Emphatic reduplication in Oroqen and its Altaic context(*).
Oroqen, a Tungusic language spoken in China, uses the partial reduplication of adjective stems to indicate intensity. This process of "emphatic reduplication" is widespread in Altaic languages, which raises the question of whether it might be a reflex of an archaic genetic trait of Altaic. We argue, to the contrary, that the presence of reduplication in Oroqen is the result of borrowing, most likely from a Mongolic language. Inter alia, the highly restrictive nature of the reduplication in Oroqen, its failure to adhere to Oroqen phonotactics, and the distribution of reduplication in other Tungusic and Mongolic languages are offered as evidence to support this position. Finally, we also provide data that point to the disappearance of emphatic reduplication in Oroqen.
Within Turkic, Mongolic, and Tungusic languages (i.e. Micro-Altaic), there is a widespread morphological phenomenon referred to as "emphatic reduplication." The process involves, among other properties to be described below, the partial reduplication of an adjective stem; thus from Turkish belli `obvious' is derived besbelli `completely obvious'. As these glosses suggest, the reduplication is typically utilized to indicate the presence of a property concept to the utmost degree. Emphatic reduplication is conceivably an archaic Altaic process that has been passed down to many languages within the sister branches of Turkic, Mongolic, and Tungusic.(1) If it could be demonstrated that emphatic reduplication were a shared genetic trait, this would represent an important piece of evidence for the existence of Micro-Altaic, a proposal that continues to attract its fair share of skeptics (among many others, see Clauson 1956; Doerfer 1985; Janhunen 1996). Unfortunately, this sort of reduplication is thoroughly described for a select few Altaic languages, particularly in the Turkic family where the process is fairly productive (e.g. Turkish by Swift 1962; Ozdemir 1986; Dobrovolsky 1987). The relative lack of information drawn from Mongolic, where the process is far more lexically restricted, and from Tungusic, where it is similarly restricted and only found in a few languages, makes it difficult to assess whether emphatic reduplication is best seen as a genetic attribute of Altaic, or whether its distribution might be better explained as the result of contact-induced borrowing.
In this paper, we provide a detailed overview of emphatic reduplication in Oroqen, a member of the northwestern branch of the Tungusic language family spoken in 12 towns in the northeast of China. After furnishing a description of the process in Oroqen and demonstrating the deleterious effect that language obsolescence has had on the use of reduplication in Oroqen, we return to the broader question of reduplication within Altaic. Based on the facts from Oroqen and other observations about the operation of reduplication in Altaic languages, we argue that emphatic reduplication is borrowed into Tungusic, most likely from Turkic via Mongolic. While our specific conclusion, when taken alone, has minimal import for deciding the question of whether Micro-Altaic languages do in fact form a genetic unit, it represents a notable example of how contact among Micro-Altaic languages has brought about unusual structural similarities.
1. Reduplication in Oroqen
The most straightforward statement of emphatic reduplication in Oroqen is that it operates by copying the first syllable of an adjectival stem onto a CVC template, and prefixing the copied material to the adjective.(2) If the first syllable of the adjective is open, than a [b] is inserted into the postvocalic slot of the CVC template. These patterns are demonstrated in (1).(3)
(1) bagdarin `white' bag-bagdarin `very white, white as snow' Sinarin `yellow' Sib-Sinarin `very yellow, golden yellow' kara `black, kab-kara `glossy black, dark' very dark' kcncrin `black' kcb-kcncrin `very black'
There is no effect on primary word stress, which is on the final syllable in all these forms. The reduplicative prefix receives secondary stress. For certain speakers, the postvocalic [b] in the last three forms has assimilated in voicing to the following obstruent, and so is pronounced [p].
Given that emphatic reduplication occurs only with a small number of adjective stems, perhaps only the four presented in (1), it is quite possible that the reduplication works in a fashion other than what we have described here. In particular, for the first form in (1) the use of [g] in the second consonant slot of the reduplicative prefix (hereafter C2) may not represent a copy of a stem consonant at all, but instead may be lexically determined. This must be recognized as a possibility because lexically governed appearances of consonants in the C2 slot arise in other languages. In Tungusic, it is also found in Xibe, where C2 can be either [v] or [k] (Li and Zhong 1986):(4) GOV-GOGXun `extremely spicy', fak-farXun `extremely dark'. The variation is greater in Solon Evenki, allowing [w], [m], or [b]: ab-aja `very good, naw-nana `very slow', nem-nermikkmn `very thin' (Chaoke 1995: 131). In Mongolic, Dagur permits both [b] and [m] in C2 (Zhong 1982; C. Wu 1996); Kalmyk has both [b] and [w] (H. Wu 1996); and what is identified only as a "Mongolian dialect" (H. Wu 1996) evinces [b] and [f]. In Turkic, it is not uncommon to find languages that manifest a two-way possibility, for example Uzbek and Tatar have [m] and [p], whereas Uighur apparently has [v] and [p] (H. Wu 1996; Zhao and Zhu 1985). It is Turkish, though, that seems to provide for greatest variation in C2 slot; [m], [r], [p], and [s] are all used. Regardless of whether emphatic reduplication is believed to be a genetic trait of Altaic or to have entered Oroqen through borrowing, one must recognize the use of [g] for C2 as a unique Oroqen innovation since it occurs nowhere else in Altaic to the best of our knowledge. The simplest way in which to explain its presence in the form bag-bagdarin `very white' is to analyze it as a full copy of the first syllable.
