Papers by Rico Sennrich

Dia-Lingle: A Gamified Interface for Dialectal Data Collection (2025.acl-demo)

Challenge:	Using a prototype, we present an automatic speech translation system for live subtitling of conference speech . the system is routinely tested in recognizing English, Czech, and German speech - and presenting it simultaneously into 42 target languages.
Approach:	They propose an automatic speech translation system aimed at live subtitling of conference presentations.
Outcome:	The proposed system is a working prototype that is routinely tested in recognizing English, Czech, and German speech and presenting it translated simultaneously into 42 target languages.

Copied to clipboard

Jiugeng Sun, Rita Sevastjanova, Sina Ahmadi, Rico Sennrich, Mennatallah El-Assady

Challenge:	Dialects suffer from the scarcity of textual resources and are largely spoken rather than written.
Approach:	They propose a gamified interface that combines active learning with gamification to enhance the dialect corpus.
Outcome:	The proposed interface demonstrates high levels of user satisfaction while requiring minimal effort.

Contrastive Conditioning for Assessing Disambiguation in MT: A Case Study of Distilled Bias (2021.emnlp-main)

Copied to clipboard

In Neural Machine Translation, What Does Transfer Learning Transfer? (2020.acl-main)

Challenge:	Lexical disambiguation is a major challenge for machine translation systems . previous work focused on automatic post-hoc analysis of translations, but rules of what makes a disambiguations correct or incorrect tend to be imprecise.
Approach:	They propose a black-box method that uses contrastive conditioning to detect disambiguation errors.
Outcome:	The proposed method is scalable and reliable for disambiguation evaluations.

Copied to clipboard

Alham Fikri Aji, Nikolay Bogoychev, Kenneth Heafield, Rico Sennrich

Parity-Aware Byte-Pair Encoding: Improving Cross-lingual Fairness in Tokenization (2026.acl-long)

Challenge:	a recent study found that word embeddings are not necessary for transfer learning.
Approach:	They perform several ablation studies that limit information transfer and measure the quality impact across three language pairs to gain a black-box understanding of transfer learning.
Outcome:	The proposed method can eliminate the need for a warm-up phase when training transformer models in high resource language pairs.

Copied to clipboard

Negar Foroutan, Clara Meister, Debjit Paul, Joel Niklaus, Sina Ahmadi, Antoine Bosselut, Rico Sennrich

Beyond Sentence-Level End-to-End Speech Translation: Context Helps (2021.acl-long)

Challenge:	Tokenization is the first step of most NLP pipelines.
Approach:	They propose a parity-aware byte pair encoder that maximizes the compression gain of the currently worst-compressed language for cross-lingual parity.
Outcome:	a new algorithm reduces tokenization inequality by 89% compared to classical BPE . the proposed algorithm is based on a fair-max rule that maximizes the compression gain of the currently worst-compressed language .

Copied to clipboard

Biao Zhang, Ivan Titov, Barry Haddow, Rico Sennrich

Evaluating Automatic Metrics with Incremental Machine Translation Systems (2024.findings-emnlp)

Challenge:	Document-level contextual information has shown benefits to text-based machine translation, but whether and how it helps end-to-end speech translation is still under-studied.
Approach:	They propose a concatenation-based ST model with adaptive feature selection for computational efficiency.
Outcome:	The proposed model improves translation quality and robustness to (artificial) audio segmentation errors.

Copied to clipboard

Guojun Wu, Shay Cohen, Rico Sennrich

The Mediomatix Corpus: Parallel Data for Romansh Language Varieties via Comparable Schoolbooks (2026.findings-eacl)

Challenge:	Existing studies have shown that neural metrics are more reliable than non-neural metrics.
Approach:	They propose to use commercial machine translations to evaluate machine translation metrics based on their preference for more recent outputs.
Outcome:	The proposed dataset confirms several previous findings, including the advantage of neural metrics over non-neural ones, and also explores the debated issue of how MT quality affects metric reliability.

Copied to clipboard

Zachary Hopton, Jannis Vamvas, Andrin Büchler, Anna Rutkiewicz, Rico Cathomas, Rico Sennrich

Improving Deep Transformer with Depth-Scaled Initialization and Merged Attention (D19-1)

Challenge:	a small-scale human evaluation confirms that the segments are highly parallel, making the dataset suitable for NLP applications.
Approach:	They present a first parallel corpus of Romansh idioms from 291 schoolbooks . they use automatic alignment methods to extract 207k multi-parallel segments from the books .
Outcome:	The proposed corpus is based on 291 schoolbook volumes, which are comparable in content for the five idioms.

