Pastabos

On endings and stem-final vowels

These lines were prompted by Prof. Vytautas Ambrazas’s commentary, published in this issue of ALL (pp. 109–111), on Jolanta Vaskelienė’s article (Vaskelienė 2006) published in issue 54 of ALL¹². Drawing on some authors’ proposal that the endings of the Lithuanian verb’s third-person forms, -a, -i, -o, be treated synchronically as inflectional endings, the author of the article attempted to apply this analysis to the description of participles and gerundives. This attempt to think through fully the consequences of choosing a particular interpretation of a phenomenon for other areas of grammatical description is undoubtedly useful and too rarely practiced. However, according to Vytautas Ambrazas, the analysis on which the article’s author relies had been widely discussed in the literature and rejected. Prof. Ambrazas recalls the work of Vladas Žulys and Tatjana Bulygina, in which “... it was firmly proved that the synchronous system of Lithuanian verb conjugation can be described transparently and consistently only if the endings of third-person forms are treated as vocalic stem-final vowels, whereas attempts to turn them into inflectional endings would obscure conjugation with a multitude of artificial inflections” (p. 110).

Indeed, a description that treats the endings of third-person forms as stem-final vowels may seem more economical, and therefore better, than one that treats them as inflectional endings. But whether its advantage can be “firmly proved,” definitively resolving the problem and closing discussion of the matter for good, seems doubtful. Besides, resolving one question gives rise to others elsewhere. Once stem-final vowels are brought back into verb inflection, the question inevitably arises: why not introduce them into noun declension as well?

Having introduced into verb conjugation the stem-final vowels excluded from the Academic Grammar, the authors of Grammar of the Contemporary Lithuanian Language (hereafter: DLKG) did not venture to describe noun declensions according to a similar principle. In the forms vyr-a-ms, žmon-o-ms, šun-u-ms, gėl-e-ms, av-i-ms, the vowels a, o, u, ė, i remain parts of unanalyzable endings in the description presented in this grammar, as they had in the Academic Grammar. Since we do not know the circumstances behind the writing of DLKG, we can only speculate as to why different principles were chosen for the description of nouns than for that of verbs. One may speculate that purely statistical arguments could have played some role. For the forms in which the historical stem-final vowel remains intact and easily recognizable, their relative number is greater in conjugation than in declension. In the present-tense to lead verb

¹² I would like to thank Dr. O. Vaičiulytė-Romańczuk (University of Warsaw), with whom I recently had the opportunity to discuss questions of Lithuanian word formation. These discussions helped me formulate some of the ideas expressed in these notes.

112

Pastabos

paradigm, of the five forms that actually differ, three (60%) retain the intact stem-final vowel -a- (veda, vedame, vedate), while only two merge it with the ending (vedu, vedi). In the paradigm of the noun gėlė, of a maximum of thirteen distinct forms (the plural vocative always coincides with the nominative), seven (just under 54%) retain the intact stem-final vowel -ė- (singular gėlė, gėlės, gėlėje; plural gėlės, gėlėms, gėlėmis, gėles); in the other cases, operating on the stem-final vowel requires introducing additional morphonological rules. The difference is not very pronounced, but the results are somewhat poorer for other declensions: in the case of -o- stems (liepa, plural dative liepoms), only 46% of forms retain the vowel -o-. Of course, all this is merely conjecture: we do not know whether the possibility of describing noun and verb inflection according to the same principles, operating on stem-final vowels, was ever considered by the authors of DLKG. It seems that Bulygina’s and Žulys’s articles prompted the authors to change the description of verbs, while the question of harmonizing the descriptions of verb and noun inflection was never raised. Yet this possibility should be considered for several reasons.

The statistical differences mentioned above might justify accepting different treatments of verbs and nouns, introducing stem-final vowels only into the description of verb inflection. However, another complex issue in the Lithuanian grammatical tradition encourages us to rethink the description of nouns: so-called suffixal derivation. This concept is a distinctive feature of Lithuanian grammar—it is almost impossible even to translate the term into other languages. Where did it come from?

The author of these lines has not conducted an in-depth study of the tradition of describing Lithuanian word formation, but it may not be too risky to claim that it arose when noun stem-final vowels were eliminated. Since the ending -us in the word stiklius is synchronically treated as an inflectional ending, one must conclude that, in relation to the base word stiklas, this word differs by just one element—the inflectional ending. Comparing the word belangė with the base word langas, two derivational devices are identified: a prefix and an inflectional ending. This description seems convincing at first glance, but it conceals a certain paradox: since each case has a different inflectional ending, the word stiklius in fact has not one but thirteen different derivational devices—one for each case, except for the identical plural nominative and vocative endings.

