The Distance Between What the Text Says and What a Reader Supplies

Consider this short passage, created for this article.

Now ask what the coat might suggest. One possible response is:

Maybe she forgot to take it off.

The passage never says Ines forgot anything, so a reader who says she did has supplied a reason for something the text only shows. That is an inference, and a sensible one: it makes her behavior make sense. What it does not do is take up anything else on the page.

What the thin answer did, and what it left

Keep observation separate from inference, because the next few minutes depend on the difference. The text states that the coat stays on through the lesson, that the radiators tick, that the boy beside her has pushed his sleeves up and is fanning himself, that she hangs her bag on the hook whose label has peeled away, that she takes it down at the bell before anyone else has moved, and that she walks home with the coat still buttoned. She forgot, she is cold, she is unsettled are inferences. The observations are what any of them has to survive.

Two first thoughts survive the whole passage. She is cold survives because people feel cold in warm rooms. She forgot survives because absent-mindedness can last a day, and a writer can show it twice. Neither is refuted by anything here, and I would be ready to hear either argued well. A reader could build a good reading out of forgetfulness: absent-mindedness set against a boy minutely aware of the heat is a contrast the passage has arranged. What maybe she forgot has not done yet is pick up the bag, the hook, the bell or the last sentence.

Set a fuller reading beside it, not as the right answer but to show the difference in kind. Suppose a reader says the passage is showing someone who stays ready to leave. The coat stays on through the first lesson, in a room the passage has gone out of its way to make warm. The bag goes up on the hook and comes down at the bell, before anyone else has moved. The coat is still buttoned on the walk home.

That reading is not better for being more elaborate. It is doing a different job. She forgot supplies a reason for one action and stops there. She stays ready to leave proposes an idea, and the idea has to hold across a coat, a bag and a bell, which is what makes it something the class can argue with. Both students have made an inference. The second connects several details rather than explaining the coat alone. What a teacher needs to know is not which answer is right but which kind is on the page, because that decides what to ask next.

One limit holds for either reading: it is a claim about the passage, not about Ines. The text does not report everything she did, and she did not settle in would go further than seven sentences allow.

Choosing a detail and saying what it does for a reading are separate jobs, and the details here do not all do the same amount of work.

The boy with his sleeves to the elbow does nothing alone. Set beside the coat, it makes a contrast: someone in the same air has adjusted to it and she has not. That weakens a reading built on a cold room without closing it off, since people do differ.

The bag can support the reading by recurrence. It comes down at the bell, before anyone else moves, and grouping that with the coat proposes one disposition behind both: keeping something near her rather than committing it to the room. That grouping is itself the claim, and it is arguable. The same detail cuts the other way too, because she did hang the bag up, and putting your bag on a hook is what settling in looks like. A student who notices that has found the strongest objection the passage offers.

The peeled label gives the weakest support. It may suggest a place that is not quite hers, which fits closely, and it may equally be ordinary wear. A detail can be consistent with a reading without carrying it, and saying so plainly is part of explaining your evidence.

That is maybe fifteen minutes of thinking. What I wanted to know is whether a Grade 7 class can be walked through it, and if not, why not.

Can a thirteen-year-old do this?

The starting point is that the age pattern is real and large. In Peskin and Wells-Jopling’s (2012)detailsFostering symbolic interpretation during adolescenceJournal of Applied Developmental Psychology, 2012 · Classroom experiment, six classes in Grades 6, 9 and 12 (45, 42 and 50 students), assigned by class, one teacher, poemsRead in the author post-print, not the version of record. No per-grade test is reported, and no delayed measure existsStudy quality: C, small classroom experiment with classes rather than students assigned and analyzed at student level · Confidence for the use made here: moderate for the overall scaffolded-versus-comparison difference and the pretest spread; low for the single new poem, whose denominators and any statistical test could not be confirmed and which is reported here as descriptive counts.View original study → study of six classes in Grades 6, 9 and 12, the proportion of poem responses that were symbolic before any of the teaching was roughly a fifth in Grade 6, approaching half in Grade 9, and about seven in ten in Grade 12.

