Toponymization in the Latvian Language

Straipsniai / Articles 117 SANDA RAPA Latvian Language Institute, University of Latvia Fields of research: onomastics, lexicology, grammar. TOPONYMIZATION IN THE LATVIAN LANGUAGE Toponimizacija latvių kalboje ANNOTATION This article deals with grammaticalization processes in the Latvian toponymy and tries to find differences in the derivation of toponymy and common vocabulary. Three types (primary, secondary and compounding toponymization) and six stages of toponymization (extension or the use of a word in new extensions, transcategorization or the change of category, decategorization or the loss of some categorical components, desemanticization or the loss or change of semes in the semantic field of a place name, clitization or merging of two words into one, erosion or the loss of phonetic substance, epexegeticalization or the acquisition of a new morphological or semantic extension in order to explain the place name) were observed. Two stages of desemanticization – partial and complete – have been distinguished. KEYWORDS: place names, toponymisation, desemanticization, grammatical changes of toponym. ANOTACIJA Straipsnyje aptariami latvių kalbos vietovardžių gramatizacijos procesai ir mėginama nustatyti vietovardžių ir bendrinių žodžių darybos skirtumus. Buvo išskirti trys toponimizacijos būdai (pirminė, antrinė ir sudėtinė toponimizacija) ir šeši lygiai (ekstencija, arba žodžio vartojimas kitame kontekste, transkategorizacija, arba kategorijos pasikeitimas, dekategorizacija, arba tam tikrų kategorinių komponentų iškritimas, desemantizacija, arba žodžio semų iškritimas ar pasikeitimas semantiniame vietovardžio lauke, klitizacija, arba dviejų žodžių sujungimas, erozija, arba fonetinių elementų iškritimas, epeksegetizacija, arba naujo morfologinio ar semantinio komponento, paaiškinančio vietovardį, įgijimas). Išskirti du desemantizacijos lygiai: dalinė ir visiška. ESMINIAI ŽODŽIAI: vietovardžiai, toponimizacija, desemantizacija, vietovardžio gramatiniai pakitimai. SANDA RAPA 118 Acta Linguistica Lithuanica LXXIII By gaining the status of a proper name the common noun does not stop its development – it is still being adapted to the language system and grammar. Grammarians normally call it grammaticalization (or grammatization) – the process whereby lexical items and constructions come to serve grammatical functions in certain linguistic contexts, and, once grammaticalized, continue to develop new grammatical functions (Hopper, Traugott 1994: XV). But, bearing in mind that proper names are a specific category of vocabulary, and thus have different derivational models and perhaps also a different further development, the process of word adaptability to the language system in onomastics should be referred to as toponymization. There are two opinions about toponymization held by onomasticians in reference to the completeness of the process. The United Nations Group of Experts on Geographical Names (UNGEGN) provides the following definition of the term: “act of producing a toponym (i.e. a proper noun) from a common noun or other part of speech” (Report 2011: 3). In Russian onomastics, the process of toponymization is explained in a more detailed way: as a type of onymization which implies not only the change of a common noun into a proper noun, but also the further grammatical and lexico-semantic development of the name (Podolskaja 1978: 143); it is a process which is most likely to occur in inflectional languages. Based on this broader understanding of toponymization, the article investigates some aspects of the grammatical development of Latvian place-names. In order to explain the process of toponymization in the Latvian language more explicitly, the terminology and methodology of grammaticalization will be used (Hopper, Traugott 2003; Heine, Kuteva 2004; Heine 1993; Fischer, Rosenbach 2000; Lehmann 2002; Meillet 1958, etc.). The material for analysis – place names with a clear origin (etymon) – was taken from the card files of the Latvian Language Institute. On the first level – when a common noun becomes a proper name – three types of toponymization can be distinguished (according to the classification elaborated by the Polish linguist Stanisław Rospond and adopted by Latvian and Lithuanian onomasticians (Rospond 1957: 235ff., Vanagas 1970, Balode 1985, Laumane 1996)): • primary toponymization – when the common noun obtains the status of a proper name without any morphological transformation (morphological