THIS STUDY REPORTS AN EXPERIMENT ON THE HYPOTHESIS THAT GRAMMAR-RELATED SENTENCE-COMBINING PRACTICE WILL MEANINGFULLY ENHANCE THE NORMAL GROWTH OF SYNTACTIC FLUENCY. TRADITIONAL STUDIES ON ERROR THERAPY AND SENTENCE STRUCTURE ARE REVIEWED, AND DESIGN AND RATIONALE ARE INFERRED FROM THE RESEARCH OF BATEMAN AND ZIDONIS. RULE LEARNING, PATTERN PRACTICE, MODELED WRITING, AND TRADITIONAL PARSING--ALL METHODS FOR ENHANCING GROWTH OF SYNTACTIC FLUENCY--ARE EXAMINED AND REJECTED. INSTEAD, STUDENTS ARE PRESENTED WITH SETS OF KERNEL-LIKE STATEMENTS WHICH ARE COLLAPSED INTO COMPLEX SENTENCES SO THAT THE STUDENTS EXPERIENCE PSEUDO-PRODUCTION INTENSIVELY. PRE-POST WRITING TESTS WERE GIVEN TO 247 SEVENTH-GRADE STUDENTS ASSIGNED TO THREE GROUPS--TRANSFORMATIONAL SENTENCE-COMBINING, CONVENTIONAL PARSING, AND NO GRAMMAR. THE WRITING WAS SEGMENTED INTO T-UNITS AND ANALYZED ACCORDING TO 12 FACTORS OF SYNTACTIC FLUENCY PERTAINING TO NOMINAL AND RELATIVE EMBEDDINGS, FREQUENCY AND DEPTH OF EMBEDDINGS, AND CLUSTERED MODIFICATION. RESULTS INDICATE THAT THE EXPERIMENTAL SUBJECTS EXPERIENCED MORE THAN ONE ADDITIONAL YEAR EXPECTED OF THEM, AND THAT THEIR PROGRESS WAS CONSIDERABLY GREATER THAN THE CONTROL AND PLACEBO GROUPS. TENTATIVE CURRICULAR IMPLICATIONS AND REQUIREMENTS FOR SUBSEQUENT LONGER-TERM RESEARCH ARE DISCUSSED. (AUTHOR)