LOCAL GHOSTS:
DUBBING BODIES IN EARLY SOUND CINEMA
...............................................................................................................................................................................

by Nataša Ďurovičová

I. Dubbing and its Relationship to the Multiple

I.1. Dubbing and Its Relationship to the Unconscious

Like E. A. Poe's purloined letter, dubbing both is and is not in the picture. Whether you see it depends on whether you hear it. And whether you hear it depends on where you stand while listening. Usually you can't stand in two places at once. Either dubbing is for you the exception that merely confirms your norm - that image and speech are congruent, redundant, wedded, indistinguishable. Or, it is dubbing that is the norm for you, and confirms your normal viewing situation as a deviant one - you see the film as it was meant to be seen, but hear as it was meant to be...well, what?
This ambivalent quality makes dubbing a prime topic for a conference on the multiple in cinema, since it raises questions regarding a text's identity for purposes at once formal, national and philological. It also provides a good occasion for reviewing one of the most firmly codified phenomenal conventions of classical narrative cinema - the synchronicity of the image and sound in speech, the inseparable voice/body weld - inviting us, in turn, to review the status, extent and true hold of that classical cinema in a global context. How normative can that weld be if most of the world's filmgoers - those in China, Germany, India, Italy and Japan most especially included - in fact (can) do without it?

I. 2. Dubbing and Crisis Historiography

Dubbing thus remains invisible to the history of cinema if and when it is assumed to be functionally equivalent, that is, paradigmatic to the original dialogue track, which is an absolutely dominant view. But such insistently "neutral" view of voice replacement privileges one aspect of speech, namely comprehension, at the expense of all of its other aspects - resonance, affect, drama, 'for-me-ness'. In the process such a view downplays - by assuming the spectator to ignore - all that undermines the fit, from the greater or lesser visual evidence of mismatch in lip-synch through the subtler perceptions of vocal quality to the most blatant yet by consensus also most invisible conspiracy, the linguistic and thereby cultural delirium inherent in the wholesale replacement of language. In sum, to take some easy examples, an American western populated by Hungarian speakers. The entire Hong Kong population speaking in English - but the Queen's, or Washington's? Russian peasants and their Napoleonic assaulters addressing each other in the same language - Italian. Scarlet shafted by Rhett in an all-Cantonese South. And so on. Such are, after all, the commonplace and basic cinematic experiences for at least as many people as not.
Of course, dubbing can be argued to be part of exhibition's hardware, on the order of posters or the height of the screen - well apparent but (in the absence of a code) not signifying. It can, on the other hand, also be compared to the impact the exhibition would have on the perception of a film if the films were always screened in a well-lit room, or to an audience lying down. In either case, however, it is indisputably an aspect of a film's event-ness, comparable, say, to a benshi's participation. But is it not also a textual feature?
To write dubbing, and through it a mark of difference, back into the picture of the classical narrative's historical reception, I propose to borrow the term functional near-equivalent from Rick Altman's revisionist "crisis historiography." Intending to challenge the core concept of functional equivalent so central to the functionalist historiography of the Classical Hollywood Cinema model, Altman proposes this modified term as a way of rendering (more) apparent the losses and substitutions that occur when moments of innovation (be they stylistic, procedural or technological) are folded into the large-scale industrial routines of standardized (film) production. Under criticism is the priority allotted by functionalism to stasis over change, the concurrent need to overemphasize the norm at the expense of deviation, indeed the general post hoc ergo propter hoc tendency inherent in the CHC paradigm. Altman's s revisionist counter-model advances a historiography of representational forms that, in innovating, are driven by ongoing institutional and procedural contestation rather than simply by recurrent 'transitional' phases. Some signals indicating such struggles are for instance contested jurisdiction (because the new elements entering the system are claimed by different social players/institutions), and evolving terminology (so that the same terms can apply to a variety of elements or phenomena, or the same phenomenon can be referred to by different terms) which indicate overlapping or competing ways of describing what in retrospect may be interpreted as "the same" procedure. Altman's overall goal is to replace the teleological trend of functionalism by a more contingency-sensitive logic of overdetermination. Dubbing, as I hope to show below, meets all these criteria.
For our purposes, then, instead of thinking about dubbing as functionally equivalent, it will here be shown to be functionally near-equivalent to the original dialogue track, as substituting almost but not quite the effect of synchronicity for a more arbitrarily constructed logic of fidelity. In tracing the research involved in the possible substitutability of voices, but alongside it also the small but chronic - in fact inevitable - deviation from a perfect lip-synch we will be better able to understand what was lost and what was found since linguistic non-identity has confronted the universal language of silent cinema from 1927 on.

I. 3. From Measurement to Code, From Synchronicity to Fidelity

What a near-perfectly lip-synchronized soundtrack hopes we will forget - and helps us too - is that sound film itself is based in an ontological non-identity. First of all, it is a mixed medium, a coordinate of two different types of waves - light and sound - each traveling at different frequency and speed. On the first level a vacuum tube or its equivalent is necessary to convert the latter into the former. On the next level a phase-converter is necessary to integrate the differential flicker of the film strip with continuous decoding of the optical soundtrack. In turn, this heterogeneous mix of recording materials (as was the case, albeit in differing technological manifestations, both for the sound-on-disk and sound-on-film methods) posed the problem of coordinating form: how to design a soundtrack that would correspond to the 'modular,' i.e. combinatory and interchangeable form, of the multishot film even while invoking the physical world's duration?
Film historians like Altman, Alberto Boschi, James Lastra, Murray Smith and Nancy Wood have carefully analyzed the steps that lead from basic technological compromises involved in synchronization to the working out of aesthetic conventions for a 'realist' acoustics conform to the 'invisible' narration of the classical film. The key idea here is synchronization not as a 'transcription' of simultaneity but rather as a process of matching the sound and the image track, with fidelity rather than identity as principal criterion. In other words, we find a perceptual protocol in which the spectator would complete an intended sound-image Gestalt, and thus agree to believe that the picture of a galloping horse and the sound of two coconut shells (to take the traditional example), or - more relevantly - a picture of hands moving on a piano keyboard plus the sound of some piano notes, or - more relevantly yet - of a face with a moving mouth plus the sound of a voice - are causally connected, that they are two homologue imprints of the same event, via two different sensory modalities.
It isn't of course that this process of conventionalizing was news. The assumption that it is possible to combine unrelated data into a Gestalt, a whole larger than the sum of its parts, had been fundamental for the silent cinema - for the medium as such, but in particular when its inner logic of modular interchangeability coincided with the external logic of mass production in a vertically integrated system.
This combinatory mode of replacement had of course been elaborated in the syntax of silent cinema, whether on the level of footage as such (the foreign negative frequently shot with a second camera near the position and angle of - but not entirely identical with - that of the 'domestic version'), between shots, as the Kuleshov effect principle (insert of a disfigured face, or of a body tied up in front of an oncoming train, or of a flag with the proper national credentials, or for that matter intertitles varying depending on either country of exhibition or even degree of sophistication of audiences, making the diegetic world almost but not exactly identical with itself); or on the level of the narrative, as the practice of multiple textual variants of one title, so that a common general sequence could contain widely differing inserts or excisions of shots or scenes, either for censorship or for export reasons (e.g. the 'Anna Karenina dodging the train and landing in Vronskij's arms after all' version for the provinces). Insofar as this practice of being non-identical - multiple, to return to our conference title - is taken to be a fundamental aspect of cinema, cinema can thus properly be termed a virtual medium even while its attraction/attractiveness is hinged from its powerful identity-based and identification-promoting machinery, i.e. its photographic basis.

