Worth noting that you can still successfully communicate while flouting one or more of the maxims. Sarcasm and irony show that tone and context play largely into understanding as well.
Totally! And this, the idea of conversational implicature, and how we draw meaning from people's variously following or flouting the maxims, is one of the reasons Grice was thinking along these lines to begin with.
He introduced the maxims in 1975 (or thereabouts) in support of his idea of a cooperative principle, which is to "make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged." [1]
He goes on, after discussing violations of the maxims leading to certain styles of communication including sarcasm, irony etc., to suggest that "to calculate a conversational implicature is to calculate what has to be supposed in order to preserve the supposition that the Cooperative Principle is being observed." [1]
It's pretty neat stuff. I like to think about communication among programs over networks in terms of the cooperative principle and conversational maxims. [2,3] HTTP/2 oversupplies information in response to a request, at first glance violating the maxim of quantity or relation; TCP acknowledgements literally mean "I heard you" but implicate "stop talking"; an HTTP request literally means "Please send me X", but implicates "I don't already have X, and I need it"; and so on.
For those who aren't familiar with formal semantics:
Conversational implicature is when the cooperative principle is used to deduce facts that were not explicitly stated.
For example, let's say I show up at your doorstep and I say "my car ran out of gas".
You reply "there a gas station just down the road"
I deduce from your statement:
- It is the nearest gas station
- The station is open and has gas to sell
- The station sells cans
- It is not too far to walk.
You did not say any of these things, but I know that they are true because if any of them were not true, the cooperative principle would have led you to say something different.
And if any of them turn out not to be true, I will feel like you deceived me, even though you never said it, because you implied it by way of Grice's maxims.
Fun fact: the Supreme Court of the United States says that deception by way of conversational implicature doesn't count as a lie and can't be used against you in a purjury case. It is legal to imply things that are not true on the witness stand. Lawyers have to word their questions very carefully as a result, and make all of the implicatures explicit.
Famous example:
Court - Have you ever had a Swiss bank account?
Bronston - The company had an account there for about six months, in Zurich.
Implicature (which later turned out to be false): I don't have any other swiss bank accounts in any other cities
The follow-up question was never asked, the matter was settled, and Bronston squeaked away from his bankruptcy proceedings with his swiss assets in tact.
This common "criticism" of Grice's Maxims isn't one, but a misunderstanding of the Maxims themselves.
In everyday conversation the Maxims do hold, that's the point. When they break down, you get "marked" effects — marked is a linguist's way of saying "unusual." This doesn't make the utterances wrong. Violating Grice's Maxims may cause miscommunication, or induce effects such as sarcasm, irony or humour, but also a slew of other, more mundane pragmatic effects.
Think of Grice's Maxims not as legislation you have to abide by. Think of them as laws of conversation (law in the scientific sense) which generally hold. I can subvert the law of gravity by jumping. But I need to invest energy, and I will cause an effect which will perturb the current reference frame in all manner of ways.
Similarly, I can lie, be facetious, tongue-in-cheek, etc. But that's not the default state of communication.
The Maxims aren't there to be obeyed. They're there to describe communication, and also describe non-cooperative communication when they're violated.
I think it's also worth noting that the maxims themselves are tongue-in-cheek. For example, "be brief" in the Maxim of Manner is given the unnecessarily prolix gloss "avoid unnecessary prolixity". I don't think Grice meant that particular list of maxims to be taken too seriously. They're just intended to illustrate how conversation is structured by the cooperative principle. In my view, this renders rather silly some of the earnest subsequent discussion of "which maxim" is violated by various example utterances.
They apply to communication. That's the fun thing. Ultimately, implicatures (a corollary of Grice's Maxims) are an information theoretic concept, though I haven't yet seen research that treats them as such. I would love to read about it, though.
Phil Resnik's dissertation ("A Class-Based Approach to Lexical Relationships ", https://ci.nii.ac.jp/naid/10010143379/) uses information theory to investigate why some direct objects are optional (e.g., "I ate [food]") while others are not (e.g., *"I brought [food]").
I extended his information theory to look at how Grice's Maxims would play in to modification of these direct objects ("The Semantics of Optionality", https://pqdtopen.proquest.com/doc/1658214586.html?FMT=ABS). In terms of Grice's Maxims + information theory, does modifying a required direct object ("marking it", as noted in a comment below) informationaly/entropically different from modifying an optional direct object? (This is further clouded by the concept of syntactic optionality being separable from semantic optionality.
