Evelyn Gius

Narration and Escalation

An Empirical Study of Conflict Narratives

This article describes the methodology and the outcomes of an empirical study of conflict narratives. The narratological analysis deployed narratological categories in the structuralist tradition based on Genette and was conducted with the help of the text annotation tool CATMA. The analysis aimed at covering as many narratological phenomena as possible by establishing 14 fields of narratological phenomena that were annotated in a corpus of 39 factual narratives about situations at the workplace with and without conflicts. The evaluation of approximately 28,000 annotations brought to light a series of interrelations between narratological phenomena and the presence or absence of conflicts in the narratives. Additionally, this approach led to the identification of some oversights of narrative theory by detecting hitherto unnoticed interrelations among narratological concepts.

1. Modelling a Narratological Approach to Conflict Narratives

1.1 Conflicts from a Narratological Point of View

This study1 was motivated by an observation from my practice as a mediator that pointed me to a previously unnoticed connection between conflict management and narratological phenomena: one of my mediation supervisors advised us to pay particular attention to recurrent statements made by the parties involved in the conflict. Recurrent statements, he argued, indicate issues that still need to be treated in the mediation in order to solve the conflict. As a narratologist, I rephrased this into the observation that one should be aware of repetitive narration during the mediation process. Moreover, assuming that repetitive narration in these cases indicates a high degree of tellability, one could reason that the events told are crucial for the conflict’s resolution.

Narratological terminology thus seems to be adequate for the description of relevant aspects of conflicts. A closer look at the core concepts of the two fields, ‘event’ and ‘conflict’ even reveals a certain analogy on the conceptual level. Jurij M. Lotman (1977, 240-241) defines an event as the movement of a character beyond the limits of a semantic field or a crossing of a forbidden border and these definitions in narratology have been conceived of as transgression of semiotic boundaries. Friedrich Glasl (2011), on the other hand, defines (social) conflicts as an interactive situation with an incompatibility in acting among the participants, together with an incompatibility in wanting, feeling and / or thinking that is perceived as incompatibility by at least one of the participants. One could therefore argue that the perception of the behavior of another person as incompatible equates to its perception as transgression, and the transgression by a character of “forbidden borders” yields a (semiotic) position that is incompatible with the position on the other side, i.e. before the crossing. While the concepts may not be based on quite the same conception of transgression, there remains a certain analogy: the crossing of borders is decisive for the presence of the crucial features in both fields, namely the eventfulness of a narrative and the conflictual character of interaction situations.

Against this backdrop, a closer analysis of the relevance of narrative structure for narration-based conflict treatment seemed promising. This led to the study described here, which was designed as a narratological approach to conflict narratives. My approach was necessarily inductive because I refrained from making a priori assumptions about the relevance of specific narratological phenomena in conflict narratives. Instead, I aimed for as comprehensive an analysis as possible of the whole bandwidth of narratological phenomena in conflict narratives, by first identifying all occurrences of the phenomena and subsequently evaluating them with regard to their frequency and distribution within the conflict narratives. Since classical narratology focuses on form rather than content, no content related aspects concerning the conflict – such as the issue of the conflict, the assessment of the overall situation by the narrator, etc. – were included in the analysis of the conflict. Such an approach is obviously not common practice in conflict management. It can therefore hopefully yield results and insights that complement existing approaches to conflict management.

On these premises my study first examined the narratological constitution of the narratives independently of the acute conflict referenced therein; in a second step, it considered narratological aspects of the conflict narratives related to a specific formal characteristic of conflicts, namely their level of escalation. Where interrelations between narratological phenomena and escalation of the narrated conflicts were detected these could then be described as the narrative profiling of conflicts, i.e., as the narrative configuration of narratives about conflicts with regard to their escalation.

The most relevant steps in the overall analysis will be presented in the remainder of this article: the analytical categories used will be described in the following subsection, whereas the methodological premises relevant to their application will be detailed in section 2. Section 3 will provide an overview of the actual narratives that were analyzed. The outcomes of the analysis will be summarized in section 4, with section 5 discussing additional findings that need to be further examined from the perspective of narrative theory.

1.2 Narratological Categories for the Analysis of Conflict Narratives

Since the research question originated in the field of conflict management, the analysis of the narrative configuration of narratives about conflicts was conducted on factual narratives. The conceptual base of my study was classical narrative theory, in particular the conceptual framework proposed by Gérard Genette (1972), which provides well-established categories (see also Lahn / Meister 2013). Compared to other concepts in literary studies, classical narratological categories tend to be well-defined, and they cover a broad range of narrative phenomena. My approach was further based on the narratological premise that literary and factual narratives generally show considerable affinity. This holds especially with regard to the underlying communication model, as Tatjana Jesch et al. (2006, 43-44) show and Uri Margolin generally observes of narratological narrator concepts: “readers process literary narratives in the same way as they do ordinary communication insofar as they assume a textually encoded conversational partner responsible for the contents of the narrative” (Margolin 2013, 4). Accordingly, one can adopt even classical narratological categories for the analysis of factual narratives – despite the fact that they were originally developed for literary narratives.

