Audiovisual Archives: Digital Text and Discourse Analysis

Today, audiovisual archives and libraries have become very popular especially in the field of collecting, preserving and transmitting cultural heritage. However, the data in these archives or libraries - videos, images, soundtracks, etc. - constitute as such only potential cognitive resources for a given public (or “target community”). One of the most crucial issues of digital audiovisual libraries is indeed to enable users to actively appropriate audiovisual resources for their own concern (in research, education or any other professional or non-professional context). This means, an adaptation of the audiovisual data to the specific needs of a user or user group can be represented by small and closed "communities" as well as by networks of open communities around the globe.
"Active appropriation" is, basically speaking, the use of existing digital audiovisual resources by users or user communities according to their expectations, needs, interests or desires. This process presupposes: 1) the definition and development of models or "scenarios" of cognitive processing of videos by the user; 2) the availability of tools necessary for defining, developing, reusing and sharing meta-linguistic resources such as thesauruses, ontologies or description models by users or user communities.
Both aspects are central to the so-called semiotic turn in dealing with digital (audiovisual) texts, corpora of texts or again entire (audiovisual) archives and libraries. They demonstrate practically and theoretically the well-known “from data to metadata” or “from (simple) information to (relevant) knowledge” problem, which obviously directly influences the effective use, social impact and relevancy, and therefore also the future, of digital knowledge archives. This book offers a systematic, comprehensive approach to these questions from a theoretical as well as practical point of view.

Contents

Part 1. The Practical, Technical and Theoretical Context
1. Analysis of an Audiovisual Resource.
2. The Audiovisual Semiotic Workshop (ASW) Studio – A Brief Presentation.
3. A Concrete Example of a Model for Describing Audiovisual Content.
4. Model of Description and Task of Analysis.
Part 2. Tasks in Analyzing an Audiovisual Corpus
5. The Analytical Task of “Describing the Knowledge Object”.
6. The Analytical Task of “Contextualizing the Domain of Knowledge”.
7. The Analytical Task of “Analyzing the Discourse Production around a Subject”.
Part 3. Procedures of Description
8. Definition of the Domain of Knowledge and Configuration of the Topical Structure.
9. The Procedure of Free Description of an Audiovisual Corpus.
10. The Procedure of Controlled Description of an Audiovisual Corpus.
Part 4. The ASW System of Metalinguistic Resources
11. An Overview of the ASW Metalinguistic Resources.
12. The Meta-lexicon Representing the ASW Universe of Discourse.

• Language: English
• Publisher: Wiley
• Release date: Mar 1, 2013
• ISBN: 9781118614099

Book preview

Preface

One of the main issues involving today’s digital libraries and archives is of allowing users an active appropriation of their textual and, more specifically, audiovisual resources. Active appropriation means adaptation of the audiovisual data to the specific needs and interests of a user or group of users. A group of users may in fact be an entire, enclosed little community, e.g. the participants in a research project or the members of a teaching team, as well as virtual community- type networks which are active the whole world over. Examining this process of active appropriation involves:

1) effectively and systematically taking account of the internal structural organization of the audiovisual text, i.e. by looking at more precisely the semiotics of the text or discourse [STO 03];

2) defining and developing models and tools to enable anyone to physically and/or intellectually process the audiovisual text (e.g. by analyzing and interpreting it);

3) the opportunity made available to anyone and everyone to become an author in the sense of someone intending to produce and publish one or more new versions of a pre-existing audiovisual text or corpus of texts – new versions which are better adapted to a specific use context or, more generally, to the (cultural) profile and expectations of a given audience.

One particularly central question in this context is of the metalanguage of description*¹, which is needed to enable anyone to carry out analyses* of all sorts of audiovisual texts or corpora of texts documenting the universe of discourse* of an audiovisual archive. This book is devoted to the presentation and critical discussion of such a metalanguage of description.

