ABSTRACT. Self-talk is well-researched in psychology, but its linguistic properties have received less attention. Unlike dialogue, where ‘I’ and ‘you’ pick out speaker vs. addressee, in self-talk ‘I’ and ‘you’ can both refer to the speaker (e.g. someone says to themself: “I am/you are such a fool”). Our experiment tested pronoun use in self-talk, and shows “I” vs. “you” rate is influenced by structural factors, and crucially also by sentiment valence (positive/negative statements about self).

ABSTRACT. The English pronoun “they” is increasingly used in the singular. E.g., in “Alex went to the movies with Liz. They bought popcorn,” “they” is ambiguous between the singular (Alex) and plural (Alex-and-Liz) interpretations. Does priming affect the relative activation of singular vs. plural meanings? If so, by what representations? We find that priming does modulate singular/plural availability through a broad “singular” or “plural” activation and not an association between “they” and a specific person.

ABSTRACT. Anaphors refer to antecedents, and the process of connecting the two is called resolution. Prior research has identified factors affecting the speed and success of anaphor resolution in humans. We investigate whether six such factors also affect the speed (i.e., indexed by surprisal) and success (i.e., comprehension question accuracy) of five LLMs: GPT-2-XL, LLaMa-3.1-8B, Pythia-12B-deduped, Mistral-7B, and Mistral-24B. We find promising cognitive alignment for some LLMs (Mistral-7B and GPT-2-XL) suggesting their potential as cognitive models.

ABSTRACT. This study examines how psycholinguistic features of spoken brand discourse on TikTok predict engagement metrics. Automated transcription and multi-feature linguistic analysis of 374 videos reveal that second-person addressivity, emotional intensity, lexical complexity, and interrogative structure show meaningfully different predictor profiles across the three engagement outcomes. The results extend processing fluency theory and research on pronoun-based referential processing to naturally-occurring promotional discourse, and illustrate the potential of corpus methods for studying spoken social media text.

ABSTRACT. Identifying fake news across media requires active epistemic monitoring. Analyzing EU data (N=25,600), this study conceptualizes “perceived exposure” to fake news as a proxy for detection capacity. Results show the strongest predictor of perceived exposure is a “Traditional Media” profile, where citizens blame legacy institutions for disinformation. Higher exposure is further predicted by lower institutional trust. The study offers implications for media literacy, emphasizing the need to move beyond technical verification to address trust breaches.

ABSTRACT. Background Document-source trustworthiness impacts perceived trustworthiness and sharing intentions of online documents. Objective Test whether prior beliefs towards embedded-sources alter this effect. Method Undergraduates read political snippets crossing document-source trustworthiness (High-trust, Low-trust) with prior beliefs towards embedded-source (Negative, No-Belief). Results High-trust document-source increased snippets trustworthiness and sharing; effect was weaker when participants had negative beliefs regarding embedded-sources. Conclusions: trustworthiness evaluation reflected an interplay between document and embedded sources while sharing depended solely on document-source trustworthiness.

ABSTRACT. In this dissertation study, we examined the extent to which different features of competing explanations affect readers’ evaluations of each explanation. We presented readers with a series of disagreements in medical explanations for fictitious patients’ physical symptoms. The explanations varied in their causal structures’ simplicity and their sources’ expertise. Participants evaluated each explanation by judging its probability. They also judged the likely cause of disagreement between sources and their own curiosity about the topic.

ABSTRACT. Meaningfully addressing low trust in science requires a nuanced understanding of its multifaceted nature, a complexity that has intensified as digital information ecosystems have fundamentally changed how people engage with and evaluate scientific information. To address this challenge, we integrate diverse perspectives on what it means to trust science from three domains—discourse processes, epistemic cognition, and science communication—to develop a multidimensional framework that captures the cognitive, communicative, and interpretive dimensions of trust in science.

ABSTRACT. The present study investigates whether mental simulation occurs in both an L1 (Japanese) and an L2 (English). We conducted two sentence–picture verification experiments with native Japanese speakers who learned English through the Japanese educational system. The findings show robust shape simulation in both L1 and L2, with a stronger effect in L1, whereas orientation simulation did not emerge in either language. Additional analyses further suggest that shape typicality may modulate the strength of simulation effects.

