Thanks

I would like to thank Thomas Chan and Mike Wright for their generous help and advice. This page summarizes many observations of Chinese that we have discussed privately. However, all errors here are my own.

Indeed, all that we know about sounds of early Chinese and its phonological history is in part due to both native Chinese linguists throughout history, and great names in modern linguistics such as Bernhard Karlgren. It is upon their shoulders that we stand. We take a slice of the pie in this page by looking at numerals across the circle of influence that China exerted in the Far East of Asia, noting briefly their geographical and historical backgrounds.

Languages and Dialects

Sino Korean, Sino-Japanese (Kan-On and Go-On), Sino Vietnamese (Quo^'c Ngu+~), Chinese Southern Dialects (Cantonese, Hakka, ChaoZhou and FuZhou), and Mandarin PuTongHua (PTH) readings are given in the table below for the characters commonly associated with numerals, and their character representation.

Cantonese has three Ru tones; upper (Yin), middle, and lower (Yang); represented by 7a, 7b and 8 respectively. Other Chinese dialects can only distinguish two Ru tones, hence upper and middle Ru tones for Cantonese have been grouped together. Where other dialects do not distinguish between upper and lower in their tones, the upper tone (Yin) has been used.

Notes on Vietnamese transcriptions and tones are given in the notes at the end of this page.

Sounds

The list of Chinese numerals above and the sounds adopted by the various CJKV countries demonstrates neatly some aspects of phonological change not only in China, but in the other countries which have used the script. Instead of viewing all the languages and dialects as isolated and separate, they can be thought of as related and derived from a common source, albeit at various periods along the development of Chinese itself.

Our primary object is to discuss endings and certain initials that crop up consistently across the South East Asian readings of these numerals and other characters. Characters which have occlusive endings -p, -t, -k, and nasal endings -n, -ng, -m are represented in the small sample above. Whether they appear in other dialects also shows the degree of change the dialects have undergone. The tone development of the Chinese dialects is compared to stage in history which can attested to through rhymebooks.

IPA representation have been provided as romanised standards of the various countries and dialects differ widely in their sound values. The IPA forms are primarily derived from Karlgren's book, Études sur la Phonologie Chinoise. Other romanisations are listed below in the "Notes" section.

Chinese Dialects

There are many regional accents or subdialects of each major dialect of Chinese. Those represented here are easily accessible as being the considered standards. They can be considered as northern (represented by PuTongHua Mandarin) and southern (represented by Cantonese, Hakka, ChaoZhou, and FuZhou dialects) and these have been chosen because each represents a stage in the development of Chinese.

Cantonese and Hakka

Location and Background

Cantonese is taken to be the dialect associated with GuangZhou city in Central GuangDong Province. MeiZhou city is the location of the MeiXian dialect of Hakka, generally accepted as the paragon. It is located in NE Guangdong Province.

There is evidence that both Cantonese and Hakka people have migrated down to these areas over time, and their origins once lay in more northerly regions. Clan genealogies and other documented evidence including the history of turbulent change in China point towards this. It is not known for certain how long each group has been in GuangDong, but during the middle of the Song Dynasty 960-1279, there was a massive population migration southwards, leaving a depopulated North because of the incursion of the northern tribes, the Jin and Mongols.

Comments

Both Southern China dialects of Cantonese and Hakka preserve many of the endings of older dialects of Chinese as they occured many centuries ago. These are the nasal endings -m (3), -n (1000) and -ng (0), and the occlusive endings -p (10) -t (1) and -k (100). The similarities between the two dialects are that they generally share common initials and ending types. They mainly differ in their vowels. Vowel dipthongs have changed to single vowel or long vowels in Cantonese. Hakka vowel dipthongs, like -ia-, -io- and -iu- (6) become Cantonese -i-/-e-, -oe-, -u- before consonant endings.

