Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos

LOÏC BOIZOU Vytautas Magnus University Research interests: computational linguistics, comparative linguistics, corpus linguistics. JOLANTA KOVALEVSKAITĖ Vytautas Magnus University Research interests: corpus linguistics, phraseology, lexicology and lexicography. ERIKA RIMKUTĖ Vytautas Magnus University Research interests: corpus linguistics, computational linguistics, automatic morphological and syntactic analysis, language culture, morphemics, multiword expressions. GRAMMATICAL CATEGORIES OF LITHUANIAN TWO-WORD MULTI-WORD EXPRESSIONS Grammatical Categories of Lithuanian Two-Word Multiword Expressions ABSTRACT This article analyzes two-word multi-word expressions that are presented in two or more forms in the Dictionary of Lithuanian Noun Phrases. It discusses the features of grammatical-form usage in these expressions, which are considered typical collocations. The aim of the study is to provide a detailed analysis of the grammatical categories of two-word expressions, with a particular focus on cases, and to describe how cases correlate with the syntactic relations they express. The article presents statistical data on the distribution of grammatical categories of verbs and nominal words and compares them with the use of grammatical forms in written and spoken Lithuanian corpora. KEYWORDS: Lithuanian, multi-word expressions, grammatical categories, Dictionary of Lithuanian Noun Phrases. Straipsniai / Articles 115 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ ABSTRACT This study focuses on two-word MWEs expressed by several grammatical forms in the Dictionary of Lithuanian Nominal Phrases. We analyse the distribution of grammatical categories in the two-word MWEs, which are mainly understood as typical collocations. The study aims to provide a detailed documentation of the usage of grammatical categories, with a specific emphasis on the case category, in order to show the correlations between each case and a particular syntactic function. The paper contributes to the description of grammatical features of Lithuanian MWEs by providing the quantitative distribution data of verbal and nominal categories, which are then contrasted with the usage of the same categories in written and spoken Lithuanian corpora. KEYWORDS: Lithuanian language, multi-word expressions, grammatical categories, Dictionary of Lithuanian Nominal Phrases. 1. INTRODUCTION Lithuanian multi-word expressions (hereafter MWEs) have been studied extensively: limiting ourselves to studies that analyze the structure and grammar of various types of MWEs, we can mention the works of Justina Bagdonaitė (2010), Agnė Bielinskienė (2010), Giedrė Jarašiūnaitė (2004), Jolanta Kovalevskaitė and Erika Rimkutė (2010), Rūta Marcinkevičienė and Ginarė Grigonytė (2005), among others. This article continues the series on the grammar of nominal two-word multi-word expressions in the Dictionary of Lithuanian Noun Phrases (hereafter LKDFŽ or the dictionary1): the authors of this study have discussed the difficulties involved in automatically determining the citation forms of multi-word expressions included in LKDFŽ (Boizou et al. 2015a), described the morphological and syntactic features of two-word MWEs in LKDFŽ (Boizou et al. 2015b), and analyzed the grammar and semantics of two-word MWEs listed in LKDFŽ in a single form (Rimkutė et al. 2015). This article analyzes two-word expressions listed in LKDFŽ in two or more forms, without additions or truncations, comprising 19,816 expressions in total. Most of these are combinations of two nouns (e.g., most residents, most of the residents; the golden age, the golden age, the golden age; Baltic 1 LKDFŽ is the first Lithuanian MWE dictionary compiled on the basis of a corpus. It consists of meaningful and grammatically correct combinations and phrases that are frequently used together and contain at least one noun (for more on LKDFŽ and its compilation principles, see Marcinkevičienė 2010; Rimkutė et al. 2012; Boizou et al. 2015b). 116 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos road, the Baltic Way, on the Baltic Way), as well as in combinations consisting of an adjective and a noun (a serious problem, serious problems; cash, cash), and a verb and a noun (opens the way, opened the way; when making a decision, will make a decision). In an earlier study (Boizou et al. 2015b), these two-word combinations were classified into the following groups, based on The Grammar of Contemporary Lithuanian (1996): 1) reciprocal relations: a) object (0.08% of all combinations), b) subject (0.33%); 2) coordinate (0.16%); 3) subordinate relations: a) object (3.42%), b) subject (0.16%), c) adverbial (0.30%), and d) attributive (95.55%)2. This study focuses primarily on the features of grammatical-form usage that emerge in the two-word combinations in LKDFŽ (which can be regarded as typical collocations). The article presents statistical data on the distribution of the grammatical categories of all verbs and nominal words, and shows how a particular grammatical category is used to form syntactic relations. Particular attention is paid to cases; the correlation between cases and the syntactic relations they express is discussed. The distribution of parts of speech and cases in the two-word MWEs presented in LKDFŽ was compared with data from spoken and written language on the basis of corpora of these two varieties, while other grammatical categories were compared with written language, since no data were found on the distribution of all grammatical categories in spoken language. To represent written language, a 1-million-word morphologically annotated corpus of Lithuanian was selected (see Rimkutė 2006, hereinafter referred to as the corpus); to represent spoken language, the Corpus of Spoken Lithuanian (Dabašinskienė 2009). It is hoped that comparing the distribution of grammatical categories in MWEs in written and spoken language will help determine what is characteristic of formulaic language composed of MWEs (formulaic language, see Wray 2002); the extent to which grammatical patterns of Lithuanian are reflected in formulaic language, and what differences and similarities can be observed. 2. DISTRIBUTION OF PARTS OF SPEECH AND GRAMMATICAL CATEGORIES Table 1 clearly shows that nouns predominate among the two-word MWEs analyzed from LKDFŽ—the proportions of parts of speech are determined by the fact that the dictionary 2 Kiti autoriai, pavyzdžiui, Vitas Labutis (1998), laikosi kiek kitokios klasifikacijos: šiame darbe prijungiamojo ryšio požymio junginius atitinka V. Labučio atributiniai junginiai, o tarpusavio sąsajos ryšio objektinius – predikatyvų (Labutis 1998: 90–94). Straipsniai / Articles 117 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ includes only combinations containing at least one noun (although it is not necessarily the main constituent of the combination)3. Thus, when analyzing two-word MWEs, no different results could be expected, and they are quite unlike the data for written and spoken language: nouns account for about 39% in the written-language corpus (Rimkutė 2006: 27) and 16.2% in the spoken-language corpus (Dabašinskienė 2009: 9). Because of the large number of nouns among the MWEs analyzed, adjectives are also abundant (since they agree with nouns)—there are twice as many as in the corpora (7.33% in the written-language corpus and 2.8% in the spoken-language corpus). The distribution of other parts of speech likewise differs from the written- and spoken-language data; cf. verbs—20.5 / 22.84%, numerals—0.96 / 1.9%, pronouns—8.7 / 16.9%. Among the main parts of speech, the number of verbs differs considerably, since the research material contains very few verb-based combinations. Although, as established in the article (Boizou et al. 2015b: 121), the verb + noun pattern accounts for 9.99%, it is still far behind the two most frequent patterns: noun + noun (49.07%) and adjective + noun (32.2%). 