Genitive-dative syncretism in the Balkan Sprachbund: an invitation to discussion.
The phenomena which are generally recognized as belonging to the so-called Balkan Sprachbund--originally signalled by the Slovenian linguist and philologist Jernej Kopitar, but formalized from the 20s and 30s of the 20th century thanks to Gustav Weigand's and Kristian Sandfeld-Jensen's studies and Alexandru Rosetti's definition (1958)--have been differently accounted for (according to the differing degrees to which they are realized) in modern Balkan linguistics. They generally include a significant simplification of the declension system, the analytic expression of the future and of comparison (of adjectives and adverbs), a partial sensitivity of the case system, particularly visible on pronominal expressions, to Vocative, the loss of infinitive, the grammaticalization of the [+def] category through a postpositive article, the form of numerals between 11 and 19 and so-called clitic doubling. (1) In addition to the traditionally recognized balkanisms, Ammann & van der Auwera (2004) questionably argue that so-called 'complementizer-headed' main clauses for volitional mood may represent a Balkanism. (2) In this paper I analyze the first of the mentioned balkanisms, i.e. the cross-linguistic occurrence of the loss of morphological case, with particular reference to the genitive-dative syncretism apparently characterizing the majority of the Balkan languages. (3) In this perspective, the question arises as to whether genitive-dative syncretism can be considered as a real balkanism or as a single realization of a tendency common to all languages which is not necessarily ascribable to the Balkan Sprachbund. (4) I will mainly focus on Romanian, Bulgarian and Greek, since these languages are representatives of the different modalities in which genitive/dative is expressed according to standard categorizations, and I will also give a brief overview on the interesting case of Albanian. In fact, an attentive analysis reveals that the homonymy between these two cases is not always a stable feature and that the origin of this phenomenon is not fully clarified at present.
2. The A's and B's of Case Syncretism
As noted by Baerman (2009: 219), case syncretism is defined in terms of the combination of multiple distinct case values in a single form (what Jakobson calls 'case neutralization'), which implies an observable asymmetry between paradigms within a language, leading therefore to case polysemy where one and the same form expresses two or sometimes more case functions. (5) The idea that two or more different morphological cases may undergo syncretism, which constitutes a widespread characteristic in the diachronic development of a number of languages, is generally supported by the assumption that there must be some underlying affinity (either semantic or morphosyntactic) between the functions which come to be expressed in the same way. To capture this generalization, we may adopt the Syncretism Principle as proposed by Alexiadou and Muller (2004: 3):
Syncretism Principle (Alexiadou and Muller 2004) Identity of forms implies identity in function (within a certain domain, and unless there is evidence to the contrary)
According to Blake (1994: 44-45), the syncretism between genitive and dative--which he analyzes starting from Latin, extending then the generalization to all other languages by virtue of the fact that Case embodies widely distributed properties of language--can be partially explained (i.e. without considering its morphosyntactic distribution) with the specifications [+oblique] and [-local] as represented in Table 1. Nevertheless, Dative often indicates actual or metaphorical state in languages other than Classical Latin (cf. Germ. Ich bin in der Kuche vs. Ich bin in der Kuche sehr eitel) (6) or metaphorical direction, corresponding to the theta-role 'benefactive' (cf. Rom. Ti am DAT cartea de lingvistica 'I gave you the linguistics book'), whereas in languages such as Romani it has no directional use in the spatial sense (Matras 2002: 88) but only encodes the function of benefactive object of particular verbs.
In fact, as observed by Blake (1994: 143), the most outstanding cross-linguistic functions of dative generally correspond to: the encoding of the indirect object or of some two-place verbs 'low on the transitivity scale'; the indirect object of three-place verbs such as give (indeed the definition 'dative' is derived from Gr. ptosis dotike 'giving case' through Latin dativus); the roles of purpose, as well as the possessor (suffice it to think of French Ce chapeau est a moi, which clearly has a 'genitive' value).
There is therefore significant uncertainty as to the nature of case syncretism, basically summarizable in two principles: if this phenomenon were exclusively dependent on what we can call 'semantic' features (regarding meaning and function), then its occurrence should be absolutely predictable, i.e. we should be able to tell which cases will exhibit syncretism cross-linguistically--because of proximity in function; if, on the other hand, this kind of syncretism were only attributable to purely morphological reasons, it would not be possible to make any predictions (Baerman 2009). However, this distinction cannot possibly be so categorical: if we take, for instance, dative as a prototypical example of morphological case undergoing syncretism, we will observe that it can absorb the functions of different other cases (suffice it to think of syncretism of the types dative-locative in Ancient Greek, dative-accusative in Hindi-Urdu, dative-nominative in OE, (8) etc.). What is more, on the one hand it is not simple to account for the different kinds of syncretism basing on the semantics of the case; on the other hand, there is a clear asymmetry between the possible variants of this phenomenon: for example, we (intuitively) assume that the genitive-nominative syncretism does not have the same statistical incidence as the genitive-dative one. Heine and Kuteva (2005: 148-149) propose three different and not mutually exclusive causes of case syncretism:
1. Owing to phonetic processes, different case forms become formally indistinguishable.
2. One case category [C.sub.1] extends its functional domain and takes over the function of another category [C.sub.2], eventually replacing the latter.
3. One of the case markers disappears and its functions are taken over by the other case marker.
From this point of view, which is evidently less emphatic, syncretism can be considered as a compounding of semantic, syntactic and morphological aspects. Interestingly enough, Manzini and Savoia (forthcoming) argue that the genitive-dative syncretism may be explained in terms of case percolation of the type Dat. > Gen., (9) as the second argument of ditransitive verbs is interpretable as some sort of 'possessive marker, (10) in the sense that it roughly corresponds to the attribution of a mental state or material possession to the 'dative object'. If we indeed consider the following sentences from Romanian (taken from Manzini/Savoia):
(1) I-l am DAT baiat-ul-u-i / fet-e-i him/her-it have given boy-the-m-obl-sg. / girl-f-obl.sg. 'I gave it to the boy/the girl'
(2) pahar-ul baiat-ul-u-i glass-the boy-the-m-obl.sg. 'The glass of the boy/ the girl'
we recognize in the first place that, although the function of the morphological case in (1) and (2) is intuitive (dative and genitive, respectively) according to communicative/pragmatic implications, (11) some kind of possessive meaning is implied. This value, which is still identifiable for instance in It. Ho lavato i capelli a Maria, Fr. J'ai lave les cheveux a Marie, Sp. Le he lavado el pelo a Maria (Eng. 'I washed Mary's hair'), where the benefactive dative inherently corresponds to a possessive (Ho lavato I capelli di Maria, etc.), does not exhaust of course all possible dative constructions. Nevertheless, following Jung/Miyagawa's (2004) analysis, this possessive function of dative can also be extended to the second argument of ditransitive verbs, such that a structure as I-am DAT cainelui sandvisul meu ('I gave my sandwich to the dog') can be argued to include an object (marked by accusative) and a 'possessive object' (marked by dative (12)).
