Large Language Models (LLMs) show remarkable performance on a wide variety of tasks. Most LLMs split text into multi-character tokens and process them as atomic units without direct access to individual characters. This raises the question: To what extent can LLMs learn orthographic information? To answer this, we propose a new benchmark, CUTE, which features a collection of tasks designed to test the orthographic knowledge of LLMs. We evaluate popular LLMs on CUTE, finding that most of them seem to know the spelling of their tokens, yet fail to use this information effectively to manipulate text, calling into question how much of this knowledge is generalizable.

The role of subword segmentation in relation to capturing morphological patterns in LLMs is currently not well explored. Ideally, one would train models like GPT using various segmentations and evaluate how well word meanings are captured. Since this is not computationally feasible, we group words according to their segmentation properties and compare how well a model can solve a linguistic task for these groups. We study two criteria: (i) adherence to morpheme boundaries and (ii) the segmentation consistency of the different inflected forms of a lemma. We select word forms with high and low values for these criteria and carry out experiments on GPT-4o’s ability to capture verbal inflection for 10 languages. Our results indicate that in particular the criterion of segmentation consistency can help to predict the model’s ability to recognize and generate the lemma from an inflected form, providing evidence that subword segmentation is relevant.

Text style transfer (TST) aims to modify the style of a text without altering its original meaning. Large language models (LLMs) demonstrate superior performance across multiple tasks, including TST. However, in zero-shot setups, they tend to directly copy a significant portion of the input text to the output without effectively changing its style. To enhance the stylistic variety and fluency of the text, we present sNeuron-TST, a novel approach for steering LLMs using style-specific neurons in TST. Specifically, we identify neurons associated with the source and target styles and deactivate source-style-only neurons to give target-style words a higher probability, aiming to enhance the stylistic diversity of the generated text. However, we find that this deactivation negatively impacts the fluency of the generated text, which we address by proposing an improved contrastive decoding method that accounts for rapid token probability shifts across layers caused by deactivated source-style neurons. Empirical experiments demonstrate the effectiveness of the proposed method on six benchmarks, encompassing formality, toxicity, politics, politeness, authorship, and sentiment.

For subjective tasks such as hate detection, where people perceive hate differently, the Large Language Model’s (LLM) ability to represent diverse groups is unclear. By including additional context in prompts, we comprehensively analyze LLM’s sensitivity to geographical priming, persona attributes, and numerical information to assess how well the needs of various groups are reflected. Our findings on two LLMs, five languages, and six datasets reveal that mimicking persona-based attributes leads to annotation variability. Meanwhile, incorporating geographical signals leads to better regional alignment. We also find that the LLMs are sensitive to numerical anchors, indicating the ability to leverage community-based flagging efforts and exposure to adversaries. Our work provides preliminary guidelines and highlights the nuances of applying LLMs in culturally sensitive cases.

We present a research agenda focused on efficiently extracting, assuring quality, and consolidating textual company sustainability information to address urgent climate change decision-making needs. Starting from the goal to create integrated FAIR (Findable, Accessible, Interoperable, Reusable) climate-related data, we identify research needs pertaining to the technical aspects of information extraction as well as to the design of the integrated sustainability datasets that we seek to compile. Regarding extraction, we leverage technological advancements, particularly in large language models (LLMs) and Retrieval-Augmented Generation (RAG) pipelines, to unlock the underutilized potential of unstructured textual information contained in corporate sustainability reports. In applying these techniques, we review key challenges, which include the retrieval and extraction of CO₂ emission values from PDF documents, especially from unstructured tables and graphs therein, and the validation of automatically extracted data through comparisons with human-annotated values. We also review how existing use cases and practices in climate risk analytics relate to choices of what textual information should be extracted and how it could be linked to existing structured data.

To democratize large language models (LLMs) to most natural languages, it is imperative to make these models capable of understanding and generating texts in many languages, in particular low-resource ones. While recent multilingual LLMs demonstrate remarkable performance in such capabilities, these LLMs still support a limited number of human languages due to the lack of training data for low resource languages. Moreover, these LLMs are not yet aligned with human preference for downstream tasks, which is crucial for the success of LLMs in English. In this paper, we introduce xLLaMA-100 and xBLOOM-100 (collectively xLLMs-100), which scale the multilingual capabilities of LLaMA and BLOOM to 100 languages. To do so, we construct two datasets: a multilingual instruction dataset including 100 languages, which represents the largest language coverage to date, and a cross-lingual human feedback dataset encompassing 30 languages. We perform multilingual instruction tuning on the constructed instruction data and further align the LLMs with human feedback using the DPO algorithm on our cross-lingual human feedback dataset. We evaluate the multilingual understanding and generating capabilities of xLLMs-100 on five multilingual benchmarks. Experimental results show that xLLMs-100 consistently outperforms its peers across the benchmarks by considerable margins, defining a new state-of-the-art multilingual LLM that supports 100 languages.

Cross-lingual alignment, the meaningful similarity of representations across languages in multilingual language models, has been an active field of research in recent years. We survey the literature of techniques to improve cross-lingual alignment, providing a taxonomy of methods and summarising insights from throughout the field. We present different understandings of cross-lingual alignment and their limitations. We provide a qualitative summary of results from a number of surveyed papers. Finally, we discuss how these insights may be applied not only to encoder models, where this topic has been heavily studied, but also to encoder-decoder or even decoder-only models, and argue that an effective trade-off between language-neutral and language-specific information is key.

