Unlike some other cultural legacies, languages leave no trace in the archaeological record. There’s often no trace in the written record, either.
Only a small portion of the world’s estimated 7,000 languages are well-documented in places like dictionaries and grammar books. Those that are least well-documented are the most endangered. [Daniel W Hieber, “Renaissance on the bayou: the revival of a lost language”]
We have argued before (see here) for the importance of including proper names data in any study of historical linguistics, and that the usual omission of proper names from standard lexicons is problematic. The quote above highlights these points.
Names, as a part of language, leave very little trace in the archaeological record — there is only the occasional name painted on a pot or carved into a stone. Of the “small portion” of languages which are well-documented in dictionaries, only a portion of each individual language is documented by these dictionaries, given the tendency to wholly omit proper names. The names which are least documented are the ones most endangered: A hapax legomenon which occurs in only one manuscript is in much greater danger of being erased forever. Even a name which has made it out of manuscript and into printed edition, say, a 19th-century edition of a cartularium, are still at risk of being “lost” if no one knows that they are in there. As we’ve also argued before (see here) one reason that people mistakenly identify certain names as modern coinages is simply that they don’t know or don’t have access to the evidence, usually in a culturally or geographically remote context, that shows the name has a much longer history. Without detailed onomastic lexicons, an important part of our linguistic history is threatened with extinction.