The Journal Club and Roland Barthes, An Introduction to the Structural Analysis of Narrative
By Rogier Brussee on 5 March 2012
Figure 1 Roland Barthes teaching
The Journal Club
The knowledge centre of the faculty of Communication and Journalism, came up with the excellent initiative of starting a Journal club. It was initiated, by Matthijs van der Schaft, and was organized by Gerard Smit with some input from Erik Hekman and me. The subject of the seminar is narrative, a subject that in many ways, runs through journalism, advertising, crossmedia strategy, and corporate communication alike. Two weeks ago we finished the first four sessions. I think it was a great success.
The goal of the journal club is to study some fundamental articles in the field and to do so with the whole faculty, rather than each of us in our own groups. It is an incarnation of the academic seminar: a venerable institution that is very important for the exchange of scientific knowledge and raising the intellectual level. It helps forming a culture of critical but respectful dealing with the body of knowledge in a field by studying a subject and one or more articles, sometimes in combination with talks on original work. I find it leads to reinterpretation of known work and reconnects your knowledge to ideas that you have somewhat forgotten, only learned in (over)simplified form or simply never learned at all. It also widens your point of view, as seminars tend to be on a subject somewhat off of your own field. In the process you tend to discover that the original articles contain a richness that rehashes in secondary work often lack and a perspective that, with hindsight underestimates some aspects and overemphasizes others. Above all it leads to discussion, and internalization of a shared body of knowledge in ways that just reading an article on your own can seldom match. Speaking for your colleagues is also an important forming experience for a young scientist, which prepares you for the less forgiving experience of talking on conferences. I for one, can still remember the first time I gave a talk on a serious article, and the first time I had to present my own work as a young PhD student.
In the four sessions we covered four articles. The first session on Hayden White, “The value of narrativity on the representation of reality” Critical Inquiry, Vol. 7, No. 1, On Narrative (Autumn, 1980), pp. 5-27 . was presented by Gerard Smit. White argues that what makes history, history and more than a collection of random chronological events (the form of the traditional annals) is that it is reworked into a story with a moral edge. The moral edge gives a sense of purpose to the main actors, rather than a description of the things that happened to a people driven by the will of an omnipotent God. An interesting viewpoint, that certainly contains some truth but one that could not quite convince me. Historians have long tried to take sufficient distance that they can see both sides, starting with Thucydides account of the Peloponnesian war. I also could not help noticing that the narrow set of examples that White gave to make his point, sometimes showed a lack of knowledge of European medieval history (example: White makes a point that the annals of the Saint Gall monastery “only” mention the battle of Poitiers and fail to mention the battle of Tours, which many historians later described as a turning point in European history stopping invading Islamic forces from conquering Gaul and paving the way for the Carolingian empire. In fact the “battle of Tours” and “battle of Poitiers” are two names for the same battle). In any case, Gerard did a good job at introducing White and explaining that many of White’s ideas were rooted in the philosophy of Kant and Hegel.
The second session on a classic work by Roland Barthes, “Introduction à l’Analyse Structurale des Récits”, or in the English translation, “An Introduction to the Structural Analysis of Narrative” was given by Madris Duric and myself. Madris introduced Roland Barthes as the philosopher and writer, stressing that the main ideas in Barthes’ work are the interplay between writer and reader. According to Barth, the reader is doing most of the work (especially in “mort de l’auteur”) with the writer “only” hinting the reader to catch his meaning based on preconceived ideas or indeed myths. He also gave some nice examples of visual signs that trigger the viewer into thinking about some associated connotation. The study of signs is of course semiotics on which more later on. Madris used some slides of his lectures in class, as the rudiments of semiotics are part of the curriculum of Communication and Media Design (CMD). The idea that signs, especially visual signs, interact with the observer, in subtle and not so subtle ways is often used in the study of visual media, in particular when teaching advertising. It was my task to explain the paper proper. More on this below.
