Some Implications of a Copy Theory of Labeling

Abstract

Abstract A copy theory of labeling makes interesting empirical predictions when combined with a revised version of Kayne's (1994) Linear Correspondence Axiom and a category‐based version of c‐command. First, it forces the pronunciation of the highest copy in a chain, all else being equal, without the need for copy deletion. Second, it predicts certain differences between underlying and derived head‐initial specifiers and between head‐final and head‐initial specifiers. This serves to extend Uriagereka's (1999) PF ‐based explanation of Huang's Condition on Extraction Domain ( CED ) to cover previously problematic or unexplained phenomena: (i) the apparent lack of subject‐island effects with head‐final specifiers; (ii) the nonislandhood of many derived specifiers (externally merged as complements); (iii) patterns of complement extraposition; and (iv) the Final‐over‐Final Constraint (see Holmberg 2000; Biberauer, Holmberg & Roberts 2008, to appear). All of the above phenomena arise under the copy theory of labeling as side effects of a revised version of Kayne's (1994) Linear Correspondence Axiom incorporating a head parameter, whereby asymmetric c‐command, defined between categories and unmediated by dominance, maps to precedence as a last resort.