Abstract
Multilingual students with developmental disabilities benefit from instruction that supports both comprehension monitoring and the development of expressive language. This study used a multiple-probe design replicated across three U.S. third-grade participants to evaluate a combined intervention involving explicit metacognitive self-monitoring routines and interactive digital versus static paper-based anaphoric cueing. Following the intervention, students demonstrated gains in comprehension-question accuracy and increased the complexity of their generated wh-questions. Gains were maintained during the maintenance phase, and accuracy was consistently higher with digital anaphoric cueing compared with paper-based anaphoric cueing. However, student feedback indicated a preference for paper-based anaphoric cueing, citing ease of use and familiarity. Students also reported high satisfaction with the intervention’s goals, procedures, and outcomes across modalities. Findings suggest that embedding multimodal supports within metacognitive frameworks can foster comprehension and language production in ways that recognize and build on multilingual learners’ cognitive and linguistic assets.
Keywords
Reading comprehension is foundational for academic achievement, social participation, and lifelong learning. However, becoming a skilled reader is not a natural process that develops solely through time and exposure to print. Seminal research by Durkin (1978) revealed that comprehension instruction was largely absent from elementary classrooms, prompting calls for more explicit teaching. The National Reading Panel (2000) affirmed that summarizing, questioning, and clarifying strategies improve comprehension when taught directly.
Even with such instruction, many students with developmental disabilities face persistent challenges in forming a cohesive understanding of text, particularly when referential relationships are unclear (Finnegan & Mazin, 2016; Joseph et al., 2023). These challenges intensify for multilingual students with developmental disabilities, whose comprehension relies on both linguistic factors and the support provided by instruction for their language development.
Too often, specially designed instruction reflects monolingual norms that overlook students’ home languages and dialects, framing linguistic differences as deficits (Cioè-Peña, 2015). Yet, multilingual learners routinely draw on flexible, strategic ways of making meaning across languages (García & Kleyn, 2016). Instruction that fails to recognize and leverage these assets can limit opportunities for comprehension and participation.
In addition to linguistic marginalization, students with developmental disabilities may experience comprehension breakdowns due to limited decoding skills, gaps in background knowledge (Cain et al., 2004), or demands on working memory (Swanson & O’Connor, 2009). For some, a detail-focused processing style (Happé & Frith, 2006) and limited use of metacognitive strategies (Ozonoff et al., 1991) further complicate reading. One common barrier is difficulty with anaphoric resolution, the process of linking pronouns or referents to their antecedents. When these connections are not made, readers may lose track of meaning.
Skilled readers construct mental models of text by making inferences across sentences, such as identifying that he refers to Jim in the sentence pair “Jim likes to run fast. He wins every race.” This type of inference, known as anaphora, is essential for comprehension (Baumann, 1986; Yuill & Oakhill, 1988). While many students with developmental disabilities decode accurately, they often struggle to resolve anaphoric references (Ehrlich et al., 1999; Tavares et al., 2015). These disruptions in referential cohesion are rarely resolved without instruction in monitoring and repair strategies. Anaphoric cueing offers one such strategy by providing visual supports that highlight referential links, such as underlining pronouns and prompting students to identify antecedents. This support can be paired with direct instruction to enhance effectiveness (O’Connor & Klein, 2004). Studies show that, when used systematically, anaphoric cueing improves comprehension for students with language-based and developmental disabilities (Roux et al., 2015; Solis et al., 2015). These findings highlight the potential value of integrating referential and metacognitive supports within a strengths-based framework.
Metacognitive Monitoring and Strategy Instruction
Many students with developmental disabilities struggle to monitor their comprehension and may not spontaneously reread or revise their understanding when texts become incoherent (Frith & Snowling, 1983; Happé & Frith, 2006; Ozonoff et al., 1991). Metacognitive strategy instruction, which teaches students to recognize breakdowns, make predictions, and verify understanding, has shown strong evidence of effectiveness in improving comprehension outcomes (Reutebuch et al., 2015; Singh et al., 2021).
