Running head: Joke Meaning Activation Getting a Joke: The Time Course of Meaning Activation in Verbal Humor

Two lexical decision semantic priming experiments examined when in the course of reading a joke the initial and the intended meanings are primed, whether the meanings overlap in time, and what happens to the initial reading when the punchline is encountered. In Exp. 1, probes related to the first activated sense (S1) vs. the second sense (S2), or true meaning, were presented at each of three temporal sites for visually displayed joke tests: shortly after joke onset, at an intermediary position and at punchline offset whereas in Exp. 2, probes were presented at joke offset following prolonged viewing. The results from Exp. 1 showed S1 priming effects at the initial and intermediary time point. Priming for S2 also emerged at the intermediary time point and persisted at the final time point. In Exp. 2, the priming effect at joke offset was reliable only for S2. The results are taken to support a concurrent meaning activation view (in line with Attardo, 1997) at incongruity detection, and a selective activation view (in line with Giora, 1991) at incongruity resolution. Joke Meaning Activation 3 Getting a Joke: The Time Course of Meaning Activation in Verbal Humor Whereas much of normal discourse is potentially ambiguous, the ambiguity in most instances is usually minimized through the use of disambiguating cues provided by the context. In contrast, jokes are an example of what Kittay (1987) has called purposive ambiguity. In such instances, words are not intended to have only a single meaning and the context-congruent meaning turns out to have very little to do with the true meaning of the joke. In fact, as Nerlich and Clarke (2001) point out, it is remarkable how much of daily linguistic interaction is structured by a play with multiple meanings. Playing with multiple meanings is a critical ingredient of jokes, which typically set up an opposition between two disparate meanings. In his treatise on creativity, Arthur Koestler (1964) argued that humor involves bisociative thinking, which he defines as perceiving a situation in terms of two self-consistent but totally incompatible frames of reference. Indeed, incongruity is commonly agreed by humor theorists to be a necessary (if perhaps not sufficient) criterion for an utterance or an event to be perceived as funny (Vaid & Ramachandran, 2001). The first and still leading application of script/frame/schema theory to humor was developed by Victor Raskin (1985) and formed the basis for subsequent cognitively oriented frameworks of humor processing (e.g., Attardo & Raskin, 1991; Norrick, 1986; Veatch, 1998). Norrick recast the bisociative analysis of humor in frame theoretic terms, proposing that jokes first evoke and then skew a schema. Joke perception is said to require that the perceiver first recognize a schema (or a script, in Attardo and Raskin's, 1991 framework) and then reinterpret it in the face of the conflict created by information that is Joke Meaning Activation 4 inconsistent with the original schema. This reinterpretation process, which has also been described as incongruity resolution (Suls, 1972) or frame-shifting (Coulson, 2001), is seen as critical to the enjoyment of humor. Cognitively oriented accounts of humor have, thus, posited distinct stages of information processing in joke comprehension (Attardo, 1997; Suls, 1972). There is an initial set up phase, which lays the groundwork for the incongruity by instantiating an initial schema and a meaning consistent with that schema. In the second phase (incongruity discovery), the expectancy created by the initial schema is violated. Finally, in the third phase (resolution), the conflict created by the expectancy violation is reduced by a reinterpretation of the joke in terms of an alternate, higher level schema in which the joke makes sense in light of the new information presented in the joke’s punchline. The various theoretical accounts of humor concur in positing abrupt shifts in the particular meanings or scripts activated in the course of comprehending a joke. Yet surprisingly little empirical research exists on the time course of meaning activation in joke processing. Attardo (1997: 416) notes that, aside from claiming "that the onset of the incongruity must be sudden, or that the simultaneous consideration of the two opposed scripts must be brief" most theories of humor "have little to say about timing". The present research was conducted to redress this gap. The questions addressed in this research concern when each reading of a joke becomes accessible to awareness, whether the two readings overlap in time, and what happens to the initial reading once the punchline meaning is encountered. Using a lexical decision semantic priming procedure, the research operationalized and tested two conceptualizations of joke perception with differing implications for the time course of multiple meaning activation in humor. We Joke Meaning Activation 5 first summarize these two conceptualizations. We have termed these the concurrent activation and the selective activation views. Concurrent Activation View. This view has its origins in the script opposition model of verbal humor, developed by Attardo and Raskin (1991) and subsequently elaborated by Attardo (1997) in terms of a three-stage model of humor processing (the stages being Setup-Incongruity-Resolution, or SIR). In this view, humor is defined as a semantic opposition between differing scripts, where scripts refer to collections of semantic information on a given subject, embodying the sum of cultural knowledge, and are represented as a set of expectations and/or weighted choices. Attardo notes that, to the extent that joke texts being linear are subject to temporal ordering, scripts presented earlier become part of the context and establish a framework of expectations against which scripts occurring later are to be processed (Attardo, 1997: 403). However, the two competing scripts