Copied to clipboard

Biao Zhang, Ivan Titov, Rico Sennrich

Improving Machine Translation of Educational Content via Crowdsourcing (L18-1)

Challenge:	Existing methods to improve NLP convergence and computational overhead are limited by stacking more layers.
Approach:	They propose a depth-scaled initialization method which reduces parameter variance at initialization and reduces output variance of residual connections to ease gradient back-propagation.
Outcome:	The proposed method outperforms the base model on translation tasks with five translation directions while matching the decoding speed of the baseline model.

Copied to clipboard

Maximiliana Behnke, Antonio Valerio Miceli Barone, Rico Sennrich, Vilelmini Sosoni, Thanasis Naskos, Eirini Takoulidou, Maria Stasimioti, Menno van Zaanen, Sheila Castilho, Federico Gaspari, Panayota Georgakopoulou, Valia Kordoni, Markus Egg, Katia Lida Kermanidis

Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned (P19-1)

Challenge:	Using crowdsourcing to train neural machine translation models is expensive and expensive . professional outsourcing of bilingual data is expensive if the translations are of a lower quality .
Approach:	They analyze the impact of crowdsourcing on the quality of in-domain training data . they use translations of MOOCs from English to eleven languages to fine-tune machine translation models .
Outcome:	The proposed method improves on general-domain training data and with pre-existing in-domain corpora.

Copied to clipboard

Elena Voita, David Talbot, Fedor Moiseev, Rico Sennrich, Ivan Titov

Evaluating Machine Translation Performance on Chinese Idioms with a Blacklist Method (L18-1)

Challenge:	et al., 2017) show that multi-head attention is important for neural machine translation.
Approach:	They evaluate the contribution made by individual attention heads to the overall performance of the Transformer model and analyze the roles played by them in the encoder.
Outcome:	The proposed pruning method removes the vast majority of heads without affecting performance.

Copied to clipboard

Yutong Shao, Rico Sennrich, Bonnie Webber, Federico Fancellu

Uncovering Hidden Consequences of Pre-training Objectives in Sequence-to-Sequence Models (2023.findings-acl)

Challenge:	idiom translation is a challenging problem in machine translation because meaning is non-compositional and literal translations are likely to be wrong.
Approach:	They propose a method to evaluate the quality of idiom translation of MT systems by a blacklist of literal translations.
Outcome:	The proposed method detects that a sizable number of idioms are mistranslated (46.1%) and that literal translation error is a common error type.

Copied to clipboard

Tannon Kew, Rico Sennrich

Improving Zero-Shot Cross-lingual Transfer Between Closely Related Languages by Injecting Character-Level Noise (2022.findings-acl)

Challenge:	Existing studies have found that some pre-training objectives have negligible impact on downstream performance, but they are not effective on all models.
Approach:	They propose a zero-shot control method that is only successful on a subset of models and experiment with different pre-training objectives.
Outcome:	The proposed method is only successful on a subset of models and unexpected interactions between the pre-training method and downstream controllability are found.

Copied to clipboard

Noëmi Aepli, Rico Sennrich

Responsible & Trustworthy NLP

Challenge:	Existing approaches to improve cross-lingual transfer do not take surface similarity into account.
Approach:	They propose to augment source language training data with character-level noise to simulate spelling variations.
Outcome:	The proposed strategy shows consistent improvements over several languages and tasks.

On Exposure Bias, Hallucination and Domain Shift in Neural Machine Translation (2020.acl-main)

Copied to clipboard

Chaojun Wang, Rico Sennrich

Self-training Reduces Flicker in Retranslation-based Simultaneous Translation (2023.eacl-main)

Challenge:	Neural machine translation suffers from exposure bias, and alternative approaches to mitigate this are under debate.
Approach:	They propose to reduce exposure bias by using minimum risk training to mitigate hallucinations . they find that exposure bias is more problematic under domain shift .
Outcome:	The proposed methods can reduce exposure bias even on in-domain test sets.

Copied to clipboard

Sukanta Sen, Rico Sennrich, Biao Zhang, Barry Haddow

Context-Aware Monolingual Repair for Neural Machine Translation (D19-1)

Challenge:	Existing approaches to reduce flicker in simultaneous translation have increased the latency through masking and specialised inference, thus losing the simplicity of the approach.
Approach:	They propose to train a machine translation system to reduce flicker by controlling monotonicity and biased beam search to achieve the same flicker-latency tradeoff.
Outcome:	The proposed approach reduces flicker by controlling monotonicity while maintaining similar translation quality to the original.

Copied to clipboard

NusaX: Multilingual Parallel Sentiment Dataset for 10 Indonesian Local Languages (2023.eacl-main)

Challenge:	et al., 2018) show that human raters prefer corrected translations over the baseline ones.
Approach:	They propose a monolingual model to correct inconsistencies between sentences . they use monolingual document-level data to train the model .
Outcome:	The proposed model improves translations of contextual phenomena in English-Russian translation task.

