2007On the Importance of Errors: Competence and Performance in Optimality Theory.
Conference talk given at: Tabu Day 2007, June 8, 2007, Groningen, Netherlands. See also http://www.rug.nl/let/onderzoek/onderzoekinstituten/clcg/events/tabudag/index. Slides.




Optimality Theory (OT, Prince and Smolensky 1993/2004) is a grammar architecture, meaning that its goal is to account for linguistic competence. Performance phenomena, such as fast speech errors, have been either ignored -- following the Chomskyan tradition of focussing on competence -- or explained by modifications of the competence model. I argue, in accord with Smolensky and Legendre (2006) among many others, that performance phenomena must be explained on the level of the implementation of the (OT) grammar. In other words, the grammar describes what forms are considered grammatical by competence, whereas the algorithm trying to calculate the grammatical forms realizes linguistic performance. The runtime, the accuracy (rate or errors) and the complexity of the algorithm should mimic the brain performing an utterance.

Without going into technical details, we compare what the ICS Architecture (Smolensky and Legendre, 2006) and the SA-OT Algorithm (Bíró 2006) say about performance. Both use simulated annealing, but in a different way, and we shall show how the SA-OT Algorithm refutes a few strong claims made by the authors of the ICS Architecture. Furthermore, we discuss the consequences of this competence-performance distinction to models of language learning and acquisition.

The talk should also be accessible to non-computational linguists interested in the competence-performance dichotomy.


Bíró, Tamás (2006). Finding the Right Words: Implementing Optimality Theory with Simulated Annealing. PhD-thesis. GroDiL 62. Groningen. Available: http://www.birot.hu/publications/dissertation/dissertation.pdf and http://dissertations.ub.rug.nl/faculties/arts/2006/t.s.biro/
Prince, Alan and Paul Smolensky (1993/2004). Optimality Theory: Constraint Interaction in Generative Grammar. Rutgers University Center for Cognitive Science Technical Report 2 (RuCCS-TR-2) (1993). Malden, MA -- Oxford, UK: Blackwell (2004).
Smolensky, Paul and Géraldine Legendre (eds.) (2006). The Harmonic Mind: From Neural Computation to Optimality-Theoretic Grammar. Cambridge MA and London, UK: Bradford Book, MIT Press.