of routines and patterns provided the basis for syntax, while morphology appeared much later:

In the development of productive structure, the children all seemed to be following the strategy of working the major constituents first and dealing with the grammatical details later.... The process of gradual analysis by which parts of formulas become freed from their original frames yields sentence patterns... in the process, some of the grammatical morphemes and processes are unnoticed and lost (p. 656).

Fillmore documents many cases of these processes, and the interested reader is urged to consult her forthcoming book. We present here only a single example. Nora, Fillmore's fastest language acquirer, had these two formulas

   (1) I wanna play wi' dese.
   (2) I don' wanna do dese.

She then discovered that the constituents following wanna were interchangeable, and that she could say

   (3) I don' wanna play dese

and

   (4) I wanna do dese.

She thus acquired the patterns "I wanna VP" and "I don' wanna VP". "Play with NP" then became a formulaic verb phrase unit" which could be used for other slots requiring VP, e.g. "Le's VP", as in:

   (5) Le's play wi' that one.

This "formula-bases analytical process... was repeated in case after case of the children's spontaneous data" (p. 645). "Rules" came about when all the constituents of a formula were "freed".

Why did these children follow such a route? This question can be subdivided into several smaller ones. First, one may ask why so many routines and patterns occurred. Fillmore points out that the linguistic environment of the classroom and playground was conducive to the learning of routines and patterns. The daily classroom routine, for example, allowed the acquirers to figure out what was being said easily--all teachers followed, to a larger extent, predictable routines. "Such language, because it is used daily and with only minor variation, becomes highly predictable. The children can figure out

94