台語ka字句的二元分析

全文

(1)國立交通大學外國文學與語言學研究所碩士論文. 台語 ka 字句的二元分析. A Non-unified Analysis of Taiwanese ka Construction. 研究生：陳佑慈指導教授：劉辰生. 博士. 中華民國九十七年七月.

(2) 台語ka 字句的二元分析 A Non-unified Analysis of Taiwanese ka Construction. 研究生：陳佑慈指導教授：劉辰生. Student：Yu-Tzu Chen Advisor：Chen-Sheng Liu. 國立交通大學外國文學與語言學研究所碩士論文. A Thesis Submitted to Department of Foreign Languages and Literatures Graduate Institute of Foreign Literatures and Linguistics National Chiao Tung University in partial Fulfillment of the Requirements for the Degree of Master In. Graduate Institute of Foreign Literatures and Linguistics July 2008 Hsinchu, Taiwan, Republic of China. 中華民國九十七年七月.

(3) 台語 ka 字句的二元分析字句的二元分析. 學生：陳佑慈. 指導教授：劉辰生博士. 國立交通大學外國文學與語言學研究所. 摘要本篇論文主要探討台語的 ka 在句中所扮演之功能。普遍認為台語 ka 為漢語 ba 之對應詞，兩者在句法上以及語意上所具有的特質都十分類似，唯 ka 有能力指派論旨角色(thematic role)予其後之論元(argument)及光桿動詞(bare verb)在 ka 句中不會造成句子的不合法。而 ba 則不能指派論旨角色予其論元亦不允許光桿動詞出現於句中，此為兩者明顯之不同點。相較於前人將 ka 單純視為 ba 的對應詞，我們提出台語的 ka 具備兩種功能，其一是作為漢語 ba 句的對應詞。其二是作為施用結構(applicative construction)之標記。當作為施用結構標記時，ka 可以引介一個額外的論元(additional argument)進入論元滿足(argument-saturated)的句中。在 Pylkkanen(2005)的分析架構下，我們相信台語的 ka 可歸類為高級施用結構(high applicative construction)。針對光桿動詞可在 ka 句中出現卻不能在 ba 句中出現之現象，本文採用 Feng(2005)所提出的 Government-Nucleus Stress Rule (G-NSR)來解釋何以漢語的 ba 句不允許光桿動詞在句中出現。我們相信由於台語本身具備了複雜的連調變化系統同時來決定重音，故台語的重音分布並不受該規則所限，而是與連調變化彼此相互牽引。至於 ka 句允許光桿動詞在句中出現之現象，我們相信追根究底仍舊與動詞本身的 boundness 有一定之關聯。關鍵字：台灣閩南語，ka，施用結構，Government-Nucleus Stress Rule (G-NSR). i.

(4) A Non-unified Analysis of Taiwanese ka Construction Student ：Yu-Tzu Chen. Advisor： Dr. Chen-Sheng Liu. Graduate Institute of Foreign Literatures and Linguistics National Chiao Tung University. Abstract The purpose of this thesis is to study the function of the Taiwanese ka in the sentence. It is widely believed that Taiwanese ka is the counter part of Mandarin ba. They share many similar syntactic and semantic features. Although ka is treated as the counter part of Mandarin ba, there still remain differences between them. For instance, ka is able to assign thematic role to its argument, but ba is incapable to assign thematic role to the argument. Also, in the ka construction, bare verb is tolerated under some circumstance. This never happens in the ba construction. Unlike the previous researches simply treat ka as the counter part of ba, we propose that ka has two functions. One is to be the counter part of ba and the other is to be the applicative marker. When ka is treated as an applicative marker, ka is able to introduce an additional argument into an argument-saturated sentence. Based on the analysis proposed by Pylkkanen (2005), we believe that Taiwanese ka construction belongs to the high applicative construction. As for the reason that the bare verb is not tolerated in the Mandarin ba construction, Government-Nucleus Stress Rule (G-NSR) raised by Feng (2005) will be applied to explain the phenomenon. In Taiwanese, we believe that because Taiwanese has a very complicated tone-sandhi system to help speakers to deicide the stress, it need not follow the G-NSR. Therefore, we believe that what really affects the existence of the bare verb in Taiwanese may still be the boundness of a verb. Keywords： Taiwanese Southern Min，ka，applicative construction， Government-Nucleus Stress Rule (G-NSR). ii.

(5) 誌謝在交大這三年，是我求學生涯裡最快樂的三年，承蒙許多老師的教誨，還有許多同學的相互激盪與鼓勵，讓我度過了充實而快樂的時光。首先我要感謝我的指導教授劉辰生老師，劉老師不僅啟發了我對句法方面的興趣，並且總是非常有耐心的跟我討論我研究中所碰到的問題與反例，在我的研究遇到瓶頸時，給我很多珍貴的建議跟想法。在我為了反例苦惱的時候，適時提醒我不要鑽牛角尖，我覺得自己很幸運，能夠遇到一位好老師。誠摯的謝謝劉老師這三年來的教導，也希望老師可以多保重身體，謝謝老師。其次我要謝謝林若望老師跟林宗宏老師，謝謝兩位老師在口試的時候給我很多建議並提出許多問題，讓我看到自己研究的盲點跟缺陷，使我有機會能夠加以改進，謝謝兩位老師。再來我要謝謝我的學長昭廷，謝謝你不管在我研究上或是人生上給我的中肯見解，在我需要幫忙的時候毫不猶豫的伸出援手，能有你這個學長我感到既驕傲又榮幸。謝謝我的好友子佳，謝謝你給了我很多很多的幫忙跟鼓勵，在每個我慌張的時刻，你總是能讓我適時的冷靜，謝謝你。謝謝我的好友孟純跟靜玉，能在研究所期間認識妳們兩個好同學，跟你們在生活與學術上相互扶持，是我的福氣，謝謝你們忍受我的直言不諱還有我的脾氣，不管我多麼任性總是順著我，也謝謝你們在我最脆弱的時候一路相挺，讓我知道我並不孤單。這些我都記在心裡。我知道我們會是一輩子的好朋友。接下來我想謝謝子玲，璦羽，學妹雯靜，學長嘉軒對我的照顧，我真的很高興可以在研究所期間認識你們這些好朋友，每次心情煩悶時跟你們聊聊就好的多，謝謝你們。謝謝廷方百忙之中幫忙看我的文章。謝謝靜儒姐姐在我慌張的時候替我出主意想辦法，用過來人的經驗讓我不再慌張。謝謝樹岡一路上給我的幫忙，非常謝謝你們。謝謝我專屬的心靈捕手小皓，謝謝你總是將我放在你前面，給我最好的一切。陪著我一起笑，一起哭，分擔我的沮喪，分享我的快樂。在每個我投出暴投或保送的時刻，謝謝你陪著我面對、克服，伴著我度過難關。每當我低潮時，因為有你的鼓勵，讓我有了勇敢往前走的動力。你總是在我需要鼓舞時幫我加油，在我需要希望時替我創造。親愛的小皓，謝謝你總是給我信心，謝謝你總是對我有信心。最後我想謝謝我最最最愛的爸爸媽媽，你們是我最堅實的後盾，在我開心愉快時比我更快樂，在我心灰意冷的時候給我溫暖跟希望。謝謝你們總是無條件的給予我最好的一切，跟你們滿滿的愛。謝謝你們總是相信我做得到，並鼓勵我，讓我無後顧之憂的去追逐自己的夢想。我很榮幸能成為你們的女兒，也很高興能有你們這一對這麼棒的爸媽。也因為有你們的愛跟支持，我才能在這條路上走的這麼遠。如果我能在人生裡能有一點什麼成就，那都是因為有你們的愛與我相伴。這篇論文是獻給你們的，親愛的爸媽，我愛你們 iii.