II. Body/Voice Duplication

II.1. A Procedural Precedent: Body Doubling

But then the question arose: how could new or added body parts and functions - whether voice alone, or the ability to make unusual and novel noises (most likely music but also ventriloquism and in the last analysis, of course, speaking a 'foreign tongue') - be subsumed to this general combinatory principle or, as the generic term for the practice went, be "doubled"?
This was not a basic problem of either technology, or technique per se. Image modularity achievable across the cut, via editing or even via the line of framing could almost from the beginning be complemented by sound editing's early near-equivalent, re-recording. A 1927 WB sound-on-disk design could combine and mix sound through a series of up to 40 synchronized disk recorders to produce a composite yet even-leveled synthetic soundtrack. In fact, the term 'dubbing' which today means 'language replacement by another voice' was at first reserved for re-recording - either of the image or of the sound track - in the sense of 'copying.' The question was, rather, under what formal rules could the talkies inherit the silent film's protocol of interchangeability now that the working material consisted of two distinct and separate types of sensory material.
A series of complementary uncertainties emerged: is the relationship between a body and its voice qualitatively different from that between, say, a body and a leg? And from there, is the metaphysical conduit to the truth with which voice is imbued in most cultures, its privileged resonance with an inner "self," something that the new medium must take into account in its representation? Put otherwise, would sound film's diegesis - its implied world - have to be conceived differently now that both the inner and the outer contour of the character have been widened?
Initially the answer to the first question - from which the rest might follow - was "not necessarily." On the level of procedure, the new doubling therefore inherited the routines of the old doubling. In service of the needs of the casting department, and in direct proportion to the value of the actor's name to the studio, any part on the "public," i.e. credited body could be replaced by an anonymous, i.e. uncredited body - whether it was a pair of musically unskilled hands or a musically unskilled or socially unacceptable voice. Thus the discussion as to whether they indeed were Laura LaPlante's hands doing the banjo picking in Showboat or whether the length of Corinne Griffith's dramatic nails was compatible with that of the hands doing the harp strumming in Divine Lady could appear both in trades like Variety and in fan magazines like Photoplay. And as long as the actor emoted and lip-serviced his performance adequately, it was acceptable to replace - double - the Swedish-accented voice of well-established character actor Warner Oland on WB's payroll by that of an ad-hoc hired Hebrew-singing cantor, by standing the latter a few feet away, off-camera, on the set of The Jazz Singer. The same procedure would go for Louise Brooks' accented French substituted by Henriette Cremieux' impeccable ditto on the Paris set of Genina's 1929 Prix de Beauté, or Joan Barry standing in the wings for Anny Ondra in Hitchcock's Blackmail. The least equivocal instance of this concealed/open status of voice doubling comes from one of UFA's earliest sound shorts in January 1929: Ich küsse Ihre Hand, Madame consisted of the well-known comedian Harry Liedtke singing in the voice of the equally well known singing star Richard Tauber (alongside the as-yet-to be known Marlene Dietrich).
In sum, the new representational technology confronted the essential modernist ambivalence: should the spectacle be located principally in the now yet more perfectly transcribed (human) performance, or should the bravura turn rather be allotted to the yet more perfected accomplishment of the apparatus?

