The selection of closed-class elements during language production: a reassessment of the evidence and a new look on new data Niels Janssen a *, Niels O. Schiller b and F.-Xavier Alario c a Departamento de Psicologı ´a Cognitiva, Social y Organizacional, Facultad de Psicologı ´a, University of La Laguna, La Laguna, Spain; b Leiden Institute for Brain and Cognition, Leiden University, Leiden, The Netherlands; c Aix-Marseille Universite ´ & CNRS, Marseille, France (Received 23 February 2012; final version received 9 May 2012) Are closed-class items such as free-standing determiners and bound inflectional morphemes selected by a competitive or a noncompetitive mechanism? Jescheniak, Schriefers, and Lemho ¨fer provide a survey of the literature on this topic involving studies employing the pictureword interference and simple picture naming tasks. They claim that the extant evidence supports a competitive lexical selection mechanism for both types of closed-class items. In this commentary, we critically evaluate their proposal and present new empirical data. We come to the conclusion that the available data do not allow a straightforward interpretation in terms of a particular hypothesis of lexical selection. We highlight the need to consider answers to the theoretical question within a broader research context. Keywords: language production; closed-class words; lexical selection; competition An ongoing debate in the language production litera- ture concerns the mechanism of lexical selection. Currently, two basic hypotheses are contrasted. The competitive lexical selection hypothesis assumes that target selection time is a function of the activation level of lexical competitors. By contrast, the noncompetitive lexical selection hypothesis assumes that target selection time is a function of the target’s own activation level irrespective of lexical competitors’ activation. The majority of studies concerning this question have focused on the production of open-class words such as nouns (e.g. Abdel Rahman & Melinger, 2009; Janssen & Caramazza, 2011). The current lack of consensus regarding this issue is evidenced by the notable number of articles published in the last years that adjust, rebut, or otherwise comment on previously published theoretical proposals (cf. most recently Mahon, Garcea, & Navarrete, 2012; Mulatti & Coltheart, 2012). Jescheniak, Schriefers, and Lemho ¨fer (in press), valuably broaden the scope of the debate regarding competitive versus noncompetitive models by reviewing evidence from studies on the production of closed-class items, rather than nouns. Specifically, they focus on the production of free-standing (e.g. determiners) and bound (e.g. inflections) gender- marked morphemes. From their review, they conclude that both types of closed-class items are selected by the same mechanism, and that this mechanism is compe- titive. In this commentary, we present a discussion of Jescheniak et al.‘s proposal as well as new empirical evidence which undermine their conclusions. Jescheniak et al. consider the production of free- standing and bound morphemes in two different tasks: the pictureword interference task (PWI; Schriefers, 1993), and the simple picture naming task (SPN; Alario & Caramazza, 2002; Schriefers, Jescheniak, & Hantsch, 2002). In the PWI task, studies have focused on the effect of gender congruency , in which pictures are named in the context of distractor words that either share or do not share the grammatical gender of the picture name. In Germanic and Slavic languages, there are reliable gender congruency effects in the production of free- standing determiners (Bordag & Pechmann, 2008; Costa, Kovacic, Fedorenko, & Caramazza, 2003; Schiller & Caramazza, 2003; Schriefers, 1993). For example, Dutch speakers named pictures with determiner noun NPs (e.g. de auto COM [the car]) slower in the context of gender incongruent (e.g. boek NEU [book]) than gender congruent (e.g. ker- k COM [church]) distractor words. 1 Both competitive and noncompetitive hypotheses can account for this effect. The competitive selection hypothesis explains slower latencies in the incongruent condition in terms of competition between divergent determiner forms. The noncompetitive hypothesis explains faster latencies in the congruent condition in terms of priming between convergent determiner forms. In other words, as *Corresponding author. E-mail: njanssen@ull.es Language, Cognition and Neuroscience, 2014 Vol. 29, No. 6, 695708, http://dx.doi.org/10.1080/01690965.2012.693617 Please note that this paper was previously published online in a different page layout. # 2012 Taylor & Francis Downloaded by [Professor Niels Olaf Schiller] at 12:44 28 April 2014