Speakers choose a particular expression based on many factors, including availability of the referent in the perceptual context. We examined whether, when expressing referents, monolingual English- and Turkish-speaking children: (1) are sensitive to perceptual context, (2) express this sensitivity in language-specific ways, and (3) use co-speech gestures to specify referents that are underspecified. We also explored the mechanisms underlying children's sensitivity to perceptual context. Children described short vignettes to an experimenter under two conditions: The characters in the vignettes were present in the perceptual context ("perceptual context"); the characters were absent ("no perceptual context"). Children routinely used nouns in the "no perceptual context" condition, but shifted to pronouns (English-speaking children) or omitted arguments (Turkish-speaking children) in the "perceptual context" condition. Turkish-speaking children used underspecified referents more frequently than English-speaking children in the "perceptual context" condition; however, they compensated for the difference by using gesture to specify the forms. Gesture thus gives children learning structurally different languages a way to achieve comparable levels of specification while at the same time adhering to the referential expressions dictated by their language. (Contains 3 figures and 14 footnotes.)