United States

Grasping the semantics of rare constructions (form--meaning pairings) has been shown to be a challenging problem that has currently only been solved by the largest LLMs. It remains an open question if open-source models have robust constructional understanding, and if so, what learning dynamics underlie the acquisition of this knowledge. Focusing on a set of rare Paired-Focus constructions in English (e.g. &quot;let alone&quot;, &quot;much less&quot;), we construct a novel dataset to test their meanings using both scalar adjectival semantics and general world knowledge. Testing a wide range of models differing in parameter count, architecture, and pretraining dataset size, we find that several modestly sized models are sensitive to both the forms and the meanings of Paired-Focus constructions. Turning to training dynamics for a set of open-checkpoint models, we find that Paired-Focus understanding emerges later in training than Paired-Focus syntactic knowledge, and that learning of Paired-Focus semantics is correlated with gains in some domains of world knowledge. Overall, our empirical results support the conclusion that modestly sized open-source models can grasp the rare Paired-Focus constructions, and demonstrate a connection between knowledge of Paired-Focus constructions and other meaning domains.

ACL 2026

Language Models Learn Constructional Semantics, Not To Mention Syntax: Investigating LM Understanding of Paired-Focus Constructions

## Welcome!
"I am excited to welcome you to this year’s edition of the Conference on Empirical Methods in Natural Language Processing! Importantly, it marks the 30th edition of EMNLP. With over 8,000 submissions, more than 3,000 accepted papers, and thousands of attendees, we have come a long way from that first
workshop, which had 14 accepted papers. As the field looks ahead, Suzhou is the fitting location for celebrating this milestone: rooted in a long literary tradition, yet modern and forward-looking, and home to a large share of the NLP community." 

*Message from the General Chair, Dirk Hovy*

EMNLP 2025

Celebrate 30 Years of EMNLP! 
EMNLP 2025 will be held in Suzhou, China from November 5th to November 9th, 2025.

Grasping the semantics of rare constructions (form--meaning pairings) has been shown to be a challenging problem that has currently only been solved by the largest LLMs. It remains an open question if open-source models have robust constructional understanding, and if so, what learning dynamics underlie the acquisition of this knowledge. Focusing on a set of rare Paired-Focus constructions in English (e.g. "let alone", "much less"), we construct a novel dataset to test their meanings using both scalar adjectival semantics and general world knowledge. Testing a wide range of models differing in parameter count, architecture, and pretraining dataset size, we find that several modestly sized models are sensitive to both the forms and the meanings of Paired-Focus constructions. Turning to training dynamics for a set of open-checkpoint models, we find that Paired-Focus understanding emerges later in training than Paired-Focus syntactic knowledge, and that learning of Paired-Focus semantics is correlated with gains in some domains of world knowledge. Overall, our empirical results support the conclusion that modestly sized open-source models can grasp the rare Paired-Focus constructions, and demonstrate a connection between knowledge of Paired-Focus constructions and other meaning domains.

workshop paper

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. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<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>

To access this event page, you need to log in with the **email address you registered with**. 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.

Visual Word Sense Disambiguation (Visual-WSD) requires ranking the correct image for an ambiguous word given a short trigger phrase. For low-resource languages, it is bottle\-necked by scarce sense-level benchmarks and limited sense-aligned multimodal supervision. We study Ukrainian and (i) extend the Ukrainian Visual-WSD benchmark from 87 to 381 instances and benchmark multilingual CLIP checkpoints and multimodal large models, and (ii) introduce two scalable Wikipedia-derived dataset construction methods. Using compute-efficient adaptation we fine-tune a multilingual CLIP backbone and show that sense-grounded supervision drives the improvements: combining our two Wikipedia-derived datasets improves HIT@1 from 37.00\% to 43.05\%.

From Sparse to Sense-Grounded: Wikipedia Training for Ukrainian Visual-WSD

Hope speech has played a vital role in online communities, yet most NLP work has focused on English and a few high-resource languages, leaving code-mixed varieties like Tulu largely unexplored. In the Shared Task on Hope Speech Detection in Code-Mixed Tulu at DravidianLangTech@ACL 2026, we have tackled two subtasks: (i) coarse-grained classification into Encouraging, Discouraging, Uninvolved and Blended categories (Task 1) and (ii) fine-grained classification into Optimistic, Realistic, Inspiring, Fading and Hopelessness (Task 2). We have fine-tuned three multilingual transformer encoders XLM-RoBERTa-base, MuRIL and mBERT on the official training splits. In Task 1, a three-way soft-voting ensemble of all three models has yielded the best performance with a macro F1 of 0.58, securing 1st place. In Task 2, XLM-RoBERTa-base alone has outperformed both MuRIL and mBERT, achieving a macro F1 of 0.42 and also securing 1st place.

