Towards a Semantic Typology of ‘Eating’, ‘Drinking’ and ‘Smoking’ in Languages of China

Languages all over the world have their distinctive strategies to express the three basic human activities of eating, drinking and smoking (Noonan 1992, Bowden 1997, Aikhenvald 2009, Newman 2009, among others). In our preliminary study of over 160 languages in China, comprising Sino-Tibetan, Tai-Kadai, Hmong-Mien, Austroasiatic, Austronesian and Mongolic-Khitan languages, we have identified five typological patterns (Table 1) regarding the relationship between these three concepts. Type I corresponds to where ‘eat’, ‘drink’ and ‘smoke’ are expressed by different morphemes, exemplified by major northern Mandarin varieties. Type II represents colexification of ‘eat’, ‘drink’ and ‘smoke’, whereby one generic verb (often the word ‘to eat’) is employed, disambiguated by different objects (being solid, liquid or air) that follows the verb. Type III, Type IV and Type V showcase a split between these three concepts, with ‘eat’ and ‘smoke’ sharing the same morpheme in Type III, ‘eat’ and ‘drink’ allocated with the same morpheme in Type IV, and ‘drink’ and ‘smoke’ colexified with the same morpheme in Type V.
In addition to the lexical typological implication, their distribution displays genetic inheritance and areal diffusion . Type I, Type II and Type III are the major types found in Sinitic languages, which demonstrates areal distribution as well. Northern Mandarin varieties, located to the north of the Yangtze River, largely belong to Type I, whereas Transitional and Southeastern Sinitic languages, including many Hui, Wu, Min, Gan, Hakka and Southwestern Mandarins typically showcase Type II. Type III, typologically uncommon cross-linguistically, is mainly seen in Far Southern Sinitic languages like Yue, and some Transitional Sinitic languages such as Hui and Jianghuai Mandarin and a few Jin Chinese, in which ‘eat’ is extended to ‘smoke’ probably via internalization as ‘pleasant inhaling’ Newman (2009: 9). Last but not least, distributions of Type IV and V are observed sporadically in some Min, Ping and Hakka varieties. It is likely that historical strata and language contact play an important role in divergence of the semantic typology on EAT, DRINK and SMOKE in Sinitic languages.
While a large number of Sinitic languages (e.g. Transitional and Southeastern Sinitic) colexify ‘eat’ and ‘smoke’, non-Sinitic languages generally avoid such colexification. In other words, Type II and Type III are not commonly found among the non-Sinitic languages, except for the Tai-Kadai languages. On the other hand, unlike Sinitic languages which seldom colexify ‘drink’ and ‘smoke’ without the link ‘eat’, many non-Sinitic languages directly extends the concept of DRINK to SMOKE, especially in the Tibeto-Burman and Hmong-Mien languages, resulting in their common presence of Type V. Our data also reveal that internal variation is the highest among the Tai-Kadai languages on the co-distribution of the three notions, while it is relative homogenous among the Tibeto-Burman languages.
As a final note, this study pinpoints two linguistic areas with intensive language contact, namely the Qinghai-Gansu sprachbund and the west Lingnan region. Languages belonging to the same phylogenetic affiliation in these two regions, may employ different source lexical items, yielding a highly mixed distribution of the types thereof.