While anaphoric cueing can signal referential relationships, combining it with metacognitive strategies may better equip students to identify and repair misunderstandings caused by unclear anaphora. For example, teaching students to recognize pronouns, predict possible antecedents, and reread if needed integrates language-specific and general comprehension-monitoring strategies. The present study builds on prior findings (e.g., O’Connor & Klein, 2004; Solis et al., 2015) by explicitly pairing metacognitive instruction with both digital and traditional anaphoric cueing to support literal comprehension and wh-question generation in multilingual students with developmental disabilities.
Framing Instruction Through a Linguistic Justice Lens
Special education interventions for multilingual students are often designed to remediate perceived deficits in English proficiency, reinforcing monolingual norms that marginalize students’ home languages and cultural identities (García & Kleyn, 2016; Migliarini & Stinson, 2021). However, research demonstrates that multilingual students with disabilities can thrive when their linguistic resources are treated as assets and instruction affirms their full repertoires (Castellón et al., 2024; Cioè-Peña, 2015).
Identity and agency play a central role in how multilingual learners engage with literacy. For example, Bird et al. (2016) showed that children with developmental disabilities can become bilingual when consistently supported by linguistically responsive instruction. These studies underscore the importance of designing instruction that affirms students’ cognitive and cultural strengths, rather than remediating perceived linguistic deficits.
The Present Study
This study investigated the effects of a combined intervention integrating explicit metacognitive strategy instruction with both interactive, digital anaphoric cueing and static, paper-based anaphoric cueing for multilingual third-grade students with autism or developmental delay. Building on research supporting metacognitive and referential scaffolds (e.g., Singh et al., 2021; Solis et al., 2015), the intervention aimed to improve comprehension-question accuracy and wh-question generation while affirming students’ agency and linguistic strengths. Instruction was delivered in English, and routines were designed to support cross-linguistic transfer, flexible meaning-making, and metacognitive reflection across modalities. This work addresses a persistent gap in the literature: Few studies have examined comprehension interventions that attend to both disability and linguistic diversity. The study offers a strengths-based alternative to conventional literacy instruction by integrating cognitively and linguistically responsive practices. Five research questions guided the study:
What are the effects of metacognitive strategy instruction combined with interactive, digital anaphoric cueing on participants’ accuracy in answering wh-questions that assess literal comprehension?
What are the effects of metacognitive strategy instruction combined with static, paper-based anaphoric cueing on participants’ accuracy in answering wh-questions that assess literal comprehension?
What are the effects of the intervention on the quality of independently generated wh-questions following reading?
To what extent are improvements maintained once metacognitive instruction and cueing supports are withdrawn?
How do multilingual students with developmental disabilities perceive the goals, procedures, and outcomes of the intervention across digital and paper modalities?
Method
Participants
Participants were three third-grade students eligible for special education services in a Midwestern U.S. public school district. All had Individualized Education Program (IEP) goals targeting reading comprehension and were eligible under autism or developmental delay. One student withdrew early, and three completed the study. Two students were emergent bilinguals who spoke Tagalog or Arabic; one was bidialectal, using both African American Vernacular English and Mainstream American English. Table 1 provides participant demographics and assessment data.
Characteristics of the Three Participants in the Study on Metacognitive Monitoring and Multimodal Cueing.
Note. Demographic data on racial-ethnic background reflect participants’ and/or families’ descriptor preferences as obtained by researchers through a recruitment survey. AAVE = African American Vernacular English; ASD = Autism Spectrum Disorder; DD = Developmental Delay; ESL = English as a Second Language; ESSA = Every Student Succeeds Act (2015); FS = Full Scale; IDEA= Individuals with Disabilities Education Improvement Act (2004); WISC-V = Wechsler Intelligence Scale for Children, 5th edition.
