Go to originalLocalized version was produced by AI. It may contain inaccuracies.
Preparing document pages…
4 short passages of this localized version are shown in the original language because they could not be localized reliably.
Page 1
ERIKA RIMKUTĖ
Vytauto Didžiojo universitetas
Research areas: corpus and computational linguistics, automatic morphological analysis and synthesis, morphological ambiguity and its resolution.
ASTA KAZLAUSKIENĖ
Vytauto Didžiojo universitetas
Research areas: experimental phonetics, synchronic accentology, dialectology, applied linguistics.
GAILIUS RAŠKINIS
Vytauto Didžiojo universitetas
Research areas: recognition methods, recognition of music and speech signals, computational and corpus linguistics.
MORPHEMIC STRUCTURE OF LITHUANIAN VERBS1
The Morphemic Structure of Lithuanian Verbs
ABSTRACT
The morphemic structure of Lithuanian words has not been studied quantitatively. The aim of this study is to identify and describe the morphemic structure patterns of Lithuanian verbs and the regularities of their realization. The article analyzes nearly 93,000 verbs. The research material consists of various excerpts from texts collected in the Corpus of the Contemporary Lithuanian Language. The texts have been morphologically annotated and their morphemes marked.
The verbs are classified into 116 morphemic structure patterns, whose productivity varies greatly: verbs in the most productive pattern (root + ending) make up one fifth of all examples examined, while as many as 26 patterns have only one or two examples each.
Verbs may consist of one to nine morphemes. Verbs with 2–4 morphemes are the most characteristic of Lithuanian (89 percent).
1 The study was funded by the Research Council of Lithuania (Agreement No. MIP-06/2010).
Although many structural patterns were identified, a quarter of them (26 patterns) account for as much as 95 percent of all the verbs analyzed. The most characteristic patterns are: (prefix) + root + ending (33 percent), (prefix) + root + inflectional suffix + ending (15 percent), (prefix) + root + inflectional suffix (11 percent). In Lithuanian, verb forms without derivational suffixes, with or without a prefix, and containing one or two inflectional affixes are the most commonly used.
In the material examined, one third (33 percent) of the verbs are simple; about 28 percent have one or two prefixes; about 15 percent have a derivational suffix; 12 percent have prefixes and a derivational suffix; and about 6 percent have a prefix and are reflexive. Verbs with other combinations of derivational morphemes account for only one percent. It was observed that verbs with more than two derivational affixes are atypical of Lithuanian.
Although the range of combinations in the morphemic structure of Lithuanian verbs is wide, the core of usage consists of verbs belonging to 10–15 patterns (about 75–85 percent).
ANNOTATION
The aim of this research was to identify and to describe the morphemic structure of Lithuanian verbs from the quantitative point of view. Quantitative analysis of the morphemic structure of Lithuanian verbs was not addressed by previous research. The investigation material consisted of all verb tokens (nearly 93 thousand) from the specially selected subset of Lithuanian language corpus. All verbs within this data set have been morphologically annotated, i.e. morphemes have been identified and their boundaries have been labeled.
Verbs have been grouped into 116 categories according to their morphemic structure. Verbs were found to be composed from one to nine morphemes. However, the majority of verbs (89 percent) were composed from two to four morphemes.
INTRODUCTION
The morphemic structure of Lithuanian words was discussed by Monika Kuosienė in the book Morphemic Analysis (1986) and by Stasys Keinys in the book Morphemics of Standard Lithuanian (2009). However, these works do not provide a quantitative analysis of the morphemic structure of Lithuanian words. Grammars and textbooks generally present only general statements about Lithuanian morphemes. This study draws on extensive data from contemporary Lithuanian and is therefore expected to reveal significant regularities in the morphemic composition of Lithuanian verbs.
The root structure of verbs included in the dictionary has been described in great detail (Kruopienė: 2000; 2001; 2002; 2005). Many aspects of the morphonological structure of verb roots are presented in exceptional detail in Audronė Kaukienė’s monograph The History of the Lithuanian Verb (1994–2002). Vida Zita Karosienė (2004) has also written about the phonotactic structure of noun roots in Standard Lithuanian.
Several works by Gražina Akelaitienė address morphonological alternation in roots and at morpheme boundaries: the book Morphonological Vowel Alternation in Word Formation (1996), and the articles
88
Acta Linguistica Lithuanica LXIV–LXV
Page 3
Lietuvių kalbos veiksmažodžių morfeminė struktūra
“Root Vowel Alternation: Morpheme Models and Structural Root Types” (2001), “Root Vowel Alternation in Derived Nouns with Consonantal Suffixes” (2008).
