Pastabos dėl sintaksinio dviprasmiškumo sąvokos ir kiti susiję dalykai

Acta Linguistica Lithuanica LV (2006), 1 – 53 Notes on the Concept of Syntactic Ambiguity and Other Related Matters Rolandas Mikulskas Institute of the Lithuanian Language The article contains a polemic with Axel Holvoet’s analysis of Lithuanian sentences of the type Onutės žalios akys. The author disagrees with Holvoet’s claim that sentences of this type are syntactically ambiguous between predicative possessive (‘belonging’) constructions and copular constructions whose subjects are discontinuous as a result of fronting. In the author’s opinion syntactic ambiguity would entail semantic ambiguity, whereas the constructions under discussion have a well-defined meaning (they give a description of a person based on a characteristic attribute), entitling us to describe them as an autonomous construction. The sentence type under discussion originated, in the author’s view, by reinterpretation of a copular construction with a subject constituent made discontinuous by syntactic movement motivated by topicalisation. The author also polemicises with Holvoet’s theory of thematic-rhematic structure, a variety of that advanced by Bogusławski, pointing out that the notion of marked pattern of thematic-rhematic structure is not clearly defined. 1.0. In the third volume of Studies in Lithuanian Grammar Studies in Grammatical Functions (Holvoet & Mikulskas 2005), Chapter 6, Types of Intransitive Sentences: Existential, Locative, and Possessive Sentences, subsection 6.6 (153–159), Axel Holvoet discusses a possessive construction with an unclear syntactic structure, exemplified by sentences such as: (1) Onutė’s green eyes. (= 31)1 (2) Peter’s difficult character. (= 32) 1 The parentheses contain the original numbering of illustrative examples quoted from the article. 1 ROLANDAS MIKULSKAS Drawing on the previously formulated parallelism between existential and locative sentences, on the one hand, and possessive sentences expressing possession and belonging, on the other, with respect to information structure and the nature of syntactic relations, the author analyzes the syntactic structure of the sentences in question and at the same time attempts to determine their place among syntactic constructions that are functionally related. Having tested the entire arsenal of syntactic parsing options, the author concludes that, because they permit two possible syntactic parses, sentences such as (1) and (2) should be considered syntactically ambiguous, and that the construction they exemplify could be described as intermediate between constructions of possession and belonging. The author believes that the type of such sentences could even be called hybrid (Holvoet 2005: 158). There are more than one or two reasons why this conclusion reached by the author should be regarded as unsatisfactory. I will try to present several below. 1.1. The formulation of the conclusion itself is problematic at first glance. Indeed, the article convincingly demonstrates that none of the syntactic parsing models currently available can be applied to the analysis of the sentences under discussion without qualification. But does this make the syntactic structure of these sentences ambiguous? After all, the author’s observation that “the construction under discussion has a special meaning” (ibid., 156) would seem to direct us toward seeking the most transparent possible syntactic interpretation of it. And syntactic ambiguity presupposes semantic ambiguity as well! Even if we agree with the author’s claim that “information structure is an aspect of a sentence’s semantic structure” (Holvoet 2003: 96), there can be no question of semantic ambiguity in the sentences under discussion in this respect, since the article analyzes only the “unmarked” variants of these sentences (i.e. those without contrastive stress, a segmental marker, or a fronted topic). 1.1.1. It seems more appropriate to speak of an ambiguous semantic structure in cases where there is a basis for assigning two different meanings to a sentence that is identical in every respect. The classic semantic basis for an ambiguous syntactic interpretation is the difference in the scope of quantifiers in a sentence. Such ambiguous sentences are widely known in the linguistic literature, for example: (3) Several children’s parents do not pay the kindergarten fee. (a) “Several parents do not pay the kindergarten fee” (for example, out of malice or because they cannot afford it), 2 Dėl sintaksinio dviprasmiškumo (b) “All parents who take more than one child to kindergarten do not pay a fee” (i.e., are exempt from it). (4) Parents of many children have little free time. (a) “Many parents have little free time” (i.e., work a lot, are busy), (b) “Parents with many children have little free time” (i.e., caring for many children takes it away). Lakoff (1970: 405) was the first to draw attention to the possibility of a dual interpretation of such sentences; they were later discussed by Jackendoff (1977: 60) and Hudson (1984: 79–81), and Langacker (2000: 82–83) revealed the cognitive causes of their semantic (and syntactic) ambiguity. The essence of the syntactic ambiguity of sentences such as (3) and (4) is that the quantifiers in their possessive phrases (several, many) are (a) included, in one interpretation, within the scope of other logical operators in the sentence (negation un-, quantifier few), whereas (b) in the other interpretation they are not included, that is, the possessive phrase remains integral with respect to the predicate negation or the second quantifier. Interpretation (a) is usually described as having “wide” scope for the predicate negation or the second quantifier (wide scope). According to Langacker, the basis for the dual interpretation of sentences of this type lies in the semantic specificity of the possessive phrases used in them, and more specifically, in the ambivalence of the kinship terms that make them up. As is well known, kinship terms are semantically relational in nature: one kinship term is defined through its relation to another. Thus, for example, the idealized cognitive model (ICM; Lakoff 1987: 68–76) of FATHER always implies a relation primarily to a child (or children) (son(s), daughter(s)) and, somewhat less strongly, to a wife (the children’s mother). In cognitive grammar, phrases such as father(s) of the child(ren), like other possessive constructions, are commonly analyzed using the notion of REFERENCE POINT (reference point) (Langacker 2000: 80–81, more broadly 171–193; Langacker 1991: 37–39, more broadly 167–180). Put simply, in cognitive analysis the possessor (child(ren)) is regarded as a reference point through which the possessed entity (father(s)) becomes accessible (identifiable) to the conceptualizer; relative to the reference point, it functions as a target (target). With respect to the syntactic relation between the components of a possessive construction understood in this way, the target generally functions as the head (head), and the reference point as its modifier (modifier). However 3 ROLANDAS MIKULSKAS when a possessive phrase consists of two kinship terms, there is reason to assume that the head of the phrase, for example, father(s), always has a “projected” place in its semantic structure for its essential “argument”—child(ren)—which may be “filled” or remain implicit. Thus, one can argue that the components of these phrases are linked by syntactic dependency in two opposite directions: primarily by a modification relation and, to a somewhat lesser extent (the likelihood of its being activated is lower), by a complement relation (in other terms, a government relation). The cognitive basis for this ambivalence is that both kinship terms—one to a greater extent, the other to a lesser extent—semantically elaborate (elaborate) each other’s substructures. Now, returning to sentences (3) and (4), we can explain more clearly why, in interpretation (a), the quantifiers in their possessive phrases are more “accessible” (accessible) to the predicate negation or the sentence’s second quantifier: the possessor of these phrases (children) is here treated as a more or less freely chosen complement to the head (parents). That the latter alone can stand for the entire possessive phrase (and thus be quantified) is also clearly shown by the paraphrases in (a) of the sentences under discussion. 1.1.2. With this rather lengthy digression, I wanted only to say that there must be a substantial semantic and cognitive basis for using the term “syntactically ambiguous” to describe sentence structure. This does not, of course, mean that whenever a sentence can be understood in two ways, its syntax must necessarily be called ambiguous. For example, even a sentence that seems unambiguous, such as Jonas developed a fever, can have two interpretations. In typical circumstances, on hearing such an utterance, we will of course first think that Jonas’s body temperature rose, that is, that he probably fell ill. However, it is easy to imagine “technical” circumstances for this utterance (for example, working in a boiler room) in which it would mean that the temperature of a device for whose readings Jonas is responsible had risen. The first actant (Jonas) and the second (temperature) in this and similar sentences are also linked by the functional model of the reference point, although the possessive relation between such actants is only implied. In the first, prototypical interpretation, the patient dative profiles the reference point itself (Jonas as a person), whereas in the second it profiles the domain associated with that reference point (reference point’s dominion), in this case, the domain of the person’s responsibility. However, it is difficult to regard a sentence analyzed in this way as syntactically ambiguous, since the causes of its dual interpretation are external and context-dependent rather than internal and determined by structural properties. 4 Dėl sintaksinio dviprasmiškumo 1.2. The article makes clear that the two constructions of predicative possessive sentences discussed earlier by the author—those of possession and belonging—offer little support for characterizing the semantic specificity of sentences such as (1) and (2). It is true that, in certain respects, the information structure of sentences representing these constructions may be compared with that of the sentences under discussion and may, for lack of a better argument, support seeking a place for the latter among the former (that is, among the two types of possessive sentences). Functionally, however, the sentence types being compared differ greatly: neither the meaning of possession nor that of belonging corresponds to the functional purpose of sentences such as (1) and (2)—“to describe a person by a certain characteristic attribute” (Holvoet 2005: 155). If we consider this argument from semantic inadequacy to be particularly important, we may question whether the construction under discussion should be regarded as possessive at all. The possibility of this hypothesis will be discussed later, but if it proves correct, then the test based on the correlation between predicates with negation and subject genitive—which helps distinguish existential sentences from locative ones and, correspondingly, possession sentences from belonging sentences—cannot be applied to sentences such as (1) and (2). In view of all this, describing the construction under discussion as hybrid may be misleading. 2.0. Before proceeding to a more detailed analysis of the syntactic structure of sentences such as (1) and (2), it would be useful to discuss several more general issues concerning the relationship between sentence constituent structure and grammatical functions. When semantic arguments lead us to suspect that the same sentence may have alternative constituent structures, it is natural to assume that alternative syntactic analyses also correspond to alternative distributions of grammatical functions. Let us see whether this is actually the case with a classic sentence involving the alternative interaction of two logical operators, where there are sound grounds for speaking of syntactic ambiguity. So, let us briefly return to the reasons for the dual interpretation of sentences (3) and (4), discussed just above. 2.1. As we have seen, the two interpretations of such sentences are essentially determined by the possibility of distinguishing two kinds of syntactic relations within their noun phrase, which consists of a possessive construction with a specific meaning: in one case, (a), its head—the possessed entity—is merely a complement of the other constituent, the possessor; in the other, (b), it is modified by the latter. It should be noted, however, that in both cases the head representing the noun phrase, and hence its grammatical function, remains the same (parents). Consequently, one cannot claim that at the sentence level the two interpretations 5 ROLANDAS MIKULSKAS The interpretation of such sentences corresponds to a twofold distribution of its grammatical functions; a certain, indirect correspondence exists only within their complex noun phrase. 2.2. That the ambiguity of sentences of this type is not related to the distribution of their grammatical functions is shown even more clearly by examples with a simpler noun phrase structure, cf.: (5) Everyone here will not take risks. (a) “Not everyone here will take risks”, (b) “Everyone here will not take risks”. (6) Few people read many books. (a) “There are few people each of whom reads many books”, (b) “There are many books that only a few people read”. In the case of the (a) interpretations of these sentences, the first logical operator takes the second within its scope, whereas in the case of the (b) interpretations, the reverse is true. For example, the (a) interpretation of sentence (5) is determined by the fact that the quantifier kiekvienas ‘each’ takes the sentence negation within its scope, i.e. the negative predication nerizikuos ‘will not take risks’ is assigned to all the elements of the assumed referential mass. Interpretation (b) is determined by the opposite disposition of the logical operators: negation takes the quantifier within its scope, i.e. the negative predication is assigned to only some elements of the assumed referential mass. 2.2.1. The fact that the twofold interpretation of the sentences under discussion cannot be associated with an alternative distribution of their grammatical functions requires us to seek the causes of their ambiguity in an alternative syntactic analysis. At least at first glance, however, it seems that we must posit the same constituent structure in both interpretations of these sentences. Of course, this is so if we do not postulate, following the generativists, some kind of deep structures. We recall that in the generative tradition, the syntactic analysis factor is extremely important for the semantic analysis of a sentence. Representatives of generative semantics (cf. Lakoff 1969) attempted to explain the ambiguity of sentences with two logical operators by postulating two different deep structures. To this end, logical operators were equated with verbal predications that define separate propositions in the deep structure. This view had to be adopted in order to show that, in the deep structure, one logical operator is brought within the scope of another by means of syntactic subordination 6 Dėl sintaksinio dviprasmiškumo by means of. Thus, through syntactic derivation (while also applying a certain rule for “lowering” logical operators, assigning them to the corresponding nouns in subordinate clauses of deep structure), two different “deep” constituent structures are derived from a “surface” structure that is unambiguous in terms of syntactic scope, corresponding to alternative interactions between logical operators. These are considered responsible for the dual interpretation of the sentences in question. 2.2.2. It was soon observed that distinguishing the alternative constituent structures of the sentences in question in terms of syntactic subordination is not sufficient to describe all the meanings they can possibly have. For instance, in principle, such an analysis cannot distinguish cases in which neither of the logical operators used in these sentences falls within the scope of the other. For example, the content of sentence (6) can also be understood to mean that simply a small number of people read a large number of books. This would be a third interpretation of the sentence, determined by a disposition of its quantifiers in which neither of the two is dominant over the other. 