Janhunen J. The Mongolic Languages

Подождите немного. Документ загружается.

tribal dialects. Khalkha itself, though relatively homogeneous, is normally divided into a

northern and a southern subdialect, with the northern dialect forming the basis of the

literary standard.

In a more general taxonomy, the six groups of Mongol dialects are often classified in

terms of three larger entities, which may be identified as the northern, eastern, and cen-

tral main dialects. In this framework, the northern main dialect may be defined as com-

prising the Outer Mongolian (Khalkha) group, as well as, possibly, the Shilingol group

of Inner Mongolian dialects. The central main dialect is essentially identical with the

Ulan Tsab (Chakhar) group, while the eastern main dialect corresponds to the Jerim

(Khorchin), Juu Uda, and Josotu (Kharachin) groups. In a further simplification of this

scheme it is perhaps possible to classify the northern (Khalkha) and central (Chakhar)

main dialects as one entity, which stands against the eastern main dialect (Khorchin,

Kharachin, and Juu Uda). It may, however, be questioned whether the Mongol dialects

can, or even should, be described in binary terms, for the actual isoglosses form an

extremely complicated non-binary network.

Many of the isoglosses dividing the Mongol dialects extend further to the neighbour-

ing Mongolic languages: Buryat and Khamnigan Mongol in the north, Oirat in the west,

and Ordos in the south, making it occasionally difficult to delimit the Mongol language

against its closest relatives. This is also reflected by the conceptions of the Mongol

speakers themselves. Ordos, in particular, is widely considered to be another (southern)

main dialect of Mongol, with which it is currently united by a common literary language

(Written Mongol) and administrative context (Inner Mongolia). Tsongol and Sartul, on

the other hand, spoken at the Russo-Mongolian border, are conventionally classified as

dialects of Buryat.

The taxonomical confusion is partly due to the arbitrary use of terms. The Mongol

terms for the concept of ‘dialect’ (vamav vayalqhu ‘oral accent’ or nuduq vayalqhu

‘local accent’), for instance, are often used without making a distinction between

‘dialect’ and ‘language’. The Chinese term for ‘ethnic group’ or ‘nationality’ (minzu) and

its Mongol counterpart (vuivdusudav) are even less specific and can be used without

almost any regard to the linguistic realities. Therefore, the officially recognized

‘Mongol’ nationality (Menggu zu) in China covers also the speakers of Buryat and Oirat.

Moreover, it also comprises populations that are at least synchronically non-Mongolic,

such as the fully Tibetanized Henan Mongols in Qinghai. An extreme case is formed by

the ‘Mongols’ of Yunnan, whose official status as ‘Mongols’ is based on the unverified

claim that their male ancestors may have included (Middle) Mongol-speaking soldiers of

the Yuan dynasty army.

The practical value of the criterion of mutual intelligibility for the classification of

Mongol dialects and Mongolic languages is to some extent obscured by the presence of

transitional phenomena. For instance, many forms of Oirat, including Kalmuck and the

dialects of Sinkiang, are more or less incomprehensible to ordinary Mongol speakers,

while other forms, notably the dialects spoken in Qinghai, are much closer to Mongol.

The Oirat in both Mongolia (Kobdo) and Inner Mongolia (Alashan) are well integrated

into the Mongol speech community through various degrees of bilingualism and

diglossia. Some originally Buryat-speaking groups, notably the Bargut of northern

Inner Mongolia, are currently also undergoing a language shift in favour of Mongol

proper.

180 THE MONGOLIC LANGUAGES

The degree of interdialectal intelligibility also depends on the extent to which any

given speaker is exposed to other dialects. In this respect, there is a considerable differ-

ence between Outer and Inner Mongolia. The dialectally diversified Inner Mongolian

Mongols are generally accustomed to communicating with speakers of other dialects.

Through public media they are even exposed to the normative variety of Khalkha. The

ordinary Khalkha speakers of Outer Mongolia, by contrast, have auditive experience

only of their own native dialect, which surrounds them from all sides. The literary stan-

dard of Khalkha is also, unlike Written Mongol, dialect-specific. Not surprisingly,

Khalkha speakers often have difficulties in understanding the dialects spoken on the

Inner Mongolian side.