Following the survey of narratological categories in Lahn / Meister (2013), more than 400 categories for the description of narratological phenomena were grouped into 14 phenomenon fields. By narratological phenomenon fields I conceive category fields in which features of narrative can be grouped together. A fifteenth field, containing categories for the description of reliability phenomena, which is also mentioned in Lahn / Meister (2013), was not considered since it doesn’t meet the prerequisites of the text-inherent approach discussed in section 2 below. The 14 analyzed fields and their central questions are: (1) narrator representation (“How is the narrating instance – implicitly or explicitly – presented?”), (2) relation narrator–world (“(To which degree) is the narrator part of the narrated world?”), (3) narrative levels (“Does the narrative consist of more – embedded – narratives?”), (4) relation narrator–event time (“What is the relation between the time of the narrated and the time of narration?”), (5) narratee representation (“How is the addressee of the narrative(s) presented?”), (6) perspective (“How and by whom is the perspective on the narrated determined?”), (7) modes of representation (“What kinds of speech and mental processes are presented?”), (8) relation discourse time–story time (“How does the time of the narrated relate to the time of its narration?”), (9) information allocation (“How is information assigned to the instances of the narrative and who knows what?”), (10) self-reflexive narration (“Does the narrative contain references to the act of narration or reflect it?”), (11) aspects of plot (“How is the plot constructed?”), (12) character (“What characters are part of the narrative and how do they contribute to the narrative?”), (13) space (“What are the spatial characteristics of the narrative?”), and (14) time (“What are the temporal characteristics of the narrative?”).

Figure 1 shows these 14 fields as well as the categories in the exemplary subset ‘order’ that is one of the three subsets of the phenomenon field ‘relation discourse time–story time’.

Fig. 1: The 14 narratological phenomenon fields and the subset ‘order’

2. Methodological Preliminaries

2.1 An Inductive Comprehensive Approach

The approach I developed differs from other narratological approaches in being directed towards a possibly comprehensive analysis of phenomena in a pragmatic domain of narration – and not towards an analysis of selected phenomena only. Additionally, narratological approaches normally focus on one or two, and at the most, a couple of narratives. A comprehensive analysis of a corpus of narratives is rather rare in narrative theory, and therefore it presupposes the development of a methodologically adequate implementation of the analysis. It follows that this methodological implementation should be described in its most important aspects, which this section undertakes to do in some detail.

Research in narrative theory has to date only provided a comprehensive analysis of phenomena in the context of the development of theoretical contributions: Some authors have based their theoretical framework on the analysis of a single literary work – the most prominent probably being Proust’s À la recherche du temps perdu, which is extensively used by Genette for the elucidation of the introduced concepts. Still, in such cases the literary work under treatment functions primarily as the illustration of the introduced concepts, and not as the objective itself of a comprehensive analysis. These cases, then, do not provide a methodology for an approach as the one described here. The methodological approach for my study had instead to be developed practically from scratch.

For this purpose, I first identified three prerequisites for the analysis, namely (1) no relation to the content of the narrated conflict, (2) text inherence, and (3) a wide renunciation of interpretation. These prerequisites derived from the research question and the breadth of the projected analysis. They are motivated as follows: since the goal of the study is the identification of relations between narratological phenomena, i.e. the form of narratives, and the degree of escalation of narratives, (1) aspects of content had to be excluded from the analysis. The other two prerequisites arise from the analysis that encompasses a broad variety of narratological concepts and is thus rather complex, due both to the number of concepts in question and to the lack of insights about relations between them. In order to make the analysis feasible nevertheless, the complexity of the applied categories needed to be kept as low as possible by (2) restricting the analysis to text-inherent phenomena and (3) keeping interpretation to a practicable minimum. Text inherence thereby meant that the analysis was limited to information available within the examined text. The adopted approach was thus text-immanent, or text-inherent, and broadly avoided drawing on extratextual (world) knowledge. In order to reduce the level of interpretation needed for the analysis, all analytical observations or statements should only be inferred directly from the text surface wherever possible. For this, linguistically marked aspects have been foregrounded.

