Causative Verbs in Formosan Languages. Preliminary Version. Working Papers in Linguistics, Vol. 5, No. 9. INSTITUTION Hawaii Univ., Honolulu
Total Page:16
File Type:pdf, Size:1020Kb
DOCUMENT RESUME ED 095 711 FL 006 182 AUTHOR Starosta, Stanley TITLE Causative Verbs in Formosan Languages. Preliminary Version. Working Papers in Linguistics, Vol. 5, No. 9. INSTITUTION Hawaii Univ., Honolulu. Dept. of Linguistics. PUB DATE Oct 73 NOTE 70p.; Paper presented at the International Conference on Comparative Austronesian Linguistics (1st, Honolulu, Hawaii, January 1974) EDRS PRICE HF-$0.75 HC-$3.15 PLUS POSTAGE DESCRIPTORS *Descriptive Linguistics; Generative Grammar; Grammar; Linguistics; *Malayo Polynesian Languages; Sentence Structure; Syntax; *Verbs IDENTIFIERS Amis; Bunun; *Formosan Languages; Rukai; Saisiyat; Seedig; Taiwan; Tsou ABSTRACT This research is the result of 10 weeks of field work in Taiwan during the summer of 1972. It consists of a description, analysis, and comparison of the morphologically marked causative verbs in Rukai, Bunrn, Tsou, Anis, Seedig, and Saisiyat. The theoretical framework employed is a type of case grammar referred to as "lexicaseIN a generative but nontransformational approach to syntax. The different approaches taken by several linguists to the problem of causation constructions are described, and then each of the six languages is considered independently. It is concluded that there is one typological characteristic common to all the languages studied, the clear and fundamental division of verbs into two classrs, active and passive. (PP) -89- CAUSATIVE VERBS IN FOALIOSAN LANGUAGES BESTCOPYAVAILABLE PRELI4INARY VERSION To be presented at the First International Conference on Comparative Lc% Austronesian Linguistics, January 2-7, 1974, Honolulu, Hawaii. ON O CONTENTS O 1.0 Introduction....89 1.4.1 2.0Amis 102 3.0 Bunun 113 4.0 Rukai 117 U 5 DEPARTMENT OF HEALTH. 5.0 Saisiyat 121 EDUCATION & WELFARE NATIONAL INSTITUTE OF 6.0 Seediq 128 EDV:ATION THIS DOCUMENT HAS BEEN REPRO 7.0 Tsou .138 DUCED EXACTLY AS RECEIVED FROM THE PERSON OR ORGANI/ATIONORIGIN 8.0 Conclusion 150 ACING IT POINTS OF VIEW OR OPINIONS STATED DO NOT NECESSARILY REPRE 9.0References 152 SENT OFFICIAL NATIONAL INSTITUTE OF EDUCATION POSITION OR POLi-Y Stanley Starosta University of Hawaii University of Edinburgh 1.0 Introduction 1.1 This paper is based mainly on ten weeks of field work which I conducted in Taiwan during the summer of 1972. I was supported by Grant No. CS-3334 from the U.S. National ScienceiFoundation while I was a Visiting Research Fellow of the Institute of History andPhilo- logy of the Academia Sinica. It forms part of a broader survey of the syntax of all the Formosan languages which I began in 1964. The languages I studied directly were Aukai (twenty-five hours of elicitation), Bunun (line hours elicitation), Tsou (about fifteen hours), Amis (fourteen hours), Seediq (twenty-five hours), and Saisiyat (seventeen and a' half hours). I conducted all of my elicitation directly in andarin Chinese, except in Lhe case of my third Saisiyat informant, with whom I had to use a Chinese-Atayal interpreter. I have also refer- red to works on Anis (Ferrell 1971W1971b), Atqyal (Egerod 1966), Bunun (Jeng 1969), Paiwan (Ferrell 1971a), Poyoma (Sprenger 1972), Rukai (Li c) 1973), Saisiyat (Tsuchida 1964), Seediq (Asai 1953), and Tsou (Tung 60 1964). Because of wide variations in outside references and time avail- able, and in the abilities of my informants, the quality of my data ranges from excellent (Amis, Tsou, Rukai) to minimal (Bunun). I have used a broad phonetic transciption for my examples, except those in the section on Tsou. Since I have not done a phonolo- gical analysis of any of these languages, I felt the material would be of more use to phonologists if I reported them essentially as I recorded th,,im, without attempting to reconcile inconsistencies or variations in vowel length and quality, consonant retroflection, etc. 1.2 In this paper, I attempt to describe, analyse, and compare LL the morphologicvliy marked causative verbs in the six languages I studied directly. I have tried to test the validity of Stevens' -90- mow claim (1973) about the general form ofindirect action causatives in Austronesian against data from Formosan andto present evidence to sup- port my claims that the formation of causativesand passives must be treated as lexical derivation rather thaninflection. 