Abstract
Instructional humor processing theory has been proposed to explain how the type of humor used by the course instructor can affect student learning. In this study, a cross-sectional design was used to test whether the relation between the instructor’s type of humor (related, unrelated, self-disparaging, offensive, and disparaging humor), and learning is mediated by variables assumed by the instructional humor processing theory (emotion, motivation, and information-processing ability). A total of 360 university and junior college students with a mean age of 19.31 years (standard deviation = 0.75) completed a questionnaire concerning their impression on the instructor humor, emotion, motivation, processing ability, and cognitive learning in a specific course. The results revealed that the instructor’s-related humor significantly predicted the cognitive learning of students, and their relation can be mediated by all variables assumed by instructional humor processing theory. Our results provided preliminary evidence for the legitimacy of the instructional humor processing theory model.
Introduction
Instructors use humor for various reasons, and the most important reason, many would argue, is to enhance learning (Wanzer, Frymier, & Irwin, 2010). The most powerful finding concerning the effectiveness of instructor humor for student learning was obtained through the naturalistic experiment by Ziv (1988). Students who participated in the study were randomly assigned to either a humor or nonhumor group, each of which received 14 weekly lessons. In the humor group, important concepts were explained using interesting stories and jokes. The results revealed that students in the humor group achieved clearly higher scores on the term-end examination. Ziv argued that the experiment yielded powerful results because the instructor humor was directly associated with the course content. This conclusion has been supported by subsequent correlational studies (Bieg & Dresel, 2018; Machlev & Karlin, 2016; Wanzer et al., 2010).
In correlational studies, instructor humor was initially divided into five types according to a survey administered to college students (Frymier, Wanzer, & Wojtaszczyk, 2008; Wanzer, Frymier, Wojtaszczyk, & Smith, 2006). Related or unrelated humor denotes humor related or unrelated to the course content; self-disparaging humor is humor directed toward the instructor himself/herself, whereas disparaging humor is humor that is meant to insult students, politicians, or other instructors; offensive humor is essentially crude and often sexual. Studies that have investigated the relations between the instructor’s type of humor and the cognitive learning of students, such as college students in the United States and 9th and 10th graders in Germany (Bieg & Dresel, 2018; Machlev & Karlin, 2016; Wanzer et al., 2010), have consistently revealed a positive correlation between related humor and student learning.
In recent years, the need to develop a theoretical model that explains the relation between instructor humor and student learning has been pointed out, and Wanzer et al. (2010) proposed the instructional humor processing theory (IHPT). This theoretical model borrows its basic framework from the Elaboration Likelihood Model (ELM; Petty & Cacioppo, 1981, 1986) and hypothesizes that for students to enhance their learning, they need the motivation and ability to process messages. We will provide an overview of the IHPT below (for details, see Wanzer et al., 2010).
According to the IHPT, messages that are perceived by students as a type of humor are evaluated as a tool that induces positive or negative affect. Humor is often accompanied by one of more incongruities that must be resolved. Messages whose incompatibilities have been resolved usually cause a positive affect in the recipient (LaFave, Haddad, & Maesen, 1996). However, aggressive humor that uses other people or members of the same group as targets is highly likely to cause a negative affect (Zillmann & Cantor, 1996). In fact, Frymier et al. (2008) discovered that students label humor that attacks individual students, sororities, fraternities, political affiliations, and men or women as a group as inappropriate. The IHPT considers that a positive or negative affect determined in this manner influences a student’s subsequent motivation. Positive affect that has been generated by appropriate humor motivates the students to process humorous messages. Negative affect, on the other hand, works to hamper this activity.
