Previous Section Contents Page Next Section



2) On narratology


To raise the question of the nature of narrative is to invite reflection on the very nature of culture and, possibly, even on the nature of humanity itself. So natural is the impulse to narrate, so inevitable is the form of narrative for any report of the way things really happened, that narrativity could appear problematical only in a culture in which it was absent - absent or, as in some domains of contemporary Western intellectual and artistic culture, programmatically refused.1

Having explored some of the theoretical discourses with which narratology has interacted over the last century, I may now turn to the question of what is ‘narrative’. In this chapter I will review some of the more important attempts to define it (in both the literary and visual fields) and to elucidate the structures that underlie the creating, telling and knowing of ‘stories’. Although the sections that constitute this chapter may seem a little disparate in places, the case studies that follow later in the text may serve not only to bring the various theoretical facets into sharper focus but will also demonstrate how these different elements of theory fit together into what is, if not a coherent model then at least an internally consistent meta-language with which to discuss narrative art.

2.a) narratives and narrativity - towards a definition

What then is a narrative? One of the axioms of narratology is that an absolute division of texts or images into categories of ‘narrative’ and ‘non-narrative’ is generally undesirable and probably unfeasible. The quality of ‘being narrative’ is more a Platonic idea than a binary characteristic. Moreover, since ‘being narrative’ is a phenomenon that derives as much from the reception of the object as from any intrinsic property of the object itself, it is difficult to conceive of any informational structure from which the receiver could not imply some degree of narrative, however limited. Even a list of names in a telephone directory carries with it a hint of biography, whilst as Nelson Goodman once observed:

... [the] description or depiction of a momentary and static situation implies something of what went before and what will come after. A picture of a forest tells implicitly of trees growing from seedlings and shedding leaves, while a picture of a house implies that trees were cut for it and that the roof will soon leak.2

[32] Clearly however ‘...some narratives are more narrative than others’.3  That is to say, some texts or images are innately more likely to be perceived as narratives by the typical reader/viewer. This degree of being narrative is generally referred to as the narrativity of a text/image.4

Narrativity is therefore a scale of sorts with, at one (rather extreme) end, telephone directories, Byzantine icons and hieratic statues, and at the other end, more conventionally understood narratives such as carefully plotted novels, epic poems, comic strips and Passion cycles. For such a scale of narrativity to be useful, it need not necessarily be a precise and deterministic measuring tool. One should however be able at the very least to define those characteristics of a text or image that would determine its relative position along the spectrum of narrativity; a task which is inseparable from that of explaining the phenomenon of narrative itself. Starting theoretically with Aristotle but more realistically with Propp, several authors have attempted to do just that.5   Unfortunately most of these efforts were hampered to some degree by a tendency to define narrative in terms of whatever genre, period or medium was most of interest to the author (be it epic poetry, Russian folk tales, short stories or the novel, in either its early or postmodern manifestations).

Some counterbalance to this narrowing of focus was provided by the Structuralist approaches of  Barthes and Genette, who worked in a more generic (or trans-genre) fashion. 6  Yet even these authors tended to be quite restrictive in the realms from which they chose their examples.  To construct usable definitions of narrative and narrativity which are truly transmedial, one must therefore cobble together elements from several different authorities.

The key elements that most definitions of narrativity share, and the model which I shall use throughout this thesis can be summarised  as follows:

1)         Time - perhaps the most fundamental characteristic of narrative is that it is diachronic, involving observations of, or references to, at least two chronologically distinct moments. As Jerome Bruner put it, ‘a narrative is an account of events taking place over time. It is irreducibly durative’.7  Thus in the story of Christ’s incarnation for example, the Nativity takes place nine months after the Annunciation and 33 1/3 years before the Crucifixion. The nature of time was [33] of great interest to medieval philosophers. Augustine in particular was fascinated by the problems of being in, and being aware of, time - Book XI of his Confessions deals with nothing else. Generally though the key distinction for medieval theologians was between those things which were eternal or ‘outside time’ and those transient things - the stuff of narrative - which could be situated somewhere on the linear thread running from Creation to Eschaton. However there was also a third kind of temporality that concerned medieval man and that was cyclical time - the circulum anni. For most ordinary people this was the ‘time’ that mattered most - the passing of the seasons and the agricultural tasks that went with them - a cycle carved into the portals of many cathedrals and illumined in the calendars that start many Psalters, either as personifications of the months, as depictions of the labours associated with those months or as the signs of the zodiac.8

2)         Change/eventfulness - some significant aspect of the story-world must have changed over time, or at the very least, something noteworthy has to happen. Christ is born, the Holy Family leave for Egypt, Herod orders the slaying of the children, etc. If time passes but nothing changes there is no narrative.

3)         Causality and emplotment - the changes or events that are described can be observed as happening as a consequence of other events or actions. Simply listing events in chronological sequence does not a satisfactory story make - it is the weaving of diachronic events into a causal chain or plot that makes them narrative.  For Hayden White, in his classic study of the development and structures of historiography,  it was precisely this process of ‘emplotment’ that marked the crucial distinction between the early medieval writing of annals (simple chronological lists of what happened) and the later writing of ‘history’ (the weaving of events into a causal chain). 9  Christ is born and because Herod has been warned about the birth of the King of the Jews, he therefore orders the Massacre of the Innocents. This aspect of narrative is also sometime referred to as teleology or orientation since it implies that there is something purposive in the story - that the ‘author’ knows where he/she is going.  
As has often been noted, the will to emplotment is so fundamental to how we make sense of the world that people often imply a causal relationship between two events, even when there is none, solely on the basis that event B is described as coming after event A, resulting in the so-called ‘post hoc ergo propter hoc fallacy’ (‘after [X] therefore because of [X]’). Clearly this phenomenon has important implications for visual narratives, particularly when scenes are arranged in sequence. [34] If the preceding three characteristics are the most fundamental, one can also identify plenty of other features that most narratives share.10

4)         Particularity - each narrative describes a particular series of events, rather than a general pattern. Those events may be typical or emblematic of more generic structures  but there is something specific and worthy of note about this particular occurrence of them. Thus a generic farmer cutting corn as part of a ‘labours of the months’ series is inherently less narrative than the very particular farmer cutting corn who is quizzed by Herod’s officers in the ‘Miracle of the Corn’ legend that became part of the story of the Flight into Egypt.

5)         Intentional state entailment - the characters within a narrative do not act at random but have a particular set of intentions, based on beliefs, values, desires, etc, which cannot be used to determine the course of events, but which can be used to interpret why things happened or why characters acted in some particular way.11 Thus Herod and Pontius Pilate each have a vested interest in protecting their respective political positions, and their actions, as reported in the Gospels, can be understood in terms of a desire to maintain the status quo.

6)         Referentiality - the story must in some way relate to, or reference, the reality experienced by the recipient. Clearly in the case of fantastical or surreal genres of literature, or with mystical theological texts like the Apocalypse, this referentiality need not be a precise reflection of experienced reality. Instead the receiver creates references between the fictional world and the real by means of which the unfamiliar becomes associated with (refers to) the familiar. This characteristic is related to the so-called ‘possible worlds’ theory discussed by Marie-Laure Ryan, Lubomir Doležel and Umberto Eco, amongst others, in which the world in which any narrative takes place can be construed as an ontologically complete entity, i.e. a world with its own internally consistent logical rules (and which can therefore, potentially, be related to the world experienced by the reader).12   Thus in the case of St John’s eschatological vision, the unimaginable horrors of the Apocalypse are imagined by reference to more familiar horrors (plague, pestilence, blight, contemporary armies, routiers,etc), while descriptions of visits to heaven or hell are described (and visualised) in terms of luxurious palaces or sooty furnaces

[35] As well as distinguishing the characteristics of narrative, one can also consider the various building blocks from which narratives are constructed, and how these components are related. A simple hierarchy of such elements was proposed by Seymour Chatman:13

Figure 3 - Seymour Chatman’s model for the elements of a narrative

The lower half of Figure 3 is straightforward. Stories deal with ‘existents’ (i.e. things which exist within the story-world) -  primarily made up of  ‘characters’ who are typically human (or if not human then at least anthropomorphised to such an extent that one can ascribe to them some degree of intentionality) and the spatial and temporal ‘settings’ within which they operate (here I mean spatial and temporal settings in their simplest physical senses - e.g. the Garden of Eden immediately after Adam bit into the Apple, Judea in the year zero, Patmos in 65AD, etc). The characters interact with each other and with the world around them in ‘events’ - either ones they initiate (‘actions’) or ones which someone or something else initiates (‘happenings’).

Less obvious but altogether more important is the higher-level distinction between ‘story’ and ‘discourse’ - a dualism which lies at the heart of all structuralist approaches to narrative. The distinction between fabula (story) and sjužet (discourse) was first drawn by the Russian Formalist, Viktor Šklovskij in his influential 1921 essay on Lawrence Sterne’s classic anti-novel ‘Tristram Shandy’ .14

[36] The ‘fabula’ is something of an abstract concept but is best conceived as the underlying raw materials of the story devoid of any act of ‘interpretation’ - i.e. what actually happened, in chronological sequence. A ‘discourse’ in this specific sense (devoid of any Foucauldian baggage) is a particular ‘telling’ or version of the story which has been configured from some underlying fabula through a mediating process of authorship or interpretation. A single fabula may be expressed in any number of discourses corresponding to different ways of telling the same underlying story. In these terms, narrativization  can be regarded as a kind of mapping process by which the events and existents of a fabula are selected and configured into a particular discourse.15

Although the concepts underlying the fabula/sjužet (or story/discourse) dichotomy are relatively simple, they have been problematized by three factors. Firstly there is the question of terminology, since narratologists (particularly in the English-speaking world) seem unable to agree on the particular binary pair of terms that should be used. Šklovskij himself paired the Latin word ‘fabula’, (literally meaning ‘story’ or ‘fable’ but traditionally used in relation to classical theatre) with the Russian word сюжет (or ‘sjužet’ - though it is also variously transliterated as ‘sjuzet’, ‘suzjet’, ‘sjuzhet’ and ‘syuzhet’) whose literal meanings include ‘subject’, ‘topic’ and (unfortunately) ‘plot’. The problem here is that the English word ‘plot’ has many common usages, some of which overlap too closely for comfort with its supposed antonym ‘fabula’. After Tzvetan Todorov, writers in the Francophone tradition drew an almost identical distinction between ‘histoire’ (or for Gérard Genette, ‘récit’)and ‘discours’, on the basis of which many English authors like Seymour Chatman have chosen to use ‘story/discourse’ as equivalent terms to ‘fabula/sjužet’.16 The problem again is that the English word ‘story’ is simply too imprecise, too broad in its range of possible meanings, to serve as part of a binary pair with ‘discourse’. As evidence of the confusion this can cause, I would point out that in both the table of contents and the introduction of an otherwise excellent collection of introductory texts on narratology, Susana Onega and Jose Angel Landa, chose as their binary pair the words ‘fabula’ and ’story’, thus reversing the normal role of the latter term.17 There is no simple solution to this terminological hotchpotch. On rather arbitrary grounds I have chosen to use  the terms ‘fabula’ and ‘discourse’ to distinguish between the underlying events and their manifestation in some particular telling.  

[37] It is important to stress that the word ‘fabula’, despite its obvious etymological links to ‘fable’ and ‘fabulous’,  is devoid of value judgements and makes no assumptions about whether the underlying story is meant to be construed as historically true or as fictional.