Setting
This study, approved by the author’s institutional review board, was conducted in a large, urban, public elementary school in the Midwest. Eighty-seven percent of students were classified as economically disadvantaged, 73% were classified as English Learners, and 12% were students with disabilities. Research activities occurred 4 days per week in a quiet setting, where a graduate student in a special education program delivered the 1:1 intervention.
Materials
Materials included five wh-question cards (i.e., who, what, where, when, how) printed in black ink and 72-point Arial font on 3″ × 3″ white cardstock and a visual flowchart illustrating the Stop–Ask–Check–Connect (SACC) strategy using image-supported steps for self-monitoring, referent identification, and question generation during reading (see Figure 1). In addition, a pool of 275 researcher-developed passages was used in the study. Each passage consisted of four sentences (6–8 words per sentence) featuring a named character, two sequential actions, and a concluding interaction or feeling. Vocabulary was high-frequency and age-appropriate, with simple syntactic structures. Passages were pilot tested with third-grade students to ensure readability and content familiarity.
Stop-Ask-Check-Connect Visual Flowchart
The Research Team
The interventionist was a bilingual special education teacher completing graduate coursework. A second bilingual educator with 24 years of experience developed materials, conducted procedural integrity checks, and collected interobserver agreement (IOA) data. A bidialectal speech-language pathologist with 25 years of experience in education consulted on instructional content.
Procedures
Pretest/Posttest: No Anaphoric Cueing
Pretest and posttest sessions included two components: comprehension question-answering and wh-question generation. Students read researcher-developed passages presented in plain text and responded to five multiple-choice, wh-questions per passage. Passages were presented in either paper or digital format, with modality randomized at the individual level within sessions. Digital versions included click-to-respond questions but no visual cues. Students completed comprehension questions independently, with no feedback provided, spending an average of 5 min per task.
Following comprehension tasks, students completed a single 12-min question-generation probe. Each probe involved reading a four-sentence passage presented in either paper or digital format, with modality randomized across sessions. The interventionist placed five wh-question cards (who, what, where, when, and how) face-up and instructed the student to “ask a wh-question about the story.” If the student generated a question, the interventionist acknowledged the attempt (e.g., “Good question” or “Thank you”) and moved on to the next card. When no response was given within 7 s, the interventionist provided a scaffolded prompt by pointing to a card and offering a partial model. If the student still did not respond, the interventionist modeled an example question and ended that opportunity before proceeding. Thus, each student was given five opportunities per probe to generate questions independently, with prompts and models used only as needed. All responses were recorded on an iPad for masked scoring, and the scoring rubric directly reflected levels of independence: a score of 2 indicated a high-quality question produced without prompting, a score of 1 indicated a partially correct or prompted response, and a score of 0 indicated an incorrect or modeled response.
Baseline and Maintenance (Anaphoric Cueing Conditions)
Procedures during baseline and maintenance phases mirrored those in pretest and posttest sessions, with only the addition of anaphoric cueing. Cueing consisted of highlighting referential relationships visually. Specifically, pronouns were underlined and antecedents placed in bold font. In these phases, comprehension question-answering and wh-question-generation tasks were connected to researcher-developed passages. In each session, students read three passages: two assigned to literal comprehension (accuracy questions) and one assigned to wh-question generation. Passages were presented in either paper or digital format, with modality randomized within and across sessions to balance exposure.
Paper-Based Anaphoric Cueing
Printed passages (36-point Arial font) were presented on 8.5″ × 11″ white paper. Pronouns were underlined, and referents were bolded. Multiple-choice questions were printed below the passage in 12-point font, and students used pencils to circle answers. Papers were coded and collected for masked scoring.
Digital Anaphoric Cueing
Students read interactive digital passages using the Genially (n.d.) platform on a Dell laptop. Pronouns were clickable, revealing referents through pop-up textboxes, and animated arrows pointed back toward the referent. See Figure 2 for an example of a response entered on a digital passage with anaphoric cueing. Multiple-choice questions appeared sequentially, with responses recorded using the Xbox Game Bar screen capture tool. All interactions were saved using unique session codes for masked scoring.
