[Nostalgia for the Coming Attraction, chapter 1]
From Sample to Model. A Brief Formal History of the Trailer
The relationship between form and function is not accidental.
Jurij Tynjanov, On Literary Evolution
During the post-production of the puppet-animation film Monty Spinneratz (Warner Bros. 1997), a German-American co-production, the editor of the film was assigned to cut a trailer together with an experienced American colleague. The American went to work on the trailer without so much as looking at the already finished rough-cut of the film. "Don't you want to look at the film first?", the
German editor asked. "No," his colleague replied, "just show me the 26 best scenes. That will do." *
This anecdote may serve to illustrate two things. First, making a film and making a trailer are two completely different matters. The production history of the film has no bearing on the advertising value of the material, which is why trailers are made by specialists, and not by the directors of the films. Second, these specialists apparently have quite specific ideas about what a trailer should look and feel like even before they have seen the film. To a large extent, making a trailer seems to be a concept-guided, top-down process that follows specific rules. But if trailer making is indeed to be considered a rule-guided cinematic practice, the question arises as to the rules by which trailer makers - and trailers - work. Which are these rules, how have they emerged, and how, if at all, have they developed over time?
In order to answer such questions, one could go directly to the trailer-makers and ask them about their work. In cases where the trailer makers are not or no longer available for interviews, information of this kind may also be obtained through articles and interviews published in trade papers and newspapers. A brief survey of such materials indicates that there are indeed certain rules. For instance, trailer makers from all periods of the
1 My informer for this anecdote is Rolf Nohr who worked as a production assistant on Monty Spinnerralz.
history of Hollywood filmmaking insist that the trailer simply has to be better than the film.2 This is of course a basic requirement of any type of rhetoric. A lawyer delivering a speech in court should, in the best interest of his client, always try to make things look slightly better than they actually are. But how exactly the trailer can be made to look better than the film is a matter of discussion.
One basic formula for the good trailer, to be found consistently in statements from the pre-1960 period, is: Show the best scenes, highlight the stars, but by all means do not give away the story (Lasky 1937,13 f.; Ross 1952,163). In an article from 1938, the New York Times writes, „The men who make [the trailersl are forbidden to reveal the salient plot points of the feature and [...] they must use
related incidents that will provoke interest in the film."3 Robert Faber, one of Hollywood's most prolific trailer producers who made trailers for Universal films from 1934 through the late seventies, claims he tried to withhold story information as much as possible and attempted to convey the mood
of the film instead.** Jack Atlas, another seasoned craftsman who worked for MGM and Columbia between 1942 and the late sixties and opened his own trailer company in the early seventies, preferred to play down the story content of the movie as well. His repertoire consisted of four standard selling
points, which he used singly or in combination: "background", "character", "plot" and "action".-* If preference was given to the plot aspect of the film, however, the trailer still wouldn't necessarily reveal the story of the film.
While it would be interesting to pursue a discussion of the concept of „mood" in relation to trailers and test it against the trailers themselves, the one specific concern which bounds all the
2 Douglas W. Churchill: "Shooting Stars in Hollywood". In: NYT, 8. Mai 1938B. John Wilson: "Impresario of the Movie Trailer". In: LAT, June 201977. Leonard Klady: "Truth About Trailers: They Work". In: Variety, 28. November 1994, p. 13-24.
3 Churchill, (see note 2).
4 Interview with Robert Faber, November 18,1994.
5 Tom Gray: "Creating Trailers - A Matter of Merchandising". In: MPH, Vol. 238, No. 43, October 23, 1968, p. 23f. Interview with Jack Atlas, October 26,1996.
Statements of classical trailer producers together is the concern with the amount and kind of information the trailer gives away about the story of the film. Accordingly, I will use thf flUfifitifín of— the distribution of story information as a point of entry for an analysis of trailer making as a rule-guided cinematic practice.
Film Advertising, Brand Image and Narrative Image
The distribution of information, or of knowledge about the film, is a problem crucial not only to the trailer, but to film advertising in general. Conventional consumer products are easier to advertise than films because they are reliable in terms of quality. Each box of Marlboro and each bottle of Coca Cola will taste the same, or so we assume. Advertising creates a brand identity for the product by constructing a brand image, which is usually done by associating the consumption of the product with a desirable experience, or by portraying the consumption itself as a desirable experience. Cowboys smoke Marlboro cigarettes; whoever smokes Marlboro cigarettes will feel like a cowboy himself; if the cowboy lifestyle appeals to you, you should smoke Marlboro, for there is no easier way west Such is, if you will, the implicit syllogism of the Marlboro brand image. As the Marlboro example shows, successful brand images tend to be extremely durable. The original campaign which established Marlboro as the brand preferred by the men of the great outdoors dates back to 1958, and the brand image has basically remained the same ever since (Ogilvy 1985,18).
