The Daily Life of the Middle Ages, Editions of Sources and Data Processing

The Daily Life of the Middle Ages, Editions of Sources and Data Processing
MANFRED THALLER
Nobody would doubt that research on the daily il fe of medieval societies is possible only, if medieval source material is easily accessible in published form; but, are there any requirements of this type of research, which ask for speci c properties of such editions? Similarly: data processing techniques – be it for the production of eheaper books, the statistical analysis of material, which is otherwise hard to come to grips with, or the storage of vast arrays of material, which shall be made more easily accessible – have in recent years been employed in a number of historical disciplines; but, is there anything in the daily life of medieval times, which asks for specific procedures in applying such tools?
This author has already in a short communication1, presented at an earlier round table of the Institut für Mittelalterliche Realienkunde, tried to point out that due to the speci c research situation of studies dealing with the material (and not so exclusively material) aspects of everyday life, certain forms of data processing are, indeed, particularly suited to this kind of research. In this paper we will try to show: (a) that the two questions at the beginning have to be answered a rmatively, (b) that from these two answers clearly de ned properties of speci c research tools can be derived and (c) how they can be integrated into a coherent technical approach.
1. The Daily Life of the Middle Ages as a Special Topic of Research
If one Iooks at the rich collection of studies, contained in the conference volumes published over the years by the institute in Krems, one could at rst glance scarcely notice anything common in the use the various contributors made of the sources accessible to them. We would like to point out that there exist three types of study, which seem to us to be particularly interesting, each of them, of course, being existent in other elds of research as weil.
(a) There are contributions2, which focus completely upon a very restricted number of sources, being dedicated to a thorough analysis of, e.g., a Iist of inventories, individual accounts, a small collection of pictorial sources and the like. This is in many ways the
1 Manfred Thaller: Mittelalterliche Realienkunde und EDV , In: Die Erforschung von Alltag und Sachkultur des Mittelalters. Methode – Ziel – Verwirklichung, Wien, 1984 (= Veröffentli chungen des Instituts r mittelalterliche Realienkunde Österreichs 6 = Österreichische Akademie der Wissenschaften, phil. Hist. Kl., Sitzungsberichte 433) 210-228.
2 E.g.: Josef Riedmann: Adelige Sachkultur Tirols in der Zeit von 1290 bis 1330, In: Adelige Sachkultur des Spätmittelalters, Wien, 1982 (= Verö entlichungen des Instituts für mittelalter liche Realienkunde Österreichs 5 = Österreichische Akademie der Wissenschaften, phil. Hist. Kl., Sitzungsberichte 400) 105-131. RudolfEndres: Adelige Lebensformen in Franken im Spätmittelal ter, ibid. 73-104; Gerd Zi erm n: Ein Bamberger Klosterinventar von 1483/86 als Quelle zur Sachkultur des Spätmittelalters, In: Klösterliche Sachkultur des Spätmittelalters, Wien, 1980 (=
6

most certainly solid kind of research: if we describe how many items of a certain dass of goods have been available at a certain castle, this is more or less what the source says – there is scarcely a chance for erroneous Interpretation. The eternal question, asked of that kind of research, however, is: so what? What relevancy does it have to know, how many cloaks of a certain colour where available at a certain castle, if we do not know, if their nurober just expresses a private whim of its Lord or Lady – or, indeed, an anxious e ort to keep up with some contemporaneous noble Jones? How representative is, whatsoever has been said?
(b) There are completely di erent contributions3, which give a sweeping picture of a broad phenomenon of material provisions, of a social behaviour or of norms imposed by the society, based upon evidence selected from a wide variety of sources. The ability, to draw upon widely scattered examples, is usually seen as proof for the universality of the phenomenon described and the validity of the Interpretation given. Nevertheless, each author has usually to acknowledge that, using such a vast array of diverse source material, she or he has not been able to go into a detailed appraisal of every source encountered; and indeed, even when we assume, that a few of them have been misunderstood, the validity of the overall picture will remain unchallenged. Still, it is a rare occassion, that such a broad synthesis of traces of a phenomenon does not evoke heated debates of specialists about the selected points they are familiar with.
(c) The third main type of argumentation consists, nally, of basing a hypothesis upon one central source, with which the author can reasonably argue to be so weil ac quainted that the basic model of Interpretation is being derived from a streamlet of tradi tion claimed t b completel understoo – suc centra source being then augmented by proof of the occurrence of the observed phenomena in other sources, other areas and periods as weil. Strengths and weaknesses of this type of study usually re ect the arguments given above, depending if the author has pui more prominence to the individual source or to the secondary material accompanying it. (Obviously the precise di erentiation between this type and the two preceding ones is di cult; we therefore avoid giving examples.)
