Chapter 5

The Analytical Task of “Describing the Knowledge Object”

 

5.1. Introduction

In this chapter, we shall more closely consider the analysis of the objects (situations, events, etc.) which make up the domain of knowledge thematized in an audiovisual text or corpus. As has already been said, this is one of the two aspects of so-called referential description* of the knowledge object dealt with in a text (in our case, an audiovisual text). The other side of referential description is the contextualization of the knowledge object, i.e. localizing it (spatio-temporally, socially, historically, etc.), which will be discussed in Chapter 6.

Also, let us not forget that referential description* constitutes one of the main tasks of a specific type of analysis of the audiovisual text, which we shall call thematic description* or description of the audiovisual content. The other tasks which make up the type of analysis called thematic description are as follows: the task of description of the discourse production* around a subject developed in the audiovisual text, the task of description of the audiovisual expression (or mise-enscène)* and/or verbal expression of the content and, finally, the task of production of a commentary by the analyst himself as regards his own description (see Chapter 4, Figure 4.4).

We shall begin this chapter by discussing a simple example of referential description (section 5.2). For this purpose, we shall return to the example developed in Chapter 3, relating to the description of an audiovisual text which deals with the technical culture of the Chavín civilization.

Then (section 5.3), we shall outline a number of theoretical and methodological distinctions which we believe to be important in order to fully comprehend the objective and the stakes involved in referential analysis. The distinctions to which we refer are those between subject* and topical configuration* (or structure) and between thematic configuration* and topical configuration* stricto sensu.

Section 5.4 is given over to a brief presentation of a library of sequences* for use in the referential description of an archive’s universe of discourse. The example used is of the CCA (Culture Crossroads Archives)¹, which is one of the experimentation workshops of the ASW-HSS project.

Finally, section 5.5 introduces the issue of conceptual modeling as an indispensable activity for defining a model of description of the content of an audiovisual corpus. This issue will be further developed in Chapter 8 of this book.

5.2. A simple example of referential description

The extract of the analytical form shown in Figure 3.2 shows the functional field reserved for the analytical task of Describing the knowledge object(s) thematized in the audiovisual text.

As pointed out in Chapter 3, the functional field in our example is equipped with a global sequence made up of two sub-sequences. Each sub-sequence, is in turn, made up of schemas of definition* at which level the actual task of analysis or description of a certain type of object is carried out. Indeed, as Figure 4.2 shows, each of the two sub-sequences:

– relates to a specific type of object from the domain of reference: the first subsequence relates to the type of objects [Civilization] and the second to the type of objects [Cultural construct];

– integrates a procedure of description* (each procedure being made up of various specific activities of description) which through using the analyst carries out his task of analysis devoted to a knowledge object: the first sub-sequence integrates the procedure of so-called free description (used to inform the conceptual term [Civilization]); the second integrates the procedure of so-called controlled description (used to inform the conceptual term [Cultural construct].

Figure 3.2 demonstrates that the conceptual term [Civilization] can be informed by the four activities of description which together make up the procedure of free description – a procedure which will be presented in greater detail in Chapter 9:

1) The activity of indicating the chosen conceptual term, i.e. the term [Civilization]. In our case, it is a question of producing the expression <Chavín civilization> by way of an indexing form associated with this activity.

2) The activity of possibly producing additional information which further contextualizes the simple citation <Chavín civilization>. Such contextualization may be temporal (historical), spatial (geographic) or indeed thematic (institutional, social).

3) The activity of (possibly) producing a textual summary presentation in the form of a short explanatory text.

4) Finally, if applicable, the activity of producing a version in the original language of expression, indicating the conceptual term.

Let us specify that only the first activity, that of indicating the conceptual term [Civilization] is obligatory – all the others are optional. Of course, the more an analysis relies on the different activities of description, the more it is enriched but also the more time it takes to carry out.

Also, as shown in Figure 3.2, once the so-called free description of the conceptual term [Civilization] has been completed, the analyst is invited to provide information about the cultural phenomenon/a the audiovisual text being analyzed deals with. Here, the description of the cultural phenomenon is performed using the procedure of controlled description*. This means that the analyst employs a thesaurus*, from which he chooses the predefined terms (descriptors) he deems to be the most apt to characterize the cultural phenomenon in question. In our specific case, the predefined term, the descriptor, that the analyst has chosen to give an account of the cultural phenomenon dealt with in his audiovisual text, is the term <Technical culture>.

