Strict Locality In Morphological Derivations

In this paper, we argue that a computational perspective can help highlight differences between similar but distinct theoretical approaches. In particular, we compare two approaches to representing morphological sequences: derived and derivational. We introduce the reader to Russian patterns — nominalization and telicity conversion — requiring particular dependencies between prefixes and suffixes. Then, we use formal language theory (in the form of the subregular class of strictly local languages) to capture the relevant pattern. We show how the computational power needed to capture these apparently complex dependencies is significantly reduced, if we allow the grammars to judge the wellformedness of morphological operations over derivational representations instead than surface forms.