Translation Base: An ambitious endeavor in corpus linguistics and translation


Chinese – English Translation Base

General Context and Rationale

English and Chinese will soon be the only global languages. In a few years, Chinese will overtake English as the most common web language. There will be an almost unlimited market for Chinese-English language technology, along with an insatiable demand for Chinese translations of English texts and vice versa.

The Chinese-English Translation Base presents a new approach to translation. It is innovative because

It is based on real source and target language.

It makes use of the practice of hundreds of translators.

It processes meaning, not form.

It replaces the (polysemous) word by (monosemous) units of meaning.

Our Translation Base will facilitate, improve and speed up human translation, it will make possible machine translation of real, unrestricted natural language, and it will be used for a range of further Chinese-English language technology, including collaborative multimedia document authoring and publishing, the creation of parallel language versions, cross-lingual terminology support, summarisation, quality assurance of translations, and also for language learning. The TranslationBase will make traditional bilingual dictionaries redundant.

Our Translation Base will change the paradigm of machine translation (MT). Classical MT systems work well in strictly defined domains with standardised terminologies which permit controlled language techniques, thereby reducing semantics to algorithmic processing. Typical examples are airplane maintenance manuals, where MT rates better than human translation. But for the bulk of texts to be translated, for general, largely unrestricted language, including legal language, newspaper language and company communication, classical MT systems have failed. They had to fail because they use the single word as the basic unit of translation. Single words often have many meanings; and MT programs are notoriously unsuccessful in selecting the required equivalent. Therefore, in many contexts they have now been replaced by translation memories. Translation memories and Example-Based MT help translating new documents by scanning previously translated documents for formal similarities. They are not rule-based like MT but statistic-based. But like MT they are not content-oriented, and therefore they do hardly better than classical MT. Depending on text types and domains, they may speed up human translation by no more than 10% to 20%.



  • 基于汉、英语平行语料库的翻译数据库设计.pdf
    444.5 KB · 浏览: 101
回复: Translation Base: An ambitious endeavor in corpus linguistics and translation

thanks a lot,oscar3.
回复: Translation Base: An ambitious endeavor in corpus linguistics and translation

英汉方面的翻译随着中国的入市和逐渐成为世界工厂以后,是越来越大。需求肯定会比较大,不一定会有"unlimited market",但是这个方向是被很多公司看好的。