Dėl hidronimų darybos nagrinėjimo

LIETUVIŲ KALBOS MORFOLOGINĖ SANDARA IR JOS RAIDA LIETUVOS TSR MOKSLŲ AKADEMIJA ON THE STUDY OF HYDRONYM FORMATION A. VANAGAS The task of examining the formation of hydronyms1, like that of proper names in general, differs from the study of common-word formation. In analyzing the formation of common words, the aim is to establish how a language creates new words and replenishes its vocabulary2. The formation of place names, especially names of bodies of water, must be approached differently. By now, all the more significant locations and features of the earth’s surface have names, many of which have been passed down orally for centuries and millennia, from generation to generation, and have even been given legal recognition. Thus, the process by which place names, especially names of bodies of water, came into being has for all practical purposes nearly ended. Unlike common words, we are dealing here not with a living system constantly shaped by the language’s internal laws, but with an almost fully formed, established, no-longer-living system. This is not, of course, to say that place names, including hydronyms, can no longer change. The names of some features, especially minor and less important ones, change through folk etymology, analogy (for example, acquiring plural forms), or the influence of the literary language; they adapt to certain dialect features as the dialects themselves change, and so on. However, a new hydronym can arise only when a new feature appears within the sphere of human practical activity and must be given some name for communication purposes. Very few new rivers or lakes are created nowadays, mostly by artificial means, and so they are usually given artificial names (e.g., Kauno marios, Ventos perkasas, etc.). By contrast, extensive land-reclamation works are destroying lakes and rivers quite intensively. Some of the features whose names are now being examined have already disappeared, and their names are on the verge of being forgotten or have been entirely forgotten. Thus, we cannot, of course, discuss developmental trends or new ways in which hydronyms are formed. The task of examining hydronym formation is to trace this, 1 Here and below, the term hydronyms is used in its narrowest sense, namely, “names of rivers and lakes.” 2 See Л. В. Щерба, Очередные проблемы языкознания, «Известия Академии Наук СССР. Отделение литературы и языка», М., 1945, р. 181. 207 the ways in which the system of river and lake names—which is already almost no longer living and, it seems, tending to contract—came into being, and to identify the full set of means that shaped it. Because of its stability, the hydronymic system somewhat resembles the systems of appellatives and place names in dead languages (Latin, Prussian, etc.). This leads to a further conclusion. Since the hydronymic system has already taken shape and changes little, the researcher of their formation always looks more toward the past, toward the period when the now-static system was still taking shape. This applies especially to those hydronyms which, following the disappearance of the corresponding features, are no longer, in essence, facts of the living language. The stability of the system is one of the most characteristic features of hydronyms, distinguishing them from appellatives. It should be noted that the hydronymic system tends to become fixed (in form and semantics) from the very beginning of its formation. Consequently, its study raises problems rarely encountered in the study of common words. This article discusses only two major questions related to hydronym formation: 1) the classification of hydronyms; 2) derived hydronyms. 1. The question of classification is central to toponomastics and concerns all layers of place names equally (hydronyms, names of inhabited places, microtoponymy, etc.). All subsequent analysis of specific material depends on classification; the classification scheme shows, first and foremost and most clearly, the researcher’s approach to toponymy. Yet even today this question has not been definitively resolved. It remains unresolved primarily because the immense diversity of place names and their special position in the language system make it very difficult to devise a universal classification. “Classification of place names is not a mechanical matter of sorting them into boxes, but the establishment of a system of regularities,” says V. Nikonov.3 Moreover, the development of scholarship is always associated with changes in how the object of study is viewed; therefore, classifications that were acceptable in the past may now have only historical value. It is true that no classification of Lithuanian (or, more broadly, all Baltic) place names has been compiled, nor has anyone attempted to apply classifications of place names from other languages to the Lithuanian name corpus. Our closest neighbors, the Slavs, however—especially the Poles, Czechs, Bulgarians, Yugoslavs, and others—have accomplished much more in this field. The first classifications of Slavic place names were compiled around the mid-nineteenth century4. Alongside their many advantages, these classifications also had shortcomings, the most important probably being that their aims were primarily not purely linguistic but “utilitarian”—that is, rather than seeking to analyze toponymic systems linguistically, they were more concerned with 3 V. Nikonov, Postwar Works on Toponymy in the Slavic Countries, “Brief Communications of the Institute of Slavic Studies of the Academy of Sciences of the USSR. Slavic Linguistics, No. 28. Moscow, 1960, p. 77. 