Yes. It is that all approaches and methods and techniques are valid and good in moderation, and that variety is everything. What should you do in class today? Whatever you didn’t do yesterday. Students will learn and improve in their own way at their own pace – and what you teach is only partially related to what they learn. The teacher’s job is to provide a language-rich environment, at the right level, with plenty of activities for noticing language and plenty of opportunities for interesting and personalized speaking activities. After that what the students learn is up to them and their own internal cognitive processes.
I believe that learning a language is very similar to learning any other skill (eg playing a musical instrument, driving a car etc). For example, the fluency and freedom of expression of a Jimi Hendrix or a John Coltrane was achieved by hours of practice at an early stage of learning (perhaps mindlessly and routinely going up and down scales). There is no magic bullet to fluency.
I believe that students cannot give attention to meaning and form at the same time, and so (for example) they can read a text for content and then discuss it, or go through the text looking at lexis. But not both at the same time. The two activities should be clearly separated in class, and students should not get mixed messages about what the point of an activity is.
In the same way, I believe that post-task rather than during-task is the best place for language feedback: post-task students are attending to form, during-task to meaning. Requiring them to switch constantly between the two modes is unhelpful to both.
I believe that students have only a certain amount of attentional resources when they speak, and that these are divided between accuracy, fluency and complexity. I believe this is a zero sum game, and that giving more attention to one of these necessarily means giving less to another.
Finally, I believe that a major factor in language learning is memory (or rather some complex process involving neurones and synapses, and connections between short and long term memory, and all sorts of other biological processes in the brain). Some people have this internal system working well, others less well, and this is yet another factor beyond our control.
One thing that is, however, very much under the teacher’s control is the use of diagnostic language feedback following speaking activities. I call this technique ROLO (Reformulate Output Lightly but Often), and I believe it is a key element in effective language acquisition. There is a slideshow about ROLO (shorter) and an article (longer) posted on the site.