Observation. Let 1 denote the language with one relation symbol called ∼1. Then I claim that for any finite language L and L-theory T, exists a 1-theory S and injection r:WFF(L)→WFF(1) such that
T⊨ϕ⟺S⊨r(ϕ)[1]
It’s worth thinking first about whether or not this is even an interesting statement. Since L and 1 are both finite, then WFF(L) and WFF(1) are both countable and hence isomorphic in Cat. For some mathematical structures, an isomorphism of underlying sets can be lifted to an isomorphism of structures: if you give me a group (G,×,e) and a set G′≅G, then I can build a group (G′,×′,e′)≅(G,×e). Following this intuition, it would seem that [1] is trivial: given that WFF(L)≅WFF(1), we should be able to lift the isomorphism to theories.
But I think it is not so, because we do not have control over the rules of logic. Note that T={∃a} and S={∀a:a∼1a} have set-isomorphic collections of formulas, but are in no way logically related.
Proof. Take L, T.
Let L1 be L but each constant symbol c∈L is replaced with a function symbol fc∈L1. Let T1 be an L1-theory constructed as follows. Take T, then for each ϕ∈T remove it and in its place include ϕ′ defined as follows. Let v be a tuple of variables not present in ϕ; define ϕ′=∃v:ψ(v) where ψ is ϕ but with each constant symbol c replaced by a variable in v. Two occurences of the same constant symbol recieve the same variable, but different constant symbols receive different variables.
https://math.stackexchange.com/questions/2261042/do-we-really-need-constant-symbols-in-first-order-theories
https://math.stackexchange.com/questions/2545007/the-constants-of-a-structure-model-theory
https://math.stackexchange.com/questions/1072035/adding-constant-symbols-in-model-theory?rq=1
https://math.stackexchange.com/questions/1147957/if-you-create-a-new-theory-of-a-model-by-augmenting-the-language-with-constants?rq=1