in a companion article on this web site titled dyslexia and biology, i present evidence from research at the university of colorado showing that many cases of dyslexia have a strong genetic origin. however, a strong genetic influence on dyslexia does not imply that it can’t be remediated or improved by extraordinary environmental intervention. for example, diabetes is a disease with a strong genetic influence, but its course can be substantially changed through environmental manipulations of patients’ insulin and diet. similarly, there is much evidence that dyslexics’ reading can be substantially improved through intensive individualized instruction that focuses both on phonological skills and accurate reading in context (e.g., torgesen et al., in press). however, the cost of this one-on-one instruction with a human tutor is very high and unavailable to many children with dyslexia. could computer-based remedial instruction be a more affordable and widely available replacement for or supplement to individual teacher-based remediation? the results from research reviewed in this article suggest that the answer is yes. i will discuss studies on the computer-based remediation of deficits in word reading, phonological skills, spelling, and reading comprehension.
accurate word recognition while reading, and learning new words
perhaps the most important factor in the development of computer technology for reading instruction has been the availability of high-quality text-to-speech translation. digital equipment corporation first introduced this technology for microcomputers in 1985. since 1986, we have explored the use of synthetic computer speech as a remedial tool for dyslexic children’s deficits in printed word recognition (e.g., olson, foltz, & wise, 1986; olson & wise, 1992; wise, ring, & olson, 2000). our initial studies selected children in the 3rd – 6th grades who were from the lower 10% of their class in reading. the children read stories on the computer for a half hour each day. when they encountered difficult words in the stories, they could click on the words with a mouse and have the computer highlight and pronounce the words. during the session the children answered occasional multiple-choice comprehension questions about the stories they were reading on the computer. at the end of the session, the computer presented some of the targeted words in a recognition test. the children were attended by a teacher for 1 of every 4 sessions, but they read alone on the computer in the other 3 sessions.
the average word-reading gains of the computer-trained children significantly exceeded those of matched poor readers who remained in their regular reading class. however, some of the computer-trained children had difficulty working alone with the programs. also, we found that the children’s phonological skills at the start of training were positively correlated with their rate of word-reading growth during training, suggesting that it might be helpful for their reading development to directly remediate their phonological-processing deficits (olson & wise, 1992). the children’s phonological decoding skill in reading was assessed by having them read pronounceable nonwords (e.g., tegwop, framble) silently and aloud. their phonological skill in language (phoneme awareness) was assessed in several ways. one method tested their ability to play a ”pig latin” game. this game required them to move the initial consonant or consonant cluster of a spoken word to the end of the word, add the /ay/ sound, and pronounce the result. for example, the word ”pig” would become ”igpay”. the children’s skill in this game and other measures of phoneme awareness was highly correlated with their phonological decoding skill in reading nonwords.
we subsequently changed the olson and wise (1992) programs and training environment in several ways. first, we trained 2nd – 5th grade children with reading problems, in small groups of 3 or 4 children, each with their own computer in a small room, and with a teacher always present (olson et al., 1997). a third of each half-hour session included group interactions in reading and related activities. during the other two thirds of the session, children worked independently on the computer programs. the programs included reading stories on the computer with help for difficult words, as in our previous studies. in addition, half of the children spent part of their training time working with programs designed to improve their deficient phoneme awareness and phonological decoding skills. the other children spent all their time accurately reading stories and applying strategies for understanding and remembering the stories.
the phonological training programs were very effective in improving children’s phonological skills, and this improvement was associated with significantly greater gains in several word-reading tests for the youngest and worst readers. however, the older and more skilled poor readers in the 4th and 5th grades seemed to gain as much or more in measures of word reading from spending all their time accurately reading the stories on the computer and practicing their comprehension strategies (wise et al., 2000).
with the new programs, standard-score gains in reading were much stronger than in earlier studies, and the gains were substantially greater than for children who remained in their regular reading class (wise, ring, & olson, 1999). in fact, the gains per hour of instruction were comparable to those from several studies that provided intensive individualized instruction from a human tutor, in addition to regular classroom instruction (torgesen et al., in press). thus, our computer-based instruction with groups of 3 or 4 children and one teacher that replaced regular classroom reading instruction was about equally effective and much more economically efficient than individual tutoring that supplemented regular classroom instruction.
