HUSCAP logo Hokkaido Univ. logo

Hokkaido University Collection of Scholarly and Academic Papers >
Graduate School of Humanities and Human Sciences / Faculty of Humanities and Human Sciences >
北方言語研究 = Northern Language Studies >
第10号 >

Reconsideration of “Major Division” of Ainu Dialects : A Statistical Reanalysis of Asai (1974)

Files in This Item:
14_231_254.pdf929.04 kBPDFView/Open
Please use this identifier to cite or link to this item:

Title: Reconsideration of “Major Division” of Ainu Dialects : A Statistical Reanalysis of Asai (1974)
Authors: Ono, Yohei Browse this author
Keywords: Ainu
Graph Theory
Spectral Clustering
Issue Date: 20-Mar-2020
Publisher: 日本北方言語学会
Journal Title: 北方言語研究
Journal Title(alt): Northern Language Studies
Volume: 10
Start Page: 231
End Page: 254
Abstract: The main objective of this studyis to reconsider Asai’s (1974) “Major Division” of Ainu dialects from the viewpoint of statistical methodologies. “Major Division”, which classifies Ainu dialects into the Hokkaido Ainu dialects, North Kuril Ainu dialect, and Sakhalin Ainu dialects, greatly influences current Ainu linguistics. Since there do not exist any other studies where Ainu linguists demonstrate the cognacy judgments among these Ainu dialects, Asai’s(1974) studyis still a landmark in Ainu dialectology. However, as a statistician, I was confronted with a question regarding the statistical methods used in Asai (1974). There were other clustering methods considered asable to produce more reliable classifications than “Large method”and“Small method” in Asai (1974), which correspond to thesingle linkage method and complete linkage method (Sørensen 1948), respectively. As I demonstrate in this study, Asai’s(1974: 92; Table 1) data contain only information on the similarity among the Ainu dialects but do not preserve the information on the dissimilarity among the Ainu dialects, oreven the order of dissimilarity. Therefore, these statistical properties of Asai (1974: 92; Table 1)violate the assumptions of classical statistical methods, including metric and non-metric Multidimensional Scaling. Furthermore, several descriptions in Asai (1974) indicate that Asai himself noticed the particularity of his data. This study proposes an alternative approach: Spectral Clustering, a typeof graph-theoretic method that can be applied to the data structure present in Asai (1974). The main results of this study demonstrate that the “Major Division”in Ainu dialects needs to be reconsidered, so far as researchers recognize the classification of Ainu dialects based on Asai’s (1974)clustering results. Rather, the Ainu dialects can be classified into two groups: Hokkaido Ainu dialects and Sakhalin Ainu dialects. Furthermore, theNorth Kuril Ainu dialect belongs to the northeastern Hokkaido Ainu dialect group in the strongest structure and to southwestern Hokkaido Ainu dialect group in the second strongest structure among the Hokkaido Ainu dialects. The statistical finding on the North Kuril Ainu dialect agrees with the philological research in both Murayama (1971) and Satō and Bugaeva (2019). These facts suggest thatAsai’s (1974) “Major Division”resultedin a great influence on current Ainu linguistics, beyond Asai’s initial purpose.
Type: bulletin (article)
Appears in Collections:北方言語研究 = Northern Language Studies > 第10号

Export metadata:

OAI-PMH ( junii2 , jpcoar_1.0 )

MathJax is now OFF:


 - Hokkaido University