As is observed above, emphatic reduplication is not at all productive in Oroqen, occurring only on the forms provided in (1). Notably, all the adjective terms are colors, and all of them are commonly associated with the coloration of certain domestic animals. For example, kara is typically used for describing dogs and horses, and Sinarin for dogs, though not exclusively. Speakers routinely reject emphatic reduplication with other adjectives, even if they are color terms, as shown in (2).
(2) gilbarin `sky blue' *gil-gilbarin tSen[g.sup.y]En `deep blue' *tfen-tSen[g.sup.y]En ??laarin `red' *??b-??laarin tSuturin `green' *tSub-tSuturin
In simplest terms the semantic outcome of emphatic reduplication is to attribute an unusually deep hue to the color being denoted. However, the emphatic adjectives have developed specific connotations, which makes their use much more restricted than the glosses provided in (1) suggest. Consider just one example: the adjective kabkara `very black', as mentioned earlier, is most often employed to describe the fur of a horse or dog. In this use, it suggests that the animal's coat has a sheen to it. The adjective, however, can also be used to describe clothes that have become blackened from dirt. In this case, the black being described need not be glossy. Kabkara can also be employed to describe the night time, in which case it implies that it is so dark that one cannot see.
The fact that adjectives that utilize emphatic reduplication have taken on specialized semantic applications is perhaps not surprising when one considers that Oroqen possesses two other productive strategies for indicating intensity of colors. These are exemplified in (3).
(3) mani kcncrin `very black' kcncc-li `very black'
The first expression consists of the degree adverb, mani, plus an adjective. The construction is fully productive with all adjectives. The second form involves the suffix -li and the lengthening of the stem vowel immediately before the suffix; it should be noted that many Oroqen adjectives, including kcncrin in (3), end with what appears to be an archaic suffix -rin, which is deleted when -li is added. Like the use of the degree adverb, the use of -li is also fully productive within the semantic domain of color terms, as well as with other adjectives describing physical appearance, such as g??gdaa-li `very tall'. The suffix is used generally to signify intensity, though, somewhat like emphatic reduplication, it does have certain specialized uses. For example, it is used to identify objects in the distance whose color is certain but whose identity is not. Colors with the -li suffix sometimes have conventionalized meanings, particularly in their use as substantives. A few examples are provided in (4).
(4) yalbaali `very white' or `white person' l??gd????li `very brown' or `low-hanging dark clouds' ??laali `very red' or `nonwhite person whose face turned red due to having had too much to drink' ulubaali `very pink' or `a white person whose face turned pink due to having too much to drink'
Since the use of the degree adverb and the suffix -li are fully productive, it is of little surprise that emphatic reduplication is dedicated to a far more restricted semantic task.
Interestingly, the highly restrictive use of emphatic reduplication appears to be the norm in Altaic, though there is a cline in the degree of restrictiveness. The greatest degree of flexibility, in terms of both the number of adjectives that can undergo reduplication and the type of concepts they denote, is found in Turkic. For example, Turkish allows the following:
(5) ince `thin' ip-ince `extremely thin' eksi `sour' ep-eksi `extremely sour' uzgun `worried' up-uzgun `extremely worried' olgun `faded' op-olgun `quite faded' islak `wet' ip-islak `completely wet' acik `open' ap-acik `wide-open' uzun `long' up-uzun `extremely long' temiz `clean' ter-temiz `spotless' beyaz `white' bem-beyaz `snow white'
(from Sebuktekin 1971: 25; Dobrovolsky 1987: 131)
As this noncomprehensive list reveals, emphatic reduplication in Turkish can occur with adjectives indicating taste, mental state, physical properties, colors, and so on.
In Mongolic generally, the process seems to be more constrained in terms of the number of adjectives that participate in the process, though this is difficult to determine in a definitive way since sources rarely indicate how exhaustive of emphatic reduplication their examples are meant to be. The degree of restriction on the semantics of the adjectives involved varies from language to language. In Khalkha, at least, several different sorts of adjective can be employed:
(6) ??la?? `red' ??b-??la?? `thoroughly red' asxa `straight' ab-asxa `extremely straight' xoldu?? `frozen' xob-xoldu?? `frozen through' gat???? `hard' gab-gat???? `rock hard'
(Khalkha from Dao 1983: 35)
In some Mongolic languages, the use of emphatic reduplication appears to be restricted largely to colors and a few physical attributes, as in Minhe Mangghue (Keith Slater and Wang Xianzheng, personal communication), Donbuyugu (Zhaona 1981: 28), and Dongxiang:
(7) xulan `red' xu-xulan `thoroughly red' qara `black' qa-qara `extremely black' smni `new' sm-smni `extremely new'
(Dongxiang from Liu 1981: 57)
In Tungusic, the number of adjectives to which emphatic reduplication applies is invariably small, though the languages vary as to which semantic types may participate. Xibe (Li and Zhong 1986) permits taste concepts, physical properties, and colors; Kile-Nanai is similar but may not include taste concepts (An 1986). Based on published data, it seems that Solon Evenki manifests a wider range than both as it includes ab-aja `very good' and naw-nana `very slow' (Chaoke 1995; Hu and Chaoke 1986).