Digital Anaphoric Cueing Passage with Accuracy Question Response
Intervention: Instruction and Anaphoric Cueing
For all intervention sessions (across two phases described below), students completed up to 18 min of wh-question answering and 12 min of wh-question generation per session, with passage modality (digital or paper-based) randomized within sessions.
Phase 1: Pronoun Resolution and Question Generation Instruction
The intervention began with two 30-min sessions focused on referent identification and wh-question generation while reading short sentence pairs. For example, the interventionist presented and read aloud, “Jordan went to the park. He played tag with his friends.” The interventionist shared that he was a pronoun, “a word that acts like a substitute and points back to the word it takes the place of.” The interventionist drew an error in connecting he and Jordan. This activity provided a natural entry point for exploring how pronouns function in English to connect ideas across sentences and how referent tracking supports comprehension.
The interventionist used a cross-linguistic reference chart (see Figure 3) as she gave examples and asked students to reflect on how pronouns might be expressed to link meaning across sentences in their home language or dialect. For the student who spoke Tagalog, the conversation focused on using the gender-neutral pronoun siya and how this compares to gendered pronouns in English. The student who spoke African American Vernacular English was supported in considering variation in subject expression and emphasis, including instances where pronouns might be repeated or omitted. The student whose home language was Arabic reflected on how subject pronouns are embedded within verb forms and are not always explicitly stated. This instructional step supported metalinguistic awareness, established an entry point for cross-linguistic transfer, and validated students’ knowledge across languages and dialects.
Pronoun Usage Across Languages and Dialects: Patterns to Notice and Understand
Students then practiced formulating who and what questions based on the sentence pairs featuring subject, object, and possessive third-person pronouns in English. Wh-question cards were used as visual scaffolds, and sentence stems for question formation were offered verbally. The interventionist modeled question formulation in English and affirmed students’ attempts to find pronoun referents and ask wh-questions. These interactions emphasized clarity of meaning and promoted students’ agency as question-askers while supporting the development of cross-linguistic strategies for tracking referents and generating questions.
Phase 2: Metacognitive Strategy Instruction
Across four intervention sessions in phase 2, the interventionist supported metacognitive monitoring by introducing the Stop–Ask–Check–Connect (SACC) strategy, which provided a structured routine for reading, self-questioning, and meaning repair. The strategy was implemented through an explicit instructional sequence outlined in a lesson plan format and delivered across four 30-min sessions. Lessons followed a consistent routine of modeling, guided practice, and independent practice.
The SACC strategy was introduced as a scaffold for resolving referential ambiguity and to affirm students’ agency and linguistic knowledge. Instruction was grounded in a language-as-resource orientation, positioning students as active problem-solvers capable of drawing on their full linguistic repertoires to construct meaning. During “Stop,” students were encouraged to pause and reflect on comprehension using inner speech or familiar expressions in any language. In “Ask,” question formulation was supported through modeling that validated and allowed students to use question formations aligned with respective home language structures (i.e., Arabic, African American Vernacular English, or Tagalog). For the next step in SACC, the interventionist guided students to “Check” by evaluating sentence-level coherence, including through self-talk or paraphrasing. In “Connect,” students were explicitly taught to link pronouns to antecedents and encouraged to explain their reasoning using all linguistic resources. See Table 2 for a sample script used to model the steps of SACC in phase 2.
Stop–Ask–Check–Connect (SACC) Strategy Instruction: Modeling Script.
Dependent Measures
Question Answering
The primary dependent variable was the total number of wh-questions answered correctly following researcher-developed passages. Each passage was four sentences long (6–8 words per sentence), used familiar vocabulary and simple syntax, and included a named character, two sequential actions, and a concluding interaction or feeling. Each passage was followed by five multiple-choice literal wh-questions (who, what, when, where, how), directly answerable from the text. Questions focused on key details such as character, setting, actions, or outcomes, with three response options (one correct, two plausible distractors).