DATA AND SOURCES

Apart from G. J. Ramstedt, who focused on the Khalkha dialect and its diachrony, the

first scholar to make a serious study of the dialectal diversity of the Mongol language

was A. D. Rudnev (1911). Rudnev’s published materials, containing texts, comparative

notes, and a glossary, cover with varying depth most of the Inner Mongolian dialects (as

well as Ordos). Ironically, the conclusion of Rudnev that ‘the time has not yet come for

a definitive solution of the problem concerning the classification of the Mongol dialects’

is still valid, with surprisingly little progress having been made after him, especially in

the field of general dialectal taxonomy.

Although the tribal and administrative framework of the Mongols of Inner Mongolia

(Manchuria) was excellently described by Owen Lattimore (1935), the only monographic

attempts to present a picture of the Inner Mongolian dialects after Rudnev seem to be

those by B. X. Todaeva (1960, 1981–5, 1997) and Sechen (1998). The works of Todaeva,

based on field materials from the 1950s, give a particularly systematic overview of all

the relevant dialects, adding both grammatical and lexical substance to the picture

presented by Rudnev. However, her materials still remain to be analysed in a more

modern linguistic framework.

In spite of the scarcity of generalizing works, a considerable number of detail studies

is available on the individual Mongol dialects and subdialects. In addition to the local

Mongolian and Inner Mongolian publications, the dialectological treatments of Darkhat

by G. D. Sanzheev (1931), Chakhar by Shirô Hattori (1951), Baarin by Chingeltei

(1961), Khorchin by James Bosson and B. Unensechen (1962), Kharachin by Masayoshi

Nomura (1950–1, 1957), Dariganga by András Róna-Tas (1960, 1961), and Üdzümüchin

by György Kara (1962, 1963), may be mentioned. Most of these works focus on phonol-

ogy, though some of them also supply grammatical and lexical material. The important

monograph by Kara on Jarut folklore contains also a general survey of the dialectologi-

cal situation in Inner Mongolia (Kara 1970: 268–78).

Important work on selected issues of Mongol phonology (umlaut and vowel reduc-

tion) has also been carried out by Hitoshi Kuribayashi (1985, 1988). More recently, com-

prehensive dialect monographs have started to be produced by native Inner Mongolian

scholars. These include the works by Chaganhada (1995) on the Khorchin dialect, by

Bayarmend Borjigin (1997) on the Baarin dialect, and by Mungungerel (1998) on the

Naiman dialect of the Juu Uda group. Several other monographs are in preparation,

including one on Chakhar by Borjigin Sechenbaatar (2003), and it may be presumed that

they will stimulate an increased interest in the general taxonomical and chronological

issues of Mongol dialectology.

MONGOL DIALECTS 181

Some aspects of the generational expansion of Mongol proper to populations

originally speaking other Mongolic languages are dealt with in the phonological case

study by Juha Janhunen (1988) of the current forms of speech of the Old Bargut. A sim-

ilar case of transition among the originally Oirat-speaking Manchurian Öelet (also

known as the Mannai-Öelet or Yeke Mingan) in Heilongjiang is discussed by Todaeva

(1988). The transitional (though essentially Mongol proper) dialects of Tsongol and

Sartul are discussed (in the context of Buryat) by Ts. B. Budaev (1965) and I. D. Buraev

(1965). A brief survey of the linguistic situation among the Yunnan ‘Mongols’ is given

by Mei W. Lee-Smith (1996).

PHONOLOGICAL TRENDS

There are several features of diachronic phonology which seem to be common to all

Mongol dialects and which, therefore, may be assumed to have characterized their com-

mon protolanguage. This protolanguage, or Proto-Mongol, appears to have a depth of

some centuries, and it must have been separated from Proto-Mongolic (with which it

should not be confused) by several more centuries. There are marked differences in the

speed of phonological evolution among the Mongol dialects, with some dialects still

remaining in a state close to Proto-Mongol and others being distanced from it by a num-

ber of additional innovations. Altogether, the Mongol dialects offer a picture of consid-

erable phonological diversity.

It is, however, difficult to find isoglosses that would be restricted only to the dialects

of Mongol proper, with the exclusion of all other Mongolic languages and dialects. In

most cases, the phenomena attested in Mongol are also present in one or more of the

neighbouring entities, notably Ordos, Oirat, or Buryat, and in some cases even more dis-

tant parallels can be found in Dagur, Moghol, and the languages of the Gansu-Qinghai

complex. Also, phenomena generally present in Mongol can be occasionally absent in

one or more Mongol dialects. This means that Proto-Mongol has to be defined in terms

of a rather diffuse bundle of isoglosses. Some of the most important features involved

include the following:

(1) The loss of initial *x, e.g. (*xulaxan >) *xulaan ‘red’ > *ulaan. Although this

development (like the diachronically somewhat earlier loss of medial *x) may well be

described as Common Mongolic, it is transitional in the direction of Dagur, which basi-

cally retains initial *x, but has lost it in the Hailar dialect, obviously because of interfer-

ence from the neighbouring dialects of Mongol proper.