Abusive language detection has drawn increasing interest in recent years. However, a less systematically explored obstacle is label imbalance, i.e., the amount of abusive data is much lower than non-abusive data, leading to performance issues. The aim of this work is to conduct a comprehensive comparative study of popular methods for addressing the class imbalance issue. We explore 10 well-known approaches on 8 datasets with distinct characteristics: binary or multi-class, moderately or largely imbalanced, focusing on various types of abuse, etc. Additionally, we pro-pose two novel methods specialized for abuse detection: AbusiveLexiconAug and ExternalDataAug, which enrich the training data using abusive lexicons and external abusive datasets, respectively. We conclude that: 1) our AbusiveLexiconAug approach, random oversampling, and focal loss are the most versatile methods on various datasets; 2) focal loss tends to yield peak model performance; 3) oversampling and focal loss provide promising results for binary datasets and small multi-class sets, while undersampling and weighted cross-entropy are more suitable for large multi-class sets; 4) most methods are sensitive to hyperparameters, yet our suggested choice of hyperparameters provides a good starting point.

The interplay of cultural and linguistic elements that characterizes metaphorical language poses a substantial challenge for both human comprehension and machine processing. This challenge goes beyond monolingual settings and becomes particularly complex in translation, even more so in automatic translation. We present VOLIMET, a corpus of 2,916 parallel sentences containing gold standard alignments of metaphorical verb-object pairs and their literal paraphrases, e.g., tackle/address question, from English to German and French. On the one hand, the parallel nature of our corpus enables us to explore monolingual patterns for metaphorical vs. literal uses in English. On the other hand, we investigate different aspects of cross-lingual translations into German and French and the extent to which metaphoricity and literalness in the source language are transferred to the target languages. Monolingually, our findings reveal clear preferences in using metaphorical or literal uses of verb-object pairs. Cross-lingually, we observe a rich variability in translations as well as different behaviors for our two target languages.

We empirically study the ability of a Large Language Model (gpt-3.5-turbo-instruct) to understand morphologically complex words. In our experiments, we looked at a variety of tasks to analyse German compounds with regard to compositional word formation and derivation, such as identifying the head noun of existing and novel compounds, identifying the shared verb stem between two words, or recognizing words constructed with inappropriately used derivation morphemes as invalid. Our results show that the language model is generally capable of solving most tasks, except for the task of identifying ill-formed word forms. While the model demonstrated a good overall understanding of complex words and their word-internal structure, the results also suggest that there is no formal knowledge of derivational rules, but rather an interpretation of the observed word parts to derive the meaning of a word.

Due to the broad range of social media platforms, the requirements of abusive language detection systems are varied and ever-changing. Already a large set of annotated corpora with different properties and label sets were created, such as hate or misogyny detection, but the form and targets of abusive speech are constantly evolving. Since, the annotation of new corpora is expensive, in this work we leverage datasets we already have, covering a wide range of tasks related to abusive language detection. Our goal is to build models cheaply for a new target label set and/or language, using only a few training examples of the target domain. We propose a two-step approach: first we train our model in a multitask fashion. We then carry out few-shot adaptation to the target requirements. Our experiments show that using already existing datasets and only a few-shots of the target task the performance of models improve both monolingually and across languages. Our analysis also shows that our models acquire a general understanding of abusive language, since they improve the prediction of labels which are present only in the target dataset and can benefit from knowledge about labels which are not directly used for the target task.

Ancient Mesopotamian literature is riddled with gaps, caused by the decay and fragmentation of its writing material, clay tablets. The discovery of overlaps between fragments allows reconstruction to advance, but it is a slow and unsystematic process. Since new pieces are found and digitized constantly, NLP techniques can help to identify fragments and match them with existing text collections to restore complete literary works. We compare a number of approaches and determine that a character-level n-gram-based similarity matching approach works well for this problem, leading to a large speed-up for researchers in Assyriology.

Although unsupervised neural machine translation (UNMT) has achieved success in many language pairs, the copying problem, i.e., directly copying some parts of the input sentence as the translation, is common among distant language pairs, especially when low-resource languages are involved. We find this issue is closely related to an unexpected copying behavior during online back-translation (BT). In this work, we propose a simple but effective training schedule that incorporates a language discriminator loss. The loss imposes constraints on the intermediate translation so that the translation is in the desired language. By conducting extensive experiments on different language pairs, including similar and distant, high and low-resource languages, we find that our method alleviates the copying problem, thus improving the translation performance on low-resource languages.

Pretrained language models (PLMs) are trained on massive corpora, but often need to specialize to specific domains. A parameter-efficient adaptation method suggests training an adapter for each domain on the task of language modeling. This leads to good in-domain scores but can be impractical for domain- or resource-restricted settings. A solution is to use a related-domain adapter for the novel domain at test time. In this paper, we introduce AdapterSoup, an approach that performs weight-space averaging of adapters trained on different domains. Our approach is embarrassingly parallel: first, we train a set of domain-specific adapters; then, for each novel domain, we determine which adapters should be averaged at test time. We present extensive experiments showing that AdapterSoup consistently improves performance to new domains without extra training. We also explore weight averaging of adapters trained on the same domain with different hyper-parameters, and show that it preserves the performance of a PLM on new domains while obtaining strong in-domain results. We explore various approaches for choosing which adapters to combine, such as text clustering and semantic similarity. We find that using clustering leads to the most competitive results on novel domains.

Pre-trained multilingual language models (PMLMs) are commonly used when dealing with data from multiple languages and cross-lingual transfer. However, PMLMs are trained on varying amounts of data for each language. In practice this means their performance is often much better on English than many other languages. We explore to what extent this also applies to moral norms. Do the models capture moral norms from English and impose them on other languages? Do the models exhibit random and thus potentially harmful beliefs in certain languages? Both these issues could negatively impact cross-lingual transfer and potentially lead to harmful outcomes. In this paper, we (1) apply the MORALDIRECTION framework to multilingual models, comparing results in German, Czech, Arabic, Chinese, and English, (2) analyse model behaviour on filtered parallel subtitles corpora, and (3) apply the models to a Moral Foundations Questionnaire, comparing with human responses from different countries. Our experiments demonstrate that, indeed, PMLMs encode differing moral biases, but these do not necessarily correspond to cultural differences or commonalities in human opinions. We release our code and models.