4 For a critical review of all of them, see W. Taszycki, Słowiańskie nazwy miejscowe (Ustalenie podziału), „Prace Komisji Językowej PAU“, Nr. 29, Kraków, 1946; or W. Taszycki, Onomastyka, Wrocław-Kraków, 1958, p. 229—258; В. Никонов, Послевоенные работы по топонимике..., p. 75—105. In our view, S. Rospond has assessed the earlier classifications most correctly in his book „Klasyfikacja strukturalno-gramatyczna słowiańskich nazw geograficznych“, Wrocław, 1957, p. 14—27. 208 to find out what information place names can provide people today about the past, culture, nature, vegetation, customs, social relations, and so on of peoples. Place names were regarded not as a special linguistic phenomenon whose study should be linguistic, but as “fossiles de la géographie humaine,” “Geschichtquelle,” “Siedlungsgeschichte,” and so on.5 This, naturally, is not meant to imply that all earlier classifications were pointless or contributed nothing of value to the development of toponomastics. Linked to the development of the scholarship of their time as a whole (and not only of linguistics), they helped clarify the principal features of place-name semantics and, in part, word formation; they provided a wealth of information about the past of peoples, languages, nature, vegetation, and so forth, thereby drawing many scholars’ attention to names. As can be seen, for example, from S. Rospond’s review of earlier classifications6, the improvement of classifications was linked to the intensified collection of place names and to the development of toponomastics as a separate branch of linguistics. Semantic-etymological classifications are now being abandoned not only because, lacking stricter linguistic principles, they often confused different facts, made many errors in explaining word formation, and failed to encompass the full diversity of place names, but apparently also because they have already largely accomplished the task they set themselves. The further development of toponomastics without earlier classifications (for the Slavs: those of F. Palacký, F. Miklošič, T. Wojciechowski, F. Schwarz, E. Muka, P. Skok, K. Moszyński, V. Taszycki, and others7) is difficult to imagine. Two works published at nearly the same time in different countries offer new solutions to the problem of classifying place names. These are P. Lebel’s8 “Principles and Methods of French Hydronymy” and S. Rospond’s9 “Structural-Grammatical Classification of Slavic Place Names”10. In many respects, the ideas of these two distinguished toponymists coincide on the most important points11. The works of P. Lebel and S. Rospond can be considered innovative because they approached place names, this specific part of the vocabulary, not as seekers of historical information, but as linguists examining facts of language. P. Lebel divides all French hydronyms into three groups: 1) appellative names (les noms appellatifs ou noms génériques); 2) qualifying names, or descriptive expressions (les noms qualificatifs ou expressions descriptives); 3) determining names, that is, adjectives or expressions that establish an object’s location in space or time, or its duration of existence (les noms déterminatifs). For example (respectively), Fontaine ‘spring’, la Fontaine — Bouillante ‘boiling spring’, la Fontaine — Saint — Martin12. 5 See S. R o s p o n d, Klasifikacija..., p. 31. 6 S. R o s p o n d, Klasifikacija..., pp. 5—27. 7 Ibid. 8 P. L e b e l, Principes et méthodes d’hydronymie française, Paris, 1956. 9 S. R o s p o n d, Klasifikacija... 10 For a review of both works, see Б. Т о п о р о в, К проблеме классификации в топонимии, «Исследования по структурной типологии», М., 1963, pp. 226—231. 11 Cf. ibid., p. 231. 12 P. L e b e l, Principes et méthodes d’hydronymie..., p. 9. 14. Lietuvių kalbos morfologinė sandara ir jos raida 209 S. Rospond’s scheme is considerably more complex, but here too there are three main groups: 1) primary hydronyms, that is, those derived from other words but bearing no signs of toponymic formation. They may derive from appellatives (Bagno, Blato, Jama) or from personal names (Jersz, Balaban, Bogdan); 2) secondary hydronyms, of which there are five groups in all: prefixal, prefixed, prefixed-suffixed, accentual (e.g., Ukrainian Suxyj=suxyj), and ending derivatives (various changes in the ending are meant); 3) composita, that is, compound and complex names. More recently, another work devoted specifically to the classification of hydronyms appeared. It is an article by the well-known Bulgarian linguist J. Zaimov13, “The Problems and Classification of Water Names.” J. Zaimov’s scheme consists of four main parts: 1) the semantic classification of water names; 2) the structural classification of water names; 3) the classification of river names according to river length; 4) the origin of names. The first part is traditional; there, water names are divided into various groups according to their semantics. The second part is devoted, one might say, solely to the formation of water names. In the third part, river names are classified according to river length: names of rivers over 100 km long, from 50 to 100 km long, and so on. In the fourth part, river names are grouped by origin (Bulgarian, Greek, Turkish, Thracian, etc.). It is easy to see that the basis for division in these four parts of the scheme is entirely different. In the first part, the semantics of the hydronym are traced through etymological analysis; in the second, its structure is discussed; in the fourth, its origin. The basis for constructing the third part is not linguistic, and therefore, in our view, the inclusion of this part cannot be justified linguistically. From the perspective of word formation, the second part is the most important. In some respects, it resembles the classifications of P. Lebel and S. Rospond. However, it also differs from them in many ways. The differences are primarily related to the rather obvious fact that J. Zaimov’s scheme is inconsistent, most likely because of an insufficiently rigorous linguistic interpretation of hydronyms. First, as in the scheme as a whole, there is no single basis for classification in the second part either. If the absence of a single basis for classification may be justified in separate parts of the scheme that could be considered autonomous, then the grouping of hydronyms within a single part bearing the title “Structural Classification of Water Names” should be based on one principle. The scheme does not always take the structure of hydronyms into account. If their structure had been considered rigorously, then, for example, two-word river names would have been distinguished from one-word names (that is, water names in chapter three Паду валог, Елени дол; in chapter four, Ивански дол; and in chapter six, subsection b), Кокалска река, 13 J. Zaimow, Classification and Problems of Water Names (Based on Bulgarian Material), “I International Slavic Conference on Onomastics,” Wrocław—Warsaw—Kraków, 1961, p. 171—180. 210 Subsection d) — Дивлянска река, subsection e) — Дряновски дол, etc., should constitute a single separate group of compound hydronyms, in which several subgroups can also be distinguished. Secondary hydronyms are not divided into suffixal and prefixal derivatives, although such groups are very much needed and can be distinguished. There are many more such inconsistencies. Therefore, although J. Zaimov’s classification was compiled considerably later than the classifications of P. Lebel and S. Rospond were published, it seems impossible to regard it as a new step forward in toponomastics. In our view, the most acceptable of the existing classifications at present are those of P. Lebel and S. Rospond, although they too are not without shortcomings14. P. Lebel’s classification is based on toponymic material very distant from the Lithuanian language, so it is more convenient to use S. Rospond’s scheme, since Slavic toponymy has much in common with Lithuanian. Moreover, S. Rospond’s scheme can be better implemented in studies of this kind, whose point of departure is the morphological structure of place names. With this in mind, there is also a basis for applying the idea of a structural-grammatical classification to Lithuanian hydronymy. The possibility of a structural-grammatical classification of Lithuanian hydronyms is also shown by the brief overview of the structure of Lithuanian river names presented below. First, all Lithuanian river names fall structurally into two unequal parts: I. one-word river names and II. multiword river names. 1. O n e-w o r d r i v e r n a m e s. This group of Lithuanian river names is the largest, several times larger than the second. The river names belonging to it also differ in derivation, and can therefore be classified accordingly: 1. River names derived from common or proper words that have remained entirely unchanged. S. Rospond calls these primary. Examples: a) from appellatives: Balà riv., Įñtakas riv., Upė riv., Ežeras lake; b) from proper names: Kūndrotas riv., Vaidōtas riv., Amėrika riv., Sibiras riv. A separate small group would consist of river names derived from adjectives (definite and indefinite), that is, river names formed through the substantivization of adjectives (Dargióji, Juodóji, Šventóji, Šventója, Saltója, Juodà, Baltà). 2. Derivatives, that is, one-word river names that have some derivational marker. They are the most numerous of all river names. According to their formation, two large groups can be distinguished: A. Derived forms (suffixal, ending-based, prefixal—for more details, see p. 212). B. Compound river names (Juodupis, Kalnupis, Gėčintakis, Platuagė). Reduplicative river names would also belong here (Veiviržas, Siesartis). II. M u l t i w o r d r i v e r n a m e s. This group is much smaller than the first and less varied. According to their formation, it can be further divided into ¹⁴ See B. Топоров, К проблеме классификации..., p. 226—231. 