His other main point is to propose an alternative to the "family tree" model of language relationships. It works well for Indeo-European (within limits) and also for the Austronesian languages of the Pacific; but he is sceptical, to put it politely, of Greenblatt's claims to have constructed family trees for the African and Amerindian languages, let alone the pretensions of Nostratic. Surely in most cases where different language groups exist side by side for centuries, it makes at least as much sense to consider a "linguistic area" where neighbouring speakers may steal vocabulary and grammar from each other. His example is Australia, the area he knows best, but I can see relevance for the Albanian / Macedonian / Bulgarian / Romanian relationship which I've always found fascinating. He makes the point that even Proto-Indo-European doesn't appear to have been homogenous - did the instrumental plural end with *-bhis or *-mis ?
Anyway, I found this rather more digestible than dear old C.-J. Bailey's essay collection. Must look out for more on this topic..