The comparative data from Tungusic, Mongolic, and Turkic reveal that the emphatic reduplication of Oroqen tends toward greater phonological, semantic, and lexical limitations than many other languages. It is now the case that the process is disappearing from the language altogether, not under a natural process of morphological decay, but due to the attrition of the language as it becomes moribund.
2. The impact of language attrition on Oroqen reduplication
Northwestern Tungusic, the language group to which Oroqen belongs, is rapidly being replaced by Russian in Siberia, and Mandarin Chinese in China. The restrictive application of reduplication has made it highly vulnerable to rapid loss as Oroqen becomes obsolescent. We offer data here from four informants that demonstrate, somewhat anecdotally, that younger speakers no longer have emphatic reduplication in their grammar. Our two older informants A (66 years old) and B (56 years old) both used Oroqen as their primary language until their early 20s. Since that time they have increasingly used Mandarin Chinese. Both women use the reduplicated adjectives in (1), though the younger of the two does not recognize or accept the last of these forms, kcb-kcncrin `very black'. Our two younger informants C (42 years old) and D (26 years old) no longer accept or recognize any of these forms. Though both of these women learned Oroqen in the home as a first language, both have grown up and been educated in a Mandarin-dominated context.
Similar patterns are found with other morphological processes, indicating the same process of loss. Consider one further example: among the nominalizers in Oroqen, -nki is by far the most productive. It occurs with practically any verb to yield a noun. Examples are given in (8).
(8) kadi-re-n `cut-PROG-3S' kadi-nki `sickle/scythe' mine-re-n `cut-PROG-3S' mine-nki `cutting board' dzik-te-n `slice-PROG-3S' dziki-nki `cutting board' g??ldzi-ra-n `lock-PROG-3S' g??ldzi-nki `lock/latch' teye-re-n `sit-PROG-3S' teye-nki `something to sit on'
Both our older informants A and B readily produced many examples using this and other less productive nominalizers. For example, they employed the nominalizer -wun (kadi-wun `knife', tukti-wun `ladder').
What is noteworthy is that quite a few of the less productive nominalizers are no longer in the repertoire of our informants C and D. Our informant C only uses the suffix -nki, even in forms where older speakers do not use it. Although she retains this most productive nominalizer -nki and in fact uses it more generally than older speakers, she alternates between having the velar nasal and leaving it out. A similar pattern held for informant D's speech. Only the most productive nominalizer was employed. However, for her, nasal deletion in the suffix -nki is compulsory, which is shown in the examples in (9) below.
(9) lcxc-rc-n `hang sth. lcxcki `hook' up-PROG-3S' teye-re-n `sit-PROG-3S' teyeki `something to sit on' mu??-le?? `water-LOC' mu??le??ki `water bucket' tukti-re-n `go up/ tuktiki `ladder' climb-PROG-3S' g??ldzi-re-n `lock-PROG-3S' g??ldziki `lock' (noun)
A clear pattern emerges in the data from the four informants A, B, C, and D. The older speakers have a much richer repertoire of derivational morphology than the younger ones, and there is a gradual phonological erosion of the productive suffixal nominalizer -nki that correlates with the age of the speakers, which is illustrated in (10).
(10) A B C D teye-nki teye-nki teye-nki/ teye- `something teye-ki ki to sit on' mu??le-nki mu??le??-nki mu??leo-nki/ mu??le??- `water mu??le??-ki ki bucket tukti-wun tukti-wun tukti-nki/ tukti- `ladder' tukti-ki ki kadi-wun kadi-wun kadi-nki/ kadi- `sickle/ kadi-ki ki scythe'
These data demonstrate, for the younger generation of Oroqen speakers, the wholesale loss of certain limited derivational strategies, such as emphatic reduplication and the replacement of low-productivity derivational morphemes such as -wun by equivalent, yet more productive, strategies. Of course, neither sort of change is unusual in the historical development of language. What is interesting, though, is the rate at which it has occurred, seemingly in the span of 25 years or so.
A complex set of conditions is responsible for such changes in Oroqen, most of which are sociological in nature. Oroqen was surrounded by several languages for centuries without suffering any attrition, which is attributable to the cohesion of the community, its self-imposed rule forbidding intermarriages with non-Tungusic ethnic groups, and its strong adherence to traditional lifestyle and values. However, dramatic social changes in the 1950s and 1960s quickly resulted in Oroqen's alarmingly rapid ceding to the superstratum language Mandarin Chinese. The construction of a railroad, massive Chinese immigration to the area, resettlement of the Oroqens, schooling in Chinese, extensive intermarriages with nearby ethnic groups, abandonment of traditional values and lifestyle, and the local government's lack of interest in maintaining the Oroqen language and cultural traditions led to a situation characterized by an increasing use of Mandarin Chinese in more and more situations, which eventually brought about a proficiency continuum determined by age among those who still speak Oroqen. The diminishing use of Oroqen severely limits the younger generation's exposure to the language, which is the major cause of their imperfect learning of their obsolescing first language.
