Abstract:
Machine Translation Systems expect target
language output to be grammatically correct
within the frame of proper grammatical
category. In Myanmar-English statistical
machine translation system, the proposed system
concerns with the target language model to
smooth the translated English sentences. Most of
the typical grammar checkers can detect
ungrammatical sentences and seek for what
error it is. However, they often fail to detect
grammar errors for translated English sentence
such as missing words. Therefore, we intend to
develop an ongoing grammar checker for
English as a second language (ESL). There are
two main tasks in this approach: detecting the
sentence pattern in chunk structure and
analyzing the chunk errors. This system identifies
the chunk types and generates context free
grammar (CFG) rules for recognizing
grammatical relations of chunks. It also presents
a hybrid approach of trigram-based markov
model with rule-based model. Experimental
results show that the proposed grammar checker
can improve the effectiveness of translation
quality.
