This study investigates the process by which young children learn to produce sequences of speech sounds in an adult-like manner. Learning to produce speech sounds requires precise motor control of several oro-facial muscles and the ability to link these motor activations both with articulatory movements in the vocal tract(via somatosensory feedback) and with acoustic realizations (via auditory feedback). During this first learning phase, a number of agonistic and antagonistic muscle synergisms aredeveloped. The acquisition of serial order in speech production implies, in addition to these skills,the ability to plan and execute sequences of motor commands with fine temporal coordination. In line with theories of optimal motor control, gesture planning is assumed to make use of neural representations (internal models) of the motor system in order to minimize some measure of cost in producing sequences of movements. In this framework, the efficiency of speech planning is considered to reflect the maturity of these neural representations.