A ROWS is a ROSE: Spelling, Sound and Reading

This experiment is based on the following paper:

Van Orden, Guy C. (1987). A ROWS is a ROSE: Spelling, sound and reading. Memory and Cognition, 15, 181-198. Reprinted by permission by the Psychonomic Society, Inc.


Dual-route reading models propose that words can be decoded in two ways: you can recognize the word on sight (direct route) or you can sound out the word using spelling-to-sound rules of the language (phonologic route). In early versions of this model there was little cross-talk between the two processing pathways, so that direct access to lexical memory would provide the correct pronunciation of a word without any phonological mediation. Furthermore, since the direct route was faster than the phonologic route, reading by directly accessing memory was considered only to involve the processing of visual or orthographic information about text. Since skilled readers have extensive sight vocabularies, and so presumably use the direct route when reading, most theories considered phonology to be important for beginning readers but not used by skilled readers.

In this study Van Orden shows that phonological mediation can occur in skilled reading. He employed a categorization task in which subjects judge whether a target word is a category exemplar. For example, “Is PAIR an exemplar of the category FRUIT?” According to Van Orden:

The phonological mediation hypothesis predicts that false positive categorization errors should occur as a function of the phonological similarity of a stimulus foil to some category exemplar. Accordingly, when the target word is a homophone foil such as ROWS, its phonological representation should strongly activate the lexical entry of the category exemplar ROSE prior to word identification. If so, the homophone foil ROWS will tend to be misidentified as the flower ROSE. (p.182)

Therefore, when a subject is asked to say whether or not the word is an exemplar and to read the word aloud, if the subject answers “YES, ROZ” for the category A FLOWER then, the subject has activated phonological information about the word.

Dual-access theories would predict categorization errors both when foils sound like category exemplars and when they are spelled similarly to category exemplars. This prediction was tested using homophone foils that varied in the degree to which they were spelled similar to their sound-alike category exemplars. Target foil MEET (for the category A TYPE OF FOOD) is very similar in spelling to MEAT. Target foil ROWS is less similar in spelling to ROSE. If an orthographic representation is used directly in word identification, then the likelihood that this process will be misled by orthographically similar foils should be a function of the number of orthographic characteristics that they share with a category exemplar. Thus, MEET is more likely to be mistaken for MEAT than ROWS is to be mistaken for ROSE.


The full text describing the experiment is available here.

The experiments are available below.

vanorden_win.zip (129 KB)

vanorden_mac.zip (178 KB)