Temporal and aspectual references in Mandarin Chinese

Available online at www.sciencedirect.com

Journal of Pragmatics 44 (2012) 2045--2066 www.elsevier.com/locate/pragma

Temporal and aspectual references in Mandarin Chinese§ Wendan Li * University of North Carolina at Chapel Hill, Asian Studies, 318 New West, Chapel Hill, NC 27599-3267, United States Received 12 January 2012; received in revised form 4 August 2012; accepted 27 September 2012

Abstract This paper proposes three descriptive generalizations about temporal and aspectual references in Chinese. First, temporal and aspectual references in a clause display a positional distinction: temporal location is indicated before the main verb and aspect after the main verb. The distinction characterizes an important pattern of information distribution within the clause; it also provides a unified explanation for a number of word order phenomena discussed in previous synchronic and diachronic studies. Second, temporal and aspectual references are managed at different levels. The former is to a large extent a discourse phenomenon, while the latter is managed at the clause level. Third, in narrative discourse, major devices to indicate perfectivity, i.e., the perfective marker --le and ‘‘bounding expressions,’’ serve as mechanisms to facilitate narrative advancement. These proposals are supported by evidence from natural discourse. © 2012 Elsevier B.V. All rights reserved. Keywords: Temporal and aspectual reference; Syntactic positions; Narrative discourse; Chinese

1. Introduction Temporal and aspectual references are a central issue in linguistic analysis. In inflectional languages, a great deal of the information is conveyed by grammatical markings on the verb, while word order is not particularly involved. Mandarin Chinese (referred to as ‘‘Chinese’’ hereafter), as an isolating language, adopts a different strategy. On the one hand, Chinese is ‘‘strikingly indeterminate on the surface’’ (Smith 1997[1991]) because of its meager grammatical marking. It is ‘‘tenseless,’’ i.e., with no systematic and overt tense marking on the verb. Temporal location of events is conveyed by other means, such as time adverbials, and through context. The language does employ a set of grammatical markers to indicate aspectual distinctions, but the use of these markers is optional rather than obligatory. In natural discourse, the majority of clauses show no use of aspectual markers. This leads to the question of how aspectual information in the ‘‘unmarked’’ clauses is conveyed and interpreted. The word order of Chinese, on the other hand, has been found to play an important role in the grammatical system. A number of observations have led to the claim that the language takes on ‘‘positional distinction in meaning.’’ That is, syntactic constituents carry specific semantic functions because of their positions in the clause alone. For example, NPs before the main verb are definite while those after the main verb are indefinite. Adverbials before and after the main verb also differ in meaning (detail in Section 4.2.3). So far in the analysis of Chinese, temporal/aspectual reference and word order have been treated separately. The present study examines these seemingly unrelated grammatical phenomena together and proposes a unified account.

§ Partial and previous versions of this paper were presented at the Annual Conference of American Counsel of Teaching Foreign Languages in Boston in 2010 and The 19th Annual Conference of the International Association of Chinese Linguistics at Nankai University, Tianjin, China in 2011. I would like to thank the participants of the panels for their invaluable comments. I would also like to thank the anonymous reviewers and Dr. Jonathan Culpeper, Editor-in-Chief of Journal of pragmatics for their insightful suggestions for revision. The remaining errors are mine. * Tel.: +1 919 843 7500. E-mail address: [email protected].

0378-2166/$ -- see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.pragma.2012.09.017

2046

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

I show that, in Chinese, there is a close relation between syntactic positions and temporal/aspectual reference. Generally speaking, situations’ temporal and physical locations are indicated before the main verb, while aspectual information is provided after the verb. The two types of information are also managed at different levels: Temporal interpretation is essentially a discourse phenomenon, managed to a large extent through cross-clausal references. Aspectual reference, by comparison, operates at the clause level. This pattern, while having important implications for language processing, characterizes Chinese at the syntax--semantics interface. It also facilitates temporal progression in narrative texts. To verify these claims, a preliminary examination of 953 clauses of narrative texts (54 excerpts from works of five contemporary Chinese writers) is included in the present study. The results show that the vast majority of the clauses support the claims of this study; roughly 5% remain unsettled and need further study.1 This article is organized as follows: Section 2 outlines the descriptive approach and the theoretical background of this study, including the nature and types of temporal and aspectual references. Sections 3 and 4 elaborate my position regarding temporal and aspectual references in Chinese, respectively. In Section 5, excerpts from natural discourse are brought in for illustration. A number of residual issues are also discussed. Section 6 demonstrates the role of perfectivity encoding in narrative advancement. Section 7 is the conclusion. 2. Descriptive approach and literature review 2.1. Temporal reference In this study, temporal reference refers to the indication of a situation’s location on the time axis. The reference can be made directly by a time expression, such as in 1980 in (1), or by encoding the relation of the situation time to a reference time (the default being the time of speech or writing), e.g., by tense marking in English (Reichenbach, 1947; Smith and Erbaugh, 2005). (1) Wo 1980 nian chusheng. I 1980 year (be)-born ‘I was born in 1980.’ A comparison of the Chinese and English versions in (1) shows that the English sentence encodes both situation time (in 1980) and tense marking (past tense). In Chinese, there is no overt tense marking on the verb. Situation time may be expressed by adverbials of time such as 1980 nian ‘in 1980’ in (1) or inferred by other means (details in Section 3). 2.2. Aspectual reference Aspectual reference indicates how a situation is viewed or presented (e.g., as a complete event or an on-going event). The present study concurs with the general consensus in the field that aspectual reference is a result of the interactions of categories both within the grammar and between the grammar and lexicon. Smith’s (1983, 1994, 1997[1991]) twocomponent view, including her terminology of viewpoint aspect and situation aspect, is adopted as the descriptive framework of this study.2 A brief sketch of this approach is laid out below using English examples. 2.2.1. Viewpoint aspect Viewpoint aspect, by a classic definition, indicates the ‘‘different ways of viewing the internal temporal constituency of a situation’’ (Comrie, 1976: 3), that is, whether a situation is viewed as a single whole or only one part of it (beginning, middle or the end) is focused on. Viewpoint aspect is indicated by grammatical markings on the main verb. For this reason, it is also called ‘‘grammatical aspect.’’ Commonly occurring and widely discussed aspectual distinctions include perfective and imperfective, durative and non-durative, inceptive and completive, and progressive and non-progressive aspects. They are part of the underlying meaning of natural languages and thus belong to universal semantic categories. This article examines two most basic aspectual distinctions, perfective aspect (viewing a situation in its entirety) and the imperfective aspect (focusing on part of an event typically without the endpoint). Below is an example from Comrie (1976: 4). (2)

(a) John read a book yesterday; (b) while he was reading it, (c) the postman came.

1 The unsettled cases mainly include special constructions such as serial verb constructions, yi followed by a verb (e.g., in [49b]), the various uses of zai (which is discussed throughout this paper) and some directional Resultative Verb Complements (Section 4.2.1) that may not indicate event endpoint. 2 See Chen (1988) for a similar approach with a three-component view of the Chinese temporal system (i.e., temporal location, aspectual marking and situation types).

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2047

In (2), John’s reading is presented from two aspectual viewpoints. In (2a), the event is presented in its entirety or as a single whole without distinguishing its various internal phases (the beginning, ending or the portion in between). Since the view is on the entire event, the event is presented as bounded. This is the perfective aspect, for which the concept of boundedness is crucial. By contrast, (2b) focuses on the middle portion of the event. It says that it was in the middle of John’s reading that the postman arrived. No reference is made to the beginning or, especially, the end of John’s reading. The event is thus presented as unbounded. This is the imperfective aspect. Note that the distinction here is not in the situation itself, but in how the situation is portrayed. In (2c), the postman’s arrival is also presented in the perfective viewpoint. Not all languages use grammatical markings to indicate aspectual distinctions and, when they do, they do not indicate the same aspectual distinctions. Whether a specific distinction is made, how it is conveyed, and the extent to which the device is grammaticalized vary from one language to another (Comrie, 1976; Friedrich, 1974). ‘‘Aspect languages’’ (such as most Slavic languages) use morphological markings on the verb to indicate a clause’s aspectual reading. Chinese is such a language (detail in Section 4.1). ‘‘Tense languages’’ (such as most modern Germanic languages) conflate the concept and grammatical markings of aspect with tense. In English, for example, the perfective versus imperfective opposition is not grammaticalized. There are no specific verbal forms to indicate the distinction. Instead, the progressive versus non-progressive distinction partially marks the perfective--imperfective opposition. As (2) shows, the progressive marking (be V-ing) indicates the imperfective aspect, whereas the plain, non-progressive verb form indicates the perfective aspect. This is, in Comrie’s (1976: 9) words, ‘‘verbal periphrasis.’’ Sentence (2) also shows that bounded events are closely related to past events, although future events may also be perfective and bounded. I will return to this point later. 2.2.2. Situation aspect Situation aspect, different from the grammatical marking of viewpoint aspect on the verb, concerns the inherent nature of situations. It is conveyed by the lexical content of the verbal predication, typically regarding the manner in which situations take place, e.g., whether they are dynamic or static, durative or non-durative, and telic or atelic. This line of research grew out of Vendler’s (1957) monumental study of situation types, which categorizes situations into four classes based on their temporal structures: states, activities, accomplishments, and achievements. The inherent nature of these situation types is related to and also contributes to aspectual meaning. States, for example, are atelic and, therefore, imperfective by nature. Accomplishments and achievements are telic; they are construed with an endpoint, hence perfective. Two important points have been made in previous studies of situation aspect. First, situation types reflect fundamental cognitive distinctions by which we, as humans, organize situations in the world. Such distinctions are language independent. Consequently, Vendler’s (1957) taxonomy has been widely adopted in the study of many languages, although details of the categorizations vary.3 Secondly, situation types, although closely related to the verb, are not solely determined by the verb. Other elements in the verb phrase, such as direct object and adverbial adjuncts, also play a role. For example, write and run designate open-ended, atelic activities, but write a letter and run for an hour are telic accomplishments, because the quantified direct object a letter and the adverbial for an hour designate the endpoints of the activities and, thus, render the situations as bounded. This means that the entire predication, rather than just the verb, should be taken into account in the analysis of situation aspect. 2.2.3. The interaction of situation aspect and viewpoint aspect Situation aspect and viewpoint aspect converge at the clause level to bring together grammatical and semantic information for the interpretation of aspect. In Smith’s (1997) words, ‘‘what information a viewpoint presents is affected, and limited, by the structure of the situation talked about.’’ The interaction is in the following manner: The temporal schema of a situation is made semantically available by situation aspect, including the inherent features of the verb and the features conveyed by other constituents. For example, walk to school is a dynamic and telic situation with an intended temporal and spatial endpoint (to school). The viewpoint aspect then puts the entire situation into perspective, viewing it either perfectively (walked to school, bounded) or imperfectively (be walking to school, unbounded), depending on the need of the discourse. By this description, telicity is treated in this paper as a feature of lexical semantics (as in Vendler’s classification of verbs). Boundedness is a feature of the perfective aspect that combines the information of viewpoint aspect and situation aspect and serves as a link between syntax and lexical semantics (Cappelle and Declerck, 2005; Dahl, 1981, 1985; Declerck, 1979, 1989, 2006; Depraetere, 1995; Jackrndoff, 1990; Sasse, 2002; Vanden Wyngaerd, 2001). I have so far demonstrated that perfectivity may be conveyed through perfective marking on the verb (viewpoint aspect) or by designating the end boundary of events explicitly (situation aspect). Generally speaking, the mechanisms employed in various languages to mark viewpoint aspect and their interplay with situation aspect lead to the complexity of

3

For Chinese, Tai (1984) proposes three categories of states, activities, and results; Smith (1997[1991]) adds a fifth category of semelfactives.