A film on the other hand poses different problems. One truism film advertisers like to propound is that each film is a product of its own. Economists agree in that films, as other forms of cultural consumption, come without reliable guarantees for quality. Every time a viewer decides to purchase a ticket at the box office or rent a video (or go see a play, a dance performance or an art show), she runs a higher risk of disappointment than an average buyer of a Coke or a box of Marlboros (Farchy 1994, 71). Accordingly, the film industries in the US and elsewhere have tried to minimize the viewer/buyer's risk of being disappointed by standardizing production from very early on in their history. On the other hand, each film is indeed a product of its own, i.e. it has to be conceived of as an original experience, and it has to be advertised as such. At every turn, then, and for
each film anew, film advertising has to find the right balance between standardization and product differentiation, i.e. between reassuring the audience about the reliability of the film in terms of quality and highlighting the film's innovative characteristics, or its „novelty", to use a classical industry jargon term.
Although films were by and large marketed by the brand up until 1912, brand images have played only a minor role in film advertising for the most part of the medium's history (Bowser 1990

103). The only studio or company brand known to have any substantial drawing power in film advertising is the Walt Disney label, and even the appeal of the Disney trademark is limited mostly to animation films. Generally speaking, however, the studio origin of a film is quite irrelevant to its
success.6 Rather than brand images, film advertising has to construct what John Ellis calls the
narrative image, i.e. it has to convey to the audience a notion of the film sufficiently specific to arouse ^\_--------------->-----------------                                             —------------------------
interest, but not specific enough to exhaust that interest (Ellis 1982,30). * - ^ / -*Wj ,Q*^ ^ ^ i ■-                              ~"^_                        -—                                                                      ^                                                                             <^j
The poster for Alien (TCF1979) is a good example of how such a notion can be conveyed. The poster shows the film's title printed in signature type.over an image of a breaking egg. A mysterious golden light shines from the egg's inside. Over the egg the tag line appears: „In space, no one can hear you scream". The typography of the title evokes notions of futuristic technology; the egg calls up images of dangerous reptiles. Also, the tag line contains two genre signals: „Space" is the preferred setting for science fiction adventures, „scream" is what people do in horror films, or what they do when they see such films. Although it may be difficult to write a story resume of the film based on the poster alone, the Alien poster tells a prospective ticket buyer all she needs to know about the film: Iťs a cross between science fiction and horror film; it's about people like you (and me) facing troubles with reptiles in the darker recesses of outer space; and iťs sophisticated enough to appeal to urban audiences beyond the usual sci-fi horror patronage (for a mere teenage gore movie would not be advertised with such sophistication). With just a logo, a title and a tag line, then, the Alien poster
6 The subordinate role of trademarks can also be seen in film posters and newspaper advertisements from the forties. These posters foreground stars and play down story, and the name of the theater were a film is shown is usually displayed more prominently than the studio label (Sweeney 1973).
4
manages to convey the concept of the film, but also to define the film's audience. One could even argue that product differentiation is achieved precisely through the poster's audience address, since i one of the defining characteristics of the film is that it is intended not just for juvenile horror/horror audiences, but also for sophisticated urban moviegoers.
In the course of a film advertising campaign, a preliminary notion of a film such as the one conveyed by the Alien poster is further refined by additional information conveyed by trailers, TV spots, newspaper and magazine articles and review's, „making of" documentaries and other
paraphernalia, or epitexts, of the film/ The theatrical trailer for Aii-w, for instanre, was composed ol shots suggesting a traveling along the surface of what seems to be an arid white planet The surface then turns out to be the shell of an egg. Thundering noise on the soundtrack accompanies the shots of the egg, which eventually cracks open and reveals the golden light on its inside. Over the rupture in the egg, the tag line appears (illustration no. 2). Redundancy and repetition are characteristic for both consumer product and film advertising, as the example of the Alicti campaign clearly shows. If the brand image of the Marlboro cowboys appears in every ad, the narrative image of the film is repeated in every poster, trailer and TV spot in order to create redundancy and to ascertain that the advertising message will register with the audience.
