The multidimensional analysis (MDA) approach was originally developed by Biber to compare written and spoken registers in English, but it has since been applied extensively in synchronic and diachronic analysis of registers in English as well as non-Western languages. While the WordSmith-style keyword analysis represents a quick and simple means of evaluating a genre against the MDA dimensions, the keyword approach nevertheless provides a less comprehensive contrast of genres and may not work for more fine-grained types of genre analysis. However, Biber's MDA model has so far been confined largely to grammatical categories, though Biber and others have started to incorporate semantic categories of some word classes in the model. In this paper, we will seek to enhance the MDA framework with semantic components and also introduce this enhanced MDA model, for the first time, in the research of world Englishes by exploring variation across twelve registers and five varieties of English in the International Corpus of English (ICE), which are annotated with grammatical and semantic categories.