In Hakka, the velar consonants endings in Cantonese manifest themselves as alvelors when preceeded by the consonant -i- consistently (eat C. sik, H. sit). This is an example of how consonants ending can move forward in the mouth depending on certain situations. Min dialects show the opposite with a progression of the end consonant towards the back of the mouth. See ChaoZhou and FuZhou dialects below.

Hakka maintains a bilabial ending -p to words which had bilabial initials. Mostly, these bilabial initials have developed into f- initials in most Chinese dialects. For instance 'law' is in Hakka fap, whilst in Cantonese, fat. Vietnamese shares Hakka's conservativeness in this resepct in the transliteration pha'p. Also, the bibilabial ending can be seen in Korean beob. Japanese has a reading of hou where the h initial developed from a bilabial, and the final -u derives from the change of *pu > *Fu > hu > u for this character. However, it is not a numeral, which we are considering in this article. It has to be said though, that one other SE dialect, namely ChaoZhou maintains the bilabial, the other Min dialects do not exhibit this.

To a lesser extent than Mandarin, Cantonese looses initials such as ng- and takes on y- as the intial, where Hakka has ng-. In some Hakka subdialects, allophone initials ng- and ny- exist, the latter palatisation agreeing with Karlgren's reconstruction. This palatisation manefist itself as the semivowel y- in Cantonese and Mandarin for the readings of the same characters, though other relationships also exist. (See Mandarin dialect below.)

For the numeral 10 000, we see that Hakka has v- and Cantonese m- as initials. Comparing this to Karlgren's reconstruction, it shows that one dialect may preserves some initials better than the other, rather like the contraction of vowel dipthongs in Cantonese where Hakka has been more conservative. In this respect, we see how different dialects of Chinese can show levels of change by viewing the contrast between them. Therefore, all modern dialects and readings from other countries hold clues to sounds of Chinese that haven't been heard for centuries or millenia.

Mandarin

Location and Background

The Chinese Language being taught in most institutions is Mandarin PuTongHua. It is a standardised form of the BeiJing dialect, spoken in the capital of China located in NE China.

After helping the Song court crush the Jin kingdom, the Mongol empire eventually conquered Song, and the Yuan era (1279-1368) began. The Mongol conquerers became sinicized by the end of their short-lived dynasty and China once again reverted back to native Chinese control under the Ming. By this time, occlusive endings had already degenerated and in some cases lost to the northern speakers. After the non-Chinese Manchu's established the Qing dynasty in the 17th Century, they also became sinicized, and interest in Chinese literature and arts flourished. The great KangXi dictionary of 1716 recorded around 48,000 Chinese characters. It was recognised early on that the northern dialects had diverged in pronunciation from earlier times because poetry of earlier times no longer rhymed. During the Qing era, many discoveries were made about early pronunciation by Chinese linguists.

Comments

Mandarin represented by PuTongHua is now standardised as a form of the Beijing Dialect. It represents the loss of occlusive endings in Chinese dialects, to the point where traditionally occlusive Ru tone characters have joined fully into other tone classes, namely Ping, Shang and Qu Sheng (1, 6, 7, 8, 10, 100, 100 million). Mandarin is considered as a Northern Chinese dialect, but Hakka vowels sounds often relate to Mandarin rather than Cantonese (1 H. yit, M. yi, C. yat). However, Hakka grammar and vocabulary mirrors Cantonese more than Mandarin. Mandarin has the vowel dipthongs -ia- and -io- before consonant endings.

The table of numerals also show that endings -m has now changed to -n (3), and the initial nasal in particular ng has been lost (2 and 5), though it endings they remain (0).

However, as can be seen in the numeral 2, ng- also derives from a palatized form of n, so whereas Hakka has "ngien2" (year), Mandarin has "nian2" and Cantonese "nin2" (tones as in table above), and Hakka "ngiu2" (cow), Mandarin "niu2" and Cantonese "ngau2". These examples also demonstrate Hakka and Cantonese initial and ending similarities, whilst vowels of Hakka coincide with Mandarin. The loss of ng in Cantonese also occurs to a lesser extend (2, H. ngi, C. yi, M. er) compared to Mandarin.

ChaoZhou and FuZhou dialects

Location

Chaozhou is located in NE GuangDong and represents a Southern Min dialect. FuZhou is on the NE coast of FuJian Province, in SE China, representing a Northern Min dialect. ChaoZhou is not far from ShanTou and the ChaoZhou Dictionary states that the romanisation covers both ShanTou and ChaoZhou cities.

In FuJian and NE GuangDong where the Min speaking population are, their population locations are both isolated due to lack of rivers and roads or communication routes, and geographically inaccessible because of the mountainous terrain [see notes]. Thus, there were little outside influences that penetrated these regions.

Comments

An aspect of Min dialects is that they possess multiple layers of readings. The main ones are the colloquial and literary Tang-Min layers. Here the Tang Min sounds of Fuzhou are given, whilst the colloquial pronunciation of ChaoZhou is given.

The ChaoZhou-ShanTou unified transcription mentions that the -n is a nasalisation of the preceeding vowel rather than a consonant ending and -h is a glottal stop, though the latter (8) is not shown in Karlgren's IPA representations.

Occlusive endings, in this dialect group, are beginning to decay into glottal stops, and so represents an intermediate stage in the progression from occlusive sounds to the loss of them (8). Min dialects therefore stand between Cantonese (and Hakka) and Mandarin in this respect. For Fuzhou and Chaozhou, the ending -t has fully mutated to -k (1, 7). There is also some nasalisation of end vowels as originally nasal endings decay (3, 1000). ChaoZhou dialect retains the ng initial (5).

In the FuZhou dialect, all -t and -p endings have gone to -k, (1, 7, 8, 10), a velar plosive, which demonstrates the gradual movement of the occlusive ending from being articulated at the lips (bilabial -p) to the teeth ridge (alveolar -t) and backwards to the palate (velar -k), an observation made by others. It is just a short step towards being glottalised as in the ChaoZhou case for numeral 8.

The Fuzhou dialect also demonstrates the movement of nasal articulation into the rear of the mouth in numerals 3, 1000, and 10 000. It may mirror the movement of occlusives, as the pairs m:p, n:d and ng:k are articulated in the same regions for each pair element. This observation is in contrast with Hakka, and its strength seem to lie in the fact that we take Mandarin as a dialect with no occlusive endings, a state which all dialects would supposedly progress to.

Interestingly, the numeral 100 million in FuZhou parallels Hakka, displaying a loss of the end consonant when it appears in other sino-xenic and conservative dialects. From the KangXi dictionary, the rhyme character displays the -k ending. A number of conclusions can be drawn.

• It was always a character with an occlusive ending that was recently lost in FuZhou, Hakka and other dialects.
• It was a character that only recently developed an occlusive ending prior to the adoption of readings in Sino-Xenic languages.
• It has multiple readings, and one was more popular in one dialect than another.

We do not propose any one as being the only explanation.

Sino-xenic Languages

These are represented by Korean, Japanese and Vietnamese. Through osmosis, these countries acquired a large body of Chinese vocabulary, the use of Chinese characters, and culture and political influences over the long intercourse with their large influential neighbour. Whilst South Korea and Japan still use Chinese characters called Hanja and Kanji respectively, Vietnam has moved over to a romanised form to transcribe their language.

Sino-Japanese

Location and Background.

Japan lies to the east of China, as a chain of islands. Due to its geographical nature, it is isolated from direct influence from Mainland China and Korea. Other than its enclave of Mimana in the 4th to 6th century, its contacts were primarily through Japanese scholars and travellers to the continent, and trade with its neighbours via sea.

The Kan-On and Go-On readings are general classes of sounds each with their own distinctions. They have been borrowed over time from many geographical regions of China, and so many Chinese dialect in the process, and some readings have gone through the filter of Korea pronunciation. They are associated with certain aspects of Japanese vocabulary. Go-On pronunciations are mainly associated with language associated with Buddhism from the sixth to late seventh century. Kan-On readings are encountered more, they represent the borrowings from the China between the eighth and tenth century.