A Grade 12 class differs from a Grade 6 class in more than age. They have been in school longer, read more, encountered more literary teaching of some kind and some quality, and they may differ in who is still in the class. I cannot tell you how much instruction in symbolic reading any of those students had received, and neither can the study. The spread is real, and its causes are not apportioned.

So the useful question is not whether older students do better, which they plainly do. It is whether something in a younger reader makes this unavailable, or whether what they are missing is knowledge, practice and a clear enough question.

One line of evidence on the first half comes from outside literary reading. Barnes, Dennis and Haefele-Kalvaitis (1996)detailsThe effects of knowledge availability and knowledge accessibility on coherence and elaborative inferencing in children from six to fifteen years of ageJournal of Experimental Child Psychology, 1996 · Two laboratory experiments, children aged 6 to 15, invented knowledge baseDifferences in vocabulary and language were not controlled, and the study does not claim they were.Study quality: B, laboratory experiments on an invented knowledge base · Confidence for the use made here: moderate for the direct-versus-unprompted and persistent-age-difference results; low for anything about literary reading, which it did not study.View original study → taught children aged six to fifteen a set of facts invented for the study about an imaginary world, drilling them until every child could recall them, then asked questions about a story that depended on those facts. Two results matter here. Children at every age answered direct questions far better than they drew the same inferences unprompted, and among the youngest, coherence inferencing roughly doubled when the question asked for it. And age differences in inferencing remained even with those facts equally recallable, which led the authors to conclude that some basic limitation remained in the younger children. The task was simple inference questions in a made-up world rather than literary interpretation, so it bears on this obliquely. Within those limits, age mattered.

Set against that is a result from the other end of the age range. McCarthy and Goldman (2015)detailsComprehension of short stories: Effects of task instructions on literary interpretationDiscourse Processes, 2015 · Two experiments, college students with no prior training in literary analysis, four task instructionsRead in the accepted author version, consulted in full by an independent verifier; the version of record was not reached. No figures are quoted. Both experiments used allegorical short stories; generalization to other fiction was not tested. The study says nothing about younger readersStudy quality: B, controlled experiments with adult readers · Confidence for the use made here: moderate for task sensitivity in adults.View original study → gave college students with no prior training in literary analysis a short story to read under one of four task instructions, and the instructions pushing toward argument and theme produced essays with more interpretive inferences than the plot instruction did. Burkett and Goldman (2016)details,“Getting the point” of literature: Relations between processing and interpretationDiscourse Processes, 2016 · Two studies, 80 undergraduates with no formal training in literature, think-alouds during reading and a post-reading interviewRead at abstract level only. The reading and interview conditions differ in several respects at once, including being asked at all, so this is not evidence about response modeStudy quality: B, descriptive and correlational with adult readers · Confidence for the use made here: moderate for noticing without production.View original study → in two studies of eighty undergraduates without formal literature training, found that their novices noticed cues to interpretive meaning while reading but were more likely to produce symbolic and thematic interpretations when asked afterward than while reading. Staying literal is not only a children’s pattern. These were adults who did reach for interpretation, and how much of it arrived on the page moved with the task.

One connection I would propose rather than assert. PISAdetailsPISA 2025 results (Volume I): Future-ready studentsOECD Publishing, 2026 · Over 760,000 students in 91 countries and economies, data collected 2025Study quality: A, the strongest descriptive evidence available · Confidence for the use made here: high for the item-level patterns as reported; the report assigns no cause, studies no literary interpretation, and is used here only as a proposed connection about which reading demand is under pressureView the report → tests fifteen-year-olds, and fifteen-year-olds have not produced the same result from one cycle to the next: reading fell across the OECD between 2018 and 2022, and again to 2025, which I wrote about in The Reading Crisis Began Before COVID. In that first interval the questions that fell most asked students to hold several pieces of information in mind at once and judge them, while those that held up best asked them to locate one stated piece of information; over the full period to 2025 the declines were more uniform across kinds of task. PISA works mostly with informational material and studies no literary interpretation, so it is not measuring what the poem and story studies measure, and the report assigns no cause. What it does show is that the same age has produced different results in different years. Whatever is moving, it is not how old the students are.

