In this work, we provide a systematic and comprehensive empirical comparison of pretrained multilingual language models versus their monolingual counterparts with regard to their monolingual task performance. We study a set of nine typologically diverse languages with readily available pretrained monolingual models on a set of five diverse monolingual downstream tasks. We first aim to establish, via fair and controlled comparisons, if a gap between the multilingual and the corresponding monolingual representation of that language exists, and subsequently investigate the reason for any performance difference. To disentangle conflating factors, we train new monolingual models on the same data, with monolingually and multilingually trained tokenizers. We find that while the pretraining data size is an important factor, a designated monolingual tokenizer plays an equally important role in the downstream performance. Our results show that languages that are adequately represented in the multilingual model's vocabulary exhibit negligible performance decreases over their monolingual counterparts. We further find that replacing the original multilingual tokenizer with the specialized monolingual tokenizer improves the downstream performance of the multilingual model for almost every task and language.
Microsoft Research（微软研究院）；Beijing Institute of Technology（北京理工大学）
Multilingual pretrained language models (such as multilingual BERT) have achieved impressive results for cross-lingual transfer. However, due to the constant model capacity, multilingual pre-training usually lags behind the monolingual competitors. In this work, we present two approaches to improve zero-shot cross-lingual classification, by transferring the knowledge from monolingual pretrained models to multilingual ones. Experimental results on two cross-lingual classification benchmarks show that our methods outperform vanilla multilingual fine-tuning.
NTIS – New Technologies for the Information Society（NTIS – 新信息技术学会）；University of West Bohemia（西波希米亚大学）
In this paper, we aim at improving Czech sentiment with transformer-based models and their multilingual versions. More concretely, we study the task of polarity detection for the Czech language on three sentiment polarity datasets. We fine-tune and perform experiments with five multilingual and three monolingual models. We compare the monolingual and multilingual models' performance, including comparison with the older approach based on recurrent neural networks. Furthermore, we test the multilingual models and their ability to transfer knowledge from English to Czech (and vice versa) with zer…
University of Edinburgh（爱丁堡大学）；Facebook AI（脸书人工智能研究院）
There has been recent success in pre-training on monolingual data and fine-tuning on Machine Translation (MT), but it remains unclear how to best leverage a pre-trained model for a given MT task. This paper investigates the benefits and drawbacks of freezing parameters, and adding new ones, when fine-tuning a pre-trained model on MT. We focus on 1) Fine-tuning a model trained only on English monolingual data, BART. 2) Fine-tuning a model trained on monolingual data from 25 languages, mBART. For BART we get the best performance by freezing most of the model parameters, and adding extra position…
Over the last few years two promising research directions in low-resource neural machine translation (NMT) have emerged. The first focuses on utilizing high-resource languages to improve the quality of low-resource languages via multilingual NMT. The second direction employs monolingual data with self-supervision to pre-train translation models, followed by fine-tuning on small amounts of supervised data. In this work, we join these two lines of research and demonstrate the efficacy of monolingual data with self-supervision in multilingual NMT. We offer three major results: (i) Using monolingu…
We use the multilingual OSCAR corpus, extracted from Common Crawl via language classification, filtering and cleaning, to train monolingual contextualized word embeddings (ELMo) for five mid-resource languages. We then compare the performance of OSCAR-based and Wikipedia-based ELMo embeddings for these languages on the part-of-speech tagging and parsing tasks. We show that, despite the noise in the Common-Crawl-based OSCAR data, embeddings trained on OSCAR perform much better than monolingual embeddings trained on Wikipedia. They actually equal or improve the current state of the art in taggin…
National Institute of Information and Communications Technology（信息通信技术研究所）
