Comprehension Instruction That Really Helps – Teaching Cohesion

  • reading comprehension cohesion
  • 05 November, 2022

Teacher question:

One of my colleagues told us that we should not be teaching guided reading lessons or comprehension skills or strategies. We’re using a core reading program that includes those kinds of things. He says that the science of reading proves that we would get higher reading achievement by teaching more social studies and science (he’s our science teacher) and dropping the comprehension instruction that we are providing. He’s really vocal about this. Can you help us shut him up?

RELATED: Shared Reading in the Structured Literacy Era

Shanahan response:

Your colleague is partly right. Knowledge about the world is a valuable commodity in reading comprehension. Education should both nurture curiosity and provide the means to fulfilling it – increasing what kids know about science and social studies (and literature and the arts, too.

He goes wrong, however, when he suggests that doing that should take the place of reading instruction. Such instruction can provide your students with a lifetime ability to increase their knowledge independently. Research does not support the idea of simply increasing knowledge about the world results in some automatic improvements in reading ability. There are types of reading comprehension instruction that can provide those kinds of results. Basically, your colleague is very unscientifically going beyond the empirical evidence and overstating his case.

Comprehension instruction is complicated and multivariate – there isn’t a single type of skill or approach that is likely to be sufficient. I can’t possibly, in the space available here, provide a complete map to what can be included profitably in a reading program.

Research shows that explicit vocabulary and morphology teaching make a difference. We should commit some of the comprehension real estate to increasing what children know about word meanings and how words express those meanings.

Likewise, instruction in how to recognize and use the organizational structure of texts has been well researched, with lots of positive outcomes. Including that is a no brainer.

Strategy instruction – that is, teaching students to execute certain actions before, during, and after reading that can increase their understanding and recall – works too, though admittedly that can be overdone. Such teaching is beneficial, but the regimens of instruction that have been evaluated have tended to be brief and focused on a narrow range of behaviors. But they work.

Not long ago I devoted a blog to the idea of teaching students to read sentences because of the need to makes sense of syntax or sentence grammar. While the general teaching of formal grammar hasn’t been found to hold water when it comes to building comprehension, teaching students to make sense of sentences certainly has.

And, what about literature? I haven’t gotten into that but hope to soon. Too often those who are promoting knowledge leave that out of the equation, as if nothing of value is gained from commerce with fictional stories and novels, poetry, and drama.

Yeah, all those things deserve instructional time, attention not likely to be provided in social studies or your colleague’s science classes.

Let me give one more example of what we need to teach in reading comprehension. Its inclusion is advocated by many of the top researchers in the field – and they have the data to prove it. I haven’t written here about this topic before, but it is one that is well worth exploring: cohesion.  

Authors try to convey a coherent message through their writing. They do that with several linguistic devices and communications conventions. Readers then try to convert this information into a mental representation of that message. Readers don’t put the text itself into their memory (we’re not computers). No, we translate what we read, and that translation requires recognition and use of those devices the author put into the text.

For example, here is a brief passage used in a study of cohesion instruction (Baumann, 1986):

Michael got the chicken pox. Then Tom did too.

Michael said, “Wasn’t that a terrible sickness?”

            “It sure was,” Tom said. “Were you as sick as I was?”

To comprehend this text the reader has to make a number of connections. It is necessary to recognize that “did too” refers to “got the chicken pox,” and that “sickness” and “It” stand for chicken pox, and that “you” is Michael, and “I” was Tom. Readers who fail to make those connections aren’t going to understand what this text is about.

Prior knowledge – the knowledge the reader started with – can help a bit; it is likely helpful to know when you start out that chicken pox is a form of sickness, though this text is simple enough I suspect someone could make the right connections in this case strictly based alone on the linguistic information the author has provided. That might be harder to do if the text was much longer and the distance between the links were greater or if there were more possible choices.

Not surprisingly, younger readers are more likely than older readers to have difficulty with those seemingly straightforward pronoun references – like you for Michael. Or, that older readers who can make those kinds of links might struggle to make causal connections in a science text such as in the following example (Best, Rowe, Ozuru, & McNamara, 2005):

            “Plants lack a nervous system. They cannot make quick responses to stimuli.”

