[tied] Abstractness (Was Re: [j] v. [i])

From: Richard Wordingham
Message: 22766
Date: 2003-06-06

--- In cybalist@yahoogroups.com, "wtsdv" <liberty@...> wrote:
> --- In cybalist@yahoogroups.com, "Richard Wordingham"
> <richard.wordingham@...> wrote:
> >
> > Apart from the side issue that in your dialect [N] seems still
> > to be an allophone of /n/, Europeans are uneasy about such
> > biphonemic analyses of English long vowels and diphthongs.
>
> But why? Is not a diphthong by definition a sequence of _two_
> sounds? Is not the analysis of a long vowel as the geminate of
> its short equivalent about the most natural assumption one can
> make?

If the equivalences are good, it does make sense to pair short and
long vowels off. But the equivalences may not be good, so pairing
them off may simply be a convenience. A diphthong is essentially a
changing vowel sound, and there can be problems in identifying the
precise end points well enough.

> > It's not an issue of dialect.
>
> But it must be an issue of dialect. Every dialect should get
> its own analysis. If a "shadow" segment, which is based on
> diachronic analysis but not actually present in a strictly
> phonetic analysis, may not be invoked, then no diachronic
> consideration may be, nor a consideration of other dialects'
> systems of contrasts either.

I think one needs to be pragmatic. A dialect is a collection of
idiolects, and even these vary over time, and, even more
inconveniently, with social situation. An analysis as vowel
sequences can work well, e.g. in Finnish, but work less well when
there are fewer contrasting sequences.

> My dialect might, in fact I
> know that it does, have a different system of contrasts than
> that of British English.
>
> In my dialect /e/ and /i/ occur only before /j/, /o/ and /u/
> only before /w/. So I still don't see how the following
> representation, grouping [E] and [e] in /e/, [O] and [o] in
> /o/, [U] and [u] in /u/, and [I] and [i] in /i/, threatens
> to reduce English's phonological system to binary code:
>
> beat /bijt/

I think you've just suggested that
beat /biit/
would be more natural!

> bit /bit/
> bait /bejt/
> bet /bet/
> suit /suwt/
> put /put/
> boat /bowt/
> toy /toj/
> buy /baj/
> bow /baw/
> yeast /jijst/
> east /ijst/ or /?ijst/
> woo /wuw/
> year /jir/
> ear /ir/ or /?ir/

One measure of the suitability of the analysis would be how much
effort had to be devoted to defining the allophonic variation of the
vowels in these contexts. The diphthongs you write /aj/ and /ou/ can
be particularly troublesome. There can be significant difficulty in
identifying a consistent starting point for these diphthongs.

Is it obvious why you group /e/ with /E/ rather than with /I/?

> saying /sejing/, or /sejiN/ if the concept of
> morpheme boundry is disallowed
> seeing /sijing/, or /sijiN/
>
> Is the more open pronuncation of /j/ in these
> last two, as opposed to that of the first /j/
> in 'yeast', part of what causes a problem with
> this sort of analysis for some? I don't see
> why.

Well, there is the syllabification issue!

Incidentally, I suspect we should move this discussion off Cybalist
onto Phonet, which was set up for discussions such as this when they
wondered too far off the core topic.

Richard.