Post-editing neural machine translation versus translation memory segments

Abstract

The use of neural machine translation (NMT) in a professional scenario implies a number of challenges despite growing evidence that, in language combinations such as English to Spanish, NMT output quality has already outperformed statistical machine translation in terms of automatic metrics scores. This article presents the result of an empirical test that aims to shed light on the differences between NMT postediting and translation with the aid of a translation memory (TM). The results show that NMT post-editing involves less editing than TM segments, but this editing appears to take more time, with the consequence that NMT post-editing does not seem to improve productivity as may have been expected. This might be due to the fact that NMT segments show a higher variability in terms of quality and time invested in post-editing than TM segments that are 'more similar' on average. Finally, results show that translators who perceive that NMT boosts their productivity actually performed faster than those who perceive that NMT slows them down. Response to Reviewers:All changes included.

Towards a Better Integration of Fuzzy Matches in Neural Machine Translation through Data Augmentation

“…MT tends to be used as a 'back-off' solution to TMs in cases where no sufficiently similar source sentence is found in the TM [12,13], since post-editing MT output in many cases takes more time than correcting (close) TM matches. This is, for example, due to inconsistencies in translation and a lack of overlap between MT output and the desired translation [14]. The level of similarity between the sentence to translate and the sentence found in the TM, as calculated by a match metric [15,16], thus plays an important role.…”

“…In contrast, MT produces a translation of any input sentence, but in spite of the recent increase in MT quality, this output is still not always completely error-free. Moreover, the perception of translators is that MT errors are often not predictable or coherent, which results in a lower confidence for MT output in comparison to TM segments [14,17].…”

OPUS-CAT: Desktop NMT with CAT integration and local fine-tuning

“…OPUS-CAT is intended for professional translators, and the utility of generic NMT models in professional translation is uncertain (Sánchez-Gijón et al, 2019), while performance improvements resulting from the use domain-adapted NMT models have been observed multiple times (Läubli et al, 2019;Macken et al, 2020). Because of this, OPUS-CAT MT Engine includes a functionality for finetuning models with small amounts of bilingual data.…”

Traducción humana vs. traducción automática: análisis contrastivo e implicaciones para la aplicación de la traducción automática en traducción jurídica

“…Según O'Brien (2012), la posedición es "la corrección que un traductor hace de traducciones en bruto generadas por un sistema de ta de conformidad con unas guías y unos criterios de calidad específicos" (propuesta de traducción del autor). Este tipo de estudios se ha hecho desde muchas vertientes y puntos de vista diferentes (Moorkens y O'Brien, 2017;Sánchez-Gijón et al, 2019). No obstante, el objetivo rector de este estudio no es ver con qué herramienta se es más productivo, sino más bien observar las implicaciones que puede tener la introducción de la ta en la formación de traductores especializados en textos jurídicos.…”