As the Tungusic languages fall into disuse, crucial data for issues such as the origin of emphatic reduplication will be lost forever. Given recent calls for the greater utilization of morphological comparison in Altaic (see Georg et al. 1998) and Tungusic (see Whaley et al. 1999), the dissipation of seemingly obscure aspects of the morphology of these languages becomes all the more problematic.
3. The origin of reduplication in Oroqen
Having laid out the fundamentals of emphatic reduplication in Oroqen, we now turn to an account of its origins. Two basic scenarios for the genesis of reduplication present themselves. First, emphatic reduplication may have been a property of Proto-Micro-Altaic that was maintained to some degree in each of its branches. The process has decayed most rapidly in Tungusic, such that it now exists only in a highly restricted, idiosyncratic form. Under the second scenario, emphatic reduplication was borrowed into Oroqen (or some ancestor of the language) as well as a number of other Tungusic languages. Either it was borrowed in a restricted form, or it was borrowed in a productive form and has since decayed.
Of course, these options are the typical possibilities put forth for nearly any similarity found among Turkic, Mongolic, and Tungusic languages, and, as is also typical, the interpretation of the data can be plausibly explained under either a genetic or a contact-based understanding of the relationship among these families. However, for this linguistic feature at least, we argue that the weight of the evidence favors a borrowing analysis.
One important consideration in this regard is the specific Tungusic languages in which emphatic reduplication is found: Xibe, Kile-Nanai (known as Hezhe `Hezhen' in China), Solon Evenki, and Oroqen. Each of these languages is found on Chinese soil.(5) As Tsumagari (1997) observes, this is an extremely suspicious geographical distribution for emphatic reduplication within Tungusic. These languages are associated by the fact that they have been in contact with one another and in contact with the same non-Tungusic languages, especially Mandarin Chinese, Khalkha Mongolian, and Dagur (also Mongolic). They do not form anything close to a genetic unity within Tungusic; Xibe falls in the Southern branch, Kile-Nanai falls in the Central branch, and Solon Evenki and Oroqen are in the Northern branch. Languages closely related to Solon Evenki and Oroqen but spoken in Siberia (i.e. numerous Siberian Evenki dialects and Negidal) do not possess the process, nor do Nanai, Ulch, or Orok despite their close affinity to Kile-Nanai. Given the general principle that a genetic trait should not be confined to an areally based set of languages within a family, the most direct explanation for such facts is that emphatic reduplication represents an areal feature of the Chinese Tungusic languages rather than a genetic trait.
A second reason to question whether emphatic reduplication might be a genetic trait of Proto-Tungusic is that the phonotactics of the process are difficult to motivate. Consider the following features of Oroqen morphology and phonology, which are representative. Other than the reduplication under discussion, the language is exclusively suffixing. There is a strong propensity for syllables to be CV, CV:, or CVS (where S is a sonorant). Other than one of the reduplicated forms, bagbagdarin, the sequence [gb] (or more generally a velar obstruent + a bilabial obstruent) does not exist in the language; and other than one of the reduplicated forms, fibfinarm, the sequence [bf] (or [pf]) does not exist. Of course, if emphatic reduplication were an archaic Altaic feature, it need only be motivated in phonotactic terms for Proto-Altaic (though even here, it is not obvious that the phonotactics of Proto-Altaic are consistent with the facts of emphatic reduplication). One could see these anomalous Oroqen forms as remnants from a much earlier era. However, they would be remarkable forms indeed, having resisted change for such a lengthy period of time.
It should be noted at this point that Dobrovolsky (1987) has demonstrated the consistency of emphatic reduplication with Turkish (and by extension Turkic) phonotactics. He observes that Turkish monosyllables are always CVC (which is decidedly not true in Tungusic) and that compounds in Turkish tend to get primary stress assigned from their left boundary (also not true in Tungusic). If one then assumes that emphatic reduplication is an instance of compounding (or first arose as compounding), both the shape of the reduplicative template and the stress of reduplicated forms (see  above) are accounted for. In terms of phonotactics, then, it is reasonable to take reduplication as consistent with Turkish, but not with Tungusic.
Finally, the lack of diversity in the C2 slot in the reduplicated syllable within the Tungusic languages as compared to Turkic, at least, leads us to doubt that emphatic reduplication is a genetic feature of the Tungusic family. It seems generally agreed upon that the choice of consonant found in C2 must be lexically determined in those languages that evince more than one possibility. Since irregularities in a morphological system are generally taken to represent historical residue rather than innovations, we should expect greater irregularity to more closely approximate the original system. As noted in section 1, Tungusic (and Mongolic) languages manifest fewer options for C2 than one finds, say, in Turkish. This alone is certainly not strong evidence that reduplication is borrowed; it is possible that Tungusic languages have simply been less conservative in this aspect of their morphology than Turkish. Nevertheless, it is a striking fact that within Tungusic (or Mongolic) there is little hint of the complexity found for C2 in Turkic.