While the "arrjrtiyejmage can actually be used to establish the film as a brand, as a highly differentiated product maintaining its commodity value throughout a long chain of release windows, it is important to maintain the theoretical differencejghveenbrand andnarrative image not least because the two kinds of image relate differently to the product they advertise. One could argue with George Gerbner that all advertising tells stories, namely „stories of value and choice", as he proposes to call them, i.e. stories that teach us how to choose.8 Conventional consumer advertising tries to teach
7 For the notion of epitext see Genette (1992,328f.) "?"H'trX   "
8 According to Gerbner, the stories „that animate our cultural environment have three distinct, but related functions": 1) to reveal how things work; 2) to describe what things are; and 3) to ell us what to do about them. Stories of the first kind are „the basic building blocks of human understanding. They demonstrate complex causality by presenting imaginary action in total situation, coming to some
5
US
how to choose by telling us stories about how we feel if and when we consume the product in question. It is just such a story that the Marlboro ads try to tell. With great insistence, Marlboro advertisements link the experience of smoking cigarettes with the outdoors lifestyle and the feeling of freedom that supposedly comes with a life in the wide-open west-Narrative images on the other hand tell stories about stories. In the case of film advertising, the product itself is, in most cases, a narrative, or, if you will, a field of narrative knowledge. Film advertising tries to teach us how to choose by introducing a lack of knowledge into the narrative field, by conveying certain information about the narrative and withholding others. „Narration comes into being when knowledge is unevenly distributed - when there is a disturbance or disruption in the field of knowledge" writes film theorist Edward Branigan in a passage that as I would like to argue, applies to film advertising as well (Branigan 1992, 66). From the point of view of narrative comprehension, then, film and film advertising are profoundly related, and one could argue that it is
precisely the procedural jffjQjtyW product and advertising discourse - or, if you want to call it that, the sfructural contiguity of the commodity form - which differentiates film advertising from
conventional consumer product advertising."
It is also the procedural affinity of product and advertising discourse which creates the problem of the distribution of information or knowledge which so concerns the trailer makers quoted above, and which I propose to make the point of entry of my analysis of the trailer. Not only are
conclusion that has a moral purpose and a social function." Stories of the second kind „are the presumably factual accounts, the chronicles of the past and the news of today. Stories of what things are may confirm or deny some conception of how things work." Stories of third kind tell us what to do. They are instructions, laws, regulations, cautionary tales, commands, slogans, sermons and exhortations. Stories of the third kind clinch the lessons of the first two and turn them into action. „Today," Gerbner writes, „ most of them are called commercials; they are the advertising messages and images we see and hear ever)' day" (Shanahan/Morgan 1999, IX-X).
9 In consumer product advertising, advertising discourse could be said to be structurally conform to the product in cases where the advertising consists of the distribution of product samples.
(<
trailers, like films, examples of cinematic narration, in the sense that they convey a specific plot with cinematic means (Bordwell 1985, 53). Trailers are in most cases also made of the material of the film. As early as the late teens industry professionals noted that this constitutes one of the strengths of the trailer. I'he trailer, they said, lets the film „speak for itself". By using excerpts, lite (railer conveys more specific information about the film, including non-verbal sensory or haptic information about what
the film might actually feel like in the theater, than any other tool of film advertising.                        ,  .
J On the other hand the structural and procedural affinity of film and trailer poses a problem.
Depending on how you define the film merchandise, some parts of it may not be given away in
advertising, lest you loose its essence or, more prosaic terms, you destroy its commodity value. The
rules and formulas quoted above indicate that at least during what is generally called the classical
period of Hollywood film production, the film's essence was the story. Stars, spectacle, setting, action
and mood could be used to sell the film, because each of these selling points alone is less than the
whole of the commodity on sale^Th^sfapry^onihe other hand, was not to be given away because it
provided the underlying unity and coherence of the film's other elements. Which is not to say that the
story was necessarily more important or valuable to producers and audiences than spectacle or stars.
It is just to say that to classical trailer producers the story was, in a way, a more perishable good than
the other elements of a film.
Let us assume, then, that the procedural affinity of product and advertising discourse does indeed put the perishable good of the story in a constant danger of being exposed to the audience - or to their activity of narrative comprehension - too soon. The question then is to what extent the work of trailer makers was actually informed by an anxiety of telling the story too soon and, accordingly, characterized by an adherence to a rule of not giving away the story in the trailer. Before I turn to a discussion of how this question may be answered, I would like to add a few more remarks on how merchandise and advertising discourse relate in the film and the trailer.
