bPAT from acta sectiont remn ains6.3.1.3 unchanged is one wh ofil esuch othe exr amCRpl'ses are. reinterpreted. Example Referring back to Figure 5.2, we are reminded of the fact that classes III, IV, recalland V canthat be thesubcate membergorishipsed of in clatermsss IVaof theverbs feature is large of lyergativity undecided. soWe it alsois futile to talk about derivation to and from an empty class. An interesting splissueit ercongativcernes thelang uagerelationships such as Am betisween. Inaccusativ section e 6.3and.2.3, erga tivan eat verbstempt inis madea to show that class IlIa is related to class IIIb, and class Va to class Vb, onlytypes in, nam an elyindir, (g)ect and way ,(h) via as the repre procesentedss of in detra Figurensit 6.1ivis wouldation be. studieTwo proced. ss Other(a) that proce wassses also that schem involveatically a rerepreintersentedpretation in Fiofgure CR 's6.1. inclu de process type

6.3.2.1 Acquiring a PAT Subject Thesimp rele latedneintransssitiv betweene verbs class can VIbe expreimpersonalssed by intra a DRnsitiv of typee verbs (a) . and Ascl weass have I deranalysedivation the proce impessrsonal linking verbs them as doesbeing not su bjeadd ctlesson a PAT but. notInst patientlead, theess pro, thecess carenlatio be nlooked in the upon input as matrixone inv olvas ingthe Patthe ientreinter subjecpretationt. There of the is TIMthe orapparent PLC case feature change from [-[ +Nom ]] to [+[ +Nom ]] and the omission of the co-occurring TIMrelation or PLC, isin netheces outpsarilyut . markedThe onPAT thecase Nominative relation , actantbeing theof thebas derivic caseed verb. 265 Normalmodel. ly, AsTIM such and, theyPLC are do notcons particideredipate oute r incase the recharaclationsterisat in theion lexiof verbscase or Thverbis classesassumption and po henceses someshou dildffic notult be y inclufor dedthe preas inputsent anfealysaturesis whichfor DRwould's. have ofto theimpo inputse an matrixinner ro whichle inter getspre reintertationpre fotedr TIM as andNominative PLC since Patient they are in the partcase classframe ofI verbs the derived canbe verbstated. asThe fo dellorivatiows: nal relationship between class VI and DR-23 +V +V -[ +Nom ] (+[ +Nom ] -[ +AGT ] >-+ -[ +AGT ] <{ +PIC l +PAT +[ +TIMJ J ClFi J ClFi ] [+Lev[ l (� ) [ ] ( :�� J The correspondence bel t-ween(=;��) the locative l-Locus or Time and the nominative needsPatient to isbe shownmade bybecause the "hthiorses shoeis a " zernotatioo derivn. ationNo . morphophExamplesonemi inc cludestatement: (6.2la) si-kawas l lomaq � �re-are-ghosts house � LCV +Nom ] (+PLC) -[ +AGT ] ] therel-[ are ghosts in the house (6.2lb) si-kawas2 kina lomaq ���re-are_ghosts this house +[ +Nom ] �� T -[ +AGT ] ] : there are ghosts in this house( Lit. thisl house is haunted J orad-an it ina rem iad (6.22a) 1 r-ain ) this dayLCV +Nom] +TIM] -[ +AGT ] ( it[�[ is rainingJ today (6.22b) orad-an2 kina remiad ra'Z-n this dayNom +NOm] +PAT :C-[ +AGT ] it is raining today( Lit. today( is j raining J 266

(6.23a) toqmanl parod dark kitchen +v 1 LCV -[ +Nom] (+LOC ] [ -[ +AGT ] it is darkJ in the kitchen (6.23b) toqman2 ko parod dark kitchenNom +Noml +PAT ] -[ +AGT ] it is dark in the( kitchen Lit. the[:[ kitchen] is dark

6.3.2.2 Passive Derivation Amis transitive verbs can be subcategorised with the feature [ ±ergv ] . As accusatshown inive Fig andure er5.2,gativ verbe subc classeslasses III. , WhIV,ile and class V canIlIa each simple be dividedaccusative into transitive verbs are uniquely characterised by the prefix mi-, class IIIb cosimpntleributes ergative to thetra nsitsplitive er verbsgativit arey ofmarked the labyng eituageher. We ma- haveor -en an. alysedThis the -en forms to be pa$sive verbs corresponding to the mi- active verbs in class laIlIa,nguag parcees. lling out the ma- forms to a separate domain belonging to ergative Thecan bere latednestatedss by be thetwe enfo ll-enowing pass derivationive forms andrule the: ir corresponding active forms DR-24 +V ) +V +[ +PAT] +[ +PAT ] +[ +AGT ] > ..... +[ +AGT ] r +Gen 1 ���;J -(-AGT ] +ACC +Nom l-( -PAT ] - -PAT( ] +pssv pssv "passive" [mi­

] ..... -en ] evenThe CR though's in the ircase assoc fraiametions are notwith recaseinter formspreted have as somebeen othechangr edcase. Therela tifeatureons [+pssv] (passive) is neither a case feature nor a semantic feature, but it i�; verbsneeded and fo r thosedisti nguithat shingare er begattweenive tothe begin "pass witiveh." forms corresponding to "active" Exfoamllowingples of sets act ofive sentencand pasessiv: e verbs related by DR-24 are given in the 267

(6.24a) mi-taes c ira to wawa n ira hit 3s child 3s +v (Nom (ACC T +trns-ergv +AGT J +PA J -pssvJ he hits1 his child (6.24b) taes-en nira kia waco hit 3s dog +v Gen Nom +trns (+AGT J (+PAT J +ps+ergvsv the dog was hit by him or , he hit the dog (6.2Sa) m i -tenge r kami to olad no ra rapa cook (until tender) gristle of beef 1 lexcl v Nom ACC +trns (+AGT (+PAT J -pssv-ergv J we are stewing beef gristle (6.2Sb) tenger r-enJ a [mi-cacak] cook+v (until tender) cook (in water) +trns (��i!1tJ +p+ergssv kia tamorak nia babahi pumpkin woman (:��T) (�:�T] the pumpkin is being cooked by the woman or, the woman is cooking the pumpkin (6.26a) mi-ki 1 im c ira to badal look-for 3s berry [+v NomT ] ACC ] +trns-ergv ( +AG (+PAT l"-PSSV he is looking fo r berries (6.26b) k i 1 im-en n i ra ko toper (n i ra) looke+v d-for 3sGen hat 3s +trns (+AGT ) (:��T] +p+ergvssv his hat is sought by him or, he is looking fo r his hat 268 It is to be noted that the passive derivation is not very productive in Amis. Notcanno allt findmi- verbsan Am isha ve-en a passivcorrespondinge to trans passivlate eint formo, it. usuallyIf an Engl can isfindh pas a sivmae­ ertheregative is eiformth erto anse rve-en pasthe spurpive oseverb. or Inan anyer gativecase, correspondingverb marked by toma- amini -theve r:::>, lexicon. In some cases, as exemplified by the trio of melaw look� see verbs: mima--melaformsw seeare , maavai-mellableaw see. byWe , canand seeme law-enfrom thewatch� gloss look that af meterlaw-en, both has-en and unde"pasrgsiveone" measemaningntic. shift. The language does not need two forms to carry the Both the ma- verbs and the -en verbs are ergative and they share identical case leaframvinges. theThe ma- devicategorysion to ascall basic the (of-en courseforms , "pitass alsoive " inclandude desrived a derived while membederivationalrship) ispot basedential on. theThe co mansider- verbsation ca nthat be detrathe ma-nsitformsivised have by procea richersses on(g ) theand ot (h)her and hand can, seems be causativised more peripheral by proce as ssesit parti (d) cipateor (e)s . in Thederivatio -en suffixnal , proceimperativesses thatforms create (see verbssectio withn 6.3.2.5 implied belo instruw). ment (see section 6.2.3) and Typologically speaking, passive verbs are accusative verbs. In the present anframealysis and, passcase ivemarkings verbs , arethey marked belong [+ erto gvthe] because category forma of ergativelly, in termsverbs of. caSE

6.3.2.3 Ergativisation accusatiIn beingve a andspl iter gativeergative constructio language, nsAmi. s opeIt hasrate beens wit hshown a mixed in sec systtionem 6.3.2.�of : formsthat the. Whereasaccusative the verbscase inframe Amis and have mark correspondinging of the active "acti vefor" msand are "pass clearlyive" theaccusativ ergativee, thever bscase. frameThe onlyand markdifferingence of bettheween "pa ssivthee" two forms typ olreogisembcallyle that of diffis markederent bycon st-enructions while a istr uein theergative mark ingverb on isthe marked verb byform ma. -. A "passive" verb In thisthe previous section , sectwe areion , goweing have to givenshow how exa, mpleswithin for a thespl itpas ersivgative derivate systemion, . accusto be ativethe sourceverbs ofand derivation ergative verbs, we canare termrela ttheseed. procAssumingesses the"un- mi-passive"accusatives Thederivations output ofin an contrast "un-passiv withe" thederivation passive dein riAmvatisio isns ex whichpected yield to be pa eitssivhere ver anbs . er(sectigativeon 6.verb3.2 .4)(sec . tion 6.3.2.3) or an intransitive verb unmarked for ergativity Here we would be making a big claim about the proto-language and its devefar aslopm weent can, if tr weace take, Pro erto-Austrogativisanestioian n liisterally in itse aslf a ashisustoricalmed to be event a spl. itAs mutuaergativelly laindependentnguage (cf. accusative Starosta , andPawley ergative, and verbsReid, . 1982)Si ncewit hI amco- existingnot ready butot makeinto theergativ claeim verbsthat, priorthe DR to prop Proto-Austroosed here nesiais to n,be accusativeregarded asverbs noth deveing moreloped thanlanguage a word-. Forcedforming by anal theogy fle whichtched may arrow be openotationrative, Ifor am so onlyme speakersmaking theof the 269 assumption that the word-forming analogy operates from the accusative to the toerg ativetell in, ifthi a s dircaseec tiowhen thermust the be assuascertainmed directed. ionBe siredesflects, it ais nativreallye speak difficerult's associafrequentitlyon asfor class the IIre lasptedecific forms-obj. ectSi nceintra classnsitives IlIa" mi-it isverbs the occurclass moreIIIb whichergative appe vearrbs more, inc frequeludingntly the in"p assthe ivetra" nsiverbstive andverb the category true er. gative verbs, Theclass fo llIIIowingb ergative DR seemingly verbs: relates class IlIa accusative verbs to corresponding DR-25 +v +v +[ +PAT] +[ +PAT] +[ +AGT] >-+ +[ +AGT] +Gen -[-AGT ] 1 +ACC +Nom - -PAT( ) -( -PAT ] -pssv pssv "passive" [mil-(� -:�;l -+ [ ma- DRare-25 produced looks very in one much si nglelike DR-2step4, as but repre I amsent noted byconvinced DR-25, thatif er ergatgativeive verbs verbs are verbsindeed inder clivedass fromIlIa thefirst accusa undergotive detra. Itnsiti is morevisa tioliken, lyand that then the de accusatrive intoive classschematised IIIb erg as ativefollows verbs:' by a transitivisation process. The path can be (detransitivisation) (transitivisation) > ------+ I > ------+ IIIb IlIa (h) (c) - In fact, this turns out to be the more natural way, as these processes can be usedand the el sesiwheremple intrato accountnsitiv efor ve therbs . relaSitionshince I psdo betweennot have thea separatesimple trasectionsitivn efor prederivationssented in thatthe nextdelet esec CRtion's, thetogether detra withnsit ivithesa detiotran procensitivisssa (h)ti onwill proce be ss (g).

6.3.2.4 Detransitivi sation In the previous section, detransitivisation is seen to provide a link between CRclass. IlIa and class IIIb. Detransitivisation is a process which deletes a In this section, DR-25 is reanalysed as DR-26 and DR-27, representing the two stepsthrough lea ading neutral from terriclasstory IlIa which accus isativ classe verbs I. to class IIIb ergative verbs 270 DR-26 +V +V +[ +PAT] +[ +PAT ] ] [ ] +[ +AGT >-+ - +AGT +AGT +PAT aFi aFi J- (+Nom-AGT } J- (+Nom-PAT )J -(+ACC-PAT } l ( } DR-26 is a zero-derivation( . } The morphogical change into the characteristic ergativdescribede ma- as prefollfixows probably: takes place during transitivisation, which can be DR-27 +V +V +[ +PAT ] +[ +PAT J

- � [ ] [ +AGT ] > + +AGT f +PAT +AGT OFi aFi ((+ Nom } (+Nom) lJ- l-PAT J- -PAT } -(+Gen-AGT ) [m i- � [ ma- ( } Theto theoutput cla ssof ofDR-2 derivations7 are class that IIIb add er gativea CR, sptraecinsitficallive y verbsan AGT. , DR-to 27the becaselong�; Actuframeally of , theas thesource follo verbwing. correAccordingsponden toces Figure show , 6.1,the seeminglyit belongs addedto type AGT (c)is . fromis add thede . original PAT through reinterpretation. It is a notional object that Nominative PAT >-+ Genitive AGT ¢ >� Nominative PAT The following set of examples show the relationships between class IlIa (or vercorrebssponding. class II specific-object intransitive), class I, and class IIIb (6.27a) mi-l iakawaYl kako tina kawp ir-an P( lsNom this tendLCVep -Leaves ns +AGT 1 +LOC -ergv ( ( J Il :�:pick l the tendep Leaves (6.27b) mi-l iakawaY2 kam i lotok pick lexclNom hiLCVll ) � rns +PAT 1 +PLc we( �pick) (leaves() in the hills( ] 271

(6.27c) ma-l iakawaY3 toay niam k ira kawoi r [+aZrAdeadyv] leGenxcl that tender-Nom Zeaves +ergv +AGT +PAT we have aZready picked the( tenderJ Zeaves( J Lit. p.[the:�:::l tender Zeaves aZready picked by us Having seen a detransitivisation process which deletes a CR, we are going to prethe sentproce a ssdetra whichns ittakeivis ationa class proce IlIass whichverb andonly turns reint iter pretsinto a CRcorresponding's. This is Figurespecific- 6.1ob asject pro intransicess (gtiv) . e veItrb is. a Thveryis proceimportantss is procewhat sswe be havecause promi, likesed in DRa bridge-26, the between detransiti the viaccussationative proce andss the (g er) whichgative will syst beems profes entedthe la hernguagee provides which prodo notcess mi accountsx. It wasfor theshown indirect in sec tionlink 6.3.1between.1.2 cla underss I causativisand classation II, andthat, more this importantly, between class IlIa and class Vb. Since intransitive verbs are unmarkedcross over for to er thegativity other. they provide a neutral ground where one system can Process (g) can be formulated in the following rule: DR-28 +v +v +[ +PAT] +[+PAT] +[ +AGT] -[ +AGT] -[ +LOC] >-+ +[ +LOC] +AGT +PAT :> aFi :> aFi -(. +Nom-AGT ] -( +Nom-PAT ] ( +PATJ ( +LOC :> BFj :> BFj J -(+ACC-PAT ]J -( +Lcv-LOC J Process (g) or DR-28 can( be illustrated by( the ] following pairs of examples: (6.28a) mi-kil iml c ira to badal 3sNom beACCrry -ergv +AGT +PAT he is Zoo king( fo r J berries( J (6.28b) mi-[:�:ki: l im2 cira itisoan Zook-flor 3s 2s J � J �rns : :T ���C he( � is Zooking( fo r you( ) 272

(6.29a) ma-olah1 kako to badal berry " , ls ;:; Nom ACC +trns-ergv r (+AGT 1 (+PAT 1 Il like berries (6.29b) ma-olah2 kako itisoan like, love ls 2s ( � rnsJ (Nom+PAT (LCV+LOC J I �love you 1 (6.30a) ma-ta1awl c ira to kawas 3s ghost (Nom (ACC J -ergv +AGT 1 +PAT he is afraid of ghosts (6.30b) ma-ta1aw2 kako itira tamdaw-an l:�::rafra+v id-o f Nomls that manLCV rl-trns 1 (+PAT J (+LOC J I am afraid of that man Whencausat theive output verbs , ofthere DR-28 is isno usedneed as to thechange input the for feature derivin ofg ergclassativi Vb tyer gativefrom [ -ergv] to [ +ergv ] which the system does not seem to tolerate. Process (g) notto theonly erg relaativete s domainclass ofIlIa the to Am classis la nguageII, but. also joins the accusative doma:_n

6.3.2.5 Imperative Imperative verbs· in Amis are treated as derived verbs because there is a ch.;mge rein lacateses theframe im perativefeatures and thehence non verbal-impe rativesubcat verbegory formsin the. process which withIn secti nominativeon 5.2.5.1 LOC above in im, perativewe have exprseeness accusations. iveThough transiti thereve locais tnoive add verbsition or deextendedletion ofcase CR fra's,me the. reCo-alignsidernment the offol CF-CRlowing mapp exaingsmple changess: the picture in the (5.81) pabel i kako iciraan (-an) to [cacay a codad] g ls 3s one book (Nom ) (LCV ) (ACC -ergvns +AGT +LOC +PAT 1 I[:� gave l one/a book to him (6.31) pabe: 1 i - i henaca cira to codad (haw) ? g please 3s book QM (Nom ) ACC ) ns +LOC ( + PAT +mprt-ergv wol:�uld: youl please give him a/the book? 273

(5.84) pa-sebanaq-i kako tia ngangan nia tamdaw let+V ...know� tell lsNom nameACC of person +trns (+LOC ) (+PAT ) +m-eprtrgv [ka-ki 1 im-en i so] looked-for 2s : T) te ll me the name(� of� the person whom you are looking fo r Theclass corre Va canspon bedences schemat betweenically the represented source and asthe fol derilowsved: imperative verbs of Nominative AGT >-+ (Genitive AGT) Locative LOC >-+ Nominative LOC Accusative PAT >-+ Accusative PAT and can be stated in the following derivation rule: DR-29 +V +v +[ +PAT] +[ +PAT ]

[ ] > -[ ] + +AGT -+ +AGT +[ +LOC ] r+[ +LOC ] +Nom +Nom -(-AGT ) - -LOC( ) J +caus

-+ - i ] coInns cotructionsnclusion, arein threislated chapter to one we anotherhave dem andonst howrated the howse intsomeer-sen Amistence verbal reisl notationships meant tocan be be anstatedexha ustiv in dere studyivation of verbalrules inder theivations lexicase in modAmisel though. Th isit 6.2)is be, lithoseeved that , addin prea CRsenting (secti onDR 's6.3 that.1) , incandorporate those thatcase only notions reint (serecprtioetn CR's orthe changeimportant the iraspe CF-CRcts ofcorre verbalspondence derivational (section proc6.3esses.2) , we. have covered most of The issue of Patient centrality has been raised. It seems that the Amisdata neitdoes hernot strfavourongly proc suppoessesrt nor that refute involve th isCR- hypothereintersispreta. tionIn ge. neralEven, witthe h laa nguage comstayspli catedwithin proce the sssame li keer gativitycausativi domsatiainon. , forIn otheexampler words, the, derivlanguageati onsalways neve r detrachangensi thetivisa featuretion proceof ergatissesvit. y,That even isthough to say it, thecan sourcebe neutral of classised Vathrough causative verbs have to be accusative (class IlIa) and the source of class Vb ercaugativitsative yverbs (class are II) ei. ther ergative (from class IIIb) or unmarked for Sichalncelenging Amis isand rich mean ining itsful morpprojecthologica to stl formsudy the of deaffrivationalixation, itrelati is aonsh veryips beginamongning. the syntactic categories of the language. This work is but a modest BIBLIOGRAPHY

ABE, Akiyoshi 1930 BangStudyo ofKenkyuk languagesai of native tribes . (In Japanese.) Taipei: ACSON, Veneeta 1979 Alexicase diachron analysic viewis . of Disscasee-markingrtation: Univsystemsersit iny Greekof Haw: aiia loca. listic­ AKMAJIAN, Adrian and Frank HENY 1975 AnCa mbridgeintrodu, ctioMassn .:to The the MIT principles Press. of transforma tional syntax. ANDERSON, John M. 1971 CambridgeThe gramrn6rUnive of casersity: Presstowards. a local istic theory . Cambridge: 1977 OnLondon ca se: Humangramrn6itiesr: pr olPregessomena. to a theory of gramrn6tical relations . ARONOFF, Mark 1976 Wo rd forma tion in genera tive grammar. Linguistic In quiry Monograph Series No . 1. Cambridge, Mass.: The MIT Press. ASAI, Erin 1936 A study of the , an spoken on Bo tel To ba go Isl and . Leiden: J. Ginsberg. 1953 Thede KanazaSedikwa lang, Kanazauage ofwa FormoUniverssa ity. Kanaza. wa, Japan: Cercle Linguistique BENDIX, Edward Herman 1966 Componential anal ysis of general vocabul ary : the semantic structLre ofIndia a naset Unofiv ersverbsity in. Engl ish, Hindi , and Japanese . Bloomington: BENEDICT, Paul K. 1966 Austro-Thai. Beha vior Science Notes 1/4:227-261. New Haven. 1967a Austro-Thai2/3:203-244 . studiNewes Hav: en1. . Material culture. Beha vior Science Notes

