Chapter 12
Character in the screenplay text
Steven Price
Now you're getting to what screenwriting's all about: character.
(Lew Hunter1)
There is no character.
(David Mamet2)
The extreme polarity between these two statements, the first by the writer of a popular manual and the second by one of the most celebrated of screenwriters, encapsulates a problem with the study of screenplay character. Within film studies, written texts are usually subordinated to or consumed within the analysis of cinema, and there have been understandably few attempts to examine character independently of the actor's performance or other aspects of the film text. (For exceptions, see Sternberg 1997: 108—30; Price 2010: 124—31.) Conversely, in screenwriting manuals, of which Lew Hunter's is a good representative example, discussion of the topic is voluminous, but conducted from the perspective of the practitioner or teacher who is concerned more with the process of developing a character suitable for filming than with retrospective textual analysis; with creative production, rather than critical consumption.
Consequendy manuals often fall foul of one of the basic principles of literary criticism by encouraging the perception of characters as real people. Hunter insists that 'fine screenwriting comes down to the characters' (1994: 71), and reproduces a number of character sketches written by his students. As Michael Hauge observes,
[m]any teachers recommend writing full biographies of all your characters, or at least the primary ones, before beginning the screenplay itself. At the very least, outiine your main characters' lives from birth until their appearance in your story to ensure that you know them at least as well as you know your best friends. Even though much of this background material will never be revealed in the screenplay itself, your characters will function much more consistendy realistically, and effectively it you know the details of their lives.
(Hauge 1989: 39)
202   Steven Price
Such a passage demonstrates the dangers of using manuals as critical studies, rather than as writing aids. Literary criticism has long insisted that characters are textual constructs, and that no more can be known about them than what the text provides. This approach can be traced back at least as far as the rejection by the 'New Critics' in the 1920s of the kind of naive assumptions about character encapsulated in A. C. Bradley's then-influential Shakespearean Tragedy (1904). New Criticism saw itself as freeing criticism of subjective speculation by focusing on the formal and linguistic properties of the textual artefact. For example, W. K. Wimsatt and Monroe C. Beardsley's (1946) 'intentional fallacy' proposed that it is impossible to reverse engineer a text to arrive at the author's intentions. Arguably, the same holds for character: all we can know about the character is what is present on the page, and from a critical (as opposed to a 'creative') perspective, such exercises as those endorsed by Hunter and Hauge are poindess. The ironic question posed in the tide of L. C. Knights's 1933 essay 'How Many Children Had Lady Macbeth?' had the welcome effect of driving the stake through the heart of biographical character study.
Novelistic and screenplay conceptions of character
So successful was this line of reasoning that by 1979, when Richard Dyer published his seminal analysis of Stars, it had become something of a problem, because 'in so far as there has been any theoretical consideration of character in fiction (in any medium), it has primarily been directed to exposing its fallacious aspects' (Dyer 1998: 89). What such an exposure failed to account for was the effect that characters often give of somehow exceeding the texts in which they appear. Consequently there had been a kind of critical short-circuit: 'having demonstrated that characters are not real people, that they are an effect of the text constructions, critics and theorists have not proceeded to an examination of how this effect, so widely known and understood, is achieved, and just what the rules of construction are' (Dyer 1998: 89).
Some of these 'rules of construction' are evident in the nine 'qualities' Dyer isolates in 'the novelistic conception of character'. These are sufnciendy familiar as to require little elaboration here, but collectively they construct what remains the dominant idea of character in Western culture, deriving from the growth of capitalism and the concurrent development of notions of liberal humanism. This ideology was fully expressed in the rise of the novel, in which the character is an autonomous individual, possessing multiple and perhaps contradictory qualities to give an impression of wholeness, roundness and uniqueness, and with motivating desires that help to propel his or her story arc.
Alternative conceptions of character are readily available, of course: among Dyer's examples is the use of 'types' to represent general classes or interests in Sergei Eisenstein's Marxist dramas of class conflict, and we should add here the
Character in the screenplay text 203
structuralist understanding of character discussed below in relation to the work of David Mamet. Moreover, the dominant paradigm is not without inherent difficulties. For example, the protagonist is often required to 'develop' and yet to stay, essentially, the same. Most significant from the present perspective of screenwriting is that the novelistic character possesses both 'interiority' (the ability to detail this 'without necessary recourse to inferences from what s/he says aloud, does or looks like' (Dyer 1998: 94) is the novel's trump card), and 'discrete identity', that paradoxical quality of appearing to exist somehow independently of the text. This 'is a problem for any narrative form' (Dyer 1998: 95), presenting the same logical absurdities as the character's 'backstory' that is recommended in screenwriting manuals but that cannot appear in the film.
