Vowels [was: Finnish KASKI]

From: Piotr Gasiorowski
Message: 53873
Date: 2008-02-21

On 2008-02-21 07:18, Patrick Ryan wrote:

> But someone who uses /a/ instead of /aw/ for <Sean> will be misleading a
> lot of people.

The COT/CAUGHT merger is nevertheless gaining ground in US English (in
Canada it was completed a long time ago). Speakers of a given accent
usually don't give a damn if their pronunciation confuses other people.
Speakers of British English may find it misleading that Americans don't
distinguish "ant" from "aunt", but who cares?

> I have no objection to using <a> for <o> (<knot>) or <a> for <æ> (<gnat>)
> but what symbol should be reserved for central /a/ in <father>? It used to
> be <ä>.

Outside of New England and the NYC/NJ area it's unusual for modern US
accents to contrast the vowels of the COT and FATHER sets. They have
merged almost everywhere. Since the COT/FATHER vowel may range anywhere
between [A] and [æ] depending on the local accent, it's reasonable to
use /a/ as a phonemic cover symbol, ignoring the phonetic details. In
the IPA transcription system, [a] stands for the front low cardinal
vowel, but in practice phoneticians use it for any low vowel which is
not distinctively back.

The "vowel space" underlying the IPA system is quadrangular, and points
within it are supposed to symbolise the position of the highest point of
the tongue surface. But in fact few languages have a phonemic contrast
between [a] and [A], and the "psychological vowel space" is triangular
rather than trapezoid (which is why the most common type of vowel
systems is something like {a, e, o, i, u}).

Some modern theories of phonological representation have given up
traditional tongue-position features like [+/-back] and use abstract
"elements", corresponding to three "extreme" vowel qualities (symbolised
as A, I, U), instead. Vowels other than [a], [i] and [u] are regarded as
combinations of those elements, e.g. (A, I) = [e], (A, U) = [o], () =
[&] (a neutral vowel, often symbolised as "@", phonetically
corresponding to a schwa). A combination may be regarded as "headed" if
one of the elements is dominant, e.g. (A_, I) = [E] (mid-low), (A, I_) =
[e] (mid-high), (I_, U) = [y] (high, front and rounded), (I, @_) = [I]
(centralised high), etc. ("@" is the "zero" of the system and doesn't
have to be explicitly present in the representation _unless_ it
functions as the head).

Piotr