In brief, the task of referential description, performed using the form shown in Figure 4.2, can be summed up in terms of the information produced by the analyst, specifying that the audiovisual text being analyzed “talks about” the technical culture of the Chavín civilization.

Note, again, that the simplest act of thematic description can be reduced to three, very specific “gestures”:

– either selecting the concept (the conceptual term) to be informed (in our case, the choice is limited to consecutively selecting the two conceptual terms [Civilization] and [Cultural construct]);

– or freely entering the nominal expression which, in the analyst’s view, represents the referent of the concept or conceptual term (in our example, the referent chosen by the analyst to expand on the concept [Civilization] is the expression <Chavín civilization>);

– or carrying out a controlled selection of one or more descriptors from a thesaurus (in our case, selecting the descriptor <Technical culture> to indicate the value of the concept [Cultural construct]).

5.3. Thematic structure, topical structure and referential objects

Analysis of the content of an audiovisual text or corpus – the subject(s) thematized and developed therein – is performed, as we know, using predefined thematic structures. Such a predefined thematic structure is represented by a conceptual term or rather, a configuration of conceptual terms. Each conceptual term must be defined in the ASW meta-lexicon. Remember that we distinguish between two main types of conceptual terms:

– those which represent the objects of analysis*,

– those which represent the activities of analysis*.

As we shall see in greater detail later on, for each of the two types we have defined a conceptual vocabulary, a descriptive ontology which covers the ASW universe of discourse (for further explanations see Part Four of this book).

A thematic structure represented by a single conceptual term (CT) is said to be condensed; a thematic structure which is represented by a collection of conceptual terms is said to be expanded (or rather, expanded to a greater or lesser extent). A thematic structure which is expanded to its full extent is that which is constituted by an aggregation of local configurations which each define one of the five major sequences of the model of description which serves for carrying out the specific type of analysis called thematic description (of the content)* (see Chapter 4).

A thematic structure* which is condensed to its full extent is that which is reduced to two conceptual terms:

– the first represents an object of analysis*;

– the second represents the activity of description* to be used to specify the object of analysis.

Figure 5.1 shows the main local configurations which make up the thematic structure * through using which we specify a subject* developed (be it verbally, visually or by purely acoustic means) in an audiovisual text. The figure also illustrates the fact that only the topical configuration forms an integral part of the thematic structure (hence the frequent synonymous usage of the expressions “topical” and “thematic”).

Let us return now to the objects of analysis*. In [STO 11a], we distinguished five main categories which we use to define the thematic models (i.e. models of thematic description or description of the content). Using those thematic models, we describe a subject thematized and developed in an audiovisual text or, rather, in a corpus of audiovisual texts:

1. referential objects (i.e. those which form part of such-and-such an empirical domain documented by an audiovisual corpus);

2. contextual objects (of spatial/temporal localization, etc.);

3. discursive objects (i.e. of discourse production);

4. objects of audiovisual expression (i.e. visual expression, acoustic expression, syncretic expression, etc.) and verbal expression;

5. so-called reflexive objects (i.e. those which serve to produce a meta-discourse concerning the analysis itself).

As shown in Figure 5.1, a model of thematic description (or content description) identifies and selects, from these five categories of objects, those it needs to give an account of its textual object.

Now let us take a closer look at referential objects – the first of the five categories of objects of analysis that we have just identified. Referential objects in fact define a sub-part of the thematic structure which is the topic, or rather, the topical structure. The topic, quite simply, is what a text “talks” about, and a topical structure is the model which we can use to describe all (audiovisual) texts whose subject (or one of whose subjects) corresponds (reasonably closely) to the underlying topical structure.

Thus, going back to our example introduced in Chapter 3, the subject, the topic of the audiovisual text being analyzed is the technical culture of the Chavín civilization during its peak period as it manifests itself in the Andean village of Huantar in Peru (see Figure 3.2).

The topical structure, the working interface of which is constituted by the working form in Figure 3.2, does, however, form a generic configuration which enables us to actually analyze not only the audiovisual text in question (thankfully!) but indeed all audiovisual texts which deal with civilizations and their cultures on American soil.

