漢語動補結構中的致使義：從論元體現而論 - 政大學術集成

全文

(1)國立政治大學語言學研究所碩士論文. National Chengchi University Graduate Institute of Linguistics Master Thesis. 指導教授：何萬順 Advisor: One-Soon Her. 立. 政治大. ‧. ‧ 國. 學. 漢語動補結構中的致使義：從論元體現而論. y. Nat. io. sit. Causativity in Chinese Resultative Compounds:. n. al. er. On the Account of Argument Realization. Ch. engchi. i n U. 研究生：黃心綸撰 Student: Hsin-Lun Huang 中華民國一百零一年六月 June, 2012. v.

(2) . Causativity in Chinese Resultative Compounds: On the Account of Argument Realization. 學. ‧ 國. 立. 政治大 BY. ‧. Hsin-Lun Huang. er. io. sit. y. Nat. n. a lA Thesis Submitted to the i v n C hInstitute of Linguistics U Graduate i engch In Partial Fulfillment of the Requirements for the Degree of Master of Arts. June 2012 .

(3) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(4) . 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. Copyright © 2012 Hsin-Lun Huang All Rights Reserverd. iii . v.

(5) . Acknowledgements. 又到了鳳凰花開的季節，正所謂天下無不散的筵席，該來的還是要來，也輪到了我踏出這校門的時候，想當初我剛進這台灣數一數二高的學府時，還是個沒沒無名的小嫩咖，多虧有了英明神武的何萬順老師的加持，讓我在句法研究的功力上大為精進，也是在何老師的調教下，腦袋才有了真正在運轉的感覺，再加上兩大句法師尊，張郇慧老師及謝富在老師的多方心訣傳授，我在語言學研究所的最終祕笈，也就是此本記載精闢分析的論文，才得以降生於世。又所謂師者，傳道、授業、解惑也。語言所掌門人─徐嘉慧所長、語言所仙女姊姊─黃瓊之老師、季陶樓扛霸子─蕭老師宇超老大、客家大仙─賴惠玲老師、及字根字首之神─莫建清老師，所賦予學生我的語言學絕學更是有如滔滔江水，連綿不絕，又有如黃河氾濫，一發不可收拾，學生所得到的收穫，都滿滿滿，滿了出來！除了上述稱霸語言所四方的天王巨頭外，語言所十八銅人，對於小弟的提攜照顧，也可說是功不可沒！帶頭的當然不能不提地震不垮、風吹不倒、古道熱腸、英氣逼人的助教惠鈴學姊了，有了學姊的庇蔭，小弟就是天塌下來，也可以面不改色，輩分同樣尊貴的，還有人稱客家雙姝的詩敏學姊及秋杏學姊，他們對於小弟的各方知識授受及語言所生活提點，上至天文，下通地理，可說是傾盡全力、醍醐灌頂！小弟在見到他們尊駕時，可也是要稱一聲詩敏姊及秋杏姊的。接著的，是同甘共苦的生死夥伴，大家一起從小嫩咖升級成老油條，這過程中的艱辛，自然是不言而喻的了；讓我最心存感激的，絕對是同門正妹，琬婷姊了，因為你太正，轉移了老師的注意，才讓我可以一帆風順地在你的掩護下飛向論文的終點，掌上明珠柏溫兄在客家庄所受到的崇敬已不可同日而語，侃彧及美杏也接續了仙女姊姊的後塵，誓不成為語言習得研究雙塔不休，書豪旋風席捲了音韻扛霸子宇超老大的地盤，以這態勢，未來音韻頭目的地位絕對非書豪兄莫屬，98 級三仙女─晉瑋、媛媜、曉貞則持續以你們的美貌普渡眾生，有你們的保庇，大家才能開心渡過苦悶的研究所閉關生活，至於油滋滋瑪莉姊 ─姿幸，還是一樣的嫩，不過用美妝照騙騙路人還是有一套的，諸位大德光芒四射，八仙過海也不過如此！在小弟人生的一頁，有各位璀璨的記錄著實讓這三年的研究生涯不虛此行，古人有云：錢財乃身外之物，生不帶來、死不帶去。和各位寶貴的回憶，卻是小弟人生莫大的資產，走到哪帶到哪！. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. iv . i n U. v.

(6) . Table of Contents. Acknowledgements .............................................................................. iv Chinese Abstract ................................................................................. vii. 政治大 English Abstract .................................................................................. ix 立. ‧ 國. 學. Chapter. ‧. 1. Introduction ...................................................................................... 1. sit. y. Nat. 2. Literature Review ............................................................................. 5 2.1 Derivational Account ................................................................................... 5. io. n. al. er. 2.1.1 Categorizations of Chinese Resultative Compounds ....................... 6 2.1.2 Syntactic Constraints on Compound Interpretations ..................... 9 2.1.3 Derivational Operations on Chinese Resultative Compounds ..... 13. Ch. engchi. i n U. v. 2.2 Lexicalist Account ...................................................................................... 21 2.3 Availability of the Accounts to Chinese Resultative Compounds .......... 27 2.3.1 Insufficiencies in the Derivational Account .................................... 27 2.3.2 Predictability of the Lexical Functional Grammar Account ......... 35. 3. Theoretical Background ................................................................. 41 3.1 Lexical Mapping Theory (LMT) ............................................................... 41 3.2 Intrinsic and Default Morphosyntactic Classification of Argument Roles (IC & DC) ................................................................................................... 44 3.3 The Unified Mapping Principle (UMP) ................................................... 45 v .

(7) . 3.4 Causative Assignment in Resultative Compounding .............................. 47. 4 Causativity in Chinese Resultative Compounds .......................... 53 4.1 Internal vs. External Causativity .............................................................. 53 4.2 Morpholexical Fusion of Chinese Causative Verbs and Resultative Compounds ................................................................................................. 58 4.2.1 Romance Causatives and Chinese Causative Resultative Compounds ........................................................................................ 62 4.2.2 Verbs Allowed in the Causative Resultative Compounds ............. 73 4.3 Newly Coined Resultative Compounds .................................................... 83. 政治大. 5 Argument Realizations of Chinese Resultative Compounds ...... 87. 立. 5.1 Argument-Function Remapping ............................................................... 87. ‧ 國. 學. 5.2 Result as “Extent” ...................................................................................... 95 5.3 Dual Status of V2 in Chinese Resultative Compounds ......................... 100. ‧. 6 Conclusion ..................................................................................... 107. sit. y. Nat. io. n. al. er. References .......................................................................................... 111. Ch. engchi. vi . i n U. v.

(8) . 國. 立. 政. 治. 大. 學. 研. 究. 所. 碩. 士. 論. 文. 提. 要. 研究所別：語言學研究所論文名稱：漢語動補結構中的致使義：從論元體現而論指導教授：何萬順. 教授. 研究生：黃心綸. 政治大論文提要內容：(共 1 冊，19,516 字，分 6 章 13 節) 立. ‧ 國. 學 ‧. 漢語的動補結構長久以來一直是漢語語言學中一個複雜難解的議題，動補結. sit. y. Nat. 構中的論元在句法上應如何體現及其相對應的語意應如何解讀更是許多語言學. n. al. er. io. 家致力解決的問題。衍生語法學派 (Derivational Grammar) 的學者 (Cheng and. Ch. i n U. v. Huang, 1994; Cheng et al, 1997; Huang, 2006, 2007; Li, 1995) 及詞彙功能語法學. engchi. 派 (Lexical Functional Grammar) 的學者 (Her, 2004, 2007, 2009; Shibagaki, 2009) 均曾試圖提出對於漢語動補結構最正確有效的分析，本論文旨在證明衍生語法學派之理論在動補結構分析上的錯誤及不足，並點出詞彙功能語法學派之理論在預測動補結構的論元體現 (argument realization) 及語意解讀 (semantic interpretation) 上有較全面的分析。然而，某些動補結構，如「這一大桌子菜胖死我了」，含有不同於其他動補 vii .

(9) . 結構的致使義 (Causativity)，進而產生特殊的論元體現模式，此情形卻未被涵蓋在詞彙功能語法學派的理論分析中，本論文也依循詞彙功能語法的理論框架，試圖提出足以預測此種特定動補結構論元體現的句法規則，以彌補詞彙功能語法理論在分析此動補結構的論元體現上之遺漏，使其整體分析更臻完善。 . 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. viii . i n U. v.

(10) . Abstract. Resultative compounds in Mandarin Chinese have long been a complicated issue in Chinese linguistics. Many researchers have put great effort in trying to solve the problem of how arguments of Chinese resultative compounds should be. 政治大 syntactically realized and how corresponding interpretations should be achieved. 立. ‧ 國. 學. Researchers of Derivational Grammar (Cheng and Huang, 1994; Cheng et al, 1997;. ‧. Huang, 2006, 2007; Li, 1995) and those of Lexical Functional Grammar (Her, 2004,. sit. y. Nat. 2007, 2009; Shibagaki, 2009) have all attempted to propose analyses that are accurate. n. al. er. io. in predicting the argument realization and compound predication of Chinese. Ch. i n U. v. resultative compounds. This thesis aims to prove the insufficiency in the prediction. engchi. power of derivational accounts and endeavors to show that the account of Lexical Functional Grammar is superior in the analysis of resultative compounds in terms of argument realization and compound interpretation. However, some resultative compounds, such as that in zhe yi da zhuozi cai pang-si wo le (‘This whole table of dishes is making me fat.’), have certain causativity that makes them different from others in terms of argument realization. Resultative ix .

(11) . compounds like these are not included in the analysis of lexicalist accounts. Based on the theoretical framework of Lexical Functional Grammar, this thesis also makes an attempt to propose a syntactic rule that predicts the correct argument realization pattern of the kind of resultative compounds mentioned above. It is the goal of this thesis that the proposed rule covers the analysis of the resultative compounds that is. 政治大 missing in previous accounts and makes the lexicalist account given in this thesis a 立. ‧ 國. 學. better solution in working on the issues of Chinese resultative compounds. . ‧. n. er. io. sit. y. Nat. al. Ch. engchi. x . i n U. v.

(12) . Chapter 1 Introduction. The complexity of resultative compounds in Mandarin Chinese has been a widely investigated issue in the course of linguistic research. Argument realization of Chinese resultative compounds has been approached by researchers either on the derivational account (Cheng and Huang, 1994; Cheng et al, 1997; Huang, 2006, 2007; Li, 1995) or the lexicalist account (Her, 2004, 2007, 2009; Shibagaki, 2009). How the. 政治大 arguments are linked to their grammatical functions and how causativity in resultative 立. ‧ 國. 學. compounds is achieved remain irresolute in the debates of researchers.. ‧. On the derivational account, resultative compounds with various argument. sit. y. Nat. structures are categorized according to the syntactic nature of the predicates of the. n. al. er. io. compounds (whether the predicates are unergative, transitive, unaccusative or. Ch. i n U. v. causative). The syntactic nature of the predicates gives rise to different argument. engchi. realizations and different interpretations of the compounds. In Li’s (1995) analysis, an example of the multiple interpretations is recorded:. (1) Zhangsan zhui-lei-le John. chase-tired-ASP1. Lisi. (transitive). Lee.                                                         1. ASP stands for aspect marker. For future reference, CL= classifier. CAU= causative verb. DE= relative marker. 1 .

(13) 2 . a. ‘John chased Lee to the extent of making him (Lee) tired.’ b. *‘Lee chased John and he (John) got tired.’ c. ‘John chased Lee and (John) got tired.’ d. ‘Lee chased John and was made tired (by John).’. Her (2004, 2007) further pursues this interpretation complexity by resorting to Lexical Functional Grammar (LFG) and claims that the different grammatical realizations of the arguments are due to the lexical mapping competition between the composite roles. 政治大. in the resultative compounds. Interpretation (1d) in his analyses is proved to be most. 立. marked and hardly accessible to most Chinese speakers. This thesis aims to account. ‧ 國. 學. for the causative reading of this interpretation in a direction similar to but slightly. ‧. different from that taken by Her (2004, 2007).. y. Nat. al. er. io. sit. This thesis starts by retrospecting the derivational analyses of Chinese. n. resultative compounds. The categorizations of the compound predicates are. Ch. engchi. i n U. v. demonstrated for a clearer view on how syntactic nature of the predicates is related to their argument realizations. The predication of the verb components of the compounds is also given to show the achievement of the compounds’ interpretations. It is proposed that several syntactic constraints are in place to regulate the argument realization; moreover, causativity, evident in some of the resultative compounds, is proved to be acquired by means of derivational operations that involve a light verb with a causative meaning. .

(14) 3 . A lexicalist approach follows the derivational approach in the next chapter for an alternative view on how Chinese resultative compounds can be analyzed. The lexicalist approach taken is Lexical Functional Grammar (LFG). Argument realization and causativity achievement in resultative compounds are also accounted for in this approach. The thesis aims to argue that the LFG approach possesses more advantages in the resolution of Chinese resultative compounds. Adopting the formalism of LFG, this thesis describes in Chapter 3 the theoretical framework of explaining how. 政治大 causativity is manifested in resultative compounds. Chapter 4 demonstrates the 立. ‧ 國. 學. analysis of causativity and its related argument structure of the compounds. A. ‧. morpholexical rule of how causativity arises and its relation to Chinese resultative. sit. y. Nat. compounds is given in this chapter. Chapter 5 contains the operations of argument. n. al. er. io. realization in resultative compounds and further supports the legitimacy of the. Ch. i n U. v. analysis in the present study. Moreover, an investigation of the identity of the V2s in. engchi. resultative compounds is also conducted for the confirmation of the legitimacy. Finally, Chapter 6 concludes the thesis.. .

(15) . 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. . i n U. v.

(16) . Chapter 2 Literature Review. 2.1 Derivational Account The issues of the argument structure of Chinese resultative constructions have been widely discussed by many researchers on the account of derivational grammar. Explanations concerning how the arguments of Chinese resultative constructions are syntactically realized are given with respect to the categorizations of the predicates in. 政治大 the constructions and certain syntactic constraints that regulate the linking between 立. ‧ 國. 學. the arguments and grammatical functions (Cheng and Huang, 1994; Huang et al, 1997;. ‧. Huang, 2006; Wang & He, 2002). In Cheng and Huang (1994), the verbs of. y. sit. n. al. er. io. the verbs:. Nat. resultative constructions are categorized as follows based on the syntactic nature of. (2) a. Zhangsan. Ch. qi-lei-le.. engchi. i n U. v. (unergative). Zhangsan ride-tired-ASP 'Zhangsan rode himself tired.'. b. Zhangsan. qi-lei-le. liang-pi. ma.. Zhangsan ride-tired-ASP two-CL horse 'Zhangsan rode two horses tired.'. 5 . (transitive).

(17) 6 . c. Zhangsan. qi-si-le.. (ergative). Zhangsan anger-dead-ASP 'Zhangsan got extremely angry.' '(Lit.) Zhangsan was angered to death.'. d. zhe-jian shi. zhen. qi-si. Zhangsan le.. (causative). this-CL matter really anger-dead Zhangsan ASP ‘This matter really angered Zhangsan.' '(Lit.) This matter really angered Zhangsan to death.'. 立. 政治大. ‧ 國. 學. As can be observed above, Cheng and Huang (1994) argue that the unergative V1, qi, in the resultative compound in (2a) can indeed demonstrate a transitive use with the. ‧. difference that the resultative state of the compound denoted by V2, lei, is predicated. sit. y. Nat. io. n. al. er. on the subject in (2a) but on the object in (2b). This is referred to as the. i n U. v. unergative-transitive alternation. On the other hand, the ergative V1, qi, in (2c). Ch. engchi. demonstrates an ergative-causative alternation where, in addition to the internal argument that is already assigned by qi, an external argument not assigned by any of the predicates in the compound can be inserted.. 2.1.1 Categorizations of Chinese Resultaive Compounds The categorization of Chinese resultative compounds in (2) is based on the first. .

(18) 7 . verb of the compound. The internal structure of resultative compounds has been widely analyzed as consisting of two verbs, where the first verb indicates an action or a state while the second verb indicates the result or change of state. In Cheng and Huang (1994), resultative compounds are indeed grouped into different classes in terms of two dimentions, aspectuality and transitivity. Therefore, a detailed description of the categorization above can be made on the account of the transitive and aspectual nature of the first predicate of the compound.. 政治大 With first predicates (henceforth, V1) that are intransitive, a more fine-tuned 立. ‧ 國. 學. classification sets in under the assumption of the Unaccusative Hypothesis proposed. ‧. by Perlmutter (1978). The Unaccusative Hypothesis assumes that one-placed. sit. y. Nat. predicates should be classified as unaccusatives and unergatives with the difference. n. al. er. io. that unaccusatives have their arguments derived from an underlying deep-structure. Ch. i n U. v. with a grammatical object but no subject while the unergatives have their arguments. engchi. derived from an underlying deep-structure with a grammatical subject but no object. As a result, under the assumption of the Unaccusative Hypothesis and Cheng and Huang’s (1994) sorting criteria (i.e. aspectuality and transitivity), the categorization of Chinese resultative compounds as that in (2) is formed2. However, some parallel relationships seem to exist between the categories of the compounds if a                                                         2. In (2c), the predicate is termed ergative by Cheng and Huang (1994). However, in Huang’s other works, he alternatively terms the predicates unaccusative. If under the syntactic definition of unaccusative and ergative, they practically mean the same thing. .

