Investigating the mechanisms of written word production: Insights from the written blocked cyclic naming paradigm

Abstract

In three experiments, we examined whether similar principles apply to written and spoken production. Using a blocked cyclic written picture naming paradigm, we replicated the semantic interference effects previously reported in spoken production (Experiment 1). Using a written spelling-to-dictation blocked cyclic naming task, we also demonstrated that these interference effects disappear when the task does not require semantically-mediated lexical selection (Experiment 2). Results are parallel to those reported for the analogous spoken production task of reading aloud. Similar results were observed in written spelling to dictation regardless of whether stimuli consisted of words with high or low probability phoneme-to-grapheme correspondences (Experiment 3) revealing the important role of non-semantically-mediated spelling routes in written word production. Overall, our results support the view that similar mechanisms underlie written and spoken production. This includes an incremental learning mechanism underlying semantically-mediated lexical selection that produces long-lived interference effects when multiple semantically similar items are repeatedly named.

Keywords: blocked cyclic naming; lexical selection; semantic interference; spelling to dictation; written production.

Figures

Fig. 1

A cognitive architecture of spoken production. Three routes can be used to produce spoken words: semantically mediated (SM) and non-semantically mediated (N-SM) lexical routes as well as the sublexical route. In reading aloud, any of the three routes can be used. All begin with visual processing to recognize the graphemes in the written word form. When the sublexical route is used, graphemes are directly converted to phonemes. When the SM lexical route is used, graphemes activate the stored word form in orthographic long-term memory, which then activates meaning features in the semantic system and then the stored word form in phonological long-term memory,. Alternatively, the N-SM lexical route (depicted with a dashed line) can be used,, bypassing the semantic system. All routes result in selected phonemes that are held in phonological working memory throughout production of the spoken word. In spoken picture naming, the SM lexical route is used as visual object recognition processes provide input to the semantic system.

Fig. 2

A cognitive architecture of written production. As in spoken production, three routes can be used to produce written words: semantically mediated (SM) and non-semantically mediated (N-SM) lexical routes as well as the sublexical route. In written spelling to dictation, any of the three routes can be used. All begin with auditory processing to recognize the phonemes in the spoken word form. When the sublexical route is used, phonemes are directly converted to graphemes. When the SM lexical route is used, phonemes activate the stored word form in phonological long-term memory, which then activates meaning features in the semantic system and the stored word form in orthographic long-term memory. Alternatively, the N-SM lexical route (depicted with a dashed line) can be used, bypassing the semantic system. All routes result in selected graphemes that are held in orthographic working memory throughout production of the written word. In written picture naming, the SM lexical route is used as visual object recognition processes provide input to the semantic system,.

Fig. 3

Results of Experiment 1: The effect of semantic similarity on response times for written picture naming of words in semantically homogeneous versus semantically heterogeneous blocks. Error bars represent the between-subjects standard errors of the means, corrected for repeated measures using the Cousineau (2005) method.

Fig. 4

Results of Experiment 2: The effect of semantic similarity on response times for written spelling to dictation of words in semantically homogeneous versus semantically heterogeneous blocks. Error bars represent the between-subjects standard errors of the means, corrected for repeated measures using the Cousineau (2005) method.

Fig. 5

Results of Experiment 3: The effect of semantic similarity on response times for written spelling to dictation of words with high and low probability phoneme-to grapheme correspondences presented in semantically homogeneous versus semantically heterogeneous blocks. The left panel shows the results of the high probability condition. The right panel shows the results of the low probability condition. Error bars represent the between-subjects standard errors of the means, corrected for repeated measures using the Cousineau (2005) method.