Anton Benz

You know that joke about economists where everybody's stuck on a desert island with a can of food and the physicist says let's smash it open with a rock and the chemist says let's heat it in the fire till it pops and the economist says let's assume a can opener? Linguists do that too. Or at least people in that weird corner of the language sciences that somehow think language is all about idealized models with preprogrammed agents bled of all real-world, let alone linguistic, content, and, like, game theory.

Large swathes of this were inaccessible to me because I don't speak calculus or algorithm or formal logic, and maybe one day I will because I don't want to be anti-knowledge, but it seems to me that when you're talking about the origin and development of language it's just not useful. You want to look at actual brains, not computer simulations with collections of "traits."

Here, Pieter de Bie programs little migrating dots to interact with like dots in certain ways and unlike in certain ways and share certain traits under certain predictable conditions and then pretends that has anything to do with how people share usages and pronunciations when they, like, go to market and like he's explained where dialects come from. Michael Franke, in a better paper, talks about "non-interference conditionals," sentences like "If you're hungry, there's cookies in the cupboard," and explains it in terms of relevance theory, but it all seems pretty obvious why we say that instead of "YOU MAY EAT COOKIES"; we're providing relevant information without putting pressure on, as a politeness matter, and all his math adds little.

These stand as examples of the kind of paper in here that doesn't do anything much and the kind that does something okay despite itself, respectively. Then there are a bunch more in the first category (the Nash equilibria fly hard and fast) and few in the second, and the occasional nugget of interest emerges but not enough to bother talking about here. But the last paper, by Thomas Scott-Phillips, Simon Kirby, and Graham Ritchie, is quite good. Distinguishing between the emergence of "informational intent" (that is, the intent to communicate something and "communicative intent" (the intent to communicate at all), they have people do a game where they have to without speaking move a little dude around a square in a way that conveys to their partner what colour spot he's gonna end up on and the longer they do it successfully the more $ they get. They show that the case where the one guy tries to develop a pattern and waits for the other guy to figure it out never works, but the case where the one guy does the same thing every time and waits for the other guy to catch on and then they build new things together one by one often works. So you can imagine cave guy going "hoo" when he stretches every morning but it's not till the others start going "hoo" when they see him coming that he can figure out that if he goes "ha" he can let them know that something is intriguingly amiss, perhaps that he's mad--the beginning of communication is having that vessel to fill with some intent. And it's an interaction. And it's difference emerging from sameness. They call it "signalling signalhood." Kinda neat.… (plus d'informations)