The
Agree. The best results comes with watching movies with subtitles. The source of video is most important, because it relates to "motivation" things. There is impossible to repeat right after actors (or other speakers) who you dislike, or even disrespect. Shadowing is a really hard exercise, because it includes coordination between Audio cortex, Motor Cortex, wide set of interpretation patterns, and finally it's related to Lobe parts of the brain, because each "story", or each "personality" has very deep affecting to our behaviour.
20 minutes video should become a 60 minutes training, and all difficulties, that students can face during this process just show that training is going effectively. Meanwhile, and this is the most important, I guess, the worst scenario for improving pronunciation is a decision to stop paying attention to specific sounds. People, who have been made this mistake wherever in the past, should pass through a long period to "retraining" ignoring patterns in the brain. It relates to the motor cortex first (how you move muscles in your mouth to pronounce words) , but as second, includes problems in audio cortex too (people just can't hear a difference between closest sounds).