Wednesday, May 04, 2005

How Should We Be Teaching Reading? Humans or Software?

The latest of issue of Harvard's Education Letter, Elizabeth Kidder takes on the Brain-Based Reading Software Fast Forword with her discussion of a recent Princeton study. What are the results? Rouse and Krueger found only that slight benefits for children in their large randomized study. Rouse and Krueger studied children in urban schools with lower socioeconomic backgrounds. They were 3rd-6th graders, 40% African Americans and 50% Hispanic and scored in the lowest 20% on a standardized reading test.

The Results reported in the Economics of Educational Review (sorry not available free online): 50-60% of students were unable to keep up with the demands of computer training (about 90 minutes a day). Of the ones remaining, slight improvement was see on the Reading Edge test, but no improvement on the CELF-3. They conclude that the slight benefit was seen because: "computers are not an effective substitute for traditional classroom instruction, or because educators have not learned how to effectively use computer technology to enhance instruction, or because there are other aspects to the school setting that make it difficult to incorporate computerized instruction into the curriculum." They further challenge: "Ask hard questions about whether to invest in educational software, brain-based or otherwise. Ask not just whether or not it works in general, in theory, but whether it is actually going to work with all the warts, etc., on a school day...and show me the empirical evidence. From the fMRI on down, does it ultimately mean kids will be able to read better?"

Oh dear, oh dear. The Princeton is not well-conceived because it never selected children with known phonological problems. There are a wide range of difficulties that can contribute to problems with reading, including impaired visual processing, impaired attention or memory, and social or behavioral problems. We never recommend auditory processing software if auditory processing problems aren't present. No data was given regarding IQ (retardation?), attention deficit, or any other confounding factors. The researchers just took all the students who performed in the lowest 20% on a standardized reading test. Aren't schools identifying which students need which particular kinds of help?

Any good reading teacher knows struggling readers may have very different problems. Schools need to abandon the idea of "one-size-fits-all" curricula and focus instead on how to better match specific needs to specific solutions. If a child with phonological problems can pick up word sounds by careful one-on-one reading with a parent or tutor, then the software isn't needed. But if one-on-one or conventional teaching isn't working and problem is clearly sound discrimination, then targeted incrementally-challenging software may really be helpful.

Also with this Princeton study, there was apparently not much warm-up time for adopting the computer technology. The study suggested computer glitches happened often, and it even mentions that Scientific Learning had suboptimal technical assistance- but the company was also not given the opportunity to rebut the remarks anywhere we could see.

It's not hard to detect a feeling of animosity in the Princeton report - and we aren't sure why. Maybe it's because Scientific Learning has decided to price their software so high ($1000-1500; they should drop their prices). Maybe because the 90 minute requirement seems too inflexible. Still another reason may be that educators feel that neuroscientists are muscling in on their area of expertise. But educators and neuroscientists have a lot to learn from each other - so we hope more cooperative opportunities present themselves, rather than groups taking sides.

Both one-on-one / small group instruction and software-based training can help struggling readers get up to speed. What's really needed are more studies to figure out how children are best matched to particular learning approaches.

The figure below shows some of brain pics in dyslexia. Some have auditory processing problems, others have visual processing problems, and still others have both. As fancy as the brain pictures look, the science is clearly in its earliest stages. Researchers are just beginning to investigate other associations of dyslexia - including impaired fluency (speed), naming, and comprehension. From the functional imaging perspective, we also do not know much about how the patterns change over time or how different strategies may compensate for particular learning blocks. And to our knowledge, no studies have yet looked at the spectrum of differences among individual dyslexics.

How Good Is Fast Forword?
Stanford Fast Forword fMRI pdf
Visual Processing dyslexia fMRI

No comments:

Post a Comment