United States

In an effort to guarantee that machine learning model outputs conform with human moral values, recent work has begun exploring the possibility of explicitly training models to learn the difference between right and wrong. This is typically done in a bottom-up fashion, by exposing the model to different scenarios, annotated with human moral judgements. One question, however, is whether the trained models actually learn any consistent, higher-level ethical principles from these datasets – and if so, what? Here, we probe the Allen AI Delphi model with a set of standardized morality questionnaires, and find that, despite some inconsistencies, Delphi tends to mirror the moral principles associated with the demographic groups involved in the annotation process. We question whether this is desirable and discuss how we might move forward with this knowledge.

NAACL 2022

Does Moral Code Have a Moral Code? Probing Delphi’s Moral Philosophy

ethical ai

bias

language models

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

ACL 2026

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

## Welcome to NAACL 2022!
The Annual Meeting of the North American Chapter of the Association for Computational Linguistics (NAACL) is taking place July 10-15, 2022 as a hybrid event, in Seattle, WA and online. We are happy to welcome all of you to conference! 
 
The main conference program features oral presentations, in-person and virtual posters and demo sessions, a plenary session for our best paper presentations and awards, keynote presentations, a plenary panel on the place of linguistics and symbolic structures in NLP, and an Industrial Track panel on Careers in NLP.
 
Posters (including Findings of NAACL 2022) and demos are grouped by areas for both the in-person and the virtual sessions. For the virtual component, the talks will be on Zoom and the posters and the demos will be in GatherTown. The Student Research Workshop and will have an oral session and a poster session.
 
The program also features 6 Tutorials and 26 Workshops.
 
 We wish you a wonderful conference!
 
[The NAACL 2022 Organizing Committee](https://2022.naacl.org/committees/organization/)

2022 Annual Conference of the North American Chapter of the Association for Computational Linguistics

Large-scale surveys are a widely used instrument to collect data from a target audience. Beyond the single individual, an appropriate analysis of the answers can reveal trends and patterns and thus generate new insights and knowledge for researchers. Current analysis practices employ shallow machine learning methods or rely on (biased) human judgment. This work investigates the usage of state-of-the-art NLP models such as BERT to automatically extract information from both open- and closed-ended questions. We also leverage explainability methods at different levels of granularity to further derive knowledge from the analysis model. Experiments on EMS---a survey-based study researching influencing factors affecting a student's career goals---show that the proposed approach can identify such factors both at the input- and higher concept-level.

Explaining Neural NLP Models for the Joint Analysis of Open- and Closed-Ended Survey Answers

Neural rationale models are popular for interpretable predictions of NLP tasks. In these, a selector extracts segments of the input text, called rationales, and passes these segments to a classifier for prediction. Since the rationale is the only information accessible to the classifier, it is plausibly defined as the explanation. Is such a characterization unconditionally correct? In this paper, we argue to the contrary, with both philosophical perspectives and empirical evidence suggesting that rationale models are, perhaps, less rational and interpretable than expected. We call for more rigorous evaluations of these models to ensure desired properties of interpretability are indeed achieved. The code for our experiments is at https://github.com/yimingz89/Neural-Rationale-Analysis.

The Irrationality of Neural Rationale Models

In this paper, we conduct an empirical study on a bias measure, log-likelihood Masked Language Model (MLM) scoring, on a benchmark dataset. Previous work evaluates whether MLMs are biased or not for certain protected attributes (e.g., race) by comparing the log-likelihood scores of sentences that contain stereotypical characteristics with one category (e.g., black) versus another (e.g., white). We
hypothesized that this approach might be too sensitive to the choice of contextual words than the meaning of the sentence. Therefore, we computed the same measure after paraphrasing the sentences with different words but with same meaning. Our results demonstrate that the log-likelihood scoring can be more sensitive to utterance of specific words than to meaning behind a given sentence. Our paper reveals a shortcoming of the current log-likelihood-based bias measures for MLMs and calls for new ways to improve the robustness of it.

