Temporal and spatial variability in speakers with Parkinson's Disease and Friedreich's Ataxia

Abstract

Speech variability in groups of speakers with Parkinson's disease (PD) and with Friedreich's ataxia was compared with healthy controls. Speakers repeated the same phrase 20 times at one of two rates (fast or habitual). A non-linear analysis of variability was performed which used some of the principles behind the spatio-temporal index (STI). The STI usually employs variation in lip displacement over repetitions of the same utterance and a linear analysis of such signals is conducted to represent the combined variation in spatial and temporal control. When working with patients, audio measures (here we used speech energy) are preferred over kinematics ones as they are minimally disruptive to speech. Non-linear methods allow spatial variability to be estimated separately from temporal variability. The results are tentatively interpreted as showing that PD speakers were distinguished from healthy control speakers in spatial variability and ataxic speakers were distinguished from controls in temporal variability. These findings are consistent with the speech symptoms reported for these disorders. We conclude that the non-linear analysis using the speech energy measure is worth investigating further as it is potentially revealing of the differences underlying these two pathologies.

Figure 1

Mean phrase duration +/− 1 sd for each of the ataxic speakers and their healthy controls at both speech rates. Participant type and rate condition can be identified by the key in the inset.

Figure 2

FDA registration graphs for a PD participant (top left) and his healthy control (top right) and for an ataxic participant (bottom left) and her healthy control (bottom right). For each of the four participants shown, fast rate is at the top and habitual rate at the bottom. For each speaker by rate panel, the top row shows the superimposed energy tracks, the second row shows these same tracks after non-linear registration. In the last row, the x axis represents a linear scaling of time so all records fit on the same (arbitrary) time frame and the y axis is the non-linear deformation of the x axis resulting from FDA registration. Full details of what is plotted are given in the text.

Figure 3

Temporal (abscissa) versus spatial variability (ordinate) for each participant and rate condition for the PD participants and their controls (left) and ataxic participants and their controls (right). Participant type and rate condition can be identified by the key in the inset.