derivation should take place in the appellative rather than in the proper name, e.g., a valley called Abra ‘kneading trough, hod’1 in Baižkalns < 1 Here and henceforth – an approximate or literal translation of a place name. Straipsniai / Articles 119 Toponymization in the Latvian Language abr-a [root + ending] and a narrow valley Abriņa in Cēsis < diminutive of abra – abr-iņ-a [root + diminutive suffix + ending]); • secondary toponymization – when the common noun is derived by topoformants (morphological derivation should take place in the proper name, e.g., Celmājs ‘a place with many stumps’ in Izvalta < celm-āj-s [root + topoformant i.e. suffix + ending]), • compounding toponymization – when the proper name consists of two or more words (sometimes it can be accompanied by secondary toponymization). Most compound place names are nominal constructions – usually a generic element in combination with another noun (89%) or adjective (10%). Other examples are very rare – prepositional, coordinative, numeral, pronominal, and interjectional constructions constitute only 1% of compound place names. The review of more than 100,000 place names from the card files of the Latvian Language Institute shows that one half of all Latvian place names (50%) is derived by compounding toponymization, 48% are formed by secondary toponymization, and only 2% are primary place names (see Picture No. 1). PICTURE 1. Three types of toponymization However, the derivation of place names is not the main object of this paper – the morphological structure of Latvian place names has already been thoroughly investigated by both Latvian (e.g., Laimute Balode, Vallija Dambe, Benita Laumane, etc.) and Lithuanian (e.g., Aleksandras Vanagas, Laimutis Bilkis, Dalia Kačinaite-Vrubliauskiene, etc.) linguists. In this study, the main processes of deeper toponymization (i.e. grammaticalization) in the Latvian language will be outlined. In order to show more explicitly the further and deeper toponymization in Latvian place names and to compare the development of proper vs. common Secondary toponymization Compounding toponymization Primary toponymization Primary toponymization S e c o n d t o p o n y d i n g z a t i o n SANDA RAPA 120 Acta Linguistica Lithuanica LXXIII vocabulary, the following scheme elaborated by Bernd Heine (Heine 1993: 58) was selected: desemanticization → decategorization → clitization → erosion This chain of grammaticalization shows the basic stages of word grammaticalization without deeper fragmentation of the process. Since place names usually maintain their grammatical form for a long period, evidence of toponymization should be found without a deeper analysis. Furthermore, each stage of grammaticalization will be described in terms of toponomastics. DECATEGORIZATION When trying to put a place name in the grammaticalization chain, the first correction should be done in the sequence of stages because in the process of onymization the category of the word is affected at the very beginning. When a common noun shifts to a proper noun, it not only acquires a graphical distinction (namely, it comes to be spelled with a capital letter), but also freezes in a certain (and often unchangeable) categorial or grammatical form. Thus, it loses at least one categorial component (for example, the nouns that become oikonyms in Latvian toponymy most frequently lose their singular form, while adjectives, as they become oikonyms, usually lose their feminine form). However, this shift between the appellative and the proper name never leads to a lower-level category (e.g., auxiliaries) – common nouns transform within an equivalent lexical category; therefore, the first stage of toponymization should be called transcategorization or recategorization. Some grammarians have discerned another sub-stage in this process before decategorization and described it as extension – the use of the word in new extensions (Heine, Kuteva 2004). Although they later abandoned this idea, it may be applied to onomastics. Thus, the first stage of toponymization may comprise the following: extension (use in new extensions) > transcategorization (change of category) > decategorization (loss of some categorial components). DESEMANTICIZATION The next stage – desemanticization – is sometimes referred to as the trigger of grammaticalization which initiates an irreversible change of a language unit, and is responsible for all the following stages of grammaticalization. If we agree with John Stuart Mill, Alan