--- title: "Soroush Vosoughi — author profile and papers" description: "Soroush Vosoughi, Dartmouth College: 3 papers on Elamur Research (2021–2022), h-index 25 — affiliations, topics, co-authors and who cites them in our…" canonical: "https://research.elamur.ai/free/en/authors/soroush-vosoughi/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/soroush-vosoughi/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Soroush Vosoughi # Soroush Vosoughi Soroush Vosoughi works in natural language processing. Their papers here, from 2021 and 2022, look at three questions: how to enlarge and diversify training data for text models, how large language models carry out step-by-step reasoning in languages other than English, and how classifiers can recognise sarcasm in social media posts. The methods range from a broad survey of existing techniques to building a translated benchmark and fine-tuning transformer classifiers for a shared task. The most cited paper here is [A Survey of Data Augmentation Approaches for NLP](https://md.research.elamur.ai/free/en/papers/feng-2021-survey-data-augmentation-approaches-nlp/) (2021). It explains why augmenting text is harder than augmenting images, since words are discrete and labels can shift after a change. It sorts the methods into rule-based edits, MixUp-style blending of examples, and generation by models, for instance through backtranslation. Next is [Language Models are Multilingual Chain-of-Thought Reasoners](https://md.research.elamur.ai/free/en/papers/shi-2022-language-models-multilingual-chain-thought/) (2022). It introduced MGSM, a set of 250 grade-school maths problems that professional translators rendered into ten typologically varied languages, among them Bengali, Swahili and Telugu. It then compared how GPT-3 and PaLM perform when prompted to answer directly, to reason in the question's language, or to reason in English. The 2022 work turns toward [multilingual and cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) and figurative language. [Dartmouth at SemEval-2022 Task 6](https://md.research.elamur.ai/free/en/papers/lad-2022-dartmouth-semeval-2022-task-6/) describes a sarcasm detection system for English and Arabic tweets, built on RoBERTa, XLM-RoBERTa and BERT. Its ideas include masking verbs and adjectives and repeating them as a second segment, softening positive labels to reflect uneven annotation, and settling pairwise questions by comparing single-item scores. Other topics covered here include [reasoning](https://md.research.elamur.ai/free/en/topics/reasoning/) and sentiment. Vosoughi is at [Dartmouth College](https://md.research.elamur.ai/free/en/universities/dartmouth-college/) in Hanover, in computer science, as an assistant professor from 2019 to 2025 and an associate professor since 2025. Earlier degrees and a postdoctoral position from 2015 to 2018 were at the [Massachusetts Institute of Technology](https://md.research.elamur.ai/free/en/universities/massachusetts-institute-of-technology/), including work linked to the [MIT Media Lab](https://md.research.elamur.ai/free/en/labs/mit-media-lab/). Co-authors on the papers here include [Freda Shi](https://md.research.elamur.ai/free/en/authors/freda-shi/), [Jason Wei](https://md.research.elamur.ai/free/en/authors/jason-wei/) and [Eduard Hovy](https://md.research.elamur.ai/free/en/authors/eduard-hovy/). Soroush Vosoughi has 3 papers in our library, in 2021–2022, most of them on [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/), [Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/), [Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/); often with [Jason Wei](https://md.research.elamur.ai/free/en/authors/jason-wei/), [Denny Zhou](https://md.research.elamur.ai/free/en/authors/denny-zhou/), [Dipanjan Das](https://md.research.elamur.ai/free/en/authors/dipanjan-das/). 3 papers in our library · 2021–2022 [OpenAlex](https://openalex.org/A5035399743) · [ORCID](https://orcid.org/0000-0002-2564-8909) · [Wikidata](https://www.wikidata.org/wiki/Q88200397) · [**h-index 25**](https://openalex.org/authors/A5035399743) · [265 works](https://openalex.org/works?filter=author.id:A5035399743) · [11,361 citations](https://openalex.org/works?filter=author.id:A5035399743&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [📄 3 works in our library](https://md.research.elamur.ai/free/en/authors/soroush-vosoughi/#papers) · [🔗 892 citations in our library](https://md.research.elamur.ai/free/en/authors/soroush-vosoughi/cited-by/) · [14 more in our library without a page yet](#also) · [⭐ Paper Star rank #1553](https://md.research.elamur.ai/free/en/paper-stars/page/16/#star-1553) The profile is the library's match of the name; two people with one name are not always told apart. Elsewhere: [Google