The acceptability of a sentence is dependent on context: some sentences look awkward in isolation but improve in an appropriate context, whereas other sentences look all right in isolation but fail to fit certain types of context. Of particular interest is the degree and specificity of textual fit of different thematic (theme-rheme, topic-comment) variants of a sentence. A hypothesis worth further testing is that thematic variants can be ranked on a scale of markedness when in isolation (for instance, sentences with topicalizations being more marked than untopicalized sentences), and that the more strongly marked a sentence structure is, the stronger must be the textual forces motivating its use in context. Considerations of this kind are relevant to error evaluation. In advanced teaching of composition and of foreign languages, judgment should not be passed on sentences mechanically and in isolation. The ways in which they fit their context and contribute to the information dynamics of their text should be considered. (Author/CLK)