The aim of this paper is to investigate ‘eînai (‘be’) plus dative’ and ‘eînai plus genitive’ possessive constructions, paying special attention to the semantic content of the verb eînai in order to identify the function and the distribution of the various combinatorial patterns of the constructions in question, and the precise role of the verbal items. In particular, the present analysis, carried out within the framework of Construction Grammar, will attempt to demonstrate that each possessive variant constitutes a semantically and pragmatically distinct pattern where the semantic content of the verb eînai is the result of form-meaning configurations over and above the morpheme and word level. From this perspective, the cluster of semantic, pragmatic and morpho-syntactic values attributed to participant slots constitutes an integral part of constructions.
The predicative encoding of possession in Ancient Greek is characterized by the grammatically important distinction between two major construction types: constructions with verbs whose semantic content has a specific possessive predicational function, e.g. ékhō ‘have’,1 and constructions with a form of the verb ‘to be’, generally eînai, in combination with an accompanying oblique case. The latter denotes the Possessor, while the Possessee NP is constructed as the grammatical subject in the nominative case controlling verb agreement. From the earliest stages on, the second construction type is realized by two distinct schemata, the Possessor NP being expressed either by the genitive or the dative.
In both genitive and dative constructions the verbal lexical item is the same (eînai), the constituent order is apparently not fixed (as in 1a–1b, and 2a–2b) and thus any position of the three constituents involved with respect to each other is acceptable. At first glance, the two constructions only seem to differ in the case form carried by the Possessor item, genitive (as in 1) vs. dative (as in 2) below.2
On closer inspection, however, it transpires that the expression of predicative possession involves two coherent, distinct constructions which include both nominal and verbal items. Nevertheless, a strictly form-based analysis of syntactic relations is not in itself sufficient to account for the precise nature of the differences between these two constructions and the relationships between them. Since both types of sentences apparently have the argument structure of the verb ‘to be’, which licenses alternative (genitive or dative) argument realisations of Possessor, two distinct research questions emerge: what is the semantic contribution of the verbal item, and is it the same in both constructions?
Indeed, as is well known, the verb eînai in Ancient Greek not only covers the expression of properties or quality ascribed to subjects, but also the semantic space of three fundamental relations: POSSESSION (see 1 and 2 above), LOCATION (3) and EXISTENCE (4).
With this in mind, an explanation of the role of the verbal items in the possessive constructions under discussion needs to be sought in the interaction between the syntactic configurations and the semantic and discourse-based linguistic features relating to the constituent parts.
Despite the fact that the expressions of predicative possession in Ancient Greek have been thoroughly investigated in a number of studies (for an overview of the range of encodings of possession in Ancient Greek see Benvenuto 2014), little attention has been given to the verbal valence and verbal lexical item role regarding constructions competing in the possessive semantic space; what is more, pragmatic aspects have received relatively scant attention.
Consequently, the aim of this paper is to investigate the two possessive constructions, ‘eînai plus dative’ and ‘eînai plus genitive’, in order to understand the interaction between the verb and the possessive constructions with which it is associated. In particular, our analysis, pursued within the framework of Construction Grammar, will attempt to demonstrate that each possessive variant constitutes a semantically and pragmatically distinct pattern where the semantic content of the verb eînai is a result of form-meaning configurations over and above the morpheme and word level. From this perspective, the cluster of semantic, pragmatic and morpho-syntactic values attributed to participant slots constitutes an integral part of constructions.
The paper is organized as follows. After the introduction to our study in section 1, some preliminary remarks are provided in section 2 alongside a presentation of the theoretical framework. Sections 3 and 4 examine argument structure constructions involving the verb eînai in possessive predication with dative and genitive respectively, taking into account various types of criteria and properties. Section 5 then concludes.
2. Preliminary Remarks
The concept of ‘possession’ is an exceptionally complex notion which is difficult to define in a uniform way. There are various linguistic renderings of it, not only cross-linguistically, but also within individual languages.3 For the purposes of this paper, we assume a general characterization of possession as a semantic concept associated with cognitive entities known as ‘relations’, as was noted, for the first time, it seems, by Aristotle (Cat. 7, 6 b34–35).
