A constraint‑based linguistic model in which a generator produces a set of candidate outputs for an input and an evaluator selects the optimal candidate by comparing how candidates violate a hierarchy of ranked constraints (typically distinguishing markedness constraints from faithfulness constraints); the optimal form is the candidate with the least serious violation pattern under the ranking.

Principle

Grammatical selection is the outcome of cross‑candidate comparison under a fixed ranking: a higher‑ranked constraint always outranks the desire to satisfy lower‑ranked constraints, so the chosen candidate minimizes violations according to that ranking rather than eliminating violations absolutely.

Demonstration

Illustrative scenario → Situation: For an underlying syllable structure, several surface segmentations are possible. Recognition: The analyst enumerates candidate outputs. Action: Each candidate is evaluated against ranked constraints (e.g., *ComplexOnset ≫ Parse), and the evaluator discards candidates losing on the highest‑ranked relevant constraint. Consequence: The surviving candidate(s) predict attested surface forms without reference to sequential derivations.

Misapplication

Interpreting constraints as independently satisfiable optimization goals (i.e., treating the model as weighted constraints or as serial derivations) or assuming constraint rankings are freely exchangeable across analyses. The error confuses ranking‑based comparative evaluation with additive weighting or stepwise rule application.

Consequence

Analyses emphasize the identification and ranking of constraints rather than ordered rules; typological claims and account of alternations follow from conceivable rankings, and differences between languages are expressed as different rankings or the presence/absence of particular constraints.

Reversal

If constraints are reinterpreted as numerical weights (harmonic grammar) or made stochastic, the strict ranking principle is qualified and comparative evaluation becomes gradient or probabilistic rather than strictly rank‑based.

Boundary

Clearly within: analyses of phonological and morphological alternations where candidate sets and constraint rankings are meaningful. Boundary case: syntactic phenomena that can be cast as constraint interactions but where candidate enumeration is impractical. Clearly outside: serial derivational accounts that model changes as ordered rule application producing intermediate outputs.

Semantic Tension

Tension between constraint‑interaction (parallel evaluation) and derivational/transformational accounts (serial application of rules); the two frameworks propose different mechanisms for how surface forms arise.

Synthesis

Optimality Theory reframes grammar as comparative evaluation under ranked constraints: grammatical generalizations become statements about which constraints are ranked above others rather than about the sequential application of deterministic rules.