(2) The spirantization of *k in all positions, e.g. (*köke >) *kökö ‘blue’ > *xöxö. This

development, with a varying degree of phonologization, is shared by Eastern Buryat,

while in Western Buryat, Oirat, and Dagur it has contextual restrictions (being mainly

limited to the position before original back vowels). Conspicuously, however, it is absent

in both Ordos and Khamnigan Mongol, leaving these languages clearly outside of the

context of Mongol proper.

(3) The merger of final *ng and *n into a single neutralized nasal, which is repre-

sented as ng (phonetically often transformed into diffuse vowel nasalization), e.g. *on

‘year’ > ong vs. *ang ‘game’ > ang. This development is absent in Ordos and Oirat, but

it is shared by several other Mongolic languages, including Buryat and Khamnigan

Mongol. The segmental identity of the neutralized nasal (n or ng) depends on the phono-

logical configuration of each given idiom.

182 THE MONGOLIC LANGUAGES

(4) Palatal breaking, e.g. *mingga/n ‘thousand’ > *myangga/n. This phenomenon is

shared by Buryat and Dagur, but not by Oirat, Ordos, or Khamnigan Mongol. There is,

however, a certain areal transition, for while palatal breaking is completely absent in

Khamnigan Mongol, it is attested in Oirat and Ordos word-initially (original vowel

anlaut) as well as after sibilant consonants, e.g. *imaa/n ‘goat’ > *yamaa/n, *sira

‘yellow’ > *syara (*shara), suggesting a connection with prebreaking (which has a

similar distribution) in cases like *mïka ‘meat’ > *maxa.

(5) The neutralization of the distinction between *A and *U in non-initial syllables,

e.g. *usu/n ‘water’ > *usa/n. With the exception of the palatal and labial harmony, this

development may be seen as the first stage in a series of neutralizations affecting the

vowels of non-initial syllables. This stage is still preserved in Buryat, while Mongol

proper and Oirat have generally proceeded further. On the other hand, no neutralization

has taken place in Ordos and Khamnigan Mongol.

(6) The metathetic anticipation of third-syllable *i, e.g. *ularil ‘turn’ > *uliral,

(*toxori- >) *toori- ‘to circle’ > *toiro-. This feature also seems to be connected with the

general tendency of vowel neutralization in non-initial syllables. Common to Mongol

proper and Oirat, it is absent in Ordos, Buryat, and Khamnigan Mongol.

(7) The loss of non-initial-syllable short (single) vowels, e.g. *tala ‘steppe’ > tal,

*arad ‘people’ > ard. This is one of the few innovations which leave Tsongol and Sartul

outside of the rest of Mongol proper, for these dialects still seem to preserve a contrast

between a vowel segment and zero in cases like tala ‘steppe’ vs. gal ‘fire’. There are

indications that a similar contrast existed until recently also in some forms of actual

Khalkha, though the general trend in both Mongol and Oirat has been to eliminate all

short vowels of non-initial syllables. Phonetically, secondary vowel segments can be pre-

sent (as indicated by the Khalkha Cyrillic orthography), but phonologically they are non-

distinctive. A secondary contrast between zero and a distinctive schwa (°) seems to be

developing in some Mongol dialects under specific morphological conditions, as in

Khalkha sawx ‘chopstick’ vs. part. fut. yaw°x ‘to go’.

(8) The monophonemization of long (double) vowels in non-initial syllables, e.g.

*ulaan ‘red’ > (*)ulang. This development, apparently shared by most Mongol dialects

as well as Oirat, is an automatic consequence of the loss of the distinctive short vowels

in the same position. As a result, there is a synchronic imbalance between the vowel sys-

tems of the initial and non-initial syllables, as the original long (double) vowels remain

distinctive only in the initial syllable. A further source of imbalance is created by the

diphthongoid sequences, which dialectally still retain their distinctivity as long vowel

elements in non-initial syllables.

