a metaphor always holds out the promise of more—more detail, more mappings, more associations—if we would only deepen our search. Finally, a paraphrase does not propose the same conceptual pact to an audience, and so is unable to serve the same communicative role as the metaphor it replaces.
One important communicative role that is not captured by an explicit paraphrase is plausible deniability, wherein a speaker uses metaphor to communicate a tendentious claim, and subsequently appeals to the ambiguity of metaphor to deflect criticism if this claim is ill-received by an audience. Consider the case of Republican presidential candidate Donald Trump, who was challenged on his alleged record of sexism by FOX-News reporter Megyn Kelly during the first televised GOP debate in 2015. Trump rebutted the charge and complained of mistreatment at the hands of the debate moderators, later claiming in an interview that “you could see there was blood coming out of her [Kelly’s] eyes, blood coming out of her wherever.” Many commentators in the media took Trump’s comments to be a potent mixture of metaphor, metonymy, hyperbole, and bombast. Although Trump painted a picture of a woman possessed by a demon, bleeding uncontrollably from every orifice, and thus lacking control over her own physical and mental state, his under-specified reference to a woman bleeding from “her wherever” was widely taken to be a metonymic reference to menstruation. Trump, it was claimed, was simply giving new form to the old charge that a woman who behaves aggressively to a man must be in the irrational grip of premenstrual tension. In a daring move, Trump was seemingly fighting charges of sexism with more sexism, yet when pressed on the matter, he defended himself with this tweet: “Re Megyn Kelly quote: ‘you could see there was blood coming out of her eyes, blood coming out of her wherever’ (NOSE). Just got on w/thought.” He elaborated further in a subsequent statement, claiming that “Only a deviant would think anything else.” Although we can debate Trump’s actual intentions for his blood metaphor, it does seem that its non-literal under-specification was an essential part of its communicative function, allowing Trump to push a provocative idea into the public sphere while denying that this idea was ever part of his intended message. Another issue with literal paraphrases then is that they sometimes render in plain speech an opinion that was never intended to be plainly expressed or openly interrogated in the first place.
Yet, for all this, there is little to be gained from being a radical skeptic on the matter of literal restatement. Paraphrases may be necessarily imperfect, but they possess a practical utility for researchers and speakers alike. For, in establishing a conceptual pact, a metaphorical utterance may prompt a listener to produce a paraphrase in response, as the listener’s own contribution to the construction of a joint meaning space. By doing so, the listener may propose a particular reading of a metaphor that enriches and elaborates upon the speaker’s own viewpoint. In effect, this exchange of paraphrases serves to align the understanding of both speaker and listener, and allows each to arrive at a deeper appreciation of a metaphor. For similar reasons, and for others that are distinctly computational, the ability to produce a literal paraphrase of a figurative statement is also of some practical value in a computational setting. Not only does paraphrasing allow a computer to explain its interpretation of a metaphor, in a way that allows users to detect a failure of interpretation, it also allows a computer to directly assign a semantic representation to a metaphor that is consistent with the system’s own axioms, so that then it can reason safely about the metaphor’s contents. Indeed, in a wide range of NLP systems—such as for sentiment analysis, information retrieval and extraction, question answering, and summarization—literal paraphrasing can support a perfectly adequate approach to metaphor meaning that is just as deep as the system’s treatment of literal texts and literal meanings.
As shown in our earlier discussion of the metaphor “wishes are children,” we need not limit ourselves to purely literal statements when paraphrasing a metaphor, and so, perhaps, neither should our NLP systems. The value of any paraphrase lies in its ability to explain a challenging turn of phrase using more conventional and less taxing language. This is so whether the text to be paraphrased uses the non-literal language of poetry or the literal language of legalistic jargon (as found, say, in a legal contract or a patent application). Indeed, the latter demonstrates that the most natural and most useful paraphrase is not always the most literal, and less challenging forms of figurative expression, such as a familiar conventional metaphor, may better convey the intended meaning of a metaphor. For example, when paraphrasing a metaphorical description of a wine as “muscular,” it is more helpful than not to exploit the conventional if metaphorical view of wines as possessing a “body.” Or consider the metaphor “to pick up a touch of the flu,” which in turn contains a pair of tactile metaphors, “to pick up” and “touch.” The first of these is perhaps more naturally paraphrased using the conventional metaphor “to catch,” which is very commonly used with infections, rather than the strictly literal and somewhat dramatic “become infected by” or “to suffer from.” The second is perhaps best paraphrased using the modifier “mild” for “flu,” even though the idea of mildness is likely a metaphor itself in this context. Indeed, it is hard to find a concise literal rendering for the tactile metaphor “touch:” a touch suggests a glancing contact, which implies a contact of short duration that lacks force. So a “touch of flu” implies a short-lived infection that does not manifest the worst symptoms of the ailment.
An acceptance of the occasional conventional metaphor in a supposedly literal paraphrase is especially useful in the context of statistical models of metaphor, for it is in the nature of conventional metaphor to be used habitually and to assume the status of normative, literal language over time, to the extent that the two are hard to separate on statistical grounds alone. Indeed, much of what we consider literal was once a newly minted figurative form that, as the philosopher Nietzsche put it, has become the base metal of literal coinage through oft-repeated usage. An NLP system that aims to remove all metaphor from language will find much—perhaps too much—to remove, and will be left with precious little with which to compose its paraphrases. In contrast to the divide between the figurative and the literal, the unconventional/conventional divide has an observable reality in large text corpora that is conducive to statistical analysis (e.g., Shutova et al. [2012], Bollegala and Shutova [2013]). It is a quantifiable reality that allows a paraphrasing system to learn to do as human paraphrasers do: to generate helpful conventional paraphrases from unconventional metaphors. If such paraphrases contain conventional metaphors, as they are very likely to do, this need not pose a serious challenge to the semantic representations of an NLP system. Conventional metaphors are part of the furniture of language, and can easily be accommodated—with some design forethought—in a system’s core representations. For instance, both the MIDAS system of Martin [1990] and the ATT-META system of Barnden and Lee [2002], and Barnden [2006] put a semantic representation of conventional metaphors at the heart of their metaphor interpretation systems. MIDAS and ATT-META each show, through their different uses of schematic structures, how an NLP system can support a wider range of inferences, and, thus, a fuller interpretation of a metaphor, by reserving a place for the metaphor in its semantic and conceptual representation of an utterance.
2.4 METAPHOR AND SIMILE
Since the worldview communicated by a metaphor is often shaped by a speaker’s perception of a situation or an interlocutor, simile—the figurative device most often called upon to simultaneously communicate and explain our perceptions—can be of particular use when paraphrasing a metaphorical worldview. Consider the metaphor “marriage is slavery:” while there is little insight to be gained by paraphrasing this metaphor with the simile “marriage is like slavery,” there is much to be gained from the paraphrase “this marriage is like the relationship between a slaver and his slave: you act like you own me, and treat me like your slave.” The use of similes imbues this paraphrase with three interesting qualities. First, the similes bring a desirable emotional distance to the description of an undesirable relationship, for the assertion “you act like you own me” conveys a very different affect than “you own me” or “you are my owner.” Second, the paraphrasing similes are not freighted with the same semantic tension as the original
