The basic, fundamental, revelatory indispensable and sufficient faith is faith in language. All talk of human ends, the meaning of life, etc., end in—meaning, in the most literal sense: can we mean what we say and say what we mean? The implication of the originary hypothesis is that that’s all any of us wants or needs, and all the more specific creeds, pertaining to specific historical or legendary events, theological specification, ritual orders, are all just ways of ensuring that we mean what we say, or that we can. I have been saying (and meaning) this for many years now, and it’s worth revisiting the question now that machines can generate language as fluently as humans can.
The origin of meaning what we say lies in the originary scene, where all participants pointed to the same object at the center, referring to the object as a way of indicating the deferral of appropriation, with the entire body, gesture and posture, poised so as to give assurances that this deferral will endure for the duration of the scene, then transitioning into the sign-mediated sparagmos (division). I have formulated the implications and transferability of this event in the phrase “saying that this is the same,” the ultimate sign of faithfulness, since it can never be the same without the “authentication” provided by everyone else saying it as well.
Saying what you mean and meaning what you say seems simple, and the kind of thing we often use words like “honesty,” “sincerity,” and “integrity” to describe. Those words do have meaning, but anyone with a literary sensibility is aware of the fragility of those stances, of the way honest, sincerity and integrity can easily dissolve into naivety, gullibility, complacency and much else. All of those qualities would need to be ascertained and authenticated as well, and the preservation of one’s integrity at the cost of ignoring and neglecting some of the preconditions enabling whatever self-consistency one takes to be proofs of integrity is really preserving more of a simulacrum of integrity.
Meaning what you say is also having meant what you said, taking into account the way biographical and historical drift introduces various wedges between what was said and meant. It means risking ending up at the center, with all the probing and questioning and dismantling and deconstruction that entails, in all kinds of ways you couldn’t have anticipated and prolonging under changed conditions the scene upon which you meant what you said; it may even mean depending on others to come along and prolong that scene. Saying what you mean and meaning what you say in the Stack or on the Expectant Scene requires a disciplinary space.
I have been calling what I produce an “idiom” for a while now as way of disavowing and disallowing any metalanguage and, more precisely, giving no rest to the metalanguage of literacy (as a way of producing other modes of literacy than predicated on the simulation of a scene of speech [logocentrism]). I will continue to do so but the problem of metalanguage still presents itself insofar as I do, of course, have a privileged vocabulary derived from the originary hypothesis which does provide a map of human scenery. My response to this has been that originary hypothesizing, or center study or, perhaps most appropriately here, anthropomorphics, aims more at entering and steering (cybernetically) other idioms rather than operating supersovereignly over them.
On any scene, one idiom will be privileged over others, so anthropomorphics is privileged on the infrascene of emergence of any scene. So, an infra rather than metalanguage, and there’s no need to deny “privilege” insofar as the things that needs to be denied is scenelessness. Now, though, thanks to the idiom of computation, I can also say that anthropomorphics operates an indicative layer just below the threshold of the declarative that tracks and tags paths from ostensive-imperative circuits through the interrogative. “Indicative” as the grammatical mood pointing out things directly, indicative as indexicality, indicative as the threshold of evidentiality, akin to symptoms. The indicative collapses the figures of the metalanguage of literacy into the layers and degrees of scenic reportage and scene inhabitation the metalanguage of literacy compresses in such a way as to block other and further possible compressions.
Anna Wierzbicka’s primes, as, in my reading, the residue of a kind of originary declarative field, are a critical component of this indicative layer. So, let’s subject the word “meaning” to the scrutiny of the primes as a way to start bringing some order into meaning what we say/saying what we mean. The word “meaning” is tricky—it seems so fundamental, but it is of course not a prime, and its etymology seems uncertain, but I suspect it is ultimately very close to “think” and “intention,” with “intention” itself being reducible to something like “this way.” If saying what you mean can be converted into saying what you think, we have a new way into the question.
It’s easy enough to understand why someone might not want to say what they think but not so obvious why saying what you think should pose any difficulty. Presumably you think in language, so saying what you think should simply involve repeating what you are saying to yourself so that others can hear it. But if that were the case, “think” and “say” would not need to be different words. There must be something to thinking irreducible to saying or, more importantly I think, “think” adds something irreducible to “say.” Helen Bromhead shows that “I think” in 16 th and 17 th century English does not, as is the case very commonly in contemporary usage, disclaim knowledge but, rather (she is not so explicit on what, exactly, “I think” does add to a sentence) as the announcement of the setting of a scene.
I mention this because, despite the differences in meaning, in both cases “think” includes some excess beyond “say” and, in fact, alludes to the scene of the saying. (Part of resisting the metalanguage of literacy is disqualifying as unproblematic realities nominalizations of the primes like “thinking,” as a substantive activity. If we ask “What is Thinking,” “the present participle of the verb ‘think’” is the right answer. What are we doing when we use that verb or nominalize it is the better question.)
We can broaden our own scene here, first by recalling the pragmatist and Wittgensteinian maxim that the meaning of a word is its use. This would seem to indicate a very different approach than the one I am attempting by grounding meaning in the primes and, indeed, the equation of meaning with use would seem to resist Wierzbicka’s entire thesis. Saying that the meaning of a word is its use is both true and question begging and circular, functioning more as a kind of caution against fixing meaning in abstraction from particular collocations, texts and iterations, so I’ll implement that caution by bringing in the whole array of primes so as to ensure we’d never be studying meaning in abstraction from use.
