Historical travel research, especially the study of travelogues, has been a popular field within literary and cultural studies. Research focuses on the exploration of the numerous cultural aspects of travel such as: the formation and rules of the literary genre of the travelogue, reception research, research on cultural contacts and stereotypes and gender aspects (see [Bödeker et al. 2004], [Maurer 1999], [Gruber 2022]). Practical and technical aspects of travel, such as: routes and roads, personal equipment, network and quality of inns, money, currencies and costs, and means of transport (examples: [Mączak 1995], [Brilli 1997], [Gräf and Pröve 1997]) are also of importance, as well as travel relationships between certain groups of people such as students, artists and scholars. Mostly the focus is not on the journey itself, but on the destinations and the perceptions of the travellers.

Despite the rise of digital humanities (including the field of digital editions, the discovery of space in social sciences and the development of geoinformation systems), there has not yet been any significant progress in the indexing, processing, and visualisation of travelogues, either in terms of content or method. Current digital travelogue editions can therefore be subsumed into three categories:

This paper presents a first draft of the ongoing work on the edition of Franz Xaver Bronner’s travelogues, in which we apply a semantic model that combines widely accepted standards (such as CIDOC CRM, Dublin Core, SKOS etc.) with project-specific elements and TEI. We address the question of category development for time, space and events through text encoding and semantic annotation, and their assets for research while focusing on the current DFG-funded project “Digital Edition of Historical Travelogues” (DEHisRe).[6]

After introducing different standards, concepts and modelling approaches in the digital edition field, we will focus on the ontology development process in section two. Edition guidelines according to TEI were also developed, but are not part of this article. We describe the development of different ontology design patterns and its application to travel events and observations. Some examples for travel events are illustrated in section three.

In section four, we present specific research questions which serve as a basis for the development and refinement of our ontology design patterns. Here we give an insight into the ongoing iterative process based on interdisciplinary exchange between historians and ontology developers. The research questions presented focus on the following issues: 1) observation of practices: ethnic groups and cultural norms, and 2) gender encounters: narrating the varied women's agency while on the road.

This approach is applied to the digital edition and additional material of the unpublished travelogue of the German expatriate in Switzerland Franz Xaver Bronner (1758–1850)[7], who in 1810 went to the Russian University of Kazan on the Volga to become a professor in theoretical physics and travelled back to Switzerland in 1817 (see [Beyer-Thoma 2017]).

To conclude the research, we will give a short overview of the possibilities of connecting TEI and ontologies. Here, suitable tools are presented: advantages and disadvantages as well as the different focal points of these tools are discussed. The ideas presented are based on initial evaluations of the implementation of concepts (textual and ontological) within the project.

According to Kudella and Jefferies (cf. [Kudella and Jefferies 2019]), two groups of data modelling can be distinguished, curation-driven and research-driven data modelling. Curation-driven describes the efforts of libraries and archives to record objects in a uniform way. Standards are used to ensure retrievability and transparency. Research driven data modelling aims to answer specific research questions and follows the interest of a researcher.

It is argued that this contrast between research and curation driven data modelling makes the re-use of data difficult or even impossible. The reason for this is the gap created by standardised data collections, as carried out by libraries, which are not sufficient for scientific interpretations and miss the actual research interest. Science-led data collections, on the other hand, are very specific and are rarely oriented towards standards. A feedback into information systems is therefore not possible “because essential information is lacking and because data is captured in non-standard ways”  [Kudella and Jefferies 2019, 42]. Eide (cf. [Eide 2014]) addresses the same issue by differentiating between “using a model” and “creating a model” when developing data models. He argues that “middle positions can be found when one takes an existing ontology and extends it, or uses terms taken from multiple ontologies in combination”  [Eide 2014, 5]. To ensure re-use and interpretation independent of a specific research project, it is important to consider current standards.

In addition, Eide [Eide 2014] uses the distinction between modelling for production and modelling for understanding in the context of TEI encoding (as a form of modelling) of texts. Our approach tries to bridge this gap by re-introducing research driven perspectives into ontology-based modelling and by extending this distinction to the use of TEI editing.

The TEI Guidelines serve as a de-facto standard for digital editions. “The Text Encoding Initiative [TEI] is the most systematic effort so far to create standards for scholarly memory in an evolving digital culture. The TEI markup language includes [...] different elements to satisfy all manner of scholarly needs in the humanities”  [TEI 2019]. The newest version of TEI includes 585 elements (see the TEI Consortium 2022) and was designed to be flexible in order to make it applicable to a wide range of research questions. It is not unambiguous, however, and the same information can be encoded and interpreted differently. This leads to interoperability problems because of the various possible ways of encoding text elements (e.g., <rs>, <placeName>, etc.) (see [Unsworth 2011]; [Kudella and Jefferies 2019]; [Burrows et al. 2021]; [Giovannetti and Tomasi 2022]). The TEI therefore reflects the dichotomy between research and curation driven modelling practice mentioned above (see 2.1).

