En-Shiun Annie Lee

En-Shiun Annie Lee, Sarubi Thillainathan, Shravan Nayak et al.

What can pre-trained multilingual sequence-to-sequence models like mBART contribute to translating low-resource languages? We conduct a thorough empirical experiment in 10 languages to ascertain this, considering five factors: (1) the amount of fine-tuning data, (2) the noise in the fine-tuning data, (3) the amount of pre-training data in the model, (4) the impact of domain mismatch, and (5) language typology. In addition to yielding several heuristics, the experiments form a framework for evaluating the data sensitivities of machine translation systems. While mBART is robust to domain differences, its translations for unseen and typologically distant languages remain below 3.0 BLEU. In answer to our title's question, mBART is not a low-resource panacea; we therefore encourage shifting the emphasis from new models to new data.

Shravan Nayak, Surangika Ranathunga, Sarubi Thillainathan et al. · utoronto

NMT systems trained on Pre-trained Multilingual Sequence-Sequence (PMSS) models flounder when sufficient amounts of parallel data is not available for fine-tuning. This specifically holds for languages missing/under-represented in these models. The problem gets aggravated when the data comes from different domains. In this paper, we show that intermediate-task fine-tuning (ITFT) of PMSS models is extremely beneficial for domain-specific NMT, especially when target domain data is limited/unavailable and the considered languages are missing or under-represented in the PMSS model. We quantify the domain-specific results variations using a domain-divergence test, and show that ITFT can mitigate the impact of domain divergence to some extent.

Aditya Khan, Mason Shipton, David Anugraha et al.

URIEL is a knowledge base offering geographical, phylogenetic, and typological vector representations for 7970 languages. It includes distance measures between these vectors for 4005 languages, which are accessible via the lang2vec tool. Despite being frequently cited, URIEL is limited in terms of linguistic inclusion and overall usability. To tackle these challenges, we introduce URIEL+, an enhanced version of URIEL and lang2vec that addresses these limitations. In addition to expanding typological feature coverage for 2898 languages, URIEL+ improves the user experience with robust, customizable distance calculations to better suit the needs of users. These upgrades also offer competitive performance on downstream tasks and provide distances that better align with linguistic distance studies.

Hannah Liu, Muxin Tian, Iqra Ali et al.

Sentence simplification aims to make complex text more accessible by reducing linguistic complexity while preserving the original meaning. However, progress in this area remains limited for mid-resource and low-resource languages due to the scarcity of high-quality data. To address this gap, we introduce the OasisSimp dataset, a multilingual dataset for sentence-level simplification covering five languages: English, Sinhala, Tamil, Pashto, and Thai. Among these, no prior sentence simplification datasets exist for Thai, Pashto, and Tamil, while limited data is available for Sinhala. Each language simplification dataset was created by trained annotators who followed detailed guidelines to simplify sentences while maintaining meaning, fluency, and grammatical correctness. We evaluate eight open-weight multilingual Large Language Models (LLMs) on the OasisSimp dataset and observe substantial performance disparities between high-resource and low-resource languages, highlighting the simplification challenges in multilingual settings. The OasisSimp dataset thus provides both a valuable multilingual resource and a challenging benchmark, revealing the limitations of current LLM-based simplification methods and paving the way for future research in low-resource sentence simplification. The dataset is available at https://OasisSimpDataset.github.io/.

Hao Yu, Jesujoba O. Alabi, Andiswa Bukula et al.

Slot-filling and intent detection are well-established tasks in Conversational AI. However, current large-scale benchmarks for these tasks often exclude evaluations of low-resource languages and rely on translations from English benchmarks, thereby predominantly reflecting Western-centric concepts. In this paper, we introduce Injongo -- a multicultural, open-source benchmark dataset for 16 African languages with utterances generated by native speakers across diverse domains, including banking, travel, home, and dining. Through extensive experiments, we benchmark the fine-tuning multilingual transformer models and the prompting large language models (LLMs), and show the advantage of leveraging African-cultural utterances over Western-centric utterances for improving cross-lingual transfer from the English language. Experimental results reveal that current LLMs struggle with the slot-filling task, with GPT-4o achieving an average performance of 26 F1-score. In contrast, intent detection performance is notably better, with an average accuracy of 70.6%, though it still falls behind the fine-tuning baselines. Compared to the English language, GPT-4o and fine-tuning baselines perform similarly on intent detection, achieving an accuracy of approximately 81%. Our findings suggest that the performance of LLMs is still behind for many low-resource African languages, and more work is needed to further improve their downstream performance.

