The order of “B yíyàng de A” and “B yìbān de A” is exactly the opposite of the typical Mandarin simile structure “A xiàng B”. More importantly, ‘A’ in the simile structure “B yíyàng de A” and “B yìbān de A” can also be ‘ground’ that expresses the similarity between the vehicle and topic sometimes. Therefore, the simile structures composed of yíyàng and yìbān deserve further investigation.

This thesis adopts a corpus-based approach as our methodology, using “B yíyàng de A” and “B yìbān de A” as the simile structures, and uses the COCT Written Corpus 2017 as our research tool. The thesis aims not only to investigate the differences between yíyàng and yìbān through the characteristics of vehicle and topic, but also to focus on the linguistic representations of ‘ground’ in these structures. The findings show that more than half of the results retrieved by yíyàng and yìbān are simile structures involving cross-domain comparisons, especially with nearly 90% of the results for yìbān. Furthermore, under the property attribution theory, the attributes of vehicles in simile structure composed of yíyàng are more concrete, while the attributes of vehicles in the simile structure composed of yìbān are more abstract, indicating that the simile structures composed of yìbān are more metaphorical than yíyàng.

This study also finds that the simile structures retrieved by “B yíyàng de A” and “B yìbān de A” can be divided into two categories: In the first structure, B is vehicle and A is topic; in the second structure, B is vehicle and A is ground. The analysis shows that even when there are occurrences of ‘grounds’ in the second type of simile structure, additional description of the grounds would also be displayed. In addition to the distribution of the corresponding ‘grounds’ in the two types of simile structures under different attributes of vehicles, this thesis also analyzes and discusses the linguistic representations of the ‘grounds. The results reveal that the linguistic representations of the ‘grounds’ differ according to the types of the ‘vehicles’ when the ‘topics’ are ‘concrete objects’. Moreover, we also find that in the case of simile structures without ‘grounds’, the images of the corresponding ‘vehicles’ are mostly ‘common senses’ in our minds; therefore, people can understand the meanings of these simile structures without the representations of the ‘grounds’ or the interactions of contexts.