Lexical Retuning Targets Features

Abstract

Previous work has shown that listeners systematically "retune" their categorical boundaries when presented with real-word tokens containing an ambiguous segment. There have been multiple previous studies on this lexically-guided retuning (Eisner and Mc-Queen 2005; McQueen et al. 2006; Norris et al. 2003), but the level of representation that retuning targets is not entirely clear. In this paper, we show that retuning occurs at the level of features for fricatives. Given that previous research has argued similarly for stops (Kraljic & Samuel 2006), we suggest that retuning generally targets featural representations.