Turkic pronouns against a Uralic background

Iran and the Caucasus 19 (2015) 79-86 Turkic Pronouns against a Uralic Background Marek Stachowski Jagiellonian University, Cracow Abstract Although nobody doubts today that a Ural-Altaic protolanguage is an obsolete idea there still exists some peculiar conformity between Uralic and Altaic that cannot easily be explained by simple borrowings. In this article some similarities between Turkic and Uralic pronouns are to the fore. Keywords Turkic, Altaic, Uralic, Pronouns, Areal Linguistics 1. INTRODUCTION The title of this article may amaze some readers. Even those who sometimes wonder whether an Altaic language family really is nothing but a mistake of older generations feel absolutely positive about the non-existence of a Ural-Altaic proto-affinity. At the same time, however, they all accept two facts: that the languages called Altaic and Uralic actually do display numerous typological (and, partially, material) similarities, as well as that the fathers of Altaic linguistics like Gustaf John Ramstedt, Martti Räsänen and Nicholas Poppe were great scholars. Of course, it is somewhat embarrassing to explain why we consider them great linguists if their ideas were wrong but I am not going to discuss this odd fact. Nor do I offer an earthshaking interpretation that will instantly convince everybody about the existence of both the Altaic and the Ural-Altaic protolanguage. My aim is far more modest. I would like to show the importance of Uralic data and parallels for our understanding of Proto-Turkic and, to some degree, Proto-Altaic (that might readily be interpreted as an areal linguistic notion rather than a genetic one).1 1 W. Kotwicz was the first scholar who interpreted the Altaic linguistic affinity as a  Koninklijke Brill NV, Leiden, 2015 DOI: 10.1163/1573384X-20150106 80 M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 2. PRONOUNS AND NUMERALS Turkic, Mongolic and Tunguzic personal pronouns can be compared with each other according to the usual methods of comparative linguistics, and the results make possible a reinterpretation of the Proto-Turkic system to the effect that the new version includes regular dual forms, absent from any modern Turkic language. This was shown by W. Kotwicz who proposed a system of regular proportions for Proto-Turkic: [a] the basic element is the singular with double phonetic forms (with the stem vowel ä or i);2 [b] the plural is derived by apophony from singular ä variants; and [c] the dual is built of singular i stems to which a dual or collective suffix *-z was added (Kotwicz 1936: 33); for instance: 2. person singular *sä(n) ~ = m + a front vowel *si(n). 2. person plural *sa(n); 2. person dual *siz.3 Apparently, no distinction was made between the singular and the plural in Proto-Uralic, e.g. *m" ‘I; we’. Later, the difference was introduced in two ways: by apophony and by suffixation (which very much resembles contact-induced one (see esp. Kotwicz 1951: 313), even though he is still incorrectly considered to have been a convinced “genetic Altaicist” (for further details, see Stachowski 2001: 217, fn. 10). Some similarities and differences between Altaic and Uralic can be now found in an article by R. Austerlitz (2009: 142 sq.). Actually, this study is older than the date suggests; it was written in 1976 or 1977. Austerlitz sent it to G. Doerfer in Göttingen for publication. However, it was never published and the reasons for this situation remain unknown (one can only venture a guess that the somewhat popular character of the article caused serious objections to be raised against publishing it in a scholarly journal). The original text was forwarded by G. Doerfer to M. Knüppel who decided to publish it in 2009. 2 Presently, we still cannot define the difference between the two vocalic variants. 3 In Kotwicz’s table (1936: 33), a variant *bä(n) is missing, certainly by accident, in the section of the 1. person singular, one that is necessary for the introduction of the 1. person plural form *ba(n). Thus, the Proto-Turkic system should have the following forms of the 1. person: singular *bä(n) ~ *mä(n) ~ *bi(n) ~ *mi(n), or just: *b/mä(n) ~ *b/mi(n) (the notation *b/mä/i(n) probably looks a little obscure); plural *b/ma(n); dual *biz. For the b- ~ malternation see fn. 5. The consonant *n must have originally been a singular marker (see below). Kotwicz probably was not quite sure about the chronology of derivation, that is, whether the singular pronouns were suffixed already in Proto-Altaic, Proto-Turkic or in a later phase. Nor are we today. Only some of these forms are asterisked by Kotwicz because he used asterisks to signal forms entirely reconstructed by comparative methods. Thus, because the Proto-Turkic dual form *biz phonetically coincides with the plural form biz known to us