TO ANALYZE,
TO SEGMENT
Raymond Bellour
1/47
The film begins. Paris, Bois de Boulogne Avenue des Acacias. The
coryphaeus (commentator) of the musical comedy presents himself and
half-speaking, half-singing, announces his theme: "Little Girls." A theme
soon centered on the apparition which calls it into being. "This story is
about a little girl in particular. . . Her name is Gigi." Gigi enters and floats
off, opening the theme of the narrative which takes her name.
The film ends. The theme is repeated: same game, same place, same
shot. But this time, behind the commentator a couple appears; it appears
in another time, a mythic time, as if beyond the narrative, since it is in
fact its solution: Gigi, accompanied by her husband Gaston.
The classic cinema (especially the classic American cinema) knows well
these rhyming effects; they are, to be precise, what constitute it. These effects
sustain the difference of the narrative across an ordered reservoir of
similarities; they flash, by the unfolding of symmetries (more or less refined
as the case may be), the dissymmetry without which there could be no
narrative. From beginning to end, the classic film ceaselessly repeats itself
because it leads to its resolution. This is the reason why its beginning and
its end often reflect each other with ultimate insistence. The film acknowledges
itself as a result, by inscribing the systematic condition of its itinerary,
by giving us its signature with a final flourish, presenting the operation
which built it piece by piece.
The effect of repetition-resolution has a specific nature, although it is
very common—therein resides all its force. It operates globally, and very
precisely, from segment to segment, that is to say from one large narrative
unit to another, from one syntagmatic unit to another.
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
332 QUARTERLY REVIEW OF FILIVI STUDIES / August 1976
But what constitutes in reality a segment for the analysis of a film?
Which is to ask, with what truth or with what practicality do the "large
syntagmatic categories of the ¡mage track" endow the definition of the
segment and the act of analysis?
Much has been written about the Large Syntagmatic. For it, against it,
growing out of it. For it: in France and especially in the U.S. with what
is perhaps excessive servitude. Against it: most recently with a real lack of
intellectual generosity or of intellectual imagination. I will not go over the
debate. I will assume you are familiar with the Large Syntagmatic. Still, I
wish to retrace, in order to better situate my own direction, the movement
which the Large Syntagmatic effects in the thought of Christian Metz.
The establishment of the Large Syntagmatic blends two complementary
movements, determined by the logic of discovery: we are in 1966, only
two years after the striking point of departure of "Cinema: Language.or
Language System?" First Movement: the attempt is made, through a code,
this code, to show that there are codes; that codicity, in cinema, has no
less efficiency than in other fields, and that one can try to master it there,
to put it into play. A point which none of the classifications, taxonomies,
or charts'of types of montage, etc., either far too formal or informal,
could show before then. In this way, the Large Syntagmatic is a sort of
theoretical operator: it actualizes the concrete possibility of a semiotics of
cinema, since it potentially materializes it Second Movement: this specific
code seems to be, from a logical as well as a historical point of view, the
code which permits to bring to light (through a closed series of comrnutable
syntagmatic types), the cultural establishment of film as the support of
fiction.
The combined force of these two movements explains how Metz could
fall partially into the trap, or the illusion (soon to be exposed), of having
found the code of cinema. Almost immediately (thereafter) the Large Syntagmatic
is no longer the code, but simply a specific code among others
(this will be the position repeatedly taken in Languageand Cinema). But;
this ambiguity is crucial; for if the Large Syntagmatic is not more or less
than a code among other codes which are juxtaposed to it and run through
it, it envelops the other codes; and in a literal sense, this code is superior
to them: forcibly, as the consequence and condition of the fiction.
