Researchers Make 3D Printed Models of Endangered Languages

3D modeling is conceived as a way to demonstrate the complexity of systems, and accessibly showcase intricacy and fine details. While 3D models might bring to mind surgical planning or engineered parts, today we bring you a project from 2022 which uses the technology to map language. Alex Pillen, PhD researcher at the School at Antholopology at UCL, collaborated with Emma-Kate Matthews from the architectural school of the same university. They produced a range of models based on four languages: Kurdish, Tarania, (an Amazonian language), an ancient Mesopotamian dialect called Akkadian and American English.

The researchers wanted to model a grammatical aspect of these languages: specifically, evidentiality. This means whether evidence exists for a statement, and if so what kind; in linguistics it is the method by which evidence is conveyed. In certain languages, speakers express evidentiality through grammar. To take the example of Tarania, the speaker expresses four types of evidence using different grammar structures. These evidence types are visual, non-visual, inferred, reported, and assumed from common knowledge. American English was chosen to compare to the other languages as a reference, because in English evidentiality is not expressed through grammar and so the models were not like the others.

Co-creators of the 3D modeling language project, Alex Pillen (left) and Emma-Kate Matthews (right), with their designs

The team relied firstly on a established theoretical link between grammar and geometry which dates back as far as the 17th century. For the practical side of the project, they used Grasshopper and Rhino3D to map specific language features onto Cartesian system of coordinates. Firstly, they manually formatted the transcript was Excel, and gave each evidential a numerical weighting. The x-axis depicts the timeline of speech as one phrase succeeds another, with the y-axis indicating the number of syllables in each line of a transcript. The x- and y-axis can thereby represent the elementary structure of language reflected in a transcript. The z-axis is the indication of evidential grammar within the sample; these markers take on a ‘virtual weight’ which gives the model its 3D “wireframe” appearance.