THEORY: [CONLANG] The Language Code (take 4)

Tom Wier:
> Quoting Dirk Elzinga <dirk_elzinga@...>:
> > On Saturday, June 14, 2003, at 07:40  AM, And Rosta wrote:
> > > To get English down to 9 vowels requires a degree of ruthless
parsimony
> > > that would be highly controversial. At a purely descriptive level
> > > (i.e. in the Code's spirit of providing a flavour of the language
> > > rather than an analysis of it), I would say English has 19-22 vowels
> > > (for my accent; 19 definites + 3 marginals). Student textbooks and
> > > modern British dictionaries would use 20. That figure of c.20 better
> > > reflects the typological eccentricity of the English vowel system
> > > & the fact that it is responsible for most dialectal variation
> >
> > 20 vowels?! Are they all distinctive? I agree that 9 vowels is too few
> > (I don't remember where that came from), but I can only get 13,
> > including diphthongs (14 if I include [O], which I don't have
> > natively). Of course, my dialect is rhotic, and I don't treat coda-r as
> > forming a diphthong with a preceding vowel nucleus
>
> My impression is that this tendency in Britain to treat (non)rhotic
> diphthongs as unit phonemes rather than two distinct segments
> is a cultural difference between British and American linguists,
> rather than differences in facts or analysis per se. (I base
> this impression on Iggy Roca's _Phonology_.)
I haven't seen that book, but I think it is more than a purely cultural
difference. For one thing, GenAm does have a less weird vowel system,
on the whole. But more importantly, the American tradition seems to
be mainly motivated by the question "What is the smallest number of
vowel phonemes consistent with English and reflecting the systemic
patterning of the vowel system?", whereas the British tradition asks
"What phonemes do we get if we apply the minimal pair test, and
take surface realization into account?" Even British English can be
analysed as containing only 6 vowels, a e i o U V. But while at some
level of analysis that might be right, it fails to capture the fact
that, say, the realization of /ai/ is unrelated to the realizations
of /a/ and of /i/.

--And.

Reply