(9) The merger of the diphthongoid sequence *e(y)i with the long (double) vowel *ii,

in non-initial syllables > i, e.g. *teime ‘so’ > *tiime >

tiim, (*bexelei >) *beelei

‘mitten/s’ > *beelii > beeli. In suffixes with vowel harmony, both *e( y)i and *ö( y)i are

normally represented as *ee > e, e.g. *köl ‘foot’ : poss. *köl-tei > Khalkha xöl-te.

Although these developments have parallels in many Mongolic languages, the sequence

*e( y)i remains distinctive in Khamnigan Mongol and Dagur, thus marking a difference

with regard to Mongol proper.

(10) The origination of a new contrast between final ng and n, e.g. *xan ‘prince’ >

xang vs. *xana ‘wall’ > xan. Since this is also a consequence of the loss of the short

vowels in non-initial syllables, no new contrast of this type is present in idioms which,

like Buryat and Khamnigan Mongol, retain the original vowel segments. The develop-

ment is present in Dagur, but it is absent in Ordos and Oirat, which retain the contrast

between the original final *ng and *n.

MONGOL DIALECTS 183

(11) The origination of a contrast between g (velar stop) and gh (postvelar stop or

velar fricative), e.g. *bag ‘band’ > bag vs. *baga ‘small’ > bagh. This contrast is

basically parallel to that between the secondary final n and ng, but, owing to morpho-

logical analogy, it has spread to medial position also, e.g. abl. (bag :) bag-aas vs. (bagh :)

bagh-aas. Unfortunately, the status of gh in Mongol dialects remains unclarified.

Although gh seems to have existed as a distinctive segment in Proto-Mongol, it has

(re)merged with g in several dialects and idiolects, especially on the Inner Mongolian

side. The contrast is, however, present in Oirat.

In addition to the above general trends, which mainly delimit the Mongol language

against its neighbours, there are phenomena characteristic of restricted groups of Mongol

dialects, thus potentially providing a basis for an interdialectal classification. A compre-

hensive analysis of these phenomena and their distribution still remains to be carried out,

but some of the better known features are listed below:

(12) The merger of short *ö with *ü, e.g. (*edür >) *ödür ‘day’ > *öder > *üder (>

üdr). This development is well known from Eastern Buryat, Khamnigan Mongol, and

Dagur, but it also occurs in Tsongol and Sartul, as well as, apparently, in some of the

neighbouring northern Khalkha dialects. It is also present as a sporadic phenomenon in

several Inner Mongolian dialects, notably in the Jerim, Juu Uda, and Josotu groups.

Ultimately, it seems to be a question of a need to eliminate *ö from the vowel system due

to the effect of rotation (which tends to make the position of *ö paradigmatically prob-

lematic). In a similar way, short *e has merged with *i in the Ulan Bator (sub)dialect of

Khalkha, and very possibly also in some dialects on the Inner Mongolian side.

(13) The dissimilatory weakening of initial strong stops before a medial strong stop

or fricative, e.g. *tata- ‘to draw’ > *data- (> dat-), *casu (*casu/n) ‘snow’ > *casa >

*jasa (> jas). Attested in the Inner Mongolian dialects of the Ulan Tsab and Shilingol

groups, the Ongniut-Naiman-Aokhan subdialect of the Juu Uda group, as well as in

southern Khalkha and Ordos, this feature involves an areal innovation that links Ordos

with the southern dialects of Mongol proper. Moreover, this innovation appears to be rel-

atively old, since it has affected initial *k before its (phonological) spirantization, e.g.

*kökö ‘blue’ > *gökö > göxö (>

göx).

(14) The deaffrication of *c into sh, e.g. (*cino >) *cono ‘wolf’ > shon. This feature

delimits rather unambiguously the Khorchin (Jerim) group of dialects against all other

Inner and Outer Mongolian dialects of Mongol proper. There may nevertheless be an

areal connection with the similar development in Buryat. In most Buryat dialects both *c

and *j are deaffricated, but in Bargut (spoken immediately north of Khorchin) the deaf-

frication affects only *c, which is also the case in Khorchin. As a sporadic phenomenon,

the deaffrication of *c, especially before *i, has a wider distribution, extending also to

Dagur.

(15) The merger of *s with *d or *t. This feature is also attested in Buryat and

Khamnigan Mongol, where syllable-final *s is represented as *d. The same phenome-

non, but apparently as a separate innovation, seems to be characteristic of at least some

subdialects of Tümet, e.g. (*ulus >) *uls ‘state’ > uld. More diagnostically, a complete

paradigmatic merger of *s with *t (in all positions) has taken place in the Jalait-Dörbet

dialect of the Jerim group, e.g. *sara ‘moon; month’ > tar, (*bos- :) *boso- ‘to rise’ >

bot-. In this dialect, the paradigmatic position of *s has apparently been taken over by

*sh (< *sh & *c), e.g. (*casu >) *shas > sat.