3. The Distribution of the Phenomenon within the Balkan Sprachbund
Case syncretism between genitive and dative (Zusammenfall von Genitiv undDativ, Schaller 1975: 100-101) is shared by Romanian, Albanian, Bulgarian, Macedonian and Greek. This phenomenon basically shows three tendencies: while in Romanian and Albanian dative can be said to have absorbed genitive or, in Banfi's (1985: 52) words, the inherited dative suffix is used to express the genitive", (13) the opposite goes for Greek, where genitive is used to express dative case (e.g. in the genitive of separation). In Bulgarian, instead, genitive/dative is expressed through so-called na-phrases. As regards the areal origin of this phenomenon, the emergence of such syncretism is difficult to ascribe to one single language or to explain through a simple model of unidirectional borrowing of features. Uncertainty in recognizing the exact emergence of genitive-dative merging is partly determined by the fact that it is present in a number of languages in and out of the Indo-European family. Three possibilities exist as to the genesis of this phenomenon: its Romance origin (Reichenkron 1962: 12) and successive spread in the Balkan Sprachbund through Romanian, in the sense that the same form for genitive and dative is attested for the feminine singular nouns of the 1st declension in Classical Latin (rosae 'rose-GEN/DAT'). The oversimplification of the case system occurring in Vulgar Latin, triggering the extension of the genitive-dative syncretism to other declensions, spelled doom for the system of declensions itself. Given that the loss of morphological case is a generalized phenomenon of Romance languages, which have retained a formal distinction between nominative and oblique only in the pronominal system, with the exception of Romanian, a language that displays a distinction of the type nominative/accusative and dative/genitive, the merging of the latter in Romanian may represent a stage of crystallization or, in Petrucci's (1999: 10) words, 'a stage in the obsolescence' of the Latin case system. (14) What is more, in Latin dative was commonly used to express possession, irrespective of the simplification of the inflectional system: for example, it was often utilized as related to inalienable possession (names, family members, body parts, etc.), which seems to suggest that dativeness is significantly connected to this kind of semantic relation. Moreover, this structure is found in different forms in a number of Indo-European languages:
(3) French Mon livre a moi My book 'to' me 'My book'
(4) Hindi/Urdu Kosar ki tin behnen hen Kausar POSP-DAT three sisters are 'Kausar has three sisters'
(5) Hindi/Urdu mujhe bhuk he I-DAT hunger is 'I'm hungry'
(6) Latin Mihi est pecunia / Mihi est filius I-DAT is money / I-DAT is son 'I have money' / 'I have a/the son'
(7) Latin Mihi pater estHippolochus [...] vir auda[chi] I-DAT father is Hippolochus [...] man daring 'My father is Hippolochus, a daring and strong man' (from: Iliad, VI, 119-236)
The Romance origin hypothesis seems to be plausible in the light of the numerous examples of genitive-dative syncretism attested by Vulgar Latin inscriptions found in Dalmatia and in the Danubian region (with DPs of the type filius regi vs. filius regis).
A second hypothesis consists in Banfi's (1985: 53) proposal that this phenomenon may have originated in Bulgarian and Macedonian, (15) which are caseless languages, although they retained vocative 'as a living category' in particular morphosyntactic contexts (a 'secondary balkanism' according to Schaller 1975). A third, possibilities-exhausting proposal is that the origins of the merging of dative and genitive are to be found in Greek 'around the first centuries of our era' (Sandfeld 1930: 186) as a means of simplifying the inflectional system. Data from Petrucci (1999) show that this syncretism appears to be fully fledged in a certain context (for the sake of clarity, let us consider the first example for each language as a genitive):
(8) a. Rom. casa batranul-ui are spus acesta batranul-ui old man-GEN old man-DAT b. Alb. shtepia eplak-ut ja thaplak-ut old man-GEN old man-DAT c. Bulg. kastata na starikat (mu) rece na starikat old man-GEN DAT old man d. Gr. to spiti tu anurop-u to eipe tu anurop-u GEN old man DAT man-DAT 'the house of the (old) 's/he said it to the (old) man' man'
In fact, as a first approximation, there seems to be complete identity between the expression of dative and that of genitive in nominal expressions (cf. also Rom. ajutor populafiei, which is ambiguous between an interpretation in which populapei is dative--ajutor requires a dative object--and one in which it is genitive and particularizes the noun). An analysis of the syntactic behaviour of genitive and dative case in the single languages will clarify that this correspondence is not complete.
As observed by Cornilescu (1995: 11), 'Romanian is correctly said to present a genitive/dative case homonymy, which it inherits from Latin [.] The homonymy in case marking is, however, only partial'. Despite the objective uncertainty about the origins of this phenomenon, some structural aspects regarding the expression of genitive and dative in Romanian are undeniable. For example, even for inflected nouns genitive--and only genitive--requires the presence of the nominal (elsewhere 'genitival article' or 'possessive article') AL, (16) which is an obligatory constituent of GenP if the specified NP is indefinite (irrespective of the definiteness/indefiniteness of the genitive), as shown in (9):
(9) a. un caiet al Mariei b. *un caiet Mariei 'A copybook of Mary's'
c. o descoperire a acestui om d. *o descoperire acestui om 'A discovery of this man'
e. alte greseli ale tuturor parinfilor f. *alte greceli tuturor parin^ilor other mistakes of all the parents'
These examples, whose interpretation cannot be dative, confirm Cornilescu's observation that dative and genitive do not equal each other perfectly from a syntactic point of view, although it is undeniable that the genitive assigning element (whatever it may be defined) does not influence the form of these two cases, which are always expressed in the same way. The status of AL is extremely difficult to identify: its prepositional value (supposable from its form, similar to prepositions in other Romance languages, and from the apparent parallelism between Rom. Casa este a vecinului 'the house is the neighbour's' and French La maison, c 'est a mon pere) can be easily contradicted through examples such as Trei studente de ale Mariei 'Two students (fem.) of Mary's', where the preposition de, clearly related to genitivity, and the agreeing element AL co-occur. On the other hand, it is not automatically ascribable to articles because in cases like u[section]ile a doua biserici 'the doors of two churches', where the genitival noun is [-specific], the definite article would not be needed. The syntactic behaviour of this element, which crucially agrees with the gender and number features of the noun modified by the genitival phrase, (17) is not easily explainable: the assumption that it appears as 'inflected', i.e. as a + article-like specifications, when the immediately preceding noun is indefinite is contradicted by the (9a-c) contrast. Similarly, a plausible expectation would be that it might appear as non-inflected after a definite noun, an expectation which is denied by the fact that when the modified noun is definite and immediately precedes the genitival phrase, a does not even appear (see example (10)), differently from a context such as that considered by Cornilescu (see below), in which an adjectival element occurs between the modified noun and the genitival phrase. I will assume that the definition of 'genitive assigning nominal' is the most appropriate one among all theoretical proposals, although not fully convincing. AL is, indeed, made up of an invariable element a, followed by the definite article, as shown in Table 2 (taken from Dobrovie-Sorin):
Table 2 Definite vs. possessive article Encl. def. art. 'poss. article' m. sg. -l, -le Al f. sg. -a a m.pl. -i ai f.pl. -le ale