274 275

1967b Austro-Thai studies: 3. Austro-Thai and Chinese. Behavior Science Notes 2/3:275-336. New Haven. 1975 AustrNew Haveno-Thai: Human: lang uagRelae tioandns cul Areature Fil, wiesth, Inc a gl. ossary of roo ts . 1976 Formosan reflexes of PAN nasal/orals. In Liem, ed. So uth-East Asian linguistic studies, vol .2. PL , C-42: BENNETT, David C. 1974 Spa tial and temporal uses of Engl ish prepositions - an essay in stLonradon:tif icaLongmantional Group semant Limicsit. ed.Long man Linguistic Series No. 17. BENTON, Richard A. 1971 PangPressasin. an reference grammars . Honolulu: University of Hawaii BLAKE, Barry J. 1977 CaseInstit mauterking of Aboriginalin Austral ianStudie languas. ges . LS23. Canberra: Australian BLOOMFIELD, Leonard 1917 LitTa galoeratureg texts 3/2- wi4.th grammaUrbanati: calUnive analrsit ysyis. of IlliStunodiesis . in Language and BLUST, Robert A. 1970 Proto-Austronesian addenda. Oceanic Linguistics 9/2:104-162. 1972 Proto-Austronesian addenda with cognates in non-Oceanic Austronesian languages: a preliminary list. Un iversity of Ha waii Wo rking Papers in Linguistics 4/8:1-43. 1974 Proto-Austronesian syntax: the first step. In Pa pers of the First In terna tional Conference on Compa ra ti ve Austronesian Li nguistics , vol .2: (Pro to-Austronesian and We stern Austronesian) . Oceanic Linguistics 13:1-15. 1977 The Proto-Austronesian pronouns and Austronesian subgrouping: a preliminary report. University of Hawaii Wo rking Papers in Linguistics 9/2:1-15. BRUGMANN, Karl 1884 Zur Frage nach den Verwandtschaftsverhaltnissen der indogermanishchen Sprachen. Interna tionale Zei tschrift fu r allgeme ine Sprachwi ssenscha ft CAPELL, Arthur I. 1964 Verbal systems in Philippine languages. Ph il ippine Journal of Science 93:231-249. 1974 Subgroups in Western Austronesian. Paper presented at the First HonoInternationallulu, Haw aiiConference, January on 2-4 Com. parative Austronesian Linguistics, 276 CENA, Resty M. 1977 AlbertaTagalog , countereCanada. xamMimpleseo . to the accessibility hierarchy. Leduc, 1978 Patient primacy in Tagalog. An earlier version read at the DecemberLinguistic 28-30 Societ, 19y 77of. America Winter Meeting, Chicago, Illinois, CHAFE, Wallace L. 1970 MeChaningicago andPress the. structure of langua ge . Chicago: The University 0:: CHAI, Chen-kang 1967 TaCambiwanridge aborigines: Harvard - Univ a geersneticity stuPredyss . of tribal va riations . CHANG, Kwang-chih 1956 A brief survey of the archaeology of Formosa. Southwestern Journal of An thropol ogy 12/4:371-386. 1963 7/1-2Introduction:195-202 to. special Taiwan section. Asian Perspec ti ves 1966 Preliminary notes on the excavations in Formosa, 1964-65. Asian Perspec tives 9:140-149. CHANG, Kwang-chih and Minze STUIVER 1966 Recent advances in the prehistoric archaeology of Formosa. Proceedings of the u.s. Na tional Academy of Sciences 55/3:539-543. CHEN, Chi-lu 1961 TaiWo od-carvingwan: Natio ofnal the Taiwan Pa iw anUn iversgroupit ofy. Ta iwan . (In Chinese.) Taipei, 1965 Age organization and men's house of the Formosan aborigines. Bu25-26lletin:93 -111of the. TaipDepartmentei, Taiwan of : ArNationalchaeol ogy Taiwan and An Univthropolersityogy. 1968 TaiwanMaterial Museum culture. of the Formosan aborigines. Taipei, Taiwan: CHEN, Chi-lu and Michael D. COE 1954 An investigation of Ami religion. Quarterl y Jo urnal of the Ta iwan Museum 7/3-4:249-262. Taipei. CHEN, Teresa M. 1969 NationalAmis texts Taiwan. 75 ppUniv. ersDepartmity, Taipent ei.of ArchaeologyMimeo. and Anthropology, CHICHIIWA, Suketaro 1960 Houses of the Fo rmosan mo untain tribes . (In Japanese.) Tokyo. CHOMSKY, Noam 1965 Aspects of the theory of syntax. Cambridge: The MIT Press. 277 1970 Remarks on nominalization. In Roderick A. Jacobs and Peter S. Rosenbaum184-221. , Wathameds, Re: adingGinn. s in Engl ish transforma tional grammar , CHRETIEN, C. Douglas 1951 Oceanic material in Bu-hwan, a hill dialict of Formosa. In Walter J. Fischel, ed. Semi tic and orien tal studies, a vol ume presented to Wil liam Popper, 65-77. Berkeley: University of California Press. CHU, Yuan-teh 1975a Amis texts. 167 pp. Taipei. Mimeo. 1975b Amis structure. 96 pp. Taipei. Mimeo. CLARK, Marybeth 1978 Coverbs and case in Vietnamese . PL , B-48. CONSTANTINO, Ernesto 1965 The sentence patterns of twenty-six Philppone languages. Lingua 15:71-124. 1971 Taged. alog1971 and:112 othe-154r. major languages of the Philippines. In Sebeok, COOK, Walter A. 1971 2:Improvem10-22. entsWashington in case , grammarD.C.: George1970 . townLanguag Univeesrsity and LingPressui. stics DAHL, Otto Chr. 1976 Pro to-Austronesian . Second revised edition. (Scandinavian StudenInstitutetlitt ofe raturAsian; StuCurzondies Press Monogr. aph Series 15.) Lund: DEGUZMAN, Videa P. 1970 A1/1 syntac:33-60tic. Quezoncompar isCityon of: Un Tagaiverlogsity, Maori of the and Ph Chamilippinorro.es . The Archi ve 1978 Syntactic deri va tion of Tagalog verbs . Oceanic Linguistics HawSpecialaii. Pu(Ublinivcationersity No of.1 Hawaii6. Honol disserulu: tationThe Un, iv1976ersit.)y Press of DIXON, Robert M.W. 1979 Ergativity. Langua ge 55/1:59-138. DYEN, Isidore 1963 The position of the Malayo- of Formosa. Asian Perspectives 7/1-2:261-271. 1965a AIndia lexicostana Univetisrsticality Puclblicassifationsicat ionin ofAnth theropo Austrlogyone andsian Lingui langsticuagess, . Memoir 19. (Supplement to Interna tional Jo urnal of American Linguistics 31/1.) Baltimore: The Waverly Press. 1965b Lexicostatictics in comparative linguistics. Lingua 13:230-239. 278 1965c Formosan evidence for some new Proto-Austronesian phonemes. Lingua 14:285-305. 1969 Reconstruction, the comparative method, and the proto-language uniformity assumption. Language 45:499-518. 1970 Maranao and Tagalic: a qualitative confirmation of a subgrouping hypothesis. Philippine Journal of Linguistics 1/1:1-11. 1971a Theed. Aus1971:trone5-54sia. n languages and Proto-Austronesian. In Sebeok, 1971b The Austronesian languages of Formosa. In Sebeok, ed. 1971:168-199. EGEROD, S¢ren 1965 Verb inflection in Atayal. Lingua 15:251-282. 1966a and word classes in Atayal. Langua ge 42:346-369. 1966b A statement on Atayal phonology. In Artibus Asiae Suppl emen tum XXIII (Fel icita tion vol ume for the Se ven ty-fifth Birthda y of Prof. C.H. Luce) 1:120-130. 1971 The origin of headhunting - an Atayal text with vocabulary. Bulletin of the Insti tute of History and Philology 39:291-325. FAGAN, Joel L. 1979 MastPreleriminaries's thesis to: Una ivfieldersit sty udyof Hawaiof Mono-i. Alu, Solomon Islands. FERRELL, Raleigh 1966 The Formosan tribes: a preliminary linguistic, archaeological anc. cult21:97-12ural 7.synthe Nankasisng,. BulTaiwanletin: Academia of the In Sinicastitut. e of Ethnology 1969 Taiclassiwan ficatioaboriginaln. Taip groueips: : Institproblemsute ofin Ethnologycultural and, Academia linguistic Sinica . 1970 Paiwan stem vocabulary (draft) . Mimeo. 1971 ErgativeIndonesia, npass langiveuag ores . focusCol-mark1oque ing?International - elicita tiodun Centreof verb National syntax in dedes laLangues Recherche sans Sci Traditionentifique: Ecrite Methode. Nics d'e, EnqueteFrance , etJune de Desc28-Julyription 2. 1972 Verb systems in Formosan languages. In Jacques Barrau, Jacquelir..e M.D. Thomas and Lucien Bernot, eds, Langues et techniques , na t ure et societe , vol.l, 121-128. Paris. 1979 Construction markers and subgrouping of Formosan languages. In Liem, ed. 1979 (SEALS 3) . PL , C-45:199-211. 1976 Pho1979nological:241-254 . subgrouping of Formosan languages. In Naylor, ed. FERRELL, Raleigh and Patricia STANLEY 1979 Aused. trone1979sia:19-31n vs.. Intro-European: the case against case. In Naylor, 279 FEY, Virginia 1962 AmMimiseo word. list, central dialect (Chengkung). Taichung, Taiwan. 1964a VerTaichubs-ngadje, Taiwanctives . (AmiMims)eo: . aspects of action or attribution. 1964b ValencesTaichung , inTaiwan Amis.: theMim eovowe. ls a i 0 are important focal point. FILLMORE, Charles J. 1966 TowardReport Noa modern.13. Colu theorymbus of: Ohio case sta. Prte ojectUnivers onity Ling. uistic Analysis 1968a The case for case. In Emmon Bach and Robert T. Harms, eds, Uniandversa Winslston in. linguistic theory , 1-88. New York: Holt, Rinehart 1968b Lexical entries for verbs. Founda tions of Language 4:373-393. 1970 Types of lexical information. In Ferenc Kiefer, ed. Studies in syntaxPublishing and semCo. antAlsoics , in109-13 Danny7. St eDordreinberqcht and, Hol Leonla ndJacobo: D. vitsReidel, Semantics : an interdisciplinary reader in philosophy, linguistics , anthropol ogy , and psychol ogy , London: Cambridge University Press. Also2:65-103 in Ohio. Sta te Un iversi ty Working Papers in Linguistics 1971 Some problems for case grammar. In Richard O'Brien, ed. Linguistics and language study , 35-56. Also in Ohio State University Wo rking Papers in Linguistics 10:245-265. 1977 The case for case reopened. In Peter Cole and Jerrold M. Sadock, Neweds , YorkGra mma: Academictical rel Pratessions. (Syntax and semantics) vol.8:59-82. FOLEY, William A. 1976 ComUniveparativersity ofsyntax Califor innia. Austrone sian. Ph.D. dissertation, Berkeley: GONZALES, Andrew B., ed. 1973 Parangal Ka y Cecilio Lopez (Essa ys in honor of Cecilio Lopez on his 75th birthday) , Philippine Journal of Lingui stics Special MoPhilnographippin esNo. .4. Quezon City: Linguistic Society of the GRACE, George W. 1953 The posi tion of the Pol ynesian languages wi thin the Austronesian in(M alaAnthroyo-Polpologynesian)y and laLingnguauisticsge fam, ilMemoy. irIndia 16. na (SUnupivplementersity Putoblic ations MDIn t: ernaWavetiorlynal Pr Journaless. of American Linguistics 25/3.) Baltimore, GRUBER, Jeffrey S. 1965 StIndiaudiesna Univein lexicalrsity Lingrelauisticstions . ClUbMIT diss. ertation. Bloomington: 280

1967 Look and see. Language 43:937-947. GUNTHER, Hartmut 1975 Valency, case forms, and case relations. University of Edinburgli Wo rk in Progress 8:64-83. HALLE, Morris 1973 4/1Proleg:3-16omena. to a theory of word formation. Linguistic Inquiry HALLIDAY, Michael A.K. 1967 Notes on transitivity and theme in English. Jo urnal of Linguistics 3/1:37-81 and 4/3:179-215. 1970 Language structure and language function. In John Lyons, ed. Ne�' horizons in linguistics , 140-165. London: Penguin. HARMON, Carol W. 1977 disserKagayanentation and, Untheiv Manoboersity subgroupof Hawai i.of Philippine languages. Ph.D. HARVEY, Mark David 1980 NatioSubgroupsnal univein Aursstityrone, Canbesian. rrB.a. A. UpdatedHonors vesubthersionsis pu, blisThe hedAu stral1982 imlin AmranThird HaliIntem,rnational Lois Carrington Conference and on S.AustroA. Wurm,nesia edsn Ling, Papersuisti csfrom, vo thel.; �: 47-99. PL , C-75. HAUDRICOURT, Andre-G. 1962 Commenton A. Capell, Oceanic linguistics today, Current Anthropol ogy 3:410. 1965 Problems of Austronesian comparative philology. Lingua 14: 315-3;�9. HIDALGO, Cesar A. and Araceli C. HIDALGO 1970 The structure of Ivatan : phonol ogical , lexical , and gramma tical componen ts . Quezon City: University of the Philippines. HOPPER, Paul J. and Sandra A. THOMPSON 1980 Transitivity in grammar and discourse. Language 56/2:251-299. HUDDLESTON, Rodney 1970 Some remarks on case grammar. Linguistic Inquiry 1/4:501-510. HUDSON, Richard A. 1976 Argumof Chentsicago forPr essa no. n-transformational grammar . Chicago: Uni versi 1:y IKRANAGARA, Kay 1980 Mel ayu Betawi grammar. Jakarta: NUSA. (Linguistic studies in Indodissernesiatation and, la1975nguag.) es in Indonesia , vol.9.) (University of Hawaii 281 INO, Yoshinori 1916 Characteristics of the languages of Formosan aborigines. (In JapanSocietesy e.)of Tokyo)Jinruig 31aku. Zasshi (The Journal of the Anthropological JACKENDOFF, Ray S. 1972 SemPressantic. in terpre tation in genera tive grammar. Cambridge: The MIT JENG, Heng-hsiung 1977 TopicSpecial and Pu blifo cuscatio inn BunNoun.72.. InsNankatituteng, ofTaiwan Hist: orAcademiay and Philol Sinicaogy. JESPERSEN, Otto 1964 EsseSeriesnti alsNo.1. of EnglUnivisersith grammy, Alaarba. ma:Al abamaUniversit Lingyuistic of Alaba andma Philoso Pressph. ical 1975 The(First phil puosophyblished of in gramma 1924r..) London: George Allen and Unwin Ltd. KANO, Tadao 1946 StudiesJapanese.) in thevol ethnology.I. Tokyo and. prehistory of Southeast Asia . (In 1952 StudiesJapanese.) in theVol et.IhnolI. ogyTokyo and. prehi story of So uthea st Asia . (In KANO, Tadao and Kokichi SEGAWA 1956 An illustra ted ethnography of Formosan aborigines . Vol.l: The Yami . Tokyo, Japan: Maruzen. KEENAN, Edward and Bernard COMRIE 1977 Noun phrase accessibility and universal grammar. Linguistic Inquiry 8:63-99. KESS, Joseph F. 1967 SyntacUniversticit yfeatures of Hawaii of. Tagalog verbs. Doctoral dissertation, KULLAVANIJAYA, Pranee 1974 HawaiTransii.tive verbs in Thai. Doctoral dissertation, University of KURODA, Shige-yuki 1965 Causative forms in Japanese, Fo undations of La ngua ge 1/1:30-50. LAKOFF, George 1970 Irregulari ty in syntax . New York: Holt, Rinehart and Winston. 282 LEE, Keedong 1974 UnKuiverssaieaityn verbal of Haw deaiirivat. ion rules. Doctoral dissertation, LEECH, Geoffrey N. 1969 Towards a semantic description of English. London: Longman. LI, Fang-kuei, CHEN Chi-lu and TANG Mei-chun 1958 Notes on . (In Chinese with English summary.) Occasional Papers of the Department of Archaeology and Anthropol ogy , Na tional Ta iwan uni versi ty, No.1: Ethnological researches among the TaiwanThao of Univers Sun-Moonit y.Lake , Formosa , 137-166. Taipei, Taiwan: National LI, Paul Jen-kuei 1972 On comparative Tsou. Bulletin of the Insti tute of History and Philology 44/2:311-337. 1973 RukaPublici stationruc tuNore.6 . 4. InsNankatituteng, ofTaiwan History: Academia and Philo Sinicalogy . Special 1975a RukaiPubli cationtexts . NoInstit.64/2. uteNanka of Histng, Taiwanory and: AcademiaPhilology SiSpniceciala. 1975b NationalSaisiyat stScructureience Counc and iltex, Taipts. ei An. annualMimeo . report presented to the 1975c Thao texts. 83 pp. Taipei. Mimeo. 1975d withResearch Bibliog materaphyrials inclanduding problems works of inFor Chinmosanese la, nguagJapanesees. and(In Western Chinese Nankanlanguagg,es Taiwan.) Bul: letinAcademia of the Sinica Ins ti. tute of Ethnology 40:51-84. 1976 Pazeh texts. 30 pp. Taipei. Mimeo. 1978a Kavalan texts. 60 pp. Taipei. Mimeo. 1978b The case-marking systems of the four less-known Formosan languagEs. In S.A. Wurm and Lois Carrington, eds, Second Interna tiona l Con569-ference615. on Austronesian Linguistics : proceedings . PL , C-61: 1982 Atayalic final voiced stops. In AmranHali m, Lois Carrington anc. S.A. Wurm, eds, Papers from the Th ird Interna tiona l Conference or. Austronesian Linguistics , vol.2:171-185. PL , C-75. LIU, Pin-hsiung, C.C. CHIU, L. SHIH and C.C. CHEN Social structure of the Hsi ukul an Amis. . 1965 of Ethnology Monograph No.8.l Nankang, Taiwan(In Ch: ineseAcademia.) (ISinstitnica.ut e LYONS, John 1968 InUntriversoductioity nPres tos. theore tical lingu istics . London: Cambridge 283 MABUCHI, Toichi 1953 Retrospect on the classification of the Formosan aborigines. (In Japan18/1-2es:1e.)-11 . Minzokugaku Kenkyu (The Japanese Journal of Ethnology.) 1954a Migration and distribution of the Formosan aborigines. Part 1. Et(Inhnology Japanese.) .)18/1-2 Minzokugaku:123-154 . Kenkyo (The Japanese Journal of 1954b Migration and distribution of the Formosan aborigines. Part 2. Ethno(In Japanlogyes.)e.) 18/ Minzokugaku4. Kenkyo (The Japanese Journal of 1960 The aboriginal peoples of Formosa. In George P. Murdock, ed. SoAnthcialropo stlogyructur, Noe .2 in9) Southea, 127-1st40. Asia , (Viking Fund Publications in 1966 Sphere of geographical knowledge and socia-political organization Monographsamong the moun 25:tain101-1 peo46. plesTokyo of . Formosa. Monumenta Nipponica MANLEY, Timothy 1972 NoOu tl.12.ine ofHonolul Sre stu:ruct The ureUniv. ersOceaitynic Press Ling uistof Hawaiics Spei. cial Publication MATTHEWS, P.H. 1974 MorphLondonol: ogyCambridge: an introducUnive rsittion y toPress the . theory of word-structure . McCAWLEY, James 1968 Lexical insertion in a transformational grammar without deep structure. Papers presen ted at the Fourth Regi onal Mee ting, Chica go Linguistic Society, 71-80. McGINN, Richard 1970 proceTagalogsses words. Ter andm Paperverbs for: a probeLInguistics into the651 nature, Univers ofit dey rivationalof Hawaii . McKAUGHAN, Howard P. 1958 PrintingThe infl ect. ion and syntax of Ma ranao verbs . Manila: Bureau of 1962 Overt relation markers in Maranao. Language 38/1:47-51. 1970 Topicalization in Maranao - an addendum. In Wurm and Laycock, eds, (PLSCapel l) . PL , C-13:291-300. 1973 Subject versus topic. In Gonzales, ed. 1973:206-213. MINTZ, Malcolm 1973 Caseof Hawai and i.sem antic affixes of Bikol verbs. Dissertation, University MIRIKITANI, Leatrice T. 1972 KapamHonolulupangan: The sy ntUnivaxersit. Oceay nicPress Ling ofuist Hawaiicsi. Spe cial Publication No.l0. 284 MONTGOMERY, Robert L. 1962 Amis text (Taiwan). Wo rk Papers of the Summer Institute of Linguistics , Un iversity of North Da kota 6:17-20. NAYLOR, Paz Buenaventura, ed. 1979 Austronesian studies : papers from the Second Eas tern Conference on Austronesian Linguistics , Ann Arbor: University of Michigan. NIKIGAWA, A. 1953 A classification of the Formosan languages. Jo urnal of Austronesian Studies 1/1:145-151. NILSON, Don L.F. 1973 The instrumen tal case in Engl ish - syntactic and semantic considerations . The Hague: Mouton. O'BRIEN, Richard 1971 Linguistics : devel opmen ts of the sixties - viewpoints for the Wasevenshingtieston. , D.C.(Monog: raphGeorgeto Serieswn Univon Languageersity Pre andss . Linguistics No.24.: OGAWA, Naoyoshi and Erin ASAI 1935 Myths and traditions of the Formosan na tive tribes . (In Japanese.) TaihokuUnivers ity(Taip. ei), Taiwan: Linguistics Department, Taihoku Imperial PAWLEY, Andrew 1966 Polynesian languages: a subgrouping based on shared innovations in morphology. Jo urnal of the Po lynesian Society 75/1:39-64. 1967 The relationship of Polynesian Outlier languages. Journal of the Pol ynesian Soci ety 76/3:259-296. 1970 Grammatical reconstruction and change in Polynesia and Fiji. In Wurm and Laycock, eds, (PLSCapel l) . PL , C-13:301-367. 1972 On the internal relationships of Eastern . In R.C. Green and M. Kelly, eds, Studies in Oceanic cul ture history , Hawvolai.3,i: (PBeracinicfice AnthropPauahi oloBishopgical Museum Records. (PNoaper.13) pr, ese1-142nted. Honoat lulu, AugWenner-Grenust 1969 .)Sympo sium on Oceanic Culture History, Sigatoka, Fiji, PAWLEY, Andrew and Lawrence A. REID 1976 The evolution of transitive constructions in Austronesian. Un iversi ty of Hawaii Wo rking Papers in Linguistics 8/2:51-74. PAWLEY, Andrew, Lawrence A. REID and Stanley STAROSTA 1978 The evolution of Philippine focus: out of the *fry-an PAN and into Societ*i-fry.y ofPaper America present helded at at thethe UnivAnnualers ityMeet ofing Il linof oithes, Lingu Julyistic, 1978. 285 PLATT, John T. 1971 Gramma tical fo rm and gramma tical meaning: a tagmemic vi ew of FiLingllmoreuistic's deepSeries st ruc5.) turAme stecaserdam conce: North-Hpts . olla(North-Hollnd. and PLEINES, Jochen 1976 Handl ung, Kausal ita t, In tention : Probl eme der beschreibung semanti scher Rel ationen . TUbingen: Gunter Narr. RAMOS, Teresita V. 1974 The case system of Ta galog verbs . PL , B-27. RAMOS, Teresita V. and Resty M. CENA 1979 Existential, locative and possessive in Tagalog. Paper presented atSalz theburg Summer, AustriMe eta, ingAugust of the2-4 Ling. uistic Society of America held at REID, Lawrence A. 1966 AnHon Ivolua tanlu: Thesyntax Unive. rsit(Oceaynic Press Ling ofui sticsHawai i.Spe cial Publication No.2.) 1974a The Igorot subgroup of Philippine languages. Paper presented at Lingthe uistiFirstcs International, Hawaii. Conference on Comparative Austronesian 1974b The Central Cordilleran subgroup of Philippine languages. Oceanic Linguistics 13:1-2:511-560. 1978 Problems in the reconstruction of Proto-Philippine construction markers. In S.A. Wurm and Lois Carrington, eds, Second Interna tional Conference on Austronesian Linguistics : proceedi ngs . PL , C-61:33-66 1979 Evidence for Proto-Philippine nominative marking. Ph il ippine Jo urnal of LInguistics 10:1-20. 1980 PANat thegeniti Lingveui alterstic nationSociety: ofthe HawaiPhilii,ppine Diachron evidenceic Synta. Paperx Fe spretivalsent, ed Apr11/2il:4 5-10.53 . uni versity of Ha waii Wo rking Papers in Linguistics ROBINS, R.H. 1974 The case theory of Maximus Planudes. In Luigi Heilmann, ed. ProceedBolognain: gsIl ofMuli theno . XI Interna tional Con tress of Linguists , 107-111. SCHACHTER, Paul and Fe T. OTANES 1972 TagPressalo. g reference gramma r. Los Angeles: University of California SEBEOK, Thomas A., ed. 1971 CurrentThe Hague trends: Mouton in . linguistics , vol.8. Lingui stics in Oceania . 286 SPRENGER, Arnold 1971 The numerals of the Puyuma language (Katipol Dialect). Fu Jen Studies 4:49-60. Taipei. 1972 Overt construction markers in Puyuma (Formosa) . In Charles T.C. Tang, Jeffrey C.H. Tung and Anthony Y.T. Wu, eds, Papers in Booklingui Co.stics in honor of A.A. Hill , 133-145. Taipei: Rainbow-Bridqe STANLEY, Patricia 1974 Syntax of Tsou verbs: implications for comparative studies. Pape:c preAustrsentedonesi atan theLing Firstuistics Int, Honernatoluluional, JanuaryConference 2-4. on Comparative 1976 Object incorporation in Tsou. In Naylor, ed. 1979:311-314. STAROSTA, Stanley 1967 TsouLinguistic focus . SocietPapery ofread Am eratica, the December42nd Annual 28. MeMimeoeting . of the 1969 Review of Tung T'ung-ho, et al.: a descriptive study of the , Formosa. Language 45/2:439-444. 1971a Review of John Lyons, Introduction to theoretical linguistics. Langua ge 47/2:429-447. 1971b Derivation and case in Sora verbs. Indian Linguistics , 32/3: 194-206. (Also appeared in University of Hawaii Wo rking Papers in Linguistics 3/7:89-110.) 1971c Lexical derivation in a case grammar. Uni versity of Ha waii Working Papers in Linguistics 3/8:83-101. 1971d Ling2/1: uistics13-27. and Anti-Science. Phil ippine Journal of Linguistics 1973 The faces of case. Language Sciences 25: 1-14. (Appeared in French as "Les visages du cas", Langages 38:104-128, 1975; also in Univer1972. ) sity of Ha waii Wo rking Papers in Linguistics 4/1:71-121, 1974 Causative verbs in Formosan languages. Oceanic Lingui stics 13/1-2:279-369. (Also appeared in University of Ha waii Working Papers in Linguistics 5/9:89-155, 1973.) 1975a Case in the lexicon. In Luigi Heilmann, ed. Proceedings of the ElBolognaeventh: IntSocietaerna tieditriconal Congree il Mulinoss of Ling. uists, II, 805-813. 1975b UnGenerativeiversity ofsyntax Hawai: ai. ca Mimse appreo. oach. Department of Linguistics, 1976a A place for case. Language Lea rni ng 26/1: 1-36. (Also appeared in University of Edinburgh Work in Progress 7:18-34, 1974.) 1976b HaLexicasewaii. MimII: eolexi. con. Department of Linguistics, University of .------�------

