Computational Measures of Speech and Language for Schizophrenia Spectrum Disorders

Schizophrenia-spectrum disorders (SSDs) are characterised by disrupted emotional processing, reflected in speech and language.In this work, I investigate automated computational measures of SSDs using the Narrative Emotion Task, a semistructured interview eliciting emotional narrative speech.I report findings on emotional alignment (EA) scores and acoustic features for distinguishing healthy controls from individuals with SSD.I then outline a plan for repeated-measures prediction of symptoms using acoustic, linguistic, and affective speech features across five timepoints, moving towards a speech-based framework for longitudinal symptom monitoring in SSDs.