There is evidence that children are able to understand different words before they are able to say them. Babies appear to have an innate ability to distinguish between linguistically different sounds, and can do this better than adults in some cases ("Japanese infants can distinguish between /r/ and /l/ while their parents cannot" - Fromkin). But having a vocabulary is not much use unless the child is able to vocalise in a recognisable way.
The mouth is capable of producing more different types of noise than any other part of the body, and from a very early age, babies make noises with their mouths (burbling, smacking lips, burping, etc.) which are typically encouraged by parents. Within the first year of life, before the baby starts to use what might be recognised as "words", he will progress through various stages, typically described as:
~0-2 months | reflexive sounds reflect the baby's state - e.g. the noises made when it breathes, swallows, burps, etc.. Sometimes referred to as 'vegetative' sounds. |
~2-4 months | the cooing stage is characterised by sounds made "generally when the baby is in a settled state" (Crystal) - typically a vowel sound preceded by a velar consonant, e.g. "koo" or "goo" |
~4-6 months | vocal play is when the infant varies back consonants with front vowels, and vice versa. E.g. /kI:/ and /ma:/. Raspberry sounds (which require a significant degree of motor control) may also occur here. |
~6-9 months | during the canonical stage, there is an emergence of syllabic type sounds "mama". Front vowels become more prominent |
~9-15 months | variegated stage sees changes of consonants and vowels in adjacent syllables, e.g. paba, or dadi. These may not be real words, or represent real words. The infant may also be using intonation throughout an utterance in a way that mirrors the expression heard in adult speech |
The infant's first "words" appear in the 9-15 month timeframe. In this context, a "word" is a stable sound representing a stable meaning - it may not be a recognisable English word, and will be constrained by the sounds that the infant is physically capable of making. These words may exhibit homophony, i.e. one utterance having different meanings, and reduplication, where a syllable is repeated, e.g. "didi" for "scissors" and "house".
There may be rules describing the way that children tend to simplify proper words, e.g.