Scholar](https://scholar.google.com/citations?view_op=search_authors&mauthors=Soroush+Vosoughi) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Soroush+Vosoughi) · [dblp](https://dblp.org/search?q=Soroush+Vosoughi) · [ACL Anthology](https://aclanthology.org/search/?q=Soroush+Vosoughi) ## Affiliations [🎓**Dartmouth College** Hanover · United States 2020–2026 · 160 papers*current*](https://md.research.elamur.ai/free/en/universities/dartmouth-college/) · [🎓**Massachusetts Institute of Technology** Cambridge · United States 2008–2021 · 35 papersEducation until 2015 · PhDEducation until 2010 · MScEducation until 2008 · BSc, Computer ScienceEmployment 2015–2018 · Postdoctoral Associate](https://md.research.elamur.ai/free/en/universities/massachusetts-institute-of-technology/) · [🔬**MIT Media Lab** Cambridge · United States 2008–2018 · 11 papers](https://md.research.elamur.ai/free/en/labs/mit-media-lab/) · [🎓**Yale University** New Haven · United States 2025 · 2 papers](https://md.research.elamur.ai/free/en/universities/yale-university/) · [🔬**International Monetary Fund** Washington · United States 2021 · 2 papers](https://ror.org/05m4rmw09) 🏛️**Dartmouth** Hanover · United States 2019–2025Employment 2019–2025 · Assistant Professor, Computer ScienceEmployment since 2025 · Associate Professor, Computer Science From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 7 more institutions on single bylines are not listed. ## Topics [Language models & pretraining (1)](https://md.research.elamur.ai/free/en/topics/language-models/) · [Sentiment & emotion (1)](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · [Multilingual & cross-lingual NLP (1)](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) · [Reasoning (1)](https://md.research.elamur.ai/free/en/topics/reasoning/) · [Semitic languages (1)](https://md.research.elamur.ai/free/en/topics/semitic-languages/) ## Works by year ### 2020s [**2022** 2 papers](https://md.research.elamur.ai/free/en/authors/soroush-vosoughi/year/2022/) · [**2021** 1 paper](https://md.research.elamur.ai/free/en/papers/feng-2021-survey-data-augmentation-approaches-nlp/) ## Papers The most cited first (OpenAlex and our library). 1. **[A Survey of Data Augmentation Approaches for NLP](https://md.research.elamur.ai/free/en/papers/feng-2021-survey-data-augmentation-approaches-nlp/)** *(2021 · 935 citations)* Data augmentation (DA) adds variety to training data without collecting new examples. It is standard in computer vision but less developed in NLP. Text is discrete, so continuous noise cannot be applied, and it is harder to keep labels unchanged after a transformation. Feng et al. present what they call the first… 2. **[Language Models are Multilingual Chain-of-Thought Reasoners](https://md.research.elamur.ai/free/en/papers/shi-2022-language-models-multilingual-chain-thought/)** *(2022 · 490 citations)* Chain-of-thought (CoT) prompting was known to unlock multi-step reasoning in large language models, but mostly in English, and existing multilingual benchmarks only test shallow reasoning. 3. **[Dartmouth at SemEval-2022 Task 6: Detection of Sarcasm](https://md.research.elamur.ai/free/en/papers/lad-2022-dartmouth-semeval-2022-task-6/)** *(2022 · 2 citations)* The paper describes Team Dartmouth's entry to SemEval-2022 Task 6 (iSarcasmEval), which asks systems to recognise sarcasm in English and Arabic tweets across five subtasks. ## Also in our library 14 more works of theirs are in our library without a page here yet; they will appear as we publish them. ## Co-authors here 12 of 14 co-authors with a page, the most papers together first. - [Jason Wei](https://md.research.elamur.ai/free/en/authors/jason-wei/) ×2 - [Denny Zhou](https://md.research.elamur.ai/free/en/authors/denny-zhou/) - [Dipanjan Das](https://md.research.elamur.ai/free/en/authors/dipanjan-das/) - [Eduard Hovy](https://md.research.elamur.ai/free/en/authors/eduard-hovy/) - [Freda Shi](https://md.research.elamur.ai/free/en/authors/freda-shi/) - [Hyung Won Chung](https://md.research.elamur.ai/free/en/authors/hyung-won-chung/) - [Markus Freitag](https://md.research.elamur.ai/free/en/authors/markus-freitag/) - [Mirac Suzgun](https://md.research.elamur.ai/free/en/authors/mirac-suzgun/) - [Sebastian Ruder](https://md.research.elamur.ai/free/en/authors/sebastian-ruder/) - [Steven Y. Feng](https://md.research.elamur.ai/free/en/authors/steven-y-feng/) - [Teruko Mitamura](https://md.research.elamur.ai/free/en/authors/teruko-mitamura/) - [Varun Gangal](https://md.research.elamur.ai/free/en/authors/varun-gangal/) ## Cited by Papers of our library cite their works 892 times (of the 140,884 papers whose bibliographies we read); [67 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/soroush-vosoughi/cited-by/).