This complication is generally overlooked: it is said that the derivational device in the word stiklius is the ending -us, while in fact what is being used here is an abstract, “generalized” ending (“the ending -’us or the corresponding endings of the same paradigm”). That in such cases what is meant is the set of noun case endings is emphasized explicitis verbis by Urbutis (1978: 221). But clarifying the formulation in this way does not change the essence of the matter: instead of operating with a constant, for all inflected forms

113

Pastabos

with a shared derivational feature, the derivational device is conflated with an element that alternates according to grammatical category—the ending—and the result is different derivational devices for every case and both numbers. This defective construction could be avoided by distinguishing from the ending a stem-final element that is common to all cases and both numbers. In that case, the derivational device used in the word stiklius would be a change in the stem-final element (-a- → -’u), while the ending would remain the same. The derivation of the word belangė would involve two devices—the prefix be- and the stem-final element -ė-; there would be no ending in the nominative form belangė (since nominative forms, like third-person verb forms in the verbal paradigm, are also unmarked, this seems entirely normal).

Another argument could be made in favor of this descriptive approach: it is already applied, to some extent, in descriptions of Lithuanian word formation, although in a disguised form. How do the words karalienė and kiauliena differ in terms of derivation? According to DLKG, they differ in their suffixes: in one case -ienė (p. 142), and in the other -iena (p. 141). But this clearly contradicts the description of noun inflection given in the same grammar: after all, -ė and -a are endings, so the suffix is common to both words. Can an ending be part of a suffix? It probably should not be, but if we agreed that the derivational devices in the words karalienė and kiauliena consist of a suffix sensu stricto and a stem-final element, we would obtain two different derivational complexes, -ien-ė- and -ien-o-, to which different derivational meanings could be assigned. This is in fact what is done when the suffixes -ienė and -iena are distinguished. The difficulty is that this then contradicts the description of inflection given elsewhere (according to which -a and -ė are endings), and, in essence, the very concept of “derivation by endings.” Comparing the words belangė (: langas) and kiauliena (: kiaulė) shows that the declension class changes in both cases, but in one case this change is treated as a derivational device, while in the other it is ignored, with the declension-class feature lumped together with the suffix. The contradictions within the grammar’s own framework remain concealed, and usually no one notices them (although Urbutis 1978: 222 briefly mentions this point).

The proposal presented here to reintroduce stem-final elements into noun inflection is not the only possible one. Instead of working with stem-final elements, one could work with an abstract derivational feature without phonetic realization—the transfer to another declension class. As early as 1978: 221, Urbutis proposed the term “paradigmatization” alongside “derivation by endings.” Smetona (2005), for example, uses the term “paradigmatization.”

But what is paradigm(iz)ation? The paradigm itself, or a change in the paradigm? If it is the paradigm itself, a certain complication arises, since then no words remain that are not derivationally derived—

114

Pastabos

—at the derivational level. The set of endings then performs not only inflectional functions but also a derivational one. The inflectional endings of the noun brolis derive the noun brolis from the stem brol’-, although it is difficult to define the derivational meaning here. On the other hand, if the derivational feature is a change in the paradigm, then the description of derivation also becomes considerably more complicated. This is because, when specifying the derivational devices in the words bro­lelis, vaikelis, sūnelis, we do not usually stop to consider which paradigms the base words brolis, vaikas, sūnus belong to: the suffix mechanically determines the inflectional type, regardless of the declension class of the base word. But if, when considering the relationship between langas and belangė, we treat a change in paradigm type as a derivational device, then we must do the same in the case of vaikelio (← vaikas). This greatly complicates the description, because it requires including a great deal of unnecessary, wholly predictable information, such as, for example, that the inflectional paradigm of the diminutive form vaikelis differs from the paradigm of the word vaikas.

