The Structure of Small Clause Predication
Total Page:16
File Type:pdf, Size:1020Kb
173 Azusa Yokogoshi The Structure of Small Clause Predication The Textual Functions of Their Heads Azusa YOKOGOSHI 1. Introduction The structure of small clauses, which are exemplified in the italicized portions of (1), has been one of the most controversial topics in generative grammar. (1) a. We consider Mary honest. b. We want Mary happy. Various analyses have been proposed for the structure of small clauses. It is widely assumed that small clauses have the same categorial status and internal architecture regardless of the matrix verbs selecting them. However, as recently discussed by Contreras (1995), Baker (1997), and Basilico (2003) among others, there is good reason to argue that the structure of small clauses should not be treated uniformly. However, their analyses differ considerably as to the classification of small clauses. Apparently, the matrix verbs in (1) seem to select small clauses with the same structure. As will be shown, however, small clauses selected by consider and want differ in their syntactic and semantic properties and therefore should be treated as having different structures. The verbs which behave like consider and want are given in (2a,b), respectively. (2) a. believe, consider, find, imagine, judge, perceive, prove, regard, suspect, take, ... b. expect, fear, hate, like, love, need, want, ... (cf. Declerck (1991)) The selection of small clauses of the matrix verbs is textual in the sense that it is determined not only by syntactic but also semantic factors.1 It will be argued that the 1 The concept of “text” would not only include discourse but also some other factors. Readers refer to Halliday (1978) as to the notion of text, although I would not strictly follow the framework of Halliday’s systemic functional grammar here. HERSETEC, 1, 1 (2007), 173–189 174 Azusa Yokogoshi categories of the small clauses in (1) are both Pred[ication] Phrases (PredPs) along the lines of Bowers (1993), but they involve different types of predicates: the small clause in (1a) involves an individual-level predicate (henceforth, ILP), whereas the one in (1b) a stage-level predicate (henceforth, SLP). In other words, there are two types of PredPs in English small clauses. There seems to be a kind of texture between the selecting verb and the predicate in the selected small clauses. The purpose of this paper is to explain the properties of small clauses and explore the textual relationships between small clauses and the matrix verbs selecting them. Our concern is mainly on syntactic and semantic texts surrounding small clauses, which are necessary in determining the types of PredPs. We also focus on the textual properties of the functional category Pred as the introducer of a subject-predicate relationship within small clauses. The organization of this paper is as follows. Section 2 overviews some properties of small clauses and examines several previous analyses of the structure of small clauses. Section 3 proposes different structures for small clause complements of consider-type verbs and want-type verbs. Section 4 extends the analysis proposed in section 3 to small clause complements of perception verbs. Section 5 is a conclusion. 2. Properties of Small Clauses and Previous Studies 2.1. Some Properties Let us begin with reviewing some properties of small clauses in English. As shown in (3) and (4), verbs like consider and want take various categories as the predicative phrases in their small clause complements. (3) a. I consider Mary intelligent. (AP) b. I consider Mary John’s best friend. (NP) c. I consider your son grown up. (VP) d. * I consider John off my ship. (PP) (Stowell (1981:259)) (4) a. * I want Mary intelligent. (AP) b. * I want Mary my best friend. (NP) c. I want this issue solved immediately. (VP) d. I want this sailor off my ship. (PP) Stowell (1981) argues that the categories of predicative phrases within small clauses depend on their matrix verbs, and therefore small clauses are maximal projections of their predicative phrases. As Kitagawa (1985) points out, however, that the categories of predicative phrases within small clauses depend on their matrix verbs is not correct. The examples in (5) and (6) show that the categories of predicative phrases are not relevant for the grammaticality of small clauses. The Structure of Small Clause Predication 175 (5) a. I consider Mary {honest/intelligent/*happy/*angry}. b. Unfortunately, our pilot considers that island off the route. (Kitagawa (1985:212)) c. * I consider John off my ship. (Stowell (1981:259)) d. We consider Mary our best friend. (6) a. We want Mary {happy/angry/*honest/*intelligent}. b. * I expect that island off the route. (Kitagawa (1985:212)) c. I expect that sailor off my ship by midnight. (ibid.:211) d. * We want Mary our best friend. It is indeed the case that properties of small clauses are determined by the matrix verbs selecting them, but what is important for the grammaticality of small clauses is not the categories, but the semantic types of their predicative phrases. Kitagawa claims that consider-type verbs s-select small clauses which express ‘state of affairs,’ while want-type verbs s-select small clauses which express ‘change of state.’ This would correctly capture the restrictions on the predicate types of small clauses observed in (5) and (6). However, Kitagawa’s analysis cannot account for the ungrammaticality of (7), where the small clause clearly denotes ‘change of state.’ (7) * I expect you an attorney by the end of the year. (cf. Contreras (1987:230)) On the other hand, Svenonius (1994) and Basilico (1997) argue that consider-type verbs select small clauses which involve ILPs, whereas want-type verbs select small clauses which involve SLPs.2 Various properties have been observed with respect to the two types of predicates since Carlson (1977), but the most fundamental difference is that SLPs denote temporary states/transitory relations, whereas ILPs denote more permanent properties. The examples in (8) and (9) illustrate the SLP/ILP distinction between small clause complements of consider-type verbs and want-type verbs. (8) a. The republics consider Zhirinovsky a threat. (Svenonius (1994:91)) b. We find him unbearable. (ibid.) c. They proved the allegations false. (ibid.) (9) a. Zhirinovsky wants reformers out of the parliament. (ibid.) b. We fear the rescue party lost in the mountains. (ibid.) c. She likes her eggs over-easy. (ibid.) As is obvious, the grammaticality of (3) – (7) can also be accounted for in terms of the 2 As pointed out by Heycock (1995) and Diesing (1992), the predicative phrase sick can be interpreted either as an ILP or an SLP. (i)? We consider Mary sick. This sentence turns out to be grammatical if sick is interpreted as an ILP meaning ‘insane’ or ‘crazy.’ It is also observed by Basilico (1997) that SLPs can express properties and behave like ILPs. The following example shows that visible can behave both as an SLP and an ILP. (ii) The visible stars are not visible tonight because it is cloudy. (Basilico (1997:288)) It seems reasonable to assume that the SLP/ILP interpretation is not determined solely by the predicate itself. But I will not pursue this matter further here. 176 Azusa Yokogoshi SLP/ILP distinction; especially, (7), which is problematic for Kitagawa’s (1985) analysis, is ungrammatical, simply because NP can only be an ILP. In addition to this semantic difference between small clauses selected by consider-type verbs and want-type verbs, there are some syntactic differences between the two types of small clauses. First, there is a difference in Condition C effects as illustrated in (10): small clause subjects of consider-type verbs cannot be coreferential with R-expressions contained in the matrix clause, whereas those of want-type verbs can. Second, (11) show that small clause subjects of consider-type verbs can be passivized, whereas those of want-type verbs cannot. Third, as the contrast between (12a) and (12b) shows, consider-type verbs do not allow topicalization of their small clause complements, while want-type verbs do.3 (10) Condition C effect a. * I consider him honest more cordially than John’s mother does. b. I wanted him dead more cordially than John’s mother did. (11) passivization a. John was considered smart. b. * John was wanted happy. (12) topicalization a. * The allegations false, they proved. (Svenonius (1994:93)) b. Dogs in the house, they hate. (ibid.:92) 2.2. Previous Analyses and their Problems There are a number of proposals which attempt to analyze properties of small clauses in terms of the assumption that the structure of small clauses is uniform regardless of the matrix verbs selecting them. For example, Contreras (1987) analyze small clauses as projections of their predicative phrases. Another line of approach pursued by Bowers (1993), and Starke (1995) among others, assumes that small clauses are headed by some functional category. The two kinds of approaches are roughly illustrated in (13) and (14), respectively. 3 As shown in (i) and (ii), the same is true of other movement operations like focalization, which are allowed only for small clause complements of want-type verbs. (i) it-cleft a. *It was Leslie in complete control of the situation that we believed. (Park (1997: 259)) b. It was Leslie in complete control of the situation that we feared most. (ibid.) (ii) pseudocleft a. *What I really consider is Mary loyal to her friend. (Svenonius (1994:93)) b. What I really want is that man off my ship. (ibid.:92) The Structure of Small Clause Predication 177 (13) Lexical Approach (14) Functional Approach Although such unitary approaches to small clauses might be preferred on the ground of theory-restrictiveness, it was observed above that small clause complements of consider-type verbs and want-type verbs show different properties, which cannot be easily accounted for under the assumption that they have the same structure.