(16) The simplication of the cluster *ngg to ng, e.g. *monggol ‘Mongol’ > *mongol

(> mongl). This feature is characteristic of several Inner Mongolian dialects, but its

184 THE MONGOLIC LANGUAGES

phonological relevance remains disputable. In principle, it could be a question of the

expansion of the new distinctive ng (< *n & *ng) from the final to the medial position.

However, all the relevant dialects seem to preserve the cluster ngg as distinctive in final

position, e.g. (*mingga >) myangg ‘thousand’ vs. xang ‘king’. Moreover, since no sim-

plification has taken place in the structurally similar clusters *mb *nd *nj (nasal plus

weak stop), it appears best to analyse the phonetically observed medial ng [] as a

variant of the cluster ngg.

SEGMENTAL PHONEMES

Among all the phonological isoglosses uniting, separating, and intersecting the dialects of

Mongol proper there are two that have a particularly profound overall paradigmatic effect.

These two isoglosses are those of palatalization and umlaut, and whatever is generally

said about the possibilities of a binary classification in Mongol dialectology, they divide

the Mongol dialects into two relatively well-defined groups that may be termed the

palatalization dialects and the umlaut dialects. Basically, it is a question of where the syn-

chronic distinctions based on the feature of palatalness are located in the phonological sys-

tem. In the palatalization dialects these distinctions are mainly connected with

consonantal contrasts, while in the umlaut dialects they are contained in the vowel system.

In terms of conventional classification, the palatalization dialects correspond to the

northern (Khalkha) and central (Chakhar) main dialects, while the umlaut dialects corre-

spond to the eastern main dialect (Khorchin, Kharachin, and Juu Uda). Since the eastern

main dialect has both geographically and demographically a dominant position in Inner

Mongolia, the phenomenon of umlaut may also be regarded as a characteristically Inner

Mongolian feature, while the phenomenon of palatalization is focused on Outer Mongolia.

Both phenomena have, however, parallels outside of the context of Mongol proper.

In Proto-Mongol (as already in Proto-Mongolic) palatalness was an inherent (primary)

property of two groups of segments: the front vowels *e *i *ö *ü, on the one hand, and

the palatal consonants and glide *c *j *sh *y, on the other. Among the front vowels, the

position of *i (< *i & *ï) was exceptional, in that it was able to occur in both front-vocalic

and back-vocalic words. Among the palatal consonants, *sh was a secondary phoneme,

which initially had distributional restrictions. All the palatal consonants had the restriction

of being unable to occur syllable-finally. There were, however, no restrictions concerning

the combining of palatal vowels and palatal consonants in a single word.

In the palatalization dialects, as exemplified by Khalkha (Table 7.2), the Proto-

Mongol consonant system was profoundly changed with the origination of a complete

series of palatalized consonants, which contain the feature of palatalness as a secondary

articulation. Apart from palatal breaking, palatalized consonants were produced by two

developments, both of which involve the vowel *i: vowel contraction, as in (*kixag >)

*kiyag ‘reed’ > xyaag; and vowel reduction, as in *gobi ‘desert’ > goby (> gowy).

Breaking and contraction seem to have been interconnected, in that both yielded palatal-

ized consonants for the syllable-initial (prevocalic) position. Vowel reduction, on the

other hand, yielded palatalized consonants for the syllable-final position. Obviously, the

vowel *i in non-initial syllables was first qualitatively reduced to an indifferent reduced

vowel preceded by a palatalized consonant which, when the vowel was lost, became

phonemic, e.g. *mori (: *mori/n) ‘horse’ > *morye > mory.

Importantly, palatal breaking had a reverse effect on the original palatal stops (sibi-

lant affricates) *c *j, which in the palatalization dialects underwent dentalization before

MONGOL DIALECTS 185

vowels other than *i, e.g. *cag ‘time’ > tsag, *jam ‘road’ > dzam. Before *i, including

cases of breaking, the segments *c *j remain phonetically palatal, but in the synchronic

system they have become incorporated into the palatalized series, e.g. *cida- ‘to be able’ >

*cyada- > c( y)ad-, *jira (: *jira/n) ‘sixty’ > *jyara- > j(y)ar. The palatal continuant *sh

did not undergo dentalization, but paradigmatically it has merged with the palatalized

reflex of *s produced by palatal breaking in cases like *sira ‘yellow’ > *syara > shar.

