Whole language was just one of many alternative programs that were merged together into a single control condition. Specifically, the whole language interventions were combined with “balanced literacy,” “whole word” and other forms of alternative programs that included unsystematic or no phonics. The key finding from the National Reading Panel was that systematic phonics was more effective than the average performance in the control group that included various forms of reading instruction. As elementary point of logic, if you compare systematic phonics to a mixture of different alternative methods, only a subset of which are whole language, then you have not tested systematic phonics compared to whole language.
More important, when Camilli et al. (2006) reanalyzed the National Reading Panel (2000) data set and directly compared systematic to unsystematic phonics (excluding studies that had no phonics, such as “whole word” interventions), the advantage for systematic phonics was greatly reduced and no longer statistically significant.
This undermines the claim that systematic phonics is more effective than whole-language instruction that includes unsystematic phonics. Nevertheless, this finding has largely been ignored. The National Reading Panel has been cited will over 22,000 times, and over 2,000 times since 2017. By contrast, the Camilli et al. (2006) paper has been cited a total of 58 times, and only 9 times since 2017 (with 3 of these citations coming from us).
This conceptual confusion persists. Bowers (2018) shows that every subsequent meta-analysis taken to support systematic phonics over whole language has made the same mistake of comparing systematic phonics to a mixture of different methods, or comparing systematic phonics to interventions that included no phonics. Accordingly, none of these meta-analyses should be taken to support systematic phonics over whole language.
Moreover, Bowers (2018) points out a host of additional fundamental problems with these meta-analyses that further undermine this conclusion. There really is little or no empirical evidence to support the conclusion that systematic phonics is best practice. The fact that this claim is repeated thousands of times in the literature does not make it so. But it is somewhat of a scandal that the research is so consistently misrepresented in the literature.
This is no victory for whole language, either. The two approaches appear to be equally good (or bad) at improving reading in schools. If you are not happy with the outcomes of whole language, you should similarly be unhappy with the outcomes of systematic phonics, and vice versa. This is the conclusion we draw, and we hope it motivates researchers and teachers to consider alternative methods of reading in the United States (where whole language is still commonplace), England (where systematic phonics has been the norm since 2006) and every other English-speaking country.