No comprehensive studies were found on the morphemic structure of words in the structurally related Latvian language. The works located (Ugis Sarkans, “Morphemic and Morphological Analysis of the Latvian Language” (1996); Kristine Levāne, Andrejs Spektors, “Morphemic Analysis and Morphological Tagging of Latvian Corpus” (2000); Pēteris Paikens, “Lexicon-Based Morphological Analysis of Latvian Language” (2007)) do not describe morphemic structure quantitatively; they discuss how information about the morphemic structure of Latvian words can be implemented in a morphological tagger. The function of determining a word’s morphemic composition is also important because it helps identify new words in contemporary Latvian.
Levāne ir Spektors straipsnyje (2000) pateikta labai apibendrinta latvių kalbos žodžių morfeminė struktūra: vienašaknių žodžių *priešdėlis + šaknis + *priesaga + (galūnė); sudurtinių žodžių *priešdėlis + šaknis + *priesaga + (galūnė) + (priešdelis) + šaknis + *priesaga + (galūnė)2.
The aforementioned article describes a prototype morphemic segmentation system developed for Latvian3. However, morphemes are not described in greater detail. The system also performs morphological analysis of Latvian words and can synthesize other inflected forms of a specified word.
The authors of the article found no detailed information on the morphemic structure of Russian from a quantitative perspective, so the information presented below should be treated with considerable caution4. In Russian, the most frequent word forms consist of two to four morphemes. The following models are possible:
R: иоссе, около, вдруг;
xR: ни-где, от-ныне;
Rx: хорош-о, вод-а;
xRx: по-утр-у, при-город-ы;
xRxx: в-пят-ер-ом, за-город-н-ый;
Rxxx: тк-ач-их-а, труд-и-ть-ся;
Rxxxx: рыб-ач-и-вши-ый;
xxRxxx: re-arma-ment-ation-ized;
xxxRxxx: re-on-you-write-out-ate;
xxxRxxxx: re-on-you-pull-out-iv-a-l-o-s5.
2 An asterisk means that an affix may be absent or there may be several affixes; affixes given in parentheses are optional.
3 It can be found at: http://eksperimenti.ailab.lv/morfo/.
4 Information from the website http://www.langust.ru/rus_gram/rus_gr03.shtml.
It is claimed that forms with as many as 11 morphemes are possible, e.g., ны́л-е-влаг-о-не-про-ныц-а-ем-ость (RxRxxxRxxxx).
In the articles “Menzerath’s Law for Morphemic Structures of Words: A Hypothesis for the Evolutionary Mechanism of its Arising and its Testing” (2000a) and “Chronological Morphemic and Word-Formational Dictionary of Russian: Some System Regularities for Morphemic Structures and Units” (2000b), Anatoliy A. Polikarpov analyzes and applies Menzerath’s law6 on the basis of a chronological dictionary of Russian morphemes and word-forming elements7, comprising more than 180,000 words. The articles describe several trends: newer words are formed with newer morphemes; the older the words, the longer their morphemes; the average length of a morpheme is inversely proportional to its distance from the root (e.g., the farther a suffix is from the root, the shorter the suffix); the longer the morphemes, the fewer of them there are in a word; the more morphemes a word contains, the shorter they are. These regularities were established on the basis of Russian-language data. It may be assumed that similar trends would also emerge in Lithuanian, but that would be the subject of another study.
RESEARCH MATERIAL
The research material consists of various excerpts from the Corpus of the Contemporary Lithuanian Language (http://tekstynas.vdu.lt) accumulated scientific publications, journalistic texts, novels by Lithuanian authors, and administrative documents (orders, resolutions, minutes, statutes). Excerpts from the Spoken Language Recordings Database (http://donelaitis.vdu.lt/garsynas/) were also used for the study. The research material comprises approximately 430,000 words (nearly 80,000 distinct forms). The texts were morphologically annotated; errors made by the morphological tagger were corrected manually, the meanings of polysemous words were identified, and morphological compounds8 were excluded.
Since the study analyzes usage data (rather than dictionary data), the term frequency is used, rather than productivity.
In determining morpheme boundaries, we followed theoretical principles formulated with great precision by Vincas Urbutis (2009: 165–166). Words were divided into as many morphemes (the smallest
6 The articles give the following title: Chronological Morphemic and Word-Formational Dictionary of Russian Language.
7 Menzerath’s law states that the longer a unit is, the shorter its parts are. The law is formulated in English as follows: “the longer some “construct” (the whole) the shorter should be its “components” (parts)” (Polikarpov 2000a; 2000b).
