[The exploitation of fine phonetic detail in the processing of L2 French, Part V. Phonology and lexicon]
French is a language that poses particular difficulties for the second language (L2) learner in the processing of continuous speech. The phonological processes of <i>liaison</i> and <i>enchainement</i> (resyllabification), can render syllable and word boundaries ambiguous (e.g. un air ‘a melody’ and <i>un nerf</i> ‘a nerve’, both [oe˜ne˜]). Some research has suggested that speakers of French give listeners acoustic cues to word boundaries by varying the duration of liaison and initial consonants and that access to mental representations in the lexicon is facilitated by these cues (e.g. Spinelli, McQueen & Cutler, 2003); however no study to date has directly demonstrated that durational differences are exploited in the online segmentation of speech. One way to directly test the exploitation of duration as a parsing cue by both native and non-native speakers is to manipulate and exaggerate this single acoustic factor while holding all other factors constant. To this end, the current study employed ambiguous French phrases in which the pivotal consonants (i.e. the /n/ in <i>un air/nerf</i>) had been instrumentally shortened and lengthened while the rest of the phrase remained unaltered. Eighteen native speakers of French and 18 advanced late learners of L2 French were tested on an AX discrimination task and a forced-choice identification task employing these manipulated stimuli. The results suggest that duration alone can indeed modulate the lexical interpretation of sequences rendered ambiguous by liaison in spoken French. In addition, although a good deal of variance was observed in both participant groups, five out of 18 non-native participants scored at or above the native mean on both perceptual tasks. These results are particularly interesting in that they suggest that not only can advanced L2 learners develop native-like sensitivity to non-contrastive phonological variation in a L2, but that these learners can exploit this information in L2 speech processing.