what speakers know

Working on a Saturday again. Contemplating linguistic knowledge:

Currently, the only theoretical framework that embeds indexicality centrally within phonological knowledge is the exemplar-based model of representation discussed by, amongst others, Goldinger (1997), Johnson (1997), Pisoni (1997), Pierrehumbert (2001, 2003a, b), Lachs, McMichael, and Pisoni (2003), Hawkins and Smith (2001), and Hawkins (2003). The exemplar model takes a very different point of departure from most other models by not assuming that lexical representations are stored solely in abstract and invariant form. Instead, knowledge of linguistic structure is built up by representing in memory the totality of linguistic experiences that an individual has. So, for example, knowledge of the sound patterning associated with the word cat is not considered to be reducible to something like a three phoneme string, /kat/. Rather, it consists of a detailed record of all of the exemplars that an individual speaker–listener has encountered of that word. In principle, a lexical representation may therefore include a potentially vast set of detailed acoustic traces based upon tokens an individual has heard, and a parallel set of traces bearing articulatory information about tokens that the individual has uttered. Each exemplar simultaneously encodes non-linguistic as well as linguistic information since the acoustic record contains reflexes of who was speaking and what the speaker’s voice sounded like (in terms of segmental features, pitch range, voice quality, etc.).

____________
Foulkes & Docherty (2006), ‘The social life of phonetics and phonology.’ Journal of Phonetics 34: 409–438