Abstract:
Machine Translation is the automatic translation of text from a source language to the target language. The demand for translation has been
increasing due to the exchange of information between various regions using different regional languages. English-Tigrigna Statistical Machine
Translation, therefore, is required since a lot of documents are written in English. This research study used statistical mac hine translation approach due
to it yields high accuracy and does not need linguistic rules which exploit human effort (knowledge). The language model, Translation model, and
decoder are the three basic components in Statistical Machine Translation (SMT). Moses' decoder, Giza++, IRSTLM, and BLEU (Bilingual Evaluation
Understudy) are tools that helped to conduct the experiments. 17,338 sentences of bilingual corpus for training, 1000 sentences for test set and 42,284
sentences for language model were used for experiment. The BLEU score produced from the experiment was 23.27% which would still not enough for
applicable applications. As a result, the effect of word factored or segmentation in the translation quality is reduced by increasing the data size of the
corpus.