Figure 5.2 provides a (cognitive) representation of part of the topical structure which enabled us to analyze (among other things) the audiovisual text which speaks about the Chavín civilization and which serves as an example here.

Every type of object of analysis is represented by a concept or, as we prefer to call it, by a conceptual term (or CT) which is displayed between squares brackets and begins with an uppercase letter. Thus, [Civilization] is a conceptual term whereas <Civilization> is a nominal expression generated by the analyst or predefined in a thesaurus;

Let us now take a closer look at Figure 5.2. It shows a small vertically oriented graph [SOW 84] between two conceptual terms which are interconnected through a chosen conceptual relation. Thus, the graph is the product of:

– a selection of 1) two conceptual terms (from the conceptual terms making up the vocabulary, i.e. the ASW meta-lexicon of terms, which represents the objects of analysis of the ASW universe of discourse*) and 2) a conceptual relation;

— a process of reciprocal positioning of the two terms using the relation {CT_i refers to CT_j} with the result of a thematic (here, topical) configuration in which the conceptual term [Civilization] is the term ab quo (that is, the original term) and the conceptual term [Cultural construct] the term ad quem (the resultant term).

Let us take from this that any referential or topical description* of a corpus of audiovisual (or other types of) texts relies (is based) on the selection of a CT or set of CTs representing the first category of objects of analysis, i.e. so-called referential objects. The selected conceptual terms also specify, from all the sequences* which make up a model of description of audiovisual content, a particular functional class which serves for the description of the domain (or rather, the objects of the domain) thematized in an (audiovisual) text or corpus. In other words, that class of sequences serves for describing what a text talks about (see section 1.6). Remember that this functionally specialized class of sequences always occupies first position in the canonic syntactic structure (see Chapter 4) of a model of thematic description. In the context of an analysis of audiovisual content, it is also the only sequence in the interactive form which necessarily has to be filled in by the analyst.

5.4. A library of sequences for referential description

Figure 5.2 is a representation of the referential part of the model of description. of the audiovisual content underlying the interactive working form shown in Figure 3.2. Figure 5.3, on the other hand, shows how an account can be given of the same model from the point of view of the organization of the interactive form which serves as a working interface between the model of description and the analyst.

The choice here is to construct particular sub-sequences for each conceptual term making up the topical configuration shown in Figure 5.1. In this case, each subsequence would be reserved for the description of a conceptual term.

Let us consider Figure 5.4, which shows an extract from the library of sequences used to describe the content of the audiovisual texts making up the collection of the Culture Crossroads Archives (CCA).² This library of sequences is defined and created by a person (or group of people) acting as concept designer (knowledge engineer), whose mission is to develop and manage the metalanguage of description for analyzing the audiovisual texts and corpora of the CCA archives. This specific task – which necessarily precedes any work of analysis – is carried out using the ASW Modeling Workshop* with the help of an xml editor called OntoEditor.³

The CCA library of sequences is organized into collections of sequences by major themes or subjects which circumscribe the universe of discourse* of the CCA archives. For instance, this library contains collections of sequences or isolated sequences which define the topical structure of themes or subjects such as:

– cultural constructs of a people,

– cultural constructs of a country,

– intangible cultural heritage,

– technical and material culture,

– research on cultures, etc.

Note that these sequences may be reused as they are or following certain modifications, to define the referential part of the models of description of audiovisual content which make up the libraries of other archives.⁴

Thus, among the collections of sequences defining the universe of discourse* of the CCA archives (Figure 5.4) is the collection of sequences Description of a civilization and its culture – a collection which contains our two sub-sequences Description of the civilization and Description of the cultural construct in question. They are accompanied by a third sub-sequence entitled Circumscription of the domain of expertise which a priori establishes the referents which the analyst must respect when carrying out a concrete analysis. In a manner of speaking, these referents define what the great sociologist Alfred Schütz [SCH 03], referring to Husserlian phenomenology, called the horizon of meaning of the epistemological structure of the lifeworld.

Remember that a sequence is necessarily defined by one or more schemas of definition*. In Figure 5.4 (at the bottom right of the image), there is a field entitled the “Field of definition” reserved precisely for selecting the appropriate schema or schemas for the structural and functional specificity of a sequence (functionally specialized either in the referential description of a topic, in the description of the discourse production from a topic, the description of the verbal and audiovisual expression of a topic, etc.).