Gardiner and others who claim that proper names have no lexical meaning and are empty differentiating signs, we should say that Straipsniai / Articles 121 Toponymization in the Latvian Language common nouns reach their complete desemanticization at the point when they become proper names. On the other hand, if we agree with their opponents, such as Otto Jespersen, Ojārs Bušs and myself, we can say that proper names have a lexical meaning which is even broader and encompasses more than common nouns do, and thus the naming of a place initiates not only desemanticization but also resemanticization of the appellative. For example, the appellative ērgļi ‘eagles’, after it became a proper noun Ērgļi, lost its appellative meaning and gained a new meaning which is, however, difficult to define: ‘a village on the bank of the Ogre river, the birthplace of the Latvian writer Rūdolfs Blaumanis, etc.’. In the card files of the Latvian Language Institute there are many instances which show place name desemanticization. Two different levels of desemanticization in the Latvian toponymy can be distinguished: partial and complete desemanticization, which are both usually initiated by metaphor or metonymy, i.e. analogy and re-analysis (these two types of semantic transposition are recognized as the most important semantically motivated mechanisms in the process of grammaticalization (Hopper, Traugott 2003: 87, Heine 1993: 96, Fischer, Rosenbach 2000: 15, etc.)). Partial desemanticization can be found in compound place names with a generic element which does not (fully) characterize the category of the object. The study of approximately 100,000 compound place names with a generic element (which should, of course, denote the category of the object) shows more than 24,000 semantic mismatches between a place name and the real geographical object – which means that more than a fifth of generics in Latvian compound toponyms do not describe the category of the real geographical object, or only partially do so (see Picture No. 2). generic term ≠ direct object name 24% generic term = direct object name 76% PICTURE 2. Generics in Latvian place names SANDA RAPA 122 Acta Linguistica Lithuanica LXXIII It could be a result of either functional change or metaphoric and metonymic transposition. But even partial desemanticization can be of a different degree: metaphorically transposed names partially break the link between the appellative and the object, but metonymically transposed place names and names which remain unchanged after the change of the object do it almost completely, while the lexical meaning of the appellative is still recognizable (for instance, a frequent metaphoric designation for a hollow place, bļuoda ‘lit.: bowl’, used in the valley name Sidraba Bļoda ‘Silver Bowl’ in Anna2, and the appellative bļoda ‘bowl’ have at least one common seme ‘round shape’, while the generic term in the meadow name Medus Strauts ‘Honey Rivulet’ in Džūkste which is named, through synecdoche, after the rivulet which flows along the meadow, and the appellative strauts ‘rivulet’ do not have any common seme). The most representative example of partial desemanticization is the appellative purvs ‘bog’ in Latvian place names. About one sixth of genitive compounds with this generic term existing in Latvian toponymy do not designate the corresponding physiographical object, i.e. a real bog or swamp, or at least a wetland. Lexeme purvs in Latvian place names is often used to designate meadows, forests, fields, ponds, bushes, villages, valleys, pits, etc. This appellative (probably, due to changes in the flora and geology of the respective objects) has undergone all types of semantic changes. An object may be called a bog through: • metaphor (when the place looks like or has some features of the respective object, e.g., a meadow name Elles Purvs ‘Hell’s Bog’ in Ārlava; characterized by an informant as “very wet”); • metonymy, i.e. synecdoche (the place is situated near a bog or a bog is inside its territory, e.g., a forest name Līkais purvs ‘Crooked Bog’ in Brenguļi; “has some little bogs inside”); • historical changes (when the territory has earlier been a real bog, e.g., Laivas purs ‘Boat Bog’ in Bilska, “was a huge bog, now ameliorated land”). The same fate has befallen other generics which are included in about a half of all Latvian place names. Probably, a bleaching of the appellative meaning of a place name is also determined by the frequency of its use. It can be observed in the use of