Verbal working memory is involved in associative word learning unless visual codes are available

[Baddeley et al] proposed a model of associative word learning in which the phonological loop, as defined in Baddeley’s working memory model, is primarily a language learning device, rather than a mechanism for the memorization of familiar words. Using a dualtask paradigm, [Papagno et al] found that articulatory suppression, loading verbal working memory, had an effect on the memorizing of word–nonword pairs, but not on the memorizing of word–concrete word pairs. The present work explored the potential for visual codes in unfamiliar word learning. In a first experiment, we replicated the results of Papagno et al. for both nonwords and highly imageable nouns. In addition, we found that articulatory suppression disrupted the memorizing of word-abstract word pairs, suggesting that phonological involvement may be triggered by the absence of visual representations for the abstract words. Experiment 2 showed that an artificially induced association between a nonword and a nonnameable visual image was sufficient to compensate for diminished verbal working memory resources due to articulatory suppression. In a third experiment, we demonstrated that our results generalize to other types of abstract words (i.e., function words), auditory stimulus presentation, and to word learning in children.