¾«Æ·SMÔÚÏßӰƬ

I work in a developing syntactic model called  (CxG). In CxG, rules of syntactic combination (descriptions of local trees) are directly associated with interpretive and use conditions, by means of semantic and pragmatic features that attach to the mother or daughter nodes in these descriptions. This amounts to the claim that syntactic rules mean things. This is a controversial claim: meaning is generally viewed as something that only words can do; syntactic rules are only supposed to determine what symbol sequences function as units for syntactic purposes. The reason is that if both words and syntactic patterns had meanings, we would create the potential for semantic conflict between (top-down) construction meaning and (bottom-up) word meaning. My contention is that this is precisely what we want to do, if we wish to model the grammatical flexibility that we actually encounter. In my research, I ask how interpreters resolve semantic conflicts between words and constructions, and how and why speakers create such conflicts. 

My focus is on tense and aspect. I view morphological markers of tense and aspect as constructions, each of which selects for a specific component in the temporal representation of a verb. In my work, I look at how speakers use tense and aspect constructions to alter verbal aspect (Aktionsart). It is my contention that semantic-conflict creation enables speakers to categorize events flexibly, which in turn enables them to express sequence and overlap relations among events in narratives. You can also find a comprehensive listing of references and news about CxG at the . 

The version of CxG that I assume is called Sign-Based Construction Grammar (SBCG), a formal extension of CxG proposed by Ivan Sag. For details, read Ivan's paper, . In SBCG, (combinatory) constructions are descriptions of local trees, or, more accurately, the mother signs of mother-daughter configurations. To the right you see an example of a construction, the Inverted Negative Adverb (INA) construction. The INA construction licenses sentences like Rarely am I chosen

. According to the INA construction, INA constructs consist of two daughters, an adverbial expression whose semantic frames include a negation frame, and a verbally headed daughter that bears the feature INV+. The INV feature is motivated independently of this construction, because it is needed to chararacterize defective auxiliaries like first-person aren't, which appears only in auxiliary-initial constructs like Aren't I invited? (as against, e.g., *I aren't invited).

At the , held at UT Austin, I gave a keynote talk in which I urged proponents of CxG to more actively promote the theory (in particular, in its sign-based incarnation) based on 'the three Fs': functionality of the theory (e.g., it encodes inheritance relations in type constraints), facts (e.g., core and periphery are interleaved during production) and fighting back (e.g., hierarchical structure and 'movement rules' are not universals of language but rather mutually reinforcing representational conventions). My 2012 article '' explores these points in some depth. 

One thing that distinguishes SBCG from other branches of CxG is its lexicalist foundation: based on the observation that most sentences, however idiomatic their meaning, have perfectly regular syntax, we leave the business of assembling words to the general-purpose phrasal constructions of the grammar, and focus most of our attention on describining the combinatoric properties of words, lexemes and lexeme classes. The lexicalist framework is a particularly useful tool for representing the syntactic flexibility of multi-word expressions (MWEs): rather than treating MWEs as 'words with spaces', we propose to treat them as bags of words—groups of idiomatically combining forms assembled in the same ways that non-idiomatic words are. Paul Kay and I offer a succinct demonstration of this approach to MWEs in the paper '', which concerns the idiomatic expression illustrated by That has nothing to do with it.