We examined acoustic differences between phonemically identical sequences arising from French elision (l'affiche 'the poster', la fiche 'the sheet', both [[image omitted]]) and listeners' discrimination and identification of these sequences. Experiments 1 (ABX paradigm) and 2 (forced choice task) demonstrate that listeners discriminate between and identify such utterances. Experiments 3 and 4 used a cross-modal identity paradigm. Although targets (fiche, affiche) were activated regardless of the intended segmentation in the prime ([[image omitted]], [[image omitted]]), priming was greater in the intended segmentation, suggesting that listeners retrieve the correct segmentation and modulate activation in favour of the correct candidate, without ruling out alternative interpretations. Experiments 5 and 6 used fragment priming to evaluate the time course of the use of acoustic cues. Consonant-initial targets (fiche) were activated regardless of intended segmentation ([[image omitted]], [[image omitted]]) whereas vowel-initial targets (affiche) were only activated in their intended segmentation ([[image omitted]]). These results provide further evidence for fine-grained lexical access.