This paper studies gender bias in machine translation through the lens of Large Language Models (LLMs). Four widely-used test sets are employed to benchmark various base LLMs, comparing their translation quality and gender bias against state-of-the-art Neural Machine Translation (NMT) models for English to Catalan (En $\rightarrow$ Ca) and English to Spanish (En $\rightarrow$ Es) translation directions. Our findings reveal pervasive gender bias across all models, with base LLMs exhibiting a higher degree of bias compared to NMT models. To combat this bias, we explore prompting engineering techniques applied to an instruction-tuned LLM. We identify a prompt structure that significantly reduces gender bias by up to 12% on the WinoMT evaluation dataset compared to more straightforward prompts. These results significantly reduce the gender bias accuracy gap between LLMs and traditional NMT systems.

Applying differential privacy (DP) by means of the DP-SGD algorithm to protect individual data points during training is becoming increasingly popular in NLP. However, the choice of granularity at which DP is applied is often neglected. For example, neural machine translation (NMT) typically operates on the sentence-level granularity. From the perspective of DP, this setup assumes that each sentence belongs to a single person and any two sentences in the training dataset are independent. This assumption is however violated in many real-world NMT datasets, e.g. those including dialogues. For proper application of DP we thus must shift from sentences to entire documents. In this paper, we investigate NMT at both the sentence and document levels, analyzing the privacy/utility trade-off for both scenarios, and evaluating the risks of not using the appropriate privacy granularity in terms of leaking personally identifiable information (PII). Our findings indicate that the document-level NMT system is more resistant to membership inference attacks, emphasizing the significance of using the appropriate granularity when working with DP.

With the increasing attention to molecular machine learning, various innovations have been made in designing better models or proposing more comprehensive benchmarks. However, less is studied on the data preprocessing schedule for molecular graphs, where a different view of the molecular graph could potentially boost the model's performance. Inspired by the Byte-Pair Encoding (BPE) algorithm, a subword tokenization method popularly adopted in Natural Language Processing, we propose GraphBPE, which tokenizes a molecular graph into different substructures and acts as a preprocessing schedule independent of the model architectures. Our experiments on 3 graph-level classification and 3 graph-level regression datasets show that data preprocessing could boost the performance of models for molecular graphs, and GraphBPE is effective for small classification datasets and it performs on par with other tokenization methods across different model architectures.

Language Models (LMs) have demonstrated exceptional performance across various Natural Language Processing (NLP) tasks. Despite these advancements, LMs can inherit and amplify societal biases related to sensitive attributes such as gender and race, limiting their adoption in real-world applications. Therefore, fairness has been extensively explored in LMs, leading to the proposal of various fairness notions. However, the lack of clear agreement on which fairness definition to apply in specific contexts (\textit{e.g.,} medium-sized LMs versus large-sized LMs) and the complexity of understanding the distinctions between these definitions can create confusion and impede further progress. To this end, this paper proposes a systematic survey that clarifies the definitions of fairness as they apply to LMs. Specifically, we begin with a brief introduction to LMs and fairness in LMs, followed by a comprehensive, up-to-date overview of existing fairness notions in LMs and the introduction of a novel taxonomy that categorizes these concepts based on their foundational principles and operational distinctions. We further illustrate each definition through experiments, showcasing their practical implications and outcomes. Finally, we discuss current research challenges and open questions, aiming to foster innovative ideas and advance the field. The implementation and additional resources are publicly available at https://github.com/LavinWong/Fairness-in-Large-Language-Models/tree/main/definitions.

Recent advancements in massively multilingual machine translation systems have significantly enhanced translation accuracy; however, even the best performing systems still generate hallucinations, severely impacting user trust. Detecting hallucinations in Machine Translation (MT) remains a critical challenge, particularly since existing methods excel with High-Resource Languages (HRLs) but exhibit substantial limitations when applied to Low-Resource Languages (LRLs). This paper evaluates hallucination detection approaches using Large Language Models (LLMs) and semantic similarity within massively multilingual embeddings. Our study spans 16 language directions, covering HRLs, LRLs, with diverse scripts. We find that the choice of model is essential for performance. On average, for HRLs, Llama3-70B outperforms the previous state of the art by as much as 0.16 MCC (Matthews Correlation Coefficient). However, for LRLs we observe that Claude Sonnet outperforms other LLMs on average by 0.03 MCC. The key takeaway from our study is that LLMs can achieve performance comparable or even better than previously proposed models, despite not being explicitly trained for any machine translation task. However, their advantage is less significant for LRLs.

Automatic speech translation (AST) modeling is usually plagued by lack of parallel training data, which limits the success of end-to-end models. Owing to their modular architecture, cascade models for AST have the advantage of leveraging the large amounts of data available to build automatic speech recognition (ASR) and machine translation (MT) models, respectively. The straightforward way of building cascade AST models is to send the 1-best ASR transcription to the text MT model. Yet another advantage of such an architecture is that it is in fact a multi-modal and multi-task one: besides speech, it also accepts text input for translation and it produces ASR output either in stand-alone mode or as a side-product of the AST task. This multi-input/modal view on the AST task is firmly anchored in the reality of practical applications, so we take it as a fundamental design choice: we aim to build a model that delivers both state of the art ASR and MT performance, while optimizing the AST performance within these constraints. Translating ASR 1-best output has the obvious disadvantage that any further training (fine-tuning) on AST parallel data specific to a given domain is unable to back-propagate cross-entropy loss gradient through the interface between the ASR and the MT model. For tighter coupling between ASR and MT modules we follow the approach of (Dalmia et al., 2021) that leverages the 1-best ASR alignment and sends the ASR encoder embeddings aligned with the 1-best ASR sequence to the MT model. This results in a cascade architecture that allows back-propagation gradient to flow from the MT model into the ASR components. The ASR model in our work uses a conformer encoder architecture (Gulati et al., 2020), pre-trained on a large amount of speech data as described in the Unified Speech Model (USM) work (Zhang et al., 2023).

Text style transfer (TST) is an important task in controllable text generation, which aims to control selected attributes of language use, such as politeness, formality, or sentiment, without altering the style-independent content of the text. The field has received considerable research attention in recent years and has already been covered in several reviews, but the focus has mostly been on the development of new algorithms and learning from different types of data (supervised, unsupervised, out-of-domain, etc.) and not so much on the application side. However, TST-related technologies are gradually reaching a production- and deployment-ready level, and therefore, the inclusion of the application perspective in TST research becomes crucial. Similarly, the often overlooked ethical considerations of TST technology have become a pressing issue. This paper presents a comprehensive review of TST applications that have been researched over the years, using both traditional linguistic approaches and more recent deep learning methods. We discuss current challenges, future research directions, and ethical implications of TST applications in text generation. By providing a holistic overview of the landscape of TST applications, we hope to stimulate further research and contribute to a better understanding of the potential as well as ethical considerations associated with TST.

Gender bias has been a focal point in the study of bias in machine translation and language models. Existing machine translation gender bias evaluations are primarily focused on male and female genders, limiting the scope of the evaluation. To assess gender bias accurately, these studies often rely on calculating the accuracy of gender pronouns or the masculine and feminine attributes of grammatical gender via the stereotypes triggered by occupations or sentiment words ({\em i.e.}, clear positive or negative attitude), which cannot extend to non-binary groups. This study presents a benchmark AmbGIMT (Gender-Inclusive Machine Translation with Ambiguous attitude words), which assesses gender bias beyond binary gender. Meanwhile, we propose a novel process to evaluate gender bias based on the Emotional Attitude Score (EAS), which is used to quantify ambiguous attitude words. In evaluating three recent and effective open-source LLMs and one powerful multilingual translation-specific model, our main observations are: (1) The translation performance within non-binary gender contexts is markedly inferior in terms of translation quality and exhibits more negative attitudes than binary-gender contexts. (2) The analysis experiments indicate that incorporating constraint context in prompts for gender identity terms can substantially reduce translation bias, while the bias remains evident despite the presence of the constraints. The code is publicly available at \url{https://github.com/pppa2019/ambGIMT}.

Complete Multi-lingual Neural Machine Translation (C-MNMT) achieves superior performance against the conventional MNMT by constructing multi-way aligned corpus, i.e., aligning bilingual training examples from different language pairs when either their source or target sides are identical. However, since exactly identical sentences from different language pairs are scarce, the power of the multi-way aligned corpus is limited by its scale. To handle this problem, this paper proposes "Extract and Generate" (EAG), a two-step approach to construct large-scale and high-quality multi-way aligned corpus from bilingual data. Specifically, we first extract candidate aligned examples by pairing the bilingual examples from different language pairs with highly similar source or target sentences; and then generate the final aligned examples from the candidates with a well-trained generation model. With this two-step pipeline, EAG can construct a large-scale and multi-way aligned corpus whose diversity is almost identical to the original bilingual corpus. Experiments on two publicly available datasets i.e., WMT-5 and OPUS-100, show that the proposed method achieves significant improvements over strong baselines, with +1.1 and +1.4 BLEU points improvements on the two datasets respectively.

Pretrained language models (PLMs) display impressive performances and have captured the attention of the NLP community. Establishing the best practices in pretraining has therefore become a major point of focus for much of NLP research -- especially since the insights developed for monolingual English models need not carry to more complex multilingual. One significant caveat of the current state of the art is that different works are rarely comparable: they often discuss different parameter counts, training data, and evaluation methodology. This paper proposes a comparison of multilingual pretraining objectives in a controlled methodological environment. We ensure that training data and model architectures are comparable, and discuss the downstream performances across 6 languages that we observe in probing and fine-tuning scenarios. We make two key observations: (1) the architecture dictates which pretraining objective is optimal; (2) multilingual translation is a very effective pre-training objective under the right conditions. We make our code, data, and model weights available at \texttt{\url{https://github.com/Helsinki-NLP/lm-vs-mt}}.