There are, in summary, three sets of facts that militate against taking the process of reduplication as an inherited micro-Altaic trait: the distribution of Tungusic languages in which reduplication is found, Tungusic phonotactics, and the restricted number of C2 consonants in Tungusic. We, therefore, see borrowing as a more likely explanation. A borrowing account, however, raises a host of issues. Assuming that the process was borrowed into Tungusic, (1) was its ultimate source Turkic, Mongolic, or some Turco-Mongolic unity?; (2) what was the immediate source of the borrowing?; (3) when did the borrowing occur?; (4) what sort of conditions were necessary to allow the borrowing of a process of reduplication?; (5) was it actually the process of reduplication or frozen forms that were borrowed? Obviously, many of these questions are interrelated, and the answer to some of them rests in how much credence one puts into certain versions of the Altaic hypothesis, for example whether one accepts the possibility of a Turco-Mongolic subbranch that excludes Tungusic (an option that runs counter to several recent proposals about a Mongolo-Tungusic unity -- e.g. Janhunen 1996; Ruhlen 1986). Even so, we find that provisional answers are possible for all of these questions except for number 3.
Regarding the ultimate source of reduplication, we find it most likely to be Turkic. It has already been noted above that important features of the reduplication are arguably consistent with Turkish phonotactics and stress facts. Mongolic phonotactics, it must be noted, also are consistent with the reduplication facts: Poppe (1955) claims syllable codas could be filled with a variety of consonants, including b, d, g, s, and m, that is, the same consonants (or at least the relevant manner and place articulations) that probably participated in the original process. Thus, the possibility of a Turco-Mongolic origin cannot be dismissed out of hand. However, like the Tungusic languages, Mongolic languages tend to have greater restrictions on C2 than do Turkic languages, manifesting at most [b] and some other labial (as in the case of Dagur and Kalmyk). Therefore, we slightly favor a Turkic origin.(6)
Although the evidence for the ultimate origin of reduplication is hardly definitive, the more immediate source for Tungusic appears more obvious. The borrowing almost certainly would have come directly from a Mongolic language or languages, and a strong case can be made that it is Dagur (with co-influence from Khalkha Mongolian also possible). Dagur is the only Mongolic language that has been in constant and intimate contact with each of the Tungusic languages that manifest emphatic reduplication. Xibe speakers (found in Xinjiang Province, China) descend from a Manchu population moved to the northwest of China by Manchu emperors in the eighteenth century. Their migration occurred in tandem with Solon Evenki (NB: this dialect of Solon is no longer spoken) and Dagur speakers.(7) The Solon Evenki, according to Janhunen (1996), originally represented a "satellite group" to the Dagur in the Amur basin. When the Dagur were forced to move south of the Amur at the end of the seventeenth century, the Solon moved with them, such that even today, wherever one finds a Solon community in China, there are also Dagurs.
Dagur speakers have also been in contact with Oroqen speakers for centuries in a symbiotic relationship. The traditionally nomadic Oroqen relied on the sedentary Dagur for certain agricultural goods, while in exchange supplying the Dagur with pelts and meat. The commercial relationship, while mutually beneficial, established Dagur as the dominant language, and it became the norm for Oroqen speakers to learn to speak Dagur.
As a result, the Oroqen lexicon has taken on many Dagur words, and Oroqen grammar has borrowed from Dagur, particularly in the area of derivational morphology. Thus, the borrowing of reduplication can be seen as part of a more general Dagur influence on Oroqen grammatical structure.
As a point of comparison, consider another example of morphological borrowing. Dagur has a plural marker -nur used for kinship terms and human nouns, illustrated in (11) below.
(11) ekee `sister' ekeenur `sister' gutf `comrade' gutfnur `comrades' (from Zhong 1982: 33)
Oroqen uses the phonologically similar suffix -n Vr, but iq is only used for kinship terms.(8) Among Tungusic languages, the suffix is only found in Oroqen and some dialects of Solon Evenki (also in contact with Dagur). This fact suggests that the suffix is a borrowing. Some examples from Oroqen are given in (12) below.
(12) naarf?? `uncle (on naatf??n??r `uncles (on mother's side)' side)' mother's amaakaa `uncle (on amaakaanar `uncles (on father's side)' side)' father's jeeje(9) `grandpa jeejener `grandpas and those of their generation'
In Dagur, the marker -nur simply signals plurality. The borrowed marker -nVr in Oroqen, however, is more restricted in two ways. It is used solely with kinship terms, and it has taken on other connotations beyond plurality. On the one hand, it can indicate an exhaustive set. The word naatf??n??r in (12) thus connotes all the uncles on one's mother's side together. The suffix can also indicate age association, as in the final form in (12), jeejener, which designates grandfathers and everybody else belonging to their generation.
