Summary #e aim of the paper is to provide an explanation for the following difference between English and Slovene: whereas in English a definite determiner and a possessor are in complementary distribution, in Slovene the two categories are perfectly compatible. Arguing that the traditional approach to determiner-possessor complementarity is inadequate, the paper proposes an explanation that has been developed within the framework of generative grammar: languages exhibiting determiner-possessor complementarity are characterized by the presence of the [~def] feature on the functional head Pos. #e generative approach also shows that (with the definite article and a demonstrative occupying different structural positions) determiner-possessor complementarity is in fact twofold, comprising (i) article- possessor complementarity and (ii) demonstrative-possessor complementarity. Povzetek Namen članka je razložiti naslednje razhajanje med angleščino in slovenščino: medtem ko se v angleščini določni element in posedovalec medsebojno izključujeta, sta v slovenščini omenjeni kategoriji med seboj popolnoma kompatibilni. Ker se izkaže, da je tradicionalni pristop h komplementarnosti določnega elementa in posedovalca neustrezen, članek predlaga drugačno razlago, zasnovano v okviru tvorbene slovnice: v jezikih, katerih značilnost je komplementarnost določnega elementa in posedovalca, je funkcijsko jedro Pos nosilec obeležja [~def]. Tvorbeni pristop pokaže tudi, da je (zaradi različnih strukturalnih položajev določnega člena in kazalnega zaimka) komplementarnost določnega elementa in posedovalca pravzaprav dvojna, in sicer vključuje (i) komplementarnost določnega člena in posedovalca ter (ii) komplementarnost kazalnega zaimka in posedovalca. DOI: 10.4312/elope.1.1-2.15-22 A Slovene learner of English may find himself puzzled by the fact that whereas in Slovene a definite determiner can be used side by side with a possessor (e.g. tista moja knjiga “that my book”), in English the two categories are incompatible (e.g. *the/*that my book). #e aim of the present paper is to provide a plausible explanation for this syntactic phenomenon. Arguing that the traditional approach to the problem of determiner-possessor complementarity is inadequate, the paper proposes a solution that has been developed within the framework of generative grammar. Possessors 1 in English are viewed as occupants of the same structural position as the definite article (Quirk et al. 1985, 326). Since one and the same structural position cannot be filled twice, the definite article (or a demonstrative 2 ) and a possessor are in complementary distribution: (1) a. my/Peter’s book b. the/that book c. *the/*that my/Peter’s book In Slovene, on the other hand, demonstratives 3 and possessors are perfectly compatible, which suggests that possessors in Slovene are not determiners but modifiers (cf. Plank 1992; Haspelmath 1999): (2) a. moja/Petrova knjiga (Slovene) my/Peter’s book b. tista moja/Petrova knjiga (Slovene) that my/Peter’s book #e view presented above may account for another difference regarding the use of possessors in the two languages. As modifiers, Slovene possessors do not affect the (in)definiteness of constructions they occur in and are thus acceptable in both definite and indefinite nominal phrases. English possessors, on the other hand, are carriers of definiteness and as such incompatible with indefinite determiners: 1 2 (3) a. neka moja/Petrova knjiga (Slovene) some my/Peter’s book b. *a/*some my/Peter’s book #e theory does not hold water though. As pointed out by Schoorlemmer (1998), possessors are not equivalent to definite determiners even when they seem to occupy the position of a determiner. Both categories exhibit specificity, however, only the latter also exhibits uniqueness. In the case of Slovene, this may explain the acceptability of examples like (4a) and the unacceptability of those like (4b). Indefiniteness and specificity are perfectly compatible whereas indefiniteness and uniqueness are not: (4) a. neka moja/Petrova knjiga (Slovene) some my/Peter’s book b. *neka tista knjiga (Slovene) some that book Examples like (5a) below should by no means be understood as counter-arguments here. As is illustrated by (5b), a nominal phrase containing a possessor may also be indefinite in English, under the condition that the possessor does not precede the headword: (5) a. *#ere was a/some my/Peter’s book on the table. b. #ere was a/some book of mine/Peter’s on the table. Following these lines, it is possible to conclude that possessors and definite determiners differ to such an extent that it seems unlikely for the two categories to share the same structural position within the nominal phrase. Consequently, the “traditional” explanation for the unacceptability of examples like (1c) above becomes inadequate. Haspelmath (1999), for example, points out that in the case of article-possessor complementarity the traditional explanation is potentially acceptable in languages like English where both elements occur on the same side of the headword. Nevertheless he discards the traditional explanation arguing that the same type of complementarity can also be observed in some languages where the definite article and the possessor occur on the opposite sides of the headword. For example (Haspelmath 1999, tab. 2): (6) Swedish: a. bok-en book the b. Karins bok Karin’s book c. *Karins bok-en Karin’s book the (7) Irish: a. an tigh the house b. tigh an tsagairt (“the house of the priest”) house the priest c. *an tigh an tsagairt (“the house of the