Multilingual pretrained models are effective for machine translation and cross-lingual processing because they contain multiple languages in one model. However, they are pretrained after their tokenizers are fixed; therefore it is difficult to change the vocabulary after pretraining. When we extend the pretrained models to new languages, we must modify the tokenizers simultaneously. In this paper, we add new subwords to the SentencePiece tokenizer to apply a multilingual pretrained model to new languages (Inuktitut in this paper). In our experiments, we segmented Inuktitut sentences into subwo…
Recent studies have demonstrated that pre-trained cross-lingual models achieve impressive performance in downstream cross-lingual tasks. This improvement benefits from learning a large amount of monolingual and parallel corpora. Although it is generally acknowledged that parallel corpora are critical for improving the model performance, existing methods are often constrained by the size of parallel corpora, especially for low-resource languages. In this paper, we propose ERNIE-M, a new training method that encourages the model to align the representation of multiple languages with monolingual …
A multilingual tokenizer is a fundamental component of multilingual neural machine translation. It is trained from a multilingual corpus. Since a skewed data distribution is considered to be harmful, a sampling strategy is usually used to balance languages in the corpus. However, few works have systematically answered how language imbalance in tokenizer training affects downstream performance. In this work, we analyze how translation performance changes as the data ratios among languages vary in the tokenizer training corpus. We find that while relatively better performance is often observed w…
Recent work on tokenizer-free multilingual pretrained models show promising results in improving cross-lingual transfer and reducing engineering overhead (Clark et al., 2022; Xue et al., 2022). However, these works mainly focus on reporting accuracy on a limited set of tasks and data settings, placing less emphasis on other important factors when tuning and deploying the models in practice, such as memory usage, inference speed, and fine-tuning data robustness. We attempt to fill this gap by performing a comprehensive empirical comparison of multilingual tokenizer-free and subword-based models…
We introduce the task of zero-shot style transfer between different languages. Our training data includes multilingual parallel corpora, but does not contain any parallel sentences between styles, similarly to the recent previous work. We propose a unified multilingual multi-style machine translation system design, that allows to perform zero-shot style conversions during inference; moreover, it does so both monolingually and cross-lingually. Our model allows to increase the presence of dissimilar styles in corpus by up to 3 times, easily learns to operate with various contractions, and provid…
iFLYTEK Research（科大讯飞研究院）；Harbin Institute of Technology（哈尔滨工业大学）
Multilingual pre-trained language models (MPLMs) not only can handle tasks in different languages but also exhibit surprising zero-shot cross-lingual transferability. However, MPLMs usually are not able to achieve comparable supervised performance on rich-resource languages compared to the state-of-the-art monolingual pre-trained models. In this paper, we aim to improve the multilingual model's supervised and zero-shot performance simultaneously only with the resources from supervised languages. Our approach is based on transferring knowledge from high-performance monolingual models with a tea…
Universidade de Vigo（维戈大学）；Cardiff University（卡迪夫大学）
Word embeddings have become a standard resource in the toolset of any Natural Language Processing practitioner. While monolingual word embeddings encode information about words in the context of a particular language, cross-lingual embeddings define a multilingual space where word embeddings from two or more languages are integrated together. Current state-of-the-art approaches learn these embeddings by aligning two disjoint monolingual vector spaces through an orthogonal transformation which preserves the structure of the monolingual counterparts. In this work, we propose to apply an addition…
Large pretrained masked language models have become state-of-the-art solutions for many NLP problems. The research has been mostly focused on English language, though. While massively multilingual models exist, studies have shown that monolingual models produce much better results. We train two trilingual BERT-like models, one for Finnish, Estonian, and English, the other for Croatian, Slovenian, and English. We evaluate their performance on several downstream tasks, NER, POS-tagging, and dependency parsing, using the multilingual BERT and XLM-R as baselines. The newly created FinEst BERT and …
University of Ljubljana（卢布尔雅那大学）；Queen Mary University of London（伦敦玛丽女王大学）；Jožef Stefan Institute（约泽夫·斯蒂芬研究所）
