CC BY
124
Juha Janhunen
University of Helsinki, Helsinki
OBSERVATIONS ON THE PARA-MONGOLIC ELEMENTS IN JURCHENIC
1. Introduction
It is well known that the Tungusic languages incorporate a large number of Mongolic loanwords. Most of these were reviewed in detail by Gerhard Doerfer [1985], who showed that their distribution indicates a gradual infiltration of Mongolic lexicon into Tungusic, with the smallest number of Mongolisms being present in geographically marginal idioms such as Western Ewenki and Ewen. In the context of the "Altaic" group of languages, the Mongolic loanwords in Tungusic are secondary to the Turkic loanwords in Mongolic, showing that the general direction of borrowing has been from west to east and from south to north. The majority of the Mongolic loanwords in Tungusic are Post-Proto-Mongolic borrowings into Post-Proto-Tungusic, that is, into the various individual Tungusic languages and dialects — naturally only those that have been in direct contact with Mongolic languages. There is also a small corpus of items shared by Proto-Mongolic and Proto-Tungusic, for which the direction of borrowing is more difficult to establish. Ultimately we cannot rule out that there are also a few items shared on a genetic basis, though evidence for a binary genetic connection between Tungusic and Mongolic in the context of a distinct language family ("Khinganic") is still very scarce [Janhunen 1996b].
Many Mongolic loanwords are present in only a single Tungusic language or dialect. In most of such cases, it is not difficult to point out the Mongolic source. It is, for instance, natural that Barguzin Ewenki has borrowed many words from Buryat, but also from Khamnigan Mongol and/or Dagur [Khabtagaeva 2010]. The Mongolic elements in Solon are, of course, mainly from Dagur, while those in Khamnigan Ewenki come from Khamnigan Mongol [Janhunen 2013: 36-38]. We also have to reckon with the possibility that some Mongolic items in Tungusic, especially in Ewenki and Ewen, have been secondarily transmitted by Yakut, a Turkic language that has a particularly large number of Mongolic
loanwords [Kaluzynski 1962]. The only really problematic cases are offered by the Jurchen-Manchu lineage, which has a considerable number of Mongolic elements not present in the other Tungusic languages. Although many of the Mongolisms in later Manchu can be directly derived from Post-Proto-Mongolic sources, there are also items that are conspicuously "different" from their historically documented Mongolic counterparts. Interestingly, many of these items were not included in Doerfer's corpus, since he focused on elements that are present in two or more separate Tungusic lineages. They are, however, discussed by Rozycki [1994], but his conclusions are often regrettably laconic.
Manchu is notoriously the most "aberrant" Tungusic language, in some respects looking even "non-Tungusic" [Vovin 2006]. Diachronically, Manchu, recorded since the early 17th century in the Manchu script, may be viewed as the more or less direct descendant of Jurchen, recorded in the Jurchen script, as well as in Chinese transcriptions, from the early 12th century till the late 16th century. Jurchen and Manchu are, consequently, two chronologically successive representatives of a single distinct lineage of Tungusic, to which also modern Sibe (Xibo), a regional variety of Manchu, belongs as a "third-generation" member. This lineage may be called "Jurchenic", and Jurchen, Manchu and Sibe represent three successively more innovative stages of this lineage (cf. also [Doerfer 1978]). In the context of the Tungusic family, Jurchenic, as a whole, may be viewed as a particularly innovative branch, at least as far as phonology is concerned. In the morphology, most of the specific features of Jurchenic seem to involve secondary losses of complexity, which often can be explained by assuming interference from Chinese and Mongolian. It has recently been maintained that the morphological "simplicity" of Manchu may in some cases be an archaic retention [Alonso de la Fuente 2011], but the evidence is not compelling.
However this may be, there has historically been one additional neighbouring language whose influence on Jurchen-Manchu has not been studied in detail — for the reason that the language is extinct. This language is, of course, Khitan, but when speaking of Khitan we also have to consider the possibility that it was only one of several "Khitanic" languages once spoken in the western neighbourhood of Jurchen. In fact, very little work has been done on this topic (cf. nevertheless [Kuz'menkov 1986, 1988]). Khitan was the dynastic language of the Liao empire in Manchuria (907-1125), and later of the Western Liao
or Kara Khitai Khanate in eastern Central Asia (1124-1218). Originally, the Khitan were only one of several tribes in southwestern Manchuria (the region of Xiliao), and we do not know how uniform the tribal languages were. In any case, Khitan was still widely used during the Jin dynasty (1115-1234) of the Jurchen, and Khitan individuals are also known to have entered Mongolian service under the historical Mongols. However, as a spoken language Khitan seems to have become extinct soon after the Mongol period, and the same fate affected the other tribal languages that belonged to the Khitanic branch. The core areas of the former Khitan political power have since several centuries been Mongolian speaking (divided among several Inner Mongolian dialects of the Khorchin group), and it may be assumed that the last Khitan speakers had changed their language to Mongolian by the end of the Ming dynasty of China (1368-1644), though the exact date is not known. Even before that, the Khitan language had been lost among the Kara Khitai in Central Asia.