2048

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

aspectual distinctions across languages. After this brief introduction to the theoretical context, the next two sections elucidate temporal and aspectual references in Chinese. 3. Indication of temporal location in Chinese This section describes the various ways by which the information of temporal location is conveyed in Chinese. Excerpts from natural discourse are used for illustration. They are from English--Chinese bilingual news reports from Voice of America (USA), China Daily bilingual news and Hengxing English (mainland China), Mingbao bilingual editorial (Hong Kong) and some excerpts from narrative texts (fiction with published English translation). The English version is used to verify the interpretation of the Chinese text. A drawback is that the two versions do not always correspond to each other word-for-word. This, however, should not affect the analysis. The examples are kept short for space considerations; the expressions under discussion are underlined. In the following I show that, with no overt tense marking on the verb, temporal location in Chinese may be conveyed by various components at the clause level and/or by mechanisms at the discourse level. At the clause level, expressions carrying (or implying) temporal information occur early before the main verb. At the discourse level, cross-clausal reference also guarantees early determination of temporal location during the processing of a clause. In general, temporal interpretation in Chinese is very much a discourse phenomenon. 3.1. At the clause level At the clause level, information of temporal location is provided most directly by adverbials of time (including subordinate clauses). Other constituents, such as adverbs and auxiliary verbs, do not denote temporal reference per se, but may carry temporal information or have implications for temporal interpretation. All these expressions occur between the subject and the main verb. They never occur after the verb. In addition, some verbs may indicate future irrealis situations and, thus, imply future time. 3.1.1. By adverbials of time The most direct way to convey temporal information is by adverbials of time. They may take the forms of noun phrases, prepositional phrases, or adverbial clauses. An example of a noun phrase is already given in (1). Shown below is a PP in (3) and an adverbial clause in (4).4 The default position of these temporal expressions is between the subject and the main verb; they never occur after the main verb. (3)

HX090425 gai yanjiu yu shang zhousan gongbu. this study on last Wednesday publish ‘(. . .) published the study on Wednesday.’ VOA120507

(4) Pu Jing yu 2008 nian cong zongtong weizhishang xieren hou yizhi Pu Jing YU 2008 year from president position.on step-down after all.the.time danren Eluosi zongli. serve.as Russian prime.minister ‘Mr. Putin had been Russia’s prime minister since relinquishing the presidency in 2008.’

3.1.2. By adverbs Some adverbs may have implication for the temporal location of situations. For example, yi(jing) ‘already’ in (5) and ceng ‘used to (verb)’ in (6) typically locate events in the past; yihou ‘later,’ jiang ‘will’ (in [7]) and yao ‘will (soon)’ locate events in the future. (5)

MB090518 . . . Zhengfu yi anshi bu hui xu yue government already hint NEG will renew contract ‘. . . the government has hinted that it will not be renewed.’

4

In the English gloss, grammatical words are presented in small cap,

M

= measure word;

MOD

= modifier.

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

(6)

2049

HX081207 Dami ceng shi Ribenren canzhuoshang zui zhongyaode shipin. rice once be Japan.people on.table most important food ‘Rice was once the most important ingredient in every Japanese meal, . . .’ VOA090403

(7) Ta jiang henkuai anpai yige yicheng, he will soon arrange one-M agenda, ‘He soon will lay out an agenda . . .’

3.1.3. By auxiliary verbs Auxiliary verbs are not full-fledged verbs, but have some verbal properties. In Chinese they include neng ‘can/be able to,’ (ying)gai ‘ought to, should,’ and hui ‘will’ (Li and Thompson, 1981). By expressing capability, intention, or expectations, their contribution to temporal interpretation is to indicate that the states of affairs they describe are projected rather than real-time occurrences. (8)

HX081209 Ni zhouwei de ren yue kuaile, ni ye hui yue kuaile. you around MOD people more happy, you also will more happy. ‘the more happy people you know, the more likely you are yourself to be happy.’ HX081207

(9) Zhiyao renmen mei dun duo chi yikou mifan, zhege mubiao jiu neng shixian. only.if people every meal more eat one.M rice, this.M goal then can realize. ‘This target can be achieved if Japanese people eat an additional bite of rice at every meal.’

3.1.4. By verbs Some verbs, such as zhunbei ‘prepare’ (10), yuqi ‘expect’ (11), xiwang ‘hope,’ and xuyao ‘need,’ may take a complement clause or verb phrase to designate intention or projected situations. The result is a serial verb construction (Li and Thompson, 1981). The interpretation of the complement clause or verb phrase is highly correlated with future time. These verbs, as Smith and Erbaugh (2005) point out, are modal in nature. (10)

HX110311 Meiguo zhunbei suishi bangzhu riben renmin. US prepare any.time help Japan people ‘. . . the United States is ready to help the people of Japan.’ VOA091001

(11) Aobama yuji xingqiwu zaoshang dida Danmai shoudu . . . Obama plan Friday morning arrive Denmark capital ‘Mr. Obama is scheduled to arrive in the Danish capital early Friday.’

3.2. At the discourse level Very often, temporal information is not provided at the clause level, but can be inferred from context through discourse anaphora or the discourse mode of the text. This requires operations at the discourse level. 3.2.1. By discourse anaphora In natural discourse, e.g., conversation or narrative texts, consecutive clauses describing a series of incidents often share the same temporal reference. Once the reference is established, inference can be made in the subsequent clauses without the need to repeat the temporal expression in every clause. Such a co-referential relation, referred to as temporal anaphora, is studied by a line of research drawing an analogy between temporal anaphora and nominal anaphora (including the null NP) (Bittner, 2008; Hinrichs, 1986; Partee, 1973, 1984). Temporal anaphora in Chinese has two basic forms. The first one is illustrated by (12), in which a time reference (future) is established in the first clause in (a) by the adverb jiang ‘will.’ The reference is then carried over to the following clauses in (b) and (c), in which the default positions for temporal expressions are left empty (indicated by Ø). Thus the temporal reference in (b) and (c) is implicit but clear. Such inference could be carried even further until a change of time is indicated. This is an important device for temporal interpretation in Chinese.

2050

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

(12) (a) Tamen jiang kaiban jinglaoyuan, (b) Ø dui 65 sui yishang de laoren (a) they will build nursing.home, (b) Ø for 65- above MOD elderly shixing tuixiu zhidu, (c) Ø lao you suoyang. . . CD110816 implement retirement system, (c) Ø elderly have care . . . ‘They will build nursing homes and implement a retirement system for those over 65 years old, so that they will be taken care of when they get old.’ The pattern in (12) shows that, as soon as a Ø is identified, a co-referential relation can be established between the Ø and its antecedent, in the same fashion as a nominal zero anaphora. Since the Ø is always before the main verb, the coreferential relation is always established early during the comprehension of the clause. The second pattern of temporal anaphora is illustrated by (13), which is constructed to contrast with (12). The overt temporal expression jinnianlai ‘in recent years’ appears in (a). Its default position is that of the Ø, but in this case, it is dislocated to the beginning of the passage. By so doing it creates a scope for temporal interpretation. All the clauses under the scope (here, [a]--[c]) are interpreted as sharing the same temporal reference. (13) (a) Jinnianlai, tamen Ø kaiban -le jinglaoyuan, (b) Ø dui 65 sui yishang (a) recent.years, they Ø build-LE nursing.home, (b) Ø for 65-above de laoren shixing tuixiu zhidu, (c) Ø lao you suoyang. MOD elderly implement retirement system, (c) Ø elderly have care ‘In recent years, they built nursing homes and implemented a retirement system for those over 65 years old, so that they are taken care of when they get old.’ A comparison of (12) and (13) shows that, although the clauses in (12b and c) and (13b and c) are identical, they are interpreted differently according to the different time reference they inherit from their respective discourse context. 3.2.2. By discourse modes Temporal location may also be inferred by the particular discourse mode the text is in. Smith (2003) categorizes passages of written texts into five major modes based on their linguistically based features: narrative, report, description, information, argument. Each mode introduces certain types of entities into the discourse and follows certain temporal principles. The narrative mode, for example, typically recounts events framed in the past. Thus the default temporal reference is the past (also see Labov, 1972; Reinhart, 1984). The excerpt in (14), for instance, is the beginning of a Chinese novel. No explicit time expression is used, but the interpretation is the past, as reflected in its English translation. (14)

(Wang, 1981: 1) (a) Beijingpai yueye qiche zai xiangcunde-gonglushang feichi, (a) Beijing.brand cross.country vehicle ZAI country.road.on dash, yidianyihuang, yaolaibaiqu, (b) chepengli youmen youre, rocking, rolling, (b) in.vehicle smothering hot, (c) zhen rang ren hunhunyushui. (c) really make one drowsy ‘A jeep made in Beijing was hurtling along a country road. Inside, it was so stuffy that the passenger felt drowsy.’

