A NOTE ON LONG EXTRACTION IN VATA AND THE ECP*
Arguments of a predicate and adjuncts behave differently with respect to Wh-extraction. Most current assumptions attribute this asymmetry to a single requirement on Wh-traces that adjunct traces fail to observe. and (henceforth HLS) suppose this requirement follows from the ECP. 1 Aoun (1984, 1986) assumes it to be the principle A of the Generalized Binding Theory (henceforth GEBI). This paper presents new data from Vata, a West-African language of the Kru family. Our main purpose here is to show that such analyses as HLS's and Aoun's are not sufficient to account for the full range of extraction possibilities and the distribution of Wh-traces. Instead, we argue that two distinct principles must be postulated: first, a principle like the ECP, requiring empty categories to be governed in an appropriate sense and, in addition, a second principle regulating the distance between a target of Wh-extraction and its immediate antecedent. Many possible alternatives come to mind to solve the problems that the Vata data reveal. Here we will limit ourselves to showing that the Condition on Long Extraction, argued for in Sportiche (1985, 1986) will handle these problems straightforwardly and better than some existing alternatives.
The paper is organized as follows. Section 1 briefly describes the HLS account. In section 2, we present the Vata data. We discuss its theoretical implications and show why they pose problems for the HLS account. In section 3, our analysis is sketched. In section 4, we show why GEBI is not sufficient by itself and we briefly discuss the extension of GEBI proposed in Wahl (1985). 2 * Research for this paper was supported in part by a grant from the Social Sciences and Humanities Council of Canada 4#410-84-503. We wish to thank three anonymous reviewers for their comments onan earlier version.
i Recall that the ECP requires empty categories to be governed either by a lexical head or by a position containing an antecedent, this last case being called antecedent government. 2 There are of course other proposals not discussed here, in particular and , but see Koopman and Sportiche (1986).