Adjunct islands are configurational

Abstract

This paper argues for the Spell Out theory of the Adjunct Condition, based on Uriagereka (1999), Johnson (2003) and Sheehan (2010), using new evidence from Balkar. The Spell Out theory makes two claims: (a) between any two phrasal sisters at least one must be spelled out and become opaque for movement; and (b) a spelled out constituent does not project its category. This predicts adjuncts to be opaque, since they are maximal projections and are merged with a phrase. The Spell Out theory predicts that modifiers can be transparent for movement, but only if they are merged with a head (as complements) or if their sister is spelled out. The argument from Balkar is based on the behavior of so-called coverbs (clausal modifiers). Balkar converbs come in three varieties: vPs attached at the vP-level or as structural complements, TPs attached between the vP and the T′ of the main clause, and CPs attached at the CP-level. vP-converbs are only transparent for scrambling if they are complements. TP-converbs are never transparent. CP-converbs are only transparent if the main clause that they modify is opaque. Thus, Balkar converbs are transparent in all and only structural configurations that are predicted to be transparent by the Spell Out theory. In the end of the paper I discuss English data from Truswell (2007) and argue that the analysis proposed for Balkar can be extended to them as well.