The conception, elaboration and actual experimentation of a metalanguage of description for analyzing audiovisual corpora, as well as the development of a software environment to test that metalanguage, was made possible thanks to funding from the Agence Nationale de la Recherche (ANR – French National Research Agency)² in the form of a research and development (R&D) project called the Audiovisual Semiotic Workshop for analyzing audiovisual corpora in Human and Social Sciences (ASW-HSS) – the English equivalent of the French Atelier de Sémiotique Audiovisuelle pour l’analyse de corpus audiovisuels en Sciences Humaines et Sociales" (ASA-SHS).³

The environment which bears the name ASW Studio (Studio ASA) is presented in the collective volume Introduction to Audiovisual Archives [STO 11a]. In another collective work entitled New uses of Audiovisual Archives [STO 11b] we present and discuss various examples showing the potential advantage of that environment for projects to compile and exploit audiovisual archives and, in particular, for analyzing specific audiovisual corpora.

The content of this book is divided into four main parts. Part One, made up of the first four chapters, sketches the general context of our work dedicated to defining and elaborating a metalanguage of description with a view to analyzing an audiovisual text or corpus of texts.

To begin with, we shall outline what we mean by analysis of a text or corpus of audiovisual texts (Chapter 1). Here, we shall focus more particularly on what, in the literature, and in reference to the work of Michel Foucault, is termed the linguistic turn in research dedicated to digital archives and/or libraries (see e.g. [CRA 08]). At the center of this approach is the question of explicitizing the universe of discourse of an archive or library, i.e. the meaning or the content of the resources (be they written, audiovisual, etc.) which make up that archive or library. This manner of defining both the (potential) interest and the active appropriation of an (audiovisual) text or archive of texts by its users corresponds exactly to the point of view developed in this book.

In the next chapter (Chapter 2), we shall give a summary presentation of the working environment called ASW Studio which our Equipe de recherche en sémiotique cognitive et nouveaux medias (ESCoM)⁴ (research group on cognitive semiotics and new media) at the Fondation Maison des Sciences de l’Homme (FMSH)⁵ (Foundation House of Human Sciences) in Paris implemented in order to test and validate the approach put forward in this book. As we shall show, the metalanguage of description peculiar to an archive’s universe of discourse manifests itself in the form of a library of models of description*, to which the analyst (the person or persons carrying out an analysis of an audiovisual text) has access by way of specialized interfaces (interactive forms for entering information or data) making up the ASW Studio. That said, the metalanguage of description as such remains independent of a specific working environment. In other words, it can be operational in the most varied of environments. Let us acknowledge here that it is to Francis Lemaitre [LEM 11a; LEM 11b], a computer engineer at ESCoM, that the great achievement of having designed and developed the ASW Studio (in spite of always very limited financial means) must be attributed.⁶

Following the summary presentation of ASW Studio, in the following chapter (Chapter 3) we shall introduce a concrete example of analysis of an audiovisual text using an approach based on a library of models of description* representing the universe of discourse* of an archive. This concrete example will, in a manner of speaking, serve as a guiding thread to develop and explicitize our approach throughout the book.

Thus, with the example presented as precisely as possible, in Chapter 4 we shall draw a number of more general consequences from it, which will be discussed in much greater detail in the remaining three parts of the book: typology of analytical practices; descriptive models appropriate for such-and-such a type of analysis; hierarchical and syntagmatic organization inherent to a descriptive model; and the main bricks making up a descriptive model.

Part Two is made up of three chapters (chapters 5, 6 and 7) and is devoted to a detailed discussion of three analytical tasks we have found to be recurrent in terms of the description/indexing of the content of an audiovisual archive. Those tasks are:

– analysis of the domain of expertise (i.e. of the referential object, the referent of an audiovisual text or corpus of texts);

– analysis of the contextualization (spatial, temporal, social, etc.) of the domain of expertise dealt with in a text; and finally,

– analysis of the strategies for conducting discourse about the referent of the text.

Along with two other tasks – that of analyzing the verbal and audiovisual expression of the content of a text, and that of the analyst’s commentary or metadiscourse about his/her (hereafter his) analysis – which will be developed in this book, the three aforementioned tasks constitute a canonic ensemble of tasks defining a specific type of analysis dedicated to the description, indexing and interpretation of the content of an audiovisual text or corpus of texts.