The second problem with the fabula/discourse distinction concerns whether the concept of the fabula has any validity or utility in relation to fictional works. Indeed it has been questioned whether the concept of an underlying, receiver-independent fabula has any validity in non-fictional or historiographic narratives either.  How, such an argument goes, can we talk about the underlying facts of a story when that story-world itself may be the invention of an author or is otherwise unknowable to us? The structuralist response to this challenge is to assert that given any specific telling of a story (which may contain flash-backs, prophecies and other deviations from natural chronological sequence, changing points of view, free indirect discourse and all manner of other narratorial trickery), it must always be possible to postulate some abstract and hypothetical sequence of events, conditions and happenings such that the given discourse is a particular telling of it.18 The existence of such an underlying fabula is foregrounded in various art-forms. The ‘multiple-tellings’ story model is an established structural topos in both literature and the cinema, such as in Akira Kurosawa’s film Rashomon, in which the details of a particular event (in this case a rape) are retold by each of the protagonists in turn. One could also point to the common publishing phenomenon of new authors extending fictional story-worlds created by now-dead ones (with or without the approval of the latter’s heirs) as an indication of how natural such an exercise is.19 More practically, Emma Kafelenos has deflected some of the objections about the fabula by redefining it simply as ‘a construct that readers make from a sjuzhet’, thereby ‘...shifting debate from problematic concepts including the intent and identity of the author to epistemological issues that can fruitfully be pursued’.20

Ultimately however the best response to this challenge is not to ask whether the distinction is ‘right’ but whether it is useful. This brings me to the third objection that has been raised in the fabula/discourse debate, namely the question of whether it matters - does the concept of an underlying fabula distinct from its specific manifestations as discourses add anything to our understanding of narratives - or is it just theoria pro gratia theoriae?  This question (which always belongs at the front of the art historian’s vade-mecum when venturing into the bosky groves of ‘Theory’) is perhaps more easily answered in relation to medieval Christian narrative art than for any other period or genre, since the definitive texts of Christianity demonstrate the [38] fabula/discourse distinction perfectly. The four Gospels each tell the same story but in different ways as the four Evangelists emphasise and elaborate different parts of the underlying fabula. They include or omit particular events or characters, presenting them as more or less important or disagreeing about where things happened and in what order. However, even when the evangelists differ on a substantive point, it is still assumed that they are describing the same underlying events and the apologetics could usually be relied upon to explain the apparent disparity. The importance of this or visual narratology is that while images may illustrate one specific Gospel text or another, more often than not they contain elements of two or more. Likewise artists and their advisers were able to interpolate details which were not mentioned by any of the evangelists but which were concordant with the assumed fabula (the ‘Miracle of the cornfield’ incident mentioned in the preceding section is a typical example). Moreover the existence of four ‘discourses’ created the possibility of generating a fifth, combining them into what was, in effect, an interpolated alternative window onto the shared underlying fabula. Such ‘harmonised’ Gospel texts appeared as early as the 2nd century AD (Tatian’s Diatesseron being the first and best-known) and remained a popular genre throughout the middle ages.21

The model of narrative discussed in this section has been concerned with what has been characterised as ‘surface structure’ -  fabulae and discourses are both concerned with the telling of specific or particular stories, populated by individual characters. An alternative approach to narrative, one more attuned to the linguistic model of Saussure, is the generalisation of stories and characters at the ‘deep structural’ level, such as we find in the analyses of Propp and Greimas - but this will be addressed shortly (chapter 2.d).

2.b) temporal ordering and pacing

2.b.1 - Anachrony

The annals and chronicles of proto-historiography, with their absolute minimal narrativity, illustrate a very particular way of arranging a sequence of events into a discourse; things are simply told in the chronological order that they happened. If we imagine eight arbitrary sequential events (consequential or otherwise) and label them A through H, then the annal will simply list them in the order A-B-C-D-E-F-G-H. In the real world however, this most innocent of ‘temporal orderings’, is rarely adhered to. Instead, most discourses flit back and forth along a hypothetical time-line representing the underlying chronology of the fabula, according to the needs of the story-teller. Gérard Genette devoted the lengthy opening chapter of his pioneering [39] study of narrative discourse to such questions of re-ordering, spawning in the process a string of neologisms with which to describe the various deviations from the annalist’s simple linear telling.22 Deviations from strict chronological order within the discourse he called ‘anachronies’ - and these he divided into two types, depending on whether the discourse was jumping forward or backwards in relation to the timeline of the fabula.

‘Analepsis’ is Genette’s more medium-neutral term for what in film is known as ‘flashback’ (where the discourse temporarily switches to an earlier point in the fabula time). Thus to return to the eight arbitrary events in our imaginary annal, if a discourse presented them in the sequence B-C-D-E-F-A-G-H then the account of event A would be analeptic, since chronologically speaking the event actually happened before  the events that precede it in the discourse. In real narratives such analepsis normally happens when the author needs to refer back to an earlier incident in order to explain or justify something that is happening at the present point in the discourse. There is of course no reason why the events described analeptically need be new elements in the discourse - the sequence A-B-C-D-E-F-A*-G-H is common in many literary genres (for example in revenge tragedies) when the retelling of an earlier event (A*) is used to explain the behaviour of characters at a later stage in the discourse (see also the discussion of ‘frequency’ below). The use of Analepsis dates back to some of the very earliest surviving examples of European literature; like most classical epics, Homer’s Iliad and Odyssey both start  in medias res - literally ‘into the middle of things’ - with the poet singing his invocation of past events, which are then recounted in various degrees of flashback.23

The other type of anachrony described by Genette was ‘prolepsis’ or ‘flash-forward’, where the discourse temporarily switches to a future point in the fabula time as a kind of narrative anticipation, or the allusion to some event that will take place later in the story (for example A-E*-B-C-D-E-F-G-H). Prolepsis tends to be more fleeting than analepsis, usually appearing as a teaser to build suspense and keep the reader interested, using phrases like ‘...little did I know what horrors awaited me...’, or in visual narratives as more subtle pointers to future events, such as the goldfinch held by Christ in assorted Madonna and Child paintings (since the bird nests in thorn bushes and has a red forehead it was a proleptic pointer to the Crown of Thorns).  Another common form of prolepsis is the use of dreams and prophecies as narrative devices - something which is particularly prominent in the Old Testament (some specific examples of which will be discussed in chapter 8b).  The prolepses of the Old Testament become the analepses of the Gospels (usually quoted verbatim), as the Evangelists constantly seek to explain the events of [40] Christ’s life in terms of earlier prophecies. This prolepsis-analepsis pairing (prophecy issued/ prophecy fulfilled) is used to validate the Gospels, drawing on Old Testament prophecy as proof of Christ’s divinity and serving as a modality marker for the New Testament discourse as a whole.24

There was another sense however in which prolepsis was quite central to medieval readings of the Bible. In effect the whole book was seen as an encrypted prolepsis - a divinely inspired crib that could help us understand God’s plan not just for the New Testament but also for the reader’s historical past and future too until the (for them very real and perhaps not too far off) anticipated end of time. The various commentaries on the Apocalypse (starting with the ninth century writings of Berengaudus but particularly in the work of twelfth and thirteenth century writers like Anselm of Havelburg and Alexander of Bremen) were in part an attempt to understand the timeline of God’s ‘universal fabula’ by unpicking the threads of St John’s discourse to calculate when exactly we could expect the ultimate narrative denouement to arrive.25

As well as distinguishing anachronies on the basis of the direction of the chronological ‘jump’, Genette also considered the duration and distance of that jump, or as he called it, the ‘extent’ of anachrony (how much time passes within the analeptic or proleptic section) and its ‘reach’ (how far into the past or future the anachrony jumps). The second of these issues is generally the more interesting, particularly in relation to whether the reach of the anachrony remains within the chronological scope of the narrative (internal anachrony) or goes beyond it (external anachrony). To borrow Genette’s example, Book XIX of the Odyssey (lines 394-466), concerning Ulysses’ scar, describes an event which happened decades before the point in the discourse where it appears (its reach), well before the earliest events described in the main narrative, (making it an ‘external analepsis’) and describes an event which took place over a few days (its extent).26 For the pedant, this idea of ‘reach’ is problematic in relation to Biblical narratives since (as Augustine discussed at length) Genesis begins with the creation of time and the Book of Revelation marks the end of it. Anachrony within the Bible as a whole is therefore always internal since the text encompasses the whole of narrative time. However, anachrony can be (and often is) external in relation to the individual books or between the New and Old Testaments.

[41] Another type of anachrony (or more accurately a narrative circumstance which necessarily gives rise to an analepsis) occurs when a discourse needs to describe two distinct strands of action which take place concurrently, or during overlapping periods of time in the fabula - usually in different places and/or involving different protagonists. The normal practice for coping with such parallelism in textual discourses is to describe one strand, then slip back (for example with a ‘meanwhile, back at the ranch...’-type framing expression) and describe the other. We find this for example in the Book of Tobit where two strands of action occur concurrently near the beginning of the story.  The discourse  from verse 2:20 (‘Whereby it came to pass that she received a young kid and brought it home...’) right though to 3:6 - Old Tobit’s lament (‘...for it is better for me to die than to live’) describes events in Nineveh which occur on a particular day at exactly the same time as the events which are happening far away as described in verse 3:7 (‘Now it happened on the same day, that Sara daughter of Raguel, in Rages a city of the Medes...’) through to Sara’s prayer which ends in verse 3:23 (‘Be thy name, O God of Israel, blessed for ever’). The parallelism here is important as it sets up the chain of events that eventually results in the happy joining of Tobit’s son to the daughter of his old friend Raguel - but because verbal narratives are limited to ‘single-threaded’ discourses, parallelism can only be expressed by the use of analepsis, coupled with suitable framing expressions. The ‘ending frame’ for parallel episodes is not always necessary, although it is sometimes used to explicitly indicate that the two story strands are connected, as in Tobit 3:24-25 which not only brings the two threads together but also gives a proleptic indication of the happy conclusion:

At that time the prayers of them both were heard [...]  And the holy angel of the Lord, Raphael was sent to heal them both, whose prayers at one time were rehearsed in the sight of the Lord.

In theory this kind of chronological parallelism should be easy to represent in visual narratives. Verbal texts are necessarily single-threaded since words are arranged in fixed sequences that can only tell one plot-line at a time. Polyscenic visual narratives, if physically arranged in two or more parallel series can in theory be ‘multi-threaded’ though in practice, the reading of them is not. The viewer can switch back and forth between two story-lines but still has to load the relevant one into their memory and cognitive processing apparatus in order to ‘read’ the scene. 27 The physical expression of the story may be multi-threaded but the reading of it is still single threaded. In this respect, the process of reading visual and textual narratives is actually much closer than G.E. Lessing realised. The narrative image is never absorbed and understood in an instantaneous flash of perception but is progressively unpicked as the eye (and the cognitive apparatus that drives it) scans the various elements of the image. The major difference is that [42] reading visual narratives (particularly polyscenic ones in anything other than a linear frieze arrangement) is always to some extent ergodic. Whereas the reader or auditor of a verbal text simply follows the words in the sequence the author provided them, the viewer of  narrative images is always required to direct the movement of the thread (in both the physical and narrative senses) on their own. All the designer of a polyscenic narrative can do is to provide clues.

The models of narrative anachrony described by Genette all assume that one is reading a text  for the first time. However, the narrative anticipation implicit in a proleptic episode changes its significance quite dramatically if one already knows the story. As Mary Carruthers has discussed at length, rote learning was routine for scholars and clerics in an age when the written word was an expensive commodity - and many members of any narrative’s audience would have been able to recite significant portions of a specific textual version from memory (in some cases, from any position and in any sequence).28 Most of the time therefore the default narrative experience of the medieval viewer or reader was one of re-experiencing; be it re-viewing, re-reading or re-hearing. All these are the kinds of activity which, as Barthes put it, ‘...draws the text out of its internal chronology (“this happens before or after that”) and recaptures a mythic time (without before or after)’.29 Rather than serving to arouse anxiety or suspense in an already-familiar story, prolepsis becomes a cue to meditation on the inevitability of some particular outcome. Re-reading becomes achronic, since the fabula time is already mapped onto the reader’s memory (something to keep in mind when considering the sometimes erratic ordering of narrative scenes, such as one often finds in twelfth or early thirteenth century stained glass).

Another difference between modern and medieval audiences in relation to anachronic narrative structures is that flexibility of ordering is in some senses innate to anyone whose schooling was primarily in Latin. John Pollini, in his account of the sometimes confusing arrangements of Roman visual narratives suggested that:

At a fundamental level, the cognitive processing of visual disiecta membra into a form of narrative would not have been difficult for a mind accustomed to the syntactic flexibility of the Latin language. This flexibility or ‘dystactic’ structure, is especially manifest in Latin poetry and rhetoric, in which words, phrases, and clauses are held in suspension until a thought is completed.30

Whilst not all of the audiences for medieval narrative art were this talented in Latin, those who devised the narrative programmes generally were.

