Current systems that use gestures to enable storytelling tend to mostly rely on a pre-scripted set of gestures or the use of manipulative gestures with respect to tangibles. Our research aims to inform the design of gesture recognition systems for storytelling with implications derived from a feature-based analysis of iconic gestures that occur during naturalistic oral storytelling. We collected story retellings of a collection of cartoon stimuli from 20 study participants, and a gesture analysis was performed on videos of the story retellings focusing on iconic gestures. Iconic gestures are a type of representational gesture that provides information about objects such as their shape, location, or movement. The form features of the iconic gestures were analyzed with respect to the concepts that they portrayed. Patterns between the two were identified and used to create recommendations for patterns in gesture form a system could be primed to recognize.