It is doubtful of the possibility of successfully reconstructing Chinese linguistic history strictly from the evidence of modern dialects by the traditional comparative method as applied to languages without written tradition. A major difficulty is that the Stammbaum or branching-tree model that is implied by the traditional comparative method is totally unrealistic in the case of Chinese. Even in remote parts of the country dialects have never developed in isolation. they have been influenced not only by their immediate neighbors but even more importantly by provincial and national standards spreading from successive political centers. The educated elite who have governed the country as the imperial bureaucracy have been the prime source of this influence, but itinerant traders have no doubt also played a role at a lower social level. The result is that all dialects are more or less multilayered. This is obvious even in Mandarin, which has alternative pronunciations from many common words. It is even more so in areas like Min that have been relatively isolated and have preserved traces of very archaic features. In some cases literary or reading pronunciations that give evidence of earlier outside influence are clearly labeled as such but in other cases pronunciations that follow the same pattern have been thoroughly naturalized into the vernacular. Sorting this out at the strictly contemporary level is difficult or even impossible. The historical depth provided by philological evidence for national standards of pronunciation at various periods in the past is indispensable. It is not a question of one or the other. The more we know about modern vernaculars the better we shall be able to understand the historical evidence, and vice versa.

Karlgren was trained in the tradition of phonetic realism of the nineteenth century and never accommodated himself to the idea of the phoneme as advocated in the structural linguistics of Saussure and Bloomfield. His work was criticized from the point of view by Y.R. Chao and Samuel martin. The world has moved on again from American Structuralist School to distinctive feature theory. It is impossible to hope for progress either in the analysis of contemporary dialects or historical reconstruction without taking account of such things as distinctive feature theory and recent developments in the theory of syllable structure and prosody.

The Qieyun rhyme dictionary, completed in 601, and the rhyme tables of the Yunjing tradition, of unknown date but probably originating in Buddhist circles in the ninth century. These traditional sources remain the essential foundation for any investigation of the history of the Chinese sound system. Of course, one needs to use all sorts of other evidence as well. This evidence includes not only modern dialects and ancient transcriptions, but also such things as the systematic borrowings of Chinese in Japanese, Korean, and Vietnamese, and the rhyming of poets. Pullyblank's reconstructions of Late Middle Chinese based on the Yunjing and Early Middle Chinese based on the Qieyun, and analysis and comparison of a wide variety of data of many different kinds.

Kargren assumed that the Qieyun was based on the current dialect of Chang'an, where it was completed in 601, and that this dialect spread over the whole empire as a koine during the Sui-Tang period asn was ancestral to most modern dialects outside the Min area.

Qieyun was not compiled by direct inquiry into current pronunciation in the manner of a modern linguistic survey but was made up by comparing five earlier dictionaries, all now lost. Four of these were from the north and one from the south. The principal compiler, Lu Fayan, had consulted other scholars about the project and one in particular, Yan zhitui, is known from his other writings to have been much interested in varieties of pronunciation and to have been well acquainted with the differences between educated speakers in the north and south in this respect. Pronunciation was, of course, not recorded in any kind of phonetic notation but simply by the way in which words were classified, first into the four major tone classes, Level, Rising, Departing and Entering, then within each tone category into rhymes, and finally within each rhyme into homophone groups for which the first character was given a Fanqie spelling, that is, its pronunciation was noted by two other characters, the first of which had the same initial sound and the second of which had same final. 

From the point of view of a modern linguist, this amounts to a phonemic analysis based on the principle of minimal pairs. Pullyblank believes that Qieyun represents an exhaustive statement of the distinctive phonological contrasts of a real language, according to distinctive feature theory. In Qieyun system, phonemic contrast certainly refers to the relation between two phonemes, but this kind of contrast might not be distinctive feature INSIDE a single phonological system of a particular language. One could theoretically attempt to reconstruct the Qieyun language by comparing these initial and final categories directly with modern dialect forms and contemporaneous evidence, but in practice sholars have mad use of the additional information provided by the so-called rhyme tables (dengyuntu) designed in later times as keys to the Qieyun.

The principal followed by Qieyun was that of maximizing distinctions. For example, Level Tone rhyme zhi 脂 was combined with zhi 之 in their southern dictionary but was separated in their northern dictionaries, they were invariably kept separate in the Qieyun. The same principle applied to the corresponding Rising and Departing Tone rhymes. 

The sound categories of the Qieyun have always been approached through the more sophisticated phonetic analysis provided by the so-called rhyme tables (dengyuntu) originally designed as keys to the Qieyun by Buddhist scholars, which must have started in Tang times but only came into prominence during the Song period. By general agreement the earliest of these is the Yunjing. The table most used by Karlgren was the Qieyun zhizhangtu, erroneously attributed to the great Northern Song statesman and historian, Sima Guang, but actually composed sometime around the end of the twelfth century. Karlgren believes that when the same distinctions in categories are observable in Qieyunzhizhangtu as in the Ts'ie Yün, we may reasonably expect that the phonetic ground for these distinctions is the same in both.