The spread is not a ceiling

The same Peskin and Wells-Jopling study that shows the grade spread also shows what happens when you teach into it. Two classes per grade, one taught by the same teacher with three concrete scaffolds over about two and a half weeks and one without. Afterward the scaffolded classes read more of six poems symbolically than the comparison classes, adjusting for pretest scores and a measure of verbal reasoning.

The scaffolds are concrete enough that a department could build them in an afternoon.

• A wheel for tracking a metaphor as it recurs through a poem.
• Venn diagrams setting the thing described against the thing it is described as.
• Two-line image metaphors scrambled, for students to reassemble.

Five of the six test poems had appeared at both pretest and posttest. On the sixth, met for the first time at the posttest, twenty-six students in the scaffolded classes gave a symbolic reading against seven in the comparison classes, which is what the paper reports descriptively for that poem at the end of the unit. Grade 6 classes took part, and the reported overall effect does not establish effectiveness separately at that grade. No later test was given, so none of this speaks to whether it lasts.

Part of what those scaffolds hand over is knowledge of where to look. Rabinowitz described conventions readers learn to use, and Goldman, McCarthy and Burkett (2015)detailsInterpretive inferences in literatureIn Inferences during reading (Chapter 17, pp. 386–415), Cambridge University Press, 2015 · Theoretical chapterRead in the authors’ pre-publication draft of 10 March 2014, not the published version. The account of reading conventions, and the quoted phrase “really more heuristics than rules,” are drawn from that draft and may differ from the published chapter. Rabinowitz (1987) was not retrieved and is attributed through this chapter.Study quality: not a study · Confidence for the use made here: moderate to high as a framework, with the version caveat aboveView the chapter → summarize them as directing readers to prioritize details in privileged positions, such as titles and first or last sentences, or that are stressed, or that repeat. They call these “really more heuristics than rules,” and the hedge matters: a detail can be marked and be doing something modest, or several things at once. Look at the Ines passage with that in mind. The coat opens it and closes it, and the bag between them repeats its logic, which is why the coat and not the worksheet is where a practiced reader’s attention goes. Some of that is a learned convention rather than a knack, and the conventional part can be told to a twelve-year-old in a sentence.

The authors read their results as not indicating a fixed developmental timetable from age eleven, which is their reading of their own data. A narrower version is enough here: where a class sits before this kind of teaching does not tell us where it can get to.

What instruction has actually taught

Two classroom studies describe what the teaching looked like in enough detail to borrow from, and they arrived at the same starting move independently.

Levine (2014)detailsMaking interpretation visible with an affect-based strategyReading Research Quarterly, 2014 · Four-week classroom study, two Grade 12 classes (19 and 18 consenting students), one high-poverty urban high school; two teachers and two sets of textsThe design cannot attribute the difference to any single feature, and the author says soStudy quality: C, quasi-experimental with intact classes and unmatched teachers and texts · Confidence for the use made here: moderate for the written-task difference, high for the description of what each class was taught.View original study → worked with two twelfth-grade classes in a high-poverty, low-achieving urban high school over four weeks. Her intervention taught a three-step heuristic, which the teacher wrote on the blackboard: identify language that feels emotionally charged; evaluate its valence, meaning whether it feels positive, negative or both; explain or justify that evaluation.

Where the practice started is the part I would copy. Before any literary text, students compared synonyms for woman, and the stage name Marilyn Monroe against Norma Jeane Mortenson, and newspaper headlines that reported one event as “Police Officer Controls Riot” or “Cop Suppresses Gathering,” and then advertisements. In each case they said which language carried more feeling, which way, and why. The teacher told them this was something they already did outside school and that they were now taking it apart and trying it on other kinds of texts.

