Vanuatu and the Museum der Kulturen Basel (MKB) share a common history. The project aims not only to revive these relationships, but to enhance them. The collaboration with the "Digitales Schaudepot" (University of Basel) aims to make the collection held in Basel and its documentation easily accessible. By creating an interactive platform, we hope to establish a foundation for lasting connections and future collaborations.

The project is a digital inventory of the photographs and objects that the Swiss ethnologist Felix Speiser brought back to Basel from his 1910-1912 trip to Vanuatu. This project aims to use a digital platform not only to facilitate access to this collection, which is now housed at the Museum der Kulturen Basel (MKB), but also to enable a multi-perspective and translocal exchange of information. A key objective is to make Felix Speiser's Vanuatu collection and its documentation transparent and to initiate a dialog about it. In doing so, the authority to interpret the photographs and objects on display is consciously handed over.

With the help of the Traditional Knowledge (TK) labels in collaboration with Local Context, the communities can determine the use and availability of the collection. Gaps and errors in the documentation can be discussed via a comment function or added directly in the corresponding field. The interactive platform invites visitors to explore the extensive and diverse collection. Using a map of Vanuatu, visitors can navigate through 83 islands and discover the collection. The platform also offers the opportunity to take a closer look at individual objects and gain new perspectives through entries by other users.

Between May 4, 1910 and July 1, 1912, Felix Speiser collected over 3,000 ethnographic objects and took around 1,500 glass plate photographs. Field diaries and notebooks from this period have been preserved. Numerous publications were published as a result of his journey. During his time in Vanuatu, Speiser never stayed in one place for more than three weeks and spent most of his time in missionary and trading stations. In addition to collecting, Speiser's research utilised other contemporary scientific practices such as anthropological measurements, anthropometric photography and the appropriation of human remains. The presence of missionaries, colonial officials, plantation owners and indigenous guides, who served as informants, prevented him from gaining the trust of Ni-Vanuatu and learning indigenous languages. As a result, the ethnologist had to rely on observations and the treatises of European authors.

Visitors should be aware that this website holds historical data created in colonial circumstances, which may include vocabulary and views that today might be considered obsolete, inappropriate, and/or racist.

The collaboration envisaged in this project makes an English translation of the research materials and the collection documentation—originally written in German and held in the Museum der Kulturen Basel—indispensable. The English translation available on the platform has been entirely machine-generated using DeepL's neural translation technology: it has not been manually reviewed or corrected by human translators.
The purpose of such a translation is solely to provide an accessibility tool assisting you in exploring and understanding information originally written in German; they are not authoritative translations. For any scholarly research, publication, or official purposes, please always refer to the original German text as the definitive source. Think of these English translations as a helpful starting point for discovery.

More detailed information about the translation process

We developed a systematic three-phase approach to make this vast German-language collection accessible in English too.

Firstly, in the preparation phase, we used Python scripts to analyze the original JSON dataset of 195,000+ records, running character encoding validation to identify and fix common issues with German umlauts (ä, ö, ü, ß) that often get corrupted during data migration. We then extracted 60 representative German-language descriptions for testing.

Secondly, in the translation phase, we built automated Python scripts that recursively traverse the JSON structure, identify text fields containing German content, and send them to the DeepL REST API for translation. The DeepL API was configured with specific parameters to maintain paragraph structure and line breaks, the neutral scholarly tone, and to ensure accurate language pair handling. Crucially, our scripts translate description fields, and all other metadata including object IDs, geographic coordinates, collector names, and provenance records remain untouched.

Thirdly, in the validation phase, we conducted blind A/B user testing through a custom Streamlit web application with SQLite database backend where 20 human evaluators reviewed side-by-side comparisons of DeepL versus Google Cloud Translation API outputs, with translation sources randomized and hidden to prevent bias. The results were stored in a structured database for statistical analysis, and DeepL demonstrated superior handling of complex ethnographic terminology and historical German phrasing.