論雅美語動詞詞綴的功能 - 政大學術集成

全文

(1)國立政治大學語言學研究所博士論文 Graduate Institute of Linguistics National Chengchi University Doctoral Dissertation. 政治大論雅美語動詞詞綴的功能立. Functions of Yami Verbal Affixes. ‧. ‧ 國. 學 er. io. sit. y. Nat. n. 指導教授：張郇慧博士 a iv l CDr. Claire Hsun-Huei Chang Advisor: n hengchi U. 黃婉婷 Stacy Wan-Tin Huang. 中華民國. 106 年 April, 2017. 4. 月.

(2) Functions of Yami Verbal Affixes Abstract. Austronesian languages have exhibited a unique subject-verb relation that has been extensively studied beginning from last century. This relation has been referred to as focus and later on the term voice has replaced it due to its pragmatic usage. Austronesian focus or voice refers to verbal affixation that signals the thematic role of the clause subject. However, this subject-verb relation is not as regular and predictable as the Englsih active-passive voice system. Hence, an alternative view has proposed that the function of these verbal affixes is to transitivize/intransitivize clauses—transitivity analysis. The present study proposes that the. 政治大. function or functions of these verbal affixes should not be limited to signal the thematic role of a clause or to transitivize/intransitivize clauses, and suggests that the verbal affixes in. 立. Yami are multi-functional. This proposal is further supported by the affixes ma- and -an in. ‧ 國. 學. Yami. Semantic properties of subject participants of clauses containing Yami verbal affixes including Ø, m-, mi-, <om>/om-, ma-, man-/mang-, ni-, -en, -an, i-, and ka- -an are examined. ‧. via the modified version of Foley’s (2005) revised macro-role hierarchy. Based on the semantic property of a clause subject, it is attested that the affix -an serves a similar function. sit. y. Nat. as the affix -en in Yami to construct clauses with an undergoer subject in addition to its original function of constructing clauses with a locative subject. There are at least three. io. n. al. er. different sets of roots/stems that are observed to interact with affixes -an and -en in. i n U. v. constructing undergoer subject clauses. One set can only be affixed with -en, one only with. Ch. engchi. -an, and the third one can be affixed with both affixes. This has implied that the lexical property of roots/stems has some influence on the choice of the verbal affixation. The semantic property of -an and -en clause subject has some subtle meaning difference when they are compared via the modified version of Foley’s (2005) revised macro-role hierarchy. The -an clause subject is causally affected and might not undergo change of state while the -en clause subject undergoes change of state.. Four sub-types of ma- clauses are identified including stative ma-, resultative ma-, potentive ma-, and activity ma-. The set of roots/stems that can only be affixed with the affix -en to construct undergoer subject clauses can be affixed with resultative ma- while the set that can only be affixed with -an cannot. The semantic properties of -an and -en clause subjects differ in affectedness have been further confirmed. The potentive mai.

(3) simultaneously serves two functions that include signaling an undergoer subject and potentiality of the event, whereas the affix -an serves two different voice functions— undergoer or locative voice. In addition to the view of multifunctionality of these verbal affixes, it is suggested that the study of the Yami verbal system has to take the lexical property of roots/stems into consideration.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. ii. i n U. v.

(4) 論雅美語動詞詞綴的功能探討摘要. 許多文獻不斷的在探討南島語言動詞詞綴的變化與句子主語的關係，動詞詞綴與主語之間的關係早期被稱為焦點系統（focus system），後來被修正為語態系統（voice system）。但南島語的語態系統與英文的主動/被動語態相似卻不雷同，南島語有主事者語態（actor voice）、受事者語態（patient voice）、地點語態（locative voice）、受惠者語態（beneficiary voice）、工具語態（instrumental voice）等等。除此之外，動詞詞綴與主語的論旨角色之間的對應並不像英文主語/被動語態般有規律，且. 政治大. 可被預測。因此，有看法認為動詞詞綴的主要功能是增加或減少句子的及物性. 立. （transitivity analysis）。雅美語是屬於南島語系的台灣原住民語言，擁有豐富的動詞. ‧ 國. 學. 詞綴，本文主要的目的就是探討這些動詞詞綴的功能。這些詞綴包括 Ø, m-, mi-, <om>/om-, ma-, man-/mang-, ni-, -en, -an, i-, 以及 ka- -an。本文採用 Foley（2005）修訂. ‧. 版的宏觀角色層級（revised macro-role hierarchy）來檢視雅美語主語的語意屬性. y. Nat. （semantic property），比較這些詞綴在不同類型的句子裡的表現，我們認為這些詞綴. sit. 的功能並不僅限於改變句子的及物性或語態，而是同時扮演著好幾個不同功能。動詞. n. al. er. io. 詞綴 -an 在雅美語中不只表示地點語態，也標示受事者語態，與 -en 有類似的功能。. i n U. v. 研究發現，至少有三組詞根/詞幹（root/stem）與 -an 跟 -en 有不同的互動。其中一組. Ch. engchi. 詞根/詞幹只能被 -an 附加，另一組只能被 -en 附加，第三組則可以被 -an 或 -en 附加。這項發現顯示詞根/詞幹的詞彙語義會影響動詞詞綴的附加選項。本文同時探討了動詞詞綴 ma- 的四項功能，包含主事者語態（activity）、狀態語態（stative）、濳在情態（potentive）跟結果語態（resultative），這四種功能包含了語態（voice）、情態（mood）、終結點（telicity）。由此，本研究認為雅美語的動詞詞綴不只擁有表語態或及物性的功能，而是多功能的詞綴。. iii.

(5) Table of contents CHAPTER 1 INTRODUCTION ............................................................................................... 1 CHAPTER 2 LITERATURE REVIEW AND GRAMMAR SKETCH OF YAMI .................. 6 2.1 INTRODUCTION .................................................................................................................................. 6 2.2 YAMI GRAMMAR ............................................................................................................................... 6 2.2.1 WORD ORDER ................................................................................................................................. 7 2.2.2 CASE MARKING SYSTEM ............................................................................................................... 11 2.2.2.1 Nominative case ........................................................................................................................ 11 2.2.2.2 Genitive case ............................................................................................................................. 12 2.2.2.3 Locative case ............................................................................................................................. 13. 政治大. 2.2.2.4 Oblique case .............................................................................................................................. 13 2.2.3 PERSONAL PRONOUN SYSTEM ...................................................................................................... 15. 立. 2.2.4 VERBAL AFFIXES .......................................................................................................................... 18. ‧ 國. 學. 2.2.5 SOUND SYSTEM AND PHONOLOGICAL CHANGES .......................................................................... 19 2.3 LITERATURE REVIEW....................................................................................................................... 21 2.3.1 FOCUS AND VOICE ANALYSES ...................................................................................................... 22. ‧. 2.3.2 TRANSITIVITY ANALYSIS.............................................................................................................. 25. y. Nat. 2.3.3 FUNCTIONAL CATEGORY OF VOICE AFFIXES AND VERBAL MEANING .......................................... 26. sit. 2.4 PREVIOUS STUDY ON YAMI VERBAL AFFIXES ................................................................................. 31. er. io. 2.5 FRAMEWORK ................................................................................................................................... 37. al. 2.6 CONCLUSION ................................................................................................................................... 43. n. v i n C CHAPTER 3 YAMI VOICE AFFIXESh.................................................................................. 45 engchi U 3.1 INTRODUCTION ................................................................................................................................ 45 3.2 VERBAL AFFIXES ............................................................................................................................. 45 3.2.1 ACTOR VOICE AFFIXES ................................................................................................................. 46 3.2.1.1 The affix Ø ................................................................................................................................ 46 3.2.1.2 The affix m-............................................................................................................................... 47 3.2.1.3 The affix mi- ............................................................................................................................. 49 3.2.1.4 The affix <om>/om- ................................................................................................................. 52 3.2.1.4.1 One participant undergoer <om>/om- clauses....................................................................... 53 3.2.1.4.2 One participant actor <om>/om- clauses ............................................................................... 53 3.2.1.4.3 Two participant actor <om>/om- clauses .............................................................................. 55 3.2.1.4.4 Section summary .................................................................................................................... 57 3.2.1.5 The affix ma-............................................................................................................................. 58. iv.