Question Generation
For each session, students completed one question-generation probe following a passage. Responses were scored for quality and completeness using a 3-point rubric adapted from Solis et al. (2015): 2 = high-quality wh-question with no prompting, 1 = partially correct or prompted response, and 0 = no question or incorrect after prompting. With five opportunities per probe, the maximum raw score was 10 points. Raw scores were multiplied by 10 to yield a percentage, with 100% representing five independently generated, high-quality wh-questions.
Experimental Design
A multiple probe replicated across participants design (Gast et al., 2018) was used to examine the effects of metacognitive strategy instruction combined with multimodal cueing. There were five phases: pretest (no cueing), multimodal cueing (baseline), strategy instruction combined with multimodal cueing (intervention), multimodal cueing (maintenance), and posttest (no cueing). Data were collected for wh-question answering and wh-question generation in each phase. The six-session strategy instruction combined with cueing (intervention) phase began for the first participant when performance on wh-question answering accuracy achieved stability, which was achieved when data showed a zero-celerating trend or when 80% percent of the data points fell within 20% of the phase median (Cooper et al., 2019). Immediately upon completion of the intervention phase, data were collected for four school days and then once per week.
Instrumentation
Procedural Fidelity
Procedural fidelity data were collected for 33% of all sessions, at least once per phase per participant. During data collection, an observer used the checklist to record whether the primary data collector correctly implemented each step. Procedural fidelity was calculated by dividing the number of correctly completed steps by the total number of steps and multiplying by 100. Across all sessions, total procedural fidelity for all students averaged 98%.
Interobserver Agreement
An independent observer collected reliability data for at least 20% of sessions for each skill in each phase. Unaware of students’ assignment to treatment conditions, the independent observer scored comprehension and question-generation probes using permanent products or video recordings. Interobserver agreement was calculated as item-by-item agreement divided by total items scored. Mean agreement was 100% for Zamir and Sadie, and 96% for Brittany.
Social Validation Questionnaire
The social validation questionnaire used a 5-point rating scale to assess students’ perceptions of the social acceptability of the intervention’s goals, procedures, and outcomes. A paper form contained 10 statements addressing the helpfulness, ease of use, and enjoyment of the metacognitive strategy and cueing enhancements in traditional and digital modalities. Sample statements included: “It is important to understand what I read,” “SACC helped me understand what I read,” “It was easy for me to click on pronouns on the computer,” “I liked reading stories on the computer,” and “I liked reading stories on paper.” Upon completing the combined strategy instruction and cueing phase, the interventionist distributed the assessment and read each statement aloud. Students used a highlighter to fill in the number of stars indicating their level of agreement (1 = strongly disagree; 5 = strongly agree) with each statement.
Results
Accuracy of Literal Interpretations
Visual analysis of wh-question answering data indicated strong intervention effects across participants (see Figure 4). All students demonstrated low, stable baseline performance across paper and digital anaphoric cueing formats. With the introduction of metacognitive strategy instruction, all three students showed immediate and sustained level increases with accelerating trends and minimal overlap with baseline data. Performance gains were maintained during the maintenance and posttest phases. Accuracy scores were consistently higher with digital anaphoric cueing than with paper-based anaphoric cueing across participants. The percentage of nonoverlapping data (PND) was 75%, indicating moderate effectiveness (Scruggs & Mastropieri, 1998).
Number of Correct Responses to Wh-Questions Across Sessions
Zamir
Zamir’s baseline scores were low and stable, with medians of 1 (paper) and 2 (digital). Following the intervention, scores increased to a median of 4 in both formats. He consistently answered all wh-questions except those beginning with how. Posttest scores reflected clear maintenance of gains, with no overlap between pretest and posttest conditions. Effect sizes (Tau-U = .77 for both formats) indicated moderate improvement. Digital cueing was associated with slightly higher accuracy across sessions.
