Adjectives That Aren’t: An ERP-Theoretical Analysis of Adjectives in Spanish

Achieving a better understanding of the brain-based representation and processing of human language is essential to an overall understanding of language, yet the two fields that examine these phenomena, linguistic theory and linguistic processing, are often studied in isolation. The present dissertation incorporates both methodologies to provide a more complete picture of the language structure in question – adjectives. Adjectives can appear both before and after the noun across languages, and Spanish is an interesting case because it allows both prenominal (e.g., el bello cuadro ‘the beautiful picture’) and postnominal (e.g., el cuadro bello ‘the picture beautiful’) modification within the same language. Further, several subgroups of adjectives, differing in their syntactic characteristics, occur in both positions. As a result, no widely accepted theory of adjective syntax exists within the theoretical literature. This dissertation aimed to clarify the controversies surrounding adjective syntax by employing a measure of processing, event-related potentials (ERPs) to help elucidate details not available using theoretical techniques alone. Specifically, the study examined three potential factors that could influence the adjectives’ distribution – syntactic movement, word class, and morphological gender agreement. In order to investigate each potential factor, six categories of adjectives were identified (three prenominal, three postnominal) based on their theoretical syntactic distribution. The electroencephalogram of native speakers of Mexican Spanish was collected as they read three-word phrases including each of these adjective types. The resulting waveforms were time-locked to the adjective, averaged across participants, and analyzed via repeated measures ANOVAs. The results indicated that one of the adjective types, specifically, one that can appear both pre- and postnominally while retaining its meaning, is a moved linguistic element, supporting theories by Demonte (2005) and Taboada (2010). The analysis into word class was inconclusive, echoing the current status of the ERP processing literature. Finally, a general pre- and postnominal division in gender agreement processing was identified that may be linked to the parser’s expectations regarding syntactic category after the appearance of a determiner. Overall, the study revealed the benefits of using a combination of methodologies to investigate linguistic debates and showed how linguistic theory and processing can combine to increase our knowledge of human language grammar.