The resulting concentration on the textual surface is at the same time adequate for a narratological approach since narrative theory concepts are primarily applied for the analysis – and not the interpretation – of texts. This has two additional methodological advantages: The focus on text-inherent aspects and text surface allows for a better reproducibility of the analysis and thus enables quantitative evaluation of the analysis. It also allowed for the subsequent re-use of this approach in a different research context: the formalizations and the weighting of categories for narratological analyses as described in subsection 2.3 were later used in the heureCLÉA project for the generation of an automated heuristics from manually annotated narratological phenomena (cf. Bögel et al. 2015). The second advantage of the concentration on the textual surface concerns the application of the outcomes: the results of an analysis are more consensual the closer the analysis is connected to the text, and the fewer interpretations it deploys that might lead to ambiguity. The targeted development of a heuristics for conflict handling benefits from this, while conforming additionally to Glasl’s demand for the priority of observation over interpretation in the beginning of conflict management: “Eine Dritte Partei muss für den Anfang vor allem beobachten und nicht zu schnell interpretieren wollen” (“a third party must first of all observe and not want to interpret too fast”, Glasl 2011, 104, my translation).

These three conditions – no relation to the content of the narrated conflict, text inherence, and a broad renunciation of interpretation – were, however, not always fully realizable. While references to aspects of the conflict can be easily excluded, mostly because conflicts themselves are not narratological phenomena, the requirements for text inherence and freedom of interpretation cannot always be met. This is due to the nature of certain narratological phenomena that cannot always be determined exclusively from the textual surface.

This was taken into account during the analysis by examining for each individual case whether additional information or interpretation would be necessary for the investigation of a phenomenon. In this way the relevance of each possible category for the description of a narratological phenomenon was weighted against the extent of both the necessary analysis of extratextual reference(s) and the necessary semantic knowledge for the linguistic interpretation. In some cases exceptions were made, especially for concepts regarding phenomena that were considered fundamental for a comprehensive narratological approach. For example, the analysis of frames and scripts in the context of action necessarily involved world knowledge or, more precisely, process knowledge. But even the analysis of phenomena falling under a supposedly formal category, such as frequency (“How often is something told in relation to how often it happens?”), also requires a certain dimension of semantic (i.e., linguistic) interpretation: as concerns determining whether something is told repeatedly, interpretation might be called for, as repetitions are often realized with different formulations on the textual surface.

2.2 Operationalization of the Narratological Analysis

In order to enable a variety of evaluations of the analyzed narratives, the annotations were realized within CATMA, an annotation tool that has been developed as a web-based platform both for the annotation and the analysis of literary and other texts.2 The analysis of the narratological phenomena was performed through a computer-aided method by (1) reading through the narratives, (2) marking up text strings where a phenomenon was encountered, and (3) assigning a narratological description to the selected text (i.e., tagging / annotating the text). Due to this computer-aided approach operationalization of the categories for analysis was a prerequisite. Yet operationalization is not a straightforward task, since many traditional theoretical concepts and terms for literary description are too vague or too abstract to allow for a straightforward formalization (cf. Meister 2003, 294).

Nevertheless, in the case of my study, in which humans – and not computers – were to apply the operationalized concepts and thus were responsible for the actual detection of the phenomena, comparably little formal operationalization was needed. The deployed narratological concepts had still to be operationalized in such a way that they could be used for an intersubjectively comprehensible annotation of narrative texts, on which further narratological analysis and interpretation of those texts could build.

In order to meet this requirement the following steps were taken to build a taxonomy-like, so-called tag set in CATMA:

  • step 1: the narratological phenomena were organized into a hierarchical concept structure in accordance with their interdependencies,
  • step 2: indicators for the presence of the phenomena were identified, and
  • step 3: these indicators were examined for their status as necessary and / or sufficient conditions for the occurrence of a phenomenon and included in the tag set accordingly.

For step 1, the hierarchical concept structure, parent-child-relations between concepts were examined. Lahn / Meister (2013) was the initial point of reference, with supplementary consultation, where necessary, of the living handbook of narratology (cf. Hühn et al. 2015). Where this was felt to be insufficient for the hierarchization, further references from both publications were taken into account (cf. figure 1 for an example of hierarchically ordered concepts in the subset ‘order’). Step 2, the identification of indicators for a given phenomenon, was realized by the actual application of the concepts for annotation. Step 3, the decision about the inclusion of the indicator as a tag in the tag set, depended on the question of whether it is a necessary or sufficient condition for the concept in question. This was determined by the examination of the relevant literature as well as by observations from the application of the tags during the annotation. The building of the tag set was thus an iterative process involving repeated testing and modification of the tag set.