One would hope that a reader, if asked “why can’t plants make a quick response to stimuli?” would answer, “Because they don’t have a nervous system” even though the text didn’t explicitly say this, but only implied this connection by arraying these sentences in this fashion.

Of course, nobody’s perfect. An author may provide a text that allows readers to easily recapture the intended coherence, while others may seemingly be more cavalier – perhaps they expected their readers to have knowledge that they didn’t.

One might assume that all readers would be best served by the authors who produce texts with the highest degree of coherence. That would appear to provide the best support for readers. Let’s face it, if the author is explicit about connections within a text, then more of us should be able to recognize and use those connections.

It turns out, according to research, that some readers earn more from relatively low coherence texts (e.g., McNamara, 2010; Ozuru, Dempsey, & McNamara, 2009). The more knowledgeable and better readers usually learn more when they are forced, by the text, to engage in more mental processing – generating more inferences and the like. Such readers may give high coherence texts a more superficial once over since it is so easy for them to understand, that they end up with lower comprehension and learning. Alternately, when reading more coherent texts the poorer and less knowledgeable readers tend to do better with comprehension. When reading less explicitly coherent texts, these readers may fail to make the necessary connections or fill the gaps that such texts require, so comprehension suffers. 

What that points out is that readers need to learn to negotiate a range of texts – texts that differ in coherence, and, consequently, in their challenge level. Increasing one’s knowledge can help with that – at least, if the knowledge and the subject of the text match – but research has found that it is possible to teach students to negotiate the cohesive ties and connections that the author’s build into their texts. It is this kind of negotiation that needs to be taught in our reading comprehension lessons.

One of the problems with the research on this topic is that it goes under a variety of titles. Some researchers focus on cohesion and cohesive ties – the linguistic repetitions and links themselves, while others categorize it as a kind of inferencing, the mental action that cohesive elements require. Still others might emphasize specific types of cohesion (e.g., referential, causal, temporal, anaphora, cataphora), or relegate this kind of thing to a superordinate category like Executive Function. I might recommend cohesion training, while a colleague may tout inferencing instruction – when we mean the same thing.

In any event, linguists have done a fine job of identifying and categorizing the many forms of cohesion (Halliday & Hasan, 1976), psychologists have developed innovative techniques for measuring the degree of unity or cohesiveness in texts (e.g., Graesser, McNamara, Cai, Conley, Li, & Pennebaker, 2014; Sheehan, 2013), and there are scads of studies showing the correlation of ability to interpret cohesion ties and reading comprehension (e.g., Duran, McCarthy, Graesser, & McNamara, 2007; Gasparinatou & Grigoriadou, 2013; Hall, et al., 2015; MacLean & Chapman, 1989; Schmitz, et al., 2017). Such studies have found these relations with a variety of text types (including science texts) and with students across a wide range of ages and abilities.

Most importantly, there are a number of studies showing that it is possible to raise reading comprehension by explicitly teaching such skills (e.g., Bauman, 1986; Best, Rowe, Ozuru, & McNamara, 2005). Elleman, 2017; Gallini, Spires, Terry, & Gleaton, 1993; Garcia-Madruga, et al., 2013). This kind of teaching has been effective with a wide range of ages and abilities – but has been especially powerful with low performing readers.

This kind of instruction should explain this aspect of language, providing many examples. Then teachers should provide direct instruction in how to deal with these kinds of connections through a text. This could include both modeling (showing students how to connect the ideas across sentences, paragraphs, and longer expanses of text), and guided reading (having students try to make these connections with explanations). This might be followed by exercises that give students more practice. These might take the form of worksheets, with the students trying to find the references for particular words or phrases or fill-in-the-blank pages. Or, another possibility, is to have students read a text and answer wh-questions about the passages.
Teachers should also raise these kinds of questions during regular group or class shared or guided reading activities. These kinds of activities should take place with a variety of texts.

Along with those earlier mentioned areas of comprehension instruction, this kind of work can have a positive impact on students’ reading ability.

Tell that to your colleague – but then also make sure that your kids are learning plenty of content, both in his science classes and in your reading lessons. Every reading assignment is an opportunity to expose children to information about their world. With a little diligent attention and effort, that information can become knowledge available to support future reading. The lessons in vocabulary, syntax, cohesion, and text structure should increase students’ ability to gain that information in future reading.

