Ensuring the verifiability of model answers is a fundamental challenge for retrieval-augmented generation (RAG) in the question answering (QA) domain. Recently, self-citation prompting was proposed to make large language models (LLMs) generate citations to supporting documents along with their answers. However, self-citing LLMs often struggle to match the required format, refer to non-existent sources, and fail to faithfully reflect LLMs’ context usage throughout the generation. In this work, we present MIRAGE – Model Internals-based RAG Explanations – a plug-and-play approach using model internals for faithful answer attribution in RAG applications. MIRAGE detects context-sensitive answer tokens and pairs them with retrieved documents contributing to their prediction via saliency methods. We evaluate our proposed approach on a multilingual extractive QA dataset, finding high agreement with human answer attribution. On open-ended QA, MIRAGE achieves citation quality and efficiency comparable to self-citation while also allowing for a finer-grained control of attribution parameters. Our qualitative evaluation highlights the faithfulness of MIRAGE’s attributions and underscores the promising application of model internals for RAG answer attribution. Code and data released at https://github.com/Betswish/MIRAGE.

Pretrained character-level and byte-level language models have been shown to be competitive with popular subword models across a range of Natural Language Processing tasks. However, there has been little research on their effectiveness for neural machine translation (NMT), particularly within the popular pretrain-then-finetune paradigm. This work performs an extensive comparison across multiple languages and experimental conditions of character- and subword-level pretrained models (ByT5 and mT5, respectively) on NMT. We show the effectiveness of character-level modeling in translation, particularly in cases where fine-tuning data is limited. In our analysis, we show how character models’ gains in translation quality are reflected in better translations of orthographically similar words and rare words. While evaluating the importance of source texts in driving model predictions, we highlight word-level patterns within ByT5, suggesting an ability to modulate word-level and character-level information during generation. We conclude by assessing the efficiency tradeoff of byte models, suggesting their usage in non-time-critical scenarios to boost translation quality.

Following multiple instructions is a crucial ability for large language models (LLMs). Evaluating this ability comes with significant challenges: (i) limited coherence between multiple instructions, (ii) positional bias where the order of instructions affects model performance, and (iii) a lack of objectively verifiable tasks. To address these issues, we introduce a benchmark designed to evaluate models’ abilities to follow multiple instructions through sequential instruction following (SIFo) tasks. In SIFo, the successful completion of multiple instructions is verifiable by examining only the final instruction. Our benchmark evaluates instruction following using four tasks (text modification, question answering, mathematics, and security rule following), each assessing different aspects of sequential instruction following. Our evaluation of popular LLMs, both closed-source and open-source, shows that more recent and larger models significantly outperform their older and smaller counterparts on the SIFo tasks, validating the benchmark’s effectiveness. All models struggle with following sequences of instructions, hinting at an important lack of robustness of today’s language models.

Recent advances in computational linguistics include simulating the emergence of human-like languages with interacting neural network agents, starting from sets of random symbols. The recently introduced NeLLCom framework (Lian et al., 2023) allows agents to first learn an artificial language and then use it to communicate, with the aim of studying the emergence of specific linguistics properties. We extend this framework (NeLLCom-X) by introducing more realistic role-alternating agents and group communication in order to investigate the interplay between language learnability, communication pressures, and group size effects. We validate NeLLCom-X by replicating key findings from prior research simulating the emergence of a word-order/case-marking trade-off. Next, we investigate how interaction affects linguistic convergence and emergence of the trade-off. The novel framework facilitates future simulations of diverse linguistic aspects, emphasizing the importance of interaction and group dynamics in language evolution.

Interpretability research has shown that self-supervised Spoken LanguageModels (SLMs) encode a wide variety of features in human speech from theacoustic, phonetic, phonological, syntactic and semantic levels, to speakercharacteristics. The bulk of prior research on representations of phonologyhas focused on segmental features such as phonemes; the encoding ofsuprasegmental phonology (such as tone and stress patterns) in SLMs is not yetwell understood. Tone is a suprasegmental feature that is present in more thanhalf of the world’s languages. This paper aims to analyze the tone encodingcapabilities of SLMs, using Mandarin and Vietnamese as case studies. We showthat SLMs encode lexical tone to a significant degree even when they aretrained on data from non-tonal languages. We further find that SLMs behavesimilarly to native and non-native human participants in tone and consonantperception studies, but they do not follow the same developmental trajectory.

Natural languages show a tendency to minimize the linear distance between heads and their dependents in a sentence, known as dependency length minimization (DLM). Such a preference, however, has not been consistently replicated with neural agent simulations. Comparing the behavior of models with that of human learners can reveal which aspects affect the emergence of this phenomenon. In this work, we investigate the minimal conditions that may lead neural learners to develop a DLM preference. We add three factors to the standard neural-agent language learning and communication framework to make the simulation more realistic, namely: (i) the presence of noise during listening, (ii) context-sensitivity of word use through non-uniform conditional word distributions, and (iii) incremental sentence processing, or the extent to which an utterance’s meaning can be guessed before hearing it entirely. While no preference appears in production, we show that the proposed factors can contribute to a small but significant learning advantage of DLM for listeners of verb-initial languages.

Multilingual large-scale Pretrained Language Models (PLMs) have been shown to store considerable amounts of factual knowledge, but large variations are observed across languages. With the ultimate goal of ensuring that users with different language backgrounds obtain consistent feedback from the same model, we study the cross-lingual consistency (CLC) of factual knowledge in various multilingual PLMs. To this end, we propose a Ranking-based Consistency (RankC) metric to evaluate knowledge consistency across languages independently from accuracy. Using this metric, we conduct an in-depth analysis of the determining factors for CLC, both at model level and at language-pair level. Among other results, we find that increasing model size leads to higher factual probing accuracy in most languages, but does not improve cross-lingual consistency. Finally, we conduct a case study on CLC when new factual associations are inserted in the PLMs via model editing. Results on a small sample of facts inserted in English reveal a clear pattern whereby the new piece of knowledge transfers only to languages with which English has a high RankC score. All code and data are released at https://github.com/Betswish/Cross-Lingual-Consistency.

Artificial learners often behave differently from human learners in the context of neural agent-based simulations of language emergence and change. A common explanation is the lack of appropriate cognitive biases in these learners. However, it has also been proposed that more naturalistic settings of language learning and use could lead to more human-like results. We investigate this latter account, focusing on the word-order/case-marking trade-off, a widely attested language universal that has proven particularly hard to simulate. We propose a new Neural-agent Language Learning and Communication framework (NeLLCom) where pairs of speaking and listening agents first learn a miniature language via supervised learning, and then optimize it for communication via reinforcement learning. Following closely the setup of earlier human experiments, we succeed in replicating the trade-off with the new framework without hard-coding specific biases in the agents. We see this as an essential step towards the investigation of language universals with neural learners.

This work studies the plausibility of sequence-to-sequence neural networks as models of morphological acquisition by humans. We replicate the findings of Kirov and Cotterell (2018) on the well-known challenge of the English past tense and examine their generalizability to two related but morphologically richer languages, namely Dutch and German. Using a new dataset of English/Dutch/German (ir)regular verb forms, we show that the major findings of Kirov and Cotterell (2018) hold for all three languages, including the observation of over-regularization errors and micro U-shape learning trajectories. At the same time, we observe troublesome cases of non human-like errors similar to those reported by recent follow-up studies with different languages or neural architectures. Finally, we study the possibility of switching to orthographic input in the absence of pronunciation information and show this can have a non-negligible impact on the simulation results, with possibly misleading findings.

We introduce DivEMT, the first publicly available post-editing study of Neural Machine Translation (NMT) over a typologically diverse set of target languages. Using a strictly controlled setup, 18 professional translators were instructed to translate or post-edit the same set of English documents into Arabic, Dutch, Italian, Turkish, Ukrainian, and Vietnamese. During the process, their edits, keystrokes, editing times and pauses were recorded, enabling an in-depth, cross-lingual evaluation of NMT quality and post-editing effectiveness. Using this new dataset, we assess the impact of two state-of-the-art NMT systems, Google Translate and the multilingual mBART-50 model, on translation productivity. We find that post-editing is consistently faster than translation from scratch. However, the magnitude of productivity gains varies widely across systems and languages, highlighting major disparities in post-editing effectiveness for languages at different degrees of typological relatedness to English, even when controlling for system architecture and training data size. We publicly release the complete dataset including all collected behavioral data, to foster new research on the translation capabilities of NMT systems for typologically diverse languages.