Let us stress that analysis of the content of an audiovisual text or corpus of texts is only one, very specific type of analysis (see Chapter 4). Other types of analysis, which are developed and made operational in ASW Studio but which we shall not speak of in this book are, for instance:

– paratextual analysis* (a type of analysis which focuses – in keeping with the Dublin Core standard⁷ – on describing the formal identity of a text: author, title, rights, genre, etc.);

– audiovisual analysis* per se (analysis which is primarily concerned with describing the visual and sound shots of an audiovisual text);

– pragmatic analysis* (which relates to the contextual anchoring – social, educational, linguistic etc. – of the text being analyzed);

– meta-description* (a particular form of analysis which serves to explicitize the objectives and stakes involved in the analysis of an audiovisual object).

Chapter 5 is given over to an in-depth discussion of the various theoretical, methodological and technical aspects of analysis of a field of expertise thematized in an audiovisual text. Here, we find the central theoretical notion – borrowed from Greimas’ semiotic theory – of the thematic configuration* which enables us to describe the subject of an audiovisual text or corpus of texts. The subject of the text is made up of both what the text talks about and how it talks about it, and how it expresses that which it is talking about. More precisely, the thematic structure is made up, among other features, of:

– the topical structure* per se (i.e. the structure defining the referential object dealt with in a text);

– the structure of the discourse production* (i.e. the structure defining the approach to the referential object dealt with in the text by its author); and

– the structure of the verbal and audiovisual expression* (i.e. the structure defining the verbal, visual or audiovisual language in which the content is expressed).

In Chapter 6, we shall discuss the analysis of the context of the knowledge object or field thematized in an audiovisual text. In particular, this relates to the spatial (geographic, geopolitical) and temporal (chronological, historical) contextualization, complemented by a third form of contextualization called thematic contextualization. This latter identifies the institutional, social/cultural, epistemological or mental context which is relevant for an object thematized in an audiovisual text.

The procedures of discourse production of an object being thematized and contextualized will be dealt with in Chapter 7. While remaining conscious of the fact that this is a complex and diversified field of research, we shall focus on a few aspects we deem important, in order to better define, e.g.:

– the interest that the referential object thematized in a text holds for its author;

– the added value of the thematization of the referential object in a given text;

– the specific (original, novel, etc.) character of the treatment of the referential object in the text;

– the author’s objectives in thematizing this-or-that referential object.

As we shall see in Part Two of this book, the performance of a specific analytical task (analysis of the object or the domain, analysis of the object’s context, analysis of the discourse production from the object, etc.) relies upon the sequences of description* which make up a given model of description. Thus, for instance, we distinguish:

– sets of sequences dedicated to the analysis of the domain or referential object thematized in an audiovisual text or corpus of texts;

– sets of sequences dedicated to the analysis of the context of a referential object being thematized;

– sets of sequences dedicated to the analysis of the discourse production of the referential object in question.

Each sequence, in turn, is made up of a canonic pair of schemas of definitions – one defines the object of analysis, the other the procedure is employed to analyze the object in question.

In addition, we distinguish between two fundamental description procedures* which can be employed by any analysis. The presentation and discussion of these is the main topic of Part Three of this book, made up of three chapters (Chapters 8, 9 and 10).

In Chapter 9, we shall discuss the first of these two procedures – the procedure of so-called free description*. The analyst – as indeed the expression suggests – is free to produce such-and-such a piece of information, such-and-such a piece of data to supplement or index an audiovisual text in order to give an account of the domain dealt with and the way in which that domain is approached and expressed by the author.

The second procedure is presented and discussed in Chapter 10 – the procedure of so-called controlled description*. This is done using an appropriate microthesaurus*, with which the analyst describes the referential object thematized in an audiovisual text, the spatial, temporal or thematic context, such-and-such a strategy of discourse production or such-and-such a way of expressing the subject developed in the text being analyzed.

Chapter 8, with which Part Three begins, is for its part, reserved for more specific questions relating to the definition of the topical structure*, i.e. that part of thematic configuration* which is reserved for analysis of the domain of expertise – of the referential object* – thematized and dealt with in an audiovisual text or corpus of texts.