II. 2. Fear of Ghosting, or, Body Properties vs. Body as Property

So, with the technology for sound interchangeability available, and the conventions of voice fidelity in dynamic flux, why not accommodate the industry's needs and introduce wholesale voice doubling as needed and when efficient? This was, after all, the path Italy took only a few years later, and an option that was always available to filmmakers in Europe - and seized on, among others, by a Hitchcock, a Clair, a Duvivier, a Feyder. In the American conditions, however, there were several factors that counter-indicated a public highlighting and advancement of this synthetic or - to use John Belton's term - "constructivist" model of sound. In the domain of representation as well as in the domain of law, interchangeability between body and voice met with resistance from a variety of sources.
First of all there was all-important competing aesthetic of live-ness (life-like-ness), which (in aiming to displace the stage show live) had after all been the industry's original raison d'etre for synchronization. The audiences were thus explicitly invited to compare Jolson on screen with Jolson on stage. Second (as historians like Altman and Lastra have thoroughly documented, drawing especially on JSMP materials), there was the strenuously upheld aesthetic ideal of liveness, carried over into the studios' sound departments by the newly-hired engineers from the talkies' genealogical precedent, radio. Thus, even though the modular principle for voice replacement remained essential for the talkers' overall economic and institutional viability (the studios' point of view), its constitutive gap, the separation between the body and voice, had to be managed very carefully. And so, when the popular press by about mid-1929 began to paraphrase voice doubling, i.e. voice replacement, as "ghosting" (i.e. 'devoid of full, authentic, presence') or as "duping" (i.e. "duplicitous"), voice substitution within the same language became instead an excellently useful (off-) limit case. It could now instead be contrasted with renewed public claims to authenticity (in the body-driven discourse of the studios) and fidelity (in the measurement-driven discourse of the engineering guilds). "In response to public outcry, WB are prohibiting any further voice doubling or substitution. Henceforth when the WB players sing, it will be the real McCoy!" promised an ad on the front page of Variety. Or: "[v]oice faking," says George Lewin of AMPAS about voice doubling in 1931 "has now been completely abandoned." When Lon Chaney post-synchronized all his five parts in The Unholy Three with eccentric accents to match, MGM prominently featured a notary-signed certificate of their provenance from his own mouth as part of its publicity release. While the playback for musical numbers remained an important exception to this policy, assurances were given that a voice would never again be separated from the body that produces it. Twenty years later Singing in the Rain still staked its happy ending on that contract.
And indeed, despite its obvious advantages in freeing the camera especially for location shooting, and despite the first-rate post-synchronization achievable thanks to the Moviola in 1931, voice replacement (in the same language), whether by the performers themselves or by another voice, remained anathema in Hollywood. What modification of the 'corporeal acoustics' was needed would be accomplished on the music track, and especially per the Foley effects (that is, effects simulating physical presence by minute and immediate noises, such as breathing, the rustle of a silk skirt, feet on steps etc), at once allowing for the acoustic focus of close-miking, and for well-modulated space around the body inhabiting it. And here, too, is of course the ontological starting point for the massive and eccentric venture known as "versioning" - i.e. the attempt to simulate the world's sound cinema live, without voice replacement by either dubbing or subtitling.
But alongside the considerable technological development, not to mention the PR efforts, aimed at reinstating confidence in the aesthetic contract of identity between body and voice, and even after voice doubling was supposedly abandoned, the cinematic institution's economic need for interchangeability remained. And so, while the ads proclaimed the return of the Real McCoy, i.e. of an indexical relationship of voice to body, other -more properly extra-textual - guarantees were being put in place to uncouple and alter that relationship.
The question of how to give credits (and how to take them) for a voice separated from its body was one important wedge with which Equity, the New York-based stage actors' union, unsuccessfully tried to gain leverage in Hollywood between 1927 ands 1929. Actors on the set of Phantom of Opera, being reshot for sound in July of 1929, thus threatened to strike unless given such Equity contract guaranteeing their assent. After 1930 voice interchangeability, so insistently denied by the exhibition departments of the majors, was fully permissible in the production departments, with a new clause in the standard studio contract, allowing the producer the right "to dub or use a 'double' in lieu of the artist." Once Equity was out of the picture, there were few other challenges to this clause. The few that there were came from German and Latin American actors hired initially to star in the German and Spanish-language language versions being shot on the West Coast, who by late 1931 were being reassigned dubbing (in one case with particular vehemence when MGM required that the Mexican actors record Spanish-language dialogue for the dog roles in its musical shorts series called Dogway Melodies). There is every reason to believe, however, that it was this clause that a few years later served as a basis for Paramount's refusal to grant Marlene Dietrich the right to post-synchronize her own voice in German for the German release of Shanghai Express, and for Fox to do the same to Lilian Harvey on I Am Suzanne and the (ominously titled) My Lips Betray. It certainly opened the way for the habitual voice replacement of British English by American-English voices for many years to come, even up to the present.
Thus, despite the proclaimed abandonment of "ghosting," and the supposed return to the aesthetic norm of vocal identity, the standard contract clause meant that whatever a voice indexed was the interiority of a fictional character copyrighted to the producer rather than a tool of trade belonging to the professional employee - or for that matter a vocal manifestation of an autonomous subject. Indeed, to anticipate one of my concluding points, it is no coincidence that it was through their assertion of control over the dubbing process that a number European countries staked their effort to regain some measure of political control over their own national cinematographies.

III. Dubbing Procedures and Techniques

III. 1. A Breach in the (Acoustic) Mirror

So far we have been mostly looking at doubling, that is, at procedures of interchangeability when the voice that replaced another voice did so in the same language: Johnny Murray's competent for Richard Barthelmess' inferior singing; Joan Barry's good for Anny Ondra's bad English. A situation, in other words, where the identity (full equivalence) between all expressive strata was asserted, and remained unquestioned, as long as it could be confirmed by the coincidence of body and voice visually, by the convention of lip synch. The temporal simultaneity "shown" per the convention of lip synch then automatically buckled, in the process (mis-) confirming, all circuits of identity, whether temporal, spatial or cultural, between the image track, the sound track and the space of the auditorium. And so, as long as language was assumed to be at zero degree difference, that is, as simply present as fully lip-synched speech used among its normal speakers, then the film screen could become something like an acoustic mirror, reflecting in its always-homogenous diegesis a unified linguistic imaginary. An imaginary that would then reflect the spectator back to themselves as not only an omnipresent viewer-eye, but also as a 'world ear' as it were. Within its homogenized, equi-balanced and all-hearing sound space, all local differences could be subsumed and sublated, creating a sensory structure that would ideally correspond to the global possibilities of (the silent) film.
But cinema's Babel-like curse of language was also there, persisting in the gap of the mismatched lip movements, accents, speech patterns, all traces hinting that the trajectories of the speaking bodies may have been other than those proclaimed by diegesis. The gap drew attention to spatial difference in the form of other fault-lines and borders, and caused cracks to erupt in the acoustic mirror's global imaginary. How that return of difference was going to play itself out technologically, representationally and politically, in other words, whether and how Babel could be overcome, was by 1928 the pressing question for just about every film producer in the world.

