Abstract

This study explored various methods of evaluating language style for integration into prompts designed for a sophisticated augmentative and alternative communication (AAC) device. A total of 395 individuals without a language disorder were recruited via a crowdsourcing platform to complete five surveys. The initial four surveys were designed to assess individuals' language style using different approaches (questionnaire developed with ChatGPT, self-written, questionnaire developed by Hartman and McCambridge, and choosing tags). Following the completion of these four surveys, a fifth survey was dynamically created based on the responses in the previous four surveys. Each possible response in the fifth survey was presented in a language style corresponding with the results of the previous four surveys. Results indicated that the most desired responses came from a machine learning method that required participants to spontaneously write responses based on given prompts. Participants were then asked about the quality of the responses selected. The majority of participants reported that the responses "definitely" or "probably" reflected their preferred style of language. While the self-written method yielded the most accurate matches, it was also rated as the most difficult, whereas the tag selection method was considered the easiest. These findings suggest that artificial intelligence (AI) has promise for enhancing the personalization of AAC devices. However, the cognitive and linguistic demands associated with more effective methods, such as spontaneous language production, may present

Degree

MS

College and Department

David O. McKay School of Education; Communication Disorders

Rights

https://lib.byu.edu/about/copyright/

Date Submitted

2025-06-13

Document Type

Thesis

Handle

http://hdl.lib.byu.edu/1877/etd13714

Keywords

augmentative and alternative communication, aphasia, artificial intelligence

Language

english

Download

Included in

Education Commons

Share

COinS