ABSTRACT. A substantial amount of research has focused on children’s reading development and reading problems but, in comparison, there has been relatively little research into children’s reading comprehension. A large part of my research career has been dedicated to finding out more about children’s reading comprehension: both development and difficulties. The overarching aim of this research has been to explore the skills and cognitive processes that support children’s understanding of text, with the ultimate aim of providing support for the development of, and remediation of, problems with reading comprehension. In this talk, I will reflect on the various hypotheses I have entertained and tested over time and will consider what avenues of investigation still need further exploration. I will end by considering the implications of these findings for helping children to develop and improve their comprehension skills.

ABSTRACT. In this study we examined whether struggling adult readers’ responses to survey questions about punctuality, reading habits, self-efficacy, and persistence are related to their attendance and reading progress in adult literacy classes. Differences and interactions based on native/non-native English-speaking status were also explored. Due to the prevalence of attendance difficulties, educational gain struggles and heterogeneity of students our findings have instructional implications which will be discussed as part of the presentation.

ABSTRACT. AutoTutor is a learning environment on the web with conversational agents who train adults comprehension strategies in 35 lessons targeting different theoretical levels (words, textbase, situation model, and rhetorical structure). Data mining procedures detect disengagement from AutoTutor based on the accuracy and response times to questions the conversational agents ask the adults. The “disengagement detectors” are not only manifested in low accuracy with short response times, but also extremely long times allegedly reflecting mind wandering.

ABSTRACT. Behavioral and self-report measures of motivation rarely agree with each other, whereas individual differences in these two domains do interact and can inform each other. Performance tracking data in AutoTutor, an autonomous learning technology tool, was analyzed to establish per-participant pacing and accuracy patterns across learning opportunities. Against these patterns, we referenced inter-individual patterns of self-reported motivation for reading (perceived difficulty, competence, interest/enjoyment, value).

ABSTRACT. We developed and administered a survey of approximately 200 interest categories to adult learners (N=51) who were randomly assigned to an interest-matching dosage level (low, medium, or high) and given e-readers with reading packs based on their interests and dosage condition. Results indicate that participants in the highest interest-matched group read approximately 30% longer than those in the lowest matched group, regardless of interest matching on a particular text.

ABSTRACT. Extending prior work demonstrating a relationship between ambiguous sentence resolution and creative problem-solving, the current study measured comprehension of two types of ambiguous sentences and compared it to success on both analytical and creative problem-solving tasks. Even after controlling for intelligence, working memory, and unambiguous sentence comprehension, individual differences in ambiguous sentence comprehension predicted creative problem-solving, but not analytic problem-solving. Results suggest a unique relationship between how people resolve ambiguity and solve creative problems.

ABSTRACT. We examined adults’ and children’s pronoun resolution following events with and without inherent endpoints (lexical aspect) when these events were described as temporarily collapsed or in progress (grammatical aspect). Participants read sentences and circled the referent of a subsequent ambiguous pronoun. Subject selections were more likely for events without than with endpoints and when events were expressed as in progress rather than temporarily collapsed. The findings support the Event Structure account of pronoun processing.

ABSTRACT. This study explored the impact of editing on understanding events in films. Participants viewed different versions of a film depicting two characters performing actions. Two versions did not have any editing and reflected single camera shots. One version used editing to emphasize the actions of the two characters. The association between event segmentation and action structure, and the recall of actions in recall protocols, increased as a function of the use of editing.

ABSTRACT. In this study, we examine whether children and adults perceptually simulate vertical object movements while listening to short narratives, and whether accompanying pictures have an impact on perceptual simulation. In our picture verification experiment, response times to matching trials were shorter than to mismatching ones, indicating that perceptual simulation takes place regardless of participants’ age. The presence or absence of pictures did not affect perceptual simulation of auditory text.

ABSTRACT. Electrophysiological activity was recorded while participants generated strongly-constrained or weakly-constrained bridging inferences for targets presented to either the right hemisphere or left hemisphere. Results showed an attenuated N400 effect (i.e., reflecting early, syntactic/semantic processes) in the left hemisphere for strongly-constrained inferences, but no differences in the right hemisphere. These findings suggest that the left hemisphere more easily integrates words into a sentence or discourse context when the word is strongly constrained in a text.