1.3 The theoretical framework employedis a type of casegrammar which may for the sake of convenience bereferred to as 'lexicase' (Taylor 1971:8). Briefly, lexicase is ageneratdveLbutnnon4transforma- tional approach to syntax. It has no distinct deep structure andno transformations, and instead relieson Phrase structure Rules (PSR), and Subcategorization (SR) and Redundancy (RR)Rules. It captures the rela- tionship between sentences bymeans of Derivation Rules (DR), rules that formally state the analogicalpattern on the basis of whichone set of lexical items may be derivedfrom another set. Case phenomena are described in terms of featureson lexical items: intrinsic features of case form and case relation on nouns, determiners,and prepositions, and contextual case frame featureson verbs and prepositions. It will be these case features and DerivationRules that will be chiefly involved in the analysis of Formosan causativesin this paper. (For other studies in the lexicase framework,see the items marked by an asterisk in the References). In syntactic terms, a causative constructionis one in which an extra agent or force actant is allowed tooccur with a verb in addition to the actants already lexically permittedby the verb's case frame. Semantically, this external agentor force is seen as causing the action, process, or state characterized by the verb. All human languagespossess such constructions, and all must choosefrom two possible types: syntac- tic (or 'paraphrastic') and morphological. A syntactic causativeconstruc- tion is one in which a usually non-finite(subject less) verb withor with- out other accompanying actants is embedded undera verb of causation. The causation verb hasan agent or instrument actant, the 'causer', and an object (often optional) acted upon by thecauser. This object is interpreted as coreferential with the missingsubject of the embedded sentence. For example in English: R S ir .% /' NNv Det N NP I I John i made the desert 1 bloIom (+fin] It_ .1 (-fin] 1-2. NP PP Det/#/\ UP Ni V 1 at...z:tieu:: causeld to John i believe in r ^ f in 1% f in I. .1 Y Mal -91- Regardless of what one Tay believe about deeigalanUAVATARfplins, etc., in the surface structure it is clear that the second NP in both structures is the object of the matrix verb rather than the sub- ject of the embedded verb, as can be seen from the correspondingpas- sives: 1-3. The desert was made to bloom by John. 1-4. *The desert bloom was made by John. 1-5. John was caused to believe in gremlins by the accident. 1-6. *John to believe in gremlins was caused by the accident. The actual subject of the embedded sentence, then, does not normally appear overtly in a syntactic causative construction, and the fact that it is :oeferential with the matrix object must be formally characte- rized in a grammar. A morphological causative construction is one containing a morphologically marked causative verb. Such a verb differs syntacti- cally from the corresponding non-causative verb in allowingan addi- tional 'causer' actant on its case frame, and morphologically in having some overt marking which distinguishes it from the non-causative counterpart. Highly inflected languages such as Japanese (Taylor 1971;228-232 ) and Korean (Yang 1972:202 217) often have morphological causatives, while isolating languages such as Thai (Kullavanijaya 1974) of course must wake do with syntactic causatives. Nany other languages have both types. 1.4 As Alan Stevens has noted, a large number of Austronesian languages have morphological causative verbs marked bya prefix pp -, and these verbs 'have to handle an added UP in terms of surfacecases and in the focus morphology of the verb (Stevens 1973). Pacausatives also occur in all the Formosan aboriginal lan,juages I have studied, In this paper, I will describe these constructions in six languages: Amis, Bunun, Rukai, Saisiyat, Seediq, and Tsou. I trill be especially concerned with explainin3 and formalizing the grammaticalprocesses of causativization in "indirect action causatives" (Stevens 1973), that is, in the formation of pp-causatives from transitive verbs. Attention will be focussed on how these languages cope with the problem ofaccom- inodating a new case relation in the case frame of a verb whichmay already be hard put to signal the case relations it already takes, with the possibly limited set of case forms the language makes available to it. This problem is especially severe in the case of ditraisitive verbs, and the way each of the languages handles this problem isespe- cially instructive. The different approaches taken by languages and linguists to the problem of indirect causation constructions differ in how they treat the case relation with corresponds to the unmarked subject of the non-causative counterpart (agent, dative, or objective for our purposes), and the new added 'causer' relation. Several approaches are possible. The easy way out from the point of view of the language is of course the syntactic causative construction. It will probably bbd always prove possible to embed a sentence under a higher verb ofcausa- tion, as previously described, instead of working out a means of accomo- dating an extra case relation in the case frame. This is a common enpe- dien in several of the languages studied, and evensome of the hardiest Formosan languages resort to this device when faced with the prospect of causativizing a ditransitivc verb. In these constructions, then, the new agent comes in as an actant (usually the subject>ef the matrix verb of causation, and the original agent of thenon causative cannot be overtly expressed at all. Instead, by a universal convention on the interpretation of syntactic causativeeconstructions (Aullevanijaya 1974), the missing subject of the embedded non-causative verb is identi- fied as coreierential with tue object of the matrix causative verb.