1 TABLE. Distribution of parts of speech Part of speech No. Percent Nouns 29559 74,55 Adjectives 6388 16,11 Verbs 2061 5,20 Numerals 738 1,86 Pronouns 640 1,61 Invariable parts of speech 31 0,08 Abbreviations 234 0,59 Total 39651 100,00 Among the invariable parts of speech, only adverbs, prepositions, and particles were found in two-word MWEs. As with adjectives, adverb MWEs would most likely be 3 Ankstesniame tyrime (Boizou ir kt. 2015b: 123) nustatyta, kad pagal pagrindinį dėmenį junginiai pasiskirsto taip: daiktavardiniai – 96,4 proc., veiksmažodiniai – 3,59 proc., būdvardiniai – 0,02 proc. 4The first figures indicate how many instances of a particular part of speech were found in the written-language corpus, while the figures after the slash indicate how many were used in the spoken-language corpus. 118 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos A higher proportion might be expected if the research material had included adverb + adjective and adverb + verb model combinations. Prepositions and particles are used mainly in multiword expressions or combinations with unclear syntactic relations (these are not discussed in detail in this article). Prepositions, adverbs, and particles make up 0.08 percent of two-word MWEs, compared with more than 13 percent in the written-language corpus; they account for an even larger share—26.4 percent—in the spoken-language corpus. Two-word MWEs contain very few invariable parts of speech, owing to the tendencies governing how parts of speech combine: in meaningful two-word combinations in which one element is a noun, the other element can usually only be a nominal or a verb. Abbreviations and shortened forms classified as parts of speech account for almost 2 percent in the corpus, but only 0.59 percent in the combinations analyzed; nevertheless, these lexical units constitute a fairly substantial share among the other parts of speech. This can be explained by the fact that abbreviations occur in combinations forming various syntactic relations, e.g.: ŽIV nešiotojų, ES šalys, narystė NATO, B limfocitai, C grupė. They are most often used as modifiers. An analysis of the variety of parts of speech occurring in different syntactic relations (see also Boizou et al. 2015b) allows the following generalizations: only nouns occur in coordinate-relation combinations; only nouns and verbs occur in reciprocal-relation subject and object combinations, as well as in subordinate-relation subject combinations. A greater variety of parts of speech was observed among subordinate-relation adverbial combinations: besides nouns and verbs, adverbs and abbreviations were also found in them. Subordinate-relation object combinations consist mainly of nouns and verbs, but adjectives, numerals, and pronouns also occur. All the parts of speech presented in Table 1 (except invariable ones) were found in subordinate-relation attribute combinations, in which nouns predominate. The fact that the distribution of parts of speech in the two-word MWEs analyzed does not match the patterns of written and spoken Lithuanian usage leads to quite different results for the distribution of grammatical categories (compared with corpus data). Nevertheless, in some cases (especially cases), usage patterns are very similar (see sections 2.1 and 2.2 for more detail). 2.1. Verb Forms 2.1.1. Distribution of Finite and Non-finite Forms The two-word combinations analyzed include 379 finite forms and 280 non-finite forms. We note that among the non-finite forms Straipsniai / Articles 119 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ verbs, participles are not included: participles are most often used as agreeing words and are therefore counted as adjectives.5 Among non-finite forms, infinitives were the most numerous (see Table 2), while semi-participles were the least common. The pattern is similar to that in the written-language corpus: infinitives are the most frequent form in the corpus (if participles are excluded), accounting for 76.73 percent; converbs rank second at 14.61 percent, and semi-participles third at 8.56 percent. The much higher proportion of converbs among the lexical combinations analyzed than in the corpus can be explained by the fact that this grammatical form occurs in various combinations: subordinate-relation subject, object, and adverbial combinations. The less frequent use of semi-participles may be due to the requirement in LKDFŽ that one element be a noun (since this was the principle used to select MWEs); consequently, only those semi-participles that most often govern nouns were included (e.g., gulėdamas lovoje, turėdami tikslą, duodamas interviu), whereas those that combine with adverbs were excluded—even though such combinations are quite common in Lithuanian (e.g., greitai bėgdamas, sunkiai nešdami). TABLE 2. Distribution of Non-finite Forms Form No. Percent infinitives 196 70,0 converbs 71 25,4 semi-participles 13 4,6 Total 280 100.0 Most commonly, converbs (189) occurred in object phrases with a subordinating relationship, e.g.: to get money, to assume obligations. Participles were also most frequently used in phrases of this type (38 instances), e.g.: having completed studies, having committed a crime. Participles are used somewhat less frequently (32 instances were found) in subject phrases with a subordinating relationship, e.g.: as time passes, after an accident occurs. Half-participles are also more common in object phrases, e.g.: giving an interview, having a goal. For the cases used with verb forms, see Section 2.2.1. The following discussion addresses finite forms and presents statistical data on the distribution of grammatical categories characteristic of verbs. 5 It is understandable that participles may be used not only as agreeing attributes but also as governing words, e.g.: shocked, gritting one's teeth. It is not accurate to classify such participles as adjectives, but instances of this usage are quite few, and therefore have little effect on the results. 120 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos 2.1.2. The Category of Tense Table 3 presents only the tenses of finite forms; the tenses of converbs and participles are not included here. These data can be compared with data from a morphologically annotated corpus (see Rimkutė 2006: 35). One important difference emerges: among the two-word MWEs analyzed from LKDFŽ, the simple past tense predominates (almost 53%), whereas in the corpus mentioned, the present tense accounts for a similar proportion (52.5%). The second most frequent tenses are distributed as follows: in two-word MWEs, the present tense (40.85%); in the corpus, the simple past tense (37.8%). The future tense is much less frequent: 6.2% in two-word MWEs and 7.9% in the corpus. The rarest is the frequentative past tense, which accounts for 2.7% in the corpus and less than one percent in two-word MWEs. 3 TABLE. Distribution of tenses Tense | No. | Percent simple past | 190 | 51.9 present | 153 | 41.8 future | 22 | 6.0 frequentative past | 1 | 0.3 Total | 366 | 100.0 2.1.3. Mood category The distribution of moods among the two-word verb forms of MWEs used in LKDFŽ (see Table 4) corresponds to the patterns of present-day written Lithuanian: more than 96% are indicative forms (cf. 91.5% in the morphologically annotated corpus (Rimkutė 2006: 35)). Most (290) forms of this mood were found in object combinations linked by subordination, e.g.: had an influence, celebrated the Eucharist. 63 indicative forms were used in subject combinations involving mutual association: the time came, music was playing. Both the subjunctive and imperative forms occur in object and adverbial combinations linked by subordination (e.g.: would create conditions, postpone until tomorrow), as well as in subject and object combinations involving mutual association (e.g.: an opportunity would arise, be a person). Straipsniai / Articles 121 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ TABLE 4. Distribution of moods Mood No. Pct. indicative 366 96,6 imperative 7 1,8 subjunctive 6 1,6 Total 379 100.0 2.1.4. The categories of number and person As mentioned earlier (see also Table 4), most indicative forms were found in object multiword expressions with a subordinating relation; consequently, most person forms were also found here. Table 5 clearly shows that the third person predominates—a tendency also characteristic of the morphologically annotated corpus, where the third person accounts for 81.3 percent (Rimkutė 2006: 35). There were 286 instances of the third person in object multiword expressions with a subordinating relation, e.g.: washes the dishes, shakes his head. 62 instances were found in subject multiword expressions with a reciprocal relation, e.g.: an opportunity arose, the conference will take place. 