ABSTRACT. This study examined the interactive effects of self-explanation training and constructed response prompt (self-explanation, think-aloud) on students’ misconceptions regarding natural selection. Undergraduate students (n=154) randomly assigned to training and prompt condition read a refutational text regarding natural selection and then completed the Conceptual Inventory of Natural Selection (CINS). There were no main effects or interactions, contradicting previous research on self-explanation and refutational texts. Possible explanations are discussed.

ABSTRACT. Via remedial teaching, the ability of word recognition and low performance of elementary school students in the rural town,Taitung, were increased by grade and the word recognition ability at first grade were indeed left behind the national norm.__The students of G1 to G3 in Taitung didn’t have Matthew effects and the disparity between national norms was scaling down by grade. But the students of G4 to G6 were existed the reverse status exactly.

ABSTRACT. In this study, natural language processing (NLP) tools were utilized to obtain linguistic indices from verbal protocols produced during reading by community college L1 & L2 readers. Next, machine learning models were explored to differentiate L1 & L2 readers. Correlations were found between the resulting predictors of L2 status and measures of reading proficiency, suggesting this approach may have utility in examining differences in L1 and L2 readers’ proficiency and potentially identifying struggling L2 readers.

ABSTRACT. We examined how adults’ pre-existing accurate and inaccurate beliefs (i.e., vaccines are helpful, harmful, and unnecessary) and prior knowledge regarding childhood vaccinations related to post-reading beliefs and intentions to vaccinate their own children one week after reading a set of belief-consistent and belief-inconsistent texts about childhood vaccinations. Findings suggest stability of pre-existing beliefs even after exposure to belief-inconsistent information and that not all vaccination beliefs are equally important in driving vaccination behaviors.

ABSTRACT. In two experiments, stories were narrated by either a reliable or unreliable narrator. Real world facts were embedded in the stories. The facts were accurate, misleading or neutral. After reading the stories, participants took a general knowledge test. Participants were more likely to produce the misinformation answer when it had been provided by a reliable than an unreliable narrator. Readers were less likely to trust information provided by an unreliable narrator.

ABSTRACT. We look at the effects of learning from multiple texts over a two-week delay. Students reported their knowledge and attitudes about a topic at three time points: immediately prior to and following multiple text use and after a two-week delay. The stability of knowledge and attitude was examined over time. While attitude stance remained consistent, attitude strength was found to dissipate, regressing to the mean. Knowledge gains persisted, although relatively little information was retained.

ABSTRACT. We examined the role of kindergartners’ (n = 69) language comprehension skills in their ability to make inferences after watching nonfiction videos of high- and low-cohesion transcripts. Students watched a series of 12 nonfiction videos on various topics and answered multiple-choice inferential questions. Performance of high- and low-language comprehension skill groups on high- and low-cohesion video texts was analyzed to determine for whom and under what conditions students learn best from video texts.

ABSTRACT. This study examines gains in second language writing ability in relation to cognitive and language resources in English as a second language undergraduate students (N = 77). Results indicated a close relationship between initial reading and writing skills. Results also indicated significant gains in writing scores over time. Greater gains in writing scores were related to higher levels of initial working memory capacity and lower levels of initial writing and reading skills.

ABSTRACT. Self-classified poetry-readers and non-readers listened to poems and rated their aesthetic experience. Half the participants were told to focus on sound, whereas the other half were told to focus on imagery. We assessed the relationship between phonemic qualities of poems (e.g., nasal dominance) and aesthetic experience. The results showed attention to sound or imagery may have lead to different relationships between the sounds of poetry and the experience of it.

ABSTRACT. Identification of contradictions between text and graph requires a multitude of inferences, especially when those representations are more complex. The purpose of the current study is to investigate participants’ identification of text/graph contradictions when the graph complexity and type of contradiction varies. Overall, participants were unlikely to identify what was contradictory and were even less likely to correctly state what should be changed to make the text and graph agree. Future research directions are discussed.

ABSTRACT. This study intended to examine the relationships among vocabulary depth, inference and reading comprehension by using structural equation model. Five hundred and eighty-four children at grades 3 to 5 were recruited from four primary schools in Taiwan. The Chinese reading comprehension test, the vocabulary depth test and the inference test were administered to participants. The results revealed that vocabulary depth was the best predictor to reading comprehension across three grade levels.

ABSTRACT. Refutation texts are well examined tools to influence factual knowledge, but little is known about their impact on meta-cognitive and behavioral variables. We conducted two experimental lab studies to examine the impact of a refutation text about misconceptions of scientific tentativeness on readers’ appraisal of journalistic science articles and on their willingness to support the research presented. Our results indicate that refutation texts are a promising method of supporting meta-cognitive understanding and influencing behavior.

ABSTRACT. Deficiency in keyboarding is likely to impede students' writing performance in an assessment. In this study, we were interested in whether keyboarding speed can be reasonably estimated using in-word typing speed recorded via keystroke logs. The results suggest that words with TASA frequency of 60 or above, or words in an expanded most frequent 100 word list, could be used to estimate a writer's typing skill.

ABSTRACT. Though prior knowledge is recognized as one of the largest contributors to comprehension performance, there is little specificity about the different aspects of prior knowledge and how they might differentially impact comprehension. The purpose of this poster is to introduce the Multidimensional Knowledge (MDK) framework to conceptualize dimensions of prior knowledge that contribute to comprehension and to facilitate a larger discussion around how we design and interpret prior knowledge research.

ABSTRACT. We investigated the relation between memory, vocabulary and listening comprehension (LC) of narrative and expository text. Three-hundred-and-eighty-eight children aged 5-6 years completed tasks tapping memory, vocabulary and LC. Path analysis was used to test direct and indirect effects between variables. There were significant direct effects of memory and vocabulary on narrative and expository LC. Additionally, narrative LC had a direct effect on expository LC. Results are discussed regarding the role of narrative for expository comprehension.

ABSTRACT. This study critically examines the construct of reading comprehension assessment as it relates to adolescents' digital literacy skills. We review recent studies that have demonstrated how memory-based and text available assessments measure different components of reading comprehension and how such tasks might not be affected by the same individual variables. Finally, we discuss how a multi-level measurement integrating both text-available and memory-based comprehension tasks can better evaluate the construct of reading comprehension

ABSTRACT. Given the importance of developing students’ competencies on the full complement of NGSS practices, the present study examined whether repeated opportunities to construct scientific explanations during science inquiry would help improve students’ explanation quality. Results showed that the quality of scientific explanations significantly improved from the first and second to the third activity. These findings have implications for designers and researchers regarding the benefits of increased opportunities to engage with inquiry practices within assessment systems.

ABSTRACT. Narratives are usually told in either first-person narration (using “I” pronouns) or third-person narration (using “he”/“she”/“they” pronouns). We examined whether readers are sensitive to changes in narrative perspective as they encode discourse models. In two experiments, participants took longer to read a first-person sentence if it introduced a novel first-person perspective into a third-person narrative rather than continuing a prior first-person perspective. These results illustrate sensitivity to narrative perspective.

ABSTRACT. We designed an experiment to analyze agreement between the answers to open ended (OE) and multiple choice (MC) questions. Students read a scientific text and answered a total of 24 low-and high-level questions, each question was first presented inOE format and after in MC format. Results show the facilitative effect that alternatives have on MC questions. Moreover, agreement is due to a correct OE response, and errors in OE are not transferred to MC answers.

ABSTRACT. Research on teaching in cultural transmission has produced inconsistent results. This study analysed the content of face-to-face teaching under experimentally manipulated conditions. Storytelling is a teaching method we expected to observe. Chains of six participants built baskets to carry rice, then taught the next-in-line either with or without their basket to hand. Displacement of the basket didn’t affect performance but stimulated innovation. Storytelling didn’t affect performance but appears specialised for discussing non-routine events.

