↵ Back to class homepage

I remember in elementary school we learned about figurative language, including similes and metaphors. Here are some examples of the sorts of thing I have in mind:

  1. This cake is as hard as a rock.
  2. This cake is like a rock.
  3. This cake is a rock.

(1) and (2) are both similes, but they differ in whether or not they explicitly mention what feature is being referred to: (1) specifies that the comparison I'm making between the cake and a rock is based on hard-ness, whereas (2) leaves that implicit, and leaves it up to context for the listener to figure out that the way in which the cake is like a rock is that both are hard. (Note that these non-literal uses of simile need to be contrasted with literal comparisons, which may have the same syntactic structure but are not figurative; "This bone is as hard as a rock" or "This bone is like a rock" might be literal comparisons rather than similes.) (3), on the other hand, is a metaphor. Metaphors and similes each involve mapping something between two concepts that usually don't go together (such as cakes and rocks). Similes do it by raising an explicit comparison, as in (1) and (2) above, while metaphors leave the comparison implicit.

For a more real and less simplistic example, let's look at a metaphor used in the below interview between New York City politician Zohran Mamdani and podcast host Briahna Joy Gray, in which they are discussing a group of New York City taxi drivers who had gone on a hunger strike in order to pressure the government to change some laws which were making it very difficult for taxi drivers to survive. In the relevant exchange shown below, Mamdani is describing an earlier hunger strike in New York City that had been successful. At one point he says, "There were excluded workers who were on hunger strike for ... 23 days I think it was. North of 20 for sure." Note how he uses "north of 20" to mean "more than 20". Here we see another metaphorical mapping between two different concepts, this time between numbers and cardinal directions—if we wanted to make this metaphor look more similar to the first example, we could restate it as something like "23 is north of 20". Larger numbers are not literally further "north" (or even literally "above" or "higher than") smaller numbers, but people often describe them in that way. In other words, a metaphor like "23 is north of 20" non-literally expresses that 23 is a bigger number than 20, just like "this cake is a rock" non-literally expresses that this cake is hard.

Figurative language like metaphors is actually very common. For example, just this morning I saw someone wearing a T-shirt that said, "KEEP YOUR EYES ON THE STARS AND YOUR FEET ON THE GROUND". The mapping between two different concepts is not as obvious here as it is in the simple "X is Y" type of metaphors (like "this cake is a rock"), but there still is such a mapping: the word "on" is being used to describe two very different relationships. Keeping your feet "on" the ground means your feet (or at least your shoes) are physically touching the ground, whereas keeping your eyes "on" the stars means your eyes are focused on, pointed towards, or paying attention to the stars. (If your eyes were "on" a star the same way your feet can be "on" the ground, you would be dead!)

(In fact the above example has more metaphors than just the use of the word on. "Keep your eyes on the stars" metaphorically means to aim for high goals or something like that, and "keep your feet on the ground" means to stay practical and realistic or something like that.)

These are far from the only examples. Lots of words show common metaphorical uses like this. For example, Hornstein et al. (2005) point out that verbs often mean very different things depending on their objects; take refers to very different actions depending on whether we take a shower, take a bus, take a phone call, take a book, etc. (They also point out that Godzilla can "take a bus" in the way that the rest of us "take a book".) You can see the same phenomenon with the Chinese verb "打" ("hit"), which means very different things in "打電話" (make a phone call), "打的" (take a taxi), "打聽" (ask around), or "打人" (hit a person). Lycan (chapter 14) suggests that almost all utterances have something metaphorical in them.

How do we explain these different meanings? How can "a rock" mean an actual rock but also also mean "anything hard"; how can "north of" mean physically north of something but also mean "a bigger number than"; how can "on" mean both "physically touching" but also "pointed towards"/"focused on"? One possible explanation could be that there are actually multiple words with the same pronunciations and spellings but different meanings. For example, maybe English has several different words on with several different meanings; e.g., "on1" means "physically touching" and "on2" means "pointed towards or focused on". But a view like this would quickly become unmanageable, because most words would turn out to have tons of meanings (for example, "stand" would have over 20 meanings). A more plausible solution is to assume that on actually has just one meaning, and is used figuratively in other contexts by metaphorically extending its meaning. This is one of the main benefits of pragmatic theory; if we can understand how certain kinds of non-literal language (like implicatures and metaphors) work, then we can explain a lot of this variation with just one or two mechanisms (things like the Cooperative Principle), rather than needing to arbitrarily list millions of different meanings in our vocabulary.

