Optional valency complementations in Czech light verb constructions
Abstract
7
Full text
Optional valency complementations in Czech light verb constructions1 Václava Kettnerová (Charles University, Prague) ABSTRACT This paper addresses Czech light verb constructions, partly revising principles of their syntactic structure formation formulated within the Functional Generative Description. It argues that obligatoriness of valency complementations should be reflected in these principles. Namely, the role of optional valency complementations of light verbs played in this process has been analyzed. This analysis has shown that in the cases where light verbs do not provide asufficient number of valency complementations for the surface expression of semantic participants of predicative nouns, semantic participants of nouns make use of optional verbal complementations; namely ORIGin, LOCative and BENefactor have been attested in the VALLEX lexicon. In such cases, semantic participants can be expressed on the surface, either as optional verbal complementation or as nominal complementation. The distribution of verbal and nominal complementations have been observed in 1,600 light verb constructions extracted from the Czech National Corpus, with the result that the surface expression of these participants through the optional verbal complementations is strongly preferred (88% of verbal complementations and 12% of nominal ones). The semantic analysis has indicated that the optional verbal complementations are overrepresented as they cover broader semantic contexts than the corresponding nominal ones. KEYWORDS coreference, Czech, light verb constructions, optional verbal complementations, surface structure DOI https://doi.org/10.14712/18059635.2021.1.1 1 INTRODUCTION Light verb constructions (henceforth LVCs) represent an advanced syntactic phenomenon the description of which is achallenge for any syntactic theory. In this paper, Ifollow the analysis of Czech LVCs within the Functional Generative Description (henceforth FGD) elaborated by Kettnerová et al. (2018). Ifurther deepen this syntactic analysis, focusing on the role of optional complementations of light verbs in the formation of the syntactic structure of Czech LVCs. Czech LVCs are similar to language-specific constructions recognized in other languages, e.g., in French (Gross 1981; Gross 1996), Italian (Cantarini 2004; Jezek 2011), Spanish (Alba-Salas 2002), Russian (Mel’čuk 1996; Apresjan 2009), Polish (Vetulani 2000), German (Polenz 1963), English (Cattell 1984; Brinton and Akimoto 1999; 1 The research reported in this paper has been supported by the grant LINDAT/CLARIAH-CZ, Digital Research Infrastructure for Language Technologies, Arts and Humanities (LM2018101); this work has been using language resources distributed by this grant.
8 LINGUISTICA PRAGENSIA 1/2021 Baron and Herslund 1998), Japanese (Grimshaw and Mester 1988; Miyamoto 2000), Korean (Choi and Wechsler 2002), Urdu (Butt 1997; 2010), Hindi (Mohanan 1994), and Persian (Karimi-Doostan 2005). LVCs are alternatively referred to as support verb constructions, constructions with function verbs, stretched verb constructions (Allerton 2002), or periphrastic verbal constructions (Wierzbicka 1982). The main similarity between these constructions is the fact that apredicate is composed of two elements: asemantically underspecified verb and another predicative element (typically anoun), both forming together asingle predication. This predication exhibits adiscrepancy in syntax and semantics: from the pair of apredicative noun and alight verb, it is the noun— despite being syntactically dependent on the verb— that represents its semantic core.2 The semantic properties of light verbs are the subject of alively debate in current linguistics. Their semantic interpretation typically varies from (i) being semantically empty, see e.g., (Grimshaw and Mester 1988), (Cattell 1984), (Mel’čuk 1996), and (Gross 1981), to (ii) having some lexical properties (Butt 2010; Butt and Geuder 2013; Sanromán Vilas 2011; Apresjan 2009). In Czech, light verbs span the scale from verbs with general meaning (e.g., mít ‘to have’, dělat ‘to do’, dát ‘to give’, dostat ‘to get’) to verbs with more distinctive meaning (e.g., poskytnout ‘to provide’, uložit ‘to impose’, přijmout ‘to receive’, upřít ‘to fix’). From the lexical semantic perspective, these verbs primarily differ in their collocational restrictions (compare, e.g., the light verbs dát ‘to give’, udělit ‘to grant’, uložit ‘to impose’, and napařit ‘to slap’ constituting LVCs with, e.g., the noun pokuta ‘fine’). The most thorough analysis of Czech LVCs was carried out by Macháčková (1979; 1994) and more recently by Radimský (2010): the former work was inspired by the German concept of function verbs, the latter applies the basic postulates of the French Lexicon-Grammar theory (Gross 1981) to Czech data. Aformal account of the syntactic structure formation of Czech LVCs has been proposed within FGD by Kettnerová (2017) and Kettnerová et al. (2018). In FGD, three layers of language description are supposed to be necessary for afull description of LVCs: alayer of cognitive content, alayer of deep syntactic structure, and alayer of surface syntactic structure (Sgall et al. 1986; Panevová et al. 2014). For each of them, different units are posited: semantic participants,3 valency complementations, and clause elements, respectively. On the 2 This interpretation is in line with Cruse’s lexical semantic analysis (1986: 103ff.), in which the semantic head of a LVC would be represented by the noun— not by the light verb— as it is the noun that semantically interacts with other parts of the sentence. For example, the semantic clash in the following sentence with the LVC dát polibek ‘to give akiss’ is produced by the interaction between the nouns polibek ‘kiss’ and policie ‘police’: Policie dala Janě polibek ‘The police gave Jane akiss’. Thus it can be rectified by replacing either the noun polibek ‘kiss’ by, e.g., the noun pokuta ‘fine’ (Policie dala Janě pokutu ‘The police gave Jane afine’) or by substituting the noun policie ‘police’ by, e.g., the noun Petr ‘Peter’ (Petr dal Janě polibek ‘Peter gave Jane akiss’). On the contrary, substituting the verb dát ‘to give’ by the verb udělit ‘to grant’ does not remove the semantic discrepancy (Policie udělila Janě polibek ‘The police granted Jane akiss’). 3 Description of the cognitive content according to FGD, which follows the distinction between meaning and content, goes beyond the scope of linguistics. However, the correspondence between semantic participants, belonging to cognitive content, and units of
