Academic journal article Memory & Cognition

Orthographic Similarity: The Case of "Reversed Anagrams"

Academic journal article Memory & Cognition

Orthographic Similarity: The Case of "Reversed Anagrams"

Article excerpt

Published online: 1 March 2012

© Psychonomic Society, Inc. 2012

Abstract How orthographically similar are words such as paws and swap, flow and wolf, or live and evil? According to the letter position coding schemes used in models of visual word recognition, these reversed anagrams are considered to be less similar than words that share letters in the same absolute or relative positions (such as home and hose or plan and lane). Therefore, reversed anagrams should not produce the standard orthographic similarity effects found using substitution neighbors (e.g., home, hose). Simulations using the spatial coding model (Davis, Psychological Review 117, 713-758, 2010), for example, predict an inhibitory masked-priming effect for substitution neighbor word pairs but a null effect for reversed anagrams. Nevertheless, we obtained significant inhibitory priming using both stimulus types (Experiment 1). We also demonstrated that robust repetition blindness can be obtained for reversed anagrams (Experiment 2). Reversed anagrams therefore provide a new test for models of visual word recognition and orthographic similarity.

Keywords Orthographic similarity · Anagrams · Word recognition · Masked priming · Repetition blindness · Letter position coding

Orthographic similarity: The case of "reversed anagrams"

A general principle underlying competitive network models of visual word recognition is that presentation of a single word partially activates the representations of orthographi- cally similar words, and the activated lexical representations then compete via mutual inhibition until the word is recog- nized. A critical question for these models, therefore, is how orthographic similarity should be defined. Intuitively, words such as home and hose (known as substitution neigh- bors) appear to be similar, as do words with transposed letters, such as salt and slat. Other forms of similarity are less obvious. For example, how similar are words such as paws and swap or live and evil? These reversed anagrams- words sharing all of their letters, but in reverse order-do not seem to be as similar as substitution neighbors, perhaps because the latter share letters in the same positions.

The intuitive notion that letter position plays a determin- ing role in orthographic similarity is supported by the results of masked form priming experiments. In these studies, a masked and briefly displayed letter string (the prime) pre- cedes a target letter string, and the viewer responds to the target by rapidly classifying it as a word or nonword (the lexical decision task). In general, responses to target words preceded by similar nonword primes (e.g., the prime glnr paired with the target BLUR) are facilitated, relative to a nonsimilar control condition; however, when target words are preceded by similar word primes (e.g., the prime blue paired with BLUR), the direction of the priming effect can be inhibitory (Davis & Lupker, 2006). As will be dis- cussed, the results of masked-priming studies have been highly influential in the development and evaluation of the letter position coding schemes used in formal models of visual word recognition (e.g., Davis & Bowers, 2006; Grainger, Granier, Farioli, Van Assche, & van Heuven, 2006; Guerrera & Forster, 2008; Peressotti & Grainger, 1999; Schoonbaert & Grainger, 2004).

Early models of word recognition such as the in teractive activation model (McClelland & Rumelhart, 1981) used a slot-coding scheme, in which each letter is assigned to a position-specific slot (e.g., the word home is coded as hx, o2, m3, e4, and hose as hx, o2, s3, e4). Under this scheme, two words are similar only if they share letters in the same slots; thus, words such as salt and slat are no more similar than are salt and spot (Andrews, 1996). Slot-coding schemes fell out of favor with the advent of studies demonstrating that prim- ing a target word such as USHER with the transposed-letter prime nhser facilitates the response to the target more than does priming it with iifner, even though both primes share exactly three slots with the target (Perea & Lupker, 2003). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.