(6) Table of Contents Chinese Abstract...................................................................................................................i English Abstract...................................................................................................................ii Acknowledgement...............................................................................................................iii Table of Contents.................................................................................................................iv Chapter 1 Introduction.........................................................................................................1 Chapter 2 Previous Analyses...............................................................................................6 2.1 Audrey Y.-H. Li (2006): The Light Verb Approach............................................6 2.1.1 The ba construction....................................................................................7 2.1.2 Application of the analysis of ba construction to the Taiwanese ka construction..............................................................................................16 2.1.3 Unsolvable problems................................................................................20 2.2 Barry C.-Y. Yang (2005): The NOP Approach .................................................23 2.2.1 The questions regarding the ka NP.........................................................23 2.2.2 The obligation of the X-factor.................................................................27 2.3 Rint Sybesma (1999): The Event Structural Approach....................................33 2.4 Summary...............................................................................................................37 Chapter 3 Proposal..............................................................................................................39 3.1 Preliminary...........................................................................................................39 3.1.1 The observations of the Taiwanese ka construction..............................39 3.1.2 The applicative construction...................................................................45 3.2 The analysis: Two functions of ka......................................................................54 3.3 Summary...............................................................................................................66 Chapter 4 Issues about the X-FACTOR: The Government-Nucleus Stress Rule.........67 4.1 Preliminary...........................................................................................................67 4.1.1 The non-X-factor sentences in the Taiwanese ka construction............68 4.1.2 The Government-Nucleus Stress Rule...................................................71 4.2 Applying the G-NSR on the Taiwanese ka and Mandarin ba construction....75 4.3 Conclusion............................................................................................................82 Chapter 5 Conclusion.........................................................................................................83 References...........................................................................................................................84 iv.

(7) CHAPTER 1 INTRODUCTION. In the previous researches of Taiwanese, most of the linguists have agreed that Taiwanese ka is the counterpart of the Mandarin ba. Ba and ka share most semantic and syntactic constraints. For example, both ka and ba may present the disposal meaning with the following form:. (1) a. NP1 ba/ka NP2 X V. b. NP1 ba/ka NP2 V X.. (Li 2006). In (1), NP1 is the subject of the sentence and NP2 is the object of the sentence. In order to get the ba/ka construction, ka and ba allow the syntactical freedom to transpose an object to an adjacent position. Although ba and ka share many similar features with each other, there still remain differences. For instance, the request of the X-factor is different between them. A bare verb is tolerated in Taiwanese ka construction but not in Mandarin ba construction. In Mandarin ba construction, the X-factor is obligatory in the sentence.. 1.

(8) The X-factor can either occur before or after the verb if only it appears in the sentence. A ba sentence without the X-factor is always ungrammatical:. (2) a. *ta. ba. wo. He ba. me. ma. scold. He scolds me. b. *wo. ba. Wang. xiao.. ba. Wang. laugh. I. I laughed at Wang.. As shown in (2a) and (2b), both sentences are ungrammatical because there are no X-factors in the ba sentences. A ba sentence with a bare verb form is always ungrammatical. It must have an X-factor. Nevertheless, a bare verb form is tolerated in Taiwanese ka construction. Unlike ba construction, a ka construction can still be grammatical without the X-factor under some circumstances. For instance:. (3) a. li. ka. you ka. goa2. ma7.. me scold. 2.

(9) You scold me. b. goa2. ka. Ong-e. chhio3.. me. ka. Ong-e. laugh. I laughed at Ong-e.. (3a) and (3b) demonstrated that bare verbs are tolerated in the sentence. This is impossible in Mandarin ba construction. The different request of the X-factor is one of the distinct features between ba and ka construction. Besides, ka is able to assign thematic role to the NP following ka. Mandarin ba does not have the ability to assign thematic role to any NP:. (4) a. li. keng3-jian5. you unexpectedly. ka. goa2. chau2. me. run. ka. khui. a.. leave. PRT. You run away on me unexpectedly. b. i. ka. he ka. goa2. thau. chau2. me. secretly. run. khi3 leave. a. PRT. Her run away on me secretly. c. i. keng3-jian5. he unexpectedly. ka ka. goa2 me. the5-cha2. ha7-ban.. earlier. duty-off. 3.

(10) He get off work on me unexpectedly.. It is impossible for the post-ka NPs in (4) to get thematic roles from the main verbs in these sentences because there is no obvious semantic relation between the main verb and the post-ka NP. The post-ka NP must obtain its theta role from ka in (4). Unlike Taiwanese ka, Mandarin ba has no ability to assign a thematic role for the argument. The argument always obtains its theta role from the main verb in the ba construction. This is the other difference between ba and ka. In sum, although Taiwanese ka is the counterpart of Mandarin ba, there seem still a lot of differences between these two constructions. As mentioned above, ka can assign thematic role to the post-ka NP and it is permitted to take a bare verb in ka sentence. The X-factor is optional in the ka construction. However, a bare verb form is not tolerated in the ba construction. The X-factor is obligatory in the ba construction and ba has no ability to assign thematic role to any argument. If ka is the counterpart of Mandarin ba, what is the feature that makes these two constructions become so different when dealing with the post-NP and the X-factor question? In this paper, I would like to discuss questions we have mentioned above and try to bring up some possibilities for the questions. I believe that ka itself has two. 4.

(11) functions, one is functioning as an applicative marker and the other is to be the counterpart of ba. The first function would be operated when an additional argument is introduced into the sentence. The second function triggers the object move to the post-ka position. As for the reason that Mandarin ba construction always need the X-factor in the sentence, I would like to use Feng’s (2005) theory called Government-Nucleus Stress Rule (G-NSR). The boundness of the event would be used to explain why the X-factor is optional in the ka construction.. 5.

(12) CHAPTER 2 PREVIOUS ANALYSES. In this chapter, previous works discussing the Mandarin ba and the Taiwanese ka constructions will be reviewed. Both the Mandarin ba and the Taiwanese ka constructions have been extensively studied. The Taiwanese ka construction is widely considered as the counterpart to the Mandarin ba construction. Both constructions take the similar form [NP1+ba/ka+NP2+V+X] to present the disposal meaning. In a sentence, ka and ba allow the syntactical freedom to transpose an object to an adjacent position; ba/ka and share most semantic and syntactic constraints. A Mandarin ba sentence can always be translated into a Taiwanese ka sentence. We review some previous research regarding the ba construction in this section. In section 2.1, we briefly discuss Audrey Li’s analysis of ba/ka constructions, and in section 2.2, Barry Yang’s analysis is discussed. In section 2.3, Rint Sybesmas’s viewpoint is introduced.. We also investigate the various deficiencies of these. analyses.. 2.1 Audrey Y.-H. Li (2006): The Light Verb Approach. 6.

(13) 2.1.1 The ba construction Audrey Li (2006) has noted that ba was initially a lexical verb meaning ‘hold, take, and handle’ in Mandarin. However, these lexical meanings of ba have turned much weakened. In other words, ba has been grammaticalized and has lost the ability of giving thematic role to its argument in Mandarin--that is to say, ba has no longer worked as a lexical verb today. When disposal meaning is expressed, the ba construction is exhibited in forms of (a) NP*+ba+NP+V+X or (b) NP*+ba+NP+X+V. The X factor is obligatory in the ba construction. X may be placed either anterior or posterior to the verb. Without the X factor the ba construction will become ungrammatical, as (5) presents:. (5) a. ta ba. ni. he ba. hai*(-le). you. hurt-le. ‘He hurts you.’ a’ . wo I. ba ta ba. he. ma*(-le) scold-le. ‘I scolded him.’ a’’. ta ba he ba. cai. *(xian). vegetable *(first). qie chop. 7.

(14) ‘He chops the vegetable first.’. The semantic meaning of (5a) and (5a’) exhibits that the ba construction is used to exhibit the disposal meaning. The NPs following ba (ba NP henceforth) in (5) are either an affected object or a victim; in either case conveying disposal. Obviously the thematic role of the ba NPs in (5) stems from the verb because the grammaticized ba has lost its ability to assign thematic roles to the ba NP. As can be seen in (5a), (5a’) and (5a’’), X can exist in either the preverbal or the postverbal position. The only requirement for X is that it must occur in the ba sentence. Without X, the sentence is ungrammatical. Since ba is no longer a lexical verb and loses its ability to assign thematic roles, it is reasonable to assume that ba has lost most of the characteristics unique to lexical verbs: (6) b. *ta ba-le he. ba-le. ni. you hurt(-le). c. *ta ba-mei/bu-ba he. hai(-le). ba-not-ba. ni. hai(-le). you hurt. d. *(mei/bu-)ba (not-)ba. 8.

(15) Both (6b) and (6c) show that unlike a lexical verb, ba can neither take an aspect marker nor form a V-not-V question. Also, ba cannot be a simple answer to a question such (6d). This strong evidence provides a clue that ba is grammaticized and no longer a lexical verb in modern Mandarin. Next, we discuss the source of the ba NP. Li (2002) pointed out that the occurrence of the ba NP is the result of movement. To obtain a ba construction, the object of the main verb is moved to the position following ba. If the ba NP is the object of the main verb in the ba sentence, the post verbal position should be unoccupied. This applies to pronouns or reflexive co-references with the ba NP. The ba sentence becomes ungrammatical if anything is inserted at the object position of the main verb under this hypothesis. This is because if the ba NP is the result of movement, it will leave a trace at the object position. To prove this hypothesis, we direct your attention to (7):. (7) a. *ta ba Zhangsani he ba. Zhangsan. hai-le tai hurt-le him. ‘He hurts Zhangsan’ b. *ta ba. Zhangsani. hai-le. ziji/tazijii. 9.

