This paper identifies the main concepts involved in the study of epigraphy and proposes the use of CIDOC CRM to encode epigraphic concepts and to model the scientific process of investigation related to the study of epigraphy. After analysing the existing CIDOC CRM entities and those provided by the CRMsci and CRMarchaeo extensions, we introduce more specific epigraphic classes to be used as the basis for creating a new extension, CRMepi, which is more responsive to the specific needs of this discipline.
Material sources bequeathed to us from antiquity represent an immense treasury of knowledge of a lost world. In addition to literary sources, numismatics, epigraphy and archaeology provide important information we would not otherwise possess, and often the objects they document constitute the only evidence we have of an ancient population. In particular, among these sources, inscriptions are particularly essential because they represent the most direct and resonant voice of the ancient people that have handed them down to us, as deathless words in stone. Generally, the preserved inscriptions were originally conceived to endure over time and to be transmitted to the future. For these reasons, these sources need to be digitised and integrated in some way, along with other cultural heritage information. Inscriptions are complex objects, and their study requires careful analysis from different points of view.
What characterises this class of objects is that they form a whole with their physical support. Indeed the meaning of an epigraph cannot be fully understood without the analysis of the object or monument or other archaeological object on which it appears, just as one cannot fully understand the nature of that particular archaeological object without thoroughly investigating the sense of the inscription or iconographic representation it hosts.
Furthermore, the inscription itself is, from a conceptual point of view, an element with physical characteristics that are themselves bearers of meaning and of valuable information going far beyond the inherent meaning of the text. The shape of the letters, their spacing, the direction, technique and other similar characteristics provide precious clues to the times, makers and functions of the inscription in question.
The tools available today seem to be insufficiently flexible and efficient to
implement a comprehensive and useful level of digitisation. Relational databases,
for instance, are not fully functional for this goal because of the rigid structure
they provide, which is quite inadequate to describe textual entities and their
possible annotations. Databases are even less suitable to describe the complex
web of relationships that links archaeological objects with the inscriptions they
bear and their meaning. Since the beginning of the studies on digitisation of
texts and other unstructured data, the XML family of technologies emerged as
the tool that could free the information from the rigid structures of relational
databases built around tables and records. Epigraphy has also largely benefitted
from XML. It is sufficient here to say that EpiDoc [
Although epigraphy is a scientific discipline with a centuries-old tradition, a
single and fully accepted definition of its object, the epigraph (or inscription), has
not yet been formulated. The definition most commonly found in manuals of the
discipline defines the inscription more or less as a direct evidence of the past
inscribed on stone or other durable materials. Guarducci, in particular, puts
emphasis on the materiality of the support, which, among other things, makes
it possible to distinguish the object of epigraphy, as opposed to papyrology. She
identifies what epigraphy is concerned with, as opposed to papyrology, and she
identifies two particular characteristics of epigraphic documentation that
differentiate it from other historical primary sources: uniqueness and authenticity [
The problem originates from the plethora of tools, techniques and purposes that can be employed in order to characterise an inscription. These constitute a complex and intertwined series of elements that do not, however, suggest a comprehensive definition. We must also consider that different cultures have given rise to different traditions of writing and thus also of epigraphy. Consequently, a consistent and comprehensive definition of the term “inscription” should consider not only epigraphic products from the major Western traditions (Latin and Greek), but also those relevant to “minor” traditions (e.g., the languages of ancient Italy) and non-Western ones (e.g., the heterogeneous legacy of the Semitic world). In addition, we should consider that there is an ambiguity of reference with respect to the term “epigraph”, which can be used to indicate both the plain text and the combination of text and physical support, especially in those cases in which the support has been created expressly for bearing the text; in fact, inscriptions may also appear on artefacts made for different purposes (such as vessels) or even on non-artefacts or on natural surfaces (caves, cliffs, etc.). It is obviously not up to us to say the last word about this topic, or to define what epigraphy is and what distinguishes it, for example, from papyrology and other textual studies. Nevertheless, it is certainly very difficult to find a comprehensive ontological definition for classes of objects that have been from time to time assigned to one category or another, mainly depending on the discipline involved in their study. 2.2
From a logical point of view, and in accordance with the tradition of epigraphic
studies, an inscription can be analysed according to three main aspects: the
text-bearing object or monument (obviously involving archaeological topics),
the text (and the obvious correlations with content and linguistic aspects lato
sensu) and the feature engraved on the support in the form of letters or other
symbols, which is the central element that characterizes and differentiates an
epigraph from any other manifestation of written communication. Whereas in
documents on papyrus or parchment (et similia), the materiality of the features
is not relevant in comparison with the morphological and typological
characteristics of the handwriting, investigated by palaeography, they are essential
in inscriptions and their analysis precedes and prepares palaeographic studies.
