# what happens when computer

Lojban is almost certainly one of the dorkiest things in existence. It's nevertheless pretty interesting.
The obsession with constructing a perfect language is an old one. There's a wonderful book about attempts to discover or invent a perfect language—it's written by Umberto Eco and called La ricerca della lingua perfetta nella cultura europea, which was translated into English as The Search for the Perfect Language. Languages like Hildegaard von Bingen's Lingua Ignota or Jon Wilkins' Real Chracter are different approaches to building a language which is somehow perfect or sublime. Other constructed languages have attempted to be optimal in some particular respect—for example, Esperanto and Volapük, for example, being culturally neutral languages At least, nominally culturally neutral. designed for international communication. The Loglan language—a predecessor of Lojban—grew out of a similar urge, but instead of being philosophically perfect or internationally optimal, it was designed to be perfectly grammatically logical and unambiguous.
What this means is sometimes misunderstood: the fact that Loglan (or Lojban) is "logical" does not, for example, mean that wordplay is impossible in the language, or that every statement is perfectly precise. It does, however, mean that a given statement can be perfectly precise. The grammar of the language is unambiguous, which means there are no ambiguous sentences like the English sentence, "John saw the man with the binoculars", and indeed, the book that describes Loglan takes great pleasure in giving a Loglan translation of every possible interpretation of the English phrase, "the pretty little girls' school." The structure of a given phrase is rigorously defined. However, this doesn't mean (for example) that you can only say things which are true, or unambiguous: you can still easily produce nonsense phrases, or be semantically ambiguous, in much the same way that you can form nonsense syllogisms or unclear mathematical equations.
Over time, Loglan's designer—writer and sociologist James Cooke Brown—attempted to enforce copyright over the language, which led to part of the Loglan enthusiast community creating their own alternate version of the language, unencumbered by copyright claims. This language is, of course, Lojban. The creators of this new language built up the vocabulary by an algorithm which ingested and spat out roots from the six most widely-spoken languages at the time, which resulted in words which are very faintly similar to existing vocabulary but only in the barest, most subtly evocative sense. The new language also added new grammatical features on top of Loglan borrowed from other sources.
In contrast to most natural languages, which develop around word classes like nouns or verbs or adjectives, Lojban is primarily concerned with predicates, which it calls brivla. Lojban reference material tends to quickly start using exclusively Lojban words for grammatical concepts, which ) find is one of the most tedious and unfortunate parts of the material. You very quickly come across sentences like, "The rafsi of these gismu combine to form this bridi," which is precise and convenient for avid learners, but opaque and difficult for casual readers. A predicate is a function which takes arguments and produces either truth or falsehood: by convention, we assume that an expression involving a predicate is expressed in such a way that it results in truth. We can express predicates in the same way we express most mathematical functions: for example, we might construct a predicate $$\textit{is-a-cat}$$ which takes an entity and tells us whether or not it is a cat. The following formula (using the $$\land$$ operator to represent and and the $$\neg$$ operator to represent not) expresses that Garfield is a cat while Jon is not a cat:
$$\textit{is-a-cat}(\textit{garfield}) \land \neg \textit{is-a-cat}(\textit{jon})$$
We could also build up predicates that operate over multiple arguments: for example, we could use $$\textit{sees}(x, y)$$ to express that $$x$$ is looking at $$y$$.
$$\textit{sees}(\textit{garfield}, \textit{jon})$$
We can also combine ideas that share variables in order to express more complicated concepts: for example, if we know that Garfield is seeing a cat, but we don't have a name for that particular cat, we can express that using our mathematical notation as well by introducing a dummy variable:
$$\textit{sees}(\textit{garfield}, x) \land \textit{is-a-cat}(x)$$
The Lojban language builds off of predicate logic like this, but provides a system for speaking these kind of formulae aloud. Instead of $$\textit{is-a-cat}(\textit{that-thing})$$, we use the word The words that appear in this colorthat is, #993366 here will have tooltips, so you can easily recall the meaning of each word. tathat to stand in for $$\textit{that-thing}$$, and we express the predicate $$\textit{is-a-cat}$$ by the word mlatuis-a-cat. So, we can express our sentence:
tathat mlatu.is-a-cat
"That is a cat."
We can also use proper nouns like "Garfield" or "Jon", but Lojban insists that, for clarity, we explicitly indicate which words are proper nouns: they must be preceeded by the word lathe-one-named, must be spelled using Lojban's phonetic conventions, and must end in a consonant. This ensures that proper nouns are easily distinguishable from other classes of words. It is likely for this reason that nobody has attempted to translate the works of Abbott and Costello into Lojban.
lathe-one-named garfildGarfield mlatu.is-a-cat
"Garfield is a cat."