Copied to clipboard

Genta Indra Winata, Alham Fikri Aji, Samuel Cahyawijaya, Rahmad Mahendra, Fajri Koto, Ade Romadhony, Kemal Kurniawan, David Moeljadi, Radityo Eko Prasojo, Pascale Fung, Timothy Baldwin, Jey Han Lau, Rico Sennrich, Sebastian Ruder

Language Modeling, Lexical Translation, Reordering: The Training Process of NMT through the Lens of Classical SMT (2021.emnlp-main)

Challenge:	In Indonesia, many languages are endangered and some are even extinct due to the unavailability of data resources and benchmarks.
Approach:	They propose a high-quality multilingual parallel corpus that covers 10 local languages from Indonesia.
Outcome:	The proposed resource includes sentiment and machine translation datasets, and bilingual lexicons.

Copied to clipboard

As Little as Possible, as Much as Necessary: Detecting Over- and Undertranslations with Contrastive Conditioning (2022.acl-short)

Challenge:	Statistical MT decomposes the translation task into distinct components that are learned separately.
Approach:	They show that neural machine translation models acquire different competences over the course of training . previous work shows how to improve some of the competences in NMT by using lexical translation probabilities, phrase memories, alignment information.
Outcome:	The proposed model improves translation quality and word-by-word translation, while learning complex reordering patterns.

Copied to clipboard

Sentence Compression for Arbitrary Languages via Multilingual Pivoting (D18-1)

Challenge:	Neural machine translation is susceptible to coverage errors such as the addition of superfluous target words or the omission of important source content.
Approach:	They propose a method for detecting Omission and addition errors with off-the-shelf translation models.
Outcome:	The proposed method is comparable to a supervised method that requires a custom quality estimation model.

Copied to clipboard

Jonathan Mallinson, Rico Sennrich, Mirella Lapata

When a Good Translation is Wrong in Context: Context-Aware Machine Translation Improves on Deixis, Ellipsis, and Lexical Cohesion (P19-1)

Challenge:	a new study advocates the use of bilingual corpora for sentence compression models . previous work focused on word deletion, while others view sentence compression as a general text rewriting problem.
Approach:	They propose to use bilingual corpora for training sentence compression models.
Outcome:	The proposed model can be trained for any language as long as a bilingual corpus is available . it performs arbitrary rewrites without access to compression specific data .

Copied to clipboard

What’s the Meaning of Superhuman Performance in Today’s NLU? (2023.acl-long)

Challenge:	et al., 2018: translation errors due to the lack of extra-sentential context are becoming more and more noticeable among otherwise adequate translations.
Approach:	They propose a context-aware translation model that uses sentence-level data to identify inconsistencies . standard metrics are not sensitive to improvements in consistency in document-level translations .
Outcome:	The proposed model shows major gains over baseline without sacrificing performance . standard metrics are not sensitive to improvements in document-level translations .

Copied to clipboard

Simone Tedeschi, Johan Bos, Thierry Declerck, Jan Hajič, Daniel Hershcovich, Eduard Hovy, Alexander Koller, Simon Krek, Steven Schockaert, Rico Sennrich, Ekaterina Shutova, Roberto Navigli

ConLoan: A Contrastive Multilingual Dataset for Evaluating Loanwords (2025.acl-long)

Challenge:	Recent research has focused on developing larger pretrained language models and introducing benchmarks such as SuperGLUE and SQuAD to measure their abilities.
Approach:	They propose to use benchmarks such as SuperGLUE and SQUAD to evaluate PLMs' abilities in language understanding, reasoning, and reading comprehension to assess their performance.
Outcome:	The proposed benchmarks have serious limitations affecting comparison between humans and PLMs and provide recommendations for fairer and more transparent benchmarks.

Copied to clipboard

Sina Ahmadi, Micha David Hess, Elena Álvarez-Mellado, Alessia Battisti, Cui Ding, Anne Göhring, Yingqiang Gao, Zifan Jiang, Andrianos Michail, Peshmerge Morad, Joel Niklaus, Maria Christina Panagiotopoulou, Stefano Perrella, Juri Opitz, Anastassia Shaitarova, Rico Sennrich

NMTScore: A Multilingual Analysis of Translation-based Text Similarity Measures (2022.findings-emnlp)

Challenge:	Lexical borrowing is a ubiquitous linguistic phenomenon influenced by geopolitical, societal, and technological factors.
Approach:	They propose a novel contrastive dataset comprising sentences with and without loanwords across 10 languages to examine how machine translation and language models process loanword .
Outcome:	The proposed dataset shows that state-of-the-art models prefer loanwords over native terms and exhibit varying performance across languages.

