The Limits of Metaphor

"This question is the devil’s work."

Discuss the issues the previous sentence raises from the perspective of the philosophy of language.

---

The biggest issue with this sentence is that the sentence is a metaphor. Metaphors are very confusing to philosophers of language. As Lycan points out, “Philosophers like language to be literal.” (Lycan, 2018, p. 197.) Metaphors are completely at odds with literal language. If we take the statement ‘this sentence is the devil’s work’ literally, then it means that the sentence is the devil’s work. It literally implies the devil wrote the sentence. At a bit more of a stretch, it implies the writer is the devil (sorry Sophie.) Taken literally, the sentence is not true. Metaphors annoy philosophers because they do not fit neatly into truth-conditional theories of language. One example would be Ayer’s verification principle (Ayer, 1946.) To

paraphrase Ayer, this principle states:

‘The truth value of a statement is related to how verifiable it is.’

With our sentence, its truth value depends on how verifiable and observable its truth conditions are. In this case, we cannot verify the devil exists, nor that the sentence is the work of the devil. The sentences truth conditions are unverifiable, so it is a false statement. However, it is not a meaningless statement. We humans use metaphors all the time. William Lycan himself points out that everyday language is more nonliteral than literal (Lycan, 2018, p. 197.) It seems far too radical and unhelpful to simply eliminate all metaphorical language then. But it also seems very unclear how metaphor relates to the mechanics of language and how metaphor relates to truth. Do we need just adjust our theories of language to include metaphors? Or do we need to construct an entirely new theory altogether?

Before discussing what other philosophers think about the mechanics of metaphor, it will be helpful to clarify what a metaphor actually is. This question itself is a philosophical debate, with definitions varying massively in the philosophy of language (Lycan, 2018.) There are several definitions of metaphor available, but the one that seems to cover the most bases is Google’s definition:

“A figure of speech in which a word or phrase is applied to an object or action to which it is not literally applicable.” (Our dictionaries, 2022.)

In other words, a metaphor is only a metaphor if it is a nonliteral statement, although Davidson (1978) would disagree with this and I will touch upon Davidson’s view later. In contrast, philosophers who subscribe to naïve simile theory, figurative simile theory and pretence theory would probably agree with Google’s definition. Philosophers disagree on and squabble incessantly over what a metaphor is, how metaphor works and how metaphor relates to truth. Stanley Cavell (1969) says metaphors are paraphrasable. Instead of saying ‘this sentence is the devil’s work’, one could paraphrase this metaphor by saying:

• This sentence resembles the devil’s work.

• This sentence is like the devil’s work.

• This sentence is very difficult… and so on.

This distinguishes literal language from nonliteral language quite well for Cavell, although there are objections to his position, especially from aesthetic philosophers. Cavell points out a flaw in his own distinction too! He thinks the ‘and so on’ part of paraphrasing a metaphor is an issue. There are an infinite number of ways one can interpret a metaphor. Cavell dubs this ‘the pregnancy of metaphors’ and he believes this muddies the waters of meaning (metaphor intended.)

This is because the more interpretations of a metaphor there are, the less clear the metaphor’s meaning becomes. This argument only goes so far though, considering there are metaphors that are supposed to have multiple meanings, as Davidson (1978) points out. For example, a double entendre is supposed to have two meanings. Donald Davidson has a completely different perspective to Cavell, Grice, Moran or Frege. He argues that metaphors do not actually have a nonliteral meaning; they just mean what they say. He does not think that metaphors have a ‘novel’ meaning beyond other words (Davidson, 1978.) In other words, metaphors do not deserve special treatment - we do not need to adjust our current linguistic theories for them. Davidson is not denying that metaphors have a point, nor that by paraphrasing a metaphor we can understand it better. He merely believes that metaphors do not transcend truth-conditional theories of language, because metaphors are just part of language.

Davidson’s position is arguably very simple and elegant, but it has its fair share of

detractors. One counter position comes from Richard Moran. His theory of metaphor is somewhat antithetical to Davidson’s. He argues metaphors cover a wide range of language (like Lycan does) and that nonliteral language operates differently to literal language. To use Davidson’s words, metaphors and nonliteral expression deserve ‘novel’ treatment. To quote Moran:

“Literal language is precise and lucid, figurative language is imprecise, and is largely the domain of poets and novelists. While literal language is the conventional ordinary or everyday way we have of talking about things, figurative language is exotic or literary and only need concern creative writers.” (Moran, 1997, p. 249).

With our sentence, Moran would argue the truth-conditions of the sentence are not that relevant to its meaning. It is the meaning of the sentence that is important, particularly in the case of nonliteral language, like with our devilish sentence. Our sentence is meaningful and that is what matters, not that it is verifiably true. Moran would not deny that there is a literal interpretation to the devil sentence too, but he would argue that the point of a metaphor is to interpret it nonliterally, and this nonliteral realm is what gives metaphor meaning. For Moran, language is two, whereas language is simply one for Davidson.

