Robert wrote this a few weeks ago. I repost it here because it is so important, in case anyone missed it the first time:
The comprehensible output (CO) hypothesis states that we acquire language when we attempt to transmit a message but fail and have to try again. Eventually, we arrive at the correct form of our utterance, our conversational partner finally understands, and we acquire the new form we have produced.
http://www.sdkrashen.com/content/articles/comprehensible_output.pdf
The above comes from an article by Krashen that is available on his website. (See link)
Basically, what Swain is saying is this:
1. Output is necessary to acquire a language
2. Output shows us what we don’t know yet
3. By TRIAL AND ERROR [an extremely inefficient and often ineffective process], we eventually discover the “correct form” because our interlocutor finally understands us
4. This conscious mental process leads to unconscious acquisition
Of the four statements above, I agree with number 2 and only number 2. Attempts at output can show us what we don’t yet know. The other day I was having a conversation with a French teacher in French. She asked me what I was going to do on the weekend. I didn’t know how to say “have the standard maintenance on my car done at the repair shop”, so I a) circumlocuted a bit and b) asked her so that I didn’t waste our time. “Discovering the correct form” might lead to memorization, but it won’t lead to acquisition. It did, however, show me an area where more comprehensible input would be helpful. That is one value of output: getting more input – but Krashen and VanPatten have been saying this for years, and it is not “comprehensible output”.
Statement #1 remains to be proven. Krashen argues persuasively and has strong anecdotal evidence that comprehensible input / comprehended intake is the sole sufficient cause of acquisition. Think of it this way. If I maintain that a student must work math problems in order to understand math, but you show me even a single student who understands math without doing any math problems, you have nullified (or “falsified”) my hypothesis. While SLA is a bit more complex, Krashen provides examples of people who have apparently acquired a language without any output whatsoever. One is “Daniel” (I think is the name), who acquired Mandarin by listening to his mother read him stories in Mandarin but never spoke a word of it until suddenly thrust into a situation in which he needed to communicate in Mandarin – he had already acquired it and so was able to communicate with ease. When I lived in Germany, I lived for a year with a family in which the mother was from Zurich and spoke Swiss German dialect. I did not have any instruction in the dialect, did not hear it much each day (only when the mother was talking to her children), and certainly did not try to speak it in any way. Nonetheless, at the end of a year, I could understand the dialect when a native German couple from the Stuttgart area could not.
(See my story at https://compellinginput.wordpress.com/2016/05/14/toward-a-philosophy-of-teaching-3/)
Swain does not call this process Trial and Error, but that is what it is. I keep trying new things until my interlocutor understands me. There are at least two significant problems with this approach: 1) As I noted above it is inefficient and ineffective, and 2) unless the interlocutor is trained to accept only a correct utterance, it will lead to acquiring incorrect forms because a “normal” (i.e. untrained, non-linguistic, non-teacher) interlocutor will understand and accept incorrect forms. For example, my school employs janitors who are not native speakers. A couple of years ago, the janitor who cleaned my room wanted to be certain he did not disturb me with his cleaning. Whenever I walked out of my room after school, he would ask me, “You go home now?” I understood what he was asking and gave him a proper response (“Yes, I’m going home”, “No, I’m just going to the office for a moment”, etc.). According to Swain’s hypothesis, by being understood, this janitor figured his output was correct (it was comprehensible) and thus acquired the form. Unfortunately, if Swain is correct, he was acquiring an incorrect form. OOPS!
Point #4 assumes that conscious, propositional knowledge somehow becomes unconscious, procedural memory. Last time I checked, though, the human brain was not equipped with a transmogrifier capable of transforming propositional knowledge into procedural memory – they are two different systems.
Just my take on “Comprehensible Output”.