There are proper names that are normal words. So I was thinking of a rule that checks if a word could be a proper name, without the starting capital.
Example: The Dutch name Ernst (also German?) is a quite common name. But it is also ‘seriousness’.
Having ‘ernst Janssen’ in a Dutch text does not trigger a rule yet.
Is it possible to check the postags for a different form of a word?
Something like:
<token modification="startupper" postag="ENM:PER:FST" case sensitive="yes">ernst</token>