11 third-person forms were found in adverbial multiword expressions with a subordinating relation, e.g.: will participate in the elections, held in their hands. TABLE 5. Distribution of number and person Number and person No. Percent 3rd person 359 94,7 1st person singular 8 2,1 1st person plural 5 1,3 2nd person plural 5 1,3 2nd person singular 2 0,5 Total 379 100,0 7 instances of the first person singular were found in object multiword expressions with a subordinating relation, e.g.: I had an opportunity, I gave a lecture. The second person singular is used less frequently; both instances were found in object multiword expressions with a reciprocal relation: be a man, be a human being. The first person plural is more frequent in object multiword expressions with a subordinating relation, e.g.: we have an opportunity, we draw attention. All instances of the second person plural were found in object multiword expressions with a subordinating relation, e.g.: press the button, do not miss 122 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos occasion. As the examples show, imperative forms predominate among the singular and plural second-person forms. The distribution of reflexive and non-reflexive forms in the two-word MWEs in LKDFŽ corresponds to general patterns of written language: reflexive forms account for 10.6% in the dictionary and 12.3% in the corpus; non-reflexive forms are predominant both in the dictionary and in the corpus, at 89.4% and 87.7%, respectively.6 To summarize the data presented in Section 2.1.1, it can be stated that the use of grammatical forms of verbs essentially follows the patterns of Lithuanian usage. Nevertheless, the overview of the categories of mood and person shows that the category member that predominates usually accounts for a larger proportion than the forms most frequently used in the corpus (for example, the past frequentative tense in MWEs). It may be assumed that these results are due either to the fact that fixed expressions were analyzed or to the number of grammatical forms being too small for clearer usage patterns to emerge. 2.2. Nominal forms 2.2.1. Case category The following describes how the use of each case is distributed, taking into account the syntactic relation it expresses. In counting cases, part of speech, gender, number, and other grammatical features were disregarded (for example, whether a noun is proper or common, reflexive or non-reflexive; whether a participle is pronominal or non-pronominal; whether an adjective is used in the positive or comparative degree, etc.). First, we present a general table of the cases of two-word multi-form and single-form MWEs in LKDFŽ, and of the frequency of cases for words occurring in the written-language corpus. Although this article analyzes two-word MWEs occurring in several inflectional forms, when discussing the distribution of cases, we also wish to draw attention to two-word expressions used in a single form7. Comparing the cases of multi-form and single-form MWEs is interesting because the case usage of multi-form MWEs, 6 All verb forms were counted, including deverbal nouns. 7 Single-form MWEs are those recorded in LKDFŽ in one inflectional form (for more, see Rimkutė et al. 2015). Some single-form MWEs are in fact completely frozen and are used as MWEs in only one form (dėmesio centre, konkurso keliu); one component of such MWEs (usually the second) is frozen, which determines the stability of the whole expression. Some single-form MWEs are recorded in LKDFŽ in only one (usually the most frequent) form, although they may inflect in actual usage, e.g.: skristi lėktuvu, atrankos rungtynių, Šventajame Rašte. Straipsniai / Articles 123 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ is likely to be closer to written language, while the use of cases in one-form MWEs should differ, since they reflect more entrenched language. TABLE 6. Distribution of cases Case | Multi-form compounds | One-form compounds | Written-language corpus No. | Proc. | No. | Proc. | No. | Proc. Nominative | 7170 | 18,66 | 5103 | 22,19 | 95075 | 26,38 Genitive | 19074 | 49,64 | 12160 | 52,89 | 147308 | 40,88 Dative | 2088 | 5,43 | 467 | 2,03 | 14664 | 4,07 Accusative | 5770 | 15,02 | 2255 | 9,81 | 59432 | 16,49 Instrumental | 1719 | 4,47 | 1628 | 7,08 | 22520 | 6,25 Locative | 2549 | 6,63 | 1310 | 5,70 | 19346 | 5,37 Vocative | 49 | 0,13 | 63 | 0,27 | 1714 | 0,48 Illative | 3 | 0,01 | 6 | 0,03 | 286 | 0,08 Total | 38422 | 100,00 | 22992 | 100,00 | 360345 | 100,00 As can be seen from Table 6, case usage in one-form compounds does not differ greatly from that in multi-form compounds. A few differences can be noted: the most frequent cases, i.e. the nominative and genitive, occur more often among one-form than among multi-form compounds, while the accusative in one-form compounds is almost twice as rare as in multi-form compounds; the dative is used much less often in one-form compounds. Attention should be drawn to the much more frequent use of the instrumental among compounds used in a single form. This can be explained by the fact that the instrumental of the second nominal constituent is often used as a postposition, i.e. as a fossilized form, e.g.: audros metu, direktoriaus teigimu, Rusijos atžvilgiu, rangos būdu, konkurso keliu. Aldona Paulauskienė (1989: 55) states that adverbialization in a particular case depends on the lexical meaning of the noun. For example, the instrumental in the compound šaukė balsu tends to become adverbialized, because this noun denotes not a concrete object but an abstract notion. A case turns into an adverb (or postposition) when its objective meaning disappears entirely and its adverbial meaning becomes primary. A. Paulauskienė (1989: 56) calls such fossilized noun forms non-paradigmatic; many such forms occur among one-form two-word MWEs. There are many more instances of the vocative among one-form than among multi-form compounds, because compounds in the vocative form are generally not inflectional forms of a particular MWE (which is why they are relatively few among two-word MWEs used in several forms), but rather specific lexical units that perform a defined communicative function: 124 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos of address or emotional expression, e.g.: Your Majesty / Excellency / Eminence; Mr President; Jesus, Mary; help me, God; oh, horror. Several compounds used in the vocative form were found in titles of works, e.g.: Dear Lithuania; Fly, little falcon; Ave Maria. Thus, the less frequently used cases (the vocative and the instrumental mentioned earlier) are more typical of more fossilized compounds. Next, let us compare the distribution of cases in two-word multi-form MWEs (as more usual and typical forms) with noun case usage in written language (see Table 6) and spoken language (drawing on the material presented by Ineta Dabašinskienė (2009) from the Corpus of Spoken Lithuanian). As I. Dabašinskienė (2009: 10) states, grammatical cases are used more often than semantic cases in both spoken and written language. In spoken- and written-language corpora, cases fall into two groups according to frequency of use: primary (nominative, genitive, accusative) and peripheral (dative, instrumental, locative, vocative). The spoken-language cases in the primary group are ranked by frequency as follows: nominative (singular nominative accounts for 31.6%, and plural nominative for 30%), accusative (singular accusative accounts for 26%, plural accusative for 22.8%), genitive (singular genitive accounts for 22.9%, plural genitive for 33.5%); in written language, the order is genitive, nominative, accusative (see Dabašinskienė 2009: 10 and Table 7). A contrast between these two language varieties emerges: the noun nominative is the most frequent case in spoken language, while the genitive is most frequent in written language. As can be seen from Table 6, the most frequent case, accounting for almost half of all the MWE cases analyzed (49.64%), is the genitive; the second most frequent is the nominative, and the third is the accusative. Compared with the frequency