+ All Categories
Home > Documents > Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

Date post: 04-Jun-2018
Category:
Upload: allan-bomhard
View: 221 times
Download: 0 times
Share this document with a friend

of 12

Transcript
  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    1/12

    nthropologicalinguisticsVOLUME 44 NUMBER 1

    Offprint

    The Nostratic Accusative in -mA: n Altaic Perspective

    PETER A. MICHALOVE

    Department ofAnthropology American Indian Studies Research InstituteIndiana University Bloomington, Indiana

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    2/12

    The Nostratic Accusative in -mA: n Altaic Perspective

    PETER A. MICHALOVEUniversity of Illinois, Urbana-Champaign

    Abstract. The suffix in mA has been posited as an accusative suffix underlying forms common to Indo-European, Uralic, Altaic, and Dravidian. V M.Illic-Svityc and others have considered this as evidence to support the Nostratichypothesis, which proposes a genetic relationship among these and some otherlanguage families of Eurasia. The Altaic component of this comparison appears,on the basis of lllic-Svityc s work, to be the weakest part of the comparison.However, a deeper examination of the Altaic forms shows them to be highlyinformative; they ultimately support the proposal of a common origin for thisformative in all of the language families considered. The usages indicate thatthe original function of the suffix was not as a direct object marker, but as anindicator of specificity.

    1. A common accusative suffix and specifier of time and space: thedata. The Nostratic theory, which proposes a genetic affiliation among manyof the languages of Eurasia, is highly controversial, largely because all of theversions of it put forth so far contain serious flaws as well as promising ideas.Rather than simply accepting any of the versions of Nostratic offered as completely correct, or dismissing the entire enterprise because of some, even significant flaws, there is a third option available to us. It would be more productiveto build on the stronger elements in the theory, to recognize and reject theelements that are simply incorrect, and most importantly, to work critically torefme and try to remedy those elements that contain both errors and promisingmaterial in order to see whether the mass of data produces something coherent.The present article is an attempt to begin this slow, detailed evaluation byexamining an important morphological isogloss in the Nostratic theory.Illic-Svityc (1976:4S-51) proposes a Nostratic su ix -mA (where the symbolA indicates an unrounded low vowel), indicating a marked direct object, represented by Indo-European -m, Uralic -m, Dravidian -m, and Altaic -ba/-bii.Of course, the comparison of these forms was not new. There had been a numberof previous proposals, mostly involving binary comparisons among various cornbinations of the forms, and Illic-Svityc (1976:50-51) gives a good summary of theliterature to the time of his death in 1966. The Indo-European-Uralic comparison had been the most frequently discussed, and it provides a particularlygood match. The comparison with the Dravidian form also appears to be wellfounded, but, since I have practically no background in Dravidian, the presentarticle will not pursue that comparison.

    The remaining family, Altaic, appears from Illic-Svityc s treatment to85

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    3/12

    86 ANTHRoPOLOGICAL LINGUISTICS N0.1present the greatest problems. AB we will see, however, a closer examination ofthe Altaic material not only supports an equation with the other forms, butturns out to be the most informative for understanding the origin of the suffix.

    Of course, the genetic relatedness of an Altaic family, consisting of Turkic ,Mongolic, Tungusic, Korean, and Japanese, is at least as controversial as Nostratic itself. However, a genetic relationship among the eastern Altaic languages(Japanese, Korean, and Tungusic) is widely accepted; the controversy, ironically, consists in trying to connect the traditional triad ofTurkic, Mongolic, andTungusic. AB we will see, the best evidence for the relatedness of the construction discussed in this article is found among the eastern languages (Japaneseand Tungusic), while the possible Turkic and Mongolic connections are moreproblematic; and this seems to reflect the cWTent state of Altaic studies as awhole. For an exposition of the stronger and better-accepted relation among thethree eastern Altaic languages, see the recent work ofVovin (2001a, 2001b).

