Do you always mean what you say?

One of the most intriguing things about language is the way that meaning is tied to context. Take this sentence for example:

“Everybody was there yesterday”

Which day is ‘yesterday’? The word ‘yesterday’ has no fixed meaning and needs a context. It refers to a different day each day that you say it. Same goes for the word ‘there’. We need a context to know where ‘there’ is. How about “everybody”? Does that literally mean every individual in the entire world? Of course not. It means something like “every person within a contextually relevant group of people”. Since this sentence has no context, you probably had to invent one, maybe by imagining a room full of people you know.

The branch of linguistics that studies how context interacts with meaning is called ‘pragmatics’, and in this post I want to introduce you to one of my favourite topics in pragmatics: implicatures.

Let’s start with an example of how the same sentence can mean different things in different contexts.

Context 1

Alice wants handwritten invitations for a party, so she’s asked her friend Bob for advice.
Alice: Any suggestions for who could do the invitations?
Bob: “Well, Paul has really nice handwriting”
Conclusion – Paul would be a good candidate for this job

Context 2

Alice owns a factory and is looking to hire a forklift operator. Paul has applied for the job, and Alice calls his former boss Bob for a recommendation.
Alice: What are some of Paul’s strengths?
Bob: “Well, Paul has really nice handwriting”
Conclusion – Paul would be a poor candidate for this job

Bob gives the same answer each time, but we draw different conclusions. This points to an important distinction we should make about how communication works: what someone says is sometimes not what they mean.

The interesting point here is that this only happens sometimes. In the context where Alice wanted invitations done up, what Bob said and what he meant were basically the same thing. Alice wouldn’t go searching for extra meaning in his answer. In the context of the forklift operator job, Bob said “Paul has nice handwriting” but Bob meant “Paul would probably not be a good forklift operator”, and that would be exactly what Alice would understand he meant. What makes this second context special that we get a difference in saying and meaning?

The philosopher Paul Grice proposed a solution to this puzzle back in the 1970s. His main insight is that conversations are cooperative ventures. In general, people make an effort to be understood and say useful and meaningful things in a conversation, and listeners are pretty generous in trying to figure out what the speaker is saying.

Now, please don’t misunderstand this as meaning that everyone is an eloquent conversationalist. It’s a much more general idea than that. When I used the word “everybody” in a sentence without context back at the beginning of this post, you (the reader) didn’t stubbornly refuse to understand me. You went the extra mile to make up a generic context to bring some sense to what you were reading. Your willingness to try and figure out a meaning is an example of  being “cooperative” in this sense.

Grice’s idea is that we expect cooperative behaviour as a the norm, so when people are obviously and intentionally uncooperative, we assume that they must be trying to tell us something other than just the literal content of their words. When Alice asked Bob about Paul’s suitability as a forklift operator, and he responded by talking about handwriting, he wasn’t being very cooperative. It’s uncooperative because it’s an off-topic response that provides no useful information. It doesn’t even get into the subject of operating forklifts, which what Alice wants to know about. Because Bob is being so obviously uncooperative, we infer that he must be trying to tell Alice something else, namely, that Paul is unsuitable for the job.

Here’s a new technical term for you: the extra information that goes beyond what was literally said is called an “implicature” and the verb is “to implicate”. Bob in this case implicated that Paul is not a good job candidate. The term “implicature” is almost never seen outside of linguistics and philosophy, and it’s more common to hear the term “imply”. However, “imply” has a different technical meaning. We would say that Sentence A implies Sentence B if whenever A is true B is also true. For example, “Dave was born in Toronto” implies “Dave was born in Canada”. The sentence “Paul has nice handwriting” does not imply “Paul is a bad forklift operator” (it only implicates that) because there’s no logical connection between the statements. It could be the case that Paul has both nice handwriting and is also an excellent forklift operator.

