What kinds of stories do we tell? The narrative structure of history.
Since the establishment of history as a discipline, narrative storytelling has been the historians’ main device to structure and disseminate results. Recently, however, the introduction of novel digital tools has extended our possibilities to represent the past dramatically. Natural language processing, social network analysis, 3D modelling and GIS applications do not only allow us to analyse our sources and data in hitherto unthought-of ways, but also provide us with innovative methods to describe and visualise our data. What are the differences between these computational methods and traditional narrative storytelling? Are they opposed to them? Does one method preclude the other? In the first part of this blog post, I am going to outline some of the features of narrative storytelling. In the second part, I will compare those features with some of the characteristics of social network analysis. Social network analysis has become a well-established method in the humanities and social sciences and is therefore well suited to serve as an example of a novel representational medium.1 Four topics in the context of historical narratives will be especially emphasised, first person experience, time, explanatory understanding and epistemic constraints.
To start off, let us look at one passage of Anita Guerrini’s Pfizer Award winning book the The Courtiers’ Anatomists,
“Under cover of night, the dead of Paris made their journey from the burial grounds to the places of dissection. In this era of recurrent plagues, their numbers never dwindled, and for three centuries from the 1530s, they did not lie quiet in their graves. The cemetery of Saints-Innocents, between the rue de la Ferronerie and the rue St. Denis, was one of the few places with streetlamps until Louis XIV ordered the installation of thousands of candle-lanterns across Paris. But its dim beacon did not deter the trade in the dead, as physicians and surgeons exhumed bodies. Shadows from their torches made the danse macabre carved into the wall of one of the bone-houses, the charniers, seem to move. Saints-Innocents had been filled many times over, and as new bodies came in, the bones of the old were disinterred and placed in the charniers. The famous Flemish anatomist Andreas Wesel, known as Vesalius, fondly remembered the piles of bones at Saints-Innocents during his days as a medical student; he and his friends blindfolded themselves and took bets as to who could identify the most bones by touch. The smell of decomposing flesh permeated the quartier.”2
Guerrini’s description of early modern Paris takes us on a journey through the city streets to the cemetery of Saints-Innocents. She conveys to the reader a vivid picture of the visual, olfactory and tactile conditions surrounding the cemetery, an overall repulsive environment. The narrative form of her description allows Guerrini to paint a picture of Paris that mediates the experiences of historical actors by triggering our own imagination. Through the description of the streets without street lamps, we imagine the city to be dark. The smell of decomposing flesh maybe causes us to be disgusted or provokes negative emotions towards the burial ground. Vesalius fond memories give us a first hint that he is a passionate anatomist and foreshadows his later fame. Descriptions like these do not only inform us but also inspire our own imagination. This is not always harmless. Given untruthful or ambiguous descriptions, readers may interpret the text in a wrong way. Because we can neither literally look into the head of historical actors nor determine what thoughts a narrative will provoke in the mind of the reader the mediation of subjective human experience is beset with difficulties and depends on the rhetorical skill of the historian.

[CC BY-SA license] On this picture, we can see Holy Innocents’ Cemetery, the same cemetery we read about in the first quote of this blog post. This is a reconstruction of the cemetery before the end of the 18th century, when the city removed the bones and transferred them to the catacombs beneath Paris. The virtual reconstruction of the visual details for the reconstruction of the cemetery, but without any kind of description it lacks explanatory content. How did the bones get there? What was it used for? These details are potentially provided by a description.
As a first approximation towards the concept of narrative, we see that narratives are a convenient way to convey human experiences. In addition to that, two other features of narratives, the dimension of time and the connectedness of events, are of central importance.
A narrative is a story that connects individual events in specific ways.3 The difference between a narrative and a chronicle depends on the way the events are or are not connected. A typical example is this. The king died and then the queen died is a chronicle, the king died and then the queen died because of grief is a narrative.4 Why did the queen die? Because the death of her husband caused her to grief. This particular explanation describes the mental state of the queen and the event that was brought about by this mental state. Events are located in time. Because narratives consist of the description of events, we can locate them in time too. Porter Abbott especially emphasised the function of narratives to convey temporal information when he writes, “[the] narrative is the principal way in which our species organises its understanding of time.”5
Inextricably connected to the temporal structure of narratives is the connectedness of events and the explanatory function of narratives. One of the main functions of a historical narrative is to provide the recipient with a better understanding of why certain events (like the death of the queen) happened. This explanatory function of narratives has a long tradition in history. In the first paragraph of Herodotus book, The Histories, he describes why he wrote the book. The overall topic consists in the retelling of memorable events, especially the events that brought about the Persian war.