Nevertheless, if AL were a nominal, it should be endowed with a semantic value which is hardly recognizable. Similarly, the function of assigning genitive does not seem definable as a compelling semantic value. We may hypothesize that AL bears a special status that is not shared by any other element. In the first place, its properties are not completely consistent with any of the existing categories: it is not an article because its use is not necessarily related to the expression of definiteness of any of the elements of the DP including the modifier and the modified item, it is not a preposition because it may co-occur with other prepositions and its nominal status does not appear to be convincing. As discussed later on for Albanian, the distribution of AL resembles that of a demonstrative: according to the data I analyzed for Standard Romanian, its presence triggers ungrammaticality when the preceding noun is definite, whilst it is obligatory when the modified NP is indefinite (except for deverbal nouns, after which this element appears as 'non-inflected'); moreover, AL is required when an AP occurs between the noun and the genitival phrase: in this context the function of AL may be hypothesized to be 'resumptive', i.e. to re-establish a connection of definiteness between the noun and the genitive, given that this accord has been 'interrupted' by the occurrence of the adjectival element. Therefore, the example in (10) represents 'the only exceptional context' (Cornilescu 1995) in which the noun theta-marking the genitive bears the definite article and immediately precedes it. In that context (i.e. in a context where there is no element between the genitive and the definite article, assuming that the 'possessive article' bears a suffixal definite article -l,-a,-i,-le) the insertion of AL triggers ungrammaticality:
(10) a. *prietenul al acestui copil b. prietenul acestui copil 'The friend of this child'
(11) Profesoara a DAT sfaturi tuturor parinfilor Teacher-the gave advice all-DAT parents-the-DAT 'The teacher gave advice to all the parents'
A comparison of the data in (9), (10) and (11) suggests that the presence of AL is obligatory in the expression of genitive, considering the discussed implications, and that the syntactic distribution of genitive and dative is not the same. Furthermore, the distribution (as well as the categorization) of AL is puzzling because it can also be used as an obligatory accompanying element for the possessive pronoun (which seems to confirm the view that AL conveys features that are demonstrative-like, since even in these contexts it agrees with the preceding noun and apparently has a resumptive function of the type 'the father of him is doctor, AL [=the father] of mine is professor'):
(12) a. Tatal lui este medic, al meu este profesor father-the he-gen is doctor al mine is professor 'His father is a doctor, mine is a teacher'
b. Parinfii tai merg la cinema, ai mei la teatru Parents-the your-m.pl. go to cinema al mine to theatre 'Your parents are going to the cinema, mine to the theatre'
Consider Cornilescu's (1995: 24-25) analysis of a DP like elevele bune ale acestui profesor, in which an AP occurs between the definite article and the genitive. The presence of AL is obligatory in this case, as it assigns case under strict c-command and adjacency. The opposite goes for a DP where no adjective appears between these two elements (e.g. elevele acestui profesor). This makes sense in that, as noted by Cornilescu, 'all the features of the Gen assigning article are also features of the definite article, which is hardly surprising, since both agree with the same N'. As a consequence, we can hypothesize that the features of AL incorporate those of the definite article, so that AL is no longer phonologically realized by virtue of the fact that the syntactic position of the definite article, with respect to that of the genitival DP, satisfies both the locality condition and the c-command requirement.
As is clear, Romanian draws a clear distinction between inflectional marking and prepositional marking. Interestingly enough, dative is realized inflectionally on [+M-Case] NPs, whilst [-M-Case] NPs take the preposition la + accusative. This element is possibly less problematic than AL, since its status is substantially prepositive and it always assigns accusative, irrespective of its semantic value and of the definiteness of the noun it governs (cf. Ma duc la piafa 'I am going downtown', Paru-i ajungea la calcaie 'He had heel-long hair', lit. 'His hair was long to the heels', etc.):
(13) a. A DAT bomboane la niste copii / lapufini copii has given sweets-acc la some children-acc / la few-children-acc 'S/he has given sweets to some children /to few-children'
b. Am recurs la un prieten have-1-ps.-sg. resorted la a friend-acc '(I) resorted to a friend'
c. Am recurs la prietenul have-1-ps.-sg. resorted la friend-the-acc '(I) resorted to the friend'
Some quantifier constructions allow both declensions: one with bare dative and one with la, bearing no semantic implications:
(14) a. (Le)-a vorbit multor studenfi they-cl-DAT has talked many-students-DAT 'He talked to many students'
b. (Le)-a vorbit la mulfi studenfi / la cafiva studenfi they-cl-DAT has talked la many-students-acc 'He talked to many students (18)'
The morphological expression of dative and genitive in Romanian may therefore be said to display evident syncretism in nominals: despite the syntactic differences in their use and the fact that genitive and dative are not always expressed in the same way, nominal morphology seems to be consistently syncretic as far as these two cases are concerned.
5. Bulgarian and Macedonian
Bulgarian and Macedonian are often treated together in that the latter is generally considered as a dialect of the former. Moreover, in the Balkan perspective, Bulgarian and Macedonian are the only two Slavic languages which are ascribed to the so-called 'Balkan languages of first degree'. For the sake of convenience, I will also consider them together, aware nevertheless of the fact that these two languages do not coincide perfectly--e.g. some phenomena related to clitic doubling are not shared by both languages--and that, as noted by Tomic (1991: 449):
Macedonian is structurally related to Bulgarian more than to any other South Slavic languages. But the core of its standard was not formed out of dialects or variants that had ever been covered by the Bulgarian standard. Consequently, its autonomy could not have resulted from a conscious distancing of a variant of a pluricentric language. Like the other South Slavic languages, the Macedonian standard was based on dialects which had never before been covered by a standard.
Sussex and Cubberley's (2006: 229 ff.) analysis, which does not seem to establish a clear relation between case syncretism in Bulgarian/Macedonian (which have 'eradicated all but a skeleton of their original case systems') and the same, generalized phenomenon in the Balkan Sprachbund, considers the development of the case system in these two languages--which are caseless, differently from all other Slavic languages--as an 'extreme case' of syncretism, in the sense that this phenomenon has been traced back to Proto-Slavic and includes all modern Slavic languages as a general means of simplification in the expression of grammatical relationships. In fact, syncretism between genitive and dative is also present, for instance, in Russian and in Czech, as exemplified in Table 3:
Table 3 Syncretism among non core cases in Czech "bone" "castle" Gen kosti hradu Loc kosti hrade Dat kosti hradu Instr kosti hradem
The caseless status of Bulgarian and Macedonian, as is the rule, has triggered analyticity in the nominal systems of these languages, which redress the lack of morphological cases by using simple apposition (Bulg. litar mijako vs. Rus. litr moloka [GenSg] 'a litre of milk') or through the use of prepositions. This is not a particularly innovative fact of language, as this substitution phenomenon is quite widespread cross-linguistically: suffice it to think of ablative case in Latin, which in its development in the Romance languages has been lost and whose function has substantially been substituted through prepositions such as It. con, Fr. avec, etc., as well as through a lot of other prepositions, depending on the type of relation expressed. Comparing the examples in (15), it becomes clear how the use of prepositions as a last resort strategy for the expression of given grammatical relationships previously expressed by morphological case does not represent a problem:
(15) a. Rus. mat' gorditsja svoim trudom mother proud her work-INSTR.
b. Bulg. majka se gordee sas svoja trud mother is proud with her work 'Mother is proud of her work'
Interestingly enough, in Bulgarian the originally dative preposition na is used to express genitive, as noted in (8) and reported here for the sake of completeness:
(16) kastata na starikat (mu) rece na starikat house na old man-gen DAT said na old man 'The house of the man' 'S/He said it to the man'
As observed by Junghanns/Lenertova (forthcoming: 2), na has traditionally been taken to be a preposition in standard literature, while generative grammar has often avoided the problem of the formal status of this syntactic element, whose semantics, just as in similar cases in the other Balkan languages, roughly corresponds to 'at', 'in'. The homonymy in case-marking seems to be complete in Bulgarian if we take na as the functional element used for the expression of both genitive and dative. I argue nonetheless that, given the non-innovative character of the use of one preposition for more functions (cf. Fr. J'ai lave les cheveux a Marie vs. J'ai donne de l'argent a Marie or even It. Ho comprato un manuale di linguistica vs. Ho detto a Maria di andarsene) and in the light of the fact that e.g. in Romanian and Bulgarian the strategies for the expression of these two cases are different, genitive-dative syncretism is not to be considered as exclusively belonging to the Balkan Sprachbund. A comparison between (17a.) and (17b.) makes this distinction clear:
(17) a. Rom. I-am DAT cartea lui Marian I gave book-the to Marian
b. Bulg. Petar dade knigata na Ivan Peter gave book-the to Ivan
In the first place, lui is doubtlessly a crystallized form of an inflected article of Latin origin (indeed, it is also used with feminine nouns if they do not bear a typically feminine ending, e.g. lui Carmen). Not only is Bulg. na not the same element (structurally, as well as etymologically), it also has completely different features as opposed to the strategies found in Romanian. Morphological case is substantially absent in Bulgarian nominal expressions, except for some remnant of vocative and instrumental in fixed expressions and in particular cases, so that na becomes the only possible device for the expression of genitive-dative, differently from Romanian, which still draws a formal distinction between the two). Rom. la seems to be more similar to Bulgarian na, although its use is not the same. By the way, judgments by native speakers consistently reveal that despite the tendency that Romanian displays towards a replacement of inflected forms by prepositional ones, a structure like Cartea Mariei can actively be substituted through Cartea lui Maria even if Maria is clearly feminine, but not for example through Cartea la Maria. Following Junghanns/Lenertova's (2009: 157 ff.) collection of the existing analyses of na, we notice that this element has been subjected to various interpretations: i.e. it has been analyzed as a phrasal affix, as a functional head K and as a semantically vacuous preposition. The analysis of na as a phrasal affix is apparently corroborated by Chomsky's 'rule of Of-insertion' and by the fact that, na, as an inflectional affix, is not typed semantically. Nonetheless, in this case we should assume the existence of a na-preposition (19) and of a na-affix occurring parallelly and independently, and this would seem too strong a claim since, as pointed out by Junghanns/Lenertova, it would be necessary to specify the features of this phrasal affix in more detail, which is not easy, given the pervasiveness of its use. As regards the 'functional head analysis', Toman (1994: 175) proposes that na is, in fact, a preposition, but that it is Kase, (20) as in the following representation (taken from Junghanns/Lenertova):
As is well-known, Bulgarian has retained inflectional morphology in the pronominal system, so that Toman's proposal is evidently contradicted by the fact that na-marked elements are assigned accusative case. This implication is of course not visible in nominal expressions, but a comparison between the use of na with a full nominal and with a pronoun, respectively, clarifies the non-validity of this claim:
(18) a. Petar dade knigata na Ivan Peter gave book-the to Ivan
b. Petar dade knigata na nego Peter gave book-the to him
By the way, the fact that na assigns accusative is also confirmed by other Slavic languages such as Czech and Russian, where the na-element is present (cf. for example Cz. napfed, na modro, etc.). What is more, Macedonian, a language with more morphology than Bulgarian, shows that even full nominals like proper names are accusative-marked when preceded by na:
(19) a. Petre mu ja dade knigata na Ivana / nemu Peter him-DAT-CL she-acc-CL gave book-the to Ivan-acc him-DAT 'Peter gave the book to Ivan'
Junghanns/Lenertova observe that this view may be supported by considering the whole KP as abstract dative case functioning as an indirect object whose projection is invisible at LF. In fact, this assumption can also be applied to Rom. la. However, although such abstraction may be interesting from a theoretical point of view, it clearly raises two main problems: in the first place, if the representation of a phrase like Macedonian na Ivana were as in (20):
(20) Mac. [KP [k [k[degrees] na [dp Ivan-a] ]]]
then the presence of accusative marking on the complement DP would be inexplicable and difficult to account for. Secondly, if we take na as K[degrees], we have to postulate that this element occurs both as a preposition (e.g. Knigata e na masata 'The book is on the table') and as K[degrees], which does not seem to be compelling. Instinctively, the question arises as to whether K[degrees]-na may be seen as a grammaticalized form of the preposition na, a phenomenon that is attested in some languages (for example in Armenian) and that would be plausible, as grammaticalization processes often retain the original function of the form subjected to the development (cf. the grammaticalization of the be going to construction as a future marker in English). Nonetheless Beukema/den Dikken (2000: 136) note that na, irrespective of its Definition (21), is structurally a non-bound item, i.e. it occurs in contexts like (21):
(21) Bulg. [[PP na [dp Ivan i Maria]]...] kusta to Ivan and Maria house 'To Ivan and Maria's house'
This aspect is crucial, as it shows that na as a K[degrees] should be assumed to have the significant feature of optionally being non-adjacent to its DP. The KP solution seems therefore not to be adequate for an explanation of the na-phrase. A third hypothesis, supported, among others, by Heim & Kratzer (1998), is that na is a 'semantically vacuous preposition' that projects a syntactic structure. I will assume this premise (as represented below) to be partly correct for a number of reasons: first of all, instinctively the syntactic behavior of this element resembles that of prepositions in other languages (cf. Bulg. Knigata na Ivan, knigata na nego vs. Germ. Das Buch von Peter, das Buch von ihm, despite the difference in accusative/dative case marking). Moreover, the recognition of its prepositional status does not pose any problem from a distributional--as well as economical--point of view, i.e. no double categorial classification is implied. This postulate would also bypass the obstacle represented by the optional presence of other elements between the preposition and the DP.
