The DOI matches its Crossref record; 38 of the 43 references it cites were found in a registry or an index.
Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translation
Open accessAbstract
Human evaluation of modern high-quality machine translation systems is a difficult problem, and there is increasing evidence that inadequate evaluation procedures can lead to erroneous conclusions. While there has been considerable research on human evaluation, the field still lacks a commonly accepted standard procedure. As a step toward this goal, we propose an evaluation methodology grounded in explicit error analysis, based on the Multidimensional Quality Metrics (MQM) framework. We carry out the largest MQM research study to date, scoring the outputs of top systems from the WMT 2020 shared task in two language pairs using annotations provided by professional translators with access to full document context. We analyze the resulting data extensively, finding among other results a substantially different ranking of evaluated systems from the one established by the WMT crowd workers, exhibiting a clear preference for human over machine output. Surprisingly, we also find that automatic metrics based on pre-trained embeddings can outperform human crowd workers. We make our corpus publicly available for further research.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“These metrics offer notable advantages in that they are (theoretically) fully reproducible, far easier and cheaper to compute, and can avoid some of the issues faced by human studies (Wei and Jia, 2021; Freitag et al., 2021; Amidei et al., 2020).”
From Lessons from the Trenches on Reproducible Evaluation of Language Models (2024) · cites this paper as Freitag et al., 2021
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“While COMET generally aligns better with human judgement than BLEU [2]”
From Metric Inversion in Generative AI Translation: Divergence Between Automated Evaluation Metrics and Human Expert Judgement in the LRM Era (2026) · cites this paper as [2]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
chrF: character n-gram F-score for automatic MT evaluation
Popović, M. (2015) · Proceedings of the Tenth Workshop on Statistical Machine Translation · doi:10.18653/v1/w15-3049 · Cited by 1,220 · cited 1×
“In general, metrics that are not solely based on surface characteristics do somewhat better, though this pattern is not consistent (for example, chrF (Popović, 2015) has a high correlation of 0.”
BackgroundCOMET: A Neural Framework for MT Evaluation
Rei, R., Stewart, C., Farinha, A. C., & Lavie, A. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.213 · Cited by 617
Findings of the 2016 Conference on Machine Translation
Bojar, O., Chatterjee, R., Federmann, C., Graham, Y., Haddow, B., Huck, M., Jimeno Yepes, A., Koehn, P., Logacheva, V., Monz, C., et al. (2016) · Proceedings of the First Conference on Machine Translation: Volume 2, Shared Task Papers · doi:10.18653/v1/w16-2301 · Cited by 539
Achieving Human Parity on Automatic Chinese to English News Translation
Hassan, H., Aue, A., Chen, C., Chowdhary, V., Clark, J. H., Federmann, C., Huang, X., Junczys-Dowmunt, M., Lewis, W. D., Li, M., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1803.05567 · Cited by 521
Findings of the 2017 Conference on Machine Translation (WMT17)
Bojar, O., Chatterjee, R., Federmann, C., Graham, Y., Haddow, B., Huang, S., Huck, M., Koehn, P., Liu, Q., Logacheva, V., et al. (2017) · Proceedings of the Second Conference on Machine Translation · doi:10.18653/v1/w17-4717 · Cited by 395
Has Machine Translation Achieved Human Parity? A Case for Document-level Evaluation
Läubli, S., Sennrich, R., & Volk, M. (2018) · Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d18-1512 · Cited by 283 · cited 1×
“Recent papers have argued that poor human evaluation practices have led to misleading results, including erroneous claims that MT has achieved human parity (Toral, 2020; Läubli et al., 2018).”
BackgroundManual and automatic evaluation of machine translation between European languages
Koehn, P., & Monz, C. (2006) · Proceedings of the Workshop on Statistical Machine Translation - StatMT '06 · doi:10.3115/1654650.1654666 · Cited by 269 · cited 1×
“The first WMT evaluation campaign (Koehn and Monz, 2006) used adequacy and fluency ratings on a 5-point scale acquired from participants as their main metric.”
MethodLommel, A., Uszkoreit, H., & Burchardt, A. (2014) · Tradumàtica tecnologies de la traducció · doi:10.5565/rev/tradumatica.77 · Cited by 233 · cited 1×
“eu) to address the shortcomings of previous quality evaluation methods (Lommel et al., 2014).”
