Three reputed advantages of using a letter-combination (LC) approach, as opposed to a grapheme-unit (GU) approach for initial word-attack and spelling instruction are critically analyzed. While LC reduces the number of elements per word, it also reduces the number of words that can be generated. LC eliminates distortion in the final vowel-consonant elements; but it allows distortion in the initial consonant, which may be a more serious problem. Although LC places vowel letters in a pronunciation-determining environment, environmental constraints are much more complex and may cover at least the whole word. The analysis suggests that the LC approach is probably not maximally appropriate for either reading or spelling instruction and that individual grapheme units and sounds can be best analyzed and synthesized within whole words. (Author/RB)