Curtains for Zeus: Mapping a multi-media offering

Annotating the surviving text of the intrepid 2nd century traveler Pausanias is a delicate task: weighing 223,000-odd words in ancient Greek and tagging people, places, events, and their relationships. The overarching goal is using the new and improved Recogito tool to generate from Pausanias a pile of open-access data that any user can use to pry loose fresh insights into ancient Greek society, religion, history, and culture. The fun part, however, is mapping Pausanias’ travels in southern and central Greece as he described them. A temple of Zeus is easy, a dot on the map (soon to be a polygon). Tag it “Paus” to say Pausanias was present at this part of his narration. String those dots together in the right way and we have Pausanias’ travels. But not every place name is a location we should map as if Pausanias were necessarily ever there. The statue of the god is made of stone from Paros or Mt. Pentele, the columns of stone from Phrygia or Libya or the quarries at Krokeai south of Sparta. Or simple “local” (ἐπιχώριον) stone, every instance of the term a different blur on the map. A statue can be, Pausanias opines, of Aiginetan workmanship (ἐργασία), whatever that means. A tune is in the Lydian mode. And let’s leave out for now the question whether a mythical but rapidly moving and dangerous Calydonian boar is usefully represented via the dot on the map our Pleiades/DARE-based gazetteer calls ancient Calydon.

An annotator’s existential crisis of the evening: At the great temple of Zeus at Olympia there is a curtain (object, parapetasma) decorated with Assyrian weaving (ὑφάσμασιν Ἀσσυρίοις) and dyed with Phoenician purple (βαφῇ πορφύρας τῆς Φοινίκων). This curtain, an offering to Zeus by King Antiochos, was (when Pausanias saw it anyway) an object with a location, even vaguely mappable at the front of the temple. Do we map it as three dots: at the temple in Olympia, but also in Assyria (presumably a high-flown Romanism for Syria) and Phoenicia?

We huddled and decided it was time to deploy a new semantic tag, “material,” and Recogito’s relationship annotator that we are all still secretly afraid of.

Thus, four distinct annotations:

                1. The Temple of Zeus: built, naos, Olympian Zeus, Paus, Place: Zeus temple (Olympia)

                2. the curtain itself: object, parapetasma, Olympian Zeus, Paus, Place: ungazetteered;

                3. the weaving: material, ufasma, Place:Syria;

                4. the dye: material, bafê, Place:Phoinike

Then, three relationship annotations:

                2 to 1 “contains”

                3 to 2 “provenance”

                4 to 2 “provenance”

These annotations will result in a relationship tree that shows the temple of Zeus, with the curtain one of its long list of offerings, and Syria and Phoenicia as two of the regions included in Olympia’s religious universe. We are cautiously accumulating enough semantic concepts to do justice to Pausanias’ complex world but not so many as to paralyze our annotation process with delicate ontological/theological distinctions.

Relational annotation in Recogito showing the relationships of materials to a monument

A typology for annotating the Periegesis

As we have set out in our previous posts, our aim in the Digital Periegesis is to identify and explore the spatial form of, and the forms of space within, Pausanias’s narrative. The challenge of analyzing spatial representation in the Description of Greece—and the reason why we were attracted to the text in the first place—is the “thickness” of that description, whether Pausanias is taking the reader on a tour of a temple precinct, stopping off along a road to take note of a statue, or recalling the mythical stories associated with a simple looking rock.

In order to capture sufficient information about those places (or objects in space), it is first helpful to establish a set of concepts and categories when annotating them in Recogito: i.e., how they function within his narrative, and, in turn, how his narrative is constructed through spatial description. With a view to building a method that can help us annotate in a systematic and uniform manner, we have so far developed the following semantic annotation typology based on the entity and tagging feature in Recogito:

1. Entities

Recogito provides three choices of entity: place, people, event. Our primary concern is place: when we identify a place in the text, we mark it and align it to an appropriate gazetteer place entry (if we can find one). However, it is also important to identify people in the text, especially for their role in certain places (or even as proxies for place): for this we use the “people” entity (and both the “place” and “people” entity if considered to be representing a place).

Figure 1: annotating a “place” in Pausanias, by identifying the character string as an entity and then aligning it to a global authority record (the gazetteer DARE or Pleiades).

2. Tags

Recogito also provides a “free” tagging features, which enables users to provide more information about those entities and construct their own schema for labelling them. For instance, for places, we want to identify: is the place human, physical, regional, or mythical; and what type of place is it? (e.g. human place might be a settlement, temple, assembly.) For people: are they mythical or historical; divine or mortal; male or female; Greek or other? Or are they a proxy for a place?

Figure 2: Annotating a “person” in Recogito, by first identifying the character string as an entity and then using tags to further define it.

3. Relations

Fundamentally, we are interested in capturing the ways in which Pausanias constructs his description of Greece. There are various different kinds of spatial relationships that can be defined in the text, as Pausanias moves through both space and time. So far we have determined the following:

  • Topographic: a place in space, as Pausanias moves through the landscape,
  • Chronotopic: a place in time, as Pausanias moves through the history of a particular place/building/statue, or
  • Analogic: places compared, as Pausanias relates one place to another in a different part of the world.

We use the “event” entity to highlight the sections of the text in which either or these three descriptive modes take place, and use the tagging feature to then specify the mode (topographic, chronotopic, or analogic). We then use an additional relational tagging feature, which is part of the Recogito UI, to further define those relations: e.g. are the topographic relations being described synoptic (a bird’s eye view) or hodologic (movement through space)? A further tag can be used to define focalisation–whether the description is from the narrator’s viewpoint or the perspective of another.

Figure 3: annotating relations in Recogito, using the “event” entity to mark the entire clause of the upper-level spatial category (topographic, chronotopic, or analogic), and then the “relational tagging” feature to mark the individual entities and their precise relationship to each other.

This represents our methodology for annotating the Periegesis at the moment. But, as we have already found, we are modifying and nuancing this typology as we move through Pausanias’s “thick” description. You may also have your thoughts about what to capture and how: we welcome your feedback!

For our latest methodological account, see: