Comparative Analysis of Collocations and Semantic Preference of “Risk” in Risk Management Reports from Selected Malaysian and American Banks

Abstract

Banks rely on risk management reports (RMRs) to detect, analyse, and manage potential threats to their operations. These reports provide insights into a bank's exposure to risk and support informed decision-making to minimise losses. Previous corpus studies have shown that the word risk is often perceived negatively. This study investigates how risk is represented in RMRs, which are intended to assure stakeholders of banks’ ability to manage risks effectively. A self-compiled corpus comprising RMRs from Malaysian and American banks was utilised in this study. Collocational patterns were examined using the Graph Coll feature in Lancs box 6.0, and collocate strength was measured through Mutual Information (MI) scores. Semantic preference was further analysed using UCREL Semantic Analysis System (USAS). The findings reveal that although the collocates of risk in both corpora differ in frequency and MI scores across the two corpora, ency the word is consistenly paired with domain-specific terms denoting risk types. Additionally, risk in Malaysian RMRs exhibits a broader range of semantic associations than in American reports. The study enhances understanding of how the concept of risk is linguistically represented across banking contexts and provides insights into the persuasive construction of expertise in financial reporting.