A partial answer to this problem is simple, if tautological: it is because the 'novelistic' character has been constructed as a free agent that s/he appears to have an existence beyond the limits of the text. It is perfecdy possible to create texts that do not generate this effect, so differing theoretical notions of character are also tied to generic distinctions between different kinds of text, as well as to different ideological conceptions of the human individual. For example, Leo Braudy in a discussion of character that Dyer analyses at length (Dyer 1998: 101—3), proposes a distinction between 'closed' and 'open' films. 'Open' films, such as those of Jean Renoir, suggest the character has a life that persists beyond the frame of the film; 'closed' films, like those of Alfred Hitchcock and Fritz Lang, do not. Several years later, the drama critic John Peter argued for a more or less identical distinction between open and closed plays (Peter 1987). He tided his book Vladimir's Carrot because that object exemplifies the workings of a closed play: it simply does not occur to an audience of Waiting for Godot to ask where Vladimir obtains the carrots he produces on stage, because we do not imagine his world to be an extension of or metonymically related to our own, but recognize it instead as a self-referential structure.
That screenwriting manuals often recommend the creation of characters with a full biography, whose decisions present a further revelation of the self, is a sure sign that the 'novelistic' conception of character has become so ubiquitous as almost to appear beyond ideology altogether: a character is an autonomous individual with the freedom to choose. What is remarkable in the present context is that this novelistic conception is in fact, and almost by definition, very clearly differentiated from what we might call the screenplay conception of character. Arguably, the screenwriter ordinarily has access to none of the most common relevant methods of characterization exploited by the novelist: detailed physical description, the ability to describe inner thought and the broader possibilities of omniscient narration.
An accumulation of physical detail, including facial features and build as well as the semiotics of fashion, clothing, hairstyles, designer brands and the like, not only gives some indication of character, but also, through the accumulation of redundant detail, helps to create a reality effect whereby the character appears securely anchored in a finely realized storyworld. Yet the screenwriter is unable to
204   Steven Price
present such detailed physical description, pardy because of the need to defer in such matters to directors, designers and actors, and partly due to the sometimes disputed convention that a page of script is equivalent to a minute of screen time. The writer simply does not have the words at his or her disposal to engage in leisurely description of people or places. In short, the screenplay is a structuring document that demands concentration on the shape of the story and the succession of events rather than on redundant physical detail.
Second, the novelist may give direct access to the thoughts and inner life of the character by such means as interior monologue and free indirect speech. The only equivalent techniques available to the screenwriter are the montage signifying a succession of thoughts, and the voiceover. Of these, the montage, besides now being rather cliched, cannot capture individual voice in the way that prose narration can. Meanwhile, voiceover has often been dismissed in film criticism as a manipulative literary device that falls victim to the 'specificity thesis'. This proposition, associated with Rudolf Arnheim among several other early film theorists, and still influential, holds that the art of cinema consists primarily in camera and editing, since these are specific to the medium, and not in dialogue, which is theatrical. The thesis has itself more recently come under sustained attack from several quarters, chiefly on the grounds that it artificially privileges one element of what has always been a hybrid and synaesthetic medium (see Carroll 1992). Nonetheless, voiceover is still widely viewed with suspicion, and in any case is generally used either in scripts for particular kinds of film, such as the 'art movie' in which there may be a conflation of not only writer and director but also protagonist, or to explain to the audience through narration what could not be satisfactorily achieved by other means. Examples are the introductory, expository voiceovers, with accompanying visual montage, that orient the spectator within what might otherwise be the confusing storyworlds of Casablanca (Michael Curtiz, 1941) and The ThirdMan (Carol Reed, 1949). Such a use of voiceover may indicate a perceived difficulty with the screenplay. The introductory narration for The Third Man, for example, was developed at a late stage when it was feared the audience would not understand the complicated division of powers in post-war Vienna; the speech does not appear in Graham Greene's original screenplay (White 2003: 7-9).3
A third, related novelistic method is authoritative narrational commentary about characters. The problem of narration in film is too complex to engage in the present context, but related if simpler difficulties bedevil the screenplay. The clearest approach to this question is perhaps that outlined by Claudia Sternberg, who identifies three different modes in the non-dialogue elements of the screenplay text. '[T]he mode of description is composed of detailed sections about production design in addition to economical slug-line reductions' (Sternberg 1997: 71). The report mode is the temporal sequence of actions, usually human. Of greatest interest here is the third, comment mode, whereby the text offers a commentary on events. As Sternberg observes, such commentary, akin to the authorial narration of prose fiction, is routinely prohibited in manuals. The assumption is that such
Character in the screenplay text 205
comments cannot be filmed, and even if they could, they would be the province of the director rather than the writer. Sternberg easily refutes this by pointing out that her sample range of Hollywood screenplays contains innumerable figures of speech and other things that cannot be shown or seen. Indeed, according to Sternberg 'screenwriters rarely miss the opportunity to use the mode of comment' (Sternberg 1997: 74). This final remark seems to overstate the case, however, since most screenplays are sparing at best in direct comment.
