Implications of the Soviet Dungan Script for Chinese Language Reform

The basic facts about Soviet Dunganese (hereafter SD) are already well known. There are about 50,000 speakers of SD who are located mostly in the Central Asiatic republic of Kirghizia while there are a lesser number in Kazakhstan and still fewer in Uzbekistan (Tsentral'noe... [1984]; Sushanlo [1989] claims upwards of 70,000). They have a very high rate of language retention, since almost 95% of them claim SD as their primary tongue (Comrie, p. 273). SD is divided into two main dialects, one with four tones and the other, which serves as the standard, with three. These two dialects derive, respectively, from Shaanxi and Gansu whence the Dungans fled from Manchu and Chinese persecution over a hundred years ago after an abortive rebellion.

The most remarkable feature of SD is that it is written with an alphabet. This would seem to give the lie to those who insist that it is impossible to write Han languages with phonetic scripts. The SD alphabet was devised at a conference convened for that purpose on May 27, 1953 in Frunze, Kirghizia (Kalimov, p. 134). It consists of the 32 letters of the Cyrillic alphabet plus 5 additional letters designed especially for sounds not in Russian. For a quarter of a century before its adoption, the present SD alphabet was preceded by a Roman SD alphabet.

There is no need here to recapitulate further the history and nature of the SD alphabet since numerous excellent studies have already been devoted to it. In chapter 5 of his seminal work on Nationalism and Language Reform in China, John DeFrancis provides a good account of Soviet efforts during the 20s and 30s to create scripts for the Dungans and other illiterate Chinese living in the U.S.S.R. He also shows how these activities of Soviet linguists and their Chinese counterparts such as Qu Qiubai had a real impact on the Latinization movement within China. It is ironic that most of the current crop of Soviet Sinologists are resolutely biased against any attempts to alphabetize Han languages and vehemently reactionary in their defense of the complicated Chinese script, Classical Chinese, and all other aspects of traditional literati culture. More recently, Heinz Riedlinger has written a very thorough and important monograph on Soviet likbez (likvidatsiya bezgramotnosti, "liquidation of illiteracy") among the Dungans beginning in 1927 and its impact on alphabetization in China up through 1988. It is, however, strictly a historical investigation and does not address the theoretical, practical, cultural, and political questions that are constantly raised by opponents of an alternative romanized orthography for China. Paul Wexler (1980) has convincingly shown how Islamic languages have enriched SD and how Soviet policies have shaped it. Detailed linguistic studies on SD have been made available by the Dragunovs, Rimsky-Korsakoff [Dyer], Hashimoto, and others. To the best of my knowledge, however, SD has never before been examined in light of the clear implications it holds for vital questions concerning current language reform in China. This is a subject of great importance, considering the continuing efforts of the Chinese government, through agencies such as the Guejia Yuyan Wenzi Gongzuo Weiyuanhui [State Language Commission] and its predecessor the Wenzi Gaige Weiytianhui [Script Reform Committee], and thousands of private citizens to find a less cumbersome script. The matter is of particularly great urgency now that an explosion of scientific and technical knowledge requires increasingly sophisticated information processing networks.

The present paper is an exercise in applied linguistics and sociolinguistics inasmuch as it focusses on the practical aspects of SD as they relate to Chinese script reform in the recent past and the near future. Hence there is no reason to enter into such purely linguistic matters as the appropriateness of the representation of the various phonemes of SD by specific Cyrillic letters. In any event, this is a dead issue for Chinese language reformers dealing with Modern Standard Mandarin (hereafter MSM) since a consistent and workable spelling scheme (the Hanyu Pinyin Fang'an) has already been in place for over 30 years. What is absolutely crucial, however, is the rich experience gained by the users of the SD alphabet with regard to the problems of word boundaries, hyphenation, tones, stress, homonyms, proper names, and all the other bugaboos that confront those who wish to provide MSM with an alternative alphabetical script.

Far and away the most signal contribution SD could make to script reform in China is in the area of orthography (early SD hsyefa, later SD orfograflya, MSM zhengcifa). Authorities in the Soviet Union and in the People's Republic of China have adopted two quite different approaches to this subject. The former would appear to believe that it is preferable to encourage regular use of the alphabet and then to extrapolate descriptive orthographical rules (Du, Yanshyans'in [1960], Imazov [1977]). The latter, on the other hand, seem to maintain that it is necessary first to establish an all-inclusive set of prescriptive rules (Wu, Bressan, Yin and Felley) and only after that to permit their widespread application. The fallacy of the latter approach is evident in the fact that even languages which have been using alphabets for hundreds of years are never able to solve all orthographical problems definitively. There is always bound to be a certain amount of disagreement on how to handle complements, compounds, idiomatic phrases, and the like. It is, furthermore, quite possible that writers of alphabetic MSM might choose one solution while SD writers might choose another. For example, in SD the subordinating morpheme di is always joined to the adjacent noun, pronoun, verb, or adjective that precedes it. MSM orthography, however, stipulates that its counterpart de stand alone. Somewhat strangely, the SD copulative verb s'i is also joined to the subject which precedes it whereas its equivalent in romanized MSM, shi, always stands by itself. Before promulgating their largely untested rules, it might be wise for Chinese language reformers to consult with their Soviet colleagues to determine how such differing usages have worked out in actual practice.

There is much that can be learned from the decades of experience gained by those who write alphabetic SD daily. Except for pedagogical purposes and scattered journals like Xin Tang, romanized MSM remains a largely theoretical construct. Having waited so long to implement Hanyu Pinyin as a functional alternative script for MSM, it would now be foolhardy not to examine carefully the lessons learned by users of Cyrillic SD in a wide variety of contexts.

One thing is certain; use of an alphabetic script to write Han languages requires a clear distinction between word (SD khua, MSM ci) and graph (SD z'i, MSM zi) or syllable (SD zhyezhyer, MSM yinjie). Square, equistantly spaced tetragraphs (MSM fangkuaizi) are patently inimical to the concept of "word." This is due to the fundamental difference between modern alphabetic scripts which are phonemic (i.e., indicating with a fair amount of precision sounds of less than syllabic length) and the tetragraphic: Chinese script which is morphosyllabic (i.e., indicating imprecisely both the sound and the meaning of syllable-length units). For example, English "telephone" consists of nine letters which taken individually signify nothing in particular, but joined together in the proper sequence determine with a high degree of accuracy, and without too complicated a relationship between sounds and symbols, the pronunciation of a specific word meaning "instrument for conveying voice or other acoustic signals over a distance." Chinese 電話, conversely, is made up of two tetragraphs that might, in different times and places, be pronounced variously as *dinh or perhaps *glins / d'ien / dian / di / dĩ / tieng / tiã / tie / tiẽ / tin, etc. and *g'wradh or perhaps *gəruats / ɣwai / hua / ho / fa / o / ik / ui / ɣo / phua / ue / ua / wa, etc., while their shapes tell us roughly that they have something to do respectively with a meteorological condition and with speech. Their association with the strings of sounds represented by MSM dianhua, SD dyankhua (much less the more common SD tyelyefon), Japanese denwa, and so forth, all of which mean exactly the same thing as "telephone," is problematic, to say the least. Only scholars acquainted with the historical reconstructions of specialists like Schuessler (pp. 127, 243, and 533) and Karlgren (nos. 385 and 302) are likely to know that there was once a closer relationship between the phonetic elements contained in the two tetragraphs for dianhua and the full tetragraphs themselves than appears in their present pronunciations. The tetragraphs for dianhua belong to the one-third whose phonetic elements now provide for ordinary readers no useful clues to the pronunciation of the tetragraphs of which they form a part. And in the case of the two-thirds in which the phonetics do provide useful information, the relationship between sounds and symbols is much more complex and obscure than in even the most irregular phonemic systems, such as English (DeFrancis [1989]).

