Jean-Claude Gardin and the Evolution of Archaeological Computing1
À l’occasion de la conférence « Computer Applications in Archaeology » (CAA 2014) qui s’est tenue à Paris, un hommage à Jean-Claude Gardin (Paris, 1925-2013) a été considéré plus qu’approprié. Avec son parcours académique éclectique, ce chercheur de réputation internationale a su dès les années 1950 aborder l’usage de techniques novatrices pour l’automatisation des processus de description et de classification des matériaux archéologiques – que ceux-ci soient des objets, des thèmes iconographiques ou des textes anciens. Dans le cadre de ses activités scientifiques, J-Claude Gardin a fondé et dirigé une série de laboratoires de recherches spécialisées, dans lesquels plusieurs projets de recherche ont été conçus, des nouvelles méthodes testées, et des événements internationaux promus. En retraçant son rôle exemplaire dans les premières phases des applications informatiques en archéologie, nous pouvons aujourd’hui explorer quelques-uns des aspects théoriques qui sous-tendent l’évolution de cette discipline, au-delà ses dimensions technologiques.
Index by keyword :archaeology, archaeological computing, history of computer applications, Jean-Claude Gardin
- 1 This paper is a revised and extended version of the inaugural lecture presented at the 2014 Paris e (...)
La naissance de l’informatique en archéologie est aujourd’hui une application très ancienne, qui peut remonter jusqu’aux années 1950. On trouverait ici des cas de toutes sortes, antérieurs même aux emplois les plus élevés du genre, touchant l’apparition du calcul en archéologie (Gardin 2009).
1Recognised worldwide as one of the promoters of the birth of archaeological computing – indeed we might say the founder – Jean-Claude Gardin (Paris 1925-2013) actively participated in the international cultural ferment that was spreading in the Fifties around the new methods of automatic processing and retrieval of scientific information and laid the foundations for the development of computer applications in the Humanities in general, and in Archaeology in particular. As early as in 1962, synthesising his experience, he could write: «Information automatique et archéologie, la juxtaposition peut paraître surprenante; son intérêt pourtant est de souligner la généralité des recherches sur le “traitement automatique de l’information” puisqu’aussi bien l’archéologie, discipline réputée conservatrice, en épreuve aujourd’hui le souci...» (Gardin 1962, 25).
2From the second half of the Fifties to the early Sixties, he contributed to the initiatives promoted by leading World and European organizations, such as UNESCO and EURATOM. He received funding from the Rockefeller Foundation to disseminate internationally the new methods of documentation that he was developing. He founded and directed scientific research laboratories in France, starting in 1957 with the CNRS Centre Mécanographique de Documentation Archéologique, which allowed him to cooperate closely with renowned French scholars such as Henri Seyrig, Fernand Braudel, Jean Leclant, André Leroi-Gourhan and Claude Lévi-Strauss, to name just a few.
3A recent research work devoted to the history of archaeological computing gave me the opportunity of investigating the first years of Gardin’s activity (Moscati 2013, with references). The consultation of the Fonds Équipe Archéologie de l’Asie Centrale et Jean-Claude Gardin (from now on JCG), preserved in Nanterre in the Archives of the Maison Archéologie & Ethnologie, René-Ginouvès, has greatly contributed to the reconstruction of a detailed picture of his pioneering impulse and tireless commitment in promoting innovative research paths, in establishing and leading ad hoc laboratories and in organising international scientific events. By focusing on these early years of cultural impulse, it is possible to highlight the most innovative landmarks of Gardin’s thought, which underlie the logicist programme he would promote in the forthcoming years and to which scholars generally associate his theoretical production.
4The earliest documents kept in the Archives refer to the establishment in 1957 of the Centre Mécanographique de Documentation Archéologique under the auspices of the French CNRS. This initiative stemmed from a previous research programme strictly related to a CNRS mission in 1955 to the Institut Français d’Archéologie de Beyrouth, which consisted of three members, including Gardin himself. The mission’s mandate was to deploy a public archive of punched cards describing different kinds of archaeological objects and making documentary research more operational. This was the beginning of a particularly fertile period that surely influenced the following phases of Gardin’s studies, but that was in turn influenced by his eclectic academic background that affected his theoretical and methodological approach.