Such features represent the peculiar object of epigraphic studies, and analysis of
them is as fundamental as that of the archaeological, palaeographic, linguistic
and historical aspects. Paramount importance must be ascribed to the
communicative purpose of the epigraphic text, since this is exactly what distinguishes
an inscription from any other feature of any kind that may be present on a
given support; in other words, given the intentionality of a feature, a figurative
decoration is distinguished from an inscription by both semiotic and linguistic
purposes (the explicit will of communicating a message), on which a different
communicative-informational structure also depends. According to a semiotic
analysis, an intentional feature (i.e., one voluntarily created by man to convey
a message) occurring on a given support can appear in the following forms (see
[
We can talk of inscriptions in the latter three cases, i.e., when we recognise signs belonging to certain writing systems; nevertheless, we can talk about written communication only in the presence of signs used per se as encoders of linguistic signs (and structures), and therefore, of a text.
Finally, we can summarize as follows: – not-glottographic feature, a figurative decoration, but also a sign of pure semasiographic systems of writing – glottographic feature, not necessary codifying a linguistic expression, since a sign can be used with different purposes than a linguistic one 3
The edition of ancient texts boasts one of the earliest and more consistent
systems of standardisation in the field of Humanities: the Leiden Conventions [
However, with the advent of the digital era, epigraphists had to face a set of problems very similar to the ones brilliantly solved by the Leiden Conventions. An electronic format that could allow digital publishing, storage and exchange of epigraphic information in a consistent and shared format was needed. From this need arose EpiDoc, a collaborative format designed to transcode in digital format the Leiden-encoded printed editions. The initiative began in the 1990s in response to the request for a free and unrestricted set of tools supporting the creation of online epigraphic archives, which was expressed during the same period in the course of a series of conferences on epigraphy and IT. The XML format was identified as being the most suitable for this purpose and the first EpiDoc DTD was released and quickly adopted by a relatively wide community of researchers. Basically, EpiDoc is an application profile of TEI specifically adapted to the needs of epigraphy. This profile has extensively evolved from its first draft. As of today, EpiDoc provides features for the recording of the materiality and history of text-bearing objects, as well as features for scholarly editions of the text, such as commentary, illustrations, bibliography, and publication data. EpiDoc also offers facilities for the detailed description and editorial representation of the texts themselves, including transcription in the technical sense of reporting readings and representing the writing system, form, appearance, layout and editorial interventions in the text.
The EpiDoc system, despite its undoubted merits, still presents some issues, especially with respect to the inline text encoding features, arising from the fact that there are no native tools fully able to support the EpiDoc format for sessions of text editing, and thus to simplify the encoding operations. Essentially, EpiDoc-based text edition still remains a manual task, which greatly complicates the digitisation of large corpora of inscriptions. From a technological point of view, the choice of reproducing the paper publication format by means of XML mark-ups also raises an issue related to the style sheet necessary for the optimal rendering of the XML encoded text and therefore inseparable from it. This could represent a further portability issue for the sharing of the information from one archive to another and on the Web. EpiDoc is also unable to guarantee the typical “relational” features offered by a database, since it is lacking in all the paraphernalia necessary to describe the complex web of relationships that characterize the various aspects of epigraphy. Only ontologies and similar semantic tools seem to be able to merge the advantages and flexibility typical of XML with the characteristic “relationality” of databases. 4 4.1
In past years, some attempts have been made to use CIDOC CRM for the
description of epigraphic entities. One of the first such projects was
VBI-ERATLVPA [
More recently, some methodological proposals have been put forward to
combine EpiDoc and CIDOC CRM features and to harmonize the features they
provide [
As mentioned, the purpose of this work is to lay the groundwork for a possible epigraphic extension of CIDOC CRM (CRMepi) through the conceptual analysis of the specific entities and problems with which epigraphy is concerned. We are aware that the lack of a comprehensive conceptual definition of the identity of an epigraph obviously complicates the formal definition of epigraphic entities using tools provided by ontologies and conceptual models at our disposal. Some general observations may however be made and the conceptual model of the CIDOC CRM can certainly be used as an intellectual guide to reasoning on the concepts involved in this process (see Figure 1). 4.2
In the case of epigraphy, an essential element that emerges from the above
discussion is the close relationship that tightly binds the inscription with its physical
support. This close cohesion between the support and the text, as already
mentioned, is, for instance, what distinguishes an epigraph from a papyrus, the study