8 Morphological compounds are formations most often consisting of several indeclinable words, having a shared meaning and performing the same syntactic function, e.g.: without a doubt, in fact, anywhere.
90
Acta Linguistica Lithuanica LXIV–LXV
Page 5
Lietuvių kalbos veiksmažodžių morfeminė struktūra
of meaningful parts of the word), insofar as they can be identified from the perspective of the contemporary language9. A word part was considered a morpheme: a) if both (or all) parts of the word (presumed morphemes) recur in other words (e.g.: house-diminutive-ending – house, house-related and little table, little child); b) in cases where only one part of the word—its root or affix—recurs in other words (e.g.: dilet-ant-as – debutant, intern).
The most difficult morphological analysis arose in cases of contraction, elision, and degemination, e.g.: neš- + -siu (cf. neša) = nešiu; lūž- + -sta (cf. lūžo) = lūžta; irz- + -sta (cf. irzo) = irzta, ne- + -esu = nesu. Such a merged sound belongs to both morphemes. This does not substantially affect morphological analysis if the morpheme also contains other sounds. However, when a morpheme consists of that one sound alone and it merges, in principle the morpheme ceases to exist. Therefore, the following symbolic notation was chosen: nešiu (RF10 structure), lūžta, irzta (RSF), nesu (PRF). This choice may be debatable, but there are few such cases (about 2 percent), and the general patterns of morphemic structure can be established. In future, these cases (marked with a special symbol in the research material) are to be examined in greater detail.
Longer verb suffixes were treated as variants of suffixes, e.g.: -ėti variants -alėti, -elėti, -dėti, -lėti, -nėti, -sėti, -vėti; -oti variants -ioti, -čioti, -l(i)oti, -n(i)oti, -snoti, -šnoti, -al(i)oti, -ailioti, -elioti, -sčioti, -šlioti, -en(i)oti, -aroti, -uoti, and so on.
Priebalsiai j, n, v kai kurių mokslininkų (Keinys 2009: 25) vadinami morfemų plėstukais, nes jie įterpiami tarp dviejų balsių naikinant hiatą, pvz.: jo-ti, joj-a, joj-o; au-ti, aun-a, av-ė, kliū-ti, kliūv-a, kliuv-a. Dabartinės lietuvių kalbos gramatikoje (1996: 327) jie vadinami intarpais. Rengiant medžiagą tyrimui šie elementai pažymėti, bet apibendrintuose struktūriniuose modeliuose neatsispindi. Juos (kaip ir įvairius morfemų sandūros garsų pakitimus) ketinama ateityje nagrinėti išsamiau.
In the words keturiasdešimt, trisdešimt, kasdien, kelintąkart, kąsyk, šiandien, visuomet and others, the highlighted segments are endings from a diachronic perspective. Since they now function as linking sounds, in this study they were marked as linking sounds.
The object of the study is the morphemic analysis of words; therefore, tebe-, nebe- were treated as two prefixes. Pronominal endings are not synchronically segmented (Keinys 2009: 55–56).
In this study, verb suffixes are divided into derivational and inflectional suffixes. This distinction is not important for morphemic analysis. However, without it, it would be impossible to determine how many verbs are primary and how many are derived.
9 Nevertheless, it must be acknowledged that when analyzing words morphologically, the boundary between synchrony and diachrony is not always easy to draw. Consequently, some words required lengthy deliberation over their morphemic structure (e.g.: viešpatauti, dievagotis).
10 The symbols used are explained at the end of this section.
In the article, morphemes are denoted by the following symbols: R – root, P – prefix, F – ending, d – reflexive particle, c – linking sounds, Sd – derivational suffix, Sk – inflectional suffix, k – contraction, denoting fused morphemes when one of them has disappeared (e.g., future-tense or imperative forms nešiu, auk). Such cases are presented only in the first table.
MAIN MODELS OF THE MORPHEMIC STRUCTURE OF VERBS
The study material contains almost 93,000 (92,468) verb tokens (of which 29,878 are distinct). The diversity of their morphemic structures is considerable: as many as 116 structural models. However, the frequency of model use varies greatly (from 1 to 18,377 verbs).
Verbs may consist of as few as one morpheme (e.g., the spontaneous-speech form reik) or as many as nine (e.g., neįsipareigokite) morphemes. However, monomorphemic verbs and those consisting of more than five morphemes are rare. Thus, the core of usage (as much as 89 percent) consists of verbs with 2–4 morphemes. Naturally, the fewer the morphemes, the less varied their combinations. The greatest number of structural models (as many as 69, about 40 percent of all models) occurs in the 5–6-morpheme group (see Fig. 1).
FIG. 1. Diversity of structural models and number of morphemes
92
Acta Linguistica Lithuanica LXIV–LXV
Page 7
Lietuvių kalbos veiksmažodžių morfeminė struktūra
Monomorphemic verbs are very rare. Only the spontaneous-speech form reik was found in the study material. This group would also include third-person future-tense forms and second-person imperative forms: grįš, leis, bėk, dink. However, as noted earlier, their morphemic structure is debatable, since the final root consonant fuses here with the imperative or future-tense suffix.
Dimorphemic verbs make up as much as a quarter (about 25 percent) of all verbs in the study material. The variety of dimorphemic models cannot be extensive (see Table 1). These are verbs consisting of a root and an ending (RF, finite forms and participles, e.g.: buvo, gali, buvęs, esąs11), a root and an inflectional suffix (RSₖ, finite forms, infinitives, and gerunds, e.g.: bus, eik, būti, dirbti, esant, siekiant), as well as isolated verbs consisting only of a prefix and a root (PR, the form nereik, occurring only in spontaneous speech).
Trimorphemic verbs (slightly more than a third of them, about 37 percent) may be formed from various combinations of a root, a derivational or inflectional suffix, an ending, and a reflexive affix. There are nine trimorphemic models: PRF (finite forms and participles, e.g.: nėra, atrodo, susijęs); RSₖF (forms with endings, e.g.: būtų, būdavo, galima, žinoma, būdamas, būdami); PRSₖ (forms without endings, e.g.: nebus, ateis, atlikti, padėti, išskyrus, sudarant); RSₖF (finite forms and participles, e.g.: turėjo, gyvena, galėjęs, turėjęs); RS_dSₖ (most of these verbs are infinitives, e.g.: daryti, kalbėti, but there are also gerunds and finite forms, e.g.: žinok, reikės, naudojant, vykdant); RFd (finite forms and a few participles, e.g.: skiriasi, laikosi, mušėsi, lankėsi); RSₖd (verbs of this structure are infinitives, along with a few finite forms and gerunds, e.g.: elgtis, kreiptis, baigsis, ruoš kis, remiantis, laikantis). Trimorphemic verbs also include the rare verbs koneveikti (RRSₖ), būsiant (RSₖSₖ). Verbs belonging to all the other models are fairly common.
1 TABLE. Models of the morphemic structure of verbs12
Number of morphemes
Model
Infinitive
Participle
Gerund
Half-participle
Finite forms
Percent
2
RF
19,874
RSₖ (RSₖk)
5,459
PR
0,003
11 In all cases, the examples most frequently found in this study will be presented.
12 Owing to space constraints, the table does not include all models, but only the more characteristic ones. It excludes models longer than six morphemes and isolated shorter models. The lightest shading marks forms occurring from 1 to 100 times, darker shading those occurring from 101 to 500 times, and the darkest shading those occurring 501 times or more.
Four-morpheme verbs (about 26 percent of them) are represented by 16 structural models. Only three of these models are rare in use. These are two-root verbs, e.g.: pusryčiavo (RRSdF), senberniauti (RRSdSk), and several dozen active-voice participles of the RSkSkF model, e.g.: trūktelėjęs, pyktelėjęs. The most frequent four-morpheme models are: PRSkF, RSdSkF (mostly participles, e.g.: negalima, įmanoma, naudojami, naudojamos), PRSdSk (especially many infinitives, e.g.: nustatyti, pasakyti, padaryti), PdRF (e.g.: atsirado, pasitaiko, pasirodė), PRSdF (e.g.: negalėjo, pasakojo, neturėjo). Other models are less frequent, but they are certainly not peripheral to the language, as their examples show: PdRSd (e.g.: pasirinkti, įsigyti, atsižvelgiant), PPRF (e.g.: nebėra, nepavyko, nesupratau), RSdSdF (e.g.: tvirtina, reikalauja, dalyvauja), RSdFd (e.g.: vadinasi, rūpinasi, formuojasi), RSdSkd (e.g.: naudotis, mokytis, laikytis), RSdSdSk (e.g.: organizuoti, reikalauti, tvirtinti), RSkFd (e.g.: kreip REPLACE
There are many models of the morphemic structure of verbs. By frequency of use, they can all be divided into three groups: a) the thirteen most frequent models, each represented by more than two thousand examples (together they account for almost 85 percent of verbs); b) another thirteen models of medium frequency, each occurring more than 400 times (together they account for almost 10 percent of verbs); c) all other models are rare (verbs in the remaining 90 models account for only 5 percent). In general, it can be stated that Lithuanian is characterized by somewhat fewer than thirty models of verb morphemic structure; all the other models are isolated and peripheral.