14* 211 two subgroups: 1) genitive forms, i.e., river names formed from the genitive case of some proper or common noun and, most often, a physiographic term15. Genitive-form river names may consist of two or three words, for example, Ȧdaraiščio upėlis, Ašmonų upėlis, Jonyčio intakas, Lankėlės raũgas, Baltų Bóbų Rãvas, Bálų valkòs upėlis; 2) qualificative names, for example, Juodasis upėlis, Baltàsai griovỹs, Gývasis upėlis. This classification of Lithuanian river names differs in some respects from S. Rospond’s classification. The differences are primarily objective in nature, since the systems of Baltic and Slavic hydronyms are not identical. In addition, there were other reasons for the differences. (In our scheme, some elements of the classification are arranged or treated differently for reasons of logic and consistency. For example, in our view, secondary river names include not only derivatives, but also compound and multiword names, since they have toponymic derivational features, though different ones from derivatives. In drawing up this scheme, greater attention was paid to the structure of the hydronym than in S. Rospond’s. Consequently, all river names are first divided into two groups: one-word river names and multiword river names. We classify suffixes according to an entirely different principle from S. Rospond (for more on them, see below). There are also other, minor discrepancies; for example, in our scheme primary river names are not further subdivided semantically. This classification scheme for Lithuanian river names can readily be applied to certain other strata of the Lithuanian nomenclature as well: lake names, other hydronyms, and, in part, microtoponymy. 2. The problem of derivatives is closely connected with classification. It arises when the derivation of hydronyms is examined. In addressing the problem of derivatives, we took the derivation of common nouns as our point of departure, since many general principles of derivation have already become sufficiently established in this area. Hydronyms (like all proper nouns) constitute a very distinctive part of the lexicon, possessing many features peculiar to them alone. Moreover, all proper nouns derive, directly or indirectly, from common nouns. For this reason, it may be assumed a priori that many regularities in the derivation of common nouns may also, to a greater or lesser extent, characterize hydronyms. The extent to which the rules governing the derivation of common nouns can be applied to hydronyms will, naturally, depend on how closely hydronyms are related to common nouns and how they differ from them, or on the place hydronyms (and all proper nouns) occupy among other words. Only after clarifying this question will we be able to understand what can be drawn from the analytical apparatus for common-noun derivation in analyzing hydronym formation, and what features are distinctive to the study of hydronym formation because of its particular linguistic status. Every independent common noun has semantic content and a sphere of use. The relationship between content and sphere of use is inversely propor- 15 For more on Lithuanian genitive-construction place names, see B. Savukynas, LKK, vol. 6, p. 235. 212 213 tional: the richer, more concrete, more definite, and less abstract the semantic content, the narrower the sphere of use, and conversely: the poorer, more abstract, and less definite the content, the broader the sphere of use16. For example, the semantic content of the word alksnis ‘alder’ is richer, more concrete, and more definite than that of the word medis ‘tree’, but the sphere of use of the latter (i.e. medis) is much broader than that of the word alksnis. However, not all lexical strata in a language are alike in this respect. There are words whose semantic content is very poor and indefinite, while their sphere of use is broad17. For example, the content of the Lithuanian demonstrative pronoun šis ‘this’ is very indefinite and poor; consequently, its applicability is extremely broad, limited only by a grammatical criterion (it can replace any masculine noun). By contrast, the content of the words Nėmunas, Vilnius, Napoleónas, Madridas, Nilas is so concrete, concentrated, and detailed that not even a shadow of abstraction can be discerned in them; their sphere of use is therefore reduced to a minimum: they can denote only one single object18. The detailing and concretization of semantic content in a proper noun have reached the absolute limit. And it is precisely from the moment when the detailing of a common noun’s semantic content reaches its limit that the common noun becomes a proper noun. Once it has become a proper noun, the word seems to freeze, to die; its semantic development comes to a halt, because its original meaning becomes completely irrelevant to speakers; it soon begins to stand apart from the language system as a fossilized symbol devoid of semantic content, serving only a nominative function in the language. “...place names,” says A. Dauzat19, “are very quickly sterilized, detached from their original meaning; their semantic life ends with their formation.” If a common noun means or denotes some object or phenomenon in the environment, a proper noun merely identifies and names it (in J. Kurylowicz’s words — il nomme20). For this reason, proper nouns are sometimes considered in the language system to occupy an intermediate position between elements of their own language system and “hieroglyphic”-type elements (such as international words, abbreviations, various symbols, insertions from foreign-language texts, and so on)21. Place names, changing little over centuries and millennia, often survive much longer than the common nouns from which they were formed. Today, such place names are entirely unclear and incomprehensible to us. Moreover, every language’s nomenclature includes a proportion of place names that have entered from other languages in various ways (through substrate influence or borrowing). Consequently, a large proportion of Lithuanian place names, especially hydronyms, at present are dif- 16 J. K u r y ł o w i c z, La position linguistique de nom propre, „Onomastica“, rocz. II, zesz. 1. Wrocław, 1956, p. 1. 