When a Topic Matters to You,Does It Matter if You Read About It in a Second Language?

Abstract

Readers’ emotions often become engaged while reading and can sometimes enhance and sometimes skew text comprehension, with most research focused on reading texts in one’s native language. This project extended the work of Gaskins to explore how adolescents’ culturally constructed emotions affected their reading comprehension, and how this effect varied when reading in their first (Korean) or second (English) language. Students (N = 477) in a Korean high school read short paragraphs on four expository topics counterbalanced to represent both languages and both an emotional and neutral version of each topic. Results were that participants were more emotionally engaged and performed better on comprehension assessments when reading emotional than neutral texts. Thus, unlike Gaskin’s study, these results indicated that emotions were generally helpful for comprehension, whether the text was in the reader’s first language or a learned language, possibly by way of increasing attention and engagement with the text.

Keywords

emotions second language reading comprehension

Purpose

This study investigated the role of emotional involvement in first- and second-language reading comprehension. Readers’ emotions often become engaged while reading, and there is evidence that aroused emotions can sometimes enhance and sometimes skew text comprehension. However, except for the rich literature taking a reader response perspective, research on the role of emotions while reading is not extensive, especially for reading in a second language, and many questions remain about how emotional and cognitive processes intertwine during reading. In this project, we were interested in extending the work of Gaskins (1996) and others to explore how adolescents’ culturally constructed emotions affected their reading comprehension, and how this effect varied when reading in their first or second language.

Theoretical Framework and Relevant Literature

Emotions in the Classroom

We begin by reviewing the burgeoning literature on emotions in academic settings. From their recognition of the narrowness of research on emotions and learning, Pekrun (2000) and Pekrun, Goetz, Titz, and Perry (2002) called for a broader perspective, moving from a nearly sole focus on anxiety to include four distinct categories of emotions: negative activating, negative deactivating, positive activating, and positive deactivating emotions. In response, educational researchers have tried to investigate the underlying mechanism of the reciprocal relationship between emotions and academic achievements. Do and Schallert (2004) explained how, triggered by personal and social factors, positive emotions facilitate engagement in discussion, whereas negative emotions are associated with disengagement, a finding echoed in Linnenbrink-Garcia, Rogat, and Koskey’s (2011) study of emotions in small group work. Pekrun, Elliot, and Maier (2009) reported that emotions mediated the relationship between achievement goals and academic performance, a finding also reported in different cultural contexts (e.g., King, 2010; Villavicencio & Bernardo, 2013). In our study, we extended the exploration of the reciprocal relationship between emotion and cognition to text processing.

Emotions When Reading in One’s Native Language

In the general literacy field, research on emotions began in reaction to a preoccupation with the cognitive aspects of reading. This research blossomed in the 80s and 90s, with differential results reported for literary and expository texts. For literary reading, readers’ emotions were explored from a reader response criticism perspective: the reader, not the text or the author, was said to determine the meaning of the text, and this meaning is imbued with an emotional response to the text (Rosenblatt, 1978; Tompkins, 1980). Another stream of emotion research was conducted with expository texts, starting as an extended branch of the role of attitude and interest in text processing. Both lines conducted research within the same cognitive traditions but differed in foci, approaches, and findings about readers’ emotions.

As a first point of contrast, research with literary texts was more concerned with emotions evoked by the text, whereas research with expository texts dealt more with emotions that readers brought to the text. Miall and Kuiken (1994, 1999) investigated how the literary stylistic feature of foregrounding arouses readers’ emotions and changes their perceptions about the world. In a similar vein, Sadoski, Goetz, and Kangiser (1988), Sadoski, Goetz, Olivarez, Lee, and Roberts (1990), and Goetz, Sadoski, Stowe, Fetsco, and Kemp (1993) studied readers’ affective responses aroused by story structures such as suspense and reported a convergence between mental imagery and affective response associated with such story structures. By contrast, in a study of issue-related emotions in expository reading, Gaskins (1996) reported that readers’ emotions caused by ego involvement biased their interpretation of the text. Ainley, Corrigan, and Richardson (2005) studied how topic-related interest predicted persistence in reading and noted that the effect of interest varied across topics and texts.

As a second issue, research on expository reading has attempted to separate the effects of prior knowledge from the role of readers’ emotional involvement. By contrast, research on literary reading has treated readers’ prior experiences as sources of emotional engagement. For example, autobiographical memory and prior knowledge were emphasized as sources for “identification,” “transportation,” and “simulation” (e.g., Miall & Kuiken, 2002; Oatley, 1999; Zwaan & Radvansky, 1998), three subprocesses associated with literary reading. Conversely, in expository reading, researchers found diminished effects of topic-related attitude when they held prior knowledge statistically constant (Fortner & Henk, 1990; Henk & Holmes, 1988; Reutzel & Hollingsworth, 1991). Thus, given the different foci and approaches, previous studies reached different conclusions about readers’ emotions, with emotional involvement seen as an interfering variable in text comprehension of expository texts but as evidence of deep engagement and better comprehension in the research with literary texts (Kneepkens & Zwaan, 1994; Komeda & Kusumi, 2006; Sadoski, Goetz, & Kangiser, 1988; Sadoski, Goetz, Olivarez, Lee, & Roberts, 1990; Zwaan, Dijkstra, & Graesser, 1993).