Finally, Part Four of this book – the most important part – consists of 6 chapters (Chapters 11, 12, 13, 14, 15 and 16) which, in as systematic and detailed a manner as possible, present the metalinguistic system and the various components of it which we need to specify the metalanguage of description* (in the form of a library of models of description) to describe the resources in an audiovisual archive.

Chapter 11 offers an overview of the various components of the aforementioned metalinguistic system and their functions in the specification of a model of description or library of models of description.

Chapters 12, 13 and 14 are given over to a detailed presentation of the heart of the system of metalinguistic resources constituted by two meta-lexicons (ontologies) of conceptual terms* (concepts):

– the first meta-lexicon identifies and designates all the objects of analysis of the ASW universe of discourse: referential objects, location objects (spatial, temporal etc. location), discursive objects or objects of expression (textual, linguistic, visual, etc.);

– the second meta-lexicon identifies all the analytical activities for describing, indexing, annotating, illustrating etc., the objects of analysis of the ASW universe of discourse.

In Chapter 15, we shall present another resource central to the ASW metalinguistic system – the thesaurus* which we need to implement regarding the procedure of controlled description. This thesaurus is made up of:

– a library of facets (i.e. of semantic dimensions), and

– a library of standardized expressions (descriptors) which are classified according to the facets of the thesaurus.

The final chapter of this book – Chapter 16 – will be devoted to a more systematic discussion of the generic building blocks* which make up a model of description or a library of models of description. There are two types of generic building blocks:

– sequences of description* (referential, contextual, discursive, of verbal expression, of audiovisual expression, etc.), and

– the schemas* which define each sequence (schemas defining the objects to be analyzed and schemas defining the activities to be carried out to analyze an object).

Sequences and schemas form configurations which select and position one or more conceptual terms* from the two meta-lexicons cited above. Along with the thesaurus and the two meta-lexicons of conceptual terms, they constitute an important third category of the ASW’s metalinguistic resources.

This book constitutes the (provisional) conclusion of a large part of the research with which we have been engaged for a long time. Indeed, our interest in the metalanguage of description of (audiovisual) texts goes back to around 1983–1984 when A.J. Greimas, with whom we have been privileged to work with for over 15 years, led us to take an interest in the theoretical and technical questions thrown up by semantic or conceptual networks, which were very much en vogue at the time in research on artificial intelligence and devoted to knowledge representation systems.

As part of our research effort at the CNRS (Centre National de la Recherche Scientifique) between 1985 and 1991, and of a very stimulating long-lived collaboration with IBM France at the end of the 1980s, we were able to work concretely on questions relating to the extraction and representation of the content in lexical and textual databases using semiotic- and structural-type models (structural in the sense of structural linguistics). From 1995/1996 onwards, thanks to a whole series of European research projects⁸, we were able to extend the problem of defining a metalanguage of description in the form of models of description⁹ to digital libraries and finally, in 2000/2001, to digital audiovisual corpora and archives.

If CISCO’s predictions¹⁰ prove true – in particular those which say that by 2014, 91% of the digital data circulating on the Internet will be audiovisual in nature – we can measure how enormously important the issue of conceptual models (and therefore of a system of metalinguistic resources* enabling us to create them) for processing (analyzing, adapting, publishing, circulating, conserving etc.) the most varied of audiovisual corpora, will become in the near future. Having such a system of metalinguistic resources indubitably constitutes a condition sine qua non for intelligent management and exploitation of these data.

We feel a certain degree of personal satisfaction in the work accomplished over nearly 25 years now, while we are aware of its limits, its gray areas and its extremely provisional nature. In view of the empirical complexity of the (audiovisual) texts and the diversity of the (very often unspoken) expectations of their audiences, our certainty of yesteryear has been replaced by a far more modest position. Any metalanguage is an intrinsically limited and fallible construct, built on often unstable foundations and deeply rooted in its designer’s cultural views and beliefs (scientific or otherwise). However, this is a tool which is indispensable for any work on audiovisual text (also see [STO 11a; STO 11b]).

By way of conclusion, we would like to express our gratitude to all those who, for a certain period of time, have formed part of the ESCoM team at the FMSH in Paris and contributed to the implementation of the Archives Audiovisuelles de la Recherche (Audiovisual Research Archives, ARA – see [STO 03c; STO 10a]) and to the advancement of our research devoted to the analysis and publication of audiovisual corpora. A particular mention goes to Elisabeth de Pablo, who has been with us in all our research projects and who has helped us so much in our scientific work.

