LINGUIST List 8.703

Mon May 12 1997

Disc: Functionalism

Editor for this issue: T. Daniel Seely <seelylinguistlist.org>

Directory

1. Wojcik, Richard H, RE: Question on Functionalism

Charles Reiss commented (LINGUIST 8.689.2):

> 1) It is true that lots of fast speech phenomena reduce 'complexity':
> clusters are simplified, consonants assimilate, etc. However, I would
> assume that a standard analysis of English must posit a constraint against
> clusters of voiceless stops in onsets (presumably because they are hard to
> say and/or hear). Yet the standard pronunciation of 'potato' in fast (i.e.
> NORMAL) speech is [pteDo], or something similar. So it seems like the
> functionalist conclusion must be that when we are trying least, we
> sometimes do what's easy (according to the principles of markedness) and
> sometimes do what's hard (according to those same principles).

This is a very good point, as it relates to generative markedness
theory. Stampean Natural Phonology, which treats linguistic processes
as direct constraints on speech production, revels in the
contradictory motivations of processes. Constraints on pronunciations
may or may not apply to phonetic targets that are byproducts of other
phonological processes. In the case of an initial [pt] target, the
fortitive epenthesis that destroys the cluster would tend not to be
triggered on the output on a lenitive syncope. For Natural Phonology,
the fortitive epenthesis gives rise to the 'intuition' that you can't
have initial /pt/ clusters. If you could, then you would have
suppressed that misarticulation during early acquisition of English.
The fact that you can't pronounce the cluster in no way means that the
cluster is banned as a side-effect in the pronunciation of some other
phonetic target. Traditional phonological theory captures constraints
of this sort as static 'phonotactic' conditions on clustering, not as
behavioral constraints on linguistic articulation.

> 3) If, as Kirchner discusses, lots of phonetic distributional patterns can
> be explained on the basis of perceptual robustness, why do we need to
> build this into the grammar? Contrasts that are hard to hear, are less
> likely to be acquired. It is otiose to offer a grammatical, as well as a
> perceptual/acquisition-based account of the same phenomenon.

Again, I think that this criticism makes sense in a generative
framework. If you have a theory that treats grammatical processes as
direct constraints on behavior (Natural Phonology's position on
phonological processes), then you need to ask why constraints that are
identical for all humans survive or fail to survive the acquisition
process of individual speakers. I think that Natural Phonology
explains this rather well, but I don't think that many linguists have
understood the explanations.

Norval Smith (LINGUIST 8.689.4) wrote in reply to Geoff Nathan:
>> 	Lastly, let me make a plea for a historical perspective on this debate.
>> Although Robert Kirchner has been happily building these notions into OT,
>> he is well aware (and so I am hereby reminding others) that these concepts
>> originate WAY before the advent of generative grammar, finding their roots
>> in the work of nineteenth century phonologists such as Baudouin de
>> Courtenay, and in early twentieth century phoneticians such as Sievers,
>> Jespersen and Grammont. Sorry, but I get a little tetchy about reinventing
>> wheels...
> No-one would want to deny that early phonologists like de Courtenay and so
> on had many acute insights. However they were not able to produce a very
> satisfactory theory of phonology. Late 20th century phonologists such as
> the natural phonologists also had some good ideas - as well as doing their
> own bit of wheel-reinventing (!) - but were also unable to integrate them
> into a satisfactory theory of phonology.

All Geoff seemed to be saying was that we need to look at historical
antecedents and trends. I agree that Baudouin had many "acute
insights". After all, he coined terms like "phoneme", "grapheme",
"alternation", etc., in something like their modern usage. Indeed,
his acute insights can easily be said to have laid the entire
foundation of modern phonological theory. That aside, I would like to
add that Natural Phonology did very little wheel-inventing, and it did
pay proper respect to its historical antecedents of the 19th and 20th
centuries. I would only point out that we are still in the "late 20th
century"! ;-) It is too bad that Natural Phonolgy is so easily
dismissed as "unsatisfying", but that is precisely the same objection
that I would have to all of the alternatives to it. ;-)

The relevant point about Natural Phonology here--what distinguishes it
from other approaches to acquistion--is that language learners acquire
pronunciations by manipulating direct constraints on speech
production, not by applying theoretical analyses (a la "little
linguists") to articulations. The appealing thing about that approach
is that it is identical to the experience that we have all had in
acquiring foreign phonological systems. Once you decide what to
pronounce, you are afflicted with hard-to-control misarticulations of
it. The acquisition process consists of bringing those
misarticulations under control, or finding some level of (adult)
comfort with them. It does not consist of finding the single best
"grammatical analysis", as calculated on the basis of markedness
theory.


Rick Wojcik Senior Principal
Scientist
Boeing ISS R&T Natural Language
Processing
425-865-3844 (fax:425-865-2965) PO Box 3707, MS 7L-43
richard.h.wojcikboeing.com Seattle, WA 98124-2207
- --Opinions expressed above are not those of The Boeing Company----