The third session on Jerome Bruner, “The narrative construction of reality”,Critical Inquirer 18 vol 1 (1991) p.1—21, was given by Marlies Havenga and Kees Winkel. Kees introduced Bruner, who is a cognitive and developmental psychologist, which makes him something of an outsider in this field. Marlies introduced his article which she had extensively studied for her master thesis. It starts with some developmental psychology arguing that narrative is an essential part of the way people represent reality and their self image. Brunner then makes a 90 degee turn and goes on with a rather concrete 10 point list of the essence of narrative: narrative diachronicity, particularity, intentional state entailment, hermeneutic composability, canonicity and breach, referentiality, genericness, normativeness, context sensitivity and negotiability, narrative accrual. Marlies had used this list for the master thesis she wrote while at the crossmedia lab. As part of her thesis she had developed a classification of the 10 points that gave this list more structure. Rather than giving a long presentation she and Kees let us work with this list to analyze a page from the children’s book Mr. Finney, arguing that in children’s books everything should be particularly clear. This was a nice break from the somewhat theoretical discussions we had so far, and shows that a presentation is ultimately determined by the presenter.
The final presentation was given by Gerard Smit once more. He discussed
John Bryce Merrill “Stories of Narrative: On Social Scientific Uses of Narrative in Multiple Disciplines” Colorado Research in Linguistics. Vol. 20. June 2007 p. 1-25. As the title suggests it is an inventory of different uses of the notion of narrative in the social sciences. After noticing that there seems to be no consensus on the definition of narrative (and arguing, not very convincingly IMHO, that this is not really necessary), he distinguishes three different fields where narrative is used: narrative construction of self and reality, narrative features, form and functions, and narrative as method. Being an overview article it contains less original ideas than the other three.
Roland Barthes, An Introduction to the Structural Analysis of Narrative
Roland Barthes, “Introduction à l’Analyse Structurale des Récits” Communications, 8, 1966. pp. 1-27.
doi : 10.3406/comm.1966.1113, or in the English translation, “An Introduction to the Structural Analysis of Narrative”
is a classic work in the area of Narrative. It is a widely cited with 1029 English + 998 French references (according to google scholar), although there is probably a significant overlap in these numbers. It lies the foundation for a theory narrative based on ideas in linguistics and semiotics and is heavily inspired by the works of the late 19th century linguist and philosopher Ferdinand de Saussure.
In the tradition of the humanities, it is written in flowery and long winding, but quite readable French, replete with references to French literature, poetry, literary analysis, linguistics, semiotics and anthropology. However, showing Barthes lighter touch, it also contains plenty of references to James Bond, in particular to the movie Goldfinger, which was quite recent at the time. I have no background in literature or visual design, so I tried to stress the logical structure of his argument and systematically describing the structural forms that Barthes describes translating in a bit more formal language. Barthes tries to set up a “structural analysis of narrative”, that is, a theory of narrative along the lines of linguistics. It looks for structural elements of language, that reveal, rather than depend on meaning. He is also interested in the semiotic part, the relation between the signs of language (including its structural elements) and the language external meaning. What this means is that he is out to construct a somewhat idealized theory of narrative, its “Langue” in the language of linguists. This is somewhat similar to describing what grammatically correct English looks like, rather than what people actually speak. Clearly there is difference between the two, but equally clearly, grammatical English is much closer to written and spoken text than say a random collection of English words distributed with the probability of their occurrence in natural language . Such a regularized model is also easier to study, and draw conclusions from. However, there is the ``minor’’ problem of finding a structural model that actually gets to the essence of the problem. This is what Barthes sets out to do.
Despite his flowery language, Barthes approach is actually rather systematic, in line with the linguistic tradition he is trying to follow. Linguistics studies the structure of language rather than its meaning and as such has a lot in common with mathematics. He points out that postulating structure has been a success story in linguistics, allowing to organize seemingly disparate observations, sometimes even predicting observations before they were made. Perhaps the most famous ones are the reconstruction of extinct proto-languages like Proto-Indo-European. Linguists like as Zellig Harris had also proposed a purely structural approach to discourse analysis based on grammatical transformations and equivalence relations between parts of speech determined by usage. (Harris, Z.S. 1952, “Discourse Analysis”, Language vol 28 (1) 1952 p.1—30).
Barthes has two main ideas. The first idea is that narrative has a syntax, a grammar, where the basic elements, the functions, can be composed in well defined ways to what he calls action sequences. In term the sequences are composed in well defined ways to narrative, similar to the way words build up parts of speech, which in turn build up sentences. There is also a recursive structure, just like in the grammar of (essentially) all human languages: each level can be used at a lower level. Thus actions can themselves have a definite function, and small sub narratives can work as an action sequence in a larger narrative. They are all narremes.