Our special thanks also go to ISTE/Wiley for giving us the opportunity to present our research on a metalanguage of description of audiovisual texts to a non-French speaking audience. We are also especially grateful to Benjamin Engel for all his work on the book.

The older we get, the more we feel we are acquiring – in the words of Claude Lévi-Strauss – the right distance, which allows us to better appreciate the value of the people who have crossed our path as researchers. Among those most dear to us, and who will always live in our memory, we count Monique and Charles Morazé, to whom our debt is so much more than merely intellectual! This book is dedicated to them in fond, faithful and filial memory.

Peter STOCKINGER

Lambach – Paris

April 2012

---

1 Those expressions followed by an asterisk (*) are defined and described in the glossary of specialized terms at the end of the book.

2 See http://www.agence-nationale-recherche.fr/.

3 The ASW-HSS project (Audiovisual Semiotic Workshop for analyzing audiovisual corpora in Human and Social Sciences; ASA-SHS in French) went on for three years. It began at the start of 2009, as part of the ANR’s Programme Blanc (White Program), and came to an end in December 2011. Various sites bear witness to the activities and results of this research project: its official website, http://www.asa-shs.fr/ (French only); the project’s research log on the portal Hypothèses.org: http://asashs.hypotheses.org/ (French only); and finally the ASWHSS portal, providing access to the experimentation workshops in the form of audiovisual archives, classified thematically and geographically: http://semiolive.ext.msh-paris.fr/asa-shs/. The ASW-HSS project is developed around and based on three major audiovisual corpora, each constituting the collection of a specific archive: the Arkeonauts’ Workshop (ArkWork) (French equivalent Atelier des Arkéonautes, ADA): http://semiolive.ext.msh-paris.fr/ada/; Literature from Here and Elsewhere (LHE) (French equivalent Atelier Littéraire d’Ici et Ailleurs, ALIA): http://semiolive.ext.msh-paris.fr/alia/; and Culture Crossroads Archives (CCA) (French equivalent Archives Rencontre des Cultures, ARC): http://semiolive.ext.mshparis.fr/arc/. In the final year of the project, these three experimentation workshops were supplemented by two other workshops which facilitated both the testing of the approach and environment developed during the ASW-HSS project on other corpora of audiovisual texts and the concrete demonstration of the advantage of using ASW Studio in projects to create and exploit bodies of scientific and cultural audiovisual heritage.

4 Official site of ESCoM: http://www.semionet.fr.

5 Official site of the FMSH: http://www.msh-paris.fr.

6 For many years, we have also been able to count on the invaluable technical and scientific expertise of our colleagues and friends, Steffen Lalande, Abdelkrim Beloued and Patrick Courounet of the INA (Institut National de l’Audiovisuel – French National Audiovisual Institute), and on the very dedicated work of Richard Guérinet, who has been collaborating with Francis Lemaitre since early 2011 to perfect ASW Studio.

7 See http://dublincore.org/.

8 More information relating to these projects can be found on ESCoM’s website (http://www.semionet.fr).

9 At the time, we spoke of semiotic scripts or thematic scripts applied, on the one hand, to analysis of the content of textual databases, and on the other, to the conception and creation of online products and services (see [STO 94]).

10 See http://newsroom.cisco.com/dlls/2010/ts_111510.html.

PART 1

The Practical, Technical and Theoretical Context

Chapter 1

Analysis of an Audiovisual Resource

1.1. Introduction

This book’s goal is to present a functional approach based on the semiotics* of the audiovisual text* [STO 03] for the analysis, i.e. the description, interpretation and indexing of digital audiovisual corpora.