To these three novelistic modes of character presentation we may add the freedom generally afforded the novelist to present speech in written forms, either dialogic or monologic, that would sound wholly unnatural if recited orally. An extreme example is Marlow's 'yarn' in Heart of Darkness, which comprises virtually all of Joseph Conrad's 1899 novella, and is replete with highly literary techniques in description, characterization and dialogue. Seemingly oral recitation can thereby become conflated with textual narration in ways that are unavailable to any screenplay that attempts to create realistic speech.
A provisional conclusion is that screenplay character is, necessarily, generically distinct from novelistic character. For example, the relative lack of access to the screenplay character's inner world, or to its contemplation of the various discourses — legal, religious, educational, etc. — that may construct and define it as a social subject, makes him or her a more consistendy existentialist being than an equivalent figure in realist prose fiction. In the screenplay we see only the actions through which the character responds to, and carves out an identity for itself within, this social world. Of course, this is also an illusion: the screenplay character's series of actions is orchestrated by the text, and to say that this character is an effect of structure is only a different way of saying that it is an effect of narration. Nonetheless, the illusion is of a different kind, or possesses a different set of emphases, than that created by the novel.
But there is more than one way of writing a screenplay, and different styles of writing produce different effects of character. To illustrate this we may contrast two celebrated scripts: David Mamet's House of Games, which adheres rigorously to the disciplines of screenwriting outlined above, and whose author conceives of screenwriting in structuralist terms; and Graham Greene's The Third Man, which not surprisingly retains much of the sensibility of the novelist.
The structuralist conception of character: David Mamet and House of Games
In a classically structuralist analysis, there can be no autonomous, individual character. The 'character' has no 'positive' or innate qualities, and instead exists only as one term within a structure of signs, assuming its identity to the extent that it differs from, and operates in relation to, the other terms. The 'hero', for example, acquires definition in relation to the 'villain'. In his Morphology of the Folktale (1928), the proto-structuralist Vladimir Propp avoids the ideological
206   Steven Price
connotations of the word 'character' and instead identifies a common structure to the tales he analyses, each of which consists of a selection of 31 possible 'functions', performed in an invariable sequence by the dramatis personae who occupy seven 'spheres of action' (villain, donor, helper, princess, dispatcher, hero and false hero) (Propp 1984: 23). The sequence of these functions remains fixed, though not all would appear in a given tale.
Mamet's comments on writing and film indicate that he views texts and stories in similarly structuralist terms. He admires the work of Joseph Campbell (see Kane 2001: 209), who identifies a 'monomyth' in Western storytelling which in many respects resembles Propp's recurrent tale: 'A hero ventures forth from the world of common day into a region of supernatural wonder: fabulous forces are there encountered and a decisive victory is won: the hero comes back from this mysterious adventure with the power to bestow boons on his fellow man' (Campbell 1949: 30).4 The fairytale itself is one of Mamet's preferred 'teaching tool[s]', because it is 'told in the simplest of images and without elaboration, without an attempt to characterize' (Mamet 1994: 396). He argues that 'all there is in a movie [is] structure' (Kane 2001: 66), affirming that the task of the writer begins with the creation of a 'logical structure', after which 'the ego of the structuralist hands the outiine to the id, who will write the dialogue' (Mamet 1994: 346). Characters themselves are 'nothing but habitual action' (Kane 2001: 40). More broadly, in an essay against realism, he observes that '[i]n general, each facet of every production must be weighed and understood solely on the basis of its interrelationship to the other elements' (Mamet 1994: 201; my emphasis). He believes that the audience finds it easier to 'identify with the pursuit of a goal' than with '"character traits'", because 'those idiosyncrasies ... divide us from [the protagonist]' (Mamet 1994: 406).
This distinction between the protagonist and his or her 'idiosyncrasies' resembles Dyer's differentiation between 'character', which 'refer[s] to the constructed personages of films', and 'personality', which is 'the set of traits and characteristics with which the film endows them' (Dyer 1998: 89—90). Unsurprisingly, both 'personality' and the comment mode are almost completely absent from a Mamet script. This is the case even in his first filmed screenplays, for The Postman Always Rings Twice (Bob Rafelson, 1981), The Verdict (Sidney Lumet, 1982), and The Untouchables (Brian De Palma, 1986). While all of these display a similar authorial method to that in the screenplays Mamet would later direct himself, the work of the other directors, and the performances of the major stars Jack Nicholson, Paul Newman, and Robert de Niro, respectively, give those characters a greater sense of 'openness', in keeping with the Method acting or New Hollywood directorial style with which most of these figures are associated. It is therefore quite possible for a Mamet script to be read or filmed in such a way that a conventional sense of 'character' emerges. For example, in one of the better screenwriting manuals Paul Lucey adopts the figure of Galvin in The Verdict as his principal 'character study' (Lucey 1996: 109—37). Lucey's decision may have been influenced by Paul Newman's unforgettable
Character in the screenplay text 207
performance, but it also indicates that character can be, as it were, 'read into' — or out of — Mamet's words.
