Research Overview
My research bridges two worlds: the formal study of language structure and the applied challenges of multilingual and culturally responsive AI.
On one side, I investigate the architecture of grammar — the primitive relations underlying complex predicates, morphology, event structure, and meaning. On the other, I develop computational approaches to multilingual NLP, narrative and event analysis, social and scientific knowledge discovery, and AI-assisted language learning, with particular attention to Persian and other underrepresented languages.
Across these directions runs a common question: how does language organize meaning, and what would it take for artificial systems to represent that complexity more faithfully?
This broader research direction is deeply influenced by the work of Jean-Roger Vergnaud, whose view of grammar as a system built from abstract primitives, structural relations, and event representations continues to shape my thinking. His ideas raise a fundamental question that guides my work: what is the underlying architecture of language, and how might it inform the future of linguistically grounded AI?
The research programs below reflect the main directions through which I pursue these questions.
Research Programs
The Architecture of Language
This research program investigates the structural foundations of human language: how grammatical structure gives rise to meaning.
My work focuses on the syntax–semantics interface, complex predicates, event structure, morphological composition, and cross-linguistic variation, particularly in Persian and Armenian. The larger goal is to identify the primitive relations and representations that underlie linguistic structure and to develop formal and computational models of how language encodes events, participants, and meaning.
Application areas
-
Computational modeling of grammar
-
Cross-linguistic analysis of complex predicates
-
Persian and Armenian linguistic analysis & resources
-
Linguistically informed NLP architectures
Linguistically Grounded AI
This program asks how linguistic theory can improve the way AI systems represent, analyze, and generate language.
Modern language models achieve remarkable performance through large-scale statistical learning, but their success does not eliminate questions about linguistic structure. This research examines whether explicit representations of morphology, syntax, semantics, event structure, and cross-linguistic variation can improve the accuracy, consistency, interpretability, and portability of AI systems — especially for multilingual and low-resource languages.
Application areas
-
Multilingual natural language processing
-
Low-resource language technologies
-
Linguistic evaluation of large language models
-
Cultural and pragmatic evaluation
-
AI-assisted language learning systems
-
Modular and hybrid NLP architectures
Language in Social Systems
Language does more than convey information. It shapes narratives, identities, ideologies, communities, and collective behavior.
This research program uses computational linguistics, narrative analysis, and network methods to examine how meaning is constructed and propagated through social and political systems. Much of the work focuses on multilingual and digital environments, where discourse, influence, and community structure interact in complex ways.
Application areas
-
Social media discourse analysis
-
Narrative analytics and ideology detection
-
Online community and influence networks
-
Computational analysis of political and cultural narratives
-
Multilingual computational social science
The Science of Knowledge
Scientific knowledge develops through networks of researchers, institutions, publications, funding, and ideas.
This research program combines natural language processing, scientometrics, network analysis, and knowledge representation to study how fields emerge, evolve, converge, and reorganize over time. By analyzing large bodies of scientific literature and research relationships, the work aims to make the structure and dynamics of knowledge itself more visible.
Application areas
-
Bibliometric and citation network analysis
-
Mapping research landscapes and emerging fields
-
Tracking intellectual convergence and divergence
-
Knowledge graphs for scientific literature
-
Tools for research discovery and knowledge synthesis