ABSTRACT. Research on discourse processing has predominantly focused on written information; however, discourse models often integrate written and spoken information. This study examined how input modality affects discourse updating in heritage Spanish English bilinguals. Participants read and/or listened to expository passages in both languages, followed by free recall protocol and comprehension questions. Findings indicate better recall and accuracy for information presented auditorily in Spanish, suggesting a language-dependent modality effect in bilingual discourse updating.

ABSTRACT. Current models of discourse comprehension indicate that information is retrieved to working memory via semantic priming. This experiment examined this form of retrieval with respect to the dominant and non-dominant languages of highly-proficient bilinguals. Our results demonstrate that within a discourse context, semantic priming does not impact retrieval of prior text and does not differ according to language dominance. Future work will examine the length of the discourse context on the strength of semantic priming.

ABSTRACT. Bilinguals adjust language use across conversational contexts that vary in cognitive demands. Dense code-switching, which involves within-sentence language switches, has been argued to be more demanding than dual-language contexts, where languages alternate between speakers. In this study, Spanish–English bilinguals (N = 72) watched conversations in both contexts and completed memory tests for conversational topics and speakers. We examined whether processing demands and prior experience with these contexts shaped episodic memory performance.

ABSTRACT. This study examined how targeted interventions impact reading subskills for middle school students over a school year. Using the ReadBasix assessment, we investigated multiple interventions and saw impacts on decoding, sentence processing, reading efficiency and comprehension. Our results suggest that foundational skill gains (i.e., decoding) often precede higher-order growth (i.e., reading comprehension), which supports the decoding threshold hypothesis. These findings also emphasize the need for longitudinal tracking of intervention efficacy.

ABSTRACT. Recent reading reforms privilege simple, two-factor characterizations of reading comprehension, comprised of two independent but correlated processes: word recognition and language comprehension. Data indicate these processes share variance and must be integrated (i.e., are not as independent as dominant models suggest). Structural modeling in a sample of 462 adults reveals poor fit for simple two-factor models, indicating roles of additional, integrative and active-self-regulatory components of adults’ reading comprehension, supporting the Active View of Reading.

ABSTRACT. This study examines whether a sentence-processing reading routine improves comprehension for ninth graders reading 2-3 grade levels below expectations and whether effects differ by gender. Teachers led 30-minute, twice-weekly sentence deconstruction/reconstruction lessons using anchor texts. ReadBasix sentence processing, efficiency, and comprehension were administered at baseline and midyear. Whole-group gains were nonsignificant, but gender moderated change: girls maintained stable performance while boys exhibited a significant gain in reading comprehension. Time-3 data will examine persistence.

ABSTRACT. Cloze (gap-fill) exercises can assess discourse comprehension when correctly filling in a gap requires integration across sentences and paragraphs. In our format, learners read a source text then complete a cloze passage derived from a summary of that text. We test three automated methods to select gaps and found that our method, which compares word probability estimates calculated conditioned on local versus global contexts, was preferred by expert raters.

ABSTRACT. This qualitative study examines how nurse educators are trained to read nursing texts and how they teach students to use text for classroom learning and clinical practice. Although disciplinary literacy is widely discussed, little is known about how pre-service nurses read or how they should read. Through in-depth interviews with nurse educators across the country, the study seeks to characterize current disciplinary-specific reading practices and generate recommendations for improving nursing instruction.

ABSTRACT. Comics are increasingly used in education, yet no study has examined their effectiveness in promoting learning compared to traditional texts. We conducted a meta-analysis of 29 empirical studies, from 4,654 screened, comparing comics versus texts on learning outcomes across Science, Technology, Engineering and Mathematics (STEM) and non-STEM domains. Comics outperformed texts in both domains, with no differences between them. These findings suggest comics as an effective tool for enhancing learning across both educational domains.

ABSTRACT. This study examined how text genre influences readers’ visual and verbal thought modalities during reading. Participants (n=161) read either an informational narrative or expository text and reported in-the-moment thought experiences. Visual imagery experience was higher than verbal thinking across genres, with only a marginal genre effect. Prior knowledge was significantly related to verbal thoughts.

ABSTRACT. What does it mean to read well? Standard accounts locate the product of reading comprehension in subpersonal mental representations—situation models that cannot be consciously inspected or directly taught. This paper proposes an alternative: reading for models (RFM). Drawing on philosophy of science, disciplinary literacy, and cognitive psychology, RFM reframes the comprehension product as purposefully constructed, revisable, and publicly evaluable models. This reconceptualization has implications for how we research and teach reading comprehension.

