Methods and apparatus for inhibiting hyphenation clustering. In one aspect, the invention identifies a sequence of consecutive text lines, one or more of which end in hyphens, and calculates a hyphenation penalty value based on the number and proximity of hyphens within the sequence consecutive text lines. The hyphenation penalty can be aggregated with other format penalties to aid the identification of a paragraph layout having a best overall visual appearance.