[43] Narratives also differ in their ‘episodicity’ and hence in the extent to which anachrony is either possible or noteworthy. Some hagiographies (such as those of St Julien the hospitaller, St Eustace or St Thomas in India) effectively have a single narrative thread running through the whole story, which needs to be told in a particular order in order to make sense overall. Others (such as the lives of St Nicholas of Bari, or St Martin of Tours) were composed over an extended period as a loose assemblage of brief episodes or segments, which could be arranged in almost any sequence, or set in any location, without losing their narrative sense. They share their chronotope with the classical Greek romance described by Bakhtin:

All adventures in the Greek romance are thus governed by an interchangeability of space; what happens in Babylon could just as easily happen in Egypt or Byzantium and vice versa. Separate adventures, complete in themselves, are also interchangeable in time, for adventure-time leaves no defining traces and is therefore in essence reversible. The adventure chronotope is thus characterized by a technical, abstract connection between space and time, by the reversibility of moments in a temporal sequence, and by their interchangeaility in space.31

The countless good deeds of Nicholas or the exorcisms of Martin can be told in almost any order, and added to or omitted as required, without changing their overall stories, making them, in effect, ‘frame narratives’ (see chapter 3.a below). Obviously this was a major advantage for designers of visual narrative programmes since individual episodes could be reduced to one or two key scenes and slotted into a series wherever required (a practice which is particularly noticeable in the triforium/lower clerestory windows at Le Mans, where some stories appear in two windows in different configurations).32 Naturally however, this flexibility of sequencing means anachrony between such segments would be hard to detect.

2.b.2 - Achrony

An altogether more confusing deviation from chronological sequence is when the discourse seems to step away from the temporal axis altogether into a form which Genette described as ‘achrony’. Achronic narratives in literature typically result from one of two circumstances - either the discourse becomes so clogged up with descriptions of some detail of place or character that the narrative temporality is temporarily lost (as happens for example in Exodus chapters 25 to 31, where the description of the Tabernacle interrupts the discourse for so long that one might easily forget the Israelites waiting not-so-patiently at the foot of Mount Sinai) - or else the author deliberately changes to a non-chronological organisation of the story. Of the latter form, Genette gives the example of Proust’s occasional segments structured on the basis [44] of geography rather than time.33   Whilst some parts of the Bible are certainly achronic in the sense that they occur ‘outside time’, in the context of the eternal, this is something of a technicality which goes rather beyond the sense of achrony envisaged by most narratologists. More relevant here though are the many sections when the Old Testament switches from conventional narrative to genealogies. For example the whole of Genesis chapter 5 (‘This is the book of the generation of Adam...’) simply lists the lineage and longevities of the male line from Adam to Japheth - analeptically from verses 1-6, then proleptically from 7-32. Although there is clearly a temporal element in Genesis 5, it is effectively achronic, since the organising principle is not chronology but genealogy - it is the passing of genes, not the passing of time that matters here.

2.b.3 - Anisochrony

So far this discussion of the mapping of fabula-time into discourse-time has concentrated on the various ways of ordering events. There is however a second set of transformations that needs to be considered; those that relate to relative pacing, or anisochrony.34  No normal narrative discourse follows an exact correlation between the duration of events in the underlying fabula and the amount of space given to those events in the discourse (where, say, one month of story time always occupies one page of the text), although the American TV network Fox Broadcasting Company did precisely that with its action thriller series 24, in which each season consists of twenty-four one hour long episodes which together present the events of a single day in ‘real time’.35

In conventional narratives, authors devote more or less space in the discourse to a given amount of fabula time, according to how interesting or relevant the events happening during that period are to the author’s overall intentions - a process referred to as ‘anisochrony’ by Genette, as ‘speed’ by Gerald Prince and as ‘rhythm’ by Mieke Bal.36 A textual discourse may slow-down (each unit of fabula-time given more space in the text) or speed-up (less text per unit of fabula-time). Note that these terms are not related to the usual notion of ‘pace’ in a discourse - an [45] exciting and racy section of a story is still considered to be ‘slowed-down’  if it occupies more lines of the narrative text than would normally be given to an equivalent amount of elapsed story-time.

Genette described four distinct modes for the mapping of fabula-time to discourse-time, to which Gerald Prince added a fifth (stretch). These modes can be summarised as follows:


Mapping of Fabula Time (FT) to Discourse Time (DT);


Periods of FT are completely omitted from discourse


Periods of FT described in less DT than usual


Consistent match between FT and DT. Generally, actions take the same relative amount of time to describe as they took in the ‘reality’ of the fabula.


More DT than FT (i.e. it takes longer to describe an action than actually elapsed while it was happening). Like filmic slow-motion.


Passing of FT temporarily suspended to allow for description of a static setting or for a digression.

Table 1 - The five types of anisochrony

The ‘stretch’ mode is more characteristic of modern narrative forms and did not play much of a role in biblical or medieval narratives. ‘Scene’ in its strictest sense (1 minute of FT = 1 minute of DT) normally only occurs in mimetic dialogue within a dramatic performance.  More generally however scene is taken to be the ‘average’ or default relationship between FT and DT - the average number of lines or pages used to describe a standard unit of time in the underlying story. In other words, scene is the normal background pace of the narration.

‘Ellipsis’ is extremely common in both Old and New Testaments - for example the lives of Cain and Abel from the birth of the latter to the offerings of both (an unspecified number of years later) are elided with the conventional formula ‘And it came to pass after many days...’ (Genesis 4:3).37 Similarly, the Gospels almost entirely elide the first thirty odd years of Christ’s life after the return from Egypt (with the exception of Luke 2:41-52, which mainly deals with the twelve year old Christ’s unaccompanied visit to the Temple) - a gap partly filled for medieval audiences by various apocryphal legends about Christ’s childhood, and occasionally made visual in works like the Tring Tiles (British Museum).38

Broadly speaking, the default anisochronic mode for the Bible is a mixture of ellipsis, summary and pause (the latter sometimes as digressions but usually as the kind of genealogical achronies discussed above), though there are also periods where considerably more discourse time than [46] usual is devoted to a given amount of fabula time - which can thus be considered as stretch. Curiously, ‘scene’, construed as the default pace in most literary narratives is relatively unusual in the Bible, which tends instead to follow a quick-quick-slow narrative tempo.

2.b.4) Frequency

The final aspect of ordering considered by Genette was frequency - the number of times an event is repeated. As he explained, there are four possible relationships between the number of times something happens in the fabula and the number of times it is described in the discourse.39 Although Genette naturally coined neologisms to describe the various permutations of frequency, these are not always helpful. Instead I will adopt a simple pseudo-mathematical formula - in Table 2 below ‘1’=once, ‘n’=more than once, ‘F’= fabula and ‘D’ =discourse. So if something which happened only once in the fabula was described several times in the discourse, this would be expressed as 1F:nD.

Number of times event occurs in the fabula

Number of times event is described in the discourse

Form of relationship
(Genette’s terms in parenthesis)



1F:1D (singulative)



nF:nD (singulative)



1F:nD (repetitive)



nF:1D (iterative)

Table 2 - 'Frequency' (repetition in fabula and in discourse)

The two forms in which events are described multiple times in the discourse (i.e. nF:nD and1F:nD) correspond to a common device of classical rhetoric also widely used in the middle ages. As has often been mentioned in recent art-historical literature, repetition, or amplificatio, was widely taught in medieval ‘preceptive grammars’ (for example in the early thirteenth century Poetria Nova of Geoffroy of Vinsauf).40 As we will see in section 3.a (reiterative groups), repetitio was also commonly used in visual narratives.


2.c) focalization

‘Focalisation’ was the term adopted by Genette to denote

...the perspectival restriction and orientation of narrative information relative to somebody’s (usually a character’s) perception, imagination, knowledge or point of view.41

Or to put it another way, whose perspective on the events of the fabula is being presented in the discourse. Here ‘point of view’ does not relate to visual perspective, nor to issues of how perception is filtered by the opinions, beliefs or prejudices of the observer; it is simply a question of who would be in a position to know the details (at the fabula level) which are actually presented in the verbal or visual account of events supplied in the discourse.  Moreover, as Genette was at pains to point out, this question of ‘who sees?’ or ‘who perceives?’ (the focalizer) is not the same as asking ‘who speaks?’ (the narrator). An external/implicit narrator may present the story as perceived by one or more particular characters, or may adopt an omniscient point-of-view, presenting scenes, emotions and states of mind experienced by multiple characters. A classic literary demonstration of the practical effects of changing the focalization is Tom Stoppard’s 1966 play Rosencrantz and Guildenstern are Dead, in which events at Elsinore are told from the viewing point of two minor courtiers, rather than from the more usual perspective of  Hamlet himself.

With the exception of a few passages in The Acts of the Apostles (the so-called ‘we-passages’), biblical texts are written in what Genette called a ‘non-focalized’ mode (corresponding to an omniscient and implicit narrator) and present whatever information is relevant, regardless of which participants within the story could conceivably be aware of them.


Genette’s original model of focalization has undergone a number of revisions, both by Genette himself and by several other theorists. In particular Mieke Bal has proposed a far more complex model of focalization which has been adopted by various other authors. However, since her improvements are only really relevant to verbal narratives, they need not concern us here.42 Indeed it is probably safe to say that of the many ‘hot topics’ that exercise text-oriented narratologists, focalization is the one which holds the least relevance for this thesis. This is not to say however that it can be ignored completely. In her treatment of the Magdalene window at Chartres (bay 46 - see Fig. 2.01), Colette Manhes-Deremble observed that the Gospels treat the visit to Bethany entirely from the perspective of Christ and the disciples, yet the artists at Chartres have included three scenes relating to Lazarus’ death and funeral which predate [48] Christ’s arrival at Bethany (panes 5-7) and which are not described in the text.43 In other words, the window’s designers have re-focalized this part of the narrative away from Jesus and onto the Magdalene, presenting events though her eyes and inferring details from the fabula which are not found in the Gospels. The reasons for this shift are complex, though they may well relate to an interest in the final sacrament - including these scenes at Bethany allowed the artist to show the important Christian rite of burial (which is depicted anachronistically, with a Christian bishop aspersing the corpse and reading from an Ordo defunctorum held by a deacon), which by this stage was being actively promoted as one of the Seven Acts of Corporal Mercy (Fig. 2.01).

2.d) background to a structural model of narrative

Any discussion of narratology must at some point turn to Vladimir Propp and his pioneering 1928 book on The Morphology of the Russian Folk Tale, which may justly be considered the first ‘structuralist’ model of narrative (even if it predated ‘Structuralism’ by some decades). Although at that stage he was largely working on his own, Propp was heavily influenced by the emerging methodologies of the Russian Formalists, particularly the strand known as ‘organic formalism’, which recognised that just as each individual organism shares certain characteristics with other individuals of that species and (more distantly) genus, so individual texts share certain elements with other texts within a given hierarchy of literary form or genre.44

Propp took as his research material one hundred out of the 600+ traditional folk/fairy tales collected by the Russian folklorist Alexander Nikolayevich Afanasyev, which he analysed line by line to identify those elements which the stories had in common. These story elements he called ‘functions’, each function being ‘... an act of a character, defined from the point of view of its significance for the course of the action’.45 His conclusions regarding the nature of these functions merit quoting in full:46

The details of points 2 and 3 are particularly surprising. Propp concluded that in fact there are only thirty-one distinct functions to be found in all of the Russian folk tales he examined - and [49] that although not all of those are present in every tale, the ones that do appear will always appear in the same order. Some of these sequences are logically necessary, while others are necessary only because of the ‘rules’ of the genre. So for example function 2-An interdiction is addressed to the hero, must logically come before function 3-The interdiction is violated. In the folk-tale genre however, these two functions will always happen before 11-The hero leaves home, which itself will always come before 18-The villain is defeated (even though there is no logical reason why this sequence should not be reversed). In spite of what may appear from some of the subsequent commentary on his work, Propp never claimed that these thirty-one functions and their sequence were applicable to all or any other genres - simply that they were sufficient to represent the structure of all conventional Russian folk tales. In listing his thirty-one functions, Propp also identified seven distinct ‘roles’ responsible for enacting the various actions associated with those functions; the hero, the villain, the princess (or more generally, the sought-for person), the dispatcher, the donor, the helper and the false hero. Note that these roles are not equivalent to characters - an individual character within the story’s dramatis personae might act as donor in relation to one function and as villain in another.