(16) he ba. Zhangsan. hurt-le self/himself. ‘He hurts Zhangsan’. In (7a) and (7b), the ba sentence is ungrammatical because the object positions are occupied by a pronoun and a reflexive pronoun, respectively. This is not allowed because the object position has already been occupied by an invisible trace. The trace stems from the movement of the ba NP. This proves that the ba NP is the result of movement. Compare the ba construction (8a) and (8b) with the lexical verb shi construction (8c):. (8) a. *ta ba Zhangsani he ba. hai-le tai. Zhangsan. hurt-le him. ‘He hurts Zhangsan’ b. *ta ba. Zhangsani. hai-le. ziji/tazijii. he ba. Zhangsan. hurt-le self/himself. ‘He hurts Zhangsan’ c. ta shi. Zhangsani. he make Zhangsan. hai-le. ziji/tazijii. hurt-le self/himself. 10.

(17) ‘ He makes Zhangsan hurt himself’. Unlike (8a) and (8b), the similar sentence is grammatical in (8c). What makes the difference is because shi is a lexical verb that has the ability to assign thematic role and Case to the noun following shi (shi-NP henceforth). The shi-NP is not the result of the movement but of the base generated process. Shi-NP gets its own thematic role and Case from the lexical verb shi. Therefore, unlike ba-NP, the shi-NP has no relationship with the main verb in this example. Examples (5) to (8) demonstrate that ba does not have the ability to carry an argument itself because it is not a lexical verb. Ba has no ability to assign thematic roles to the ba NP. The ba NP must be the result of movement. Although ba is incapable of assigning thematic role to the ba-NP, ba still has the ability to give the exceptional Case to the ba-NP. Case is also what triggers ba-NP moving to the following ba position. In summary, since ba is not a lexical verb and is unable to provide a thematic role to the ba NP, the ba NP cannot be base generated. Li (2006) has stated that the ba NP is the result of movement and originates from the verbal phrase. She proposed that the ba NP is the V object or the V’ object of the verb and is thus moved to the position posterior to ba. The fact that a trace is occupying the object position of the verb also. 11.

(18) provides an explanation as to why one cannot put a pronoun or a reflexive pronoun there in a ba construction. Another issue that has been widely discussed regarding the ba construction concerns the necessity of the X factor. It is known that in the Mandarin ba construction, the existence of a bare verb is disallowed. There must be an X factor preceding or following the verb in the sentence. Without the X factor, the sentence becomes ungrammatical:. (9) a. *wo I. zuotian. zhe-ge zhuozi ca. yesterday this-Cl table. wipe. ‘ I, this table, wiped yesterday.’ b. wo I. zuotian. zhe-ge zhuozi ca-le. yesterday this-Cl table. wipe-le. ‘ I, this table, wiped yesterday.’ c. * zhe-ge zhuozi, this-Cl table,. wo. zuotian. ca. I. yesterday wipe. ‘This table, I wiped yesterday.’ d. zhe-ge zhuozi,. wo. this-Cl table,. I. zuotian. ca-le. yesterday wipe. 12.

(19) ‘This table, I wiped yesterday.’. Both (9a) and (9c) are ungrammatical because they contain a bare verb without any X factor. Li claimed that this due to a widely observed phenomenon in Mandarin that if the object of a sentence has been preposed, a bare verb is not tolerated. Furthermore, Li stated that X is necessary because the ba construction requires it to intensify the disposal meaning. Li believed that a longer X can strengthen disposal meaning. If a verb has a very weak disposal meaning, a long X is necessary in a ba construction. On the contrary, if the verb itself has a very strong disposal interpretation, then the X does not need to be too long. Li and Thompson also mentioned that the more elements added to elaborate the nature of disposal, the more likely the sentences are to appear in the ba form. The following tree diagram represents the structure of ba sentences:. 13.

(20) (10). (Li 2006). In (10), vP1 is the verb phrase that follows the ba NP (which is NP2 here). NP1 is the external argument. To obtain the surface structure of the ba sentence, ba should be moved from the specifier of ba’ to v*, and V3 should be moved to v1. ba is no longer a lexical verb, and thus unable to assign thematic role here because of the grammaticization process. Hence, the ba NP derives its thematic role from V3. v* disappears and the external argument NP1 occupies the specifier position of the baP and the ba NP occupies the specifier of vP1. Subsequently, baP loses its light verb projection because ba is not a lexical verb and is unable to assign a thematic role to the ba NP. Ba still has the ability to assign an exceptional case to the ba NP. This is also why when we add ba to the sentence, to trigger the object movement. Henceforth, the structure of the ba sentence becomes as (11): 14.

(21) (11). (Li 2006). The structure of a non-ba counterpart is given as (12): (12). (Li, 2006). It has been proven that a Mandarin ba sentence always has a non-ba 15.

(22) counterpart. The non-ba counterpart has a structure similar to the ba construction. The only difference is that the non-ba counterpart has no baP.. 2.1.2 Application of the analysis of ba construction to the Taiwanese ka construction Ka, which is the Taiwanese counterpart of the Mandarin ba, shares many similar features with the ba construction. They both use similar forms [Subject+ba/ka+NP+V+X] to represent the disposal meaning. Ka has the ability to assign case to the NP following ka (ka NP henceforth). This feature is exactly the same as the Mandarin ba. Since they share so many similar features, this is also the reason that linguists believe that ka is the Taiwanese counterpart of Mandarin ba. Nevertheless, close observation of ka and ba constructions reveals differences. A Mandarin ba sentence always has a Taiwanese ka counterpart:. (13) a. ta ba He ba. wo. ma. san-ci. me scold three-times. le. PRT. ‘ He scolded me three times.’ b. i. ka. he ka. goa2 me. ma7 scold. saN-pai2. a.. three-times PRT. 16.

(23) ‘He scolded me three times’. However, a ka construction cannot always find its counterpart in ba construction. Moreover, though a bare verb cannot exit in a ba construction, its existence in certain conditions does not turn the ka construction ungrammatical:. (14) a. i. cin su-iau. He very need. lang. ka i thiann. people ka he care. He needs other people’s care b. *ta hen. xuyao. He very need. ren. (Lin 2007). ba ta teng. people ba he care. He needs other people’s care c. i. ka. goa2. He ka. me. (Lin 2007). phah/pian2/ma7 hit/cheat/scold. He hit/cheat/scold me d. *ta ba wo da/pian/ma he ba me hit/cheat/scold He hit/cheat/scold me. 17.

(24) In the previous discussion we have mentioned that a bare verb in ba construction is not tolerated in Mandarin. Nevertheless, the bare verb in Taiwanese ka construction is undoubtedly acceptable, as shown in (14a) and (14c). In the Taiwanese ka construction, the bare verb is obviously accepted when the verb takes an affectee as its object. Here comes the question. Ba-NP in the ba construction is also an affectee. Now that ka in Taiwanese is the counterpart of the ba in Mandarin, why does ka allow a bare verb when ka-NP is an affectee, while ba does not? While Li has not offered an explanation, she emphasized that the Taiwanese ka is more lexical than the Mandarin ba. She claimed that ka is not only able to assign case, but also has the ability to assign thematic role to the ka NP. Since ka can assign both thematic role and case to the ka NP, it is reasonable to assume that ka NP is base generated, not the result of movement. Li has proposed that although ka is the counterpart of ba, ka has some abilities that ba has not. Ka can assign thematic role to the ka NP and is assumed to be a light verb carrying the affect meaning. Therefore, according to Li, the ka-NP is not necessary to have any thematic relation with its following V or V’. For example:. (15) a. li you. mai don’t. ka ka. goa2. the5-cha2. me earlier. ha7-ban off work. 18.

(25) ‘Don’t get off work earlier (than you should) on me.’ b. li. na. you if. kaN2. ka. dare ka. goa2 me. chau2, goa2 leave, I. chiu7 then. ho. li. let you. ho2-khoaN3 good-look. ‘If you dare to leave on me, I will show you the consequence.’. In (15), both thematic roles of the ka-NPs in (15a) and (15b) are given by ka, but not the verb. The ka-NP in (15a) and (15b) obviously has no thematic relation with the main verb. This is convincing evidence suggesting that a ka NP can be base generated. Being the counterpart of the ba construction, the structure of the ka construction is quite similar to it, as shown in the following example:. (16). (Li 2006). 19.

