Underlying and surface representations
The variation in the pronunciation of words is truly mind-boggling. Biological differences between speakers lead to different acoustic outputs, and independently of whether they are adult or child and male or female, people speaking the same language will have different accents, depending on their social class and the region they grew up in. Within the speech of any one speaker, the pronunciation of words will vary with the degree of formality. And even if we tried to remove all variation from our speech, we could not produce acoustically identical pronunciations of the same word, due to the inevitable variation in the physiological mechanics and physical conditions of our environment. One specific type of within-speaker variation is central to our concern. The same speaker speaking in the same style will systematically vary the pronunciation of the same word as a function of the phonological context. In the English spoken in North America, the final consonant in an expression like Right! will be an unreleased [t], but in Right on! , this same /t/ will be an alveolar flap [ɾ]. The pronunciation of the English past tense suffix is either [t] or [d], depending on the voicing of the preceding segment, giving [lυkt] as the past tense of look [lυk], but [bεgd] as the past tense of beg [bεg]. And if the verb stem ends in [t] or [d], we find it is pronounced [d], as in [bfətd] and [ni dd], the past tense forms of buffet [bfət] and need [ni d].