The proposal that na is a semantically vacuous preposition (cf. Heim & Kratz 1998), however, does not seem to be appropriate for the description of this item. The assumption of this empty status, apparently motivated by a grammaticalization process, is basically useless. Na is a dative preposition whose primary meaning is locative and the fact that it originally only indicated dative (and therefore often a benefactive theta-role) with its complement DP appears to be motivated and confirmed typologically: suffice it to think of It. andare a casa (andative-based meaning) vs. dare un libro a qualcuno (dative) vs. lavare i capelli a Maria ('possessive', in the sense discussed above). The exclusively dative function of the na-PP later absorbed genitive, but this development does not seem to have modified the semantics of na, whose 'to'-meaning is consistent with a benefactive function. Hence I argue that na is not semantically empty: on the contrary, it has retained its original meaning by specializing itself in the expression of dative (and later also of genitive), as is the case in many other languages: cf. Eng. I am going to the beach vs. I gave the book to Peter, Germ. Ich bin zu Hause vs. Ich habe meine Hauskatze zu meiner Mutter gegeben, Fr. Je vais a la maison vs. J'ai donne ma poche a ta cousine, etc.. The pronominal system of Bulgarian is consistent with genitive/dative syncretism, as can be seen in Table 4:
Table 4 The pronominal system in Bulgarian (22) INDIRECT OBJECT no preposition with DATIVE preposition full short 1. p.s. mene mi men 2 p.s. tebe ti teb 3 p.s. m. nemu mu nego 3 p.s. f. nej i neja 3 p.s. n. nemu mu nego 1 p. p. nam ni nas 2 p. p. vam vi vas 3 p. p. tjam im tjah
The pronouns indicated as appearing with a preposition refer of course to the na + accusative type, which reveals that, as is often the case, the pronominal system of Bulgarian has not undergone complete syncretism but has preserved as distinction between nominative, accusative and dative, where the latter is also used to indicate genitive.
As already observed, Greek seems to display case syncretism of the type Dat > Gen, since this language has lost the morphological dative case of Ancient Greek, which has been absorbed by genitive. As can be seen in the following examples (Dimitriadis 1999: 97), formerly dative structures are formed with the preposition se (roughly bearing the meaning of 'to', 'into', 'at'), followed by an accusative NP (23), just as na in Bulgarian and la in Romanian:
(22) a. Edosa to vivlia sto Yorgho gave-1-sg the books to-the George-acc 'I gave the books to George'
b. Eghrapsa sti Maria wrote-1-sg to-the Maria-acc 'I wrote to Mary'
In fact, this construction, often defined as a 'periphrastic dative', is mainly formed by se + accusative case, but it also occurs with a number of other prepositions (e.g. apo 'from' and me 'with'), depending on the verb. As for 'dative/genitive' elements in other Balkan languages, the actual status of se in Greek (i.e. whether it is to be considered as a real case assigner or 'a reflex of dative morphology') is not resolved. Leaving the question open, I propose that the same goes for languages like English, where the much-discussed notion of 'indirect object' (roughly indicating a 'benefactive/goal dative') is sometimes expressed not by the canonical preposition to, but by other prepositions, possibly as a last resort strategy, given the ungrammaticality of constructions like (23c). For example, the following structures show that the same theta-role (that is, the same semantic value) is expressed by different prepositions:
(23) a. John made me / Mary a cake
b. He made a cake for me / for Mary
c. *He made a cake to me / to Mary
As underlined by Dimitriadis (1999: 95 ff.), in all southern dialects of Greek a synonymous construction with indirect objects--both full nominals and clitics, as well as indirect objects in clitic doubling contexts--marked by morphological genitive occurs. In particular, in clitic doubling constructions both the case of the clitic and that of the nominal must be genitive (note the contrast in 24d.-24e.):
(24) a. Edosa tou Yorghou ta vivlia Gave.1-sg the George-gen the books-acc 'I gave George the books'
b. Egraphsa tis Marias Wrote-1-sg the-Maria-gen 'I wrote to Maria'
c. Tou edosa ta vivlia He-cl.gen gave-1-sg. the-books-acc 'I gave him the books'
d. Tou edosa ta vivlia tou Yorghou He-cl-gen gave-1-sg the-books-acc the-George-gen 'I gave George the books'
e. * Tou edosa ta vivlia sto Yorgho He-cl-gen gave-1-sg the-books-acc to-the-George-acc
Besides revealing that genitive/dative is consistently expressed in nominal expressions and clitic doubling contexts only in some varieties of Modern Greek, the data in (24), especially (24d)-(24e) are interesting because they clearly contradict the validity of Kayne's generalization on clitic doubling, requiring that 'doubled elements must appear affixed by a morpheme (dative-looking in Spanish, the object markerpe in Romanian)' (Sportiche 1996: 222). Indeed, as we observe in (24), no preposition or generic case marker is required in southern Greek dialects. On the contrary, the presence of a prepositional element blocks clitic doubling, i.e. this phenomenon does not take place when the IO is a PP.
7. A Brief Overview of Albanian: The Last Straw?
Interestingly enough, Spencer (2007: 219 ff.) calls into question the existence of a genitive case in Albanian, substantially basing on Koptjevskaja-Tamm's research on Bantu languages. In fact, in Albanian the genitive construction implies agreement between the possessor and the possessum in exactly the same way as in a number of Bantu languages and in Hindi-Urdu. As insisted on by Turano (2005: 195), in Albanian the element preceding the genitival phrase, which is generally assigned the controversial status of 'article', displays the same gender and number features as the head noun. The balkanism, as it is normally intended in Balkan linguistics, substantially consists in the morphological identity of the case inflection attached to the noun indicating the possessor, as we can observe in (25) [taken, and slightly modified, from Turano 2005: 195-196]:
(25) a. libri i studentit book-the-m-sg. 'the'-m-sg. student-m.sg.-gen 'The student's book'
b. canta e studentit bag-the-f-sg 'the'-f-sg. student-m-sg.-gen 'The student's bag'
c. *libri studentit book-the-m-sg. student-the-m-sg.-gen
Crucially, this 'genitival article' (24), whose presence is obligatory, agrees in number and gender with the preceding noun (as the AL element in Romanian, despite clear differences in their syntactic distribution, for example as far as the [in]definiteness of the preceding item is concerned). This aspect is significant for two reasons: in the first place, as I argued for Romanian, it indicates an asymmetry between the preceding noun and the genitival phrase in that the 'article' displays a much stronger relation to the former than to the latter--which is not the case in the other languages of the Balkan Sprachbund (except Romanian) and is not so widely diffused from a typological perspective. Nonetheless, the fact that this phenomenon is present in languages like Hindi-Urdu shows that its origin may be Indo-European. Secondly, this controversial element is the same found in the AP, which suggests that a demonstrative-like feature may be attributed to this 'article'. Observing the following examples [taken from Turano 2005: 169], it is simple to hypothesize that a correlation between the element in the AP and that in the GenP (if we assume that it structurally belongs to the GenP) is definitely present:
(26) a. libri i book-the 'the' his 'his book'
b. libri i studentit book-the 'the' student-gen 'the student's book'
c. libri i kuq book-the 'the' red 'the red book'
These three instances of the use of this element reveal that i has a special function: it seems to particularize the noun it follows by restricting the semantic domain ('the book, precisely HIS book' < 'that of him', 'that book which belongs to him'; the same goes for the other cases). The genitival phrase in Albanian may also be introduced by other elements (basically te and se, whose occurrence in Albanian is predictably not limited to the GenP), as we see in Table 5 (the declension of the masculine noun mal is taken as an example: the morphology for dative and genitive is equally the same in the other declensions):
Table 5 The nominal case system in Albanian Indefinite Singular Indefinite Plural Nominative nje mal (a mountain) male (mountains) Accusative nje mal male Genitive i/e/te/se nje mali i/e/te/se maleve Dative nje mali maleve Ablative nje mali malesh Definite Singular Definite Plural Nominative mali (the mountain) malet (the mountains) Accusative malin malet Genitive i/e/te/se malit i/e/te/se maleve Dative malit maleve Ablative malit maleve
Considering the paradigm of nominal inflection in Albanian (a language which, according to most grammars, has retained six cases: nominative, genitive, dative, accusative, ablative and vocative (25) + locative in some dialects), we cannot but note that the morphological noun marking for genitive and dative is the same, although, again, the syntactic strategy of expression of these two cases is different (that is probably why they are officially kept separate in normative grammars). This idea, which has been particularly insisted on throughout this paper, also represents the pivot of my discussion: pragmatically, genitive and dative (except for their morphological endings) do not coincide perfectly in Albanian. Recognizing such a controversial fact does not correspond to an underestimation of the value or of the function of case morphology, but is objective, (26) which becomes clear, for instance, by observing the following examples (for the sake of clarity, I indicated the case morphology on Marise in a. and b. as dative and genitive, respectively):
(27) a. Ja dhashe librin Marise It-cl-acc gave-1-sg. book-the-acc Mary-DAT 'I gave the book to Mary'
b. Libri eshte i Marise book-the-nom is 'i' Mary-gen 'The book is Mary's'
Notice that, interestingly, Spencer (2007: 221) defines the i element present in the genitival construction either as a morphologized clitic--or as a weakly adjoined affix--allowing a NP to be used attributively. The fact that a significant structural parallelism exists between the genitival phrase and the adjectival phrase in Albanian, however, does not seem to be particularly striking if we consider that from a purely theoretical point of view they accomplish the same function.