of the genitive in written language (40.88%), the figures are quite similar. The use of cases in two-word multi-form MWEs in LKDFŽ is closer to written language than to spoken language; this is unsurprising, since written-language texts make up the majority of the Corpus of Contemporary Lithuanian, from which the data for LKDFŽ were collected. The frequent use of the genitive among two-word LKDFŽ compounds (and in written language generally) can be explained by the genitive’s use as an uninflected modifier. This claim will also be supported by the frequency data presented below for the usage of all cases classified according to their use in expressing a particular syntactic relation (see Tables 7–12). The use of the nominative differs quite markedly between two-word MWEs and the corpus: 18.66% and 26.38%, respectively. This can be explained by the fact that only a small proportion of subject–predicate combinations can be considered fixed expressions, so relatively few nominative instances were included in the dictionary, most of them occurring in compounds characterized by a subordinating relation. The usage of all cases is discussed in detail below: how a particular case is distributed among compounds with a particular syntactic relation, and in which position—left or Straipsniai / Articles 125 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ on the right – is used. In the tables presented, cases are listed and discussed further in the text according to the highest number of occurrences in combinations with a particular syntactic relation. 2.2.1.1. Nominative Because the nominative is a frequently used case, and because combinations expressing the feature of a subordinative relation predominate among all the analyzed two-word combinations, the frequent use of the nominative is unsurprising. TABLE 7. Distribution of the nominative No. Syntactic relation Left, no.8 Left, percent Right, no. Right, percent Total Share among all combinations, percent9 1 Subordinative: feature 2159 30,9 4823 69,1 6982 18,7 2 Mutual relation: subject 15 23,4 49 76,6 64 100,0 3 Subordinate: object 52 98,1 1 1,9 53 6,6 4 Mutual relation: object 16 47,1 18 52,6 34 100,0 5 Coordinate 14 50,0 14 50,0 28 37,8 6 Subordinate: circumstance 8 88,9 1 11,1 9 11,4 Total 2264 4906 7170 As can be seen from Table 7, in two-thirds of cases the nominative is used on the right, e.g.: administration employees, the Esperanto language, while the left constituent is an uninflected modifier, most often expressed by a noun in the genitive. In those phrases where agreeing modifiers are used, both words agree fully, e.g.: both teams, engineering networks. Of course, here it may 8Left refers to the left constituent of an MWE, right—to the right constituent. In the case tables, percentages indicate what proportion of a particular syntactic relation a given case constitutes when occurring as the left and right member. 9In the last column of this and other case tables, percentages indicate what proportion of all analyzed two-word MWE phrases forming a particular syntactic relation is accounted for by that case (see also Table 13). 126 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos Other cases may also be used (see the examples for the other cases). In some combinations, the grammatical forms of both the first and the second component may change, e.g.: excise rates, excise rate. As expected, only the nominative is used in subject combinations expressing mutual relations. As Table 7 shows, the nominative is more commonly used on the right; that is, the usual word order in such combinations is verb + noun, e.g.: an accident happened, rain is falling. Nevertheless, there are also examples (admittedly, much rarer) in which the nominative is used on the left: children grow, the sun shines. Some combinations use invariable10 forms, e.g.: the sun rises; in others, the forms of the verb or noun change, e.g.: the conference will take place / took place, danger threatened / threatens; a person / people live, a person / people strive. When the distribution of cases was counted, the relatively frequent use of the nominative in object combinations with a subordinating relation was surprising: 53 were found (the third most frequent combination type). As Table 7 shows, the nominative occurs on the left in these combinations, e.g.: most residents, a multitude of people, nineteen years, a million litas. The examples suggest that these are not entirely typical object combinations (such as, for instance, combinations with participles: having thrown back his head, having defended a dissertation); they could partly be regarded as examples of adverbials of manner (quantity). In the Grammar of Contemporary Lithuanian (1996: 512), it is stated that nouns denoting quantity, substantival numerals, pronouns, and adverbs of quantity form object combinations, and that the genitive governed by them is called the genitive of quantitative content. Thus, on the basis of the grammar, the combinations mentioned above were assigned to object combinations with a subordinating relation. Only the nominative is used in object combinations expressing mutual relations, of which there are almost half as many as subject combinations—34. Most examples follow the noun + participle pattern, e.g.: voting begins, the decision has been made. As the examples show, both the noun and the participle are in the nominative; the noun may occur on either the right or the left. In only three combinations do the nouns and participles vary in number, e.g.: the issue is being resolved / the issues are being resolved; the study is being conducted / the studies are being conducted. Because of the specific nature of the coordinating relation, both components must agree in case, e.g.: left, right; Marxism, Leninism. The number of nouns may vary, e.g.: men, women; man, woman (the cases may also vary; see the subsections devoted to the other cases). 10 The point is that in LKDFŽ a particular combination is used in only one form, or one component is always used in a fixed form. It should be noted that different forms of the same combination may be assigned to different syntactic relations: for example, lyja lietus ‘rain is falling’ is assigned to subject combinations expressing mutual relations, whereas lyjant lietui ‘while rain is falling’ is assigned to subject combinations with a subordinating relation (for more details, see Boizou et al. 2015b: 126). Straipsniai / Articles 127 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ The nominative is likewise not particularly characteristic of adverbial combinations with a subordinating relation—these are the least frequent type of combinations containing the nominative, e.g.: approximately half, the war in Chechnya, NATO membership, ambassador in Lithuania / to Lithuania. In combinations of this type, the nominative is almost always used on the left. 2.2.1.2. Genitive Since non-agreeing modifiers are most often expressed in the genitive, the large number of genitive instances in attributive combinations with a subordinating relation is entirely unsurprising—nearly 19,000 (50.4% of all MWEs) (see Table 8). A non-agreeing modifier normally occurs on the left, a pattern also confirmed by the two-word combinations in LKDFŽ, e.g.: pedestrian crossing, radios. If both components of a combination are inflected, and if these inflected forms are recorded in LKDFŽ as forms of the same combination, the genitive may also occur on the right, e.g.: Radviliškis district, human body. TABLE 8. Distribution of the genitive No. | Syntactic relation | Left, number | Left,% | Right, number | Right,% | Total | Proportion of all combinations,% 1 | Subordinating: attribute | 12444 | 66,2 | 6358 | 33,8 | 18802 | 50,4 2 | Subordinating: object | 41 | 17,7 | 191 | 82,3 | 232 | 28,8 3 | Coordinating | 14 | 50,0 | 14 | 50,0 | 28 | 37,8 4 | Subordinating: adverbial | 10 | 83,3 | 2 | 16,7 | 12 | 15,2 Total | 12509 | | 6565 | | 19074 | Table 8 shows that the second most frequent type of combinations containing the genitive is object combinations with a subordinating relation. The first component is most often a verb form, e.g.: to allocate money, there was no space left, but combinations in which the genitive is required by a noun denoting quantity are also common, e.g.: in many countries, for ten minutes, fifteen years old. The genitive used on the right predominates in these combinations; the genitive occurs on the left when the first component is a noun denoting quantity, e.g.: of many countries, nineteen years. 