VáCLAVA KETTNEROVá 9 basis of the analysis introduced in Kettnerová (2017) and Kettnerová et al. (2018), aLVC can be defined as anoun-verb pair such that: (i) the noun occupies avalency complementation of the verb, (ii) none of the remaining valency complementations in the valency frame of the verb corresponds to asemantic participant of the verb,4 (iii) the semantically underspecified valency complementations of the verb are semantically saturated by semantic participants of the noun, (iv) those semantic participants of the noun that semantically saturate valency complementations of the light verb are structured on the surface as the respective valency complementations of the verb, i.e., as verbal dependents. It follows from the definition that light verbs allow predicative nouns to employ their semantic participants in verbal structures. The number of semantic participants of the predicative noun employed as verbal dependents depends on the number of semantically unsaturated valency complementions of the light verb. As the annotation of alarge number of LVCs in the VALLEX lexicon5 revealed, at least one nominal participant and at most two nominal participants are syntactically structured as verbal modifications in LVCs, one of them is typically that nominal participant that is mapped onto ACTor of the noun. Here Ishow that when avalency frame of alight verb does not provide asemantically unsaturated valency complementation for the semantic participant of the noun mapped onto its ACTor (sporadically PATient as well), this semantic participant still displays atendency to be structured on the surface as adependent of the verb, making use of an optional complementation of the light verb, which stands outside the valency frame of the verb.6 In these cases, the surface expression of semantic participants, however, cannot fully comply with characteristic (iv) as these participants can be expressed on the surface as dependents of either the verb or the noun. Ithus refine the formal rules allowing for generation of well-formed LVCs, taking into account the role of optional valency complementations. Ilimit my study to the most frequent Czech LVCs, identified on the basis of the above given definition of LVCs, that are composed of alight verb and apredicative noun expressed in active LVCs as the non-prepositional direct object of the light the other two syntactic layers is crucial for an analysis of LVCs. Here Idetermine the set of semantic participants primarily on the basis of the semantically oriented lexicons included in the CzEngClass project (Urešová at al. 2019). 4 For the only exception see Section 2.2. 5 The VALLEX lexicon, version 4.0, contains 3,000 collocations of predicative nouns and light verbs (counted as combinations of alemma of alight verb and alemma of apredicative noun), obtained from the Czech National Corpus, SYN2010, using the Sketch Engine (Kilgarriff et al. 2014). These collocations correspond to almost 1,500 LVCs (counted as individual combinations of alexical unit of alight verb and alexical unit of apredicative noun). The lexicon is available at http://ufal.mff.cuni.cz/vallex/4.0. 6 The role of obligatory valency complementation of light verbs in LVCs have been thoroughly described by Kettnerová et al. (2018).
10 LINGUISTICA PRAGENSIA 1/2021 verb. From these cases, Ifocus on such LVCs in which asemantic participant of the predicative noun semantically specifies an optional valency complementation of the light verb; this viewpoint narrows down the range of discussed LVCs and it should be stressed that these cases are not numerous. 2 REPRESENTATION OF LIGHT VERB CONSTRUCTIONS IN FGD In this section, Ibriefly outline the main tenets of the valency theory of FGD relevant for the description of LVCs (Section 2.1) and the present model of their representation within this framework (Section 2.2). The model builds on the assumption that LVCs are syntactically compositional constructions, i.e., their syntactic structures can be derived from the information on syntactic behavior of predicative nouns and light verbs.7 2.1 VALENCY THEORY OF FGD FGD is adependency-oriented framework with amultilayered design. The primary focus of FGD is the so-called tectogrammatical layer, i.e., alayer of linguistically structured meaning, roughly corresponding to the deep syntactic layer, see esp. Sgall et al. (1986). One of the core components of the tectogrammatical layer is valency, the ability of some words (primarily of verbs, but also of some nouns, adjectives and adverbs) to require anumber of dependent words. The valent words are usually referred to as predicates. The dependent words, filling valency positions of predicates, represent their valency complementations.8 In the valency theory of FGD, valency complementations are divided into actants (inner participants) and free modifications, see esp. Panevová (1994). With verbs, we distinguish five types of actants characterized by functors, i.e., labels representing the type of the dependency relation of an actant to its governing verb: ACTor, PATient, ADDRessee, ORIGin and EFFect. With nouns, the actant MATerial can occur in addition (Piťha 1984). Actants are determined primarily on asyntactic basis. By contrast, free modifications, e.g., LOC (locative), TWHEN (temporal when), CAUS (cause), and MEANS (means), are identified primarily according to their semantics (see esp. Panevová et al. 2014). Each complementation is either obligatory or optional, see esp. Panevová (1974; 1975). The valency frame of apredicate captures its valency structure by listing its obligatory and optional actants and obligatory free modifications. Optional free modifications are not part of the valency frame but they can be listed for whole groups of semantically similar predicates. 7 Cf. asimilar approach to LVCs in the Meaning-Text Theory (Mel’čuk 1982; 1996). Nevertheless, it is worth noting that the fact that verbs and nouns within LVCs have their own syntactic structure is questioned by some scholars, see e.g., (Thompson and Hopper 2001). 8 In some works, this term is reserved for constructions with the predicate taking an argument of the propositional character expressed as acomplement clause, see e.g., Dixon and Aikhenvald (2006) and Noonan (2007).