Previous work has shown that the representations output by contextual language models are more anisotropic than static type embeddings, and typically display outlier dimensions. This seems to be true for both monolingual and multilingual models, although much less work has been done on the multilingual context. Why these outliers occur and how they affect the representations is still an active area of research. We investigate outlier dimensions and their relationship to anisotropy in multiple pre-trained multilingual language models. We focus on cross-lingual semantic similarity tasks, as these are natural tasks for evaluating multilingual representations. Specifically, we examine sentence representations. Sentence transformers which are fine-tuned on parallel resources (that are not always available) perform better on this task, and we show that their representations are more isotropic. However, we aim to improve multilingual representations in general. We investigate how much of the performance difference can be made up by only transforming the embedding space without fine-tuning, and visualise the resulting spaces. We test different operations: Removing individual outlier dimensions, cluster-based isotropy enhancement, and ZCA whitening. We publish our code for reproducibility.

Despite advances in multilingual neural machine translation (MNMT), we argue that there are still two major challenges in this area: data imbalance and representation degeneration. The data imbalance problem refers to the imbalance in the amount of parallel corpora for all language pairs, especially for long-tail languages (i.e., very low-resource languages). The representation degeneration problem refers to the problem of encoded tokens tending to appear only in a small subspace of the full space available to the MNMT model. To solve these two issues, we propose Bi-ACL, a framework which only requires target-side monolingual data and a bilingual dictionary to improve the performance of the MNMT model. We define two modules, named bidirectional autoencoder and bidirectional contrastive learning, which we combine with an online constrained beam search and a curriculum learning sampling strategy. Extensive experiments show that our proposed method is more effective than strong baselines both in long-tail languages and in high-resource languages. We also demonstrate that our approach is capable of transferring knowledge between domains and languages in zero-shot scenarios.

We study whether linguistic information in pre-trained multilingual language models can be accessed by human language: So far, there is no easy method to directly obtain linguistic information and gain insights into the linguistic principles encoded in such models. We use the technique of prompting and formulate linguistic tasks to test the LM’s access to explicit grammatical principles and study how effective this method is at providing access to linguistic features. Our experiments on German, Icelandic and Spanish show that some linguistic properties can in fact be accessed through prompting, whereas others are harder to capture.

Many natural language processing (NLP) tasks are naturally imbalanced, as some target categories occur much more frequently than others in the real world. In such scenarios, current NLP models tend to perform poorly on less frequent classes. Addressing class imbalance in NLP is an active research topic, yet, finding a good approach for a particular task and imbalance scenario is difficult. In this survey, the first overview on class imbalance in deep-learning based NLP, we first discuss various types of controlled and real-world class imbalance. Our survey then covers approaches that have been explicitly proposed for class-imbalanced NLP tasks or, originating in the computer vision community, have been evaluated on them. We organize the methods by whether they are based on sampling, data augmentation, choice of loss function, staged learning, or model design. Finally, we discuss open problems and how to move forward.

Large multilingual models trained with self-supervision achieve state-of-the-art results in a wide range of natural language processing tasks. Self-supervised pretrained models are often fine-tuned on parallel data from one or multiple language pairs for machine translation. Multilingual fine-tuning improves performance on low-resource languages but requires modifying the entire model and can be prohibitively expensive. Training a new adapter on each language pair or training a single adapter on all language pairs without updating the pretrained model has been proposed as a parameter-efficient alternative. However, the former does not permit any sharing between languages, while the latter shares parameters for all languages and is susceptible to negative interference. In this paper, we propose training language-family adapters on top of mBART-50 to facilitate cross-lingual transfer. Our approach outperforms related baselines, yielding higher translation scores on average when translating from English to 17 different low-resource languages. We also show that language-family adapters provide an effective method to translate to languages unseen during pretraining.

Pre-trained multilingual language models are the foundation of many NLP approaches, including cross-lingual transfer solutions. However, languages with small available monolingual corpora are often not well-supported by these models leading to poor performance. We propose an unsupervised approach to improve the cross-lingual representations of low-resource languages by bootstrapping word translation pairs from monolingual corpora and using them to improve language alignment in pre-trained language models. We perform experiments on nine languages, using contextual word retrieval and zero-shot named entity recognition to measure both intrinsic cross-lingual word representation quality and downstream task performance, showing improvements on both tasks. Our results show that it is possible to improve pre-trained multilingual language models by relying only on non-parallel resources.

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, leveraging the high quality alignment of our static embeddings to better align the representation space of XLM-R. We show positive results for multiple complex semantic tasks. We release the static embeddings and the continued pre-training code. Unlike most previous work, our continued pre-training approach does not require parallel text.

We present a literature and empirical survey that critically assesses the state of the art in character-level modeling for machine translation (MT). Despite evidence in the literature that character-level systems are comparable with subword systems, they are virtually never used in competitive setups in WMT competitions. We empirically show that even with recent modeling innovations in character-level natural language processing, character-level MT systems still struggle to match their subword-based counterparts. Character-level MT systems show neither better domain robustness, nor better morphological generalization, despite being often so motivated. However, we are able to show robustness towards source side noise and that translation quality does not degrade with increasing beam size at decoding time.

Multilingual neural machine translation models (MNMT) yield state-of-the-art performance when evaluated on data from a domain and language pair seen at training time. However, when a MNMT model is used to translate under domain shift or to a new language pair, performance drops dramatically. We consider a very challenging scenario: adapting the MNMT model both to a new domain and to a new language pair at the same time. In this paper, we propose m^4Adapter (Multilingual Multi-Domain Adaptation for Machine Translation with a Meta-Adapter), which combines domain and language knowledge using meta-learning with adapters. We present results showing that our approach is a parameter-efficient solution which effectively adapts a model to both a new language pair and a new domain, while outperforming other adapter methods. An ablation study also shows that our approach more effectively transfers domain knowledge across different languages and language information across different domains.