(6) 3.2.1.5.1 One participant stative ma- .................................................................................................... 58 3.2.1.5.2 Activity ma- ........................................................................................................................... 59 3.2.1.5.3 Potentive ma- ......................................................................................................................... 61 3.2.1.5.4 Resultative ma- ...................................................................................................................... 63 3.2.1.5.5 Section summary .................................................................................................................... 64 3.2.1.6 The affix man-/mang- ............................................................................................................... 66 3.2.1.7 Section summary ....................................................................................................................... 72 3.2.2 NON-ACTOR VOICE AFFIXES ......................................................................................................... 74 3.2.2.1 The affix -en and ni- ................................................................................................................. 74 3.2.2.2 The affix -an ............................................................................................................................. 76 3.2.2.2.1 Causally affected -an clauses ................................................................................................. 77 3.2.2.2.2 Locative -an ........................................................................................................................... 79. 治政 3.2.2.3.1 Affected i- and beneficiary i- clauses..................................................................................... 81 大立 3.2.2.3.2 Instrument i- Clauses ............................................................................................................. 83 3.2.2.3 The Affix i- ............................................................................................................................... 81. ‧ 國. 學. 3.2.2.3.3 Causative pa- and pi- within i- clauses .................................................................................. 84 3.2.2.4 The affix ka- -an ....................................................................................................................... 86 3.2.2.5 Section summary ....................................................................................................................... 87. ‧. 3.3 CONCLUSION ................................................................................................................................... 89. y. Nat. sit. CHAPTER 4 UNDERGOER -EN AND -AN CLAUSES AND MULTI-FUNCTIONAL MA-. er. io. IN YAMI.................................................................................................................................. 93. al. v i n 4.2 THE UNDERGOER AFFIXES -EN ANDC -AN .......................................................................................... 94 hengchi U 4.2.1 EVIDENCE FROM CASE MARKING AND ROLE OF INVOLVED PARTICIPANT(S) ............................... 97 n. 4.1 INTRODUCTION ................................................................................................................................ 93. 4.2.2 EVIDENCE FROM NEGATIVE AND NEGATIVE IMPERATIVE CONSTRUCTIONS ................................ 99 4.2.3 EVIDENCE FROM INTERROGATIVE CONSTRUCTION .................................................................... 103 4.2.4 EVIDENCE FROM THE SEMANTIC PROPERTY OF THE INVOLVED PARTICIPANTS ......................... 106 4.2.5 TOTAL VS. PARTIAL EFFECT OF AN UNDERGOER PARTICIPANT—EVIDENCE FROM THE MARESULTATIVE ...................................................................................................................................... 110. 4.2.6 SECTION SUMMARY .................................................................................................................... 111 4.3 MULTI-FUNCTIONAL MA- ............................................................................................................... 112 4.3.1 STATIVE MA- AND ACTIVITY MA- ................................................................................................ 113 4.3.2 POTENTIVE MA- .......................................................................................................................... 115 4.3.2.1 Evidence from sound changes after negator ji ........................................................................ 116 4.3.2.2 Evidence from verbal affixation after negator ji ..................................................................... 118. v.

(7) 4.3.2.3 Evidence from negative imperative construction .................................................................... 121 4.3.2.4 More evidence for potentive mood from interrogative constructions ..................................... 123 4.3.3 RESULTATIVE MA- AND ACTOR MAN-/MANG- ALTERNATION ...................................................... 126 4.3.4 SECTION SUMMARY .................................................................................................................... 130 4.4 CONCLUSION ................................................................................................................................. 131. CHAPTER 5 YAMI VOICE AFFIXES IN COMPLEX CLAUSES .................................... 135 5.1 INTRODUCTION .............................................................................................................................. 135 5.2 THREE-PARTICIPANT CLAUSES ...................................................................................................... 136 5.2.1 LOCATION (GOAL/BENEFICIARY) ............................................................................................... 136 5.2.2 INSTRUMENT .............................................................................................................................. 139 5.2.3 CAUSATIVE PA- CONSTRUCTION................................................................................................. 140. 治政 ....................................................................................................................... 145 大立 ................................................................................................................ 146. 5.3 SERIAL VERB CONSTRUCTIONS ..................................................................................................... 142 5.4 RELATIVE CLAUSES. 5.5 SUBORDINATE CLAUSES. ‧ 國. 學. 5.6 CONCLUSION ................................................................................................................................. 148. CHAPTER 6 CONCLUSION................................................................................................ 151. ‧. REFERENCES ...................................................................................................................... 155. y. Nat. al. er. io. sit. APPENDIX I ......................................................................................................................... 161. n. APPENDIX II ........................................................................................................................ 168. Ch. engchi. vi. i n U. v.

(8) List of tables Table 1 Correspondence between voice affix and thematic role of clause subject. .................. 3 Table 2 Yami case marker ....................................................................................................... 11 Table 3 Yami personal pronouns ............................................................................................. 15 Table 4 Yami voice affixes ...................................................................................................... 19 Table 5 Yami consonants ......................................................................................................... 20 Table 6 Yami vowels ............................................................................................................... 20 Table 7 Verb type and nominative NPs in Tsou NAV clauses ................................................ 28 Table 8 Puyuma verb classes (after Teng 2008: 120-123, 165-178, 181-183) ........................ 29 Table 9 Proto-role hierarchy, adapted from Foley (2005: 391) ............................................... 30 Table 10 Ho’s (1990) classification of Yami verbal affixes .................................................... 32. 治政大structure ..................................34 Table 12 Meaning of ma- affixed predicate and its argument 立 Table 13 Foley and Van Valin (1984) macro-role hierarchy ................................................... 39 Table 11 Yami pivot, mood, and aspect morphemes ............................................................... 32. ‧ 國. 學. Table 14 Revised macro-role hierarchy ................................................................................... 41 Table 15 Proto-role hierarchy, adapted from Foley (2005: 391) ............................................. 42. ‧. Table 16 The modified version of Foley’s (2005) revised macro-role hierarchy .................... 42 Table 17 Case and semantic property of involved participant(s) of Ø clauses ........................ 47. y. Nat. sit. Table 18 Case and semantic property of involved participant(s) of m- clauses ...................... 49. al. er. io. Table 19 Case and semantic property of involved participant(s) of nominal mi- clauses ....... 50. n. Table 20 Case and semantic property of involved participant(s) of verbal m- clauses ........... 51. Ch. i n U. v. Table 21 Case and semantic property of involved participant(s) of one-participant om clauses. engchi. .......................................................................................................................................... 53 Table 22 Case and semantic property of involved participant(s) of one-participant om clauses .......................................................................................................................................... 55 Table 23 Case and semantic property of involved participant(s) of two-participant om clauses .......................................................................................................................................... 56 Table 24 Case and semantic property of involved participant(s) of stative ma- clauses ......... 59 Table 25 Case and semantic property of involved participant(s) of activity ma- clauses ....... 60 Table 26 Case and semantic property of involved participant(s) of potentive ma- clauses .... 62 Table 27 Case and semantic property of involved participant(s) of resultative ma- clauses ... 64 Table 28 Different types of ma- Clauses ................................................................................. 65 Table 29 Case and semantic property of involved participant(s) of man-/mang- clauses ....... 69 Table 30 Functions of the affixes Ø, m-, mi-, <om>/om-, ma-, and man-/mang- ................... 74 vii.

(9) Table 31 Case and semantic property of involved participant(s) of -en and ni- clauses ......... 76 Table 32 Case and semantic property of involved participant(s) of the -an clause ................. 78 Table 33 Case and semantic property of involved participant(s) of the locative -an clause ... 81 Table 34 Role and case marking of verbs of giving toro and panta in Yami .......................... 82 Table 35 Case and semantic property of involved participant(s) of instrument i- clauses ...... 84 Table 36 Case and semantic property of involved participant(s) of ka- -an clauses ............... 87 Table 37 Property of clause subject and number of participant involved in i- and -an clauses .......................................................................................................................................... 89 Table 38 Functions of Yami verbal affixes.............................................................................. 92 Table 39 Yami pivot, mood, and aspect morphemes (adapted from Rau 2004, 2005) ........... 96 Table 40 Affixation of -an and -en .......................................................................................... 97. 治政大 Table 42 Affixation of undergoer clauses .............................................................................. 103 立 Table 43 Affixation of the predicate within nominalized phrase .......................................... 105 Table 41 Affixation for negative and negative imperative -an and -en clauses .................... 102. ‧ 國. 學. Table 44 Affixation of UV predicate in different types of clauses ........................................ 106 Table 45 Functions of undergoer voice affixes -an, -en, and ni- ........................................... 112. ‧. Table 46 Different types of ma- clauses ................................................................................ 113 Table 47 Stative ma- and activity ma- after negation ............................................................ 115. y. Nat. sit. Table 48 Sound change after negator ji ................................................................................. 117. al. er. io. Table 49 Affixation for negative undergoer and potentive clauses ....................................... 121. n. Table 50 Yami voice and mood affixes ................................................................................. 125. Ch. i n U. v. Table 51 Case marking of participants in English causative, man-/mang-, and -en clauses . 130. engchi. Table 52 Types of affix ma-................................................................................................... 130 Table 53 Functions of four different types of main Yami ................................................... 131 Table 54 Voice, tense, aspect, and mood of Yami verbal affixes .......................................... 134 Table 55 Voice, tense, aspect, and mood of Yami verbal affixes .......................................... 153. viii.

