Computational methods for descriptive and theoretical morphology: a brief introduction

Bonami and Sagot talk about the computational methods for descriptive and theoretical morphology. While computational morphology is a well-established sub-field of computational linguistics with important applications in Natural Language Processing (NLP), it has long been somewhat isolated from descriptive and theoretical morphology. Historically, computational approaches to morphological analysis predominantly relied on finite-state transducers, either as a framework for manually developing morphological analysers and generators--also for generating large-scale lexica-or for the unsupervised acquisition of morphological analyses. Computational approaches to the unsupervised acquisition of morphological structure have also failed to convince theoretical and descriptive morphologists despite the early adoption of information-theoretic approaches.