cmc.acts

CMC Act Taxonomy

Background:

The CMC Act Taxonomy is a classification scheme developed in 2005 by Susan C. Herring, Anupam Das, and Shashikant Penumarthy for coding "speech" acts in computer-mediated discourse. It is an amalgam and distillation of Bach and Harnish's (1979) classification of speech acts, which is based on Searle's (1976) classification, and Francis and Hunston's (1992) classification of conversational speech acts. The taxonomy was designed to apply both to genres of CMC that are closer to traditional writing, such as email and blog posts, and to more conversational genres such as real-time text chat and text message exchanges. Consisting of 18 act categories and two meta-act categories, the CMC Act Taxonomy makes more fine-grained distinctions than Searle's taxonomy, while being easier to apply reliably than the 33 acts identified by Francis and Hunston.

The original taxonomy comprised 16 acts and two meta-acts. In 2024, Susan Herring and Jing Ge-Stadnyk added two additional acts to the taxonomy: congratulate and behave.

The 18 acts and two meta-acts are defined and illustrated in this document. Note that Congratulate and Desire are umbrella acts that subsume multiple act functions.

How to Cite:

Herring, S. C., Das, A., & Penumarthy, S. (2005). CMC act taxonomy. http:/homes.luddy.indiana.edu/herring/cmc.acts.html

Herring, S. C., & Ge-Stadnyk, J. (2024). Emoji and illocutionarity: Acting on, and acting as, language. In: M. Gill, A. Malmivirta, & B. Warvik (Eds.), Structures in discourse: Studies in interaction, adaptability, and pragmatic functions (pp. 124-165). John Benjamins. Prepublication version: https://homes.luddy.indiana.edu/herring/festschrift.prepub.pdf

Related References:

Bach, K., & Harnish, R. M. (1979). Linguistic communication and speech acts. MIT Press.

Francis, G., & Hunston, S. (1992). Analysing everyday conversation. In M. Coulthard (Ed.), Advances in spoken discourse analysis (pp. 1-34). Routledge.

Searle, J. R. (1976). A classification of illocutionary acts. Language in Society, 5(1), 1-23.

CMC Act Taxonomy (Herring et al., 2005; Herring & Ge-Stadnyk, 2024)

CMC Act	Description	Textual Examples
Accept	Concur, agree, acquiesce, approve; acknowledge	Definitely; I agree I see your point.
Apologize	Humble oneself, self-deprecate	I'm sorry. Oops my fault :(
Behave	Perform a virtual action (that does not primarily function as another CMC act)	dances with joy *sips tea
Claim	Make a subjective assertion that is unverifiable in principle, e.g., because it describes a state of mind, feeling, or belief; assert, opine, speculate	I love you. That's the nicest thing you ever said.
Congratulate	Celebrate/praise an accomplishment Express confidence in future success, encourage Validate, support	Well done! Congratulations! You've got this! What you did was perfectly logical.
Desire (Irrealis)	Want, hope, wish Promise, predict, speculate Hypothetical Counterfactual	I would like to meet him. She will join us after class. If you're driving, I'll pay for gas. If she hadn't left, I would've stayed longer.
Direct	Command, demand; prohibit; permit; require; advise	Provide evidence for your claim. You have to register first. They should stop doing that.
Elaborate	Comment on, explain, or paraphrase a (usually one's own) previous utterance	(His position is untenable.) It will never work. (I did what you said.) I got a haircut.
Greet	Greeting and leave takings Formulaic inquiries about and wishes for well-being Formulaic responses to 1 and 2.	Hello; Bye; See you later How are you? I'm fine, thanks. And you? Happy birthday!
Inform	Provide "factual" information that is verifiable in principle, even if untrue; inform, state, report	Paris is the capital of France. My uncle just bought a jet. I've never been here before.
Inquire	Seek information, ask; make neutral or marked proposals	What are you guys eating? There's still time, right?
Invite	Invite, seek participation Suggest; solicit input Offer	Please join us. What if we did it this way? Let me help you.
Manage	Manage the discourse, e.g., organize, prompt, focus, open or close discussions	I have two thoughts about that. First, ... Second, ... That's my final word on the matter.
React	Show simple listenership, engagement (positive, negative, or neutral)	That's great/terrible! hahaha; lmao; wow; hmm; ugh; yeah
Reject	Disagree, dispute, challenge; insult	No way is that accurate. Gayyyy
Repair	Clarify or seek clarification; correct misunderstanding	Did you mean "school holiday"? Just kidding.
Request	Seek action politely, make direct or indirect request	Can you plz send pics? Would you mind sharing the link?
Thank	Appreciate, express gratitude	Thanks so much. It's very nice of you to offer.

Meta-Acts

In addition to the acts in Table 1, each utterance can be coded for two meta-act categories, each of which adds an additional layer of nuance to CMC act analysis:

Self-perspective vs. reported perspective:
- Self-perspective refers to content that comes from the utterance producer him or herself. Self-perspective is the default in speech act analysis; the acts produced typically represent the perspective of the utterance producer.
- Reported perspective refers to content that is explicitly or implicitly attributed to a different person than the utterance producer. A reported perspective can be expressed in a direct quotation or an indirect quotation, as in the following examples:
- The president said, "I want to end the war in the Middle East."
- The president said that he wants to end the war to the Middle East.
- In an utterance containing a reported act, the reported act (embedded clause: quote or indirect quote) itself is often the most important, new information in the utterance, and thus it is most relevant for act analysis. In that case, assign the act code to it, rather than to the main (embedding) clause. Thus the above examples would be coded as DESIRE [reported].
- However, if what the utterance producer intends to emphasize or present as new information is the fact that the president said something, and what he said was of secondary importance, the examples would be coded as INFORM [self].

Bona fide vs. non-bona fide communication:
- Bona fide communication is the default in speech act analysis; that is, the utterance producer is presumed to be producing acts sincerely and in good faith.
- In non-bona fide communication, the surface form of the utterance does not match the utterance producer's actual intended meaning; in that sense, it is insincere. Non-bona fide communication includes sarcasm, irony, joking, teasing, lies, and rhetorical questions, as in the examples below:
- The human failed the Turing test, haha.
- Have you ever heard of anything more ridiculous than that?
- Code non-bona fide utterances as if they were sincere, and add a meta-code to indicate that the utterance is non-bona fide. Thus the first example would be coded as INFORM [non-bona fide] and the second example as INQUIRE [non-bona fide].
- Non-bona fide utterances often function in their discourse context as indirect claims. The first example above could be paraphrased as, e.g., 'The human isn't very intelligent,' and the second example could be paraphrased as, e.g., 'Nothing is more ridiculous than that.' Thus it is also possible to code them both as CLAIM [bona fide]. The choice to assign the act and meta-act codes to the surface form of an utterance as opposed to its (presumed) underlying function depends on the analyst's preference. However, whichever strategy is adopted should be applied consistently.

Please direct any questions or feedback regarding this act taxonomy to herring @ iu.edu.

Last updated May 3, 2025