A would-be monk's guide to perfect languages

Post Reply
User avatar
Scott Alexander
Posts: 1778
Joined: Sun Dec 19, 2010 1:20 am

A would-be monk's guide to perfect languages

Post by Scott Alexander »

The Galisyin monks have devoted themselves to recovering one of Raikoth's most powerful tools: the so-called "perfect language". So far they have made slow but significant progress.

In order to clarify exactly what a perfect language is, and how the Galisyin monks are going about their task, here is a translation of one of Kayi Klasperion's review articles on the subject. I've translated most of the Galisyin and Pelagian examples to ones from English and Earth to make it more readable.
Some people think a perfect language is one in which it is impossible to lie. That's absurd. Any language that lets you say "The sky is blue" and "The grass is green" must also let you say "The sky is green".

The only difference between a perfect language and a regular language is that a perfect language forces you to either lie or tell the truth. A regular language lets you say all sorts of meaningless blather that sounds important, that can even change the course of people's lives, without being true or false at all.

Let's take an example from politics. Someone says "The Democrats are crooks."

This sentence has perfect grammar, and it seems to make sense in a way that ridiculous phrases like "colorless green ideas sleep furiously" doesn't. But the sentence, as it stands, is neither true or false.

Does it mean all Democrats are crooks? If so, how likely is it that every single member of a party of tens of millions of people is a crook? Does it mean most Democrats are crooks? This encounters the same problem. Does it mean at least some Democrats are crooks? Given tens of millions of people, it's ridiculous to think that none of them are crooks, so this conveys no extra information.

And what does it mean by "crook", anyway? If a Democrat once went a few miles over the speed limit, does that make her a crook? What if she engaged in some complicated form of political corruption that was not technically illegal? We can look up police records and find out exactly how many Democrats have been convicted of felonies in the last year - is the sentence only true if that number is above a certain amount? If a certain percent of Democrats have been convicted of felonies, but a much higher percent of Republicans and independents were convicted of those same felonies, is the sentence still true?

This questions are futile attempts to explain the inexplicable. The sentence has no meaning. It is a verbal gambit trying to trick the listener into associating the concept of Democrat with the concept of criminality, in the hopes that the association will subconsciously stick. It's not a lie - there is a sort of honesty in lies that this sentence totally lacks. It's more of a language failure, a loophole in semantics that lets it fly under the radar.

And the problem is that when someone hears this sentence, they don't ignore it as meaningless. They answer with "No they're not!" or "No, all Republicans are crooks!" or "You only say that because you've been brainwashed" and start a debate which is impossible to settle even in principle. And these meaningless debates over meaningless statements shape not only people's opinions, but religions, cultures, governments, and lives.

A perfect language is one in which "All Democrats are crooks" and a host of similarly flawed propositions sound as confused and agrammatical as "Him goed two they're house" or "the of should chair" or "yarble garble".

In order to start designing a perfect language, we need to identify a comprehensive list of the sort of problems that cause language failure.

1. Missing quantifiers: "Democrats are crooks". Are all Democrats crooks, most Democrats crooks, or only a few Democrats crooks?

2. Nonprobabilistic statements. "God exists". Can you only say this sentence if you're 100% sure God exists? Can you say if if you're 90% sure? What about 51% sure? 20% sure?

3. Overly broad loaded terms. "You are a crook". Is it fair to say this to a murderer? A person who drove 2 mph above the speed limit? A political prisoner in jail for speaking out against a dictator?

4. Scope of metaphors. "The Democrats are like the Communists." This may be true in one limited domain (for example, they have broad support among poor people) but it implies that they may resemble them in other domains as well (being oppressive). A perfect language would limit its metaphors to the domain intended.

5. Failure to separate empirical statements from random personal opinion. "Gay people raise children poorly". Are there any statistics about how the children of gay people do in relation to criminality, school grades, mental disorders, and other usual measures of whether someone is being raised badly? In imperfect languages, it is possible to express strong opinions about the statement above without even considering the existence of evidence; in a perfect language, an assertion should immediately imply need for evidence.

6. Poor handling of comparisons. "Life in the United States is difficult." Compared to what? Life in other countries? Life in other first world countries? Life in the United States a generation ago? Being dead?