287 1976c Case forms and case relations in Sora. In Philip N. Jenner, Lawrence C. Thompson and Stanley Starosta eds, Austroasiatic stud1069-ies,1108 .(Ocea Honicnolulu Ling: Theuistic Univers Specialsity Pr Pubessli ofcation Hawai Noi..1 3)(Also, Part II, ap5/1peared:133-1 53in , Univer1973.)si ty of Hawaii Wo rking Papers in Linguistics 1978 The one per sent solution. In Werner Abraham, ed. Va lence, semantic ca se, and gramma tical relations, 459-576. Studies in Languagein German Comaspanion "Die l-Pro-SentSeries 1. LosungAmster",dam: in JochenJohn Ben Pleinesjamins., ed.(A ppeared Beitrage zum Stand der Ka sustheorie. TUbingen: Gunter Narr.) 1979a The end of phrase structure as we know it. Universi ty of Hawaii Wo rking Papers in Linguistics 11/1:59-76. 1979b MannLexicaseheim. I:Mim sceoience. and syntax. Institut fur deutsche Sprache, 1982a Lexical decomposition: features or atomic predicates? Linguistic Anal ysis 9/4:379-393; revised from Is lexical decomposition a lot of12/2 rot?:61-, 75,univer 1980.si ty (Eofxtracted Hawaii Wofromrking Starosta Papers 1976in b.Lingui) stics 1982b Affix hobbling. L.A.U.T. , Series A, Paper No.95. Trier: University of Trier. Reprinted from Un iversi ty of Hawaii Wo rking Papers in Linguistics 9/1:62-158, 1977. 1982c Case relations, perspective, and patient centrality. Universi ty of Hawaii Wo rking Papers in Linguistics 14/1. 1982d Lexicase62-79. : an attempt at a generative reformation. Ki swahili 49/2: to The derivation of combining forms and normal pseudo-compounds in appear a Sora. In Jan-Olof Svantesson, ed. Proceedings of the Symposi um onSt udAusiestro-A. sia tic Languages . Lund: Scandinavian Institute of Asian to Sora noun inflection. To appear in Gerard Diffloth, Bijoy P. appear b Mahapatra and David L. Stampe, eds, Proceedings of the Second CentraInternal tioInstinal tuteCongress of India on Ausn Languagetroasiatics. Linguistics . Mysore: STAROSTA, Stanley, Andrew PAWLEY and Lawrence A. REID 1978 The evolution of Philippine focus: out of the *fry-an PAN and into Societ*i-fry.y ofPaper Ame ricpresa heldented atat thethe UniverAnnualsit Mey etofing Illinois of the , LingJulyuist, 19ic78. 1982 The evolution of focus in Austronesian. In ArnranHali m, Lois Carrington and S.A. Wurm, eds, Papers from the Third Interna tional Conference on Austronesian Linguistics , vol.2: Tra cking the travel lers . PL , C-75. STEVENS, Alan 1969 Case grammar in Philippine linguistics. Paper read at the Annual MeetingDecember of27- the29. Linguistic Society of America held at San Francisco, 288 1973 Case relations of pa-causatives. Abstract. First International ConferenceJanuary 2-4 on. Comparative Austronesian Linguistics, Honolulu, STOCKWELL, Robert, Paul SCHACHTER and Barbara Hall PARTEE 1973 andThe maWinstjoron synt. actic structures of Engl ish. New York: Holt, Rinehart SUNG, Margaret M. Yan 1966 Phonetic andphonemic system of the Kanakanavu language, Formosa. Bulletin of the Institute of History and Philology 36/2:783-800. 1969 WoUnivrd ersitstructy masterure of's the th Kaesisnakanavu. language . Ithaca: Cornell SUNG, Wen-hsun 1967 The megalithic 's east coast. (In Chinese.) Newsl etter of Chinese Ethnology 7:20-21. Taipei. TAYLOR, Harvey 1971 CasePress in. Ja panese . South Orange, New Jersey: Seton Hall University TCHEKHOFF, Claude 1974 OnAus thetronesia inapplican languagbilites.y ofPaper the active- presentpassed ative thedicho Firsttomy Ininternational certain ConferenceJanuary 2-4. on Comparative Austronesian Linguistics, Honolulu, THOMPSON, Lawrence 1965 A Vietnamese grammar. Seattle: University of Washington Press. TORJESEN, Edvard P. 1964a Amisto the articula /d/ phonemetory and. Taichu acousngtic, patterninTaiwan. Mimeogs: wi. th special referencl� 1964b Amis stress patterns. Taichung, Taiwan. Mimeo. 1964c Phonemic problems in Amis. Taichung, Taiwan. Mimeo. 1964d Amis word list. Taichung, Taiwan. Mimeo. 1964e Outline headings of Amis morphology. Taichung, Taiwan. Mimeo. TSUCHIDA, Shigeru 1964 ofPreli theminary Lingui reportsstic Soci onety Sais ofiyat Japan: phonology.) 46:42-52. Gengo. Kenkyu (Journa.L 1968a Bibliography of Formosan languages. 35 pp. Mimeo. 1968b Linguistic questionnaire for Philippine languages. Tokyo. Mimeo. 1971 List of words of Formosan languages. 40 pp. Mimeo. 1976 Reconstruction of Pro to -Tsouic phonology . Study of Languages and TokyoCultur Gaikokugoes of Asia Daiga and Afkurica,. (Yale Monogra Universph Seriesity di sserNo.5.tation Tokyo, 1975, Ja.)pan : 2� TUNG, T'ung-ho 1964 A descriptive study of the Tsou language , Formosa . (Institute of AcademiaHistory and Sinica.Phi lo logy Special Publication No.48.) Nankang, Taiwan: Utrecht Manuscripts manuscriptVocabularuims. Formosanum. Utrecht University Library. Unpublished UTSUSHlKAWA, Nenozo, Nobuto MIYAMOTO and Toichi MABUCHI 1935 The(In FormoJapanesanse .) na tivTaihokue trib es(Tai: pei a gene), alTaiwanogical. and classifica tory study . VOEGELIN, Carl F. and Florence M. VOEGELIN 1964 Languages of the wo rld: Indo-Pacific Fa scicle One . Anthropologica l Linguistics 6/6. 1965 Languages of the Wo rl d: Indo -Pacific Fa scicl e Fo ur . Anthropological Linguistics 7/2. WALTON, Charles 1979 A Philippine language tree. Anthropological Linguistics 21/2:70-98. WANG, Samuel H. 1975 Fatahan phonology: a preliminary approach. Term paper presented to Prof. Shuan-fu Lin, National Taiwan Normal University, 29pp . 1976 TaiwanThe syllaNormal ble strUnivuctureersit y,of FatTaipaanei. Amis. Master's thesis, National WANG, Sung-hsing 1961 Religion and myth among the Vataan Ami. (In Chinese.) Bulletin of theSini Instica. tute of Ethnology 12:107-178. Nankang, Taiwan: Academia WEI, Hwei-lin 1958 Lineage systems among the Formosan tribes. (In Chinese.) Bulletin ofSini theca . Institute of Ethnology 5:1-44. Nankang, Taiwan: Academia 1965 Tribal organization and authority system of the Formosan aboriginal societies. (In Chinese.) Bulletin of the Depa rtment of Archa eology andUniv Anthersitropy. ology 25-26.71-92. Taipei, Taiwan: National Taiwan WEI, Hwei-lin and Jen-ying WANG 1966 A survey of population growth andmigration patterns among Formosan aborigines. (In Chinese with English summary.) Occa sional PapersTaipei , ofTaiwan the Depa: Nationalrtment ofTaiwan Archaeology Univers itandy. Anthropol ogy No.3. WOLFF, John U. and Ida WOLFF 1970 The classification of Cebuano verbs. Phi lippi ne Journal of Linguistics 1/1:74-91. 290 1972 Review of Llamzon: a subgrouping of nine Philippine languages. Jo urnal of the American Oriental Society 92/2:368-370. 1973 Verbal71-91. inflection in Proto-Austronesian. In Gonzales, ed. 1973: WU, Peter A. 1969 AGeorge descritownptive Un ivaneralyssity,is Washiof Bununngton lan. guage. Master's thesis, WURM, S.A. and D.C. LAYCOCK, eds 1970 Paci fic linguistic studies in honour of Arthur Capell . PL , C-13. (PLSCapell) . YAMADA, Yukihiro and Ying-chu LIAO 1974 A23 pho/6:1nolog-9. y of Tayal. Research Reports of Kochi Un iversity . YUAN, Chang-rue 1969 The Ma kutaai Ami of eastern Ta iwan - an ethnographic report . Academia(Institute Si nicaof Ethnogra. phy Monograph No.18.) Nankang, Taiwan: ZORC, R. David 1968 laInnguagformationes. Kalsheetibo and, Ak lalin.nguistic Mimeo check. list for Philippine 1973 The western subgroup of Bisayan. Oceanic Linguistics 11/2:110-139. 1974a Towards a definitive Philippine wordlist - the qualitative use of vocabulary in identifying and classifying languages. Oceanic Linguistics 13/1-2:409-456. 1974b Internal and external relationships of the Mangyan languages. Oceanic Linguistics 13/1-2:561-600. 1977 The Bisa yan dialects of the Phi lippines : subgrouping and reconstruction . PL, C-44. AUTHOR INDEX

ACSON, Veneeta 16, 26, 47-49, 123- DIXON, Robert M.W. 16 AKMA124JIAN, 126, Adrian, 134, and139 Frank HENY, DYEN, Isidore 6, 8-11, 13-14, 16 116 EGEROD, S¢ren 16 ARONOFFANDERSON, , MarkJohn 36M., 47164, 49, 123, 235 FAGAN, Joel L. 16, 43, 46, 104, 118, ASAI, Erin 3, 14-16, 21 FERR121EL, L,145 Rale, 171igh 10-11, 13, 15-16, 93 BENNETTBENEDICT, , DavidPaul K.C. 12810 FEFIY,LLMORE Virginia, Charles 15 J. 16, 29, 41, 43, BENTONBLOOMFIELD, Richar, Leonardd A. 24 16 45-78, 47,84 , 50-52113, , 12155, 60124, , 6516, 2,70-71 236 , 76- BLUBRUGMANNST, Ro, bertKarl A.9-1 6,0 8-10, 16 FOLEY, William A. 16 CAPELL, Arthur 6, 16 GRACE,GRUBER, GeorgeJeffr eyW. S.9 16, 235 CENACHAFE,, ReWastlly aceM. L.16 , 6071 GUNTHER, Hartmut 16 CHCHAAING,, Chen-kangKwang-chih 15 15 HALLEHALLIDAY, Mo, rriMichaels 16 A.K. 16, 93 CHASTUIVERNG, Kwang-chih 15 and Minze HARMON151 , Carol W. 16, 26, 43, 118, 140, CHEN, ChChii-lu-lu and15 Michael D. COE HAUDRICOURHIDALGO, CesarT, Andr A. � andG. Arace6 li C. CHIC15 HIIWA, Suketaro 15 HOPPEHIDALGOR, Pau 16l J. and Sandra A. CHOM93SKY , Noarn16 , 38, 42, 46, 77, HUDDLESTTHOMPSONON , 16Rodney, 69, 1677 , 98, 190 CHUCLARK, Yuan-teh, Marybeth 16 16 , 26, 43, 47, HUDSON, Richard A. 16 49171-50, 173, 12 3, 125, 134, 160, 163, IKRANAGARA164 , Kay 16, 26, 36, 43, 60, CONSTANTINO, Ernesto 16, 45 INa 4 COOK, Walter A. 16, 46, 77 JACKENDOFF, Ray S. 16, 116 DADEGUZMHL, OttoAN, Videa Chr. P.6, 9-1016, 26-, 1627,, 9329 , JEJESPENG, RSENHeng-, hsOttoiung 16 16, 77, 113 , 151 10133, , 35121-1, 43,23, 48 -49128,, 71,139 , 94142, 96, , KANO, Tadao 15 231460,, 253163-1 64, 173, 202-203, KANOKEENA, N,Tadao Edward and andKok Bernardichi SEG AWACOMRIE 15 16

291 292 KULKESSLAVAN, JosephIJAYA, F. Pranee17 16, 26, 43, SCHACHTE112, 136R, Paul and Fe T. OTANES 17, KURODA46, 76-7, Shig7,e -yuki114, 12316 , 171 STANLSPRENGEYER,, PatArnoldricia 16 10 , 16 LAKOFF, George 16 STAROSTA32, 35-, 38,St an42-46ley 10, 48-49, 16, , 2653,, 28 -29, LEELEECH, Keedong, Geoffrey 16 , N.26 16, 43 9355-56, 104, 58-60, 110, 11370-7, 1,11 6,73, 11875-79, 121-122, 84, , LI, Fang-kuei,i 12 Ch3,i-lu 171 CHEN and STAROSTA133, 146, St, an160ley, 195, Andrew, 235 -2PAWLEY36, 253 and LIMe, Pauli-chun Jen-kue TANG i16 16, 26, 29, 34, 20Lawrence4, 245-246 A. REID, 24875 , 83, 85, 91, 94, LIU43,, Pi 56n-hs iung 15 STOCKWSTEVENSELL, Alan, Robe 17rt, , 75Paul, 253 SCHACHTER and LYONS, John 16, 60 SUNGBarbara, Margaret Hall M.PARTEE Yan 1615 , 16 MANMABUCHLEYI,, Timothy Toichi 453, 15 SUNG, Wen-hsun 15 McCAWLMATTHEWSEY, JameP.H. s 1616 TAYL173OR-174, Har vey 16, 26, 43, 118, 171, McKAUGHAN,McGINN, Rich Howardard 17 P. 17, 24, 45-47, TCHEKHOFFTHOMPSON, , LawrenceClaude 17 160 MINTZ52 , Malcolm 17, 24 TORJTSUCHESENIDA,, EdvardShigeru P.2, 15 6, 10, 16 MIRIKITMONTGOMEANIRY, RoLebertatric L.e T.15 17 TUNG, T'ung-ho et al 16 NILSON, Don L.F. 113 UTSandUSHlK ToichiAWA, NenoMABUCHzoI, Nobuto3 MIYAMOTO O'OGAWABRIEN, Naoyoshi, Richard and16 Erin ASAI 9, 14 WAWALTONNG, Sam, Chuearlesl H. 1015, 1716 PAWLEY, Andrew 9 WEI, Hwei-li-linn and3, 15Jen-ying WANG 1-3, PAWLE10, Y,74 Andrew, 91 and Lawrence A. REID WOL15FF , John U. and Ida WOLFF 10 PAWLEY, Andrew, Lawrence A. REID WU, Peter A. 16 PLAandTT, StaJohnnl eyT. STAROSTA16, 70 17 YAMADA, Yukihiro PLEINES78, 84, Jochen 16, 43, 55, 70-71, YUANYAMADA, Chang-rue, Yukihir o 3 and Ying-chu LIAO 16 RAMOSRAMOS,, TereTeresitsita V. and17, 163Resty M. CENA ZORC, R. David 10, 17 REID,17 Lawrence A. 10, 13, 16-17, 45, ROBINS136 , R.H. 124 LANGUAGE INDEX

