Cleft and pseudo-cleft constructions in English

ELSEVIER Journal of Pragmatics 26 (1996) 699-710 Book reviews Peter C. Collins, Cleft and pseudo-cleft constructions in English. London: Routledge, 1991, xvi + 230 pp. ISBN 0-415-06328-0, £35.00. Reviewed by Andreas H. Jucker, FB 10, Anglistik, Justus Liebig University, Giessen, Otto-Behaghel-Str. 10, D-35394 Giessen, Germany Cleft and pseudo-cleft constructions in English have always been a testing ground for linguistic frameworks: generative in the sixties and early seventies, presuppositional in the seventies and early eighties, functional or discourse analytical in the eighties, and cognitive pragmatic in the nineties. Collins offers a functional/systemic account in the Hallidayan tradition. His analysis is unique, however, in that it is based on a large corpus of actual examples drawn from the London-Lund corpus of spoken English and the Lancaster-Oslo-Bergen (LOB) corpus of written English. "As well as providing information on frequencies of occurrence across a range of genres, the corpusbased approach has the salutary effect of requiring that attention be paid to "untidy" data of a type often overlooked, or ignored, in studies based on introspectivelyderived examples." (p. 1) What Collins refers to as "untidy data" are constructions that share certain properties with cleft or pseudo-cleft constructions but which do not have their prototypical form and are therefore usually neglected by other researchers. His aim is no less than a "comprehensive approach" to the study of clefts and pseudo-clefts. He deals in some detail with prosodic, syntactic, semantic, and pragmatic aspects of clefts and pseudo-clefts, but he devotes the bulk of his attention to functional and textual aspects. He argues that the different cleft and pseudo-cleft constructions all have unique communicative values, which are based on their logico-semantic, thematic, and informational properties. 1785 actual examples drawn from the LOB and the London-Lund corpus provide the empirical evidence for his claims. In an analysis that relies heavily on statistical data, it is crucial to give precise and explicit definitions of the constructions under analysis. Collins provides these definitions in chapter 3. In most cases he adopts wider definitions than traditional treatments of clefts and pseudo-clefts. Thus he includes not only constructions such as (1) but also constructions such as (2) and (3) among the pseudo-cleft constructions. (1) What the car needs is a new battery. (2) The thing the car needs is a new battery. (3) All the car needs is a new battery. 0378-2166/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved, 700 Book reviews /Journal of Pragmatics 26 (1996) 699-710 These examples are called wh-clefts, th-clefts and all-clefts, respectively. The lexically-headed th-clefts - such as example (2) - rely on the Hallidayan distinction between pro-nouns (thing, one, place, time, reason, way, and so on), and general nouns (thing, person, man, place, kind, and so on), where thing and place appear in both lists (p. 29) because they share properties with both types. Example (4) is a pseudo-cleft, according to Collins, but (5) and (6) are not. (4) Frank Morgan was the one who started all this. (5) Frank Morgan was the man who started all this. (6) Frank Morgan was the person who started all this. Collins also includes an unusually wide range of constructions in the class of clefts. Example (7) illustrates a prototypical cleft construction, while (8) and (9) are more controversial. (7) It was a sherry that Tom offered Sue. (p. l, ex. (lc)) (8) is it that you you're not in a different post there? (9) it may well have been that we'd have been able to help her to see the damaged bit of her ... (p. 35, ex. (28) and (29), simplified) Many researchers would probably be happier to call (8) and (9) extrapositions rather than clefts. In these cases there is no "highlighted element with experiential function" (p. 34). What is highlighted here is tense, modality, aspect, and polarity. I do not want to take issue with individual decisions for inclusion or exclusion of specific constructions as clefts. Casting a wide net has the advantage of providing a rich collection of very similar constructions that highlight each other's distinct properties, but it has the disadvantage of making it difficult to compare Collins' results with similar results from other corpora. The very specific definitions given by Collins seem to have precluded any kind of automatic data searching. Even prototypical cleft constructions are hard to find with any of the available text retrieval programs. No element in a cleft construction is unique to clefts; on the contrary, all the characteristic elements are extremely common in the English language (it, is or are and that, who or which in clefts; and what and is or are in pseudo-clefts). It is only the specific combination in which they occur that makes them cleft constructions. Collins' decision to include lexically headed clefts increases these problems considerably. He used a computer to highlight all elements in the corpus that are characteristic of clefts, that is to say, all forms of the verb 'to be', the interrogatives (what, who, etc.), their pro-forms (thing, one, and so on), it and all (p. 25). After that it was still a matter of hand-searching the entire corpus. Collins argues that thematic and informational considerations are paramount for an analysis of cleft constructions, and he offers one of the most precise and detailed taxonomies. He distinguishes between several types of givenness and newness, and classifies the different cleft constructions according to the distribution of these types of old and new information in the highlighted element and in the relative clause, respectively. Book reviews /Journal of Pragmatics 26 (1996) 699-710 701 Clefts are "newness-oriented" and therefore require a delicate classification of newness ("fresh newness" versus "contrastive newness", etc.). In the unmarked type 1 clefts, the highlighted element is new and the relative clause given or inferable. In