Copied to clipboard

On Romanization for Model Transfer Between Scripts in Neural Machine Translation (2020.findings-emnlp)

Challenge:	Neural machine translation (NMT) enables several similarity measures to estimate the probability of translations.
Approach:	They propose to rank the similarity of short text segments using translation-based similarity measures . they use the NMTScore library to analyze translation-related similarity .
Outcome:	The proposed measures show a relatively high correlation to human judgments when used for reference-based evaluation of data-to-text generation in 2 tasks and 17 languages.

Copied to clipboard

Chantal Amrhein, Rico Sennrich

Analyzing the Source and Target Contributions to Predictions in Neural Machine Translation (2021.acl-long)

Challenge:	Using romanization to improve low-resource machine translation is not always the best strategy.
Approach:	They propose to use romanization to improve transfer between languages with different scripts . they compare two romanization tools and find that they exhibit different degrees of information loss, which affects translation quality.
Outcome:	The proposed method improves transfer between languages with different scripts while entails information loss.

Copied to clipboard

Context-Aware Neural Machine Translation Learns Anaphora Resolution (P18-1)

Challenge:	Neural Machine Translation models are influenced by two types of context, source and target, but none explicitly evaluates relative contribution to generation decision.
Approach:	They propose to adopt a variant of Layerwise Relevance Propagation which evaluates relative contributions to the generation decision by a proportion of token influence.
Outcome:	The proposed model can evaluate the relative contribution of source and target to the generation decision by using a variant of Layerwise Relevance Propagation (LRP)

Copied to clipboard

Elena Voita, Pavel Serdyukov, Rico Sennrich, Ivan Titov

Information Extraction & Text Mining

Challenge:	Standard machine translation systems process sentences in isolation and ignore extra-sentential information.
Approach:	They propose a context-aware neural machine translation model that controls flow of information from extended context to the translation model.
Outcome:	The proposed model improves on an English-Russian subtitles dataset over its context-agnostic version (+0.7) and over simple concatenation of context and source sentences (+0.6).

Zero-Shot Crosslingual Sentence Simplification (2020.emnlp-main)

Copied to clipboard

Jonathan Mallinson, Rico Sennrich, Mirella Lapata

On Biasing Transformer Attention Towards Monotonicity (2021.naacl-main)

Challenge:	Recent approaches to simplification have shown promising results with encoder-decoder models trained on large amounts of parallel data which often only exists in English.
Approach:	They propose a model which transfers simplification knowledge from English to another language while generalizing across languages and tasks.
Outcome:	Empirical results show that the proposed model performs better than unsupervised and pivot-based methods.

Copied to clipboard

Annette Rios, Chantal Amrhein, Noëmi Aepli, Rico Sennrich

Encoders Help You Disambiguate Word Senses in Neural Machine Translation (D19-1)

Challenge:	Existing work has focused on learning monotonic attention behavior via specialized attention functions or pretraining.
Approach:	They introduce a monotonicity loss function compatible with standard attention mechanisms and test it on sequence-to-sequence tasks.
Outcome:	The proposed monotonicity loss function can achieve largely monotonic behavior on grapheme-to-phoneme conversion, morphological inflection, transliteration, and dialect normalization tasks.

Copied to clipboard

Gongbo Tang, Rico Sennrich, Joakim Nivre

Identifying Weaknesses in Machine Translation Metrics Through Minimum Bayes Risk Decoding: A Case Study for COMET (2022.aacl-main)

Challenge:	Neural machine translation models can perform word sense disambiguation (WSD) however, it is unclear which component dominates the process of disambiguating words.
Approach:	They evaluate hidden states and investigate distributions of self-attention in NMT encoders and decoders to disambiguate word senses.
Outcome:	The proposed model outperforms encoder hidden states on large datasets . the model outpersforms decoders on large data sets .

Copied to clipboard

Chantal Amrhein, Rico Sennrich

Efficient CTC Regularization via Coarse Labels for End-to-End Speech Translation (2023.eacl-main)

Challenge:	Neural metrics have a high correlation with human judgements but they are hard to eliminate due to their "black box" nature.
Approach:	They propose to use minimum bayes risk decoding to explore and quantify weaknesses in COMET models.
Outcome:	The proposed model is not sensitive enough to discrepancies in numbers and named entities, and is hard to remove by training on additional synthetic data.

Copied to clipboard

Biao Zhang, Barry Haddow, Rico Sennrich

Speech & Audio in NLP

Challenge:	Developing techniques to support end-to-end speech translation is non-trivial because of the speech-text modality gap.
Approach:	They propose a coarse labeling approach that merges vocabulary labels via simple heuristic rules . they propose to use 256-bit truncation, division or modulo operations to regularize the encoder .
Outcome:	The proposed method can increase training efficiency while delivering better performance.