Asiatic 10 Melayu Betawi 26, 44 Austrone75-76, si81,an 146languages, 245 6, 8-9, 17, Mono-Alu, Solomon Is. 44, 171 Bikol 24 Oceanic language 10 English 26, 44, 51, 79, 98-99, 116, PaiPangaswanicinan la nguage24 II, 13 136, 163 Ph23-24ilippine, 46 languages, 52, 71, 1074, 8113, 16-1784-85, , 9120, Formosan (aboriginal) languages I, 94, 144 6,134 8-17, 26, 44, 46, 52, 71, 76, Proto-Austro268 nesian 6, 8, 91, 94, 245, GeGreekrman 45-4644, 79, 48 , 123 Rukai 10, 26, 29, 44, 171 Ilokano 81 SrSorae 45 26 , 44, 79, 160 Indonesian 6, 14, 17 Tagalog 26, 44, 48, 81, 94, 96, 121- Japanese 26, 44, 171 171122, 245128, 144, 146, 160, 163-164, KagayanenKusaiean 2626, , 4444 , 151 TeThailugu 26 50, 44 , 76, 114, 171 Latin 46, 50 Vietnamese 26, 44, 47, 160, 163, 171 Malayo-PoMaranao 24lyne, 144sia n 6, 8, 14 Yarnila nguage 10

293 TOPIC INDEX

Acabcusasolutitivvee Agent45-47 139-141, 147 case forms 31-33, 41-42, 45-53, 56-57, the Accusative case form (Acc) 45-50, case107 -1fra08me, 123features-124, 30-31139-141, 47-50 , 53- 60138-64, 143-1, 67-7047,, 80-14983,, 15 6-159107-109, 164-, 125- 57163-, 97-98189, 191-196, 128-134, 198, 143-144-208, 21, 0,153-15 2216,- 258171, 196, 198-200, 220-221, 253- 224, 230, 235-240, 245-251, 259-262, Accusative Instrument 85-87, 146- case264 mark ing 45-50, 56-57, 123-134, Accus147,ative 196 , Locus198- 20067-7, 0,224-225 147 case139-1 re41,lations 146 -130-4731, 156, 41-, 57,208 , 67-25970, Ac97,cusativ 101e-1 Patient02, 134-1 67-738, 0,143 80, -83,146 , 96- 102171-178, 107, , 191-196110-113, , 205116-12,210,3, 23 139-15-234c,3, 245-189-124991,, 253203--258205 , 219-220, 224, case250- re251,lati on264 assignment 53-55, 67-70, Accusativ178-186e, si189mple, 21 9-tra221nsit ive verbs 116-118, 235-238 Accusative Time 108-109, 146-147 Comitativecausativisa casetion form171-178 (Com), 253 45--26247 , 118- Accusativverbs 197-1e trans98, it224ive instrumental 120, 128-134, 139-141, 146-147, 149- Accusative transitive locative verbs Comi150ta, tiv156-e/c159ompan ion case notion 42-45, Accusativ200-203e, tra226-227nsiti ve verbs 67-70, com125-1panion27, verbs 134-1 42-4538, 147 72-7156-1593, 76-, 189-77, 191,80- 83,196 143-1, 200-20344, 147, , com196plem, entatio200 n 109, 164-178, 186-188, Accusa253-258tive-t ype languages 76-77, concimitant/comitativerelation 42-45, 113-114/com, panion118-120 ca, se Agent/Ag80-83 entive case relation (AGT) consonant128-134 inventory 17, 22-23 14431-33 , 42-45, 60-64, 113-114, 143- contextual55, 110- 113features, 160- 16226-35, 164-191, 38-42, , 19653-, Amisambient sett verbslements 178 -11-686 contr200,astiv 205-21e case0, 230-23form 45-8, 47,253-2 150-15258 anaspeti-ctpassual iveadverb 64-70 164 -171 co-re118-120ferent iality 106-107, 110-113, aspiration 20-21 Correspondent94, 113-114 , case121-1 rel22ation, 145 42-45, 93- Bene104-1fact05ive, 125-1case 41,form 147-1 (BEN48) , 42-15047,

294 295 Dative case notion 42-50, 93-94, 113- Genitive case form (Gen) 45-50, 64, Dative114, movement121-122 116-118 72-73, 75-76, 87-89, 96-97, 121-122, definite 30-31 198-125-120527, 208-2134-14110, 245-249144-145 ; 160-162, Deriv76-77ation, 163-1 Rule64 26-, 2331,5- 24136-38 , 53-55, Genitiv191-196e ins, 198trument-200, 83-8221-2255, 14, 5-1259-26246, DerDerivedived nounverb 11118-128-1200,, 125156-159 , 235- Genitiv156-159e Locus 87-89, 139-141, 145, De238rived, 245-249 verbs with, 259 -262implied, 264-266 goalGeni tive47-50 patient, 123-124 208-210, 128- 141, 230-234 descriinstrptiveument verbs 128-134 178-, 178-1186 86 habitual place 105-106 Determiners123-127, 134-130-3341,, 45- 14447,, 146-1 110-11347, , impersonal ins transitive verb 205-208 detra153-1nsit56ivisa, 160-1tio62n , 269-272245-249 impe228-230rsonal transitive verb 208-210, direct205 object 45-47, 134-138, 200- impeimpliedrsonal instr verbsument 205-210 89-93 dir12ect3-124ion/di, 134-138rectional 30-31, 45-47, inchoativeincrement case30-31 re lation 42-45, 85-87, disylla17-18bic and polysyllabic words indirec116-118t ob ject 45-47, 93-94, 121-122, endocentric construction 38-41 Inflec134-1tio38nal, 200-203 feature s 31-35, 164-171 epistergativeemolog simpicalle verbstransitiv 121e-122 verbs Inf26-29lectional, 31-38 Redundancy, 160-162 Rules (IRR) erga17tiv8-186e tra, 189nsitiv, 19e1-196 instr, um219-ental223 In31flect-32 ional subcategorised rules (ISR) erverbsgative 198-200transitiv, 224-225e locativ e verbs inner205- case208, re264latio n 93-94, 116-118, ergativ200-2e05 tra, 227ns-228itive verbs 64, 76-77, Instr42-um45ent/I, 70-nstr71, ume78-85ntal, 87-8case9, re92-93lation, 97191--98196, 114-1, 198-15200, 143-1, 221-45,223 171, 253--178, Instr98-99ument, 113-1 verbs14 , 161444-171 ergativ258, e266-268 transit ive instrumental InstrInstruumementalntal sucasebj ectsform 80-8345-47, 89-93, erga153-tive-t156 ype languages 76-77 intr164-17ansitiv1, e196 loca, 24tive4-249 verbs 36-38, 186- erg178ativit, 189y , 76-200-277, 0380-, 83,208- 2197-0,98, 225 171-, intr188ans, 21it7-219ive verbs, 23 0-23417-1 8, 60-64, 67-70, er253-gativisa258, tio264n , 242-243272-273 76-14477, 17, 80-1-18883, 92-203-2093,8, 101-1 210-20219, 142-, 228- exocen134 tric construction 38-41, 128- 230, 239-243, 251-262, 264-266 Exp76er, ience93-94 case, 121-1 re22latio, 145n 42- 45, 75- lexical features 30-41 Factive 42-45, 113-114 lelexicxicalal heaRedundancyd 38-42 , Rule128-134 (RR ) 36-41 force 83-85, 113-114 lexilexicalcase rulmodeesl 26-3114-17, , 36-3826-30 , 33-45, Geni96-tiv97e, Agent134-1 38,71 -7143-16, 83-45,85 171-178, 87-89, lexi53-55con , 6-10171-, 17826-41, 23, 5-23853-55 , 164-171, 211910,-1 96221, -225198-, 200253-262, 203-205 , 208- lexi235-cost238atist ics 1-10 296 liloca1quidsis ticand analnasaysisls 23 47-50 , 99-101, nomi156nativ, 22e0-22 agent1, 22471-7, 245-2493, 142-1, 4325, 3-258153- loc12a1ist3-124ic features 30-31, 47-53, nomi50,native 60-64 ca, se80 , form94, 123-1(NoIll)24 30, -31128-134, 45-, 97-98144, 153-1, 103-10656, 160-1, 121-16225, 21, 0,128-1 21427-, nomi142-143native, 156-instru159me, nt171 80-83-178 , 92-93, loca219,tion 221, di-22rection3, 225 , or23 orienta0-234 tion 221142-1, 43245-, 145-1249 46, 153-156, 196, 220- location/locat42-45, 47-50ive, 67 verbs-70, 123-12447-50, , 93-150 nomi249native locus 139-143, 203-205, 245- locatio95, 225n nouns, 230-2 94-3897 , 147-148, 160- nomi144native, 146 , patient191-196 87-8, 203-2059, 92-, 93,219-2 142-25, Locative162 case form (LCV) 45-50, 60- 253-258, 264-266 64138, , 93143-94,, 147-1102-10349, , 160-162113-114 , 128- oblOmegaique rule cases 33-35 134-1, 38-4138, 142, 205-208 Locative101-102 locus, 146 , 67-71480,-149 87-, 89164-17, 94-951, , one-pe103, r-sent106-107 (l, /sent)110-113 co, nst116-118raint , 42-121-45, Locative189-196 pl, ace200-20 1035,-1 05245-249, 149 orthogra122, 146phy, 15173-, 15619-20, 171-178, 235-238 Locative110-113 pr, edic149, ates153-156 98-99 , 103-104, pa1ata1isation 21 Locative109, 123-1 rela25tor, 128-1nouns34 94-95, 147-1, 99-10149, , Patient64, 70-7 case3, 83rel-85ation, 87 -8930-31, 92, 42--9445,, 113- 59- Locative160-162 subcase forms 123-124 Patient114, 116-118primary/, patient121-122 centrality 42- LocusLocativ casee tim ree la107tio-109n (L, OC146) 42-, 14947 , 67- phenom45, ena59-60 verbs 205-208 70124, , 93146-94 , 96-97, 102-104, 123- pho29netic interpretation rules (PIR) 26- Locus-focus 96-97 phrasePlace casestr ucturerelation rules 42-45 26-, 3567-7, 38-0, 41102- Mann80er, 113-1 case 18notion, 128-134 42-47 , 58-59, 78- po108sses, sor110-113 73-75, , 11126-1181-122 , 128-141, 144- MannerMeans caseverbs notion 128-1 42-34, 45164-, 67-17171 , 85- prepr145ositions, 153-159 and, 245-2 combi49nat orial case meteorolo87 gical verbs 60-64, 178-186, prepmarkersositio ns123 -124and po, st128-134posit ions 45-47, morp21hological0-216 causative verbs 75-76 23123-19-24141 , 147-150, 156-159, 164-171, morphophonmorphologiemiccal rulerule 33-35 33-35, 163-164 pronouns147-150 123-127, 156-162, 134-142, 144-145, motion216, verbs221-22 973,-98 230-2, 128-134 34, 210- psycho178, logi191-cal196 verbs, 198-200 121-1 22, 148, 171- Neutrnegatioal n caseverbs forms 164-17 42-1 45, 87-89, 150, RangeReason casecase formnotion45-47 42-47 nominal153-159 com, plements230-234 118-120, 156-159 Redundancy142, 160- Ru17les8, 1826-316-188, , 33-35210-21, 6,38- 41,230, nominal109, 113-predi114cates, 116- 96-99118, , 128-1102-10541, , Re23fer5-238ence/Referent case notion 42-45, 144, 149-156 Result104-105 case, 107 notion 42-45 297 sema102ntic, 12 feature3-124, s163-1 30-3164, 171-133-3586, 101-, tertemporalminus pr47-50edicates, 123- 101249-113 , 149 sema191ntic-196 inter, 205-2pre08tation, 210-225 rules, 23 (SI0-2R)38 Timetheme case42-45 re lation 42-45, 60-64, 107- simp26-29le intr, 186-188ansitiv, e191-1 verbs96 , 212210-2-2216 3 topic113 , 30-31116-118, 118-1 20, 150-159 simp219-220le non-ag entive verbs 178-186, trtransansitformative ionsinstrum 235-ental238 verbs 178-186, simple189-196 tra, nsi21t9-220ive verbs 178-186, tra191-196nsitive, 224locative verbS 178-188, 225- sistsistererhead consti hypottuenthes is38- 4138 -41 tra228nsitiv, 23e0-23 verbs4 38-41, 64, 71-72, 75- spat160-1ial62 orienta, 225 tion 99-101, 123, 128-13477, 80-83, 14, 101-13, 14602, , 171-1114-18615, , 189-198121-122, , specificecific -o101-102bject intransitives 94-95, trans205-iti21vity0, 219- 67-70223, 2376-0-27734, 17, 2531-186-262, Sta101-ndard102 theory/Chomsky's Model 26- trans205-210itivi, satio251-262n 251 -252, 269-272 subcatego29, 42-45risatio n rules 26-38, 160- verbal complements 98-99, 102, 110-120, subj178ect, 210-225 36-38, 45-47 238128-134-239,, 242-156-124459 , 164-178, 205-208, subj93ect choice hierarchy 87-89, 92- verbs239, derived 244-249 from derived nouns 238- sursyllafactble ststructureructure 42-17-1845 voicevowel inveninflectorytion 1730-31 , 45-47 word formation analogies (WFA) 36-38

PACIFIC LINGUISTICS - LIST OF PUBLICATIONS

After each entry, ISBN numbers have been added . Where there is more than one number, the first refers to the complete set of volumes and the second and/or subsequent numbers to indiv idual volume s or parts.

Note : the earliest works were called LINGUISTIC CIRCLE OF CANBERRA PUBLICAT IONS , however all now bear the name PACIFIC LINGUISTICS . All volumes are softbound unless otherwise indicated .

Unmarked publicat ions wh ich come under 'Languages for intercultural communi­ cation in the Pacific area ' project are: A-54, A-57, B-26, B-61, B-73, C-34, C-40, C-52, 0-3, 0-5, 0-12, 0-23, 0-29 (note the change of proj ect name) .

SERIES A - OCCASIONAL PAPERS

No . 1. WURM , S.A. Some remarks on the role of language in the assimilation of Australian abor igines . 1963; 12pp . Repr inted 1966. ISBN a 85883 006 X No . 2. HEALEY , Alan Handling unsophisticated linguistic informants. 1964 ; iii+30pp. Repr inted 1967, 1972, 1973, 1975. ISBN a 85883 007 8 No . 3. Papers in New Gu inea linguistics No .1. 1964; iv+42pp . Reprinted 1971. ISBN a 85883 008 6 Papers by Alan Pence ; Ellis Deibler Jr; Phyllis M. Healey ; Bruce A. Hooley . No . 4. Papers in New Gu inea linguistics No .2. 1964; iv+41pp . (incl. 1 map) . Reprinted 1971. ISBN a 85883 009 4 Two papers by S.A. Wu rm . No . 5. Papers in New Gu inea linguistics No .3. 1965; iv+53pp . Repr inted 1972. ISBN a 85883 010 8 Two papers by Phyllis M. Healey . No . 6. papers in New Gu inea linguistics No .4. 1965; iv+68pp . Reprinted 1971. ISBN a 85883 all 6 Two papers by Darlene Bee . No . 7. Papers in New Gu inea linguistics No . 5. 1966; viii+93pp . Repr inted 1971. ISBN a 85883 012 4 Papers by Chester I. and Marjorie E. Frantz ; Des and Jennifer Oatridge; Richard E. Lov ing ; Joyce Swick ; Alan Pence ; Philip Staalsen ; Helen and Maurice Boxwell. No . 8. papers in Philippine linguistics No .1. 1966; iv+38pp . Reprinted 1971. ISBN a 85883 013 2 papers by Jo Shetler; Richard Pittman ; Vivian Forsberg; Jean Hussey . No . 9. Papers in South East Asian linguistics No . 1. 1967; iii+43pp . + 30 charts, 24 tables . Reprinted 1970. ISBN a 85883 014 a Papers by Nguy�n ��ng Liem (2) ; A. Tran Huong Ma i; Dav id W. Dellinger . No .lO. Papers in Au stralian linguistics No .1. 1967; v+59pp . Repr inted 1972. ISBN a 85883 015 9 Papers by Dav id and Kathleen Glasgow ; Jean F. Kirton; W.J. Oates; B.A. and E.G. Somme r. No .ll. Papers in Australian linguistics No .2. 1967; iii+73pp . + 7 maps . Reprinted 1971. ISBN a 85883 016 7 Papers by C.G. von Brandenstein; A. Capell (2) ; Kenneth Hale . No .12. Papers in New Guinea linguistics No. 6. 1967; iii+48pp . Reprinted 1971. ISBN a 85883 017 5 Papers by K.A. McElhanon ; G.L. Renck . No .13. papers in New Guinea linguistics No .7. 1967; iv+59pp . Repr inted 1971. ISBN a 85883 018 3 Papers by Jean Goddard; Karl J. Franklin.

'" PACIFIC LINGUISTICS 2

Series A - Oc casional Papers (continued ) No .14. Papers in Australian linguistics No .3. 1968; iii+46pp. + 1 map . Repr inted 1971. ISBN 0 85883 019 1 papers by E.F. Aguas ; D.T. Tryon . No .15. Papers in linguistics of Melanesia No .1. 1968; iii+52pp. + 1 map . Repr inted 1971, 1980 . ISBN 0 85883 020 5 Papers by A. Capell; G.J. Parker ; A.J. Schutz (2) .

No .16. Papers in New Guinea linguistics No .8. 1968; iv+62pp . (incl. 2 maps) . Reprinted 1971. ISBN 0 85883 021 3 Papers by C.L. Voorhoeve; Karl J. Franklin; Graham Scott . No .17. Papers in Australian linguistics No .4. 1969; viii+97pp . (incl . 3 maps) . Repr inted 1971. ISBN 0 85883 022 1 Papers by Joy Kinslow Harris; S.A. Wurm; Don Laycock . No .18. Papers in New Gu inea linguistics No .9. 1969; vi+llOpp. (incl. 1 map) . Reprinted 1971. ISBN 0 85883 023 X Papers by A. Capell; Alan Healey ; Darryl Wilson (3) . No .19. Papers in Philippine linguistics No .2. 1969; iii+32pp . Re printed 1971. ISBN 0 85883 024 8 Papers by Jeanne Miller; Helen W. Miller . No .20. Papers in Borneo linguistics No .1. 1969; iv+41pp . Repr inted 1971. ISBN 0 85883 025 6 Papers by D.J. Prentice (3) . No .21. papers in linguistics of Melanesia No .2. 1969; v+105pp . (incl. 5 maps) . Repr inted 1979. ISBN 0 85883 002 7 papers by A. Capell; Ann Chowning ; S.A. Wurm. No .22. Papers in New Gu inea linguistics No .lO. 1969; v+84pp . ISBN 0 85883 026 4 Papers by Don Laycock ; Richard G. Lloyd ; Philip Staalsen . No .23. Papers in New Gu inea lingu istics No .ll. 1970; v+78pp . (incl . 1 map) . ISBN 0 85883 001 9 Papers by Gordon and Ru th Bunn ; Alan Pence , Elaine Geary and Dor is Bjorkman ; Harry and Natalia We imer ; O.R. Claassen and K.A. McElhanon . No .24. Papers in Philippine linguistics No .3. 1970; vi+77pp . ISBN 0 85883 000 0 Papers by Norman Abrams ; Jannette Forster ; Robert Brichoux . No .25. Papers in New Gu inea linguistics No .12. 1970; iv+60pp . + 1 map . ISBN 0 85883 027 2 Papers by C.L. Voorhoeve ; K.A. McElhanon ; Bruce L. and Ruth Blowers. No .26. Papers in New Guinea lingu istics No .13. 1970; iv+48pp . ISBN 0 85883 028 0 Papers by Bruce L. Blowers; Margie Griffin; K.A. McElhanon . No .27. Papers in Australian linguistics No .5. 1971; iv+7 0pp . ISBN 0 85883 029 9 Two papers by Jean F. Kirton . No .28. Papers in New Gu inea linguistics No .14. 1971; vi+172pp . (incl . 8 maps ). ISBN 0 85883 030 2 Papers by T.E. Dutton ; C.L. Voorhoeve ; S.A. Wurm . No .29. Papers in South East Asian lingu istics No .2. 1971; iv+7epp . (incl. 1 map) . ISBN 0 85883 031 0 Papers by Warren W. Glover ; Maria Hari; E.R. Hope . No .30. Papers in South East Asian linguistics No.3. 1973; iv+82pp . ISBN 0 85883 091 4 Papers by D.W. Dellinger ; E.R. Hope ; Makio Katsura; Tatsuo Nishida . No .31. Papers in New Guinea linguistics No .15. 1972; v+69pp . ISBN 0 85883 032 9 Papers by R.K. Lewis; Sandra C. Lewis; Shirley Litteral; Philip Staalsen . No .32. Papers in Philippine linguistics No .4. 1971; iv+32pp . ISBN 0 85883 033 7 Papers by R.M. Hohulin; Lou Hohulin . PACIFIC LINGUISTICS 3

