Fiction genre in both COCA and BNC best predicts (i.e. highest correlation with) learner collocation knowledge, Durrant (2014). This could be useful to select collocations to use with lower proficiency students. For higher proficiency students filter the previous list by mutual information (MI) score in COCA as learners are known to be weak with low (word) frequency items that MI scores pick out. Or use the KWIC function of COCA to easily see any idiomatic uses that higher proficiency learners may want to learn.
For example searching on collocations of all uses of get in the Fiction section in COCA shows highly frequent collocations include get out (of) + location and get up. These can be appropriate for low proficiency learners.
Filtering by MI score we have contraction of (have) got to i.e. gotta and get rid (of). Looking more closely in a KWIC search in COCA we see uses of the idiom get out of hand. These may be appropriate for higher proficiency students to learn.
Durrant, P. (2014). Corpus frequency and second language learners’ knowledge of collocations: A meta-analysis. International Journal of Corpus Linguistics, 19, 443–477.