From Roger Lass, The Shape of English, p. 121:
Looked at in cold blood, the modifications [of the segmental sounds of English words] are so radical that it’s hard to imagine a text containing a lot of them would be comprehensible; yet casual speech is characteristically modified in this way. The reason it can be processed without difficulty is that speakers “know the rules” for converting lento [slow, precise] representations into allego [fast/normal conversational speed] ones; they can ’re-synthesize’ them, given an allegro input. It is in fact typical for an untrained speaker hearing a modified allegro sequence to be quite convinced that he heard the lento form. And for a speaker pronouncing an allegro form to insist that he’s produced a lento form.”
What this indicates to me is that learners of L2 must internalize the full forms or strong forms of words before they can process the allegro or colloquial forms e.g. ’did you’ > ’didju’, ’can’t you’ > ’cantchu’, etc. Therefore, teachers might consider speaker lento, i.e. using lento forms like ’did you’ before using the typical sandhi or modified forms. Without that foundation, learners may find it difficult to untangle the massive sound modifications when words are strung together in rapid, allegro speech.