Robert Gorman
Those attending past panels sponsored by the DCA have been introduced to the Ancient Greek and Latin Treebanks and the digital tools that support them. These databases have been discussed primarily for their value for research. But their implications for language pedagogy have also been touched on, and with good reason. The technology of treebanking can be a valuable classroom resource throughout all stages of the Greek and Latin curriculum. I propose to discuss treebanking as a teaching method and illustrate how it can be integrated into language instruction.
I have been incorporating treebanking into our Latin sequence for over 5 years at all levels of instruction from the first week of Latin 1 to our most advanced classes. In my presentation, I intend to discuss aspects of treebanking at all these stages, but I will concentrate on how the technology works in the first-year classroom.
I will begin with a few words about the nature of dependency syntax and the characteristics of a dependency graph. Then I will introduce the Perseids Collaborative Platform and the Arethusa Annotation Framework. I will focus in particular on the features of most value to teachers. The system offers tremendous flexibility, rather than forcing a single approach to language onto its users. This quality is apparent, for example, in its handling of syntax labels. The labels used by research scholars are not well suited to the classroom, where the paramount concern is that text book and technology should agree. The Perseids/Arethusa system makes such agreement easy, since it supports user-defined tag sets. I will demonstrate how these labels are set up. Likewise, it allows great latitude in handling morphology. For beginning-level students, one may turn off Arethusa’s morphological analysis, so that the students must supply the lemma, part-of-speech, and form for each word. At the intermediate level, one may set the parser to offer choices of possible forms for the students to select. My advanced students have used the system several times in lieu of a textbook and lexicon.
I will also demonstrate how the student interface allows one to grade student homework automatically against a gold-standard template. For each word, data from a wide range of student choices (both correct and erroneous) are recorded and displayed. I will close by touching upon the efforts currently underway to determine a rating of sentence difficulty and make these data comparable among different college language programs.