Nevertheless, several themes about readers’ emotions have emerged consistently from the previous research. First, emotions play an active role in reading. Readers’ emotions in the form of attitudes and interest function as “a frame of reference” in expository reading (Ainley, Corrigan, & Richardson, 2005; Baldwin, Peleg-Bruckner, & McClintock, 1985; Henk & Holmes, 1988) and as integral to literary reading (Miall & Kuiken, 1994, 1999, 2002; Oatley, 1999). Second, readers’ emotions are viewed as multifaceted and dynamic. Emotions in literary reading have been distinguished into fiction and artifact emotions (Dijkstra, Zwaan, Graesser, & Magliano, 1994; Kneepkens & Zwaan, 1994) that transform into evaluative, aesthetic, narrative, and self-modifying feelings (Miall & Kuiken, 2002), whereas emotions in expository reading show distinctions among such affective responses as attitudes, interest, and ego involvement (e.g., Baldwin et al., 1985; Gaskins, 1996; Henk & Holmes, 1988). Lastly, emotions are associated with prior knowledge and personal memories for both kinds of reading (e.g., Baldwin et al., 1985; Miall & Kuiken, 2002; Oatley, 1999).

Much of this earlier work regarded readers’ emotions as an individual phenomenon, but, starting in the 2000s, researchers have increasingly used sociocultural perspectives to portray readers’ emotions as culturally situated and socially constructed (Dutro, 2008; Gee, 2000, 2001; Lysaker, Tonge, Gauson, & Miller, 2011; Trainor, 2008). These views hold that readers construct situated meanings from the text using their own cultural models and become emotionally involved through representations of cultural experiences. Readers’ emotions are not solely individual but originate from institutional practices and social relations invited by the text. Thus, the overall implication from prior research in first language reading is that readers’ emotions are multifaceted, playing an active role in the reading process intertwined with readers’ prior knowledge and sociocultural backgrounds.

Emotions When Reading in a Second Language

As for reading in a second language, there has been much less research on readers’ emotions except for anxiety (Brantmeier, 2005; Saito, Garza, & Horwitz, 1999; Sellers, 2000). Second language reading is often illustrated as a complex process that consists of the integration of linguistic and nonlinguistic resources. Linguistic resources include lexical, syntactic, and discourse knowledge about a second language, whereas prior knowledge, attention, reading strategies, and motivation represent nonlinguistic resources (Bernhardt, 2011; Grabe, 2009; J. W. Lee & Schallert, 1997). However, these areas address readers’ emotions only tangentially. Although there is little research focused on emotions associated with reading in a second language, a strong interest has emerged on emotional aspects of second language learning in general. The implication is that emotions in second language situations are socioculturally constructed and experienced (Benesch, 2012; Pavlenko, 2007) and that individuals experience transactions in a second language as having lower emotionality than in a first language (Dewaele, 2010; Pavlenko, 2007). Benesch (2012) connected this lack of emotional resources in a second language with the lack of agency experienced by second language learners.

Another issue with respect to second language readers’ emotions, as implied by the research on reading in one’s native language, is that emotions aroused when reading in a second language are associated with prior knowledge. Here, there are conflicting results about prior knowledge and its effect on second language reading (Bernhardt, 2011; Grabe, 2009; Nassaji, 2007). Some studies (Abu-Rabia, 1998; Jimenez, García, & Pearson, 1996; J. F. Lee, 1986) showed that readers comprehend better when reading culturally relevant texts about which they have prior knowledge, whereas others found that prior knowledge played a weak or debilitating role in reading in a second language (Bernhardt, 1991; Carrell, 1983; Elder, Golombeck, & Stott, 2004).

In addressing the above contradictory findings, Nassaji (2007) suggested that Kintsch’s (1988, 2004) “construction-integration model” can help bring some coherence to the role of prior knowledge in second language reading. The construction–integration model makes a distinction between a “text-base model” and a “situation model” level of understanding. The text-base model emerges through the construction phase of text-based information, whereas the situation model is built through the integration phase with prior knowledge. Nassaji (2007) suggested that different levels of text representations are necessary to understand the distinct contribution of prior knowledge. In our study, we used Kintsch’s model to design our measures of reading comprehension and posited that, by examining adolescent readers’ emotions as they read a text in their first language (Korean) and in their second language (English), we could contribute to the debate about how prior knowledge participates in meaning making.

Our Study

For this project, we focused on how emotional involvement affected comprehension in expository reading. Two versions of texts were developed to manipulate emotional involvement and to see its effect on comprehension as indicated by different levels of comprehension. Our research questions were the following:

Are there differences in the emotional response of adolescent readers reading different text versions (emotional vs. neutral) in their first and second languages (Korean and English)?

Do text-base and situation model measures of comprehension differ for emotional and neutral text versions presented in Korean and English?

Method

Participants and Context

Participants were 477 students (268 girls, 209 boys; age range: 15–17 years) learning English as a foreign language at a public high school in a southern city of South Korea. These students had received school-based English instruction starting in the third grade, 1 hr per week in elementary school, and 4 hr per week in middle and high school, for an accumulated amount of around 612–748 hr, supplemented by after-school instruction of about 2 hr per week for most secondary students. Note that only students whose parents had consented to their participation in the study were included.

As is typical for Korean academic high schools, English instruction was focused on the receptive skills of listening and reading, partly due to large class size (35–40 students), and the strong “wash back” effect of the Korean college entrance exam that did not assess productive skills of English (Choi, 2008; Shin, 2012). Although participants were familiar with reading tasks from their previous English instruction, they were not as familiar with free recall or open-ended questions as with multiple choice and short answer questions. Regarding their computer literacy, these students were skillful and comfortable with reading texts on screen, a mode of reading that had been a part of learning activities in regular content classes from elementary grades on.

Materials

Reading passages

All texts were excerpted from articles found on Internet news websites such as BBC (http://www.bbc.com/), Korea Herald (http://www.koreaherald.com/), HITC sport (http://hereisthecity.com/en-gb/sport/), and Mail online (http://www.dailymail.co.uk/news/). These news articles were selected because they addressed issues deemed controversial in South Korea at the time of the study, and they had evoked diverse emotions among Korean high school students in a pilot study. The topics included (a) a dispute between South Korea and Japan about an island, (b) a comparison between a Korean and a Japanese soccer player in the British Premier league, (c) the potential harm of social networking websites for children’s brain development, and (d) plagiarism in Korean education. These texts were expected to evoke topic-related emotions among our participants when they were read in their emotional version (see Appendix A for a sample reading text, the island dispute between South Korea and Japan).