(10) List of figures Figure 1 Dynamic-stative continuum....................................................................................... 28 Figure 2 Semantic property of the Ø clause subject ................................................................ 47 Figure 3 Semantic property of the m- clause subject ............................................................... 49 Figure 4 Semantic property of the mi- clause subject .............................................................. 52 Figure 5 Semantic property of the <om>/om- clause subject ................................................ 57 Figure 6 Semantic property of the ma- clause subject ............................................................. 65 Figure 7 Semantic property of the man-/mang- clause subject................................................ 70 Figure 8 Semantic property of Ø, m-, mi-, <om>/om-, ma-, man-/mang- clause subject ....... 73 Figure 9 Semantic property of -en and ni- clause subject ........................................................ 76 Figure 10 Semantic property of causally affected -an clause subject ...................................... 79. 治政大 Figure 12 Semantic property of ka- -an clause subject............................................................ 87 立 Figure 13 Semantic property of -en, ni-, -an, i-, and ka- -an clause subjects .......................... 88 Figure 11 Semantic property of causally affected i- and beneficiary i- clause subject ........... 83. ‧ 國. 學. Figure 14 Semantic property of clause subject ........................................................................ 90 Figure 15 Semantic property of Yami clause subject ............................................................ 152. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. ix. i n U. v.

(11) Abbreviations and Symbols. 1. first person. RES. resultative. 2. second person. S. singular. 3. third person. STAT. stative. ABS. absolutive. TEL. telic. AV. actor voice. TR. transitive. AUX. auxiliary. UV. undergoer voice. BV. beneficiary voice. CAUS. causative. CONJ. conjunction. EXCL. exclusive. GEN. genitive. INCL. inclusive. IPFV. imperfective. IV. instrument voice. KA. ka-. LA. locative applicative. L. linker. LOC. locative. LV. locative voice. NEG. negative. NOM. nominative. OBL. oblique. P. plural. PI. pi-. PFV. perfective. PN. proper name. POT. potentive. PV. patient voice. QM. question marker. RED. reduplication. 學 ‧. ‧ 國. 立. 政治大. n. er. io. sit. y. Nat. al. Ch. engchi. x. i n U. v.

(12) Chapter 1 Introduction Austronesian languages have exhibited rich verbal morphology and unique subjectverb agreement that have been extensively studied in the early 1990’s. The uniqueness of Austronesian subject-verb relations refers to the property that the verbal affix always signals the thematic role of the clause subject, and this has been referred to as focus in earlier studies (McFarland 1976, Harvey 1979, Starosta et al. 1982, Ramos & Bautista 1986, Huang 1995, Holmer 1996, DeGuzman 1997, among others) and has been treated as voice by Austronesian linguists in the last few decades (Chang 1997, Ross & Teng 2005, Wu 2007, among others). The term focus has been rejected due to its pragmatic usage, opaqueness for the research. 政治大. outside of this subfield, and difficulty for typological study (Teng & Ross 2005). The term voice seems to be a better choice and has been adopted for studies in recent years on. 立. Austronesian languages including Formosan languages.. ‧ 國. 學. The use of the term voice is not without controversy; first, the voice variants for verbs. ‧. in Austronesian languages are not as regular and predictable as in English (Starosta 2002). A verb might have two or more voice variants, but few exhibits a full set of the voice variants. Nat. sit. y. that are available in the language. Second, Starosta (2002) has suggested that the voice. io. er. affixes are actually transitive and applicative affixes that attach to a verb to alter the argument structure of the verb. The appropriateness of using the term voice to refer to the unique. n. al. Austronesian subject-verb relation remains debatable.. Ch. engchi. i n U. v. Unlike the well-known English active-passive, the Austronesian voice system exhibits more than two voice variants as illustrated by Tagalog examples in (1). Clauses with actor subject are usually referred to as actor voice (AV), as in (1a), clauses with patient subject as patient voice (PV), as in (1b), clauses with location subject as locative voice (LV), as in (1c), clauses with instrument subject as instrument voice (IV), as in (1d), clauses with beneficiary subject as beneficiary voice (BV), as in (1e).. 1.

(13) (1) a. actor voice (AV) 1 b-um-ili Ng isda sa VC 2-buy CORE fish OBL ‘The man bought fish in the store.’. tindahan store. b. patient (PV) bi-bilh-in ng lalake sa tindahan IRR-buy-VC CORE man OBL store ‘The man will buy the fish in the store.’ c. location (LV) bi-bilh-an ng lalake ng IRR-buy-VC CORE man CORE ‘The man will buy fish in the store.’. ang. lalake man (Foley 2008: 23). ang. isda fish (Foley 2008: 23). isda fish. ang. tindahan store (Foley 2008: 23). 政治 isda大 fish. d. instrument (IV) ipam-bi-bili ng lalake ng VC-IRR-buy CORE man CORE ‘The man will buy fish with the money.’. ng CORE. isda fish. ang. bata child (Foley 2008: 23). ‧. e. beneficiary (BV) i-bi-bili ng lalake VC-IRR-buy CORE man ‘The man will buy fish for the child.’. salapi money (Foley 2008: 23). 學. ‧ 國. 立. ang. sit. y. Nat. al. er. io. While AV refers to actor voice, the other voice variants, including PV, LV, IV, and. n. BV, are referred to as non-actor voice (NAV) in the literature as the subject role of these. Ch. voice variants shares a common property—non-actor.. engchi. i n U. v. Starosta (2002: 581) has observed that the correspondence between verbal affix and thematic role of a clause subject in Austronesian languages is irregular and unpredictable and has offered an alternative view on the function of these verbal affixes. He has suggested that the so-called voice affixes in Austronesian languages are derivational and their affixation to the predicate is to alter valency of clauses. Accordingly, the function of the so-called voice affixes is to alter transitivity of clauses in Austronesian languages.. 1. Foley’s Tagalog examples are adopted from his paper in 2008 to illustrate the correspondence between the verbal affixation and the thematic role of a clause subject. 2. VC – voice, IRR – irrealis, CORE – term, OBL – oblique. 2.

(14) While the function of verbal affixes is claimed to show agreement with the thematic role of the clause subject in voice analysis—grammatical voice (Reid 1992, Huang 1993, Chang 1997, … etc.), the correspondence between the voice affix and the thematic role of the clause subject is not as regular and predictable as the English active-passive voice (Starosta 2002). In addition to that, the affixation of the so-called voice affixes has been observed to alter valency, change meaning, and co-occur with other voice affixes across Austronesian languages (H. Y. Chang 2011, Shih 2013). Hence, transitivity analysis that attempts to account for these defects of the voice analysis—co-occurrence of two voice affixes and changes in meaning and valency has proposed that the so-called voice affixes are transitivity affixes. The voice analysis might have adequately explained the correspondence between verbal morphology and the thematic role of the clause subject, but transitivity analysis can. 治政 valency after verbal affixation. Both voice and transitivity 大 analyses attempt to treat all the 立 verbal affixes as the same type of affixes, but these affixes behave differently in grammar and better account for the co-occurrence of two verbal affixes and changes in meaning and. ‧ 國. 學. have different appearances—showing up as prefixes, infixes, and suffixes.. ‧. In Yami, a full-fledged verb that exhibits all voice variants is seldom observed and the correspondence between the verbal affix and the thematic role of a clause subject is observed. y. Nat. sit. to be irregular and unpredictable in some instances. Hence, the present study aims to. al. er. io. investigate the function(s) of the verbal affixes that appear to signal the thematic role of the. n. clause subject and alter valency of clauses.. Ch. engchi. i n U. v. The verbal affixes in Yami are observed to participate in different types of clauses and the correspondence between the verbal affix and the thematic role of the clause subject appears to be irregular when it is compared with the English active-passive voice system. More than one affix, such as m-, mi-, <om>/om-, man-/mang-, etc., are observed to participate in constructing a clause with an actor subject (many-to-one correspondence), and one affix, such as ma-, is observed to participate in constructing clauses with both actor and patient subject (one-to-many correspondences), as summarized in Table 1.. Table 1 Correspondence between voice affix and thematic role of clause subject. mam-, mi-, om-/<om>, man-/mangThematic Role of Clause Subject Actor Actor Patient Correspondence One-to-many Many-to-one 3.

