Quantifying Semantic and Affective Flow Between and Within Speakers:
Two New Tools for Integrating Language and Neuroscience Research
Many theoretical inferences in psycholinguistics are drawn from experimental paradigms divorced from real world language processing demands. One substantive challenge involves bridging data from single words with more ecologically valid approaches that better account for connections between words in stories and in conversation. I will present two new tools for analyzing semantic transitions between words (semdistflow) and computing alignment between speakers engaged in dyads (ConversationAlign). Both approaches transform raw language data to quantitative time series, allowing for causal modeling between and within speakers (e.g., Speaker A causes changes in Speaker B’s average word length, concreteness, happiness, etc.). We will discuss numerous applications to language and neuroscience, including modeling semantic deficits using implicit language samples, yoking biosignals (e.g., pupil diameter) to word transitions, and measuring alignment (and causes of misalignment) in intergenerational and intercultural dyads (e.g., older adults speaking with teenagers).