ABSTRACT. This experiment examined factors that influenced knowledge revision for vaccine misconceptions on social media (i.e., Facebook). Participants read refutations on individual users’ Facebook profile that varied in source credibility (high vs. low) and justification (personal opinion, authority, multiple sources) and then completed a knowledge revision post-test and argumentative essay about vaccines. Results show that 1) the refutations facilitated knowledge revision, and 2) readers’ essays were more semantically similar to refutations in the low-credibility condition.

ABSTRACT. This study explores the potential interactions between on-line and off-line explanatory prompts on science text comprehension. On-line, participants were prompted to self-explain or think aloud (control). Off-line, they were prompted to explain or recall the text. Preliminary results suggest a possible interaction between on-line and off-line tasks, such that subsequent test scores may be facilitated only when explanation prompts align during and after reading. Additional data collection is ongoing to fully test this potential interaction.

ABSTRACT. This study examined the effects of relevance instructions on Japanese EFL learners’ post-reading memory when focusing on specific information rather than general comprehension. Participants read two expository texts with different difficulty levels and then engaged in a written recall task. Results showed that the relevance instructions facilitated recall of textual information. However, such effects were not observed when the participants read a demanding text because readers allocated many cognitive resources to lower-level processing during reading.

ABSTRACT. When drawing conclusions from media, critical evaluation is important. The purpose of this study was to investigate detection of contradictions between text and graph for pseudoscience topics when questions were increasingly specific and text was more/less complex. Results showed no effects of complexity or question specificity on detection, partly because overall detection rates were quite low. Further, participants rarely identified what was contradictory between text/graph when asked. Future work should investigate the lack of detection.

ABSTRACT. This paper explores the development of local knowledge during an online conversation. Local knowledge is characterized by the sharing, distribution, and certainty of mutual knowledge that develop in online conversation. Data is collected of students sustaining an online conversation throughout a semester. The analysis examines the emergence of local knowledge under two conditions. Each condition is tied to the formation of virtual co-presence that the interlocutors use to manage their conversation.

ABSTRACT. We adopted an error-detection paradigm based on textual errors on the situation model level in expository texts as a marker of metacognitive comprehension monitoring in secondary school children. Inconsistent and consistent versions of each text differed only in one word, which led to self-contradictory information within the inconsistent version. Reading times were longer for correctly classified inconsistent than for correctly classified consistent sentences in fifth-graders, while there was no difference in children of higher grades.

ABSTRACT. This study investigated the monitoring of situational intersentential links in L2 narrative reading. Fifteen Japanese college students read English narrative texts including information inconsistent along three situational dimensions (protagonist, temporality, spatiality). Their during-reading processes were recorded with an eye tracker and the collected data were examined with eye movement matrices. The results showed that L2 readers looked back to the inconsistent information, but their monitoring differed by dimension.

ABSTRACT. Research consistently demonstrates that readers reproduce inaccurate information from texts to complete post-reading tasks. We assessed whether opportunities for online search would reduce the effects of exposure to inaccuracies. In Experiment 1, participants utilized inaccurate information at reduced rates when allowed to perform online searches during a post-reading task. In Experiment 2, participants were more likely to consult information when search results were made readily available during the post-reading task, albeit without any additional reductions.

ABSTRACT. We examined the role prospective content (information relevant in the future) has on the time-course of processing difficulties in narratives. Readers encountered target sentences that were consistent with a protagonist’s present but inconsistent with prospective information associated with the protagonist. When the prospective information was local, processing difficulties were immediate. When the prospective information was backgrounded, processing disruption was delayed until the spillover sentences. Results will be discussed within the memory-based processing model, RI-Val.

ABSTRACT. While readers typically read from an omniscient point of view, both task-based and text-based manipulations can result in readers adopting the perspective of the protagonist. Taking the protagonist’s perspective results in readers becoming sensitive to perspective-relevant inconsistencies they otherwise fail to detect. However, consistent with the RI-Val model of comprehension (O’Brien & Cook, 2016a; 2016b), even when readers adopt the protagonist’s perspective, they are unable to do so to the exclusion of other information.

