Food and Environment in Early and Medieval China. E. N. Anderson

to the sea, the Austronesians exploded over the vast realms of the Pacific and Indian Oceans. Today, somewhat closely related Austronesian languages are spoken from Madagascar to Hawai‘i and from New Zealand to Micronesia. Wherever these people went, they took agriculture. Cognate words for chickens, coconuts, root crops, and dozens of other agricultural items and techniques are found all over their vast realm, indicating that the early Austronesians had all these things.

      It occurred to Bellwood, and to other scholars, that other linguistic spreads might also be associated with agriculture. This has been the subject of much research, culminating in a volume edited by Bellwood and Colin Renfrew (2002). Independent genetic evidence confirms that many migrations occurred and that agriculture released spectacular demographic expansions. Agricultural peoples moved rapidly from the Near East through Europe and into Africa, expanding in numbers as they did so. Evidence from Europe, Southeast Asia, and western Africa confirms striking demographic expansions directly after the introduction of agriculture in these regions (Gignoux et al. 2011). Farmers multiplied fast and moved out to new lands. Local people were not wiped out but rather merged with the expanding farmers, leaving varying degrees of genetic admixture.

      Of course, not all linguistic spreads were accompanied by farming. The Inuit, Athapaskans, and several other hunting peoples spread over thousands of miles without benefit of agriculture. Moreover, having agriculture does not guarantee spread; the Georgian-language speakers have probably had agriculture almost since it began, 11,000 years ago, but have remained confined to a tiny area in the Caucasus. Within eastern Asia, the Yao-Mian phylum has recently spread from southeast China into Southeast Asia, but before that it seems to have been narrowly confined to a small part of south China. The Miao-Hmong phylum started in northwest China, according to some Miao myths. It survives in central and south China, with recent radiation into northern Southeast Asia. It has certainly spread with agriculture but has never gained much territory.

      But some groups do spread. Bellwood and others have made a very convincing case for the association of the Tibeto-Burman (Sino-Tibetan) language phylum with the spread of millet agriculture. The dates and geography make this seem reasonable. The Tibeto-Burman languages, including the ancestor of the Chinese languages of today, are about as different as you would expect if they branched off from each other 7,000 or 8,000 years ago. I find the association convincing, but it is controversial. G. Van Driem thinks the stock originated in Sichuan (van Driem 1999, 2002). Others (myself included) think it originated further north but then differentiated in Sichuan. Either way, the stock originated very close to the point of origin of millet agriculture.

      The spread of the Thai-Kadai phylum is clearly associated with the spread of rice agriculture (Bellwood and Renfrew 2002). We know that Thai-Kadai languages diversified in, and probably spread from, the Yangzi Valley area, where rice was domesticated. Their routes of spread and the probable timing of the spread fit well with the spread of rice agriculture south and southwest. The Austronesian phylum was associated with rice agriculture early and has some very Thai-sounding words; it may be related to Thai-Kadai (P. Benedict 1975), or, more likely, it simply may have become connected with rice agriculture and a few loanwords in very early times. The Thai-Kadai languages branched from each other perhaps 6,000 years ago. Their speakers were, however, probably not the only rice-growers, and Hmong/Miao and Yao/Mien languages were in the right area then, too, and have been associated with the spread of rice by some scholars.

      A significant fact is the spread of the Thai root for “chicken,” kai. This word was borrowed into Chinese early, becoming ji in Mandarin but remaining kai in Cantonese. (The Cantonese language is likely the result of Thai speakers switching to Chinese in the Tang Dynasty and since. The Cantonese word for “chicken” is far from the only Thai-sounding word in that language.) Not stopping there, kai went on—increasingly distorted—into Korean, Japanese, the Central Asian languages, and thence into the Western world, eventually as far afield as Morocco (Blench 2007). It is awfully hard to escape the conclusion that the Thai peoples domesticated the chicken, which is native to south China and Southeast Asia. Borrowed words surely indicate borrowed chickens. Other local peoples in Southeast Asia, such as the Austronesians, have their own words for the bird, implying that they were aware of wild chickens before domestication.

