Behavioral correlates of language production in non-verbal tasks: Inner speech is observable via word frequency...

This study demonstrates that unprompted inner speech use is reflected in gaze durations during an audio-visual probe-recognition task. In Experiment 1, participants studied visually presented object arrays for six seconds and subsequently decided whether an auditorily presented word corresponded to an object in the study array. One object in the study array was designated as the target object. Target objects were selected based on the word frequency of their names, forming two categories: high- and low-frequency object names. Gaze durations were shorter for objects with high-frequency names than for those with low-frequency names. In Experiment 2, the same target objects were used in an overt object-naming task. Speech onset times mirrored the gaze duration effect observed in Experiment 1, thereby validating the interpretation of the gaze data. We argue that gaze durations offer a direct, real-time measure of covert linguistic encoding in a memory task, providing a new index of the established phonological rehearsal process. This enhances the observability of inner speech on a behavioral level compared to previous studies relying on articulatory suppression paradigms. We discuss the methodological implications of these findings in relation to psycholinguistic and other perspectives on inner speech.