From a cognitive perspective a prototypical relation of possession comprises two entities: the ‘Possessor’ (henceforth Pr), i.e. the entity, prototypically human and highly individuated, that has something at its disposal, and the ‘Possessee’ (Pe), that is to say, the entity actually, or metaphorically, at the Pr’s disposal and which usually denotes a concrete, not necessarily highly individuated, entity.4 The nature of the relation between the two entities is essentially asymmetrical since their role or status involves different specifications in terms of empathy (see Lehmann 2002), control and/or agentivity as well as pragmatic-cognitive salience. Various subtypes of possession can be characterized on the basis of the different combinations of these parameters and in terms of the different values they assume in relation to factors such as the duration of the possession (temporary or permanent), the physical proximity of the Pe to the Pr and the alienability/inalienability of the Pe. Many languages, for instance, distinguish between inalienable (or inherent) possession and alienable (or established) possession (Seiler 1983; Heine 1997: 17–25). Even though this distinction (and related linguistic encodings) can ultimately be defined on the basis of culture specific knowledge, there are entity concepts, regarding, for example, kinship, body parts, or organic part/whole, which are intrinsically relational at a universal cognitive level. Because of this property, it can be said that a possessive relationship is inherently given in such concepts and consequently determines a lexical interpretation of the possessive relationship.5
The conventionalized encodings of the concept of possession are language specific and differ according to which syntactic and/or informational structures are ascribed to the two participants in the construction. From a cross-linguistic perspective, the possessive constructions differ particularly in how they represent the participants of a possessive situation in the morphosyntactic and/or informational structure. When one of the participants is part of the predicative statement and the possessive relation is the main assertion of the sentence, we are dealing with predicative possession; the predicate can be a lexical verb which specifies the possessive semantic function or a copular/existential verb, while the Pr and Pe fill the argument slots of the predicate.6
2.2. Predicative Possession in Ancient Greek: Previous Studies
The expression of the possession relation in Ancient Greek has been investigated in previous studies which have focused on various aspects of the issue (Benveniste 1960; Kahn 1973; Kulneff-Eriksson 1999; Bauer 2000; Guilleux & Bortolussi 2011). In Benveniste’s seminal work (1960), predicative constructions with dative and with genitive are defined as “prédicat de possession”, expressing “possession”, and “prédicat d’appartenance” expressing “belonging” respectively. Kulneff-Eriksson’s study (1999) is also worthy of mention; here occurrences of the dative construction are largely analysed contrastively with the ékhō construction in order to identify any differences between the two modes of expressing the notion of ‘have’ in Ancient Greek; the genitive construction, however, is not considered. Finally, Guilleux & Bortolussi’s paper (2011), based on a limited corpus (Mycenaean Greek, 5th and 4th century B.C. inscriptions from Attica and a selection of Attic literary texts), uses an essentially typological approach for a brief examination of certain functions, including expressions of possession, in which eînai and ékhein apparently compete.7
All in all, traditional literature on the issue does not provide a satisfactory account of the verbal lexeme role in the organization of argument realization. Furthermore, research focusing exclusively on the verb ‘to be’ in Ancient Greek possessive constructions is limited to the seminal essay by Kahn (1973). Even though the analysis is based on a detailed and exhaustive study of the various uses of eînai, Kahn maintains that the only satisfactory general classification of such uses is their formal division into copulative and non-copulative constructions. He does not, however, put forward consistent semantic and discourse-pragmatic parameters when discussing existential uses (Kahn 1973: 228–229). Finally,—as mentioned above—it is worth noting that none of the previous studies provides a detailed analysis of the pragmatic aspects of predicative possessive constructions.
The present paper primarily examines the ‘eînai plus genitive/dative’ constructions, by paying special attention to the semantic content of the verb. As such, this study is a continuation of the research which began with Benvenuto & Pompeo (2012). The earlier paper, based on a quantitative analysis of corpus data, consists of an evaluation of the distribution of the constructions and accounts for their different functions: the genitive possessive construction has a precise functional area, is contextually restricted and semantically specialized, while the semantic-functional profile of the dative construction is relatively loose. These constructions are identified by different configurations of fundamental syntactic relations (in the case of Pr) and different discourse-related features of constituents that allow us to define the constructions as ‘have-’ and a ‘belong- constructions’ respectively.8
What still remains to be explored more thoroughly, however, is the precise role of the verbal items in the constructions. The two different configurations of these constructions depend on the occurrence of the lexeme eînai which has both a copular and an existential value. Indeed, as is well known, the Indo-European verbal root *h1es- ‘to be’ is vague in meaning. As a matter of fact, there are two major readings/functions which can be determined by the immediately surrounding discourse context and by the type of clause in which the verb is used: a grammatical reading, as a linking verb (or copula) and a lexical one, as an abstract verb of existence (see Benveniste 1960). The distinction between these two patterns of use has not been lexicalized in the majority of Indo-European languages.
2.3. Theoretical Framework
Given such conceptual and linguistic complexity, the present study adopts a constructional approach in line with the work of Adele Goldberg (1995, 2006) and William Croft (2001) as well as drawing on the fundamental principles of Cognitive Grammar (Ronald Langacker 1987, 1991).9 Constructional and cognitive grammar approaches are similar in that constructions are considered as symbolic units equipped with conventionalized form-meaning configurations at varying levels of complexity and abstraction. Consequently, the construction itself “contribute[s] semantics not attributable to the lexical items involved” (Goldberg 1995: 141; see also Croft 2001). In particular, the constructional approach assumes that constructions can influence the valence patterns of verbs, over and above the verb’s lexical value.10 Consider for instance Goldberg’s (1995: 9) example, He sneezed the napkin off the table, where the verb sneeze, usually associated with a single participant argument role (the sneezer), receives a transitive reading when it is integrated in the caused motion construction. The constructional approach can thus account for the variable behaviour of verbs, and their multiple argument realization can be explained by verb occurrence in multiple constructions. In other words, it is not necessary to stipulate a specific sense of sneeze which is unique to the caused motion construction. Schematic argument structure constructions are capable of contributing argument.
A further assumption of Construction Grammar is that constructions are organized in taxonomic networks (Croft 2001: 25–26), ranging from substantive constructions (idiomatic) to highly abstract ‘schematic constructions’ i.e. sentence structure templates. Moreover, “several constructions can be shown to be associated with a family of distinct but related senses, much like the polysemy recognized in lexical items” (Goldberg 1995: 4).
The constructional approach is also concerned with accounting for the pragmatic factors that are crucial to understanding the constraints on grammatical constructions. Since semantics, information structure and pragmatics are closely interrelated, we will focus our attention on issues regarding information packaging in line with Lambrecht (1994, 2000). Two notions that play a central role in the packaging of information structure are Topic and Focus, Topic being a “matter of current interest which a statement is about and with respect to which a proposition is to be interpreted as relevant” (Lambrecht 1994: 119), while Focus is an “element of a pragmatically structured proposition whose occurrence makes it possible for the sentence to express a ‘pragmatic assertion’” (Lambrecht 2000: 612).
