|Home||Linguistik online 50, 6/2011|
θ-structures of Mandarin Resultative Verb Compounds*
Fengqi Li (Chongqing)
Please note that some of the syntactic representations in this paper may be displayed incorrectly in some cases, depending on the browser that is being used. Please see the PDF-paper for guaranteed correct display.
A Resultative Verb Compound (RVC) is a Mandarin construction composed of a verb and its complement denoting the result (cf. Lv 1980)1. The verb, which can be either transitive or intransitive, comes before the complement, which can be a verb (as in gan-pao, chase-run, meaning drive away), or an adjective (as in ku-hong, cry-red, meaning crying so much that one’s eyes turn red), or a bound morpheme (as in mai-zhao, buy-ASP, meaning having been bought, where zhao is a bound morpheme indicating the present state). Semantically, the verb indicates the causing action or event and the complement the resulting state or event (cf. Her 1999). In gan-po, for example, the verb gan, or chase, is the causing action which leads to pao, or run, the resulting event. That is to say, one’s chasing has caused the running of another entity. According to this kind of cause-effect relation, we represent the verb and its complement respectively as Vcaus and Cres.
RVCs are much more complicated than any simplex verbs in Mandarin in terms of their thematic-role structures (or θ-structures) since both Vcaus and Cres have the capacity of assigning their own thematic roles. This, of course, does not apply to RVCs with a bound morpheme in the complement position, which only serves the verbs preceding them, indicating the result of the action, and assigns no thematic roles to either the subject or object. In this sense, RVCs containing a bound morpheme have no difference with other simplex verbs like xiangshou enjoy, kefu overcome, etc. Since this study focuses on the complexity of thematic roles in RVC constructions, this kind of RVCs would not be taken into account in the following discussion. So whenever RVCs are mentioned hereinafter, those with bound morphemes are automatically excluded.
Though there have been a lot of studies on RVCs, few of them have thoroughly examined their θ-structures. The perspectives of previous studies include but not limited to2: the Head of RVCs (Lv & Zhu 1952; Ding 1961; Li 1984; Ma 1987; Zhan 1989; Song 2004a; Xiong/ Liu 2005; Lin 2009), transitivity of RVCs (Zhu 1982; Li 1994), sentence patterns in which RVCs occur (Li 1980; Fan 1987; Song 2004c; Huang 2008), implications of Cres (Fan 1985; Wang 1996), semantic reference of Cres (Zhan 1989; Li 1990; Mei 1994; Jiang 2007), valency of RVCs (Fan 1985; Huang 1993; Guo 1995; Wang 1995; Song 2004b; Shi 2006), correlation between different levels of RVCs, like lexicon-syntax correlations (Lin 1996; Lin 1999), and syntax-semantics correlations (Lu 1977; Huang/Lin, 1993; Li 1995; Claudia 1990; Her 1999).
The current study is aimed to present a clear picture of the θ-structures of RVCs. It is observed that frequently a RVC can assign more than two thematic roles, and sometimes a composite role is formed when a single entity bears two different roles. This is further compounded by ambiguity and causation of RVCs.
2 Thematic Roles
To study the θ-structures of RVCs, we have to first have a look at the thematic roles in general. Saeed (1997: 140–145) identifies 9 generally accepted thematic roles (The relevant role-bearing nominal is italicized in the examples below each role):
1. Agent: the initiator of some action, capable of acting with volition, e.g.
2. Patient: the entity undergoing the effect of some action, often undergoing some change in state, e.g.
3. Theme: the entity which is (literally or figuratively) moved by an action, or whose location is described, but constitutionally unchanged, e.g.
(4) He gives lessons to a group of young pupils.
4. Experiencer: the entity which is aware of the action or state described by the predicate but which is not in control of the action or state, e.g.
(6) The boy was frightened by the dog.
5. Beneficiary: the entity for whose benefit the action was performed, e.g.
(8) He bought me a watch.
6. Instrument: the means by which an action is performed or something comes about, e.g.
7. Locative: the place in which something is situated or takes place, e.g.
8. Goal: the entity towards which something moves, either literally or metaphorically, e.g.
(12) Pat told his story to everyone he knew.
9. Source: the entity from which something moves, either literally or metaphorically, e.g.
(14) This clever idea is from Thomas Jefferson.
It has been noted that there is a hierarchical relation between thematic roles as far as prominence is concerned. For example, the agent has primacy over the patient or beneficiary; where both are present, the patient or beneficiary is affected as a result of something the agent does. Bresnan and Kanerva (1989) propose the thematic hierarchy (see below), where the left most role is the most prominent and the right most the least prominent.
agent > beneficiary > experiencer/goal > instrument > patient/theme >locative
The hierarchy is not an independent stipulation but a consequence of the conceptualization.
Accordingly, the thematic roles in the θ-structure are arrayed by their relative prominence in the hierarchy: the most prominent is put left most, and the least right most.
3 Θ-structures of RVCs in General
As has been stated, RVC constructions are complex in their θ-structures. Take da-bai (defeat) for example. Vcaus da (literally beat) is transitive, and has two thematic roles: agent and patient, hence has such θ-structure as <ag pt> (ag = agent, pt = patient). Cres bai (literally defeated) is intransitive, and has the θ-structure of <th> (th = theme)3. As a whole, the θ-structure of the compound can be represented as
where the inner angled brackets indicate “embedded” since it is the θ-structure of the embedded predicate Cres, and the broken line indicates that the two roles are borne by a single entity. We regard those two different roles like the patient and theme in the example which are co-borne by one single entity as forming a composite role, i.e. pt and th here together are called a composite role. The roles both in the outer angled brackets and the inner ones are arrayed respectively according to the thematic hierarchy mentioned above. What’s more, the main θ-structure always precedes the “embedded” one.
In what follows, RVCs will be classified roughly based on the composition of the θ-structures: whether they contain composite roles or not. Different pictures of the θ-structures under each category will be presented by giving typical examples. All the examples to appear come from the online Chinese Corpus of Center for Chinese Linguistics PKU (http://ccl.pku.edu.cn:8080/ccl_corpus/index.jsp?dir=xiandai), unless otherwise stated.
3.1 RVCs without composite roles
A number of RVCs can fall into this category: ku-hong cry-red, ku-sha cry-foolish, bing-si fall-sick-die, etc.
In the examples above, no roles are co-borne by one entity, therefore, no composite roles can be identified. Note that (16a) and (16b) have the same θ-structures but different syntactic representations. Also note the subscripted “i” in (17) which indicates the coindexation of the two roles, or the identical reference of them: both referring to ren. They do not form a composite role because they are identical roles instead of different ones. See another example of coindexation:
Let’s see one more example.
Note that not all the entities bearing a certain role show themselves in the syntactic representation: since kan cut is transitive, its θ-structure should contain two arguments: agent and patient, but in (19) both the entities bearing the agent role and the patient role do not appear. (Such roles are italicized in the θ-structures.).
3.2 RVCs with composite roles
It is typical of RVC constructions that one entity often bears more than one thematic role. These roles form (or merge into) a composite role. Sometimes it is the most prominent role in the main predicate θ-structure, that is, , that merges with the role in the embedded predicate θ-structure, and sometimes it is the role other than it. For example,
The agents in the two sentences merge with the theme and experiencer respectively.
What makes things more complicated is where inversion is concerned. See below:
The patient yao is not in its usual position—the OBJ position; instead, it is in the position of SUBJ. Hence inversion occurs (cf. Her 1996; 2007). Inversion does not occur within this subcategory only; it also occurs with RVCs without composite roles. (17), for example, is another kind of inversion: locative inversion. That is, the locative role is not in its usual position (at the end of the sentence) but in the initial position and serves as SUBJ (cf. Her 1996).
Most examples given till now have the agent as in the main predicate θ-structure. Of course, it can also be some other roles like the theme, experiencer, etc.
in the above example is the theme, but in the embedded θ-structure, it is the agent. The two roles form a composite role.
On other occasions, does not form a composite role with any other roles; it is the roles other than it that merge. For example,
From those examples, we get such composite roles like pt-th, loc-th, etc.
To sum up, the θ-structure of a RVC comprises two parts: the main structure and the embedded structure. It is the thematic roles assigned by Vcaus that form the main structure, and those assigned by Cres that form the embedded structure. Sometimes an entity plays a role in the main structure, and at the same time plays a role in the embedded structure. If the two roles are identical, they are “coindexed”; if they are different, they form a composite role. Not only ô, the most prominent role in the main structure, but any other roles in it can merge with a role in the embedded structure to form a composite role.
4 Θ-structures of RVCs with Ambiguity
It is observed that ambiguity exists in RVC constructions. Li (1995), for example, presents the following sentence and claims that it has three different readings.
