Do people structure words and sentences using shared mechanisms?

Abstract

The same sequence of linguistic units can be structured differently: Relative clauses in sentences such as “I met the students of the teacher who played the violin” can either have a high attachment (HA; i.e., the students played the violin) or low attachment reading (LA; i.e., the teacher played the violin). Similarly, morphological attachment in noun phrases such as “social psychologist” can have either an HA (i.e., someone who studies social psychology; [[social psycholog(y)][ist]]) or LA reading (i.e., a psychologist who is social; [[social][[psycholog(y)][ist]]]). Thus, abstractly at least, sentences and words have analogous internal hierarchical structures. Using a structural priming paradigm, we investigated in three experiments whether shared mechanisms process the internal structures of both sentences and words, despite the difference in grain size. Overall, we only observed priming effects when the primes and targets were of the same grain size: Participants produced more HA sentences or rated the HA readings of ambiguous sentences as more probable following HA sentence primes (compared to following LA sentence primes), while the attachment structure of morphological primes did not affect subsequent sentence production and comprehension. Participants also rated the HA readings of ambiguous noun phrases as more probable following HA morphological primes (compared to following LA morphological primes), while the attachment of sentence primes did not affect subsequent morphological comprehension. We suggest that at least as reflected by structural priming, structural operations are not shared across morphological and syntactic levels.