Ever since the beginning, or almost, the Large Syntagmatic comprised a
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BELLOUR / To Analyze, To Segment 333
patient and rigorous self-criticism on Metz's part; its elaboration was submitted
to internal contradictions which (according to an ultimately improbable
logic) undermine, in a sense, the initial positivism of the code, but
which add to its potential force. From his very first writings in Film
Language,Metz confronted three fundamental objections. First, the distinction
between the autonomous shot and the seven other syntagmatic
categories: the shot-sequence (as its name indicates) contains within itself,
through a sort of spatio-temporal expansion, all the possibilities contained
in the seven other types; the shot sequence therefore imposes the necessity
of reworking the initial chart, dividing it into two branches. Secondly,
there is a discrepancy in the seven other syntagmatic types between those
which are "hard" and "fast" and those which are "less clear." (This discrepancy
becomes evident in the syntagmatic analysis of Adieu
Philippines.) For example, the bracket syntagma and the diegetic insert are
said to be "distinct" configurations which are easily recognized and unmistakable,
while the ordinary sequence and the scene are said to have
"vague contours"—Metz says it is sometimes difficult "to isolate them from
the general filmic flow." Thirdly, and this seems to me the most important
point: "all the problems raised by alternation" which can not simply be
resolved by the two complementary formulas—the a-chronological formula
of the alternating syntagma. Metz writes: "The solution would be perhaps
the introduction of a semiotically rigorous theory which would take
into account two facts which are both very evident in films, but as yet
poorly accounted for: First, the phenomenon which we can call the transformation
of an insert (an autonomous segment with a single insert can be
'transformed' into an autonomous segment comprised of multiple inserts
and thus into a type oí alternating syntagma). . . Secondly the distinction
between true alternation (which establishes a narrative branching, or doubling
in the film) mû pseudo-alternation (which is only a visual alternation
within a unified space, or else is derived from a vaguely alternating aspect
inherent in the film subject under certain circumstances)/'
Finally, we come to a Third Movement, which logically grows out of the
first two: through the methodical, expanded study which comprises
Language and Cinema, this code really becomes a code; that is to say, radically
detached from the filmic text, the various categories become only abstract
entities evidenced in the film in the concrete form of the autonomous
segment.
This concept was integral to the formulation of the three criteria of
demarcation of the autonomous segment, which Metz undertook in an article
written soon after the one on the Large Syntagmatic. He says: "The
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
334 QUARTERLY REVIEW OF FILM STUDIES / August 1976
analyst of the classic film is correct to consider as a single autonomous segment
any passage from a film which is not interrupted by a major change
in the course of the intrigue nor by a sign of punctuation nor by the abandonment
of one syntagmatic type for another/' Thus the question of the
autonomous segment has been deliberately transposed to consideration in
the text, that is to say, towards its imperative intercodicity, towards textual
analysis.
Having recently discussed this with Metz, I know that he now feels that
a new version of the Large Syntagmatic is possible. (You will ask: "Why not
do it then, if it's possible?" I think, simply, because science always rests on
desire, and desire is displaced—as Freud discovered. Only the scientific
Utopist can believe that one always attempts to finish—at least within a
practical time—everything which is begun.) This new version of the Large
Syntagmatic would have to overturn the positivist illusions which often accompany
formalization in its initial stage. It would equally guard against
the new positivism that the evolution of linguistics could throw over film
theory by trying to substitute too simplistically the generative and transformational
model for the structural model. By protecting itself equally
against both, the new version would be able to combine their advances* It
would challenge the plurality of levels which forbids the strict Chomskyan
model from ever rejoining its object, the film, in the singularity of its
textual system. Cinema will never be a language, nor film a grammar. It's
not by chance that the poetic and musicological analyses of Nicholas
Ruwet owe so much, if not more, to the structural model than they do to
the generative and transformational model. But on the other hand, this new
version of the Large Syntagmatic should possess increasing abstraction
which would permit it to definitively erase any structural effect of flattening,
of descriptive application between the code and the text, and finally
to mitigate the insufficiencies of the previous effort. The new
Large Syntagmatic would have to constitute a collection of spatio-temporal
matrices where the present syntagmatic types would be arranged with
their necessary complements and modifications in an ordered series.
The surface level, the textual actualization, would then alone merit the
name of autonomous segment, offering the analysis the singularity, continually
renewed, of an exact decomposition of the filmic chain.
This very particular situation of the Large Syntagmatic, this problematic
code, unfinished, in one sense first and foremost code, in another a
code among others, seems to have had two opposite consequences in the
field of filmic analysis. First, numerous studies have been formulated
which developed directly from the Large Syntagmatic sometimes strictly
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BELLÖUR / To Analyze, To Segment 33!
applying the code, sometimes seeking to perfect it, by trying to diversify,
to enrich or to make more flexible one or another of the categories (but
always remaining within the bounds of Metz's own self-criticism, which
implies a transformation). These studies have reaffirmed the determining
richness of the code, its capacity for stylistic and historic induction, its
specific instrumentality, its practical and analytical function in textual
study of large narrative units.