2. Khitanic and Jurchenic
Recent progress in the decipherment of the Khitan Small Script, as summarized by Kane [Kane 2009, cf. also Wu Yingzhe, Janhunen 2010: 13-48], has confirmed that Khitan was a distinct language, related to Mongolic, but clearly separate from the Proto-Mongolic lineage, from which all the other historical and modern Mongolic languages have evolved. Khitan is therefore best classified as "Para-Mongolic", meaning that it represented a branch collateral to Proto-Mongolic. Since Khitan was the language of an important ethnic group, or tribal union, which controlled the most powerful political state of its time in East Asia, it must have exerted considerable influence on the neighbouring languages, including Jurchen. Very probably, a part of the Jurchen elite of the 10th to 11th centuries were bilingual in Khitan. The power relationship changed after the collapse of the Liao dynasty, and in the 12th century an increasing number of Khitan speakers are likely to have become bilingual in Jurchen, though this bilingualism was extinguished by the subsequent expansion of the historical Mongols.
There is no historical information concerning the separation of Khitan and "regular" Mongolic from each other, but judging by the available linguistic data, the separation must have taken place at least several
centuries before the founding of the Liao dynasty, possibly even earlier. The Proto-Mongolic lineage seems to have differentiated from Para-Mongolic by way of gradual diffusion towards the north. While the Khitan speech community remained in the original Mongolic homeland in southwestern Manchuria, the new Proto-Mongolic homeland, from where the historical Mongols started their expansion, was located in northwestern Manchuria. In the intermediate zone there may have been transitional idioms: we do not know, for instance, what type of Mongolic was spoken by the Tatar confederation, which occupied the territory between the Khitan and the Mongols. The Proto-Mongolic lineage itself underwent at least some dialectal differentiation before it was unified once again during the Mongol Empire (1206-1368) [Janhunen 2008a: 130-134]. In any case, in view of the considerable difference between Proto-Mongolic and Khitan, the linguistic boundary between the two branches is likely to have been sharp, and the two types of Mongolic are unlikely to have been mutually intelligible.
The geographical history of Tungusic is in many respects analogous to that of Mongolic, in that the original Tungusic homeland was located in the south, in southeastern Manchuria, probably extending to northern Korea [Janhunen 2012a]. This was the later historical location of Jurchen-Manchu, from where this lineage extended northwards to Central Manchuria, including probably the Amur basin, where the Jurchen speakers became known to the Russians by the Dagur name "Jucher" [Janhunen 2004]. It is, of course, also possible that the historically documented Jurchen language was only one of several Jurchenic languages, and we do not know how close the mutual relationship of these languages would have been. Even so, soon after the breakup of Proto-Tungusic, a sharp boundary seems to have been formed between Jurchenic and the other major branch of Tungusic, Ewenic, which had a secondary homeland in the Middle Amur region, from where the Ewenic languages (Ewenki-Ewen) subsequently diffused all over Siberia and to some extent back to Manchuria (Khamnigan Ewenki, Orochen and Solon). Both geographically and chronologically, Jurchenic corresponds closely to Khitanic, while Ewenic corresponds to Mongolic (in the sense of the Proto-Mongolic lineage). There are only two differences between the two families: on the one hand, Jurchenic survived in the form of Manchu until modern times, while Khitanic became extinct; and, on the other, the Tungusic family also incorporates two intermediate branches known as "Amur
Tungusic" (Nanai and Udeghe), while similar intermediate branches on the Mongolic side, if they ever existed, were lost before being documented.
The historical and geographical situation also gives us some insights into the question concerning the types of Mongolic elements that are, or can be, present in Jurchen-Manchu. Since the Jurchenic lineage was originally located in southeastern Manchuria, while the Proto-Mongolic lineage was secondarily formed in northwestern Manchuria, it is unlikely that any direct contacts between these two lineages took place before the rise of the historical Mongols. During the political dominance of the Mongols in East Asia in the 13th to 14th centuries, in Manchuria extending to the 16th century, lexical elements naturally flowed into Jurchen-Manchu first from Middle Mongol and then from the eastern dialects of modern Mongolian. Later, during the Manchu Empire of the Qing (1636-1911), Manchu borrowings also entered Mongolian, and some modern groups of the eastern Mongols (the Khorchin, the Mongoljin, and possibly others) even seem to have been formed from a mixture of Manchu and Mongolian speaking elements. However, before the rise of the historical Mongols, the source of Mongolic loanwords in the Jurchenic branch of Tungusic would not have been Mongolic proper, but Khitan or some other Para-Mongolic languages.
3. Proto-Mongolic and Para-Mongolic
It is, consequently, reasonable to assume that the expansion and differentiation of Mongolic and Tungusic took place from the same general source region (southern Manchuria), in the same direction (from south to north), and, moreover, at approximately the same time and with the same speed. Although our knowledge of the historically documented Khitan language is still fragmentary at best, there are indications that Khitan linguistically relates to Proto-Mongolic in a way that is more or less analogous to the relationship of Jurchen-Manchu to Ewenic. The breakup of Proto-Tungusic into the Jurchenic and Ewenic branches (with the subsequent formation of the Amur Tungusic branches) may on the basis of lexical evidence roughly be dated to the "Iron Age" and can, with some certainty, be assumed to correspond to an absolute time depth of c. 2000 years. Quite possibly, then, the separation of the Para-Mongolic and Proto-Mongolic branches of Mongolic can have had an approximately similar depth.
