There is a set of books by the author Mark Z. Danielewski, that features a very irregular (but not necessarily absolutely unique) style. For example, in his debut novel, the word house (and counterparts of that word in other languages) is always colored blue. Or (this is more in a series of books he started, but did not finish, working on many years later), there are quite a few calligrams that show up. I'll focus on the house-case for now.
So, it doesn't seem totally clear to me that the coloration affects the semantics of the word. No house in the story is actually blue, if I remember correctly; especially not the house that is at the heart of the entire text. The reason the author used that color, if not quite the reason he used it on that particular word all the time (in his debut), had something to do with use of blue background framing in some cinematography, IIRC. But it's not like the point, then, is to interpret all mentions of houses, in that book (or in later books, when the word house is sometimes, but not allways(!), colored over in blue again), as implicit references to blue backgrounds in cinematography.
But it also isn't clear to me that the coloration is a syntactic matter, either. (By contrast, the calligraphy examples look kind of like syntactic variations; yet at the same time, the images given calligraphically, have a sort of imagistic semantic value, then, too. And yet even so, the calligrams are usually of objects, mostly animals, already referenced by name in the text used to form the calligrams; so at best is there a sort of "semantic overdetermination" at play, there, as such?)
Also, some parts of the text require physical or at least mental rotation, to read in proper order; or use of a mirror. This is the "ergodic" side of the book. Again, it is difficult for me to see this feature as a purely semantic or syntactic matter, if either at all.
Philosophically speaking, then: if issues of logical semantics and logical syntax are abstractions from the semantics-syntax distinction in natural language, to what logical concepts/functions/what-have-you, do things like font/typeface color, calligraphy, ergodics, etc. correspond, if anything? Can we interpret ineffability in these terms, no less, as if to say that we might not be given to rationally communicate something in the plain semantics or syntax of our propositions and arguments involving them, yet even so we might communicate the otherwise ineffable by means of the logical counterpart "purpose" of these "signiconic" literary styles? (This is not idle speculation, on my part: I was working one time on a story in which there is a scene similar to the one in Dante's Paradiso, where Dante finally "sees God face to face," and I wanted to use signiconic methods to express what was ineffable if expressed only in terms of the words and their mere grammatical arrangements proper.†)
†The first form the idea took was to have the start of the scene involve a black-to-white gradient shift, per letter, at the point of the pivotal word in the scene. So that the first letter would be colored black, the next a very dark shade of grey, etc., until finally the word "faded out" into white. Then, a large number of pages would be left blank, except perhaps for "flickers" of phrases fading in and out again, until eventually the text faded back in conclusively. The idea would be that the blank sections would refer to the otherwise perfectly ineffable, and even therefore comprise a sort of icon of this ineffability.
However, an overarching hyperpremise of the book's style was supposed to be adjoint calligraphy-kineography. In other words, all the words on every page would form a shape, and then all the shapes in sequence would form a "flipbook" superimposed over the interior of the narrative. I never decided whether the blank-out section would then interrupt the procession of the "flipbook." I never made the decision, because I wondered whether the signiconic dimension of the adjoint hyperstyle would already satisfy the ineffability parameters to be encoded into the text, so that blanking the adjunction out, even temporarily, would defeat the purpose of having the relevant signiconic dimension in place. More specifically, the "story" being "told" on the flipbook-level of the text, would not itself directly involve the scene that justified the blanking-out of that level of the text. Now granted, I also considered having microkineographs in the margins surrounding the bulk of the text, on every page, so maybe I could've just had it be that the smaller "flipbooks" remained in place during the blank-out. Without explaining the actual story at issue, I suppose I can't explain the reason why there would be an "ineffable" event therein...