the marked type 2 clefts it is the relative clause that contains new information while the highlighted element contains given or inferable information. The marked type 3 clefts contain new information in both elements. Clefts with given information in both parts are not attested in Collins' corpus. Basic pseudo-clefts are "givenness-oriented". They are concerned with backgrounded information and therefore need a correspondingly delicate subclassification of types of givenness. Reversed pseudo-clefts typically present the main point of newness in the comment, while the head is normally text-anaphoric and thus prototypically given information as in (p. 10). (10) that' s how it operates (p. 59, ex. (21.a)) In chapter 7, Collins discusses the distribution of the various types of clefts and pseudo-clefts in the different genres of his vast corpus. To my mind, this is the most exciting part of the entire book. He can show that the varying textual functions of these constructions make them more or less suitable for individual genres or text types. In the spoken English of the London-Lund corpus, pseudo-clefts are more frequent than clefts. Both the basic pseudo-clefts and the reversed pseudo-clefts are more frequent than the clefts. In the written English of the LOB corpus, on the other hand, the situation is reversed. The clefts are now even more frequent than the basic and the reversed pseudo-clefts taken together. Clefts are a means of directing the reader's interpretation. The specific information structure of clefts assigns the theme a higher information value than it would have in an unmarked non-cleft construction. Thus they can be used in the written language to fulfill similar functions as stress and intonation in the spoken language. The basic pseudo-cleft specifies in the subject relative clause what the addressee should be prepared to accept as non-controversially recoverable. "Basic pseudoclefts thus offer the speaker a means of specifying precisely, before the announcement of the 'message', the background knowledge to which the addressee is expected to have access" (p. 181). They are therefore particularly appropriate in the spoken language. They function as indices that point both backwards (the relative clause is presented as given and/or presupposed), and forward (the relative clause is also the theme of the basic pseudo-cleft). In reversed pseudo-clefts, the subject is often a text-anaphoric element (usually that), while the relative clause presents again given or presupposed information. Reversed pseudo-clefts, therefore, have only a very minimal level of informativity. They are used to summarize stretches of previous discourse, which is again something that is particularly suitable in spoken language (e.g. that's what happened, that's what he said (p. 182)). On a more detailed level, Collins can show, for instance, that the ratio of clefts to pseudo-clefts differs significantly between dialogic and monologic samples in the 702 Book reviews /Journal of Pragmatics 26 (1996) 699-710 London-Lund corpus, and that in the informative categories of the LOB corpus the frequency of clefts varies along a scale from factual descriptive registers such as documents and reports and press reportage (with very few clefts) to the rhetorical persuasive registers such as press reviews, editorials, and religion (with many clefts). In conclusion, I want to stress that this book is a major contribution to the ongoing discussion of cleft constructions. One of its outstanding features is the fact that it is rigorously empirical. It is based on a close analysis of a large corpus of actual examples. Not surprisingly, it turns out that the corpus reality of cleft constructions is considerably messier than introspectionists would like to have it. K n n d L a m b r e c h t , Information structure and sentence form: Topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press, 1994. xvi + 388 pp. $54.95. Reviewed by Yael Ziv, Department of English, The Hebrew University, Mt. Scopus, Jerusalem, Israel 91905. E-mail: [email protected] The nature of the relation between form and function in language is the major concern of Lambrecht's book. Radical iconicity (as in Bolinger's, 1985, view), transparent segmentation of sentences into informationally functional sub-components, as well as the establishment of a direct causal relationship between functional considerations and formal realizations, are very convincingly rejected in favor of a more sophisticated view of the factors which are relevant in motivating sentence form. Rejected is also the view implicit elsewhere (e.g. Sperber and Wilson, 1986, and Vallduvf, 1990) that information amounts to a mere increase in the knowldge store of the addressee. This overly simplistic conception, which gives rise to erroneous analyses of information chunks containing 'given' material as displaying redundancies and hence as not being informationally relevant, is replaced by a more realistic model of information structure, whereby for 'new' information to be integrated into the 'knowldge store' of the addressee it must necessarily utilize given information. There is no consequent redundancy or irrelevance; rather, the coherent flow of information, according to Lambrecht, requires such grounding in givenness. I In the same spirit, various restrictive relative clauses are shown to be effective in the referent identification function, precisely because they contain familiar information. Anchoring to that which is known (to use Prince's, 1981, term) makes the task of referent determination easier, since there is no need to process and integrate new material for this purpose. Lambrecht's approach explicitly dispels yet another myth concerning information: the philosophically slanted view whereby information is concerned with truth. In a framework where information is not co-extensive with See also Ziv (1996), where certain linearization principles concerning sentence-initial position in Modem Hebrew are shown to be insightfullydescribable in terms of anchoring or grounding in linguistic or situational givenness.