While other researchers like Her (2007) and Pan (1998) also agree upon the existence of this kind of ambiguity, we find (28c) above difficult to accept for many native speakers: we interviewed some native speakers about their understanding of (27), they unanimously said (28a) and (28b) are acceptable, but all rejected (28c) as a justifiable reading. Hence here we take (28a) and (28b) as possible meanings of (27), excluding (28c) from our discussion.
A careful study of the θ-structure of the RVC zhui-lei would reveal that different combination of thematic roles would lead to different readings.
When the agent merges with the experiencer, we get (28a), and when the theme merges with the experiencer, we get (28b).
In addition to zhui-lei, there are other examples of ambiguous RVCs. See the sentence below:
5 Causation and Θ-structures of RVCs
It has been observed that RVC constructions are semantically composed of two events: the causing event and the caused event, hence causative relation exists between them. To present an overall picture of the θ-structures of RVCs, we have to take causation into consideration.
5.1 Representation of causation in RVC constructions
According to the standard assumptions (Alsina 1992: 521), a causative predicate would contain the functor or predicator CAUSE, which takes two arguments: a causer, which may be an individual or an event, and the caused event. The latter contains the argument structure of the predicate on which the causative is based. Under this approach, the causative predicate would be represented as:
where ag (agent) represents the causer, and the inner angled brackets indicate that the θ-structure of the caused predicate is embedded in that of the causing predicate.
Based on this assumption, we provide a modified model to represent the causation of RVCs. Since all the RVC constructions are composed of two events (not individuals): the causing event and the caused event, we substitute ag in (31) for causing event:
Here “θa, θb …” represents the θ-structure of the causing predicate, or Vaus, “θx, θy …” represents the θ-structure of the caused predicate, or Cres, and the inner angled brackets indicate “embedded.” The broken line beneath shows which two roles are borne by one single entity and form a composite role.
Let’s apply this representation to a specific compound: da-bai beat-defeated to see how it works.
5.2 Assignment of the causer and causee to thematic roles
Note that in (33) two causative roles are identified: the causer and causee, and associated with the agent ta (he) and theme duishou (opponents) respectively. The causer is the entity that causes something to happen, and the causee is the entity that is affected by the causer.
The causer may be frequently associated with the agent, it is not always the case, however. For example, in the inverted sentences like (22) (repeated in (34) below), the causer is associated with the patient rather than the agent:
The causer is yao, which is patient instead of agent; the agent, on the other hand, co-borne with the theme by ni, is associated with the causee. So we may argue that (31) is inadequate in that it identifies the causer with the agent, which is not always the case, at least for mandarin RVC constructions.
But are there any general rules for this causative-thematic association in RVC constructions? From the examples above and many others which could not be stated here because of the limited space, we can get at least two rules:
Based on this observation, we can assign the causer and causee to thematic roles by assuming the following constraints4
According to (35), one of the two components of a composite role may be assigned the causer or causee. When this occurs, the other component is automatically assigned the causer or causee since they constitute a whole body, and are borne by the same entity which could not be the causer and causee at the same time.
Let’s use these constraints to reexamine the example of (34). We assign the causer to the patient and the causee to the theme because the patient is a role of Vcaus chi eat, and the theme is a role of Cres si die; and what’s more, they do not form a composite role although the theme forms one with the agent. We do not assign the causer to the theme because the latter is of Cres; we do not assign it to the agent because the latter forms a composite role with the theme which has to be assigned the causee according to (35)a. Since the theme is assigned the causee, automatically the agent is assigned the causee too, because ag-th is a composite role.
The analysis of the θ-structures of RVCs provides us with a unique picture:
This study is a static description of the θ-structures of RVCs. But we have noticed that the same θ-structures may have different syntactic representation. How do the θ-structures, especially those of RVC inversions like (17) and (34), and of RVC constructions with ambiguity like (27), map onto syntax? Do we have a general rule to govern the mapping? These are the questions needing further exploration, although they have been partially answered.
Alsina, A. (1992): "On the argument structure of causatives". Linguistic Inquiry 23: 517–555.
Bresnan, J./Kanerva, J. (1989): "Locative inversion in Chicheŵa: a case study of factorization in grammar". Linguistic Inquiry 20,1: 1–50.
Claudia, R. (1990): "Resultative verb compounds". Journal of the Chinese Language Teachers Association 25: 61°83.
