About

I’m Chihiro Taguchi, a full-time language lover. I’m a PhD student in Computer Science and Engineering (CSE) at the University of Notre Dame, USA, as part of the Natural Language Processing group led by Dr. David Chiang.

I am interested in human language and computation. How are we able to use language, what are the underlying structure of language, and how can we model it computationally/theoretically? I have also been working on documentation of Kichwa (Ecuador) and Ikema Miyakoan (Japan).

My name in Japanese is 田口智大, and roughly speaking, it’s pronounced as /tágɯ̀t͡ɕì t͡ɕìçíɾó/ (á: high pitch, à: low pitch).

Quick links

Interests

Linguistics
- Turkic (Tatar), Quechuan (Kichwa), Japonic (Japanese)
- Syntax (Minimalism, Lexical-Functional Grammar, Construction Grammar)
- Semantics (Discourse Representation Theory)
- Language documentation
- Anything about languages
Natural Language Processing & Computational Linguistics
- Speech recognition
- Machine translation
- Large language model

Thoughts

News

(June 3, 2026) My first-author paper Pretrained self-supervised speech models can recognize unseen consonants” has been accepted at Interspeech 2026.
(April 7, 2026) Our paper “Creating ConLangs to Probe the Metalinguistic Grammatical Knowledge of LLMs” has been accepted at ACL2026.
(March 16, 2026) I started an internship at Sakana AI (again).
(February 15, 2026) Our paper “Automatic Speech Recognition for Documenting Endangered Languages: Case Study of Ikema Miyakoan” has been accepted at LREC2026.
(February 6, 2026) I passed the Oral Candidacy Exam and am now a PhD candidate!
(January 9, 2026) Happy New Year! I won the Shared Task “Mozilla Common Voice Spontaneous Speech ASR” in the “Best Improvement over Baseline” subtask.
(August 21, 2025) Two of my first-author papers have been accepted at EMNLP 2025 Main! Efficient Context Selection for Long-Context QA: No Tuning, No Iteration, Just Adaptive-k and Languages Still Left Behind: Toward a Better Multilingual Machine Translation Benchmark.
(July 7, 2025) I am participating in the Linguistic Society of America’s 2025 Summer Institute with the support of the General Institute Fellowship!
(May 19, 2025) I started an internship at Sakana.ai
(March 11, 2025) I presented a poster about machine translation for the Jinghpaw language at the 2025 Annual Meeting of the Association for Natural Language Processing in Nagasaki.
(March 7, 2025) I presented our work “Digital documentation for diasporic data: Challenges, opportunities, and solutions for working with diaspora communities” at ICLDC 2025
(January 22, 2025) Our paper “A Soft and Fast Pattern Matcher for Billion-Scale Corpus Searches” has been accepted at ICLR 2025.
(January 13, 2025) I started my internship at Megagon Labs.
(October 4, 2024) I was awarded the Lucy 2024 Societal Impact Award!
(September 11, 2024) I gave a talk at the NLP Colloquium on my ACL 2024 paper.
(August 14, 2024) At ACL 2024 I received two awards (Outstanding Paper Award and Senior Area Chair’s Award) for my paper “Language Complexity and Speech Recognition Accuracy: Orthographic Complexity Hurts, Phonological Complexity Doesn’t”!
(August 14, 2024) I gave poster and oral presentation at ACL 2024! Check out our paper: arXiv
(July 7, 2024) I completed my first fieldwork in Quito, Ecuador, to research the Kichwa language. The fieldwork involved collecting spoken Kichwa data and analyzing Kichwa morphosyntax.
(June 21, 2024) I presented my (previously presented) work “Killkan: The Automatic Speech Recognition Dataset for Kichwa with Morphosyntactic Information” at AmericasNLP 2024 hosted at NAACL 2024!
(May 22, 2024) I presented my work “Killkan: The Automatic Speech Recognition Dataset for Kichwa with Morphosyntactic Information” at LREC-COLING 2024! [Kichwa ASR Model] [Killkan Dataset] [Code]
(May 16, 2024) Our paper “Language Complexity and Speech Recognition Accuracy: Orthographic Complexity Hurts, Phonological Complexity Doesn’t” was accepted at ACL 2024!
(May 15, 2024) Our paper “Non-discourse-configurationality in Imbabura Kichwa” has been published in the Proceedings of the Linguistic Society of America. [Paper]
(April 21, 2024) Our joint work “Japanese Rule-based Grapheme-to-phoneme Conversion System and Multilingual Named Entity Dataset with International Phonetic Alphabet” has been accepted at SIGMORPHON 2024 at NAACL 2024!
(April 16, 2024) I presented my speech-to-IPA project at the Midwest Speech and Language Days at the University of Michigan, Ann Arbor.
(April 13, 2024) Alianza Quechua 2024 tantanakuypi, kichwa shimipak “rimashkata killkak anta”ta ruray proyectomanta rimarkanimi. Presenté mi trabajo sobre el reconocimiento automático del habla para el idioma kichwa en la reunión anual de la Alianza Quechua. [Diapositivas]
(April 2, 2024) I was awarded the Kellogg Institute Graduate Student Research Grant!
(April 2, 2024) Our joint work “Strategies for the Annotation of Pronominalised Locatives in Turkic Universal Dependency Treebanks” has been accepted at the Joint Workshop on Multiword Expressions and Universal Dependencies at LREC-COLING 2024!
(March 14, 2024) 言語処理学会2024にて口頭発表しました。スライド
(February 20, 2024) Our paper “Killkan: The Automatic Speech Recognition Dataset for Kichwa with Morphosyntactic Information” has been accepted at LREC-COLING 2024! [Preprint]
(February 20, 2024) Our joint work “J-SNACS: Adposition and Case Supersenses for Japanese Joshi” has been accepted at LREC-COLING 2024!
(January 20, 2024) Our joint project “Unifying the Annotations in Turkic Universal Dependencies Treebank” has been accepted at UniDive 2nd meeting.
(January 10, 2024) Hice una presentación titulada “El conjunto de datos de reconocimiento automático de voz para kichwa” en el Taller Tecnologías Digitales y Lenguas Indígenas.
(January 7, 2024) I presented a poster titled “Non-discourse-configurationality in Imbabura Kichwa” at the Linguistic Society of America 2024 Annual Meeting. This is a joint work with Jefferson Saransig.
See here for the news in 2023.
See here for the news in 2022.

Contact

Notre Dame: c[ my surname ] at nd.edu

itscalledhymmnosfyi