According to Giovanetti and Tomasi (cf. [Giovannetti and Tomasi 2022]), linking Linked Open Data (LOD) and TEI could address the problem of interoperability: “Enhancing TEI-encoded texts with linked open data represents a possible way to overcome this limitation”  [Giovannetti and Tomasi 2022]. We therefore try to solve the interoperability and expressivity problems of TEI and CIDOC “Conceptual Reference Model” (CRM) by means of “Deutsches Textarchiv Basisformat” (DTABf) and frame-based ontology design patterns. This, in turn, makes the categories of travel events and travel observations in historical travelogues explicit for further analysis and visualisation.

We use the CIDOC CRM for the enrichment of the text with knowledge made explicit and stored in a database. It is an event-oriented and well-accepted ontology in the cultural heritage sector. The CIDOC CRM has been developed since 1996 as an Application Ontology for the documentation of museum artefacts and was accepted as an ISO standard for data modelling in the cultural heritage sector in 2006. It is intended to serve as a transdisciplinary “lingua franca” for the standardisation of knowledge on cultural heritage objects and structures and is therefore not limited to pure museum documentation. Based on this harmonised, machine-processable cultural knowledge, the CIDOC CRM aims to facilitate the exchange of relevant information between museums, libraries, archives and comparable institutions as well as between different research projects (see [Meyer 2019]).

The event-centred modelling approach of CRM (see 2.2) enables us to model the fabula and sujet of a travelogue (cf. [Bartalesi et al. 2017]).[12] As Maurer (cf. [Maurer 2015, 391f]) notes, travelogues written by scientific explorers often follow a thematic organisation to present “travel results” instead of simply following a chronological order (see 1.1 and 4.).

To build the annotation scheme and associated ontology design patterns, i.e. to interlink text and data, we apply frame semantics as a theoretical framework. Frames can be represented as n-ary relations[13] and used as skeletal structure in the development of ontology design patterns.[14] Moreover, frames can be considered as “knowledge patterns” for validation of the ontology design patterns (cf. [Presutti et al. 2012]).

The ontology design patterns are the basis for the ontology-based text enrichment and editing workflow. The reason for the frame-like structure of classes for representing travel events and activities during travelling and stopovers is motivated by the semantic annotation use case of interlinking text and data. Our ontology design approach is based on requirement analysis, whereby the competency questions are derived from initial user stories to define data modelling and information retrieval requirements.

From an ontological perspective, a travel or journey has to be considered as a kind of event. To be more specific, it is a composite event where a certain person in the role of the traveller is involved throughout all its subsequent events, i.e. travel segments as parts of the travel as a mereological whole. In order to make explicit the segments of a journey reported in a historical travelogue, we need an ontology design pattern to model a journey and all its travel segments including the mereological, chronological, as well as their continuation relations. The traveller conducts several activities (typically observations, arrangements, etc.), which are reported in the travelogue and must therefore be modelled in the database of its digital edition. This includes not only activities when on the move, but also during stopovers. As we have to be able to distinguish between these two modes, we need two major classes to represent a travel as a complex whole composed not only of journey segments, but also of temporary stays at places in between travelling: a class to model travels/travel segments and a class to model stopovers.

In a first step, we added to the types of activities the item “(he/she/them) compares”[23] since Bronner often made comparisons, especially after crossing the Russian border behind Memel (now: Klaipėda). A visualisation of the results can be seen on Figure 4:

Before Bronner migrated to Switzerland in 1794, he had lived in the Swabian-speaking region centred around Höchstädt, Donauwörth and Augsburg. After the Swiss Revolution of 1797, he got a position in the administration of the newly founded Helvetic Republic and in 1803, when it was dissolved, he had accepted an ill-paid and insecure position as a teacher in Aarau, the capital of the newly created Swiss canton of Aargau, which Bronner regarded liberal enough to be a safe place for a former revolutionary. Bronner had become a Swiss patriot [Radspieler 1967, 13–79][24] and the country, in his comparisons, stood for a lovely landscape, flourishing agriculture and free, wealthy people. Germany, meanwhile, represented in Bronner's view a higher state of culture. Bronner’s home region and the “Upper Germany” (i.e., Southern Germany), however, very seldom appeared as a reference during his journeys. The contrast with Switzerland got marked for Bronner when he entered the Baltic provinces of the Russian Empire, the more so since the serfdom that prevailed there had already reached the attention of the educated public of Germany [Heeg 1992]. Bronner also made comparisons with places formerly seen on his travels.

These findings spurred us to refine our ontology to more abundantly reflect all kinds of perceptions made by Bronner. To start with a narrowly travel-related subject: Bronner's descriptions of officials change at the border with Prussia in Treuenbrietzen where he was vexed by toll officials. Consequently, we classified professional groups and added various types of observations regarding his perception of officials to the travel activities classification system to be used in the pattern for the Perception_active class (see Figure 5).

Figure 6 shows the area of distribution of Bronner’s utterances of dissatisfaction with state officials that started at the Prussian frontier and continued throughout Russia and – in a slightly milder way – Austria. The last instance was in Braunau before Bronner crossed the Austrian-Bavarian frontier. Bronner, nevertheless, noted in Braunau in his draft travelogue: “Compare the honest character of Austrians and Bavarians with [that of] Poles and Russians” (Bronner 1817, December 1).