Then they took it to a version of Cinderella. Given “It was afternoon. Cinderella was left at home, weeping by the ashes,” students judged the second sentence the more affect-laden, picked out left, weeping and ashes, and justified the judgment by saying that in context those words created abandonment and hopelessness. Most read the story’s overall effect as positive, explaining it with thematic statements like hope will win out in the end. A few read it as negative, on the grounds that in Cinderella the only way a woman succeeds is through a man.

That disagreement is worth sitting with. One class, the same procedure, the same text, and two defensible opposite readings arrived at from evidence rather than from inattention. It is what a room looks like when the thinking is actually happening.

The comparison class was not left alone, and that is what makes the contrast informative. That teacher also taught literary interpretation, explicitly: the same schoolwide argument template, imagery and symbols defined and pointed out, discussions of what the symbols meant in relation to the poems’ themes, and an explicit heuristic of her own, looking for patterns of imagery, modeled on the first stanza of Langston Hughes’s “Harlem Sweeties.”

What she did not do, in Levine’s account, was connect that heuristic to anything students already did outside school, or model the move from concrete imagery to abstract meaning. On a whole-class written task the two groups were comparable before the unit and differed afterward, the intervention class gaining and the comparison class not.

What that comparison cannot do is tell us which difference mattered. The classes had different teachers and different texts, which Levine counts among the study’s main limitations. What it does do is move the question off instruction against no instruction. Both classes were taught to interpret literature, explicitly, and they did not end in the same place.

Sosa, Hall, Goldman and Lee (2016)detailsDeveloping symbolic interpretation through literary argumentationJournal of the Learning Sciences, 2016 · Mixed-methods case study, two Grade 9 sections in a support program, 22 students with both essays, no comparison groupShifts were tested with Wilcoxon signed-ranks tests, one per dimension against that dimension’s own baseline, so the results support no comparison between dimensions and no conclusion that a nonsignificant dimension did not change; the paper names the tests but prints no test statistics. Verified against the journal-formatted Project READI copy, pp. 122–123 and Table 5.Study quality: C, single case study without a comparison group · Confidence for the use made here: moderate as a description of what one module taught and what this cohort’s essays showed.View original study → ran a five-week module, twenty-one lessons, with two ninth-grade sections in a program for students who needed extra support, identified partly by reading scores below the thirtieth percentile.

That module also began away from literary text. Students first worked on elements of argument and on what counts as coming of age. Symbols were then introduced through what Carol Lee calls cultural data sets: the American flag, hearts, roses, and pop songs with the lyrics in front of them, texts in which the symbolic reading is already available. The question put to students was metacognitive. Not what does this mean but how do you know. They explained, for instance, how lines in Alicia Keys’s “Girl on Fire” rule out a literal reading of the title.

From there the module moved in three phases: identify a symbol and say why it is one, from the associations it carries; make an interpretive claim grounded in those associations and in evidence from the text; then do the same from patterns of evidence across a longer text. Three worksheets carried the phases, each adding a column. The first asked for the symbol, the associations and what it represents. The second added what the text actually says, with a page number, and then a column asking what the words and the associations together lead the student to think the symbol means.

Neither module was shy about direct teaching. Sosa’s taught what a symbol is and how to claim one, in as many words; Levine’s taught a named procedure and wrote it on the board. The difference was not explicitness but what the explicit teaching was routed through. Two studies, in different years and with different age groups, independently built on material where their students were already reading for significance. Neither design can tell us this was the feature that worked. It gives me a place to start: not with what a symbol is, but with a text where the class is already doing the thinking, and how do you know put to work on it.

What the essays still did not show

Sosa’s twenty-two students wrote an essay before the module and another afterward, on stories they had not studied, scored on several dimensions. Three shifted significantly: whether students made accurate claims, what those claims were about, and what they wrote about symbolism. No student identified anything as symbolic in the first essay; afterward thirteen of the twenty-two did, and six of those thirteen interpreted what they had identified.

Five dimensions showed no significant shift, among them the use of evidence and the explicitness of reasoning. After the module about a fifth of the students included explicit reasoning about why evidence supported a claim, and about two fifths had nothing indicating why their evidence supported anything. Two limits ride with those numbers. With twenty-two students, a nonsignificant result does not establish that nothing changed; and because each dimension was tested against its own baseline, none of this shows that one dimension improved more than another. What the essays do show is that explicit explanation was still uncommon at the end of a module that taught argument and interpretation directly.

