Scivillage.com Casual Discussion Science Forum

Full Version: This simple structure unites all human languages
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
http://nautil.us/issue/76/language/this-...-languages

EXCERPT: . . . Human language is amazingly creative. If you make up a sentence of any complexity, and search for that exact sentence on the Internet, it’s almost never there. Virtually everything we say is novel. Yet at the heart of this capacity of ours lies an incredibly simple piece of mental technology: Merge. Merge takes two bits of language, say two words, and creates out of them another bit of language. It builds the hierarchical structures of language.

Merge was proposed by Noam Chomsky in the early 1990s. He argued that this single piece of mental technology, plus language specific constraints that children could learn from their linguistic experiences, was enough to capture the syntax of all human languages. [...] Merge has created a self-similar structure: a larger bit of language containing two smaller bits of language. Boxes within boxes. An invisible hierarchy. Now, in spoken or written language, we have to put one word after the other. That’s just the way these channels of language work. We can’t say two words at the same time. Speaking, and to a lesser extent signing, flattens the hierarchy that Merge builds into a sequence, and that sequence has an order.

[...] Merge isn’t very complex, but it does a lot of what linguists need it to do. It applies to discrete units of language (words or their parts). It combines these, not sequentially, but hierarchically. It doesn’t state what order the words have to be pronounced in, so it allows variation across languages. The hierarchical structure is the same in all four types of language we just looked at, but the order of the corresponding words is different.

The fact that the hierarchical structure is the same but the word order is different allows us to express an important idea. The way that languages build up meaning is through Merge. Each Merge comes along with an effect on the meaning of the sentence, and that effect is generally both stable and systematic. That’s why it makes sense to say that the Japanese and English sentences mean the same. They have different orders, but there is a deep commonality. Merge builds both structure and meaning in the same basic way in both kinds of language. Languages are deeply similar, not deeply different.

Merge is also open-ended. As we just saw, it can reuse something it has already created.

Because Merge reuses its own output, it is a recursive process. Recursive processes are well known in mathematics, and form the foundation of modern theories of computing. Merge is a quite specific recursive process—it uses exactly two units and it creates structures that are linguistic. If our human sense of linguistic structure is guided by Merge, that will explain why all human languages are hierarchical and none are sequential. It also explains why human language is so unbounded, why sentences don’t have a natural upper limit. Merge both creates, and limits, the infinite potential of language... (MORE - detailed elaboration)