BackgroundBLEU might be Guilty but References are not Innocent
Freitag, M., Grangier, D., & Caswell, I. R. (2020) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.5 · Cited by 201
Self-citation
Research works printed but not linked: 1
- ALPAC, 1966, Language and Machines: Computers in Translation and Linguistics; a Report
Cited by
Anil, R., Dai, A. M., Firat, O., Johnson, M., Lepikhin, D., Passos, A., Shakeri, S., Taropa, E., Bailey, P., Chen, Z., et al. (2023) · arXiv · doi:10.48550/arxiv.2305.10403 · Cited by 834
“MQM represents the average errors per segment, with lower numbers indicating better results.”
MethodInfluentialSelf-citationSurvey of Low-Resource Machine Translation
Haddow, B., Bawden, R., Barone, A. V. M., Helcl, J., & Birch, A. (2022) · Computational Linguistics · doi:10.1162/coli_a_00446 · Cited by 352
One Embedder, Any Task: Instruction-Finetuned Text Embeddings
Su, H., Shi, W., Kasai, J., Wang, Y., Hu, Y., Ostendorf, M., Yih, W.-T., Smith, N. A., Zettlemoyer, L., & Yu, T. (2023) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2023.findings-acl.71 · Cited by 404
Large Language Models Are State-of-the-Art Evaluators of Translation Quality
Kocmi, T., & Federmann, C. (2023) · arXiv · doi:10.48550/arxiv.2302.14520 · Cited by 122
“(2020) and Freitag et al. (2021b) changed this back to the 2011 version.”
BackgroundMethodInfluentialRepairing the Cracked Foundation: A Survey of Obstacles in Evaluation Practices for Generated Text
Gehrmann, S., Clark, E., & Sellam, T. (2023) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.13715 · Cited by 123
Exploring Human-Like Translation Strategy with Large Language Models
He, Z., Liang, T., Jiao, W., Zhang, Z., Yang, Y., Wang, R., Tu, Z., Shi, S., & Wang, X. (2024) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00642 · Cited by 271
To Ship or Not to Ship: An Extensive Evaluation of Automatic Metrics for Machine TranslationPreprint
Kocmi, T., Federmann, C., Grundkiewicz, R., Junczys-Dowmunt, M., Matsushita, H., & Menezes, A. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.10821 · Cited by 173
Prompting PaLM for Translation: Assessing Strategies and Performance
Vilar, D., Freitag, M., Cherry, C., Luo, J., Ratnakar, V., & Foster, G. F. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.859 · Cited by 343
Self-citationTRUE: Re-evaluating Factual Consistency Evaluation
Honovich, O., Aharoni, R., Herzig, J., Taitelbaum, H., Kukliansy, D., Cohen, V., Scialom, T., Szpektor, I., Hassidim, A., & Matias, Y. (2022) · arXiv · doi:10.48550/arxiv.2204.04991 · Cited by 84
“This is in line with the findings of Freitag et al. (2021), who showed that in some cases, metrics may be better than non-expert annotators.”
BackgroundResultFindings of the IWSLT 2022 Evaluation Campaign
Anastasopoulos, A., Barrault, L., Bentivogli, L., Boito, M. Z., Bojar, O., Cattoni, R., Currey, A., Dinu, G., Duh, K., Elbayad, M., et al. (2022) · International Conference on Spoken Language Translation (IWSLT) · doi:10.18653/v1/2022.iwslt-1.10 · Cited by 175
“MQM has been used in recent MT evaluation studies (Freitag et al., 2021a) and WMT Metrics shared task (Freitag et al., 2021b).”
MethodInfluentialBartsch, A., Neuberger, C., Stark, B., Karnowski, V., Maurer, M., Pentzold, C., Quandt, T., Quiring, O., & Schemer, C. (2025) · Communication Theory · doi:10.1093/ct/qtae020 · Cited by 88
Findings of the WMT 2023 Shared Task on Quality Estimation
Blain, F., Zerva, C., Ribeiro, R., Guerreiro, N. M., Kanojia, D., de Souza, J. G. C., Silva, B., Vaz, T., Jingxuan, Y., Azadi, F., et al. (2023) · Conference on Machine Translation (WMT) · doi:10.18653/v1/2023.wmt-1.52 · Cited by 148
“As training data , we used the annotations released for the Metrics and QE shared tasks in the previous years (Freitag et al., 2021a,b).”