The instance of the -nur borrowing is reminiscent of the facts surrounding the borrowing of reduplication. In both cases, a morphological strategy is borrowed, but imperfectly, such that in Oroqen it can only be applied to a subset of those nouns to which it can be applied in Dagur. Furthermore, perhaps as part of its limited distribution in Oroqen, it takes on connotations that it did not have in Dagur. This pattern, which we have only discussed with respect to two morphological borrowings, appears to hold true for all other cases of probable borrowing that we have identified to date.
Though the influence of Dagur on Oroqen is undeniable, it is important to point out that the actual pattern of borrowing between the two languages may be more complex than we have suggested so far. In the case of emphatic reduplication, it is also possible that the process entered into Oroqen via another Tungusic language, which in turn had borrowed it from Dagur (or another Mongolian language). If so, the candidates for the source language would be either Khamnigan Evenki or Solon Evenki. Little is known of the morphology of the former (Janhunen 1991 is the best overview but makes no reference to emphatic reduplication). The Oroqen had much more extensive contact with the Solon Evenki, who themselves were typically bilingual in Dagur. Because of the nature of multilingualism, which has pervaded this region for centuries, it may not be the case that the emphatic reduplication entered into any one of the Tungusic languages first. If Tungusic speakers were fluent in one or more Mongolian languages and one or more Tungusic languages, it might be most accurate to say that emphatic reduplication entered the speech repertoire of multilingual individuals where it was used in each of the languages spoken.
Regardless of whether the borrowing of emphatic reduplication was directly and solely from Dagur or mitigated by multilingualism in other Tungusic languages, the sort of structural influence that Dagur has had on Oroqen requires an extended period of relatively intimate contact. It is useful here to review the history of these groups, which points to just this type of interaction.
Both the Dagurs and the Oroqens are believed to originate from the region north of the Amur River, that is, in present-day Russia. They both crossed the Amur several hundred years ago (most likely in the 1600s) and spread over the Greater and Lesser Hinggan Mountains in Manchuria. Janhunen (1997) suggests that migrations of small Dagur populations occurred in tandem with the Oroqen and Solon migrations. Regardless of whether this is true or not, it is widely accepted that all these groups have coexisted harmoniously in Inner Mongolia and the northeast of China for several centuries.
Trading among them was ubiquitous, usually involving the exchange of fur, game products, and animal-hide handicrafts from the Oroqens and Solon Evenkis and animal husbandry or agricultural products from the Dagurs. The frequency of the trading led to multilingualism and to some mixing of the populations through intermarriages. Since the borrowing being discussed here is structural rather than lexical in nature, massive bilingualism on the part of the Oroqen speakers in the lending language (s) persisting over a long period of time is a crucial prerequisite.
Another social factor frequently invoked in accounting for borrowings across languages, especially languages belonging to different families, is a prestigious status of the source language. The Dagur people have historically held the necessary position of prestige: according to Janhunen (1997: 128), "since Qing times (1644-1911), the Dagurs have been known as an ethnic group interested in acquiring higher learning through dominant languages such as Manchu and Chinese." This may help explain why the Qing emperor entrusted the Dagurs with the control of the diaspora army sent to guard Chinese borders against the potential invaders in bordering areas.
Although some linguists would consider structural similarity and functional congruence as important factors in cases of grammatical borrowing (e.g. Weinreich 1953), we find that the Oroqen borrowing of reduplication provides a counter-example. Dagur has both suffixes and prefixes, whereas Oroqen is purely suffixal, with the exception of emphatic reduplication. At least in terms of morphological structure, Oroqen and Dagur are not structurally similar. In this respect, we agree with Thomason and Kaufman (1988: 35), who contend that "it is the sociolinguistic history of the speakers, and not the structure of their language, that is the primary determinant of the linguistic outcome of language contact. Purely linguistic considerations are relevant but strictly secondary overall." The case of Oroqen borrowing provides evidence in support of Thomason and Kaufman's contention.
There is one final issue in the borrowing of emphatic reduplication that needs to be raised. It is clear that it is the PROCESS of reduplication that has been borrowed, rather than frozen reduplication forms, since Oroqen applies the reduplication to native color terms rather than borrowing these terms from Dagur (cf. bag-bagdarm Oroqen vs. tfiva??n Dagur). Because the process is borrowed, and not lexical items that contain frozen instances of reduplication, it is surprising that the reduplication is so highly restricted in Oroqen. We propose that the restrictive nature of this process arose from two factors: first, the presence of other highly productive strategies to indicate intensity relegated the reduplication to a highly specialized semantic domain and inhibited its spread (see Whaley and Li  for an overview of qualitative suffixes in Oroqen); second, the borrowing of the reduplication most likely occurred in the context of trade. Since domesticated animals were one common trade item, this would account for the semantic realm in which the reduplication entered into Oroqen. It is after entry into Oroqen that emphatically reduplicated adjectives were applied idiosyncratically in other contexts.
