Lexhnology: joint linguistic and NLP discourse structure modeling of legal texts for language pedagogy – Lexhnology

Given the growing extraterritoriality of American law, national law is increasingly impacting countries outside of its normal jurisdiction. Comprehension of law texts by English second-language (L2) legal learners and professions, therefore, is of prime importance. Indeed, there is a growing premise that making the divisions of texts explicit in terms of communicative functions, also known as moves, helps L2 learners of languages for specific purposes (LSP) to identify the information structure and communication objectives necessary for understanding legal documents.
Despite this context, consensus about the linguistic definition of moves in case law does not yet exist. In addition, no Natural Language Processing (NLP) techniques are currently able to automatically identify moves in case law. Finally, the effectiveness of making moves explicit to L2 learners has not been measured experimentally.

These gaps raise several research questions:
1. What typology of moves and prototypes for judicial opinions can be established using move analysis combined with corpus linguistics augmented by NLP methods ? 2. Does interaction with a corpus with annotated moves improve reading comprehension of case law? 3. What needs and use cases exist for L2 learners of legal English and LSP teachers concerning the task of reading case law? 4. How does access to a corpus with annotated moves foster learner and teacher autonomy in reading case law in L2 students and jurists? 5. How do current neural language models represent the discourse structure underlying the moves? How can self-supervised learning take discourse information into "attention"? How to design predictive neural architectures to make them discourse aware ? How can linguistic knowledge be injected into computational models, making them more self-explanatory?

To answer these questions, Lexhnology will take an innovative interdisciplinary approach – linguistic, NLP, LSP teaching/learning. It will define moves in case law, annotate a corpus, automatize move recognition, while measuring the effectiveness of using moves in LSP teaching/learning. These deliveries will be integrated into an open access platform for students, teachers, translators and legal professionals and packaged to be released as an open educational resource.