When Mamet directs his own films, however, the full consequences of his theoretical conception of character emerge. This is well illustrated in the screenplay for his directorial debut, House of Games (1987), in which Margaret Ford, an academic who has just published a book on psychology, is drawn into the increasingly complex and dangerous world of a team of confidence men headed by the antagonist, Mike. The drama comes to revolve around the erasure of an initial structural opposition of Margaret (female, professional middle class, student of psychology) and Mike (male, criminal underclass, practitioner of confidence tricks). At the heart of this opposition lie the questions of whether Margaret or Mike has the better understanding of the mind, and whether or not there is a deep psychology to be unearthed. In other words, Mamet's first film as writer-director is an interrogation of what the human — and what the 'character' — is.
The script opens with the following sequence:
People hurrying to work across a crowded plaza. Camera moves  forward toward a coffee cart in the background.
A young woman walks into the frame in the foreground. She takes a book out of her purse,   looks down at the book.5
The above description simply reports the sequence of events that the imagined spectator (or camera) is imagined to observe on, or record for, an imagined screen. It does not comment on these actions, nor does it give any authorial or narratorial insight into the woman's character, grant her any 'traits', or even provide any physical description bar the kind of approximation of age — 'young' — that would occur to any observer.
This method persists throughout the text. The reader does not receive any direct indication of the age of the protagonist, Margaret, whose autograph the young woman solicits as the opening scene continues. Mamet is, apparently, neither creating nor referring back to any visualization or interpretation of 'character'. He is simply describing a series of actions, and it is as if the names alone are sufficient to distinguish one figure from another. We know the gender, we are sometimes told that a person is 'about thirty', for example, but otherwise there is simply a series of actions involving several figures who interrelate in ways that form the structure that is the screenplay.
Yet it is not quite true that there is no comment in the script. Margaret's apartment, for example, is '[ojbviously the abode of a single woman' (28). Is that a description, or an interpretive comment? It appears to be a subtle direction to the readers, or to the designer (who is also a reader), indicating that the apartment must be set in such a way that the cinema spectator will interpret the shot as the author intends. At such moments the text is interpretive, but only so that it can describe the effect the reader is to imagine will be generated by the screen image.
208   Steven Price
This is more significant than at first appears. Later, once Margaret has been drawn more deeply into the world of the confidence men, the camera adopts her point of view when she and Mike arrive on the scene of a complicated con, in which Mike and his gang attempt to trick a businessman into giving them his money in exchange for a suitcase he thinks contains a fortune. Her involvement in this begins when she sees two men in the street 'conversing, as after a good meal' (42). One of these is the 'Vegas man', whom Margaret already knows to be part of the gang; the other is the businessman. The depiction of how the two men speak ('as after a good meal') is another example of a clause that is both a description and a comment, again with the apparent aim of indicating the required effect on the screen. Then '[t]he cab drives away. But the Vegas man has forgotten his suitcase' (43). Conjunctions such as 'but' should logically be omitted in the rigorously paratactic style for which Mamet strives, because, by providing a connection between the material in two sentences or clauses, they comment on the action. Moreover, there is a trace of interiority in the declaration that 'the Vegas man has forgotten his suitcase'. Once it becomes clear that the businessman is sufnciendy greedy to have taken the bait, Mamet will supply another descriptive-interpretive phrase when 'Mike turns to Ford, [and] nods slightly, sadly, meaning you see what human nature is?' (46).
These tiny modifications to Mamet's paratactic style are remarkably suggestive. They can be reconciled with his purist conception of screenwriting by noting that cinema has always been able to signal interior thought through the juxtaposition of shots. The writing at this point in the script implies the use of the Kuleshov effect: the meaning of Mike's nod will emerge on the screen because the businessman has just shown that he wants something for nothing, and is therefore a natural 'mark'. Later in the script, after the businessman has been shot (having been exposed as an undercover policeman), there is an explicit revelation of interiority when memories of the shooting pass as visual images through Margaret's mind, revealing her horror, guilt and anxiety. It seems, then, that no matter how hard Mamet insists both in theory and in practice that there is no such thing as character, the concept cannot finally be dispensed with.
A similar argument could be made about his dialogue. In a self-deprecating preface to the published text, he records that to prepare for directing House of Games he used a simple version of Eisenstein's theory of montage 'to reduce the script, a fairly verbal psychological thriller, to a silent movie' (p. vii), following the principle that the juxtaposition of two shots creates a third, unspoken and unvisualized idea. Mike's nod provides an illustration, although it also helps to indicate why the comment mode sometimes has to intrude, since otherwise his gesture could appear ambiguous in the written text.
The belief that a film should be directed as if it were a silent movie, with the concomitant devaluation of dialogue, suggests an endorsement of the specificity thesis. Yet a glance at any Mamet script will show that, partly because it attempts to eschew comment, it is dominated by dialogue; and it is hardly surprising that the dialogue of this celebrated dramatist turns out to be essential to our understanding
Character in the screenplay text 209
of character. To take a simple example, Margaret first encounters Mike in the otherwise exclusively masculine domain of the title location, in which pool, poker and con tricks are the major currencies. She apparendy tries to adopt the idiom of the confidence men, but is comically inept at doing so ('Let's talk turkey, Pal', 13). Such lines indicate things about Margaret's character, particularly in relation to Mike: she is awkward and, in this environment, inferior to him, and although she is an expert on psychology she appears remarkably superficial and inauthentic.