In this paper I provocatively asserted that genitive/dative syncretism, a phenomenon which is present in Romanian, Albanian, Bulgarian, Macedonian and Greek, is not necessarily to be related to the phenomena ascribed to the Balkan Sprachbund, primarily because it appears to be a pervasive fact of language, found in a number of different languages (within the Indo-European family, e.g. in Russian and Armenian, but also in Australian and in Dravidian languages (27)); in the second place, my strong claim was that, although the case morphology for dative and genitive in the Balkan languages coincides in nominals, the syntactic expression of the functions conveyed by these two cases, except for occasional instances, is basically different in the Balkan Sprachbund. Summing up, we may conclude that:
(i) As regards Romanian, in particular contexts the element AL is required for the expression of genitive (and not of dative), which is followed by morphological genitive; the NPs defined by Cornilescu (1995) as [-M-Case], instead, take the accusative-assigning preposition LA (whose status is definitely less controversial than that of AL), whereas some quantifier constructions allow both a bare genitive structure and a LA[+Acc]-structure;
(ii) In Bulgarian/Macedonian, which are caseless languages, the originally dative preposition na ((see (3.) for a discussion on its status)) is utilized to express both the dative and the genitive phrase. An analysis of Macedonian, a language that has retained more case morphology than Bulgarian, reveals that the complement of na is marked by accusative. The expression of genitive/dative in these two languages does not imply the presence of endings (except for rare contexts in Macedonian) and the pronominal system shows that [+wh]-phrases such as komy/na kogo (dative) and cni (genitive) still preserves formal differences. The use of na can be explained as a last resort strategy due to the loss of case morphology (for example, in German the preposition von, among a wide range of functions, serves as a means for the expression of agents, of the origin and of genitival phrases, cf. Jeder Verstofi wird von der Polizei zivilrechtlich verfolgt vs. Alles Gute kommt von unten vs. Die Frau von meinem Vater ist echt eiskalt!).
(iii) In Greek, the genitive/dative homonymy, expressed through morphological genitive, displays the same level of consistency as in the other languages. In Standard Greek genitive/dative phrases are made up of the preposition se + accusative, although Southern dialects show a tendency to also use a bare genitive construction.
(iv) The genitive (but not dative) phrase in Albanian always displays the to-be-defined element (possibly a demonstrative-like item) i/e, which agrees in gender and number with the preceding noun and is also present in the AP. Dative bears the same morphology as genitive, which is also partially shared by ablative, as is the case in a number of other languages.
The crucial point is that case syncretism is an extremely widespread phenomenon. The strategies adopted by the languages belonging to the Balkan Sprachbund for genitive/dative constructions are substantially different and have diverse--and still mostly unexplained--origins. What is more, according to Urban (2007: 151), two of the main criteria identifying a linguistic league are that we have to be able to prove that the similarities are due to borrowing and that the core languages of the Sprachbund have to be clearly distinguishable, which is not the case as far as genitive/dative syncretism is concerned.
ABL = ablative
ACC = accusative
Bulg. = Bulgarian
CL = clitic
DAT = dative
Def = definite
Eng. = English
GEN = genitive
Germ. = German
Gr. = Greek
IO = indirect object
It. = Italian
KP = Kase Phrase
NOM = nominative
pl. = plural
ps. = person
Rom. = Romanian
Rus. = Russian
sg. = singular
Sp. = Spanish
VOC = vocative
ALEXIADOU, Artemis, MULLER, Gereon. 2004. Class Features as Probes. Ms., University of Stuttgart-Mannheim; available at <http://www.uni-leipzig.de/~muellerg/mu55.pdf>.
ALEXIADOU, Artemis, HAEGEMAN, Liliane, STAVROU, Melita. (2007). Noun Phrase in the Generative Perspective. Berlin: Mouton de Gruyter, 2007.
ALLEN, Cynthia L. 1995. Case Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Oxford University Press, 1995.
AMMANN, Andreas, VAN DER AUWERA, Johan. 2004. Complementizer-Headed Main Clauses for Volitional Moods in the Languages of South-Eastern Europe: A Balkanism?. In TOMlC, O. Balkan Syntax and Semantics. Amsterdam: Benjamins, 2004, pp. 293-314.
BAERMAN, Matthew. 2009. Case Syncretism. In MALCHUKOV, A, SPENCER, A. The Oxford Handbook of Case. Oxford: Oxford University Press, 2009, pp. 219-230.
BANFI, Emanuele. 1985. Linguistica balcanica. Bologna: Zanichelli, 1985.
BEUKEMA, Frits, DEN DIKKEN, Marcel. 2000. Clitic Phenomena in European Languages. Amsterdam: John Benjamins, 2000.
BLAKE, Barry J. 1994. Case. Cambridge: Cambridge University Press, 1994.
COMRIE, Bernard. 1991. Form and Function in Identifying Cases. In PLANK, F. Paradigms. The Economy of Inflection. Berlin/New York: Mouton de Gruyter, 1991, pp. 41-55.
CORNILESCU, Alexandra. 1995. Rumanian Genitive Constructions. In CINQUE, G., GIUSTI, G. Advances in Roumanian Linguistics. Amsterdam: Benjamins, 1995, pp. 1-54.
DANILIUC, Laura, DANILIUC, Radu. 2000. Descriptive Romanian Grammar--An Outline. Munchen: Lincom Europa, 2000.