Contextualized word embeddings have emerged as the most important tool for performing NLP tasks in a large variety of languages. In order to improve the cross- lingual representation and transfer learning quality, contextualized embedding alignment techniques, such as mapping and model fine-tuning, are employed. Existing techniques however are time-, data- and computational resource-intensive. In this paper we analyze these techniques by utilizing three tasks: bilingual lexicon induction (BLI), word retrieval and cross-lingual natural language inference (XNLI) for a high resource (German-English) and a low resource (Bengali-English) language pair. In contrast to previous works which focus only on a few popular models, we compare five multilingual and seven monolingual language models and investigate the effect of various aspects on their performance, such as vocabulary size, number of languages used for training and number of parameters. Additionally, we propose a parameter-, data- and runtime-efficient technique which can be trained with 10% of the data, less than 10% of the time and have less than 5% of the trainable parameters compared to model fine-tuning. We show that our proposed method is competitive with resource heavy models, even outperforming them in some cases, even though it relies on less resource

Bilingual Word Embeddings (BWEs) are one of the cornerstones of cross-lingual transfer of NLP models. They can be built using only monolingual corpora without supervision leading to numerous works focusing on unsupervised BWEs. However, most of the current approaches to build unsupervised BWEs do not compare their results with methods based on easy-to-access cross-lingual signals. In this paper, we argue that such signals should always be considered when developing unsupervised BWE methods. The two approaches we find most effective are: 1) using identical words as seed lexicons (which unsupervised approaches incorrectly assume are not available for orthographically distinct language pairs) and 2) combining such lexicons with pairs extracted by matching romanized versions of words with an edit distance threshold. We experiment on thirteen non-Latin languages (and English) and show that such cheap signals work well and that they outperform using more complex unsupervised methods on distant language pairs such as Chinese, Japanese, Kannada, Tamil, and Thai. In addition, they are even competitive with the use of high-quality lexicons in supervised approaches. Our results show that these training signals should not be neglected when building BWEs, even for distant languages.

We propose the neural string edit distance model for string-pair matching and string transduction based on learnable string edit distance. We modify the original expectation-maximization learned edit distance algorithm into a differentiable loss function, allowing us to integrate it into a neural network providing a contextual representation of the input. We evaluate on cognate detection, transliteration, and grapheme-to-phoneme conversion, and show that we can trade off between performance and interpretability in a single framework. Using contextual representations, which are difficult to interpret, we match the performance of state-of-the-art string-pair matching models. Using static embeddings and a slightly different loss function, we force interpretability, at the expense of an accuracy drop.

This paper summarizes the results of our test suite evaluation with a main focus on morphology for the language pairs English to/from German. We look at the translation of morphologically complex words (DE–EN), and evaluatewhether English noun phrases are translated as compounds vs. phrases into German. Furthermore, we investigate the preservation of morphological features (gender in EN–DE pronoun translation and number in morpho-syntacticallycomplex structures for DE–EN). Our results indicate that systems are able to interpret linguistic structures to obtain relevant information, but also that translation becomes more challenging with increasing complexity, as seen, for example, when translating words with negation or non-concatenative properties, and for the morecomplex cases of the pronoun translation task.

We present the findings of the WMT2022Shared Tasks in Unsupervised MT and VeryLow Resource Supervised MT with experiments on the language pairs German to/fromUpper Sorbian, German to/from Lower Sorbian and Lower Sorbian to/from Upper Sorbian. Upper and Lower Sorbian are minoritylanguages spoken in the Eastern parts of Germany. There are active language communitiesworking on the preservation of the languageswho also made the data used in this Shared Taskavailable.In total, four teams participated on this SharedTask, with submissions from three teams for theunsupervised sub task, and submissions fromall four teams for the supervised sub task. Inthis overview paper, we present and discuss theresults.

We consider two problems of NMT domain adaptation using meta-learning. First, we want to reach domain robustness, i.e., we want to reach high quality on both domains seen in the training data and unseen domains. Second, we want our systems to be adaptive, i.e., making it possible to finetune systems with just hundreds of in-domain parallel sentences. We study the domain adaptability of meta-learning when improving the domain robustness of the model. In this paper, we propose a novel approach, RMLNMT (Robust Meta-Learning Framework for Neural Machine Translation Domain Adaptation), which improves the robustness of existing meta-learning models. More specifically, we show how to use a domain classifier in curriculum learning and we integrate the word-level domain mixing model into the meta-learning framework with a balanced sampling strategy. Experiments on English-German and English-Chinese translation show that RMLNMT improves in terms of both domain robustness and domain adaptability in seen and unseen domains.

Cross-lingual word embeddings (CLWEs) have proven indispensable for various natural language processing tasks, e.g., bilingual lexicon induction (BLI). However, the lack of data often impairs the quality of representations. Various approaches requiring only weak cross-lingual supervision were proposed, but current methods still fail to learn good CLWEs for languages with only a small monolingual corpus. We therefore claim that it is necessary to explore further datasets to improve CLWEs in low-resource setups. In this paper we propose to incorporate data of related high-resource languages. In contrast to previous approaches which leverage independently pre-trained embeddings of languages, we (i) train CLWEs for the low-resource and a related language jointly and (ii) map them to the target language to build the final multilingual space. In our experiments we focus on Occitan, a low-resource Romance language which is often neglected due to lack of resources. We leverage data from French, Spanish and Catalan for training and evaluate on the Occitan-English BLI task. By incorporating supporting languages our method outperforms previous approaches by a large margin. Furthermore, our analysis shows that the degree of relatedness between an incorporated language and the low-resource language is critically important.

