Language is full of ambiguity. This fuzziness is often cited as a sign of imperfection, leading some to try to develop more precise languages of their own. But ambiguity actually serves a purpose, and is frequently exploited in human interactions.
Part of recognizing the utility of ambiguity requires understanding that language is more than just a tool for conveying information between agents. Rather, language is a form of social action (Austin, 1962; Levinson, 2012). We use language to make requests, make jokes, insult or compliment others, express frustration, and more. Put another way, language is one of the many tools we recruit to navigate social interactions.
Language: A Tool for Negotiation
Often, this involves tacitly negotiating the nature of one’s relationship to the listener. Given that social relationships are often uncertain, and that violating the terms of a relationship can result in significant repercussions, we see that ambiguity could be a helpful tool for preserving these relationships or one’s position in a social hierarchy.
The sociologist Erving Goffman called this saving face (Goffman, 1978). Face refers to one’s position or “dignity” in some social context. Goffman notes that people go to great lengths to save face, and will even intervene on behalf of others; for example, if someone trips at a party, stumbles over their words, or even passes gas, others will sometimes pretend not to have noticed. Of course, sometimes other people do point it out, but this is also a form of social action and often serves to subject the original party to embarrassment.
Later, linguists Penelope Brown and Stephen Levinson applied this concept to politeness theory (Brown & Levinson, 1987). They argue that speakers use different politeness strategies to save face, depending on the face-threatening situations that could arise. For example, negative politeness strategies are used to avoid imposing on the listener. Perhaps the best-known are indirect requests, in which a speaker makes a request, but softens the blow by formatting the request as a question or a statement of fact (e.g. “I was wondering if you’d be able to give me a ride to the airport?”, or even “My ride to the airport canceled on me…”). This gives the listener the opportunity to “opt out” without an awkward situation ensuing.
Ambiguity as Plausible Deniability
Pinker et al (2008) describe a more formal model of the utility of indirectness, or ambiguity, in language. The Rational Briber model is rooted in the idea that indirectness allows the speaker (and the listener) plausible deniability in how their utterance is interpreted. I discussed this model in a previous post, but I’ll go into a bit more detail here.
Navigating ambiguous relationships can be difficult. Whether it’s bribing a possibly dishonest cop out of a traffic ticket, or inviting a date up for a sexual encounter, the misalignment of intentions and interpretations can result in awkwardness, social blowback, or in the former case, arrest. Pinker et al (2008) illustrate this difficulty via a 2×2 game-theoretic model:
Bribing could either result in freedom or arrest; not bribing will certainly result in a traffic ticket. Neither of these are optimal strategies. But a speaker can also imply a bribe by speaking indirectly, e.g. “maybe we can just settle this here?” Note that the outcomes of an implied bribe are either freedom or a traffic ticket. The expected value, then, is higher than either of the direct strategies:
This explains why indirect requests are useful as bribes, but Pinker et al (2008) argue that the same logic applies to nonlegal situations. Ambiguous language preserves social relationships via the preservation of face; just as we might ask a police officer, “maybe we could just settle this here?”, we might ask a date, “Would you like to come up to see my etchings / listen to some records / Netflix and chill?”
Of course, the degree of plausible deniability depends on the indirectness of the statement. Indirect bribes are often still detectable as bribes – and indeed, they should be, if they’re to be successful – but as Pinker points out, any reduction in the “certainty” of a statement’s intended meaning will assist in saving face. That is, any amount of indirectness will have a greater chance of saving face than direct speech.
The Takeaway (Or: “Why we should care”)
This might all seem like academic hand-wringing, but it has serious implications for how we think about law, politics, and many other facets of human interaction. Consider the following scene from HBO’s The Wire. (Note: If you haven’t seen The Wire and intend to, note that the clip and text below contain spoilers for season 1.)
In this clip, drug kingpin Stringer Bell makes a request to corner-boy Bodie; Stringer wants Bodie to kill Wallace, another corner-boy and potential snitch. Note, however, that at no point in the interaction does Stringer, the master of indirect speech, explicitly say this.
Stringer first alludes to Wallace indirectly, exploiting the common ground between them:
STRINGER: What’s up with that boy?
Once it’s established that Wallace is “back in the game”, Stringer refers to Wallace’s supposedly weak constitution (and elicits a confirmation from Bodie):
STRINGER: I heard he damn near shit his pants when he heard what happened to Omar’s bitch.
BODIE: He just ain’t built for this, y’know? His heart pump Kool-Aid.
Stringer now makes something akin to a pre-request by inquiring about Bodie’s abilities and strengths:
STRINGER: What about you? You built for this shit?
BODIE: No doubt.
