PeopleofSuperintelligence

Researcher

Robert Östling

Robert Östling is a computational linguistics researcher whose AI work develops and evaluates systems for many languages. In 2015, he applied Bayesian word alignment and part-of-speech transfer to the Swedish Sign Language Corpus. His language-vector work tested multilingual models on Bible translations in 990 languages, and his work on efselab produced fast part-of-speech and named-entity tagging. He later studied multilingual language representations and evaluated GPT-3 and other large language models for Swedish grammar correction, historical entity recognition and conflicting information. As of October 2026, he is a docent in Stockholm University's Department of Linguistics and represents its computational linguistics section.

Now at
Department of Linguistics
Works
65
Citations
747
h-index
14

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2015

    Part-of-speech tags for sign language

    In 2015, Östling and coauthors use a Bayesian model for joint word alignment and part-of-speech transfer to add tags to the Swedish Sign Language Corpus.

  2. 2017

    Language vectors for 990 languages

    In 2017, he evaluates multilingual language models on 1,303 Bible translations in 990 languages. The language vectors capture genetic relationships between languages.

  3. 2018

    efselab for fast tagging

    In a 2018 paper, Östling presents efselab, a compiler for sequence-labeling tools that produces part-of-speech taggers and named-entity recognizers. The system is three orders of magnitude faster than the neural network model and more accurate than other systems on 47 of 65 treebanks.

  4. 2019–2023

    Representations and typological structure

    In 2019, Östling and Jörg Tiedemann examine correlations between language representations learned from translations and languages' genetic, geographic and structural similarity. In 2023, he and coauthors train neural models for NLP tasks on Bible translations in 1,295 languages.

  5. 2024

    GPT-3 for Swedish correction

    In 2024, he evaluates grammatical error correction for Swedish. The study finds that few-shot GPT-3 outperforms previous Swedish systems; Swedish accounts for about 0.1% of GPT-3's training data.

  6. 2025

    AI and historical named entities

    In June 2025, Östling researches and teaches at Stockholm University and coordinates its Språkbanken CLARIN node; he is interested in using AI to explore questions. A 2025 study investigates prompt-answering LLMs for named-entity recognition in historical texts and finds that hallucinations affect performance.

  7. April 2026

    Conflicting information in multilingual LLMs

    In April 2026, Östling and Murathan Kurfali report that tested LLMs, including GPT-5.2, usually ignore conflicting information and confidently assert one answer.

  8. As of October 2026

    Docent and section representative

    As of October 2026, he is a docent in Stockholm University's Department of Linguistics and the representative for its computational linguistics section.

Career

  1. Since 2025Stockholm UniversityCoordinator
  2. NowDepartment of LinguisticsRepresentative

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Efficient Word Alignment with Markov Chain Monte Carlo2016 · ˜The œPrague Bulletin of Mathematical Linguistics · 106 citations
  2. Continuous multilinguality with language vectors2017 · Conference of the European Chapter of the Association for Computational Linguistics · 72 citations
  3. Stagger: an Open-Source Part of Speech Tagger for Swedish2013 · Northern European Journal of Language Technology · 62 citations
  4. Word Order Typology through Multilingual Word Alignment2015 · 55 citations
  5. Distribution and duration of signs and parts of speech in Swedish Sign Language2016 · Sign Language & Linguistics · 46 citations

From OpenAlex, as of 21 October 2026.

  1. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Stockholm University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Stockholm University. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  3. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. The United States employs an estimated 48% of people shaping AIWe estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.

Sources and credits

Photo: We have no published photo of Robert Östling yet. Send us a free-licensed photo.

  1. Enriching the Swedish Sign Language Corpus with Part of Speech Tags Using Joint Bayesian Word Alignment and Annotation Transfer ep.liu.se. Checked 20 October 2026.
  2. Continuous multilinguality with language vectors aclanthology.org. Checked 20 October 2026.
  3. Part of Speech Tagging: Shallow or Deep Learning? nejlt.ep.liu.se. Checked 20 October 2026.
  4. What Do Language Representations Really Represent? aclanthology.org. Checked 20 October 2026.
  5. Language Embeddings Sometimes Contain Typological Generalizations aclanthology.org. Checked 20 October 2026.
  6. Evaluation of Really Good Grammatical Error Correction aclanthology.org. Checked 20 October 2026.
  7. Månadens profil: Robert Östling sprakbanken.se. Checked 20 October 2026.
  8. Prompting the Past: Exploring Zero-Shot Learning for Named Entity Recognition in Historical Texts Using Prompt-Answering LLMs aclanthology.org. Checked 20 October 2026.
  9. Language Bias under Conflicting Information in Multilingual LLMs arxiv.org. Checked 20 October 2026.
  10. Contact - Stockholm University su.se. Checked 20 October 2026.