Genta Indra Winata, David Anugraha, Emmy Liu et al. · amazon-science

High-quality datasets are fundamental to training and evaluating machine learning models, yet their creation-especially with accurate human annotations-remains a significant challenge. Many dataset paper submissions lack originality, diversity, or rigorous quality control, and these shortcomings are often overlooked during peer review. Submissions also frequently omit essential details about dataset construction and properties. While existing tools such as datasheets aim to promote transparency, they are largely descriptive and do not provide standardized, measurable methods for evaluating data quality. Similarly, metadata requirements at conferences promote accountability but are inconsistently enforced. To address these limitations, this position paper advocates for the integration of systematic, rubric-based evaluation metrics into the dataset review process-particularly as submission volumes continue to grow. We also explore scalable, cost-effective methods for synthetic data generation, including dedicated tools and LLM-as-a-judge approaches, to support more efficient evaluation. As a call to action, we introduce DataRubrics, a structured framework for assessing the quality of both human- and model-generated datasets. Leveraging recent advances in LLM-based evaluation, DataRubrics offers a reproducible, scalable, and actionable solution for dataset quality assessment, enabling both authors and reviewers to uphold higher standards in data-centric research. We also release code to support reproducibility of LLM-based evaluations at https://github.com/datarubrics/datarubrics.

Quang Phuoc Nguyen, David Anugraha, Felix Gaschi et al.

Realignment is a promising strategy to improve cross-lingual transfer in multilingual language models. However, empirical results are mixed and often unreliable, particularly for typologically distant or low-resource languages (LRLs) compared to English. Moreover, word realignment tools often rely on high-quality parallel data, which can be scarce or noisy for many LRLs. In this work, we conduct an extensive empirical study to investigate whether realignment truly benefits from using all available languages, or if strategically selected subsets can offer comparable or even improved cross-lingual transfer, and study the impact on LRLs. Our controlled experiments show that realignment can be particularly effective for LRLs and that using carefully selected, linguistically diverse subsets can match full multilingual alignment, and even outperform it for unseen LRLs. This indicates that effective realignment does not require exhaustive language coverage and can reduce data collection overhead, while remaining both efficient and robust when guided by informed language selection.

Luke Zhang, Justin Vasselli, Aditya Khan et al.

We propose Dynamic Meta-Metrics (DMM), a framework for machine translation evaluation that learns source-sentence conditioned combinations of existing metrics. Rather than relying on a single static ensemble or language-specific weighting, DMM adapts the metric combination based on properties of the source segment. We study hard conditioning, which fits an interpretable combiner per cluster, and an exploratory soft-conditioned extension whose weights vary continuously with source-cluster responsibilities. We evaluate DMM on the WMT Metrics Shared Task data across multiple language pairs using pairwise agreement measures at the system and segment levels. Across settings, MLP-based combinations outperform linear and Gaussian process-based ensembles, and introducing soft conditioning yields gains over linear models.

Genta Indra Winata, Frederikus Hudi, Patrick Amadeus Irawan et al.

Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WorldCuisines, a massive-scale benchmark for multilingual and multicultural, visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.

Eric Khiu, Hasti Toossi, David Anugraha et al.

Fine-tuning and testing a multilingual large language model is expensive and challenging for low-resource languages (LRLs). While previous studies have predicted the performance of natural language processing (NLP) tasks using machine learning methods, they primarily focus on high-resource languages, overlooking LRLs and shifts across domains. Focusing on LRLs, we investigate three factors: the size of the fine-tuning corpus, the domain similarity between fine-tuning and testing corpora, and the language similarity between source and target languages. We employ classical regression models to assess how these factors impact the model's performance. Our results indicate that domain similarity has the most critical impact on predicting the performance of Machine Translation models.