Also, several textual analyses which have been developed as part of
the effort to establish a semiology of cinema, with more or less explicit
references to Metz, found themselves not ignoring, but skirting the
proof of the syntagmatic categories. Some of them, because they isolate
a segment or fragment of a film in order to consider smaller units.
Others,, because they evaluate, taking into account several segments or
fragments, their functioning in the productivity of the text as a whole.
Efforts which consider an entire film situate themselves more or ¡ess
intentionally in the perspective defined by Metz as that of the global
textual system. Here I am purposely grouping together works of varied
intentions and nature. Foetus astral by Jean Monod and Jean-Paul
Dumont, the collective text by Cahiers du Cinema on Young Mr.
Lincoln, the book by Claude Bauble, Michel Marie and Marie-Claire
Ropars on Muriel, Stephen Heath's long study of Touch of Evil, and
my own analysis of North by Northwest. The detour is remarkable in
these last two examples as the development of the analyses (otherwise
quite different) both depended on segmentation; a sort of rough, or at
least partially rough segmentation, which justifies Heath's note following
his decoupage: "The segmentation here operates at the level of the
narrative signified according to the simple criteria of unity of action,
unity of characters, unity of place; it has no analytic status other than
that of allowing reference to the film as narrative." What is evident in
these two analyses is the fact that they do not cease to touch the Large
Syntagmatic without seeking to constrain it and certainly without running
the risk of being constrained by it.
This is the risk I shall try to take now. Working from the syntagmatsc
categories and well beyond them, it seems to me I can show you the
systematic modelization of narrative units in the classic American film.
Perhaps Gigi, which I present here, does not aim at filling the gap between
the present state of the Large Syntagmatic and a second still unrealized
state. Certainly this analysis will situate, and project the gap
through the logic of its own movement. But nothing more. The Large
Syntagmatic is thus fully used here (this is to say through its lacks) as
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
Operator in the analysis: first, through the effect of descriptive logification
operating within it; and then, through the syntagmatic solicitation
which it opens, by the fact it incites to a generalized diegetic segmentality.
Segmentation, as we shall see, is a mise-en-abyme, a process
which theoretically is infinite—which is not to-say that it lacks meaning.
By the shifted play which operates between its different levels, segmentation
permits us to sense the increased plurality of textual effects.
But this analysis is only a beginning and by definition, in the limited
form of this presentation, it is much more the framing of an analysis,
its placing into perspective, than the analysis itself. I am largely limited
(and on the level of small units, very minimally limited) to the enumeration
of rhyming effects, that is to say the effects of differential repetition
which structure the development of the narrative. Previous analyses
have shown me the fundamental determination of these effects in the constitution
of the classic film, on the level of the fragment, the segment,
or the entire film.
It seems striking to me to have them appear, in this crossing of the
levels which is done here. But, things being as they are, I can only classify
the rhyming effects, enumerate them; cannot produce them within
the logic of their textual progression, the material work of their unfolding:
analysis is not reducible to its frame. Intellectual imagination, mine,
yours, must give to the elements which I will decompose the space which
forms and constitutes them: their textual volume.
The syntagrnatic decomposition of Gigi which causes 47 autonomous segments
to appear in the image-track of Gigi (numbers 1-47 in the summarizing
table at the end of the article) calls for certain remarks:
a) The major effect of redundancy between the three criteria of demarcation
manifests here a high degree of classicalness.
b) In this sound film, this musical film in particular, the image is sovereign
as concerns the segmental demarcation. The sung numbers are smaller
or equal to the segmental limits, with one exception: Gigi's song ("I
don't understand the Parisians") which crosses segments 9 and 10, creating
a sort of autonomous sound segment. The voice, the dialogues
are strictly submitted to the phenomenology of the image, its temporal
336 QUARTERLY REVIEW OF FILM STUDIES / Aygust 1976
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BEL LOUR / To Analyze, To Segment 337
partition. As for the music, if the musical dissolves sometimes do not
coincide with the visual dissolves of segmental demarcation, they still
globally reinforce, since they are tied to the songs, the dominance of
the autonomous segments of the image-track. The Large Syntagmatic
of the image-track in the classic film, due to the force of the diegesis
(story) itself, is, therefore, not far from being the code, the one which
permits the others.
c) This strict application of the Large Syntagmatic is confronted by one
major difficulty: the impossibility of taking into account, in a film
which significantly contains no alternating syntagmas, of the alternances
which structure numerous segments. These segments (1,8, 13, 15, 17,
21—an episodic sequence which is at the same time an alternating syntagma,
but without simultaneity between the temporal series as in the
example of an alternating syntagma in Adieu Philippines from which
Metz raises the question of alternation) 24, 25, 26 (in which the segmentai
level loses a true scene by Honoré and Mamita) 30, 35, (a bastardized
episodic sequence, of which the first two episodes are a brief
form of alternating syntagma) and 47.
