SIGTYP 2023 Shared Task


In 2023, SIGTYP is hosting a Shared Task on Cognate and Derivative Detection for Low-Resourced Languages. Cognates are etymologically related word pairs across languages which may or may not have similar spelling, pronunciation and meaning. Cognates can be traced back to a single ancestral word form in a common earlier language stage. On the other hand, derivatives are words which have been derived directly from another word form or root, including both words inherited from an earlier stage of the same language and words borrowed from distinct languages. Cognates have been studied in various fields of linguistics with different purposes. In historical linguistics, cognates are useful in the reconstruction of proto-languages and can aid in establishing the relationship between languages; in lexicography, cognates are helpful in the development of multilingual dictionaries. Moreover, in recent years, NLP researchers have shown interest in using cognates to enhance the performance of multilingual tasks such as machine translation, lexical induction, word embeddings and many more.
As there has been little work on automatic cognate identification, it is still a challenging task, especially for less-resourced languages. Supervised identification of cognates and derivatives is expensive as we need more linguistic information and annotation; at the same time, finding linguists and annotators for less-resourced languages is impractical. Thus we propose a shared task which aims to provide a new benchmark for differentiating between cognates and derivatives and introduce new unsupervised approaches for cognate and derivative detection in less-resourced languages.

The tasks are multiclass classification tasks which require that the relationship between pairs of words be identified as either a cognate relationship, a derivative relationship, or no relationship.
a. Supervised: Cognate and Derivative Detection
b. Unsupervised: Cognate and Derivative Detection

We will provide 232,482 annotated word pairs in 34 languages, including both high-resourced and less-resourced languages. Each word pair is annotated with a language for each word, and with one of the three categories based on the relationship between the pair; cognate, derivative or none. This data was collected and annotated using wiktionary. In addition to this, the shared task allows participants to use external resources as long as they are openly available and can be, in theory, used by other participants for research purposes. In case participants decide to use external resources and data in their system they should use proper citation and provide the details of the dataset in the system description.

Participants will be invited to describe their system in a paper for the SIGTYP workshop proceedings. The task organizers will write an overview paper that describes the task and summarizes the different approaches taken, and analyzes their results.

Important Links

  ↣  Shared Task GIT Page (provides all the details and regular updates)!  

  ↣  Register for the Task!  

Important Dates

  Training data Release: ↣  9 January 2023
  Test data Release: ↣  24 February 2023
  Submissions Due: ↣  10 March 2023
  System descriptions are Due: ↣  20 March 2023
  Camera-ready Due: ↣  27 March 2023

Task Organizers

Priya Rani, PhD Student, SFI Centre for Research and Training in AI, Data Science Institute, University of Galway
Koustava Goswami, Adobe Research, Bangalore, India; Former PhD Student, Insight SFI Research Centre for Data Analytics, Data Science Institute, University of Galway
Adrian Doyle, Research associate, Insight SFI Research Centre for Data Analytic, Data Science Institute, University of Galway
Bernardo Stearns, Research associate, Insight SFI Research Centre for Data Analytic, Data Science Institute, University of Galway
Theodorus Fransen, Postdoctoral Researcher, Insight SFI Research Centre for Data Analytics, Data Science Institute, University of Galway
John P. McCrae, Assistant Professor, Insight SFI Research Centre for Data Analytics, Data Science Institute, University of Galway

Contact

    Please contact priya.rani@insight-centre.org if you have any questions