A Lightweight Recurrent Network for Sequence Modeling (P19-1)

Copied to clipboard

Biao Zhang, Rico Sennrich

Green & Sustainable NLP

Challenge:	Recent studies show that recurrent networks suffer from severe computational inefficiency due to weak parallelization.
Approach:	They propose a lightweight recurrent network (LRN) that uses input and forget gates to handle long-range dependencies and gradient vanishing and explosion.
Outcome:	The proposed recurrent network yields the best running efficiency on six NLP tasks.

Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decoding (2024.eacl-short)

Copied to clipboard

Rico Sennrich, Jannis Vamvas, Alireza Mohammadshahi

Adaptive Feature Selection for End-to-End Speech Translation (2020.findings-emnlp)

Challenge:	Hallucinations and off-target translations remain unsolved problems in machine translation, especially for low-resource languages and massively multilingual models.
Approach:	They propose two methods to mitigate hallucinations and off-target translations with a modified decoding objective without retraining or external models.
Outcome:	The proposed methods reduce translation errors with segment-level chrF2 below 10 by 67-83% on average across 57 tested translation directions.

Copied to clipboard

Biao Zhang, Ivan Titov, Barry Haddow, Rico Sennrich

Understanding Pure Character-Based Neural Machine Translation: The Case of Translating Finnish into English (2020.coling-main)

Challenge:	End-to-end speech translation (E2E) models that directly maps audio to a foreign text are not efficient.
Approach:	They propose adaptive feature selection (AFS) for encoder-decoder based E2E ST.
Outcome:	The proposed model outperforms the existing model on LibriSpeech En-Fr and MuST-C with a BLEU score of 18.56.

Copied to clipboard

Gongbo Tang, Rico Sennrich, Joakim Nivre

Leveraging In-Context Learning for Political Bias Testing of LLMs (2025.acl-long)

Challenge:	Recent work shows that deeper character-based neural machine translation models outperform subword-based models.
Approach:	They propose to investigate the ability of character-based models to learn word senses and morphological inflections and the attention mechanism in Finnish into English translation.
Outcome:	The character-based models outperform subword-based model in Finnish to English translation.

Copied to clipboard

Patrick Haller, Jannis Vamvas, Rico Sennrich, Lena Ann Jäger

Responsible & Trustworthy NLP

Challenge:	Existing probing methods for evaluating LLMs with political questions have limited stability and are unreliable.
Approach:	They propose to use human survey data as in-context examples to query LLMs with political questions to evaluate their potential biases.
Outcome:	The proposed task improves the stability of question-based bias evaluation and may be used to compare instruction-tuned models to their base versions.

Robust Native Language Identification through Agentic Decomposition (2025.emnlp-main)

Copied to clipboard

Ahmet Yavuz Uluslu, Tannon Kew, Tilia Ellendorff, Gerold Schneider, Rico Sennrich

Information Retrieval

Challenge:	Large language models (LLMs) often achieve high performance by leveraging superficial contextual clues rather than the underlying linguistic patterns indicative of native language (L1) influence.
Approach:	They propose an agentic NLI pipeline where specialized agents accumulate and categorize diverse linguistic evidence before an independent final assessment.
Outcome:	The proposed pipeline significantly improves robustness against misleading contextual clues and performance consistency compared to standard prompting methods.

Turning English-centric LLMs Into Polyglots: How Much Multilinguality Is Needed? (2024.findings-emnlp)

Copied to clipboard

Tannon Kew, Florian Schottmann, Rico Sennrich

BlonDe: An Automatic Evaluation Metric for Document-level Machine Translation (2022.naacl-main)

Challenge:	Existing models that target a single language are not seen during finetuning, but are able to respond in multiple languages once deployed in downstream applications.
Approach:	They investigate the minimal amount of multilinguality required during finetuning to elicit effective cross-lingual generalisation in English-centric LLMs.
Outcome:	The proposed model can respond in as few as two to three languages to a user's query in English, but the degree to which a target language is seen during pretraining is limiting.

Copied to clipboard

Yuchen Jiang, Tianyu Liu, Shuming Ma, Dongdong Zhang, Jian Yang, Haoyang Huang, Rico Sennrich, Ryan Cotterell, Mrinmaya Sachan, Ming Zhou

SwissSLi: The Multi-parallel Sign Language Corpus for Switzerland (2024.lrec-main)

Challenge:	Standard evaluation metrics, e.g., BLEU, TER and METEOR, focus on the quality of translations at the sentence level and do not consider discourse-level features.
Approach:	They propose to use a metric to take discourse coherence into consideration by categorizing discourse-related spans and calculating the similarity-based F1 measure of categorized spans.
Outcome:	The proposed metric possesses better selectivity and interpretability at the document-level, and is more sensitive to document- level nuances.

Copied to clipboard

Zifan Jiang, Anne Göhring, Amit Moryossef, Rico Sennrich, Sarah Ebling

Multimodality

Challenge:	Using a CC BY-NC-SA 4.0 license, this corpus contains parallel sign language videos and spoken language subtitles.
Approach:	They introduce SwissSLi, the first sign language corpus that contains parallel data of all three Swiss sign languages.
Outcome:	The proposed corpus contains parallel sign language videos and spoken language subtitles.