During the building of the tag set, certain adaptations of narratological concepts became necessary. This was the case when the description of the concepts was not differentiated enough, was not precise enough, or the described phenomenon could not be fully determined from the text surface. These cases were treated as follows:

  1. Concepts that failed to differentiate between sub-phenomena that needed to be taken into account for the analysis of the concept were further differentiated to allow a more accurate description.
  2. Concepts that were not formulated precisely enough to be applied straightforwardly, or didn’t take into account phenomena that should be included, were revised accordingly.
  3. Concepts whose determination was only partially possible were either adapted with the introduction of additional assumptions, or alternatively the analysis was restricted to non-problematic phenomena in the texts.

The yielding tag set was finally described with regard to its application for text analysis and could be deployed for the annotation of the conflict narratives in CATMA. In this tag set every category within the 14 narratological phenomenon fields was mapped as a tag (i.e., every leaf of the mind map in fig. 1 above). The resulting tag set consisted of more than 500 tags.

2.3 Weighting of the Fields of Narratological Phenomena

Even though the concepts for the analysis were now operationalized hierarchically each within its respective field, the 14 phenomenon fields were all located at the same, uppermost level of the tag set (cf. figure 1) since in narrative theory there is no prioritization of one field over others. How, then, could the tag set be applied in a sensible manner?

This question was resolved by arranging the fields in accordance with the degree of complexity of the phenomena: it is both more efficient and more effective – i.e., conceptually precise – if the more easily applicable categories are analyzed and annotated first, followed by the more complex ones. In this way, a basis for determining more complex categories building upon simpler phenomena can be provided. Additionally, this approach prepares for the analysis of complex phenomena that entail a more intensive examination of the narrative since every annotation of phenomena within a specific narratological field entails the necessity of going over the narrative in its entirety once more, thereby systematically improving the annotator’s acquaintance with it.

Thus, the above-discussed conditions were the starting point for a weighting of narratological phenomena regarding the order of their application. The first condition, the disregarding of information about the conflict, was not considered because it is not relevant for the description and operationalization of narratological concepts for the phenomenon fields. The demand for text inherence, however, had to be taken into account for the weighting, just as the interpretation requirements had for the determination of a phenomenon. Additionally, the dependence of a phenomenon on other phenomena was included in the weighting.

For the definition of the degree of text inherence of a narratological concept, two aspects were considered: (1) the amount of text affected by the category in question and (2) the extent of the context necessary for determining the category. The range of text was set according to the amount of text typically annotated with a tag of the tag set. In this regard, it is important not to confuse the average annotated text range with the question of how much of a narrative may be generally affected by the category. A narrative text can, for example, be entirely annotated with tags for order phenomena concerning the relation between the order of the happenings and the order of their narration. Nevertheless, a single occurrence of an order phenomenon such as an analepsis or flash-forward includes typically only one or a few paragraphs – and not the entire text.

The amount of annotated text ranges from one or more words (e.g., for space) up to the entire narrative (e.g., for narrative levels in the case of a story that has only one narrative level).

The scale for the classification of the amount of text was accordingly:

  • 1 = word to word group;
  • 2 = word (group) to sentence;
  • 3 = sentence to paragraph;
  • 4 = paragraph to section;
  • 5 = section to entire text.

The intratextual context that is additionally required for determining a narratological category was also described according to these five gradations. Accordingly, ‘1’ stands for a category with no context or reference to the context of a word or a phrase (e.g., in the case of speech representation that can normally be identified by the introducing verba dicendi and / or quotation marks) while ‘5’ indicates that a large part or even the whole text has to be included (e.g., in the case of the relation between narrator and time of narration, where the analysis of temporal position of the narrator needs to be inferred from the whole narrative). The classification of text range and intratextual context can correlate, but they can also diverge significantly: for example, the category of self-reflexive narration typically concerns entire sentences (= 2) and its determination requires the context of sentences (= 2), as well. In contrast, the determination of speech and thought representation is virtually independent of the context (= 1), but usually affects sentences (= 2) and the analysis of characters involves virtually the whole text (= 5), but concerns only a text range from sentence up to paragraph (= 3).

The second weighting criterion, the interpretation in the sense of the semantic interpretation of the linguistic indicators, is set on the scale of 1 (very low need for interpretation) to 5 (very high need for interpretation). Accordingly, categories such as speech and thought representation, which can be determined by analyzing the text surface, and the determination of which is correspondingly uncontentious, are classified as barely interpretive. Categories such as the narrative perspective, which are non-deterministic and often subject of discussions in the field of narratology (cf. Niederhoff 2011), are in turn considered highly dependent on interpretation. As for the interpretation dependency, the 14 fields were each assigned a value from 1 (very little) to 5 (very much) according to the respective insights from their application.