Beginning with these English-language authentic articles about potentially emotionally involving articles, we modified them to be equivalent in length (253–257 words), lexical density, and syntactic complexity as measured using a web-based text analyzer (Ai & Lu, 2010; Lu, 2010). Guided by Gaskins (1996), we created the neutral version of each text by changing only the names of locations, people, and years so as to distance the text from the students. Once the emotional and neutral versions of each text were developed, the texts were translated into Korean and assessed by two high school language arts teachers for Korean accuracy and fluency. There were some readability differences between the English and Korean versions: the English versions ranged from 10th to 12th grade for English-language learners but the Korean versions ranged from 8th to 10th grade for Korean students reading in Korean, their native language.

Prior knowledge assessment

Guided by previous research (Fortner & Henk, 1990; Gaskins, 1996; Reutzel & Hollingsworth, 1991), we measured self-perceived prior knowledge of the four topics using 5-point Likert scales. In this prior knowledge assessment, students were asked about eight topics in total, with the four target topics intermixed among four other distraction topics. A sample item asked the student to rate the degree to which he or she agreed with the statement: “I can explain the Dokdo Island dispute with some specific details.”

Emotional involvement assessment

Guided by Pekrun et al. (2002), we used four categories of emotions to measure emotional involvement while reading: positive activating (pride, hope, and excitement), positive deactivating (gratitude, satisfaction, and relief), negative activating (anger, worry, and shame), and negative deactivating (boredom, sadness, and disappointment). The items were mixed and presented without category labels. Immediately after reading each text, students indicated the degree to which they felt each emotion during reading, again on 5-point scales.

Comprehension assessment

Guided by Kintsch (1988, 2004) and Zwaan and Radvansky (1998), we used two sets of questions to measure students’ comprehension of the texts: five true or false questions to measure their text-base level of understanding and one open-ended question to measure their situation model level of understanding. The five text-base model questions consisted of three to four microstructure statements such as “Korea stationed a coast-guard detachment on the island,” and one to two macrostructure questions as in, “Korea insists on their sovereignty over the island for historical reasons.” The open-ended questions asked students to use what they had learned from the text to solve a given problem, such as, “Complete the following debate between Korea and Japan based on what you read in the text.” The students were then given sentence starters that began the debate between Korea and Japan, with open boxes inviting their responses.

Procedures

Data were gathered in a computer lab, with materials delivered by computer, during regular English classes. Participants within the same English achievement level, as determined by their semester grades, were randomly assigned to one of four groups. In the first session, students responded to the demographic questions, including the prior knowledge questions. One to three days later, participants read four texts on four different topics: two topics were read in Korean and two topics were read in English; also, two topics were read in the emotional version (one in English and one in Korean) and the other two topics were read in their neutral version (in English and Korean). Every student experienced all four text conditions in counterbalanced order, and across students, all four topics appeared approximately the same number of times in all four orders. Table 1 shows the means of each group in Korean and English achievement level and summarizes how language and text versions were distributed across students.

Table 1.

Means (SD) for Students’ Korean and English Achievement Scores and Distribution of Text Versions (Language × Emotion) Across Groups of Participants.

	Group A (n = 134)	Group B (n = 108)	Group C (n = 121)	Group D (n = 114)
Language achievements scores
Korean	5.08 (1.97)	4.84 (1.80)	5.16 (1.90)	4.98 (1.94)
English	5.10 (1.87)	4.92 (1.91)	5.10 (1.82)	4.92 (1.87)
Text Version × Language Assignment
Soccer players	English/emotional	Korean/neutral	English/neutral	Korean/emotion
Social networking	Korean/neutral	English/neutral	Korean/emotional	English/emotional
Island dispute	English/neutral	Korean/emotional	English/emotional	Korean/neutral
Plagiarism	Korean/emotional	English /emotional	Korean/neutral	English/neutral

Note. Group A attracted more participants accidentally because it was located at the top of the menu page, and some participants who had been assigned to a different group carelessly clicked on it. Each group had equivalent Korean and English achievement levels as shown by one-way analyses of vairance, F(3,456) = .60, p = .61 for Korean, and F(3,456) = .33, p = .81 for English.

Students were told they could read the text at their own pace and that there would be some comprehension questions to check their understanding. After reading the first text, participants were asked to rate the degree to which they felt each of 12 emotions (see Pekrun, Goetz, Titz, & Perry, 2002). After indicating their emotional response, they answered the five text-base model questions and the one open-ended question assessing their situation model understanding. They then went on to read and respond to questions about each of the other three texts in turn.

All measures, including the prior knowledge questionnaire and the comprehension tests, were presented in Korean, the participants’ first language.

Data Analysis

Emotion ratings

The students’ emotion ratings were grouped and averaged into the four categories of emotions suggested by Pekrun et al. (2002): positive activating (excitement, hope, and pride), negative activating (anger, worry, and shame), positive deactivating (gratitude, relief, and satisfaction), and negative deactivating (boredom, sadness, and disappointment) emotions.

Comprehension coding

Students’ responses to the text-base model questions were totaled, resulting in scores ranging from 0 to 5. Compared to the simplicity of scoring the text-base model questions, scoring situation model understanding as measured by the open-ended questions involved a long process of trial and error to produce a final scoring rubric. We first created a scoring rubric based on the propositions in the text, guided by Kintsch (1988, 2004). Before attempting to score all protocols against our rubric, we tested its viability by scoring a random sample of 30 responses. We found that sometimes we had to condense propositions and other times we had to separate them. In addition, we continued to have discussions to distinguish inferences and elaborations based on text-based information from those based on background knowledge only, eventually settling on taking such statements based solely on background knowledge to construct a separate measure, background knowledge intrusions.

