Complex Predicates in Control

  • C.-T. James Huang
Part of the Studies in Linguistics and Philosophy book series (SLAP, volume 48)


One of the most important results of early research in generative grammar is the discovery that superficially very similar constructions should be distinguished on the basis of whether they involve a construal rule of Control, or a movement rule of Raising. Thus, although the sentences in (1) are respectively very similar to those in (2) in their unanalyzed surface forms, it is generally agreed that they should be analyzed in different ways, as indicated below:
  1. (1)(a)

    John i tried [PRO i to be honest].

  2. (b)

    John persuaded Bill i [PRO i to be honest].

  3. (c)

    Bill was persuaded t i [PRO i to be honest].

  4. (d)

    Who did you persuade t i [PRO i to be honest]?

  5. (2)(a)

    John i seemed [t i to be honest].

  6. (b)

    John believed [Bill to be honest].

  7. (c)

    Bill i was believed [t i to be honest].

  8. (d)

    Who i did you believe [t i to be honest]?



Argument Structure Linguistic Inquiry Small Clause Complex Predicate Matrix Subject 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    The morpheme de is a suffix (or clitic, depending on one’s analysis) that developed historically from the full verb de meaning ‘obtain (the result of)’. It will be simply glossed as DE in the examples below.Google Scholar
  2. 2.
    Here and throughout I will not make a distinction between PRO and pro, but represent all instances of the empty pronominal as Pro. Following the proposal made in Huang (1984, 1989) and more explicitly in Borer (1989), I assume that there is only one pronominal category, Pro, and that the two empty pronominals PRO and pro postulated in Chomsky (1981) and other works are instances of a pure empty pronominal with the feature [—anaphor, +pronominal]. In Huang (1984, 1989) it is argued that the essential distributional and referential properties of both PRO and pro may be derived from a theory that subjects them to a somewhat enriched theory of control — the Generalized Control Theory — which eliminates the need for a featural distinction between them. The general line of approach is supported, not only on general a priori grounds, but also by empirical considerations. Foremost among these is the fact that PRO and pro exhibit very similar properties, both in their distribution and their reference. In particular, the occurrence of PRO is restricted to the subject position, and so is that of pro in a language that exhibits subject-non-subject asymmetries. Furthermore, both PRO and pro may be shown to be obligatorily controlled in certain positions and uncontrolled in others. For example, in many languages (e.g., Japanese, Korean, Persian) a finite clause with a pro subject may be embedded in structures that exhibit obligatory control (e.g., under ‘try’). Like the ungoverned PRO in the corresponding constructions in English, such a pro is obligatorily controlled. Another advantage of the Generalized Control Theory is that it eliminates the need to determine whether the empty subject of a resultative clause in Chinese should be a pro or a PRO. This question is a difficult one to answer since there is, as yet, no explicit standard criterion to distinguish ‘finite’ from ‘non-finite’ clauses in Chinese. The proposal not to distinguish PRO from pro is also motivated on learnability grounds.Google Scholar
  3. 3.
    This assumption is necessary also in the theory according to which the ba-phrase is brought to its surface position by movement. The case-marker ba must not increase the depth of embedding, so that the moved NP following it will c-command its trace.Google Scholar
  4. 4.
    In contrast to (20), it has been widely known that sentences like Bill is promised to be allowed to leave are well-formed. In addition, although John promised Bill to leave requires subject control, the sentence John promised Bill to be allowed to leave permits object control. Larson (1990) argues that the cases that are not predicted by Visser’s Generalization arise out of entailments of the verbs involved. Under his analysis, the sentences with infinitival passive complements are assimilated to Bill is promised permission to leave and John promised Bill permission to leave.Google Scholar
  5. 5.
    A common plausible explanation for Visser’s Generalization is that when a sentence is passivized, the agent phrase is either missing or demoted to a status that makes it incapable of control. A sentence that requires subject (or agent) control thus cannot be passivized. An object-control sentence can be passivized, however, since the trace of the passive subject remains in its D-Structure object position where it controls the Pro, in accordance with the MDP.Google Scholar
  6. 6.