READ MORE ARTICLE HERE: Shanahan On Literacy's Blogs


Baumann, J. F. (1986). Teaching third-grade students to comprehend anaphoric relationships: The application of a direct instruction model. Reading Research Quarterly, 21(1), 70-90. doi:

Best, R. M. , Rowe, M. , Ozuru, Y. & McNamara, D. S. (2005). Deep-Level Comprehension of Science Texts. Topics in Language Disorders, 25 (1), 65-83.

Cox, B. E., Shanahan, T., & Sulzby, E. (1990). Good and poor elementary readers' use of cohesion in writing. Reading Research Quarterly, 25(1), 47-65. doi:

Duran, N. D., McCarthy, P. M., Graesser, A. C., & McNamara, D. S. (2007). Using temporal cohesion to predict temporal coherence in narrative and expository texts. Behavior Research Methods, 39(2), 212-223. doi:

Elleman, A.M. (2017). Examining the impact of inference instruction on the literal and inferential comprehension of skilled and less skilled readers: A meta-analytic review. Journal of Educational Psychology, 109(6), 761–781. 00180

Gallini, J. K., Spires, H. A., Terry, S., & Gleaton, J. (1993). The influence o macro and micro-level cognitive strategies training on text learning. Journal of Research & Development in Education, 26(3), 164-178.

García?Madruga, J. A., Elosúa, M. R., Gil, L., Gómez?Veiga, I., Vila, J. Ó., Orjales, I., . . . Duque, G. (2013). Reading comprehension and working memory's executive processes: An intervention study in primary school students. Reading Research Quarterly, 48(2), 155-174. doi:

Gasparinatou, A., & Grigoriadou, M. (2013). Exploring the effect of background knowledge and text cohesion on learning from texts in computer science. Educational Psychology, 33(6), 645-670. doi:

Graesser, A. C., McNamara, D. S., Cai, Z., Conley, M., Li, H., & Pennebaker, J. (2014). Coh-metrix measures text characteristics at multiple levels of language and discourse. Elementary School Journal, 115(2), 210-229. doi:

Halliday, M. A., & Hasan, R. (1976). Cohesion in English. London: Routledge.

Hall, S. S., Kowalski, R., Paterson, K. B., Basran, J., Filik, R., & Maltby, J. (2015). Local text cohesion, reading ability and individual science aspirations: Key factors influencing comprehension in science classes. British Educational Research Journal, 41(1), 122-142. doi:

MacLean, M., & Chapman, L. J. (1989). The processing of cohesion in fiction and non-fiction by good and poor readers. Journal of Research in Reading, 12(1), 13-28. doi:

McNamara, D. S. (2010). Strategies to read and learn: Overcoming learning by consumption. Medical Education, 44(4), 340-346. doi:

McNamara, D. S., Crossley, S. A., & McCarthy, P. M. (2010). Linguistic features of writing quality. Written Communication, 27(1), 57-86. doi:

Ozuru, Y., Dempsey, K., & McNamara, D. S. (2009). Prior knowledge, reading skill, and text cohesion in the comprehension of science texts. Learning and Instruction, 19(3), 228-242. doi:

Schmitz, A., Gräsel, C., & Rothstein, B. (2017). Students’ genre expectations and the effects of text cohesion on reading comprehension. Reading and Writing: An Interdisciplinary Journal, 30(5), 1115-1135. doi:

Sheehan, K. M. (2013). Measuring cohesion: An approach that accounts for differences in the degree of integration challenge presented by different types of sentences. Educational Measurement: Issues and Practice, 32(4), 28-37. doi:


See what others have to say about this topic.

L Schroeder Nov 08, 2022 01:26 PM

Thank you. When I taught home economics I made many connections as to the meaning of words when cooking or sewing - baste is and example of one with multiple meanings. F

Harriett Janetos Nov 08, 2022 03:44 PM

Thanks, Tim, for a complex blog about a complex topic. There's a lot of confusion out here as well as a great deal of simplification: don't teach strategies--establish background knowledge. The Elleman research you cite states:

"Many models consider the role of knowledge-based inferences in comprehension (Kintsch, 1988; Kendeou & O’Brien, 2014), whereas a few focus only on the impact of text-based inferences for maintaining coherence (e.g., Zwaan, Magliano, & Graesser, 1995). Despite these different perspectives, nearly all discourse comprehension models assume that readers establish coherence by generating text-based inferences to make sense of causal, temporal, and spatial relationships in discourse (McNamara & Magliano, 2009)."

