It is asserted that the optimum method of describing the stages of an interlanguage (IL) and the development of learner competency toward the target language (TL) is one of self-analysis and subsequent reanalysis. A nine-page letter written by a native speaker of German in 1987 is compared with her correction of the letter in 1991. Analyses indicate that, while some stabilized idiosyncrasies originate in interlingual identifications, and some corrections are toward target language norms, others do not correspond to any forms outside the IL. The latter seem to be improvements on the TL and are examined in detail, especially restrictive relative clause structures. That none of the idiosyncrasies greatly affect the perceived interpretability of the IL by native speakers would inhibit their identification as errors by the subject, thereby encouraging stabilization. These findings imply that IL development involves some degree of internal analysis independent of external linguistic stimuli, lending credence to the treatment of interlanguages as independent of both the native and the TL. There also seems to be a threshold that must be reached before idiosyncrasies are perceived as errors to be reanalyzed and corrected. (Author/JP)