Re: Redefining [a] and [æ]? or using a better vowel quadrilateral?

On Aug 9, 8:30 pm, "Peter T. Daniels" <gramma...@xxxxxxxxxxx> wrote:
On Aug 9, 6:38 pm, "ranjit_math...@xxxxxxxxx"



<ranjit_math...@xxxxxxxxx> wrote:
On Aug 9, 4:30 pm, "Peter T. Daniels" <gramma...@xxxxxxxxxxx> wrote:

On Aug 9, 4:17 pm, "ranjit_math...@xxxxxxxxx"
<ranjit_math...@xxxxxxxxx> wrote:
On Aug 9, 11:27 am, "Peter T. Daniels" <gramma...@xxxxxxxxxxx> wrote:

On Aug 9, 12:09 pm, "ranjit_math...@xxxxxxxxx"

<ranjit_math...@xxxxxxxxx> wrote:
On Aug 8, 11:32 pm, "Peter T. Daniels" <gramma...@xxxxxxxxxxx> wrote:

On Aug 8, 7:39 pm, "ranjit_math...@xxxxxxxxx"

<ranjit_math...@xxxxxxxxx> wrote:
On Jul 27, 10:38 pm, "Peter T. Daniels" <gramma...@xxxxxxxxxxx> wrote:

On Jul 27, 4:54 pm, Nathan Sanders <nsand...@xxxxxxxxxxxx> wrote:

Almost no linguist uses the IPA cardinal vowels exclusively to
represent true cardinal vowels.

As is obvious from the CD accompanying Ladefoged's *Vowels and
Consonants*, which includes a number of distinguished phoneticians'
demonstrations of the cardinal vowels, including Daniel Jones and
Peter Ladefoged, no two phoneticians, even those trained directly by
Daniel Jones (let alone those trained by his pupils), make the same
noises as either each other or him.

If there's a synthetic demonstration, two speech sythesizers would
presumably produce the same sounds.

And what would be the basis of those sounds?

Possibly an averaging of several people's speech, producing sounds
that are not articulated by any one person. Possibly a quadrilateral
with some formants being synthetic such as cardinal vowels being
exactly so many Hz apart in one formant.

And what would that have to do with sounds actually used in speech?

An averaging of accents of several people in a group would describe
the pronunciation of that group. Listening to a synthetic accent that
averages that group's accents would prescribe a pronunciation that
would be acceptable to that group. As for synthetic formants,
exaggerating the difference between [&] and [E] by making [E] an equal
number of Hz apart from [e] and [&] would help [&],[E] mergers
separate their [&]s from their [E]s. The same goes for [O] and [A.].
For example, when trying to teach a Hindi speaker an accent which,
while foreign, would be more comprehensible to Brits than is a Hindi
speaker's ESL accent:
Synthesize an [E] exactly halfway between the Indian's [e:] and a
Brit's [&] in "mat" for describing a pronunciation of words like
"met". That would help 'em not mix up their "mat" and their "met".
Synthesize an [E:] exactly halfway between the Indian's [e:] and a
Brit's [&:] in "mandible" for describing a monophthongal [E:] usable
in words like "aerate". That would help 'em not make their "aerate"
sound like "arrate".

None of that would end up sounding like a human language.

"hour" and "air" were typically [A:] and [E:] in high register Indian
English and Brits didn't seem to have trouble understanding them.
("hours" was [A:z] in high register but [A:s] in mid-register which
pronunciation carried an unfortunate connotation.) The low-register
[Vw@r.] and [e:j@r.] were considerably more difficult for Anglos to
understand.

None of which has anything to do with your proposal to make reference
noises by computer generation.

These would be reference noises for a given audience. Hindi speakers
don't have [E] or [&]; they have something inbetween and a little
retracted. So, both their "ax" and "ex" would come out like French
"Aix". To generate an [E] exactly 3/4 of the way from their [e:] to
their merged [E]/[&] would take a computer. To generate an [&] such
that the generated [E] is exactly halfway between the generated [&]
and their [e:] would require a computer too.

"Aerate" is three syllables.

The 3 syllable pronunciation in Hindi speakers' English would be
[e:j@re:t.] which tends to sound much grosser to Anglos

Nonetheless, it is the (American?) pronunciation of the word.

The Hindi speaker's /e:j/ is realized as [e:i] which is very different
from anything preceding a vowel in American pronunciation. Americans
probably have something between [EI] and [eI].

than a 2
syllable pronunciation of [E:re:t].

Which I suppose might be a weird rendition (with final stress) of the
name of the Afroasiaticist and Africanist Chris Ehret.

He's at UCLA. Even with stress, Americans wouldn't pronounce [ret]
with the first vowel in a German's pronunciation of "Beethoven" which
is the vowel I mean by [e:].
.

Relevant Pages
Quantcast