Quoting "Mark P. Line" <[log in to unmask]>:

> Andreas Johansson said:
> > Quoting Nik Taylor <[log in to unmask]>:
> >
> >>
> >> The Cambridge Encyclopedia of Language lists a language Amuesha
> >> (Andean-Equatorial family) with a vowel system /e a o/.
> >
> > I'd be itching to analyze that as /i a u/.
> Why?

Because that would be a much less typologically marked set.

Actual judgment would, of course, be postponed till having seen some real
phonetic data on the language - if it turns out that /e/ and /o/ simply never
have high allophones ever, well, typology be damned. But languages with small
vocalic sets often allow their vowel phonemes to cover alot of space in the ol'
tetragon, and if [i] and [u] are valid realizations of /e/ and /o/, there would
seem to be little reason not to reanalyze them as /i/ and /u/.