Graphemic cohesion effect in reading and writing complex graphemes

Abstract

AU /o/ and AN // in French are both complex graphemes, but they vary in their strength of association to their respective sounds. The letter sequence AU is systematically associated to the phoneme /o/, and as such is always parsed as a complex grapheme. However, AN can be associated with either one phoneme (/A)/ in e.g., CRAN /kR/ “notch”) and be parsed as a complex grapheme; or with two phonemes (/an/ in e.g., CANE /kan/ “duck”), thus being parsed as two simple graphemes. As a consequence, AU would be a more cohesive grapheme than AN, for which there is a parsing ambiguity. We examined whether the reading and writing systems take into account this potential parsing ambiguity due to the graphemes’ degree of cohesion when processing complex graphemes. Experiment 1 consisted of a letter detection task. The participants had to detect, for example A in strongly cohesive complex graphemes (e.g., AU /o/) or weakly cohesive complex graphemes (e.g., AN //). A was detected faster in weakly cohesive complex graphemes than in strongly cohesive ones. In a handwriting task (Experiment 2) we found that weakly cohesive complex graphemes (e.g., ON) yielded longer programming times than strongly cohesive ones (e.g., OU), suggesting that the handwriting system also takes into account the potential decomposability of the complex graphemes into either one (ON //) or two (O+N /on/) units. Overall, our results show an effect of parsing ambiguity due to graphemic cohesion of complex graphemes; these results should be accounted for by current models of written word processing and spelling.

Keywords:

• Graphemes
• Letter-detection task
• Handwriting
• Parsing

Acknowledgements

We thank Piennie Zwitserlood and two anonymous reviewers for their helpful comments on earlier versions of this manuscript. We are also grateful to Melinda Woodley for her valuable comments.