United States

This paper addresses the problem of fine-grained error classification in stepwise algebraic problem solving, with the objective of enabling accurate and timely feedback in large-scale educational environments. Using authentic student response data, we compare a carefully engineered rule-based baseline with large language models (LLMs) in zero-shot and few-shot configurations, as well as multistep LLM-based approaches. We further consider hybrid architectures that combine symbolic computation with LLM inferential processes, with particular emphasis on enhancing the robustness and faithfulness of intermediate representations and mitigating error propagation across successive stages of the computational pipeline. Our empirical results indicate that, although the baseline model delivers strong and reliable performance for narrowly defined error categories, structured multi-step approaches improve performance relative to single-step methods by achieving superior precision, F1 scores, and overall accuracy.

ACL 2026

Multi-step Large Language Model for Fine-Grained Feedback in Stepwise Linear Equation Solutions

sign-error

linear-equations

multi-step-llm

feedback

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. <br>
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) <br><br><br>
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>
<br><br><br><br>

To access this event page, you need to log in with the **email address you registered with**. <br>Access credentials will be sent to your email from Underline -  subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

Registration Required

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

workshop paper

Russian grammar correction models can improve
on aggregate benchmarks
while getting worse at specific grammar rules.
We show this through per-rule evaluation
on a diagnostic benchmark of 48 prescriptive rules:
finetuning on synthetic data improves overall F$_{0.5}$
while driving subordinate-clause comma accuracy
from 14\% to 1\%.
The suppression is invisible under corpus-level metrics
and undetectable with existing coarse, corpus-specific tagsets;
it is recoverable only when diagnosed at rule granularity.
To enable this analysis,
we develop a 98-category error taxonomy
grounded in Rozental's reference grammar
and SyntErr, an open-source synthetic data generator
whose per-rule distribution is an explicit parameter,
designed to support arbitrary rule sets and languages.
Finetuning eight open models (0.8B--12B)
on 39K synthetic examples
yields up to 75.3 F$_{0.5}$,
approaching frontier API models
with models small enough to run on device.
We release the taxonomy, generator,
per-rule evaluation data, and all training artifacts.

What Aggregate Scores Hide: Per-Rule Evaluation of Russian Grammatical Error Correction

This paper looks at the impact of annotators sharing a language background with learners when annotating learner data using the Universal Dependencies (UD) framework. We perform a study comparing annotations by two different annotators working on sets of L2 Swedish sentences (learner sentences and target corrections) from the Swedish Learner Language corpus (SweLL) written by learners for whom French is a main writing language. The annotators are both L2 speakers of Swedish but have different knowledge of French: one is a native French speaker and the other has no knowledge of French. We find high annotator agreement, which may indicate an non-significant impact, though we qualitatively observe an advantage in sharing language background.

Sharing is Caring: Advantages of Sharing a Language Background with Learners as an Annotator of Learner Data in UD

Automated Writing Evaluation (AWE) platforms have become common, but a significant gap remains between automated assessment and expert human feedback. We address this gap by introducing a supervised learning method that uses a multi-component student writing profile (comprising estimated CEFR levels, grammatical error rates, and vocabulary distribution) to align AI scoring with expert human judgments. In the context of an online essay-writing platform for second language learners of English, our model achieves a 36% reduction in RMSE for holistic essay scoring and an 84% improvement in similarity to human-expert annotation of grammatical errors compared to automarker scores (26% and 57% improvement from the best-performing comparable earlier work, by Zaidi et al. (2019) . Furthermore, we demonstrate that the model can predict a student's final submission profile (CEFR level and grammatical error rate) from earlier drafts and that predictions generalize to a subsequent task, offering new possibilities for automated curriculum planning. Finally, we introduce a visualization tool that provides educators with clear expert-aligned longitudinal views of student development.

Multi-component student writing profiles for expert-aligned automated evaluation of English learner essays.

When programming students encounter errors in their code, compiler messages or static analysis output often provide limited guidance, particularly for novice programmers. Personalized feedback from instructors can be effective but does not scale well. Recent advances in large language models (LLMs) enable automated feedback generation at scale.
This study examines whether LLM-generated feedback with different levels of guidance is associated with differences in students' problem-solving behavior. We analyze effects on time to solution and number of attempts, and examine whether these effects differ by programming experience. We design three feedback types and compare them to a baseline in which students receive only compiler error messages. Results from an online programming course show that LLM-generated feedback is associated with faster time to solution compared to the no-feedback baseline, with less guided feedback showing slightly stronger effects. 
Overall, the findings suggest that feedback structure plays an important role in how students progress toward correct solutions and motivate further work on adaptive feedback designs and longer-term learning outcomes.

The Effects of Structured LLM-Generated Feedback on Programming Assignment Performance

This paper investigates whether automated short answer grading can reliably support teachers when marking conceptual physics responses in settings with limited labelled data. Using free-text responses derived from Force Concept Inventory-style questions, the study shows that incorporating subject-specific knowledge improves grading consistency, particularly in early deployment scenarios. The system reduces grading errors and provides more reliable agreement with reference judgement's, especially for more challenging questions. These results suggest that automated grading can assist teachers by supporting marking decisions and prioritizing responses for review, while still requiring human oversight.