Mason Shipton, York Hay Ng, Aditya Khan et al.

The URIEL+ linguistic knowledge base supports multilingual research by encoding languages through geographic, genetic, and typological vectors. However, data sparsity remains prevalent, in the form of missing feature types, incomplete language entries, and limited genealogical coverage. This limits the usefulness of URIEL+ in cross-lingual transfer, particularly for supporting low-resource languages. To address this sparsity, this paper extends URIEL+ with three contributions: introducing script vectors to represent writing system properties for 7,488 languages, integrating Glottolog to add 18,710 additional languages, and expanding lineage imputation for 26,449 languages by propagating typological and script features across genealogies. These additions reduce feature sparsity by 14% for script vectors, increase language coverage by up to 19,015 languages (1,007%), and improve imputation quality metrics by up to 33%. Our benchmark on cross-lingual transfer tasks (oriented around low-resource languages) shows occasionally divergent performance compared to URIEL+, with performance gains up to 6% in certain setups. Our advances make URIEL+ more complete and inclusive for multilingual research.

Hasti Toossi, Guo Qing Huai, Jinyu Liu et al.

In the pursuit of supporting more languages around the world, tools that characterize properties of languages play a key role in expanding the existing multilingual NLP research. In this study, we focus on a widely used typological knowledge base, URIEL, which aggregates linguistic information into numeric vectors. Specifically, we delve into the soundness and reproducibility of the approach taken by URIEL in quantifying language similarity. Our analysis reveals URIEL's ambiguity in calculating language distances and in handling missing values. Moreover, we find that URIEL does not provide any information about typological features for 31\% of the languages it represents, undermining the reliabilility of the database, particularly on low-resource languages. Our literature review suggests URIEL and lang2vec are used in papers on diverse NLP tasks, which motivates us to rigorously verify the database as the effectiveness of these works depends on the reliability of the information the tool provides.

Tong Su, Xin Peng, Sarubi Thillainathan et al.

Parameter-efficient fine-tuning (PEFT) methods are increasingly vital in adapting large-scale pre-trained language models for diverse tasks, offering a balance between adaptability and computational efficiency. They are important in Low-Resource Language (LRL) Neural Machine Translation (NMT) to enhance translation accuracy with minimal resources. However, their practical effectiveness varies significantly across different languages. We conducted comprehensive empirical experiments with varying LRL domains and sizes to evaluate the performance of 8 PEFT methods with in total of 15 architectures using the SacreBLEU score. We showed that 6 PEFT architectures outperform the baseline for both in-domain and out-domain tests and the Houlsby+Inversion adapter has the best performance overall, proving the effectiveness of PEFT methods.

Steve Bakos, Félix Gaschi, David Guzmán et al.

Realignment techniques are often employed to enhance cross-lingual transfer in multilingual language models, still, they can sometimes degrade performance in languages that differ significantly from the fine-tuned source language. This paper introduces AlignFreeze, a method that freezes either the layers' lower half or upper half during realignment. Through controlled experiments on 4 tasks, 3 models, and in 35 languages, we find that realignment affects all the layers but can be the most detrimental to the lower ones. Freezing the lower layers can prevent performance degradation. Particularly, AlignFreeze improves Part-of-Speech (PoS) tagging performances in languages where full realignment fails: with XLM-R, it provides improvements of more than one standard deviation in accuracy in seven more languages than full realignment.

Bo-Han Lu, Yi-Hsuan Lin, En-Shiun Annie Lee et al.