    In lllic-Svityc's work, the Altaic distribution of this form is limited toTungusic, with an unconvincing Mongolic possibility (1976:49). Since then, however, Miller (1971) and Itabashi (1988), building on the ideas of Murayama(1957), have identified the Tungusic suffix in -bA (with vowel harmony alter-nants -ba -bii., and with the consonantlenited t w in intervocalic position insome of the Tungusic languages) with the Old Japanese case suffix wo

    Itabashi (1988) points out three functions of the suffix in Old Japanese:definite direct object marker, exclamatory marker, and prosecutive case suffiX.Whiie the use of the accusative suffix to indicate a (definite) direct object isobvious, the exclamatory usage is less significant because it may be onomatopoetic. However, the prosecutive function is of particular interest and meritsfurther discussion here. It expresses specificity of time or space, or a point intime or space through which something passes, as in (1) and (2).1) nagaki yo wo (Old Japanese)long night PROSECUTIVE

    'duringthelongnights' (Itabashi 1988:200)(2) nagi n kaFa e wo (Old Japanese)

    Nagi GENITIVE river side PROSECUTIVE'along the Nagi riverside' (Itabashi 1988:201)For Tungusic, Itabashi illustrates the same functions. The use of the accusa

    tive to specify time or space (what Benzing calls the accusative of zeitliche undriiumliche Erstreckung [1955:1097]), as in (3) and (4), parallels precisely that ofOld Japanese.(3) bu ala Ci l la w dolboni wa (Ewenki)we wait DURATIVE INGRESSIVE PRETERITE 1PL night-ACC

    'We started to wait all night.' (Konstantinova 1964:49)

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    4/12

    2002 PETERA. MICHALOVE 87(4) mi duente we pulsi he mbi (Nanai)

    I forest-ACC walk-PRETERITE-1SGI walked through the forest . (Benzing 1955:1028)This construction, of course, corresponds exactly to the Indo-European accu

    sative of extent of time or space, as in (5)-(7).(5) makrits hodous (Greek)long-ACC-PL road-ACC-PL

    over long roads(6) ga tva tri n ahoratran (Sanskrit)go-GERUND three-ACC complete.day-ACC-PL

    having traveled three complete days (Whitney 1973:92).(7) man kan lJUR.SAGTeQ8ina n sara pa un (Hittite)

    CONDITIONAL-PARTICLE mountain-Tehsina-ACC up go-1SG-PRETERITEI would have gone up the mountain Tehsina. (Friedrich 1974:120)In Uralic, the use of the accusative to indicate not only a definite direct

    object, but also specificity of time and space, is primarily seen only in BaltoFinnic. Yet this usage is not a borrowing from Germanic, but clearly an indigenous one, since the expression of time and space exactly parallels the complexexpression of the direct object in Balto-Finnic. Karlsson (1999:105-6) cites Finnish usages such as those in (8) and (9).(8) Ole n ollut Suome ssa viiko n (Finnish)be-1SG PERFECT Finland-INESSIVE week-ACC

    I have been in Finland for a week. (Karlsson 1999:105)(9) Viren juokse e kilometri n (Finnish)

    Viren run-3SG-PRES kilometer-AceViren will run a kilometer. (Karlsson 1999: 106)

    Here the extent of time and space is expressed by the accusative in n (< m ,formally now the same as the genitive in n. These constructions have negativecounterparts, as in (10) and (11), with the negative length of time or space in thepartitive, mirroring the expression ofnegative direct objects.(10) E n ole ollut Suome ssa viiko a (Finnish)NEG-1SG be PERFECT Finland-INESSIVE week-PARTITIVE

    I have not been in Finland for a week. (Karlsson 1999: 105)(11) Viren e i juokse kilometri ii (Finnish)Viren NEG-3SG run kilometer-PARTITIVE

    Viren will not run a kilometer. (Karlsson 1999: 106)

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    5/12

    88 ANTHRoPOLOGICAL LINGUISTICS 44NO.lKarlsson also illustrates the imperative use of the uninflected (nominative) formto indicate definite amounts of time and space, as in (12) and (13), again paral-leling the expression of the definite object of an imperative verb.(12) ~ Suome-ssa IJiikko-0be IMP Finland-INESSIVE week NOM

    Stay a week in Finland. (Karlsson 1999:105)(13) Juokse-0 kilometri-0run-IMP kilometer-NOM

    Run a kilometer. (Karlsson 1999:106)

    (Finnish)

    (Finnish)

    Thus, the use of the accusative of time or space in Finnish follows the sameintricate pattern as that of a direct object, indicating that this is an indigenoususage, even though it is confined to Balto-Finnic. We will discuss the absence ofthis usage elsewhere in Uralic in section 4.