MethodLu, Q., Qiu, B., Ding, L., Xie, L., & Tao, D. (2023) · Preprints.org · doi:10.20944/preprints202303.0255.v1 · Cited by 110
Karpinska, M., & Iyyer, M. (2023) · Conference on Machine Translation (WMT) · doi:10.18653/v1/2023.wmt-1.41 · Cited by 226
TRUE: Re-evaluating Factual Consistency Evaluation
Honovich, O., Aharoni, R., Herzig, J., Taitelbaum, H., Kukliansy, D., Cohen, V., Scialom, T., Szpektor, I., Hassidim, A., & Matias, Y. (2022) · Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies · doi:10.18653/v1/2022.naacl-main.287 · Cited by 58
x comet : Transparent Machine Translation Evaluation through Fine-grained Error Detection
Guerreiro, N. M., Rei, R., van Stigt, D., Coheur, L., Colombo, P., & Martins, A. F. T. (2024) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00683 · Cited by 521
FINDINGS OF THE IWSLT 2023 EVALUATION CAMPAIGN
Agarwal, M., Agrawal, S., Anastasopoulos, A., Bentivogli, L., Bojar, O., Borg, C., Carpuat, M. J., Cattoni, R., Cettolo, M., Chen, M., et al. (2023) · International Conference on Spoken Language Translation (IWSLT) · doi:10.18653/v1/2023.iwslt-1.1 · Cited by 156
“We applied the error weighting by Freitag et al. (2021a).”
MethodTRUE: Re-evaluating Factual Consistency Evaluation
Honovich, O., Aharoni, R., Herzig, J., Taitelbaum, H., Kukliansy, D., Cohen, V., Scialom, T., Szpektor, I., Hassidim, A., & Matias, Y. (2022) · Proceedings of the Second DialDoc Workshop on Document-grounded Dialogue and Conversational Question Answering · doi:10.18653/v1/2022.dialdoc-1.19 · Cited by 56
“This is inline with the findings of Freitag et al. (2021), who showed that in some cases, metrics may be “better” than non-expert annotators.”
BackgroundLearning Compact Metrics for MT
Pu, A., Chung, H. W., Parikh, A. P., Gehrmann, S., & Sellam, T. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.58 · Cited by 140
“These metrics deliver superior performance over those based on lexical overlap, outper-forming even crowd-sourced annotations (Freitag et al., 2021; Mathur et al., 2020b).”
On Decoding Strategies for Neural Text Generators
Wiher, G., Meister, C., & Cotterell, R. (2022) · arXiv · doi:10.48550/arxiv.2203.15721 · Cited by 48
Evaluating Attribution in Dialogue Systems: The BEGIN Benchmark
Dziri, N., Rashkin, H., Linzen, T., & Reitter, D. (2022) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00506 · Cited by 141
Evaluating Attribution in Dialogue Systems: The BEGIN Benchmark
Dziri, N., Rashkin, H., Linzen, T., & Reitter, D. (2021) · arXiv · doi:10.48550/arxiv.2105.00071 · Cited by 47
Leveraging GPT-4 for Automatic Translation Post-Editing
Raunak, V., Sharaf, A., Wang, Y., Awadalla, H. H., & Menezes, A. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.804 · Cited by 166
Bridging the Gap: A Survey on Integrating (Human) Feedback for Natural Language Generation
Fernandes, P., Madaan, A., Liu, E., Farinhas, A., Martins, P. H., Bertsch, A., de Souza, J. G. C., Zhou, S., Wu, T., Neubig, G., et al. (2023) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00626 · Cited by 132
“Annotator expertise : Depending on task and training (Snow et al., 2008; Sheng et al., 2008; Clark et al., 2021; Gillick and Liu, 2010; Freitag et al., 2021), annotators can be domain experts to crowdworkers or even models.”
InfluentialOn Decoding Strategies for Neural Text Generators
Wiher, G., Meister, C., & Cotterell, R. (2022) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00502 · Cited by 141
LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods
Li, H., Dong, Q., Chen, J., Su, H., Zhou, Y., Ai, Q., Ye, Z., & Liu, Y. (2024) · arXiv · doi:10.48550/arxiv.2412.05579 · Cited by 40
INSTRUCTSCORE: Towards Explainable Text Generation Evaluation with Automatic Feedback
Xu, W., Wang, D., Pan, L., Song, Z., Freitag, M., Wang, W. Y., & Li, L. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.365 · Cited by 184
Self-citationGEMBA-MQM: Detecting Translation Quality Error Spans with GPT-4
Kocmi, T., & Federmann, C. (2023) · Conference on Machine Translation (WMT) · doi:10.18653/v1/2023.wmt-1.64 · Cited by 307
Quality-Aware Decoding for Neural Machine Translation
Fernandes, P., Farinhas, A., Rei, R., de Souza, J. G. C., Ogayo, P., Neubig, G., & Martins, A. F. T. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.100 · Cited by 178
Kocmi, T., Avramidis, E., Bawden, R., Bojar, O., Dvorkovich, A., Federmann, C., Fishel, M., Freitag, M., Gowda, T., Grundkiewicz, R., et al. (2023) · Conference on Machine Translation (WMT) · doi:10.18653/v1/2023.wmt-1.1 · Cited by 295
“…with the metrics shared task (Fre-itag et al., 2023), human annotation data for the Chinese → English and English → German direction was collected using two different approaches: the source-based DA+SQM approach, and the Multi-dimensional Quality Metrics (MQM) framework (Freitag et al., 2021).”
