Although many speech perception studies have suggested that long-term memory representations of phonemes induce categorical perception along an acoustic continuum (e.g., voice-onset time; VOT) when identifying speech sounds, other studies have suggested that acoustic information is preserved and that graded responses can be observed in within-category comparisons. Using subjective confidence reports, we present findings that support the use of both acoustic and phonemic cues during speech perception. Replicating earlier findings, we observed evidence for two well- defined phoneme categories along the voice-onset time continuum. Additionally, we also observed overconfidence in responses suggesting that the explicit representation of phonemes differs from the representations used to make identification and discrimination responses. Taken together with results from other studies, our findings support the claim that listeners can access both phonemic and acoustic representations, with explicit knowledge of the former but not the latter.