Final Report Summary - VERBAGRENCODING (Verb agreement encoding during language production and comprehension)
The main objective of our research project was to make a cross-linguistic characterization of agreement encoding processes, both during language production and comprehension, and in both native and non-native speakers. For this purpose, we used a phenomenon that has been extensively used to investigate the way in which agreement is computed during language production and, to a lesser extent, language comprehension: the agreement attraction phenomenon (Bock & Miller, 1991). Processing agreement during production, or detecting ungrammatical agreement during comprehension appears effortless, but it is affected by the presence of more than one possible agreement candidate in a sentence. For instance, in a sentence such as ‘The key to the cabinets...is/*are’ both key and cabinets are potential candidates with which the verb can agree in number (Bock & Miller, 1991; Nicol, Forster, & Veres, 1997), yielding errors generally known as agreement attraction. Importantly, this phenomenon has been mostly investigated in subject-verb agreement structures, ignoring other verb agreement configurations. In several series of experiments we aimed at providing new evidence on whether the encoding of object-verb agreement relations in Basque and the encoding of object-clitic agreement relations with their antecedents in Spanish are similarly affected by attraction effects and thus whether they are governed by the same principles as subject-verb agreement relations. We also intended to explore the similarities of agreement encoding during language production and comprehension. Finally, in order to explore the way agreement relations are explored by bilinguals, we compared the performance of non-native vs. native speakers.
Our data on number agreement encoding in Basque (Santesteban, Pickering & Branigan, 2013) revealed that object-verb agreement (a relation not studied before) is affected by the same syntactic factors that affect the encoding of subject-verb agreement. Basque native speakers were asked to repeat and complete preambles containing singular or plural subjects and objects in sentences with canonical subject-object–verb (SOV) or non-canonical object–subject–verb (OSV) order. Participants were equally likely to produce erroneous plural subject agreement following singular subjects and plural object agreement following singular objects, revealing similar number attraction effects in both cases. Both subject and object errors were more common for OSV than SOV sentences but the increase in errors from SOV to OSV sentences was greater for object agreement than for subject agreement. These findings suggest agreement encoding is affected by both sentential word order and proximity of agreeing elements; furthermore, object agreement involves similar processes to those of subject agreement.
In a follow up experiment (Santesteban, Pickering, Branigan, & Laka, in preparation) we showed that highly proficient Spanish-Basque early bilinguals (L2 acquired before 4) are more prone to error attraction effects than Basque native speakers. No differences were found in their performance on subject and object agreement encoding, suggesting that there are no differences in bilinguals' acquisition and use of subject-verb (a relation present in their L1-Spanish) and object-verb (a relation not present in their L1) agreement encoding. These results suggest that acquiring a second language at an early age allows speakers to properly acquire agreement rules not present in their first language, even though their agreement encoding rules seem to be "weaker" than those of native speakers.
Similarly, our data on number agreement encoding in Spanish (Santesteban, Zawiszewski, Erdocia, & Laka, in preparation) revealed that object-clitic agreement (a relation not studied before) is affected by similar syntactic factors that affect the encoding of subject-verb agreement. In a grammaticality judgment self-paced reading experiment Spanish native speakers showed larger reading times and lower accuracy rates in sentences containing number mismatching than matching attractors, revealing main number attraction effects. These effects were only present in singular (but not plural) object-clitic agreement (replicating previous number markedness effects) and were larger in ungrammatical sentences than in grammatical ones (replicating the grammatical asymmetry of attraction effects). Moreover, the electrophysiological responses to object-clitic agreement violations elicited similar ERP patterns to those previously shown for subject-verb agreement: a biphasic frontal negativity-P600 pattern following number-matching attractors. However, number attraction following non-matching attractors led to the absence of frontal negativity (only P600 elicited). Importantly, attraction effects obtained only in ungrammatical sentences (larger negativity following number matching attractors than non-matching ones). These findings suggest that agreement encoding of object-clitics involves similar processes to that of subject-verb agreement. These findings are argued to support cue-based retrieval mechanisms for agreement comprehension, as proposed by Wagers, Lau, and Phillips (2009).
Finally, in another series of experiments we explored number attraction effects in the production of both subject-verb and object-clitic agreement relations in Spanish. In two experiments we showed that the presence of a local noun mismatching in number with the head noun led to agreement errors in both subject-verb and object-clitic agreement relations. Similarly to our findings in comprehension, main number attraction effects were larger in singular than plural agreement in both subject-verb and object-clitic relations, replicating the number markedness effects in agreement error attraction. In sum, we showed that similar attraction effects occur in subject-verb and object-clitic agreement, both in sentence comprehension and production modalities.
In sum, in several series of experiments we have shown that number attraction effects similarly disrupt the production and comprehension of different types of agreement relations (subject-verb, object-verb and object-clitic agreement) in typologically different languages (Basque and Spanish). In our cross-linguistic characterization of agreement encoding processes, our results suggest that the encoding of object-verb agreement relations in Basque and the encoding of object-clitic agreement relations with their antecedents in Spanish are governed by the same principles as subject-verb agreement relations. Furthermore, data from highly proficient early bilingual speakers suggest that, although they seem to process agreement in a similar way as native speakers, their agreement encoding system seems to be more prone to attraction errors. Hence, our research provided results that would allow us perform a more broad cross-linguistic characterization of agreement encoding processes during speech production and comprehension.
List of Keywords: Verb agreement; Speech errrors; Sentence Production; Sentence Comprehension; ERPs.