[Susan notes: Here is example of letter than explains the article so people can understand the argument even if they did not see the article.]
Published in Scientific American
"How Reading Should be Taught" (Raynor, Foorman, Perfetti, Pesetsky and Seidenberg, March 2002) presents an inaccurate definition of whole language and an incomplete view of the research on learning to read.
Contrary to what Raynor et. al. claim, whole language is not a derivative of the "whole word" approach and it does not forbid the teaching of phonics. The whole word approach simply asks students to memorize sight words, and is an inefficient way to teach reading. Whole language is based on the well-supported hypothesis that we acquire language when we understand messages and we learn to read when we understand what is on the page. The core principle of whole language is providing children with interesting texts; the teacher's task is to help make the texts comprehensible. As Raynor et. al. note, context is one way of doing this, but it is not the only way. Some direct teaching of phonics also helps make texts comprehensible. There are, however, severe limits on how much phonics can be taught and learned: As Raynor et. al. point out, some rules are extremely complex, many rules do not work well, and different commercial series teach different rules. Most of our knowledge of phonics is the result of reading, not the cause.
Raynor et. al. claim that research shows that phonics-based methods are superior to whole language methods in helping children learn to read. This is not correct. Much of this research (eg some of Chall's review and the Arabic study Reynor et. al. describe) actually compares phonics-based teaching to whole-word methodology, not whole language. As Raynor et. al. note, the National Reading Panel (NRP) recently reviewed studies comparing phonics-based methods with those labeled whole language. The NRP did not consider the crucial variable of how much reading the children did. I reviewed the same studies, as well as those the NRP inappropriately excluded, and concluded that children in classes in which more meaningful reading was done outperformed those in classes in which less reading was done on tests of reading comprehension. Meaningful reading is the crucial variable in interpreting these studies and it is the core of whole language.
Raynor et.al. reported that in Evans and Carr's 1985 study children in "traditional" reading instruction (termed "routinized performance") did better on tests of reading comprehension than children in what Raynor et..al. considered to be "whole language" type classes (Evans and Carr termed this group "language-oriented" and did not use the terms "whole language.") Children in the routinized performance class read more, spending about five minutes more per day on silent reading, which amounts to about 18 more hours of reading over the year. Also, the routinized performance classes emphasized "contextual meaning" more (p. 333). The language-oriented group did more oral reading of stories the children had written themselves or had dictated to the teacher, an activity that entails less new meaning. Thus, a careful look at what the children actually did reveals that the results of this study are consistent with those of other studies showing that children who read more do better on tests of reading comprehension.
Raynor et. al. also appear to have missed Elaine Garan's recent review of the NRP research on phonics, which appeared in the Phi Delta Kappan last year. Garan concluded that NRP's own analysis shows that heavy phonics instruction primarily impacts tests of regularly spelled words presented in isolation, and has little impact on tests of reading comprehension given to children beyond grade 2.
Stephen Krashen, Ph.D.
University of Southern California