Comment by nwatson | Hacker News Reader

Comment by nwatson | Hacker News Reader

I submit this post in response to the article on scripting by Larry Wall (inventor of Perl) (see http://news.ycombinator.com/item?id=1033677, linked article http://www.perl.com/lpt/a/997).

In his article Mr. Wall says: "So it's rather ironic that my views on Postmodernism were primarily informed by studying linguistics and translation as taught by missionaries, specifically, the Wycliffe Bible Translators. One of the things they hammered home is that there's really no such thing as a primitive human language. By which they mean essentially that all human languages are Turing complete. When you go out to so-called primitive tribes and analyze their languages, you find that structurally they're just about as complex as any other human language. Basically, you can say pretty much anything in any human language, if you work at it long enough. Human languages are Turing complete, as it were."

The former Wycliffe Bible Translator/Summer Institute of Linguistics member Dan Everett reportedly counters this in the New Yorker article. "Everett, once a devotee of Chomskyan linguistics, insists not only that Pirahã is a severe counterexample to the theory of universal grammar but also that it is not an isolated case." This article is fascinating.

My grad school experience is too far in my past to exactly distinguish the nuances between "non-Turing-complete" and "not conforming to Chomsky's universal grammar" -- so have at it discussing the differences. In any case, this article describes a language where it's impossible to describe some common human ideas -- and not through lack of vocabulary, and perhaps not even because of inadequate grammar. Rather, the language betrays a completely different mindset where such ideas may not be relevant to survival.

Dan Everett, former Wycliffe missionary, reached crisis when he found no way to describe the notion of a God or Jesus Christ and their relation to humans to this Amazonian tribe. Mr. Everett decided that if there are people to whom one cannot convey a Christian message, the message must not be universal and cannot be real.

Really brilliant article. Thanks!

I'm wondering whether the premise of the language as not Turing complete holds up though. The article claims that the Piraha language lacks abstractions, including counting. Perhaps that could count as incompleteness, but I'm not so sure, and the article doesn't say enough to clear it up. Even if a person cannot use recursive linguistic forms, such as "I saw the dog by the river", they can say, "I saw the dog. The dog is by the river". Can they also say, "Another dog came to the river"? Is that enough context, without tracking the precise count?

Another thought, just because they don't care about a historical fact, is the fact that the story can be told in the language enough for completeness? Even without precise numeric counts?

Does Turing completeness even make sense as a description for a natural language?

I'm using "Turing complete", perhaps incorrectly, in the sense that Larry Wall (inventor of Perl) did in a related article at http://www.perl.com/lpt/a/997, where he says: "Basically, you can say pretty much anything in any human language, if you work at it long enough. Human languages are Turing complete, as it were."

Pirahã seems to be a language where expressing certain kinds of complex ideas doesn't work. Sure, one might say "I saw the dog. The dog is by the river." I'm not sure that's the same as "I saw the dog which bit me while I was down at the river yesterday looking for fresh-water oysters to feed my mother-in-law to impress her because she's telling my wife I'm a loser because my alligator hunting sucks." My impression is it's hard to express the latter thought in the language, even breaking it up into separate sentences, without using recursion, which the language lacks, at some point.

It is not in fact impossible to express that sort of sentence in Piraha. Embedding is accomplished by "noun-ifying" a clause by adding "-sai". The debate is whether or not this suffix represents actual embedding/recusion. This debate has been discussed at length in Language and other sources.

I'm not sure if this applies, but... can't you convert any recursive algorithm into a looping one? In which case iterating over the loop will let you say anything. But this seems to take the metaphor too far :-)

One of the most remarkable parts of the language is that is does not have recursion. Nor does it have ways of explicitly differentiating temporally distant events.

This is an old article, but there are new ones a-comin'; the implications of Dan's work are just starting to be realized.

Reply on news.ycombinator.com