For the purposes of an analysis based on linguistically marked elements, little dependence on interpretation is a main concern. Therefore, the narratological phenomenon fields were first weighted according to their dependence on interpretation and subsequently categorized according to the criterion of text inherence, i.e., the amount of tagged text and their context dependency. This weighting eventuated in the order shown in Table 1 below.

Table 1: Weighting of the narratological fields by need for interpretation and text inherence

This weighting procedure yielded ambiguous results only for the categories ‘time’ and ‘space’ that have exactly the same values for the three aspects. Since “most definitions, by characterizing stories as the representation of a sequence of events, foreground time at the expense of space” (Ryan 2012, 2), time was ranked above space.

This weighted order of application now needed to be adapted according to the third criterion such that connections between the concepts in the narratological fields were taken into account, which had become apparent during prior applications of the tag set on narrative texts. The following interdependencies were encountered:

  • The identification of narrative levels is a prerequisite for the analysis of all phenomenon fields (with the exception of time, space and speech representation), since their determination depends on the narrative level on which they occur.
  • Phenomena of character are also partly relevant for the addressee and the narrator and should therefore be analyzed first.
  • The analysis of the temporal relationship between discourse and story relies – in addition to narrative levels – on the segmentation of the narratives, which is included in the phenomenon field ‘aspects of plot’.
  • Finally, the analysis of metanarrative elements in the phenomenon field ‘self-reflexive narration’ is often closely linked to the analysis of narrative levels and, therefore, should immediately follow it.

These findings were integrated into the weighting of the 14 phenomenon fields. Phenomenon fields whose analysis builds on other fields, or is otherwise closely linked with these, were ordered directly after these phenomenon fields. This finally resulted in the moving up of some fields and yielded the following weighted analysis sequence (in bold: modifications to the original weighting by moving the field up):








narrator representation


modes of representation


narratee representation


narrative levels


aspects of plot


self-reflexive narration


time relation discourshistoire


relation narrator-world




relation narrator-event time


information allocation

Although the applied criteria are of a rather general nature, the weighting needs to be considered a simplification. This is due to the heterogeneity of the categories in question with respect to the classification criteria: within the category ‘character’, for example, there are both categories that can be determined based on single words in the text (such as the designation of a figure with a proper name) and categories whose determination presupposes knowledge of the entire narrative (such as the determination of character features). The classification of the category ‘character’ as a very context-dependent one is accordingly an inevitable generalization that doesn’t consider some (simpler) aspects of the category. This restriction holds for virtually all categories with respect to all weighting criteria.

However, if one considers the sequence created with regard to the approach stipulated above, it seems appropriate: the progression from categories with little relation to content towards those with stronger content references, as well as the progression from smaller amounts of text and context towards larger ones has all in all been accomplished. The resulting sequence could therefore be used in my study. Moreover, it is probably generally an adequate approach for an analysis of narratological phenomena with a potentially broad focus.

3. The Analyzed Corpus of Conflict Narratives

The analysis of the narrative constitution of conflicts was performed on a corpus of conflict narratives. These narratives are transcripts of telephone interviews conducted with 17 people who, at the time of the interview, were experiencing a conflict with one or more other persons at their workplace. The interviews were based on the method of narrative interviews by Schütze (1977) and were conducted in accordance to a semi-structured interview guide. After the narration of the current conflict in the first part of the interview, the interviewees were asked in the second part to narrate a situation without conflicts, possibly with the same person or persons.

A total of 18 conflict narratives and 15 narratives about a situation without a conflict was collected. One person did two interviews about different conflicts, and some interviewees failed to narrate a situation without a conflict – mainly because they repeatedly returned to the narration of the conflict situation instead of narrating a situation without conflict. Moreover, six interviewees consented to a follow up interview after one or two years, in which we talked again about the then resolved conflicts. The analyzed corpus thus consisted of 39 narratives with a total of about 52,000 words (transcribed from audio recordings with a length of 6 hours and 13 minutes) with an average of approximately 2,800 words / narrator.

Table 2 provides an overview of some features of the narrated conflicts in accordance with Glasl (2011). An additional description of the conflict issues can be found in Gius (2015, 57-65). The transcripts of the interviews are available online,3 as well as the annotated corpus, which can be accessed on request within the free text annotation platform CATMA.

Table 2: Overview of main features of the narrated conflicts (narrator names: pseudonyms)

The annotation of the 39 narratives resulted in about 28,000 annotations, of which 20,000 pertained to narratological categories.4 The remainder consisted of tags that were used for the subsequent evaluation (e.g., tags for the degree of escalation etc.). These annotations were then evaluated with regard to relations between the presence or absence of a distinguished narratological phenomenon and the degree of escalation in the narrative.