5Perplexed about the repetitiveness of the archaeological cataloguing procedures – he speaks of “monotony” – and aware of the need to adopt new documentation strategies to speed up the acquisition of factual data and organise bibliographic work, in Beirut Gardin found Henri Seyrig’s support in the application of mechanical techniques for the storage and retrieval of data, with the purpose of facilitating the sorting and searching tasks that are at the basis of any archaeological classification (Gardin, 1955; for a recollection of this period, see also Gardin, 1996 and Gardin, 1997). At that time, he had already realised that documentary systems would have resulted in a more complex theoretical research activity than the simple collection and search of bibliographic data. A theoretical activity that is set in motion «lorsque l’ensemble de données mises en mémoire est soumis à une série d’opérations logico-mathématiques visant à mettre en évidence certains phénomènes d’“ordre” (modèles, classifications, etc.) jusqu’alors inaperçus» (Gardin 1962). All this, thanks to the combined use of a descriptive language, methodically refined, and a mechanical punched card system allowing specialists to organise the elements of this language in any selected order, was envisaged with a view to detecting, on the basis of a wide range of interdependent observations, some underlying complex cultural phenomena (Gardin, 1960).
6In 1956 Gardin published the first file developed by the members of the mission in a brochure entitled Le fichier mécanographique de l’outillage, in order to describe metal tools of the Bronze Age discovered in a large area ranging from the Balkans to the Indus (Gardin, 1956). The earliest five “Codes” for the analysis of archaeological objects, which involved pottery, weapons and tools, Greek coins, oriental seals and iconographic themes, were also created in 1956. These analytical codes, which are stored in the Archives with the related documentation (JCG 157-161), were not published until a few decades later (Christophe, Deshayes, 1964; Digard et al., 1975; Le Rider, 1975; Gardin, 1976; Gardin, 1978).
7The main task of the Centre was to develop the institution of international archaeological corpora, combining the formula of printed catalogues with that of mechanic files, which turned out to be easy to update as opposed to the traditional edition of supplements. They were also far more flexible, as they were oriented towards an objective formalised description of archaeological artefacts, without taking into account the most complex conceptual constructions useful for historical interpretation. On reading this research programme more than fifty years later, one finds the seeds of those aspects of description and formalisation of archaeological information that are at the core of the subsequent development of archaeological data banks, which Gardin himself shall define in the years to come as a systematic gathering of all kinds of archaeological information in a form suited to their retrieval and scientific processing (Gardin, 1974). A programme of this kind also rested on a clear distinction between the descriptive and the interpretative parts of archaeological scientific constructions, which will feature many of Gardin’s speculations and, more in general, the early debate on the selection of archaeological information to be submitted to computer analysis (for a general overview, see Moscati, 1994).
8The method was based upon a coding procedure allowing for the segmentation and description of the single component of each object and of their relationships. It could therefore be standardised and applied to different objects and to various levels of descriptive complexity. Through this formalised description, its transcription on punched cards and the use of a device of automatic selection, it was possible to obtain a sum of «virtual catalogues» (Gardin, 1956, 16) or «potential classifications» (Gardin, 1960, 9), which could be multiplied at will. This is undoubtedly a striking anticipation of the concept of “virtual representation”, which overwhelmingly came into use in the archaeological terminology since the so-called computer graphics revolution began to affect the visual reconstruction process of archaeological features. Gardin’s approach, instead, leads back to the concept inherited from the scholastic Latin – virtualis as opposed to formalis – i.e. of something that only potentially exists, but that has the virtus to be actualised through the construction of logical models, whose effectiveness and usefulness is determined by their ability to approximate the “real world” and recreate cultural processes.
9According to Gardin, attention should be placed on the method and its theoretical logic implications, regardless of the underlying physical system. This was a key concept for Gardin who, in 1971, wrote to René Ginouvès – another French protagonist of the application of computers in classical archaeology – that since 1955 he had realised that the comparative merits of such or such a model of punched cards or machines was less of a concern as opposed to the methods of data representation and reasoning specific to archaeology, from the perspective of a “mechanization” conceived without reference to any of these cards or machines (JCG 205, letter of 12 January 1971). In the following years, he reiterated this concept, by adding that, since the beginning, his primary concern had been «the study of mental processes at work in archaeological reasoning, with a view to making them amenable to machine handling in a Turing sense – that is, with or without computers» (Gardin, 1989a, 7).