VáCLAVA KETTNEROVá 11 Each valency frame comprises aset of valency positions, each of which stands for one valency complementation, described by afunctor and in terms of obligatoriness.9 Moreover, possible morphemic form(s) are listed where necessary: with verbs they determine the surface realization of valency complementations, i.e., clause elements, in active, non-reflexive and non-reciprocal constructions.10 With nouns, they indicate the surface expression of valency complementations in non-reflexive and non-reciprocal nominal structures.11 The valency theory of FGD has been applied in several lexical resources, esp. in the Prague Dependency Treebank family (Hajič et al. 2018; Hajič et al. 2012) and valency lexicons (Lopatková et al. 2016; Urešová et al. 2014; Urešová et al. 2016). 2.2 LIGHT VERB CONSTRUCTIONS IN FGD LVCs have been studied within FGD esp. by Cinková (2009), Kolářová (2010), Kettnerová (2017) and Kettnerová et al. (2018). The theoretical results have then been applied in the annotation of alarge amount of linguistic data in the VALLEX lexicon, see esp. Kettnerová et al. (2018), and they are partly reflected in the Prague Dependency Treebank (Hajič et al. 2018) as well. Let me sketch the present model of the syntactic structure formation of LVCs as elaborated within FGD. 2.2.1 DEEP SYNTACTIC STRUCTURE It is possible to capture the deep syntactic structure formation of a LVC by rules relying on: (i) the valency frame of the predicative noun, (ii) the frame of the light verb combined with the noun, and (iii) coreference between valency complementations of the light verb and the predicative noun: (i) The valency frame of the predicative noun underlies both its usage in nominal structures and in LVCs. Valency complementations of the noun are in one-toone correspondence with semantic participants (the correspondence is the same regardless of whether the predicative noun is used in nominal structures or in LVCs). (ii) The valency frame of the light verb underlies its usage in LVCs. It is typically identical with the valency frame of its full verb counterpart, which can be usually identified for each light verb. The only difference in the valency frame 9 As obligatoriness of valency complementations of nouns is often questioned and as it is not crucial for my explanation, Ileave this information aside here. 10 Reflexive constructions refer here to those constructions that are characterized by coreference of the reflexive pronoun and avalency complementation of averb expressed in the subject position, typically the ACTor (e.g., Petr se vidí jako vůdce ‘Peter perceives himself as aleader’). Under the term reciprocal constructions, Iunderstand those constructions that involve two propositions (e.g., Petr aMarie se milují ‘Peter and Mary love each other’ involves Petr miluje Marii ‘Peter loves Mary’ and Marie miluje Petra ‘Mary loves Peter’), see esp. Kettnerová and Lopatková (2019). 11 Valency of nouns within FGD has been described esp. by Kolářová (2010; 2014), Kolářová et al. (2019) and Panevová et al. (2014).
12 LINGUISTICA PRAGENSIA 1/2021 of the light verb is that one of its valency complementations is reserved for apredicative noun (in FGD labelled with the CPHR functor, Compound PHRaseme). Valency complementations of light verbs— with respect to the semantic underspecification of these verbs— do not correspond to any semantic participants, cf. Alonso Ramos (2007) and Grimshaw and Mester (1988). Only with causative light verbs (e.g., budit ‘to arouse’, vyvolat ‘to cause’), one of their valency complementations corresponds to the semantic participant Instigator, instigating the event expressed by the predicative noun. (iii) Coreference between valency complementations of the light verb and the predicative noun is afundamental characteristic of LVCs. Via coreference with valency complementations of predicative nouns, semantically underspecified valency complementations of light verbs acquire their semantic capacity. Information (i), (ii) and (iii) is stored in the lexicon where light verbs and predicative nouns forming individual LVCs are linked together. In the grammar, rules operating on individual valency frames of predicative nouns and light verbs make use of the information on coreference, see esp. Kettnerová et al. (2018). These rules stipulate first that the deep syntactic structure of a LVC consists of all valency complementations of the light verb and all complementations of the predicative noun, occupying the CPHR complementation of the verb, and second, that respective pairs of nominal and verbal valency complementations are linked by coreference. For example, the deep structure of the LVC chovat nenávist ‘to hold hatred’ is described by the valency frame of the light verb chovat ‘to hold’ (1) and by the valency frame of the noun nenávist ‘hatred’ (2). The valency frame of the light verb chovat ‘to hold’ in (1) underlies its usage in other LVCs as well (e.g., chovat cit ‘have affection’, chovat nedůvěru ‘to have distrust’, chovat podezření ‘to have asuspicion’, chovat úctu ‘to have respect’). The valency frame of the noun nenávist ‘hatred’ (2) describes its usage in nominal structures and in other LVCs as well (e.g., pojmout nenávist ‘to start to feel hatred’, vzbudit nenávist ‘to arouse hatred’, vyvolat nenávist ‘to arouse hatred’). In the deep syntactic structure of the LVC chovat nenávist ‘to hold hatred’, valency complementations of both the light verb chovat ‘to hold’ (ACTor and Compound PHRaseme) and the predicative noun nenávist ‘hatred’ (ACTor and PATient semantically corresponding to Experiencer and Stimulus, respectively) are contained. The noun nenávist ‘hatred’ occupies the valency complementation CPHR of the light verb. The other complementation of the light verb— the ACTor— acquires semantic specificity via coreference with the ACTor of the predicative noun, see the scheme of the mapping of semantic participants in (3). The simplified deep dependency tree of example (4) is provided in Figure 1 (upper part a.) below.12 12 The numbers indicate cases, pos stands for possessive adjectives or pronouns, obl marks obligatoriness.