The ELM, moreover, identifies the causes of distraction as a variable that influences people’s processing ability. Humor has often been described as a strategy that attracts attention (Gorham & Christophel, 1990). If an instructor uses humor in class, students need to pay even closer attention since, by definition, humor is often accompanied by unease and uncomfortable feelings that must be resolved. One of the claims of the IHPT is that learning is enhanced if an instructor’s humorous messages to attract students’ attention include elements related to the content of the course that increase the students’ processing ability. An instructor’s humor, however, might also distract students’ attention from the course content and lower their processing ability. An insulting and aggressive type of humor that causes negative affect is likely to lower students’ motivation and, at the same time, may reduce their processing ability by distracting them from the course content. In summary, the IHPT claims that the effects of instructor humor—whether it enhances student learning or not—differ according to the type of such humor. Moreover, the differences in such effects may be explained by three mediators: emotion, motivation, and processing ability.
Although the IHPT has been widely recognized as a useful model in explaining the effect instructor humor has on student learning (Banas, Dunbar, Rodriguez, & Liu, 2011), its theoretical verification has not been adequately addressed. Wanzer et al. (2010), who proposed the IHPT, only studied the relation between the type of instructor humor and student learning and did not test the mediators assumed by the theory. Based on the IHPT, Bieg, Grassinger, and Dresel (2019) showed in longitudinal studies that “related humor” induced an increase in positive affect (enjoyment) and a decrease in negative affect (boredom and anger), but that disparaging humor had the opposite influence. However, the researchers did not examine whether these functioned as mediators with learning and did not investigate other mediators (motivation and processing ability) that the IHPT hypothesizes. Among other mediators between instructor humor and student learning presupposed by the IHPT, Bolkan and Goodboy (2015) focused on and tested emotion and processing ability (measured as sustained attention). They then demonstrated that emotion serves as a mediator, but processing ability does not, and concluded that the IHPT lacks legitimacy. Although the study of Bolkan and Goodboy (2015) is valuable as a preliminary attempt to verify the IHTP, it did not include student motivation, which is a central and essential mediator assumed by the theory. Although some studies have shown that student motivation and learning (measured as elaboration) were mediated by the interestingness of the teachers’ instruction as perceived by the students (Bieg & Dresel, 2018), they did not necessarily test whether motivation mediated the relationship between instructor humor and student learning, as the IHPT hypothesizes. Following the ELM (Petty & Cacioppo, 1981, 1986), the IHTP explains that a student’s motivation to actively scrutinize information exerts a great influence on their learning. Another issue is that Bolkan and Goodboy (2015) did not take into account the type of humor. The literature (Frymier et al., 2008; Wanzer et al., 2006) has identified five dimensions of instructor humor. The IHTP conceptualizes instructor humor as a multidimensional construct and argues that appropriate and inappropriate humor affect student learning through different processes. There are currently no studies that have investigated all the mediators that the IHPT hypothesizes while taking into account the five types of instructor humor.
Studies inspired by the IHPT have sought new models to explain the relationship between instructor humor and student learning. Bieg and Dresel (2018) have proposed a model in which the relationship between instructor humor and learning is mediated by the teachers’ instructional dimensions (teachers’ care, interestingness of instruction, and clarity of instruction) as perceived by the students and investigated whether they mediated instructor humor and student learning. The results showed that the positive association between related humor and learning (measured as elaboration) was mediated by all instructional dimensions. On the other hand, the negative association between unrelated humor and learning was shown to be mediated by the interestingness of instruction, and the negative association between aggressive humor and learning was shown to be mediated by the teachers’ care and clarity of instruction. Bolkan and Goodboy (2015) proposed a model in which the associations between instructor humor and student learning were mediated by the “basic needs” (autonomy, competence, and relatedness) of self-determination theory (SDT; Deci & Ryan, 1985). Their results support this. Although proposals of new models such as these have value, researchers will need to evaluate the individual models appropriately and verify them carefully. As mentioned previously, however, this has not necessarily been done in the case of the IHPT. Therefore, the legitimacy of the IHTP clearly requires further testing.