ABSTRACT. Our objective was threefold: to identify the variety of markers used to signal causal relations in Spanish; to describe the functionality of those markers; and, to determine whether there is a relationship of specificity between markers and particular types of causal relations. We analyzed a corpus of 2,514 causal coherence relations. 40 linguistic devices were identified. 8 of the most frequent markers signal different relations. 10 markers show a high degree of specificity.

ABSTRACT. Since an RST (rhetorical structure theory) tree can be converted into a syntactic dependency tree as well as the network data equally, the data extracted from RST-DT can be used to calculate the discourse distance (DD). This study finds that English discourse structure has its minimum mean DD and discourse units are arranged in patterns. The two methods merged should work as a computational model for measuring text comprehension and complexity.

ABSTRACT. Concepts in text activate relevant and associated knowledge in memory and this activation can facilitate reading comprehension6. But less is known about what types of knowledge are most predictive of comprehension. High-school students took a comprehension test on the topic of immigration and a set of associated background knowledge measures that varied in specificity (domain to topic). Only topic specific and topical vocabulary measures accounted for significant variance in reading comprehension, supporting the specificity hypothesis.

ABSTRACT. This poster addresses research exploring the hypothesis that good reading comprehension instruction designed to teach students to learn from complex informational texts should include pedagogy that problematizes the texts. In this manner, students learn skills to address text shortcomings found in the wild and maximize their learning. A rationale and taxonomy for problematizing texts in reading comprehension curriculum will be presented along with classroom intervention outcomes.

ABSTRACT. Eliciting self-explanations during source code comprehension activities have the potential to help students learn core computer science skills such as reading code and building accurate mental models of the code. At the same time, self-explanations in the context of source code comprehension pose new challenges such as combining language with references to code statements and mathematical expression evaluations. This paper presents a preliminary analysis of self-explanations generated by college-level students during source code comprehension activities.

ABSTRACT. We propose to present a brief literature review on important characteristics of early reading text; show how we have addressed these characteristics in a set of early readers developed by the first author and collaborators; and discuss practical methods for selecting and adapting text for struggling readers.

ABSTRACT. This study examines the assumptions that text comprehensibility positively influences readers’ perceived competence, intrinsic motivation during reading, and interest in the texts’ topic. 164 students from 6th to 9th grade randomly received one of four texts with comparable content. After reading, they completed a comprehensibility questionnaire and several motivation questionnaires. As expected, results showed statistically significant effects of comprehensibility on perceived competence (r_alerting = .94) and interest in the topic (r_alerting = .80).

ABSTRACT. Researchers in art education have postulated trans-symbolic processes which are thought to occur across media. In this study, we test this hypothesis by having participants think aloud to paintings and stories. Participants also rated each on aesthetic experiences (pleasure, interest, understanding). The protocols were categorized on the processes (e.g., paraphrases, elaborations, bridges, metacognition, responses and evaluations). Although we did not find evidence of profiles across media, the frequencies of the thought categories predicted aesthetic responses.

ABSTRACT. This study investigated which linguistic features assessed by Coh-Metrix play a role in predicting the text difficulty of graded readers for young learners of English. The multiple regression analysis indicated that a combination of the variables of lexical sophistication, temporal cohesion, and co-referentiality explained 80% of the variance in the text grade levels assigned by the publisher. These results are discussed in terms of readers' age level to increase our understanding of text simplification processes.

ABSTRACT. The present study attempts to sketches a theoretical model entitled ‘a Performative Theory of Solidarity Discourse’ to analyze solidarity as a discourse. The data of this study is selected speeches that were delivered in the 2016 AIPAC Conference by seven influential American politicians. This theory puts forward eight assumptions and six hypotheses. These assumptions along with the hypotheses of this theory viewed solidarity discourse as performing a number of acts: identifying, assertive, regrouping, and commissive.

ABSTRACT. Generalized conversational implicature (GCI) is a context-independent class of pragmatic meaning, such as “some” implicating “some but not all." We tested whether comprehenders incorporate GCIs into their memory representations of sentences. In a running recognition task, participants judged whether they had seen exact sentences previously. GCI-relevant changes were falsely recognized as “old” at rates matching other semantic changes. Meanings derived from GCIs appear to become part of sentence representations, especially as the verbatim trace degrades.