OHIO with the Enochian Language

So, fun fact: OHIO is the Enochian word for "woe". John Dee and Edward Kelly were truly ahead of their time.¹

The URL title of this page is OHIO A Enochian, where "A" means "in" or "with", which makes it Woe with Enochian. I'm not trying to flex that I am somehow fluent in Enochian—I'm not—but that actually gets at one of the big problems facing the Enochian language: no one can be fluent with it, ever, because of so many limiting factors.

What I want to do in this entry is express what some of those difficulties are and ways that I have tried to address them in my project.

The corpus is really tiny

I've mentioned a few places that the Enochian language has a limited array of translated words:

• from the Angelic Keys: ~1090
• from the diaries: probably ~30 or less, especially when not counting names
• from Liber Loagaeth, probably ~50 or less

This means that, in total, we're looking at less than ~1500 words—and that's including words where the translation has been reverse-engineered or possibly "divined". This might sound like a lot, but it really isn't.

For instance, consider some of the living and dead languages currently in existence: Latin, a dead language with no real new texts being generated, has 215 million words in the Corpus Corporum alone, a digital Latin library. This is 14,333,300% more than Enochian. Compare that to English, Spanish, or Chinese (which obviously have much more written in them available to us) and it puts in perspective how small the corpora of translated Enochian words is.

Why do you need a large corpus?

Let's leave AI out of this for a moment. With humans, the more examples you have of a word used in a sentence, the better we can make sense of what it might mean.

For instance, consider the word defenestrate, which means "to throw out of a window". If the only appearance of that word is "they done got defenestrated," with no surrounding text, our context is too limited to really figure out what the word means. All we can really decipher is:

• it's probably a verb
  • note the "-ed" ending, which usually makes things past-tense in English
  • also "they ... got...", which we may know from other texts that what follows is something performed upon an object
• it's probably past-tense
  • besides the "-ed" ending, "done" and "got" are also past-tense, making it so "defenestrated" conforms

Again, leaving AI out of it, having translations of a corpus can highlight even more what a word might mean. I'll illustrate this with an example:

There's a global phenomenon where people are encouraged to work longer and longer hours. I remember hearing a Partner at a company I contracted to say, "It's truly special to see someone's face the first time they've worked all night and see the sunrise the next day." Regardless of how I feel about it, increasingly in the USA, unions are crushed via right-to-work laws and other efforts, legislators are on the side of the executives and profits, and workers are left without any bargaining power. The status quo has become live to work, not work to live.

Yet, we want to live. We want our lives back. We want to do the things that are meaningful to us, that enrich our inner-beings.

This sentiment is captured in a Chinese expression: 报复性熬夜 (bàofùxìng áoyè), which is the inspiration for the English "revenge bedtime procrastination", where one takes revenge upon the day through procrastinating bedtime.

However,² if you break down what's happening with the Chinese characters literally, the expression is more like retaliatory (报复性) staying up all night (熬夜). To break it down further, 报复 is akin to to retaliate and 性 hints at a thing's essence—in English, it's closest to "-ness" or "-ity". So, when trying to understand the meaning of this phrase, while we might be better describing it as "in the nature of 'to retaliate'".

All of this is to say that we can learn so much about a word by examining it from different angles—especially with translations.

So how does this differ from AI?

Large Language Models, which some people consider nigh-sentient,³ don't actually "understand" words; they only output words that relate to other words based on things they've been trained on previously.

AI has been input enough typos that it recognizes nonstandard spellings; it has had enough made-up words input that it can make up words in similar ways; it has been fed grammar that doesn't grammar the way grammar should so it knows atypical grammars.

This, again, requires millions upon millions of words. Children require communication to learn communication—and a lot of it. And, I'd like to remind you, you can read through almost all of the translated Enochian available to humanity in about half an hour—less if you don't get distracted.

But that's the only issue, right? It's just that there's a small corpus to work with—but there couldn't possibly be any other challenges, right?

Other problems and complications

Olde Enochiann speling and uords

One of the biggest problems with the Enochian language is its spelling inconsistencies. Let me explain.

The Enochian words Dee transcribed are not... spelled the "Enochian way"; they are spelled in a way convenient to communicating some English sense about them. For instance, sometimes he wrote "ENAY" (lord) and sometimes wrote "NA" (lord). You might think like I initially did—"Oh, this means that NA (pronounced like "knot" without the "t") and ENAY (pronounced "en-ay", like the letters N and A) are two pronunciations for the same word."

You might also, much more quickly than I did, spot how "NA" is a clever shorthand for "ENAY"—"N.A." On one-hand, it's nice to have some indication of how it could be pronounced, but it's also frustrating when you're trying to codify someone else's made-up language.

