Brown corpus divergences

We should examine and validate where the copy of SemCor here diverges from Brown.

There should be a script that loads Brown from NLTK and verifies each sentence looking for any divergences in the text (excluding spaces and underscores). The script should produce a report in Markdown.

Start by locating the SemCor corpus copy and the project’s existing Python entry points, then inspect how NLTK exposes the Brown corpus. Compare each sentence while ignoring spaces and underscores, and produce a Markdown report listing divergences. Done means the script runs against Brown and reports the validation results reproducibly.

Written by the indexing model from the issue text.