ABSTRACT. We assessed reference as a function of audience design. Speakers named items in displays for partners who were either themselves, peers, or older adults. Competitor items, when included, differed from targets in size, color, or type. Competitors were either visible mutually (common ground) or to speakers only (privileged ground). Speakers elaborated more for themselves than others, especially peers, varying systematically across addressees. This suggests self-communication relies more on speaker-specific memory than perspective-taking during referential communication.

ABSTRACT. Verbal irony is when intended meaning contrasts with expressed meaning. This study investigates how producing verbal irony affects emotional states. Participants viewed negative images and described them ironically or literally via typing, then rated the intensity of their negative emotion. Participants reported less intense negative emotion in the irony condition compared to the literal condition. Possible mechanisms for this effect derived from text analysis of typed responses include language production, word affect, and cognitive engagement.

ABSTRACT. Alignment has been theorized to play a functional role in human coordination and communication. We used nonlinear methods on a large conversational dataset to explore the interaction between age, sex, and status, and linguistic alignment (lexical, syntactic, and turn-taking). Our results show these social variables have varying relationships with the different dimensions of alignment. To account for this pattern of results and guide future research, we put forward the notion of compensatory alignment.

ABSTRACT. Across two self-paced reading experiments, we examined how personal common ground (PCG; close vs. stranger) and informational common ground (ICG; shared knowledge) influence verbal irony comprehension. Reading times and Likert judgments indexed readers’ inferences. Irony was read faster under high PCG and low ICG, but slower than literal endings when both were low or high.

ABSTRACT. Despite decades of research on comprehension strategy instruction, the specific mechanisms that explain its impact remain unspecified. The goal of this study is to create a taxonomy of these mechanisms by systematically analyzing published articles testing the effects of strategy instruction in upper-elementary and middle grades. Findings can help researchers intentionally test specific mechanisms that might explain the positive effects observed in the literature and resolve some of the confusion surrounding text comprehension instruction.

ABSTRACT. In this study we utilize a think aloud procedure to capture Grade 9 students’ reading strategy use. We assess whether this way to capture reading strategy use in real time explains variance in different reading comprehension outcome measures over and above a questionnaire designed to capture knowledge and use of reading strategies, and word reading ability.

ABSTRACT. Digital reading dominates educational instruction and assessment, yet many platforms replicate print formats without leveraging supportive interactive features. This scoping review synthesizes 45 empirical studies examining how images, interface design, annotations, and gamification influence comprehension and learning outcomes in high school and college learners. Findings indicate digital outcomes are design-depended: features that stabilize navigation, guide attention, and support generative processing improve comprehension, while poorly integrated multimedia, cognitive overload, and competitive gamification can hinder deep learning.

ABSTRACT. Reading comprehension arises from interactions between readers and test items. This study examines how reader characteristics (sex, race/ethnicity, English learner status, free or reduced lunch) and item features (text complexity, question type) jointly influence 9th graders’ reading comprehension using Explanatory Item Response Modeling. Significant main effects and interactions emerged (e.g., Craft and Structure questions were more difficult for females than males). Findings highlight the importance of considering reader-test interactions in assessment design and interpretation.

ABSTRACT. This study characterizes how adults with varying literacy skills interpret narrative symbolism in a low-level, written, fictional short story. Think-aloud responses are leveraged in tandem with post-reading comprehension questions to analyze performance and strategy use. Implications will provide valuable insights for refining the understanding of how interpretation is related to constructing mental models, mental model levels, and transactional reading by juxtaposing different types of narrative symbolism (explicit versus ambiguous) across a simple narrative.

ABSTRACT. This study explores summary writing as a way of improving multiple text comprehension. In Experiment 1, 81 undergraduates either summarized or reread each text in a set. Comprehension tests revealed that summarization supported within-text representations (intratextual) but did not support intertextual integration. Experiment 2 explores the potential of combining summary writing with intertextual reflection writing. This study informs how tasks can promote comprehension within single texts and guide integration of those ideas across multiple texts.

ABSTRACT. Braasch & Bråten (2017) identified the Discrepancy-Induced Source Comprehension model should be adapted for younger adolescents, given their documented difficulties with detecting contradictions. Theoretical and empirical work remains sparse. However, a long tradition of comprehension monitoring research has demonstrated that processing contradictions places considerable demands on baseline capacities D-ISC assumes are reliable. We bridge these relatively isolated literatures to specify how monitoring capacity may contribute to D-ISC processes in younger readers, and outline research directions.