(15) The affixation of the locative voice affix -an might have a patient clause subject, or the affixation of ma- could have either an actor or patient clause subject. These mismatches have also cast doubt on voice analysis in Yami.. The present study proposes that the function of the so-called voice affixes in Yami is not limited to signal the thematic role of the clause subject (voice analysis) or to intransitivize/transitivize clauses (transitivity analysis) via investigation of clauses containing these verbal affixes in Yami and the semantic property of the clause subject via the modified version of Foley’s (2005) revised macro-role hierarchy. This hypothesis is demonstrated by comparing the number of participant(s) involved, case marking of involved participant(s),. 政治大. and the sematic properties of the involved participant(s) determined via the modified version of Foley’s (2005) revised macro-role hierarchy in different types of constructions. These. 立. constructions include declarative, negative, negative imperative, and interrogative.. ‧ 國. 學. The investigation of the function(s) of Yami verbal voice affixes is based on data. ‧. from previous research, Yami bible, and field notes. The field notes were mainly collected from middle-aged native speakers of Yami on Orchid Island located near the southeastern. sit. y. Nat. coast of Taiwan. The following lists the language consultants who have contributed their knowledge on the language over the years of my study: Si Vagyatan, Syaman Manidong,. io. n. al. er. Sinan Manidong, Syaman Mapanod, Syaman Yokalaman, Syapen Jinopna, Si Iking and. i n U. v. many other friendly villagers that I have met during my visits to Orchid Island.. Ch. engchi. Our investigation has revealed that some voice affixes serve more than one function in Yami, and the lexical property of roots/stems appears to have some influence on the choice of verbal affixation. These might be the reasons that makes the correspondence between the verbal affix and the thematic role of clause subjects appear to be unpredictable and shaky, and leads to meaning and valency change. We have found that the affix -an that is originally a locative voice affix has served another function in Yami and that is to construct clauses with a patient subject—the same as the well-known patient voice affix -en whose function has been widely observed across the Austronesian literature. It is highly probable that the function of the so-called voice affixes is not simply to signal the thematic role of clause subjects or to transitivize clauses as our study has shown that some of them serve more than one function in Yami. Hence, a thorough study of these affixes is required before one could 4.

(16) fully understand the relation between these affixes and the clause subject. The present study has focused attention on affixes -an and ma-, a locative voice affix and a multi-functional affix, respectively, to confirm our proposed property of the voice affixes in Yami—multifunctionality.. This dissertation consists of six chapters. In the following chapter, related literature on voice analysis and transitivity analysis is discussed with issues regarding mismatches, valency and meaning changing, and the one-voice-per-clause constraint. A brief grammar sketch of Yami is provided, and Foley’s revised macro-role hierarchy and Dowty’s proto-role entailments are also discussed in chapter two. In chapter three, the clauses containing the voice affixes Ø, m-, mi-, ma-, <om>/om-, man-/mang-, -en, -an, ka- -an, and i- are. 治政大 of Foley’s revised macroproperties of clause subjects determined via the modified version 立 role hierarchy are investigated. In chapter four, the function(s) of the affix -an and ma- are. investigated. The number of participants, the case marking of participants, and the semantic. ‧ 國. 學. investigated via different types of clauses, including negative, negative imperative, and interrogative. In chapter five, the affixation of the verbal affixes and its relation with the. ‧. semantic property of clause subject, including case marking of involved participants within complex clauses, are investigated to further support the multi-functionality of the Yami. y. Nat. n. al. er. io. is one of their functions. Chapter six concludes the dissertation.. sit. verbal voice affixes and to further confirm that signaling the thematic role of clause subjects. Ch. engchi. 5. i n U. v.

(17) Chapter 2 Literature Review and Grammar Sketch of Yami 2.1 Introduction The present study aims to investigate functions of Yami verbal affixes including m-, mi-, ma-, <om>/om-, man-/mang-, -en, -an, ka- -an, and i-. These affixes have been claimed to signal the thematic role of clause subjects (voice analysis) or to intransitivize/transitivize clauses in Yami (transitivity analysis). In addition to these two functions, the present study proposes that some if not all of these affixes serve other functions simultaneously, and their multi-functionality has sometimes blurred studies of the verbal system in Yami. This chapter. 政治大 includes a brief summary on the 立previous literature on voice and transitivity analyses and. consists of two parts. The first part is a brief sketch of Yami grammar and the second part. ‧ 國. 學. ends with a discussion of the framework that is adopted in the present study. In the following section, Yami word order, verbal affixes, case markers, sound system, and other relevant grammatical information are discussed and illustrated with examples. Section three starts. ‧. with a discussion of previous studies on Austronesian voice and transitivity analyses and ends. y. Nat. with a discussion on the works that have been done on Yami. The last section introduces the. n. al. er. io. be used in the present study.. sit. framework that is employed to examine the Yami data and defines terminologies that are to. 2.2 Yami grammar. Ch. engchi. i n U. v. Yami, also known as Tao, is an Austronesian language spoken on Orchid Island, which is located near the southeastern coast of Taiwan. It has about 3000 speakers. The land of Yami speakers, Orchid Island, geographically belongs to Taiwan, is linguistically related to the Malayo-Polynesian language family—a primary subgroup of the Proto-Austronesian language (Blust, 2009).. According to Blust (2009), Austronesian languages could be divided into at least ten primary subgroups, and nine out of the ten are represented only in Taiwan. The tenth one is Malayo-Polynesian that could be further divided into two primary subgroups—Western Malayo-Polynesian (WMP) and Central-Eastern Malayo-Polynesian (CEMP) (Blust 2009: 30). According to Blust, Yami belongs to the WMP subgroup that has 500-600 languages 6.

(18) spreading from Orchid Island through the Philippines, the Greater Sunda Island of Indonesia (including Sulawesi), and mainland Southeast Asia and Madagascar. In other words, Yami and other Austronesian languages spoken in Taiwan (also known as Formosan languages) are not as closely related with Philippine languages. This could be further justified by a shared feature of Yami and other WMP languages. Yami and other WMP languages share an areal feature and that is the presence of nasal substitution in active verbs or actor voice verbs. Examples of nasal substitution in Malay, Chamorro, and Yami are provided in (2). The affixation of active or actor voice affixes results in nasal substitution of the initial segment of the root/stem.. (2). 政治大. Malay pukul ‘hit’ (base form): mukul ‘to hit’ (active verb). 立. (Blust 2009: 30). Chamorro saga ‘stay’ (base form): ma-ñaga ‘to stay’ (active verb). ‧ 國. 學. (Blust 2009: 30) Yami rotong ‘cook’ (base form): ma-notong ‘to cook’ (actor voice). ‧. y. Nat. Yami, as in many other Austronesian languages, is a verb initial language. Its word. sit. order, case marking system, pronominal system, verbal affixes, voice variants, sound system. al. n. illustrated with examples.. er. io. and relevant phonological changes are discussed in the following subsections and are. Ch. 2.2.1 Word order. engchi. i n U. v. Yami clauses initiate with a predicate that could be verbal or nominal as illustrated in (3) and (4), respectively.. (3) verbal predicate ya m-itkeh AUX AV-sleep. si NOM. namet 3 PN. 3. Translations of Yami examples into English cannot appropriately express actual meaning as the subject of Yami examples could include location, instrument, beneficiary, etc. that do not exist in English. Hence, the clause subject of each example is underlined in its English translation to show the difference between an English translation and the meaning of Yami the example. 7.