10This is one of the reasons that in 1958 led Gardin to propose a new title for the Centre: Centre d’Analyse Documentaire pour l’Archéologie (CADA), placing in this way greater emphasis on the method rather than on the techniques (Centre d’Analyse, 1959; Gardin, 1966). Gardin pointed out this change of viewpoint in several occasions, both from a strategic and a methodological perspective. In particular, in the Archives of the Bibliothèque Claude-Lévi-Strauss at the Laboratoire d’Anthropologie Sociale in Paris (Archives du LAS (1960-1982). Direction Lévi-Strauss, Dossier Jean-Claude Gardin, fr/cdf/las/FLAS.D.04.05, from now on Dossier Jean-Claude Gardin), some interesting notes and documents can be found regarding the relationships between the Laboratoire, the CNRS Section d’Automatique Documentaire and CADA. In a letter to Claude Lévi-Strauss, who was one of the members of the Centre’s Scientific Committee, Gardin underlined that he had been asked to modify the title of the Centre in order to make more simple and less ambiguous a choice for those times – in other words to make the reception of the new laboratory and its methods more immediate and appealing to the promoting institutions involved (Dossier Jean-Claude Gardin, n. 024, letter of 16 March 1959). In a subsequent letter to René Ginouvès, he traced this choice back to a personal desire, to demonstrate his reduced interest in the physical equipment as opposed to the analytical methods (JCG 205, cit.).
11In his early methodological works, some of which were published in America (Gardin, 1958, 1959), Gardin devoted much attention to the identification of the steps required to set up a file of punched cards: a) development of the analytical code; b) editing of the descriptions; c) perforation of the punched cards; d) optionally, duplication of the cards, in view of making them fully or partially available. The new-born Centre would be responsible for overseeing all four phases. However, Gardin’s attention was mainly directed to the first phase, the most sensitive one, which requires a “tournure d’esprit logique” and involves an in-depth knowledge of the subject matter and the acquisition of a rigorous logic that allows for the maximum amount of information to be expressed with the minimum amount of terms. Data encoding, in the sense of data representation according to a set of rules and a system of symbolic designation, is therefore the cornerstone of the entire procedure, since it constitutes the essential preliminary stage for entering data into the process of documentary analysis, either mechanically or manually, and for correctly representing the complexity of the system under analysis. All mental operations typical of documentary research were assembled by Gardin under one single label: metalanguages (Gardin, 1981, in particular chap. 3). They have three main components: a lexicon of terms for indexing purposes, a semantic organisation of these terms and a grammar indicating their syntactic relations (Gardin, 1997, 68).
12In this respect, the words often used by Gardin in describing his own method (see e.g Gardin, 1959) allow us today to better comprehend his classificatory approach as a way to breaking down the information into various elements, and in expressing in different ways the relations according to which they are assembled. By fragmenting the continuous representation of objects in a discrete series of characteristic elements, Gardin moved towards a classification approach based on a “relational” structure, overcoming the cataloguing process centred on objects and promoting a representation of data whose uniqueness is made explicit through a specific combination of words that speak of both the constitutional elements and their relationships. By following Gradin’s lines of thought, we can suggestively close our eyes and have the feeling that we are listening to a modern advocate of data standardisation and knowledge sharing in a semantic web environment.
13In a typewritten document kept in the Archives of the Bibliothèque Claude-Lévi-Strauss (Dossier Jean-Claude Gardin, Trois projets d’analyse structurale intéressant des phénomènes humains), Gardin clearly claimed that automatic analysis, once refined, could become a method of structural analysis allowing the detection, in the general organisation of human phenomena, of a series of laws similar, by nature, to those that linguistics have used since about twenty years, through the application of the same methods. Indeed, all descriptive formulas should be provided with rigor, conciseness and objectivity, all qualities that make them useful as material for a precise formal study, theoretically appealing to topology and practically implying the use of machines.
