We propose procedures for evaluating and strengthening contextual embedding alignment and show that they are useful in analyzing and improving multilingual BERT. In particular, after our proposed alignment procedure, BERT exhibits significantly improved zero-shot performance on XNLI compared to the base model, remarkably matching pseudo-fully-supervised translate-train models for Bulgarian and Greek. Further, to measure the degree of alignment, we introduce a contextual version of word retrieval and show that it correlates well with downstream zero-shot transfer. Using this word retrieval task, we also analyze BERT and find that it exhibits systematic deficiencies, e.g. worse alignment for open-class parts-of-speech and word pairs written in different scripts, that are corrected by the alignment procedure. These results support contextual alignment as a useful concept for understanding large multilingual pre-trained models.
Recently, multilingual BERT works remarkably well on cross-lingual transfer tasks, superior to static non-contextualized word embeddings. In this work, we provide an in-depth experimental study to supplement the existing literature of cross-lingual ability. We compare the cross-lingual ability of non-contextualized and contextualized representation model with the same data. We found that datasize and context window size are crucial factors to the transferability.
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；University of Virginia（弗吉尼亚大学）；Microsoft Research（微软研究院）；Stanford University（斯坦福大学）
Word embeddings, which represent a word as a point in a vector space, have become ubiquitous to several NLP tasks. A recent line of work uses bilingual (two languages) corpora to learn a different vector for each sense of a word, by exploiting crosslingual signals to aid sense identification. We present a multi-view Bayesian non-parametric algorithm which improves multi-sense word embeddings by (a) using multilingual (i.e., more than two languages) corpora to significantly improve sense embeddings beyond what one achieves with bilingual information, and (b) uses a principled approach to learn …
RIKEN AIP（理化学研究所先进智能项目中心）；Tohoku University（东北大学）；Heidelberg Institute for Theoretical Studies gGmbH（海德堡理论研究所）
Pretrained contextual and non-contextual subword embeddings have become available in over 250 languages, allowing massively multilingual NLP. However, while there is no dearth of pretrained embeddings, the distinct lack of systematic evaluations makes it difficult for practitioners to choose between them. In this work, we conduct an extensive evaluation comparing non-contextual subword embeddings, namely FastText and BPEmb, and a contextual representation method, namely BERT, on multilingual named entity recognition and part-of-speech tagging. We find that overall, a combination of BERT, BPEmb…
Distributed word representations (word embeddings) have recently contributed to competitive performance in language modeling and several NLP tasks. In this work, we train word embeddings for more than 100 languages using their corresponding Wikipedias. We quantitatively demonstrate the utility of our word embeddings by using them as the sole features for training a part of speech tagger for a subset of these languages. We find their performance to be competitive with near state-of-art methods in English, Danish and Swedish. Moreover, we investigate the semantic features captured by these embed…
Static and contextual multilingual embeddings have complementary strengths. Static embeddings, while less expressive than contextual language models, can be more straightforwardly aligned across multiple languages. We combine the strengths of static and contextual models to improve multilingual representations. We extract static embeddings for 40 languages from XLM-R, validate those embeddings with cross-lingual word retrieval, and then align them using VecMap. This results in high-quality, highly multilingual static embeddings. Then we apply a novel continued pre-training approach to XLM-R, l…
Recent work on learning multilingual word representations usually relies on the use of word-level alignements (e.g. infered with the help of GIZA++) between translated sentences, in order to align the word embeddings in different languages. In this workshop paper, we investigate an autoencoder model for learning multilingual word representations that does without such word-level alignements. The autoencoder is trained to reconstruct the bag-of-word representation of given sentence from an encoded representation extracted from its translation. We evaluate our approach on a multilingual document…
Facebook AI Research（脸书人工智能研究院）；CREST ENSAE, Institut Polytechnique de Paris（巴黎理工学院ENSAE经济与统计研究中心）；Google Brain（谷歌大脑）
We consider the problem of aligning continuous word representations, learned in multiple languages, to a common space. It was recently shown that, in the case of two languages, it is possible to learn such a mapping without supervision. This paper extends this line of work to the problem of aligning multiple languages to a common space. A solution is to independently map all languages to a pivot language. Unfortunately, this degrades the quality of indirect word translation. We thus propose a novel formulation that ensures composable mappings, leading to better alignments. We evaluate our meth…