(26) The biggest difference between (16) and (11) is that in (16) we keep the light verb construction for ka. Li pointed out that ka is able to assign the thematic role to the ka NP; therefore, to keep a light verb projection for ka is reasonable. The Mandarin ba is unable to assign thematic role. For the reason, it lacks a light verb projection in the structure. Since ka has the ability to assign thematic role and case to the following NP, we could assume that the ka-NP is base generated, but not the result of the movement. Unlike the Mandarin ba, Taiwanese ka has more independent status. (16) also explains why not all of the ka sentences have non-ka counterparts. It is because that ka can assign thematic role and Case to the ka-NP. We cannot simply delete ka in the sentence to return to the non-ka form. Instead, we have to delete both ka-NP and ka. If we simply delete ka, the existence of ka-NP will result in an ungrammatical sentence. Ba in Mandarin, however, will not lead to the same outcome because ba-NP is always related to the following V or V’ thematically. Consequently, a non-ba sentence can be derived by deleting ba and moving ba-NP back to the postverbal position. The difference between ba and ka strongly prove that ka-NP does not result from the movement, but the base generated process.. 2.1.3 Unsolvable problems. 20.

(27) There are several problems germane to Li’s analysis which we have been unable to solve. First, is the problem of X. Li mentioned that in Mandarin, a bare verb is not tolerated in a sentence having a moved NP. Thus a ba sentence must have an X following the verb. However, she did not explain why in a sentence with a moved NP, a bare verb is disallowed. Even if we accepted Sybesma’s1 point of view, we still could not understand why an X is necessary to strengthen the disposal meaning. As the counterpart to the Mandarin ba, why the Taiwanese ka construction allows a bare verb form following the ka NP remains unsolved. In Li’s essay she, noted that such a phenomenon exists, but she did not attempt to provide a reasonable explanation. Li further believed the ka NP in the ka construction to be base generated. However, if ka NP were base generated, it would be difficult to determine the location of the argument of verbs such as ‘hit, cheat, and scold’ in a sentence such as (14). Obviously, the ka NP in (14) must be transferred from the object position to the position following ka. The moved object has a thematic relation with the main verb. If we accepted Li’s proposal that ka is able to assign thematic role and case, then the moved object would assume two thematic roles and cases; one being from the original verb and the other from ka. Nevertheless, it has been widely proven that one NP can only assume one thematic role and one case. Thus, we are left with a dilemma as Li’s 1. Sybesma accepted Li and Tompson’s point of view. He believed that the existence of X is to strengthen the disposal meaning. The longer X is, the stronger is the disposal meaning the X factor can provide. (this is still unclear to me). 21.

(28) analysis yields no convincing explanation for the sentences like (14). We discovered another type of Taiwanese ka construction in which the which the thematic role of the ka NP is not affected. The thematic role of this type of ka NP could be a theme marker, a goal marker, or a source marker. For example:. (16) a. Abing. ka. Asan. ka3. Eng-bun5. Abing. ka. Asan teach English. ‘Abing teaches Asan English.’ b. Abing Abing. ka ka. Asan. phian3. Asan cheat. lak8-pah six-hundred. kho dollar. Abing cheated Asan of six hundred dollars. The thematic role of the ka NP Asan in (16a) is goal; while in (16b) it is source. Although according to Li’s analysis there can be thematic roles such as affectee or beneficative (broadly speaking, a beneficative is also a type of affectee), seen in (14) and (15). However, in (16a) and (16b) the ka NPs are neither affectees nor beneficatives. Can ka construction only produce thematic roles such as affectee and beneficative? This is another unsolved problem regarding ka construction. Li mentioned that in ba construction, ba assigns an exceptional case to the ba NP.. 22.

(29) However, she did not explain why the ba NP takes the exceptional Case but not the verbal case. Since the case of the ba NP can be satisfied when the NP is at the post verbal position, there is no need for a NP movement here. To sum up, Li’s analysis tried to explain why bare verbs are disallowed in ba construction and whence the thematic role of the ba NP derives. She also explained ba NP as being the result of NP movement, and the reason why ka construction does not always have a non-ka counterpart. However, she did not explain the location of the case that the original verb has given to the ba NP. She has provided unclear explanations as to why the ka construction can bear a bare verb form and the origin of different thematic roles such as in (16) if the ka NP is base generated and ka is able to assign thematic roles. Another question that may arise is that if ka NP is base generated, how can sentences such as (15) be explained? The ka NP in (15) should not be treated as the result of movement since the ka NP and the main verb are obviously thematically related. Next we are going to discuss Barry Yang’s analysis, and consider the problem that may happen under his analysis.. 2.2 Barry C.-Y. Yang (2006): The NOP Approach 2.2.1 The questions regarding the ka NP. 23.

(30) Yang’s point of view is basically similar to Audrey Li’s. He has treated ka as a light verb carrying affect meaning. He also considered that total acceptance of Li’s argument will lead to a dilemma. According to Li, ka is more a lexical verb than ba, and capable of assigning the thematic role. Ka can carry its own argument and the ka NP will be licensed by ka. In other words, ka-NP is base generated.. (17) a. Abu Mother Abing b. Asan. ka Asan. se2. saN. ka Asan wash clothes washes clothes for Asan ka. Asan ka. goa2. chau2-khui. me run-away. Asan run away on me / I was affected by Asan’s running away. Adding an additional pronoun to the sentence can easily prove that ka-NP is the result of movement:. (18) a. *Abing. ka. Asani phah-si2. Abing. ka. Asan. Abing. beat Asan to death.. ii. a. beat-dead him Perf.. 24.

(31) b. *Abing ka Asani ka3 Abing ka Asan teach. ii. Eng-bun5. him English. Abing teaches Asan English.. (18a) and (18b) are ungrammatical because in each example, an invisible trace following the ka NP is left right after the main verb after the movement of the ka NP. If another pronoun is added, the sentence will be ungrammatical. This proves that the ka NP is the result of the movement. The above-mentioned phenomena bring a paradox when exploring the origin of the ka NP, because a NP cannot be base generated as well as from the result of the movement. Therefore, Yang has offered a NOP approach, by which Yang has considered workable to deal with the paradox. The NOP approach is presented as follows: If we are to explore the gapped ka construction2 , we will treat the lower embedded IP as the secondary predicate induced by the null operator as (19):. 2. The gapped ka construction here means the ka-NP has a gap for it to resconstructed back. This kind of ka-NP is the result of movement, for example: a. Goa2 ka Abing phah Me ka Abing hit I hit Abing b.. goa2 ka Abing ma7 Me ka Abing scold I scolded Abing Abing in sentence (a) and (b) can be reconstructed back to the object position and the sentence is still grammatical.. 25.

(32) (19) gauj ka. Abingi [IP OPi [IP PROj phah ti]].. I. ka. Abing. I. beat Abing. beat. When the non-gapped construction3 such as (20) happened, we can use the null ‘outermost object’ construction to explain it:. (20) Ij. ka goa2 [OPi [ti [ PROj chau2-khui]]]. He ka. I. run-away. He ran away (on me) or I was affected by his running away. In (20), a null outermost object is adjoined to the complex IP position, and moved to the IP adjoined position through the NOP movement. ka has two complement, one is the ka-NP and the other is the embedded IP. By using the null outermost object construction we can also explain why the ka-NP will become a benefactive or an adversative when we deal with the non-gapped ka construction Treating ka as a light verb in the NOP approach, we can exert the affectedness. 3. The non-gapped ka construction means the ka-NP has no gap to reconstructed back. This kind of ka-NP is base gerenated like (1). 26.

(33) on the object. Yang proposed that the different theta roles are nature consequences of the secondary predication achieved by the NOP movement so we can solve the problem of the different theta roles that happened in the ka construction. The null operator movement results the IP complement in a lambda predicate. This lambda predicate will strengthen the properties of the predicate through strong binding. Since the ka NP and the IP predicate has such relationship, the thematic relationship is established in a nature way. 2.2.2 The obligation of the X-factor By solving the problem of whether ka NP is the result of movement or base generated, another problem is raised: the Mandarin ba cannot take a bare verb as its complement like (21a) presented, but in a ka construction, a bare verb complement is allowed like (21b) presented. Since it is widely recognized that Taiwanese ka is the counterpart of the Mandarin ba, the reasons behind this difference are of interest.. (21) a.*wo I. ba. Zhangsan. da. ba. Zhangsan. beat. I beat Zhangsan. b. goa2 I. ka ka. Abing Abing. phah beat. 27.