7. The passive voice. Classically, "Mistakes were made". Less classically "This beloved leader" - beloved by whom?

8. Inability to handle morality. English handles morality by categories "good" and "bad" and assigning objects to these categories, but doesn't explain standards for doing so - one can simply say, for example "Democracy is good so we should have more of it." A perfect language would include a built-in metaethics.

9. For that matter, non-instrumental "should" is a linguistic failure. "We should raise taxes". Well, whether we should or shouldn't raise taxes depends on who we are and what we want. "We should raise taxes if we want X" is closer to the mark.

10. Too many words with strong connotations. There is a conjugation game about this: "I have firm principles. You're stubborn. He's pig-headed." Or "I'm sensitive. You're emotional. He's melodramatic."

These ten examples are only the tip of the iceberg. Instead of going through more, I want to give an example of how a perfect language could outdo a normal language.

Let's say someone says "Gay people are bad at raising children."

A perfect language wouldn't allow this phrasing. You could still express a similar point, but it would have to be something like "I believe with >75% probability that on average children raised by gay people will do significantly worse academically and socially than children raised by straight people."

The first sentence is a suggestion to feel uncomfortable about gay people, which can only be countered by flinging names at the person involved. The second is an empirical statement to which the obvious response is to look up a study on whether children raised by gay parents get worse grades in school (or are more likely to go to jail, or what). If the study doesn't show what the speaker thinks it shows, he would retract the statement. There is no opportunity for unresolveable heated debate at all.

In the next few chapters, I will discuss first the role of a perfect language (would it exclude poetry? normal conversation?) and some beginning features of what a perfect language's grammar might look like.

User avatar
ari
Grim Auxiliator
Posts: 1802
Joined: Sun Nov 21, 2010 9:45 pm

Re: A would-be monk's guide to perfect languages

Post by ari »

Ah, so it's a language for expressing yourself in boring ways!
Lord Furniture
Not even partially responsible for Malarboria; will take the blame for Caverden, though!
Tallest and therefore Greatest of the Janitors
Eternal Watcher of #micronations

User avatar
Daniel
Posts: 1161
Joined: Fri Dec 31, 2010 3:55 pm

Re: A would-be monk's guide to perfect languages

Post by Daniel »

So it doesn't have sarcasm, irony and hyperboles?
Ghost of Dâniyal the Dead
Disturb not my sleep!

User avatar
Scott Alexander
Posts: 1778
Joined: Sun Dec 19, 2010 1:20 am

Re: A would-be monk's guide to perfect languages

Post by Scott Alexander »

I don't know. Such a language would be more for scientific, philosophical, or political discussion than for everyday speech - kind of a special language of the technocracy. You could probably work those kinds of things in, but it would be stretching the design specifications.

User avatar
Aryeztur
Posts: 2121
Joined: Mon Dec 13, 2010 11:16 am
Contact:

Re: A would-be monk's guide to perfect languages

Post by Aryeztur »

I like this a lot. But as you say, it is only desirable for political and scientific types of speech. The lack of precision of termenology is something I can't stand and is probably one of the leading causes for the misuse of philosophy.
The Great King, King of Kings, King of the World

User avatar
Aryeztur
Posts: 2121
Joined: Mon Dec 13, 2010 11:16 am
Contact:

Re: A would-be monk's guide to perfect languages

Post by Aryeztur »

But under this scheme, I don't think there is any perfect language in real life...
The Great King, King of Kings, King of the World

Andreas the Wise
Posts: 2189
Joined: Tue Nov 23, 2010 11:40 pm

Re: A would-be monk's guide to perfect languages

Post by Andreas the Wise »

Scott Alexander wrote:I don't know. Such a language would be more for scientific, philosophical, or political discussion than for everyday speech - kind of a special language of the technocracy. You could probably work those kinds of things in, but it would be stretching the design specifications.
I don't think you'd need to worry about things like humour in that sort of language, it's really not the purpose.

You always want to deal with vague terms, like bald, heap etc (see the Sorites Paradox).
The character Andreas the Wise is on indefinite leave. But he does deserve a cool war ribbon.
Image
However, this account still manages:
Vincent Waldgrave - Lord General of Gralus
Manuel - CEO of VBNC. For all you'll ever need.
Q - Director of SAMIN
Duke Mel'Kat - Air Pirate, Melangian, and Duke of the Flying Duchy of Glanurchy
Cla'Udi - Count of Melangia
Vur'Alm Xei'Bôn - Speaker of Nelaga, Minister of Interior Affairs, and a Micron

User avatar
Octavius
Posts: 2410
Joined: Sun Dec 19, 2010 7:50 am
Location: Utasia
Contact:

Re: A would-be monk's guide to perfect languages

Post by Octavius »

Sounds a little like the use of New Latin for writing scientific and philosophical texts. The original, classical Latin, processed until it became clean.
Аїомидє Оћтавион
Hesam wrote:Trains are very efficient at making long lines of people dead, and I suppose they could also be used for transportation or something.