DE GROOT, Albert W. 1956. Classification of Cases and Uses of Cases. In HALLE, M., LUNT, H., McLEAN, H., SCHOONEVELD, C. For Roman Jakobson. Essays on the Occasion of His Sixtieth Birthday. The Hague: Mouton de Gruyter, 1956, pp. 187-194.
DIMITRIADIS, Alexis. 1999. Clitics, Prepositions and Case Licensing. In ALEXIADOU, A., HORROCKS, G. & STAVROU, M. Studies in Greek Syntax. Dordrecht: Kluwer, 1999, pp. 95-112.
DIMITROVA-VULCHANOVA, Mila, GIUSTI, Giuliana. 1999. Possessors in the Bulgarian DP. In DIMITROVA-VULCHANOVA, M., HELLAN, L. Topics in South Slavic Syntax and Semantics. Amsterdam/Philadelphia: John Benjamins, 1999, pp. 163-192.
DOBROVIE-SORIN, Carmen. (in press). Genitive DPs and Pronominal Possessors. Universite de Paris-Diderot.
FISIAK, Jacek. 1988. Historical Dialectology: Regional and Social. Berlin: Mouton de Gruyter, 1988.
GIUSTI, Giuliana. 2002. The Functional Structure of Noun Phrases. In CINQUE, G. Functional Structure in DP and IP: The Cartography of Syntactic Structures, Vol. 1. Oxford: Oxford University Press, 2002, pp. 54-90.
GOLDBLATT, Harvey. 1984. The Language Question and the Emergence of Slavic National Languages. In SCAGLIONE, A. The Emergence of National Languages. Ravenna: Longo Editore, 1984, pp.119-173.
HARIZANOV, Boris. 2011. Non-Initiality within Spellout Domains: Unifying the Post-Syntactic Behaviour of Bulgarian Dative Clitics. In LACARA, N., THOMPSON, A., TUCKER, M.A.. Morphology at Santa Cruz: Papers in Honor of Jorge Hankamer. UC Santa Cruz: University of California, Linguistics Research Center, 2011, pp. 1-30.
HEIM, Irene, KRATZER, Angelika. 1998. Semantics in Generative Grammar. Malden, MA: Blackwell, 1998.
HEINE, Bernd, KUTEVA, Tania. 2005. Language Contact and Grammatical Change. Oxford: Oxford University Press, 2005.
JOSEPH, Brian D. 2000. Processes of Spread for Syntactic Constructions in the Balkans. In TZIZILIS, C., SYMEODINIS, C. Balkan Linguistik: Synchronie und Diachronie. Thessaloniki: University of Thessaloniki, 2000, pp. 139-150.
JUNG, Yeun-Jin., MIYAGAWA, Shigeru. 2004. Decomposing Ditransitive Verbs. In JUNG, Y, MIYAGAWA, S., Proceedings of SICGG, Summer 2004, pp. 101-120.
JUNGHANNS, Uwe, LENERTOVA, Denisa. 2009. On the Status of 'na' Marking Indirect Objects in Bulgarian. In ZYBATOW, G. et al. Studies in Formal Slavic Phonology, Morphology, Syntax, Semantics and Information Structure. Proceedings of FDSL-7, Leipzig 2007 (= Linguistik International; 21). Frankfurt am Main: Peter Lang, 2009, pp. 155-166.
KALLULLI, Dalina, TASMOWSKI, Liliane. 2008. Cliting Doubling in the Balkan Languages. Amsterdam/Philadelphia: John Benjamins, 2008.
KAYNE, Richard. 1984. Datives in French and English. In KAYNE, R., Connectedness and Binary Branching. Dordrecht: Foris, 1984, pp. 193-202.
LAMONTAGNE, Greg, TRAVIS, Lisa. 1987. The Syntax of Adjacency. Proceedings of the 6th West Coast Conference on Formal Linguistics, 1987, vol. 6, pp. 173-186.
LARSON, Richard K. 1988. On the Double Object Construction. In Linguistic Inquiry, 1988, vol. 19, no. 3, pp. 335-391.
MANZINI, Maria R., SAVOIA, Leonardo M. (unpublished). Case Inflections as Denotational Elements Syncretisms as Ambiguity: In Defence of the Lexicon. Universita di Firenze.
MATRAS, Yaron. 2002. Romani: A Linguistic Introduction. Cambridge: Cambridge University Press, 2002.
NESS, Ashild. 2009. Varieties of Dative. In MALCHUKOV, A., SPENCER, A. The Oxford Handbook of Case. Oxford: Oxford University Press, 2009, pp. 572-580.
PANCHEVA, Roumyana. 2004. Balkan Possessive Clitics. In TOMlC, O. Balkan Syntax and Semantics. Amsterdam: John Benjamins, 2004, pp. 175-219.
PETRUCCI, Peter P. 1999. Slavic Features in the History of Romanian. Munchen: Lincolm Europa, 1999.
REICHENKRON, Gunter. 1962. Der Typus der Balkansprachen. In Zeitschrift fur Balkanologie, 1962, vol. 1, pp. 91-122.
RIVERO, Maria-Luisa, RALLI, Angela. 2001. Comparative Syntax of the Balkan Languages. Oxford: Oxford University Press, 2001.
SANDFELD, Kristian. 1930. Linguistique Balkanique: problemes e resultats. Paris: Champion, 1930.
SCHALLER, Helmut W. 1975. Die Balkansprachen. Eine Einfuhrung in die Balkanphilologie. Heidelberg: Carl Winter Universitatsverlag, 1975.
SPENCER, Andrew. 2007. The Possessum-Agreement Construction or 'Does Albanian Have a Genitive Case?'. In Martin Atkinson - The Minimalist Muse, 2007, vol. 53, pp. 219-250.
SPORTICHE, Dominique. 1996. Clitic Constructions. In ROORYK J., ZARING L. Phrase Structure and the Lexicon. Dordrecht: Kluwer, 1996, pp. 213-276.
SUSSEX, Roland, CUBBERLEY, Paul. 2006. The Slavic Languages. Oxford: Oxford University Press, 2006.
TOMAN, Jindrich. 1994. Case as a Functional Projection: A Note on an Issue in Parametrization. In TOMAN, J. Formal Approaches to Slavic Linguistics, 1.. Ann Arbor: Michigan Slavic Publications, 1994, pp.173-181.
TOMIC, Olga. 1992. Macedonian as an Ausbau Language. In CLYNE, M. Pluricentric Languages: Differing Norms in Differing Nations. Berlin: Mouton de Gruyter, 1992, pp. 437-454.
TOMlC, Olga. 2006. Balkan SprachbundMorphosyntactic Features. Dordrecht: Springer, 2006.
TOSO, Fiorenzo. 2006. Lingue d'Europa-La pluralite linguistica dei paesi europei tra passato e presente. Milano: Baldini Castoldi Dalai Editore, 2006.
TURANO, Giuseppina. 2002. On Modifiers Preceded by the Article in Albanian DPs. University of Venice Working Papers in Linguistics, 2002, no. 12, pp. 169-215.
