MORITA Takashi

Starting year 2025

Chubu University
Academy of Emerging Sciences / Center of Mathematical Science and Artificial Intelligence
Designated Senior Assistant Professor

Research fields

Computational Linguistics
Computational Cognitive Science

Research Interests

Unsupervised Learning
Vocalization
Ethology
Data Science

Main research topics

Dr. Takashi Morita is a computational linguist, studying computational models of human language learning. His research primarily focuses on the unsupervised learning of speech recognition. Current industry-oriented machine learning methods rely on text-annotated speech recordings for model training. However, this approach is implausible as a cognitive model of spoken language acquisition because humans acquire their native spoken language before learning to read and write. Moreover, writing systems were developed relatively recently compared to the evolution of spoken languages, and some existing languages stil lack orthography. Given these facts, a cognitively plausible model of speech recognition should be capable of inferring a consistent symbolic representation of speech from auditory data alone, despite substantial variations among and within individuals.

Dr. Morita is also engaged in a machine learning-based analysis of animal vocalizations. Similar to unwritten languages, text annotations for animal vocalizations are inherently unavailable. Nonetheless, categorical representations of these sounds are essential and useful for various statistical analyses, leading researchers to classify animal vocalizations based on auditory/visual inspection. Such manual classifications, however, are often criticized for their lack of objectivity and reproducibility. Dr. Morita is addressing these concerns by employing unsupervised machine learning techniques, yielding statistical analyses of animal vocalizations without relying on manual annotations.

 

Representative papers

Takashi Morita & Timothy J. O’Donnell. “Statistical Evidence for learnable lexical subclasses in Japanese”, Linguistic Inquiry, 53 (1), pp. 87-120, 2022.

Takashi Morita, Hiroki Koda, Kazuo Okanoya, & Ryosuke O. Tachibana. “Measuring context dependency in birdsong using artificial neural networks”, PLOS Computational Biology, 17 (12), p. e1009707, 2021.

Research URL

Researchmap　https://researchmap.jp/takashi_morita

Interview

  • No interview

News

  • No related news

Back