by Nell K. Duke
University of Michigan
April 16, 2014

Periodically, I still hear students told that, “When two vowels go walking, the first one does the talking.” In other words, so the generalization goes, when there are two vowels side-by-side in a word, they represent the long sound of the first of the two vowels, as in the word rain, for example. The problem is, this generalization actually holds true less than half the time—considerably less, depending on whose analysis you read.

I recently had the opportunity revisit a classic analysis by Francine P. Johnston reported in the 2001 article The Utility of Phonic Generalizations: Let's Take Another Look at Clymer's Conclusions. Johnston reviewed a landmark study on how often phonics generalizations apply that was first published by Theodore Clymer in 1963, as well as several replication studies that followed. She also reported on her own analysis, which employed the American Heritage Word Frequency Book (Carroll, Davies, & Richman, 1971), which was created from a wide variety of reading materials used in grades 3 through 9. Johnston too found that the “when two vowels. . .” generalization often did not apply, nor did any related broad generalization. However, when she looked at generalizations for specific vowel pairs, she found   that regularity was much higher. For example, in her database

• aw made the sound in saw 100% of the time
• oy made the sound in boy 100% of the time
• oi made the sound in join 100% of the time

Ay, oa, ee, ai, ey, and au also represented one particular sound more than three-quarters of the time. Another set of vowel pairs represented either one of two sounds three-quarters or more of the time: ow (snow or how), ew (blew or view), oo (book or boot), and ei (eight or either).

The well-worn “final e” or “silent e” generalization (variously stated to mean something along the lines that in a word ending with a vowel, one or two consonants, and an e, the first vowel represents its long sound and the second e is silent) behaves similarly. In her database, it was true 77.7% of the time for a-e, as in cake, but only 16.6% of the time for e-e, as in these (i-e met the generalization 74.2% of the time, o-e 58.4% of the time, and u-e 76.9% of the time).

What do we make of all this? Johnston concludes, “Broad phonic generalizations are not especially useful in very many cases. However, that should not be interpreted to mean that phonics instruction is not useful or that English orthography is too irregular to be the subject of study. Research in the years since Clymer's study offers ideas about alternatives to the teaching of generalizations” (p. 140). Johnston goes on to suggest strategies that are still recommended (and research-supported) today, such as teaching specific sound-letter relationships (e.g., the sound commonly represented by oi), teaching phonograms or rimes (e.g., -ake), developing orthographic knowledge through word sorting, and teaching students to be flexible in decoding, trying one likely vowel sound and then, if that doesn’t work, trying another. We can’t boil down English orthography into a few simple generalizations, but we have lots of tools to help students deal effectively with its complexity.

Nell Duke is a member of the International Reading Association’s Literacy Research Panel. Reader response is welcomed. E-mail your comments to LRP@/.