Compared with the earlier discussion about the fabula/discourse distinction, Propp’s functions and roles are operating at an altogether deeper, more abstract level. Essentially what he was attempting to do was to identify the full set of atomic building blocks (functions) from which any fabula within a particular genre could be constructed - and then to show that the particular genre could be expressed in terms of a single generalised structure based on the sequence of those building blocks. Expressed in terms of some hypothetical narrative syntax, these deep-structural functions are purely syntagmatic categories - they are in effect place-holders which may or may not be filled but which, if present, always join together in a particular ordered sequence. The paradigmatic axis exists only in the surface structure of fabula and discourse - in terms of, for example, who the villain is or how and when they are defeated.

The Morphology of the Folk Tale remained relatively unknown in the west until its 1968 translation into English. It thus emerged into an academic world enthusiastically embracing Structuralism, a context in which Propp’s unabashed Formalism seemed distinctly old hat. Criticism of Propp (of which there has been a great deal) has come from two main directions. On one hand are those who rejected the Formalist method outright in favour of approaches that showed more respect for the ‘textuality’ and individuality of narratives - or else who looked for more deeper, more universal structures.47 On the other hand are those who broadly accepted Propp’s methodology but disagreed about the required number or scope of the functions.  Of the latter, perhaps the most important critic was the Lithuanian semiotician Algirdas Greimas, who [50] sought to make the model universally applicable to all narrative forms.  Whereas the set of general ‘roles’ identified by Propp were specific to the genre he was interested in (i.e. Russian folk tales), Greimas abstracted the model one stage further and identified a base set of six ‘actants’ (Subject, Object, Sender, Receiver, Helper, Opponent) which, he believed could account for all of the roles found in all different narrative genres.48

One of the problems with the approaches of both Propp and Greimas is that they seek to describe narrative solely in terms of agents and agency-effects (i.e. the causers of actions and the actions thus caused). The major difference between their approaches is that Propp starts with the agency-effect (the ‘function’) and derives from this the necessary agent (the ‘role’) whilst Greimas works the other way round, privileging the actant (or ‘type of agent’) as the basis of his genre theory. Whilst both approaches have their uses and have been fundamental to the development of Structuralist approaches to narratology, these ‘universal’ roles/actants are generally too deeply abstracted to be directly useful to the understanding of medieval narrative art. Moreover both of them exclude the role of the reader/viewer. In these models, the significance of a function for the course of the action is regarded as an innate property of the genre, not the product of a receiver’s interpretation.49 This is not to say however that deep-structural models don’t have their uses. One can for example postulate a more directly relevant set of roles, specific to the needs of biblical and hagiographic narratives, as I have attempted to do in Table 3.


Characters often expressing this role

Visual characteristics


Christ, Saints, Disciples



Disciples, Three Magi

Various ‘positive’ features


Prophets, Apostles, Longinus

Book or scroll, eye-gesture


Satan, Herod, assorted guards, executioners, heretics, etc

Various negative features, including Semitic facial types, partial undress, inelegant posture etc.51

Table 3 - Deep-structural 'roles' in Biblical and hagiographic narratives

These roles become particularly relevant to the present thesis when one considers them in terms of the semiotic clues employed to identify and distinguish them in visual narratives (third [51] column of the table). Such clues index the role, rather than signifying a specific character at the fabula level. Instead the latter generally relies on additional visual clues, such as a textual titulus, a personal attribute (e.g. St Laurence’s griddle, Stephaton’s sponge, Baalam’s ass), an action (Pilate’s hand-washing) or on other contextual information. Note also that characters can and do perform multiple roles within any given narrative - for example John the Baptist takes the roles of Witness (as a prophet), Helper (when baptising Christ) and Hero (in the story of his own martyrdom), while Christ, who would normally be associated with the Hero role also appears in various saintly narratives as Helper (arranging Thomas’ trip to India, bringing St Denis his last communion, etc.)

As well as considering role/actant-based models, deep-structural models based around the idea of the narrative function also have their practical uses. The obvious similarities in the lives of various martyr saints is an open invitation to some kind of Proppian analysis, employing such functions as Nativity, Conversion, Trial (test of faith), Imprisonment, Miracle (proof of sanctity), Martyrdom, etc. I will return to these matters later when discussing hagiographic topoi (chapter 2.i). Apart from any utility they have as analytical tools, the real attraction of these deep-structural models of the characters and actions that feature in Christian narratives is the fact that medieval exegetes, in their obsessive quest for precursors and ‘types’, also tended to discuss their narratives and characters in generalised terms. Thus all saints were seen as Alter Christus, persecutors were all ‘types’ of Satan and so on.

While Propp may have pioneered the structural analysis of narratives, it was Roland Barthes who developed it into a potentially useful methodology. His Introduction to the Structural Analysis of Narrative , which proposed the adopting of a linguistic model as the basis of narrative theory, might justly be considered one of the foundational charters of structuralism in general and structuralist narratology in particular.52  The Introduction...  was always seen by Barthes as a ‘work in progress’ and in the years that followed, his attitude to narrative, together with the associated analytical method, developed and shifted quite radically. These shifts culminated in the publication in 1970 of S/Z (his close textual analysis of Balzac’s Sarazine) - an event that signalled his turning away from the supposedly ‘closed’ analytical approach of structuralism towards the poststructuralist ideal of textual plurality (something he felt was being ignored or denied by the emerging field  of narratology). In 1969, while still on his journey from Introduction... to S/Z, Barthes also wrote a far less well-known paper (not published until 1971) [52] entitled The Structural Analysis of Narrative: A Propos Acts 10-11, which marked a further development of his original method.53

The following section will attempt to set out a hybrid approach to structural analysis, drawing upon elements of all three of Barthes’ contributions to the topic, as well as on some of the developments of Barthes’ scheme proposed by James Rushing in his study of Ywain imagery.54 By proposing yet another slight variant on the existing models for the analysis of narratives, I am not suggesting that the existing ones are wrong or deficient in any way - simply that they do not suit my requirements. One of the most frequent criticisms aimed at structuralist narratology is its supposed desire to define a single all-encompassing ‘Grand Unified Theory’ of narrative, applicable to all genres, periods and media. Roland Barthes, somewhat belatedly recognising the monster he had created, wrote about this in the opening paragraph of S/Z:

There are said to be certain Buddhists whose ascetic practices enable them to see a whole landscape in a bean. Precisely what the first analysts of narrative were attempting: to see all the world’s stories […] within a single structure: we shall, they thought, extract from each tale its model, then out of these models we shall make a great narrative structure, which we shall reapply (for verification) to any one narrative: a task as exhausting […] as it is ultimately undesirable, for the text thereby loses its difference.55

Such criticism is largely justified, and ultimately it was the impossibility of developing models that could fully account for all texts (including deliberately anti-narrative postmodern literature) that led many to a general dissatisfaction with structuralism. Fortunately I need not concern myself with such problems of universalism. All I require of an analytical methodology is that it should help to make sense of just three closely related ‘texts’; the Old Testament, the New Testament and Christian hagiography. Provided that this methodology can accommodate these narratives in any textual or visual form, I shall be content. This is perhaps one model of how structuralism can still be useful - not in the search for universals but as a loose tool-set or paradigm for investigating specific types of narratives.


2.e) a structural model of narrative

The structural model adopted in the remaining sections of this thesis can be defined thus:

  1. A narrative can be broken into sections, call them ‘functional units’, whose function is determined not by their literary or pictorial form but by what they contribute to the meaning of the narrative as a whole (for the supposed target audience).
  2. Narratives are constructed from four types of function; ‘nuclei’, ‘catalyses’, ‘indices’ and ‘informants’. The first two are elements of emplotment - they determine the direction and movement of the story-line. The other two contribute to the mood and meaning of the story but without changing the plot.
  3. Nuclei’ are the junctions or nexus points in a story.  These are the plot decision points where choosing one option or another will lead the overall story off in one or other direction.  Complete narratives normally begin with a nucleus and most will also end with one.
  4. Catalyses’ extend the narrative along its chronological axis but do not involve the possibility of branching off.  Catalyses often provide the filling-in of plot details and the extension of the story but without actually changing the plot.
  5. A series of nuclei and catalyses can constitute a ‘segment’ - effectively a meaningful and self-contained sequence of linked episodes or fragments forming part of a larger narrative, complete with a beginning and an end of its own. Nuclei can initiate or conclude segments but catalyses can only continue or extend an existing one.
  6. An ‘informant’ provides contextualising data about the temporal and spatial conditions of the story (‘It was a dark and stormy night...’ in a text or stylised trees indicating an exterior setting in a narrative image). They are the seasonings of narrative; adding richness to a plot without changing it.
  7. An ‘index’ is a pointer to meanings or interpretations beyond the immediate story level. Indices can be internal (e.g. proleptic or analeptic references to other parts of the story) or external (references to something outside the story). Recognition of an external index is culturally specific, being dependent on the reader/viewer’s familiarity with a complementary semiotic system that extends beyond the discourse.
  8. Nuclei and catalyses are primarily functions of the fabula (though of course they will also be reflected in the discourse) - they provide it with the change/eventfulness that [54] was discussed mentioned in chapter 2.a as a necessary condition of narrative. Indices and informants are primarily functions of the discourse level, contributing whatever is distinctive about different tellings of a story.
  9. Each functional unit can exercise one or more function - for example a catalysis can also be an index.

These various points will become clearer over the course of the case studies that come later - but for the meantime a quick example may help to begin the clarification (particularly of point 9) Take for example Genesis 22:6, where Abraham is preparing to comply with God’s command that he should sacrifice his only son:

And he took the wood for the holocaust, and laid it upon Isaac his son: and he himself carried in his hands fire and a sword...

Here there is no nucleus - Abraham’s decision to obey has already been taken in the preceding verses and this verse only continues the thread. Nor are there any informants in the text - the Bible is often short on details about settings (though this never stops artists adding them when appropriate). The two functional units of Isaac carrying the wood for his own funeral pyre and Abraham carrying the fire and sword are both catalyses, since they continue the plot without changing it. However the former is also an external index - a reference to Christ’s carrying of the Cross. Both the catalysis and the external index relationship are made explicit in panel 04 of the New Alliance window at Bourges (Fig. 2.02), where the continuation of the story leads into panel 05, while the external index relates to panel 06.

2.f) on visual narratives

Having considered what is meant by narrative in general, how it is to be measured and how it might be analysed, I am almost in a position to turn to the specifics of narrative images. Paradoxically it may be easiest to begin by considering what other types of images there are. After a brief assessment of ‘non-narrative’ images, I will discuss the existing typologies of narrative images, before proposing some criteria of my own.

2.f.1) Prelude - a typology of non-narrative images

In 1996, Gunther Kress and Theo Van Leeuwen published a highly original study of visual semiotics which included a simple categorisation of the various possible types of images (or ‘visual representational structures’, to use their splendidly inclusive term) that one might [55] encounter.56   With some minor adjustments, the categorisation they proposed can be expressed thus:

Figure 4 - A classification of types of images (after Kress & van Leeuwen, 1996)

Kress and Van Leeuwen presented this largely as an ‘either or’ classification of image types. I prefer to think of all these categories as representing the various different communicative or information storage functions a particular image is able to exercise, since images often participate in more than one category of informational system. Their sub-division of ‘Conceptual’ image types is a useful precursor for considering what constitutes visual narrativity because it helps to isolate, identify and classify those image functions which are essentially extra-narrative. As discussed above, all images can potentially be read as having a narrative element - extra-narrative in this context simply means those images that are more likely to be interpreted as exercising a communicative mode other than narrative.

The typology of ‘Conceptual images’ set out in Figure 4 may be better understood by considering some typical examples from medieval art. The ‘narrative’ image type we shall return to shortly but for the moment can be thought of simply as images which deal with the kind of diachronic processes of change discussed in chapter 2.a.  By contrast, conceptual images express stability and timelessness. For example the iconic, hieratic figures of saints that stand as jamb statues in early Gothic portals, or the angels and ancestors of Christ that appear in the [56] archivolts. Although most of the characters represented in these statues are also capable of participating in narrative scenes, when they appear in these contexts, it is their timeless and eternal character that is emphasised.