    A plausible explanation for Bach’s Generalization is that in a sentence like (25) there is no object to fulfill the requirement of object control. An empty object controller is out of the question here since English does not allow a pro in object position. On the other hand, it is well known that Chinese permits null objects. Given this latter fact, (29) might be expected to be grammatical, with the embedded Pro controlled by a matrix Pro object, contrary to fact. However, as I have argued in (1984, 1989), a null object in Chinese should be more properly analyzed as a null topic. Furthermore, there is independent evidence that a topic is incapable of controlling a resultative clause subject: (i) Lisi, Zhangsan ku-de [Pro hen shangxin]. Lisi, Zhangsan cry-DE very sad (As for) Lisi, Zhangsan cried until he (Zhangsan, not Lisi) became very sad. Thus, even if (29) is used in a discourse in which a null object is permitted, that null object (as a null topic) cannot control the Pro in (29). The ungrammaticality of (29) thus provides further evidence for the null-topic analysis of null objects proposed in Huang (1984).Google Scholar
  7. 7.
    An alternative is to assume that the verb moves into the head of a functional category dominating VP (such as the functional category PrP (Predicate Phrase) proposed in Bowers (1989)). Still another possibility is for the verb to raise to I0, as is assumed in Chomsky (1986b), though in light of Emonds (1978) and Pollock (1989) this possibility is excluded at least for English.Google Scholar
  8. 8.
    For a study of how the argument structure of a V-V compound may be built up from the argument structure of each component V, see Li (1990). My contention is that generalizations about the argument structure of V0 categories also obtain with complex predicates that the V′-categories. Superficial differences between compound and complex predicates are largely attributable to the principle of lexical integrity.Google Scholar
  9. 9.
    This is probably due in part to the fact that the element ba is historically derived from a full verb that has the meaning of ‘take x and do something onto x’, or ‘to dispose of x’. The ba-construction is often known as the ‘disposal construction’.Google Scholar
  10. 10.
    Since non-ba forms like (39) may have, in addition to the meaning according to which Lisi is an affected Theme, also the meaning of a mere description of the extent to which Zhangsan cried, a representation in which Lisi directly appears as the subject of the result clause is also possible. This is irrelevant to our discussion here. My contention is that in the ba forms the ba phrase cannot be related to the result clause subject position by a raising relation, but only by a control relation.Google Scholar
  11. 11.
    An alternative to talking about this process as one of reanalysis is to consider it an instance of verb-incorporation or V-movement (Baker (1988)), by which the second verb moves to the first. A movement analysis does not explain why the two verbs must retain the same precedence relation as they do before movement takes place, i.e., why the movement must be string-vacuous. On the other hand, string vacuity directly follows if a reanalysis rule of V → V0 is assumed.Google Scholar
  12. 12.
    Sentences like (51) and (52) have been reported to be grammatical in the dialect of older Northern Mandarin speakers, and in the literature representing the historical period of Pre-Modern Chinese (Ming-Ching dynasties). (See Li (1963) and Zhu (1982) [1965].) As Peyraube (1989) has shown, such examples represent a historical period where the Phrase Structure Constraint (alluded to above; see the discussion in connection with (16)–(17)) was not yet a property of Chinese. That is, in the historical dialect of Chinese where (51)–(52) are grammatical, such sentences may have a structure in which the verb is followed by an object (‘rice’ in (51)), and a higher-order result clause (‘very full’) minimally c-commanded by the matrix subject but not by the postverbal object — a structure specifically ruled out by the Phrase Structure Constraint of Modern Mandarin.Google Scholar
  13. 13.
    Note that, although the Patient argument in a causative sentence like (53) may receive Case under any one of several ways (as Oblique under ba-insertion, as Accusative under V-movement, or as Nominative under passivization), the Patient argument in an unaccusative sentence like (55) (= (61)) can receive Case only in one way, as Nominative under NP-movement to the SPEC of IP position. This is true because a subjectless sentence in which the Patient follows the verb or the element ba must be interpreted as a causative sentence whose subject is omitted in discourse (a pro, but not an expletive subject). This situation is not unexpected: Case assignment by V under V-movement is excluded because, as the head of an unaccusative predicate, V is intransitive in the traditional sense, hence not a Case-assigner. The option of ba-insertion might be excluded under some appropriate version of Buzio’s Generalization. A similar case in English has to be excluded also: *It/there broke of the window.Google Scholar
