Проблемы китайского и общего языкознания. К 90-летию С. Е. Яхонтова

 480  George van Driem   Starosta’s evidence is meagre, yet primarily morphological in nature. Languages change so fast that traces of a genetic relationship between two languages are either obliterated or obscured beyond recognition after about a dozen millennia, give or take a few thousand years. Regular phonological correspondences and a common morphological system are the most compel- ling types of evidence for a genetic relationship between languages. When the time depth of a linguistic phylum is very great, morphological correspon- dences may be the only remaining vestige that evinces a genetic relationship. Despite a scarcity of lexical correspondences between Itelmen a.k.a. Kamchadal and Chukchi-Koryak, the inclusion of Itelmen within a Luoravet- lan a.k.a. Chukotko-Kamchatkan family is based on formal and semantic similarities in a small number of flectional morphemes. Yet such morphologi- cal evidence would necessarily be lacking altogether if the proto-language of a given family just happened to have been typologically analytical, like Mandarin is today. Most historical linguists modestly resign themselves to a maximal time depth beyond which the comparative method is unable to distinguish between correspondence and coincidence and therefore unable to establish a genetic relationship between languages. Yet a few linguists strive to gaze beyond the linguistic event horizon. [Nichols 1986, 1992, 1995, 1998] attempts to salvage empirical evidence for deep genetic relationships between linguistic phyla and to detect the vestiges of ancient substrate influence exerted by one language phylum on another in the form of typological diagnostics. She argues that languages which are distant on a morphosyntactic spectrum ranging from wholly head- marking to wholly dependent-marking are unlikely to be genetically rela- ted. Nichols’ diagnostics embolden her to speculate about ancient linguistic spread zones and bottlenecks. However, many of the world’s languages ex- hibit both head-marking and dependent-marking morphology and are thus situated somewhere in the middle of the spectrum. Sino-Austronesian has evolved since Sagart first explained his theory to me at the Univer- sity of Hawai‘i in August 1989. After abandoning his pre-1994 position that Sinitic was unrelated to Tibeto-Burman, Sagart was compelled to adhere to the outdated and empiri- cally unsupported Sino-Tibetan family tree model, which he requires as an ingredient for his model. Sagart identifies the Middle Yǎngsháo culture in the 5th and 4th millennia bc as the Sino-Austronesian homeland based on ‘regular correspondences’ in the four words for pig, rice, net and millet, one of which is not reflected in Sinitic. The kindest assessment of Sagart’s theory in print [van Driem 2005] evaluates his evidence as failing to meet the conventional standards of proof. Sino-Austronesian does not merit serious consideration as a hypothesis about linguistic phylogeny. Sagart’s arguments for his Sino-Austronesian hypothesis have now been totally demolished by [Blust 2014].