Abstracting Suffixes: A Morphophonemic Approach to Polish Morphological Analysis1

This paper presents a morphophonology-based Item-and-Process approach to the finite-state lemmatization and morphological analysis of Polish. Unlike current text-based techniques, which search for all possible orthographic representations of Polish morphological suffixes, the multi-level algorithm presented here extracts morphophoneme arrays from graphemic word suffixes, independent of their surface representation. This makes it possible to use a simple mono-lemmatic dictionary, as well as to distinguish between homographic suffixes, and to carry out various phonological and morphological investigations using suffix fields in corpora.