Ding Shengdhu (1961): Xiandai hanyu yufa jianghua (Speeches on modern Chinese grammar). Beijing: The Commercial Press.
Fan Xiao. (1985): "Luelun V-R (On V-R)". In: Yufa yanjiu he tansuo 3 (Research and exploration of grammar). Beijing: Beijing University Press.
Fan Xiao. (1987): "V-R jiqi suo goucheng de jushi" (V-R and the sentence patterns in which it occurs). In id. (ed.): Yuyan yanjiu ji (Anthology of language study papers). Shanghai, Fudan University Press: 230–247.
Guo Rui. (1995): "Shujieshi peijia jiegou yu chengfen zhenghe" (RVC valency structure and constituent integration). In: Shen Yang/Zheng Ding’ou (eds.): Xiandai hanyu peijia yufa yanjiu (The study of modern Chinese valency). Beijing: Beijing University Press.
Her, O. S. (1996): "Chinese inversion constructions within a simplified LMT". In: Selected papers from the second international symposium on languages in Taiwan. Taipei, Crane Publishing: 287–304.
Her, O. S. (1999): "The Lexical Mapping Theory and Mandarin Resultative Compounds". Unpublished.
Her, O. S. (2004): "Argument-function linking in resultatives". Concentric: Studies in Linguistics 30/2: 1–34.
Her, O. S. (2007): "Argument-function mismatches in Mandarin Resultatives: A lexical mapping account". Lingua 117/1: 221-246.
Huang, C. R., & Lin, F. W. (1993): "Composite event structures and complex predicates: a template-based approach to argument selection". In: L. S. Stvan et al. (eds.): FLSM III: Papers from the third Annual Meeting of the Formal Linguistics Society of Midamerica. Bloomington, IULC Publications: 90–108.
Huang Huijuan. (2008): Dongjieshi ji xiangguan jushi de jufa yuyi yanjiu (The syntactic and semantic of the VC and constructions relevant to it). MA Dissertation, North-east Normal University.
Huang Jinzhang. (1993): "Xingwei lei keneng shi V-R weiyuju luoji jiegou yu biaoceng jufa xianxiang" (The logic structure and surface syntactic representation of the sentences having as predicate “action RVCs” denoting possibilities). Yuwen Yanjiu (Linguistic Researches) 2: 57–62.
Jiang Hong. (2007): "Dongjieshi zhong buyu yuyi qizhi xianxiang fenxi" (An analysis of the multi-semantic-orientation in VCR complement). Journal of Anhui University (Philosophy & Social Sciences) 1: 31–34.
Li Linding. (1980): "Dobu ge jushi" (Sentence patterns of RVCs). Zhongguo Yuwen (Studies of the Chinese Language) 2: 75–77.
Li Linding. (1984): "Jiujing nage “bu” nage – “dongbu ge” guanxi zaiyi" (Which serves which? – reexamination of verb-complement relations). Hanyu Xuexi (Chinese Language Learning) 2: 1–10.
Li Xiaorong. (1994): "Dui dongjieshi dai binyu gongneng de kaocha" (An investigation into RVC’s capacity of taking object). Hanyu Xuexi (Chinese Language Learning) 5: 32–38.
Li, Y. F. (1995): "The thematic hierarchy and causativity". Natural Language and Linguistic Theories 13: 255–282.
Li Ziyun. (1990): "Buyu de biaoshu duixiang wenti" (The issue of the complement’s semantic reference). Zhongguo Yuwen (Studies of the Chinese Language) 5: 23–26.
Lin, H. L. (1996): "A lexical-syntactic analysis of resultative compounds in mandarin Chinese". NACCL 8,1: 189–201.
Lin, H. L. (1998): "The syntax-morphology interface of verb-complement compounds in mandarin Chinese". University of Illinois at Urbana Champaign, Ph.D. Dissertation.
Lin Yan. (2009): "Dongjieshi fuheci de hexin wenti" (On the head of Resultative Verb-compounds). Journal of Hebei North University 1: 23–27.
Lu, J. H-T. (1977): "Resultative verb compounds vs. directional verb compounds in mandarin". Journal of Chinese linguistics 5: 277–313.
Lv Shuxiang. (1980): Xiandai haiyu babai ci (800 words in modern Chinese). Beijing: The Commercial Press.
Lv Shuxiang, & Zhu Dexi. (1952): Yufa xiuci jianghua (Speeches on grammar and rhetoric). Beijing: China Youth Press.
