Cross-lingual transfer allows models to perform tasks in languages unseen during training and is often assumed to benefit from increased multilinguality. In this work, we challenge this assumption in the context of two underexplored, sense-aware tasks: polysemy disambiguation and lexical semantic change. Through a large-scale analysis across 28 languages, we show that multilingual training is neither necessary nor inherently beneficial for effective transfer. Instead, we find that confounding factors - such as fine-tuning data composition and evaluation artifacts - better account for the perceived advantages of multilinguality. Our findings call for more rigorous evaluations in multilingual NLP. We release fine-tuned models and benchmarks to support further research, with implications extending to low-resource and typologically diverse languages.
This paper addresses the critical need for high-quality evaluation datasets in low-resource languages to advance cross-lingual transfer. While cross-lingual transfer offers a key strategy for leveraging multilingual pretraining to expand language technologies to understudied and typologically diverse languages, its effectiveness is dependent on quality and suitable benchmarks. We release new sense-annotated datasets of sentences containing polysemous words, spanning nine low-resource languages across diverse language families and scripts. To facilitate dataset creation, the paper presents a demonstrably beneficial semi-automatic annotation method. The utility of the datasets is demonstrated through Word-in-Context (WiC) formatted experiments that evaluate transfer on these low-resource languages. Results highlight the importance of targeted dataset creation and evaluation for effective polysemy disambiguation in low-resource settings and transfer studies. The released datasets and code aim to support further research into fair, robust, and truly multilingual NLP.
We propose LAGO - Language Similarity-Aware Graph Optimization - a novel approach for few-shot cross-lingual embedding inversion attacks, addressing critical privacy vulnerabilities in multilingual NLP systems. Unlike prior work in embedding inversion attacks that treat languages independently, LAGO explicitly models linguistic relationships through a graph-based constrained distributed optimization framework. By integrating syntactic and lexical similarity as edge constraints, our method enables collaborative parameter learning across related languages. Theoretically, we show this formulation generalizes prior approaches, such as ALGEN, which emerges as a special case when similarity constraints are relaxed. Our framework uniquely combines Frobenius-norm regularization with linear inequality or total variation constraints, ensuring robust alignment of cross-lingual embedding spaces even with extremely limited data (as few as 10 samples per language). Extensive experiments across multiple languages and embedding models demonstrate that LAGO substantially improves the transferability of attacks with 10-20% increase in Rouge-L score over baselines. This work establishes language similarity as a critical factor in inversion attack transferability, urging renewed focus on language-aware privacy-preserving multilingual embeddings.
We present the first comprehensive study of Memorization in Multilingual Large Language Models (MLLMs), analyzing 95 languages using models across diverse model scales, architectures, and memorization definitions. As MLLMs are increasingly deployed, understanding their memorization behavior has become critical. Yet prior work has focused primarily on monolingual models, leaving multilingual memorization underexplored, despite the inherently long-tailed nature of training corpora. We find that the prevailing assumption, that memorization is highly correlated with training data availability, fails to fully explain memorization patterns in MLLMs. We hypothesize that treating languages in isolation - ignoring their similarities - obscures the true patterns of memorization. To address this, we propose a novel graph-based correlation metric that incorporates language similarity to analyze cross-lingual memorization. Our analysis reveals that among similar languages, those with fewer training tokens tend to exhibit higher memorization, a trend that only emerges when cross-lingual relationships are explicitly modeled. These findings underscore the importance of a language-aware perspective in evaluating and mitigating memorization vulnerabilities in MLLMs. This also constitutes empirical evidence that language similarity both explains Memorization in MLLMs and underpins Cross-lingual Transferability, with broad implications for multilingual NLP.
Hausa Natural Language Processing (NLP) has gained increasing attention in recent years, yet remains understudied as a low-resource language despite having over 120 million first-language (L1) and 80 million second-language (L2) speakers worldwide. While significant advances have been made in high-resource languages, Hausa NLP faces persistent challenges, including limited open-source datasets and inadequate model representation. This paper presents an overview of the current state of Hausa NLP, systematically examining existing resources, research contributions, and gaps across fundamental NLP tasks: text classification, machine translation, named entity recognition, speech recognition, and question answering. We introduce HausaNLP (https://catalog.hausanlp.org), a curated catalog that aggregates datasets, tools, and research works to enhance accessibility and drive further development. Furthermore, we discuss challenges in integrating Hausa into large language models (LLMs), addressing issues of suboptimal tokenization and dialectal variation. Finally, we propose strategic research directions emphasizing dataset expansion, improved language modeling approaches, and strengthened community collaboration to advance Hausa NLP. Our work provides both a foundation for accelerating Hausa NLP progress and valuable insights for broader multilingual NLP research.