III. 2. Three Speech Functions: a Typological Sketch

The language replacement experiments that began proliferating almost immediately, in Italy, Germany, the US and France but also in Japan and India, all issued from the same dilemma: what is the absolutely minimal (i.e. least expensive) modification necessary to detach a film from its place of origin in order to make it again potentially infinitely mobile and infinitely global? What, in other words, is the minimal necessary aspect of language in cinema? And which aspects can be dispersed with? What we can see in the 1928-1933 international dubbing rush is the effort to answer this question by an unruly working out of a paradigmatic of sorts.
In all of the many and varied experiments (some of which we shall next lay out), three functions of spoken language were taken into account; and the differences among the translation strategies reflected the differing importance attached to each of these three. Once the minimal requirement of overall acoustics had been satisfied (i.e. the question of how the entire soundtrack, whether voices, noises or music, would be handled acoustically - especially with regard to clarity of recording and reproduction), speech itself posed problems in the following three areas:
a) The semantic dimension - comprehensibility of the words' content, their communicative function. An exemplary question here would be "should 'native' dialogue spoken in exotic travelogues be translated, or should it be left as exotic sound?" But squarely in this category belonged concerns with translation, paraphrasing, and the phenomenon of dubbing as a technique of censorship. An extreme (and later) case in point being the ardent anti-dubber Jean Renoir who was known to shout out the correct dialogue translation from the back of screening rooms if he found the dubbed version to be not to his liking.
b) The expressive dimension - voice's "inner" tie to the body, its indexical property. The anticipatory excitement encapusuled in the "Garbo Speaks!" moment or the kind of problems arising in dubbing for instance a Welles film, where a character's vocal signature would be among its determining traits, illustrate this aspect most obviously. Voice texture, accents, speech mannerisms. How should a distinction between a British and an American, or a New York and a Texan accent be dubbed?
c) The phatic dimension - (as in "em-phatic"); a speaking body's reverberation in the social sphere, its 'placing' function, its 'for-me-ness." Designates the sphere of audience inclusion, within which there is a continuity of under-standing in the broadest sense. A function that is normally (intended to be) transparent, at "zero degree," but which, like contrast dye, becomes apparent in any "crisis of address." Defined negatively, the phatic zone is what was preserved in the FLVs, and what disappears in dubbing. Consider Las Hurdes whose pompous American voice-over (written by Bunuel while he was working as a dubbing supervisor for Warner Brothers in Madrid) is what spins the film's address into its completely fantastic social continuum: dubbed into French the film's address would be at best oblique. A fully bi-lingual film like Kameradschaft (or later Godard's polyphonic/polylingual Le Mépris) has invested all its drama in its phatic dimension, which indeed vanishes in any dubbed variant (and nearly to an equal degree in subtitling, though there a change in typeface could hypothetically mark the distinction between the German and the French characters' dialogue lines). René Clair, in turn, in his attempts to avoid dubbing and to create a technique for a linguistically interchangeable phatic zone included in his early 1930s films (e.g. Le Million) minor additional characters wandering around in the diegesis and conversationally summarizing the dialogue and action in the language of the audience.
With this typology on hand we should now see more clearly which aspects of language in cinema were perceived as being at stake, and which were considered expendable. Put another way, at what cost and through which processes was dialogue's role downgraded from the "(full) equivalent" of a new direct-shot version (achieved in the fully reshot but therefore prohibitively expensive and star-weakened FLVs) to the "near-equivalent" of a dub?
Again it is worth underscoring that while the question might seem at once miniature, and at best archaeological, its reverberations established precedents of conventions shaping the larger institutions until quite recently. For it was on the assumption that a star's body must always resonate with its authentic voice that Hollywood from the mid-Thirties on established the firm dual standard of exporting its films to be dubbed (abroad) even while insisting on the inevitability of remaking films from abroad: the endlessly reiterated claim that an American spectator couldn't, wouldn't, accept a dubbed (not to mention subtitled) film was always the rhetorical device used by the Hollywood majors (or their front, the MPPDA, today MPA) for quarantining their domestic exhibition territory without having to accept charges of protectionism and unfair trade practices. Thinking about the relationship between the original and the dubbed dialogue as "functionally equivalent " thus directly obscures the tourniquet function that a particular stylistic element (the fully lip-synched speech) has been assigned to police an economic terrain.