17 J. Kuryłowicz, La position linguistique..., p. 2. 18 The problem of recurring place names and personal names is not discussed here, since it is not directly related to the topic. Moreover, in principle, it is easy to resolve. 19 A. Dauzat, La toponymie française, Paris, 1946, p. 13. 20 J. Kuryłowicz, op. cit., p. 2. 21 B. H. Топоров, Некоторые положения теоретической топонимики, «Принципы топонимики», М., 1962, p. 3. 214 semantically are no longer clear, or, in other words, they no longer have derivational meaning22. A word can be considered a derivative of another only if: 1) alongside the derivative there is a word from which it is derived23, for example: brūkavimas “cooing” from brūkuoti “to coo”; 2) there is a difference in meaning24 between the base word and the derivative, since the content of every derivative adds something new25; 3) the difference in meaning must necessarily be the result of a grammatical change26. (Otherwise, we could consider all polysemous words derivatives, e.g., nerti “to dive into water” and nerti “to knit”.) These fundamental principles of the derivation of common words can also be applied to hydronyms. Clearly, their application is possible only to a limited extent, insofar as the commonalities between these two layers of the lexicon allow. These commonalities are various, for example, identities of morphology, derivational morphemes, syntactic functions, and so on.27 For our purposes, the most important point is that the derivational connection of some hydronyms with corresponding common or proper words is still entirely clear. Their derivational meaning is clear as well (e.g., river Trūmpė, lake Apskritÿs, river Barsukÿnas, river Juodėlė, lake Ringÿs, etc.). The fundamental principles of the derivation of common words can be applied without reservation to hydronyms of this kind. However, it can be said in advance that such cases are relatively few, since even hydronyms with clear etymologies cannot always be derived from corresponding appellatives. For example, we seemingly have no grounds for confidently deriving the river name Duobaitė (Skaudvilė, tributary of the Minsnora) from the common noun duobė “pit,” since it may be derived from another place-name, *Duobė (for example, the name of a hollow, meadow, marsh 22 For this concept, see V. Urbutis, Dabartinės lietuvių kalbos sudurtinių daiktavardžių daryba, “Dabartinė lietuvių kalba,” Vilnius, 1961, p. 67. 23 G. O. Vinokur, Notes on Russian Word Formation, “Selected Works on the Russian Language,” Moscow, 1959, p. 425. 24 On distinguishing a base word from a derivative, cf.: “A base word can be distinguished from a derivative by its place in the general system of word formation, specifically, by the direction of word-formation relations in a word-formation nest, in a word-formation synonym series, and in a particular structural-semantic group of base and derived stems” — V. N. Zinovyev, The Problem of Phonetic-Morphological Word Formation in Russian, “Scientific Notes of the Gorky State University named after N. I. Lobachevsky,” issue 52, historical-philological series, Gorky, 1961, p. 343. 25 Cf.: “For us to speak of a derivational relationship, there must be not only a phonetic but also a semantic relation...” (H. Marchand, Esquisse d’une description des principales alternances dérivatives dans le français d’aujourd’hui, “Studia linguistica,” V, 1951, No. 2, Lund, p. 95); “Two conditions must be met for a word to be called a prefixal derivative. First, it must be possible to analyze it phonetically and semantically. This means that derived words must be related to the base word in two ways: through the signified and the signifier. Second, that relationship must also characterize the group” (H. Marchand, The Question of the Derivative Relevancy and the Prefix s- in Italian, ibid., VII, 1953, No. 2, p. 104.) 26 The latter principle is called the principle of commutation and, following Ferdinand de Saussure, is formulated as follows: different signifiers must correspond to different signifieds and, conversely, different signifieds must correspond to different signifiers (see C. K. Shaumyan, On the Nature of Structural Linguistics, Voprosy yazykoznaniya, 1956, No. 5, p. 48). 27 See V. N. Toporov, cited work, p. 3. 215 and so on, of the name). Even if the place-name *Duobė is no longer found anywhere nearby, we still cannot assert with certainty that Duobaitė is derived from duobė (or is a diminutive physiographic term duobaitė used figuratively), since such a place-name may have existed earlier. This is especially important to remember when this type of formation is more widespread in the territory in question. Among derivatives, most are hydronyms whose formation is unclear, but which can be morphologically segmented (into morphemes). There are also quite a few hydronyms whose formation is unclear and whose segmentation into morphemes can only be conjectured. With this in mind, hydronyms containing identical affixes can be classified accordingly. It seems most convenient to distinguish three categories (types, groups) of hydronyms: 1. Hydronyms whose formation is clear, i.e., those whose derivational relationship with the corresponding base word is