Good quality monolingual word embeddings (MWEs) can be built for languages which have large amounts of unlabeled text. MWEs can be aligned to bilingual spaces using only a few thousand word translation pairs. For low resource languages training MWEs monolingually results in MWEs of poor quality, and thus poor bilingual word embeddings (BWEs) as well. This paper proposes a new approach for building BWEs in which the vector space of the high resource source language is used as a starting point for training an embedding space for the low resource target language. By using the source vectors as anchors the vector spaces are automatically aligned during training. We experiment on English-German, English-Hiligaynon and English-Macedonian. We show that our approach results not only in improved BWEs and bilingual lexicon induction performance, but also in improved target language MWE quality as measured using monolingual word similarity.

Sentence weighting is a simple and powerful domain adaptation technique. We carry out domain classification for computing sentence weights with 1) language model cross entropy difference 2) a convolutional neural network 3) a Recursive Neural Tensor Network. We compare these approaches with regard to domain classification accuracy and and study the posterior probability distributions. Then we carry out NMT experiments in the scenario where we have no in-domain parallel corpora and and only very limited in-domain monolingual corpora. Here and we use the domain classifier to reweight the sentences of our out-of-domain training corpus. This leads to improvements of up to 2.1 BLEU for German to English translation.

Successful methods for unsupervised neural machine translation (UNMT) employ cross-lingual pretraining via self-supervision, often in the form of a masked language modeling or a sequence generation task, which requires the model to align the lexical- and high-level representations of the two languages. While cross-lingual pretraining works for similar languages with abundant corpora, it performs poorly in low-resource and distant languages. Previous research has shown that this is because the representations are not sufficiently aligned. In this paper, we enhance the bilingual masked language model pretraining with lexical-level information by using type-level cross-lingual subword embeddings. Empirical results demonstrate improved performance both on UNMT (up to 4.5 BLEU) and bilingual lexicon induction using our method compared to a UNMT baseline.

We address the task of automatic hate speech detection for low-resource languages. Rather than collecting and annotating new hate speech data, we show how to use cross-lingual transfer learning to leverage already existing data from higher-resource languages. Using bilingual word embeddings based classifiers we achieve good performance on the target language by training only on the source dataset. Using our transferred system we bootstrap on unlabeled target language data, improving the performance of standard cross-lingual transfer approaches. We use English as a high resource language and German as the target language for which only a small amount of annotated corpora are available. Our results indicate that cross-lingual transfer learning together with our approach to leverage additional unlabeled data is an effective way of achieving good performance on low-resource target languages without the need for any target-language annotations.

Achieving satisfying performance in machine translation on domains for which there is no training data is challenging. Traditional supervised domain adaptation is not suitable for addressing such zero-resource domains because it relies on in-domain parallel data. We show that when in-domain parallel data is not available, access to document-level context enables better capturing of domain generalities compared to only having access to a single sentence. Having access to more information provides a more reliable domain estimation. We present two document-level Transformer models which are capable of using large context sizes and we compare these models against strong Transformer baselines. We obtain improvements for the two zero-resource domains we study. We additionally provide an analysis where we vary the amount of context and look at the case where in-domain data is available.

This paper describes the submission of LMU Munich to the WMT 2021 multilingual machine translation task for small track #1, which studies translation between 6 languages (Croatian, Hungarian, Estonian, Serbian, Macedonian, English) in 30 directions. We investigate the extent to which bilingual translation systems can influence multilingual translation systems. More specifically, we trained 30 bilingual translation systems, covering all language pairs, and used data augmentation technologies such as back-translation and knowledge distillation to improve the multilingual translation systems. Our best translation system scores 5 to 6 BLEU higher than a strong baseline system provided by the organizers. As seen in the dynalab leaderboard, our submission is the only fully constrained submission that uses only the corpus provided by the organizers and does not use any pre-trained models.

The performance of NMT systems has improved drastically in the past few years but the translation of multi-sense words still poses a challenge. Since word senses are not represented uniformly in the parallel corpora used for training, there is an excessive use of the most frequent sense in MT output. In this work, we propose CmBT (Contextually-mined Back-Translation), an approach for improving multi-sense word translation leveraging pre-trained cross-lingual contextual word representations (CCWRs). Because of their contextual sensitivity and their large pre-training data, CCWRs can easily capture word senses that are missing or very rare in parallel corpora used to train MT. Specifically, CmBT applies bilingual lexicon induction on CCWRs to mine sense-specific target sentences from a monolingual dataset, and then back-translates these sentences to generate a pseudo parallel corpus as additional training data for an MT system. We test the translation quality of ambiguous words on the MuCoW test suite, which was built to test the word sense disambiguation effectiveness of MT systems. We show that our system improves on the translation of difficult unseen and low frequency word senses.

We present the findings of the WMT2021 Shared Tasks in Unsupervised MT and Very Low Resource Supervised MT. Within the task, the community studied very low resource translation between German and Upper Sorbian, unsupervised translation between German and Lower Sorbian and low resource translation between Russian and Chuvash, all minority languages with active language communities working on preserving the languages, who are partners in the evaluation. Thanks to this, we were able to obtain most digital data available for these languages and offer them to the task participants. In total, six teams participated in the shared task. The paper discusses the background, presents the tasks and results, and discusses best practices for the future.

We present our submissions to the WMT21 shared task in Unsupervised and Very Low Resource machine translation between German and Upper Sorbian, German and Lower Sorbian, and Russian and Chuvash. Our low-resource systems (German↔Upper Sorbian, Russian↔Chuvash) are pre-trained on high-resource pairs of related languages. We fine-tune those systems using the available authentic parallel data and improve by iterated back-translation. The unsupervised German↔Lower Sorbian system is initialized by the best Upper Sorbian system and improved by iterated back-translation using monolingual data only.