VáCLAVA KETTNEROVá 13 (1) chovat ‘to hold’: ACT1obl CPHR4obl (2) nenávist ‘hatred’: ACT2,pos PATk+3,proti+3,vůči+3 (3) chovat nenávist ‘to hold hatred’: Experiencern ⇒ ACTn ↔ACTv13 Stimulusn ⇒ PATn (4) Muž-Ø choval-Ø nenávist-Ø karmád-ě. man-NOM.SG.M held-SG.M hatred-ACC.SG.F to army-DAT.SG.F ‘The man held hatred for the army.’ 2.2.2 SURFACE SYNTACTIC STRUCTURE The surface structure of an active LVC is governed by principles operating on its deep syntactic structure. These principles say that from each pair of coreferring verbal and nominal valency complementations, only the verbal complementation is expressed on the surface as the morphemic forms determine for this complementation in the valency frame of the light verb; the nominal complementation from this coreferring pair is elided from the surface. Other valency complementations outside coreference, be it verbal or nominal, are realized in the surface structure by the morphemic forms prescribed in the respective valency frames.14 For example, according to the principles given above, the valency frame of the light verb (1’) and the valency frame of the predicative noun (2’) underlying the surface structure of the LVC chovat nenávist ‘to hold hatred’ are as follows: (1’) chovat ‘to hold’: ACT1obl CPHR4obl (2’) nenávist ‘hatred’: ACTØ PATk+3,proti+3,vůči+3 From the valency frame of the light verb (1’), both the ACTor and CPHR are realized on the surface as their morphemic forms determine: ACTor as the subject (muž ‘man’ in example (4)) and CPHR as the direct object (nenávist ‘hatred’ in (4)). From the valency frame of the predicative noun (2’), only the PATient is expressed on the surface as an adnominal attribute (karmádě ‘to army’ in (4)). The ACTor of the predicative noun, being coreferring with the ACTor of the light verb, see scheme (3), is deleted from the surface (see Ø in the lower index). See Figure 1 (lower part b.) displaying the simplified surface dependency tree of example (4). 13 The symbols n and vin the lower index with semantic participants and valency complementations distinguish whether they are nominal or verbal. The rightward double arrow shows the mapping of semantic participants onto valency complementations, the left right arrow marks coreference between nominal and verbal complementations. 14 These principles have been verified in the Prague Dependency Treebank 3.0 (Kettnerová and Bejček 2016). Let me stress that the surface structure governed by these principles can be subsequently subject to other types of systemic as well as non-systemic textual ellipsis; on ellipsis in FGD see esp. Panevová et al. (2014).
14 LINGUISTICA PRAGENSIA 1/2021 The principles formulated for the surface structure of LVCs ensure that each semantic participant is expressed on the surface only once.15 The principles of the surface expression of semantic participants within LVCs can be summarized as follows: (ia) Nominal semantic participants that are mapped within LVCs only onto nominal valency complementations are expressed on the surface as the respective nominal complementations. (ib) Nominal semantic participants that are mapped within LVCs onto both nominal and— via coreference— verbal complementations are realized as the respective verbal complementations. (ii) The verbal semantic participant Instigator contributed to LVCs by causative light verbs is expressed on the surface as the verbal complementation onto which it is mapped. 15 In this respect, LVCs do not differ from constructions with full verbs. Only in very rare cases, asemantic participant mapped onto the ACTor of apredicative noun and at the same time— via coreference— onto the ACTor of the light verb combined with the noun can be expressed twice (e.g., SváACTn podezření mají imístní obyvateléACTv ‘Local peopleACTv have theirACTn suspicions as well’). However, these cases are rather infrequent and their stylistic appropriateness might often be questionable. Figure 1.The simplified deep (a) and surface dependency tree (b) of sentence (4). The fine dashed arrow pointing from the ACTor of the noun to the ACTor of the verb in (a) represents the coreferential link. The lemma #Cor assigned to the nominal ACTor marks that this node is added to the deep dependency tree and has no surface counterpart. The dashed arrows pointing from (a) to (b) show correspondence between the deep and surface nodes. chovat ‘to have’ PRED muž ‘man’ ACT #Cor ACT nenávist ‘hatred’ CPHR armáda ‘army’ PAT k ‘for’ AuxP nenávist ‘hatred’ Obj b. nominal verbal ACTn-ORIGv 0 50 150 250 350 ACTn-LOCv ACTn-BENv PATn-LOCv Frequency a. choval ‘had’ Pred muž ‘man’ Sb armádě ‘for army’ Atr
VáCLAVA KETTNEROVá 15 For example, the LVC chovat nenávist ‘to hold hatred’ is characterized by two semantic participants— Experience and Stimulus. Both are licensed by the noun nenávist ‘hatred’, see their mappings in (3). The surface expression of Experiencer (muž ‘man’) follows from principle (ib) and Stimulus (armáda ‘army’) is realized in accordance with principle (ia).16 3 OPTIONAL VERBAL COMPLEMENTATIONS IN ACTIVE LVCS Here Ishow that principle (ib), introduced in Section 2.2.2, requires further modification, taking the obligatoriness of valency complementations of light verbs into account (Section 3.1). This modification is supported by acorpus analysis (Section 3.2). Finally, Isuggest an interpretation of differences in the distribution of verbal and nominal complementations in the surface structure of LVCs (Section 3.3). 