Re: THEORY: phonemes and Optimality Theory tutorial

On Sat, 25 Nov 2000, And Rosta wrote:

> > I would not explain that as "extra complexity". I would explain that as
> > perceptual saliency. The perceptual distinction between A and I, U is
> > greater than the distinction between A and E,O and E,I; O,U. Strong
> > syllables allow for enhanced perceptability when compared to weak
> > syllables; therefore, strong syllables can support less acoustically
> > distinct forms than weak syllables can.
>
> That's certainly food for thought. It is a very good functional explanation,
> but I can't decide whether it works as a formal explanation and whether at
> a formal level the two analyses wouldn't come down to the same thing.
> For example, you could formalize your observation along the lines of mine,
> and then explain the 'complexity' of complex segments as the complexity
> of introducing additional perceptual contrasts.
Here is another area that can be explained by perceptual distinctiveness,
but not (I think) "complexity". It was pointed out by Hayes and Steriade
in the introduction to a book they are editing.

The perceptual salient portion of a retroflex stop is during the onset of
the sound, ie, as the tongue curls back. On the other hand, the salient
portion of a "regular" stop is at the end, ie the Voice Onset Time. In
both cases, perceptibility is easiest when the strongest cue for the sound
is against a vowel. Thus, retroflexes should be easier to hear following a
vowel, but "regular" stops are best heard immediately preceding a vowel.

Final devoicing is very common in languages of the world. Under the
reasoning advanced here, this is because the distinction between voiced
and voiceless is more difficult to perceive without a following segment.
On the other hand, retroflexes tend not to neutralize word finally
followng a vowel. They neutralize word intially instead. Again, this is
because their perceptual cues are less salient in this position.

If this is correct, then perceptibility cues should be a driving force in
Phonology. This, in fact, is the approach taken by Steriade and (to a much
lesser extent) Hayes.

Marcus