Every schema of definition which is selected or merely selectable to furnish specifications about a sequence is itself part of a library of conceptual schemas of the universe of discourse of the CCA archives. This library defines generic microconfigurations either of objects of analysis* or of activities of analysis*.

A schema of definition in itself is characterized either by a conceptual term or by a combination of two or more conceptual terms and/or one or more referents. As we know, the conceptual terms are part of the meta-lexicon of conceptual terms representing the ASW domain of expertise. In Part Four of this book, we shall take a closer look at the rather complex organization of a model of description in the form of sequences, schemas, conceptual terms and referents.

5.5. Alternative functional architectures to define sequences of referential description

That said, formatting an interactive form as per the model shown in Figure 5.3 represents an important conceptual choice.

Indeed, using such a form obliges the analyst to describe the topical structure step-by-step, in an analytical manner – specifying each conceptual term which makes it up, one-by-one. As our example with the extract from the content analysis form (Chapter 3, Figure 3.2) shows, the analyst first describes freely, in the first subsequence, the conceptual term [Civilization], indexing it with the expression <Chavín civilization>; then, in the second sub-sequence, he describes the conceptual term [Cultural construct] in a controlled manner, using a specialized thesaurus.

Figure 5.5, on the other hand, shows the functional architecture of an alternative model to the one whose interface is shown in Figure 3.2. While it keeps the global sequence (reserved for the referential description of an audiovisual text) distinct, the architecture shown in Figure 5.5 reserves:

– the first sub-sequence for the analyst to select the appropriate conceptual term(s) to define the subject or topic he is attempting to describe in the audiovisual text that he is analyzing; and

– the second sub-sequence for the procedures of description (free and/or controlled) per se of the configuration of conceptual terms selected.

Figure 5.6, in turn, shows the interface which is a realization of the alternative functional architecture represented in Figure 5.5 and which also takes account (as does that shown in Figure 5.3) of the generic configuration defining the part of the domain of knowledge Civilizations and their cultures (see Figure 5.2).

This alternative interface is indeed far more powerful than that shown in Figure 3.2 but also requires more “technical” knowledge from the analyst, more ease when manipulating metalinguistic concepts.

While the interface in Figure 3.2 only requires the analyst to actually describe the conceptual term, the interface shown in Figure 5.6 actually requires the analyst:

– to first configure his domain of analysis or expertise,

– before actually describing it.

The example shown in Figure 5.6 is still fairly simple: it shows two conceptual terms which already make up a very restrictive generic configuration defining (part of) the topical structure for thematizing a knowledge object in the text. In this particular case, the analyst can select either one conceptual term or the other, or indeed both at once.

The analyst (where he selects both conceptual terms), by doing so, in the later case, signals that the topic he has identified and described in the audiovisual text relates to the actual configuration of the two conceptual terms, more than to just one or the other conceptual terms taken in isolation. In our example (Figure 5.6), the analyst decides that the freely-entered verbal expression <Chavín religious construct> indexes the two conceptual terms [Civilization] and [Cultural construct] together!

Of course, the analyst may select only one term – either [Civilization] or [Cultural construct]. This being the case, he carries out a project of description similar to that which he carries out using the interface shown in Figure 3.2.

Bearing in mind that this is a very simple example (i.e. a definitional configuration made up of only two conceptual terms), one can fairly easily see that allowing the analyst (a certain degree of) freedom to configure his knowledge object (that is, to select from a set of conceptual terms a sub-set of pertinent terms, or even a single conceptual term), opens up very interesting possibilities in the development of content analysis forms for audiovisual collections with very varied themes.

More particularly, here, we think of the audiovisual collection which constitutes the Audiovisual Research Archives Program.⁵ This collection covers a wide variety of disciplines in human and social sciences. In this context, it is difficult to envisage the elaboration of interactive analytical forms similar to that shown in Figure 3.2. This would certainly involve the development of hundreds of forms which would be difficult to control by the different parts involved in the process of modeling (and monitoring) of forms, analysis of audiovisual corpora and publication of these analyses in the form of thematic or special access portals.