128 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos There are nearly ten times fewer coordinate constructions than subordinate object constructions. Given the specific nature of constructions with this relation, both the first and the second constituent must be expressed in the same case. Examples with the genitive include: the mayor, the board; fruit, vegetables. A few examples in which the genitive is used in subordinate adverbial constructions:year after year, the war in Chechnya. As can be seen from Table 8, in these constructions the genitive occurs more often on the left. In only two cases is the genitive used on the right: a year’s time, yesterday’s day. 2.2.1.3. Dative In both spoken and written language, the dative is one of the least frequently used cases; it is also relatively rare among the two-word MWEs in LKDFŽ, accounting for only 5.43% of all cases (see Table 6). As already mentioned, this dictionary contains mostly constructions expressing the attribute relation in subordinate constructions, so it is unsurprising that most instances of the dative were also recorded in these constructions. As can be seen in Table 9, in two thirds of cases the dative occurs on the right, e.g.: a debt to customs, for eternal life. When constructions contain agreeing modifiers, both words are in the dative form. As noted earlier, constructions with non-agreeing modifiers are very common, which means that the dative occurs on the left less often than on the right. TABLE 9. Distribution of the dative No. | Syntactic relation | Left, no. | Left,% | Right, no. | Right,% | Total | Share of all constructions,% 1 | Subordinate: attribute | 649 | 32,7 | 1337 | 67,3 | 1986 | 5,3 2 | Subordinate: object | 35 | 58,3 | 25 | 41,7 | 60 | 7,5 3 | Subordinate: subject | 4 | 12,5 | 28 | 87,5 | 32 | 100,0 4 | Subordinate: adverbial | 2 | 33,3 | 4 | 66,7 | 6 | 7,6 5 | Coordinate | 2 | 50,0 | 2 | 50,0 | 4 | 5,4 Total | 692 | | 1396 | | 2088 | Subordinate object constructions are characterized by the fact that the dative can occur either on the left or on the right, although its use on the left is more usual, e.g.: Straipsniai / Articles 129 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ to represent the state, for several days, to finance the project. When the dative occurs on the left, the constructions most often express quantity, and their first constituent is a noun (as in the example for several days). In subordinate subject constructions, the dative is more usual on the right, e.g.: when spring arrived, when it rains. Constructions such as as the child grows, while the sun shines, as can be inferred from the limited data, are not a structure usual in Lithuanian for expressing subject relations. Only six instances were found in which adverbial constructions are expressed in the dative: approximately half, membership in the EU / NATO, postpone it until tomorrow; therefore, no more comprehensive generalizations can be made about the use of the dative in this type of construction. The same number of datives was also found in coordinate constructions, e.g.: to brothers, sisters; to the mayor, the board. 2.2.1.4. Accusative As noted more than once, because among two-word MWEs, combinations expressing the subordinating relation of attribute predominate, the accusative is most commonly used in precisely this type of combination (see Table 10), e.g.: considerable sums, Lithuanian society. The fact that in two-thirds of cases the accusative occurs on the right can be explained by the predominance of combinations with non-agreeing modifiers; consequently, only the second constituent can be inflected, e.g.: Lithuanian society / society / to society / society. TABLE 10. Distribution of the accusative No. | Syntactic relation | Left, no. | Left, percent | Right, no. | Right, percent | Total | Share among all combinations, percent | 1 | Subordinating: attribute | 1880 | 35,3 | 3442 | 64,7 | 5322 | 14,3 | 2 | Subordinating: object | 39 | 9,1 | 390 | 90,9 | 429 | 53,3 | 3 | Coordinating | 5 | 50,0 | 5 | 50,0 | 10 | 13,5 | 4 | Subordinating: adverbial | 3 | 33,3 | 6 | 66,7 | 9 | 11,4 | Total | 1927 | 3843 | 5770 A comparison of the use of all cases (see Table 13) shows that the accusative is the most typical case for expressing subordinating object combinations (accounting for 53.3 percent of all MWEs with this relation). It should be mentioned that in more than 90 percent 130 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos of cases the accusative occurs on the right, since the most common combinations of this type have the structure verb + noun, e.g.: to prevent, had an opportunity. The accusative is used on the left when the first constituent is a numeral expressing quantity, e.g.: a hundred years, a thousand litas. However, a few somewhat atypical combinations were found in which the accusative is used on the left: impact on the environment (because the noun governs it), the problem to solve (because of inverted word order). Only 9 subordinating adverbial combinations and 10 coordinating-relation examples were found; these combinations are illustrated by the following examples: bank accounts, yesterday morning (adverbial) and windows, doors; arms, legs (coordinating). 2.2.1.5. Instrumental Most combinations expressing the subordinating relation of attribute are inflected, and therefore they also occur in the instrumental form, e.g.: with the pleasures offered, king of Poland. As noted, combinations with a non-agreeing modifier predominate; therefore, the instrumental occurs twice as often on the right as on the left (see Table 11). The summarized data for all cases (see Table 13) show that the instrumental (excluding the vocative) is the least frequent case used to express attribute combinations, since it accounts for only 4.5 percent. TABLE 11. Distribution of the instrumental No. Syntactic relation Left, no. Left, percent Right, no. Right, percent Total Share among all phrases, percent 1 Subordinate: attribute 628 37,2 1060 62,8 1688 4,5 2 Subordinate: object 3 10,7 25 89,3 28 3,5 3 Subordinate: circumstantial adjunct 1 33,3 2 66,7 3 3,8 Total 632 1087 1719 Because of the rare use of the instrumental case, only 28 instances of this case were found in subordinate object phrases. The most common structure of such phrases is verb + noun, e.g.: to find guilty, shrugged their shoulders. A few examples were found in which verbal nouns govern the instrumental case, e.g.: being oneself, self-confidence. On the left, the instrumental is used only in phrases expressing quantity, e.g.: one hundred percent. It should be noted that among subordinate circumstantial phrases with the instrumental case, two time-expressing combinations were found: month after month, year after year. Straipsniai / Articles 131 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ Another phrase in which the instrumental is used on the left is ambassador in Lithuania. 2.2.1.6. Locative When classifying the data for the study described in Boizou et al. 2015b, many questions arose about where to assign phrases with the locative; these uncertainties arose because two-word phrases were analyzed. If three-word phrases had been analyzed, verbs would probably have been used with many of them, and their meanings would have made it possible to determine the phrase type more precisely. When classifying two-word phrases with the locative, as shown in Table 12, most of them were assigned to phrases expressing an attributive relation, e.g.: in beauty contests, in the upper part, in a department store. At first glance, these may appear to be phrases denoting location. They would be, if verbs were used alongside them, e.g.: participate in beauty contests, shop in a department store. On the other hand, these phrases are autonomous and can be considered typical collocations, and are therefore treated as inflectional forms of phrases expressing an attribute. Thus, all forms of attribute-expressing phrases (regardless of case) were assigned to the aforementioned phrase type, e.g.: third place, in third place, the third place, in the third place, third places, the third places, a third place, the third place, the third places. For this reason, phrases with the locative were also included among phrases expressing an attributive relation. As the examples given above show, the locative is used in phrases with both agreeing and non-agreeing modifiers; this case occurs more often on the right. TABLE 12. Distribution of the locative No. Syntactic relation Left, no. Left,% Right, no. Right,% Total Share among all phrases,% 1 Subordinating: attribute 716 28,6 1786 71,4 2502 6,7 2 Subordinating: circumstance 2 5,0 38 95,0 40 50,6 3 Coordinate 2 50,0 2 50,0 4 5,4 4 Subordinate: object 3 100,0 0 0,0 3 0,4 Total 723 1826 2549 As might be expected, the locative was used most frequently of all cases in subordinate circumstantial phrases—accounting for 50.6 percent (see 132 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos Tables 12 and 13). In almost all cases, the locative occurs on the right; it can be governed by both verbs and nouns, e.g.: lying in bed, accounts in banks, hangs in the air, left in peace, having in mind. It occurs on the left only in two forms of one phrase: planned / envisaged in the project. Among the analyzed two-word MWEs with the locative, four examples of coordinate relation were found: in Vilnius, in Kaunas; in Latvia, in Estonia. As the examples show, a specific place is expressed in every case. Only three object phrases in a subordinating relation with the locative were found: in many states, in many countries, in the Letter to the Romans. As noted earlier, although these phrases could be considered expressions of an adverbial of place, taking into account the meaning of the whole phrase (rather than that of any one of its inflected forms), such phrases are classified as object phrases. 2.2.1.7. Other cases Since the vocative is one of the least frequently used Lithuanian cases (cf. it accounts for 0.48% in the written-language corpus; in the spoken-language corpus, the plural vocative accounts for 0.3%, while the singular vocative accounts for 3.3%), it is not surprising that only 49 occurrences of this case were found in the LKDFŽ in total. All of them occurred in phrases expressing a subordinating relation, e.g.: honorable chairman, dear reader, lord God, dear colleagues. Twenty-four occurrences of the vocative appeared on the left, and 25 on the right; thus, in almost all cases the vocative is used in phrases with agreeing attributes. Three occurrences of the illative (distinguished from the locative) were also found in phrases expressing a subordinating relation, e.g.: into the ownership of the municipality, into the ownership of the state. 2.2.1.8. Summary of case usage Summarizing the results for case usage (excluding the vocative and illative), Table 13 shows which cases predominate in particular types of phrases. As noted earlier, because phrases expressing a subordinating relation predominate among the two-word LKDFŽ phrases, the statistical data on case usage are determined by the usage of this phrase type; nevertheless, conclusions can also be drawn about case usage in expressing phrase types other than those expressing an attribute. Table 13 shows that only the nominative was used in the LKDFŽ two-word MWEs to express mutual relations in both subject and object phrases. This can be explained by the fact that, according to the DLKG classification, phrases of these types are most often formed only with the nominative. Subject phrases in a subordinating relation are formed only with the dative. Based on the data for two-word LKDFŽ MWEs, it can be stated that the most typical Straipsniai / Articles 133 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ The structure of subordinate subject combinations is a converb + a dative noun. If we did not adhere to the classification presented in the traditional DLKG, we could classify the related subject and object combinations, as well as subordinate subject combinations, together, thus better highlighting the cases used to express subject relations: the nominative and the dative (with converbs). These three types of combinations are also united by the fact that they consist only of nouns and verbs. The clearest are combinations with a subordinating relation (except for subject combinations), in which the distribution of cases is most evident. In object, circumstantial, and attributive combinations, one dominant case stands out (accounting for more than half of all usage: the accusative, locative, and genitive, respectively). In object combinations with a subordinating relation, the accusative dominates, accounting for more than 53 percent. The second most frequent is the genitive (28.8 percent), and the third is the dative (7.5 percent). As noted earlier, the use of the nominative in object combinations was somewhat surprising (the nominative accounts for 6.6 percent here), but this can be explained by the fact that quantity-expressing combinations occur. If the traditional grammatical approach were not followed and quantity-expressing combinations were classified as circumstantial, the distribution of cases in object combinations would be clearer and more consistent. More than half of circumstantial combinations with a subordinating relation are formed with the locative; the second most frequent is the genitive (15.2 percent), followed by the nominative and accusative (both cases account for 11.4 percent each). The use of the nominative to form circumstantial combinations is somewhat surprising, but perhaps because of the small number of combinations (only 77 circumstantial combinations were found in total), patterns of case use in this type of combination do not emerge. Circumstantial combinations also have the characteristic that one constituent (usually the left-hand one) is inflected (its case, tense, number, or person changes), while the other constituent (usually a locative, sometimes an adverb) remains unchanged, e.g.: atstovybės / atstovybė Lietuvoje; dalyvavo / dalyvauti / dalyvaus rinkimuose; vakar ryte / rytą; maždaug pusė / pusę. Words used in other types of combinations are characterized by greater inflectional variability. In combinations expressing an attribute with a subordinating relation, all cases occur, since this type of combination was found most frequently (accounting for more than 95 percent of all combinations), making it possible to identify patterns of case use. These confirm (and almost reproduce) the data presented in Table 6 on the distribution of cases in two-word multi-form PJs; by frequency, the cases rank as follows: genitive – 50.4 percent, nominative – 18.7 percent, accusative – 14.3 percent, locative – 6.7 percent, dative – 5.3 percent, and instrumental – 4.5 percent. 134 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos As has been noted repeatedly, the dominance of the genitive in these and other combinations is due to the fact that two-word combinations in LKDFŽ are most often formed with non-agreeing modifiers. The frequency of the other cases confirms the case-frequency patterns of spoken and written Lithuanian described in earlier studies. Coordination can be expressed by all cases; it is true that the instrumental does not occur in this type of combination, but that does not mean that such usage is impossible for the instrumental. Rather, because the instrumental is rarely used, two-word combinations with a coordinate relation and this case were not included in LKDFŽ. Since coordinate combinations are not characterized by a specific expression of grammatical forms, only general patterns of case frequency can be discussed here (these reflect case usage in a corpus more than in a dictionary). These patterns are also confirmed by the data presented in Table 13: the most frequent are the nominative and genitive (almost 38 percent each), followed by the accusative (more than 13 percent). 2.2.2. Other Nominal Categories This subsection discusses the categories of gender, number, degree, and definiteness. In counting these categories, some verb forms were also included. As already noted, most of the two-word PJs analyzed belong to combinations expressing an attribute with a subordinating relation; consequently, this type of combination largely determines the distribution of the grammatical categories discussed below. 