The disparity between morphosyllabic Chinese tetragraphs and phonemic writing systems becomes even greater, and even more apparent, when the comparison is made not with a bad morphophonemic system like English but with a good phonemic system like Dungan. Let us now examine how some of the differences in the relationship of writing systems are revealed by the adoption of a simple alphabetic script for SD instead of the tetragraphs.

As one embarks upon the study of SD, perhaps the first and most prominent difference from languages written exclusively in tetragraphs that one notices is the arrangement of its dictionaries (Yanhsyans'in [1968]; cf. Yanlisyans'in [1959] and Imazov [1981]) and other types of word-lists in a single-sort alphabetical order such as that advocated by the author for MSM (Mair [1986]). No longer is one compelled to determine the radical and count the residual strokes of "head characters." As a result, looking up words in a SD dictionary is much faster than in a typical dictionary for tetragraphic Han languages. But revolutionary changes in lexicographical methods are only the start. The phonology (Imazov [1972, 1975]), grammar, morphology (Imazov [1982]), and syntax (Imazov [1987]) of SD are all treated in a fashion similar to that for other alphabetical languages. Once again, the chief reason for these starkly dissimilar methods of analysis are due to the perception engendered by alphabetization of word as the fundamental unit of discourse as opposed to syllable. It becomes quite natural in SD morphology, for example, to speak of prefixes, suffixes, and infixes and in grammar it is possible to speak unambiguously of tense, voice, and mood. Since all of these aspects of language are considered to be restricted in Chinese writing to individual semantically or modally pregnant tetragraphs, it is difficult to envisage them as being integral components of units (viz. words) of larger than syllable length.

One of the plainest divergences between alphabetic SD and tetragraphic: MSM is the operation of etymology. Before we turn to this subject, however, we need to establish a reliable transliteration for Cyrillic. Since we have already begun to encounter single SD words and will soon be reading whole passages, this is the proper moment to introduce our romanization of the Cyrillic alphabet. The system of transliteration adopted in this paper is designed to represent both Russian and SD. It should be emphasized that this is a transliteration, not a transcription (i.e. it is neither phonemic nor strictly phonetic with regard to SD). The purpose of the present system is to provide a consistent set of Roman letter conversions for the Cyrillic alphabet and the expanded SD alphabet without having to resort to awkward diacriticals and other special symbols. The same transliterations are used for both Russian and SD with the exception of е which is transliterated e in Russian and ye in SD, ж which is transliterated zh in Russian and rzh in SD, щ which is transliterated shch in Russian and hs in SD, and the eight additional letters and combinations of letters which are used only in SD. Letters in parentheses (as is the case with the transliterations for е, й, and у) may be dropped in SD when used in combination with certain other vowels or are absent from Russian (as is the case with the transliteration for э).

Transliteration Table

Cyrillic Roman
А а a
Б б b
В в v
Г г g
Д д d
Е е e/(y)e
Ё ё yo
Ж ж zh/rzh
З з z
И и i
Й й (y)u or y(i)
К к k
Л л l
М м m
Н н n
О о o
П п p
Р р r
С с s
Т т t
У у u
Ф ф f
Х х kh
Ц ц ts
Ч ч ch
Ш ш sh
Щ щ shch/hs
Ъ ъ ``
Ы ы `i
Ь ь `
Э э e(i)
Ю ю yu
Я я ya
Ә ә eh
Ң ң ng
Җ җ zh
Ў ў wu
Ү ү (y)u
Уә уә yeh
Ён ён yon
Ян ян yan

We are now prepared to take a rather detailed look at a few examples of etymological analysis in SD and then contrast them with a typical etymological excursus or two as might be found in a scholarly Chinese commentary.

DUNTS'I (III-I). Tasi grammatikashon siyundi khua. Tas'i byofeh dunzhuehdi yuyan buf'in, suikhwu vurusdi "glagol" khuani. Lyan tu yigeh khuag'in zamu gotu rzh'insh'ili (kan "dun" II).

Di eirge khuag'in dwuli bu s'iyun, tadi yis'is'i "khua", "termin", ta zei yudi grammatika terminmu ch'inf'inni tsanzhyadini, bilyun: mints'i, furts'i, deimints'i z.d., ta zei "ts'i dyan" (terminologicheskiyi slovar') khua litu ye tsanzhyadini. (Tsunvaz'i, p. 93)

verb (IIIA). A word used in grammar. It is a component of language expressing movement and accords with the Russian word glagol. We have already become acquainted with the first root above (see "dun" II).

The second root, whose meaning is "word," "term," is not used alone. It occurs as an element in some grammatical terms, such as mints'i ("noun"), furts'i ("numeral"), deimints'i ("pronoun"), and so forth. It also occurs in the term "ts'i dyan" [sic] ("terminological dictionary").

KHAN-YAN (III-1). Tu yigeh bufindi yisisi "gandi", di eirgeh bufindi yis'isi "yikhozi tsomyo" (bilyun: khuon-yan, khiyan). (TsunvazI, pp. 259-260)

tobacco (III-I). The meaning of the first part is "dry," the meaning of the second part is "a kind of grass" (for example: tobacco [plant], black smoke).

Never mind that the etymologist falters near the end by failing to observe that the primary meaning of yan is "smoke" and the derived meaning is "tobacco [plant]." It is heartening, notwithstanding, to witness his attempt to explain the word khan-yan as a whole

In opposition to this concentration on the word in SD etymologies, when we refer to etymology as it is applied to Han languages written with the tetragraphic script, we intend the decomposition of single graphs into their constituent elements. This is the usual sort of tetragraphic etymologizing as it is carried out paradigmatically in the Shuowen Jiezi [Explanation of Simple and Compound Graphs] completed by Xu Shen in the year 100 of the International Era. The first case is that of MSM ming which, as every schoolboy knows, can be broken down into pictographs for "sun" and "moon," hence "bright." This is wrong, of course, because the oracle bone, bronze, and seal forms of the graph show the moon shining through what seems to be a window. Once the word ming ("bright") was assigned to this particular graphic configuration, various permutations could be worked upon it, thus meng ("alliance" -- the modern semantic classifier is "vessel" but the original form showed "blood"), meng ("to bud, sprout" -- the semantic classifier is "grass"), and so forth. "Alliances" and "budding" assuredly have precious little to do with the moon shining through a window, and yet the tetragraphs used to represent them embrace a tetragraph which apparently does convey that sense but is manifestly only being used for its phonetic value in the expanded forms. The real mystery concerning ming ("bright") lies in our knowing next to nothing about it before it became attached to the tetragraph .

The second case is that of the MSM homophone ming ("name"). Here we are on still less firm ground than with ming ("bright") concerning the derivation of the tetragraph with which it is written. Among the various more or less frantic guesses are: mouth plus evening, because it is necessary to give one's name when it is dark (!); sacrificial meat plus the vessel in which it is placed, because a child is named at a ceremony held three months after it is born (!); loud sound plus mouth, because one calls out his name (!); and so on. Fortunately, our understanding of the antecedents of the word that lie behind the tetragraph is much solider. There is little doubt that it is cognate with various other words for "name" in Central, South, and Southeast Asian languages: Gyami minn, Gyarung (tir)ming, Takpa myeng, Manyak ming, Tibetan ming, Sherpa min, Gurung ming, Munni min, Magar ming, Thaksya min, Limbu ming, Chepang myeng, Bhramu min, Vayu ming, Bhutani ming, Bodo mung, Dhimal ming, Garo mung, Tablung Naga min, Namsang Naga min, Singpho ming, Burmese (a)min, Pwo-karen maing, Toungh-thu min, and dozens of phonologically related words in languages that are remote from the realm of tetragraphs (Hunter, p. 146). There can be no doubt that Chinese ming ("name") came from an ancient Asian root that predates the tetragraph and can have had nothing whatsoever to do with it. Judging from the tentative archaic (early Zhou period) reconstruction *myaŋ < *ymaŋ, it is conceivable that the Chinese word for "name" is linked to a Proto-Nostratic (c. 15,000-10,000 BIE) root that encompasses languages spanning the entire Eurasian continent: Church Slavonic imę, Serbo-Croatian ime, Bohemian jméno, Polish imię, Russian imya, Old Prussian emmeno, Armenian anun, Albanian émën, Hittite lāman (with dissimilation) (Buck, pp. 1263-1264). Considering the manifold gradations of the Indo-European root (*enmen-, *n̥men-, *nō̆men, etc.), it is evident that English "name" and all of its IE cognates (cf. Sanskrit and Avestan nāman-, Tocharian A ñom, Tocharian B ñem, Greek ónoma, to mention only a few) as well as Finnish nime-, Lapp namma, Japanese namae, and Hungarian név are probably also derived from the same Proto-Nostratic root. The above data clearly prove that is only adventitiously linked to the string of sounds pronounced ming which has a history that long antedates the tetragraph. What matters is the word; the graph is only an arbitrary vehicle that serves to record it. Apart from the word ming, the tetragraph is meaningless. The nearly universal belief that there is an inalienable semantic bond between Han words and the tetragraphs which are used to write them is in serious need of reexamination.

Though we have strayed far from SD, these etymological ruminations are pertinent because the shift in focus from tetragraph to word occasioned by the alphabetization of this Han language has led to an entirely new attitude about the ultimate location of meaning. To summarize, linguistic meaning lies in the word, not the tetragraph which is originally but a device for recording the word. This is not to deny that the tetragraph is capable of assuming a semi-independent role of its own (e.g. as an aesthetic object in calligraphy), only to attempt to reassert the rightful priority of spoken language over script. Traditional Chinese etymologies are concerned almost wholly with explicating the meanings of single tetragraphs insofar as they can be discerned from analysis of their visual shapes. The sounds of the tetragraphs are not normally taken into account as information that is of fundamental significance in extracting their meanings.

The deficiencies of this procedure, even for elucidating single graphs, are patent. It is not without reason that Liu Xi in his Shiming [Explanation of Terms], completed sometime before 273 IE, attempted to pursue a more purely phonological approach. Unfortunately, his own method failed still more spectacularly than did that of Xu Shen because he had no means to annotate the sounds of the tetragraphs than by resorting to other tetragraphs (whose meanings tended to get in the way of his analysis) nor did he have a sufficiently sophisticated understanding of the phonology of archaic Chinese (the stage when it first came to be written down in tetragraphs, more than a thousand years before his own time) to determine true cognates. This resulted in an enormous circularity completely lacking in philological rigor and validity. To give only a few examples, Liu Xi claimed that xing ("star") actually meant san ("scattered") because the two words sounded somewhat similar when he was compiling his dictionary. Likewise, he believed that dong ("winter") meant zhong ("end") and chun ("spring") meant chun ("[time of] wriggling worms") for the same type of strained reason. A more critical deficiency of traditional Chinese etymology is that neither Xu Shen nor Liu Xi was able to account systematically for the meanings of polysyllabic words such as junzi ("princely man"), daolu ("way"), shanhu ("coral"), xishuai ("cricket"), pipa ("[Persian] oud"), and so on. While the evidence is too intricate to present here, it can be demonstrated that all of these early words were originally binomes or monosyllables with consonant clusters that had to be broken up if they were to be written with tetragraphs. By the Song period, Zhang You (b. 1054) had become aware of the existence of so-called lianmianzi ("conjoined graphs [sic words]," also referred to in modern Sino-English linguistic terminology as "disyllabic roots") such as paihuai ("hesitate"), qiaocui ("pallid"), hulian ("lingering"), pufu ("crawl") and so forth. The origin and nature of such terms, which number in the thousands, are still imperfectly understood. During the Qing period, gigantic collections of polysyllabic expressions were compiled (Peiwen Yunfu [1704], Pianzi Leibian [1719]), but no attempt was made to provide definitions or etymological analyses. Even today, polysyllabic Chinese terms are more often than not designated by linguists as "compounds." All of this serves to underscore the power of the tetragraphs to influence one's view of language.

Once Han languages are divorced from the tetragraphs, all of this morphosyllabic centering ceases. Cyrillic SD compels the etymologist to look at the function of syllables in Chinese words in ways that are very different from the normal tetragraphic approach. Even the elaborate bound and free analysis of Y. R. Chao and L. S. Yang cannot really account for the presence, for example, of the syllableddo in such words as MSM didao ("genuine"), gongdao ("fair, impartial," cf. SD gundo), weidao ("flavor, interest," cf. SD vido), and zhidao ("to know," cf. SD zh'ido). Stripped of its tetragraphic carrier whose basic meaning is "way," the syllable dao offers a challenge to those who wish to comprehend its multifarious usages. Ultimately, it will be discovered that dao as the second or subsequent syllable of a word often has no legitimate connection at all with "way" or its derived meanings. The tetragraph with that meaning has simply been chosen for convenience to carry the sound dao whose source lies elsewhere in the spoken language. This is even more true of MSM syllables such as la (usually written with the tetragraph meaning "pull") and luo (usually written with the tetragraph meaning "fall, drop") where it is completely impossible to offer even a forced interpretation based on the superficial meaning of the tetragraph as it is with dao. Frequently, we are forced to admit that, given our present level of understanding, we just do not know the origin of SD words such as dunhsi ("thing") (Tsunvaz'i, p. 93) which is written with the tetragraphs for "east" and "west" in MSM.

Having investigated in some depth the vital place of the word in SD, we may now move on to longer utterances. In order for those who are not familiar with the Cyrillic script to be able to gain a direct impression of the language and the way it works, several transliterated sketches will be given together with their English translations.

