Labeling Algorithms Based on Categorial Feature Sharing and New Approaches to Studies on Compounds

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K12984
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 02060:Linguistics-related
• Research Institution: Hokkaido University of Education (2022-2023); Kochi University (2021)
• Principal Investigator: Sato Ryosuke 北海道教育大学, 教育学部, 講師 (60859685)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 複合語 / 場所句 / 文主語 / ラベル決めアルゴリズム

Outline of Final Research Achievements

In this study, I explored a method of label determination based on categorial features and, as a consequence, provided explanations for various phenomena observed in compounds.
In order to scrutinize the features required for the labeling algorithm, I studied not only compounds but also Locative Inversion constructions and and sentential subject constructions in which prepositions/that-clauses occur at the beginning of sentences.
Using the findings, I conducted a study of Japanese compound nouns and compound verbs. As a result of the study, I found that the approach based on categorial-feature sharing, which I had initially assumed, was not enough and difficult to explain linguistic data, and instead proposed a method of label determination based on chain uniformity.

Free Research Field

統語論と形態論

Academic Significance and Societal Importance of the Research Achievements

本研究の主目的は範疇素性共有に基づくラベル決めアルゴリズムを用いて複合語に見られる諸現象に説明を与えることであったが、その過程で場所句倒置構文や文主語構文に見られる束縛現象（＝代名詞の解釈）についても説明を与えた。
日本語の複合語だけでなく、上記のような英語の構文についてのあらたな事実を解明したことで、日本語学と英語学に貢献しただけでなく、日本語教育と英語教育の両方に貢献できたと言える。