Paper short abstract:

The graphemic nature of the Japanese writing system from a comparative perspective is addressed. Recently proposed criteria of grapheme are too strict for the Japanese writing system. It is proposed that those insufficiency can be solved to allow variation of referencing to the linguistic units.

Paper long abstract:

This paper concerns the graphemic nature of the Japanese writing system (JWS) from a comparative perspective. The recent growing attention to the world’s writing systems demands comparable units in those systems. In his proposal to refine the definition of the term grapheme, Dimitrios Meletis (Writing System Research 11, 2019) argues that it distinguishes meaning, has a linguistic correspondence, and minimal. Accordingly, a German letter consequence ‹ch› comprises a grapheme whilst that of ‹ng› does not, because the former refers to a phoneme /ç/ and the latter is the combination of the already existing graphemes ‹n› and ‹g›.

This view cannot be maintained so long as JWS is considered. One of the critical difficulties is that JWS has an unmanageable amount of orthographic variants, such as variation in okurigana and homophonous kanji possibilities. For example, one may write a verb tori-simaru in several ways such as a more favoured form, 取り締まる and more historical one, 取締る, or one can even write almost in hiragana, 取りしまる. This irregularity raises a question of what the characters 取 and 締 correspond with the language: to, or tori? Or, si or sima? If we allow both to be variants, we encounter another problem that 取 only represents tori in compounds. Similarly, to a lesser extent, though, homophonous kanji possibilities raise another problem. For example, one can write a-u ‘to meet’, as 合う, 会う, 逢う, to name a few. These irregular varieties hinge Meletis’s minimality and linguistic value criteria.

One of the major difference with minimality distinction lies in that the JWS orthographic variation is related to the loaned character kanji and supplied reading in hiragana. The present paper suggests loosening the linguistic value criterion and allowing partial reference to linguistic units. Not only solve this change the graphemic problems in JWS, but also this would work for all the writing systems that inherit morphophonic writing system from their parent writing systems.