On the left periphery of DP – Evidence from possessives (Abstract selected for LESCOL 2006, Leipzig)

(cf. Program at: http://lingweb.eva.mpg.de/LSSLD/files/lescol.htm)

1. Introduction
Following the DP analysis proposed by several authors (Szabolcsi (1983), (1987), Abney (1987), Valois (1991), Giorgi & Longobardi (1991), Cinque (1994), Longobardi (1994), among others) and the proposals for the architecture of the left periphery of the sentence (Ambar (1996/1998), Rizzi (1995/1997), a.o.), the aim of this paper is to show some evidence from the syntax and semantics of possessive pronouns in Romance languages that support the idea that:
(i) possessives behave like adjectives, overtly agreeing with the noun;
(ii) although being basically generated inside NP, possessive pronouns end up in a high position inside the DP where they maintain a relation with the subject of the discourse.
The syntax of some adjectives, in Romance languages, will support the idea that an extra high projection will receive adjectives, namely adjectives displaying an evaluative reading when preposed to the head noun. Assuming Rizzi (1995/1997) and Ambar (1996/1998), among others, we will argue that the referred projection is Evaluative Phrase. Both phenomena are related to discourse either by visible agreement (possessives) or by subjective meaning (evaluative adjectives) obtained in pre-nominal position.

2. Data and analysis
Possessives in Romance languages display a uniform pattern always showing morphological agreement with the head noun (cf. (1)), a behaviour similar to the one typical of adjectives (cf. (2)):
(1) a. o nosso [possessive, masculine] livro [masculine]
the our book
b. a minha [possessive, feminine] mesa [feminine]
the my table
(2) a. a menina [feminine, singular] bonita [feminine, singular]
the girl beautiful
b. o livro [masculine, singular] espesso [masculine, singular]
the book thick
Besides the adjectival agreement shown in possessive pronouns, Romance languages also show an agreement between the possessive and a subject of discourse. Thus if the possessive is projected inside NP (either in Spec or in Compl position) it will move out from there to check a person feature. Therefore the possessive must move up to the position where that feature may be checked: gP in (5). Furthermore we also assume that Romance languages always display a movement of the head noun (Cinque (1994) a.o.) to an agreement position, where Nº checks its number and gender features and, consequently, shares them with the Dº.
Based on identical evidence, Gonzaga (2003) has argued that both adjectives and possessives are base generated inside the NP: in [Spec, NP] and in [Compl, NP] positions. With nominalized nouns, where the argument structure is active, the Spec and Compl positions inside NP will be occupied by arguments interpreted as subject-agent and object-theme, respectively, that may have an adjective (cf. (4)) or possessive (cf. (3)) shape. In non nominalized nouns we assume that the two positions inside NP are also available for possessives as well as for adjectives.

(3) a. a invasão [da península Ibérica]theme [pelos bárbaros]agent
the invasion [of-the peninsula Iberian] [by-the barbarian]
b. a suatheme invasão [pelos bárbaros]agent
the its invasion by-the barbarians
(4) a. a invasão [bárbara]agent da península Ibéricatheme
the invasion barbarian of-the peninsula Iberian
b. a [sua]theme ocupação [bárbara]agent
the its occupation barbarian
We argue that the base structure for both (3) and (4) is the one in (5). After derivation we will obtain structures (6)-(8).
(5) [DP [Dº a [gP [EvaluativeP [AgrP [NP os bárbaros q-Agent] [Nº invasão [península Ibéricaq-Theme]]]]]]]]
bárbara sua
sua
(6) [DP [Dº a [gP suaj [AgrP invasãoi [NP suaj q-Agent] [Nº ti [(da) península Ibérica q-Theme]]]]]]]]
(7) [DP [Dº a [gP suaj [AgrP invasãoi [NP pelos bárbaros q-Agent] [Nº ti [suajq-Them]]]]]]]]
(8) [DP [Dº a [gP suaj [AgrP invasãoi [NP bárbaraq-Agent] [Nº ti [da península Ibéricajq-Theme]]]]]]]]
As far as adjectives displaying an evaluative reading are concerned, data will be given with adjectives moving to Evaluative Phrase (EvP) (pre-nominal) and with other that are base generated there (mero ‘mere’).
(9) a sua oportuna ocupação da península Ibérica
We argue for the existence of evaluative meaning when there is no factual evidence that the adjective has a role on delimiting the denotation of the noun. Furthermore, we will show that both possessive and adjective may co-exist in the DP structure supporting the idea that the two projections are separate and that they really exist.The proposal will be made upon data from different Romance languages, namely European vs. Brazilian Portuguese, Italian, Spanish and French, and from Germanic languages like English.