  14. 14.
    The following sentence is ambiguous, exhibiting both transitivization and causativization (from Zhu (1982)): (i) xiaohai zui-de wo [Pro zhi chuanqi]. child chase-DE I straight pant (a) The child chased me until I kept panting. (b) The child got me to chase (him) until I kept panting. The reading (a) is the transitive reading, and (b) the causative reading. This ambiguity arises from the possibility of interpreting (ii) as a pure intransitive or an unaccusative: (ii) NP zui-de [Pro hen shangxin]. chase-DE very sad That is, the NP in (ii) may be understood as Agent or as Patient/Experiencer (as someone made to chase another). If NP is Agent, then the addition of a Theme argument would turn the sentence into the transitive (ia). If NP is Patient/Experiencer, then the addition of an external Causer argument would turn it into the causative (ib).Google Scholar
  15. 15.
    The relevant binding facts can also follow from the small-clause or raising analysis, given familiar assumptions of CP-deletion and ECM. However, the ECM account does not lend itself to an explanation of similar binding facts observed with complex V-O constructions discussed below. See the discussion around (83)–(84) and note 21.Google Scholar
  16. 16.
    The expression is more normally used in an exaggerative way of expressing one’s disbelief, etc. For example, the following may be said of a miser: (i) yaoshi ta gei ni qian, lian tieshu dou hui kai-hua le. if he give you money, even iron-tree all will blossom ASP ‘If he gives you any money, even the iron tree will blossom.’ This is again a literal use of the expression. An exaggeration is not an idiom.Google Scholar
  17. 17.
    The representation of (70) may be as follows: (i) Zhangsani xia-de [Proi [huli lu-chu-le yiba]]. where Zhangsan controls a Pro subject or topic, to which the expression huli lu-chu-le yiba is a clausal predicate or a comment. That this representation is possible is evidenced by the availability of independent sentences like (ii): (ii) tarnen huli lu-chu yiba le. they fox-reveal tail ASP ‘The property of a fox revealing its tail because true of them.’ That the representation (i) is necessary, and not one in which the huli ‘fox’ is directly represented as the subject of the result clause coindexed with Zhangsan, follows from binding theory. This coindexing would be ruled out by Principle C. In fact, as we just saw in the text, even a proximate overt pronoun is excluded from the subject position of the clause.Google Scholar
  18. 18.
    The ba- sentence below refers to someone referentially established as a sly person who was frightened by Zhangsan to such an extent as to reveal his own secrets: (i) Zhangsan ba huli xia-de lu-chu-le yiba. Zhangsan BA fox frighten-DE reveal-ASP tail Lit. ‘Zhangsan got the fox so frightened as to reveal its tail.’ On this meaning (i) is on a par with (71) and (72) in the text, where huli is a fully referential NP that is capable of control.Google Scholar
  19. 19.
    This analysis is in the spirit of Thompson (1973).Google Scholar
  20. 20.
    Reanalysis may be assumed to occur also with the resultative constructions after verb-raising. For example, one might assume that the output of V-movement given in (36) above may be reanalyzed so that Lisi does form a clause with hen shangxin ‘very sad’, but this may be assumed to occur only at a superficial level, in a way inconsequential to the thematic interpretation of the sentence.Google Scholar
  21. 21.
    One might attempt to derive the binding facts here from the small-clause or raising analysis by postulating a nominal small clause of some kind. Suppose that the object NP in (83) and (84), or his leg in John pulled his leg, undergoes NP-deletion (on a par with CP-deletion), resulting in ECM. In this case the governing category for the possessive pronoun would be the root clause, and disjointness of reference follows. But note that the pronoun receives Genitive Case in these cases. This means that it is not exceptionally Case-marked (or governed) from outside the NP. Therefore, the small-clause or raising analysis really does not have a good way to account for (83)–(84).Google Scholar
  22. 22.
    In contrast to (86)–(88), the following sentences are well-formed: (i) John considers there to be a smart person here. (ii) I consider advantage to have been taken of John. (iii) I consider tabs to have been kept on Jane Fonda. These sentences are genuine raising/ECM constructions, however, and not complex predicate constructions. The expletive and idiom chunks in these sentences occur in the non-thematic subject position of IP. Incidentally, extraposition it may occur in a theta-position since it is coindexed with an antecedent with a thematic role: (iv) John considered it obvious that Bill was stupid. Examples like (iv) cannot be taken as showing that the object position of a complex predicate can be a non-theta position.Google Scholar


