Harley's conclusion that "grammar instruction can help to promote lasting improvement in second language proficiency" is based on her study of English speaking sixth graders an in early total immersion French program in Canada (Harley, 1989). Her subjects received eight weeks of instruction focused only on two forms: the passé composé and imparfait, totaling about 12 hours of direct instruction. The final tests used clearly focused the students on the form they had been studying and was very similar to some of the activities they had done in class. There was no difference at all between instructed students and comparisons on a composition that forced students to use the targeted forms. Instructed students did a little better than comparisons on an oral test and on a cloze (fill-in-the-blank) test, but there was no difference between the groups on similar tests given three months later. Harley's study only shows that grammar teaching leads to modest short-term gains that disappear in a short time.

Haver cites Day and Shapon's conclusion that systematic instruction in grammar is helpful, especially with verb tenses. A closer look shows that this claim is based on weak evidence. In Day and Shapson (1991,1996), seventh graders in French immersion in Canadia studied the French conditional for six weeks (17 hours total). They did somewhat better than comparisons on a grammar test and on a composition given 11 weeks after the instruction ended, but not on an oral interview. Even on the grammar test, however, gains were hardly spectacular after so much instruction. Instructed students improved from 19 percent correct on the pretest to 41 percent correct on both the posttest and delayed posttest. The grammar test had 27 items, which means that students improved from five correct to eleven. After 17 hours of instruction, this is disappointing. It would mean about a C+ in the average classroom.

The third source Haver cites is Lightbown, Halter, White and Horst (2002), who compared the impact of a comprehension-based method with traditional methodology in ESL for French-speaking children in New Brunswick, Canada. According to Haver, "the only difference between the groups was that one group was taught with little or no focus on form, whereas the other group received form-focused instruction and feedback on error correction regularly" (xiv). The differences were greater than this. The core activity in the comprehension-based classes was independent listening to English books on tape while following along in the text, for 30 minutes per day, starting in grade 3. In grades 5 and 6, students also watched video programs and followed the transcripts, and did some workbook activities. The comparison groups were described as audio-lingual.

Before discussing the results, it is important to note that it was not clear that the "comprehension groups" received high-quality comprehensible input. There was individual variation in involvement in the listening/reading activity. Lightbown et. al. note that "both teachers and program evaluators observed that students in the experimental program varied greatly in their approach to the listening-reading activity. Some students followed the words with their fingers, while others allowed their gaze (and their mind?) to wander around the room. Students were encouraged to read each book at least twice, and some happily read the same book more than twice, while others chafed at this restriction and insisted on reading new books as quickly as they could..." (p. 454). In my view, the requirement that students simultaneously read the text while reading was quite unnatural. Lightbown et. al. insist, however, that the children really did get absorbed in stories (p. 457).