Downloaded from http://camqtly.oxfordjournals.

org/ at National Chung Hsing University Library on April 11, 2014

Roland Barthes's Narratology

I should have thought that Barthes, who has gone into his posthumous
slump, will certainly endure simply because he writes so beautifully
and because although he says many silly things, as we all do, some
of his work is of lasting importance for the study of literature.
Frank Kermode, Traditions and Literary Studies, (1990) p. 82

TRADITIONALLY, LITERARY THEORY, from Plato onwards, has set itself

the task of establishing a conceptual framework for practice - the practice
of writing poems and narratives and also (over the last three centuries, at
least) the practice of reading them. Hostile critics of recent developments
have sometimes objected that this essential linkage between theory and
practice has now been broken. There is an element of truth in this, at
any rate in relation to the kind of structuralist poetics which flourished
in the 1960s, where the objective was almost invariably seen as that of
formulating a comprehensive theory of literary discourse in the abstract.
In line with the scientism then in vogue, structuralism claimed to be a
scientific method capable of identifying the general laws governing the
phenomena of language, culture and society, so that attention came to
be targeted not on activities specific to a particular human grouping but
rather on die underlying structures of which these were said to be merely
manifestations. Thus in linguistics die focus was not on actual speech-
events (parole) but on die internal systems sustaining language as a social
institution (langue). In a similar way in literary dieory die focus of concern
moved away from die individual poem, story or play as such, this being
now seen as 'simply the means by which one can describe die properties of
literature in general'.1 Thus, as Tzvetan Todorov put it quite explicidy,
die aim of structural analysis can never be 'the description of a concrete
work' [of literature]; instead, die work will be considered 'as die
manifestation of an abstract structure, merely one of its possible

1
Tzvetan Todorov, Pokique, 1973.
« THE CAMBRIDGE QUARTERLY

realisations', and the 'real goal will be understanding of that structure'. In

practice', he continued, 'structural analysis will also refer to real works . . .
But such analysis will discover in each work what it has in common with
others (study of genres, of periods, for example), or even with all other
works (theory of literature); it would be unable to state the individual
specificity of each work . . .' 2 There can be no doubt that Roland

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
Barthes's earliest involvement in narrative theory was deeply imbued with
diese attitudes (and was indeed tinged with the scientism implicit in them).
But as he moved, radier uncertainly, from 'structural analysis' to 'textual
analysis' a certain tension between the claims of abstraction and of
specificity became increasingly apparent in his writing. In what follows,
moreover, I shall hope to show that in addition to making some vast
generalisations about the nature of literature, he also offered (in that
small part of his varied oeuvre which I am here concerned with) some
highly unrealistic prescriptions for both the writing and the reading of
it - prescriptions which are peculiarly difficult to understand in their full
complexity and some of which are in consequence sometimes acted upon
in disastrously over-simplified ways.
Even in the 1960s Bardies did not see himself as turning away from the
concrete and specific quite so whole-heartedly as some of his fellow-
structuralists. In his 1966 essay, 'Introduction to the structural analysis of
narratives' his aim had been (he told an interviewer five years later) to
'reconstruct a sort of grammar of narrative, or a logic of narrative' from
which 'would then be derived analyses of contingent texts'. The 'general
structure' he posited at this time is both complicated and highly abstract,
and it is based quite explicidy on a model provided by the structural analysis
of language. Thus as his starting-point he takes die hierarchy of levels
(phonetic, phonological, grammatical, contextual) employed by linguistics
to describe die sentence. By analogy he proposes to distinguish in narrative
a hierarchy of three levels for analysis: the level of Junctions, die level of
actions, and die level of narration (or discourse). These levels are said to
be closely linked togedier in practice, in that 'a function only has meaning
insofar as it occupies a place in the general action of an actant; and diis
action in turn receives its final significance from die fact diat it is narrated,
entrusted to a discourse which possesses its own code'. While he does
accord some space to a discussion (radier cursory) of die two higher levels,
Bardies gives his most detailed attendon to die functional units; and diese
he subdivides (in a manner which is terminologically a litde confusing)
into two main categories - die functions proper and the indices.

' Tzvetan Todorov, 'Structural Analysis of Narrative'. Novel, III, i (1969),

pp. 70-1.
 ROLAND BARTHES'S NARRATOLOGY

The Junctions proper arc elements which acquire their meaning and
significance through their relationship to something either earlier or later
on in the story: an example (rather hackneyed) used by Barthes is 'the
purchase of a revolver (which) has for correlate the moment when it will
be used'. In terms of the linguistic analogy their role is, in fact,

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
syntagmatic, not paradigmatic. They are, moreover, further subdivided
into (a) cardinal functions (sometimes referred to by their alternative title
of nuclei), which represent what Barthes calls the 'hinge-points' of the
narrative, those points at which what happens, whether seemingly
important or not, can make a significant difference to what happens
subsequently: and (b) calalysers, incidents which though functional are
essentially complementary (or even 'parasitic') in that their role is to fill
the space between two nuclei.
The indices, by contrast, contribute their significance not in relation
to the sequential flow of incident, but by reference to the implications
they carry within them on one of the two other levels - providing, for
example, some indication on the one hand about the character and
personality of one of the actants, or on the other about the mood and
atmosphere engendered by that 'discourse' which controls the overall
meaning of the narrative. It is in this figurative sense that the indices are
said to be paradigmatic, in that they contain an implicit reference to a
range of possibilities which define and constrain certain more extended
and diffuse features of the narrative. Though the comparison is rather
a loose one, these have some affinity with the range of meanings
embodied, in linguistic analysis, in the paradigms of tense in verbs or
of comparison in adjectives.
However, Barthes contends that, despite the differences between them,
catalysers and indices can from another point of view be seen to have a
quality in common: they have their existence in every case as expansions
(another term and concept borrowed from linguistics) of one or more of
the nuclei. It is thus the nuclei which constitute the indispensable framework
of the narrative structure; being 'at once necessary and self-sufficient'
they include all the items which would need to be retained in any adequate
synopsis or plot-summary.
Continuing still on the level of the functions, Barthes offers to organise
the smaller segments (the nuclei, many of which may be quite trivial) into
a hierarchy, an 'organisation of relays', in which the basic unit is what
he calls a sequence. A sequence is a logical and self-contained succession
of nuclei whose distinctive feature is that it is 'nameable'; indeed it is
specifically implied that in the process of reading the reader necessarily
supplies (internally) the appropriate name ('to read is to name'). These
names (borrowed in the more frequendy occurring instances from Propp
44 THE CAMBRIDGE QUARTERLY

and Br6mond) seem designed to enforce a belief that reading is in effect

a matter of recognising stereotypes (Fraud, Betrayal, Struggle, Contract,
Seduction, etc.). As Barthes puts it quite explicitly: 'any function which
inaugurates a seduction, say, prescribes upon its appearance, in the name
which it produces, the whole process of seduction that we have learned

