When Input Is Not Enough: Cognitive Load, Dyslexia, and Instructional Architecture in ESL Contexts

Introduction

Exposure-based pedagogies, including communicative language teaching and task-based language teaching, continue to occupy a dominant position in English as a Second Language (ESL) instruction. Within these traditions, language learning is often understood as a process of gradual internalization driven by meaningful input, interaction, and repeated exposure to language in context (Ellis, 2019; Long, 2015). Learners are expected to extract regularities from use, stabilize patterns over time, and develop competence through communicative engagement.

In practice, this orientation is reflected in classrooms where vocabulary is introduced contextually, grammar is expected to emerge inductively, and learners are encouraged to engage in communicative production early in instruction. Such approaches are often viewed as learner-centered, authentic, and pedagogically desirable. Yet they also rest on an assumption that is not always made explicit: learners possess sufficient cognitive resources to process, store, and integrate linguistic input under conditions of simultaneous demand.

From the perspective of cognitive load theory, this assumption deserves closer scrutiny. Working memory is limited, and instructional tasks vary in the extent to which they impose extraneous demands on learners. When students must decode unfamiliar orthographic forms, maintain unstable phonological information, construct meaning, and produce language at the same time, learning may be constrained not by lack of exposure but by the structure of the task itself (Sweller, 2011; Sweller et al., 2011).

This issue becomes especially salient for learners with dyslexia. Dyslexia is typically associated with persistent difficulties in accurate and fluent word recognition, often linked to differences in phonological processing (Snowling et al., 2020). In first-language literacy, these differences affect decoding, spelling, and reading fluency. In additional-language learning, however, the consequences may be broader. If literacy mediates access to vocabulary, grammar, and classroom participation, then phonological instability may shape not only reading outcomes but also the conditions under which language input becomes learnable (Kormos, 2020; Kormos & Indrarathne, 2026).

These constraints are particularly important in English, whose orthography is comparatively deep and inconsistent. Learners must often coordinate phonological, orthographic, lexical, and semantic information simultaneously, and the burden of this coordination is not evenly distributed. For learners whose phonological processing is already effortful, repeated exposure may not automatically yield stable representations. Instead, input may remain only partially encoded, context-bound, or fragile under changing task demands (Ehri, 2014; Share, 2008; Verhoeven & Perfetti, 2022). The present article does not reject the value of input. Rather, it asks a narrower and more consequential question: under what cognitive conditions does input become durable knowledge? The argument advanced here is that instructional sequencing matters. Specifically, the order in which learners are required to coordinate phonological, orthographic, and communicative demands may shape whether exposure leads to robust literacy learning or merely to temporary task performance.

To examine this claim, the article reports findings from a classroom-based randomized controlled study comparing two instructional conditions: an exposure-oriented condition emphasizing contextual learning and early communicative use, and a structured condition emphasizing explicit phonological and orthographic stabilization before broader integration. The central hypothesis is that structured sequencing will support stronger literacy gains, particularly for learners with dyslexic profiles, because it reduces unnecessary cognitive competition during early encoding.

Theoretical Framework

Cognitive Load and the Processing Limits in ESL Learning

Cognitive load theory provides a useful framework for examining how instructional design interacts with learning processes. According to Sweller (2011), learning is constrained by the limited capacity of working memory, and instructional effectiveness depends partly on whether tasks direct cognitive resources toward relevant processing rather than unnecessary coordination. There is a crucial distinction between intrinsic load, which is tied to the inherent complexity of the material, and extraneous load, which is imposed by the way information is presented or tasks are organized. 

The following sections apply this framework to ESL learning. First, they identify the main sources of cognitive load in typical ESL tasks. Next, they examine how different instructional approaches either reduce or increase this load. Finally, they outline implications for designing instruction that supports efficient learning.

