Hachmann, W. 1 & Moeller, M. 2
1 Max Planck Institute for Human Development Berlin
2 Technical University Berlin
We investigated naming performance during artificial script learning on differentially complex orthographic rules and ran an error analysis on the actual utterances to evaluate the effects of rule complexity over three sessions. Participants learned to read familiar words in a new script including four grapheme changes. These changes represented a design of 2 new polygraphs and 2 new monographs crossed with 2 monophones and 2 polyphones.
Naming accuracy results reflected rule complexity according to the hypothesized criteria: context sensitivity (polygraphs), phonological plausibility, and rule-to-sub-rule relation. In latencies, however, new polygraphs elicited longer response times irrespective of phonological plausibility.
Zooming into accuracy, the error analysis provided a clear pattern of increasing differentiation of rules and graphemes. The implicit differentiation tapped visual interferences (similarly looking new letters), proactive interferences (new letters that look like familiar letters), and rule-based interferences. The latter covered confusions between letters involved in the new rules, interference between new and familiar rules, and spill-over to thus far unaffected letters. Differentiating these rule-based interferences required more training than resolving other interferences. This shows how that the kind and the combination of rules affect the speed and accuracy with which new letters and orthographic systems are learned.