2.3. Cognitive Grammar devotes considerable attention to questions concerning the interaction of quantifiers and negation in the sentence (cf. Langacker 1991: 118–141). Unlike generativists, proponents of this linguistic approach do not emphasize the importance of a sentence’s syntactic form for its semantic analysis; in the case under discussion, they do not associate the different interactions between quantifiers and negation with alternative constituent structures—especially since such structures can be distinguished only at the level of deep syntactic analysis of the sentences in question, which is precluded by the methodological framework of Cognitive Grammar. For here, only the conceptualizations of structures that have phonological expression are recognized2. Avoiding the complex diagrams drawn by cognitivists and steering clear of specialized terminology, I will briefly outline below how they explain the reasons for the different interpretations of so-called ambiguous sentences, highlighting only those aspects of their analysis that may help resolve the questions raised in our discussion. 2.3.1. First, it should be noted that proponents of Cognitive Grammar reject reducing the semantic analysis of the sentences in question to logical propositions 2 The central postulate of Cognitive Grammar requires that only phonological structures, semantic structures, and the symbolic relationship between them—that is, symbolic structures—figure in a description of language. Taylor (2002: 292) formulated this restriction as the symbolic thesis. 7 Rolandas Mikulskas interaction, considering such an approach overly restrictive. Accordingly, in their analysis, so-called logical elements are not regarded as logical operators in the strict sense of the term, but are defined as units of complex semantic structure. The interaction between two quantifiers, or between a quantifier and negation, is explained by conceptual links connecting their substructures (correspondences). Which compositional path is followed in constructing a complex linguistic structure—a sentence—is not very important to its interpretation. Thus, one or another segmentation of a sentence structure with the same phonological expression does not determine the possibilities for interpreting that sentence differently. Such possibilities arise only through different ways of conceptualizing the sentence as a complete compositional structure. Different conceptualizations of the same sentence depend not on its taxonomic segmentation, or on differing priorities among levels of composition, but on the possibilities of the type of (complex) process profiled by that sentence to form different combinations of its instantiations at certain stages (levels) along the compositional path, whose specifications are inherently endowed with a distinctive hierarchical structure and can subsequently be modified and quantified. 2.3.2. To clarify what is meant, let us examine sentence (6) as an example. It is reasonable to take as the basis for conceptualizing sentences of this kind with quantifiers not a single instantiation of a process type (reads), but a replicated variant of it, whose multiplicity is indicated by the plurality of the process’s participants. In this case, such a multiple instantiation is expressed by the complex structure people read books. We can immediately say that this structure can be conceptualized in three ways, or that three specifications of the complex process type can be distinguished on its basis: (i) ‘a few people read a few books’, (ii) ‘the same person reads a few books’, (iii) ‘the same book is read by a few people’. The simplest case to explain is the further conceptualization of specification (i): the quantifiers few and many simply specify, respectively, the social group size of the people reading books and the books read by them. The alternative scope relation between the quantifiers depends on which of the two specifications—type (ii) or (iii)—is chosen as the basis for the sentence’s final conceptualization. In other words, at a higher level of conceptual organization, one of the two complex process types is replicated by relating it, respectively, 8 Dėl sintaksinio dviprasmiškumo with instantiations of the first or second arc. By choosing the (ii) specification of the replicated-process type (skaito), we thus obtain interpretation (a) of sentence (6), in which the quantifier mažai has the quantifier daug within its scope. By choosing a type (iii) specification, we obtain interpretation (b) of sentence (6), in which the arrangement of quantifier scopes is reversed. 2.3.3. It is important to understand correctly what is meant when it is said that the different interpretations of such sentences are determined by a certain hierarchical structure inherent in the process-type specifications selected in advance, operating at a higher level of conceptual organization and producing different quantifier-scope effects. This entire hierarchical order of conceptual organization should be understood dynamically, that is, as a certain procedural asymmetry in conceptualization, whose direction is associated with the choice between specific and nonspecific reference to the participants in the quantified replicated process. Suppose that, as the basis for interpreting sentence (6), we choose process-type specification (ii). At the subsequent stage of conceptualization, the referents of the quantified subject noun (people) are then construed as specific individuals, whereas the referents of the object noun (books) remain nonspecific, even when it is quantified (many books). In cognitive terms, the referents of the first sentence arc are instantiated by the conceptualizer in physical space, whereas those of the second arc remain in the mental space defined by the type specification. At the final stage of sentence conceptualization, the first quantifier (few) distributes this element of the replicated-process type (ii) specification, ‘many books’, among all the specifically construed individuals (people). In the case of interpretation (b) of the sentence under discussion, where process-type specification (iii) is selected as the basis for further conceptualization, the referents of the object noun (books), by contrast, are quantified as specific individuals, while the referents of the subject noun (few people) remain in the mental space of the type specification. In the language of logic, this strategy of alternative conceptualizations associated with specific and nonspecific reference is called interaction between different quantifier scopes. 2.3.4. Somewhat differently, though again through the interaction of specific and nonspecific reference, we can explain the alternative conceptualizations of sentences with a subject quantifier and a negated predicate. The content of the negative predicate in such sentences can be defined as a situation localized in discourse space with respect to the speaker resp. hearer (in grammatical tense), from which a certain process is eliminated. For example, the negative predicate of sentence (5) 9 ROLANDAS MIKULSKAS will not risk can be defined as denoting a future situation from which the process ‘(someone) risks’ is eliminated. At a certain stage of sentence conceptualization, this negatively defined situation may be selected as a situation-type specification for the final conceptualization, during which the subject quantifier replicates that specification and distributes it among specific (specifically referenced) individuals. Thus, in interpretation (a) of sentence (5), the quantifier each assigns to every individual in the mass of assumed referents a description of the type of future situation from which the process ‘(someone) risks’ is eliminated. In the alternative interpretation of such sentences, their negated predication denotes a situation from which what is eliminated is not an abstractly construed process, whose actant can be defined only schematically, but rather a process whose actant is described by the type specification expressed by the sentence subject. For example, in interpretation (b) of sentence (5), the process ‘risks’ eliminated from the future situation is one whose actant represents all the elements of the mass of referents assumed by the subject, which here is also the quantifier (for the type specification of quantifier each is constituted by the representativeness of the element it singles out (profiles)3). 2.4. We would like to conclude this discussion of the cognitive analysis of sentences involving the alternative interaction of two logical elements with a few interim conclusions that will be useful for the further analysis of the type of sentences represented by (1) and (2), which concerns us here. Language often exhibits cases in which the semantic contrast between pairs of sentences that apparently convey the same situation is supported by external structural differences, such as a different distribution of grammatical functions (the passive construction) or clearly different constituent structures (cf. Pabučiuok mamai ranką 3 This characterization of the meaning of the quantifier kiekvienas, -a is considerably simplified. In Cognitive Grammar, the entity profiled by this quantifier is defined, as usual, on the basis of a certain base concept, which, while remaining outside the profile, is only implied. It is argued that the semantic base of the quantifier kiekvienas, -a (English every, everyone, everybody) consists of the concept of a replicated process that it activates, comprising multiple instantiations of a schematically construed process type. The higher-level (collective) trajector (TR) of such a process (or another figure in the relation expressed by the verb) coincides with the assumed mass of reference (Ri). Within this base concept, the quantifier kiekvienas, -a profiles one freely selectable element ti of that collective trajector TR, which represents all the elements of the mass of reference Ri insofar as it is equivalent to all other TR instantiations with respect to their role in the process assumed (see Langacker 1991: 111–114; 136–138 for more). 10 Dėl sintaksinio dviprasmiškumo and Pabučiuok mamą ranką), or a reversed word order (cf. Pieštukas yra stalčiuje ir Stalčiuje yra pieštukas). These structural factors can serve as significant criteria for classifying such sentences. However, it is obvious that sentences of the type represented by (1) and (2) offer few opportunities for applying similar criteria. These criteria are even less applicable to the analysis of the ambiguous sentences just discussed. As may already be apparent, it is therefore no coincidence that the internal possibilities for their alternative interpretations have been discussed at considerable length. In Cognitive Grammar, these possibilities are quite convincingly associated with different ways of conceptualizing a sentence. It remains only to briefly discuss the factors that prompt the speaker resp. hearer to choose one or another conceptualization. Once these factors have been established, it will also be easier to understand the possibilities for different ‘readings’ of sentences such as (1) and (2). 2.4.1. It should be noted that in sentences involving the alternative interaction of two quantifiers, only one participant in the profiled process is directly accessible to discourse participants; that is, the speaker resp. hearer establishes direct mental contact with this actant. The interlocutors treat the referents of this quantified actant as specific individuals. The reference of the other actant does not extend beyond the bounds of the type specification and is merely part of the semantic characterization of the first actant. This relationship between actants with specific and nonspecific reference represents one of the fundamental cognitive abilities: identifying one entity through another in discourse space. The latter performs the aforementioned reference-point (reference point) function. It is natural to assume that a cognitively more prominent figure is generally chosen as the reference point for identifying discourse referents. The fact that the subject is the distinctive figure of the relation profiled by a sentence is now widely acknowledged in linguistics4. It is therefore unsurprising that, ceteris paribus, specific reference in the sentences under discussion is assigned precisely to the subject noun, and the quantifier accompanying it is interpreted as having wide scope. More specifically, in the unmarked sentence (6) 4 Čia minėtinas ne tik pamatinis Keenano ir Comrie (1977) darbas, kuriame šie autoriai reletyvizacijos testais patikimai įrodė, kad subjektas užima aukščiausią padėtį „prieinamumo hierarchijoje“ (accessibility hierarchy). Išskirtinę subjekto padėtį tarp kitų gramatinių funkcijų patvirtino ir van Hoek (1995, 1997) darbai, kuriuose ji parodė, kad labiausiai pronominalines anaforos referencijos poreikiams yra „prieinamas“ subjektas. Tomlino (1995, 1997) ir Forrest (1996) eksperimentiniai darbai įtikinamai parodė, kad diskurso dalyvių dėmesį pirmiausia patraukia sakinio subjektu koduojama informacija ir diskurso dalyvių dėmesį patraukusi informacija pirmiausia koduojama sakinio subjektu. 11 ROLANDAS MIKULSKAS in this case (default case), interpretation (a) is chosen (mutatis mutandis, such an interpretation also applies to sentence (5), in which the subject quantifier interacts with the predicate negation). It should also be noted that other factors may affect the cognitive prominence of the noun chosen as the reference point, such as the noun’s precedence in the phonological sequence (word order), the new information it conveys, and the contrastive stress and corresponding intonation that often indicate this. All these factors can be associated with the functional sentence perspective, which is understood as a discourse category, that is, topic and comment are traditionally defined as “topic vs. comment,” “given vs. new information,” or “presupposition vs. focus” (cf. Lyons 1999: 227–232; Holvoet adopts a different conception of the functional sentence perspective; for a discussion of the advantages and disadvantages of the latter for sentence analysis, see below). 2.4.2. The communicative purpose or contextual presuppositions may discourage choosing the conventional interpretation of so-called ambiguous sentences. By means of various marked devices of the functional sentence perspective, such as word order, contrastive stress, segmental markers, etc., a listener may be directed to an alternative interpretation of a sentence. Thus, the interpretation (b) of sentence (5) may be prompted by contrastive stress on the negative predicate on the comment side, cf.: (5b) Everyone here will NOT TAKE RISKS (only the most experienced). 2.4.3. We could probably also find an accentual contour corresponding to the alternative interpretation of sentence (6), one that leads the quantified referents of the object noun to be treated as specific individuals. But here, it seems, we would more readily achieve the desired interpretation by switching the subject and object of the sentence, cf.: (6b) Few people read many books. The word order of this sentence does not seem marked. If its object (many books) is not emphasized by contrastive stress, then its functional sentence perspective should not be considered marked either, even though it is the reverse of the previous one: this time, the object noun phrase constitutes the topic of the sentence. This sentence would be most suitable in a context dealing with the low readability of many books (for example, scholarly ones). The “reversed” word order in sentence (6ᵇ) could be considered functionally motivated (this notion will be discussed later) and counted among its typical features. 12 Dėl sintaksinio dviprasmiškumo 2.4.4. The particular interpretations of sentences (3) and (4) discussed earlier may also be supported by corresponding logical accents within the complex possessive phrase. Of course, the choice of these accents depends on the relevant presuppositions, cf.: (3a) Do the parents of all the children pay the kindergarten fee? No, the parents of A FEW children (maliciously) do not pay the kindergarten fee. (3b) Are all parents of children required to pay the kindergarten fee? No, the parents of a few CHILDREN do not pay the kindergarten fee (they are exempt from it). Of course, the ambiguity of similar sentences can also be resolved by choosing appropriate paraphrases, but then, in place of some sentences, we will have other sentences. 