Loading...
Citations
Altmetric:
Genre
Thesis/Dissertation
Date
2023-12
Advisor
Committee member
Group
Department
Communication Sciences
Permanent link to this record
Collections
Research Projects
Organizational Units
Journal Issue
DOI
http://dx.doi.org/10.34944/dspace/9495
Abstract
Measures drawn from language samples (e.g., discourse measures) are used in clinical and research settings as a functional measure of language and cognitive abilities. In narrative elicitation tasks, discourse measures reliably vary by the type of prompt used to collect a language sample. Additionally, language features tend to very along with communicative context, topic, and modality (e.g., oral vs. written). However, until recent years, technology had not advanced sufficiently to support large-scale study of spoken language data. In this project, we used natural language processing and machine learning methods to examine the intersection of discourse measures, language modality, and cognition (i.e., working memory) in healthy young adults. In Experiment 1, we used a computational approach to examine discourse measures in spoken and written English. We achieved >90% accuracy in binary classification (e.g., spoken/written). In Experiment 2, we took a behavioral approach, studying working memory and narrative discourse measures in a cohort of healthy young adults. We predicted that working memory would predict informativity in participants’ narrative language samples. We found mixed results for our two measures of informativity (e.g., the Measure of Textual Lexical Diversity and Shannon entropy). We attributed the observed differences in these two measures to the fact that, while both serve to measure new or unique information, MTLD indexes additional linguistic information (e.g., semantic, lexical). In contrast, Shannon entropy is based on word co-occurrence statistics. We interpret our overall results as support for the potential utility of machine learning in language research and potential for future research and clinical implementations.
Description
Citation
Citation to related work
Has part
ADA compliance
For Americans with Disabilities Act (ADA) accommodation, including help with reading this content, please contact scholarshare@temple.edu