John Quijada and Ithkuil, the Language He Invented | Hacker News Reader

John Quijada and Ithkuil, the Language He Invented

newyorker.com

78 pointsph0rque27 comments

Classical Sanskrit was designed with this exact goal in mind.

http://www.aaai.org/ojs/index.php/aimagazine/article/view/46...

There are still native speakers of Sanskrit. The cool thing about Sanskrit is that not only is its syntax formalized, but people have tried to give an account of its semantics over thousands of years.

There are no native speakers of Sanskrit - some people have tried to revive Sanskrit by speaking it - sometimes in large groups. Usage of Sanskrit as a day to day language probably ended hundreds if not thousands of years ago. There are still books written in the language and quiet a number of people can understand it. It is a matter of debate if Sanskrit was any time used as a primary language for communication.

It is a matter of debate if Sanskrit was any time used as a primary language for communication.

It is not a matter of debate as there are speakers even today. Declining definitely, but still significant.

http://en.wikipedia.org/wiki/Sanskrit#Contemporary_usage

The 1991 and 2001, census of India recorded 49,736 and 14,135 persons, respectively, with Sanskrit as their native language.[1]

Yes - but native language in census terms do not mean mother tongue. There are few villages in India which adopted Sanskrit as first tongue - That is an experiment not akin to being part of a continuity. At the most, you can say these native speakers revived a dead language.

I know this is a very sensitive topic given the language chauvinism that is prevalent in India. There are at least two villages which speak Sanskrit (just Google it). It was definitely the case that there were more speakers of Sanskrit (Vedic+Classical) and people who have Sanskrit as their mother tongue. People sort of resort to gymnastics to deny the obvious. As long as there are people speaking a language natively and people genuinely interested in it, I don't think you can call that language dead. But I agree, Sanskrit speakers are on the decline and it is a bad thing.

http://www.mangalorean.com/news.php?newstype=broadcast&b...

But surprisingly, today in Mangalore, there is one unique family where all the members are using Sanskrit as their mother tongue and they all interact with each other in this language only. Seems highly improbable but nevertheless, it is true.

I think one reason for the decline is that you need to be pretty intelligent to learn Sanskrit (vs say Tamil or English). I sort of feel that the human race itself was more smart a few thousand years ago. (See http://rt.com/usa/news/intelligence-stanford-years-fragile-5...)

I can pretty well understand Sanskrit and have learned it for 10 years. It is definitely not difficult to learn. 'Sanskrit' died several hundred years ago as a spoken language - and efforts to revive it by adopting it has yielded mixed results. (The villages you mention are adopters). A language is said to be living when 1) There are people who speak it for daily purposes 2) It continues to evolve to address needs of the present. Sanskrit passes criteria 1 with some determined followers, but fails criteria 2. Point to be noted is that characteristics that make Sanskrit orderly and strict, is an offshoot of the language being ear marked for restricted scholarly usage and not the tenet of a living breathing language. That said, there is a vast literary treasure awaiting people who are determined to learn Sanskrit and there are enough avenues available in India and abroad for the same.

And it comes with the added benefit of making one able to read ancient literature in it's original form. Thanks for the link, fascinating stuff that such a language would be the perfect match for AI. Stranger than fiction.

what does that mean that NLP and sanskrit go so well together? Is that accidental?

It is definitely accidental given that they had no computers back then. It makes sense as Classical Sanskrit was an effort to remove ambiguity in Vedic Sanskrit and improve precision while still being a natural language. A lot, not all, of modern NLP is about removing ambiguity.

The way we describe grammars can be attributed to one of the first Sanskrit grammarians, Pāṇini.

http://en.wikipedia.org/wiki/Backus%E2%80%93Naur_Form

The idea of semantic roles can also be traced back to him.

http://elies.rediris.es/elies11/cap5111.htm

It is odd to say NLP is about "removing ambiguity". Rather, it is about _coping_ with ambiguity (of the linguistic variety). Using a language with (supposedly) less ambiguity would just be a cop-out, not a real solution.

Agreed. We need to understand how humans do language fully with ambiguity. We should also study Sanskrit and understand it fully from an AI perspective.

