First language attrition: The methodology revised

This article presents a criticism of the methodology most frequently used by language attrition studies. In particular, the preoccupation of such research with ?errors? in the data from attriters is questioned. It is proposed that approaches which focus on overt deviance cannot come to a full understanding of the attritional process, and that a full investigation of lexical, morphological and syntactic complexity and richness of the data produced by attriters is necessary in order to achieve a truly balanced view. The limitations to the insights gained on the basis of an error-based approach to language attrition, and the potential of an analysis that takes into account all aspects of proficiency, are illustrated on the basis of an investigation of autobiographical narratives from German Jews.