Casanave (2004) emphasizes the importance of defining improvement and good writing before discussing what works and what does not when teaching ESL writing. She divides improvement into two categories: improving quality of writing and developing L2 proficiency (p.66), which represent broadly the two constructs of L2 writing. Depending on which of these is a teacher's (and institution's) belief and goal, the teacher will see and pursue improvement in different ways identified first and foremost in assessment criteria. According to Casanave (2004), assessing quality of writing can be too ambiguous and non-satisfying for teachers, who tend to focus on criteria that are easier to count and correct (i.e., vocabulary and grammar). However, the question that I would ask is why is L2 proficiency necessarily related to grammar and vocabulary? Definition of language proficiency is problematic in itself (e.g., fluency vs accuracy debate), and choosing the countable ways of assessment does not always imply that the teacher aims for developing L2 proficiency in (through) students' writing.
Besides their own beliefs about good writing, English teachers should also take into account students' beliefs and institutional policy. Within our department, good writing was seen first and foremost in relation to accuracy in using language elements (grammatical and organizational features and vocabulary items) being targeted and studied at that particular time. Fluency was expected to increase with practice of writing and re-writing the essays. So, error correction and feedback played a very important role.
I would like to share two interesting cases I had. One was when I tutored a student (I will refer to her as Marina) who was accepted onto the major where students were expected to have studied English at school. However, Marina's foreign language at school was German and she had studied English only for two summer months with a tutor prior to the beginning of the school year. In other words, she had little acquaintance with English (e.g. extremely limited vocabulary), while certain level of knowledge of and about the English language was expected. For her, learning to write (as well as read and speak) implied translating and learning almost every word in the textbooks, understanding absolutely new grammar rules and sentence structure. Our tutoring sessions often involved my going through her draft, negotiating the meaning and discussing the word choices and grammar issues. Nevertheless, while the required vocabulary was often more difficult and sophisticated (e.g., reciprocate) for other students (who had studied English before), for Marina it was often the only vocabulary at her disposal. Therefore, she had no problem using that vocabulary and structures in her essays if she understood their function and meaning, while her peers sometimes resisted using the imposed forms. This meant that Marina ended up getting good grades according to the construct of good writing promoted by the department, even though her overall language proficiency, general fluency and probably quality of writing were not "good". Within the given definition, her writing improved dramatically over a semester. Her style of learning and department policy implied a lot of error correction on my part and re-writing and submitting a perfect product on hers.
Another case was the tutoring of Anna, who had studied English before, was confident in using English both in speaking and writing, but 99 % of her writing was impossible to understand. In spite of the fact that we shared L1 and her handwriting was intelligible, the sentences for some reason did not make much sense. Part of it could be attributed to spelling, but only part. Her choice of words, sentence structure and grammar was beyond comprehension, sometimes even for her. In order for her to be able to pass the English course, Anna had to be confined to a very rigid paragraph and sentence structure. In the end, she got a passing grade.
So, in the first case, the restrictions in form and content were taken up voluntarily, while in the second case, they were imposed. In the first case, the language learned served as a springboard for the student to strive for fluency, in the second – fluency was controlled by the rules which denied any freedom of expression. However, both cases showed improvement in students' performance, at least in accordance with the department's definition of improvement.