Abstract
The development of a computational morphological analyser for Setswana necessitates the accurate modelling and implementation of, among others, compounding as a word formation process. Compounding is known to be an area of Setswana morphology that has sadly been neglected and still requires much investigation and research. The main purpose of this article is to investigate the formation of noun + noun compounds by computational morphological means in order to understand how this process should be formalised, modelled and subsequently implemented. In particular, an empirical study based on a collection of Setswana noun + noun compounds is reported on. The computational morphological analysis of these compounds revealed linguistic deviations from the standard morphological rules governing the formation of nouns and deverbatives. Examples, computational results and a discussion of the main findings are included.
Highlights
Setswana, as a semi-disjunctively written, agglutinating South African language, is characterised by a complex morphology
Reiterating, the research question was: “How should the morphological analysis of compounds be formalised or modelled and implemented so that compounding in Setswana can be handled as accurately as possible by the morphological analyser under development?” In order to address this question for noun + noun compounds an empirical study was conducted
From a computational perspective the experiments confirmed the accuracy of the noun and deverbative morphology as modelled and implemented in the current Setswana analyser prototype
Summary
As a semi-disjunctively written, agglutinating South African language, is characterised by a complex morphology. In achieving the long term goal of developing a broad coverage computational morphological analyser for Setswana it is necessary to ensure that, ideally all and only valid Setswana words are analysed correctly. Among others this means that all productive word formation processes should be modelled accurately. Setswana pronouns and some Setswana particles are class bound and are considered closed (morphologically unproductive) word classes. Setswana pronouns and particles may be explicitly included in a computational morphological analyser. Further complexity is added to Setswana morphology by productive processes such as compounding. Krüger (2006:41) states that the analysis of Setswana compounds “is an area of morphology that is sadly
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
