Phase III Development of a Valid, Reliable, Clinically Feasible Measure of Transactional Success in Aphasic Conversation: Modernizing Methods of Acquisition and Analysis of Discourse Data - SUMMARY Models for assessing and treating aphasia, a language disorder affecting at least 2.5 million Americans, have gradually moved from an impairment-based to a participation-based framework. However, there exist no valid, reliable, clinically practical tools for directly measuring gains in conversation, the most frequent communicative activity of daily life in older adults. The long-range goal of the proposed study is to provide a theoretically grounded, practical, automated tool for clinicians and clinical researchers to assess changes in transactional success in conversation without transcribing and analyzing aphasic discourse, a time- and resource-intensive skill that hinders the inclusion of conversational discourse in treatment outcomes. We propose to fill this gap with the recently piloted Brief Assessment of Transactional Success in conversation (BATS). After watching and/or listening to short (2.5 minute) BATS video/audio stimuli, people with aphasia (PWA) retell the story in a monologic discourse task and then engage with a conversation partner who is naïve to the story, who then retell the story. The objectives of the proposed study are twofold: Aim 1 will demonstrate the reliability of the BATS and its validity in assessing transactional success. Reliability coefficients will be calculated for alternate forms of the video/audio stimuli as assessed by core lexicons, main concepts and comprehensiveness ratings. The magnitude of multiple potential sources of measurement error will be estimated to identify optimal data collection conditions (e.g., conversation partner familiarity). Evidence of converging and discriminant validation will evaluate the degree to which the BATS measures constructs such as joint action, total communication, and context, the three core elements that characterize the situated language use model of communication, and that enable PWA to communicate better than they talk. Aim 2 will advance automated methods of studying discourse in aphasia. Text analysis tools from open-source natural language processing platforms will interface with Python programs to assess core lexicons and topic similarity in conversation partner story retells. Pilot data suggests that use of automated methods for assessing micro- and macrolinguistic features of discourse to reveal transactional success in conversation in aphasia is feasible. This research is innovative because it applies 21st century videoconferencing technology to deliver a novel tool for assessing transactional success in conversation, while developing an automated software application to make the tool accessible to clinicians and clinical researchers who serve PWA, even in remote areas. The contribution is significant in that it will overcome a major obstacle in measuring real-world functional response to therapy in aphasia. Incorporating the three levels of BATS discourse (PWA story retells, dyadic conversations, and conversation partner retells) into AphasiaBank will create a rich, shared online database for studying how PWA succeed in co-constructing meaning in conversation. This project reflects current priorities in Areas 3 and 4 in the NIDCD strategic plan for Voice, Speech, and Language Research.