Ch'inshon In the City
Zuehr veh Iyan Kheichehr nyonnyon zei ch'inshon lonni. Bazarshon khoshodi mashnei dei cheh. Shonvu Kheichehr nyonnyon ba veh lindo yig'i da chonz'inili. Chonz'ini khosho rzh'in du peidi dui zudini. Tamu du zhudi khuho dei twuyon. Veh Kheichehr nyonnyon fehde. Zh'is'i Khun chonz'i. Yesterday Aunt Kheichehr and I were strolling in the city. There were a lot of trucks and cars at the bazaar. At noon Aunt Kheichehr took me to a big open area. In the open area lots of people were walking in ranks. They were holding up slogans and paintings. Aunt Kheichehr told me that this was Red Square.
(Du [1959], p. 48)
Khun chi Red Flag
Hsyuehs'iwnu du peili dui, zandi chichirdi. M'inm'indi Chish'ir lyan lyong'i yatu zhudi khun chi chwuiei zando vamudi tunili. Da khun chishon hsyedi: "Sh'iyueh gehmin vansui!" The students were all standing neatly lined up in rows. Suddenly Chish'ir and two girls came out holding aloft a red flag and stood before the children. On the red flag was written, "Long live the October Revolution!"
(Du [1959], p. 48)
Hsyatyan Summer
Hsyatyan tyanchi chon,
Hwuzhya bu hsinkhuon.
Tanni tei chinhsyuan,
Chyochyor luan zhyokhuan.
Zhuonzhya ye kho kan,
Hwuzhya ye hsikhuan.
In summer the days are long,
The farmers are at ease.
In the fields all is freshness,
And birds are calling everywhere.
The crops are beautiful,
And the fanners are happy.
(Du [1959], p. 49)
Fichin Ferma Poultry Farm
Vehmu kolkhozdi fichin ferma zei kheizi byannini. Ngeh dei yaz'i du zei kheizi litu gohsindifudini. Tyanngeh ye zei kheizi lituni. Our kolkhoz's poultry farm is next to a lake. Ducks and geese swim merrily in the lake. There is also a swan in the lake.
B'ii zhir, lyan hsyueh yiyon, du zei kheizi yanshonni. Yigi ngehloyin fidi leili, zhir luan du zhyokhuantuehli. Chickens white as snow are along the shore of the lake. A hungry hawk comes flying toward the chickens and they run away clucking wildly.
Zhivar donvehr du chondo sh'in tso lituli. The chicks immediately rush into the deep grass.
(Du [1959], pp. 49-50)
Radio Radio
Ganzo veh tinfl radioli. Radiofehdi gunfu da. Nwurzh'in lyan nanrzh'in khuandi, khuandi fehdini, Myan fehli yizh'inzi zavodshon zwuli duehsho traktordi, fabrikashon zh'ili duehsho bupidi sichin. Zikhu khuatur doli kolkhozshonli. Ba vehmu rayionshon Aun Auonshyadi s'ichin ye fehli, fehs'i vehmu kolkhoz kehzhya ba 300 gektar lyonsh'i zhonshonli. Danlindi yi, lyong'i kolkhoz khan gan vehmu zhondi dueh. In the morning I listened to the radio. The radio broadcasted for a long time. Male and female announcers alternated, speaking first for awhile of such things as how many tractors were made in a factory and how much cloth was woven in a mill. After that the subject changed to kolkhozes. They also mentioned the planting of crops in our district, saying that our kolkhoz had already planted 300 hectares of grain. One or two other kolkhozes had planted even more than us.
Shukur fehdi: Zh'i dus'i pindi kommunist partiyadi linshu zamu, d'iikhadi sh'in. In conclusion, they said that we had achieved all of these victories through the leadership of the Communist Party.
(Du [1959], p. 50)

From the above passages, it is readily apparent that Cyrillic SD permits discussion of a wide variety of subjects. It is also capable of more literary applications as well. Folk songs (Khasanov, Rimsky-Korsakoff Dyer [1987]), legends (Rimsky-Korsakoff Dyer [1981-1983]), tales (Riftin), collections of proverbs (Yusurov), and other types of literature including short stories and novels are possible in SD. One of the most interesting developments is in poetry where the typically blockish syllabic structure of traditional Han verse is no longer evident:

Vehmwus'i minda shonyan
Sovyet guonyinshondi
Zei keikuardi guyitwuni
Lyan khuar yiyon zhonni
We are fortunate youths
During the Soviet period
In a land that is blossoming
Growing like flowers.

Ya. Hsivaza (Du [1957], pp. 284-285)

Many different poetic genres and effects have now become feasible that were previously ruled out by the mandatory equidistant spacing of tetragraphic syllables. Conversely, customary poetic devices such as parallelism are no longer relied upon so heavily as they were in traditional tetragraphic verse.

To give some indication of just how much SD diverges from MSM, we may compare the opening from a short story by Lu Xun to its translation in SR Note that a simple transcription of MSM into SD would be both unidiomatic and largely unintelligible to SD speakers who are unfamiliar with MSM. Translation into English of the MSM original is also provided as a reference point.

Zhufu Chyuzhi Yiunchi

New Year's Blessings
Jiuli de niandi bijing mi xiang niandi, cunzhenshang bubi shuo,jilu zai tiankongzhong ye xianchu jiang dao Xinnian de qixiang lai. Huibaise de chenzhong de wanyun zhongjian shishi fachu shanguang, jiezhe yi sheng dunxiang, shi songzao de baozhu; jinchu ranfang de ke jiu geng qiangliele, zhen'er de da yin hai meiyou xi, kongqili yijing sanmanle youwei de huoyao xiang. Wo shi zhengzai zhe yi ye huidao wo de guxiang Luzhen de. Suishuo guxiang, raner yi meiyou jia, suoyi zhide zhan yu zai Lu Si laoye de zhaizili. Ta shi wo de benjia, bi wo zhang yi bei, yinggai cheng zhi yue "Sishu," shi yige jiang lixue de lao jiansheng.

Lo litwudi linyirdi nei yityan tei hsyon gueh Hsin nyan zhyechidik'in Lwuzh'in zhuonz'ishon guon ye bus'i vi zhyeyin Hsin nyan monkhuondini, tadi chis'ii litu ba Hsin nyandi vido du n'in v'inzhyan.

Tyechin yuntsei zhunt'int'indi zei banhsyukunni dyodini. Yuntsei litu khuehzhyan bu zhwudyar rzhodadini. Lyan khuehzhyan yitun, na z'i khwukhadi danz'i yis'ir, yis'ir ye zei kunzhunni byedi, gi hsyonsh'indini. Zh'is'i ba zhyanidi khehda sundo tyanshon, zhyo zwu mannyandi zunzhye bogochini. Z'i danz'i zhyali zhin byetuehli, kunzhunni hsyonsh'in yuehhsin duehkhali. Chis'ii litu khuehyueh vido ye chwuleili.

