Do diacritic mismatches disrupt lexical access more for consonants than for vowels?

Diacritical marks across Latin-based orthographies serve different linguistic functions, ranging from suprasegmental stress marking (e.g., Spanish) to phonemic contrast (e.g., German, Finnish), and this distinction shapes how readers process diacritic mismatches during visual word recognition. Turkish is particularly well suited to examine this question because it includes an equal number of diacritical vowels (e.g., o /o/ vs. ö /ø/) and diacritical consonants (e.g., s /s/ vs. ş /ʃ/), with diacritical and non-diacritical forms functioning as distinct letters. This balanced distribution makes Turkish ideal for examining whether consonant-vowel status modulates the processing cost of diacritic mismatches relative to intact words. We conducted two semantic categorization experiments (is the word an animal name? ) in which native Turkish readers responded to words with a diacritical mark either omitted (Experiment 1) or added (Experiment 2) relative to intact words, with vowel and consonant diacritics included in the same experiment. Bayesian mixed-effects models showed a processing cost for both diacritic omission and addition relative to intact words, and the cost was greater for addition than for omission. Critically, we found no evidence that the response-time cost was modulated by consonant-vowel status. These findings are consistent with an account in which a substantial component of diacritic mismatch effects arises during visual feature-to-letter identity mapping, before consonant-vowel differences in lexical constraint exert a measurable influence.