Janhunen J. The Mongolic Languages

Подождите немного. Документ загружается.

Todaeva, B. X. [H.] (1959) ‘Über die Sprache der Tung-hsiang’, Acta Orientalia Hungarica 9:

273–310.

Todaeva, B. X. [Ch.] (1960) ‘Mongolische Dialekte in China’, Acta Orientalia Hungarica 10:

141–69.

Todaeva, B. X. [Ch.] (1963) ‘Einige Besonderheiten der Paoan-Sprache’, Acta Orientalia

Hungarica 16: 175–97.

Vladimircov, B. Ya. (1929) Sravnitel’naya grammatika mongol’skogo pis’mennogo yazyka i xalx-

askogo narechiya: Vvedenie i fonetika, Leningrad: Vostochnyi institut.

Weiers, Michael (1986) ‘Zur Herausbildung und Entwicklung mongolischer Sprachen’, in Michael

Weiers (ed.) Die Mongolen, Darmstadt: Wissenschaftliche Buchgesellschaft, pp. 29–69.

Yaxontova, N. S. (1997) ‘Mongol’skie yazyki’, in Mongol’skie yazyki – Tunguso-man’chzhurskie

yazyki – Yaponskii yazyk – Koreiskii yazyk [Yazyki Mira], Moskva: Rossiiskaya Akademiya

Nauk & Izdatel’stvo Indrik, pp. 10–18.

Yu Shichang (1983) Lun Menggu Yuzu de Xingcheng he Fazhan, Beijing: Minzu Chubanshe.

390 THE MONGOLIC LANGUAGES

391

CHAPTER NINETEEN

PARA-MONGOLIC

Juha Janhunen

The conventional understanding of the Mongolic language family is based on the infor-

mation available from the living and historically documented Mongolic languages, all of

which may be considered descendants of Proto-Mongolic. The only pieces of direct evi-

dence of any other kind of Mongolic are supplied by the few traits in Written Mongol

that seem to reflect traces of Late Pre-Proto-Mongolic dialectal variation. On the basis of

this evidence, it may be concluded that in Pre-Chinggisid times there existed dialectal

forms of Mongolic which in some relatively minor respects deviated from the dialect that

came to form the foundation of Proto-Mongolic. It may further be assumed that such

dialectal differences were conditioned by social, cultural, and geographical divisions

among the direct ancestors of the historical Mongols.

From historical sources, such as the ‘Secret History of the Mongols’, we even know

the names of many social units which once coexisted with the Mongols of Chinggis

Khan. It may be taken for certain that the idioms spoken by tribes and confederations

such as the Kereit, Ongniut, and Naiman were not perfectly identical with the dialect of

the original Mongol unit. On the other hand, they certainly represented variant forms of

a single language, as is also suggested by the non-Proto-Mongolic traits of Written

Mongol, which with some likelihood may be connected with the specific influence of the

ancient Naiman dialect. The problem is that the historical sources also mention many

other populations in the vicinity of the Mongols, and we do not know how closely the

idiom of each given group was associated with the Proto-Mongolic lineage.

Obviously, many of the ethnic groups surrounding the historical Mongols were not

linguistically related to the Mongols at all. Rather, they may have spoken, for instance,

Turkic, Tungusic, Uralic, or Yeniseic idioms. Between the alternatives of immediate

dialectal relationship and total genetic difference there is, however, the tantalizing possi-

bility that some of the ethnic neighbours of the early Mongols may actually have spoken

languages that were collaterally related to Proto-Mongolic. Such languages may be

called Para-Mongolic. Like Proto-Mongolic itself, the Para-Mongolic languages would

have represented descendant branchings of Pre-Proto-Mongolic. We have no way to

know how diversified Para-Mongolic could have been internally, but externally

there could well have been at least a considerable difference, if not a total absence of

intelligibility, against the Proto-Mongolic lineage.

