Phonological Optimality Theory Constraints
A Listing of Established and Accepted Constraints with Explanation of OT
Optimality theory is a model for solving phonology problems as well as explaining variation over the languages of the world. Instead of making rules anew each time for languages based on the loose set of observations of supposed phonological rules, universal tendencies are formalized abstractly. These tendencies are called constraints, and they conflict with each other such that they cannot all be true all the time. In Optimality theory, languages differ not in what constraints they have, but in the relative importance or ranking of constraints in relation to other ones. This ranking is used for a particular language to grade possible pronunciations (or outputs) of some input. The winning or “optimal” outputs will hopefully be the pronunciations we observe in the language.
“Faithfulness Constraints” judge outputs on how similar they are to the input, i.e., how easy it is for the listener to reconstruct the string from the pronunciation.
“Markedness Constraints” judge outputs on how easy they are to say.
“Alignment Contraints” judge outputs on how far away each feature is from the edges of some important rhythmic unit, e.g., how far [nasal]s are from the right side of the word (or whatever the language does).
Markedness and Faithfulness are inherently in conflict. If markedness were all there was, the speaker would just reduce all syllables to the least marked segments, and language would probably consist only of something like “ta ta ta” or “ti ti ti.” If faithfulness was all there was, speakers might try to put an infinite amount of distinctiveness in every sound, and they would never actually get to pronouncing anything. To resolve the conflict, languages make some decision about which constraints are more and less important.
After some ranking has been determined, it becomes possible to easily run each possible pronunciation through the constraint hierarchy. These possible pronunciations are called candidates, and the process is called evaluation. When this is done, many losers are ruled out immediately (especially the ones that could never win no matter what the ranking). Other candidates are good contenders, and those can be compared against each other in an OT tableau.
Segments have identical values of feature F in the input and output.
The value of F from the input is the same in the output.
Segments have identical values of feature F in the base and the reduplicant.
The value of F from the base is the same in the reduplicant.
NB: Since reduplicative morphemes have no phonological content of their own, this is an Output-Output constraint, where different parts of the output are compared to each other.
Max = MaxIO(C/V)
All segments in the input are also present in the output.
Every segment in the input has an output correspondent.
Don’t delete anything.
Dep = DepIO(C/V)
No segments in the output are also present in the input.
Every segment in the out has an input correspondent.
Don’t add anything.
There are no syllable codas.
All syllables are open.
All consonants get syllabified in the onset.
All syllables have an onset.
Every vowel has a consonant before it.
There are no consonant clusters (sequences of two or more consonants in a syllable).
NB: [C.C] does not incur a violation of *Complex.
Adjacent consonants or vowels share the same value of feature F.
All instances of feature F on its tier connect to the left or right edge of the prosodic unit Unit.
All segments have a value (+/-) for binary feature F.
Sequences of segments XYZ are prohibited (if XYZ stand for segments).
The combination of features XYZ on one segment is prohibited (if XYZ stand for features).
This is how language-specific phonotactics (the rules for legal and illegal sound combinations) are done in Optimality theory.
Obligatory Contour Principle: Also known as the “OCP,” as in the phrase “OCP effects.” In phonology, OCP is the requirement on adjacent pieces of sound to change by some minimum amount. In other words, over some unit of time, there must be a certain minimal change in some feature. The OCP was first observed to operate for tone, i.e., in adjacent syllables for instance, there cannot be a certain number of hi tones, mid tones, etc. in a row. To avoid this, languages sacrifice faithfulness and alter the output.
Subset Principle: When there are two rules – a general rule, and a more specific one that applies to a subset of the cases of the general rule – the specific rule must be applied first. Otherwise, the general rule would apply and there would be no way for the specific rule to have any effect.