(19) ‘Namet is sleeping.’. (Ho 1990: 60). (4) nominal predicate si Salang NOM PN ‘I am Salang.’. ko 1.S.NOM (Rau and Dong 2006: 89). The examples in (3) and (4) involve only one participant. When a clause contains more than one participant, the order of the participants seems to be flexible in Yami. Examples with the subject, the nominative case-marked NP, taking the clause final position are found in four voice constructions in the data, as illustrated in (5a-d).. 治 lasey si mapapo 政大 mat NOM PN. (5) a. ya man-linas so AUX AV-wipe OBL ‘Mapapo is wiping mats.’. 立. 學. kanakan o soli child NOM taro (Ho 1990: 78). c. ya na linas-an AUX 3.S.GEN wipe-LV ‘Mapapo is wiping the mats.’. ni GEN. mapapo PN. o lasey NOM mat (Ho 1990: 72). sit. y. no GEN. Nat. b. ya na kan-en AUX 3.S.GEN eat-IPFV.PV ‘The child is eating the taro.’. ‧. ‧ 國. (Ho 1990: 92). n. al. er. io. d. i-akan na ni Salang o among ya IV-eat 3.S.GEN GEN PN NOM fish this ‘Salang took this fish and ate it. (lit.) What was given for Salang to eat was this fish.’ (Rau and Dong 2006: 87). Ch. engchi. i n U. v. Examples with the subject taking the position right after the predicate can also be found in AV, PV, LV, and IV constructions in the data, as illustrated in (6a-d), respectively.. (6) a. ya man-bakbak si AUX AV-hit NOM ‘Mapapo is hitting a child.’. mapapo PN. b. na zeyt-en o ayob 3.S.GEN sew- IPFV.PV NOM dress ni namet ayob na dress 3.S.GEN GEN PN ‘Namet is sewing my dress and her dress.’ 8. so kanakan OBL child (Ho 1990: 69). ko 1.S.GEN. aka L. no GEN. (Ho 1990: 106).

(20) c. na osos-an 3.S.GEN drill-LV ‘Mapapo drills the tree.’. o NOM. kayo tree. ni GEN. d. na i-pa-naing o ipangan 3.S.GEN IV-CAUS-cut NOM knife ‘Mapay cut the fish with the knife.’. Mapapo PN (Shih 1996: 75). so among ni Mapay OBL fish GEN PN (Shih 1996: 29). When a clause contains three full noun phrases, the subject can take any position after the predicate—post-predicate position, clause-final position, or the second position of the three NP sequence, as illustrated in (7a-c). However, a clause that contains more than one full noun phrase is rare in discourse data and narratives. When a clause involves more than one or two participants, pronominals are used. The relative order of full noun phrases and. 政治大. pronominals will be discussed in the following subsection on the Yami pronominal system.. mapapo PN. so OBL. ayob cloth. y. n. Ch. si mapapo NOM PN (Ho 1990: 105). engchi. sit. io. c. i-bakbak na ji namet o kayo IV-hit 3.S.GEN LOC PN NOM tree ‘Mapapo will hit Namet with the stick.’. al. so ayob OBL cloth (Ho 1990: 86). ‧. ‧ 國. 學. Nat. b. ya ma-narang ni namet AUX AV-buy GEN PN ‘Mapapo will buy a dress for Namet.’. ni GEN. ni mapapo ya this GEN PN (Ho 1990: 104). er. 立. (7) a. i-panta na si namet BV-give 3.S.GEN NOM PN ‘Mapapo will give a dress to Namet.’. i n U. v. Except for the clause-initial predicate, a few other elements are permitted in the clause initial position. These elements include auxiliary, temporal adverbial, negator, and bound pronouns, as illustrated in (8a-d), respectively.. (8) a. Auxiliary pia ya AUX ø-good ‘The boat is good.’. o NOM. tatala boat. b. Temporal adverbial ni-ma-nabang nokakyab ya yesterday AUX PFV-AV-sew ‘Olongen sewed clothing yesterday.’. (Ho 1990: 60). so milakayob OBL clothing. 9. si NOM. Olongen PN.

(21) c. Negator ko ni-bakbak-an ji NEG 1.S.GEN PFV-hit-LV ‘I did not hit your children.’ d. Bound pronoun man-linas ko 1.S.NOM AV-wipe ‘I am wiping mats.’. o NOM. so OBL. anak child. mo 2.S.GEN (Ho 1990: 128). lasey mat (Ho 1990: 92). There might be more than one element taking the pre-predicate position, as in (8b). The temporal adverbial is observed to take the clause initial position with the presence of the auxiliary ya. In most cases, the auxiliary ya always takes the clause initial position, as illustrated in (9). The auxiliary ya precedes the negator ji in (9a), another auxiliary to in (9b),. 政治大. and the bound pronoun na in (9c).. si NOM. mapapo PN. (aN) QM. (Ho 1990: 125). o cimoy NOM rain. c. ya na ni-kan AUX 3.S.GEN PFV.PV-eat ‘Mapapo has eaten up the taroes.’. ni mapapo GEN PN. y. n. Ch. engchi. sit. io. al. (Rau and Dong 2006: 103) o soli NOM taro (Ho 1990: 74). er. Nat. b. ya to mipá-kdeng AUX AUX gradually-heavy ‘The rain is becoming heavier.’. ‧. ‧ 國. 立. 學. (9) a. ya ji tava AUX NEG ø-fat ‘Isn’t Mapapo fat?’. i n U. v. The order of bound pronouns and the negator ji are not fixed. The negator ji is found to precede a bound pronoun, as in (10a), and follows a bound pronoun, as in (10b).. (10) a. ji ni-bakbak-an ko NEG 1.S.GEN PFV-hit-LV ‘I did not hit your children.’. o NOM. b. ya ji ni-ma-nakew ko AUX 1.S.NOM NEG PFV-AV-steal ‘I have not stolen any pigs.’. 10. anak child. so OBL. mo 2.S.GEN (Ho 1990: 128) kois pig (Ho 1990: 128).

(22) The order of the elements that could take a position before the predicate is summarized in (11). Note that the order of the negator ji and bound pronouns is not distinguished in (11), as the negator ji could either precede or follow bound pronouns.. (11) Temporal Adverbial > Auxiliary ya > other Auxiliary > Negator/Bound Pronoun > Predicate. 2.2.2 Case marking system Noun phrases in Yami are always case marked. Two sets of case markers have been identified—one for common nouns and the other for proper nouns, as summarized in Table 2.. 政治大 identified. These include nominative, genitive, locative, and oblique case markers. 立 Agent/Possessor GEN no ni. Location LOC do ji. Oblique OBL so ---. n. al. er. io. sit. y. Nat 2.2.2.1 Nominative case. ‧. Table 2 Yami case marker Subject Case NOM Common noun o Proper noun si. 學. ‧ 國. Within these two sets of case markers, seven case markers and four different cases can be. i n U. v. The nominative case marker si or o marks the most prominent noun phrase of the. Ch. engchi. clause—subject, as illustrated in (12) and (13). The only noun phrase within one-participant clauses is always nominative case-marked, as in (12). Within two participant clauses, the nominative case-marked NP is the subject of the clause, and its thematic role appears to be signaled by the verbal affix—man- or ni-, as illustrated in (13). The clause with a manaffixed predicate has an actor subject, as in (13a), and the clause with a ni- affixed predicate has a patient subject, as in (13b). Note that the correspondence between verbal affix and the thematic role of a clause subject in Yami is not always as regular as in these two examples and will be further discussed in the following sections.. (12) one-participant clause ya t-om-ava si AUX <AV>fat NOM. mapapo PN 11.

(23) ‘Mapapo is getting fat.’. (Ho 1990: 60). (13) two-participant clause a. actor subject ya man-bakbak si AUX AV-hit NOM ‘Namet is hitting a child’. namet PN. b. patient subject na ni-bakbak ni 3.S.GEN PFV.PV-hit GEN ‘The child was hit by father.’. yama father. so OBL. kanakan child (Shih 1996: 41). o NOM. kanakan child (Chang 1997: 53). 政治大. 2.2.2.2 Genitive case. The genitive case marker, ni and no, have two major functions in Yami. First, it marks. 立. the doer of an action when it is not the clause subject, as illustrated in (14). Second, it. ‧ 國. 學. expresses a possessive relationship, as illustrated in (15). Note that the possessor follows the possessed item and the genitive case marker takes the position between the possessed item. y. Nat. n. al. no GEN. (15) cinai kois no intestine GEN pig ‘intestines of pigs’. Ch. kois kanakan o child NOM pig (Ho 1990: 64). sit. io. ya na ni-zakat AUX 3.S.GEN PFV.PV-kill ‘The child has killed the pig(s).’. er. (14). ‧. and the possessor.. engchi. i n U. v. (Ho 1990: 104). The genitive case marker is observed to mark instrument and beneficiary in some instances, as illustrated in (16). The instrument kayo ‘stick’ that is used to carry out the action is case marked by no—a genitive case marker in (16a); the beneficiary, the one who benefits from the action done by the actor, namet ‘a personal name’ is genitive case-marked in (16b).. (16) a. instrument bakbak-an na si namet hit-PV 3.S.GEN NOM PN ‘Mapapo will hit Namet with a stick.’. 12. no GEN. kayo stick. ni mapapo GEN PN (Ho 1990: 105).

