A THREE TIMES MARKED POSSESSIVE STRUCTURE IN HUNGARIAN

The paper presents a unique construction of possessing, in which the fact of the possessing is marked three times. On the regent of the construction (which refers to the possessed thing) there is a possessive suffix, and two other linguistic elements refer to the possessor. Both are personal pronouns, the first is in nominative, the second is in dative case. The study looks through the role of this structure in the secondary grammaticalization of the suffix. The paper suggests that the grammaticalization is not a one way, linear path. First, because there can be junctions in the grammaticalizational paths. Second, because in the semantic extension, the newly appearing functions may have an effect on the already existing system. The paper offers a corpus based analysis on how emphasizing the possessor can be related to the ethical dative function. The ethical dative usually occurs on the personal pronoun in the first or second persons (me, you) and refers to a figure that is highly involved in the communicative situation. Thus, both structure types (the three times marked possessing and the ethical dative) elaborate emphasizing personal relations in the communicative situation.


Introduction
The paper discusses a particular structure type in relation to the semantic structure of the Hungarian dative case suffix -nak/-nek, with special regard to the role of this structure type in the grammaticalization of the suffix. The structure type described in the title of the paper is the following: (1) neki az ő élete he/she.DAT 1 the he/she.NOM life.PX.3SG 'his/her life' The uniqueness of the structure is in the fact that the possessive relation is triple marked: the possessor is marked by a possessive personal suffix (élet-e life-PX.3SG) on the head of the structure and also by two pronouns: one personal pronoun in dative case form (neki) and another one in nominative form (ő 'his/her').
The main question of the study is how this construction relates to other functions in the semantic structure of the suffix -nak/-nek, and how it provides a link to other functions arising via meaning extension. The semantic analysis of the structures shows that the dative element appears in the discourse more as an involved figure or perspective marker. Therefore, the structure will be presented in this context in the paper.
The research is corpus-based, using mainly the Old Hungarian Corpus (OHC), but also other historical corpora (Old and Middle Hungarian Corpus of informal language use, abbreviated OMHC, and Hungarian Historical Corpus, abbr. HHC) as well as contemporary text databases (Hungarian gigaword corpus, abbr. HNC).
The paper starts with the theoretical framework (2), in which the process of grammaticalization and studies on secondary grammaticalization are discussed, with a review of the literature on the meaning extension of -nak/-nek. Section 3 presents the methodology used for the study, followed by a discussion of the results (4). The paper concludes with a summary (5). The study was conducted within a functional cognitive theoretical framework.

Theoretical framework
The paper presents one chain in the extension of the meaning of -nak/-nek. In the study, I examined this semantic change within a functional cognitive framework (cf. Langacker 1987;Tolcsvai Nagy ed. 2017). According to the functional cognitive description, most linguistic items have multiple meanings, and this is true not only for lexical items with conceptual meanings, but also for grammatical items such as case markers in Hungarian. The system of semantically related meanings of a linguistic element is called polysemy (Tolcsvai Nagy 2013: 232;Langacker 2008: 38).
Functional cognitive language description examines polysemy in the context of prototype categorisation. According to prototype categorization (cf. Rosch 1973), categories are created in the course of cognition, based on the abstraction of common properties of experienced phenomena, and category elements are organized around a central, prototypical instance. The prototype has most of the characteristic properties of a category. New meanings in the network of meanings arise by the extension of the central category element, which provides the cognitive motivation for meaning extension (Lewandowska-Tomaszczyk 2007: 148).

Grammaticalization
Grammaticalization is a type of linguistic change in which a lexical meaning of a linguistic element develops into a more grammatical meaning, or the meaning of an already grammatical linguistic element becomes more grammatical (Heine-Claudi-Hünnemeyer 1992: 2). On this basis, we can talk about primary and secondary grammaticalization processes (Traugott 2002: 26-27). The present study falls into the latter, i.e. it investigates the semantic change of a linguistic element whose meaning is already highly abstract (Narrog-Heine 2012: 3). In the early stages of grammaticalization research, attention focused primarily on formal changes, but more recently the pragmatic perspective has also become a focus of investigations (Dér 2008). For example, the function of a case marker may be extended by a change in its distribution: it may appear in contexts in which it did not before, and it may be able to take on new meanings with new kinds of heads. These are what Diewald calls 'untypical contexts' (Diewald 2006). Diewald also draws attention to the fact that the new meanings appear in use.