Contextual word embeddings (e.g. GPT, BERT, ELMo, etc.) have demonstrated state-of-the-art performance on various NLP tasks. Recent work with the multilingual version of BERT has shown that the model performs very well in zero-shot and zero-resource cross-lingual settings, where only labeled English data is used to finetune the model. We improve upon multilingual BERT's zero-resource cross-lingual performance via adversarial learning. We report the magnitude of the improvement on the multilingual MLDoc text classification and CoNLL 2002/2003 named entity recognition tasks. Furthermore, we show…
Multilingual Word Embeddings (MWEs) represent words from multiple languages in a single distributional vector space. Unsupervised MWE (UMWE) methods acquire multilingual embeddings without cross-lingual supervision, which is a significant advantage over traditional supervised approaches and opens many new possibilities for low-resource languages. Prior art for learning UMWEs, however, merely relies on a number of independently trained Unsupervised Bilingual Word Embeddings (UBWEs) to obtain multilingual embeddings. These methods fail to leverage the interdependencies that exist among many lang…
The lack of annotated data in many languages is a well-known challenge within the field of multilingual natural language processing (NLP). Therefore, many recent studies focus on zero-shot transfer learning and joint training across languages to overcome data scarcity for low-resource languages. In this work we (i) perform a comprehensive comparison of state-ofthe-art multilingual word and sentence encoders on the tasks of named entity recognition (NER) and part of speech (POS) tagging; and (ii) propose a new method for creating multilingual contextualized word embeddings, compare it to multip…
University of Pennsylvania（宾夕法尼亚大学）；IBM Research（IBM研究院）
We develop high performance multilingualAbstract Meaning Representation (AMR) sys-tems by projecting English AMR annotationsto other languages with weak supervision. Weachieve this goal by bootstrapping transformer-based multilingual word embeddings, in partic-ular those from cross-lingual RoBERTa (XLM-R large). We develop a novel technique forforeign-text-to-English AMR alignment, usingthe contextual word alignment between En-glish and foreign language tokens. This wordalignment is weakly supervised and relies onthe contextualized XLM-R word embeddings.We achieve a highly competitive performa…
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…
While recent work on multilingual language models has demonstrated their capacity for cross-lingual zero-shot transfer on downstream tasks, there is a lack of consensus in the community as to what shared properties between languages enable such transfer. Analyses involving pairs of natural languages are often inconclusive and contradictory since languages simultaneously differ in many linguistic aspects. In this paper, we perform a large-scale empirical study to isolate the effects of various linguistic properties by measuring zero-shot transfer between four diverse natural languages and their…
Robust state tracking for task-oriented dialogue systems currently remains restricted to a few popular languages. This paper shows that given a large-scale dialogue data set in one language, we can automatically produce an effective semantic parser for other languages using machine translation. We propose automatic translation of dialogue datasets with alignment to ensure faithful translation of slot values and eliminate costly human supervision used in previous benchmarks. We also propose a new contextual semantic parsing model, which encodes the formal slots and values, and only the last age…
Shanghai University of Finance and Economics（上海财经大学）；Southern University of Science and Technology（南方科技大学）
Multilingual pretrained language models (mPLMs) have shown their effectiveness in multilingual word alignment induction. However, these methods usually start from mBERT or XLM-R. In this paper, we investigate whether multilingual sentence Transformer LaBSE is a strong multilingual word aligner. This idea is non-trivial as LaBSE is trained to learn language-agnostic sentence-level embeddings, while the alignment extraction task requires the more fine-grained word-level embeddings to be language-agnostic. We demonstrate that the vanilla LaBSE outperforms other mPLMs currently used in the alignme…
In this paper, we propose a new approach to learn multimodal multilingual embeddings for matching images and their relevant captions in two languages. We combine two existing objective functions to make images and captions close in a joint embedding space while adapting the alignment of word embeddings between existing languages in our model. We show that our approach enables better generalization, achieving state-of-the-art performance in text-to-image and image-to-text retrieval task, and caption-caption similarity task. Two multimodal multilingual datasets are used for evaluation: Multi30k …
Amazonas State Univ.（亚马孙州立大学）；Federal Univ. of Amazonas（亚马孙联邦大学）
This paper describes the system submitted by our team (BabelEnconding) to SemEval-2020 Task 3: Predicting the Graded Effect of Context in Word Similarity. We propose an approach that relies on translation and multilingual language models in order to compute the contextual similarity between pairs of words. Our hypothesis is that evidence from additional languages can leverage the correlation with the human generated scores. BabelEnconding was applied to both subtasks and ranked among the top-3 in six out of eight task/language combinations and was the highest scoring system three times.
