I’ve been looking for this for quite a while, and of course the information was right under my nose in a book I’d bought a while back but never actually gotten around to even open: James Card’s idiosyncratic Seductive Cinema. Here it is, then, the killer example of someone, in American cinema, trying to have titles flashed over the image of characters talking in a silent film: The Chamber Mystery, 1920, directed by a guy from Pinsk. Here’s what imdb.com has to say about him:
Abraham S. Schomer
Date of Birth
2 August 1876, Pinsk, Russia
Date of Death
16 August 1946, Los Angeles, California, USA
Mini Biography
One time leader in the Jewish Congress, Schomer was a well-known Yiddish novelist and playright. A New York attorney specializing in immigration law, he gave up his law practice in 1915 to write for motion pictures.
I’ve been on the lookout for instances of words used over photographic images as you can see in my series on Words over Images, but this is the purest example of a narrative use I have found – the other uses were more in pseudo-documentary contexts or in comedy situations where the comic book subtext was more obvious.
Carr offers this image 
and adds that this was Schomer’s last film and the last time this experiment was tried. Was it ? I’d love to see other photograms of the film, and I’m happy to report that there is a three-minute excerpt of the film on DVD from Flicker Alley (Discovering Cinema, 2007) – along with some Caruso recordings that I’d very much like to get, also.
But was it the last time this was tried ? If so, why ? Is there some sort of ontological reason why written words could not be mixed with the visuals — an hypothesis that bodes well for the bifocal nature of Hollywood silent narratives : the text brings in another, often dissonant voice that never fully meshes with the visual flow and has always been described as a “problem” for silent film, even at the time, with the goal being that of the title-less silent film. That this ideal was rarely achieved in the 1920s (litterature always mentions the same two or three examples: Ol’ Swim Hole, Murnau’s Last Laugh…) is itself a tell-tale sign that this was an ideal but not a particularly important one from a concrete point of view of cinematic pleasure. 99.99% of silent films had titles, and not just because audiences (real or imagined by promotional campaigns) were “dumb”. I’d argue that titles, in their non-synchronic nature, because of the very nature of the juxtaposition and discontinuity they offered, were part and parcel of cinematic pleasure. Flashing titles on the image, from the point of view of narrative efficiency, is the smartest way to go: all the info one needs is given all at once. American cinema did not go that route, however, because narrative efficiency is not its be-all and end-all: the pleasure was (is) to experience the image and then to experience the often strong narrative voice that speaks through the titles, to feel that there’s a strong pull to bring the images back into a 19th century narrative fold, but that images always escape that fold, because they give out much more information and open themselves to all sorts of non-directed gazes.
In other words, the solution “titles on the image” was not long-lasting because it went agaist the fundamental heterogeneous pleasure of silent cinema, because it brought images down to a mere level of being a support for narrative information carried out through the titles. On their own, images were much more fascinating because they gave out (give out) much more than mere narrative information. In-between the images, the titles attempt to reimpose some sort of order on what promises to be an orgy of visual stimuli, with the audience granting, or not, the authority to do so to the written narrative voice – thus remaining on the edge, always, of narrative incoherence and madness. Just, you see, on the edge.