Speaking of Para-Mongolic would remain at the level of mere speculation were it not

that there are indications that such languages actually existed both before and after the

appearance of the historical Mongols. Evidence on Para-Mongolic is threefold: histori-

cal, philological, and linguistic. The historical evidence comes from the fact that the

Mongols originally seem to have represented a single relatively marginal group among a

diversity of interrelated populations that once existed in the zone extending from

Southern Manchuria to Northern Mongolia. The early populations in this zone are only

known by generic names, notably Donghu (from c.300

BC) and Xianbei (to c. AD 300).

392 THE MONGOLIC LANGUAGES

There is reason to assume that both the Donghu and the Xianbei were large

conglomerations of local populations which were not linguistically homogeneous, but

which certainly comprised the contemporary speakers of Pre-Proto-Mongolic. Moreover,

since the conditions for population growth and ethnic expansion under these entities

were favourable, and had been so already before the Donghu, it is very likely that Pre-

Proto-Mongolic broke into parallel branches, only one of which led to Proto-Mongolic,

while the others belonged to the context of Para-Mongolic. The ultimate fate of all the

Para-Mongolic branches was to become extinct under the demographic impact of the

subsequent political developments.

Before their extinction the Para-Mongolic populations were involved in the creation

of two well-documented political states: the Northern Wei (Chinese Bei Wei) of the

Tabghach (Tuoba, 386–550) and the Liao of the Khitan (Qidan, 907–1125). It is from the

time of these states that philological information is available on Para-Mongolic. This

information is of two kinds: on the one hand, there are occasional Para-Mongolic words

transcribed and glossed in Chinese sources, and, on the other hand, there is a corpus of

preserved texts written in the Para-Mongolic Khitan language. Unfortunately, both kinds

of information involve considerable problems of interpretation. The most important

immediate conclusion from the philological sources is that they do, indeed, represent

various linguistic stages that may be classified as Para-Mongolic.

Finally, the former existence of Para-Mongolic in the border zone between Mongolia

and Manchuria is confirmed, although only indirectly, by the linguistic fact that there are

actual Para-Mongolic loanwords in the modern languages of the region. It is very likely

that Proto-Mongolic also received such loanwords, especially from the language of the

Khitan, but these remain difficult to identify. The main information therefore comes from

the Tungusic languages, which, in addition to a large corpus of regular Mongolic loan-

words of various periods, contain a number of Mongolic-like items with otherwise atyp-

ical or seemingly anachronistic phonological or derivational characteristics. These

loanwords can only derive from Para-Mongolic, and one of the contexts in which they

were borrowed must have been the Liao dynasty of the Khitan.

DATA AND SOURCES

Historical information on Para-Mongolic is mainly provided by the Chinese dynastic his-

tories, such as the Wei Shu, Sui Shu, Jiu Tang Shu, and Liao Shi. These sources also con-

tain Para-Mongolic lexical items preserved in Chinese transcription and translation. The

philological problems connected with these items are numerous and often insurmount-

able. For one thing, the dynastic histories were compiled long after the times they refer

to. Also, as the authors were no linguists, their attempts to record non-Chinese lexical

items with the help of Chinese characters inevitably remained imperfect and inconsistent

both as far as the transcriptions and the translations are concerned. In many cases, the

exact source language of the data is unclear. Moreover, apart from the initial mistakes

made in the process of recording, additional mistakes have often accumulated during the

transmission of the records. One trivial source of such mistakes is contained in the

frequent confusion between certain Chinese characters.

Under such circumstances it is surprising that anything at all can be said of the Para-

Mongolic items preserved in Chinese rendering. Inevitably, there are many sources of

controversy. Peter A. Boodberg (1936) and Louis Bazin (1950), for instance, analysing

the Tabghach words of Chinese sources, came to the conclusion that Tabghach was

a Turkic language, a mistake at least partly caused by the fact that some of the items

concerned are Turkic names and titles, widespread as cultural words in Central and East

Asia. However, the few genetically diagnostic items in the Tabghach corpus have obvious

Mongolic cognates, as was noted by Louis Ligeti (1970) and confirmed again by Gerhard

Doerfer (1992, 1993). Even so, a large portion of the Para-Mongolic data transmitted by

Chinese historical sources remains beyond the reach of critical research.