Cross-lingual transfer has become a crucial aspect of multilingual NLP, as it allows for models trained on resource-rich languages to be applied to low-resource languages more effectively. Recently massively multilingual pre-trained language models (e.g., mBERT, XLM-R) demonstrate strong zero-shot transfer capabilities[14] [13]. This paper investigates cross-linguistic transfer through the lens of language families and morphology. Investigating how language family proximity and morphological similarity affect performance across NLP tasks. We further discuss our results and how it relates to findings from recent literature. Overall, we compare multilingual model performance and review how linguistic distance metrics correlate with transfer outcomes. We also look into emerging approaches that integrate typological and morphological information into model pre-training to improve transfer to diverse languages[18] [19].
Large language models (LLMs) have demonstrated impressive performance across a wide range of Natural Language Processing (NLP) tasks. However, ensuring their effectiveness across multiple languages presents unique challenges. Multilingual prompt engineering has emerged as a key approach to enhance LLMs' capabilities in diverse linguistic settings without requiring extensive parameter re-training or fine-tuning. With growing interest in multilingual prompt engineering over the past two to three years, researchers have explored various strategies to improve LLMs' performance across languages and NLP tasks. By crafting structured natural language prompts, researchers have successfully extracted knowledge from LLMs across different languages, making these techniques an accessible pathway for a broader audience, including those without deep expertise in machine learning, to harness the capabilities of LLMs. In this paper, we survey and categorize different multilingual prompting techniques based on the NLP tasks they address across a diverse set of datasets that collectively span around 250 languages. We further highlight the LLMs employed, present a taxonomy of approaches and discuss potential state-of-the-art (SoTA) methods for specific multilingual datasets. Additionally, we derive a range of insights across language families and resource levels (high-resource vs. low-resource), including analyses such as the distribution of NLP tasks by language resource type and the frequency of prompting methods across different language families. Our survey reviews 36 research papers covering 39 prompting techniques applied to 30 multilingual NLP tasks, with the majority of these studies published in the last two years.
Social bias in language models can potentially exacerbate social inequalities. Despite it having garnered wide attention, most research focuses on English data. In a low-resource scenario, the models often perform worse due to insufficient training data. This study aims to leverage high-resource language corpora to evaluate bias and experiment with debiasing methods in low-resource languages. We evaluated the performance of recent multilingual models in five languages: English (\textsc{eng}), Chinese (\textsc{zho}), Russian (\textsc{rus}), Indonesian (\textsc{ind}) and Thai (\textsc{tha}), and analyzed four bias dimensions: \textit{gender}, \textit{religion}, \textit{nationality}, and \textit{race-color}. By constructing multilingual bias evaluation datasets, this study allows fair comparisons between models across languages. We have further investigated three debiasing methods-\texttt{CDA}, \texttt{Dropout}, \texttt{SenDeb}-and demonstrated that debiasing methods from high-resource languages can be effectively transferred to low-resource ones, providing actionable insights for fairness research in multilingual NLP.
Hallucinations in generative AI, particularly in Large Language Models (LLMs), pose a significant challenge to the reliability of multilingual applications. Existing benchmarks for hallucination detection focus primarily on English and a few widely spoken languages, lacking the breadth to assess inconsistencies in model performance across diverse linguistic contexts. To address this gap, we introduce Poly-FEVER, a large-scale multilingual fact verification benchmark specifically designed for evaluating hallucination detection in LLMs. Poly-FEVER comprises 77,973 labeled factual claims spanning 11 languages, sourced from FEVER, Climate-FEVER, and SciFact. It provides the first large-scale dataset tailored for analyzing hallucination patterns across languages, enabling systematic evaluation of LLMs such as ChatGPT and the LLaMA series. Our analysis reveals how topic distribution and web resource availability influence hallucination frequency, uncovering language-specific biases that impact model accuracy. By offering a multilingual benchmark for fact verification, Poly-FEVER facilitates cross-linguistic comparisons of hallucination detection and contributes to the development of more reliable, language-inclusive AI systems. The dataset is publicly available to advance research in responsible AI, fact-checking methodologies, and multilingual NLP, promoting greater transparency and robustness in LLM performance. The proposed Poly-FEVER is available at: https://huggingface.co/datasets/HanzhiZhang/Poly-FEVER.
