Date: Mon, 10 Apr 1995 10:53:13 -0400


Subject: Peripherality

Terry Irons asks, with regard to the Southern Shift:

Could you clarify for me what peripheral/nonperipheral mean in acoustic

and articulatory terms?

As a matter of fact, no. I don't mean to be flip. It's just that the

peripheral/non-peripheral distinction, like the tense/lax distinction, has no

empirical basis. They're both ways of labeling two sets of vowels that we all

*know* are different. Labov, in his new book, has lots of diagrams showing

peripheral vs. non-peripheral vowels in terms of the boundaries of the vowel

quadrilateral. Ocke Bohn (of the University of Kiel) recently gave a talk here

at Haskins in which he appealed to the same peripheral/non-peripheral

distinction for very different purposes, to explain Pat Kuhl's

magnet/prototype effect (in a nutshell, it's harder to hear small differences

between vowel tokens that are near some posited prototype than it is to hear

differences of comparable physical magnitude that are not near the prototype).

Giving theoretical support to the notion of peripherality would depend on

providing a theoretical interpretation of the vowel quadrilateral. In intro

books, we talk about it terms of the position of the highest portion of the

tongue. However, physiological studies (x-ray, etc), especially those by

Sidney Wood, have shown that this is not an appropriate way to describe vowel

articulation. In acoustic terms, there's a reasonably good fit between the

vowel quadrilateral and F1/F2 plots of acoustic measures (especially if you

use a logarithmic scale and plot F2-F1 rather than simply F2). But it's really

unlikely (in my opinion, others may disagree) that speakers produce and

understand speech in terms of formant frequencies (we understand people with

very different sized vocal tracts...).

As I said, the peripheral/non-peripheral distinction is (nearly) isomorphic to

the phonological distinction between tense and lax vowels. You'll read in

various places that tense vowels are produced with greater articulatory force

than their lax counterparts. The problem with this is that the one attempt to

test this by using electromyography to measure electrical activity in tongue

muscles found no consistent relationship between vowel type and activity; for

some speakers, /iy/ had activity of longer duration than that for /ih/, as if

there was activity for the nucleus and activity for the glide separately.

(This study was by Freddie Bell-Berti and Larry Raphael).

A problem with the tense/lax opposition is its replicability. Most people who

appeal to it agree that the nuclei of beat, bait, and boot are tense, and

those of bit, bet, and put are lax. However, there is much less agreement

about the rest of the vowels: bat, but, pot, bought, boat. They're clearly

partitioned in English between those that can be the nuclei of stressed C(C)V

monosyllables (so long as you ignore duh and the Knights who say "nih").

Calling the raising of /ae/ in MAN "tensing" implies that it's lax, but it's

still peripheral, no?

Anyway, this has all been a long way around to justify my claim that there's

no real theoretical motivation for peripheral/non-peripheral. That doesn't

mean it's not a real partitioning of the vowels. It just means that we haven't

figured it out yet.

I am unaware of any principles of sound change. All I am familiar

with are hypotheses within the context of certain paradigms and then

certain facts.

Perhaps you disagree with Labov as to what constitutes a principle...

Alice Faber