Machine translation focuses mainly on high-resource languages (HRLs), while low-resource languages (LRLs) like Taiwanese Hokkien are relatively under-explored. The study aims to address this gap by developing a dual translation model between Taiwanese Hokkien and both Traditional Mandarin Chinese and English. We employ a pre-trained LLaMA 2-7B model specialized in Traditional Mandarin Chinese to leverage the orthographic similarities between Taiwanese Hokkien Han and Traditional Mandarin Chinese. Our comprehensive experiments involve translation tasks across various writing systems of Taiwanese Hokkien as well as between Taiwanese Hokkien and other HRLs. We find that the use of a limited monolingual corpus still further improves the model's Taiwanese Hokkien capabilities. We then utilize our translation model to standardize all Taiwanese Hokkien writing systems into Hokkien Han, resulting in further performance improvements. Additionally, we introduce an evaluation method incorporating back-translation and GPT-4 to ensure reliable translation quality assessment even for LRLs. The study contributes to narrowing the resource gap for Taiwanese Hokkien and empirically investigates the advantages and limitations of pre-training and fine-tuning based on LLaMA 2.

York Hay Ng, Aditya Khan, Xiang Lu et al. · utoronto

Existing linguistic knowledge bases such as URIEL+ provide valuable geographic, genetic and typological distances for cross-lingual transfer but suffer from two key limitations. One, their one-size-fits-all vector representations are ill-suited to the diverse structures of linguistic data, and two, they lack a principled method for aggregating these signals into a single, comprehensive score. In this paper, we address these gaps by introducing a framework for type-matched language distances. We propose novel, structure-aware representations for each distance type: speaker-weighted distributions for geography, hyperbolic embeddings for genealogy, and a latent variables model for typology. We unify these signals into a robust, task-agnostic composite distance. In selecting transfer languages, our representations and composite distances consistently improve performance across a wide range of NLP tasks, providing a more principled and effective toolkit for multilingual research.

Hannah Liu, Junghyun Min, Ethan Yue Heng Cheung et al.

Despite major advances in machine translation (MT) in recent years, progress remains limited for many low-resource languages that lack large-scale training data and linguistic resources. Cantonese and Wu Chinese are two Sinitic examples, although each enjoys more than 80 million speakers around the world. In this paper, we introduce SiniticMTError, a novel dataset that builds on existing parallel corpora to provide error span, error type, and error severity annotations in machine-translated examples from English to Mandarin, Cantonese, and Wu Chinese. Our dataset serves as a resource for the MT community to utilize in fine-tuning models with error detection capabilities, supporting research on translation quality estimation, error-aware generation, and low-resource language evaluation. We report our rigorous annotation process by native speakers, with analyses on inter-annotator agreement, iterative feedback, and patterns in error type and severity.

Syed Mekael Wasti, Shou-Yi Hung, Christopher Collins et al.

Machine translation (MT) post-editing and research data collection often rely on inefficient, disconnected workflows. We introduce TranslationCorrect, an integrated framework designed to streamline these tasks. TranslationCorrect combines MT generation using models like NLLB, automated error prediction using models like XCOMET or LLM APIs (providing detailed reasoning), and an intuitive post-editing interface within a single environment. Built with human-computer interaction (HCI) principles in mind to minimize cognitive load, as confirmed by a user study. For translators, it enables them to correct errors and batch translate efficiently. For researchers, TranslationCorrect exports high-quality span-based annotations in the Error Span Annotation (ESA) format, using an error taxonomy inspired by Multidimensional Quality Metrics (MQM). These outputs are compatible with state-of-the-art error detection models and suitable for training MT or post-editing systems. Our user study confirms that TranslationCorrect significantly improves translation efficiency and user satisfaction over traditional annotation methods.

Sarubi Thillainathan, Songchen Yuan, En-Shiun Annie Lee et al.

Fine-tuning multilingual sequence-to-sequence large language models (msLLMs) has shown promise in developing neural machine translation (NMT) systems for low-resource languages (LRLs). However, conventional single-stage fine-tuning methods struggle in extremely low-resource NMT settings, where training data is very limited. This paper contributes to artificial intelligence by proposing two approaches for adapting msLLMs in these challenging scenarios: (1) continual pre-training (CPT), where the msLLM is further trained with domain-specific monolingual data to compensate for the under-representation of LRLs, and (2) intermediate task transfer learning (ITTL), a method that fine-tunes the msLLM with both in-domain and out-of-domain parallel data to enhance its translation capabilities across various domains and tasks. As an application in engineering, these methods are implemented in NMT systems for Sinhala, Tamil, and English (six language pairs) in domain-specific, extremely low-resource settings (datasets containing fewer than 100,000 samples). Our experiments reveal that these approaches enhance translation performance by an average of +1.47 bilingual evaluation understudy (BLEU) score compared to the standard single-stage fine-tuning baseline across all translation directions. Additionally, a multi-model ensemble further improves performance by an additional BLEU score.