Differences of opinion have also concerned the linguistic identity of the materials

preserved in Khitan writing. The existence of such materials was first established by Toru

Haneda (1925), but the dispute as to whether Khitan was a Mongolic, Turkic, or

Tungusic language continued several decades. Some proponents of the Mongolic identi-

fication have assumed that Khitan was close to Middle Mongol, or the language under-

lying Written Mongol. Today, however, the conception is gaining support that Khitan

was a language in some respects radically different from the historically known

Mongolic languages. If this view proves to be correct, Khitan is, indeed, best classified

as a Para-Mongolic language.

The historical context of the Liao empire, including its internal and external ethnic

relationships, is investigated in detail in Karl A. Wittfogel and Fêng Jia-Sheng (1949),

a monumental work that remains an unsurpassed secondary source on the Khitan. More

specific treatments of the linguistic situation in the Liao empire, and of the Khitan

language materials preserved in Chinese rendering, are provided by Herbert Franke

(1969, 1982). A general evaluation of the sources on Khitan is also offered by

Ye. A. Kuz’menkov (1997), complemented by a similar treatment of the language of the

Xianbei by A. Luwsandendew (1997).

Work on the Khitan language materials preserved in native writing was long ham-

pered by the confusion resulting from the historically known fact that the Khitan actual-

ly used two systems of writing, termed the Khitan Large Script (Chinese dawen) and the

Khitan Small Script (xiaowen). It was only Goro Toyoda (1964) who convincingly iden-

tified the two scripts and their principles. We now know that the Khitan Large Script was

a basically logographic (perhaps partly logosyllabic) system of writing with a large num-

ber of separate characters (logograms), while the Khitan Small Script was a basically syl-

labic (perhaps partly phonosyllabic) system of writing with a much smaller number of

symbols (syllabic signs). From this fundamental conclusion, it has been possible to pro-

ceed further towards a more comprehensive understanding of the Khitan scripts and their

historical context, as preliminarily summarized by Juha Janhunen (1994), Chingeltei

(1997a, 1998), and Liu Fengzhu (1999).

The corpus of the Khitan Small Script was first systematically collected, analysed,

and published (in coded form) by a Russian team comprising V. S. Starikov (1982),

M. V. Arapov (1982), as well as M. A. Probst et al. (1986). Apart from the syllabic nature

of the script, their results give information on the typology (morphology, morphosyntax,

and syntax) of the language underlying the script. They do not, however, give any actu-

al ‘readings’ (phonology). In a separate project, an updated version of the same corpus

was published (both in the original Khitan script and in coded form) by a Chinese team

comprising Chingeltei, Liu Fengzhu, and Chen Naixiong (1985), and others. Most of the

subsequent work on Khitan has been based on the results of the Chinese team, whose

members continue to be active in the field. The corpus is still growing, with new

epigraphic documents found on an almost yearly basis.

The first attempts to decipher the Khitan Small Script were made already soon after

the identification of the underlying language as Khitan. Not much of value remains of

these attempts, but a new tradition of ‘reading’ Khitan has gradually emerged from the

PARA-MONGOLIC 393

394 THE MONGOLIC LANGUAGES

corpus studies of the Russian and Chinese teams. It is today possible to identify and

‘read’ most of the Chinese elements (names and terms) in the Khitan texts, and on the

basis of this information, attempts have been made to find reasonable sound/meaning

correspondences for native Khitan words. Even these ‘readings’, as exemplified by

György Kara (1975, 1986–7), Chingeltei (1997bc), Goro Toyoda (1998), and Alexander

Vovin (2000), can still only be considered preliminary at best, as is also evident from the

fact that they are often mutually contradictory. Clearly, future work will require a more

careful consideration of not only the philological but also, and perhaps in particular, the

linguistic aspects of the problem.