the most widespread appellatives such as aste ‘narrow territory’ (lit.: ‘tail’), gals ‘end’, kalns ‘hill, kāja ‘narrow territory’ (lit.: ‘leg’), leja ‘valley’, lauks ‘field’, mala ‘edge’ mežs ‘forest’, upe ‘river’, vidus ‘middle’, zeme ‘land, soil’. There are several indications of semantic bleaching of the most widespread appellatives in place names: 2 Here and henceforth the place name and its literal translation is followed by the name of the parish (administrative entity after 1939) where the place name was registered. In some cases, some additional information given by an informant in quotation marks is also quoted. Straipsniai / Articles 123 Toponymization in the Latvian Language a) they are often used in the genitive (usually singular) form: e.g., kalna-, lejas-, lauka-, meža-, vidus-, and have acquired a particular meaning which is used only in onomastic compounds – in most cases, to differentiate a new farmhouse from an older one (e.g., farm names Dribas, Kalna Dribas ‘Hill Dribas’, and Lejas Dribas ‘Valley Dribas’ in Baižkalns, Kalna Uodiņi in Drabeši, as the informant said, “were separated from Uodiņi”); b) usually they have lost some phonetic substance: kalna- > kal-, lejas- > lejs-, lauka- > lau-, vidus- > vids- (e.g., farm name Kalmaļi [< kalns ‘hill’ + mala ‘border’] in Vārve, village name Laugals [< lauks ‘field’ + gals ‘end’] in Nīca); c) they are often used in a particular morphological form which does not have a specific meaning and cannot be found in the common vocabulary: gals > -gale [derivation with a feminine ending], -gali, -gaļi [nom. pl.]; mala > -male [feminine], -maļi [nom. pl.]; upe > -upi, -upji [ nom. pl.]; vidus > -vidi [nom. pl.]), zeme > -zemi, -zemji [masculine, nom. pl.] (e.g., Mežvidi [< mežs ‘forest’ + vidus ‘middle’] in Dignāja, Ērģeme, Gaujiena, etc., Ciemgali in Dundaga, Ciemgaļi [< ciems ‘village’ + gals ‘end’] in Bauņi, Turlava, etc.); d) they sometimes contain a reduplication which is considered by Lehmann (Lehmann 2002: 116–117) to be a reliable indicator of desemanticization: e.g., hill names with the reduplicated generic kalns ‘hill’: Kalniņkalns, Kalnijkalns, farm names Kalnkrievkalns, Kalnkalniņi in Ērgļi; e) sometimes an additional connotation can appear: e.g., leja ‘valley’ can also denote ‘a farm servant’s house’ and kalns ‘hill’ – ‘a farm owner’s house’ (ME II 143, 447) (e.g., Lejasvērzemnieki ‘Valley Vērzemnieki’ in Cirgaļi described by an informant as “the servants’ house”). These indications show that the most widespread lexemes have almost acquired the status of an affixoid (in most cases, a prefixoid) – a regular word which can also function as a morphological component of a name, as an affix (prefix). Even if an independent word loses its phonetic substance and becomes unrecognizable, it may still be used as a topoformant in derivation of new place names. Sometimes a coincidence with regular auslauts can help (for example, a frequent appellative aste ‘tail, narrow territory’ is reminiscent of auslaut -a(i)ste or suffix -stin Baltic Finnic languages). Many lexemes in place names have reached complete desemanticization. This phenomenon can be observed in the so-called obscure place names where the specific meaning has been long forgotten and now can be discovered only through etymological analysis. For example, the frequent ending -ava in Latvian hydronyms (e.g., in river names Abava, Daugava, Rucava) has probably arisen from Indo-European root *a(e)- and originally had the meaning ‘to moisten, to flow’ (cf. Latv. avots ‘spring [of water]’) (Karulis I 93–94; Buck 1949: 44; Pokorny 1959: 78). SANDA RAPA 124 Acta Linguistica Lithuanica LXXIII There are many topoformants which have entered Latvian toponymy as independent words from other languages. Due to subsequent unrecognizability, these borrowed appellatives (mainly generics) have become grammatical formants. Thus, Latvian lake names Kaņieris in Riga, Ninieris in Cēsis, Spicieris in Smiltene probably contain Estonian järv ‘lake’ or Livonian jōra, järu ‘lake’ (Dambe 1987: 34; Hauzenberga 1932: 134). This ending has become a regular topoformant; therefore, it is included not only in hydronyms, but also in meadow names (e.g., Piniers in Liezēre) and farm names (e.g., Poņeri in Piltene). Borrowed postfixoids are also -nīte in Kurzeme region (cf. Estonian niit, Livonian nīt ‘meadow’) and -pole in Latgale (cf. Russian поле ‘field’ which, according to Fasmer III 307, is related to Old Russian полъ ‘open, empty’), etc. Not only desemanticization