Zoorna Institute Presents at NACIL4: Exploring Narrative AI for Persian and Kurdish Languages

Zoorna Institute was honored to present at the fourth North American Conference on Iranian Linguistics (NACIL4), held at the University of Toronto Mississauga this spring. Our talk, entitled “AI-Enabled Narrative Analytics for Persian and Kurdish,” highlighted our latest work on using large language models (LLMs) to analyze narrative structure and temporal reasoning in underrepresented languages.

Narratives—whether found in news articles, memoirs, or oral histories—are central to human communication. Yet they remain one of the most challenging areas for natural language processing (NLP), particularly in low-resource languages like Persian and Sorani Kurdish.

Our goal in this project was to explore whether LLMs could support narrative understanding in these languages without requiring traditional, resource-intensive NLP pipelines. We focused on identifying events, actors, causal relationships, and timelines using prompt-based techniques and multilingual models.

Our results showed that LLMs—when guided with effective prompts—performed surprisingly well, even out-of-the-box, on both Persian and Sorani Kurdish. Notable successes included:

• Identifying implicit and nested events

• Resolving temporal sequences

• Handling compound verbs and indirect expressions

However, challenges remain, particularly in coreference resolution, counterfactual analysis, and consistency across Sorani Kurdish inputs.

This research supports Zoorna Institute’s broader mission: building inclusive language technologies that reflect and respect linguistic and cultural diversity.

Narrative analytics has applications in areas like human rights monitoring, education, and AI literacy, particularly in regions where underrepresented languages intersect with geopolitical sensitivity.

We’re continuing this work by expanding our annotated dataset for Sorani Kurdish and developing hybrid approaches that combine the strengths of LLMs with traditional linguistic insights.

Download presentation

💡 Interested in collaborating or learning more? Contact us or sign up for our newsletter to follow the project.