The findings reported herein shed light on our understanding of the complexity of language change in terms of both internal and external factors leading to the development and loss of morphological processes. The interaction between internal phonological changes and external language contact/multilingualism-induced changes coupled with factors of language obsolescence provide a special set of insights into the intricate complexity of language change in general, and morphological change in particular.
This case study also illumines some general principles in identifying areally based similarities between languages in instances where a genetic relationship is also suspected. Recall two of the basic justifications that were offered for believing the presence of emphatic reduplication in Oroqen, and Tungusic languages more generally, to be due to borrowing rather than to be an archaic genetic feature. First, when genetic traits are not found throughout a family (or phylum, or branch), we expect the languages that share (or lack) the trait either to characterize a genetic subunit or to occur randomly. In the former case, the shared feature would simply be an innovation of the proto-language that unites the genetic subgroup. In the latter instance, the shared feature would represent an independent convergence. What we do not expect, if a shared feature is genetic, is for it to be found in an areally defined cluster of languages within the family, but this is precisely the situation in Tungusic, where emphatic reduplication seems to be restricted to Tungusic languages spoken in China.
Second, innovations and retentions in a language are more sensitive to the structural properties of that language than borrowings are. Over time the structure of languages can obviously change in dramatic ways due to a long series of innovations, but each innovation (or each step in one innovation) will adhere in large measure to the structural characteristics of the language in which it occurs. Oroqen emphatic reduplication is suspicious in this regard since the forms it produces depart from the phonotactics of the language as well as the proto-language from which it comes. If emphatic reduplication were used with most adjectives in Oroqen, it would be easier to take the phonotactic pattern as an irregularity that has resisted conforming to the structure of language. Given the peripheral nature of the process in Oroqen, however, the maintenance of the structural irregularity from a distant Altaic source is harder to accept.
California State University, Chico Dartmouth College
Received 12 July 1999 Revised version received 14 February 2000
(*) Research for this paper was sponsored by NSF grant SBR-9710091. We would like to acknowledge the helpful comments of the audiences present at the 22nd Annual Penn Linguistics Colloquium and the 34th Regional Meeting of the Chicago Linguistic Society, where we presented shorter versions of the material in this paper (published as Li and Whaley 1998a; Li and Whaley 1998b). We are also grateful for the excellent suggestions for improvement made by two anonymous reviewers. We would especially like to thank the many Oroqen speakers who have been teaching us their language and who supplied the Oroqen data in this article. Correspondence address: Lindsay J. Whaley, Program in Linguistics and Cognitive Science, 6086 Reed, Dartmouth College, Hanover, NH 03755, USA. E-mail: email@example.com.
(1.) Notably, the process is not found in Japanese or Korean, so it is much more difficult to make a case for it being a marker of Macro-Altaic.
(2.) We use the term "prefixing" here in a nontechnical sense. Dobrovolsky (1987) has effectively argued that emphatic reduplication in Turkish is actually a case of compounding rather than prefixation. His arguments are based on facts about word stress and phonotactics in Turkish that do not hold true for Oroqen. Nevertheless, we leave it as an open question as to whether the process is one of prefixing or compounding.
(3.) For some speakers, the fourth form in (1) is pronounced as kcnnorin. However, even though this produces a closed first syllable, the reduplicated form remains the same as in (1).
(4.) Rather than [k], H. Wu (1996) treats the consonant as [q]. Moreover, some of the instances of [v] found in Li and Zhong (1986), Wu treats as [b].
(5.) Emphatic reduplication is not found in every Chinese Tungusic language. It is apparently not found in classical Manchu or contemporary spoken Manchu. It is unclear whether it occurs in Khamnigan Evenki or Yakut Evenki.
(6.) According to Menges (1968) emphatic reduplication is also found in a number of languages outside of Altaic. He names Suomi (i.e. Finnish), Armenian, and "Uralic" languages, though we have been unable to find data to support this claim. If true, this complicates a genetically based analysis of reduplication even further since one can no longer claim the morphological process to be characteristic of Altaic uniquely. Of course, it is possible a language phylum larger than Altaic was involved -- Nostratic or Eurasiatic would be obvious candidates -- but this creates the additional problem of accounting for the absence of reduplication in most branches of these macro-families as well as all the problems already mentioned for Tungusic.
(7.) We should note that this migration of the Xibe also put them in contact with speakers of Turkic languages. Xibe is thus the only Tungusic language that possesses emphatic reduplication that has been in sustained contact with Turkic over the last several hundred years.
(8.) Most Oroqen suffixes are subject to vowel harmony with the preceding root vowel. Vowels subject to harmony are orthographically represented in this paper by capital V.
(9.) The word jeeje is itself a borrowing from Mandarin Chinese.
An, Jun (1986). Hezhe Yu Jianzhi [A Grammatical Sketch of Hezhen]. Beijing: Minzu Chubanshe [Nationalities Press].
Chaoke (1995). Ewenke Yu Yanjiu [A Study of Ewenki]. Beijing: Minzu Chubanshe [Nationalities Press].
Clauson, Gerard (1956). The case against the Altaic theory. Central Asiatic Journal 2, 181-187.
