Grammatical Entries

Perseus 6.0 illustrates how we can link particular words in a text to a traditional grammar (which contains explicit language-specific information not in the Perseus treebank). See this in Perseus 6.0 here.

Zooming in calls up brief descriptions with a handful of examples from Smyth and cross references to Smyth, to the Pedalion Grammar (developed in Leuven) and to the Cambridge Grammar of Classical Greek.

We cannot, of course, quote the Cambridge Grammar extensively because of copyright. All of Smyth is, and has for years been, available online but there were two reasons not to rely primarily on Smyth. First, Farnoosh wanted students to be working from a more compact base — 15 varieties of dative are, for example, more than enough, at least to start with.

The second reason is, however, particularly compelling. Farnoosh teaches Ancient Greek in Persian and her dissertation research focuses on how to do so using techniques such as translation alignment as a learning exercise. There are few Persian language resources about ancient Greek — there are also few direct translations from Greek into Persian, even though Greek historians such as Herodotus and Xenophon provide irreplaceable information about ancient Persia and even though Plato and Aristotle are seen as foundational thinkers within Persian culture. Farnoosh needs to be able both to translate Didakta into Persian and to localize it, editing its content to reflect the particular challenges that Persian speakers face when learning Greek (e.g., the fact that Persian has no participles).

The point is not simply to make Ancient Greek more accessible to speakers of Persian but, in so doing, to develop methods by which to localize knowledge about Greek in other languages in which little or no information about this language may be available. More generally still, we are feeling our way forwards towards representing more and more information about historical languages such as Ancient Greek in a machine actionable form that is in turn linked to the actual corpora of these languages.

Background

Didakta is not the first attempt to add a layer of grammatical annotation that goes beyond the treebank annotation scheme. More than one careful annotator has found the need to add a layer of this type. Matthew Harrington of Tufts University created an extended treebank tagset so that his student annotators could encode more fine-grained distinctions: e.g., D-Inter, D-Poss, D-Refer for the dative of interest, the dative of possession, and the dative of reference. Giuseppe Celano left the treebanking annotations untouched but added what he called an “advanced syntactic layer / semantic layer” based on Smyth’s Greek Grammar.