This study targeted Japanese college students to test whether there is any relation between the type of instructor humor perceived by students and cognitive learning as perceived by students as well as whether the mediators assumed by the IHTP, including emotion, motivation, and processing ability, do in fact mediate this relation. Bolkan and Goodboy (2015) found that processing ability is not a mediator between humor and learning. However, we found that if we account for the type of instructor humor, there is a positive association between related humor and student learning. We then assumed that this association is mediated by all variables assumed by the IHPT (Hypothesis 1). Bieg and Dresel (2018) found a negative relation between unrelated humor and learning, while Wanzer et al. (2010) found no such correlation. In addition, Wanzer et al. (2010) found a positive relation between self-disparaging humor and learning, while Bieg and Dresel (2018) found no such correlation. As Bieg et al. (2019) have stated, however, these types of humor do not appropriately support clarity of instruction or teaching practices. The authors therefore hypothesized that they are unrelated to learning, since they generate neither positive nor negative affect in the classroom and do not influence motivation or processing ability (Hypothesis 2). Similarly, the IHPT (Wanzer et al., 2010) regard disparaging and offensive humor as inappropriate and anticipates that negative affect reduces motivation and impedes the processing of messages as “distractions.” The authors therefore hypothesized that these types of humor show negative associations with learning and that these associations are mediated by all the variables that are hypothesized by the IHPT (Hypothesis 3).
Method
Participants and procedure
A total of 360 participants (114 males, 244 females, and 2 unknown) recruited from three schools—a private university, a private junior college, and a national university—were enrolled in this study. The participants were aged 18 to 23 years, and the mean age was 19.31 years (standard deviation (SD) = 0.75). A briefing on the survey was given after the end of the lessons, and students who signed the informed consent completed the self-administered survey. Most previous studies have used the approach introduced by Plax, Kearney, McCroskey, and Richmond (1986) to increase variability in the sample of instructors, whereby the course given immediately before the data collection date is recalled. In this study, the following approach was taken to further increase the variability of instructors: Students with either 0 or 1 in the last digit of their student number were instructed to think about the instructor of the Monday course; students with 2 or 3, the instructor of the Tuesday course; students with 4 or 5, the instructor of the Wednesday course; students with 6 or 7, the instructor of the Thursday course; and students with 8 or 9, the instructor of the Friday course. Moreover, students with an even last digit (including 0) were instructed to think about the first course of the day, students with an odd number the last. The survey was conducted during either the 14th or 15th week of a 15-week program.
Measures
Instructor humor
The scale developed by Frymier et al. (2008) was translated and used. This scale comprises five subscales with a total of 25 items. The five subscales are as follows: Related Humor (e.g., uses humor related to the course material), Unrelated Humor (e.g., tells jokes unrelated to the course content), Self-Disparaging Humor (e.g., makes fun of him/herself in class), Offensive Humor (e.g., uses vulgar language or nonverbal behaviors in a humorous way), and Disparaging Humor (e.g., picks on students in class for their intelligence). The participants were asked to rate the degree to which the description in each item was true of the course instructor based on a five-point Likert-type scale (1 = strongly disagree to 5 = strongly agree). The internal consistency of this study (Cronbach’s α) was .85 for related humor, .76 for unrelated humor, .74 for self-disparaging humor, .90 for offensive humor, and .87 for disparaging humor.
Affection
The Japanese version (Sato & Yasuda, 2001) of the Positive and Negative Affect Schedule (Watson, Clark, & Tellegen, 1988) was used. This scale comprises two subscales with a total of 20 items, namely, Positive Affect (e.g., “Active,” “Proud,” “Strong”) and Negative Affect (e.g., “Afraid,” “Scared,” “Upset”). The participants were asked to rate on a six-point Likert-type scale (1 = strongly disagree to 6 = strongly agree) the degree to which the description of each item was true of them during the course. The internal consistency was .89 for Positive Affect and .90 for Negative Affect.