The skip-gram (SG) model learns word representation by predicting the words surrounding a center word from unstructured text data. However, not all words in the context window contribute to the meaning of the center word. For example, less relevant words could be in the context window, hindering the SG model from learning a better quality representation. In this paper, we propose an enhanced version of the SG that leverages context information to produce word representation. The proposed model, Contextual Skip-gram, is designed to predict contextual words with both the center words and the con…
Previous work on cross-lingual sequence labeling tasks either requires parallel data or bridges the two languages through word-byword matching. Such requirements and assumptions are infeasible for most languages, especially for languages with large linguistic distances, e.g., English and Chinese. In this work, we propose a Multilingual Language Model with deep semantic Alignment (MLMA) to generate language-independent representations for cross-lingual sequence labeling. Our methods require only monolingual corpora with no bilingual resources at all and take advantage of deep contextualized rep…
This paper describes Luminoso's participation in SemEval 2017 Task 2, "Multilingual and Cross-lingual Semantic Word Similarity", with a system based on ConceptNet. ConceptNet is an open, multilingual knowledge graph that focuses on general knowledge that relates the meanings of words and phrases. Our submission to SemEval was an update of previous work that builds high-quality, multilingual word embeddings from a combination of ConceptNet and distributional semantics. Our system took first place in both subtasks. It ranked first in 4 out of 5 of the separate languages, and also ranked first in…
There has been significant interest recently in learning multilingual word embeddings -- in which semantically similar words across languages have similar embeddings. State-of-the-art approaches have relied on expensive labeled data, which is unavailable for low-resource languages, or have involved post-hoc unification of monolingual embeddings. In the present paper, we investigate the efficacy of multilingual embeddings learned from weakly-supervised image-text data. In particular, we propose methods for learning multilingual embeddings using image-text data, by enforcing similarity between t…
Peking University（北京大学）；Huazhong University of Science and Technology（华中科技大学）；Microsoft Research（微软研究院）
Transformer has demonstrated its great power to learn contextual word representations for multiple languages in a single model. To process multilingual sentences in the model, a learnable vector is usually assigned to each language, which is called "language embedding". The language embedding can be either added to the word embedding or attached at the beginning of the sentence. It serves as a language-specific signal for the Transformer to capture contextual representations across languages. In this paper, we revisit the use of language embedding and identify several problems in the existing …
RWTH Aachen（亚琛工业大学）；L3S Research Center（L3S研究中心）；University of Mannheim（曼海姆大学）；GESIS（盖泽斯社会科学研究所）；CSH Vienna（维也纳复杂科学中心）
Adding interpretability to word embeddings represents an area of active research in text representation. Recent work has explored thepotential of embedding words via so-called polar dimensions (e.g. good vs. bad, correct vs. wrong). Examples of such recent approaches include SemAxis, POLAR, FrameAxis, and BiImp. Although these approaches provide interpretable dimensions for words, they have not been designed to deal with polysemy, i.e. they can not easily distinguish between different senses of words. To address this limitation, we present SensePOLAR, an extension of the original POLAR framewo…
In this paper, we advance the current state-of-the-art method for debiasing monolingual word embeddings so as to generalize well in a multilingual setting. We consider different methods to quantify bias and different debiasing approaches for monolingual as well as multilingual settings. We demonstrate the significance of our bias-mitigation approach on downstream NLP applications. Our proposed methods establish the state-of-the-art performance for debiasing multilingual embeddings for three Indian languages - Hindi, Bengali, and Telugu in addition to English. We believe that our work will open…
University of Wolverhampton（伍尔弗汉普顿大学）；University of Surrey（萨里大学）
Most studies on word-level Quality Estimation (QE) of machine translation focus on language-specific models. The obvious disadvantages of these approaches are the need for labelled data for each language pair and the high cost required to maintain several language-specific models. To overcome these problems, we explore different approaches to multilingual, word-level QE. We show that these QE models perform on par with the current language-specific models. In the cases of zero-shot and few-shot QE, we demonstrate that it is possible to accurately predict word-level quality for any given new la…
Multilingual pre-trained models exhibit zero-shot cross-lingual transfer, where a model fine-tuned on a source language achieves surprisingly good performance on a target language. While studies have attempted to understand transfer, they focus only on MLM, and the large number of differences between natural languages makes it hard to disentangle the importance of different properties. In this work, we specifically highlight the importance of word embedding alignment by proposing a pre-training objective (ALIGN-MLM) whose auxiliary loss guides similar words in different languages to have simil…
