Some time ago I promised David a post on Altaic and why I’m sceptical. Here goes.
Firstly, I’m sceptical because Turkic is so messy. There is not a lot of solid Turkic reconstruction that doesn’t involve appeal to shared archaisms, for example. The methodology for the papers I’ve seen on comparative Turkic has been rather hodge-podge, and often an appeal to Old Turkic at the expense of anything else. (I’ve read literature in French, German and Russian but I don’t read Turkish well enough so I’m quite prepared to eat my words on this.)
Secondly, the relationship between Turkic, Mongolian and Tungusic doesn’t seem to be very clear either. I’ve never seen anything that tries to sort out what the areal issues are versus shared inheritance. There’s pronouns and general typological characteristics, and a heap of things which might be loan words.
However, I’m perfectly willing to be persuaded that this is just me not knowing very much about it. (After all, without Barry Alpher’s paper on Pama-Nyungan cognates, that’s about the size of Pama-Nyungan published materials at present.) A minimal Altaic that comprises Turkic, Mongolian and Tungusic seems to be reasonable, if difficult to demonstrate without more detailed a view of areal versus genetic history (which someone may well have done).
The position of Japanese and Korean is different. I’ve been to a number of talks about the relationship between Japanese and Korean and I’ve never been convinced. The “cognates” adduced in favour of the relationship always seem to be concentrated in semantic fields that are highly borrowable, and the grammatical features seem to be nothing more than one would expect from languages which are strongly SOV.
So, that’s why I’m currently extremely sceptical about “Altaic”, particularly the version that includes Japanese and/or Korean, but also to a certain extent the core version too. Let me now go and read the materials that David sent me. I’ll post an update about whether I’ve changed my mind.