Evaluating different translation methods: a case study of Chinese graduate students in an MTI program

This paper compares three approaches to translation: (i) from-scratch human translation (HT), (ii) machine translation combined with post-editing (MTPE), both produced by two independent groups of translation students, and (iii) machine translation (MT), produced by Baidu Chinese-to-English Translat...

Full description

Saved in:
Bibliographic Details
Main Authors: Tingting Lei, Jeroen van de Weijer
Format: Article
Language:English
Published: Taylor & Francis Group 2025-12-01
Series:Cogent Arts & Humanities
Subjects:
Online Access:https://www.tandfonline.com/doi/10.1080/23311983.2025.2511386
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849471813123309568
author Tingting Lei
Jeroen van de Weijer
author_facet Tingting Lei
Jeroen van de Weijer
author_sort Tingting Lei
collection DOAJ
description This paper compares three approaches to translation: (i) from-scratch human translation (HT), (ii) machine translation combined with post-editing (MTPE), both produced by two independent groups of translation students, and (iii) machine translation (MT), produced by Baidu Chinese-to-English Translation, employing various evaluation methods. A number of unsupervised automated models (BLEU, METEOR and ROUGE) were used to compare the quality of these three types of translation outputs. Complementary to this, a subjective evaluation method was applied to calculate the frequency of various translation errors in a semi-automatic way. It was found that the quality of the three types of texts assessed by these two methods displayed similar characteristics: In this test, both MT and MTPE significantly outperformed from-scratch human translation in terms of quality, while the differences between MT and MTPE were not significant. Lexical and grammatical mistakes accounted for the highest proportion of errors in the different translated texts and after data and textual analysis different PE styles were identified. The results have implications for translation training and the further development of MT research.
format Article
id doaj-art-0b625feba8ef4c1a833f0639a9ef7e28
institution Kabale University
issn 2331-1983
language English
publishDate 2025-12-01
publisher Taylor & Francis Group
record_format Article
series Cogent Arts & Humanities
spelling doaj-art-0b625feba8ef4c1a833f0639a9ef7e282025-08-20T03:24:43ZengTaylor & Francis GroupCogent Arts & Humanities2331-19832025-12-0112110.1080/23311983.2025.2511386Evaluating different translation methods: a case study of Chinese graduate students in an MTI programTingting Lei0Jeroen van de Weijer1College of International Studies, Shenzhen University, Shenzhen, ChinaCollege of International Studies, Shenzhen University, Shenzhen, ChinaThis paper compares three approaches to translation: (i) from-scratch human translation (HT), (ii) machine translation combined with post-editing (MTPE), both produced by two independent groups of translation students, and (iii) machine translation (MT), produced by Baidu Chinese-to-English Translation, employing various evaluation methods. A number of unsupervised automated models (BLEU, METEOR and ROUGE) were used to compare the quality of these three types of translation outputs. Complementary to this, a subjective evaluation method was applied to calculate the frequency of various translation errors in a semi-automatic way. It was found that the quality of the three types of texts assessed by these two methods displayed similar characteristics: In this test, both MT and MTPE significantly outperformed from-scratch human translation in terms of quality, while the differences between MT and MTPE were not significant. Lexical and grammatical mistakes accounted for the highest proportion of errors in the different translated texts and after data and textual analysis different PE styles were identified. The results have implications for translation training and the further development of MT research.https://www.tandfonline.com/doi/10.1080/23311983.2025.2511386Error frequencieshuman translationmachine translationpost-editingtranslation trainingunsupervised evaluation metrics
spellingShingle Tingting Lei
Jeroen van de Weijer
Evaluating different translation methods: a case study of Chinese graduate students in an MTI program
Cogent Arts & Humanities
Error frequencies
human translation
machine translation
post-editing
translation training
unsupervised evaluation metrics
title Evaluating different translation methods: a case study of Chinese graduate students in an MTI program
title_full Evaluating different translation methods: a case study of Chinese graduate students in an MTI program
title_fullStr Evaluating different translation methods: a case study of Chinese graduate students in an MTI program
title_full_unstemmed Evaluating different translation methods: a case study of Chinese graduate students in an MTI program
title_short Evaluating different translation methods: a case study of Chinese graduate students in an MTI program
title_sort evaluating different translation methods a case study of chinese graduate students in an mti program
topic Error frequencies
human translation
machine translation
post-editing
translation training
unsupervised evaluation metrics
url https://www.tandfonline.com/doi/10.1080/23311983.2025.2511386
work_keys_str_mv AT tingtinglei evaluatingdifferenttranslationmethodsacasestudyofchinesegraduatestudentsinanmtiprogram
AT jeroenvandeweijer evaluatingdifferenttranslationmethodsacasestudyofchinesegraduatestudentsinanmtiprogram