The process of grammaticalization is usually referred to as a path (e.g. Dér 2008) or a chain (e.g. Heine-Claudi-Hünnemeyer 1991: 220-229). These terms refer to the overlapping stages of the process and to the fact that each stage cannot be broken down into discrete elements. Narrog and Heine (2012: 2) note that one of the driving forces of the grammaticalization process is frequency of use along with variation in uses. In the case of case markers, this means that a linguistic element that has progressed along the grammaticalization path, previously occurring after certain types of nouns defined according to semantic criteria, loses this specificity and occurs with an increasing number of nouns (Heine 2008: 463). The semantic consequence of this frequency is what Heine calls desemanticisation, by which he means that the linguistic element going through the process loses its lexical meaning (cf. bleaching). According to Hopper and Traugott (2003: 94), this approach reflects the fact that grammatical descriptions usually view linguistic change from the perspective of the origin of the process, rather than from the point of what the end result will be. In the context of the latter viewpoint, Sweetser (1988: 392) notes that it is more meaningful to view change from both directions: as the lexical meaning is obscured, the linguistic element is also enriched; namely, by the expression of a new grammatical function (cf. loss and gain).
The grammaticalization of -nak/-nek is typically described in the literature from the historically primary lative meaning to the expression of possession. The historical grammar reports the following for the functions of the suffix -nak/-nek: Korompay (1991a: 311-315) refers to the socalled dative suffix as -nak/-nek 1 , and to the genitive suffix as -nak/-nek 2 . The case marker -nak/-nek 1 is originally shown as a suffix answering the question where to?, i.e. as a spatial complement with a lative orientation; after that, it is reported in dative, essive and dative possessive structures, a type of structure in which the noun with -nak/-nek at the end (and which refers to a possessor) is a dependent of the possessive substantive verb van, nincs or sincs. At the end of the line, the 'abstract relations' appear, with -nak/-nek 2 treated as the possessive marker. The Historical grammar (Korompay 1991a) does not report on the processes and tendencies of meaning expansion, only on the functions that can be discerned.
From a cognitive semantic point of view, Gábor Tolcsvai Nagy (2005) reviews certain functions of the suffix. His description focuses on how the Hungarian possessive structure can be understood as a metaphorical extension of a spatial relation. Thus, he shows the divergence of the meaning of the suffix in four steps, from the direct, spatial relation involving contact (falnak megy -wall.DAT go.PRES.3SG 'going to the wall') through the contactless spatial relation expressing direction (délnek tart -south-DAT aim-PRES.3SG 'going south') and -nak/-nek denoting the recipient of a transactional process (ad egy könyvet a fiúnak -give-PRES.3SG a book.ACC the boy.DAT 'giving a book to the son'; in metaphorical space: üdvözletét küldi az apjának -greetings.PX.3SG.ACC send.PRES.3SG the father.PX.3SG.DAT 'sending greetings to his father') to the expression of the possessive relation (a fiúnak a könyve -the boy.DAT the book.PX.3SG 'the boy's book').
The present paper argues that grammaticalization is not a strictly one-way process: new functions can refer back to functions already in the semantic structure, either structurally or semantically. The paper places the triply marked possessive structure on the grammaticalization map of the -nak/-nek suffix.

Method
For this paper I used the Hungarian Historical Generative Syntax Corpus (Old Hungarian Corpus) (cf. Simon 2014). Part of the corpus is only in original orthograph, a smaller part is normalized, i.e. it is also transcribed according to contemporary orthography. Moreover, some texts are also morphologically analysed, i.e. the morphological properties of individual word forms are labelled, which can thus be searched for in a targeted way.