However, irrespective of the issue of absolute dating, we may
distinguish between four sources of Mongolic elements in Tungusic,
as defined in terms of relative chronology:
(1) Pre-Proto-Mongolic, which terminologically refers to the common protolanguage of the Proto-Mongolic and Para-Mongolic branches. Since Pre-Proto-Mongolic ("Proto-Khitano-Mongolic") is likely to have been contemporary to Proto-Tungusic ("Proto-Jurcheno-Ewenic"), the loanwords in question would have been transmitted into the Tungusic protolanguage. The place of interaction must have been southern Manchuria, where the two protolanguage-level speech communities existed side by side, separated only by the river Liao [Janhunen 1996a: 232-233]. Very probably, the territory of central Manchuria was at this time still occupied by non-Mongolic and non-Tungusic languages, among which Amuric (the genetic context of later Ghilyak) would seem to have played an important role, as is possibly also suggested by the very name of the river Liao [Janhunen 2008b]. Since Pre-Proto-Mongolic loanwords were transmitted into Proto-Tungusic, they are equally likely to be present in all branches of Tungusic, including both Jurchenic and Ewenic (as well as the intermediate Amur Tungusic branches).
(2) Para-Mongolic, which refers to the Khitanic branch after the breakup of Pre-Proto-Mongolic. As already mentioned above, Para-Mongolic items were transmitted mainly into the Jurchenic branch of Tungusic, but not into the other branches. The place of interaction must still have been southern Manchuria, where both Khitanic and Jurchenic remained in the previous locations of their original homelands, while the Proto-Mongolic and Ewenic lineages had already moved northwards to their secondary homelands. It cannot be ruled out, of course, that some Para-Mongolic items can have been transmitted within the Tungusic family from Jurchenic into the other branches, but this must have been a minor phenomenon. The issue is complicated by the possibility that Proto-Mongolic itself may also have contained Para-Mongolic loanwords in addition to those lexical items that it shared with Para-Mongolic on a genetic basis.
(3) Proto-Mongolic, which refers to the common protolanguage of the extant Mongolic languages, and which linguistically corresponds to the Middle Mongol level of evolution. Originally, Proto-Mongolic would have been geographically adjacent to the early form of Ewenic (Proto-Ewenic), but during the height of the Mongol period, Middle Mongol elements did penetrate also into Jurchen, as they did even into Korean [Lee Ki-moon 1964], though direct contacts between Proto-Mongolic and the Amur Tungusic branches must have been very limited. The Proto-Mongolic loanwords in Tungusic show typically Middle Mongol features, which in the Mongolic context are archaisms, lost in the later Common Mongolic branch of Mongolic.
(4) Post-Proto-Mongolic, which refers to the premodern and modern Mongolic languages that developed from Proto-Mongolic during the last several centuries. In practice, the languages that have transmitted loanwords into Tungusic include Mongolian proper (the Khalkha and Khorchin groups of dialects), Buryat, Dagur and Khamnigan Mongol. On the receiving side we have the individual Tungusic languages, including both Manchu and Ewenki. Direct contacts of Post-Proto-Mongolic with the Amur Tungusic branches have been very limited, though there are occasional examples of possible Dagur loanwords in Amur Tungusic [Janhunen 1998].
4. Para-Mongolic and Pre-Proto-Mongolic
Focusing on the Para-Mongolic elements in Jurchen-Manchu, we may immediately note that they are not always distinguishable from the other layers of Mongolic loanwords. One problem is that some items have inherently such a "simple" phonological structure that they do not contain any feature that could be unambiguously identified as either Pre-Proto-Mongolic, Para-Mongolic or Post-Proto-Mongolic. To count as a verifiable Para-Mongolic element in Jurchenic, an item has to fill at least one of the following two types of criteria:
(1) Distributional: A Para-Mongolic item is likely to be attested only in the Jurchen-Manchu branch of Tungusic, since the Ewenic branch is unlikely to have been in direct contact with Para-Mongolic. Moreover, a Para-Mongolic item would ideally have to be attested not only in Manchu, but also in Jurchen, which
would confirm that it was present in the language already at the time when Para-Mongolic was still spoken. This latter criterion can, however, be ignored if the item otherwise fits into the patterns exhibited by Para-Mongolic elements.
(2) Formal: A Para-Mongolic item ideally shows differences as compared with the Proto-Mongolic lineage. Such differences may, in principle, be either semantic, morphological or phonological, but in practice only phonological features can be identified with any reliability at the present state of knowledge. Due to our incomplete knowledge of Khitan, it is still very difficult to reconstruct the Pre-Proto-Mongolic stage with any precision, but it is, at least, possible to identify items among the Mongolic elements in Tungusic that show substantial differences with regard to Proto-Mongolic. Such items are, then, likely to be of Para-Mongolic origin.
However, phonological features can be either archaisms or innovations, and strictly speaking only innovations peculiar to the Para-Mongolic lineage can be used as certain evidence of the Para-Mongolic origin of a given loanword. An item that contains a phonological archaism can, of course, also be of Para-Mongolic origin, provided that it can be shown that the Para-Mongolic lineage preserved that particular archaism, but it can likewise derive from Pre-Proto-Mongolic, which in any case would also have contained that archaism. In such cases, we may also consider the distributional criterion: if the item is attested only in the Jurchenic branch of Tungusic, then it is likely to be a later (mediaeval) loanword from Para-Mongolic, rather than an earlier (protohistorical) loanword from Pre-Proto-Mongolic. Even so, as far as the item contains no specifically Para-Mongolic feature, the dating must remain uncertain.
To take a concrete example: There are several both internal and external reasons to assume that Pre-Proto-Mongolic had the phoneme *n- (palatal nasal), which was later paradigmatically lost in the Proto-Mongolic lineage due to neutralization with *n (dental nasal). Manchu shows initial (*)n, orthographically <ni> or <niy>, in a few items that are clearly of a Mongolic origin, while Khitan also preserves a distinctive (*)n in the same words, as in niohe 'wolf' (< 'dog', by taboo) = Proto-Mongolic *noka.i 'dog' (incorporating the secondary class suffix *.i) = Khitan n.qo for fno£o (approximate reading, supported
also by Chinese transcriptions). Technically, such items could represent either a Para-Mongolic or a Pre-Proto-Mongolic source, but the fact that the word is attested on the Tungusic side only in the Jurchenic lineage would nevertheless support the assumption of a borrowing from Khitanic to Jurchenic. This may, in fact, be the case with all the Mongolic words containing an initial n in Manchu, although by far from all these items are attested in the currently available corpus of Khitan lexicon.
5. Archaisms and innovations in Para-Mongolic
The linguistic interaction between Para-Mongolic and Jurchen is likely to have concerned not only the lexicon, but also structural features. Unfortunately, although we know relatively well how Jurchen-Manchu developed from Proto-Tungusic [Benzing 1956], our understanding of the structural history of Khitan and its Pre-Proto-Mongolic background is still very imperfect, and, in particular, we have no information on how substantial the difference between Khitan (proper) and the possible other Para-Mongolic languages may have been. In this situation, it is not always clear whether the special features, especially phonological ones, exhibited by a Mongolism in Jurchen-Manchu are due to internal developments in Jurchen-Manchu, or to developments that had taken place already in the Para-Mongolic donor language.
Even so, we may assume that since Khitan and Jurchen were adjacent languages, with apparently relatively similar morphosyntactic structures, their phonological systems are also likely to have been very similar, if not identical. For instance, the types of suffixal vowel harmony that can be observed in Khitan [Kane 2009: 30-32] seem to have been very similar to what we know of the vowel harmony of mediaeval Jurchen, of which only traces are preserved in later Manchu [Kiyose 1997]. It may also be assumed that many phonological innovations were shared by Khitan and Jurchen on an areal basis. For instance, the Northeast Asian type of vowel rotation, as present in Korean and Manchu, as well as in several other languages [Janhunen 1981], seems to have been present also in Khitan, as suggested by Khitan "readings" such as funi 'ox' = Mongolic *üni-xe/n 'cow' vs. fon- 'to ride' = Mongolic *unu- id. [Janhunen 2012b: 122-123]. On the other hand, Khitan preserved some archaic features that are also typical of Jurchen-Manchu. For instance, initial *p (strong labial stop) is preserved as a labial in Khitan, as also in Jurchen (in Manchu > f), while in the Proto-
Mongolic lineage, as also in the Ewenic branch of Tungusic, it developed into a laryngeal (later mostly lost), as in fpar(a) 'people' = Mongolic *xara.n : pl. *xara-d (> modern Mongolian arad > ard) id. However, as was already mentioned above, archaisms in Khitan are not useful for distinguishing Para-Mongolic elements in Jurchenic from those borrowed already from Pre-Proto-Mongolic. For this purpose, only innovations are relevant.
The problem is that many idiosyncratic Mongolic items that are present in Jurchen-Manchu are not attested in the currently available Khitan corpus, although it is fully possible that they were used in the language. For this reason it is often not clear whether an innovation shown by a Mongolic element in Jurchen-Manchu is due to an internal development in Jurchenic or to an earlier development in the Para-Mongolic donor language. There are, for instance, a few words in which Jurchen-Manchu shows a single vowel in the first syllable instead of a diphthongoid sequence on the Mongolic side, as in Manchu fomo-ci 'stockings' = Mongolic *xoima-su/n id., Manchu niome-re 'octopus' = Mongolic *naiman > *naima/n 'eight' : *naima-ljin 'crab' (the latter item also shows an unexplained vowel correspondence, though the etymology is likely to be correct). The same relationship is observed also in non-initial syllables, as in Manchu ferhe 'thumb' (also in Nanai perxe) = Mongolic *xereke.i id., Manchu arfa 'barley' = Mongolic *arba.i id., but in these cases it is possible that the Mongolic items are secondary derivatives in * -i. It may be noted that diphthongoid sequences as such are not alien to Jurchen-Manchu, though they occasionally exhibit unexpected correspondences with regard to Mongolic, as in Manchu meihe 'snake' = Mongolic *moga.i id.; in this example the Manchu shape of the word could be due to metathesis, i. e. meihe < *mehei < *muhei, but this remains unconfirmed.