Two raters who were blind to text conditions coded the open-ended responses of all protocols into two categories of propositions: situation model propositions (idea units directly stated in the text or closely inferable from the text) and background knowledge intrusions (idea units reflecting background knowledge that extended far beyond the text). An interrater reliability analysis using intraclass correlation was performed to determine consistency among raters (Multon, 2010) and revealed reliabilities of .89 on average across the four topics, ranging from .84 for the social networking text to .93 for the soccer text.

Results

We present our results in two sections. First, we report on participants’ emotional responses to the texts, presenting two sets of analyses to determine whether students experienced different emotions depending on whether they were reading in Korean (their native language) or in English (their foreign language) and depending on whether the text was written in its emotional version or not. Second, we examine the effect of text version (emotional vs. neutral) and language on comprehension measures using a within-subject approach. Here, a first analysis ignored topic, and a second disaggregated the data topic by topic to see whether the effects of text version and language held for all four topics, with prior topic knowledge used as a covariate.

Emotional Responses to Texts

We began by establishing, as a manipulation check, that students had experienced more emotions when reading the emotional text versions (see Table 2 for means and standard deviations for all measures). A repeated-measures 2-way multivariate analysis of variance (MANOVA) with four dependent variables (the four categories of emotions) showed that there was a significant main effect of text version (emotional vs. neutral) and language (Korean vs. English), but no significant interaction between text version and language. Univariate F-tests for each emotion category showed that the effect of text version was strongest for the negative activating emotions, p < .001, η_p ² = 1.24, moderate for both positive activating and negative deactivating emotions, p < .001, η_p ² = .03, and p < .001, η_p ² = .02, respectively, and was not significant with positive deactivating emotions as the dependent variable, p = .093. The effect of language was only significant for negative activating emotions, p = .002, η_p ² = .02.

Table 2.

Means (SDs) for Different Types of Emotions Following Reading Emotional and Neutral Texts in Korean and English.

Text Language	Text Version	Negative Activating	Negative Deactivating	Positive Activating	Positive Deactivating
Korean	Emotional	8.47 (3.0)	7.63 (2.4)	7.54 (2.7)	6.74 (2.7)
Korean	Neutral	7.40 (2.6)	7.30 (2.3)	7.15 (2.7)	6.55 (2.6)
English	Emotional	7.96 (2.9)	7.63 (2.5)	7.40 (2.8)	6.82 (2.7)
English	Neutral	7.31 (2.6)	7.46 (2.5)	7.09 (2.7)	6.76 (2.6)

Note. These means are totals across three emotions on 5-point rating scales, and thus the minimum and maximum values are 3–15. There was a significant text version effect, F(4,473) = 24.49, p < .001, η_p ² = 1.72, and language effect, F(4,473) = 9.11, p < .001, η_p ² = .07, and no interaction, F(4,473) = 0.43, p = .84.

Next, we tested whether topic made a difference in the emotional response of our participants, now testing topic by topic the two independent variables of text version and language. Here, we used MANOVA with the four emotion categories as dependent variables. Results were similar to the within-subject overall analysis only for the island text, with text version effects for all categories of emotions this time (positive activating emotions, p < .001, η_p ² = .05; positive deactivating emotions, p < .05, η_p ² = .01; negative activating emotions, p < .001, η_p ² = .19; and negative deactivating emotions, p < .001, η_p ² = .03), and a language effect only for the negative deactivating emotions, p < .05, η_p ² = .01 (see Table 3). For the other three topics, there was a language effect only for the negative activating emotions (soccer: p < .05, η_p ² = .01; social networking: p < .05, η_p ² = .02; and plagiarism: p < .05, η_p ² = .02), with an additional emotion category for the soccer text, negative deactivating emotions, p < .05, η_p ² = .01. Somewhat surprisingly, text version only had an effect for one of the emotion categories of the plagiarism topic, the negative activating emotions, p < .001, η_p ² = .03. Finally, the only topic that showed a Language by Text Version interaction effect was the social networking text, with the Korean text eliciting higher emotions than the English text in the emotion version and no such language effect evident in the neutral version. This two-way interaction was significant for positive activating p < .05, η_p ² = .02, positive deactivating p < .05, η_p ² = .02, and negative activating p < .05, η_p ² = .01 emotion categories but not for negative deactivating emotions.

Table 3.

Means (SDs) for Different Types of Emotions Following Reading Emotional and Neutral Texts in Korean and English Korean and English for the Island Text.

Text Language	Text Version	Negative Activating	Negative Deactivating	Positive Activating	Positive Deactivating
Korean	Emotional	3.48 (0.9)	2.73 (0.8)	2.64 (1.0)	2.21 (1.0)
Korean	Neutral	2.51 (0.9)	2.42 (0.8)	2.18 (0.8)	1.97 (0.8)
English	Emotional	3.26 (0.8)	2.85 (0.7)	2.54 (0.8)	2.25 (0.8)
English	Neutral	2.56 (0.9)	2.62 (0.8)	2.21 (0.8)	2.15 (0.8)

Note. There were a significant text version effect, F(4,470) = 33.27, p < .001, η_p ² = .22, and language effect, F(4,470) = 5.69, p < .001, η_p ² = .05, and no interaction F(4,470) = .98, p = .42.

In sum, these results suggest that, when ignoring topic and run as a within-subject analysis, students were more emotionally aroused when they read the emotional versions of texts, as shown by a main effect of text version averaged across language versions. There was also a language main effect, but it occurred only for the measure of negative activating emotions (anger, shame, and worry) and showed that the intensity of negative activating emotions was higher for the Korean than English versions. However, topic made a difference in these results: only the island text generally echoed the findings from the overall within-subject analysis. The soccer text only showed a language effect for two of the emotion categories, and the other two texts for only one of the emotion categories. There was no text version effect for the soccer and social networking texts, although the latter had a significant interaction effect with language. For the plagiarism text, there was a text version effect but only for the negative activating emotions.

