Acta Linguistica Asiatica, 10(2), 2020. ISSN: 2232-3317, http://revije.ff.uni-lj.si/ala/ DOI: 10.4312/ala.10.2.9-48 CONTEXTUAL CONDITIONS AND CONSTRAINTS IN CHINESE DANGLING TOPICS: SYNTAX-DISCOURSE INTERFACE ANALYSIS WU Jiayi Fu Jen Catholic University, Taiwan wujiayi.fju@gmail.com Abstract This study verifies the aboutness condition proposed by Chao (1968), Chafe (1976), Li and Thompson (1981) Xu and Langendoen (1985) and many others as a relation that holds between sentence-initial NPs and comment clauses of alleged gapless topic-comment constructions in Mandarin Chinese. Four often-cited types of dangling topics in Chinese are revisited and the arguments from both the semantic (Pan & Hu 2002, 2008, 2009) and syntactic (Shi, 2000; Huang & Ting, 2006) views are examined. A critical scrutiny from contextual perspective, in particular Nomi Erteschik-Shir’s (2007) notions of topic, reveals that there exist dangling topics in Chinese where the sentence-initial NPs at issue are either contrastive or old topics which need not be syntactically or semantically licensed. The contextual contrast set and commentative conditions are proposed as the licensing conditions with constraints dictating that proper selection and relevance must be observed. At the end the study raises a hypothesis regarding language typology. It is proposed that discourse configurational languages exhibit phenomena of dangling NP topics which are properly licensed by the aboutness relation under which either one of the contextual conditions is satisfied. Keywords: dangling topics; aboutness relation; licensing conditions; language typology Povzetek Ta študija preverja pogoj »govorjenja o« (angl. aboutness), ki ga predlagajo Chao (1968), Chafe (1976), Li in Thompson (1981) Xu in Langendoen (1985) in drugi kot povezavo med samostalniškimi frazami na začetku stavkov in komentarji domnevno neprekinjenih konstrukcij tema – komentar v mandarinščini. Članek ponovno prouči pogosto obravnavane štiri vrste visečih tem v kitajščini in pregleda tako pomenski (Pan & Hu 2002, 2008, 2009) kot tudi skladenjski vidik (Shi, 2000; Huang & Ting, 2006). Kritični pregled z vidika konteksta, še zlasti opredelitev pojma teme v Nomi Erteschik-Shir (2007), razkriva, da v kitajščini obstajajo viseče teme, pri katerih so sporni začetki stavkov bodisi primerjalni bodisi stare teme, ki so pomensko oziroma skladenjsko pogojene. Kontekstualne primerjave in pogoji ocenjevanja so predlagani kot licenčni pogoji z omejitvami, ki narekujejo, da je treba upoštevati pravilno izbiro in ustreznost. Na koncu študija postavlja hipotezo o tipologiji jezika in poda predlog, da v jezikih konfiguracijskega diskurza pride do pojava visečih tem v obliki samostalniških fraz, ki jih pravilno pogojuje »govorjenje o«, s čimer je izpolnjen eden od od kontekstualnih pogojev. Ključne besede: viseče teme; odnosi pri »govorjenju o«; pritrditveni pogoji; jezikovna tipologija 10 WU Jiayi 1 Introduction Dangling topics in Mandarin Chinese have drawn considerable attention in recent decades. Six type of utterances presented below have often been cited and debated as to the questions of whether there exist dangling topics in Chinese and how they are properly licensed (Shi, 2000; Pan & Hu, 2002; Huang & Ting, 2006). (1) 他們 大魚吃小魚。 Tamen da-yu chi xiao-yu. They big-fish eat small-fish ‘They act according to the law of the jungle.’ (2) 他們 誰都不來。 Tamen shei dou bu lai. They who all not come ‘They (topic), none of them are coming.’ (3) 那場火 幸虧 消防隊來得快。 Na-chang huo xingkui xiaofang-dui lai-de-kuai. that-CL fire fortunately fire-brigade come-DE-fast ‘As for that fire, fortunately the fire brigade came quickly; (otherwise)…’ (4) 這件事情 你不能 光麻煩一個人。 Zhe-jian shiqing ni bu neng guang mafan yi-ge ren. this-CL matter you not can only bother one-CL person ‘This matter (topic), you can’t just bother one person.’ (5) 那種豆子 一斤 三十塊錢。 Na-zhong douzi yi-jin san-shi-kuai qian. that-CL bean one-catty 30 CL money ‘That kind of bean (topic), one catty is thirty dollars.’ (6) 物價 紐約 最貴。 Wu-jia Niuyue zui gui thing-price New-York most expensive ‘Speaking of the price of things, New York is the most expensive.’ 1.1 Literature review To begin with, the alleged gapless constructions above are often interpreted as dangling topics by aboutness relation: the comment clause says something about the topic (Chao, 1968; Chafe, 1976; Li & Thompson, 1981; Xu & Langendoen, 1985). Contextual Conditions and Constraints in Chinese Dangling Topics: … 11 Another line following semantic approach is a theory of formal semantics proposed by Pan and Hu (2002). A handful of semantic relations are proposed to account for the occurrence of the sentence-initial noun phrases (NPs). Shortly after the semantic analysis, Pan and Hu (2008; 2009) go one step further to propose a semantic-pragmatic account that unifies both the moved and base-generated topics.1 Opposite proposals are syntactic views primarily from Shi (2000), and Huang and Ting (2006). On the basis of the evidence in (7), Shi rejects the aboutness relation and proposes a structural account to argue against the semantic view that there exist dangling topics. And in favor of Shi’s structural approach Huang and Ting hold that the so-called dangling topic NPs are in fact subjects, NP topics, NP adverbials, and prepositional phrase (PP)-reduced forms. (7) a. *這件大事 我知道 張校長辭職了。 *Zhe-jian da-shi wo zhidao Zhang-Xiaozhang cizhi le. this-CL big-issue I know Zhang-Principal resign ASP ‘As for this big issue, I know that Principal Zhang has resigned.’ b. 這件大事 就是 張校長辭職了。 Zhe-jian da-shi jiu shi Zhang-Xiaozhang cizhi le. this-CL big-issue exactly be Zhang-Principal resign ASP this-CL big-issue ‘This big issue is that Principal Zhang has resigned.’ 1.2 Motivation, Methodology, and Purpose The motive for conducting this research arises as follows. First, arguments raised within previous literature, be it the syntactic or semantic account of Chinese topics, have their share of problems. Second, the issue has rarely been looked at from contextual perspective. They suffer from difficulty and controversy in identifying topics solely based on the traditional criterion of (in)definiteness. Some of them support the existence of dangling topics in Chinese without a holistic justification that explains all the possible examples, while others reject the existence based on ad hoc assumptions without much contextual consideration. Third, the puzzle of whether the aboutness relation is a valid condition under which the so-called dangling topics can be properly licensed remains open for debates. This paper thus aims to explore the contextual licensing conditions in Chinese and what it implies about language typology. 1 In their paper, a semantic-pragmatic interface account of dangling topics in Mandarin Chinese, Pan and Hu (2008) argue that topic structures be accounted for at the semantic–pragmatic interface in Mandarin Chinese. Specifically, in Chinese a topic is licensed if there is a variable in the comment and the set generated by this variable produces a non-empty set when intersecting with the set represented by the topic. 12 WU Jiayi This paper adopts a syntax-discourse analysis approach. The notions of topic are primarily drawn from the work of Nomi Erteschik-Shir (2007), Information Structure: Syntax-Discourse Interface. Her definitions of topic are relevant and instrumental in shedding light on the puzzles when identifying the Chinese topics in question throughout this paper. As a subfield of linguistics dealing with how sentences are structured with respect to topic and focus, Information structure (IS for short) fares better in identifying topics since it put topics in a broader context. Topics are given or old, and, as Erteschik-Shir argues, are identified by context.2 Pronouns, defintes, specific indefinites, generics, and contrastive elements qualify as topics.3 All sentences must have a topic and a sentence can have multiple topics. They can occur anywhere in a sentence. This paper argues that the initial NPs at issue are dangling topics because they are not only what the utterances are about in discourse but also occupy syntactically in left-peripheral position higher than subject. Specifically, they are either contrastive or old topics that dangle syntactically or semantically unrelated to the following clauses. The relation is simply pragmatic. Four of the often-debated gapless topics in Chinese, specifically the material/instrument type, as in (4), the nominal-predicate type, as in (5), the event type, as in (3), and finally the (non-)idiom chunk type, as in (1), will be re-visited respectively in section 2, 3, 4, and 5. A rich variety of examples in Chinese will be provided to challenge the previous semantic and syntactic views. And a pragmatic hypothesis will be postulated. There two context-induced licensing conditions that guarantee the well- formedness of dangling topics in Chinese and their constraints will be investigated in section 6 and 7. Specifically, they are contrast set and commentative conditions with constraints dictating that proper selection and relevance have to be observed. Finally, the allegedly vague aboutness relation will thus be justified and claimed cross- linguistically as a valid licensing mechanism under which either one of the contextual conditions is satisfied. 