Q: speech synthesis and historical phonology

[Alice Faber]

----------------------------Original message----------------------------
Jacob Baltuch asks:

>Has anyone ever seen any work drawing both on speech synthesis
>and historical phonology? Do historical phonologists show any
>interest in results and technologies from the field of speech
>synthesis? Can any see any use in them at all for their field?
>(If yes, I'd be interested in examples)

Do you mean speech synthesis in particular, or instrumental phonetics in
general? If the former, I'm not sure I see the application, except insofar as
perception experiments using synthetic stimuli might shed light on particular
phonetic confusions that might have in a particular instance given rise to a
sound change.

Speaking only for myself, now, I started off as a historical phonologist,
working primarily on Semitic languages. As a result of trying to limit myself
to phonetically plausible reconstructions of sound systems and phonological
changes, I found myself very interested in questions of phonetic naturalness,
and would have to describe my current work as more phonetics than historical
phonology. I see a continuity in my work that others, perhaps, might not see.
I view this kind of use of phonetic data in reconstruction as comparable to
use of typological information; if you're reconstructing a phonological
inventory that doesn't look like any attested modern inventory, you'd better
have some pretty good evidence for bucking the typological trend.

There are also phoneticians who see in the inventory of sound changes that
have demonstrably taken place a "laboratory" for testing theories of
phonetics. I'm thinking here primarily of Ohala, though he's certainly not the
only one to make heavy use of such evidence.

This is all relevant, of course, for the near merger/meat-mate/Basque
discussion, which I haven't commented on because it overlaps so heavily with
work that I should be putting the finishing touches on for submitting for
publication. However, it's worth pointing out that in some of the instances
under discussion, acoustic measurements have been made and perceptual
judgements solicited. I know that some Histling-ers are familiar with much or
all of that literature, but I'd be glad to provide references for anyone who
isn't. The short version is that the results are not unambiguous and that, as
Labov and others have shown, native speaker intuitions about whether they
pronounce two words differently or not aren't always reliable. I'd love to see
some instrumental data from Basque dialects in which the putative fricative
merger took place; in my view, that's the most reliable way of knowing for
sure that there was in fact a merger. Of course, the fact that the relevant
dialect might have been spoken several hundred years ago would make that
somewhat difficult, alas.

[posted to histling & (I hope) mailed to original questioner]

Alice Faber
Haskins Laboratories
http://www.haskins.yale.edu