Sylvia Sotomayor and I went to a faculty talk by Mary Paster at
Pomona College today. As a faculty talk (i.e. a talk by faculty to
other faculty from different disciplines [or anyone else who wants
to attend]), it wasn't as technical as a usual linguistics talk, but
she did present a bit of data I found fascinating and wanted to

Gã, a Kwa language spoken in Ghana, is a level tone language
with high and low tones. Contour tones can emerge for various
reasons. For example, imperatives are associated with a high
tone. When words with an inherent low tones are put in the imperative,
the result is a rising tone:

High Tone: è-dú "he cultivates" / dú "cultivate!"
No Tone: è-bà "he (somethings)" / bá "(something)!"
Low Tone: è-wò "he (somethingelses)" / wǒ "(somethingelse)!"

(Sorry, I don't remember all the glosses; there wasn't a handout.)

The perfect, though, is associated with a low tone. For low tone
and no tone words, this works as one might expect:

No Tone: è-bà "he (somethings)" / é-bà "he (has somethinged)"
Low Tone: è-wò "he (somethingelses)" / é-wò "he (has somethingelsed)"

(Note: Not sure why the subject marker has a high tone in the
perfect, but, crucially, it does.)

For high tone words, though, the second high tone is downstepped--
rather than a contour tone. So:

High Tone: è-dú "he cultivates" / é-!dú "he has cultivated"

I would have expected a contour tone. I've seen downstep in
level tone languages, and it usually occurs when either (a) the
language doesn't allow surface contour tones of any kind, or (b)
the language doesn't allow that particular combination of contour
tones (for example, if it allows falling tones, but not rising, what
would be a rising tone will be realized as downstep)--or, now that
I think about it, (c) the language doesn't allow spreading in that
direction, or a floating tone can't be associated with any available
tone bearing unit (because, for example, the language has a
restriction on contour tones in short syllables). In this case,
the language allows contour tones, allows spreading, allows
contour tones on short syllables, *and* allows that particular
contour tone (LH)--and yet there's still downstep!

I talked to Mary after the talk, and she guesses that the reason
the tone melody doesn't surface as a contour tone is because
the low tone is morphological, rather than phonological. It seems
to be allowed for imperatives, though, but in that case, there would
be no other way to realize the imperative (there's no guarantee
that something will follow it, and there might be nothing to upstep).

Anyway, if it's the case that the phonological rules work differently
for this particular morphological category, I think it's further evidence
for the kind of unique morpho-phonological level we discussed
awhile back regarding overlong vowels. In that case, even though
there are minimal pairs in Estonian, Dirk argued that the overlength
was not phonemic. I think the downstep in Gã is similar: there will
be minimal pairs, but even so, it looks like the downstep phenomenon
is not phonemic.

"A male love inevivi i'ala'i oku i ue pokulu'ume o heki a."
"No eternal reward will forgive us now for wasting the dawn."

-Jim Morrison

LCS Member Since 2007