Date of Award

Winter 12-15-2018

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



Previous research has found that adding different forms of variability during study can affect later memory at test. For example, having words spoken by different talkers has been shown to improve recall of known and novel words (Goldinger et al., 1999; Barcroft & Sommers, 2005), and varying the cues in cue-target related word pairs has been found to improve recall of the targets (Glenberg, 1979; Bevan et al., 1966). It was unclear, however, whether benefits of variability would extend to more naturalistic stimuli, such as sentences, which have higher working memory demands. The present set of experiments investigated how talker and contextual variability, both individually and combined, affect free recall of target words that appear in semantically-related sentences.

Target words were sentence-final items, and all stimuli in Experiment 1 were presented auditorily and orthographically. For each participant, targets appeared in one of the following four conditions: the same sentence spoken three times by the same person (no variability), three different sentences spoken by the same person every time (contextual variability), the same sentence spoken by three different talkers once each (talker variability), or a different sentence spoken by a different talker at each of three exposures (combined contextual and talker variability). Conditions with contextual variability resulted in significantly worse memory performance than constant-context conditions. There was no significant effect of talker variability and no significant interaction between talker and contextual variability.

Experiment 2 further investigated the unexpected negative effect of contextual variability observed in Experiment 1 by changing the presentation modality to auditory-only (all with a constant talker). The switch from combined auditory-orthographic to auditory-only presentations was designed to both decrease working memory demands and encourage processing of the sentence as it unfolded in time. In addition, working memory measures were collected in order to test two predictions—that working memory would be a significant predictor of target word recall and that it would be a significantly better predictor in the variable-context compared to the constant-context condition. No recall differences between the constant- and variable-context conditions were found, but there was a significant positive relationship of working memory on target word recall. Lastly, although positive relationship of working memory on target word recall was stronger in the variable- than the constant-context condition, the interaction was not statistically significant.

These findings suggest that the benefits of talker and contextual variability that have previously been found for lists of words or word pairs (e.g., Glenberg, 1979; Barcroft & Sommers, 2005) do not necessarily extend to semantically-related sentences. The results are discussed with regard to working memory demands and how this may interact with variability.


English (en)

Chair and Committee

Mitchell Sommers

Committee Members

Joe Barcroft, Mark McDaniel, Henry Roediger, Kristin Van Engen,


Permanent URL: