A Comparative Study of Linguistic Bias and Fairness in Large Language Models for Automated Essay Scoring

Authors

  • Leila Diaz Graduate School of Education, Rutgers University–New Brunswick, USA
  • Liam Sato College of Education, Clemson University, USA
  • Jasmine Yoon College of Education, Purdue University, USA

DOI:

https://doi.org/10.54097/navkme51

Keywords:

Automated essay scoring, large language models, linguistic bias, counterfactual fairness, African American English, English learners, debiasing

Abstract

Large language model (LLM) representations have become the default backbone of automated essay scoring, yet the resulting scores may react to construct-irrelevant features of African American English (AAE) and learner (L2) English. We present a comparative counterfactual study of linguistic bias across five representation families—word and character n-grams, static word vectors, LLM-derived static token embeddings (WordLlama) and RoBERTa examined layer by layer—on 5,569 ASAP essays. Real essays are rewritten with eight AAE morphosyntactic rules and with learner errors whose rates and misspellings are estimated from JFLEG and TOEFL-Spell, and every edit is labelled so that score shifts can be attributed to fourteen phenomena. All unmitigated scorers lower the scores of counterfactual variants, but the mean shifts stay below 0.06 standard deviations while 4.0–9.7% of individual reported scores change, so group-level screening would accept scorers that alter individual outcomes. Within RoBERTa, sensitivity to dialect grows with depth: the AAE penalty increases from −0.015 standard deviations at the embedding output to −0.077 at layer 10, and the contextual model penalises 13 of the 14 phenomena. Construct-aligned invariance regularisation lowers counterfactual flip rates on content scores for every family (by 28–60%) with QWK changes between −0.007 and +0.004. We extend it with variety-selective construct maps and sensitivity anchoring: on the Conventions trait of RoBERTa, the extension removes 80% of the AAE penalty while retaining 98% of the learner-error penalty, whereas a uniform map retains 66%. A geometric bound and the measured overlap of dialect and learner displacement directions show why surface-identical phenomena, such as a missing third-person -s, cannot be treated differently by any linear scorer.

Downloads

Download data is not yet available.

References

[1] Shermis, M. D. (2014). State-of-the-art automated essay scoring: Competition, results, and future directions from a United States demonstration. Assessing Writing, 20, 53–76. https://doi.org/10.1016/j.asw.2013.04.001

[2] Taghipour, K., & Ng, H. T. (2016). A neural approach to automated essay scoring. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 1882–1891). https://doi.org/10.18653/v1/D16-1193

[3] Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics (NAACL-HLT) (pp. 4171–4186). https://doi.org/10.18653/v1/N19-1423

[4] Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019). RoBERTa: A robustly optimized BERT pretraining approach. arXiv:1907.11692.

[5] Mayfield, E., & Black, A. W. (2020). Should you fine-tune BERT for automated essay scoring? In Proceedings of the 15th Workshop on Innovative Use of NLP for Building Educational Applications (BEA) (pp. 151–162). https://doi.org/10.18653/v1/2020.bea-1.15

[6] Mizumoto, A., & Eguchi, M. (2023). Exploring the potential of using an AI language model for automated essay scoring. Research Methods in Applied Linguistics, 2(2), 100050. https://doi.org/10.1016/j.rmal.2023.100050

[7] Williamson, D. M., Xi, X., & Breyer, F. J. (2012). A framework for evaluation and use of automated scoring. Educational Measurement: Issues and Practice, 31(1), 2–13. https://doi.org/10.1111/j.1745-3992.2011.00223.x

[8] Bridgeman, B., Trapani, C., & Attali, Y. (2012). Comparison of human and machine scoring of essays: Differences by gender, ethnicity, and country. Applied Measurement in Education, 25(1), 27–40. https://doi.org/10.1080/08957347.2012.635502

[9] Loukina, A., Madnani, N., & Zechner, K. (2019). The many dimensions of algorithmic fairness in educational applications. In Proceedings of the 14th Workshop on Innovative Use of NLP for Building Educational Applications (BEA) (pp. 1–10). https://doi.org/10.18653/v1/W19-4401

[10] Yang, K., Raković, M., Li, Y., Guan, Q., Gašević, D., & Chen, G. (2024). Unveiling the tapestry of automated essay scoring: A comprehensive investigation of accuracy, fairness, and generalizability. Proceedings of the AAAI Conference on Artificial Intelligence, 38(20), 22466–22474. https://doi.org/10.1609/aaai.v38i20.30254

[11] Schaller, N.-J., Ding, Y., Horbach, A., Meyer, J., & Jansen, T. (2024). Fairness in automated essay scoring: A comparative analysis of algorithms on German learner essays from secondary education. In Proceedings of the 19th Workshop on Innovative Use of NLP for Building Educational Applications (BEA) (pp. 210–221).

[12] Wang, J., Ge, Y., Liu, J., & Jiang, B. (2024). Early warning of liquidity stress in U.S. banks with seasonally aligned frailty hazards and delayed-feedback conformal alarms. Journal of Trends in Finance and Economics, 1(1), 85–100.