2.5. To summarize the foregoing, if the two interpretations of so-called ambiguous sentences are associated with conceptualizations of a particular hierarchical order, then it seems possible to analyze the structure of those sentences without recourse to the notion of syntactic ambiguity. Differences in the compositional path (constituent structure) are not very significant for these conceptualizations. The conventional conceptualizations of such sentences are supported by the relative prominence of grammatical functions (primarily that of the subject noun) and functionally motivated word order; alternative conceptualizations may be prompted by factors associated with a marked functional sentence perspective. All these factors should support unambiguous analyses of syntactic structure even in more complex cases, where the two interpretations of the sentences under analysis may be associated with alternative constituent structures, as, for example, in sentences (1) and (2). 3.0. Typologically, the locative existential and possessive (ownership and belonging) sentences discussed by Holvoet in the same article are closer to the sentences (1) and (2) that concern us. In analyzing the syntax of these sentences and drawing certain typological parallels between them, the author relies extensively on the notion of the functional sentence perspective. Before deciding how far the semantic structure and syntax of such sentences can be associated with the functional sentence perspective in general, and how far their typological classification may depend on it, we should briefly recall how the article’s author understands the functional sentence perspective itself. He presented his distinctive conception of it in the first volume of Studies in the Grammar of the Lithuanian Language (Holvoet 2003: 79–98). 13 ROLANDAS MIKULSKAS 3.1. Drawing on Bogusławski’s (1977) ideas, Holvoet proposes viewing the actual division of the sentence as a layered thematic–rhematic structure. The division of the predication expressed by a sentence into levels, or layers, is primarily associated with the dichotomy between predication and reference. For it is clear that no proposition, i.e. the attribution of a property to an entity, is possible without an act of reference. The stepwise division of the sentence’s semantic structure is also based on the assumption that only one act of reference is possible at a time. Thus, in effect, layered actual division means the sequential unfolding of acts of reference and predication, corresponding in many cases to the linear structure of the sentence. At the first stage of division, the primary theme and primary rheme are distinguished. As the sentence is further divided, one or more secondary themes and one or more secondary themes and one or more secondary themes and one or more secondary themes may be distinguished 3.2. There is no room here to discuss in greater detail the advantages and disadvantages of layered actual division of the sentence in comparison with the traditional conception of semantic division 5 Šiai minčiai vienaip ar kitaip pritaria ir kiti autoriai. Antai Li ir Thomson (1976: 484) minėtą sugramatinimo procesą vaizduoja taip: „In the process of being integrated into the case frame of the verb (at which point we call them subjects), topics become somewhat impure, and certain of their topic properties are weakened, but their topicness is still recognizable“. Gyvōnas taip pat pastebi, kad bent jau prototipiškai 14 Dėl sintaksinio dviprasmiškumo conception. However, some aspects of the newly proposed division<...> ROLANDAS MIKULSKAS division, the descriptor “actual” is also not very apt, since it is dependent on the current discourse space only in part and only in “special cases”—on its presuppositions8. The categories of theme and rheme can be regarded as discourse categories only in cases of marked contrastive division or theme fronting. 3.2.2. It seems that the theme (themes) distinguished in cases of layered actual division of the sentence has little in common with the theme distinguished in cases of marked contrastive division or theme fronting. Only the latter is associated with discourse logic and can be defined using the traditional terms of given vs. new information or presupposition vs. focus. By contrast, the components of layered actual division are defined in more cognitive terms—reference vs. predication. In the latter case, the choice of theme is determined by the noun’s stable categorial properties: animacy, personhood, definiteness, and so on. According to Holvoet, these properties, together with the semantic roles performed in the sentence, constitute the notion of “inherent topicality” (topicworthiness). But these same properties also determine the choice of grammatical functions (primarily the subject) (sic!) (Holvoet 2003: 87). Thus, when actual division is understood as the sequential unfolding of acts of reference and predication, and especially when its unmarked variant is singled out, it effectively becomes inseparable from the distribution of grammatical functions in the sentence. Therefore, layered actual division can do little to assist syntactic sentence analysis: by relying on it, we risk generating tautological claims. For a certain tautology is already inherent in the definition of the theme in this kind of actual division: the sentence subject reflects the grammaticalization of the unmarked theme, while the latter is defined as the theme that coincides with the subject position.... It is no wonder that formal criteria for distinguishing grammatical functions are invoked to get out of this predicament. For example, it is claimed that “the subject, although functionally reflecting the grammaticalization of the unmarked theme, is to be defined by formal features (which most often encode the unmarked theme)” (Holvoet 2005: 146). the distribution of grammatical functions in them. In other words, the semantic contrast between these two sentences is primarily constituted by the contrasting distribution of grammatical functions, corresponding to the specific nature of the concrete situations conveyed, in relation to the lexical or referential values of the same verb’s actants (Jonas, a famous journalist). The theme, however, is the same in both sentences—Jonas—only in the second sentence it should be considered marked, since it does not coincide with the subject. But its “markedness” can still be disputed (see below). 8 Apie šią sąvoką žr. Langacker (1991: 97), plg. Mikulskas (2006: 37). 16 Del sintaksinio dviprasmiškumo 3.2.3. Although the introduction of a marked variant of layered actual division of the sentence acknowledges that this division sometimes does not coincide with the distribution of grammatical functions, thereby avoiding a complete identification of these two phenomena, the marked variant of layered actual division is, on the one hand, also dependent on the distribution of grammatical functions, albeit a different one; on the other hand, the impression is created that these “deviations” from the canonical model are treated merely as exceptions that prove the rule (cf. Holvoet 2005: 145). As we shall see later, this is contrary to the facts of language. 3.2.4. Since it is emphasized “that the notion of ‘marked (layered—R. M.) actual division’ is based on deviation from a certain typical model of actual division encoded by grammatical functions” (Holvoet 2003: 88–89), it becomes clear that this “markedness” is of a different nature from that found in cases of contrastive division or theme fronting. For in the latter cases, deviation from the neutral accentual frame or the usual word order is motivated by discourse logic (contextual presuppositions), rather than by the distribution of grammatical functions corresponding to one situation or another in reality. Of course, the prototype of actual division of the sentence is undoubtedly constituted by cases in which the primary theme of the sentence coincides with its subject, and the secondary theme with its object. The linear arrangement of these figures of verbal relation, which have a distinctive grammatical status and correspond to the canon of the transitive sentence, also testifies to the relative cognitive importance of the information encoded by the subject in relation to the information conveyed by the object. There is reason to believe that when this asymmetry (most often already encoded in the verb’s semantic structure itself) is reflected in the actual division of the sentence, its conceptualization (processing) proceeds most smoothly9. But is it meaningful to call other actual-division patterns that deviate from this one marked, especially if the distribution of grammatical functions in the sentences representing them, in relation to the semantic roles or lexical resp. referential values10 of the verb’s actants, is functionally motivated and the word order appears usual? 9 Plg. Langackerio (2001: 44) išsakytą mintį: „In particular, discourse tends to flow most smoothly when the participant chosen as subject is also the pivot corresponding to a salient topic“. 10 It is not always possible to discuss the distribution of grammatical functions in a sentence in relation to the semantic roles of its verb’s actants. When the predicates of different sentences 17 ROLANDAS MIKULSKAS 3.2.4.1. In what sense do I call the distribution of grammatical functions functionally motivated11? First, when speaking of the adequate conceptualization of a fragment of reality profiled by the corresponding sentence patterns. We could reasonably suppose that the adequacy of the chosen sentence pattern to the designated situation ultimately depends on the disposition of the archetypal semantic roles performed by its actants in that pattern and on the degree to which they are grammaticalized. For example, the canonical transitive sentence is based on the so-called ENERGY TRANSFER CHAIN ICM12, cf.: (7) The boy kicked the ball and (with it) broke the window. In this sentence, the subject position is prototypically occupied by an agent, while the object position(s) is/are occupied by inanimate entities that undergo the action of an active agent (and consequently change their spatial position or state), namely, the mover (which is also an instrument) and the patient, respectively. There is little need to demonstrate that the distribution of grammatical functions in such a sentence is cognitively motivated. The structural asymmetry of such sentences is already inherent in the meaning of their verbs and is confirmed by their prototypical actants. It can be said of the structure of a canonical transitive sentence that it iconically reproduces the asymmetry of the ENERGY TRANSFER CHAIN model. 3.2.4.2. In other cases, the asymmetry in the disposition of grammatical functions is not so pronounced (we might say that the degree of grammaticalization of semantic roles is lower here). The structural asymmetry necessary for sentence conceptualization may then be inferred by the conceptualizer from the corresponding cognitive models of reality, cf.: when expressed by the same verb, one most often has to speak of a different distribution of grammatical functions with respect to the lexical or referential values of the verb’s actants. 11The concept of functional motivation can be understood in various ways. The term itself can also be used diachronically, to denote so-called system-pressure phenomena, for example, when the emergence of newer, more grammaticalized structures motivates the reorganization of older, inherited language structures. Holvoet also uses the term functional motivation in this sense when he explains the emergence of newer habeo-type possessive constructions in place of older mihi est-type ones by the former’s greater grammaticalization. The newly selected ‘have’ predicate makes it possible to assign subject status to the theme of possessive sentences and thus to “even out” the mismatch in this respect that exists in analogous mihi est-type sentences (cf. Holvoet 2005: 149). 12See Langacker (2000: 23–34; 2002: 209–225) for a fuller discussion. 18 Dėl sintaksinio dviprasmiškumo (8) The child has a fever. (9) *The fever has a child. The distribution of grammatical functions in such and similar sentences with respect to semantic roles is determined by the inherent asymmetry of possessive relations, which is perhaps most pronounced when a whole is related to its part. 3.2.4.3. The syntactic asymmetry of some possessive constructions, for example those that profile kinship relations, depends on the speaker’s choice of conceptualization perspective, cf.: (10) These two children have a mother. (11) This mother has two children. It may reasonably be assumed that the semantic base for the conceptualization of these two sentences consists of approximately the same situation in reality, but that they encode two different profiles of that situation. The circumstances in which one or the other sentence may be uttered also differ, of course. Thus, we could assert that the semantic contrast between such sentences consists of two different perspectives on the conceptualization of the same situation in reality. The choice of one perspective or the other is captured by the alternative distribution of grammatical functions, but in this case with respect not to the semantic roles performed by the verb’s actants, but to their lexical or referential values. The asymmetry in the disposition of grammatical functions here is associated not so much with the specific nature of the fragment of reality being conceptualized as with the very nature of the conceptualization, which in this case constitutes the content of functional motivation. 3.2.4.4. In other cases, the alternative choice of a kinship-relation figure as the subject of a sentence is determined by the encoding of two different situations in reality and accordingly generates two sentences with different semantic content, cf.: (12) The bride’s father arrived. (13) Father’s bride arrived. The choice of the head and its modifier in the possessive constructions used in these sentences (and hence the functional motivation for their arrangement) is determined by two entirely different states of affairs. 3.2.4.5. The situation with respect to syntactic-distribution motivation in possessive constructions profiling kinship relations is reminiscent of that in 19 Rolandas Mikulskas equative constructions. For example, in sentences (14) and (15), the distribution of grammatical functions with respect to the lexical resp. referential values of the participants is determined by the choice of a perspective for conceptualizing essentially the same situation (appropriate to the circumstances of the utterance), cf.: (14) My wife is the editor of the publication. (15) The editor of the publication is my wife. However, this choice may be constrained by the way phenomena in reality are categorized. For example, asymmetry in the arrangement of grammatical functions in only one direction is possible in equative sentences expressing an inclusion relation, cf.: (16) Lions are a kind of cat. (17) *Cats are a kind of lion. 3.2.4.6. In some cases, two different perspectives for conceptualizing essentially the same state of affairs can be expressed in language by predicates with opposite directionality, forming pairs of so-called converses: buy vs. sell; give vs. take; give as a present vs. receive as a present; lend vs. borrow and so on, cf.: (18) Jonas gave Ona flowers as a present. (19) Ona received flowers as a present from Jonas. In such sentences, the distribution of grammatical functions with respect to semantic roles depends on the choice of predicate that serves the communicative purpose. It may be worth generalizing that speakers prototypically choose the so-called agent perspective (for the naturalness of this strategy, cf. Langacker 2001: 43), as in sentence (18); but why should choosing the recipient as subject in sentence (19) be considered less functionally motivated if it meets the communicative requirements? 