Six talkers (two men, two women, and two children) listened to recordings of their own vowels, and to vowels produced by five other speakers both in isolation and pVp contexts. On the whole, subjects perceived their own vowel productions more accurately than those of the other speakers. Further, they perceived vowel tokens in a pVp context more accurately than in isolation even when they were presented with their own vowel tokens. Thus both speaker normalization and formant‐transition information can be considered important variables in determining vowel identification.