We release to the community six large-scale sense-annotated datasets in multiple language to pave the way for supervised multilingual Word Sense Disambiguation. Our datasets cover all the nouns in the English WordNet and their translations in other languages for a total of millions of sense-tagged sentences. Experiments prove that these corpora can be effectively used as training sets for supervised WSD systems, surpassing the state of the art for low-resourced languages and providing competitive results for English, where manually annotated training sets are accessible. The data is available …
Some Transformer-based models can perform cross-lingual transfer learning: those models can be trained on a specific task in one language and give relatively good results on the same task in another language, despite having been pre-trained on monolingual tasks only. But, there is no consensus yet on whether those transformer-based models learn universal patterns across languages. We propose a word-level task-agnostic method to evaluate the alignment of contextualized representations built by such models. We show that our method provides more accurate translated word pairs than previous method…
College of Electrical Engineering and Computer Science, National Taiwan University（台湾大学电机资讯学院）
Recently, multilingual BERT works remarkably well on cross-lingual transfer tasks, superior to static non-contextualized word embeddings. In this work, we provide an in-depth experimental study to supplement the existing literature of cross-lingual ability. We compare the cross-lingual ability of non-contextualized and contextualized representation model with the same data. We found that datasize and context window size are crucial factors to the transferability. We also observe the language-specific information in multilingual BERT. By manipulating the latent representations, we can control t…
In this work we approach the task of learning multilingual word representations in an offline manner by fitting a generative latent variable model to a multilingual dictionary. We model equivalent words in different languages as different views of the same word generated by a common latent variable representing their latent lexical meaning. We explore the task of alignment by querying the fitted model for multilingual embeddings achieving competitive results across a variety of tasks. The proposed model is robust to noise in the embedding space making it a suitable method for distributed repre…
Nara Institute of Science and Technology（奈良先端科学技术大学院大学）；NTT Communication Science Laboratories（NTT通信科学实验室）
We propose an unsupervised method to obtain cross-lingual embeddings without any parallel data or pre-trained word embeddings. The proposed model, which we call multilingual neural language models, takes sentences of multiple languages as an input. The proposed model contains bidirectional LSTMs that perform as forward and backward language models, and these networks are shared among all the languages. The other parameters, i.e. word embeddings and linear transformation between hidden states and outputs, are specific to each language. The shared LSTMs can capture the common sentence structure …
We describe the Uppsala NLP submission to SemEval-2021 Task 2 on multilingual and cross-lingual word-in-context disambiguation. We explore the usefulness of three pre-trained multilingual language models, XLM-RoBERTa (XLMR), Multilingual BERT (mBERT) and multilingual distilled BERT (mDistilBERT). We compare these three models in two setups, fine-tuning and as feature extractors. In the second case we also experiment with using dependency-based information. We find that fine-tuning is better than feature extraction. XLMR performs better than mBERT in the cross-lingual setting both with fine-tun…
We propose a novel data augmentation for labeled sentences called contextual augmentation. We assume an invariance that sentences are natural even if the words in the sentences are replaced with other words with paradigmatic relations. We stochastically replace words with other words that are predicted by a bi-directional language model at the word positions. Words predicted according to a context are numerous but appropriate for the augmentation of the original words. Furthermore, we retrofit a language model with a label-conditional architecture, which allows the model to augment sentences w…
Thai Nguyen University（太原大学）；Vietnam National University（越南国立大学）；Karlsruhe Institute of Technology（卡尔斯鲁厄理工学院）；JAIST（日本北陆先端科学技术大学院大学）
Prior works have demonstrated that a low-resource language pair can benefit from multilingual machine translation (MT) systems, which rely on many language pairs' joint training. This paper proposes two simple strategies to address the rare word issue in multilingual MT systems for two low-resource language pairs: French-Vietnamese and English-Vietnamese. The first strategy is about dynamical learning word similarity of tokens in the shared space among source languages while another one attempts to augment the translation ability of rare words through updating their embeddings during the train…
