LIETUVIŲ KALBOTYROS KLAUSIMAI, XXXIV (1994)

LIETUVIŲ KALBOS TARMĖS IR JŲ TYRINĖJIMAI

Albertas ROSINAS

THE HIERARCHY OF NOUN DECLENSION TYPES IN SIMONAS STANEVIČIUS’S TEXTS

More than 15 years ago, the idea was put forward in linguistic literature that Lithuanian noun declension, in both synchronic and diachronic studies, is examined according to the traditional order of Lithuanian declensions, and that this order is entirely unmotivated and arbitrary, since it takes into account neither noun gender nor the similarity of forms ¹.

The existence in the language system of not only an arbitrary but also a natural order of declension types is demonstrated by diachronic studies, which show that only words with certain stems influence one another: a-stems influence u-stems, įa- and ė-stems influence i-stems, į- and partly įa-stems influence consonant stems, and so on. However, for example, not the slightest influence of ā- and ē-stems on u- or įu-stems has been observed. Those declension types between which there is diachronic and synchronic interaction must be adjacent, or at least close to one another, in the language system [Girdenis, Rosinas 1977, p. 338].

Given this premise, it becomes clear that the most reliable criterion for determining a natural classification of declension types and their mutual hierarchy is the frequency of shared endings.

Endings are considered identical not only when they are wholly alike, but also when they differ through automatic phonological alternation a: e, ā: ē, as well as endings that arose through systematic shortening and have an “older” and a newer phonetic variant (e.g., the plural dative endings in S. Stanevičius’s texts for nouns with i)ā- and ē-stems, -uoms (written -om’s) and -ums) ².

With this in mind, I chose the number of shared endings (m) in adjacent paradigms (“stems”) as the principal indicator of declension types for investigating the hierarchy of noun declension types in S. Stanevičius’s texts. For example, the paradigms of the common-language nouns spiečius and papūpys have 8 cases in common—thus, the indicator of similarity between these paradigms (m) is equal to 8. However, such a simple index would not be entirely accurate, because: 1) endings such as the plural genitive are common to all declension types; 2) the singular vocative varies considerably, depending on particular words or even situations; 3) the frequency of not all cases in the text (together

¹ Plačiau žr. [Girdenis, Rosinas, 1977, p. 338].

² In this study, endings are understood in the traditional way—they are not divided into the stem-final element (“pre-ending”) and the “true” ending. For the possibility of such an interpretation, see [Girdenis, Rosinas, 1977, p. 340].

112

and weight in the system) is the same; the frequencies (and weights) of such central cases as the singular nominative, genitive, and accusative cannot be compared with those of such a peripheral case as the plural locative. The first and second inaccuracies can be avoided by eliminating from the paradigms the cases that are anomalous in this respect—the vocative and the plural genitive. The third inaccuracy can be mitigated by introducing a certain adjustment (pᵢ), which directly or indirectly reflects the frequency of syncretic cases in connected text [Girdenis, Rosinas, 1977, p. 339].

In examining the hierarchy of noun declension types in S. Stanevičius’s texts, after various attempts I determined that, for singular cases, the role of such an adjustment can be performed by a fraction approximately equal to twice the relative frequency of the corresponding cases in connected text, while for plural cases it can be performed by half the corresponding adjustment for singular cases. In addition, I rounded the raw case-frequency data substantially (see Table 1).

Table 1

Adjustments for the similarity index of declension types in S. Stanevičius’s texts

Singular Plural Nom. 0,50 0,20 Gen. 0,60 — Dat. 0,10 0,05 Acc. 0,50 0,25 Instr. 0,15 0,05 Loc. 0,15 0,05

These adjustments are added to the index indicating the identity of the endings of two cases belonging to different stems.

In abbreviated form, these calculations can be described by the formula

I = m + ∑ pᵢ,

where m is the number of matching case endings, and pᵢ is the adjustment for a particular case.

Case-frequency calculations were carried out using the following texts by S. Stanevičius: ”Szeszes pasakas Symona Stanewiczes” (Wylniuje, 1829, with preface, 16 pp.), ”Daynas žemaycziu” (Wylniuje, 1829, 63 pp.), ”Apey darima walge ysz kiarpiu islandu” (Wyjlniuy, 1823, 22 pp.) and ”Historyia Szwenta” (Wylniuje, 1823, 70 pp.). The texts comprise 171 pages.

