Book review

Constructions at work

Goldberg, Adele E. 2006. Constructions at work. The nature of generalization in language. Oxford: Oxford University Press

Reviewed by Martin Hilpert, Department of Linguistics, Rice University (current affiliation as of 2007-2008: ICSI, U.C. Berkeley)

Constructions at work, as a sequel to the highly influential Constructions (1995), is Adele Goldberg's book-length answer to the question: So, what's been going on in your lab lately? Besides presenting a wide range of empirical studies, the book also reports on the current state of Construction Grammar as a syntactic theory, and discusses the relation of Construction Grammar to other cognitive theories of syntax.

Constructions at work consists of three parts. The first part introduces the reader to the basic notions of Construction Grammar. The second part discusses how generalizations are acquired and constrained. The third part focuses on particular syntactic phenomena to explain the role of constructions in linguistic generalizations.

Part I: Constructions

The first chapter lays out the notion construction, which is defined more broadly than in Goldberg's earlier publications. As before, constructions are viewed as linguistic patterns for which some aspect of either form or function is not strictly predictable from its component parts. In addition, even fully predictable patterns are now called constructions, provided that they are frequent enough to become entrenched as units. This re-definition gives a nod to usage-based models of grammar (Barlow and Kemmer 2000, Bybee and Hopper 2001), in which even regular patterns are redundantly stored. Goldberg emphasizes that knowledge of language is knowledge of constructions all the way down: from abstract argument-structure constructions to individual morphemes.

Chapter 2 contrasts Construction Grammar with derivational theories of syntax. In favor of the constructional approach, Goldberg argues that linguistic generalizations can always be stated more broadly about surface forms than about the same surface form and a supposed underlying form. As a case in point, the ditransitive construction has a benefactive reading in Mina bought Mel a book and a transfer reading in Mina sent Mel a book. Accounts such as Baker (1988) derive these two sentences from distinct input expressions (the for-Dative and to-Dative respectively), thereby failing to explain why both transfer and benefactive ditransitives exhibit the exact same syntactic and semantic behavior (both disallow the questioning of the recipient, both disallow pronominalization of the theme, both require animate recipients, etc.).

Chapter 3 reviews the evidence for a usage-based model of grammar, which allows for storage of item-specific knowledge alongside with generalizations. While grammars exhibit great productivity, they are riddled with idiosyncrasies, which can only be explained through a usage-based account. Phonological evidence comes from the fact that discourse frequency correlates with reduction, such that the word memory is often pronounced with two syllables whereas its quasi-homophone mammary is not (Bybee 2000). Evidence from the lexicon includes the fact that speakers store idiosyncratic distributional facts, as for example that the adjective mere does not function predicatively (*the child seems mere). Syntactic evidence stems from the partial productivity of particular constructions. Goldberg illustrates this point with English Serial Verb Constructions such as Won't you come sit with me?, which cannot be tensed (*She came sat with me). Goldberg concludes that generalizations are learned gradually, and that they coexist with item-specific knowledge.

Part II: Learning Generalizations

Chapter 4 addresses how argument structure constructions are learned. In the spirit of the research program of Tomasello (1992, 2003), Goldberg presents evidence against the poverty-of-the-stimulus argument and points out similarities in the acquisition of constructions in children and adults. English corpora of child-caretaker interaction show that argument structure constructions are predominantly used with one highly typical verb, which strongly connects the semantics of that verb to the semantics of the construction as a whole. For example, the intransitive motion construction is overwhelmingly often instantiated with the verb go, the caused motion construction favors put, while the ditransitive construction occurs with give. Children induce the constructional meaning from these verbs, and generalize it once they hear the construction used with other verbs.

Chapter 5 presents an account of how generalizations, once they are acquired, are constrained. A problem all construction-based models have is the generation of sentences that actual speakers judge as ungrammatical or odd. For example, how is it that John wiped the table dirty is not an acceptable instance of the English resultative? Goldberg presents three factors that constrain the productivity of constructions. For a construction to be productive, it needs to have a high enough discourse frequency, a large enough variety of co-occurring lexical elements, and the particular extension must not be pre-empted by an already existing form.

Chapter 6 asks why generalizations are learned. It is argued that speakers are encouraged to learn constructions because they are a reliable predictor of sentence meaning. In fact, constructions indicate 'who did what to whom' more reliably than verbs alone. Corpus analyses of the caused motion construction and the ditransitive construction show that argument structure has a high cue validity for sentence meaning. While some verbs (e.g. put for the caused motion construction) are very reliable predictors of sentence meaning, other verbs require the listener to rely on argument structure. The importance of constructions as predictors of sentence meaning is further underscored by evidence from sorting tasks, where subjects ordered sentences according to 'overall sentence meaning'. Expectedly, sentences with the same verbs were sorted together, but subjects also sorted sentences together if they instantiated the same construction.