of which mainly concerns textual analysis. For some inscriptions, however, the
shape, the materials, the production techniques and all the attributes of the
physical object that hosts the epigraph can become fundamental not only for
their understanding but also for the definition of their nature. If we focus on
the support on which the inscription was engraved, we note that CIDOC CRM
offers plenty of concepts with which we could describe it. The physical support
is in fact very often an archaeological object, a class of objects which has
frequently been investigated in a CIDOC CRM perspective. In terms of integration
and interoperability it is also important to note that, thanks to its nature, the
support constitutes one of the main points of contact between epigraphy and
archaeology. The specific archaeological aspects (discovery, provenance,
archaeological context etc.) relating to the physical support can be documented using
the CRMarchaeo extension [
In relation to epigraphy, it should be noted that very often the physical support has been designed and built specifically to accommodate the inscription. In this case the CIDOC CRM E84 Information Carrier entity seems definitely to be an optimal choice. However, this condition does not always occur: certain inscriptions may in fact even have been placed on objects not specifically designed to accommodate an inscription, as in the case of buildings, vessels or other objects of daily use on which the inscription may have been placed at a later time. In this case, the use of a more generic class, like E22 Man-Made Object, seems more appropriate. There are also cases in which the inscription is placed on natural surfaces not created by human activities, such as inscriptions on rocks, in caves or other similar natural places. The use of the superclass E19 Physical Object sufficiently broad so as to include every possible kind of physical support and would be in this case a more suitable choice. Each of these classes can still be linked with the physical features they bear, via the P56 bears feature property, having the E19 class as domain and thus being inherited by all its subclasses. The EpiDoc elements used to mark archaeological information concerning physical objects or monuments (such as the supportDesc, material, objectType and dimension tags) can easily be mapped using these CIDOC CRM entities. 4.3
CIDOC CRM provides a specific class to model the concept of inscription: E34 Inscription. The scope notes of this class state that “this class comprises recognisable, short texts attached to instances of E24 Physical Man-Made Thing ”. We need at this point to make sure that this class is consistent enough with the concept of inscription in the epigraphic sense, so as not to risk incurring conceptual ambiguities. Although many inscriptions bear short texts, the brevity or length of an inscription is not among its main characteristics. In fact, there are inscriptions occupying entire walls (the Gortyn Law Code or the Res Gestae Divi Augusti, for example) and in any case the “short text” of the E34 class remains too vague and undefined for the purposes of our investigation. The E34 class also belongs to the classes of conceptual objects which in turn are defined as “non-material products of our minds and other human produced data”, something that renders only in part the essence of an epigraph, not taking into any account its “materiality” which is a fundamental component of its identity. The study of epigraphy typically moves from the analysis of the physical characteristics of inscriptions before getting to their archaeological, palaeographic, linguistic and historical characteristics. In this sense, an inscription intended only as a conceptual object does not seem to capture fully the very nature of the epigraph itself. Thus, the etymology of the word “epigraph” indicates as a fundamental condition of its identity its being written on something. In all these ways it seems to present a much closer resemblance to the classes created for the description of physical features, and more specifically the E25 Man-Made Feature. We have managed to create some new and more appropriate classes to be used in documenting epigraphic concepts, and in particular: – EPI1 Epigraph. Subclass of E25 Man-Made Feature intended to describe a particular feature created by humans, in various ways and on various kinds of support, mostly rigid ones, with the declared purpose of conveying a specific message towards a given recipient or group of recipients. – EPI2 Engraving. Subclass of E12 Production indicating the activity of creating inscriptions in an epigraphic sense by using various techniques (painting, sculpture, graffiti etc.) and by means of specific tools on a given physical carrier. The definition of this activity allows us to make a better distinction between the creation of inscriptions and the production of the physical carriers that host them (two activities that are not always and not necessary contemporary), and to distinguish more accurately the creation of an epigraph from that of a story or poetry or other literary texts written, for example, on a papyrus. – EPI3 Epigraphic Field. Subclass of E25 Man-Made Feature. This represents another important element of epigraphy, usually understood as the surface or portion of the physical carrier reserved, delimited and arranged for the purpose of accommodating an inscription, to highlight it and to isolate it from the other parts of the object or building to which it belongs. There are various types of epigraphic fields, and among the most important of these epigraphists usually distinguish those created during the same production event of the carrier and those added to it at a later time. From a conceptual point of view, the epigraphic field is a feature designed to accommodate another feature (the inscription). EpiDoc also provides specific entities for the description of these elements (the tag layoutDesc for example) that can be easily mapped on this class. To define the relationship occurring between EPI3 Epigraphic Field and EPI1Epigraph, the new property EPP2 is included within has been proposed as a sub property of P56 bears feature, which is in turn a shortcut of the full path relating E19 Physical Object through P59 has section (is located on or within), E53 Place, P53 has former or current location (is former or current location of ) with E26 Physical Feature as described in the related scope notes. 4.4
