(Or it's real, but it ain't exactly there).
Something is bothering me. I've been working on a paper dealing with historical linguistics and Whitehead's process philosophy. It's something I've been working on for some time: I had a go at it in my Ph.D. thesis but looking back, the idea seems quite half-baked and impressionistic as I presented it there. Things are a lot more clearer right now.
The attraction to the idea is that it might enable one to hold both a Platonistic view on linguistic structure with a thoroughly diachronic, historical view on language-as-such. The basic dichotomy Whitehead draws is between universals or "eternal objects" which are, in and of themselves, pure potentials, and events or "actual occasions", in which universals may actualize. Space and time is not given: an actual occasion or event gains its spatiotemporal location during its "becoming", during which it essentially relates to its whole relevant universe (past events, universals inferred from those, etc.).
Whitehead has been applied to language by, to my knowledge, only Michael Fortescue (2001, "Patterns and Process"). He basically regards language as a complex "eternal object" and defines the linguistic equivalent of an event with a "moment of linguistic cognition". I happily follow him here. Fortescue's book is a bit smorgasbord-like, though: virtually all of linguistics from discourse studies to language change and acquisition is dealt with, and historical linguistics relatively briefly.
Now, the problem.
I think there are compelling reasons to hold some kind of linguistic Platonism, either a Whiteheadian version, or a non-Whiteheadian one as has been developed by Jerrold Katz. By doing so, we avoid hypostasizing grammar, i.e. regarding the work of theoretical linguistics as relating to a "language organ" or to linguistic cognition, etc. Autonomous linguistics would be autonomous linguistics; neurolinguistics would be neurolinguistics. We would avoid subordinating linguistics to some kind of "first order" discipline. Also, I think there are very good reasons to hold that language is a normative system analogous to mathematics, logic, etc.: it's about "correct" and "incorrect" rather than "is" or "not-is".
There's a bit of a snag, though, when we try to account for (historical-)linguistic labels such as "Uralic", "Indo-European" or "Finnish" on such a conception of language as a Platonic object. Because "Finnish" refers to a determinate spatiotemporal location. OK, so we regard the phonology, morphology, etc. of Finnish as a Platonic object which exists eternally (as a potential or in some kind of Platonic realm) but state that it becomes Finnish only due to its actualization at a certain time and place in the actual world of events. Problem with that it of course that "Finnish" as a historical, linguistic entity has a lot of "temporal smear", has undergone quite a bit of change during its history, and the issue of genetic identity, i.e. on what grounds do we issue a label such as "Finnish" or "Uralic" on a spatiotemporally extended community of "linguistic events" which does not embody exactly the same structures through space and time?
An obvious solution would be to use Popper's World 3 - but I dislike it. It's too obvious, and I can't help thinking that regarding historically specific things as for instance Mozart's requiem (and by extension, a given language) as World 3 objects is a bit confused. (Rather, Mozart's requiem would be a Platonic object which happened to be discovered and actualized by Mozart, as well as in subsequent performances, etc.). Also, I believe it merely displaces the problem of genetic identity and genetic relationships.
I think Whitehead referred to this problem when he wrote that:
"(...) the complex nexus of ancient imperial Rome to European history is not wholly expressible in universals. It is not merely the contrast of a sort of city, imperial, Roman, ancient, with a sort of history of a sort of continent (...). The nexus in question does involve such a complex contrast of universals. But it involves more. For it is the nexus of that Rome with that Europe." (Process and Reality, p. 324).
A "nexus" in Whiteheadian is a spatiotemporally extended community of events, which exhibit the same universals through their relationships with each other. It's also part of my currently favoured solution to the problem mentioned above. Linguistic structures would exist Platonically but without any reference to their "togetherness", not to speak of labels such as "Finnish": this togetherness emerges in actual usage, and in the presence of linguistic events "in" each other (my comprehension of an utterance by another person would be a past linguistic event immanent in a present one: and the diachronic dimension created in such a manner, with the actualization of the same linguistic structures in both events, would account for genetic relationships).
However, it seems to me intuitively difficult to hold a Platonic view on linguistic structure with an emergentist view on language as a social, historical system of rules, which emerges from linguistic usage, with language learners relating to their language as they would to rules, etc. I think the intuitive difficulty is only apparent, though, and resulting from somehow unconsciously holding Platonic objects to be more concrete than they really are. There seems to be no real contradiction between linguistic structure being a Platonic object; and it's actualization in space and time to emerge from language usage.
But this leads to something else, and quite unrelated to the specifically linguistic issue above: is the relationship between Platonic objects and their actualization in space and time something to be explained? Does it need an explanation?
Individual languages of course have a great deal of contingency due to the arbitrariness of the relationship between meaning and phonological form: it is only within a linguistic system in which we "cannot doubt" that a cat is called a cat and in which calling it tac would be wrong (in others it might be kissa, katt, mácska, etc.). But the impossibility of even conceiving 2 and 2 to be 5 is I think, an important ground for holding mathematics to be Platonic.
The thing is that mathematics, logic, reason all appear to be working very well in making sense of the world. Peirce seemed to feel this was worth mentioning a number of times, at one point regarding the validity of abductive reasoning as conditioned upon the "hope" that "there may be some natural tendency toward an agreement between the ideas which suggest themselves to the human mind and those which are concerned in the laws of nature" (CP 1:80), at another point stating that "(...) every scientific explanation of a natural phenomenon is a hypothesis that there is something in nature to which the human reason is analogous; and that it really is so all the successes of science in its applications to human convenience are witnesses. They proclaim that truth over the length and breadth of the modern world. In the light of the successes of science to my mind there is a degree of baseness in denying our birthright as children of God and in shamefacedly slinking away from anthropomorphic conceptions of the universe." (CP 1:316).
The issue is, obviously, religiously relevant. Nagel, in "The Last World", mentions that while he remained a metaphysical materialist, the irreducibility of reason does make us be more at home in the universe than secularly comfortable. But it should be handled carefully: the irreducability of Platonic objects (to which we respond normatively) to the empirical, physical realm says nothing about the impossibility or possibility of, for example, human cognition and all that evolving naturalistically. It would seem to me to be eminently possible for the normative (such as language) to emerge in a community of people predisposed to look for it - though I think mind itself (with intentionality, valuation, etc.) would be presupposed.