The current dominance of deep neural networks in natural language processing is based on contextual embeddings such as ELMo, BERT, and BERT derivatives. Most existing work focuses on English; in contrast, we present here the first multilingual empirical comparison of two ELMo and several monolingual and multilingual BERT models using 14 tasks in nine languages. In monolingual settings, our analysis shows that monolingual BERT models generally dominate, with a few exceptions such as the dependency parsing task, where they are not competitive with ELMo models trained on large corpora. In cross-l…
University of Groningen（格罗宁根大学）；NAVER LABS Europe（NAVER LABS 欧洲）
We consider the problem of multilingual unsupervised machine translation, translating to and from languages that only have monolingual data by using auxiliary parallel language pairs. For this problem the standard procedure so far to leverage the monolingual data is back-translation, which is computationally costly and hard to tune. In this paper we propose instead to use denoising adapters, adapter layers with a denoising objective, on top of pre-trained mBART-50. In addition to the modularity and flexibility of such an approach we show that the resulting translations are on-par with back-tra…
Singapore University of Technology and Design（新加坡科技设计大学）
With the development of several multilingual datasets used for semantic parsing, recent research efforts have looked into the problem of learning semantic parsers in a multilingual setup. However, how to improve the performance of a monolingual semantic parser for a specific language by leveraging data annotated in different languages remains a research question that is under-explored. In this work, we present a study to show how learning distributed representations of the logical forms from data annotated in different languages can be used for improving the performance of a monolingual semant…
In recent studies, it has been shown that Multilingual language models underperform their monolingual counterparts. It is also a well-known fact that training and maintaining monolingual models for each language is a costly and time-consuming process. Roman Urdu is a resource-starved language used popularly on social media platforms and chat apps. In this research, we propose a novel dataset of scraped tweets containing 54M tokens and 3M sentences. Additionally, we also propose RUBERT a bilingual Roman Urdu model created by additional pretraining of English BERT. We compare its performance wit…
Modular Sentence Encoders: Separating Language Specialization from Cross-Lingual Alignment
Ubiquitous Knowledge Processing Lab (UKP Lab) Department of Computer Science Technical University of Darmstadt and National Research Center for Applied Cybersecurity ATHENE, Germany（技术大学达姆施塔特计算机科学系通用知识处理实验室（UKP实验室）和应用网络安全国家研究中心ATHENE德国）；Center for AI and Data Science, University of Würzburg（人工智能与数据科学中心，乌尔姆大学）
Pompeu Fabra University（庞培法布拉大学）；Catalan Institute for Research and Advanced Studies (ICREA)（加泰罗尼亚研究与高级研究所）
Multilingual Transformer-based language models, usually pretrained on more than 100 languages, have been shown to achieve outstanding results in a wide range of cross-lingual transfer tasks. However, it remains unknown whether the optimization for different languages conditions the capacity of the models to generalize over syntactic structures, and how languages with syntactic phenomena of different complexity are affected. In this work, we explore the syntactic generalization capabilities of the monolingual and multilingual versions of BERT and RoBERTa. More specifically, we evaluate the synt…
Johns Hopkins University（约翰斯·霍普金斯大学）；New York University（纽约大学）
Structural probing work has found evidence for latent syntactic information in pre-trained language models. However, much of this analysis has focused on monolingual models, and analyses of multilingual models have employed correlational methods that are confounded by the choice of probing tasks. In this study, we causally probe multilingual language models (XGLM and multilingual BERT) as well as monolingual BERT-based models across various languages; we do this by performing counterfactual perturbations on neuron activations and observing the effect on models' subject-verb agreement probabili…
University of Oslo（奥斯陆大学）；University of Copenhagen（哥本哈根大学）
In order to preserve word-order information in a non-autoregressive setting, transformer architectures tend to include positional knowledge, by (for instance) adding positional encodings to token embeddings. Several modifications have been proposed over the sinusoidal positional encodings used in the original transformer architecture; these include, for instance, separating position encodings and token embeddings, or directly modifying attention weights based on the distance between word pairs. We first show that surprisingly, while these modifications tend to improve monolingual language mode…
