to have a more limited use in actual speech performance than propositional language, all performers must eventually come to generate creative utterances at some point in their language development.

Automatic Speech in Child Second Language Acquisition

The impression one gets from the literature on child second language acquisition is that the second language performer relies far more on routines and patterns than does the first language acquirer. As we shall see below, Hatch and Hakuta provide evidence and slightly different, although complementary, reasons for this phenomenon. Then we will examine a study by Wagner-Gough which supports position 2. After this, we will review what is easily the most thorough treatment of routines and patterns in the language acquisition literature, L. Fillmore's documentation of routines and patterns in five Spanish-speaking children acquiring English as a second language.

Hatch (1972) describes the case history of Paul, a 4-year-old acquirer of English as a second language (see also Huang, 1971; Huang and Hatch, 1978). Paul, a Chinese speaker, was exposed to English in an informal environment, interacting with American children in school and in the playground. He made early use of the gestalt style. During the first month, "it seemed as if Paul was learning by imitation. He might repeat the sentence immediately after the other person said it, or he might remember it and use it later in the appropriate situation" (p. 31). "Propositional" speech appeared in the second month and looked quite similar to the analytic speech one generally sees in descriptions of child first language acquisition. Some typically analytic sentences included

   This kite.
   Ball no.
   Paper this.
   Wash hand?

At the same time, Paul was using complex routines such as

   Get out of here.
   It's time to eat and drink,

90