Investigating Wakun “Miru” #

Ikeda Shoju

January 4–11, 2025

Summary #

This page is an English-language summary of 和訓「ミル」の調査, a Japanese case study investigating attestations of the wakun “miru” (ミル) in the Kanchi-in manuscript of the Ruiju Myōgishō (KRM). The full account, including the character-by-character evidence, is available only in Japanese; this page presents only the abstract.

This note uses the wakun “miru” (ミル) as a test case to examine the reliability of the hyōki (表記, “orthographic forms”) column in the Nihon Kokugo Daijiten, 2nd edition (Nikkoku). After tabulating the source editions and dates of the 17 pre-modern dictionaries cited in Nikkoku’s own dictionary-source column, I confirmed that 15 of them, including KRM, do record “miru,” but found that the Shinsen Jikyō — despite being listed — does not actually attest the reading, an error in Nikkoku’s source citations. Nikkoku’s hyōki column lists 104 attestations for “miru” (103 single-character, 1 compound), of which 53 characters (about half) were entered as bitmap images in the 2007 JapanKnowledge edition because they could not be input in Unicode at the time; nearly all can now be represented in Unicode. Checking these against the Kanchi-in manuscript’s text turned up seven apparent input errors (e.g., 呪 should be 眖) and one omission — 瞩, which does appear in KRM but is missing from the hyōki column. Cases of apparent confusion between graphically similar characters, such as 覞 and 䁤, were also examined individually.

Surveying, in order of appearance, the characters in KRM bearing “miru” that are not listed in Nikkoku’s hyōki column turned up over 100 characters: 51 have textual grounds confirmable in dictionaries such as the Shuowen Jiezi, the Tenrei Banshō Meigi, and the Guangyun, and could reasonably be added to the column; 11 appear to be misreadings caused by graphic confusion with similar characters; and no textual basis could be identified for the remaining 32. A comparison with the Ekirin-bon Setsuyōshū also found that 9 of Nikkoku’s compound-word attestations do not appear in that dictionary’s “Bi” (美) and “Genji” (言辞) sections.

In conclusion: (1) the hyōki column’s entries generally have reliable semantic correspondence; (2) it nonetheless contains outright errors (6 characters); (3) it has room for defensible additions (about 50 characters); (4) some entries are misreadings from graphically similar characters (11); and (5) some entries have no identifiable textual basis (32). Identifying character forms and judging the correctness of glyph input are unavoidable tasks in annotating KRM; the supporting evidence for each case is recorded in the remarks column of KRM_wakun.tsv.