Just as a raised 'h' indicates aspiration, the circle below or above an IPA  symbol means that the sound is not voiced. Given that amongst the stops and affricates the fundamental opposition is between aspiration and non-aspiration, in broader transcription, [ b̥ ], [ d̥ ], [ g̊ ], [ d̥z̥ ], [ d̥ʐ̊ ] and [ d̥ʑ̊ ] are shown very adequately just as [p], [t], [k], [ts], [tʂ], and [tɕ]. Using Pinyin letter values to 'analyse' zh, ch, sh and j ,q x, they could be said to be "dsh, tsh, sh" on the one hand, and "dx, tx, x" on the other.

What Pinyin writes as  r- was -- and among some speakers still is -- pronounced [ ʐ ]. The considerable friction involved with this latter sound explains Wade-Giles use of the letter j, the Russian system's use of ж and the fact that the first draft of Beila used not rh but jh. The j analogises from the value of of this letter in French -- [ʒ].

Zhuyin Fuhao has no zero initials (Pinyin's y- and w-). The Russian system is asymmetric here: there is no zero-initial conforming to Pinyin y- (й- being possible), but for all syllables beginning with w- in Pinyin, the Russian system uses в-, with the exception of Pinyin wu, which is simply (cyrillic) у without a в in front.  Further, what Pinyin writes as wei (by the above logic, either уй or вуй) is вэй.    

Aside from exclamations, o only occurs in diphthongs (Pinyin's ou and uo). It is perhaps for this reason that the Russian system economises /o/ and /uɔ/ as one undifferentiated о.

Before n or ng, i is [ɪ]. Before ng, o can be either [u] or [o].

Rarely, the syllable finals uan (after y, j, q, x), ian, and iong are pronounced [yan], [ian], and [yuŋ] rather than [yɛn], [iɛn], and [iuŋ]. This gives the impression of being old-fashioned or ultra-correct.

The apical dental vowel that Pinyin writes as i after s, z, c and the apical retroflex vowel that it writes as i after sh, zh, ch are shown above as ɿ and ʅ . These are unofficial but in some quarters widely used IPA symbols. Using official IPA the same two sounds are sometimes shown as [ z̩ ] and [ ɻ̩ ], a notation that seem to exaggerates the consonantal character of the two sounds. Another expediency sometimes seen is the use of [ ɨ ] or even [ ɯ ], modified by diacriticals of one sort or another.

Let us see -- in tandem with the problem of semi-palatals and retroflexes  -- how different systems represent  [ i ], [ ɿ ] and [ ʅ ]. (For those systems -- namely EFEO and Beila -- that have Pinyin ji, qi, xi divided into a velar series and a dental series, we choose their dental correspondent to xi.)
 

IPA	[ ɕi ]	[ sɿ ]	[ ʂʅ ]
Hanyu Pinyin	xi	si	shi
Beila	si	s ( [ ɿ ] is default vowel)	sh ( [ ʅ ] is default vowel)
Gwoyeu Romatzyh	shi	sy	shy
Yale	syi	sz	shr
Russian	си	сы	ши
Wade-Giles	hsi	ssŭ (or szu)	shih
EFEO	si	sseu	che
Bortone-Allegra	sci	se	sce
Zhuyin Fuhao	ㄒㄧ	ㄙ ( [ ɿ ] is default vowel)	ㄕ ( [ ʅ ] is default vowel)

Using the numbers from 1 to 5, with 1 representing the lowest pitch register of a given person's normal speaking voice and 5 representing the highest, the four tones of the standard language based on Pekinese are approximately as follows: 55, 35, 21 (or 214), 51. In other words all but the first tone has a rising or falling pitch-contour. In actual linked speech there are further variations, with 21(4) becoming 35 when followed by another third tone, and 11 when followed by any other tone.

Below, the syllables 'tang', 'qian' and 'hu' are put into their Hanyu Pinyin and Gwoyeu Romatzyh orthographic representations. Note that GR shows tone through spelling variation -- either changing letters or adding them -- but not in the same way for all syllables.
 

Also called inter-regional romanisation. It means a more-or-less historical or 'etymological' spelling for cognates shared amongst the various Sinitic languages and dialects, and so necessarily involves a much larger repertoire of initials and finals than found in any one presently existing variety. The result would be a relatively complex orthography that cannot be read as direct transcription but only as an abstraction to which different sets of transformation rules must be applied per language variety. For example, of such possible spellings as (a) zren, dren, jren; (b) gian, giam, gien, giem, zien ziem; (c) gon, gom; (d) ghip, ghit, ghik, each set would from the point of view of Pekinese consist of homophones --  (a) zhēn, (b) jiān, (c) gān, (d) jí -- while in Cantonese, Hakka, Southern Min, or the language of Shanghai there would be variously differing sets of homophones and non-homophones. For example, read into Cantonese, all three members of (a) would be [tsɐn], but (b) would be differerentiated into [kan] [kam] [kin] [kim] [tsin], and [tsim]. In Southern Min, dren would be diffierentiated from zren and jren, as [tin] vs. [tsin]. In the language of Shanghai, all members of (b) would be [tɕi], but with gian and giam also pronounced [kɛ] in some lexical contexts.

Aside from what seems from the viewpoint of any single language variety to be a superfluous number of initials and finals, another key to how such a romanisation is 'decodable' is, as the following schematic shows, the relationship between the four etymological tone-classes and four types of etymological initials (aspirated unvoiced obstruents, non-aspirated unvoiced obstruents, voiced obstruents, and 'sonorants'  -- respectively something along the lines of [pʰ] [tʰ] [ʦʰ] [tʲʰ] [ʧʰ] [ʨʰ] [kʰ], [p] [t] [ʦ] [tʲ] [ʧ] [ʨ] [k] [f] [x] [s] [ʃ] [ɕ], [b] [d] [ʣ] [dʲ] [ʤ] [ʥ] [g] [v] [ɣ] [z] [ʒ] [ʑ], and [ɱ] [m] [l] [n] [nʲʑ] [nʲ] [j] [ŋ] ).
 

Ⅰ	Ⅱ	Ⅲ	Ⅳ	Pekinese:
tung	tŭng	tùng	tuk	tōng tŏng tòng tu
dung	dŭng	dùng	duk	dōng dŏng dòng du
dhung	dhŭng	dhùng	dhuk	tóng dòng dòng dú
lung	lŭng	lùng	luk	lóng lŏng lòng lù

In the case of Pekinese, tonal predictability breaks down where a syllable ending in etymological -p, -t, or -k begins with an etymologically unvoiced initial, but for the southern languages -- as well as many of the northern dialects other than Pekinese -- there would neat sets of correspondences, no matter if they correlate to the three tonemes of Tianshui, the four of Xi'an and Wuhan, the five of Nanjing, or the eight of Guangzhou and Chaozhou.

Both Wáng Lì and Zhào Yuánrèn (Y.R. Chao) have designed inter-regional romanisations.
 
 

TO INDEX