Series A - Occasional Papers (continued) No .33. Papers in Borneo and western Austronesian linguistics No .2. 1977; vi+132pp . + 1 map . ISBN 0 85883 164 3 Papers by C. Court; Robert A. Blust; F.S. Watuseke. No .34. Papers in New Gu inea linguistics No .16. 1972; iii+46pp . ISBN 0 85883 081 7 Papers by Janice Allen ; Marshall Lawrence . No .35. Papers in linguistics of Melanesia No .3. 1972; vii+113pp . (incl . 6 maps) + 6 maps . ISBN 0 85883 083 3 Papers by C.H. Beaumont; D.T. Tryon ; S.A. Wurm. No .36. Papers in Australian linguistics No .6. 1973; iv+72pp . + 4pp . photographs, 2 maps . ISBN 0 85883 095 7 Papers by B. Schebeck ; Luise A. Hercus and Isobel M. White . No .37. Papers in Australian linguistics No .7. 1974; iv+73pp . (incl . 1 map , 3 photographs) . ISBN 0 85883 116 3 Pape rs by Christine E. Furby; Lu ise A. Hercus ; Chr istine Kilham. No .38. Pape rs in New Gu inea linguistics No .17. 1973; iii+78pp . ISBN 0 85883 097 3 Papers by K.G. Holzknecht (3) ; Donald J. Phillips . No .39. Papers in Australian linguistics No .8. 1975; v+78pp . ISBN 0 85883 126 0 Papers by M.C. Sharpe; Lothar Jagst; David B.W. Birk. No .40. Papers in New Gu inea linguistics No .18. 1975; iv+l02pp . (incl . 5 maps ). ISBN 0 85883 118 X Papers by Robert Conrad and Wayne Dye ; N.P. Thomson ; Leslie P. Bruce Jr. No .41. Papers in Philipp ine linguistics No .5. 1974; iv+74pp . ISBN 0 85883 114 7 Papers by Donna Hettick Chandler ; Edward Ruch; Jeannette Witucki. No .42. Papers in Australian linguistics No .9. 1976; iv+79pp . ISBN 0 85883 140 6 Papers by Joyce Hudson ; Barbara J. Sayers. No .43. Papers in Philipp ine lingu istics No .6. 1974; iii+74pp . (incl. 1 map ) . ISBN 0 85883 108 2 Papers by Thomas N. Headland and Alan Healey ; Jeannette wituck i. No .44. Papers in Philippine linguistics No .7. 1975; iv+60pp . ISBN 0 85883 135 X Pape rs by Betty Hooker ; Dietlinde Behrens; Patricia M. Hartung . No .45. Papers in New Guinea lingu istics No .19. 1976; v+l05pp . (incl. 2 maps) . ISBN 0 85883 156 2 Pape rs by Ger P. Reesink ; Lillian Fleischmann and Sinikka Turpeinen; Peter C. Lincoln . No .46. Papers in Philippine linguistics No .8. 1976; iv+89pp . ISBN 0 85883 146 5 Pape rs by Jeannette witucki; Michael R. Walrod ; Jean Shand . No .47. Pape rs in Au stralian linguistics No .lO. 1976; iv+78pp . (incl. 3 maps, 11 photographs). ISBN 0 85883 153 8 papers by Jean F. Kirton ; Bruce A. Somme r; S.A. Wurm and L. Hercus; P. Austin, R. Ellis and L. Hercus. No .48. THOMAS , David, Ernest W. LEE and NGUY�N �XNG LItM, eds Papers in South East Asian linguistics No .4: Chamic studies. 1977; ix+124pp . ISBN 0 85883 163 5 Papers by Alice Tegenfeld t Mundhenk and Hella Goschnick ; Timothy Fr iberg and Kvoeu Hor ; Dor is Walker Blood ; David L. Blood ; Eugene Fuller ; Ernest W. Lee; Hella Goschnick . No .49. Papers in South-East Asian linguistics No .5. 1977; iv+98pp . ISBN 0 85883 158 9 Three papers by David Bradley . PACIFIC LINGUISTICS 4

Series A - Occasional Papers (continued) No.50 Papers in Philippine ling uistics No. 9. 1979; v+l08pp . (incl . 1 map) . ISBN 0 85883 186 4 Papers by E. Clay Johnston ; Hartmut Wiens; Jo Ann Gault with Sulaiman and Fatima Barhama ; Peter Green ; Bruce Grayden ; Jeannette Witucki. No .51. Papers in Australian linguistics No . 11. 1978; vii+199pp. (inc1. 1 map) + 3 maps , 4 photographs . ISBN 0 85883 179 1 Papers by Jean F. Kirton ; R. Wood ; L.A. Hercus; Chester S. Street and Harry Palada Kulampurut; Dianne Buchanan; Jean F. Kirton and Bella Charlie.

u A No. 52. GETHING , T.W. and NGUYEN'/( BANG LIEM, eds Papers in South-East Asian linguistics No. 6: Tai stud ies in honour of William J. Gedney . 1979; vi+149pp . ISBN 0 85883 188 0 Papers by Pongsri Lekawatana; Leslie M. Beebe ; Thomas W. Gething ; Carol J. Compton ; Ma ry Sarawit; Thomas Scovel ; John F. Hartmann ; Wilaiwan Khan i ttanan ; James R. Chamberlain; Beatr ice T. Oshika . No. 53. Papers in South-East Asian linguistics No. 7. 1980; v+130pp . (inc l. 4 maps) ISBN 0 85883 206 2 Papers by Ronald L. Trail with Harisingh T. Rathod , Geeta Chand , Chaudhary Roy , Indira Shrestna, Nirmal Man Tuladhar; Peter J. Grainger; Wa rren W. Glover and John K. Landon ; Austin Hale and Thakurlal Manandhar ; Austin Hale ; Burkhard SChottelndreyer (3) . No .54. Papers in pidgin and creole linguistics No .1. 1978; vi+197pp. Reprinted 1980. ISBN 0 85883 178 3 Papers by Loreto Todd and Pe ter Muhlhausler; S.A. Wu rm (2) ; John T. Platt; Peter Muhlhausler (2) ; D.S. Walsh. No.55. Papers in Philippine linguistics No .lO. 1979; vi+142pp . ISBN 0 85883 193 7 Papers by Andrew F. Gallman; E. Joe Allison ; Carol M. Harmon; Jeannette Witucki. No.56. Papers in New Guinea linguistics No .20. 1980; v+2l4pp . (incl. 4 maps) . ISBN 0 85883 215 1 Papers by Maurice Boxwell; Jean Goddard; Malcolm Ross; Arden G. Sanders and Joy Sanders (2) ; Joy Sanders and Arden G. Sanders; H.J. Davies. No.57. Papers in pidg in and creole linguistics No .2. 1979; x+290pp . Reprinted 1985. ISBN 0 85883 198 8 Papers by Peter Muhlhausler; Elsa Lattey ; Ellen Woolford; William G. C�mden; Margaret S. Steffensen; M.G. Clyne; William Peet Jr ; Ulrike Mosel; Ian Sm ith; Jerry G. Gebhard; Nguy�n Bang Liem; Gail Raimi Dreyfuss and Djoehana Oka; Maria Isabelita o. Riego de Dios. No .58. Pape rs in Australian linguistics No .12. 1980; vi+l13pp . ISBN 0 85883 208 9 Papers by Chester S. Street (2) ; Helen Geytenbeek; Kathleen Glasgow and Mark Garner . No .59. RIGSBY , B. and P. SUTTON , eds Papers in Australian linguistics No. 13: Contributions to Australian linguistics . 1980 ; viii+314pp. (incl. 5 maps) . ISBN 0 85883 205 4 Papers by Alan Rumsey ; Patrick McConvell; Peter Sutton (2) ; Tamsin Donaldson ; L. Hercus; Peter Austin, Corinne Williams and Stephen Wurm; Paul Black ; Terry Crowley ; Harold J. Koch ; D.T. Tryon ; A.K. Chase and J.R. von Sturmer. No.60. Papers in Australian linguistics No .14. 1980; v+178pp . (incl . 33 maps) . ISBN 0 85883 230 5 Papers by Bruce E. Wa ters (2) ; Peter A. Busby . No .6l. Papers in New Guinea linguist ics No .21. 1981; v+209pp . ( incl . 10 maps) . ISBN 0 85883 236 4 Papers by H.J. Davies; Dieter Osmers; John Lynch; S.A. Wurm. PACIFIC LINGUISTICS 5

Series A - Occasional Papers (continued )

No .62. BRADLEY , David , ed . Papers in South-East Asian linguistics No .8: Tonation . 1982; viii+159pp . (incl . 1 map) . ISBN 0 85883 256 9 Papers by A.V. Diller ; David Bradley; Philip John Rose ; VU Thanh Phddng ; U Thein Tun ; Jack and Mary Jane Gandour.

No .63. Papers in New Guinea linguistics No .22. 1985; iii+373pp . (incl 5 maps) . ISBN 0 85883 322 0 Papers by Karen Adams and Linda Lauck; J. Miedema and F.I. We lling ; W.A.L. Stokhof and Don A.L. Flassy; Hiroko Og uru (2) ; Kenneth Collier and Kenneth Greger son ; Thomas R. Phinnemore ; David Scor za; John Davies and Bernard Comr ie ; Stan Abbott (2) .

No .64. LYNCH , John , ed . Papers in linguistics of Melanesia No .4. 1982; vi+l67pp. (incl. 2 maps) . ISBN 0 85883 276 3 Papers by John Lynch (2) ; Daniel D. Dodenhoff.

No .65. Papers in pidg in and creole linguistics No .3. 1983; v+206pp. (incl. 2 maps , 2 photographs) . ISBN 0 85883 305 0 Papers by Lo is Carr ington ; Jeff Siegel; Pe ter Muhlhausler (3) ; Linda Simons; Alan Baxt er ; Joyce Hudson ; Alan Rumsey; Ann Chown ing .

No .66. AUSTIN, Peter , ed . Papers in Austral ian linguistics No .15: Australian Abor iginal lexicography . 1983; xii+173pp . (incl. 3 maps) . ISBN 0 85883 291 7 Papers by Peter Austin; Patrick McConvell, Ron Day and Paul Bla ck; R. David Zorc ; Be rnhard Schebeck; G.R. Mc Kay ; Kenneth Hale ; Mary Laughren and David Nash; Anna Wierzbicka ; Ma ry Laughren; H. Koch .

No.67. BRADLEY, David , ed . Papers in South-East Asian linguistics No .9: Language policy , language planning and soc iolinguistics in South-. 1985 ; vi+214pp . (incl. 1 map , S photographs). ISBN 0 85883 325 5 Papers by Anthea Fraser Gupta ; John T. Platt; Ng Bee Chin; Asmah Haji Om ar ; A.V.N. Diller ; Prapart Brudhiprabha ; David Bradley; Thel Thong ; Nguyen Dinh-Hoa ; Anna J. Allott; R.K. Bar z and A.V.N. Diller ; Yogendra Yadav .

No .68. Papers in Austral ian linguistics No .16. 1984 ; v+327pp. (incl. 2 maps) . ISBN 0 85883 311 5 Papers by Kathleen Glasgow ; A. Capell ; G.R. McKay (2) ; Rod Kennedy; D. Trefry. No .69. Papers in New Guinea linguistics No .23. 1984; vi+283pp . (incl. 4 maps ). ISBN 0 85883 313 1 Papers by Malcolm Ross; Jeff Siegel ; Robe rt Blust; Michael A. Colburn; W. Se iler .

No .70. Papers in pidgin and creole linguistics No .24. 1986 ; iv+312pp . (incl. 4 maps) . ISBN 0 85883 350 6 Papers by D.C. Laycoc k; Walter Seiler ; Le s Bruce ; M.A. Chlenov ; R. Daniel Shaw; Susann e Ho lzknecht ; Graham Scott; Otto Ne kitel; S.A. Wurm; L.R. Goldman ; J.S. Fingleton .

No .72. Papers in pidg in and creole linguistics No .4. 1985; iv+238pp. (incl. 3 maps) . ISBN 0 85883 337 9 Papers by Susanne Roma ine ; Pe ter Muhlhausler ; Linda Sim ons ; Christine Jourdan ; Roger M. Keesing ; Doug Munro ; wa lter Seiler; Anna Shnukal ; Margaret C. Sharpe ; John R. Sandefur ; Philip Baker and R. Ramnah . PACIFIC LINGUISTICS 6

Se ries A - Occasional Papers (continued )

No .74. Papers in New Guinea linguistics No .25. 1986 ; iv+261pp . (incl. 10 maps) . ISBN 0 85883 338 7 Papers by Linda K. Jones; Larry B. Jones; Francesca Me rlan and Alan Rumsey; Martha Reimer ; David L. Martin; De idre Shelden; Hiroko Oguri.

IN PREPARATION Papers in Philippine linguistics No .ll. Papers in New Guinea linguistics No .26. Papers in Austral ian linguistics No .17. Papers in western Austronesian linguistics No .3. Papers in We stern Austronesian linguistics No .4. Papers in South-East As ian linguistics No .lO. PACIFIC LINGUISTICS 7

SERIES B - MONOGRAPHS

No . 1. WURM , S.A. and J.B. HARRIS Police Motu : an introduction to the trade language of Papua (New Guinea) for anthropolog ists and other fieldworkers. 1963; vi+81pp . Repr inted 1964 , 1965 , 1966, 1967 , 1969, 1970, 1971, 1973 . ISBN 0 85883 034 5

No . 2. WURM S.A. Phonolog ical diversification in Australian New Guinea highlands languages. 1964 ; iii+87pp . + 1 map. Repr inted 1971. ISBN 0 85883 035 3 No . 3. HEALEY , Alan Te lefol phonology. 1964; ii+53pp . + 2 figures, S tables . Reprinted 1972, 1981. ISBN 0 85883 036 1

No . 4. HEALEY , Phyllis M. Telefol noun phrases. 1965 ; iii+51pp. Repr inted 1972 . ISBN 0 85883 037 X No . 5. HEALEY , Phyllis M. Levels and chaining in Te lefol sentences. 1966 ; iv+64pp. Repr inted 1971. ISBN 0 85883 038 8 No . 6. TRYON , Darrell T. Nengone grammar . 1967 ; x+91pp. Reprinted 1971. ISBN 0 85883 039 6 No . 7. TRYON , D.T. Dehu grammar . 1968 ; xi+lllpp . Re pr inted 1971. ISBN 0 85883 040 X

No . 8. TRYON, Darrell T. Iai grammar . 1968 ; xii+l25pp. Repr inted 1971. ISBN 0 85883 041 8 No . 9. DUTTON, T.E. The peopl ing of Central Papua : some preliminary observations . 1969; viii+l82pp . Repr inted 1970 , 1971. ISBN 0 85883 042 6 No .lO. FRANKLIN, K.J. The dialects of Kewa . 1968 ; iv+72pp. (incl. 20 maps) . Repr inted 1971. ISBN 0 85883 043 4 No .ll. SOMMER, B.A. Kunjen phonology: synchronic and diachronic . 1969; iv+72pp. (inc 1. 3 maps) . ISBN 0 85883 044 2 No .12. KLOKEID, Terry J. Thargari phonology and morphology. 1969; viii+56pp . (incl. 1 map) . Repr inted 1981. ISBN 0 85883 045 0 No .13. TREFRY , D. A comparative study of Kuman and Pawaian . 1969 ; v+94pp . (incl. 1 map) . Reprinted 1980 . ISBN 0 85883 046 9 No .14. McELHANON , K.A. Se lepet phonology . 1970 ; v+47pp. (inc l. 1 map) . ISBN 0 85883 003 5 No .15. TRYON, D.T. An introduction to Ma ranungku (Nor thern Australia) . 1970; x+lllpp . (inc1. 1 map) . ISBN 0 85883 047 7 No .16. McELHANON , K.A. and C.L. VOORHOEVE The Trans-New Guinea Phylum : explorations in deep-level genetic relationships . 1970; v+l07pp. (incl. 4 maps) . Repr inted 1978 . ISBN 0 85883 048 5 No .17. KUKI, Hiroshi Tuamotuan phonology. 1970; ix+119pp . + 2 maps . ISBN 0 85883 049 3 No .18. YOUNG , R.A. The verb in Bena-bena : its form and function . 1971; v+68pp . ISBN 0 85883 050 7 No .19. PATON, W.F. Ambrym (Lonwolwol) grammar . 1971; xi+128pp . (incl. 1 map) . ISBN 0 85883 051 5 No .20. CAPELL , A. Arosi grammar . 1971; iv+90pp. (inc l. 1 map) . ISBN 0 85883 052 3 No .21. McELHANON, K.A. Selepet grammar. Part I: From root to phrase . 1972; vi+116pp . ISBN 0 85883 085 X and 0 85883 086 8 No .22. McELHANON , K.A. Towards a typology of the Finiste rre-Huon languages, New Guinea. 1973 . vii+73pp. (inc1. 1 map) . ISBN 0 85883 094 9 No .23. SCOTT , Graham Higher levels of Fore grammar . Ed i ted by Robert E. Longacre. 1973 ; x+88pp . ISBN 0 85883 088 4 PACIFIC LINGUISTICS 8

Series B - Monog raphs (continued) No .24. DUTTON, T.E. A checklist of languages and present-day villages of central and south-east mainland Papua . 1973; iv+80pp . (incl. 1 map) . ISBN 0 85883 090 6 No .25. LAYCOCK , D.C. Sepik languages - checklist and preliminary classification . 1973 ; iv+130pp. (inc l. 1 map) . ISBN 0 85883 084 1 No .26. MUHLHAUSLER , P. Pidginization and simpl ification of language . 1974 ; v+161pp . Repr inted 1978 , 1981. ISBN 0 85883 113 9 No .27. RAMOS , Teresita V. The case system of Tagalog verbs. 1974; viii+168pp. ISBN 0 85883 115 5 No .28. WEST , Dorothy Wo jokeso sentence , paragraph , and discourse analysis. Edited by Robert E. Longacre . 1973 ; x+181pp . ISBN 0 85883 089 2 No .29. ELBERT , Samuel H. Puluwat grammar . 1974 ; v+137pp. ISBN 085883 103 1 No .30. METCALFE , C.D. Bardi ve rb morphology (northwestern Australia) . 1975; x+215pp. (incl. 1 map) . ISBN 0 85883 121 x No .31. VOORHOEVE , C.L. Languages of Irian Jaya : checklist . Preliminary classification , language maps , wordlists . 1975; iv+129pp. (incl. 17 maps) . Repr inted 1980 . ISBN 0 85883 128 7 No .32. WALTON , Janice Binongan Itneg sentences. 1975; vi+70pp . ISBN 0 85883 117 1 No .33. GUY , J.B.M. A grammar of the nor thern dialect of Sakao . 1974; ix+99pp . (incl. 2 maps) . ISBN 0 85883 104 X No .34. HOPE , Edward Reginald The deep syntax of Lisu sentences: a transformational case grammar . 1974; viii+184pp. (incl . 1 map) . ISBN 0 85883 110 4

No .35. IRWIN, Barry Salt-Yui grammar . 1974 ; iv+151pp. ISBN 0 85883 III 2 No .36. PHILLIPS , Donald J. Wahgi phonology and morphology . 1976; x+165pp. ISBN 0 85883 141 4 No .37. NGUYEN �ANG LIEM Cases, clauses and sentences in Vietnamese . 1975; v+89pp . ISBN 0 85883 133 3 No .38. SNEDDON , J.N. Tondano phonology and grammar . 1975; viii+264pp. ISBN 0 85883 125 2 No .39. LANG , Ad rianne The semantics of classificatory verbs in Enga (and other Papua New Guinea languages) . 1975; xii+234pp. (inc l. 2 maps) . ISBN 0 85883 123 6 No .40. RENCK , G.L. A grammar of Yagar ia. 1975; xiii+235pp . (incl. 1 map) . ISBN 0 85883 130 9 No .41. Z'GRAGGEN, John A. The languages of the Madang District, Papua New Guinea. 1975; vi+l54pp . (incl. 1 map) . Repr inted 1979. ISBN 085883 134 1 No .42. FURBY , E.S. and C.E. FURBY A preliminary analysis of Garawa phrases and clauses . 1977; viii+lOlpp . ISBN 0 85883 151 1 No .43. STOKHOF , W.A.L. Preliminary notes on the Alor and Pantar languages (East Indonesia) . 1975; vi+73pp . (incl. 2 maps) . Reprinted 1979. ISBN 0 85883 124 4 No .44. SAYERS , Barbara J. The sentence in Wik-Munkan : a descript ion of propositional relationships . 1976; xvi i+185pp. ISBN 0 85883 138 4 No .45. BIRK , D.B.W. The MalakMalak language , Daly River (Western Arnhem Land) . 1976; xii+179pp . (incl. 1 map) . ISBN 0 85883 150 3 No .46. GLISSMEYER , Gloria A tagmemic analysis of Hawaii English clauses. 1976; viii+149pp. ISBN 0 85883 142 2 PACIFIC LINGUISTICS 9

