Human and Robot Narrator: How Gestures Affect Comprehension and Recollection

Abstract: This thesis investigated if a human or a robot narrator affected listeners’ comprehension and memory recollection similarly in a narrative story. 103 participants (MeanAge = 35) were recruited to an online experiment to investigate if gestures affected the participants’ narrative comprehension and recollection more than a narrator that did not produce gestures. Participants were presented with the same narrative setting in four different conditions; either with a human or a robot narrator, with or without gestures. Participants answered questions in a questionnaire, and in total, there were 21 questions regarding the narrative, of these, 13 questions were directly related to a gesture, and 8 questions were related to the narrative without any accompanying gesture. The gestures produced were congruent redundant iconic and deictic gestures. Participants were graded and analyses were conducted on three dependable variables; total sum, the sum of the gesture-related questions, and the sum of no-gesturerelated questions. The total sum was used to test the hypotheses. The sum of gesture-related and no-gesture-related questions were analyzed to see if gestures affect comprehension and recollection of specific and/or general information. A second experiment was also conducted on the same participants as in the first experiment to see how participants perceived the human and robot gestures. Most participants perceived the gestures produced by the narrator as similar to each other. The results found in the experiment of this thesis showed a small indication that participants who viewed the robot narrator, both in the condition with gestures and the condition without gestures, had a stronger benefit in comprehension and recollection compared to the human narrator conditions. Similar indications were found in the gesture-related questions and a strong indication were found in the no-gesture related questions. The credible intervals overlapped between the human and robot narrator conditions, meaning that the conditions with gestures scored similar to each other, regardless of the narrator, and the same applies to the conditions without gestures. The gesture conditions scored significantly higher compared to the conditions without gestures, showing that human and robot narrators producing gestures affected the listener’s compression and recollection in a similar way.