    Returning to Altaic, Itabashi (1988) points not only to the Japanese-Tungusic parallels, but also to possibly related fonns in Turkic and Mongolic.These are evidenced most clearly in derivatives from the inteiTOgative forms,such as Old Turkic nii-mii thing, something from nii what? , kim kim miiwhoever it is from kim who? ; and Written Mongolian ken ba whoever fromken who, which , yambar ba whatever from yambar what?, what kind?

    This combination of functions suggests that the use of the suffix to indicatea definite direct object is secondary, and the earlier Altaic accusative in i-gi,which is preserved in some form in all of the Altaic languages except Japanese,confirms this suspicion. Itabashi (1988:211) proposes that the exclamatoryusage must be the original function, and he appears to claim that the variety ofusages cannot derive from an earlier function that is not represented in theattested languages.

    But we can indeed propose such an original function. In phonological reconstruction, we can compare forms like Greek thermos, atin formus, and Gothicwarm-jan, and posit an initial consonant in gwh_ although none of the attestedforms preserves all of the voiced, velar, labial, and aspirate features that wereconstruct for the original form. For the Altaic suffix in -bA, the combinationof definite direct object, exclamatory, and prosecutive functions as a group suggests an original function as a marker of specification.

    In this case, the Turkic and Mongolic usages, i they are related to theTungusic and Japanese fonns, imply a semantic development in those lan-guages from specificity to its apparent opposite, generality. This development,although quite possible, makes these fonns less secure than their Tungusic andJapanese counterparts.

    The function of specificity seems to be the one from which it is easiest toderive all of the attested usages, and it is certainly easier to imagine the accusative of specificity of time or space from an original marker of specificity than

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    6/12

    ......

    2002 PETERA MICHALOVE 89from an exclamatory usage. Thus, we do not need to suppose, as ltabashi does,that the prosecutive function and the exclamatory function are two different

    functions in origin, since each tends to exclude the other (1988:201). In fact,their complementary distribution (in Japanese) is strong evidence in favor of acommon origin.

    Interestingly, Vertes (1960), considering only the Uralic languages, in whichthe accusative in m has a more restricted scope of usage than the Altaic form,concluded, primarily on morphological evidence, that the original Uralic function of the suffix was not to mark a definite direct object. (See Wickman [1955]for a good summary of the distribution and usage of this formative.) Vertes seesthe origin of the Uralic suffix in a deictic or determinative element of the sameorigin as the first-person pronoun, or rather possessive suffix, since the possessive suffixes frequently serve to indicate specificity. Analogously, she sees theaccusative suffix -t found in Hungarian and some Khanty dialects as well as inthe Balto-Finnic pronominal accusative, as a deictic element of the same originas the second-person possessive suffix. While this explanation of the origin ofthe Uralic accusative formative is highly speculative, it is not inconsistent withour thesis for Indo-European and Altaic, which, as we will see below, share afirst person pronominal marker with Uralic. But what is significant for our purposes is that the Uralic data alone indicate that the Uralic accusative suffixoriginated, not as a marker of the definite direct object, but as some sort ofdeictic or determinative element. This is in line with our view of the marker as aspecifier in origin.2 Bomhard and Kerns (1994:185-86) reach a similar conclusion, reconstructing -m (without the following low vowel) as a Nostraticaccusative of specification.

    2. Denasalization of *mAin Altaic. While Itabashi has clearly shown thatthe Altaic suffix in bA is not isolated within Tungusic, the comparison of thisform with Uralic and Indo-European (and possibly Dravidian and Afroasiatic)leads us directly to another apparent problem in Illic-Svityc's formulation: thephonological aspect. The development m > b or even vice versa, is an unremarkable phenomenon, and one that would not require a great deal of comment in itself. But i we suppose, as appears to be the case, that m is theoriginal phoneme at the Nostratic level, and that denasalization occurred atsome point in the development of Altaic, then the common occurrence of initialm in attested Altaic forms means that an inherited m was denasalized only ina particular environment, but not in others.