Linguistic work relevant to Para-Mongolic has so far been focused on two issues: the

dialectal diversity of Proto-Mongolic (and the language underlying Written Mongol), as

discussed by L. L. Viktorova (1961) and Ye. A. Kuz’menkov (1988), and the Mongolic

loanwords in Tungusic, as comprehensively analysed by Ligeti (1960), Doerfer (1985),

and William Rozycki (1994a). Two specific categories of lexical borrowing are discussed

in Okada (1962) and Janhunen (1993), while an important phonological detail is taken

up by Rocyzki (1994b).

THE KHITAN SCRIPTS

It is very likely that the texts preserved in Khitan writing, once fully deciphered, will

provide the single most extensive and reliable database on Para-Mongolic. This is, indeed,

not a collection of isolated words in obscure transcription, like the Para-Mongolic items

preserved in Chinese rendering, but a large corpus of texts originally meant to be read by

literate native speakers. It is also obvious that the key to the Khitan language is contained

in the Khitan Small Script. The Khitan Small Script remains the most crucial unsolved

problem of comparative Mongolic studies. It is also one of the last remaining undeci-

phered major scripts in the world for which the prospects of full deciphering are good.

Historically, this script is attributed to Prince Yelü Tiela, who is said to have created

it in the year 925 upon the request of his elder brother, the Liao emperor Taizu (reigned

907–26), after having familiarized himself with the Uighur language and script. This

information is probably correct in so far as the Uighur script may well have served as a

stimulus to simplify the Khitan Large Script, which, allegedly was designed by Liao

Taizu himself only five years earlier (920). However, the influence of the Uighur script

(later adopted for Written Mongol) on the Khitan Small Script can only have been very

indirect, as the two systems are fundamentally different as far as their functional proper-

ties and external graphic orientations are concerned. In fact, there were certainly also

other models used in the creation of the Khitan Small Script.

As it is, both systems of Khitan writing can be safely classified as belonging to the

Sinitic type of scripts, with the Chinese script as their most immediate model. The

syllabic signs of the Khitan Small Script were apparently designed by selecting and

simplifying a sufficient number of the logographic symbols of the Khitan Large Script,

though the exact derivations of most of the syllabic signs remain still obscure. The basic

innovation involved by the Khitan Small Script is, however, not contained in its graphic

forms, but in the fact that the paradigm of syllabic signs is essentially a closed system.

The number of the currently known syllabic signs is just over 500, but this preliminary

paradigm is likely to contain a considerable amount of allography due to local and

temporal variation in the shape of the symbols. Another innovation involved by the

Khitan Small Script is that the syllabic signs are (in the normal writing style) grouped

into blocks, corresponding to linguistic words.

Although several Khitan words have been tentatively ‘read’, the most crucial

evidence of the Para-Mongolic identity of the underlying language comes from a pun

between the words for ‘five’ and ‘hare’. In the Khitan Small Script, the word for ‘five’ is

written with an isolated syllabic sign, while the word for ‘hare’ is written as a block of

three signs, the first of which is identical with the sign for ‘five’. Of all language fami-

lies in the region, this pun makes sense only in Mongolic, where ‘five’ is *tabu/n and

‘hare’ is *taulai. Obviously, the same syllabic sign was used to write the Khitan equiva-

lents of both the *tabu- of ‘five’ and the *tau- of ‘hare’. However, these elements are not

identical in Proto-Mongolic. They may also not have been identical in Khitan, but this

would mean that the writing system was phonologically rather inexact. A more likely

possibility is that they were identical in Khitan (perhaps to be read as tau), which, on the

other hand, would prove that Khitan was characterized by phonological innovations dif-

ferent from Proto-Mongolic.

The word for ‘hare’ illustrates well the problems of decipherment and reconstruction.

The most commonly accepted ‘reading’ for the whole block of syllabic signs is tau.li.a.