Stylistically, the texts are diverse. They comprise original literary works and folklore—more precisely, folk-song texts—and some can conditionally be classified as popular-scientific prose. These are texts written by the author himself (the fable ”Prakalba”, the song ”Pratar ymas”). Two texts (”Apey darima

113

“walge ysz kiarpiu islandu” and “Historyia Szwenta” are translations. The first of them could be assigned to texts of popular science, and the second to fiction.

After calculating the frequency of cases, it turned out that the most frequent case in S. Stanevičius’s texts is the genitive (gen. sg. 29%, gen. pl. 38%). The second most frequent are the singular nominative (27%) and the plural accusative (25%). The high frequency of the genitive in S. Stanevičius’s texts was most likely determined by the scholarly character of some texts ³.

When examining the declension of S. Stanevičius’s texts, certain difficulties arose: 1) S. Stanevičius’s texts were written on the basis of the Southern Samogitian Viduklė dialect; in that dialect (as throughout the Samogitian dialect), unstressed nominal endings are shortened and often have a vowel quality different from that of stressed endings; 2) this shortening made some endings of different origins identical and caused various morphological processes, as a result of which “baritone” i-stem nouns became intertwined with “baritone” ē- and iā-stem nouns; 3) therefore, wherever possible, I distinguished declension types only on the basis of “oxytonic” words, making an exception for the iu stem, which has no “oxytonic” forms, and for cases where no “oxytonic” forms of a given stem were attested in S. Stanevičius’s texts; 4) I present some case forms reconstructed according to the declension system of S. Stanevičius’s texts.

The “oxytonic” nouns in S. Stanevičius’s texts decline as follows:

a stem ia₁ stem V. szin’s jawaw sweczias *swetey K. szina jawù kiala *swecziu N. Diewuy jawåns swecziuy *swetēm G. Diewa jawus kiala kialus Įn. layku jaways keliu klejeys Vt. kalne jawùse kialie kraujuse įa₂ stem ā stem V. žaltis riszey rasa dinas K. *žalce sopulu žimòs dinu N. žalcz iuy *petēms kalbay daynùms G. buri burius žima dinas Įn. buriu peteys kalbà mierom’s Vt. burie Žemaycziuse dino kalbose įā stem ē stem V. paty, žynia paczes dubie žiwes K. žynios kancziu dubies žiw iu N. paczey *pacziums *dubey gymyniems G. žyne *paczes dube žoles Įn. kancze marcz ioms garbe gism iemys

³ Statistical calculations show that case frequency depends on the functional style to which a text belongs. For example, S. Klavina’s calculations of case frequency in Latvian show that the genitive is most commonly used in scholarly texts (about 40%), whereas the nominative is most frequent in literary texts (about 37%; the genitive accounts for only about 21% here); see [Клявина, 1976, с. 51—52].

114

Vt. waldžio *kancziose *dubie *dubiese im stem if stem V. *dantys žwierys dalys dãlys K. gialžis dantu dalis dalù N. *dancziuuy *dantyms *szaley awyms G. zwiery *dantys daly dalys Įn. wagymy wagymys ugnymy szalymys Vt. *dantie *dantyse dalie szalyse u stem iu stem V. sunus sumus waysius waysey K. sunaus sunu waysiaus ãmžiu N. sunuy *sunams karaluy *waysems G. sunu sunus waysiu waysius Įn. sunumy *sunumys karalum *wayseys Vt. dangu *turguse pawojų *pawojuse Cm stem Cf stem V. wandù wandenys sesu *seserys K. wandens akmenu sesers *seseru N. *pimeniuy pimen yms dukterey *dukter yms G. akmeni wandenys duktery *dukterys Įn. wãndeniu pimenymys *sesere *dukterymys Vt. wandenie *wandenu se⁴ *seserie *dukteryse

Having compared binarily all the noun-declension paradigms in S. Stanevičius’s texts and performed the calculations, I obtained the following results (see Table 2). In the table, the numbers at the intersections of columns and rows indicate the similarity index of the corresponding stems.

The table shows that, in order of similarity, the declension types can be arranged as follows: u, įu, a, įa₁, įa₂, im, if, Cm, Cf, ē, įā, and ā, and divided into three classes, with boundaries between the įa₂ and im stems, and between the Cf and ē stems.

