What Metaphors Mean

Consider the metaphorical sentence: 

(1) God is my copilot.

Now suppose that someone asked you to explain precisely what this sentence meant. You might be able to say various things that would provide a sense of more or less what the sentence was supposed to convey, but it seems that you would never be able to capture the full richness of the metaphor itself: inevitably, you would end up leaving something out. A number of philosophers have pointed to this striking difficulty we face in explaining the meaning of metaphors and used it as the basis for philosophical theories about the nature of metaphor itself. 

In a forthcoming paper, Mark Phelan takes up this supposed fact about our inability to paraphrase metaphors and subjects it to actual empirical study. But Phelan introduces a surprising new twist. Instead of simply asking people to explain what metaphors mean, he compares people's ability to explain the meanings of metaphors with their ability to explain the meanings of perfectly literal sentences. So, for example, people's ability to explain the meaning of sentence (1) could be compared with their ability to explain the meaning of the sentence: 

(2) Bill Thompson is my copilot.

Phelan asked experimental subjects to give paraphrases for metaphorical sentences; then he took these paraphrases and presented them to other experimental subjects, asking them whether the paraphrase captured the full meaning of the original or whether it left something out. Just as one might predict, subjects felt that the paraphrases inevitably left something out. This confirms the usual philosophical view. However, when Phelan did the same with paraphrases of perfectly literal sentences, he got exactly the same result! In fact, he sometimes found that subjects were, if anything, more likely to think that something had been left out of the paraphrases of literal sentences than they were to think that something had been left out of the paraphrases of metaphorical sentences. 

In short, it looks like it really is pretty impossible to explain what a metaphor means. But that is not because of anything special about metaphors. It is merely a reflection of the fact that we can't explain what any sentence means. 

Comments

[Brandon Watson]

I'm glad this research is being done, but I'm not sure how much this particular set of inquiries actually shows on its own. The usual position is not that metaphor resists paraphrase, simply speaking, but that metaphor resists paraphrase because the paraphrase will always have to make precise assumptions that distort the relations of ideas -- in Black's famous line, it says too much and with the wrong emphasis. But paraphrase inadequate due to excessive precision will always maintain a high degree of similarity to the original -- it's the way they are dissimilar more than the degree of dissimilarity that matters. In any case, the fact that it is a particular form of inadequacy in view is the reason why inadequacy of paraphrase is usually connected to the open-endedness of metaphor. Thus the research here only directly affects the inadequacy of paraphrase view if the kind of inadequacy is the same in the case of both the metaphorical and the literal paraphrases. After all, the proponents of the view are not claiming that the only paraphrases that are inadequate are paraphrases of metaphor, but that paraphrases of metaphor are always inadequate in a way more detrimental to meaning than paraphrases of literal statements are. That is, a specific kind of inadequacy is in view. The results still might cut against this view, but it's hard to tell without a better understanding of what people are doing when deciding whether two utterances are similar or whether one utterance leaves out an idea. For instance, we really need to compare the paraphrastic cases with straightforward equivalences in order to make sure that people are not doing something unexpected in their evaluations of similarity. Would we get similar doubts about the similarity of 1+3=4 to 2+2=4? If we did, that would suggest that the experiment is picking up on something else entirely. Likewise, one needs to ask not merely whether an idea is left out but also whether an idea is added -- in ordinary conversation saying of two claims that one leaves an idea out is not the same as saying that one of them adds an idea because each can suggest a different baseline for what is normal -- and not merely whether the utterances are similar but also whether they are substitutable.

[Alève Mine]

Absolutely! Any research about the relationship between this and our ability to acquire knowledge or to reason?

[Eddy Nahmias]

Sorry, I haven't read the paper (yet), but can Josh or Mark explain if this interpretation of the results is addressed: with the "literal sentences" the subjects reading other people's paraphrases will themselves have a more complete and "filled out" sense of what the sentences mean, so they will have higher standards when interpreting others' paraphrases than they have regarding paraphrases of metaphorical sentences (for which the readers also have an incomplete sense of what they mean and how to paraphrase them).

That is, when I'm reading someone's paraphrase of "Bill is my co-pilot" I have a clear sense of how that *should* be paraphrased, so I may find anyone else's paraphrase inadequate. But since I don't have a clear sense of what "God is my co-pilot" really means, I'll think others' grasping attempts as good as I could do. Make sense?