  1. Aoun, Joseph and Y.-H. Audrey Li: 1988, ‘Minimal disjointness’, Proceedings of the 7th West Coast Conference on Formal Linguistics, Department of Linguistics, Stanford University, pp. 29-40.Google Scholar
  2. Baker, Mark: 1987, Incorporation, University of Chicago Press, Chicago.Google Scholar
  3. Bach, Emmon: 1979, ‘Control in Montague Grammar’, Linguistic Inquiry 10, 515–531.Google Scholar
  4. Borer, Hagit: 1989, ‘Anaphoric Agr’, in Osvaldo Jaeggli and Ken Safir (eds.), The Null Subject Parameter, Kluwer Academic Publishers, Dordrecht, pp. 69–110.CrossRefGoogle Scholar
  5. Bowers, John: 1989, ‘The Syntax and Semantics of Predication’, ms., Cornell University.Google Scholar
  6. Burzio, Luigi: 1986, Italian Syntax, Academic Publishers, Dordrecht.CrossRefGoogle Scholar
  7. Chao, Yuen-Ren: 1968, A Grammar of Spoken Chinese, University of California Press, Berkeley.Google Scholar
  8. Cheng, Lisa L.-S.: 1986, ‘Clause Structures in Mandarin Chinese’, MA thesis, University of Toronto.Google Scholar
  9. Chomsky, Noam: 1955, [1975] Logical Structure of Linguistic Theory, Plenum, New York.Google Scholar
  10. Chomsky, Noam: 1985, Aspects of the Theory of Syntax, MIT Press, Cambridge, Massachusetts.Google Scholar
  11. Chomsky, Noam: 1980, ‘On Binding’, Linguistic Inquiry 11, 1–46.Google Scholar
  12. Chomsky, Noam: 1981, Lectures on Government and Binding, Foris, Dordrecht.Google Scholar
  13. Chomsky, Noam: 1982, Some Concepts and Consequences of the Theory of Government and Binding, MIT Press, Cambridge, Massachusetts.Google Scholar
  14. Chomsky, Noam: 1986a, Knowledge of Language, Praeger, New York.Google Scholar
  15. Chomsky, Noam: 1986b, Barriers, MIT Press.Google Scholar
  16. Emonds, Joseph: 1972, ‘Evidence that Indirect Object Movement is a Structure-Preserving Transformation’, Foundations of Language 546-561.Google Scholar
  17. Emonds, Joseph: 1976, A Transformational Approach to English Syntax, Academic Press, New York.Google Scholar
  18. Emonds, Joseph: 1978, ‘The Verbal Complex V’-V in French’, Linguistic Inquiry 9, 151–175.Google Scholar
  19. Goodall, Grant: 1987, ‘On Argument Structure and L-marking with Mandarin ba’, NELS 17, 232–243.Google Scholar
  20. Goodall, Grant: 1989, ‘Evidence for an Asymmetry in Argument Structure’, Linguistic Inquiry 20, 669–674.Google Scholar
  21. Hashemipour, Peggy: 1988, ‘Finite Control in Modern Persian’, Proceedings of the 7th West Coast Conference on Formal Linguistics, Department of Linguistics, Stanford University, pp. 115-128.Google Scholar
  22. Higgins, Roger: 1974, ‘On the Use of Idioms as Evidence for Movement: A Cautionary Note’, paper presented at the LSA annual meeting; ms., University of Massachusetts, Amherst.Google Scholar
  23. Huang, C.-T. James: 1977, ‘Against Particle Movement’, paper presented at the LSA annual meeting, Chicago.Google Scholar
  24. Huang, C.-T. James: 1982, ‘Logical Relations in Chinese and the Theory of Grammar’, MIT PhD dissertation.Google Scholar
  25. Huang, C.-T. James: 1984, ‘On the Distribution and Reference of Empty Pronouns’, Linguistic Inquiry 15, 531–574.Google Scholar
  26. Huang, C.-T. James: 1988, ‘wo pao de kuai and Chinese Phrase Structure’, Language 64, 274–311.CrossRefGoogle Scholar
  27. Huang, C.-T. James: 1989, ‘Pro Drop in Chinese: A Generalized Control Theory’, in Osvaldo Jaeggli and Ken Safir (eds.), The Null Subject Parameter, Kluwer Academic Publishers, Dordrecht.Google Scholar
  28. Huang, Shuanfan: 1984, ‘Morphology as a Cause of Syntactic Change: The Chinese Evidence’, Journal of Chinese Linguistics 12, 54–85.Google Scholar