The phenomenon of consonant palatalization is shared by Buryat and Dagur. Unlike

in these languages, however, palatalized consonants in Mongol proper are generally not

distinctive in original front-vocalic words, a restriction that seems to be due to the merger

of *i (the ultimate source of palatalization) and *e in non-initial syllables already before

the phonemization of the palatalized consonants. This merger is also attested (though

perhaps not fully completed) in Ordos, e.g. *beri ‘daughter-in-law’ > Ordos bere >

Mongol ber, *üliger ‘tale’ > Ordos üliger ~ üleger > Mongol ülgr, *üniye (: *üniye/n)

‘cow’ > Ordos ünee > Mongol üne. The only palatalized consonants that synchronically

occur in front-vocalic words are the segments c j sh y, which thus still manifest their orig-

inal status as inherently palatal (rather than secondarily palatalized) consonants, cf. e.g.

*cidör ‘hobble’ > *cyödör > cödr (dialectally also > cüdr, shüdr), *küci (: *küci/n)

‘power’ > xüc (dialectally >

xüsh).

In the umlaut dialects of Mongol proper, the segments c j sh y fully retain their status

as inherently palatal (not palatalized) consonants. The segments c j sh are systematical-

ly opposed to the dentals t d s, but there are no dental (dentalized) affricates of the type

ts dz. The consonant system of these dialects, as perhaps best exemplified by the dialects

of the Juu Uda group (Table 8.1), is therefore still synchronically more or less identical

with that of Proto-Mongolic.

In difference from Proto-Mongolic, the synchronic system in the umlaut dialects con-

tains the Common Mongolic marginal phonemes p k w, to which f and possibly lh have

been added in some subdialects or idiolects. The modern velar fricative x (< *k) and nasal

ng (< final *n) are also new phonemes. There is no certain information concerning the

distribution and status of the postvelar segment gh in the umlaut dialects, but it may well

exist as a separate phoneme for some speakers.

Slight dialectal variations in this system are caused by dialect-specific innovations,

including the developments *c > sh in the Jerim group and *s > t (as well as *sh > s) in

the Jalait-Dörbet dialect. The dialects of the Josotu group, especially Kharachin, are

known to exhibit the Mandarin type of differentiation between palatal (before i) and

retroflex (before other vowels) realizations of c j sh. This difference, adopted directly

from Chinese, is, however, not phonemic in native Mongol words, and the segments c j

186 THE MONGOLIC LANGUAGES

TABLE 8.1 JUU UDA CONSONANTS

pt c k

bd j g

sshx

mn ng

sh may still, even when pronounced as retroflexes, be phonologically classified as

palatals.

Whatever the dialectal details, the general picture is that the more than thirty consonant

phonemes of the palatalization dialects constitute a much larger selection of paradigmatic

resources than the less than twenty consonant phonemes of the umlaut dialects. This

difference is counterbalanced by the fact that the umlaut dialects have a richer vowel

system, which, most importantly, contains the secondary palatal vowels ä [æ ] and

ö [œ ø]. The origin of these vowels is connected with the phenomenon of umlaut,

according to which the original back vowels *a *o of the initial syllable were fronted

under the impact of a subsequently lost second-syllable *i, e.g. *tabi (: tabi/n) ‘fifty’ >

täb (> täw), *mori (: mori/n) ‘horse’ > mör. In other words, the umlaut vowels bear the

same distinctive load as the syllable-final palatalized consonants in the palatalization

dialects.

The phenomenon of umlaut was also active in the diphthongoid sequences ending in

*i, but in these cases the resulting palatal vowels are phonologically long (double), e.g.

*a(y)il ‘camp’ > ääl, *no( y)ir ‘sleep’ > nöör. In non-initial syllables, the long palatal

vowels (like all long vowels) yield synchronically short (single) segments, e.g. *dalai

‘sea’ > *dalää > dalä, *oroi ‘top’ > *oröö > orö. Dialectally, in the Josotu and Juu Uda

groups, an umlaut vowel of the second syllable can also regressively influence the vowel

of the first syllable (double umlaut), cf. e.g. Jerim dalä vs. Juu Uda dälä, Jerim orö vs.

Juu Uda örö.

