engleski

Object Clitic Agreement in Croatian: An ERP study

Among the relationships established within the domain of the morphosyntactic agreement the one between the verb and its arguments stands out as particularly complex and intriguing. The information contained in the lexical entry of the verb determines the number, the nature and the form of its obligatory complements. If the complement is a pronoun, however, the verb determines its form only partially, since all the phi-features (number, gender, person) are inherited from the pronoun’s antecedent. In Croatian, a language with an elaborate case- marking system, the verb will thus dictate the case of the pronoun that serves as an argument, but all the other characteristics of its form are determined by the characteristics of the full NP it is co-indexed with. This is why pronouns as verb arguments represent an especially interesting case of morphosyntactic agreement. Present study aims to investigate the processing of pronouns as direct objects in Croatian. In Croatian the object pronouns are cliticized (unless they are the focus of the utterance), i.e. they come in a shortened and unstressed form, e.g. tebe – youACC vs. te – youACCunstressed, whereas in the most frequently investigated languages such as English they remain in the form of the full pronouns. Participants were presented with two types of sentences: 1) incorrect sentences, i.e. sentences that contained a violation on the object clitic following the verb, and 2) parallel correct sentences. Each experimental sentence was preceded by an introductory sentence, as in examples 1 (a-d): (1) a) Branko izrađuje drveni stolac. Već pola sata brusi ga u radionici. Branko-masc.nom.sg. make- pres.3.sg. wooden- masc.acc.sg. chair-masc.acc.sg. Already half hour-masc.gen.sg. sand-pres.3.sg. he- acc.sg. in workshop-fem.locc.sg. Branko is making wooden chair. He has been sanding it for already half an hour. b) Mislav pije crni čaj. Svakog jutra priprema *ju u velikoj šalici. Mislav- masc.nom.sg. drink-pres.3.sg. black- masc.acc.sg. tea-masc.acc.sg. Every-masc.gen.sg. morning- masc.gen.sg. prepare-pres.3.sg. *she- acc.sg. in large-fem.locc.sg. cup-fem.locc.sg. Mislav drinks black tea. Every morning he prepares *her in a large cup. c) Anton uređuje svoju garažu. Već dva tjedna čisti *joj s prijateljima. Anton- masc.nom.sg. decorate-pres.3.sg. his- refl.acc.sg. garage-fem.acc.sg. Already two week- masc.gen.sg. clean-pres.3.sg. *she-dat.sg. with friend- masc.ins.pl. Anton decorates his garage. He has been cleaning *to her for two weeks already. d) Jana je kupila modni časopis. Prije spavanja čita *joj u krevetu. Jana-fem.nom.sg. buy-perf.3.sg. fashion- masc.acc.sg. magazine- masc.acc.sg. Before sleeping-neu.gen.sg. read- pres.3.sg. *she- dat.sg. in bed. Jana bought a fashion magazine. Before sleeping she reads *to her in the bed. In the sentences with the violation on the object clitic it was either the gender of the pronoun (1b) or the case of the pronoun (1c) that was violated, or both (1d). Sentences were presented visually. The EEG activity on the critical word, i.e. clitic pronoun, was measured in correct and incorrect sentences in two time windows: 300-500 ms and 500-800 ms post-stimulus onset. The ERPs were computed and then compared. Since both single- violation and double-violation condition were included in the experiment, it was possible to examine the relative contribution of each of the single-violation types to the ERP response in the double-violation condition. Expectations on the ERP response were based on the electrophysiological studies on language processing that had revealed that violations on morphosyntactic agreement elicit a biphasic ERP response: LAN followed by P600 (Friederici et al. 2002, Molinaro et al. 2008). This type of ERP response indicates that there are two stages of processing of violation on an argument: the early, fairly automatic stage, during which parser detects the violation, and which is reflected by the ERP component named LAN – left anterior negativity (Friederici and Kotz 2003, Kuperberg et al. 2003), and a later stage during which parser attempts to reanalyze the structure and which therefore elicits the P600, a component that is considered to reflect syntactic processes of integration, reanalysis and repair (Osterhout and al. 1997, Friederici et al. 2002, Kuperberg 2007). In the present study the ERP methodology was chosen because of its high temporal resolution and its ability to differentiate between various types of functional processing, which provide us with an excellent opportunity to investigate linguistic features that participate in building the structural relationships among sentence constituents.

object clitics; gender; case; agreement; ERP; LAN; P600

nije evidentirano