Understanding Fluency in Aphasia

• One of the most common ways of describing aphasia is by fluency. • Both the BDAE and the WAB attempt to capture the multidimensional nature of fluency using ratings on dimensions such as prosody, paraphasia, grammaticality, word‐finding difficulty, and articulatory effort. • However, they have demonstrated poor agreement on aphasia classifications, including fluent vs non-fluent distinctions.1 • Perceptual ratings by practicing speech-language pathologists have identified several features that predict fluency judgements, including:  Grammaticality, articulatory effort, and word‐finding difficulties2  Speech productivity, speech rate, and audible struggle3 • We propose a shift away from the fluent/non-fluent dichotomous categorization toward a focus on identifying the underlying contributors to disrupted speech fluency. • Comparison of Spontaneous Speech Measures for PwA with Mismatching Clinician Fluency Impressions. Spontaneous speech measures were transformed to z-scores to facilitate comparison. • Mismatches (n=36) by Aphasia Type: Agreement on fluency category was 86% overall, similar for fluent (84%) and nonfluent (88%) aphasia.


Conclusions & Future Directions Results
• One of the most common ways of describing aphasia is by fluency.
• Both the BDAE and the WAB attempt to capture the multidimensional nature of fluency using ratings on dimensions such as prosody, paraphasia, grammaticality, word-finding difficulty, and articulatory effort.
• However, they have demonstrated poor agreement on aphasia classifications, including fluent vs non-fluent distinctions. 1 • Perceptual ratings by practicing speech-language pathologists have identified several features that predict fluency judgements, including:  Grammaticality, articulatory effort, and word-finding difficulties 2  Speech productivity, speech rate, and audible struggle 3 • We propose a shift away from the fluent/non-fluent dichotomous categorization toward a focus on identifying the underlying contributors to disrupted speech fluency.
• Comparison of Spontaneous Speech Measures for PwA with Mismatching Clinician Fluency Impressions.Spontaneous speech measures were transformed to z-scores to facilitate comparison.
• Participants included 254 people with aphasia (PwA) from the AphasiaBank database, representing a range of WAB aphasia types: • Objective measures of connected speech predicted to underlie fluency were extracted from the Cinderella stories using CLAN. All inter-correlations were <.500 or >-.500 to reduce collinearity.Aphasiology, 12(7/8), 673-688  Clinicians were influenced by apraxia, complex grammar, lexical diversity, and sex of the PwA.
• Linear Regressions: Fluency is often measured with the Spontaneous Speech rating scale of the WAB.This largely reflects severity.Lexical specificity and accuracy and grammatical complexity also contribute. Different variables affect ratings for fluent and nonfluent PwA.
• Logistic Regression*: WAB Fluency was predicted primarily by aphasia severity, empty speech, and the use of verb inflections.

Results (continued)
• Linear Regressions: Fluency is also measured by mean utterance length or speech rate.Both indices are themselves influenced by multiple (lexical, grammatical, and speech) dimensions.
 MLU is most strongly affected by severity and grammatical complexity.
 WPM is further influenced by lexical variables (VocD, empty speech).• Fluency categories based on the WAB largely reflect aphasia severity.

References
• Clinicians are sensitive to differences in a variety of spontaneous speech dimensions that the WAB does not capture. In making fluency judgements, clinicians are influenced by variables contributing to underlying components of fluency: grammatical competence, lexical retrieval, and speech production.
• By providing objective and standardized methods of capturing these underlying variables, we aim to improve diagnostic reliability of fluency.
This research was funded by a 2017 New Century Scholars Research Grant from the American Speech-Language-Hearing Foundation, awarded to both authors (Fluency Forward: Developing a More Reliable and Clinically Useful Assessment of Fluency in Aphasia).We would also like to acknowledge the founders of and contributors to AphasiaBank and all of the people with aphasia who have participated.Particular thanks to Davida Fromm for her help in navigating AphasiaBank and CLAN.