Comments

Japanese preserve many rhyme endings which have gone from Chinese dialects altogether, so it serves as a good reference for ancient sounds. We see the occlusive -t and -k endings as represented by -tsu, -chi and -ku, -ki endings in readings today. However, over the centuries, Japan has lost the -p ending, though it can be traced by looking at old kana spellings prior to modern language reforms of the last 50 years. Also, the nasal ending -ng in Chinese has changed and now readings usually end in a vowel dipthong such as -ou or -uu, where the final u is the remanent in modern Japanese spellings. Depending on the time of borrowing, the Chinese -ng ending can also be represented by the Japanese -n ending.

Japanese is not a tonal language in the same sense as Chinese is, so tones have no real meaning when applied to classes of characters.

Some interesting relationships between Kan-On <=> Go-On are in the initials, m- <=> b-, d- <=> n-, z- <=> n-, k <=> w, and unvoiced <=> voiced initials, and in the endings like Chinese -t show -tsu <=> -chi. For instance, the numeral 2 has ji (otherwise written as zi in kunreishiki) and ni, and itsu and ichi for Kan-On and Go-On repsectively.

In our small sample of Chinese characters, only -ku is representative of -k in Chinese, as in 100 and 100 million.

Numerals 1, 7, 8 represent the Chinese -t endings as -chi and -tsu (Kunreishiki -ti and -tu).

The loss of the -p ending is shown by the numeral 10. The result is the vowel dipthong, once represented by -fu, (Kunreishiki -hu) itself derived from pu (which shows indirectly that even the initial p- has changed). In the numeral 10 the ipa u: represents this as an elongation of the u vowel of the preceeding kana sound.

The representation of -n (3, 1000, 10 000) in Japanese depends on the following consonant. When it is followed by b, p, or m, it has the sound m. When followed by k, g, or ng (nasalised g), it has the sound of nasalised g or ng. Before other consonants, t, ts, ch, s, d, j ,z, n and r, it has the sound n.

The ending -ng is represented by numeral zero, 0, and like the -p ending has also decayed. It is shown as the residual u in kana representation and in the IPA rio: as an elongation of vowel o.

Sino-Korean

Location and Background

Korea is a peninsula located to the NE reaches of China, sharing a common border with North Korea.

During the Three Kingdoms era of China, 220-265AD, the Korean Peninsula was split between local kingdoms (Kokuryo, Mahan, Jinhan, Byeonhan, and the Wei Kingdom of the Three Kingdoms), so there was early cultural intercourse between early China and early Koreans. When the Three Kingdoms collapsed, China fragmented into the many smaller states, its grip on Korea loosened, and the early Japanese invaded the peninsula to form the enclave of Mimana, 369-538. Japan now had direct access to Chininese culture, literature, sciences amongst other thngs. Later, Korea unified under the Shilla Period, (676-935) and succeeded by the Koryo Dynasty (918-1392) from whence the name Korea derives, the Choson Dynasty, (1392-1910), a 35 year interlude under Japanese colonial rule, and then Korea divided into North and South.

During the Japanese occupation of 1910-1945, Korea also acquired a new subset of vocabulary of specifically Japanese derived character terms (wristwatch, shi-gye; photograph, sa-jin; etc), with Sino-Korean readings.

Comments

Sino-Korean pronunciation of character retains endings similar to the Southern Chinese dialects. However, characters with Chinese -t rhyme endings show -l in Korean. Like Japanese, it is not a tonal language in the sense that Chinese is. Some striking similarities in the pronunciations of the characters between Korea and Southern Chinese dialects occur, even though the geographical distance of some 2000km (1200miles). This also reinforces the notion that modern Southern Chinese dialects must have had some proximity to Korea, and therefore, ancestors of these dialect speakers must once have resided in more northerly locations.

Names of people and places primarily use Sino-Chinese readings, (e.g Kim Il Sung, PyongYang etc), but native readings for characters also occur (e.g. Seoul).