Massively multilingual models are promising for transfer learning across tasks and languages. However, existing methods are unable to fully leverage training data when it is available in different task-language combinations. To exploit such heterogeneous supervision, we propose Hyper-X, a single hypernetwork that unifies multi-task and multilingual learning with efficient adaptation. It generates weights for adapter modules conditioned on both tasks and language embeddings. By learning to combine task and language-specific knowledge, our model enables zero-shot transfer for unseen languages and task-language combinations. Our experiments on a diverse set of languages demonstrate that Hyper-X achieves the best or competitive gain when a mixture of multiple resources is available, while on par with strong baseline in the standard scenario. Hyper-X is also considerably more efficient in terms of parameters and resources compared to methods that train separate adapters. Finally, Hyper-X consistently produces strong results in few-shot scenarios for new languages, showing the versatility of our approach beyond zero-shot transfer.

Recent advances in multilingual language modeling have brought the idea of a truly universal parser closer to reality. However, such models are still not immune to the “curse of multilinguality”: Cross-language interference and restrained model capacity remain major obstacles. To address this, we propose a novel language adaptation approach by introducing contextual language adapters to a multilingual parser. Contextual language adapters make it possible to learn adapters via language embeddings while sharing model parameters across languages based on contextual parameter generation. Moreover, our method allows for an easy but effective integration of existing linguistic typology features into the parsing model. Because not all typological features are available for every language, we further combine typological feature prediction with parsing in a multi-task model that achieves very competitive parsing performance without the need for an external prediction system for missing features. The resulting parser, UDapter, can be used for dependency parsing as well as sequence labeling tasks such as POS tagging, morphological tagging, and NER. In dependency parsing, it outperforms strong monolingual and multilingual baselines on the majority of both high-resource and low-resource (zero-shot) languages, showing the success of the proposed adaptation approach. In sequence labeling tasks, our parser surpasses the baseline on high resource languages, and performs very competitively in a zero-shot setting. Our in-depth analyses show that adapter generation via typological features of languages is key to this success.1

The scarcity of parallel data is a major limitation for Neural Machine Translation (NMT) systems, in particular for translation into morphologically rich languages (MRLs). An important way to overcome the lack of parallel data is to leverage target monolingual data, which is typically more abundant and easier to collect. We evaluate a number of techniques to achieve this, ranging from back-translation to random token masking, on the challenging task of translating English into four typologically diverse MRLs, under low-resource settings. Additionally, we introduce Inflection Pre-Training (or PT-Inflect), a novel pre-training objective whereby the NMT system is pre-trained on the task of re-inflecting lemmatized target sentences before being trained on standard source-to-target language translation. We conduct our evaluation on four typologically diverse target MRLs, and find that PT-Inflect surpasses NMT systems trained only on parallel data. While PT-Inflect is outperformed by back-translation overall, combining the two techniques leads to gains in some of the evaluated language pairs.

Neural machine translation (NMT) systems are nowadays essential components of professional translation workflows. Consequently, human translators are increasingly working as post-editors for machine-translated content. The NWO-funded InDeep project aims to empower users of Deep Learning models of text, speech, and music by improving their ability to interact with such models and interpret their behaviors. In the specific context of translation, we aim at developing new tools and methodologies to improve prediction attribution, error analysis, and controllable generation for NMT systems. These advances will be evaluated through field studies involving professional translators to assess gains in efficiency and overall enjoyability of the post-editing process.

Neural semantic parsers have obtained acceptable results in the context of parsing DRSs (Discourse Representation Structures). In particular models with character sequences as input showed remarkable performance for English. But how does this approach perform on languages with a different writing system, like Chinese, a language with a large vocabulary of characters? Does rule-based tokenisation of the input help, and which granularity is preferred: characters, or words? The results are promising. Even with DRSs based on English, good results for Chinese are obtained. Tokenisation offers a small advantage for English, but not for Chinese. Overall, characters are preferred as input, both for English and Chinese.