As already noted, conceptually an epigraph significantly differs from the definition of the class E34 Inscription of CIDOC CRM. To avoid semantic and linguistic ambiguities we decided to use its superclass E33 Linguistic Object to describe the text (intended as a linguistic production), which the inscription records. Before proceeding any further, it would be appropriate to clarify the relationship between the epigraph, intended as a feature consisting of a set of signs, and its text as obtained through observation and decoding of those signs and the interpretation of the linguistic signs they refer to.
One of the most important operations carried out by epigraphists for their study, and especially for publication, is the so-called “reading” of the epigraph, consisting of a deep and accurate analysis and study of the surface and the signs followed by establishing as faithful as possible what is shown by the physical feature. This scientific process can be modelled by means of the concept of observation documented in the scientific extension of CIDOC CRM (CRMsci), and more precisely by means of the S4 Observation class. This class seems particularly appropriate to document this kind of scientific analysis. It could also avail itself of specific instrumentation to assist reading, such as microscopes or magnifying glasses, especially in case of inscriptions of reduced dimensions. It is also possible, thanks to this class, to document the different processes of analysis and study carried out over the years on a particular object by various scholars, and to report details on the tools and methods used. The observation (S4 ) class is used to define and identify (O16 observed values) the graphemes (i.e., the symbolic object E90 used to encode, on an abstract level, the linguistic units in the text) of which the engraved signs on an EPI1 Epigraph are the concrete graphical manifestation; this relation is made explicit by means of the P128 carries property. The opportunity to instantiate a more specific subclass of E90 Symbolic Object in order to provide a better description of these ideal graphemes requires a more thorough discussion.
The graphemes inferred from the observation of the epigraph represent the level of the intellectual decoding and understanding of the signs and constitute the basis for the subsequent operations of transcription usually carried out by epigraphists, in particular for so-called diplomatic transcription (i.e., a specific transcription recording only the characters as they appear on the support, without any editorial intervention or interpretation), which is also of great importance from the point of view of publication. The publishers of an epigraph, for practical reasons, generally those of typeface, perform these transcriptions using Latin or Greek characters, even in case of non-Latin and non-Greek inscriptions (Etruscan inscriptions, for example). To document this process we have created some specific classes: – EPI4 Transcription . Subclass of E7 Activity describing the specific operation of transliteration that, starting from the symbols observed on the epigraph (E90 Symbolic Object ->P16 was used for ), leads to the creation (P94 created ) of a set of instances of E73 Information Object recording the transcription(s) performed. – EPI5 Writing System . Subclass of E29 Design or Procedure, which refers to a specific sequence of characters (graphemes, E90 Symbolic Object ) used both to write and to transliterate the text of the epigraph (e.g. Latin letters). As mentioned, EpiDoc provides an entire section of tools for the encoding of the diplomatic transcriptions, implementing the Leiden Conventions in the form of specific inline XML tags designed to mark and describe each part of the text. An instance of the E62 String class, in association with one of the E73 Information Object (with type = diplomatic), might accommodate the EpiDoc-encoded text by means of a P3 has note property, to strengthen integration. The E33 Linguistic Object class, as we already mentioned, was used instead to represent the text of the inscription, i.e., the intended linguistic entity as resulting from an intellectual and linguistic process of creation, witnessed by the epigraph and also inferred from the various transcriptions (E73 Information Object ). The latter relation is established via the EPP2 has transcription property, a sub property of P130 shows features of. The connection between E33 Linguistic Object and EPI1 Epigraph is instead represented by using the P62 depicts property, which perfectly renders the nature of this relationship. It should be noted that the graphemes (E90 ) also remain in constant and close connection with the E33 Linguistic Object. We have used the P67 refers to property to describe this connection. In future developments, however, we shall assess the question of whether it is opportune to create a new sub property that better expresses this type of link.
