Construction Grammar

Abstract

Construction Grammar was originally developed as a monotonic, constraint-based framework whose conceptual basis rests on the fundamental assumption that grammatical patterns are complex signs, in principle not much different from lexical signs: a grammatical pattern is treated as a conventional association between form and function or meaning, i.e., a grammatical construction in a technical, theoretically grounded sense. The framework is the brainchild of Charles Fillmore, who started publishing his constructional research in the mid- to late 1980s. The approach developed out of a confluence of interests—linguistic, cognitive, anthropological, philosophical, computational—that all revolved around the idea that linguistic form is inextricably bound to its meaning and its communicative function and that this connection must be the basis for any descriptively and explanatorily adequate theory of language. Constructions are multidimensional symbolic entities that represent hypotheses about speaker’s linguistic knowledge and, as such, they allow for both the Gestalt view of linguistic patterning (unlike mainstream generative theories of language) and for keeping track of the internal properties of larger patterns (like any other grammatical theory). Constructional analysis thus makes a systematic distinction between what conventionally identifies a construction as a whole versus what is characteristic of its constituents, thereby giving a theoretical status to the well-known slogan that a construction is not just the sum of its parts: it may indeed have its own idiosyncratic properties, unpredictable from the properties of its constituents. Grammar, then, is seen as consisting of networks of constructions, related through shared properties. Constructionally based analysis is now a widely known and widely accepted approach, which continues to develop and extend in various new directions. It is not a monolithic enterprise with uniform goals, interests, or even representational conventions. There are various strands of constructional analysis, some more formally oriented (Head-Driven Phrase Structure Grammar and its most recent variant, Sign-Based Construction Grammar), some more cognitively focused (usage-based models), not all of them monotonic, not all of them sharing the same formalism. Nevertheless, all constructional scholarship reflects the universal interest of constructional grammarians in studying grammar in its use, rather than as an abstract entity independent of its communicative and cognitive grounding. Different strands overlap in various ways and to a varying degree, but differences among scholars are more a matter of emphasis than a fundamental disagreement about the nature of grammatical structure. The focus of this article is the Fillmorean strand and its most directly related extensions. While originally designed for the purpose of synchronic syntactic descriptions, Construction Grammar has now found its way into many other areas of linguistic research, including morphology, discourse and text analysis, diachrony, acquisition, computational applications, and corpus linguistics. It follows that the field is also developing methodologically, including emphasis on sophisticated statistical methods and some interest in experimental research as contributions toward improving the empirical grounding of constructional analyses.