[13] Ziems, C., Chen, J., Harris, C., Anderson, J., & Yang, D. (2022). VALUE: Understanding dialect disparity in NLU. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 3701–3720). https://doi.org/10.18653/v1/2022.acl-long.258

[14] Hofmann, V., Kalluri, P. R., Jurafsky, D., & King, S. (2024). AI generates covertly racist decisions about people based on their dialect. Nature, 633(8028), 147–154. https://doi.org/10.1038/s41586-024-07856-5

[15] Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., & Zou, J. (2023). GPT detectors are biased against non-native English writers. Patterns, 4(7), 100779. https://doi.org/10.1016/j.patter.2023.100779

[16] Blodgett, S. L., Barocas, S., Daumé III, H., & Wallach, H. (2020). Language (technology) is power: A critical survey of “bias” in NLP. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 5454–5476). https://doi.org/10.18653/v1/2020.acl-main.485

[17] Fang, Q. (2024). Bias-aware large language models for equitable educational feedback and assessment. Trends in Social Science and Humanities Research, 2(1), 41–50. https://doi.org/10.61784/tsshr4253

[18] Tenney, I., Das, D., & Pavlick, E. (2019). BERT rediscovers the classical NLP pipeline. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 4593–4601). https://doi.org/10.18653/v1/P19-1452

[19] Liu, N. F., Gardner, M., Belinkov, Y., Peters, M. E., & Smith, N. A. (2019). Linguistic knowledge and transferability of contextual representations. In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics (NAACL-HLT) (pp. 1073–1094). https://doi.org/10.18653/v1/N19-1112

[20] Napoles, C., Sakaguchi, K., & Tetreault, J. (2017). JFLEG: A fluency corpus and benchmark for grammatical error correction. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics (EACL) (Vol. 2, pp. 229–234).

[21] Flor, M., Fried, M., & Rozovskaya, A. (2019). A benchmark corpus of English misspellings and a minimally-supervised model for spelling correction. In Proceedings of the 14th Workshop on Innovative Use of NLP for Building Educational Applications (BEA) (pp. 76–86). https://doi.org/10.18653/v1/W19-4407

[22] Kusner, M. J., Loftus, J., Russell, C., & Silva, R. (2017). Counterfactual fairness. Advances in Neural Information Processing Systems, 30, 4066–4076.

[23] Garg, S., Perot, V., Limtiaco, N., Taly, A., Chi, E. H., & Beutel, A. (2019). Counterfactual fairness in text classification through robustness. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (AIES) (pp. 219–226). https://doi.org/10.1145/3306618.3317950

[24] Zmigrod, R., Mielke, S. J., Wallach, H., & Cotterell, R. (2019). Counterfactual data augmentation for mitigating gender stereotypes in languages with rich morphology. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 1651–1661). https://doi.org/10.18653/v1/P19-1161

[25] Teng, D. (2025). RSRBench: Reproducibility and performance benchmarking for regional seismic risk workflows. Academic Journal of Architecture and Civil Engineering, 3(1), 40–52.

[26] Ravfogel, S., Elazar, Y., Gonen, H., Twiton, M., & Goldberg, Y. (2020). Null it out: Guarding protected attributes by iterative nullspace projection. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 7237–7256). https://doi.org/10.18653/v1/2020.acl-main.647

[27] Honnibal, M., Montani, I., Van Landeghem, S., & Boyd, A. (2020). spaCy: Industrial-strength natural language processing in Python. Zenodo. https://doi.org/10.5281/zenodo.1212303

[28] Hoerl, A. E., & Kennard, R. W. (1970). Ridge regression: Biased estimation for nonorthogonal problems. Technometrics, 12(1), 55–67. https://doi.org/10.1080/00401706.1970.10488634

[29] Touvron, H., et al. (2023). Llama 2: Open foundation and fine-tuned chat models. arXiv:2307.09288.

[30] Miller, D. L. (2024). WordLlama: Recycled token embeddings from large language models. https://github.com/dleemiller/wordllama

[31] Wang, Z., et al. (2026). FairPromote: Explainable and Fairness-Aware Talent Promotion Prediction via Adversarial Debiasing and SHAP-Based Interpretation. IEEE Access. https://doi.org/10.1109/ACCESS.2026.3692578

[32] Zou, J., Feng, E., Li, P., & Wang, G. (2026). Horizon-aware group-conditional conformal prediction: Post-hoc shrinkage for sparse multi-horizon coverage auditing. Machine Learning with Applications, 26, 101038. https://doi.org/10.1016/j.mlwa.2026.101038

Downloads

Published

17-11-2026

Issue

Section

Articles

How to Cite

Diaz, L., Sato, L., & Yoon, J. (2026). A Comparative Study of Linguistic Bias and Fairness in Large Language Models for Automated Essay Scoring. International Journal of World Economic Research, 4(1), 34-42. https://doi.org/10.54097/navkme51