E f+AOk-sto Garth Wallace <[log in to unmask]>: > Muke Tever wrote: >> >> Don't forget the dialects that use [h] for syllable-final /s/. >> In such a case you could actually have a new /h/ phoneme from the merger >> of /x/ > [h] and /s/ > [h] (there's a rule somewhere that a single phone >> cannot be an allophone of two different phonemes) > > Are you sure about that? What about neutralization? I know I understated the rule there to avoid overcomplicating the sentence. IIRC, the rule is that if a phone occurs as an allophone of two different phonemes, that phone must itself be a phoneme. In such a case you have alternation of phonemes and not just allophony, which is why we can inherit paradigms like knife +AH4- knives, when there is nothing phonetically unacceptable about *knifes (we can make do with _laughs_ after all). As far as neutralization goes (if I understand correctly) /ra:d/ and /ra:t/ realized as [ra:t] only has the rule telling us there must be a phoneme /t/ (which in this case isnt very helpful). *Muke! -- http://frath.net/ E jer savne zarj+AOk- mas ne http://kohath.livejournal.com/ Se imn+AOk- koone'f metha http://kohath.deviantart.com/ Brissve m+AOk- kol+AOk- ad+AOI-.