Finally, the request – or as close as we get to a request – is made:
STRINGER: You ready to put the work in?
STRINGER: You got heat?
(BODIE gestures to his beltline, implying that he possesses a gun.)
STRINGER: All right, soldier.
It’s clear to both Stringer and Bodie that a request has been made, and what the nature of that request is. And it’s also clear to us, as viewers – knowing what we know about the context of this interaction – that Stringer is asking Bodie to kill Wallace. But does Stringer say anything we can definitively point to as a “command” to murder somebody? If a prosecutor were to bring this in as evidence, how would it fly in court? Maybe it’s obvious enough to be “beyond a reasonable doubt” to a jury – “you got heat?” is a pretty direct reference to having a gun – but maybe it’s not; much of what’s conveyed in the interaction is beyond the scope of what’s actually said. Thus, for a prosecutor to argue their case successfully, they’d need to draw upon an understanding of how intentions are conveyed indirectly in interaction. The Wire is full of indirect speech, whether it’s in the streets or mayoral offices, and whether it’s via indirect requests (as Stringer makes to Bodie) or ambiguous reference (e.g., using the word “product” to refer to drugs).
If that’s not convincing, consider a more recent example – this one happened in reality, not on HBO. In reference to the investigation of General Flynn, President Trump told former FBI director James Comey: “I hope you can let this go”. To many, this sounds like a bribe (or a veiled threat, another form of indirect speech). But many Trump supporters were comically steadfast in their literal interpretation of the statement as an actual statement of “hope”, as opposed to an implied command to Comey.
In both cases, indirectness gives the speaker some amount of plausible deniability, even though the intention seems pretty clear to many people. The lesson here is not that indirect speech is somehow bad; it’s an essential and useful part of human communication. But there is a sinister and slippery side to indirectness, and it rears its head when ambiguity is employed by those in power, who are particularly well-positioned to defend (and define) this plausible deniability.
The takeaway, then, should be this: when someone – particularly someone in power, or someone supporting them – defends themselves against accusations by invoking the “plausible deniability” argument, don’t just take them at their word. Maybe they really did mean what they said, but maybe they meant something else as well. Cite the relevant research, and build a case defending your interpretation.
As to how indirect speech should be handled under the law, the solution remains unclear. The nature of indirectness is that it’s highly nuanced, which means it’s very difficult to detect and enforce in a systematic way. This is almost certainly a topic for a separate blog post, but for now, I’ll simply say that lawmakers need to start thinking about the role of nuance in language.
Borges, J. L. (1988). The Analytical Language of John Wilkins. Other Inquisitions 1937-1952, (1910), 101–105.
Eco, U. (1995). The search for the perfect language. Wiley-Blackwell.
Austin, J. L. (1975). How to do things with words. Oxford university press.
Levinson, S. C. (2012). Action formation and ascription. The Handbook of Conversation Analysis, 101–130.
Goffman, E. (1978). The presentation of self in everyday life (p. 56). London: Harmondsworth.
Pinker, S., Nowak, M. A., & Lee, J. J. (2008). The logic of indirect speech. Proceedings of the National Academy of Sciences of the United States of America, 105(3), 833–838. http://doi.org/10.1073/pnas.0707192105
Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage (Vol. 4). Cambridge university press.
Curl, T. S., & Drew, P. (2008). Contingency and action: A comparison of two forms of requesting. Research on language and social interaction, 41(2), 129-153.
 These attempts are well-documented in Eco’s The Search for the Perfect Language (1995); one particularly well-known attempt is discussed by Borges (1988).
 These functions could perhaps also be analyzed through the lens of information transfer, but their purpose – as discussed above – seems to be more than just about transferring information to the listener’s mind.
 The concept of “face” has been very well-studied in many different contexts, from linguistics to political science. The Wikipedia article has a very good breakdown of the study of face by country and by field.
 The question of how the speaker formats this request is of deep interest to linguists. For example: why do speakers say “could you open the window?” in some situations, but simply complain, “It’s so hot in here!” in others? And how do these formatting choices change across speakers? One theory, supported by extensive analysis of actual conversations, is that the way in which speakers make a request depends on how entitled they feel to make a request (Curl & Drew, 2008). The authors found that a more entitled speaker is more likely to use the form, “Could you X?”, whereas a less entitled speaker is more likely to use the form, “I was wondering if you could X“. I’m currently running an experiment testing out this very theory — once the data is in, I’ll have something to report on this blog.
 This means that the main choice a speaker faces is how indirect to be. They want the meaning to be plausibly deniable, but also interpretable in the first place.
 Technically, at this point, Stringer is second-in-command to Avon Barksdale, but I’m eliding the difference for the sake of brevity.