How would you explain Bill Gates to a German? He is associated with founding a company in the United States, so perhaps the German founder Carl Benz could stand in for Gates in those contexts. This type of translation is called adaptation in the translation community. Until now, this task has not been done computationally. Automatic adaptation could be used in natural language processing for machine translation and indirectly for generating new question answering datasets and education. We propose two automatic methods and compare them to human results for this novel NLP task. First, a structured knowledge base adapts named entities using their shared properties. Second, vector-arithmetic and orthogonal embedding mappings methods identify better candidates, but at the expense of interpretable features. We evaluate our methods through a new dataset of human adaptations.

Recent high scores on pronoun translation using context-aware neural machine translation have suggested that current approaches work well. ContraPro is a notable example of a contrastive challenge set for English→German pronoun translation. The high scores achieved by transformer models may suggest that they are able to effectively model the complicated set of inferences required to carry out pronoun translation. This entails the ability to determine which entities could be referred to, identify which entity a source-language pronoun refers to (if any), and access the target-language grammatical gender for that entity. We first show through a series of targeted adversarial attacks that in fact current approaches are not able to model all of this information well. Inserting small amounts of distracting information is enough to strongly reduce scores, which should not be the case. We then create a new template test set ContraCAT, designed to individually assess the ability to handle the specific steps necessary for successful pronoun translation. Our analyses show that current approaches to context-aware NMT rely on a set of surface heuristics, which break down when translations require real reasoning. We also propose an approach for augmenting the training data, with some improvements.

Bilingual dictionary induction (BDI) is the task of accurately translating words to the target language. It is of great importance in many low-resource scenarios where cross-lingual training data is not available. To perform BDI, bilingual word embeddings (BWEs) are often used due to their low bilingual training signal requirements. They achieve high performance, but problematic cases still remain, such as the translation of rare words or named entities, which often need to be transliterated. In this paper, we enrich BWE-based BDI with transliteration information by using Bilingual Orthography Embeddings (BOEs). BOEs represent source and target language transliteration word pairs with similar vectors. A key problem in our BDI setup is to decide which information source – BWEs (or semantics) vs. BOEs (or orthography) – is more reliable for a particular word pair. We propose a novel classification-based BDI system that uses BWEs, BOEs and a number of other features to make this decision. We test our system on English-Russian BDI and show improved performance. In addition, we show the effectiveness of our BOEs by successfully using them for transliteration mining based on cosine similarity.

We describe the WMT 2020 Shared Tasks in Unsupervised MT and Very Low Resource Supervised MT. In both tasks, the community studied German to Upper Sorbian and Upper Sorbian to German MT, which is a very realistic machine translation scenario (unlike the simulated scenarios used in particular in much of the unsupervised MT work in the past). We were able to obtain most of the digital data available for Upper Sorbian, a minority language of Germany, which was the original motivation for the Unsupervised MT shared task. As we were defining the task, we also obtained a small amount of parallel data (about 60000 parallel sentences), allowing us to offer a Very Low Resource Supervised MT task as well. Six primary systems participated in the unsupervised shared task, two of these systems used additional data beyond the data released by the organizers. Ten primary systems participated in the very low resource supervised task. The paper discusses the background, presents the tasks and results, and discusses best practices for the future.

This paper describes the submission of LMU Munich to the WMT 2020 unsupervised shared task, in two language directions, German↔Upper Sorbian. Our core unsupervised neural machine translation (UNMT) system follows the strategy of Chronopoulou et al. (2020), using a monolingual pretrained language generation model (on German) and fine-tuning it on both German and Upper Sorbian, before initializing a UNMT model, which is trained with online backtranslation. Pseudo-parallel data obtained from an unsupervised statistical machine translation (USMT) system is used to fine-tune the UNMT model. We also apply BPE-Dropout to the low resource (Upper Sorbian) data to obtain a more robust system. We additionally experiment with residual adapters and find them useful in the Upper Sorbian→German direction. We explore sampling during backtranslation and curriculum learning to use SMT translations in a more principled way. Finally, we ensemble our best-performing systems and reach a BLEU score of 32.4 on German→Upper Sorbian and 35.2 on Upper Sorbian→German.

We present our systems for the WMT20 Very Low Resource MT Task for translation between German and Upper Sorbian. For training our systems, we generate synthetic data by both back- and forward-translation. Additionally, we enrich the training data with German-Czech translated from Czech to Upper Sorbian by an unsupervised statistical MT system incorporating orthographically similar word pairs and transliterations of OOV words. Our best translation system between German and Sorbian is based on transfer learning from a Czech-German system and scores 12 to 13 BLEU higher than a baseline system built using the available parallel data only.

The task of Bilingual Dictionary Induction (BDI) consists of generating translations for source language words which is important in the framework of machine translation (MT). The aim of the BUCC 2020 shared task is to perform BDI on various language pairs using comparable corpora. In this paper, we present our approach to the task of English-German and English-Russian language pairs. Our system relies on Bilingual Word Embeddings (BWEs) which are often used for BDI when only a small seed lexicon is available making them particularly effective in a low-resource setting. On the other hand, they perform well on high frequency words only. In order to improve the performance on rare words as well, we combine BWE based word similarity with word surface similarity methods, such as orthography In addition to the often used top-n translation method, we experiment with a margin based approach aiming for dynamic number of translations for each source word. We participate in both the open and closed tracks of the shared task and we show improved results of our method compared to simple vector similarity based approaches. Our system was ranked in the top-3 teams and achieved the best results for English-Russian.

This paper investigates the use of bilingual word embeddings for mining Hiligaynon translations of English words. There is very little research on Hiligaynon, an extremely low-resource language of Malayo-Polynesian origin with over 9 million speakers in the Philippines (we found just one paper). We use a publicly available Hiligaynon corpus with only 300K words, and match it with a comparable corpus in English. As there are no bilingual resources available, we manually develop a English-Hiligaynon lexicon and use this to train bilingual word embeddings. But we fail to mine accurate translations due to the small amount of data. To find out if the same holds true for a related language pair, we simulate the same low-resource setup on English to German and arrive at similar results. We then vary the size of the comparable English and German corpora to determine the minimum corpus size necessary to achieve competitive results. Further, we investigate the role of the seed lexicon. We show that with the same corpus size but with a smaller seed lexicon, performance can surpass results of previous studies. We release the lexicon of 1,200 English-Hiligaynon word pairs we created to encourage further investigation.