Domain-Adaptive Pre-training for Automated Short Answer Grading in Conceptual Physics: Reliability, Question-Level Analysis, and Error Reduction

Large audio language models (LALMs) integrate audio representations with large language models to enable unified understanding of spoken content. Their capabilities have been increasingly investigated across several benchmarks; however, the examination of their use in rating L2 speech is still in its infancy. This study explores the abilities of LALMs in scoring three L2 speech global dimensions: foreign accentedness, comprehensibility, and intelligibility. Ninety audio samples produced by L2 speakers were rated by ten native speaker raters as well as five LALM models. Model performance was evaluated against the human composite mean using Pearson r, Spearman p, mean absolute error (MAE), and systematic bias, with the human leave-one-out correlation (r = .46-.73 across dimensions) serving as an empirical performance benchmark. The results showed that no LALM reached human-level performance on any dimension. Only one model (i.e., Gemini) achieved a significant correlation with human ratings on comprehensibility (r = .28, p < .01), while Qwen2-Audio showed modest correlation on intelligibility (r = .32, p < .01). MAE ranged from 0.75 to 3.99 for accentedness (human: 1.24), 1.35 to 3.00 for comprehensibility (human: 1.24), and 12.03 to 15.43 for intelligibility (human: 8.49). All models exhibited systematic biases, with deviations ranging from -9.31 to +13.19 points. The paper concludes with a discussion of the implications for automated L2 speech assessment.

Assessment of L2 speech global dimensions using large audio language models

Large Language Models (LLMs) are widely applied in educational practices, such as for generating children's stories. However, the generated stories are often too difficult for children to read, and the operational cost of LLMs hinders their widespread adoption in educational settings. We used an existing expert-designed children's reading curriculum and its corresponding generated stories from GPT-4o and Llama 3.3 70B to design different experiments for fine-tuning three 8B-parameter LLMs, which then generated new English reading stories that were subjected to quantitative and qualitative evaluation. Our method prioritizes controllability over scale, enabling educators to target reading levels and error patterns with a compact, affordable model. Our evaluation results show that with appropriate fine-tuning designs, children's English reading stories generated by 8B LLMs perform better on difficulty-related metrics than those from GPT-4o and Llama 3.3 70B, with almost no discernible safety issues. Such fine-tuned LLMs could be more broadly used by teachers, parents, and children in classrooms and at home to generate engaging English reading stories with children's interests, controllable difficulty and safety.

Children's English Reading Story Generation via Supervised Fine-Tuning of Compact LLMs with Controllable Difficulty and Safety

Automatic readability assessment for Arabic remains under-explored compared to English, and existing models are typically evaluated only within the training domain. We introduce the Jordanian School Textbook Corpus (JSTC), comprising approximately 87,000 text segments extracted from 240 official textbooks spanning grades 1–12, annotated with 127 handcrafted linguistic features. We compare XGBoost models trained on linguistic features, fine-tuned CAMeLBERT transformers, and hybrid architectures under both in-domain and cross-domain evaluation, using the BAREC benchmark for out-of-domain testing. While CAMeLBERT achieves strong in-domain performance (QWK = 0.830), its cross-domain QWK drops to 0.085—a 90% relative decrease. In contrast, XGBoost with linguistic features alone maintains the highest cross-domain QWK (0.240), outperforming all transformer and hybrid configurations. Notably, adding contextual [CLS] embeddings to linguistic features actively harms cross-domain transfer. These findings suggest that domain-robust readability models for educational deployment should prioritize linguistically-grounded features over contextual embeddings.

Transformer-based readability classifiers are worse than you think: Evidence from cross-domain Arabic readability assessment

Prediction of item difficulty from its text content is of substantial interest for automated generation of test items. In this paper, we focus on the related problem of recovering IRT-based difficulty when the data originally reported item p-value (percent correct responses). We model this item difficulty using a repository of reading passages and student data from US standardized tests from New York and Texas for grades 3-8 spanning the years 2018-23. This repository is annotated with meta-data on (1) linguistic features of the reading items, (2) test features of the passage, and (3) context features. Using a penalized regression model, we achieve an RMSE of 0.59 (compared to a 0.92 baseline) and a 0.77 correlation between true and predicted difficulty. We further evaluated the impact of LLM embeddings (ModernBERT, BERT, and LLaMA), finding that they marginally improve performance but function effectively as standalone alternatives to traditional linguistic features. Finally, we demonstrate how this difficulty prediction model powers a publicly available, human-in-the-loop tool for generating reading comprehension items.

Predicting Item Difficulty and Generating Reading Comprehension Items via an Annotated Repository

When the same LLM generates assessment items, simulates student responses, and scores them, the validation loop is self-referential. We introduce \textbf{Generative-Evaluative Agreement (GEA)}, a validity criterion measuring whether an LLM's scoring function recovers the skill levels its generative function was instructed to produce. In the first direct measurement of GEA on a two-stage adaptive assessment, the model recovers roughly half the intended variance ($r = 0.698$) with systematic positive bias. GEA is strong ($r > 0.7$) for syntactically verifiable skills but near zero for design-level skills, and low-skill overestimation inflates scores near the routing threshold. We show that granular, skill-decomposed rubrics are the primary mechanism for strengthening GEA and outline complementary mitigations.

Downloads

Next from ACL 2026

What Aggregate Scores Hide: Per-Rule Evaluation of Russian Grammatical Error Correction

Similar lecture

Using Interaction Log Data to Evaluate and Improve Feedback Accuracy in an Intelligent Language Tutoring System