ABSTRACTResearch has demonstrated that second language (L2) vocabulary learning improves when target words are presented in acoustically varied compared with acoustically consistent formats. The present study investigated the extent to which this benefit of acoustic variability is a consequence of difficult encoding demands (cognitive effort hypothesis) versus increased representational quality. Experiment 1 compared L2 vocabulary learning for words produced in normal (easier encoding) or nasal (more difficult encoding) voice. Vocabulary learning was superior in the normal-voice condition, arguing against a simple cognitive effort hypothesis as the basis for improved L2 vocabulary learning with increased acoustic variability. Experiment 2 assessed the resistance of newly acquired L2 word forms to the effects of acoustic degradation. Participants heard six repetitions of each item in either a single-talker or multiple-talker condition. The robustness of the new word-form representations was assessed by measuring the accuracy and latency of L2 to first language (L1) translation as a function of the signal/noise ratio. At all four signal/noise ratios, accuracy and latency of L2 to L1 translation were significantly better for words learned in the multiple-talker as opposed to the single-talker condition. Of particular importance, the difference between single talkers and multiple talkers increased systematically as signal/noise ratio decreased. These findings suggest that the benefits of acoustic variability are a consequence of learners' ability to retain and use indexical information during the earliest stages of word learning and provide support for the representation quality hypothesis.