    The problem is identical to that of the first-person pronoun, with Uralicmi and Indo-European oblique me-, but Altaic bi. Vovin (1998) addresses

    this problem and correctly shows that, in the well-known suppletive pattern ofthe Turkic, Mongolic, and Tungusic first-person pronouns, the form in biis original. The oblique forms in mi- have only secondarily developed aninitial nasal by assimilation to a following nasal in the form. To remain within

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    7/12

    9 ANTHRoPOLOGICAL LINGUISTICS 44NO. lTungusic, for example, this paradigm is illustrated by Ewenki, which hasnominative bi, but accusative mine, dative mindu and so on.

    This state of affairs leads Vovin to conclude that the [Proto-Altaic] personalpronouns are unrelated to the Indo-European or Uralic personal pronouns(1998:269). In the same passage, Vovin also rejects an equation of the Altaicaccusative discussed in this article with the Indo-European and Uralic forms

    on the same basis. Indeed, to equate the Altaic forms for the first-personpersonal pronoun and for the accusative suffix with their Indo-European andUralic counterparts, we would have to establish a specific phonological environment in which Altaic ~ n s l i z e d inherited m-.

    Illic-Svityc (1976:50) proposes that initial m- was denasalized in monosyllabic forms in Altaic; the accusative or specifying formative was undoubtedlya free form originally, and so represented an initial b-. This is a promising idea,although one that does not completely reach the mark. Altaic still contains alarge number of monosyllabic forms with initial m-, so we must look for a morespecific environment in which the denasalization occurred. I propose, on thebasis of the data below, that Altaic denasalized inherited initial m- to b- inmonosyllabic forms with a short vowel, but that initial m- remained in monosyllables with a long vowel or diphthong. A further constraint is that the denasalization either failed to occur or was reversed in forms with a subsequentnasal.

    The positive evidence for this development is scanty. We have the firstperson personal pronoun and the accusative suffix discussed above. Illic-Svityc(1976:70) proposes one other example, the demonstrative pronoun reflected inTurkic as bu, as, for example, in Old Turkic bu 'this (one)'. He adds an incoiTectreference to Mongolic, and no attestation at all in the other branches of Altaic.For Uralic, lllic-Svityc compares it to mu mo 'other' (attested only in FinnoUgric, e.g., Finnish muu 'something else') and a Kartvelian demonstrative pronoun, with possible correspondences in Afroasiatic and Indo-European. n thebasis of the Uralic form, Illic-Svityc posits a Nostratic long vowel. He musttherefore assume an unexplained vowel shortening in the Turkic forms, sincewe have short u in Yakut and Turkmen bu. Because of the phonological andsemantic problems, as well as the lack of good Altaic attestation outside ofTurkic (and the absence of Uralic attestation in Samoyed), this comparisonseems extremely unreliable for our purposes.

    But there is significant negative evidence in Altaic for the development.That is, if initial m- developed to Altaic b- in monosyllabic forms with shortvowels and no subsequent nasal, then we should not fmd initial m- in thisenvironment in Altaic. Since Altaic vocalic length is reflected only in Turkic andTungusic, and since initial m- developed to b- in Proto-Turkic in all cases(leaving no cases of initial m-in native Turkic forms), the Tungusic languagesare our best source of data.

    Cincius (1975) cites a large number of monosyllabic Tungusic forms with

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    8/12

    2002 PETER A. MICHALOVE 91initial m-. Disregarding the interjection ma., forms attested in only oneTungusic language, and of course forms identified as loanwords, we find that theremaining 1\mgusic monosyllabic forms with initial m- can be classed into thefollowing three categories. (The numeral in parentheses after each fonn is areference to the page number in Cincius [1975].)