MethodSelf-citationMENLI: Robust Evaluation Metrics from Natural Language Inference
Chen, Y., & Eger, S. (2023) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00576 · Cited by 113
Fernandes, P., Deutsch, D., Finkelstein, M., Riley, P., Martins, A., Neubig, G., Garg, A., Clark, J., Freitag, M., & Firat, O. (2023) · Proceedings of the Eighth Conference on Machine Translation · doi:10.18653/v1/2023.wmt-1.100 · Cited by 31
Self-citationPerturbation CheckLists for Evaluating NLG Evaluation Metrics
Sai, A. B., Dixit, T., Sheth, D. Y., Mohan, S., & Khapra, M. M. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.575 · Cited by 80
Lankford, S., Afli, H., & Way, A. (2023) · Information · doi:10.3390/info14120638 · Cited by 70
“Description of error categories within the core MQM framework [25].”
BackgroundMethodResultInfluentialTransparent Human Evaluation for Image Captioning
Kasai, J., Sakaguchi, K., Dunagan, L., Morrison, J., Bras, R. J. L., Choi, Y., & Smith, N. A. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.254 · Cited by 65
A Critical Evaluation of Evaluations for Long-form Question Answering
Xu, F., Song, Y., Iyyer, M., & Choi, E. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.181 · Cited by 33
Reliable and Safe Use of Machine Translation in Medical Settings
Mehandru, N., Robertson, S., & Salehi, N. (2022) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3531146.3533244 · Cited by 76
“Evaluation in MT is notoriously difficult, due to the nuanced and subjective nature of human language translation [10, 18].”
BackgroundBeyond the Chat: Executable and Verifiable Text-Editing with LLMs
Laban, P., Vig, J., Hearst, M., Xiong, C., & Wu, C.-S. (2024) · Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology · doi:10.1145/3654777.3676419 · Cited by 31
“The inefciency in the NoEx Chat condition is probably caused by the need for participants to split their time and attention between the chat and the manual integration of the chat’s suggestions into the draft.”
BackgroundRay, M., Kats, D. J., Moorkens, J., Rai, D., Shaar, N., Quinones, D., Vermeulen, A., Mateo, C. M., Brewster, R., Khan, A., et al. (2025) · Archives of Pediatrics and Adolescent Medicine · doi:10.1001/jamapediatrics.2025.1729 · Cited by 40
An Evaluation of ChatGPT's Translation Accuracy Using BLEU Score
Ghassemiazghandi, M. (2024) · Theory and Practice in Language Studies · doi:10.17507/tpls.1404.07 · Cited by 40
“Consequently, extensive human evaluation is required to provide a dependable benchmark for assessing the quality and advancement of MT (Freitag et al., 2021).”
BackgroundEmbeddings from language models are good learners for single-cell data analysis
Liu, T., Chen, T., Zheng, W., Luo, X., Chen, Y. T., & Zhao, H. (2026) · Patterns · doi:10.1016/j.patter.2025.101431 · Cited by 44
“43 The BLEU score is used to evaluate the similarity between observed string ̂ y and ground-truth string y based on the n - grams strategy.”
High Quality Rather than High Model Probability: Minimum Bayes Risk Decoding with Neural Metrics
Freitag, M., Grangier, D., Tan, Q., & Liang, B. (2022) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00491 · Cited by 152
“We hired 9 professional translators (4 for En → De and 5 for De → En) and measure translation quality with an document context version of MQM (Lom-mel et al., 2014) which mimics the setup proposed in Freitag et al. (2021a).”