A closer examination of the most productive models presented in Table 2 shows that the first six models form pairs, the first member of each pair lacking a prefix and the second having one: the most frequent RF and PRF models, the RSkF and PRSkF models, which are half as frequent, and the still less frequent RSk and PRSk. Three more pairs follow a similar pattern (the first without a prefix, the second with one, although they are no longer adjacent): RSdSkF and PRSdSkF, RSdF and PRSdF, RSdSk and PRSdSk. After this, models of reflexive verbs appear, and it becomes more difficult to discern clear distributional patterns.
At the bottom of the list of all 116 models (not presented here because of its length) were disyllabic verbs: as many as twelve verbs of different structures, each used once or twice—twelve rarely used models. There are also verbs derived from international words (e.g., komplikuosis, užsifiksuodavo, susimontavo, išprovokuos), whose suffix segmentation is also debatable. This merely shows that these models are peripheral.
KEY REGULARITIES IN THE MORPHEMIC STRUCTURE OF VERB FORMS
The study makes it possible to establish which structural models are characteristic of individual verb forms13. On the other hand, analysis of the morphemic structure of verb forms cannot yield major surprises, since it is partly determined by the formation of the forms themselves: the infinitive and the half-participle have an inflectional suffix, while participles and gerunds have an inflectional suffix and an ending. Both types may also have a reflexive particle at the end. Greater variety is possible only among finite forms. Naturally, coherent texts contain the largest proportion (56 percent) of finite forms, and their structural models display the greatest variety (85 models). Finite verb forms usually end in an ending, an inflectional suffix and an ending, or, in some cases, only
13 At the beginning of the study, no such goal had been set, so the data were classified into five groups: finite forms, infinitive, participle, half-participle, and gerund. In the future, finite forms could be subdivided by tense and mood.
an inflectional suffix (third-person future tense, second-person singular imperative), and less often a reflexive particle. The ten most characteristic models14:
Participles are half as numerous (about 24 percent of all verbs) as finite forms, but in the abundance of their morphemic structure models they almost equal finite forms (74 models). Participles end in an inflection or a derivational suffix and an inflection:
There were few gerunds and half-participles in the material analyzed. The frequency of use and the diversity of structural models for these forms are very similar. Gerunds (3 percent in total, 24 models) always end in an inflectional suffix:
When inflectional affixes are excluded, the five most characteristic models are the same for all forms; only their order by frequency of use differs:
a) finite forms:
R; PR; RSd; PdR; PRSd;
b) participles:
PR; R; PRSd; RSd; PdR;
c) infinitives and adverbial participles:
PR; R; RSd; PRSd; PdR;
d) semi-participles:
RSd; R; PR; PRSd; PdR.
As can be seen, the first three models for almost all forms are: R; PR; RSd. These five models account for as much as 86 percent of occurrences: R – 33 percent, PR – 25 percent, RSd – 14 percent, PRSd – 11 percent, PdR – 3 percent.
THE RELATIONSHIP BETWEEN SIMPLE VERBS AND DERIVATIVES
Although word formation is not the subject of this study, the material collected nevertheless allows a few observations to be made from this perspective as well. It is true that the usual word-formation terms (suffixal derivative, prefixal derivative) are not used here, since the verbs are not analyzed from a word-formation perspective. As already mentioned, almost 93 thousand verbs were analyzed. Simple verbs account for one third (33 percent) of all verbs used. These are verbs consisting of a root and an ending or (and) an inflectional suffix: R (the aforementioned spontaneous-speech reik), RF (about 20 percent of all verbs used, e.g.: yra, buvo, gali, turi, žinai, reikia, sako), RSk (about 6 percent, e.g.: bus, eik, te ks, duok, taps, gaus), RSkF (about 8 percent, e.g.: būtų, būdavo, tarsi, tarkim, būčiau, būsi, reiktų), RSkSk (būsiant), RSkSkF (very rare, e.g.: būsimos, būsimų).
