Complexity of nominal group adjectivization: a self-information approach

Abstract Nominal group adjectivization has been extensively investigated in relation to its distribution across registers relying on the normalized frequency of its occurrence. This frequency-based approach does not consider the repetitive use of a specific instance, thus assuming that a text with more nominal group adjectivizations is of more complexity than the one with less such cases. Although the entropy-based approach takes the repetitive use of lexical items into consideration, the entropy value counters our expectation that a more frequent instance of nominal group adjectivization conveys a smaller entropy value, because the linguistic items around such an instance will definitely mitigate the complexity that this instance conveys. Against this backdrop, this research proposed the employment of the self-information value as an index to measure the complexity that nominal group adjectivization conveys. It is uncovered that the self-information value efficiently measures the complexity of nominal group adjectivization not only in different sentences but also in various types of texts. The significance of this research lies in that it is possible to quantitatively analyze the complexity of nominal group adjectivization, which will further facilitate the quantitative investigation of grammatical metaphor, a superordinate term of nominal group adjectivization, in systemic functional linguistics.