ABSTRACT
This chapter demonstrates the strengths and weaknesses of statistical and rule-based machine translation systems and suggests that Sustainable language technology (SMT) between such languages as Bantu languages and English encounters no insurmountable problems. The problem is that the major need for text-based machine translation is between African and European languages and not between two African languages. The main trend in translation technology is to develop translation systems based on SMT and neural machine translation. The research shows that neither SMT nor rule-based machine translation produces fully correct text. Both have weaknesses, although of very different kinds. The chapter demonstrates the differences in performance between Google Translate and Swahili and English is the Swahili Language Manager (SALMA) using example sentences from the news media. The structure of SALAMA resembles the processing method of OpenLogos including a modular pipeline structure. The SALAMA translates long noun phrases according to grammatical rules.