III. 3. Some Early Dubbing Procedures

It is in looking at the earliest practical experiments with language replacement - precisely coincident with same-language voice replacement practices - that the differing, as yet quite fluid "dimensions" of comprehension, expressivity and address make themselves noticed. The cases below - that of Roy Pomeroy at Paramount, of Friedrich Zelnick for UA, the Vivigraph method of Edwin Hopkins, and the Rhytmograph method developed in Germany - are to my best knowledge among the very first procedures working out the conventions of dubbing, certainly among those described with at least some degree of detail.
After having triumphantly synchronized the complex sound effects for Paramount's road-show of Wings, and directed the studio's very first sound film Interference, Roy Pomeroy, its resident sound expert, was given the assignment to sonorize the studio's major imported star Emil Jannings. This involved, on the one hand, solving the problem of Jannings' heavily accented English for his enthusiastic American fans, and, on the other, rendering him generally comprehensible in the many European countries where he was a huge box office draw for Paramount. "Avoid[ing] duplication" by the "off-screen lip synch" of the Jazz Singer variety for patent reasons (which could have been possible with the disk recording method), Pomeroy seems to have intended to record Jannings' - no doubt sparse - dialogue in 5 languages by other voices, and then coordinate each with the image strip in the form of something referred to as "foreign tongue overlay." The particulars are sketchy. But it seems as if Pomeroy's plan was to cover up any visible deficiencies of lip synchronization in post-production (i.e. without making the actor reenact his lines during the production phase by mouthing them in another language) by altering the scale of the actor's body with the help of a zoom-lens like addition to the printer, presumably making him appear further away. This was apparently then to be further supplemented by transparencies, perhaps designed to fill in shot sequences with inserts in which Jannings' body is superimposed onto suitable backdrops.
Here all that matters is that the dubbed words get audibly communicated to a mono-lingual audience; attention to the semantic function seems to be absolute while the phatic and expressive function of Jannings' voice (soon to be put to such brilliant use by von Sternberg in Der blaue Engel ) seem to be of no importance. But particularly intriguing here is that voice replacement for Pomeroy is not really approached as a sound problem, related to acoustics and executed in recording under the sign of transcription, but rather as related to special effects, i.e. to simulation. Rather, sound here is approached as an element in the construction of a virtual space. His subsequent work for RKO alongside the innovative cameraman Leo Planskoy, probably on Son of Kong, points both to sound's role in the creation of simulacra, and to this role's logical extension toward the creation of synthetic characters (precisely of the King Kong variety) where sound is used to ground and thereby "confirm" the appearance of physical reality in the monstrous. It may even be argued that Pomeroy's approach is an avatar of today's development of dubbing (accomplished with the help of digital morphing) which has moved one step closer to dispensing with the limitations of the human corporeal shell, hyle , even while evoking it via synchronized voice. Still capitalizing on the ontological realism of "recording," the digitalized image track and the sound track here nonetheless edge one step closer to their joint origin as electronic signals.
My second example comes from an equally eccentric case, namely Friedrich Zelnick's dubbing work on Herbert Brennon's 1929 The Lummox (under the German title Der Tolpatsch). Here the veteran German director (who had earlier that year supervised the "ghosting " of Liedtke by Tauber in Berlin) gave absolute priority to the expressive aspect of language, affording maximal respect to the actors. No doubt driven by the key economic imperative of eliminating any "doubling" i.e. repetition of performances to facilitate lip-synching, in particular in close-ups, Zelnick left the image track intact. With absolute priority allotted the lip movement, he then rewrote the German dialogue so as to conform, as much as possible, with the English lip movements, regardless of semantic consequences (presumably thus the English "hat" as the German "hat" (=[s/he] has). Judging from the press (as well as from Zelnick's subsequent interviews), the catastrophic reception of the film was directly related to the near-nonsense of the German dialogue so produced. Another way of putting this would be that language was here treated more as expressive than as communicative, more akin to, say, an echo than as speech. However, the basic premise of such "squeezing" of the translated words into the pre-shaped mouth of the original speaker on screen is of course au fond not different from the basic convention of dubbing still in use today. Any semantic "truth" of the dialogue still remains fundamentally "captive" to its visible evidence, its sound shape. So, though Zelnick's method fell short because it granted words all too much arbitrariness (as if taking seriously the point that sound/image relationship is one of merely conventional fidelity), its ruthlessly mechanistic innovation lies precisely in treating language as no more and no less than a category of noise, untrammeled by inwardness, projection or any ruse of 'character.' As we'll see, the German Rhytmograph method took off precisely from this assumption to become the dominant dubbing procedure.
Though the Vivigraph method of "revocalization" seems to have remained (like the Pomeroy experiment) only a hypothetical proposition described at length in mid-1928 in JSMPE by its author Edwin Hopkins, it must have been known well enough to be referred to in trade press for several subsequent years. And just as was the case for Pomeroy's and Zelnick's trials, some features of his prototype contributed toward an analytic splitting up of the language function in the service of the efforts of the Hollywood industry's subsequent abandoning of the "phatically overburdened" FLVs. Hopkins' ideas came to be take up especially at MGM, the most dubbing-reluctant AND the most dubbing-dependent of all the US majors, given its dependence on its star lineup.
Hopkins' schema presupposed an experimental recording technology consisting of a cylindrical phonograph record, 'built up of thin laminations' and thus, in contrast to a disk, capable of being edited minutely for optimum synchronicity. But in addition to relying on this flexible storage medium's manipulation to achieve synchronicity, the "Vivigraph" method's recording process also called for a complex reenactment scenario complementing it. In Hopkins' schema, footage would first be shot of the original actors performing the action repeatedly even while mouthing the foreign dialogue "mechanically," in any number of required languages, without necessarily being even minimally competent speakers of that language. This same action would then be reenacted "at least 20 times" by the dubbing crews, dressed in full costume and on the same sets as the original performance so as to perfect - thanks to the repeated reenactments - at once the expressiveness and the routine, i.e. pacing, of the replacement dialogue. And to truly guarantee the "liveness" of their dubbers' performance, a paying audience could be admitted on the set (covering thereby also some of the costs of the dubbing procedure!). Clearly, in this prototype the semantic function - the comprehensibility of the dialogue - is understood as inseparable from the expressive function, which in turn is underwritten by a phatic contract. It is the presence of a genuine audience on the set that turns the repetition /rehearsal/ mechanical routine of duplicated recitation into a meaningful, therefore expressive, and thereby comprehensible, dialogue delivery...We have here in fact completely abandoned the modular principle of cinema and returned the full circle, back to the stage and the auditorium of legitimate theatre - albeit reformulated as a serial arrangement of comprehension, affect and address - in order to fully honor the phatic dimension of the spoken voice.
It was thus indeed the nexus of aesthetic assumptions behind the Vivigraph idea - the central role of the voice's expressive function, corporeal presence as necessarily underwriting voice, mechanical reproduction as merely transcribing a human(ized) routine - which led to the generalized conclusion, in the second half of 1929, that dubbing simply was not a viable large-scale procedure for film industry. Ergo, it was then the FLVs, with their full reenactment procedure, that appeared to be the as it were regrettable but inescapable course of sound cinema. As Geoffrey Shurlock, the head of the Paramount Foreign Department, said at the beginning of 1930, in justifying the studio's planned move to Joinville: "Who knows how important it is that the whole crew - from director and actors through the technicians to the last gaffer boy - speak the same language?" It is easy to say (as has been the film-historical routine) that this was an "experiment bound to fail" because of the undertaking's expense. The question remains, nonetheless, why Shurlock would even think to ask such a question, which today seems so quaint and ephemeral. In other words, why exactly the versioning was abandoned in some cases (like Joinville) but not in others (like the MGM Chevalier avatars into the mid-1930s, not to mention the now so well documented Franco-German FLV collaboration until the eve of la drôle de guerre). What was it that had changed in the Gestalt of body, voice, and space that made dubbing the either tolerated or mandated (or both) standard in Europe by mid-1930s, even while remaining firmly rejected in the US?
We have until now been looking at the technological leap year 1928/9 with special attention to Hollywood's pioneering role in research, design and wiring. But the problem of voice replacement in general, and of language replacement in particular was, if anything, even more apparent to film producers in Europe. Their very recent joint advances in the inter-national Film-Europe movement in late 1920s had, after all, begun to provide a glimmer of hope for a joint - multinational - resistance to the American dominance. Some of the technical and technological procedures were similar on both continents (we have already mentioned same-language voice replacements for Anny Ondra, Louise Brooks and Harry Liedtke). But one particular procedural paradigm, mainly known under the trademark name 'Rhytmograph method' (as patented to the sound equipment manufacturers Lignose-Brausing) handled speech distinctly enough to advance the viability of dubbing over and against the US studios' procedures.
The Rhytmograph method - "a mechanical guide method" (in counterdistinction to, say, Hopkins' fundamentally human-dependent analytic "guiding" via routinization) was a three-step process. The first was the detection of phonetic components of the original version, which were then electromechanically transcribed in the form of a graph (like a cardiogram, or in fact like the sound-on-film method). This abstract graph was then transcribed on paper, a la musical notation, with syllables instead of notes. This 'analytic' phase was then followed by a 'synthetic' moment in which a parsing mechanism (whether a disk, a telegraph-type strip, or a graph projected onto an adjacent screen, depending on the patent) paced and shaped the sounds for the dubber reading the printout aloud. Thus the dubber would sit, watching at once the original version's image and reading the portioned-out translation, projected for her onto another surface.
Where Hopkins' synchronization scheme had depended on the living actors' duplicate performance, their memory and their ability to emulate, the German approach mechanized that process. Comparably with Zelnick's wholesale homophonic "echoing," it first separated the semantic from the phonetic aspects of speech, and then locked them into a new whole by pacing the "Nachsprechende" (the "after-speaker") in a line delivery entirely separated from any expressive, and certainly any phatic dimension, more akin to conveyor belt assembly of words than to a felt performance.
Some variant of the Rhytmograph method was therefore probably the innovative technique that began to be used at the Paramount studios in Joinville around 1931, introduced by its German division head Jacob Karol, to give it a competitive edge against MGM's reenactment procedure. While Joinville, in an effort to find the right ratio for its productions between repetition and variation had frantically been varying strategies for their FLVs output, MGM (whose main effort was concentrated on staying put on its Glendale lot while establishing the widest possible global recognition for its new roster of sound stars) was in the meanwhile trying to find a compromise solution with the Vivigraph procedure. Starting in early 1931, MGM's experimental dubbing work thus involved reshooting key actors while they were mouthing the versioned language, with their designated dubbers (often, as has been noted, former FLV actors hired just a few months earlier to perform such roles "in their own person") then emulating the native stars' performance by watching them repeatedly on-screen. Still, this did not solve the problem of the repeat shooting, i.e. of the infinitely multiplied performance (not to mention the problem of needing to find in an ever-growing number the individualized dubbers selected for their expressive features) matching optimally each star's designated qualities. . It would instead take the mechanized abstraction of the Rhytmograph method - which phased out expressivity in the performative moment altogether - to reach the near-industrial precision and efficiency of work at Joinville. The first film to have been praised in the otherwise extremely hostile French press for the quality of its dubbing (in itself, of course, a paradox, that is a proof of the non-equivalency between the dubbed version and the original) was Karol's dubbing of Derelict (Rowland Lee, 1930), released in France as Desemparé.