Hence, the possibility mentioned above (of displaying reasoned sets of conceptual terms in an interactive working form and allowing the analyst to select those which are pertinent for his work) seems to constitute one possible solution to drastically limit the number of interactive forms dedicated to analysis of audiovisual content. At any rate, this is the path we chose to follow, developing a library of basic forms for analysis of the ARA collection’s audiovisual content.

Let us return to our example from Figure 5.6. We said that the analyst could define his knowledge object by jointly selecting [Civilization] and [Cultural construct]. However, he may also content himself with choosing only one conceptual term – either [Civilization] or [Cultural construct]. By opting for one or other of these two solutions, the analyst signals that, depending on the situation, he is carrying out either a description of a civilization without reference to a particular cultural construct, or a description of a cultural construct without reference to a particular civilization.

In terms of modeling the referential domain of knowledge of the universe of discourse of an archive (in our case, that of the CCA archives), this may, however, pose problems.

By selecting only the conceptual term [Civilization] or only [Cultural construct], the analyst essentially throws off the very specific structure which defines the knowledge object of the form shown in Figure 3.2, i.e. referential description of subjects and topics which deal with (native American) civilizations in relation to a particular cultural construct.

Thus, a priori, the analyst can use the form shown in Figure 4.2 to describe any audiovisual text which is about civilizations (on the American continent) “as such” and with no relation to a cultural construct – or indeed in relation to completely different knowledge objects from those for which the form in question is intended to be used, and for the analysis of which there may be other forms in the library of forms for analyzing audiovisual content.

Similarly, the analyst may opt to use the aforementioned form to define just a cultural construct (on the American continent) without relation to a particular civilization. For instance, he may use the form to describe an audiovisual text which thematizes technical culture in a certain era in this-or-that region of the American continent.

This ambiguity is common to the functional architecture underlying the form shown in Figure 5.6 and that underlying the form shown in Figure 3.2. If need be, it may be got rid of by further specifying the particular requirements of a configuration defining a knowledge object such as that which represents the object Civilizations and their cultures on the American continent (Figure 5.2).

Thus, in Figure 5.2, nothing is said if, for instance, one or other or indeed both conceptual terms are needed to define the object in question, nothing is said if one of the two terms depends “critically” on the other (in the sense that the dependent term cannot be selected unless the term on which it depends is already asserted or selected), nor is anything said if the choice of this term or that precludes such-andsuch another term, and so on.

All this stems from conceptual analysis of the domain of reference of the universe of discourse of an archive. The generic configuration such as it is presented in Figure 5.2 is “toothless” in relation to all these possible constraints. In its current state, it does not provide the necessary guidance to be able to decide on the most optimal functional architecture of the interactive working form which is supposed to “embody” the model of description underlying it as precisely as possible.

In Chapter 9, we shall return to this highly important question of a more restrictive conceptual model on a domain of knowledge, which takes account of the requirements listed above: dependence of a (set of) conceptual term(s) on another (set of) conceptual term(s), whether or not the choice of a conceptual term is exclusive, etc.

---

1 See http://semiolive.ext.msh-paris.fr/arc/.

2 http://semiolive.ext.msh-paris.fr/arc/.

3 OntoEditor is a tool designed and developed at ESCoM by Francis Lemaitre as part of various R&D projects – both European (particularly the LOGOS project) and French (particularly the SAPHIR and ASW-HSS projects). Today this tool is used to specify and create ASW metalinguistic resources which, as previously stated, are made up of: two metalexicons of conceptual terms which represent the objects and the activities of analysis respectively; a thesaurus comprising a set of facets and several thousand predefined expressions; blocks of models of description (in the form of functionally specialized schemas of definition and sequences of description); models of description; and, finally, libraries of models of description. We shall give a brief presentation of this in Chapter 11.

4 We can clearly see here that the production and monitoring of descriptive models can become a professional and economic activity of crucial important in the knowledge industry and economy.

5 An R&D program we set up at ESCoM in 2001, devoted to scientific, technical and practical issues relating to the constitution, analysis, publication, exploitation and conservation of bodies of scientific and cultural heritage using information and audiovisual technologies (for further information, see [STO 03c; STO 07; STO 11a]); the portal of the ARA (AAR) program is: http://www.archivesaudiovisuelles.fr/.