From August 25th 2002 to February 22nd 2003 I have been at the Humanities Information Technologies Centre (Hit Centre) undertaking research on Corpus Linguistics, Corpus Design, Alignment and Translations Studies. This has been possible thanks to the BATMULT scholarship as part of a Marie Curie Training Site in Bergen (Norway).
During these 23 weeks (between December 20th and January 8th I was back in Spain) I have been working on my Phd Thesis, especially on those chapters that could benefit from the expertise of corpus alignment research. I have been able to attend several seminars on this subject, which proved great help and encouragement to go on with the investigation. Furthermore, I have been trying to apply an aligning program to the texts under study. Knut Hofland, from the Hit Centre, has been cooperating on this, rewriting his own program for my purposes, proofreading the output of the aligning and providing useful tools and materials for the needs of this research. At the same time, we have been experimenting with sound files and digital image, and the Phd Thesis will actually integrate some links to sound and image from the aligned materials.
The texts I am working with are film texts, mainly film scripts, transcripts or film dialogues. My idea was building up a parallel corpus, including English originals and their translations into Spanish. The translations are of two main types: a) the typed versions done by a translator, b) the actual transcriptions of the dubbed or subtitled versions into Spanish. In order to do this, first it was essential to have the whole input in machine-readable form. The scanner was not helpful at this stage, since the generally poor quality of the written materials made it difficult to apply this (otherwise) useful tool. Thus it was necessary to type in all the written manuscripts (translations and originals) onto the computer. At the same time, some symbols were added, following Knut Hofland's suggestion, to make the texts machine-friendly. It was crucial that the aligning program could recognise some symbols in advance of being run. The subtitled versions were also transcribed directly from the screen. The dubbed material was written down thanks to the use of a program called TRANSCRIBER. Having first the whole sound file of a film put into a cd rom, this program easily allowed the rendering of the dialogues. This program was also employed for transcribing the original dialogues of a couple of films, from which I did not have access to the final written scripts.
Once the input was in the computer as Word document, I would pass them on to Knut Hofland, who would convert them into MS-DOS format so as to work with them. We have made use of anchor lists, lists of equivalents in both languages which allowe a better aligning of the versions. We have run the program first on bi-texts, and then I would do the proofreading, spotting errors, “jumps” and so on. Especially at the beginning of this stage, the aligning went out of face quite easily, due to the special features of film scripts, especially those censored or self-censored. These censored translations involved additions, suppressions and modifications of the original dialogue, thus causing difficulties which we have been able to overcome.
After some weeks running the alignment program with film texts, proofreading and improving the program so as to better apply to film scripts, I found out that it was actually working and that my main purpose in applying for the BATMULT grant was being achieved. With one of the film texts, Life of Brian, we first started running the alignment program with several versions in different columns and adding the link to sound (dubbed version) and to digital image. It was only a pilot study we made, but it showed how this kind of aligning actually worked. Therefore, the main findings I have been able to obtain are a) the possibility of applying an aligning program otherwise used for literary texts in other languages to transcripts from films; b) the implementation of this program allowed its application to multi-texts such as one original scripts and several translations (going beyond bi-texts); c) by creating anchor lists and doing some manual alignment on the Word documents, it was attainable to deal with censored versions (shorter lines, things added) as well as with typical items of dialogues such as interjections, very brief phrases or inarticulate sounds; d) the possibility of playing with the sound to complete this corpus, since it was demonstrated that by getting access to the time code (from the transcribed dubbed versions) it was easy to have a link from the line to its spoken equivalent; e) the use of digital image with sound is to be used at least in a case study in the Phd Thesis; thus the whole film Life of Brian was put into a cd so as to find the links to the written material.
In these weeks, at the same time as working on corpus alignment design and on inputting all the data into the computer, I have been completing various chapters of my Phd thesis. There were mainly related to the theoretical framework having to do with Translations Studies, Audiovisual Translation and Film Semiotics. For this purpose, it was useful to borrow several books and publications included in the resources of the university library in Bergen. I also worked on some other chapters of my thesis, taking advantage of the computing and printing facilities at the Hit Centre.
I have attended four seminars during my stay in Bergen, presenting some of the work in progress and some other materials such as videotapes and slides used in previous research. In September I and my colleague Ignacio Pérez had an introductory meeting with the staff of the Hit Centre on “Translation and Censorship in Spain”. This proved useful so that the professors and other staff members involved could give us feedback on what we were doing. It was practical for us to be aware of what kind of human and technical resources they were able to offer us. In October, I had a second presentation, this time solely focused on textual analysis, and I showed scholars involved in different departments (Linguistics, Spanish, English, etc.) how I would proceed as to analyse the equivalence of translated film texts and their originals. The speech was called “Textual Analysis in Translation and Censorship: the Translation of Film Texts from English into Spanish 1975-1985”, thus connecting the issue of textual and translation analysis with other relevant topics such as film censorship in Spain and the characteristics of film texts as such. The necessary feedback also took place, with attendants coming up with lots of ideas on how to improve my model of analysis, derived from the use of conflictive concepts such as “equivalence”, “shift of meaning”, etc.
In November Ignacio Pérez and I were invited by professor Stig Johansson to the University of Oslo to give a talk on “Translation and Censorship in Spain”. I focused on “The Translation of Film Texts from English into Spanish (1975-1985)”, showing the audience the applicability of my model of analysis as well as illustrating the talk with slides covering the range of problems faced in the research, from the theoretical methodology to the problem of the term “auto-censorship”, etc. Some films extracts were displayed on screen, in which the effects of official external censorship in Franco's Spain were clearly manifested.
In December our supervisor in the BATMULT project professor Koenraad de Smedt organized an “International Seminar on Corpus Alignment”. In this seminar, in which experts of the field presented their newest achievements, we had the chance to deliver a paper on “Some Problems in Aligning Multiple Versions of Censored Translations”. Here I focused on how the program was being applied to film texts, both original and translations, and the difficulties implied by the specific features of film dialogues and the study of more than one translation together with the original. It seemed to be clear after this seminar that once I had decided on building a parallel corpus as the main tool and object of study in my thesis, this had to be obligatorily aligned. Moreover, this alignment had to cope with the basic needs of my research overcoming barriers like the impossibility of aiming at a “perfect” aligning of film scripts and the specific linguistic features of the various texts: the summary involved in subtitling, the omissions and additions allowed by the (auto)censored material, the overlapping involved in some dubbed dialogues, etc.
All in all, the work carried out under this Marie Curie Training Site in Bergen has been entirely positive. In my application submitted at the end of May 2002, I wrote:
The main problems we are now facing... are related to the creation of corpora, the selection of textual units as comparable units of analysis and the alignment of pieces of text. We find here a need to devise useful tools which may account for both the size and the shape of the corpus and which would define the criteria for structuring and determining portions of texts.
I can now assert that the creation of my parallel corpus of 8 original film texts in English and their counterparts in Spanish (both written and spoken material) have been completed. The texts have been segmented according to the basic unit namely “sentence”, but some other syntactic and punctuation criteria have been employed. I have reached further than I thought I would in the beginning, since at first I only had portions of texts under study, and now I have “whole texts” “wholly aligned” at the sentence level. I have certainly met my needs in this six-month stay in Bergen. Now the Phd Thesis on film translations and originals will be a corpus-based study instead of just a merely work inspired by portions of texts.