Part I · Chapter 4

The Code

Putting the object together

You now have all the pieces. Reading a word endolinguistically means making three moves in a row:

1. Remove what isn't root: endings, prefixes, suffixes (chapter 2).
2. Pull out the skeleton of the root: its consonants, as they sound, in order (chapter 2).
3. Turn it into classes: each consonant to its region (chapter 3).

What's left is a very short sequence: almost always two or three classes, in an order. That, exactly that, is an endolinguistic code. If it has two classes we call it binary; if it has three, ternary.

The idea of the code is the heart of endolinguistics, and it comes from its founders, Meulemans and Elias. Comparing dozens of languages, they discovered that certain patterns of consonants repeat over and over, cross centuries and borders, and gather around them words whose meanings have something in common. The classes we use here to write those patterns are my way of organizing them.

Order matters

Now a detail that looks minor and is the soul of the whole thing.

21 isn't the same as 12, even though it uses the same digits. Codes work the same way: Θ·Λ is not the same as Λ·Θ. Same two classes, different order, different code.

To talk about this precisely we need two words.

• The nucleus of a code is its set of classes, regardless of order. It's written in curly brackets: the nucleus of Θ·Λ is {Θ, Λ}, and so is the nucleus of Λ·Θ.
• The code is the sequence with its order. Θ·Λ and Λ·Θ have the same nucleus, but they're different codes.

When two codes have the same nucleus in reverse order, we say one is the inversion of the other.

An inversion you can hear

Read these two lists out loud, one after the other.

(In rueda, the u is the vowel of a diphthong, so it doesn't count: the skeleton is r · d.)

On one side, words of ground, of settling, of what stays put. On the other, the same pair of classes reversed: words of turning, of going around, of motion. The same two notes; played in one order they sound like rest, in the other they sound like movement.

Now comes the honest part, and it matters as much as the pretty one.

Terrain and rotate aren't relatives. Terrain comes (through French) from the Latin terra, which goes back to a Proto-Indo-European root meaning "dry." Rotate comes from the Latin rota, "wheel," from a different root meaning "to run, to roll." Two separate histories. No etymological dictionary connects them.

So what did you just hear? Not a kinship: an endolinguistic reading. It's the observation that words with different histories carry the same nucleus in reverse order, and that their fields of meaning (settling, turning) also seem to be related. Endolinguistics proposes that this isn't entirely by chance: that the order of the classes orients meaning. But that's a proposal, not a proven fact, and it has to be treated as one.

A hint for the challenge. One example that doesn't fit doesn't destroy a proposal on its own, but it can't be hidden either. What decides whether a reading holds is counting: how many words fit it and how many don't, in many languages, compared with what pure chance would produce. Two or three pretty examples prove nothing, and one ugly one refutes nothing. That's why, in this manual, every time I show you a reading like this I'll tell you it's a reading. That habit is the difference between a method and a collection of coincidences.

A code doesn't mean: it orients

If, reading the table of terrain and rotate, you thought "so Θ·Λ means ground," this is the moment to brake. That thought is natural, and it's the wrong door.

A dictionary tells you what a word means. A code doesn't do that. A code orients: it points to a zone ("words that carry me tend to live around here") without saying what's in the zone.

It's the difference between a treasure map and a compass. The map promises the chest. The compass only swears where north is; what's up north, the terrain will tell you.

There's an example that shows this better than any explanation. Look at these words:

There are two families here. Calorie, caliente, and caldo belong to one: the Latin root cal-. Gélido, cold, and kalt belong to another: they're relatives of each other (gélido is, in fact, the Spanish cousin of your cold), and they come from a root that meant cold. The two families aren't related. Calorie, caliente, and gélido begin their root with the same pair Χ·Λ.

And here's the finest part: Italian caldo, which means "hot," and your English cold have exactly the same code, Χ·Λ·Θ. They come from two unrelated families, and each one sits at an end of the same axis: temperature. German kalt, the sibling of cold, carries Χ·Λ·Θ too.

One nuance, to be honest about it. In Latin, the d of calidus belonged to the suffix, not to the root. In Italian the suffix fused, and today that d is part of the root cald-. We didn't do that when we took the word apart: the language did it by itself.

Does the code contradict itself, showing up in both hot and cold? Only if we expected it to mean something. If we understand that it orients, the apparent contradiction becomes information. That code points to an axis (temperature) and each word walks that axis to a different end. A compass that marks the north–south axis isn't disproved because one traveler goes north and another goes south.

(This last part, again, is a reading. What's documented is that gélido, cold, and kalt are relatives and that calorie, caliente, and caldo belong to another family. That caldo and cold meet in the whole code Χ·Λ·Θ, and the two families on the axis, is the endolinguistic observation.)

Use is the referee

If the code doesn't say what a word means, who does? Use. The real sentences a word lives in, what people actually do with it.

The code tells you where to look; use tells you what you see. That division of labor is the most important rule of the method, and whoever flips it (whoever first decides what a code "means" and then forces words to obey) is no longer observing the language. They're dictating to it.

How not to read a code

Back to the trap from chapter 3, because with codes it becomes even more tempting.

Take the word star. Its code is Σ·Θ·Λ. Someone might reason like this: "Σ is the hiss, which feels like going out; Θ is the limit; Λ is what flows. So a star is something that goes out, gets defined, and flows: a star!"

That reasoning is wrong, and it's worth understanding why, because the problem isn't just that the conclusion is doubtful.

1. It treats each class as if it had a fixed meaning. We already saw it doesn't.
2. It adds up the pieces, as if the code were a phrase made of three words. But classes don't add up: they bond, and the result depends on the whole and on the order, the same way water looks like neither hydrogen nor oxygen.
3. It closes what the code opens. A code points to a wide field; a two-word gloss shuts it inside a label. If a reading fits in two words, it was almost certainly built by adding up.

What's the right way, then? Go to the words. Star has documented relatives: Latin stella, which gave Spanish estrella, and German Stern. What those words share, and what people do with them in real sentences, is what tells you which field that code orients. Meaning is read from the relationships between words, not from the pieces of the code.

Summary of Part I

With this you have the complete instrument. Let's review:

• The skeleton is the set of consonants in a word, taken from the sound (chapter 2).
• The root is the piece that carries the meaning; the code comes from it (chapter 2).
• The classes are seven regions of the mouth, each with its own feeling; on their own they mean nothing (chapter 3).
• The code is the sequence of classes in the root, in order; the nucleus is the same set without order; inversion changes the order and keeps the nucleus (this chapter).
• The code orients, it doesn't mean. Use is the referee.

In Part II you'll use all this for what I promised at the start: learning languages. And the first thing you need to know is who your real relatives are.