The present study investigated an extension of the emotional design hypothesis in multimedia learning for textual parts of multimedia instruction. In an one-factorial experimental mixed-methods design with three groups, participants learned with multimedia instruction incorporating a positive or negative emotional text design or the original learning text. Both the positive and negative emotional text design led to better learning outcomes compared with the control group. Further, the emotional text design facilitated elaboration processes but suppressed metacognitive processes during learning. Learners’ emotional state was not affected by a positive emotional text design, but participants in the group with the negative emotional text design showed a worse emotional state after learning. Qualitative data showed that even though both emotional text designs facilitated learning, cognitive mechanisms for these effects differed between the groups. Results of the present study support the extension of the emotional design hypotheses with regard to textual parts of learning environments.