Workshop on Freedom of Analysis

Start date:  01.09.2005

End date:  02.09.2005

Venue:  University of Tromsø, Humanities building, room B1003

For information on how to get to Tromsø and where to stay see our Practical Information section.

INVITED SPEAKERS
Chris Golston (CSU Fresno)
Bruce Morén (CASTL, Tromsø)
Marc van Oostendorp (Meertens Institute, Amsterdam)
Curt Rice (CASTL, Tromsø)
Christian Uffmann (Philipps Universität Marburg)

Please click here for registration
Please click here for conference programme
abstracts, handouts and slides
NEW!!finally pictures
Call for Papers:

Freedom of Analysis?

The bulk of contemporary research in OT focuses on constraints and their interaction, yet other aspects of the overall OT model remain largely unexplored. This workshop takes up the problem of Freedom of Analysis in its broadest sense, and asks to what extent the very phonological properties of candidate outputs are restricted by representational considerations, ‘output viability’, the content of CON and the ranking of constraints in EVAL. The issue of restrictions is an important one, since if it turns out to be possible to formulate principled restrictions on the space of inputs and output candidates in a way that does not duplicate the job of the constraints in EVAL, a more restrictive view of CON and the typology emerges. Sensible discussion of variation presupposes some conception of the limits on the universal space of variation.

Much work in OT fails to spell out its representational assumptions adequately, using representations in a way that sacrifices long-term explanatory to short-term descriptive goals. The fundamental choice between binary features and unary elements is just as pregnant with ramifications for analyses in OT as, say, a rule-based framework, both as regards the content of CON and the resulting typology. The same holds true of theories of the hierarchical organisation of phonological primitives. Yet, despite the continued relevance of the prosodic hierarchy and the recent renaissance of feature geometry, many analyses simply rely in practice on an SPE-style conception of segments as unordered feature bundles or flat autosegmental structures.

A related question involves the extent to which inputs and candidate outputs must be ‘viable outputs’, i.e. phonetically interpretable as they are.

Representational issues figure in defining the absolute variation space. On the classical conception of freedom of analysis, the absolute variation space must be one and the same as the space of candidates for any given input.

To take stock of these issues, we invite abstracts addressing the implications of hard constraints for OT.