The research, conducted by Jagdeep Panchal of Koneru Lakshmaiah Education Foundation in Hyderabad, India, examined 115 Grade 4 and Grade 5 learners at a CBSE-affiliated private English-medium school in Karnataka. The participants, aged roughly nine to eleven, came from multilingual homes in which Kannada was the primary language for most, with smaller groups speaking Tamil, Telugu, Malayalam, Marathi, or Hindi. English served as the medium of instruction but not as the home language, making the classroom a realistic testing ground for how young English language learners negotiate narrative texts.

The study’s central innovation lies in its refusal to treat reading comprehension as a single score. Drawing on decades of comprehension theory, including Walter Kintsch’s construction-integration framework and Paul van den Broek’s landscape model of reading, the study separated reading performance into four dimensions. Two were processing-oriented: Inferential Processing, the ability to interpret implicit meanings and unstated relationships, and Structural Coherence Processing, the ability to recognize causal and organizational links across narrative events. Two were comprehension-oriented outcomes: Micro-Level Comprehension, covering character perspectives, emotional nuance, and localized meaning, and Macro-Level Comprehension, covering plot development, thematic relationships, and global narrative meaning.

To isolate the effect of narrative perspective, Panchal adapted two graded-reader stories, one realistic and one non-realistic, into parallel first-person and third-person omniscient versions. Storyline, event sequence, characters, settings, and themes were held constant across versions; only pronouns, perspective markers, and narrative voice were altered, with minimal linguistic adjustment. Each version ran approximately 220 to 250 words, and three specialists in English language teaching and reading instruction reviewed the materials for equivalence in length, vocabulary, and sentence complexity. The goal was to ensure that any performance differences could be plausibly attributed to narration condition rather than to broader textual variation.

The design was a counterbalanced quasi-experiment embedded in ordinary classroom routines. Four intact class groups completed reading sessions under both narration conditions, with Groups 4A and 5A starting with first-person texts and Groups 4B and 5B starting with third-person versions. Sessions were separated by two to three days to reduce immediate recall effects, learners read independently within supervised 30 to 40 minute periods, and the same instructor administered every session under standardized instructions. Participants were not told that narrative point of view was the focus of the study, a precaution designed to minimize response bias. Each text was followed by a 32-item assessment comprising eight items per dimension, mixing multiple-choice, short constructed-response, and extended-response formats.

The statistical results revealed a striking crossover pattern. On composite scores, third-person omniscient narration produced a modest but significant overall advantage, with a mean of 53.68 compared with 51.12 under first-person narration, a difference of roughly 2.6 points. A repeated-measures analysis of variance confirmed a significant main effect of narration condition, F(1, 113) = 83.70, p < 0.001, with a large effect size of partial eta squared = 0.425. But the more revealing finding emerged when the four dimensions were analyzed separately: the Narration by Dimension interaction was highly significant, F(3, 342) = 132.24, p < 0.001, indicating that the two narration conditions produced genuinely different performance profiles rather than a uniform boost.

Under first-person narration, learners scored higher on Inferential Processing, with a mean of 13.60 versus 12.80, and on Micro-Level Comprehension, 13.50 versus 12.80. Under third-person omniscient narration, the pattern reversed: Structural Coherence Processing rose to 14.10 versus 12.00, and Macro-Level Comprehension reached 14.00 versus 12.10. The interpretation aligns with the study’s conceptual framework. Because first-person narration filters information through a single character-bound viewpoint, readers must work harder to infer motivations and unstated relationships, engaging perspective-taking and local inference. Third-person omniscient narration, by contrast, offers broader access to events, multiple perspectives, and narratorial explanation, providing scaffolding for coherence building and global integration.

A second significant interaction involved development. Grade level alone showed no main effect on overall performance, F(1, 113) = 0.84, p = 0.363, yet the Narration by Grade interaction was significant, F(1, 113) = 13.50, p < 0.001, with a moderate effect size of 0.107. Grade 5 learners outperformed Grade 4 learners more noticeably under first-person narration, where inferential and perspective-dependent demands are greatest, while grade differences shrank under third-person narration, whose structural support appears to flatten developmental gaps. This suggests that developmental differences in reading are not fixed traits but are expressed more or less visibly depending on the narrative context in which reading occurs.

The reliability of the assessment supported these conclusions. Cronbach’s alpha coefficients for the composite scores reached 0.90 under first-person and 0.92 under third-person conditions, and dimension-level coefficients ranged from 0.66 to 0.77, acceptable for a classroom-based multidimensional instrument. Scoring consistency was checked by re-evaluating roughly 20 percent of responses after an interval, yielding approximately 88 percent agreement. The study’s limitations are acknowledged: it involved a single school, a narrow grade range, and only two narration types, leaving open questions about third-person limited narration, second-person narration, shifting focalisation, and multimodal formats, as well as about generalization across proficiency levels and educational settings.

The pedagogical implications may be the study’s most consequential contribution. Reading materials in English language teaching are typically chosen by topic, vocabulary level, or readability formula, with narration type rarely considered. These results suggest that narrative point of view deserves a seat at the selection table. Teachers aiming to develop inference, character interpretation, and perspective-taking may find first-person texts naturally suited to those goals, while those targeting summarization, thematic analysis, and narrative organization may benefit from third-person omniscient material. Neither perspective is superior; each supports a different dimension of meaning construction. As the author concludes, narrative perspective may function not merely as a stylistic feature of stories but as a meaningful component of instructional text design, one that interacts with learner development in ways that reading pedagogy is only beginning to map.

Subject of Research: Effects of narrative point of view on multidimensional reading comprehension in primary English language learners

Article Title: Narrative point of view as an instructional variable in English reading classrooms: Differential patterns of multidimensional reading performance

Article References: Panchal, J. (2026). Narrative point of view as an instructional variable in English reading classrooms: Differential patterns of multidimensional reading performance. Frontiers in Language Sciences, 5, Article 1813257. https://doi.org/10.3389/flang.2026.1813257

Image Credits: AI Generated

DOI: 10.3389/flang.2026.1813257

Keywords: narrative point of view, reading comprehension, English language teaching, first-person narration, third-person omniscient narration, inferential processing, structural coherence, primary education, multidimensional assessment, young learners, text design, quasi-experimental study