The Complexity Argument: Johnson (2001)

Date: 2015-10-07; view: 390.

Clymer (1963, 1966) investigated 45 phonic generalizations of words in four basal series and concluded that many did not work very well. This result has been a central part of the argument against over-teaching phonics. Here are two well-known examples: The rule "when two vowels go walking the first does the talking" (when two vowels appear side by side, the long sound of the first is heard and the second is silent, as in "bead") worked in only 45% of the cases Clymer examined, and the final e rule (first vowel is long, final e is silent, as in "cake") worked in only 63% of the cases.

Johnson (2001) re-examined Clymer's conclusions. On reading the title of her paper ("The utility of phonics generalizations: Let's take another look at Clymer's conclusions") and the short summary under the title ("English orthography is not easily reduced to a few rules, but there are some general recommendations for teaching about vowels that can be helpful."), one gets the impression that Clymer's results will be contested, and that a new case for direct phonics instruction will emerge from this article. Johnson promises to review what she considers neglected studies that followed Clymer, and promises to present a new analysis.

The neglected studies, however, replicate Clymer's results, with only a few alterations, a conclusion Adams (1990) also arrived at in her discussion of Clymer's work. And Johnson's new analysis confirms that extensive phonics teaching is a hopeless endeavor. She provides, in fact, dramatic evidence that English phonics is extremely complex, which was, in fact, Clymer's point.

Here is an example. As noted above, Clymer concluded that the "two vowels walking" rule applied only 45% of the time. In her re-analysis, Johnson concludes that this rule works well for five two-vowel combinations: ay (96.4%), oa (95%), ee (95.9%), ai (75%) and ey (77%). Of course, one can easily dispute that 75 or 77% accuracy is enough to justify this conclusion, but more serious is the fact that the situation is a disaster for the 14 other two-vowel combinations Johnson presents in her table 2. While four other two-vowel pairs are regular, none of these meet the criteria of the "walking" rule. Four additional pairs have two possible pronunciations and four more pairs have three possible pronunciations. Finally, another two pairs with two possible pronunciations were considered "very rare." (One of them appears in "fruit," "suit," and "build" and the other in "Asia," "piano," and "official," hardly arcane words.) Very few of these alternative pronunciations follow the "walking rule."

As noted earlier, Clymer concluded that the final e rule worked only 63% of the time. Johnson concludes that for some combinations, it does a bit better (a-o, as in "cake", 77.7%, i-e as in "five," 74.2%, u-e, as in "rule," 76.9%) but it is less efficient for others (for o-e, as in "stove," 58.4%, e-e, as in "these," 16.6%). She concludes that this rule is "surprisingly reliable when restated" (p. 139) and that teachers can teach this rule "with confidence" (p. 138) as long as they encourage a "flexible strategy." This does not seem to be a real step forward: we are trading a simple rule that works 63% of the time for a far more complex rule that is only slightly more efficient.

Johnson's work, in other words, is a strong confirmation of Clymer's: Simple rules don't apply to a large percentage of words. Yes, one can come up with rules that cover more words, but they also have numerous exceptions and are very complex.

Johnson has no clear program for teaching sound-spelling correspondences, other than the suggestion that we also consider rules for higher order units, such as rimes and "vowel patterns," also, as she points out, numerous and complex.

Buried deep in Johnson's article is the suggestion that some children can acquire phonics generalizations by reading. As noted earlier, Smith (e.g. 1994) has hypothesized that most of our knowledge of phonics is the result of reading and not the cause. Johnson's view differs somewhat from Smith's in that she claims that some children can indeed acquire sound-spelling correspondences by reading, while others "need systematic instruction" (p. 141). No evidence is provided for this extremely important claim, a claim that runs counter to current official state and federal government policy that all children must have systematic, intensive phonics instruction.

To support such a claim, one would have to show that there are substantial numbers of children who have learned to read without extensive phonics training (this is easy to find), and also substantial numbers of children who cannot "learn to read by reading," who require extensive phonics instruction. The existence of this second group has never been demonstrated: To do so, one must find large numbers of children who have been read to, who have substantial exposure to comprehensible and interesting texts, and who nevertheless fail to learn to read.

One sees, of course, some children who learn to read less quickly than others do, but this is a statistical necessity in any phenomena that exhibits any degree of variability. No matter what, we will always have children who fall in the lowest 25% in rate of learning to read. What we very rarely see are those who never learn to read despite the availability of comprehensible and interesting print.

Which phonics generalizations are useful? Which ones really help children understand text? I suggest we ask the real experts: Teachers who have helped children learn to read for many years. A consensus of experienced practitioners will tell us if it is worthwhile to tell children that the a-e combination is pronounced with the long vowel and the final e silent (except when the final syllable is unaccented - then the vowel is pronounced with a short-i sound, as in "palace," or the combination is "are," with words such as "have" and "dance" as exceptions). How many of us who easily and fluently read words with the a-e combination were ever aware of this rule?¹