Performative arts and choreographed dancing motion are powerful medium for nonverbal emotional communication, allowing humans to convey affective states capturing and enchanting a wide audience. The integration of robotics into this domain, using AI, motion capture, and movement-mapping algorithms, offers new opportunities to investigate how artificial agents can express emotions. However, this raises important questions about the authenticity and fidelity of robotic emotional expression. The challenge becomes particularly relevant with non-humanoid robots, where the structural disparity with the human body complicates the mapping of expressive gestures. In prior work, we introduced a PCA-based projection method to transfer human dancing movements to robotic arms. This technique adapts the high degrees of freedom of human motion to the robot more limited kinematic structure, aiming to preserve dance nuance. While earlier results showed structural coherence between human and robot motion, the present study focuses on emotional fidelity. We examine whether the embedded space generated by PCA preserves perceptual distinctions between emotions or introduces distortions, particularly in the intensity (arousal) or pleasantness (valence) dimensions. Our analysis consists of two steps: first, we assess the clarity of emotional differentiation post-mapping; second, we analyze whether robot movements maintain the expressive signature of each emotion. This approach sheds light on how dimensionality reduction and mapping complexity influence emotional authenticity in robotic dance.