Reading about Reading
I’ve been traveling a lot lately. Luckily, while standing around in airports and sitting on planes, I’ve had a very interesting book to read, namely Language at the Speed of Sight: How We Read, Why So Many Can’t, and What Can be Done About It. Fairly technical in many places, opinionated in parts, and sure to be controversial, it’s also a fascinating, informative, and fun read about reading, especially for anyone who is a language nerd (like me).
The book’s author, Mark Seidenberg, a University of Wisconsin cognitive neuroscientist, writes in a voice that is both humorous and authoritative. He says he has been studying reading “since the disco era” and I believe him. His deep knowledge is apparent and disco isn’t the only 70s cultural reference he drops.
The first two thirds of the book address the history of language development (including reading and writing) and the science of how we read (and why reading is more difficult for some than others). There are many points to ponder, including 1) the reasons why reading is much more difficult than speaking, and 2) the 10,000 year old evolutionary pathway our brains may have taken (moving from face recognition to symbol recognition) to get to the point where reading occurs.
The last third of the book is devoted to Mark Seidenberg’s takes on a number of educational issues that impact whether or not children in America learn to read. This section of the book is already creating some controversy. For a taste of this, as well as for a sad reminder of how the digital age has made it so easy to post ideological, ill-informed, and nonsensical opinions, rants, and whines, see the Customer Reviews for this book on Amazon. As for this reader, I’m ambivalent about some of Seidenberg’s viewpoints (I feel like I need to know more) and I debate others (I think his critique of balanced literacy probably suffers from a lack exposure to current-day reading classrooms that are staffed by highly trained and knowledgeable teachers who do engage in the literacy best practices that Seidenberg thinks should be taught). However, I am in agreement with a few of his critiques, especially the one concerning how American elementary school teachers are prepared for classroom jobs (the process is lacking) and what the field of education can do to improve it.
Anyway, controversy and opinions aside, there are dozens of science-based points in the book that are really interesting to think about. Here are three of my favorites, culled from my top ten:
For the remainder of this post, I’m going to ponder the first point: the basic process for learning to read is a statistical one. More specifically, I’m going to consider sight words, a subject I’ve written about before.
No Sight Words
No Sight Words is a subheading in the book’s chapter entitled Reading, The Eternal Triangle. The subheading sets the stage for this idea: because a reader’s brain statistically processes words, sounds, and meanings (encountered daily, over weeks and months), sight words aren’t really learned through memorization. To quote the book (page 143): “Words with atypical pronunciations such as HAVE and GIVE are usually treated as sight words that must be memorized. However, HAVE’s pronunciation is not arbitrary; it overlaps with HAD, HAS, HAVING, HAVEN’T, HIVE, and other words. Thus, what the child learns about words such as HAVE is relevant to many other words in which the vowel has an atypical pronunciation…”
Just to be clear, because we don’t end English language words in U or V, teachers know that words like HAVE and BLUE, GIVE and TRUE end with an E. It’s a convention of English spelling. Seidenberg doesn’t dispute this. Rather, he focuses on how the brain works during reading. That brain’s process for coming to a point in which it can read a word is not a process of remembering and applying a spelling principle (although this might help a speller spell the word correctly). Nor is it a process of strict memorization, in which a child repeatedly looks at the word HAVE in isolation until it is remembered. Rather, the brain’s process for reading HAVE is one of looking at and hearing many different words, analyzing their word features and pronunciations, applying weights to these features, and then, over time, coming to recognize (reading) the word as “have.” In the end, the word has become wholly encoded in the brain dictionary or lexicon, ready for instant recall for spelling and, more importantly, ready for instant recognition for reading (by sight, as it were). None of this is done consciously, at least not by beginning readers.