Furthermore, given the essentially asymmetrical nature of the possessive relation (see previous section), Langacker’s reference-point model (Langacker 1991, 1995, 2003, 2009, among others) merits attention. According to Langacker, the reference-point model is an idealized cognitive model, representing “simply the idea that we commonly invoke the conception of one entity for purposes of establishing mental contact with another [the target]” (Langacker 1995: 58), the entity invoked being the most salient, and constituting a sort of natural reference point. In particular, in possessive expressions, the Pr is usually the reference point which affords mental access to the Pe (the target). Consequently, the various possessive constructions will differ in which element(s) they render prominent through profiling and trajector/landmark assignment (see, among others, Langacker 1987, 1991, 1995, 2003, 2009).
Profiling and trajector/landmark organization (or alignment) are two basic notions in Langacker’s Cognitive Grammar. According to Langacker, profiling is “one kind of prominence: within the overall conception it evokes as its base, an expression profiles some substructure, i.e. puts it in focus as the entity it designates (refers to). An expression’s grammatical category is determined by the nature of its profile” (Langacker 2009: 111). On the other hand, trajector and landmark are the profiled participants in a relational predication: a trajector is the primary focal participant, “the figure in a relational profile”, while a landmark is a secondary focal participant, which functions as a ground to the trajector (Langacker 1987: 217). Finally, “the figure within a scene is a substructure perceived as “standing out” from the remainder (the ground) and accorded special prominence as the pivotal entity around which the scene is organized and for which it provides setting” (Langacker 1987: 120). It is important to note that in certain special circumstances the less salient entity may constitute the reference point (Langacker 1995: 59), as we will see in section 4.2.
Finally, as far as the cognitive-semantic level is concerned, the following aspects will be taken into consideration in order to identify the function of the constructions under discussion:
the semantic characterisation of the Pr and the Pe, identifying the possible combinations of the features [±animate], [±human], [±definite]11 and, only in the case of the Pe, [±concrete] and [±relational]12
the variables determining the diverse feature distributions, that is to say, the constraints that relate to the various distributions, and, in particular, how these distributions are associated with the expression of various relationships of possession in the two constructions analysed.
3. The Dative Pattern: Syntactic, Pragmatic and Semantic Features
3.1. Phrasal Syntax and Information Structure
In dative possessive constructions the Pe NP is the grammatical subject in the nominative case controlling verb agreement; Pr’s encoded by the dative are generally previously mentioned referents and thus can be considered given.
This feature is consistent with the high presence of pronominal expressions of the Pr, a phenomenon borne out by the data provided in the literature (Kulneff-Eriksson 1999; Benvenuto & Pompeo 2012). Pronouns, indeed, nearly always indicate that their referents are fully discourse-activated and presupposed, and therefore offer an indication of topichood.
A useful indicator of the topic function of the dative Pr is provided in the following example:
The topic referent in (5) is expressed by the anaphoric pronoun which identifies the Pr as a co-referent of the contextual antecedent ‘Indians’.
However, the clearest illustration of the essentially pragmatic functions of the dative Pr is provided by wh-questions, as in example (6) from Anabasis.
It is clear from this example that the focus element of the question is an interrogative pronoun demanding information about the Pe referent.
An important result of these observations is the confirmation that the Pr in dative constructions is a previously mentioned referent, while the Pe is a newly introduced element providing particularly salient new information. In other words, the Pe, the subject of the sentence, conveys the assertion in the dative construction and is the unmarked focus argument. Consequently, the syntactic and pragmatic structure is [Pr DAT ] TOP [to be Pe NOM ] FD.
This observation sheds light on another important aspect of our interpretation. By examining the features of possessive clauses with ‘eînai plus dative’, it is possible to state that they are presentative-existential clauses of the kind ‘there is a Y Pe to X Pr ’/‘to X Pr exists a Y Pe ’. Indeed, focused subjects are typical of existential-presentative clauses (Givón 2001: 255–261).
As far as the predicate is concerned, the verb eînai not only introduces an entity into the discourse, but also asserts the existence of the Pe explicitly, as do existential sentences proper, with respect to a Pr (see section 3.2). The latter is an entity that specifies the area of validity of the statement of existence as pertaining to the personal sphere of Pr (Lehmann 2002: 5).13 This assertion can be supported by a negation test, as in (7).
It is intuitively clear that ékhthra in (7) is a focus rather than topic expression since it is completely unpredictable and makes the assertion informative. The Pr’s, on the contrary, are both active in the discourse and are present in the presupposition. Moreover, the subject is in the scope of negation and this is a diagnostic test to identify a focus domain.14
Thus, in example (8), the existence of a human Pe (Smerdis) is negated and his death is asserted.
The observations made so far indicate that constructions with eînai (or rather, its third-person form) plus dative are existential clauses. They are syntactically intransitive, the intransitive predicate filling the slot ([V to be]) in the schematic existential construction [NP dat V to be NP nom ]. The Pe, constructed as the grammatical subject of the predicate, is depicted as a Theme, i.e., the argument whose existence is asserted with reference to the Pr, which is an Internal Obligatory Participant (IOP) of predication and is encoded as a locative/adessive15 complement. This state of affairs is represented in Figure 1, following Croft’s figurative formalism (2001: 18).
In this template for the encoding of predicative possession, the verbal slot can be filled in various ways. As a matter of fact, few other verbs apart from eînai function in constructions expressing ‘possession’; these include e.g. hupárkhein ‘exist’ (see example 9), keîsthai ‘lie, be in a place’ and gígnomai ‘come into a certain state, become’ (cf. Kulneff-Eriksson 1999: 16–17).
Possessive constructions have the same characteristics as presentative/existential construction,16 illustrated, for example, by the occurrences in (4) quoted above and in (10) below, where the existence of a subject’s referents is asserted in the universe of discourse:
In particular, the predicative possessive sentence is very similar to the expressions that in the present paper we have decided to call expressions of ‘referential existence’, illustrated in (11):
In (11) the existential predicate introduces the referent of its primary argument into the discourse by ascribing a domain of instantiation to it. In other words, a certain entity is introduced as existing with respect to a location, which functions as a reference point, rather than as an entity endowed with the quality of being located somewhere, as in the copular locative construction in example (3) above.
