the test set including human-scored sentences will be available soon here
Distributed datasets contain sentences that have been collected adopting different strategies:
sentences deriving from naturalistic corpora, in particular extracted from different sections of the Italian Universal Dependency Treebank;
sentences that include specific morphosyntactic manipulations tested in independent psycholinguistic experiments;
sentences that have been procedurally generated with the aim of having represented within the corpus different levels of grammaticality/acceptability and complexity.