By contrast, passages in the general description and information mode are atemporal. By default the interpretation assumes the present time (Comrie, 1976). Thus, in (15), despite the lack of explicit mention of time (same as [14] above), the interpretation is the present. This is also reflected in its English translation. (15)

CD090924

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2051

(a) Jihu meiyou boke shi tongguo dingyue yinlide; (b) qi jingji (a) almost no blog be by subscription earn.profit (b) its economic fazhan moshi shi yikao zaixian guanggao. (c) ruguo duzhe dianji wangzhanshangde online ad (c) if reader click website’s develop model be by guanggao, bokezuozhe jiu neng zhuan qian. (d) yixie wangzhan shenzhi hai ad blogger then can earn money. (d) some website even also yi dianjiliang zuowei xiang zuozhe fufei de biaozhun. by click.count as to blogger pay.fee MOD rate ‘Almost no blogging is by subscription; rather, it owes it economic model to on-line advertising. Bloggers make money if their consumers click the ads on their sites. Some sites even pay writers by the click, . . .’ In summary, information of temporal location may be provided or implied in Chinese by various constituents in the clause. It may also be conveyed by mechanisms at the discourse level that create contexts for temporal interpretation. The length or size of such contexts may vary. In general, temporal information is made available early on in the interpretation of a discourse unit. Overt temporal expressions override the implicit temporal information. Given that verbs in Chinese are not marked to show tense, these are important devices that guarantee efficiency in the system and keep the temporal reference of designated situations covert but, nevertheless, clear. 4. Aspectual reference in Chinese To start this discussion with a general picture and to see the Chinese system in a typological perspective, let’s refer back to a distinction made earlier in Section 2.2.1. That is, languages differ in how temporal and aspectual information are conveyed through grammatical markings. ‘‘Tense languages’’ (such as English) do not mark aspect specifically but conflate the marking of aspect with that of tense. ‘‘Aspect languages’’ (such as Slavic languages) use grammatical morphemes, separate from those of tense, to mark aspect explicitly. Chinese, on the one hand, is distinct from tense languages in that it has no tense marking. On the other hand Chinese, as an aspect language, is also distinct from other aspect languages in that, although a set of aspectual markers are used, the use of these markers is optional rather than obligatory. Aspect indications do not have to rely on these markers. They can be accomplished by an array of other means. This section elucidates the various means of aspectual reference in Chinese, starting with viewpoint markers and then moving on to situation aspect. The discussion aims to show two distinct features: (a) its post-verbal position and (b) its management at the clause level. Constructed examples are used for the initial illustration. 4.1. Viewpoint aspect markers In Chinese, aspectual markers play a specialized role in providing aspectual information. For this reason, they have been the focus of most studies of Chinese aspect (Chao, 1968; Li and Thompson, 1981; Ross, 1995; Smith 1997[1991]; Wang, 1965; Xiao and McEnery, 2004, among others). Since a vast literature on these markers already exists, I focus this discussion on the ones relevant for the present study: the perfective aspect marker -le and two markers of the imperfective aspect, the progressive zai and the durative --zhe.5 -Le, as a perfective aspect marker, ‘‘signals the actualization of a situation with respect to a past, present or future reference time and presents the situation as a single whole’’ (Xiao and McEnery, 2004: 91).6 In (16), for example, the --le indicates that the event is already materialized or completed by the reference time (the default being the time of the speech or writing) and that the event is being looked at as a single bounded whole with its effect applying to the present. Without the --le, the sentence is ambiguous, because the temporal and aspectual information of the event are unclear. (16) Wo xue-le hongwen. I study-LE Chinese ‘I studied Chinese.’ (I know Chinese now.)

5 In addition to -le, Smith (1997[1991]) and Xiao and McEnery (2004) also treat verb duplication (Section 4.2.2) and RVCs (Section 4.2.1) as viewpoint aspect markers. Some researchers also discuss --guo as a perfective marker. Regarding --guo, I adopt Smith’s (1997: 266--267) view that the situation it makes is more of a state rather than a dynamic event. Therefore, -guo is not discussed in this paper. 6 Some researchers, e.g., Ross (1995), maintain that --le also indicates past time reference. My view is that Xiao and McEnery’s (2004) analysis (and their term the ‘‘actual’’ --le) offers a more flexible account. For the close relation between tense and aspect, see Bull (1971) and Hopper (1982).

2052

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

This analysis of --le offers flexibility in interpretation according to different reference time. While perfective events may materialize in the future with a future reference time or context, -le by default typically locates an event in the past (Smith and Erbaugh, 2005). The two imperfective markers, the progressive zai and the durative -zhe, are illustrated in (17) and (18). Generally speaking, zai is used before dynamic activity verbs (such as da (qiu) ‘play (ball)’), while -zhe is used after verbs of posture and placement (such as zhan ‘stand’ and dai ‘wear’) to denote a posture or a state as a result of a dynamic activity. (17) Tamen zai da qiu. they ZAI play ball. ‘They are playing ball.’ (18) Ta dai-zhe yanjing. he wear-ZHE glasses ‘He is wearing glasses.’ In complex sentences, --zhe may also be used in one of the two (or a series of) clauses to signal that the event it marks provides a durative background for the other verb event(s) (Li and Thompson, 1981). In (19), there are two action verbs, he ‘drink’ and liaotian ‘chat.’ The --zhe puts the former in a secondary status. Syntactically, the verb becomes subordinate in nature, as shown in the English translation. (19) Women he-zhe cha liaotian. drink-ZHE tea chat we ‘We chatted while drinking tea.’ Aspectual markers in Chinese are mostly verb suffixes. The only exception is the progressive zai, which appears before the verb and is unhyphenated in transcription. This will be discussed in more depth in Section 5. 4.2. Situation aspect and a broader view of perfectivity Situation aspect, as stated earlier, is the lexical means of presenting a situation. It involves the verb and other constituents such as the verb’s complements and adverbial adjuncts. Earlier in Section 2.2.3 it was shown that, to present an event in the perfective aspect, the event has to be bounded, i.e., to be presented with an endpoint. In Chinese this may be done in two ways. One is to mark the verb grammatically by the perfective marker --le in the component of viewpoint aspect. The other is to explicitly designate the event’s endpoint, result or the extent to which the event occurred -- in the component of situation aspect. The expressions involved in situation aspect are not grammatical aspect markers, but they convey aspectual meaning. This part of the Chinese aspectual system has just begun to be investigated. With Smith’s (1997[1991]) two-component theory of aspect as the theoretical foundation, Xiao and McEnery (2004) present a refined characterization of the interaction between the main verb and other constituents to convey situation aspect. This section characterizes two important features of Chinese situation aspect: (a) the postverbal position of the expressions involved and (b) its operation at the clause level. Again, I focus on perfectivity and show that expressions contributing to perfectivity share an important function, i.e., they impose a bounding effect on the event they describe. For this reason, they are referred to collectively as ‘‘bounding expressions.’’ Note that bounding expressions are not grammatical markers but, nevertheless, they render the situations bounded.7 The following discussion divides bounding expressions into three categories: resultative verb complements (RVCs) (Section 4.2.1), event quantifications (Section 4.2.2) and prepositional phrases (PPs) (Section 4.2.3). Constructed examples are used for illustration, before excerpts from natural discourse are brought in for verification in Section 5. 4.2.1. Resultative verb complements (RVCs) RVCs are part of a two-element verbal structure. The first element is an activity verb; the second, an RVC, may be a verb, an adjective denoting the result of the action, or a word indicating the direction of a motion. RVCs are abundant and

7

See Section 5 for a further discussion of the differences between the grammatical marker --le and bounding expressions.

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2053

Table 1 Categories of RVCs. Chinese

Gloss

English

(a) Resultative RVCs

kan-qingchu ning-jin ting-wen

see-clear twist-tight stop-stable

see clearly twist tight stop

(b) Achievement RVCs

xue-hui kan-dao da-po ting-dong

study-know look-attain hit-break listen-understand

learn see break understand

(c) Completion RVCs

xiu-hao zuo-wan huan-diao

fix-good do-finish return-attain

fix finish doing return

(d) Directional RVCs

na-zou chuan-shang ai-shang na-qi(-lai)

take-away put-on love-on pick-up(-come)

take away put on fall in love pick up

used frequently in Chinese. The two elements and their relation show rich syntactic and semantic features. For this reason, RVCs constitute one of the particularities of the Chinese language and have been widely discussed (Chao, 1968; Li and Thompson, 1981; Lu, 1977; Packard and Jerome, 1998; Packard, 2000; Ross, 1990; Thompson, 1973). Below is a common categorization, based on what is conveyed by the RVCs (underlined) (Table 1). The contribution of RVCs to perfectivity, recognized by Mei as early as in 1981, is now well agreed upon, although views vary with regards to many other aspects of RVCs.8 One issue for further research is in which part of the aspectual system RVCs play a role. Smith’s (1997: 265) proposal is that they ‘‘convey viewpoint and/or situation aspect, depending on individual forms and syntactic context’’ (also see Xiao and McEnery, 2004). In this paper, RVCs are discussed together with other ‘‘bounding expressions’’ because they share some important features: they all have lexical content and they designate event endpoint explicitly. They are not as highly grammaticalized as aspectual markers. RVCs also differ from grammatical markers in that they are large in number and form an open class (Packard, 2000; Ross, 1990), while aspectual markers are a small and closed set. Comparing with the highly grammaticalized markers such as --le and -zhe, one may reasonably say that the meaning and use of RVCs have not quite reached the status of grammatical markers yet. Their current status is probably somewhere between content words and grammatical markers.9 4.2.2. Event quantification Activities may be bounded by quantifying expressions to specify the extent to which the activities occur ([20] and [21]), the number of times they happen ([22]), or the distance a motion covers ([23]).10 (20) Wo zou zou. I walk walk ‘I (will) take a walk.’ (21) Wo kan ta yi yan. I look he one glance ‘I (will) take a look at him.’