Brittany
Brittany’s baseline scores were also low and stable, with medians of 1 (paper) and 2 (digital). After the intervention, her median scores rose to 5 (digital) and 4 (paper). She consistently answered who, what, and where questions accurately. No overlap was observed between pre- and posttest conditions. Tau-U values indicated strong improvement (digital = 1.00; paper = 1.00). Her growth was more rapid and sustained in the digital format.
Sadie
Sadie’s baseline performance showed low scores with some variability (medians: 1 for paper, 2 for digital). With intervention, scores increased to a median of 5 in both formats. She maintained these gains after the intervention was withdrawn. Tau-U values showed strong effects (digital = .98; paper = 1.00), with slightly faster gains in the digital format despite a stronger effect size for paper-based cueing.
Quality of Generated Questions
Figure 5 shows students’ performance in generating high-quality wh-questions about the content of digital or paper passages. Only one question-generation probe was administered per session. As a result, modality varied across sessions and was not equally distributed. In Figure 5, data points were marked by modality to illustrate exposure. However, modality effects cannot be claimed; therefore, results can only be interpreted in terms of overall growth across phases rather than by format. During the baseline (anaphoric cueing only) phase, all students produced few or no complete wh-questions without prompting. With anaphoric cueing and strategy instruction, students demonstrated immediate level increases and positive trends, which were sustained following the withdrawal of support. Although students did not reach the full instructional target of generating five questions per probe, all showed meaningful growth in the number of wh-questions independently generated.
Percentage of Complete, High-Quality Wh-Questions Generated Across Sessions
Tau-U analyses indicated strong to very strong effects: Zamir = 1.00, Brittany = 0.88, and Sadie = 0.80. These results support the effectiveness of the combined intervention in improving question generation skills across all participants.
Social Validity
Across all participants, the mean social validity rating was 4.47 (SD = 0.57), indicating high satisfaction with the intervention’s goals, procedures, and outcomes. Although students responded well overall, they rated paper-based cueing more favorably than digital formats, suggesting a preference for familiar, tangible tools.
Discussion
This study demonstrated that combining metacognitive strategy instruction with multimodal anaphoric cueing improved literal comprehension for multilingual students with developmental disabilities. The intervention was effective across both digital and traditional formats, though students responded differently to each modality. These findings extend prior research on anaphoric cueing (e.g., O’Connor & Klein, 2004; Solis et al., 2015) and metacognitive strategy instruction (e.g., Reutebuch et al., 2015; Singh et al., 2017) by investigating their combined effects in a population that has been underrepresented in such research, those from bilingual and bidialectal backgrounds.
Gains in question answering were robust, with clear functional relations demonstrated across all participants. Though effects on question generation were more modest, students moved from near-zero performance to consistently producing one to two high-quality wh-questions without prompting. Given the brief duration of the intervention and its emphasis on who and what questions, these expressive gains are meaningful and signal the potential of extended, scaffolded instruction to deepen metacognitive engagement and language production.
While modality was not a manipulated variable, students generally performed better with digital cueing, suggesting that interactive features may have enhanced attention or memory. However, all students expressed a preference for paper-based materials. This discrepancy may reflect greater familiarity with print tasks, digital fatigue from prolonged online learning, or increased cognitive demands associated with digital navigation. These findings suggest the need to balance student preferences with performance-based data, while building learners’ metacognitive awareness of how different formats support comprehension.
Although the intervention was delivered in English, it incorporated representations of students’ home languages and dialects. Cross-linguistic comparisons appeared to support metalinguistic awareness and may have contributed to improved comprehension and question generation. Future interventions should expand this integration by affirming students’ full communicative repertoires and encouraging strategic use of home languages and dialects during reading and self-monitoring. Such approaches align with linguistic justice and may support both comprehension and agency.
Limitations and Future Research
Several limitations warrant consideration. Given the small sample size and heterogeneity of learners (autism, developmental delay, bidialectal, and bilingual profiles), the findings cannot be generalized beyond this context.
