Test case: ICE_NG, query string [boy|girl], output columns Query string and Gender. With no substituted values for the Gender column, the G-text matrix works, but if for example m is substituted by male, it fails when attempting to retrieve the log-likelihoods.
If value substitution is used for corpus features, the subcorpus functions don't work anymore because the substituted value is used to determine for example the subcorpus size.
Possible solution: get_subcorpus_size() could be made sensitive to substitutions, and could reverse the substitutions.
Test case: ICE_NG, query string
[boy|girl], output columnsQuery stringandGender. With no substituted values for theGendercolumn, the G-text matrix works, but if for examplemis substituted bymale, it fails when attempting to retrieve the log-likelihoods.If value substitution is used for corpus features, the subcorpus functions don't work anymore because the substituted value is used to determine for example the subcorpus size.
Possible solution:
get_subcorpus_size()could be made sensitive to substitutions, and could reverse the substitutions.