8

A small number of RVCs, mostly directional RVCs, may not designate event endpoint. Their role in aspectual marking awaits further study. In fact, fuzzy boundaries among grammatical categories are a common phenomenon. See Chen (2005) for an analysis that divides viewpoint aspect markers into core and peripheral categories. 10 Quantified object NPs may also contribute to event bounding (Li and Thompson, 1981). However, because they are different in nature and may work differently, they will be studied in another project. 9

2054

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

(22) Wo zixi ting san bian. I carefully listen three times ‘I (will) listen to it carefully three times.’ (23) Naliang che kai shi li. That-CL vehicle run ten miles ‘That car (will) go ten miles.’ The bounding effect of most of these quantifying expressions is self-explanatory, except verbal duplication in (20), which limits the designated activity to a short duration or a light degree. This is reflected in the English translation ‘‘take a walk.’’ Without the duplication, wo zou means ‘‘I (will) leave’’ or ‘‘I (will) walk (to a place).’’ Chao (1968: 205) analyzes this type of verb duplication as ‘‘tentative aspect.’’ Smith (1997: 271) considers such verb duplications as ‘‘semantically perfective because they present closed situations.’’ 4.2.3. PPs and the ‘‘rule of positional meaning’’ PPs may appear either before or after the verb in Chinese depending on their semantic roles. This has been described in earlier studies, but so far no explanation has been given as to the motivation behind. Here I provide a unified account of the word order variations. Some background first. 4.2.3.1. The ‘‘rule of positional meaning’’. A great deal of research in Chinese linguistics, synchronic and diachronic, has been devoted to word order. This is hardly surprising given the analytical nature of the language. Word order has been found to play a role in making semantic distinctions. For example, NPs in preverbal and postverbal positions differ with regard to definiteness. Preverbal NPs are usually definite, while postverbal NPs tend to be indefinite (Mullie, 1932; Chao, 1968; Tai, 1975; Teng, 1975; Li and Thompson, 1975). Temporal and locative expressions also differ in meaning depending on their positions. General adverbials (e.g., time and location) appear before the verb, whereas temporal duration or the location of participants as a result of the verb event is expressed after the verb. In (24), sandian ‘three o’clock’ is the time when the event of shuijiao ‘sleep’ takes place, hence the preverbal position. In (25), the time expression shi fenzhong ‘‘ten minutes’’ denotes temporal duration. Therefore, it is after the verb. (24) Wo sandian shuijiao. I three.o’clock sleep. ‘I go to sleep at three.’ (25) Wo shui shi fenzhong. I sleep ten minute ‘I sleep for ten minutes.’ Light (1979: 167), based on these observations and similar word order variations with locative expressions (to be discussed below), proposes the ‘‘rule of positional meaning’’: ‘‘The meaning of nouns and adverbs depends on their location before or after the main verb.’’ It is now commonly recognized that word order, especially the verb vis-à-vis the preverbal and postverbal positions in the clause, is an important device to encode semantic distinctions in Chinese (Huang, 1981; Li, 1990; Tai, 1983). In diachronic studies, there was a discussion from mid 1970s to 1990s about the historical change of postverbal PPs to the preverbal position and whether that change can serve as evidence for a more fundamental word order change of Chinese from SVO to SOV. Although views vary on many accounts, there has been a consensus on the historical change of some postverbal PPs to the preverbal position. Relevant to the present study are the PPs headed by yu. Yu-PPs were prolific in classical Chinese with a large number of functions. They appeared mostly in the postverbal position. During the course of the change, their various functions were differentiated, moving some to the preverbal position and leaving the others after the verb. During this process, yu gradually withdrew from many of its functions; in its place now are a number of more specific prepositions in Mandarin. After the change, the majority of the prepositions replacing yu are now preverbal; only a small group remains after the verb (He, 1985; Sun, 1991, 1996). The three examples below (all from Mengzi, written about 300 BCE, adopted from He, 1985) illustrate three functions of yu-PPs in classical Chinese relevant

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2055

to the present study: goal or destination in (26), recipient in (27) and location in (28). These yu-PPs would later be replaced by several prepositions in modern Chinese whose use follows the ‘‘rule of positional meaning’’.11 (26) yi qi min yu he dong. move its people YU river east ‘(I will) move the people to the east of the river.’

(YU: destination)

(27) daifu neng jian ren yu zhuhou officer can recommend people YU prince ‘A great officer can recommend a man to his prince.’

(YU: recipient)

(28) wang zuo yu tang shang prince sit YU hall up ‘The prince was sitting aloft in the hall.’

(YU: location)

These functions of yu-PPs in classical Chinese are taken over by several other prepositions in modern Chinese. They are dao-PPs for goal or destination, gei-PPs for recipients and zai-PPs for location. These PPs may appear either before or after the verb, but obvious constraints have developed on postverbal PPs to exert a bounding effect on the events and contribute to perfectivity. Before diving into the details, I should mention that prepositions in modern Chinese are a complex phenomenon. They used to be verbs and are now in a grammaticalisation process to becoming prepositions. Many of them can now still be used as verbs (Li and Thompson, 1974, 1981; Li, 1980). For this reason, they are also known as ‘‘coverbs.’’ A typical example of their multiple functions is the different uses of zai, which are discussed in several sections of this paper. The discussion below shows the most basic distinction between preverbal and post-verbal positions of the three PPs. 4.2.3.2. Dao-PPs. Dao as a verb means ‘to arrive.’ As a preposition, it means ‘to/till (a destination or time).’ In the preverbal position, a dao-PP specifies the physical or temporal location of the verbal action. The one in (29), for example, indicates the location of the chatting event (unbounded).12 (29) Tamen dao jiuba liaotian. they (go) to bar chat ‘They chat in a bar.’ By contrast, a postverbal dao-PP designates the endpoint of the verbal event. In (30) the dao-PP indicates that the event of sleeping terminates at five o’clock. Note that the postverbal position contrasts with the preverbal position of cong sandian ‘from three o’clock,’ which specifies the starting point of the activity. In (31), the dao-PP denotes the destination of the flight. In both (30) and (31), the situations are bounded by the postverbal dao-PPs. (30) Wo cong sandian shui dao wudian. I from three.o’clock sleep to five.o’clock ‘I sleep from three to five.’ (31) Tamen fei dao Niuyue. they fly to New York They fly to New York.

11 This is one of the results of the historical development regarding preverbal and postverbal PPs. It should not be seen as the only motivation or cause. Part of the complexity is discussed in Section 5. 12 Alternatively, this dao may be analyzed as a verb in a serial verb construction. The structure of serial verb constructions is beyond the scope of this paper.

2056

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

4.2.3.3. Gei-PPs. The use of gei-PPs shows a similar pattern. A pre-verbal gei-PP encodes a beneficiary, often translated into English as ‘‘for + NP.’’ In (32), for example, the preverbal gei-PP indicates beneficiary; it does not impose an end-boundary to the event. (32) Wo gei ni jiao niupai. I for you order steak ‘I’ll order a steak for you.’ When a gei-PP appears after a verb, the verb usually designates the transfer of an object. In such cases, the gei-PP plays a bounding role by indicating the recipient of the transferred object, as in (33). Such a gei-PP renders the event (and the clause) as bounded. (33) Wo ji-le yiben shu gei ni. I send-LE one-M book to you ‘I sent you a book.’

4.2.3.4. Zai-PPs. Locative zai-PPs also show a semantic distinction before and after the verb. Preverbal locatives denote the general location of actions and states of affairs; the situations are presented as unbounded (e.g., [34]). Postverbal locatives denote the location of participant(s) as a result of the event. Thus postverbal zai-PPs are restricted to the verbs of (dis)placement, posture and (dis)appearance (e.g., ting ‘park’ in [35]). By describing the result, the events are presented as bounded. As Tai (1975) observes, ‘‘this functional difference is the governing principle for the word order of Chinese place adverbials.’’ (34) Ta zai canguan gongzuo. he ZAI restaurant work ‘He works in a restaurant.’ (35) Wode che ting zai waibian. my car park ZAI outside My car is parked outside. Sometimes the same zai-PP may occur either before or after the same verb. Li and Thompson (1981) use the examples in (36) (with their translation) to show that in such cases the preverbal and postverbal locatives make no difference in meaning. Here I disagree. I argue that the different word order does entail semantic differences. (36)

a. Ta zai chuangshang shui. he ZAI in.bed sleep/lie ‘S/He sleeps on the bed.’ b. Ta shui zai chuangshang. he sleep/lie ZAI in.bed ‘S/He sleeps on the bed.’

Generally speaking, preverbal zai locatives tend to designate the location of a situation. For cases such as (34) and (36a), i.e., without context or any other indication, the default interpretation is that the situation is habitual rather than a specific event. Or, it may be an intended event. Thus the interpretation of (36a) is that the subject ta ‘he’ either habitually sleeps in bed, or intends to sleep in bed. In either case, ‘‘sleeping’’ is presented as unbounded. Li and Thompson’s English translation reflects such interpretation. By contrast, the postverbal zai-PP in (36b) indicates that the subject ta ‘he’ is in bed as a result of the specific activity. Thus, the sentence could be translated as ‘‘S/He has lain down on the bed’’ or ‘‘S/He is in bed (sleeping).’’ By designating the result, the event is presented as bounded.

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2057

Table 2 The order of the verb, -le and bounding expressions in the clause. Verb

Endpoint (A)

(-le) (B)

RVCs

(a) zuo (b) kan (c) xue (d) da

wan dao hui po

(-le) (-le) (-le) (-le)

Event quantification

(e) kan (f) qiao (g) pao (h) fa

Prepositions (or coverbs)

(i) song (j) fei (k) zuo

(-le) (-le) (-le) (-le) gei dao zai

(-le) (-le) (-le)

Quantification (C)

kan yixia yihuir sanci

Object

Gloss

(fan) (ta) (Fawen) (beizi)

cook-finish (meal) look-attain (him/her) study-know (French) hit-break (cup)

(shu) (men) (bu) (kuan)

read a bit (book) knock once (on the door) jog a while (steps) punish three times (money)

(ta) (Beijing) (dishang)

give to (him/her) fly to (Beijing) sit on (the floor)

Two more examples are given in (37) and (38) to further explicate the difference. Because the default position for general adverbials is before the verb, locatives in that position by default designate the location of habitual and intended events (unbounded). This accounts for the low acceptability of (37), as accidents such as falling on the stairs are usually not habitual or intended.13 When the zai locative occurs after the verb as in (38), the sentence becomes felicitous because in that position the zai-PP designates the location of ta ‘he’ as a result of the falling, which is congruent with the interpretation of an accident. (37) ?

