One Process, Not Two, in Reading Aloud: Lexical Analogies Do the Work of Non-Lexical Rules

Abstract

It is widely held that there are two (non-semantic) processes by which oral reading may be achieved: (a) by known words visually addressing lexical storage of their complete orthography and phonology; (b) by parsing a letter string into graphemes which are translated by rule into phonemes. Irregular words (HAVE) rely on the former, new and non-words rely on the latter. Recent evidence casts doubt on this view; to meet some of this data a revised version is presented. An alternative view is that the phonology of both words and non-words, at each encounter, is retrieved by analogy with all known words having matching segments. In a mixed list of words and non-words, presented singly for pronunciation, phonologically ambiguous non-words (NOUCH) were preceded critically by words with the same ambiguous segments, either pronounced regularly (COUCH) or irregularly (TOUCH). Standard (and revised) dual-process theory predicts that preceding words will not affect pronunciation of non-words; analogy theory predicts that they will. Significant biasing effects, compared to control conditions, support analogy theory, but a further modification to dual-process theory enables it to deal with these results. However the presence in critical non-words of morphemes pronounced consistently or inconsistently with the biased pronunciations significantly affected biasing. This makes the case for lexical analogy theory even stronger. Formal knowledge (descriptive spelling-sound rules) may be used consciously, but does not reflect tacit processes in oral reading, which are better described by a single-process lexical analogy model.