3.2.4.7. The perspective for conceptualization, and the corresponding distribution of grammatical functions with respect to semantic roles, can be expressed in language not only by lexical means but also by grammatical means, of which passivization is the first to be mentioned, cf.: (20) The program is hosted by a well-known journalist. (21) The program is being hosted by a well-known journalist. Preference for passive constructions is often also prompted by discourse 20 Dėl sintaksinio dviprasmiškumo logic. Suppose that determining the referents of pronouns in anaphora, and thus interpreting them appropriately, is facilitated by a distribution of grammatical functions in which the pronoun is also the subject of the sentence; cf.: (22) Purchase this computer program with confidence. It was substantially updated recently. Although, from the perspective of grammatical structure, passive constructions are often justifiably regarded as marked variants of the corresponding active constructions (cf. Langacker 2002: 229–230), examples such as (22) show that, from a discourse perspective, the distribution of their grammatical functions can also often be regarded as motivated and natural. 3.2.4.8. To sum up what has been said, we can claim that various sentence patterns are functionally motivated if the asymmetry expressed by their distribution of grammatical functions is iconic with respect to the profiled situation. As we have seen, the direction of this asymmetry may be determined by objective factors, most often arising from the nature of the process denoted by the verb itself, but it may also be determined by subjective factors, such as the choice of perspective from which a fragment of reality is conceptualized, a choice often supported in turn by discourse logic. 3.2.5. The notion of functional motivation for the distribution of grammatical functions is relevant to our discussion only insofar as it can be grounded in the unmarkedness (resp. markedness) of the corresponding sentence patterns. If we consider only the motivations of grammatical sentence structure, we should first call all those sentences marked whose distribution of grammatical functions exhibits an asymmetry in the direction opposite to that of the aforementioned ENERGY TRANSFER CHAIN ICM vector. As we recall, the latter is iconically reflected by the canonical transitive sentence. Structural asymmetry in the opposite direction is perhaps clearest in passive constructions (of course, the distribution of grammatical functions in these constructions is not a mirror image of that in the corresponding active construction: agents in passive constructions never become their direct objects). Thus, the “grammatical” markedness of passive constructions is the most obvious. 3.2.5.1. Strictly considering grammatical motivations, sentence patterns such as (23) and (24) should also be regarded as marked; cf.: (23) The residents recovered the confiscated property. (24) This hat belongs to Antanas. 21 Rolandas Mikulskas There is reason to believe that the semantic basis for conceptualizing sentences whose predicates form converse pairs is constituted by the same idealized cognitive model. However, the direction of the process underlying such an idealized model coincides only with the direction of the process profiled by one member of the converse pair. Thus, although the action performed by the subject of sentence (23) (gyventojai) (atgavo) can, with certain reservations, be treated as active in both a physical and a mental sense (the subject’s agency is correspondingly expressed through the syncretic role of recipient–beneficiary), the “direction” of such activity is nevertheless opposite to the primary direction of the mover’s (property’s) movement, which coincides with the energy-transfer chain in the GIVING–RECEIVING ICM (see Langacker 2002: 227 for its schema). This direction corresponds only to the direction of the process profiled by duoti, the other member of the converse pair under discussion. In the possessive relation expressed by sentence (24), Antanas is the “active” member not only because he is a living being, but also because a beneficiary role can be assigned to him here, with his mental-energy vector directed toward the possessed entity (the hat). However, the noun Antanas is not the subject of this sentence. The subject (and possessee) position in sentences of this type (cf. also Ši kepurė priklauso Antanui) is usually occupied by a noun denoting an inanimate object. It is, of course, difficult to speak without substantial reservations of the direction of the process profiled by the verb in such sentences’ possessive relation. Their structural asymmetry is based on the asymmetry of the cognitive model of the possessive relation. The latter is grounded in the fact that the possessee is always identified with respect to the possessor. In other words, the noun denoting the possessor becomes the reference point for this cognitive operation (reference point). Clearly, the direction of conceptualization of the possessive relation expressed by sentences such as (24) is opposite to the asymmetry in the arrangement of grammatical relations, which coincides with the direction in which their phonological sequence is “read” (processing in the narrow sense). Therefore, all sentences that profile a possessive relation should also be considered marked, despite their conventionality. Many other sentence patterns should likewise be considered marked (including the one of interest to us, represented by sentences (1) and (2)), whose distribution of grammatical functions, or merely their arrangement13 13 Cf.: (i) Vaikui įkando gyvatė. (ii) Gyvatė įkando vaikui. In both sentences, the distribution of grammatical functions among their verb’s arguments 22 Dėl sintaksinio dviprasmiškumo deviates in one way or another from the transitive-sentence canon, although their conventionality and adequacy for conveying the situations (functional motivation) are beyond doubt. We might reasonably suppose that the conventionality of many such sentences arises from the repeated conceptualization of the same real-world situations, driven by communicative needs. This is why there is reason to speak of distinct sentence types. By calling such sentences marked solely because their grammatical structure deviates from the canonical model, we violate the empirical facts of language in many respects. It seems far more meaningful to apply the distinction “marked vs. unmarked” to the description of sentences (or sentence types) on the basis of the conception of functional motivation set out above. 3.2.5.2. Of course, classifying sentences as marked or unmarked solely on the basis of the criterion of canonical grammatical structure is much simpler. It may be assumed that applying the criterion of functional motivation in this classification may be problematic in many cases, and therefore the possibilities for its application remain to be investigated. Even now, however, we can anticipate that the concept of functional motivation is useful in cases where there is reason to speak of established sentence patterns, and especially of sentence types. The very existence of conventional sentence patterns indicates that their emergence has been functionally motivated in one way or another. Therefore, conventional sentence patterns and their corresponding word orderde definitione should be considered unmarked. From this perspective, only particular realizations of various sentence patterns would be called marked, primarily those associated with an unusual information structure, the latter, of course, being understood as a discourse category. Accordingly, information structure that PALAIKANTI a functionally motivated distribution of grammatical functions would be unmarked. It goes without saying that such an unmarked information structure of a sentence need not coincide with the hierarchy of grammatical functions14. It is easy to surmise that, if we consistently adhere to the functional- with respect to the semantic roles they perform is the same, but their arrangement in the phonological sequence is reversed (as is the information structure). Although the arrangement of grammatical functions is motivated in its own way in both cases, there is reason to believe that it is more conventional in sentence (i) than in sentence (ii), despite the fact that in the latter the direction of the asymmetry it expresses coincides with the vector of the ENERGY-TRANSFER CHAIN ICM. 14 Here, no account is taken of the fact that the hierarchy of grammatical functions itself is in some respects defined ambiguously. These include indirect objects encoding various archetypal semantic roles. For example, a dative indirect object often occupies a higher position in the aforementioned hierarchy than 23 ROLANDAS MIKULSKAS The principle of communicative motivation would eliminate the distinction between unmarked, marked, and layered actual division. In seeking to preserve this distinction, we would be unable to avoid contradictions: we would have to speak of sentence models whose grammatical functions are distributed in an unmarked way, while their actual division is marked. Only cases of contrastive division and focus highlighting would then be considered marked, primarily those identifiable from an unusual word order in the sentence. Marked actual division into the predication focus or rheme highlights only the word whose marked information addresses doubts or questions expressed in the sentence’s presupposition. We could say that such actual division of the sentence is closely associated with its SPECIFIC (in Holvoet’s terms, supplementary) presuppositions. 3.2.6. It is no secret that word order in a sentence is closely related not only to the distribution of grammatical functions, but also to actual division (cf. Lyons 1999: 230). However, intuition suggests that in languages such as Lithuanian, where word order in the sentence is relatively free, there is a certain specificity in this respect. Let us examine these matters more closely. Holvoet (2003: 96) states that “actual division is an aspect of the semantic structure of a sentence, which may or may not be reflected in word order, BUT IS ONTOLOGICALLY INDEPENDENT OF WORD ORDER” (emphasis mine – R. M.). Indeed, some cases of marked division are identifiable only through contrastive stress, appropriate intonation, and/or the use of segmental markers. At the same time, it should be noted that all cases of layered actual division, which, as is already easy to understand, encompass what I propose to call functionally motivated sentence models, are essentially connected with word order in those sentences. Holvoet indirectly acknowledges this when he states that “an important difference between marked non-contrastive and marked contrastive actual division is that the latter is not always reflected in word order (ibid., 90). Hence, marked non-contrastive (i.e., layered) actual division is ALWAYS reflected in word order in the sentence¹⁵. It is understandable that unmarked instrumental or accusative objects. This is explained by the fact that the experiencer role comes immediately after the agent role in the hierarchy of initiative, and is therefore ranked higher in this respect than the instrument or patient roles (see Langacker 2002: 238–239; cf. Holvoet 2003: 88). 15 Cf. also ibid., 91: “If actual division deviates from the unmarked model encoded by grammatical functions, word ORDER BECOMES THE SOLE OR AT LEAST THE PRIMARY INDICATOR OF ACTUAL DIVISION” (emphasis mine – R. M.). 24 Dėl sintaksinio dviprasmiškumo In the case of layered actual division, it is all the more appropriate to speak of the unmarked word-order model corresponding to it (ibid., 86). 3.2.6.1. Holvoet (2003: 96–97) illustrates the relative independence of actual division from word order in the sentence with the following pair of sentences, cf.: (25) Jonukas is very talented. (= 48) (26) Very talented is that Jonukas. (= 49) It is claimed that the actual division of these sentences is identical. In both cases, the topic is the noun Jonukas, whose referent is said in both sentences to be talented. In making this claim, one important detail is overlooked: these two sentences can be equated in terms of semantic division only if, when the word order in sentence (26) is reversed, the demonstrative pronoun that, which provides a definiteness description for the postverbal noun phrase, is added to the name Jonukas. Although it is often rightly asserted that proper names are inherently definite (cf. Mikulskas 2006: 41), in this case the inherent definiteness of the proper name is not enough to prevent the undesirable interpretation of sentence (26), according to which Jonukas would convey new, rhematic information in the sentence and would almost automatically be pronounced with contrastive stress, cf.: (27) Very talented is JONUKAS, (not Petriukas). This undesirable word-order effect is related to the specific way Lithuanian ensures the coreferentiality of discourse entities. As we know, Lithuanian is not among the languages in which the definiteness or indefiniteness of noun phrases is regularly marked by corresponding articles. In Lithuanian, the coreferentiality of entities mentioned in the current discourse space is ensured by other means, such as demonstrative pronouns, definite adjectives, relative quantifiers functioning as indefinite or generic pronouns, and so on (cf. Mikulskas 2006: 33). Some facts indicate that one of the more important means of coordinating the reference of discourse entities in Lithuanian is word order in the sentence. It is entirely understandable that a noun phrase placed at the beginning of a sentence, in the prototypical topic position, is usually definite, since its referent has already been introduced into the current discourse space: generally, only a noun phrase whose referent has been identified can become a point of reference for subsequent sen- 25 ROLANDAS MIKULSKAS for clause predication16. Indeed, sometimes, when a sentence’s presupposition is extremely general in content, an indefinite noun phrase can be uttered at its beginning. But most often such sentences (at least in the traditional sense) cannot be divided into topic and comment: all the information they convey is new, cf. (situation in a traffic jam): (28) Why aren’t we moving forward? Some kind of car is blocking the road, apparently there’s been an accident. Meanwhile, the referent of a noun phrase on the comment side is regularly interpreted as a new, and thus indefinite, element in the unfolding discourse. In the literature, this kind of requirement, imposed by context or sentence position, to assign a particular value of the feature [± definite] to a noun phrase is often described as a (in)definiteness effect and distinguished from the expression of “genuine” definiteness (see Lyons 1999: 27 ff.; cf. Holvoet, Tamulionienė 2006: 30–32). However, this classificatory assessment does not diminish the functional value of the expression of such a definiteness category. On the contrary, we could argue that in articleless languages these “definiteness effects” associated with word order in the sentence are no less important for referential coordination than the corresponding lexical or morphological means. In any case, examples (25) and (26) clearly show that we cannot change the word order in a Lithuanian sentence, and thus its information structure (and hence its interpretation), without consequences for meaning: the indefiniteness effect produced by the inverted word order in sentence (26) can be neutralized only by inserting a definiteness description before the name Jonukas, which occupies a typically comment position, indicating that the sentence predication describes the mental abilities of a person mentioned earlier. Only when explicitly definite in this way can the noun phrase remain the sentence topic in comment position. 3.2.6.2. Such a close connection between information structure and sentence word order associated with the expression of definiteness calls into question the previously cited claim that information structure is ontologically independent of sentence word order 16 The identification of referents of nouns used generically, however, has a cognitive basis other than the current discourse space. There is reason to argue that the corresponding noun phrases are interpreted as indefinite, especially since in article-bearing languages they are often used with an indefinite article (Lyons 1999: 233). However, this kind of “generic” indefiniteness can in a certain sense be equated with a definiteness description (cf. Mikulskas 2006: 52–54). 