Trailer and Film: Mapping a Field of Mutual Determinations
7
If the poster is an index for the film, the trailer is a sample. „Showing a few scenes" was the original
gesture of the trailer when it first emerged in the mid-teens. The logic of this apparently simple gesture is in fact rather complex. Excerpts from a film shown in a trailer represent the film in a manner similar to the way a textile sample represents a textile or the way a food sample represents a dish. Textile and food samples differ from the textiles and the food they represent primarily in terms of quantity. They indicate that there is more of the same, and they give an idea of the supply's specific quality. Excerpts in trailers on the other hand first and foremost represent themselves. In addition, they indicate the possibility of a replay or a repetition of the same material in a different context.
The difference between an excerpt in a trailer and an excerpt in a film first of all pertains to a micro structural level. In his analysis of narrative comprehension of films, Edward Branigan lists eight different levels of narration - text, fiction, story world, event/scene, action, verbal utterance, perception and thought - and three types of the distribution of knowledge: the narration of the "narrator", the narration of the "actor", and the narration of the "focalizer": "In a strict sense, a narrator offers statements about; an actor/agent acts on or is acted upon; and a focalizer has an experience of (Branigan 1992,105). In a film, the interplay between these agents of narration creates a complex texture, of which little remains if an excerpt is transferred to a trailer. In a trailer, a reaction shot may appear as a mere close-up, or a point-of-view shot can be transformed into a neutral shot. Only the narration of the diegetic agent, the actor, remains largely unaffected by the transfer. Even the actor's narration, however, may be given a different meaning when appearing in a trailer. Harmless supporting actors may suddenly reappear as the evil antagonist; excerpts from a friendly conversation may turn into a call for murder; random gestures become decisive ones.
Trailers, then rearticulate the material on which they are based in a fashion similar to that of found footage films. While the found footage film could be characterized as a metatext, the frailer       ) belongs to what Gerard Genette proposes to call the paratexfe i.e. the textual elements that surround     Item the actual text and cue its reading. Other than a metatext, a paratext, if it is made up from materials of the text, may not completely alter the meaning of these materials. If the trailer, for example, creates expectations that diverge too much from what the film eventually offers, it can seriously damage the film's commercial prospects. The promises the trailer makes the film must keep. Most importantly.
while an experimental filmmaker like Bruce Connor may well take the artistic license ofjevealing the comedy inherent in military training films, the trailer, in his particular reassembling of found footage, may not misrepresent the genre of the film.
Trailers feature a number of relatively stable genre signals in the form of conventionalized markers that indicate certain types of experiences to be expected from a movie. Thus for instance, shrieking female voices on the trailer's soundtrack indicate horror and thrills, or rather they anticipate the action on screen and in the theater, where male adolescents are promised the entertaining and perversely satisfactory spectacle of their girlfriends shrieking along with the female on-screen victims in orgiastic fear. On a less interactive note, song medleys in trailers indicate the musical, while cliffhanger shoot-outs at the end of the trailer confirm the viewer's suspicion that the film advertised is indeed a western. From the point of view of film advertising then genres are not deep structures, but rather types of experiences and gratifications to be expected from a film, and it is the task of the   ■        V       )/
trailer to offer a reliable map of those experiences and gratifications to the prospective viewer of the
- ■.,.{                                                                                     '/<-,
film. In order to determine in more detail how excerpts, which are not conventionalized genre signals, contribute to the mapping of the film experience in the trailer, the work of Nelson Goodman may prove useful.
Nelson Goodman begins his book on the Languages of Art with a critique of the Peircean notion of the iconic sign (Goodman 1976, 3ff.). According to the proponents of that notion, an iconic sign - a painting, or a photograph, for instance - refers to its object in a relation of similarity. Goodman points out that in logical terms, a relation of similarity implies reversibility between sign and object. Such reversibility, however, is neither to be found in figurative painting, photographs nor in film. We are fully capable of apprehending a portrait of a historical personality such as general Wellington without knowing who the person represented is, without, that is, having to recur to a perceived similarity between object and symbol or sign. Accordingly, an image does not denote its object in a manner substantially different from the way in which a spoken or written word refers to its object, which also implies that an image is no less a conventional sign or symbol than a spoken or written word. Goodman proposes to account for the difference between images and words not in
*"2
9
i . ; terms or their denotation or their motivation, as a classical semiotician would, but in terms of their
structure.                                                                                                                                     >~*A
>
In Goodman's taxonomy of sings, or symbols, the crucial difference is the difference between dense and differentiated symbols, which he illustrates by using the example of the thermometer. Thermometers come in graded or ungraded varieties. A graded thermometer indicates a difference in temperature through a scale of discrete numbers and the passage from one number to the next. An ungraded thermometer on the other hand, indicates temperature through a quicksilver column without reference to a scale of discrete numbers. In an ungraded thermometer then, every difference makes a difference. An ungraded thermometer, in which every difference makes a difference, would be an example for a dense symbol system. A graded thermometer, on the other hand, would be an example for a differentiated symbol. Along this same divide between dense and differentiated symbol systems, images are dense and words are differentiated symbols.
