Constructions at work: The nature of generalizations in language (review)

Adele Goldberg's Constructions at work is a welcome sequel to her 1995 Constructions, by now a landmark in linguistics. The new book extends her previous analyses and explores new and exciting territories. Since G is arguably the leading figure in construction grammar, the theory is here pitched against the background of alternative approaches, functional syntax and generative grammar, in order to convince the reader that whether one agrees with G or not, the book is essential reading.

Part 1 (Chs. 1–3) sums up the basic claims of construction grammar. Constructions are form-function pairings, and the idea is that grammar is 'constructions all the way down' (18). In other words, clause-level syntactic constructions, phrases, collocations, words, and morphemes are all analyzed in a similar, construction-based fashion. In addition, linguistic expressions (constructions) are taken to reflect both high-level, abstract generalizations (elegant rules) and low-level generalizations, as well as even idiosyncratic phenomena. These analyses, moreover, are not mutually exclusive, so the same linguistic string may receive multiple representations. The assumption is that we store 'redundant' representations (e.g. both the general, abstract -ed rule for forming English past-tense verbs and concrete representations for individual, regular past-tense verbs—verb + ed).

G's main research program concerns argument-structure constructions, those clause-level form/ function pairings used to express basic meanings, such as TRANSFER (as in the ditransitive con-struction I gave him a red pepper, SBC:011),¹CAUSED MOTION (subject, verb, object, a directional oblique: I was thinking of [sending Matt up there], SBC:014), and so on. Her argument has been that the computation of meaning must take into account constructional meaning, and not just the individual lexical items according to their syntactic roles (the usual way compositionality is applied).

As G acknowledges, hypothesizing a constructional meaning in the above two examples seems unnecessary (redundant). After all, the verbs give and send in any case denote TRANSFER and CAUSED MOTION (respectively). But, as G has convincingly demonstrated, verbs that do not denote TRANSFER (e.g. throw) or CAUSED MOTION (e.g. squeeze) can nonetheless be so interpreted if incorporated into the appropriate constructions (e.g. They usually throw me stale corn bread, See if [I can squeeze that in], LSAC). Once we assume constructional meanings, we avoid unnecessary multiple meanings and argument structures for verbs. Throw need not (also) be analyzed as a transfer verb requiring a recipient, nor is squeeze treated as (also) a motion verb requiring a locative. The recipient in the former and the directional oblique in the latter are contributed by the constructions instead.

G does not shy away from assuming constructions that have only limited productivity (e.g. the goVPing construction, as in So they go barging in on Mar, SBC:006). At the same time, metaphorical interpretations (e.g. Let's throw her a good party, LSAC) can extend construction meaning and use. No physical party is handed over in this ditransitive construction, but a metaphorical act of giving is conveyed, despite the lack of a transfer verb. This is due to the construc-tion. G notes that pragmatic functions are quite relevant in accounting for how and why speakers use specific linguistic constructions. For example, goVPing is associated with negative actions. In fact, G argues that both individual grammars and crosslinguistic 'universals' are motivated extralinguistically.

Now, Gis not the first to have argued for grammatically specified 'added values' (meanings) for syntactic constructions. Kuno (1972), Prince (1978), Fillmore and colleagues (1988), Thompson (1990), and others have convincingly shown that syntactic constructions convey more than what their components amount to. G shares some assumptions with these functional approaches. She too values subtle meaning distinctions (see the analysis of goVPing and other constructions in Ch. 3). She also fully recognizes the importance of procedural meanings (Ch. 7). Nonetheless, she has in effect challenged many of the assumptions and practices of functionalists.

G's is a...