Few readers will probably argue with the description just given; many however will say that what we said is noi at all specific to research on everyday’s life, but an inherent di culty of historical research, so that the same typology might as weil be used to describe
Verö entlichungen des Instituts für mittelalterliche Realienkunde Österreichs 3 = Österreichische Akademie der Wissenschaften, phil. Hist. Kl., Sitzungsberichte 367) 225-245.
3 E.g. Gernot Kocher: Spätmittelalterliches städtisches Rechtsleben, In: Das Leben in der Stadt des Spätmittelalters, Wien, 1977 (= Verö entlichungen des Instituts für mittelalterliche Realienkunde Österreichs 2 = Österreichische Akademie der Wissenschaften, phil. Hist. K l . ,
Sitzungsberichte 325) 51-75.
Ages – A General View and Problems of Research, In: Frau und spätmittelalterlicher Alltag, Wien, 1986 (= Veröfef ntlichungen des Instituts für mittelalterliche Realienkunde Österreichs 9 = Österreichische Akademie der Wissenschaften, phil. Hist. Kl., Sitzungsberichte 473) 9-18. Werner Rösener: Zur sozialökonomischen Lage der bäuerlichen Bevölkerung im Spätmitttelalter, Wien, 1984 (= Verö entlichungen des Instituts für mittelalterliche Realienkunde Österreichs 7 = Österreichische Akademie der Wissenschaften, phil. Hist. Kl., Sitzungsberichte 439) 9-47.
7
Sh amith Shabar: T h e History o f Women in t h e Later Middle
the kind of studies produced by any kind of historical research4• This is indeed so; still we consider the situation of the researcher of the daily life as a peculiar case: in almost all other elds of historical study we can, and rightly so, assume that knowledge tends to accumulate. That the relevance of a source discussed in isolation will ultimately become clear by other isolated sources; that the broad synthesis can become a frame of reference, to be used afterwards for checking successive ndings, if they t it or not; that an analysis of a central source can be expected to serve as a starting point for the understanding of other sources.
This will work wonderfully in (in the broadest possible sense) prosopographical5 work: the activities of persons being restricted to a certain geographical area, we have no reason to doubt, that the immatriculation lists of a given university will be consulted by everybody who wants to trace the fate of the citizens of a community nearby. It will work great with all kind of legal systems and proceeedings; the kind of legal relationships between any two people can, after , only take a limited number of forms. Even in a streamlined history of the economy, knowledge is constantly accumulated; a „price“ being a very weil de ned quality, that lends itself perfectly to comparisons between studies6•
As a result, all classical editions of source material strive for three aims: (a) to make more easily accessible the text of a document, which eise could be consulted only at a spe ci c archive, (b) to add to it l information about the entities mentioned in the sources, by hinting at other sources, where, e.g., the same persons and/or places have been mentioned as weil and (c) to blaze a trail for the historian encountering the same task with another corpus later, by providing extensive registers of the entities mentioned in this source.
Now, all historians will agree that the nutrition of the populace has been a central question in daily life: is there any imaginable source, where a register of passages dealing with problems of nourishment would be provided by even the most careful editor, as part of his or her intellectual duty except in cases where su interest has been part of the reasons for the edition? (As would be the obvious duty of every editor to provide a register of persons mentioned in a source, though being interested only in the legal formulae it employs). Is it, intellectually and economically, feasible, that editors make themselves acquainted with all aspects of material culture, so they can be relied upon, to develop the habit of indicating all exceptional references to food, architecture, clothing, furniture etc., in the same degree every trained editor of a corpus of sources will consider it necessary to indicate exceptional legal constructions being encountered? And, particularly: is there
4 Cf. M fred Th e r: Praktische Probleme bei der interdisziplinären Untersuchung von Gemeinschaften ‚Langer Dauer‘, In: Gerhard A. Ritter und RudoJf Vierhaus (Edd.): Aspekte der historischen Forschung in Frankreich und Deutschland, Göttingen, 1981 (= Veröfef ntlichungen des Max-Planck-Instituts für Geschichte 69) 172-189.
5 Which therefore has produced a disproportionally !arge nu ber of computer applications: Karl Ferdinand Werner (Ed.): L’histoire medievale et les ordinateurs, München etc., 1981; Helime MiJJet (Ed.): Prosopagraphie et lnformatique, Paris, 1985.
1 lndeed, a pilot project to make a directory of medieval and early mode currency exchange ratios, available on-line is currently implemented in the United States. See: The Medieval and Early Modern Data Bank, in: The Research Library Group News 12 (January 1987) 8-10.
8

any chance of this being done in sources, where such references would not be expected in the rst place?
To sum it up: we tried to point out that historical knowledge in traditional disciplines, as political, prosopographical or legal history, h s a tendency to accumul te, which is considerably stronger than just the obvious phenomenon that a later historian is able to consult what an e lier one h written. In many historical disciplines knowledge gained from one source will in uence the treatment another source gets during the very process of being edited critically; if possible at all, this happens only very restrictedly with regard to the history of the daily life nd the sources relevant to it.