26 Dėl sintaksinio dviprasmiškumo does not depend on it. After all, one of the conditions for effective communication is ensuring the coreference of discourse entities. Of course, a proponent of layered information structure need not associate the primary sentence topic they identify with the presupposition of unique identification of its referent, since this analysis allows the topic to be distinguished from its antecedent. For this reason, the sentence topic identified traditionally may not coincide with the primary topic in a layered analysis, cf.: (29) Do you remember the tall linden tree at our homestead? A storm broke it down. Following the traditional conception of information structure, in example (29) we should identify the pronoun it as the topic of the second sentence; its definiteness derives from its coreference with the antecedent the tall linden tree. The comment group would then consist of the predication a storm broke down. Applying the method of layered information structure, the subject noun a storm is necessarily identified as the primary topic already at the first stage of the analysis. Despite its initial position in the sentence, this noun is indefinite because it denotes an element newly introduced into the discourse. Nevertheless, although it is not closely connected with the definiteness of the topical noun phrase, layered information structure is tied to certain sentence word-order patterns in another way—by expressing the sequential distribution of acts of reference and predication. 3.2.6.3. As we shall see later, sentence word order, its information structure, and the definiteness or indefiniteness of the corresponding nouns are particularly closely interrelated in locative, existential, and be possessive sentences. Apparently, the very conditions for predication in such sentences are closely connected with certain typical contextual presuppositions. This contextual dependence of these sentence types can even be identified as a general typological feature they share. In each of these intransitive sentence types, the position of the subject relative to the predicate depends on the information structure specific to that sentence type. In other words, a structural asymmetry is necessary for the predication resp. interpretation of such sentences; this is usually ensured by a particular arrangement of grammatical functions, which here derives less from the semantics of the predicate—which is especially impoverished here—than from the information structure determined by contextual presuppositions and, correspondingly, from the word order characteristic of each of the sentence types mentioned. Conversely, it is precisely the definiteness or indefiniteness effects associated with word order that help properly interpret the predicates of these sentences, whose meaning is overly general. For example, the appropriate, “existential” interpretation of the verb be in existential sentences is essentially 27 ROLANDAS MIKULSKAS depends on the indefiniteness of the noun phrase that follows it (which is also the subject in Lithuanian), which in turn is implied by its rhematic position at the end of the sentence and may additionally be marked by the genitive of indefinite quantity (in the case of plural or mass nouns) or by lexical means (cf. There are some kind of ladder in the basement). 3.2.7. Layered actual sentence division makes it possible to separate its theme as fully as possible from contextual presuppositions. It is claimed that this new understanding of the theme has “considerably greater explanatory power” (Holvoet 2003: 80). This greater explanatory power should apparently be applicable to both the syntactic and semantic description of the sentence. However, such separation comes at a cost. First, layered actual sentence division necessarily identifies a theme (or themes) even in cases where division in the traditional sense is impossible because the sentence’s presuppositions are too general. Such a semantic division of the sentence merely reproduces its syntactic division, so it is difficult to call it actual. Second, in Lithuanian, where word order in a sentence is not strictly fixed, a considerable number of cases of layered actual sentence division inevitably deviate from the model encoded by the hierarchy of grammatical functions and become “marked.” Such markedness in the actual division of sentences often conflicts with the fact that their word order appears neutral, i.e. unmarked. Consequently, distinguishing unmarked and marked layered actual sentence division in practical sentence analysis gives rise to certain difficulties. For example, when comparing the patterns of existential and locative sentences, respectively There is a notebook in the drawer (= 1) and The notebook is in the drawer (= 2), it is claimed that their contrast is related to the unmarked model of actual division (Holvoet 2005: 140). This means that only the actual division of the second, locative, sentence is unmarked, whereas that of the first, existential, sentence is marked, because its theme does not coincide with its subject. At the same time, it should be noted that each of these sentence types has its own characteristic word order, which appears neutral in both cases. This fact most likely leads the author of the article to treat the actual division of the analogous existential sentence discussed below, There are candies in the box (= 3), as unmarked when this sentence needs to be compared with its genuinely marked variants, Candy is in the box (= 7) and There is candy in the box (= 8) (ibid., 141–142). 3.2.7.1. Layered actual sentence division is impractical for applied use in one further respect: its unmarked and marked variants are usually associated with different sentence patterns, making it difficult to relate such division to different reading conditions for the same sentence. On the basis of the previously 28 Dėl sintaksinio dviprasmiškumo explained concept of functional motivation, one could dispense with dividing layered actual sentence division into marked and unmarked. Understandably, this proposal runs counter to the very spirit of layered actual sentence division, since the motivation for the word order of some sentence patterns, and correspondingly for their actual division, would have to be sought in contextual presuppositions. Yet proponents of layered actual sentence division rely on these presuppositions only in cases of marked contrastive division or theme fronting. 3.2.7.2. It is uncertain how much basis there is for claiming that in Lithuanian the word order resp. actual division of sentences whose theme does not coincide with their subject is more often motivated by contextual presuppositions than that of sentences in which these figures coincide. At first glance, the so-called unmarked cases of actual division appear less dependent on context. However, we could easily find counterexamples in language. The nature of the relationship between various actual-division patterns and contextual presuppositions in Lithuanian remains to be investigated. Nevertheless, it is already possible to foresee certain problems arising in the practice of semantic sentence division in this regard. For example, one might suppose that without a certain typology of presuppositions, it may sometimes be difficult to distinguish cases of theme fronting from cases of simply noncanonical word order resp. actual division. It would seem that this issue should be of least concern to proponents of layered actual division. However, by calling actual sentence division that deviates from the model based on the hierarchy of grammatical functions marked, the problem is merely concealed, not eliminated. Nevertheless, a preliminary analysis of several noncanonical examples shows that there is no need to worry excessively about problems of this kind; cf.: (30) Juozą graužia sąžinė. (31) Kaimyną išvežė greitoji. (32) Mamai pakilo (kraujo) spaudimas. (33) Pušį išvertė vėjas. The layered actual division of these sentences should be considered marked, despite their usual word order. It is worth noting that in Lithuanian such a coincidence of the theme with the direct or indirect object is not necessarily associated with theme fronting17. The latter interpretation requires a specific, context-related 17 In an English sentence, where word order is strictly fixed, the extreme leftmost position of the object inevitably marks theme fronting (fronting, topicalization) and signa- 29 ROLANDAS MIKULSKAS modalumu susijusi presupozicija (Holvoet 2003: 94–95), temos iškėlimas dažnai paryškinimas tam tikru segmentiniu rodikliu, plg.: (34) How does Juozas feel after that event? Juozas is (thus) troubled by his conscience. The naturalness of word order is particularly evident in sentences (30)–(32), in which people are chosen as the topic. It could reasonably be assumed that such sentence patterns are motivated not only by the relevant contextual presuppositions but also by their adequacy to ordinary life situations. Naturally, far more predications expressed in spoken sentences concern people than things or abstractions. One can readily see the certain unnaturalness of choosing the reverse topic, cf.: (35) ?Conscience is troubling Juozas. A “neutral” reading of this sentence, in which the subject noun conscience would also express the unmarked topic, is difficult to imagine. It sounds natural only in a particular, specific context in which the person named Juozas is a new discourse element and thus the rheme of the sentence. It is natural to expect Juozas’s name in sentence (35) to be uttered with contrastive stress, i.e. Conscience is troubling Juozas (not Petras). By the way, the same information can also be conveyed in a word order that is closer to the animacy hierarchy, cf.: (36) (Thus) Juozas is troubled by his conscience. The word order in sentence (33) is the same as in the first three sentences, but the semantics of its actants provide less basis for the “inherent” asymmetry of the sentence, since neither of them has the feature of animacy and, apparently, at least in this respect, they can claim the role of topic to approximately the same degree, cf.: (37) The wind uprooted a pine tree. It seems that the choice of topic in sentences (33) and (37) depends only on contextual presuppositions, yet both sentences can easily be read “neutrally,” and their information structure can accordingly be considered unmarked. In other words, in both cases the topic position in the sentences is occupied by definite noun phrases, although, it goes without saying— —tically discusses marked information structure, e.g., Bill I admire; That friend of yours I really don’t like. (cf. Lyons 1999: 230). 30 Dėl sintaksinio dviprasmiškumo Indeed, their definiteness is supported by different antecedents. Although in examples (33) and (37) the distribution of grammatical functions is the same, and thus their identity and syntactic structure as well, semantically they could be regarded as separate sentences because the definiteness disposition of their actant nouns is diametrically opposed. By contrast, sentences (35) and (36) are semantically identical, although their word order differs in a way similar to that of the first two. Examples (30) and (36), however, should be regarded as different conceptualizations of the same sentence pattern, since their semantic contrast arises from their different information structures, which is unmarked in the first case and marked (contrastive) in the second. A brief analysis of sentences whose word order diverges from the canonical transitive-sentence pattern shows that marked cases of information structure—whether contrastive structure or topic fronting—are not particularly difficult to distinguish from unmarked ones. 3.3. Methodological problems arising from the excessively close interdependence between layered information structure and the distribution of syntactic figures are addressed in cognitive grammar by defining the subject and object of the sentence on new grounds. This makes it possible to treat topic exclusively as a discourse category, while reserving for the subject the role of the sentence’s primary actant. The need to distinguish sentence topic from subject, despite the identity between them often emphasized in the linguistic literature, was recently acknowledged by Ronald Langacker (2001: 11–48), the author of the Foundations of Cognitive Grammar himself. I would like to conclude the discussion of sentence information structure by briefly recounting his views on this issue. It should immediately be noted that this author does not deny a certain functional commonality between topic and subject, for which convincing evidence has been presented in the work of more than one recent linguist. For analytical purposes, however, these categories should be kept distinct, even though in some extreme cases they may be realized by the same element of sentence structure. 3.3.1. In cognitive grammar, the basis for such a distinction is provided by defining the principal syntactic figures of the sentence—the subject and object—on new, psychological grounds. The essence of the process profiled by a finite sentence lies in the dynamic nature of the conceptualization of the relation expressed by its verb, which, among other things, makes it possible to speak of certain processing-time costs. It is these time costs that primarily account for the asymmetry of the relation expressed by the verb. The endpoints of this asymmetry 31 Rolandas Mikulskas are fixed (the asymmetry is formalized) by the so-called figures of first and second cognitive prominence—the trajector (trajector) and landmark (landmark). Both figures, or at least one of them—the trajector, in the case of an intransitive verb—are inseparable elements of the verb’s conceptual structure. They should be understood as reference points for nominal reference in the broadest sense, without which it is difficult to conceive of any process. Defined in cognitive terms, the trajector and landmark of a verbal relation form the conceptual basis of the principal grammatical functions—subject and object. Additional semantic characteristics of these syntactic figures (the so-called archetypal roles of agent, patient, mover, instrument, and experiencer) are induced from language use, which reflects the prototypical participants in the corresponding cognitive models (the principal one, the ENERGY TRANSFER CHAIN ICM, was mentioned earlier). 3.3.2. Let us now briefly return to layered sentence information structure. As we recall, it is in fact related to sentence-processing problems rather than to discourse logic. The claim underlying the concept of layered information structure—that two acts of reference cannot occur simultaneously—is linked to the nature of attentional focus and thus has a psychological basis. At the same time, it indirectly points to the temporal parameter that is important for conceptualizing complex (syntactic) structure. It is precisely these processing-time costs that make it possible to distinguish primary and secondary topics in information structure. It is easy to see that, in essence, these are very close to the figures of relative prominence distinguished in cognitive grammar—the trajector and the landmark. Their arrangement is precisely what the sentence subject and object encode grammatically. Formally, this encoding is manifested by the nominative resp. accusative cases or by the corresponding word order. It is also easy to see that the same parameters relevant to the “independent topicality” of a noun phrase also constitute the grammatical characterization of the sentence subject. From what has been said, it becomes clear that once the sentence subject and object are defined as the figures of first and second cognitive prominence in a syntactic relation, these concepts essentially take over the functions of primary and secondary topics distinguished in layered information structure, while the concept of topic itself is thereby freed up for analytical discourse purposes. In other words, the concepts of trajector and landmark developed by cognitive linguists solve the sentence-processing problems important to proponents of layered information structure that arise from the reference vs. predication dichotomy, from 32 Dėl sintaksinio dviprasmiškumo the reference vs. predication dichotomy. 3.3.3. The concepts of topic and subject, understood in this way, are united by the shared function of the noun phrases they define—that of a point of reference. The nouns functioning as topics and subjects perform this function only within their respective domains. The domain of a single topic may consist of anywhere from one sentence to an entire sequence of sentences. It is precisely the same topical noun that helps to properly interpret the content of the propositions expressed by all of them. The domain of the subject as a point of reference does not extend beyond the sentence. In other words, the subject noun helps to properly understand the content of the relationship expressed by the verb. In this respect, the object of the sentence is second in importance. Thus, in interpreting the content of the proposition expressed by a sentence, the subject and object nouns form a kind of chain of points of reference, comparable to those formed by the elements of complex possessive or locative constructions (cf. my mother’s sister’s husband’s lawyer; you can find the third volume of works on Lithuanian grammar on the website, on the top shelf of the side bookcase, next to the neatly arranged issues of the journal Acta Linguistica Lithuanica numbers). 