Gueh Hsin nyandi nei zhityan veh dondor doli Lwuzh'in zhuonz'ishonli. Suirzhan zh'is'i s'in-yonli vehdi zhuonz'i yeba, kehs'i zhehr chyuanli mehyu vehdi chinchin-lwuzhyanli, yinvi neigeh mehfor veh zando Lwu S'izhyali. Ta bi veh dadi bonzhyer yileiz'ini yinvi neigeh, veh ba ta ch'inkhwuligeh zhyuzhyu. Lo guonyinshon tas'i kuehhsyuehzh'in, zochyan zei D'iigui kuehhsyueh akademiya litu zwugueh kheuh. (Eirbudwudi, p. 3)

New Year's Eve of the old calendar, after all, seems more like the end of the year. Even in the air, to say nothing of the towns and villages, there is an atmosphere of the approaching New Year. Light flashes sporadically amidst the heavy , gray clouds of evening, followed by a rumbling reverberation from the firecrackers for sending off the Hearth God. Those that explode nearby are even louder. Before the deafening sound dissipates, the faint smell of gunpowder fills the air. It was on this very night that I returned to myoId hometown of Luzhen. Although I call it my hometown, there was no home left, so all I could do was stay at Mr. Lu the Fourth's house for awhile. He was a member of my clan, a generation older than me, so I should have called him "Fourth Uncle." He was an old student of the imperial academy who stressed Neo-Confucianism.

(Lu Hsun [1973, originally completed February 7, 1924], p. 3)

It is obvious from these passages and from a great deal of other evidence that SD, while clearly related to MSM, in many respects is as different from it as Dutch is from German, Spanish from Portuguese, Russian from Ukranian, or Hindi from Urdu. If we were to examine comparable passages from MSM and current SD dealing with politics, science, or religion where borrowings from other languages are frequent, the contrast would be even sharper. In terms of morphology, lexicon, grammar, syntax, and idiomatic usage, there are significant disparities between MSM and SD. It is unfortunate that many Sinologists who have studied SD, including some who have published most extensively on the subject, are incapable of reading SD on its own terms. Instead, they try to force it into an MSM straitjacket by equating SD syllables with tetragraphs. By erroneously assuming that all SD morphemes can be written with appropriate tetragraphs, they often resort to ludicrously forced equations. Consequently, their translations into Russian and English, which are based on these faulty tetragraphic transcriptions instead of on the original SD texts, are extremely unreliable. The simple fact of the matter is that SD morphemes are not always identifiable with any of the 60,000+ tetragraphs. Speakers of Dungan have bitterly complained about this myopically procrustean procedure which not only distorts their literature but seriously misrepresents the very nature of their language.

The patent differences between SD and MSM inevitably lead to the question of their linguistic relationship to each other. Are they dialects or separate languages? The Dungans insist that SD is an independent language, not a dialect of Mandarin, and, indeed, even deny vehemently that they are Chinese at all. The Chinese, contrarily, assert that there is only one Ran language and that SD, Cantonese, Taiwanese, Mandarin, and so forth are but dialects of it. This is, of course, much too sensitive and complicated an issue to be discussed adequately here (see Mair [1987] for a more thorough investigation), but we may note that the criteria for the Chinese lumping of many languages as a single language in general are more political than linguistic. To defuse the issue slightly, from the Chinese point of view, we may refer to MSM, Cantonese, Taiwanese, and even SD -- if we accept the Dungan view -- as "topolects" (the exact English equivalent of the Chinese designation fangyan) which makes it seem slightly more neutral. From the Dungan point of view, we must concur that there are many features of SD which set it apart from MSM, not the least of which is its extensive borrowing from Arabic, Persian, Kazakh, Kirghiz, Uighur, Uzbek, Turkish, and latterly Russian. Even very common words such as the names of the days of the week are totally unrelated to MSM: Shanbe ("Saturday"), Yekshanbe, Dwushanbe, Hsyesfianbe, Chashanbe. Panshanbe, and Zhuma ("Friday"). These are derived from Persian Šambe, Yekšambe, Dosambe, Sešambe, Čāharšambe, Panjšambe, and Arabic Jom`e.

This leads to several very important phenomena concerning word borrowing in tetragraphic scripts. In the first place, it requires the syllabification, distortion, or partial dropping of all consonant clusters ("masochism" mazhaofuzhuyi, "Kropotkinism" Kelupaotejinzhuyi, "gram" gelamu, "clutch" kelazi, etc.). Secondly, the tetragraphic: components employed are prone to cause various types of semantic dissonance, whether felicitous or not ("miniskirt" miniqun ["skirt which entrances you"], "husband" heiqibandeng ["black lacquer board stool"], "Broadway" Bailaohui ["place where all the old people congregate"], etc.). Thirdly, there is a tendency to abbreviate so drastically that any connection with the original term is hopelessly obscured ("aluminum" lyu, "manganese" meng, "uranium" you, "Bodhisattva" Pusa, "speculation" siban, etc.). Fourthly, multiple forms cause mass confusion. Here we need only mention that 22 different renditions of "microphone" have been reported (maikefeng, maigefeng, maike, mi, huatong, chuanshengqi, kuoyinqi [also used for "megaphone," "amplifier," and "loudspeaker"], weiyinqi, etc.). Fifthly, borrowing is discouraged in favor of the creation of calques, neologisms, and translations ("thermometer" wendubiao ["warmth-degree-gauge"], "geometry" jihe ["how much," i.e. "quantity," coined by Xu Guangqi, 1562-1633], "eugenics" youshengxue ["superior-birth-study" < Japanese yseigakul or shanzhongxue ["good-seed/race-study"], "Xerox" quanlu ["complete-record"], yingyin ["shadow-print"], fuyin ["reprint"], etc.].

The situation is altogether different with SD which borrows freely and naturally without any tetragraphic masking or interference. Above the nameplate of the Dungan newspaper Sh'iyudi Chi [October Banner] is the following slogan: "Chyuan sh'izhyedi proletarmu, lyankhehchyelei! [Workers of the world unite!]" It will be observed that the borrowed term "proletariat" is readily capable of taking a plural suffix. Borrowing thus becomes productive within the morphological framework of SD. Another example is that of kolkhoz, also borrowed into English, which is a Russian contraction (from kollektivnoe khozyaistvo) meaning "collective farm." In SD, it is used flexibly without any distortion from tetragraphs. Kolkhoztnu means kolkhozes, kolkhozzhya (where zhya is a suffix signifying a person who follows a certain profession or occupation) refers to an individual belonging to a kolkhoz, kolkhozzhyamu means "members of a kolkhoz," and kolkhozzhyamudi means "belonging to the members of a kolkhoz."