Effects of Emotions on Comprehension

Overall effects, ignoring topic

A repeated-measures MANOVA with three dependent comprehension measures (text-base model scores, situation model scores, and background intrusion scores) showed that there were significant main effects of text version and language, but no significant interaction between text version and language (see Table 4 for means and standard deviations). As a follow-up, we conducted analyses of variance (ANOVAs) with each comprehension measure.

Table 4.

Means (SDs) for Comprehension Measures for Emotional and Neutral Texts in Korean and English.

Text Language	Text Version	Text-Base Model Scores^a	Situation Model Scores^b	Background Knowledge Intrusions
Korean	Emotional	3.96 (1.0)	2.43 (2.6)	.82 (1.6)
Korean	Neutral	3.75 (1.1)	2.11 (2.3)	.70 (1.4)
English	Emotional	3.43 (1.1)	1.33 (1.8)	.84 (1.3)
English	Neutral	3.11 (1.2)	1.25 (1.7)	.57 (1.1)

Note. There was a significant text version effect, F(3,474) = 17.33, p < .001, η_p ² = .10, and language effect, F(3,474) = 94.48, p < .001, η_p ² = .37, and no interaction, F(3,474) = 1.61, p = .185.

^aBased on students’ answers to 5 true–false statements reflecting the text.

^bBased on coding of students’ responses to one open-ended essay question.

For text-base model scores, the univariate F-test resulted in significant main effects for text version, p < .001, η_p ² = .06, and language, p < .001, η_p ² = .23. Students received higher scores when they read emotional than neutral texts, and this was true in both languages. As for language, students earned higher scores with Korean than English texts, and this was true whether they read emotional or neutral versions. These results suggest that students were better at text-base model comprehension when reading emotional than neutral texts and when reading Korean rather than English texts.

Next, a univariate F-test with situation model scores resulted in significant main effects for text version, p = .014, η_p ² = .01, and language, p < .001, η_p ² = .25. Students had higher comprehension scores for the situation model assessment when they read emotional than neutral texts. Students also showed higher comprehension when they read Korean than English texts, and this was true for both emotional and neutral versions. There was a suggestion that the effect of emotional texts on situation model scores decreased when they read texts in English.

Lastly, a univariate F-test of background intrusion scores resulted only in a main effect of text version, p < .001, η_p ² = .02. Students made more insertions of background knowledge propositions in their recalls when they read emotional texts than neutral texts, with a suggestion that this occurred more when reading the English than Korean texts, though not significant.

Topic effects on the comprehension measures

We ran four initial multivariate analyses of covariate (MANCOVAs; one per topic) with prior knowledge as a covariate to examine the effect of language and text version on the three comprehension measures: text-base and situation model scores and background knowledge intrusions. For the soccer and social networking texts, there was no interaction between factors and the covariate, thereby meeting the MANCOVA assumptions. However, for the island and plagiarism texts, an interaction between text version and prior knowledge was found for the background knowledge intrusion measure, which violates the assumption of MANCOVA. For these two topics, we ran MANCOVAs with only two comprehension variables and used regression analysis for the background knowledge intrusion measure.

For the island text, results of the MANCOVA generally replicated the findings from the overall analysis. There were significant main effects for language and text versions (p < .001, η_p ² = .11 and p < .001, η_p ² = .09, respectively) that were still significant when the follow-up univariate F-tests were run (language: p < .001, η_p ² = .09 and p < .001, η_p ² = .05 for text-base and situation model scores, respectively; text version: p < .001, η_p ² = .08 and p < .001, η_p ² = .04 for text-base and situation model scores, respectively; see Table 5 for means and standard deviations). Scores were higher when students read the Korean rather than the English texts and also higher when they read the emotional rather than neutral version. There was also an interaction between language and text version for situation model scores, p < .05, η_p ² = .02. Post hoc tests showed that students earned higher situation model scores when reading the emotional texts than neutral texts but the difference was significant only with Korean texts.

Table 5.

Means (SDs) for Comprehension Measures for the Island Text.

Text Language	Text Version	Text-Base Model Scores^a	Situation Model Scores^b	Background Knowledge Intrusions
Korean	Emotional	3.89 (0.9)	3.30 (3.4)	.60 (1.4)
Korean	Neutral	3.46 (1.0)	1.81 (2.1)	.04 (0.2)
English	Emotional	3.41 (0.8)	1.63 (1.9)	.65 (1.1)
English	Neutral	2.74 (1.1)	1.30 (1.9)	.13 (0.9)

Note. Multivariate analysis of covariance with two comprehension measures, text-base and situation model score, indicated a significant main effect of language and text version, F(2,471) = 28.18, p < .001, η_p ² = .11 and F(2,471) = 22.93, p < .001, η_p ² = .09, respectively, and a significant interaction between language and text version, F(2,471) = 5.33, p < .05, η_p ² = .02. The regression model with background knowledge intrusions as dependent variable was significant, R ² = .10, F(4,472) = 13.22, p < .001, showed that text version and the interaction between text version and prior knowledge were significant predictors of background knowledge intrusions, β = .54, t(472) = 6.48, p < .001, and β = .24, t(472) = 2.81, p < .05, respectively.

Recall that the covariate of prior knowledge interacted with the independent variables for the background intrusion measure. Here, the recommended approach is to use regression analysis. Results examining the effects of language, text version, and prior knowledge revealed that text version and the interaction between text version and prior knowledge were significant predictors of background knowledge intrusions, b = .54, t(472) = 6.48, p < .001 and b = .24, t(472) = 2.81, p < .05, respectively. Thus, text version was a significant predictor only for students with prior knowledge above 2.3 (2.3–5), indicating that there were more intrusions for the emotional version only if students had enough background knowledge.

