In a discussion with many teachers, one of which was me, Betty Azar said:
"What we DO mean when we say that 'grammar teaching works' is that students develop their interlanguages faster and with better results when a grammar component is included in a balanced program of second language instruction. This is clear not only to experienced teachers, but is clear in the cumulative research into grammar teaching during the past 20 years."
My knowledge of the research on this subject is not complete. From what I understand, much of it actually shows that "grammar teaching" will result in gains in "grammar testing" and only modest gains at that. This doesn't reflect acquisition. Could you share some references to any research where acquisition has been demonstrated through direct grammar teaching?
Consider this question:
How is it possible that students cannot acquire grammar solely through grammar teaching but students can acquire grammar solely through extensive reading and exposure to the language?
This seems to indicate that grammar teaching can only play the most minor role, if any, in language acquisition. Stephen Krashen recommends grammar teaching to only deal with anything the student has learned incorrectly, what I would call a sort of post-acquisition experience fine-tuning.