[Mark Phelan]

Thanks for the comments! Brandon, you're quite right about the nature of the inadequacy being asserted. As I write: "The inadequacy assumption that has been central to debates concerning metaphorical meaning is the claim that, although a metaphor and its purported paraphrase may have somewhat overlapping content, the paraphrasing utterance generally expresses content that leaves out some important idea present in, or adds in an important idea absent from, the content of the target metaphor." My experiments may fail to address this thesis, but that at least is what I was trying to get at--and I tried to get at it by asking, not only how similar the statement and its paraphrase are, but also whether one leaves out an idea that the other includes. I like how you put the inadequacy view as asserting "that paraphrases of metaphor are always inadequate in a way more detrimental to meaning than paraphrases of literal statements are." (Of course, that can't quite be the view of some proponents, since some who embrace metaphorical meanings also endorse the inadequacy...but never mind that.) What's essential to this view is that the literal and the figurative have paraphrases that are inadequate in distinctive ways. I try to argue against this sort of view in general in the paper by pointing out that the two kinds of utterances (or the two groups included in my study, at least) are inadequate to exactly the same degree. Thus, it's more parsimonious to suppose that the inadequacy of literal statement paraphrases and the inadequacy of metaphor paraphrases are explained in the same way than to suppose that these admit of different explanations. Of course, this is not a decisive case. I'm happy to shift the burden of proof (and I think the most important part of the paper for doing that is the section raising philosophical objections to previous assessments). And I'm interested in discussing better ways of getting at the purported inadequacy. (There's a lot more experimental work to be done here!) Maybe a good method would be to examine how many ideas people come up with as being left out of (or added into) the paraphrase of one kind of statement or the other in a set period of time, or to look at how long it takes participants to generate ideas that are left out between the statement and its paraphrase. Finally, I see your point about the difference between leaving an idea out versus adding one in--the paraphrase is naturally conceived as the statement that either leaves out or adds in because of the direction of fit. But I'm not sure "leaves an idea out that the other includes" (my wording) suffers from the same problem, and, anyway, participants here don't know which is the paraphrase, so shouldn't presume a direction of fit.

Eddy: I don't consider that view specifically. But I do consider the view that it's hard to paraphrase the literal statement because it's so simple--thus there's no other way to put the idea it expresses; but the metaphor paraphrase is inadequate because it is open-ended. In arguing against this I make the general point (mentioned above) that the mean adequacy ratings are so similar its not plausible to suppose unique explanations.

Alève: None yet. Though I'm interested in exploring the idea that we reason about the mental states of others using certain conceptual metaphors. Do you have specific proposals?

[Metaphorhacker]

Personally, I think Mark Phelan doesn't go far enough in challenging the special status of metaphor. Here's my take on it in a separate blog post (sorry, got too long for comment): http://metaphorhacker.net/2010/08/metaphor-is-my-co-pilot-how-the-literal-and-metaphorical-rely-on-the-same-type-of-knowledge/

[Alevemine]

About language, knowledge and reasoning in general, not limited to metaphors or the mental states of others: our not being able to explain what any sentence means should interfere with our ability to acquire knowledge or to reason. Unfortunately I do not have a specific experiment - a catch 22 ? - in mind, and if some research in that direction is nevertheless done, I'd love to be informed! Indeed this issue had been bugging me for a while and I was very excited to read your abstract.

[Mark Phelan]

Metaphorhacker: Thanks for the interesting post. I tend to agree with your conclusion "that there’s nothing special about metaphors when it comes to meaning, understanding and associated activities like paraphrasing" (as I discuss in my dissertation). Do you think there's something special about metaphor in other respects? Also, I'm familiar with the work of Gibbs and Glucksberg on comprehension and processing time. Do they have work on the inadequacy of generated paraphrases too? I think both are separate contributions to the conclusion that there's nothing special about metaphorical meanings.

Alève: I don't think problems with paraphrase necessarily reflect a problem with comprehension. We could grasp the thought conveyed by any sentence or utterance and just be unable to find a different sentence or utterance that also captured that thought. If we thought in natural language and there were no synonymous terms this would be the case. But it could also obtain in other situations.