T-Unit Analysis in Second Language Research: Applications, Problems and Limitations

T-unit analysis has been widely applied to measure the overall syntactic complexity of speech and writing samples. It has been used both in crosssectional descriptive studies and in experimental studies to measure the effect of sentence-combining as a curricular activity designed to enhance normal developmental trends in syntactic maturity. The claim that mean T-unit length is a valid measure of overall syntactic complexity is well supported. In recent years, T-unit analysis has been applied in studies of second language learning to characterize the syntactic nature of linguistic input and to assess the syntactic maturity of learners' written production. In all studies, T-unit analysis was seen to provide an objective and reliable method of determining the overall syntactic complexity of language samples. Nevertheless, the validity of the T-unit as a measure and the significance of overall syntactic complexity as a variable have been questioned. The present paper is a critical review of arguments for and against T-unit analysis in second language research; suggestions are made about the validity, usefulness, and limitations of this and other measures of overall syntactic complexity.