August 2012

Document Type


Degree Name



Dept. of Biomedical Engineering


Oregon Health & Science University


As the prevalence of neurological disorders such as dementia, autism, and language impairment increases, so will the demand for simple, objective, and unobtrusive screening tools for these disorders. The automated analysis of narratives shows potential as a component of such a screening tool, since the ability to produce accurate and meaningful narratives is impaired in these populations. This dissertation investigates the reliability and diagnostic utility of automatically analyzing responses to a clinically elicited narrative retelling task, in which a subject listens to a brief story and must verbally retell the story to the examiner. In order to establish the utility of using narrative retellings for diagnostic classification, we first demonstrate that manually assigned narrative recall scores can be used to accurately identify subjects with mild cognitive impairment and language impairment. We then present a method for extracting narrative recall scores automatically and highly accurately from a word-level alignment between a retelling and the source narrative. We propose improvements to existing machine translation-based systems for word alignment, including a novel method of word alignment relying on random walks on a graph of interconnected nodes, which achieves alignment accuracy superior to that of standard expectation maximization-based techniques for word alignment in a fraction of the time required for EM. In addition, the narrative recall scores and related narrative fidelity features extracted from these high quality word alignments yield classification accuracy comparable to that achieved using manually assigned scores and significantly higher than that achieved using automated scoring techniques proposed in the literature. Finally, we apply these automated scoring and classification methods to spontaneous language samples elicited in other neuropsychological instruments, thereby demonstrating the flexibility and generalizability of these methods.




Center for Spoken Language Understanding


School of Medicine



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.