What we would need to overcome this restriction, is essentially a technique o{ edition, where, Iogether with the classical tools of critical apparatus and identifying registers, a possibility exists to address directly every portion of a source, which is interesting {rom the point of view of such a discipline as the study of the daily life- in its most primitive m nner by having immediate access to every context, in which a certain word appears. In a utopian sense we imagine a situation, where a historian having found in a newly encountered source a term for some item of material culture not yet known to him or her, has a tool, llowing to access the sa e term immediately in a !arge number of sources.
The remainder of this paper argues that, while not feasible within the next year, the current developments in data processing make it right now at least possible to sketch, how such a tool h to Iook like, and how we can arrive at it. We are convinced, furthermore, that this prelimina.ry conceptual work should Iake place now, that is before technical fait accomplis have been created.
This necessity, to discuss in detail the possibilities of modern information technologies now, we see particularly in two respects:
• We are at a stage now, where data processing has developed a dynamic of its own that allows it to dominate elds like type setting and publishing, quite without any concern for the feelings of the uthors – or critic editors- whose labours are to be published. lndeed, one of the few prophecies about the development of the next years one might feel reasonably sure to make, is that within a short range of time every critical edition that becomes printed, will per de nition be available at the sa.me time in machine readable form, as every text to be printed will have been composed on some electronic device. We should explore now, if this situation creates additional possibilities for historical research, which go beyond books remaining (or becoming gain) at least marginally a ordable. As any argument like that is known to raise certain fe rs, it should be made emphatic y clear th t we do not spe k of the end of the printed book: we speak about possibilities resulting out of new ways of producing printed books. We should do so the more, because we eise might discover some da.y, that the publishing trade h s done everything th t is lucrative from the point of view of publisher, blocking other avenues in the process.
• This author seriously believes- and has discussed with regard to a number of topics in a recent series of papers7- that there are fundamental di erences between information occurring in our present day discourse and information Iransmitted to us from earlier
7 M1mfredTha er: WarumbrauchendieG isteswiss nschafi nfachspezi schedatentechnische Lösungen? Das Beispiel kont xtsensitiver Datenbanken in der Geschichtswissenschaft, In: Man·
9

times; if any of these suspected di erences are true, they will have to be considered before applying any formal treatment of information to historical sources such.
2 . Retrieving Information Out of Historical Sources
Before we discuss, how an edition of a historical source might be made more trans parent to difef rent classes of historians, we would like to describe rst what precisely it means to search for information8 in a corpus of written material. All the tools described in the following sections have been applied to historical research8: when we describe them
fred Tballer and Albert Müller (Edd.): Computer in den Geisteswiuenschaften, Frankfurt/New York, 1987 (= Studien zur Historischen Sozialwiesenschaft 7); Ungefähre Exaktheit. Theoretische Grundlagen und praktische Möglichkeiten einer Formulierung historischer Quellen als Produkte ‚unscharfer‘ Systeme, In: He a Nagl-Docekal and an Wimmer (Edd.): Neue Ansätze in der Geochichtswiuenschaft, Wien, 1984 (= Conceptus Studien 1), 77-100; Vom Beleg zum Begri . Der Beitrag der Datenverarbeitung zur Lösung von Terminologieproblemen, In: Gerhard Dienes et al. (Edd.), Ut populua ad hiatoriam trahatur. Graz, 1988; aecundum manuscript. Zur Daten verarbeitung mehnchichtiger Editionen, : Reinbard H rlel el al. (Edd.): Geschichte und ihre Quellen. Festschri für Friedrich Hausmann zum 70. Geburtstag, Graz, 1987; Geographische Angaben in einer historischen Datenbank, In: Eratosthene-Sphragide 2 (1987).
1 The author is weil aware ofthe exi1tence ofthe concepts ofa data bank and, more speci cally, of data baae manipulation syatems. Indeed some of his publications deal with modi cationa of current concepts in the information aciencea neceuary to make some of the more technical of these conceph applicable to hiatorical research. This paper is directed to historians however; the author abstain1 intentionally from any diacuuion of information acience topica. Indeed he ia aware that he does – and intenda to – blur di erencea between varioua Ievels of software products, like fully grown data baaea, ctual knowledge administered by data banks, full text retrieval systems and so on. He does so, because he believes it necessary to concentrate among historians upon the needs of the discipline, to de1cribe – abstracting from existing ao ware -, what kind of toola we need; when the practical con1equences of auch substantial decisions are discussed and implemented, he agreu fully with the neceuity to discu., data baaing and other areaa of aoftware on the same Ievel ofabstraction, u1ed in the professional development of aoftware, and thinks he has done so. By oimply popularizing, however, what is deacribed in a computer science text book for the rst years of a cour1e in information 1cience, one does not so much enlighten hittorians about data processing in his opinion, one rather obacures the fact that there are proble e of data proces1ing, which are speci c for our discipline1. To the historian, who wants to inform her- or hi self more thoroughly about the technical background, we recommend: On Data Basing: Toby J. Teorey and James P. y: Design of Databa1e Structures, Englewood Cli s, 1982; David R. Howe: Data Analysis for Data Baae Design, London, 1983. On non-trivial data proceasing conceph: William Kent: Data and Reality, Amsterdam etc., 1978; Jobn F. Sowa: Conceptual Structures: Information Processing in Mind and Machine, Reading, Mass. etc., 1984. On non-elementary programming techniques:
Jean-Paul Tremblay and Paul G. Sorenson: Introduction to Data Structures with Applications, Auckland etc., 21984; Billy G. Claybrook: File Management Techniquea, New York etc., 1983.
