Halliday & Matthiessen (1999: 466):
… we have to keep in mind that we are dealing with a semantic system and not with a collection of unrelated items. … For example, if we recognise figures of saying … this goes together with certain other features: with the distinction between phenomena and metaphenomena, and between ideas and sayings, with the organisation of projections as sequences rather than figures, with the identification of symbol sources as a kind of participant, and with the recognition of circumstances of matter. In other words, we have to consider syndromes of features that occupy a region of semantic space.