Acoustic packaging, intersensory redundancy, motherese and motionese are the main concepts, which describe how acoustic features and body movements of caretakers are

adapted intuitively in adult-child-interaction. This thesis will approach the question,whether feature characteristics highlight certain parts of child-directed actions. The main part focuses on the statistical examination of four features, Eye-Gaze, Intonation,

Word, Velocity of Hand Motion of child-directed demonstrations. The auditory and visual

signals of the recorded toy demonstrations were divided into meaningful segments for this examination. Then the composition of each feature and its values (fc) in each

segment is analysed. A conditional probability model P(Segment_x|fc_j) is computed to answer the question: in which segment (Segment_x) is, an already appeared feature

characteristic fc_j , likely to occur? Patterns of high conditional probability values have been identified for distinct feature characteristics and provide a basis for a future automatic classification.