The authors saw the same thing in the room: students appeared to lack the language for expressing how and why evidence supported a claim, and the teacher supplied that language on several occasions in the lessons they analyzed.

Taking that as a teaching problem rather than a stage to wait out is a professional judgment, not a finding. Lewis and Ferretti (2011)detailsTopoi and literary interpretation: The effects of a critical reading and writing intervention on high school students’ analytic literary essaysContemporary Educational Psychology, 2011 · Six high school students taught literary topoi, argument from topoi, quotation and tie-in sentencesRead at abstract level only; the full text, including the design and measures, was not retrieved. Used as a reason to try teaching the link between evidence and claim explicitly, not as evidence that doing so is necessary or generally effective. The authors’ framing of high school students as generally struggling to justify interpretations is their reading of the field, citing Marshall (2000), not something six participants establishStudy quality: C, six participants · Confidence for the use made here: low.View original study → went at it directly, teaching six high school students to build an argument from a pattern in the text, use direct quotation, and write tie-in sentences linking the quotation to the argument, and they report higher-quality essays with more textual evidence. Six students, and I have read only the published abstract, so I offer it as a reason to try something rather than as a method. I have written about that sentence before, in The Sentence After the Quotation. Sosa’s third worksheet column reaches the same object from another direction: it gives the explanation a place on the page, which is the simplest thing here to copy tomorrow.

Three things to look at separately

Working the Ines passage the long way separates three things that a single mark in a margin usually collapses.

Proposing a reading is offering an account of what the details suggest. She is keeping herself ready to leave is a proposal.

Selecting evidence is deciding which details the reading rests on. Here the coat, the bag and the walk home carry the pattern, while the radiators and the boy’s sleeves carry the contrast that gives it force. Evidence does not have to repeat the same motif to support a claim, and one detail connected exactly can do more than four gathered loosely.

Explaining why the evidence supports it is saying what the chosen detail does for the claim, and why it favors this reading over a rival. The three paragraphs about the boy, the bag and the label are three attempts at that, and they are not equally successful.

A student can do the first and not the second, or the first two and have nothing to say about the third, which produces a kind of answer every English teacher recognizes: a plausible claim, a correctly chosen quotation, and then a sentence that restates the claim instead of connecting them. They are not stages, and nothing here shows that one is harder than another. What matters at a desk is that an answer can show one without the others.

Back to the coat

So what does a thin answer to what might the coat suggest? give me to work with? An explanation offered for one action, and an open question about how much more this reader could do with the passage. Three things to look for, each with its own next move.

If no reading has been proposed, I would ask again with the passage in view: not why Ines keeps her coat on, but what the passage is doing by keeping it on her through to the last sentence. The first question invites an explanation, the second a reading. Often enough, the thin answer was an accurate answer to the question I actually asked.

If a reading arrives with nothing under it, I would ask which details it rests on, and listen for whether the boy with his pushed-up sleeves arrives alongside the coat. Choosing evidence is a separate thing to teach from having something to say, and what I am listening for is fit, not a count.

If a quotation arrives and the sentence after it restates the claim, I would ask what makes the link visible. How does that detail support what you said? How much weight will it carry? And what in the passage makes your reading harder to hold?

The last two are not the same question, and the passage answers them differently. The peeled label is a question of weight, and calling it consistent without being decisive is the work least visible in a finished essay. Hanging up the bag is the complication. Saying what cuts against your reading is not a concession, and where it leads is open: the reading may survive with a reason, narrow to something more defensible, or give way to a better one. Any of those is the work going well.

Age is in this. A twelve-year-old and an eighteen-year-old do not arrive with the same reading history, and nothing here says everything is available at every age. What the evidence does not show is a ceiling. Classes taught with three scaffolds for two and a half weeks read more of six poems symbolically than classes taught without them, and in Sosa’s posttest essays the explaining stayed rare even after a module that gave it a column on the worksheet. One of those is a reason to teach this directly; the other is a reason to keep checking whether it arrived. Where a class starts is not where it has to finish, and none of these questions requires knowing which ability a student lacks. They tell me where this response needs more work, which is the part I can do something about.

