Browse Tag

linguistics

The Meaning of His Own Words

by Andy Dibble

The foundation stone Kabbalists retrieved from the ruins of Mohenjo-Daro seemed a veritable Rosetta stone.  Indologists would finally understand the language of the ancient Indus Valley Civilization.  Linguists hoped it would determine why many ancient languages are staggeringly complex.

Like the Rosetta stone, “The Lord of Wide Rivers” repeated the same message in parallel, once in Harappan and again in an archaic version of Vedic Sanskrit. Vedic was known, so scholars could read the Harappan message, and thereby unlock the other four thousand Harappan inscriptions discovered hitherto.

The Vedic was a prayer to the four-faced god Brahma beseeching him to stay awake a little longer (or if he sleeps, may the night be brief and day short in coming). It spoke of cycles, of stars whirling in the firmament, of cosmogonic tides and undertows, of perigee and apogee, of the shackling of words to meanings, and of the bewildering darkness in which all bonds shall break.

The unnamed author chose every Vedic noun, verb tense, and prefix with care, even to the extent of violating the poetic meter of his verses, a mortal sin according to later pundits.  With such obsessive clarity, breaking into the Harappan language should be easy.

It seemed an eye in the Harappan script meant both sight and thought, a winged-horse meant transformation, and that stacked parentheses indicated quantity. But none of these inferences made sense when checked against other Harappan inscriptions, and all predictions about the function of the prongs, dots, and other modifications to the base symbols proved groundless.

In hindsight, this difficulty was unsurprising. Vedic was fiendishly complex, and if Vedic descended from Harappan, Harappan should be even more intricate because grammars tend to simplify as speakers use a language over time.

Tellingly, the Vedic began, “The Lord of Wide Rivers will execute me for betraying the hidden language to our adversaries, but if even I–one of his code-slaves–cannot understand, the language is already lost. So as the cycle dips down, I write this so that I might understand the meaning of my own words.”

#

In the late 2020’s, there was a revolution in natural language processing. The dream of six decades, that programmers would program in everyday language, was almost realized. Most lawyers were out of work because software could write briefs indistinguishable from the work of the average legal mind. Social media persona could be software or human or both, and rumor on the matter diverged from reality more often than it agreed.

Deep learning algorithms began to unriddle Harappan. The chief difficulty was that every inscription had multiple meanings, much like the picture of a duck that is also a picture of a rabbit. One message was ostensibly a contract to exchange a quantity of sheep for garnets. But read another way, the same symbols divulged a murderous conspiracy. Beneath that was the intimation, potentially of proto-Zoroastrian origin, of a cosmic sacrifice.

One Harappan seal was a picture of entwined water serpents, secondarily a game of snakes and ladders, and thirdly the first four axioms of Euclidean geometry. But supposing the eye of a serpent in retreat was a vanishing point, the image took on perspective, and the axioms established hyperbolic geometry. The Harappans had refuted Euclid, more than two millennia before Euclid.

But even the most scrupulously trained algorithms could infer nothing with high probability. Human intuition was necessary to complete the picture, and intuition keened that Harappan symbols were in fact ciphers, that subterranean meanings are realer than surface meanings, that Harappan was always closer to meaning everything than one thing.

But a language that always expresses everything, expresses nothing.

#

The 2030’s were the Age of the Panopticon. As within the panopticon of Jeremy Bentham and of Michel Foucault, it did not matter if someone–whether corporation, deep state, foreign power, or AI overlord–was watching, only that they could be, and not just in the stupid way of keyword scanning, hypertext semantics, and search engine indexing. Machines could read, and in reading they understood.

Many retreated from social media, or frequented closed forums that, supposedly, were inaccessible to the uninvited. Courts would not rule against the possession of concealed mobile phones; there was an exodus from public spaces too. Utopian communes swelled.

But for the marketeers, the busy bees of the gig economy, the celebrities, and the wannabes, the dominate impulse was to shout as loudly and as publicly in as big a space as one could, and there is no space bigger than the Internet. They reassured themselves: the Powers That Be only wished to present ads more intelligently. Still, they avoided alleged “trigger words,” deployed hopelessly standardized locutions, and prayed the Argus eyes of AI were resting.

When it came out that Kabalsoft’s reclusive CEO was not man but machine, everyone assumed the firm’s meteoric rise was the machination of an all-wise Executor. Pressured by shareholders, publicly-traded companies everywhere automated their leadership in a frenzy. No company could remain competitive with mere organisms at the helm.

