Verification of nonwords: The baseword frequency effect in children's pseudohomophone reading

In this study, we investigated the baseword frequency effect in children and its implications for models of visual word recognition. The baseword frequency effect reflects the finding that response latencies in the lexical decision task to nonwords derived from high-frequency basewords (e.g., GREAN derived from GREEN) are shorter than for those derived from low-frequency basewords (e.g., SLEAT derived from SLEET). Importantly, the baseword frequency effect presents a challenge to current activation-based models of visual word recognition. One explanation for this effect is that the orthographic representations of high-frequency basewords are easier to access. This allows a quick progression to a verification stage in which the exact spelling of a stimulus is checked, upon which the lexicality decision is then based. The main goal of this study was to investigate whether such a verification mechanism is specifically modulated by the quality of the orthographic lexicon. We tested whether the baseword frequency effect was evident in children's lexical decisions to pseudohomophones (PsH) and whether verification accuracy varied as a function of children's orthographic knowledge. The baseword frequency effect in response latency was observed in both German-speaking adults and children. Children's spelling skills significantly influenced the accuracy of the verification stage in their responses to PsH. These findings imply that verification is an integral part of word reading and thus should be included in computational models of visual word recognition.