2.2.2.1. Gender category When calculating the gender category of nominal words, data on participles and half-participles were also included, since these forms were most often used as agreeing words in the analyzed two-word MWEs. The distribution of genders in the two-word combinations in LKDFŽ is shown in Table 14. TABLE 14. Distribution of gender forms Gender No. Percent masculine 21039 54,58 feminine 17500 45,4 neuter 6 0,02 Total 38545 100.00 The frequency of masculine forms hardly differs from the corpus data—here they account for 57.2 percent; the use of feminine and neuter genders differs considerably—in the corpus, they account for 36.6 and 2.6 percent, respectively (Rimkutė Straipsniai / Articles 135 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ TABLE 13. Generalized distribution of cases (the only or most frequent case forming a particular syntactic relation is marked in gray) Syntactic relation / case Nom., no. Nom., percent Gen., no. Gen., percent Dat., no. Dat., percent Acc., no. Acc., percent Inst., no. Inst., percent Loc., no. Loc., percent Total, no. 1 For mutual relation: subject 64 100,0 0 0,0 0 0,0 0 0,0 0 0,0 0 0,0 64 2 Mutual relation: object 34 100,0 0 0,0 0 0,0 0 0,0 0 0,0 0 0,0 34 3 Subordinate: subject 0 0,0 0 0,0 32 100,0 0 0,0 0 0,0 0 0,0 32 4 Subordinate: object 53 6,6 232 28,8 60 7,5 429 53,3 28 3,5 3 0,4 805 5 Subordinate: circumstance 9 11,4 12 15,2 6 7,6 9 11,4 3 3,8 40 50,6 79 6 Subordinate: attribute 6982 18,7 18802 50,4 1986 5,3 5322 14,3 1688 4,5 2502 6,7 37282 7 Coordinate 28 37,8 28 37,8 4 5,4 10 13,5 0 0,0 4 5,4 74 Total 7170 19074 2088 5770 1719 2549 38370 136 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos 2006: 31–32). These results were caused by the fact that forms of pronouns and numerals that do not have the category of gender (e.g., kas, mano, aš, dešimt, penkiolika, dvejetas¹¹) were also included in the corpus for the gender category—these forms make up 3.6%. Nevertheless, it can be argued that the distribution of gender in two-word MWEs corresponds to the patterns of use of these forms in written language. 2.2.2.2. Number category When counting singular and plural forms, the number forms of finite verb forms were excluded, since they were discussed together with the category of person (see Table 5), while participle and half-participle forms were included. Two thirds of the two-word inflectional MWEs in LKDFŽ are singular forms; see Table 15. TABLE 15. Distribution of number forms Number No. Percent singular 25650 66,95 plural 12665 33,05 Total 38315 100,00 The figures presented in Table 15 are almost identical to the written-language corpus data: in this source, singular forms account for 67.27%, and plural forms for 32.73% (Rimkutė 2006: 32). Regarding the plural, it is important to emphasize that some MWEs are used exclusively in the plural form, cf. Jungtinės Tautos, mirties aplinkybės, meteorologinės sąlygos (Boizou et al. 2015a: 46). The corpus also contained dual forms, and forms of number that are atypical for some nominals (e.g., dešimt, dvylika, keliolika—these were classified into a separate group in the corpus), so the data for the two-word MWEs analyzed and the corpus data cannot be compared with complete accuracy. Nevertheless, counting only the singular and plural forms of nouns, adjectives, pronouns, numerals, participles, and half-participles found in the corpus shows that the number-category data for the two-word MWEs and the corpus are almost entirely in agreement. 2.2.2.3. Category of degree Some MWEs are used in only one degree form (e.g., Aukščiausiasis Teismas), while others can be used in several degree forms (e.g., įvairūs būdai, įvairiausi būdai) (Boizou et al. 2015a: 47). The degree forms of the two-word MWEs analyzed ¹¹ There are various opinions concerning the genders of these and other forms (as well as other grammatical categories) (cf. Paulauskienė 1994; Judžentis 2012). The data provided by the Lithuanian morphological annotator were used (see http://tekstynas.vdu.lt/). Straipsniai / Articles 137 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ forms conform to the usage patterns of this category in written language—positive degree forms predominate; see Table 16. TABLE 16. Distribution of degree forms Degree | No. | Percent; positive | 6300 | 95,93; superlative | 135 | 2,06; comparative | 132 | 2,01; Total | 6567 | 100,00 Nevertheless, the percentage distribution of degree forms differs considerably between the analyzed LKDFŽ expressions and the forms used in the corpus: the positive degree accounts for 87.5 percent in the corpus; among two-word MWEs, comparative and superlative forms account for almost equal proportions, whereas the corpus shows a greater difference—the comparative degree accounts for 8 percent, and the superlative for 4.5 percent (Rimkutė 2006: 36). 2.2.2.4. Category of definiteness As might be expected, indefinite forms are the most numerous among the expressions analyzed; see Table 17. TABLE 17. Distribution of definite and indefinite forms Forms | No. | Percent; indefinite | 6714 | 81,9; definite | 1480 | 18,1; Total | 8194 | 100,0 Palyginus su rašytinės kalbos tekstyno duomenimis (Rimkutė 2006: 33), matyti, kad vartosenos dėsningumai tokie patys, bet procentinė išraiška gana skirtinga: tekstyne neįvardžiuotinės formos sudaro 93,6 proc., o įvardžiuotinės – 6,4 proc. Šiuos skirtumus gali lemti tai, kad, kaip minėta poskyrio pradžioje, giminės, skaičiaus, laipsnio, apibrėžtumo kategorijas iš esmės lemia prijungiamojo ryšio požymio raiškos junginių gramatinės kategorijos, taip pat tai, kad tarp dvižodžių PJ labai dažnas morfologinės struktūros modelis būdvardis + daiktavardis ir retesnis modelis skaitvardis + daiktavardis (Boizou ir kt. 2015b: 121). Tiek būdvardžiai, tiek kelintiniai skaitvardžiai gali būti įvardžiuotiniai ir tai galima laikyti priežastimi, dėl ko tarp analizuotų dvižodžių PJ gana daug 138 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos of pronominal forms. The higher frequency of pronominal forms is also due to the fact that many of the two-word MWEs included in the LKDFŽ are terms, designations of positions, and names of institutions, which are usually pronominal, e.g.: gyvenamasis namas, lengvoji atletika, pirmojo vicemero, Raudonasis Kryžius. Summarizing the grammatical categories discussed in Section 2.2.2, the two-word MWEs analyzed in this article display a distribution of grammatical forms similar to that found in the written-language corpus. Nevertheless, the percentages for some categories differed because a greater variety of forms was found in the corpus (for example, dual forms and forms lacking number or gender categories, etc.), whereas the variety of such forms among the two-word MWEs analyzed was much more limited, or such forms were not found at all. In addition, differences in some forms are determined by the characteristics of the MWEs (terms, institution names, etc.). 3. CONCLUSIONS 1. Nouns predominate (74.55%) among the two-word MWEs included in the LKDFŽ and analyzed in this article; adjectives are the second most frequent (16.11%), and verbs the third (5.2%). The distribution of parts of speech does not correspond to patterns in written and spoken language. This difference in usage is due to the dictionary’s compilation principle (every phrase must contain one noun) and to the fact that the two-word fixed expressions analyzed cannot exhibit a wide variety of parts of speech. 2. The analyzed phrases contain more finite verb forms; among non-finite forms, infinitives predominate (70%). More than half of all finite forms are in the past simple tense, while the present tense accounts for about 42%. The indicative mood constitutes the largest share of the mood category (96.6%); the third person predominates (94.7%). The use of grammatical verb forms in the analyzed phrases does not differ greatly from the written-language corpus data, although the dominant member of the respective category accounts for a somewhat larger share than in the corpus (this also applies to the grammatical categories of nominals). 