      Bellwood’s correlation of advanced agriculture with the spread of the Austronesian languages in the islands east of Asia is no longer controversial. Millet reached Taiwan by 3,000–2500 BCE; a recent find revealed large amounts of foxtail millet and rice at Nan-kuan-li. This and related sites probably represent the ancestors of today’s Austronesian-speaking “aborigines” of Taiwan, recently arrived from south China with seeds in hand (Tsang 2005). There is clear archaeological evidence for an explosive radiation of advanced farming and pottery-making people from south China to Taiwan and thence to the Philippines and the islands south and east—the lands inhabited by Austronesian peoples today (Bellwood 1997, 2002, 2005; Donohue and Denham 2010 dispute this, but Bellwood has a very effective answer in the commentary section of their article). However, subsequent profound changes in both language and agriculture took place when Austronesians mingled with Papuans in Melanesia (Paz 2002), with the result that Oceanic Austronesian agriculture looks much more Papuan than Chinese.

      When we move to western Eurasia, however, we are in a very different situation. Bellwood and Renfrew hypothesized that the Indo-European (IE) phylum was present at the birth of agriculture in the Near East and spread along with it. This is certainly false. Agriculture in the Near East began at least 11,000 years ago, and the IE phylum is a very close-knit one. Suffice it to say that the Hittite for water is wadar. This and hundreds of other close pairs prove that IE cannot possibly have split up more than about 6,000 years ago. Languages change very fast, especially in the days before books, radio, and television. Languages diverge and differentiate faster than we once thought (Brown 2010), and this process probably happened even faster in preliterate times.

      Moreover, we know that agriculture began in the dry Levantine back country. But the IE phylum has shared cognates for a whole host of cool-temperate plants and animals, including laks for salmon. (No, that word isn’t of Jewish origin.) These biota firmly fix the IE origin somewhere between northeast Europe and the Caucasus—most likely in and around what is now the Ukraine. Conversely, IE significantly lacks words for dry Levantine commodities.

      Also, there is plenty of evidence for pre-IE farmers in Europe. The surviving Basque language is the most obvious piece, but there are also the host of agricultural and rural words in Germanic that have no IE roots: “wheat,” “sheep,” “eel,” “delve,” “roe” (deer), “boar,” and even “land,” among others (Witzel 2006). Greek also borrowed from non-IE languages a whole host of agricultural and settlement words. Speakers of IE languages would hardly have borrowed such words from hunter-gatherers. Spreading in the other direction, IE speakers of the language ancestral to today’s Iranic and north Indian (Sanskritic) languages borrowed a similar large range of words, including terms for camel, donkey, and brick as well as a whole host of religious terms (including names of gods, like Indra) and literary usages (Witzel 2006).

      A recent hypothesis, based on virus epidemiology, has the IE languages originating in Anatolia and spreading with agriculture (Bouckaert et al. 2012). But again the timing is wrong; agriculture had already spread widely by the later, and more believable, timing they reconstruct, and one wonders what happened to the earlier propagators. Viruses do not make a very good model for humans.

      What, then, accounts for the spread of the IE peoples? The traditional explanation was that they developed riding, horse traction, and horse-based warfare (chariots and, probably later, riding). This explanation has received a powerful boost lately from increasingly clear evidence that the horse was domesticated in Kazakhstan around 5,500 years ago (Anthony 2007; Harris 2010)—just east of the place and time reconstructed for the IE homeland. The horse was probably a food animal first. Only after domestication could it be milked and ridden. Anyone familiar with wild equines will know that they would not stand still for either process! Horses, unlike ruminant livestock, are neither stolid nor intellectually limited. They are high-strung, sensitive, extremely intelligent animals, and to this day it takes a tremendous amount of empathy and skill to work with them. Instead of dull servants like cows, they can become super-smart companions. In Mongolia, my wife saw small boys riding bareback,