Specific lexical choices in narrative text reflect both the writer's attitudes towards people in the narrative and influence the audience's reactions. Prior work has examined descriptions of people in English using contextual affective analysis, a natural language processing (NLP) technique that seeks to analyze how people are portrayed along dimensions of power, agency, and sentiment. Our work presents an extension of this methodology to multilingual settings, which is enabled by a new corpus that we collect and a new multilingual model. We additionally show how word connotations differ acros…
``Classical'' word embeddings, such as Word2Vec, have been shown to capture the semantics of words based on their distributional properties. However, their ability to represent the different meanings that a word may have is limited. Such approaches also do not explicitly encode relations between entities, as denoted by words. Embeddings of knowledge bases (KB) capture the explicit relations between entities denoted by words, but are not able to directly capture the syntagmatic properties of these words. To our knowledge, recent research have focused on representation learning that augment the …
Bilingual word embeddings represent words of two languages in the same space, and allow to transfer knowledge from one language to the other without machine translation. The main approach is to train monolingual embeddings first and then map them using bilingual dictionaries. In this work, we present a novel method to learn bilingual embeddings based on multilingual knowledge bases (KB) such as WordNet. Our method extracts bilingual information from multilingual wordnets via random walks and learns a joint embedding space in one go. We further reinforce cross-lingual equivalence adding bilingu…
The Hong Kong University of Science and Technology（香港科技大学）；EMOS Technologies Inc.（EMOS科技公司）
We propose a multilingual model to recognize Big Five Personality traits from text data in four different languages: English, Spanish, Dutch and Italian. Our analysis shows that words having a similar semantic meaning in different languages do not necessarily correspond to the same personality traits. Therefore, we propose a personality alignment method, GlobalTrait, which has a mapping for each trait from the source language to the target language (English), such that words that correlate positively to each trait are close together in the multilingual vector space. Using these aligned embeddi…
Center for Information and Language Processing, LMU Munich（慕尼黑大学信息与语言处理中心）；Faculty of Mathematics and Physics, Charles University（查尔斯大学数学与物理学院）
Multilingual contextual embeddings, such as multilingual BERT and XLM-RoBERTa, have proved useful for many multi-lingual tasks. Previous work probed the cross-linguality of the representations indirectly using zero-shot transfer learning on morphological and syntactic tasks. We instead investigate the language-neutrality of multilingual contextual embeddings directly and with respect to lexical semantics. Our results show that contextual embeddings are more language-neutral and, in general, more informative than aligned static word-type embeddings, which are explicitly trained for language neu…
As a key natural language processing (NLP) task, word sense disambiguation (WSD) evaluates how well NLP models can understand the lexical semantics of words under specific contexts. Benefited from the large-scale annotation, current WSD systems have achieved impressive performances in English by combining supervised learning with lexical knowledge. However, such success is hard to be replicated in other languages, where we only have limited annotations.In this paper, based on the multilingual lexicon BabelNet describing the same set of concepts across languages, we propose building knowledge a…
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…
We present LINSPECTOR WEB, an open source multilingual inspector to analyze word representations. Our system provides researchers working in low-resource settings with an easily accessible web based probing tool to gain quick insights into their word embeddings especially outside of the English language. To do this we employ 16 simple linguistic probing tasks such as gender, case marking, and tense for a diverse set of 28 languages. We support probing of static word embeddings along with pretrained AllenNLP models that are commonly used for NLP downstream tasks such as named entity recognition…
NTT Communication Science Laboratories, NTT Corporation（日本电信电话株式会社NTT通信科学实验室）
We present a novel supervised word alignment method based on cross-language span prediction. We first formalize a word alignment problem as a collection of independent predictions from a token in the source sentence to a span in the target sentence. As this is equivalent to a SQuAD v2.0 style question answering task, we then solve this problem by using multilingual BERT, which is fine-tuned on a manually created gold word alignment data. We greatly improved the word alignment accuracy by adding the context of the token to the question. In the experiments using five word alignment datasets amon…
École Polytechnique Fédérale de Lausanne（洛桑联邦理工学院）
We present a new method for estimating vector space representations of words: embedding learning by concept induction. We test this method on a highly parallel corpus and learn semantic representations of words in 1259 different languages in a single common space. An extensive experimental evaluation on crosslingual word similarity and sentiment analysis indicates that concept-based multilingual embedding learning performs better than previous approaches.
