SIGTURK 2027 Workshop
The Third Meeting of the ACL Special Interest Group on Turkic Languages
Co-located with ACL 2027 · Kyoto, Japan · August 21 or 22, 2027 (exact date TBA)
Introduction
The Third Meeting of the ACL Special Interest Group on Turkic Languages (SIGTURK 2027) will be held in conjunction with the 65th Annual Meeting of the Association for Computational Linguistics (ACL 2027) in Kyoto, Japan.
SIGTURK provides a venue for research on computational linguistics and natural language processing (NLP) for Turkic languages. Building on the previous editions, SIGTURK 2027 welcomes research across all areas of Turkic NLP, with particular emphasis on resources and benchmarks, adaptation and evaluation of large language models, and multimodal language technologies. We especially encourage contributions addressing under-resourced Turkic languages, regional varieties, and diverse writing systems.
The main objectives of the workshop are to:
Promote the creation, curation, standardization, evaluation, and open sharing of corpora, benchmarks, and software libraries that support reproducible research across the Turkic language family.
Encourage participation by researchers working on Turkic languages, particularly those based in under-represented communities.
Host a shared task that establishes strong baselines and fosters collaboration on key challenges in Turkic NLP.
Provide a venue for discussing tokenization, morphology, evaluation, and modeling methods suitable for the typological features of Turkic languages.
Strengthen collaboration among academia, industry, and local communities to advance NLP methods and tools across Turkic languages.
Important Dates
| Event | Date |
|---|---|
Direct paper submission deadline | May 2, 2027 |
ARR commitment deadline | TBA |
Notification of acceptance | May 29, 2027 |
Camera-ready submission deadline | June 19, 2027 |
Workshop | August 21 or 22, 2027 (TBA) |
All dates are tentative until confirmed with the ACL 2027 workshop chairs. Deadlines are at 23:59 UTC-12:00 (“anywhere on Earth”). Shared task and benchmark contribution deadlines are listed on the Shared Task page.
Topics of Interest
We invite submissions on all aspects of Turkic computational linguistics and NLP. Topics of interest include, but are not limited to:
Creation, curation, documentation, and evaluation of datasets, corpora, and benchmarks for monolingual and multilingual tasks
Computational modeling of Turkic morphology, syntax, semantics, and other linguistic phenomena
Tokenization, morphology-aware modeling, and diagnostic probing of language models
Large language model pretraining, continual adaptation, and parameter-efficient fine-tuning for Turkic languages
Rigorous evaluation of language models, including cultural and linguistic capabilities across Turkic languages and varieties
Cross-lingual transfer between Turkic languages and between Turkic and other languages, including zero-shot and low-resource adaptation
Multimodal NLP, including speech, optical character recognition, and document processing
Responsible NLP, including privacy, safety, fairness, and bias mitigation
NLP applications, open-source software libraries, and tools for Turkic languages
Shared Task: Culture and Understanding Across Turkic Languages
The workshop hosts a shared task on adapting multilingual large language models to improve their cultural and linguistic capabilities in Turkic languages. Systems will be evaluated on tasks such as cultural knowledge reasoning, reading comprehension, and cross-lingual cultural understanding across diverse Turkic languages and varieties.
Languages, tracks, rules, evaluation, and the timeline are on the Shared Task page.
Community Benchmark Contributions
Alongside the shared task, we invite community contributions to the Culture and Understanding Across Turkic Languages benchmark: cultural knowledge questions, dialect identification examples, and reading comprehension passages for under-resourced Turkic languages.
Accepted contributions will be integrated into a unified benchmark suite, and contributors will be invited to co-author a joint overview paper, subject to contribution and authorship criteria. Instructions on submission, expert validation, data licensing, and deadlines will be published on the Shared Task page.
Benchmark contributions and shared task participation are separate, each with its own submission instructions and timeline.
Submissions
We invite:
Research papers of up to 8 pages, excluding references. Shorter papers are welcome; we make no formal distinction between long and short papers. Accepted papers receive one additional page for the camera-ready version.
Shared task system descriptions of 4–6 pages, excluding references.
All papers must use the official ACL templates. Research papers submitted directly must be anonymized for double-blind review.
Papers can be submitted through two routes:
Direct submission via the workshop’s OpenReview venue, with double-blind review by the program committee. The submission link will be posted here.
ARR commitment of papers already reviewed through ACL Rolling Review, by the ARR commitment deadline. Submitting to ARR alone does not count as a submission to the workshop. The commitment link, eligible ARR cycles, and review requirements will be posted here.
Policies on submission language, concurrent submissions, and previously published work will also be posted on this page.
We encourage authors to share code, datasets, models, and documentation where possible, using appropriate licenses and preserving anonymity during review.
Accepted papers will be published in the workshop proceedings and presented as talks or posters. Any presentation-only or non-archival options will be announced separately.
Workshop Format
SIGTURK 2027 is planned as a one-day hybrid event, in person in Kyoto with remote participation, subject to confirmation of the conference arrangements. Registration and presentation requirements will be announced on this page.
We aim to provide an inclusive venue for NLP researchers, linguists, industry practitioners, and members of Turkic-speaking communities, and particularly welcome students and researchers working on under-represented languages and varieties.
Outstanding contributions will be recognized with a Best Paper Award and Honorable Mentions.
Organizers
Abdullatif Köksal, Google DeepMind, United Kingdom
Dilara Keküllüoğlu, Sabancı University, Türkiye
Sherzod Hakimov, University of Potsdam, Germany
Jonne Sälevä, Brandeis University, United States
Çağrı Toraman, Middle East Technical University, Türkiye
İlker Kesen, Humboldt-Universität zu Berlin, Germany
Contact
For questions, email sigturk.secretary@gmail.com. For general SIGTURK news, see sigturk.org.