A motor learning perspective on phonetic syllable kinships: How training effects transfer from learned to new syllables in severe apraxia of speech

Abstract

Background: There is strong evidence for a syllable-based pathomechanism in patients with apraxia of speech (AOS). The apraxic impairment is localised to the phonetic encoding level where Levelt, Roelofs, and Meyer (1999) postulate a mental store of holistic, ready-made motor programs for syllables. In a recent learning study we showed that complex target syllables benefit from a training of a set of phonologically related, less complex syllables (Aichert & Ziegler, 2008). This outcome is incompatible with the view that syllabic motor plans are holistic phonetic entities, but rather suggests an internal structuring of speech motor programs. Aims: The present study further explored the structural relationship between syllables exercised in a learning trial and new syllables profiting from these exercises, in order to gain deeper insight into the internal architecture of syllabic speech motor plans. We used four different syllable structure models to test a range of theoretically relevant cross-syllabic transfer conditions. Methods & Procedures: We conducted a learning experiment with three patients with severe AOS. During the learning trials each patient practised four different syllables with a high number of repetitions. The training syllables were systematically related with an untrained set of transfer syllables which were controlled for their subsyllabic overlap with the training syllables. Accuracy of the training and transfer syllables was tested before and after the training. Outcomes & Results: Whereas one patient did not improve after the training, two patients showed significant learning and transfer effects. The results revealed that the transfer was confined to syllables with a position-true overlap on syllable constituents of the learning and the transfer syllables. More specifically, the inclusion of coarticulatory adjustments was crucial for the training success. Conclusions: The phonemes constituting the training syllables were not learned as segments, but rather as syllable components. The results provide further evidence for internally structured syllabic motor plans in which phonemes are associated to syllabic nodes such as onset or rhyme. Due to the small number of patients further research is necessary to replicate these results.