Proceedings of the Annual Meeting of the Cognitive Science Society

This study introduces a novel approach for quantifying individual differences in print exposure through the integration of distributional semantics with the Author Production Test (APT). By employing the Universal Sentence Encoder to generate vector representations of authors from their works, we constructed 'participant vectors' reflecting the aggregated author vectors individuals produced in the APT and 'genre vectors' capturing the representative characteristics of each literary genre. By analyzing the cosine similarities between participant and genre vectors, we objectively estimated individuals' genre preferences. The results demonstrated a significant correlation between these objective measures and self-reported genre preferences, particularly for older frequent readers, highlighting the method's effectiveness. Our findings offer a promising avenue for the objective measurement of print exposure, with potential implications for developing personalized models of lexical behavior.