The numerals above show that the Korean pronuciations show some semblance to Southern Chinese, in particular Hakka. Ryug (6) and ryeong (0), both with r- initials in the romanisation does not appear as a retroflex in the IPA transcription, therefore it may be due to the romanisation. In Chinese these numbers has the initial l- and they are zero initial in the Sino-Korean readings.

The nasal ng- does not seem to appear for Sino-Korean readings of Chinese characters, but has a zero initial (non-consonant initial) rather like a large proportion of Mandarin readings of such characters. For Sino-Korean placenames such as PyongYang, -ng occurs as a final as in the numeral 0. Other nasal endings -m (3) and -n (1000, 10 000) are retained.

Sino-Vietnamese

Location

Vietnam shares a common border with China, with Yunnan to the NW and GuangXi to the NE, both provinces being in the SW of China. True Vietnamese were once resident in Guangdong Province region of China, but due to Chinese expansion, they had to move and came to settle in the north of modern Vietnam on the Red River now called the Tonkin area over two millenia ago. At that time, the South was populated with the Cham and Khmer peoples. Vietnamese during the 17th Century moved down to the Tonle Sap basin and absorbed these groups, though it was not until the arrival of Europeans that they colonised the length of modern Vietnam.

Comments

Vietnam was called by the Chinese, Annam, covering only the Tonkin Basin. The common border with China afforded the dissemination of Chinese literature and other cultural influences. Early Vietnamese documents used Classical Chinese and later they developed their own characters to aid the recording of their language and continued until recently, Chinese characters were used officially. There is still a large body of literature which is written in Chinese characters, and many of these have solely been originated in Vietnam. These are the so-called chu+~ no^m or chu+~ nho characters.

Today, the sounds of Vietnamese are transcribed in a romanised form called Quo^'c Ngu+~, and it incorporated tone indications. Traditional Ru Sheng occlusive ending class characters can take on a number of Vietnamese tones. However, most occlusive tone dialects of Southern China only distinguish two levels of the Ru tone, ie, high and low. Therefore, Vietnamese tones may not relate to Chinese in any particular rigid way, or that waves of borrowings that have introduced tone pitches which have changed over time.

Vietnamese retains aspects of the older Chinese endings as in Southern Chinese dialects, i.e. nasals -m (3) and occlusive endings -p (10), -t (1, 7, 8), -k (6, 100 million). For the expected -k ending in the numeral 100, there is some palatisation. (Nguyen's dictionary gives a "Guide to Pronunciation" lists rhymes -ach as Northern -ayk, Central -at and Southern -at). In fact from our sample, Vietnamese ending -t is palatised, due to the Vietnamese sounds system.

We note that Vietnamese vowels seem to correspond to the vowels of Cantonese. Cantonese is also spoken in areas of GuangXi (bordering Vietnam as mentioned above) which may account for this coincidence.

Another pecularity of Vietnamese is that some Chinese s-, sh- initial words (3, 10) occur with t- initial in Vietnamese. The plosive t- must have deleveloped from fricative s- sometime along its history, as Vietnamese indicate the fricatice s using x- in its orthography.

Below some correspondence between Vietnamese broken and level tones to Chinese Ru and Ping tones is shown, however, the character sample is too small to make this a definite correlation.

Tones

Chinese linguistic tradition derives from an Indic tradition due to the early need for transcribing the names and appelations of Gods and terms from Sanskrit.

Ru tones in Chinese dialects are shown the occlusive endings -p -t -k (-b -d -g) or the glottal stop (-?). Most Chinese dialects only differentiate between two registers or levels of each tone, however for Cantonese, there are three Ru tones as mentioned above. For Vietnamese, traditional Ru tone characters take more than two types, so there may not be an obvious correspondence to any modern dialect of Chinese. However, we see that Ping and Ru tone classes seem to correspond to Vietnamese level and broken tones repectively.

