On the processing of thematic features in deverbal nominals

Abstract

The primary motivation for the research reported in the present dissertation was to investigate the status of thematic features (TFs) in deverbal nominals (DNs) in Modern Greek. The investigation addressed two independent issues with respect to TFs of DNs. The first was whether the processing of TFs of DNs constitutes a necessary step in accessing their mental representation. The second concerned the status of thematic constraints in deverbal word formation. Three on-line lexical decision tasks and one off-line grammaticality judgment task were carried out. The stimuli for these tasks included deverbal nouns, deverbal adjectives and pseudo-words violating thematic constraints. The findings showed that TFs appear to increase processing load only for those DNs with an increased eventive character (e.g. plysimo 'washing', kallymenos 'covered'), with a decomposition access route possibly playing a facilitatory role. In contrast, TFs do not appear to affect processing in the case of DNs with a diminished 'verb-like' character (e.g. conqueror). Furthermore, lexical access results for pseudo-words indicated that TFs impose constraints which operate at a later stage of word formation compared to other constraints, such as categorial specifications of the base. This strongly suggests that TFs play a crucial role in the creation of new DNs, independently of the type of nominal. The findings of the present study have implications for both psycholinguistics and theoretical linguistics. The psycholinguistic implications relate to the stage-like nature of lexical access, the existence of a general representational component called feature representation, and the role of grammatical class in both lexical access and the organization of the lexicon. The linguistic implications principally inform theories of word formation postulating feature percolation, as well as the role of various constraints operating during derivation. The experimental results support the view that the creation of a new word is subject to constraints specific to the morphological operations involved in it, such as thematic constraints for DNs. Furthermore, constraints seem to apply sequentially, with degrees of violability even for those constraints which are considered to be strong. More importantly, there appears to be a relationship between violability and late application, with those constraints that apply at a later stage being more violable.