Abstract

According to Construction Grammar (CxG), what we know as native speakers of a language is both more concrete and more elaborate than what is implied by the prevailing model of grammar as a highly general symbol-concatenating device (Goldberg 1995, 2006, Lambrecht & Michaelis 1996, Kay & Fillmore 1999, Croft 2001). CxG sees grammar as an inventory of grammatical patterns of varying grains – from the most open (e.g., the head-complement construction) to the least open (e.g., the VP idiom fall off the wagon). In this talk, I will argue that, despite the controversies surrounding construction-based analysis, there is no sensible conception of meaning composition without constructions. I will then briefly describe the conceptual foundations and architecture of a formal version of CxG called Sign-Based Construction Grammar (SBCG), using the SBCG treatment of verbal argument structure as an illustration. I will then discuss six phenomena that seem to substantiate a construction-based view of language:

• Productive idiomatic patterns (Fillmore & Kay 1999, Kay 2002, Culicover 1999)
• Paradigmatic effects in morphosyntax (Ackerman 2003, Birner, Kaplan & Ward 2007)
• Failures of transconstructional filters (Zwicky & Pullum 1991, Van Valin & LaPolla 2000, Croft 2001)
• Head-phrase mismatch (Zwicky 1995, Michaelis & Lambrecht 1996)
• Product-oriented generalizations (Barlow & Kemmer 2000, Bybee 1995, 2001)
• Coercion (Jackendoff 1997, De Swart 1998, Michaelis 2004, Lauwers 2008).

In concluding, I will suggest that CxG might ultimately deliver on the unfulfilled promise of generative syntax: to provide a model of human creative potential as manifested in language.