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
from all the narratives which have formed in us the language of narrative'.
(On a more trivial level Barthes puts forward 'to order a drink, to receive
it, to drink it, to pay for it' as an example of an 'evidently closed sequence'
to which he gives the name 'Having a drink'.) The functions are thus
seen to be organised within a mounting series of sequences into a kind
of pyramid whose apex touches on the next higher level, that of Actions.
Barthes's treatment of these two higher levels (Actions and Narration)
is both more perfunctory and more derivative. It does not call for much
comment here, except to note the extent to which Barthes is prepared
to go along with those essentially reductive narratological procedures
which would like to boil down all the world's narratives into three main
semantic axes (communication, desire or quest, and ordeal) and would
similarly confine 'the infinite world of characters' within a narrowly
limited cast of 'actants'. It will have been noticed, of course, that in
this early foray Barthes's whole approach has been not only highly
systematised but also highly abstract; the only approaches to specificity
are in fact contained within a few scattered and fragmentary references
to Ian Fleming's Goldfinger. In these circumstances some doubt must arise
as to whether the carefully-elaborated analytic system would prove flexible
enough to encompass the complexity of a more genuinely sophisticated
narrative work.
Some such question may have presented itself to Barthes, for in the
1971 interview already mentioned, he spoke as follows:
In the former (1966) text ['Introduction to the structural analysis
of narratives'] I appealed to a general structure from which would
then be derived analyses of contingent texts . . . I postulated the profit
there would be in reconstructing a sort of grammar of narrative, or
a logic of narrative, (and at that period, I believed in the possibility
of such a grammar - I do not wish to deny it) . . . In S/Z I reversed
this perspective: there I refused the idea of a model transcendent to
several texts (and thus all the more so, of a model transcendent to
every text) in order to postulate . . . that each text is in some sort
its own model, that each text, in other words, must be treated in
its difference . . .
How far this is accurate in its proclamation of a complete reversal of
perspective on Barthes's part will be a matter for subsequent discussion.
 ROLAND BARTHES'S NARRATOLOGY 45

What the comment does show is that Barthes regarded S/Z as a radically
new departure - and therefore as itself a text which needs to be treated
'in its difference'.
S/Z is indeed unique; it is also uniquely difficult to get to grips with.
Its 217 pages consist almost entirely of a detailed analysis of a litde-known

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
short story by Balzac entided 'Sarrasine', die complete text of which (some
12,000 words) is printed as an appendix at die end of the volume.3 For
the purposes of his commentary Barthes has cut up Balzac's text into
561 consecutively-numbered fragments which he terms units of reading
(or iexias'). This division is professedly arbitrary, based only on the
principle diat each lexia should have 'at most three or four meanings to
be enumerated'; consequendy die Iexias vary considerably in length, some
containing only a few words, others several sentences. The main structure
of the book consists of Barthes's notes on each lexia in turn, but these
notes are interrupted periodically by a series of 93 'divagations' each of
which discourses radier more fully upon some issue raised by the Iexias.
In die discussion diat follows (as in Barthes's text) the Iexias will be
signalled by Arabic numerals, and die divagations by Roman numerals.

' Balzac's tale begins with an evening party at the Lantys' Parisian mansion
where the narrator observes a strange almost ghost-like old man, known to be
a prot6g6 of the Lantys', whose decrepitude both scares and intrigues the young
Marquise he has escorted to the party. Her curiosity extends also to a painting of
Adonis ('too beautiful for a man', she pronounces) which according to the narrator
was copied from die statue of a woman yet also has as its subject a relative of
Madame de Lanty: it is agreed between them that the mystery shall be unveiled
at an assignation the following evening. The narrator's 'tale widiin a tale' relates
the history of a young provincial named Sarrasine whose phenomenal artistic talent
takes him to Paris as the pupil of the great sculptor Bouchardon. After six years
there he travels to Rome where within a fortnight he conceives a violent passion for
La Zambinella, the star of the Opera, whose ideal beauty seems to combine in a
single human form all the womanly perfections he has hitherto only dreamed of;
admitted by the other singers to a wild party at which she is present, he finds that
her timidity only enhances her feminine appeal to him. Since his advances have
been determinedly repulsed he plans to kidnap her after a concert at the French
ambassador's palace, but before this can be accomplished he is informed by an
Italian nobleman that there are no women on the Roman stage and that Zambinella
is really a castrate Unwilling to believe this, Sarrasine nevertheless removes the
kidnapped Zambinella to his studio where the frightened youth recognises his
own features on the statue of a beautiful woman. Unutterably disgusted at the
trick diat has been played on him, Sarrasine welcomes his death at the hands
of murderers sent to assassinate him by the Cardinal who is the castrato's protector.
The Adonis, the narrator explains, was a painting inspired by Sarrasine's statue
of Zambinella; while Zambinella, the source of the Lantys' fabulous wealth, was
himself the old man seen at the Lantys' party. Repelled by the story of the castrato,
the Marquise declares her intention of remaining virtuous after all.
46 THE CAMBRIDGE QUARTERLY

However, despite this apparent concentration upon the specificity of

a single text, Barthes announces from the outset (see I) an intention to
attain a more widely generalised goal - namely to laud the virtues of what
he calls 'plurality' in both writing and reading. For him the acme of
achievement in writing has now become what he calls the 'writerly'

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
[scriptible] - an ideal (rarely to be found, it seems, even in ultra-modern
literature) which in its infinite openness of signification makes the reader
'no longer a consumer, but a producer of the text'. 4 By contrast the
overwhelming majority of literary works are merely 'readerly' [lisible] ; 5
in reading them the unhappy reader (instead of being able to enjoy die
'magic pleasure' of doing the writing himself) has only 'die poor freedom'
of being able to accept or reject what has been already written. To round
off his first divagation Barthes declares widi a grand flourish: 'We call any
readerly text a classic text.' This undoubtedly includes 'Sarrasine' - but
die bent of all die ensuing commentary is to demonstrate ingenious ways
in which die reading of a classic and readerly text can nevertheless

4
The reader is invited to sample for himself the power of the rhetoric with
which Barthes clothes his ideal of the (non-existent?) plural text:
The writerly text is a perpetual present, upon which no consequent language
(which would inevitably make it past) can be superimposed; the writerly text
is ourselves writing, before the infinite play of the world (the world as function)
is traversed, intersected, stopped, plasticised by some singular system
(Ideology, Genus, Criticism) which reduces the plurality of entrances, the
opening of networks, the infinity of languages. The writerly is the novelistic
without the novel, poetry without the poem, the essay without the dissertation,
writing without style, production without product, structuration without
juncture . . . Let us first posit the image of a triumphant plural,
unimpoverished by any constraint of representation (of imitation). In this
ideal text, the networks are many and interact, without any one of them being
able to surpass the rest; this text is a galaxy of signifiers, not a structure of
signifieds; it has no beginning; it is reversible; we gain access to it by several
entrances, none of which can be authoritatively declared to be the main one;
the codes it mobilises extend asjar as the eye can reach, they are indeterminable
(meaning here is never subject to a principle of determination, unless by
throwing dice); the system of meanings can take over this absolutely plural
text, but their number is never closed, based as it is on the infinity of language.
Speaking for myself, I am not gready enticed by the prospect thus held out.
5
Richard Miller's choice of writerly and readerly as equivalents for scriptible and
lisible is not entirely happy; thus writerly certainly loses the overtones of 'film-
script' which seem to be present in the coinage. In her Beautiful Theories (1982),
pp. 371-9, Elizabeth W. Breuss has some pertinent pages on the problems
attendant on the translation into English of Barthes's highly distinctive prose.
I have used the standard translations by Stephen Headi, Richard Miller and
Richard Howard, but on occasion have inserted the original French word or
expression in square brackets.
 ROLAND BARTHES'S NARRATOLOGY 47

encompass a considerable degree of plurality. As Barthes himself puts

it, in discerning and pointing out meanings which are not immediately
apparent the aim is to establish not 'the truth of the text (its profound,
strategic structure), but its plurality (however parsimonious)' (VIII). In
thus 'unraveling' the text Barthes explicitly disclaims any intention of