Otherwise, it is like trying to go to a planet in the Andromeda galaxy without first ever going to the moon. Never gonna happen.

well... I mean do you think that a language evolved that is this easy to parse accidental? I mean our brains are computers using mimicry and heuristics... why isn't that accounted for more in the creation of this language?

Thank you for this!

This reminds me of an interesting short story entitled "Understand" that I read a few years ago. The premise of the story deals with the question "what would it be like to find meaning and order in everything you saw?". Naturally, language is one facet of this.

"I'm designing a new language. I've reached the limits of conventional languages, and now they frustrate my attempts to progress further. They lack the power to express concepts that I need, and even in their own domain, they're imprecise and unwieldy. They're hardly fit for speech, let alone thought."

Full story here: http://www.infinityplus.co.uk/stories/under.htm

This article is a great read, I read it a while back and was blown away. I never gave any thought to the advantages of any language over any other, I always looked at it with the programmer's lens. Just as all modern programming languages are Turing Complete and therefore equivalent (preferences aside), I assumed that the same would be true for most spoken languages. Ithkuil and the like really show that there is a lot more room for variety than most people imagine.

To say that all languages are equivalent because they are Turing Complete is just not true. That means they can decide the same class of problems, it does not mean they can or do decide them in the same way. In particular some languages are inherently faster than others.

Not sure where I read it, but it was recently - languages that sound faster really just tend to have more syllables per concept, and all languages generally tend to have the same speed of communicated concept.

Whoever said that was not a competent programmer, discussing programming languages.

Research into programming languages says that lines of code written/day tends to be surprisingly consistent. But the amount of functionality embedded in that code varies widely depending on the language.

Faster languages are ones that take over basic mechanics for you (eg memory management), infer things for you (eg compare C++ to go), or offer more convenient abstractions (eg compare scripting languages to C or Java). The difference is usually an order of magnitude, but can be more in the right circumstances.

I'm talking about spoken languages. Spanish, for instance - low amount of concepts per syllable, which is why it sounds so fast.

I read the same thing, the article I read also said that Chinese encodes more info per sound so it sounds like the speakers are speaking more slowly. English was somewhere in the middle.

> That means they can decide the same class of problems, it does not mean they can or do decide them in the same way.

OK so far...

> In particular some languages are inherently faster than others.

Which is an example of deciding the same class of problems, just doing in in a "different way."

So your example is not providing a counterexample to the claim; it's providing an example.

Measuring time of computation is counting the number of primitive operations. I'd say a language is a collection of capabilities that play nicely. Given one language, we can create a second that is also RE by taking each instruction in the first and adding a nop, or calculate ack(10,10) and throws it away, or whatever. Any algorithm that the second language computes with at least one step runs in fewer primitive steps in the first language by design.

Beyond a theoretical consideration, in practice there is much more than preference providing the difference between languages. If they were the same, you shouldn't have any trouble implementing combinators in brainfuck.

I understand, I'm saying that this article suggests that there may be languages which enable us to solve entirely different problems.

Maybe the comparison of a classical computer vs. a quantum computer is more apt. There are many problems which they can both do, but for some a quantum computer is infinitely more efficient. For some linguistic problems, a certain language may be much more efficient.

In the same way Lisp or Haskell enable us to solve entirely different problems from writing directly in assembly perhaps, since you avoid having to deal with all the mundane details and can reason about your program at a much higher level?

Some years ago I read a wonderful - and very opinionated - book entitled "The Loom of Language", which opined that had Wilkins' "Real Character" (briefly described in the article) become the norm for scienfific discourse then scientific progress would have ground to a halt, because Wilkins' scheme systematised human knowledge into a fixed hierarchy of concepts, but made no allowance for new concepts. I wonder if Ithkuil might suffer from the same essential weakness (which is not to detract from the awe in which one must hold such an exquisitly intricate creation)

"Ithkuil has two seemingly incompatible ambitions: to be maximally precise but also maximally concise". I immediately thought about perl.

What? You thought of perl as maximally precise? Have you ever used it? I wish I could find the link about MJD doing some perl consulting, fixing an apparent bug and then finding that what appeared to be a bug was yet another perl special case.

Reply on news.ycombinator.com