It depends on what you mean by "traditional theory of optimality". Well, in relation to the very first version of OT, I think that there have been three majors changes.

1) While OT was born as a parallel framework, serialism was reintroduced by Harmonic Serialism. Among other reasons, HS is supposed to constrain the number of possible candidates, and account for the many problems posed by opaque processes.

2) Stratal OT, for its part, is supposed to incorporate into the theory what lexical phonology had previously done in dealing with the interface between phonology and morphology and the lexicon.

3) Finally, pure ranking was replaced, e.g. by Maximal Entropy grammars, with weight coefficients assigned to constraints. This has proved particularly interesting in the study of variation.

I hope this very brief description gives you an idea of the changes that have taken place. Googling "Harmonic Serialism", "Stratal OT" and "MaxEnt grammars" should lead you to the relevant publications.

Joaquim Brandão de Carvalho

The recent changes and developments in Optimality Theory (OT) within phonology:

• Conspiracies and Soft Universals: OT addresses two main issues: Conspiracies: Some languages exhibit constraints that appear to be satisfied in multiple ways, as if rules conspire to achieve a single target. Soft Universals: Constraints that show evidence across unrelated languages but don’t hold universally. Researchers explore these phenomena to refine OT’s understanding of constraint interaction.
• Output-Based Model: OT shifted from rule-based to an output-based model, emphasizing surface forms and constraints in shaping linguistic patterns.
• Historical Sound Change: OT considers underlying representations as a locus of change. Surface structures derived by phonological processes may be reanalyzed as new input forms.
• Connectionist Roots: OT has connections to neural network research, emerging as an alternative to harmonic grammar.
• Stratal OT: Stratal OT models phonology with ranked violable constraints, revealing universals and typological generalizations.