Visa tai turint omenyje, ne toks jau blogas atrodo senas, dar Skardžiaus (1943) taikomas principas, kai (kad ir nenuosekliai) išskiriami tokie dar ybiniai elementai, kaip -ja- (aukštis ← aukštas), -ju- (anglius ‚anglininkas, Köhler‘ ← anglis), arba (kai kamiengalis sudaro kompleksinį vienetą su priesaga) -ala- (augalas), -inja- (sidabrinis) ir pan. Tokia traktuotė gali atrodyti pernelyg istorinė, tačiau ne mažiau istoriškai motyvuotas yra teiginys, jog formoje augu slypi kamiengalis -a-. Panašiai, kaip ir veiksmažodžių asmenuotės atveju, sprendimas įsivesti kamiengalius į daiktavardžio linksniuotės aprašą galėtų būti motyvuojamas ne prielaida, jog sinchroninis aprašas turėtų sutapti su istoriškai pagrįstu, o siekiu dabartinės lietuvių kalbos kaitybinę ir darybinę morfologiją aprašyti paprastai ir neprieštaringai. Daiktavardžio atveju tokio aprašo privalumas būtų trigubas: (i) pačios daiktavardžio kaitybos aprašas įgytų tuos pačius „skaidrumo ir nuoseklumo“ požymius, kuriuos prof. Ambrazas pripažįsta ir pabrėžia veiksmažodžio asmenuotės atveju, (ii) daiktavardžio ir veiksmažodžio kaitybos aprašai būtų paremti vienodais principais, (iii) paprastesnis taptų žodžio darybos aprašas ir neliktų prieštaravimų, kai tas pats segmentas -ė- tos pačios gramatikos puslapiuose vienur laikomas galūne, o kitur – priesagos (-ienė) dalis.

The authors of DLKG decided not to make use of these advantages, and we do not intend to criticize their decision here. However, we see no basis for claiming that the solutions adopted in this grammar are more transparent and consistent than those applied in Jolanta Vaskelienė’s article. On the contrary, the view that the endings of both the form augal-a-ms and the form aug-a (-ams and -a) are endings is more consistent than the view of the authors of DLKG, according to which only the ending -ams in the first form is an ending.

115

Pastabos

BIBLIOGRAFIJA

DLKG – Dabartinės lietuvių kalbos gramatika. 4-oji papildyta laida. Vilnius: Mokslo ir enciklopedijų leidybos institutas.

Skardžius, P. 1943/1996. Lietuvių kalbos žodžių daryba. Vilnius: Lietuvių kalbos institutas, 1943. Perspausdinta: Rinktiniai raštai 1, par. A. Rosinas, Vilnius: Mokslo ir enciklopedijų leidykla, 1996.

Smetona, A. 2005. Pagrindinių darybos būdų skyrimo klausimu. Acta Linguistica Lithuanica 52, 83–89.

Urbutis, V., 1978. Žodžių darybos teorija. Vilnius: Mokslas.

Vaskelienė, J. 2006: Dėl esamojo ir būtojo kartinio laiko dalyvių bei padalyvių darybos ir morfeminio skaidymo. Acta Linguistica Lithuanica 54, 83–93.

Axel Holvoet

Lietuvių kalbos institutas

P. Vileišio g. 5, 10308 Vilnius, Lietuva

Dėl sintaksinio dviprasmiškumo ir teminės-reminės struktūros

Su Rolando Mikulsko „Pastabomis dėl sintaksinio dviprasmiškumo sąvokos“ (šis ALL numeris, pp. 1–53) papolemizuoti paskatino ne tik straipsnio pavadinime minėtas klausimas, bet ir „kiti susiję dalykai“, ypač šių eilučių autoriaus 2003 m. (remiantis Bogusławskio 1977 m. koncepcija) suformuluotos aktualiosios skaid os koncepcijos kritika. Diskusija čia paliečia fundamentalius kalbos gramatinės sandaros klausimus, ir Mikulsko pastabos iškelia fundamentalias problemas.

Visų pirma dėl sintaksinio dviprasmiškumo: Rolando Mikulsko nuomone, mūsų aptariama konstrukcija Onutės žalios akys negalinti būti įvertinta kaip sintaksiškai dviprasmiška, nes ji turinti ryškią, tik šiai konstrukcijai būdingą reikšmę; sintaksinis dviprasmiškumas implikuojąs ir semantinį dviprasmiškumą, kurio čia kaip tik nepastebime.

Kognityvinėje lingvistikoje, iš kurios pozicijų Mikulskas kritikuoja mūsų analizę (panašiai kaip ir konstrukcijų gramatikoje, kurią galima laikyti kognityvinės

116