Although umlaut itself is a simple process, its paradigmatic impact is complicated by

many additional factors, because of which the actual vowel systems of the umlaut

dialects vary considerably. One complicating factor is vowel rotation, because of which

the original front vowels *ö *ü have in the umlaut dialects the fully velarized values

[o u], while the original back vowels *o *u have the lowered values [ ] (accompanied

by a varying degree of pharyngealization). Possibly for this very reason, the umlaut

counterpart of *u has generally merged with that of *o, e.g. *kubi ‘share’ > xöb (> xöw)

(possibly through *

xobi). The diphthongoid sequences *u(y)i and *ü( y)i, however, yield

the distinctive long umlaut vowel üü [y], e.g. *ku( y)ika ‘skin; peel’ > xüüx, *ü( y)ile

‘deed’ > üül, though the sequences wii and ii are also attested dialectally, e.g. *u( y)ila-

‘deed’ > wiil- ~ üül-, *kü( y)iten ‘cold’ > xiitng (possibly through *xwiitng).

The monophthongization of the diphthongoid sequences *a( y)i *o( y)i *u( y)i *ü( y)i,

as well as, theoretically, *ö( y)i, together with the even more widespread elimination of

the sequence *e( y)i, means that the umlaut dialects, unlike the palatalization dialects,

have synchronically only short (single) and long (double) vowels, but no diphthongoid

sequences. The paradigm of long vowels is, however, larger than the paradigm of short

vowels, since the long vowel üü has no short counterpart. Although this imbalance has

analogies in Buryat (long öö with no short counterpart) and Ulan Bator Khalkha (long ee

with no short counterpart), its phonological implications remain unclear. Possibly, it

means that the long vowels (with a larger paradigm) should be understood as more basic

(less marked) than the short vowels (with a smaller paradigm). In any case, the imbal-

ance suggests that the vowel system of the umlaut dialects is unstable and potentially

open to restructuring. One possible way of balancing the system would be the emergence

of a short ü

as a distinctive segment.

Disregarding the paradigmatic imbalance between short and long vowels, and

focusing only on the qualitative distinctions, the vowel system of the umlaut dialects, as

exemplified by the Juu Uda group (Table 8.2), comprises ideally ten distinctive qualities,

which may be compared with the only seven qualities of Khalkha (Table 7.1).

MONGOL DIALECTS 187

In the general context of Mongolic, the umlaut dialects of Mongol proper are unique,

in that they combine the effects of rotation and umlaut in a single system. Thus, the new

umlaut vowels *ö *ü are synchronically distinct not only from the rotated values of the

original back vowels *o > o resp. *u > u, but also from those of the original front

vowels *ö > ó resp. *ü > u, cf. e.g. xön ‘sheep’ < *koni (: *koni/n) vs. xól ‘foot’ < *köl

vs. xor ‘poison’< *kor, jüül ‘sort’< *jü(y)il vs. juung ‘left’< *jexün vs. juur ‘on the way’

< *jaxura. This is a substantial difference compared with Oirat, which also has umlaut,

but which merges the umlaut vowels with the original (non-rotated) front vowels.

It seems that the rather intricate phonetic distinctions between the vowels o ó u u are

easily open to various dialectal or idiolectal mergers and reinterpretations, though the

details remain to be investigated. The vowel *ö > ó, in particular, is likely to be absent

in the synchronic system of some umlaut dialects due to its merger with either *ü or *e

(whose rotational value is dialectally accompanied by roundedness). Under such condi-

tions, the vowels uu o (including their long counterparts) may be dialectally regrouped

to a set of rounded (non-pharyngealized) back vowels with three degrees of opening,

contrasting systematically with the original unrounded vowels i e a and the secondary

umlaut vowels ü ö ä.

A major problem in the diachronic phonology of the umlaut dialects concerns the

status of breaking. While the umlaut vowels primarily bear the distinctive load that in

the palatalization dialects is borne by syllable-final palatalized consonants, they also

occur in the place of the broken and contracted vowels of the palatalization dialects. In

practice, it is always a question of the umlaut vowel ä, e.g. *biraxu ‘calf’ > bäru vs.

Khalkha byaru, *uliyasu (: *uliyasu/n) ‘poplar’ > u

läs ~ ulas vs. Khalkha ulyas. Thus, in

addition to its basic function as the marker of the regressive influence of *i on a preced-

ing *a, the umlaut vowel ä also indicates the regressive influence of *a on a preceding

*i. In the latter case, the vocalic palatalness corresponds to the feature of syllable-initial

(prevocalic) consonantal palatalization in the palatalization dialects.