Conceptual images can be further divided according to the kinds of relationships they participate in. ‘Classificatory’ images (of which Figure 4 is itself an example) deal with taxonomies, or hierarchies of types, in which the defining relationship is the shared membership of some class or group. These taxonomies can be ‘overt’, with the hierarchical relationships between the members forming part of the representational structure itself, or they can be ‘covert’, in which case the relationship to a shared super-ordinate is not shown but is implied by the similarities between the subordinates. Common overt taxonomies in medieval art include charts of genealogy and consanguinity, diagrams of the Pseudo-Dionysian Celestial Hierarchy, the familiar ‘Tree of Jesse’ and images of the ‘Holy Kinship’ (see Fig. 2.03 for examples).  Covert taxonomies are even more familiar in medieval Christian art, which abounds in ‘sets’ of entities which belong to a particular category (or which, in the language of set theory, share an implied superordinate). Examples include the virtues and vices, the labours of the months, the signs of the zodiac, the twenty-four Elders of the Apocalypse, the twelve Apostles, the four Evangelists, the twelve minor and four major prophets, the seven deadly sins and seven acts of corporal mercy,  etc. The increasing complexity of Gothic portal sculptural programmes bred an insatiable appetite for such groupings. Covert taxonomies are limited to two hierarchical levels, in which the superordinate is the ‘general class’ and the subordinates are the members of that class. For example the class ‘Evangelist’ has the four subordinate members ‘Matthew’, ‘Mark’, ‘Luke’ and  ‘John’. When visualised, it is normally the subordinates that are depicted, rather than the superordinate class - one sees statues representing the individual Apostles but never a statue representing ‘apostleness’. Even the Wise and Foolish Virgins, though they derive from a parable (Matthew 25:1-13), generally appear in portal sculpture as covert taxonomies, stripped of their narrative context and instead standing as markers of liminality and as warnings of the imminence of judgement. Apart from the way they hold their lamps (and occasionally their exaggerated emotional responses), the only element of the parable retained by these statues is their location by the doorway.

If classificatory images deal with relationships in the form ‘[X] is a kind of [Y]’, analytical images describe ‘[X] is a part of [Y]’ relationships - rather like those engineering block diagrams that show how the various components of a system fit together. Such images are perhaps less common in medieval art, though there are still plenty of important examples, both in scientific and medical manuscripts and also in theological texts.  Mary Carruthers has written extensively about such images and their roles both as mnemotechic devices and also as tools for [57] expressing concepts.57 The example in Fig. 2.04A is a typical example, although arguably medieval maps and the portolan charts (Fig. 2.04B) that emerged in the late thirteenth century also belong in this category since it is the relationships between the places represented on them that matters. All of these images correspond to Charles Saunders Pierce’s notion of ‘diagrammatic iconicity’, since what matters for the image’s communicative effect is not an iconic resemblance of parts of the image to whatever they represent individually, but the fact that the relationships of those sign-components to each other resembles the relationships between their respective referents.

The final image function in Kress and Van Leeuwen’s model is the ‘symbolic’. This is the more familiar form of image, such as the iconic/hieratic jamb statues, in which the sculpture stands as a symbol for the saint thus represented (usually by means of a suitable attribute - an iconic sign functioning as a conventional sign). As with the Wise and Foolish Virgins, the characters represented by these statues may also appear in narratives - but there is usually a clear demarcation between their narrative and non-narrative roles.58

As mentioned earlier, these categories of image functions need not be exclusive. For example a jamb figure may individually constitute the symbolic representation of a saint or patriarch whilst also participating in a covert taxonomy along with its neighbours. In some thirteenth century portal programmes, such as central portal at Reims and the right portal at Amiens, jamb figures began to incline towards each other and participate in narrative groups, albeit in a rather restrained fashion.

2.f.2) Early models for classifying visual narratives - Wickhoff and Weitzmann

Various Enlightenment authors discussed the suitability of the visual arts for depicting narrative, most famously Gotthold Ephraim Lessing in his famous essay on the Laocoön. Such discussions invariably followed the paragone tradition; concerning themselves with the relative merits of text and image, and how the latter should be used, rather than how it actually had been used. As such they are of limited interest here. One of the first serious attempts to analyse narrative art as an historical phenomenon came with the 1895 publication of Die Wiener Genesis, by the Austrian art historian Franz Wickhoff, who was primarily interested in the origins and development of late antique art.59 Wickhoff distinguished three ‘modes’ by which distinct events within a literary text could be represented pictorially: [58]

Writing some fifty years later than Wickhoff, Kurt Weitzmann also considered the relationships between textual and visual narratives but based on a much broader range of material. Building in particular on the work of Carl Robert (which in many respects had prefigured that of Wickhoff by twenty, though the latter was apparently unaware of it), Weitzmann proposed a slightly different trio of narrative modes:

Although Weitzmann’s terminology has proved more widely influential and longer lasting than Wickhoff’s, the  problem with both their classifications of narrative image types is that each was developed in order to support the author’s own particular ‘grand narrative’ concerning the early evolution of narrative art. Although they differed over the details, both authors assumed a teleology in which artists developed their illustrative methods from the ‘primitive’ (because of its lack of respect for the Renaissance ideal of unity of time and space) complementary/ simultaneous mode, through the use of monoscenic and cyclic images in scrolls and on classical monuments, towards their adaptation for the codex format.

In reality however, all three modes of narrative depiction continued to coexist and thrive throughout the late classical and medieval periods and (for certain subjects) survived well beyond into the Renaissance, when a strict ‘unity of time and place’ in images was rather less common than is commonly supposed. The biggest problem with both Wickhoff’s and Weitzmann’s classifications is that in practice, for medieval narrative images at least, there is no clear binary distinction between the Simultaneous (Complementary) and Monoscenic (Isolating) modes - distinctions which seem to have been irrelevant to the producers and consumers of such images alike. As well as the more extreme cases like the Betrayal/Arrest of Christ, or Moses striking the rock of Horeb (where, as Schapiro pointed out, we normally see the striking of the rock, the surprised reaction of the Israelites and the resultant watering of the herds, all presented as if in a single momentary scene), a high proportion of narrative images contain elements which are internal indices of earlier or later moments of the story (visual prolepses and analepses) or else elements which, like Wickhoff’s complementary mode, are extra-narrative.

A classification based on whether or not key characters are depicted only once, or more than once, within the image field reflects an overly simplistic attitude to frames; should we see the repetition of Potiphar’s wife in panel 11 of the Auxerre west façade Joseph cycle (see chapter 6.e) as being ‘primitive’ because it ignores the supposed ‘ideal’ of unity of time, action and place - or should we recognise the narrative skill of an artist whose inventive and flexible approach to framing allowed him to create more dramatic effects than would be possible by following some arbitrary rule? Ultimately then, my reason for rejecting both these systems for classifying narrative images is that regardless of whether Wickhoff, Weitzmann or neither were correct in their teleologies of early Christian art, their terminology (and the assumptions that underlie it) had no relevance to the makers and consumers of narrative art in twelfth and thirteenth century Europe.


2.f.3)  An alternative approach to classifying visual narratives

Delimitation, classification, typology, it is all very nice as a remedy to chaos-anxiety, but what insights does it yield? [...] the pervasive taxonomical bend of narratology is epistemologically flawed.60

Whilst Mieke Bal’s belated realisation that the field of narratology has suffered from an excessive fondness for inventing categories (rather than using those categories for anything useful) is welcome, it should not be taken as an excuse to abandon such taxonomic activities once and for all. As an alternative to the simple tripartite classifications of Wickhoff and Weitzmann, I would instead suggest that narrative images should be considered in terms of their positions on five distinct axes. These axes are summarised in Table 4 and discussed at length in the pages that follow.




Possible/Typical values


The number of distinct images and/or episodes depicted as a recognisable group.

monoscenic, oligoscenic, polyscenic


The mechanism of narrative semiosis; how  the images relate to a supposed underlying story.

denarrated element, charade image, word image, pseudo-narrative, closely tied

Visual Narrativity

The apparent narrative ‘richness’ specific to the image (regardless of the narrativity of any underlying textual source or fabula)

high, average, low, minimal


For image cycles, the physical arrangement of individual frames; geometry/tessellation

grid, frieze, complex, irregular, banded


For image cycles, the trace followed between individual frames, in order to read them in narrative sequence.

left->right and top->bottom, boustrophedonic, irregular

Table 4- Criteria for categorising narratives images

My goal in proposing these criteria has been to devise a model that is as far as possible, purely descriptive (making no value judgements or assumptions about teleology) but which encompasses all the axes of variation that account for the narrative characteristics of images. With respect to Mieke Bal’s  warning quoted at the start of this section, the value of any taxonomy lies not in its ability to define, categorise and mould its objects into a closed system but in its propensity to suggest a series of questions that help to elucidate their underlying similarities and differences.


Axis 1 - Extent

Extent, in this context, refers simply to the number of separate images or frames used to depict the story. The obvious distinction here is between single images and multi-image cycles; what I shall describe as monoscenic and polyscenic narrative. The re-use of Weitzmann’s term ‘monoscenic’ is unfortunate but not really avoidable - in using this term I do not in any way imply an image which aspires to a unity of time, action and place - simply one that functions as a distinct and unitary image field. Monoscenic narratives can be used to tell the story generally (by focussing on the ‘pregnant moment’) or they can pick out some easily recognised action or event in the life of a character - either for its own sake or to help identify an associated non-narrative image (such images were often used on socles to identify the jamb figure above).

The polyscenic narrative type encompasses most of the conventional forms of image cycles, such as narrative windows, Psalter prefatory cycles and portal friezes.  In all these cases, one can identify multiple image fields which function in a spatially sequential manner to tell the chronologically sequential stages of a single story, or a closely related group of stories. The latter often applies with hagiographic windows, particularly those of confessors, where a series of unrelated miracles are allotted a few panels each (see for example the St Nicholas windows at Chartres and Le Mans).

In terms of extent, there is also a third, less obvious possibility, which I will describe as ‘oligoscenic’, in which a narrative cycle is reduced to a very small number of images (the exact number will depend on medium and context but is generally just two or three). Typically these oligoscenic narratives are made up of scenes abstracted from a polyscenic cycle and serve not to tell the story in detail but to identify an associated non-narrative image, such as the standing saint figures in the clerestory at Chartres, or the jamb prophets at Amiens. As an example, Fig. 2.05 shows one of the clerestory windows at Chartres (bay 142) which features extra-narrative iconic/hieratic standing figures of St Laudomarus and St Mary of Egypt. Laudomarus is identified by a monoscenic narrative image of his visit from the Bishop of Chartres (rather obscure perhaps but a key moment in his vita), while Mary is identified by an oligoscenic depiction of two highly distinctive and important events in her life - her first receipt of the sacrament and her burial by Zosimas, aided by one of the finest pantomime lions in all of medieval art. Such ‘cut-down’ narratives often serve a dual function of identifying the accompanying figure whilst also reminding the viewer of why they are worthy of pious devotion - in effect the narrative images help to valorise the extra-narrative ones.


Axis 2 - Mode

Narrative mode is a little more complicated to explain. Essentially it is an indication of the type of relationship between narrative image and its associated story - of the mechanism by which the former represents the latter. It is the how, but also to some extent the what, of narrative semiosis. To make sense of this it might be better to consider the various different modes and some illustrative examples.


Axis 3 - Visual Narrativity

Just as narrativity was discussed in section 2.a as a measure of the ‘degree of being narrative’, so visual narrativity is a measure of how much ‘narrative content’ is expressed in an image itself (regardless of the narrativity of its underlying story). Again, this is not an easy concept to explain directly but it can be envisaged as a measure partly of how ‘richly’ an image tells its underlying story and partly in terms of the richness of the viewing experience. In the case of oligoscenic and polyscenic narratives, it makes sense to consider the visual narrativity of the whole cycle as well as for individual frames.

In terms of the structural model presented in section 2.e, any narrative image will contain at least one nucleus or catalysis. As a minimum, narrative images are ones in which something can be seen to be happening. As visual discourses they depict actions or events representing the nuclei (decision-points) or catalyses (continuations) that combine in a particular sequence to represent the underlying fabula.  In this respect even a frame in a polyscenic narrative that shows a character standing alone doing nothing represents a minimal catalysis since it continues the narrative thread in spatial form. However one could also regard visual narrativity as a measure of how many informants and indices are present within the image since, although they don’t contribute to the progression of the plot, they contribute to the richness of its telling. Often such indices refer to elements outside the actual story; Christian typological references or even just extra-narrative cultural concerns of the work’s target audience (such as the meticulously observed military equipment in the Morgan Picture Bible) - yet these elements can still be thought of as contributing to the image’s visual narrativity, as can any other elements or devices which encourage a more active viewer-response. In the same way, the depiction of highly distinctive events, the quirky depiction of familiar ones, or the use of metalepsis and other paradoxical devices (see chapter 6) all contribute to viewer engagement and hence to the narrativity of the image.