As listeners, how do we arrive at implicatures? The general reasoning process goes like this:

1. People are usually cooperative
2. The person I’m talking to is not being cooperative
3. Therefore they must be trying to tell me something else

Grice proposed there are four ways in which this cooperative principle played out, which he called the “Maxims” of conversation. This gives a slightly more fine-grained way of discussing how people are (or are not) cooperative.  For each example, there’s a short conversation where Alice asks a question and Bob answers with statement that implicates something. I’ll go over how the reasoning works from the perspective of Alice, who is trying to understand what Bob told her.


“Say what you know to be true”

Alice: Want to go see Letters to Juliet?
Bob: Sure, then later we can poke sharp stick into my face.
Implicature: Bob does not want to see the movie.

Reasoning: Normally people tell the truth. Bob obviously doesn’t want sharp sticks in his face for real, so why is he telling me this falsehood? He must actually be trying to tell me something else. Since he is proposing something really unpleasant, it probably means he doesn’t want to go to the movie


“Say as much as you can”

Alice: Do you know when the final exam is?
Bob: It’s in May.
Implicature: Bob doesn’t know which day it is.
Reasoning: If Bob knew which day it was, he would have told me because that would have been the most useful answer. He didn’t tell me, so he probably doesn’t know (or possibly doesn’t want me to know)


“Stay on topic”

Alice: Want to come to the party tonight?
Bob: My mother is flying into town this afternoon.
Implicature: Bob can’t come.
Reasoning: I told Bob about a party, and he answered back with an unrelated statement about his mother. If he could come to the party, he would have just stayed on track and answered my question, so he must be trying to tell me that he can’t come.


“Say things in a brief and ordinary way”

Alice: Did Monica sing ‘O Canada’ last night?
Bob: Monica produced a series of notes resembling the tune to ‘O Canada’
Implicature: Monica didn’t sing very well
Reasoning: If Monica had sung ‘O Canada’, Bob would have just said so. He answered in a weird way so there was probably something weird about the way that Monica sang.

A key point here is that you don’t get an implicature just because you broke a Maxim. It has to be obvious that you were doing it. Your listener has to know that you are intending to implicate something. Consider this exchange:

Alice: Who is the President of Finland?
Bob: Mauno Koivisto

Bob’s answer is false, and so it violates the Maxim of Quality (say what is true). However, Mauno Koivisto is a Finnish name, and, in fact, it is the name of a former President of Finland. In other words, it’s a plausible answer to the question. It could be that Bob is just not up on his Finnish politics, and he is just giving his best answer. There isn’t necessarily  any reason for Alice to assume something else. Now consider this:

Alice: Who is the President of Finland?
Bob: Santa Claus

This time, Bob is clearly implicating that he doesn’t know who the President is. Alice can reason that if Bob knew the answer, he would have given it, but he gave an obviously false answer, so he must be trying to say that he doesn’t know the answer. Note, though, that implicating something with a Quality violation like this is very different from lying.  It would be weird if Alice took Bob’s answer at face value:

Alice: Who is the President of Finland?
Bob: Santa Claus
Alice: You liar!

Once you are aware of the Maxims, you’ll see their use all over the place. In many cases where people are trying to avoid saying something uncomfortable, they will break Maxims and hope the implication is enough. Let’s say that Bob was supposed to be watching Alice’s pets for the afternoon, but he’s really messed up, and the cat ate the fish. Alice comes home and this conversation ensues:

Alice: Where’s the fish?
Bob: Let’s just say the cat had a good afternoon.
Alice: Get out of my apartment.

Bob’s answer is a clear violation of the Maxim of Relevance, because it doesn’t answer the question at all. It’s something of a Manner violation as well, since it is an oddly phrased answer. Alice can probably reason from here that something bad has happened, especially since Bob chose to mention “the cat” in his answer.

Another example is the “it’s interesting” response that people use when they are trying to avoid hurting someone’s feelings.