(34) I beat Abing.. The other problem needs to be discussed is that ba always has its non-ba counterpart, as shown in (22) and (23). However, this does not always occur in the ka construction, as shown in (24) and (25):. (22) a. wo ba I. ba. Zhangsan. ma-le. yi-dun. Zhangsan scold-Perf one-CL. I scolded Zhangsan severely. b. wo ma-le Zhangsan. yi-dun. I. scolded Zhangsan one-CL. I. scolded Zhangsan. (23) c. wo ba I. juzi. severely.. bo-le. pi. ba tangerine peel-Perf skin. I peeled the tangerine. d. juzi,. wo. tangerine, I. bo-le. pi. peel-Perf skin. Tangerine, I peeled off the skin. (24) a. i. ka goa2. chau2-khui. 28.

(35) he ka me run-away He ran away (on me) b. *i. chau2-khui. he run-away (25) c. Abu Mother. goa2 me. ka Abing. se2. ka. wash clothes. Abing. saN. Mother washes clothes for Abing d. *Abu. se2. saN. Abing. Mother wash clothes Abing Mother washes clothes for Abing. Yang has started that we can also use the NOP movement approach to solve these problems. Regarding bare verbs, Yang has quoted Liu’s analysis (1997), pointing out that ba sentence requires its predicate to denote a bounded event or situation. To get a bound event, the bare verb needs to be followed by some additional forms, as are usually observed in the ba construction.. (26) a. V+ complement. 29.

(36) b. V+ de c. V+ retained object d. V+ perfective marker le e. V+ PP f. V+ quantified phrase g. V+ yi+ V h. V+ durative marker zhe i. Adv+ V. Liu has argued that ba is unable to take a bare verb because the predicate of the ba sentence is required to be a bound event. To achieve this condition, an X element has to be occurred with the main verb, in the preverbal or the postverbal position. Through the addition of the X element the main verb thereby becomes a bound event. This is why an X element is always needed in Mandarin ba construction. However, this interpretation results in some problems. Firstly, why does Mandarin ba sentence require its predicate to be a bound event? In addition, since Taiwanese ka construction is the counterpart of Mandarin ba construction, why Taiwanese ka construction does not need an X element to have the verb become a bound event? Yang proposed these problems can also be coped with through the NOP approach:. 30.

(37) We should initiate by comparing the syntactic structures between the ba and ka constructions:. (27). (Yang 2006). (Yang 2006) In (27) we notice that in the ka construction, the ka-NP is farther away from the main predicate than the ba NP is. Besides, ka-NP here is situated in the higher VP 31.

(38) shell and the main predicate is located in the embedded IP closed off by the null operator in the ka construction. This is why ka-NP cannot require the main predicate to be bounded. For the same reason, a bare verb is allowed to occur in the ka construction. Broadly speaking, ka is more analytic and lexical than ba. Yang’s argument has suggested that the position of the ka NP is far away from the main predicate, and thus fails to require a bound event. But Yang did not explain why. In addition, such hypothesis also fails to give a persuasive interpretation explaining why sentences like “i ka goa2 phah” and “i ka goa2 phah chit8-e7” in Taiwanese are grammatical. There seems to be no direct connection between the two phenomena. To sum up, the NOP approach can solve the problem how various theta roles are assigned in the ka construction, and whether ka NP is the result of the movement or base generated. But when it comes to X-factor, the NOP approach still cannot give us a convincible explanation. Another unsolved question is why a ba-NP always has a non-ba counter part but the ka construction does not always has a non-ka counter part? These questions are left to be further explored. In 2.3 we will discuss the event-structural approach provided by Sybesma, and explore whether this approach could help answer the questions above.. 32.

(39) 2.3 Rint Sybesma (1999): The Event Structural Approach Sybesma has suggested that there are two kinds of ba construction: one is causative ba construction and the other is canonical ba construction. These two kinds of ba construction can be differentiated by subject. A canonical ba sentence has an animate subject and a causative ba sentence has an inanimate subject. Regardless which kind of ba construction used, Sybesma has proposed that a ba sentence is always a CAUS-sentence in an abstract sense, and that these two kinds of ba sentences share most syntactic and semantic characteristics. (28) is the construction that Sybesma devised for both causative ba construction and canonical ba construction:. (28). Sybesma has proposed that the meaning of the ba sentence can be paraphrased as ‘the subject causes the ba-NP to undergo the event denoted by the VP ’, so he. 33.

(40) choose a CausP as its projection. The VP embedded under the CAUSP is unaccusative in this sense. An unaccusative VP has no external argument. VP here only has a resultative small clause as its complement. The thematic role of NP1 stems from the head CAUSE, not from the verb. NP2 is the ba NP and NP3 is NP2’s trace. NP2 must be the subject of the embedded clause and has no direct semantic relation with the matrix verb. It is thematically dependent on the constituent embedded under V. Since an unaccusative verb has no objective Case, the object must move and derive its case from CAUS. Sybesma also suggests that the head of CAUS must be filled. There are two possibilities: one is move V to the CAUS, the other is to insert ba into the position of the CAUS. The first one produces a non-ba sentence, and the second results in a ba sentence. Sybesma adopt P. Wang’s analysis to explain why ba construction always requires an X-factor following the verb. He believes that the ba-NP is always related to the X-factor. Whether this is interpreted as ‘disposal of’ or not depends on the nature of the post verbal constituent. A longer X-factor results in a stronger disposal meaning demonstrated by the construction. Sybesma attempts to fit is analysis to all kinds of ba construction. However, there still remain problems. The first is regards NP2; Sybesma mentioned that the. 34.

(41) NP2 must be the subject of the embedded clause. For example:. (29) a. nei-ge. nuhai ku-de Zhang San nian-bu-xia. That-CL girl. nei-ben-shu.. cry-DE Zhang San read-not-on that-CL book. ‘That girl cried so that Zhang San could not continue reading that book.’ b. nei-ge. nuhai ba Zhang San ku-de. that-CL girl ba. nian-bu-xia. Zhang San cried-DE. nei-ben-shu.. read-not-on that-CL book. ‘That girl cried so that Zhang San could not continue reading that book.’ c.* nei-ge. nuhai ba. that-CL girl. ba. nei-ben-shu. ku-de. Zhang San. nian-bu-xia.. that-CL book cried-DE Zhang San read-not-on. However, while this may be true under most circumstances, not all ba sentences follow such restrictions, for example:. (30) a. nei-ge. nuihai ku. That-CL girl. de. cry de. mei ren no. gan. mo. people dare. na shoupa. touch that handkerchief. ‘That girl cries so hard that no people dare to touch that handkerchief.’ b. nei-ge. nuihai ba na shoupa. That-CL girl. ku. de. mei ren. ba that handkerchief cry de no. 35. gan. mo. people dare touch.

(42) c.*nei-ge. nuihai ba. That-CL girl ba. mei ren no. ku. de gan. mo. na. shoupa. people cry de dare touch that handkerchief. In (30a), shoupa obviously is not the subject of the embedded clause; it is the object of the embedded clause. According to Sybesma’s analysis, if shoupa becomes the ba NP, the sentence is ungrammatical. However, in (30b) the sentence remains grammatical with shoupa as the ba NP. If we raise the subject of the embedded clause mei ren here as (30c), the sentence becomes ungrammatical. This suggests that the assumption Sybesma made is too strong. Secondly, Sybesma believes the ba structure always has a CAUSP. This is because the meaning of ba sentences can be paraphrased as ‘the subject causes the ba-NP to undergo the event denoted by the VP’ and this is why he put ba on the specificier of CAUSP. Nevertheless, this may be questioned if we use the actual CAUSP ‘shi’ in Mandarin to instead of ‘ba’:. (31) a. ni you. ba ba. zhe-kuai rou this-Cl. meat. qie-qie. ba!. cut-cut. Par!. ‘Cut the meat!’. 36.