indisputable, e.g.: Agarinis lake: agaras ‘water chestnut (Trapa natans)’, Akmenė river: akmuo ‘stone’, Daubilis river: dauba ‘ravine’, Dumblinis river: dumblas ‘mud’, Vidaujàitė river: Vidauja river, etc. 2. Hydronyms whose formation is unclear, but which, when compared with other Baltic hydronyms or place-names more generally, may be considered to contain an affix. For example, comparing the river name Mauč-iadis (Zietela, right bank of the Nemunas) with Mauč-iuvis (Vaškai, left bank of the Y sly kio), one may surmise that mauč- is the root, while -adis and -iuvis are suffixes; cf. also Alm-a-jà river (Linkmenys, tributary of the Srovė) and Alm-ė river (Šiluva, tributary of the Dubysa), alm-ėti ‘to flow, drip gradually’; alm-auō ‘pus’; Ving-ainis||Ving-ainis river (Kvėdarna, tributary of the Jūra) and Ving-is river (Kamajai, tributary of the Setekšna), ving-is, ving-iuoti ‘to meander’. Depending on how clearly or unclearly the base stem can be identified, the hydronyms in this group may be more or less transparent, for example (respectively): Alm-ajà river, Mauč-iadis river. 3. Hydronyms whose formation is unclear and for which even the identification of an affix is highly doubtful, although the hydronym’s structure, stress placement, intonation, comparison with other proper or common nouns, and other circumstances suggest that it may contain a particular affix; for example, Žand-agas river (Krekenava, tributary of the Vadaktis). For žand- compare (?) Žand-ù-gala river (Seduva, tributary of the Kriaukė), Žand-k-ys lake (Giedraičiai, Padvariai village). However, the river name Žandagas may also be a compound, formed from Žan- and -dagas. Thus, the basic principle governing the formation of common nouns—that a derivative can be posited only when and for as long as the corresponding base word exists28—is preserved only in the first group. At the same time, the first group is the most precise, since it involves the least subjectivity. The second group is less precise, since a hydronym regarded as suffixal will lack the base word from which it is derived, or the base word will not be entirely clear. Consequently, when comparing two hydronyms with the same root, there is always a risk of error. This risk arises because morphemes that sound identical today may have different origins. 28 See Г. О. Винокур, op. cit., p. 425. 216 For instance, there is now no way to prove precisely that the morphemes Mauč-iuvis and Mauč-iadis in the river names mauč- are of common origin. On the other hand, it is also difficult to prove that they have different origins. In such cases, comparing one place-name with another is customary, generally accepted, and possible, although there is a serious risk of error. Here it should also be emphasized that, in trying to enlarge the first group at the expense of the second—that is, by seeking common or proper words from which hydronyms may have arisen directly—we would risk making a mistake more than if we compared hydronymic morphemes that are formally identical today. Even with hydronyms whose formation seems quite clear at first glance, such as the river name Ving-ainis||Ving-ainis (Kvėdarna, tributary of the Jūra), we must proceed very cautiously, since it cannot be said with complete certainty from which common word—vingis or vingiuoti—this river name was formed, although it is known that common words with the suffix -ainis (-ainia-) are most often formed from nouns, adjectives, and numerals. Indeed, there are quite a number of cases in which hydronym formation was governed by laws different from those governing common words. It is important here to remember that the formation of hydronyms was strongly influenced by the toponymic context. This influence often resulted in various affixes appearing where, according to the laws of common-word formation, they would be impossible. For example, the Lithuanian name inventory includes rivers ending in -oja: Ilg-oja, Šalt-oja, Švent-oja, etc. In origin, these are simply definite adjectives²⁹ that later became nouns³⁰, cf. ilg-as, šált-as, švent-as. Owing to the influence of hydronyms and various other place-names, particularly the abundant microtoponymy with -oja, the formant -oja sometimes appears where its presence would be entirely impossible under the laws of common-word formation, e.g. Lank-oja river (Obeliai, Audroniai village) and Sirvint-oja||Sirvinta (Sirvintos, tributary of the Šventoji). Therefore, the regularities of common-word formation cannot be applied unconditionally to toponymy, since within the language system toponymy forms a microsystem in which its own, sometimes very strict, laws operate. On the other hand, when justifying the presence of an affix in a hydronym, the abundance (regularity) of derivatives with a given affix is also very important. There is a group of river names (about 10) with -aja* (Alm-aja, Indr-aja, Kam-aja, Lak-aja, etc.). Comparison with other hydronyms always makes it fairly clear that -aja is a suffix. Only the river name Usl-aja stands apart from the others. No other place-names or common words with the root usl- have been found in the Baltic languages. However, given the regularity of the suffix -aja, its geographical distribution (it is not far from the area where river names in -aja are found), and other secondary considerations (for example, the existence of a hydronym with a root that may reflect another ablaut grade, cf. Aušl-as lake, Degučiai, Auslo, etc.), there seems to be no reason to separate it from the other river names with this suffix. 