Alice: What do you think of my haircut?
Bob: It’s interesting.
Alice: You don’t like it.

Bob is violating the Manner of Relevance, and maybe there’s a Quantity violation in here too. It’s a very short answer that said very little about the haircut. If Bob liked it, he either would have said ‘yes’ or said something specific about it.

Implicatures can plausibly be denied or cancelled without contradicting yourself. They were never actually uttered or asserted, so you aren’t liable for them in the same way. Implicatures are re-created entirely “on the other side”, in the mind of the listener. Bob can try and back-peddle here, by focusing on what he said instead of what he meant, and respond “No, I never said I didn’t like it! I said it was interesting”. (He can *try*, he probably won’t succeed in this particular case.)

Implicatures don’t just happen with answers to questions. The questions themselves may implicate something.

Alice: Do you have a watch?
Bob: It’s quarter to eleven.

Alice didn’t really want to know if Bob had a watch. She actually wanted to know what time it was. Bob understood the implicature, and he responded directly to it. He ignored what she said entirely, which is what Alice wanted him to do anyway. In fact, Bob looks like a bit of jerk if he does answer the literal question that was asked.

Alice: Do you have a watch?
Bob: Yes, it’s a Timex.

Make a game of this and go implicature hunting today. In the next few conversations you have, take a moment to reflect at some point about what’s actually happening. Are you asking direct questions, or hoping your listener will get the right implicature? Think about the answers that people are giving you. How much of what you understood was actually said, and how much was an implication? Do you notice that some people are more direct, and others prefer to do a lot implicating?  Some knowledge of pragmatics can bring you a new perspective on communication.


Filed under Linguistics

7 responses to “Do you always mean what you say?

  1. richardmendacks

    Reblogged this on Richard Mendacks.


  2. My immediate thought to the first few lines of this post was: do you do (English) cryptic crosswords? English, because they are rather different to, say, American puzzles.

    The setter’s idea is something like, “I must say what I mean, but need not mean what I say’.

    For example, the word ‘flower’. You might think of a rose or a tulip, or some other flower on a plant. But the crossword setter is using a ghastly pun; something that ‘flows’ is a ‘flower’, such as a stream or river. I’m not sure where this fits into your typology; it’s just amusing, an exercise in mental gymnastics.


    • I’ve tried North American cryptic crosswords, and to be honest I just can’t figure them out. I give them a try every few years, and I then I end up quitting without solving any. I didn’t know that English-style puzzles were different, so maybe I’ll have a look around for some.


      • Cryptic crosswords, such as those in The Times, until a few years ago really required an education in a typical English ‘public’ boarding school. A lot of the references were to bits of literature that most people simply didn’t know. They are more egalitarian today, but still need a good knowledge of obscure and archaic words. You need to know that a ‘flag’ might mean an ‘iris’ (the plant); you needed to know the Book of Jeremiah to understand the ‘Flower in Jeremiah’ refers to a flower in bloom hidden in the full name of that book, ‘The Lamentations of Jeremiah’ which conceals ‘amenta’. There’s more of a trend today to clues which use codewords to conceal letters; ‘even’ means the even letters in a word (‘odd’ similarly), ‘supported’ in a down clue means letters at the end of word, and so on.You also have to understand the mind of the setter to see how he thinks (and almost all setters are men). But the initial statement, saying what you mean but not meaning what you say still holds. Some papers have blogs which describe how clues can be deconstructed and understood.


  3. Your blog is very interesting. My student days are long gone, but the valuable content in your posts keeps my love of language alive.

    Liked by 1 person

  4. eso es verdadera lectura


  5. This actually answers something I had wondered. In the movie Independence Day, in what passes for the lab below Area 51, when the general barks, “Is this glass bulletproof?” It’s obvious he wants his men to shoot, though a subordinate does answer “No, Sir!” before ordering the privates to fire. I always wondered what that process was called.

    Liked by 1 person

Tell the world what you think!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s