I carried out several queries on the corpus. First, I performed a statistical analysis by function. To do this, I searched for nouns ending with -nak and -nek in the normalised, analysed texts. This resulted in a total of 10,106 tokens. From the data, I selected a random sample of 400 tokens using Excel (which turned out to be a sufficient number to be representative of the whole sample). These were then analysed manually by function. The list of 400 resulted in 11 blind hits, so the figures should be understood in the light of the 389 valuable hits. In addition to these, I also performed targeted searches, these are indicated next to the data with a TS mark, indicating the corpus from which they originated. The sources of the targeted searches are the Hungarian Historical Corpus (HHC), the Old and Middle Hungarian corpus of informal language use (OMHC), the Hungarian National (gigaword) Corpus (HNC v2.0.5) and the Old Hungarian Corpus.
The corpus data should be interpreted in the light of the specificity of the texts. Most of the examples analysed are from codices, many of them are from the biblical or religious literature translated from Latin, and the specific features of these examples will be discussed in the respective examples.
In the morphological labelling I indicate possessive relations, noun cases and the information of the verbs.

Results
Among the randomly selected data, I have distinguished the following functions (1-9). In the list, I name the function or semantic role that the suffix -nak/-nek plays in its construction. Each function is followed by an example. So I have primarily considered function in the separation, but of course some of the functions occur in constructions with typical structures. The obvious exception to this separation is group 3, namely the postpositional phrases. However, it is worth treating this as a separate group because of several peculiarities of its genesis. Groups 5, 6 and 8 also have a typical constructional pattern. Figure 1 shows the whole grammaticalizational map of the suffix. However, this paper only examines one of the chains. (For a larger, comprehensive study on the grammaticalization of the suffix, see Pomázi 2021: 23-50.)

A brief description of the abstraction of meaning up to possessive constructions
The abstraction of the meaning of the suffix can be briefly outlined below. The marking of the spatial relation in the case of the suffix indicates the construing of the target in a directional relation ( (2): lábaimnak 'in the direction of my legs'), and directionality is also present in the expression of the RECIPIENT.
( 2)  In (3), azoknak 'to those' elaborates the endpoint of a transactional process: it designates the persons in space to whom the thing moved, the bread, is delivered. Attention may also focus on whether the scene is beneficial (4) or harmful (5)  'What profit is it to a man, should they gain the whole world, and should they find peril to his soul in exchange?
The transitions between the dative and the genitive functions are stages in the process of meaning expansion that have a characteristic constructional pattern.
(6) gonoszabbak az rágalmazók az orvaknál és az tolvajoknál, kik embereknek ez világi jószágukat elorozzák wicked.COMP.PL the slanderer.PL the blind.PL.ADE and the thief.PL.ADE, who.PL people.DAT this wordly goods.PX.3PL.ACC take.away.PRES.3PL 'more wicked are the slanderers than the blind and the thieves, who take away the worldly goods of people'.
(7) ezeknek gyökerük nincs these.DAT root.PX.3PL not.be.PRES.3SG 'they have no roots' Sentence (6) is an example of a possessive dative construction. The possessive dative (embereknek 'of people') is subordinated to a verb (eloroz 'take away') and is grammatically in a possessive relation with a noun having a possessive personal suffix (jószág-uk-at -goods-PX.3PL-ACC 'their goods') that agrees in number and in person with it. The participant elaborated by the -nek suffixed word can thus be processed in the scene as both possessor and beneficient/maleficient (see Elekfi 1993). With regard to the word order pattern, the study does not pay special attention to where the noun with the possessive suffix and the one with -nak/-nek at the end occur in the actual clause in relation to the verb. The dative possessive in (7) represents a structure in which the noun with -nak/-nek at the end is a dependent of the existential verb van having a possessive meaning, nincs/sincs 'not to be', and it expresses the possessor. The former is thus closer to the dative node of meaning, the latter to the genitive. A more abstract stage in the grammaticalization is the marking of the possessor in a certain structure, where the head of the structure is always a noun.