Research cited

Barnes, M. A., Dennis, M., & Haefele-Kalvaitis, J. (1996). The effects of knowledge availability and knowledge accessibility on coherence and elaborative inferencing in children from six to fifteen years of age. Journal of Experimental Child Psychology, 61(3), 216–241. Two laboratory experiments, children aged 6 to 15, invented knowledge base. https://doi.org/10.1006/jecp.1996.0015

Burkett, C., & Goldman, S. R. (2016). “Getting the point” of literature: Relations between processing and interpretation. Discourse Processes, 53(5-6), 457–487. Two studies, 80 undergraduates with no formal training in literature, think-alouds during reading and a post-reading interview. https://doi.org/10.1080/0163853X.2016.1169969 Read at abstract level only.

Goldman, S. R., McCarthy, K. S., & Burkett, C. (2015). Interpretive inferences in literature. In E. J. O’Brien, A. E. Cook, & R. F. Lorch, Jr. (Eds.), Inferences during reading (Chapter 17, pp. 386–415). Cambridge University Press. Theoretical chapter. https://doi.org/10.1017/CBO9781107279186.018 Read in the authors’ pre-publication draft, not the published version.

Levine, S. (2014). Making interpretation visible with an affect-based strategy. Reading Research Quarterly, 49(3), 283–303. Four-week classroom study, two Grade 12 classes (19 and 18 consenting students), one high-poverty urban high school; two teachers and two sets of texts. https://doi.org/10.1002/rrq.71

Lewis, W. E., & Ferretti, R. P. (2011). Topoi and literary interpretation: The effects of a critical reading and writing intervention on high school students’ analytic literary essays. Contemporary Educational Psychology, 36(4), 334–354. Six high school students taught literary topoi, argument from topoi, quotation and tie-in sentences. https://doi.org/10.1016/j.cedpsych.2011.06.001 Read at abstract level only.

McCarthy, K. S., & Goldman, S. R. (2015). Comprehension of short stories: Effects of task instructions on literary interpretation. Discourse Processes, 52(7), 585–608. Two experiments, college students with no prior training in literary analysis, four task instructions. https://doi.org/10.1080/0163853X.2014.967610

OECD (2026). PISA 2025 results (Volume I): Future-ready students. OECD Publishing. Over 760,000 students in 91 countries and economies, data collected 2025. https://doi.org/10.1787/73451bc5-en

Peskin, J., & Wells-Jopling, R. (2012). Fostering symbolic interpretation during adolescence. Journal of Applied Developmental Psychology, 33(1), 13–23. Classroom experiment, six classes in Grades 6, 9 and 12 (45, 42 and 50 students), assigned by class, one teacher, poems. https://doi.org/10.1016/j.appdev.2011.08.002

Sosa, T., Hall, A. H., Goldman, S. R., & Lee, C. D. (2016). Developing symbolic interpretation through literary argumentation. Journal of the Learning Sciences, 25(1), 93–132. Mixed-methods case study, two Grade 9 sections in a support program, 22 students with both essays, no comparison group. https://doi.org/10.1080/10508406.2015.1124040

Each citation above carries two short judgments, visible when you open it. Study quality rates the design’s rigor: A for preregistered or peer-reviewed work with strong designs or large meta-analyses, B for solid single studies or secondary reports, C for preprints or smaller, quasi-experimental designs. Confidence in the conclusion is Northlight’s own judgment of how much weight a finding should carry for the specific point it supports here, which can be lower than the study quality alone would suggest.

Research & AI transparency

AI tools assisted with locating, organizing, comparing, and checking research, and with drafting and revising parts of this article. The tools were not treated as sources. Sources were checked against the originals. The Ines passage and sample response are invented examples created for this article. Northlight Studies determined the question, interpretation, strength of the claims, position taken, and classroom implications.