Numerous studies have proposed computational methods to detect hate speech online, yet most focus on the English language and emphasize model development. In this study, we evaluate the counterfactual fairness of hate speech detection models in the Dutch language, specifically examining the performance and fairness of transformer-based models. We make the following key contributions. First, we curate a list of Dutch Social Group Terms that reflect social context. Second, we generate counterfactual data for Dutch hate speech using LLMs and established strategies like Manual Group Substitution (MGS) and Sentence Log-Likelihood (SLL). Through qualitative evaluation, we highlight the challenges of generating realistic counterfactuals, particularly with Dutch grammar and contextual coherence. Third, we fine-tune baseline transformer-based models with counterfactual data and evaluate their performance in detecting hate speech. Fourth, we assess the fairness of these models using Counterfactual Token Fairness (CTF) and group fairness metrics, including equality of odds and demographic parity. Our analysis shows that models perform better in terms of hate speech detection, average counterfactual fairness and group fairness. This work addresses a significant gap in the literature on counterfactual fairness for hate speech detection in Dutch and provides practical insights and recommendations for improving both model performance and fairness.

Large language models (LLMs) excel in tasks that require complex linguistic abilities, such as reference disambiguation and metaphor recognition/generation. Although LLMs possess impressive capabilities, their internal mechanisms for processing and representing linguistic knowledge remain largely opaque. Previous work on linguistic mechanisms has been limited by coarse granularity, insufficient causal analysis, and a narrow focus. In this study, we present a systematic and comprehensive causal investigation using sparse auto-encoders (SAEs). We extract a wide range of linguistic features from six dimensions: phonetics, phonology, morphology, syntax, semantics, and pragmatics. We extract, evaluate, and intervene on these features by constructing minimal contrast datasets and counterfactual sentence datasets. We introduce two indices-Feature Representation Confidence (FRC) and Feature Intervention Confidence (FIC)-to measure the ability of linguistic features to capture and control linguistic phenomena. Our results reveal inherent representations of linguistic knowledge in LLMs and demonstrate the potential for controlling model outputs. This work provides strong evidence that LLMs possess genuine linguistic knowledge and lays the foundation for more interpretable and controllable language modeling in future research.

Counterfactual explanations can be used to interpret and debug text classifiers by producing minimally altered text inputs that change a classifier's output. In this work, we evaluate five methods for generating counterfactual explanations for a BERT text classifier on two datasets using three evaluation metrics. The results of our experiments suggest that established white-box substitution-based methods are effective at generating valid counterfactuals that change the classifier's output. In contrast, newer methods based on large language models (LLMs) excel at producing natural and linguistically plausible text counterfactuals but often fail to generate valid counterfactuals that alter the classifier's output. Based on these results, we recommend developing new counterfactual explanation methods that combine the strengths of established gradient-based approaches and newer LLM-based techniques to generate high-quality, valid, and plausible text counterfactual explanations.

Mitigating explicit and implicit biases in Large Language Models (LLMs) has become a critical focus in the field of natural language processing. However, many current methodologies evaluate scenarios in isolation, without considering the broader context or the spectrum of potential biases within each situation. To address this, we introduce the Sensitivity Testing on Offensive Progressions (STOP) dataset, which includes 450 offensive progressions containing 2,700 unique sentences of varying severity that progressively escalate from less to more explicitly offensive. Covering a broad spectrum of 9 demographics and 46 sub-demographics, STOP ensures inclusivity and comprehensive coverage. We evaluate several leading closed- and open-source models, including GPT-4, Mixtral, and Llama 3. Our findings reveal that even the best-performing models detect bias inconsistently, with success rates ranging from 19.3% to 69.8%. We also demonstrate how aligning models with human judgments on STOP can improve model answer rates on sensitive tasks such as BBQ, StereoSet, and CrowS-Pairs by up to 191%, while maintaining or even improving performance. STOP presents a novel framework for assessing the complex nature of biases in LLMs, which will enable more effective bias mitigation strategies and facilitates the creation of fairer language models.

In this paper, we explore strategies to detect and evaluate counterfactual sentences. Since causal insight is an inherent characteristic of a counterfactual, is it possible to use this information in order to locate antecedent and consequent fragments in counterfactual statements? We thus propose to compare and evaluate models to correctly identify and chunk counterfactual sentences. In our experiments, we attempt to answer the following questions: First, can a learned model discern counterfactual statements reasonably well? Second, is it possible to clearly identify antecedent and consequent parts of counterfactual sentences?

Counterfactual reasoning requires predicting how alternative events, contrary to what actually happened, might have resulted in different outcomes. Despite being considered a necessary component of AI-complete systems, few resources have been developed for evaluating counterfactual reasoning in narratives. In this paper, we propose Counterfactual Story Rewriting: given an original story and an intervening counterfactual event, the task is to minimally revise the story to make it compatible with the given counterfactual event. Solving this task will require deep understanding of causal narrative chains and counterfactual invariance, and integration of such story reasoning capabilities into conditional language generation models. We present TimeTravel, a new dataset of 29,849 counterfactual rewritings, each with the original story, a counterfactual event, and human-generated revision of the original story compatible with the counterfactual event. Additionally, we include 80,115 counterfactual "branches" without a rewritten storyline to support future work on semi- or un-supervised approaches to counterfactual story rewriting. Finally, we evaluate the counterfactual rewriting capacities of several competitive baselines based on pretrained language models, and assess whether common overlap and model-based automatic metrics for text generation correlate well with human scores for counterfactual rewriting.