What are the criteria for deciding when a new lexical entity must be added to the grammar? For instance, given the sentence
We tested our findings by comparing acceleration response spectra of recent and historical intraslab and megathrust earthquakes along a longitudinal profile.
ERG produce the predicates
_megathrust/JJ_u_unknown (also sometimes
_megathrust/NN_u_unknown). The term
intraslab is considered an adjective by https://en.wiktionary.org/wiki/intraslab, but despite that, ERG was able to produce a good analysis for the sentence considering, as far as I understood, the whole fragment
recent and historical intraslab and megathrust earthquakes as a compound. So my personal belief would be that despite unknown words beeing correctly guessed, it would be better to always complete the lexicon, is that the right approach?