3.1 SURFACE REALIZATION OF NOMINAL SEMANTIC PARTICIPANTS We can observe that semantic participants of predicative nouns in LVCs can be mapped— via coreference— onto optional complementations of light verbs as well. In these cases, the coreferring pair onto which the nominal semantic participant is mapped consists of anominal valency complementation and an optional verbal complementation (be it actant or free modification).17 For example, the semantic participant Speaker in the LVC dostat pokyn ‘to receive an instruction’ is mapped onto the nominal ACTor and— via coreference— onto the verbal actant ORIGin, which is optional, see the mapping in this LVC in (7) and the valency frames (5) and (6). There are two options for expressing such anominal participant on the surface: besides the possibility to be expressed as the respective optional verbal complementation, it can also be expressed as the respective nominal complementation. For example, in the LVC dostat pokyn ‘to receive an instruction’, the participant Speaker (tajemník ‘secretary’) can be expressed either as the verbal ORIGin (od tajemníka ‘from the secretary’ in (8a)) or as the nominal ACTor (tajemníka ‘of the secretary’ (8b)). Let me stress, however, that only the first variant in (8a), not the latter one in (8b), complies with principle (ib) given in Section 2.2.2. (5) pokyn ‘instruction’: ACT2,od+2,pos ADDR2,3,pos PATk+3,inf,aby,ať,zda,že,cont (6) dostat ‘to receive’: ACT1obl CPHR4obl ORIGod+2opt 16 Let me remark that the order of nodes in the surface structure of LVCs is often non-projective: e.g., the expression sostatními ‘with other people’ depends on the noun soucit ‘sympathy’ as its PATient in both sentences Měl soucit sostatními ‘He had sympathy with other people’ and Soucit, jaký měl sostatními, byl pozoruhodný ‘Sympathy which he had with other people was remarkable’; in the latter sentence, PATient is, however, placed non-projectively; on projectivity see (Uhlířová and Kučerová 2017). 17 Let me remind the reader that only optional actants are part of valency frames. In contrast, optional free modifications stand outside valency frames. See Section 2.1.
22 LINGUISTICA PRAGENSIA 1/2021 Let me return back to the LVCs with the coreferring pair ACTn-BENv. The similar semantic difference between the variant with the nominal ACTor in the form of the possessive pronouns/adjectives or the genitive case and the variant with the verbal BENefactor in the form of the possessive dative can be observed here as well. The nominal ACTor implies only the already existing possession of an entity denoted by predicative nouns. In contrast, the verbal BENefactor can entail not only the already existing possession but its future possession as well. From this it follows that in the contexts concerning the already existing possession, either the nominal ACTor or the verbal BENefactor can be used, compare examples (11a) and (11b) in Section 3.2. In contrast, the verbal BENefactor is restricted to the contexts in which the future possession is at play, compare the semantic difference between the examples (15a) and (15b) found in the data. (15a) with the possessive dative, the verbal BENefactor, can be interpreted in two ways: Russian companies and banks have already been entering the capital markets or they have not entered them yet, while (15b) with the nominal ACTor can rather have only the former interpretation, that Russian companies and banks have already been trading on the capital markets. (15) a. Na nových opatřeních, která by například ruským firmámBENv či bankámBENv uzavřela přístup kevropským kapitálovým trhům … ‘On new measures which would close access for Russian companiesBENv or banksBENv to European capital markets’ b. Na nových opatřeních, která by například uzavřela přístup ruských firemACTn či bankACTn kevropským kapitálovým trhům … (SYNv7) ‘On new measures which would close Russian companies’ACTn or banks’ACTn access to European capital markets’ Similar differences in the variants with the other coreferring pairs involving the nominal ACTor can be found as well. The expression of the nominal ACTor on the surface is associated with the already existing possession of an entity denoted by predicative nouns (in (16b), for example, it is the sympathy that people have that is aroused by swindlers) while the expression of LOCative can additionally entail the future possession of this entity (in (16a), it can be the sympathy that is evoked in people by swindlers either as an already experienced feeling or as anew feeling). (16) a. Podvodníci budí ulidíLOCv soucit. ‘Swindlers arouse sympathy in peopleLOCv.’ b. Podvodníci budí soucit lidíACTn. (SYNv7) ‘Swindlers arouse people’sACTn sympathy.’ The semantic difference between the already existing possession (expressed by either the nominal ACTor or the corresponding verbal complementation) and the future possession (expressed exclusively by the respective verbal complementation) would account for the complete lack of the nominal ACTor expressed on the surface with the LVCs vytvořitpf, vytvářetimpf příležitost ‘to create an opportunity’ (see Table 2,