(19) 8 . closer inspection of their syntactic behavior is made:. (3) a. Zhangsan ku-lei-le.. (unergative). Zhangsan cry-tired-ASP ‘Zhangsan cried himself tired.’ b. Zhangsan. ku-lei-le. Lisi.. (transitive). Zhangsan cry-tired-ASP Lisi ‘Zhangsan cried and made Lisi tired.’. 政治大. 立. It is clear to see that the predicates of the compounds in (3a) and (3b) are the same.. ‧ 國. 學. Syntactically speaking, if unergative predicates are deemed to have only grammatical. ‧. subjects but no objects in D-structure, the object position is left open for them to carry. y. Nat. al. er. io. sit. an object as their internal argument and thus demonstrate transitivity. This is how the. n. unergative-transitive alternation in Cheng and Huang (1994) comes into place.. Ch. engchi. i n U. v. Likewise, since the unaccusative predicates in the compounds are deemed to have only grammatical objects but no subjects in D-structure, there exists a possibility for them to carry a subject as their external argument. But this time, things are different with respect to their unergative counterparts. The external argument comes with the effect of causativity for it plays the role Causer and causes the event depicted by the unaccusative predicates:. .

(20) 9 . (4) a. Zhangsan. lei-si-le.. (unaccusative). Zhangsan tired-dead-ASP ‘Zhangsan is really tired.’ b. zhe-ge gongzuo this-CL job. lei-si. Zhangsan le.. (causative). tired-dead Zhangsan ASP. ‘This job made Lisi really tired.’. This is what is termed unaccusative-causative alternation in Cheng and Huang (1994).. 政治大. It is worth noticing that despite both being transitive, sentences (3b) and (4b) show. 立. one difference that the subject in (4b) actually causes the event of Zhangsan being. ‧ 國. 學. tired while the subject in (3b) does not cause the event of Lisi crying.. ‧ y. Nat. al. er. io. sit. 2.1.2 Syntactic Constraints on Compound Interpretations. n. With the syntactic structures of Chinese resultative compounds sorted out, it is. Ch. engchi. i n U. v. now time to see how the interpretations of the compounds are achieved in terms of their syntactic structures. Supported by the evidence from English, a constraint is proposed by Simpson (1983) to explain how resultative constructions are to be interpreted based on which predicate should be predicated on which argument:. (5) a. John hammered the metal flat. b. *John hammered the metal tired. (under the resultative reading) .

(21) 10 . Due to the ungrammaticality of (5b), Simpson proposes the Direct Objet Restriction (DOR), which states that the Result of resultative constructions should be predicated on the direct object of the sentence:. (6) The Direct Object Restriction (Simpson, 1983): In a resultative construction, the result is predicated on an object, not the subject.. 立. 政治大. At first glance, the DOR seems right for resultative constructions. But what happens if. ‧ 國. 學. Chinese resultative compounds are put into consideration? The interpretation of. ‧. Zhangsan riding two horses and making the horses tired in (2b) is rendered by the. Nat. io. sit. y. DOR. However, any native speaker of Chinese can tell that there are indeed more. al. er. interpretations to it besides that glossed in (2b). For one, (2b) can mean Zhangsan. n. v i n C h the Result of theUresultative compound is rode two horses and got tired, where engchi predicated on the subject.. Another problem to the DOR is that the DOR implies that an object is required in resultative constructions since the result of the constructions has to be predicated on the object. English resultative constructions would be rendered ungrammatical without an object. Compare examples (7) and (8):. .

(22) 11 . (7) a. *John laughed silly. b. *Mary ran tired.. (8) a. John laughed himself silly. b. Mary ran herself tired.. In order to avoid being filtered out by the DOR, the ‘fake’ reflexives have to be inserted for the predication of the result. Things are completely different in Chinese. 治政大 nature of the predicates resultative constructions. The unergative and unaccusative 立 ‧ 國. 學. shows that intransitive use of resultative constructions, where the subject-predication of the result is the only option, is absolutely grammatical.. ‧. To solve the problems posed by the DOR, Rappaport Hovav and Levin (2001). sit. y. Nat. io. n. al. er. propose two explanations. In situations where an object is missing, the resulative. i n U. v. constructions are claimed to have a simplex underlying event structure, i.e. the. Ch. engchi. sub-events denoted by the predicates of the constructions are temporally and spatially coextensive and are treated as a simple event:. (9) A man gabbed and groped her…, but she kicked free and fled.. In (9), since the moment of her kicking is the moment of her being free, the resultative construction is taken to be a simple event that functions intransitively. .

(23) 12 . The second explanation for the subject-predication of the Result in Chinese transitive resultative compounds are proposed by Rappaport Hovav and Levin (RHL) (2001) and dubbed by Cheng and Huang (1994) the Force Recipient Principle (FRP):. (10) The Force Recipient Principle (paraphrasing RHL 2001): a. In a resultative construction, the Result is predicated on the argument bearing the role of Force Recipient of the relevant action or event, if such. 政治大 If no Force Recipient exists, the Result is predicated on the subject. 立 a recipient exists.. 學. ‧ 國. b.. It seems to have successfully explained the transitivity and Result predication. ‧. differences between English and Chinese. However, Cheng and Huang (1994) have. sit. y. Nat. io. n. al. er. discovered that the first explanation of English unergative predicates being analyzed. i n U. v. as having a simplex event structure is insufficient to account for Chinese unergative. Ch. engchi. resultative compounds for many of them are composed of predicates that do not denote temporally and spatially coextensive events. Take (2a), Zhangsan qi-lei-le (Zhangsan rode himself tired.), for example, it is generally conceived that the point of Zhangsan being tired only occurs some time after the point at which the action of riding starts. Cheng and Huang (1994) see it more like an analogy of the use of English unaccusative predicates:. .

(24) 13 . (11) The river froze solid.. (unaccusative). The point of the river being solid happens some time after the point the river starts to freeze. Therefore, Cheng and Huang (1994) hold the idea that Chinese unergative resultative compounds “may be alternatively analyzed as unaccusatives.” As for the second explanation, the default of subject being the target of Result predication when Force Recipient is missing is called into question by Cheng and. 政治大. Huang (1994) since what makes the subject the target is left unclear. Cheng and. 立. Huang state that it is argument prominence that determines Result predication. In their. ‧ 國. 學. account, the Result of resultative constructions should be predicated on the closest. ‧. prominent argument. As a result, if the prominent argument, Force Recipient is not. y. Nat. al. er. io. sit. realized in the syntactic structure of the constructions, the subject becomes the most. n. prominent on which the Result is predicated. Although it seems to have successfully. Ch. engchi. i n U. v. explained the Result predication on Subject in example (11), the cases of Chinese resultative compounds, where the Result could be predicated on both Object and Subject and lead to semantic ambiguity, remain unsolvable under the explanation.. 2.1.3 Derivational Operations on Chinese Resultative Compounds Recall the idea Cheng and Huang hold that some Chinese unergative resultative compounds “may be alternatively analyzed as unaccusatives.” Under this claim, in .

(25) 14 . addition to the unergative-transitive alternation and the unaccusative-causative alternation, another compound alternation can be formed; that is, unergative-causative alternation because if the unergative predicates are analyzed unaccusatively, they can alternate with the causative structure like genuine unaccusatives:. (12) a. Zhangsan chang-lei-le.. (unergative). Zhangsan sing-tired-ASP. 政治大. 'Zhangsan sang himself tired.'. 立. b. Zhangsan chang-lei-le. na-shou ge.. ‧ 國. 學. Zhangsan sing-tired-ASP that-CL song. (transitive). ‘Zhangsan sang that song and made himself tired.’. ‧. c. na-shou ge chang-lei-le. Zhangsan.. sit. y. Nat. that-CL song sing-tired-ASP Zhangsan. n. er. io. ‘Singing that song made Zhangsan tired.’. al. (causative). Ch. engchi. i n U. v. This pattern of causativization is also found in the way English encodes causativity in its predicates. Basically, causativity is conveyed by means of periphrastic causatives, where an apparent verb marking the transmission of force from the causer to the causee is explicit and the causee ends up with a change of state, as in (13):. (13) John made the vase broken.. .