but also transsemanticization or resemanticization can be observed in Latvian place names if a new meaning appears as the result of grammaticalization. For example, farm name Rīskalni in Renda has arisen as a result of contamination from rijas- [gen. sg.] ‘threshing barn’ and kalni ‘hills’, not from the words rīss ‘rice’ and kalni ‘hills’; farm name Dardedži has arisen from dar(v)dedži ‘tar-boilers’, not from dardedze ‘rainbow’; farm name Sālskalns – from saules kalns ‘hill of the sun’, not from sāls kalns ‘hill of salt’, farm name Piesargi means pievsargi ‘watchmen of meadow’, not ‘near/at Sargi’. In some cases, it is probably due to folk etymology, not grammaticalization, but the reason is still desemanticization. CLITIZATION The most widespread trigger of affixoids is clitic, when one word in rapid speech blends with the next word and phonologically depends on another word. It does not necessarily follow desemanticization, as it was shown in the grammaticalization chain, because here the component of a toponym usually keeps its meaning – it is always recalled by the connotation with the corresponding object. The most characteristic process in Latvian toponymy is proclitic when a word (usually a preposition) without an independent accent merges with the following word in the process of speech. There are still many place names in Latvian toponymy which resemble an onomastic description. Most of them are prefixal constructions, and many of them are combined in proclitic words: Aizezere < aiz + ezers ‘behind the lake’, Aizupe < aiz + upe ‘behind the river’, Aizpurve < aiz + purvs ‘behind the bog’, Pierobeža < pie + robeža ‘by the border’, Pārdaugava < pār + Daugava ‘over/across the river Daugava’, pārmežs < pār + mežs ‘behind the forest’, Pārpurvs < pār + purvs ‘behind/over the bog’. In these Straipsniai / Articles 125 Toponymization in the Latvian Language constructions (more often than in the common vocabulary), the accent shifts from the independent word to the prefix (according to the rule of the typical first syllable stress in Latvian). At the same time, the new lexeme adapts (by taking the respective derivational ending) to the corresponding denotatum: for example, a meadow name Pieakmene has become feminine, because it refers to a feminine noun pļava ‘meadow’, while a fishing place Pie akmeņa still contains the masculine ending of the noun akmens ‘stone’ in Jumurda. An enclitic, where a word phonetically merges with the preceding word, seems to be rather infrequent in Latvian toponymy. However, it is also less visible and recognizable (because, as Christian Lehmann has noted, it is often followed by agglutination and fusion (Lehmann 2002, 145ff.)). Frequent topoformants which have arisen from independent words ezers ‘lake’ and arājs ‘plowman’ in Latvian toponymy are, for example, auslauts -eris or -ars and -rāji - e.g. in the names of settlements Meldzere in Nīgrande and Sauzeri in Stūri, meadow name Melzere in Skrunda, lake name Zuosars in Gaigalava, and farmhouse names Dziļrāji in Code, Jaunrāji in Padure, Muižrāji in Īvande – these place names probably have earlier been Meld-ezere (< ‘reed lake’), Saus-ezeri (< ‘dry lake’), Mel(n)-ezere (< ‘black lake’), Zuos-ezers (‘lake of geese’), Dziļ-arāji (< ‘deep plowing plowmen’), Jaun-arāji (< ‘young/new plowmen’), Muiž-arāji (< ‘manor plowmen’). Besides clitics, different kinds of phonological changes can affect place names. The Latvian linguist Edīte Hauzenberga-Šturma has described more than 150 types of phonological changes in approximately 2000 Latvian oikonyms, including loss of vowels, consonants, or syllables, as well as qualitative vowel changes, epenthesis, assimilation, dissimilation, metathesis, alliteration, etc. (see Hauzenberga 1932). EROSION Erosion or loss of phonetic substance (Heine, Kuteva 2002: 2) always goes hand in hand with the process of clitic and usually follows desemanticization (Meillet 1958: 135–139). Commonly, a toponymic collocation (onomastic lexeme + generic) becomes a compound place name, and then the second component of the name becomes a topoformant (as we could see in the enclitic examples above), but often the last component (mostly a generic) is completely dropped. Probably, it is due to language economy which is more effective in toponymy than in common language, because in the case of toponymy one has to memorize many specific, often non-standard language units (Superanskaja 1973: 87). Besides, the more evident loss or the so-called toponymic ellipsis