## Important links ##
- Shared Task webpage: https://field-matters.github.io/sharedtask2027
- Announcements and general contact (Google group): https://groups.google.com/g/fm-shared-task-glossing-2027
- GitHub repository: https://github.com/field-matters/ST2027
## Description of the task ##
The Field Matters 2027 workshop at EACL 2027 (https://field-matters.github.io) offers a shared task on automatic interlinear glossing for field linguistic annotations.
1 Zbz sc tqwaipele tqnginipenge kraesz'mz
2 zbz sc tq-wai-pe-le tq-ngini-pe-nge kraesz'mz
3 so.then PFV RL-do-COS-3MINIA RL-become-COS-1MINII sixth
4 'So then that's how I became the sixth.' (Natugu)
Figure 1: An Interlinear Glossed Text (IGT) for a sentence in Natugu.
1: Source sentence in Natugu; 2: Morphological segmentation; 3: Interlinear gloss; 4: Translation in English.
The task is to predict interlinear glosses, which are morpheme-level linguistic annotations, as shown in the third line of the figure above, notably used in field linguistics.
Ultimately, such glossing systems aim to help field linguists with the annotation task for languages under documentation, as it is mainly performed manually with little automation.
This Shared Task follows the SIGMORPHON 2023 Shared Task on Interlinear Glossing (https://github.com/sigmorphon/2023GlossingST).
In our Shared Task, the input is mainly the unsegmented source sentence (first line). Some instances will be accompanied by their translation in a high-resource language (usually, in English; fourth line).
The goal is to predict the glossed sentence (third line).
## Languages ##
The task will cover the following languages:
- [Phase 1] 4 from the previous SIGMORPHON 2023 Shared Task on Interlinear Glossing: Gitksan (git), Lezgi (lez), Natugu (ntu), and Nyangbo (nyb).
- [Phase 1] + 1 language from the Glossed Audio Corpus of Ainu Folklore (https://ainu.ninjal.ac.jp/folklore/en/): Hokkaido Ainu (ain).
- [Phase 2]: TBA on 03 November 2026 (cf. website)
## Summary of the shared task rules ##
Submissions to the shared task should follow the rules below:
- Model: Any type of model is allowed, provided that the system is based on open-source or open-weight models (if LLM-based).
Closed-source models can be used for other purposes (e.g., summarising a grammar book, generating pseudo-training data).
- Dataset: Regarding the glossing dataset, we consider the GlossLM/PolyGloss dataset as the main labelled dataset to use. Any data without gloss information (e.g., monolingual sentences, dictionaries) can be used otherwise when publicly available.
- Evaluation: We will mainly evaluate the quality of the predicted gloss. However, we encourage submissions to be able to handle morphological segmentation (second line of the example figure), since field linguists usually analyse sentences into morphemes before producing glosses.
We will mainly use the Morpheme Error Rate (MER) to evaluate the predicted gloss quality.
- Participation tracks: We consider two sets of languages:
- existing_set (from phase 1; SIGMORPHON languages + Ainu) and
- new_set (from phase 2; new languages).
Participation can be in only one of the two tracks, or in both. However, in any case, a submission should contain predictions for all languages of a set.
## How to participate ##
If you are interested in the shared task or for more details, please join our Google Group: https://groups.google.com/g/fm-shared-task-glossing-2027 .
## Important dates ##
- Phase 1: Beginning of the development phase: 26 October 2026
- Phase 2: Information about the new languages: 03 November 2026
- Phase 3: Test set release (beginning of the test phase): 17 November 2026
- Output submission deadline: 08 December 2026
- System description deadline: 15 December 2026
- Notification of acceptance: 11 January 2027
- Camera-ready deadline: 22 January 2027
- Workshop dates: 9-14 March 2027 at EACL 2027
All deadlines are 23:59 UTC-12 ('Anywhere on Earth').
## Organisers ##
Field Matters workshop organisers and dataset contributors
## Contact ##
We suggest using the Google group for general questions that may interest all participants.
For private questions, please write to Shu Okabe (shu dot okabe at tum dot de).