  29. Jackendoff, Ray: 1977, X’ Syntax, MIT Press, Cambridge, Massachusetts.Google Scholar
  30. Koopman, Hilda: 1984, The Syntax of Verbs, Foris, Dordrecht.Google Scholar
  31. Ladusaw, William and David Dowty: 1988, ‘Toward a Non-Grammatical Account of Thematic Roles’, in Wendy Wilkins (ed.), Syntax and Semantics 21: Thematic Relations, Academic Press, New York, pp. 61–73.Google Scholar
  32. Larson, Richard: 1988, ‘Light Predicate Raising’, Lexicon Project Working Papers 27, Center for Cognitive Sciences, MIT.Google Scholar
  33. Larson, Richard: 1990, ‘Promise and the Theory of Control’, Linguistic Inquiry 22, 103–140.Google Scholar
  34. Li, Charles N.: 1975, ’synchrony and Diachrony in Language Structure’, Language 51, 873–886.CrossRefGoogle Scholar
  35. Li, Linding: 1963, ‘Dai de zi de buyu ju [sentences with de-complements]’, Zhongguo Yuwen, 396-410.Google Scholar
  36. Li, Yafei: 1990, ‘On V-V Compounds in Chinese’, Natural Language and Linguistic Theory 8, 177–208.CrossRefGoogle Scholar
  37. Li. Y.-H. Audrey: 1985, ‘Abstract Case in Chinese’, USC PhD dissertation.Google Scholar
  38. Mei, Kuang: 1978, ‘Guoyu yufa zhong de dongcizhu buyu [Verb phrase complements in Chinese grammar’, in A Festschrift for Professor Chu Wan-Li, Lianjing, Taipei.Google Scholar
  39. Montalbetti, Mario: 1984, ‘After Binding’, MIT PhD dissertation.Google Scholar
  40. Pollock, Jean-Yves: 1989, ‘Verb movement, Universal Grammar, and the Structure of IP’, Linguistic Inquiry 20, 365–424.Google Scholar
  41. Peyraube, Alain: 1989, ‘zaoqi baziju de jige wenti [Some Problems of Early ba-Sentences]’, Yuwen Yanjiu 30, 1–9.Google Scholar
  42. Reuland, Eric: 1983, ‘Governinging’, Linguistic Inquiry 14, 101–136.Google Scholar
  43. Rosenbaum, Peter S.: 1970, ‘A Principle Governing Deletion in English Sentential Complementation’, in Roderick Jacobs and Peter S. Rosenbaum (eds.), Readings in English Transformational Grammar, Ginn and Company, Waltham, Massachusetts.Google Scholar
  44. Saito, Mamoru and Hajime Hoji: 1983, ‘Weak Crossover and Move a in Japanese’, Natural Language and Linguistic Theory 1, 245–260.CrossRefGoogle Scholar
  45. Stowell, Timothy: 1981, ‘Origins of Phrase Structure’, MIT PhD dissertation.Google Scholar
  46. Tang, Ting-chi: 1977, Guoyu Bianxing Yufa [Chinese Transformational Grammar], Student Book Company, Taipei.Google Scholar
  47. Teng, Shou-Hsin: 1975, A Semantic Study of Transitivity Relations in Chinese, University of California Press, Berkeley.Google Scholar
  48. Thompson, Sandra A.: 1973, ‘Transitivity and Some Problems with the ba-Construction in Mandarin Chinese’, Journal of Chinese Linguistic 1, 208–221.Google Scholar
  49. Visser, F. T.: 1973, An Historical Syntax of the English Language, Brill, Leiden.Google Scholar
  50. Wang, Li: 1958, Zhongguo Yufa Lilun [A Theory of Chinese Grammar], Zhonghua Press, Beijing.Google Scholar
  51. Wang, Peter C.-T.: 1970, ‘A Transformational Approach to Chinese Ba and Bei’, University of Texas PhD dissertation.Google Scholar
  52. Williams, Edwin: 1980, ‘Predication’, Linguistic Inquiry 11, 203–238.Google Scholar
  53. Williams, Edwin: 1983, ‘Against Small Clauses’, Linguistic Inquiry 14, 287–308.Google Scholar
  54. Zhu, Dexi: 1982, [1965] Yufa Jiangyi [Lectures on Grammar], Commercial Press. Beijing.Google Scholar

Copyright information

© Springer Science+Business Media New York 1992

Authors and Affiliations

  • C.-T. James Huang

There are no affiliations available

Personalised recommendations