Product designers seek to create novel designs that elicit aesthetic responses from consumers. In this paper, we re-visit the much investigated relationship between novelty and aesthetic preference, but with a new focus on what it is that constitutes novelty in product design. Based on prior research on consumer perception of product appearance, we included in our study three fundamental dimensions of product semantics - trendiness, complexity, and emotion - and investigated their influences on novelty and aesthetic preference. For the study, we selected as stimuli 88 chairs, all highly varied in shape but corresponding to a common prototype: a chair with back support and without arms. We then conducted a semantic differential study on the 88 chairs, in which we measured trendiness, complexity, and emotion by using three sets of bipolar adjectives ("traditional-modern," "simple-complex," and "rational-emotional," respectively), and evaluated novelty and aesthetic preference with an additional set of bipolar adjectives ("typical-unique" and "beautiful-ugly," respectively.) The results confirmed Berlyne's hypothesis that the relationship between novelty and aesthetic preference resembles an inverted-U curve, in which the chairs perceived to be most beautiful were those with a moderate level of novelty. Each of the three dimensions -trendiness, complexity, and emotion - formed a positive linear relationship with novelty. These results show that the three fundamental dimensions of product semantics can be regarded as predictor variables for novelty; of the three dimensions, trendiness was shown to have the greatest influence, followed by complexity, and last by emotion. The three dimensions influenced aesthetic preference differently: Both complexity and emotion exhibited inverted-U relationships with aesthetic preference, while trendiness had a small positive linear relationship with aesthetic preference.