The Perfect Language

Since at least the twelfth century, people have been trying to build a language from scratch that fixes what's broken about the ones we grew up with. The history of this quest is almost entirely a history of failure — and the failures are more interesting than any success could be.

The Dream

Hildegard von Bingen's Lingua Ignota, created in the twelfth century, is the earliest constructed language on record: a thousand words listed in hierarchical order from Aigonz (God) to Cauiz (cricket). We don't know what she used it for, though "some form of mystical communion" is the usual guess. More than nine hundred languages have been invented since, and almost all have foundered.¹

The serious philosophical attempts began in the seventeenth century. Bacon, Descartes, and Leibniz were all fascinated by how natural languages clouded thought, and wondered if an artificial substitute could mirror reality more faithfully. Inspired by Chinese ideograms — which seemed to convey concepts rather than sounds — they imagined a universal written language that anyone could read, the way Arabic numerals work for mathematics. "Instead of disputing, we can say that 'we calculate,'" Leibniz wrote in 1679.¹

Bishop John Wilkins tried to actually build this in his 1668 Essay Towards a Real Character. His system was a sprawling taxonomic tree where every concept had a precise address: De meant element, Deb meant fire, Debα meant flame. Robert Hooke was so impressed he published a discourse on pocket watches in it. The language was simply too burdensome to use and vanished into obscurity — though its organization-by-meaning scheme was a precursor to the modern thesaurus.

By the nineteenth century, the ambition shifted from philosophical perfection to universal communication. Solresol (1827) used only seven syllables — Do, Re, Mi, Fa, So, La, Si — so that sentences could be sung, played on a violin, or woven as colors into a textile. Esperanto (1880s) was far more successful, reaching perhaps two million speakers at its peak, including about a thousand native speakers. George Soros grew up speaking it. But two world wars and the ascent of global English killed the dream of a universal auxiliary language.

Ithkuil

John Quijada, a former employee of the California DMV, spent thirty years building something much stranger. Ithkuil has two seemingly incompatible ambitions: to be maximally precise and maximally concise. "On the contrary, I think it may turn out that this rugged mountain range trails off at some point" collapses into a single word: Tram-mļöi hhâsmařpţuktôx.¹

Where natural languages have six grammatical categories for verbs (tense, aspect, person, number, mood, voice), Ithkuil has twenty-two, plus eighteen hundred distinct suffixes. Every utterance requires the speaker to specify things that natural languages leave vague: evidentiality (are you speaking from direct experience, inference, or hearsay?), expectation of outcome, contextual appropriateness. When Quijada translated the word "gawk" into Ithkuil, it took him fifteen minutes of introspection to assemble the word from its constituent meanings — adding the "second degree of the affix for expectation of outcome" and the "third degree of the affix for contextual appropriateness."¹

Quijada's inspiration was partly the discovery that "every individual language does at least one thing better than every other language." Guugu Yimithirr (Australian Aboriginal) uses only cardinal directions, never "left" or "right." Wakashan languages require every sentence to encode evidentiality. Niger-Kordofanian languages have aspectual systems more nuanced than anything in European languages. Ithkuil tries to combine the best features from all of them.

Why It Doesn't Work (and Why It Matters)

Ithkuil was never meant to be spoken casually — it was "an attempt to demonstrate what language could be, not what it should be." But the Russian internet discovered it and got excited. A community in Vladivostok translated the whole website. A Ukrainian psychonetics movement invited Quijada to a conference in Kalmykia. The language escaped its creator's intentions.

The deeper issue is the one Quijada was trying to address: the Sapir-Whorf hypothesis, the idea that language shapes thought. The strong version — that language constrains what you can think — has been largely discredited. But weaker versions hold up: speakers of gendered languages really do think about objects differently depending on grammatical gender, even when speaking a second, non-gendered language.¹

If the weak Sapir-Whorf hypothesis is right, then a more precise language might genuinely produce more precise thinking. Quijada was inspired by Lakoff and Johnson's Metaphors We Live By — the argument that conceptual metaphors (life is a journey, time is money) are so deeply embedded in language that they shape cognition invisibly. Ithkuil was supposed to force its speakers to "precisely identify what they mean to say. No hemming, no hawing, no hiding true meaning behind jargon and metaphor."¹

The problem, which every constructed language eventually discovers, is that ambiguity isn't a bug. It's a feature. Irony, metaphor, double meaning — these are the instruments that let us mean more than we say. In Ithkuil, an ironic statement is tagged with the verbal affix 'kçç. Hyperbole is inflected by the letter 'm. But once you've labeled your irony, it stops being ironic. The very precision that makes Ithkuil philosophically interesting makes it humanly impoverished.

This is the deepest lesson of the perfect language quest. Every natural language is, in some sense, broken — riddled with irregularities, ambiguities, and words like "knight." But those imperfections aren't just tolerated; they're the medium through which meaning happens. A language that eliminates ambiguity would eliminate art. The mess is the point.