BackgroundMethodInfluentialSelf-citationJiao, W., Huang, J.-T., Wang, W., He, Z., Liang, T., Wang, X., Shi, S., & Tu, Z. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.1001 · Cited by 113
Underreporting of errors in NLG output, and what to do about it
van Miltenburg, E., Clinciu, M., Dušek, O., Gkatzia, D., Inglis, S., Leppänen, L., Mahamood, S., Manning, E., Schoch, S., Thomson, C., et al. (2021) · International Conference on Natural Language Generation · doi:10.18653/v1/2021.inlg-1.14 · Cited by 57
scELMo: Embeddings from Language Models are Good Learners for Single-cell Data Analysis
Liu, T., Chen, T., Zheng, W., Luo, X., Chen, Y., & Zhao, H. (2023) · doi:10.1101/2023.12.07.569910 · Cited by 30
“Furthermore, we loaded the gene functional annotation information provided by Geneformer [9] and used the known information to predict the functional information for the rest of genes based on a k-nearest-neighbor (kNN) classifier.”
InfluentialDiscoScore: Evaluating Text Generation with BERT and Discourse Coherence
Zhao, W., Strube, M., & Eger, S. (2023) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2023.eacl-main.278 · Cited by 81
A survey of context in neural machine translation and its evaluation
Castilho, S., & Knowles, R. (2024) · Natural language processing. · doi:10.1017/nlp.2024.7 · Cited by 30
Currey, A., Nădejde, M., Pappagari, R. R., Mayer, M., Lauly, S., Niu, X., Hsu, B., & Dinu, G. (2022) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2022.emnlp-main.288 · Cited by 27
Learning with Different Amounts of Annotation: From Zero to Many Labels
Zhang, S., Gong, C., & Choi, E. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.601 · Cited by 38
“Even after thorough quality control, it is often infeasible to reach complete annotator agreement, as annotators make mistakes (Freitag et al., 2021) and ambiguity is a key feature of human communication (Asher and Lascarides, 2005).”
BackgroundHuman Evaluation of English–Irish Transformer-Based NMT
Lankford, S., Afli, H., & Way, A. (2022) · Information · doi:10.3390/info13070309 · Cited by 31
“Furthermore, the non-translation category was allocated a weight of 25, an approach which is line with the best practice established in previous studies [33].”
BackgroundResultInfluentialThe Eval4NLP Shared Task on Explainable Quality Estimation: Overview and Results
Fomicheva, M., Lertvittayakumjorn, P., Zhao, W., Eger, S., & Gao, Y. (2021) · Workshop on Evaluation and Comparison of NLP Systems (Eval4NLP) · doi:10.18653/v1/2021.eval4nlp-1.17 · Cited by 46
Angler: Helping Machine Translation Practitioners Prioritize Model Improvements
Robertson, S., Wang, Z. J., Moritz, D., Kery, M. B., & Hohman, F. (2023) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3580790 · Cited by 29
Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand
Kasai, J., Sakaguchi, K., Bras, R. J. L., Dunagan, L., Morrison, J., Fabbri, A. R., Choi, Y., & Smith, N. A. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.259 · Cited by 34
RetGen: A Joint Framework for Retrieval and Grounded Text Generation Modeling
Zhang, Y., Sun, S., Gao, X., Fang, Y., Brockett, C., Galley, M., Gao, J., & Dolan, B. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i10.21429 · Cited by 27
Ties Matter: Meta-Evaluating Modern Metrics with Pairwise Accuracy and Tie Calibration
Deutsch, D., Foster, G. F., & Freitag, M. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.798 · Cited by 118
“Second, ties in MQM scores are very common.”
BackgroundMethodInfluentialSelf-citationUncertainty-Aware Machine Translation Evaluation
Glushkova, T., Zerva, C., Rei, R., & Martins, A. F. T. (2021) · Findings of the Association for Computational Linguistics: EMNLP 2021 · doi:10.18653/v1/2021.findings-emnlp.330 · Cited by 17
“This problem can be mitigated by averaging over a sufficient number of references, but this brings considerable annotation costs (Freitag et al., 2021; Mathur et al., 2020).”
BackgroundMethodInfluentialResults of WMT23 Metrics Shared Task: Metrics Might Be Guilty but References Are Not Innocent
Freitag, M., Mathur, N., Lo, C.-K., Avramidis, E., Rei, R., Thompson, B. J., Kocmi, T., Blain, F., Deutsch, D., Stewart, C., et al. (2023) · Conference on Machine Translation (WMT) · doi:10.18653/v1/2023.wmt-1.51 · Cited by 202
Self-citationAziz, M., Rehman, U., Danish, M. U., & Grolinger, K. (2025) · IEEE Transactions on Human-Machine Systems · doi:10.1109/thms.2025.3527397 · Cited by 34
“However, we acknowledge that a ranking-based approach could offer an alternative perspective by focusing on relative comparisons between images, potentially reducing subjectivity in responses [44].”
