ABSTRACT
The “grain” of coding is determined by the goals of the investigation. For example, suppose one were interested in studying the acquisition of casemarking in Russian and Turkish, and that one were concerned only with the semantics of case categories. For such an analysis, it would be irrelevant to code for phonological form or for language-specific features (e.g., gender and number, which figure in Russian, but not in Turkish). Thus, for such a project, the relevant form category would be simply the case category (e.g., accusative). Depending on the goals of the study, formal coding of other parts of the utterance might also be necessary (e.g., person, tense, aspect, modality). However, note that overcoding is time-consuming and expensive and should be carefully avoided. The level of coding can always be made more fine-grained on a second pass, if additional features turn out to be important.