Results were similar for the soccer text in that the initial MANCOVA, with all three dependent measures, indicating significant main effects for language and text version: p < .001, η_p ² = .16, and p < .001, η_p ² = .05, and a significant interaction between language and text version, p < .05, η_p ² = .03. Follow-up univariate tests showed that the effect of language appeared for all comprehension measures, p < .001, η_p ² = .10 for text-base model, p < .001, η_p ² = .12 for situation model, and p < .05, η_p ² = .01 for background knowledge intrusions, respectively. Post hoc comparisons showed that students scored significantly higher with Korean texts than English texts on all three measures. The univariate tests for the text version variable was significant for text-base model and background knowledge intrusion scores, p < .001, η_p ² = .04 and p < .05, η_p ² = .01, respectively, with post hoc comparisons showing that emotional texts led to higher scores than neutral texts. As for the interaction between factors, univariate tests showed that it appeared significant for background knowledge intrusions, p < .05, η_p ² = .02, with post hoc comparisons indicating that students made significantly more background knowledge intrusions for emotional texts only when they read English texts (only when they were having some linguistic difficulty).

For the other two topics, results were less interesting, with only language as a significant main effect in the initial MANCOVA (plagiarism: for text-base model and situation model scores; social networking sites: for all three dependent measures). Follow-up univariate F-tests with each dependent measure for both topics indicated that scores were higher for the Korean version than for the English version. The background knowledge intrusion measure for the plagiarism topic did show an interesting connection to the prior knowledge covariate, with the interaction between text version and prior knowledge a significant predictor, b = .48, t(472) = 2.95, p < .05. Specifically, text version was a significant negative predictor for students with low levels of prior knowledge (0–3), whereas it was a significant positive predictor for students with high level of prior knowledge above 4.

In sum, results suggested that secondary students reading expository texts in both their native language and a learned language and in both an emotional and neutral version comprehended the Korean versions better than the English versions and performed better on the comprehension measures for the emotional rather than the neutral versions. Although students’ emotional responses and comprehension performance were stronger when reading a text in their first rather than their second language, the contrast between emotional and neutral versions of the text in the second language was less than when reading in the native language but nevertheless still evident. When performance was examined topic by topic, results for the island and soccer topics were generally similar to the overall results. The other two topics only showed a language effect, thus serving as a reminder to attend to topic differences, especially when investigating emotional response to text reading and comprehension performance.

Discussion and Study Significance

Like the students in the Gaskins (1996) study, the Korean adolescent readers in our study responded with stronger emotions to text versions that were written to arouse culturally relevant emotional scripts when compared to more neutral texts that placed some distance between the young reader and the topic. For example, when they were reading a text that addressed the island dispute between Japan and Korea, students’ emotions were more intense than when they read an account about an island dispute between Spain and Algeria for which these young readers felt little connection. Thus, when they were asked in an open-ended question to complete a debate between two countries about sovereignty over the island, participants who read the emotional texts constructed debate points preferential to Korea and showing some hostility to Japan. By contrast, those who read about the island dispute between Spain and Algeria showed an impartial, even nonchalant, attitude toward the issue and responded to the debate question in a detached manner, suggesting less ego involvement and less interest in the neutral version.

It was noteworthy that the personal engagement encouraged by the emotional versions of the texts led to better comprehension, suggesting that emotional involvement helped these readers with their comprehension. This result stands somewhat in contrast to previous research with expository texts showing that emotional involvement decreased comprehension as measured by delayed recall of the text (Reutzel & Hollingsworth, 1991), text interpretation (Gaskins, 1996), and answers to open-ended questions (Fortner & Henk, 1990). However, this contradictory result may be connected to the fact that we separated background knowledge intrusions from our situation model assessments. Thus, the negative effect of emotional involvement in previous research may originate in background knowledge intrusions. Like Fortner and Henk (1990) and Gaskins (1996), our results showed that students relied on their background knowledge when they read the emotional version, but in our case, the elicited emotional involvement seemed to lead to greater engagement and better comprehension, even though it also was accompanied by more background knowledge intrusions. In fact, emotional involvement seemed to rely on prior knowledge as indicated by the fact that for some of the topics, low background knowledge led to a negative effect of emotions on comprehension, not a positive one.

In second language reading, the contrast between emotional and neutral versions seemed modulated somehow. We attribute this finding, at least in part, to the linguistic difficulty of the text as suggested by significantly higher degrees of negative deactivating emotions (sadness, boredom, and disappointment) reported with English texts. In addition, the emotionality of text construction when reading in a second language as predicted by Pavlenko (2007) and Dewaele (2010) seemed to indicate that reading in a second language is accompanied by less emotional resonance, perhaps due to a different socioeducational history for second language learners. Our high school participants had learned their second language in classroom environments with little experience of full immersive socialization. Thus, they may have treated English texts as simply another classroom activity, with little connection to issues they should care about.

Recall that the emotional and neutral versions of each text were identical in structure and wording except for names of places or individuals or years. Nevertheless, our second language readers responded differently not only in their emotional involvement but also in their comprehension. This result implies that cultural elements as well as linguistic features of the text affect how readers engage with the text as argued by Jimenez, García, and Pearson (1996). Our adolescent second language readers comprehended the emotional version better, not because it was linguistically easier than the neutral version, but because they could make meaning out of the text by engaging with their culturally situated emotions. Even if the emotional involvement learners experience or the sense of frustration at difficulties arising from lack of proficiency need some careful guidance by a teacher, emotional engagement should be supported in second language reading. In response to a call by Benesch (2012) and Pavlenko (2007) for second language teachers to help young people see their endeavors as more than a cold intellectual enterprise, our study suggests the benefits of including texts on issues that engage students.

