Grammatical gender does not constrain lexical access: Evidence from ERPs and behavioral results

Grammatical gender is a pervasive feature of many natural languages. It provides coherence among grammatical elements, for example between a noun and elements in the determiner phrase (DP). In French, bare nouns are generally not permissible such that all nouns must be preceded by a determiner, which must agree in gender and in number. Hence, the French auditory fragment “la sta…” can only be completed by a feminine noun such as “la statue.” Given its pervasive nature, several studies have examined whether and, if so, when, grammatical gender plays a role in lexical access. Some have claimed that grammatical gender guides lexical access and restricts activation to candidates that match in gender with the preceding determiner or adjective (1,2,3). Subsequent work has challenged this claim (4). It is noteworthy that all of these studies have used behavioral measures to examine processing whereas work using ERPs has shown that initial lexical access may indeed differ as a function of the lexical properties of a target word (5). In the present study, we provide evidence from two cross-modal priming experiments in French, one measuring RTs and the other ERPs, which confirm that lexical access is not constrained by grammatical gender. Primes were auditorily presented non-homographic French homophones preceded by a definite determiner (eg. la pique /lɑ pik/ “the spear” which has the homophonic mate le pic /lə pik/ “the peak”). All homophones had both a dominant and a subordinate meaning but only the subordinate meaning was presented, as specified by the gender of the determiner (eg. /lɑ pik/). A 2 x 2 design was used: auditory Primes were either the subordinate homophone or a frequency matched control (/lɑ pik/ vs. /lɑ fuʁʃ/ “the spear vs. the pitchfork”) and visual Targets were associated to either the dominant or subordinate meaning of the homophone (montagne/lance “mountain/lance”). Prime-Target SOA was zero milliseconds. In Experiment 1, RTs were recorded as participants made lexical decisions to target words and filler items. In Experiment 2, ERPs were recorded and a go/no-go task was used in which participants responded only to probe items (city names). Both Experiment 1 and 2 showed activation of both the dominant and subordinate meaning of the homophone following the gender-specified subordinate homophone prime versus an unrelated control prime. In Experiment 1 this was shown by a significant reduction in RT to both types of target following subordinate primes whereas in Experiment 2 it was shown by a significant reduction in the N400 component as well as in a later 500-800 ms window (5) to both types of target. No interactions were found. Our results thus support models that purport an exhaustive account of lexical access and, moreover, demonstrate such using methods that do not require a direct response to either the homophone or related targets. Indeed, our ERP results reveal that independent of whether gender appropriate or not, both meanings of homophonic primes are immediately activated and affect the cortical response to subsequent targets. Our ERP results also belie the findings from previous ERP work in sentential context showing that the N400 response to the contextually inappropriate meaning of a homophone is delayed (5). To conclude, we provide unequivocal evidence from complementary techniques that grammatical gender does not play a role in the early stages of lexical access in French