David Anugraha, Patrick Amadeus Irawan, Anshul Singh et al.

Vision-language models (VLMs) have achieved strong performance in visual question answering (VQA), yet they remain constrained by static training data. Retrieval-Augmented Generation (RAG) mitigates this limitation by enabling access to up-to-date, culturally grounded, and multilingual information; however, multilingual multimodal RAG remains largely underexplored. We introduce M4-RAG, a massive-scale benchmark covering 42 languages and 56 regional dialects and registers, comprising over 80,000 culturally diverse image-question pairs for evaluating retrieval-augmented VQA across languages and modalities. To balance realism with reproducibility, we build a controlled retrieval environment containing millions of carefully curated multilingual documents relevant to the query domains, approximating real-world retrieval conditions while ensuring consistent experimentation. Our systematic evaluation reveals that although RAG consistently benefits smaller VLMs, it fails to scale to larger models and often even degrades their performance, exposing a critical mismatch between model size and current retrieval effectiveness. M4-RAG provides a foundation for advancing next-generation RAG systems capable of reasoning seamlessly across languages, modalities, and cultural contexts.

Junghyun Min, York Hay Ng, Sophia Chan et al. · utoronto

Cantonese, although spoken by millions, remains under-resourced due to policy and diglossia. To address this scarcity of evaluation frameworks for Cantonese, we introduce \textsc{\textbf{CantoNLU}}, a benchmark for Cantonese natural language understanding (NLU). This novel benchmark spans seven tasks covering syntax and semantics, including word sense disambiguation, linguistic acceptability judgment, language detection, natural language inference, sentiment analysis, part-of-speech tagging, and dependency parsing. In addition to the benchmark, we provide model baseline performance across a set of models: a Mandarin model without Cantonese training, two Cantonese-adapted models obtained by continual pre-training a Mandarin model on Cantonese text, and a monolingual Cantonese model trained from scratch. Results show that Cantonese-adapted models perform best overall, while monolingual models perform better on syntactic tasks. Mandarin models remain competitive in certain settings, indicating that direct transfer may be sufficient when Cantonese domain data is scarce. We release all datasets, code, and model weights to facilitate future research in Cantonese NLP.

York Hay Ng, Phuong Hanh Hoang, En-Shiun Annie Lee

Linguistic feature datasets such as URIEL+ are valuable for modelling cross-lingual relationships, but their high dimensionality and sparsity, especially for low-resource languages, limit the effectiveness of distance metrics. We propose a pipeline to optimize the URIEL+ typological feature space by combining feature selection and imputation, producing compact yet interpretable typological representations. We evaluate these feature subsets on linguistic distance alignment and downstream tasks, demonstrating that reduced-size representations of language typology can yield more informative distance metrics and improve performance in multilingual NLP applications.

Kosei Uemura, David Guzmán, Quang Phuoc Nguyen et al.

Large language models excel in English but still struggle with complex reasoning in many low-resource languages (LRLs). Existing encoder-plus-decoder methods such as LangBridge and MindMerger raise accuracy on mid and high-resource languages, yet they leave a large gap on LRLs. We present MERLIN, a two-stage model-stacking framework that applies a curriculum learning strategy -- from general bilingual bitext to task-specific data -- and adapts only a small set of DoRA weights. On the AfriMGSM benchmark MERLIN improves exact-match accuracy by +12.9 pp over MindMerger and outperforms GPT-4o-mini. It also yields consistent gains on MGSM and MSVAMP (+0.9 and +2.8 pp), demonstrating effectiveness across both low and high-resource settings.

David Anugraha, Genta Indra Winata, Chenyue Li et al.

