Optimality Theory posits that languages are systems of ranked, violable constraints. Phonological patterns emerge from constraint conflict resolution: the optimal output best satisfies the hierarchy, even if lower-ranked constraints are violated. This approach unifies cross-linguistic patterns (e.g., consonants prefer onset position) while capturing language-specific variation through constraint ranking, moving phonology from rules to optimization.
From your study of phonological systems, you know that languages are not random collections of sounds — they have systematic patterns. Some syllable structures are preferred over others; certain sound sequences recur across unrelated languages; sounds change predictably in specific environments. Earlier phonological theory explained these patterns with rules: ordered instructions that transform one representation into another. Optimality Theory (OT), developed by Prince and Smolensky in the early 1990s, proposes a fundamentally different architecture: instead of rules that apply sequentially, there are constraints — and instead of derivations, there is optimization.
The key conceptual move in OT is treating constraints as ranked and violable. This is counterintuitive coming from rule-based thinking, where a rule either applies or doesn't. In OT, every candidate output violates some constraints — the question is *which* constraints it violates. The optimal output is the one that best satisfies the constraint hierarchy: it avoids violations of high-ranked constraints, even at the cost of violating lower-ranked ones. Think of it like a priority ordering in decision-making: you satisfy your top priority first, then your second, and so on. A candidate that violates a high-ranked constraint loses to a candidate that violates only lower-ranked ones, even if the winner violates more constraints in total. What matters is the *ranking* of the violated constraint, not the count.
OT constraints divide into two types. Markedness constraints evaluate the output itself — they penalize cross-linguistically dispreferred structures like codas (syllable-final consonants), consonant clusters, or complex onsets. Faithfulness constraints evaluate the relationship between the underlying form (what the speaker intends) and the surface form (what gets pronounced) — they penalize deletions, insertions, or changes. The constant tension between markedness and faithfulness drives phonological patterns. A language with high-ranked NOCODA (no syllable-final consonants) will delete or restructure final consonants, violating faithfulness to achieve a marked-free output. A language with high-ranked faithfulness will preserve final consonants even though they are marked. Language-specific phonological patterns emerge from language-specific rankings of the same universal constraint set.
The power of OT is that it explains cross-linguistic universals and language-specific variation with a single mechanism. Universals emerge because some markedness constraints dominate in most languages — consonantal onsets are almost universally preferred, for example. Variation emerges because the same constraints are ranked differently across languages. Japanese has more severe restrictions on coda consonants than English because NOCODA outranks faithfulness in Japanese; English tolerates them because faithfulness outranks NOCODA. Learning a language, in OT terms, is partly the process of discovering its constraint ranking — the same inventory of universal constraints, ordered differently, produces different grammars. OT also elegantly captures conspiracies: cases where multiple apparently different phonological processes in a language all serve the same structural goal, which OT attributes to a single high-ranked markedness constraint driving multiple different repair strategies toward the same output preference.