Effects of grapheme-to-phoneme probability on writing durations

The relative involvement of the lexical and sublexical routes across different writing tasks remains a controversial topic in the field of handwriting production research. The present article reports two experiments examining whether or not the probability of a grapheme-to-phoneme (G-P) mapping affected production during copy of polyvalent graphemes embedded in French (Exps. 1a and 1b) and Spanish (Exp. 2) known words. The relative probabilities of two different G-P mappings associated with the same polyvalent grapheme were manipulated (higher vs. lower probability). In Experiment 1a, we used the polyvalent French grapheme E. Writing durations revealed that the interletter intervals (ILIs) located before and after this letter were shorter and that the letter itself was executed faster in the condition of higher probability of the G-P mapping (e.g., S E RVICE, "service") than in the lower-probability condition (e.g., S E MAINE, "week"). In Experiment 1b, we used the sequence TI (e.g., VIC TI ME-MAR TI EN, "victim-Martian"), which is less frequent. In this case, we failed to observe significant differences between the conditions. In Experiment 2, effects similar to those obtained in Experiment 1a were found with Spanish words using different pronunciations of the letter C (e.g., DES C ANSO-DES C ENSO, "rest-descent"). Altogether, these results reveal that the link between a grapheme and a phoneme is weighted according to its probability in the language. Moreover, they suggest that a two-phase route linking graphemes to phonemes and phonemes to graphemes is functional during copy.