Online Learning From Input Versus Offline Memory Evolution in Adult Word Learning (Storkel et al., 2014)

Purpose In this study, the authors investigated adult word learning to determine how neighborhood density and practice across phonologically related training sets influence online learning from input during training versus offline memory evolution during no-training gaps.
Method Sixty-one adults were randomly assigned to learn low- or high-density nonwords. Within each density condition, participants were trained on one set of words and then were trained on a second set of words, consisting of phonological neighbors of the first set. Learning was measured in a picture-naming test. Data were analyzed using multilevel modeling and spline regression. Results Steep learning during input was observed, with new words from dense neighborhoods and new words that were neighbors of recently learned words (i.e., second-set words) being learned better than other words. In terms of memory evolution, large and significant forgetting was observed during 1-week gaps in training. Effects of density and practice during memory evolution were opposite of those during input. Specifically, forgetting was greater for high-density and second-set words than for low-density and first-set words.
Conclusion High phonological similarity, regardless of source (i.e., known words or recent training), appears to facilitate online learning from input but seems to impede offline memory evolution.