Matthew Gordon gordonmj at MISSOURI.EDU
Thu Aug 28 21:10:04 UTC 2008

I looked at the spectrograms for the m-w files for 3 sets of triplets:


Each triplet appears to be spoken by the same person, but I can't be sure.
I didn't bother taking formant measurements b/c at least one vowel in each
set lacked a steady state. Just eyeballing the formants, though, it's pretty
clear that the F1s of the <ing> items are closer to those of the /I/ items
(win, sin, kin) and the clear /i/ tokens (ween, scene, keen) have lower F1s
than the others.

With F2 there's more variability in the contour (due to consonantal context)
but it looks like the <ing> tokens generally have F2s that are in between
those for the other classes (at least on average).

So, the <ing> items appear to be produced at roughly the same height as
(other) /I/ words but are slightly fronter than /I/ words. Still, they're
clearly not as front as /i/ words. And, of course, this is just going off of
the couple of speakers recorded by m-w.

Anyway, it seems to me that the real question here isn't whether the vowel
of wing, sing, king, etc. is PHONETICALLY identical to that of win, sin, kin
or to that of ween, scene, keen. It obviously isn't phonetically identical
to either. The only interesting question (for me, and even I'm rapidly
losing interest) is whether the <ing> vowel is PHONEMICALLY part of the /I/
class or the /i/ class. Historically there's no question, right? These are
short i's and that class gives us /I/. Still, if we take seriously
native-speaker intuitions, doesn't the fact that many people identify the
pre-velar nasal vowel as the same as other clear cases of /i/ argue that the
allophone has been reallocated to /i/?

On 8/28/08 9:47 AM, "Tom Zurinskas" <truespel at HOTMAIL.COM> wrote:

> We can reduce variability by using a standard pronunciation source like
>, a very good one.
> I've got praat.  Somebody tell us how to copy the voicing and compare
> waveforms and formant numbers.  But it's obvious to my ear that the vowel
> sounds for "i" are different in the following,
> ching vs chin
> wing/wink vs win
> king/kink vs kin
> think/think vs thin
> It's more than an allophone thing.  It's a full phoneme shift.
> Here's a thought.  Suppose you take the word "sheep" ~shee.  That's a true
> long e, right, as pronounced in  And you replace "p" with "ng" to
> make nonsense word ~sheeng (with a true long e).  Then say to other folks:
> "Spell this word - ~sheeng."  I predict they would spell it "shing" and say it
> rhymes with all the other "ing" words, like wing, sing.
> Tom Zurinskas, USA - CT20, TN3, NJ33, FL5+
> See - and the 4 truespel books plus "Occasional Poems" at
>> Date: Thu, 28 Aug 2008 08:59:18 -0500
>> From: gordonmj at MISSOURI.EDU
>> Subject: Re: Chinglish
>> ---------------------- Information from the mail header
>> -----------------------
>> Sender: American Dialect Society
>> Poster: Matthew Gordon
>> Subject: Re: Chinglish
>> -----------------------------------------------------------------------------
>> --
>> You can't ever take variability out of the equation. Each human vocal tract
>> is unique. Your [i] will have different acoustic characteristics from mine
>> due simply to the fact that they're played on different instruments. In
>> fact, there is intraspeaker variation too; each time you pronounce an [i] it
>> will differ acoustically from the last time you pronounced one.
>> On 8/27/08 7:57 PM, "Tom Zurinskas"  wrote:
>>> I would like to trust the numbers to identify vowels. I've played with Praat
>>> (I assume the "aa" is pronuounced "ah" like "Saab" (foespeld ~aa in
>>> truespel).
>>> I'm not practiced at it but I find it hard to determine vowel identity by
>>> numbers or wave forms. If this is possible with practice, it would be a good
>>> thing, taking human bias and variability out of the equation.
>>> Tom Zurinskas, USA - CT20, TN3, NJ33, FL5+
>>> See - and the 4 truespel books plus "Occasional Poems" at
>> ------------------------------------------------------------
>> The American Dialect Society -
> _________________________________________________________________
> Get thousands of games on your PC, your mobile phone, and the web with
> Windows®.
> ------------------------------------------------------------
> The American Dialect Society -

The American Dialect Society -

More information about the Ads-l mailing list