Spelling

Communication by spelling would, of course, be painfully slow, and the listener would presumably find it nearly impossible to organize the phonologic segments into the larger units of words and sentences. It is also relevant to the rate problem that even if it could somehow be solved in production, the result would defeat the ear. Speech delivers phonologic information at rates of 10 to 20 consonants and vowels per second. But if each consonant and vowel were a unit sound, as it would be in an acoustic alphabet, rates that high would strain the temporal resolving power of the ear and overreach its ability to keep sequential order straight. Thus, an acoustic alphabet is impossible if people are to speak and listen as fast as they must.

What knowledge does the reader need in order to accomplish this feat? In the case of an alphabetic orthography like English, we argued that the reader needs two kinds of information. One is knowledge of the spelling-sound correspondences of the language, or what we have called the orthographic cipher.

Where does this knowledge come from? We're not certain, but we're confident that it does not result from direct instruction (at least it does not result directly from direct instruction). For one thing, the rules the child is taught in phonics are too few. Moreover, the rules of phonics are context-free rules, whereas nearly every spelling-sound correspondence in English depends on its context. The rules of the cipher, if they exist, are context-dependent rules.

Acquisition of the cipher not only changes the child's reading ability, it also changes the child's spelling. The cipher reader spells phonetically (camel as kaml), whereas the child who does not have the cipher seldom does (camel as qzmny). In keeping with this, the cipher reader's spellings correspond in length to the target word, whereas the noncipher reader's errors are only loosely related in length (rain might be misspelled with eight letters). Most dramatically, the cipher reader spells only with letters, whereas the child without the cipher may insert nonalphabetic matter into his or her spellings.

The cipher, then, is of central importance to both the child and the adult, but it is not enough. Students of English orthography (notably spelling reformers) have noted that there are many exception or irregular words in our language, words that depart from the regularities of the orthography. However, we should note that even more numerous than the irregular words are what might be called polyphonic words-words containing letter sequences that can be pronounced in more than one way.

he term orthographic processing is used to describe performance on tasks that could not be accomplished using only the cipher. What these tasks require is knowledge of the spelling or pronunciation of a specific word. For example, in Richard Olson's ( Olson, Kliegl, Davidson, & Foltz, 1985) orthographic task, the subjects had to decide which of two homophonic strings (e.g., rane-rain) was actually the word. Or in Stanovich and West ( 1989) homophone choice task, the subjects had to decide which of two homophones was semantically related to a third word (e.g., rose: flour or flower?). In neither task does deciphering yield the answer, because instead it yields two identical phonological forms; something more, namely knowledge of how that specific word is spelled, is required.

The first of these tasks was a naming task. Following Baron, we asked subjects to name regular, exception, and pseudowords, and measured their latencies. In a second task, we used the same three kinds of words in a task we called spelling verification: Each subject heard a word, then saw a spelling of that word and decided whether it was correct or incorrect. The task was performed three times, once with regular words, once with exception words, and once with pseudowords.