The zero-shot cross-lingual ability of models pretrained on multilingual and even monolingual corpora has spurred many hypotheses to explain this intriguing empirical result. However, due to the costs of pretraining, most research uses public models whose pretraining methodology, such as the choice of tokenization, corpus size, and computational budget, might differ drastically. When researchers pretrain their own models, they often do so under a constrained budget, and the resulting models might underperform significantly compared to SOTA models. These experimental differences led to various …
HiTZ Center, University of the Basque Country (UPV/EHU)（巴斯克大学 HiTZ 中心）；DeepMind
State-of-the-art unsupervised multilingual models (e.g., multilingual BERT) have been shown to generalize in a zero-shot cross-lingual setting. This generalization ability has been attributed to the use of a shared subword vocabulary and joint training across multiple languages giving rise to deep multilingual abstractions. We evaluate this hypothesis by designing an alternative approach that transfers a monolingual model to new languages at the lexical level. More concretely, we first train a transformer-based masked language model on one language, and transfer it to a new language by learnin…
We present a family of neural-network--inspired models for computing continuous word representations, specifically designed to exploit both monolingual and multilingual text. This framework allows us to perform unsupervised training of embeddings that exhibit higher accuracy on syntactic and semantic compositionality, as well as multilingual semantic similarity, compared to previous models trained in an unsupervised fashion. We also show that such multilingual embeddings, optimized for semantic similarity, can improve the performance of statistical machine translation with respect to how it ha…
We propose a straightforward vocabulary adaptation scheme to extend the language capacity of multilingual machine translation models, paving the way towards efficient continual learning for multilingual machine translation. Our approach is suitable for large-scale datasets, applies to distant languages with unseen scripts, incurs only minor degradation on the translation performance for the original language pairs and provides competitive performance even in the case where we only possess monolingual data for the new languages.
Applied Linguistics and Technology Program（应用语言学与技术项目）；Iowa State University, USA（美国爱荷华州立大学）；Department of Informatics（信息学系）；University of Oslo, Norway（挪威奥斯陆大学）
The Common European Framework of Reference (CEFR) guidelines describe language proficiency of learners on a scale of 6 levels. While the description of CEFR guidelines is generic across languages, the development of automated proficiency classification systems for different languages follow different approaches. In this paper, we explore universal CEFR classification using domain-specific and domain-agnostic, theory-guided as well as data-driven features. We report the results of our preliminary experiments in monolingual, cross-lingual, and multilingual classification with three languages: Ge…
Nowadays, digital news articles are widely available, published by various editors and often written in different languages. This large volume of diverse and unorganized information makes human reading very difficult or almost impossible. This leads to a need for algorithms able to arrange high amount of multilingual news into stories. To this purpose, we extend previous works on Topic Detection and Tracking, and propose a new system inspired from newsLens. We process articles per batch, looking for monolingual local topics which are then linked across time and languages. Here, we introduce a …
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Microsoft（微软）
While monolingual data has been shown to be useful in improving bilingual neural machine translation (NMT), effectively and efficiently leveraging monolingual data for Multilingual NMT (MNMT) systems is a less explored area. In this work, we propose a multi-task learning (MTL) framework that jointly trains the model with the translation task on bitext data and two denoising tasks on the monolingual data. We conduct extensive empirical studies on MNMT systems with 10 language pairs from WMT datasets. We show that the proposed approach can effectively improve the translation quality for both hig…
