This research project aims at analysing the cognitive complexity of compound figures of speech, that is, complex non-literal expressions which serve more than one purpose. Specifically, my research will concentrate on the interaction between metaphor, irony and hyperbole (i.e. ironic metaphor, hyperbolic metaphor or hyperbolic irony). The research questions concern the type of meanings conveyed and the logical, psychological and temporal order of interpretation: is an ironic metaphor first interpreted as a metaphor and then as irony, for instance, or is one interpretation conditioned on the other, and if so, how? The hypothesis is that the actual interpretation of these multiple figure uses is context-dependent, both in terms of contextual attractors and contextual salience. The theoretical claim, however, is that individual tropes are not represented in the mind as rigid categories but rather as multidimensional spaces including flexible sets of factors that may group together differently on different occasions to let one or the other interpretation emerge. The complexity of compound figures is analysed empirically in a Paraphrase Test. A total of 155 participants took part in the experiment. Results showed that stimuli containing simple figures of speech recorded higher accuracy scores and lower response times than short texts with compounds. There were, however, some exceptions, as some compound stimuli recorded better results than simple ones with an equal complexity score. This led to the conclusion that several other factors, which will likely be analysed in future studies, can influence participants’ interpretation of compounds and figurative language in general.