We present a neural network architecture based on bidirectional LSTMs to compute representations of words in the sentential contexts. These context-sensitive word representations are suitable for, e.g., distinguishing different word senses and other context-modulated variations in meaning. To learn the parameters of our model, we use cross-lingual supervision, hypothesizing that a good representation of a word in context will be one that is sufficient for selecting the correct translation into a second language. We evaluate the quality of our representations as features in three downstream tas…
The surge of social media use brings huge demand of multilingual sentiment analysis (MSA) for unveiling cultural difference. So far, traditional methods resorted to machine translation---translating texts in other languages to English, and then adopt the methods once worked in English. However, this paradigm is conditioned by the quality of machine translation. In this paper, we propose a new deep learning paradigm to assimilate the differences between languages for MSA. We first pre-train monolingual word embeddings separately, then map word embeddings in different spaces into a shared embedd…
The capabilities and limitations of BERT and similar models are still unclear when it comes to learning syntactic abstractions, in particular across languages. In this paper, we use the task of subordinate-clause detection within and across languages to probe these properties. We show that this task is deceptively simple, with easy gains offset by a long tail of harder cases, and that BERT's zero-shot performance is dominated by word-order effects, mirroring the SVO/VSO/SOV typology.
Carnegie Mellon University（卡内基梅隆大学）；University of Washington（华盛顿大学）
We introduce new methods for estimating and evaluating embeddings of words in more than fifty languages in a single shared embedding space. Our estimation methods, multiCluster and multiCCA, use dictionaries and monolingual data; they do not require parallel data. Our new evaluation method, multiQVEC-CCA, is shown to correlate better than previous ones with two downstream tasks (text categorization and parsing). We also describe a web portal for evaluation that will facilitate further research in this area, along with open-source releases of all our methods.
In this paper we propose and carefully evaluate a sequence labeling framework which solely utilizes sparse indicator features derived from dense distributed word representations. The proposed model obtains (near) state-of-the art performance for both part-of-speech tagging and named entity recognition for a variety of languages. Our model relies only on a few thousand sparse coding-derived features, without applying any modification of the word representations employed for the different tasks. The proposed model has favorable generalization properties as it retains over 89.8% of its average PO…
Recent progress on unsupervised learning of cross-lingual embeddings in bilingual setting has given impetus to learning a shared embedding space for several languages without any supervision. A popular framework to solve the latter problem is to jointly solve the following two sub-problems: 1) learning unsupervised word alignment between several pairs of languages, and 2) learning how to map the monolingual embeddings of every language to a shared multilingual space. In contrast, we propose a simple, two-stage framework in which we decouple the above two sub-problems and solve them separately …
Computer Science Department, Bar Ilan University（巴伊兰大学计算机科学系）；Allen Institute for Artificial Intelligence（艾伦人工智能研究所）
Recent works have demonstrated that multilingual BERT (mBERT) learns rich cross-lingual representations, that allow for transfer across languages. We study the word-level translation information embedded in mBERT and present two simple methods that expose remarkable translation capabilities with no fine-tuning. The results suggest that most of this information is encoded in a non-linear way, while some of it can also be recovered with purely linear tools. As part of our analysis, we test the hypothesis that mBERT learns representations which contain both a language-encoding component and an ab…