• For a rat orientation: iedricb Hausmann et al. (Edd.): Datennetze für die historischen Wiuenscha en?, Graz, 1987; Manfred Tballer (Ed.): Datenbanken und Datenverwaltungssysteme
10

isolated, we do so, because in most, though not , conventional computer systems10, some or all of these approaches are mutually contradictory. If you have choosen one that is, you preclude yourself from using the others.11
2.1. Structural Access
Structural access we call that way of retrieving information from a source, which is probably the most fa liar to all historians, who have had any exposure to computers at all. It works roughly according to the following logic:
(a) Historical source material – or information collected (rom a number of sources and/or from Iiterature dealing with historical phenomena – is entered into a computer
als Werkzeuge historischer Forschung, St. Katharinen, 1986 (= Historisch Sozialwissenschaftliche Forschungen 20).
10 In historical research, there exists virtually no agreement about the infrastructure necessary to apply a computer to historical research- therefore the moot unreliable claim, that can be made in computer supported historical studies, is that a particular system is „extremely easy to use“. lndeed, very prominent advocates of particular approaches have not formulated any command to a computer for years, have been convinced, however, by the ability of their otaff of (mostly student) assistants to produce in due time the required results that, whatsoever this staff uses, it is very easy to use indeed. All our following remarks assume a difef rent concept: that of historians chronically short of funds, who, far from bcing able to hire additionalstudents, employ the computer precisely to reduce the workload encountered by themselves. We concentrate in the following paragraphs, therefore, not upon the possibility to program computers in procedural steps: „read the data until you reach the rst line, which contains an ampersand. Extract the following data until you encounter a co , which is not contani ed between quotation marks. Store them, until you have read all the data. Now sort the data.“ Rather we tacitly assume that all the software, which we will explore, has the property to handle „logical objects“: „Create a register of rst names“. (We are aware that this is only a very intuitive introduction to the meaning of „object oriented“: a short explanation what the difef rence between the old „procedural“ and the new „object oriented“ style is like (though on a more elementary Ievel) can be got from: Patrick H. Wimton: The LISP Revolution, in: Byte 10/4 (April 1985) 209-218.)
11 We apologize that the following sections may seem as bewildering to information scientists as, unfortunately, also to historians, who have just read a short introduction into data basing. The author is fully aware that the following classi cation does not run along the lines on which data structures are usually described within the classical „data models“ of information science. The following approach has been chosen, however, because the author has experienced during the last decade that these are the di culties, which are encountered by historians. The whole theory of data models of information science is outspokenly (and in the meantime also almost subconsciously) dominated by properlies of data models, which ultimately stem from their difef ring behaviour, when exposed to problems of frequent on line updating. While these problems certainly can occur as weil with historical data bases, an edition of a historical source, which shall be accessible by data base oriented tools, will have updates among its least central problems. We focus intentionally on problems, which are not related to update problems. Readers which, by exposure to dBase or aimilar systems, thought that the relational data model is the only one, might get a broader peropeclive by Dionysios C. Tsichritzis and Frederick H. Lochovsky: Data Models, Englewood Clifsf , NJ, 1982.
11

according to logical structure, which is assumed to be at least latently available in all units of material to be administered by the computer. An example for part of a charter entered in this way, might be:
sender*Friedericus/de Brunnens•
recipient$Willelmus/de Guiningen
(b) To make use of any information contained in the data, yon have to de ne its
complete structurallocation. That is, you have for example to specify name of sender, if you want to address the name of the sender; a very useful feature, indeed, if you want easy means to distinguish between names of senders and names of recipients: a potentially calamitous situation, however, if you want to produce, say, a register of all „persons“, that are both senders and recipients, in the data. In many data bases which provide for what we have called structured access, you would need for such a register at the very least one explicit command for every kind of persons you want to put into this register; in some it would not work at all.