Performance prediction is a method to estimate the performance of Language Models (LMs) on various Natural Language Processing (NLP) tasks, mitigating computational costs associated with model capacity and data for fine-tuning. Our paper presents ProxyLM, a scalable task- and language-agnostic framework designed to predict the performance of LMs using proxy models. These proxy models act as surrogates, approximating the performance of the LM of interest. By leveraging these proxy models, ProxyLM significantly reduces computational overhead in task evaluations, achieving up to a 37.08x speedup over traditional methods, even with our smallest proxy models. Our results across multiple multilingual NLP tasks and various robustness tests demonstrate that ProxyLM not only adapts well to previously unseen languages in pre-trained LMs, but also generalizes effectively across different datasets, outperforming the state-of-the-art by at least 1.78x in terms of root-mean-square error (RMSE).

David Ifeoluwa Adelani, Jessica Ojo, Israel Abebe Azime et al.

Despite the widespread adoption of Large language models (LLMs), their remarkable capabilities remain limited to a few high-resource languages. Additionally, many low-resource languages (\eg African languages) are often evaluated only on basic text classification tasks due to the lack of appropriate or comprehensive benchmarks outside of high-resource languages. In this paper, we introduce IrokoBench -- a human-translated benchmark dataset for 17 typologically-diverse low-resource African languages covering three tasks: natural language inference~(AfriXNLI), mathematical reasoning~(AfriMGSM), and multi-choice knowledge-based question answering~(AfriMMLU). We use IrokoBench to evaluate zero-shot, few-shot, and translate-test settings~(where test sets are translated into English) across 10 open and six proprietary LLMs. Our evaluation reveals a significant performance gap between high-resource languages~(such as English and French) and low-resource African languages. We observe a significant performance gap between open and proprietary models, with the highest performing open model, Gemma 2 27B only at 63\% of the best-performing proprietary model GPT-4o performance. In addition, machine translating the test set to English before evaluation helped to close the gap for larger models that are English-centric, such as Gemma 2 27B and LLaMa 3.1 70B. These findings suggest that more efforts are needed to develop and adapt LLMs for African languages.

Surangika Ranathunga, En-Shiun Annie Lee, Marjana Prifti Skenduli et al.

Neural Machine Translation (NMT) has seen a tremendous spurt of growth in less than ten years, and has already entered a mature phase. While considered as the most widely used solution for Machine Translation, its performance on low-resource language pairs still remains sub-optimal compared to the high-resource counterparts, due to the unavailability of large parallel corpora. Therefore, the implementation of NMT techniques for low-resource language pairs has been receiving the spotlight in the recent NMT research arena, thus leading to a substantial amount of research reported on this topic. This paper presents a detailed survey of research advancements in low-resource language NMT (LRL-NMT), along with a quantitative analysis aimed at identifying the most popular solutions. Based on our findings from reviewing previous work, this survey paper provides a set of guidelines to select the possible NMT technique for a given LRL data setting. It also presents a holistic view of the LRL-NMT research landscape and provides a list of recommendations to further enhance the research efforts on LRL-NMT.

Mehrdad Valipour, En-Shiun Annie Lee, Jaime R. Jamacaro et al.

Recent advancements in language representation models such as BERT have led to a rapid improvement in numerous natural language processing tasks. However, language models usually consist of a few hundred million trainable parameters with embedding space distributed across multiple layers, thus making them challenging to be fine-tuned for a specific task or to be transferred to a new domain. To determine whether there are task-specific neurons that can be exploited for unsupervised transfer learning, we introduce a method for selecting the most important neurons to solve a specific classification task. This algorithm is further extended to multi-source transfer learning by computing the importance of neurons for several single-source transfer learning scenarios between different subsets of data sources. Besides, a task-specific fingerprint for each data source is obtained based on the percentage of the selected neurons in each layer. We perform extensive experiments in unsupervised transfer learning for sentiment analysis, natural language inference and sentence similarity, and compare our results with the existing literature and baselines. Significantly, we found that the source and target data sources with higher degrees of similarity between their task-specific fingerprints demonstrate a better transferability property. We conclude that our method can lead to better performance using just a few hundred task-specific and interpretable neurons.