We are pleased to announce the release of the FR-MIGR-TWIT Corpus 2.0.
The MIGR-TWIT Corpus is a multilingual corpus of tweets developed within the framework of OLiNDiNUM — the Linguistic Observatory of Online Debate. With the aim of documenting and analyzing online public discourse on (im)migration in contemporary European politics, this corpus makes it possible to observe diachronic evolution over the past decade (2011–2022) and linguistic variation across the political spectrum and in two national contexts: France and the UK.
The corpus has previously been published through the following modules: • Tweets of right and far-right politics in Europe (Battaglia, Blandino, Jeon & Pietrandrea, 2022) o UK-R-MIGR-RA-TWIT-2012-2022 o FR-R-MIGR-TWIT-2011-2022 • Tweets of French left-wing politics (Pietrandrea & Jeon, 2023) o FR-L-MIGR-TWIT-2011-2022
Taken together, the published corpora have attracted research interest from, among others, corpus linguistics, discourse studies, and communication and information sciences, reaching more than 2K downloads on Zenodo by 2025.
Compiled from the FR-R and FR-L modules, the FR-MIGR-TWIT Corpus comprises 17,395 tweets posted by 39 French political figures and parties (16 right-wing and 23 left-wing) between 2011 and 2022. Tweets containing migr- derivatives were retrieved via the Twitter API v2 Academic Research, and truncated retweets (>140 characters) were restored through targeted verification.
The FR-MIGR-TWIT Corpus 2.0 offers: • multilayer linguistic annotations associated with each occurrence of a migr- derivative (MIGR-LEXICON), including semantic roles (ROLE_SEM), syntactic functions (FUNC_SYN), lemmatised forms (LEMMA), as well as features and collocational items related to modification (MODIFICATION, LEMMA_MODIF_*, LEMMA_NOUN-1) and list/parallelism constructions (LIST_PAR, LENGTH-1, #forme#_MIGR-LIST_PAR) (Non-exhaustive list); • tweet URLs (tweet_url) and 44 types of data retrieved through the Full Archive Search endpoints of the Twitter API v2, such as the textual content of tweets, posting date, user ID, number of retweets, likes, replies, and quotes. (Non-exhaustive list)
The corpus is available in CSV and TEI-XML formats, with TEI files providing the text layer and CSV files providing stand-off linguistic annotations.
Following version 1.0 (Jeon, Battaglia & Pietrandrea), FR-MIGR-TWIT Corpus 2.0 was developed by Sangwan Jeon (Université de Lille) and Paola Pietrandrea (Université de Lille, IUF).
Changelog version 2.0 (© 2025 Jeon & Pietrandrea) – Added multilayer linguistic annotations – Corrected delimiter-related errors – Added TEI-XML format – Added a basic Python query script – Added README.md
The creation of the corpus and the inter-annotator agreement process were funded by Université de Lille, Campus France, and the Institut Universitaire de France. The corpus is accessible through Zenodo and is scheduled to be archived in Ortolang.
Sangwan Jeon (University of Lille) & Paola Pietrandrea (University of Lille, Institut Universitaire de France)