User avatar
Harvey
Ghost of Christmas Past
Posts: 1780
Joined: Thu Nov 25, 2010 3:27 pm

Re: A would-be monk's guide to perfect languages

Post by Harvey »

Odd that someone who focuses so much on poetry would come up with these sorts of objections to language. Flowery prose is chuck full of unusual yet interesting comparisons and metaphor.

User avatar
ari
Grim Auxiliator
Posts: 1802
Joined: Sun Nov 21, 2010 9:45 pm

Re: A would-be monk's guide to perfect languages

Post by ari »

But... it's the opposite of odd! Wouldn't you expect someone who's good at understanding connotation to be concerned about its misuse (whether deliberate or accidental)?
Lord Furniture
Not even partially responsible for Malarboria; will take the blame for Caverden, though!
Tallest and therefore Greatest of the Janitors
Eternal Watcher of #micronations

User avatar
Scott Alexander
Posts: 1778
Joined: Sun Dec 19, 2010 1:20 am

Re: A would-be monk's guide to perfect languages

Post by Scott Alexander »

(translated from Kayi Klasperion's Psidion Silkl Chamri by myself)

Some natural languages convey evidential strength through verb forms. We will want to replicate that. Although the perfect language could be either inflexional or agglutinative, the agglutinative makes most sense when we expect to be combining a multitude of data about each verb.

I have been considering the following set of verb forms:

1. a mood indicating an assertion has statistically valid empirical evidence
2. a mood indicating an assertion has anecdotal evidence
3. a mood indicating an assertion is based on widely-accepted common authority
4. a mood indicating an assertion is based on hearsay
5. a mood indicating an assertion is based on direct sensory perception or memory
6. a mood indicating an assertion is mathematical, or tautological, or a definition
7. a mood indicating an assertion is an extrapolation to a hypothetical situation
8. a mood indicating an assertion is meant metaphorically, to illuminate a concept
9. a mood indicating an extraction of information, or summation, from a larger work.
10. a mood indicating an hypothesis or assertion that does not require justification

In a true perfect language, each of these would have its own morpheme; in our sketch, I will represent each with a dash and a number, like "I ate-5 the cake."

Before continuing, I must also introduce (using the # sign) a convention in which we merely contemplate gaining a certain level of evidence. For example, "Dogs are-1# responsible for more attacks on humans each year than cats" means that the speaker believes that if we were to gather statistical evidence, we could support this assertion.

We can now begin analyzing language. Thus, the paragraph:

"I went to the herpetarium yesterday. A herpetarium is a park or building for keeping and looking at snakes. At this herpetarium, there were over ninety different kinds of snakes. Snakes are scary. However, the overwhelming majority of snake species are not poisonous."

Could be written as:

"I went-5 to the herpetarium yesterday. A herpetarium is-6 a park or building for keeping and looking at snakes. At this herpetarium, there were-4 over ninety different kinds of snakes. The overwhelming majority of snake species are-1 not poisonous."

Note that the sentence "Snakes are scary" cannot be translated, because it its evidence base does not correspond to any of these five categories. In fact "snakes are scary" is meaningless - even if we discovered everyone in the world was afraid of snakes, one could still make an argument that everyone in the world was making a mistake, in the same way a child who is scared of a bag waving in the wind is making a mistake, since "bags are not scary". An effective perfect-language translation of "Snakes are scary" might be something like "I am-4 scared of snakes" or "Most people are-1 scared of snakes"

Note also that in many cases an assertion may fit into more than one evidence class. For example, if a leading physicist told you that all physicists agreed that protons were made out of quarks, this is both (1), an assertion with scientific evidence behind it; (3), widely accepted, and (4), something you have heard. In this case, if you believe the physicist, the lower-numbered mood, (1), takes precedence.