Differentiated symbol schemata denote objects. Dense symbol schemata denote objects, but they also exemplify qualities. These are primarily the symbol's own. A greyish watercolor painting of a coastline not only denotes a particular landscape, but also exemplifies the quality „grey". The word „grey", on the other hand, denotes a particular quality of the painting, or exemplified by the painting. Exemplification, then, works like denotation, only the other way around, or in the other direction. Furthermore, if the quality exemplified by a dense symbol is described in metaphorical terms - for instance, the quality exemplified by a gray painting of a coastline can be said to exemplify the quality „sad"-, exemplification becomes expression, and the symbol expresses a metaphorical quality. In this example, it would be the metaphorical quality denoted by the word „sad".
Goodman's terminology of dense and differentiated symbol systems is useful for the analysis of trailers and film advertising in several ways. First of all, it allows us to describe the way in which advertising typography works. Graphically elaborate writing is a symbol system that is simultaneously differentiated and dense because in elaborate writing or print, typography acquires a pictorial quality. Graphically elaborate film titles such as those for the movie poster of Francis Ford Coppola's Dracula (Columbia 1991) or for Alien (TCF1979) denote the film, but they also exemplify, or rather express, specific qualities which by way of the title's denotation are also attributed to the film.
10
Trailers use graphical elements and particularly elaborate titles on a regular basis, to the point where writing in conjunction with or superimposed on images may be said to define to a large extent the trailer as a cinematic format. Roll-on titles in classical trailers exemplify experiential qualities of the film through their movement, and a title with the word „thrills" set in a blurred typography expresses the shivering horror which spectators may expect to feel once they see the entire film (III. 3).
Furthermore, Goodman's notion of dense and differentiated symbol systems can be used to describe the relationship between excerpts and the film they are taken from. A textile sample exemplifies the cloth it belongs to insofar as the qualities of the cloth will essentially be the same as those of the sample. Excerpts, on the other hand, are essentially visual tropes and express the qualities of the film ox the fictional world they denote. The trailer for Casablanca (Warner Bros. 1942) begins with an excerpt from a chase scene in the Kasbah that ends in a shooting. Shown in the film, this scene denotes a particular occurrence in the fictional world of the film. Shown in the trailer, this same scene expresses qualities such as „violence", „action" or „adventure". What we see is not only a black-and-white image of a fleeing man getting shot in the back. It is also a „suspenseful" image, an image that denotes suspense. Thus the image works not only to denote the occurrence in the film, but it also works as a metaphor, in this particular case as a metaphor for the generic qualities of thefilm (cf. the analysis of the Casablanca trailer in chapter 4). Furthermore, excerpts also quite often work as synecdoche. In the same way that „sail" can be a synecdoche for „ship", a single song in a trailer can be a synecdoche for „musical". Genre signals in trailers, then, can be highly conventionalized, such as female voices in thriller trailers, shoot-outs in_western and medleys in musical trailers, but they can alsobe construed through highly specific reframing of excerpts as visual tropes.
What remains to be answered is the question that qualities are preferentially expressed in trailers, or, in less terminological terms, according to which criteria excerpts are selected. Censorship rules and the self-regulation frameworks of the industry impose some of these criteria. The Advertising Code of Ethics, an appendix to the Production Code, was comparatively vague about what may be shown in trailers and what was to remain out of bounds.10 The MPPA's Ratings Board
10 Production Code of Ethics, MPA 1938.
11
current rules for advertising on the other hand are quite specific. In order to obtain the vital G-rating that allows trailers to be shown in all theaters, coming attractions may not feature profane language (with the exception of the expressions „hell" and „damn"), allusions to menage-ä-trois situations or to drug use. Naked breasts and other forms of full frontal nudity are barred, as well an explicit love scenes and the obvious display of gore. The MPPA's book of rules further contains an explicit injunction against scenes in which firearms and potential victims are shown in the same frame.