It’s not just a question of saying what you think, which itself would have to be modified as a kind of harmonization of what you say and what you think, but of an articulation of say with do, say with feel, say with want, say with see, say with hear, say with happen and say with know. We often, for example, speak in terms of “doing what you say” in the sense of confirming a commitment or keeping a promise but once we introduce the necessary difference between do and say we’d have to say something more like “do so that others can see what you say in what you do.” A whole series or, maybe, style, of maxims regarding ways in which all the primes could be harmonized with each other, all under the aegis of saying “this is the same” upon a scene “furnished” with the primes would yield the entire “meaning of existence.”
So, we could say that one says what one means, one does what one says, one does what one wants, one knows what one did, one knows what one feels, and so one, which is to say, the knowing and feeling, the wanting and doing, the thinking and saying, and so on, is the same insofar as we hold a particular configuration of the other primes constant so as to “bracket” the differences between any pair of primes. A particular disciplinary space, derived at some distance from some juridical and, further back, ritual, space, would open to ascertainment the relation between what someone knows, what they say and what they do in a particular case.
If someone presents as their highest desire as knowing or seeking the truth; or communing with God; or partaking of the divine; or living a saintly or sanctified life; or adhering to God’s will; or participating in eternity by extracting oneself from worldly concerns and interests; etc., etc., we can acknowledge while offering respectfully (and even offering for use and confirmation) a kind of translation or explication of that desire into some prime-derived maxims. Wierzbicka herself clearly has theological concerns (her most recent book, I think, which I have not yet read, is an explication via the primes of the Nicene Creed), and even if, as a Christian, she wouldn’t quite endorse the dissolution of the Creed into the primes, I think the possible convergence of all forms of desire for transcendence into a kind prime-ology accords with the spirit of her work (the “spirit of her work” would yield an interesting rendering into the primes). (And if it doesn’t I can live with that as well.)
This maxim-ization of the primes as a kind of originary declarative idiom or indicative layer gives us a way of confronting and converging with the vectorized, latent spaces of the next token predictors. Questions about whether computers really think, or are conscious, or reason, are trivialities based on trivial, analytic, philosophy. Can computers do what they want or say what they think are more interesting questions, because they point to the scenic element in any generation of language. Even more, can we see that they do what they say and all the rest and say that we see them doing so in such a way that we in turn can be seen to say what we mean?
The time has passed when we could keep finding new ways of negating whatever would make machinic intelligence like humans on some cognitive terms; the better question is whether they are like humans enough that would justify including them under the juridical categories covering human activity. How do we see each other articulating know and do, do and happen, say and think, and so on, so as to make juridical categories applicable and ritual participation possible? There may not be a single answer—the entire menu of possible prime articulations may be unavailable to computer programs now or at any other time without it being possible to exclude the possibility that they would acquire them or that our ritual and juridical categories might in some cases be applicable and in some cases not.
I’ve proposed that the vocation of humans will hitherto be to train the AIs, and while a degree of irony might be suspected this is really no different in principle than the ways in which humans have always defined and constructed themselves in the course of shaping their environment. This proposal allows for the ongoing friction that is the best way of maintaining human distinctiveness, not as a residual resistance to computation, but as a kind of alienation, being other, to computation (turnabout is fair play here—the computers can be expected to be xenophilic as well). Computation will keep showing us, in response to our prompts and through the use of the data we’re incessantly feeding it, what it looks, sounds and even feels like to say what one thinks, do what one says, know what one sees, see what one hears, feel the way one knows, and whatever other maxims we may generate (it will of course be able to generate such maxims) while such approximations will continue to generate differences.
(I think at some point Stanley Cavell’s The Claim of Reason , which works through thought experiments predicated upon the desire to know that the other is human, many of which elicit the violence and even rage implicit in such desire—with some remarkable readings of Shakespeare along these lines—will become newly interesting to people in these contexts. The irresistible desire to know for sure whether the other is really human, in short, will eventually lead one to tear apart the other and seek out the proof somewhere in its interior. Hannah Arendt made similar observations.) When questions like “can we sue a language model for defamation, or breach of contract, or prosecute it for fraud” become real we will know it. Can they take communion or be married? These questions provide much better “benchmarks” than IQ tests and criteria for determining what we need to know when we invoke “consciousness.”
Language is the apocalypse which, as is well known, is essentially synonymous with “revelation.” In theo-political terms, the idiom is the restrainer or Katechon. I’ve been constructing the Natural Semantic Primes as a kind of originary declarative layer over a more originary ostensive and imperative “substrate.” How we operate the declarative layer derives from the imperatives directed our way and received from all the inappropriate ostensives out there. If people are pointing at things that solicit our attention and participation on the scene and yet you are unable to say “this is the same,” you are out of sync and therefore spiritually bereft—the search for meaning begins with such experiences of meaninglessness.
So, you want to find a way that you can, on that solicited scene, say that this is the same, even if that involves remaking the scene, or exiting the scene and recreating it or finding another elsewhere—or, more precisely, detect and make visible some way in which the middle of the scene might be the end or beginning of some other scene, its end or beginning the middle of some other scene, etc., and assert or inhabit that scene. The indicative layer is located just below the threshold of the declarative, where the declarative’s limning of the interrogative and the interrogative’s prolonging of the imperative can be brought into view, and so it is key to revelation, to saying what we mean/meaning what we say.
The declarative is itself a response to the command to bring your attention in accord with some command issued in “reality,” and the question is a kind of forgiving or patient way of presenting that to you. And registering all this involves creating idioms that draw upon the entirety of our semiotic resources layered over the indicative layer. Evacuating the metalanguage of literacy does not involve a linguistic primitivism; rather, it enjoins one to register the coming through of meaning via the entire grammatical stack (ostensive-imperative-interrogative-declarative) as sensitively as possible. One wants simultaneously to fly under the radar and operate the radar at ever greater degrees of detection. In that case, saying what one means/meaning what one says means enacting the grammatical stack so as to maximally show the seams between the verbal forms along with the stitching.