Text this: Sparse and continuous sampling approaches to fMRI of overt vocalization tasks