URBAN, Mateusz. 2007. Defining the Linguistic Area/League: An Invitation to Discussion. Studia Linguistica Universitatis Iagellonicae Cracoviensis, 2007 vol. 124, 137-159.
(1) For an exhaustive discussion of clitic doubling cf. Kallulli/Tasmowski (2008, 80 ff.).
(2) In this paper I argue that in fact the presence of a given feature in a number of languages is not necessarily ascribable to a language family or to a Sprachbund: structures like Mod. Gr. Na zisete! 'May you live (a long life)' and Rom. Sa ne vedem sanatosi! 'May we see each other healthy!', displaying 'complementizer-headed main clauses' are present, for instance, both in Italian and Spanish (Che mi venga un colpo!; Que se vayan todos!).
(3) In the present paper I will basically follow Schaller's (1975) subdivision of the Balkan languages, according to which Albanian, Bulgarian, Romanian and Macedonian are "primary" Balkan languages, Serbian and Greek are "secondary" Balkan languages, while Turkish is a Balkan language of third degree.
(4) A further distinction to be drawn is that between Balkan languages (Balkansprachen), i.e. the languages sharing these phenomena and the general definition of 'languages of the Balkans' (Sprachen des Balkans), which obviously corresponds to the geographical disposition of the countries, irrespective of linguistic implications.
(5) Following Comrie (1991: 44), the identification of case functions depends on a clear correlation between morphosyntactic case and a formal inflectional distinction.
(6) In fact, in many languages dative and ablative (which is already the result of a process of syncretism with locative) undergo syncretism. By expression of 'metaphorical state' we mean the occurrence of the morphological marking associated with locative contexts in cases in which the location is not necessarily physical, but rather assumed or implicit (the real semantic value of the example from Germ. Ich bin in der Kuche sehr eitel 'I am very vain in the kitchen' does not imply the subject's physical presence in a kitchen, but refers to the concept of 'I am very vain when it comes to cooking').
(7) The notion of locality assumed in Blake's work is very general and includes accusative (a morphological marker of the target of motion) and ablative (indicating origin or static location).
(8) According to Allen (1995: 158) this syncretism triggered the emergence of a new type of passive (cf. The king was given a gift, where the subject would have been clearly dative in OE).
(9) Cf. Banfi's (1985) interpretation in (3).
(10) At least from Kayne (1984) the dative argument of this class of verbs has been linked to possessives (cf. R. Kayne, Datives in French and English, in: Connectedness and Binary Branching, R. Kayne (ed.), Dordrecht: Foris), pp. 193-202.
(11) If the superficial equality of the two forms systematically led to incomprehension, the linguistic system would probably find alternative means to express the function encoded by these two cases or one of them, responding to a need for clarity which generally characterizes languages.
(12) This is generally true in Romance languages (except for Romanian verbs displaying double accusative constructions, which seem to be a Balkan phenomenon). As regards English, the status of give-verbs selecting two accusatives (although dative and accusative are not morphologically distinct) for which passives of the type Mary was seen by John (vs. John saw Mary) are possible is in all likelihood related to the dative-nominative syncretism in OE. For an exhaustive discussion of the origin of the 'new passive' in OE cf. Allen (1995: 158 and ff.).
(13) Banfi hypothesizes the direction Gen > Dat in opposition to Manzini & Savoia's theory of case percolation Dat > Gen (cf. 2.).
(14) Recall that the Latin language spoken in the Byzantine Empire--and generally in the marginal areas of the Empire--had different features and developed differently (from a lexical point of view, cf. Lat. magis > Rom. mai, Sp. mas 'more' vs Lat.plus > It. piU, Fr. plus, etc.).
(15) It is interesting to notice that Bulgarian and Macedonian are emblematic from a number of points of view, as they are the only two languages in the Slavic family which have developed the article category, which is postponed or agglutinated, after the loss of morphological case. The case of Bulgarian and Macedonian represents a further confirmation of the development of the definite article from demonstratives, since this category in the two languages evolved out of the Old Slavic demonstrative pronoun *jb, *ja, *je ('this'), a form which disappeared from the spoken language by the 16th century but was maintained in the literary language. Gebert (1996: 11) also claims that a striking characteristic of the Slavic languages which have not developed the definite article is the presence, attested in Old Slavic, of long and short forms of the adjective, which linguists generally define as definite and indefinite, respectively, and are today still productive in Slavic languages. Long forms of the adjective are interestingly composed of the adjective and the demonstrative pronoun *jb, *ja, *je, that, in turn, developed from the Indoeuropean deictic *-io and belongs to the same class as the Latin demonstrative is. Also recall that Macedonian is generally treated as a dialect of Bulgarian.
(16) For the sake of convenience, I will indicate a in all its forms as AL.
(17) I find this aspect to be extremely significant (although not much insisted on in the literature) because it represents an evident indication that AL is more related to the preceding noun than to the NP present in the genitival phrase.
(18) As the brackets indicate, clitic doubling is optional here.
(19) Recall that na is also actively used as a locative and directional preposition.
(20) The KP Hypothesis (Lamontagne/Travis 1987: 180 ff.) implies the assumption that the category selecting NP/DP is called K(ase)P(hrase), a projection whose head checks D-/N-features and is licensed by the feature [+Case]. The nominal structure appears therefore as follows: [VP [KP [DP [NP]]]].
(21) Beukema/den Dikken, in fact, give for granted that na is a preposition, just as in standard literature.
(22) The part of the pronominal system of Macedonian related to the expression of the indirect object, which I have not reported here, is identical to Bulgarian, except for the 3rd person forms. It is to be noted that the dative clitic forms are also used to indicate possession (i.e. genitivity) and in fact most Bulgarian and Macedonian grammars describe them as short forms of the possessive pronouns.
(23) When se is followed by a determiner--as is often the case with proper names in Greek--the two elements merge and se appears as s-.
(24) As in other parts of this paper, for the sake of convenience I will assume the most widespread term in the literature to refer to the treated item.
(25) Vocative has not been reported here because it only occurs with a very limited number of words.
(26) As we see in Table 1.5., genitive and dative are also partly syncretic with ablative, a tendency which is common to a number of languages in the world.
(27) As reported by Noess (2009: 578), many languages in the world display this phenomenon with tendencies that are similar to those characterizing the various Balkan languages (e.g. in Bengali genitive extends to cover some dative functions or in Dravidian languages, where dative is used for the expression of some possessive relations). What is more, crucially syncretism between genitive and dative was already present both in Old Church Slavonic and in Latin (see (2.) for Reichenkron's explanation).
Dipartimento di Studi Linguistici e Culturali Comparati
Dorsoduro 3462--VENEZIA 30123 (VE)--Italy
Table 1 Feature analysis of the Latin case system Voc Nom Acc Gen Dat Abl addressee + - - - - - oblique - - + + + + peripheral - - - + + + [local.sup.7] - - + - - + possessor - - - + - -
|Printer friendly Cite/link Email Feedback|
|Publication:||SKASE Journal of Theoretical Linguistics|
|Date:||Dec 1, 2011|
|Previous Article:||The syntax of yes-no questions in Embosi.|
|Next Article:||The role of morphological naturalness in the production of innovative verbs in English and Polish: a comparative study.|