Parsing in context

Token, lemma and role

Copy the token as printed in the edition. Identify possible lemmata and morphological analyses, keeping each as a separate proposal. A lemma is a dictionary filing form; it is not necessarily a spelling present on the object. Test candidates against agreement, government, word order, formulae and the surrounding passage. Translate only after this test so that an attractive English phrase does not decide the parse in advance.

A single ending can be syncretic, and a damaged sign can create additional possibilities. Mark which ambiguity belongs to the writing and which belongs to grammar. A parser or glossary can suggest analyses, but its output depends on its lexicon, normalization rules and editorial input.

A decision record

For a disputed token, make four short lines: candidate A and its evidence; candidate B and its evidence; evidence against each; remaining uncertainty. Note whether a comparison is from the same date and genre. If a parallel passage is restored at the relevant word, it is weaker support than an intact attestation.

The result can be a bounded conclusion: “both parses remain possible,” “syntax favours A,” or “the image could change the reading.” Another reader can test such a conclusion.

Practice

In an Oracc text linked from the ATF Primer, pick one token for which the project glossary supplies a lemma. Compare token, glossary headword and surrounding line. Write the reasoning that joins them; if the glossary only supplies a guide translation, do not treat that gloss as a complete parse.