There are, consequently, many Mongolic elements in Jurchen-Manchu which may well have been borrowed from Para-Mongolic, but which could also represent earlier borrowings from Pre-Proto-Mongolic. In as far as such items are present only in Jurchen-Manchu — or also in the other Tungusic languages as secondary loanwords from Jurchenic — a Para-Mongolic source would appear likely for most of them and would be more or less confirmed if the items concerned were also attested in Khitan. In this respect, the recent growth of information on Khitan has turned out to yield unexpected results, in that certain
formally aberrant items that otherwise would look like good candidates for Para-Mongolic elements in Jurchen-Manchu seem to have been absent in the language of the Khitan inscriptions. There are two possible explanations for this: either (a) the items were lost in Khitan only after they had been transmitted to Jurchenic, or (b) they were borrowed from a Para-Mongolic language other than Khitan.
6. Para-Mongolic but not Khitan
Two coherent groups of Mongolic elements in Jurchenic that suggest a Para-Mongolic origin different from the Khitan language of the Liao period are the numerals for the 'teens' (the numerals of the second decade) and the terms for the basic colours.
(1) Numerals for the teens: These are attested as a complete set in Jin dynasty Jurchen in the shapes 11 fomshon, 12 fjorgon, 13 fgorhon, 14 fdurhon, 15 ftopohon, 16 fnolhun, 17 fdalhon, 18 fnohun,
19 fonohon. Rudiments of this system are preserved in later Manchu, from where some items were also secondarily transmitted to other Tungusic and non-Tungusic languages (notably Solon Ewenki and Dagur). The whole set from 12 to 19 is clearly based on Mongolic synthetic forms composed of the corresponding numeral roots plus an element f-hon ~ f-hun < *-kOn [Janhunen 1993], which may or may not be connected with the Mongolic item for
20 *korin (< *xorin). The numerals for the corresponding digits in Mongolic incorporate the element *-pAn, which, again, may or may not be connected with the Mongolic item for 10 *xar-ba/n < *par-pa/n. Importantly, the set for the teens is not attested in the Proto-Mongolic lineage, and also not in Khitan. Although the basic numerals of Khitan correspond to Mongolic [Janhunen 2012b: 118-119], the phonetic shapes of the Khitan items for the teens are unknown, but the way they are written with logograms suggests that they were expressed analytically (10 + 1 etc.).
In any case, the Jurchen data confirm that the numerals for the teens were originally present in some form of Mongolic. If they were borrowed from Para-Mongolic, then the donor language would have to have been either an idiom different from Khitan or a form of Khitan predating the language of the preserved inscriptions. If they were
borrowed from Pre-Proto-Mongolic, then we would have to assume that they were subsequently lost both in Khitan and in the Proto-Mongolic lineage. None of these alternatives can be fully ruled out, but certain phonetic details in the data would nevertheless seem to suggest a Para-Mongolic origin. For instance, the item for 18 fno-hun corresponds to Mongolic 8 nai-ma/n in a way that is analogous to that observed in niomere (no-me-re) 'octopus' (as discussed above). Of course, some of the items exhibit developments, including the vowel rotation in 13 fgorhon = Mongolic *gur- and 14 fdurhon = Mongolic *dör-, that can have taken place secondarily on the Jurchenic side.
(2) Basic colour terms: Most colour terms in Manchu have parallels in Mongolic, and the phonetic relationships indicate clearly that they are loanwords and not genetic cognates. There are four items of this type: Manchu: fulgiyan [fulgJan] = /f.u.l.gy.a.n/ 'red', shanyan < shanggiyan [saqgJan] = /sh.a.ng.gy.a.n/ 'white', niowanggiyan [^waqgJan] = /nw.a.ng.gy.a.n/ 'green' and genggiyen [geqgJen] = /g.e.ng.gy.e.n/ 'bright'. These correspond systematically to Proto-Mongolic *xulaxan, *cagaxan, *nogaxan and *gegexen, as well as to modern Mongolic (*)ulaan, (*)cagaan, (*)nogoon and (*)gegeen, respectively. However, the shapes of the Manchu items exhibit several both archaic and innovative features [Okada 1962] that cannot be explained by the known sound laws of either Mongolic or Tungusic, but which might point to Para-Mongolic origin. Among the archaisms we note the preservation of *n (in niowanggiyan), and among the innovations the deaffrication of *c (in shanggiyan), which may have been characteristic of the Mongolic donor language. Most importantly, we see the systematic representation of Mongolic *-xA- as Manchu -gya-, a correspondence that cannot be due to an internal Jurchenic development. On the other hand, the loss of the original second-syllable vowel, as in fulgiyan < * pulagyan, is a well-known Jurchenic development ("Mittelsilbenschwund") that may even have taken place after the Jin period. The curious representation of Mongolic *-gVx-as *-gg- > -nggy- in Jurchen-Manchu would, on the other hand, seem to be a Para-Mongolic feature.
It is, consequently, likely that the colour terms were borrowed from Para-Mongolic into Jurchenic after the breakup of Proto-Tungusic.
