What is Optimality Theory?
If you’re not a linguist and/or you have never heard of OT, that’s alright! This post will introduce you to the important concepts of OT so you can understand what’s going on on this blog. :D
So, basically, Optimality Theory is a theory of linguistics that uses a set of constraints (Constraint Hierarchy or CH) to explain the phonology of languages. This set of constraints is the same for every language, and the way that each language’s phonology differs from each other is simply in the order or rankings of each individual constraint. (i.e. For Constraints A, B, and C: Language 1’s CH could be C » B » A & Language 2’s CH could be B » A » C). Within this set of constraints, there are constraints that directly interact with each other (perform or promote contradictory situations) and because of this, those constraints are ranked in different stratum with respect to each other, while constraints that do not directly interact with each other can be on the same stratum as each other (this will become apparent in my examples).
So now that you’ve got the idea of the CH down, let’s move on to the other aspects of OT.
The CH is no good if it doesn’t have candidates to weed through to pick out the correct output. There’s a feature of OT, called GEN, which basically takes the input phoneme(s) and alters it in every possible way (little changes, big changes, etc), and then all these different output forms are the set of candidates that is weeded through, by the CH, to come upon the correct output form.
Now, you should be asking yourself, how does this list of candidates and constraints interact with each other? Let me tell you! :)
Each candidate is essentially scrutinized by each constraint, one at a time, in order going through the CH. If it does not violate that constraint, it moves on to the next constraint, if it does violate it (and some constraints allow candidates to violate them multiple times) they receive a violation mark (*) and do not move forth down the CH. Each candidate also is tested against each other (b/c how else would the correct one be picked in the end?). So if there are just two candidates: 1 and 2; and two constraints A and B…
/input/ | A | B |
1 | | * |
2 | *! | |
In this situation, candidate 1 wins over 2, because 2 violates constraint A in the first stratum. Since 1 wins in the first stratum, the first candidate’s violation in B is irrelevant to determining the outcome since the ranking A » B means that whatever happens in A takes precedence over what takes place in B (and so forth for more constraints).
(OH! I totally forgot to add… the “!” after the * denotes that that’s the fatal violation, meaning that’s the violation that makes that candidate lose to the other(s)… Note how candidate 1’s violation in B does NOT have that… that’s because that violation isn’t the one that’s making it lose (though, granted it’s not losing at all anyway… sooooo jaja))
If A and B are switched, then 2 would win.
I shall work at explaining it a little more and better later on… but until then, this should suffice :) - also, I’ll put in a link to a page that should better explain this, and will show examples of what OT looks like.
You may also see me reference Harmonic Serialism, or HS for short. This is essentially a modified version of OT that takes a more generative approach to what OT does. But I won’t get into that until I need to!