The role of prominence in Spanish sentence comprehension: an ERP study

GATTEI, CAROLINA ANDREA; TABULLO, ÁNGEL JAVIER; PARÍS, LUIS; WAINSELBOIM, ALEJANDRO JAVIER

Ámsterdam

Encuentro; Annual Meeting of the Society for Neurobiology of Language; 2014

Society for the Neurobiology of Language

One of the major tasks involved in comprehending sentences is understanding ?who did what to whom?. In order to do so, readers need to link sentential syntactic constituents to the appropriate thematic role for each verb. Evidence from German, an SOV language, has shown that the parser does not wait until the verb is read in order to form a prediction about how such linking should proceed (Bader and Bayer, 2006). Moreover, a number of studies shows that appearance of an unexpected verb after prominence (the hierarchical relationship between sentential arguments) is wrongly computed entails differential electrophysiological correlates due to thematic reanalysis effects (Bornkessel, Schlesewsky and Friederici, 2003, Bornkessel- Schlesewsky and Schlesewsky, 2009 among others). However, up to the best of our knowledge, little is known about the role of prominence in SVO languages that also allow word order variation. In order to fill this gap, we conducted and ERP study in Spanish (N=23), and examined the role of prominence for comprehension in this language by testing the interplay between word order (SVO vs. OVS) and type of verb (Activity vs Object Experiencer psych verbs). While the verb in sentences (a) and (b) assign the most prominent role (i.e. Agent) to the nominative constituent, the verb in sentences (c) and (d) assign the most prominent role (i.e. Experiencer) to the dative constituent. Agentive SVO: (a) María le grita a Juan (MaríaNOM yells at JuanDAT); Agentive OVS: (b) A María le grita Juan (JuanNOM yells at MaríaDAT). OE SVO: (c) María le gusta a Juan (MaríaNOM appeals to JuanDAT); OE OVS: (d) A María le gusta Juan (JuanNOM appeals to MaríaDAT); Predictions: If computation of prominence indeed plays a role for comprehension in languages that allow word order variation, an ERP effect should be found in both SVO sentences with object experiencer (OE) verbs, and OVS sentences with activity verbs, due to difficulty of integration of these verbs after a constituent that comprises the least prominent argument of the event. Results: Results show that at the disambiguating region of the verb, subject-initial sentences showed a positivity for OE verbs (maximum at 600 ms; centroparietal topography); and object-initial sentences showed a negativity (maximum at 400 ms, centroparietal topography), and a positivity (maximum at 600 ms, broadly distributed topography) for activity verbs. Conclusion: Results suggest that in context-free declarative sentences, subject-initial arguments are firstly linked to the Agent role, the highest ranked thematic role for nominative case in Spanish. When an OE verb is encountered, thematic role for the previous argument needs to be revised. Conversely, object-initial arguments are firstly linked to the role of Experiencer, the most prominent thematic role for this constituent in initial position. When an agentive verb is encountered, this decision needs to be amended. All in all, results demonstrate that thematic expectations -and thus, linking- are generated incrementally, and may even be based upon the appearance of a single argument, as it occurs in SVO languages.