Data & Analytics

Brown Student-Developed Module Speeds Archaeological Analysis

• By Dian Schaffhauser
• 09/29/16

To non-professionals, archaeology appears to be a tedious business. Not only is there the excruciating work of identifying and surveying sites, but there's the painstaking process of excavation and mind-numbing job of documenting the many objects and artifacts that have been pulled out of the ground. A team of participants at Brown University has undertaken a project to expedite archaeological laboratory research with the use of a customized digital recording system.

The research is being led by Parker VanValkenburgh, an assistant professor of anthropology. Last year, VanValkenburgh applied for a Brown U "Interdisciplinary Team Undergraduate Teaching and Research Award." The I-Team UTRA, as it's called on campus, is intended to increase research opportunities for first and second year students, encourage their intellectual development and growth through peer mentorship and group learning, and boost interdisciplinary scholarship among members of Brown's faculty.

VanValkenburgh's idea was to pull together a team of students who could take an existing open source Android-based digital recording system developed at an Australian university and modify it for site work he was involved in. The Federated Archaeological Information Management System (FAIMS) was developed at Macquarie University. He wanted a similar tool that could be used to record data from excavated artifacts in his field laboratory.

From three seasons of excavating reducciones, colonial planned towns, in Peru, VanValkenburgh and his research team had collected tens of thousands of small ceramic potsherds or "sherds." These tiny pieces of pottery, dating back hundreds of years, could provide insights on the lives and habits of indigenous populations — what they ate, how they organized their living spaces and how they were connected to the rest of the world — after being forcibly resettled into towns by Spanish colonialists.

"A single sherd on its own is often mute, but with 80,000 sherds you notice patterns and establish lines of evidence that enable you to tell really interesting stories," said VanValkenburgh, in a Brown article about the project.

Traditionally, performing the analysis of the pieces and recording that data requires researchers to search through a "codebook" that offers up to 40 specific descriptive labels for each artifact. Those details are then entered into a spreadsheet. A two-person team could typically enter data at the speed of between 40 and 100 items a day, suggested VanValkenburgh. He thought an automated system of artifact analysis could speed that up considerably.

As VanValkenburgh explained in his I-Team UTRA proposal, he wanted funding for a team of four students — one with experience in computer programming, another with good hand-drawing and digital design skills, a third with an interest in social sciences and human-computer interaction and a fourth with archaeology field experience to help with design and testing. All should have at least a working knowledge of Spanish since the recording module envisioned by VanValkenburgh would contain component labels in both Spanish and English. The proposal was accepted.

In a three-week summer flurry of activity on campus, the team designed, developed and refined what they called the Proyecto Arqueológico Zaña Colonial (PAZC). Through the use of drop-down menus and scrollable lists of illustrations and photographs showing representative artifact forms, decorations and technological characteristics, the use of codebooks could be eliminated and multiple students could work in parallel on tablets to identify sherds. Each tablet would sync to the project's server for data aggregation and import into CSV files.

The student team then headed to Peru to test out the ceramics module. The larger research team started the season on the old system and then made the move to the new module for the sake of comparison. According to VanValkenburgh, those users found the tablet-based approach "to be universally more engaging than the conventional system we'd been using." Eventually, the research team deployed seven tablets in the lab there, allowing users to analyze 30,000 sherds over a six-week period.

One member of the team, linguistics and classics student Chiara Repetti-Ludlow, said that an anonymous survey taken of the larger team found that "the system was overall hugely successful in terms of speeding up classification, helping with accuracy, clarifying the classification process and maintaining interest."

Now that they're back on campus in Rhode Island, the student team is working with VanValkenburgh to finalize data analysis and draft a paper that they hope to submit to a peer-reviewed journal for publication.

Said VanValkenburgh, "We want to share not just the discussion but the code itself, and maybe make it open source. It's important to be generous and magnify the impact of the work you do."