How Does the Post-Editing of Neural Machine Translation Compare with From-Scratch Translation? A Product and Process Study

Yanfang Jia, Michael Carl, Xiangling Wang

Research output: Contribution to journalReviewResearchpeer-review

Abstract

This study explores the post-editing process when working within the newly introduced neural machine translation (NMT) paradigm. To this end, an experiment was carried out to examine the differences between post-editing Google neural machine translation (GNMT) and from-scratch translation of English domain-specific and general language texts to Chinese. We analysed translation process and translation product data from 30 first-year postgraduate translation students. The analysis is based on keystroke logging, screen recording, questionnaires, retrospective protocols and target-text quality evaluations. The three main findings were: 1) post-editing GNMT was only significantly faster than from-scratch translation for domain-specific texts, but it significantly reduced the participants’ cognitive effort for both text types; 2) post-editing GNMT generated translations of equivalent fluency and accuracy as those generated by from-scratch translations; 3) the student translators generally showed a positive attitude towards post-editing, but they also pointed to various challenges in the post-editing process. These were mainly due to the influence of their previous translation training, lack of experience in post-editing and the ambiguous wording of the post-editing guidelines.
Original languageEnglish
JournalJournal of Specialised Translation
Issue number31
Pages (from-to)60-86
Number of pages27
ISSN1740-357X
Publication statusPublished - Jan 2019
Externally publishedYes

Keywords

  • Post-editing process
  • Translation quality
  • Neural machine translation
  • Text types

Cite this

@article{85c0e14026a648d69bdf81d7958b3d64,
title = "How Does the Post-Editing of Neural Machine Translation Compare with From-Scratch Translation?: A Product and Process Study",
abstract = "This study explores the post-editing process when working within the newly introduced neural machine translation (NMT) paradigm. To this end, an experiment was carried out to examine the differences between post-editing Google neural machine translation (GNMT) and from-scratch translation of English domain-specific and general language texts to Chinese. We analysed translation process and translation product data from 30 first-year postgraduate translation students. The analysis is based on keystroke logging, screen recording, questionnaires, retrospective protocols and target-text quality evaluations. The three main findings were: 1) post-editing GNMT was only significantly faster than from-scratch translation for domain-specific texts, but it significantly reduced the participants’ cognitive effort for both text types; 2) post-editing GNMT generated translations of equivalent fluency and accuracy as those generated by from-scratch translations; 3) the student translators generally showed a positive attitude towards post-editing, but they also pointed to various challenges in the post-editing process. These were mainly due to the influence of their previous translation training, lack of experience in post-editing and the ambiguous wording of the post-editing guidelines.",
keywords = "Post-editing process, Translation quality, Neural machine translation, Text types, Post-editing process, Translation quality, Neural machine translation, Text types",
author = "Yanfang Jia and Michael Carl and Xiangling Wang",
year = "2019",
month = "1",
language = "English",
pages = "60--86",
journal = "Journal of Specialised Translation",
issn = "1740-357X",
publisher = "Journal of Specialised Translation",
number = "31",

}

How Does the Post-Editing of Neural Machine Translation Compare with From-Scratch Translation? A Product and Process Study. / Jia, Yanfang; Carl, Michael; Wang, Xiangling.

In: Journal of Specialised Translation, No. 31, 01.2019, p. 60-86.

Research output: Contribution to journalReviewResearchpeer-review

TY - JOUR

T1 - How Does the Post-Editing of Neural Machine Translation Compare with From-Scratch Translation?

T2 - A Product and Process Study

AU - Jia, Yanfang

AU - Carl, Michael

AU - Wang, Xiangling

PY - 2019/1

Y1 - 2019/1

N2 - This study explores the post-editing process when working within the newly introduced neural machine translation (NMT) paradigm. To this end, an experiment was carried out to examine the differences between post-editing Google neural machine translation (GNMT) and from-scratch translation of English domain-specific and general language texts to Chinese. We analysed translation process and translation product data from 30 first-year postgraduate translation students. The analysis is based on keystroke logging, screen recording, questionnaires, retrospective protocols and target-text quality evaluations. The three main findings were: 1) post-editing GNMT was only significantly faster than from-scratch translation for domain-specific texts, but it significantly reduced the participants’ cognitive effort for both text types; 2) post-editing GNMT generated translations of equivalent fluency and accuracy as those generated by from-scratch translations; 3) the student translators generally showed a positive attitude towards post-editing, but they also pointed to various challenges in the post-editing process. These were mainly due to the influence of their previous translation training, lack of experience in post-editing and the ambiguous wording of the post-editing guidelines.

AB - This study explores the post-editing process when working within the newly introduced neural machine translation (NMT) paradigm. To this end, an experiment was carried out to examine the differences between post-editing Google neural machine translation (GNMT) and from-scratch translation of English domain-specific and general language texts to Chinese. We analysed translation process and translation product data from 30 first-year postgraduate translation students. The analysis is based on keystroke logging, screen recording, questionnaires, retrospective protocols and target-text quality evaluations. The three main findings were: 1) post-editing GNMT was only significantly faster than from-scratch translation for domain-specific texts, but it significantly reduced the participants’ cognitive effort for both text types; 2) post-editing GNMT generated translations of equivalent fluency and accuracy as those generated by from-scratch translations; 3) the student translators generally showed a positive attitude towards post-editing, but they also pointed to various challenges in the post-editing process. These were mainly due to the influence of their previous translation training, lack of experience in post-editing and the ambiguous wording of the post-editing guidelines.

KW - Post-editing process

KW - Translation quality

KW - Neural machine translation

KW - Text types

KW - Post-editing process

KW - Translation quality

KW - Neural machine translation

KW - Text types

UR - https://sfx-45cbs.hosted.exlibrisgroup.com/45cbs?url_ver=Z39.88-2004&url_ctx_fmt=info:ofi/fmt:kev:mtx:ctx&ctx_enc=info:ofi/enc:UTF-8&ctx_ver=Z39.88-2004&rfr_id=info:sid/sfxit.com:azlist&sfx.ignore_date_threshold=1&rft.object_id=1000000000281223&rft.object_portfolio_id=&svc.holdings=yes&svc.fulltext=yes

M3 - Review

SP - 60

EP - 86

JO - Journal of Specialised Translation

JF - Journal of Specialised Translation

SN - 1740-357X

IS - 31

ER -