from their
original homeland of
Taiwan via China to
Mainland Southeast Asia, according to Matthias Gerner's "Northeast to Southwest Hypothesis". James R. Chamberlain (2016) proposes that the Tai–Kadai (Kra–Dai) language family was formed as early as the 12th century BCE in the middle of the
Yangtze basin, coinciding roughly with the establishment of the
Chu fiefdom and the beginning of the
Zhou dynasty. The high diversity of Kra–Dai languages in southern China, especially in
Guizhou and
Hainan, points to that being an origin of the Kra–Dai language family, founding the nations that later became Thailand and Laos in what had been
Austroasiatic territory. Genetic and linguistic analyses show great homogeneity among Kra–Dai-speaking people in Thailand. Although the position of Kra–Dai in relation to Austronesian is still contested, some propose that Kra–Dai and Austronesian are genetically connected. Weera Ostapirat (2005) sets out a series of regular sound correspondences between them, assuming a model of a primary split between the two; they would then be co-ordinate branches. Ostapirat (2013) continues to maintain that Kra–Dai and Austronesian are sister language families, based on certain phonological correspondences. On the other hand, Laurent Sagart (2008) proposes that Kra–Dai is a later form of what he calls "FATK" (Formosan Ancestor of Tai–Kadai) a branch of Austronesian belonging to the subgroup "Puluqic", developed in Taiwan, whose speakers migrated back to the mainland, to Guangdong, Hainan, and north Vietnam, around the second half of the 3rd millennium BCE. Upon their arrival in this region, they underwent linguistic contact with an unknown population, resulting in a partial relexification of FATK vocabulary. Erica Brindley (2015) supports Sagart's hypothesis, arguing that the radically different Kra-Dai history of migration to the mainland (as opposed to the Philippines for Proto-Austronesian) and extended contact with Austro-Asiatic and Sinitic speakers would make the relationship appear more distant. She also suggests that the presence of only the most basic Austronesian vocabulary in Kra-Dai makes this scenario of relexification more plausible. Besides various concrete pieces of evidence for a Kra–Dai existence in present-day Guangdong, remnants of Kra–Dai languages spoken further north can be found in unearthed
inscriptional materials and non-Han
substrata in
Min and
Wu Chinese. Wolfgang Behr (2002, 2006, 2009, 2017) points out that most non-Sinitic words found in Chu inscriptional materials are of Kra–Dai origin. For example, the Chu graph for 'one, once' written as (? 22nɯŋ, Dai 33nɯŋ, Longzhou nəəŋA etc.) 'one, once'. In the early 1980s, Wei Qingwen (韦庆稳), a
Zhuang linguist, proposed that the
Old Yue language recorded in the
Song of the Yue Boatman is in fact a language ancestral to Zhuang. Wei used reconstructed
Old Chinese for the characters and discovered that the resulting vocabulary showed strong resemblance to modern Zhuang. Later,
Zhengzhang Shangfang (1991) followed Wei's proposal but used Thai script for comparison, since this orthography dates from the 13th century and preserves archaisms not found in modern pronunciation. Zhengzhang notes that 'evening, night, dark' bears the C tone in Wuming Zhuang
xamC2 and
ɣamC2 'night'. The item
raa normally means 'we (inclusive)' but in some places, e.g., Tai Lue and White Tai, it means 'I'. However,
Laurent Sagart criticizes Zhengzhang's interpretation as anachronistic, because however archaic the Thai script is, the Thai language was only written 2,000 years after the song had been recorded; even if
Proto-Kam–Tai had emerged by the 6th century BCE, its pronunciation would have been substantially different from Thai. , in the 5th century BC. ==Internal classification==