Saturday, March 13, 2010

replacing Chomsky's SPE rule 43


D: My apologies for yesterday's entry! It was a mess. I was changing my thoughts on the fly.

I'll try again.

http://en.wikipedia.org/wiki/The_Sound_Pattern_of_English

D: SPE uses a simple number system to denote stress. By simple, I mean that the #s assigned to syllable stress can be nada, and 1,2,3 etc. 1 denotes the most stressed syllable. The higher the number, the less stressed the syllable.
Syllables that are not stressed at any stage may have their vowel reduced. Schwa is a common example of a reduced vowel.

"Relaxation" is a good example. We first examine the stem "relax" and find that the second syllable is stressed. There are rules for this, but they can be found in SPE, and are not relevant to my observation. So re-lax is initially stressed as 2 - 1. So the second syllable is stressed. Rules regarding the affix "-Ation" mean the affix is more stressed than the stem. "Ation" is considered a single entry. So we get re-lax-Ation as the three 'syllables'. Stress mid-way through the rule process is 3-2-1.

And then we encounter "Rule 43". It works, sure. It just doesn't feel elegant. I know that is a silly thing to be concerned about.
It feels like an afterthought, a stop-gap measure.
Basically, Rule 43 says there all other syllables other than the most stressed one are de-stressed an additional step. This means re-lax-ation becomes stressed as 4-3-1.
OK.

----
My proposal.
I propose using an integer number system instead. This involves both positive and negative whole numbers. For example, -2, -1, 0, +1, +2.
In my proposal, a stress never changes once assigned. The higher positive value is the most stressed syllable.
The result of Rule 43 is instead applied as the very first step. For example, take "relaxation" and use it as to test my proposal.
We will use my revised system. The initial system resulted in inappropriate stress distance between syllables. It did not duplicate Rule 43.
The revised system takes note of the -Ation affix. The stem is assigned to the left of zero. NO syllable is assigned to zero. By this process, we create the distance of 2 stress levels away from the primary stressed syllable as per rule 43.

(My initial proposal involved looking at "relax" - the stem - in isolation initially. It assigned" re"- to -1 and "lax" to +1. But this does not recreate the result of Rule 43. It would result in "lax" only one stress away from the affix stress. I suppose we could assign a value to the affix of '2 more than the nearest stress', but this seems no more elegant than rule 43.)

My integer system NEVER changes a stress # assigned to a syllable. Ergo "-Ation" occupies the next higher unoccupied slot. In the revised system, this will be +1. In the original system, it would have been +2, since "-lax" would have been +1.
This permanent # assigned to a syllable reduces demands on working memory.
Because we always skip zero on the integer system, we duplicate the effects of Rule 43 in my revised system.

D: In my revised system, only values of -2 or lower can have the vowel reduced. In the initial system, that would have been true of any negative value.

OK, here is a caveat. I only started reading SPE a week ago, and am presently at page 60.
My proposal is only a tentative one, based on what I have read so far.
I'll likely tweak it.
For example. I don't quite grasp how stress 4 3 1 is different from 4 2 1 right now. I need to learn more.

D: sadly, this blog font screws up any attempt at regular spacing. I'm not sure I can access Courier Font for that.
I'll scan more examples and some pics of the rule steps tomorrow. I gotta work all day, starting in an hour.
Cheers.
D

3 comments:

Unknown said...

If you need to apply stress in any degree to relaxation, you aren't doing it right!

Dino Snider said...

LOL - thanks.

Haha and the discussion of a "deep structure" for the following ambiguous sentence is amusing.
"He liked exciting women." <:

dino snider said...

Well I've gone as far with this thought-experiment as I can. Conceptually, I had to use the idea of atoms with positive, negative and neutral charge. Graphically it ended up a mess. But this was all hindered by the terrible mess of editing (lack thereof) that is SPE. I eventually gave up even trying to cite examples from SPE. It is an unmitigated nightmare. I'll post it tomorrow.