Abstract
Background: Although many speech errors can be generated at either a linguistic or motoric level of production, phonetically well-formed sound-level serial-order errors are generally assumed to result from disruption of phonologic encoding (PE) processes. An influential model of PE predicts that speaking rate should affect the relative proportion of these serial-order sound errors (anticipations, perseverations, exchanges). These predictions have been extended to, and have special relevance for persons with aphasia (PWA) because of the increased frequency with which speech errors occur and because their localisation within the functional linguistic architecture may help in diagnosis and treatment. Supporting evidence regarding the effect of speaking rate on phonological encoding has been provided by studies using young normal language (NL) speakers and computer simulations. Limited data exist for older NL users and no group data exist for PWA.
Aims: This study tested the phonologic encoding properties of Dell’s model of speech production, which predicts that increasing speaking rate affects the relative proportion of serial-order sound errors (i.e., anticipations, perseverations and exchanges).
Methods & Procedures: The effects of speech rate on the error ratios of anticipation/exchange (AE), anticipation/perseveration (AP) and vocal reaction time (VRT) were examined in 16 normal healthy controls (NHC) and 16 PWA without concomitant motor speech disorders. The participants were recorded performing a phonologically challenging (tongue twister) speech production task at their typical and two faster speaking rates.
Outcomes & Results: A significant effect of increased rate was obtained for the AP but not the AE ratio. Significant effects of group and rate were obtained for VRT.
Conclusion: Although the significant effect of rate for the AP ratio provided evidence that changes in speaking rate did affect PE, the results failed to support the model-derived predictions regarding the direction of change for error-type proportions. The current findings argued for an alternative concept of the role of activation and decay in influencing types of serial-order sound errors. Rather than a slow activation decay rate, the results of the current study were more compatible with an alternative explanation of rapid activation decay or slow build-up of residual activation.
Acknowledgements
The data presented in this manuscript represent data collected for the doctoral dissertation of the first author. Although this study was completed at the University of Pittsburgh, Malcolm R McNeil and Sheila R. Pratt provided resources and facilities through the Geriatric Research Education and Clinical Center in the VA Pittsburgh Healthcare System. Nonetheless, the contents of this manuscript do not represent the views of the Department of Veterans Affairs or the United States Government. A portion of the data presented in this manuscript was also included in an abstract submitted to the Academy of Aphasia and previously published in Brain & Language (2007). The authors wish to acknowledge the multiple recruitment sites involved in facilitating data collection for disordered participants, with special acknowledgement to Drs. Nadine Martin and Myrna Schwartz at Temple University and Moss Rehabilitation Hospital, respectively, and Dr. Julie Wambaugh, at the VA Salt Lake City Healthcare System. Additionally, they wish to thank Dr. Julie A. Fiez in the Department of Psychology at the University of Pittsburgh for her contributions to the project.
Notes
1. Distortions can also be perceived of as sound-level substitutions and therefore substitutions cannot unambiguously be attributed to the PE level of production (McNeil et al., Citation2009). These authors proposed that distorted sound substitutions or distorted sound sequencing errors occur when both phonological and motor-level processing are disturbed. The differential diagnostic value of undistorted sound sequencing errors is that they, as compared with other error types, can more confidently be attributed to disruption of the PE (linguistic-symbolic) level of production (McNeil et al., Citation2009). The models of Dell (Citation1986, 1988) and Dell and colleagues (Dell, Burger, et al., Citation1997) are consistent with this level of error assignment.