It allows this because negation is what allows language to refer to itself (it introduces recursion into language) insofar as to say 'not-x' is to refer to one's use of language, rather than some positively existing entity. — StreetlightX
I agree in general, but I think it is more technically precise to talk of dichotomies or symmetry breaking rather than negation alone, and of hierarchies rather than simply recursion.
The point is that a logically crisp idea like "never" has to arise not just as a negation, but as a dichotomous division. It's logical counterpart - always - has to arise in mutually grounding fashion.
So we can imagine a basic distinction in language - frequently~rarely. And then through inductive generalisation or abstraction, this becomes the absolute distinction of always~never.
Now this shift to proper abstraction is hierarchical rather than simply recursive as recursion is already happening at the level of the dichotomy. Even the notions of frequently and rarely are negating each other in self-referential fashion. And always and never are a stronger version of this self-referential act of mutual definition.
But the strong version of the dichotomy brings in the further thing of the concrete representation of the global symmetry they break. Always and never appeal now to the backdrop notion that is eternal time.
Frequently and rarely speak about the occurrence of events - the foreground action. Always and never make it clear that they are the absolute poles marking the extremes of some idea still larger than themselves. They point hierarchically to the third thing which is "time", the global symmetry from which they could spring as a dichotomy.
So a hierarchy is about the memory, the backdrop higher level idea, that can fix a local distinction in a definite fashion. It stabilises a negation.
No doubt this might seem a pedantic analysis, but it makes an important shift from a dyadic to a triadic logic of sign relations. Negation and recursion frame matters in terms of this against that - A vs not-A, and the repetition of a distinction. Complex reference might emerge as a result, but it is essentially unaccounted for. It simply is treated as emergent in an open-ended fashion.
But a triadic sign relation closes the story. Unlike negation, the dichotomy has recursion built in as each half of the dichotomy refers to its "other". And unlike recursion, the hierarchy explains how dynamical uncertainty (where is recursion going to lead?) gains generalised stability. The third player in the triad - the global symmetry that the local symmetry-breaking claims to break - is itself now named. Always and never get their meanings fixed in terms of the further notion of time, a temporal dimension.
Triadic sign relations also introduce vagueness and asymmetry in natural fashion.
Dyadic logic always demands counterfactual crispness. The middle gets excluded. It only wants to speak of either/or. But triadic logic creates room for middles, both as points of departure and places of arrival. Middles are what get developed by dichotomisation or symmetry breakings. You start off with a vague potential and break it into a definite spectrum of states bounded by two complementary poles of being.
So frequently~rarely is a little vague as a dichotomy as it simply states that some thing is either more or less. And then always~never takes that nascent relation to its crisp or absolute limit - a polar pairing that then admits of every intervening shade of "occasionally".
Likewise, triadic logic is large enough - it has enough dimensionality - to speak directly about asymmetry.
Symmetry breaking comes in degrees of hierarchically-fixed definiteness. The simplest and most unstable symmetry breaking - because it is single-scale and easily reversible - is a negation. It is like positive and negative charge. You can produce both for free from the splitting of "nothing", but then they are so weakly separated (so eager to get back together) that they annihilate back to nothing in the next blink of an eye.
So to fix symmetry-breakings, a separation must be achieved across hierarchical scale - an asymmetry must be formed. And this is what a hierarchy does. It disconnects the global from the local, the global becoming a state of "memory" for the system, its long-term prevailing constraints, while the local becomes its individual degrees of freedom.
In the example of always~never, time is the general idea that stands orthogonally to the notion of "the event". The event itself can freely either be or not be on the local view. It seems a perfectly reversible state of affairs - a fluctuation - at that level. But step back into the background notion of time and now the event can be seen as either always or never (or occasional, periodic, intermittent, unpredictable, etc).
When it comes to language evolution, the triadic point of view allows for negation and recursion always to exist vaguely in any language use. It is there in weak form even indexically. I could shake my head to signal negation, in the way any infant would twist away from food it didn't like. As a metaphoric sign, it could gain meaning quite naturally, building on already dichotomised and hierarchically integrated reactions of approach and avoidance that we all share as part of the same biological inheritance.
But language proper is a machinery for a social memory. Habits of abstraction can become fixed in a way disconnected from the individual and held collectively as named ideas. So as you say, that makes all the difference in the world. There is the open-ended meta-possibility of infinite levels of recursion or self-reference.
So the idea that basic language - some kind of proto-speech - must have preceded more advanced language is problematic. The essential trick - the division of communicative intent into words and rules - must have been there from the start.
Again, the standard approach to language evolution relies on dyadic logic. So either the habit of naming, or the habit of grammatical organisation, must have come first, in this view. There is the classic chicken and egg dilemma that dogs anthropological speculation.
But a triadic logic has the advantage that if words and rules are the dichotomous elements of speech acts, then they must co-arise, being each other's context. The habit of abstraction is already built in from the get-go, even if its first expression is a vague as hell.