How much can you say in a tweet? An approach to political argumentation on Twitter

The paper investigates whether and how political argument is possible within Twitter’s 280-character constraint and whether its nature varies by the news outlet to which it is addressed. The authors pose foundational theoretical questions for studying argument on Twitter: Is argument attempted on the platform? What constitutes argument and its content? If argumentative content is understood as speech acts, what kinds can comprise an argument under strict character limits? The study aims to establish a theoretical framework for subsequent quantitative content analysis. Focusing on illustrative cases—responses to Fox News and MSNBC tweets about the October 2018 arrest of Cesar Sayoc—the authors explore how single speech acts, including non-assertives, can convey enthymematic arguments and how humor functions as a rhetorical device. They also examine whether responses differ by audience orientation toward conservative (Fox) versus progressive (MSNBC) outlets.

A broad Twitter literature spans politics, social movements, sentiment analysis, network structure, and platform features (e.g., Karam et al., 2020; Zimbra et al., 2018). Quantitative work often examines structural features (network analysis, retweeting practices, hashtags and their linguistic/cultural functions). Qualitative work explores lexical, interactive, and ideational features, identity formation, and sociolinguistic argumentation. Speech act research on Twitter includes both computational detection/classification (e.g., email-to-speech-act algorithms; customer service corpora; celebrity speech acts) and theoretically oriented distinctions among assertives, directives, expressives, etc. (Zhang et al., 2012). The paper draws on Speech Act Theory (Austin, Searle) and the concept of enthymemes in popular discourse (Aristotle), arguing that much argument is implicit, permitting single speech acts to carry argumentative weight. Prior work identifies humor and irreverence as core elements of political Twitter discourse (Highfield, 2016; Davis et al., 2018), with humor serving to discredit opponents, build political identity, and bolster support; sarcasm, irony, and benign violation theory help explain its rhetorical force.

Design: An illustrative qualitative discourse analysis to outline a framework for future quantitative study of argument on Twitter using speech acts as the units of analysis. Data: From a larger dataset of ~64,000 tweets directed at 36 news sites around the 2018 US midterm elections, the authors focus on responses to Fox News and MSNBC about the October 26, 2018 arrest of Cesar Sayoc (mailing pipe bombs to Trump critics). Fox discussion concentrated across 10 tweets between 16:50:42 and 18:39:18; MSNBC discussion across 7 tweets between 15:15:15 and 15:33:39. Sampling for tables: First 100 response tweets to Fox’s initial tweet (“A male suspect arrested. Covered van from garage” at 16:50:42, Oct 26) and all 42 responses to MSNBC’s initial tweet (“Major response by FBI and other law enforcement” at 15:20:47, Oct 26). Some tweets were deleted; analyses distinguish visible tweets and those containing identifiable speech acts. Coding: Broader speech act categories were used for reliability—Assertion/Representative, Interrogative, Expressive, Commissive, Directive, Declaration—coding the first speech act per tweet. Additional coding noted whether tweets contained multiple speech acts, whether they made an argumentative point, political orientation (left/right), presence of images, and humor (attempted but unreliable). Reliability: Interrater agreement for speech act type was 88% (Cohen’s κ = 0.81). For presence of multiple speech acts per tweet, 88% agreement (κ = 0.74). Political orientation coding had 84% agreement (κ = 0.65). Humor coding proved insufficiently reliable. Ethics: The study follows AOIR (2012) guidelines emphasizing context, harm avoidance, and privacy expectations; Twitter content is generally public; data quoted are political commentary from non-vulnerable populations.

- Single speech acts can convey substantive arguments on Twitter due to the enthymematic nature of popular argumentation. Arguments can be effectively communicated even in very brief tweets. - Non-assertive speech acts (interrogatives, expressives, directives/commands) frequently carry argumentative force, not merely assertions. - Distribution of first speech act type (from the first 100 Fox responses and all 42 MSNBC responses; totals refer to visible tweets with identifiable speech acts): Assertion 38 (53.5%), Interrogative 15 (21.1%), Expressive 6 (8.4%), Directive 10 (14.1%), Commissive 0 (0%), Declaration 0 (0%). Assertions dominate but other types are substantial. - Many tweets consisted of a single speech act, yet approximately two-thirds still made an argumentative point: ~64% for Fox responses and ~72% for MSNBC responses. - Responses to Fox and MSNBC were dominated by Trump critics/left-wing voices. In one characterization (Table 2), left-wing tweets comprised 58.8% (25) and right-wing 6.6% (1) among the visible sample (figures presented as reported). - A greater percentage of responses to Fox included images, often used to challenge perceived framing (e.g., claims that Fox avoided showing the suspect’s van covered with pro-Trump stickers), highlighting the rhetorical interaction of text and image. - Humor—especially sarcasm and irony—appeared to be a powerful rhetorical device, frequently targeting Fox News, Trump, the administration, or Trump supporters. Although formal coding was unreliable, the authors estimate 20–60% of tweets displayed humorous elements. - Intertextual context and sequencing (prior tweets and embedded media) play crucial roles in how singular tweets convey arguments via implied premises and moralized descriptors.

The findings support that political argument is not only possible within Twitter’s character limit but is often achieved through terse, enthymematic constructions. Even a single speech act can advance an argument by leveraging shared contextual knowledge, intertextual cues, and morally charged descriptors. Non-assertive forms—questions, expressives, and directives—regularly function rhetorically, signaling accusations (e.g., dishonesty, hypocrisy), calls to action, or evaluations without explicit syllogistic structure. Differences by outlet suggest that responses to Fox News contained more suspicion toward the source, often using images and sarcasm to challenge perceived framing, whereas responses to MSNBC directed criticism more toward Trump or the administration than at the outlet itself. Humor and morality are intertwined: sarcasm highlights incongruities with norms of honesty, consistency, or fairness, thereby making moral points while amusing in-group audiences. Overall, the study demonstrates that the enthymematic character of public discourse allows substantial argumentative content to be communicated briefly, with context and media (text-image relations) amplifying argumentative effect.

The study lays theoretical groundwork for analyzing political argumentation on Twitter through the lens of speech acts and enthymemes. It shows that: (1) even single, brief speech acts can carry arguments; (2) non-assertive forms (interrogatives, expressives, directives) often function argumentatively; (3) humor can be a potent rhetorical tool tied to moral critique; and (4) responses in the examined case were dominated by left-leaning interlocutors. The authors advocate future research using rigorous quantitative content analysis to test representativeness, refine reliable humor detection, systematically compare argumentative styles across the political spectrum, and analyze the rhetoric of images in conjunction with text. They argue that understanding how implication and context operate—beyond explicit logical form—will be essential for grasping political argument in social media environments.

- Illustrative, qualitative focus limits generalizability; the sample emphasizes specific event-based threads (Cesar Sayoc arrest) and two outlets (Fox, MSNBC). - Reported counts in tables are constrained to visible tweets and initial speech acts; some inconsistencies and deletions affect denominators. - Political orientation coding is skewed (dominated by left-wing responses), limiting conclusions about comparative left/right styles. - Humor coding lacked sufficient interrater reliability, preventing robust quantification. - The complex relationship between text and images—central to many arguments—was acknowledged but not systematically analyzed. - Reliance on the first speech act per tweet may overlook argumentative content in subsequent clauses or media. - The study precedes a full-scale content analysis; findings are preliminary and hypothesis-generating.