A topic in the Open Knowledge Graph — a free, open map of 15,495 topics and the order to learn them in.
Construction Grammar is a family of theories holding that grammar consists of constructions: learned pairings of form and meaning at every level of size and abstraction, from words and idioms to abstract sentence patterns. In this view there is no sharp division between the lexicon and the rules; "kick the bucket," "the X-er, the Y-er," and the English ditransitive pattern are all stored constructions. Adele Goldberg's 1995 book argued that argument-structure patterns themselves carry meaning, which explains sentences like "She sneezed the napkin off the table," where the verb alone does not license the extra arguments. The approach is closely tied to usage-based accounts of acquisition and to cognitive linguistics.
Start with partly fixed idioms that standard phrase-structure rules handle awkwardly, such as "let alone" or "What's this fly doing in my soup?", and ask what a speaker must know to use them. Then compare a lexicalist account, which lists argument frames in each verb's entry, with a constructional account, which attaches meaning to the sentence pattern. Test both on novel verb uses, such as "He texted me the address." Read the foundational papers alongside a usage-based acquisition study to see how the theory links grammar to learning.
Mainstream generative syntax has often divided linguistic knowledge into two parts: a lexicon of words with their properties, and a general system of rules that combine them. Idioms such as "kick the bucket" were treated as exceptions stored in the lexicon. Construction Grammar challenges that division.
The starting point was a set of patterns that are neither fully fixed nor fully general. In a 1988 paper, Charles Fillmore, Paul Kay, and Mary Catherine O'Connor analyzed "let alone," as in "I can't afford a bicycle, let alone a car." Its use depends on special conditions of meaning and context that no general rule predicts. Similar cases include the comparative correlative ("The more you read, the more you know") and the "What's X doing Y?" pattern ("What's this fly doing in my soup?"), which expresses surprise or disapproval, not a literal question about activity. Speakers must learn each of these as a unit with its own form and meaning.
Construction Grammar generalizes this point. A construction is any conventional pairing of form and meaning. Constructions range from fully specific (the word "avocado") through partly open ("the X-er, the Y-er") to fully abstract (a subject-verb-object clause). The whole inventory is sometimes called the constructicon.
Adele Goldberg's *Constructions* (1995) extended the idea to argument structure. Compare:
A lexicalist approach needs a new sense of "sneeze" or "bake" for each frame. Goldberg proposed instead that the caused-motion and ditransitive constructions carry meaning themselves. The verb fills in manner, and the construction supplies the event structure and argument slots. When a verb's meaning conflicts with a construction, the construction can coerce a new interpretation, or the sentence becomes odd.
Construction Grammar is a family of approaches. Berkeley Construction Grammar and Sign-Based Construction Grammar use formal feature structures similar to HPSG. Goldberg's version and William Croft's Radical Construction Grammar are closer to cognitive linguistics.
The theory connects closely to usage-based acquisition research. Michael Tomasello and others argue that children first learn item-specific patterns, such as "more ___" or "Where's the ___?", and gradually abstract general constructions. On this view, grammar is learned from experience with constructions rather than derived from innate parameters, a point of continuing debate with generative approaches.
No topics depend on this one yet.