A Novel Application of Natural Language Processing for the Assessment of Communication Outcomes Post-Stroke

Aphasia is an acquired neurogenic communication disorder resulting in an impairment across one or many modalities of communication. When assessing aphasia, speech-language pathologists (SLPs) utilize a blend of quantitative and qualitative measures to aid in clinical diagnoses and treatment outcomes. Analysis of discourse is a particularly important component to document language recovery. Computer aided text analysis (CATA) utilizing natural language processing (NLP) is an intersection of quantitative and qualitative research that aims to draw the thoughts and emotional attitudes from individual narratives and written texts. Due to the advancement and accessibility of software programming and computational powers, CATA has the ability to both investigate the superficial and latent semantic attributes of language embedded in a text sample. Furthermore, sentiment analysis, the automated process of deriving positive, negative, or neutral opinions from text, is one specific application of CATA. Past studies have applied sentiment analysis towards consumer-driven and marketing research. Fewer studies have researched how sentiment analysis can be applied to healthcare domains. The purpose of this exploratory study is to apply a methodology for programmatic analysis of the sentiment of transcribed post-stroke speech samples (text) and assess change over time.