This is, however, poorly compatible with Proto-Mongolic *taula.i. Although there is no

question that the two words are cognates, we do not know for the time being whether

they are complete cognates or only partial ones (perhaps up to *taul-). It is also unknown

what the exact phonological shape underlying the proposed ‘reading’ of the Khitan word

might have been. For these reasons, it cannot be taken for certain that the ‘reading’ is

correct in the details (especially as far as the second and third syllabic sign are

concerned). One might conclude that, although the current ‘readings’ are clearly going in

the right direction, the final breakthrough remains to be made.

The currently known corpus in the Khitan Small Script comprises twenty-one large

epigraphic (and mainly epitaphic) texts, many of them well, or even perfectly, preserved.

Additionally, there are several shorter texts cast on bronze objects, coins and tallies,

engraved on pottery, or painted on walls. Most of this corpus, dated between 1057 and

1170, derives from tombs and occasional finds in the region of the Upper and Middle

Capitals (Shangjing and Zhongjing) of the Liao empire, in modern central Inner

Mongolia (Chifeng). The Khitan Small Script survived well beyond the collapse of the

Liao empire, until its use was officially discontinued in 1191 in the Jin Empire of the

Jurchen (1115–1234), which replaced the Liao as the dominant power in Northern China

and Manchuria. Unfortunately, there is only one actual bilingual text, the so-called

Langjun Xingji inscription, which also dates from the Jin period (1134).

Compared with the Small Script, the Khitan Large Script was long considerably less

well known. However, recent excavations have revealed a growing corpus of texts also

in this variety of Khitan writing. It is now clear that the Small Script, though created

later, never fully replaced the Large Script, for during the whole duration of the Liao

empire, the two scripts were used in parallel. This gives rise to many contextual ques-

tions. Most importantly, it is not immediately clear why the Khitan continued to use the

Large Script, which must have been much more cumbersome in practice than the lin-

guistically more advanced Small Script. Like the Chinese script, the Large Script was

essentially an open system, in which each word was written with a special character. The

number of the currently known ‘large’ characters is well over 1,000. Another question is

why the Khitan, if they wished to have a logographic writing, did not simply adopt the

Chinese script, as many other peoples in peripheral China had done.

One possible answer to these questions is that the Khitan Large Script was actually

not an invention of the Khitan, though it may have undergone some normalization in the

context of the Liao empire. Rather, it is likely to have represented an old local variety of

PARA-MONGOLIC 395

396 THE MONGOLIC LANGUAGES

the Chinese script that may also have been in use among earlier Para-Mongolic popula-

tions, including the Tabghach. The assumption that the Khitan Large Script was part of

a local tradition is also supported by the fact that it was adopted, with minor modifica-

tions, as the official script for the Jurchen language in the Jin empire (1119). Without the

pressure of a historical continuity, the Jurchen would hardly have chosen the Large

Script, for they must have been aware of the technical superiority of the Small Script,

which also continued to be used by the Khitan living under Jurchen rule.

It is therefore obvious that the Khitan Large Script and the Jurchen Script ultimately

reflect the same local (Southern Manchurian) tradition of writing. The written symbols

in both of these scripts are clearly based on the Chinese characters, with some symbols

being even completely identical with their Chinese counterparts, while others represent

modifications of the regular Chinese characters. Thus, both the Khitan Large Script and

the Jurchen Script should be seen as the results of gradual evolution, rather than unique

invention. The Khitan Small Script, by contrast, seems to have involved an element of

conscious innovation.

SEGMENTAL PHONEMES

With the language of the Khitan Small Script still phonologically uninvestigated, and

with the shape of Para-Mongolic words in Chinese rendering being extremely distorted,

the most reliable source on Para-Mongolic segmental phonology is provided by the cor-

pus of loanwords in Tungusic, especially Manchu. Although possible Para-Mongolic

items are present in all Tungusic languages, they seem to be more abundant in Manchu,

which points to the possibility that many of these words were borrowed during the his-

torical contacts between the Khitan and the Jurchen (the direct ancestors of the Manchu).