An Empirical Study on Pseudo-log-likelihood Bias Measures for Masked Language Models Using Paraphrased Sentences

Conversational recommender systems offer a way for users to engage in multi-turn conversations to find items they enjoy. For users to trust an agent and give effective feedback, the recommender system must be able to explain its suggestions and rationales. We develop a two-part framework for training multi-turn conversational recommenders that provide recommendation rationales that users can effectively interact with to receive better recommendations. First, we train a recommender system to jointly suggest items and explain its reasoning via subjective rationales.
We then fine-tune this model to incorporate iterative user feedback via self-supervised bot-play. Experiments on three real-world datasets demonstrate that our system can be applied to different recommendation models across diverse domains to achieve state-of-the-art performance in multi-turn recommendation. Human studies show that systems trained with our framework provide more useful, helpful, and knowledgeable suggestions in warm- and cold-start settings. Our framework allows us to use only product reviews during training, avoiding the need for expensive dialog transcript datasets that limit the applicability of previous conversational recommender agents.

Self-Supervised Bot-Play for Transcript-Free Conversational Recommendation with Rationales

In this work, we propose a knowledge enhanced generative model to mitigate
these two issues. Our generative model is a unified framework to sequentially generate relational triplets under various relation extraction settings and explicitly utilizes relevant knowledge from Knowledge Graph (KG) to resolve ambiguities.

Enhanced Knowledge for Joint Entity and Relation Extraction

Structured Knowledge has recently emerged as an essential component to support finegrained Question Answering (QA). In general, QA systems query a Knowledge Base (KB) to detect and extract the raw answers as final prediction. However, as lacking of context, language generation can offer a much informative and complete response. In this paper, we propose to combine the power of transfer learning and the advantage of entity placeholders to produce high-quality verbalization of extracted answers from a structured KB. We claim that such approach is especially well-suited for answer generation. Our experiments show 44.25%, 3.26% and 29.10% relative gain in BLEU over the state-of-the- art on the VQuAnDA, ParaQA and VANiLLa datasets, respectively. We additionally provide minor hallucinations corrections in VANiLLa standing for 5% of each of the training and testing set. We witness a median absolute gain of 0.81 SacreBLEU. This strengthens the importance of data quality when using automated evaluation.

Transfer Learning and Masked Generation for Answer Verbalization

Traditional approaches for creating natural language inference datasets from semi-structured datasets either involve manually constructing datasets, which is a computationally expensive and time-consuming operation which limits generation to small-scale datasets, or use fully automated methods, which while capable of producing large-scale datasets frequently produce basic statements that may lack reasoning. In this paper, we produce a semiautomated framework for data generation from entity based tabular data. We use the INFOTABS (Gupta et al., 2020) dataset to generate a large-scale human-like synthetic data that includes counterfactual entity-based tables. Through thorough experiments we demonstrate
the performance of the generated dataset

Efficient Realistic Data Generation Framework for Semi-Structured
Tabular Inference

Presentation on the paper "Multi-sense Language Modelling": recap, observations and findings

Multi-sense Language Modelling

Automated scoring technology for short-answer questions has been attracting attention to improve the fairness of scoring and reduce the burden on the scorer. In general, a large amount of data is required to train an automated scoring model. The training data consists of the answer texts and the scoring data assigned to them. It may also include annotations indicating key word sequences. Many previous studies have created models with large amounts of training data specific to each question. This paper aims to achieve equivalent performance with less training data by utilizing a BERT model that has been pre-trained on a large amount of general text data not necessarily related to short answer questions. On the RIKEN dataset, the proposed method reduces the training data from the 800 data required in the past to about 400 data, and still achieves scoring accuracy comparable to that of humans. Annotating 400 data is still costly, but it is beneficial to reduce the number of data needed.

Automatic scoring of short answers using justification cues estimated by BERT

The role of an author's L1 in SLA can be challenging for automated CEFR classification, in that texts from different L1 groups may be too heterogeneous to be combined as training data. We experiment with recent debiasing approaches by attempting to devoid textual representations of L1 features. This results in a more homogeneous group when aggregating CEFR-annotated texts from different L1 groups, leading to better classification performance. Using iterative null-space projection, we marginally improve classification performance for a linear classifier by 1 point. An MLP (e.g. non-linear) classifier remains unaffected by this procedure. We discuss possible directions for future work to attempt to increase this performance gain.

Downloads

Next from NAACL 2022

Explaining Neural NLP Models for the Joint Analysis of Open- and Closed-Ended Survey Answers

Similar lecture

Do VLMs Have a Moral Backbone? A Study on the Fragile Morality of Vision-Language Models