Abstract
In previous studies Lin (2009) and Yang (2010) used the method of longest common subsequence (LCS) to help aligning parallel sentences in a parallel corpus between Taiwanese and Mandarin. Contrast to the traditional writing of no space inside a sentence, the parallel corpus consists of sentences written in so-called “LangGeh”(讓格) orthography, or, using simple short phrases(SSP) as a unit and having spaces in-between. This paper continues the alignment study on parallel sentences between Mandarin and English. With the help of chunking, we first segment an English sentence into sequence of simple short phrases, and align the Mandarin-English parallel sentences in SSP using the same LCS method. In sentence alignment using LCS, a gain function between the SSP’s is required. We again use LCS to compute a score for each pair of simple short phrases. The method is used to help aligning a news parallel corpus in Mandarin and English.