Series B - Monographs (continued) No .47. SCOTT , Graham The Fore language of Papua New Guinea. 1978; xv+210pp. (incl. 2 maps) . ISBN 0 85883 173 2 No .48. CLARK , Marybe th Cove rbs and case in Vietnamese . 1978; xi+215pp . ISBN 0 85883 162 7 No .49. FILBECK, David T'in: a historical study . 1978; vi+lllpp . (incl. 2 maps) . ISBN 0 85883 172 4 No .50. SMITH , Kenneth D. Sedang grammar ; phonolog ical and syntactic structure . 1979; xix+191pp . (inc!. 3 maps) . ISBN 0 85883 180 5 No .51. WELLS , Margaret A. Siroi grammar . 1979; vii+2 18pp. ISBN 0 85883 181 3 No .52. KILHAM , Chr istine A. Thematic organization of Wik-Munkan discourse . 1977 ; xix+280pp. (incl. 1 map) . ISBN 0 85883 168 6 No .53. VESALAINEN, Olavi and Ma rja VESALAINEN Clause patterns in Lhomi. 1980 ; vii+lOOpp. ISBN 0 85883 210 0 No .54. SNEDDON , J.N. Proto-Minahasan : phonology , morphology and wordlist. 1978; x+204pp. (inc!. 1 map) . ISBN 0 85883 169 4 No .55. LYNCH , John A grammar of Lenakel. 1978; vii+135pp . (incl. 1 map) . ISBN 0 85883 166 X No .56. ROSS , Malcolm with John Natu PAOL A Wa skia grammar sketch and vocabulary. 1978 ; v+119pp . ISBN 0 85883 174 0 No .57. BLAKE , Barry J. A Kalkatungu grammar . 1979; xii+198pp. (incl. 1 map) . ISBN 0 85883 197 X No .58. BEAUMONT , Cl ive H. The of New Ireland . 1979; xi+163pp. (incl. 2 maps) . ISBN 0 85883 187 2 No .59. STOKHOF , W.A.L. Wo isika II: phonemics. 1979; xi+188pp. (incl. diagrams , photographs , 3 maps) . Repr inted 1981. ISBN 0 85883 190 2 No .60. FOX , G.J. Big Nambas grammar . 1979; xii+139pp. (incl. 2 maps) . ISBN 0 85883 183 X No .61. HAWK INS, Em ily A. Hawaiian sentence structures . 1979; iii+lllpp. ISBN 0 85883 195 3 No .62. HEATH , Jeffrey Basic materials in Ritharngu: grammar , texts and dictionary. 1980; ix+249pp. (inc!. 1 map) . ISBN 0 85883 204 6 No .63. LUZARES , Casilda Ed rial The morphology of selec ted Cebuano verbs : a case analysis. 1979; xii+208pp . ISBN 0 85883 199 6 No .64. VooRHOEVE , C.L. The Asmat languages of Irian Jaya . 1980; x+177pp. (inc!. 5 maps) . ISBN 0 85883 207 0

No .65. McDONALD , M. and S.A. WURM Basic materials in WaQ kumara (Ga!ali) : grammar , sentences and vocabulary. 1979; ix+lllpp. (incl. 2 maps) . ISBN 0 85883 202 X No .66. WOOLFORD , Ellen B. Aspec ts of Tok Pisin grammar . 1979; v+118pp . ISBN 0 85883 203 8 No .67. HERCUS, L.A. The Bagandji language . 1982; xvi ii+329pp . (incl. 6 maps , 10 photographs) . ISBN 0 85883 263 1 No .68. DAVIES , H.J. Kobon phonology. 1980; v+80pp. (incl. 3 maps) . ISBN 0 85883 211 9 No .69. SOBERANO , Rosa The dialects of Mar induque Tagalog . 1980 ; xii+232pp . (incl. 42 maps) . ISBN 0 85883 216 X No . 70 . JOHNSTON , Raymond Leslie Nakanai of New Br itain : the grammar of an Oceanic language . 1980; xiii+310pp. (inc!. 2 maps) . ISBN 0 85883 209 7 No .7!. CHAN YAP , Gloria Hokkien Chinese borrowings in Tagalog . 1980; viii+155pp. ISBN o 85883 225 9 No .72. HEATH , Jeffrey Basic materials in Warndarang : grammar , texts and dictionary. 1980 ; xii+174pp. (inc!. 1 map) . ISBN 0 85883 219 4 PACIFIC LINGU ISTICS 10

Se ries B - Monographs (continued)

No .73. MOSEL , Ulrike Tolai and Tok Pisin: the influence of the substratum on the development of New Guinea Pidg in . 1980 , viii+146pp. ISBN a 85883 229 1 No .74. WILLIAMS , Corinne J. A grammar of Yuwaalaraay . 1980 , vii i+200pp. (incl. 1 map) . ISBN a 85883 221 6 No .75. GUY , J.B.M. Experimental glottoc hronology: ba sic methods and results. 1980, vii+217pp. ISBN a 85883 220 8 No . 76. PERCIVAL , W.K. A grammar of the urbanised Toba-Batak of Medan. 1981, vi+125pp . ISBN a 85883 237 2 No.77. HERCUS, L.A. Victorian languages: a late survey. 1986, xx ix+303pp. (incl. 6 maps , 19 photographs) . ISBN a 85883 322 a

No .78. TSUNODA , Tasaku The Djaru language of Kimberley , We stern Australia. 1981; xxi+290pp . (inc l. 3 maps , S photographs) . Repr inted 1987. ISBN a 85883 252 6

No .79. GUY , J.B.M. Glottochronology without cognate recognition . 1981; viii+134pp. ISBN a 85883 235 6 No .80. DURANTI , Alessandro The Samoan fono : a sociolinguistic study. 1981, xi+195pp. (incl . 3 maps , S photographs) . ISBN a 85883 248 8 No .81. TCHEKHOFF, Claude Simple sentences in Tongan. 1981, iv+95pp. ISBN a 85883 251 8 No .82. TIPTON , Ruth A. Nembi procedural and narrative discourse . 1982; v+87pp . ISBN a 85883 259 3 No .83. THURSTON, William R. A comparative study in Anem and Lusi. 1982, ix+l07pp . (inc! . 3 maps) . ISBN a 85883 260 7 No .84. CAUGHLEY , Ross Charles The syntax and morphology of the verb in Chepang . 1982, xvi+269pp. (inc!. 3 maps) . ISBN a 85883 278 X No .85. WILSON , William H. Proto-Polynesian possessive marking . 1982, xv+137pp . ISBN a 85883 270 4 No .86. RUMSEY , Alan An intra-sentence grammar of Ungar injin, nor th-western Australia . 1982, xii+l79pp . (inc!. 1 map) ISBN a 85883 272 a No .87. CROWLEY , Te rry The of Vanuatu . 1982, xii+268pp. (inc! . 3 maps) . ISBN a 85883 279 8

No .88. KEESING, Roger M. Kwaio grammar . 1985 , vii+299pp . ISBN a 85883 327 1 No .89. MERLAN , Francesca Ng alakan grammar , texts and vocabulary. 1983, xi+220pp . (inc!. 1 map) . ISBN a 85883 282 8

No .90. HARTMANN, John F. Linguistic and memory structures in Ta i-Lue oral narratives . 1984 ; ix+230pp. (inc !. 8 maps) . ISBN a 85883 296 8 No .91. SNEDDON, J.N. Proto-Sang iric and the . 1984, viii+138pp . (inc!. 2 maps) . ISBN a 85883 306 9 No .92. MOSEL , Ulrike Tolai syntax and its histor ical development. 1984, iv+223pp . ISBN a 85883 309 3 No .93. SEILER, Wal ter. Imonda, a Papuan language . 1985 , v+236pp. (inc!. 1 map) . ISBN a 85883 326 3 PACIFIC LINGUISTICS 11

Series B - Monog raphs (continued)

No .96. FAGAN , Joel L. A grammatical analysis of Mono-Alu (Bougainville Straits, Solomon Islands). 1986 ; vii+158pp. (incl. 1 map) . ISBN 0 85883 339 5

IN PREPARATION:

AUSTIN, Peter , Luise A. HERCUS, and Stephen A. WURM Basic materials in Malyangaba : grammar , sentences and vocabulary.

FELDMAN, Harry A grammar of Awtuw.

HURLBUT , Hope M. Ve rb morphology in Eastern Kadazan .

NOTE : For further monog raphs on Indonesian languages see under Ser ies D - the subseries Materials in languages of Indonesia. PACIFIC LINGUISTICS 12

SERIES C - BOOKS No . 1. LAYCOCK , D.C. The Nd u language family (Sepik District, New Guinea) . 1965 ; xi+224pp . (inc1. 1 map) . ISBN 0 85883 053 1 No . 2. GRACE , George W. Canala dictionary (New Caledonia) . 1975; ix+128pp. ISBN 0 85883 122 8 'X .... 1\ No . 3. NGUYEN �ANG LIEM English grammar : a combined tagmemic and trans- formational approach. (A contrastive analysis of English and Vietnamese , vol.l.) . 1966 ; xliv+177pp . Repr inted 1970 . ISBN 0 85883 054 X and ISBN 0 85883 055 8 No . 4. NGUY�N �ANG LIEM Vietnamese grammar : a combined tagmemic and transformational approach. (A contrastive analysis of English and Vietnamese , vol.2.) . 1969 ; xlvi+209pp . Repr inted 1975 . ISBN 0 85883 054 and ISBN 0 85883 056 6 1< A No . 5. NGUYEN �Av NG LIEM A contrastive gramma tical analysis of English and Vietname se . (A contrastive analysis of English and Vietnamese , vol . 3.) . 1967; xv+151pp. Repr inted 1971. ISBN 0 85883 054 X and ISBN 0 85883 057 4 No . 6. TRYON, Darrell T. Dehu-English dictionary. 1967; v+137pp. Reprinted 1971. ISBN 0 85883 058 2 No . 7. TRYON , Darrell T. English-Dehu dictionary. 1967; iii+162pp . Reprinted 1971. ISBN 0 85883 059 0 No . 8. NGUY�N �ANG LI�M A contrastive phonological analysis of Eng lish and Vietnamese . (A contrastive analysis of English and Vietnamese , vol.4.) . 1970; xv+206pp . ISBN 0 85883 054 X and ISBN 0 85883 004 3 No . 9. TRYON, D.T. and M.-J. DUBOIS Nengone dictionary. Part I: Nengone­ English . 1969 ; vii+445pp . (out of pr int) ISBN 0 85883 060 4 and ISBN 0 85883 061 2 No .10. OATES , W. and L. OATES Kapau pedagog ical grammar . 1968 ; v+178pp. Reprinted 1971. ISBN 0 85883 062 0 No .ll. FOX , C.E. Arosi-English dictionary. 1970 ; iv+4 06pp. (inc l. 1 map) . ISBN 0 85883 063 9 No .12. GRACE , George W. Grand Couli dic tionary (New Caledonia) . 1976; vii+113pp . ISBN 0 85883 154 6 No .13. WURM , S.A. and D.C. LAYCOCK, eds Pac ific linguistic stud ies in honour of Arthur Capell. 1970 ; xi+1292pp . (Hardbound : incl. 25 maps , 1 photograph) . Repr inted 1974 , 1975 , 1978 , 1979. ISBN 0 85883 005 1 Articles by Byron W. Bender, Catherine H. Berndt, Ronald M. Berndt, H. Bluhme , J.E. Bolt, C.G. von Brandenstein, C. Douglas Chretien, J.R. Cleverly, Christopher Cour t, R.M.W. Dixon , Wilfrid H. Douglas , T.E. Dutton , Isidore Dyen , Samuel H. Elbert, A.P. Elkin , E.H. Flint, Karl J. Franklin, Marie Godfrey , George W. Grace , Kenneth Hale , Joy Harris, Alan Healey , Henry Hershberger , Ruth Hershberger , W.G. Hodd inott, Patrick W. Hohepa , Nils M. Holmer , B.A. Hooley, Dorothy J. James, Hans Kahler , Susan Kaldor , Harland Kerr, Jean F. Kirton , D.C. Laycock, K.A. McElhanon , Howard McKaughan , Nguy�n ��ng Li�m , Geoffrey N. O'Grady , Andrew Pawley , Eunice V. Pike , Richard Pittman , D.J. Prentice , Albert J. Schut z, M.C. Sharpe , W.E. Smythe , A.J. Taylor , D.T. Tryon , E.M. Uhlenbeck, C.F. Voegelin, F.M. Vo egelin, C.L. Voorhoeve , S.A. Wurm , John A. Z'graggen. No .14. GEERTS , P. 'Are 'are dictionary. 1970; iv+187pp. (inc l. 2 maps) ISBN 0 85883 064 7 No .15. McELHANON, K.A. and N.A. McELHANON Selepet-English dictionary. 1970; xxi+144pp . ISBN 0 85883 065 5 No .16. FRANKLIN, K.J. A grammar of Kewa , New Guinea. 1971; ix+138pp. ISBN 0 85883 066 3 No .17. PARKER, G.J. Southeast Ambrym dictionary. 1971; xiii+60pp. ISBN 0 85883 067 1 PACIFIC LINGUISTICS 13

Se ries C - Books (Continued ) No .18. PRENTICE, D.J. The Murut languages of Sabah . 1971; xi+311pp. (incl. 1 map) . ISBN 0 85883 068 X No .19. Z'GRAGGEN, J.A. Classificatory and typolog ical studies in languages of the Madang District. 1971; viii+l79pp . (incl. 4 maps) . ISBN 0 85883 069 8 No .20. LANG , Ad rianne Enga dictionary, with English index. 1973; lxi+219pp. (Hardbound : incl . 1 map) . Repr inted 1978. ISBN 0 85883 093 0 No .21. PATON, W.F. Ambrym (Lonwolwol) dictionary. 1973; ix+337pp . (Hardbound ) + 1 map. ISBN 0 85883 092 2 No .22. LONGACRE , Robert E. , ed . Philippine discourse and paragraph studies in memory of Betty McLachlin . 1971; xv+366pp. (incl. 1 photograph). ISBN 0 85883 070 1 Ar ticles by Barbara Blackburn, R.E. Longacre, Betty McLachlin, Charles Walton , Claud ia Wh ittle , Hazel J. Wr igglesworth . No .23. TRYON, D.T. and M.-J. DUBOIS Nengone dictionary. Part II: English­ Nengone . 1971; iii+202pp. ISBN 0 85883 060 4 and ISBN 0 85883 071 X No .24. ELBERT , Samuel H. Puluwat dictionary. 1972; ix+401pp. (Hardbound) . ISBN 0 85883 082 5 No .25. FOX , Charles E. Lau dictionary, with English index . 1974; vi+260pp . (Hardbound) . Repr inted 1976, 1978 . ISBN 0 85883 101 5 No .26. FRANKLIN, Karl, ed . The linguistic situation in the Gulf District and ad jacent areas, Papua New Guinea. 1973; x+597pp. (Hardbound : incl. 8 maps) . Reprinted 1975. ISBN 0 85883 100 7 Articles by H.A. Brown , T.E. Dutton , Karl J. Franklin, Richard G. Lloyd , George E. MacDonald , Karen Shaw, R. Daniel Shaw, Clemens L. Voorhoeve , S.A. Wurm . No .27. SOHN, Ho-min and B.W. BENDER A Ulithian grammar . 1973; xv+398pp. (Hardbound : incl . 2 maps) . Repr inted 1980 . ISBN 0 85883 098 1 No .28. HEADLAND , Thomas N. and Janet D. HEADLAND A Dumagat (Casiguran) ­ English dictionary. 1974 ; lxiii+2 32pp. (Hardbound : incl . 1 map) . ISBN 0 85883 107 4 No .29. DUTTON , T.E. , ed . Stud ies in languages of Central and South-East Papua. 1975; xvii+834pp . (Hardbound : incl. 5 maps) . Repr inted 1978 . ISBN 0 85883 119 8 Articles by John Austing , Russell E. Cooper , T.E. Dutton , Cynthia Farr, James Farr, Roger Garland , Susan Garland , J.E. Henderson , J.A. Kolia, Mike Olson , Andrew Pawley, Ernest L. Richert, N.P. Thomson , Randolph Upia, Harry We imer , Natalia We imer . No .30. LOVING, Richard and Aretta LOVING Awa dictionary. 1975 ; xlv+203pp . (Hardbound : incl . 1 map) . ISBN 0 85883 137 6 No .31. NGUY�N �ANG LIEM, ed . South-East Asian linguistic studies , vol .l. 1974 ; vii+213pp. (Hardbound) . Repr inted 1978 . ISBN 0 85883 144 9 and ISBN 0 85883 099 X Ar ticles by Ma rybe th Clark, Ar thur G. Crisfield , Soenjono Dardjowidjojo, Cesar A. Hidalgo , Philip N. Jenner , Nguy@n �ang Li@m , Saveros Pou . No .32. TRYON, D.T. Daly Fami ly languages, Australia. 1974 ; xvii+305pp. (Hardbound : incl. 1 map) . Reprinted 1980 . ISBN 0 85883 106 6

No .33. WURM, S.A. and B. WILSON English finderlist of reconstructions in Austronesian languages (post-Brandstetter) . 1975; xxxii+246pp . (Hardbound ). Reprinted 1978. ISBN 0 85883 129 5 No .34. GUY , J.B.M. Handbook of Bichelamar - Manuel de Bichelamar . 1974; iii+256pp . (Hardbound). Reprinted 1975, 1979 . ISBN 0 85883 109 0 No.35. KEESING, R.M. Kwaio dictionary. 1975; xxxv+296pp . (Hardbound : incl. 1 map) . Reprinted 1981. ISBN 0 85883 120 1 PAC IFIC LINGU ISTICS 14

Se ries C - Books (continued) No .36. RE ID, Lawrence A. Bontok-English dictionary . 1976; xxiii+500pp. (Hardbound) . ISBN 0 85883 145 7 No .37. RENCK , G.L. Yagaria dictionary, with English index . 1977; xxix+327pp . (Hardbound : incl . 1 map) . ISBN 0 85883 161 9

No .38. WURM, S.A. , ed . New Guinea area languages and language study , vol .l: Papuan languages and the New Guinea linguistic scene . 1975; xlvi+l038pp. (Hardbound : incl . 28 maps) . Repr inted 1977 . ISBN 0 85883 131 7 and ISBN 0 85883 132 5 Articles by A. Capell, T.E. Dutton , Karl J. Franklin , Harland B. Kerr, D.C. Laycock, K.A. McElhanon , Evelyn M. Todd , C.L. Voorhoeve , S.A. Wurm, John A. Z'graggen .

No .39. WURM, S.A. , ed . New Guinea area languages and language study , vol .2: Austronesian languages . 1976; xxxv+736pp. (Hardbound : incl. 21 maps) . Repr inted 1979. ISBN 0 85883 131 7 and ISBN 0 85883 155 4 Articles by C.H. Beaumont, A. Capell , Ann Chowning , T.E. Dutton , George W. Grace , Alan Healey , Bruce A. Hooley , D.C. Laycoc k, Peter C. Lincoln , David R. Lithgow , Andrew Pawley, A.J. Taylor , S.A. Wurm, John A. Z'graggen.