2 Erteschik-Shir argues that the choice of topic is context dependent. Consider the examples below: (1) Q: What happened? A: John washed the dishes. (2) Q: What happened to the dishes? A: John washed them. (3) Q: Why does John look so pleased with himself? A: He washed the dishes. The question in (1) forces a reading in which the sentence must be predicated of a stage topic, the current spatio-temporal parameter of the sentence. The question in (2) not only refers to “the dishes” but also asks what happened to them. In (3), “John” is the only given element in the answer, hence the topic. 3 Erteschik-Shir provides examples of Danish topicalization to exhibit readers a variety of topics. Readers can refer to page 8 of the book. Contextual Conditions and Constraints in Chinese Dangling Topics: … 13 2 The Material/instrument type This section deals with the examples, as in (8), which are analyzed as dangling topics by Pan and Hu, whereas sentences involving PP-reduction by Shi (2002) and then NP- movement by Huang and Ting (2006). Their analyses are not without their problems. A careful examination reveals that the sentence-initial NP in (8b) is in fact a contrastive topic that need not be syntactically licensed. (8) a. 這件事情 你不能 光麻煩一個人。 Zhe-jian shiqing ni bu neng guang mafan yi-ge ren. this-CL matter you not can only bother one-CL person ‘This matter (topic), you can’t just bother one person.’ b. 西紅柿 我 炒了雞蛋。 Xihongshi wo chao (le) ji-dan. Tomato I fry ASP chicken-egg ‘I fried eggs with tomatoes.’ 2.1 The semantic account Standing in opposition to Shi’s preposition-dropping analysis, Pan and Hu postulate an asymmetry between (9b) and (9c) and suggest that the difference between them lies in whether the initial NP is subcategorized. (9) a. 為這件事情 你不能 光麻煩一個人。 Wei zhe-jian shiqing ni bu neng guang mafan yi-ge ren. for this-CL matter you not can only bother one-CL person ‘For this matter, you cannot just bother one person.’ b. 這件事情 你不能 光麻煩一個人。 Zhe-jian shiqing ni bu neng guang mafan yi-ge ren. this-CL matter you not can only bother one-CL person ‘This matter (topic), you can’t just bother one person.’ c. 為這件事情 張三 打架了。 Wei zhe-jian shiqing Zhangsan dajia le. for this-CL matter Zhangsan fight ASP ‘For this matter, Zhangsan fought.’ d. *這件事情 張三 打架了。 *Zhe-jian shiqing Zhangsan dajia le. this-CL matter Zhangsan fight ASP ‘For this matter, Zhangsan fought.’ 14 WU Jiayi Arguing that the initial NP in (9b) is subcategorized by the verb mafan ‘bother’ in the thematic structure, Pan and Hu claim that the subcategorized NP is a dangling topic semantically licensed by bearing the theta role of Instrument or Material of the predicate. Conversely, the initial NP in (9d) cannot be a dangling topic since it is not subcategorized and semantically licensed by the predicate dajia ‘fight’. The analysis prompts Pan and Hu to further suggest that the sentence in (8b) is also a construction of dangling topic. Nevertheless, their semantic analysis incurs a fundamental problem that, given subcategorized, the initial NPs in (8) should have syntactic relations with their predicates and hence should not be considered as dangling topics per se. 2.2 The syntactic account Following Shi’s structural perspective, Huang and Ting propose an NP-movement solution. However, their analysis only half solves the issue. Providing sentences such as (10), they first argue that (8a) involves a double-object construction. (10) 我 想麻煩你 一件事。 Wo xiang mafan ni yi-jian shi. I want bother you one-CL matter ‘I want to bother you with something.’ The initial NP zhejian shiqing ‘this matter’ in (8a) is analyzed as moving from the direct-object position subcategorized by the three-place predicate mafan ‘bother’, and therefore should not be treated as a dangling topic. We concur with the NP movement analysis to (8a) thus far. To further justify that (8b) also involves movement, they adopt Huang’s (1982) complex predicate analysis of ba-construction in Chinese, as exemplified in (11), and reason that the initial NP in (8b) is in fact syntactically licensed by being the outer complement of the complex predicate chao jidan ‘fry egg’. Contextual Conditions and Constraints in Chinese Dangling Topics: … 15 (11) a. 他 [把紙門] 踢了一個洞。 Ta [ba zhi-men] ti le yi-ge dong. He [BA paper-door] kick ASP one-CL hole ‘He kicked a hole in the paper-door.’ b. This account that analogizes (8b) to the complex predicate structure is compelling, especially when we realize that both xihongshi ‘tomato’ and ji-dan ‘chicken-egg’ in (8b) are roles of material which can equally be the objects of the predicate chao ‘fry’. One more similar example is given below: (12) 玉米 我 炒(了)火腿。 Yumi wo chao (le) huotui. Corn I fry ASP ham ‘I fried hams with corns.’ However, challenges emerge when it comes to the example in (13) where the sentence-initial NP is not role of material but that of instrument. (13) 平底鍋 我 炒(了)青菜。 Pingdi-guoInstrument wo chao (le) qingcai. flat-pan I fry ASP vegetable ‘I fried vegetable with flat-pan.’ Imagine that the sentence can be uttered when a speaker was asked to list the instruments he/she had used when preparing a meal. Note that pingdi-guo ‘flat-pan’ S NP ta ‘he’ VP V’ V NP yige dong ‘a hole’ NP ti ‘kick’ zhimen ‘paper-door’ 16 WU Jiayi here cannot be the object of the predicate chao ‘fry’. This suggests that the extension of the complex predicate analysis to (13) would not be a satisfactory state of affairs. There should be a unified solution to accommodate both cases. 2.3 The pragmatic proposal Following a similar line to Erteschik-Shir’s notion of contrastive topic which indicates that contrast is contextually constrained to occur only if a contrast set is available,4 we argue that prior to the utterances of (8b), (12) and (13) there should be contrast sets discoursually available. In other words, the initial NP is a contrastive dangling topic licensed by a pragmatic condition requiring that one or each member of a contrast set be selected. Take (14) for further illustration. (14) a. 烤箱 我 烤(了)雞。 Kaoxiang wo kao (le) ji. Oven I bake ASP chicken ‘I baked chicken with oven.’ b. 電鍋 我 燉(了)肉。 Dian-guo wo dun (le) rou. Electric-cooker I stew ASP meat ‘I stewed meat with electric cooker.’ The sentences above can be uttered when an apprentice cook was asked by his/her chef to clarify how he/she prepared a meal using the instruments (utensils) available in the kitchen. Adopting Erteschik-Shir’s notion of contrastive topic, we are therefore convinced that the sentences in (8b) and (12) involve a contrast set concerning the topic of culinary material, as represented in (15). And the sentences in (13) and (14) involve a contrast set regarding culinary instrument, as in (16). The occurrences of the initial NP are apparently not syntactically but pragmatically licensed. 4 Erteschik-Shir explained that this is the case both when the focus is contrasted as in (1) and when the topic is contrasted as in (2). (1) a. Which laundry did John wash, the white, or the colored? b. He washed the white laundry. (2) a. Tell me about your brothers John and Bill. b. John is the smart one. In both responses, one member of the contrast set provided in the context is selected. In (1b), the contrasted element answers a wh-question and is therefore a contrasted focus. In (2b), the contrasted element is diagnosed as a topic because it is one member of the topic set in the discourse. Contextual Conditions and Constraints in Chinese Dangling Topics: … 17 (15) [Material1, Material2, Material3,…] Contrast set: the topic of culinary material (16) [Instrument1, Instrument2, Instrument3,…] Contrast set: the topic of culinary instrument To sum up, some might adopt a syntactic recovery approach and speculate that the material or instrument NPs in the sentence-initial position are syntactically licensed because they can be found in the recovered sentences. We disagree with this theory because to recover those NPs one has to employ additional phases and often ends up with anomalous sentences that people do not usually use. And the recovered sentences do not necessarily guarantee the availability of the un-recovered sentences. For example, both instrument and location NPs can be found in a recovered sentence but it is instrument rather than location that is able to be syntactically licensed in the sentence initial position. And some might even adopt a semantic role approach and argue that they are semantically licensed because both material and instrument are semantic roles denoted by the predicates. Again, the question boils down to why other adjunct roles such as location cannot be licensed in the same position. The answer is clearly a pragmatic requirement and only material and instrument can be contrastive in such context. 3 The nominal-predicate Type This section addresses another type of topic which involves quantity phrase (QP), as in (5). A discreet review of the problems from both the semantic and syntactic sides indicates that the initial NPs of this type are dangling topics which also involve the pragmatic requirement that one or each member of a contextually available contrast set be selected for evaluation. 3.1 The semantic account Disagreeing with Shi’s (2000) structural analysis that the sentence-initial NP of the so- called nominal-predicate construction in (17) is a subject, Pan and Hu (2002) provide the unacceptability of the sentence in (18) and contend that the initial NP in (17) must be a topic on the basis of a well-known criterion that a topic must be definite while a subject need not be (Li & Thompson, 1979/1981; Tan, 1990; Shyu, 1995). The analysis leads them to further suggest a set-member account5 that the NP at issue is a dangling topic semantically licensed by the set-member relation between the NP and the QP yijin ‘one catty’. 