Retour vers liste

Détail de la contribution

Auteur: Anna GREENWOOD

Questioning anti-c-command: A comparative study of parasitic gaps

Abstract/Résumé: Sentences like (1)-(2) exhibit a phenomenon where a single A'-extraction leaves more than one semantically coreferent gap. The considerable literature (Engdahl 1983, Chomsky 1986) agrees broadly that in such cases one of the gaps is, in some sense, primary (marked as t). The others are parasitic (pg), in that they somehow depend on the primary gap for existence. 1. Who did John hug t after firing pg? 2. Who did you warn t that the police were going to arrest pg? (Culicover 2001) This paper observes how the structure of PGs inside finite complement clauses, such as (2), is problematic for current analyses of PGs with island hosts (1). It is generally agreed upon that anti-c-command is a condition on the well formedness of island PGs. Sentences such as (3) are said to be ungrammatical because the primary gap c-commands the PG. 3. *Who t hugged John after firing pg? This condition, however, does not hold for complement PGs. Using standard binding principles (Barss and Lasnik 1986), I demonstrate that complement PGs are c-commanded by their primary gap, which the current theories predict to produce ungrammaticality. The anti-c-command condition has been built into many analyses of PG creation. This paper adopts the null operator movement analysis of island PGs (Chomsky 1986) and raises the question of whether this is fitting for complement PGs as well. In this analysis, PGs are analyzed as R-expressions, and so c-command between the two gaps produces a Condition C violation. This means either that complement PGs as in (2) cannot be accounted for by null operator movement, or that the anti-c-command condition is faulty and must be reworked. I call these two consequences "distinctiveness" and "uniform" hypotheses, respectively. The distinctiveness hypothesis predicts that complement PGs are created either by a non-movement process, or by A'-movement of the wh-phrase. Observing that complement PGs undergo island violations, I posit that complement PGs are created by movement. Furthermore, evidence from binding rejects the notion that the wh-phrase in constructions like (2) is base-generated inside the complement. Upon these grounds, I abandon the distinctiveness hypothesis and argue that both types of PGs are created by null operator movement. This finding means that PGs should not be analyzed as R-expressions, as no Condition C violation occurs in (2). The anti-c-command condition is therefore inapplicable. This paper invites further discussion of what mechanisms cause the ungrammaticality of sentences like (3). The theoretical work provides the foundation for future experimental work concerning the acceptability of PGs in islands vs. complement clauses.