These days when I am not trying to get my taris straight from my naris and keris, I am usually reading commentaries to Chinese texts. There has always been a lot of writing about writing, and even writing about writing about writing, in China. Even today, when you buy a copy of a Chinese classic--say, the Analects of Confucius--it is typically accompanied by some notes either in the margins or actually written in the body of the text in a smaller font or something. Some of these commentaries have become canonized right along with the texts they accompany. The thing that makes them interesting, though, is not the light they shed on the original text, but the way they capture an understanding of that text at a point in time. You can go through layers of commentaries and see how the understanding of a text has shifted through time.
Some texts, and some certain passages of texts, seem to attract more commentaries than others. Perhaps we can say those texts are more "open": with archaic language or obtuse subject matter, they lend themselves to interpretation.
What do the comments say? I have been trying to think of rough categories for them. Individual comments seem to have one or more of the following functions:
1. Textual/editorial: Stuff like "Character X should really be character Y."
2. Phonetic glosses: Notes on how to pronounce a character, usually via the fanqie 反切 ("reverse cut") system in which two characters are supplied, one with the same initial as the character in question and one with the same final. (ping+fang=pang, etc.)
3. Backgrounding: Notes providing historical or other allegedly factual information to flesh out the text.
4. Rephrasings: Using clearer wording for difficult passages.
5. Intertextual references: Using another text to explicate the one at hand.
6. Systematizing and generalizing: Drawing general moral or philosophical principles from the text.
As the written language was largely unstandardized for most of its history, #1 shouldn't surprise us much. I was a bit surprised by the frequency of #2, though. Perhaps they are similar to #1, in that once you could pronounce it you'd have a better guess at what the character was supposed to mean. But for that to work, you'd have to be using the same dialect as the commentator. Maybe there was some ancient form of what in China they now call putonghua 普通話, "common speech." (Confucius is said to speak his own dialect as well as yayan 雅言 "elegant speech" for more formal occasions, such as reciting the Odes.) Maybe these commentaries served as some kind of crib sheet for people who had to recite these texts in the common speech rather than their dialect. Numbers 3 and 4 are fairly obvious. Number 5 is interesting, though, because it assumes the texts are consistent with one another due to their canonical status--they're usually explaining, say, the Odes in terms of the Analects or something like that. Sometimes they explain parts of a text with other parts from the same one. Zhu Xi explained Confucian analects with other ones, assuming a consistent message and use of terminology. That brings us to #6. People like Zhu Xi extrapolated from the various canonical works things like, say, the meaning of humaneness 仁, whether that was an internal or external quality, and how to go about cultivating it.
I am betting that the later the commentary, the higher the number of fives and sixes in there. The texts had found more or less stable forms and needed less "editing," and standard commentaries made the meanings more or less understood even if the backgrounds could be argued over (people might disagree over references, etc.--one says the "prince" of a poem refers to King Wen and another says no, it's [somebody else]). Later, by the Song Dynasty, people like Zhu Xi were taking these texts and remixing and sampling them into their own works.
What did they get from doing that? Why slice'n'dice the canon to prove your point? Perhaps the use of canonical texts lent authority to their own ideas. Perhaps, and to me more likely, classical education provided a set of semantic building blocks any literate person would be familiar with, making its use natural. Whatever the reason, it is remarkable how the Confucian canon retained such a prominent place in the Chinese intellectual realm for so long.