5 Pan and Hu propose a set-member account in which the dangling topic nazhong douzi ‘that kind of bean’ is analyzed as a set among which yijin ‘one catty’ is delimited as the member of the dangling topic. 18 WU Jiayi (17) 那種豆子 一斤 三十塊錢。 Na-zhong douzi yi-jin san-shi-kuai qian. that-CL bean one-catty 30 CL money ‘That kind of bean (topic), one catty is thirty dollars.’ (18) *一種豆子 一斤 三十塊錢。 *Yi-zhong douzi yi-jin san-shi-kuai qian. one-CL bean one-catty 30 CL money ‘One kind of bean (topic), one catty is thirty dollars.’ However, their statement appears unsatisfactory, especially when we consider the sentence in (19) where the initial NP is construed with an existential marker you ‘have’. That is, the sentence in (18) is saved when the NP in (18) is construed with the marker such as you ‘have’. Along a similar line to Hsin’s (2002) analysis that the initial NP of a sentence in Chinese can be indefinite when preceded by the existential marker,6 we argue that it is fairly possible that the unacceptability of (18) is simply because it is not preceded by the existential marker you ‘have’ to produce an indefinite specific reading. (19) 有一種豆子 一斤 三十塊錢。 You yi-zhong douzi yi-jin san-shi-kuai qian. have one-CL bean one-catty 30 CL money ‘One kind of bean (topic), one catty is thirty dollars.’ 3.2 The syntactic account Conversely, in light of the fact that there is a strong tendency for clause-initial NPs in Chinese to be definite,7 Huang and Ting (2006) argue for Shi’s structural theory and 6 Hsin notes that, if we assume the indefinite NPs in the predicate is assigned the existential meaning by the VP enclosure, then all indefinite NPs outside VP enclosure would need a syntactic marker to be licensed. 7 Huang and Ting note that there is a strong tendency for clause-initial NPs in Chinese to be definite. The definiteness constraint on a subject can only be relaxed in a limited number of contexts. For example, when the nominal at issue denotes a quantity (Li 1998) as in (1a), expresses a unit reading (Shi 2000) as in (1b), or occurs with stage-level predicates in a root context (Shyu 1995) as in (1c), they can occur in the grammatical subject position. Contextual Conditions and Constraints in Chinese Dangling Topics: … 19 propose that the initial NP in (17) is a subject rather than a topic. Citing the unacceptability of (20) without the alleged QP yijin ‘one catty’, they argue that the problem for the unacceptability of (18) lies in the indefiniteness of the subject NP rather than that of the topic NP. (20) *一種豆子 三十塊錢。 *Yi-zhong douzi san-shi-kuai qian. one-CL bean 30 CLmoney ‘One kind of bean is thirty dollars.’ As they further contend, Pan and Hu’s set-member account is in fact untenable because it is not applicable to other similar constructions such as (21). Namely, there is clearly no such a set-member relation between the initial NP yiliang che ‘one car’ and the QP yitian ‘one day’ or sanqian ‘three thousand (dollars)’. (21) 一輛車 一天 三千。 Yi-liang che yi-tian san-qian. one-CL car one-day 3000 ‘One car is three thousand dollars a day.’ Their challenge to the set-member account based on the evidence in (21) appears perspicacious; however, the argument is not without problems. First, in their analysis it seems that all sentence-initial NPs are subjects regardless of whether they are definite or indefinite. This indicates that there should be no topic at all, contrary to the fact. Topic and subject are not mutually exclusive. It has been widely observed that subjects are unmarked topics across languages (e.g., Li &Thompson 1976; Reinhart 1981; Andersen 1991; Lambrecht 1994; Winkler & Gӧbbel 2002; Erteschik-Shir 2007). It follows that distinguishing subject from topic, and vice versa, based on the traditional criterion of (in)definiteness is inadequate. Second, the syntactic argument again does not take into account the possibility that the unacceptability of (18) in fact results from the fact that the initial NP is not preceded by the existential marker you ‘have’ either. It should be fairly possible that the reason why (18) is unacceptable is simply because (1) a. San-ge ren chi liang-wan fan. three-CL person eat two-bowl rice ‘Three people eat two bowls of rice.’ b. Yi-zhi qingwa si-tiao tui. one-CL frog four-CL leg ‘A frog (has) four legs.’ c. Yi-wei yi-sheng xiang wo jieshao ta-de bing-ren. one-CL doctor toward I introduce he-GEN sick-person ‘A doctor introduced me to his patients.’ Therefore, if the sentence-initial NP is a subject, the sentence may be ruled out because the NP does not meet the requirement for an indefinite subject. Please refer to their work “Are There Dangling Topics in Mandarin Chinese?” on page 139 for more details. 20 WU Jiayi it is not construed with the existential marker, rather than because it does not meet the requirement for an indefinite subject. 3.3 The topichood of initial NP Along Huang’s (1984) claim that Chinese is one of the topic-drop languages,8 we argue that the sentence-initial NPs are in fact topics due to their contextual droppability. This is especially true when we compare the examples without the QPs and those without the initial NPs. As shown in (22-23), the droppability of the initial NPs in (22b) and (23b) entails their topichood. (22a) and (23a) are equally acceptable only when QPs are mutually understood as basic units of measurement and thus are often omitted in such context. A distinction then has to be made that sentence b involves topic drop while sentence a QP drop due to common ground (CG). (22) a. 那種豆子 (一斤) 三十塊錢。 Na-zhong douzi (yi-jin) san-shi-kuai qian. that-CL bean one-catty 30 CL money ‘That kind of bean (topic) is thirty dollars per catty.’ b. (那種豆子) 一斤 三十塊錢。 (Na-zhong douzi) yi-jin san-shi-kuai qian. that-CL bean one-catty 30 CLmoney ‘One catty is thirty dollars.’ 8 Huang (1984) distinguishes subject drop from topic drop as follows: subject drop is dependent on the availability of rich inflectional agreement morphology; topic drop does not exhibit such a dependency. Instead, topic is recoverable from discourse. Topic trop is illustrated in the Chinese examples in (1) below (e stands for the omitted topic pronoun). (1) a. e lai-le. come-LE ‘(He) came.’ b. Lisi hen xihuan e. Lisi very like ‘Lisi likes (him) very much.’ c. Zhangsan shuo e bu renshi Lisi. Zhangsan say not know Lisi ‘Zhangsan said that he did not know Lisi.’ d. Zhangsan shuo Lisi bu renshi e. Zhanfsan say Lisi not know ‘Zhangsan said that Lisi did not know (him).’ Contextual Conditions and Constraints in Chinese Dangling Topics: … 21 (23) a. 一輛車 (一天) 三千。 Yi-liang che (yi-tian) san-qian. one-CL car one-day 3000 ‘One car is three thousand dollars a day.’ b. (一輛車) 一天 三千。 (Yi-liang che) yi-tian 3000. one-CL car one-day three-thousand ‘One day is three thousand dollars.’ According to the paradigm exemplified in (22-23), we further argue that the sentence-initial NPs at issue are by no means subjects. They are topics because they are not the actual units for the prices to be evaluated. The subjects are the elements elsewhere, such as the QPs themselves or price NPs, e.g. jia-qian ‘price’, ding-jia ‘fixed price’, etc. The former assumption that the subjects could be the QPs themselves is made from the conversation below: (24) a. 那種豆子 一斤 三十 兩斤 五十。 Na-zhong douzi yi-jin san-shi liang-jin wu-shi. that-CL bean one-catty 30 two-catty 50 ‘That kind of bean, one catty is thirty dollars and two catties fifty dollars.’ b. 我 要買 一斤。 Wo yao mai yi-jin. I want buy one-catty ‘I would like to buy one catty.’ Both yi-jin ‘one catty’ and liang-jin ‘two catties’ are the actual units for selling and buying in (24). In other words, The QPs can be viewed as genuine NPs or quantificational units for real price evaluation while the initial NP Na-zhong douzi ‘that kind of bean’ is the topic being talked about. Another observation is that it is fairly plausible that the subject is price NP such as jia-qian ‘price’ which is often omitted in discourse since price negotiation is a CG in such context. Its occurrence is cumbersome. To wit: 22 WU Jiayi (25) a. 鳳梨 一箱 五百塊錢。 Fengli yi-xiang wu-bai-kuai qian. pineapple one-CL five-hundred-CL money ‘As for pineapple, one box is five hundred dollars.’ b. ?鳳梨 一箱 價錢 五百塊錢。 ?Fengli yi-xiang jia-qian wu-bai-kuai qian. pineapple one-CL price five-hundred-CL money ‘As for pineapple, One box is five hundred dollars.’ c. 鳳梨 一箱 定價/特價 五百塊錢。 Fengli yi-xiang ding-jia/te-jia wu-bai-kuai qian. pineapple one-CL fixed-price/bargain-price five-hundred-CL money ‘As for pineapple, the fixed/bargain price is five hundred dollars per box.’ As exemplified in (25b-c), unless it is contextually necessary to make a distinction between a fixed price, ding-jia ‘fixed price’ and a bargain price, te-jia ‘bargain price’, the price NP is often dropped to avoid redundancy. Having seen the assumptions concerned with the candidates for subject above, we are convinced at this point that the QPs at issues are the subjects with the price NP jia- qian being presupposed. The price NP claims the subjecthood only when there is a particular price to make. 3.4 The contrast set proposal Taking into account Erteschik-Shir’s notion of contrastive topic we have already elaborated in the previous section, we propose that the sentences at issue are in fact contrastive dangling topics. Specifically, the occurrences of the initial NPs are in fact licensed by the pragmatic condition requiring that one or each member of a contextually available contrast set be selected. Consider the following exchanges. (26) a. 