A generative study of Latin word order allows for the assumption of a basic syntactic
structure generated by the interaction of universal principles and language-specific
parameters, which can be further extended by movements triggered by discourse
(semantic and pragmatic) features. A corpus-based investigation supports
the validity of this method, confirming the predictions spelled out in a qualitative
analysis from the quantitative perspective. We give an example of this integrated
qualitative-quantitative approach in the area of nominal expressions, showing
that the basic orders are overwhelmingly attested, while the other orders are less
frequently attested as they are “costly” in a theory-internal scale.