Bronner describes unpleasant encounters with Russian people during his journeys but refrains from commenting on them. The encounters with Polish and Jewish people, on the other hand, are more explicitly commented on. Figure 7 shows the authors' perceptions of these two ethnic groups according to the travel activity classification system. Bronner first encountered Poles in the former Polish territories of Prussia, thereafter in the Baltic and in the Ukraine, and the Eastern part of Austria. He commented on these encounters, with one exception, always in a negative way. His critical observations on Jews, however, focused on Eastern Jewish lifestyles, and Jewish prevalence on the whole, in Polish towns. Bronner mentioned Jewish people and communities eastbound from Prussia onward and on the way back until Bohemia.

As a result of close reading of the text and a refinement and completion of our ontology, we can observe that Bronner’s view on “Germany” as a stronghold of higher culture and on “Germans” and “Austrians” as more “honest” people is mixed up by his own observations.

Bronner’s text, for the most part written only as a draft to be revised later, nevertheless testifies to his abilities as an accurate observer especially of human relations. He scarcely commented on the many, mostly short, stories with women he reported on. This is why we decided to examine how Bronner represented gender relations in his travelogue.

Bronner described the regional costumes of women and by applying the method of geographical mapping, we already found some interesting features concerning gender relations, as can be seen in Figure 8. Remarks on women can be divided into three categories, here visualised by diamonds: mostly not commenting on them (white diamond), emphasising various characteristics of women (light green diamond), negative remarks about a woman’s character or appearance (various kinds of red diamonds). In the culturally divided Baltic (starting already in Eastern Prussia), the number of observations increased in opposite directions. Bronner saw women dressed in the European or “French” manner and he was astonished at the incomplete, by his standards, clothing of the Baltic and later the Russian peasant women (“just a shirt”). Likewise, he disapproved that their clothes were not figure-hugging.[25] In the Baltic, Bronner for the first time encountered prostitutes and prostitution, but he stops mentioning them, as well as observations of flirting with a few exceptions, in Russia proper, only to start anew on the way home after the Austrian border (various shades of blue). Interestingly, on the way home Bronner noted with negative remarks that Russian host peasant women demanded a lot of money for boiling water and importunately, as he saw it, asked for gifts for their children and the elderly. Accustomed to the restrained behaviour of German middle-class women, he, as it seems, could not approve that Russian peasant women strongly defended the interests of their family in the sphere which was under their uncontested control.[26]

These observations aroused the question of female agency within these encounters and of the impact of the social and cultural norms. To attain results, attention had to be centred less upon the perceptions of the author, but upon how women acted and why – at least as it is portrayed by Bronner as an observer who undoubtedly was determined by the norms of his time.

One way of retaining compatibility with the ontology pattern is to apply the methodological approach of the Actor-Network Theory (ANT). It also uses graph-conform data consisting of vertices (“actors”) and edges, but ANT networks are not set out for being operationalized with mathematical methods nor even for visualisation. Instead, they are an instrument for reflection and constant refinement. A central claim of the ANT is that non-human actors of any kind must be considered, though not as preconceived constructs of the researcher. Instead, they must be discernible in the text as objects, motives, ideas, or clusters of them and human actors, conceived by the author or actors themselves at least in an indirect way.[27] The non-human actors to be defined by us according to the ANT mostly were too vague to be palpable and unequivocally identifiable in the text of the travelogue. Consequently, they could not be tagged and the tight connection of TEI tags and the database could be lost. Furthermore, our design pattern so far did not provide non-human actors. Let us give an example: When mentioning that women in a Riga banquet (Bronner 1810, August 30) sat around the housewife, Bronner clearly spoke about a kind of segregation of sexes. This was not the only time Bronner mentioned it, also in the Baltic (Bronner 1810, August 26), he wondered that domestic servants and poor travellers of all sexes shared for the night the same main room of an inn. In this example, too, he refers to a norm of segregation of sexes he was accustomed to from German households regardless of their wealth. We therefore have, according to ANT, to consider different norms of segregation of sexes as “actors”: two norms of Baltic origins and Bronner’s accustomed norms which he brought with him from Germany and Switzerland. As non-human actors these norms interact with people who are sharing them or not acting accordingly.

Since agency can be assumed as a unitary action of exerting influence, we decided to treat it with the frame-inspired modelling approach. The representation consists so far of two actors, an activity (using the Activity Types classification system), a determinator of strength ranging from 1 to 3, and the definition of direction. We also provided for the possibility of clustering several actors into one as an important requirement of ANT. Some actor-centred activity types had to be added, among them, of course, such that relate to non-human actors. In ANT, they are conceived vaguely in terms of “participation” which can be grasped via thematic roles in the frame semantics framework. But a person can set pressure upon another person by intentionally appealing to, say, a legal or moral norm, and that norm, in consequence, may become coercive – or just for this reason provoke resistance.