Gold's Theorem and Natural Language Learning

[LauLuna]

I've recently known of Gold's theorem (1967).

It states that any class C of languages containing an enumerable
number of them L1, L2, L3,... and still a language L-omega, such that

1.for any n, Ln is strictly included in Ln+1 and in L-omega;
2. whatever is in L-omega is in some Ln;

is unlearnable, i.e there is no learner able to learn each L-alpha in
C given any environment of sentences of L-alpha.

But I don't see exactly what bearing the theorem should have on
natural language learning. Is there any result to the effect that
natural language can be conceived of as an unlearnable class of
languages?

Thanks