To test and demonstrate the potentialities of the proposed model, we chose an
epigraph in the Oscan language from Pietrabbondante (Isernia, Italy), stored
at the Archaeological Museum of Naples (Italy). It is a cippus (MANN.2525)
bearing a dedicatory inscription (VE.150) [
The cippus was created specifically to host the epigraph. In this case, it seems more appropriate to use the E84 Information Carrier class rather than its E22 Man-Made Object superclass to encode it. The use of EPI1 Epigraph and the related production event (EPI2 Engraving ) allows us to distinguish the event of creation of the epigraph from that of the archaeological object, although in this case the two events happened simultaneously, since the inscription was in fact sculpted (P32 used general technique ->E55 Type = sculpture) contextually to the production of the cippus. The definition of the same E52 Time Span for both the events indicates their contemporaneity.
The E90 Symbolic Object is represented by a set of characters of the Oscan alphabet evidenced through the property P3.1 has type, a sub property of P3 has notes usually used to record specific notes concerning peculiar aspects of a given entity such as the writing system used, as happens in this case. The alphabet to which the graphemes of the E90 Symbolic Object belong is an element of capital importance, especially in the study of non-Latin and non-Greek inscriptions, which needs to be expressed in a richer and more specific way. We will consider the definition of adequate classes and properties for modelling this concept in future revisions of the model. Through analysis of the inscription (O4 Observation), in 1850 Theodor Mommsen provided a reading and a transcription of it (EPI4 Transcription ->P94 created ->E73 Information Object ) using the Latin alphabet (EPP3 used writing system ->EPI5 Writing System) to create both the diplomatic and interpretative transcriptions of the text. This same set of classes and properties can be instantiated several times in case of new or different readings, transcriptions and interpretations of the same epigraph by other scholars, in order to create a chain of events able to represent the history of studies of the object.
Diplomatic transcription constitutes the basis for the interpretative edition of the E33 Linguistic Object, i.e., the text intended as a linguistic production, encoded by means of a given writing system in the epigraph. The E33 Linguistic Object is therefore linked both to the E90 Symbolic Object (i.e. the Oscan graphemes, units of writing system on an abstract level) and to the EPI1 Epigraph (the concrete manifestation of such Oscan units as physical features), through the P67 refers to and the P62 depicts properties respectively. Since the E33 Linguistic Object is an expression of the Oscan language (P72 has language), it can be provided with a translation into any other language (P73 has translation), for example, into English, in order to make the content more understandable.
This study has just scratched the surface of what is certainly a complex problem. Although the bases for creating an epigraphic extension of CIDOC CRM (CRMepi) have been laid, much of the work still remains to be done. If we consider that EpiDoc not only provides entities for the description of the text and its structural characteristics but also provides a series of tags for the identification of actor and place names inside it, for example, we must notice that the text itself may contain semantically relevant elements that need to be captured in some way. Actor appellations can for instance relate to the commissioners of a given monument or to the people to whom a certain epigraph was dedicated; furthermore, place appellations could also refer to places where the inscription was located in the past, or to which the text refers in various ways. These appellations may also evoke and prescribe the use of thesauri and gazetteers (like Pleiades, for example) to operate a further enrichment of the descriptions or for the creation of terminology tools starting from the very texts of the inscriptions. A special case, but one that is very frequent in epigraphy, is that of the so-called “talking objects”, in which the inscription becomes a “declaration” made by the object itself (“I am the cup of Aphrodite”, “I was made by Ergotimos”) that in this particular case acts as an “actor”, as though it were a living entity. This type of data allows us to enrich our information network about the object and to expand our archaeological and historical knowledge.
Further developments will start from these bases to plan cycles of semantic
enrichment of epigraphic information from textual data. For example, it will be
possible to deduce and instantiate E39 Actor and E53 Place classes from the
appellations found in the text, a process already attempted for discovering
relevant semantic entities in ancient literary sources [
The present work has been partially supported by the ARIADNE project, funded by the European Commission (grant 313193) under the FP7 INFRA-2012-1.1.3 call.