Constraint-based Coanalysis: Portuguese cliticisation and morphology-syntax interaction in HPSG

Berthold Crysmann

Saarbrücker Dissertationen in Computerlinguistik und Sprachtechnologie DFKI LT 2003.


In this book, I argue in favour of a morphology-syntax interface that opens up the possibility that surface strings may be licensed simultaneously by morphological and syntactic well-formedness conditions. The approach, which I have dubbed Constraint-based Coanalysis, builds on the assumption inherent in monotonic constraint-based formalisms, such as HPSG, that interaction within a module and across modules proceeds via combination of partial information contributed from different sources. Thus, the perspective assumed here can best be described in terms of cooperation, instead of competition. The main evidence in favour of morphosyntactic coanalysis derives from a detailed study of cliticisation in European Portuguese, a phenomenon which, as I argue, cannot be satisfactorily described by reference to a single grammatical module alone. While the application of most of the diagnostic criteria suggested by Zwicky & Pullum (1983) indeed suggests affixal status for these elements, their placement and the fact that they can take wide scope over a coordination of hosts demand that these pronominal affixes be granted some limited syntactic visibility. The present work makes two major contributions to the field: the incorporation of the concept of coanalysis into current unification-based theories of linguistic knowledge, and the formulation of a monotonic theory of morphological paradigms, which captures basic insights from current non-monotonic theories of realisational morphology and which can serve as a convenient basis for a lexically constrained morphology-syntax interface. The present work also contributes, though probably to some lesser degree, to the representation of phonological processes in a monotonic constraint-based framework.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence