Papers by David Wu

How Far can 100 Samples Go? Unlocking Zero-Shot Translation with Tiny Multi-Parallel Data (2024.findings-acl)

Challenge:	We extend the Yawipa Wiktionary Parser to extract and normalize translations from etymology glosses and morphological form-of relations.
Approach:	They extend Yawipa to extract and normalize translations from etymology glosses . they propose a method to identify typos in translation annotations based on extracted morphological data .
Outcome:	The proposed method improves on a standard attention baseline by using copy attention.

Copied to clipboard

Di Wu, Shaomu Tan, Yan Meng, David Stap, Christof Monz

Neural Transduction for Multilingual Lexical Translation (2020.coling-main)

Challenge:	a common solution to zero-shot translation is to add as many related translation directions as possible to the training corpus.
Approach:	They show that a small amount of multi-parallel data can achieve significant zero-shot improvements . they say that the resulting non-English performance is close to the complete translation upper bound .
Outcome:	The proposed model achieves +21.7 ChrF++ non-English translation improvements on EC30 dataset . the resulting non- English performance exceeds M2M100 by an average of 5.9 ChrF+ .

Copied to clipboard

Dylan Lewis, Winston Wu, Arya D. McCarthy, David Yarowsky

Image Difference Captioning via Adversarial Preference Optimization (2025.emnlp-main)

Challenge:	a method for completing multilingual translation dictionaries is proposed . a 27% relative improvement in whole-word accuracy is achieved when multilingual data is unavailable .
Approach:	They propose a method for completing multilingual translation dictionaries using multilingual inputs and multilingual decoding objective.
Outcome:	The proposed method can synthesize new word forms in multilingual translation dictionaries . it can perform in settings where correct translations have not been observed in text .

Copied to clipboard

Zihan Huang, Junda Wu, Rohan Surana, Tong Yu, David Arbour, Ritwik Sinha, Julian McAuley

Visual Data in NLP

Challenge:	Existing supervised approaches to image difference captioning overfit to dataset-specific language patterns and fail to capture accurate preferences.
Approach:	They propose an adversarial direct preference optimization framework that aligns captioning policy with pairwise difference preferences via Direct Preference Optimization.
Outcome:	The proposed approach outperforms baselines on benchmark IDC datasets in generating fine-grained and accurate difference descriptions.

Beyond Label Attention: Transparency in Language Models for Automated Medical Coding via Dictionary Learning (2024.emnlp-main)

Copied to clipboard

John Wu, David Wu, Jimeng Sun

Responsible & Trustworthy NLP

Challenge:	Current efforts in interpretability of medical coding rely heavily on label attention mechanisms, which often leads to the highlighting of extraneous tokens irrelevant to the ICD code.
Approach:	They propose to leverage dictionary learning to extract sparsely activated representations from dense language models embedded in superposition to facilitate accurate interpretability.
Outcome:	The proposed model extracts sparsely activated representations from dense language models in superposition, even when the highlighted tokens are medically irrelevant.

Creating Large-Scale Multilingual Cognate Tables (L18-1)

Copied to clipboard

Linear Script Representations in Speech Foundation Models Enable Zero-Shot Transliteration (2026.findings-acl)

Challenge:	Low-resource languages often suffer from a lack of high-coverage lexical resources.
Approach:	They propose a method to generate cognate tables by clustering words from existing lexical resources.
Outcome:	The proposed method outperforms baselines on the Romance and Turkic language families.

Copied to clipboard

Ryan Soh-Eun Shim, Kwanghee Choi, Kalvin Chang, Ming-Hao Hsu, Florian Eichin, Zhizheng Wu, Alane Suhr, Michael A. Hedderich, David Harwath, David R. Mortensen, Barbara Plank

Semantic Text Processing

Challenge:	We show that script information is linearly encoded in the activation space of multilingual speech models . modifying activations at inference time induces script change even in unconventional pairings .
Approach:	They propose to add script vectors to activations at test time to induce script change . they also show that script information is linearly encoded in the activation space of multilingual speech models .
Outcome:	The proposed approach can induce script change even in unconventional language-script pairings.

ACT2: A multi-disciplinary semi-structured dataset for importance and purpose classification of citations (2022.lrec-1)

Copied to clipboard

Suchetha Nambanoor Kunnath, Valentin Stauber, Ronin Wu, David Pride, Viktor Botev, Petr Knoth

Information Extraction & Text Mining

Challenge:	Existing methods for classifying citations rely on bibliometric measures to consider the semantics of citation.
Approach:	They propose to use a Citation Context Classification (3C) shared task dataset to classify citations according to their purpose and importance.
Outcome:	The proposed model can be used to link research works to graphs and enable efficient knowledge discovery.

Sequence Models for Computational Etymology of Borrowings (2021.findings-acl)

Copied to clipboard

Winston Wu, Kevin Duh, David Yarowsky

Detecting Latin in Historical Books with Large Language Models: A Multimodal Benchmark (2026.eacl-long)

Challenge:	a computational model of word borrowing can be useful for lexicon expansion and language preservation.
Approach:	They propose to use neural sequence models to model word borrowings from a donor word to an incorporated word.
Outcome:	The proposed model beats baseline models in both directions, with the quantity of data strongly influencing performance.

Copied to clipboard

Yu Wu, Ke Shu, Jonas Fischer, Lidia Pivovarova, David Rosson, Eetu Mäkelä, Mikko Tolonen

The Johns Hopkins University Bible Corpus: 1600+ Tongues for Typological Exploration (2020.lrec-1)

Challenge:	This paper presents a novel task of extracting low-resourced and noisy Latin fragments from mixed-language historical documents with varied layouts.
Approach:	They propose to extract low-resourced and noisy Latin fragments from mixed-language historical documents with varied layouts using a multimodal dataset.
Outcome:	The proposed model lacks a functional comprehension of Latin, but reliable detection is achievable with zero-shot models.

Copied to clipboard

Arya D. McCarthy, Rachel Wicks, Dylan Lewis, Aaron Mueller, Winston Wu, Oliver Adams, Garrett Nicolai, Matt Post, David Yarowsky

Akan Cinematic Emotions (ACE): A Multimodal Multi-party Dataset for Emotion Recognition in Movie Dialogues (2025.findings-acl)

Challenge:	Our corpus spans 1611 diverse written languages, with constituents of more than 90 language families.
Approach:	They propose to scrape and merge online resources and merge them with existing corpora to create a verse-parallel scheme for all translations.
Outcome:	The results show that the Bible provides high coverage of core vocabulary.

Copied to clipboard

David Sasu, Zehui Wu, Ziwei Gong, Run Chen, Pengyuan Shi, Lin Ai, Julia Hirschberg, Natalie Schluter

Sentiment Analysis

Challenge:	Akan Cinematic Emotions (AkaCE) is the first multimodal emotion dialogue dataset for an African language . it contains 385 emotion-labeled dialogues and 6162 utterances across audio, visual, and textual modalities, along with word-level prosodic prominence annotations.
Approach:	They propose to use AkaCE to analyze African cinematic emotions using word-level prosodic prominence annotations.
Outcome:	The Akan Cinematic Emotions (AkaCE) dataset addresses the significant lack of resources for low-resource languages in emotion recognition research.

CReSE: Benchmark Data and Automatic Evaluation Framework for Recommending Eligibility Criteria from Clinical Trial Information (2024.findings-eacl)

Copied to clipboard

Siun Kim, Jung-Hyun Won, David Lee, Renqian Luo, Lijun Wu, Tao Qin, Howard Lee

Semantic Text Processing

Challenge:	Eligibility criteria (EC) are defined as a set of conditions an individual must meet to participate in a clinical trial.
Approach:	They propose to recommend EC based on clinical trial information, including trial titles, and introduce an automatic evaluation framework to assess clinical validity of the EC recommendation model.
Outcome:	The proposed model outperforms existing language models pre-trained on the biomedical domain in EC clustering.

Computational Etymology and Word Emergence (2020.lrec-1)

Copied to clipboard

A Comparative Study of Extremely Low-Resource Transliteration of the World’s Languages (L18-1)

Challenge:	etymology is the study of words' origins.
Approach:	They develop an extensible Wiktionary parser that predicts the etymology of a word across the full range of ethymological types and languages in Wiktionaries.
Outcome:	The proposed parser predicts the etymology of a word across the full range of ethymologies and languages in Wiktionary, and shows the application of tymatics in modeling this phenomenon.

Copied to clipboard

Multilingual Dictionary Based Construction of Core Vocabulary (2020.lrec-1)

Low-Resource NLP

Challenge:	a phrase-based MT system performs better than other methods for transliterating Bible names . combining data and training a single neural system yields significant gains .
Approach:	They compare several machine translation methods for transliterating Bible names . they find a phrase-based MT system performs better than other methods .
Outcome:	The phrase-based MT system performs better than other methods, the study finds . but, the single-language system outperforms the phrase-backed MT systems .

Copied to clipboard

Winston Wu, Garrett Nicolai, David Yarowsky

Fine-grained Artificial Neurons in Audio-transformers for Disentangling Neural Auditory Encoding (2023.findings-acl)

Challenge:	Existing methods for core vocabulary lists for multiple applications are lacking coverage in sparse dictionaries . we propose a new method for definition and construction of core vocabulary sets based on coverage in dictionary dictionaria .
Approach:	They propose a functional definition and construction method for core vocabulary sets based on relative coverage of a target concept in bilingual dictionaries.
Outcome:	The proposed method achieves high overlap with existing vocabulary lists . it uses a cognate prediction method to recover missing coverage of the vocabulary .

Copied to clipboard

Mengyue Zhou, Xu Liu, David Liu, Zihao Wu, Zhengliang Liu, Lin Zhao, Dajiang Zhu, Lei Guo, Junwei Han, Tianming Liu, Xintao Hu

Semantic Text Processing

Challenge:	Existing studies treat each transformer encoding layer as a single artificial neuron . layer-level embeddings aggregate multiple types of contextual attention captured by multiple head modules .
Approach:	They propose to embed each transformer encoding layer as a single artificial neuron . they propose to couple those ANs with their biological-neuron counterparts in the human brain .
Outcome:	The proposed models can be used to link representations to brain activity, the authors say . their results show that the proposed models carry meaningful neurolinguistic information .

Academics Can Contribute to Domain-Specialized Language Models (2024.emnlp-main)

Copied to clipboard

Mark Dredze, Genta Winata, Prabhanjan Kambadur, Shijie Wu, Ozan Irsoy, Steven Lu, Vadim Dabravolski, David Rosenberg, Sebastian Gehrmann

On the Robustness of Cognate Generation Models (2022.lrec-1)

Challenge:	Commercially available models dominate academic leaderboards, focusing on creating and adapting general-purpose models . however, general- purpose models often underperform in specialized domains, and domain-specific models yield superior results.
Approach:	They advocate for a renewed focus on developing and evaluating domain- and task-specific models . they advocate for an adapted or adapted model that can be used to improve academic leaderboard standings .
Outcome:	The proposed model can do well on professional and linguistic examinations, college-level knowledge questions, and collections of reasoning tasks.

Copied to clipboard

Responsible & Trustworthy NLP

Challenge:	We examine different types of noise generated by human errors and how these noisy inputs affect the performance of cognate generation models.
Approach:	They evaluate two popular neural cognate generation models’ robustness to human-plausible noise.
Outcome:	The proposed models are robust to deletion, duplication, swapping, keyboard errors, and a new type of error, phonological errors.

Efficient Citer: Tuning Large Language Models for Enhanced Answer Quality and Verification (2024.findings-naacl)

Copied to clipboard

Marzieh Tahaei, Aref Jafari, Ahmad Rashid, David Alfonso-Hermelo, Khalil Bibi, Yimeng Wu, Ali Ghodsi, Boxing Chen, Mehdi Rezagholizadeh

MobileLLM-Flash: Latency-Guided On-Device LLM Design for Industry Scale Deployment (2026.acl-industry)

Challenge:	Existing models with explicit citations lack the ability to verify information generated by these models.
Approach:	They construct a citation training dataset and fine-tune two models to address the challenge of explicit citations efficiently.
Outcome:	The proposed models surpass ChatGPT and exhibit exceptional out-of-domain generalization in both human and automatic evaluation.

Copied to clipboard

Hanxian Huang, Igor Fedorov, Andrey Gromov, Bernard Beckerman, Naveen Suda, David Eriksson, Maximilian Balandat, Rylan Conway, Patrick Huber, Chinnadhurai Sankar, Ayushi Dalmia, Zechun Liu, Lemeng Wu, Tarek Elgamal, Adithya Sagar, Vikas Chandra, Raghuraman Krishnamoorthi

Green & Sustainable NLP

Challenge:	MobileLLM-Flash is a family of foundation models for efficient on-device use with strong capabilities.
Approach:	They propose a method for designing on-device large language models under mobile latency constraints using hardware-in-the-loop architecture search.
Outcome:	The proposed model is amenable to industry-scale deployment and is compatible with mobile runtimes like Executorch.

Modeling Color Terminology Across Thousands of Languages (D19-1)

Copied to clipboard

Arya D. McCarthy, Winston Wu, Aaron Mueller, William Watson, David Yarowsky

Challenge:	Existing studies on what constitutes a "basic" color term and its acquisition sequence are flawed . a pan-lingual approach may reveal general color trends more reliably than smaller datasets.
Approach:	They propose to operationalize and critique the Berlin and Kay color term hypotheses . they use 14 empirically-grounded computational linguistic metrics to analyze cross-linguistic data .
Outcome:	The proposed measures correlate strongly with the Berlin and Kay color term partition and their hypothesized universal acquisition sequence.

An Analysis of Massively Multilingual Neural Machine Translation for Low-Resource Languages (2020.lrec-1)

Copied to clipboard

Aaron Mueller, Garrett Nicolai, Arya D. McCarthy, Dylan Lewis, Winston Wu, David Yarowsky

Walk in Others’ Shoes with a Single Glance: Human-Centric Visual Grounding with Top-View Perspective Transformation (2025.acl-long)

Challenge:	In this study, we explore massively multilingual low-resource neural machine translation.
Approach:	They propose to use Bible translations to train models with up to 1,107 source languages and create multilingual corpora varying the number and relatedness of source languages.
Outcome:	The proposed approach is highly language-specific and can be tailored to the source language and its typology.

Copied to clipboard

Yuqi Bu, Xin Wu, Zirui Zhao, Yi Cai, David Hsu, Qiong Liu

Multimodality

Challenge:	Existing VLMs are insensitive to information differences induced by slight perspective changes.
Approach:	They propose a visual perspective-taking task that requires robots to interpret human-centric instructions and identify corresponding objects from robot perspectives.
Outcome:	The proposed method improves performance by up to 18% and generalizes effectively to robotic and dynamic scenarios.

Massively Translingual Compound Analysis and Translation Discovery (L18-1)

Copied to clipboard

MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies (2023.acl-long)

Challenge:	Morphological compounding is one of the most common and productive methods of word formation across the world's languages.
Approach:	They propose a model for compounding using bilingual dictionaries and no annotated training data . they also release a massively multilingual dataset of compound words and their decompositions .
Outcome:	The proposed model generates novel translations of English concepts on a multilingual dataset . the model can be applied to a wide range of languages and is highly reproducible.

Copied to clipboard

Shiyue Zhang, Shijie Wu, Ozan Irsoy, Steven Lu, Mohit Bansal, Mark Dredze, David Rosenberg

Fine-grained Morphosyntactic Analysis and Generation Tools for More Than One Thousand Languages (2020.lrec-1)

Challenge:	Autoregressive language models are trained by minimizing the cross-entropy of the model distribution Q relative to the data distribution P. However, these systems still struggle in many openended generation settings, where they are asked to produce a long text following a short prompt.
Approach:	They propose to combine forward and reverse cross-entropy to train autoregressive language models by minimizing the cross-Entropy of the model distribution Q relative to the data distribution P.
Outcome:	The proposed model overgeneralizes and produces non-human-like text without complex decoding strategies.

Copied to clipboard

Garrett Nicolai, Dylan Lewis, Arya D. McCarthy, Aaron Mueller, Winston Wu, David Yarowsky

Syntactic Text Processing

Challenge:	Using morphosyntactic tools, we train and distribute tools for approximately one thousand languages.
Approach:	They train and distribute morphosyntactic tools for approximately one thousand languages.
Outcome:	The results show that the tools generalize well across rare and common forms alike.

Scaling Collaborative Effort with Agents (2026.findings-acl)

Copied to clipboard

Shannon Zejiang Shen, Valerie Chen, Ken Gu, Alexis Ross, Zixian Ma, Jillian Ross, Alex Gu, Chenglei Si, Wayne Chi, Andi Peng, Jocelyn J Shen, Ameet Talwalkar, Tongshuang Wu, David Sontag

Challenge:	Current evaluations of agents focus on producing high-quality, final outputs in one shot, failing to account for the inherently iterative nature of many real-world problems.
Approach:	They propose a framework that captures how an agent’s utility grows with increasing user involvement.
Outcome:	The proposed framework captures how an agent’s utility grows with increasing user involvement, revealing a missing ingredient in agent design: the ability to sustain engagement and scaffold user understanding.

Creating a Translation Matrix of the Bible’s Names Across 591 Languages (L18-1)

Copied to clipboard

Winston Wu, Nidhi Vyas, David Yarowsky