In this paper, we show that Multilingual BERT (M-BERT), released by Devlin et al. (2018) as a single language model pre-trained from monolingual corpora in 104 languages, is surprisingly good at zero-shot cross-lingual model transfer, in which task-specific annotations in one language are used to fine-tune the model for evaluation in another language. To understand why, we present a large number of probing experiments, showing that transfer is possible even to languages in different scripts, that transfer works best between typologically similar languages, that monolingual corpora can train mo…
This paper presents BiPaR, a bilingual parallel novel-style machine reading comprehension (MRC) dataset, developed to support multilingual and cross-lingual reading comprehension. The biggest difference between BiPaR and existing reading comprehension datasets is that each triple (Passage, Question, Answer) in BiPaR is written parallelly in two languages. We collect 3,667 bilingual parallel paragraphs from Chinese and English novels, from which we construct 14,668 parallel question-answer pairs via crowdsourced workers following a strict quality control procedure. We analyze BiPaR in depth and…
Cross-lingual transfer of multilingual models on low resource African Languages
School of Computer Science and Engineering, Vellore Institute of Technology（韦洛尔理工学院计算机科学与工程学院）；Department of Computer Science, University of Pretoria（比勒陀利亚大学计算机科学系）
University of Cambridge（剑桥大学）；University of Mannheim（曼海姆大学）
In this work we present a systematic empirical study focused on the suitability of the state-of-the-art multilingual encoders for cross-lingual document and sentence retrieval tasks across a number of diverse language pairs. We first treat these models as multilingual text encoders and benchmark their performance in unsupervised ad-hoc sentence- and document-level CLIR. In contrast to supervised language understanding, our results indicate that for unsupervised document-level CLIR -- a setup with no relevance judgments for IR-specific fine-tuning -- pretrained multilingual encoders on average …
The use of multilingual language models for tasks in low and high-resource languages has been a success story in deep learning. In recent times, Arabic has been receiving widespread attention on account of its dialectal variance. While prior research studies have tried to adapt these multilingual models for dialectal variants of Arabic, it still remains a challenging problem owing to the lack of sufficient monolingual dialectal data and parallel translation data of such dialectal variants. It remains an open problem on whether the limited dialectical data can be used to improve the models trai…
Many recent works have demonstrated the benefits of knowledge graph embeddings in completing monolingual knowledge graphs. Inasmuch as related knowledge bases are built in several different languages, achieving cross-lingual knowledge alignment will help people in constructing a coherent knowledge base, and assist machines in dealing with different expressions of entity relationships across diverse human languages. Unfortunately, achieving this highly desirable crosslingual alignment by human labor is very costly and errorprone. Thus, we propose MTransE, a translation-based model for multiling…
We present Emu, a system that semantically enhances multilingual sentence embeddings. Our framework fine-tunes pre-trained multilingual sentence embeddings using two main components: a semantic classifier and a language discriminator. The semantic classifier improves the semantic similarity of related sentences, whereas the language discriminator enhances the multilinguality of the embeddings via multilingual adversarial training. Our experimental results based on several language pairs show that our specialized embeddings outperform the state-of-the-art multilingual sentence embedding model o…
MIT Computer Science and Artificial Intelligence Laboratory（麻省理工学院计算机科学与人工智能实验室）；Le Mans University（勒芒大学）
We propose a simple and effective cross-lingual transfer learning method to adapt monolingual wav2vec-2.0 models for Automatic Speech Recognition (ASR) in resource-scarce languages. We show that a monolingual wav2vec-2.0 is a good few-shot ASR learner in several languages. We improve its performance further via several iterations of Dropout Uncertainty-Driven Self-Training (DUST) by using a moderate-sized unlabeled speech dataset in the target language. A key finding of this work is that the adapted monolingual wav2vec-2.0 achieves similar performance as the topline multilingual XLSR model, wh…
To Each Language Its Tokenizer: Modular Tokenizers for Efficient Multilingual LLMs
Kyutai；Sorbonne Université（索邦大学）；CNRS（法国国家科学研究中心）；Institute of Intelligent Systems and Robotics（智能系统与机器人研究所）；Univ. Grenoble Alpes（格勒诺布尔阿尔卑斯大学）；Grenoble INP（格勒诺布尔国立理工学院）；LIG（格勒诺布尔信息学实验室）