In ESL learning, the demands placed on working memory are frequently substantial. Learners need to process unfamiliar sound patterns, map these patterns onto orthographic forms, retrieve meanings, interpret syntax, and respond communicatively within the same instructional episode. When several interacting elements must be coordinated simultaneously, the risk of overload increases, especially for novice learners (Sweller et al., 2011).

The multicomponent model of working memory further clarifies why these demands matter. The phonological loop is central to the temporary maintenance and rehearsal of verbal information (Baddeley, 2012). If phonological representations are unstable or effortful to sustain, then more cognitive resources are consumed simply by maintaining verbal material in an active state. This leaves fewer resources available for integration, consolidation, and retrieval. Research in second-language reading has likewise shown that working memory is closely related to comprehension efficiency, especially when reading demands are high (Chow et al., 2021; Jeon & Yamashita, 2014).

Differences in phonological processing can increase the effort required to maintain and manipulate verbal information (Melby-Lervåg et al., 2012; Snowling et al., 2020). Under such conditions, tasks that appear ordinary for some learners may become cognitively dense for others. When decoding, comprehension, and production are combined too early or too loosely structured, the problem may not be insufficient exposure but insufficient processing stability.

Orthographic Mapping and Durable Learning

While cognitive load theory explains why learners may struggle to process instruction efficiently, orthographic mapping helps explain how durable word learning occurs. Orthographic mapping refers to the process through which phonological, orthographic, and semantic information become bonded in memory, enabling automatic word recognition and recall (Ehri, 2014). Once a word has been securely mapped, it no longer needs to be laboriously decoded each time it is encountered.

The process is necessary for fluent reading and accurate spelling. It also matters for vocabulary growth, because words that are efficiently encoded become easier to retrieve, recognize across contexts, and integrate into broader language use. On the other hand, when words are not mapped, learners tend to rely on context, partial visual familiarity, or short-term guessing strategies. Such strategies can support temporary task success but do not necessarily result in durable lexical knowledge.

This issue is particularly acute in English. When compared with more transparent orthographies, English presents learners with deep orthography, which increases the complexity of mapping print to sound (Share, 2008). Cross-linguistic work has confirmed that reading development varies substantially across orthographies and that orthographic depth shapes the demands learners face during literacy acquisition (Verhoeven & Perfetti, 2022).

In exposure-oriented instructional environments, vocabulary is often introduced through context before phonological and orthographic relationships have been stabilized. It facilitates semantic approximation, but it does not guarantee accurate encoding. Learners may recognize words in familiar tasks without developing the representational precision necessary for independent reading or spelling. 

These challenges are further intensified for learners with dyslexia, whose processing constraints make them particularly sensitive to increases in cognitive load and instructional inefficiencies.

Dyslexia, Variability, and Instructional Access

The implications of dyslexia for additional-language learning extend beyond low reading scores. Specific learning differences shape how learners interact with instructional environments, particularly those that rely heavily on implicit pattern extraction or simultaneous processing (Kormos, 2020; Kormos & Indrarathne, 2026). If the architecture of instruction assumes intact phonological processing, then learners with dyslexia may encounter barriers built into the design of the lesson itself.

One important consequence is variability in performance. Learners with dyslexia may appear inconsistent, succeeding in one task while struggling in another that seems superficially similar. Such inconsistency is often misread as inattentiveness, low effort, or lack of mastery. However, it may also reflect unstable encoding under changing processing conditions. A learner may perform adequately when contextual support is high and task demands are tightly constrained, yet they struggle when those supports are reduced.

From this perspective, instructional design affects not only average outcomes but also the reliability of performance. When tasks impose excessive simultaneous demands, small differences in timing, memory load, or representational clarity may produce large differences in observable performance. Structured instruction may therefore matter not simply because it raises scores, but because it stabilizes access to what learners know.

Building on these theoretical considerations, the present study examines whether structured instructional sequencing leads to stronger gains than exposure-oriented approaches in ESL learning, and whether these effects are more pronounced for learners with dyslexic profiles.