from historical and modern sources, Kotwicz writes biz without an asterisk for Proto-Turkic. M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 81 the Turkic pronouns). Examples for apophony are, for instance, as follows: Komi me ‘I’ ~ mi ‘we’, Hung. te ‘thou’ ~ ti ‘you’. The Old Hungarian form min ’we’, as well as Vogul dial. män ~ man and Mordvin miń ‘we’ display a “pronoun suffix *-nɜ” (UEW 3: 295 sq.). Now, is the Proto-Uralic suffix *-nɜ actually different from the Proto-Turkic *-n in *b/mä(n), etc. (see above)?4 I wonder whether the following scenario is too far-fetched: In some area in Northern Asia a continuum of dialects existed. These dialects made no difference between the singular and the plural in their pronominal systems. Their other common features were: [a] anlaut consonants (*b- or *m- for the first person,5 and *s- or *t- for the second6) and [b] the same front vowels in the singular (Turkologists usually put an *ä, *e or *i here whereas Uralists invented a practical device, viz. the symbol “denoting a front vowel; neither Turkologists nor Uralists seem to have ever proposed labial front vowels for the reconstructed pronouns). Besides, every pronominal stem had the structure of an open syllable both in Turkic and in Uralic. As time went on the general differentiation of persons but none of grammatical numbers evidently appeared cumbersome or at any rate uncomfortable and impractical. All the dialects in this area launched a reform of the pronominal system and they seem all to have 4 K. Rédei connects the Uralic *m" ‘I’ with the Altaic *bi, attested, for instance, in Turkic “min ~ män (< bi-n)” (UEW 3: 294), and the Uralic *m" ‘we’ with the Altaic *bi, attested, for instance, in Turkic biz ’we’ (ibid.). All this is incorrect and very peculiar. 5 The Turkic b- ~ m- alternation is probably nothing but our ignorance. We were the ones who traditionally could not decide whether b- or m- was the original anlaut here because the modern reflexes seemed to admit both reconstructions. However, if Kotwicz’s reconstruction of Proto-Turkic pronouns is correct (and it seems, it is), the dual form *bi+z has no m- variant and no auslaut –n, which shows that the possible *m- variants are relatively later phonetic forms resulting from the *b – n > *m – n assimilation. This solution matches the prevailing Turkological approach (*b- is original) and contradicts the Nostratic tendency of reconstructing Proto-Turkic *m- and comparing the Turkic pronouns with those in other linguistic families that use m- as a typical signal of the 1st person singular pronoun. Since we do not know the time of the *b – n > *m – n change one could also use some other notation like *Bä/in and *Ban (with *B = for “*b or *m”) in Proto-Turkic because one cannot exclude the possibility of the existence of both forms already in Proto-Turkic. 6 The S ~ t alternation (S = any sibilant) is a frequent phenomenon in various Siberian languages. = m + a front vowel M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 82 had the same methods at their disposal: apophony and the “pronominal suffix *-n(ɜ)”.7 A comparison of some modern reflexes, as, for instance, Hung. te ‘thou’ ~ ti ‘you’ with Turkish se(-n) ‘thou’ ~ si(-z) ‘you’ shows that only the general lines and the types of devices used really coincided in the changing process. Also the grammatical function of the suffix *-nɜ was understood differently in various dialects. Old Hungarian, Vogul and Mordvin (see above) assigned the suffix the function of a plural marker whereas Turkic and some other Uralic languages (cf. Ostyak mä ~ mȧ ~ mȧn ‘I’ [UEW 3: 294]) used it as a sign of singularity. In other words: general methodology and instruments were the same but the specific “recipes” were a private matter of each dialect and that is why the results of this reform yielded somewhat different products in the present-day languages. The situation with numerals is very different. A glance at the three first numerals in Turkic, Mongolic and Tunguzic8 suffices: ‘1’: ‘2’: ‘3’: Tkc. *bīr, Mo. *niken, Tng. *ämün; Tkc. *ik(k)i, Mo. *qojar,9 Tng. *ǯör; Tkc. *üč, Mo. *γurban, Tng. *ilan. Reconstructing common protoforms of numeral systems in the three families does not appear very promising. Thus, if one starts thinking of the possibility of some genetic affinity of the languages called Altaic based on their pronominal systems he may accept this hypothesis as quite believable. However, if his starting point is the numerals he will certainly deny any protolinguistic affinity. Interestingly enough, a parallel situation is known from Uralic. The Fenno-Ugric and the Samoyedic pronominal systems are similar while numerals are not. Only two numbers coincide in Fenno-Ugric and Samoyedic, and even that result is not necessarily accepted by every linguist 7 That is, *-n in what was going to become Proto-Turkic, and *-nɜ in the future ProtoUralic. For Pre-Proto-Turkic (Proto-Altaic?) a still older form *-nɜ can also be accepted if the original auslaut vowel had neither dynamic nor pitch accent (cf. Poppe 1960: 146). 