Influence of social and semantic contexts on phonetic encoding in naturalistic conversations

Social interactions occupy a significant part of life, and understanding others' conversations is key to navigating our social world. While the role of semantics in speech comprehension is well-established at the word or sentence level, its influence on larger conversational time scales, alongside social context, is less understood. The present study examined how semantic and social contexts modulate phonetic encoding during natural conversations using a speech-in-noise paradigm. Participants listened to AI-generated dialogues (two speakers) or monologues (one speaker) in an intact or sentence-scrambled order. Each trial contained five sentences, with the fifth sentence embedded in multi-talker babble noise. The same sentence was then repeated without noise, with one word either altered or unchanged. Healthy adults identified whether the sentence matched the in-noise version. Through several online experiments (N = 211), both social and semantic contexts showed influences on speech-in-noise processing, with improved performance for dialogues over monologues and for intact over sentence-scrambled conversations. These results suggest that both semantic and social factors shape speech comprehension, emphasizing their role in auditory cognition. This finding raises important questions about predictive and other mechanisms involved in processing complex, multi-sentence conversations, underscoring the critical role of social interaction in communication.