ONE OF FOUR THESES WHICH FORM THE "SUPPLEMENT TO STUDIES IN LANGUAGE AND LANGUAGE BEHAVIOR, PROGRESS REPORT V," SEPTEMBER 1967, THIS STUDY WAS DESIGNED TO RESEARCH (A) WHAT SORT OF INFORMATION ABOUT GRAMMATICAL CONSTITUENTS IS NECESSARY IN A GRAMMAR, AND (B) HOW THIS INFORMATION CAN BE PRESENTED IN A GENERATIVE TAGMEMIC GRAMMAR. THE FIRST CHAPTER DESCRIBES THE NOTION OF "SUBJECT" IN ENGLISH ACCORDING TO SEVERAL PROMINENT LINGUISTS AND PROCEEDS TO DISCUSS FOUR TYPES OF GRAMMATICAL UNITS--(1) GRAMMATICAL FORM (SUBJECT, OBJECT), (2) GRAMMATICAL MEANING (AGENT, GOAL, INSTRUMENT), (3) LEXICAL FORM (NOUN, VERB), AND (4) LEXICAL MEANING (ANIMATE, MALE, HUMAN). CHAPTER TWO PRESENTS R. LONGACRE'S MODEL OF TAGMEMICS WHICH THE AUTHOR CLAIMS IS ADEQUATE FOR DISCERNING SURFACE STRUCTURE (GRAMMATICAL AND LEXICAL FORM), BUT INSUFFICIENT FOR DEEP STRUCTURE ANALYSIS (GRAMMATICAL AND LEXICAL MEANING). THE THIRD CHAPTER INVOLVES A DESCRIPTION OF ENGLISH SUBJECT TAGMEMES AS WELL AS A DETAILED DISCUSSION OF CLAUSE-LEVEL GENERATIVE RULES. THE FINAL CHAPTER CORRELATES THE FOUR TYPES OF GRAMMATICAL UNITS (TAGMEMES) WITH THE DESCRIPTIVE RELEVANCE OF GRAMMATICAL LEVELS (WORD, PHRASE, SENTENCE), AND THE SUFFICIENCY OF THE GRAMMAR IN PROVIDING INFORMATION FOR SEMANTIC INTERPRETATION (DEEP STRUCTURE) IS DISCUSSED. THIS STUDY WAS PUBLISHED AT THE CITY CENTER BUILDING, 220 EAST HURON STREET, ANN ARBOR, MICHIGAN 48108. (FB)