While ‘an’ is a common error for ‘aan’, there is also the valid ‘An’, which is a name.
Apparently, the postagger assigns the postag of ‘An’ to ‘an’, which is not correct.
It is okay if the postagger tags a alllower word with a tag for a firstupper word (because of the forst word of a sentence) , but not the other way around. Is there a way to achieve that?
Or maybe make postagging fully case sensitive (resulting in a larger dictionary)