The arguments given to predicates are ordered, so for a predicate of more than one argument, we can list the arguments in the correct order. Using the Lojban predicate viskasees, which is analogous to our predicate $$\textit{sees}(x, y)$$:
lathe-one-named garfildGarfield viskasees lathe-one-named djan.Jon
"Garfield sees Jon."
What if we want to express something like, "Garfield sees the cat," though? Well, we could say something like
lathe-one-named garfildGarfield viskasees ta.that tathat mlatu.is-a-cat
"Garfield sees that thing. That thing is a cat."
But even though it's logically similar to what we want to express, it's awkward linguistically. To that end, Lojban lets us turn a predicate into a "thing" using the article lothat-which, which means something like, "The thing such that [a predicate] is true." Therefore, the phrase lothat-which mlatuis-a-cat means, "the thing such that it is a cat," or more idiomatically, "the cat."
lathe-one-named garfildGarfield viskasees lothe-one-which mlatu.is-a-cat
"Garfield sees the cat."
If we don't need or want to specify an argument to some predicate, we can supply it with the word zo'e[unspecified], which stands in for something that's not relevant to the current discourse. Thus, we can translate the more vague statement, "Garfield sees something," as
lathe-one-named garfildGarfield viskasees zo'e.something unspecified
"Garfield sees something unspecified."
If the word zo'e[unspecified] comes at the end of a sentence, we can safely omit it; otherwise, predicates with a large number of arguments would always end in a tedious string of zo'e[unspecified] zo'e[unspecified again] zo'e[this one is unspecified too]. In fact, I've already been doing this without mentioning it: the predicate mlatuis-a-cat takes not one but two arguments—x mlatux1 is a cat of species x2 y means that x is a cat of species y—and viskasees takes three arguments—x viskax1 sees x2 under x3 y z means that x sees y in the condition z. Some predicates take as many as five arguments: for example, klamacomes/goes, according to the standard Lojban reference documents, means:
x1 comes or goes to destination x2 from origin x3 via route x4 using means or vehicle x5
That means we can express the sentence, "Jon is coming," as:
lathe-one-named djanJon klamacomes zo'e[to somewhere unspecified] zo'e[from somewhere unspecified] zo'e[by an unspecified route] zo'e[using an unspecified vehicle].
"John is coming."
or, more concisely, as
lathe-one-named djanJon klamacomes.
"John is coming."
We can also rearrange the order of arguments to a predicate: the word se[swap x1 and x2] is used to swap the first two arguments to a predicate, so that
lathe-one-named garfildGarfield viskasees lathe-one-named djanJon.
"Garfield sees John."
is identical in meaning…although perhaps not in emphasis! to
lathe-one-named djanJon se viskais seen by lathe-one-named garfildGarfield.
"John is seen by Garfield."
We can use those argument-swapping words with lothat-which, as well: lothat-which viskasees means "the one seeing" or "the see-er", while lothat-which se[swap x1 and x2] viskasees means "the one being seen." This is a convenient way of building up a very large amount of vocabulary: from the five argument positions of the word klamacomes/goes, we can derive five "nouns": lothat-which klamagoes "the go-er", lothat-which seswap x1 and x2 klamagoes "the destination", lothat-which teswap x1 and x3 klamagoes "the origin", lothat-which veswap x1 and x4 klamagoes "the route", and lothat-which xeswap x1 and x5 klamagoes "the means of transportation".
Now we have the grammar necessary to tell relatively basic and banal children's stories:
lathe-one-named djanJon viskasees lathe-one-named garfildGarfield. lothat-which mlatuis-a-cat ciskaeats lathe-one-named lazaniaslasagna.
"Jon sees Garfield. The cat eats the lasagna."
Lojban of course is more than just what I've introduced above: tenses, In fact, I've been acting as though these sentences are in the present tense, but by default Lojban sentences do not specify time at all: they could contextually be interpreted as past, present, or future. pronouns, compound clauses, and so forth. Lojban also has a relatively large set of "discursives", which are words used to structure conversations in a rich and descriptive way, and several other interesting features. But I've shown the three major components of the Lojban language: proper nouns, structure words like lothat-which and se[swap x1 and x2], and predicates, which are the principal distinguishing feature of Lojban.