Irony processing within a linguistic framework: a corpus-based theoretical analysis

Figurative (or nonliteral) language presents a persistent challenge for researchers in different scientific fields, ranging from linguistics to philosophy and psychology. Its intricate nature, particularly evident in phenomena like irony, has led scholars to outline hypotheses and theories that could potentially account for its complexity and specificity. Early attempts to explain ironic meaning through a linguistic (and, more specifically, pragmatic) lens included H.P. Grice’s theory of conversational implicature, according to which listeners or readers have to first process an utterance’s literal meaning to subsequently access and process its opposite non-literal meaning. Following Grice’s account, dubbed the “’literal-first’ model” (Sperber; Wilson, 2012, p. 125), an addressee would access the nonliteral meaning of a given utterance only if the utterance’s literal meaning is in contradiction with the communicative context. Dissatisfaction with the Gricean model led to the formulation of influential alternative theories, such as Sperber and Wilson’s (1981; 1992; 1995) “relevance theory”, which emphasizes the notion of “second-degree interpretation” in irony processing, Gibbs’s (1983; 1989; 1994; 2002) “direct access view”, which argues that ironic meaning may be processed before literal meaning, and Giora’s (2002) “graded salience hypothesis”, whose foundation is the degree of salience of a given meaning, which would further dictate processing order and cost. Although these theories may seem, at first glance, contradictory, and diverge largely in terms of the emphasis they concede to a listener’s or reader’s interpretation, the role the degree of familiarity plays in irony processing, and whether irony processing requires further cognitive effort in comparison to literal language, we argue that they share much common ground and may even be complementary in their respective assertions. Thus, rather than continuing to frame these theories as oppositional, this article advocates for a reassessment of their potential as complementary frameworks, which may shed further light on the nature and processing of irony. To that end, we provide a concise literature review of the four theories identified, outlining their main underlying assumptions, followed by an approximation among them, highlighting possible points of convergence. To illustrate our theoretical stance with empirical data, we conducted a small-scale qualitative corpus analysis using EPIC (Frenda et al., 2023), a corpus of short online conversational exchanges composed of parent posts and replies. Replies were independently annotated by evaluators from different English-speaking countries and classified as ironic or non-ironic. This study is expected to contribute to the scientific understanding of figurative language from a pragmalinguistic and psycholinguistic standpoint.