A very good question! Unfortunately I do not know any paper on this topic. Maybe one reason could be, that the correctness of proper names can not be proved objectively. The examiner can not verify whether certain proper names actually exist or are possibly just thought up. Verbal fluency tests are intendet to assess "divergent problem solving", but not creativity or inventiveness.
If this is the reason, it should be considered as our test problem (which we should find a solution for it), not our subject’s problem or a kind of insufficiency for their verbal fluency ability, because we want to exclude all of the proper nouns (or in other words a group of common words). I don’t know but it may also fade the superiority of some subjects. I think at least for children this method can be ignored.
I kept thinking about your question. In the German RWT (Regensburger Wordfluency Test) there are phonematic tasks (internationally also known as the FAS-test), and also semantic subtasks, which include, among others, the following task/category: first names. the subjects should name as many first names as possible. Thus, at least the RWT is able to asses (semantic) verbal fluency using proper names.
Dear Majid, I don't know the original reason, but since the norms are all based on this rule, to do otherwise would deprive us of past norms and accurate research comparisons.
You are right, almost all of previous articles on verbal fluency have emphasized on this phrase: "excluding proper names" in their procedure part. But, I think at least for doing research specially when subjects are children or for those subjects which are not from different cultures we can ignore this rule. Although, still the problem you raised will remain.
Dear Majid, It would be an interesting piece of quick research to determine whether the norms vary at all if we eliminate that requirement (no proper nouns). My suspicion is that there will be no significant difference in norms. It would be nice to see it verified though !