    The first category consists ofform.s with a long vowel or diphthong. Since itis precisely vocalic length that we propose blocked the denasalization of thesefonns, it is not surprising that this is the largest group.

    e 'to set up a fishing net' (p. 533). This form goes back to a diphthong, asindicated by the Nanai fonns mia.- 'set up a fishing net', mfa.c f- 'catch fishwith a net'.me- 'to awaken' (p. 534). This is also the reflex of a back diphthong, asindicated by the e in Ewenki mel- 'to awaken'.mi 'to cut' (p. 535). This may represent a front diphthong on the basis ofManchu mejje- 'to crumble, unwind, tear, etc.'

    mir 81Tow (p. 537). This form may best'be omitted, as it is isolated inArman Ewenki. The Nanai fonn mora.lt- crossbow does not belong here.mo wood (p. 540)

    mu I 'water' (p. 548)mu II rim (of a shaman's tambourine)' (p. 549). This is likely a Mongolian

    loan.mub 'to intercede for (p. 549)mej- 'to stagger' (p. 564). The Ewen fonns mei meji- indicate vocalic

    length.mek- 'to obtain' (p. 565). his fonn is isolated in Ewenki. The Oroch, Udehe,Ulchi, and Nanai forms cited by Cincius (1975) under this lemma repre-sent a different, bisyllabic form.men 'one's own, suus' {p. 568 . his fonn, of course, also includes a followingnasal.

    The second category consists of fonns with a short vowel and a subsequentnasal.

    man flock (p. 526) and mo ne 'impudent' (p. 544). The second form isattested in Ulchi as mo ne but in Manchu simply as mo and Cincius doesnot list it in any other Tungusic languages. f we take the Ulchi combinationas a single form, as Cincius does, then the initial m- would be expected toremain because of the bisyllabic structure of the form, as well as thefollowing nasal. It is the shorter Manchu fonn mo that raises questions hereand suggests the only possible counterexample to our rule. However, Stefan .Georg (p.c. 2000) has suggested that Manchu mo (in the construction moseme 'saying mo ') may be an onomatopoetic form, unrelated to Ulchi.The third category consists of fonns with a short vowel and no subsequent

    nasal.

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    9/12

    9 ANTHRoPOLO GICAL LINGUISTICS 44NO.lThis is precisely the environment in which we would expect initial m to bedenasalized to b-, so that any forms of this shape would be true counterexamples to our rule. The only such form listed in Cincius (1975) is mit we(inclusive)' (p. 539). This form, of course, is inseparable from the singular i I and takes its initial m- from the oblique forms of that pronoun.Indeed, the inclusive plural is attested with an initial'b- in Oroch and somedialects of other Tungusic languages.This rule, then, allows us a phonological justification for the equation of theAltaic specifier in -bA with the co Tesponding Indo-European and Uralic forms

    in -m. But before leaving the issue, we should clarify that the long vowels" wehave been speaking of as blocking the denasalization are only long vowels inTungusic. Starost in (1995) shows that long vowels in Tungusic usually co Tespond to short vowels in Turkic, while long vowels in Turkic co Tespond to shortvowels in Tungusic. In addition, both languages share a large number of formswith short vowels.3. Loss of the final vowel in Altaic. A further phonological question israised by the final vowel in the Altaic form -bAas opposed to the vowelless mfound in Indo-European and Uralic. As we have seen above, it is best to considerthe suffix as an originally free form, which would require a vocalic element.Thus, the form takes the shape CV in Japanese (which requires an open-syllablestructure in any case) and Tungusic, where it is still orthographically a separateword in Manchu (as are all case suffixes in written Manchu). Similarly,ltabashi s comparison with Written Mongolian ken ba 'whoever' and yambar bawhatever (1988:216) shows the form remaining as a free form. The evidencefrom these languages supports the view that -bA remained a free form untilrelatively late in Altaic, which explains why the final vowel was retained even inthose cases where it is now exclusively a bound form, such as the Ewenki suffixbs (intervocalically -wa), or Old Turkic n m thing, something', where theform appears graphically to be a bound form. The Turkic form also indicates

    that b became a bound form only after the phonological development of wordinitial m b-in Turkic.In Indo-European and Uralic, the accusative suffix functions only as a

    bound form, and shows evidence (especially in Indo-European, of course) ofhaving developed into a bound form much earlier. Significantly, it is these casesin which the final vowel has been lost, while it remains only in Altaic, where theform continued as a free form until considerably later.4. Further developments in Indo-European and Uralic. This informationfrom Altaic leads us to answer two further questions about the use of the su ixin the other languages. First, why was the marker of specification, which cameto indicate a definite direct object in Uralic and Altaic, generalized to refer to