Research Questions

The present study addressed the following research questions:

1. Does structured instructional sequencing produce stronger gains than exposure-oriented sequencing in ESL reading fluency and spelling accuracy? 

2. Are the effects of instructional sequencing more pronounced for learners identified with dyslexic profiles?

Method

Research Design

The study employed a classroom-based randomized controlled design. This design was selected to balance internal validity with ecological validity by enabling systematic comparison of instructional conditions within authentic ESL learning environments. Rather than comparing entirely different curricula, the study compared two instructional sequences applied to the same lexical and grammatical content. This allowed differences in learning outcomes to be interpreted primarily in relation to instructional architecture rather than differences in content coverage or time on task. 

Participants

The study involved 320 English learners aged 10–12 enrolled in primary schools in Poland. All participants had received at least three years of prior English instruction. The sample represented a range of proficiency levels typical for this age group in school-based EFL settings. 

The sample comprised 160 learners with dyslexic profiles and 160 learners without dyslexic profiles. Following stratified randomization, each instructional condition included 80 learners with dyslexic profiles and 80 learners without dyslexic profiles, resulting in 160 participants in the exposure-oriented condition and 160 participants in the structured condition.

Learners with dyslexic profiles were identified on the basis of formal educational or psychological documentation available within the Polish education system, including persistent difficulties in decoding, reduced reading fluency, and weaknesses in spelling accuracy. This classification served as an operational research category rather than a substitute for independent clinical diagnosis.

Parental consent was obtained for all participants in accordance with ethical guidelines for research involving minors. Instructional implementation and data collection were conducted by trained classroom teachers following standardized procedures. The researcher had no direct contact with the child participants.

Instructional Conditions

Exposure-Oriented Condition

The exposure-oriented condition reflected widely used communicative and meaning-focused practices. Instruction emphasized contextualized language input, inductive grammar learning, task-based interaction, and early communicative production. Vocabulary and structures were introduced through meaningful context, and learners were expected to infer patterns through repeated exposure and use (Ellis, 2019; Long, 2015).

Structured Instructional Condition

The structured condition was designed to reduce unnecessary cognitive competition during early learning and to support stabilization of phonological and orthographic representations before broader communicative integration. Instruction emphasized explicit phoneme–grapheme alignment, systematic modeling of decoding processes, cumulative and spaced retrieval practice, controlled progression from recognition to production, and delayed introduction of open-ended communicative tasks.

This condition did not eliminate communication; rather, it sequenced communicative demands after learners had established more stable foundational representations. This sequencing logic is consistent with cognitive load accounts of instructional design (Sweller, 2011; Sweller et al., 2011).

Procedure

The intervention lasted 12 weeks and consisted of three instructional sessions per week, each lasting 25 minutes. Both groups received equivalent instructional time and were exposed to the same lexical and grammatical content. The difference between conditions lay in how that content was organized and sequenced. 

Instruction was delivered by trained teachers who completed pre-intervention training focused on condition-specific procedures, sequencing principles, and implementation consistency. Treatment fidelity was supported through written instructional protocols and was monitored using an implementation checklist completed throughout the intervention.

Participants completed pre-test measures before the intervention and post-test measures after the instructional period.

Measures

Reading fluency was assessed through timed English word-reading tasks measuring the number of words read correctly per minute. This measure was selected as an indicator of increasingly efficient decoding and automatized word recognition. Spelling accuracy was assessed through English spelling dictation tasks consisting of dictated words and sentences aligned with the instructional content. This measure was selected because accurate spelling requires stable and retrievable phonological–orthographic representations. Both outcome variables were administered at pre-test and post-test under standardized conditions.

Data Analysis

Data were analyzed using mixed-design analyses of variance (ANOVA), with Time as a within-subjects factor and Instructional Condition and Dyslexia Status as between-subjects factors. Analyses examined the main effect of Time, the interaction between Time and Instructional Condition, and the three-way interaction between Time, Instructional Condition, and Dyslexia Status.