On Sparsifying Encoder Outputs in Sequence-to-Sequence Models (2021.findings-acl)

Copied to clipboard

Biao Zhang, Ivan Titov, Rico Sennrich

SignCLIP: Connecting Text and Sign Language by Contrastive Learning (2024.emnlp-main)

Challenge:	Using sequence-to-sequence models, encoder outputs are usually transferred to the decoder for generation, but in this study, encoded outputs can be compressed to shorten the sequence for decoding.
Approach:	They propose to use a stochastic gate-based algorithm to mask encoder outputs to shorten the sequence delivered for decoding.
Outcome:	The proposed model can be used to shorten encoder outputs to short a sequence . the proposed model yields a speedup of up to 1.65 on document summarization and 1.20 on character-based machine translation tasks.

Copied to clipboard

Zifan Jiang, Gerard Sant, Amit Moryossef, Mathias Müller, Rico Sennrich, Sarah Ebling

PARME: Parallel Corpora for Low-Resourced Middle Eastern Languages (2025.acl-long)

Challenge:	SignCLIP is an efficient method of learning useful visual representations for sign language processing from large-scale, multilingual video-text pairs without optimizing for a specific task or sign language of limited size.
Approach:	They propose a method for learning visual representations for sign language processing from large-scale video-text pairs without directly optimizing for a specific task or sign language.
Outcome:	The proposed model can learn from multilingual video-text pairs without optimizing for a specific task or sign language of limited size.

Copied to clipboard

Sina Ahmadi, Rico Sennrich, Erfan Karami, Ako Marani, Parviz Fekrazad, Gholamreza Akbarzadeh Baghban, Hanah Hadi, Semko Heidari, Mahîr Dogan, Pedram Asadi, Dashne Bashir, Mohammad Amin Ghodrati, Kourosh Amini, Zeynab Ashourinezhad, Mana Baladi, Farshid Ezzati, Alireza Ghasemifar, Daryoush Hosseinpour, Behrooz Abbaszadeh, Amin Hassanpour, Bahaddin Jalal Hamaamin, Saya Kamal Hama, Ardeshir Mousavi, Sarko Nazir Hussein, Isar Nejadgholi, Mehmet Ölmez, Horam Osmanpour, Rashid Roshan Ramezani, Aryan Sediq Aziz, Ali Salehi, Mohammadreza Yadegari, Kewyar Yadegari, Sedighe Zamani Roodsari

Improving Massively Multilingual Neural Machine Translation and Zero-Shot Translation (2020.acl-main)

Challenge:	UNESCO has identified 60 varieties of Middle Eastern languages as underrepresented . a limited availability of language technology perpetuates a cycle of digital exclusion .
Approach:	They develop a parallel corpora for eight severely under-resourced varieties in the region . they evaluate machine translation capabilities through zero-shot approaches and fine-tuning experiments .
Outcome:	The proposed model aims to improve the processing of the eight under-resourced languages in the Middle East.

Copied to clipboard

Biao Zhang, Philip Williams, Ivan Titov, Rico Sennrich

Examining Multilingual Embedding Models Cross-Lingually Through LLM-Generated Adversarial Examples (2025.findings-emnlp)

Challenge:	Existing approaches to improve multilingual neural machine translation (NMT) are weak, and lack robustness to support language pairs with varying typological characteristics.
Approach:	They propose to deepen NMT models to support language pairs with varying typological characteristics by random online backtranslation.
Outcome:	The proposed approach narrows the performance gap with bilingual models and improves zero-shot performance by 10 BLEU, approaching conventional pivot-based methods.

Copied to clipboard

Andrianos Michail, Simon Clematide, Rico Sennrich

Measuring the Effect of Disfluency in Multilingual Knowledge Probing Benchmarks (2025.emnlp-main)

Challenge:	Cross-Lingual Semantic Discrimination (CLSD) is a lightweight evaluation task that requires only parallel sentences and a Large Language Model (LLM) to generate adversarial distractors.
Approach:	They propose a lightweight task that requires only parallel sentences and a Large Language Model (LLM) to generate adversarial distractors.
Outcome:	The proposed task requires only parallel sentences and a Large Language Model (LLM) to generate adversarial distractors.

Copied to clipboard

Kirill Semenov, Rico Sennrich

Revisiting Negation in Neural Machine Translation (2021.tacl-1)

Challenge:	Several datasets have been created for factual knowledge assessment of LLMs . MLAMA uses templated sentences that do not take into account grammatical and semantic information of named entities .
Approach:	They use MLAMA to prompt LLMs for factual knowledge using template translations . this leads to numerous instances of ungrammaticality or wrong wording of the final prompts - authors say .
Outcome:	The results show that the MLAMA dataset has higher knowledge retrieval scores than previous versions.