Hate speech detection in low-resource languages like Telugu is a growing challenge in NLP. This study investigates transformer-based models, including TeluguHateBERT, HateBERT, DeBERTa, Muril, IndicBERT, Roberta, and Hindi-Abusive-MuRIL, for classifying hate speech in Telugu. We fine-tune these models using Low-Rank Adaptation (LoRA) to optimize efficiency and performance. Additionally, we explore a multilingual approach by translating Telugu text into English using Google Translate to assess its impact on classification accuracy. Our experiments reveal that most models show improved performance after translation, with DeBERTa and Hindi-Abusive-MuRIL achieving higher accuracy and F1 scores compared to training directly on Telugu text. Notably, Hindi-Abusive-MuRIL outperforms all other models in both the original Telugu dataset and the translated dataset, demonstrating its robustness across different linguistic settings. This suggests that translation enables models to leverage richer linguistic features available in English, leading to improved classification performance. The results indicate that multilingual processing can be an effective approach for hate speech detection in low-resource languages. These findings demonstrate that transformer models, when fine-tuned appropriately, can significantly improve hate speech detection in Telugu, paving the way for more robust multilingual NLP applications.
Multilingual language models have significantly advanced due to rapid progress in natural language processing. Models like BLOOM 1.7B, trained on diverse multilingual datasets, aim to bridge linguistic gaps. However, their effectiveness in capturing linguistic knowledge, particularly for low-resource languages, remains an open question. This study critically examines MLMs capabilities in multilingual understanding, semantic representation, and cross-lingual knowledge transfer. While these models perform well for high-resource languages, they struggle with less-represented ones. Additionally, traditional evaluation methods often overlook their internal syntactic and semantic encoding. This research addresses key limitations through three objectives. First, it assesses semantic similarity by analyzing multilingual word embeddings for consistency using cosine similarity. Second, it examines BLOOM-1.7B and Qwen2 through Named Entity Recognition and sentence similarity tasks to understand their linguistic structures. Third, it explores cross-lingual knowledge transfer by evaluating generalization from high-resource to low-resource languages in sentiment analysis and text classification. By leveraging linguistic probing, performance metrics, and visualizations, this study provides insights into the strengths and limitations of MLMs. The findings aim to enhance multilingual NLP models, ensuring better support for both high- and low-resource languages, thereby promoting inclusivity in language technologies.
Wikipedia's perceived high quality and broad language coverage have established it as a fundamental resource in multilingual NLP. In the context of low-resource languages, however, these quality assumptions are increasingly being scrutinised. This paper critically examines the data quality of Wikipedia in a non-English setting by subjecting it to various quality filtering techniques, revealing widespread issues such as a high percentage of one-line articles and duplicate articles. We evaluate the downstream impact of quality filtering on Wikipedia and find that data quality pruning is an effective means for resource-efficient training without hurting performance, especially for low-resource languages. Moreover, we advocate for a shift in perspective from seeking a general definition of data quality towards a more language- and task-specific one. Ultimately, we aim for this study to serve as a guide to using Wikipedia for pretraining in a multilingual setting.
The common practice of preprocessing text before feeding it into NLP models introduces many decision points which have unintended consequences on model performance. In this opinion piece, we focus on the handling of diacritics in texts originating in many languages and scripts. We demonstrate, through several case studies, the adverse effects of inconsistent encoding of diacritized characters and of removing diacritics altogether. We call on the community to adopt simple but necessary steps across all models and toolkits in order to improve handling of diacritized text and, by extension, increase equity in multilingual NLP.
This paper explores the problems of Question Answering (QA) and Named Entity Recognition (NER) in five diverse languages. We tested five Large Language Models with various prompting methods, including zero-shot, chain-of-thought reasoning, and translation techniques. Our results show that while some models consistently outperform others, their effectiveness varies significantly across tasks and languages. We saw that advanced prompting techniques generally improved QA performance but had mixed results for NER; and we observed that language difficulty patterns differed between tasks. Our findings highlight the need for task-specific approaches in multilingual NLP and suggest that current models may develop different linguistic competencies for different tasks.
Training large language models (LLMs) in low-resource languages such as Hebrew poses unique challenges. In this paper, we introduce DictaLM2.0 and DictaLM2.0-Instruct, two LLMs derived from the Mistral model, trained on a substantial corpus of approximately 200 billion tokens in both Hebrew and English. Adapting a pre-trained model to a new language involves specialized techniques that differ significantly from training a model from scratch or further training existing models on well-resourced languages such as English. We outline these novel training methodologies, which facilitate effective learning and adaptation to the linguistic properties of Hebrew. Additionally, we fine-tuned DictaLM2.0-Instruct on a comprehensive instruct dataset to enhance its performance on task-specific instructions. To rigorously evaluate our models, we introduce a new benchmark suite for Hebrew LLM evaluation, covering a diverse set of tasks including Question Answering, Sentiment Analysis, Winograd Schema Challenge, Translation, and Summarization. Our work not only addresses the intricacies of training LLMs in low-resource languages but also proposes a framework that can be leveraged for adapting other LLMs to various non-English languages, contributing to the broader field of multilingual NLP.