Dao, Bu (1983). Mongu Yu Jianzhi [A Grammatical Sketch of Khalkha Mongolian]. Beijing: Minzu Chubanshe [Nationalities Press].
Dobrovolsky, Michael (1987). Why CVC in Turkish reduplication? In The Nordic Languages and Modern Linguistics 6, 131-146. Helsinki: Helsinki University Press.
Doerfer, Gerhard (1985). Mongolo-Tungusica. Wiesbaden: Harrassowitz.
Georg, Stefan; Michalove, Peter A.; Manaster Ramer, Alexis; and Sidwell, Paul J. (1998). Telling general linguistics about Altaic. Journal of Linguistics 35, 65-98.
Hu, Zengyi (1986). Elunchun Yu Jianzhi [A Grammatical Sketch of Oroqen]. Beijing: Minzu Chubanshe [Nationalities Press].
--; and Chaoke (1986). Ewenke Yu Jianzhi [A Grammatical Sketch of Ewenki]. Beijing: Minzu Chubanshe [Nationalities Press].
Janhunen, Juha (1991). Material on Manchurian Khamnigan Evenki. Castrenienumin toimitteita 40. Helsinki: Finno-Ugrian Society.
--(1996). Manchuria: An Ethnic History. Helsinki: Finno-Ugrian Society.
--(1997). The languages of Manchuria in today's China. Northern Minority Languages: Problems of Survival, Senri Ethnological Studies 44, 123-146.
Li, Fengxiang; and Whaley, Lindsay J. (1998a). The life and death of derivational morphology: reduplication in Oroqen. U Penn Working Papers in Linguistics 5.
--; and Whaley, Lindsay J. (1998b). The rise and fall of derivational morphology: reduplication in Oroqen. Paper presented at the 34th Regional Meeting of the Chicago Linguistic Society.
Li, Shulan; and Zhong Qian (1986). Xibe Yu Jianzhi [A Grammatical Sketch of Xibe]. Beijing: Minzu Chubanshe [Nationalities Press].
Liu, Zhaoxiong (1981). Dongxiang Yu Jianzhi [A Grammatical Sketch of Dongxiang (Santa)]. Beijing: Minzu Chubanshe [Nationalities Press].
Menges, Karl H. (1968). The Turkic Languages and Peoples. Wiesbaden: Harrassowitz.
Moravcsik, Edith A. (1978). Language contact. In Universals of Human Language, vol. 1, J. H. Greenberg (ed.), 93-122. Stanford: Stanford University Press.
Ozdemir, E. (1986). Emphatic reduplication in Turkish. Paper presented at the Third Conference on Turkish Linguistics, Tilburg University, The Netherlands.
Poppe, Nicholas (1955). Introduction to Mongolian Comparative Studies. Helsinki: Finno-Ugrian Society.
Ruhlen, Merrit (1986). A Guide to the World's Languages. Stanford: Stanford University Press.
Sebuktekin, Hikmet I. (1971). Turkish-English Contrastive Analysis: Turkish Morphology and Corresponding English Structures. The Hague: Mouton.
Swift, L. B. (1962). A Reference Grammar of Modern Turkish. Uralic and Altaic Series 19. Bloomington: Indiana University Press.
Thomason, Sarah Grey; and Kaufman, Terrence (1988). Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press.
Tsumagari, Toshiro (1997). Linguistic diversity and national borders of Tungusic. Northern Minority Languages: Problems of Survival, Senri Ethnological Studies 44, 175-186.
Weinreich, Uriel (1953). Languages in Contact. The Hague: Mouton.
Whaley, Lindsay J.; Grenoble, Lenore A.; and Li, Fengxiang (1999). Revisiting Tungusic classification from the bottom up: a comparison of Evenki and Oroqen. Language 75, 60-95.
--; and Li, Fengxiang (1998). The diminutive-intensive suffix -kan in Oroqen. Studies in Language 22, 447-471.
Wu, Chao Lu (1996). Dagur. Languages of the World. Munich and Newcastle: LINCOM.
Wu, Hegejilitu (1996). A common method by which all Altaic languages indicate emphasis of adjectives. Minzu Yuwen 50-56.
Zhao, Xiangru; and Zhu, Zhining (1985). Weiwuer Yu Jianzhi [A Grammatical Sketch of Uyghur]. Beijing: Minzu Chubanshe [Nationalities Press].
Zhaona, Situ (1981). Dongbuyugu Jianzhi [A Grammatical Sketch of Dongbuyugu]. Beijing: Minzu Chubanshe [Nationalities Press].
Zhong, Suchun (1982). Dagur Yu Jianzhi [A Grammatical Sketch of Dagur]. Beijing: Minzu Chubanshe [Nationalities Press].
|Printer friendly Cite/link Email Feedback|
|Author:||LI, FENGXIANG; WHALEY, LINDSAY J.|
|Publication:||Linguistics: an interdisciplinary journal of the language sciences|
|Date:||Mar 1, 2000|
|Previous Article:||Acceptability and preference in the interpretation of anaphors(*).|
|Next Article:||The notion of current relevance in the Thai perfect(*).|