Does Phonetic Detail Guide Situation-specific Speech Recognition?

The evidence is overwhelming that listeners use both statistical patterns about abstract properties of their language, and its phonetic detail, to make informed predictions and decisions. Listeners seem to learn about new phonetic detail when it does not contradict other important cues to communicating meaning. Though we do not always use available phonetic detail, we probably monitor it constantly. An eye-tracking experiment shows that we use weak, subphonemic information to predict the morphemic status of words. This and other data suggest that we use phonetic detail when it is relevant to the task at hand, even when there is little apparent advantage in doing so. Since perceptual learning about speech seems central to interpreting or systematising a speaker‟s or group‟s use of phonetic detail, these various observations encourage speculation about how to model speech perception as part of a biologically-grounded, situated theory of human interactive behaviour.