Ma Xiwen. (1987): "Yu dongjieshi dongci youguan de mouxie jushi" (Some RVC-related sentence patterns). Zhongguo Yuwen (Studies of the Chinese Language) 6: 12–16.
Mei Lichong. (1994): "Ye tan buyu de biaoshu duixiang wenti" (Also on the issue of the complement’s semantic reference). Yuyan Jiaoxue yu Yanjiu (Language Teaching and Linguistic Studies) 2: 79–89.
Pan Haihua. (1998a): "Thematic hierarchy, causative hierarchy, and Chinese resultative verb compounds". Presented at the 1998 Annual Research Forum of the Linguistic Society of Hong Kong, the Polytechnic University of Hong Kong, 5-6 December. Retrieved 12 Nov. 2008. http://ctlhpan.cityu.edu.hk/haihuapan/pan/pan-conf-prest, accessed November 9, 2011.
Pan Haihua. (1998b): "Lunyuan jiegou shang de cihui caozuo guize yu hanyu de juzi jiegou" (The lexical operating rule on argument structures and the structures of Chinese sentences). Paper submitted to Beijing International Conference on Grammar. Beijing: Beijing University, Aug. 26–31.
Saeed, J. I. (1997): Semantics. Cambridge: Blackwell Publishers Inc.
Shi Chunhong. (2006): "Dongjieshi de peijia cengji jiqi qijia xianxiang" (The Valence Hierarchies of Verb-Resultative Constructions and the Related Issue of Multivalence). Language Teaching and Linguistic Studies 4: 45–57.
Song Wenhui. (2004a): "Zailun xiandai hanyu dongjieshi de jufa hexin" (Reexamining Heads of Chinese V-R compounds). Modern Foreign Languages 2: 163–172.
Song Wenhui. (2004b): "Buyu de yuyi zhixiang wei dongci de dongjieshi de peijia" (On the compound valence of the semantic-oriented complementation). Journal of Hebei Normal University (Social Science Edition) 3: 94–99.
Song Wenhui. (2004c): "Dongjieshi zai jige jushi zhong de fenbu" (On the location of V-R compounds in several sentence patterns). Linguistic Researches 3: 13–19.
Wierzbicka, A. (1988): The semantics of grammar. Amsterdam: J. Benjamins Pub. Co.
Wang Hongqi. (1995): "Dongjieshi shubu jiegou peijia yanjiu" (A study on RVC valency). In: Shen Yang/ Zheng Ding’ou (eds.): Xiandai hanyu peijia yufa yanjiu (The study of modern Chinese valency). Beijing: Beijing University Press.
Wang Hongqing. (1996): "Dongjieshi shubu jiegou de yuyi shi shenme" (What is the meaning of RVC?). Hanyu Xuexi (Chinese Language Learning) 1: 24–27.
Xiong Zhongru/ Liu Liping. (2005): "Hanyu dongjieshi de hexin" (The Head of the Resultative Construction in Mandarin). Journal of College of Chinese Language and Culture of Jinan University 4: 39–49.
Zhan Renfeng. (1989): "Dongjieshi duanyu de biaoshu wenti" (The semantic reference of RVCs). Zhongguo Yuwen (Studies of the Chinese Language) 2: 14–20.
Zhu Dexi. (1982): Yufa jiangyi (Lecture notes on grammar). Beijing: The Commercial Press.
* My sincere thanks go to the reviewers of LO for their insightful and constructive comments. This article is developed on the basis of my MA thesis "A study on Mandarin Resultative Verb Compounds: An LMT approach" (Chongqing University, 2003). back
1 As one of the anonymous reviewers has pointed out, in many sinoligustic articles there is no clear distinction between syntax and semantics. In this case here, the definition of RVCs is a mix of both. Even so, we still follow this tradition on the grounds that the focus of this study – thematic roles – inevitably involves both syntax and semantics. back
2 Here we provide a brief literature view that majorly focuses on the studies conducted by researchers in Mainland China, which are mostly presented in Chinese and published intra-nationally, so little known to overseas researchers. Hopefully, this will be beneficial for overseas researchers to find out what is going on in Mainland China in the field of RVC research. back
4 We are inspired by Her (1999) who also assigns the causer and causee to thematic roles but on condition that not all RVC constructions are causative. So his assignment has stricter constraints in order to rule out those constructions which he thinks bear no causative meaning. We hold that all RVC constructions are causative in that they can undergo the test of causativity proposed by Wierzbicka 1988. back