Thus, a common schematic characterization as a ‘reference point relationships’ marks the particular affinity between the referential existential construction and the possessive predicative construction designating “location” in the Pr’s dominion (Langacker 2003).
To conclude, all constructions considered here are characterized by an existential intransitive verb and by an indefinite non-anaphoric referent as a subject introduced into the discourse for the first time. Moreover, when a reference point is present, this element is conceptualized as an Internal Obligatory Participant, about which the lexical predicate makes a statement. This inheritance hierarchy is illustrated in figure 2.
3.2. Semantic Features
Pr’s in dative constructions are typically [+human] and [+definite], and can be represented either by nouns or by personal pronouns, that is, by elements denoting well-defined, generally human, entities. The following examples have been selected insofar as they are representative of various types of Pr and/or Pe, and of the possessive relations relevant to our analysis. For a better understanding of the most salient aspects of the examples quoted, each one is introduced by a label (in small caps) referring to the significant semantic feature(s) and to the possessive relation, classified according to Heine’s typology (1997: 34–35).
The passage in (12) provides an example of the possessive relation that Heine (1997: 35) calls inanimate inalienable possession, where the Pr and the Pe, regarded as inseparable, are both [-animate]. This is a kind of relation which occurs very rarely in Ancient Greek due to the infrequency of [-animate] Pr’s.
The situation is more varied regarding Pe’s. Indeed, even though they are typically [-definite] and [-human], they can be both [+concrete] (e.g. khrusós, khalkós próbata, híppoi in 13) and [-concrete] nouns (e.g. kakón in 14), as well as [+human] (e.g. dmōiaí in 13, paîdes in 16 and adelpheós in 17) and [+definite]; additionally, Pe’s can be [+relational] (e.g. glôssai, stómata in 15 and adelpheós in 17) and [-relational] nouns (e.g. khrusós, khalkós in 13 and kakón in 14).
A good indicator of this scenario is provided in the following examples, starting with (13) below—with [+concrete] Pe’s—which is an instance of the possessive relation Heine calls permanent possession; this “may be said to correspond most closely to the legal notion of ownership as used in western societies” (Heine 1997: 34).
Conversely, in (14) the Pe is [-concrete], within a kind of possessive relation which—following Heine’s classification (1997: 34)—expresses abstract possession.
The following three examples express relations of inalienable possession (Heine 1997: 35), where the Pe’s are perceived as inseparable from their Pr. In these occurrences Pe’s are [+relational] nouns—body part terms (in 15) and kinship terms (in 16 and 17)—modified by numerals in the first two examples and unmodified in (17).
If we take into account the types of possessive relations that can be expressed by the dative construction, we may conclude that ‘eînai plus dative’ in Ancient Greek is used to encode a broad range of possessive relations. This variety correlates with the distribution of the semantic features of the two relata, and, in particular, with the range of characteristics as regards the Pe’s.
Among the types of relation rarely expressed by the dative construction are those of ‘inanimate inalienable possession’ (cf. example 12 above), and ‘physical’ or ‘momentary possession’, which may be interpreted as the association between a Pr and a Pe for a specified time, as in example (18) below:
Aristoph. Peace 259–261
A further important point regarding the function of the ‘eînai plus dative’ construction is that it essentially establishes the existence of a relation of possession, asserting the possessive link between two entities.
In this respect, if the wide variety of Pe’s occurring in the dative construction provides an initial hint of this construction’s predicative possessive function, a further useful diagnostic can be found in the constraints which characterize the use of [+relational] nouns. According to Herslund & Baron (2001: 13), as far as relational nouns—that is to say, inherently possessed entities—are concerned, the use of predicative possessive constructions is rather limited, while it is frequent in attributive possessive constructions (e.g. Mary’s mother). Indeed, the predication of the existence of a possessive link with this kind of noun would be redundant and uninformative due to the fact that possessive relations are inherent in relational nouns. This is the reason why we do not find unmodified affirmative sentences simply meaning ‘Y has a father/ a mother’ without there being further semantic nuances. Nevertheless, when the Pe is modified (e.g. Mary has a wise mother) or negated (e.g. Mary does not have a mother), the possessive expressions are meaningful and provide new information. In our opinion, however, it is important to observe that what Herslund & Baron noted does not apply equally to the whole class of relational nouns but largely to ascending kinship and body-part terms,17 that is to say, to nouns denoting entities that every human being must have: everyone has a (biological) father and a mother, and everyone has a body with its specific parts. Given this, we have made the decision in the present paper to call possessive relations involving these kinds of terms “necessary possessive relations” since they are relations given in nature where a Pe necessarily requires a Pr. The comparison between a sentence such as ‘Mary has a mother’, which in its basic semantics is essentially meaningless, and a sentence such as ‘Mary has a brother’, which, on the contrary, provides information, makes this distinction evident. It follows that, not infrequently, any given language can use an unmodified construction with parental terms to express a different, more specific meaning: for example, the English sentence ‘Mary has a mother’ is best interpreted as ‘Mary has a mother still living’.
Data concerning the dative possessive construction in Ancient Greek fit this scenario perfectly. As regards kinship, in fact, while sentences involving non-ascending kinship terms are possible and frequent, as in examples (16) and (17) above, an affirmative sentence involving unmodified parental kin terms means ‘Y has a father/a mother still living’, as in example (19):18
Not unexpectedly, we can find ascending kinship terms as Pe’s in negative sentences, as in (20). In this case it is evident, as observed by Kahn (1973: 267), that “Andromache means that her parents are dead”.
