Progress and tensions in modelling register as a semantic configuration
Halliday (1978: 111) defines register as "the configuration of semantic resources that the member of a culture typically associates with a situation type." Elsewhere, however, he stresses that when we talk of "a register" this is a term of convenience: register is more properly theorised as continuous variation along many linguistic dimensions. In this paper I review progress in our capacity to describe register and context of situation and ask whether the tension between discrete and continuous models of register might hinder such progress. I then consider Hasan's (1983, 2013) contextually-open networked model of message semantics, arguing that in conjunction with context networks it has potential to map register variation but still needs to be tested across a large and varied set of domains. Examples from healthcare interaction ground the discussion.