The possessive structures
More than half of the data express a relationship of a genitive nature, therefore relatedness. This relatedness is, semantically speaking, highly varied (cf. Hadrovics 1969: 139-177). For ease of treatment, I will treat them here as belonging to one category. In her study of the frequency of the genitive role of -nak/-nek suffixed nouns in codices, Klára Korompay came to the following conclusions (Korompay 1991b(Korompay : 375, 1992b. The -nak/-nek suffix in this function is the second most frequent noun suffix in the codices after the accusative suffix -t, which is also due to the influence of Latin texts. Since the Latin genitive is always realised as an overt morpheme, translators striving for faithful morpheme-by-morpheme translation used this variant in Hungarian instead of expressing the possessor with a default nominative form. In the Old Hungarian Corpus, there are also structures expressing multiple possession where several of these functions are formally marked with the -nak/-nek suffix: There are three possessive structures in (8), two of which are marked with a -nak/-nek suffix. In (9), both possessive structures are marked with -nak/-nek, although the omission of one of them would not affect intelligibility, the multiple markedness (one possessive suffix on the head, and another one on the dependent) being motivated by the multiple possessive relations themselves (Korompay 1991a: 301). In (10), in the structure kijelentésének ideiglen, the relation is apparently marked once, since it is not read idejéiglen. At the same time, the root idő is a type that can alternate the tone and duration of its last vowel, so that the possessive personal suffix -i inside the form idejéiglen and the i of the terminative case marker -ig(len) coincide. This construction kijelentésének ideiglen is related to the postpositional structures that require a noun with -nak/-nek at the end. The structure is embedded into the syntactic structure of the sentence by ideiglen, and kijelentésének ideiglen together marks the time of the predicate lakozik vala 'was dwelling'. This is also consistent with the fact that, according to the findings of language historians, the marking of the possessive relation has progressed from simpler to more marked (Korompay 1991a: 300). Therefore the relationship between other meanings of the suffix (especially the lative and dative as the historically primary and one of the most common functions, respectively) and the above mentioned meaning of -nak/-nek should be considered as motivated.
In the random sample, there were some items in which the possessor's person was marked more than once. These are: In examples (11)-(13), the possessor marked with the case suffix -nak/-nek is accompanied by the personal pronoun ő 'she/he' in nominative form, which also refers to the person of the possessor. The relation between the two is coreferential, i.e. they refer to the same character in the discourse universe. The inclusion of the nominative personal pronoun in this case may serve the role of emphasis (Sipos 1991: 362). This means that it has a pragmatical role by putting an emphasis on the person of the possessor. If we accept that valakinek ő in (13) is the dependent of élete, the construction cannot be interpreted as dative possessive, because in that case they would be the dependents of nincs. Also, nincs in (13) has an existential rather than a possessive meaning.
