Shared Task Announcement: MultiLexNorm2 @ W-NUT 2026 (EMNLP 2026)

[JinYeong Bak]

Dear All,

We are pleased to announce MultiLexNorm2: Multilingual Lexical Normalization, the shared task of the 11th Workshop on Natural User-generated Text (W-NUT), to be held at EMNLP 2026.

MultiLexNorm2 builds upon the original MultiLexNorm shared task conducted in 2021, which covered 12 languages. This new edition expands the scope to 17 languages, with particular emphasis on non-Indo-European languages, including Thai, Vietnamese, Korean, Japanese, and Indonesian.

The goal of the task is to develop systems for lexical normalization — the conversion of non-canonical, noisy text into its canonical equivalent form. As user-generated content continues to play a central role in real-world NLP applications, robust and multilingual normalization remains a fundamental challenge.

We warmly invite researchers and practitioners from academia and industry to participate. We especially encourage contributions that advance inclusive multilingual NLP beyond Indo-European language settings.

Further details, including important dates, are available on the workshop website:

https://noisy-text.github.io/2026/multi-lexnorm.html

We look forward to your participation.

Yours sincerely,

JinYeong Bak, on behalf of the MultiLexNorm2 Organizing Committee