Not all Para-Mongolic loanwords in Tungusic need, however, derive from Khitan, for

they may also reflect the diversity that possibly existed within Para-Mongolic.

Using the combined information of linguistic and philological sources it is possible to

establish some phonological properties that are likely to have been characteristic of at

least some forms of Para-Mongolic, perhaps specifically Khitan (though this remains to

be verified). Some of these properties may be seen as archaisms, which reflect a linguistic

stage identical with Pre-Proto-Mongolic. Such features are not unambiguously Para-

Mongolic. Other features are, however, clearly innovatory, and they serve to distinguish

Para-Mongolic from the lineage represented by Proto-Mongolic. Most interestingly,

some of the features of Para-Mongolic seem to involve innovations that also affected the

Proto-Mongolic lineage, but only much later, in Post-Proto-Mongolic times. This might

mean that the Para-Mongolic territory was the primary centre of linguistic innovation,

from where the innovative features diffused slowly towards the periphery, where the

Proto-Mongolic lineage was spoken.

Among the apparent phonological archaisms of Para-Mongolic, two features

connected with the consonant paradigm may be mentioned:

(1) Preservation of the strong labial stop *p as a separate phoneme, as in Khitan †pon

‘time’, probably cognate with Proto-Mongolic *xon ‘year’. The fact that *p can still be

restored for Proto-Mongolic on the basis of internal reconstruction suggests that the

development *p > *x took place in Late Pre-Proto-Mongolic not much prior to the emer-

gence of the historical Mongols. Quite possibly, Para-Mongolic, before its extinction,

was also embraced by this innovation. The analogous change *p > f in Manchu seems to

have been initiated in medieval Jurchen.

PARA-MONGOLIC 397

(2) Preservation of a palatal nasal *ny as a separate phoneme, as in *nyoka ‘dog’,

borrowed into Manchu as nyoxe (Written Manchu niohe) ‘wolf’, as opposed to Proto-

Mongolic *noka.i ‘dog’. Correspondences of this type suggest that the Proto-Mongolic

lineage was characterized by the depalatalization development *ny > *n. However, since

the date of this development remains unknown, it is difficult to determine whether the

relevant loanwords in Tungusic derive from Para-Mongolic or simply from some early

stage of the Proto-Mongolic lineage.

When these two archaisms are projected on what is otherwise known of the Mongolic

consonantism, it seems possible to postulate a paradigm of seventeen consonant

phonemes for Pre-Proto-Mongolic (Table 19.1).

At the syntagmatic level, there are two other consonantal archaisms that may have

been characteristic of some forms of Para-Mongolic. Both features are suggested by

loanwords documented in Tungusic:

(3) Preservation of a primary velar spirant *x as an initial segment distinct from *p

and *k, as in *xorin (*xorïn) ‘twenty’, borrowed into Proto-Tungusic as *xorïn. In this

case, the Proto-Mongolic shape *kori/n (*korï/n) would seem to be irregular

(probably influenced by *koyar ‘two’). Although the reconstruction of *x is still con-

nected with problems, the postulation of this segment for Pre-Proto-Mongolic is sup-

ported by the areal fact that Proto-Tungusic also had an analogous (native)

segment. However, it remains unclear whether the items suggesting *x were

borrowed into Tungusic from Para-Mongolic, or already from the common ancestor of

Para-Mongolic and the Proto-Mongolic lineage.

(4) Preservation of a distinctive dental *t (as opposed to palatal *c) before the high

unrounded vowels *i *ï, as in *gutin (*gutïn) ‘thirty’, borrowed into Proto-Tungusic as

*gutïn, as opposed to Proto-Mongolic *guci/n (*gucï/n). Again, it is impossible to deter-

mine the chronological limits of this feature, leaving open the possibility that it reflects

the original Pre-Proto-Mongolic, rather than the Para-Mongolic situation. Since the

development *ti > ci is also attested in Manchu, it may have affected some forms of the

neighbouring Para-Mongolic, as well.