(26) 15 . “Make” is the verb that indicates the transmission of force from John to the vase. However, the transmission of force could be encoded in the predicates without apparent causative verbs. Such predicates are termed “lexical causatives” (Fodor, 1970; McCawley, 1978; Comrie, 1985; Jackendoff, 1990). In most cases, an alternation of argument realization exists in lexical causatives and is deemed the causative-inchoative3 alternation:. 政治大. (14) John broke the vase.. (lexical causative) (inchoative). ‧. ‧ 國. 學. The vase broke.. 立. In (14), ‘broke’ is a lexical causative in which causativity is encoded. Its difference. y. Nat. al. er. io. sit. with the inchoative ‘broke’ lies in that the subject of the inchoative ‘broke’ is the. n. object of the lexical causative ‘broke’. The unaccusative-causative alternation and the. Ch. engchi. i n U. v. unergative-causative alternation in Chinese resultative compounds are in fact inchoative-causative alternation in this respect. Traditionally, the derivational analysis of Chinese lexical causatives is that they are syntactically derived from the combination of periphrastic causatives and inchoatives (Wang and He, 2002):.                                                         3. Without clearly indicating the force of causation, the inchoatives simply denote the change of state of the argument. .

(27) 16 . (15) a. na-ben shu. shi. [wo gandong-le.]. (periphrastic causative). that-CL book CAU me touched-ASP ‘That book touched me.’ b. na-ben shu. V[CAUSE]. c. na-ben shu. gandong-le-V[CAUSE]. [wo gandong-le] [wo. t]. (lexical causative). Lexical causatives are analyzed as having a similar D-structure to that of periphrastic. 政治大. causatives. However, the head position of the VP is occupied by a phonologically null. 立. causative verb instead of a substantial one. The predicate in the embedded clause is. ‧ 國. 學. later moved to V for pronunciation factors and hence has the causative reading. In the. ‧. same logic, the causative categories of Chinese resultative compounds are claimed to. Nat. n. al. er. io. sit. y. demonstrate the same derivation from their unergative and unaccusative counterparts:. (16) a. ta. he-zui-le. Ch. he drink-drunken-ASP. engchi. i n U. v. (unergative). ‘He drank and was drunk.’ b. na-ping. jiu. V[CAUSE]. [ta he-zui-le]. that-bottle wine ‘That bottle of wine’ c. na-ping jiu. . he-zui-le-V[CAUSE]. [ta. t]. (causative).

(28) 17 . (17) a. ta. xia-hun-le. (unaccusative). he shock-faint-ASP ‘He was shocked and fainted.’ b. zhanglang. V[CAUSE]. [ta xia-hun-le]. cockroach ‘cochroaches’ c. zhanglang. xia-hun-le -V[CAUSE]. [ta. t]. (causative). 治政大the derivational analysis Based on the categorization of compound predicates and 立 ‧ 國. 學. mentioned above, Huang (2006) further provides templates for all possible syntactic structures of Chinese resultative compounds as follows:. ‧ sit. y. Nat. io. er. (18) Huang (2006):. A. Inchoative (1): [BECOME<UNERGATIVE> [ x <STATE>]]. n. al. Zhangsan. Ch. zhui-lei-le.. Zhangsan chase-tired-ASP. engchi. ‘Zhangsan got tired from chasing.’. . i n U. v.

(29) 18 . VP1 DP. V1’ V1. V2P DP. V2’. [BECOME] <UNERGATIVE>. V2. zhui-lei-le t chase-tired-ASP x Zhangsan Zhangsan. <STATE>. 政治大. 立. ‧ 國. 學. B. Inchoative (2): [BECOME<UNACCUSATIVE> [ x<STATE>]] Zhangsan lei-bing-le.. ‧. Zhangsan tired-sick-ASP. VP1. n. al. Ch. DP. er. io. sit. y. Nat. ‘Zhangsan got sick from exhaustion.’. e n g cV1’ hi V1. i n U. v. V2P DP. V2’. [BECOME] <UNACCUSATIVE>. lei-bing-le t tired-sick-ASP x Zhangsan Zhangsan. . V2 <STATE>.

(30) 19 . C. Pure causative (1): [ x CAUSE [BECOME<UNERGATIVE> [ y <STATE>]]] baozhi. kan-hua-le wo-de yanjing.. newspaper read-blur. my. eyes. ‘The newspaper made my eyes blurred from reading it.’ VP1 DP. V1’ V1. V2P. [CAUSE]. 立. DP 治 V2’ 政大. ‧ 國. V2. VP3. 學. kan-hua-le read-blur-ASP. DP. V3’. [BECOME] t. y. <STATE>. sit. io. n. al. V3. er. y wo-de yanjing my eyes. Nat. x baozhi newspaper. ‧. <UNERGATIVE>. Ch. engchi. i n U. v. D. Pure causative (2): [ x CAUSE [BECOME<UNACCUSATIVE> [ y <STATE>]]] wu xiuzhi de jianku gongzuo lei-bing-le no rest DE hard work. Zhangsan.. tire-sick-ASP Zhangsan. ‘Continuous hard work got Zhangsan sick from over-exhaustion.’. .

(31) 20 . VP1 DP. V1’ V1. V2P DP. V2’. [CAUSE] lei-bing-le tired-sick-ASP. V2. VP3 DP. V3’. [BECOME] <UNACCUSATIVE>. 立. y 治 t 政大 Zhangsan. <STATE>. Zhangsan. ‧. ‧ 國. 學. x wu xiuzhi de jianku gongzuo continuous hard work. V3. E. Causing with a manner: [ x CAUSE<UNERGATIVE> [BECOME [ y <STATE>]]]. y. qiuxie.. sit. ti-po-le. Nat. Zhangsan. n. al. ‘Zhangsan kicked the sneakers thread-bare.’. . Ch. engchi. er. io. Zhangsan kick-broken-ASP sneakers. i n U. v.

(32) 21 . VP1 DP. V1’ V1. V2P DP. V2’. [CAUSE] <UNERGATIVE>. V2. ti-po-le kick-broken-ASP. VP3 DP. V3’. [BECOME] x Zhangsan Zhangsan. 立. V3. y 治 t 政大 qiuxie. <STATE>. sneakers. ‧. ‧ 國. 學. With the derivational templates and the syntactic constraints that regulate the. y. Nat. al. er. io. sit. predications of the verb components in resultative compounds, the argument. n. realizations and interpretations of the compounds should be comprehensively. Ch. engchi. i n U. v. accounted for. However, problems arise in this methodology of analysis. It undergenerates and the causativity of resultative compounds described in it does not conform to the interpretations of the compounds. These problems will be discussed in later chapters.. 2.2 Lexicalist Account The issues of causativity and argument realization of Chinese resultative .

(33) 22 . compounds have also been approached from a lexicalist point of view. But before moving to the issues of resultative compounds, how the lexicalist approach treats the relationship between argument structure and grammatical functions is to be made clear. Within the framework of Lexical Functional Grammar (LFG) raised by Bresnan (1990, 2001), the diverse constituent structures (c-structures) of different languages are not derived from a unified underlying structure, unlike the traditional view taken by Generative Grammar like Chomsky’s (1981, 1995) Government and Binding (GB). 政治大 or the Minimalist Program (MP). No syntactic derivations are involved in the 立. ‧ 國. 學. structures of languages, be it a hierarchical language like English or a flat language. ‧. like Warlpiri4 (Bresnan, 2001). The marking of constituency in flat languages is. sit. y. Nat. indeed a lexical process. Since no syntactic derivation is involved in languages, what. n. al. er. io. constituents appear in what grammatical functions is treated as a lexical relational. Ch. i n U. v. issue in LFG. For example, passivization is seen as a syntactic transformation caused. engchi. by the search of Case in Generative Grammar (Burzio, 1986). However, in LFG, it is more of a lexical operation that involves the mappings of arguments to different grammatical functions. Also a relational lexical operation, causativization could be analyzed similarly. Through the process of causativization, the causativized predicate may have its                                                         4. Warlpiri is an indigenous language in Australia. The language has a ‘flat’ constituent structure, where the constituents in the sentence can occur in any order as long as the auxiliary tense marker occurs in the second position. .

