Argument Linearization in Dutch: A Multi-factorial Approach

Abstract

AbstractIn the combinatory process of building a meaningful expression, verbs establish a semantic and syntactic relationship between their arguments, i.e. the participants in the eventualities expressed by the verb. A meaningful message is only conveyed if the hearer comes to the semantic and syntactic relation between the verb and its arguments intended by the speaker. For getting the arguments right the hearer has different syntactic and semantic sources of information at his disposal. This chapter focuses on the role of such factors on argument linearization and the comprehension of the argument structure of different types of verbs (i.e. agentive verbs, subject-experiencer verbs, causative and unaccusative psych verbs). More specifically, it is argued that the linearization of arguments is a multi-factorial phenomenon including different sorts of disambiguating information (i.e. selectional restrictions, number agreement, case marking, animacy), as well as other types of syntactic and semantic factors (i.e. syntactic function, thematic dependency). To assess the relevancy of these factors for the comprehension of Dutch sentences, an ease of comprehensibility rating study was performed. The results showed clear effects of the order of the arguments and verb type. Structures starting with a subject were easiest to comprehend. This subject-first preference was strongest for verbs selecting an animate subject (agentive/subject-experiencer verbs). The observed difference between the two types of psych verbs indicates that besides the selectional restrictions other verb specific characteristics influence comprehensibility. The data were evaluated by mapping constraint violation patterns onto the results indicating that the advocated multi-factorial approach provides suitable means to explain differences in comprehension of Dutch sentences with different argument linearizations.KeywordsWord OrderThematic RoleSelectional RestrictionPreference RuleAgentive VerbThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.