Why we Think that Syntax is Construction-Based

Laura A. Michaelis
Department of Linguistics and Institute of Cognitive Science

University of Colorado at Boulder

October 28, 2002

Room 182 Dwinelle

4 p.m.

ABSTRACT

With the decline of 'rule-free grammar', construction-based syntax is on its way (back) to respectability. Constructions play a central role in recent models of idiomaticity (Kay & Fillmore 1999, Kay 2002), sentence types (Ginzburg & Sag 2000, Michaelis & Lambrecht 1996), argument structure (Goldberg 1995, Jackendoff 1997, Rappaport Hovav & Levin 1998), typological variation (Van Valin & LaPolla 1997, Croft 2001), prosody (Lambrecht & Michaelis 1998), sentence processing (Hare & Goldberg 2000, Jurafsky & Narayanan 1998), inflectional morphology (Bybee 2001, Ackerman forthcoming), language acquisition (Bates & Goodman 2001, Tomasello & Brooks 1999), and language breakdown (Gahl et al. 2001). As rich as this literature is, it has not fundamentally changed the theory and practice of syntax. This is due in part to the inductive, case-based nature of the argumentation, in part to the perception that constructions are useful primarily for linguistic marginalia,  and in part to general confusion about just what constructions are, e.g., whether they differ from the functional projections of EGB (Borer 1998). In order to demonstrate how we might make the case for constructions, I will present a survey of (a) construction types and (b) the arguments, both linguistic and psycholinguistic, for a construction-based grammatical architecture. We will discuss the following syntactic arguments for constructionality:

• Many, if not most, constraints on the form-function interface are special-case constraints;
  
• Many morphosyntactic constraints and affordances are explicable only as paradigmatic effects;
  
• There are no transconstructional filters (i.e., 'parameters');
  
• Syntactic licensing relationships need not be syntactic sisterhood relationships;
  
• Syntactic heads do not necessarily determine the combinatory behaviors of their projections;
  
• Syntactic alternations, e.g., lexical rules, permit more 'product-oriented' generalizations than 'source-oriented' ones;
  
• As shown by coercion effects, semantic licensors need not be syntactic heads, and in fact include skeletal syntactic structures.

We will then look at supporting evidence from linguistic cognition, including studies of syntactic priming, syntactic ambiguity resolution, grammar acquisition, agrammatic comprehension, frame-based speech errors, and syntactic frequency effects.

Laura Michaelis in associate professor in the Department of Linguistics and a faculty fellow in the Institute of Cognitive Science at the University of Colorado at Boulder. She is the author of Aspectual Grammar and Past-Time Reference (1998) and co-author, with Josef Ruppenhofer, of Beyond Alternations: A Construction-Based Account of the Applicative Pattern in German (2002). Her published works include articles in Language, Linguistics & Philosophy, the Journal of Pragmatics and the Journal of Semantics. She is currently a visiting researcher at ICSI in Berkeley, where she is collaborating with Profs. Fillmore and Kay on a Construction Grammar textbook.