Identifying factors that make certain languages harder to model than others is essential to reach language equality in future Natural Language Processing technologies. Free-order case-marking languages, such as Russian, Latin, or Tamil, have proved more challenging than fixed-order languages for the tasks of syntactic parsing and subject-verb agreement prediction. In this work, we investigate whether this class of languages is also more difficult to translate by state-of-the-art Neural Machine Translation (NMT) models. Using a variety of synthetic languages and a newly introduced translation challenge set, we find that word order flexibility in the source language only leads to a very small loss of NMT quality, even though the core verb arguments become impossible to disambiguate in sentences without semantic cues. The latter issue is indeed solved by the addition of case marking. However, in medium- and low-resource settings, the overall NMT quality of fixed-order languages remains unmatched.

It is now established that modern neural language models can be successfully trained on multiple languages simultaneously without changes to the underlying architecture, providing an easy way to adapt a variety of NLP models to low-resource languages. But what kind of knowledge is really shared among languages within these models? Does multilingual training mostly lead to an alignment of the lexical representation spaces or does it also enable the sharing of purely grammatical knowledge? In this paper we dissect different forms of cross-lingual transfer and look for its most determining factors, using a variety of models and probing tasks. We find that exposing our LMs to a related language does not always increase grammatical knowledge in the target language, and that optimal conditions for lexical-semantic transfer may not be optimal for syntactic transfer.

We present an end-to-end neural approach to generate English sentences from formal meaning representations, Discourse Representation Structures (DRSs). We use a rather standard bi-LSTM sequence-to-sequence model, work with a linearized DRS input representation, and evaluate character-level and word-level decoders. We obtain very encouraging results in terms of reference-based automatic metrics such as BLEU. But because such metrics only evaluate the surface level of generated output, we develop a new metric, ROSE, that targets specific semantic phenomena. We do this with five DRS generation challenge sets focusing on tense, grammatical number, polarity, named entities and quantities. The aim of these challenge sets is to assess the neural generator’s systematicity and generalization to unseen inputs.

Dravidian languages, such as Kannada and Tamil, are notoriously difficult to translate by state-of-the-art neural models. This stems from the fact that these languages are morphologically very rich as well as being low-resourced. In this paper, we focus on subword segmentation and evaluate Linguistically Motivated Vocabulary Reduction (LMVR) against the more commonly used SentencePiece (SP) for the task of translating from English into four different Dravidian languages. Additionally we investigate the optimal subword vocabulary size for each language. We find that SP is the overall best choice for segmentation, and that larger dictionary sizes lead to higher translation quality.

We study the problem of domain adaptation in Neural Machine Translation (NMT) when domain-specific data cannot be shared due to confidentiality or copyright issues. As a first step, we propose to fragment data into phrase pairs and use a random sample to fine-tune a generic NMT model instead of the full sentences. Despite the loss of long segments for the sake of confidentiality protection, we find that NMT quality can considerably benefit from this adaptation, and that further gains can be obtained with a simple tagging technique.

Natural languages display a trade-off among different strategies to convey syntactic structure, such as word order or inflection. This trade-off, however, has not appeared in recent simulations of iterated language learning with neural network agents (Chaabouni et al., 2019b). We re-evaluate this result in light of three factors that play an important role in comparable experiments from the Language Evolution field: (i) speaker bias towards efficient messaging, (ii) non systematic input languages, and (iii) learning bottleneck. Our simulations show that neural agents mainly strive to maintain the utterance type distribution observed during learning, instead of developing a more efficient or systematic language.

This paper describes our submission for the English-Tamil news translation task of WMT-2020. The various techniques and Neural Machine Translation (NMT) models used by our team are presented and discussed, including back-translation, fine-tuning and word dropout. Additionally, our experiments show that using a linguistically motivated subword segmentation technique (Ataman et al., 2017) does not consistently outperform the more widely used, non-linguistically motivated SentencePiece algorithm (Kudo and Richardson, 2018), despite the agglutinative nature of Tamil morphology.

Recent advances in multilingual dependency parsing have brought the idea of a truly universal parser closer to reality. However, cross-language interference and restrained model capacity remain major obstacles. To address this, we propose a novel multilingual task adaptation approach based on contextual parameter generation and adapter modules. This approach enables to learn adapters via language embeddings while sharing model parameters across languages. It also allows for an easy but effective integration of existing linguistic typology features into the parsing network. The resulting parser, UDapter, outperforms strong monolingual and multilingual baselines on the majority of both high-resource and low-resource (zero-shot) languages, showing the success of the proposed adaptation approach. Our in-depth analyses show that soft parameter sharing via typological features is key to this success.