One more point Moran makes is he believes metaphors are image-centred and override our logical faculties when we hear them. In his words, metaphors have an ‘imagistic’ capacity that ‘has a force that goes beyond agreement with what it asserts.’ (Moran, 1989.) His beliefs on metaphor are what Lycan calls a ‘pretence theory’ of metaphor (Lycan, 2018.) It is so-called because from this perspective metaphor is based on pretence i.e., pretending. Metaphors are asserting one thing but are really saying something else. In other words, they are pretending to assert one thing through an image, but saying another.

Whilst Moran’s pretence theory is compelling and ties in nicely with other philosopher’s positions on metaphor e.g., Cavell’s paraphrase theory, it has a big shortcoming. It seems paradoxical to have a sentence that says one thing but actually means another. It also seems paradoxical that there are sentences that override logic. A law of grammatical language is not a law if it can just be contradicted or ignored! Another issue that arises is this: how would we know which sentences are metaphorical and which are not? Why would this very sentence not be a metaphor that actually means something else? There does not really seem to be a criterion for when a sentence becomes ‘imagistic’ and is supposed to override literal interpretation. This is why some people do and do not get the exact same metaphor – sometimes someone is unaware a statement is supposed to be metaphorical at all!

Searle has a remedy for Moran’s issue, but his remedy also has problems. John Searle takes what others have dubbed a ‘pragmatic’ position on metaphor, or what Lycan dubs a ‘pragmatic theory of metaphor.’ To summarise Searle’s position, he believes metaphor depends on the speaker’s intended meaning, not the sentence’s meaning (Searle, 1979.) For Searle, a metaphor is defined by what the speaker's intentions are. He points out that there are non-verbal cues a speaker might use when deploying a metaphor in conversation e.g., an exaggerated tone of voice, to make it more obvious that a metaphor with nonliteral meaning is being used. This is what some philosophers and psycholinguists call a ‘pragmatic twist.’ In other words, the speaker’s intentions are ‘twisting’ the literal meaning of a statement (Hills, 2022.) This is why we can read, ‘this sentence is the devil’s work’ and understand that it has a nonliteral meaning i.e., ‘this self-referring sentence is very annoying.’

The pragmatic position is problematic though. Searle’s thought expands upon Grice’s thought, particularly his ideas of conversational implicature and his 4 Maxims (Grice, 1989.) Grice’s ideas are all based upon his idea of the cooperative principle, a now well-respected concept in social sciences and linguistics. To paraphrase Grice, this principle states:

‘Whenever you communicate, state what is as relevant, as minimal and as

understandable as possible.’

Grice further elaborates upon the cooperative principle with 4 maxims, which aim to make communication as clear as possible. As briefly as possible, these are:

1. Quantity – state the necessary amount of info; not too little, nor too much

2. Quality – try to state what is true, accurate and evidence-based

3. Relation – state what is relevant and on topic

4. Manner – avoid being obscure and ambiguous

   
   

Metaphor is problematic for Gricean models of language and for a pragmatic theory of language. One reason is metaphor often violates one or more of Grice’s 4 maxims of implicature. Take our example: ‘This sentence is the devil’s work.’ This statement clearly flouts maxim 4, since the sentence is nonliteral and requires more thought to decipher its meaning than a literal statement would e.g., this sentence is difficult. The statement also violates maxim 2. We do not know if the predicate exists at all; the devil might not exist for all we know. Therefore, the statement is not being as true or as fact-based as possible and violates maxim 2. It seems people can use nonliteral language which violates Grice’s maxims yet understand the speaker’s intent. Grice’s maxims only go so far, and conversational implicature seems to be much broader than Grice’s maxims infer.

Another issue for Searle’s pragmatic theory of language and conversational implicature theory is people willingly violate the cooperative principle. This happens in espionage, negations and in court. It happens anywhere else people find it advantageous to deceive with language. Conversational implicature can never be verified because people can and do lie. If liars use metaphors to deceive too, then it muddies the waters of metaphorical meaning even more. This is because metaphors already often have more than one meaning, so if metaphors also have ‘opposite’ meanings to their intended meaning due to deceitful implicature and/or violating the cooperative principle, it makes metaphorical language horribly unclear. For example, if somebody who is pretending to be cooperative says:

‘This sentence is the devil’s work. Discuss the issues…’

It could actually mean from a pragmatic perspective:

‘This sentence not hard and is very simple. Nevertheless, discuss why it is incredibly complicated.’

It starts to become impossible to know what is actually being said with this train of thought. The way people use language is inordinately complex, and one could argue quite successfully that Searle’s and Grice’s ideas make metaphorical matters even more complex than they already are. One more differing perspective on metaphor is this: perhaps metaphor is not about the speaker at all, nor mutual agreement, nor implicature… metaphor is about the language itself.