Howcroft, D. M., & Rieser, V. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.703 · Cited by 33
“Similar studies have been conducted for machine translation (Freitag et al., 2021).”
ResultThe statistical advantage of automatic NLG metrics at the system level
Wei, J., & Jia, R. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.533 · Cited by 16
Non-parametric Online Learning from Human Feedback for Neural Machine Translation
Wang, D., Wei, H., Zhang, Z., Huang, S., Xie, J., & Chen, J. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i10.21395 · Cited by 29
Findings of the Second WMT Shared Task on Sign Language Translation (WMT-SLT23)
Müller, M., Alikhani, M., Avramidis, E., Bowden, R., Braffort, A., Camgöz, N. C., Ebling, S., España-Bonet, C., Göhring, A., Grundkiewicz, R., et al. (2023) · Conference on Machine Translation (WMT) · doi:10.18653/v1/2023.wmt-1.4 · Cited by 50
“We employed the source-based direct assessment (DA; Graham et al., 2013; Cet-tolo et al., 2017) methodology with document context, extended with Scalar Quality Metric (SQM; Freitag et al., 2021).”
MethodTranslation Error Detection as Rationale Extraction
Fomicheva, M., Specia, L., & Αλέτρας, Ν. (2022) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2022.findings-acl.327 · Cited by 25
Source Code Vulnerability Detection Based on Joint Graph and Multimodal Feature Fusion
Jin, D., He, C., Zou, Q., Qin, Y., & Wang, B. (2025) · Electronics · doi:10.3390/electronics14050975 · Cited by 17
Extrinsic Evaluation of Machine Translation Metrics
Moghe, N., Sherborne, T., Steedman, M. J., & Birch, A. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.730 · Cited by 16
SOYSAL, F. (2023) · Karamanoğlu Mehmetbey Üniversitesi Uluslararası Filoloji ve Çeviribilim Dergisi · doi:10.55036/ufced.1402649 · Cited by 30
UScore: An Effective Approach to Fully Unsupervised Evaluation Metrics for Machine Translation
Belouadi, J., & Eger, S. (2023) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2023.eacl-main.27 · Cited by 22
“We do not evaluate at the system level because metrics there often perform very similarly and achieve very high correlations, making it difficult to determine the best metric (Mathur et al., 2020b; Freitag et al., 2021b).”
BackgroundMethodInfluentialGENIE: Toward Reproducible and Standardized Human Evaluation for Text Generation
Khashabi, D., Stanovsky, G., Bragg, J., Lourie, N., Kasai, J., Choi, Y., Smith, N. A., & Weld, D. S. (2022) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2022.emnlp-main.787 · Cited by 80
“Human evaluation for machine translation (Graham et al., 2013, 2014; Sak-aguchi and Van Durme, 2018; Freitag et al., 2021) typically involves crowdsourcing where qualified crowd workers score output translations given the reference text.”
BackgroundThe Eval4NLP 2023 Shared Task on Prompting Large Language Models as Explainable Metrics
Leiter, C., Opitz, J., Deutsch, D., Gao, Y., Dror, R., & Eger, S. (2023) · Workshop on Evaluation and Comparison of NLP Systems (Eval4NLP) · doi:10.18653/v1/2023.eval4nlp-1.10 · Cited by 40
Krüger, R. (2022) · Translation Spaces · doi:10.1075/ts.21026.kru · Cited by 13
Yan, Z., Li, Y., Zhang, P., Yang, Z., & Zhao, R. (2023) · The Journal of Supercomputing · doi:10.1007/s11227-023-05224-0 · Cited by 13
“NER is a task to detect real entities from text and classify them by predefined types, and it is an indispensable part of various natural language processing technologies such as information extraction [2], information retrieval [3], machine translation [4], question answering system [5], and so on.”