Motivation
The Richmond (1990) Scale was translated and used. The participants were asked to respond to five items (e.g., motivated–unmotivated, excited–bored, uninterested–interested) on seven-step bipolar scales (1 and 7 = very much, 2 and 6 = moderately, 3 and 5 = slightly, and 4 = neither) with reference to the statement “My feelings about studying the content in the course.” The internal consistency was .77.
Processing ability
Bolkan and Goodboy (2015) measured processing ability as sustained attention. Although the ability to sustain attention to a message can be considered an aspect of processing ability, the essential aspect of this concept is, according to Wanzer et al. (2010), the ability to think positively about and understand the message. Therefore, the following three items were assessed in this study: It was easy to understand what the instructor talked about, I was able to listen closely to the lecture, and I was distracted and the lecture content did not enter my mind (reverse item). The participants were asked to rate the degree to which the description in each item was true of them during the course based on a five-point Likert-type scale (1 = strongly disagree to 5 = strongly agree). The internal consistency of the three items was .64, which was inadequate. However, after we excluded one item (“I was distracted and the lecture content did not enter my mind”), the internal consistency increased to .88. Therefore, this item was excluded and only the other two items were used in the subsequent analysis.
Cognitive learning
Although the method used in this study and that in Plax et al. (1986) have the advantage of maximizing the variability of instructors, they make it difficult to demonstrate a scale that is objective and reliable in the measurement of cognitive learning (Frymier & Houser, 1999). In other words, they cannot use objective scores obtained from course tests. Thus, the global item developed by Richmond, McCroskey, Kearney, and Plax (1987) to measure perceived cognitive learning was translated and used. The participants were asked to answer the question “How much did you learn in this course?” on a 10-point Likert-type scale (0 = nothing to 9 = more than in any other class). This item have been used in a variety of studies as it offer estimated values of cognitive learning (e.g., Christophel, 1990; Frymier, 1994; Richmond, 1990).
Statistical analysis
First, descriptive and bivariate analyses of the main study variables were conducted. Second, multiregression analysis examined the effect that the five types of instructor humor had on students’ perceived cognitive learning. Finally, regarding the types of instructor humor found to have significant effects on student learning, Process Macro for SPSS (Hayes, 2013) model 4 was used to examine the indirect effect of the mediators (affect, motivation, and processing ability) assumed by the IHPT. We present the path model in Figure 1. Specifically, multiple mediation analyses were carried out to examine the significance of the indirect effect using the bootstrapping method. The bootstrap method used 5000 resamples. Indirect effects were deemed significant if the 95% confidence intervals (CIs) did not include zero.

Path model illustrating our mediation hypotheses including labels of the direct and indirect paths.
Results
Means, SDs, and intercorrelations among the study variables are presented in Table 1.
Means, SDs, and intercorrelations among study variables.
Note: N = 360. RH: related humor; UH: unrelated humor; SDH: self-disparaging humor; OH: offensive humor; DH: disparaging humor; PAf: positive affect; NAf: negative affect; Mot: motivation; PAb: processing ability; CL: cognitive learning; SD: standard deviation.
*p < .05. **p < .01.
Next, we investigated the effect of the explanatory variable (type of instructor’s humor) on the target variable (perceived cognitive learning) using multiple regression analysis (Table 2). Only related humor showed a positive effect on perceived cognitive learning.
Multiple regression analysis of the effect of teacher’s humor on cognitive learning.
Note: SE: standard error.
Accordingly, affect (positive and negative), motivation, and processing ability were tested as mediators of the relation between related humor and perceived cognitive learning (Table 3). Thus, the other four types of humor (i.e., unrelated, self-disparaging, offensive, and disparaging) were entered as covariates.
Multiple mediation analyses of the relation between instructor humor and cognitive learning.
Note: CI: confidence interval; SE: standard error; RH: related humor; PAf: positive affect; NAf: negative affect; Mot: motivation; PAb: processing ability; CL: cognitive learning.
The indirect effect was significant for all humor types, as the 95% CIs did not include zero. Although the total effect was significant, the direct effect ceased to be significant following the introduction of the parameters, suggesting the plausibility of complete mediation by the four parameters.