No .40. WURM, S.A. , ed . New Guinea area languages and language study , vol.3: Language , culture , soc iety , and the modern wor ld . 1977; lxxxvi+1449pp . (Hardbound : incl . 3 maps , 40 photographs) , in two fascicles. Repr inted 1981, 1985 . ISBN 0 85883 131 7 and ISBN 0 85883 159 7 Articles by C. Abel , Henry L. Bell, Catherine H. Berndt, H. Myron Bromley , H.A. Brown , A. Capell, Lo is Carrington , Em ily Clarke , Anne M. Cochran , E.W. De ibler Jr , T.E. Dutton , Irenaus Eibl-Eibesfeldt, Franz-Josef Eilers, Joice Franklin , Karl J. Franklin, Paul G. Freyberg , E. Fry, Alan Healey, L.R. Healey , R.K. Johnson , Ad rianne Lang , Ranier Lang , Ralph S. Lawton , D.C. Laycock, Peter C. Lincoln , John Lynch , Howard P. McKaughan , Francis Mihalic , John Minogue , Peter Muhlhausler , A.K. Neuendor f, Ebia Olewale , Andrew Pawley, G.L. Renck, Joan Rule , W.M. Rule , Gillian Sankoff, Robert P. Scott, Pe ter J. Silzer , A.J. Taylor , W.E. Tomasetti , Donald F. Tuzin , C.L. Voorhoeve , David Y.H. Wu , S.A. Wurm, John A. Z'graggen . Calvin Zinkel. No .41. FLIERL , W. and H. STRAUSS , eds K�te dictionary. 1977 ; xxxv+499pp. (Hardbound : incl. 1 map) . ISBN 0 85883 149 X No .42. NGUYftN�ANG LI�M , ed . South-East Asian ling uistic studies, vol .2. 1976; iv+262pp. (Hardbound : incl . 2 maps) . ISBN 0 85883 144 9 and ISBN 085883 143 0 Ar ticles by Paul K. Benedict , G. Diffloth , Eugenie J.A. Henderson , Judith M. Jacob, Philip N. Jenner , Joseph F. Kess, A. Llamzon , Ma . Te resita Martin, Malcolm Warren Mintz, Lili Rabel-Heymann , H.L. Shorto , Sidharta (Sie Ing Dj iang) , John U. Wol ff. No .43. SMALLEY, William A. , ed . Phonemes and orthography : language planning in ten minor ity languages of Thailand . 1976 ; xiii+347pp . (Hardbound) ISBN 0 85883 144 9 Ar ticles by C.W. Callaway , Lo is Callaway , Joseph R. Cooke , David Filbeck, David Hogan , E.R. Hope , J. Edwin Hudspith, Beulah M. Johnston , James A. Mor ris, Donald Schlatter , William A. Smalley , Peter Wyss. No.44. ZORC , David Paul The Bisayan dialects of the Ph ilippines: subgrouping and reconstruction. 1977; xxiii+328pp. (Hardbound : incl. 9 maps) . ISBN 0 85883 157 0 No.45. NGUytN �ANG LltM, ed . South-East Asian linguistic studies, vol .3. 1979; ix+326pp. (Hardbound) + 3 maps . ISBN 0 85883 144 9 and ISBN 0 85883 177 5 Ar ticles by A. Capell, Soenjono Dardjowidjojo, Raleigh Ferrell, Jackson T. Gandour , Eugenie J.A. Hender son , Joseph F. Kess , Ng uy�n �ang Liem, A. Kemp Pallesen , Alejandr ino Q. Perez, Lawrence A. Re id , Al fonso o. Santiago , Patr ic ia Stanley , Norman H. Zide , R. David Zorc . PACIFIC LINGUISTICS 15

Series C - Books (continued) No .46. HEALEY , Phyllis and Alan HEALEY Telefol dictionary. 1977; xix+358pp. (Hardbound ). ISBN 0 85883 160 0 No .47. PEREZ, A.Q. , A.O. SANTIAGO and NGUyfN �ANG LI�, eds Papers from the Con ference on the Standardisation of Asian Languages, Manila, Philippines, December 16-21, 1974 . 1978 ; ix+386pp . (Hardbound : incl . 1 map) . ISBN 0 85883 176 7 Ar ticles by Abdullah Hassan , S. Takdir Alisjahbana , Asmah Haji Omar , Ne lia G. Casambre , James R. Chamberlain, Sisir Kumar Das, Virgilio G. Enriquez, Andrew Gon zalez, Am ran Halim, Astuti Hendrato-Darmosug ito, Yahaya Ismail, Hans Kaehler , Har imurti Kridalaksana , Ernest W. Lee , Ferdinand E. Marcos , S.W. Rudjiati Mul jadi, Nguy�n �ang Li�m , Fe T. Otanes, Al Q. Pe rez, Ponc iano B.P. Pineda, Wissanu Rawanking , Jack C. Richards , Al fonso o. Santiago , Takuji Sasaki , Bonifac io P. Sibayan , Hengtse Tu , Lars S. Vi kor . No .48. GONZALEZ , Andrew Pampangan : towards a meaning-based description . 1981; xiii+402pp. (Hardbound) ISBN 0 85883 244 5 No .49. NGUyfN �ANG LIEM, ed . South-East Asian linguistic studies, vol .4. 1979; iv+436pp. (Hardbound ). ISBN 0 85883 144 9 and ISBN 0 85883 201 1 Articles by Donald F. Barr, Maxwell Cobbey, James T. Collins, Joseph R. Cooke , Anthony Diller , Jack Gandour , John F. Hartmann , SamAng Hiranburana , Lou Hohulin, Mary E. Honts , Hope M. Hurlbut, Ph ilip N. Jenner , Wilaiwan Kanittanan , Michael Kenstowicz, Joseph F. Kess , Anna Kwan-Terry, Nguy�n �inh-Hoa , Patcharin Peyasantiwong , Saveros POu , Joan M. Ro sen , Marmo Soemarmo, Tham Seong Chee , Michael R. Thomas, Udom Warotamasikkhadit. No .50. TRYON, D.T. New Hebrides languages: an internal classification . 1976; v+545pp. (Hardbound : incl . 7 maps) . Repr inted 1979 . ISBN 0 85883 152 X No .51. GLOVER, Wa rren W. , Jessie R. GLOVER and De u Bahadur GURUNG Gurung-Nepali-English dictionary, with English-Gurung and Nepali-Gurung indexes. 1977 ; xiii+316pp . (Hardbound ). ISBN 0 85883 147 3 No .52. MUHLHAUSLER , Peter Growth and structure of the lexicon of New Guinea Pidgin. 1979; xx+498pp. (Hardbound) . Repr inted 1984 . ISBN 0 85883 191 0 No .53. FRANKLIN, Karl J. and Joice FRANKLIN, assisted by Yapua KIRAPEASI A Kewa dictionary, with supplementary grammatical and anthropological materials. 1978; xi+514pp. (Hardbound : incl . 10 maps) . ISBN 0 85883 182 1

No .54. WORM, S.A. , ed . Australian ling uistic studies . 1979; xv+753pp . (Hardbound : incl. 18 maps , 7 illustrations). ISBN 0 85883 185 6 Articles by Barry J. Blake , A. Capell, Lo is Carr ington , Ne il Chadwick, Je ffrey Heath, L.A. Hercus , Geoffrey N. O'Grady, Bruce Rigsby , M.C. Sharpe , Peter Sutton , Michael J. Wa lsh. No .55. LYNCH , John Lenakel dictionary. 1977; vii+167pp. ISBN 0 85883 165 1 No .56. CAPELL , A. Futuna-Aniwa dictionary, with grammatical introduction . 1984 ; iv+252pp. ISBN 0 85883 316 6 No.57. FOX , Charles E. Arosi dictionary. Revised ed ition with English-Arosi index prepared by Mary Craft. 1978 ; iv+598pp. (Hardbound : incl . 1 map) . ISBN 0 85883 170 8 No .58. THARP , J.A. and Y-Bham �UON-YA A Rhade-English dictionary, with English-Rhade finderlist. 1980 ; xi+271pp . (Hardbound) . ISBN 0 85883 217 8 No .59. BAUTISTA , Ma ria Lourdes S. The Filipino bilingual 's compe tence : a model based on an analysis of Tagalog-English code switching . 1980 ; vi+386pp . (Hardbound) . ISBN 0 85883 212 7 PACIFIC LINGUISTICS 16

Series C - Books (continued ) No .60. HEATH , Jeffery Basic materials in Mara: grammar , texts and dictionary. 1981; xiii+522pp. (Hardbound : incl . 1 map) . ISBN 0 85883 247 X

No .6l. WURM, S.A. and Lois CARRINGTON , eds Second International Con ference on Austronesian Linguistics: proceed ings. Fascicle one : we stern Austronesian 1978 ; xxii+1-688pp. (Hardbound : incl. 1 map , 2 photographs) Repr inted 1983. ISBN 0 85883 184 8 Articles by Natalia Al ieva , J.C. Anceaux , , Alice Cartier , Sandra Chung , Otto Chr . Dahl , Soenjono Dardjowidjojo, Isidore Dyen , Ra leigh Ferrell , Jo-Ann Flora, Jeanne D. Gibson , A. Hakim Usman , R. Hardjadibrata, Hans Kahler , Joseph F. Kess, Don Laycock, Paul Jen-kuei Li,. Paz Buenaventura Naylor , D.J. Prentice , Lawrence A. Re id , J.P. Sarumpaet , U. Sirk, H. Steinhauer , Claude Tche khoff, Michael R. Thomas, John W.M. Ve rhaar , Fran90is Zacot, R. David Zorc . Fascicle two : Eastern Austronesian . 1978; xxii-xxvi+688-l497pp . (Hardbound : incl . 6 maps) Repr inted 1983. ISBN 0 85883 184 8 Articles by David G. Arms , Bruce G. Biggs, Ann Chowning , Ro ss Clark, Anne Cochran , Tom Du tton , Bryan Ezard , Jacques Bernard Michel Guy , S.P. Harrison , Marianne Haslev , Raymond L. Johnston , Yves Lema�tre , P.C. Lincoln , John Lynch , Donald S. Marshall , Rodney F. Moag , Pe ter Muhlhausler , Andrew Pawley, Albert J. Schutz, William J. Seiter , Andrew Taylor , Evelyn M. Todd , D.T. Tryon , D.S. Walsh, S.A. Wurm. No.62. SCOTT , Graham Fore dictionary. 1980 ; xiii+243pp . (Hardbound : incl. 1 map) . ISBN 0 85883 226 7 No.63. BROMLEY , H. Myron A grammar of Lower Grand Valley Dani . 1981; xiv+424pp . (Hardbound) . ISBN 0 85883 223 2 No .64. COPPELL , W.G. Austronesian and other languages of the Pacific and South-East Asia: an annotated catalogue of theses and dissertations. 1981; xiii+52lpp . (Hardbo und) ISBN 0 85883 238 0 No .65. RANBY , Peter A Nanumea lexicon . 1980 ; xi+243pp. (Hardbound) . ISBN 0 85883 227 5

No .66. WURM , S.A. and Shir8 HATTORI , eds Language atlas of the Pacific area, part 1: New Guinea area, Oceania , Australia. 1981; ii+74pp . (incl. 25 multicoloured maps , appropr iate text materials, indexes). (Boxed set) ISBN 0 85883 239 9 and ISBN 0 85883 240 2 (Distributed by GeoCenter , Honigwiesenstrasse 25, D-7 000 Stuttgart 80, Postfach 80 08 30, We st Germany. Price DM 250 .-)

No .67. WURM, S.A. and Shir8 HATTORI , eds Language atlas of the Pac ific area , part 2: Japan area , Philippines and Formosa , mainland and insular South-east Asia. 1983; ii+72pp. (incl. 23 multicoloured maps , appropr iate text mater ials, indexes) . ISBN 0 85883 239 9 and ISBN 0 85883 290 9 (Distributed by GeoCenter , Hon igwiesenstrasse 25 , D-7000 Stuttgart 80 , Postfach 80 08 30 , We st Germany . Pr ice DM 250 .-) No .68. STREICHER, J. -F. Jab&m-English dic tionary. 1982; xii+674pp . (Hardbound : incl . 1 map) . ISBN 0 85883 274 7 No .69. CAPELL , A. and H.H.J. COATE Comparative studies in No rthern Kimberley languages. 1984 ; xvi+258pp. (incl. 1 map) . ISBN 0 85883 314 X

No .70. WURM , S.A. and P. Muhlhausler, eds Handboo k of Tok Pisin (New Guinea Pidg in) . Languages for intercultural communication in the Pacific area project of the Australian Academy of the Humanities , publ ication No .1. 1985 ; iv+725pp. (Hardbound) . ISBN 0 85883 321 2 Articles by Tom Dutton ; Susanne Hol zknecht; Don Laycock; P. Muhlhausler ; Julie Piau ; Malcolm Ro ss; Je ff Siegel; S.A. Wurm . No .7l. WORDICK, F.J.F. The Yind j ibarndi language . 1982; xii+390pp. (Hardbound : incl . 1 map) . ISBN 0 85883 265 8 No .72. TRYON, D.T. and B.D. HACKMAN Solomon Islands languages: an internal classification . 1983; viii+490pp. (Hardbound : incl . 5 maps) . ISBN 0 85883 292 5 PACIFIC LINGUISTICS 17

Series C - Books (continued )

No .73. FERRELL , Raleigh Pa iwan dictionary. 1982; x+503pp. (Hardbound) . ISBN 0 85883 264 X No .74. HALIM, Am ran , Lo is CARRINGTON and S.A. WURM , eds Papers from the Third International Conference on Austronesian Linguistics. vol.l: Currents in Oceanic . 1982. vi+314pp. (inc!. 8 maps) . ISBN 0 85883 285 2; 0 85883 271 2 Ar ticles by Tom Dutton , Jacques B.M. Guy , S.P. Harrison , Raymond L. Johnston , Don Laycoc k, John Lynch, Malcolm Ross, D.T. Tryon , D.S. Wa lsh. No .75. HALIM, Amran , Lo is CARRI NGTON and S.A. WURM , eds Papers from the Third International Conference on Austronesian Linguistics. vol .2: Tracking the travellers. 1982; vi+331pp. (inc!. 3 maps) . ISBN 0 85883 285 2; o 85883 275 5 Articles by J.C. Anceaux , James T. Collins , Isidore Dyen , Farid M. Onn , Mark Harvey , Hans Lapoliwa , Paul Jen-kuei Li , Teodore A. Llamzon , J. Noorduyn , Andrew K. Pawley, Lawrence A. Re id , Nicole Revel-Macdonald , Mangantar Simanjuntak , Stanley Starosta , C.L. Voorhoeve , John U. Wol ff, Colin Yallop , R. David Zorc .

No .76. HALIM , Am ran, Lo is CARRINGTON and S.A. WURM , eds Papers from the Third International Conference on Austronesian Linguistics. vol.3: Accent on variety. 1982; vi+324pp . (inc !. 48 maps) . ISBN 0 85883 285 2; 0 85883 281 X Ar ticles by Ro ss Clark, Precy Espiritu-Reid , James J. Fox , C.D. Grijns, Kay Ikranagara, Joseph F. Kess , E.A. Kondrashkina, Threes Y. Kumanireng , Bernd No thofer , D.J. Prentice , J.P. Sarumpaet, Gary F. Simons , Istiati Sutomo , KMA M. Usop , Roland Walker . No .77. HALIM, Amran , Lo is CARRI NGTON and S.A. WURM , eds Papers from the Third International Con ference on Austronesian Linguistics, vol .4: Thematic variation . 1983; viii+415pp. (inc!. 6 maps) . ISBN 0 85883 285 2; o 85883 304 2 Ar ticles by Na talia F. Al ieva , Azhar M. Simin , Wayan Bawa , Joel Bradshaw , Al ice Cartier , Terry Crowley , Soenjono Dardjowidjojo, Raleigh J. Ferrell , R. Hardjadibrata, Marit Kana , Bambang Kaswanti Purwo , Ke ith McCune , Tams in Medan , Claire Moyse-Faurie, I Gusti Ng urah Bagus, Fran�oise Ozanne-Rivierre , Ramelan , Sri Wulan Rujiati Mulyadi , Soed jarwo, Dendy Sugono , Ignatius Suharno , Michael R. Thomas, Henny Lomban Ticoalu, E.M. Uhlenbeck, John W.M. Ve rhaar , Husein Widjajakusumah , Yeoh Chiang Kee . No .78. KING, Julie K. and John Wayne KING, eds Languages of Sabah : a survey report. 1984; vi+359pp . (inc!. 39 maps) . ISBN 0 85883 297 6 No .79. LYNCH , J. ed ., Stud ies in the languages of Erromango . 1983; vii+222pp . (incl 2 maps ) . ISBN 0 85883 284 4 No .80. BENDER, Byron W. , ed . Studies in Micronesian linguistics. 1984; xii+515pp. (incl. 1 map) ISBN 0 85883 307 7 Ar ticles by Byron W. Bender , Jo-Ann Flora, Sheldon P. Harrison , Robert W. Hsu, Frederick H. Jackson , Roderick A. Jacobs , John Thayer Jensen , Lewis S. Josephs , Kee-dong Lee , Ann M. Pe ters, Kenneth Rehg , Ho-min Sohn , Anthony F. Tawerilmang , Jud ith W. Wang .

No .81. BRUCE , Les The Alamblak lan�uage of Papua New Guinea (East Sepik) . 1984; vi+361pp . (Hardbound : �nc! . 2 maps) . ISBN 0 85883 303 4

No .83. MORRIS, Cliff Tetun-English dictionary. 1984 ; xx+194pp. (inc!. 1 map) . ISBN 0 85883 308 5

No .84. BROWN , Herbert A. A comparative dictionary of Orokolo , Gulf of Papua . 1986; xxi+254pp. (inc! . 1 map) . ISBN 0 85883 317 4 PAC IFIC LINGUISTICS 18

Series C - Books (continued) No .85. CHEN, Teresa M. Ve rbal constructions and verbal classification in Nataoran-Amis. 1985; iii+297pp . (incl. 4 maps) . ISBN 0 85883 331 X

No .86. RENSCH , Karl H. Tikisionalio fakauvea-fakafalani -- Dictionnaire wallisien-fran�ais. 1984; Spec ial Number ; xiv+439pp . (incl. 1 map) . ISBN 0 85883 310 7

No .87. THURGOOD , Graham , James A. MATISOFF , David BRADLEY , eds, with the assistance of Grace Sh iang-Jiun Lin and Keith Record. Linguistics of the Sino-Tibetan area: the state of the art. Papers presented to Paul K. Benedict for his 71st birthday. 1985; vii+498pp. ISBN 0 85883 319 0 Articles by William L. Ballard , William H. Baxter III, Alton L. Becker , Nicholas C. Bodman, David Bradley, Chr istopher Court, S¢ren Egerod , William J. Gedney, Inga-Lill Hansson , Mantaro J. Hashimoto , Andr�-Georges Haudricourt, Eug�nie J.A. Henderson , Franklin E. Huffman , F.K. Lehman , Charles N. Li , Lorenz G. Leffler , James A. Matisoff, Martine Mazaudon , Tsu-li Me i, Boyd Micha ilovsky, Yasuhiko Nagano , Ng uy�n Blnh-hoa , Tatsuo Nishida , Jerry No rman , A. Schussler , R.K. Spr igg , Sandra A. Thompson , Graham Thurgood , Jul ian K. Wheatley, Paul Fu-mien Yang .

No .88. PAWLEY , Andrew and Lois CARRINGTON, eds Austronesian linguistics at the 15th Pacific Sc ience Congress. 1985 ; iv+353pp. (inc l 2 maps) . ISBN 0 85883 329 8 Ar ticles by Byron W. Bender , Joel Bradshaw , Ann Chowning , Ross Clark, George W. Grace , Rob in Hoope r, Robert Hsu, Paul Jen-kuei Li , Frantisek Lichtenberk, Lamont Lindstrom , John Lynch, Stanley Starosta , D.T. Tryon , D.S. Walsh , Jud ith W. Wang , Jack H. Ward .

No .89. HARRIS, J.W. Northern Territory pidg ins and the origin of Kriol . 1986; vii+418pp. (incl. 8 maps) . ISBN 0 85883 334 4

No .90. RENSCH , Karl H. ed . Tikisionalio fakafutuna-fakafalani -- Dictionnaire futunien-fran9ais. 1986 ; Spec ial Number ; xi+327pp . (incl. 1 map) . ISBN 0 85883 348 4

No .91. CROWLEY , Susan Smith Tolo dictionary. 1986 ; xii+107pp . ISBN 0 85883 346 8

No .92. CLYNE , Michael , ed ., Australia, meeting place of languages. 1985 ; v+328pp. (incl. 3 maps) . ISBN 0 85883 330 1 Ar ticles by Ed ith Bavin, Camilla Bettoni , Maya and David Bradley, Michael Clyne , John Harris, Malcolm Johnston , Susan Kaldor , Manfred Klarberg , Patr ick McConvell, Ian G. Malcolm , Peter Muhlhausler , J.V. Neustupny , Uldis Ozolins , Anne Pauwels, Marta Rado , John Sandefur , Anne tte Schmidt, M.J. Secombe , Anna Shnukal , Tim Shopen, J.J. Smolicz.

No .93. GERAGHTY , Paul , Lo is CARRI NGTON and S.A. WURM, eds FOCAL I: papers from the Fourth International Conference on Austronesian Linguistics. 1986 ; x+403pp . (incl 1 map , 3 photographs) . ISBN 0 85883 343 3; o 85883 344 1 Articles by David G. Arms , Niko Be snier , Susanna Cumming , Otto Chr . Dahl , Videa P. De Guzman , Carol Georgopoulos , Vo lker Graefe , Ray Harlow , Robin Hooper, Bambang Kaswanti Purwo , Paul R. Kroeger , Juliette Levin, Diane Massam , G.B. Milner, Susan Mo rdechay , Paz Buenaventura Naylor , Stanley Starosta , He in Ste inhauer , Geraldine Tr iffitt, Fay Wouk , Ayako Yasuda-Graefe , R. David Zorc . PACIFIC LINGUISTICS 19

Series C - Books (continued)

No.94. GERAGHTY , Paul, Lois CARRINGTON and S.A. WURM , eds FOCAL II: papers from the Fourth International Conference on Austronesian Linguistics . 1986; vi+463pp. (incl. 15 maps , 4 photographs) . ISBN 0 85883 343 3; o 85883 345 X Articles by Ann Chowning , Ross Clark, James T. Collins, Tom Dutton , Isidore Dyen , Paul Geraghty , George W. Grace , S.P. Harr ison , Even Hovdhaugen, Frederick H. Jackson , Paul Jen-kuei Li, John Lynch , Bernd Nothofer, A.K. Ogloblin, Malcolm ROSS, J.P. Sarumpaet, Jeff Siegel, Geoff P. Smith, J.N. Sneddon , D.T. Tryon , Shigeru Tsuchida , R. David Zorc.

No. 95. LINDSTROM , Lamont Kwamera dictionary -Nikukua Sai Nagkiari ien Nininife. 1986 ; ix+195pp . (incl. 1 map) . ISBN 0 85883 340 9

IN PREPARATION :

LAYCOCK , D.C. Basic ma terials in Buin: grammar , texts and dictionary .