Many attempts have been made in multilingual NLP to ensure that pre-trained language models, such as mBERT or GPT2 get better and become applicable to low-resource languages. To achieve multilingualism for pre-trained language models (PLMs), we need techniques to create word embeddings that capture the linguistic characteristics of any language. Tokenization is one such technique because it allows for the words to be split based on characters or subwords, creating word embeddings that best represent the structure of the language. Creating such word embeddings is essential to applying PLMs to other languages where the model was not trained, enabling multilingual NLP. However, most PLMs use generic tokenization methods like BPE, wordpiece, or unigram which may not suit specific languages. We hypothesize that tokenization based on syllables within the input text, which we call syllable tokenization, should facilitate the development of syllable-aware language models. The syllable-aware language models make it possible to apply PLMs to languages that are rich in syllables, for instance, Swahili. Previous works introduced subword tokenization. Our work extends such efforts. Notably, we propose a syllable tokenizer and adopt an experiment-centric approach to validate the proposed tokenizer based on the Swahili language. We conducted text-generation experiments with GPT2 to evaluate the effectiveness of the syllable tokenizer. Our results show that the proposed syllable tokenizer generates syllable embeddings that effectively represent the Swahili language.
The Universal Dependencies (UD) project aims to create a cross-linguistically consistent dependency annotation for multiple languages, to facilitate multilingual NLP. It currently supports 114 languages. Dravidian languages are spoken by over 200 million people across the word, and yet there are only two languages from this family in UD. This paper examines some of the morphological and syntactic features of Dravidian languages and explores how they can be annotated in the UD framework.
Large language models (LLMs) often inherit and amplify social biases embedded in their training data. A prominent social bias is gender bias. In this regard, prior work has mainly focused on gender stereotyping bias - the association of specific roles or traits with a particular gender - in English and on evaluating gender bias in model embeddings or generated outputs. In contrast, gender representation bias - the unequal frequency of references to individuals of different genders - in the training corpora has received less attention. Yet such imbalances in the training data constitute an upstream source of bias that can propagate and intensify throughout the entire model lifecycle. To fill this gap, we propose a novel LLM-based method to detect and quantify gender representation bias in LLM training data in gendered languages, where grammatical gender challenges the applicability of methods developed for English. By leveraging the LLMs' contextual understanding, our approach automatically identifies and classifies person-referencing words in gendered language corpora. Applied to four Spanish-English benchmarks and five Valencian corpora, our method reveals substantial male-dominant imbalances. We show that such biases in training data affect model outputs, but can surprisingly be mitigated leveraging small-scale training on datasets that are biased towards the opposite gender. Our findings highlight the need for corpus-level gender bias analysis in multilingual NLP. We make our code and data publicly available.
While multilingual language models (MLMs) have been trained on 100+ languages, they are typically only evaluated across a handful of them due to a lack of available test data in most languages. This is particularly problematic when assessing MLM's potential for low-resource and unseen languages. In this paper, we present an analysis of existing evaluation frameworks in multilingual NLP, discuss their limitations, and propose several directions for more robust and reliable evaluation practices. Furthermore, we empirically study to what extent machine translation offers a {reliable alternative to human translation} for large-scale evaluation of MLMs across a wide set of languages. We use a SOTA translation model to translate test data from 4 tasks to 198 languages and use them to evaluate three MLMs. We show that while the selected subsets of high-resource test languages are generally sufficiently representative of a wider range of high-resource languages, we tend to overestimate MLMs' ability on low-resource languages. Finally, we show that simpler baselines can achieve relatively strong performance without having benefited from large-scale multilingual pretraining.
Rapidly growing numbers of multilingual news consumers pose an increasing challenge to news recommender systems in terms of providing customized recommendations. First, existing neural news recommenders, even when powered by multilingual language models (LMs), suffer substantial performance losses in zero-shot cross-lingual transfer (ZS-XLT). Second, the current paradigm of fine-tuning the backbone LM of a neural recommender on task-specific data is computationally expensive and infeasible in few-shot recommendation and cold-start setups, where data is scarce or completely unavailable. In this work, we propose a news-adapted sentence encoder (NaSE), domain-specialized from a pretrained massively multilingual sentence encoder (SE). To this end, we construct and leverage PolyNews and PolyNewsParallel, two multilingual news-specific corpora. With the news-adapted multilingual SE in place, we test the effectiveness of (i.e., question the need for) supervised fine-tuning for news recommendation, and propose a simple and strong baseline based on (i) frozen NaSE embeddings and (ii) late click-behavior fusion. We show that NaSE achieves state-of-the-art performance in ZS-XLT in true cold-start and few-shot news recommendation.