the colorado computer-based training studies were experimental. as such, they were of limited duration (25-30 hours total over 4 months), and they were not systematically integrated with on-going classroom instruction or home activities. the computer-trained children made significant reading gains in this brief period, but few of them improved to normal reading levels. it was clear that many of the children would require much more remedial training and continued reading to reach a normal reading level. for most children with dyslexia, their attainment and long-term maintenance of a normal reading level will depend on continued reading activity at school and in the home, both on and off the computer.
some characteristics of our computer-training programs have recently appeared in commercial products. the kurzweil 3000 (www.lhsl.com/kurzweil3000) program allows users to scan books and other materials for reading on the computer. the scanned images, including both text and pictures, are presented on the screen just as they are in the book. a character-recognition program converts the text in the images to digital representations that can be converted to synthetic speech. kurzweil has generally marketed their programs as compensatory or prosthetic tools to bypass reading difficulties in dyslexia, but passive listening to texts read aloud by the computer is not likely to improve reading skills. children need to actively read and receive support for decoding difficulties to reinforce words they know and to learn new words.
the kurzweil program can be set to allow the targeting of individual words for speech support while reading, although the targeting involves the inconvenient use of a tool bar. the program could be easily modified to increase the convenience of targeting words, and these words could be kept in a file for later review. it would also be helpful to support the inclusion of questions to monitor comprehension, as we have done in our research. kurzweil appears to be working on these modifications (personal communication).
the complete kurzweil 3000 program including color scanning is currently quite expensive in the u.s. ($1895 u.s. as of 18/5/01). a very similar program, including the spell checker described in the next section and a screen reader for web content, has been developed by the texthelp! company (www.texthelp.com) . the cost of this program ($549 u.s. as of 18/5/01) makes it more accessible to schools and individual users. texthelp! is also working to make their program more supportive for reading growth. i believe that eventually the kurzweil and texthelp! programs will be widely used to improve reading skills in children and adults with dyslexia.
word-reading and phonological problems in dyslexia are usually accompanied by severe deficits in spelling. we designed a program (spello) that pronounces children’s spelling attempts for spoken words based on the most common rules of english orthography, allowing them to modify their spellings for further feedback (wise & olson, 1992). unfortunately, the inconsistency of grapheme-phoneme correspondences in frequent english irregular or exception words limits the usefulness of this approach.
spell checkers in most word processing programs will flag misspelled words and some will automatically change common misspellings to the correct forms. when this is done automatically, there is not much support for learning the correct spelling. to support the learning of correct spellings, it may be more beneficial for the computer to provide a range of possible choices in addition to highlighting a misspelled word. however, it is difficult for both computers and people to predict the desired word from many dyslexic spellings. texthelp! is one company that has made significant advances in this area, but its program still has problems interpreting many dyslexic spellings.
one of the most advanced programs has been developed for swedish spelling
by andersson and holtsberg (1998). this program incorporates the authors’ knowledge of dyslexics’ spelling errors, allowing it to correct about 90% of dyslexics’ errors, compared to about 40% for word 98. for example, in swedish, the program can correct errors such as *skokla for ”choklad” (=chocolate), which no other swedish spelling program can manage, the letter string being too disparate from the target. for an analogous example in english, the authentic english error *citon for ”kitten” would bring up a range of possible choices beginning with the letter ”c” in the texthelp program, but not the intended word kitten (andersson and draffan, 2001). andersson and holtsberg’s program incorporates the type of knowledge that ”c” is sometimes used for ”k” in dyslexics’ (and young children’s) spelling errors, so it also offers choices that begin with ”k” and would find ”kitten” in this case. the student’s active selection of the correct choice should improve their independent spelling of that word in future attempts.the spell checker also contains some 8,000 explanatory sentences with words shown in context, to help the student pick the right alternative for homophones and almost homophonic words. ”
learning to read, write, and type
many educators have recognized that early writing attempts can play an important role in learning to read. however, spelling words and composing texts on the computer requires typing skills that most beginning readers and children with dyslexia do not have. there is one excellent and very affordable program that teaches beginning readers to touch-type at the same time that it teaches them to read and spell. read, write, and type (www.talkingfingers.com ), was developed by jeannine herron (see a review of this and other programs by russel smith at www.electronic-school.com/2001/01/0101poweruser.html ). this program teaches the most common letter sounds rather than letter names to correspond to appropriate key strokes. it quickly moves on to the typing of words that can be made from the learned key-letter sound combinations, and eventually to the composition of short stories. children learn to type and to read through spelling and reading their own writing. the short-term benefits of the program for reading development in beginning readers at risk for dyslexia had been demonstrated by torgesen et al., (in press). following this program with continued writing activities on the computer is likely to have further benefits for word reading skills as well as for writing and reading comprehension.