(34) 23 . arguments end up in different grammatical functions as opposed to its active counterpart. It is along this line of reasoning that Chinese resultative compounds are inspected by many researchers (Her, 2004, 2007, 2009; Li, 1995; Shibagaki, 2009). Not all of the Chinese resultative compounds are related to causativity; for those that do incorporate causativity in them, researchers have tried to capture the causative reading in the resultative compounds by various means. Li (1995), as one of the researchers, has proposed the concept of Causative Hierarchy to resolve the intricate. 政治大 interpretations of the resultative compounds in question. 立. ‧ 國. 學. The resultative compound of this kind first documented in Li’s (1995) analysis. ‧. was zhui-lei (chase-tired). Three possible interpretations could result from the. n. al. er. io. sit. y. Nat. sentence ‘Zhangsan zhui-lei-le Lisi’:. (19) Zhangsan zhui-lei-le John. Ch. Lisi. engchi. chase-tired-ASP Lee. i n U. v. a. ‘John chased Lee to the extent of making him (Lee) tired.’ b. *‘Lee chased John and he (John) got tired.’ c. ‘John chased Lee and (John) got tired.’ d. ‘Lee chased John and was made tired (by John).’. Interpretations (19a) and (19d) both have a causative meaning. Since none of the verbs in the resultative construction are causative themselves, Li (1995) claims that .

(35) 24 . the causative reading is achieved through the combination of the two verbs into a resultative compound. The causative reading is due to the assignment of the causative roles (C-roles) received from Rs in the V-R compounds to the grammatical functions, based on Causative Hierarchy; i.e., the more prominent Cause to the more prominent Subject and the less prominent Affectee to the less prominent Object. In his account, the more prominent Cause role could be assigned to the more prominent Subject even though the Cause as well as the theta role of the R is a less prominent role on. 政治大 Thematic Hierarchy (Bresnan and Kanerva, 1989, 1992) as opposed to the theta role 立. ‧ 國. 學. of the V in the V-R compound:. ‧. n. al. Ch. engchi. er. io. ag ＞ ben ＞ go/exp ＞ inst ＞ pt/th ＞ loc. sit. y. Nat. (20) Thematic Hierarchy (Bresnan and Kanerva, 1989, 1992):. i n U. v. In the prediction of Thematic Hierarchy, a more prominent theta role should be linked to a more prominent grammatical function. Therefore, Agent is usually linked to Subject since Agent is the most prominent role and Subject the most prominent function. But Li’s (1995) analysis of linking a less prominent theta role with the Cause role to a more prominent function indicates that Causative Hierarchy overrides Thematic Hierarchy. With an extra syntactic construct of Causative Hierarchy, Li .

(36) 25 . indeed poses a con to the elegance of his theory if the solution to the problem of interpreting causativity in resultative compounds could be sought entirely lexically without having to resort to any extra device. Since causativization in many languages is viewed to be a lexical relational operation like passivization (Aissen, 1979; Alsina, 1996; Bresnan, 1990; Falk, 2001), Her (2007) has further pursued this issue of resultative compounds with causative meanings by means of LFG, which treats lexical relational operations as the alternations of argument-function mapping.. 政治大 In Her’s (2007) analysis, the various argument-function mappings of the theta 立. ‧ 國. 學. roles of Chinese resultative compounds, along with causative readings, result from the. ‧. competition between the arguments for the two available syntactic positions. Using. sit. y. Nat. the aforementioned resultative compound, zhui-lei, in Li’s (1995) analysis, Her (2007). n. al. er. io. demonstrates how the arguments of the verbs form a composite role and explicates. Ch. i n U. v. how they manifest in the grammatical functions by introducing the syntactic concept. engchi. of suppression. Having inspected the argument structure of resultative compounds, Her (2007) comes up with the argument compositions of the verbs as follows:. (21) Resultative Compounding (Her, 2007): Vcaus ＜x y＞+ Vres ＜z＞→ Vcaus Vres＜α β＞, where ＜α β＞= (i) ＜x y-z＞ (ii) ＜x-z y＞ .

(37) 26 . As can be seen in (21), the first verb component of resultative compounds, Vcaus requires two arguments while the second verb component, Vres requires one. When they combine to form a resultative compound, their argument structures merge and two possibilities of the argument structures ensue. With the syntactic operation of suppression introduced for a strict one-to-one mapping requirement in LFG, Her (2007) successfully predicts all the possible interpretations a transitive resultative compound like that in (19) could contain5:. 立. 政治大. ‧ 國. John. Lisi. 學. (22) Zhangsan zhui-lei-le. chase-tired-ASP Lee. ‧. Vzhui ＜ag pt＞+ Vlei ＜th＞→. y. Nat. Vzhui Vlei＜α β＞, where ＜α β＞=. er. io. sit. (i) ＜ag pt-th6＞. ‘John chased Lee to the extent of making him (Lee) tired.’. n. al. Ch. (ii) ＜ag pt-th＞. engchi. i n U. v. ‘John chased Lee to the extent of making him (Lee) tired.’ (iii) ＜ag-th pt＞ ‘John chased Lee and (John) got tired.’                                                         5. Mechanisms regulating how the argument-function mapping takes place in LFG will be given with more details in Chapter 3. 6 The concept of suppression is merely a syntactic device and functions in order to provide proper argument realization patterns of Chinese resultative compounds. Which argument is suppressed does not come through as a semantic difference in the interpretations of the compounds. Take (22i) and (22ii) for example, the interpretations are practically the same despite that different arguments are suppressed in the two argument structures. Although one of the arguments in the composite role is suppressed syntactically, its meaning is still understood with its syntactically unsuppressed companion. In other words, the objects in (22i) and (22ii) all represent the patient of zhui and the theme of lei no matter which argument is syntactically realized. Suppression of the role is indicated by crossing out the role. .

(38) 27 . (iv) ＜ag-th pt＞ ‘Lee chased John and was made tired (by John).’. 2.3 Availability of the Accounts to Chinese Resultative Compounds After reviewing both the derivational and lexicalist accounts of argument realization on Chinese resultative compounds, this chapter describes the problems with which the derivational account is faced and the relatively sufficient ability of the. 政治大. LFG account in the comprehensive prediction of compound argument realization.. 學. ‧ 國. 立. 2.3.1 Insufficiencies in the Derivational Account. ‧. Recall that in the derivational account, the argument realization of Chinese. y. Nat. al. er. io. sit. resusltative compounds is analyzed by means of derivational operations that lead to. n. various templates of syntactic structures of the compounds and syntactic constraints. Ch. engchi. i n U. v. that regulate the interpretations of the compounds. This analysis encompasses almost all of the resultative compounds there is to be found in Chinese. However, if one takes a closer look at the interactions between the operations and constraints and the possible outcomes of the analysis, one would find that the derivational account does not produce satisfactory results in accounting for the argument realization of Chinese resultative compounds. With the Chinese resultative compounds that exhibit intransitivity and .

(39) 28 . subject-predication and are different from the English resultative constructions, a syntactic constraint named the Force Recipient Principle proposed by Rappaport Hovav and Levin (2001) and later modified by Huang (2006) was given to prevent the compounds from being ruled out by the Direct Object Restriction (Simpson, 1983). This constraint successfully explains the substantiality of Chinese unergative and transitive resultative compounds with subject-predication of the Result:. (23) Zhangsan kan-lei-le. 立. 政治大. ‧ 國. 學. Zhangsan read-tired-ASP ‘Zhangsan read himself tired.’. ‧. baozhi.. (subject-predication). sit. y. Nat. (24) Zhangsan kan-lei-le. (unergative). io. ‘Zhangsan read the newspaper and got tired.’. n. al. Ch. engchi. er. Zhangsan read-tired-ASP newspaper. i n U. v. In (23), no object exists. Therefore, the predication of the Result in the unergative compound falls on the prominent argument, i. e. the subject. On the other hand, although there is an object in the sentence, the “Force” of reading newspaper transmits not to the object being read but back to the reader; hence the subject-predication of the Result. However, a big part of the picture would be missing if we live simply on this constraint: .

(40) 29 . (25) Zhangsan zhui-lei-le. Lisi.. (transitive). Zhangsan chase-tired-ASP Lisi a. ‘Zhangsan chased Lisi and made him (Lisi) tired.’. (object-predication). b. ‘Zhangsan chased Lisi and (Zhangsan) got tired.’. (subject-predication). c. ‘Lisi chased Zhangsan and was made tired (by Zhangsan).’. (causative). Since the effect of the chasing action functions on the object, Lisi, the interpretation of object-predication is successfully predicted. But other interpretations of this sentence. 治政 would be missed out. Further, in the causative reading大 of (25), Zhangsan is the one 立 ‧ 國. 學. being chased and the recipient of force during the act of chasing. If so, the constraint would wrongly predict that the Result should be predicated on Zhangsan, thus. ‧. producing an ungrammatical interpretation of the sentence:. n. Ch. er. io. sit. y. Nat. al. (26) Zhangsan zhui-lei-le. Lisi.. engchi. Zhangsan chase-tired-ASP Lisi. i n U. v. *‘Lisi chased Zhangsan and he (Zhangsan) got tired.’. Some problems can as well be seen from the test of adverb incorporation in the derivational operations. In the causative templates of the derivational accounts, resultative compounds have undergone movements from the innermost embedded verb head to the outermost null causative verb to acquire the resultative and causative. .