BackgroundUsing chat GPT to evaluate police threats, risk and harm
Halford, E., & Webster, A. (2024) · International journal of law, crime and justice · doi:10.1016/j.ijlcj.2024.100686 · Cited by 20
Evaluating factual accuracy in complex data-to-text
Thomson, C., Reiter, E., & Sundararajan, B. (2023) · Computer Speech & Language · doi:10.1016/j.csl.2023.101482 · Cited by 32
Mehandru, N., Agrawal, S., Xiao, Y., Gao, G., Khoong, E. C., Carpuat, M. J., & Salehi, N. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.712 · Cited by 42
A Natural Diet: Towards Improving Naturalness of Machine Translation Output
Freitag, M., Vilar, D., Grangier, D., Cherry, C., & Foster, G. F. (2022) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2022.findings-acl.263 · Cited by 27
“However MT output still can be improved: it currently trails human translators in expert evaluation (Toral et al., 2018; Freitag et al., 2021) and its language is perceived as poorer and more synthetic (Vanmassenhove et al., 2021).”
BackgroundSelf-citationMiret, J. R., Farré-Maduell, E., López, S. L., Vigil, L., Brivá-Iglesias, V., & Krallinger, M. (2024) · Information · doi:10.3390/info15100585 · Cited by 13
The quality optimization of English–Chinese machine translation based on deep neural networks
Lu, P., & Xu, F. (2025) · Discover Artificial Intelligence · doi:10.1007/s44163-025-00319-4 · Cited by 15
Chen, S., & Lin, Y. (2025) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2025.1619489 · Cited by 35
English-Catalan Neural Machine Translation: state-of-the-art technology, quality, and productivity
Brivá-Iglesias, V. (2022) · Tradumàtica tecnologies de la traducció · doi:10.5565/rev/tradumatica.303 · Cited by 12
NLG Evaluation Metrics Beyond Correlation Analysis: An Empirical Metric Preference Checklist
Ni’mah, I., Fang, M., Menkovski, V., & Pechenizkiy, M. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.69 · Cited by 15
Toxicity in Multilingual Machine Translation at Scale
Costa‐jussà, M. R., Smith, E., Ropers, C., Licht, D. J., Maillard, J., Ferrando, J., & Escolano, C. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.642 · Cited by 47
Hierarchical template transformer for fine-grained sentiment controllable generation
Li, Y., Wang, J., Yu, L.-C., & Zhang, X. (2022) · Information Processing & Management · doi:10.1016/j.ipm.2022.103048 · Cited by 17
Toward Human-Like Evaluation for Natural Language Generation with Error Analysis
Lu, Q., Ding, L., Xie, L., Zhang, K., Wong, D. F., & Tao, D. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.324 · Cited by 36
Machine Translation Meta Evaluation through Translation Accuracy Challenge Sets
Moghe, N., Fazla, A., Amrhein, C., Kocmi, T., Steedman, M. J., Birch, A., Sennrich, R., & Guillou, L. (2024) · Computational Linguistics · doi:10.1162/coli_a_00537 · Cited by 24
Enhancing pathogen identification through AI-assisted metagenomic sequencing
Peng, X., Wei, Y. X., & Zhou, X. (2025) · Frontiers in Microbiology · doi:10.3389/fmicb.2025.1634194 · Cited by 14
Not All Errors are Equal: Learning Text Generation Metrics using Stratified Error Synthesis
Xu, W., Tuan, Y.-L., Lu, Y., Saxon, M., Li, L., & Wang, W. Y. (2022) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2022.findings-emnlp.489 · Cited by 33
Popović, M. (2021) · Conference on Computational Natural Language Learning · doi:10.18653/v1/2021.conll-1.18 · Cited by 9
“Most publications dealing with human evaluation of MT, such as (Vilar et al., 2007; CallisonBurch et al., 2008; Klubička et al., 2018; Kreutzer et al., 2020; Popović, 2020; Castilho, 2020; Freitag et al., 2021), report an overall inter-annotator agreement (IAA) score such as percentage of equal labels, Koehn’s κ, Flei…”
BackgroundIndicMT Eval: A Dataset to Meta-Evaluate Machine Translation Metrics for Indian Languages
Ananya Sai, B., Dixit, T., Nagarajan, V., Kunchukuttan, A., Kumar, P., Khapra, M. M., & Dabre, R. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.795 · Cited by 52
Joint Retrieval and Generation Training for Grounded Text Generation.Preprint
Zhang, Y., Sun, S., Gao, X., Fang, Y., Brockett, C., Galley, M., Gao, J., & Dolan, B. (2021) · arXiv (Cornell University) · Cited by 13
