讓格書寫以及台華互譯初探

LangGeh Orthography and an Initial Study of Statistical Translation Between Taiwanese and Mandarin

讓格書寫是新近提議的書寫方式，主張在語句中適當的地方加上空白字元，適合台客華語等使用漢字ｅ文字系統。大體上講，讓格書寫是分簡短詞組，對比的是英語的分詞書寫，傳統華語的分句書寫，以及語言技術的分詞技術。讓格書寫有減少模糊、方便閱讀、利益語言技術等優點。我們甚至認為，空白字母的地位如同數字系統的零。我們使用讓格書寫，製作一套主要是台華語的平行語料庫，各約15萬字，並且用來初步探討台華語詞典製作、以及台華語互譯問題。在此過程中，我們利用了台華語的二大類似、共同詞多、詞序類似。比較現時詞組為基礎的統計式翻譯趨向，讓格書寫實質上有讓台語的語言技術站在較佳的基礎上。

LangGeh orthography is a new writing style proposed by. For Han family languages such Taiwanese or Mandarin that uses Chinese character, LangGeh proposes writing with spaces in-between, using simple short phrase as a unit. This is in contrast to word-based orthography in English and sentence-based orthography in traditional Mandarin. Easy to add spaces, LangGeh has the advangtages of reducing ambiguity, easier to read, and easier for text processing in Chinese characters. Using the LangGeh orthography, we produce a parallel corpus in Taiwanese and Madarin, about 150 thousand characters each. We then explore the extraction of “phrase dictionary” from the parallel corpus, and begin the study of statistical translation between Taiwanese and Mandarin.