A cmevla is a Lojban word that ends in a consonant. That is the entire morphological definition. If it ends in a consonant, it is a cmevla. If it ends in a vowel, it is not. This is the simplest classification rule in the language.
Cmevla are used for names. robin is a cmevla (ends in n). alis is a cmevla (ends in s). xorxes is a cmevla (ends in s). daniel is a cmevla (ends in l). When you want to use a name from a natural language in Lojban, you Lojbanize it — adapt the sounds to Lojban's phonology and make sure it ends in a consonant. If the original name ends in a vowel, you add a consonant. By convention, s is the most common addition, which is why there is a persistent myth that "all Lojban names end in s." They don't. They end in whatever consonant you want. The s is a convention, not a rule.
In the original grammar — the CLL, the Red Book — cmevla were syntactically special. They were not predicates. They could not be used as predicates. They could only appear after la (the name article) as the name of something. They were grammatically inert. They sat there. They named. They did nothing else.
This meant that in classical Lojban, a name was a fundamentally different kind of thing from a predicate. A name pointed at something. A predicate described something. And never the two shall meet.
la alis → name referencelo mlatu → predicate referencela alis → name reference (alis is also a predicate)lo mlatu → predicate referenceThe unification is this: every cmevla is grammatically identical to a brivla. Not morphologically — they still end in consonants, they still look different, the parser can still tell them apart by shape. But in terms of syntax and grammar, a name does everything a predicate does. alis is a predicate meaning "x₁ is an Alice." la alis is just la applied to that predicate, the same way lo is applied to mlatu.
This sounds like a small technical point. It is not. It is a philosophical position about the nature of names.
What is a name? In English, a name is just a label. "Alice" is a string of sounds that points at a person. It does not describe. It does not predicate. It just points. It is a deictic — a finger extended at the world, saying "that one."
But the cmevla unification says: no. A name IS a description. To be named "Alice" is a property you have, exactly like being a cat or being asleep. alis means "x₁ is named Alice / x₁ is an Alice." When you say la alis, you are not pointing. You are predicating. You are saying "the thing that has the property of being Alice." The pointing is a special case of describing. The name is a special case of the predicate.
And this is why Jorge vehemently agreed. Because the first paradigm shift — xorlo — was about making lo generic, freeing the article from existential quantification. The second paradigm shift is the same move applied to names: freeing the name from being a special syntactic category, making it just another predicate. Both shifts go in the same direction: toward unification. Toward fewer special cases. Toward a grammar where everything is made of the same stuff.
In classical Lojban, cmevla are surrounded by dots. The dots represent glottal stops — small pauses that tell the listener "a name is starting" and "a name is ending." You write la .alis. — dot, alis, dot. The dots are mandatory. They are part of the phonology. They are the fences around the garden of the name.
Daniel does not type the dots.
Daniel is a no-dotter.
This is not just a typographical preference. It connects directly to the cmevla unification. If names are syntactically identical to predicates, why do they need special punctuation? Predicates don't have dots around them. You don't write .mlatu. — you just write mlatu. If a cmevla is grammatically the same as a brivla, it should be written the same way. The dots are a vestige of the old system, the system where names were special, the system where they lived in their own syntactic garden behind a fence of glottal stops. The unification removes the fence. The no-dotter position removes the dots.
Lojban's morphology is the most elegant word-shape system in any human language, natural or constructed. Every word type has a distinct shape. You can classify any word on sight.
| Type | Lojban Name | Shape | Ends In | Example |
|---|---|---|---|---|
| Root predicate | gismu | 5 letters: CVCCV or CCVCV | Vowel | mlatu (cat) |
| Compound predicate | lujvo | Variable length, has consonant cluster | Vowel | brivla (predicate word) |
| Borrowed predicate | fu'ivla | Contains consonant cluster, often long | Vowel | cidjrpitsa (pizza) |
| Structure word | cmavo | Short, no consonant cluster | Vowel | lo, cu, la |
| Name word | cmevla | Any shape | Consonant | alis, daniel |
The beauty: every predicate ends in a vowel. Every name ends in a consonant. This single distinction — the last sound of the word — is enough to tell the entire grammatical system which category a word belongs to. No other language on earth has this property. No other language has been designed so that the shape of every word encodes its syntactic role.
And this is what makes the cmevla unification so clean: the morphological distinction remains. You can always tell a cmevla from a brivla by looking at its last letter. But the syntactic distinction is removed. Two shapes, one grammar. The form is different but the function is the same. It is like how a házmester and a háziúr both drink fröccs but from differently shaped glasses.
