Swahili Language Manager: A Storehouse for Developing Multiple Computational Applications

Abstract

Swahili Language Manager (SALAMA) is a computational environment for managing written Swahili language and for developing various kinds of language applications. Having been subject to development since 1985, it currently (2004) contains the Standard Swahili lexicon as fully as possible. As it is a system for managing the language, it includes also the full morphological and morpho-phonological description of Swahili, a rule-based system for solving the word level ambiguities, a rule-based system for tagging text syntactically (including alternatively a shallow Constraint Grammar parsing or a deep Dependency Grammar parsing), a rule-based system for handling idiomatic expressions, proverbs and other non-standard clusters of words, and a semantic tagging and disambiguation system for defining correct semantic equivalents in English. SALAMA facilitates also a raw translation from Swahili to English, including the correct surface forms in English (e.g. verbs, nouns and adjectives) and transfer rules for the correct English word order. An essential part in developing and testing SALAMA is the Helsinki Corpus of Swahili, which has been under construction since 1988 and is currently globally available at the Language Bank of Finland (www.csc.fi). The paper discusses all these features in detail.