Why it is a question
THEORY-1153 reports the PETra corpus (LIT-1798, read in NOTE-1344): among additions a named-entity and null-alignment filter and a classifier trained on its finds turned up, system transfers, chiefly unit conversions, and entity changes are most of the labels. The filter requires a named entity, so entity-linked additions were bound to be common, three quarters of the pairs were found by the classifier, and no agreement is reported. How often translators add material for their audience, and of which kinds, outside that net is not known.
What would answer it
An annotation of a random sample of English-source human translations, drawn without a named-entity or alignment filter, by at least two annotators working from the same guidelines with their agreement per category reported, that gives per-language rates of explicitation per sentence and says whether measurement and entity explicitations are the most frequent types. Rates within PETra’s own filtered pool, or classifier counts, do not settle it.