Ta zai loutishang shuaidao. he ZAI on.stairs fall

(38) Ta shuaidao zai loutishang. he fall ZAI on.stairs He fell on the stairs. The bounding function of the postverbal zai locative has been noted by a number of researchers. Li and Thompson (1981: 205--206) consider it as one type of ‘‘perfectivizing expressions.’’ Chen (1978) claims that a postverbal zai-PP presents an event in the ‘‘terminal aspect’’ when no overt grammatical aspectual marker is used. 4.3. The syntactic slots for bounding expressions The bounding expressions discussed above and the perfective marker -le are put together in Table 2 to show their relative positions vis-à-vis the main verb. As can be seen, bounding expressions and --le all occur after the main verb in the three slots labeled (A), (B), and (C). Slot (B) is for --le, which may or may not appear in a clause, hence in parentheses. The other two slots contain bounding expressions. The object NPs are included as reference. In the last column is the English gloss with the role of the bounding expressions underlined. Since bounding expressions occupy different slots after the verb, they are related to the verb in different degrees of tightness. In Table 2, the first group in rows (a)--(d) has RVCs in slot (A). RVCs have a tight relation with the verb. They form verb compounds and the verb suffix --le occurs after them.14 The expressions of event quantification in slot (C) are positioned after --le. These expressions, with a relatively loose relation with the verb, are adverbial adjuncts or peripherals in the verb phrase. It can be seen from Table 2 that PPs as bounding expressions also have close affinity with the preceding verbs, such that the verb suffix -le may occur after the prepositions. It seems that, in structure, these prepositions are treated the same as RVCs. But they differ from RVCs in important ways. RVCs (rows a--d) and the preceding verbs form compounds, which

13 Additional expressions such zong ‘always’ may be added in front of the locative in (37) to increase the acceptability and to express the meaning that the person does habitually fall on the stairs. 14 Some RVCs, e.g., directional RVCs, may appear after -le.

2058

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

then take the following NPs as objects. With prepositions (rows i--k), however, the following NPs are the object of the prepositions and are limited by their semantic roles, e.g., to designate the recipient or goal of the events or the resultative location of the participants. When the main verb is followed by an object NP, as in (35), the postverbal PPs have another option of occurring after the object NP. Because of space limitations, I will not elaborate further on the differences between RVCs and postverbal prepositions. Suffice it to say that they both function as bounding expressions and contribute to perfectivity. Before ending this section, I would like to point out that --le and bounding expressions, although both convey perfectivity, work differently. Bounding expressions do not mark perfectivity equally either. How they interact with each other and with other features of the clause is a topic that calls for further refinement.15 5. Discussion In Sections 3 and 4, temporal and aspectual references were discussed separately using constructed examples. In this section, excerpts from natural discourse are brought in to illustrate how they work together in real communication. Major studies in this area are reviewed briefly to show how the present paper contributes to this line of research. The last issue to be dealt with in this section is the imperfective marker zai, which posts a counter-example to the word order claim of the present study. First, examine (39) extracted from a news report. It describes an air accident. Various types of temporal and aspectual information are packaged into one sentence. General adverbials specifying the time and location of the accident (i.e., xingqiyi zaochen ‘Monday morning’ and zai Baxi xibeibu waihai shangkong zaoyu fengbao hou ‘after encountering a storm in the air off Brazil’s northeastern coast’) are put between the subject/topic NP 447-hao hangban ‘flight No. 447’ and the verb zhui ‘crash.’ Note that these adverbials may be extracted from that position and put at the beginning of the sentence. When this happens, the pre-posed expression establishes a scope at the discourse level, within which not only the current clause but also a stretch of discourse may be interpreted (also see [13]). By contrast, the bounding expression, ru haiyang ‘into the ocean,’ appears after the verb. As aspectual information, this expression cannot be extracted from its post-verbal position. (39)

VOA090608 447-hao hangban xingqiyi zaochen zai Baxi xibeibu waihai shangkong NO. 447 flight Monday morning ZAI Brazil north.west ocean sky zaoyu fengbao hou zhui-ru haiyang. encounter storm after crash-into ocean. ‘The plane crashed into the sea early Monday off Brazil’s northeastern coast after flying into heavy storms.’

In (40) below is an excerpt with multiple clauses; no time expression can be found. To account for temporal interpretation in such cases, Smith and Erbaugh (2005: 713--717) have recourse to the aspectual information in the clause. They explicate that, since Chinese does not have overt tense marking, aspectual information plays a key role in temporal interpretation. ‘‘When no direct temporal information appears, aspectual information, such as whether a situation is ongoing or closed, gives pragmatic cues to the temporal location of the situation expressed.’’ One of their default deictic patterns of temporal interpretation states that ‘‘bounded events are in the past time: located before the speech time.’’ This account, although logically viable and which may ultimately arrive at the correct interpretation, cannot be (and probably is not meant to be) a description of how temporal interpretation actually works. Since aspectual information is always provided after the main verb, as shown by the present study, Smith and Erbaugh’s (2005) account would entail that, in every clause, the temporal location of the event remains unclear until aspectual information is provided. (40)

15

(Wang 1999: 59) (a) Wenguang na-zhe yizhi wan yishuang kuai (b) zou dao shuichizi genqian, (a) Wenguang hold-ZHE one.M bowl a.pair.of chopsticks (b) walk to sink front, (c) ning-kai shuolongtou (d) chong-le yixia, (e) shou-jin wangui. (c) turn-on faucet (d) rinse-LE a.bit, (e) put-into cupboard. ‘Wenguang walked over to the sink with a bowl and a pair of chopsticks in his hand. He turned on the faucet, rinsed the utensils and put them into the cupboard.’

Thanks to an anonymous reviewer for this point.

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2059

The account offered by the present study takes into consideration mechanisms at the discourse level. Since overt time expressions always appear early in a clause or in a discourse unit, early determination of temporal location is guaranteed: When no overt expression occurs in a clause, the interpretation is based by context, either through cross reference with an explicit expression in the preceding text (Section 3.2.1) or by the discourse mode (Section 3.2.2). This is an economical and efficient system, in which temporal interpretation operates in a global fashion without the need to wait for aspectual information to appear in every clause. For aspectual reference, (40) illustrates the non-obligatory use of the aspect markers. Among the five activities described in the excerpt, only two are marked by aspect markers, a --zhe in (a) and a --le in (c). The question is how the aspectual value of the ‘‘unmarked’’ clauses is determined. This is an important issue for the study of Chinese aspect, which has generated a great deal of research. Since it is also an area the present paper makes a contribution to, it is worthwhile to briefly review the line of research that leads to the present study. Up to recently, the study of Chinese aspect has focused exclusively on aspectual markers. The non-obligatory use of aspect markers has been ‘‘a very disturbing fact’’ (Chu, 1998). Although Chinese has a relatively rich set of aspect markers, the majority of clauses in actual language use, regardless of genre, are not marked by the aspectual morphemes. This has been characterized statistically. Yang (2002), in a study of 1067 clauses in narrative texts, reports that only 21.55% of the clauses use aspect markers. A similar result is also reported by Christensen (1994). The non-obligatory use of aspectual markers is discussed in most major works on Chinese aspect (Chao, 1968; Li and Thompson, 1981; Klein et al., 2000; Smith, 1997; Xiao and McEnery, 2004). Smith (1997: 279) observes that ‘‘the most important pragmatic fact about the viewpoint morphemes of Mandarin is their optionality.’’ Chu (1998) comments that ‘‘the perfective --le has been a source of great interest for linguists, but it also has been one of tremendous puzzlement and confusion for language teachers and students. It behaves sometimes like the past tense and other times like the present perfect in English; but it is not quite the same as either of them. Especially intricate and puzzling is the non-use of --le where the theory predicts otherwise.’’ A breakthrough in the analysis is made by Smith (1997[1991]). Her two-component approach to aspect takes into account not only grammatical marking but also situation aspect. For perfectivity in Chinese, Smith (1997) recognizes the role of RVCs and verb duplication (rows [a]--[e] in Table 2) in perfectivity encoding, in addition to the grammatical marker --le. This approach works in a promising direction, although Smith’s proposal solves only part of the problem. Take (40) for example, in which three clauses in (b), (c), and (e) are ‘‘unmarked.’’ Smith’s (1997) account is able to solve the problem in (c) and (e), where the verbs are followed by RVCs (ning-kai ‘turn-on’ and shou-jin ‘put-into’, respectively). Thus, the two events are bounded and perfective. But the problem in (40b) remains, because the bounding effect of PPs (dao shuichizi genqian ‘to the sink’ in [40b]) is not recognized in Smith’s (1997) analysis. The present study moves further along this line and explores additional structures and patterns involved in Chinese situation aspect. By recognizing the role of PPs and other types of bounding expressions (rows [f]--[k] in Table 2), it provides a better account of natural data. To illustrate, (41) below is an excerpt that describes a series of moving activities. While no -le is used in any of the clauses, it is the bounding expressions that render the situations bounded (dao-PPs in [a], [b] and [d], RVCs --xia ‘down’ in [a], chu ‘out’ and jin ‘enter’ in [c]). (41)

(Chen 1981: ch 6) kai dao Fu Jiajie jia menkou ting-xia. (b) deng Fu Jiajie chan-zhe (a) ‘‘130’’ (a) (vehicle) drive to Fu Jiajie home doorway stop-down. (b) wait Fu Jiajie assist-ZHE Lu Wenting yibuyiaidi zou dao chebian shi, (c) siji mang shen-chu Lu Wenting one.step.one.pause walk to vehicle.side time, (c) driver quickly reach-out da shoulai ba Lu Wenting fu-jin jiashishi, (d) yizhi xiaoxindi ba che big hand BA Lu Wenting assist-enter vehicle, (d) all.the.way carefully BA vehicle kai dao yiyuande jizhenshi. drive to hospital’s emergency.room. ‘They drove up to Jiajie’s door and stopped. The driver waited until Jiajie had carefully assisted Wenting to the side of the truck. Then he reached out a big hand to help her into the cab and they cautiously drove straight to the emergency room.’

In the next section, additional examples of PPs and other types of bounding expressions (e.g., headed by zai in [44c] and gei in [45d], respectively) can be found when additional excerpts from natural discourse are presented. Before leaving this section, a remark is in order concerning the progressive aspect marker zai, which appears before the verb and posts a counter-example to the current proposal. At this point, I do not have a satisfactory explanation for its behavior. But I’d like to highlight some of the complexities involved with its use. Zai as a coverb/preposition has already been discussed in Section 4.2.3.4. Here I focus on its use as an aspectual marker.