Thanks for your pointers. I had actually read (parts of) Resnik's diss and further work, but apparently it had slipped my mind, or I just forgot about it.
Your work sounds really interesting, too. I hadn't heard of Firth's model yet; it sounds like it's in the spirit of distributional semantics. I hope I can find some time to give it a read!
One can probably shoehorn the notion, that one should take into account the knowledge, beliefs and attitude of one's audience, into these four maxims, though I think it could stand as one on its own.
10 comments
[ 2.7 ms ] story [ 33.7 ms ] threadHe introduced the maxims in 1975 (or thereabouts) in support of his idea of a cooperative principle, which is to "make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged." [1]
He goes on, after discussing violations of the maxims leading to certain styles of communication including sarcasm, irony etc., to suggest that "to calculate a conversational implicature is to calculate what has to be supposed in order to preserve the supposition that the Cooperative Principle is being observed." [1]
It's pretty neat stuff. I like to think about communication among programs over networks in terms of the cooperative principle and conversational maxims. [2,3] HTTP/2 oversupplies information in response to a request, at first glance violating the maxim of quantity or relation; TCP acknowledgements literally mean "I heard you" but implicate "stop talking"; an HTTP request literally means "Please send me X", but implicates "I don't already have X, and I need it"; and so on.
--
[1] Grice, H. Paul. "Logic and Conversation." In Syntax and Semantics 3: Speech Acts, 41–58. New York: Academic Press, 1975. PDF: http://lefft.xyz/psycholingAU16/readings/grice1975-logic-and...
[2] Garnock-Jones, Tony. "Conversational Concurrency." PhD dissertation. Northeastern University, 2017. http://syndicate-lang.org/tonyg-dissertation/html/#x_2_1_0_0...
[3] https://eighty-twenty.org/2018/02/01/lying-to-tcp
Conversational implicature is when the cooperative principle is used to deduce facts that were not explicitly stated.
For example, let's say I show up at your doorstep and I say "my car ran out of gas".
You reply "there a gas station just down the road"
I deduce from your statement:
- It is the nearest gas station
- The station is open and has gas to sell
- The station sells cans
- It is not too far to walk.
You did not say any of these things, but I know that they are true because if any of them were not true, the cooperative principle would have led you to say something different.
And if any of them turn out not to be true, I will feel like you deceived me, even though you never said it, because you implied it by way of Grice's maxims.
Fun fact: the Supreme Court of the United States says that deception by way of conversational implicature doesn't count as a lie and can't be used against you in a purjury case. It is legal to imply things that are not true on the witness stand. Lawyers have to word their questions very carefully as a result, and make all of the implicatures explicit.
Famous example:
Court - Have you ever had a Swiss bank account?
Bronston - The company had an account there for about six months, in Zurich.
Implicature (which later turned out to be false): I don't have any other swiss bank accounts in any other cities
The follow-up question was never asked, the matter was settled, and Bronston squeaked away from his bankruptcy proceedings with his swiss assets in tact.
In everyday conversation the Maxims do hold, that's the point. When they break down, you get "marked" effects — marked is a linguist's way of saying "unusual." This doesn't make the utterances wrong. Violating Grice's Maxims may cause miscommunication, or induce effects such as sarcasm, irony or humour, but also a slew of other, more mundane pragmatic effects.
Think of Grice's Maxims not as legislation you have to abide by. Think of them as laws of conversation (law in the scientific sense) which generally hold. I can subvert the law of gravity by jumping. But I need to invest energy, and I will cause an effect which will perturb the current reference frame in all manner of ways.
Similarly, I can lie, be facetious, tongue-in-cheek, etc. But that's not the default state of communication.
The Maxims aren't there to be obeyed. They're there to describe communication, and also describe non-cooperative communication when they're violated.
Note that these apply to programming languages as much as they do to natural language.
I extended his information theory to look at how Grice's Maxims would play in to modification of these direct objects ("The Semantics of Optionality", https://pqdtopen.proquest.com/doc/1658214586.html?FMT=ABS). In terms of Grice's Maxims + information theory, does modifying a required direct object ("marking it", as noted in a comment below) informationaly/entropically different from modifying an optional direct object? (This is further clouded by the concept of syntactic optionality being separable from semantic optionality.
Your work sounds really interesting, too. I hadn't heard of Firth's model yet; it sounds like it's in the spirit of distributional semantics. I hope I can find some time to give it a read!