Copied to clipboard

Gongbo Tang, Philipp Rönchen, Rico Sennrich, Joakim Nivre

The Bottom-up Evolution of Representations in the Transformer: A Study with Machine Translation and Language Modeling Objectives (D19-1)

Challenge:	Negation is an important linguistic phenomenon in machine translation, as errors in translating negation may change the meaning of source sentences completely.
Approach:	They evaluate the translation of negation in English–German (EN–DE) and English– Chinese (EN-ZH) . they find that NMT models can distinguish negation and non-negation tokens very well and encode a lot of information about negation .
Outcome:	The accuracy of manual evaluation in ENDE, DEEN, ENZH, and ZHEN is 95.7%, 94.8%, 93.4%, and 91.7% respectively.

Copied to clipboard

Sparse Attention with Linear Units (2021.emnlp-main)

Challenge:	a recent study has shown that deep neural networks are effective with various tasks . a new study examines how representations of tokens evolve between layers under different learning objectives .
Approach:	They use canonical correlation analysis and mutual information estimators to study how information flows across Transformer layers.
Outcome:	The proposed model outperforms untrained models on word identity prediction tasks . the model outpersforms models trained on other linguistic tasks based on the model's objective .

Copied to clipboard

Biao Zhang, Ivan Titov, Rico Sennrich

Has Machine Translation Achieved Human Parity? A Case for Document-level Evaluation (D18-1)

Challenge:	Recent studies have suggested that sparse attention mechanisms can be made more interpretable by replacing the softmax activation with its sparser variants.
Approach:	They propose a method to replace softmax activation with a ReLU to achieve sparsity in attention by layer normalization with either a specialized initialization or an additional gating function.
Outcome:	The proposed model is easy to implement and more efficient than previously proposed sparse attention mechanisms.

Copied to clipboard

Samuel Läubli, Rico Sennrich, Martin Volk

Cheap Character Noise for OCR-Robust Multilingual Embeddings (2025.findings-acl)

Challenge:	Recent research suggests that neural machine translation achieves parity with professional human translation on the WMT Chinese–English news translation task.
Approach:	They empirically test neural machine translation on a Chinese–English news translation task . they show human raters prefer human over machine translation when evaluating documents .
Outcome:	The proposed method shows that human translators prefer document-level evaluation over machine translation . the results highlight the need to shift towards document- level evaluation as machine translation improves .

Copied to clipboard

Andrianos Michail, Juri Opitz, Yining Wang, Robin Meister, Rico Sennrich, Simon Clematide

Machine Translation Models are Zero-Shot Detectors of Translation Direction (2025.findings-acl)

Challenge:	Optical character recognition (OCR) is a key component of the digitization of historical documents.
Approach:	They propose a method that fine-tunes existing multilingual models using noisy texts and a contrastive loss.
Outcome:	The proposed model improves on the training data of existing models using noisy texts and a contrastive loss.

Copied to clipboard

Michelle Wastl, Jannis Vamvas, Rico Sennrich

How Suitable Are Subword Segmentation Strategies for Translating Non-Concatenative Morphology? (2021.findings-emnlp)

Challenge:	Existing approaches to detect the translation direction of parallel text are lacking in the machine translation community.
Approach:	They propose an unsupervised approach to detection of translation direction of parallel texts . they use a simple hypothesis that p(translation\|original)>p(original\|translation) they confirm the approach is effective for high-resource language pairs .
Outcome:	The proposed approach achieves document-level accuracies of 82–96% for NMT-produced translations and 60–81% for human translations, based on the model used.

Copied to clipboard

Chantal Amrhein, Rico Sennrich

Syntactic Text Processing

Challenge:	Data-driven subword segmentation is the default strategy for open-vocabulary machine translation but may not be sufficiently generic for learning non-concatenative morphology.
Approach:	They propose to test data-driven subword segmentation on non-concatenative morphological phenomena in a controlled, semi-synthetic setting.
Outcome:	The proposed model can translate non-concatenative morphological phenomena in a controlled, semi-synthetic setting.

Understanding the Properties of Minimum Bayes Risk Decoding in Neural Machine Translation (2021.acl-long)

Copied to clipboard

Mathias Müller, Rico Sennrich

Robustness in NLP

Challenge:	Neural Machine Translation (NMT) currently exhibits biases such as producing translations that are too short and overgenerating frequent words.
Approach:	They propose to use minimum bayes risk decoding instead of beam search to investigate the effects of beam decoding on unbiased samples.
Outcome:	The proposed method improves on a number of previously reported biases and failure cases of beam search on unbiased samples.