And yet, in the brilliant twists of House of Games, such notions are confounded. First of all, despite what the text appears to say, the Vegas man has not 'forgotten' his suitcase: he has remembered to appear to forget it, this being the opening move in the con. The text does not reveal what is going on in the Vegas man's mind; it constructs what is going on in the reader's mind. This is a clue as to what is really happening in the 'Mamet movie'. We understand quickly enough that the Vegas man did not really forget anything. Only when the businessman is revealed as a police officer, however, do we have to reinterpret Mike's unspoken comment about human nature: the businessman's words in fact reveal nothing at all about human nature because he was only pretending to fall victim to the gang. Even this is not the end of the matter. At the climax of the film, Margaret discovers that the 'police officer' is alive and well and just another member of the gang, that she herself has been the 'mark' all along, and that everything we have seen has been a performance that revealed nothing about anyone's nature save her own. Mike's nod was just another deception. Although there are still further twists to come, the film appears to conclude with Margaret learning something about her own character: she herself is capable of becoming a thief and a con artist, and was latendy so from the beginning.
If this were really the conclusion we are supposed to reach, then for all its trickery House of Games would be an example of the Aristotelian principles of recognition and reversal that Mamet has frequendy insisted lie at the heart of drama. It would not sit easily, however, with the notion that 'there is no character'. That notion transmits itself readily to most spectators who recognize that it establishes something distinctive about Mamet's work. Everyone notices, for instance, that there is something peculiar about the delivery of the lines in a Mamet-directed film: the actor appears to be reciting, rather than simply speaking, the dialogue. This is one of the ways in which a Mamet film never possesses the illusion of reality, but instead calls attention to the film as a record of a script. His celebrated dialogue is not just audible but also, as it were, visible. This is literally so in the book cover, diary entries, notebooks, and building signs that pepper House of Games, but it is also in a different sense true of the spoken dialogue. One reviewer of the film felt the presence of 'the man, just off-screen, who wrote the screenplay and is monitoring everything the actor does' (Canby 1987); another remarked, 'you feel as if Mamet were in the seat next to you repeating, "This isn't real, this is . . . artifice'" (Hinson 1987).6
Although it appears to be a drama played out between Margaret and Mike, these figures have no real substance. The real drama is that between Mamet and the
21 0   Steven Price
audience or reader. This is absolutely in accordance with Mamet's understanding of film. Exposing character as an illusion created by the structure of the text, he locates the source of meaning not in the character, the actor, or even the director, but in the writer: 'The words are set and unchanging. Any worth in them was put there by the author' (Mamet 1997: 62). His method of working with a familiar ensemble — which may pardy explain the absence of description (Margaret Ford was played by Mamet's then-wife, Lindsay Crouse) — means that his actors 'will trust that the line's going to work and read the line as it's written' (Mamet, in Kane 2001: 158). Directing is simply an extension of writing; it is a record of the pro-filmic event, 'the work of constructing the shot list from the script' and of 'recording] what has been chosen to be recorded' (Mamet 1994: 349). His ideas about filmmaking therefore protect the writer's voice at the expense of the director's, just as, in John Lahr's words, 'his ideas about acting protect the author's voice at the expense of the actor's' (Lahr 1997: 78).
Character that exceeds the text: The Third Man
Mamet's screenplays suggest that the structuralist and paratactic style is not simply one way of rendering character with the aim of eliminating material that cannot be filmed. Instead, it constructs a generically distinct form of characterization that differs from that produced by other approaches to screenwriting, such as Graham Greene's in writing The Third Man. Greene's Harry Lime is a criminal racketeer who fakes his own death amidst the ruins of post-war Vienna. For most of the time his deception fools the audience, as well as both the authorities, headed by Colonel Calloway of the British Military Police, and Harry's naively innocent friend Holly Martins, whom Harry has invited to join him. Even though Harry does not appear until two-thirds of the way through and is on screen for barely 10 percent of the duration, he famously dominates the film. He therefore provides a fine example of the notion of character as something that appears to exceed the textual structure that seemingly confines it, and although Orson Welles's performance and persona undoubtedly contribute to this effect, it is also bound up with the way the character is created in the screenplay text.