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
setting forth 'die criticism of a text' [la critique d'un texte]; instead he
wants to disclose 'the semantic substance (divided but not distributed)
of several kinds of criticism (psychological, psychoanalytical, thematic,
historical, structural)', each of which he describes as 'hearing one of the
voices of the text'. In reality, nevertheless, there is implicit throughout
S/Z's account of 'Sarrasine' a distinctive (and somewhat idiosyncratic)
critique of Balzac's tale, one which is moreover very close in its main
oudines to the rather crudely psychoanalytical interpretation set out in
the 1967 article by Jean Reboul which had guided Barthes to this story
in die first instance.6
Barthes's key instruments in this project of sketching 'the stereographic
space of writing' are die five 'codes' - concepts die precise status of which
(in S/Z) will need fuller consideration later. For die moment it will perhaps
be enough to quote the characterisation accorded to diem by Jonadian
Culler. 'Each code', he writes, 'is the accumulated cultural knowledge
that enables a reader to recognise details as contributions to a particular
function or sequence'. If diis seems less dian crystal clear, more
illumination may appear as we look at each of the codes in turn.
The first to appear (Barthes refuses to put them in any order of
importance) is die hermeneutic code (HER), which.sets out 'the various
(formal) terms by which an enigma can be distinguished, suggested,
formulated, held in suspense, and finally disclosed'. The point being made
here is mat a narrative text often holds die reader's attendon by presenting
some puzzle, and then keeping curiosity alive by a succession of partial
disclosures or false clues, until the solution.is finally revealed. The, first
example given by Barthes is perhaps somewhat untypical. The very tide,
he claims, of Balzac's tale 'raises a question: What is Sarrasine? A noun?
A name? A diing? A man? A woman? This question will not be answered •
until much later . . . " More representative is die example (28 et.seq.)
of die mystery of the identity of die old man encountered at. die Lantys'
party, a puzzle die solution to which is held back till late in.die narrative. '
Next comes the semic code (SEM) - a matter of picking up secondary
meanings (semes) carried by words, in their relationship with, odier
segments of die text. These semes work essentially dirough connotation

' Jean Reboul, 'Sarrasine ou la castration personnifiee', Cahiers pour VAnalpe,

mars-avril 1967, 91-6.
48 THE CAMBRIDGE QUARTERLY

rather than denotation, but the connoted meanings are usually specific
to the context rather than inherent in die word's general linguistic usage;
and they tend towards building up an impression of some character or
object or place. Culler states it rather too crudely in saying that: 'The
semic code provides cultural stereotypes (models of personality, for

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
example) that enable readers to gather pieces of information to create
characters'; but it is true diat in XII Barthes refers to die semes as 'the
Voice of die Person', and diat many of his examples relate to character
(details for instance which tend to suggest a quality of 'femininity' in
Zambinella). A slightly different example is that of the seme Wealdi (4)
as signalled by the chimes of 'die clock of die Elys6e-Bourbon', which
indicate diat die Lantys' mansion is situated in a wealdiy neighbourhood.
The diird code to be mentioned is die symbolic code, and diis is defined,
with mystifying casualness, as 'the place for multivalence and for
reversibility'. It is perhaps assumed diat we all know what symbolism
is anyway; in fact, die examples diat accumulate suggest diat die symbolic
systems in question are mosdy on a grandiose, perhaps even archetypal,
scale. Thus die first one mentioned (2) opens up an extended 'series of
antidieses: garden and salon, life and death, cold and heat, outside and
interior'; while a later sequence recorded under die heading 'SYM. Axis
of Castration' (from 21 **** onwards) divides up die characters radier
arbitrarily into diose who are symbolically castrated and those who
symbolically castrate.
The fourth code is that of the code of actions (ACT) - or, to dignify
it widi Barthes's somewhat fancy tide for it, die proairetic code. This
code is clearly to some extent a successor to the 'cardinal functions' as
set out in the 1966 'structural analysis'; diere is die same preoccupation
widi 'named sequences' (a kidnapping, for example) in which a self-
contained grouping of data (nuclei?) enables die reader to pick up, on
die basis of stereotypes derived from previous reading ('die already-read'),
a preliminary sense of what is going on, and whidier die action is tending.
The sequences can be large-scale and melodramatic, but they can also
be quite trivial. Thus in regard to 2 ('/ was deep in a daydream') Barthes
comments 'The state of absorption formulated here (I was deep in . . .)
already implies (at least in "readerly" discourse) some event which will
bring it to an end . . . Such sequences imply a logic in human behaviour.'
This reference to logic, too, picks up die drift of Barthes's earlier attempt
to construct a generalised 'grammar of narrative'.
The final code is die cultural code (REF) which embodies 'references
to a science or a body of knowledge' and 'affords the discourse a basis
in scientific or moral authority'. It seems clear diat die terms 'science'
and 'knowledge' are to be interpreted widi some freedom, however; and
 ROLAND BARTHES'S NARRATOLOGY 49

indeed the examples again suggest reliance on a variety of stereotyped

notions, as in 24 REF where the phrase 'dark as a Spaniard, dull ds'a banker'
is cited as a reference to 'psychology of peoples and professions'. A
stereotype may also, however, be invoked merely by departing from it,
as in 170 R E F : 'He worked all day, and in the evening went out to beg for his
living.' Here Barthes's comment is: 'Stereotype: the poor, courageous

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
artist (earning his living by day, creating at night or, as here, vice
versa).'
In choosing 'code' as the name for his five interpretative instruments,
Barthes raised problems which are not easily resolved. The term was,
of course, a fashionable one in the structuralist climate of the 1960s, and
was often used rather loosely. (One recalls, for instance, Basil Bernstein's
dubious claim to have identified in the speech of working-class and middle-
class children two distinct 'codes', the 'restricted' and the 'elaborated'.)
More typically, the word was used to denote those elaborate conventions
of non-verbal behaviour whereby the members of any human society are
constantly communicating with each other in ways that either reveal, or
lay claim to, their membership of a particular social or class grouping.
Thus LeVi-Strauss had developed a complex analysis of the socio-cultural
significance of different types of eating habit and food-menu; while Barthes
himself had attempted to describe the code of fashion in which differing
colours, shapes and materials in clothes convey a social message. It is
difficult to see that the codes of S/Z quite fit in with this use of the term.
Nor on the other hand do the hermeneutic code (HER) and the proairetic
code (ACT) match up at all obviously with Jonadian Culler's formulation
(quoted above). As first presented, they don't seem to depend on prior
cultural knowledge, and their organisation seems to be intrinsic to the
narrative flow, rather than extrinsically-derived. It is relevant to note diat
Barthes himself recognised a distinction between these two codes and the
other three. In XV he writes: '. . . of the five codes, only three establish
permutable, reversible connections, outside the constraint of time (the
semic, cultural, and symbolic codes); the other two impose their terms
according to an irreversible order (the hermeneutic and the proairetic
codes).' We shall return to this issue later.
First, however, what can be said about the (literally) thousands of
detailed comments offered by Barthes on the 561 lexias? It must surely be
agreed that they are a very mixed bag, some being moderately perceptive,
others elaborately fanciful, others frankly banal. How otherwise than
under the latter heading can one describe, for instance, the apparent
poindessness of 92, where the total entry reads as follows:

'He smells like a graveyard,' cried the terrifiedyoung woman, * S E M . Deadi.

