Tuesday, October 02, 2007

A sub-optimal output

Several weeks in a practicum I had suggested that a very basic introduction to Optimality Theory might be a worthwhile for the few students who might go on to study more linguistics. It seems to me that if the OT table makes enough sense that an introduction could work very easily. The general reception was cautious disagreement. It was definitely disagreement but it was cautious. I imagine the only thing that held people back from calling it a stupid idea was a regard for my feelings.

So I did a presented a quick mock lesson in class today. The assignment was to introduce a topic as if I was teaching an intro to linguistics course. I introduced OT as a model for the way we might make any decision--using the example of purchasing a car. I mentioned very highly ranked constraints (such as 'cannot explode on ignition') and lower constraints (such as 'must be red'). That part went fine.

Then I tied it into phonology where the names of the constraints had to change (unless there is risk of some word exploding on utterance--Monty Python comes to mind). This is where my lesson fell apart because I had made some assumptions about what students might ask. I never assumed they would ask why faithfulness constraints should be expected. And my fellow students put me on the spot asking 'why should we assume that nothing will be added to an underlying form?' and I made the fateful mistake of trying to argue that faithfulness constraints function on principle. As if we have a reason for expecting faithfulness constraints to play an important part in language without seeing them in action. How do we know that DEP-IO and MAX-IO are ranked somewhere? Only because that would explain some of what we see.

How do we know that an output form will in some ways correspond to the input? Because we see places where that constraint makes sense. Not because it should correspond. And obviously it doesn't always.

Two things you don't want to hear from a professor after you've done a presentation:

  1. As a student I was willing to accept what you said. As a phonologist I have some issues.
  2. It was certainly brave of you to try.
I'm not sure which one is worse.
_

5 comments:

  1. Awww, hang in there. Rest assured if *I* tried to give ANY kind of presentation about OT, it would be 20 times worse than whatever you could produce. Also, I think it is a topic that intro students can understand...it's more intuitive than most generative approaches to syntax, at any rate :)

    Grad school's hard.

    ReplyDelete
  2. I made the fateful mistake of trying to argue that faithfulness constraints function on principle. As if we have a reason for expecting faithfulness constraints to play an important part in language without seeing them in action. How do we know that DEP-IO and MAX-IO are ranked somewhere? Only because that would explain some of what we see.

    But we do have a principled reason (within the OT framework) for assuming the existence of faithfulness constraints a priori. If there were no IO faithfulness constraints, then the input might as well not exist. All utterances would be neutralized to the most phonotactically harmonious possible surface form. So if our model is going to include a lexicon that is at all relevant to the determination of the output, we need faithfulness (or something like it).

    ReplyDelete
  3. Yes. We would see a change. And with the knowlege that there is some stability--ie analysis of the output--we then see that faithfulness constraints are at work.

    Granted it is one of the first observations we make. But it's the analysis of data (even just cursory) that leads us there--not a mere principle. Right?

    Something about this reminds me of the voice onset time issue with obstruents. Hmmm...

    (Thanks pc. This crap is hard you know.)

    ReplyDelete
  4. In my opinion, OT wouldn't be nearly so menacing if it weren't for the atrocious names given to the constraints.

    I think I'm imagining the correct phonologist your quotations, and I can even hear the intonation in her voice. I got some comments like these when I was working on my thesis--I must say, she certainly has a gift for criticizing kindly.

    ReplyDelete
  5. One time Hammond responded to the person who present before me with something like "porque estas repitiendo esta mierda"...let me just say after that I was really freaked out about presenting...

    I'm sorry your prof didn't have the most wonderful response to your presentation...

    Sometime you'll have to explain OT to me...in a really dumbed down version because I have no clue and I keep hearing the term!

    ReplyDelete

Thanks for reaching out.

You can also contact me at wishydig[at]gmail[d0t]com.