Sensitivity of statement-level annotation syntax (feature set vs. forgivefulness)

[chrfrantz]

How sensitive should the IG Parser be to syntax when it comes to annotations?

Using the stylized example, the following coding expresses a statement-level annotation [annotation] that is associated with the leading institutional statement (i.e., A(regulators) ...; not the condition).

[annotation] {Once [left] E(policy) F(comes into force) [AND] [right] E(policyRight)} A(regulators) D(must) I(act).

Coded without the separating whitespace between annotation and condition (i.e., annotation immediately preceding the brace), should this be interpreted as an annotation on the condition, or the statement (leading statement)?

[annotation]{Once [left] E(policy) F(comes into force) [AND] [right] E(policyRight)} A(regulators) D(must) I(act).

Note that this second form is currently not explicitly considered, and simply ignored in the output (i.e., the annotation would not appear), pending this design decision. However, its benefit would be to allow a nuanced annotation of a combination, combining branch-level annotations ([left] and [right] in this example) and a annotation that is shared amongst both ([annotation]).

The proposed adjustment is a design decision trading off feature richness (shared annotation on nested component combination vs. statement-level annotation) and sensitivity to syntax (i.e., error-minimization in encoding).

Any comment is welcome.

[chrfrantz]

Upon discussion with Ute: Approach 2 (i.e., [annotation]{[left] [XOR] [right]} A(actor) I(act)) might be useful to be treated as different from [annotation] {[left] [XOR] [right]} to retain nuanced ability to variably annotate statement entire, nested component, branch of statement combination and component (i.e., four levels).
Idea: Highlight whitespaces in the parser to explicitly visualize gaps between annotations and boundaries (i.e., brackets, braces, etc.).