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    10/12

    2002 PETER A. MICHALOVE 93any direct object in Indo-European? At first glance, t is may appear to be atrivial question, since such a generalization is hardly an unusual morphologicalor semantic development. In fact, the Old Japanese accusative suffix wo marking a definite direct object, has come (in its modem fonn o to indicate any directobject in Modern Japanese.

    lllic-Svityc (1976:50) proposes that in Indo-European, the accusative of thedefinite direct object was generalized to all the objects in the masculine andfeminine genders, while the neuters, for which direct object is the unmarkedfunction, further generalized the accusative fonn to the nominative. This is goodreasoning, and it may well have been a factor in the development of the IndoEuropean accusative. But in light of the typological research that has occurredsince that time, we can add a further reason. Indo-European is unusual amongthe languages of the world in having a marked nominative. It is typologicallyeven more unusual for a language to have a marked nominative and an unmarked direct object case, although there are some cases of this structureattested. Hence it is not surprising that Indo-European should have developedan overtly marked accusative.

    And finally, why was the accusative of specification of time and space lost inmost of Uralic when it was retained to the east and west in Altaic and IndoEuropean? A likely explanation is that many of the Uralic languages havesecondarily developed a profuse system of spatial cases, replacing the accusativeof space or time, so that the various Uralic languages express the extent of timeor space in a number of separate manners. Many of the Uralic languages thatexpress specificity of time and space through grammatical case do so withdifferent case suffixes, which only secondarily acquired that function. For example, Udmurt expresses the concept with the dative case, as in ki k nunal lt fortwo days ; Hungarian uses the terminative case for the same function, as het igfor a week .

    Of course, the development of secondary local cases does not guarantee thatspecificity of time and space will e expressed by novel means; the originalmarking by the suffix -bA is retained in the morphologically rich northernTungusic languages, as we saw in the Ewenki example above, not to mention inFinnish, as we have lso seen. But where elaborate secondary systems of spatialcases do occur, they provide a structural reason for the varying means of expressing specificity of time and space, as in Udmurt and Hungarian. Structuralexplanations such as this for the loss of the accusative of time and space in mostof the Uralic languages, or for the generalization of the accusative to expresseven indefinite direct objects as in Indo-European, provide a rationale for thespecific developments we fmd, and add further support to the idea of a commonorigin for the form in these languages.Thus, a closer examination of this suffix, especially the Altaic material,which originally appeared to be the weakest link in the correspondence, servesonly to support its status as an inherited morpheme, common to Indo-European,

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    11/12

    ........

    94 ANTHRoPOLOGICAL LINGUISTICS 44NO. 1Uralic, Altaic, possibly to Dravidian and, with much less assurance, to Afroasiatic.

    Perhaps the most persuasive aspect of this isogloss is not simply that wehave a common accusative suffix, or a common specifier of time and space, butrather that we see the agreement of these functions across all three of the language families we have considered here. In evaluating proposals of languagerelationship, many scholars have emphasized the importance of morphologicalevidence as particularly significant in establishing a common genetic origin,much as Bopp (1816) established the Indo-European relationship once and forall on the basis of verbal morphology. I agree entirely with the importance ofmorphological parallels. But in addition, this article has tried to show that aconsideration of the entire package of phonological, morphological, syntactic,and semantic components of a reconstruction is necessary to understand theform s origins and development.

    NotesAcknowledgments Many thanks to Fabrice Cavoto, Stefan Georg, and the tworeviewers for their valuable comments on an earlier version of this article. Of course they

    bear no responsibility for the conclusions stated here.Abbreviations The following abbreviations are used: 1 =first person; 2 =secondperson; 3 = hird person; ACC = accusative; IMP = imperative; NOM =nominative; PL =plural; so singular.1 The modem notation used here for Benzing's 2. Fabrice Cavoto (p.c. 2000) has kindly pointed out that Diakonoff (1988:64-67)considers mimation in Semitic as having a determining function, similar to a definitearticle, and that in Old Akkadian, where it is most clearly attested, it is absent in thestatus indeterminatus and often in proper names. Whether this feature can be

    attributed back to Afroasiatic as a whole and is comparable to the other markers ofspecification discussed here is questionable and, like the Dravidian case, will not bepursued further in this article.

