More Bad Ideas about Why We Should Avoid Complex Text Reading Instruction

Welcome to 2018.

  During the interim, several intriguing questions have been submitted and soon I’ll be taking those on. This posting responds not to your questions, but to some public comments made by various colleagues concerning complex text and its use in instruction. My comments are responses to their handwringing over the requirement that we teach kids to read complex text. 

1. We should be concerned about the use of complex text for instruction because text complexity has a negative correlation with reading comprehension and reading fluency.

  The premise here is correct, but the conclusion is false. This is what logicians refer to as the non-sequitur fallacy. I know of no one who rejects the idea that complex text is harder to read and understand than simple text.

  There are definitely quibbles over whether some approaches to measuring text complexity are adequate, but even then the notion is that more complex texts will be harder to understand. Researchers long ago showed that while shorter sentences tend to be easier to comprehend than long sentences, there are many exceptions to this correlation. If sentences are shortened to omit explicitly stated causal connections, for instance, the brevity tends to reduce understanding more than it facilitates it. (That kind of work was not an argument about whether text complexity made comprehension more difficult, but over where such complexity resided.)

  Conceding that the use of more complex texts will likely lower students' daily fluency and comprehension performance with the instructional passages—which it will—tells us nothing about whether or not we should require complex text instruction.

  To get to there, one needs to add another premise—an unstated one in this case. The missing premise is the claim that students learn or learn best from texts they comprehend easily and that they don’t from relatively harder texts.

  As I’ve written before, evidence for that hidden premise doesn’t actually exist, though educators cling to it as a matter of faith. The experimental studies—studies where difficulty levels of texts are systematically varied to determine if that factor affects learning--have not found that working with easier texts improves learning.

  The confusion between how well kids can read the instructional texts and how much they learn is misleading. Indeed, complex texts tend to be more difficult, tend to elicit reading performances that are not as polished as what can be demonstrated with easier texts. However, that does not mean that kids learn less from the more challenging texts. In fact, it appears to be the opposite!

  2. Educational standards that require teachers to teach students to read complex text are over-emphasizing the role of text in reading comprehension.

  This one surprised me a bit. The theory my colleague works within claims that reading comprehension is the result of an interaction between reader, text, and task. Not a bad idea that one, and one that I would have expected to support the idea of complex text--given its place in that holy trinity. My colleague's fear seemed to be that requiring complex text would lead to an undervaluing of the role of task (e.g., question types) in classroom reading work.

  Accordingly, her argument was that text complexity is not such a central issue in reading comprehension. Her evidence that text complexity was being overvalued? She noticed that different National Assessment questions about the same text elicited different levels of student performance. Seventy percent of students could answer a question about a given passage, while only 35% could answer other questions from the same passage. She attributed this difference to the fact that some tasks or questions required literal recall and others depended upon inferencing (though assessment research has long found performance levels on those tasks not to vary in any consistent way across passages).

  Her analysis is shaky again because of an unstate and indefensible assumption—the assumption that all portions of a text should be equally challenging.

  Obviously, sentences that include especially challenging vocabulary words will be harder than ones that use plain language. Questions about such sentences--even if drawn from the same text--are likely to differ in how well students do with them. We’ve long known that ideas that are high in the information hierarchy of a text will be better grasped than ideas that are lower in that hierarchy. That means you would usually see big variations in performance if they tapped an understanding of different information from the same text. Thus, the idea that 70% of kids can answer a question about one aspect of a text, but only 35% of kids can answer a question about another aspect doesn't necessarily tell us anything about readers or tasks since all of that variation may--or may not--be attributable to the text alone.

  I don’t think that problem with her argument means that we, therefore, need to teach text of a particular Lexile level, though it suggests to me the need to engage kids in the of reading texts that they will not comprehend easily—texts that may trip them up in particular ways. That means focusing more attention on what makes texts difficult and less on teaching question types and the like.

3. We shouldn't be stampeded into teaching complex text based upon flawed research studies--studies that varied not just text complexity but the instructional methods.

  Again, a correct premise, but a shaky conclusion.

  This claim points out that those experimental studies that I tout do more than just compare the impacts of the book levels on learning. The experimental groups that were being taught with the harder texts were sometimes being taught differently than the students who worked with the easier texts.

  For example, in one of those studies, the experimental groups working with the harder grade level texts did more fluency work than the guided reading comparison group that read the easier books. Perhaps the study outcomes have been due to those variations in instruction rather than to the text levels themselves.

  That is a fair concern, but I have two fair responses to it.

  First, while this charge is true of some of the studies that I cite, there are others in which it is not the case, but with the same outcome results. The same finding—that teaching kids with instructional level texts either doesn’t help or actually hinders readers to learn to read—is obtained whether the instruction has been held constant or not. Thus, if this concern is used to impeach the evidence, it can only do so for part of the evidence (and still doesn't explain why no studies that have tested teaching kids with easier texts have found any benefit).

  Second, frankly, I would expect good instruction with complex text to look different from good instruction with simple text. This point is extremely important for educators to understand.

  If I were trying to teach kids to read with books that they already could read with a high degree of accuracy and without much need of instructional support (the so-called “instructional level”), then I would expect the kids to spend a large amount of time reading those texts—doing substantially more reading then I see in a typical guided reading classroom. And I would expect much more required engagement with the texts than is afforded by the shallow small-group discussions that are usual in those classrooms. Perhaps I’d increase the amount of classroom reading by 3-5 times over the current amounts and would greatly reduce the small group talks in favor having kids writing much more and more extensively about those easy texts and doing more extended projects (probably a lot more work would be pushed away from the school to be done at home given the low demands of the texts and the notion that kids would learn so much just from doing the easy reading).

  If, on the other hand, I were working with really hard texts such as those required by more than 40 states then my students would certainly do more fluency work, and they would read shorter texts more intensely (with a lot of group discussion and the like). I’d ask the kids lots of questions about ideas in those texts, especially questions that got at the ideas that I thought might be difficult to gain due to particular aspects of the texts’ complexity. I would have the kids engaged in much more rereading than is common, too.

  Of course, good instruction is likely to be a blend of both of these visions—since no one in their right mind would expect instruction to only emphasize hard to read texts (though it is odd that so many of my colleagues have long held to the idea that kids should read only what for them would be easy texts--prohibiting so many learning disabled, second language, and racial minority from the opportunity to struggle with the grade level ideas).

  Friends, I think you protest too much. Reading comprehension instruction should not be focused on how to answer particular questions. It should teach students to recognize and gain control over those aspects of text that serve as barriers to comprehension. To accomplish that, kids have to be asked to read texts that they cannot already read easily—even though that will reduce their initial reading comprehension and fluency with the texts we are using for that teaching. Finally, that teaching reading with complex texts means that you will need to make other changes to instruction beyond the book choice is not a problem; it is a reality. Teachers need professional development that goes beyond how to select complex text; it should teach teachers how to teach kids to read complex text.