Proof-nets as Logical Forms for LFG

Although ‘glue semantics’ appears to be generally accepted as LFG’s present account of semantic composition, its uptake amongst LFG practitioners doesn’t actually seem to be that extensive (and a number of well-known syntacticians have claimed to me that they are pretty much baffled by it). A possible reason for this is that the standard presentation of glue is based on term-assigning deductions in linear logic, a kind of apparatus that isn’t very congruent with normal LFG mechanisms. Here I will show how proof-nets, an alternative format for deductions in certain forms of linear logic, can be construed as a linguistic level of ‘logical forms’, related to the other levels in LFG by correspondence relations in a manner reminiscent of the c-structure-f-structure correspondence φ, as discussed for example in Kaplan (1995). Perhaps presenting glue in a manner more characteristic of linguistics in general and LFG in particular will make it easier to grasp, especially for LFG syntacticians.