---
title: RFC: The Next Level, or Discourse Categories
author: porpoiseless
---Discourse categories in Ithkuil '11 provide a wealth of interesting ideas. However, like the basic categories critiqued in RFC 0001, they suffer from non-orthogonality.
[2020-07-31 Fri]
WIP Draft.
- @porpoiseless
Conversation relates two logically distinct contexts:
- the content or subject matter of our our conversation, and
- the people involved in having the conversation; speaker and audience plus the performative, rhetorical or communicative actions performed between them.
I will call the former 'content level' and the latter 'discourse level'. While persons and events might coincide between these two levels, they are indeed very different and the grammar of Ithkuil's successor should reflect this fact.
As an extreme example of overlap between the contexts, someone might make a sandwich and narrate to themselves every step of the process: "I slice some bread. I spread peanut butter on one slice and jelly on another slice..."
In this case, the same individual is the subject spoken about at the content level, and both speaker and audience at the discourse level. The former is a description of something in the world, in this case the activity of the speaker as they assemble a sandwich, while the latter is simply a person talking to themselves. We can change the discourse context independently, for instance by having the speaker use the same series of utterances as a demonstration or tutorial of how to make a sandwich.
If the two contexts share persons or events, they can be often distinguished by time, modalization, and role. We often speak about our past or possible future selves. For example, when giving a command, "(You) spread peanut butter on the bread", the content is a possible future state where the addressee is performing the named action. The discourse level is an instruction or command.
Even if, as in our example of narrating the assembly of a sandwich, there is no difference in time and modalization, we can make the distinction that at the content level there is a sandwich maker, and at the discourse level there is a speaker and a listener. The fact that all three are the same person in no way diminishes our ability to differentiate what that person is doing in each role (making a sandwich, vs. speaking/listening).
The view I want to adopt here is that all language is performative: every utterance does something. Truth claims are only one kind of speech act that happens to have received preferential treatment at the hands of the logicians. But let us not be fooled: even a simple claim like, "The sky is blue" entails a new fact about the world. Namely it is now true that "I claim..." or "I judge that the sky is blue".
And these facts about what people claim to be true have important consequences. It is on their basis that we call someone honest or dishonest, well- or ill-informed, fair-minded or hypocritical, and so on. Because we are committed to truth, and there is a whole system of trust and consensus reality interwoven with it, it makes sense to mark truth claims as a distinct category.
Illocution can only be used to indicate speech acts made in the first person. I cannot, of instance, use Illocution to say, "Alice instructed Bob to make a sandwich." For speech acts made by 2nd and 3rd persons, the language will require verbs for "warn", "promise", "inform", "command", and so on. It is therefore sufficient to have only two values for Illocution: Assertive and Performative. The former marks truth claims and latter all other speech acts.
Consider these:
"I warn you"+Assertive
"I warn you"+Performative
The first is a claim that presumes that some other sign or speech act carries the warning, while the sentence merely asserts that warning takes place. The second really is the warning. And having only two values means an open class of verbs naming speech acts that may be either mentioned (in Assertive sentences) or used (in Performative sentences).
It may be helpful here to consider that a sentence like, "I warn you..." seems like a speech act, while "I warn them..." is a claim. "I warn them..." cannot be the same sentence the speaker used as a warning—otherwise it would refer to the recipient of the command in the 2nd person: "I warn you..."
The category of Expectation marks what kind of response the speaker desires from their audience. Broadly speaking, the three expectations work as generalized versions of the classic trio of sentence purposes in English (Declarative, Interrogative, and Imperative), and may also be thought of as corresponding to an expected "channel" in which the response will take place (as thought, speech, and action).
| Cognitive | Discursive | Motive | |
|---|---|---|---|
| intended response | thought | speech | action |
| sentence purpose | declarative | interrogative | imperative |
Expectation is not to be used for things like polite "question-commands" or rhetorical questions: rather, it should mark how the speaker actually intends their conversational partner to respond. The reason for this is that the verbs naming speech acts (which have replaced many of the Illocutions) can be used in tandem with Expectation and Illocution to produce both direct and indirect commands.
Speech acts like warnings and commands may be, but are not necessarily, truth claims. Consider the following utterances:
If you touch the stove, you will get burned.
I warn you not to touch the stove.
The first sentence seems like a truth claim, but it can be used as a warning. The second is explicitly a warning, but it may be only the mention of a warning previously given—in other words, it may be an assertion that a warning was issued.
| Assertive | Performative | |
|---|---|---|
| Cognitive | truth claim inviting only mental consideration | stipulation or declaration entailing cognitive change (c.f. let x = y) |
| Discursive | truth claim inviting rebuttal, comment, or explanation | request for comment |
| Motive | truth claim inviting listener to take action | warning, command, specific call to action |
Consider, for example, the following sentence in each of the 6 Illocution×Expectation combinations listed in the preceding table. This sentence does not name a speech act, so it demonstrates the power and flexibility of the new Illocution×Expectation scheme.