14In the early Sixties, the Centre’s activity rested mostly on three main pillars: objects, iconography and texts. We will not dwell here on his work on textual analysis, although it has accounted for a large part of Gardin’s initial interests. Whether it carries out the analysis of clay tablets from the ancient Near East (Christophe, Digard, Gardin, 1958; see also infra) or of a religious text as the Quran (Allard et al., 1963; see also Gardin, 1989b), or still of the myths of the Zuni Indians (Dossier Jean-Claude Gardin, L’analyse structurale des myths: ébauche d’une méthode), the study does not directly address the literal form of the document but its semantic content. This is why Gardin argued that the exegeses carried out at the CADA differed from the research works undertaken with the help of punched cards and focused on the philological rather than the semantic aspect of some texts, e.g. the collected works of St. Thomas Aquinas analysed by Father Roberto Busa (Gardin, 1960; Dossier Jean-Claude Gardin, L’analyse sémantique et la mécanographie).
15In the study of the morphology and the iconographic repertoire of archaeological artefacts, what is surprising is the utter modernity of Gardin’s scientific approach so much so that some of his early studies gave birth to archaeological computer projects in the years that followed. An exemplary case study can be quoted that shows the growing interest of the French academic environment in the new methods. In 1958, in accordance with the Musée du Louvre, a project was launched which was designed to set up a mechanical index of the figurative scenes on Greek vases, starting from those published in the Corpus Vasorum Antiquorum. In the archival documents, Lilly Kahil, who in the early Seventies promoted the editorial initiative of the Lexicon Iconographicum Mythologiae Classicae, is already referred to as one of the most authoritative scholars for this task (JCG 1/1, 28 October 1958).
16However, what is most striking about Gardin’s approach is his lucid and insightful analysis of the new techniques that science had made available for humanists in general and for archaeologists in particular. The rigorous formalisation of archaeological description, as well as the use of a minimum set of core variables able to uniquely describe any part of a more complex system, led Gardin directly to the first computer applications in archaeological research, making him internationally recognised as a pioneer (Cowgill, 1967). Using an IBM 650, two case studies were promoted with the help of EURATOM’s technicians and machines: the first consisted in the experimentation of a deductive method for the automatic classification of a corpus of Eurasian axes of the Bronze Age, while the second concerned the study of the commercial network established by Assyrian merchants in Cappadocia at the beginning of the second Millennium BC (Gardin, Garelli, 1961; Ihm, 1961; Ihm, 1965).
17These two pioneering projects are intertwined with the institutional commitment of Gardin, who at the beginning of the Sixties founded the Groupe d’études Documentaires pour les Sciences Humaines, working in close collaboration with Paul Braffort and his Groupe de Recherche sur l’Information Scientifique Automatique. In this context, Gardin acted as one of the architects of the Syntol system, not merely a documentary language but a general model for information retrieval (Cros, Gardin, Lévy, 1964; Le Syntol, 1964; see also Gardin, de Grolier, Levéry, 1964). The EURATOM bibliographic Fund, preserved at the Biblioteca dell’Accademia Nazionale dei Lincei e Corsiniana in Rome, makes it possible to enucleate, from the most extensive body of scientific works, a number of titles on documentation systems and machine translation, focusing on what Braffort and Gardin called the “problèmes non numériques” (see in particular Braffort, Fonzi, 1963; Braffort et al., 1964).
18In this supranational context, cooperation between Humanities scholars on the one hand and engineers and mathematicians on the other was greatly encouraged. In this regard, Gardin clearly stated that this interdisciplinary collaboration would not be successful unless the former party became knowledgeable at least in general terms about the outlook of the latter party and vice versa. In a sort of role play and with continuous alternations, he pointed out that it was the duty of the historian to first become an “expert” in his own field of reasoning – something that is unlikely or hardly likely for computer scientists – and to modestly abandon to the latter the expertise in programming – something that the historian, even if a programmer, is hardly or never provided with.
19This is a position common to other Mediterranean archaeologists, who in those days were witnessing the rising of new technologies and, therefore, the convergence of the “two cultures” (Donato, 1969). For example, the famous Italian Etruscologist Massimo Pallottino, who at the beginning of the Sixties was committed, along with other Italian scholars, to establishing new centres of archaeological research promoted by the National Research Council, wrote that investigation tools, whatever their nature, should always remain ‘tools’ with respect to the purpose of the historical interpretation, because the interpreter of the discoveries, whatever his education, cannot effectively and definitely fulfil his task except in the capacity of and through the mindset of an archaeologist and historian (Pallottino, 1963, 113).
