Abstract

Event integration – the conflation of multiple events into a unitary event – plays a vital role in language and cognition. However, the conditions under which event integration occurs in linguistic representation and the differences in how linguistic forms encode complex events remain unclear. This corpus study examines two types of Japanese complex predicates – compound verbs [V1-V2]V and complex predicates consisting of a deverbal compound noun and the light verb suru ‘do’ [[V1-V2]N suru]V – using an original “related-event approach”. Findings indicate that [[V1-V2]N suru]V can be established based on coextensiveness alone, whereas [V1-V2]V typically requires direct or shared causality (“the inevitable co-occurrence constraint”). The related-event approach examines related events of linguistic concepts, such as causes and purposes of an event, identified through “complex sentences” from ultra-large-scale web corpora. This study demonstrates that such an approach is effective in clarifying causal relationships between verbs. Furthermore, this paper contributes to the “iconicity versus frequency” debate by showing that conceptually more accessible events (causality plus coextensiveness) tend to be represented in a simpler form than less accessible events (coextensiveness only), due to “efficiency”. The frequency of usage is a result of the nature of concepts rather than the driving force of coding asymmetries.