8 The Tunguzic languages are the least known ones in the Altaic family. The reconstructed numerals are adduced here after an old but still usable comparative grammar by Benzing (1956: 101). 9 A Middle Mongolian word ǯirin ‘two; pair’ (Poppe 1987: 243) may be possibly somehow connected with Tng. *ǯör ‘id’. M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 83 because the conformity is not absolutely perfect. The first near conformity is between ‘5’ in Fenno-Ugric (e.g., Finnish viisi, Hungarian öt) and ‘10’ in Samoyedic (e.g., Nenets jūt and Nganasan bīʔ). These can all be derived from a Proto-Uralic form *witte but some seem to be worried about the semantic difference. The other equation concerns the word for ‘2’: Finnish kaksi, Hungarian két, Mordvin kavto vs. Selkup sitte ~śede, and so on. The semantic aspect is quite easy here but the degree of backness of the original vowel remains unsettled―should the protoform be reconstructed *kakta or *käktä? One of possible explanations of the disagreement of Fenno-Ugric and Samoyedic data is that the Proto-Uralic language had no numerals yet; they developed only after the differentiation of Proto-Fenno-Ugric and Proto-Samoyedic was definitely completed. Despite the lack of common numerals attributable to a protolinguistic phase the common origin of Fenno-Ugric and Samoyedic is widely accepted. The question is whether the same can be achieved with Turkic, Mongolic and Tunguzic. The category of grammatical number is generally a rather late invention in both Turkic and Uralic. The Bulgarian Turkic branch separated from all the other Turkic dialects when its speakers abandoned their original homeland in present-day Northern Mongolia approximately in the 5th century. Because the modern Chuvash language, being the only living reflex of the Bulgarian Turkic group, is at the same time the only Turkic language that does not possess the Turkic plural suffix -lar (using sĕm instead) a conjecture is possible that the plural suffix -lar did not yet exist in the 5th century. It will have been created at a time after the Bulgarian Turkic exodus and before the period of the Old Turkic stone epitaphs (8th century) but its use in Old Turkic still was restricted to nouns denoting men of higher social status like representatives of tribal aristocracy (nota bene, the plural marker -lar occurs distinctly less often in Turkish than in the Indo-European languages even today). The relatively late distinction between the singular and the non-singular (i.e., the dual or the plural) is a common feature of Turkic and Uralic. One cannot but wonder whether this fact is a mere coincidence or, maybe, rather points to original mutual influences. M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 84 In the two language families the starting point of the pronominal system will have probably been a sort of the transnumeral. The second feature common to both families is the use of the very similar singular suffix (Proto-Turkic *-n, Proto-Uralic *-nɜ). The third feature is apophony used to distinguish plural pronouns from singular ones. The numeral systems in Turkic and Uralic are far less similar. However, at least one feature points to their originally common area and mutual influences: they are combined with nouns in the singular, e.g., Hungarian két macska = Turkish iki kedi ‘two cats’, lit. ‘two cat’. For some other (mainly) grammatical similarities see Stachowski 2001: 215-219. Let us move on now to two lexicological cases. 3. PRONOUNS AND WORD STRUCTURE It is an obvious fact that words can easily be borrowed from one language into another and, thus, cannot be used as sufficient arguments for a genetic affinity of two languages. Common words do not even necessarily point to particularly intensive contacts between languages, a case in point being the Latin influence on the lexis of all modern European languages. Nevertheless, some words display structural and/or material features that cannot readily be reduced to simple borrowing processes. One of such words is modern Turkish işte ‘voilà’, originally being an intensified (*iš-) variant of a Proto-Turkic demonstrative pronoun *te ‘this/that in view, one in range of vision’. The reconstructed pronoun *te does not exist as such any more in modern Turkic but its traces can still be found in the Siberian Turkic languages, e.g., Proto-Turkic derivative *te+kü (adj.) ‘being in the range of vision’ > Chulym tég, Tofalar tē, and so on (Stachowski 2007: 174). But the crucial point in our context is that the Proto-Turkic *te seems to have a good equivalent in the Proto-Uralic pronoun *t" ‘this’, sometimes also noted *tä ~ *te ~ *ti (UEW 5: 513)10. But that is not all. = m + a front vowel 10 As for the different vowels in the Proto-Uralic reconstructions see UEW (5: 514): “Der Stammvokal kann nicht auf eine einzige U [= Uralic] […] Vorform zurückgeführt werden […]. Das bunte Bild im Vokalismus―das bei Pronomina keine Seltenheit ist―kann auch uralt sein (vgl. *m" ‘wir’ U, *t" ‘ihr’ U). Die unregelmäßige Vokalvertretung im Stamm ist noch kein Grund dafür, in der U […] Grundsprache mit solchen aus t + palatalem Vokal bestehenden Demonstrativpronomina zu rechnen, die verschiedenen Ursprungs sind”. = t + a front vowel M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 85 When publishing a study on short and long declensional forms of the Yakut interrogative pronoun tuox ‘what?’ < *tōx < *tōk < pronoun *to + intensifier *-ok (Stachowski 1990) I felt somewhat uneasy about the fact that *to is unknown in the other Turkic languages. But now a system-oriented solution appears possible. My hesitation in 1990 was caused by the fact that I was seeking some traces of *to in Turkic only and failed to notice the existence of the ProtoUralic pronoun *to ‘that’ (UEW 5: 526). Even though no other solution of the Proto-Turkic *to exists one might doubt whether borrowing of a rather basic pronoun is actually realistic. But in this case we are dealing with two or, maybe, even three common pronouns. The third one, unmentioned up to now, is Turkic ne ‘what?’. Its most striking feature is the word-initial n- because the Turkic languages do not generally tolerate the four sonorant consonants m, n, r, l in the anlaut of native Turkic words but in some onomatopoeiae. Ramstedt’s (1957: 74) suggestion of a possibility of reconstructing an affricate *mb- for ProtoTurkic words whose modern reflexes display a m ~ b alternation, although questionable in and of itself, makes an idea arise that Proto-Turkic *ne and *to could both be traced back to a still older form like *ntV. All this looks very uncertain, indeed. I would, therefore, rather suggest explaining the Proto-Turkic *ne ‘what?’ as an interrogative reflex of the Proto-Uralic pronoun *nä ~ *ne ~ (?) *ni ‘this’ (UEW 3: 300), i.e. semantically presumably: ‘this’ → ‘this one?’ → ‘what?’ or with a simile: ‘this’ → ‘(something) like this?’ → ‘what?’. The same evolution is also valid for the previous example: Uralic *to ‘that’ and Proto-Turkic *ti+ok > Yaku tuox ‘what?’. In so doing we arrive at a possibility of reconstructing the following pronominal forms (a bit tendentiously simplifying the Proto-Uralic vowel reconstructions): Proto-Turkic *te ‘this’ ~ Proto-Uralic *te id. Proto-Turkic *to ‘that’ (> Yakut ‘what?’) ~ Proto-Uralic *to id. Proto-Turkic *ne (‘this’ >) ‘what?’ ~ Proto-Uralic *ne ‘this’. 4. CONCLUSIONS It is absolutely clear to me that the few examples cannot revolutionarily change the common opinion of the non-existence of the Ural-Altaic protolanguage. On the other hand, however, the examples presented above *to+ok M. Stachowski / Iran and the Caucasus 19 (2015) 79-86 86 must not be ignored or forgotten. The best solution probably is an extension of Kotwicz’s idea of a Siberian conglomerate11 consisting of the Altaic and the (some?) Uralic languages. As it seems, a clear (albeit not altogether simple) answer to the Altaic and Ural-Altaic question will presumably come one day from Siberia and from areal rather than genetic linguistics. BIBLIOGRAPHY Austerlitz, R. P. (2009), “Ural-Altaic languages”, Sibirische Sprachstudien 4/2: 129-144. Benzing, J. (1956), Die tungusischen Sprachen. Versuch einer vergleichenden Grammatik, Wiesbaden. Kotwicz, W. (1936), Les pronoms dans les langues altaïques, Kraków. ——(1951), “Studia nad językami ałtajskimi”, Rocznik Orientalistyczny 16: 1-317. Poppe, N. (1960), Vergleichende Grammatik der altaischen Sprachen, vol. 1: Vergleichende Lautlehre, Wiesbaden. ——(1987), Introduction to Mongolian Comparative Studies, Helsinki. Ramstedt, G. J. (1957), Einführung in die altaische Sprachwissenschaft, vol. I: Lautlehre, Helsinki. Stachowski, M. (1990), “Der Lautwandel in der Deklination des jakutischen Interrogativpronomens tuox? ‘was?’”, Zeszyty Naukowe UJ. Prace Językoznawcze 101: 115-120. —— (2001), “Uralistik und Turkologie – geschieden und doch verliebt”, J. Pusztay (ed.), Vade mecum! A huszonötödik óra, Szombathely: 209-225. —— (2007), “Die Etymologie von ttü. şimdi ‘jetzt’ und işte ‘voilà’ vor gemeintürkischem Hintergrund”, Studia Etymologica Cracoviensia 12: 171-176. UEW 3 = Rédei, K. (1986), Uralisches etymologisches Wörterbuch, Lieferung 3, Budapest. UEW 5 = Rédei, K. (1988), Uralisches etymologisches Wörterbuch, Lieferung 5, Budapest. 11 I owe the term conglomerate that is free of associations with the terms league or Sprachbund to Robert Woodhouse (Brisbane).