My colleagues and I explored modality-specific properties unique to signed languages, investigating how three signed languages (Italian, French, and American) marked deixis and anaphora as text cohesion and tracking devices in signed discourse (Antinoro Pizzuto, Rossini, Sallandre, & Wilkinson, 2008). Not only did the typological study show that deixis and anaphora may be expressed by pointing signs, lexicalized signs, and signed forms produced with complex manual and non-manual components (all depictive constructions on hand, body, and the face), both deictic and anaphoric functions appeared to be similar across the three signed languages. Although other studies have analyzed deictic and anaphoric functions with specific lexical and morphological forms, our study found that depictive constructions of highly iconic structures (including classifier constructions and role-shifting/constructed action) can function as anaphora (role-shifting/constructed action in this context refers to the adoption of a character that does not refer to the ego speaker in the present time.) This new discovery of the anaphoric function in depicitive constructions brings more understanding to how signers and listeners keep track of referents in signed language discourse.