Feature Extraction of Onomatopoeia Considering Phonotactics for Automatic Semantic Usage Classification of Onomatopoeia

抄録

Onomatopoeia is a language that imitates sounds and state of things. In particular, Japanese onomatopoeia has the merit of being able to communicate intuitively and concisely. On the other hand, onomatopoeia is a language that is hard to understand for people who do not speak Japanese natively. Therefore, this paper aims at estimating the meaning of onomatopoeia to support Japanese learners. Towards this end, I design a learner that automatically classifies Japanese onomatopoeia into their semantic usage. The input feature of this learner is extracted by focusing on the sound symbolism which is assumed to be the most important property of onomatopoeia. Therefore, this paper proposes a method to extract features of onomatopoeia considering their phonotactics. In the evaluation experiment, the efficiency of the proposed method is demonstrated by the fact that the precision of the proposed feature classifier exceeded those of the existing research.