While I do hope to learn Japanese one day to read the works of such great novelists as Kawabata and Mishima, I doubt I will ever be more than the most casual of dilettantes in Japanese historical linguistics. That said, I do like skimming books on the field, and have heard enough of Roy Andrew Miller’s view on the language in the context of general Altaic studies.
One reader has recently brought to my attention a fairly new view on the history of Japanese, set out by Christopher I. Beckwith in his book Koguryo: The Language of Japan’s Continental Relatives (Leiden: Brill, 2004). I haven’t actually read the book yet, as it is not in the university library in Helsinki, and I can’t afford a typical Brill hardcover, and though the book is completely readable for free at Google Books, it hurts my eyes. Nonetheless, there is a substantial review of the book on the web at the blog Néomarxisme. In introducing us to this language which may possibly be related to Japanese, the reviewer writes:
From around 100 B.C. to the 7th century A.D., modern day Korea was divided into three kingdoms: Koguryo, Shilla, and Paekche. The three states were eventually unified under Shilla in 668, and the modern Korean language originates from the language spoken in Shilla. Koguryo and Paekche, however, had different languages which are posited to be related to each other. Scholars thus make two groupings of Korean peninsula languages: the Han languages — spoken in Shilla and among the subjugated class in Paekche — and the Puyo-Koguryoic languages of Koguryo, Puyo (another Northern Korea state), and Paekche’s ruling class. The latter family is now totally extinct and probably made a minor impact on modern Korean. The lack of written records and remaining vocabulary items from these languages make it difficult to learn much about the nature of the “Koguryoic” family.
I’ve encountered two separate complaints about unreliable reconstructions in Beckwith’s book. On an Amazon review, one Asier Gabikagojeazkoa writes:
But Beckwith has made some considerably big mistakes in his comparative Japanese-Koguryo work (how could possibly yama, mountain in Japanese, and ɣapma ‘big mountain’ in Koguryo, be related? The Proto-Japanese word for mountain is *dama, as every y- in Japanese comes from d-, as noted in some Ryukyuan dialects).
John R. Bentley at the Northern Illinois University writes:
Beckwith reconstructs Proto-Japanese-Ryukyuan ‘eye’ as *mika or *miak (2004:157-58). This is based entirely on the Hateruma island word miŋ ‘eye’. He sees the velar nasal going back to a velar -k. This is then compared with Old Chinese *mek ‘eye’ and Old Tibetan myig ‘eye’. Gary Oyler, in 1997, did an MA thesis on the problem of -ŋ in Hateruma, and concluded that the -ŋ is secondary. It only occurs word finally, and there is no pattern to which words have the ŋ and which do not. It is completely random. My own work has found this same phenomenon in Yonaguni, an island not far from Hateruma, but the ŋ is attached to different nouns than those in Hateruma (so the development of this nasal was independent on the two islands). This velar nasal is simply a relic of morphology that the speakers have reanalyzed as part of the noun, kind of like American speakers spelling ‘hafta’ (< ‘have to’), where they treat two original words as one. It cannot be reconstructed as part of the proto-form. The true etymology of ‘eye’ would be *ma-i or perhaps *ma-Ci. So Beckwith also has the vowels wrong.