Effect sizes were reported using partial eta squared ηₚ². Statistical significance was evaluated at the conventional threshold of p < .05.

Results

Reading Fluency

Analysis revealed a significant main effect of Time (F(1,316) = 182.47, p < .001, ηₚ² =.37), indicating overall improvement in reading fluency across the instructional period. The interaction between Time and Instructional Condition was also significant (F(1,316) = 24.83, p < .001, ηₚ² = .07). This interaction indicated that the magnitude of improvement differed between instructional conditions. Learners in the structured condition demonstrated greater gains in reading fluency than learners in the exposure-oriented condition.

Spelling Accuracy

A similar pattern emerged for spelling accuracy. There was a significant main effect of Time (F(1,316) = 156.91, p < .001, ηₚ² =.33), indicating overall improvement across participants. The interaction between Time and Instructional Condition was also significant (F(1,316) = 31.56, p < .001, ηₚ² =.09), indicating that gains differed by instructional condition. Learners in the structured condition demonstrated stronger improvement in spelling accuracy than learners in the exposure-oriented condition. No significant pre-test differences between groups were observed, supporting baseline comparability across conditions. 

Differential Effects for Learners with Dyslexic Profiles

A significant three-way interaction between Time, Instructional Condition, and Dyslexia Status was observed (F(1,316) = 9.62, p = .002, ηₚ² =.03). The interaction indicated that the advantage of the structured condition was amplified for learners identified with dyslexic profiles.

Thus, while structured sequencing benefited the sample as a whole, it was particularly beneficial for learners whose phonological processing differences would be expected to increase sensitivity to instructional load.

The results indicate that both instructional conditions led to measurable improvement over time, but structured sequencing produced substantially stronger gains in both reading fluency and spelling accuracy, particularly for learners with dyslexic profiles.

Pedagogical Implications

The findings have direct implications for ESL instruction, particularly in literacy-mediated classrooms involving learners with dyslexia or other phonological processing differences. The instruction should prioritize the stabilization of phonological and orthographic representations before introducing heavy communicative complexity. Learners may benefit from establishing accurate sound–symbol relationships and controlled decoding routines before being expected to interpret, manipulate, and produce new language in open-ended tasks.

Explicit attention to grapheme–phoneme correspondences should play a more central role in ESL instruction, particularly in English. Meaning-focused exposure is valuable, but it does not reliably guarantee precise phonological–orthographic encoding (Ehri, 2014; Verhoeven & Perfetti, 2022). Instructional tasks should be designed to manage cognitive load by limiting simultaneous demands. Rather than requiring learners to decode, interpret, and produce at the same time, instruction may be more effective when these demands are sequenced in a deliberate progression from recognition to integration to production (Sweller et al., 2011).

Cumulative retrieval practice should be incorporated systematically; retrieval strengthens memory traces and supports the movement from fragile knowledge to more stable, automatized performance (Roediger & Karpicke, 2006). Communicative instruction should be reframed rather than discarded. Communication remains the ultimate goal of language learning. The present findings suggest, however, that the route to communicative competence may need to be differently sequenced for learners whose phonological systems do not support efficient implicit extraction under high processing load.

Limitations and Future Research

The study focused on Polish English learners aged 10–12, which may limit its generalizability to other age groups, first-language backgrounds, or instructional settings. Secondly, although the randomized design strengthens internal validity, some variation in instructional implementation across classroom sites was unavoidable. Future studies should incorporate stronger fidelity measures and, where possible, multilevel analytic approaches that account for classroom-level clustering.

Moreover, the term dyslexia was operationalized through available educational and psychological documentation rather than through independent diagnostic assessment for research purposes. It increases ecological validity but may also introduce heterogeneity within the dyslexic-profile subgroup.