priest”) the house the priest Even in (8c) below, where the two elements are used on the same side of the headword, the traditional explanation is not acceptable because the definite article is realized as a suffix and not as a separate word (Haspelmath 1999, tab. 3): (8) Bulgarian: a. sestra-ta sister the b. sestra mi sister my c. *sestra-ta mi sister the my Haspelmath argues that such patterns of article-possessor complementarity are economically motivated: articleless possessive constructions allow for more economical (i.e. shorter) utterances. It follows from the very nature of possession that a possessed nominal phrase is likely to be definite, and this makes any additional marker of definiteness redundant. #e reason why not all languages or possessive constructions exhibit article-possessor complementarity may be that some languages simply prefer explicit (i.e. transparent but uneconomical) constructions to inexplicit (but economical) ones. Another reason suggested by Haspelmath is a diachronic one. If the definite article was introduced into a language later than a given possessive construction, the article did not become established in that construction because there was no need for an additional marker of definiteness. If, on the other hand, the definite article was established in the language before a given possessive construction, the possessive construction was later syntacticized together with the definite article. #is explains the article- possessor complementarity in the case of English prenominal possessive constructions (e.g. my book) and the non-complementarity in the case of postnominal ones (e.g. the roof of the house). #e former type of construction was established in the language before the definite article whereas the latter is of more recent origin. Haspelmath’s explanation of article-possessor complementarity is based on the universal tendency to economize, i.e. to make linguistic utterances as short as possible. Another, purely syntactic explanation has been proposed within the framework of generative grammar. What follows is a brief outline of this approach. Within the generative framework, nominal phrases like the book are labelled determiner phrases (DPs) with the basic structure as illustrated below: (9) [DP Spec [D’ D [NP Spec N ]]] [DP Spec [D’ the [NP Spec book ]]] Due to certain functional projections intervening between D and NP (I will not go into details here), the structure of a DP is in fact much more complex. For the purposes of this paper, two of such intervening functional projections will be included in my presentations – PosP and NumP: (10) [DP Spec [D’ D [PosP Spec [Pos’ Pos [NumP Spec [Num’ Num [NP Spec N ]]]]]]] For DPs like my book, the following structure has been proposed (Schoorlemmer 1998): nouns are base-generated in N, moving to Num in overt syntax, whereas possessors 4 are base- generated in Spec,NP , moving to Spec,PosP: (11) [DP Spec [D’ D [PosP Spec [Pos’ Pos [NumP Spec [Num’ Num [NP my book ]]]]]]] [DP Spec [D’ D [PosP myj [Pos’ Pos [NumP Spec [Num’ booki+Num [NP tj ti ]]]]]]] #e movement is induced by the need for feature-checking, a mechanism whose key concepts are the following (cf. Chomsky 1995; Radford 1997): i) A given syntactic element carries a set of features. ii) A given feature may be variable (optional) or fixed (obligatory). For example, the feature for definiteness is variable on N [~def] and fixed on D [±def]. iii) A variable feature must be checked, i.e. associated with a fixed feature of the same type. iv) #e need for feature-checking triggers movement. v) If the feature is variable and strong, the movement is overt. As can be seen from (9) and (11) above, a determiner and a possessor are assigned two distinct structural positions. With the possessor occupying Spec,PosP , it is still unclear why D cannot be filled by a determiner in (11). #e only plausible explanation seems to be the assumption that D is already filled by something else. As proposed by Schoorlemmer (1998), D carries a strong feature [±def], and in English, Pos has a variable feature [~def], which is assumed to be strong. T o check this strong feature, Pos has to move to D in overt syntax, which makes it impossible for a determiner to occupy its position: (12) [DP Spec [D’ D [PosP Spec [Pos’ Pos [NumP Spec [Num’ Num [NP my book ]]]]]]] [±def] [~def] [DP Spec [D’ Posk+D [PosP myj [Pos’ tk [NumP Spec [Num’ booki+Num [NP tj ti ]]]]]]] 4 It has been assumed that, on the other hand, in languages that do not exhibit determiner- possessor complementarity Pos has no [~def] feature. Since there is no need for checking, there is no movement of Pos to D, D remains empty and can be filled by a determiner. An example from Italian: (13) [DP Spec [D’ D [PosP Spec [Pos’ Pos [NumP Spec [Num’ Num [NP mio libro ]]]]]]] [DP Spec [D’ D [PosP mioj [Pos’ Pos [NumP Spec [Num’ libroi+Num [NP tj ti ]]]]]]] il mio libro the my book #ere is another difference between English and Italian that can be accounted for by the presence/absence of the [~def] feature on Pos. Contrary to their Italian counterparts, English possessive pronouns display different forms in non-elliptical and elliptical DPs: (14) a. Take my book. Take mine. b. Prendi il mio libro. Prendi il mio. (Italian) take the my book take the mine In an elliptical DP , N is filled by “pro”, an empty element behaving like a pronoun and as such carrying a fixed feature [+def]. #e variable [~def] feature on Pos in an English DP can now be checked simply by means of N-to-Pos movement, which makes the movement of Pos to D superfluous: (15) [DP Spec [D’ D [PosP Spec [Pos’ Pos [NumP Spec [Num’ Num [NP my pro ]]]]]]] [±def] [~def] +def] [DP Spec [D’ D [PosP minej [Pos’ proi+Pos [NumP Spec [Num’ ti+Num [NP tj ti ]]]]]]] #e difference between non-elliptical and elliptical DPs lies in the form of the element in the head position of PosP . In non-elliptical DPs, where Pos moves to D, this element is the trace (t) of Pos (cf. 12). In elliptical DPs, where no Pos-to-D movement takes place, the element is Pos (cf. 15). #is difference is overtly reflected by the existence of short and long forms of possessive pronouns in English. In languages where (due to the lack of the [~def] feature on Pos) no movement takes place, the element in question is Pos in both types of DPs. #is explains why Italian possessive pronouns have one form only. Since in elliptical DPs Pos does not move to D, D is empty and can be filled by a determiner: (16) Prendi il mio. (Italian) take the mine Although the system allows both options, a language may grammaticalize only one. English is a case in point: (17) a. Take mine. b. *Take the mine. Slovene obviously belongs to the same type as Italian, lacking the [~def] feature on Pos: (18) Vzemi mojo knjigo. Vzemi mojo. (Slovene) (cf. 14b) take my book take mine Slovene differs from Italian in that it lacks the category of articles. It can nevertheless be assumed that if, hypothetically speaking, the definite article was used in Slovene, it would be compatible with possessors in the same way the definite article in Italian is. In accordance with the traditional view that demonstratives are occupants of the same structural position as the definite article, the presence/absence of the [~def] feature on Pos should also account for the difference in grammaticality between (19a) and (19b) below: (19) a. *this my book b. ta moja knjiga (Slovene) this my book Within the generative framework, on the other hand, demonstratives are regarded as occupants of the specifier position Spec,DP (Campbell 1993). As argued by Bruge and Giusti (1996), one of the arguments in favour of this hypothesis is the fact that in some languages (e.g. Spanish, Hungarian and Modern Greek) a demonstrative and the definite article may co-occur. Furthermore, Plank (1992) points out that demonstratives used to co-occur with possessors in English and that the following examples, found in the 18 th and 19 th century prose, are not entirely ungrammatical even today: (20) a. this their first journey to London b. this my savage crew c. this their church (Plank 1992, 1) It follows from the above observations that demonstrative-possessor complementarity cannot result solely from the presence of the [~def] feature on Pos. If Pos has to move to D (in order to check its [~def] feature), this cannot preclude the use of a demonstrative since demonstratives do not occur in D but in Spec,DP . #e question remains how to account for the incompatibility of demonstratives and possessors in English then. It has been proposed (Bruge and Giusti 1996) that demonstrative-possessor complementarity results from a parametric principle ruling the lexical insertion in a functional head when its specifier is filled. In some languages (e.g. Hungarian), if Spec,DP is occupied by a demonstrative, D must be filled by a lexical item. In languages like English, by contrast, if Spec,DP is occupied by a demonstrative, D must be lexically empty. If a possessor is used in English, D cannot remain lexically empty because the need to check the strong [~def] feature on Pos against the [±def] feature on D triggers overt Pos-to-D movement. #us the use of a demonstrative in English precludes not only the use of the definite article but also the use of a possessor. If, however, Pos in English had no strong [~def] feature, the use of a possessor would not violate the above-mentioned principle: there would be no overt movement of Pos to D, so D would remain lexically empty. Demonstrative-possessor complementarity obviously results from both (i) the principle that rules out the lexical insertion in D when Spec,DP is filled by a determiner and (ii) the presence of the [~def] feature on Pos. Determiner-possessor complementarity is in fact twofold, comprising article-possessor complementarity on the one hand and demonstrative-possessor complementarity on the other. A plausible explanation for this syntactic phenomenon has been proposed within the framework of generative grammar. Languages exhibiting either type of complementarity are characterized by the presence of the [~def] feature on Pos. #ose exhibiting demonstrative- possessor complementarity are at the same time subject to the (parametric) principle that if the specifier of a functional head is filled, the functional head must be lexically empty. Slovene, in contrast to English, exhibits no complementarity of the kind although in examples like tista knjiga (“that book”) the lexically empty D with a determiner in Spec,DP suggests that Slovene is subject to the same principle as English. #e only possible explanation for the compatibility between demonstratives and possessors in Slovene seems to be that Pos in Slovene has no [~def] feature. It is worth noting that this goes hand in hand with the traditional view that possessors in Slovene are modifiers, not affecting the (in)definiteness of constructions in which they occur. Similarly, the fact that a possessor in English (with the [~def] feature on Pos) moves to the head of the determiner phrase, in a way accounts for the traditional classification of possessors in English as definite determiners.