3.3.4. The concepts of topic and subject can be distinguished not only by the domains of their nouns, which serve as different reference points and coincide only in a limiting case, but also by the different positions of these reference points in relation to their domains. A topic noun usually remains outside the interpretive field it creates. Sentences are incorporated into its domain as a reference point solely in accordance with the internal logic of discourse development. By contrast, the subject noun serving as a reference point is an immanent element of sentence structure. This element is present in abstract form already in the very concept of the verb (its conceptual structure), and is only specified semantically when a finite sentence is constructed and the actants corresponding to the situation it encodes are selected. 3.3.5. If we assume that a shared topic helps us properly understand the propositional content of an entire sequence of sentences, then in each of these sentences we will find a noun phrase (often its pronominal variant) that is coreferential with that topic18. We could say that such a topic functions as the common referential anchor for all 18 This situation is somewhat idealized. In language use, sentences in which the noun phrase coreferential with the topic is not expressed and can only be approximately recovered implicitly may also be included in the interpretive field of a shared topic (especially if we consider an entire paragraph). 33 ROLANDAS MIKULSKAS The common antecedent of these noun phrases. The noun phrases under discussion become a kind of anchor points (pivots), enabling the properties predicated by sentences to be assigned to the appropriate referent, which is what the topic indicates. It is clear that this coreferential function of the noun phrase becomes apparent only in the course of discourse. On the basis of his preliminary observations, Langacker (2001: 12–13) believes that, in order of priority, the first candidates for the coreferential anchor-point function are the subject, the object, or the possessor of one of these two. In his view, more extensive research would only confirm this tendency. The data discussed above show that it is quite common in Lithuanian to place a direct or indirect object at the beginning of a sentence. This fact suggests that the hierarchy of sentence constituents formulated by Langacker—candidates for performing the anchor-point function—is probably less rigid in Lithuanian; cf.: (38) You should try our cake: (a) it is baked here according to a special recipe, (b) they bake it here according to a special recipe, (c) its BATTER is mixed here according to a special recipe, (d) its BATTER is mixed here according to a special recipe. At first glance, it is not easy to say which of the comment variants is preferred in language. Cf. also: (39) Yesterday we brought our father home from the hospital. A week ago, they successfully removed his appendix there. (40) Suspected individuals were detained today on the Lithuania–Russia border. A large quantity of smuggled cigarettes was confiscated from them. We could speak with greater confidence about priorities in selecting sentence constituents for the coreferential anchor-point role only after conducting more extensive research. 3.4. To summarize this somewhat protracted discussion of the actual division of the sentence, we can say that the development of cognitive grammar not only makes it possible to redefine the notion of topic specifically for the purposes of discourse analysis, but also allows us to distinguish more clearly between unmarked and marked cases of semantic organization in the sentence. This, in turn, helps us better model the conditions under which the same syntactic structure can be “read” in alternative ways. 4. Taking a more flexible view of the semantic organization of the sentence than the notion of layered actual division allows, let us examine more closely the existential- 34 Del sintaksinio dviprasmiškumo …tive and locative sentences with the predicate būti ‘be’. We have previously noted that the different conceptualization of a sentence with the same syntactic structure (the same distribution of grammatical functions) depends to some extent also on its actual division. Let us now try to answer the question of whether a marked actual division can change the sentence type. 4.1. Let us first recall what the article under discussion says about sentences of the types in question (Holvoet 2005: 139–147). Let us rewrite the sentences analyzed here as (41) and (42): (41) There is a notebook in the drawer. (= 1) (42) The notebook is in the drawer. (= 2) These two sentences seem to consist of the same semantic elements. According to the author of the article, their common feature is that both mean “the presence of a certain thing in a certain place.” As the author writes, “the contrast between sentences (1) and (2) is not manifested in a different distribution of grammatical functions: in both cases the same noun (sąsiuvinis) functions as the subject. The contrast is related to the actual division of the sentence, or more precisely, to its unmarked pattern <...> In an existential sentence, the theme is the place where a certain thing is asserted to exist; the rheme is the assertion of that thing’s existence. In a locative sentence, the existence of the thing is presupposed, and the rhematic assertion consists only in its being in a certain place” (ibid. 140). He correctly goes on to note that the difference in actual division is not the only difference between these sentences. In fact, according to the author of the article, there are more such differences—“first and foremost, they have to do with the meaning of the verb būti ‘be’. In an existential sentence, the meaning of this verb is specifically existential—it asserts the existence of a thing (entity)” (ibid. 140–141). 4.1.1. By making this claim, one also acknowledges that the difference in actual division is not the sole cause of the semantic contrast between these sentences. However, it remains unclear what circumstances lead the verb būti ‘be’ to have specifically an “existential” meaning in sentence (41), whereas in (42), the locative sentence, the meaning of this verb is, in the author’s words, “difficult to specify at all” (ibid. 141). It is easy to see that the definiteness of the subject noun in the locative sentence and its indefiniteness in the existential sentence remain the only constant elements of the semantic structure of these sentences that are independent of actual division (in the sense that they are primary with respect to it), and that make it possible to conceptualize their verbs accordingly, i.e. 35 Rolandas Mikulskas to assign them a “locative” meaning in the first case and an “existential” meaning in the second. To explain the connection between the “existential” interpretation of the verb būti ‘be’ and the indefiniteness of the postverbal noun phrase, it is enough to examine the ontology of existence predication itself. It seems that one can assert the existence of an entity only when its status in reality is in doubt. We may reasonably suppose that this ontologically necessary doubt is formally expressed by the indefiniteness of the aforementioned noun phrase (the (in)definiteness effect associated with existential predication has, after all, been widely discussed in the literature; see Lyons 1999: 236–246). Of course, in making this claim, one should not forget that the main function of definiteness descriptions is to ensure the coreferentiality of the denotatum of a noun phrase in the current discourse space, or, more simply, to indicate the identity of the referents of the noun phrase and its antecedent. It was shown earlier that in Lithuanian sentences, the indefiniteness of the postverbal noun phrase is prototypically implied by its rhematic position. A closer look at the expression of existential predication allows us to clarify that such an indication of the noun phrase’s indefiniteness, associated with the actual division of the sentence and the corresponding word order, only iconically reflects the ontological conditions of the proposition in question, but is not directly related to the position of the phrase’s referent in reality. After all, the definite description itself merely indicates the possibility of uniquely identifying the referent of a noun phrase that has already been introduced into discourse or is being introduced anew. It is clear that the status of an entity in reality and its status as the referent of a noun phrase in discourse space are not identical: the referent of a definite noun phrase may be instantiated and identified in mental space as well (for example, the space of assumptions, conjectures, or expectations). However, the strong correlation between existential propositions and the indefiniteness of postverbal noun phrases, which allows us to speak of the iconicity of existential sentence expression, suggests that speakers often identify these statuses with one another. Usually there is no reason to doubt the existence in reality of the referent of a definite noun phrase; we might say that it is “naively” presupposed. Conversely, a referent newly introduced into discourse (by means of an indefinite noun phrase) does not have this kind of “existential” presupposition, and can therefore be asserted when necessary. 4.1.2. We can see the prototypical nature of the correlation between an existential proposition and the indefiniteness of a postverbal noun phrase by trying to imagine that sentence (41) asserts not just any notebook, but a specific one (recently 36 Dėl sintaksinio dviprasmiškumo the existence (i.e., being in the drawer) of the notebook mentioned in the conversation. Such an interpretation of the sentence is possible, but less likely; cf.: (43) Is there an account book in the drawer? (a)?Yes, the account book is in the drawer. (b) Yes, the account book is in the drawer. This example shows, among other things, that when confirming the existence of its referent, it is more usual to place a definite noun phrase at the beginning of the sentence. This is understandable, since the rheme of sentence (b) consists solely of the verb is, which expresses that confirmation. 4.1.3. If sentence (41) nevertheless refers to a specific, previously mentioned notebook, then the location expressed in it will most likely be emphasized by contrastive stress; cf.: IN THE DRAWER is a notebook (but not on the table). It is easy to see that, in terms of information structure, this sentence is the marked variant of the locative sentence (42): this is indicated not only by the contrastive stress on the rheme (IN THE DRAWER), but also by its unusual position at the beginning of the sentence. Thus, sentence (44) has nothing to do with an existential statement. 4.1.4. If we correctly relate the ontological conditions of existential predication to the indefiniteness of the postverbal noun phrase, then, at least at first glance, it is difficult to explain the entirely ordinary use in existential sentences of proper names regarded as inherently definite; cf.: (45) Jonas is in the room. But the inherent definiteness of a name used in this way merely means that its bearer is familiar to the interlocutors. Such definiteness is semantic in nature (its basis is the ICM OF PROPER NAMES) and in no way points to a coreferential noun phrase in the current discourse space, or, in other words, it does not imply an antecedent. Since this implication is a necessary condition for a noun phrase to be topical, a proper name, lacking it, can freely occupy the rhematic position in an existential sentence. 4.1.5. The specific nature of proper-name use in existential sentences is further highlighted by another aspect of these sentences’ iconicity. In a naïve worldview, the question of the existence of a human being, or of a living creature in general, is equated with the question of whether it is alive: a dead person does not exist! The type under discussion 37 ROLANDAS MIKULSKAS In sentences asserting the existence of a person, the bearer of a name, what is meant is rarely their being alive: existence here should be understood as that person’s being in a particular place. The person’s being alive, that is, existence in general, is usually not called into question in sentences of this type; it is simply presupposed, or accepted as given. Thus, when speaking about people, we could distinguish between their existence (being) in a particular place and their existence (life) in general. When the existence of things is asserted in sentences of the type under discussion, it too is usually localized. In this case, however, it seems that a thing’s being in a particular place is equated with its existence in the world in general. 4.1.6. From these observations we can draw an important conclusion: existential sentences with the verb būti rarely concern the existence of a thing or person in general; the assertion of their existence matters to the speaker or listener only insofar as these entities are accessible to them or available to their control. In other words, the ‘existential’ meaning of the predicate (būti) in such sentences prototypically implies a beneficiary. If this is true, then the emergence and use of the possessive mihi est-type construction alongside existential sentences with the verb būti is entirely understandable. The use of locatives in the latter is also easily explained: indicating the location of an entity simultaneously marks its degree of accessibility. 4.1.7. We could reasonably assume that, in sentences of the type under discussion, the prototypically localized existence of a thing is iconically matched by the indefiniteness of the noun naming it—or rather, by the non-topicality expressed in this way, which signals that the noun has no antecedent in the current discourse space. In other words, a thing whose existence is asserted in such a sentence is usually introduced into the discourse for the first time. The iconic relationship between the existence of a thing localized in reality and a noun “unlocalized” in the current discourse space (its referent is not uniquely identifiable in this space) is inverse. It is easy to see that this relationship is grounded in the equation of a thing’s accessibility to the participants in a conversation in reality with the “accessibility” (i.e. recognizability) of the corresponding noun’s referent in the current discourse space. In view of all this, the information structure of existential sentences of the type under discussion, and the corresponding word order in them, can be regarded as functionally motivated. Therefore, not only (42), the locative sentence, but also (41), the existential sentence with reverse word order, can rightly be called unmarked. 38 Dėl sintaksinio dviprasmiškumo 4.2. Reverse information structure is the most immediately apparent distinguishing feature of the existential and locative sentences of the type under discussion. If this feature is essential to their typological classification, then the various marked information-structure variants of these sentences should not alter their meaning so much as to change the sentence type, that is, turn an existential sentence into a locative one or vice versa. Examples such as (44), mentioned earlier, should not mislead us. Although the word order of this sentence is the same as that of (41), the existential sentence, it would be incorrect to group it with the latter, since in fact it is a marked variant of (42), the locative sentence. It seems that the information structure of the sentences in question, and the corresponding word order, determine their typological affiliation insofar as they help express or identify the indefiniteness of the subject noun in one case and its definiteness in the other19. For example, the thematic position of the subject noun at the beginning of the locative sentence (42) also indicates the definiteness of that noun, which is necessary for expressing location: only an identified referent can be localized. To interpret such a sentence as existential, we must first “remove” the definiteness of its subject noun. This is usually possible only by giving it a contrastive accent and thus making it rheme, cf.: (46) THE NOTEBOOK is in the drawer (not the pencil). But this sentence is a marked variant of (41), the existential sentence. 