Abstract
Classic research in psychology has shown that comprehenders in recall tasks ‘chunk’ word lists in order to facilitate later retrieval. Here we ask how chunking mechanisms apply in grammatically structured sentence stimuli. How do comprehenders make use of this grammatical structure to guide chunking of a sentence’s component words in memory? In previous studies of multi-clause sentence recall, many researchers have argued that chunking is clause-based, as evidenced by scalloped serial accuracy curves in which primacy and recency effects are observed within each clause. Here we suggest that these previous findings may have been driven by the correlation of sentence boundaries with nouns, a lexical class known to demonstrate an independent accuracy advantage in recall. Using immediate serial recall of naturalistic biclausal English sentences and manipulating noun placement, we find evidence that nouns indeed confer a large accuracy boost at the internal clause boundary that can explain prior results. We discuss the difficulty of applying serial position logic to highly structured representations such as those that underlie sentence forms.