finite-state morphological analysis

The Computational Morphological Analysis project is a unique multidisciplinary research project, requiring knowledge, expertise and skills associated with the disciplines of linguistics of the South African indigenous languages and of computing science.

This project is NRF (National Research Foundation) supported within the Focus Area of Information and Communication Technology (cf. https://www.nrf.ac.za).

The significance of morphological analysis as a basic enabling application for further kinds of NLP is well known. The primary aim of the overall project therefore is the development of computational morphological analysers for Bantu languages, using the natural language independent Xerox Finite-State Tools (https://www.fsmbook.com). This integrated set of tools is used to model and implement the complexities of word-formation rules as well as morphophonological alternations by means of finite-state networks, which in turn are combined together algorithmically into larger networks that perform morphological analysis. Lexical challenges are addressed by means of the development of machine-readable lexicons in XML format, containing knowledge about individual words in the languages.

Project Leader:
Sonja Bosch
boschse [at] unisa [dot] ac [dot] za