Knowledge management has recently hit on the metaphorical underpinnings of knowledge, as is attested by this special issue, but this discovery is in fact just another instantiation of the general cognitive--scientific interest in the metaphorical structure of many abstract concepts in a wide range of socio-cultural domains, including management and organization, education and science, politics and government and health and care (Ortony, 1979/1993; Gibbs, 2008). This general attention to metaphor in thought is partly due to the cognitive-linguistic postulation of conventional metaphorical structures in all cognition by Lakoff and Johnson (1980) in their well-known Metaphors we live by. These conventional metaphorical structures in cognition are called conceptual metaphors and have been studied in many different ways over the past 30 years (cf. Kovecses, 2002/2009; Lakoff, 2008). Metaphor 'in thought' is often modelled via this notion of conceptual metaphors, which can be defined as conventionalized and systematic mappings (sets of correspondences) between distinct conceptual domains; this includes seeing knowledge (or ideas or understanding) as food, movement or perception (Lakoff and Johnson, 1999). This is one important part of the general background which has led to the current special issue, as may be shown by the references in Morgan (2006), Andriessen (2006, 2008) and others.
However, a number of problems with the cognitive--linguistic approach to metaphor in thought and language have emerged over the past 30 years (Steen, 2007, 2008, in press). The following four problems are of particular relevance to metaphor in knowledge management. First, the problem of what counts as a metaphor in language (and other codes of expression such as visuals) has led to groundbreaking methodological work on linguistic metaphor identification. Criteria for metaphor identification were seldom explicit, often diverging between researchers and never tested for interanalyst agreement in order to examine the reliability of data analysis. Since researchers aim their claims to be open to criticism, comparison and validation, standardization was clearly needed. The Pragglejaz Group, an ensemble of ten international metaphor scholars, worked for seven years to produce a reliable procedure for metaphor identification (Pragglejaz Group, 2007), which has since been applied by many metaphor researchers working on metaphor in natural language use in all kinds of settings. The method has been further developed, improved and applied by Steen et al. (2010) in their linguistic analysis of a substantial amount of data from four registers of English and Dutch (academic texts, news texts, fiction and conversations).
The issue of linguistic metaphor identification naturally leads to a second problem: how can such linguistic expressions of metaphor be reliably related to the specific underlying conceptual metaphors that they are supposed to express? This problem is possibly even more intimidating than the one of linguistic metaphor identification: which metaphorical model in thought, exactly, is being used when people speak or write in particular metaphorical ways? For instance, when they use expressions like win, defend or lose in the context of argumentation, do they conceptualize argument as sports, fighting, war or yet another concept? Equally well-developed methods and techniques for addressing this type of question, about conceptual metaphor identification, are currently not available (cf. Steen, 2007). If knowledge management researchers aim to model the metaphorical nature of the conceptual structure of knowledge with social-scientific adequacy, for fundamental or applied purposes, they have to address this methodological problem in all its complexity.
A third issue has to do with the relation between this type of structural analysis of the meanings of linguistic forms and conceptual structures, on the one hand, and the way in which these linguistic forms and conceptual structures are processed and mentally represented in the psychology of individual language users in concrete settings, on the other. The fact that language data may be analysed, both linguistically and conceptually, in such a way as to point to specific metaphorical meanings and conceptual structures is one thing, but it does not automatically imply that these metaphorical semantic structures are in fact mentally processed as metaphorical representations by all or even most individual language users. Psycholinguists have formulated fundamental criticism of this conclusion and offered alternative models. For instance, language may be processed in a shallow way which selects the appropriate conventionalized, abstract, figurative senses of words like win, defend or lose without activating …