Auditory Modeling Improves Aphasic Speech Production Recovery

Recent research has shown that melodic intonation therapy enhances speech output in non-fluent aphasic patients. Improvements have occurred in the successful production of words and phrases as well as spontaneous speech output. However, these patients typically also have deficits in the articulatory implementation of speech, often making it difficult to understand what they are trying to say. Here we developed a protocol adapting the melodic intonation procedure to examine whether modeling the speech output of the examiner results in an improvement in the production of the sounds of language. Four participants (2 nonfluent and 2 fluent aphasics) were trained on the production of English words beginning with voiced and voiceless alveolar stop consonants ([t-d]) in two exposure sessions in which patients produced utterances modeled by the examiner, one with normally intoned but slowed input and the other with intoned and slowed input. Acoustic analysis of a set of words produced in a carrier phrase before and after training showed improved articulatory implementation of voiced and voiceless stop consonants in the trained words as well as generalization to a new set of untrained words beginning with a  different place of articulation, i.e. [p b], [k g]. In addition, patients produced fewer speech sound errors and had fewer stutters in both the trained and untrained words after the exposure sessions.