CUET_SYNTHETICA@DravidianLangTech 2026: Multilingual Transformer Based Hope Speech Detection for Coarse and Fine-Grained Classification in Tulu

Abusive language targeting women has been a serious problem on Tamil social media and building systems to detect it automatically is harder than it looks. Tamil is morphologically complex, people have written it mixed with English in ways no dictionary has accounted for and a lot of the hostility has been indirect enough that has slipped past models trained on surface patterns. In the Shared Task on Abusive Tamil Text Targeting Women on Social Media DravidianLangTech@ACL 2026, we have worked on classifying Tamil YouTube comments as Abusive or Non-Abusive. We have trained three transformer models four times each with different learning rates, giving us 12 models total. Their predicted probabilities have been averaged to make the final decision. The 12-model ensemble has achieved a macro F1 of 0.8086, outperforming all individual models and securing 4th place in the shared task. Combining Tamil-specialized and multilingual transformer models has outperformed any single-architecture approach.

CUET_SYNTHETICA@DravidianLangTech 2026: Multi Architecture Transformer Ensemble for Detecting Abusive Tamil Text Targeting Women

This paper describes Team CHMOD_777's system for the DravidianLangTech@ACL 2026 shared task on political multiclass sentiment analysis of Tamil Twitter comments. The task requires classifying Tamil political tweets into seven sentiment categories under severe class imbalance (8:1 ratio). We address this challenge through LLM-based data augmentation using Gemini 2.5 Flash, expanding training data from 4,352 to 15,316 samples (3.5x the original). Our best system, MuRIL fine-tuned on augmented data with Focal Loss (gamma=3.0) and weighted sampling, achieves 35.79% Macro F1 on the development set, a 67% relative improvement over the non-augmented baseline. On the official test set, our system achieves 34.25% Macro F1, ranking 12th out of 22 participating teams. We find that (1) language-specific pre-training (MuRIL, 236M) outperforms larger general models (IndicBERT-v3, 1B), (2) smaller models benefit disproportionately from augmentation, and (3) Substantiated is the hardest category (F1=10.7%) due to its requirement for factual reasoning.

CHMOD_777@DravidianLangTech 2026: LLM Augmented Transformer Fine-tuning for Tamil Political Sentiment Analysis

This paper describes Team CHMOD_777's system for the DravidianLangTech@ACL 2026 shared task on detecting abusive Tamil text targeting women on social media. We fine-tune three transformer backbones (MuRIL, XLM-RoBERTa, IndicBERT-v3) with Focal Loss and weighted sampling, systematically evaluating the effects of context length, hyperparameter tuning, and language-specific pre-training. Our best system, MuRIL with 256-token context, achieves 82.76% Macro F1 on the development set and 80.61% on the official test set, ranking 6th out of 24 teams. We find that (1) extending context from 128 to 256 tokens improves F1 while converging 2.4x faster, (2) language-specific pre-training (MuRIL, 236M) outperforms larger models (IndicBERT, 270M), and (3) default hyperparameters are optimal, with every tuning attempt degrading performance.

CHMOD_777@DravidianLangTech 2026: Context-Aware Fine-tuned MuRIL for Abusive Tamil Text Detection on Social Media

This paper describes Team CHMOD_777’s system for the DravidianLangTech@ACL 2026 shared task on Tamil dialect speech recognition and classification. The task comprises two subtasks: classifying Tamil speech into four regional dialects (Northern, Southern, Western, Central) and transcribing dialectal Tamil speech to text. For dialect classification, we fine-tune MMS-1b-all with Focal Loss and weighted sampling, achieving 83.04% Macro F1 on the development set (5th out of 11 teams on the test set). For speech recognition, we fine-tune a Tamil-specific Whisper model (763M parameters), achieving 53.72% WER on the development set and 49.75% on the official test set, ranking 1st out of 13 teams. Our key finding is that domain-specific pre-training significantly outperforms larger general-purpose models: Tamil Whisper (763M) beats Whisperlarge-v3 (1.5B) by 8 WER points despite having half the parameter

CHMOD_777@DravidianLangTech 2026: Tamil-Adapted Whisper and MMS for Dialect Speech Recognition and Classification

Hate speech detection in low-resource, code-mixed languages is a challenging task as people often switch between scripts and languages in a single post. Code-Mixed scripts can take the form of explicit slurs, subtle insults, or fragmented abuse, and is often hidden by spelling variants and Romanized script. These datasets are also subjected to class imbalance with hate speech being a minority class of interest. To mitigate the imbalance, targeted data augmentation of minority class samples can help learn better representations to aid hate speech detection despite the naturally expected imbalance. We propose FLAICOL, a flip-point method which identifies the minimal embedding perturbation that moves an input across the decision boundary, map it back to discrete text, and retrain on those focused examples. Empirical results show that these interpretable augmentations strengthen Transformer classifiers on low-resource, code-mixed low resource hate datasets (Experiments were conducted on the Tamil-English, Malayalam-English, and Kannada-English splits in the Dravidian CodeMix Benchmark).