Thus, not surprisingly, we also found that the topic of the text made a difference in whether these young people could become emotionally involved in what they were reading and could comprehend the text better. Our results from the topic-specific MANCOVA showed that the positive effect of the emotional text on comprehension appeared most strongly with the island topic, which may have triggered socially constructed emotions (Gee, 2000, 2001; Trainor, 2008) among our participants and captured their attention inherently. A similar pattern occurred with the soccer text but with reduced effects, probably because the issues in the soccer text did not appeal as much to female students as to their male counterparts. For the other two topics, we could not see a significant effect of emotional texts on comprehension. This result is consistent with what Ainley et al. (2005) noted, that the effect of affective involvement varies across topics. The variance in emotional involvement across topics also implies that emotions experienced by our young readers were not only individually generated but also culturally constructed. The two texts that aroused these adolescents emotionally, the island dispute and the debate about which soccer player was better, may have been more central and important to their Korean identity, whereas the social networking and plagiarism texts were not only less emotionally arousing but also showed no significant difference between emotion and neutral versions on the comprehension measures.

From a socioconstructivist view of reading, our participants responded to the texts from within their sociocultural perspectives, whether they were reading in their native language or in a newly learned foreign language. As Gee (2001) suggested, these students made situated meanings in response to the text, drawing on their prior knowledge and emotional response. The main effect of emotional texts on emotional response and comprehension scores demonstrated that affective responses were culturally constructed and evoked. As Gee (2000) noted, reading is not a neutral process; it is always a perspectival process that draws on not only cognitive resources such as prior knowledge but also affective resources such as beliefs and values.

Readers employ all resources in their reading, opportunistically making use of cognitive and affective resources without prejudice, relying on both languages as resources and prior knowledge as needed. As a resource, emotions fall in a special category as they are sometimes associated with experiences of failure and sometimes associated with greater involvement and better comprehension. When reading in a second language, such emotional energy may as much derail effort at comprehension if anxiety is high and self-efficacy low, as it may fuel a commitment to keep working with the text until a sense of comprehension is achieved. Our study, of course, faced limitations such as a lack of an emic perspective from the participants that would have enhanced our understanding of the motivational qualities of emotions and the fact that we only used one open-ended question to assess their situation model understanding. We are nevertheless encouraged by the clear endorsement from our results of the need to attend to the emotional processes that infuse reading comprehension of both first and second language texts.

Footnotes

Appendix A

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Abu-Rabia

(1998). Social and cognitive factors influencing the reading comprehension of Arab students learning Hebrew as a second language in Israel. Journal of Research in Reading, 21, 201–212.

(2010, 6). A web-based system for automatic measurement of lexical complexity. Paper presented at the 27th Annual Symposium of the Computer-Assisted Language Consortium (CALICO-10), Amherst, MA.

Ainley

Corrigan

Richardson

(2005). Students, tasks and emotions: Identifying the contribution of emotions to students’ reading of popular culture and popular science texts. Learning and Instruction, 15, 433–447.

Baldwin

R. S.

Peleg-Bruckner

McClintock

A. H.

(1985). Effects of topic interest and prior knowledge on reading comprehension. Reading Research Quarterly, 20, 497–504.

Benesch

(2012). Considering emotions in critical English language teaching: Theories and praxis. New York, NY: Routledge.

Bernhardt

E. B.

(1991). Reading development in a second language: Theoretical, empirical, and classroom perspectives (Vol. 1). Norwood, NJ: Ablex Publishing Corporation.

Bernhardt

E. B.

(2011). Understanding advanced second-language reading. New York, NY: Routledge.

Brantmeier

(2005). Anxiety about L2 reading or L2 reading tasks? A study with advanced language learners. The Reading Matrix, 5, 67–85.

Carrell

P. L.

(1983). Three components of background knowledge in reading comprehension. Language Learning, 33, 183–203.

10.

Choi

I. C.

(2008). The impact of EFL testing on EFL education in Korea. Language Testing, 25, 39–62.

11.

Dewaele

J. M.

(2010). Emotions in multiple languages. London, England: Palgrave McMillan.

12.

Dijkstra

Zwaan

R. A.

Graesser

A. C.

Magliano

J. P.

(1994). Character and reader emotions in literary texts. Poetics, 23, 139–157.

13.

S. L.

Schallert

D. L.

(2004). Emotions and classroom talk: Toward a model of the role of affect in students’ experiences of classroom discussions. Journal of Educational Psychology, 96, 619–634.

14.

Dutro

(2008). “That’s why I was crying on this book”: Trauma as testimony in responses to literature. Changing English, 15, 423–434.

15.

Elder

Golombek

Stott

(2004). Familiarity breeds contempt: Reading texts from learners’ own cultures does not guarantee recall. TESOL Quarterly, 38, 345–352.

16.

Fortner

B. H.

Henk

W. A.

(1990). Effects of issue-related attitude on readers’ comprehension and judgments of unbiased text. Reading Research and Instruction, 30(2), 1–16.

17.

Gaskins

R. W.

(1996). “That’s just how it was”: The effect of issue-related emotional involvement on reading comprehension. Reading Research Quarterly, 31, 386–405.

18.

Gee

J. P.

(2000). Discourse and sociocultural studies in reading. In Pearson

P. D.

Barr

Kamil

M. L.

(Eds.), Handbook of reading research (Vol. 3, pp. 195–207). Hillsdale, NJ: Erlbaum.

19.

Gee

J. P.

(2001). Reading as situated language: A sociocognitive perspective. Journal of Adolescent and Adult Literacy, 44, 714–725.

20.

Goetz

E. T.

Sadoski

Stowe

M. L.

Fetsco

T. G.

Kemp

S. G.

(1993). Imagery and emotional response in reading literary text: Quantitative and qualitative analyses. Poetics, 22, 35–49.

21.

Grabe

(2009). Reading in a second language: Moving from theory to practice. Cambridge, England: Ernst Klett Sprachen.