50 THE CAMBRIDGE QUARTERLY

His perceptiveness is less easy to exemplify since it requires reference

to the narrative context. One might point to the adroitness with which
Barthes picks up in some of the opening lexias (5,8,9,10 under SYM and
SEM) the suggestions carried within Balzac's text by the Antithesis between
salon and garden, warmth and cold, the Living and the Dead - but what
looks at first sight like sensitivity is less impressive when it turns out to be

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
merely an anticipatory recognition of one of Balzac's more melodramatic
touches (31), the sensation of cold carried around with him by the ghost-
like old man at the party. Again one senses a certain shrewdness in
Barthes's comments on the description of the old man's voice (66):
[He sat slowly down on his seat, with circumspection,] muttering some
unintelligible words. His worn-out voice was like the sound made by a stone
falling down a well. * The noise of a stone falling down a well is not
a 'worn-out sound'; however, the chain of connotation of the sentence
is more important than the exactitude of the simile; this chain links
the following elements: the inanimate inertia of the stone, the
sepulchral depth of the well, the discontinuity of the aged voice,
antinomic to the perfect, unified, 'lubricated' [lubrifie'e] voice;
the signified is the 'thing', artificial and creaky as a machine
(SEM.Mechanicalness).
But the responsiveness to verbal nuance displayed here seems to have
been straitjacketed by a compulsion towards generalising definition, an
ambition born of scientism. A provisional conclusion might be that a
literary (or linguistic?) sensitivity is there, but is often misused owing
to an absence of responsible evaluative concern.
This irresponsibility can be studied at length in the extraordinarily far-
fetched attribution of symbolism in the 'Antithesis' labelled 'SYM. Marriage
of the Castrato', as set out in 64, 91 and 97, and expatiated upon even
further in XXVII under the title 'ANTITHESIS II: THE MARRIAGE'.
In Balzac's tale there has, of course, been no marriage. All that has
happened is that at the Lantys' party the old man whose identity is
ultimately to be revealed as that of Sarrasine's castrato has sat down close
to the narrator's lady-friend; out of horrified curiosity she nerves herself
to touch him, at which point there issues from his dried-up throat 'a cry
like a rattle'. This sound is apparently the first 'catastrophic conclusion'
to follow from the so-called marriage; in the cloudily portentous prose
of XXVII Barthes goes on to pronounce the 'marriage' of the young
woman and the castrato as 'doubly catastrophic', in that 'Antithesis
cannot be transgressed with impunity' and that, symbolically, later in
the tale 'the castrato will transgress morphology, grammar, discourse,
and because of this abolition of meaning Sarrasine will die'. It is in
 ROLAND BARTHES'S NARRATOLOGY 51

extravagant ways such as this, evidently, that the 'text' is to be 'written'

(or re-written) instead of being 'read'. It must be said that Barthes does
seem to have an obsession with castration, which assumes a central
importance in his version of the tale as an all-embracing theme; few of
Balzac's characters (whether male or female) are allowed to escape the

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
fate of castration, symbolically at least. Another example of this kind of
distortion may be noted in 373 where La Zambinella's threat that she
will defend herself against rape by plunging her dagger into Sarrasine's
heart is described by Barthes as a threat of 'mutilation' and categorised
under the heading of 'SYM. castration, the knife'.
This issue of 'rewriting' is faced more openly in relation to Sarrasine's
infatuation with La Zambinella and in particular his first encounter with
her. This is the opening of LIII (EUPHEMISM):

Here is one story about Sarrasine: he goes into a theatre; the beauty,
voice, and art of the star enrapture him; he leaves the hall overcome,
determined to recapture the enchantment of the first evening by
renting a box near the stage for the entire season. Now here is another
story about Sarrasine: he enters a theatre by chance (206), by chance
he is seated near the stage (209); the sensual music (213), the beauty
of the prima donna (220) and her voice (231) fill him with desire;
because of his proximity to the stage, he hallucinates, imagines he
is possessing La Zambinella (242); penetrated by the artist's voice
(243) he achieves orgasm [iV en vient a I'orgasme] (244); after which,
drained (247), sad (248), he leaves, sits down and muses (249); this
was in fact his first ejaculation [sa premiere jouissance]; he decides to
reexperience this solitary pleasure every evening, domesticating it
to a point where he can experience it whenever he wishes.

Now what one notices first about this 're-writing' is the extent to which
Barthes is perverse in relation to matters of detail. Thus 'Jomelli's sublime
harmony' (213) becomes 'sensual music'; 'the innermost depth of his being . . .
what we call the heart' (233) becomes 'the sexual organ'; and various other
indications in Balzac's text which signal Sarrasine's fascination by the 'ideal
beauty' of the prima donna are gathered together as members of a proairetic
sequence on which Barthes bestows the name 'ACT. "Seduction"' (219,
231, 232, 233, 234). The most remarkable of these transformations,
however, is Barthes's flat assertion: 'he achieves orgasm'. This is a
re-writing of the following sentence:

(243) Last, this agile voice, fresh and silvery in timbre, supple as a thread
shaped by the slightest breath of air rolling and unrolling, cascading and
scattering, this voice attacked his soul so vividly (244) that several times he
52 THE CAMBRIDGE QUARTERLY

gave vent to involuntary cries torn from him by convulsive feelings of pleasure
(245) which are all too rarely vouchsafed by human passions.

On the centra] part of this (244) Barthes's comment is 'ACT. "Pleasure":

ejaculation' [jouissance] - and this despite the fact that Sarrasine is at

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
the time seated in the theatre 'squeezed between two notably fat abbati'. Even
more remarkable is Barthes's repeated assertion (211, 244) that this is
the young sculptor's 'first' ejaculation [sa premiere jouissance] - an assertion
based largely on the report that Sarrasine's patron Bouchardon (cast
thereby in the role of 'castrator') had kept his pupil 'in total ignorance of
the facts of life f I'ignorance profonde sur Us choses de la vie]'' (186), and held
to despite the explicit testimony that in Paris Sarrasine had had as mistress
for a short time ' Clotilde, one of the luminaries of the Opera' (192). Compared
with this it is perhaps only a minor eccentricity that leads Barthes to
describe the 'inexplicable sadness' which overtakes Sarrasine on leaving
the theatre as "'post coitum" sadness' (248). (Coitus? masturbation?
what difference does it make?)
Barthes would undoubtedly argue (LIII) that it is not the details that
matter; what counts is the structure that holds them together, their
'systematic cohesion'. There is, it is claimed, a'congruence of relationships'
linking the two versions, and to assert any interpretation as the correct
one is merely an attempt to 'intimidate' a reader into reading in only
one among a number of possible ways. As against this it needs to be said
that if one follows Balzac's text attentively, and takes the tale as a whole
(and not as a series of discrete lexias), one is led ineluctably to a certain
constellation of experience which 'hangs together' in one's mind and
excludes the possibility of competing versions (Barthes's 're-writing'
among them). The achievement which marks Balzac's Sarrasine (and this
is perhaps its sole real distinction) is that of capturing some of the
inwardness of a young man's infatuation for an imagined 'ideal woman'
and realising in fictional terms the extent to which this ideal is tied in
with a fragile and essentially unreal conception of 'femininity'. (Barthes's
obsession notwithstanding, Sarrasine is not in love with 'castration'
(cf 277), but with an imagined prima donna whose charm is destroyed
once 'she' is revealed to be a castrated boy.) I would not claim any very
great merit for this minor Balzac work, but it is at least of considerably
greater interest than Barthes's crude and clich6d rewriting.
It is evident, however, that the liberties Barthes allows himself to take
are not at all random; there is a system, or at any rate a strategy,
underlying them. To gain an understanding of how it works we may start
with a lexia (343) relating to Sarrasine's conversation with La Zambinella
during his first actual meeting with her:
 ROLAND BARTHES'S NARRATOLOCY 53

and she revealed herself to be weak and superstitious. * Weakness and

superstition are equivalent to pusillanimity (SEM. Pusillanimity).