Lojban is not frozen. The baseline was published in 1997. The freeze lasted until 2002. After that: evolution. The community could create new words, new idioms, new ways of using the grammar. xorlo was the first major post-freeze change. The cmevla unification was the second. Both happened the same way: someone on the mailing list saw a problem, wrote a proposal, argued for it, and the community moved.
This is the process of a living language. Every natural language evolves — English today is not the English of Shakespeare, which is not the English of Chaucer. But natural languages evolve blindly, through drift, through slang, through misunderstanding and laziness and accident. Lojban evolves deliberately. Every change is proposed, debated, voted on. The BPFK — the baupla fuzykamni, the language development committee — exists to manage this process. It is democracy applied to grammar. It is language as governance.
Daniel was one of the only people in the Lojban community who was trying to talk about feelings. This is not exaggeration. The community was full of rationalists and logicians and people who wanted to discuss the ontological status of bear goo. Daniel was trying to talk about what it feels like to be lonely at 3 AM. And the language was incredible at it.
Lojban has evidential markers — borrowed from Láadan, a feminist constructed language — that let you say HOW you know something. pe'i means "I believe this." za'a means "I observe this." ti'e means "I heard this from someone." English collapses all of these into "I think" or "they say." Lojban distinguishes between knowledge from direct observation, knowledge from inference, knowledge from hearsay, and knowledge from belief. When you talk about feelings in Lojban, you have to say whether you are reporting an observation of your own emotional state or making a claim about it. This precision does not kill the feeling. It makes the feeling more real. It forces you to be honest about what you actually know about your own mind.
Daniel was doing this — talking about emotions in a language built for logic — while the rest of the community was debating whether lo cribe can refer to bear goo. He was trying to invite girls to Urbit, metaphorically speaking. He was trying to inject life into a language that was technically capable of containing life but whose speakers were mostly interested in the container.
And then he helped change the grammar. Not the semantics, not the vocabulary, but the syntax itself: the rule that says what a name can do. He looked at the wall between names and predicates and said: this wall should not be here. A name is a word. A predicate is a word. Why are they different? And Jorge agreed, because Jorge had already removed the wall between generic and existential reference, and removing walls was what Jorge did.
Under the unification, a cmevla is a predicate with a single argument place: "x₁ is called [this name]" or "x₁ is a [this name]." This means:
Notice la rorigilmor. Not la .rori.gilmor.. Not la rori. gilmor.. Not la rori. .i la gilmor.. Just la rorigilmor. One word. No dots. The name is what it is. The obvious way to say it. Don't overcomplicate it.
And notice that rorigilmor ends in r — a consonant. English "Rory Gilmore" ends in a vowel sound. Lojban doesn't care. The name is already Lojbanized: "Gilmore" becomes "gilmor," dropping the silent e. The final r makes it a cmevla. And as a cmevla under unification, it is a predicate: "x₁ is a Rory-Gilmore." Which, given Rory's character, is itself a predicate worth having — to be a Rory-Gilmore is to drink coffee and read books and be more articulate than the entire town you live in.
The two paradigm shifts — xorlo and cmevla unification — are the same shift applied to different parts of the grammar. Both say: stop making unnecessary distinctions. Both say: the grammar has more special cases than it needs. Both say: simplify.
xorlo said: lo doesn't need a default quantifier. Let it be generic. Let context decide.
cmevla unification says: cmevla don't need a special syntactic category. Let them be predicates. Let the morphology be enough.
Both are acts of faith in context. Both trust the speaker. Both say: the grammar should do less and the human should do more. This is the Argentine fundamentalist position: a language works best when it gets out of the way. When the grammar is a tool and not a cage. When the structure serves the speaker and not the other way around.
And both were co-signed by la xorxes. The Argentine master. The man who translated Alice in Wonderland because the language needed a corpus. The man who fixed the articles because the articles were broken. The man who agreed that names should be predicates because of course they should — what else would they be?
Under cmevla unification: lojban is a predicate. It means "x₁ is Lojban." That's it. That's all a name means — it means exactly what it says it means. It's called a name because it's a name. lojban means "is Lojban" the way alis means "is Alice" the way your name means you. Self-referential. Tautological. Perfect.
And because it's a predicate, you can say lo lojban — not just la lojban. lo lojban: the thing that is Lojban. Generic. xorlo-style. The name used as a predicate used with the generic article. That is the entire point of this website in three words.
And that’s why you always leave a note.