| Fluent speech involves producing sound sequences that are composed from a finite alphabet of learned words, syllables, and phonemes. The brain thus requires machinery to organize and enact properly ordered and timed motor command sequences that correspond to the desired phonological plan. This dissertation seeks to provide an enhanced mechanistic understanding of this system through a combination of computational neural modeling and neuroimaging.; The first portion of the dissertation describes an experiment using sparse event-triggered functional magnetic resonance imaging (fMRI) to measure brain responses due to preparation and overt production of non-lexical three syllable sequences of varying complexity. The network of brain regions related to initiation, motor execution and hearing one's own voice was found to include the primary motor and somatosensory cortices, auditory cortices, supplementary motor area (SMA), insula, and portions of the thalamus, basal ganglia, and cerebellum. Additional stimulus complexity led to increased engagement of the basic speech network and recruitment of additional areas known to be involved in control of non-speech motor sequences, including the left hemisphere inferior frontal sulcus region and posterior parietal cortex, and bilateral regions at the junction of the anterior insula and frontal operculum, the pre-SMA, basal ganglia, anterior thalamus, and cerebellum.; These experimental results as well as previous clinical, behavioral, and imaging data were used to guide the development of a neural model of speech syllable sequencing based on a "competitive queuing" architecture. The new GODIVA (Gradient Order DIVA) model extends the DIVA model of speech production, which describes how individual speech items are learned and produced, to include explicit parallel representations for forthcoming utterances. GODIVA posits detailed neuroanatomical substrates and neurobiologically plausible mechanisms for its components. The model can thus account for a database of clinical and neuroimaging results beyond the scope of previous non-biological models.; Finally, preliminary efforts using magnetoencephalography (MEG) and surface electromyography (EMG) to obtain neuroimaging data that complements fMRI results and offers further modeling constraints are described. A novel algorithm was applied to detect neural source components that could be used to reliably discriminate between stimuli that necessitated the preparation of one, two, or three syllable plans. |