The most likely explanation of the dialectal correspondences is that palatal breaking

was, indeed, a Proto-Mongol feature, which was originally also present in the ancestral

forms of the umlaut dialects. This is, in particular, suggested by the cases of initial break-

ing (vowel anlaut) of the type *imaa/n ‘goat’ > *yamaa/n, which are present in all

Mongol dialects. However, postconsonantal breaking may initially have existed at the

phonetic level only, without involving phonemic palatalization in the consonant system.

Umlaut must also have been a phonetic phenomenon until it became phonological with

the reduction and loss of all short vowels in non-initial syllables. Obviously, as soon as

the umlaut vowels became phonemized, they offered a point of identification for the

palatalized sequences produced by breaking and contraction.

Because of their dual origin, the umlaut vowels are sometimes (though in practice

very rarely) diachronically ambiguous, for they can imply both actual umlaut and palatal

breaking, as in xäru ‘answer’ < *karixu (umlaut, cf. Khalkha xaryu) and ‘rimfrost’

< *kiraxu (breaking, cf. Khalkha xyaru). In most cases, however, the umlaut dialects and

188 THE MONGOLIC LANGUAGES

TABLE 8.2 JUU UDA VOWELS

üu ui

öo ó e

äa

the palatalization dialects are in a transparent one-to-one correspondence, which allows

speakers of both types of Mongol to communicate with each other. The position of the

central main dialect (especially Chakhar) in this framework remains to be worked out;

there are indications that this dialect, though basically featuring the phenomenon of con-

sonant palatalization, may nevertheless have vocalic distinctions reminiscent of the

umlaut dialects.

MORPHOLOGY

Morphologically, all Mongol dialects are largely congruent with the patterns exhibited by

the Khalkha literary language. The literary standard of Modern Written Mongol, as used

in Inner Mongolia, also basically incorporates the Mongol type of morphological

structure, as is evident from such crucial sections of morphology as the systems of case

endings, verbal suffixes, and pronominal declension. The Mongolic language closest to

Mongol proper in the morphological respect is Ordos, which for this reason is particu-

larly well positioned for being included within the cultural and educational framework

offered by Written Mongol.

There are more differences against the other neighbouring languages. The apparently

most diagnostic morphological feature of Mongol proper and Ordos is the absence of

verbal predicative endings. Since predicative endings (based on the pronominal nomina-

tives) are present in Buryat, Khamnigan Mongol, Dagur, and Oirat, their absence in

Mongol and Ordos may actually involve an innovation, perhaps connected with the influ-

ence of Manchu and Chinese. (It should be noted, though, that the personal endings in

Buryat, Khamnigan Mongol, Dagur, and Oirat are also an areal feature shared with

neighbouring non-Mongolic languages.)

Another diagnostic morphological feature of Mongol proper is its system of finite

temporal-aspectual forms, which synchronically includes finitely used participles. In par-

ticular, the use of the perfective participle as an unmarked (plain) past tense form,

e.g. Khalkha yaw-sng ‘[he] went’ < *yabu-gsan, marks a clear difference against Buryat

and Khamnigan Mongol, in which this function is filled by the imperfective participle,

e.g. Khamnigan Mongol yab-oo id. < *yabu.xa. Importantly, the Sartul and Tsongol

dialects conform with Mongol proper in this respect. On the other hand, Dagur, which

also uses the perfective participle in the function of a finite past tense, differs from

Mongol proper in many other respects, including, for instance, the functions of the

terminative form (past tense in Mongol proper vs. present-future tense in Dagur).

There are also minor morphological differences among the Mongol dialects. These

are due to occasional archaisms, such as, for instance, the preservation of the original

comitative case (> -lä) in some Inner Mongolian dialects of the Juu Uda and Josotu

groups (as also in Ordos), or innovations, which are particularly numerous in the system

of converbs. It is more difficult to establish what kind of functional differences, if

any, there are in the use of the morphological forms in the various dialects. It may well

be, though the matter remains to be investigated, that such differences exist in, for

instance, the modal and aspectual details of the original finite forms of verbal conjuga-

tion (especially the terminative, confirmative, and resultative, all which have mainly a

past tense reference with a possibly dialectally varying modal and aspectual content).

The most easily accessible morphological differences are those connected with the

phonological shapes of morphological markers. Perhaps the most important such

difference involves the accusative marker, which in Khalkha invariably contains

MONGOL DIALECTS 189