Constructing pronoun forms in English to Swahili machine translation

Research output: Working paperScientific


It is hard to find a linguistic feature, where English and Swahili are identical. While English verb has only five possible forms, Swahili verb has theoretically millions of valid forms. Not all of them can be found in practice for each verb lemma. However, it is not difficult to find over two thousand forms for a single verb. The complexity derives from the noun class system and from the fact that Swahili encodes morphological and also syntactic features as part of the verb. Most of these features are prefixes, but also verb derivation by means of suffixes is common. In this report we investigate one specific problem deriving fron differences in language structure. We try to find out how Swahili pronoun forms can be produced in cases, where the referent of the pronoun is difficult to identify.
Original languageEnglish
Place of PublicationHelsinki
PublisherUniversity of Helsinki, Institute for Asian and African Studies
Number of pages18
Publication statusPublished - 2019
MoE publication typeD4 Published development or research report or study

Fields of Science

  • 6121 Languages
  • 113 Computer and information sciences

Cite this