Axis 4 - Arrangement

By arrangement I mean the physical disposition of the individual image fields that together constitute an oligoscenic or polyscenic narrative. The actual arrangement of scenes in a narrative cycle is often constrained by the medium and context of display; capital friezes (such as the one on the west façade of Chartres Cathedral) and lintels are, like the oft-postulated classical scrolls, effectively limited to a one dimensional, linear arrangement of image fields.66 Dado friezes, such as one finds at the cathedrals of Auxerre and Amiens also privilege linearity, though the greater height in both these cases allows for two rows of scenes. The tall, narrow lancet [65] forms of stained glass windows apply similar constraints but in the vertical axis. In the early-thirteenth century glazing programmes at Chartres, Bourges and elsewhere, glaziers excelled at finding complex tessellations of medallions, though by the end of the century these had almost entirely given way to simple grid or band arrangements of rectangular panels.

A tympanum offers more of a two-dimensional space, albeit one that narrows towards the summit. This format favoured the traditional triangular arrangement of Last Judgement scenes (Judge at the apex with the damned and elect in the lower corners) but it also encouraged the development of banded narratives with a climactic scene at the top, such as one finds in the north transept portal at Reims and in both the south transept and the north-western portal at Amiens. Round image fields such as rose windows, tables, or parts of vessels such as the foot of the ciborium shown in Fig. 7.10 needed to arrange their individual scenes in segments rotating around the centre - here we can distinguish between centrifugal (heads towards the centre) and centripetal (feet towards the centre) arrangements. Wall paintings generally offer more flexibility, though they may be constrained by doors, windows and other architectural elements. Manuscript illuminations offer a far wider range of possible arrangements, depending on whether they occupy a full page, an historiated initial or marginal or bas-de-page setting. Other historiated objects such as enamel reliquary chests (see chapter 8.a) and liturgical vestments tended to have the arrangement of their image fields determined by a delicate balance between the functional needs of the object and the desire to ensure that their narrative scenes may be viewed in an appropriate fashion.

Scale is another important factor in the arrangement of polyscenic narratives - with small objects such as manuscripts or ivory diptychs, the viewer can take in all the image fields without moving their head (facilitating a more personal and intimate viewing experience), whereas reading a sculptural frieze, mural painting cycle or narrative window will require a degree of physical action (possibly even a change of position) as part of the viewing process, making the act of reading a more public or communal affair.

Axis 5 - Direction

Direction refers to the physical trace or pathway that would be taken through the individual panels of a polyscenic narrative by a viewer following the story in narrative sequence, or by connecting typological / extra-narrative hypotactic frames to their related narrative scenes. Whilst direction works in conjunction with axis 4 - arrangement, it remains to some extent independent of it. A linear frieze can read from left to right, from right to left, or a mixture of the two, jumping back and forth as the physical setting and narrative content require. Such deviations from a normal linear pattern can be significant for the discourse, as with the well-known examples in the Bayeux Tapestry, where, for example, Edward the Confessor’s burial [66] precedes his death. More often however such variations in the direction are driven by the physical setting and by the need to orient the narrative action in relation to it. Thus in the Incarnation cycle in the capital frieze of the west façade at Chartres, the direction of reading has been re-arranged so that, for example, the movement of the Holy Family fleeing to Egypt or the Magi approaching the Virgin and Child are directed towards one of the doorways, the story leading the viewer quite literally into the body of the Church.

Lancet windows can read top to bottom (the norm in twelfth and thirteenth century England) or bottom to top (the norm in France). When there is room for two or three panels across the width of the lancet, one enters the realm of properly two-dimensional arrangements, which offer far greater scope for directional complexity.67 As well as the obvious text-like ‘left to right then top to bottom’ readings, together with its seven topological variants (right to left then bottom to top, top to bottom then left to right, etc), there is also the boustrophedon form, which reads from left to right then right to left on alternate rows. This latter figure is surprisingly common, appearing in a range of media but particularly in ivory diptychs (e.g. the Soissons Diptych - Fig. 2.08) and in stained glass (e.g. the St Eustace window at Sens - Fig. 2.09). Derived from epigraphy, the  boustrophedonic form (from the Greek, ‘as the ox turns in ploughing’) was not uncommon in archaic languages, where it seems to have had thaumaturgical  associations, something that survived in medieval palindromic or chiastic formulations such as the famous ‘SATOR-AREPO-TENET’ square. Whether the form also retained such associations in its medieval polyscenic narrative manifestations is impossible to say, though there is also a purely pragmatic explanation. Its success, particularly in some of the larger narrative windows with multiple vertical registers, may derive from the ease with which a viewer can follow the story. On reaching the end of a row, the eye simply moves one frame upwards or downwards, rather than having to scan back horizontally to find the start of the next row (a task which is easy enough if following rows of text but which becomes quite challenging when nearing the top of a lancet window, far from the reach of an index finger or stylus).


2.g) ordering and pacing in the second transformation

Although it is an oversimplification to think in terms of a simple two-stage mapping operation, from fabula to text and then from text to image, it is nevertheless helpful to consider the transformation from ‘primary discourse’ (the ‘implied text’ as discussed in section 1.b) to a visual ‘secondary discourse’. Section 2.b discussed the various temporal aspects of the transformation from a fabula to a (nominally textual) discourse - what Genette referred to as [67] ‘order’.68 In the following section I will re-visit those same themes in relation to the the formation of  visual narratives.

2.g.1 - Anachrony in visual narratives

Anachronies (see section 2.b.1) can be difficult to detect between the frames of a polyscenic narrative since in practice our assessment of directionality (spatial sequence) is primarily driven by our prior knowledge of the fabula-sequence of the events therein depicted. Moreover in contexts like stained glass where panels have sometimes been swapped around, deliberately or accidentally, during the last 700 years or so of restorations, the original makers’ intentions concerning sequencing are often effaced.69 The only clear exception is when a story is presented within the linearity of a frieze-format in a medium less prone to re-arrangements, though even here, deviations from the normal directional sequence may result from extra-narrative factors, such as the need to position a particular scene adjacent to a portal or facing into the church.

An altogether more fruitful area for considering anachronies in visual narratives is in relation to their role in extending the chronological scope of monoscenic narratives, or interpolating events between the individual scenes within a cycle, by means of proleptic or analeptic details. Such elements function as internal indices; they may or may not be relevant to the scene in which they appear but they also serve to reference events elsewhere within the story.70 A particularly elegant example of this, which will be explored further in chapter 4, appears in the Charlemagne window at Chartres (bay 07). In panel 07, where the King, accompanied by his young nephew Roland, kneels before the altar of his chapel at Aachen to present the relics he brought back from Constantinople, an oliphant (ivory horn) is shown hanging from a tie-beam above his head. Anyone familiar with the legends of Charlemagne and Roland could not help but see in that detail a proleptic reference to Roland’s heroic death at Roncesvalles, where he famously blows that horn in a forlorn attempt to summon help after being overwhelmed by his enemies. Precisely this scene is indeed shown in another roundel, four panels up along the central axis (panel 17). In this respect, Roland’s oliphant is similar to the various goldfinches, pomegranates and other Passion symbols held by the infant Christ in countless Virgin and Child images from across European medieval and Renaissance art - an unsettling element within an otherwise happy scene, which serves to remind the viewer or the tragedy that is to come. Conversely, in [68] many images of the Crucifixion, the bleeding wounds which cover Christ’s body are an analeptic index, pointing back to the scourging which happened at an earlier stage in the fabula.

2.g.2 - Achrony in visual narratives

As discussed in section 2.b.2 above, achrony is the temporary suspension of the temporal basis of narrative, marked in textual narratives by a switch to discursive modes like ekphrasis or self-reflexion. Its most obvious visual manifestation is the inclusion of one or more frames within a polyscenic narrative which are ‘conceptual’ (see section 2.f.1) rather than narrative. For example, most of the prefatory cycle of the Křivoklàt Psalter is occupied by scenes from the Incarnation, the Passion and the death of the Virgin. This polyscenic narrative series is framed however by two achronic images which serve as an extra-narrative framing device - a prologue and epilogue contextualising the events depicted in between. At the start of the cycle, folio 6 recto (Fig. 6.17) shows an elaborate depiction of the ‘Holy Kinship’, or Anna trinuba et tripara as it is sometimes known, (St Anne with all her husbands, daughters and grandchildren) - an example of Kress and Leeuwen’s ‘overt taxonomy’ (see section 2.f.1). The closing frame of that same cycle, f.13v (Fig. 6.25-lower), shows one of the most common achronic images found amongst medieval narrative cycles, namely the Coronation of the Virgin, complete with Angelic musicians. Although it is normally presented as the final image in the story of the Virgin’s life and/or dormition, the Coronation is in fact quite distinct from it; an achronic symbolic scene drawing upon imagery from the Psalms (45:9 and 21:3 in the KJV numbering).71

2.g.3 - Anisochrony in visual narratives

Just as the implied transition from fabula to discourse involves decisions about relative pacing or anisochrony (amount of discourse per hypothetical unit of fabula-time - see section 2.b.3), so does the transition from primary discourse (implied text) to image cycle. In other words, rather than allowing say ‘one image per ten verses’ or even ‘one image per three days of fabula-time’, the artist gives more weight to those parts of the story which are most ‘important’, according to the needs of the patrons, the context of display, or the general needs of visual narratives.

One of the few studies to consider issues of pacing in medieval art (and also the relationships between content and form - between story matter, arrangement and directionality, albeit not in these terms), was Alyce Jordan’s discussion of the Esther and Judith windows in the Sainte-Chapelle.72 As she noted, the story of the all-action heroine Judith is told in a limited number of scenes, contained within clearly defined circular medallions arranged on a simple vertical axis. [69] This is in stark contrast to the adjacent window telling the story of Esther, who also conquers an enemy and overcomes a threat to the Israelites but this time through a long slow process of careful diplomacy, and through her gentle wooing of King Ahasuerus. Esther’s story is told over a large number of panels of more complex shape, within which the variable disposition of scenes across panels requires a slow and thoughtful reading that seems to match the content of the story. If one sees these polyscenic narratives as segments within the Ste Chapelle’s broader telling of the Old Testament, the Esther panels constitute an example of ‘stretch’, while the Judith story is told as ‘summary’ (see Table 1 in section 2.b.3).

If the Esther and Judith windows are gendered narratives presenting ideal models of queenship, an equally gendered but distinctly masculine narrative artwork which also makes good use of anisochrony is the so-called Morgan Picture Bible (Pierpont Morgan Library, M.638). Compared with any other Old Testament narrative cycle, this manuscript devotes an extraordinarily high proportion of its images to battle-scenes, lovingly depicting the minute details of military equipment and siege engines. This fascination with warfare also manifests itself in narrative stretch (see section 2.b.3 above) whenever the Old Testament concerns itself with military conflict, compensated for by treating many of the less violent passages in the narrative in summary mode.73 Another warlike visual narrative whose anisochrony has attracted comment is the Bayeux Tapestry. As Grape noted:

The feast at Hastings and its preparations occupy approximately the same amount of space as the - historically far more important - process of building the fleet and embarking the army. The frequent transition from leisurely depiction to tight narrative (...) create a rhythmic variation in pictorial narrative such as had been unknown since late antiquity. Unmistakably the designer was looking for new narrative techniques that would maintain the beholder’s interest as never before, as he or she strolled the length of the tapestry.74

Given the inherent gappiness of visual discourses, anisochrony is one area where the specific requirements of particular audiences can be accommodated by polyscenic narratives more readily than they can by textual narratives.

2.h) topoi - literary and visual commonplaces

In classical schooling, the Greek term topos (pl. topoi ), or its Latin equivalent locus communis,  referred to a commonplace formula to be learnt by rote as part of a student’s training in rhetoric.  Both the Latin and the Greek names refer to a mnemotechnic method alluded to by Aristotle, later described in detail by Cicero and Quintillian (and more recently re-popularised by Francis [70] Yates and Mary Carruthers), for memorising complex lists of things by associating each one with a particular place on a well known street or building, so that recalling the information thus memorised is as simple as mentally strolling along that street and picking through the facts therein stored.75  The topoi learnt by rote in this way were  stock  components or building blocks of disputation which could be bolted together as required to build up an argument on almost any subject.  Throughout medieval Christian Europe, scholars continued to rely on topoi, learnt from standard texts like the Rhetorica ad Herennium or the Topica of Boethius.

