> Andreas Johansson wrote:
> > Quoting Dirk Elzinga <dirk_elzinga@...>:
> >
> >
> >>Hey.
> >>
> >>In my dissertation on Goshute consonant phonology
> >>(
http://roa.rutgers.edu/searchlist.php3?
> >>num=7&detail=&pointer=0&search=elzinga&ids=431), I represented all long
> >>affricates with a doubled stop consonant symbol followed by a
> >>homorganic fricative symbol; i.e., [ttT, tts, ttS]. I did this because
> >>it is the stop closure which is lengthened, not the affricate as a
> >>whole.
> >
> >
> > It's physiologically possible to lengthen either part of the affricate, so
> I
> > just imagined a conlang that distinguishes tS~t:S~tS:~t:S:. Anadewism
> demands
> > some natlang sports the same contrast!
>
> But if the stop and fricative are treated separately for length, would
> that really be considered an affricate or just a stop-fricative cluster?