To explain this grammar we should realize that the linear structure of language imposes itself on narrative: if we use language (or for that matter, a temporally structured medium like film) we have no choice but to set one narreme after the other. The order matters, and some types of narremes can be put after the other while others cannot without ruining the narrative. That virtually forces you to think of some sort of grammar, some production rules.
Production rules are an old idea in grammar going back at least to the great Indian grammarian Panini in the 4th century BC. For example, there is a rule in English grammar that states that an article is followed by an adjective, a noun, or (in rare cases) a modifier. Linguists and computer scientists think of this as a consequence of the grammatical production rules for a nounphrase. pieces of the sentence that grammatically qualify to be used as a subject e.g. “a boy” or “the black cat that crossed our way” or “the among living poets greatest poet”. Such production rules are the heart of a grammar and can be describe in so called Backus Naur notation, a notation already used, in essence, by Panini, but formally invented to denote the mother of all modern programming languages, Algol 60. However If you like algebra you can also think of reading production rules from right to left and think of them a “multiplication” or “composition” rule. This point of view says that you have a “grammar” whenever you can juxtapose “things” of some type to get “thingies” of possibly different type. The simplest possible grammar may then describe the multiplication of positive integers: “1” is a number and multiplying two numbers gives a new number. Thus integer multiplication follows the very simple “grammar”
::= “1” |
Actually for integer multiplication you need additional rules that say that that multiplication by “1” does nothing and that multiplication is commutative and associative i.e. that when doing multiplication you may compute products of numbers in any order you like (in case you wondered, this grammar with rules does not really characterize the positive integers but accepts a much more general algebraic structure called a commutative monoid). In the grammars used by linguistics or computer science such extra rules are usually not included, but it is an entirely respectable thing to do in algebra. In any case we are exactly in the situation that Barthes describes when he introduces narremes.
Reading through Barthes more literary language we can set up a grammar of Narrative. I will just describe this grammar in words rather than using a formal Backus Naur grammar with some extra rules.
• A consists of s
• A consists of s
• A is a (called an action)
• A starts and ends with a
• s and s are function
• s and s combine freely
• < Catalyst>s come before a
A is irreducible (or prime as in prime number) if it can not be decomposed in a juxtaposition of s. which gives a somewhat ineffective termination condition. At this point all of the words in fish angles are merely abstract symbols. However Barthes gives a mostly semantic (i.e. extra lingual) definition of the different functions
Function : part of a text that is a potential source or target of a correlation with other part of the text
The functions themselves are then classified in four categories corresponding to the elements of the narrative “grammar” above which tells you which ways they combine:
• Catalysts: delays, quickens the pace , sums up, anticipates, confuses …
• Nucleus: (potential) branch point of the narrative (some important some minor)
• Indice: has metaphoric relations
• e.g. part standing for a whole
• Informant: has metanomic relations
• example stands in for a whole concept
The first two and the last two should really be thought of as part of two different classification schemes. Thus functions can be put in a two by two matrix. What seems to be missing here are actions and actors, but as Gerard Smit pointed out, in Barthes view the actor is only important as the function it has and the action that the actor is involved in in some sense defines the actor.
The dual idea of production rules, and the reason that linguists write in the left to right notation, is the parse tree. The parse tree of a sentence is a natural way to organize all the possible ways in which you can decompose the sentence by applying production rules to generate the sentence out of terminators (at least if the parse tree terminates!) For example, the parse tree of the sentence “the cat with the hat is wet” starts with using the production rule
Ofcourse we can now go on and parse the and further giving rise to a tree structure . Barthes creates similar parse trees to describe the narrative of a sequence in a James Bond scene.
For narrative, it is important to realize that once you start parsing a text with the grammar and you have found just a part of the parse tree, there are almost always far far fewer possibilities that can apply for the rest. Thus once you are using a grammatical structure for communication, the reader (or listener) is subject to a certain amount of determinism: once fed with a few leaders by the writer, the reader is expecting one or just a very few things to follow. In music this is well known: given just a few notes, we recognize the tonality, and after a few chords we feel an urgency for the chords to dissolve in accordance with the musical scheme as the logical conclusion. Alternatively, the writer, or composer, can use this effect to create surprise and tension, by choosing the possible and interesting, but more unlikely continuation. In natural language this idea works on many levels: from sound to words, form words to sentences. Barthes point is in narrative a similar mechanism is at work, one that straddles all levels, from the phonetic (rhime and rhythm) all the way through functions and action sequences.
