Andrew Haynes
Howdy, I'm Andrew! I'm a freshman studying Linguistics and Computer Science at Princeton. Broadly, I'm interested in low-resource and endangered languages, multilinguality, as well as bringing linguistic theory and psycholinguistics more directly into computational linguistics and natural language processing.
I also love learning languages: I speak English (Native), Japanese (Heritage/N2), Mandarin (Intermediate), and Korean (Intermediate-Mid/ACTFL)!
If you're working on something related, I'd love to hear from you!
✉️ andrewhaynes [at] princeton [dot] edu github linkedin instagram youtube substack
news
May 05, 2026
Excited to share that I'll be presenting my first paper on low-resource OCR at ComputEL-9 @ ACL 2026 in San Diego!
publications
CoRSAL-OCR: Evaluating Zero-Shot OCR for Language Archive Materials
Luke Gessler and Andrew Haynes. 2026. In Proceedings of the Ninth Workshop on the Use of Computational Methods in the Study of Endangered Languages (ComputEL-9), pages 125–135, San Diego, California, USA. Association for Computational Linguistics.
Language archives contain valuable linguistic materials that are undigitized and therefore difficult to access. Modern optical character recognition (OCR) systems have great potential to make these collections more accessible, but there are few system evaluations which can assess the quality of an OCR system specifically for language archive materials. We present CoRSAL-OCR, an OCR evaluation dataset of over 200 document pages with gold-standard transcriptions from two South Asian languages: Bodo (written in Devanagari) and Garo (written in Latin script). Using this dataset together with the 8-language AILLA-OCR benchmark, we evaluate four OCR systems: Tesseract, Google Cloud Vision, Gemini 3 Flash, and Qwen3.5-27B (an open-weight model). We find that vision language models (VLMs), when given appropriate prompts, achieve the lowest error rates on these datasets. However, prompt design has a large effect on VLM performance, with a detailed generic prompt reducing CER by up to six-fold compared to a minimal prompt. We release our dataset at github.com/larc-iu/corsal-ocr to support further research on OCR for language archives.
gallery