(43) b.(?) ni you. shi. zhe-kuai rou. qie-qie. shi. this-CL. cut-cut. meat. ba ! Par !. ‘You cause the meat to be cut!’. In (31a) and (31b), the meaning of the sentence is totally changed. In fact, (31b) is not intuitively grammatical to most native speakers. Thus, it is not appropriate for ba to be put on the specifier of CAUSP is been strongly proved. In fact, Sybesma’s suggestion that the X-factor following the main verb functions to make the disposal meaning stronger is an ad hoc solution. If the X-factor is to give the disposal meaning, then why can the Taiwanese ka construction tolerate a bare verb form, give that ka is the counterpart of the Mandarin ba construction? Why does the Taiwanese ka construction not need the X-factor to intensify the disposal meaning? The existence of the X-factor remains an unsolved question.. 2.4 Summary In section 2 we provide a discussion of the previous analyses that Audrey Li, Barry Yang, and Rint Sybesma has developed. We found that although these analyses can solve portions of the problems inherent in the Mandarin ba and Taiwanese ka construction, none of them are able to provide a satisfactory solution to all other. 37.

(44) questions surrounding the Taiwanese ka construction. This is especially the case with respect to the different uses of X-factor between Mandarin ba and Taiwanese ka construction. There seems to be no convincing explanation for this phenomenon. The origin of the ka-NP is also a controversial issue deserving further investigation. In sentences such as, “Abu ka Asan se sann“(“Mother washes clothes for Asan.”), the ka NP is obviously not the result of movement, but base generated. In the sentences such as, “i ka goa2 pah” (“He hits me.”), the ka- NP is obviously the result of movement. We provide our proposal and a reasonable solution for the questions regarding the origin of the ka-NP in Section 3. We use the Government-Nucleus Stress Rules to explain the different uses of and requirements of the X-factor in Section 4.. 38.

(45) CHAPTER 3 PROPOSAL. In this Chapter, basic observations of ka-construction and some theoretical concepts of applicative construction will be introduced as preliminaries of our analysis. Specifically, we propose that ka in Taiwanese not only is the counterpart of Mandarin ba but also can be an applicative marker that introduces an additional argument.. 3.1 Preliminary In 3.1.1 some observations of ka-construction in Taiwanese will be described. It was found that ka is more lexical than ba because ka is capable to assign theta role to the following NP. In the ka construction, not all of the ka-NPs are the result of movement. Some of the ka-NP is obviously base-generated. We will bring up our assumptions for the Taiwanese ka construction. In 3.1.2 we are going to introduce the applicative construction for our further analyses as the background knowledge.. 3.1.1 The observations of the Taiwanese ka construction. 39.

(46) Most of the previous analyses suggested that ka in Taiwanese could be understood as the counterpart of the Mandarin ba. For example:. (32) a. li. ka. chit-nia2. saN. se2. se2. e. You ka this-CL clothes wash wash PRT (You) wash this clothes a’. ni. ba. zhe jian yifu. You ba this-CL. xi. xi. ba. clothes wash PRT. (You) wash this clothes b. Ong-e ka. phoe. Ong-e ka. letter. sia2. hoo a. write done-PRT. Ong-e finished writing the letter b’. Ong-e ba Ong-e. ba. xin. xie. hao le. letter write done-PRT. Ong-e finished writing the letter. c. i. ka hit siang e5-a ching-phua a. He ka that-CL shoes wear-threadbare PRT He wore his shoes threadbare d. ta. ba na shuang xie-zi chuan-po le. 40.

(47) He ba that-CL shoes wear-threadbare PRT He wore his shoes threadbare. Judging from the parallelism shown in (32), ka could be treated as the counterpart of ba in this type of construction. Moreover, we believe that like ba-NP, the post-ka object NP (ka-NP henceforth) undergoes Case-driven from the object position of the verb to the surface position after ka. The main verb is unaccusativized and thus incapable of Case-marking. We assume that the Case of the ka-NP is from ka. Our assumption could be proved by the observation that ka-NP is inseparable from ka in the sentence. In order to get the Case, ka-NP must follow adjacency condition. In addition, ka is found to be capable of introducing an additional argument into a thematically-saturated sentence:. (33) a. li. m-hoN ka goa2 the5-cha2. you. don’t. ka. ha7-ban.. me earlier. off work. ‘Don’t get off work earlier (than you should) on me.’ b. li. na kaN ka goa2. you if. dare ka. me. chau2, goa2 leave,. I. chiu. ho. li. ho2-khoaN3. then. let you good-look. ‘If you dare to leave on me, I will show you the consequence.’. 41.

(48) (Li, 2006). Obviously, the ka-NP goa2 in (33) is thematically unrelated to the main verb ha-ban/tsao. However, according to Theta Criterion, goa2 must receive a theta-role or it cannot be used in the sentence grammatically. Nevertheless, it is not possible for the main verb to assign a theta role to the ka-NP because the main verb is already thematically saturated. Witness the fact that sentences in (33) remain grammatical without the presence of ka and ka-NP as shown in (34a/b/c) ), and the ill-formed sentences (34c/d/e) obtained after removing only ka from the sentence, we assume that ka is responsible for assigning theta role to the applicative ka-NP and thereby licenses its appearance.. (34) a. i. ka. He ka. goa2. se2. saN. me wash clothes. He washs clothes for me/ He helps me wash clothes b. i. ka goa2. he ka me. chau2 khui a leave-go-PRT. He leaves from me c. i. ka. goa2. be2. chhia. 42.

(49) He ka. me. buy car. He helps me buy car d.* i φ goa2 se2 saN Heφ me wash clothes e. *i φ goa2 chau2 khui a Heφ me leave-go-PRT f. *i φ goa2 be2 chhia Heφ me buy car. More sentences with applicative ka are provided in (34a/b/c). Given the fact that the ka-NP in these sentences is thematically unrelated to the main verb, it is reasonable to suspect that the ka-NP in applicative ka-construction is base-generated at its surface position instead of moving from certain position within VP like those in (32). So far, the ka-construction can be differentiated into two types according to how ka-NP is produced. On the one hand, the applicative use of ka introduces/licenses an additional argument NP that is thematically unrelated to the main verb and such ka-NP is assumed to be base-generated at its surface position. On the other hand, when the ka is treated as the counterpart of ba, the ka-NP undergoes Case-driven A-movement to its surface position from the complement position of the main verb.. 43.

(50) Though it is generally assumed that ka is the counterpart of ba, there is one vital distinction between them as pointed out by Li (2002). That is, there is an obligatory post-verbal X factor in ba-construction. However, such constraint does not hold in ka-construction:. (35) a. goa2 ka I. ka. Abing. phah. Abing. beat. I beat Abing b. *wo I. ba. Zhangsan. ba. Zhangsan. b’ wo ba Zhangsan I. ba Zhangsan. da beat. da. le. beat. le. I beat Zhangsan c. goa2 ka i ma7 Me ka i scold I scolded him d. *wo ba ta ma I ba him scolded d’ wo ba ta ma-le. 44.

(51) I ba him scolded-PRT I scolded him. Sentences in (35) demonstrated the X-factor is obligatory in the ba construction but not in the ka construction. The ka sentence remains grammatical under some circumstances. In order to explain this phenomenon, I would like to suggest Government Nucleus Stress Rules (G-NSR) to provide another path to find out the different need of the X factor in Mandarin and Taiwanese. There will be more detailed discussion about the X factor in section 4.. 3.1.2 The applicative construction In Section 3.1.1 we proposed the two functions of ka. First, it is the counterpart of the Mandarin ba. Second, ka can function as an applicative marker. In Taiwanese, an additional argument can be added into a sentence with the help of ka. In this section, the analyses of applicative construction will be introduced. Typically, applicative construction provides an extra position for placing an additional argument in a sentence. Applicative construction can be used to increase the transitive of verbs (Peterson 2007)4. For example, if applicative construction is 4. Peterson (2007) mentioned that applying applicative construction may intensify the transitivity of the verb or simply result in the rearrangement of an argument structure. The applicative construction we discuss here is the one that gives a verb the phenomenon of argument augmentation.. 45.

(52) used in a sentence having a transitive verb, the transitive verb will be able to carry an additional object. In English, sentence like ‘I cooked Mary a meal’ is treated as a sentence applying applicative construction. Applicative construction can be introduced either by a word or by an affix. A word or affix which introduces the applicative argument, is known as the applicative marker. In Austronesian languages, an affix is used as the applicative marker to introduce an additional argument:. (36) no-helo’a-ako. te. 3R-cook-APPL core. ina-no. [Tukang Besi]. mother-3-POSS. They cooked for their mother. (Bresnan and Moshi 1990). Some languages do not use applicative markers to present applicative construction. Instead, they use double-object construction. For instance, non-marked double-object construction functions as applicative construction in English. The function of applicative construction does not change, regardless of whether it is introduced by an applicative marker or by the double-object construction. The purpose of the applicative construction is to intensify the transitivity of a verb and enable the verb to carry an additional participant of the event:. 46.