This is no unsurmountable obstacle, if you are going to di erentiate in charters be tween sender and recipients: but what about the private charter, where it is also mentioned that the recipient has a nephew, who, as soon as of marriageable age, shall marry the sec ond cousin of the sender; after which event the legal proceedings, which are documented in the charters, are to be considered void? Data structured to take account of that, have usually to be made up of dozens – in extreme cases hundreds -of structurally di erent kinds of entities (Iet us for the sake of simplicity and clarity assume for this paper that an „entity“ is something like an individual person). At the very best, it means that you have to write down dozens or hundreds of commands, if you want to access of them – , e.g., necessary to create a register. That quite a few software systems of that type are not able to combine information out of more than one kind of entity into a meaningful analysis, we just mention in passing.
We will not follow this in detail: Iet us just keep in mind that there are programs, which transform information into entities, between which very clear structural relationships exist; that this approach is very useful, if those structural relationships become central for the analysis; that it is de nitely troublesome, if you have to ignore the structural relationships at least some times.
2.2. Semantic Access
So in the rst approach, you divide your source into entities, which roughly correspond to people: and what type of people they are, becomes the central concern. The same approach, of course, can be taken with material objects: for example, a piece of furniture can be described by a structure in the sense given above, which simply divides the object further:
Object.Bed/Wooden Part.Bedstead/Carved Part.Coverlet/Silk Part-of-Part. broidery/Italian Part$Cushion/Red
Pie e notice, that we do not give any detailed instructions, how such data are formed: still m t of the readers will have immediately understood that we just described a wooden
12

bed, which has as remarkable features a carved bedstead, a coverlet made of silk and a red cushion; the coverlet, in due turn, being distinguished further by having an embroidery of ltalian origin (presumably stiched to it). The proble is that, while we obviously have to avoid the situation of the computer telling us that we have embroidered beds (due to this we made the embroidery a part-of-part of the bed only), it is entirely feasible that in the same group of sources we also have something like this:
Object$Cloak/Purple Part$Embroidery/Italian
the embroidery, indeed, being a part of a cloak in a sense, in which it never could be part of a wooden bed.
According to what we said so far, this unfortunately would mean that, when asking for all embroideries appearing within our sources, we would have to memorize that an embroidery can occur as the name of an object, the name of the part of an object, the name of the part-of-part of an object, probably the name of the part-of-part-of-part of an object as weil, and only heaven knows, as what eise.
Why would a human being have less di culties nding all embroideries? Because a human would di erentiate between cases, where you are interested in the relationships between various Ievels of subdividing an object, and cases, where these subdivisions can pro tably be ignored, tacitly assuming that the whole data above consist of a sequence of unrelated items, each of which is described by two fields (or descriptors, or variables, or … ) „name“ and „quality“.
For the purpose of this paper, we will it call semantic access, if a data base has the property of being able to access all entities, which have a field (or descriptor, or variable, or …) known as „name“ without regard to the structural relationships between such entities. The gain is obvious: in our preceding example, we would have not to bother any more about which kinds of persons are available, we would be able simply to say the computer: „produce a register of entities which have a name“ (or, more probably: „produce a register of all entities which have a personal name“). In theory, there is no di culty with such an approach at all- unfortunately, however, in commercial software, there is usually a kind of trade-o between the two approaches: almost all existing systems are either good at preparing structural access to data or at giving semantic access; very few of them o er one without restricting severely the other.
2.3. Access by Content
Even if we are ready to ignore this unpleasantry, however, there is still another prob le . Let us assume, we do research upon the Cashion of the 15th century. We device for that purpose data, which consist of pieces of clothing, which are made up of variant struc tures of entities, which are interrelated somehow among each other, all having in common three fields, however, name, colour and material, as in
item$cloak/red/wool
Unfortunately this is not such a universal solution as we might imagine: in quite a few cases, the colour of material is an integral part of the description of its quality. A „cloak of red cloth from Flanders“ might not only be of another colour than a „blue cloak of cloth from Flanders“, but also be made of a distinct quality of Flandrian cloth, being
13

so ne that it is always associated with presti ous red. Now, dividing the information we have distinctly between colour and material, we give away some important information on the quality of the material; entering „red“ alternatively into colour or material, we have to think less than Straightforward in convincing the computer that we want to access all „red objects“. (If we enter the colour in both, all theoretical problems are settled; we just get a headache in keeping the data consistent – particularly during corrections after initial data preparation.)
Similar problems are quite frequent in medieval data, by the way: is „Hubertus iburger“ „Hubertus“, migrated to bis current residence from „Friburg“, or does he have a surname?
We solved structural ambiguities by concentrating on the names of the elds; now we have to ignore even these names. Indeed, there exists software, which is able to ignore all categories completely and to treat the content of the data base simply as a set of words, all of which are equally accessible; so we are interested only in the words occurring, in the contents of the data base; neither in its logical structure, nor in the di erences between the various elds (or descriptors, or variables, or …) describing the individual entities.
