The place of PLACE in Persian

Abstract

In this talk I present an analysis of Persian spatial P2 Prepositions and show that their nominal properties can be explained by assuming a silent place element in the P projection. Background: Traditionally, Persian (spatial) prepositions are divided into two classes, with respect to which of them take the Ezafe linker when followed by a complement: P1 — prepositions which directly take a complement, and P2 — those which require an Ezafe before the complement (Samiian 1994). Except for the availability of the Ezafe linker, Persian spatial P2s manifest a range of other distinctions compared to spatial P1s in that they can (i) occur without complement, (ii) follow a demonstrative, (iii) take plural morphology, (iv) host pronominal clitics, which typically express the possessor when attached to nouns, and, finally, (v) can be NP modifiers (Pantcheva 2006). P1s, in their turn, can precede any P2P to specify a Goal or a Source/Route Path and Location, depending on which P1 is used (see (1)). However, all P1s, except the ones expressing Source, can be omitted in front of a P2P. A P1 is omitted also when it is followed by a NP denoting a conventional place, e.g proper place names, places for activities (school, work), etc. Previous accounts: As can be seen from (i)–(v), P2s exhibit nominal properties and their categorial status is therefore controversial. According to Ghomeshi (1996) and Larson and Yamakido (2005) they are nouns, whereas Samiian (1994) argues against this. The arguments she gives in favor of the distinction between nouns and P2s are based on the restrictions which hold for the latter but not for true nouns, e.g. the unavailability of adjectival modification and the ungrammaticality of quantifiers. Proposal: In this talk, I first consider further evidence that P2s are not real nouns. Then, I suggest that spatial P2s are representatives of the syntactic category Ax(ial)Part, as proposed by Svenonius (2006). AxPart elements are part of the PP projection and are used instead of adpositions to express spatial meanings. Such AxPart elements can be found in a variety of languages, e.g. Japanese, Greek, Hebrew, Korean, Kiitharaka (Bantu), etc. and they all exhibit some or all of the properties (i)–(v). An important cross-linguistic property of AxParts is that they are noun modifiers and they exhibit nominal properties while at the same time being distinct from nouns. For instance, they can be preceded by a demonstrative. Crucially, then the demonstrative gets the interpretation of here or there. Katz and Postal (1964) decompose the words here and there into at this place and at that place, respectively. Interestingly, the Persian adverbs here and there consist of exactly the demonstrative in ‘this’ and un ‘that’, respectively, and the noun dja ‘place’. Therefore, following Katz and Postal (1964) and Kayne (2004), I suggest that there is a silent place selected by a spatial P1, the latter being usually omitted since place doubtlessly expresses conventional place. The nominal place element is then modified by the P2 Phrase, which is allowed, since P2s can be noun modifiers (see (v)). The Ground complement is the possessor of the space thus specified and this requires an Ezafe to attach to the last element of the phrase expressing the possessed thing, i.e. to the P2. Adopting Samiian’s (1994) proposal that Ezafe is a case marker, and Svenonius’ (2006) PP structure, I put Ezafe under K (Case). The proposed structure for the PP in (2a) then would be as in (2b). A similar approach is adopted by Terzi (2006) for Greek locative prepositions, Botwinik-Rotem (to appear) for Hebrew, and Bendjaballah and Haiden (2005) for Berber. Extension: The presence of a silent place accounts for several aspects of the behavior P2s exhibit. First, place being a noun, it can be preceded by a demonstrative (cf. (ii) above). Since place is silent, the demonstrative will appear to directly precede the P2P at Spell-out, still the interpretation of here and there will remain. Secondly, the possessive relationship between the Ground DP and place allows the former to be replaced by a possessive clitic (iv). When the possessor is omitted, the P2 will appear without a complement (i). Finally, plural endings will attach to the P2, since they need a phonologically realized element to attach to (iii). In any of these cases, the whole phrase will behave as a PP because it is headed by an overt or covert P1.