Matthew Nazari

matthewnazari@college.harvard.edu

+1 (408) 438-8442

I recently graduated Harvard University where I studied computer science and the Middle East. I am interested in language technologies and society. I am currently working on a dialect atlas of the Northeastern Neo-Aramaic dialects.

en aten surayet baba qa mu aha beqrayet khush men lakha. aha akhchi ctivon qa rabbiyye qat bayyi khazzi qat ana raba gura nasheven.

Updates

Publications

  1. G. Khan, M. Nazari, et al. The North-Eastern Neo-Aramaic Dialects. Studies in Semitic Languages and Linguistics. Leiden-Boston: Brill, 2025. (in preparation)
  2. M. Nazari. “‘No, that’s my fault. I can’t read.’: Understanding young speakers of Arabic and Armenian experiences using language models.” 2025. (in preparation)
  3. M. Nazari. “NENAMorph: Creating a morphology dataset of 56+ million inflections without webscraping.” In ACL, 2025. (submitted)
  4. D. Ley*, L. Tang*, M. Nazari*, H. Lin, S. Srinivas, H. Lakkaraju. “Consistent explanations in the face of model indeterminacy via ensembling.” ICML Workshop on Interpretable Machine Learning in Healthcare, 2023. [arXiv, code]
  5. H. Lin*, M. Nazari*, D. Zheng*. "PCTreeS: 3D Point Cloud Tree Species Classification Using Airborne LiDAR Images." AAAI Workshop on AI for Social Good, 2023. [arXiv, code]
  6. M. Nazari. “NoLoR: An ASR-based framework for expedited endangered language documentation.” AAAI Workshop on AI for Social Good, 2023. [arXiv]

Community Projects

I am developing an online learning space for several Northeastern Neo-Aramaic (NENA) dialects at shlama.io. I am slowly uploading materials such as textbook lessons and machine learning models.

I developed NENAMorph, a dataset of so far 56 million inflections capturing village-level morphophonological microvariations across several NENA dialects.

I developed NENACrowdsource, a crowdsourcing project to curate datasets for developing language technologies for our communities.

Teaching