Such systems, usually known as full text or, more recently ee text (retrieval) systems, have just one shortcoming: they usually insist that you ignore any structures that might be available – they will not only give you words, which consist of the same characters and might mean any of two things, they will give you words, whlch consist of the same characters, though we can be absolutely sure, that they have di erent meanings in their various contexts.
2.4. Access by Meaning
This last mentioned proble completely aside, there remains at least one additional di culty.Whenwewanttogetal.linformationavailableabout“furniture“,inallthe access methods we have discussed so far, we would have to specify very explicitly what we mean by furniture; in fact we would have to specify „beds“ and „tables“ and „chairs“ and …; and, unfortunately, we would also have to specify „beddes“, „tabls“ or whatsoever other variation in spelling we encounter in our sources. Now, when we do this once, we can in very many computer systems tell the machine, that it shall reme her these definitions of „furniture“ – a capability, whlch is usually implemented as a „thesaurus“, often clever enough not only to understand that certain things are related to a common concept, but also that they form part of each other and more sophisticated relationships between the meaning of words12, contained within a corpus of sources.
All of the three access methods described so far can be enhanced considerably by the use of such thesauri; even more so, when we employ a system, which combines a thesaurus with what is known as a lemmatization program, which, roughly, reduces any wordform encountered to a word stem before using it. This reduces significantly the e ort needed to specify minor variations resulting from di erent grammatical usage as being equivalent in meaning for the purpose of the historian. With software, that allows structural access, we would therefore be able to ask for „objects with a name that signi es ‚furniture‘ „; with
12 On specialized toolo for the administration of terminological information see J. Goetschalckx (aic) and L. Rolling (Edd.): Lexicography in the Electronic Age, Amsterdam etc., 1982.
14

software allowing semantic access, we could ask for „something with a name signifying ‚furniture‘ „; and with content-oriented access, we nally would be able to ask for „contexts in which ‚furniture‘ occurs“.
Still, this access method, too, is not without drawbacks: quite besides thesauri being usually administered in a way, which makes changing them not too easy, such software often is somewhat restricted with regard to the number of terms it can handle: that makes thesaurus supported systems, which handle – not in theory, but in practical operation – very restricted vocabularies, less useful for historians, who want to transcribe a source literally, and favour those, who use a pre-de ned number of keywords.
And here we would like to put an axiomatic statement: a very di erent situation is created by any computer system, which accepts the words the user needs to employ without Iimitation, when compared to systems, which force the user to select phrases from a limited and restricted vocabulary. There is no conceptual di erence, however, between systems, where you simply have to enter code numbers and such, where you have to select out of a prede ned set of keywords. The first distinction may create the ability to handle a source as it is; the second provides an optical advantage, which may be very important psychologically, but does not bring the data closer to the original at .
2.5. Data Revisited: Input Conventions
Before we discuss, how out of these techniques eventually a tool could be derived, which is appropriate to turn sources into computer administered editions that can be understood to consist of an almost in nite number of cross referencing systems, for every aspect of the daily life, we should go back to the question of input13.
2.5.1 The „Structured Input“ Approach
We closed the preceding section with a very strong statement in favour of source
orientation: in favour of entering source material as close to the original as possible. Now:
Object$Bed/Wooden Part$Bedstead/Carved Part$Coverlet/Silk Part-of-Part$ broidery/Italian Part$Cushion/Red
will certainly not appear as part of a medieval document, which rather would read:
„Item, a bed made of wood, with a picture carved nicely into the bedstead, a covered with a cloth of silk, in the middle of which a beautifui embroidery, done by a Florentine maiden, appears; also a red cushion on top of the bed.“
Now „sticking closely to the source“ can mean, as we will see soon, that indeed we keep the text in the second form. Most historians will admit, however, that the quotation as such contains quite a few terms, which are of doubtful relevance to but purely philological research. If the „beautiful embroidery“ „appears“ in the „middle of which“ or simply „is there“, may have some relevance for intellectual history – or indeed, as mentioned, for
13 M fred Tha er: A Draft Proposal for a Standard for the Exchange of Machine Readable Sources, in: Historical Social Research / Historische Sozialforschung 40 (October 1986) 3-46.
15

philolo sts using the text -, but will not usually Iead to important insights regarding the daily life of the inhabitants of the bed in question.