VáCLAVA KETTNEROVá 23 Section 3.2) in which only the future possession is relevant, compare examples (17a) and (17b).22 (17) a. Město občanůmBENv vytváří pracovní příležitosti. ‘The city creates job opportunities for citizensBENv.’ b. ?Město vytváří pracovní příležitosti občanůACTn. ‘The city creates citizens’ACTn job opportunities.’ The semantic account of the surface variants with the coreferring pair PATn-LOCv is rather different. The use of the nominal PAT semantically restricts the entity denoted by predicative nouns to their particular kinds, e.g., pleasure from food in (18b). In contrast, the verbal LOCative can be used even in the contexts when there is no such kind of the entity available, compare (18a) and (19a); in (19b) the use of the nominal PATient is semantically blocked as relief from sweeping as aparticular kind of relief does not exist. (18) a. Chcete ipři přísné dietě nacházet potěšení vjídleLOCv? ‘Do you want to find pleasure in foodLOCv even on astrict diet?’ b. Chcete ipři přísné dietě nacházet potěšení zjídlaPATn? (SYNv7) ‘Do you want to find pleasure from foodPATn even on astrict diet?’ (19) a. Nacházím útěchu vzametáníLOCv. (SYNv7) ‘Ifind relief in sweepingLOCv.’ b. ?Nacházím útěchu ze zametáníPATn. ‘?Ifind relief from sweepingPATn.’ 4 CONCLUSION Ihave shown here that even in the cases when alight verb does not have asufficient number of valency positions in its valency frame for semantic participants of the predicative noun combined with the verb, the nominal participants in LVCs have astrong tendency to be expressed on the surface as the dependent on the verb, making use of optional complementations of the light verb as well. This tendency is attested in the corpus data by those semantic participants of predicative nouns that 22 ACTor is completely missing with the LVC získatpf, získávatimpf půjčku ‘to get aloan’ as well. However, in this case, amore plausible explanation for its lack is that the form of possessive pronouns/adjectives or the genitive case with the noun půjčka ‘loan’ ambiguously encodes either the ACTor or the ADDResse of the noun (e.g., in the nominal structure Petrovapos půjčka ‘Peter’s loan’, Petr ‘Peter’ can be either the one who gives the loan or the one who receives it). It can be hypothesized, however, that in the similar cases where ambiguity is at play, the possibility to express the nominal ACTor on the surface may not be completely excluded but may depend on the meaning of nouns and the contexts in which they are used.
24 LINGUISTICA PRAGENSIA 1/2021 primarily correspond to their nominal ACTor as the most prominent valency complementation, and partly to their PATient. Although the speaker’s choice between the nominal complementation and the verbal one for the surface realization of these nominal participants is semantically justified, the verbal complementations are overrepresented on the surface, probably covering broader semantic contexts than the corresponding nominal one. The fact that the analyzed semantic participants of predicative nouns are primarily mapped onto ACTor (sporadically onto PATient) draws our attention to the question whether there is some hierarchy between semantic participants that determines which of them is expressed on the surface as verbal dependents. Acrucial task in further research of the syntactic structure formation of LVCs thus lies in identifying which semantic participants of predicative nouns tend to be expressed on the surface as verbal dependents. REFERENCES Alba-Salas, J. (2002) Light Verb Constructions in Romance: ASyntactic Analysis. Ph.D. thesis. Cornell University. Allerton, D.J. (2002) Stretched Verb Constructions in English. London/New York: Routledge. Alonso Ramos, M. (2007) Towards the Synthesis of Support Verbs Constructions. In: Wanner, L. (ed) Selected Lexical and Grammatical Issues in the Meaning Text Theory, 97–137. Amsterdam/Philadelphia: John Benjamins. Apresjan, J. (2009) The Theory of Lexical Functions: An Update. In: Beck, D., K.Gerdes, J.Milićević and A.Polguère (eds) Proceedings of the fourth international conference on MeaningText Theory, 1–14. Montreal: OLST. Baron, I. and M.Herslund (1998) Support Verb Constructions as Predicate Formation. In: Olbertz, H., K.Hengeveld and J.Sánchez García (eds) The Structure of the Lexicon in Functional Grammar, 99–116. Amsterdam/ Philadelphia: John Benjamins. Brinton, L.J. and M.Akimoto (1999) Collocational and idiomatic aspects of composite predicates in the history of English. Amsterdam/ Philadelphia: John Benjamins. Butt, M. (1997) Complex Predicates in Urdu. In: Alsina, A., J.Bresnan and P.Sells (eds) Complex Predicates, 107–149. Stanford, Calif.: CSLI Publications. Butt, M. (2010) The Light Verb Jungle: Still Hacking Away. In: Amberber M., B.Baker and M.Harvey (eds) Complex Predicates: Crosslinguistic Perspectives on Event Structure, 48–78. New York: Cambridge University Press. Butt, M. and W.Geuder (2013) On the (Semi) lexical Status of Light Verbs. In: Cover N. and H. van Riemsdijk (eds) Semi-lexical Categories: The Function of Content Words and the Content of Function Words, 323–370. Berlin/New York: Walter de Gruyter. Cantarini, S. (2004) Costrutti con verbo supporto. Italiano etedesco aconfronto. Bologna: Patron. Cattell, R. (1984) Composite Predicates in English. Sydney/New York/London: Academic Press. Choi, I. and S.Wechsler (2002) Mixed Categories and Argument Transfer in the Korean Light Verb Construction. In: Eynde, F. van, L.Hellan and D.Beerman (eds) Proceedings of the 8th International HPSG Conference, 103–120. Stanford, Calif.: CSLI Publications Cinková, S. (2009) Words that Matter: Towards aSwedish-Czech Colligational Dictionary of Basic Verbs. Praha: ÚFAL. Cruse, D.A. (1986) Lexical Semantics. Cambridge: Cambridge University Press. Dixon, R.M.W. and A.Y.Aikhenvald (2006) Complementation. ACross-Linguistic Typology. New York: Oxford University Press.