Although it was never directly discussed in such terms, the topic mindset also permeated the construction of narratives, both in the reworking of well known fabulae and in the forging of ‘new’ stories, particularly in the rich field of hagiography.76 This is part of why we find certain themes or narrative fragments, such as imprisonment, refusal to worship a pagan idol, or decapitation and cephalophory reappearing in the stories of countless different saints.  To modern audiences, conditioned by  the apologetics of Renaissance humanism to regard originality as the Holy Grail of narrative invention, this deliberate rejection of novelty in favour of an endless recycling of existing story elements has sometimes been presented as a bad thing; a reprehensible symptom of the ‘Gothic’ mentality.  Such a view does however miss the point. For the hagiographer, individual saints were supposed to be ‘copies’ of earlier saintly figures. The similarity of an event in some new saint’s vita to events in the vita of an earlier saint was in part a modality marker (a way of asserting the narrative truth-value of the new story) and in part a means to encourage the audience to think about the exegetical correlations with the prototypes.

The copying of saintly topoi followed a chain of inheritance with Christ as the ultimate role-model. This view had received biblical sanction in St Paul’s epistles, where he implores his readers; ‘be ye followers of me, as I also am of Christ’ (1 Corinthians 4:16 and 11:1). As John Alford puts it, ‘Christ was a pattern for Paul, Paul for the Thessalonians, and the Thessalonians for their countrymen in general’.77  In fact, thanks to St Paul’s temporary inability to travel, there is a further demonstration of the inheritance of the sacred role-model in 1 Corinthians 4:17;  ‘For this cause have I sent to you Timothy, who is my dearest son and faithful in the Lord; who will put you in mind of my ways, which are in Christ Jesus’. Thus an extra link was inserted into the chain of copying, in the form of Paul’s locum, Timothy. The ‘added bonus’ of this [71] inheritance of role-models was that the humble devotee could more easily imagine him- or her-self joining the end of the chain by aspiring to emulate the saints whose topos-rich narratives were depicted in all the narrative spaces of Christian life. St John Chrysostom, in a prescient nod towards the assorted literary specula that flourished in the later middle ages, wrote:

...there is a mirror, spiritual, and far more excellent, and more serviceable than that other one; for it not only shows our own deformity, but transforms it too, if we be willing, into surpassing beauty. This mirror is the memory of good men, and the history of their blessed lives; the reading of the Scriptures; the laws given by God.78

To bend the metaphor slightly, the rich narrative windows of the great cathedrals, with their panoply of saints, were an exemplary mirror in which the viewer might see ‘through a glass, brilliantly’ towards their own salvific path.

All Christians were encouraged to emulate Christ and since the saints were those who had excelled in this emulation it is only natural that written stories of the saints were relatively uniform, endlessly recycling a stock of narrative topoi. Indeed sacred biographers borrowed more than just narrative structures from their predecessors, often re-using entire passages of text verbatim - not from the laziness of a plagiarist but as a self-conscious act of emulation. When the Cistercian monk Walter Daniel wrote his Vita Sancti Aelredii in the 1170’s, describing the life and death of his close friend, the Abbot of Rievaulx, he created what seems to modern eyes like a work of bricolage. Much of it was constructed out of lengthy quotations from earlier saints’ vitae, most of which would have been well-known to his contemporary readers.79 His other surviving works show Walter to have been an accomplished writer, well versed in inventio and perfectly capable of finding novel ways to describe events to which he was a close eye-witness. Yet the recycling of existing text fragments (frustrating as it may be to those historians who read him in search of practical information about life in twelfth century England) was fundamental to his task. Rather than being an insult to his friend’s memory, Walter’s lack of originality was a way of aggrandising it by stressing the extent to which, in his life, Aelred had successfully copied the examples of earlier saintly figures.80 Hagiographers, like the saints they documented, also strove to emulate their predecessors. Just as every  martyr-saint ultimately inherited his role from the proto-martyr Stephen, so every hagiographer inherited his topoi through a chain that started with St Luke, who had described Stephen’s martyrdom in the Acts of the Apostles. [72]

What holds for the writers of vitae is equally true for the designers of image cycles. At Chartres for example, where some twenty-four hagiographic narrative windows survive, it is striking how many near-identical episodes are depicted in the lives of otherwise unrelated saints.  The {imprisonment} topos appears at least nine times in the windows at Chartres (Fig. 2.10). In all cases the basic ingredients are the same - a guard raising a weapon to drive the saintly figure towards an open doorway. Far from being a local phenomenon, this same image type serves for imprisonment scenes in the stained glass of other cathedrals (see Fig. 2.10H and Fig. 2.10I ), as well as in a range of other media. One can do the same thing with any number of other hagiographic topoi, such as {decollation} (often followed by {cephalophory}) or {refusal to worship an idol} - which is usually followed either by {imprisonment} or else by {destruction of an idol}. Even such apparently distinctive events as the failed attempt to torture a saint on a wheel can be recycled. This episode, most firmly associated in most people’s minds with St Catharine of Alexandria, also appears in the vitae of various other saints, including St George, St Euphemia and St Pantaleon (an obscure eastern healer and martyr celebrated in a window at Chartres). Nor are topoi reserved only for martyrs - the lives of confessors routinely feature {exorcism}, {preaching to a crowd} and {baptism of converts} - and each of these scenes appears in several different windows at Chartres.

The systematic reuse of visual topoi in hagiographic image cycles has a number of benefits, above and beyond the constant reiteration of saintly exemplarism. On the simplest, practical level, it meant that artists could get by with a relatively small portfolio of image types (birth, education, disputation, imprisonment, torture, martyrdom, etc), which could easily be reused or adapted for all but the most unusual stories.  Just as importantly it made life considerably easier for the viewers of those images. Once a viewer has absorbed the basic elements of a particular visual topos, it becomes part of their stock of cognitive frames, while its relations to adjoining frames are resolved as scripts (see chapter 1.a). The same type of episode in a different saint’s life would then be instantly recognisable, even if the artistic style or medium was very different. This kind of storage is highly efficient since whenever something out of the ordinary happens, we don’t have to store every detail of the entire event in memory - we just store a pointer to the default script, plus the handful of details in which this particular event differed from the norm.


Previous Section Contents Page Next Section


1 H. White, 'The Value of Narrativity in the Representation of Reality' in On Narrative, ed. W. J. T. Mitchell, Chicago 1980, p.5.

2 N. Goodman, 'Twisted tales; or Story, Study and Symphony', in Critical Inquiry, 7(1), 1980, p.115

3 G. Prince, Narratology: the form and functioning of narrative, Berlin 1982b, p.145

4 According to the OED, the English term ‘narrativity’ (from the French narrativité)  was first used in this specific sense in G. Prince, Grammar of Stories : An Introduction., The Hague 1973, p.41.

5 The most significant of the more modern attempts at writing a ‘poetics of narrative’ are V. Propp, Morphology of the folk-tale, Austin 1968 [1928]; R. Barthes, 'Introduction to the Structural Analysis of Narratives', in New Literary History, 6(2), 1975, pp. 237-72; G. Prince, 'Narrative Analysis and Narratology', in New Literary History, 13(2), 1982a, pp. 179-88; M. Fludernik, Towards a 'Natural' Narratology, London 1996, and J. Bruner, 'The Narrative Construction of Reality', in Critical Inquiry, 18(1), 1991, pp. 1-22.

6 See for example R. Barthes, 'Introduction to the Structural Analysis of Narratives', in New Literary History, 6(2), 1975 and G. Genette, Narrative Discourse, Oxford 1980.

7 J. Bruner, 'The Narrative Construction of Reality', in Critical Inquiry, 18(1), 1991, p.6.

8 The changing medieval notions of time, and their implications for narrative art, are discussed in the concluding chapter of S. Whatling, Narrativity in French Gothic Portal Sculpture (MA Thesis), University of London (Courtauld Institute of Art), 2005, pp.36-40.

9   Of his many writings on this topic, the most useful is H. White, The Content of the Form - Narrative Discourse and Historical Representation, Baltimore 1987.

10 The following list draws heavily on J. Bruner, 'The Narrative Construction of Reality', in Critical Inquiry, 18(1), 1991.

11  ‘Intentionality’ in this sense used by  Husserl, i.e. that aspect of a mental state  (in terms of thoughts, desires, hopes, etc)  which is directed towards some object or goal.

12 See U. Eco, The Role of the Reader: Explorations in the Semiotics of Texts, 1979; L. Doležel, Heterocosmica: Fiction and Possible Worlds, Baltimore 1998; M.-L. Ryan, Possible worlds, artificial intelligence, and narrative theory, Bloomington 1991.

13 S. Chatman, Story and Discourse, Ithaca, NY 1978, p.19. If the preceding discussion of narrativity dealt with the dynamic characteristics of narratives, this might be regarded as a model of the ‘static’ underlying components.

14 This essay by Šklovskij, together with his earlier one on ‘Art as technique’ is among the most important texts from the infancy of modern narratology. In the entry on ‘Story-Discourse Distinction’ in the Routledge Encyclopedia of Narrative Theory, (p.566) Dan Shen mistakenly attributes the first discussion of the story-discourse distinction to Tsvetan Todorov’s 1966  article on the categories of literary narrative, yet it was already well established in the writings of  Šklovskij and Propp and had subsequently been used by various members of the Prague School. For English translations of both of Šklovskij’s pioneering essays, see L. T. Lemon and M. J. Reis, Russian Formalist Criticism: Four essays, Nebraska 1965. 

15 It is important to distinguish between ‘narrativization’ and ‘narration’ - which is simply an element at the discourse level whereby a character relays events to the audience

16 See for example S. Chatman, Story and Discourse, Ithaca, NY 1978.

17 S. Onega and J. A. Landa, Narratology: An Introduction, London 1996,  page v and pp. 6-7.

18 Chapter 8 of U. Eco, Semiotics and the Philosophy of Language, Indiana 1984, offers a dense but insightful and generally convincing study of the logical nature of the fabula and its relationship with ‘Possible Worlds’ theory.

19 For example Jean Rhys’ critically acclaimed 1966 novel Wide Sargasso Sea, which is a ‘prequel’ to Charlotte Brontë’s Jane Eyre. The recent success of Seth Graham-Smith’s 2009 ‘Jane Austen gore-fest’, Pride and Prejudice and Zombies  is a measure of how popular - and flexible - such story-world extensions can be.

20 E. Kafalenos, 'Not (Yet) Knowing: Epistemological Effects of Deferred and Supressed Information in Narrative' in Narratologies: New Perspectives on Narrative Analysis, ed. D. Herman, Ohio 1999, p.37.

21 One such text, Zaccharius of Besançon’s mid twelfth century ‘In unum ex quator’ became particularly popular in the thirteenth century and has been proposed by at least one author as ‘the source’ of the Bibles moralisées’ peculiar visual narratives. See K. Reinhart, 'The Texts of the Bible Moralisée' in The Bible of Saint Louis - Commentary Volume, ed. R. G. Ruiz, Barcelona 2004, p.283.

22 G. Genette, Narrative Discourse, Oxford 1980 [1972], Chapter 1 - ‘Order’ pp.33-85.

23 The term in medias res is a rather more recent invention than the literary practice it describes, though still old. It first appears in line 148 of Horace’s Ars poetica, written around 18BC.

24 I use the term ‘modality marker’ to mean those meta-communicative elements within a message (in any medium or genre) which are not necessarily parts of the story proper but which serve to affect the epistemic ‘truth value’ or credibility that the recipient is likely to assign to the associated semiotic content. I have borrowed both term and concept from G. Kress and T. van Leeuwen, Reading Images: The Grammar of Visual Design (2nd Edition), London 2006, pp. 154-74.

25 See M.-D. Chenu, Nature, Man and Society in the Twelfth Century, Chicago 1968, particularly chapter 5 for contemporary attitudes to temporality and the Eschaton.

26 Gérard Genette, 1980 [1978], p.48

27 The ‘threading-model’ terminology used here is borrowed from the computer programming world, where it is used to describe the various strategies for allocating processor time to multiple software programs running concurrently. The actual experience of reading narrative images in parallel would be analogous to the so-called ‘apartment threading’ model, which is the basis of multi-threaded processing in Microsoft Corporation’s ‘Component Object Model’ (introduced in 1993).