The innovative features of the Para-Mongolic consonantism are likely to have included,

at least, the following:

(5) Spirantization of the palatal stop *c to sh. This feature is well attested in the Para-

Mongolic loanwords of Manchu, e.g. shanggyan ‘white’ (Written Manchu shanggiyan),

as opposed to Proto-Mongolic *cagaxan. Since, however, the same development is also

observed in native Manchu items, it is difficult to determine whether it affected the loan-

words concerned already in the Para-Mongolic donor language, or only later in Manchu.

TABLE 19.1 PRE-PROTO-MONGOLIC CONSONANTS

*p *t *c *k

*b *d *j *g

*s *x

*m *n *ny *ng

398 THE MONGOLIC LANGUAGES

In any case, the development *c > sh was probably connected with the similar spiranti-

zation of the velar stop *k to x (h) in both Para-Mongolic and Jurchen. This innovation,

which was initially valid only for the position before original velar vowels, spread sub-

sequently to several modern descendants of the Proto-Mongolic lineage.

(6) Loss of medial *x, as in †eulen ‘cloud’, as opposed to Proto-Mongolic *exüle/n.

Although this innovation is well known from all Modern Mongolic languages, it is

absent in the language underlying Written Mongol and is therefore not likely to have

affected the Proto-Mongolic lineage until in Late Pre-Proto-Mongolic times. In Para-

Mongolic, by contrast, it seems to have begun at least several centuries earlier. A related

development may be the apparent loss of medial *b, as in †tau ‘five’ (Proto-Mongolic

*tabu/n), though other interpretations are also possible.

Partly due to Chinese influence, but possibly also as the result of indigenous devel-

opments, some forms of Para-Mongolic, including Khitan, are likely to have had a num-

ber of secondary (initially perhaps marginal) consonant phonemes, notably the labials †w

and †f. If this was so, the typical Para-Mongolic consonant system may have comprised

at least 19 distinct segments (Table 19.2).

It is considerably more difficult to say anything definite concerning the Para-

Mongolic vowel system. There are indications that Para-Mongolic, in general, and

Khitan, in particular, may have been characterized by several innovations that affected

the Proto-Mongolic lineage only shortly before or even after its breakup. Such innova-

tions include labial harmony, the paradigmatic merger of the high unrounded vowels *i

*ï, the phenomenon of rotation, and the loss of *ö (and possibly also *ü) as a distinctive

segment. Many of the ‘readings’ proposed for Khitan also seem to imply that there was

vowel reduction or even vowel loss in non-initial syllables. However, information on the

presence or absence of these phenomena in Para-Mongolic is still insufficient. It is there-

fore virtually hopeless to attempt any schematization of the Para-Mongolic vowel

system. Moreover, there were probably many different vowel systems depending on

geographical and chronological circumstances.

Two concrete and specifically Para-Mongolic innovative phenomena connected with

vowels can nevertheless be established on the basis of loanwords borrowed into Tungusic:

(7) Contraction of diphthongoid sequences ending in *( y)i or *( y)ï, as in *po( y)ïma/n

‘sock/s’ > *pomo/n, represented in Manchu as fomon : fomo.ci, as opposed to Proto-

Mongolic *xo(y)ima.su/n. Contractions of this type would also suggest the more gener-

al conclusion that there were no monophthongoid long or double vowels in

Para-Mongolic. If such vowels had arisen through contraction they are likely to have

been shortened (as also happened in Jurchen-Manchu).

TABLE 19.2 PARA-MONGOLIC CONSONANTS

†p †t †k

†b †d †j †g

†f †s

†sh †x

†m †n †ny †ng

†l

†r

†w †y

PARA-MONGOLIC 399

(8) Raising of *o to *u in the words *mori/n (*morï/n) ‘horse’ and *koni/n (*konï/n)

‘sheep’, borrowed into Tungusic as *murï/n and *kunï/n, respectively. The shape *murï/n

(†murin) is also attested in medieval Jurchen, which confirms that it is, indeed, an old

borrowing dating back to Para-Mongolic times. However, it is unclear whether or not the

vowel raising was a simple combinatory development, for it is contradicted by the item

*xorin (*xorïn) ‘twenty’, represented in Tungusic as *xorïn. Possibly, the different rep-

resentations derive from two varieties of Para-Mongolic, separated either areally or

chronologically.