4. Findings on the Narrative Constitution of Conflict Narratives

The analysis of the annotated material brought to light more than 60 interrelations between narratologically definable phenomena of narration and the presence or absence of conflicts in the relevant narratives. The comprehensive list of interrelations between narratological phenomena and features of (narratives about) conflicts can be found in Gius (2015, 317-328). Most of these interrelations are of the form

If phenomenon X is present / absent / dominant in comparison to related phenomena, the narrative is about an acute conflict / a resolved conflict / no conflict.

The interrelations are in principle of two different types. The first type establishes a correlation between the linear or inverse-linear progression of one or more phenomenon complexes and the degree of escalation. For example, the amount of partial analepses increases with the degree of escalation of the conflict. Some of the interrelations of this type only affect the escalation in conflict situations and can be used to make graduated distinctions from acute through to resolved conflicts. Time expressions, for example, are predominantly deictic in the case of acute conflicts while in dissolved conflicts they are mostly anaphoric. A simplified overview of narratological phenomena that increase or decrease with the further escalation of conflicts is outlined in table 3 below.

Table 3: Narratological phenomena and degree of escalation

The second type of interrelation concerns not a progression, but a distribution of phenomena that is typical for a specific conflict narrative type. This indicates that the narrative type in question is particularly marked by a phenomenon, while the other two narrative types differ little or not at all. In this case, the phenomena themselves do not function as indicators of escalation, but their configuration is discriminatory for a certain narrative type. For example, even though for all narrative types the overall proportion of prospective narration, i.e., the narration of events which from the narrator’s temporal standpoint occur in the future, is smaller than the retrospective one, a relatively high degree of prospective narration is typical for narratives without conflicts. Narratives about resolved conflicts, on the other hand, contain a comparably high number of anachronies (in Genette’s terminology) and are thus less chronological than narratives about acute conflicts and narratives about situations without conflicts.

Some of the discriminatory phenomena can also be applied without comparing the narrative in question with the other two types of narrative. This holds for example for conflict-free narratives where – in contrast to the other two types – speech representation outweighs the representation of mental processes. Table 4 provides an overview of the narrative configuration of the three types of (no) conflict narratives, both for absolute indicators and comparative indicators.

Table 4: Narrative configuration of narratives without conflicts, about no longer acute
and about acute conflicts

In addition to the findings about the narrative constitution of narratives about conflicts, a closer look at the annotated passages revealed that some of the outcomes do not concern any kind of relation, but rather seem to mark points of the narratives where information of major relevance for the understanding – and thus for the resolution – of the narrated conflict is available. The comprehensive list of content markers is available in Gius (2015, 328-330). Among them are the following (together with a short description of their possible relevance for conflict management):

  • Tense switch: passages with a change of the grammatical tense and no change in the semantic tense, i.e., passages in which the tense switch is not motivated by temporal aspects of the narrated content, may refer to central issues for the narrator in the context of the conflict.
  • Negative interpersonal characterizations: the content of interpersonal characterizations can be used for the assessment of the conflictual nature of a situation (i.e., its degree of escalation) since negative contents are mitigated following the resolution of the conflict.
  • Collapse of the figural and narratorial functions of the narrator: passages in which the two functions of the narrator – namely the narrating and the narrated I, i.e., the function of the narrator as narrator or as character – may contain references to the perception of the narrator of the conflictual nature of the narrated situation.
  • Repetitive passages: repetitive passages always refer to something that the narrator is concerned about. In some cases these can also be positive elements, but mostly the passages contain negative events that are central to the conflict. The repetitive passages particularly emphasize those aspects from which an interpretation of the overall situation and assumptions about the central issue from the perspective of the narrators can be inferred. Additionally, the last repetition in a narrative seems to correspond most closely to the actual perception of the narrator.
  • Temporal perspective: passages containing a temporal perspective may include references to the evolution of the conflict and therefore help to assess the narrated conflict.

The majority of these phenomena can be analyzed even during oral ad hoc conversation and can therefore easily be used as an additional heuristics during conflict management.

