In the first nine months of life, infants are far from silent, as any sleep-deprived parent can attest. Around nine months of age, however, a significant change in the quality of sounds produced by the infant occurs. In addition to screaming, cooing, laughing, and other non-linguistic sounds, the child begins to utter speech-like sounds such as "ba" or "da," often in long nonsense strings of syllables. This "pre-linguistic speech" is called babbling. Babbling is thought to be the first stepping-stone to spoken language. By producing a variety of language-like sounds, the child is exercising his or her mental and vocal skills in a way that will provide the foundation for true speech.

Between nine months and one year of age, the child usually begins to produce whole words. At first, words are produced in isolation: the child will say "mama" in response to the presence of the mother or "cookie" to request a snack. Typically, the first words learned are monosyllabic; as the child grows, the phonological complexity of the words produced increases. In English, at least, nouns are learned earlier than verbs. Early in the word- learning process, children often make mistakes of overextension or underextension. In overextension, the child applies a word to a broader class of objects or actions than in adult usage. For instance, the word "dog" might be used to describe any four-legged animal, or the word "eat" might be used to describe both eating and drinking. Underextension is the exact opposite: the child applies the word to a smaller class of objects than he or she should. For instance, the word "cat" might be used to describe the family pet but not unfamiliar cats. Around eighteen months, the child begins to use multi-word utterances such as "give ball." Around two or three years, a "naming burst" occurs during which an enormous number of new words are acquired. One estimate is that one to two words are learned per waking hour between the ages of three and five.

The problems of grammar first appear when the child begins to produce multi-word utterances and, in English and many other languages, to alter single words in order to indicate number (the –s ending in English), gender (the –e ending in French), tense (-ed in English), or other inflections of meaning. Just as in word learning, children frequently make errors of overgeneralization. For instance, when English-speaking children learn that the –ed ending indicates past tense, they tend to use it for all verbs, including those that are irregular and do not take the –ed ending in adult speech (such as "go" or "think"). Only after extensive practice with both the rule and its exceptions does the child learn to speak as an adult.

In English, word order is a crucial part of grammar and the learning of word order rules can be tracked as infants produce increasingly large numbers of multi-word utterances between the ages of two and five. Even two-word utterances in the second year include some kind of grammatical information: the phrase "hit Bob" means something different to a child than "Bob hit." Children appear to be able to produce this kind of difference as soon as they begin to produce multi-word utterances, and they comprehend it even earlier. This brings up an important point in language acquisition and other parts of developmental psychology: an inability to produce a certain behavior does not mean that the corresponding cognitive structures are absent. Children are able to understand grammatically complex sentences and words long before they are able to produce them.