\'Do\'-support revisited: A Construction Grammar approach

Karen Malcolm
Linguistics Department, University of Canterbury, New Zealan

     Full text: Not available
     Last modified: March 1, 2004
     Presentation date: 07/15/2004 2:00 PM in SS Common Room
     (View Schedule)

Abstract
Do-support revisited: A Construction Grammar approach

Do-support has received numerous analyses in the syntax literature; mostly in transformational grammar (Chomsky 1957, 1991 and others). The framework of Construction Grammar (CG: Fillmore 1988, Goldberg 1995, Kay & Fillmore 1999) offers a view of syntax which is radically different from that of TG. I will here present an analysis of do-support in CG. The CG analysis allows us to view do-support as a unified phenomenon: it is a single construction, even though it can combine with numerous other constructions.

Do-support appears in the following general constructions:

(1) What do you think Frank is here for? (Subject-Auxiliary inversion)

(2) I do not think John is a nice guy. (Sentential negation)

(3) Sandy knew the answer, and Chris did too. (Conjoined-Clause Elision)


The do-support phenomenon has the following characteristics:

(a) Dummy do is always finite.
(b) Dummy do never appears immediately adjacent to its VP complement, and that complement is always an uninflected non-auxiliary verb.
(c) Dummy do always appears adjacent to the sentential subject

All of these characteristics are represented in the do-support construction being presented here:

(PLEASE NOTE: construction unavailable on web form, please see email version for the construction)
FIGURE 1: The do-support construction.

The do-support construction is inherited by the other constructions exemplified in examples (1), (2) and (3) because they each require an auxiliary verb with no semantic content, and dummy do fulfils this requirement. The analysis presented here is that all sentences have a verbal semantic (v-sem) value which limits which verbs may appear. Sentences with do-support require an auxiliary but have a v-sem requirement that does not include the semantic value for a normal auxiliary. This ensures the inheritance of the do-support construction, which supplies an auxiliary with no semantic value to complete the construction. The CG analysis provided thus offers a new perspective on do-support as a unified phenomenon, while it also captures the old insights that dummy do is semantically empty and that it only appears when ‘necessary’.