7.4 Stereotyped compound given names. Suppose a person carries a stereotyped compound name like Jean-Baptiste. The person from history was John the Baptist, where Baptist makes little sense without the John. Would it be advisable to view this as two given names, one a token of <Jean> and the other a token of <Baptiste>? The difficulty of predicting the proper value specific comparison weight seems even more problematic for this case. There would be a low agreement weight for <Jean> and an unusually high agreement weight for <Baptiste> since any use by itself is very rare. Would the sum of these two weights be anything like what the weight for <Jean-Baptiste> ought to be? Probably much too high. Here the average of the two would probably be more accurate. This kind of analysis in German countries has the effect of seeing a name like <Maria Magdalena> given at baptism as corresponding to the name <Magdalena> at marriage. In this situation it may be possible to get relative frequencies to produce weights that will work.
In practice there is often little difference between separate given names compounded and stereotyped compounds. The Pierre of Jean-Pierre may be an distinctive modifier like the stereotype "Baptiste," and the modifier Baptiste may become a given name in its own right. The difference is thus only apparent from our intuitions about derivations probably reflected in the relative frequencies. Suppose X Y is more frequent than Y by itself. In this situation Y is probably the modifier in a stereotyped compound.