III. 4. Dubbing and the Politics of Language

This is by no means to say that dubbing was thereby somehow a solved problem. Rather, the moment of technological compromise (when its practice achieved a near-mechanical precision of synchronization without having to depend on the gearbox of proximity and fine-tuning between the dubbed and the dubbing performer) led immediately to a next set of jurisdictional battles. These concerned on one hand a proper identification of a dubbed film over and against other kinds of "textual variants," and on the other rules for attribution of a dubbed film's national identity, or rather its national provenance.
Once the production of the dialogue track could be conceived as divisible between two different speakers, the conflict became instead redefined as a battle over a nation's acoustic fly-over zone, that is, over the right to admit films speaking another language and thus 'traveling on a stolen passport. If in the earliest sound period (let's say until the end of 1930) the term "version" pretty consistently referred to a film re-shot in its entirety, that is, a F (oreign) L(anguage) V(ersion), by the end of 1931 the same term can be found referring to a film's 'original version' that was dubbed. This shift in terminology - one of the prime symptoms of crisis in Altman's historiographical method - wasn't merely trailing a change in practices. The gap in the definition of was "equivalent" and what was "near-equivalent" but definitively not "equivalent" could be put to good use. In 1933 German exhibitors thus protest vehemently against being pre-sold, in the framework of the prevalent American block-booking practice, "versions" which they claim to have understood to be FLVs proper (that is, films remade with German actors and German dialogue), complaining that the films they have received are, instead, the original American films with a dubbed soundtrack grafted onto the Hollywood actors' lips, much less popular with the German audiences. A claim was even made that this tactic of terminological confusing was a deliberate move on the part of the US members of the German umbrella film trade organization SPIO.
Even while lip-synching was by mid-1932 becoming perfected enough to meet the conventions of fidelity, uneasiness about this "deceptive" practice remained. Most states in Europe made sure that additional markers were put in place so that the substitution between the original and the national dialogue track wouldn't be completely invisible, in other words, that some trace of the linguistic superimposition could always be detected. The legal and political practices of this marking varied considerably. The gamut went from a complete "hijacking" of dubbing as a kind of permanent domestic supplement to each imported film, and a source of considerable revenue and employment (as well as a linguistic filter, at once homogenizing all local dialects and blocking out the lure of 'foreign sounds'), as was the case in Italy to, say the situation in Sweden. There dubbing was never a real option (probably because of the smallness of the domestic market) and so it was the band of subtitles that provided the permanent, highly visible and most reliable tag of an import.
France stands as the most complex, but also the precedent-setting, case. In light of widespread attacks on dubbing as "theft of national identity" by prominent figures in the film-critical establishment, the actors' unions and a portion of trade press (in particular that which stood to gain by cooperation with the FLV-oriented German producers), all of which were also lightly veiled manifestations of the more generalized French anti-Americanism, legislation was put in place to vigorously upkeep the distinction between the FLVs and the dubs. From 1932 on, all dubbing had to take place on French territory, the dubbers be given full credits, and a limit of 12 theatres for all France was established where non-dubbed films (versions originales) could be shown. As Martine Danan has proposed, a two-tier system as established: the original sound track for the urban cosmopolites and the art and essai theatres, a "nationalized," dubbed dialogue track for the provincial mass public. The French cultural institution thus literally arranged to 'take credit' for, and so share in a portion of Hollywood's success. And it as thus building on France's trend-setting insistence on the non-identity of the two textual variants - on their status as non-interchangeable i.e. as significant variants - that the first Venice film festival started requiring, in 1936, that all submissions be the original and not the dubbed version. Behind both countries' policies lay a decided effort to underscore the nationalness of any give cinema - that of Hollywood included - over and against (the fear of) the medium's Americanization under the guise of global reach.
In the US, on the other extreme end of the spectrum, dubbing of imported films (not to even mention subtitling) was continuously said to be unacceptable to the American spectators by the same majors whose balanced books depended, to a greater or lesser extent, exactly on the widest possible spread of that practice abroad. At the close of the transitional period of intense experimentation we have just surveyed, and resulting by late 1933 in the closing of MGM's West Coast "re-vocalization" projects, dubbing work became completely separated from other production: "(F)rom now on all dubbing will be done by local ghosts," Variety declared. Perhaps because neither the specter of ghosting, nor the latent anti-realist modernism of the montage effect underlying dubbing, were stances compatible with the 'high-fidelity' aesthetic of the classical Hollywood film, the practice was best uncoupled from production and left to the gray(er) zones of distribution and exhibition. Preferably abroad; if not, then at least in New York, far away from Hollywood's main lots.
In surveying the situation as a whole, it is clear that this account of repossession of language, made mobile by technology and then secured anew in a symbolically proper site by essentially political fiat (New York for the Americans, Paris for the French, Rome for the Italians, etc), is not a linear narrative of cultural imperialism, nor of resurgent nationalism, but rather - to return to the terms our historiographical revision - an overdetermined convergence of both. Its outcome as presented here points to a compromise solution in the form of a period of mutual accommodation - a precarious compromise ready to be tipped as soon as either a new technology or a new political configuration arose.