20But what is the real shift that Gardin perceived in the transition towards computer tools? In terms of research planning and the need for a formalised description of data, Gardin detected no differences except for the amount of data that could be analysed. However, he focused on some aspects that would help to overcome the results of documentary research by suggesting new opportunities in data interpretation. Computers were suitable for sorting procedures, which are distinctive of a documentary approach, as well as for more complex “ordinations”, by the combined use of mathematical methods and automatic calculation procedures, according to the two lines of investigation that Gardin defined as “classifications” and “networks”.
21The use of computers is justified not only by the size of the data matrix, but also by the capacity to perform multiple operations at the same time, and with a single “programme”, combining the unintelligent phases of the analysis – i.e. simple data tabulations – with the intelligent ones, namely the inferences that can be drawn and then incorporated in the computer programme, in the form of conditional statements. Therefore, Gardin was already well aware of the fact that, as was later claimed by Tito Orlandi (1996), the proper use of the computer, as a physical machine that allows researchers to perform logical processes on data, requires a full knowledge of the theoretical aspects underpinning the same machine.
22Here, therefore, is the substance of what would animate Gardin’s approach throughout all his life, and which was already clearly outlined in his main theoretical work at the end of the Seventies: “Une archéologie théorique” (Gardin, 1979; Gardin, 1980). In fact, a spiral cycle is at the heart of his “practical epistemology”, according to which the phases of data analysis and interpretation determine a new formalisation of the same data, whose results are to be included in the subsequent analyses. A spiral, where validation procedures constantly dash the hopes that the structural “units” of the subject matter under study have been reached. Any other scheme, using Gardin’s words, would mean interrupting and putting the scientific research to an end.
23In conclusion, I would like to close with few but significant and auspicious words by Gardin, which he pronounced during his inaugural speech at the Symposium “I modelli nella ricerca archeologica” held in Rome in 2000 at the Accademia Nazionale dei Lincei (I modelli, 2003), still recorded in an audio file. By prefiguring the final advent of knowledge-based logical constructs in historical and archaeological sciences, he reaffirms some concepts that have been characteristic of his approach since the years of his youth:
«Pendant encore quelques décennies on va assister à la coexistence, que j’espère pacifique, des modèles discursifs de type logique et des constructions narratives qui cachent cette logique sous-jacente sous un discours naturel. Les deux types de constructions vont certainement cohabiter pendant longtemps, les premières extrêmement minoritaires aujourd’hui ... et les autres qui constituent le flot dominant des constructions historiques dans la littérature de nos disciplines. Mais ce que je crois, c’est que les premières vont petit à petit – je ne veux pas employer un terme trop fort – éliminer ou peut-être réduire le nombre des secondes. Je pense que les constructions cognitives fortes, avec ou sans nouvelles technologies, mais probablement avec, vont petit à petit, en tout cas, poser des questions qu’on n’avait pas l’habitude de poser avant sur les vertus cognitives des constructions narratives stricto sensu. Je laisse le débat ouvert ...».
1 This paper is a revised and extended version of the inaugural lecture presented at the 2014 Paris edition of the international Conference Computer Applications and Quantitative Methods in Archaeology. I am grateful to my colleagues in the Organizing Committee and, in particular, to François Djindjian, for giving me the honour of recalling some aspects of the life and work of Jean-Claude Gardin in the Amphithéâtre Richelieu at the Université de Paris 1 Panthéon-Sorbonne. Since the earliest years of my academic career, I had the privilege to benefit from Gardin’s wise guidance and unquestioning generosity. This is the reason why I was particularly grateful to retrace the initial stages of his scientific activity, to show younger generations how modern he was in his contribution to the formation of a discipline such as archaeological computing, not so much for its technological aspects but for its theoretical background promoting the development of formalised procedures that were to have a profound impact on traditional archaeological research methods.Haut de page
Pour citer cet article
Paola Moscati, « Jean-Claude Gardin and the Evolution of Archaeological Computing », Les nouvelles de l'archéologie, 144 | 2016, 10-13.
Paola Moscati, « Jean-Claude Gardin and the Evolution of Archaeological Computing », Les nouvelles de l'archéologie [En ligne], 144 | 2016, mis en ligne le 25 août 2016, consulté le 28 mars 2017. URL : http://nda.revues.org/3457 ; DOI : 10.4000/nda.3457Haut de page
© FMSHHaut de page