NUMERALS

Numerals are only one of several closed systems of Para-Mongolic lexical items for

which systematic information is today available both from the Khitan Small Script and

the Para-Mongolic loanword corpus in Tungusic. Some allegedly Khitan numerals have

also been transmitted in Chinese transcription. Other closed systems, for which informa-

tion can mainly be gathered from the Khitan Small Script, include the names for the four

seasons (Toyoda) and the twelve animals of the East Asian Zodiac (Chingeltei). The

names of the basic colours (Okada), on the other hand, can be approached mainly with

the help of Tungusic.

The most reliable source on the Para-Mongolic numerals is provided by the Jurchen-

Manchu set for the teens (11–19). These were systematically borrowed from an idiom

clearly related to Mongolic, but different from the lineage of Proto-Mongolic. The shapes

of these items, as reconstructable from Jurchen-Manchu, seem to have been: 11

†omshon, 12 †jir.hon (> jorhon ~ jorgon), 13 †gor.hon, 14 †dur.hon ~ †durhun,

15 †tobu.hon (> tofohon), 16 †nil.hun (> nyolhun), 17 †dal.hon, 18 †nyo.hun ~ †nyohon,

19 †onyo.hon. Assuming that these items had a deeper history, they may be reconstructed

as Pre-Proto-Mongolic 11 *omcon, 12 *jï.r.ku/n, 13 *gu.r.ku/n, 14 *dö.r.kü/n,

15 *tabu.ku/n, 16 *nil.kü/n, 17 *dal.ku/n (or *dal.u.ku/n), 18 *nya(y)i.ku/n, 19 *o+

nya(y)i.ku/n.

This set for the teens, unattested in any living or historically documented idiom of the

Proto-Mongolic lineage, allows several interesting conclusions to be made. Most impor-

tantly, the items for 12 to 19 are clearly derived from the corresponding basic numeral

roots by adding the suffixal element †.hU/n < *.kU/n (or perhaps *.xU/n) ‘-teen’.

Similarly, the Proto-Mongolic numerals for 3–4 and 6–8 of the first decade contain the

suffixal element *.pA/n ‘of the first decade’. It is not far-fetched to see in these suffixal

elements the roots of the numerals 20 *xo.r.ï/n (Proto-Mongolic *korïn) and 10

*pa.r.pa/n (> Proto-Mongolic *xarba/n), respectively. As a curious detail, the item 16

†nil.hun suggests the presence of an original Pre-Proto-Mongolic root 6 *nil, which was

replaced in Proto-Mongolic by the secondary innovation 6 *jir+gu.xa/n (= 2 x 3).

Some other aspects of the set for the teens seem to have involved innovations on the

Para-Mongolic side. Thus, 11 †omshon is possibly connected with the nominal root

*onca ‘special, additional’, as attested in the Proto-Mongolic lineage, while 19

†onyo.hon is clearly based on 18 †nyo.hun ~ †nyohon. Since Proto-Mongolic 1 *nike/n

(> *nige/n) and 9 *yer.sü/n are also likely to be innovations, the original roots for these

numerals (if there were any original roots) remain obscure. The roots for 2–5 and 7–8

are, however, common to Para-Mongolic and Proto-Mongolic, allowing them to be

reconstructed tentatively as Pre-Proto-Mongolic 2 *jir (or *jïr), 3 *gur, 4 *dör, 5

*tab(.u), 7 *dal(.u), 8 *nyay(.ï ). Importantly, the final consonant *r of the roots for 2–4,

which also occurs in 10 *par, is observed both in Para-Mongolic and Proto-Mongolic,