Triangulating the Early Language History of Japan

The genetic relationship of the Japanese language to other areal languages has been subjected to many scholarly and less-than-scholarly theories. To this day, there is little in the way of a consensus, other than that Japanese and Ryūkyūan are related. Much of this previous work has been hampered by the lack of an accurate picture of what earlier states of the languages of Japan were like. Part of the reason for this is a general underappreciation of the importance of the languages of the Ryūkyūan archipelago in elucidating the history of the Japanese language. Motivated by nationalistic concerns, some scholars have designated the languages of the Ryūkyūs as hōgen (dialects), and this often has the unintended consequence of lowering people's expectations of the significance of these languages. The other problem has been a lack of appreciation for evidence gleaned from internal reconstruction (IR) of Japanese, where "synchronic, language-internal variation" is reduced "to an earlier, prehistoric stage of invariance."¹ Thus, alternations such as ki "tree" versus kokage "shade of a tree," or ame "rain" versus harusame "spring rain," demonstrate that it behooves us to take care to understand the historical nature of these internal changes before we begin comparing Japanese to other languages.

It is therefore exciting to have a flood of new scholarship on the language history of Japanese within the course of roughly 12 months. The volume edited by Bjarke Frellesvig and John Whitman, Proto-Japanese, is a compilation of cutting-edge research by a variety of scholars who tackle numerous issues. The work is organized into four sections: reconstructing the basic phoneme inventory, use of dialects in reconstruction, reconstructing accent, and reconstructing morphology and syntax.

J. Marshall Unger's monograph, The Role of Contact in the Origins of the Japanese Language and Korean Languages, reassesses several theories that have been put forth in recent years arguing against the theory that Japanese and Korean have a common origin. He first analyzes Christopher Beckwith's theory that the languages of Koguryŏ and Japan are genetically related and then looks into Alexander Vovin's new theory which argues that Japanese and Korean are not related—most of scholars' proposed cognates are borrowings. Unger presents evidence arguing that abandoning the proto-Japanese-Korean hypothesis is hasty and creates more problems than it solves.

One of the nagging problems in reconstructing earlier states of Japanese, and comparing these with surrounding languages, has been how many proto-vowels to posit. Earlier work by Roy Andrew Miller, John Whitman, and others has argued for four vowels; Marc Miyake's work argues for six vowels.² The chapter coauthored by Frellesvig and Whitman in their edited volume offers evidence that seven vowels should be reconstructed for proto-Japanese (PJ). Of primary importance are the reconstructions of *e and *o, the two vowels most commonly dropped from the four-vowel hypothesis. Frellesvig and Whitman reconstruct both vowels, primarily on Ryūkyūan data: in many dialects in the various islands of the Ryūkyūs, an earlier ki palatalized to something like t∫i (written as či), while earlier ke simply underwent vowel raising and became ki. Thus a word like "tree" is reconstructed as proto-Ryūkyūan *ke (cf. mainland Japanese ki) or "wound" as *kezu (cf. mainland Japan kizu). Hence, this new hypothesis proposes that a number of earlier vowels raised and merged with other vowels, or pushed mid-vowels higher, and this push-chain eventually resulted in the loss of vowels, yielding the five-vowel system of modern Japanese.

The somewhat controversial proposal in this Frellesvig and Whitman essay is the reconstruction...