This injunction is particularly interesting because it seems to prove the validity of André Bazin's theory of „montage interdit". According to Bazin, one may not cut if the essence of the action depends on the simultaneous presence of two or more of its elements within the same frame, as in the recurrent scene in safari films where the hunter is shown in the same frame as his prey before the moment of the shooting (Bazin 1985, 59). Fully in line with Bazin, the MPAA's rule implies that showing a weapon and a potential victim in one frame lends a decisive credibility to the threat, and one that is too strong for general audiences. The same rule does not apply for trailers with an R-rating, by the way. R-rated trailers, however, may still not feature mutilations, genitals, pubic hair and intercourse. Up until the mid-nineties trailer makers were also clearly warned against showing, or even implying, homosexuality.^^
Within the framework of such guidelines, trailer producers select excerpts according to criteria, which apparently have changed little over the years. Trailer producer Andrew Kuehn, a veteran of the industry with a forty-three year career spanning trailer work for such films as Dr. Zhivago (MGM 1966), }aws (Universal 1975) or Titanic (Paramoimt/TCF 1997), recalls selecting excerpts for a re-release trailer for Casablanca in 1992. Kuehn claims that when he got to see the original trailer after the completion of his own trailer, he realized that his colleagues at the Warner Brothers trailer department of fifty years before had essentially chosen the same excerpts to sell the film.12
One could try to elaborate a historical criteriology of the selection of excerpts based on analysis of a significant number of case studies comparing trailers to films. Such an analysis, however,
11 MPAA Advertising Handbook, p. 16f.
12 Interview with Andrew Kuehn, Los Angeles, December 7,1997.
12
is beyond the scope oř this study. What can be determined on the basis of an analysis of a significant number of trailer examples is how trailer producers have handled the problem of the distribution of information and knowledge over the course of time. As a general rule, trailers address an audience that has not yet seen the film. They must be able to stand on their own as representations of the film. How trailers attempt to convey a sufficient and adequate notion of the film can thus be studied on the basis of the trailers alone, without reference to the films.
A Serial History of the Trailer as Narrative Form
What I would like to propose in the remaining part of this chapter is a serial history of the trailer in the   * sense suggested by French film historian Michěle Langy. The point of a s^rialhistory, which is based
on the analysis of a large series of examples of the same format or genre, according to Lagny

is not to build a theoretical model, a structural scheme, which is supposed to explain a mode
of textual (in our case, visual) functioning. What matters instead is to submit the entire series
to the same kind of questioning and to the same viewpoints, which may lead us to formulate
conclusions on permanence, evolutions or ruptures (Lagny 1994,33f.).
[^   L-L-.r,   t~-L^
■
A serial history thus allows us to describe not only textual or audiovisual modes and to establish models or ideal types against which specific examples may be judged as typical or deviant for a given situation. A serial history also makes it possible to closely trace the historical development of audiovisual modes and to account in some detail for the shifts, the discontinuities and ruptures in their development.
The first question one has to answer before embarking on a serial history of an entire filmic format such as the trailer is how to select the examples and how to compose the corpus on which the analysis will be based. It is of course quite impossible to include every trailer ever made. Not only would there be too many, but many of the examples are also lost, particularly a majority of trailers from the silent feature era. On the other hand, trailers were and are produced industrially by a limited
13
number of companies. Most American trailers of the years between 1919 and 1965 actually came from
one company, the New York-based National Screen Service Corporation, and while in the late sixties, a group of small companies entered trailer production, their number has remained relatively limited and stable, with Andrew Kuehn's Kaleidoscope Films, a company with 180 employees, as the market leader since the early seventies. Furthermore, trailer producers, like film producers, will tend to achieve economies of scale. One of the ways in which economies of scale can be achieved is by basing their work on preconceived and relatively stereotypical notions of what a trailer is. Given the industrial nature of trailer production, then, any randomly selected corpus of examples will be sufficiently indicative of the general trends and shifts in the trailer's development.
In order to gain as complete an overview as possible over the development of the trailer, 1 chose to include examples from the entire period for which there are records of the production and the use of trailers in American cinema, i.e. from 1912 through the end of the nineties, when this study was conducted. Trailers were chosen irrespective of the genre of the film. On the one hand, recent studies have convincingly argued that genres are a lot less stable than structuralist genre theorist of the Seventies tended to assume (Altman 1998; 1999,30f.). On the other hand, 1 would like to argue that J advertising not only communicates the genre of a film, but also actively contributes toward the construction of genre as a system of types of gratifications to be expected from films. In order to understand how trailers map genre, then, one has to include examples from as many different genres as possible.