We investigate whether off-the-shelf deep bidirectional sentence representations (Devlin et al., 2019) trained on a massively multilingual corpus (multilingual BERT) enable the development of an unsupervised universal dependency parser. This approach only leverages a mix of monolingual corpora in many languages and does not require any translation data making it applicable to low-resource languages. In our experiments we outperform the best CoNLL 2018 language-specific systems in all of the shared task’s six truly low-resource languages while using a single system. However, we also find that (i) parsing accuracy still varies dramatically when changing the training languages and (ii) in some target languages zero-shot transfer fails under all tested conditions, raising concerns on the ‘universality’ of the whole approach.

Recent work has shown that neural models can be successfully trained on multiple languages simultaneously. We investigate whether such models learn to share and exploit common syntactic knowledge among the languages on which they are trained. This extended abstract presents our preliminary results.

Neural sequence-to-sequence models have proven very effective for machine translation, but at the expense of model interpretability. To shed more light into the role played by linguistic structure in the process of neural machine translation, we perform a fine-grained analysis of how various source-side morphological features are captured at different levels of the NMT encoder while varying the target language. Differently from previous work, we find no correlation between the accuracy of source morphology encoding and translation quality. We do find that morphological features are only captured in context and only to the extent that they are directly transferable to the target words.

Recent work has shown that recurrent neural networks (RNNs) can implicitly capture and exploit hierarchical information when trained to solve common natural language processing tasks (Blevins et al., 2018) such as language modeling (Linzen et al., 2016; Gulordava et al., 2018) and neural machine translation (Shi et al., 2016). In contrast, the ability to model structured data with non-recurrent neural networks has received little attention despite their success in many NLP tasks (Gehring et al., 2017; Vaswani et al., 2017). In this work, we compare the two architectures—recurrent versus non-recurrent—with respect to their ability to model hierarchical structure and find that recurrency is indeed important for this purpose. The code and data used in our experiments is available at https://github.com/ketranm/fan_vs_rnn

Intelligent selection of training data has proven a successful technique to simultaneously increase training efficiency and translation performance for phrase-based machine translation (PBMT). With the recent increase in popularity of neural machine translation (NMT), we explore in this paper to what extent and how NMT can also benefit from data selection. While state-of-the-art data selection (Axelrod et al., 2011) consistently performs well for PBMT, we show that gains are substantially lower for NMT. Next, we introduce ‘dynamic data selection’ for NMT, a method in which we vary the selected subset of training data between different training epochs. Our experiments show that the best results are achieved when applying a technique we call ‘gradual fine-tuning’, with improvements up to +2.6 BLEU over the original data selection approach and up to +3.1 BLEU over a general baseline.

Distributed word representations are widely used for modeling words in NLP tasks. Most of the existing models generate one representation per word and do not consider different meanings of a word. We present two approaches to learn multiple topic-sensitive representations per word by using Hierarchical Dirichlet Process. We observe that by modeling topics and integrating topic distributions for each document we obtain representations that are able to distinguish between different meanings of a given word. Our models yield statistically significant improvements for the lexical substitution task indicating that commonly used single word representations, even when combined with contextual information, are insufficient for this task.

The quality of a Neural Machine Translation system depends substantially on the availability of sizable parallel corpora. For low-resource language pairs this is not the case, resulting in poor translation quality. Inspired by work in computer vision, we propose a novel data augmentation approach that targets low-frequency words by generating new sentence pairs containing rare words in new, synthetically created contexts. Experimental results on simulated low-resource settings show that our method improves translation quality by up to 2.9 BLEU points over the baseline and up to 3.2 BLEU over back-translation.

A major challenge for statistical machine translation (SMT) of Arabic-to-English user-generated text is the prevalence of text written in Arabizi, or Romanized Arabic. When facing such texts, a translation system trained on conventional Arabic-English data will suffer from extremely low model coverage. In addition, Arabizi is not regulated by any official standardization and therefore highly ambiguous, which prevents rule-based approaches from achieving good translation results. In this paper, we improve Arabizi-to-English machine translation by presenting a simple but effective Arabizi-to-Arabic transliteration pipeline that does not require knowledge by experts or native Arabic speakers. We incorporate this pipeline into a phrase-based SMT system, and show that translation quality after automatically transliterating Arabizi to Arabic yields results that are comparable to those achieved after human transliteration.

