研究背景Backgorund伯尔曼(2000 年)指出,翻译过程包括变形过程,他将其中一个过程称为 “数量贫化”,即词汇丰富性和多样性的损失。虽然人工翻译可以减轻这种损失,但这种损失在某种程度上是不可避免的,因为在将一种语言翻译成另一种语言时,很难尊重众多的符号和结构。Berman (2000) observed that the translation process consists of deformation processes, one of which he refers to as ‘quantitative impoverishment’, a loss of lexical richness and diversity. Although mitigated by a human translator, this loss is to some extent inevitable as it is hard to respect the multitude of signifiers and constructions when translating one language into another.研究目的Purpose通过比较不同 MT 系统的词汇多样性并进一步分析词频,研究旨在揭示多样性的丧失与某些词的加剧或丧失之间的关系,验证:
1. NMT 与 SMT 和 HT 在词汇丰富度或词汇丰富度损失方面的程度;2. 量化不同的 MT 架构在多大程度上有利于在训练数据中更常见的翻译报告及原文Download论文原文和报告可在微信公众号“翻译圈”后台回复文献精读获取!全文展示Passage