This brings us right to Barthes second main idea, that functions or more generally narremes, are mentally triggering existing concepts in the reader. This is a core idea in semiotics; it says that narremes function as signs in a semiotic relation with a language external referent, mediated by reader held concepts.
I like to think of this as the concepts of the reader forming a Baysian network and that what the narrative does is, carefully conditioning us to get one likely answer. Moreover it seems reasonable that we take only a few things into account simultaneously. The stack is thus rigged against us and we tend to believe a very few things while discarding other possibilities that don’t fit our worldview. Barthes argues, that while reading, a reader builds up a new network that gets updated while reading a text starting from defaults that represents a readers preconceived ideas, but
one (at least that is my interpretation).
Now to our mind the “grammatical” necessity, works like causality, perhaps because we are actually using the neural pathways used to predict what comes next. Thus what follows in the narrative, seems to follow because of what preceded it (“post hoc ergo propter hoc”). The experience of a form of “causality”, is a seductive internal “logic” of cause and effect driven by subtle or not so subtle hints in the narrative and preexisting conceptual frameworks. Good narrative uses this to create a feeling of coherence and a suggestion of a rich structure of a narrated world that goes beyond the sequential structure of the narrative. The internal “logic” may well be completely illogical in a mathematical or even a common sense sort of way: it may involve magic, extremely unlikely heroism or winning the lottery. Just everyday experience shows that it is probably more important that a good story is consistent with our desires and with conventions like” the good guy wins and gets the girl”,” the pain was not in vain” or for that matter “the invisible hand of god the market is punishing for the sins of sun and garlic” “plantasterol is good for your heart”, (to my surprise plantesterol http://en.wikipedia.org/wiki/Phytosterol actually exists as a Dutch word http://nl.wikipedia.org/wiki/Plantensterolen, but for marketing purposes it is above all important that it sounds scientific and like good cholesterol vegetable healthyness). However, Barthes extends this idea to a much finer scale where each functional element or even words and phonemes dynamically reinforce or damp associations between concepts.
Dependence on preexisting ideas in the reader makes the narrative depend on what these ideas are. This is especially important for the grand narrative of society. Barthes, like many French postmodern thinkers of his time (e.g. Derrida, Foucault, Levi Strauss) was highly politically conscious (which at the time was a synonym for left wing), and acutely aware of preconceived “myths” that are right in the skeleton of society. Some of his interest was therefore that such “myth”s get continuously self reinforced by constant repetition to a grand narrative, a system of shared believes, values and petty habits that “defend” the status quo and are a great obstacle to change. From my poorly informed point of view, it seems that, as the sixties gave rise to the seventies, postmodern thinking became increasingly obsessed by narrative, the thinkers of Paris 68 became increasingly opaque and increasingly part of the new order that had started the long march through the institutions. French thinkers became fashionable in some academic circles in the USA and generations of English literature students and faculty got exposed to rehashed and over simplified versions of the original ideas. As a result they deconstructed all ideas to become male chauvinist text based on social convention. In particular the idea that narrative creates a personal and social context dependent representation of reality that, since people act upon this image, has some very real consequences was ``simplified’’ to the (IMHO extremely dangerous) idea that reality is a mere social convention. Meanwhile, left leaning political ideas (like the need to pay taxes) has been systematically discredited in general, while fact free “reality is just an opinion of the liberal media” politics has become the norm. In this climate, Barthes ideas have been discredited as backward. However, while he may oversimplify things his article seems driven driven by an intellectual honest attempt to find the mechanisms, structure and essence of narrative irrespective of what that narrative is or has to say, and I found them having deep insights.
In the discussions that followed my presentation it was pointed out that Barthes ideas where used in the visual arts where the linear structure of language does not apply. That means that a different two or three dimensional grammatical composition rules apply. Of course “the composition” is a well known term in the visual arts, architecture and design where it has been studied extensively. The idea that composition and composition rules interact with visual “signs” certainly seems to apply in that context which to me showed once more the usefulness of having different traditions to look at the same problem.