A first-pass approach for evaluating machine translation systems

This article presents a probabilistic sub-tree alignment model and its application to tree-to-tree machine translation. Unlike previous work, we do not resort to surface heuristics or expensive annotated data, but instead derive an unsupervised model to infer the syntactic correspondence between two languages. More importantly, the developed model is syntactically-motivated and does not rely on word alignments. As a by-product, our model outputs a sub-tree alignment matrix encoding a large number of diverse alignments between syntactic structures, from which machine translation systems can efficiently extract translation rules that are often filtered out due to the errors in 1-best alignment. Experimental results show that the proposed approach outperforms three state-of-the-art baseline approaches in both alignment accuracy and grammar quality. When applied to machine translation, our approach yields a +1.0 BLEU improvement and a -0.9 TER reduction on the NIST machine translation evaluation corpora. With tree binarization and fuzzy decoding, it even outperforms a state-of-the-art hierarchical phrase-based system.

Download Full-text

Interlingua-based Machine Translation Systems: UNL versus Other Interlinguas

The Egyptian Journal of Language Engineering ◽

10.21608/ejle.2014.59863 ◽

2014 ◽

Vol 1 (1) ◽

pp. 42-54

Author(s):

Sameh Alansary

Keyword(s):

Machine Translation ◽

Translation Systems

Download Full-text

Is automation changing the translation profession?

International Journal of the Sociology of Language ◽

10.1515/ijsl-2020-0015 ◽

2021 ◽

Vol 0 (0) ◽

Author(s):

Anthony Pym ◽

Ester Torres-Simón

Keyword(s):

Machine Translation ◽

Communication Skills ◽

Service Providers ◽

Time Data ◽

Market Segments ◽

Interactive Communication ◽

Neural Machine Translation ◽

History Of ◽

Translation Services ◽

Translation Systems

Abstract As a language-intensive profession, translation is of frontline interest in the era of language automation. In particular, the development of neural machine translation systems since 2016 has brought with it fears that soon there will be no more human translators. When considered in terms of the history of automation, however, any such direct effect is far from obvious: the translation industry is still growing and machine translation is only one instance of automation. At the same time, data on remuneration indicate structural wage dispersion in professional translation services, with some signs that this dispersion may increase in certain market segments as automated workflows and translation technologies are adopted more by large language-service providers more than by smaller companies and individual freelancers. An analysis of recent changes in discourses on and in the translation profession further indicates conceptual adjustments in the profession that may be attributed to growing automation, particularly with respect to expanding skills set associated with translation, the tendency to combine translation with other forms of communication, and the use of interactive communication skills to authorize and humanize the results of automation.

Download Full-text