5. Theoretical Findings: Interdependencies of Narratological Concepts

From a theoretical or methodological point of view, the broad application of narratological concepts and especially their operationalization not only yielded the adaptation of concepts and the weighting of their respective narratological fields described above; it also brought to light hitherto unnoticed interdependencies between narratological concepts. The reason for this is the already discussed prevailing focus on selected phenomena in one or several narratives: only a broad application of narratological concepts yields a comprehensive discussion of concepts that, in turn, is fostered by its application to a certain number of narratives. Even though some of these interdependencies might be felt as self-evident, they are rarely discussed in narrative theory. For example, it seems obvious that the narratological field of time, containing the subfields ‘time expressions’ and ‘tense’, is related to all other fields for which time is relevant, namely the fields ‘time of narration’, ‘temporal relationship between histoire and discours’ and ‘temporal perspective’. The field ‘space’ is analogously a basis for determining the spatial perspective. Additionally, time and space are relevant for the analysis of almost any other phenomenon field because all of them rely more or less explicitly on the determination of space and time within – and to an extent even outside – the narrative.

But there are many more connections than these two, most of them less obvious and thus even more meriting of further consideration. The following interrelations emerged during the methodological research and / or the analysis in the course of my study (cf. figure 2 that illustrates all interdependencies – with exception of narrative levels since they are an overall interdependency):

  • Narrative levels: the determination of narrative levels is a prerequisite for the determination of any other phenomenon field. For some phenomena, narrative levels do not necessarily need to be analyzed in advance, but the evaluation of the analysis of these phenomena must be based on narrative levels. This applies for example to order. In some cases – such as in the representation of narrators or addressees, the segmentation of the narrative, etc. – the identification of narrative levels is, however, also a prerequisite for the analysis.

  • Narrative levels and speech representation: all additional narrative levels in the analyzed narratives were representations of speech, and it seems plausible that most speech representations are separate narrative levels.

  • Duration and other phenomena: the pace of narration can be inferred from the following phenomena:

    • direct speech is isochronic, i.e., the time of narration and the time of the narrated correspond,
    • iterative passages are summaries, and
    • passages with present functions and metanarrative elements are scenes, i.e., they slow down the act of narration.
  • Repetitive telling and temporal perspective: repetitive passages may point to a temporal perspective of the perspectivizing instance (i.e. the narrator or a character).

  • Metanarrative elements and narrator: for the analysis of metanarrative elements, the determination of the uttering instance (i.e., the narrator, or, in case of a homodiegetic narrator, the narrating I) is a prerequisite. Therefore, an anticipation of the analysis of characters and narrators is to a certain extent necessary.

  • Means and types of metanarration: in the analyzed narratives, there was also a consistent correlation between metanarrative means and metanarrative types, i.e., every occurrence of a type was realized with the same means, all three types of metanarration being connected to different means.

  • Speech / mental processes and actions: speech as well as mental processes of characters can also be considered actions of these characters.

  • Speech / mental processes, character representation and narrator position: the determination of the narrator position was not possible due to a lack of text inherent features for the analysis. Therefore it was operationalized by a combined analysis of character representation and representation of speech and mental processes. This approach is also sensible in other contexts of narratological analyses.

  • Character analysis and representation of the narrator: in principle, the categories of character analysis can also be used for the analysis of the representation of the narrator. For homodiegetic narratives, an additional distinction between the narratorial and figural function of the narrator, i.e., the narrating I and the narrated I, needs to be made.

  • Metanarrative means and representation of the narrator: in principle, all metanarrative passages of a text are relevant for the analysis of the representation of the narrator. This applies particularly to the self-description of the narrator.

  • Character analysis and representation of the addressee: analogously to the analysis of the representation of the narrator described above, the categories of character analysis can be used for the analysis of the representation of the addressee.

  • Present functions and narrator and addressee: passages with present functions are always – at least implicitly – indicative of the narrator and the addressee of the passage.

  • Mental processes and knowledge: mental processes are relevant for information allocation because they also represent knowledge of the instance that originates the mental process.

Fig. 2: Interdependencies of narratological concepts (the hierarchy of the nodes corresponds to the tag set ordering, with solely concepts featuring interdependencies being depicted; the uppermost, colored nodes are the respective fields, concepts of the field are bordered in the same color)

These interdependencies need to be discussed in greater detail. The detection and examination of the interdependencies could lead to a reduced narratological framework that relies on fewer concepts, but has the same analytical power. This is not only a theoretical desideratum: a (more) consistent conceptual framework also is of eminent importance for computer-aided, and especially for automated approaches to narrative analysis.

6. Conclusion

In summary, the study described here has yielded results for the analysis of conflict narratives, developed a methodology for a comprehensive analysis of narratives, and detected a series of theoretically significant relations between narratological concepts. The outcomes point to a systematic connection in narratives about conflicts between narrative properties and certain qualities of the conflict itself, such as the degree of escalation. Even though the analysis was performed on only a rather small corpus within the domain of labor situations, the findings can function as a point of departure for further investigation. For such a purpose, the analysis of additional narratives should prepare the groundwork for the development of a model for analysis that can be applied within conflict management.

