When the last speaker of a language dies, the language is said to be extinct (maybe that should be ‘fluent native speaker’, but I’m going with the more general condition). Can a language be resurrected? That is, can a language come to life again if people start speaking it again with sufficient fluency? I think an answer to this question turns on a number of philosophical and political/pragmatic concerns. From the philosophical side, it depends on what sort of thing a language is, and what persistence or identity conditions things of that type have.
For example, if languages are Platonic abstract entities, they are by nature timeless and cannot go extinct at all. ‘Going extinct’ for an abstract object, just means, ‘has no current exemplars’. Having no current exemplars or instances doesn’t affect the existence or identity of Platonic objects. The abstract existence of the perfect circle, as a mathematical object, is not affected in the least by the fact that no concrete physical objects are perfectly circular. So the existence of a language is not affected by whether it has any speakers or not. Of course, it is also unaffected by whether the language ever had any speakers.
If a language is a complex mental construct, as some Chomskians think, then languages cease to exist when their speakers cease to exist, and there can be no question of languages being resurrected. In the same way that a memory is no more (pace JK Rowling’s penseive) once the mind that held that memory disappears, the particular and individually calibrated mental language faculty of an individual speaker passes away with that speaker.
Only if language is (something like) a body of speech, or of written and spoken utterances, can a language go extinct and then be resurrected. Actual (and potential utterances) are datable particulars, that is particular events or happenings that can be assigned a definite (actual or potential) date and time. Since all actual and potential utterances have a date (and an author or person who spoke them or might have spoken them), a language could go extinct when there is a long temporal discontinuity among the utterances dates. Suppose a language goes extinct in the 18th Century (as Cornish did), and then people attempt to revive it now. All the utterances of the language have a date and the period of extinction is just a long gap in the sequence of dates. A language that goes extinct and stays extinct is a language with a last dated utterance.
Don’t be distracted by the inclusion of potential utterances. I mean only to allow for utterances, for example, that I might have made but didn’t, or would be able to make but won’t. These ought to be just as much part of a language as the utterances people actually make. The inclusion of potential utterances in the body of a language permits the nominalist to say a language never dies, just as the Platonist can say this, because even after the last speaker has died, there is an attenuated possibility that it could go on being spoken (if the last speaker had lived longer, or hadn’t been the last speaker then there would have been other utterances, so these utterances are potential utterances). But if the definition of the language corpus allows only actual and nearly actual or closely possible utterances, then languages will be able to go extinct after all.
This ‘corpus picture’ presupposes a criterion of language identity for utterances in that corpus, since only if they are ‘alike’ or of the right kind will the utterances compose a language. The point is essentially the same as that made against John Locke’s account of personal identity in terms of continuity of memory. In his Essay Concerning Human Understanding, you will remember, he took continuity of a person’s mental life as a criterion for when two individuals considered at different times are really one and the same individual seen over time. But as was pointed out, continuity of the same memory and mental life presupposes personal identity, since if the memories belong to different persons, there isn’t personal identity. Same mental life, means ‘mental life of the same person’. In a like way utterances can only compose a language if they already belong to it. Or to put it another way, not just any old collection of utterances will do. The cacophony of utterances heard during a coffee break at the United Nations do not compose a language, for example.
The way around this observation is to locate the criterion of identity in something else. The criterion that suggests itself is the mutual intelligibility of utterances to speakers. This move draws in political and pragmatic considerations. Obviously appealing to the effect of utterances on speakers presupposes identity criteria for speakers and groups of speakers. If we have random groups of speakers, there’s not much chance that anything resembling a language as we know it will emerge. But speakers helpfully group themselves politically as Welsh or Cornish or … and so on. If we take the identities of the groups as politically determined, then utterances belong to the same language when they mutually intelligible to speakers of the same group. Or better, two token utterances belong to the same language when they belong to a sequence of utterances that preserves a continuity of mutual intelligibility over time among members of a group.
On this version of the ‘corpus picture’ then, Cornish counts as going extinct and then being resurrected because there is a group of people who identify as a distinctive Cornish group, and there is mutual intelligibility among the members of the group. They all understand the Cornish utterances. In particular, I take it as a necessary condition on the success of the resurrection project that if an 18th Century native speaker of Cornish were somehow transported to the present day, they could converse without difficulty with people who claim to speak Cornish now. The per force hypothetical intelligibility to one of the earlier speakers of Cornish is what allows us to say that it is the same Cornish the modern group is speaking.
Perhaps certain groups can understand each others’ utterances but distinguish themselves politically from each other all the same (e.g. Norwegians and Swedes). In that case, the criteria will say that there are two languages even though the criteria would have said there was only one language if the groups hadn’t distinguished themselves from each other.
Anyway, that is one way of making sense of the idea of a language dying and being resurrected from a Philosophical perspective.