Now what, one may ask, is achieved by thus substituting for Balzac's
two adjectives a single word which is only very loosely equivalent in

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
meaning? The explanation must be that at this stage of the game Barthes
is collecting a series of instances of La Zambinella's behaviour which can
be labelled, roughly, as manifestations of'pusillanimity' (347, 377, 433,
438) and which can be amalgamated, in 439 and 440, with another string
of semic meanings signifying 'Femininity', and additionally, in 440, with
a third string labelled 'Timidity'. In justification of this procedure it might
be said that it draws attention (however rough-handedly) to a significant
feature of Balzac's artistry, namely one of the means whereby he keeps
alive that ambiguity about the castrato's sexual identity on which
Sarrasine's misapprehension depends. What is significant, however, is
the terms in which Barthes in the immediately ensuing LXXIV
characterises the apparent activity of the author in this instance (yes,
inconsistently he does allow the author, elsewhere alleged to be 'dead',
an active and visible presence in the text at this point):

A classic [sc. readerly] narrative always gives this impression: the

author first conceives the signified (or the generality) and then finds
for it, according to the chance of his imagination, 'good' signifiers,
probative examples; the classic author is like an artisan bent over
the workbench of meaning and selecting the best expressions for the
concept he has already formed. For instance, timidity: one selects the
sound of a champagne cork, a story of a snake, ar story of
highwaymen.

It is far from evident that this hypothesis (it can be no more) about the
audior's way of working fits the particular chunk of Balzacian text to
which Barthes has tied it; Barthes has, after all, himself arrived at his
'generality' by arbitrarily grouping together a number of specific
descriptive or narrative touches to form a stereotyped general concept
of his own. But however this may be, there can certainly be no warrant
for Barthes to extend his questionable hypothesis to all classic texts. The
passage points to a radical uncertainy in S/Z as to whether its observations
relate to a single text which is 'its own model' or whether-they lay claim
to a wider application.
Moreover it is not only in relation to character-portrayal (which falls
within the province of SEM or semic meaning) that Barthes sets out an
account of the genesis of the text as a matter of starting with a generalised
.(and. stereotyped) concept which is then clothed with appropriate
54 THE CAMBRIDGE QUARTERLY

illustrative detail. Under the heading of the proairetic code (ACT) a

remarkably similar pattern is said (in LXVII HOW AN ORGY IS
CREATED) to be discernible behind Balzac's description of the party
at which Sarrasine meets Zambinella. In this case, however, Barthes takes
care to avoid allotting the author any active role by presenting his

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
diagnosis in the passive voice (the 'orgy is announced . . . it is named . . . "
and so on). Balzac does in fact use the word 'orgy' twice, rather en passant,
in relation to the party's development (310 'ready to begin an artists' orgy'
and 356 ' When the supper became an orgy'). Barthes fastens on the term as
the title for a 'named sequence' which he categorises as ACT. 'Orgy',
and under which he lists no fewer than thirteen lexias (310, 313, 337,
339, 345, 356, 357, 358, 360, 361, 362, 390) which correspond to the
details constituting his (Barthes's) concept of an orgy. To quote from
LXVII itself: 'These parts of the orgy are stereotyped items of behaviour,
created by a repetition of experiences (supper, uncorking champagne,
singing, licentiousness) . . . Further, a certain moment of this orgy, given
under its generic name, abandonment, can no longer be analysed like the
orgy itself, but illustrated by certain typical items of behaviour (sleeping,
spilling wine) . . . " Again it may seem that this interpretation of the
novelist's way of working (from generalised concept to selection of
illustrative detail) is somewhat strained - a wrenching of the data to fit
in with a pseudo-scientific parti pris. In this instance, however, it is of
some interest to note the conclusion to which Barthes is led:
What we have called the proairetic code is thus itself made up of other
diverse codes based on different disciplines.
In other words the (proairetic) code of actions is to be seen in the end
to rely upon the other cultural codes (SEM, SYM and REF) and to be
dependent like them on what Culler called 'the accumulated cultural
knowledge which enables a reader to recognise details as contributions
to a particular function or sequence'. 7

7
In support of Culler's interpretation one might cite the following from the
later (1974) analysis of 'A Tale by Edgar Allan Poe':
The word code itself should not be understood in therigorous,scientific sense
of the term. The codes are simply associativefields,a supratextual organisation
of notations which impose a certain notion of structure; the instance of the
code, for us, is essentially cultural; the codes are certain types of already-
seen, of already-read, of already-done . . .
There does seem however to be a certain not uncharacteristic vacillation (or
confusion?) between two opposed standpoints. On the one hand there is said to
be an inner logic to an acrional sequence: if the telephone rings, it must be either
answered or ignored; if a character commences on a journey, the journey must
 ROLAND BARTHES'S NARRATOLOCY 55

What, more precisely, is the nature of this 'accumulated cultural

knowledge'? Barthes is at times inclined to hint that it can be reduced
simply to a knowledge of what has been previously written (the 'already
written, already read'). Thus he writes (LXXII) that 'the "realistic" artist
never places "reality" at the origin of his discourse, but only and always,

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
as far back as can be traced, an already written real, a prospective code,
along which we discern, as far as the eye can see, only a succession of
copies.' Again, he describes the 'cultural codes' (in LXXXVII) as
'entirely derived from books' so that they can only appear to establish
reality; while a similar line of thought seems to underlie the tortuous
argument whereby, in XXIII, he claims that in descriptions the most
a writer can ever do is to 'de-depict' [aV-peindre] an object or a view,
so that realism 'consists not in copying the real but copying a (depicted)
copy of the real'. Yet he doesn't follow this dirough to the uncompromising

be either completed or interrupted. On the other hand such expectations are said
to arise not from logic but from cultural experience (and in particular from
experience of the 'already-written'). Thus in his 1972 essay 'Apropos of Acts
10-11' Barthes writes:
Personally, I incline toward the notion of a sort of cultural logic which owes
nothing to any mental datum, even on an anthropological level; for me, the
sequence of narrative actions are encased in a logical appearance which comes
solely from the already-ivritten: in a word, from the stereotype.
Now the examples of actional sequences mentioned by Barthes in this context
are of a rather trivial kind; someone knocks on a door, for instance, and the door
must either be opened or not opened. But even in 'stereotypes' of this nature
the logic involved is surely not merely a matter of appearance only; an author
would rarely see any point in mentioning that a character embarked on a journey
unless the outcome of the journey (its completion or its interruption) was also
going to be announced later. Or to take as an example an actual (and less
insignificant) action, when in Wuthering Heights Isabella hands to Nellie Dean a
letter for Edgar Linton surely it is narrative logic which demands that the letter
be either delivered or withheld. And similarly in the case of the hermeneutic code
there would be little point in posing an enigma unless it were intended eventually
to disclose an explanation of the puzzle. It is difficult to avoid the conclusion that
both these codes differ from the other three in containing within their own structure
a logic of expectation. This does indeed seem to be the side on which Barthes
comes down in his last reflections on the issue at the end of 'A Tale by Edgar
Allan Poe':
despite the floating character of structuration, in the classical, readerly
narrative (such as Poe's tale), there are two codes which maintain a vectorised
order, the actional code (based on a logico-temporal order) and the code of
the Enigma (the question is crowned by its solution); an irreversibility of
narrative is thereby created.
It is indeed these two codes that Barthes expects the truly 'modem' text to subvert,
in order to achieve the desired goal of complete plurality.
56 THE CAMBRIDGE QUARTERLY