Other endings (vowel and nasals) do not belong to any particular tone classes, Ping, Shang, and Qu.

During the Song Dynasty, Liu2 Yuan1 in 1252 condensed the 206 rhyme GuangYun Rhyme Dictionary to 107 Rhymes, called the PingShui Rhymes (PingShui was the geographical origins of Liu Yuan). It was later reduced to 106 in the Yuan Dynasty. GuangYun of 1008 and JiYun of 1037 built upon the tradition of Rhyme dictionaries from earlier times, such as QieYun of 601. The whole classification of tones began with Shen Yue's (441-513) Shi Sheng Pu.

By the Yuan Dynasty, works such as ZhongYuan YinYun (Rhymes of the Central Plains) of 1324 showed that the Ru sheng tone characters had already degenerated and classed with other tones, and the sounds represented is not far from modern dialects of the North of China.

PingShui rhymes became known as ShiYun (Poetry Rhymes) and represent the last period where the culmination of all linguistic traditions, still preserved the Ru tone distinctions. It is taken as our control tone sample to measure the change in modern dialects since the Song Dynasty. A note of caution: whilst ShiYun indicated Shang Ping and Xia Ping, they are not different registers of the Ping tone; they meerly mean that these characters belonged to different volumes of the GuangYun rhyme dictionary, these others were Shang, Qu and Ru, making five volumes in all. Therefore, for Shang Ping and Xia Ping, these should be seen as a Ping or level tone overall.

Tones
Character	Chinese Dialect					Other	Comments
Number	Cant	Hak	CZh	FZh	PTH	Viet	Shi Yun Class	Ru Tone	Dialects
	7a	7a	8	7a	1	High Rising '	Ru	*	The southern dialects share the same Ru tone class as found in ShiYun (Poetry) rhymes, though they belong to different registers. Mandarin has lost its occlusive ending, and ends up in a Ping tone.
	6	5	4	6	5	Low Falling `	Qu		Chaozhou pronunciation has changed tone class as compared with ShiYun, but the other dialects seem to retain the original category, though at different registers.
	1	1	1	1	1	Mid Level unmarked	Xia Ping		All four example dialects are in upper level tone, the ShiYun tone was a Xia Ping (level) tone, whilst those of the dialects show upper level tone. Vietnamese also has a level tone.
	2	2	2	5	2	High Rising '	Qu		ShiYun Qu tone class shows that the Chinese dialects have changed tone to a lower level tone, except for the Fuzhou dialect which is in the Yin Qu register.
	4	3	2	3	3	High Broken ~	Shang		Chaozhou has changed to a lower Ping tone, whilst other dialects seems to have kept to the same tone class as given by ShiYun rhymes.
	8	7a	8	8	5	Low Broken .	Ru	*	The lower broken tone of Vietnamese seems to correspond here to the lower Ru tone in Chaozhou and Cantonese. Hakka has low pitch for this type of tone in the standard Meixian dialect.
	7a	7a	7a	7a	1	High Broken ~	Ru	*	All the southern dialects share the same tone type. The Vietnamese high brokem tone appears to follow Cantonese and Chaozhou upper Ru tone.
	7b	7a	7a	7a	1	High Rising '	Ru	*	The southern dialects agree only in tone class, and Cantonese has a mid Ru tone which has been grouped with the Upper Ru tones for the other dialects.
	3	3	3	3	3	Low Rising ?	Shang		The lower Shang tone is reflected in Vietnamese by the low rising one in this case.
	2	2	2	2	2	Mid Level unmarked	Xia Ping		The ShiYun Xia Ping level tone is shows no change in tone type for the dialects, and Vietnamese agrees in a level tone. All modern Chinese dialect here show a Yang Ping tone.
	8	8	8	8	2	Low Broken .	Ru	*	The southern dialects agree in tone class (lower Ru) and Vietnamese by the lower broken tone.
	7b	7a	7a	8	1	High Rising '	Ru	*	Like the character for eight, this 100 also shows the same varieties of tone for each dialect. The Fuzhou dialect shows a lower Ru register.
	1	1	1	1	1	Mid Level unmarked	Xia Ping		Both Vietnamese and the Chinese dialects show a level tone type. However, all Chinese dialects represented here show the Yin Ping tone.
	6	5	6	6	5	Low Broken .	Qu		The ShiYun tone class has been perserved by the Chinese dialects, though at different registers.
	7a	5	8	5	5	High Rising '	Ru	*	Hakka, FuZhou and Mandarin, do not exhibit an occlusive ending for this case. The ShiYun Ru tone is kept by Cantonese and ChaoZhou, although at different registers. Why the conservative Southern dialect of Hakka and Fuzhou has lost the -k ending which appears in Sino-Xenic readings is unsure. The right portion of this character is often used as a phonetic, and only displays the occlusive end in the other dialect and languages when used this way in some cases.