22.

Henk

W. A.

Holmes

B. C.

(1988). Effects of content-related attitude on the comprehension and retention of expository text. Reading Psychology: An International Quarterly, 9, 203–225.

23.

Jimenez

R. T.

García

G. E.

Pearson

P. D.

(1996). The reading strategies of bilingual latina/o students who are successful English readers: Opportunities and obstacles. Reading Research Quarterly, 31, 90–112.

24.

King

R. B.

(2010). What do students feel in school and how do we measure them?: Examining the psychometric properties of the S-AEQ-F. Philippine Journal of Psychology, 43, 161–176.

25.

Kintsch

(1988). The role of knowledge in discourse comprehension: A construction-integration model. Psychological Review, 95, 163–182.

26.

Kintsch

(2004). The construction-integration model of text comprehension and its implications for instruction. Theoretical Models and Processes of Reading, 5, 1270–1328.

27.

Kneepkens

E. W.

Zwaan

R. A.

(1994). Emotions and literary text comprehension. Poetics, 23, 125–138.

28.

Komeda

Kusumi

(2006). The effect of a protagonist’s emotional shift on situation model construction. Memory & Cognition, 34, 1548–1556.

29.

Lee

J. F.

(1986). Background knowledge and L2 reading. The Modern Language Journal, 70, 350–354.

30.

Lee

J. W.

Schallert

D. L.

(1997). The relative contribution of L2 language proficiency and L1 reading ability to L2 reading performance: A test of the threshold hypothesis in an EFL context. TESOL Quarterly, 31, 713–739.

31.

Linnenbrink-Garcia

Rogat

T. K.

Koskey

K. L.

(2011). Affect and engagement during small group instruction. Contemporary Educational Psychology, 36, 13–24.

32.

(2010). Automatic analysis of syntactic complexity in second language writing. International Journal of Corpus Linguistics, 15, 474–496.

33.

Lysaker

J. T.

Tonge

Gauson

Miller

(2011). Reading and social imagination: What relationally oriented reading instruction can do for children. Reading Psychology, 32, 520–566.

34.

Miall

D. S.

Kuiken

(1994). Foregrounding, defamiliarization, and affect: Response to literary stories. Poetics, 22, 389–407.

35.

Miall

D. S.

Kuiken

(1999). What is literariness? Three components of literary reading. Discourse Processes, 28, 121–138.

36.

Miall

D. S.

Kuiken

(2002). A feeling for fiction: Becoming what we behold. Poetics, 30, 221–241.

37.

Multon . (2010). Interrater reliability. In Salkind

N. J.

(Ed.), Encyclopedia of research design (pp. 627–629). Thousand Oaks, CA: Sage.

38.

Nassaji

(2007). Schema theory and knowledge-based processes in second language reading comprehension: A need for alternative perspectives. Language Learning, 57, 79–113.

39.

Oatley

(1999). Meetings of minds: Dialogue, sympathy, and identification, in reading fiction. Poetics, 26, 439–454.

40.

Pavlenko

(2007). Emotions and multilingualism. Cambridge, England: Cambridge University Press.

41.

Pekrun

(2000). A social-cognitive, control-value theory of achievement emotions. In Heckhausen

(Ed.), Motivational psychology of human development (pp. 143–163). Oxford, England: Elsevier.

42.

Pekrun

Elliot

A. J.

Maier

M. A.

(2009). Achievement goals and achievement emotions: Testing a model of their joint relations with academic performance. Journal of Educational Psychology, 101, 115–135.

43.

Pekrun

Goetz

Titz

Perry

R. P.

(2002). Academic emotions in students’ self-regulated learning and achievement: A program of qualitative and quantitative research. Educational Psychologist, 37, 91–105.

44.

Reutzel

D. R.

Hollingsworth

P. M.

(1991). Investigating topic-related attitude: Effect on reading and remembering text. The Journal of Educational Research, 84, 334–344.

45.

Rosenblatt

(1978). The reader, the text, the poem. Carbondale: Southern Illinois University Press.

46.

Sadoski

Goetz

E. T.

Kangiser

(1988). Imagination in story response: Relationships between imagery, affect, and structural importance. Reading Research Quarterly, 23, 320–336.

47.

Sadoski

Goetz

E. T.

Olivarez

Lee

Roberts

N. M.

(1990). Imagination in story reading: The role of imagery, verbal recall, story analysis, and processing levels. Journal of Literacy Research, 22, 55–70.

48.

Saito

Garza

T. J.

Horwitz

E. K.

(1999). Foreign language reading anxiety. The Modern Language Journal, 83, 202–218.

49.

Sellers

V. D.

(2000). Anxiety and reading comprehension in Spanish as a foreign language. Foreign Language Annals, 33, 512–520.

50.

Shin

S. K.

(2012). “It cannot be done alone”: The socialization of novice English teachers in South Korea. TESOL Quarterly, 46, 542–567.

51.

Tompkins

J. P.

(Ed.). (1980). Reader-response criticism: From formalism to post-structuralism. Baltimore, MD: The Johns Hopkins University Press.

52.

Trainor

J. S.

(2008). The emotioned power of racism: An ethnographic portrait of an all-white high school. College Composition and Communication, 60, 82–112.

53.

Villavicencio

F. T.

Bernardo

A. B. I.

(2013). Positive academic emotions moderate the relationship between self-regulation and academic achievement. British Journal of Educational Psychology, 83, 329–340.

54.

Zwaan

R. A.

Dijkstra

Graesser

A. C.

(1993, 6). Components of understanding literary narrative. Paper presented at the Third Annual Meeting of the Society for Text and Discourse, Boulder, CO.

55.

Zwaan

R. A.

Radvansky

G. A.

(1998). Situation models in language comprehension and memory. Psychological Bulletin, 123, 162–185.