![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Subtitled "How We Read, Why So Many Can't, And What Can Be Done About It" this fascinating book provides some great insight into the science of reading. Not linguistics or language, really, but reading.
There's some great observational science about how people actually read, and this goes towards exploding many common myths. Both of the 98% scams of 'speed-reading' and the education wars of how to teach children to read.
Perhaps one of the most interesting little bits is probably quite useless. As you might know, your eye makes little jumps or saccades as you read. Obviously it takes time to move your eyes, so one way to increase reading speed is to remove the saccades. Systems have been devised for this so that a text is presented one word at a time at a fixed location on a screen, and you can power your way into a significant speed increase. But a drawback is that you can't keep up that focus for long. I think the author describes it as trying to win a staring contest with a book.
One of the interesting insights comes through the development of neural network type models. While not necessarily exactly the same as how brains process written language, it's still very interesting and I think does teach us something. A naïve look at reading is just to say that it's language translated into the visual rather than into sound. But what the models find (by training them in different ways) is that there are three or four interacting pieces to the puzzle. The brain is juggling with the orthology (the written characters) the phonology (the sounds of the phonemes and graphemes and morphemes) and the semantics (the meaning of the language), and the best neural networks have connections that run both ways through all three of these subsystems.
And if you cripple a neural network by not training it on the phonology, you get a less able reader. And this leads into the more contentious issues in the latter part of the book. Namely that 'whole language' reading instruction that deemphasizes skills-based learning is, shall we say, not well-supported by the evidence. Worse, its proponents misunderstand the evidence they use to support it.
Seidenberg naturally feels that it would be a useful thing to base a curriculum of "teaching people to teach people how to read" on the science of how people learn to read. Sadly, this is not the case at the moment, and I share his frustration with what education curricula actually do (and don't) teach prospective teachers. The entrenched inertia of ed school is likely preventing us from making evidence-based advances in the teaching of reading.
There's some great observational science about how people actually read, and this goes towards exploding many common myths. Both of the 98% scams of 'speed-reading' and the education wars of how to teach children to read.
Perhaps one of the most interesting little bits is probably quite useless. As you might know, your eye makes little jumps or saccades as you read. Obviously it takes time to move your eyes, so one way to increase reading speed is to remove the saccades. Systems have been devised for this so that a text is presented one word at a time at a fixed location on a screen, and you can power your way into a significant speed increase. But a drawback is that you can't keep up that focus for long. I think the author describes it as trying to win a staring contest with a book.
One of the interesting insights comes through the development of neural network type models. While not necessarily exactly the same as how brains process written language, it's still very interesting and I think does teach us something. A naïve look at reading is just to say that it's language translated into the visual rather than into sound. But what the models find (by training them in different ways) is that there are three or four interacting pieces to the puzzle. The brain is juggling with the orthology (the written characters) the phonology (the sounds of the phonemes and graphemes and morphemes) and the semantics (the meaning of the language), and the best neural networks have connections that run both ways through all three of these subsystems.
And if you cripple a neural network by not training it on the phonology, you get a less able reader. And this leads into the more contentious issues in the latter part of the book. Namely that 'whole language' reading instruction that deemphasizes skills-based learning is, shall we say, not well-supported by the evidence. Worse, its proponents misunderstand the evidence they use to support it.
Seidenberg naturally feels that it would be a useful thing to base a curriculum of "teaching people to teach people how to read" on the science of how people learn to read. Sadly, this is not the case at the moment, and I share his frustration with what education curricula actually do (and don't) teach prospective teachers. The entrenched inertia of ed school is likely preventing us from making evidence-based advances in the teaching of reading.
![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif){kind=small}