Summary

Characters with Occlusive Endings

Ru class tones are found to retain the same tone class compared with ShiYun for Chinese tones. However, from the 106 Shi Yun Rhymes, there are no distinctions for upper or lower registers for this particular tone. Vietnamese broken tones tend to map onto occlusive ending sounds.

We have seen that most Sino-Xenic languages preserve some vestige of the older Chinese dialects that are still present in some Southern Chinese dialects. However, as in the case of Japanese and Korean, developments in native sound systems also attribute to the loss of these endings, and can be seen on a small scale in Min Dialects, and wholly in Mandarin.

Characters with Nasal Endings

There is no one tone category for nasal tones as in the special case of occlusives and Ru tones. Therefore, they are spread over the three remaining tone classes, Ping (level), Shang (rising) and Qu (departing). The meaning of these tone names were once applicable, but over time, characters originally assigned to them have changed as dialects develop tonally. The most extreme example is the merging of once Ru Sheng tones (1, 6, 7, 8, 10, 100, 100 million) for Mandarin into the other tone classes completely.

In Mandarin, -m has now merged into -n, and -ng endings are well preserved. On the surface, Japanese also seems to exhibits this. However, if one looks at the sounds which follow the -n sound, it changes, and in some cases where definite -n sounds are used, -m is heard. In Korean and Vietnamese, readings of the characters distinguish all three nasal types according to the sounds of their language.

Characters with Vowel Endings

Like nasal endings, vowel endings, so far uncommented upon, are can belong to any of the Ping, Shang and Ru tone class. In short, their endings are less prone to change (2, 4, 9). Tonally, they have have moved class for instance for the numeral 4, all the modern Chinese dialects have shifted into the Ping class; the numeral 9 retains the ShiYun class across the dialects; and the numeral 2 appears to belong to different registers of the same tone class, though ChaoZhou exhibits movement into the Shang tone class.

Table 4. Summary of endings comparing Chinese dialects and Sino-Xenic Languages (not in IPA)
Please see notes about the use of n in Japanese.

Conclusions

Karlgren's reconstructions shown here are from 1915, and may have been superceeded by other reconstructions by other eminent linguists of the Chinese language. We have not relied too much on Karlgren's readings, rather, they were used as a handy reference to possible sounds of Chinese many centuries ago.

A knowledge of the background history proved useful in our understanding in why there are reservoirs of conservative readings in the South of China, Vietnam, Korea, and Japan. The export of the Chinese written language to other languages which have different sentence and grammar structures from Chinese, also showed the strengths of such a writing system.

Even with so small a number of examples, the various Chinese dialects and readings in Sino-Xenic languages show some important phonological relationships between them. In particular, the readings themselves show that the endings persist in borrowings, and in conservative Southern Chinese dialects, whilst Northern Chinese dialects already exhibit that endings may be lost. Likewise, some comment about the intials of the readings of these characters have been noted. Most readings seem to preserve a vestige of the tone class in which they were used almost a millenium ago in Chinese, and in a tonal language such as Vietnamese, this is less certain. However, we do not pretend that these are the only changes undergone through the development of readings in China internally and externally, rather, they are but the tip of an iceberg waiting to be revealed.

