翻譯記憶系統的發展歷程與未來趨勢

Translation Memory Systems: A Historical Sketch and Future Trends

本文對翻譯記憶系統的發展歷程進行了簡單回顧，指出其原型思想肇始於20 世紀60 年代，商用系統出現於20 世紀90 年代初，近20 年來取得了迅猛發展。本文認為，當代翻譯記憶系統可以分為句級翻譯記憶和低於句級翻譯記憶的兩代模式，並批評了當代句級匹配演算法的弊病，進而指出，新興的低於句級翻譯記憶極大地提高了翻譯記憶庫的利用率，成為不可逆轉的趨勢。本文分析了當前翻譯記憶系統發展的五大趨勢，並指出低於句級翻譯記憶模式尚未成熟，在語種、系統結構等方面尚有有一些內在的缺點和不足。此外，新型翻譯記憶系統對譯者的翻譯活動和素質要求將產生何種影響，亦值得學界予以進一步關注。

This paper begins with a historical sketch of translation memory (TM) systems—though the prototype of the TM module can be traced back to the 1960s, TMs were not commercialized until almost 30 years later. Since then, they have enjoyed rapid development. This paper proposes that TM systems can be divided into two generations, namely sentential TMs and sub-sentential TMs. Based on a critique of the former, we see that a sub-sentential segmentTM system, as an effective booster of TM leverage efficiency, has become an irrevocable trend. Despite the bright prospect of TMs as shown in five basic trends, the new-generation sub-sentential TM system is still premature and thus susceptible to defects and deficiencies such as the limitation of language pairs and modular structures. This study concludes that, greater attention is called for with regard to what possible impacts the new TM system will have on the translating behaviors and qualifications of translators today.