This conclusion is not affected by the fact that all of these items have in Jurchen-Manchu parallel forms ending in -hUn, i. e. Manchufulahun, shahun, niohon and gehun, which are used as moderative and/or feminine (in the zodiac) counterparts of the forms in -(ng)gyan. This formal dichotomy is itself likely to derive from a Para-Mongolic language, since it is not attested in the Proto-Mongolic lineage. Obviously, the colour terms are ultimately based on the simple roots *pula-, *ca-, *no- and *ge-, which are nowhere attested in their plain shapes. It is true that Mongolic has the essive-transitive verbs *xula-i-'to be(come) red', *ca-i- 'to be(come) white', *ge-i- 'to be(come) bright', but these may actually be based on forms containing the medial consonant *-g-, i. e. *pula-g-i-, *ca-g-i-, *ge-g-i- (with *-g- > *-x- > -y- > -0-). It may be noted that some of these roots, with other derivational suffixes, are also attested in the other Tungusic languages, as in Ewenki xula-riin, xula-ma 'red'; such data suggest a contact different from that underlying the Jurchen-Manchu colour terms.
What is, however, most intriguing is that the Jurchenic colour terms do not seem to have any counterparts in Khitan. The Khitan corpus does contain several colour terms (as used in the zodiacal calendar), but they are, surprisingly, completely different from those known from Mongolic and Jurchenic [Kane 2009: 176]. A case in point is the Khitan word for 'red', which is written l.iau.qu, suggesting a reading like fliauku and based on the root f liau = lyaw. In view of the fact that Khitan preserved initial *p as intact, there is no way to connect this item etymologically with Mongolic *pula- (cf. also [Kane 2006: 130-131]). This means that, very probably, the Jurchenic colour terms were borrowed from a Para-Mongolic language other than Khitan.
7. Lexical and derivational incompatibilities
Similar cases of non-correspondence, or only partial correspondence, between Jurchenic and Khitan are often encountered among random lexical items, as well as in the realm of derivation. For instance, the Manchu word aisin 'gold' shows an unexplainable relationship to Turko-Mongolic *alton (> Turkic *altun = Mongolic *altan) [Rozycki 1994: 13]. One could think that aisin reflects a Para-Mongolic shape of the word (though it has also been proposed that it has no relationship to the Turko-Mongolic data), but the problem is that this etymon is not attested
in Khitan, which used a totally different word, fnogu 'gold'. The history of Manchu aisin is further complicated by the fact that mediaeval Jurchen had fancun 'gold' (in Manchu > 'earring'), which may or may not belong to the same etymological family as aisin and/or *alton. On the other hand, a reflex of (*)aisin is also present in Ghilyak, which has aysng < *aisin 'gold', and it is not immediately clear whether this is a Jurchenic loanword in Ghilyak or an Amuric (Pre-Proto-Ghilyak) loanword in Jurchenic [Janhunen 2005: 75]. However this may be, there remains a possibility that both Manchu aisin and Jurchen ancun represent otherwise lost traces of items once used in Para-Mongolic languages other than the language of the Khitan inscriptions.
There are many other Jurchen-Manchu items that show unexplained and apparently irregular correspondences with regard to Mongolic, as in Manchu eihen 'donkey' = Mongolic *eljige/n id. (= Turkic *eshik < *eshike) and Manchu dolcin ~ dolgin 'wave' = Mongolic *dolgixa/n id. In other examples, an internal Pre-Proto-Jurchenic development might explain the idiosyncracy of the data, but this is difficult to confirm, as in Manchu enggemu 'saddle' = Mongolic *emexel < *emegel id.; in this case the Manchu data could be derived from the Mongolic by assuming an irregular assimilation (* emegel > *emegem), final vowel addition (> *emegemu), regular second-syllable vowel deletion (> *emgemu) and also a more or less regular nasal assimilation (> enggemu). Even so, the irregularities in the Jurchen-Manchu data could also be due to developments that had taken place in the Mongolic donor language, which, then, would have to have been a Para-Mongolic idiom — either Khitan or non-Khitan.
Some items show also unexplained, but systematic, derivational differences between Jurchen-Manchu and Mongolic. It is, for instance, not clear why there is variation in the presence or absence of a final *n in items such as Manchu fehi 'brain' = Mongolic *xeki/n (< *peki-n) 'head; beginning', Manchu bokson 'threshold' (by metathesis from *boskon) = Mongolic *bosuga (< *bos/u-ka) id. Interestingly, Jurchen fpon > Manchu fon 'time' corresponds to Mongolic *xon < *pon 'year' as far as the final *n is concerned, but semantically the Jurchen-Manchu data come closer to Khitan fpo 'time', which lacks the final nasal [Kane 2006: 127]. (It may be noted that Khitan and Proto-Mongolic
often show differences with regard to the final nasal, which itself may be identified as a derivative suffix with class-marking-related functions.)
A whole bundle of unsolved questions is connected with the representation of the Mongolic intervocalic consonant *x in Jurchen-Manchu. At the Proto-Mongolic stage, *x was a phoneme that contrasted in this position with both *g and *k [Janhunen 1999]. In Post-Proto-Mongolic and, in fact, already in late Middle Mongol, *x was lost, yielding sequences of either identical or non-identical vowels (long monophthongoids and diphthongoids, respectively). In Pre-Proto-Mongolic, however, *x is likely to have developed from *g, with which it still alternated in many paradigms in Middle Mongol (the type * cerig 'soldier' : plural *cerixlü-d). In the Mongolic loanwords transmitted to Jurchen-Manchu, *x is variously represented either as zero or as -g-, occasionally also as -h-, as in (zero:) adun 'herd' = Mongolic *aduxuln id.
