A reflective, speculative piece...
There is a general principle in teaching and learning that you tend to get better at what you practice. In language learning, if you do lots of listening, you become a better listener. If you read a lot, you become a better reader, and so on. By this token, it would seem logical to assume that if you practise speaking, you will become a better speaker. But is this the case?
If you accept the claims made by proponents of the comprehension hypothesis, first elaborated by Stephen Krashen back in around 1980 and still exerting considerable influence today, you would say that you do not get better at speaking by speaking, but by doing more listening and reading. This claim is based on the assumption that acquisition only occurs through receiving comprehensible input and, as far as acquisition is concerned, speaking just performs the role of getting more input from interlocutors. You could put it this way: you cannot become more proficient without getting new input.
Most applied linguists do not accept this claim even though they acknowledge that input is the number one factor in making progress with language learning. I was talking about this with my wife the other day, who is a linguistics graduate, speaks a number of languages and who is a bit sceptical of Krashen's work. She is convinced that the very process of speaking helps you improve. We speculated that, in the process of forming new spoken utterances, partly by engaging unconscious, tacit knowledge of grammar, partly by referring to conscious awareness of rules, you are continually building up connections in your brain which either reinforce underlying mental representations of the language, or forming new ones. What interesting chats we sometimes have!
So, let's say that you have reached a good intermediate or advanced level and are conversing, often struggling to produce fluent and accurate utterances. You are using previously rehearsed chunks of language, successfully applying rules of syntax, grappling with others which are not unconsciously mastered, referring to rules you may have been taught, monitoring what you say, aware that some is accurate, some is not, making the same fossilised errors you always make - all of these processes, as they move from brain to mouth, are establishing more and more connections and, most probably, enabling you to improve over time. In a sense, you are providing your own input, practising it and honing it, with more or less accuracy and fluency.
In this way, you could argue that practising speaking makes you a better speaker. That is certainly the impression that many language learners have and an assumption nearly all teachers make. Some applied linguists (but by no mean all) accept this view and place it within the framework of 'skill acquisition', where language learning is viewed as analogous to the learning of any complex skill - you learn the skill and practise it until it becomes automatised (i.e. part of your underlying, unsconscious knowledge). This is what most teachers assume to be the case and they may well be right. We cannot know for certain.
What do you think?