"di?nt" (with glottal stop)

[Arnold M. Zwicky]

On Nov 15, 2004, at 10:48 AM, Laurence Horn wrote:

> At 12:38 AM -0500 11/15/04, Alice Faber wrote:
>> --On Sunday, November 14, 2004 10:57 PM -0500 "Mark A. Mandel"
>> <mamandel at LDC.UPENN.EDU> wrote:
>>
>>> A question from a friend in Boston:
>>>
>>> ---------- Forwarded message ----------
>>>
>>> I was on the Green Line, and there were a couple of African-American
>>> teenage girls talking loudly to one another (they were about twenty
>>> feet
>>> away, almost out of sight, but I could hear every word they said).
>>>
>>> And it occurred to me as I was listening that there's a linguistic
>>> artifact that I've only heard from urban African-Americans my age or
>>> younger, mostly girls.  It's a sort of glottal stop used in place of
>>> t or
>>> d; eg. di-unt instead of didn't.  Do you know where this might have
>>> come
>>> from?
>>>
>>> =====================
>>
>> I've heard this often from callers to NY sports talk radio who have no
>> overt AAVE phonology (other than this, if it *is* AAVE).
>>
> Indeed.  And I've also heard it in Connecticut, as the counterpart of
> the phenomenon we discussed a couple of months ago in voiceless
> examples, as in "kitten", "mitten", "New Britain", etc.  But this is
> perhaps restricted to this particular voiced context; I don't think
> I've heard it with "hidden", where "a [hI?In] message" would be
> perceived as the implausible "hittin' message".  Arnold can probably
> explain all this (if anyone can).

background: most (though not all) u.s. speakers have a glottalish
(glottalized in some way, or just a glottal stop) variant of /t/ after
an accented syllable and before syllabic n (mitten, cotton,...), and
also in syllable-final position (hat, hothead, fitness,...).  these two
phenomena are probably related: /t/ in an accented syllable before
syllabic n is syllabified with the preceding syllable, and so gets
syllable-final allophony.  (in contrast, /t/ after an accented syllable
and before other syllabics  is [depending on your religion]
ambisyllabic or syllabified entirely with the following syllable, and
so gets a voiced allophone, [d] or a voiced tap or whatever.)

meanwhile, syllable-final /d/ in american dialects is weakly voiced.
in some dialects -- notably african-american, but also some southern
and south midlands varieties (i don't understand the
geographical/social pattern, but the variant does seem to be spreading)
-- syllable-final /d/ is in fact glottalized, in one way or another,
and as a result is fully devoiced.  (this is a subtle racial indicator,
largely unrecognized by everybody involved, in the speech of african
americans who are otherwise fluently bidialectal.)

now to negative inflection for bases ending in /d/ (didn't, couldn't,
wouldn't, shouldn't, hadn't).  while there are good arguments (due to
geoff pullum and me) that these are single inflected words, they are
nonetheless transparently analyzable into two meaningful parts, an
auxiliary verb ending in /d/ and a negative element /nt/, and so can be
seen as subject to a syllable division reflecting this analysis, with
syllable-final glottalization in the first part as a result.  other
occurrences of /d/ followed by syllabic n, even in "wooden" and
"hidden", are not so clearly analyzable.

that's an account of the historical origin of a medial glottalish
element in "didn't" etc.  i suspect, however, that once the glottalish
variants were produced, they became lexicalized: in effect, "did" in
"didn't" etc. were interpreted as having an alternant ending in /t/
rather than /d/.  (this would allow, by the way, for different
developments for different auxiliaries, and even for some modest
extension to other analyzable words, like "wooden".)  lexicalization is
suggested by the fact that the correlation between syllable-final
glottalization (in general) and medial glottalization in  contracted
negatives looks very poor to me.

a medial glottalish element in "didn't" is widespread (but not
universal) in AAVE, or at least the AAVE of the last forty years or so,
and is attested in lots of places (mostly deep south, but now south
midlands and elsewhere) for white speakers; my impression is that for
southern speakers, black and white, it is at best very modestly related
to social class.

but surely this one has been studied...  and probably discussed in the
pages of American Speech.  does anyone remember, or is anyone able to
figure out how to search for it?

arnold (zwicky at csli.stanford.edu), once again ignorant