A non-structural model is proposed for quantifying and analyzing the dynamics of language attrition, particularly among immigrants in a second language environment, based on examination of disfluencies (hesitations, errors, and repairs). The first chapter discusses limitations of the conventional synchronic textual approach to analyzing language loss, defines first-language competence, looks at the reliability of hesitation phenomena, and outlines details of the proposed model. Methodology of a study is described in the second chapter. Several chapters relate findings related to silent pauses, filled pauses, and rate of speech and verboseness, correlating these with length of stay and educational level and socioeconomic background variables. A chapter examines the relationship between first-language loss and speech guardedness. Subsequent chapters discuss the correlation of hesitation frequency with speech error rates, a time analysis of changes in speech error/repair rate, correlations between social/psychological variables and first-language loss, and implications for language loss theory and other applications of the model. (MSE)