VáCLAVA KETTNEROVá 25 Grimshaw, J. and A.Mester (1988) Light Verbs and θ-Marking. Linguistic Inquiry 19/2, 205–232. Gross, G. (1996) Prédicats nominaux et compatibilité aspectuelle. Langages 30/121, 54–72. Gross, M. (1981) Les bases empiriques de la notion de prédicat sémantique. Langages 15/63, 7–52. Hajič, J., E.Bejček, A.Bémová, E.Buráňová, E.Hajičová, J.Havelka, P.Homola, J.Kárník, V.Kettnerová, N.Klyueva, V.Kolářová, L.Kučová, M.Lopatková, M.Mikulová, J.Mírovský, A.Nedoluzhko, P.Pajas, J.Panevová, L.Poláková, M.Rysová, P.Sgall, J.Spoustová, P.Straňák, P.Synková, M.Ševčíková, J.Štěpánek, Z.Urešová, B.Vidová Hladká, D.Zeman, Š.Zikánová and Z.Žabokrtský (2018) Prague Dependency Treebank 3.5. Available at http://hdl.handle. net/11234/1-2621 [last accessed 22August 2020] Hajič J., E.Hajičová, J.Panevová, P.Sgall, O.Bojar, S.Cinková, E.Fučíková, M.Mikulová, P.Pajas, J.Popelka, J.Semecký, J.Šindlerová, J.Štěpánek, J.Toman, Z.Urešová and Z.Žabokrtský (2012) Announcing Prague Czech-English Dependency Treebank 2.0. In: Calzolari, N., K.Choukri, T.Declerck, M.U.Doğan, B.Maegaard, J.Mariani, A.Moreno, J.Odijk and S.Piperidis (eds) Proceedings of the 8th International Conference on Language Resources and Evaluation (LREC 2012), 3153–3160. İstanbul: European Language Resources Association. Ježek, E. (2011) Verbes supports et composition sémantique. Cahiers de Lexicologie 98/1, 29–43. Karimi-Doostan, G. (2005) Light verbs and structural case. Lingua 115/12, 1737–1756. Kettnerová, V. (2017) Syntaktická struktura komplexních predikátů včeštině. Slovo aslovesnost 78/1, 3–24. Kettnerová, V. and E.Bejček (2016) Distribution of Valency Complements in Czech Complex Predicates: Between Verb and Noun. In: Calzolari, N., K.Choukri, T.Declerck, S.Goggi, M.Grobelnik, B.Maegaard, J.Mariani, H.Mazo, A.Moreno, J.Odijk and S.Piperidis (eds) Proceedings of the 10th International Conference on Language Resources and Evaluation (LREC 2016), 515–521. Paris: European Language Resources Association (ELRA). Kettnerová, V. and M.Lopatková (2019) Reflexives in Czech from aDependency Perspective. In: Gerdes, K. and S.Kahane (eds) Proceedings of the Fifth International Conference on Dependency Linguistics (Depling, Syntaxfest 2019), 14–25. Paris: Association for Computational Linguistics. Kettnerová, V., M.Lopatková, E.Bejček and P.Barančíková (2018) Enriching VALLEX with Light Verbs: From Theory to Data and Back Again. The Prague Bulletin of Mathematical Linguistics 111, 29–56. Kilgarriff, A., V.Baisa, J.Bušta, M.Jakubíček, V.Kovář, J.Michelfeit, P.Rychlý and V.Suchomel (2014) The Sketch Engine: ten years on. Lexicography. Journal of ASIALEX 1/1, 7–36. Kolářová, V. (2010) Valence deverbativních substantiv včeštině (na materiálu substantiv sdativní valencí). Praha: Karolinum. Kolářová, V. (2014) Special valency behavior of Czech deverbal nouns. In: Spevak, O. (ed) Noun Valency, 19–59. Amsterdam/ Philadelphia: John Benjamins. Kolářová, V., A.Vernerová and J.Verner (2019) Non-systemic valency behavior of Czech deverbal nouns based on the NomVallex lexicon. Jazykovedný časopis 70/2, 424–433. Křen, M., V.Cvrček, T.Čapka, A.Čermáková, M.Hnátková, L.Chlumská, T.Jelínek, D.Kováříková, V.Petkevič, P.Procházka, H.Skoumalová, M.Škrabal, P.Truneček, P.Vondřička and A.Zasina (2017) Korpus SYN, verze 7 z29.11.2018. Praha: Ústav Českého národního korpusu FF UK. Available at http://www.korpus.cz [last accessed 22August 2020] Lopatková, M., V.Kettnerová, E.Bejček, A.Vernerová and Z.Žabokrtský (2016) Valenční slovník českých sloves VALLEX. Praha: Karolinum. Macháčková, E. (1979) Analytické konstrukce typu sloveso + abstraktní substantivum (Analytické