Second, the use of researcher-developed passages and questions, though aligned with instructional targets, limits external validity and comparability with standardized measures.
Third, the short intervention period may have limited full internalization of the strategy, particularly for question generation. Future studies should extend the duration, diversify participant samples, and examine long-term maintenance and generalization.
In addition, the mismatch between performance and format preference highlights the need to investigate how modality, cognitive effort, and self-regulation intersect in comprehension tasks. Students with executive functioning differences may find digital interfaces more cognitively taxing. Future work should explore how explicit metacognitive reflection can help students evaluate their learning across formats and increase self-awareness.
Implications for Practice
The results of this study offer several important considerations for educators working with multilingual students who have autism or developmental delays. First, incorporating metacognitive strategies into reading instruction can support students in monitoring their comprehension and resolving ambiguity as they read. Structured routines such as Stop–Ask–Check–Connect (SACC) can be introduced through teacher modeling, supported with visual scaffolds, and reinforced through repeated opportunities for students to pause, reflect, and apply self-questioning techniques. These strategies encourage students to become more active readers who are capable of identifying when meaning breaks down and taking steps to repair it.
Second, anaphoric cueing provides a valuable scaffold for helping students make referential connections across sentences, particularly when understanding how pronouns relate to prior information. Although often overlooked in reading instruction for students with disabilities, teaching pronoun resolution explicitly can significantly strengthen literal comprehension. Educators can embed this work into existing routines by highlighting pronouns in text, modeling how to trace them to their antecedents, and prompting students to explain who or what each pronoun refers to. These practices help students construct coherent mental models of text and support their engagement with sentence-level meaning.
Third, instruction that affirms students’ full linguistic repertoire can promote both comprehension and engagement. In this study, the intervention integrated explicit comparisons across Arabic/English, AAVE/Mainstream American English, and Tagalog/English pronoun systems. These reflective conversations helped students notice differences such as subject omission, gendered pronouns, and emphatic repetition. For example, students connected English he to pronouns embedded in Arabic verb forms, reinforcing comprehension strategies while validating their home language knowledge. Such opportunities to examine how pronouns function across languages and dialects fostered contrastive awareness, positioning students’ everyday communication practices as resources for meaning making. Teachers can extend this approach by encouraging students to draw on familiar words, expressions, and background knowledge as entry points to comprehension, thereby affirming home languages as assets.
Finally, these findings underscore the value of student voice in instructional decision-making. Although students in this study performed more accurately with digital cueing, they consistently expressed a preference for traditional, paper-based materials. The divergence between accuracy and preference highlights the need for educators to balance student voice with performance data. Students may favor formats that feel familiar or less effortful, even when alternative modalities yield stronger outcomes. Embedding reflective conversations about modality may help students recognize how tools support comprehension. This is especially critical for students navigating complex linguistic, cognitive, and sensory demands in classrooms. When educators center student perspectives and explicitly support metacognitive growth, they foster environments that promote both access and agency in literacy development.
Conclusion
This study demonstrated that combining multimodal anaphoric cueing with metacognitive strategy instruction supported both comprehension and question generation. Embedding opportunities for cross-linguistic reflection positioned students as resourceful meaning-makers, while attending to their modality preferences acknowledged student agency. These findings have several implications. For classroom practice, teachers can integrate simple routines such as Stop–Ask–Check–Connect into shared reading lessons, using visual cues to model how pronouns connect across sentences. For IEP decision-making, teams can recognize and incorporate students’ bilingual or bidialectal strengths when setting literacy goals, affirming home language knowledge as an asset. For researchers, replication of this intervention with larger, more diverse samples is needed to examine generalization and sustainability, as well as to further explore how modality, metacognitive reflection, and cross-linguistic awareness intersect to shape comprehension growth. Taken together, the findings suggest that interventions combining multimodal cueing with metacognitive strategy instruction show promise, especially when paired with metalinguistic conversations that help students reflect across languages, deepen comprehension, and expand their agency as readers and thinkers.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
