Automated analysis of lexical features in Frontotemporal Degeneration

We implemented an automated analysis of lexical aspects of semi-structured speech produced by three patient groups with Frontotemporal degeneration (FTD): behavioral variant FTD (n=74), semantic variant Primary Progressive Aphasia (svPPA, n=42), and nonfluent/agrammatic PPA (naPPA, n=22). With a natural language processing program, we automatically tagged part-of-speech categories of all words and rated nouns for lexical measures, and computed the cross-entropy estimation, which is a measure of word predictability. Our automated analysis was a valid reflection of manual scoring. For svPPA patients, we found fewer unique nouns and more pronouns and wh -words than in the other patient groups and the controls; high abstractness, ambiguity, frequency, and familiarity for nouns they produced; and the lowest cross-entropy estimation among the groups. These measures were associated with cortical thinning in the left temporal lobe. In naPPA patients, we found increased speech errors, which were associated with cortical thinning in the left middle frontal gyrus. bvFTD patients were similar to the controls. Our results underline distinct word use profiles in subgroups of PPA patients and validate our automated method of analyzing FTD patients’ speech.