Language

How can phrases mean things their words don't appear to mean? For example, if I'm eating a salad, my friend asks how it is, and I say "not bad," the words "not bad" seem to be extremely open - the salad could be amazing, it could be okay, it could be great or it could be totally neutral; it might even be horrible, so long as it isn't "bad." However, I would normally be understood as saying the salad was okay, rather than any of the other logically plausible alternatives. How does that work?

Comments

I fear the answer will disappoint. "Not bad," in the kind of context you describe, means the same as "okay" because we use the words "not bad" in contexts like that to mean "okay." How we came to use those words that way is something that may be lost in unrecorded linguistic history. And how we keep track of contexts and adjust our understanding to them is a very big and very interesting question that people in a variety of fields are working on. If we knew how we can do what we do, we'd be a lot closer to full-blown artificial intelligence than we actually are. But the short answer is that words mean what we use them to mean, and how we use them changes and varies in complicated ways.

I think there's rather more that can be said here (and, for what it's worth, I don't actually agree that "words mean what we use them to mean").

We probably need to distinguish a couple different things here. One kind of case is that of idiom. These are linguistic expressions, like "kicked the bucket", whose meaning has nothing to do with the component words. These sorts of phrases are really just single words, but long ones, and there are good tests for when you have an idiom. Note, e.g., that I cannot say "The bucket was kicked by John" and have it mean the same as "John kicked the bucket", where the latter is the idiomatic use meaning "John died".

It might well be that "not bad" in this kind of case is an idiom, but the case seems to me to have many features of a case of implicature. Here's a standard kind of example. Suppose Professor Jones writes a letter of recommendation for Mr Smith. The letter says:

To whom it may concern:

Smith has excellent handwriting and was never late for class.

Yours, Prof Jones

Now Jones certainly hasn't said that Smith isn't qualified for whatever the letter was supposed to recommend him for. But he's made his view pretty clear. Why? Well, there's a story to be told about that. Jones knows what kind of letter he's expected to write, but he's totally failed to do that. Why? The obvious thought is that Jones is just saying something positive, and that's all he's got to say that's positive. So you can infer what Jones actually thinks from what he does say and the situation in which it's said.

Here's another kind of case. Suppose I say, "Most of the students passed the test". I do not, when I say that, also say that not all of them did. You can see that because I could continue, "In fact, all of them did". But if I don't continue that way, then you might reasonably infer that not all of them did. Why? Because, if all of them did, I could just have said so. But I didn't. Now, maybe in certain circumstances, that wouldn't be relevant. Maybe all that matters is that most of them passed and, if so, then you shouldn't draw that inference. But in a normal case, you could draw it, and reasonably so.

This kind of example illustrates what's known as the "conversational maxim of quantity", which says, more or less: When you say something, say the most informative thing you can say on the topic, given the general parameters of the conversation.

The case of "not bad" is like that. If the salad were delicious, then you could have said so. Indeed, it's reasonable to suppose you would have said so. But you didn't say so, so it must not really be delicious. Rather, it's merely okay. Minimally not bad. That's it.

Read another response by Allen Stairs, Richard Heck

Read another response about Language