We use language to teach a child how to spell shoelaces, but not how to tie them. Children learn how to tie shoelaces by observing us do it. One problem with learning how to speak is that speaking involves complex internal mouth and vocal system movements that children can't see. What's wonderful about our brain is that information that enters one sensory modality easily transfers to others. So if we hear the word shoelace, we can visualize it in our mind, even though no shoelace is visible. Learning to speak makes use of this cognitive transfer property.

Hearing articulate speech activates the same mirror neurons in the child's brain that the speaker used to sequence the uttered sounds and words. Speech (like learning to tie shoelaces) is a complex motor activity, so the infant initially babbles in awkward incoherent imitation, but over time in a verbal environment, the child begins to correctly utter simple phonemic combinations, and finally smooth articulate speech emerges.

We enhance this learning process with motherese, a universal behavior in which an adult holds an infant in a close face-to-face position, and then speaks in a slowed-down, high-pitched, exaggerated, repetitive, melodic format that engages the infant's attention and so easily activates her mirror neuron system.

Children who have specific auditory deficiencies will be delayed in the development of their ability to understand speech, and so to develop speaking competence. Computerized interventions such as Fast ForWord products can slow down incoming speech to match the child's auditory rate, and then gradually increase it as the child's brain adapts to the faster rates.