Abstract : Due to the specific sociolinguistic situation in China where very divergent linguistic varieties for a long time have been "roofed" by a common culture and a common writing system, the development of the Chinese dialects reflects a 'complex and intertwined evolu- tionary history' (Hamed and Wang 2006: 31). Chinese dialects are therefore a good test case for methods which no longer take the family tree as their primary evolutionary model. In this talk, we shall introduce and discuss the implications of network approaches by applying them to a recently compiled dataset covering translations of 180 concepts for 40 different di- atopic varieties of Chinese (Hóu 2004). Apart from the fact that cognacy information can be extracted from the data, this dataset has the great advantage of giving full phonetic records of all dialect words. This makes it possible to apply methods which work on cognate sets (Nelson-Sathi et al. 2011) as well as methods for computational sequence comparison (List 2010).