Furthermore, as no example was found of an ascending kinship term modified by an attribute as in a sentence such as “Andromache has a very clever mother”, we can hypothesize that in Ancient Greek the sequence ‘patḗr/mḗtēr SUBJECT eînai plus Dat PR ’ came to be used to express only the ‘being alive of the father and/or the mother’ and, therefore, cannot be used if not for this purpose.
In a similar way to ascending kinship, a possessive predication concerning parts of the body—the second type of “necessary possessive relations”—very rarely occurs and when it does it is in sentences, with modified Pe’s, that express hypothetical or unusual situations. Example (15) quoted above and (21) below—part of the description of Scylla—are illustrative to this effect.
If we consider our data according to Langacker’s reference-point model, the ‘eînai plus dative’ construction profiles the whole possessive relationship, whereby—not unexpectedly—the Pe is the target and the trajector, while the Pr is the reference point and the landmark.
4. The Genitive Pattern: Syntactic, Pragmatic and Semantic Features
4.1. Phrasal Syntax and Information Structure
In the construction ‘eînai plus genitive’, the Pe NP is construed as the grammatical subject in the nominative case controlling verb agreement. The Pe item is typically a definite element, already part of the universe of discourse, which plays the topic role in the information structure of the utterance, whereas the genitive Pr is in the focus domain. The syntactic and pragmatic structure can be schematically represented as [Pe NOM ] TOP [to be Pr GEN ] FD.
The clearest illustration of this statement is the question-and-answer pair. Here, the question word, and the corresponding word group in the response, are the Focus of their respective clauses, as in (22):
[CONTEXT: In the previous passage, Dicaeopolis had built an enclosure around his house, within which there is a peaceful, free market for the neighbouring people. The dialogue takes place between a Beothian merchant and Nicarchus, a war party informer.]
Aristoph. Ach. 910–911
What is of greatest interest in such passages is what appears early on in the questions. In the wh-questions, where the speaker knows one specific referent and the related idea ‘of belonging to’, the interrogative pronoun seeks information about the Pr referent.
On the other hand, consistent with the informational structure of this kind of sentence is the frequent omission of the Pe which characterizes genitive patterns as in (23).
As is well known, an interesting property of topics is the fact they can be omitted altogether (cf. Lambrecht 1994: 136; Matić 2003; Erteschik-Shir 2007: 23). Indeed, the omission of the Pe (Archias) in (23) is possible because Archias, the discourse referent, is already active in the spatio-temporal frame and is the continuous topic throughout the sentence, whereas the assertion is confined to the identification of the focused referent of Pr.
Example (22) also provides a useful syntactic testing ground for our description. The answer has a possessive pronoun emà in the same position and conveys the same properties as genitive Pr’s. This potential for substitution offers strong evidence that the Pr is a predicate nominal in a copular clause where the Pe is the subject. The possessive predication can be realised by predicate nouns in the genitive case or by possessive pronouns in place of the constituent in the genitive. Clearly, the two constructions are not wholly synonymous, as the genitive can refer to specific referents and can also be modified by adjectives. However, the syntactic function is the same: both constructions tend to encode the Pr as an intrinsic attribute or feature of the Pe. In this construction, the verb eînai can be considered the copula, filling the slot [V to be] in the schematic construction [NP nom V to be NP gen ].
This copular construction contains a non-verbal predicate as a complement of a non-agentive, non-volitional and definite Pe/subject. The main function of the verbal slot filler is to transfer grammatical categories, such as tense, aspect and mood into the predicate phrase. This type of construction is intrinsically stative with a semantically underspecified one-place predicate which is subsidiary to the nominal complement that carries most of the lexical-semantic load of the predication and receives focus. This subsidiary status of the verb can be verified through a contrastive focus and negation test (cf. Goldberg 2006: 130), which invariably negates only elements within the potential focus domain. As such, in (24), what is negated is the identity of the Pr, not his existence (see section 4.2).
To sum up, the pragmatic and syntactic phenomena observed above confirm that genitive constructions predicate a state of a certain entity (Pe) that is somehow related to another (Pr) whose relation to the former is already taken for granted; consequently, the informational core of the predication is the identification of the Pr and not the existence of a relation. This statement can be schematized as follows:
This construction can be regarded as a pairing of form and meaning that inherits such characteristics from the copular construction with its complex constructional network of different non-verbal predicates (i.e., nominal, adjectival and prepositional), as represented in figure 4 below.
In all copular sentences the only participant of the predication is the non-agentive, non-volitional and definite Subject, while the verb eînai co-constitutes the predicate with the Subject Complement; this is the predicate nucleus and receives focus. The verbal item remains constant in constructional occurrences, except for some infrequent uses of the near synonymous verb, gígnomai (25).
4.2. Semantic Features
As far as the semantic aspects of the two relata are concerned, Pr’s are typically characterized by the features [+human, +definite] (see examples 26, 27 and 28). Thus, in this respect, there is no difference between genitive Pr’s and dative equivalents. However, the situation is quite different when we consider the data regarding Pe’s, particularly in relation to the ‘definiteness’ parameter. As a matter of fact, where the ‘eînai plus genitive’ construction occurs, Pe’s are generally [+definite, +concrete, -relational] (see example 24 above and the examples below); moreover, Pe’s can be [+human] beings (example 27).
The distribution of semantic features—and in particular the recurrent characterization of Pe as a [+definite] element—is consistent with the pragmatic functions outlined in section 4.1 above. In fact, since the Pe is given, it must also be [+definite]. The Pr, as we shall see, constitutes an intrinsic attribute of the Pe and is in turn [+definite].
