in general, any breaks in the tab-input format before the first token of a document can probably be ignored: document boundaries ("file breaks") should always coincide with every other break-type (every doc boundary is also a sentence- and paragraph-boundary): we can insert these implicitly (i.e. at beginning- or end-of-document) and not require the breaks to be explicitly listed in the tab-files to be imported. They should however be allowed in the tab-files to be imported, even before the first token of a document (or after the last token).