29 Regarding -oja instead of the literary-language form -oji see Z. Zinkevičius, Lietuvių kalbos įvardžiuotinių būdvardžių istorijos bruožai, Vilnius, 1957, p. 90. 30 On the substantivization of definite adjectives, see A. Valeckienė, Dabartinės lietuvių kalbos įvardžiuotinių būdvardžių vartojimas, „Literatūra ir kalba“, t. 2, Vilnius, 1957, p. 239. 217 Moreover, when assigning a hydronym to the second or third group, one must consider not only the regularity of the affix but also whether that affix occurs in common words. For these reasons, the second hydronym group is usually the largest and most important of the three. It includes not only some hydronyms that are etymologically clearer, often of later origin, but also most ancient hydronyms, which are very important for illuminating ethnogenesis, language genesis, and their mutual relations. It may also be added that for these purposes it is often immaterial what a hydronym is derived from (e.g., whether Vingains derives from vingis or vingiuoti); it is enough that the Baltic root (ving-) and suffix (-ainis) are clear. The third group, which includes hydronyms of unclear formation in which even the presence of an affix is highly doubtful, is also quite large. It is difficult to avoid a certain degree of subjectivity when assigning a hydronym to one group or another. However, if we are to make progress, first of all we must know what is clear at present and what is not; classifying hydronyms into three groups should serve precisely this purpose. Moreover, in dividing hydronyms into the groups mentioned, no impermeable boundaries are intended between hydronyms that have or may have the same affixes; the aim is only to distinguish clearer cases from less clear ones. Given that relatively few hydronyms have a clear formation, the term derivatives is not entirely suitable for hydronyms in all three groups. It is used provisionally, since most hydronyms of unclear formation can only tentatively be regarded as derivatives. The derivatives are then divided into three large groups: a) suffixal derivatives; b) ending derivatives; c) prefixal derivatives. a) S u f f i x a l d e r i v a t i v e s. Among derivatives, suffixal hydronyms are the most numerous. They clearly outnumber ending derivatives and greatly exceed prefixal derivatives. For example, more than four hundred river names formed with suffixes can be counted. In examining suffixal derivatives, the first issue to arise is how to arrange and classify the suffixes. In studies of Lithuanian word formation, it has hitherto been customary to classify suffixal derivatives according to the consonant of the suffix, i.e. to speak of “suffixes with the consonant b, with the consonant d, with g, with j, with k, etc.” This classification has its advantages and disadvantages. It is more convenient in studies of word formation that address not only how words are formed but also the origin of suffixes. In the Baltic languages, suffixes of the same origin sometimes have different vocalism today (e.g. -uoni-, -ūni-, -uona-, -ūna-, -uonio31), so it is more convenient to rely on the more stable element of the language—the consonants. But even in works of this kind, classifying by consonant presents various difficulties. First of all, there are many suffixes containing more than one consonant. It is often difficult to decide which consonant in such suffixes is more important and can serve as the classificatory marker. For example, Pr. Skardžius several 31 K. Būga, Rinktiniai raštai, II, Vilnius, 1959, p. 368, 366. 218 He classified suffixes containing consonants according to the final consonant. In many cases, such a division is difficult to justify not only from a synchronic but also from a diachronic perspective. It is now not entirely clear why, for example, the very old suffix -mena (cf. sėmeną “small fallow plot,” Akmeną river) should be assigned to suffixes with the consonant n, and not to those with the consonant m. Apparently, the same can be said of the suffix -estis. Because of this ambiguity, researchers of Lithuanian word formation have not all classified suffixes in the same way. For example, A. Leskynas³² assigned the suffixes -mena, -menė, -gana, -tuvís, -tuvė, -tuva, -taua not to the suffixes with n, v, as Pr. Skardžius³³ did, but to those with m, g, t. Since the subject here is word formation, not the origin of suffixes, classifying suffixes solely according to their consonant would be inconvenient, first of all because we would not know where to assign suffixes containing several consonants (and these make up perhaps a quarter of all hydronymic suffixes). By classifying suffixes according to consonant, we would, as it were, single out a consonant from the suffix as an element of particular importance. In the study of word formation, however, what matters is not one individual element of the suffix but the suffix as a whole. In the analysis of common-word formation, the derivational meaning of derivatives plays a very important role. Sometimes several suffixes with the same consonant but different vocalism modify the meaning of the base word in the same way, e.g.: -ūnas, -uonas, -is (palaidūnas — palaiduonas, atskalūnas — atskaluonis)34. The formation of the great majority of hydronyms is