3. The usage of more and less frequent cases confirms the patterns of case frequency in spoken and written Lithuanian described in previous studies. The cases most frequently used in the analyzed MWEs are the genitive (49.64%), nominative (18.66%), and accusative (15.02%). The dominance of the genitive is due to the fact that the two-word phrases in the LKDFŽ are most often formed with non-agreeing modifiers expressed in the genitive. In one-form two-word MWEs, the use of the instrumental and vocative is markedly more frequent (than Straipsniai / Articles 139 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ in multi-form MWEs). Thus, less frequently used cases are more typical of more fossilized phrases. 3.1. Only the nominative is used in the two-word MWEs of the LKDFŽ to express mutual relations in both subject and object phrases. Dative alone forms subject phrases linked by subordination. 3.2. In object, circumstantial, and attributive phrases linked by subordination, one dominant case stands out, accounting for more than half of all usage: respectively, the accusative, locative, and genitive. 3.3. Coordinate phrases reflect patterns of case usage in the written-language corpus more closely than subordinate or mutually related phrases do: the most frequent cases are the nominative, genitive, and accusative. Nevertheless, the number of such MWEs in the study material is too small to allow more reliable conclusions. 4. The distribution of other grammatical categories of nominals generally coincides with the corpus data: the largest shares are masculine forms (54.58%), singular forms (66.95%), positive-degree forms (95.93%), and non-pronominal forms (81.9%). Nevertheless, some categories, such as pronominal forms, are used more frequently because many of the analyzed MWEs are terms. LITERATURE AND SOURCES Bagdonaitė Justina 2010: Lietuvių kalbos samplaių sintaksinės funkcijos. Bakalauro darbas. Kaunas: Vytauto Didžiojo universitetas. Bielinskienė Agnė 2010: Sujungiamojo ryšio semantika, pragmatika ir vartosenos ypatumai. Daktaro disertacija. Kaunas: Vytauto Didžiojo universitetas. Boizou Loïc, Kovalevskaitė Jolanta, Rimkutė Erika 2015a: Automatic Lemmatisation of Lithuanian MWEs. – Proceedings of 20ₜₕ Nordic Conference of Computational Linguistics NODALIDA 2015. NEALT Proceedings 23, 41–49. Boizou Loïc, Kovalevskaitė Jolanta, Rimkutė Erika 2015b: Lietuvių kalbos dvižodžių junginių morfologinių ir sintaksinių ypatybių sąsajos. – Darbai ir dienos 64, 115–131. Grammar of Contemporary Lithuanian, ed. V. Ambrazas. Vilnius: Mokslo ir enciklopedijų leidykla, 1996. 140 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos Dabašinskienė Ineta 2009: Šnekamosios lietuvių kalbos morfologinės ypatybės. – Acta Linguistica Lithuanica 60, 1–15. Jarašiūnaitė Giedrė 2004: Morfologinių samplaikų semantika ir gramatika. Magistro darbas. Kaunas: VDU. Judžentis Artūras 2012: Lietuvių kalbos gramatinės kategorijos. Vilnius: Vilniaus universiteto leidykla. Kovalevskaitė Jolanta, Rimkutė Erika 2010: Sudėtinės ir suaug­tinės lietuvių kalbos morfologinės samplaikos. – Kalbų studijos 16, 79–88. Labutis Vitas 1998: Lietuvių kalbos sintaksė. Vilnius: Vilniaus universiteto leidykla. Marcinkevičienė Rūta, Grigonytė Gintarė 2005: Lexicogrammatical Patterns of Lithuanian Phrases. – Proceedings of 2nd Baltic Conference on Human Language Technologies, 299–305. Marcinkevičienė Rūta 2010: Lietuvių kalbos kolokacijos. Kaunas: Vytauto Didžiojo universitetas. Paulauskienė Aldona 1989: Gramatinės lietuvių kalbos vardažodžių kategorijos. Vilnius: Mokslas. Paulauskienė Aldona 1994: Lietuvių kalbos morfologija. Paskaitos lituanistams. Vilnius: Mokslo ir enciklopedijų leidykla. Rimkutė Erika 2006: Morfologinio daugiareikšmiškumo ribojimas kompiuteriniame tekstyne. Daktaro disertacija. Kaunas: Vytauto Didžiojo universitetas. Rimkutė Erika, Bielinskienė Agnė, Kovalevskaitė Jolanta (sud.) 2012: Lietuvių kalbos daiktavardinių frazių žodynas. Kaunas: Vytauto Didžiojo universiteto leidykla. Prieiga internete: http://donelaitis.vdu.lt/lkk/pdf/ daikt_fr.pdf. Rimkutė Erika, Kovalevskaitė Jolanta 2015: Lietuvių kalbos daiktavardinių frazių žodyno vienaf or miai pastovieji junginiai. – Darbai ir dienos 64, 155–169. Wray Alison 2002: Formulaic Language and the Lexicon. Cambridge: Cambridge University Press. Straipsniai / Articles 141 LOÏC BOIZOU, JOLANTA KOVALEVSKAITĖ, ERIKA RIMKUTĖ Grammatical Categories of Lithuanian Two-Word Multiword Expressions SUMMARY The focus of this study is on two-word MWEs used in several grammatical forms in the Dictionary of Lithuanian Nominal Phrases. We analyse the distribution of grammatical categories in the two-word MWEs, which are mainly understood as typical collocations. The study aims to provide a detailed documentation of the usage of grammatical categories, with a specific emphasis on the case category, in order to show the correlations between each case and a particular syntactic function. The usage data of grammatical categories are compared with the data in written and spoken Lithuanian corpora. The analysis has revealed that in case of the part of speech, the research data are dominated by nouns (74.55%), which are followed by adjectives (16.11%); verbs take the third position (5.2%) by their frequency. The data in written and spoken language corpora reveal a different situation, which could be determined by the fact that we analyse the MWEs consisting of at least one noun and only two-word phrases; thus, we cannot expect such MWEs to demonstrate a high variability in the part-of-speech categories. In the research data the inflective verb forms are used more frequently than non-inflective. The analysis of inflective and non-inflective categories led to the following results: the most frequent category is infinitive (70%); more than half of all inflective forms are those of past tense and 42% – present tense forms. The mood category is expressed by indicative mood (96.6%), and the 3rd person (94.7%) dominates. It can be stated that the grammatical categories of the analysed MWEs are rather similar to the written language usage, although the most frequent member of a respective category constitutes a larger part in the dictionary than the corpus. The distribution of more frequently and less frequently used cases confirms similar tendencies in written and spoken language documented in previous studies. The most frequent cases are as follows: genitive (49.64%), nominative (18.66%), and accusative (15.02%). The research data provide an explanation to the exceptionally frequent use of genitive: more than half of MWEs in the dictionary are attributive phrases with a non-agreeing attribute expressed by a noun in the genitive case. In MWEs, which are included in the dictionary in one form only, the cases such as instrumental and vocative show a higher frequency than it is usual for these cases. This entails that the occurrence of infrequent cases is higher in fixed expressions. Only the nominative case is used in both the interdependent subject phrases and the interdependent object cases. The subject phrases with the subordination relation include nouns in the dative case only. The dominant case is detected in the subordination relation: genitive for attributive phrases, accusative for object phrases, locative for adverbial phrases. In contrast to the phrases with subordination, phrases with coordination express 142 Acta Linguistica Lithuanica LXXV Lietuvių kalbos dvižodžių pastoviųjų junginių gramatinės kategorijos case usage in a similar order as it is in the corpus, not in the dictionary: nominative is used most frequently; it is followed by genitive, and accusative takes the third position by frequency. The distribution of other nominal categories coincides with the usage observed in the corpus data: the biggest part is made up of the categories of the masculine gender (54.58%), singular number (66.95%), positive degree (95.93%), and definite form (81.9%). Received 12 July 2016. LOÏC BOIZOU Vytauto Didžiojo universitetas Centre for Computational Linguistics Vlადო Putvinskio g. 23-216, LT-44243 Kaunas [email protected] JOLANTA KOVALEVSKAITĖ Vytauto Didžiojo universitetas Centre for Computational Linguistics Vlადო Putvinskio g. 23-216, LT-44243 Kaunas [email protected] ERIKA RIMKUTĖ Vytauto Didžiojo universitetas Center for Computational Linguistics 23-216 V. Putvinskio St., LT-44243 Kaunas [email protected] Straipsniai / Articles 143