In order to progress toward fluency, a language learner must continuously more closely approximate a native's pronunciation and understanding of the language. This claim is almost tautological. But you can easily see how this necessary aspect to language learning can be (and often does get) corrupted. Most people have experienced teachers stuck on simply fixing bad pronunciation or grammar; these teachers see their task as developing more and more grammar worksheets, and diagrams of the shape your mouth should make for different sounds. But even in these often absurd activities, there is a kernel of truth.

Once we can't distinguish between our pronunciation and understanding and a native's, or once we are satisfied the difference is not worth overcoming, we are blocked from progress. We have plateaued [link to plateau article].

As a teacher, there is a place for correcting a language learner's pronunciation, and highlighting differences between his and a native's understanding. But such correcting, if it makes the learner hesitant to speak unless he is positive it is spoken with a native accent and with perfect grammar, is detrimental to the learner's progress. If the correcting is reduced to, or perceived to be reduced to providing lists of correct sentences, it can give the impression that language learning is tantamount to memorizing a set of grammatically-correct sentences.  This flawed system is reliant on the teacher providing the sentences and it will start the learner down the wrong road.

One SmartNoise method that has a good instructional basis looks like this:

Before/After

Shopped/shopping Read(past tense)/reading
Prepared/preparing Saw/seeing
Cleaned/cleaning heard/hearing
Understood/understanding

I The for
article/book
Movie/photograph
Sound/music Closed/closing
Opened/opening
Cooked/cooking
Door/window
Meal/food
House/room

After I read the article, I understood.
Before I saw the movie, I read the book.
Before seeing the movie, I read the book.
*Before I saw the movie, I reading the book.
After I shopped, I prepared the food.
After shopping, I prepared the food.
*After I shopping, I prepared the food.
*After I shopped, I preparing the food.

First there is a "scatter-chart" with different words or groups of words. At the bottom, there are several sample-sentences (the incorrect ones have an asterisk) generated from the scatter-chart. The task given to students is to generate as many sentences as they can from this limited set of words, using the sample sentences as a guide. The sentences highlight frequent errors students make. The students figure out what the mistakes are, or they get a feel for why some expressions work and others don't.

Learners can gain a feel for an aspect of a native's understanding, and work to generate many novel expressions, all the while, a limited set of typical differences between his and a native's understanding are highlighted.

In a sense, every activity should work toward facilitating the learner approximating a native's understanding of or use of the language. Some time (and this time can be a tricky thing--teachers and programs often go way overboard) should be devoted to how the learner's language differs from the natives. This reflection can be part of an activity primarily focused on exposure to the language, or on opportunity to speak, or it could be an independent exercise; however, the more abstracted it is from actual, everyday use, the less helpful and practical it becomes.