Abstract
A hotly debated issue concerning spoken word production is whether lexical selection is by competition or not. Recently, Mahon, Garcea, and Navarrete (2012) claimed that associative facilitation from color-related words in the Stroop task challenges lexical competition accounts of word production, such as implemented in the WEAVER++ model (e.g., Levelt, Roelofs, and Meyer, 1999; Roelofs, 1992). Associative facilitation concerns the finding that color naming (e.g., say “red”) is faster with associatively related words (e.g., the word fire in red ink color or combined with a red rectangle) than with unrelated words (e.g., lawn). However, in a comment (Roelofs and Piai, 2013), we demonstrated through WEAVER++ simulations that associative facilitation is fully compatible with a lexical competition account. In a response to our comment, Mahon and Navarrete (2014) argued that these simulations are problematic and that recent event-related brain potential (ERP) data provide evidence against lexical selection by competition. In the present article, we first briefly discuss the associative facilitation in the Stroop task and our demonstration by computer simulations that the facilitation is fully compatible with a lexical competition account of word production (Section 2). Next, we argue that the rejection of this demonstration by Mahon and Navarrete (2014) is based on a mischaracterization of the competition account and its computational implementation in WEAVER++ and other models in the literature. We make clear what competition models really maintain and how competition is computationally implemented in these models, and we show that under the correct interpretation of what competition entails in these models, the rejection of lexical competition by Mahon and Navarrete is unwarranted (Section 3).