Since the approach adopted was empirical and inductive, it deviates from conventional narratological approaches. This required a methodological enhancement of narratological analysis and thereby fostered two additional outcomes that should be considered within narrative theory. First, the classification of narratological fields according to their complexity can also be deployed for other analyses. Second, the interdependencies of concepts within the narratological fields should be taken into account by narrative theory. A more comprehensive view on narratological concepts might reveal redundancies and thus lead towards a more consistent theory of narrative. Interestingly, while these two aspects have been facilitated by a decisively empirical approach, their ramifications are relevant for theoretical reflection. This questions the opposition of empirical vs. theoretical approaches in an unexpected way.

Accordingly, this work is an example of a case where “the unprecedented empirical power of digital tools and archives offers a unique chance to rethink the categories of literary study” (Moretti 2013, 119). Both the methodological insights and the study itself were made possible only with the adoption of a computer-aided approach – and they reach far beyond the digital realm.


Bögel, Thomas et al. (2015): Collaborative Text Annotation Meets Machine Learning: heureCLÉA, a Digital Heuristic of Narrative”. In: DHCommons Journal 1. URL: http://dhcommons.org/journal/issue-1/collaborative-text-annotation-meets-machine-learning-heurecl%C3%A9-digital-heuristic (12.02.2016).

Genette, Gérard (1972): “Discours du récit. In: Gérard Genette (Eds.), Figures III. Paris, pp. 67-282.

Gius, Evelyn (2015): Erzählen über Konflikte: Ein Beitrag zur digitalen Narratologie. Berlin et al.

Glasl, Friedrich (2011) Konfliktmanagement: Ein Handbuch für Führungskräfte, Beraterinnen und Berater. Bern.

Hühn, Peter et al. (2015) (Eds.): the living handbook of narratology. Hamburg. URL: http://www.lhn.uni-hamburg.de/ (12.02.2016).

Jesch, Tatjana et al. (2006): “Patientenerzählungen wie Literatur verstehen: Vom Nutzen der Narratologie für die psychodiagnostische Hermeneutik”. In: Vera Luif et al (Eds.), Qualitative Forschung im klinischen psychotherapeutischen und psychoanalytischen Kontext. Lengerich, pp. 39-65.

Lahn, Silke / Meister, Jan Christoph (2013): Einführung in die Erzähltextanalyse. Stuttgart.

Lotman, Jurij M. (1977): The Structure of the Artistic Text. Ann Arbor.

Margolin, Uri (2013): “Narrator”. In: Peter Hühn et al. (Eds.), the living handbook of narratology. Hamburg. URL: http://www.lhn.uni-hamburg.de/article/narrator (12.02.2016)

Meister, Jan Christoph (2003): Computing Action: A Narratological Approach. Berlin et al.

Moretti, Franco (2013): Distant Reading. London, New York.

Niederhoff, Burkhard (2011): “Perspective – Point of View. In: Peter Hühn et al. (Eds.), the living handbook of narratology. Hamburg. URL: http://www.lhn.uni-hamburg.de/article/perspective-%E2%80%93-point-view (12.02.2016).

Ryan, Marie-Laure (2012): “Space. In: Peter Hühn et al. (Eds.), the living handbook of narratology. Hamburg. URL: http://www.lhn.uni-hamburg.de/article/space (12.02.2016).

Schütze, Fritz (1977): Die Technik des narrativen Interviews in Interaktionsfeldstudien – dargestellt an einem Projekt zur Erforschung von kommunalen Machtstrukturen. Bielefeld.

Dr. Evelyn Gius
Department for German Studies
University of Hamburg

Please do not cite the HTML version but only the PDF file:

URN: urn:nbn:de:hbz:468-20160607-153200-2

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

1 The study presented here was originally realized between 2008 and 2013 as my dissertation project and was published in German in 2015 in the Narratologia series by De Gruyter (cf. Gius 2015). This article summarizes some of the major aspects of the dissertation and thus makes my work available in English. Additionally, the article focuses on methodological aspects of my approach that, even though present, are not emphasized in the publication of the dissertation.

4 Since the tag set amounts to a total of 500 concepts, a complete analysis could be performed only exemplarily on three narratives. For the remainder of the corpus the focus was set on two narratological dimensions that have proven particularly relevant in preliminary studies: phenomena of time, and the representation of speech and mental processes. The related four phenomenon fields (e.g., time, relation narrator–event time, time relation discours–histoire and modes of representation) comprise about a quarter of the narratological concepts in the tag set.