As for the autonomous shot, despite its frequency (9 segments) it does
not pose severe problems herebecause these particular autonomous shots
fully remain within the classical model.
This decomposition manifests two arrangements essential to the development
of textual logic.
First, the large number of repetitions, of rhymes which operate from segment
to segment, inside the mirror effect of segments 1 and 47. This obviously,
does not concern their syntagmatic forms. The syntagmatic type,
on this level, is only one characteristic among others, even though profoundly
different: as is a specific index of the textual surface it affects,
the form through which pertinences of relative importance are inscribed
in the cinematic signifier. These pertinences are such that their
systematic arrangement in each segment carries the narration from segment
to segment,
a) One only has to look at the summarizing diagram to recognize the
operations which bring unity, a sort of repetition full of differences,
between the segments:
4 and 37
6, 16, and 20
7 and 31
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
8, 28, and 29
14 and 40
15 and 41-42
19, 27, and 32
23 and 46
24, 25, and 26
36 and 39
b) Some of these operations, starting with segment 36, concern many of
the last segments of the film (37, 39, 40, 41-42, 46, 47). This concentration
is notable: it shows how the film, through its segmental partition,
resolves by repeating, by a sort of generalized narrative condensation,
which carries it formally towards its end.
But these effects of repetition-resolution are inscribed within the
limits of the autonomous segments in order to overcome them in many
ways. Even if the classic film, as Metz points out, leans towards the sequence
(the autonomous segment) rather than the shot, it does so in favor
of an equally profound tendency, which inscribes the segment in a system
of narrative commutation with units both larger and smaller than the segment,
decoupling the effects of segmentation of the filmic chain. These
two simultaneous movements attract and repulse, contradict and complement
each other: it's what makes it necessary and yet difficult to distinguish
them; their confusion (which is nonetheless decomposable)
which turns film into the space of a generalized segmentaiity, is a fundamental
condition which transforms the filmic surface into a textual volume.
The classic film is thus marked by a tendency towards units which are often
superior to the segment (suprasegment or macrosegment). These units
are often called sequences (in every day as well as professional language),
and they often correspond to units in the scenario. They are generally determined
by a global unity of time, place, and action—for example, in
Gigi all three segments (24, 25, and 26) take place in Trouville, in the
course of one day, with the same characters (creating a kind of episodic
suprasegment). Segments 12 and 13 constitute another kind of suprasegment.
Segment 12 is a short autonomous shot depicting the facade of.the
ice-skating palace (we see neither Gaston nor Gigi but know that they are
338 QUARTERLY REVIEW OF FILM STUDIES / August 1976
Suprasegmentai/Sybsegmental
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BELLOUR / To Analyze, To Segment 339
going towards the building and are about to enter); segment 13 is a scene
taking place inside. But the unity of place does not necessarily determine a
segment, when narrative movements prove too dissimilar despite the
transitions which bind them. Thus it seems justifiable not to combine segment
27, which depicts Alicia talking Mamita into speeding up Gigi's education,
and segments 28 and 29 (VII) which are devoted to these pedagogical
sessions. The suprasegment is a single minor dramatic flow; this explains
how it can in other cases encompass several different locations:
thus segments 3, 4, 5, and 6 (III) which concern Gaston's and Honore's
meeting, or segments 30 and 31 (XXVI) which present in continuity the
contradictory reactions of Gaston to Gigi's transformation.
The determination of suprasegments is obviously less rigorous than the
determination of autonomous segments, because it is not indicated by any
cinematic specificity, but rather depends on the force of the textual system.
Thus it is totally derived from the singularity of each film, and conditioned
by the analysis which turns it into an intermediary function
between the segments and the major divisions (A, B, C, D, E, in the case
of Gigi) which block out the overall dramaturgy of the narrative. As is the
case with all efforts of textual decoupage, the interest of suprasegments is
primarily descriptive; they serve to isolate rhyme effects which otherwise
are merged or superimposed with the functioning of the autonomous segments.