This paper studies strategies to model word formation in NMT using rich linguistic information, namely a word segmentation approach that goes beyond splitting into substrings by considering fusional morphology. Our linguistically sound segmentation is combined with a method for target-side inflection to accommodate modeling word formation. The best system variants employ source-side morphological analysis and model complex target-side words, improving over a standard system.

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 neutrality. Contextual embeddings are still only moderately language-neutral by default, so we propose two simple methods for achieving stronger language neutrality: first, by unsupervised centering of the representation for each language and second, by fitting an explicit projection on small parallel data. Besides, we show how to reach state-of-the-art accuracy on language identification and match the performance of statistical methods for word alignment of parallel sentences without using parallel data.

Applying the Transformer architecture on the character level usually requires very deep architectures that are difficult and slow to train. These problems can be partially overcome by incorporating a segmentation into tokens in the model. We show that by initially training a subword model and then finetuning it on characters, we can obtain a neural machine translation model that works at the character level without requiring token segmentation. We use only the vanilla 6-layer Transformer Base architecture. Our character-level models better capture morphological phenomena and show more robustness to noise at the expense of somewhat worse overall translation quality. Our study is a significant step towards high-performance and easy to train character-based models that are not extremely large.

Using a language model (LM) pretrained on two languages with large monolingual data in order to initialize an unsupervised neural machine translation (UNMT) system yields state-of-the-art results. When limited data is available for one language, however, this method leads to poor translations. We present an effective approach that reuses an LM that is pretrained only on the high-resource language. The monolingual LM is fine-tuned on both languages and is then used to initialize a UNMT model. To reuse the pretrained LM, we have to modify its predefined vocabulary, to account for the new language. We therefore propose a novel vocabulary extension method. Our approach, RE-LM, outperforms a competitive cross-lingual pretraining model (XLM) in English-Macedonian (En-Mk) and English-Albanian (En-Sq), yielding more than +8.3 BLEU points for all four translation directions.

Mining parallel sentences from comparable corpora is important. Most previous work relies on supervised systems, which are trained on parallel data, thus their applicability is problematic in low-resource scenarios. Recent developments in building unsupervised bilingual word embeddings made it possible to mine parallel sentences based on cosine similarities of source and target language words. We show that relying only on this information is not enough, since sentences often have similar words but different meanings. We detect continuous parallel segments in sentence pair candidates and rely on them when mining parallel sentences. We show better mining accuracy on three language pairs in a standard shared task on artificial data. We also provide the first experiments showing that parallel sentences mined from real life sources improve unsupervised MT. Our code is available, we hope it will be used to support low-resource MT research.

Unseen words, also called out-of-vocabulary words (OOVs), are difficult for machine translation. In neural machine translation, byte-pair encoding can be used to represent OOVs, but they are still often incorrectly translated. We improve the translation of OOVs in NMT using easy-to-obtain monolingual data. We look for OOVs in the text to be translated and translate them using simple-to-construct bilingual word embeddings (BWEs). In our MT experiments we take the 5-best candidates, which is motivated by intrinsic mining experiments. Using all five of the proposed target language words as queries we mine target-language sentences. We then back-translate, forcing the back-translation of each of the five proposed target-language OOV-translation-candidates to be the original source-language OOV. We show that by using this synthetic data to fine-tune our system the translation of OOVs can be dramatically improved. In our experiments we use a system trained on Europarl and mine sentences containing medical terms from monolingual data.

We tackle the important task of part-of-speech tagging using a neural model in the zero-resource scenario, where we have no access to gold-standard POS training data. We compare this scenario with the low-resource scenario, where we have access to a small amount of gold-standard POS training data. Our experiments focus on Ukrainian as a representative of under-resourced languages. Russian is highly related to Ukrainian, so we exploit gold-standard Russian POS tags. We consider four techniques to perform Ukrainian POS tagging: zero-shot tagging and cross-lingual annotation projection (for the zero-resource scenario), and compare these with self-training and multilingual learning (for the low-resource scenario). We find that cross-lingual annotation projection works particularly well in the zero-resource scenario.

We describe LMU Munich’s machine translation system for German→Czech translation which was used to participate in the WMT19 shared task on unsupervised news translation. We train our model using monolingual data only from both languages. The final model is an unsupervised neural model using established techniques for unsupervised translation such as denoising autoencoding and online back-translation. We bootstrap the model with masked language model pretraining and enhance it with back-translations from an unsupervised phrase-based system which is itself bootstrapped using unsupervised bilingual word embeddings.

We describe LMU Munich’s machine translation system for English→German translation which was used to participate in the WMT19 shared task on supervised news translation. We specifically participated in the document-level MT track. The system used as a primary submission is a context-aware Transformer capable of both rich modeling of limited contextual information and integration of large-scale document-level context with a less rich representation. We train this model by fine-tuning a big Transformer baseline. Our experimental results show that document-level context provides for large improvements in translation quality, and adding a rich representation of the previous sentence provides a small additional gain.

Bilingual word embeddings are useful for bilingual lexicon induction, the task of mining translations of given words. Many studies have shown that bilingual word embeddings perform well for bilingual lexicon induction but they focused on frequent words in general domains. For many applications, bilingual lexicon induction of rare and domain-specific words is of critical importance. Therefore, we design a new task to evaluate bilingual word embeddings on rare words in different domains. We show that state-of-the-art approaches fail on this task and present simple new techniques to improve bilingual word embeddings for mining rare words. We release new gold standard datasets and code to stimulate research on this task.

