Between data-mining all things Etruscan and pondering on the issues of Pre-Proto-Indo-European, I'm becoming increasingly enchanted by Old Chinese. Old Chinese is absolutely nothing like modern Mandarin or Cantonese. In fact, all of the tones that we now take for granted in Chinese come purely from sound changes in the language that occured since Augustus Caesar crowned himself divine emperor of the newly forged Roman Empire. (I've wrote about Chinese tonogenesis before.) Vis-à-vis the number of tones, Mandarin was the more economical of the Sinitic languages, settling with four tones; the so-called "fifth tone" (as in the interrogative particle, 吗 ma) being in reality a misnomer for what is the reduced variant of any of these four tones. In Cantonese, the number of distinct tones is six.
The reconstruction of Old Chinese depends on who does the reconstructing but there is no question that Old Chinese had interesting consonant clusters that would make Russian speakers feel right at home. You can see these same clusters in Classical Tibetan, a language which shares a common ancestry with Chinese within the past 6000 years. In 1992, William Hubbard Baxter authored A Handbook of Old Chinese phonology, clarifying all of the interesting changes that took place. We can tell in part what happened during the course of history of the language based on old rhymes. Words that originally rhymed now no longer do. In fact, in many cases, the changes are absolutely dramatic. Below, I will talk about some of these "used-to-but-no-longer" rhymes, showing pronunciation of the characters in Modern Mandarin.
For example, we know that 人 rén "person" and 年 nián "year" used to rhyme but they don't now. Even in Cantonese, the two words have deviated a fair amount: /jɐn/ "person" & /nin/ "year". So some propose that the two words in Old Chinese were pronounced *znen and *ʕznen (yes, with pharyngeal onset!), respectively. This is what is represented on Wikipedia... but I would recommend that people take Wikipedia with a large pinch of salt. Baxter shows instead *njin "person" (where j is for IPA /j/ as in the y- of "yeller") and *nin "harvest, year" (see p.424).
Likewise, we are told that *b-rjɨŋ and *prjɨŋ both mean "ice" in Old Chinese (on p.273) and that they are undoubtedly related. However in Modern Mandarin, these two words are quite different: 凌 líng and 冰 bīng. In the first word, the prefix *b- has been omitted in Middle Chinese while *p- in the second word is not a prefix and therefore remains intact. Prefix deletion is a common change in Middle Chinese which is half the reason why modern Chinese seems so completely removed from its original pronunciation. Note that the voicing or voicelessness of the initial and final consonants of these words played a part in determining the eventual tone of the word and also keep in mind that different Chinese languages reflect different tones. What is usually high flat tone in Mandarin (which sounds like "singing" to English ears and which is identical in sound with the French tonal accent) will often be high falling tone in Cantonese (sounding like an exclamation in English).
(Mar 25 2008) Based on the input from Movenon, I realize now that my initial statement in the first paragraph of this entry ("Mandarin was the more conservative of the Sinitic languages, settling with four tones,[...] ") is too vague and may cause confusion. I was thinking of changing it to "Mandarin was the more conservative (in the number of tones)" but I guess the word "conservative" is the crux of the problem since I'm using it to mean "reserved, limited" while it can also be interpreted as meaning "traditional". Arrgh. Damn the English language. So I guess we'll have to try this: "Vis-à-vis the number of tones, Mandarin was the more economical of the Sinitic languages, settling with four tones,[...]".