THE PURPOSE OF THIS RESEARCH WAS TO EXPLORE SOME OF THE VARIABLES THAT INFLUENCE INTRAINDIVIDUAL PHONETIC VARIATION IN CERTAIN ASPECTS OF AMERICAN ENGLISH SPEECH. FORTY COLLEGE STUDENTS PARTICIPATED IN THE EXPERIMENT. EACH PERFORMED TWO TASKS--(1) READING ALOUD WORDS (16 IN EACH OF FIVE CATEGORIES) FROM FLASHCARDS AS PART OF AN OSTENSIBLE LEARNING TASK AND (2) READING THROUGH THE SAME LIST OF WORDS WITH INSTRUCTIONS TO PRONOUNCE THEM "CLEARLY AND ACCURATELY." TWO OF THE CATEGORIES INVOLVED SUBPHONEMIC OR ALLOPHONIC DISTINCTIONS AND THREE INVOLVED THE RETENTION OR LOSS OF ONE PHONEME. EACH CATEGORY HAD TWO STYLISTIC VARIANTS--A FORMAL ALTERNANT (FA) AND A CASUAL ALTERNANT (CA). A SIGNIFICANTLY HIGHER USE OF FA PRONUNCIATIONS WAS FOUND IN THE SECOND CONDITION (CLEAR PRONUNCIATION) THAN IN THE FIRST FOR ALL FIVE CATEGORIES. THE CATEGORIES INVOLVING SUBPHONEMIC DISTINCTIONS ELICITED MUCH LOWER FA RESPONSES THAN THE CATEGORIES INVOLVING THE LOSS OF A PHONEME, AND THEIR VARIANCE INCREASED FROM CONDITION 1 TO CONDITION 2. THIS WAS ATTRIBUTED TO THE EMIC/ETIC DIFFERENCE. FINDINGS ALSO INDICATED THAT WORD FREQUENCY AND STATUS ARE RELEVANT VARIABLES TO THE CHOICE BETWEEN FA AND CA, AS IS A PERSON'S SPEED OF ARTICULATION. (DO)