Karine Megerdoomian, PhD

Linguist
Natural Language Processing Researcher
Artificial Intelligence Engineer

Founder, Zoorna Institute for Language, AI and Society

I'm a linguist and Artificial Intelligence specialist, working at the intersection of linguistics, AI, and societal impact. I've always been fascinated by the scientific analysis of language and the way language shapes our world—and how technology can enhance our understanding of it.

My work explores how AI systems, especially those driven by Natural Language Processing (NLP), can be designed to better understand and reflect the complexities of human language, and how language can be analyzed to shed light on socio-political issues. My particular areas of expertise are social media analytics, detection of radicalization, and narrative analysis. I also specialize in Persian and Armenian linguistics, where I’ve had the opportunity to combine my passion for these languages with cutting-edge NLP research and development. My research in theoretical linguistics focuses on the analysis of complex predicates, the syntax-semantics interface, lexical semantics, and the computational representation of language.

After nearly 18 years at MITRE, I have recently founded the Zoorna Institute for Language, AI and Society, a non-profit research group aiming to strengthen and spur research in low-resource languages of the Middle East and the Caucasus, and to apply our solutions for social good.

About Me

Research Interests

My research focuses on computational approaches for less commonly investigated languages with strong specialization on Persian (Farsi, Dari, Tajiki) and Armenian, semantic approaches to AI, linguistic studies of complex predicates and causativity, the syntax-semantics interface, and the investigation of the linguistic characteristics of heritage speakers.

My main passion is in bringing together formal linguistic theories and approaches within NLP applications and language pedagogy. This has allowed me to work on a number of interdisciplinary projects developing linguistically-informed computational models of natural language; the application of NLP and Artificial Intelligence to the study of complex social processes; and the discovery of knowledge through the study of semantic, temporal, and causal relations in unstructured text. More recently, my work explores the applications of technology and artificial intelligence in the legal domain and in knowledge discovery of health information.

My formal linguistic research has focused on the close association between syntactic structure and semantic interpretation across languages, the investigation of natural laws in language, and the study of the linguistic characteristics of heritage speakers. I have written several publications on complex predicates, causatives, and verb phrase structure in Persian and Armenian, and I am currently studying the role of recursion and scaling in linguistic structure.

Learn More

Recent Publications

Linguistic data-driven approach to Persian language pedagogy: Practical application to compound verbs (forthcoming)

(forthcoming). In Persian Language Pedagogy: New Trends and Innovations, edited by Azita Taleghani and Pouneh Shabani-Jadidi; Routledge.

There exists a significant disconnect between linguistic research and pedagogical practice in the Persian classroom. Although theoretical and computational studies of Persian have made substantial advances in recent years in understanding the linguistic patterns of Persian, identifying the characteristics of Persian language learners, and developing the Artificial Intelligence tools for studying authentic Persian language material, these insights have not been integrated within the language teaching curriculum. This chapter introduces Data-Driven Pedagogical Linguistics (DDPL), a collaborative program of practice for the Persian language, which advocates for the integration of linguistic theory, formal language acquisition research, and Artificial Intelligence-driven approaches within the Persian language teaching curriculum.
By focusing on Persian compound verbs as a practical case study, the chapter highlights how research insights and computational approaches can inform language pedagogy, by enabling learners to better understand language patterns and related nuances in meaning that are not typically explained in traditional grammars or taught in language textbooks and utilizing technology to further increase the learners’ exposure to authentic material. DDPL is a collaborative and multidisciplinary practice that establishes a cohesive, linguistically-informed, and student-centered language teaching methodology for the Persian language.

Automated extraction of substance use and co-occurring disorders from probation records (2023)

2023.

Linguistic competence of Persian heritage versus second language speakers (2020)

2020

View All Publications