Now, what is usually irrelevant in this sense, is also hardest to understand for a computer: that is, the relationships created between various parts of a text with the means of gr ar, associative arguments, elliptical expressions, and so forth. Therefore, a weil trodden path of source conservative or source oriented data proc�Rsing consists simply of replacing these grammatical relationships by a structure of abstractly de ned units like „object“, „name of such“ and so on, which are indicated by a system of limiting characters (in our example dollar signs and slashes). The vocabulary we put into these structures is completely ee; so
ObjecttBed/made of vood
ParttBedstead/a picture carved nicely into the Parttcovered vith a cloth of/Silk Part-of-Parte broidery/done by a Florentine maiden Parttcushion/red
would be as acceptable as our rst construction for almost all systems which accept this rst one at all.
This decision, of restricting oneself to the semantics of a source, forgeUing about the syntactical intricacies by which the language has been formed, we will refer to henceforth the technique of strudured input; its Iimitations are probably obvious. The reason, why such a lo c is often recommended, is simply, because there are some computer systems14 already, which are perfectly able to understand it. – As we will see that those other solutions have their shortcomings as weil, however, for all sources, which have a relatively orderly structure, like account books or similar, structured input has indeed some virtues of its own.
2.6.2 The „Pre-edited Input“ Approach
Still, there may be some reason, that one really wants to catch the source just as it
is. A solution for this might be the following:
(0
“Item, a (0 (N bed N) made of (M vood M), vith a picture (0 (Q carved Q) nicely into the (N bedstead N) 0), all
(0 (N covered vith a cloth ! of (M silk M), in the middle of vhich a beautiful (0 (N embroidery N), (Q done by a Florentine maiden Q) 0) 0), appears; also a (P (C red
C) (N cushion N) 0) 0) on top of the bed.“
Indeed software systems understanding this kind of text, which consists of objects
0), which can be described by name (N . . • N), material (M . . . M), quality
14 d Tha r: AELW • A Data Base System Speci c for the Historical Disciplines, Göttingen, 1987. The current version is available from the Max-Planck-Institut fur Geschichte, Hermann-Föge-Weg 11, D 3400 Göttingen; future editions will be distributed by the regular channels of the publishing trade. A German version of the manual is available as weil; a French one currently being prepared. Cf. Kevin Schurer: Historical Research in the Age of the Computer: An Aueument of the Present Situation , Historical Social Research f Historische Sozialforschung 36 (October 1985) 43-54.
16
(Q • . • Q) and colour (C . . • C), and may in turn contain further objects, have been implemented 1 5 •
This form of preparing a source for the computer, known as pre-edited input, is undoubtedly closer to the original than the earlier one. Also, it saves the necessity to restructure the text – you can simply lranscribe it, as it occurs. The number of programs able to cope with such input, is incomparably smaller, however, than the ones that can handle the earlier one. Beyond that, this form of entering data has one major shortcoming, which may depend on the experience, one has got with using it, is viewed as considerable by many people, however: if you expand the principle to more complex constructions – and the ability to do so is the main point of starting with it – systems of logical brackets or start-stop symbols can become easily very complex, and it is extremely time consuming to nd out, where in a lengthy expression, which the computer considers erroneous, a particular bracket has been forgotten.
2.5.3 Mask Oriented Input
„So why do we not simply buy an ATARI? There it is all so much easier; you just are presented with a mask and ll into it whatever you have to. That is, the computer generates on its screen a form, as would be a questionnaire on paper, where you are asked for every piece of furniture, which name, quality, material and colour it has, and which parts it consists of, repeating the same question „for every part.“ Weil; – one answer might be: try it. We have intentionally not mentioned a very trivial fact so far: any kind of truly source oriented program assumes flexibility in a number of ways. Reasonably advanced systems of source oriented data processing all allow each eld to have an almost arbitrary length; commercial (and that includes most mask oriented) programs usually allow only predetermined lengths of elds, often heavily penalizing users who ll in only part of the eld. Systems of structured and pre-edited input all allow constructs of more or less unlimited complexity, as long as they are made up of the structural items or start / stop symbol sequences you de ned; for mask oriented input you have to prepare the most complex structure to be possibly handled much more in detail in advance.
However, we introduced the question at the beginning of this chapter as colloquially, as we did, to avoid even the slightest hint at intellectual propriety. lndeed, asking the question as we did, in the context we introduced it, is one of the easiest ways to convince everybody with more than a super tial knowledge of data processing that the historian asking it does certainly not know anything about computers at all.
One of the most important concepts in data processing, which is often obscured by introductions into the eld gained via the user’s manual of a software system or a micro computer, is the di erence between the background system and the user shell. lndeed, we blurred this di erence ourselves so far: important, from a data processing point of view, are the classes of access to be possible within a system (and on a Ievel too technical to be discussed here, the logical properties a data structure has to have to make the cho.sen ways of access possible). The parts of a programming system responsible for the administration of the system, and those, which decide, which of the characters the user
15 See Kevin Schurer in the last note; Timothy J. King: The Use of Computers for Storing Records in Historical Research, in: Historical Methods 14 (1981) 59-64.
