And Rosta wrote:
> Dick:
>>> Has the
>>> idea of a phonological network been developed by any of you WG
>>> experts?]
>>
>> *## Not really - it's an enormous gap in the WG literature.
>
> I've never thought of it as an enormous gap. That is, i've tended to
> think that it could be filled by turning to any representational (i.e.
> nonderivational) constraint-based theory of phonology and
> straightforwardly slot it into WG -- "plug-and-play". In a not dissimilar
> way, had you not been a syntactician, WG could also have borrowed, say,
> HPSG in similar plug-and-play style.
The problem is that neither constraint-based phonology nor HPSG are
neuropsychologically plausible.
> However, although GP & DP are very nice theories of pure phonology,
> I'm not convinced they fit in a cognitivist model, since it is
> nowadays becoming clearer and clearer that a speaker's phonological
> knowledge can't be separated from their phonetic knowledge -- or
> rather, there is no natural boundary between them. Like other
> linguistic categories, phonological categories must be learnt
> inductively from phonetic patterns in usage. That doesn't mean
> that the proper unit of phonology is the phoneme, but it does mean
> that there does have to be a type-token relation between phonological
> forms and their phonetic realizations.
Yep. If you allow both the formal types and their phonetic realizations to
be of any length (as opposed to forgetting about phonetics once you've
built segments out of gestures or features), you come pretty close to my
view.
-- Mark
Mark P. Line
Polymathix
San Antonio, TX
|