Furthermore, she associated strings with particular situations. For instance, See you meant 'I'll be seeing you,' and it was used on occasions when her friends were parting. Attempts to lead her to combine it with I can to form I can see you were not only unsuccessful but confusing" (p. 79).

Ordinary analytic style grammatical development began subsequently. Hanania and Gradman report a slow development of sentence structure and morphology not at all unlike that seen in child first and second language acquisition: "Starting with simple structures made up of essential substantive units, she proceeded to build up her constructions by enlarging these units and linking them together...." (p. 82). Hanania and Gradman's summary statement concerning Fatmah's syntactic development is also worth repeating:

The adult in the present study proceeded to learn the language creatively. She did not simply imitate models of the language but acquired elements selectively and built them into syntactic units which became progressively more complex. The pattern of her linguistic development was similar to that of first language learners. Early constructions were constrained to two-term utterances, and the growth of sentence complexity occurred along the same lines (pp. 87-88).

The similarity of natural adult language acquisition has been noted elsewhere in terms of the acquisition order of certain structures (see Bailey, Madden, and Krashen, 1974; and also Krashen 1977a).

While Hanania and Gradman do not note the disappearance or reanalysis of routines and patterns in Fatmah's speech, propositional language appears to be independent of gestalt in her case, as in Brown and Wagner-Gough's subjects and in Paul, the child L2 acquirer discussed above. Language development appears to proceed analytically, in the "one word at a time" fashion, and routines and patterns are not mentioned after the early stages. (We are presupposing that the "gestalt" mode is available to both the child and the adult. It is possible, however, that this mode is manifested somewhat differently for these two classes of acquirers: the child L1 acquirer, as Peters notes, sometimes approximates whole sentences which are characterized by "a 'melody' unique enough so that it can be recognized even if badly mumbled" (Peters, 1977, p. 562). The adult may use routines and patterns, but not "tunes". This child-adult difference may have a psychological-affective basis (see Krashen, 1980a).)

Other than Hanania and Gradman's study, there are no direct data

97