The experience of listening to someone speak a totally unfamiliar language is a common one. The strange language appears to be merely a flow of sounds, virtually indistinguishable from each other, definitely not reproducible by the listener. Often one is aware that he cannot even find the boundaries between the words. In fact, Neisser (1967) points out that the flow of speech does not really have physical boundaries where we listen to them once we have learned to aud. He writes:
When the stream of speech is displayed in spectrograms. . . it displays a rather disconcerting continuity. We think of speech as made up of successive words, and words are composed of successive "sounds," but such parts are by no means always obvious when a spectrogram is examined. (p. 179)
We might expect that the child has a similar experience when first encountering adult speech. Somehow he must learn to distinguish the recurring sounds of the language, and to reproduce them. Then he must learn to segment the speech flow, that is, to find the divisions between the words.
Although it is true that at least some portion of the speech the child hears is more slowly spoken, more clearly articulated, and grammatically simpler than normal adult conversation (Snow, 1972), it is not currently clear how the infant learns to produce the sounds and to segment the speech flow into words. It is known, however, that the child's listening processes must give him the ability to recognize recurring verbal sounds, and his uttering processes must give him the ability to reproduce these sounds. Perhaps they work together and he gradually approximates the phonemes by comparing the sounds he produces with those he hears, and altering the productions to obtain a closer and closer match. This problem is discussed further in Chapter IV, pp. 55-58.
Sometime after the babbling stage, the child begins to produce utterances composed of single words. One view of these early one-word utterances is that there is a full conceptualization underlying them. The child is attempting to express what would require a full sentence; he has the cognitive capacity to conceptualize it, as well as the intention to express it, but lacks the linguistic capability. That is what is generally meant when one-word speech is called "holophrastic." This view has long been popular, having been supported by Stem and Stem (1907), De Laguna (1927; see also Brown, 1973), Leopold (1949), and numerous others.