Structural linguistics teaches that one of the principal properties of speech sounds is being different from each other: the essence of a phoneme is that it is not any other phoneme. In addition, many structuralist theories of sound change are based on notions of preservation of contrast between phonemes. From this one might expect that it should be possible to observe speakers' efforts at maintaining contrasts in speech. One situation where this would be expected is in repeated speech, i.e., where a speaker repeats a word after receiving feedback that it has been misapprehended as another similar word. This paper reports results of an analysis of such repeated speech samples of 10 English speakers when they produced one of 12 near-minimal syllables (bayed, paid, bed, ped, bet, pet, bid, bit, etc) under two conditions: control and repetition (in response to feedback that their initial production had been misunderstood as one of the other syllables). Contrary to expectations, there was no significant contrastive exaggerations in VOT or vowel duration as a function of the word presented in feedback.