was said, which became a routine for Adam. After several weeks of hearing this, Adam was heard to say

   (2) Wait for it to dry

when hanging up a washcloth. Also, many of Adam's long sentences were the result of juxtaposing existing routines. For example,

   (3) I want you to get a biscuit for me

consists of the pattern "I want _______" plus the underlined portion, which existed as an independent routine for Adam.

Clark claims that routines and patterns were primary in Adam's development of language: "In Adam's speech at this time a number of routine sequences seemed to coexist with a few simple productive rules. Many, though not necessarily all, the productive rules originated as invariable routines, which were in use for some time with the original lexical items before new lexical items were inserted" (p. 4). Thus Clark, while not denying the existence of a separate creative construction process, supports position 3 as the central means for language development for Adam. This seems to be in conflict with Brown's conclusion.

The work of A. Peters (1977) helps to resolve this apparent conflict. Peters distinguishes "analytic" and "gestalt" styles of first language development. The analytic style, which is used for referential, labeling functions, is the "one word at a time" development described in most studies of child language acquisition, such as those by Brown and his co-workers. The "gestalt" style, on the other hand, is the attempt to use whole utterances in a socially appropriate situation; it is thus used in more "conversationally defined" contexts. Peters suggest that there is individual variation among children as to which style will predominate. One of the factors that may determine which style a child will prefer is input type: the "analytic" child may have received clear caretaker speech, while the "gestalt" child may have received more rapid, conversational input.

A given child may use one style for one situation and the other for other situations: Peters' subject Minh appeared to speak analytically when naming, and used the gestalt style for social functions. Other investigators (Dore, 1974; Nelson, 1975) have also noted this kind of

88