(of horses), (-g---h-:) argan ~ arhan 'fang, molar, tooth' = Mongolic
*araxa id. (also *arixa > *ariya), (-h-:) Manchu gosihon 'bitter' = Mongolic gasixun id. The alternation of -g- with -h- may in most cases be discarded as a secondary phenomenon, but the difference between -g-and zero in the Jurchen-Manchu data is potentially significant. In Khitan, *-x- seems also to have been lost, as in Khitan fjau 'hundred' = Mongolic *jaxuln. It seems that the loss of * -x- in Khitan preceded the similar development in Post-Proto-Mongolic by several centuries. This means that items that show zero in place of *-x- in Jurchen-Manchu can have been received either from Khitan (or other Para-Mongolic) or from Mongolian of the late Middle Mongol or even later Post-Proto-Mongolic period. Leaving a more comprehensive treatment of this problem for the future, it may only be noted here that at least some of the items concerned exhibit archaic features, such as a preserved initial *p > f, as in Manchu fujuri 'origin' = Mongolic *ijaxur < *pijaxur ~*pujaxur. In cases where the Khitan cognate is actually attested we may confidently assume that the word was borrowed from Khitan into Jurchen, as in Manchu fulu 'surplus' < Khitan fpulu id. > 'intercalary month' = Mongolic *xülexü ~ *xilexü 'much, more' (probably ultimately from Pre-Proto-Mongolic *pile-gu).
8. Conclusion
It may be concluded that Jurchen-Manchu is likely to contain many Para-Mongolic loanwords. A large part of these must derive
from Khitan, which was the dominant language of Manchuria in the 10th to 12th centuries. However, only a few Mongolic items in Jurchen-Manchu can be definitively identified as having been borrowed from Khitan (proper), for this requires that they are attested in a corresponding shape in the extant Khitan corpus. It is likely that more items can be identified in the future. Still other items seem to have been borrowed from some other Para-Mongolic language or languages different from Khitan, though certainly also different from the Proto-Mongolic lineage. In such cases, the items show features that differ from the forms actually attested in Khitan. This suggests that there was some diversity within the Para-Mongolic group, though only a single variety of Khitan (with some chronological variation) is preserved in written documents.
Of course, many Khitan and other Para-Mongolic loanwords must also be hidden among those Mongolic elements in Jurchen-Manchu that do not show any unambiguous Para-Mongolic or other "irregular" features. For such items, only distributional criteria (attestation only in Jurchen-Manchu) and chronological information (attestation already in Jin period Jurchen and possibly attestation in Khitan) may serve as criteria for dating.
Bibliography
Alonso de la Fuente 2011 — J. A. Alonso de la Fuente. Tense, Voice and Aktionsart in Tungusic: Another Case of «Analysis to Synthesis»? [Tunguso-Sibirica 32]. Wiesbaden: Harrassowitz, 2011. Benzing 1956 — J. Benzing. Die tungusischen Sprachen: Versuch einer vergleichenden Grammatik [Akademie der Wissenschaften und Literatur in Mainz: Abhandlungen der geistes- und sozialwissenschaftlichen Klasse 11]. Wiesbaden: Steiner, 1956. Doerfer 1978 — G. Doerfer. Classification problems of Tungus // G. Doerfer, M. Weiers (eds.). Beiträge zur Nordasiatischen Kulturgeschichte [Tungusica 1]. Wiesbaden: Harrassowitz, 1978. S. 1-26. Doerfer 1985 — G. Doerfer. Mongolo-Tungusica [Tungusica 3]. Wiesbaden:
Harrassowitz, 1985. Janhunen 1981 — J. Janhunen. Korean vowel system in North Asian perspective //
Hangeul 172, 1981. P. 129-146. Janhunen 1993 — J. Janhunen. The teens in Jurchen and Manchu revisited // U.-M. Kulonen (ed.). Festschrift für Raija Bartens zum 25.10.1993 [Mémoires de la Société Finno-Ougrienne 215]. P. 169-184.
Janhunen 1996a — J. Janhunen. Manchuria: An Ethnic History [Mémoires de la Société Finno-ougrienne 222]. Helsinki: The Finno-Ugrian Society, 1996.
Janhunen 1996b — J. Janhunen. Prolegomena to a comparative analysis of Mongolic and Tungusic // G Stary (ed.). Proceedings of the 38th Permanent International Altaistic Conference (PIAC), Kawasaki, Japan: August 7-12, 1995. Wiesbaden: Harrassowitz, 1996. P. 209-218.
Janhunen 1998 — J. Janhunen. On chopsticks in Manchuria // Saksaha: A Review of Manchu Studies 3, 1998. P. 13-17.
Janhunen 1999 — J. Janhunen. Laryngeals and pseudolaryngeals in Mongolic: Problems of phonological interpretation // Central Asiatic Journal 43, 1, 1999. P. 115-131.