(24) b. beneficiary ya ma-narang ni nament AUX AV-buy GEN PN ‘Mapapo will buy a dress for Namet.’. so OBL. ayob dress. si mapapo NOM PN (Ho 1990: 105). 2.2.2.3 Locative case The locative case markers, ji and do, mark noun phrases that denote location, as illustrated in (17). The location is used as a cover term to refer to goal, source, recipient, etc. as the locative marker in Yami also marks goal, as illustrated in (18a), and recipient, as illustrated in (18b). Note that the locative case marker is not limited to an inanimate noun phrase and could be used to mark an animate noun phrase, as illustrated in (18a-b).. io. a l ji LOC Ch. n. b. recipient ko ni-toro 1.S.GEN PFV.PV-give ‘I gave money to Mary.’. mali PN. ya ni mapapo this GEN PN (Ho 1990: 105). y. kayo stick. sit. Nat. (18) a. goal i-bakbak na ji namet o IV-hit 3.S.GEN LOC PN NOM ‘Mapapo will hit Namet with this stick.’. o mavakes. NOM woman (Rau and Dong 2006: 116). ‧. ‧ 國. da 3.P.GEN. 學. ya m-ian do vahay AUX AV-be LOC house ‘The woman is at their house.’. o NOM. engchi. er. 立. (17). 政治大. niezpi iv money n U. (Chang 1997: 4). 2.2.2.4 Oblique case The nominative, genitive, and locative case markers have two forms—one for proper noun and the other for common nouns. Only the so case marker does not have a corresponding form for proper nouns, as the ungrammaticality of (19a) illustrates. Ho (1990) has reported that a non-subject patient noun phrase is locative case-marked when it is a proper noun, as illustrated in (19b). It has been shown that the so case marker is found to mark an indefinite noun phrase only and always shows up in a two-participant AV construction to mark the patient participant, as illustrated in (20).. 13.

(25) (19) a. *ibakbak. na. o. kayo. so. b. i-bakbak na o kayo ya IV-hit 3.S.GEN NOM stick this ‘Mapapo will use this stick to hit Namet.’. namet. ni. mapapo (Ho 1990: 112). ji LOC. mapapo namet ni PN GEN PN (Ho 1990: 112). (20) ya ni-man-zakat o AUX PFV-AV-kill NOM ‘The child has killed a pig/pigs.’. kanakan child. so OBL. kois pig (Ho 1990: 64). A pair of examples that only differs with the case marker of the theme noun phrase—. 治政大common function. case marker so and the nominative case marker o share some 立 vahay house. (Ho 1990: 113). io. y. sit. do vahay LOC house. (Ho 1990: 113). er. Nat. b. ya m-ian o kanakan AUX AV-be NOM child ‘The child is in the house.’. do LOC. ‧. (21) a. ya m-ian so kanakan AUX AV-be OBL child ‘There is a child in the house.’. 學. ‧ 國. kanakan ‘child’—is observed in the language data, as shown in (21). It seems that the oblique. al. n. v i n Cmarker Furthermore, the oblique case h e nsogiscfound i toUmark the added participant in h causative constructions, as illustrated in (22a). The one-participant predicate lavi ‘cry’ in (22b) is affixed with the causative affix pa- in (22a) that has increased the valency of the predicate by one, and the added participant kanakan ‘child’—the causee—, is case marked by the oblique case marker so.. (22) a. ma-pa-lavi ko AV-CAUS-cry 1.S.NOM ‘I made a child cry.’. so OBL. kanakan (added participant) child (Chang 1997: 22). b. am/om-lavi ko AV-cry 1.S.NOM ‘I want to cry.’. (Chang 1997: 24). 14.

(26) The properties of lacking a corresponding form for proper nouns and marking an indefinite noun phrase have led researchers to treat the so case marker as an oblique case marker (Ho 1990; Deng 2005; Rau and Dong 2006), but it appears to serve a similar function as the nominative case marker as the examples shown in (21). Hence, its function still remains questionable. However, the present study will follow previous studies and treat it as an oblique case marker for the sake of simplicity and comparative studies in the future.. 2.2.3 Personal pronoun system There are five sets of pronouns in Yami; two sets are bound and three sets are free. Within these five sets of pronouns, three case contrasts are made, as shown in Table 3. These. 政治大 corresponding bound forms. Also, the oblique case marker so discussed in the previous 立 subsection does not have corresponding forms in the Yami pronoun system. include nominative, genitive, and locative cases. Only locative pronouns do not have. n. Ch. nio. e ndag c h i. Genitive (Free) niaken nimo nia niamen niaten. y. sit. er. io. al. Genitive (Bound) ko mo na namen ta. ‧. ‧ 國. 學. Nat. Table 3 Yami personal pronouns Nominative Nominative (Bound) (Free) 1S ko yaken 2S ka imo 3S (ya) (iya) 1P (EXCL) namen yamen 1P (INCL) ta, tamo, yaten takamo 2P kamo/kanio inio 3P (sia) sira. iv n U ninio nira. Locative jiaken jimo jia jiamen jiaten jinio jira. The major difference between bound and free pronouns is that the bound pronouns cannot stand alone. They have to attach to verbs or nouns, as shown in (23). The bound pronoun can take the position either before or after the main predicate. The bound pronoun in (23a) and (23b) takes the position before the predicate and the bound pronoun in (23c) and (23d) takes the position after the predicate. The bound pronouns that take the position before or after the predicate specify the actor of the action and are sometimes co-referential with the genitive case-marked noun phrases, if there is one, in the clause. In (23a), the clause contains a genitive case-marked phrase and a genitive bound pronoun that are co-referential.. 15.

(27) (23) a. nai kan-en o 3.S.GEN eat-IPFV.PV NOM ‘Mapapo is eating the fish.’ b. ya ni-bakbak-an mo AUX 2.S.GEN PFV-hit-PV ‘You (sg.) have hit Mapapo.’. among fish. [ni GEN. si NOM. mapapo PN (Ho 1990: 79). c. ni-m-itkeh do rarahan ka PFV-AV-sleep 2.S.NOM LOC road ‘You (sg.) slept on the road yesterday.’ d. k-om-an so ko <AV>eat 1.S.NOM OBL ‘I will eat taro.’. mapapo]i PN (Shih 1996: 40). nokakyab yesterday (Ho 1990: 116). soli taro. 治政 According to Ho (1990), a bound pronoun attaches大 to the clause initial auxiliary if 立 there is one, as illustrated in (24a). If the clause initial auxiliary is not present, the bound ‧ 國. 學. pronoun attaches to the predicate, as in (23a). It is not permitted for the bound pronoun to take the position after the predicate when the clause initial auxiliary ya is present, as. ‧. illustrated by the ungrammaticality of (24b). In other words, the bound pronoun that specifies the actor of the action has to take the second position of the clause in Yami. Furthermore, the. y. Nat. sit. position of a bound pronoun signals tense information of the clause. The presence of the. al. er. io. auxiliary ya indicates that the clause is in present tense. Clauses that are not initiated with the. n. auxiliary ya are non-present—past or future. A post-predicate bound pronoun signals a non-. Ch. i n U. v. present event that is either past or future. With the presence of the perfective prefix ni-, four. engchi. different clause types can be identified. (23a) is an example of a present imperfective sentence—the event is taking place; it describes an on-going action. (23b) is an example of a present perfective sentence—the event has just taken place. (23c) is an example of a nonpresent perfective—the event has taken place in the past. (23d) is an example of a nonpresent imperfective—the event will take place in the future.. (24) a. ya bakbak-an na AUX 3.S.GEN hit-LV ‘He is hitting his child.’. o NOM. anak child. na 3.S.GEN (Ho 1990: 91). b. *ya bakbak-an o anak na AUX hit-LV 3.S.GEN NOM child Intended meaning: ‘He is hitting his child.’. na 3.S.GEN (Ho 1990: 91). 16.