IV. Summary

Dubbing is - and is bound to remain - an ontological crisis for cinema, marking as it does the fault line between the medium's absolute mobility (as image, as goods) and its attachment to residual representational forms - in particular language, with its own powerful agenda and resistant history. This is not a new point - Arnheim called the talking film an "artistic composite" - but one that film historiography yet has to fully address, even though this built-in representational heterogeneity has widely influenced the global cultural landscape. It is, after all, on the strength of a supposed impossibility to show a dubbed film to a US audience that the vertically integrated US majors have - intentionally, I would venture - kept out the non-American film off of American screens, big and little, from the early 1930s on. The history of remakes follows to a considerable degree from this claim of the intractable and unique resistance to dubbing on the part of the American audiences - a resistance which, if it ever existed, was not tolerated for any of the world's audiences of the Hollywood product. Like the pivot bar of a tourniquet, dubbing has the capacity to regulate the flow of images, itself functioning either as a barrier or a conduit for such flow. The absolute one-way street historically charted by the Hollywood institution and its products has locked the imperial United States into the constricted space of a cultural province.

In: Anna Antonini (ed.): Il film e I suoi multipli/Film and Its Multiples.
IX Convegno Internazionale di Studi sul Cinema (Forum: Udine, 2003)

NOTES

1. Three exceptions, all bearing on European cinemas, must be noted. Martine Danan's dissertation From Nationalism to Globalization: France's Challenges to Hollywood's Hegemony (PhD thesis, Michigan Technological University, 1994) has studied in detail the politics of dubbing in France, especially from the vantage point of a contest between exhibitors and the cultural elites. Charles O'Brien has recently written about technology and style of dubbing practices in France, advancing the hypothesis that dubbed films might represent a distinct aesthetic category. Charles O'Brien, "'Un individu brisé?": The Dubbed Actor in 1930s France," L'uomo visibile/The visible man (Udine: Universita degli Studi di Udine, 2001). The one synoptic film history touching the stylistic and institutional impact of dubbing is Maurice Bardeche and Robert Brasillach, Histoire du cinéma (Givors: A. Martel, 1953-1954).
2. As formulated for instance in Rick Altman, "Penser l'histoire (du cinéma) autrement: un modele de crise," Vingtiéme siécle 46 (1995), 65-74.
3An instance of the legal application of this principle of non-identity was the UK situation in the 1990ies. To bypass the law prohibiting direct reproduction of terrorist statements the then-outlawed Sinn Fein leader Gerry Adams could be shown moving his own lips while his voice was 'doubled' by that of some more legal stand-in, presumably that of some BBC employee. Sight and Sound, 3/94, 5.
4. On the consequences of that distinction see Edward Branigan, "Sound and Epistemology in Film," The Journal of Aesthetics and Art Criticism, 47:4 (Fall 1989), 311-324.
5. Rick Altman, "The Technology of the Voice" parts I and II, Iris II/1 and 2 (1985); Alberto Boschi, L'avvento del sonoro in Europa: Teoria e prassi del cinema negli anni della transizione (Bologna: CLUEB, 1994); James Lastra, Sound Technology and the American Cinema: Perception, Representation, Modernity (New York: Columbia U P, 2000), Murray Smith, "Technological Determination, Aesthetic Resistance, or, Cottage on Dartmoor, Goat Gland or Masterpiece? Wide Angle, 12/3 (July 1990), and Nancy Wood, "Towards a Semiotics of the Transition to Sound: Spatial and Temporal Codes," Screen 25/3 (May-June 1989).
6. Photoplay, July 1929, 32-33. One historian attentive to this function of doubling is Donald Crafton in his The Talkies: American Cinema's Transition to Sound 1926-1931 (Berkeley: U C Press, 1997), 509-515. Crafton, however, attributes the abandonment of voice replacement in Hollywood exclusively to the influence of the fans' preference for authenticity.
7. It is as if "fidelity" is here tenuously "implied" through the temporally simultaneous action of singing of the two
performers, one silently one out loud, near-identical but for the slight spatial separation of a few feet. Henriette Cremieux in personal interview with Dudley Andrew, Cassis, May 1998. Tom Ryall, Alfred Hitchcock and the British Cinema (Urbana, Il.: U of Illinois P, 1986), 29. Klaus Kreimeier, Die UFA-Story: Geschichte eines Filmkonzerns (Műnchen: Hanser , 1992), 216.
8. Films highlighting the multimedia- and apparatus-driven aspect of talkies by amplifying the gap between the body and the voice as a source of amusement and pleasure complemented the mainstream synchronizeds: most important among them was animation in general. Indeed animated cinema's total compatibility with dubbing - which was simply one special case of the generalized rule of post-synchronization - made it, and Mickey Mouse in particular, a key engine of the international wiring of theatres.
9. John Belton, "Technology and Aesthetic of Film Sound," 70-71, in Elizabeth Weiss and John Belton, eds., Film Sound (New York: Columbia U P, 1985).
10. Variety, 31 July 1929, 1.
11. George Lewin, "Dubbing and Its Relation to Sound Picture Production," JSMPE 16/1 (January 1931), 38-9.
12. Thus for instance sections of Joseph Korngold's scores were composed with the actor's breathing as underlying rhythmic patterns. David Helvering, "Korngold's Early Approach to the Synchronization of Music with Narrative," presentation, University of Iowa Sound Seminar, Spring 2000.
13. For additional background on the foreign language versions ("FLVs") see for instance Ginette Vincendeau, "Hollywood Babel," Screen 29/3 (Spring 1988), Nataša Ďurovičová, "Translating America: the Hollywood Multilinguals 1929-1933," in Rick Altman, ed: Sound Theory/Sound Practice (New York: Routledge, 1992); Joseph Garncarz, "Die bedrohte Internationalität des Films," in Sibylle Sturm and Arthur Wohlgemut eds., Hallo?Berlin? Hier spricht Paris!:Deutsch-französische Filmbeziehungen 1918-1939 (München: text + kritik, 1996).
