Study on the well-formedness conditions of Japanese Sign Language by the assistance of machine learning

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K02536
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Linguistics
• Research Institution: Toyota Technological Institute
• Principal Investigator: HARA Daisuke 豊田工業大学, 工学部, 教授 (00329822)
• Co-Investigator(Kenkyū-buntansha): 三輪 誠 豊田工業大学, 工学(系)研究科(研究院), 准教授 (00529646)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 日本手話 / 音節 / 音素配列論 / 適格性

Outline of Final Research Achievements

We have made two types of databases, one of which has 2,600 well-formed syllables of Japanese Sign Language(JSL), and the other of which 600 ill-formed syllables of JSL. In the databases, syllables are recorded as strings so syllable-constituting elements. Using the two types of databases thus made both for the inputs to machine learning and for linguistic phonotactic analyses, we have found some combinations of syllable-constituting elements that cause JSL well-formed and ill-formed syllables. We have also found what types of elements are involved in the syllable formation of JSL and also a prototypical combination of syllable-constituting elements for the type III syllable, in which one hand moves while the other hand is still.

Free Research Field

手話言語学