comprehension is the ultimate goal in reading. dyslexics’ comprehension of text may be improved as they develop their word-reading skills, but there is much more involved in reading comprehension. many dyslexics will still have significant problems in reading comprehension even after their basic word-reading skills have improved.
continued reading will help build the vocabulary and general knowledge base that is so important for reading comprehension. programs such as accelerated reader (http://www.renlearn.com) provide comprehension questions on the computer for a wide range of children’s literature. children who use this program typically read from books and then take multiple-choice comprehension tests on the computer to demonstrate that they have read and understood the material. their scores and the material they have read are recorded on the computer so that teachers and parents can track their reading progress. this program is widely used in u.s. elementary and middle schools to promote, monitor, and reward reading. the multiple-choice questions and record-keeping functions of the accelerated reader program are rather simple but surprisingly useful applications of computer technology for literacy development.
multiple-choice questions on the computer may not be the ideal way to assess reading comprehension. many educators argue that the ability to produce a summary of a text that highlights the most important information is a much better indication of a readers’ comprehension of the material. in addition, the prospect of having to write (or speak) a summary gives readers practice in thinking about the most important points of a text while they are reading. the problem with oral or written summaries is that they are difficult to score reliably, even for human raters, and human grading of summaries is very time consuming and expensive.
a remarkable new computer program involving latent semantic analysis (lsa) can automatically score summaries for their semantic overlap with the material that has been read (landauer, foltz, & laham, 1998; http://lsa.colorado.edu ). this program has proved to be as reliable as humans in grading written summaries of a text. the program is initially trained on a very large body of text to represent the semantic relations among words in a high-dimensional space. it then applies this knowledge to assess the semantic overlap of words in the material that is read and words in the reader’s summary. they do not have to be the same words, but they have to be words from semantic spaces shared by the text and the summary to achieve a high score.
latent semantic analysis is a centerpiece in the development of a computer- and web-based reading tutor at the university of colorado that is jointly supported by the national institutes of health, the national science foundation, and the department of education. preliminary studies of its application in the classroom for providing feedback on children’s summary drafts have been very promising (steinhart, 2001). we aim to combine this program with programs similar to those that we have studied experimentally for supporting word decoding and phonological skills (wise et al., 2000). these programs will ultimately be disseminated to schools throughout colorado and their use will be monitored through the world wide web.
web-based assessment of and support for targeted needs, including those of ”poor” and ”normal” readers
computer-based testing and the web have introduced exciting new opportunities in the diagnosis and remediation of dyslexia and other reading problems. i often receive inquiries from concerned parents about their children’s reading difficulties. most schools do not have a systematic assessment program or adequate personnel to administer diagnostic tests. i usually refer them to private clinics for diagnostic testing, but it is not unusual for them to have to pay $500 or more for this service. then, there is the expense of private tutoring. for those who have the resources, that is fine, but many can’t afford these private services.
current computer technology can readily support some basic diagnostic procedures. for example, we have been using computer-administered and scored tests of word recognition, phonological decoding, orthographic coding, spelling, and reading comprehension for the last 20 years in our studies of identical and fraternal twins (olson et al., 1994). similar measures are being developed commercially in the u.s. and europe for computer-based diagnosis that will be much more affordable and assessable than a visit to the clinic. some companies are even offering free diagnostic tests on the web (http://www.thrass.com ). the federally-funded diagnostic and remedial programs under development at the university of colorado will eventually be available on the web at minimal cost.
when tests have been thoroughly standardized and validated, the computer-based diagnosis of different areas of strengths and weaknesses in reading can then be matched with a prescription for and delivery of appropriate remedial programs over the web. for example, a diagnostic program might reveal severe deficits in accurate phonological decoding and phoneme awareness that should be addressed first. if phonological skills and word recognition are reasonably accurate but slow, programs to increase speed might be prescribed. if word-reading skills are reasonably accurate and fluent but comprehension is weak, programs to ensure adequate reading practice, growth in vocabulary, appropriate comprehension strategies, and summarization skills might be prescribed and delivered over the web.