Discussion
This study tested the legitimacy of the IHPT by taking into account five types of instructor humor identified in the previous studies (Frymier et al., 2008; Wanzer et al., 2006). Contrary to the results of Bolkan and Goodboy (2015), processing ability was shown to fully mediate the relation between related humor and student learning. This is seen from the fact that its indirect effect is the largest of all four mediators. In addition, Bieg and Dresel (2018) showed that the positive association between related humor and learning was mediated by the instructors’ clarity of instruction as perceived by the students. However, clarity of instruction was measured in this study using more or less the same items as for processing ability. Further, whereas Bolkan and Goodboy (2015) did not test student motivation as a mediator, this study demonstrated that student motivation functions as a mediator, as assumed by the IHPT. These results support our Hypothesis 1 and were in agreement with the theoretical model proposed by the IHPT. Thus, the process explained in the IHPT regarding related humor at least (i.e., the process in which appropriate humor generates a positive emotion in students and improves their learning by enhancing processing ability and motivation) seems to have a rational basis. The result of the positive association between related humor and student learning was completely in agreement with the previous studies (Bieg & Dresel, 2018; Machlev & Karlin, 2016; Wanzer et al., 2010). Bieg and Dresel (2018) concluded that from the practical point of view, instructors can promote student learning using course-related humor. This study results supported this claim. Related humor has been known to exhibit a positive association with learning, regardless of culture and the developmental stage of children.
The other four types of humor did not significantly predict learning. Therefore, Hypothesis 2 was supported, but Hypothesis 3 was not. This study posited Hypothesis 2: that unrelated humor and self-disparaging humor were unrelated to learning. The results supported this hypothesis. However, a study of U.S. college students by Wanzer et al. (2010) demonstrated a positive association between instructors’ self-disparaging humor and learning, and a study of German children by Bieg and Dresel (2018) reported a negative association between unrelated humor and learning. Therefore, the pedagogical effects of these types of humor may differ according to one’s culture and developmental stage, as has been suggested in the previous work (Banas et al., 2011; Bieg & Dresel, 2018). Thus, this point must be verified in detail in the future research by collecting samples from diverse cultures and different age groups.
In Hypothesis 3, we hypothesized that disparaging humor and offensive humor showed negative associations with learning, and that these associations were mediated by all the variables hypothesized by the IHPT. This hypothesis, however, was not supported. An experimental study conducted in Japan (Tsukawaki, 2018) showed that humor generated positive affect in the receiver if the closeness level was high and aggression level was low. The educational effects of these types of humor may vary according to the degree of existing rapport with the instructor and the degree of aggression in instructor humor.
This study conducted a preliminary evaluation of the IHPT, which is a pioneering model explaining the association between instructor humor and student learning. The explained variance ratio of student learning using the five types of instructor humor was by no means high, at 8%. We consider this a limitation of this study. However, related humor was shown to significantly predict student learning, and these associations were mediated by the variables hypothesized in the theory. As noted above, several models have been proposed in recent years to explain these associations using the teachers’ instructional dimensions (teachers’ care, interestingness of instruction, and clarity of instruction) as perceived by the students (Bieg & Dresel, 2018), as well as basic needs (autonomy, competence, and relatedness) (Bolkan & Goodboy, 2015) based on the SDT (Deci & Ryan, 1985). These models have been proposed individually but overlap in certain areas. For example, clarity of instruction, which is included in instructional dimensions, measures more or less the same concepts as processing ability in the IHPT; and relatedness, which is included in SDT, measures more or less the same concept as teachers’ care, which is included in instructional dimensions. In the future, researchers will need to categorize and integrate these models in order to construct a more appropriate model. Moreover, it must be noted that as a large majority of studies, including this study, have a cross-sectional design, they cannot explicitly identify causal relations. An experimental study to replicate the results suggested by correlational methods is clearly necessary.
