Imagery and Meaningfulness Ratings of Sentences: Reliability and Relationship to Learning

Abstract

Exp. 1 assessed the test-retest reliability with which individuals and groups can rate simple sentences for image-evoking power and meaningfulness. For 10 different Ss rating 100 sentences on each dimension, the mean reliability coefficients were .69 and .67 for imagery and meaningfulness respectively. Within-group test-retest reliability of mean ratings was excellent (.90, imagery; .92, meaningfulness). Rating of both dimensions was also performed by a third group of 19 Ss, and inter-group agreement (estimated through correlation of group mean ratings) was substantial (.76, imagery; .87, meaningfulness). However, agreement between individual Ss was low for both dimensions (.18, imagery; .27, meaningfulness), though significant. Exp. 2 examined Paivio's (1971) dual-coding hypothesis applied to sentence memory through correlation of both individual ratings and group mean ratings with sentence recognition scores. Neither method yielded a significant correlation between learning and imagery, or meaningfulness. Apparently the dual-coding hypothesis cannot be straightforwardly applied to memory for sentences, despite its predictive strength for single-word learning tasks.