Finite Automata for Processing Word Order

This paper is concerned with applications of-nite state automata (FSA) to word order processing. As the surface order of constituents is a function of many interacting factors, an adequate formalisation seems fairly complicated. I argue that despite this complexity, most ordering constraints observed in free word order languages can be treated by lower-level methods such as nite state automata. In order to illustrate my claim, I will present a nite state formalisation of some ordering constraints for German. Since the acceptability of an utterance very often depends on the context it occurs in, the approach taken does not only distinguish between grammatical and ungrammatical structures, but also takes into account some discourse information for-malised in terms of information packaging (ground/focus). Furthermore, the paper discusses the performance of a grammar of German augmented with an FSA-based word order processing module. Finally, the problem of morphological and syntactic ambiguity is addressed.