2060

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

Zai as a progressive aspect marker is a recent development of modern Chinese roughly during the last century. This function does not seem to be as clear and as fully established as that of the other aspectual markers. Currently, it still carries a tone of the written genre and, in speech, is often replaced by other durative particles such as --zhe and --ne (Zhang, 2011). In terms of developmental path, the other aspectual markers, such as --le, -guo and --zhe, follow a similar route from verbs first to verbal complements and then to aspect markers through grammaticalisation. The history of zai, while having some of these features, is also closely related to adverbs zhengzai and zheng, both indicating ongoing situations and appearing before the verb. The fact that zai is part of the word zhengzai shows their close relation. Consequently, many researchers treat zai as an adverb rather than an aspectual marker (Dai, 1997: 105; Lyu, 1980; Zhang, 2011, 2002; Zuo, 2007).16 In terms of syntactic position, zai differs from other aspect markers in two accounts. One is that, while all the other aspect markers appear after the verb, zai occurs in the preverbal position. The other is that the other aspect markers are verb suffixes, while zai may be separated from the verb by various adverbials. In (42), for example, the zai is separated from the verb pao (cha) ‘make (tea)’ by a beneficiary gei-PP; in (43) an adverb buduandi ‘continuously’ appears between the zai and the verb nuli ‘make effort’ (also see [50d] below). This explains why zai as an aspect marker is not analyzed (and not hyphenated in gloss) as an affix to the verb as all the other aspect markers are. (42) Ta zai gei keren pao cha. he ZAI for guest make tea ‘He is making tea for the guest(s)’ (43) Tamen zai buduandi nuli. They ZAI continuously make.effort They are making continuous efforts. This, combined with the other uses of zai, seems to show that zai is a word whose many uses have come into existence under the pressure from multiple directions. The different uses are entangled and may also be in a state of flux. As many scholars have observed, grammar of a language and diachronic changes are the result of competing motivations (e.g., analogy, generalization, reanalysis, semantic shift, economy, and expressivity) (Haiman, 1985; Peyraube, 1988, 1999). Lexical items and structural patterns may have different developmental history and be at different developmental stages. The grammar of modern Chinese, compared to that of the classical period, has been streamlined to a large extent. But, because of the competing motivations and principles, few generalizations or even grammar rules work 100%. While the study of the word zai and other prepositions is far from complete, the competing motivations and principles involved in their current multiple functions and the potential counter-examples they may pose for any analysis should not be a surprise. Next, I move on to examine a pattern in Chinese narrative discourse, in which aspectual reference plays a crucial role. 6. Aspectual reference and narrative advancement In this section, excerpts from narrative and news reports are presented to demonstrate a key role of perfectivity encoding in Chinese: It facilitates the advancement of narrative time. Before presenting the details, a few words about narrative texts are necessary. The narrative mode of text (also see Section 3.2.2) may be divided generally into two portions: main on-storyline events and background descriptions (including comments and elaberations). Main events are temporally related to each other and typically recounted in the order in which they occur. As the events unfold in sequence, narrative time advances. Temporal dynamism is a vital feature. Research has found that temporal adverbials and/or perfective events facilitate narrative advancement. Unbounded or imperfective predications do not have such capability (Cappelle and Declerck, 2005: 896; Declerck, 1989; Depraetere, 1995; Smith 1997: 36). Narrative time typically advances in the following manner: the text or passage begins with a given reference time, explicit or implicit. When no explicit time expression is present and two perfective clauses are temporally ordered with respect to each other, the two events are separated by a temporal juncture that specifies the endpoint of the preceding event. As the events unfold in sequence, the reference time advances one step at a time to a point just after that event (Dry, 1981, 1983; Hopper, 1979a,b; Labov, 1972; Labov and Waletzky, 1967; Smith, 2004). Dowty (1986: 37) describes

16

In Chinese, aspect markers are a closed set of grammatical morphemes, whereas adverbs are an open class of content words.

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2061

this mechanism as follows: ‘‘If a sentence in a narrative contains an accomplishment or achievement predicate but no definite time adverb, that sentence is understood to describe an event occurring later than the time of the previous sentence’s event (or in the literary analyst’s terms, narrative time ‘moves forward’ in the second sentence).’’ ‘‘The general conventions for narrative texts hold for Mandarin,’’ as Smith (1997: 279) observes. ‘‘Sentences with the perfective viewpoint tend to advance a narrative by the convention of Sequentiality: perfectively presented events follow one another in sequence. In contrast, sentences with the imperfective viewpoint provide background information.’’ In the examples below, temporal location of the events is implicit and by default in the past (Section 3.2.2), we can see that --le and bounding expressions (underlined) are used to identify temporal junctures for narrative advancement. The passage in (44) consists of three clauses (a)--(c) describing three events in a temporal sequence. The clauses are all perfective, although only one --le is used in the first clause. Between (a) and (b) the temporal juncture is the endpoint of the first event, designated by both the --le and the RVC chulai ‘out.’ The narrative time advances to the point just after that moment. The event in (b) is bounded by the RVC ganjing ‘clean.’ The narrative time again advances to the moment just after that. The same pattern is repeated in (c), with the bounding expression being the zai-PP. (44)

(Lu 1984: 130) (a) Wo ba nage zhubangzi zhao-le-chulai, (b) kaifu-ganjing, (c) fang zai shoubian. (a) I BA that clapper find-LE-out, (b) wipe-clean, (c) put ZAI close.by. ‘I got the clapper out and wiped it clean. I held it in my hands.’

In (45) are four events narrated in the order in which they occur. The narration advances one step in each clause in the same pattern as above. Temporal junctures are identified by a --le in (a), a bounding expression in (b) and (c), and combinations of the two in (d). (45) (a) Wenwen xia-le che, (b) tiao-jin menlang, (c) dongshou jiu jie-kai (a) (name) descent-LE bicycle, (b) jump-into porch, (c) start then untie-open (Wang, 1999: 23) yupi, (d) jiao-gei-le xiaohuozi,.. . . raincoat, (d) hand-to-LE young.man, . . . ‘Wenwen got off and ran into the portico where she took off the raincoat and handed it back to the young man . . .’ The excerpt in (46) below consists of six clauses that narrate a series of actions in the same pattern. Each action is bounded by a --le and/or a bounding expression, which also identify the temporal juncture between the two consecutive events. The clause in (c) has two verbs. The first one deng ‘wait’ is marked by the imperfective marker --zhe and is subordinating in nature (Section 4.1). The second verb in the perfective form chi-jing-le ‘eat-finish-LE’ is the main verb. (46)

(Lao She 1936: ch 10) (a) laozhe chi-wan ziji de fenr, (b) ba huaizhong de jiu he-gan, (a) old.man eat-finish self MOD share, (b) BA hand-in MOD liquor drink-dry, (c) deng-zhe Xiao Mar chi-jing-le baozi. (d) tao-chu kuai bu lai, (c) wait-ZHE Litte Horse eat-finish-LE roll. (d) bring-out a.piece.of cloth (come), (e) ca-le-ca zui, (f) ta you xiang dajia dian-le-dian tou: ‘‘. . .’’ (e) wipe-LE-(dupl.) mouth, (f) he then toward everyone nod-LE-(dupl.) head: ‘‘. . .’’ ‘‘The old man finished eating his share, drained the last drop of wine from his cup, and sat waiting for Little Horse to eat the rest of the meat rolls. He brought out a piece of tattered cloth and wiped his mouth with it. Nodding his head to the crowd, he said: ‘‘. . .’’

The pattern seen in (44)--(46) is by no means unusual. It is typical of the narration of major events in general, spoken or written. In fact, some of the earlier examples, e.g., (40) and (41), already demonstrate this pattern. In (47) below is one from oral story-telling. Of the four clauses, three are bounded by --le and/or bounding expressions. Clause (b) uses a special structure ‘‘yi ‘one’ + verb,’’ which designates an instantaneous action bounded by nature.17

17 The clause is part of a complex structure with the following clause denoting the result of the action. Space limitations here preclude detail of such special constructions.

2062

(47)

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

(SAVP 2000) (a) Ta ganjin zou-chu dong lai, (b) you yi tiao, (c) chuan-guo pubu, (a) he hurriedly walk-out cave (come), (b) again one jump, (c) go-through waterfall. (d) ba ta kanjian de gen dahuo shuo-le yibian. (d) BA he see MOD to everyone say-LE once. ‘He hurried out of the cave. He made another jump through the waterfall and told the others what he saw.’

From these examples a clear pattern emerges: Temporal advancement in each clause follows a two-step procedure. First, the verb identifies the dynamic event and then a --le and/or a bounding expression encodes the end boundary of the event. The two steps alternate like two footsteps moving the narrative time forward as a series of events unfolds. Since perfectivity can be signaled either by a --le or a bounding expression, a --le is not required. This explains the nonobligatory use of --le in perfective clauses. Very often, the series of clauses narrating such consecutive events form a topic chain, in which the topic/subject/agent NP (wo ‘I’ in [44], Wenwen in [45], laozhe ‘old man’ in [46] and ta ‘he’ in [47]) appears only in the first clause. The tightly knit structure of such topic chains reflects the close relation of the encoded events iconically.18 By contrast, clauses with neither --le nor bounding expressions tend to be imperfective. They designate unbounded events or states overlapping in time. These clauses tend to be part of the background descriptions. To see this, examine (48), in which two of the clauses, (a) and (f), describe dynamic events. The action ta ‘to pedal’ in (a) is presented imperfectively by the durative marker --zhe; the clause in (f), also imperfective, depicts a repetitive up and down motion. The rest of the clauses are all state predications. Neither --le nor bounding expressions are used.19 The narrative time remains static. (48)

(Wang 1999: 17) ta-zhe chezi, (b) ding feng, (a) Xiaohuozi fenli (a) young.man struggle pedal-ZHE bike (b) go.against wind, de fudan, (d) kanqilai youdian chili. (c) you zengjia yige ren (c) also add one.M person MOD burden, (d) look a.bit difficult (e) ta shenti qian qing, (f) kuankuande jianbang yishang yixia. (e) he body forward lean, (f) broad shoulder up down (g) er Wenwen zuo zai-zhe-kuan-jianbang-houtou, (g) and (name) sit ZAI-behind-this-broad-shoulder, (h) dao neng bibi yu le. (h) on.contrary can take.shelter.from rain LE ‘The young men pedaled hard against the wind. With the added burden it looked like he was struggling. He was leaning forward, his broad shoulders were going up and down. Wenwen, sitting behind such shoulders, was protected from the rain.’

The narrative mode of text may consist of a mixture of perfective and imperfective clauses, as shown in (49) below. Among the nine clauses, seven are perfective. One ([c]) is bounded by the perfective --le only, four ([b], [f], [g], and [i]) use bounding expressions without --le, and two ([d] and [e]) use both. By contrast, the other two clauses, (a) and (h), are imperfective, both marked by the durative --zhe. Note that, when a verb is marked by --zhe, no bounding expressions can be used for the same verb, apparently because of their opposing functions. (49) (Chen 1981: ch 6)

18

The structure of topic chains will be dealt with in another paper. Two things need to be explained in this example. First, the zai-PP in (g) here does not encode event bounding. In general, whether a postverbal zai-PP encodes event bounding depends on whether the clause occurs in a dynamic or non-dynamic context. In this example, the context is non-dynamic, thus the zai-PP (and the clause) describes a state. In comparison, the zai-PP in (46c) is in a dynamic context. It is one of a series of clauses describing dynamic actions. In that context, the zai-PP serves as a bounding expression (also see Liu, 2009). Secondly, the le in (h) is a sentence-final le, which encodes a state, rather than a verb suffix --le. 19

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2063

(a) Zhao Tianhui wan-zhe Sun Yiminde shoubi, (b) kua-jin Lu Wentingde bingfang, (a) Zhao Tianhui pull-ZHE Sun Yimin’s arm, (b) step-into Lu Wenting’s room, (c) cai ting-le huatou. (c) only.then stop-LE talking. (d) Fu Jiajie zaoyi zhan-le-qilai, (e) Zhao Tianhui chong ta hui-le hui shou, (d) Fu Jiajie already stand-LE-up, (e) Zhao Tianhui to him wave-LE (dupl.) hand, (f) jiu yizhi zou-jin chuangbian, (g) wan-xia yao qu, (h) duanxiang-zhe (f) then straight walk-to bed.side, (g) bend-down back (go), (h) examine-ZHE bingrende lianse, (i) you cong zhiban yisheng shoushang jie-guo bingli. patient’s look, (i) also from on.duty doctor hands receive-over medical.record. ‘Zhao Tianhui did not stop talking until he had stridden into Wenting’s room, pulling Sun Yimin by the arm. Fu Jiajie stood up as Zhao Tianhui, passing by him with a wave of the hand, headed straight to the patient’s bedside and bent over her. As he scrutinized her face, he reached for the case record from the attending physician.’ Example (50) is extracted from a news report about a visit by the Obamas to Georgetown. The excerpt is in the narrative mode. The preceding text states that the visit was made on a Saturday evening. (50)

HX090504 (a) Obamafufu daofang Qiaozhicheng xiyin-le zhongduov weiguanzhe, the.Obamas’ visit Georgetown attract-LE many onlooker (b) jingfang yong jingjiexian lanzhu renqun, (c) bing shi’er ming di, police use tape block-hold crowd, (c) also occasionally blow whistle (d) ling you yiming kangyizhe zai canguanwai gaohu kouhao. also exist one-M protester ZAI restaurant.outside shout slogan ‘The Obamas’ visit to Georgetown attracted thick crowds of onlookers, who were held back by police tape, while sirens occasionally wailed and a protester chanted outside the restaurant.’

Again, with no explicit indication of time in (50), the temporal interpretation of the events is determined at the discourse level by inference from the previous text. Aspectual reference is managed at the clause level. The --le in (a) encodes the event of attracting onlookers as bounded and actualized. In (b), the event is delimited by the RVC (lan)-zhu ‘(block)-hold.’ These two bounded events stand in contrast to the unbounded and on-going events in (c) and (d), where neither --le nor bounding expressions are used. 7. Concluding remarks This article describes the general patterns of temporal and aspectual references in Chinese. It has demonstrated that syntactic positions before and after the main verb have semantic and pragmatic functions. Temporal (and physical) location of events is expressed before the main verb, while aspectual information is provided after the verb. This is an important pattern of information distribution at the clause level. It also provides an optimal explanation for a number of word order phenomena observed in previous studies, including the historical change involving the positioning of PPs. This study has also shown that the perfective aspect may be indicated by the perfective marker -le (viewpoint aspect), bounding expressions (situation aspect), or combinations of the two. The multiple means provide an explanation for the non-obligatory use of --le, which has so far been a thorny issue in the analysis. It is evident that the crucial role of bounding expressions should be duly recognized in an adequate account of perfectivity in Chinese. Temporal reference in Chinese is to a large extent a discourse phenomenon. Context and pragmatics play a significant role. The early appearance of overt temporal expressions in a discourse segment, be it a clause or a stretch of discourse, serves as an anchor or creates a scope for the temporal interpretation of the unit. The length of such segments varies. Later segments may also be interpreted with reference to this anchor. By contrast, aspectual information pertains to the individual event and is thus managed locally at the clause level. This contrast has psycholinguistic implications. In discourse, the information of temporal location is generally made available before that of aspect. Therefore, in language processing, temporal location is determined and processed before aspect. The former also serves as a temporal scope within which the latter is interpreted. Time advances in the narration of major on-storyline events. This study contributes to this line of research by demonstrating the pattern of temporal advancement in Chinese narrative. It shows that Chinese adopts an iconic two-step procedure in each narrative clause. In the first step, the verb specifies the dynamic event; in the second one, a --le and/or a

2064

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

bounding expression encodes the event endpoint. The --le and bounding expressions also serve to identify temporal junctures between events. They are indispensable devices to facilitate narrative advancement. Such details have so far evaded due attention in the analysis. Questions remaining include when and why --le and a bounding expression co-occur. This is a crucial issue for the treatment of aspect in Chinese. A number of proposals have been made taking discourse factors into account. Andreasen (1981), for example, holds that --le has the function of marking foreground in narrative discourse. Chang (1986) proposes that --le marks the peak clause in a discourse segment, which very often falls at the end of a discourse unit. Further investigations would require in-depth studies that incorporate discourse structure and aspect marking. A statistical analysis of a larger corpus would also be desirable. Because of the absence of the overt grammatical marking of tense and the non-obligatory use of aspectual markers in the clause, Chinese shows a different division of labor between grammar, lexicon, and mechanisms at the discourse level when indicating temporal and aspectual distinctions. Lexicon, discourse pragmatics and even word order take on some of the functions served by grammatical morphology in other languages. By examining how the different types of temporal and aspectual information are integrated and encoded in Chinese, we have a better understanding of the linguistic repertoire and its inner working in the language. The division within the clause for encoding temporal and aspectual information in Chinese adds a new dimension in the analysis and opens a fruitful perspective for future research. Conflict of interest This article reports on original research that has not been published elsewhere. References Andreasen, Andrew John, 1981. Backgrounding and foregrounding through aspect in Chinese narrative literature. Stanford University Dissertation, California. Bittner, Maria, 2008. Aspectual universals of temporal anaphora. In: Susan Rothstein (ed.), Theoretical and Crosslinguistic Approaches to the Semantics of Aspect. John Benjamins, Amsterdam, Netherlands, pp. 349--385. Bull, William E., 1971. Time, Tense, and the Verb. University of California Press, Berkeley. Cappelle, Bert, Declerck, Renaat, 2005. Spatial and temporal boundedness in English motion events. Journal of Pragmatics 37 (6), 889--917. Chang, Vincent Wuchang (1986). The Particle LE in Chinese narrative discourse. University of Florida dissertation, Gainesville, Florida. Chao, Yuen-Ren, 1968. A Grammar of Spoken Chinese. University of California Press, Los Angeles. Chen, Chung-yu, 1978. Aspectual features of the verb and the relative positions of the locative. Journal of Chinese linguistics 6 (1), 76--103. Chen, Ping, 1988. Lun xiandai hanyu shijian xitong de sanyuan jiegou (On the three-component structure of the temporal system in modern Chinese). Zhongguo Yuwen 1988 (6), 401--422. Chen, Qianrui, 2005. Dangdai timao lilun yu hanyu sicengji de timao xitong (On the aspectual theories and the four-level chinese aspectual system). Hanyu Xuebao (Chinese Linguistics), 2005 (3), 20--27. Christensen, Matthew Bruce (1994). Variation in spoken and written Mandarin narrative discourse. Ohio State University Dissertation, Ohio. Chu, Chauncy C., 1998. A Discourse Grammar of Mandarin Chinese. Peter Lang, New York. Comrie, Bernard, 1976. Aspect: An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge University Press, Cambridge. Dahl, Ӧsten, 1981. On the definition of the telic-atelic (bounded-unbounded) distinction. In: Philip, J., Tedeschi, Annie Zaenen, (Eds.), Tense and Aspect (Syntax and Semantics, Vol. 14) Academic Press, New York, pp. 79--90. Dahl, Ӧsten, 1985. Tense and Aspect Systems. Basil Blackwell, Oxford, UK. Dai, Yaojing, 1997. Xiandai Hanyu Shiti Xitong Yanjiu (A Study of Aspect in Modern Chinese). Zhejiang Education Press, Hangzhou. Declerck, Renaat, 1979. Aspect and the bounded/unbounded distinction. Linguistics 17, 761--794. Declerck, Renaat, 1989. Boundedness and the structure of situations. Leuvense Bijdragen 78, 275--308. Declerck, Renaat, 2006. The Grammar of the English Tense System: A Comprehensive Analysis (The Grammar of the English Verb Phrase, Vol. 1) Mouton de Gruyter, New York. Depraetere, Ilse, 1995. On the necessity of distinguishing between (non)boundedness and (a)telicity. Linguistics and Philosophy 18, 1--19. Dowty, David R., 1986. The effects of aspectual class on the temporal structure of discourse: semantics or pragmatics? Linguistics and Philosophy 9, 37--61. Dry, Helen Aristar, 1981. Sentence aspect and the movement of narrative time. Text 1, 233--240. Dry, Helen Aristar, 1983. The movement of narrative time. The Journal of Literary Semantics 12 (2), 19--54. Friedrich, Paul, 1974. On aspect theory and Homeric aspect. International Journal of American Linguistics 40 (4), (Part 2, Memoir no. 28). Haiman, John, 1985. Natural Syntax: Iconicity and Erosion. Cambridge University Press, Cambridge. He, Leshi, 1985. Shiji yufa tedian yanjiu (A study of the grammar in Shiji). In: Cheng Xianqing (ed.), Lianghan hanyu yufa yanjiu.(Studies in the Grammar of Han Chinese). Shandong Jiaoyu Chubanshe, Jinan (PRC), pp. 1--261. Hinrichs, Erhard, 1986. Temporal anaphora in discourses of English. Linguistics and Philosophy 9, 63--82. Hopper, Paul, 1979a. Some observations on the typology of focus and aspect in narrative language. Studies in Language 3, 37--64. Hopper, Paul, 1979b. Aspect and foregrounding discourse. In: Talmy Givon (ed.), Discourse and Syntax (Syntax and Semantics 12). Academic Press, New York, pp. 213--241. Hopper, Paul (Ed.), 1982. Tense-aspect: Between Semantics and Pragmatics. John Benjamins, Amsterdam. Huang, Shuan-fan, 1981. The scope phenomena of Chinese quantifiers. Journal of Chinese Linguistics 9 (2), 226--243.

