variation. Nelson, for example, distinguishes a "referential style" used by children who are more oriented to things, objects, and actions on objects from an "expressive" style, used by children who are personal-social oriented. Perhaps the latter is related to Peters' gestalt style and the former to the analytic style.

Peters' analysis is strengthened by Dore's (1974) analysis of two child L1 acquirers, M (female) and J (male). While M produced words during the period her speech was studies, J produced more primitive speech acts, that is, he tended to make more use of language for communication, often using intonation alone. J's language use also tended to involve other people more than M's did; he used language more instrumentally than M, who was more prone to "label, imitate, and practice words" (p. 628). Input for the two children was, to some extent, different. M's mother "set up routines in which she would pick up an item, label it, and encourage her daughter to imitate it" (p. 627). "J and his mother did not participate in word-learning routines" (p. 628). Dore suggests that "there may be two partly separate lines of development--word development versus prosodic development" (p. 628). The diagram below depicts parallels in terminology among Peters, Nelson, and Dore.

________________________________________________________________________
            Analytic language               Gestalt language
________________________________________________________________________
--"one word at a time" development   --Whole utterances in conversational
--Referential, labeling functions      contexts
  at first                           --Rapid, conversational input
--Clear mother-ese
________________________________________________________________________

"Analytic": Peters                   "Gestalt": Peters
"Referential": Nelson                "Expressive": Nelson
"Word development": Dore             "Prosodic development": Dore
________________________________________________________________________

Peters suggest that gestalt users may "have to convert slowly and painfully to a more analytic approach to language" (p. 13), holding that "creative language" (analytic language) eventually predominates. This is most consistent with position 2: gestalt language, which involves the heavy use of routines and patterns, may be a temporary strategy for the performer to outperform his analytic competence, to solve certain communication problems that his creative language has not evolved far enough to handle. Yet, since automatic speech appears

89