FLAICOL: Flip-Point-Led Augmentation for Imbalanced Code-Mixed Offensive Language Detection

Tokenization is fundamental to neural language modeling, yet for Tamil it remains largely adapted from general-purpose multilingual models without systematic consideration of its rich agglutinative morphology. We introduce TamilMorph, a large-scale dataset of over 500,000 morphologically segmented Tamil word forms. With this new morphological resource, we develop TamilTok, a morphology-aware tokenization framework that incorporates explicit morpheme structure into tokenizer training. We benchmark Tamil tokenization quality across multiple segmentation algorithms and vocabulary configurations and find that our approach improves both morphological alignment and downstream performance compared to previous approaches. Together, these contributions establish both a foundational linguistic resource and empirically grounded guidance for tokenizer design in morphologically rich languages, contributing to more inclusive language technologies.

TamilTok: Morphologically-Informed Tokenization for Tamil

Recovering writing style prompts in low resource languages has been daunting due to diverse morphology, culturally cognizant language patterns and deficient annotated resources. As previous works have predominantly focused on binary sentiment or single attribute transfer, extensive multi-class style classification in under-resourced languages like Telegu has been vastly underexplored. In this study, we have addressed this chasm through the Telugu Prompt-Style Recovery Shared Task at DravidianLangTech@ACL 2026 (Premjith et al., 2026), framing prompt reconstruction as a nine-class classification problem with Formal, Informal, Optimistic, Pessimistic, Humorous, Serious, Inspiring, Authoritative and Persuasive as prompt styles. We have evaluated three input configurations—Change Style, Original Transcripts and Merged input style—while training three transformer based models-MuRIL, XLM-RoBERTa and IndicBERT v2 under identical conditions. Our most promising model, IndicBERT v2 with partial layer freezing and weighted cross-entropy loss, has obtained a macro-F1 of 0.2987 and accuracy of 0.299. The Change Style configuration has significantly outperformed Original and Merged inputs, indicating that explicit style changes have made tonal and meaning cues more distinctive. These results have showcased the importance of language-specific pretraining and careful input design for style-sensitive NLP in low-resource settings, ultimately securing 1st rank on the shared task.

ERROR_500@DravidianLangTech2026: Automatic Prompt Style Classification in Telugu Using Transformer-Based Language Models

Mayangoli errors are context-sensitive errors in Tamil that arise from confusion among phonetically similar graphemes (e.g., ல/ள/ழ, ர/ற, ந/ன/ண). These errors are challenging for conventional spell checkers because both incorrect and correct forms are valid dictionary words, making dictionary lookup insufficient and requiring contextual modelling. We present TamilMayangoliSpell, a reproducible framework for Mayangoli error correction that combines (i) Tamil-specific preprocessing for sentence segmentation and normalisation, (ii) linguistically grounded error induction for generating training data constrained by dictionary validity, and (iii) fine-tuning of multilingual sequence-to-sequence models. Using 30,000 sentence pairs derived from TamilCorp, a massive multi-genre Tamil corpus and split 80/10/10 into train/validation/test, we fine-tune mBART, mT5, and NLLB under a small hyperparameter grid using greedy decoding with a maximum sequence length of 128. mT5 achieves the best performance (BLEU 99.28; Exact Match Accuracy 93.50%) and remains strong in a cross-genre evaluation on short stories. The preprocessing scripts, generated parallel datasets, and trained models are publicly available in a GitHub repository.

Downloads

Next from ACL 2026

From Sparse to Sense-Grounded: Wikipedia Training for Ukrainian Visual-WSD

Similar lecture

Unpacking Let Alone: Human-Scale Models Generalize to a Rare Construction in Form but not Meaning

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ACL 2026

From Sparse to Sense-Grounded: Wikipedia Training for Ukrainian Visual-WSD

Similar lecture

Unpacking Let Alone: Human-Scale Models Generalize to a Rare Construction in Form but not Meaning

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Downloads