Identifying and treating phonological and motor errors in acquired speech impairment

Introduction Determining whether an error arises during phonological or motor-based processing is an important theoretical and clinical issue. Buchwald and Miozzo (2011, 2012) demonstrated that acoustic/articulatory properties of errors may indicate the level the errors arise. They reported two individuals who made similar errors (/s/-deletion from /s/-stop clusters, speak → [_pik]) and distinguished errors reflecting the acoustic/articulatory properties of a singleton (i.e., aspirated [p]) and those reflecting the properties of a cluster (i.e., unaspirated [p], similar to American English [b]). They reasoned that errors with the timing of singletons reflected deletions taking place during phonological processing (before motor planning was initiated) whereas errors with the timing of clusters reflected deletions that took place after motor planning initiated, reflecting the target cluster. In a post hoc analysis, they discovered that weekly repetition sessions benefited the cluster production accuracy of the individual with motor planning errors, but not the individual with phonological errors. Here we report an expansion of these findings in which we both examined additional consonant cluster errors, and systematically investigated the differential effect of the repetition task on accuracy in producing consonant sequences. We present a series of cases for whom we identified error types on the basis of the acoustic analyses discussed above and evaluated change in cluster production accuracy after repetition-treatment sessions. The individuals who predominantly produced phonological errors showed no accuracy change, whereas the individuals with predominantly motor-level errors improved in cluster production accuracy. Participants Basic information about the five participants is presented in Table 1. Each individual suffered a single left-MCA stroke. Method and procedure We designed a repetition list containing words with 12 different stop-initial and fricative-initial consonant clusters (N=244). Two baseline repetition sessions were conducted where we tested each stimulus. During training, we presented words beginning with each cluster (N=183), and training lasted 10 weeks. Training consisted solely of repetition of the words in randomized order each week. Two post-training sessions were conducted to test both trained and untrained words. The accuracy change from pre-test to post-test was computed. All deletions within any cluster were extracted for acoustic analysis; we compared the duration of the remaining segment in the deletion (e.g., /s/ in slum→s_um) with that same segment in accurately produced clusters (slum) and in singletons (sum) where possible. Results Table 1 presents the results from both components of this analysis. The two individuals who produced phonological processing errors (i.e., singleton timing) did not change in accuracy at post-test. The three other individuals whose errors indicated motor planning errors (i.e., cluster timing) improved their cluster production accuracy for trained and untrained words (M1, M2) or substantially decreased their self-corrections. Discussion This study replicated and expanded on previous work indicating that the acoustic/articulatory properties of errors can help identify error types and predict responsiveness to treatment via repetition. These crucial findings are consistent with speech motor learning research (Maas et al., 2008) and suggest a promising direction for exploring the differential diagnosis and treatment of acquired sound production disorders.