4.3. When the defining description of the subject noun is expressed explicitly, the typological independence of the sentences in question from their marked information structure is even more evident. At least this is so when the noun— 19 Cf.: (i) Antanas returned home late in the evening. THE YOUNG MAN was completely exhausted. (ii) Sigitas got back only at night. His colleagues drove THE FOREMAN home. It is easy to see that the shortest possible distance in the phonological sequence separates the sentence-initial anchor noun phrase from its antecedent topic. We could reasonably assume that this is the most economical way to ensure the definite interpretation required for its coreferentiality. Although definiteness is an essential feature of the anchor noun phrase, traditionally called simply the topic, this information-structure figure should not be equated with a definite noun phrase—the concepts are defined on different grounds. It is understandable that the information structure correlating preverbal definiteness and postverbal indefiniteness of the noun phrase is unmarked. In fact, only this kind of information structure could be considered an inherent feature of the semantic structure of sentences classified in typological terms. 39 ROLANDAS MIKULSKAS The indefiniteness of the noun phrase in existential sentences is marked by the partitive genitive, cf.: (47) There are notebooks in the drawer. This sentence could be an answer to the question What is in the drawer? If the noun denoting the subject of such a sentence is moved into topic position, its existential character remains readily discernible, cf.: (48) Are there any (more) notebooks? There are notebooks (in the drawer). The frame of this sentence continues to be formed by the existentially affirming predicate yra. A sentence with the same word order can also have another kind of marked information structure as its frame, with only the location singled out by contrastive stress (STALČIUJE). Of course, the presupposition of such a sentence is somewhat different, cf.: (49) Kur (dar) yra sąsiuvinių? (Jų (= sąsiuvinių) yra) STALČIUJE. It is interesting that in a sentence with this kind of information structure, the existential predicate ends up on the topic side and thus has presuppositional status. Holvoet (2005: 142) also draws attention to this when discussing an analogous sentence, There are sweets in the box (= 8). Although the information structure of sentences such as (49) and (= 8) resembles that of the corresponding locative sentences, following Holvoet, we could regard them as existential sentences. And here is why: in these sentences, the presupposed existence of the noun phrase referent is of a different kind than in locative sentences. In the latter, the reality of the localized referent (i.e., its existence in reality) is not in doubt. In the former, by contrast, only an assumption is made that the referent of the noun denoting their subject exists in reality. We might even say that the existence of this referent is presupposed only in the mental space of assumptions, or conjectures. The partitive genitive most clearly signals the presupposition of this hypothetical existence. Undoubtedly, the function of sentences such as (49) and (= 8) is to confirm this conjectured existence of the noun phrase referent. Nevertheless, it must be noted that this confirmation of existence is expressed in a distinctive way in the sentences under discussion. It is likely that the answer we would most often receive to the question posed in example (49) is in the drawer. But the locative alone cannot confirm the conjectured existence of the thing; it can do so only together with the verb būti (let us recall that if a thing IS SOMEWHERE, then it EXISTS AT ALL). Consequently, although in sentence (49) the predicate yra belongs 40 Del sintaksinio dviprasmiškumo to the topic, together with the locative in the rheme, stalčiuje it forms a single predicate. In general, we could render this predicate as be in a place. Although formally such a predicate resembles the typical predicate of locative sentences (for a fuller discussion, see Holvoet, Čižik-Prokaševa 2005: 74–78), because of its additional function, which becomes apparent at the discourse level—confirming the conjectured existence of the referent—it should be regarded as existential. In making this claim, we could in fact endorse the possible alternative interpretation of sentence (= 8) offered by Holvoet (2005: 142, note 2), according to which its predicate expresses both existence and location syncretically. In other words, the verb būti in this sentence is simultaneously existential and locative. The type of sentences such as (49) and (=8) could then be described as intermediate between existential and locative. But we would then also have to acknowledge that marked information structure can at least seriously “undermine” the foundations of a sentence’s typological classification. In any case, if, on grounds that are more functional than formal, we nevertheless assign such sentences to the existential type, we will hardly be able—without doing violence to the linguistic facts—to call the locatives that are obligatory in them peripheral (or optional) sentence constituents (cf. Holvoet 2005: 146). 4.4. Thus, however much they may vary in terms of information structure, existential sentences remain existential, and locative sentences remain locative. Marked information structure alone cannot change their typological affiliation. We shall see below whether this claim can be applied more broadly. 5. Finally, we can turn to the analysis of sentences such as (1) and (2). For the sake of simplicity, we will discuss only the first of them, presenting it as (50): (50) Onutė’s green eyes. Discussing the structural distinctiveness of sentences such as (50), Holvoet observes that the adjectival modifier cannot be omitted in them, “because sequences of words such as Onutė’s eyes, Petras’s character could probably not be regarded as sentences under any circumstances.” According to the author of the article, “<...> the adjective is necessary here because it expresses the main predicate of the sentence” (Holvoet 2005: 153). Indeed, this claim is correct only if the adjective in the sentence under discussion is understood as a predicative. This interpretation means that the subject noun akys and the possessive genitive Onutės form a discontinuous phrase, cf.: 41 ROLANDAS MIKULSKAS [Onutė’s eyes] are green. Although these two nouns are separated from each other in the phonological sequence, the possessive (body–part) relation linking them allows them to be treated as components of a single constituent. If this interpretation of sentence (50) is chosen, it should be regarded as a marked variant of the copular sentence Onutė’s eyes are green, in which the adjective should be pronounced with contrastive stress, cf.: (51) Onutė’s (those) green eyes (not blue ones). If this constituent structure is chosen for the interpretation, the possessive construction Onutė’s eyes, as in the unmarked case, is the subject of the sentence, while (are) green is the predicate. It should also be noted that the rheme of the sentence then consists solely of the adjective green, while the noun eyes, despite its postverbal (i.e. typically rhematic) position, is assigned to the sentence topic. The “creation” of such a discontinuous topic phrase is especially facilitated by the inalienable-possession relation linking its components. 5.1. The markedness of sentence (50), when interpreted as a copular construction, makes another constituent structure more acceptable, namely Onutė’s [green eyes]. For one must agree with Holvoet’s observation that the word order in this sentence “seems entirely neutral and unmarked: it is not associated with marked intonation or contrastive stress” (ibid., 154). One must also agree with the other arguments the article’s author presents in support of this choice of sentence segmentation: first, the bracketed words are adjacent; second, when the verb būti ‘be’ is not omitted, it clearly separates these words from the noun at the beginning of the sentence, cf.: (52) Onutė’s eyes were green. Nevertheless, we should not assume that, if this constituent structure is chosen for the interpretation of sentence (50), the possessive relation between the referents of the words Onutė’s and eyes is severed: the relation linking a body and its part is always presupposed, but in this case it is overshadowed by another conceptual link. 5.2. The distribution of grammatical functions in a sentence with this constituent structure (green eyes—subject, Onutė’s—obligatory modifier) differs in no way from the syntactic distribution of the well-attested possessive-ownership sentence type in the language, cf.: (53) This scarf is Onutė’s. (= 37) 42 Dėl sintaksinio dviprasmiškumo Since the word order in this sentence model is the reverse of that in the one under discussion, the possibility should be considered of interpreting sentence (50) as a marked variant of the corresponding possession sentence. But then we would have to disagree with Holvoet's claim that the sentence Green eyes belong to Onutė (= 38) is impossible in the language. In fact, under special circumstances, for example, while doing puzzles or considering which of the relatives a child inherited their eyes from, such a sentence could sound quite natural. The author of the article also acknowledges this, while distancing himself from such an interpretation (ibid., 155, footnote 10). The circumstances mentioned clearly show that the marginality of the possession interpretation of the sentence under discussion is due to cognitive factors. In prototypical possessive possession sentences, the possessed entity is treated as an object physically separate from the possessor. In the case of inalienable possession (prototypically, the relationship between a body and its parts), asserting a relation of possession would seem redundant and illogical. Therefore, only if we treat eyes as a relatively independent object can we regard sentence (50) as a marked variant of the corresponding possessive possession sentence. Given the limited possibilities for such a treatment, it is also unlikely that the sentence under discussion could be interpreted as a predication expressing a relation of possession; cf.: (53) This kerchief belongs to Onutė. (54) This kerchief is Onutė’s, (not Marytė’s). ir (55) ?Green eyes belong to Onutė. (56) ?Onutė’s are green eyes, (not Marytė’s). 5.3. Since the proper noun Onutė in sentence (50) is not marked by any special intonation or contrastive stress, it could still be regarded as the detached topic of the previously mentioned copular sentence20. Let us recall that moving a topic to the first position in a sentence indicates which sentence element’s presupposition is in doubt (Holvoet 2003: 94–95). We could formulate such a doubt as a corresponding question, cf.: 20 The simplest variant of topicalization from the copular construction Onutė’s eyes (are) green would be Onutė, her eyes are green. The segmental marker tai is necessary here, since without it we would not recognize the topicalization of (Onutė), given that even in the unmarked sentence this proper noun occupies the first position. Since topicalization in such a sentence practically amounts to moving the possessor out of the possessive construction, the same effect can also be achieved by placing the possessed entity at the end of the sentence. 43 ROLANDAS MIKULSKAS (57) Do you remember what Onutė looks like? Hers (= Onutė’s) (tai) are green eyes, brown hair, beautiful hands, and so on. Apparently, such topicalization does not disrupt the syntactic structure of the copular sentence because its possessive construction Onutė’s eyes remains readily comprehensible despite the phonological distance separating its components, owing to the aforementioned inalienable-possession relation connecting their denotata. For this reason, the discontinuous phrase Onutė’s <...> eyes is easily interpreted as the subject of the sentence. However, it seems that for reasons of information structure the possessor can also be moved out of other types of possessive construction; cf.: (58) And what sort of land did the parents have? The parents (tai) had fertile land. 5.4. If our earlier conclusion is correct—that marked information structure cannot change the semantic sentence type—then we should continue to regard sentences with a detached topic such as (57) and (58) as copular constructions. To treat sentence (50) differently, we would have to provide strong semantic arguments attesting to its structural separation from the source sentence. From a diachronic perspective, this would mean that there is reason to speak of a reanalysis of sentence (50). The latter would first of all mean that we can no longer regard the discontinuous phrase Onutė’s <...> eyes as the subject of this sentence. Semantically, this would mean that the possessive relation connecting the components of this phrase is relegated to the background by the new predication. We could define this new predication on the basis of Holvoet’s correct observation that the purpose of such sentences is to “describe a person in terms of a certain characteristic attribute” (Holvoet 2005: 155). By noting that sentences serving this purpose are common in the language, this observation also allows us to speak of their relative functional autonomy and classify them as a distinct semantic type. We should now examine how far these functional considerations are supported by a possible new syntactic analysis of these sentences. Clearly, if sentence (50) is classified as a distinct type, its proper noun Onutė loses its status as a detached topic, although it no longer remains the sentence topic either. The main predication of the sentence is expressed by the phrase forming the sentence frame (has green eyes (if the sentence frame consisted only of the property of the eyes (green), as Holvoet argues (ibid. 158), then we would have to regard sentence (50) as a copular construction with a detached topic). In other words, the sentence topic is assigned the phrase that characterizes it 44 Dėl sintaksinio dviprasmiškumo the attribute is ‘green eyes’. In seeking a more precise characterization of the predication expressed by sentences serving this purpose, we could call it THE ASSIGNMENT OF A CHARACTERISTIC ATTRIBUTE TO A PERSON. The verbal component of such predication, būti, would then correspondingly have the meaning of ‘attribution’ or ‘assignment’. This name for the meaning of the verb būti, derived from the construction, is no different from the names for its meanings of ‘existence’ or ‘location’ in the corresponding constructions. We could formally represent the construction of the sentence type under discussion as follows: [NGEN[VBŪTI[ADJ N]NOM]] Thus, an adequate syntactic interpretation of the predication expressed by sentence (50) would be as follows: the topic Onutės is the subject of the sentence, while (yra) žalios akys ‘(are) green eyes’ is the sentence’s complex predicate, consisting of the verb būti ‘to be’ with the meaning of assignment and the so-called predicate nominative phrase žalios akys ‘green eyes’, consisting of a head noun and its adjectival modifier. It should be emphasized that if, in syntactically interpreting sentences of this type, we consistently rely on their functional purpose attested in usage, then we must inevitably regard the genitive noun as their subject, despite its failing to meet the formal criterion (the nominative case). To mitigate this formal inadequacy, we might try to interpret the genitive as a partitive, cf.: (59) Vyšnių užderėjo ir šįmet. (60) Parodos lankytojų buvo gausu ir švenčių dienomis. However, it must be acknowledged that the partitive meaning of the genitive of non-mass nouns in subject position is difficult to grasp; an appropriate supplement is needed, cf.: (61) Didžioji dalis laivo korpuso paniro po vandeniu. (62) Išskirtinis Onutės bruožas yra (jos) žalios akys. It is interesting to note that, when we compare the sentence under discussion (50) with sentence (62), we could assign the former to the type of equative sentences, that is, sentences in which the identity of the referents of two noun phrases is asserted. But then we would also have to assign the meaning of ‘identity’ to the verb būti in the sentence, and, following Holvoet and Čižik-Prokaševa (2005: 79), we would have to regard the noun phrase žalios akys as an argument of such an identity predicate or- 45 Rolandas Mikulskas argument21. Incidentally, it should be recalled that these authors also propose assigning the meaning of ‘identity’ to the verb of the copular construction (ibid., 80). 