computer-based diagnosis and remediation will work best in conjunction with excellent teaching and social support for reading. we have clearly seen the importance of social interaction and teacher support in our experimental computer training programs (olson et al., 1997). what computers can do is considerably extend the reach of the classroom teacher and therapist to help many more children receive the intensive and individualized support they need in their schools and homes.
finally, i would like to consider the needs of poor readers who may not meet some of the traditional criteria for ”dyslexia”. dyslexia or ”specific learning disability” has traditionally been defined in the u.s. based on low reading, despite normal iq and educational support. the presence of an iq discrepancy is still widely used as a criterion for providing special remedial services in the schools. however, there is increasing recognition that non-discrepant poor readers and dyslexics may have very similar deficits and remedial needs in phonological decoding and word recognition, and they show similar response rates to intervention (lyon et al., 2001, www.edexcellence.net ). there is no reason to limit the benefits of computer-based instruction to the traditional dyslexic. children who read poorly because of poor schools, second-language learning difficulties, or general impairment of intellectual function may also benefit from the extra instructional support that can be provided through computer technology.
even normally progressing readers may benefit from good computer-aided instruction. this could raise the ”normal” reading and writing levels in the population as a whole, posing even greater challenges for children with learning difficulties, and increasing their need for affordable and assessable computer technology to help them reach higher expected levels of literacy.
andersson, b., & draffan, e.a. (2001). assistive technology for better or worse: from swedish and uk speech therapist viewpoints. presentation at the meeting of the british dyslexia association, april 20, york, u.k.
andersson b & holtsberg a: att stava med stöd. om en ny, datorbaserad stavningshjälp för dyslektiker (”stava rätt”.) /supported spelling. on a new computerized spelling aid for dyslexics (”spell it right”)/ project report for the swedish national labor board, the swedish national insurance office and the swedish handicap institute. 1998.
landauer, t. k., foltz, p., & laham, d. (1998). an introduction to latent semantic analysis. discourse processes, 25, 259-284.
lyon, g.r., fletcher, j.m., shaywitz, s.e., shaywitz, m.d., torgesen, j.k., wood, f.b., schulte, a., & olson, r. (2001). rethinking learning disabilities. washington, d.c.: fordham foundation ( www.edexcellence.net ).
olson, r. k., foltz, g., & wise, b. (1986). reading instruction and remediation with the aid of computer speech. behavior research methods, instruments, and computers, 18, 93?99.
olson, r., forsberg, h., wise, b., & rack, j. (1994). measurement of word recognition, orthographic, and phonological skills. in g.r. lyon (ed.) frames of reference for the assessment of learning disabilities: new views on measurement issues (pp. 243-277). baltimore: paul h. brookes publishing co.
olson, r.k. & wise, b.w. (1992). reading on the computer with orthographic and speech feedback: an overview of the colorado remedial reading project. reading and writing: an interdisciplinary journal, 4, 107-144.
olson, r.k., wise, b.w., ring, j., & johnson, m. (1997). computer-based remedial training in phoneme awareness and phonological decoding: effects on the post-training development on word recognition. scientific studies of reading, 1, 235-253.
steinhart, d. (2001) summary street: an intelligent tutoring system to improve students’ writing through the use of latent semantic analysis. unpublished dissertation, university of colorado.
torgesen, j.k., alexander, a.w., wagner, r.k., voeller, k., conway, t., & rose, e. (in press). intensive remedial instruction for children with severe reading disabilities: immediate and long-term outcomes from two instructional approaches. journal of learning disabilities.
wise, b.w. & olson, r.k. (1992). how poor readers and spellers use interactive speech in a computerized spelling program. reading and writing: an interdisciplinary journal, 4, 145-163.
wise, b.w., ring, j., & olson, r.k. (1999). training phonological awareness with and without attention to articulation. journal of experimental child psychology, 72, 271-304.
wise, b.w., ring, j., & olson, r.k. (2000). individual differences in gains from computer-assisted remedial reading with more emphasis on phonological analysis or accurate reading in context. journal of experimental child psychology, 77, 19