ABSTRACT

Reading requires access to word-specific knowledge about spelling patterns—an “orthographic lexicon.” This paper reports on an attempt to explore the question of whether the orthographic lexicon (or lexica) used during reading is (or are) also accessed during the performance of skills that depend on orthographic knowledge but need not involve visual input. The tasks examined are: (1) writing a defined or spoken word; and (2) identifying a word spelled out letter-by-letter.

The methodology is based on long-term repetition priming. It is first argued that, under some conditions, the facilitation of visual recognition by a prior visual encounter with the word can be interpreted as reflecting a persistent change in the accessibility of that word’s representation in a lexicon. Given such conditions, one can compare to this visual → visual priming effect the effect produced by a prior nonvisual task requiring access to the word’s orthography. If the nonvisual task fails to prime later visual identification, this suggests it does not result in access to the same lexicon. If priming occurs, this is compatible either with the hypothesis that both tasks access the same orthographic lexicon, or with a hypothesis of separate lexica so linked that activation of a word’s representation in one can activate its representation in another. The latter hypothesis allows the possibility that there will be conditions under which priming does not occur.

Three experiments on writing are described. In one (already reported elsewhere), writing a word blind in response to a (visual) definition failed to prime a later visual lexical decision. But in two others, writing a spoken word blind while matching it to a (visual) definition did prime later visual lexical decision or semantic categorisation. In combination these conflicting outcomes are, prima facie, compatible only with the hypothesis of separate but linked lexica. However, other aspects of the data suggest that only the first experiment fully met the criteria for a priming effect associated solely with lexical identification.

300A final experiment shows that naming an item spelled letter-by-letter primed later visual lexical decision for words (but not for nonwords), suggesting that this task does activate the lexicon used for reading.