This kind of spelling inconsistency shows up all over the place. One of the easiest examples to highlight this with is a set of words that center around meaning something dealing with fire.

• DIALPRT, third flame
• IALPRG, burning flame
• IALPURG, burning flames
• IALPIRGAH, flames of the first glory

But phonetic glosses aside, there are also interchangeable letters. For instance:

• CABA, to govern
• TABAS, to govern

Another of these is OD, the most used word in the Enochian Keys (again, almost the entire corpus), which means "and". In A True & Faithful Relation of what passed between Dr. John Dee and some Spirits,⁴ Dee and Kelly are told given the spelling of this word by Nalvage.⁵ They are told:

D — The 4th descending 212 ...... This may be T or D.

O — The 6th ascending from the center to the left corner 1907.⁶

Call it OD or OT.

This means the most common word in the Angelic Keys can technically be spelled one of two ways—even though it's only spelled OD in the Keys themselves. That might sound like a good thing, but that means the language as intended does not contain all of its permutations within the corpus.

To put it simply: if the English of Dee's day did not have consistent spelling, which it did not, one could only imagine how much worse a made-up language would fare in Dee and Kelly's hands.

When grammar is option

I'm going to lift an example you can find on Wikipedia—not entirely out of laziness, but because it's actually one of the best (and only) examples available in the Enochian language.

A verb that gets used a lot in the Enochian Keys is "to say" or "to speak". Here are some of the ways it shows up in the text:

• GOHUS, I say
• GOHE or GOHO, he says
• GOHIA, we say
• GOHOL, saying
• GOHON, they have spoken
• GOHULIM, it is said

From this, it seems fair to infer that "GOH" is the root word and anything that follows is conjugating the verb. Awesome—it looks regular enough (as opposed to "to be", which is notoriously awful as far as conjugation goes in languages that do that sort of thing) so we should be able to find evidence of this in other words, right?

Wrong.

Here are almost all of the Enochian words ending in...

There were additional -ON words, but none were verbs (not that these are all verbs to begin with—which is kind of my point), so for ease of reading, I chose to leave them out. As for the rest...

Of this sampling of 23 words, we have a mix of things, but only 5 verbs. This means one of two things:

• the morphology of words is highly irregular
• the corpus is too small to find regularity

Unfortunately, I think both of these things are true.

So what's one to do?

Possible solutions

I've chosen to tease out a few possibilities. If any of this sounds like a bad idea of otherwise foolish, please contact me and let me know a better way. So far, these have seemed reasonable enough to me.

First of all, I have created a substitution map of most of the different variant spellings that exist. It looks something like this currently (just a sample, not the whole thing because that's not kind on the eyes):

{
  "f": {
    "canonical": "f",
    "alternates": [
      { "value": "ef", "confidence": "high", "direction": "both" },
      { "value": "or", "confidence": "low", "direction": "to" }
    ]
  },
  "n": {
    "canonical": "n",
    "alternates": [
      { "value": "en", "confidence": "medium", "direction": "both" },
      { "value": "drux", "confidence": "low", "direction": "to" },
      { "value": "drun", "confidence": "low", "direction": "to" },
      { "value": "dru", "confidence": "low", "direction": "to" }
    ]
  },
  "q": {
    "canonical": "q",
    "alternates": [
      { "value": "qu", "confidence": "high", "direction": "to" },
      { "value": "qua", "confidence": "medium", "direction": "to" },
      { "value": "ger", "confidence": "low", "direction": "to" }
    ]
  },
}

Code scares people a lot of times, but this is actually really easy to make sense of. What I've done is made a map of all the letters that show up (diacritics aside) in the corpus, their canonical form, and alternate versions that could occur in words. For instance, F could be a stand-in for the sound EF and vice versa—which is why the transformation is indicated as going both directions.

F and EF both mean "to visit"—in fact, EF shows up in the twelfth key as "... bind up your girdles and visit us..." while F shows up as part of a compound word: FBLIARD, which means "to visit with comfort," from the passage in the sixth key: "... visit with comfort the earth...".

What's the significance of this?

In a roundabout way, this means we can artificially inflate the corpus. If EF and F can be used interchangeably, FBLIARD and EFBLIARD are both valid representations of the same word.

This is incredibly valuable for my project. As a reminder, I'm trying to programmatically identify root words in Enochian by comparing words with similar definitions that have the same root candidate. With the different kinds of substitution mapping I've done, this enables me to successfully track PRG, PIRG, and PURG all as dealing with fire.

In some ways, this gets around the spelling irregularities by identifying the patterns in the irregularities.

Conclusion

At any rate, I just wanted to talk about one aspect of the project. I may expand and update this entry later because I'm pretty sure there was more I wanted to talk about, but I forgot it and have been writing this for a while, so yeah—good night.