Evaluation — the function and usage of language to express the speaker’s or writer’s opinion — has only relatively recently become the object of systematic linguistic research, for example in stance or appraisal analysis. This paper proposes an alternative, corpus-based approach to evaluation which assumes that there are at least ten different meaning dimensions (parameters) along which speakers can evaluate aspects of the world. This framework helps to explain the complexity of evaluation, and in particular what is here called evaluative interplay or combination: the expression of more than two evaluative parameters at the same time, which can be realized by evaluative conflation (evaluations signalled by one and the same linguistic item) and evaluative collocation (evaluations signalled by different linguistic items). Both the parameter-based framework and evaluative interplay are illustrated with a number of examples from authentic discourse (mostly from British newspapers).