Language Log

What you’re asking for would require a treatise, or maybe even a monograph on these things, and I must pass on that assignment right now. But I can help you out a little. First of all, these points are dealt with in two handy sources. The first is Jerry [Norman]’s book Chinese, Chapter 2. The index to the book will lead you to the relevant parts of the chapter. The other source is a full exposition of traditional medieval Chinese phonology by Guillaume Jacques. You will find it here.  Start reading on p. 6 and then read as much as you find useful.

Now, briefly, regarding your specific questions:
 
1. when and how it arose

This is a bit ambiguous, since  it is not clear to me whether you mean when the actual distinction arose in the language (an impossible question to answer) or when the distinction in terminology arose. As to the latter the terms apparently date from mid to late Tang times and had fully assumed their accepted current usages and meanings by the time of the rime tables. The earliest received tables date from Song times but are thought to be evolved versions of earlier more primitive prototypes. The tables are specifically connected to the Qieyun (601 AD) by some of their internal structure and terminology, so they cannot be any earlier than the QY text. And, obviously, they must be at least somewhat later than that, since they were apparently intended to be used in various ways to interpret the QY. It is generally agreed that the rime table inventory and arrangement of consonant types is based on Indic models (i.e., the vargas), and Chinese accounts specifically say that Chinese Buddhist monks were the ones who created, or at least inspired the creation of, the rime tables.

 
2. how it functions within traditional Chinese phonology

In the Yunjing there are three categories, qing, zhuo, and qingzhuo. The only way one could be 100% certain what these categories denoted in real phonetic terms for the rime table compilers would be to listen to them pronounce relevant syllables in these classes, which is of course impossible. So all descriptions of their phonetic characteristics are a priori conjectural. But based on the parallels between the consonant series in the tables and the corresponding arrangements of consonants in Indian grammatical tradition, one can begin to make educated guesses. Also, one can look at the modern dialects to see where they make parallel distinctions found in the traditional nomenclature. And, then, one can look at Sino-Xenic reading systems like Japanese, Korean, Vietnamese character reading traditions , etc. From all this, it seems virtually certain that qing just meant “voiceless”. Zhuo is harder to deal with, but it would seem logical to assume that voicing of some kind was what was meant. Based on inspection of examples qingzhuo then comprised sonorant or resonant elements, which in the modern dialects means liquids, nasals, and pure vocalic or semi-vocalic  onsets (where no initial consonant classifiable as qing or zhuo was present). In somewhat later traditional phonology qing and zhuo are each divided into two subtypes, i.e., 全清,次清,全濁, and 次濁, i.e., “completely clear, secondary clear, completely turbid, and secondary turbid”. In this breakdown, completely clear means voiceless plain, while secondary clear means voiceless aspirated. This is easy to determine by inspection. Likewise, secondary turbid means sonorants or resonants and vocalics. Then we have completely turbid. What was that? Some modern dialects, like Southern Wu and Central Xiang have pure voiced initials here. The consonants sound just like voiced consonants in English. But in Northern Wu, syllables with completely turbid initials are not at all like this. Instead, they tend to have a voiceless lenis onset, followed by murmur or breathy voice on the portion of the syllable beginning immediately after the initial consonant. The general view of historical phonologists is that in earlier stages of the language the quanzhuo initials were probably true voiced sounds, while in somewhat later periods, in the north at least, they were perhaps more often of the Northern Wu type. Of course, this is not a blanket pronouncement for the whole Sinitic speech area, because, as I have just noted, there are modern dialects that still have fully voiced quanzhuo initials today, and as well as dialects that have the Northern Wu type, with voiceless lenis onset accompanied by breathy voice or murmur. And, of course, there are modern northern  dialects that have lost voicing altogether, though we now know that one or two Southern Mandarin dialects still have the Wu-type initials.

3. how it correlates with concepts in modern linguistics

I think what I have said above pretty much answers this, but I’ll sum it up as follows. Modern phoneticians would class all qing consonants in the traditional system as voiceless. Zhuo is more complex and depends on what sort of pre-modern Chinese one is speculating about. Central Xiang zhuo initials today sound just like English voiced consonants to me, and Central Xiang speakers with whom I have worked agree with that. For example, a lady from Hunan I am working with right now feels that her word for “skin” 皮 [bi2] and my English word for “bee” or the name of the letter “B” are identical to her word, as long as I say my words with the apposite tone contour of her word. The northern Wu type zhuo is special and is accordingly given special treatment by modern phoneticians like Ladefoged and Maddieson. These people class the consonants in question as a sub-type of murmured or breathy voiced obstruent. It is interesting to compare these Wu consonants with the so-called “voiced aspirated” initials in Indic languages. The two types are perceptually quite different to my ear. You might like to read what  Ladefoged and Maddieson say about both types. I personally think that the difference is in voicing onsets. The Indic sounds have voiced onsets followed by murmur or breathy voice, while the Northern Wu sounds have voiceless lenis onsets followed by murmur or breathy voice. Anyway, as you can see, “zhuo” is conceptually a considerably more complex category than qing. Qingzhuo is pretty straightforward and just means resonant/sonorant, including vocalic and semi vocalic or glide-like entities. That, I suspect is what modern phoneticians would say about them.