Mining parallel sentences from comparable corpora is of great interest for many downstream tasks. In the BUCC 2017 shared task, systems performed well by training on gold standard parallel sentences. However, we often want to mine parallel sentences without bilingual supervision. We present a simple approach relying on bilingual word embeddings trained in an unsupervised fashion. We incorporate orthographic similarity in order to handle words with similar surface forms. In addition, we propose a dynamic threshold method to decide if a candidate sentence-pair is parallel which eliminates the need to fine tune a static value for different datasets. Since we do not employ any language specific engineering our approach is highly generic. We show that our approach is effective, on three language-pairs, without the use of any bilingual signal which is important because parallel sentence mining is most useful in low resource scenarios.

Bilingual tasks, such as bilingual lexicon induction and cross-lingual classification, are crucial for overcoming data sparsity in the target language. Resources required for such tasks are often out-of-domain, thus domain adaptation is an important problem here. We make two contributions. First, we test a delightfully simple method for domain adaptation of bilingual word embeddings. We evaluate these embeddings on two bilingual tasks involving different domains: cross-lingual twitter sentiment classification and medical bilingual lexicon induction. Second, we tailor a broadly applicable semi-supervised classification method from computer vision to these tasks. We show that this method also helps in low-resource setups. Using both methods together we achieve large improvements over our baselines, by using only additional unlabeled data.

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.

Cross-sentence context can provide valuable information in Machine Translation and is critical for translation of anaphoric pronouns and for providing consistent translations. In this paper, we devise simple oracle experiments targeting coreference and coherence. Oracles are an easy way to evaluate the effect of different discourse-level phenomena in NMT using BLEU and eliminate the necessity to manually define challenge sets for this purpose. We propose two context-aware NMT models and compare them against models working on a concatenation of consecutive sentences. Concatenation models perform better, but are computationally expensive. We show that NMT models taking advantage of context oracle signals can achieve considerable gains in BLEU, of up to 7.02 BLEU for coreference and 1.89 BLEU for coherence on subtitles translation. Access to strong signals allows us to make clear comparisons between context-aware models.

We describe LMU Munich’s unsupervised machine translation systems for English↔German translation. These systems were used to participate in the WMT18 news translation shared task and more specifically, for the unsupervised learning sub-track. The systems are trained on English and German monolingual data only and exploit and combine previously proposed techniques such as using word-by-word translated data based on bilingual word embeddings, denoising and on-the-fly backtranslation.

We present the LMU Munich machine translation systems for the English–German language pair. We have built neural machine translation systems for both translation directions (English→German and German→English) and for two different domains (the biomedical domain and the news domain). The systems were used for our participation in the WMT18 biomedical translation task and in the shared task on machine translation of news. The main focus of our recent system development efforts has been on achieving improvements in the biomedical domain over last year’s strong biomedical translation engine for English→German (Huck et al., 2017a). Considerable progress has been made in the latter task, which we report on in this paper.

In this paper we describe LMU Munich’s submission for the WMT 2018 Parallel Corpus Filtering shared task which addresses the problem of cleaning noisy parallel corpora. The task of mining and cleaning parallel sentences is important for improving the quality of machine translation systems, especially for low-resource languages. We tackle this problem in a fully unsupervised fashion relying on bilingual word embeddings created without any bilingual signal. After pre-filtering noisy data we rank sentence pairs by calculating bilingual sentence-level similarities and then remove redundant data by employing monolingual similarity as well. Our unsupervised system achieved good performance during the official evaluation of the shared task, scoring only a few BLEU points behind the best systems, while not requiring any parallel training data.

We present a generative model that efficiently mines transliteration pairs in a consistent fashion in three different settings: unsupervised, semi-supervised, and supervised transliteration mining. The model interpolates two sub-models, one for the generation of transliteration pairs and one for the generation of non-transliteration pairs (i.e., noise). The model is trained on noisy unlabeled data using the EM algorithm. During training the transliteration sub-model learns to generate transliteration pairs and the fixed non-transliteration model generates the noise pairs. After training, the unlabeled data is disambiguated based on the posterior probabilities of the two sub-models. We evaluate our transliteration mining system on data from a transliteration mining shared task and on parallel corpora. For three out of four language pairs, our system outperforms all semi-supervised and supervised systems that participated in the NEWS 2010 shared task. On word pairs extracted from parallel corpora with fewer than 2% transliteration pairs, our system achieves up to 86.7% F-measure with 77.9% precision and 97.8% recall.

Translating into morphologically rich languages is difficult. Although the coverage of lemmas may be reasonable, many morphological variants cannot be learned from the training data. We present a statistical translation system that is able to produce these inflected word forms. Different from most previous work, we do not separate morphological prediction from lexical choice into two consecutive steps. Our approach is novel in that it is integrated in decoding and takes advantage of context information from both the source language and the target language sides.

Many errors in phrase-based SMT can be attributed to problems on three linguistic levels: morphological complexity in the target language, structural differences and lexical choice. We explore combinations of linguistically motivated approaches to address these problems in English-to-German SMT and show that they are complementary to one another, but also that the popular verbal pre-ordering can cause problems on the morphological and lexical level. A discriminative classifier can overcome these problems, in particular when enriching standard lexical features with features geared towards verbal inflection.

Translating prepositions is a difficult and under-studied problem in SMT. We present a novel method to improve the translation of prepositions by using noun classes to model their selectional preferences. We compare three variants of noun class information: (i) classes induced from the lexical resource GermaNet or obtained from clusterings based on either (ii) window information or (iii) syntactic features. Furthermore, we experiment with PP rule generalization. While we do not significantly improve over the baseline, our results demonstrate that (i) integrating selectional preferences as rigid class annotation in the parse tree is sub-optimal, and that (ii) clusterings based on window co-occurrence are more robust than syntax-based clusters or GermaNet classes for the task of modeling selectional preferences.