28 See M. Carruthers, The Book of Memory: A study of memory in medieval culture (2nd Ed), Cambridge 2008.

29 R. Barthes, S/Z, Oxford 1990 [1974], p.15.

30 J. Pollini, 'The Gemma Augustea: Ideology, Rhetorical imagery and the creation of a dynastic narrative' in Narrative and Event in Ancient Art, ed. P. J. Holliday, Cambridge 1993, pp. 258-98.

31 M. Holquist, The Dialogic Imagination: Four Essays by M.M. Bakhtin, Austin, 1981, p.100. The only difference between the chronotope of the Greek romance and the episodic hagiography is that in the latter, the specific location of the action becomes important because of the localisation of sanctity resulting from the miracle (with its consequences for the pilgrim industry).

32 For the Le Mans windows, which are about as far from being a coherent integrated programme as it is possible to get, see;

33 G. Genette, Narrative Discourse, Oxford 1980 [1972], pp.84-5.

34 Consideration of ‘Duration’ and anisochrony occupies the whole  of the second chapter of  Genette, 1980 (ibid), pp.86-112.

35 The actual run-time of each episode is around 44 minutes - the on-screen clock remains and continues ticking during the average 16 minutes of commercial breaks. (Robert Hanks, Jack's back: The clock ticks for 24's antihero, published in  The Independent, Arts & Entertainment section,  21/11/2008.) Paradoxically this makes it perhaps the only television format whose narrative ‘integrity’ suffers from the elimination of advertisements.

36 Genette’s term an-iso-chrony  (not-equal-time), though easily confused with ‘achrony’ and ‘anachrony’ does at least make good sense etymologically. Bal’s use of the term ‘rhythm’ (see M. Bal, Narratology: Introduction to the Theory of Narrative (2nd Edition), Toronto 1997, p.99ff) may derive from Prince’s use of that term (G. Prince, Narratology: the form and functioning of narrative, Berlin 1982b, p.55 - though he used it as a metaphor, rather than as a terminus techincus) but it adds nothing and the familiarity of the word ‘rhythm’, together with its various other meanings, is likely to lead to confusion. For this reason I will stick to Genette’s terminology

37 As Mieke Bal has pointed out (1985, p.72), such ‘indicated ellipses’,  are, strictly speaking, not ellipses at all but minimal summaries. In practical terms however if no more information about a passage of time is given but that time passed, it seems sensible to treat it as an ellipsis.

38 For the Tring Tiles and associated bibliography, see J. Alexander and P. Binski, Age of Chivalry: Art in Plantagenet England 1200-1400, London 1987, Cat.217.

39 G. Genette, Narrative Discourse, Oxford 1980 [1972], p.113ff. In fact Genette overlooked two other variations on frequency, corresponding to when the number of occurrences is 0, rather than 1 or n, although both of these cases correspond to terms covered elsewhere in his treatment of ‘tense’. In this sense 1F:0D (or nF:0D) constitutes ellipsis, while 0F:1D is digression

40 See for example A. Jordan, 'More is Better: Amplification and Design Theory in the Windows of the Sainte-Chapelle in Paris' in Proceedings of the 19th International Colloquium of the Corpus Vitrearum, ed. L. Kalinowski, Krakow 1998. For an excellent introduction to medieval teaching and practice of Rhetoric, see J. J. Murphy, Rhetoric in the Middle Ages: A history of the rhetorical theory from Saint Augustine to the Renaissance, Tempe, Arizona 2001.

41 Focalisation is discussed in G. Genette, Narrative Discourse, Oxford 1980 [1972], pp.189-94, but this excellent potted definition by Manfred Jahn comes from the Routledge Encyclopedia of Narrative Theory, p.173.

42 M. Bal, Narratology: Introduction to the Theory of Narrative (2nd Edition), Toronto 1997, pp.100-118

43 J.-P. Deremble and C. Manhes, Vitraux de Chartres, Paris 2003, p.220

44 See P. Steiner, 'Russian Formalism' in The Cambridge History of Literary Criticism, Vol.8, ed. R. Selden, Cambridge 1995, pp. 11-29, (particularly p.19ff) for the background to Russian Formalism(s).

45 V. Propp, Morphology of the folk-tale, Austin 1968 [1928], p.21

46 Summarised from V. Propp, ibid, pp.21-23

47 Claude Lévi-Strauss was perhaps the most vocal critic. See particularly his essay ‘Structure and Form: Reflections on a Work by Vladimir Propp’, published in C. Lévi-Strauss, Structural anthropology (Trans. Claire Jacobson and Brooke Grundfest Schoepf), New York 1963.

48 A. J. Greimas, Structural Semantics: An Attempt at a Method, Lincoln, Nebraska 1983 [1966], p.200ff

49 This deficiency was partly addressed in E. Kafalenos, 'Functions after Propp: Words to Talk about How We Read Narrative', in Poetics Today, 18(4), 1997, pp. 469-94, which seeks to develop a more reception-oriented functional model.

50 Arguably the ‘Witness’ could be considered as just a specialisation of the ‘Helper’ role. I have however separated the two since a) the functions of the witness in the verbal narratives is often distinct (especially in the case of the Old Testament prophets) and b) their visual characteristics are also very specific.

51 R. Mellinkoff, Outcasts : signs of otherness in northern European art of the late Middle Ages, Berkeley 1993, provides a useful and extensive (if sometimes slightly tendentious) discussion of such marks of ‘otherness’.

52 R. Barthes, 'Introduction to the Structural Analysis of Narratives', in New Literary History, 6(2), 1975 [1966], pp. 237-72. This pivotal paper was first published in Communications, 8 (1966) as Introduction à l’analyse structural des récits. It was translated into English by Lionel Duisit and in 1975 was one of a number of influential essays to appear in a special volume of New Literary History devoted to narrative theory. Since then it has been reprinted widely in various anthologies dedicated either to the subject or to the author, including S. Onega and J. A. Landa, Narratology: An Introduction, London 1996, (pp.45-60), and S. Sontag (Ed.),  Barthes: selected writings, London 1983, (pp.251-95).

53 Reprinted in D. Jobling, T. Pippin and R. Schleifer, Eds., The Postmodern Bible Reader, Oxford 2001,( pp.58-77). Barthes’ other major foray into the field of Biblical studies was  ‘Wrestling with the Angel: Textual Analysis of Genesis 32:23-33’, which is reprinted in R. Barthes, The Semiotic Challenge, Berkeley 1988 [1985], (pp.246-60).

54 J. A. Rushing, Jr., Images of adventure: Ywain in the visual arts, Philidelphia 1995.

55 R. Barthes, S/Z, Oxford 1990 [1970], p.3

56 G. Kress and T. van Leeuwen, Reading Images: The Grammar of Visual Design, London 1996,

57 See particularly M. Carruthers, The Book of Memory: A study of memory in medieval culture (2nd Ed), Cambridge 2008.

58 For the relationship between the narrative and ‘eternal’ elements in portal sculpture, see S. Whatling, Narrativity in French Gothic Portal Sculpture (MA Thesis), University of London (Courtauld Institute of Art), 2005, pp.36-40.

59 Translated into English by Mrs S. Arthur Strong and published in 1900 as Roman art: some of its principles and their application to early Christian painting.

60 M. Bal, 'Close Reading Today: From Narratology to Cultural Analysis' in Transcending Boundaries: Narratology in Context, ed. W. Grünzweig and A. Solbach, Tübingen 1999, p.20.

61 The issue of topoi in both textual and visual narratives is discussed in more detail below (chapter 2.g).

62 The term ‘reductive epitome’ was coined by Richard Brilliant (R. Brilliant, Visual narratives: storytelling in Etruscan and Roman art, Ithaca 1984, p.52) to describe the ‘Megarian’ bowls (Athenian mass-produced moulded drinking vessels of c.225-150 BCE) which were decorated with ‘degenerated’ narrative cycles. Although his application of the term to Greek art has been challenged (see for example the review by Malcolm A. R. Colledge in The Classical Review,  35:1, 1985, pp.169-71), I find the concept highly relevant to some types of medieval art. I have however favoured the adjectival form ‘epitomic’ rather than the tautological ‘reductive epitome’.

63 See W. Noel, 'Medieval Charades and the Visual Syntax of the Utrecht Psalter' (pp. 34-41); and L. F. Sandler, 'The Images of Words in English Gothic Psalters' (pp. 67-86), both in Studies in the Illustration of the Psalter, ed. B. Cassidy and R. Muir-Wright, Stamford 2000. Sandler uses the term ‘word-images’, whilst Noel talks about ‘charade-images’ - both of these are highly apposite terms denoting what I would regard as different sub-types of pseudo-narrative imagery.

64 M. Camille, Image on the Edge - the Margins of Medieval Art, London 1992.

65 Photographs of the full set of 42 choir stall spandrels from Poitiers Cathedral and all of the 118 quatrefoils from the west façade of Amiens Cathedral are available on the author’s website;  go to and follow the links. The meaning of the boy with the basilisk remains unknown.

66 For the definitive study of the Chartres west façade capital frieze, see  A. Heiman, 'The Capital Frieze and Pilasters of the Portail Royal, Chartres', in Journal of the Warburg and Courtauld Institutes, 31, 1968, pp. 73-102.

67 In her study of Italian mural decoration, Marilyn Aronberg Lavin defined eight distinct ‘patterns’ for the reading sequence in narrative wall paintings. Most of these are however specific to her particular medium and geographical area of interest and need not be considered here. (See M. A. Lavin, The Place of Narrative: Mural Decoration in Italian Churches, 431-1600, Chicago 1990, particularly pp.6-9)

68 G. Genette, Narrative Discourse, Oxford 1980 [1972], p.33ff.

69 For the issues behind nineteenth century restoration of medieval glass and its effects on narrative arrangements, see A. Jordan, 'Rationalizing the Narrative: Theory and Practice in the Nineteenth-Century Restoration of the Windows of the Sainte-Chapelle', in Gesta, 37(2), 1998, pp. 192-200.

70 The use of  anachronic elements within narrative images would be championed in the 18th century by that master of narrative art, William Hogarth. The individual scenes within his ‘Progresses’ positively abound with inconspicuous details which serve not as informants to enrich the description of the particular moment but as internal indices, to ‘fill in the gaps’ between the frames.

71 See E. Mâle, Religious Art in France: The Thirteenth Century, Princeton 1984 [1898], p.255

72 A. Jordan, 'Material girls: Judith, Esther, narrative modes and models of queenship in the windows of the Sainte-Chapelle in Paris', in Word and Image, 15(4), 1999, pp, esp. pp.47-50.

73 For this extraordinary manuscript, see W. Noel and D. Weiss, Eds., The Book of Kings: Art, War, and the Morgan Library's Medieval Picture Bible, London 2002.

74 W. Grape, The Bayeux Tapestry: Monument to a Norman Triumph, Munich 1994, pp.69-70.

75 See F. A. Yates, The Art of Memory, London 1966;   M. Carruthers, The Craft of Thought: Meditation, rhetoric and the making of images, 400-1200, Cambridge 1998;  and M. Carruthers, The Book of Memory: A study of memory in medieval culture (2nd Ed), Cambridge 2008.

76 Of course it did not stop with Christian hagiography - Ernest Robert Curtius saw the topos as fundamental to the development of practically all Western literary forms (see E. R. Curtius, European Literature and the Latin Middle Ages, Princeton 1973 [1953]).

77 J. Alford, 'The Scriptural Self' in The Bible in the Middle Ages: Its Influence on Literature and Art, ed. B. S. Levy, Binghampton, NY 1992, pp. 1-21.

78 St John Chrysostom, Homily IV (on Matthew 1:17), verse 16. From Nicene and Post-Nicene Fathers, Series I, Vol. X Available on-line at; (checked 05/01/2010)

79 See T. Heffernan, Sacred Biography: the Saints and their Biographers in the Middle Ages, Oxford 1988, p.177ff.

80 In his fascinating study of medieval hagiography, Thomas Heffernan cites a letter Walter Daniel sent to his friend Maurice bemoaning the fact that some readers of his Vita Sancti Aelredii had actually failed to recognise the excerpts he had used from the Life of St Martin (ibid).


Previous Section Contents Page Next Section