The range of relationships which can be expressed by the genitive possessive construction includes two main features which seem to be characteristic of this pattern:
defining pre-existing relations of kinship by specifying the Pr/relative’s identity, the latter always being the father and/or the mother, often referred to by their own name, as in example (27) above and in example (30) below;
specifying or, more often, assigning legal ownership to a specific Pr rather than to someone else (see example 24, above). In this instance, two Pr’s in opposition to one another are frequently seen in the same context; one refers to present ownership, the other to past ownership. Example (29), below, provides an illustration of such use.
Logically speaking, it is important to note that both ownership and kinship with parents are marked by a solid bond which links the Pr to the Pe, a bond which clearly holds even if the entities involved are not both present in the same place at the same time. In the first case (kinship), the connection with parents is a natural one and is the only natural relationship that every human being must have; in the second case (ownership) by contrast, the bond is dictated by socio-cultural convention. Moreover, both types also imply exclusiveness in the relationship between Pe and Pr and, consequently, the uniqueness of the Pr: any given human being will be the son/daughter exclusively of one single father and one single mother; any given house will belong exclusively to a specific owner (or owners, though they are always identified specifically and in a relationship of exclusivity with the Pe) and so on. Given the uniqueness of the Pr, it follows that this becomes a necessary element in qualifying or identifying the Pe, especially in the case of parents.
The impossibility for a [+relational] noun to occur as Pe is interesting for two different reasons. First of all, just as with the dative construction, it is a sign of the predicative function of this type of sentence. However,—and this is the other significant aspect—unlike the dative possessive construction, the restriction systematically applies to the whole class of relational nouns. This behaviour most likely correlates with the Pe’s [+definite] feature in genitive possessive sentences, as well as with the function of this construction as defined in the previous section. As a matter of fact, given that a relational noun implies a possessive relation with another entity at the universal cognitive level, the definiteness of a relational noun entails that the Pr is known, because ignorance of the Pr’s identity—an intrinsic attribute of the relational noun—is incompatible with the definiteness of the noun itself. In other words, using the expression the son—and not a son—means that the speaker and/or the hearer already know who is the father/mother of the son. Consequently, in this case, the genitive possessive construction is not at all necessary.
To summarize, the data presented in this section and in section 3.2 suggest that when an utterance somehow pertains to kinship and ownership, two very different situations may arise.
Of these, the first involves a context in which the Pr is known and given, and where the existence of a relationship of kinship—with the exception of the ascending line—or of ownership is established. In Ancient Greek this kind of relationship, as observed above, is typically expressed by the dative possessive construction.
In contrast, the second kind of situation implies a context in which the Pe is known and given, and where the existence of an exclusive relationship between Pe and Pr (parent or owner) is presupposed. Thus, the new information in such cases is the identity of the Pr. Relationships of kinship—in this case only regarding the ascending line of a [+definite] individual—and ownership fall into the latter category, whose typical expression is the ‘eînai plus genitive’ construction.
In negative sentences—as observed by Kahn (1973: 169) and discussed in the previous sections—the difference between the two kinds of assertion, that is, between the use of the dative or the genitive, seems more evident. This is revealed by comparing example (24) above (regarding a relation of ownership) with (30) and (31):
Example (30) is particularly significant as regards the function of the genitive construction. Here Trygaeus and Hermes are talking about Cleonimus, and while they have no doubt about the existence of Cleonimus’s father, they do have doubts about his identity.
When, by contrast, we are dealing with a dative possessive construction, as in example (31), the negative form of the sentence should be interpreted as ‘There is not X for Y’ and, consequently, ‘Y does not have X’, that is to say, as the negation of the existence of a relation of possession, which in this specific case is of not ‘having children’.
Data regarding the various functional profiles, alongside the kind of possessive relationship expressed by genitive and dative possessive constructions, are presented in Table 1:
Finally, if we re-examine our data following Langacker’s reference-point model, the two constructions under examination differ significantly not only with respect to profiling and trajector/landmark alignment, but also with respect to the roles of reference point and target. As a matter of fact, where a generic possessive relationship and the existence of a generic Pr are presupposed, the ‘eînai plus genitive’ construction entails a specific possessive relationship with a specific Pr, thus identifying the Pr. In other words, this construction profiles a relationship with the specific Pr in whose dominion the Pe is already situated. Consequently, even though the Pr commonly constitutes the reference point in possessive constructions, as regards the specific level expressed by the genitive construction, the Pr is best interpreted as the target and the trajector, while the Pe constitutes the reference point with a landmark status.
The various semantic and pragmatic configurations of genitive and dative constructions outlined in the previous sections determine the types of predicate with which each form co-occurs.
Our analysis shows that the status of the verbal item eînai is inextricably linked to that of the constructions which constitute its typical environment of usage. In particular, the copular status of eînai is due to its occurrence in a construction where the informational core of the predication is the identification of the Pr as a property of a typically definite Pe. This construction is characterized by the co-occurrence of a specific set of pragmatic, semantic and syntactic features such as the focalized definite Pr and the topical definite Pe.
As far as the existential possessive construction is concerned, the occurrence of the verb ‘to be’ is also implied, but the specific constructional context, with an indefinite and focalized Pe encoded by the nominative subject and a dative Pr constituting a reference point, resolves the ambiguity of the verb eînai. In addition, this analysis also supports the primacy of possessive constructions over the individual items which comprise it. The interplay between two morphosyntactic strategies of encoding Pr’s, the typical pragmatic saliency of both Pr’s and Pe’s and the verbal lexeme gives rise to two different constructions with a pairing of form and meaning that schematizes beyond their actual instantiations.
Baldi, Philip and Pierluigi Cuzzolin. 2005. Considerazioni etimologiche, areali e tipologiche dei verbi di “avere” nelle lingue indoeuropee. In Latin et langues romanes. Études de linguistique offertes à József Herman à l’occasion de son 80ème anniversaire, Sándor Kiss, Luca Mondin and Giampaolo Salvi (eds.), 26–36. Tübingen: Niemeyer.
