Overview:
Inferential thinking is essential to reading comprehension, yet many students struggle to connect ideas, interpret implied meaning, and explain what a text leaves unsaid. Here’s how teachers can help.
If you were given an article to read and then asked what a particular sentence meant, how would you respond? You would do your best to provide the meaning. And if you were asked to explain the main message of the article, you would recall the information, consolidate your understanding, and then give the main idea.
So far so good.
But, how would you respond if you were asked to critically comment on whether that same sentence provided an effective conclusion to the main message of the article?
To answer fully, there are two steps: First, explain the sentence to unpack its meaning. You get 1 out of 2. But there is a second.
Use the meaning of the sentence and link this to the article’s main message. You’ve done it. But subconsciously, you’ve also employed inference: the ability to derive meaning that is implied rather than explicitly stated.
The Data
In a recent English test, 28% of our learners got 2 out of 2; that is 17 learners out of 60. Against the research, though, that statistic is actually above average.
Every three years, the Organization for Economic Co-operation and Development (OECD) undertakes the Program for International Student Assessment (PISA). The study evaluates education systems by measuring the performance of 15-year-olds in mathematics, science, and reading. In one of its latest assessments, PISA 2022, only around 7% of students, on average across OECD countries, reached Levels 5 and 6 – the highest levels of reading proficiency.
Learners like these are characterized with the ability to understand lengthy texts, infer relevant information and perform causal reasoning within abstract concepts. It is not that only 7% of students are capable of inference, but that few reach higher levels of reading proficiency where a depth of understanding is shown. Beginning in 2000, and now across more than two decades of PISA, reading performance has shown a concerning decline: the OECD-average reading performance peaks around 2012 before falling in subsequent assessments.
Inferential skills- and how they grow.
The ability to infer meaning is multi-faceted. A reader must draw on vocabulary, context, language knowledge, syntax, working memory and the ability to monitor whether what they are reading makes sense. When literal meaning is not explicit, the reader must combine what they see on the page with what they already know, and from there, join the dots to see whether their interpretation fits the evidence. Naturally, it is often the teacher marking the work who realizes whether these dots have been joined.
In the learning arena, skills like these need to be developed. Classroom practice must have teachers modelling these types of cognitive and linguistic resources. Examples include teacher think-alouds, or read-alouds that draw attention to textual evidence and invite comparison:
“Okay, Johnny, so once Macbeth has heard the prophecy of the witches and waves goodbye, what is he actually feeling?”
“Confused”, Johnny suggests.
“Okay, given what you know from the text, show me why you think this”.
Like this, we’re not making random guesses, but using clues within a text to foster inference.
Because of the importance of modeling inference, genuine classroom conversation becomes a crucial part of the process.
Where has good old conversation gone?
I am certain that the millennials among us can remember the trips to Granny and Grandpa’s house and the ensuing conversation. They were full of protracted explanations of fairly normal events. More personally, I recall listening to my uncle relay the story of a regular trip; ‘the journey began at 6am; we set off, the three of us…’
We would listen: one by one, each person would relay and then respond. At a subconscious level, we were doing three things contemporary learners must be practicing in the classroom:
1. Decoding incomplete meaning
2. Inferring intention or implication
3. Getting feedback quickly
Has conversation really disappeared?
Yet walking through the school corridors and on the break fields, I had to believe that decent conversation was still happening. Of course, learners still speak. What about conversations with friends and parents? Surely the space to develop inference was still there? Of course, the answers are all yes.
But here it is necessary to understand that while conversation has not disappeared, the landscape of young people’s language exposure and practice is changing.
In his work on conversation analysis, UCLA Associate Professor Giovanni Rossi shows how much everyday speech is organized around actions: the asking, requesting, answering, instructing, agreeing and responding all reveal how rich and meaningful conversation is.
Yet, more often, young people engage in conversation that consists of shorter exchanges designed to accomplish something quickly: ask a question, give an instruction, get an answer, react and move on. Because younger learners live in a world where conversation often competes with multitasking, the potential for any narrative form of explanation is reduced. It is certainly not the only example, but the online dialogue exemplifies this.
On a recent school tour, I sat chatting with one of my students. Looking at his phone, we shared a laugh; I couldn’t quite understand the lingo of his online texts. One of the messages – he doesn’t mind me sharing – read:
Student A: I tripped
Student B: 💀💀💀 Bro 😭
The skulls confused me, but I got the point. An exchange like this actually demonstrates how often adolescents are making inferences. Neither student explains the meaning of the skull emojis, yet both can appreciate the implied reaction: inferences are still being made.
Perhaps the better question, then, is whether these quicker, socially familiar conversations transfer to the longer inferential demands of academic reading.
Let’s talk reading then
If conversation provides one area to develop inference, reading is where those skills are tested. A significant disparity exists between live conversation and the deliberate cognitive act of reading: we tend to understand far more when someone helps us speak through an idea than when we must work it out on a page ourselves.
When learners share verbal forms of conversation, they do so within a network of visual and gestural clues. Even the atmosphere of a group facilitates comprehension. In reading, though – and certainly exam-style reading – these clues are removed. Now the learner must hold the text in working memory, gauge implicit meaning, determine author intent, and do so all without the clues of visual or verbal feedback. The result? Inference skills become more critical. And often, where these skills are not as developed, learners tend to answer poorly.
Pondering this, I was reminded of a departmental discussion around the 2025 barriers to learning statistics. Of the thousands of learners with accommodations, the largest percentage – outside of additional time and a separate venue – required readers. Granted, students receive reading assistance for all sorts of reasons, and the existence of that accommodation does not explicitly suggest an inferential deficit. But it is interesting to consider what a skilled reader, through reading aloud, actually provides: they reinvest the necessary verbal clues and reduce some of the cognitive load needed to construct meaning independently. They become the helpful, live conversation.
So, what can we do?
Because inference is developed through varieties of language exposure, the solution may not require cave-bound, Crood-like discussions with granny and grandpa.
As parents, our first instinct is to adopt instructional language (do this, do that): it is both comfortable and necessary. But as a tool to shape inference, it is limited, and an explanation for decisions is better: “No, you can’t” is not as effective as “No, you can’t because I’m worried about you being outside on your own.” The former breeds compliance while the latter triggers inference: there is a reason attached to instruction.
As helpful, and as discussed, is reading aloud. When parents read to their children, immediate feedback and conversation is invited. Parents are also prone to reading books above their child’s independent reading level, and the possibility of richer language and the excitement of deeper plots and motives become prevalent.
Earlier suggestions have been provided for a school setting. Teachers modelling thinking and asking questions that require learners to connect ideas across a text are crucial. But allow me to offer another suggestion: one where I fondly recall an attempt in my own classroom.
Tired of regurgitating past papers, I gave the class a drawing with a single caption: ‘Explain what could have happened.’ The image was of a fallen chair, surrounded by strewn paper.
After some silence, the first response came: “Sir, this was after so and so stole the examination papers!” Naturally, so and so responded abruptly and then followed with his own attempt. As wild as the opinions were, at least the class was enjoying inference. But there was also something light and human about the lesson, giving me a chance to reflect.
In a world increasingly carved for speed and brevity, meaning-making should be more than a curriculum objective. Learners are, first and foremost, human beings. And given their innate need for connectedness, personal relationships, and even humour, making meaning is more than a task. It is actually something they love to do.