ABSTRACT. We explored the dimensionality of inference processes using responses from 1,714 U.S. kindergarten students on a video-based assessment that measures inference-making as it occurs. Using latent space item response modeling, we identified two distinct item clusters—one primarily supporting local coherence and the other global coherence—revealing subdimensions within inference processes. Our findings further indicate that, despite similar overall performance levels, students can differ in their profiles across local and global coherence inferences.

ABSTRACT. The Scene Perception and Event Comprehension Theory proposes that situation models guide attentional selection when comprehending visual narratives. Ambient-to-focal processing (ATF) is a pattern of eye movements that shows the relationship between attention and comprehension, but has only been demonstrated with single images. The present study investigated whether ambient-to-focal processing is sensitive to how attention is affected by situation model construction for picture stories, and in particular during mapping and shifting phases of structure building.

ABSTRACT. We present LANTERN, a network-science framework for quantifying reading comprehension from text retell. Retell responses from 121 children were transcribed and modeled as word co-occurrence networks using multiple window sizes. Network metrics indexing semantic organization predicted passage-level comprehension, standardized reading scores, and high-stakes state assessment performance, controlling for fluency. Findings demonstrate that retell network structure captures meaningful individual differences in comprehension, offering a scalable, theory-driven approach to retell scoring in educational assessment contexts.

ABSTRACT. Fluent reading requires simultaneous identification and comprehension of words. Although automaticity in word recognition is assumed to be the cognitive basis of fluency, automaticity is insufficient for skilled comprehension. This study examined contributions of graphophonological-semantic cognitive flexibility (GSF) to reading fluency (both accuracy/automaticity and prosody) in 78 university students. After controlling for verbal ability and automatic word identification, GSF contributed uniquely to both measures of reading fluency. Implications for theory and practice will be discussed.

ABSTRACT. Reading is an active process that requires the coordination of linguistic and cognitive resources, especially for bilingual readers. Data from 32 independent samples (N = 4,274) was used to examine the relation between executive functions (EFs) and second language (L2) reading. Meta-analytic models showed that EFs are positively associated with L2 reading. Findings are discussed with the Active View of Reading framework and contribute to the call for more investigations of EFs in bilingual literacy.

ABSTRACT. Reading comprehension in middle school increasingly depends on constructing coherent representations of expository discourse, placing high demands on executive control. Socioeconomic status and executive function both predict reading, yet their integration at the level of discourse remains unclear. We examined whether working memory and cognitive flexibility mediate socioeconomic differences in expository comprehension, conceptualizing retell and multiple-choice performance as covarying expressions of discourse comprehension. Executive function partially mediated socioeconomic effects, highlighting control capacity as a mechanism.

ABSTRACT. Two eye-tracking experiments examined pronoun resolution biases and referential adaptation using the Visual World Paradigm. Experiment 1 established robust subject preference: participants fixated significantly more on subject antecedents (61%) versus nonsubject antecedents (32%), p < .001. Experiment 2 tested whether systematic exposure modulates these biases. Statistical analyses revealed no significant effects, though descriptive patterns showed nonsubject-exposure reduced first-mention preferences. Findings demonstrate strong default subject biases with suggestive but non-significant evidence for adaptation in online processing.

ABSTRACT. How does experience guide referential processing? We hypothesize that discourse biases stem from adaptation to the most frequent referential patterns, building on evidence that exposure to subject- vs. nonsubject-antecedent patterns modulates pronoun interpretation. We ask: by what mechanism? Two studies test which representations are activated, suggesting that people track antecedents based on syntactic categories but not specific to semantic role. Three studies test the mechanism underlying adaptation, supporting frequency-tracking over error-based models.

ABSTRACT. This study examined how personal common ground (PCG) and informational common ground (ICG) influence processing of literal and ironic stories) across two Reading Times (RTs) experiments. Experiment 1 showed a PCG × ICG interaction for familiarity RTs and story-type main effects on irony, sense, and emotion RTs. Experiment 2 replicated slower emotion RTs for ironic stories and PCG × story-type interaction.

ABSTRACT. Much research has explored meaning activation in ambiguous words when only one meaning is contextually appropriate but has largely neglected contexts in which multiple meanings are appropriate. Puns are one example of these contexts. The time course of how meaning is activated and selected in puns is being examined using cross-modal lexical decisions to observe the activation levels of meanings of ambiguous words in puns that bias both meanings. Results are pending.