“This is the display of the inquiry of Herodotus of Halicarnassus, so that things done by man not be forgotten in time, and that great and marvelous deeds, some displayed by the Hellenes, some by the barbarians, not lose their glory, including among others what was the cause of their waging war on each other.”6
Similar to the king queen passage above, where we saw the difference between chronical and narrative, Herodotus too explicitly mentions his intentions to write about the causes of the war, another example of the long-standing tradition of narrative storytelling.
Historical storytelling is an extension of storytelling in everyday life. The representational medium of most narratives is ordinary language, but in principle, nothing excludes nonlinguistic forms of representations. The close relation between every-day life and historical narrative makes history an easy accessible subject but can also lead to confusions about the scientific status of the story told. In an academic context, narrative storytelling is subject to specific epistemic requirements. The story told should accurately resemble the events that happened in the past. The main method to fulfill this requirement is to abide the interpretative restrictions constituted by the available sources. (What Reinhart Koselleck called Vetorecht der Quellen). Our narrative representation of the past has to withstand the confrontation with all the available sources.
For the larger part of history’s own institutional development, the prevalence of narratives has not only been a given, but also formed an important point of self-identification for the discipline. Historians thought of themselves not just as the collectors of individual facts but the author of stories. In The Historian’s Craft Marc Bloch describes this point in the following terms, “To neglect to organize rationally what comes to us as raw material is in the long run only to deny time – hence, history itself.”7
Some of the main features of narrative storytelling:
- It provides a flexible form to represent historical facts with the help of a natural language.
- It is easy to incorporate descriptions of idiosyncratic first person human experiences.
- They allow to leave gaps open. If certain details of the story cannot be reconstructed or are uncertain, the narrative can be continued without problems.
- Because they describe events, narratives have an inherently temporal structure.
- Narratives increase our understanding of the past by showing which factors were responsible for the occurrence of an event.
Some of the advantages of narratives can turn out to be highly problematic if we do not handle them with care:
- The flexibility of a narrative can be a problem. Unrestricted storytelling invites biases or distorted descriptions of past events. The desire to make sense of the world can also lead to the creation of narratives which show more coherence than is warranted, this is sometimes called story bias.
- Gaps in the story are easily brushed aside and create the appearance of cohesion that cannot be justified by the evidence/sources we have.
Social network analysis
In most cases, natural language provides the basis for narrative story telling. Let us have a look at another method to represent our data. This picture shows the relationship between a group of boys belonging to the same class in the 1880s.
The dataset used to create this network was collected by the German primary school teacher Johannes Delitsch and published in 1900.8 Between 1880 and 1881, he observed the relationship of the pupils in his school class. I created the picture with the help of the graph visualisation software Gephi.9
Data: https://github.com/gephi/gephi/wiki/Datasets
Social network analysis quantitatively assesses the connections between different entities. These entities do not have to be individual humans, but also can include words, institutions, material things and so on. The only important thing is that the relationship between the entities can be expressed in a quantitative form. Delitsch for example observed his pupils during class, based on his observations he then created a table with friendship requests, reciprocal relationship and other measures. Graph theory provides the backbone of network analysis, its rigorous mathematical framework can be used to formulate explicit definitions about the constituents of a network. A graph consists of a set of vertices (nodes) and lines (edges) between those vertices. Different kinds of centrality measures can be used to describe and visualise how the nodes are related. The nodes in the picture above were ordered according to the degree of connectedness to other nodes, with the nodes with the highest degree in the middle and colored in a darker green than the other nodes.
There are some striking differences between this kind of representation and traditional narrative. I will compare it according to the three features of narratives I talked about in the first part of this text, first person experience, time and explanatory understanding.
With regard to first person experience, we can see that the network does not necessarily tell us anything about the way the boys in the network experienced their friendship to others. No historical actor necessarily perceived or knew about the relationships between the children in the same way as it is represented here. Nonetheless, the network can provide us with important information about the circumstances they lived in and which influenced the children’s behavior. It also can lead us to new questions, for instance, why are the repeaters in the class, Pfeil, Vetter and Schnabel at the center of the friendship relations?