Derridean view that a text can have meaningful existence only as reference
to or commentary on other texts, keeping instead one foot at least in the
world of common experience. Thus in XII while describing the code as
'a perspective of quotations', he goes on to call the 'units that have resulted
from it' not only 'so many fragments of something that has always been

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
already read . . . " but also fragments of something that' [has always been
already] . . . seen, done, experienced . . .', thereby leaving a space to
be occupied by living as well as by reading. And in LXXXVI after
referring to the proairetic [action] sequences as 'deriving from a
patrimonial hoard of human experiences', he goes on more explicitly to
comment that they
can be assigned no logic other than that of the probable, of empirics,
of the 'already-done' or 'already-written', for the number and order
of their terms vary, some deriving from a practical reservoir of trivial
everyday acts (to knock at a door, to arrange a rendezvous) and others from
a written corpus of novelistic models (the Abduction, the Declaration of
Love, the Murder) . . .

Formally at least, this does recognise and spell out the role played by
remembered experience as well as by remembered quotations; but in the
spelling out it also suggests a relative unimportance for that which has
merely been 'seen, done, experienced', and this emphasis can fairly be
held to represent the impression left by S/Z as a whole. Moreover the
'novelistic models' offered as representative of 'the already-written' are
so blatantly stereotypical as fatally to weight the argument against
admitting any vestige of originality which the author (before his 'death')
might have been suspected of introducing into his version of the 'text'.
One can see that, thus conceived, the codes - 'these voices (whose origin
is "lost" in the vast perspective of the already-written)' - must indeed
'de-originate the utterance' (XII).
The whole of the analysis which Barthes sets out in this book at such
confused and confusing length does, in fact, seem most applicable to that
compulsive shuffling around of worn-out stereotypes that constitutes
popular or 'commercial' fiction. Yet Barthes is not here writing about
an Ian Fleming confection; his theme is 'classic narrative', and his
specific example is a story by Balzac whose departures from the expected
he sometimes chronicles in minor matters (as in 170 REF, mentioned
above) while at the same time he turns a determinedly blind eye to that
breath of originality which distinguishes the central theme of Balzac's
tale. But despite some protestations to the contrary Barthes seems to have
a thoroughgoing contempt not only for Balzac but for almost all literature
of the past, as these sentences from LXXXVII surely reveal:
 ROLAND BARTHES'S NARAATOLOGY 57

'Life', then, in the classic text, becomes a nauseating [koeurant]

mixture of common opinions, a smothering layer of received ideas:
in fact, it is in these cultural codes that what is outmoded in Balzac,
the essence of what, in Balzac, cannot be (re)written, is concentrated.
What is outmoded, of course is not a defect in performance, a personal

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
inability of the audior to afford opportunities in his work for what
will be modern, but rather a fatal condition of Replete Literature
[ Pleine Litterature], mortally stalked by the army of stereotypes it
contains.
It is a strange paradox that when die writer of these words 'rewrites' a
portion of his chosen classic text die result turns out to be more unfailingly
clich6d than anything his audior from the past ever perpetrated even on
his worst off-days.
Bardies has been much praised for what Geoffrey Strickland has termed
'die power and charm of his prose' - a quality which is obvious only radier
intermittendy in S/Z. What has been insufficiendy remarked on is a certain
strain of elegandy recherche silliness which is literally at die centre of
S/Z, placed diere widi self-conscious contrivance in Divagation XLVII
(the middle one of die ninety-diree), in order to reveal not only die
author's celebrated charm but also die solution to die 'enigma' proposed
by his eccentric title. I shall quote it in full, widiout comment:
XLVII S/Z
SarraSine: customary French onomastics would lead us to expect
SarraZine: on its way to die subject's patronymic die Z has encountered
some pitfall. Z is the letter of mutilation: phonetically, Z stings like
a chastising lash, an avenging insect; graphically, cast slantwise by
the hand across the blank regularity of die page, amid die curves
of the alphabet, like an oblique and illicit blade, it cuts, slashes, or
as we say in French, zebras; from a Balzacian viewpoint, diis Z (which
appears in Balzac's name) is die letter of deviation (see die story
Z.Marcas); finally, here, Z is die first letter of La Zambinella, die
initial of castration, so diat by diis orthographical error committed
in die middle of his name, in the centre of his body, Sarrasine receives
the Zambinellan Z in its true sense - die wound of deficiency. Further,
S and Z are in a relation of graphological inversion: the same letter
seen from the other side of die mirror: Sarrasine confronts in
La Zambinella his own castration. Hence die slash (/) confronting
the S of SarraSine and the Z of Zambinella has a panic function;
it is the slash of censure, the surface of die mirror, the wall of
hallucination, die verge of antidiesis, die abstraction of limit, the
obliquity of die signifier, die index of die paradigm, hence of meaning.
58 THE CAMBRIDGE QUARTERLY

Clearly, whatever one makes of it, this is (even in translation) sheer

Roland Barthes, Barthes sui generis.
Two other features which deserve comment in S/Z seem to be not so
individually distinctive, but more like manifestations of a trend common
to much post-structuralist critical writing. The first is a readiness to

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
personify aspects of a text as though words were living entities capable
of doing things to each other - a tendency which seems even odder when
considered in conjunction with the decision to remove from the ('dead')
author his power to do anything to or with words. Thus in LX Barthes
writes that: 'This [the reticence of the discourse] shows that the discourse
is trying to lie as little as possible . . .'; while in LXXXI we are told, with
seeming authority, that 'semes can migrate [emigrer] from one figure
[personnage] in the text to another' - under their own steam, it appears.
The second way in which S/Z falls into line with much post-structuralist
criticism is its eagerness to treat a work of imaginative literature not as
a human document but as a demonstration of some abstruse literary
theory. Thus according to Barthes (in XXXIX) 'Sarrasine's adventure
is not the main story', and 'Sarrasine is not the story of a castrato, but
of a contract'. In support of this strange assertion Barthes offers the theory
that 'in order to be produced, narrative . . . must subject itself to an
economic system' - a claim exemplified in this instance by the implied
contract between the narrator and Madame de Rochefide that in exchange
for the story of the old man she will yield up to him her body ('a night
•of love for a good story'). In the interests of theory the compelling human
content of the young sculptor's misguided infatuation for a castrato is
thus displaced from the centre of attention, its place to be taken by a
frame narrative which Balzac had sketched in only rather clumsily, and
indeed somewhat perfunctorily. It seems doubtful if literature, whether
readerly or writerly, would continue to be much read if its true subject-
matter were indeed little more than the nature of its own structures.
One is inclined to doubt, for that matter, whether many of those who
profess acquaintance with S/Z can have had the persistence to read
steadily all the way through it. But despite its irritating obscurities and
eccentricities, does it not offer some worthwhile insights to justify its wide
influence and high reputation? For my own part I can testify to no more
than two of these which seem likely to prove of significant and lasting
benefit. In the first place there is the way Barthes brings home to one,
by sheer accretion of detailed comment, the extent to which the reader's
response to a narrative depends on bringing up out of one's experience
of living a vast store of accumulated cultural knowledge - knowledge of
language and linguistic forms, of human nature and psychology, social
conventions and social history, economic realities, climate of ideas, local
 ROLAND BARTHES'S NARRATOLOGY 59

and national customs, and so on. (And, yes, knowledge of what had been
previously written under these various heads both in narrative and in
non-narrative contexts - though this is not to be taken as endorsing any
suggestion that the 'already written' remains, in worthwhile literature,
inert and untransformed by the new purpose brought to it by a genuinely

