What determines syllabification -- sonority (Zec 1988) or perceptibility (Steriade 1997)?  We elicited syllables from 16 English speakers via the 'sounding-out' language game (used to teach reading), and tried to predict the odds that an original coda consonant would be followed, rather than preceded, by a vowel in game outputs (e.g., [d] in 'seed' -> [s@ + i + d@] vs. [s@ + id]).  For each consonant, sonority was estimated following Parker (2002), and the perceptual advantage of CV over VC was estimated from confusion matrices (Cutler et al. 2004).  Results from 16 speakers show that the odds of a CV vs. VC output are closely related to sonority, but unrelated to CV-VC perceptual advantage.