In the general workflow, DBpedia Spotlight returns the types of the entities. These are stored in a column types which contains a list of lists for each entity. This column can be resolved to individual columns with the types_src argument in get_dbpedia_uris(). Currently, this argument is only available for character vectors.
For other input data such as quanteda corpora, columns of the default value of types_src are still created. For subcorpora based on CWB corpora, this is not the case.
In this case, the columns are not provided. This is an issue for the current implementation of entity_types_map() which relies on these columns.
On a related note: When implemented, the README should be updated to account for these developments.