Finally, the study focused on reading fluency and spelling accuracy rather than broader outcomes such as oral proficiency, written composition, or long-term retention. Future research should examine whether the benefits of structured sequencing extend to these domains and whether they remain stable over time.

The study also did not directly measure cognitive load during task performance. Process-based measures could help clarify the mechanisms through which instructional sequencing influences learning.

Conclusion

Exposure is a necessary but not sufficient condition for ESL learning. The findings of the present study suggest that the effectiveness of input depends on the cognitive conditions under which it is processed. For dyslexic students, phonological processing differences and working-memory constraints may shape whether exposure leads to durable encoding or only temporary performance.

Structured instructional sequencing appears to support stronger literacy outcomes without increasing instructional time. The central implication is not that communication should be abandoned, but that communicative goals may be better served when instruction first secures the representational foundations on which fluent reading, accurate spelling, and stable language learning depend. For learners whose processing systems differ from the assumed norm, instructional architecture is not secondary, but fundamental.

Anna Karwowska is an educational researcher and language specialist, and the author of the Dyslexia in the ESL Classroom method, focused on evidence-based instruction for learners with dyslexia and ADHD.

References

Baddeley, A. D. (2012). Working memory: Theories, models, and controversies. Annual Review of Psychology, 63, 1–29. https://doi.org/10.1146/annurev-psych-120710-100422

Chow, B. W.-Y., Mo, J., & Dong, Y. (2021). Roles of reading anxiety and working memory in second language reading comprehension. Learning and Individual Differences, 92, Article 102092. https://doi.org/10.1016/j.lindif.2021.102092

Ehri, L. C. (2014). Orthographic mapping in the acquisition of sight word reading, spelling memory, and vocabulary learning. Scientific Studies of Reading, 18(1), 5–21. https://doi.org/10.1080/10888438.2013.819356

Ellis, R. (2019). The study of second language acquisition (2nd ed.). Oxford University Press.

Jeon, E. H., & Yamashita, J. (2014). L2 reading comprehension and its correlates: A meta-analysis. Language Learning, 64(1), 160–212. https://doi.org/10.1111/lang.12034 

Kormos, J. (2020). Specific learning difficulties in second language learning and teaching. Language Teaching, 53(2), 129–143. https://doi.org/10.1017/S0261444819000442

Kormos, J., & Indrarathne, B. (2026). Specific learning differences in learning, teaching, and assessing additional languages. Language Teaching, 59(2), 139–163. https://doi.org/10.1017/S0261444825100803

Long, M. H. (2015). Second language acquisition and task-based language teaching. Wiley-Blackwell.

Melby-Lervåg, M., Lyster, S. -A. H., & Hulme, C. (2012). Phonological skills and their role in learning to read: A meta-analysis. Psychological Bulletin, 138(2), 322–352. https://doi.org/10.1037/a0026744 

Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x

Share, D. L. (2008). On the anglocentricities of current reading research and practice: The perils of overreliance on an “outlier” orthography. Psychological Bulletin, 134(4), 584–615. https://doi.org/10.1037/0033-2909.134.4.584

Snowling, M. J., Hulme, C., & Nation, K. (2020). Defining and understanding dyslexia: Past, present and future. Oxford Review of Education, 46(4), 501–513. https://doi.org/10.1080/03054985.2020.1765756

Sweller, J. (2011). Cognitive load theory. In J. P. Mestre & B. H. Ross (Eds.), Psychology of learning and motivation (Vol. 55, pp. 37–76). Elsevier. https://doi.org/10.1016/B978-0-12-387691-1.00002-8

Sweller, J., Ayres, P., & Kalyuga, S. (2011). Cognitive load theory. Springer. https://doi.org/10.1007/978-1-4419-8126-4 

Verhoeven, L., & Perfetti, C. (2022). Universals in learning to read across languages and writing systems. Scientific Studies of Reading, 26(2), 150–164. https://doi.org/10.1080/10888438.2021.1938575