The central notion used for this approach is of the model of description* of an audiovisual object, such as a video, based on a set of criteria which serve the semiotics to process the text object* and will be presented in greater detail in Chapter 3 of this book. Primarily, it is a question of the following criteria:

– the criterion of the text as a compositional entity (a text can, in principle, be broken down into smaller textual units, and in turn forms part of a textual environment, of what is, metaphorically speaking, a textscape or mediascape);

– the criterion of the text as a structural entity possessing a set of characteristic constituents (such as the thematic constituent, the narrative constituent, the rhetorical and discursive constituent, the multimodal expression of content, or the formal and physical organization of the content in the text); and finally,

– the criterion of the text as a historical entity (the text as a genre) and an evolutive entity (the text as the product of savoir-faire, in principle always modifiable).

The hypothesis behind this book is that any project of analysis of a textual corpus in general and an audiovisual corpus in particular – whatever its level of specialization – relies on representations, visions, or theories: 1) about the object text and 2) the activity of the analysis* of the text.

Thus, all told, a model of description is nothing more or less than the explicitized, formalized (in the broader sense of the word) part of a theory or vision which guides the task of analyzing a textual corpus (in our case, audiovisual).

The gap of satisfaction which may exist between the model and the theory or vision underlying the work of analysis can be explained either as a more or less significant implicit factor which guides the analyst in his work and which the model is not capable of taking into consideration, or by imperative simplifications which must be carried out in relation to a theoretical referential to develop an explicit and functional approach to the analysis of textual, or audiovisual, corpora.

The work of definition, development, validation and tracking of models of description of textual and, particularly, audiovisual corpora, still represents an entire occupation, i.e. a set of specialist skills and knowhow calling on a varied body of culture and knowledge which cover not only the practical and technological domains such as information and knowledge technology, applied sciences of documentation, archiving, library sciences or the management of cultural heritage lato sensu, but also – and, in our opinion, crucially – a set of disciplines in human sciences such as text sciences (and particularly semiotics*), linguistic sciences or even that heterogeneous emerging set of approaches and problems classified under the general umbrella label of cultural sciences (Kulturwissenschaften, in German).

The occupation in question is that of the concept designer*, sometimes also called concept-designer*, or information technician or engineer; indeed, the terminology is still very fuzzy and unstable. However, it is a central role of the workflow* [STO 11e] defining the constitution, analysis and publication/diffusion of bodies of knowledge heritage which are channeled by audiovisual corpora. The modelizer prepares, develops and manages all the metalinguistic resources necessary for the other actors involved to carry out their work.

1.2. Functionally different corpora

As part of the process of digitizing knowledge heritage, we can distinguish a series of categories of models (i.e. metalinguistic resources) needed to accomplish the various activities making up that process.

As set out in [STO 11e], the process of constituting a body of knowledge heritage in the form, e.g. of a digital archive, takes place in various canonic stages – notably:

1) the stage of preparation of a field for collection of data documenting a body of cultural heritage;

2) the stage of the realization of the field work¹;

3) the stage of technical and auctorial treatment of the data collected (including, amongst other things, the derushing of audiovisual data, the montage and postproduction of the audiovisual data collected);

4) the stage of analysis (description, indexing but also pragmatic adaptation) of the data collected and documenting a terrain;

5) the stage of the publication and diffusion of the data collected and/or analyzed and, finally,

6) the stage of conservation of the data collected/analyzed/published.

However, each stage in this process of digitization of a body of knowledge heritage necessarily has to do with a certain functionally specialized type of corpus* (in our case, an audiovisual corpus):

1) The stage of preparation of a field for collecting audiovisual data can only be conceived of in reference to a pre-existing corpus, or by compiling the knowledge and sources of information necessary to the proper functioning of the field work (knowledge and sources which could cover bibliographical references, online resources, personal information, ‘good practices’, examples of similar projects underway or already carried out, directories, etc.).²

2) The stage of data collection leads to the creation or updating/enriching of a pre-existing field corpus*. The field corpus is made up not only of data produced within the boundaries of the field. Take the example of the recording of a field as circumscribed as a research seminar whose sessions to be filmed are spread out over a whole academic year. The corpus of data documenting the field research seminar is not (necessarily) restricted to the audiovisual recordings of the various sessions. It covers all the data deemed pertinent either to give an account of that field (i.e. to make it an archive of knowledge in the true sense of the term), to facilitate a highquality analysis of such-and-such an aspect of the filmed session, to have a documentary base in view of one or more publications (online) of the seminar, or to transform it (as it is, or after a process of selection of documents which must be preserved absolutely) into a heritage corpus (see below) documenting, e.g. the history of a discipline or of a research institution.³