Following a sort of internal tautology, segments connected in a
suprasegment rhyme strongly within this new unit. What is notable about
this operation is how it concentrates the rhyming effects in the narrative
succession. This is the case in the examples cited above: segments 24, 25,
and 26, the "sequences" which all take place at Trouville, held together
by the same characters; and segments 28 and 29 which group together the
two episodic sequences of Gigi's education in Alicia's apartment. But more
eloquent examples of suprasegmental rhyme are built over the course of
the film from one suprasegment to another: the only clear example is the
two groups of segments occurring in Maxim's, which combines in suprasegment
VIII, segments 14 and 15, and in suprasegment XV, segments
40, 41, and 42. Other effects function between suprasegments, but those
are more partial and can be analyzed in the complementary work of sub-
segmentation.
In this film as is generally the case with classic film, the level of units
smaller than a segment is where the multiple play of echoes which structures
and defines the progressive resolution of the textual system is systematized.
Let's first look at what I call elementary subsegmentation:
that which refers to two or more successive times within the continuity of
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
340 QUARTERLY REVIEW OF F1LSVI STUDIES / August 1976
a single segment, each one of which forms a sort of small scene—but in a
dramatic and not a syntagmatic sense of the term scene, obviously.
Recall the third criterion invoked by Metz to determine the demarcation
of autonomous segments: "a major change in the course of the plot."
Obviously this criterion lacks precision: what is amajor change? But this
is the wrong way to pose the question, since this demarcation by the plot
itself is practically always evident: the limitations of the other two criteria
almost automatically lead to the transparency of the third. This
grows out of the fact that the classic film determines its segmental units
by a series of ruptures in the signified of temporal denotation: when
punctuation and variation of the syntagmatic treatment do not indicate
the rupture, only a change in the course of the plot can do so. That's why
this imprecise criterion is the most dependable: by a sort of tautological
reasoning, a narrative change which does not lead to segmentation can
only be a minor change. However minor changes are often less "minor"
than many other changes in the course of the plot, others which are elevated
to the status of segmental demarcation only by the intervention of one
or both of the other two criteria (this is obviously not the case when the
inscription of multiple criteria is purely redundant, which is quite common).
Compare, for example , the mutation introduced in the narrative by the
appearance or disappearance of a character with the slight change which is
denoted by the punctuation between the autonomous shot of the facade
of the ice-skating rink and the scene which follows inside. Which is to say
that the segmental demarcation determined by the multiple inscription of
the signified of temporal denotation within the filmic signifier only half
corresponds (though sometimes more, sometimes less) with the unfolding
of the plot, the succession of narrative actions. These dissociations make
it necessary to introduce an operation of subsegmentation.
The episodic sequence (as is the case of the bracket syntagma) is singularly
privileged in this perspective, as a result of the precise demarcation of
each episode. Thus demarcation occurs at this smaller level as a sort of
subdemarcation, both by the diegesis (story) and for the most part by an
internal punctuation. The episodes are almost always linked (as are some
segments within a suprasegment) by an effect of successive rhyme: this is
the case with the episodic sequences 14, 21, 28, and 29 (that is to say, all
the episodic sequences except 35, which, as I have emphasized, is somewhat
bastardized). But a given episode may form a distanced rhyming effect,
as in the 4th episode of segment 1 where Gaston and Liane enter
Maxim's, which rhymes with segment 40 where Gaston and Gigi enter it.
Inversely, the subsegments of other syntagmatic types are determined
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BEL LOUR / To Analyze, To Segment 341
without specific inscription within the cinematic signifier. Their indisputable
"vague" demarcation is not a problem—some real scenes are less
clearly marked. These subsegments are defined by the entries and exits of
characters, particularly evident in this genre film, which is very strongly
marked by theatrical representation. The location, actions, musical and
sung motifs, obviously powerfully reinforce these scenes which are organized
between characters. But these other elements do not have the
same degree of significance: the musical and sung motifs, because their
boundaries are almost always smaller or equal to the segment, only partially
coincide with those of the subsegments; the actions, because they
cannot be segmented in the same way, tend to dissolve into the global
mass of signifiers; the locations, because the temporal form which determines
them is already taken into account by the syntagmatic categories.
