Effects of Literary-Specific Knowledge on the Generation of Interpretive Inferences During Reading

:speech_balloon: Speaker: Dr. Kathryn McCarthy

:classical_building: Affiliation: Georgia State University
Title: Effects of Literary-Specific Knowledge on the Generation of Interpretive Inferences During Reading

Abstract (long version below): In two studies, we examined how providing information about how experts use literary-specific knowledge of conventions and themes to make sense of a short story. Specifically, we were interested in how this knowledge would influence readers’ generation of interpretive inferencing about the deeper meaning(s) of the work. Participants were asked to think-aloud as they read and to rate their appreciation of the story after reading. In Study 1, we examined how different types of knowledge (conventions, theme, both) influenced interpretive inferencing and appreciation. In Study 2, we examined how allowing readers to reread the text might augment these effects.

:movie_camera: [example to be replaced]

:newspaper: Long abstract

One goal of reading literary works is to go beyond the page to gain a deeper understanding of our own world and experiences. This literary mode of reading invites the generation of interpretive inferences. Prior research suggests that literary experts readily produce interpretive inferences and attend to the stylistic variations to support their interpretations. By contrast, novice readers often struggle to go beyond a surface, literal understanding of the text (e.g., Claassen, 2012; Graves & Frederiksen, 1991; Peskin, 1998). In prior work, researchers have demonstrated that providing novice literary readers with discipline-specific knowledge about literary text processing led to more evidence of interpretation in post-reading essays (e.g., Authors, 2015, 2019). The current work extends this research to examine the extent to which providing literary-specific knowledge to novice readers affects the comprehension processes that occur during reading.

Study 1
Procedure. Participants (n = 31) were randomly assigned to one of four pre-reading instructions. The instructions were developed in previous research (Authors, 2019). These instructions were built based on expert readings of the target story, The Elephant. In the Conventions instruction condition, readers were given information about the literary devices (juxtaposition, deviation from the norm, disruptions) in the text. In the Theme instruction, readers were given information about the satiric nature of the story. In the Combined condition, participants received both the convention and theme information. Finally, a Control condition was given only the title to serve as a baseline. After reading, participants completed a three-item appreciation measure (Dixon et al., 1993).

Results. The think-alouds were coded for evidence of interpretive inferences relative to more textbase processing (e.g., paraphrasing, story-world inferences). Analyses revealed no effect of the instruction manipulation on the amount of interpretive inferences. However, there was a modest effect of instruction on the appreciation measure, such that those in the control condition showed less appreciation for the text than those who had been given any of the three instructions.

Study 2
One possible explanation for the overall lack of interpretive inferences in Study 1 is that readers were devoting their limited cognitive resources to lower-level processes related to constructing a textbase representation (e.g., Millis et al., 1998). Thus, one potential way of supporting novice readers’ ability to leverage the information in the pre-reading instructions might be to allow them to read the text once through prior to thinking aloud. This first pass allows for the reader to construct a strong base upon which they can then elaborate upon with more literary-specific processes, such as attention to specific language and interpreting. Indeed, rereading has been shown to support literary processes and effects (e.g., Dixon et al., 1993; Kuijpers & Hakemulder, 2018).

Procedure. Participants (n = 60) were randomly assigned to a 2(pre-reading instruction: control, combined) x 2(think-aloud: first pass, rereading) between-subjects design. The procedure was the same as Study 1 except that those who were assigned to the rereading condition were given the opportunity to read the text silently to themselves once before being asked to think-aloud.

Results. On average, participants’ think-alouds included less than 1000 words (M = 944.28, SD = 390.71). An ANOVA revealed no significant differences in think-aloud length, although the effect of rereading approached significance, F(1,56) = 3.59, p = 0.06. This pattern suggested that those who had already read the text once silently prior to think-aloud (reread condition) generated slightly longer think-alouds (M = 1039.33, SD = 422.76) than those who were asked to think-aloud on their first pass of the text (M = 849.23, SD = 336.39).

To evaluate the effects of instruction and first pass vs. rereading, we submitted the data to a 2(instruction: control, combined) x 2(think-aloud: first pass, reread) x 3(process: paraphrase, text-based inference, interpretive inference) mixed ANOVA. This omnibus test revealed a significant three-way interaction, F(2,112) = 6.45, p < .01. Follow-up analyses revealed that the first pass readers relied heavily on paraphrasing, regardless of whether or not they received literary-specific pre-reading information. By contrast, those who were given the chance to read silently first and produced their think-aloud upon rereading showed an instruction by process interaction, F(2,56) = 3.93, p = .03, such that those who were given no information about the text relied heavily upon paraphrasing and demonstrated almost no evidence of interpretation. Those who reread and were given the literary-specific knowledge information produced significantly fewer paraphrases and generated slightly more interpretive inferences.

Inconsistent with other work (e.g., Kuijpers & Hakemulder, 2018), there was no effect of rereading on appreciation measures. There was, however, a significant main effect of information condition on appreciation F(1,17.07) = 9.52, p < .01, such that readers given literary rules information prior to reading found the story more enjoyable than those given no pre-reading information.