How do we understand metaphors?

While some metaphors are "conventional" and their meanings might just be memorized, we clearly have the ability to understand metaphors we have never heard before. For example, the below scene from Game of Thrones includes at least four novel metaphors:

While these might not make sense without context, people who had been watching the show could understand what all these metaphors meant. The first two metaphors both express the idea that "the realm" is not a real thing, just an idea (the first refers to the claim that the king's throne is made up of "a thousand blades taken from the hands of [the first king]'s fallen enemies", even though in reality there are really much fewer than that). The third metaphor seems pretty easy to understand even without specific knowledge of the show. And the fourth metaphor, the most famous one from this scene, expresses that chaos is an opportunity for someone to increase his power and social position.

There must, therefore, be some explanation for how we understand the meanings of metaphors we haven't heard before.

Early pragmatic theory argued that metaphor is a kind of conversational implicature. Specifically, when a speaker utters something that's literally false (like "This cake is a rock"), we recognize that they have violated the maxim of quality, and then we search for a related meaning that would not violate the maxim. To find that related meaning, we recognize some feature that is typical of rocks and we reinterpret the utterance as meaning that this cake has that feature that is typical of rocks.

There are several problems with this account. One is that not all metaphors come from utterances that are literally false. Zufferey et al. give the example "No man is an island". This sentence is literally true, but nevertheless it's metaphorical; therefore, it cannot have been derived from a violation of the maxim of quality. (An extra complication here is that the sentence has negation. So you might be able to rescue the pragmatic account of metaphor by assuming that the way we understand "No man is an island" is to start by thinking of what it would mean for a man to be an island, then we have to figure out the metaphorical meaning of that (i.e., a person who doesn't depend on anyone else], and then we negate that. Experimental pragmatics research, though, challenges the idea that negation is processed in a two-step fashion like this [see Noveck, chapter 13, for details]. Furthermore, there are other examples of metaphors that come from literally true sentences and that don't involve negation [see, e.g., Levinson, chapter 3].) Some have suggested that metaphors can be realized from violations of the maxims of quality or relevance. But it is not clear to me how this account distinguishes between quality/relevance flouts that lead to metaphors versus those that lead to other types of implicatures.

Another problem is that, even if we assume metaphors are triggered by a violation of Gricean maxims, we still need an understanding of how people figure out the metaphorical meaning—i.e., how they figure out which features are the ones that are being highlighted. There are lots of complicated theories about this (see Levinson, chapter 3, or Lycan, chapter 14, for a review). Where most have settled is this aspect of metaphor comprehension is not a purely linguistic phenomenon, but relies on general psychological mechanisms of analogical reasoning. In this sense it is similar to other aspects of pragmatics—for example, the Cooperative Principle is also something that's not specifically about language, but rather is a domain-general principle of communication that happens to also apply to language.

Video summary

In-class activities

The "pragmatic" account of metaphor we briefly sketched above necessarily assumes that people first access the literal meaning of a metaphorical utterance, and only after they access that meaning (and notice that it violates some Gricean maxim) do they figure out the metaphorical meaning. So, for instance, when reading a sentence like "Those lobbyists are hyenas", you would first recognize the literal features of hyenas (furry, mammal, etc.) and only later recognize the metaphorical features (aggressive, ruthless, etc.).

Another view of metaphor proposes that metaphorical meanings are access directly, i.e., without needing to first recognize the literal meaning.

Have students discuss which view they support, and why. (For a contemporary review with experimental evidence, and discussion of more recent theories than these, see Noveck, chapter 10).

⟵ Non-restrictive modification
Experimental pragmatics ⟶

by Stephen Politzer-Ahles. Last modified on 2022-03-23. CC-BY-4.0.