26 LINGUISTICA PRAGENSIA 1/2021 vyjadřování predikátů). Praha: Ústav pro jazyk český ČSAV. Macháčková, E. (1992) Je posesívní dativ volný, nebo vázaný? Slovo aslovesnost 53/3, 185–192. Macháčková, E. (1994) Constructions with verbs and abstract nouns in Czech (analytical predicates). In: Čmejrková, S. and F.Štícha (eds) The syntax of sentence and text, 365–374. Amsterdam/Philadelphia: John Benjamins. Mel’čuk, I.A. (1982) Lexical Functions in Lexicographic Description. In: McCaulay, M. et al. (eds) Proceedings of the Eighth Annual Meeting of the Berkeley Linguistics Society, 427–444. Mel’čuk, I.A. (1996) Lexical Functions: ATool for the Description of Lexical Relations in aLexicon. In: Wanner, L. (ed) Lexical Functions in Lexicography and Natural Language Processing, 37–102. Amsterdam/Philadelphia: John Benjamins. Miyamoto, T. (2000) The Light Verb Construction in Japanese. The Role of the Verbal Noun. Amsterdam/Philadelphia: John Benjamins. Mohanan, T. (1994) Argument Structure in Hindi. Stanford, Calif.: CSLI Publications. Noonan, M. (2007) Complementation. In: Shopen, T. (ed) Language Typology and Syntactic Description, 52–150. Cambridge: Cambridge University Press. Panevová, J. (1974–75) On Verbal Frames in Functional Generative Description I-II. The Prague Bulletin of Mathematical Linguistics 22 and 23, 3–40, 17–52. Panevová, J. (1994) Valency Frames and the Meaning of the Sentence. In: Luelsdorff, P.A. (ed) The Prague School of Structural and Functional Linguistics: AShort Introduction, 223–243. Amsterdam/Philadelphia: John Benjamins. Panevová, J., E.Hajičová, V.Kettnerová, M.Lopatková, M.Mikulová and M.Ševčíková (2014) Mluvnice současné češtiny 2. Syntax češtiny na základě anotovaného korpusu. Praha: Karolinum. Piťha, P. (1971) Existuje dativ posesívní? Slovo aslovesnost 32/4, 301–312. Piťha, P. (1984) Case frames of nouns. In: Sgall,P. (ed) Contributions to functional syntax, semantics, and language comprehension, 225–238. Amsterdam/Philadelphia: John Benjamins. Piťha, P. (1992) Posesivní vztah včeštině. Praha: AVED. Polenz, P. von (1963) Funktionsverben im heutigen Deutsch: Sprache in der rationalisierten Welt. Düsseldorf: Pädagogischer Verlag Schwann. Radimský, J. (2010) Verbo-nominální predikát skategoriálním slovesem. České Budějovice: Jihočeská univerzita. Razímová, M. (2004) Funkce adverbálního dativu vhloubkové apovrchové stavbě české věty. Diploma thesis, FF UK. Sanromán Vilas, B. (2011) The Unbearable Lightness of Light Verbs: Are They Semantically Empty Verbs? In: Boguslavsky, I. and L.Wanner (eds) Proceedings of the 5thInternational Conference on Meaning-Text Theory, 253–263. Barcelona. Sgall, P., E.Hajičová and J.Panevová (1986) The Meaning of the Sentence in its Semantic and Pragmatic Aspects. Dordrecht: Reidel. Thompson, S.A. and P.J.Hopper (2001) Transitivity, clause structure, and argument structure: Evidence from conversation. In: Bybee, J. and P.Hopper (eds) Frequency and the emergence of linguistic structure, 27–60. Amsterdam/Philadelphia: John Benjamins. Uhlířová, L. and I.Kučerová (2017) Projektivnost. In: Karlík, P., M.Nekula, and J.Pleskalová (eds) CzechEncy— Nový encyklopedický slovník češtiny. Available at https://www.czechency.org/ slovnik/PROJEKTIVNOST [last accessed 30October 2020] Urešová, Z., J.Štěpánek, J.Hajič, J.Panevová and M.Mikulová (2014) PDT-Vallex: Czech valency lexicon linked to treebanks. Available at http:// hdl.handle.net/11858/00-097C-0000-00234338-F [last accessed 22August 2020] Urešová, Z., E.Fučíková and J.Šindlerová (2016) CzEngVallex: abilingual CzechEnglish valency lexicon. The Prague Bulletin of Mathematical Linguistics 105, 17–50. Urešová, Z., E.Fučíková, E.Hajičová and J.Hajič (2019) Parallel Dependency Treebank
VáCLAVA KETTNEROVá 27 Annotated with Interlinked Verbal Synonym Classes and Roles. In: Candito, M., K.Evang, S.Oepen and D.Seddah (eds) Proceedings of the 18th International Workshop on Treebanks and Linguistic Theories (TLT, SyntaxFest 2019), 38–50. Paris: Association for Computational Linguistics. Vetulani, G. (2000) Rzeczowniki predykatywne języka polskiego. Wkierunku syntaktycznego słownika rzeczowników predykatywnych. Poznań: Wydawnictwo naukowe UAM. Wierzbicka, A. (1982) Why Can You Have aDrink When You Can’t *Have an Eat? Language 58/4, 753–799. Václava Kettnerová Matematicko-fyzikální fakulta Univerzity Karlovy Ústav formální aaplikované lingvistiky Malostranské nám. 2/25, 118 00 Praha 1 ORCID-ID: 0000-0001-9694-1304 kettnero[email protected].cuni.cz