The relationships between the paradigms can be brought into sharper relief by means of hierarchical clustering. The most convenient of the three possible methods is the so-called maximum method. Having carried out all the clustering procedures using this method, I obtained the following binary classification of noun-declension types in S. Stanevičius’s texts (see Figure 1).

As can be seen, the nouns in S. Stanevičius’s texts do indeed divide into three main classes according to the similarity of their declension types. The first comprises nouns with u, įu, a, and įa stems; the second, nouns with i and consonant stems; and the third, nouns with ē, įā, and ā stems. It is easy to see that all the words in the first class are masculine, those in the second class are masculine and feminine, and those in the third class are feminine (with rare exceptions such as dėdė, valkata, and the like). Thus, the language of S. Stanevičius’s texts is characterized by three main noun-declension types: I — masculine, II — mixed, III — feminine. Each

⁴ Cf. wandenuose StR 331, which is most likely an Aukštaitianism.

115

Fig. 1. A hierarchical grouping “tree” of the declension types in S. Stanevičius’s texts.

116

Within the type, two subclasses can be distinguished: in the masculine, the opposition between the u, įu and a, įa stems is clear; in the mixed type, the opposition between i and consonant stems; and in the feminine, the opposition between the ā, iā and ē stems.

Table 2

Similarity indices of declension types in S. Stanevičius’s texts

Type a | iā1 | iā2 | ā | iā | ē | ¹m | ¹f | u | įu | Cₘ | Cբ a | — | 12,45 | 9,45 | 1,5 | 1,5 | 1,5 | 1,1 | 0 | 4,45 | 6,7 | 3,3 | 0 iā1 | — | 11,6 | 1,5 | 1,5 | 2,65 | 2,65 | 1,15 | 4,45 | 6,7 | 4,45 | 1,15 iā2 | — | 0 | 0 | 1,5 | 3,75 | 2,65 | 4,45 | 6,8 | 5,95 | 2,65 ā | — | 12,1 | 6,2 | 0 | 1,1 | 0 | 0 | 0 | 2,25 iā | — | 6,2 | 0 | 1,1 | 0 | 0 | 0 | 2,25 ē | — | 1,15 | 2,25 | 0 | 0 | 1,15 | 3,4 ¹m | — | 12,5 | 1,1 | 1,1 | 8,3 | 8,25 ¹f | — | 0 | 0 | 7,2 | 9,25 u | — | 11,35 | 1,1 | 0 įu | — | 1,1 | 0 Cₘ | — | 10,4 Cբ | —

The research conducted allows us to draw the well-founded conclusion that the “tree” of noun declension types in S. Stanevičius’s texts essentially almost coincides with the corresponding “tree” of the modern standard language, except that it is somewhat less developed and indicates that noun declension types in S. Stanevičius’s dialect had only just begun to converge. If “baritone” nouns were also taken into account, the similarity between declension types would be greater.

LITERATŪRA

G i r d e n i s, R o s i n a s, 1977 — G i r d e n i s A., R o s i n a s A. Lietuvių kalbos daiktavardžio linksniavimo tipų hierarchija ir pagrindinės jos kitimo tendencijos (bendrinės kalbos ir žemaičių tarmės duomenimis). — Baltistika, 1977, T. 13(2). P. 338—348.

К л я в и н я, 1976 — К л я в и н я С. П. Сопоставление статистических характеристик морфологических категорий в функциональных стилях латышского языка. — Valodas sistēma. Rīga, 1976. Lpp. 49—80.

HIERARCHY OF THE DECLENSION TYPES OF THE NOUN IN SIMONAS STANEVIČIUS’ TEXTS

Summary

Statistics indicate that the nouns in S. Stanevičius’ texts fall into three main classes according to the isomorphism of their declension. The first class is represented by masculine nouns, the second — by masculine and feminine and the third — by feminine nouns. It follows that there are three main declension types in the nouns of S. Stanevičius’ texts, i. e. masculine, mixed and feminine, respectively. Each of these three types allows, furthermore, binary subdivision: u, įu:: a, įa (the masculine type), i:: consonant themes (the mixed type) and ā, iā:: ē (the feminine type). The dendrogram of the declension types almost correlates with the corresponding dendrogram of the contemporary literary language, however, the former being a bit lower; this shows that the inflectional types in S. Stanevičius’ dialect entered the process of mutual assimilation.

117