The dynamics of spoken word recognition are acutely sensitive to competition among similar-sounding words. Here, we take advantage of this sensitivity to examine the manner in which native Italian speakers who are late learners of English perceive English vowels. Native Italian speakers and native English speakers listened to recordings of naturally produced words (pin, pen, and pan) and used a computer mouse to select the matching stimulus from an array of two pictures. The same participants also performed a similarity judgment task. The perceptual similarity space for these vowel categories differed between groups, and these differences were also reflected in the dynamics of performance in the online measure. The results are largely interpretable in terms of models of second-language speech perception that predict performance from patterns of assimilation to native language categories. The results suggest, however, that there are also effects of graded differences in the perceptual similarity of categories as measured in native speakers of the target language.