Given the terrific direct borrowing capacity of SD and the fact that it was already thought to be "nonstandard" (from the MSM vantage) when its speakers left China over a hundred years ago, one is bound to be frustrated if one thinks in terms of Chinese tetragraphs when trying to read and write this language. What is more, the longer SD continues to be written in a fully phonemic script, the more difficult it will be to write in tetragraphs. New words are constantly emerging in all Han languages without any necessary connection to the tetragraphs. As Robert Cheng has shown for Taiwanese and Robert Bauer for Cantonese, there are many instances where it is impossible to write these languages in tetragraphs. The speech of Chengdu, classified as a Mandarin dialect, is chock full of words that would be completely incomprehensible to someone who was familiar only with MSM and for which there are no secure tetragraphs: zuazi ("what [are you] doing?"), jinjin ("torn cloth"), diaoqiao ("to be exceedingly choosy"), and so on (Luo Yunxi). In the rural areas of Sichuan, the situation is even more hopeless, so that someone who is conversant only in MSM cannot fathom the local topolects at all. Even in Pekingese, the supposed model for MSM, there are hundreds of common expressions whose tetragraphic form is not fixed (Chen Gang). In such instances, one is often forced to make arbitrary choices (e.g. gai maor ["splendid" or "to block a shot in basketball"], cuiber ["lackey"], qiemer ["stage decoration for traditional opera"], etc.). Numerous expressions used in the daily life of Peking Muslims are still considered by many of them to be irreducible to fixed tetragraphic segments (sab ["thanks"], wubair ["congratulations"], niyaht ["alms"], koufan ["vow"], wus ["bath, shampoo"], and dozens of others) (Jin; Hu; also see Wexler [1976] for a study of Persian, Arabic, and other borrowings in various Chinese languages and dialects).

Let us now address some of the potential disadvantages of the romanization of MSM in light of SD. Perhaps the one aspect of phoneticization that frightens language reformers more than any other is the presumably intolerable levels of homonymy that will ensue. There is no doubt that romanized Classical Chinese would be gibberish, but that is because it is not a spoken language. Classical Chinese can only be read with the aid of the semantic components of the tetragraphs; their sounds alone will not suffice. Such is not the case with the vernacular Han languages which are used by millions for oral communication without any reference whatsoever to the tetragraphs. It is a truism that people do not speak in tetragraphs. What they speak, rather, are strings of sounds without any visual components except sporadic nonessential gestures. Anything that can be spoken and understood without ambiguity can also be recorded phonetically and understood without ambiguity so long as the same amount of context is provided. In the romanization of MSM, however, there are several caveats that need to be made. The written model must first of all be truly vernacular. Classical (wenyan) or semiclassical (banwenbanbat) styles simply will not work. Only those classical expressions and foreign terms that are immediately recognizable when spoken can be included without explanatory notes. That is to say, they must be firmly embedded in the living language of the people or, in other words, they must be "sayable."

A key lesson to be learned from SD is that romanization of MSM must take advantage of the polysyllabicity of the language. The average length of a word in MSM, the monosyllabic myth notwithstanding, is almost exactly two syllables (Mair [1986], p. 140). Once spelled out polysyllabically, even without taking into account the tones, there are not nearly so many homonyms in MSM as commonly believed. Indeed, if Han languages had the sort of homophone problems attributed to them by opponents of phoneticization, there would be unworkable ambiguity in speech. Were this to be the case, surely the users of Han languages would have the resources to remedy such a serious obstacle to effective communication.

This leads to the question of how to represent tones in romanized MSM or, indeed, whether they need to be indicated at all. Once again, the practices of SD are instructive and merit investigation by Chinese language reformers. Those who designed the SD alphabet wisely refrained from requiring the indication of tones in running text. There are several advantages to this, not the least being that it is easier to type the language without having to insert ungainly diacriticals, numerals, or extra letters that are also hard to order in computer sorts. It also means that dialectical differences in tonal usage are not brought into conflict when standardizing the script. Tones are, however, specified in dictionaries (they are even given for borrowings such as brigadir I-I-III), in much the same fashion that we provide pronunciation guides in our English dictionaries. SD also has an elaborate system of stressed and unstressed syllables, somewhat comparable to full and neutral tones in MSM or accent in English, but these are not designated in the orthography nor in dictionaries. It is expected that the native reader will automatically produce the correct tones, sandhi, and stress when vocalizing a text. This is, indeed, what transpires in actual practice. Presumably, as in Russian texts for beginning students, fuller indication of the minutiae of pronunciation might be provided for learners of Cyrillic: SD and romanized MSM as well. We should remember that tetragraphs do not even provide an accurate guide to the pronunciation of the syllables they are meant to represent, so they are often supplied with phonetic annotations (bopomofo [zhuyin fuhao], pinyin, furigana), usually including indication of tones where appropriate, for the novice.

It is interesting to see how SD handles words which are often considered problems in using Pinyin to write MSM. There are supporters of Pinyin who insist that with non-representation of tones, and sometimes even with tone representation, there would be so much ambiguity that it is necessary to distinguish a few frequently used words by creating distinctive spellings, as in the case of zai ("at") versus zay ("again"), you ("have") versus iu ("again"), and xiang ("toward") versus xiaang ("think"). In SD orthography, tones are never marked in running text and no special spellings are used to distinguish homonyms. As they devised their orthography, Dungan writers realized that homophonous confusion could be avoided, instead, by careful attention to context and by substitution of polysyllabic words when necessary. The following quatrain will serve to show that, even in highly elliptical verse which is fraught with inversion and other poetic effects, homophony does not pose a problem so long as the author is attentive to making himself clear:

Vehmu kolkhoz yu yigeh rzh'in,
Da chui, mazhon bu nanv'in.
Tinzhyan zwu khueh ba bin zhuon,
Yityan do khi yu s'i fon.
There is a man on our kolkhoz
Who is always fighting, never stops running around;
He hears the workers loading ice,
But the whole day long pokes his nose into things.
(cited in Sushanlo and Imazov, p. 48)

Taken by themselves, many of the words in this quatrain are indeed ambiguous: yu ([tone] I stroll, go, wander, travel; oil, grease, fat; II have; III right; again, still), rzh'in (I person; II bear, endure, suffer; III recognize, acknowledge), da (I catch, hunt, place, put, set, detain hold, keep, increase, throw on or over; father, from, out of; II beat, strike, hit, and dozens of idiomatic usages; III big), chui (I pound, beat), ma (I mother; hemp; take [away]; play; numb; II horse; III curse, scold, swear), zhon (I open [up or wide]; look at with wonder; II to grow; support, maintain, control; shoe; III to swell, blow up; a unit of length; account [book]), and so forth for all of the monosyllabic words.

The same phenomenon exists in other languages as well. Take, for example, the English sentence 'We can ring up the operator right away and have her tell the highway patrol that a drunk bear from the state park is creating a traffic jam at the cloverleaf." Over half of the words in this sentence are possessed of a plurisignification that can only be disambiguated through juxtaposition with other words in set semantic structures. It would take a perverse or seriously deficient reader to insist that "can" here signifies "metal container" or "toilet," that "ring" signifies "a small circular band," and so on. Conversely, unless an author is being intentionally obscure or is inattentive to the needs of his reader, he will naturally employ various devices at his command to ensure that his message is conveyed accurately. Thus, if there is any chance of misunderstanding ma ("horse") for ma ("mother") in a given sentence, SD authors will choose mama to specify the latter. The same holds for dada instead of da ("father") when the situation warrants greater specificity.

