Back to list

Detail of contribution

Auteur: Georg HÖHN

Unagreement is an Illusion: The distribution of person features in the extended nominal projection

Abstract/Résumé: The term unagreement (Hurtado 1985) describes a configuration observed in various pro-drop languages (e.g. Spanish, Modern Greek, Bulgarian), involving an agreement mismatch between an apparently third person plural subject DP and first or second person plural subject agreement on the verb (1). Other pro-drop languages like Italian lack this construction (2). (1) Ftiaksame i fitites ena oreo keik. (Greek) made.1pl the students a good cake 'We chemists baked a good cake.' (2) *Gli studenti abbiamo fatto una torta. (Italian) the students have.1pl made a cake In languages with articles, the definite article is obligatory in adnominal pronoun constructions (APCs; "we linguists") in langages with unagreement, while non-unagreement languages proscribe an article in APCs (3). (3) a. emis *(i) fitites (Greek) we the students b. noi (*gli) studenti (Italian) we (*the) students My analysis adopts a pronominal determiner analysis of the second type of APCs (Postal 1969, Rauh 2003, Roehrs 2005) and Panagiotidis' (2002) analysis of pronouns as involving a silent noun eN. I suggest that the cross-linguistic variation in the availability of unagreement results from a difference in the structure of the extended nominal projection (xnP). The structure in (4) corresponds to (3a), unagreement is available when person and definiteness features are hosted on separate heads in xnP. The data in (3b) suggest that in non-unagreement languages regular pronominal determiners encode definiteness and person on a single head, yielding (5). (4) [PersP Pers [DP D [NumP Num NP ]]] (5) [DP D [NumP Num NP ]] Null spell-out of Pers in (4) leads to unagreement. Null realization of D in (5) would not result in the characteristic definite plural DP subject, but in a bare noun. There seems to be an independent constraint in both types of languages that definite D cannot be silent if there is other overt material in the DP. This prevents null realization of D in (5) exactly for the relevant contexts for unagreement, i.e. if there is overt material in NumP (while allowing it with a silent noun eN, yielding "pro"). If the realization of Pers depends on demonstrativity (drawing on Rauh's 2003 investigation of German pronominal determiners), the proscription of overt pronouns in quantificational unagreement (6) is also explained: quantified phrases cannot be demonstrative. (6) (*Emis) poli taksidiotes aghapame ti Thessaloniki. we many travellers love.1pl the Thessaloniki 'Many of us travellers love Thessaloniki.' The present theory offers an account for the cross-linguistic variation and a solution for a serious problem for asymmetric theories of agreement.