5.4.1. These formal difficulties of analysis lead us to regard the encoding of syntactic functions in the sentence under discussion as inadequate to the meaning it conveys. In this respect, the following syntactic structure would be adequate for such a sentence: (63) Onutė yra žalių akių (mergina). It is easy to see that the purpose of sentences of this type is to assign the referent of the subject noun to a broader class of referents; therefore, in the literature they are also called ascriptive sentences. Holvoet and Čizik-Prokaševa propose interpreting their syntax in much the same way as that of equative sentences (ibid., 80). Thus, we would have to assign the meaning of ‘identity’ to the verb būti in sentence (63), and regard the postverbal phrase žalių akių (mergina) ‘green-eyed (girl)’ as a complement of this identity predicate, despite the fact that it actually expresses the main predication of the sentence (the predicative nature of this phrase is especially evident when its head—mergina ‘girl’ or some other noun—remains implicit). 5.4.2. If we continue to maintain that the purpose of sentences such as (50) is to characterize a person by an attribute characteristic of them, then sentence (63) conveys approximately the meaning of sentence (50). Since nothing can formally be objected to regarding the subject of sentence (63), Onutė (it is prototypically marked by the nominative), we could reasonably suppose that the emergence of sentences of this type is motivated by the inadequacy of grammatical-function encoding in sentences of the type under discussion: in sentence (50), the subject noun, which prototypically has topic status, is marked by a non-prototypical case for a subject—the genitive; in sentence (63), the situation is ‘levelled out’—the subject noun has acquired the nominative case characteristic of it. 5.5. We should consider separately the possibility of treating the indefinite noun phrase žalios akys ‘green eyes’, which occurs in the rheme position of sentence (50), as the sentence subject. If this phrase is interpreted as definite, it has topic status and is unambiguously understood as the subject of the sentence. In that case, however, we have a marked variant of the possessive-existence sentence—a possibility of this interpretation under specific circumstances has already been discussed. But the existential sentences discussed earlier (Stalčiuje yra sąsiuvinis ‘There is a notebook in the drawer’) and typologically related to them 21 For the possibility of interpreting such a noun phrase as a predicate nominative (predicative), see Langacker (1991: 67–68). 46 Del sintaksinio dviprasmiškumo Closely related possessive sentences (Vaikui yra sloga) show that even in postverbal position, a noun phrase that retains its rhematic status (and is therefore prototypically indefinite) becomes the subject of the sentence. What possibilities are there for understanding the sentence under consideration (50) as existential or possessive? Is the similarity in information structure alone sufficient for such an interpretation? 5.5.1. For us to understand sentence (50) as an existential proposition, the referent of the description žalios akys should be understood “substantially.” In other words, attributes describing a person—“green eyes,” “light-brown hair,” “beautiful hands,” and so on—would have to be granted the status of a relatively independent existence, which can be affirmed or denied. The referent of the description žalios akys has such a status in this sentence: (64) Among people, green eyes are nothing unusual. In sentences denying the existence of such an attribute, the noun phrase naming it regularly takes the genitive case, cf.: (65) Green eyes are not found (do not exist) among the inhabitants of Ghana. Examples (64) and (65) clearly show that asserting the existence of an object is difficult to conceive of without its location. It would be difficult to assign a locative meaning to the genitive Onutės in sentence (50). Perhaps that is why an existential interpretation of this sentence is hardly possible. This is made even clearer by the fact, also noted by Holvoet (2005: 156), that in the corresponding sentence with a negative predicate, the noun phrase žalios akys does not take the genitive case, cf.: (66) Onutė did not have green eyes. (67) *Onutė did not have green eyes (genitive). 5.5.2. Having rejected the existential interpretation of sentence (50) and set aside the meaning of belonging, one possibility remains without changing the constituent structure: to treat the postverbal noun phrase žalios akys as the subject—that is, to interpret the sentence as a possessive possession construction. The corresponding noun phrase, in which the possessive relation holds between a person and an attribute characteristic of them (not a body part!), is easy to imagine, cf.: (68) Andrius was captivated by ONUTĖ’S GREEN EYES. 47 ROLANDAS MIKULSKAS Interpreted as a possessive-ownership construction, sentence (50) would then recall a similar construction with a dative, which in Lithuanian is merely a relic (Holvoet 2005: 150), cf.: (69) The child has a runny nose. (= 23) (70) ?Onutė has green eyes. The mihi est type of construction is still used in Lithuanian to express a relation of possession in cases where the possessor can also be understood as the experiencer of a certain state. In principle, sentence (70) is also possible, but it sounds archaic, although a similar construction—Bijūnėliui žali lapai (Šukys 1998: 162)—is attested in dialects. Citing this sentence in his article, Holvoet (ibid., 159) conjectures that the dative in this construction is older than the genitive. But if the claim made earlier is correct—that the existential sentence gave rise to the mihi est type of possessive construction—then, at least in terms of its origin, it would be inappropriate to compare it with a construction in which the possessor is expressed by a noun in the genitive22. For the origin of the latter construction is associated with topicalization and, according to Holvoet (ibid., 159), also with the raising of the possessor from the corresponding noun phrase (to the level of sentence arguments). Given this origin, the negative-genitive test associated with existential predication is also inapplicable to the construction in question. 5.5.3. Certain specific contexts help in interpreting the predication expressed by sentence (50) as a relation of possession between a person and an attribute characteristic of that person 22In principle, the noun genitive in such sentences of the type under discussion, which have been reinterpreted “possessively,” could be replaced by a dative, especially in cases where a subject of the possessive relation (the possessor) marked by the genitive is also understood as the beneficiary of that relation. We could even argue that these sentences, having acquired a possessive meaning, reflect an intermediate stage in the so-called raising of the possessor to the level of sentence arguments. It is easy to see that replacing the possessor’s genitive in such sentences with a beneficiary dative would in effect make them identical to the mihi est type of possessive construction. One may conjecture that such a replacement is now difficult to imagine (or, more accurately, blocked) only because the dative possessive-ownership construction has been almost displaced in contemporary Lithuanian by the newer habeo type of construction. It is nevertheless likely that in the past, for semantic reasons, some sentences of the type under discussion that had been reinterpreted “possessively,” with their genitive replaced by a dative, coincided with possessive mihi est type sentences that had arisen from existential sentences. Further diachronic research would be needed to confirm this hypothesis. 48 Dėl sintaksinio dviprasmiškumo contexts in which those personal attributes become “in demand,” and having them may in one way or another be useful to the person, cf.: (71) Onutė has shapely legs and a slender waist—she could enter a beauty contest. In similar circumstances, we could confidently replace sentence (71) with a possessive construction using a possession predicate, cf.: (72) Onutė has shapely legs and a slender waist—she could enter a beauty contest. 5.6. Summarizing all this, we can assert that the basis for the various possible interpretations of sentence (50) is provided by the models of two sentence types: a copular construction and a possessive-ownership construction (typologically close to locative sentences). In both cases, the sentence under discussion is marked in terms of information structure. We can understand sentence (50) as a marked variant of the possessive-ownership sentence with inverted word order Green eyes belong to Onutė only under special circumstances that allow the object of possession (the eyes) to be regarded as relatively autonomous. When the basis for interpreting sentence (50) is the copular construction Onutė’s eyes are green, it may be marked in two ways relative to the latter. Since the author of the article discussed emphasized that sentence (50) is neutral with respect to intonation and contrastive stress, it is most appropriate to regard it as a copular construction with a fronted topic. However, the fact that the word order of sentences like (50) also appears neutral cautions against placing too much emphasis on their underlying structure and suggests seeking a motivation other than topicalization to account for that order. 5.6.1. If we regard the word order of sentence (50) as unmarked, it could be justified only by a constituent structure in which the initial noun genitive (Onutės) is no longer considered part of a possessive noun phrase. Analysis of the possible interpretations showed that the likelihood of this sentence structure is linked, on the one hand, to the possibility of conceptualizing the sentence as an equative or ascriptive construction and, on the other, as a possessive-ownership construction. The alternative ways of conceptualizing the sentence under discussion also entail alternative distributions of grammatical functions within it. 49 ROLANDAS MIKULSKAS 5.6.2. If it is rightly emphasized that the primary function of sentences such as (50) is to describe a person in terms of an attribute characteristic of them, then their equative or ascriptive interpretation should be regarded as more acceptable. Functional outlines of such an interpretation attested in the language in fact make it possible to speak of a distinct semantic sentence type. However, given a favorable context, the possibility of a possessive interpretation of such sentences cannot be ruled out either. Yet it is difficult to overlook the formal inadequacy of the sentence under consideration with respect to both of the aforementioned interpretations. In the first case, the noun in the genitive (Onutės) has to be treated as the subject of the sentence; in the second case, it has to be treated as an argument of the possession predicate. Coding both the subject and the possessor in a possessive construction with the genitive is not characteristic of an active sentence in Lithuanian. 5.6.3. Holvoet also discusses almost all the aforementioned possibilities for interpreting sentence (50) in his article, but in seeking its single correct interpretation he arrives at a methodologically incorrect conclusion, namely, that the syntactic structure of sentences of this type is ambiguous. It would seem more accurate to suppose that, once one or another direction for conceptualizing these sentences is chosen, their syntactic structure, too, becomes unambiguous. Moreover, if there is any basis at all for speaking of a distinct type of sentence under discussion, it could be called hybrid not because “the whole construction could be described as intermediate between possessive and belonging constructions” (ibid., 158), but because the inherited form of these sentences does not correspond to new trends in their conceptualization. LITERATŪRA BOGUSŁAWSKI, A. 1977: Problems of the Thematic-Rhematic Structure of Sentences. Warszawa: Państwowe Wydawnictwo Naukowe. DOWNING, P., NOONAN, M., eds., 1995: Word Order in Discourse. Amsterdam: John Benjamins. GIVÓN, T. 1984: Syntax: A Functional-Typological Introduction. Vol. 1. Amsterdam: John Benjamins. GOLDBERG, A. E., ed., 1996: Conceptual Structure, Discourse and Language. Stanford, CA: Center for the Study of Language and Information. 50 Del sintaksinio dviprasmiškumo Forrest, L.B. 1996: Discourse Goals and Attentional Process in Sentence Production: The Dynamic Construal of Events. In: Goldberg, ed., 1996, 149–161. Holvoet, A. 2003: Aktualiosios skaidós problemos. In: Holvoet & Judžentis, red., 2003, 79–98. Holvoet, A. 2005: Intranzityvinių sakinių tipai: egzistenciniai, lokatyviniai ir posesyviniai sakiniai. In: Holvoet & Mikulskas, red., 2005, 139–160. Holvoet, A., Čižik-Prokaševa, V. 2005: Papildiniai ir aplinkybės. In: Holvoet & Mikulskas, red., 2005, 65–92. Holvoet, A., Judžentis, A., red., 2003: Sintaksinių ryšių tyrimai. Vilnius: Lietuvių kalbos institutas. (Lietuvių kalbos gramatikos darbai, 1.) Holvoet, A., Mikulskas, R., red., 2005: Gramatinių funkcijų tyrimai. Vilnius: Lietuvių kalbos institutas. (Lietuvių kalbos gramatikos darbai, 3.) Holvoet, A., Mikulskas, R., red., 2006: Daiktavardinio junginio tyrimai. Vilnius: Lietuvių kalbos institutas. (Lietuvių kalbos gramatikos darbai, 4.) Holvoet, A., Tamulioniene, A. 2006: Apibrėžtumo kategorija. In: Holvoet & Mikulskas, red., 2006, 11–32. Hudson, R. A. 1984: Word Grammar. Oxford: Basil Blackwell. Jackendoff, R. 1977: X-Bar Syntax: A Study of Phrase Structure. Cambridge, MA / London: MIT Press. Lakoff, G. 1969: On Derivational Constraints. Papers from the Fifth Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 117–139. Lakoff, G. 1970: Repartee, or a reply to “Negation, conjunction and quantifiers”. Foundations of Language 6, 389–422. Lakoff, G. 1987: Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago / London: University of Chicago Press. Langacker, R. W. 1991: Foundations of Cognitive Grammar. Vol. 2, Descriptive Application. Stanford, CA: Stanford University Press. 51 Rolandas Mikulskas Langacker, R. W. 2000: Grammar and Conceptualization. Berlin / New York: Mouton de Gruyter. Langacker, R. W. 2001: Topic, Subject, and Possessor. In: Simonsen & Endresen, eds., 2001, 11–48. Langacker, R. W. 2002: Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin / New York: Mouton de Gruyter. Li, Ch. N., ed., 1976: Subject and Topic. New York: Academic Press. Li, Ch. N., Thomson, S. A. 1976: Subject and Topic: A New Typology of Language. In: Li, ed., 1976, 457–489. Lyons, Chr. 1999: Definiteness. Cambridge: Cambridge University Press. Keenan, E. L., Comrie, B. 1977: Noun phrase accessibility and Universal Grammar. Linguistic Inquiry 8 (1), 63–99. Mikuls kas, R. 2006: Apibrėžiamųjų būdvardžių aprašo perspektyva. In: Holvoet & Mikulskas, red., 2006, 33–65. Nuyts, J., Pederson, E., eds., 1997: Language and Conceptualization. Cambridge; Cambridge University Press. Simonsen, H. G., Endresen, R. Th., eds., 2001: A Cognitive Approach to the Verb: Morphological and Constructional Perspectives. Berlin / New York: Mouton de Gruyter. Šukys, J. 1998: Lietuvių kalbos linksniai ir prielinksniai: vartosena ir norma. Kaunas: Šviesa. Taylor, J. R. 2002: Cognitive Grammar. Oxford: Oxford University Press. Tomlin, R. S. 1995: Focal Attention, Voice, and Word Order. In: Downing & Noonan, eds., 1995, 517–554. Tomlin, R. S. 1997: Mapping Conceptual Representations into Linguistic Representations: The Role of Attention in Grammar. In: Nuyts & Pederson, eds., 1997, 162–189. 52 Del sintaksinio dviprasmiškumo VAN HOEK, K. 1995: Conceptual Reference Points: A Cognitive Grammar Account of Pronominal Anaphora Constraints. Language 7 (2), 310–340. VAN HOEK, K. 1997: Anaphora and Conceptual Structure, Chicago: University of Chicago Press. Rolandas Mikulskas Lietuvių kalbos institutas P. Vileišio 5, 10308 Vilnius, Lietuva [email protected] 53