你們 機車 怎麼賣？ Nimen jiche zeme mai? your motorbike how sell ‘How much is your motorcycle?’ b. ? 輕型 (一台) 五萬 重型 (一台) 六萬。 ?Qingxing (yi-tai) wu wan zhongxing (yi-tai) liu wan. Light-type one-CL 50000 heavy-type one-CL 60000 ‘Scooter is fifty while motorcycle is sixty thousand dollars.’ Contextual Conditions and Constraints in Chinese Dangling Topics: … 23 c. ? 三陽機車 (一台) 五萬。 ? Sanyang jiche (yi-tai) wu wan. Sanyang motorbike one-CL 50000 ‘Sanyang motorbike is fifty thousand dollars.’ d. ?? (一台) 五萬。 ?? (yi-tai) wu wan. one-CL 50000 ‘One motorbike is fifty thousand dollars.’ e. 三陽機車 輕型 一台 五萬； Sanyang jichetopic1 qingxingtopic2 yi-taisubject wu wan; Sanyang motorbike light-type one-CL 50000 兩台 特價 七萬。 liang-tai topic3 te-jiasubject qi wan. two-CL bargain-price 70000 ‘As for Sanyang motorbike, scooter is fifty thousand dollars. Two scooters are seven thousand dollars as discount.’ (26a) is a question while (26b-e) are possible responses. (26b-d) are not ungrammatical. They are contextually awkward because the topics for example motor brand in (26b) and motor type in (26c) which are expected to be selected are missing. (26d) is even worse because both topics are absent. (26b-d) can be saved only when the topic selections are carried out either in the responses themselves or in the question (26a) prior to the responses. One the other hand, (26e) is satisfactory because it exhibits multiple topics necessary with the motor brand Sanyang jiche ‘Sanyang motorbike’ being the main while the others such as the motor size and quantity secondary. The subscripts indicate the identifications of topic and subject. The multiple topics involved are shown below: (27) [Motor Brand]topic1; [Motor Type]topic2; [Motor Quantity]topic3 Along a similar line to this analysis, sentence (21) that Huang and Ting use to argue against Pan and Hu’s set-member theory can also be accounted for. That is, in (28) the occurrence of the initial NP in fact forces a contrastive reading, entailing that there exists a contrast topic set regarding quantity for evaluation: the initial NP yi-liang che ‘one car’ is a contrastive dangling topic licensed by a contrast set regarding quantity and the quantity yi-liang che ‘one car’ is first selected, providing as a basic unit for the hearer(s) to infer the rental values of other quantities. Based on this unit, the speaker continues to utter another sentence in which the quantity liang-liang che ‘two cars’ is subsequently selected as a discountable unit. This echoes with Erteschik-Shir’s analysis that even (non-specific) indefinites can be topicalized if they are contrastive. What is 24 WU Jiayi old and given here is not a particular quantity of car but rather the contrast set discoursally available. (28) 一輛車 一天 三千； Yi-liang chetopic yi-tiansubject san-qian; one-CL car one-day 3000 ‘One car is three thousand dollars a day.’ 兩輛車 (一天) 特價 五千。 liang-liang chetopic (yi-tian)CG te-jia subject 5000. two-CL car (one-day) bargain-price five thousand ‘Two cars are five thousand dollars a day as discount.’ The analysis of contrastive topics in section 2 and 3 above is just to facilitate discussion. Both types of topic are demonstrated to involve topic selection from contrast set. However, it does not matter when and how the contrast set condition is satisfied in discourse. For example, one may easily find seeming counterexamples, as in (29) and (30), and argues that the initial NPs are not necessarily contrastive. They are old topics9 that can be derived in the discourse. However, these examples do not amount to present a contradiction to our contrast set proposal at all. They only present another context where the contrast set condition can be satisfied. That is, the presence of the NPs which are often dropped in (29b) and (30b) entails the contrast sets discoursally available. The selection requirement of the sets however has been satisfied already in the previous immediate discourse. The presence of the NPs which can be dropped directly without compromising acceptability is simply to repeat and confirm what has been selected. This explains why they are licensed by the condition but less contrastive than the ones discussed extensively above. (29) a. 玉米呢？ Yumi ne? corn question-particle (Q) ‘How about the corns?’ b. (玉米) 我 炒(了)火腿。 (Yumi) wo chao (le) huotui. Corn I fry ASP ham ‘I fried hams with the corns.’ 9 In Erteschik-Shir’s classification of topics, old topics are the referents that must have been mentioned in the immediate discourse, or else they can be derived from a previously mentioned topic. Contextual Conditions and Constraints in Chinese Dangling Topics: … 25 (30) a. 那種豆子 多少錢？ Na-zhong douzi duoshao qian? that-CL bean how-much money ‘How much is it for that kind of bean?’ b. (那種豆子) 一斤 三十塊錢。 (Na-zhong douzi) yi-jin san-shi-kuai qian. that-CL bean one-catty 30 CL money ‘That kind of bean (topic) is thirty dollars per catty.’ In a nutshell, the initial NPs discussed above are dangling topics properly licensed by the contrast set condition. They are what the sentences are about in discourse and are the loci for further assessment. 4 The event type This section revisits the event type which is often associated with the connective, xingkui ‘fortunately’. The often cited is the example below. (31) 那場火 幸虧 消防隊來得快。 Na-chang huo xingkui xiaofang-dui lai-de-kuai. that-CL fire fortunately fire-brigade come-DE-fast ‘As for that fire, fortunately the fire brigade came quickly; (otherwise)…’ 4.1 The debate In the literature, the initial NPs of this type are analyzed as either syntactically-licensed NPs related to a position in recovered main clause in discourse (Shi, 2000), 10 10 Shi claims that the initial NP in such construction is related to a position in the recovered main clause and thus cannot be analyzed as a dangling topic, as exemplified in (1) and (2). 26 WU Jiayi semantically-licensed topic NPs by the cause-effect relation (Pan & Hu, 2002), or NP adverbials which need not be subcategorized (Huang & Ting, 2006). The debate begins with arguing against Shi’s recovering analysis by Pan and Hu, and then Huang and Ting. First, Pan and Hu argue that the recovering analysis is undesirable since the gap of the recovered main clause for the initial NP does not necessarily guarantee the availability of the un-recovered sentence, as exemplified in (32). A semantic account, namely a cause-effect relation, is thus proposed to address the issue. (32) a. 這件大事 幸虧 張校長來了， Zhe-jian da shi xingkui Zhang-Xiaozhang lai le, this-CL big issue fortunately Zhang-Principal come ASP 要不然 我還不知道 如何處理。 yaoburan wo hai bu zhidao ruhe chuliøi . otherwise I still NEG know how deal-with ‘As for this big issue, fortunately Principal Zhang has come; otherwise I do not know how to deal with it.’ (1) Na-chang huoi xingkui xiaofang-dui lai-de-kuai, that-CL fire fortunately fire-brigade come-DE-fast buran ∅i jiu hui shao-si bu-shao ren. Otherwise really will burn-die not-few person ‘As for that fire, fortunately the fire brigade came quickly, or (it) would have killed many people.’ (2) Na-chang huoi xingkui xiaofang-dui lai-de-kuai, that-CL fire fortunately fire-brigade come-DE-fast Buran na-cii women dou hui shao-si. Otherwise that-time we all will burn-die ‘As for that fire, fortunately the fire brigade came quickly; otherwise we would all have been burnt to death at that time.’ The sentence-initial NPs nachang huo ‘that fire’ in both (1) and (2) are related to certain positions in the recovered main clauses, so nachang huos ‘that fire’ are not dangling topics. Shi further continues that if the adverb xingkui ‘fortunately’ in such construction is eliminated, the sentence becomes unrecoverable but is still well-formed. (3) Na-chang huoi xiaofang-dui lai-de-kuai, that-CL fire fire-brigade come-DE-fast buran ∅i bu-shao ren. Otherwise burn-die not-few person ‘At the time of that fire, the fire brigade came quickly.’ In this case, the initial NP which is not related to any position inside the following clause is a sentential adverbial. Contextual Conditions and Constraints in Chinese Dangling Topics: … 27 b. *這件大事 幸虧 張校長來了。 *Zhe-jian da shi xingkui Zhang-Xiaozhang lai le. this-CL big issue fortunately Zhang-Principal come ASP ‘As for this big issue, fortunately Principal Zhang has come.’ Second, Huang and Ting also consider Shi’s analysis problematic in that there is unnecessarily a gap in the recovered clause for the sentence-initial NP to be syntactically licensed, as illustrated in (33). (33) 那場火 幸虧 消防隊來得快， Na-chang huo xingkui xiaofang-dui lai-de-kuai, that-CL fire fortunately fire-brigade come-DE-fast 要不然 他 早就死了。 yaoburan ta zao jiu si le. otherwise he early really die ASP ‘As for the fire, fortunately the fire brigade came quickly; otherwise he would have been dead.’ Nevertheless, Huang and Ting do not concur with Pan and Hu’s semantic solution because evidence such as (34) suggest that the proposed cause-effect relation does not hold at all. (34) a. 那場溫布頓網球賽 幸虧 大雨停了。 