Barðdal, Jóhanna and Thórhallur Eythórsson. 2009. The origin of the oblique subject construction: an Indo-European comparison. In Grammatical Change in Indo-European Languages, Vít Bubeník, John Hewson and Sarah Rose (eds.), 179–193. Amsterdam: John Benjamins.
Barker, Chris. 2011. Possessives and relational nouns. In Semantics: An International Handbook of Natural Language Meaning, Vol. 2, Claudia Maienborn, Klaus von Heusinger and Paul Portner (eds.), 1109–1130. Berlin: Mouton de Gruyter.
Baron, Irène, Michael Herslund and Finn Sørensen (eds.). 2001. Dimensions of Possession. Amsterdam: John Benjamins.
Bauer, Brigitte. 2000. Archaic Syntax in Indo-European: The Spread of Transitivity in Latin and French. Berlin: Mouton de Gruyter.
Benveniste, Émile. 1960. “Être” et “Avoir” dans leurs fonctions linguistiques. Bulletin de la Société de Linguistique de Paris 55 (1): 113–134. Reprinted in Benveniste, Émile. 1966. Problèmes de linguistique générale, 187–207. Paris: Gallimard.
Benvenuto, Maria Carmela. 2014. Possession. In Encyclopedia of Ancient Greek Language and Linguistics, Georgios K. Giannakis (ed.), Vol. 3, 115–122. Leiden: Brill.
Benvenuto, Maria Carmela and Flavia Pompeo. 2012. Expressions of predicative possession in Ancient Greek: “εἶναι plus dative” and “εἶναι plus genitive” constructions. Annali del Dipartimento di Studi Letterari, Linguistici e Comparati. Sezione linguistica, N.S. 1: 77–103.
Conti, Luz. 2014. Non-canonical subjects. In Encyclopedia of Ancient Greek Language and Linguistics, Georgios K. Giannakis (ed.). Leiden: Brill.
Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press.
Croft, William and D. Alan Cruse. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press.
Erteschik-Shir, Nomi. 2007. Information Structure: The Syntax-Discourse Interface. Oxford: Oxford University Press.
Givón, Talmy. 2001. Syntax: An Introduction (rev. edn.). Amsterdam: John Benjamins, 2 vols.
Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press.
Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford: Oxford University Press.
Guilleux, Nicole and Bernard Bortolussi. 2011. Être et avoir dans les langues classiques. Lalies 31: 241–251.
Heine, Bernd. 1997. Possession: Cognitive Sources, Forces, and Grammaticalization. Cambridge: Cambridge University Press.
Herslund Michael and Irène Baron. 2001. Introduction: dimensions of possession. In Dimensions of Possession, Irène Baron, Michael Herslund and Finn Sørensen (eds.), 1–25. Amsterdam: John Benjamins.
Isačenko, Alexander. 1974. On have and be languages: a typological sketch. In Slavic Forum: Essays in Linguistics and Literature, Michael S. Flier (ed.), 43–77. The Hague/Paris: Mouton.
Kahn, Charles H. 1973. The Verb ‘be’ in Ancient Greek. Dordrecht: Reidel.
Kulneff-Eriksson, Karin. 1999. On ‘have’ in Ancient Greek: An Investigation on ἔχω and the Construction εἶναι with a Dative as Expressions for ‘have’. Lund: Lund University Press.
Lambrecht, Knud. 1994. Information Structure and Sentence Form: Topic, Focus, and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Press.
Lambrecht, Knud. 2000. When subjects behave like objects: an analysis of the merging of S and O sentence-focus constructions across languages. Studies in Language 24(3): 611–682.
Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1 Theoretical Prerequisites. Stanford: Stanford University Press.
Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, Vol. 2 Descriptive Application. Stanford: Stanford University Press.
Langacker, Ronald W. 1995. Possession and possessive constructions. In Language and the Cognitive Construal of the World, John R. Taylor and Robert E. MacLaury (eds.), 51–79. Berlin: Mouton de Gruyter.
Langacker, Ronald W. 2003. Strategies of clausal possession. International Journal of English Studies 3(2): 1–24.
Langacker, Ronald W. 2009. Investigations in Cognitive Grammar. Berlin: Mouton de Gruyter.
Lehmann, Christian. 2002. Possession in Yucatec Maya (2nd, rev. edn.) [ASSidUE 10]. Erfurt: Seminar für Sprachwissenschaft der Universität.
Lyons, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press.
Lyons, John. 1967. A note on possessive, existential and locative sentences. Foundations of Language 3(4): 390–396.
Matić, Dejan. 2003. Topic, focus and discourse structure: Ancient Greek word order. Studies in Language 27(3), 573–633.
Meillet, Antoine. 1923. Le développement du verbe “avoir”. In Ἀντίδωρον. Festschrift, Jacob Wackernagel zur Vollendung des 70 Lebensjahres am 11 Dezember 1923, Gewidmet von Schülern, Freunden, und Kollegen, 9–13. Göttingen: Vandenhoeck and Ruprecht.
Napoli, Maria. 2009. Aspects of definiteness in Greek. Studies in Language 33(3): 569–611.
Rappaport Hovav, Malka and Beth Levin. 1998. Building verb meanings. In The Projection of Arguments, Miriam Butt and Wilhelm Geuder (eds.), 97–134. Stanford: CSLI Publications.
Seiler, Hansjakob. 1983. Possession: As an Operational Dimension of Language. Tübingen: G. Narr.
Sorace, Antonella. 2004. Gradience at the lexicon-syntax interface: evidence from auxiliary selection. In The Unaccusativity Puzzle, Artemis Alexiadou, Elena Anagnostopoulou and Martin Everaert (eds.), 243–268. Oxford: Oxford University Press.
