Year: 2025
Type: article
Abstract: Tip-of-the-tongue states (TOTs) commonly occur when a word cannot be retrieved, despite the conviction that it is known and will emerge imminently. Past research has probed access to target information during TOTs, but these tasks typically use definitions, common knowledge questions, or pictures to cue recall. This mimics naturalistic language use, but this semantic input creates challenges for observing the retrieval of semantic information. In Experiment 1a (N = 85), participants studied a list of uncommon words before completing a cued-recall task using vague clues to elicit TOTs thus limiting semantic input. When recall failed, participants rated their mental state before attempting to pick a semantic associate in a 4AFC task. In Experiment 1b (N = 88) recall was cued by presenting one item from studied word pairs, providing no direct information about the targets. The proportion of associate choice was significantly higher after reporting TOTs compared to forgetting in Experiment 1a but not 1b. The latter design was replicated in Experiments 2 (N = 90) and 3 (N = 86) using phonological and orthographic associate words, respectively. Finally, Experiment 4 (N = 86) included a semantic, phonological, and orthographic associate in the 4AFC task to place all three modalities in mutual competition. The results showed that TOTs benefited only phonology in isolation, but orthography was preferred when more lexical information was accessible. The experimental data provided a basis to evaluate a computational model of lexical retrieval, with the aim of simulating the observed recall and TOT phenomena. The model’s lexicon included vector representations for the semantic, phonological, and orthographic relationships between 52,176 English words. Lexical retrieval required an initial semantic step which acted as a gate to the phonological and orthographic features of the word. Simulations of the empirical results using the same experimental stimuli showed promise, but the model struggled to capture the recall and 4AFC outcomes with a single set of parameters. Adaptations are proposed to better capture the progress of TOTs over time and the shift in attention from semantics towards phonology and orthography as retrieval unfolds.
Source: Mspace (University of Manitoba)
Author: Ball, Angus
Institution: University of Manitoba
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7137464095