One way of demonstrating the difference between the two kinds of screenplay considered in this chapter is by borrowing a distinction developed by Steven Maras between two 'discourses' surrounding screenwriting (Maras 2009). The first sees the text as a blueprint that completes the 'conception' stage; the filming is merely the execution of the idea. Mamet's ideas about the relationship between screenwriting and production offer the starkest possible illustration of this discourse, with the near-total absence of the comment mode in House of Games suggesting that the text is a self-sufficient document that requires no further elaboration. The second sees screenplay and filming as a continuous, evolving process that cannot be divided so easily into two stages. For this reason, Maras offers the term 'scripting' as a way of blurring the distinction, and to suggest that filming may itself be seen as a form of writing. The published text of The Third Man, which uses parentheses to indicate
Character in the screenplay text    21 I
material from the screenplay that was unnlmed or omitted from the release print, and footnotes to present material found in the film but not in Greene's screenplay, preserves a textual record of this 'scripting' process in The Third Man.
Different production practices produce different kinds of script. Mamet worked with his own screenplay and acting ensemble, so the text remained relatively stable, whereas in The Third Man Lime's character was altered significantly from Greene's original conception due to the nature of the collaboration. Most significantly, while the producer David O. Selznick initially wanted Noel Coward for the role, in the end the director Carol Reed won the day and Orson Welles was cast (Thomson 1997: 293—4). The resulting 'problem of fit' (Dyer 1998: 116) between actor and character was then partially resolved by alterations in the script, most memorably in Welles's authorship of the 'cuckoo clock' speech, which appears in the published text merely as a footnote, since it formed no part of Greene's creation.
There is a related sense of 'fit' in the relationships between different characters in the text. The casting of Joseph Cotten (rather than Selznick's preferred choice, Cary Grant) as Holly Martins reconstituted an old partnership familiar to cinema-goers from Citizen Kane (Welles, 1941), and this adds an extra-textual dimension to the understanding of the film. Even within the text, however, Harry Lime and Holly Martins each derive at least some of their meaning from the structural relationship to the other. Pardy this is a matter of narrative, with Harry the object of Holly's pursuit; but it is also an effect of a kind of doubling, whereby each character becomes more complex (more 'realistic', in this sense), either by taking on traits of the other or by coming to be seen as representing conflicting forces in a dynamic between two figures. Originally named 'Rollo' throughout Greene's script, the change of Martins's first name to 'Holly' in the film has the happy effect of suggesting this connection even at the level of sound; indeed, Anna frequendy calls Holly 'Harry' by mistake. This is doubly irritating to Martins once he realizes that he wishes to occupy his friend's former position as the lover of Anna, the girl Harry leaves behind.
This drama, in which one character unknowingly begins to take on the characteristics of another who is either dead or presumed to be so, particularly by moving into the physical spaces previously occupied by another, is common in literary fictions of the uncanny or the doppelganger, such as many of Poe's tales. It is also frequendy found in ghost or horror stories, perhaps the best cinematic example being The Tenant (Roman Polanski, 1976). Its significance in the present discussion is that it shows how the structural method of distinguishing characters by means of parallels and contrasts does not only establish one character's identity as an effect of its difference from others. It also shows how one character acquires identity by taking on aspects of the identities of others, and in so doing gains the appearance of greater substance than it would possess if it were really an autonomous and unique individual.
For example, our understanding of the child-like, playful aspect of Harry Lime, so sharply in contrast to the crimes he has committed, is prompted by the likeable Holly entering into and performing Harry's role from the beginning. As with any
21 2   Steven Price
literary doppelganger, the doubling presents not contrast but uncanny repetition: like Harry, Holly is a seemingly innocent American, a lover of westerns, wandering through a bomb-shattered Vienna, trying to survive by his wits, and falling in love with Anna. Each figure seeps into the other, so that when Harry finally emerges from the shadows it is as if we have known him already: not because of what other people have said about him, much of which is lies, but because we have already seen Holly acting out a version of the '[b]est friend I ever had' (25).
It is unusual for one character almost literally to embody another in this way, but much more common, of course, for the various personages in a screenplay to comment upon one another in ways that contribute to the construction of character. Still, if Mamet is right in arguing that character is 'nothing but habitual action' — a belief derived from Aristode's Poetics, which is an almost ubiquitous authority in screenwriting circles — then oral comment in the screenplay is likely to be minimal compared to the stage play. In House of Games, such comment does not take the form of extended reflection; instead, one character may offer a cutting epithet to define another, as when Mike dismisses Margaret simply as 'an addict (61).
But if a character is absent or even dead, as for example in Last Orders (Fred Schepisi, 2001), the verbal commentary of others becomes much more significant. This technique is remarkably extensive in The Third Man due to Harry's delayed appearance and the conflicting accounts of what has happened to him. Until his emergence from the shadows, he has existed not as an autonomous individual but as a series of verbal or textual stories or ideas created by others. Anna idealizes him as the romantic lover who could not have betrayed her, Holly is convinced of the essential decency of his childhood friend (his changing loyalties an index of the developing drama) and the textual fabric constructed by his underworld associates Kurtz and Tyler (Popescu in the film) slowly unravels under critical interrogation. Even Calloway's revelation of the extent of Harry's criminal depravity, while 'true', is a narrative construct nonetheless. As Holly remarks after hearing it, 'He never existed, we dreamed him' (83). Yet this dream-figure who appears so briefly has a striking tangibility and roundedness, pardy because of the rather literary and theatrical method whereby conflicting conceptions of Harry are articulated by different characters.