To show how the writer constructs his language in such a fashion that multivalent components acquire explicit meanings, let us examine briefly a few of the collocations in the quoted verse. In the first line, the verbal position of yu permits it to mean "there is" or "to have" but not "oil" or "again." Because rzh'in follows a number plus attached measure word, it can only mean "person," not "endure" or "recognize." Da and chui together constitute a virtual compound that conveys the sense of "fight." Similarly, ma and zhon following one another and occurring just before a verb with its attendant adverb, if any, are limited to the single interpretation "horseshoe." Bu coming before a stative verb ("to remain quiet") must be the negative adverb instead of the verb "repair" or the nouns "cloth" and "step." Zwu and khueh together can only mean "work" which enables the reader to understand that the final three words of the third line must refer to a common form of labor in the kolkhoz. Thus ba becomes the pretransitive marker instead of "to pull out," "to climb," "uncle," "eight," "to guard," "to take," "dam (usually occurs only in the form fiba)," or "harrow." The object bin and its accompanying verb zhuon then are readily comprehended as "ice," not "soldier" or "ill," and "load" not "decorate," "pretend," "stake," "stout," "fill," or "plump. Yityan do khi is a common expression which prohibits the misinterpretation of do as "to peck," "knife," "overturn," "to thrust," "to change," "pour," "road," or "island" instead of "up to (khi can only mean "dark[ness]"). Fon s'i is a closely bound phrase meaning "investigate affairs," which eliminates about a half dozen other possibilities for each of the two syllables if they were isolated. The phrase is here inverted for the rhyme. In contrast to its verbal function in the first line, yu here plays an adverbial role.

All of this seems elaborate and complicated when explained step by step, but for an experienced reader of SD, the correct choices are made automatically and without hesitation. It must further be remembered that alternative meanings for each of the given monosyllables have their own delimiting constructions. For instance, do in the sense of "knife" will usually follow immediately after a measure word and is often combined with another syllable, which narrows the range of its meaning, hence mado ("sword") and dobaz'i ("knife haft"), etc. The adjectives ("sharp," "dull") and verbs ("slice," "cut") with which it customarily appears also make clear that do means "knife" not "island" or the like.

Those who stubbornly maintain that SD must be unworkably ambiguous because of homophony merely display their own condescension toward the Dungans. In effect, they are saying that all of the stories, poems, textbooks, and newspapers of the Dungans do not make sense. It is hard for a sensitive observer to impute such gross stupidity to a people as to imagine that day after day for over half a century they would read and write voluminous nonsense. Yet there are indeed many prominent Sinologists who entertain such a preposterous opinion. Most of them do not know a single word of SD, and the rest only pretend that they do. In truth, they are singularly unqualified to pontificate on the workability of SD. The Dungans, who are fluent and literate in their own language and who use it for a variety of useful purposes, know better. Their quiet and persistent faith in their own script exposes the folly and intolerance of the uninformed experts.

The same people who doubt the practicability of SD are even more vociferous in their denunciations of Pinyin MSM. Oblivious of the fact that the Chinese navy has been using Pinyin -- without tones indicated -- to send semaphoric and other types of messages for over two decades, that New China News Agency reporters overseas file their stories in Pinyin, that Pinyin is routinely employed in machine translation research, and that much private correspondence is already being carried out in Pinyin, they declare that a romanized orthography is impossible for MSM. To put the matter bluntly, the experts are wrong. Pinyin MSM is already a reality among certain restricted circles. Its versatility, expressiveness, and utility have been proven repeatedly though not among the broad populace. Still, there is much that proponents of Pinyin MSM might learn from the history and practice of SD. For example, instead of proposing separate spellings for mǎi ("to buy") and mài ("to sell") or worrying about the need for diacriticals in running text, they might consider the Dungan practice of substituting polysyllabic words when necessary. Thus mei II replaced by meishon ("to buy") and mei III is replaced by meidyo ("to sell"). Likewise, MSM could use mailai, goumai, shoumai, shougou, dinggou, dinghuo, caigou, etc. for mǎi , and maidiao, chumai, chushou, huomai, xiaoshou, neixiao, waixiao, jingxiao, jingshou, etc. for mài depending on the circumstances. In most cases, the context is sufficiently clear that mai alone will be enough. Nor do such apocryphal stories as the misshipment of xiāngjiāo ("bananas") for xiàngjiāo ("rubber") constitute an obstacle to the Dungans, who use the international terms rezina or kauchuk for rubber and banan or bazhyo for banana. If there is any likelihood of a mistake (which seems highly unlikely anyway), even MSM can easily avoid the problem by specifying renzao xiangjiao, hecheng xiangjiao, ziran xiangjiao, shujiao, dadoujiao, alabojiao, dingben xiangjiao, etc. for specific types of rubber and reserving xiangjiao for banana, adopting the scientific term ganjiao or using the international word banana (often heard in Hong Kong and Singapore). The point is that conscientious writers, like thoughtful speakers, are resourceful and fully capable of molding their language into a tool for communicating their thoughts and ideas efficiently and accurately. Above all, advocates of Pinyin MSM should emulate SD by adopting a thoroughly vernacular style. Bastardized semiliterary Mandarin texts are pretentious, infelicitous, and ineffective for precise communication.

Another highly personal qualm besetting those who contemplate the alphabetization of MSM is that names and surnames will no longer be distinguishable. Many Chinese worry that, since there are only an extremely limited number of syllables in MSM (between 398 and 418, depending on dialect, or 1,277 if tones are taken into account -- DeFrancis [1984], p. 42 -- compared to over 8,000 in English) and since Chinese names are either one or two syllables in length, they would be reduced to drab monotony. Here, as well, SD allays our fears. First of all, it must be reiterated that when Chinese address each other, they do so with strings of sounds, not with tetragraphs, so the tetragraphs do not serve to disambiguate names and surnames except when written or described visually. With SD, a very interesting phenomenon occurs. Han names and surnames are modified by the addition of Russian endings or Islamic names are adopted. It also becomes easier to identify an individual as male or female on the basis of his or her name (in contrast to the perennial complaints against the genderless quality of most Chinese names). In any event, the range of possibilities for naming oneself and one's children are actually much broader in SD than in MSM and other tetragraphic Han languages (see, for example, Yanhsyans'in [1968], pp. 160-169).

Foreign personal names and place-names also benefit from alphabetization by not having to undergo the syllabic deformation imposed by the tetragraphs. For example, in SD we find Nikolayi Mikhayilovich Przheval'skiyi instead of MSM Nikelamikayiluoweichipuerrewaersiji, Abdurakhman instead of Abudulaheman, Tokmak instead of Tuokemake, S'ir-dar'ya instead of Sierdaliya, Armyan instead of Aermingniyaren, Gruzin instead of Gelujiyaren. This is a decided boon in an increasingly international world which. shares essentially the same geographical, scientific, technical, social, and economic terminology.

Conclusion

When they first arrived in Russia over a century ago, except for a few religious instructors (SD akhun, MSM ahong = Persian ākhūnd) who were able to read parts of the Koran in Arabic, virtually the entire population of the Dungans was illiterate. Now the adult population is almost wholly literate, but in a way their ancestors would have found hard to predict, for the Soviet Dungans read neither Chinese nor Arabic. Instead, they read a combination of materials written in SD and in Russian, Kazakh, Kirghiz, or Uzbek -- all in the Cyrillic alphabet. In spite of their small population base, the Cyrillic alphabet has served the Dungans well in helping them to preserve their language and their identity. They are prosperous, their population is growing steadily, and they are respected as superb kolkhozzhyamu, not to mention other professions in which they have succeeded.