Teaching a third grade class over the past four years has made it abundantly clear to me just how difficult it is for many students to "make sense of causal, temporal, and spatial relationships in discourse." When we read a piece about biomimicry that explained how scientists look to nature to solve problems, my students had no problem learning this new term (thereby building background knowledge) and what it meant based on the clear explanation in the article. However, in reading the paragraph about the Japanese engineer who looked to the kingfisher bird's beak to solve the noise of the bullet train, many students struggled to make connections between the bird and the bullet train--though they all had knowledge of both beaks and trains. So it's not teach either background knowledge or contextualized strategies, it's both/and.

I hope you continue to revisit this interplay between knowledge and text when it comes to comprehension.

Timothy Shanahan Nov 08, 2022 07:14 PM


Your illustration is a good one. People definitely can read successfully about things that they have no immediate prior knowledge of -- and there are various reasons for that (including that whatever you already know you don't need to relearn, thus reducing the amount of attention and effort required). However, knowledge is something like vocabulary -- I can teach a ton of vocabulary with little impact on comprehension if I don't manage to teach the vocabulary that appears in the texts that you are reading (and the vocabulary the meanings of which are central to understanding those texts). Building knowledge should be a lifetime pursuit and education should increase it at every turn (including reading). However, the purpose of reading instruction needs to be much more than increasing knowledge -- if that isn't the case, there is no point in reading instruction once decoding has been mastered to the point that it improves through practice alone. Teaching how language works, how to use your prior knowledge during reading, fundamentals of written language (as opposed to oral language), how to maintain attention and effort in what is a monologue (as opposed to dialogue) situation, etc. all should be part of reading instruction.


Joan Sedita Nov 10, 2022 01:37 AM

Thank you Tim for a number of topics and points you include in this post. I have spent over 40 years addressing instructional practices that improve student comprehension, reviewing the research that has developed over this time, and shared what I have learned with teachers.
First, thank you for addressing head-on the unfortunate conclusion that many educators have drawn related to an emphasis (over-emphasis in my opinion) that some have placed on the role of background knowledge. You note, "He goes wrong, however, when he suggests that doing that should take the place of reading instruction." Another misleading message that is given by some in our field who emphasize background knowledge is that comprehension strategy instruction should be avoided, often attacking in particular "main idea skills". However, as the recent IES research guide "Providing Reading Interventions in Grades 4-9" notes (Vaughn and team), explicit instruction of comprehension strategies is very beneficial (see the report's recommendation #3 that specifically suggests teaching question generation and answering, and determining the gist, i.e., main idea). The National Reading Panel identified comprehension strategies that include summarization (which requires students to identify essential main ideas), and the 2021 piece by Duke, Ward, and Pearson "The Science of Reading Comprehension" also summarized the findings that support comprehension strategy instruction. See this blog post I wrote last year about their article:

Thank you also for pointing that there are multiple components of comprehension instruction that should include vocabulary, text structure (at multiple levels), and sentence work. In particular, I thank you for bringing people's attention to teaching cohesive devices, a topic I addressed in another blog post that your readers might find helpful:

Timothy Shanahan Nov 10, 2022 01:43 AM

Thanks, Joan. I guess we all just have to keep banging at it... someday people may get that following the science of reading means basing practice on scientific studies.


Pete Bowers Nov 13, 2022 08:30 PM

Thanks for the very interesting post, Tim. Your discussion about "cohesion" points to an area that deserves close attention. My main research interest is about word-level study, so I have not spent a lot of time on this level of literacy research. But your discussion reminds me of discussions with my PhD supervisor, John Kirby, who has a much wider expertise/research interest. I don't recall the term "cohesion" (I may have missed it), but he was addressing similar ideas in trying to understand reading comprehension and ways to improve it through instruction.

I was pleased (and not at all surprised) by your emphasis on morphology and vocabulary instruction to support reading comprehension. In particular, I wanted to highlight the intersection of morphology/vocabulary instruction with subject area study which was central to the question you are responding to. This intersection is one of the things that got me interested in the instruction I was exploring as a Grade 4 teacher that grew into structured word inquiry (SWI) (Bowers & Kirby, 2010). Once we learn how to analyze the morphology and phonology of related words, we notice connections in spelling structure and meaning in words about any topic. Students and teachers can draw on the generative morphological problem-solving knowledge supported by SWI to leverage understanding of concepts and terms in any subject area. Perhaps you and your readers would be interested in a few examples.

