To explore the mechanisms of speech articulation, which is one of the most sophisticated human motor skills controlled by the central nervous system, we investigated the force-generation dynamics of the human speech articulator muscles [orbicularis oris superior (OOS) and inferior (OOI) muscles of the lips]. Short-pulse electrical stimulation (300 micros) with approximately three or four times the sensation threshold intensity of each subject induced the muscle response. The responses of these muscles were modeled as second-order dynamics with a time delay (TD), and the model parameters [natural frequency (NF), damping ratio (DR), and TD] were identified with a nonlinear least mean squares method. The OOS (NF: 6.1 Hz, DR: 0.71, TD: 14.5 ms) and OOI (NF: 6.1 Hz, DR: 0.68, TD: 15.6 ms) showed roughly similar characteristics in eight subjects. The dynamics in the tongue (generated by combined muscles) also showed similar characteristics (NF: 6.1 Hz, DR: 0.68, TD: 17.4 ms) in two subjects. The NF was higher, and the DR was lower than results measured for arm muscles (NF: 4.25 Hz, DR: 1.05, TD: 23.8 ms for triceps long head), indicating that articulatory organs adapt for more rapid movement. In contrast, slower response dynamics was estimated when muscle force data by voluntarily contraction task were used for force-generation dynamics modeling. We discuss methodological problems in estimating muscle dynamics when different kinds of muscle contraction methods are used.