14. Variety 10 July 1929, 7.
15. Variety June 30 1931, 11.
16. On Dietrich being disallowed to dub herself into German see Variety, 10 May 1932, 48. On a possible lawsuit between Harvey and Fox regarding the British/German star's right to dub herself to German see Variety 9 January 1934, 11.
17. Thus yelling "sale Juif!" from the back of the Melnitz Theatre at UCLA at a screening of Grand Illusion some time in the early 1950s, to fill in the full rendition of an anti-Semitic remark of Maréchal (Gabin) to Rosenthal (Dalio) when he noticed that it had been dubbed as a milder and less offensive insult on the 'normalized' American English soundtrack. Personal communication from Alexander Sesonske to author.
18. H.G.Weinberg, "Foreign Language Film in the US," Close Up X/2 (June 1933), 173-4. Variety (9/22/31, 14) claims that Paramount used this "Le Million method" for some releases of Morocco. On the Italian release of Morocco see Mario Quargnolo, La parola ripudiata (Gemona: La cineteca del Friuli, 1986), 34.
19. Probably for Sins of the Fathers, guest-directed by Ludwig Berger on loan from UFA. Variety 4 January 1928, 13; 7: March 1928, 10; 16 May 1928, 45.
20. Variety 10 October 1928, 4. Also Farciot Edouard, "Economic Advantages in Process Photography," Bulletin of the AMPAS , Supplement # 9 (20 July, 1932), n/p.
21. There Jannings - as if prophylactically countering rumors of his linguistic failure in the US - was cast as a professor of English, torturing his students by drilling them endlessly, though not very successfully, in the pronunciation of the word "the."
22. There is an intriguing trail from Pomeroy's firing in fall 1929 from Paramount (for "insubordination") first to his subsequent collaboration with the special effects expert Carroll Dunning on a project of FLVs achieved entirely "synthetically," via rear view projections, and then, alongside Dunning and the dubbing maven Friedrich Zelnick, on an "entirely synthetic" version of Beau Ideal for RKO.
23. In German contexts his name usually retains its original spelling 'Zelnik'.
24. Interview, Lichtbildbühne 19 November 1929, 4.
25. Edwin Hopkins, "Re-vocalized Films," Journal of the Society of Motion Picture Engineers (JSMPE) 12 (1928), 845-852. Hopkins' professional background was likely phonography.
26. A storage medium possibly related to the Stille-Blattner magnetic strip just coming into use in Britain. Joe May and Erich Pommer may have been aiming to use this in their early dubbing experiments. See Hans-Michael Bock , "Ein Instinkt- und Zahlenmensch," in Hans Michael Bock and Claudia Lenssen, Joe May: Regisseur und Produzent (München: text+kritik, 1991), 140-141.
27. Geoffrey Shurlock, "Versions," American Cinematographer 11/9 (1931), 10.
28. W.A. Pozner, "Synchronization Technique," JSMPE 47/3 (September 1946) distinguishes between the German-model "mechanical" and the American "visual synchronization method " model (which was, inversely, referred to in German literature as the "nach- Schnauze-Methode," roughly "seat-of- the-pants-method").
29. A full description of the Rhytmograph method is for instance Paul Hatschek, "Die Rhytmographie," Filmtechnik/Filmkunst, 7/7 (7 Februar 1931), 6-8.
30. Thus both Crawford or Shearer had a personal dubber into French as well as German, much like Shearer's role was played by Huguette Duflos in Le proces Mary Dugan, by Nora Gregor in in Der Mordprozeß Mary Dugan and Maria Guerrera Ladron in El proceso Mary Dugan in the French, German and Spanish versions of The Trial of Mary Dugan. Crawford was said, however, to be administered intense French lesons by MGM in the hopes that she could eventually play herself in the French versions despite a possible residual accent.
31. Variety 29 April 1931, 33. Pour Vous 7 May 1931, 2 and La revue du cinéma 3/23 (Juin 1931), 62, didn't quite share the enthusiasm of the American trades.
32. Variety, 24 January 1933, 15. For a while the shifts in terminology became in fact quite chaotic, with FLVs renamed "directs" or "direct-shots" or "straight-shots" in retro-contrast to the "dubbed ," or sometimes "synced" "versions." It seems almost as if in the American trade usage the word "dub" at this point still dragged the negative charge of "duping." The German trades themselves meanwhile distinguished between an "Orginal-Fassung" (an original version), an "akustische Version" (a dubbed version with original lip-movement and substituted dialogue),) and an "optische Version" (a version made by reshooting the same performer, Vivigraph style, and then dubbing to match the reenacted lip movement). N/a, "Versions-Terminologie", Filmtechnik/Filmkunst, (28 November 1931), 20.
33. Again, however, a straightforward market, that is, economist argument doesn't carry sufficient explanatory power. Though the "Latin American market" (as the US Department of Commerce would designate it) was enormous, and on the face of it best served by either direct-shot FLVs or dubs (in part because of the high illiteracy rates), the variety of dialects of Spanish, combined with weak local cultural policies, led Hollywood to export its films in the cheaper, subtitled versions. This, however, became a problem for instance in Argentina when the German and Italian films, with their more immigrant-friendly dialogue track, became more active competitors in the politically confrontational late 1930s.
34. Danan, op. cit., Ch.II.
35. The term "significant variant" is elaborated in Josef Garncarz, Filmfassungen: Zu einer Theorie signifikanter Filmvariation (Frankfurt a/Main: Peter Lang, 1992). On the contradictory aspects of cultural nationalism vis a vis the universalizing claims of American movies see for instance Victoria de Grazia, "Mass Culture and Sovereignty: The American Challenge to European Cinemas, 1920-1960. Journal of Modern History, 61/1 (March 1989), 53-87.
36. Variety (7 June 1932), 11.
37. Dubbing (was) thus returned to the Hollywood sound facilities at the very end of WWII, when the Office of War Information needed to shore up the advancing Allied front with a steady dose of American films, dubbed to suit the local needs.