Stassen, Leon. 2009. Predicative Possession. Oxford: Oxford University Press.
Taylor, John R. 1989. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press.
In the present paper, Maria Carmela Benvenuto is responsible for sections 1, 2.1, 2.2, 3.1 and 4.1, while Flavia Pompeo is responsible for sections 2.3, 3.2, 4.2 and 5. This paper has benefited from the helpful remarks and suggestions of the anonymous reviewers. To them, our sincere thanks.
On the distinction between ‘have’ and ‘be’ languages and the development of Indo-European verbal roots meaning ‘to have’, the reader is referred to Meillet (1923); Benveniste (1960); Isačenko (1974); Heine (1997: 138–142, 209–211); Bauer (2000) and Baldi & Cuzzolin (2005) among others.
Greek examples are taken from Oxford Classical Text editions, whereas English equivalents are taken from Greek-English Loeb editions. In some cases, the translations have been adapted. However, all examples have been glossed following the Leipzig Glossing Rules (http://eva.mpg.de/lingua/files/morpheme.html). Judging from the Oxford critical editions considered for the examples given in this paper, genitive or dative forms have been transmitted without any variation.
See, for example, Seiler (1983); Taylor (1989); Heine (1997); Baron, Herslund & Sorensen (2001); Stassen (2009).
The abbreviations used in the present text are: Comp ‘complement’, Cop ‘copula’, Dat ‘dative’, FD ‘focus domain’, Gen ‘genitive’, Intr ‘intransitive’, IOP ‘Internal Obligatory Participant’, Nom ‘nominative’, NP ‘noun phrase’, Pe ‘Possessee’, Pr ‘Possessor’, Subj ‘subject’, Top ‘topic’, V ‘verb’.
See, among others, Seiler (1983: 13); Heine (1997: 20); Lehman (2002: 6) and Barker (2011: 4).
It is on the basis of these two kinds of predicative structures that Isačenko (1974) divided the languages of the world into ‘have-’ and ‘be-languages’.
For a diachronic consideration of the distribution of the three constructions, see also Kulneff-Eriksson (1999); Benvenuto & Pompeo (2012) and Benvenuto (2014).
These notions have been developed in typological studies, cf. Heine (1997).
For a useful survey pertaining to the current theories of Construction Grammar in Cognitive Linguistics, see Croft & Cruse (2004).
Theoretical linguistic research in recent decades has been characterized by the debate between projectionist (e.g. Rappaport Hovav & Levin 1998) and constructional (e.g. Goldberg 2006) approaches to argument structure. We cannot provide an extensive discussion of this issue here, but will limit ourselves to pointing out that the projectionist approach assumes that the semantic and/or syntactic information specified by the main verb is the basis for basic sentence patterns. For a detailed account of both perspectives, see Rappaport Hovav & Levin (1998), while a discussion of the limitations of both approaches can be found (among others) in Sorace (2004).
As regards definiteness in particular, we have followed Lyons (1999) and Napoli (2009: 577–583), considering not only its grammatical encoding but also logical/semantic and pragmatic aspects. Logical definiteness is dependent on the inherent semantics of nouns and primarily concerns proper nouns, common nouns of unique entities and generics, while the pragmatic aspect regards cases where the definiteness is inferred mainly from the context (linguistic or extra-linguistic). It is worth noting that a noun can be definite even if it occurs without the article and this is precisely the case in Ancient Greek. As a matter of fact, the definite article in Homer is not yet fully developed as an independent category, and in Classical Greek the picture is neither consistent nor systematic: the definite article occurs in all cases of pragmatic definiteness, while it is optional with generics, proper nouns, nouns of unique entities and with mass and abstract nouns, that is, the entire domain of logical definiteness (Napoli 2009: 582–583).
As already mentioned, this label applies to those concepts that imply a possessive relation with another entity at the universal cognitive level. According to Barker (2011: 4), “the most common relational concepts cross-linguistically include family relations (mother, uncle, cousin); body parts (hand, head, finger); and intrinsic aspects of entities such as color, speed, weight, shape, temperature”.
On this topic, see also Lyons (1967); Taylor (1989); Langacker (1995, 2009); Heine (1997) among others.
As noted by Lambrecht (1994: 153), “since the topic is an element of the pragmatic presupposition evoked by the sentence, there is a sense in which the topic itself must be taken for granted, hence must be outside of the scope of negation or modality in assertion”.
The dative in Ancient Greek encodes a wide range of syntactic functions and semantic roles and, because of its complex functional profile as a syncretic case, its role in the expression of possession needs to be considered with some caution. As is well known, it essentially codifies a generally animate entity, partially involved in the state of affairs licensed by verbal semantics; in some impersonal constructions it can also function as a non-canonical subject (see e.g. Barðdal & Eythórsson 2009; Conti 2014). Nevertheless, given the spatial function inherited from the Indo-European locative, it seems plausible to assume a locative/adessive function of the dative in possessive constructions.
Existential sentences vary crosslinguistically in their structure, semantics, and pragmatics. Nevertheless they appear to share certain characteristics. We cannot provide an extensive discussion of this issue here, but limit ourselves to pointing out that existential/presentational sentences are characterized by the indefiniteness of the subject referent (Givón 2001: 255–261).
Herslund & Baron (2001: 13) do not specify this distinction, even though the examples they provide all regard the noun mother.
This structure can be considered an extension of the existential use, which Kahn calls “the vital use” (1973: 240–245).
The plus sign in parentheses indicates that the number of definite/indefinite Pe’s is relatively unimportant (cf. Benvenuto & Pompeo 2012).
See for example Seiler (1983); Taylor (1989); Heine (1997); Baron Herslund & Sorensen (2001); Stassen (2009).
On this topic see also Lyons (1967); Taylor (1989); Langacker (1995 2009); Heine (1997) among others.