(28) Genitive bound pronouns are observed to attach to nouns to express a possessive relationship, as illustrated in (25). The first person genitive bound pronoun ko attaches to the noun phrase ai ‘foot’ to express a possessive relation.. (25) na ni-songit o ai 3.S.GEN PFV.PV-bite NOM foot ‘The dog has bitten my foot.’. no GEN. ko 1.S.GEN. ino dog. Note that the first person nominative and genitive bound pronouns have the same. 政治大. form. This might sometimes cause confusion in identifying the participants in the clause, as illustrated by the pair examples in (26). The same form at the same position plays different. 立. roles in the clause. In (26a) and (26b), the bound pronoun ko is in its nominative case. ‧ 國. 學. indicating the subject of the clause is first person singular, whereas ko in (26a’) and (26b’) is in genitive case indicating the doer of the action is first person singular.. n. o NOM. Ch. b. ko agek so OBL 1.S.NOM ø-drink ‘I drink water.’ (Chang 1997: 12) b’.. sit. y. wakay sweet_potato. (Shih 1997: 103). ranom water. engchi. er. io. al. a’. agek-en ko drink-IPFV.PV 1.S.GEN ‘I drink the water.’. ‧. Nat. (26) a. koma-k-om-an so ko RED<AV>eat 1.S.NOM OBL ‘I often eat sweet potatoes.’. i n U. v. (Chang 1997: 12). ranom water. i-pa-nazang(paN-sazang) ko 1.S.GEN IV-CAUS-buy ‘I buy clothes for Manlock.’. so ayob OBL clothes. si manlock NOM PN (Shih 1996: 51). Furthermore, there are three slots—third person plural nominative bound pronoun, third person singular bound pronoun, and free nominative pronouns that are in parentheses in Table 3. Yami seems to lack these three forms and the forms in parentheses are sometimes used as substitutes. Compared to the pair examples in (27), the third person singular. 17.

(29) nominative bound pronoun is not observed in (27a), whereas the third person singular genitive bound pronoun is observed in (27b). The pair examples show two facts. First, Yami lacks a third person singular nominative bound pronoun. Second, the reason that the auxiliary ya is sometimes used as a substitute for the lacking third person singular nominative bound pronoun might be due to the position it takes within the clause. It seems that the function of the present tense auxiliary ya has been extended to other usages and serves other functions in the language, in this example—third person singular bound pronoun. This issue requires further investigation in a separate paper on the distribution of the auxiliary.. (27) a. ya k-om-alat do tokon AUX <AV>climb LOC mountain ‘He is climbing in the mountain.’. 立. 政治大 o tokon. b. na kalat-en 3.S.GEN climb-IPFV.PV NOM mountain ‘He is climbing the mountain.’. (Shih 1996: 122). ‧. ‧ 國. 學. (Shih 1997: 122). There is only a set of locative pronouns that is bound and is used to mark a definite. n. al. b.. si NOM. Ch. akay. grandfather. engchi. i-bakbak na o kayo IV-hit 3.S.GEN NOM stick ‘Mapapu will use this stick to hit you.’. ya this. er. io. (28) a. ma-níring jiaken AV-speak 1.S.LOC ‘Grandfather said to me.’. sit. y. Nat. noun phrase that undergoes influence of the action, as illustrated in (28a) and (28b).. iv n U (Rau and Dong 2006: 92) ni jimo 2.S.LOC GEN (Ho 1990: 121). mapapo PN. 2.2.4 Verbal affixes At least four voice variants can be identified in Yami. These include actor voice (AV), patient voice (PV), location voice (LV), and instrument/beneficiary voice (IV/BV). The voice affixes are summarized in Table 4. Patient and location voice affixes are suffixes, whereas the instrument/beneficiary voice affix is a prefix. For the actor voice, there are at least five variants that can be found from previous studies on Yami. Most of the actor voice affixes are prefixes except <om> that appears as an infix and sometimes as a prefix. Note 18.

(30) that the perfective prefix ni- and the patient voice affix -en are in complementary distribution. The presence of ni- signals that the clause is in patient voice while no other voice affix is observed. Hence, the prefix ni- is included in the patient voice affix column.. Table 4 Yami voice affixes AV Affix -om-/m-/mi-/ma-/man-. PV -en/ni-. LV -an. IV/BV i-. As has been mentioned previously, there are at least four voice variants that can be identified in Yami. However, it is difficult to find a verb that exhibits a full set of the voice variants. Rau and Dong (2006) have provided examples of the predicate kan ‘eat’ to illustrate four voice variants, as cited in (29). Note that a verb with a full set of voice variants in Yami is not found except this one.. 立. 政治大. ‧. ‧ 國. 學. (29) a. k-om-an so wakay si Salang <AV>eat OBL sweet_potato NOM PN ‘Salang wants to eat a sweet potato. (lit.) The one who wants to eat a sweet potato is Salang.’ (Rau and Dong 2006: 87). n. al. er. io. sit. y. Nat. b. kan-en na ni Salang o wakay eat-IPFV.PV 3.S.GEN GEN PN NOM sweet_potato ‘Salang ate the sweet poatao. (lit.) What Salang ate was the sweet potato.’ (Rau and Dong 2006: 87). i n U. v. c. ni-akan-an na o mogis ori ni Salang PFV-eat-LV 3.S.GEN NOM rice that GEN PN ‘Salang ate some rice from there. (lit.) What Salang ate a little bit from there was rice.’ (Rau and Dong 2006: 87). Ch. engchi. d. i-akan na ni Salang o among ya IV-eat 3.S.GEN GEN PN NOM fish this ‘Salang took this fish and ate it. (lit.) What was given for Salang to eat was this fish.’ (Rau and Dong 2006: 87). 2.2.5 Sound system and phonological changes According to Ho (1990) and Rau and Dong (2006), there are twenty consonants and four vowels in Yami, as listed in Table 5 and Table 6. There are twelve obstruents and eight sonorants. Obstruents include seven stops—four voiceless /p, t, k, ʔ/ and three voiced /b, ɖ, g/, three fricatives—one voiceless /ʂ/ and two voiced /v, ʁ/, and two affricates—voiceless /tʃ/ 19.

(31) and voiced /dʒ/. Eight sonorants include three nasals /n, m, ŋ/, two liquids /l, r/, one trill /r̆/, and two glides /w, y/. There are four vowels—/i, e, a, o/ in Yami. The high rounded vowel /o/ could sometimes be pronounced as [u] in Yami, but these two vowels are non-distinctive in Yami and are consistently spelled out as [o] in examples. The sounds are spelled out differently in our examples and the spelling is included in the parentheses beside each sound.. Table 5 Yami consonants labial alveolar stop pb t fricative v affricate nasal m n liquid l trill r̆ (z) glide w. 立. Table 6 Yami vowels. r. palatal. velar kg. tʃ (c) dʒ (j). uvular ʁ (h). glottal ʔ (’). ŋ (ng). 政 y治大 Central. ‧ 國. Front i. 學. ə (e) a. Back o. ‧. High Mid Low. retroflex ɖ (d) ʂ (s). y. Nat. There are two phonological changes that are frequently observed in the language data. sit. and are related to the present study. These two phonological changes are discussed and. n. al. i n U. maN-, and the other one involves affixation of <om>/om-.. Ch. engchi. er. io. illustrated with examples in this section. The first phonological change involves affixation of. v. While the affix maN- is attached to roots/stems initiated with /p, t, ɖ, k, ʂ, r, tʃ/, the initial segment of the roots/stems is replaced by the final nasal segment of the affix that retains the place of articulation of the initial segment of the roots/stem. Examples of this phonological change are provided in (30). In (30a), the root/stem initiates with /p/, the affixation of maN- replaces it by /m/ that shares the same place of articulation with bilabial /p/. In (30b-f), the roots/stems initiate with /t, ɖ, ʂ, r, tʃ/, the affixation of maN- replaces the initial segment by alveolar /n/ that shares the same place of articulation with /t, ɖ, ʂ, r, tʃ/. In (30g), the initial velar obstruent /k/ is replaced by velar /ŋ/ after affixation of maN-.. (30) /maN-/ + pareng ‘build’  ma-mareng ‘to build’ /maN-/ + takao ‘steal’  ma-nakao ‘to steal’ 20.