3) The stage of technical and auctorial processing relies on a selection of collected data forming part of a field corpus, or else of several field corpora, or on data stemming from different periods in the life of a field corpus (a field corpus can be updated, enriched, etc.). In any case, a processing corpus* is composed of data selected, e.g., with a view to being cut together to constitute a new audiovisual creation corresponding to an authorial intention to publish (i.e. to a scenario defining such a creation). Thus, an intention to publish the recordings of a research seminar may be aimed at diffusing a certain problem dealt with during the said seminar. In this case, not the entire seminar is the object of an intention to publish, but rather just those parts of it in which the problem chosen is dealt with. Yet even when a decision is taken to publish the entirety of the seminar, the recordings made during the field phase have to undergo technical processing (encoding, checking of the image and sound quality, deletion of unusable passages, etc.) before being made available for publication of the seminar and its various sessions in the form, e.g., of a website. Hence, no matter whether the processing stage is reduced to a simple activity of processing or whether it also covers a genuine authorial activity per se, the question of the definition and constitution of the processing corpus arises every time. Note that in addition, in the context of digital archives of knowledge, a processing corpus can be fed not only by data from one or more field corpora, but also by data already published and re-injected, reused in the context of a new technical, and above all authorial, treatment. In concrete terms, a corpus documenting a scientific problem which is dealt with in a seminar and which is the object of a montage with a view to publication online, alongside original data (e.g. from a new field corpus), may perfectly well include parts from pre-published contributions.

4) The stage of analyzing the data collected is the one which interests us most, and to which this book is dedicated. For the moment, let us highlight that the analysis of a piece of textual information (or, in our case, audiovisual information) cannot be reduced to a simple free indexation, nor to indexation controlled according to this-or-that standard, this-or-that documentary language. The analysis includes all intellectual activities – from documentary indexation to the most personal interpretation, through the various forms of professional assessment of the information – which use and exploit the object text* to satisfy a need (a desire, or a simple curiosity) for knowledge. However, such a need or desire may stem from very variable motivations, and arise in extremely different social and cultural contexts. It is still true that analysis as an activity to satisfy a need or desire for knowledge can only be successfully carried out if the right object is available to it, as its primary material which is the text* or rather, the corpus of texts. In the context of the constitution and diffusion of a body of cultural/knowledge heritage, the analysis corpus*, i.e. in our case the corpus of audiovisual data being analyzed, is not necessarily coextensive with a field corpus – far from it, in fact. Indeed, everything depends on the goal of the analysis* and, more generally, on the analytical policy* (e.g. in the context of exploitation of the contents of an archive* of knowledge). If the analysis is conceived as an activity of description and classification of data collected beforehand and documenting a particular field with a view, e.g. to their publication online, the field corpus and analysis corpus become similar – although they do not merge. If the analysis is conceived independently of the activity of collection, the corpus of audiovisual data needed for the analysis to fulfill its goals, obviously, no longer has anything to do with this-or-that field corpus. The analysis corpus is constructed and enriched solely according to the objectives of the analysis itself. In [STO 11b], two examples are provided of the constitution of an analysis corpus fed by data from different field corpora: the first example is of the analysis of traditional bread-making in France and Portugal [DEP 11d]; the second of the comparative analysis of the view of the Arabian Nights and the creative uses certain artists make of the tales [CHE 11b]. In both cases, the analysis corpus is composed⁴ not only of data derived from different field corpora (fields created as part of the ARA Program)⁵ but also possesses its own dynamic of enrichment or updating.

5) The stage of publication/diffusion of the data in turn relies on a corpus of data – the publication corpus* – which is not necessarily coextensive with the field corpus or the analysis corpus. For instance, in the process of publishing a research seminar or interview, filmed and analyzed, the four functionally distinct corpora – the field corpus, the processing corpus, the analysis corpus and finally the publication corpus – may become similar, or even (partially) overlap. However, they remain functionally distinct, obey their own motives and objectives, evolve