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
'living' writer.) A recognition of the sheer amount of such cultural reference
throws light on the nature of the disadvantage one feels under when trying
to engage with the literature of a culture one has never lived in - a
disadvantage surely proportionate to the distance between that culture
and one's own. I am not convinced that Barthes's attempts to categorise
such knowlege under the heading of various codes are particularly helpful,
unless perhaps as a means of drawing attention to its diversity; and the
danger of such unremitting concentration upon references external to the
text is of course that of reducing one's capacity for attention to the specific
transforming significance which the author has given to them in the
context of his own creative reshaping. Nevertheless I am inclined to believe
that in this respect die experience of exposing myself at length to S/Z will
make a permanent difference to all• my reading'of fiction in the future.
The second insight of lasting benefit is that centred round the
hermeneutic code. Barthes offers here a perceptive account of the ways
in which compelling involvement in a narrative may be generated not
so much by the question 'What next? What next?' as by curiosity to
discover the answer to some puzzle or enigma which the writer has put
before the reader. Once fully alerted to this, one becomes aware'that this
method of getting a reader 'hooked' is more widespread than one had
thought - and of course by no means confined to detective stories. In his
Recognitions (1988) Terence Cave has suggested that Barthes's hermeneutic
code is essentially a variant on the Aristotelian concept of anagnorisis;
and it is certainly true that the 'enigmas' that come most readily to mind
(the source of Pip's mysterious benefaction in Great Expectations, or the
provenance of Jane Fairfax's piano in Emma) can usually be reduced to
die recognition, discovery or unveiling of some actant's true identity. On
die other hand it seems desirable to keep clear in our minds die distinction
between an anagnorisis in which what is revealed is an identity concealed
from other actants but already known to die reader (as in Conrad's Under
Western Eyes, for instance) and the true' Barthesian enigma in which the
hermeneutic code makes play with a curiosity which the text has excited
in die reader himself. Barthes's decidedly original formulation in S/Z of
this aspect of narrative technique surely carries us some way beyond his
earlier dieories. of narrative structure. • • . . , •
The publication in 1970 of S/Z did not mark the. end of Barthes's
involvement widi structural analysis. While still engaged in writing it
60 THE CAMBRIDGE QUARTERLY

he had published (in 1969) an article entided 'The Structural Analysis

of Narrative; Apropos of Acts 10-11' which was to turn out to be the
first of two analyses of excerpts from the Bible. In 'Acts 10-11' he is
still embroiled in the structuralist ambition to reconstruct 'a general
structure . . . a sort of grammar of narrative . . . from which would dien

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
be derived analyses of contingent texts'; though he is now 'embarrassed'
to find himself 'discussing a single text' because of die 'principle of
abstraction in whose name we are trying to establish a language of
narrative . . . " And at the end of diis analysis (which is pursued on lines
very similar to die mediod of S/Z) he offers, in apology for a performance
which he regards as a compromise, the 'excuse' that 'die goal of die
investigation, of the research, is not the explication or interpretation of
a text, but die interrogation of diis text (among others) with a view to
die reconstitution of a general language of narrative'.
A similar conflict of aims is traceable in die two subsequent essays,
'Wresding with die Angel' (1972) and 'A Tale by Edgar Allan Poe'
(1974). In each case Bardies introduces into his tide a new term ('textual
analysis') to describe the activity to be offered; yet in 'Wresding with
die Angel: Textual Analysis of Genesis 32:23-33' the audior also writes
about 'the structural analysis presented here' (diough conceding it to be
'anydiing but a pure one'), while at the same time including a lengdiy sub-
section actually entitled 'Structural Analysis'. This, though commended
for its 'interest', is contrasted widi his own work which is said to be
'oriented in a somewhat different direction'. Insofar as anything coherent
is to be extracted from all this it has to be sought in the following extract:
diis textual analysis seeks to 'see' die text in its difference - which
does not mean its ineffable individuality, for such difference is 'woven'
in die known codes; for diis analysis, die text is caught in an open
network, which is the very infinity of language, itself structured
widiout closure; textual analysis seeks to say, no longer where die
text comes from (historical criticism), nor even how it is made (structural
analysis), but how it is unmade, how it explodes, disseminates:
according to what coded avenues it goes.
Even so, it is hardly possible to see this as significandy different from
die 'Principle of Plurality' put forward in die 1969 'Apropos of Acts
10-11', where we were told diat, in die Structural Analysis of Narrative,
'what die analyst wants to establish in die language of narrative is die
possible site of meanings, or again die plurality of meaning, or meaning
as plurality'. In each case (as also in S/Z) die over-riding purpose of the
exercise is to assert diat, far from establishing any single meaning, analysis
can only reveal a multiplicity of meanings, an 'explosion' of meanings.
 ROLAND BARTHES'S NARRATOLOGY 61

(Significandy, diis claim never gets beyond mere assertion; die most that
is ever demonstrated is the possibility of a second meaning. Presumably
it is implied diat true plurality could be demonstrated in die writerly text - if
one could ever be found.) To add another dimension of confusion Barthes
writes in die opening paragraph to 'Textual Analysis of A Tale by Edgar

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
Allan Poe':
Hence, in a certain way, we must distinguish structural analysis from
textual analysis, diough we are not prepared to declare diem to be
antagonistic: structural analysis, stricdy speaking, is chiefly applied
to oral narrative (to mydi); textual analysis which we shall attempt
to perform in die following pages, applies exclusively to written
narrative.
This wholly new distinction seems to have been dirown off, however,
on a passing impulse; so far as I know it has not been pursued anywhere
else.
More consonant with die general drift of Barthes's shifting emphases
is die distinction he draws in die same essay between structure ('object
of structural analysis proper') and structuration (announced as object of
die textual analysis to be performed on Poe's Tale). This distinction,
related as it is to die 'frayed character of die codes' which is 'an integral
part of structuration', seems suggestive of a wish to move away from
analysing narrative into a limited number of elements which recur in
various combinations (as in linguistic structuralism) towards a more
flexible model responsive to die individual text 'in its difference'.
Indicative of diis tendency is die way in which die five quite-firmly-defined
codes of S/Z proliferate in the essay on Poe's tale dirough die addition
of a miscellany of new codes - metalinguistic, rhetorical, narrative, socio-
ethnic, scientific-deontological, phatic, chronological, etc. It is difficult
to see, however, diat die greater freedom made possible by this multiplicity
of codes results in a commentary which follows more faidifully die details
of die text. The tale in question ('The Facts in die Case of M. Valdemar')
is a short one, and Bardies divides it into only 150 lexias. He dien confines
his commentary to two 'fragments of analysis' (one concerned widi lexias
1 to 17, die other widi lexias 103 to 110) plus a brief linking 'actional
analysis' of die lexias in between. It would hardly be fair to pounce too
heavily on an analytical treatment which is so professedly scanty. I will
say only diat it does not seem to me to engage very closely widi die
distasteful sensationalism of diis story or to deal adequately widi die
unpleasantness of die culminating episode in which M. Valdemar, having
been mesmerised on the direshold of deadi, emerges from his seven
mondis of catalepsy, only to rot away immediately leaving behind no more
62 THE CAMBRIDGE QUARTERLY

than 'a nearly liquid mass of loathsome - of detestable putridity'. Nor

does it explain why, out of the whole array of 'classic' texts, this cheap
shocker should have been selected by Barthes for the rare dignity of a
'textual analysis'.
In view of Barthes's rather strange criteria in the choice of texts (the

