Date of Degree


Document Type


Degree Name



Speech-Language-Hearing Sciences


Douglas H. Whalen

Subject Categories

Cognitive Psychology | Communication


fluency; speech; stability; stuttering; variability


It is well known that people who do and do not stutter produce speech differently, at least some of the time, even when perceived as fluent. One way that investigators have assessed these differences is by measuring variability, or the inconsistency of repeated speech movements. Variability in speech has typically been quantified using linear analysis techniques (e.g., measures of central tendency), and results have indicated that people who stutter produce speech that is (sometimes) characterized by increased variability. However, variability is a complex phenomenon, one that cannot be assessed by linear methods alone. This dissertation employs linear and nonlinear analysis techniques to examine the nature of variability, stability, and flexibility in stuttering and non-stuttering speakers.

Two experiments are reported in this dissertation. The first is a pilot study in which 11 participants judged short utterances that were manipulated in gap (or pause) duration to be fluent or disfluent. This preliminary study facilitated the selection of 'fluent' utterances for the primary experiment, which measured lip aperture kinematics and acoustics for 20 speakers who stutter and 21 speakers who do not stutter, under two manipulations: 1) audience and non-audience; 2) increasing linguistic complexity.

Results from the primary experiment corroborated results from prior studies that used linear techniques to show that 1) adults who stutter exhibit more effector variability than adults who do not stutter when target utterances are embedded in sentences of increased linguistic complexity, and 2) linear acoustic measures are as effective as linear kinematic measures for quantifying variability. Nonlinear analysis techniques demonstrated that adults who stutter exhibit more deterministic structure in lip aperture dynamics. Furthermore, cognitive-emotional stress (i.e., the presence of an audience) resulted in decreased surface variability, increased deterministic structure, decreased stationarity, and decreased signal complexity in speakers who stutter, but not in those who do not stutter. Thus, adults who stutter appear to exhibit less overall stability, which leads to a more rigid, less flexible approach to speech production, especially when cognitive-emotional stressors are placed on their speech motor systems.

These findings highlight the benefits of using nonlinear analysis techniques to examine variability in speech production. Specifically, the results demonstrated that speech movements that appear to be less variable on the surface, may in fact be overly deterministic and nonstationary'two attributes that indicate system instability in complex biological systems. Thus, a combination of linear and nonlinear approaches is warranted in future investigations of speech production.