T5Score: Discriminative Fine-tuning of Generative Evaluation Metrics
Qin, Y., Yuan, W., Neubig, G., & Liu, P. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.1014 · Cited by 24
Common Flaws in Running Human Evaluation Experiments in NLP
Thomson, C., Reiter, E., & Belz, A. (2024) · Computational Linguistics · doi:10.1162/coli_a_00508 · Cited by 8
“Experimental design issues such as the choice of evaluation metric (e.g., Kocmi et al. 2021) and human evaluation method (e.g., Freitag et al. 2021) are often discussed in the NLP literature, as is, more recently, the question of how these impact the reproducibility of results (e.g., Belz et al.…”
BackgroundMountantonakis, M., Mertzanis, L., Bastakis, M., & Tzitzikas, Y. (2024) · Language Resources and Evaluation · doi:10.1007/s10579-024-09745-9 · Cited by 10
“Moreover, there exists b) traditional string-matching based metrics based on n-grams such as [31, 32], BLEU [33], METEOR [34] ROUGE [35], and chrF [36] which is better for high-morphology languages, since it does not penalise very much differences related to endings, prefixes, etc. Finally, there are…”
Aldosari, L. A., & Altuwairesh, N. S. (2025) · Perspectives · doi:10.1080/0907676x.2025.2464120 · Cited by 16
Mukherjee, A., & Shrivastava, M. (2025) · ACM Computing Surveys · doi:10.1145/3735970 · Cited by 19
“[38] 2021 MQMAnnotation Pearson[58],Kendall[49], Spearman[101] en-de,zh-en WMT20 Pearson[58] zh-en,de-en,ru-en,cs-en,ja-en,uk-en,ha-en,pl-en,is-en,ta-en, km-en,ps-en,en-de,en-ruen-hi,en-gu,en-mr,en-ml,en-ta WMT-DA(20-22),WMT-MQM(20-22), IndicMTEval22”
Dancing Between Success and Failure: Edit-level Simplification Evaluation using SALSA
Heineman, D., Dou, Y., Maddela, M., & Xu, W. H. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.211 · Cited by 15
SEAHORSE: A Multilingual, Multifaceted Dataset for Summarization Evaluation
Clark, E., Rijhwani, S., Gehrmann, S., Maynez, J., Aharoni, R., Nikolaev, V. A., Sellam, T., Siddhant, A., Das, D., & Parikh, A. P. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.584 · Cited by 60
Aleedy, M., Alshihri, F., Meshoul, S., Al-Harthi, M., Alramlawi, S., Aldaihani, B. H., Shaiba, H. A., & Atwell, E. (2025) · PeerJ Computer Science · doi:10.7717/peerj-cs.2788 · Cited by 16
“Evaluators utilized a standardized rubric inspired by the Multidimensional Quality Metrics (MQM) framework ( Freitag et al., 2021 ; Lommel et al., 2024 ), and were familiarized with the evaluation criteria, which covered addition, omission, and mispairing errors.”
InfluentialValido, J. V. (2021) · Tradumàtica tecnologies de la traducció · doi:10.5565/rev/tradumatica.285 · Cited by 16
Yang, F., & Deng, J. (2023) · Scientific Reports · doi:10.1038/s41598-023-31053-5 · Cited by 8
FRMT: A Benchmark for Few-Shot Region-Aware Machine Translation
Riley, P., Dozat, T., Botha, J. A., Garcia, X., Garrette, D., Riesa, J., Fırat, O., & Constant, N. (2023) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00568 · Cited by 35
Metadata
- Authors
- Freitag, M., Foster, G., Grangier, D., Ratnakar, V., Tan, Q., & Macherey, W.
- Affiliations
- Google (United States) (US)
- Corresponding author
- Freitag, Markus
- Foster, George
- Grangier, David
- Ratnakar, Viresh
- Tan, Qijun
- Macherey, Wolfgang
- ORCID
- Grangier, David
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2021
- Volume
- 9
- Pages
- 1460-1474
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W3159892921
- Cited by
- 271
- References
- 43
- Citations by year
- 2026: 45 · 2025: 48 · 2024: 36 · 2023: 73 · 2022: 45 · 2021: 24
- FWCI
- 15.59
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Locations
- Transactions of the Association for Computational Linguistics (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Social Sciences › Arts and Humanities › Language and Linguistics › Translation Studies and Practices
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- human evaluation of machine translation, Multidimensional Quality Metrics, error analysis, evaluation methodology, professional translators, crowd workers, automatic evaluation metrics, pre-trained embeddings, human-machine comparison
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.