A favourite early story is from a Grade 4/5 class in a disadvantaged neighourhood in Ontario. A year or two before this story, this school had the lowest literacy scores in Ontario. A teacher friend of mine did great work with SWI in this school. One day he was taking up a great science activity on the concept of condensation as part of their unit on states of matter. At the end of the activity he wrote that word on the board. Before he could say anything about the word, the kids asked, “What’s the word sum?”

The teacher had not planned on using their new orthographic knowledge in this science lesson. Clearly his students already saw value in applying what they were learning to new words.

Click the following link to see a fuller account of this story and how the teacher guided the students' inquiry to discover the base and the word sum con + dense/ + ate/ + ion —> condensation.

This student-led morphological investigation led to a class discussion linking the concept of “condensation” to the word “density” and the base spelled DENSE that builds both of both those words. If you read the linked account, you will see that a student correctly proposed the base needed that final E to act as a plural cancelling marker.

This kind of self-motivated study of orthography, guided by a knowledgeable teacher, has students engaging in extremely focused study about how orthography represents morphemes and grapheme-phoneme correspondences. This is a means of building mental representations of the spellings, meanings and pronunciations of words. This is exactly the kind of activity recommended by Perfetti’s (2017) Lexical Quality Hypothesis and our theory of morphology as a binding agent (Kirby & Bowers, 2017) which builds on Perfetti’s concept of a “constituent binding feature” of lexical quality. It also fits perfectly with the goal of Ehri’s (1998, 2005) Orthographic Mapping to build “sight words” by binding the spellings, pronunciations and meanings of words.

Our Grade 4/5 intervention (Bowers & Kirby, 2010), which introduced the phrase “structured word inquiry,” found generative vocabulary learning. Students who studied morphological families with the matrix and word sums and the interrelation of morphology and phonology were better than controls at defining words they were never taught, but which had a base that they were exposed to.

This is good news for vocabulary learning, which — as you point out — contributes to reading comprehension. But I think the richest learning in that study is not often recognized. It is not just that children showed generative vocabulary learning. In my view, the recognition that spelling is ordered enough that you can make sense of it if you learn how to study it. That understanding is a necessary condition to explain the story above of a Grade 4/5 class in a public school in a disadvantaged neighbourhood. It is this understanding that motivates continued study and a potential for a virtual cycle of learning.

One reason your post grabbed my attention is that it seems to me that this kind of study can also be leveraged for the topic of “cohesion” for reading comprehension you address. Discussions of implied meanings in sentences and paragraphs could be be sparked and catalyzed by orthographic knowledge like what we see in this story. The more that study occurs, the easier it becomes to notice subtle orthographic cues.

For anyone interested, here are a few illustrations of such teacher-guided, student-led learning.

6-year-olds investigate “carnivore” using etymology and make many connections including one to the “carnation” flower — so named for being a “meat coloured flower” as they discover.

A Grade 5 Student creates a video explaining the orthography of “resistance” related to history study and makes many fascinating discoveries.

A Grade 2 student learns about the words “endure” and “durable”

And here is a story of a high school discussion of the related meanings and history of the words “subjugation” and “jugular” in the context of a study of a famous presidential speech addressing a racially motivated assassination.

I hope people find the above a productive addition to your important post Tim.


Melissa Apr 27, 2023 11:56 PM

Are there any studies showing whether this kind of work on cohesion improves writing?

Timothy Shanahan Apr 28, 2023 02:46 PM

I'm aware of no experimental studies of writing that focused on cohesion (several studies measuring cohesion in writing that show its importance and its development). It would make sense that cohesion instruction would help both reading and writing, but I would encourage teachers to include some cohesion in writing exercises -- having kids compose sentences and paragraphs that make certain kinds of links.


What Are your thoughts?

Leave me a comment and I would like to have a discussion with you!

Comment *

Comprehension Instruction That Really Helps – Teaching Cohesion


One of the world’s premier literacy educators.

He studies reading and writing across all ages and abilities. Feel free to contact him.