Research in statistical machine translation (SMT) is largely driven by formal translation tasks, while translating informal text is much more challenging. In this paper we focus on SMT for the informal genre of dialogues, which has rarely been addressed to date. Concretely, we investigate the effect of dialogue acts, speakers, gender, and text register on SMT quality when translating fictional dialogues. We first create and release a corpus of multilingual movie dialogues annotated with these four dialogue-specific aspects. When measuring translation performance for each of these variables, we find that BLEU fluctuations between their categories are often significantly larger than randomly expected. Following this finding, we hypothesize and show that SMT of fictional dialogues benefits from adaptation towards dialogue acts and registers. Finally, we find that male speakers are harder to translate and use more vulgar language than female speakers, and that vulgarity is often not preserved during translation.

Defining the reordering search space is a crucial issue in phrase-based SMT between distant languages. In fact, the optimal trade-off between accuracy and complexity of decoding is nowadays reached by harshly limiting the input permutation space. We propose a method to dynamically shape such space and, thus, capture long-range word movements without hurting translation quality nor decoding time. The space defined by loose reordering constraints is dynamically pruned through a binary classifier that predicts whether a given input word should be translated right after another. The integration of this model into a phrase-based decoder improves a strong Arabic-English baseline already including state-of-the-art early distortion cost (Moore and Quirk, 2007) and hierarchical phrase orientation models (Galley and Manning, 2008). Significant improvements in the reordering of verbs are achieved by a system that is notably faster than the baseline, while bleu and meteor remain stable, or even increase, at a very high distortion limit.

This paper compares techniques to combine diverse parallel corpora for domain-specific phrase-based SMT system training. We address a common scenario where little in-domain data is available for the task, but where large background models exist for the same language pair. In particular, we focus on phrase table fill-up: a method that effectively exploits background knowledge to improve model coverage, while preserving the more reliable information coming from the in-domain corpus. We present experiments on an emerging transcribed speech translation task – the TED talks. While performing similarly in terms of BLEU and NIST scores to the popular log-linear and linear interpolation techniques, filled-up translation models are more compact and easy to tune by minimum error training.

This year FBK took part in the BTEC translation task, with source languages Arabic and Turkish and target language English, and in the new TALK task, source English and target French. We worked in the framework of phrase-based statistical machine translation aiming to improve coverage of models in presence of rich morphology, on one side, and to make better use of available resources through data selection techniques. New morphological segmentation rules were developed for Turkish-English. The combination of several Turkish segmentation schemes into a lattice input led to an improvement wrt to last year. The use of additional training data was explored for Arabic-English, while on the English to French task improvement was achieved over a strong baseline by automatically selecting relevant and high quality data from the available training corpora.

This paper reports on the participation of FBK at the IWSLT 2009 Evaluation. This year we worked on the Arabic-English and Turkish-English BTEC tasks with a special effort on linguistic preprocessing techniques involving morphological segmentation. In addition, we investigated the adaptation problem in the development of systems for the Chinese-English and English-Chinese challenge tasks; in particular, we explored different ways for clustering training data into topic or dialog-specific subsets: by producing (and combining) smaller but more focused models, we intended to make better use of the available training data, with the ultimate purpose of improving translation quality.

We tried to cope with the complex morphology of Turkish by applying different schemes of morphological word segmentation to the training and test data of a phrase-based statistical machine translation system. These techniques allow for a considerable reduction of the training dictionary, and lower the out-of-vocabulary rate of the test set. By minimizing differences between lexical granularities of Turkish and English we can produce more refined alignments and a better modeling of the translation task. Morphological segmentation is highly language dependent and requires a fair amount of linguistic knowledge in its development phase. Yet it is fast and light-weight – does not involve syntax – and appears to benefit our IWSLT09 system: our best segmentation scheme associated to a simple lexical approximation technique achieved a 50% reduction of out-of-vocabulary rate and over 5 point BLEU improvement above the baseline.