(12) Explaining the gender effect in spoken-word recognition: Evidence of L1 interference during L2 lexical competition

In a recent eyetracking study, Paris and Weber (AMLaP 2004) showed for cognate nouns that native gender information can influence non-native spoken-word recognition. It is known that during the recognition of L2 nouns, onset-overlapping nouns in the L1 are also activated (e.g., Weber and Cutler, JML, 2004). However, when the competitor's gender in the participants' L1 did not match the gender of the target, cohort activation was supressed: French-speaking participants did not look at the picture of a canon when instructed in German to click on a cassette (die _[feminine] Kassette), despite onset overlap in the two languages, because in French canon is masculine.

This was interpreted as evidence that gender's influence on lexical access is due to grammatical gender categories, not to the sound-based co-occurrence of articles and nouns. However, competitor nouns in Paris and Weber (2004) were cognates, and L2 gender could be stored differently for cognates and non-cognates. In visual-word recognition, for example, Lemhöfer (2004, PhD Dissertation) obtained an effect of L1 gender for cognates but not for non-cognates. Thus, a more stringent test of the gender effect's origin would involve non-cognate competitors.

Eighteen proficient German learners of French took part in this experiment. They were instructed in French to click on a target in a display also containing a competitor and 2 unrelated distractors. As in Paris and Weber (2004), the target was preceded by its gender-marked article. Here, however, the competitor was a non-cognate, whose German name overlapped in onset with the French target, although its French name did not. In the listeners' native language, German, the competitor shared gender with the target in the same-gender pairs (1) and differed in gender in the different-gender pairs (2). The gender of the competitor's name in French always differed from that of the target.

German non-native listeners looked more often at the competitor than at the unrelated distractors in the same-gender pairs: French la _[fem] table (`table') activated German Tanne _[fem] (`fir'). In the different-gender pairs, however, no such difference was found: le _[masc] radis (`radish') did not activate Rakete _[fem] (`rocket'). A French control group of 12 participants showed no competitor activation in either condition.

These results thus confirm the findings of Paris and Weber (2004). Also for non-cognates, L1 gender categories seem to interfere with lexical competition in L2. Given that effects of gender from the non-presentation language are unlikely to be caused by form-based regularities, the present results further strengthen a grammar-based origin of the gender effect.

		French	German	English
(1) same-gender	target	table _[fem]		'table'
	competitor	sapin _[masc]	Tanne _[fem]	'fir'
(2) different-gender	target	radis _[masc]		'radish'
	competitor	fusée _[fem]	Rakete _[fem]	'rocket'