Abstract
The paper provides fine-grained evidence concerning the development of syllabic consonants /r l/ in Czech, that is only sketched in the existing literature. The evidence is based on an automatic parser that identifies potential syllable-projecting segments according to sonority. The parser was applied to six verse texts from the 14th–16th centuries, which show a strong tendency towards octosyllabicity. The data provided by the parser newly reveal that the shift from non-syllabic to syllabic /r l/ is position-dependent: word-medial non-syllabic strings C(r/l)C change more rapidly than non-syllabic word-final ones C(r/l)#. This finding is in line with a cross-linguistic observation that non-syllabic C(r/l)C are marked, hence they are regularly syllabified prior to less marked C(r/l)#.