Chinese writing has had an unbroken history for about four thousand years. Ancient artifacts such the tortoise shell oracle bones have been found with inscriptions on them. Out of some 2500 characters, about 1400 can be identified with modern characters. In the history of China, the users of chinese were located in many kingdoms and they were situated in many regions of present day China, with each region having its own local dialects and vocabularies.
Confucius (c. 500 B.C.E.) lived in the period just before the Warring States. Although his thoughts were recorded by his disciples, they were nonetheless written in characters. He is quoted as saying that 'dishonest scribes invented new characters for those that they did not know instead of leaving blanks'. He named them 'strange characters'. This lead to the use of many characters for a single word, and so the numbers of the characters grew unabated.
Several hundred years later out of the chaos, China was united by Qin Shi Huang Di of the Qin dynasty. A political and literary purge took place; opposition suppressed and books were gathered and burnt, and all but a few escaped this censorship. During this time the Great Wall of China was built. The effort was done by press ganging the intellectual elite in to physical labour. Many died and those that did ended their days entombed in the structure of the wall, making it the longest grave site in the world. Ironically, the Qin emperor commisioned a lexicon of characters and their meanings from his advisor, Li Si in 213 B.C.E. For the very first time, characters now had fixed meanings. Although he got some of these associations wrong, it provided a start for all future lexicons of the chinese language.
When scholars got around to interpreting the old text, much had changed in terms of dialect and way language was written. Chinese itself has relatively few sounds. In modern Chinese, there are approximately 1130 sounds (taking into account the tones also). There are around four to five thousand characters in everyday use. Hence many written characters are homophonous or have the same sound. Over the course of the centuries, with dialectal pronunciations changing, once homophonous groups of words were no longer so. Poetry that was once written to rhyme now read did not. Some works did not make sense since the words had changed in meaning during the course of time. Or, with Confucius' observation, words with the same sound were substituted for the actual by mistake. With such mis-spellings, interpretations therefore varied and hence over the centuries, many literary discussions and treatise were written.
This provides us with a useful means to test whether indeed a present day dialect closly resembles the older dialects. In comparing modern dialects by using them to read old or ancient chinese poetical texts, rhyming of the passages would indicate some bearing of this relation. Now, in Modern Mandarin Chinese spoken by a vast 90 percent plus of the population of China, large number of words have lost their end consonants as compared to various minority chinese dialect such as Hakka, and Cantonese. We can say this because when the Chinese exerted a major intellectual influence on its neighbouring countries of Korea, Japan and Annam (Vietnam) to the south, these countries imported the spoken sounds at the time of their introductions. Many of the words kept a semblance of their original sound values. For instance take the following words, notice that the sounds 't' or 'k' at the end of the word has been lost in Mandarin;
Mandarin Hakka Cantonese Japanese English
li lit lik riku power
ji git gat kichi; kitsu good fortune
chu tut chut shutsu; sui to exit
That is not to say that in Mandarin, all endings have been lost. In Hakka, ~ing has mutated into ~in whilst in japanese, ~ing has mutated consistantly into ~ei. Whereas ~t seems to have changed into ~l in Korean.
Mandarin Hakka Cantonese Japanese English
xing sin sing sei star
cheng sin sang sei,jo to be completed
Such lost consonants can help in the reconstruction of the older dialects. We can also survey other surviving documents such as dictionaries. There are various classification of characters. Two ways they may be indexed are by rhyme in rhyming dictionaries, and also by using radicals, which are parts of a character that appear in common with other characters. Thus, these are arbitary ways in which they can be grouped together. So, after looking at the rhymes and also the endings of the words, it leaves then the tones of the words themselves.
Since language is a freely changing medium of communication, there are no real boundaries in which we can say for certain when such and such a change took place, so linguists have sectioned history into five arbitary parts, namely; proto-chinese, archaic, ancient, middle and modern corresponding to: before Confucious and Mencius, after Confucius around 500BC to the end of the Three Kingdoms, Han Dynasty to the end of the Sung Dynasty, and modern era as Yuan Dynasty to the present. Since we have the modern pronunciation of modern dialects, using documents contemporaneous to the periods concerned, it is possible be reconstuct more older dialects. So after a constuction of middle chinese pronunciation, a construction of ancient the archaic chinese and finally proto chinese.
The derived pronunciation can be never be absolutely 100 percent accurate. Centers of political influences changed with wars and so the local dialect of the center also changes. Since chinese is a tonal language, and since the reconstruction is based on modern dialects, fixing the tone structure and selecting the correct change in the word's sound will depend on many assumptions.
One of these is the survival of consonant endings as we have mentioned above. Chinese was first introduced into Japan as early as the first century AD, through Korea mainly through the vehicle of Buddhism. Hence Buddhist pronunciation of chinese words in modern Japan are the oldest contemporaneous evidence of earlier sounds still preserved. Since Japan has had long relations with China, it had borrowed sounds and meanings throughout its long history. Thus you will notice that there are often more than one pronunciation to each word, reflecting the periods of linguistic borrowings.
Mutation of sounds over the ages vary at different rates between the different dialects. Documented sources may not be from the locality for which the reconstructed language was designed. However the construction gives a rough idea as to what speech was like in the past. This proto-chinese from which all modern chinese dialects descend is at best a guess. It would be interesting to hear someone speak it.