In the Old Hungarian corpus, I searched for the following types of structures in the plural possessive structures: nekem az én (me.DAT the I, 'my'), neked a te (you.DAT the you, 'your'), neki az ő (he/she.DAT the he/she 'his/her'). In this way, I wanted to find structures in which the person of the possessor is marked in multiple ways: with a personal pronoun in the dative case and a personal pronoun in the nominative case. Klára Korompay (1992a: 347) points out that multiple marking may be a stylistic peculiarity of the text: in the codex texts, the translator wants to render the original Latin text in a structurally faithful way. The equivalents of én 'my' and te 'your' in the Latin original may be in the dative case, but in the third person singular, the structure does not contain a dative pronoun. Although the Latin pattern is not followed in every case (G. Varga 1992: 461), its influence is still predominant in the Old Hungarian period. I have investigated this in the following first and second person possessive structure obtained by a targeted search (from the normalized part of the corpus, the source is mentioned in brackets at the end of the example): (17) "Tu quid dicis de illo, qui aperuit oculos tuos?" In example (15), note that neked could also designate a beneficiary, and in (14) nekem could be interpreted in the Hungarian version as 'for me', too. The Latin equivalents do not include the personal pronoun me, you in dative. This means that in these particular cases, it was not the translator's concern for structural fidelity that could have been the reason for the inclusion of the dative inflected form of the personal pronoun and the formally unmarked form of the pronoun expressing the possessor in the Hungarian version. In these cases, the nekem, neked components of the structure seem to be close to the dativus ethicus function. About dativus ethicus, Rácz and Szemere (1985: 86) note that it occurs in Hungarian in the form nekem (so the first person singular pronoun in dative case) and is a feature of impulsive speech. German descriptive grammarians emphasize that dativus ethicus expresses the personal, inner interest of the speaker (Pete 1998: 302), i.e. here too its use is reduced to first person singular. In Czech, the dative form of the second person singular personal pronoun can also express this function. Fried (2014: 13-14) emphasizes the discourse-building function of the pronoun: the actor it expresses is a participant in the discourse itself, it emphasizes them. Pete (1998: 302-303) furthermore points out that this structure-function type exists in most European languages (French, German, Polish, Czech, etc.). She thus assumes the influence of a mediating language behind the phenomenon. As a source structure, she identifies the following type (with the relevant correspondences marked in italics): (18) Quid mihi Celsius agit? 'What does my Celsius do for me?' Ágnes Pete derives the dativus ethicus from such constructions, which foreground one of the participants of the discourse. However, the situation of the third person is different in Hungarian, too. Although lexical nouns can be substituted for nekem/neked, in this case the article is omitted (cf. sentences (11)-(13)). The following example can be found by searching for "neki az ő"he/she-DAT the he/she in the normalized part of the corpus (ignoring those where neki clearly refers to a RECIPIENT/BENEFICIENT, and not a possessor): The structure type "nekem az én -I-DAT the I" can also occur in sentences in which the personal pronoun in dative case is the dependent of a verb, and which designates the addressee of the process in the sentence (20): (20) elég nekem az én nyomorom (HNC) enough I.DAT the I misery.PX.1SG 'my misery is enough for me' (HNC) In terms of dependencies, (20) is different from the structures presented in sentences (14), (15), (17). If we accept that, because of following the Latin pattern, the dative personal pronouns (nekem, neked, neki) in the latter sentences reinforce the person of the possessor, they are in a chain with the nominative case pronouns (én, te, ő). In sentence (20), the personal pronoun nekem 'for me' is a dependent of the predicate elég 'enough', while én 'my' refers to the possessor and is the dependent of nyomorom 'my misery', so they are not in the same chain. However, it also occurs in sentences where it is not easy to decide whether it is an ethical dative or a pronoun that makes the possessor's person more explicit (20) Here we find cases in which a) the inflected noun and the one in nominative occur in different functions, b) the dative accentuates the nominative in its possessive function, c) the noun in dative emphasises one of the participants in a communicative situation. Thus, whether the function of the structure type is to indicate the possessive or to emphasise the role of the participant, this function involves foregrounding one of the participants in the situation: its function is thus primarily pragmatic, as it is able to display the point of view of the conceptualizer. The examples also show that the specific pragmatic function of the dative form of the personal pronoun is to emphasize the deeper involvement of the discourse participant in the scene. This is also reflected in uses where it is not easy to decide whether the role of the pronoun in the sentence is closer to that of marking the RECIPIENT/BENEFICIENT or EXPERIENCER or the possessor. The construction schema of the structures presented is as follows: a personal pronoun in dative case form, followed (either after an attribute or a chain of attributive syntagmatic structures) by a noun with the suffix -nak/-nek at the end and also having a possessive suffix, which is in number and person agreement with the pronoun. And in examples (23) to (25), the anchored verb (or core sentence, see Imrényi 2017) of the clause requires a dative dependent. The analytical problem is that in examples (23)-(25) the pronouns nekem, neked, etc. can be interpreted as markers of the possessor in terms of dependency relations, but it is also important to note the extension of involvedness, as they do not necessarily mark the BENEFICIENTs described by the clauses but rather possibly the conceptualizers of the scene, by which the pronouns highlight the conceptualizers' points of view. Analysing them as marking the possessors is also supported by the fact that the pronouns agree in person with the nouns ending with -nak/-nek. However, the dative interpretation can also be explained by this extension of involvedness.