Janhunen 2004 — J. Janhunen. From Choson to Jucher: On the possibilities of ethnonymic continuity in Greater Manchuria // Studia Etymologica Craco-viensia 9, 2004. P. 67-76.
Janhunen 2005 — J. Janhunen. The lost languages of Koguryo // Journal of Inner and East Asian Studies 2, 2, 2005. P. 65-86.
Janhunen 2008a — J. Janhunen. Mongolic as an expansive language family // T. Kurebito (ed.). Proceedings of the International Symposium: Past and Present Dynamics: the Great Mongolian State: December 22-23, 2006, Tokyo University of Foreign Studies. Tokyo: Tokyo University of Foreign Studies, 2008. P. 127-137.
Janhunen 2008b — J. Janhunen. Liao: A Manchurian hydronym and its ethno-historical context // Studia Etymologica Cracoviensia 13, 2008. P. 89-102.
Janhunen 2012a — J. Janhunen. The expansion of Tungusic as an ethnic and linguistic process // A. L. Malchukov, L. J. Whaley (eds.). Recent Advances in Tungusic Linguistics [Turcologica 89]. Wiesbaden: Harrassowitz, 2012. P. 5-16.
Janhunen 2012b — J. Janhunen. Khitan: Understanding the language behind the scripts // Scripta 4, 2012. P. 107-132.
Janhunen 2013 — J. Janhunen. The Tungusic languages : A history of contacts // Kim Juwon, Ko Dongno (eds.). Current Trends in Altaic Linguistics: A Festschrift for Professor Emeritus Seong Baeg-in on his 80th Birthday. Seoul: Altaic Society of Korea. P. 17-60.
Kaluzynski 1962 — S. Kaluzynski. Mongolische Elemente in der jakutischen Sprache [Prace Orientalistyczne 10]. Warszawa: Panstwowe Wydawnictwo Naukowe, 1962.
Kane 2006 — D. Kane. Khitan and Jurchen // A. Pozzi, J. A. Janhunen, M. Weiers (eds.). Tumen jalafun jecen akü: Manchu Studies in Honour of Giovanni Stary [Tunguso-Sibirica 20]. Wiesbaden: Harrassowitz, 2006. P. 121-132.
Kane 2009 — D. Kane. The Kitan Language and Script [Handbook of Oriental Studies 19]. Leiden — Boston: Brill, 2009.
Khabtagaeva 2010 — B. Khabtagaeva. Mongolic elements in Barguzin Evenki // Acta Orientalia Academiae Scientiarum Hungaricae 63, 1, 2010. P. 9-25.
Kiyose 1997 — G. N. Kiyose. The collapse of palatal-velar vowel harmony from Jurchen to Manchu // Ä. Berta (ed.). Historical and linguistic interaction between Inner-Asia and Europe: Proceedings of the 39th Permanent International Altaistic Conference (PIAC), Szeged, Hungary, June 16-21, 1996 [Studia uralo-altaica 39]. Szeged: Department of Altaic Studies, University of Szeged, 1997. P. 147-152.
Kuz'menkov 1986 — E. A. Kuz'menkov. Kidan'skie elementy v man'czurskom i dialektnaja baza staromongol'skoj pis'mennosti // V. M. Solncev (ed.). "Istoriko-kul'tumye kontakty narodov altajskoj jazykovoj obscnosti": Tez. dok. XXIX sessii Postojannoj mezdunarodnoj altaisticeskoj konferencii (PIAC), Taskent, sentjabr', 1986 g. T. 2. Lingvistika. Moskva: Nauka, 1986. P. 71-72.
Kuz'menkov 1988 — E. A. Kuz'menkov. Mongol'skie elementy v man'czurskom i dialektnaja baza staromongol'skoj pis'mennosti // Voprosy jazykoznanija 4, 1988. P. 117-122.
Lee Ki-moon 1964 — Lee Ki-moon. Mongolian loan-words in Middle Korean // Ural-Altaische Jahrbücher 35, 1964. P. 188-197.
Okada 1962 — H. Okada. Color-names in Manchu // N. Poppe (ed.). American Studies in Altaic Linguistics [Indiana University Publications: Uralic and Altaic Series 13]. Bloomington: Indiana University, 1962. P. 225-227.
Rozycki 1994 — W. Rozycki. Mongol Elements in Manchu [Indiana University Uralic and Altaic Series 157]. Bloomington: Indiana University, Research Institute for Inner Asian Studies, 1994.
Vovin 2006 — A. Vovin. Why Manchu and Jurchen look so non-Tungusic? // A. Pozzi, J. A. Janhunen, M. Weiers (eds.). Tumen jalafun jecen akü: Manchu Studies in Honour of Giovanni Stary [Tunguso-Sibirica 20]. Wiesbaden: Harrassowitz, 2006. P. 255-266.
Wu Yingzhe, Janhunen 2010 — Wu Yingzhe, J. Janhunen. New Materials on the Khitan Small Script: A Critical Edition of Xiao Dilu and Yelü Xiangwen [Corpus Scriptorum Chitanorum 1 / Languages of Asia 9]. Folkestone: Global Oriental, 2010.
