I came across an on-line experiment the other day regarding the use of pragmatics in disambiguating references to visual scenes. Pragmatics is the aspect of language where we "read into" what the speaker says and extract more meaning than is seemingly there. For example, when you say "Can you pass the salt?" you're really doing more than asking whether someone has the ability to pick up the salt and move it toward you. Movie aliens and robots may take it literally, but humans are pretty good at making inferences about additional meaning.

Anyway, the experiment used made up words and made up figures, and I won't explicitly link to the site or the experiments because I don't want to muck with their results by talking about the experiment here and sending biased people their way. Maybe in a few weeks when their experiment is finished.

So they'd have a picture kind of like this:



And an audio recording would say something like: "Show me the blue greep." Then you'd select A, B, or C by clicking on it with your mouse, and you'd move to the next item.

So which one would you choose? There are no "correct" answers, though at the end of the experiment they do tell you which responses would be most pragmatic. In this case, it seems pretty obvious that the choice should be B. But what if the voice had said: "Show me the vreck." In this case you would probably infer that since there are two of one object and only one of the other, that the vreck is A. Otherwise the speaker would have given you enough information to distinguish between B and C.

What about this one:



What would you choose if the voice said: "Show me some jeebles." Respond in the comments, please. I disagreed with the authors about what they considered the pragmatic response, and I want to see what other people think.