Researchers have hypothesized that words that are highly related semantically are more likely to occur within the same intonational phrase (F. zzaq;, 1988; E. O. Selkirk, 1984). D. Watson and E. Gibson (2004) proposed that semantic closeness can be captured by using the argument/adjunct distinction, such that intonational boundaries are more likely to occur before adjuncts than before arguments. In the current experiment, the authors compared two aspects of argumenthood: semantic relatedness and obligatoriness. In a production study, speakers were more likely to place an intonational phrase boundary between a word and a dependent if the dependent was optional (e.g., after "investigation" in "The reporter's investigation [of the crash] unnerved the officials") than if the dependent was obligatory (e.g., after "investigated" in "The reporter investigated [the crash], and this unnerved the officials"). These data suggest that obligatoriness is a better predictor of intonational boundary placement than semantic closeness.