Notes:

• ZhGYYXYJ = Zhon1gGuo2 Yin1Yun4Xue2 Yan2Jiu1 or Études sur la Phonologie Chinoise, which was originally in French, written by Bernhard Karlgren in 1915. This is the translated edition in Chinese, ISBN 7-100-00018-1, March 1995.
• IPA symbols represented in the table of pronunciations above, differs in a number of ways from Phonologie. These are due in part to their not being found in the IPA tables at http://www.arts.gla.ac.uk/IPA/fullchart.html . These are mainly the alveolo-palatal plosives, t and d which are represeted by t and d here; and the alveolo-palatal fricative z in combination with alveolo-palatal plosive d, here represented dz in the readings for Japanese j-; and finally, the alveolo-palatal nasal as in the initial character of the IPA transcription here used in numerals 1 and 2 in Vietnamese and in other places.
• Readings of Sino-Korean are from Charles Muller's CJK Dictionary which is available on-line.
• Kan-On and Go-On readings which are neither Go-on or Kan-On are shi, go, sen and oku; san is a kan-on reading, as is yoku.
• Round brackets ( )indicate the modern kana representations, whilst square brackets [ ] represent old kana writing.
• Sino-Vietnamese readings are in the internet VIQR representation. Some symbols used are given below. Romanised forms are transcribed from:
  NTC's Vietnamese ~ English Dictionary by Ðinh-hoa Nguyen, ISBN 0-8442-8357-6, NTC Publishing Group, 1995.
• The Cantonese romanisation found here is from the YueYin (Yuet Yam) input method as used in Twinbridge software for Chinese.
• The HakGa romanisation found here is not standardised in any form, but tries to use the initials similar to the Cantonese mentioned above.
• The ChaoZhou dialect uses the romanisation in ChaoZhouYin ZiDian (CZYZD), ISBN 7-80521-710-6, by Da2 Fu3, and Zhang1 Lian2hang2, after the September 1960 Education Ministries' guidelines. The glottal stop is represented by the -h ending. ChaoZhou and ShanTou dialects are represented by CZYZD, and the IPA transcrion is of the ShanTou dialect found in Karlgren.
• The FuZhou dialect romanised form comes from Dictionary of the Foochow Dialect, by Maclay, Baldwin, and Leger, was kindly supplied by Mike Wright
• Jerry Norman, Chinese, 1988, ISBN 0-521 +22809-3 (hdbk) +29653-6 (ppbk). Chapter 9.4, The Min Dialects.
• Two forms of Mandarin romanisation are reproduced here, Wade/WadeGiles and Pinyin as represented by the PuTongHua standard.
• Li WANG
  • HanYu YuYinShi (Chinese Phonological History), ISBN 7-8004-2017 (April 1997 reprint of May 1985). ZhongGuo SheHui KeXue ChuBanShe.
  • HanYu ShiGao (Monograph of Chinese Linguistics), ISBN 7-101-01553-0 (December 1996 reprint of June 1980). ZhongGuo SheHui KeXue ChuBanShe.
• Kanwa Jiten, Bennesse, ISBN 4-8288-0422-6, (Februrary 1996 reprint of November 1st 1990).
• KangXi ZiDian, ZhongHua ShuJu, ISBN 962-231-006-0, (March 1997 reprint of January 1958).

Vietnamese sounds and tones are written in the VIetnamese Quotable Readable (VIQR) form using ASCII keyboard characters (some information about this standard used here, can be found at http://www.nonsong.org/viqr.html).

Tones are coloured in red.

It also possesses other letters which are not found on the keyboard but which represented by similar characters that are modified by the following set of characters (in blue).

In NTC's Vietnamese English Dictionary by Nguyen, gh comes under g, ngh comes under ng.

• Nguyen's V. Dictionary as above.
• The Rough Guide to Vietnamese compiled by Lexus, ISBN 1-85828-172-8, distibuted by The Penguin Group.