How flabbergasting is Extragrammaticality?

Recently, I have been forced to think a bit harder about the concept of Extra-grammaticality, which is basically the idea that some elements of a linguistic utterance are not analyzed as part of the abstract, underlying sentential representation (even if they are legitimate lexical items that may, on another occasion, be so analyzed). The context for this concern is a paper that I recently submitted to a leading journal, which—the paper, that is—was tossed back at me after being savagely rejected by one of the reviewers. One of the many things the reviewer objected to was a section in which, in passing, I entertained the proposal that utterance final Q-morphemes in East Asian languages such as Vietnamese and Mandarin might be extra-grammatical in the sense defined above; in the particular case at hand,  which concerns the analysis of interrogative không, I was in fact rejecting such a proposal for không. But no matter: the mere suggestion of extra-grammaticality was enough to horrify, indeed flabbergast, the reviewer of the afore-unnamed journạl. Verbatim, if not literally (whatever the literal meaning of flabbergast might be, I think he was exaggerating). Quoth he: (“[The] statement ...that “many languages have lexical elements that are extra-metrical in this sense—present in utterances but not in sentences..."... left me flabbergasted.”

For this, and doubtless other, sounder, reasons, the reviewer was minded to urge rejection of the paper, and the editor duly complied.
(It should be said that this particular reviewer has issues with almost everything that I have ever written on Vietnamese—and this is mutual—so it's not clear whether the flabbergasting came from the content of this assertion or from the fact that I was the one to make it; the generative syntax sand-box is rarely a congenial place to play.)

In my defence—as I have written in a footnote to a revised version of the paper that will appear early next year—I wrote:

I know of no formal analysis that treats elements such as final ‘alright’, ‘ok’, ‘yeah’, etc—as in (i) below—as structurally integrated into English clauses, even though these elements also are functionally clause-typing, signaling a (rhetorical) question:

(i)     a.    I’m coming, alright?!
         b.    She’s my sister, yeah?!
         c.    I know what I’m doing, ok?!

Nor are these the only elements that can be viewed as extra-grammatical, but nonetheless linguistic objects: from affective noises (Brrr!, Whoosh, Zoom-Zoom) to the ubiquitous, conversational like, to paralinguistic gestures, including nods, head-shakes, turn-taking uh-huhs etc., natural language utterances are populated with morphemes—unique conventionalized pairings of sound and meaning—that show no signs of grammatical integration. Some of these elements are of course treated in theories of pragmatics or in cognitive theories of communication—see, for example, Kita & Ide (2007)—but not as syntactic objects.

So, what’s your verdict? I am genuinely interested in receiving your comments. Is extra-grammaticality real, as I naively, and pragmatically supposed, or ‘is every ***** thing that appears in an utterance analyzed as sentential **** constituent,?’—even asterisked expressions that can appear inside lexical****compounds?! Personally, I doubt it, but then, like, I would ****** say that, wouldn’t I…?!