Thus, scenes appear and provide numerous rhyming possibilities. From
the beginning, in segment 1, the 15 shots which occur before Gigi appears,
put Honore in the role of commentator (1a), as he will be again in the
first episode of segment 14 (14a), then again in subsegment 34b after the
disappearance of Gaston. In the same way in segment 32, the disappearance
of Gaston in shot 15 opens a short scene between Mamita and Gaston
which corresponds to one which occurred earlier in segment 7, and one
which will occur late in subsegment 30b before reappearing in segment 31.
Subsegmental divisions also permit new commutations to be performed
between intermediary groupings: they go from subsegments to segments
and from segments to suprasegments. Thus, the first group concerning "education"
which includes segment 8 and subsegment 9a which follows it,
finds its exact echo in suprasegment XVII (28 and 29). In the same way,
the extremely long subsegment 30c, subsegment 43b, and the short segments
44 and 45 respectively indicate a very exact repetition of Gaston's
two voyages which serve as preludes to his decisions first to keep Gigi,
then to marry her. Thus, a global rhyming effect is established between
subsegment 30c plus segment 31 on the one hand, and segments 43, 44,
45, and 46 on the other, which is suprasegment XXVI.This last rhyme
largely builds the effect of condensation, uniting in the final part of the
film (E) a series of elements which appeared earlier in the four other parts.
But in order to expose this productive circulation of the text, it is necessary
to go much further in the decomposition of its elements: it is necessary
to subsegment the subsegments.
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
342 QUARTERLY REVIEW OF FILM STUDIES / Aygust 1976
Segmentai (Sub-supra) Textual
Subsegmentation goes far beyond what I've done so far. Until now, I've
restrained it to scene effects corresponding to one or more shots; and always,
in the case of the single shots, to very long ones (like 43b) or to the
more specifically determined units in the episodic sequences. Also, I have
only performed successive dissociations, in a linearity which sometimes
corresponds to the true presentation within the fact, but often only does
so partially, mimicking its truth with a representative approximation which
is neither altogether false nor altogether true.
This is why I qualified this first subsegmentation as elementary. Complex
subsegmentation goes far beyond this. It can be termed microsegmentation
to indicate a movement, a progressive work of the textual and analytical
force. There is no longer anything on this level corresponding to
the strict distinction, there are only degrees of narrative expansion. Therefore,
complex segmentation has nothing to do with the shot as a limit:
even though the demarcative boundary of the shot constitutes in (and of)
itself a textual pertinence and a stylistic index.
I can give only two very summary examples here.
a) Segmen133—Shot302, which shows Gigi in her bedroom (the decor is
visible for the first time); she is alone, lying on her bed, stroking her cat,
and gets up to let Gaston in, within the continuity of the same shot.
This very fragmentary scene is soon echoed by a fill scene, a long
autonomous shot which comprises segment 38, where Gigi is singing in
her bedroom, holding her cat in her arms. In its simplicity, this effect
shows very well the process of the film, a process of varied reduplication
of its successive elements. It completes the condensation which
operates between the fifth part of the film and the four preceding
parts: from segments 37 through 47, there is not one which does not
rework a segment, a subsegment, a moment of the four others,deepened
by the differentiation which carries the film towards its con-
clusion.
The three completely unequal subsegments which follow (33b, c,
d) are inscribed along three axes which the lack of subsegmental development
prohibits working out completely: the first which takes up the
large part of the segment (shots 302-307) is one of the series of scenes
between Gaston and Gigi; the second, determined by the appearance of
Mamita in shot 308, is part of the multiple scenes or fragments of
scenes between Gigi, Mamita, and Gaston (this microsegment announces
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BELLOUR / To Analyze, To Segment 343
in particular the dramatic and equally very short microsegment, scene
43a); the third, after Gigi has fled into her bedroom, is inscribed by a
second, interior rupture in shot 308, in the series of scenes, either
peaceful or dramatic, between Mamita and Gaston.
b) Segment47—hereagain, a division in the shot creates three minuscule
subsegments in the two last shots of the film: in shot 348, Honore
sings, leaning against a tree in medium shot; in shot 349, Gigi and
Gaston diagonally cross the grass towards a carriage, which they enter.
Their movement is accompanied by a camera movement which loses
them to return, in the continuity of the same final shot, to Honore in
the same fixed framing as in the previous shot
This last example shows how the progressive decomposition of dramatic
occurrences in the filmic chain, opened by the decomposition of the
Large Syntagmatic, leads more or less inevitably to the internal analysis of
segments and to a comparative analysis of echoing segments. This fina!
segment which displays a refined classicism is thus perceived as constructed
according to an a/b/a alternation which reproduces the pattern of the first
segment of the film of which this final segment constitutes the resolution:
in both cases the narrative alternation is between Honore and the others.