(41) 30 . meanings, as in Template D:. (27) yi- ke. jidan. ye-si-le. yi-sui. yinger. one-CL egg choke-dead-ASP one-year baby ‘An egg choked a one-year-old baby to death.’ D. Pure causative (2): [ x CAUSE [BECOME<UNACCUSATIVE> [ y <STATE>]]]. VP1 DP. V2P DP. V2’. 學. [CAUSE]. io. n. al. yi-ke jidan one-CL egg. <UNACCUSATIVE>. y. sit. Nat. ye-si-le DP choke-dead-ASP [BECOME]. VP3. er. V2. ‧. ‧ 國. 立V1. 治政 V1’ 大. v i yi-sui yinger t n Ch e nbaby one-year gchi U. V3’ V3. <STATE> t. Since an adverb is an adjunct, it can be attached as the sister of a V’. In a structure like the above, there are three available positions for the attachment of an adverb:. .

(42) 31 . (28) VP1 DP. V1’ V1’. 3 AdvP ○. V1. tongku-de painfully. V2P DP. V2’. [CAUSE] V2’. 2 AdvP ○. ye-si-le choke-dead-ASP. 立. VP3 DP. V3’. ‧ 國. 學. V3’. 1 AdvP ○. [BECOME]. sit. <STATE>. n. er. io. al. V3. tongku-de t painfully. y. yi-sui yinger one-year baby. ‧. <UNACCUSATIVE>. Nat. yi-ke jidan one-CL egg. 治 V2 政tongku-de 大 painfully. Ch. engchi. i n U. v. On a closer inspection, one can see that the sentence with the adverb is acceptable before the derivation but ungrammatical after. The following are comparisons between the acceptable “before” sentences and the ungrammatical “after” sentences, depending on the available positions for the adverb:. . t.

(43) 32 1 a. before [BECOME] derivation: ○. yi- ke. jidan shi yi-sui yinger. one-CL egg. tongku-de7. ye-de. CAU one-year baby choke-become painfully. si-le dead-ASP. ‘An egg choked a one-year-old baby and caused him a painful death.’ b. after [BECOME] derivation: *yi- ke jidan shi yi-sui yinger. ye-si-le. tongku-de. one-CL egg CAU one-year baby choke-dead-ASP painfully ‘An egg choked a one-year-old baby and caused him a painful death.’. 政治大. 2 a. before [CAUSE] derivation: ○. 立. jidan shi yi-sui yinger tongku-de. ye-si-le. 學. ‧ 國. yi- ke. one-CL egg CAU one-year baby painfully choke-dead-ASP. ‧. ‘An egg choked a one-year-old baby and caused him a painful death.’. ye-si-le. yi-sui. yinger tongku-de. n. al. painfully. er. io. one-CL egg choke-dead-ASP one-year baby. sit. Nat. * yi- ke jidan. y. b. after [CAUSE] derivation:. i n U. v. ‘An egg choked a one-year-old baby and caused him a painful death.’. Ch. engchi. It is clear that after the derivations, the adverb ends up being in the wrong position with respect to the verb. This fact provides evidence for a re-evaluation of the substantiality of the derivational accounts. Furthermore, Problems also arise in the aforementioned templates used for classifying the syntactic structures and derivational operations of Chinese resultative                                                         7. De in ye-de indicates the becoming of the result depicted by si while de in tongku-de is an adverb marker. Adverbs with de appear before verbs in Mandarin Chinese. .

(44) 33 . compounds. Take the sentence in (25) for example, if according to the templates, interpretation A should fit into the Causing-with-a-manner template, Template E:. (29) Zhangsan zhui-lei-le. Lisi.. (transitive). Zhangsan chase-tired-ASP Lisi ‘Zhangsan chased Lisi and made him (Lisi) tired.’. (object-predication). E. Causing with a manner: [ x CAUSE<UNERGATIVE> [BECOME [ y <STATE>]]]. 政治大. 立. DP. 學. ‧ 國. VP1. V1’ V2P V2’. y. Nat. DP. ‧. V1. V2. n. Ch. zhui-lei-le chase-tired-ASP. e n[BECOME] gchi. VP3. er. io. al. sit. [CAUSE]. <UNERGATIVE>. i n U DP. v. V3’ V3. Zhangsan Zhangsan. Lisi Lisi. t <STATE> t. However, other interpretations of this sentence fit into other templates:. .

(45) 34 . (30) Zhangsan zhui-lei-le. Lisi.. (transitive). Zhangsan chase-tired-ASP Lisi ‘Lisi chased Zhangsan and was made tired (by Zhangsan).’. (causative). C. Pure causative (1): [ x CAUSE [BECOME<UNERGATIVE> [ y <STATE>]]] VP1 DP. V1’ V1. 立. V2P. 政治大 DP V2’. 學. ‧ 國. [CAUSE] V2 DP. ‧. zhui-lei-le chase-tired-ASP. VP3. [BECOME]. n. al. Ch. engchi. y. Lisi Lisi. t. V3. sit. io. Zhangsan Zhangsan. er. Nat. <UNERGATIVE>. V3’. i n U. ‘Zhangsan chased Lisi and (Zhangsan) got tired.’. v. <STATE> t. (subject-predication). Template: Not applicable. The causative reading of Zhangsan zhui-lei-le Lisi fits into Template C under Huang’s (2006) categorization, where the internal argument of zhui, Zhangsan, is thought to be generated in the subject position of the sentence without any movement. And the .

(46) 35 . transitive reading does not fit into any of the templates because it does not convey causativity. In brief, a resultative compound with multiple interpretations structurally fit into different templates. This being so, the point of generalization in argument realization of Chinese resultative compounds seems to be lost and the identity (to what category a resultative compound belongs) of the compounds blurred because the transitive use of a single compound renders the result of the compound being in many different categories. Moreover, the causativity encoded in the templates shows two. 政治大 different kinds of nature, depending on the syntactic nature of V1. Not covering this 立. ‧ 國. 學. point also whittles down the credibility of the derivational analysis. The identification. ‧. of causativity in Chinese resultative compounds is significant to the meaning and. n. al. er. io. sit. y. Nat. argument realization of the compounds. More details will be given in Chapter 4.. Ch. i n U. v. 2.3.2 Predictability of the Lexical Functional Grammar Account. engchi. Compared to the derivational account, Lexical Functional Grammar (LFG) actually works better in explicating the argument realization of Chinese resultative compounds. As long as the argument structures of V1 and V2 in the compounds are identified, all possible syntactic realizations of the arguments and their related interpretations will arise. Take Her’s (2007) analysis of the sentence Zhangsan zhui-lei-le Lisi for example (Example (22) repeated here as example (31) : .

(47) 36 . (31) Zhangsan zhui-lei-le John. Lisi. chase-tired-ASP Lee. Vzhui ＜ag pt＞+ Vlei ＜th＞→ Vzhui Vlei＜α β＞, where ＜α β＞= (i) ＜ag pt-th＞ ‘John chased Lee to the extent of making him (Lee) tired.’ (ii) ＜ag pt-th＞ ‘John chased Lee to the extent of making him (Lee) tired.’ (iii) ＜ag-th pt＞. 政治大. ‘John chased Lee and (John) got tired.’. 立. (iv) ＜ag-th pt＞. ‧ 國. 學. ‘Lee chased John and was made tired (by John).’. ‧ sit. y. Nat. Zhui-lei here is in transitive use. With zhui and lei’s argument structures identified, the. n. al. er. io. argument realization and interpretations of zhui-lei can be acquired exhaustively.. Ch. i n U. v. Under the same rule of the Resultative Compounding (Her, 2007), intransitive verbs,. engchi. such as unergative and unaccusative verbs, are able to undergo the same process and subsume grammatical argument realizations and interpretations:. (32) ta. ku-xia-le.. he cry-blind-ASP Vku ＜ag＞+ Vxia ＜th＞→ Vku Vxia＜α＞, where ＜α＞= . (unergative).