There is toilet paper on your shoe.
| Assertive | Performative | |
|---|---|---|
| Cognitive | I claim there is toilet paper on your shoe. | Consider yourself informed that there is toilet paper on your shoe. |
| Discursive | Comment on the toilet paper on your shoe. | Isn't that toilet paper on your shoe? |
| Motive | I claim there is toilet paper on your shoe, (do something about it). | Do something about the toilet paper on your shoe! |
Note the capacity here for making indirect, but clearly marked commands and questions. Honestly this system is so alien to me that I'm not sure if it will work, but it also manages to cover all the bases I can think of. Also note that you can turn those into explicit commands and questions simply by adding the right verb, so instead of "There is toilet paper on your shoe"+Performative+Motive, you could easily say, "You clean your shoe"+Performative+Motive to make explicit what you want your listener to do.
JEFF: Yeah, we really need wifi!
FRANKIE: Okay, let's not get carried away. We /need/ oxygen.
ELROY: We /have/ oxygen, we /need/ wifi!
— Queer Studies & Advanced Waxing (Community S6E4)
There are 3 sorts of modality in language, of which only one is content level. Epistemic and Deontic modality refer respectively to the speaker's certainty about the content, and the normative/desiderative force embodied by the utterance. Alethic modality relates to the modality of the content itself. Since alethic modality is not a discourse level category, it belongs elsewhere in the scope order.
Consider a sentence like, "They can't have children", which might be interpreted as any of the three kinds of modality enumerated above.
| Modality | "They can't have children." |
|---|---|
| Alethic | They (biologically) aren't capable of having children. |
| Epistemic | Evidence suggests they don't have children. |
| Deontic | They ought not/shouldn't have children. |
For each of these kinds of modality, we need a minimum of 3 modal operators: actuality, possibility, and necessity. We can add more specific modalities as needed: however, other grammatical machinery can express affective distinctions like, "I fear/hope they have children".
The Epistemic and Deontic modalities operate in tandem with other discourse level categories.
Epistemic modality is associated with Evidentials—it expresses the speaker's commitment to the knowledge embodied in their claims: whether they judge that something is necessary, possible, or actually the case.
Deontic modality works with Illocution and Expectation to convey the speaker's needs and desires.
Intersection between modality types and other categories
| Alethic | Epistemic | Deontic | |
|---|---|---|---|
| actuality | being, reality | thinking | is |
| possibility | ability, capacity, capability | doubt | should/may |
| necessity | need, requirement | certainty | must |
| Related categories: | [content categories] | Evidential | Illocution, Expectation |
Example sentence:
You must practice.
| Alethic | You need/require practice. [c.f. "You require air."] |
| Epistemic | (I am certain) you practice. |
| Deontic | You must practice. [obligation/duty to practice] |
Because there can be an epistemology of ethical judgments as well as an ethics of epistemological states, there arises a question of the scoping of the modalities. Now, there are three kinds of modality, each with three values, meaning that there are nine of these modalizations. However we may also suppose that we can stack them.
- A huge affix with 162 values for all combinations of scoping order;
- Implement an is-ought distinction: epistemic modalities only modify Assertive claims, whereas deontic modalities only modify Performative speech acts;
- Something like the Organizational Affix (i.e. stackable, possibly with an adjunct);
The 162 values of option 1 seem unacceptably high, even for so important and
expressive a category as modality. Option 2 is appealing to me, but it forces
one to rigidly distinguish between Assertions of what is and Performatives
expressing what ought to be. So I propose a default scoping that could then be
modified/evaded using some sort of adjunct or particle, much like the Ca
Organizational Affix with stacking.
Let's think about what an utterance does vis a vis the discourse level categories and modality. Illocution comes first, since whenever I utter something it is immediately the case that I have uttered it it, whatever its modality or content.
I am certain I should want to fly.
How is this rendered? Well first let's say the speaker is not making a truth claim. (I mean, they could be, but it's kind of pointless to make truth claims about one's internal states.) So the Illocution is Performative.
I must be certain that our prisoner committed the crime.
The speaker should be able to vary the scoping among these, but only in certain parameters. For instance, Illocution must have the outermost scope.
"I can say, 'The Moon is a moldy wheel of cheese'"
Now the content of this utterance is my capacity to make a certain statement: I do not actually make the statement, but I predicate of myself that it is (alethically) possible for me to do so. Illocution is for actual speech acts (of either the Assertive or Performative variety). So the preceding example requires a different wording.
"The Moon may be a moldy wheel of cheese."
I mean not that the moon has this capacity (alethically), but rather that my epistemic state does not preclude that state of affairs.
| Vote | Name |
|---|---|
| +1 | @porpoiseless |
| +1 | @uakci |
Draft.
A list of persons to CC about this RFC.