But, in the beginning segment there are 21 shots which must be decomposed:
neither the first nor the last depict Honore—the last shot of the
first segment shows Gigi alone since the narrative is opening; in the final
segment, two shots begin and end on Honore, but between these images,
Gigi, this time with her husband Gaston.
At this level, which we can call a microsystem, the analysis encounters
the multiplied constraint and the dispersion of the specific cinematic codes
(codes of camera movement, glances, etc.) which are deployed in the interior
and all throughout the macrosystem, the code of syntagmatic
categories. The analysis again encounters the voluminous force of the
textual system, the full organic play of differential repetition. For example:
from segment to segment, as from suprasegment to subsegment,
the segmental analysis thus built up would end by constructing, through
the numerous scenes which take place at Gigi's, the immense paradigm of
entries and exits which supports (along with others which are less obvious)
the microsystematic structuration of the narrative units of the film.
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
I want to insist on finally a last, fundamental effect common to most
American film. This effect, through the textual volume, multiplies and
closes doubly the area of its own expansion. The systematic stockpiling
of symmetries and dissymmetries throughout the filmic chain decomposed
by generalized segmentation faithfully copies, (because they, in fact, produce
one another) the schema of familial relations which constitute the
space of the narrative,
Gigi and Gaston are both children raised, each according to the custom
befitting their sex, by a substitute mother (Mamita) and a substitute father
(Honore). A three-way dissymetry is inscribed within this symmetry
which, from the beginning, makes Gigi and Gaston destined for one
another.
a) One man, Honore, corresponds to two women, Mamita and Alicia, in
the role of adopted substitute parent.
b) A genealogical gap makes Gaston Honore's nephew, but Gigi, Mamita's
granddaughter and Alicia's great-niece.
c) The clear age differences between Gigi and Gaston reproduce this
genealogical separation, making one already a man, the other still a
child.
It's in suprasegment XV at Trouville that the reciprocal feeling between
Gigi and Gaston is crystallized. These three segments are in part constructed
on a narrative alternation which juxtaposes Gigi and Gaston on one
hand, Honore and Mamita on the other, who meet for the first and last
time in the entire film during this suprasegment. They seem to be linked
by an old romance, and go so far as to evoke the marriage which failed to
unite them. This remembered marriage only serves to mirror the potential
marriage of Gigi and Gaston. But it suggests much more. For at this moment,
the line of dissymmetry between the generations disappears in favor
of a matching counterpart. Mamita tells Honore, with an insistence as useless
to the plot as it is necessary to the symbolic functioning: "Gigi is my
granddaughter," Honore's gallantry, appropriately structural, pushes him
to answer: "Granddaughter, no, daughter." How to better suggest that the
children they never had are obviously those which they each adopted and
344 QUARTERLY REVIEW OF FILM STUDIES / Aygust 1976
Segmentai/Familial/Conjugal
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BELLOUR / To Analyze, To Segment 345
whom the film will reunite by marriage? Thus, an incestuous fiction is presented,
the kind for which the classic-romantic imaginary displays such a
predilection. Dissymmetry obviously reappears as does Aunt Alicia, absent
from this structure containing four terms. But the dissymmetry of structures
serves that which it hides and permits to be resolved: through the
difference in ages, Gigi finds in Gaston a substitute father for the one who
is even more strangely absent than the mother, who is present, you will
remember, only as a singing voice-off—and Gaston will find in Mamita the
substitute for a mother whom no one has mentioned.
This is what the story of the film tells us, in a narrative which makes
the segmental the textual condition of a happy sliding from the familial
to the conjugal, and thus presents in its formal achievement the resolution
of the Oedipus complex. Such is the effect of textual production
which Tve tried to focus on in terms at once different and yet strictly
complementary in my analysis of Hitchcock's North by Northwest, an effect
I have called the symbolic closure.
Translated by Maureen Turim
Raymond Bellour is a Professor at the Centre National de la Recherche
Scientifique.
Downloadedby[UNIVERSITYOFADELAIDELIBRARIES]at12:4409November2017
BELLOUR, Raymond. The Analysis of Film. Bloomington - Indianapolis: Indiana University Press, 2000,
s. 207-215