Max Black argues it is the interaction between words in a statement/sentence that makes a literal statement become a metaphorical statement (Black, 1979.) By analysing the ‘logical grammar’ of sentences, Black discovers a pattern. Metaphors juxtaposes two things against one another. He calls this the ‘focus’ and the ‘frame’ of a metaphor. The focus is the part of a sentence that is the metaphorical part, and the frame is everything else that ‘frames’ the metaphor. With our sentence, the focus is ‘the devil’s work’ and the frame is ‘this sentence is…’

From this perspective, Black argues that we find the two differing parts (focus and frame) of a metaphor and compare them with one another to see what they have in common. He dubs this their ‘system of commonplaces’ (Black, 1979.) This point is illustrated by Emily Ayoob in her excellent essay on Davidson and Black (Ayoob, 2007.) See below for a visual representation of Black’s frame-focus distinction. Ayoob uses the metaphor ‘John is a wolf’, but Black’s focus-frame distinction applies to any metaphor, including our devilish one:

In other words, we are seeing what overlaps between the focus and frame of a metaphor. With our example, the two things the focus and frame i.e., ‘the devil’s work’ and ‘this sentence’ have in common is they are both difficult, tricky, prickly, nasty, hellish, etc.

Unsurprisingly, ‘semantic theory’ of metaphor has its detractors. One objection comes from Aloyisus Martinich. He points out that the biggest flaw with each of our prior theories (besides from Davidson’s) is that they do not fit into any general theory of language or within general language use (Martinich, 1984.) A theory can only solve the issue of metaphor if it comports with a more general theory of language, an idea that has eluded us so far (except for Davidson.) Martinich’s position is more in line with Searle’s and Grice’s, albeit more radically in the ‘pragmatic’ camp than even they are. I would describe his position as more ‘contextual’ or context-based than Searle or Grice.

For a sentence to become metaphorical instead of literal, circumstances, surroundings and context have to be taken into account. He uses the example of a downtrodden farmer living in a poor country who says, ‘This is a fine country.’ It becomes much more obvious that the farmer is being nonliteral, because of the circumstances this statement is uttered in. Martinich distinguishes between what the speaker ‘makes-as-if-to-say’ and what the speaker actually implies. With our example, it is more obvious the devilish sentence is metaphorical because of the context. I am a student in a philosophy of language class, I have studied the issues of implicature and metaphor, I have chosen to answer a question relating to linguistic issues, etc. It becomes more obvious that ‘this sentence is the devil’s work’ is a metaphor due to the circumstances I find myself in.

Whilst I believe Martinich’s insight is a useful one, especially with our devilish sentence, Martinich’s theory does not answer everything, nor does his ‘makes-as-if-to-say/implies to say’ distinction work with every metaphor. One thorn in Martinich’s side is the issue of idioms. However, our sentence is not an idiom, so idioms will not be discussed here.

It is worth mentioning that our devilish sentence provokes other problems in philosophy of language. One issue is the fact that the sentence is self-referential, which muddies truth-condition and truth-value theories of language. Another issue is that our sentence arguably has no truth-value. However, this sentence is problematic first and foremost because it is a metaphor. Philosophers of language cannot agree upon what a metaphor is, what metaphors are supposed to say, nor how metaphor relates to truth. Also, metaphorical meaning can change over time, which complicates matters further (Martinich, 1984.) It seems Ludwig Wittgenstein had a point when he said:

“The limits of my language means the limits of my world.”

References

Ayer, A.J. (1946) Language, truth and logic. London: Penguin Books.

Ayoob, E. (2007) Black & Davidson on Metaphor. Macalester Journal of Philosophy, 16(1).

Black, M. (1979) How Metaphors Work: A Reply to Donald Davidson. Critical Inquiry, 6(1), pp. 131-143.

Cavell, S. (1969) Aesthetic Problems of Modern Philosophy.

Grice, H.P. (1989) Logic and Conversation, Lecture 2: Studies in the Way of Words. Cambridge, Massachusetts and London: Harvard University Press, pp. 22–40.

Hills, D. (2022) Metaphor, Stanford Encyclopedia of Philosophy. Stanford University. Available at: https://plato.stanford.edu/entries/metaphor/ (Accessed: January 6, 2023).

Lycan, W.G. (2018) Philosophy of language: a contemporary introduction. New York, NY: Routledge Taylor & Francis Group.

Martinich, A.P. (1984) A Theory for Metaphor. JLSE, 13(1), pp. 35–56. Available at:

https://doi.org/10.1515/jlse.1984.13.1.35.

Moran, R. (1989) Seeing and believing: Metaphor, Image, and Force. Critical Inquiry, 16(1), pp. 87–112. Available at: https://doi.org/10.1086/448527.

Moran, R. (1997) “Metaphor,” A Companion to the Philosophy of Language [Preprint]. Available at: https://doi.org/10.1002/9781118972090.ch16.

Our dictionaries: Oxford languages (2022) Our Dictionaries | Oxford Languages Dictionary. Available at: https://languages.oup.com/dictionaries/ (Accessed: January 3, 2023).

Searle, J.R. (1979) Metaphor. Expression and Meaning: Studies in the Theory of Speech Acts. Cambridge and New York: Cambridge University Press, pp. 76–116.