Trailers were also chosen irrespective of studio origin. Given the standardized and centralized character of production, it is safe to guess that studio styles in trailers are less pronounced even in the classical studio era than one might assume. Again, a comparative analysis is called for, which necessitates the inclusion of examples from as many studios as possible. In view of a number of case studies 1 also included all trailers I could find for films featuring James Stewart, directed by Cecil B. DeMille or Alfred Hitchcock and produced by David O. Selznick. In order to determine he point of saturation, a criterion established by Vladimir Propp proved helpful. In his study on the Russian magic tale, Propp suggests that a corpus has reached the point of saturation and may be called
14
representative when neither the addition nor the subtraction of a few examples significantly changes the results of the analysis (Propp 1987,34).
This study was eventually based on a corpus of 2039 examples, of which 1512 were theatrical trailers, 253 TV-spots, 157 so-called teaser trailers, 63 so-called featurettes or „making of'-trailers, 2U institutional trailers or theatrical and on-screen public service announcements, 13 so-called product-reels or trailers destined for an audience of exhibitors, and 7 Vitaphone shorts featuring veiled announcements for films. The corpus further includes eight radio spots and two manuscripts for pre--scripted radio interviews. Also, there are a newsreels segment, a pre-produced TV talk show clip, a manuscript for a pre-scripted TV interview, and a trailer for a vaudeville attraction from the silent era (which was not an unusual feature of theater programming so long as films were still combined with stage attractions). The corpus averages twenty examples for each year (check tables 1-7 of the corpus file on the accompanying CD-ROM for a detailed description of the corpus).
The analysis was based on two theoretical assumptions. First I consider the trailer to be a form of cinematic narration, which means that it is a process in which stylistic and other filmic means work together to construct a specific plot and convey notions of a fictional world. Second I take the problem of the distribution of information, of conveying a sufficient and adequate notion of the film, to be the crucial problem of film advertising, and I take the configuration of plot and style in trailers to be, in each case, a specific solution to that particular problem. Accordingly, I asked two basic questions of each example: How is the trailer's plot structured, i.e. how much, and what, docs it tell us about the plot of the film? And secondly, which stylistic and rhetorical means does the trailer use to construct its plot?
In order to answer the first question, we have to develop a working hypothesis. Whether a trailer conveys a comprehensive notion of what a story the film tells depends less on whether the trailer foregrounds the story as a selling point, as the analysis of the trailer for Casablanca in chapter 4 will show. Rather, what is important is whether the trailer's plot suggests a sequence of events that can be construed to represent the story of the film. A model for such a sequence of events may be found in the literature about drama theory and screenwriting. Screenwriting manuals by authors such as Syd Field and Robert McKee teach us that most mainstream films are structured in three acts: Exposition, confrontation and resolution (Eder 1999, 26f.).
/
15
According to Field's model, exposition and confrontation are linked by what is called „plot point 1", also known as the initiating event (Field 1988,14). Far from being original in his formulaic prescription, Field actually draws on the drama theory of German playwright and theorist Gustav Freytag. first published as Die Technik des Dramas in 1870, a book which in turn draws on Aristotle's Poetics. In Freytag's terminology, exposition is the introduction, plot point 1 is the „moment of excitement", the confrontation is the „enhancement", and plot point 2 is the „tragical moment." Freytag further calls the resolution a „descending action". Like Aristotle, he argues that the descending action is composed of two elements, the peripateia or turnaround and the catastrophe (Freytag 1983,105ff.).
Formal analyses of American mainstream movies convincingly show that a majority of films closely adhere to such formulae. There is some disagreement over whether the model they follow is that of a basic three-act structure or, as Kristin Thompson has argued a few years ago, that of a four-act structure along the lines of Freytag or Aristotle (Bordwell/Staiger/Thompson 1985; Thompson 1998). There is no disagreement however with regard to the fact that most mainstream movies are protagonist-centered, i.e. built around one central, goal-oriented character. Given the almost canonical status of the protagonist centered three-act (or three/four-act) plot structure in mainstream cinema, trailers may well be measured against this particular model in order to determine whether they suggest it sequence of events which could be construed to represent to the story of the film. The question then is whether trailers convey story content in formulae like „James Stewart stars in a story full of violence and passion..." or whether they use linear story resumes along the lines of „James Stewart is x, who finds himself in situation y, when z happens, from which conflict p ensues, at the height of which endings q or non-q suggest themselves as possible outcomes."