Let's examine a real piece of writing now: the first part of an editorial from the politics site NewsMax. In cases where I can't put a mood to the verb, I've used a question mark.
Tom Brokaw wrote-3 a book called “The Greatest Generation. It is-? about a generation that grew up with fathers in the home. They saw-? it as their duty to instill in their sons a work ethic. The greatest generation went-3 on to win World War II.

Newt Gingrich is-? right when he warns that the newest generation does-? not understand or appreciate the value of good, hard work.

Tragically, 40 million children will-1 go to bed tonight without a father in the home to teach them the economic facts of life. One wonders-? how exactly these children will ever learn any kind of work ethic.
We see that even in this relatively innocuous piece of writing, most assertions cannot be supported even in this relatively permissive tenfold scheme. Can we rewrite the editorial in a more exact way?
Tom Brokaw wrote-3 a book called "The Greatest Generation". It describes-3 a generation that grew up with fathers in the home. These fathers placed-1# more importance than those of the current generation on instilling a work ethic in their children.

Newt Gingrich has warned-4 that the newest generation works-1# fewer hours than their grandparents' generation.

Tragically, 40 million children will-1 go to bed tonight without a father in the home. They are-1# less likely than other children to work long hours.
We notice a few things here. First, the editorial now makes several testable predictions:

1. If we survey elderly people and younger people on eg a Likert scale, the elderly people will place more importance on instilling a work ethic.
2. Teenagers work fewer hours today than they did in the 1930s.
3. Children who grow up without a father in the home will work fewer hours.

Unlike the original piece, we can investigate whether this piece is right or wrong. If we determine all four points are correct, even the editorial's strongest opponents will have to admit its case; if we determine all four points are false, the writer would be forced to retract it or face complete ridicule.

(in fact, I've found research showing that point 2, at least, is correct)

But there's a major problem with the rewriting of the editorial: "Today's teens work fewer hours" is an inexact translation of "Today's teens have a poor work ethic." This is not my fault, but the writer's. "Teens have a poor work ethic" is meaningless unless defined. It may be that the writer has a better criterion for work ethic, for example "They refuse to work even when extra money is necessary to keep the family above the poverty line", but in that case the writer should have specified it more explicitly. Writing in a perfect language would have forced him to do so.

User avatar
Harvey
Ghost of Christmas Past
Posts: 1780
Joined: Thu Nov 25, 2010 3:27 pm

Re: A would-be monk's guide to perfect languages

Post by Harvey »

Perfect language = 10 times more words. Well, okay then.

User avatar
Scott Alexander
Posts: 1778
Joined: Sun Dec 19, 2010 1:20 am

Re: A would-be monk's guide to perfect languages

Post by Scott Alexander »

Sorry, could you clarify that? I don't understand.

User avatar
Harvey
Ghost of Christmas Past
Posts: 1780
Joined: Thu Nov 25, 2010 3:27 pm

Re: A would-be monk's guide to perfect languages

Post by Harvey »

Not sure if language-related joke...

Your solution to the issue of ambiguity seems to be to split verbs into ten categories, essentially creating ten possible definitions for each verb.

User avatar
Scott Alexander
Posts: 1778
Joined: Sun Dec 19, 2010 1:20 am

Re: A would-be monk's guide to perfect languages

Post by Scott Alexander »

I'm calling them "moods", which puts them in the category of grammar. An example of English moods would be the indicative "You love me." versus the interrogative "Did you love me?" versus the imperative "Love me!" versus the subjunctive "You would have loved me" .

English handles this through word order and adding a few extra words, but many other languages would handle it through changes in morphology: what I think you're calling "different verb categories".

We see an example of this in English with tense: "I love you" versus "I loved you" versus "I'm loving you".

So I'm not doing anything to verbs that other languages don't do already. There are actually some Australian aboriginal languages that have moods a lot like the system I've mentioned.

I could compensate for the extra complexity by getting rid of tense or something boring like that. Still would be fewer hard-to-remember categories than Finnish.

User avatar
Aryeztur
Posts: 2121
Joined: Mon Dec 13, 2010 11:16 am
Contact:

Re: A would-be monk's guide to perfect languages

Post by Aryeztur »

Moods are fine if you're born as a native speaker of this language, but it would make it hard to learn for outsiders. I recall that most people run into some of the biggest problems when learning other languages when learning how to use their moods correctly.
The Great King, King of Kings, King of the World

Post Reply

Return to “Galinomai”