The search on Files works the same way as the search on Items and Collections – it generally just looks at whatever has been entered in the metadata fields. Files can have their own metadata, distinct from the Item itself, but fairly few sites use that feature. Thus, in most cases, File search doesn’t produce much of a difference.
The PDF to Text plugin extracts the text data and stores it in a metadata field, which makes that content available to the search. It only works on PDF files, though, not Word documents.