Sugar or Arsenic in your tea?

The readings this week dealt with pragmatic inference, and the leveraging of scalar alternatives to infer information, in both adults and children. What’s an example of this sort of scalar implicature? Well, when your best friend tells you “Some Stanford students are idiots,” you probably won’t feel insulted, because another thing implied there is “Not all Stanford students are idiots.” You can assume this, because your friend probably would have used the stronger ‘all’ in her first statement, if she wanted to say that all Stanford students are idiots (though that is perfectly logically consistent with ‘some Stanford students are idiots’)

In Ad-hoc Scalar implicature in adults and children, Stiller makes an interesting point about the way we leverage features and our knowledge of the world to give others more information. We might point out that our friend has a Mohawk, to distinguish him from others, rather than mentioning that he has, say, legs. Likewise, we assume that the people who speak to us are rational agents. For instance, if there are two faces, one with glasses and one with glasses and a tophat (like in Stiller’s experiment) and your friend mentions “the face with glasses” they mean the one with only glasses, because otherwise they would have distinguished the other option by his top hat, which is a stand alone feature (like Experiment 1 for Stiller).

In Stiller’s experiments, even young kids were able to make these inferences, performing reliably better than chance at “selecting the one-feature item for which a more informative description was not available” (Stiller).

What does this mean really? Well, everyday we are sorting through information in speech. Pragmatic inference, to me, seems to be the act of understanding the intentions of the speaker, and inferring what is truly meant, even when both options are a logical possibility (as in the top hat + glasses example). In natural language, we make these sorts of inferences all the time. If I say “Well, the exam was not, not hard,” my double negation should logically lead you to believe the statement “The exam was hard.” That would be an exact translation. But, if I had truly meant “The exam was hard,” you can probably assume I would have just said that. Therefore, you can infer that I’m trying to impart to you some other meaning, and by saying “The exam was not, not hard” I mean that it wasn’t easy, but maybe it wasn’t quite hard either- and I’m communicating something in the middle.

I’m in a Boolean logic class at the moment, and my favorite class was the one we had on natural language and logic within it. Saying ‘the speaker with the glasses’ is of course a logical way to refer to the ‘speaker with the glasses and tophat’, but why wouldn’t you provide the more differentiating feature, as a rational speaker? This sort of ambiguity of conjunctions, can be explicated further by a funny example my teacher gave in class. It referred to a logical conditional (if-then statement): “If I put sugar in my tea, it will taste good.” From this statement, it is logically entailed that “If I put sugar and arsenic in my tea, it will taste good” because there is in fact still sugar in that tea. But, you wouldn’t expect a rational agent to tell you, “Oh, the tea has sugar in it, so it tastes good” if they knew there was also arsenic in it. Why? Perhaps, because the arsenic seems to be the MOST RELEVANT piece of information, so wouldn’t our speakers have mentioned that instead of/alongside the sugar? It might seem like a crude analysis to Stiller’s experiments, but in some ways it does shed light on the way we sort out the most important information.

When working with a dearth of information, we seem to assume a kind of benevolence on the part of the speaker, that they would have the good-will to refer to the most relevant pieces of information (the tophat in Stiller’s scalar experiment, the arsenic in the tea) so that we might not be left confused, or dead.

In the study discussed by Barner, kids watched boys choose both a bike and a skateboard. Though both “bike or skateboard” and “bike and skateboard” are technically logically consistent, it was clearly shown that “children prefer stronger, more informative descriptions,” or they prefer the 'and' in this scenario, even though when denied the alternative, they are fine using 'or' (Barner 89). This seems to show that kids do make a distinction between what is entailed and what is implied, even though they aren’t very good at generating scalar alternatives on their own, as shown in the Barner study. Because if someone tells me “Sugar and arsenic are in the tea” it gives me a lot more information then “Sugar or arsenic are in the tea…” although, in either scenario I don’t think you’d find me drinking that tea.