Social Priming in L1 Japanese Speech Perception

Research on social priming in speech perception has focused on Western contexts and typically examined aggregate effects rather than temporal dynamics. To address these gaps, we investigated how social primes modulate L1 Japanese speech perception over time using a visual matched-guise transcription-in-noise task. University students in the Tokyo area transcribed native Japanese sentences while viewing one of four visual primes: an East Asian face, a Southeast Asian face, a Black face, or an East Asian face with a Chinese name. Results revealed nonlinear learning trajectories, with accuracy improving rapidly in early trials before gradually plateauing. Across the full session, learning rates did not differ significantly across prime conditions, although a sensitivity analysis indicated that condition differences in improvement smaller than roughly 15 percentage points could not have been reliably detected. An exploratory split-half analysis revealed that learning rates differed across conditions during the first 20 trials, with the Black prime yielding the steepest early improvement. These differences had disappeared by Trials 21 to 40, and overall trajectory shapes did not robustly differ across conditions. This pattern is consistent with the account that, in L1 speech perception, socio-phonetic information carried in the acoustic signal overrides listeners' initial expectations about the speaker's social group membership derived from the visual prime and generates the testable hypothesis that social information primarily influences the initial phase of exposure, a phase that aggregate measures systematically miss.