Spotting vowels is easy; even a computer can do it. The vowels (a e i o u y) may not predict reading levels as reliably or as accurately as syllables can. But being closely associated with the syllables, vowels can measure text difficulty.
A syllable may have one or more vowels: by has one, tie has two, course has three and queue has four. In ‘The Vocalic Cloze Procedure’, I wrote: “The average syllable has three letters, of which two are usually consonants and one is a vowel.” I chanced upon a table of relative frequencies of alphabetic characters in Simon Singh’s The Code Book. H. Beker and F. Piper’s table had first appeared in Cipher Systems: The Protection Of Communication.
Based on a sample of 100,362 letters, the authors calculated the frequency of each letter of the alphabet. I summed the frequencies of only the vowels and obtained the figure 40.2%. This means that there are 1.2 vowels per syllable and 2 vowels per word.
That should suffice for us to derive the Vocalic Readability Index (VRI) = AVS / 4. The AVS is the average vowels per sentence, which is divided by 4 to match the text to the reading or grade level from 1 to 17+. The VRI can be easily derived from the W-Index or the S-Index or the L-Index; these indices of mine are discussed in another article titled ‘Seven Indices Of Readability’.
I tested the VRI on the 10 graded samples found in the appendix of Jeanne S. Chall and Edgar Dale’s Readability Revisited (the new Dale-Chall readability formula). The VRI predicts within two grade levels on all the tested samples; and within one grade level on 50 % of the samples. The VRI was able to predict exactly the reading level of the passage beginning ‘The controversy over the laser-armed satellite …’, which has a reading level 9-10. There were 189 vowels in 5 sentences. Therefore, the AVS is 189/5 = 37.8 and the VRI is 37.8/4 = 9.45.
To obtain a better estimate, let V25 be the number of vowels in 25 sentences. Then the VRI = V25 / 100. What is more, this formula can be easily computerised.