Tip-of-the-tongue states reoccur because of implicit learning, but resolving them helps

In six experiments, we elicited tip-of-the-tongue (TOT) states, to investigate the novel finding that TOTs on particular words tend to recur for speakers, and examine whether this effect can be attributed to implicit learning of the incorrect mapping from a lemma to phonology for that word. We elicited TOTs by asking participants to supply the word that fit a given definition, and then retested participants on those same definitions in a second test. In Experiments 1-3 we investigated the time course of learning that occurs during TOTs, and found that TOTs are likely to recur with a five-minute test-retest interval, that this error learning can still be measured following a one-week interval, and that they recur for both monolingual and bilingual speakers. We also report the novel finding that error learning can be corrected when individuals resolve their TOT, making them less likely to re-experience a TOT for that word on a later test. In Experiment 4 we showed that these learning effects are not modulated by a priori knowledge of future tests. In Experiments 5a and 5b we show that orthographic cues can help individuals resolve their TOTs, and that these cued-resolutions lead to corrective learning in much the same way as self-resolutions. In Experiment 6 we demonstrate that effortful retrieval is critical to finding differences in error learning when manipulating the duration of unresolved TOTs. We conclude that the error-repetition effect is highly robust, and is best explained by implicit learning of the erroneous state. These findings reinforce the notion that the language production system is dynamic, and continually learning from experience, even when that experience is errorful.