Apart from determining the narrativejjrucjm-e of the trailer, the analysis has to seek out other patterns ol construction that may be used independently of the respective strategy of distributing or divulging information about the film's story content. In particular, one has to pay close attention to basic structural patterns below the level of narrative plot, as well as to micro structural elements, such as the already mentioned medley part in musical trailers, or the often-used lists of attractions, detailing the various spectacular scenes in a film (cf. illustration 4). The analysis conducted for this

16
study also included checking off a list of 23 formal and 35 rhetorical parameters that may or may not
be used by trailers. Formal parameters include length or the trailer, the average shot length,'3 the emplacement of the film's title and the naming of the cast, the use of roll-on titles, graphic elements, wipes, voice-over or stills, freeze-frames etc. Rhetorical parameters include topoi such as „never before..." or „now" usually used by extra fictional enunciations such as the roll-on titles or the voice-over (for a detailed list of the parameters analyzed see CD-ROM, corpus, section II).
Trailers and the Narrative Turn in Film Advertising
The main results of the serial history of the trailer I propose may be summed up as follows:
Most trailers from the classical period14 feature what you might call a mystery plot. Instead of suggesting a linear sequence of events, which may be construed to represent the story of the film, classical trailers tend to raise a bundle of questions about the film and about its story, leaving their addressees all but in the dark about what is really going on in the film. Thus the film promises to be the solution to the riddle presented by the trailer. Contemporary trailers on the other hand, and particularly trailers produced after the mid-seventies, feature what you might call a suspense plot.
13 Length of the trailer in seconds/(number of cuts + 1).
14 Bord well/Sta iger/Thompson (1985) define the classical period of Hollywood cinema in terms of
style. According to them, the classical era covers the period from 1917 to 1960, when a narrative mode V -*?$ dominated the industrial production of films that merited the term classical not least thanks to its
longevity and its coherence. Other scholars such as Noll Brinckmann (1997, 280) prefer to limit the ** **"£.
classical era to the period from the full development of the sound film to the mid 50ties, because it is      hw /*-only in this relatively short period that a homogeneity in the use of stylistic parameters and genre                  . 'JC*
conventions can be observed which merits the label „classical". I propose to use the term classical Hollywood cinema primarily in an economic sense. The classical era thus covers the first period of full vertical integration, which begins in the late teens and ends with the divorcement of studios and theater chains and the divestiture of theatre chains in the early fifties.
17

They are usually based on story resumes closely following the canonical three-act structure, but break off with a cliffhanger after the second act and the second plot point. Furthermore, regardless of the plot form used, a vast majority of examples from the pre-1970 period feature a similar sequence or basic elements, a pattern of construction. The story resume characteristic of more recent trailers first occurs within the framework of this classical structure. In the early eighties, however, the story resume evolves into a basic structural pattern in its own right. Also, the criterion of structure (does (he trailer feature a classical structure?) and the criterion of plot (does the trailer feature a suspense plot?) can be combined in the analysis. Bracketed together, they form a set of four basic types or models of trailers. As it turns out, every example in the corpus - as well, indeed, as every example outside of the corpus - can be attributed to one of those four types (which is partly due of course to the fact that the fourth type, defined by both the absence of classical structure and suspense plot, is an omnibus category which includes all examples not covered by types 1 to 3; see discussion below).
Furthermore, the analysis shows that the length of trailers is surprisingly stable at around two minutes and twenty seconds for almost the entire period under analysis. Trailers from the seventies      A   :(-tend to be longer, a period, coincidentally, when American car design favors expansiveness as well.                    ^y
While length remains stable, significant change can be observed in the average shot length. From the       .   a twenties to the late nineties, the average shot length decreases steadily, although not in a continuous f -■> ' *>-' ~ curve, from nearly five seconds tojusXovgxgnefffiQgd- Also, while roll-on titles and wipes for all                 '
( practical purposes define the trailer as a format in the classical era, they quickly disappear in the years / after 1960, to resurface only in trailers clearly intended as parodies of classical trailers. Voice-over     ' ' '•' ~ t/^~t-^ ^narration, technically and economically feasible after 1933, increases over time and becomes all but ubiquitous in the eighties. This is also the time when the stentorian bass of one speaker, Don Lafontaine, can be heard at least once in virtually every pre-show trailer program in American movie theaters. Still photographs are mainly used in trailers from the thirties and from the late sixties and early seventies, for reasons that will be discussed in chapters 4 and 6 respectively. Significant changes may also be observed with regard to the editing of trailers. Classical trailers show excerpts more or less intact, with a certain license for shortening scenes in order to speed up delivery. More recent
L8