Wednesday, December 16, 2009

More Adventures in Phonology (Part II)

It's time for another exciting edition of More Adventures in Phonology!

In our class today we analyzed the problem of stress distribution in Southern Paiute. This problem can be found on p. 181 of Halle's Problem Book in Phonology. A set of 24 nonsense words with various stress patterns are given, and in question 2 one is asked to give a formal statement to account for these patterns.

Once again I found myself completely disagreeing with the analysis of my teacher (and presumably most everyone else in the class)! If I've learned one thing from this class, it's that I tend to approach phonology issues in a very different way.

The analysis given by my teacher can be summarized as follows:

(1) Final syllables are to be considered extra-metrical
(2) Trees are constructed from left to right and have two layers if the word has both primary and secondary stress
(3) Lower layers are right-headed while higher layers are left-headed

So, no. 8 paáppàA is analyzed as follows:

pa-áp-pà-(A)
** *
\/ |
*
\/

I immediately objected to such an analysis as it requires the head to be located on the right in the lower layer and on the left in the upper layer. Such an ad hoc assignment is not allowed in syntax, as far as I am aware--you can't start off with a head on the right and then change mid-course to the left.

I presume that my teacher opted for such an analysis as it allows disyllablic words (no.1 páwA, no. 5 páppA) to be analyzed in a consistent fashion: the last syllable is extra-metrical and therefore stress naturally is assigned to the first syllable. This may be the only redeeming feature of such an analysis, but considering the inconsistencies in head assignment, I believe it should ultimately be rejected in favor of the following analysis:

(1) Except in disyllablic words, initial syllables are to be considered extra-metrical
(2) Trees are constructed from right to left and have two layers if the word has both primary and secondary stress
(3) Lower layers and higher layers are both left-headed

My analysis solves the problem of arbitrarily changing the position of the head. Some people in the class objected to my specification in (1) that disyllabic words are to be excepted from the extra-metrical rule, saying that setting such a limitation would be ad hoc. In other words, if we say that either final or initial syllables are extra-metrical, we have to apply this across-the-board to all forms, regardless of their syllable count.

However, such objections can be easily shown to have no basis. Many languages have a requirement that every word contain at least one foot (the well-known "minimal word syndrome"). If we assume that Southern Paiute also has such a constraint, this explains why disyllabic words cannot have an extra-metrical syllable--they would in fact become monomoraic and would therefore be degenerate-size words. Note that my teacher's analysis violates this constraint.

A look at p. 88 of Bruce Hayes' Metrical Stress Theory lists Southern Paiute as one of the languages subject to the "minimal word syndrome." Hayes also provides an analysis of Southern Paiute on p. 266, and he seems to take the word-final syllable as extra-metrical. He also allows for head position to be changed from layer-to-layer. So now I'm confused...head really doesn't mean head in the syntax sense of the term. It is simply a place holder to mark where stress is assigned.

If phonologists want to use syntax terms such as head, etc., they need to be a little more rigorous with their definitions. The analysis I gave above for Southern Paiute still seems to me to be entirely consistent, at least with regard to the data I was presented with.

Aargh...

No comments: