AI-Driven Language Pedagogy for Less Commonly Taught Languages: Mina Golestani

Developing a Persian Collocation Resource Using AI and Computational Methods for Language Pedagogy

The Elahé Omidyar Mir-Djalali Institute of Iranian Studies in collaboration with the Department of Middle Eastern Studies and the Center for Middle Eastern Studies, University of Chicago  jointly present "Developing a Persian Collocation Resource Using AI and Computational Methods for Language Pedagogy" on Saturday, February 7, 2026, 1 p.m. (Eastern Time: Canada and US).

Abstract
This study addresses a significant gap in Persian language pedagogy by leveraging artificial intelligence (Al) and computational tools to create a structured, up-to-date, and publicly accessible resource of Persian collocations. Collocation words are essential for fluency and naturalsounding language use. Through a survey among both Persian learners and experienced Persian language teachers, the strong need for such an important resource was identified. Therefore, this project aims to extract Persian collocations from two selected corpora using complementary techniques. To build this resource, two corpora were selected for analysis: Persian Twitter NER (ParsTwiNER), which represents informal and social media language, and Persica, a corpus of formal Persian texts suitable for Multipurpose Text Mining and natural language processing. These corpora were chosen based on their availability, licensing, and suitability for text mining for the goal of this project.

After evaluating the advantages and limitations of various collocation extraction techniques, including statistical measures (like PMI and t-score) and embedding-based methods (such as Word2Vec), the project ultimately adopts a hybrid approach, combining the most effective computational and Al-driven methods to extract collocations that are both frequent and semantically meaningful.

The final product is intended to support both classroom instruction and selfstudy, while also serving as a foundation for further research in Persian linguistics. The paper discusses the current state of Persian collocation resources, compares alternative extraction methods, presents the methodology used, and evaluates the output. By integrating Al-driven methods into language pedagogy, this study contributes to the growing field of technology-enhanced learning for less-resourced and less commonly taught languages such as Persian.

Bio:
Mina Golestani is a student research assistant and a Master's student in Iranian studies and Digital Humanities at the University of Gottingen. She holds a Bachelor's degree in Persian Language and Literature, which laid the foundation for her continued interest in Persian pedagogy. With four years of experience collaborating with the education sector of the Academy of Persian Language and Literature on three different projects, she has developed a growing interest in Language Pedagogy and digital methodologies. Currently, she is working on two projects aimed at developing digital tools that support Persian pedagogy using Al and computational approaches

Zoom Meeting Registration: https://utoronto.zoom.us/meeting/register/i5HR3liYShGHU9rXmcg6cw
After registering, you will receive a confirmation email containing information about joining the meeting.