This point is of particular importance in the interpretation of sentence (25). The core of the clause (problémát jelent 'cause a problem') requires the elaboration of the dative complement, i.e. it must be made explicit to whom the subject (bekapcsolódni 'to join') is harmful. However, múltamnak 'to my past' is not an appropriate candidate for elaborating this, since it is incorporated into the syntactic structure of the sentence with the postposition köszönhetően 'thanks to', and the múltamnak köszönhetően 'thanks to my past' composite structure thus participates in the dependency representation as an adverbial of cause. Thus the only suitable candidate for the role of the dative complement is nekem 'for me', which is thus processed both as a participant and as a possessor in the background (since the possessor of múltam(nak) 'my past' is also the first person singular speaker: *nekem nekem a múltamnak köszönhetően… 'for me, thanks to my past'..., but of course double elaboration would be superfluous).
In (26)-(27) the situation is different in that the nouns ending with -nak/-nek (szüleimnek 'my parents'', kislányomnak 'my daughter's') are not dependents of a verb, but of another noun (szülei 'parents', bukása 'failure'). Thus, the structure is more likely to be interpreted as the first possessive marker of a double possessive structure than as a dative complement; the latter interpretation would be supported by the sentence position (in this case, the speaker of the sentence would communicate as a lived experience that their grandparents are musicians or that their daughter failed at school).
The examples in (23)-(27) also show that there is no rigid boundary between the different functions of the suffix -nak/-nek. A structurally constructed possessive component may also have properties that semantically bring the occurrence closer to the dative meaning node. The dynamic processing of the sentence also plays a role: the appearance of a noun having a possessive suffix, for example, opens up the interpretation that there is a possessive relation in the sentence.
In particular, this ambiguity may be true of pronouns, which have a schematic meaning and whose function in a sentence depends to a large extent on the context in which they occur.
Thus, the study has argued that this pragmatic function of the suffix provides a link between meanings in the extension of meaning. In this way, the following grammaticalization path from the historically primary, locational to possessive relation marking can be outlined: The diagram shows how the extension of meaning from the lative meaning to the expression of possession and belonging happens. The paper suggests that grammaticalization should not be understood as a linear process, but should be considered in a more dynamic model. On the one hand, grammaticalization should be considered as a process that may involve branching into multiple pathways. This is particularly important in the case of the extremely rich meaning network of the suffix -nak/-nek, especially when looking back at the functions isolated in the corpus analysis. On the other hand, any new features that emerge may also reflect back on the already existing system as a whole. Thus, for example, both the ethical dative and the possessive structure may share the pragmatic function of emphasising a human participant in the scene (the discourse actor in the case of the ethical dative, and the person of the possessor in the possessive structure).

Conclusion
The study examined aspects of the grammaticalization of the suffix -nak/-nek. The central issue of the study was a type of structure in which the person of the possessor is marked twice: with a base-form (nominative) and a dative case-form personal pronoun. The novelty of the study was that it provided a corpus-based analysis of the grammaticalization of a case suffix.
The study has shown that the accentuation of the person of the possessor with the personal pronouns nekem, neked, neki (which coincide morphologically in Hungarian, so can both mean 'for me, you, him/her' and 'my, your, his/her') etc. can have a pragmatic linking function in the grammaticalization of the suffix, which can also extend to other functions. By foregrounding the actor of the speech situation, it can be related to the ethical dative, whose most typical form of elaboration is also the first and second person dative personal pronoun. In addition, the study has provided examples of how the foregrounding of participants in a speech situation or the point of view of the speaker by means of the personal pronouns nekem, neked can be situated on the continuum between dative and genitive functions.
The investigation will continue by constructing a grammaticalization map of the entire meaning network, which can be supported by the parameters of secondary grammaticalization processes.