must in some sense coexist temporally in the mind of the humor comprehender. This coexistence is needed at the incongruity phase, Attardo notes, “for the time necessary to the evaluation of the dissimilarity between the expected and the observed data" (p. 412). Coexistence is also needed at the resolution phase, where "the two interpretations must be conserved in 'working memory' for a certain period of time necessary for the research of a cognitive rule capable of solving the incongruity" (Attardo, 1997: 412). Consistent with this view, Nerlich and Clarke (2001) suggest that humorous uses of incongruity are examples of an 'ambiguated' use of context in which language is kept alive by "keeping several meanings of a word or expression in mind simultaneously" (p. 9). They refer to research that shows that understanding of polysemous words results in the Joke Meaning Activation 6 continued activation of inappropriate senses for some time after the word has been encountered. Selective Activation View. A somewhat different formalization of humor is presented by Giora (1991), who offers two criteria for successful or well-formed jokes involving semantic ambiguity (rather than syntactic or pragmatic violations). The first criterion is a marked informativeness requirement. Giora notes that all texts involve a progressive increase in informativeness. In non-joke texts this increase is held to progress gradually. In joke texts, however, the increase occurs suddenly and abruptly and involves a shift from an initial unmarked to a final marked meaning of the text. Giora’s second criterion for well-formedness of jokes is that the joke “...causes the reader to perform a linear shift: the reader is made to cancel the first unmarked interpretation upon processing the second marked interpretation,” (Giora, 1991: 470). While the exact timing of this linear shift is not specified, what is important to note here is the claim that, once processed (i.e., upon resolution of the joke), the true joke meaning has the effect of erasing the impact of the original reading of the joke text. Indeed, Giora asserts that the canceling of the original reading is what makes a joke funny: “where ambiguity is not cancelled but sustained, the effect is witty but not funny,” (Giora, 1991: 476). Colston, Giora and Katz (2000 as cited in Giora, in press) go so far as to suggest that the initial reading of the joke text is actually suppressed once the joke’s actual meaning is realized. Consistent with this view, Coulson and Kutas (2001) propose that a semantic reinterpretation process characterizing joke processing is facilitated if aspects of the initial interpretation are suppressed in working memory. Joke Meaning Activation 7 On the basis of the above theoretical positions, two complementary predictions about meaning activation during joke processing may be put forth. The position associated with Attardo (1997) would predict that both the first activated sense of the joke (S1), what we also refer to below as the ‘favored’ meaning, and the second activated sense (S2), or the surprising, true joke meaning, should be concurrently (and perhaps even simultaneously) activated as early as incongruity discovery but certainly by incongruity resolution. Giora (1991) would predict that S1 should be less active than S2 and that S1 would perhaps even be suppressed upon joke resolution. The Present Research The present research was conducted as an empirical test of these two predictions. Lexical decision reaction times (RT) taken at different points during the reading of a joke text were used to determine whether and at what specific time point a target word related to either the favored or the surprising meaning would be semantically primed and whether the priming occurs gradually or abruptly. Two experiments were conducted. In the first experiment, probes related to S1 (the favored meaning) vs. S2 (surprising meaning) were presented at three discrete temporal points in joke texts: immediately after the joke was introduced (i.e., during joke setup), at an intermediate point in the joke text when information discrepant with the initial reading is introduced (i.e., during the incongruity phase), and immediately after the onset of the punchline (resolution phase). Baseline response times were measured by presenting target words prior to joke onset. The difference in RT between baseline responses and responses at each of the three temporal sites allowed for a comparison of the size of priming effects. The second experiment explored the fate of S1 after punchline offset (i.e., post resolution), Joke Meaning Activation 8 using a slightly different priming procedure in which lexical decision RTs to targets related to the favored vs. surprising meaning were compared relative to semantically neutral controls. EXPERIMENT 1 This experiment addressed two issues: 1) whether the shift from the initial sense to the joke reading of the text occurs gradually or abruptly, and 2) whether the initial (S1) and joke readings (S2) are concurrently or selectively primed at or beyond the incongruity stage. The working hypotheses for the rate of meaning activation were as follows: a) b) If the activation of the joke meaning occurs gradually, one would expect a progressively increased activation for S2 over the three time points, i.e., from setup to incongruity and from incongruity to resolution. If joke meanings involve a sudden frame shift, there should be no priming of the S2 at setup and a high level of priming of S2 at incongruity, which should persist at resolution. The working hypotheses regarding the time course of meaning activation during verbal humor processing were as follows: a) Concurrent activation will be evidenced by equivalent priming of S1 and S2 as early as the incongruity stage and upon presentation of the punchline. b) Sequential activation will be evidenced by selective priming of S1 at setup, followed by selective priming of S2 at the incongruity stage and at punchline onset. Joke Meaning Activation 9