The purpose of this paper has been to discuss some of the advantages and disadvantages of cyrillicization of SD but, above all, to suggest that the experience of the Soviet Dungans with their alphabet merits the most intense scrutiny of Chinese language planners, reformers, and private activists. The Soviet Dungans are by no means the only speakers of a Han language to have experimented with a phonetic script. Thousands of elderly people who were taught the vernacular script of South Fukien (Minnan Baihuazi) by Christian missionaries during the first half of this century are still literate only in that alphabetic script (Huang). The sisterhood of Jiangyong county, Hunan province were clever and determined enough to devise their own syllabary known as "Women's Writing" (Nyushu) and based on skewed forms of a limited number of select graphs. It is likely that there were many other similar attempts to create demotic phonetic pts in China. The fundamental difference with SD is that it grew up outside of the Chinese polity and hence, far from being suppressed by government and elites, it has been actively fostered Soviet authorities.

Several general points need to be emphasized before drawing this paper to a close. The most important one is to recognize that Chinese language reformers are faced with a genuine dilemma, to wit, which should come first, unification of the spoken Han languages or romanization? The study of SD has shown how very different the so-called "dialects" of China are -- and here we are dealing with a language that is ostensibly (or was once) a very close relative of Northwest Mandarin. So far, research on the topolects has been restricted almost entirely to phonological studies. More intensive investigations on the grammar, lexicon, idiomatic usage, and syntax of the topolects is sure to reveal startling disparities among them (cf. Moser's insightful comments in his 1985 book). Sanders has already rightfully called our attention to the great differences between MSM and all Han dialects, topolects, and languages as they are actually spoken in real life.

The contributions of the tetragraphs to the cultural and political continuity of China are undeniable. At the same time, however, they have inhibited unification of spoken Han languages by perpetuating a vast congeries of topolects, most of which have never been written down. The tetragraphs permit individuals from different topolectical backgrounds to pronounce them in wildly varying fashions. For example, MSM chen ("array") is read as tsan in Hangzhou, dzang in Shanghai, dzing in Ningpo, teng in Fuzhou, tin in Amoy and Swatou, and zhen in Canton. Through a judicious and well-planned introduction of Pinyin, standardization of MSM could be achieved within a reasonable period of time, whereas the present policy of benign neglect ensures that the mutually unintelligible Han topolects will probably persist indefinitely. It is noteworthy that few of the leaders of the Pinyin movement during the past 30 years, men such as Zhou Youguang, Ni Haishu, and Yin Binyong, were native speakers of MSM and yet they all could write beautifully correct romanized Mandarin. A similar situation obtains with SD where speakers of the non-standard Tokmak dialect are able to read and W-rite the standard language. So long as related speech forms are mutually intelligible (i.e., are truly dialects and not separate languages), it is possible to select one of them as standard even before complete unanimity of pronunciation and usage is attained among all the members of a linguistic community (indeed, absolute unanimity is impossible because each speaker inevitably has his or her own idiolect). This is also illustrated by the relationships that obtain among West Texas, Eastern Maryland, Boston Brahmin, and Midwestern varieties of American speech. They each have a unique pronunciation and special expressions but they remain, nonetheless, mutually intelligible and all employ the same standard for written English. From this and other evidence, it would appear that appropriately phased romanization of MSM would actually stimulate unification of the Han topolects rather than prevent it.

There are many other benefits of romanization. Aside from all those we have cited above, perhaps the most important in the present age of advanced electronic information processing is the ability to arrange large quantities of data in a single alphabetical listing, manipulate them in various useful ways, and to retrieve them readily and inexpensively. Another important advantage is the capacity for alphabetical scripts to employ modern terminology directly from other languages without having to stumble about while waiting for a consensus on an appropriate translation. SD, as a functioning alphabetical Han language, has shown its adaptability to the needs of its users by developing a complex, up-to-date vocabulary for dealing with modern agriculture (machines, plants, irrigation, fertilizers, and so forth).

In some respects, it was easier for the Dungans to alphabetize than it will be for the Chinese. For one thing, they were not weighed down by three thousand years of tetragraphic civilization as is the mainstream of literate Chinese now made up of those who can read and write MSM with proficiency. Also, as pointed out above, the Dungans were actively supported by the Soviet government in their efforts to create an alphabet. On the other hand, the Chinese are in some respects in a better position to phoneticize MSM than were the Dungans when they started out to do so in the early part of this century. MSM has already possessed a neat, coherent spelling system for over 30 years. It can be typed on a standard keyboard and, with the substitution of yu for ü, requires no special symbols or diacriticals. Best of all, the advocates of the alphabetization of MSM are able to learn from the long and fruitful experience of SD.

Examination of the history of SD brings to light the necessity in alphabetization for making script emphatically subordinate to spoken language. For successful alphabetization, it is essential to abandon the attempt to sustain a one-to-one correspondence with the tetragraphs since so much that is written in Chinese is composed in a banwenbanbai style removed in varying degrees from any actual speech. Surely, Han languages preceded the tetragraphs and, while the latter indubitably had an impact on the former during the more than three millennium in which they interacted, it is obvious that speech has always been primary and script secondary. We need only recall that right up to the present the vast majority of Chinese have been wholly or functionally illiterate, whereas all Chinese except a tiny and statistically insignificant percentage of mutes can speak one or another Han language.

The cardinal rule Chinese script reformers must always keep in mind is this: do not panic! If something of substance can be said without ambiguity in the spoken language, then it most assuredly can be written with suitable phonetic symbols. Unless we assume that the content of spoken Han languages is decidedly less colorful and interesting than that of written Chinese, then, as the Dungans have shown us, we need not fear that a written language based on phonetically transcribed speech will be necessarily inferior to tetragraphic writing and may even be superior in some aspects. Unless we assume that the lectures of Chinese professors are babyish and the tales of Chinese storytellers are bland, then there is nothing to prevent the emulation of SD by MSM. Just as Cyrillic SD is already a reality, so can MSM gain an auxiliary Roman expression by following its path.

Acknowledgement

The author wishes to express his special gratitude to I. Yusupov of Alma Ata (Khazakh SSR) and M. Sushafflo, Director of the Dungan Research Institute of the Academy of Sciences in Frunze (Kirghiz SSR) for providing him with many of the SD materials that were used in the preparation of this paper. He would also like to thank John DeFrancis for much helpful guidance.

This paper was originally prepared for the conference on "The Legacy of Islam in China: An International Symposium in Memory of Joseph F. Fletcher" held at Harvard University, April 14-16, 1989. Revised June 22, 1989 and subsequently.

Select Bibliography

Webmaster's note: For the bibliography, see the printed edition, which can be purchased for US$7, plus shipping, from the editor of Sino-Platonic Papers. The relevant issue (No. 18) of Sino-Platonic Papers includes another important essay: "Who Were the Gyámi?"