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

2065

Jackrndoff, Ray, 1990. Semantic Structure. MIT Press, Cambridge, MA. Klein, Wolfgang, Ping, Li, Henriette, Hendriks, 2000. Aspect and assertion in Mandarin Chinese. Natural Language and Linguistic Theory 18, 723--770. Labov, William, 1972. The Transformation of Experience in Narrative Syntax. Language in the Inner City: Studies in the Black English Vernacular. University of Pennsylvania Press, Philadelphia. Labov, William, Waletzky, Joshua, 1967. Narrative analysis: oral versions of personal experience. In: June Helm (ed.), Essays on the Verbal and Visual Arts. University of Washington Press, Seattle, pp. 12--44. Li, Ying-Che, 1980. The historical development of the coverb and the coverbial phrase in Chinese. Journal of Chinese Linguistics 8, 273--293. Li, Yen-hui Audrey, 1990. Order and Constituency in Mandarin Chinese. Kluwer Academic Publishers, Boston. Li, Charles N., Thompson, Sandra A., 1974. Co-verbs in Mandarin Chinese: verbs or prepositions? Journal of Chinese Linguistics 2 (3), 257--278. Li, Charles N., Thompson, Sandra A., 1975. The semantic function of word order: a case study in Mandarin. In: Charles, N., Li, (Eds.), Word Order and Word Order Change. University of Texas Press, Austin, TX, pp. 457--490. Li, Charles N., Thompson, Sandra A., 1981. Mandarin Chinese: A Functional Reference Grammar. University of California Press, Berkeley, CA. Light, Timothy, 1979. Word order and word order change in Mandarin Chinese. Journal of Chinese Linguistics 7 (2), 149--181. Liu, Feng-hsi, 2009. Aspect and the post-verbal zai phrase in Mandarin Chinese. In: Janet Zhiqun Xing (ed.), Studies of Chinese Linguistics, Functional Approaches. Hong Kong University Press, Hong Kong, pp. 103--129. Lu, John H.T., 1977. Resultative verb compounds vs. directional verb compounds in Mandarin. Journal of Chinese Linguistics 5, 276--313. Lyu, Shuxiang, 1980. Xiandai Hanyu Babai Ci (Eight Hundred Words in Modern Chinese). Commercial Press, Beijing. Mei, Zulin, 1981. Xiandan hanyu wandhcnemao jushi de ciwei de laiyuan (The origin of the perfective aspect construction and the perfective suffix in Modern Chinese). Yuyan Yanjiu 1981 (1), 65--77. Mullie, Joseph, 1932. The Structural Principle of the Chinese Language, vol. 1. The Bureau of Engraving and Printing, Peking. Packard, Jerome L., 2000. Morphology of Chinese: A Linguistic and Cognitive Approach. Cambridge University Press, New York. Packard, Jerome L. (Ed.), 1998. New Approaches to Chinese Word Formation: Morphology, Phonology and the Lexicon in Modern and Ancient Chinese. Mouton de Gruyter, New York. Partee, Barbara, 1973. Some structural analogies between tenses and pronouns in English. The Journal of Philosophy 70 (18), 601--609. Partee, Barbara, 1984. Nominal and temporal anaphora. Linguistics and Philosophy 7 (3), 243--286. Peyraube, Alain, 1988. Syntactic change in Chinese: on grammaticalization. The Bulletin of the Institute of History and Philology 59, 617--652. Peyraube, Alain, 1999. Historical change in Chinese grammar. Cahiers de linguistique -- Asie orientale 28 (2), 177--226. Reichenbach, Hans, 1947. Elements of Symbolic Logic. Macmillan, New York. Reinhart, Tanya, 1984. Principles of gestalt perception in the temporal organization of narrative texts. Linguistics 22, 779--809. Ross, Claudia, 1990. Resultative verb compounds. Journal of the Chinese Language Teacher’s Association 25 (3), 61--83. Ross, Claudia, 1995. Temporal and aspectual reference in Mandarin Chinese. Journal of Chinese Linguistics 23 (1), 87--136. Sasse, Hans-Jürgen, 2002. Recent activity in the theory of aspect: accomplishments, achievements or just non-progressive state? Linguistic Typology 6, 199--271. Smith, Carlota S., 1983. A theory of aspectual choice. Language 59, 479--501. Smith, Carlota S., 1994. Aspectual viewpoint and situation types in Mandarin Chinese. Journal of East Asian Linguistics 3, 107--146. Smith, Carlota S., 1997 [1991]. The Parameter of Aspect. Kluwer, Dordrecht. Smith, Carlota S., 2003. Modes of Discourse. Cambridge University Press, Cambridge. Smith, Carlota S., 2004. The domain of tense. In: Jacqueline Guéron, Jacqueline Lecarme, (Eds.), The Syntax of Time. MIT Press, Cambridge, MA, pp. 597--619. Smith, Carlota S., Erbaugh, Mary, 2005. Temporal interpretation in Mandarin Chinese. Linguistics 43 (4), 713--756. Sun, Chaofen, 1991. The adposition Yi and word order in Classical Chinese. Journal of Chinese Linguistics 19 (2), 202--218. Sun, Chaofen, 1996. Word Order Change and Grammaticalisation in the History of Chinese. Stanford University Press, California. Tai, James H.-Y., 1975. On two functions of place adverbials in Mandarin Chinese. Journal of Chinese linguistics 3 (2/3), 154--179. Tai, James H.-Y., 1983. Temporal sequence and Chinese word order. In: John Haiman (ed.), Iconicity in Syntax (Typological Studies in Language, Vol. 6) John Benjamins, Amsterdam, pp. 49--72. Tai, James H.-Y., 1984. Verbs and times in Chinese: Vendler’s four categories. In: Papers from the Parasession on Lexical Semantics Chicago Linguistic Society, Chicago, 289--296. Teng, Shouxin, 1975. A Semantic Study of Transitory Relations in Chinese. University of California Publications, Berkeley, Linguistics 80. Thompson, Sandra A., 1973. Resultative verb compounds in Mandarin Chinese: a case for lexical rules. Language 49 (2), 361--379. Vanden Wyngaerd, Guido, 2001. Measuring events. Language 77 (1), 61--90. Vendler, Zeno, 1957. Verbs and Times. The Philosophical Review 66, 143--160. Wang, William S.-Y., 1965. Two aspect markers in Mandarin. Language 41, 457--470. Xiao, Richard, McEnery, Tony, 2004. Aspect in Mandarin Chinese: A Corpus-based Study. John Benjamins, Amsterdam and Philadelphia. Yang, Jun (2002). The acquisition of temporality by adult second language learners of Chinese. University of Arizona Dissertation, Tucson, AZ. Zhang, Yajun, 2002. Shijian fuci ZHENG, ZHENGZAI, ZAI jiqi xuhua guocheng kaocha (The study of the temporal adverbs zheng, zhengzai and zai in Chinese and their grammaticalisation). Journal of Shanghai Teachers University 31 (1), 46--55. Zhang, Jie, 2011. Putonghua fuci ZAI yuanliu kaobian (An investigation of the origin and evolvement of the adverb zai in Mandarin). Language Teaching and Linguistic Research 2011 (1), 76--81. Zuo, Simin, 2007. Hanyu shiti biaoji xitong de gujin leixong bianhua (A diachronic study of tense and aspect marking in Chinese). Hanyu Xuebao (Chinese Linguistics), 2007 (2), 11--23.

Sources of narrative examples Chen, Rong, 1981. Ren Dao Zhongnian (At Middle Age). Baihua Wenyi Chubanshe, Tianjin, (translation by Margaret Decker, 1984. In: Perry Link (Ed.), Roses and Thorns. University of California Press, Los Angeles).

2066

W. Li / Journal of Pragmatics 44 (2012) 2045--2066

Lao, She, 1936. Luotuo Xiangzi (Rickshaw Boy). Haitai Publishing, Beijing, (translation by Evan King, 1945. Reynal and Hitchcock, New York). Lu, Wenfu, 1984. Xiaofan shijia (The Man from a Peddlers’ Family). Chinese Literature: Distributed by China International Book Trading Corp., Beijing, China, (translated by Ralph Lake, 1986. In: Lu Wenfu (Ed.), A World of Dreams). Shanghai Audio and Visual Publishing, 2000. Huaguoshan [Flower and Fruit Mountain, Stories from A Journey to the West, ML-11] (Audio). Shanghai Audio and Visual Publishing, Shanghai. Wang, Anyi, 1999. Wang Anyi Xiaoshuo Xuan (Bilingual Version) (Selected Stories by Wang Anyi). Chinese Literature Press and Foreign Language Teaching and Research Press, Beijing. Wang, Meng, 1981. Hudie (The Butterfly). In: Wang Meng Xiaoshuo Baogaowenxue Xuan (Selected Novels and Reportages by Wang Meng). Beijing Chubanshe, Beijing, (translated by Gladys Yang, 1983. In: Wang Meng (Ed.), The Butterfly and Other Stories. Panda Books, Beijing). Wendan Li is an associate professor of Chinese language and linguistics at the University of North Carolina at Chapel Hill. Her research focuses on the syntax-semantics interface and discourse structure of Chinese. She has also conducted research in second language teaching and learning of Chinese.