Part III: Explaining Generalizations

Chapter 7 offers a fresh look at a classic syntactic problem. Ross (1967) developed the notion of syntactic 'islands' whose constituents cannot be questioned or extraposed. For example, sentential subjects as in That she knew Bob bothered John, do not allow questioning of their internal object (*Who did that she knew bother John? ). While traditional accounts posit constraints on movement rules ('island constraints') for this phenomenon, Goldberg attempts an explanation in terms of constructional mismatch: The sentential-subject construction and the interrogative construction are each associated with particular information structure properties. These properties clash in the starred sentence above and render it unacceptable. Goldberg argues that 'islands' are pragmatically backgrounded constructions that encode presupposed information. Hence, it is pragmatically anomalous to make these constructions prominent through questioning.

In Chapter 8, Goldberg explores the functional underpinnings of English Subject-Auxiliary Inversion (SAI), which occurs in questions, conditionals, wishes, exclamatives, comparatives, and other constructions. The wide variety of constructions has been taken as evidence that SAI is a purely syntactic generalization with no meaning attached to it. Challenging this view, it is shown that SAI constructions can in fact be arranged in a radial network around a semantic prototype.

Chapter 9 shifts the focus away from just English and presents an argument against the purported cross-linguistic universality of linking rules. Goldberg argues that argument realization follows general functional principles. For example, agents and undergoers are cognitively salient, and therefore expressed in prominent syntactic slots. An innately specified mapping of agent and undergoer onto grammatical subject and object is problematic, because it presupposes the cross-linguistic existence of subjects and objects. Another tendency is that the number of arguments tends to align with the number of overtly expressed participants. Again, this is no innate feature of grammar, but results from the pragmatic pressure to express relevant and non-recoverable participants.

Chapter 10 briefly outlines the main differences between Construction Grammar and derivational theories of syntax. It then turns to a comparison of different brands of Construction Grammar itself. Under the new definition of construction (as outlined in Chapter 1) Goldberg's own model differs from the unification-based approach (Fillmore et al. forthcoming), which does not represent redundant information and hence is not fully usage-based. Another criticism concerns the use of feature AVMs, which may not capture semantic detail finely enough. The main difference between Goldberg's model and Cognitive Grammar (Langacker 1987) is that Cognitive Grammar aims to reduce grammatical categories such as 'noun' or 'subject' to their cognitive underpinnings, while Construction Grammar keeps the labels. This, however, does not amount to an endorsement of autonomous syntax. The third approach discussed here is Radical Construction Grammar (Croft 2001), which merely has a stronger cross-linguistic focus than Goldberg's model.

Chapter 11 presents a brief summary and the conclusion that a usage-based Construction Grammar is well-suited to explain generalization in language.

Evaluation

Constructions at work is an important book in several respects. First, its introductory part constitutes the most accessible general introduction to Construction Grammar that exists to date. It will be welcomed as a reference by students and Construction Grammarians, as well as linguists with other theoretical orientations.

Second, Parts II and III document the evolution of Construction Grammar as an empirically driven enterprise. Corpus resources and psycho-linguistic experimentation are increasingly gaining in importance in cognitive linguistics. Goldberg (and her collaborators, who deserve to be mentioned here), shows how these methods can be put to the task of investigating theoretically relevant questions. While Chapters 4 to 6 summarize a number of previous studies, Chapters 7 and 8 explore exciting new territory. In the spirit of Constructions (1995), phenomena that were thought to be purely syntactic receive a compelling functional account.

A possible criticism concerns the discussion of cross-linguistic generalizations in Chapter 9. Here, the universalist position is associated with the Isomorphic Mapping Hypothesis, which, as a literal one-to-one correspondence of arguments and NPs, is obviously too strong a position. Likewise, the critique of three different generative approaches on seven pages in the final chapter is probably too short to do justice to the individual authors. The section on the different brands of Construction Grammars is very interesting, and would have deserved some more space as well. That being said, Constructions at work is a book to keep somewhere close to your desk.

Acknowledgement: I'd like to thank Suzanne Kemmer for discussing the issues in this review with me. -MH.

References

Baker, Mark. 1988. Incorporation: A Theory of Grammatical Function Changing. Chicago: University of Chicago Press.

Barlow, Michael and Suzanne Kemmer (eds.). 2000. Usage-based models of language. Stanford: CSLI.

Bybee, Joan L. 2000. The phonology of the lexicon: evidence from lexical diffusion. In Barlow & Kemmer, Usage-based models of language. Stanford: CSLI, 65-85.

Bybee, Joan L. and Paul Hopper (eds.) 2001. Frequency and the emergence of linguistic structure. Amsterdam: John Benjamins.

Croft, William. 2001. Radical Construction Grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press.

Fillmore, Charles J., Paul Kay, Laura Michaelis, and Ivan Sag. (forthcoming) Construction Grammar. Stanford: CSLI.

Goldberg, Adele E. 1995. Constructions. A Construction Grammar Approach to Argument Structure. University of Chicago Press.

Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. I. Stanford: Stanford University Press.

Ross, John R. 1967. Constraints on Variables in Syntax. PhD Dissertation, MIT.

Tomasello, Michael. 1992. First verbs: A case study of early grammatical development. Cambridge: Cambridge University Press.

Tomasello, Michael. 2003. Constructing a language: a usage-based theory of language acquisition. Cambridge, MA: Harvard University Press.

Links

© 2002-present ICLA; all rights reserved.
Site admin information and contact