Na-chang Wenbudun wang-qiu sai xingkui da-yu ting le. that-CL Wimbledon net-ball match fortunately big-rain stop ASP ‘As for that Wimbledon match, fortunately the heavy rain stopped.’ b. 台中縣松鶴部落 幸虧 土石流 Taizhong-xian Songhe-buluo xingkui tu-shi-liu Taichung-county Songhe-village fortunately soil-stone-slide 已停止氾濫。 yi tingzhi fanlan. already stop overflow ‘As for the Songhe village in Taichung County, fortunately the landslide already stopped spreading.’ Instead, by analogy to NPs in English such as yesterday, that way, etc., a conclusion is thus made that the initial NPs of this type are NP adverbials. Thus far, we side with their objection to the semantic solution. However, in what follows, we will argue against any possible structural analyses based on acceptability downgrade. A dangling topic hypothesis will then be analyzed that the initial NPs are in 28 WU Jiayi fact nominalized event NPs and, under Erteschik-Shir’s framework, old topics. They occur in context where a speaker feels an urge to express a comment relevant to a topic. 4.2 The PP-reduced form account One structural possibility is to consider the sentence-initial NPs in (31) and those below are PP-reduced form adverbials. (35) a. 那場比賽 幸虧 大雨停了。 Na-chang bisai xingkui dayu ting le. that-CL match fortunately big-rain stop ASP ‘As for that match, fortunately the heavy rain stopped; (otherwise)…’ b. 那次地震 幸虧 房子堅固。 Na-ci dizhen xingkui fangzi jiangu. that-CL earthquake fortunately house solid ‘As for that earthquake, fortunately the houses were firm and solid; (otherwise)…’ c. 那次搶案 幸虧 行員機警。 Na-ci qiangan xingkui hang-yuan jijing. that-CL robbery fortunately bank-staff alert ‘As for that robbery, fortunately the bank-staff was alert; (otherwise)…’ This assumption is untenable because it would be a brute force deleting the claimed preposition, such as zai ‘at’. Specifically, when an adverbial marker such as zai ‘at’ is added to all the initial elements in (35), we find that the sentences incur slight infelicity compared to the un-recovered ones, as illustrated in (36). It is the redundancy of the adverbial marker zai ‘during’ that contributes to the acceptability downgrade. Given the undesirable results in (36), it is questionable that the initial NPs are derived from the reduction of the PP adverbials. Put differently, it is suspicious that the sentences in (31) and (35) are derived from the redundancy of the recovered sentences in (36) for such a derivation is not economical. (36) a. ? 在那場比賽 幸虧 大雨停了。 ? Zai na-chang bisai xingkui dayu ting le. at that-CL match fortunately big-rain stop ASP ‘During that match, fortunately the heavy rain stopped; (otherwise)…’ Contextual Conditions and Constraints in Chinese Dangling Topics: … 29 b. ? 在那次地震 幸虧 房子堅固。 ? Zai na-ci dizhen xingkui fangzi jiangu. at that-CL earthquake fortunately house solid ‘During that earthquake, fortunately the houses were firm and solid; (otherwise)…’ c. ? 在那次搶案 幸虧 行員機警。 ?Zai na-ci qiangan xingkui hang-yuan jijing. at that-CL robbery fortunately bank-staff alert ‘During that robbery, fortunately the bank-staff was alert; (otherwise)…’ The untenable of the PP-reduced form analysis is further supported by the examples below in which the connective, xingkui ‘fortunately’, is eliminated under the assumption that the sentences without the adverb xingkui ‘fortunately’ should be as perfectly acceptable as the un-recovered ones. (37) a. ?? 在那場比賽 大雨停了。 ?? Zai na-chang bisai dayu ting le. at that-CL match big-rain stop ASP ‘During that match, fortunately the heavy rain stopped; (otherwise)…’ b. ?? 在那次地震 房子堅固。 ?? Zai na-ci dizhen fangzi jiangu. at that-CL earthquake house solid ‘During that earthquake, fortunately the houses were solid; (otherwise)…’ c. ?? 在那次搶案 行員機警。 ??Zai na-ci qiangan hang-yuan jijing. at that-CL robbery bank-staff alert ‘During that robbery, fortunately the bank-staff was alert; (otherwise)…’ When the adverb xingkui ‘fortunately’ is dropped, the sentences obtained in (37) are even more downgraded. The sentences in (37) might be acceptable only in a limited contrastive context in which the presence of zai ‘at’ is used for locative emphasis. 4.3 The NP adverbial account Another structural possibility is that the initial NPs are adverbials themselves. However, this analysis does not hold either. Consider the sentences below. 30 WU Jiayi (38) a. *那場比賽 大雨停了。 *Na-chang bisai dayu ting le. that-CL match big-rain stop ASP ‘During that match, fortunately the heavy rain stopped; (otherwise)…’ b. *那次地震 房子堅固。 *Na-ci dizhen fangzi jiangu. that-CL earthquake house solid ‘During that earthquake, fortunately the houses were firm and solid; (otherwise)…’ c. *那次搶案 行員機警。 *Na-chang qiangan hang-yuan jijing. that-CL robbery bank-staff alert ‘During that robbery, fortunately the bank-staff was alert; (otherwise)…’ Again, when the adverb xingkui ‘fortunately’ is removed, the sentences obtained in (38) are the least acceptable compared to all the sentences discussed above. The adverbial analysis apparently cannot explain why the sentences without the adverb sound even worst. 4.4 The dangling topic hypothesis Aside from the issue of acceptability downgrade, there is another critical reason to reject the sentence-initial NPs at issue as PP-reduced forms or NP adverbials. The initial NPs unnecessarily entail the spatio-temporal parameters: (39) a. 那場比賽 你一定沒有興趣。 Na-chang bisai ni yiding meiyou xingqu. That-CL match you certainly NEG interest ‘You are certainly not interested in the match.’ b. 那場比賽 李四早就知道結果了。 Na-chang bisai Lisi laozao jiu zhidao jieguo le. That-CL match Lisi very-early already know result ASP ‘Lisi has already known very early the result of the match.’ The initial NPs above have little to do with the spatio-temporal parameters. They apparently refer to the events per se. They are nominalized event NPs and old topics which have been mentioned in the immediate discourse or can be derived from a previously mentioned topic. They are licensed in context where a speaker feels an urge to express a comment relevant to a topic. Contextual Conditions and Constraints in Chinese Dangling Topics: … 31 5 The (non-)idiom Chunk Type This section covers the (non-)idiom chunk dangling topics, as in (40), which are argued first by Shi (2000), and then Huang and Ting (2006) as subject-predicate constructions. Under Erteschik-Shir’s framework of topic, we will argue instead that the chunks are fixed clauses predicated of the old topic tamen ‘they’. (40) a. 他們 [大魚吃小魚]。 Tamen [da-yu chi xiao-yu]. They big-fish eat small-fish ‘They act according to the law of the jungle.’ b. 他們 [我看你，你看我]。 Tamen [wo kan ni, ni kan wo]. They I look you you look me ‘They look at each other.’ 5.1 Posing the problems Shi reasons that the idiom chunks are predicates or verbs, given the fact that the preverbal NPs contained in the idiom chunks do not refer to any NP in the sentences. Adverb zhuanmen ‘specifically’ is particularly used to justify the subjecthood of the initial NPs and the predicatehood of the idiom chunks. However, those claims are challenged by Pan and Hu (2006) who provide the following examples below in which the non-idiom chunks have referential subjects and the sentence in (41b) can pass the zhuanmen ‘specifically’ test even though the chunk is obviously a chunk with two full- fledged sentences. (41) a. 他們 [你指責我不對，我抱怨你不好]。 Tamen [ni zhize wo bu dui, wo baoyuan ni bu hao]. They you blame I NEG right I complain you NEG good ‘They blame each other and complain about each other.’ b. 他們倆 專門[小王先來，小李後到]。 Tamen liang zhuanmen [XiaoWang xian lai, XiaoLi hou dao]. they two specifically XiaoWang first come XiaoLi after arrive ‘As for them two, XiaoLi always comes after XiaoWang.’ Following Shi’s structural approach, Huang and Ting add another aspect marker test, specifically zai ‘at’ test, and argue that the bracketed idioms listed in (40) should be analyzed as predicates that should have undergone a certain kind of reanalysis as unanalyzable chunks, viewed as a complex V-node in the hierarchical representation. Moreover, they go a step further and argue that the examples in (41) are not dangling 32 WU Jiayi topics but normal sentences with initial PP-reduced form adverbials: the initial NP tamen ‘they’ is reduced from the PP-adverbial zai tamen zhizhong ‘among them’. Appealing as it might be, this account nevertheless has a number of problems. First, the syntactic analysis is again uneconomical to assume because it entails that the sentences in (41) are generated from unnatural sentences, as in (42), and, likewise, the deletion of zai ‘at’ and zhizhong ‘midst’ would be a brute force. (42) a. ??在他們之中 [你指責我不對，我抱怨你不好]。 ??Zai tamen zhizhong [ni zhize wo bu dui, wo baoyuan ni bu hao]. at they midst you blame I NEG right I complain you NEG good ‘Among them, one blames and complains about another.’ b. ??在他們倆之中 專門[小王先來，小李後到]。 ??Zai tamen liang zhizhong zhuanmen [XiaoWang xian lai, XiaoLi hou dao]. at they two midst specifically XiaoWang first come, XiaoLi after arrive ‘Between them, XiaoLi always comes after XiaoWang.’ The second problem concerns the different treatments of syntactic representation between (40b) and (41). It is rather contradictory that the chunk in (40b) should be analyzed as a complex predicate under V-node whereas the ones in (41) as normal sentences. In what follows, the proposal that the chunks are fixed clauses will be re- considered and the zai ‘at’ test, developed by Huang and Ting to support Shi’s subject- predicate viewpoint and argue for their complex V-node analysis, will be refuted. A pragmatic topic-comment structure will be proposed instead to better represent the syntactic structures of the sentences in question. 