THE AUTHORS HAVE CONCERNED THEMSELVES IN THIS STUDY WITH IMITATION AS A DEVICE BY WHICH THE INVESTIGATOR CAN LEARN ABOUT CHILD LANGUAGE. THE DATA EXAMINED ARE PART OF A LONGITUDINAL THREE-MONTH STUDY OF LINGUISTIC DEVELOPMENT IN A TWO-YEAR-OLD CHILD. THE REPORT IS BASED ON 1,000 ELICITED IMITATIONS. CERTAIN PHENOMONA WERE OBSERVED GENERALLY WHEN THE MODEL SENTENCES WERE SOMEWHAT BEYOND THE CHILD'S NORMAL SENTENCE PROCESSING SPAN AND WERE NOT ANOMALOUS. THESE SEEMED TO DEMONSTRATE THAT SENTENCE RECOGNITION AND IMITATION ARE FILTERED THROUGH THE INDIVIDUAL'S PRODUCTIVE LINGUISTIC SYSTEM. WHILE THE CHILD COULD UTTER SENTENCES SPONTANEOUSLY WHICH SHE COULD NOT IMITATE, SHE COULD ALSO GIVE "RECODED" IMITATIONS OF MODEL SENTENCES WHICH EXCEEDED HER PRODUCTIVE CAPACITIES. SOME INTERESTING SIDE-PHENOMONA WERE OBSERVED--(1) OMITTED ITEMS, SUCH AS ARTICLES, COPULA, EMBEDDED CLAUSES, MAY SIMPLY NOT HAVE BEEN HEARD, (2) HESITATION PAUSES WERE IMPORTANT CUES, (3) PRESERVATION OF STRESS SEEMED TO BE GENERAL THOUGH ITS POSITION WAS NOT ALWAYS PREDICTABLE, AND (4) THE PRESERVATION OF RHYTHMIC AND INTONATIONAL ASPECTS IN IMITATION MAY BE BASIC. AN IMPORTANT EXCEPTION TO THE LAST OBSERVATION WAS THAT THE CHILD IGNORED REPEATED WORDS UNLESS THE REPEATED WORD COULD BE INTERPRETED AS AN APPROPRIATE LEXICAL ITEM IN THE SENTENCE. THIS PRELIMINARY ANALYSIS INCLUDES A REFERENCE LIST OF STUDIES IN CHILD LANGUAGE. (AM)