In the picture above, the network depicts a static picture of the class between 1880 and 1881, but it is easily conceivable that with sufficient data we extend the network to include changes over time. In this way, we could animate the graph.
We noted above that narratives often include explanatory information regarding the reasons why an event happened. The network in itself does not provide this information, but can be one part of such an explanation. We could for example ask ourselves why the repeaters were more popular than the other children were. In this case, we could combine ethnographic observation with the network structure we want to explain. Combined, they form the complete explanation and the network plays an auxiliary role to formulate a more detailed narrative about the past.
Some of the distinctive features of social network analysis:
- Networks represent the relationship between entities.
- Those relationships have to be describable in mathematical terms. (It also possible to represent qualitative data in quantitative terms, for example, a gesture of friendship as one, the absence as zero)
- Because of the mathematically explicit properties of the elements in the networks, they we can analyse them with the help of algorithms. Algorithms can for example calculate some of the features of the network. The use of algorithms in this context ranges from the calculation of the shortest path between two nodes to more complex measures. Here are some basic examples:
-
- Degree Centrality, describes the direct connections between a node and the rest of the nodes
- Betweenness centrality, describes how often a node lies between two other nodes
- Closeness centrality, describes how close a node is to the other nodes in a network
Summary
The difference between narrative storytelling and alternative forms of representation like social network analysis lies in the different kinds of restrictions and possibilities that come with the respective method. Restrictions on the side of social network analysis include the representation of first-person experiences, the depiction of time and the missing possibility to causally explain events. There are new opportunities when it comes to the algorithmic analysis of networks, either as a heuristic tool, or to understand network effects, and the explicit meaning of all the elements involved in the representation.
Historians use narrative storytelling usually to describe what happened in the past. Social network analysis can serve a similar descriptive function, but it is more restrictive. Networks describe and visualise only the relationships between entities. This comes at the cost of certain kinds of information that is usually part of narratives. As we have seen, the inclusion of first-person experiences, that is the way somebody felt about something, plays an important role in traditional narratives. The relationship depicted in networks can coincide with the experiences of historical actors, for example in the case when the boy at the center of the class friendship network also is aware of his function within the network. However, it is also possible that this is not the case and that the network does not resemble the perspective of the people in the network. If we are instead interested in the social structure of the group network, social network analysis provides methods to make this structure visible and analyse it. Because the elements and methods in this kind of analysis are explicitly defined, we are less likely to build ambiguities into our description.
Another central difference between narrative and network concerns the question what we can do with both entities. Because networks are built according to certain rules (the restrictions I described above), they can also be analysed with the help of algorithms. Through this analysis, it is then possible to discover those structures that would have been overlooked otherwise.
- As can be seen by the establishment of specialised journals. For example https://jhnr.uni.lu/index.php/jhnr/index
- Guerrini, Anita. The Courtiers’ Anatomists: Animals and Humans in Louis XIV’s Paris. University of Chicago Press 2015. p. 17.
- Herman, David. The Cambridge Companion to Narrative. Cambridge University Press 2007. Sometimes definitions of narratives explicitly include human experiences, but to me it seems to be sensible to speak of narratives about non-human or even non-animated objects.
- This is a much-quoted example and originally comes from Edward Forster. Forster, Edward Morgan. Aspects of the Novel. Reprinted. Penguin Classics 2005.
- Abbott, Horace Porter. The Cambridge Introduction to Narrative. Cambridge Introductions to Literature. Cambridge University Press 2008. p. 3.
- http://www.perseus.tufts.edu/hopper/text?doc=urn:cts:greekLit:tlg0016.tlg001.perseus-eng1:1.1.0
- Bloch, Marc. The Historian’s Craft. Manchester University Press 2002. p. 121.
- For more information regarding Delitschs’ work see, Heidler, Richard, Markus Gamper, Andreas Herz, and Florian Eßer. ‘Relationship Patterns in the 19th Century: The Friendship Network in a German Boys’ School Class from 1880 to 1881 Revisited’. Social Networks 37 (May 2014): 1–13.
- Bastian M., Heymann S., Jacomy M. Gephi: an open source software for exploring and manipulating networks. International AAAI Conference on Weblogs and Social Media (2009).