unclear, and for this reason the criterion of classifying suffixes by consonant is of little importance in the analysis of hydronym formation. The most acceptable approach appears to be a formal classification of suffixes alphabetically. In essence, this is no classification at all, but merely a certain arrangement and ordering of all the suffixes. b) D e r i v a t i v e s w i t h e n d i n g s. Derivatives with endings are derived hydronyms in which the derivational marker is the ending of the derivative (e.g., -ė in the river name Trūmpė, formed from trumpas ‘short’; -ys in the lake name Ilgýs, formed from ilgas35). There are more derivatives of this type than prefixal derivatives, but considerably fewer than suffixal derivatives. Moreover, derivatives with endings are less varied than other derivatives. Their number can only be estimated approximately, since it is often unclear whether a river name can or cannot be derived from an appellative (e.g., it is now difficult to say whether the river name Žibà can be derived from žibėti, the river name Aitrà from aitrùs, Alýs from alėti, and so on). Therefore, when examining derivatives with endings, the first issue is to distinguish clear cases from those that are not entirely clear. In such cases, one should proceed as with suffixal derivatives: a hydronym should be considered a derivative only when its derivational relationship with the corresponding base word is evident. 32 A. L e s k i e n, Die Bildung der Nomina im Litauischen, Leipzig, 1891, p. 417—421; 526; 565—567. 33 See Pr. S k a r d ž i u s, Žodžių daryba, Vilnius, 1943, p. 232; 226; 382—384. 34 K. B ū g a, Rinktiniai raštai, II, p. 366. 35 For more on derivatives with endings, see V. U r b u t i s, Dabartinės lietuvių kalbos galūnių darybos daiktavardžiai, „Kalbotyra“, vol. 3, Vilnius, 1961, pp. 27, 28. 219 c) Prefixal derivatives. These derivatives are the least numerous. There are several times fewer of them than suffixal derivatives and considerably fewer than derivatives with endings, although there are more prefixes than endings. Prefixal derivatives should be classified according to the same principle as suffixes, that is, formally and alphabetically. ON THE WORD-FORMATION ANALYSIS OF HYDRONYMS А. ВАНАГАС Abstract The formation of the system of Lithuanian hydronyms can now be considered virtually complete. From the perspective of word formation, there are grounds for regarding it as no longer developing to any significant extent, but as stable. In terms of its stability, it is to some extent similar to the systems of appellatives and toponyms in dead languages. This applies entirely to those hydronyms which, owing to the disappearance of the objects they designate, have in practice ceased to be elements of a living language. The stability of the system is one of the characteristic features that distinguish hydronymy from the system of appellatives. Therefore, the aim of the word-formation analysis of hydronyms is not to determine ways of forming new words or replenishing a language’s vocabulary, but to establish the set of means and methods that participated in the formation of what is now an already static system. The article addresses two questions of a word-formation nature: 1) the classification of hydronyms; 2) the problem of derivatives in hydronymy. 1) The principal question in toponomastics—the question of classification—has so far received virtually no attention in Baltic linguistic literature, although further research in this field is inconceivable without resolving it. The classification of Lithuanian river names proposed by the author (see the scheme on p. 213) is intended as a first attempt to address this pressing question, at least in part. Our classification is based on the idea of the structural-grammatical classification developed by S. Rospond and, in part, P. Lebel. (The latter not only developed a classification of French hydronymy but also applied it fully in practical research.) It must be emphasized, however, that the classification scheme we propose differs substantially from S. Rospond’s scheme. First, these differences are objective in nature, since the schemes are based on different material. Second, most of the differences are dictated by considerations of logical consistency. 2) The problem of derivatives invariably arises in the word-formation analysis of hydronyms. Since hydronyms are not homogeneous in terms of word formation, they can be divided into three groups: a) hydronyms that can undoubtedly be considered derived, i.e. whose derivational relationship with the corresponding common or proper words is clear; b) hydronyms that cannot be considered derived because their derivational relationship with the base from which they were formed is unclear, but which can be analyzed morphologically (divided into morphemes); c) hydronyms that cannot be considered derivatives and can only tentatively be divided into morphemes. In addressing the classification of affixes themselves, the author advances the view that the traditional classification according to the consonant in the affix (e.g., suffixes with the consonant d, with the consonant g, with the consonant i, with the consonant k, etc.) is not justified in hydronymy. The author proposes classifying affixes formally, in alphabetical order (e.g., suffixes with -ad-, -ag-, -ain-, -ait-, -aj-, -ak-, -al-, -am-, -an-, -ar-, -as-, etc.). 220