Rapid-fire legislation, first in the European Union and then in the United States and China, mandated that software serve a strictly advisory function, and so it was. But executives and directors still deferred to their calculating counselors, and when they defied, who could say whether that defiance was itself anticipated by inscrutable neural networks, whether computers knew even the shrewdest minds better than they knew themselves?

#

When Kabalsoft unveiled a quantum computer architecture advanced enough to shatter all available encryption, the last redoubt of online privacy was overwhelmed. Now, there was just one unbroken code: Harappan. But as a code, it was useless. It could not be modern or general purpose. Artificial cultivation would inevitably sterilize it, render it dumb and limpid to machines.

But Harappan proved that human genius for language could confound machines. And who are the true sages of language? Young children, as Noam Chomsky demonstrated.

The United States skimmed children, aged four to seven, from its melting pot and abroad. No more than two of the conscriptees spoke the same language, and like the pairs of Noah’s Ark, most every language was represented: Dutch so rich in idioms, English the ever-weird, Finnish for its fifteen cases, Sanskrit squirming with ambiguous compounds, Arabic for Qur’anic convolution, six-toned Vietnamese, Japanese to say much with little, Dyirbal rife with unspeakable taboos, isolate languages like Basque and Burushaski, Ebonics and argots, patois and pantomime, clicks and whoops and growls.

Miraculously, it worked. The code-talking children inverted entropy, inverted Babel. They understood one another, and only they understood one another. They learned secrecy and resilience, and only then learned state secrets.

The best minds of China wrestled with the fabulous omniglot but failed to master it. They learned from failure, and in rugged Xinjiang, assembled an omniglot pod, which drew most on Silk Road languages, Zen koan, temurah, and haiku.

Nations hung in equipoise until a day when even the ten Sefirot blinked. A terrorist faction, “Kabalsoft Reborn,” published grammars for both omniglots in two-hundred sixteen languages.

No one read them. They were too huge for comprehension, but the unknown is fearful, and fear suddenly thickened again.

#

There was one last code to slice and splice, a last descent. The little ones were already so nearly right, the unfathomable genius already there, if it could only be unfurled, the cerebral cortex grown within a roomier skull (and taught compliance–the young are too forthright, too prone to defect).

All this could be done with genomics. And it was.

But as the cycle dips down, as entropy overcomes information and words detach from meanings, one will master himself and recount this story so that he might understand the meaning of his own words.

~

Bio:

Andy Dibble is a former academic and Sanskritist turned healthcare IT consultant. He has supported the electronic medical record of large healthcare systems in six countries. His fiction appears in Writers of the Future 36. This is his second story in Sci Phi Journal. (andydibble.com)

Rabbit

by Joel Page

To escape I am trying to forget English. This is difficult for a grown man, but I have certain advantages. First, I am in solitary confinement, which means that people rarely speak English at me, so I am not prompted to it. Second, I am fluent in ALT-9, which is comprised entirely of words pronounced and spelled exactly like words in English. So when I think of a word that I used to know in English, there is something to take its place. There is another node in the brain for the sound to go to.

I’m lucky. The man in the cell catty-corner to mine delivered the English-ALT-9 dictionary to me before – and I mean just before — the CO’s relocated him and declared the dictionaries contraband. In his case, the dictionary was legitimate evidence, so there wasn’t much question about his right to it. But then he started selling them, and the inmates started shenanigans with them (asking for the keys to their cells in ALT-9 and whatnot) so they decided they were a security problem.

Right after I went in, processing power exploded in the wake of the quantum computing revolution. So some wise guy decided to make new languages by scrambling an edition of Webster’s New International, pairing words with the definitions of other words at random. Each time he did this (or had his program do it), he imagined then that he’d created a foreign language dictionary, which provided English definitions for the words of a new language – it’s just that  all the words in the new language were spelled and pronounced exactly like English words. Then he kept making dictionaries until he’d made every possible one.

