Lliana Loewus:

The tide has turned on reading instruction. Nearly all states have passed “science of reading” laws, and most researchers and educators now agree students need to learn letters and sounds explicitly and systematically to become proficient readers. The Washington Post’s editorial board recently went so far as to proclaim that the reading wars have ended. “The victor is clear: Phonics is the best way to teach kids how to read.”

And yet a look inside K–3 classrooms reveals surprising variation in exactly how these letters and sounds are taught. Along with the many research-based methods in use, there’s another practice taking hold, and at great cost to students: over-teaching.

Mark Seidenberg, a professor emeritus at the University of Wisconsin–Madison who studies reading science, helped persuade the public of the need for science-based instruction—and now he’s among those sounding the alarm on over-instruction. Reading teachers need not aim to teach every single pattern students will encounter in text, he says; they simply need to teach enough that students can achieve “escape velocity,” or the ability to start cracking the code on their own.

“You do teach them about words, about print. You teach them enough simple phonics patterns so they can start sounding out some words. And then there’s supposed to be a light bulb that goes on,” Seidenberg says.

Researchers call this ability to implicitly pick up patterns and apply them the “self-teaching mechanism,” or “statistical learning”—and many say it’s underrecognized within the science-of-reading movement.