Permalink : http://hdl.handle.net/2115/73540

Linguistic data are generally recorded using many "symbols." Therefore, linguistic research based on these symbols is, in practice, confronted with a sort of quantification problem (i.e., to assign appropriate values to these symbols), consciously or implicitly. However, humanities data, including the process of how a researcher in the substantive field records the matter of concern and quantifies those records, contain a complex mechanism in various ways that current statistical methods developed in psychology and science do not necessarily assume. Hence, statisticians addressing humanities data should reconsider the assumptions present in statistical analysis and develop new statistical methodologies, which will relax the present assumption, add some new aspects to the humanities, and bring new insights into the humanities. As its starting point, this paper is an attempt to address this new interdisciplinary research among the humanities and statistics, taking lexicostatistical data in Hattori and Chiri (1960) as an example. The five symbols (i.e., "±", " ◯", "?", "・", and "()"), which Hattori and Chiri (1960) introduced to record some unavoidable uncertainty in the linguistic environment of Ainu in the 1960's, led the author to extend the data type for lexicostatistics and the hypothesis of the ordinal scale in lexicostatistics. The application of statistical analysis with the assumption of the ordinal scale clearly demonstrates the significance and uniqueness in and around Biratori and Samani dialects.