The CRITICAL DIFFERENCE in models of speech production: A response to Roelofs and Piai

In our original article (Mahon, Garcea, & Navarrete, 2012), we replicated Dalrymple-Alford’s (1972) observation of semantic facilitation at zero SOA when participants name the ink color (e.g., ‘red’) of semantically related non-color words (e.g., ‘fire’) compared to semantically unrelated non-color word distractors (e.g., ‘lawn’; see also Glaser & Glaser, 1989). We replicated this classic finding to draw attention to the implications of this semantic facilitation effect, and semantic facilitation effects generally, for a model of word selection in speech production. The implication is straightforward: The theory of lexical selection by competition predicts that the distractor ‘fire’ should interferemorewith saying ‘red’ than thedistractor ‘lawn’. This (and other) observation(s) of semantic facilitation are incompatible with the theory of lexical selection by competition. WEAVERþþ (Roelofs, 1992, 2003) is the most developed model that implements lexical selection by competition, and as itmakes both qualitative and quantitative predictions, its ability to simulate response time effects has been taken as a litmus test of the viability of the hypothesis of lexical selection by competition. As Roelofs and Piai (2013) write of the original (i.e., Roelofs, 2003) WEAVERþþ simulations, “.facilitation of 41 msec or more was obtained at preexposure SOAs and no effect at zero SOA” (p. 1768). In other words, the previously published simulations of WEAVERþþ did not resemble, in relevant ways, the empirical data. On that