C.M. Downey

C.M. Downey

@cmdowney.com

Assistant Prof. in Ling & DataSci/AI, University of Rochester Director: ur2nlp lab (ur2nlp.io) NLP for low-resource and endangered languages cmdowney.io

323 followers88 following21 posts

C.M. Downey
@cmdowney.com

Tried in class to roll with the metaphor of RNNs as a mixing bowl, ingredients becoming less distinct as you add more, but proceeded to torture the metaphor until I was explaining biRNNs as a sort of aggregation over alternate worlds of ingredient order.. like actually forget I said any of this

C.M. Downey
@cmdowney.com

I'm recruiting Linguistics PhD students! Come work with me on all things multilingual and low-resource NLP! NLP background welcome, but commitment to scientific study of language is essential. My recent work has included ASR for field linguistics and compact LLMs for heritage languages

sas.rochester.eduGraduate Program
C.M. Downey
@cmdowney.com

Personal and lab website in bio. You can email me, but my responses are aggressively filtered for quality-at-a-glance, since faculty get many AI-generated inquiries. Lack of response *does not* imply you shouldn't apply. Be aware that the university is not currently supporting remote study

C.M. Downey reposted

Boulder NLP
@bouldernlp.bsky.social

Congratulations to @mginn.bsky.social, @covetedfish.bsky.social, Ali Marashian, @mvalentini.bsky.social, @alexispalmer.bsky.social & friends for receiving an Outstanding Paper award at #ACL2026 for the paper "Massively Multilingual Joint Segmentation and Glossing". aclanthology.org/2026.acl-lon...

aclanthology.orgMassively Multilingual Joint Segmentation and GlossingMichael Ginn, Lindia Tjuatja, Enora Rice, Ali Marashian, Maria Valentini, Jasmine Xu, Graham Neubig, Alexis Palmer. Proceedings of the 64th Annual Meeting of the Association for Computational Linguist...
C.M. Downey
@cmdowney.com

More students to brag about: congrats to @beichengbei.bsky.social, @hannahchun8011.bsky.social, and @cguo3.bsky.social, who had their Deep Learning term paper accepted to the Workshop on Computational Models of Narrative! arxiv.org/abs/2604.10786

arxiv.orgDo BERT Embeddings Encode Narrative Dimensions? A Token-Level Probing Analysis of Time, Space, Causality, and Character in FictionNarrative understanding requires multidimensional semantic structures. This study investigates whether BERT embeddings encode dimensions of fictional narrative semantics -- time, space, causality, and...
C.M. Downey
@cmdowney.com

It features 1) The first ASR baseline for Uneme (endangered Edoid lg) 2) A Tone Error Rate (TER) extension to Feature Error Rate (FER), calculated by carefully scoring phonetic feature vectors 3) Evidence that WER misrepresents ASR performance by collapsing feature-level errors into lexical failure

C.M. Downey
@cmdowney.com

I'm recruiting Linguistics PhD students this cycle! You should apply to work with me if you're interested in anything low-resource NLP. Some topics I'm particularly excited about currently:

C.M. Downey
@cmdowney.com

- ASR for extremely low-resource langs: making it useful and practical for language documentation - The role of Linguistic relatedness in transfer learning - "Roll-your-own" modeling frameworks (i.e. community-trained and -maintained solutions) - LMs for ancient/archaic language understanding

C.M. Downey
@cmdowney.com

One week left to apply: I'm recruiting (Ling) PhD students! Come work with me on getting NLP to work for as wide a range of languages as possible! Some perks of UR: - Only have to TA 2 semesters total - At least $3k discretionary funds for e.g. conferences - Easy access to upstate NY nature

You are up to date.