(53) (37) a. I cooked a meal.. [English]. b. I cooked him a meal.. Marantz (1993) proposed that a sentence like (37) is an example of hiding applicative construction in a double-object construction. In (37a), ‘cook’ carries one object, ‘a meal,’ in the sentence. A transitive verb can only carry one object. Therefore ‘cook’ is a transitive verb. Applying applicative construction to (37a) allow the transitive verb ‘cook’ to carry an additional new argument, ‘him’, as shown in (37b). This is an example of using non-mark double-object construction as applicative construction. Marantz believed that sentences such as (37b) are applicative constructions because, within applicative construction, the applied argument will asymmetrically c-command the direct object. This c-command irregularity is a defining property of a double-object and of applicative construction (Barss and Lasnik 1986, Marantz 1993). In applicative construction, the theme object must follow a beneficiary object. (Alex Alsina and Sam A. Mchombo 1993) The usage of applicative construction is not an isolated phenomenon. It is widely found in various world languages. In African tongues, such as Bukusu, Luganda and Kinchaga, applicative construction is observed:. 47.

(54) (38) a. n-a-i-lyi-i-a Foc-1s-PR-eat-app-FV. m-ka. k-elya. 1-wife. 7-food. [Kinchaga]. He is eating food for/on his wife b. wanjala. a-mu-kul-il-a. sii-tabu. Wanjala 3ss-CL10-buy-APP-FV. CL7-book. Wanjala bought her a book c. Mukasa. ya-tambu-le-dde. Katonga. Mukasa PAST-walk-APPL-PAST. Katonga. Mukasa walked for Katonga. [Luganda]. (Pylkkanen 2002). (38a), (38b) and (38c) represent that, in African languages, applicative construction can be denoted by the addition of the affix. The main verbs of the sentences (38a), (38b) and (38c), ‘eat’ and ‘bought,’ are both transitive and can carry only one object. After applicative construction is applied, both verbs have the ability to carry an additional argument. Applicative construction can also be found in languages like Japanese, Korean, and Albanian:. 48.

(55) (39) a. Taroo-ga. Hanako-ni. Taro-NOM Hanako-DAT. tegami-o. kaita. [Japanese]. letter-ACC wrote. Taro wrote Hanako a letter. b. John-i. Mary-hanthey pyunci-lul. John-NOM Mary-DAT. sse-ess-ta. letter-ACC. [Korean]. wrote-PAST-PLAIN. John wrote Mary a letter c. Drita. i. poqi Agimit. Drita ACC-CL. kek. [Albanian]. baked Agim.DAT cake. Drita baked Agim a cake (Pylkkanen 2002). An argument introduced by applicative construction is known as an affectee, or a beneficative if it is animate5. This is because an argument introduced by applicative construction is usually influenced by the event described by the verb phrase or by the object in the sentence. Therefore, it is reasonable to assume that the thematic roles of the animate applicative NPs are either that of benefactive or an affectee. For example:. (40)a. Katonga. ya-kwaant-i-dde. Mukasa ensawo. 5. [Luganda]. ‘Animate' here means a living thing or a thing that is living at the moment we talk. This feature is specifically for human.. 49.

(56) Katonga PAST-hold-APPL-PAST Mukasa bag Katonga held the bag for Mukasa b. Nd-aka-uray-ir-a I-PST-steal-APPL-FV. nyoka. pa-dombo. 1-mother. [Changa]. 9-money. I stole money form my mother (Pylkkanen 2002). In (40a), the applicative affix ‘-i-’ introduces an additional argument. ‘Mukasa’, is the beneficative of the event ‘Katonga held the bag.’ In (40b), the applicative affix ‘–ir-’ makes ‘mother’ the affectee of the event ‘I stole money.’ Both of the introduced arguments are influenced by the action of the main verb. In English, double object construction, as seen in (37) and (40), is used as applicative construction:. (41) a. Elmer baked Hortense a cake. (Marantz 1993). b. I wrote John a letter. (Pylkkanen 2002). Marantz explained that, in English, a double-object construction like (41a) is an applicative construction. He claimed that, in a double object construction like (41a),. 50.

(57) the benefactive is outside the event, affecting the theme, whereas the affectee is located inside the event on the specifier of the applicative phrase. In (41b), the benefactive object, ‘John,’ is above the original object, ‘the letter,’ leading us to believe that it is applicative construction as well. While many languages use the same applicative marker to indicate both affectees and benefactives, others use different applicative markers to differentiate between the two. If a language has only one applicative marker, the semantic meaning of the sentence can be used to distinguish between a benefactive applied object and an affectee applied object. If the applicative marker is removed from the sentence, the additional argument must be omitted concurrently, or the sentence will become ungrammatical. Only with the help of the applicative marker can the transitivity of a verb be intensified. Without the applicative marker, a verb can only bring the grammatically allowable number of arguments. Semantically speaking, applicative construction can be separated into high applicative and low applicative. Pylkkanen (2002) pointed out that high applicatives denote a relationship between an event and an individual, while low applicatives denote a relationship between two individuals. She has proposed that low applicative construction can only occur with a transitive verb. High applicative construction does. 51.

(58) not have such rules. High applicative construction can occur with various verb forms, despite its transitivity. The theory is reasonable because the high applicative denotes a thematic relationship between the event described by the verb and the applied object; therefore, the transitivity of the verb is irrelevant. On the other hand, a low applicative describes the relation between two individuals. Without a transitive verb, only one object will exist in the sentence:. (42) a. Mukasa. ya-som-e-dde. Katonga ekitabo. Mukasa 3G.PAST-read-APPL-past b. Mukasa Mukasa. ya-tambu-le-dde. Katonga. [Luganda]. book. Katonga. [Luganda]. PAST-walk-APPL-PAST. Mukasa walked for Katonga c. Kantonga Katonga. ya-kwaant-i-dde. Mukasa. PAST-hold-APPL-PAST. Katonga held the pot for Mukasa (43). Mukasa. ensawo. [Luganda]. bag (Pylkkanen 2002). a. I baked him a cake *b. I run him. In example (42), the transitivity of the verb does not influence the grammaticality of. 52.

(59) applicative construction. A multitude of verbs are compatible with applicative construction. Based on this reason, we confirm that Luganda has high applicative construction. In (42), because the transitive verb is only able to use applicative construction, to avoid making the sentence ungrammatical, we conclude that English has low applicative construction. If we discuss the syntactic structure of high and low applicative construction, we may find that the location in which each construction places its applicative argument is different. High applicative places its applicative argument out of the VP, like in (44a). Low applicative places its applicative argument inside the VP, like in (44b). The only similarity between the two is that no matter whether high applicative or low applicative is used—both of their applied objects must c-command the direct object.. (44a) high applicative. (44b) low applicative. 53.

(60) (Peterson 2007). In summary, applicative construction does exist and is applied widely in many languages. Although there may be different ways to present applicative construction, its purpose is always to intensify the transitivity of a verb and introduce an additional argument into a sentence. In the next section, the method of applying applicative construction and its function in the Taiwanese ka will be shown.. 3.2 The analysis: Two functions of ka In Section 3.1.1, we proposed that ka has two functions. One is as an applicative marker and the other is as a counterpart of the Mandarin ba. In this section, we start by introducing an analysis of the applicative ka construction. Then, we prove the other function of ka. We claim that ka is capable of being an applicative marker for introducing an additional argument into a sentence. Our proposal can be proven by the following sentences:. (45) a. i. chau2. a. He run PRT. 54.

(61) He ran away a’. i. ka. goa2 chau2 a. He ka. me ran PRT. He ran away (and makes me become an affectee) a’’* i. φ goa2. He φ b.. i. chau2 a. me. poan. run PRT i2-a2. He move chair He moved the chair b’ i. ka. goa2. He ka. me. poan move. i2-a2 chair. He moved the chair (and makes me become the benefactive) b’’* i. φ. He φ c. O-eng O-eng. goa2 me. poan. i2-a2. move chair ha7 pan. off the duty. O-eng is off duty c’ O-eng. ka. goa2. ha7 pan. O-eng. ka. me off the duty. 55.