Two thirds of the verbs in the research material have derivational affixes (although, upon derivational analysis, some of them may also be non-derived). As Table 3 shows, no verbs were found that had all four derivational morphemes¹⁵; verbs with three different types of derivational morphemes are also rare (about 2 percent of all verbs; these are usually: a prefix, a suffix
15 The number of derivational affixes is not taken into account here; only the type of affix is considered: prefix, suffix, reflexive affix.
and a reflexive affix, e.g., įsivaizduoji, įsigalioja, apsigyveno, susirūpino; a prefix, a suffix and two roots, e.g., nufotografuoti, nepiktnaudžiaujant). Verbs having two different types of derivational affix are quite common (about 20 percent); the more characteristic combinations are: a prefix and a suffix, e.g., negalėjo, neturėjo, pastebėjo, sumažėja, nežinojo; a prefix and a reflexive affix, e.g., atsirado, pasitaiko, pasirodė, pasidaro, prasidėjo; a suffix and a reflexive affix, e.g., laikykis, saugokis. The largest group consists of verbs with prefixes (about 28 percent) and suffixes (15 percent). Thus, from this perspective, verbs in actual usage could be arranged in the following descending order: with a prefix (about 28 percent), with a suffix (about 15 percent), with a prefix and a suffix (12 percent), with a prefix and a reflexive affix (about 6 percent), with a reflexive affix (about 2 percent), with a prefix, a suffix and a reflexive affix (about 2 percent), with a suffix and a reflexive affix (about 1 percent). Verbs with other combinations of derivational morphemes account for only one percent.
3 TABLE. Distribution of verbs by derivational morpheme type and of two-root verbs
Prefixal Suffixal Reflexive Two-root Total (percent of all verbs) 27.74 12.18 5.77 0 2.16 0.02 0 0 15.21 1.35 0.07 0.00116 2.26 0.00117 0.00118
16 Only fotografuojamasi.
17 Only dievagojosi.
18 Only koneveikti and visaregintis, which also occurs in the texts examined.
100
Acta Linguistica Lithuanica LXIV–LXV
Page 15
Lietuvių kalbos veiksmažodžių morfeminė struktūra
Nearly half of the verbs used in the material under study have one prefix (48%), two (only 3%), and rarely three (just 0.1%), e.g.: atrodo, nežinau, nebuvo, priklauso, sudaro; nesupratau, pripažino, neišėjo, nepritaria; nebesulaukė, nepripažino. Most often, they also have other derivational morphemes. Consequently, the variety of morphemic structures among these verbs is considerable, comprising as many as 77 models. Nevertheless, as many as 42 percent of the verbs are accounted for by just ten models, so the other 67 models are certainly not well represented. The most characteristic models are: PRF (13%, e.g.: priklauso, sudaro, negali), PRSₖF (7%, e.g.: nebūtų, aptarsime, padėtų), PRSₖ (5%, e.g.: nebus, ateis, padės), PRS_dSₖF (4%, e.g.: pamatysi, pasakysiu, neturėtų), PRS_dSₖ (4%, e.g.: pažiūrėk, pasakys, padarys), PdRF (3%, e.g.: atsirado, pasitaiko, pasirodė), PRS_dF (3%, e.g.: negalėjo, pasakojo, neturėjo), PdRSₖ (1%, e.g.: pasidės, pasiimk, prisimink), PPRF (1%, e.g.: nebėra, nepavyko, nesupratau), PdRSₖF (1%, e.g.: pasimindavo, atsirasdavo, prisiminkime).
Slightly less than a third (31%) of the verbs used have derivational suffixes. Verbs with two derivational suffixes are uncommon (only 2% of all verbs), while three suffixes are very rare indeed, e.g.: turėjo, gyvena, galėjo, galvojo, kalbėjo, norėjau; tvirtina, vadovauja, abejoja, tikrina, žingsniavo; mokytojau, girtuokliauja. The structural variety of verbs with derivational suffixes is also considerable (79 models), but, once again, the ten most productive models account for as much as about 30 percent. These are: RS_dSₖF (5%, e.g.: turėtų, galėtų, reikėtų), RS_dF (5%, e.g.: turėjo, gyvena, galėjo), RS_dSₖ (5%, e.g.: daryti, kalbėti, matyti), PRS_dSₖF (4%, e.g.: pamatysi, pasakysiu, neturėtų), PRS_dSₖ (4%, e.g.: nustatyti, pasakyti, padaryti), PRS_dF (3%, e.g.: negalėjo, pasakojo, neturėjo), PdRS_dSₖ (1%, e.g.: atsisakyti, išsiaiškinti, įsivaizduoti), PdRS_dF (1%, e.g.: įsivaizduoji, įsigalioja, susiklostė), RS_dS_dF (1%, e.g.: tvirtina, reikalauja, vadovauja), RS_dFd (1%, e.g.: vadinasi, rūpinasi, formuojasi).