(32) /maN-/ + dengdeng ‘cook’  ma-nengdeng ‘to cook’ /maN-/ + sidong ‘help’  ma-nidong ‘to help’ /maN-/ + rotong ‘cook’  ma-notong ‘to cook’ /maN-/ + cita ‘see’  ma-nita ‘to see’ /maN-/ + kali ‘dig’  ma-ngali ‘to dig’ According to Rau and Dong (2006), the affix om is inserted into roots/stems whose initial consonants are /ʂ/, /t/, /k/, or /g/, as in s-om-dep ‘enter’, t-om-anek ‘stand’, k-om-ala ‘search’, etc. and can either be infixed or prefixed to roots/stems whose initial consonants are /l/, /d/, /r/, or /z/. In other cases, om can only prefix to the roots/stems.. In this section, a brief sketch of Yami grammar, including word order, case markers, pronouns, verbal voice affixes, sound system, and two related phonological changes, is. 政治大 discussed before discussion of the framework that is to be adopted in the present study. 立 discussed. In the next section, some related studies on focus, voice, and transitivity are. ‧ 國. 學. 2.3 Literature review. In earlier studies, the unique relation between verbal affixation and the thematic role. ‧. of clause subjects is referred to as the focus system, and the languages that exhibit this system. y. Nat. are referred to as Philippine-type languages. Later, researchers (Chang 1993, Ross & Teng,. io. sit. 2005) suggested that this phenomenon is better treated as voice instead of focus due to the. er. pragmatic usage of the term focus. The disadvantages of using the term focus include the. al. n. v i n C been adopted U and comparative works. The term voicehhas e n g c h i and is used extensively in the. opaqueness for the research outside of this subfield and the difficulty for typological studies. literature to refer to this phenomenon in the last few decades. However, the term voice also raises some objections that include having more than one passive in accusative analysis. Some studies still use focus to refer to the relationship between the verbal affixation and the thematic role of clause subjects. The terms focus and voice are widely found across the Austronesian literature. In the present study, the term voice is adopted to refer to the relation between the verbal affixation and the thematic role of a clause subject.. In this section, the traditional focus analysis, voice analysis, and transitivity analysis are briefly discussed. All of these analyses attempt to address the issue of the relation between verbal affixation and the thematic role of clause subjects as this issue has remained controversial over the years.. 21.

(33) 2.3.1 Focus and voice analyses Blust (2009) in working on the description of the syntax of Austronesian languages has pointed out two dilemmas. First, Austronesian languages have wide geographical distribution and are in contact with languages of other families. Researchers who try to provide a broad description of Austronesian languages commonly encounter a great variation in syntactic types among these languages. Second, it is not an easy task to describe the syntax of a language without reference to a theoretical framework, and the framework choice might not be agreed upon by all the researchers working in this subfield. The description of the Austronesian voice system can be dated back to the early nineteenth century, but the relationship between predicate and the nominative noun phrase still remains controversial due to its unique property that is not found in other languages of the world and due to work using. 政治大. different theoretical frameworks (Blust 2009: 431).. 立. Liao (2004) has discussed and classified the diverse linguistic descriptions on voice of. ‧ 國. 學. Philippine-type languages from the early nineteenth century until the present day into three chronological periods and these include traditional, classical, and modern periods. According. ‧. to her, the description of Austronesian focus/voice constructions in the traditional period that begins in the early nineteenth century has been highly influenced by Latin grammar. The. y. Nat. sit. linguistic descriptions in this period have suggested that Austronesian languages exhibit one. al. er. io. active and three passives. The -um- clauses are considered as active; the -en clauses, -an. n. clause, and si- clauses are considered as passives. In this period, it was claimed that the. Ch. i n U. v. Austronesian languages exhibited the accusative-nominative system in that the subject of an. engchi. intransitive clause and the agent of a transitive clause are coded the same, whereas the patient is coded differently.. In the classical period, under the influence of structuralism and the work of the Summer Institute of Linguistics (SIL), the uniqueness of Austronesian focus/voice recognized, and the term focus was introduced by SIL linguists and has been used extensively in the literature (Starosta et al. 1982, Huang, 2001, Zeitoun, 1992, and many others). The modern period starts in the mid-seventies with the influence of generative linguistics and the work of Starosta and his students. Various analyses have been proposed in this period; these analyses include active analysis, ergative analysis, fluid voice analysis, hybrid voice analysis, and symmetrical voice analysis.. 22.

(34) In active voice analysis, Drossard (1984, 1994) has proposed a two-way voice system—active and stative for Tagalog voice constructions. The verbs that are marked by ma- are stative; verbs that are not marked by ma- are active. In ergative analysis, the -en clause (PV) is treated as the basic transitive clause, whereas the -um- clause (AV) is considered to be the intransitive clause of the language. The subject of the intransitive clause and the object of the transitive clause are coded the same, whereas the agent is coded differently. Hence, it has claimed that Austronesian languages exhibit an ergative system.. In fluid voice analysis, Shibatani (1988) has suggested that Austronesian languages are neither accusative nor ergative, noun and verb distinction in the root form is not observed, and all the focus/voice constructions are derived. In hybrid voice analysis, Maclachlan (1996). 治政大three non-oblique cases have AV and PV are both treated as the basic transitive clause, and 立 been identified. In other words, there is no antipassive or passive construction in Tagalog. working on Tagalog has suggested that it exhibits neither an accusative nor ergative system.. ‧ 國. 學. within Maclachlan’s analysis.. ‧. The symmetrical voice system, unlike the asymmetrical voice system such as accusative-nominative English or ergative-absolutive Dyirbal that has a preferred NP type. y. Nat. sit. (actor for English and undergoer for Dyirbal) for the sentence pivot, does not have a. al. er. io. preferred NP type for the sentence pivot. In the symmetrical voice system, any type of NP (e.. n. g. actor, undergoer, etc.) that is chosen as the sentence pivot is signaled by overt verbal voice. Ch. i n U. v. morphology (e. g. -um-, -en, -an, i-, etc) and all the voice constructions are equally marked.. engchi. Liao (2004) has summarized seven analyses on Austronesian focus/voice constructions in chronological order. The study of Austronesian focus/voice constructions can be traced back to the early nineteenth century, and the works on Austronesian focus/voice constructions still remain pervasive and diversified in the last few decades due to its uniqueness compared to other languages of the world, the complexity of the verbal morphology and clause structures, and works using different theoretical frameworks.. While diversified works on voice analysis keep expanding, flaws of voice analysis have been detected. The relation between verbal affixation and thematic role of clause subjects is not as regular as the English active-passive voice system. The same verbal affix might have a subject bearing different thematic roles, and a different verbal affix might end 23.

(35) up with a subject bearing the same thematic role. Teng’s (2007) work on Puyuma has suggested that the voice affix in Puyuma does not necessarily match the thematic role of the clause subject, as shown in (31). The clause subject of the LV affix -ay could be patient, as in (31a), source, as in (31b), goal/location, as in (31c), or causee, as in (31d). That the verbal voice affix does not always match the thematic role of a clause subject, and this is referred to as the mismatch problem, and the correspondence between the verbal affixation and subject role remains mysterious as the verbal affixation does signal the thematic role of clause subject in some instances.. (Teng 2007: 166). io. d. causee tu=pasisi-ay=ku 3.GEN=force-LV=1S.NOM ‘She forced me to work.’. n. al. pa-karun CAUS-work. Ch. dekal village. engchi. y. Nat. c. goal/location tu=u-sabak-ay nanta 3.GEN=go-inside-LV DF.NOM/1P.PSR ‘They invaded our village.’. ‧. ‧ 國. paisu money. 學. da ID.OBL. (Teng 2007: 166). (Teng 2007: 166). sit. 立. b. source tu=takaw-ay-ay=ku 3.GEN=steal-LV=1S.NOM ‘He stole money from me.’. 政治大. er. (31) a. patient 4 tu=salpit-ay=ku 3.GEN=flog-LV=1S.NOM ‘He flogged me.’. i n U. v. (Teng 2007: 166). While the mismatch problem has been widely reported cross Austronesian languages (Teng 2007, H. Y. Chang 2011, Shih 2013), transitivity analysis has started to win recognition among Austronesian linguists as the evidence has shown that voice analyses cannot fully account for the correspondence between the verbal affixation and thematic role of a clause subject.. 4 ID - indefinit, DF - definite, PSR - possessor.. 24.