Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Health and Rehabilitation Sciences


JB Orange


Background: The effects of disease (PD) on cognition, word retrieval, syntax, and speech/voice processes may interact to manifest uniquely in spoken language tasks. A handful of studies have explored spoken discourse production in PD and, while not ubiquitously, have reported a number of impairments including: reduced words per minute, reduced grammatical complexity, reduced informativeness, and increased verbal disruption. Methodological differences have impeded cross-study comparisons. As such, the profile of spoken language impairments in PD remains ambiguous.

Method: A cross-genre, multi-level discourse analysis, prospective, cross-sectional between groups study design was conducted with 19 PD participants (Mage = 70.74, MUPDRS-III = 30.26) and 19 healthy controls (Mage = 68.16) without dementia. The extensive protocol included a battery of cognitive, language, and speech measures in addition to four discourse tasks. Two tasks each from two discourse genres (picture sequence description; story retelling) were collected. Discourse samples were analysed using both microlinguistic and macrostructural measures. Discourse variables were collapsed statistically to a primal set of variables used to distinguish the spoken discourse of PD vs. controls.

Results: Participants with PD differed significantly from controls along a continuum of productivity, grammar, informativeness, and verbal disruption domains including total words F(1,36) = 3.87, p = .06; words/minute F(1,36) = 7.74, p = .01 , % grammatical utterances F(1,36) = 11.92, p = .001, total CIUs F(1,36) = 13.30, p = .001, % CIUs (Correct Information Units) F(1,36) = 9.35, p = .004, CIUs/minute F(1,36) = 14.06, p = .001, and verbal disruptions/100 words F(1,36) = 3.87, p = .06 (α = .10). Discriminant function analyses showed that optimally weighted discourse variables discriminated the spoken discourse of PD vs. controls with 81.6% sensitivity and 86.8% specificity. For both discourse genres, discourse performance showed robust, positive, correlations with global cognition. In PD (picture sequence description), more impaired discourse performance correlated significantly with more severe motor impairment, more advanced disease staging, and higher doses of PD medications.

Conclusions: The spoken discourse in PD without dementia differs significantly and predictably from controls. Results have both research and clinical implications.