Revisiting Low-Resource Neural Machine Translation: A Case Study (P19-1)

Copied to clipboard

Rico Sennrich, Biao Zhang

Exploiting Biased Models to De-bias Text: A Gender-Fair Rewriting Model (2023.acl-long)

Challenge:	Recent research has shown that neural machine translation models are highly data-inefficient and underperform phrase-based statistical machine translation (PBSMT) in low-resource settings.
Approach:	They propose to use auxiliary data to train low-resource neural machine translation systems without auxiliary monolingual or multilingual data.
Outcome:	The proposed methods outperform PBSMT and other statistical machine translation models in Korean–English with minimal data.

Copied to clipboard

Chantal Amrhein, Florian Schottmann, Rico Sennrich, Samuel Läubli

Towards Unsupervised Recognition of Token-level Semantic Differences in Related Documents (2023.emnlp-main)

Challenge:	Existing work has explored using sequence-to-sequence rewriting models to transform biased outputs into more gender-fair language by creating pseudo training data through linguistic rules.
Approach:	They propose to use machine translation models to create gender-biased text from real gender-fair text via round-trip translation to eliminate rule-based data creation.
Outcome:	The proposed approach matches the performance of state-of-the-art rewriting models for English.

Copied to clipboard

Responsible & Trustworthy NLP

Challenge:	Existing studies on word-level predictions and highlighting semantic differences in natural language documents did not focus on semantic differences as the main target.
Approach:	They propose to perform a token-level regression task to highlight semantic differences between two documents . they use word alignment and sentence-level contrastive learning to evaluate the approaches .
Outcome:	The proposed approach has a robust correlation to gold labels, but all unsupervised approaches leave a margin of improvement.

Vision Matters When It Should: Sanity Checking Multimodal Machine Translation Models (2021.emnlp-main)

Copied to clipboard

Jiaoda Li, Duygu Ataman, Rico Sennrich

Multimodality

Challenge:	Multimodal machine translation models outperform text-only models when visual context is available, but recent studies have shown that the performance of MMT models is only marginally impacted when the associated image is replaced with an unrelated image or noise.
Approach:	They propose to use visual data to highlight the importance of visual inputs in MMT models to enhance their leverage.
Outcome:	The proposed models outperform text-only models when visual context is available, but the results show that the visual context might not be exploited by the models at all.

Why Self-Attention? A Targeted Evaluation of Neural Machine Translation Architectures (D18-1)

Copied to clipboard

Gongbo Tang, Mathias Müller, Annette Rios, Rico Sennrich

SwissGov-RSD: A Human-annotated, Cross-lingual Benchmark for Token-level Recognition of Semantic Differences Between Related Documents (2026.acl-long)

Challenge:	Recent studies show that non-recurrent architectures outperform RNNs in neural machine translation.
Approach:	They hypothesize that CNNs and self-attentional networks could extract semantic features from source text.
Outcome:	The proposed architectures outperform RNNs on two tasks: subject-verb agreement and word sense disambiguation.

Copied to clipboard

Michelle Wastl, Jannis Vamvas, Rico Sennrich

Source-primed Multi-turn Conversation Helps Large Language Models Translate Documents (2025.findings-emnlp)

Challenge:	Recognizing semantic differences across documents is crucial for text generation evaluation and content alignment . but as a standalone task, it has received little attention, a new study shows . evaluating semantic differences between documents is an underexplored challenge in natural language understanding .
Approach:	They introduce SwissGov-RSD, the first naturalistic, document-level, cross-lingual dataset for semantic difference recognition.
Outcome:	The proposed dataset shows that current approaches perform poorly on monolingual, sentence-level and synthetic benchmarks.

Copied to clipboard

Hanxu Hu, Jannis Vamvas, Rico Sennrich

Natural Language Interfaces

Challenge:	Document-level machine translations have paved the way for truly simple document-level translation, but challenges such as omission errors remain.
Approach:	They propose a method for document-level machine translation that leverages previous contexts in a multi-turn conversational manner by decomposing documents into segments and iteratively translating them while maintaining previous turns.
Outcome:	The proposed method outperforms translations of entire documents in a single turn and translations independently according to multiple automatic metrics in representative LLMs.

Wino-X: Multilingual Winograd Schemas for Commonsense Reasoning and Coreference Resolution (2021.emnlp-main)

Copied to clipboard

Denis Emelin, Rico Sennrich

Evaluating Discourse Phenomena in Neural Machine Translation (N18-1)

Challenge:	Winograd schemas are well-established tools for evaluating coreference resolution and commonsense reasoning capabilities of computational models.
Approach:	They present a dataset of German, French, and Russian schemas aligned with their English counterparts.
Outcome:	The proposed model improves in English and German, while the model improve in other languages.

Copied to clipboard

Rachel Bawden, Rico Sennrich, Alexandra Birch, Barry Haddow