(48) 37 . (i) ＜ag -th＞ ‘He cried and went blind as a result.’ (ii) ＜ag-th＞ ‘He cried and went blind as a result.’. (33) ta. xia-hun-le.. (unaccusstive). he shock-faint-ASP Vxia ＜th＞+ Vhun ＜th＞→ Vxia Vhun＜α＞, where ＜α＞= (i) ＜th -th＞. 立. 政治大. ‘He got shocked and fainted.’. ‧ 國. 學. (ii) ＜th-th＞. ‧. ‘He got shocked and fainted.’. sit. y. Nat. io. n. al. er. As one can observe, transitivity of the resultative compounds determines the syntactic. i n U. v. realizations of the arguments and the nature of the predicates gives rise to the multiple. Ch. engchi. readings the sentence could have. The connection between arguments and grammatical functions involves a mechanism called the Lexical Mapping Theory (LMT) with several constraints that define the workings of LMT. Due to the simplicity and the comprehensiveness of the LFG analysis, this thesis follows the formalism of LFG in order to deal with Chinese resultative compounds. The details of LMT and its related constraints will be in full description in the next chapter as a. .

(49) 38 . theoretical basis for this thesis. Despite the prediction power exemplified in the examples above, there is still one missing link in the analysis of resultative compounds in LFG. A type of resultative compound is recorded in the works of the derivational linguists but not in the map of LFG: the causative category of resultative compounds. The causative category involves two alternations with compounds of other categories, as stated in the derivational accounts:. 立. 政治大. chang-ya-le.. Zhangsan sing-coarse-ASP. chang-ya-le. Zhangsan.. io. this-CL song sing-coarse-ASP Zhangsan. al. (causative). er. b. zhe-shou ge. sit. y. Nat. ‘Zhangsan sang and made his voice coarse.’. (unergative). ‧. ‧ 國. a. Zhangsan. 學. (34) Unergative-Causative Alternation:. n. v i n ‘Singing this songC made voice coarse.’ h eZhangsan’s ngchi U (35) Unaccusative-Causative Alternation: a. Zhangsan lei-huai-le.. (unaccusative). Zhangsan tired-bad-ASP ‘Zhangsan was exhausted.’ b. zhe-ge gongzuo lei-huai-le this-CL job. Zhangsan.. tired-bad-ASP Zhangsan. ‘This job made Zhangsan really exhausted.’ . (causative).

(50) 39 . The causative category of the unergative-causative alternation actually conforms to the fourth reading documented in Her’s (2007) analysis of resultative compounds as in (31). Since the subject in this category is the internal argument of the unergative predicate, chang, the possibility of its realization in the subject position is viable in the analysis of LFG. However, the causative category of the unaccusative-causative alternation tells a different story in attributing its argument realization to the argument structures of its predicates. The subject of the sentence is the argument of neither. 政治大 predicates of the resultative compound. Since LFG needs to operate on the argument 立. ‧ 國. 學. structures of the compounds to reveal possible interpretations, this causative structure. ‧. and reading will never appear in the predictions of LFG.. sit. y. Nat. Moreover, although the derivational account manages to notice the existence of. n. al. er. io. this causative category, it does not endeavor to differentiate the causativity in the. Ch. i n U. v. unergative-causative and unaccusative-causative alternations. This causativity. engchi. difference is the key to solving the prediction problem in LFG and explaining the identity of the subject argument. Pertaining to this issue, this thesis aims to propose a rule as the missing piece to the puzzle of resultative compounds in LFG.. .

(51) . 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. . Ch. engchi. i n U. v.

(52) . Chapter 3 Theoretical Background. Having weighed the pros and cons of both derivational and lexicalist theories about argument realization in Chinese resultative compounds, this thesis adopts the lexicalist approach to see what role causativity plays in argument realization. Instead of being recognized as the result of some derivational operations, different argument realizations of a compound are seen as the argument-function remapping in LFG.. 學. ‧ 國. 立. 政治大. 3.1 Lexical Mapping Theory (LMT). ‧. In LFG, when a predicate is extracted from the mental lexicon and put into a. sit. y. Nat. sentence, a syntactic device of mapping the lexical forms of the predicate to their. n. al. er. io. appropriate grammatical functions takes place. This device is termed the Lexical. Ch. Mapping Theory (LMT) (Bresnan, 1990: 646).. engchi. i n U. v. By assigning the features of grammatical functions to the arguments, LMT is able to map the arguments onto their proper grammatical functions:. (36) Lexical Mapping Theory (LMT): A. θ-structure to a-structure mapping Patients and Themes map to [-r] “secondary” Patients and Themes map to [+o] as a marked option 41 .

(53) 42 . Non-Theme/Patient arguments map to [-o]. B. a-structure to f-structure mapping SUBJ Mapping 1:. A [-o] argument which is Ө^ maps to SUBJ.. SUBJ Mapping 2:. [-r] may map to SUBJ.. NonSUBJ Mapping: Add positive values of features where possible.. The mapping process of LMT concerns three independent levels of structures: the. 政治大. lexical forms of the predicates (θ-structure), the intermediate level of argument. 立. structure (a-structure), and the grammatical function realizations of the arguments. ‧ 國. 學. (f-structure). In order to follow a strict one-to-one mapping relationship as required by. ‧. Theta Criterion and to fulfill the requirement of having a subject in a sentence in. y. Nat. al. er. io. sit. certain languages, two well-formedness conditions, Function-Argument Biuniqueness. n. Condition and the Subject Condition, have been proposed to achieve the goal:. Ch. engchi. i n U. v. C. Function-Argument Biuniqueness Condition: Each a-structure role corresponds to a unique f-structure function, and each f-structure function corresponds to a unique a-structure role.. D. The Subject Condition Every verb must have a SUBJ.. .

(54) 43 . It is with these mechanisms that the mappings of the lexical forms of verbs to the grammatical functions in the sentences can properly work. For example, if the verb, place, is to be used in a sentence, its argument realization would be as follows under the rules of LMT mentioned above (Falk, 2001):. (37) θ-structure: place ＜ Agent, a-structure: f-structure:. 立. Patient/Theme,. [-o]. [-r]. (-r, -o). (-r, +o). Location ＞. 治政 SUBJ OBJ 大. [-o] OBLθ (+r, -o). ‧ 國. 學 ‧. According to LMT, the Patient/Theme of place is assigned the feature [-r] while the. sit. y. Nat. other Non-Patient/Theme arguments are assigned the feature [-o]. The leftmost. n. al. er. io. argument in the lexical form is the most prominent argument, Ө^. Being Ө^ and. Ch. i n U. v. having the feature [-o], the Agent is mapped to the grammatical SUBJ by SUBJ. engchi. Mapping 1 in (36B). The rest of the arguments are added positive values of features where possible by NonSUBJ Mapping. The Patient/Theme and the Location end up with the features [-r, +o] and [+r, -o], respectively. [-r, +o] are the features of OBJ and [+r, -o] are the features OBLθ in LFG. Hence, the Patient/Theme maps to the function OBJ and the Location maps to the function OBLθ. After the mappings of the arguments are sorted out, the well-formedness conditions C and D set in to check. .

(55) 44 . whether the syntactic realization is viable. Since each argument maps to each function in a strict one-to-one fashion and the position of SUBJ is filled, the mappings pass the check and surface as a grammatical sentence.. 3.2 Intrinsic and Default Morphosyntactic Classification of Argument Roles (IC & DC) Considering that the process of mapping from θ-structure to f-structure can be. 政治大 better generalized in LMT and that not all languages share the characteristic of an 立. ‧ 國. 學. obligatory subject, Her (2009) restates (36A) above as two rules, the Intrinsic. ‧. Morphosyntactic Classification of Argument roles (IC) and the Default. n. al. er. io. sit. y. Nat. Morphosyntactic Classification of Argument roles (DC):. i n U. v. (38) Intrinsic Morphosyntactic Classification of Argument Roles (IC): θ,. θ = pt/th. Ch. engchi. [-r] (39) Default Morphosyntactic Classification of Argument Roles (DC): θ,. θ ≠ Ө^. [+r]. These two rules simplify the mapping process in LMT. Instead of determining what arguments should be assigned minus features and what arguments should be assign .

(56) 45 . plus features, Her (2009) gives only a rule that specifies the feature assigned to Patient/Theme, i.e. [-r]. As for the rest of the arguments, a default rule sets in and everybody gets the feature [+r] except for Ө^. In Her’s theory, Ө^ is the most prominent argument having no features and can map to the most prominent grammatical function available. Patient/Theme, with the feature [-r], can only map to the functions with compatible features. Last, other arguments, with the feature [+r], map to functions with the same feature.. 學. ‧ 國. 立. 政治大. 3.3 The Unified Mapping Principle (UMP). ‧. However, the Intrinsic Morphosyntactic Classification of Argument roles (IC). sit. y. Nat. and the Default Morphosyntactic Classification of Argument roles (DC) only tell half. n. al. er. io. of the story. In LFG, the grammatical functions that are basic in a sentence are called. Ch. i n U. the argument functions and possess the features as follows:. (40). engchi. v. [–o]. [+o]. [–r]. SUBJ. OBJ. [+r]. OBLθ. OBJθ. One argument has to possess two features to be able to map to a corresponding function. That is why the original LMT has the rule of assigning positive features to .

(57) 46 . arguments in order to complete feature specification. IC and DC only assign one feature to the arguments and thus leave the arguments multiple options to choose from. Patient/Theme, after IC, has the functions SUBJ and OBJ to map to since they all share the feature [-r]. And other arguments, after DC, have the functions OBLθ and OBJθ to map to since they all share the feature [+r]. This goes against the strict one-to-one mapping requirement in LMT. As a result, Her (2009) comes up with another constraint that solves the problem and further meets other well-formedness. 政治大 conditions in LMT. This constraint is termed the Unified Mapping Principle (UMP): 立. ‧ 國. 學 ‧. (41) The Unified Mapping Principle (UMP):. Map each argument role, from the most prominent to the least, onto the. y. Nat. sit. highest compatible function available.. n. al. er. io. (*A function is available iff it is not linked to a role.). Ch. engchi. i n U. v. The UMP strictly states that each argument should map to the most prominent function available. This cuts down the options for Patient/Theme and other arguments to only one that is most prominent on the list according to the Markedness Hierarchy of Argument Functions:. .

(58) 47 . (42) Markedness Hierarchy of Argument Functions: SUBJ (-r -o) ＞ OBJ (-r +o)/OBLθ (+r –o) ＞ OBJθ (+r +o). The problem of one-to-one mapping requirement is solved. Moreover, since SUBJ is the most prominent function and will always be the first on the list of argument-function mapping, the well-formedness condition in (36D) is no longer needed. The UMP actually incorporates the two well-formedness conditions and. 政治大. simplifies the mapping mechanisms of LMT.. 3.4 Causative Assignment in Resultative Compounding. 學. ‧ 國. 立. ‧. With the mechanisms so far, the argument realization of predicates in Chinese. y. Nat. al. er. io. sit. resultative compounds can now be properly processed. Mappings of the resultative. n. compound, zhui-lei, are constructed as thus in Her (2009):. (43) Zhangsan John i.. zhui-lei-le. engchi. v. Lisi. chase-tired-ASP Lee. ‘John chased Lee to the extent of making him (Lee) tired.’ ＜x IC DC CF UMP. . Ch. i n U. y-z＞ [-r]. ---------------S/O… S/O S O. (x=ag, y=pt/th) (non-causative). (causative).

(59) 48 . y-z[af]8＞. ＜x[caus] IC DC. [-r] ---------------------------S/O… S/O S O. CF UMP. ‘John chased Lee and (John) got tired.’ y＞. ＜x-z IC DC. (x=ag, y=pt/th) (non-causative). [-r] ---------------S/O… S/O S O. CF UMP. 政治大. ‧ 國. 立. 學. ‘Lee chased John and was made tired (by John).’. (z=pt/th, y=pt/th) (causative). ‧. [-r]. n. al. Ch. engchi. y. --------------------------S/O S/O O S. io. CF UMP. [-r]. y[caus]＞. Nat. IC DC. ＜x-z[af]. (causative). sit. iii.. (non-causative). er. ii.. (x=ag, z=pt/th) (causative). i n U. v. Note that the various interpretations of the resultative compound are reached by the different function mappings of the arguments, but the causative readings in the interpretations are still unaccounted for. Furthermore, one problem remains in that how the two arguments with the same feature in the final interpretation are assigned their grammatical functions is still a mystery.                                                         8. . [Caus] is short for the concept of Causer, and [af], Affectee. .

(60) 49 . Her (2007) tries to solve this problem along with the causative readings by submitting to his proposed rule of Causativity Assignment in Resultative Compounding:. (44) Causativity Assignment in Resultative Compounding: An unsuppressed role from Vres receives [af] iff an unsuppressed role from Vcause exists to receive [caus].. 立. 政治大. This rule accounts for the causative readings in the resultative compound by. ‧ 國. 學. distributing Causer and Affectee to the corresponding arguments. Moreover, it leads. ‧. to the final interpretation in (43) by adding a semantic feature of Causer to one of the. Nat. io. sit. y. two arguments that receive the same [-r] feature. By resorting to Dowty’s (1991) idea. al. er. of proto-agent and proto-patient, Her (2007) shows that the argument that receives the. n. v i n C h the one that receives feature [caus] is more prominent than e n g c h i U [af] and should therefore map to SUBJ, which is a more prominent grammatical function.. The complex issue of the mappings of resultative compounds with the relevant causativity seems to be well-answered. However, remember the causative alternation of unaccusative resultative compounds, where the subject is not an argument if under the analysis of LMT that the argument-function mappings all come from the arguments of the predicate. Yet, the subject still receives the causative reading that has .

(61) 50 . to be given by the Vres in the compound. To make things worse, the Vres of the compound does not function as a real intransitive verb that assigns its theta role to the entities in the sentence, nor does it assign causativity. The Vres is more like an expression of extent in the compound:. (45) yi. zhuozi de. one table. cai pang-si wo le. DE dish fat-die me. ASP. 政治大. ‘A table full of dishes is making me fat.’. 立. ‧ 國. 學. Yi zhuozi de cai (a table full of dishes) is not an argument of any of the predicates in. ‧. the resultative compound, pang-si, nor does si function as a real verb that denotes the. Nat. io. sit. y. action of dying. Her’s (2007) account seems insufficient to explain resultative. er. compounds like this. How, then, can the resultative compounds of this kind be. al. n. v i n C h accounted for?UThis thesis is dedicated to analyzed and their relevant causativity engchi answering this question by examining causativity in Mandarin Chinese and solving the problem of causativity and lexical mappings of Chinese resultative compounds. But before that, this special kind of resultative compounds needs to be examined to see if the compounds indeed carry a seemingly extra argument outside of their argument structures. A kind of construction would look like the causative resultative compounds in question here but turn out to be something different under close .

(62) 51 . inspection, as pointed out by Wu (2010):. (46) na-chang jihuang that-CL. e-si-le. henduo ren. famine hungry-dead-ASP many people. ‘Many people starved to death in that famine.’. In (46), jihuang (famine) is the argument of neither e (hungry) nor si (dead). Yet, it appears in the subject position. Wu (2010) claims that the sentence should be. (47) henduo ren. e-si. yu na-chang jihunag in. that-CL famine. ‧. many people hungry-dead. 學. ‧ 國. 治政大the thematic role of Locative: paraphrased as (47) and the subject argument is actually 立. al. er. io. sit. y. Nat. ‘Many people starved to death in that famine.’. n. Yu, in Mandarin, can serve to specify time and place. Jihuang in (46) functions as. Ch. engchi. i n U. v. Locative and has undergone a syntactic process that has been noted in English as well as in Mandarin Chinese and termed the Locative Inversion (Her, 2009). In this case, what appears to be a causative resultative compound is indeed an example of Locative Inversion. However, the Locative Inversion account does not suffice to explain all possible instances of causative resultative compounds, where the subject argument cannot be seen as Locative or any argument required by the argument structures of the compound predicates: .

(63) 52 . (48) a. zhe-jian yangzhuang this-CL dress. mei-fan. ta le.. beautiful-flip her ASP. ‘This dress makes her look extremely beautiful.’ b. *ta. mei-fan. yu zhe-jian yangzhuang. she beautiful-flip in this-CL dress ‘This dress makes her look extremely beautiful.’. Since the causative resultative compound in (48a) cannot be paraphrased as (48b) and. 政治大. semantically speaking, the subject argument, zhe-jian yangzhuang, cannot be. 立. interpreted as a location where the incident depicted by the compound happens, there. ‧ 國. 學. is no way causative resultative compounds like this are to be viewed as. ‧. Locative-Inversion sentences in disguise. Therefore, it is for sure that this special kind. y. Nat. al. er. io. sit. of Mandarin causative resultative compounds exists and proper measure of analysis. n. should be taken to account for its presence.. Ch. engchi. . i n U. v.