Slightly more than a tenth (12%) of the verbs are reflexive. They can be grouped into 47 morphemic structure models, but ten models account for nearly ten percent: PdRF (3%, e.g.: atsirado, pasitaiko, pasirodė), RFd (1%, e.g.: skiriasi, laikosi, stengiasi), PdRSₖ (1%, e.g.: pasirinkti, atsirasti, atsižvelgti), PdRSₖF (1%, e.g.: atsimindavo, atsirasdavo, pasimsiu), PdRS_dSₖ (1%, e.g.: atsisakyti, išsiaiškinti, pasižiūrėti), PdRS_dF (1%, e.g.: susiklostė, apsigyveno, susirūpino), RSₖd (0.5%, e.g.: elgtis, kreiptis, imtis), RS_dFd (0.5%, e.g.: vadinasi, rūpinasi, naudojasi), PdRS_dSₖF (0.5%, pasidarydavo, įsivaizduokite, pasitaikydavo), RS_dSₖd (0.5%, e.g.: naudotis, mokytis, laikytis).
Lithuanian has few compound-root verbs (only 89 out of nearly 93,000; they make up less than half a percent). Most often, these verbs also have a derivational
suffix; some also have a prefix or are reflexive. The structural variety of these verbs is considerable (24 models), but the models are unproductive, e.g.: bendradarbiauti (RcRSdSk), senberniauti (RRSdSk), nufotografuoti, nepiktnaudžiaujant, padvigubėjus (PRRSdSk), pusryčiavo, badmiriauja, viešpatavo (RRSdF), nufotografavo (PRRSdF), bendradarbiaudamas, duofikuojama, keliaklupščiaujančių (RcRSdSkF), bendradarbiaujant (RcRSdSk), netušči[a]žodžiaudamas (PRcRSdSkF), sudaiktavardėję (PRcRSdF).
CONCLUSIONS
The 93,000 verbs examined were classified into 116 morphemic structure models, whose productivity varies considerably: verbs belonging to the most productive model (RF) make up one fifth of all examples studied, while as many as 26 models have only one or two examples each.
Verbs may consist of between one and nine morphemes. Verbs with one morpheme and those with more than five morphemes are rare. Thus, verbs with 2–4 morphemes are most characteristic of Lithuanian (as many as 89%).
The formation of the forms themselves affects the morphemic structure of verb forms: infinitives and converbs can end only in an inflectional suffix.
Although many structural models were identified, a quarter of them (26 models) account for as much as 95 percent of all the verbs analyzed. The most characteristic models are: (P)RF – about 33 percent, (P)RSkF – about 15 percent, (P)RSk – about 11 percent. Thus, the verb forms most commonly used in Lithuanian are those without derivational suffixes, with or without a prefix, and containing one or two inflectional affixes.
In the material examined, one third (33%) of the verbs are simple; about 28% have one or two (much less often) prefixes; about 15% have a derivational suffix; 12% have prefixes and a derivational suffix; and about 6% have a prefix and are reflexive. Verbs with other combinations of derivational morphemes account for just one percent. It was observed that verbs with more than two derivational affixes are not characteristic of Lithuanian.
In summary, the variety of combinations in the morphemic structure of Lithuanian verbs is extensive, but the core of usage consists of 10–15 models (about 75 and 85 percent, respectively) of verbs with 2–4 morphemes.
102
Acta Linguistica Lithuanica LXIV–LXV
Page 17
Lietuvių kalbos veiksmažodžių morfeminė struktūra
LITERATURE AND SOURCES
A k e l a i t i e n ė G r a ž i n a 1996: Morfonologinės balsių kaitos žodžių daryboje. Vilnius: Vilniaus pedagoginio universiteto leidykla.
A k e l a i t i e n ė G r a ž i n a 2001: Šaknies balsių kaita: morfemų modeliai ir struktūriniai šaknies tipai. – Žmogus ir žodis. Didaktinė lingvistika 1(3), 3–9. Prieiga internete: http://www.biblioteka.vpu.lt/zmogusirzodis/PDF/didaktinelingvistika/2001/akel3-9.pdf.
A k e l a i t i e n ė G r a ž i n a 2008: Priebalsines priesagas turinčių išvestinių vardažodžių šaknies balsių kaita. – Žmogus ir žodis. Didaktinė lingvistika 10(1), 5–9. Prieiga internete: http://www.minfolit.lt/arch/14001/14402.pdf.
A m b r a z a s V y t a u t a s, red., 1996: Dabartinės lietuvių kalbos gramatika. Vilnius: Mokslo ir enciklopedijų leidybos institutas.
Corpus of Contemporary Lithuanian. Available online: http://tekstynas.vdu.lt.
K a r o s i e n ė V i d a Z i t a 2004: Bendrinės lietuvių kalbos vardažodžio šaknies struktūra. Vilnius: Vilniaus universiteto leidykla.
K a u k i e n ė A u d r o n ė 1994–2002: History of the Lithuanian Verb 1–2. Klaipėda: Klaipėda University.
K e i n y s S t a s y s 2009: Bendrinės lietuvių kalbos morfemika. Vilnius: Vilniaus pedagoginio universiteto leidykla.
K r u o p i e n ė I r e n a 2000: Bendrinės lietuvių kalbos veiksmažodžio šaknies struktūra. Dak-taro disertacija. Vilnius: Vilniaus universitetas.
K r u o p i e n ė I r e n a 2000: The root structure of verbs with a stable center. – Kalbotyra 48(1), 71–82.
K r u o p i e n ė I r e n a 2001: The coordination of initial and final consonants in verb roots. – Kalbotyra 50(1), 67–82.
K r u o p i e n ė I r e n a 2002: The root structure of verbs with a variable center. – Kalbotyra 51(1), 89–97.
K r u o p i e n ė I r e n a 2005: Features of the triple initial consonants of verb roots. – Kalbotyra 54(1), 73–82.
K u o s i e n ė M o n i k a 1986: Morfeminė analizė. Šiauliai: Šiaulių Kazio Preikšo pedagoginis institutas.
L e v ā n e K r i s t ī n e, S p e k t o r s A n d r e j s 2000: Morphemic Analysis and Morphological Tagging of Latvian Corpus. – LREC2000 Proceedings, 1095–1098. Available online: http://gandalf.aksis.uib.no/non/lrec2000/pdf/107.pdf.
Paikens Pēteris 2007: Lexicon-Based Morphological Analysis of Latvian Language. – Proceedings of the 3rd Baltic Conference on Human Language Technologies, 235–240.
Polikarpov Anatoliy A. 2000a: Menzerath’s Law for Morphemic Structures of Words: A Hypothesis for the Evolutionary Mechanism of its Arising and its Testing. – An abstract of a paper for presenting at Qualico’2000 – Quantitative Linguistics Conference. Prieiga internete: http://www.philol.msu.ru/~humlang/articles/menz_law.htm.
Polikarpov Anatoliy A. 2000b: Chronological Morphemic and Word-Formational Dictionary of Russian: Some System Regularities for Morphemic Structures and Units. – Linguistische Arbeits Berichte 5. Leipzig, 201–212.
Spoken language recordings database. Online access: http://donelaitis.vdu.lt/garsynas/.
Sarkans Ugis 1996: Morphemic and Morphological Analysis of the Latvian Language. – Proceedings of the Forth conference on Computational Lexicography and Text Research, 219–225. Prieiga internete: http://www.ailab.lv/users/usarkans/complex.html.
Urbutis Vincas 2009: Žodžių darybos teorija. Vilnius: Mokslo ir enciklopedijų leidybos institutas.
The Morphemic Structure of Lithuanian Verbs
SUMMARY
The aim of this research was to identify and to describe the morphemic structure of Lithuanian verbs from the quantitative point of view. Quantitative analysis of the morphemic structure of Lithuanian verbs was not addressed by previous research. The investigation material consisted of all verb tokens (nearly 93 thousand) from the specially selected subset of Lithuanian language corpus. All verbs within this data set have been morphologically annotated, i.e. morphemes have been identified and their boundaries have been labeled.
Verbs have been grouped into 116 categories according to their morphemic structure. Frequencies of these categories were found to be very different. The most frequent category was the root-suffix model which represented one fifth of all verb occurrences, whereas 26 least frequent models each covered just one or two verbs.
Verbs were found to be composed from one to nine morphemes. However, the majority of verbs (89 percent) were composed from two to four morphemes.
Despite the greater number of models of morphemic structure, one quarter of them (26 models) was enough to cover 95 percent of all analyzed verbs. The most typical models were: (prefix)-root-flexion (33 percent), (prefix)-root-suffix-flexion (15 percent), (prefix)-root-suffix (11 percent). In summary, the non-suffix both prefixed and non-prefixed verb forms composed
104
Acta Linguistica Lithuanica LXIV–LXV
Page 19
Lietuvių kalbos veiksmažodžių morfeminė struktūra
of one to two inflectional affixes were the most common models of morphemic structure of Lithuanian verbs.
Investigations showed that primary verbs amount to one third of all verbs, verbs with one or two prefixes amount to 28 percent, verbs with a suffix amount to 15 percent, verbs with prefixes and a suffix amount to 12 percent of all verbs. In general verbs have no more than two compositional affixes.
Although the diversity of morphemic structures of Lithuanian verb is great, the core of the usage is covered by 10 to 15 models (75 and 85 percent coverage respectively).