WURM , S.A. , John G. MEALUE and John Ini LAPLI and Frank Bollen Yopuse Northern Santa Cruzan dictionary, Solomon Islands.

WURM S.A. , Patrick BWAKOLO and Martin MOY IYO Ayiwo dictionary, Reef Islands, Solomon Islands .

LEE , Jennifer R. Tiwi today : a study of language change in a contact situation .

HODDINOTT , W.G. and F.M. KOFOD The Ngankikurungkurr language (Daly River area, Nor thern Te rritory) . PACIFIC LINGUISTICS 20

SERIES D - SPECIAL PUBLICATIONS

(BULLETINS , ARCHIVAL MATERIALS AND OTHER PUBLICATIONS)

No. 1. Bulletin No.1. 1964; 9pp . ISBN 0 85883 072 8 Contributions by A. Capell, R.D. Eagleson , E.H. Flint, Susan Kaldor , J. Lyons, S.A. Wurm. No. 2. Bulletin No .2. 1965; v+84pp . ISBN 0 85883 073 6 Contributions by J.R. Bernard, H. Bluhme , Christopher A.F. Court, Robert R. Dyer, E.H. Flint, F.W. Harwood , Susan Kaldor , E.M. Ligg ins, A. Mu r tonen , Anita Pincas , Hans Pollak , C.W. Ruhle , R.J. Zatorski. No . 3. WURM, S.A. New Guinea Highlands Pidgin: course materials. 1971; vii+175pp . ISBN 0 85883 074 4 No. 4. WURM , S.A. Languages : Eastern, we stern and Southern Highlands, Territory of Papua & New Guinea. (Map in four teen colours.) 1961. ISBN 0 85883 075 2 No . 5. LAYCOCK , Don Materials in New Guinea Pidgin (Coastal and Lowlands) . 1970; xxxvii+62pp . Repr inted 1974. ISBN 0 85883 076 0 No . 6. NGUYEN �ANG LIEM Four-syllable idiomatic expressions in Vietnamese . 1970; v+60pp . ISBN 0 85883 077 9 No. 7. ELBERT , S.H. Three legends of Puluwat and a bit of talk . 1971; ix+85pp. (incl. 1 map , 1 photograph) . ISBN 0 85883 078 7 No. 8. LANG , Adrianne , Katherine E.W. MATHER and Mary L. ROSE Information storage and retrieval : a dictionary project . 1972; vii+151pp. ISBN 0 85883 087 6 No. 9. PACIFIC LINGUISTICS Index to Pacific Linguistics, Series A-D, as at the end of 1970. 1971; iv+75pp . ISBN 0 85883 079 5 No. 10 . PATON , W.F. Tales of Ambrym. 1971; xii+8 2pp . (incl . 1 map) . Reprinted 1978. ISBN 0 85883 080 9 No. 11. WURM , S.A. , ed ., with P. BRENNAN , R. BROWN , G. BUNN , K. FRANKLIN, B. IRWIN, G. SCOTT , A. STUCKY , and other members of the Summe r Institute of Linguistics, New Guinea Branch Language maps of the Highlands Provinces , Papua New Gu inea . 1978; iii+21pp . (incl . 6 maps) . ISBN 0 85883 171 6 No. 12. DUTTON , T.E. Conversational New Guinea Pidgin. 1973; xv iii+292pp . Reprinted 1974, 1977, 1979, 1981. ISBN 0 85883 096 5 Rev ised course available . See 067. No .13. GLOVER , Jessie R. and Deu Bahadur GURUNG Conversational Gurung . 1979; vii+216pp . ISBN 0 85883 192 9 No. 14 . BARNETT , Gary L. Handbook for the collection of fish names in Pacific languages . 1978; v+10lpp . (incl. 1 map , 47 photographs, 3 drawings) . ISBN 0 85883 175 9

No.15. TRYON , D.T. & R. GELY , eds Gazetteer of New Hebrides place names / Nomenclature des noms geographiques des Nouvelles-Hebrides. 1979; xxxiii+ 155pp . (incl. 1 map) . ISBN 0 85883 194 5 No .16. YOUNG , Ma ribelle Bwaidoka tales. 1979; viii+136pp. (incl. 1 map) . ISBN 0 85883 200 3 No. 17. STOKHOF , W.A.L. , ed . ...wit h Lia SALEH-BRONKHORST Holle lists : vocabularies in languages of Indonesia, vol.l: Introductory volume . (Materials in languages of Indonesia, No. 1: W.A.L. Stokhof , Series ed .) 1980 ; v+149pp . (incl. 2 maps) . Reprinted 1981. ISBN 0 85883 213 5 and ISBN 0 85883 214 3 PACIFIC LINGUISTICS 21

Series D - Spec ial Publications (continued)

No .18. STANHOPE , John M. The language of the Rao people, Grengabu , Madang Province , Papua New Guinea. 1980; vii+28pp . (incl . 3 maps , 5 photographs) . ISBN 0 85883 222 4

No .19. STOKHOF , W.A.L. Woisika I: an ethnogr aph ic introduction . 1977: ix+74pp. (incl . 3 maps) . Reprinted 1980 . ISBN 0 85883 167 8

No . 20. CAPELL , A. and J. LAYARD Ma terials in Atchin, Malekula : grammar, vocabulary and texts. 1980 ; v+260pp . (incl . 1 map) . ISBN 0 85883 231 3

No .2l. SCHEBECK , B. Texts on the soc ial system of the AtYnYamatana people , with grammat ical notes . 1974; xviii+278pp . + 1 photog raph: ISBN 0 85883 102 3

No . 22. PATON , W.F. Customs of Ambrym (texts, songs, games and drawings). 1979; xv+98pp . (incl . 1 map , 4 photographs). ISBN 0 85883 189 9

No . 23 . CLYNE , Michael, ed . Australia talks: essays on the soc iology of Australian immigrant and aboriginal languages. 1976; viii+244pp . Repr inted 1978, 1980 , 1983. ISBN 0 85883 148 1 Ar ticles by M. Anne Bolitho , Michael Clyne , Robert D. Eagleson , R. McL . Harris, Ruth Johnston , Susan Kaldor , Manfred Klarberg, Stephen Muecke , Ma rta Rado , John Sandefur, Margeret C. Sharpe , J.J. Smol icz , Bruce A. Sommer , Br ian A. Taylor , Elizabeth Thuan , Darrell T. Tryon .

No .24. DUTTON , T.E. and C.L. VOORHOEVE Beg inning Hiri Mo tu . 1974; xv ii+259pp . Set of 6 cassettes (optional) . Reprinted 1975. ISBN 0 85883 112 0

No .25. Z'GRAGGEN , John A. The languages of the Madang Distr ict, Papua New Gu inea . (Map) 1973. ISBN 0 85883 105 8

No . 26 . LAYCOCK , D. Languages of the Sepik Region , Papua New Guinea . (Map) 1975. ISBN 0 85883 136 8

No .27. WURM , S.A. Spread ing of languages in the South-western Pacific. (Map) 1975. ISBN 0 85883 127 9

No. 28. STOKHOF , W.A.L. , ed ., ...wit h Lia SALEH-BRONKHORST Holle lists : vocabularies in languages of Indonesia, vol. 2: Sula and Bacan Islands , North Halmahera, South and East Halmahera . (Mater ials in languages of Indonesia, No . 2: W.A.L. Stokhof, Ser ies ed .) 1980 ; iv+325pp . ISBN 0 85883 213 5 and ISBN 0 85883 218 6

No . 29 . DUTTON , Tom Queensland Canefields Eng lish of the late nineteenth century (a record of interview with two of the last surviv ing Kanakas in Nor th Queensland , 1964) . 1980; xiii+147pp . (incl . 3 maps , 2 photographs) . ISBN 0 85883 224 0

No . 30 . Z'GRAGGEN , J.A. A compa rative word list of the Ra i Coast languages, Madang Province , Papua New Guinea . 1980 ; xv+18lpp . (incl. 1 map) . ISBN 0 85883 232 1

No .3l. Z'GRAGGEN , J.A. A compa rative word list of the Nor thern Adelbert Range languages, Madang Province , Papua New Guinea. 1980; xvi+178pp . (incl . 1 map) . ISBN 0 85883 228 3

No. 32. Z'GRAGGEN , J.A. A compa rative word list of the Mabuso languages, Madang Province , Papua New Guinea. 1980; xv+ 184pp . (incl. 1 map) . ISBN 0 85883 233 X No . 33. Z'GRAGGEN , J.A. A compa rative word list of the Southern Adelbert Range languages, Madang Province , Papua New Gu inea . 1980; xvi+97pp . (incl . 1 map) . ISBN 0 85883 234 8 PACIFIC LINGUISTICS 22

Series D - Spec ial Publ ications (continued)

No . 34 . LAPOLIWA , Hans A generative approach to the phonology of bahasa Indonesia. (Mater ials in languages of Indonesia, No .3: W.A.L. Stokhof , Ser ies ed .) 1981; v+ 155pp . ISBN 0 85883 245 3

No .35. STOKHOF, W.A.L. , ed . , ...wit h Lia SALEH-BRONKHORST and Alma E. ALMANAR Holle lists : vocabular ies in languages of Indonesia, vol .3/1 : Southern Moluccas; Central Moluccas: Seram (1) . (Mate rials in languages of Indonesia, No . 4: W.A.L. Stokhof , Series ed. ) 1981; iv+201pp . (inc1. 1 map) . ISBN 0 85883 213 5; 0 85883 242 9; 0 85883 243 7.

No . 36 . HALIM, Amran Intonation in relation to syntax in Indonesian. (Mater ials in languages of Indonesia, No . 5: W.A.L. Stokhof , Series ed .) 1981; vii+149pp. ISBN 0 85883 246 1

No .37. NABABAN , P.W.J. A gramma r of Toba-Batak . (Mate rials in languages of Indonesia, No. 6: W.A.L. Stokhof , Ser ies ed .) 1981; xxiv+146pp . (inc1. 2 maps) . ISBN 0 85883 254 2

No. 38 . POEDJOSOEDARMO , Soepomo Javanese influence on Indonesian. (Mater ials in languages of Indonesia, No. 7: W.A.L. Stokhof , Ser ies ed .) 1982; viii+187pp. (inc 1. 1 map) . ISBN 0 85883 261 5

No .39. KARTOMIHARDJO, Soeseno Ethnography of commun icat ive codes in East Java. (Materials in languages of Indonesia, No . 8: W.A.L. Stokhof , Ser ies ed .) 1981; xi+212pp . (inc 1. 1 map) . ISBN 0 85883 255 0

No . 40. CARRINGTON , Lo is and Miriam CURNOW Twenty years of Pacific Linguistics: an index of contr ibutions to Pacific lingu istic studies 1961-1981. 1981; vi+161pp . ISBN 0 85883 249 6

No . 41. STOKHOF , W.A.L. Wo isika riddles. (Materials in languages of Indonesia, No .9: W.A.L. Stokhof , Ser ies ed .) 1982; iii+74pp. (inc 1. 2 maps) . ISBN 0 85883 257 7

No.42. McGREGOR , Donald E. and Aileen McGREGOR 010 language materials . 1982; viii+155pp . ISBN 0 85883 262 3

No. 43. VERHEIJEN . J.A.J. Dictionary of Manggarai plant names. 1982; iii+140pp . (inc 1. 1 map) . ISBN 0 85883 273 9

No . 44. STOKHOF , W.A.L. , ed . , ..•wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists: vocabularies in languages of Indonesia, vol .3/2: Central Moluccas : Seram (II) . (Materials in languages of Indonesia, No .lO: W.A.L. Stokhof , Ser ies ed .) 1981; iv+ 207pp. ISBN 0 85883 213 5; 0 85883 242 9; 0 85883 253 4

No . 45. SUHARNO , Ignatius A descr iptive study of Javanese . (Mate rials in languages of Indonesia, No .ll: W.A.L. Stokhof, Ser ies ed .) 1982; xiv+ 175pp . ISBN 0 85883 258 5

No . 46. VOORHOEVE , C.L. , ed. The Ma kian languages and the ir ne ighbours. (Mater ials in languages of Indonesia, No. 12 : W.A.L. Stokhof , Series ed .) 1982; viii+l48pp . (inc1 . 4 maps) . ISBN 0 85883 277 1 Articles by C.L. Voorhoeve ; J.T. Collins (2) ; D. Te ljeur . No . 47. COLLINS , James T. The histor ical relationships of the languages of Central Maluku , Indonesia. (Materials in languages of Indonesia, No .13: W.A.L. Stokhof , Series ed. ) 1983; ix+ 168pp . (incl. 10 maps) . ISBN 0 85883 289 5 No .48. TAMPUBOLON , Daulat Purnama Ve rbal affixat ions in Indonesian : a semantic exploration. (Mate rials in languages of Indonesia, No. 14 : W.A.L. Stokhof , Ser ies ed .) 1983; v+156pp . (inc1 . 1 map) . ISBN 0 85883 280 1 PACIFIC LINGUISTICS 23

Series D - Spec ial Publ icat ions (continued)

No .49. STOKHOF , W.A.L. , ed ., ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists: vocabularies in languages of Indone sia, vol .3/3 : Central Moluccas : Seram (III) ; Haruku; Banda ; Ambon (I) . (Materials in languages of Indonesia No. 15 : W.A.L. Stokhof , Series ed .) 1982; vi+214pp . ISBN 0 85883 213 5; 0 85883 242 9; 0 85883 288 7

No . 50 . STOKHOF , W.A.L. , ed . , ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists : vocabular ies in languages of Indonesia, vol .3/4 : Central Moluccas : Ambon (II) ; Buru ; Nusa Laut; Saparua . (Materials in languages of Indonesia No .16: W.A.L. Stokhof , Series ed .) 1982; iv+179pp . ISBN 0 85883 213 5; 0 85883 242 9; 0 85883 267 4 No . 51 . STOKHOF , W.A.L. , ed . , ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists : vocabularies in languages of Indone sia, vol.4: Talaud and Sang ir Islands . (Mate rials in languages of Indonesia No .17: W.A.L. Stokhof , Series ed .) 1982; iv+313pp . (incl . 1 map) . ISBN 0 85883 213 5; 0 85883 268 2

No . 52 . STOKHOF , W.A.L. , ed ., ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists: vocabular ies in languages of Indonesia, vol .5/1 : Irian Jaya : Austronesian languages; Papuan languages , Digul area . (Materials in languages of Indonesia No . 18 : W.A.L. Stokhof , Series ed .) 1982; iv+186pp . (inc l. 2 maps) . ISBN 0 85883 213 5; 0 85883 269 0; o 85883 293 3. No .53. STOKHOF , W.A.L. , ed . , ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists : vocabular ies in languages of Indonesia, vol .5/2 : Irian Jaya : Papuan languages, Nor thern languages, Central Highlands languages. (Materials in languages of Indonesia No . 19 : W.A.L. Stokhof , Ser ies ed .) 1983; iv+245pp . (incl . 2 maps) . ISBN 0 85883 213 5; 0 85883 269 0; o 85883 294 1 No .54. HOLMER , Nils M. Lingu istic survey of south-eastern Queensland . 1983; vii+485pp . ISBN 0 85883 295 X No . 55. DJAWANAI, Stephanus Ngadha text tradition : the collective mind of the Ngadha people , Flores . (Materials in languages of Indonesia No .20: W.A.L. Stokhof , Series ed .) 1983; vii+278pp . (incl. 1 map) . ISBN 0 85883 283 6 No . 56 . AJAMISEBA , Danielo C. A classical Malay text grammar : insights into a non-western text tradition. (Materials in languages of Indonesia No . 21 : W.A.L. Stokhof , Series ed .) 1983; v+116pp . ISBN 0 85883 286 0

No . 57 . McELHANON , K.A. A lingu istic field guide to the Morobe Province , Papua New Guinea. 1984 ; iii+67pp . (incl . 1 map) . ISBN 0 85883 287 9

No .58. JASPAN , M.A. Ma terials for a Re jang-I ndones ian-English dictionary , with a fragmentary sketch of the Rej ang language by W. Aichele, and a preface and add itional annotations by P. Voorhoeve . (Materials in languages of Indonesia No . 27 : W.A.L. Stokhof , Series ed .) 1984 ; x+ 162pp . ISBN 0 85883 312 3

No .59. STOKHOF , W.A.L. , ed ., ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists: vocabular ies in languages of Indonesia, vol.6: The Lesser Sunda Islands (Nusa Tenggara) . (Materials in languages of Indonesia No . 22: W.A.L. Stokhof , Ser ies ed .) 1983; iv+337pp . (incl . 2 maps) . ISBN 0 85883 213 5; 0 85883 302 6 PACIFIC LINGUISTICS 24

Series 0 - Spec ial publ ications (continued)

No .60. STOKHOF , W.A.L. , ed . , ...wit h Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists : vocabularies in languages of Indonesia, vol.7/2 : North Sulawesi : Philippine languages . (Mater ials in languages of Indonesia No.23: W.A.L. Stokhof , Ser ies ed .) 1983; v+328pp . (incl . 1 map) . ISBN 0 85883 213 5; 0 85883 298 4; 0 85883 299 2

No. 61. STOKHOF , W.A.L. , ed . , ... with Lia SALEH-BRONCKHORST and Alma E. ALMANAR Holle lists: vocabular ies in languages of Indonesia, vol 7/1 : North Sulawesi : Gorontalo group and Tontoli. (Mater ials in languages of Indonesia No .24: W.A.L. Stokhof , Ser ies ed .) 1984 ; iii+240pp . (inc l. 2 maps) . ISBN 0 85883 213 5; 0 85883 298 4; 0 85883 300 X No .62. STOKHOF , W.A.L. , ed . , with Alma E. ALMANAR Holle lists : vocabular ies in languages of Indonesia, vol.7/3 : Ce ntral Sulawesi, South-west Sulawesi. (Materials in languages of Indonesia No . 25 : W.A.L. Stokhof , Series ed .) 1984 ; iv+251pp. (inc!. 3 maps) . ISBN 0 85883 213 5; 0 85883 298 4; o 85883 301 8 No .64. GROVES , Terab 'ata R. , Gordon W. GROVES and Roderick JACOBS Kiribatese : an outline descr ipt ion . 1985; vi+155pp . ISBN 0 85883 318 2

No. 65. HARDJADIBRATA , R.R. Sundanese : a syntactical analysis. (Mater ials in languages of Indonesia No .29: W.A.L. Stokhof , Series ed .) 1985; vi+159pp . (inc!. 1 map) . ISBN 0 85883 320 4

No . 66. STOKHOF , W.A.L. , ed . , ... with Alma E. ALMANAR Holle lists : vocabular ies in languages of Indonesia, vol .7/4 : South-East Sulawesi and ne ighbou ring islands , west and North-East Sulawesi. (Mater ials in languages of Indonesia No .28: W.A.L. Stokhof , Ser ies ed .) 1985; iv+290pp. (incl 4 maps) ISBN 0 85883 213 5; 0 85883 298 4; 0 85883 328 X No .67. DUTTON , Tom , in collaboration with Dicks THOMAS A new cou rse in Tok Pisin (New Guinea Pidg in) 1985; xxviii+407pp . (incl . 3 maps , 110 photographs) . with a set of 14 cassettes (optional ). ISBN 0 85883 341 7

No. 68. MOELIONO Anton Language development and cultivation : alternative approaches in language planning . (Materials in languages of Indonesia, No .30: W.A.L. Stokhof , Series ed .) 1986 : ix+135pp . ISBN 0 85883 332 8

No . 69. STOKHOF , W.A.L. , ed . , ... with Alma E. ALMANAR Holle lists: vocabular ies in languages of Indonesia, vol.8: Kalimantan (Borneo) . (Materials in languages of Indonesia No. 31: W.A. L. Stokhof , Series ed .) 1986; iv+207pp . ISBN 0 85883 213 5; 0 85883 333 6

No . 70 . VERHE IJEN , Jilis A.J. , SVD The Sama/Bajau language in the Lesser Sunda Islands . (Mater ials in languages of Indonesia No . 32: W.A.L. Stokhof , Ser ies ed .) 1986; viii+209pp . (inc! . 1 map) . ISBN 0 85883 335 2

No. 71 . STOKHOF , W.A.L. , ed ., ... with Alma E. ALMANAR Holle lists : vocabular ies in languages of Indonesia, vol.9: Nor thern Sumatra. (Materials in languages of Indonesia No . 33, W.A.L. Stokhof , Se ries ed .) 1985; iv+315pp . (incl 4 maps) . ISBN 0 85883 213 5; 0 85883 336 0

IN PREPARATION :

OETOMO , Dede The Chinese of Pasuruan : their language and identity. (Materials in languages of Indonesia No . 26 .)

HUSEN ABAS Indones ian as a unifying language of wider communication .

TRYON , Darrel T. Let 's talk Bislama

STOKHOF , W.A.L. , ed ., ... with Alma E. ALMANAR . Holle lists : vocabular ies in languages of Indonesia, vol.lO.

Chen, T.M. Verbal constructions and verbal classification in Nataoran-Amis. C-85, iv + 300 pages. Pacific Linguistics, The Australian National University, 1967. DOI:10.15144/PL-C85.1 ©1967 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative.