THIS STUDY COMPARED THE EFFECTIVENESS OF TWO METHODS OF ACQUISITION OF GRAMMAR BY CHILDREN. SPEECH DEVELOPMENT OF CHILDREN EXPOSED TO "EXPANSION" (REPEATING NEAREST COMPLETE SENTENCE APPROPRIATE TO CHILD'S PARTIAL STATEMENT) AND "MODELING" (EXPOSURE TO WELL FORMED SENTENCES) WERE COMPARED TO SPEECH DEVELOPMENT OF CHILDREN WHO RECEIVED NEITHER TREATMENT. THE STUDY HYPOTHESIZED THAT EXPANSION WOULD BE THE MOST EFFECTIVE. TWELVE CHILDREN, BETWEEN THE AGES OF 28 AND 38 MONTHS, WERE DIVIDED INTO THREE GROUPS ON THE BASIS OF (1) CHRONOLOGICAL AGE, (2) TALKATIVENESS, AND (3) INITIAL LEVEL OF LANGUAGE DEVELOPMENT. TAPE RECORDINGS WERE MADE AT THE BEGINNING, MIDDLE, AND END OF A 3-MONTH PERIOD. SIX MEASURES OF LANGUAGE DEVELOPMENT SERVED AS DEPENDENT VARIABLES. FIVE VARIABLES MEASURED ASPECTS OF SPONTANEOUS SPEECH AND ONE MEASURED THE ABILITY TO REPEAT SENTENCES. TWO TYPES OF STATISTICAL ANALYSES WERE USED AND THE VARIABLES WERE CONSIDERED BOTH TOGETHER AND SEPARATELY. TWO MAJOR RESULTS EMERGED--(1) ADJUSTED MEANS WERE HIGHER FOR THE INITIALLY LESS ADVANCED SUBJECTS ON ALL SIX MEASURES, AND (2) MODELING, NOT EXPANDING, WAS THE MORE EFFECTIVE TREATMENT. SUGGESTIONS FOR FURTHER RESEARCH INCLUDED (1) THE TESTING OF A LARGER SAMPLE OF SUBJECTS, AND (2) ALTERNATE HYPOTHESES WHICH TAKE INTO ACCOUNT THE EFFECTIVENESS OF INCREASED LANGUAGE EXPOSURE OVER EXPANSION. (PM)