Modeling lexical decision and word naming as a retrieval process.

Abstract

We argue that rule-like phenomena in naming and lexical decision reflect the collapsing of information that occurs during retrieval from the lexicon, and that complex patterns in performance reflect the pattern of correlation that exists in the reader's lexicon rather than mapping rules wired into, or learned by, the processing apparatus. By using a lexicon built to scale, we show that simple retrieval operations applied to a large corpus of words correctly predict an interaction of word frequency by spelling-tosound regularity in naming, a frequency main effect in lexical decision, sensitivity to orthographically defined syllable-like structures in lexical decision, and an interaction of number of syllables with word frequency in naming. Seidenberg (1985) found a regularity effect, but it was limited to cases in which "strange" words (their adjective), such as colonel, were included. Later, Besner, Twilley, McCann, and Seergobin (1990) reported a regularity effect when low-frequency homophones were included in the stimuli. Although the referee cited Andrews (1982) for evidence of a regularity effect in lexical decision, she reported a consistency effect, but not a regularity effect.

Phonographic neighbors, not orthographic neighbors, determine word naming latencies

“…The primary constraint arising is that generalization of print-sound correspondence can be influenced by individual words. This arises in IA models (see, e.g., McClelland & Rumelhart, 1981) and the IAbased portion of the DRC (Coltheart et al, 2001), as a consequence of the error arising from individual words in backpropagation PDP models (e.g., Plaut et al, 1996;Seidenberg & McClelland, 1989;Zorzi et al, 1998), and from the retrieval of individual words in memory-based models (e.g., Kwantes & Mewhort, 1999).…”

“…This is because they do not have orthographic word representations; rather, letters are processed immediately into intermediate or phonemic representations. Such models include parallel-distributed processing (PDP) models (see, e.g., Plaut et al, 1996;Seidenberg & McClelland, 1989;Zorzi, Houghton, & Butterworth, 1998) and retrieval models (e.g., Kwantes & Mewhort, 1999). Moreover, PDP models do produce orthographic neighborhood size effects (Sears, Hino, & Lupker, 1999).…”

An instance theory of associative learning

“…Minerva 2 is an instance theory of human memory for explaining episodic recognition and judgment of frequency. The theory has since been applied to a wide range of memory phenomena (e.g., Arndt & Hirshman, 1998;Benjamin, 2010;Clark, 1997;Dougherty, Gettys, & Ogden, 1999;Goldinger, 1998;Hicks & Starns, 2006;Jamieson, Holmes & Mewhort 2010;Jamieson, Hannah & Crump 2010;Jamieson & Mewhort, 2009a, 2009b, 2010Kwantes, 2005;Kwantes & Mewhort, 1999;Kwantes & Neal, 2006;Thomas, Dougherty, Sprenger, & Harbison, 2008).Informally, Minerva 2 is a theoretic framework that describes the memorial processes involved in representing, storing, and retrieving instances of experience. The model assumes that each experience is stored in memory as a unique trace (i.e., an instance).…”

When parallel processing in visual word recognition is not enough: New evidence from naming

“…However, Zorzi (2000) also simulated these data with a model that operates solely in parallel. Furthermore, Kwantes and Mewhort (1999) simulated these data with a serial processing model that has no rules for converting orthography to phonology. The human data reported by Rastle and Coltheart therefore neither require a serial processing account, nor successfully discriminate among a number of computational models of reading aloud.…”

“…Implemented computational models of reading aloud include the parallel distributed processing (PDP) model (Plaut, McClelland, Seidenberg, & Patterson, 1996), the connectionistdual process (CDP) model (Zorzi, Houghton, & Butterworth, 1998), the LEX model (Kwantes & Mewhort, 1999), and the dual route cascaded (DRC) model (Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001). With several competing models on the table, a program of study devoted to adjudicating among the contenders is necessary in order for one to achieve a better understanding of human word recognition processes.…”