    ReferencesBenzing, Johannes1955 Die tungusischen Sprachen: Versuch einer vergleichenden Grammatik.Abhandlungen der Geistes- und Sozialwissenschaftlichen Klasse 11:949-1099. Mainz: Akademie der Wissenschaften und der Literatur.Bomhard, Allan R., andJohn C Kerns1994 The Nostratic Macrofamily: A Study in Distant Linguistic Relationship.

    Bopp,FranzTrends in Linguistics, Studies and Monographs 74 Berlin: Mouton deGruyter.

    1816 Ober das Conjugationssystem der Sanskritsprache in Vergleichung mitjenem der griecheschen, lateinischen, persischen und germanischenSprache. Frankfurt am Main: Andreiische Buchhandlung.Cincius, V 1., ed.1975 Sravnitel'nyj slovar' tunguso-man'czurskix jazykov: materialy k etimologiceskomu slovarju. Vol. 1 Leningrad: Nauka.

  • 8/13/2019 Michalove - The Nostratic "Accusative" in *-mA - An Altaic Perspective

    12/12

    2002 PETER A. MICHALOVE 95Diakonoff, I M1988 Afrasian Languages. Translated by A A Korolev and V Ya. Porkhomovsky. Moscow: Nauka.Friedrich, Johannes

    1974 Hethitisches Elementarbuch. Erster Teil: Kurzgefasste Grammatik. 3d ed.Heidelberg: Carl Winter.Illic-Svityc, V M1976 Opyt sravnenija nostraticeskix jazykov (semitoxamitskij, kartvelskij,indoevropejskij, uralskij, dravidijskij, altaijskij). Sravnitel nij slovar . Vol.2. Moscow: Nauka.Itabashi, Yoshizo1988 A Comparative Study of the Old Japanese Accusative Case Suffix w with

    the Altaic Accusative Case Suffixes. Central Asiatic Journal32 3-4):193-231.Karlsson, Fred1999 Finnish: An Essential Grammar. 2d ed. London: Routledge.Konstantinova, 0. A1964 Evenkijskij jazyk: fonetika, morfologija. Moscow: Nauka.Miller, Roy Andrew1971 Japanese and the Other Altaic Languages. Chicago: University of ChicagoPress.

    Murayama, Shichiro1957 Vergleichende Betrachtung der Kasus-Sufflxe im Altjapanischen. StudiaAltaica. Wiesbaden: Harrassowitz.Starostin, SergeiA1995 On Vowel Length and Prosody in Altaic Languages. Moscow LinguisticJournal1:191-235.vertes, Edit1960 Beitriige zur Frage des flnnisch-ugrishen bezeichneten Akkusativ-objekts.Acta LinguisticaAcademiae Scientiarum Hungaricae 10:181-94.Vovin, Alexander1998 Nostratic and Altaic. n Nostratic: Sifting the Evidence, edited by Joseph CSalmons and Brian D Joseph, 257-70. Amsterdam Studies in the Theoryand History of Linguistic Science, Series IV, Current Issues in LinguisticTheory 142. Amsterdam: John Benjamins.2001a Japanese, Korean and Tungusic: Evidence for Genetic Relationship fromVerbal Morphology. n Altaic Affinities: Proceedings of the Fortieth Meeting of the Permanent International Altaistic Conference (PIAC), edited byDavid B. Honey and David C Wright, 183-202. Bloomington: ResearchInstitute for Inner Asian Studies.2001b North East Asian Historical-Comparative Linguistics on the Threshold ofthe Third Millennium. Diachronica 18:93-137.Whitney, William Dwight1973 Sanskrit Grammar, Including Both the Classical Language, and the OlderDialects, of Veda and Brahman [1889]. 2d ed. Cambridge: Harvard University Press.Wickman,Bo

    1955 The Form of the Object in the Uralic Languages. Uppsala: Almqvist andWiksells.


Recommended