(62) O-eng is off duty (and makes me become an affectee) c’’ O-eng O-eng. φ φ. goa2. ha7 pan. me off the duty. Chau2/run is a one-place predicate, as (45a) indicates. A one-place predicate should have only one argument in a sentence. In (45a’) chau2 is able to carry an additional argument goa2/me with the help of ka. If we omit ka and leave the additional argument in the sentence, the additional argument makes the sentence ungrammatical. The same phenomenon occurs in (45b). The main verb in (45b), poan/move, is a transitive verb. A transitive verb has only one object. An additional argument goa2 /me is introduced into the sentence with the help of ka. All of these phenomena explain one thing—the transitivity of a main verb becomes stronger with the addition of ka in a sentence. With the help of ka, we can correctly introduce an additional argument into an argument-saturated sentence. The phenomenon presented in (45) demonstrates our assertion－ka is capable of introducing an additional argument into a sentence and intensifying the transitivity of the main verb. The characteristics of introducing an additional argument to the sentence and intensifying the transitivity of the main verb are identical to those of an applicative marker. Thus, we suggest that ka can function as an applicative marker in. 56.

(63) Taiwanese. However, if ka is purely an applicative marker in Taiwanese, the origin of the ka-NP in sentences like ‘goa2 ka chhai3 be2 a/ I buy the vegetables’ and ‘i ka goa2 phah/ He hit me,’ is difficult to explain. The ka-NP in these sentences is obviously not an additional argument. It is the object of the main verb. In order to explain sentences like ‘goa2 ka tsa ibei a/ I buy the vegetables’ and ‘i ka goa2 poan i2-a2/ He moves the chair for me at the same time,’ we propose that ka has two syntactic functions when added to a sentence:. (a) To be the counterpart of the Mandarin ba6 (b) To be the applicative marker7. Function (a) operates when we present the disposal meaning. If function (a) is utilized, ka would render the main verb unable to assign the inherent Case to its object. According to Burzio’s generalization8, the main verb will lose its external subject.. 6. The reason that we do not treat the function (a) as the applicative construction is because that the relationship between the arguments does not change after we add this kind of ka into the sentence. 7 Tsai (2007) has pointed out that Taiwanese applicative ka construction is believed to be the counter part of Mandarin gei construction. Taiwanese applicative ka construction can be used to exhibit the usage of the affectee and the benefactive, which is very similar to the Mandarin gei construction. In this thesis we focus on the two functions of ka and the differences between these two functions. Detailed discussion about the relationship between ka and gei see Tsai (2007). 8 Burzio’s generalization: (i) A verb which lacks an external argument fails to assign accusative Case. (ii) A verb which fails to assign accusative Case fails to theta-mark an external argument. TA (Burzio’s 1986:185) T represents the external theta role, assigned indirectly. A stand for accusative Case.. 57.

(64) Because the object loses the Case given by the main verb, it is forced to move to the post-ka position and receive the Case given by ka. Due to the adjacency condition that ka-NP must follow so that the NP will recieve the Case from ka, the phenomenon of why in the ka construction, ka and ka-NP are always inseparable becomes explainable. The movement of ka-NP is because of its need for the Case. Why putting emphasis on ka keep the main verb from assigning the Case to its complement NP is because that only if the main verb fails to assign the Case to the object, we would be able to explain why, in sentences like ‘goa2 ka chhai3 be2 a/ I buy the vegetables’ and ‘i ka goa2 phah/ He hit me,’ the object needs to move to a position following ka. The phenomenon of an added element which makes the main verb lose its external argument and simultaneously, its ability to assign the inherent Case, is not an isolated occurence. We find similar circumstances in the passive form of English. Burzio’s generalization states that a verb in passive form loses its external argument and its ability to assign the Case to an object. The second function of ka in a sentence, as (b) presents, is to operate as an applicative marker. When ka is treated as an applicative marker, it introduces an additional argument into the sentence. Ka is responsible and necessary for grammatically introducing an additional participant into a sentence. When ka is added to the sentence and when an appropriate candidate for the additional argument is. 58.

(65) found in the lexical array, sentences with saturated arguments may add this additional argument after ka. It is not necessary for ka to have the main verb fail to assign the Case to the object. ka itself is already satisfied with an additional argument. Ka does not need to assign the Case to the object of the main verb. The specific phenomenon is called feature-checking. If we can satisfy the demand to assign the Case for ka under the local domain, ka would not need to take the strategy of making the main verb fail to assign the Case. Ka would instead move the verb-complement upward to do the feature checking. The applicative ka construction is believed to be similar to the Mandarin gei construction, for both of them are able to bring an additional argument to be an affectee or a benefactive9. There are two kinds of applicative constructions. One is high applicative construction, and the other is low applicative construction. We propose that the Taiwanese applicative ka construction is an example of high applicative. This is because the position of the Taiwanese applicative ka construction is located in a position higher than that of the VP. We use a manner adverb to prove our assumption:. (46) a. i. ka he ka. 9. goa2. kai. me. spend. ji money. For more detail discussion, see Tsai (2007).. 59.

(66) He spend money (and makes me an affectee) b. i. ka. he ka. goa2 me. luan. kai. without concern. spend. chiN5 money. He spends money without concern (and makes me an affectee) c.(?) i. luan. ka. He without concern ka. goa2 me. kai spend. chiN5 money. The structure of (46b) is presented as follows:. To confirm that the kaP is above the VP, we perform a test using a manner adverb. As (46b) presents, we find that the manner adverb always follows the kaP. The sentence will become less acceptable if we put the manner adverb before the kaP. Thus, the kaP should be located above the VP. Because of this conclusion, we adopt the structure that Li (2002) proposed: that the Taiwanese ka construction is, in fact, the applicative. 60.

(67) ka construction. Because ka functions as the theta-role assigner, as Li mentioned, we keep its light verb structure on its construction. Owing to the ka construction being at a position higher than the VP, we conclude that the Taiwanese ka construction belongs to high applicative construction. Another reason which leads us to our conclusion, is the semantic meaning of ka. The semantic meaning exhibited by the Taiwanese applicative ka construction is always found between an event and an individual introduced by ka; not between two individuals:. (47) a. i. ka. He ka. goa2. be2. me. buy. chheh books. He helps me to buy the book b. Abu. ka. Mother ka. goa2. se2. saN. me wash clothes. Mother helps me to wash my clothes c. goa2. ka. Me ka. i him. poan. i2-a2. move chair. I help him to move the chair.. 61.

(68) In (47), the relationship between the applicative ka-NP and the event described by the verb confirms the high applicative construction we mentioned. In (47a), goa2 is the benefactive of the event ‘i be2 chheh/He buys the book.’ In (47b), goa2 is the benefactive of the event ‘Abu se2 saN/ Mother washes clothes.’ In (47c), i is the benefactive of the event ‘goa2 poan i2-a2/ I moved the chair.’ The applicative ka NP is always influenced by the whole event described by the main verb. Following the definition proposed by Pylkkanen, the high applicative always introduces the relationship between an event and an individual. This confirms that the Taiwanese applicative ka construction belongs to high applicative construction. Also, the Taiwanese applicative ka construction is able to combine with a unergative verb. Following Pylkkanen’s definition, only high applicatives can combine with unergative verbs.. (48) a. he le phaiN2-lang5 That bad guy. keng3-jian5. ka goa2 si2-a. unexpectedly. ka me die. That bad guy die unexpectedly (and makes ‘me’ an affectee) b. i. ka. He ka. goa2. chau2-a. me run-PRT. He runs (and makes ‘me’ an affectee). 62.

(69) Because the Taiwanese applicative ka construction always presents the semantic relationship between an event and an individual, and its structural location is above the VP, we find that it belongs to the high applicative. Owing to the conclusion that the Taiwanese ka applicative construction belongs to high applicative construction, no specific restriction on the verb selection exists. If only a qualified argument exists in the lexical array, the applicative ka construction would be allowed to produce a multitude of verbs. In summary, the function of ka, wherein the main verb loses its external argument and fails to assign the Case to the object, would not work if we aim to present applicative construction. On the contrary, if we wish to present a disposal meaning like the Mandarin ba construction, ka will take the object of the main verb as its argument. To determine whether or not the result of the ka NP is NP movement or applicative construction is quite easy: If we remove ka and the following ka NP, and the sentence is still grammatical, it is an applicative ka construction10. Take ‘i ka mih8 a2 kun hoo a/He ka things put already’ as an example. The original sentence should be ‘i kun hoo mih8 a2 a/He put things already.’ This is a 10. Beyond the examples presented above, the other kind of applicative ka construction is discovered, for example: i ka goa2 chin san/ He helped me to dress up. Unlike the examples presented above, if the ka+NP had been deleted, although the sentence remains grammatical, the meaning of the sentence will become totally different. Unlike the applicative ka construction we have introduced above, this kind of applicative ka construction may be considered as the mid applicative (Tsai 2007). For the more detail discussion, see Tsai (2007).. 63.