Research on serious games shows that complementary design measures are often needed to increase effectiveness. The present study investigated the role of scripted collaboration. In such a collaboration, players are assigned roles or given specific tasks to stimulate them to communicate more about essential game aspects (eg, arguing why a move should be made versus merely suggesting a move), and thereby raise learning. A risk of scripted collaboration is that it can constrain communication so much that motivation drops. An experiment with eighth-grade students is reported in which a scripted and non-scripted condition were compared for their effects on the level of dialogic acts, motivation and learning outcomes. In the scripted condition, players were assigned opposing roles in a game that revolved around making smart choices by considering each perspective. The findings supported the prediction that scripting would raise the level of the dialogues. Scripting led to significantly more dialogic acts about deeper level game features. In addition, players in the scripted condition achieved a significantly higher mean score on a knowledge posttest. An exploratory analysis showed that higher level dialogic acts correlated with higher posttest scores. Motivation was the same in both conditions.