On the Semantic Relations and Functional Properties of Noun-Noun Compounds in Mandarin

摘要

The goal of this research was to determine the distribution of compounds in Mandarin Chinese from the aspect of semantics. In particular, the focus was on two types of compounds: compounds interpreted by semantic relations or by functional properties between constituents. We collected 880 compounds from a dictionary and categorized them into two types of noun-noun compounds in Mandarin, including relation-based compounds (e.g., 中國菜 zhongguocai "Chinese food") and functional property-based compounds (e.g.,柳葉眉 liuyemei "arched eyebrows"). Finally, the frequency of occurrence of the two types of compounds was determined. The results showed that relation-based compounds occurred much more frequently than property-based compounds in our data (96.1% vs. 3.8%). In addition, it was found that within the relation-based compounds, noun-noun compounds using the FOR relation (e.g.,信紙xinzhi "letter paper") had the highest rates of occurrence (37.6%), while the CAUSE relation (e.g.,刀傷daoshang "wounds by a knife") had the lowest rates of occurrence (0.7%). On the other hand, the functional property-based compounds, almost always referring to objects in the NATURAL KIND domain, take metaphorical meanings from individual constituents. Our study suggests that the relation-based meanings for interpreting compounds are common in daily conversation, which could be the dominant strategy people use to interpret novel compounds. This research has practical implications for natural language processing in dealing with segmentation of compounds and multiword expressions in Mandarin and even recognition of novel word combinations.