My understanding of the statistical nature of learning to read a word (such as HAVE) is this: Certain letter feature aspects of HAVE, such as VE, AVE, HAV, are analyzed in conjunction with the aural presentation of the word (as read by a parent, teacher, or the reader herself). These aspects are compared and contrasted with other aspects of other words. The brain considers the parts VE, AVE, and SAV of the word SAVE, the parts VE, IVE, GIV of the word GIVE, the parts AD and HA of the word HAD, and so on. These word aspects (or parts) are constantly weighted by the brain. In the end, the brain comes to read the target word by comparing and contrasting all of these weighted parts, allowing some to fade away and others to strengthen.
Dr. Seidenberg’s research involves computer modeling. The computer model of reading that he and others have developed sheds light on the invisible thought processes that take place in the brains of readers all over the world. Consider this quote from the book: “Although the model doesn’t learn ‘sight words’ by memorizing them, once it has learned a word very well, it behaves as though it is read ‘by sight’… Early in training, the model’s performance on a word such as HAVE is affected by its overlap with other words. It is not memorized ‘by sight.’ However, like most ‘sight words,’ HAVE is a very high-frequency word. The model learns it relatively quickly because common words are trained more often than less common ones.”
So, how does knowing that a brain learns words via statistical analysis help me teach kids to read, spell, and write? What can I do to more effectively teach struggling readers, writers, and spellers? First of all, its important to provide opportunities for kids encounter lots of words. This means we must have them read, read, read and write, write, write. Secondly, it’s important to have effective skill instruction in place.
For years I taught 4th and 5th graders with IEPs. Later, I taught low achieving 3rd graders. Some of the children stumbled over high frequency, atypically pronounced words like WAS and WANT. Past practice was this: give kids a big ring of “sight” word cards and have them repeatedly practice these flash cards at home with their parents, on their own during independent work time, and with a teacher (but just for a few minutes a week). The thought was that students would memorize the words through repeated practice, cramming them into their mental dictionary and securing it for use when they saw it in a piece of text.
As I learned more about the Wilson Reading System (based on Orton-Gillingham, which teaches via instruction that supports pattern recognition), I came to believe that the rote memorization method for learning sight words needed to be pushed to the background and a method that taught students to notice patterns, read patterns, and spell patterns needed to be emphasized. In other words, the main method for teaching “sight” words should be to present these words alongside other words with similar features. Spelling and phonics instruction should be aligned, spelling and phonics should be carefully taught, and the time devoted to them should be intensive.
Effective instruction involves presenting “sight words,” such as the high frequency words on a Dolsch list, within a context of patterns that make words and understanding "how words work. Sure, to give kids multiple exposure to difficult to learn words, we can have them roll through a string of unrelated high-frequency words on flash cards. But more importantly, we must present “irregular” sight words (which really aren’t so irregular) alongside words connected by a shared feature and presented during spelling and phonics instruction. For example, to teach SAID, present it alongside SAY and SAYS. Teaching that SAID is spelled with an A (as opposed to an E as in SED) helps connect the word, through meaning, to the words SAY, SAYING, and SAYS. It also helps the brain tease out spelling and pronunciation features that are statistically significant.
Here’s another example. Consider the schwa sound of A in the word WAS. This word isn’t so irregular in its pronunciation. Its schwa A sound is also present in the high-frequency words WHAT and WANT, as well as in contractions and inflectional variations (WASN’T, WANTED, WANTING). Additionally, the schwa A is found in other Dolsch list words such as ABOUT, AGAIN, and AROUND, as well as in commonly encountered words like ANOTHER and ALONE. Finally, the sound and spelling is present in engaging, common knowledge words such as ALASKA and BANANA.
In the end, words are read in entirety and by parts. Thus, my language instruction (reading, writing, spelling) should give attention to both: noticing and comparing parts, working to move whole words into the brain dictionary for instantaneous recognition during reading. To quote Language at the Speed of Sight one last time: “So, is HAVE read “as a whole,” or do the parts matter? This is like asking if light is a wave or a particle. The answer is it isn’t one or the other; it’s both.”
I am a teacher, literacy consultant, author, musician, nature lover, and life long learner.