What Bierwisch is stating has been called the "universal primitives hypothesis." E. Clark looks for clues about the kinds of semantic features that might underlie early word acquisition in studies of the development of perception in infants. She cites studies reviewed by Gibson (1969), such as those of the infant coming to recognize faces, as evidence that perceptual development begins with the isolation of single perceptual features and proceeds to the use of configurations of features. Gibson summarizes the work on infants' recognition of faces:

Development seems to proceed from simple contours to differentiated features to structural relations or patterns to unique patterns of individual faces and finally to higher order properties invariant over different individual faces. (1969, p. 347)

E. Clark believes that this type of development of looking is recapitulated when the child begins interpreting his perceptions to use them in attaching a meaning to a word. She writes:

The child therefore begins by using a single general feature, such as shape or contour, and considers that to be the "meaning" of some term. As he becomes compelled to differentiate more meanings, he can no longer simply use a single perceptual feature: he must begin to use more than one and eventually will encode the information from a bundle or combination of features. (1971a, p. 71)

Both E. Clark (1971a, 1973a) and H. Clark (1970, 1973) apply the semantic feature hypothesis, not only to nouns, but also to relational terms. E. Clark (1971b, 1973a) has also attempted to apply it to verbs, as has Kleiman (1973). There are many additional difficulties in trying to define the semantic features of verbs, so these will not be discussed further.

As an example of relational terms, the finding of Donaldson and Wales (1970) that three-year-olds often interpret "less" to mean "more" is useful. H. Clark (1970) claims that "less" and "more" have a common feature, which he calls +Amount. When the child is confusing the two words, he has only this feature attached to both. Later, to differentiate them, he must learn to attach a feature Clark calls +Polar to "more" and -Polar to "less." H. Clark (1973) also has done an excellent, highly detailed analysis of children's acquisition of terms designating space and time. He also details how the child's perceptions influence the acquisition of these terms: