We are able to push the spatial resolution of dynamic speech magnetic resonance imaging to 2-mm near-isotropic level with 64 mm coverage of 32 3D slice locations that are spaced 2-mm apart with 35 fps. We choose to analyze lingual differences of American English voiced lateral [l] and (central) [t]. Several analysing methods are utilized such as magnitude comparison, t-test and deformation map comparison. The results give us detailed observations of lingual articulatory differences such as tongue grooving, twisting and coarticulation. Through this high spatial and temporal resolution, we demonstrate that this method will show great potentials on linguistic research.