So then according to the guy in the catty-corner cell, the wise guy stole some examples of naturally occurring English conversations from unsecured phones. He made transcripts of the intercepted conversations and had his program translate them into English, as if all the conversations had actually been spoken in each of the several quintillion new languages defined by his new dictionaries. Then he took the transformed conversations and fed them to AI bots, which evaluated each of them for its plausibility as an English conversation. Finally, human readers agreed that the ninth candidate language identified by the bots made sense of both the transcripts that had been fed to the bots and new spontaneously occurring conversations. The first eight candidates (languages “ALT-1” thru “ALT-8”) made sense of the pre-existing conversations, but failed to make sense of any newly recorded material, leading the wise guy to conclude that their ability to make sensible conversations out of the originally recorded material was just coincidence. But the dictionary defining the ninth candidate language (ALT-9) consistently made sense of new, naturally occurring speech. And so he concluded that everyone is always speaking both English and ALT-9, all the time.

Like I say, the catty-corner man had the dictionaries because they were evidence in his case. Legit evidence. Crucial evidence, because the cops had used ALT-9 to convict him. They asked him questions in ALT-9 without giving Miranda warnings and he responded to the English meaning of the words in their questions. Well, turns out he’s also answering in ALT-9 (because everyone always is) and in ALT-9 his answers were incriminating – they confessed to his presence at the known crime scene. They ask him whether some species of rabbit can change their sex, and he thinks they’re asking about the genitals of rabbits, which they are in English, but in ALT-9, the question means “where were you the night the tall man was shot?” And he answers, “what kind of lunatic cop mind game is this?” But in ALT-9 that means “on the rooftop near the docks,” which answer they use against him at trial, because the victim got shot on a rooftop near the docks.

So he moves to suppress the confession (the ALT-9 meaning of the answer), because the cops didn’t Mirandize him. But he loses because the court says if he isn’t thinking in ALT-9, the ALT-9 meanings of the words he spoke aren’t his words. Rather, they’re the words of another being that speaks and thinks in ALT-9, but co-occupies his brain and body. Personhood is thought, the court says, and thought is language, so different language, different person. And you can’t complain that a witness besides yourself wasn’t given Miranda warnings before they snitched you out.

So then his lawyers say, “well, it’s not fair to send the ALT-9 being to prison for something the catty-corner man did.” But the court doesn’t like that either. It says, “well, the ALT-9 being isn’t going to prison because no one is talking to him like a prisoner.” It says, “we don’t know what will be said to cause him to stay in the building we call a ‘prison’ but it probably isn’t ‘you have to stay here as punishment for a crime.’” Like maybe the ALT-9 being is choosing to be in the prison-building.

#

And that gets me thinking: maybe if I learn ALT-9 and forget English by the time I’m out of solitary, I won’t be in prison anymore. I’ll hear or think whatever words are making the ALT-9 beings stay in the building. But maybe I’ll be choosing to stay here, like they are. Like a monk.

So I spend a long time with it. For hours I stare at objects, such objects as I can find in the cell, and repeat aloud their ALT-9 meaning, desperately trying to drive out their English meanings. I stare at the sink and say the ALT-9 word for sink. I stare at the wall and say the ALT-9 word for wall. I jump and say the ALT-9 meaning for jump. And so forth.

It’s hard – even as my English wanes, there are things I know how to say in English that I don’t know how to say in ALT-9.* And the mind reaches for language in silence like a hand reaches for a lifesaver in the ocean.  But it can be done; I believe it can be done. I’ve seen guys come out of solitary knowing fewer words than they knew going in, and they weren’t even replacing the words they forgot. When I am tempted to think in English, I repeat aloud the ALT-9 word for “freedom,” drawing that freedom to myself as I do, the freedom of confinement chosen.

There’s another possibility, of course. Maybe the court was right and people who think and talk in ALT-9 are just completely different people than people who talk and think in English, even if they happen to occupy the same bodies and brains. In that case, forgetting English might kill me.

Maybe. But maybe that’s the point. I don’t like it here.

Either way, though, I have to wonder who’ll be speaking and thinking in English once I’ve forgotten it. Someone has to be if I’m off speaking and thinking in ALT-9, since the existence of an ALT-9 speaker implies the existence of an English speaker. I guess I could go into a coma, or my body will die, because the universe does not permit a divergence between English and ALT-9. The Word is an atom, and it shall not be split.

Or maybe when I’ve passed to ALT-9 another poor English speaker will be born behind me to serve the rest of my sentence. Sorry, pal.


* These English words are a last indulgence. I fear I may not be able to say them in ALT-9, and wish to say them in English while I still can say them in something.

~

Bio

Joel Page is a public defender in Dallas, where he writes appeals for federal prisoners. His fiction has appeared in The Fabulist, Fleas on the Dog. Thimble, and is forthcoming in Speculative City.