As far as I know, tonogenesis occurs when consonants merge. The merging of initial consonants results in register tones and the merging of final consonants results in contour tones.
What are concrete examples of this in Chinese?
Linguistics Stack Exchange is a question and answer site for professional linguists and others with an interest in linguistic research and theory. It only takes a minute to sign up.Sign up to join this community
The formation of the "four tone" system of Middle Chinese, which resulted in a historically attested distinction (see the various rime dictionaries compiled in the Sui, Tang and Song dynasties), is meant to have derived from the "cheshirisation" of the final consonants of Later Old Chinese.
Finding the "reflexes" of these ancient consonants requires comparative evidence from corresponding cognates across the Sino-Tibetan family. One such cross-comparison (found on pages 30-31 of the ABC Etymological Dictionary of Old Chinese) comes from comparing Middle Chinese "tone B" or 上聲 (Mandarin Pinyin: shǎngshēng) with Chepang, from the Mahabharat ranges of Nepal; the connection is strengthened with evidence from tone 1 of Tiddim Chin in northwestern Myanmar. Examples cited include：
Additionally, it is also possible to glean hints of the original glottal stop that gave rise to 上聲 within the Sinitic varieties. Just as page 322 of Baxter's A Handbook of Old Chinese Phonology states:
There are in fact modem Chinese dialects which have a final glottal stop in words which are shǎngshēng in Middle Chinese; the dialects cited by Mei are Wēnzhōu 溫州 in Zhèjiāng, generally assigned to the Wú group, and four Mǐn dialects: Pǔchéng 浦城 and Jiànyáng 建陽 in northwestern Fújiàn, and Dìng'ān 定安 and Wénchāng 文昌 on Hǎinán island.
Baxter goes on to talk about loanword adaptation, involving shǎngshēng syllables being chosen to represent the short vowels of Sanskrit (it must be remembered Sanskrit did not have final glottal stops) when transcribing early Buddhist texts into Chinese; he states that:
This shortness is a natural concomitant of the final glottal stop.
This line of evidence has also been found for the qùshēng 去聲 tone of Middle Chinese, which is presumed to have derived from Old Chinese *-s. Sanskrit and many other Indic languages had final -s (or one of the other sibilants, conventionally romanised as -ś and -ṣ in IAST), and these were transcribed with characters that would later have the qùshēng tone of Middle Chinese. One example is:
Bṛhaspati= 毗里害波底 which in Baxter's reconstruction of Middle Chinese is
bjij-liX-hajH-pa-tejX; -स्प-, the syllable -has-, has been transcribed with 害, which would have been qùshēng 去聲, as marked in the transcription with the capital
H, and corresponds to the final -s of the Sanskrit.
The same principles apply to the much later development of the cheshirisation of voicing leading to the lower register tones. With comparative cognate evidence, the Wu group, including Shanghainese and Suzhounese, still retains a marginal form of voicing (usually breathy voice in practice) where the Yue group such as Cantonese has one of the lower-pitched tones (tones 4, 5, 6 in Jyutping). Some loanword adaptations are also evidence for this: compare the modern tones of the transliteration of the above Sanskrit word (e.g. 毗, which corresponded to voiced बृ in Sanskrit).
Furthermore, we actually have documentary evidence: the three-way distinction between 全清、次清、全濁 in the obstruents (plus the 次濁 for the sonorants) was recorded in the rime dictionaries, and comparing these forms with the modern forms is straightforward. The classic example:
However, you can see that cheshirisation has left aspiration as well as tone change in Mandarin and in Cantonese. It is not always the case that the net result is tone change: in Mandarin, what was shǎngshēng 上聲 and qùshēng 去聲 was split according to voicing and then re-merged, losing the voicing contrast but without net tonogenesis. Both Cantonese and Hokkien exhibit (or rather maintain the result of) tonogenesis though.