Dewaele, Jean-Marc2008-08-292008-08-292000International Review of Applied Linguistics, 38(1): 31-470019-042Xhttp://hdl.handle.net/10315/1446The utterance is a widely used linguistic unit. It seems, however, to escape every attempt to define it unequivocally. There seems to be a consensus in applied linguistics that a fuzzy combination of syntactic, semantic and prosodic clues are necessary to identify the boundaries of an utterance. Drawing upon our own research in advanced French interlanguage, we present an analysis of the measure "Mean Length of Utterance" (MLU), widely used in studies on first language acquisition and speech disorders and, to a lesser extent, in second language acquisition. MLU is shown to be methodologically unreliable for adult speech. We argue that other measures of utterance length, like the MLU3, are sounder and can help to gain a better understanding of synchronic variation in speech.frSecond language acquisitionMLU IndexFirst Language AcquisitionMean Length of UtteranceSecond Language Learning IndexArticle