A Reconsideration of Phonic Generalizations

Abstract

The most frequently occurring 18,000 words from the Kucera and Francis (1967) corpus were analyzed by a computer program. The output consisted of the frequence of occurrence of bigrams, trigrams, tetragrams, pentagrams, hexagrams and heptagrams (called ngrams), and the words in which each occurred. From this analysis two kinds of frequency for each unit were available: 1) word frequency, the number of words in which the ngram occurred and 2) sheer frequency, the frequency of the ngram in running text. The consistency of 23 commonly taught phonics rules was examined using both kinds of frequencies. While there was general agreement with previous studies, there were notable exceptions. In addition to frequency there is evidence that rule complexity is also a crucial variable in determining utility. When the rules were analyzed in terms of consistency and complexity, only nine appeared to be useful. Rules that are complex but consistent may be efficiently taught with a higher-order unit approach.