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
two Biblical stories are after all only decontextualised extracts) it would
certainly be of interest to see what happens when the Barthesian method
is applied to a complete work of some distinction. The sole published
example I have managed to find is a 1987 essay in which Elaine Millard
'appropriates' Barthes's S/Z textual strategy for an avowedly feminist
reading of D. H. Lawrence's St Maivr.8 There can be no real doubt that
Ms Millard has soaked herself thoroughly in the Barthesian ethos and
has come up with a textual analysis which is very faithful to its source
of inspiration. She is indeed able at times to formulate the approved
doctrine with a clarity which the master himself does not often achieve,
pointing out for example that her purpose (like that of S/Z) is 'not to
say what texts mean but to unpick the way that meaning is produced'.
She makes use of the five S/Z codes, and adds to them a further 'code
of misogyny' (labelled MIS), an extension which seems to be legitimate
enough in terms of Barthes's own theory. She provides detailed comment
on only a few of the earlier lexias (the first 19, plus another 6 from a
dozen pages further on), but her observations follow closely the pattern of
S/Z, even to the extent of finding an 'enigma' in the story's title. ('. . . who
or what is St Mawr? A place? A religious leader or martyr? etc.' The
conclusion is that: 'StMawr, as tide, at once announces and demonstrates
for the woman re-reading, a covert phallocentric and misogynist intent.';
but this, based as it is on an association between Mawr and the saint's
name nearest to it in spelling, St Mawnan, seems to partake of the
extravagance that was noted in some of Barthes's commentary.) There
is even the occasional Barthesian cavalier attitude to detail, as in the
assertion that 'the epithet "handsome" ' (said to be equivalent to 'beau'
in Barthes's analysis) marks Rico as 'feminised' when taken in conjunction
with the statement: 'He flirted widi other women still' - a clear mistake,
this, about Lawrence's customary usage of die word 'flirt'. Nevertheless
setting aside cavils of this kind, Ms Millard's use of 'Barthes's metnod
of structuration in order to establish die codes and myths which constitute
gender within die text' is surprisingly effective in bringing into the
foreground of consciousness those stereotypes about sexual difference on
which Lawrence's characterisation in the early part of the novella is built.

* Elaine Millard, 'Reading as a Woman: D. H. Lawrence, St Mawr', in

D. Tallack (ed.) Literary Theory at Work: Thru Texts, 1987.
 ROLAND BARTHES'S NARRATOLOGY 63

What is not, I believe, brought out is the use the (living) author makes
of them - the subtle ways in which he plays with them for his own creative
purposes. Before approaching this issue, however, I would like to dwell
for a moment on a local example of the pitfalls which attend on the 'lexia'
method of reading (what Barthes himself has called 'filming the reading in

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
slow motion'). Here is a passage from Lawrence's text with Ms Millard's
lexia numbers inserted in it:
(9) and her quaint air of playing at being well-bred, in a sort of
charade game; and her queer familiarity with foreign cities and
foreign languages; and the lurking sense of being an outsider
everywhere, like a sort of gypsy, who is at home anywhere and
nowhere: all this made up her charm and her failure. She didn't quite
belong.
(10) Of course she was American: Louisiana family, moved down
to Texas. (11) And she was moderately rich, with no close relations
except her mother. But she had been sent to school in France when
she was twelve, and since she had finished school, she had drifted
from Paris to Palermo, Biarritz to Vienna and back via Munich to
London, then down again to Rome. Only fleeting trips to America.
So what sort of American was she, after all? And what sort of
European was she either? She didn't 'belong' anywhere. Perhaps
most of all in Rome, among artists and the Embassy people.
Ms Millard's comment on lexia 10 reads as follows:
*REF: ethnic psychology (American, worldly, duplicitous).
**MIS: woman as deceiver.
Now while this interpretation might have some degree of plausibility if
the sentence were taken on its own (and in slow motion), it is surely
evident when it is read at normal speed in the larger context I have printed
out that the idea being pursued has nothing to do with ethnic psychology
but is a matter of Lou's displaced rootlessness, her quality of not quite
belonging, either in America or in Europe.
I have taken the trouble to point out this rather trivial instance of
misreading because it seems to me to show with exceptional clarity what
is liable to happen on a larger scale when the Barthesian method is applied
to a whole work. In effect, by taking a microscope to specific lexias or
local sets of lexias the reader's vision is blinded to patterns in the larger
context which are readily perceived by the normal reader - and these
patterns of significance are, of course, the very ones which are uniquely
imposed by a creative author and cannot be subsumed under the 'already
written' or 'already read' stereotypes of Barthesian theory. To take an
64 THE CAMBRIDGE QUARTERLY

example from Ms Millard's interesting and thoughtful essay, it would

seem that her preoccupation with 'the codes and myths which constitute
gender' in the earlier pages of St Mawr has prevented her from realising
that these stereotypes are far from being endorsed in the larger pattern
of the novella. Here indeed the three male characters are shown to be

Downloaded from http://camqtly.oxfordjournals.org/ at National Chung Hsing University Library on April 11, 2014
just as much at sea as the two women in their responses to the frustrating
pressure of modern civilisation, so that the positive value we are led to
at the end is one of sympathy and understanding for Lou's intention to
spend her life 'keeping herself on her lonely ranch 'deep in America'
where there is a 'wild spirit' more real to her than men. 9 Can a method
which fails to identify so central a meaning as this be said in any real
sense to have 'unpicked' the meaning of a text?
I recognise of course that in putting it in this way I set myself in radical
opposition to the Barthesian dogma that it is always impossible to arrive
at any single agreed meaning for a 'text', and that an attempt to argue
for the superiority of any one reading is merely an attempt to intimidate
others into a dull conformity. Here perhaps one lights on a major reason
why Barthes's narratological oeuvre, for all its opacities, incoherencies
and inconsistencies, has gained for itself such a wide and raptly partisan
audience. Many of the younger generation, alienated by the authoritarian
values of our educational system, repudiate any requirement to subordinate
their reading of a work of literature to the interpretation and judgment
of any critical 'authority': others resist even the requirement to submit
themselves (for the duration of the work's reading) to a value-system
'imposed' by the author himself. For both of these groups the doctrine
of'plurality' offers deliverance from their dilemma; and all too often the
qualifications with which it is hedged around are soon discarded, to leave
behind only a crudely confident assumption that the meaning of a work
is 'whatever you think it is'.
Barthes would perhaps have agreed that such over-simplifications are
regrettable; but I believe he must bear some responsibility for their current
prevalence, and consequently for their trivializing influence upon the
study of literature.
Frank WhiUhead

9
It may seem that Ms Millard has come to a certain intuitive awareness of
this when, in her penultimate paragraph, she tentatively discovers 'a repressed
but radically feminist consciousness in this most phallocentric of texts'. It is not
at all clear, however, that this insight has come to her through her Barthesian
analysis rather than from a pre-existent ability to read a work of literature.