Linguistics Stack Exchange is a question and answer site for professional linguists and others with an interest in linguistic research and theory. It only takes a minute to sign up.
Sign up to join this community
I understand how basic Optimality Theory (as applied to phonology) works, but I've never understood how it came into popularity. I'm guessing, though, that there are good reasons why it arose.
So what phonological patterns does Optimality Theory explain that rule-based phonology doesn't? (Or, what phonological patterns does OT do a better job of explaining?) In particular, I'm curious about the original motivating examples (if any) that Prince and Smolensky provided when they introduced OT.
Prince and Smolensky's original paper about OT is available for free online. The examples that they have extensive discussion of (viz. a whole chapter in the table of contents) are Berber syllabification and Lardil word-final deletion.
A main thing the Optimality Theory explains with the previous theories did not, is how different phonological 'rules' which do not seem to have something in common are actually a part of a 'conspiracy' which is meant to serve a single purpose (the prevention of a form the language phonology does not allow etc.)
Within the Optimality Theory, these restrictions are formed as constraints and allow a description which unites different processes serving the same purpose as parts of the same process.
I'm not aware of anything explained by OT that cannot be explained by rule-based phonology. The reasoning in OT is generally circular. A claim is made that there is a constraint of some sort in a certain language, it is argued that the constraint cannot be adequately described with rules of standard generative phonology, and the conclusion is drawn that this language has this constraint.
Let me review that. There is a constraint which is not describable in generative phonology, therefore this constraint exists. It may be interesting that rule-based phonology does not correctly account for everything, and it's good to know about such cases. Maybe we can find a better theory. But how does it follow that OT is a better theory? Or any theory at all?
It is a characteristic of scientific theories that they make predictions. When we find something that generative phonology can't handle, that tells that (1) generative phonology did make a prediction, and (2) it's not correct. When we find something that OT can't handle (if that is even possible), what do we do? We add a new type of constraint to OT and claim to have made a theoretical advance. Isn't that a little weird?
I am completely in sympathy with OT's contention that surface constraints are relevant to phonological structure and that this cannot be appropriately described in the original form of generative phonology. But this is not exactly news that was first brought to light in OT. Phonologists have been looking at such cases for a while -- for instance, there is Kiparsky's description of rule and rule-interactional transparency.
My own theory about transparency is that a theoretical constraint of the original generative phonology, the restriction of conditions on rule application to the inputs of phonological rules was incorrect, and not even well thought out. Generative phonology was supposed to be "neutral between speaker and hearer", so what sense did it ever make to restrict conditions on rule application to rule inputs?
I gave some detail about this in this forum in an earlier post on eliminating intermediate forms.