5.2 The dangling topic hypothesis In the literature various grammatical functions of Chinese idioms, including subject, predicate, object, determiner, adverbial, complement, etc., have been widely discussed (Li, 1997; Zheng, 2005). Away from the traditional view that analyzes all Chinese idiom chunks as lexicons or phrases, Ma (1998), Wen (2006) and many others propose that some Chinese idioms alone can be sentences themselves. Their viewpoint supposes that there must be at least two kinds of idiom in terms of morphosyntactic characteristics. One is the predicate type that is verbal in form while the other is the one that is sentential in structure. Both can be predicated of the initial NP tamen ‘they’, as exemplified in (43) and (44) respectively. Contextual Conditions and Constraints in Chinese Dangling Topics: … 33 (43) a. 他們 [以德報怨]。 Tamen [yi de bao yuan]. They use kindness reply resentment ‘They render good for evil.’ b. 他們 [吃裡扒外]。 Tamen [chi li pa wai]. they eat inside claw outside ‘They live on somebody while helping others secretly.’ (44) a. 他們 [毛遂自薦]。 Tamen [Maosui zi-jian]. They Maosui self-introduce ‘They introduce themselves.’ b. 他們 [鶼鰈情深]。 Tamen [jian-die qing-shen]. They lovebirds-flatfish emotion-deep ‘They love each other very much.’ What’s more, there exist another chunks called proverbs in Chinese that can also be predicated of the initial NP tamen ‘they’: (45) a. 他們 [塞翁失馬焉知非福]。 Tamen [sai-weng shi ma yan zhi fei fu]. They frontier-man lose horse how know NEG fortune ‘Their misfortune might be a blessing in disguise.’ b. 他們 [牆頭草兩面倒]。 Tamen [qiang-tou-cao liang-mian dao]. They wall-head-grass two-side fall ‘They bend with the wind.’ Aligning with the classification made above, we argue that the idioms in (43) are predicated of the “subject” topic NP tamen ‘they’ while the sentential chunks in (44) and (45) are predicated of the “dangling” topic NP tamen ‘they’. 5.3 Justifying the hypothesis The hypothesis can be justified by the fact that the preverbal NPs contained in the chunks can be as referential as normal subjects and need not hold any syntactic and semantic relations with the sentence-initial NPs. The complex V-node analysis made from the zai ‘at’ test will also be rejected as an additional support. 34 WU Jiayi 5.3.1 The subjecthood of the preverbal NPs To begin with, the preverbal NPs of the chunks in (44) and (45) often idiomatically refer to the sentence-initial NPs, as shown in (46). This shows that there is no need to reject them as non-subjects since they are as referential as normal subjects. (46) a. 他們 [鶼鰈情深]。 Tameni [jian-diei qing-shen]. They lovebirds-flatfish emotion-deep ‘They love each other very much.’ b. 他們 [牆頭草兩面倒]。 Tameni [qiang-tou-caoi liang-mian dao]. They wall-head-grass two-side fall ‘They bend with the wind.’ Second, one may also suspect that they are cases of left-dislocation or topic promoting construction. However, consider the paradigm in (47-48): (47) a. 他們 [虎視眈眈]。 Tameni [hui shi dandan]. They tiger see covetously ‘They glare like a tiger eyeing its prey.’ b. 他們 視眈眈。 *Tamen shi dandan. They see covetously ‘They glare like a tiger eyeing its prey.’ (48) a. 他們 [塞翁失馬焉知非福]。 Tameni [sai-wengi shi ma yan zhi fei fu]. They frontier-man lose horse how know NEG fortune ‘Their misfortune might be a blessing in disguise.’ b. *他們 失馬焉知非福。 *Tamen shi ma yan zhi fei fu. They lose horse how know NEG fortune ‘Their misfortune might be a blessing in disguise.’ In (47b) and (48b) the co-referential preverbal NPs contained in the chunks are replaced by the alleged left-dislocated NPs. The unacceptability reveals that the sentence initial NPs are not left-dislocated elements and therefore does not bear any syntactic relation with the preverbal NPs. This analysis is further supported by the Contextual Conditions and Constraints in Chinese Dangling Topics: … 35 examples below where there is by no means any co-referential relationship between the sentence-initial and the preverbal NPs. (49) a. 他們 [鬼打牆]。 Tameni [gui*i/j da qiang]. They ghost hit wall ‘They are in a loop.’ b. 他們 [禍不單行]。 Tameni [huo*i/j bu dan xing]. They misfortune NEG alone go ‘Their misfortunes never come alone.’ c. 他們 [罪不可赦]。 Tameni [zui*i/j bu ke she]. They crime NEG can remit ‘Their sentences cannot be remitted.’ 5.3.2 The unreliable of zai test To further support their claim that the idiom chunks at issue belong to one particular type of predicate, viewed as an unanalyzable complex V-node in the syntactic representation, Huang and Ting (2006) continue that the predicatehood of the idiom chunks can be further verified by the fact that they can be preceded by aspect markers such as progressive zai : (50) a. 他們 一直在[大魚吃小魚]。 Tamen yizhi zai [da-yu chi xiao-yu]. They continually ASP big-fish eat small-fish ‘They have been bullying the weaker.’ b. 他們 正在[我看你，你看我]。 Tamen zheng zai [wo kan ni, ni kan wo]. They right ASP I look you you look me ‘They are looking at each other.’ However, the assumption that the idiom chunks should be predicates syntactically under V-node is questionable. We provide three major reasons below to reject the zai test. First, the applicability of the zai test is highly restrictive and, as shown below, the constructions at issue cannot be construed with other aspect markers, such as le, zhe and guo, etc. 36 WU Jiayi (51) *他們 [大魚吃小魚] 了/著/過。 *Tamen [da-yu chi xiao-yu] le/zhe/guo. They big-fish eat small-fish ASP ‘They already bullied the weaker.’ Second, the zai test is unreliable since, as illustrated in (52), it would predict that sentences such as (41a) are acceptable subject-predicate utterances while similar constructions such as (41b) are only barely acceptable. This prediction in turn contradicts Huang and Ting’s initial position that both are normal subject-predicate constructions with initial PP-reduced form adverbials: (52) a. 他們 常常在 [你指責我不對，我抱怨你不好]。 Tamen changchang zai [ni zhize wo bu dui, wo manyuan ni bu hao]. They often ASP you blame I NEG right I complain you NEG good ‘They often blame and complain each other.’ b. ? 他們倆 專門在 [小王先來，小李後到]。 ?Tamen liang zhuanmen zai [XiaoWang xian lai, XiaoLi hou dao]. they two specifically ASP XiaoWang first come XiaoLi after arrive ‘As for them two, XiaoLi is always coming after XiaoWang.’ Third, the test is misleading, especially when we consider that the progressive marker zai in fact modifies the verb phrase (VP) of the chunk: (53) a. 他們 一直在 [黑吃黑]。 Tamen yizhi zai [hei chi hei]. They continually ASP black eat black ‘They have been always bullying others.’ b. *他們 一直在 [媳婦欺負婆婆]。 *Tamen yizhi zai [xifu qifu popo]. They continually ASP daughter-in-law bully mother-in-law ‘They have been always bullying their mother-in-laws.’ c. 他們 一直在 [欺負婆婆]。 Tamen yizhi zai qifu popo. They continually ASP bully mother-in-law ‘They have been always bullying their mother-in-laws.’ The paradigm above reveals not only that the sentence following the progressive marker zai in (53a) is a chunk while the one in (53b) is not, but also that the progressive marker zai is merely used to modify the VP, as exemplified in (53c). As it stands, an explanation with respect to the acceptability contrast between (53a) and (53b) Contextual Conditions and Constraints in Chinese Dangling Topics: … 37 proceeds as follows. Given that the subject of the subordinate clause in (53a) is indispensable of the “unanalyzable” clause, there should be no surprise that the sentential chunk as a whole can be construed with the progressive marker zai that is in general used to modify the VP of the chunk. Conversely, the unacceptability of (53b) results from the interference of the subject in the “analyzable” clause. 5.3.3 The topic-comment structure In view of the issues we have elaborated above, we are urged to reject the complex V- node analysis and argue that Rizzi’s (1997) TopP proposal, specifically a topic-comment structure, better accounts for the relation between the sentence-initial NPs and the following clauses. Consider the topic-comment structure in (54) and see how the construction at issue is represented in (55) where IP stands for a sentence: (54) TopP (Rizzi, 1997): The specifier is the topic and its complement is the comment. (55) Top-Comment Structure The topic-comment analysis has its explanatory power in that it adequately captures the relations among the syntactic units of “double-chunk” constructions. TopP XP Top’ Top ﾟ YP XP = topic YP = comment TopP NP Top’ Top ﾟ IP tamen ‘they’ dayu chi xiaoyu ‘act according to the law of the jungle’ 38 WU Jiayi Consider the double-chunk constructions in (56) where the second chunk is adjoined. Unlike (40b) or (41), the second chunk is adjoined merely for reinforcement: (56) a. 他們 [黑吃黑] [狗咬狗]。 Tamen [hei chi hei] [gou yao gou]. They black eat black dog bite dog ‘They act according to the law of the jungle.’ b. 他們 [花開富貴] [子孫滿堂]。 Tamen [hua kai fu-gui] [zi-sun man tang]. They flower bloom fortune offspring full house ‘They grow with prosperity and have plenty of offspring.’ c. 他們 [好漢不吃眼前虧] [識時務者為俊傑]。 Tamen [hao-han bu chi yan-qian kui] [shi shi-wu zhe wei jun-jie]. They good-man NEG eat eye-front loss know reality person is hero ‘They are the wise men who suit their actions to the time.’ Under the subject-predicate framework, the second unanalyzable complex V- chunk gou yao gou ‘dog bites dog’ in (56a) is adjoined to the VP, as represented in (57). However, the structure in (57) is inadequate since the representation is structurally ambiguous in that either one of the complex V-chunks could be the head of the VP. Specifically, the geometry shows that the second chunk is an adjunct predicate; however, the categories represented in the structure are incompatible with the geometry since the categories indicate that either one of the complex-Vs could potentially be the head of the VP. This representation thus suggests that there should be an interpretation burden, contrary to the fact. Contextual Conditions and Constraints in Chinese Dangling Topics: … 39 (57) Subject-Predicate Structure By contrast, analyzed under the topic-comment geometry, the IP-chunk gou yao gou ‘dog bites dog’ is only adjoined to TopP, as shown in (58). This representation clearly distinguishes the difference between complement and adjunct. (58) Topic-Comment Structure The top-comment structure above properly specifies the legal relations among the syntactic units of the sentence: the head, Top ﾟ, subcategorizes the topic tamen ‘they’ as its specifier, the IP-chunk hei chi hei ‘dark eats dark’ as its complement, and finally the IP-chunk gou yao gou ‘dog bites dog’ as its adjunct. This analysis is also flexible in IP NP tamen ‘they’ I ’ Infl VP V’ Complex-V NP hei chi hei ‘dark eats dark.’ Complex-V gou yao gou ‘dog bites dog.’ V’ NP Top’ IP tamen ‘they’ IP hei chi hei ‘dark eats dark.’ gou yao gou ‘dog bites dog.’ Top’ Top ﾟ TopP 40 WU Jiayi that it allows for a distinction between verbal and sentential chunks and a possible construction where the first chunk is an IP while the other a VP, or vice versa. Chunks for most language users are opaque and not commonly used. They are used only when a speaker is confident in usage and feels an urge in context to express a comment relevant to a topic in a succinctly figurative or idiomatic way. 6 The licensing conditions and constraints The present paper has thus far analyzed that there exist dangling topics in Chinese. They are either contrastive or old topics. This section elaborates the licensing conditions and constraints that guarantee the well-formed dangling topic structures. 6.1 The literature review In light of the fact that the acceptability contrast presented below cannot be adequately accounted for by the traditional aboutness condition, Pan and Hu (2009) postulate set intersection as the topic licensing condition.11 (59) a. *幼兒園的小孩 張三 教兒子畫畫。 *Youeryuan de xiaohai Zhangsan jiao erzi huahua. Kindergarten MOD children Zhangsan teach son draw-poctures ‘As for the children in the kindergarten, Zhangsan teaches his son to draw pictures.’ b. 幼兒園的小孩 張三 只教兒子畫畫。 Youeryuan de xiaohai Zhangsan zhi jiao erzi huahua. Kindergarten MOD children Zhangsan only teach son draw-poctures ‘As for the children in the kindergarten, Zhangsan only teaches his son to draw pictures.’ As Pan and Hu contend, (59a) is unacceptable because no set intersection occurs in it. (59b) is acceptable because the intersection of the topic set with the set projected from the focus element erzi ‘son’ is not empty. In addition, aware of the fact that the licensing condition alone cannot guarantee the well-formedness of all topics, they further add the interpretation condition (predicate-subject condition),12 whereby topics can be properly interpreted because 11 Topic Licensing Condition (Pan & Hu, 2008, 1970). A topic can be licensed iff (i) there is a set Z induced by a variable x in the comment, and (ii) the set Z thus generated does not produce an empty set when intersecting with the set T denoted by the topic. 12 Topic Interpretation Condition (Pan & Hu 2009). In a configuration Σ = [TopP X [IP … Y …]], the topic X is properly interpreted if it can form a subject-predicate relation with an element Y in the comment clause, where Y is the subject and X the predicate. Contextual Conditions and Constraints in Chinese Dangling Topics: … 41 they can form a predicate-subject relation with elements contained in comment clauses. This explains why (60a) is well-formed while (60b-c) is ill-formed. (60) a. 水果 我喜歡蘋果。 Shuiguo wo xihuan pingguo. Fruit I like apple ‘As for fruits, I like apples.’ b. *蘋果 我喜歡水果。 *Pingguo wo xihuan shuiguo. Apple I like fruitfascinating ‘*As for apples, I like fruits.’ c. *蘋果 我喜歡香蕉。 *Pingguo wo xihuan xiangjiao. Apple I like banana ‘*As for apples, I like bananas.’ Specifically, in (60a), the initial NPs is able to form a predicate-subject relation with the element in the comment clause. However, the relation does not hold in (60b-c). Consider the contrast below: (61) a. 蘋果 是水果。 Pingguo shi shuiguo. Apple be fruit ‘Apples are fruits.’ b. *水果 是蘋果。 *Shuiguo shi pingguo. Fruit be apple ‘*Fruits are apples.’ c. *香蕉 是蘋果。 *Xiangjiao shi pingguo. Banana be apple ‘*Bananas are apples.’ The analysis is fascinating; nevertheless, the conditions would be too strict to accommodate other potential topics. It for instance may not provide an adequate explanation to some of the nominal-predicate topics such as (21), repeated in (62). Neither the set intersection nor the interpretation condition is able to account for the well-formedness of the topic. 42 WU Jiayi (62) 一輛車 一天 三千。 Yi-liang che yi-tian san-qian. one-CL car one-day 3000 ‘One car is three thousand dollars a day.’ 6.2 The contrast set condition In the previous sections we have proposed that the contrast set condition is one of the topic licensing conditions in context which requires that one or each member be selected. This is when a speaker assumes that a hearer expects her/him to select from a discoursally available set or provide an assessment to a topic which has been previously selected. The acceptability contrast in (59) is thus explicable. (59a) is unacceptable because two potential set members are selected simultaneously from the contrastive topic set regarding the potential candidates to teach drawing; on the other hand, (59b) is acceptable because only one member is selected from the contrast set youeryuan de xiaohai ‘the children in the kindergarten’. The same holds true for the acceptability contrast in (60). (60a) is acceptable because the requirement is satisfied: there exists a contrast set which is explicitly expressed by shuiguo 'fruit’, from which one member of the set is selected. On the other hand, (60b) and (60c) are unacceptable because the requirement is not met. In fact, two discoursal questions can be re-constructed below to potentially elicit responses (60b-c). However, whichever question they respond to, the requirement for the contrast set condition is not met. Consider the questions and how the contrast set condition works to eliminate both (60b-c). (63) a. 你喜歡什麼水果？ Ni xihuan sheme shuiguo? you like what fruit ‘What fruit do you like?’ b. * c. * Contrast set shuiguo Member pingguo shuiguo Contrast set shuiguo Member pingguo xiangjiao Contextual Conditions and Constraints in Chinese Dangling Topics: … 43 The unacceptability of (60b-c) proceeds as follows. As represented in (63b), there exists a contrast set, namely shuiguo ‘fruit’; however, one member and the set itself, which is not one of the members, are selected in (60b). As also shown in (63c), two members are selected simultaneously and this leads to a selection conflict in (60c). (64) a. 你喜歡什麼蘋果？ Ni xihuan sheme pingguo? you like what apple ‘What kind of apple do you like?’ b. * c. * Likewise, as represented in (64b-c), there exists a contrast set pingguo ‘apple’ in (60b-c); however, neither shuiguo ‘fruit’ nor xiangjiao ‘banana’ is the member of the contrast set. Both exemplify an improper selection of contrast set. 6.3 The commentative condition Another condition we are convinced to propose is a commentative relation held between the sentence-initial NPs and the following clauses, a condition when a speaker assumes that a hearer expects her/him to express a contextually relevant comment (point of view) to a topic. The well-formedness of the event and (non)idiom chunk types is licensed by this condition, with the difference being whether the condition is satisfied directly or indirectly. So far as the event type is concerned, the sentence-initial NPs and the comment clauses form an indirect commentative relation in which the latter is indirectly commentative to the former. Take robbery event for example. The acceptability contrast in (65) strongly supports this idea. (65) a. 那場搶案 幸虧 行員機警。 Na-chang qiangan xingkui hang-yuan jijing. That-CL robbery fortunately bank-staff alert ‘As for the robbery, fortunately the bank-staff was alert; otherwise…’ Member shuiguo Contrast set pingguo Member xiangjiao Contrast set pingguo 44 WU Jiayi b. *那場搶案 行員機警。 *Na-chang qiangan hang-yuan jijing. That-CL robbery bank-staff alert ‘As for the robbery, the bank-staff was alert; otherwise…’ c. *這個小偷 (幸虧) 警察來得早。 *Zhe-ge xiaotou (xingkui) jingcha lai-de-zao. this-CL thief (fortunately) police come-DE-early ‘As for this thief, (fortunately) the police came early.’ In (65a) the commentative relation is connected indirectly via the speaker’s comment on one of the salient participants’ actions in the event. The participant is often considered as having an impact on the event. The connective adverb xingkui is often required in order to connect the topic NP and its following comment clause. In other words, by making use of the connective, the two are somehow commentatively related. The indirect relation can be captured schematically as below. (66) Indirect commentative relation On the other hand, the relation in (65b) is unclear because, without the adverb xingkui ‘fortunately’, the meaning expressed in the comment clause is ambiguous: the situation described in the clause is most likely to be interpreted as a statement about the participant’s personality traits irrelevant to the event itself. The relation in (65c) is not connected at all: the speaker’s view on one participant’s performance bears no relevance to another participant in the event. It is simply illogical to comment a person by commenting another. Relevance does play an essential role. Connective Adverb | | Event Participant | | Comment Predicate Event topic Initial NP Contextual Conditions and Constraints in Chinese Dangling Topics: … 45 The same goes for the (non-)idiom chunk type. The commentative relation in this type is however direct: the following chunk is directly commentative to the sentence- initial topic NP. Relevance has to be observed. Consider the examples below. (67) a. 他們 [罪不可赦] /*[公道自在人心]。 Tamen [zui bu ke she] /*[gongdao zi zai renxin]. They crime NEG can remit justice naturally in man-heart ‘Their crimes are not forgivable.’ b. 這件事 [公道自在人心] /*[罪不可赦] Zhe-jian shi [gongdao zi zai renxin] /*[zui bu ke she]. this-CL matter justice naturally in man-heart crime NEG can remit ‘As for this matter, facts speak louder than words and justice will prevail.’ The paradigm in (67) shows that the chunks used to comment on the topics have to be relevant. The second chunk that follows in each sentence is simply irrelevant. 7 Concluding remarks To recap, in this paper four alleged types of topic have been analyzed as dangling topics in Chinese. The sentence-initial NPs at issue are either contrastive or old topics. The analysis of this study suggests that pragmatically licensing NPs in left-peripheral position is one of the ways that Chinese employs to reflect information structure of topic. The conditions and constraints elaborated are the linguistic and contextual knowledge that a speaker possesses to calculate and package her/his utterances. Two contextual licensing conditions that guarantee the well-formedness of dangling topics in Chinese have been identified. Specifically, the material/instrument and nominal-predicate types are properly licensed under the contrast set condition. And the event and (non-)idiom chunk types are licensed via the commentative condition. The following table shows the topic types, the licensing conditions, and the constraints. Table1: Dangling topics and licensing conditions Dangling type Licensing condition Constraint Material/Instrument Contrast set Proper selection Nominal-predicate Contrast set Proper selection Event Commentative Relevance (Non-)idiom chunk Commentative Relevance 46 WU Jiayi This table provides insights to the other two types of topic that we have not yet discussed about. That is, the sentence in (2) is uttered when ‘neither’, one of the possible options, is selected. Likewise, (6) selects New York as the topic for further assessment. Note that there is no rigid one-to-one relation that each condition is exclusively corresponds to specific types of topic. The relation is rather defined as context-dependent in that one condition is more relevant and decisive than the other. This is especially true as one may find examples where the first two types of topic occur as old topics. Similarly, the last two types of topic appear as contrastive topics. They are contextually licensed regardless of contrastive or old topics they appear in discourse. The bottom line is neither the contrast set nor the commentative condition can be violated. The analysis thus decomposes and lends a further support to the traditional aboutness relation purportedly held between the sentence-initial NPs and the following clauses. The vague aboutness relation can be further justified as a valid licensing mechanism in general, under which either one of the conditions is satisfied. Shi’s refusal to the aboutness relation on the basis of the evidence in (7a), repeated in (68b), is therefore misleading in this regard. Consider the following discourse exchanges: (68) a. 你知道 張校長辭職了？ Ni zhidao Zhang-Xiaozhang cizhi le? you know Zhang-Xiaozhang resign ASP ‘Do you know Zhang-Xiaozhang has resigned?’ b. *這件大事 我知道 張校長辭職了。 *Zhe-jian da-shi wo zhidao Zhang-Xiaozhang cizhi le. this-CL big-issue I know Zhang-Principal resign ASP ‘As for this big issue, I know that Principal Zhang has resigned.’ b’. 這件大事 我知道 李四受了很大的委屈。 Zhe-jian da-shi wo zhidao Lisi shou-le hen-da de weiqu. this-CL big-issue I know Lisi encounter-ASP very-big MOD grievance ‘As for this big issue, I know that Lisi felt so wronged and unjustly treated. In (68b-b’) the contrast set condition is irrelevant and not violated. However, (68b) is ill-formed because the commentative condition is not observed. What was brought up in the following clause is not a relevant comment but simply an unnecessary reiteration of the topic itself. The occurrence of the topic NP even causes a licensing conflict with the object NP since they refer exactly to the same thing. On the other hand, the response in (68b’) is well-formed because what was brought up is the speaker’s comment on the perception of the participant in the event, an indirect comment to the topic. Contextual Conditions and Constraints in Chinese Dangling Topics: … 47 Finally, the result of this study poses an intriguing inquiry that pertains to language typology. Erteschik-Shir discusses language typology in terms of word order and concludes that discourse configurational languages, e. g. Chinese, feature a ranking hierarchy where information structure dominates syntax in determining word order; on the contrary, in configurational languages, e. g. English, syntax dominates information structure. Along that line, we suppose that, in contrast to configurational languages, discourse configurational languages exhibit phenomena of dangling NP topics which need not be syntactically (or semantically) but pragmatically licensed by the aboutness relation. For example, the sentence in (5) suggests that as a discourse configurational language Chinese allows multiple NPs in sentence-initial position as long as they are properly licensed: the topic NP is pragmatically licensed in the left- most position while the QP is syntactically licensed in the subject position. However, as a configurational language English tends to not allow pragmatically licensed NPs. Therefore, similar topic NP in English has to be in subject position while QP in prepositional phrase, or the other way around, as long as they are syntactically licensed. References Chafe, W. L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In C. N. Li (ed.), Subject and Topic (pp. 25--55). New York: Academic Press. Chao, Y.-R. (1968). A Grammar of Spoken Chinese. Berkeley: University of California Press. Erteschik-Shir, N. (2007). Information structure: The syntax-discourse interface. Oxford, UK: Oxford University Press. Gregory, M. L., & Michaelis, L. A. (2001). Topicalization and Left-Dislocation: A Functional Opposition Revisited. Journal of Pragmatics, 33, 1665-1706. Hsin, A.-L. (2002). On Indefinite Subjects in Chinese. Chinese Studies, 20(2), 353-376. Hu, J., & Pan, H. (2009). Decomposing the aboutness condition for Chinese topic constructions. The Linguistic Review, 26, 2-3. Huang, C.-T. J. (1982). Logical relations in Chinese and the theory of grammar. Cambridge, Massachusetts: MIT Press. Huang, C.-T. J. (1984). On the Distribution and Reference of Empty Pronouns. Linguistic Inquiry, 15(4), 531-574. Huang, R.-h., & Ting, J. (2006). Are There Dangling Topics in Mandarin Chinese? Concentric: Studies in Linguistics, 32(1), 119-146. Kiss, E. (2004). The EEP in a Topic-Prominent Language. In P. Svenonius (Ed.) Subjects, Expletives, and the EEP (pp. 107-124). Oxford: Oxford University Press. Larson, R. K. (1985). Bare-NP Adverbs. Linguistic Inquiry, 16(4), 595-621. Li, C., & Thompson, S. (1976). Subject and topic: A new typology of language. In Li, C. N. (1976). Subject and topic: [papers] (pp. 457-489). New York: Academic Press. McCawley, J. D. (1988). Adverbial NPs: Bare or Clad in See-Through Garb?. Language, 64(3), 583-590. 48 WU Jiayi Pan, H., & Hu, J. (2002). Licensing dangling topics in Chinese. Paper presented at the 2002 LSA Annual Meeting in San Francisco, CA, USA. Pan, H., & Hu, J. (2008). A semantic-pragmatic interface account of (dangling) topics in Mandarin Chinese. Journal of Pragmatics, 40(11), 1966-1981. Rizzi, L. (1997). The Fine Structure of Left Periphery. In L. Haegeman (Ed.) Elements of Grammar (pp. 281-337). Dordrecht: Kluwer. Shi, D. (1992). The Nature of Topic Comment Constructions and Topic Chains. Los Angeles: University of Southern California dissertation. Shi, D. (2000). Topic and topic-comment constructions in Mandarin Chinese. Language, 76(2), 383-408. Shyu, S. (2011). The syntax of focus and topic in Mandarin Chinese. Los Angeles: University of Southern California. Strawson, P. E. (1964). Identifying Reference and Truth-Values. Theoria, 30, 86-99. Tsao, F.-F. (1990). Sentence and Clause Structure in Chinese: A Functional Perspective. Taipei: Student Book Co. Xu, L., & Langendoen, D. T. (1985). Topic structures in Chinese. Language, 61, 1-27.