More subdy however, two contradictory qualities of his nature are expressed by two different methods that do not require explicit verbalization. The doubling with Holly lends Harry associations of light-heartedness and an essentially American optimism; conversely, a submerged connection to Conrad's Heart of Darkness, which is present from the outset, thickens the characters of both Holly and Harry and helps to weave around the unseen Harry the intimations of horror that will be fully redeemed at the children's hospital, where Calloway confronts Holly with the irrefutable consequences of Harry's diabolical trade in diluted penicillin. If Holly's first name is similar to Harry's, his surname, Martins, recalls that of Marlow, Conrad's embedded narrator who is similarly in search of a mysteriously disappeared figure, reports of whom occupy much of the first half of the text, and
Character in the screenplay text    2 I 3
whose persona becomes no less enigmatic with his tangible yet fleeting appearance towards the end. In Heart of Darkness the character Marlow pursues is Kurtz; in The Third Man another Kurtz, the phoney 'Baron', contacts Holly to begin the long process of deceiving him. In both texts, Kurtz is associated with a lie about the last words of a dying man: Conrad's Marlow cannot bring himself to tell Kurtz's fiancee the horrifying truth about Kurtz's last words, and tells her instead that he spoke her name; in The Third Man Kurtz similarly embroiders the fiction about Lime's death by telling Holly that 'Even at the end his thoughts were of you . . . he was anxious I should look after you' (35—6). The echo is unmistakable. Perhaps because screenplays are rarely considered as literature, perhaps because they are often adaptations of a single privileged precursor text, it is easy to forget that they can participate in intertextual worlds just as readily as novels. In The Third Man, the effect is the same as that produced by similar references in more 'literary' texts: the reader who spots the connection, even unconsciously, begins to extend the understanding of character beyond the limits of 'the words on the page'.
The Third Man also exploits various more direct modes of comment that Mamet eschews. Among the most conspicuous and unusual of these is that Greene's screenplay is prefaced by short descriptions of each of the main characters:
Harry Lime has always  found it possible to use his devoted friend   [Holly].  A light,   amusing, ruthless character,   he has always been able to find superficial excuses  for his own behaviour.  With wit and courage and immense geniality,   he has inspired devotion both in Rollo Martins and the girl Anna,  but he has never felt affection for anybody but himself. (7)
Although this is a thumbnail sketch, it hints at a backstory of sorts, and certainly predisposes the reader to conceive of Harry in certain ways. It might be objected that this kind of paratextual or supplementary material is not part of the screenplay proper, and that these qualities in Harry should be inferred from the screenplay itself: if they can be, the paratextual materials are redundant; if they cannot, there is a fault in the text. Alternatively, however, this kind of material may act as an interpretive guide for producers or actors: it does not necessarily retain for the writer a privileged interpretation of the character, but instead offers a concession to the collaborative nature of film. In any case, as we saw in House of Games, even the most purist approach to screenwriting finds it difficult to eliminate directions about character altogether.
That Harry's character is defined by storytelling is confirmed when he finally speaks: he is less an active agent within the story world than a commentator upon it. Aside from the game of cat-and-mouse in the shadows and the climactic sequence in the Viennese sewers, the only scene in which he makes a substantial appearance is one that does not advance the plot at all, and if anything represents a hiatus in the action. Like the giant Ferris wheel that is its location, it describes a circle,
214   Steven Price
beginning and ending with Holly's conviction of the depravity of Harry, whose own character similarly undergoes no change within the scene. Instead, what make the moment unforgettable are Harry's two speeches about the reduction of human life to dots seen from afar, and about peace and democracy in Switzerland producing nothing more significant than the cuckoo clock. As noted earlier, the routine affirmation in manuals that character is denned by a series of actions and decisions amounts to an ideological belief in the autonomy and freedom of the individual. Such notions certainly inform many contemporary Hollywood genres, among them Holly's beloved westerns. Yet this is quite at odds with the world-view of The Third Man, which instead dramatizes the helplessness of individuals in the face of post-war realpolitik. One of the reasons why Harry is so memorable, so much more than a cameo, is that he recognizes this and in the Ferris wheel scene expresses it with aphoristic clarity: 'In these days, old man, nobody thinks in terms of human beings. Governments don't, so why should we?' (98).
On the one hand, this speech represents conclusive proof of Harry's psychopathic unconcern for the human individual; on the other, much like the contemporaneous writing of George Orwell, Greene articulates a fear that Harry's understanding of power may be right. One of the ways in which The Third Man dramatizes this is by questioning whether the human individual any longer possesses the agency and potential that conventional notions of 'character' ascribe to it. So Holly is an essentially passive victim of Harry's plot; Anna delusionally refuses to act in accordance with what she now knows about Harry, and instead is condemned to life behind the Iron Curtain; Harry dies unheroically in the sewers of Vienna.
House of Games and The Third Man both demonstrate that the presentation of character in the screenplay text must be carefully distinguished from that in the dominant paradigm of the realist novel. At the same time, however, these screenplays indicate that diametrically opposed approaches to screenplay character are not just possible but inevitable, not least because of the demands of differing modes of film production. House of Games presents character as contained within a closed system, as an effect of structure; The Third Man is an open text, acknowledging some of the many ways in which character can appear to exceed this structure to produce something akin to, but distinct from, the reality effect of the novel. In each case, the interplay of character and structure remains a persistent dynamic. That two such radically different approaches have each produced a text that challenges the most familiar notions of screenplay character as autonomous, active and denned by choice is further evidence that the time for a properly critical examination of screenplay texts is long overdue.
Notes
1 Hunter 1994: 71; italics in the original. Hunter has previously discussed ideas for, and oudine plotting of, the story. As with Michael Hauge's arguments, the implication is that at least some aspects of character are separable from, or can exist independendy of, the story.
Character in the screenplay text    2 I 5
2 Mamet 1997: 9; italics in the original. Although expressed in the course of an argument against Stanislavskian notions of inner-directed acting in the theatre, this remark, and the following sentence ('There are only lines upon a page') are wholly consistent with Mamet's views on writing for both theatre and film.
3 The text of The Third Man used in this discussion (Greene 1988) was originally published by Lorrimer in 1973. For detailed discussion of the development of the screenplay see White 2003, and, in particular, Drazin 1999.
4 Not surprisingly this paradigm has proved influential both in Hollywood and in manuals, most prominendy in Vogler 1998 and also in Voytilla 1999.
5 Mamet 1988: 5. Subsequent page references are to this edn. To avoid confusion, I have not retained the italicization of the scene (non-dialogue) text that is used in the printed editions of both House of Games and The Third Man.
6 For a more detailed discussion of these aspects of the film, and the critical response to them, see Price 2009.
References
Campbell, Joseph (1949) 77k Hero with a Thousand Faces, Princeton, NJ: Princeton University Press.
Canby Vincent (1987) 'Mamet Makes a Debut with House of Games', New York Times (11 Oct., sec. 1): 94.
Carroll, Noel (1992) 'The Specificity Thesis', in Gerald Mast, Marshall Cohen and Leo
Braudy (eds), Film Theory and Criticism, 4th edn, Oxford: OUPress, 278-85. Drazin, Charles (1999) In Search of The Third Man, London: Methuen. Dyer, Richard (1998) Stars, rev. edn, London: BFI. Greene, Graham (1988) The Third Man, London: Faber. Hauge, Michael (1989) Writing Screenplays that Sell, London: Elm Tree. Hinson, Hal (1987) 'House of Games', Washington Post (19 Dec). Hunter, Lew (1994) Screenwriting, London: Robert Hale.
Kane, Leslie (ed.) (2001) David Mamet in Conversation, Ann Arbor, MI: University of Michigan Press.
Knights, L. C. (1933) How Many Children Had Lady Macbeth? An Essay in the Theory and Practice
of Shakespeare Criticism, Cambridge: Gordon Fraser. Lahrjohn (1997) 'Fortress Mamet', New Yorker (17 Nov.): 70-82.
Lucey Paul (1996) Story Sense: Writing Story and Scriptfor Feature Films and Television, New York: McGraw-Hill.
Mamet, David (1988) House of Games, London: Methuen.
— (1994) A Whore's Profession, London: Faber.
— (1997) True and False: Heresy and Common Sense for the Actor, New York: Random House. Maras, Steven (2009) Screenwriting: History, Theory and Practice, London: Wallflower. Peter, John (1987) Vladimir's Carrot, London: Deutsch.
Price, Steven (2009) 'Televisuality in the Films of David Mamet', in Johan Callens (ed.), Crossings: David Mamet's Work in Different Genres and Media, Newcastie: Cambridge Scholars, 33-48.
— (2010) The Screenplay: Authorship, Theory and Criticism, Basingstoke: Palgrave.
Propp, Vladimir (1984) Morphology of the Folktale, Austin, TX: University of Texas Press. Sternberg, Claudia (1997) Written for the Screen: The American Motion-Picture Screenplay as Text,
Tubingen: Stauffenburg. Thomson, David (1997) Rosebud: The Story of Orson Welles, London: Abacus.
21 6   Steven Price
Towne, Robert (1998) Chinatown and The Last Detail, London: Faber.
Vogler, Christopher (1998) The Writer's Journey: Mythic Structure for Storytellers and Screenwriters,
2nd edn, London: Pan. Voytilla, Stuart (1999) Myth and the Movies: Discovering the Mythic Structure of 50 Unforgettable
Films, Studio City, CA: Michael Wiese Productions. White, Rob (2003) The ThirdMan, London: BFI.
Wimsatt, W. K., and Beardsley, Monroe C. (1946) 'The Intentional Fallacy', Sewanee Review 54: 468-88.