17

has entered shall be stored where, c n be seen completely independent of e ch other. At least conceptually, every program system parses through the input data to divide them into meaningful segments of information, which have to be delimited by some convention, and, as soon a suitable number of such segments has been encountered, they are stored in a data structure that has few similarities to the way the characters are being typed into the machine.
So, for the programs responsible for selecting out of the source those items of infor mation we need, it is perfectly unimportant, if we entered the data via a le of structured data, as a pre-edited text or by using the mask-oriented „shell“ of a particularly user friendly program – as long that mask generator is exible enough to handle the data structureu.
„Parsing“, as the proceas of deciding, which part of the input data shall be consid ered to mean what, is called, indeed can go to very great complexities: the term is most frequently employed in computational Jinguistics, where the „conventions limiting the dif ferent entities and signifying relationships between them“ are the grammatical rules of natural language; even on such a Ievel of complexity we still have to dccide, however, what are the modes of access, which we want to employ. (And computational linguistics is very far om a point, where we could expect programs, which „simply understand“ arbitrary text to store it safely into a logical structure, even in contemporaneous, leave alone historical, language.)
2.5.4 Books vs. Data Bases
Why, then, this lengthy discussion? We mentioned already, that one of the reasons to present this paper now, is, that many historical sources are typeset by computers, that is, are available in machine readable form: and what we said before is necessary to understand, how a source prepared to become a book, could by specialized software be made accessible along the lines we mentioned for research into the daily life of the Middle Ages, s weil.
To explain this, we have to poinl out, that the conventions of pre-edited input, which we described before, are, indeed, rather similar to certain conventions being used for the preparation of computer supported typesetting. For this purpose we compare two exam ples.
18 We hold the frequent miaunderstanding that a beautifully looking acreen implies great power of the data otructurea being adminiatered, for au ciently dangeroua to feel responaible for a warning in clear terma. Of course the relationahipa are more complicated: in caaes, where aofiware ayatema, which handle atructured input, simply expect texts entered with a text editor (or word proceuor) there ia nothing that prevents a historian to uae an arbitrary mask generatot for •moothing the input proceu, aimply storing intermediately characters indicators for structural relationahips, which are to complex for the mask oriented shell to underatand. Still, we would like to repeat our warninga against programs which are all beauty and no substance. lt is very typical that in Roland und M cus Gröber: Die Anwendung des Computers bei der Auswertung von Kirchenbüchern, in: Computergenealogie 2/5 (1986) 135·139, the use of a mouse and even a light pen is aeen more or less mandatory, but the authors – otherwise very computerwise indeed – have obvioualy no knowledge at all about even such trivial techniquea of historical data proceuing the handling of namea with varying orthography.
18

Huius testes sunt: {\italics Friedericus de {\bold Cawenberghe}}, {\italics Cuonradus de {\bold Gravestetten}} et {\italics Markwardus de {\bold Seyfring}}.
would in the typesetting system17, that has also been used to produce the present article, result in
Huius testes sunt: Friedericus de Cawenberghe, Cuonradus de Gravestetten et Markwardus de Seyfring.
Now, we do not want to argue that the idea to print names of persons in italics and names of places hold, is a necessarily good one: what shall be illustrated, however, is that a convention like the one above, which just uses di erent fonts to emphasize systematically portions of edited source, can be immediately translated into the input convention of a programming system, which is able to parse pre-edited text, as in:
Huius testes sunt : (person Friedericus de (place Cawenberghe) ) , (person Cuonradus de (place Gravestetten)) et (person Markwardus· de (place Seyfring)).
That is: the kind of information, that has to be provided to produce a book, which uses any kind of semantically meaningfu1 rule for switching between di erent fonts, can at the same time also be used for the various modes of access we discussed before.
2.6. Access Revisited: Retrieval Lnnguages
We introduced first di erent properlies of various ways to access the information within historical sources; then we described di erent ways to prepare source material for that purpose and explained that the two areas are completely independent of each other. Now, access does not only consist of the e ort needed to prepare data for it, but also of the knowledge one has to have to specify, what infor ation to access. Once again, the speci c tools provided by a programming system, to specify what it has to do, are independent of the kind of data structures it can handle. lndeed, there are progra ming systems, which provide more than one way to access information.
We will illustrate this point with the help of a program ing system18, which in turn is discussed further below. For this illustrative comparison, we assume that we want to select out of a data base sentences appearing in a set of charters, which contain a word starting with rubr („rubrum“ et sim.), to nd out all descriptions of red objects – this would be necessary, if we want to evaluate the importance of an object having a speci c colour, an interest for whkh the editor of the chartulary has almost certainly provided no tool at .
2.6.1 Menu Driven Systems
The rst way of asking the computer to do so, consists of telling it vaguely that we are interested in examining the text of the chartulary, being then supplied by the program
17 . Besides the beautiful description given by Donald E. Knuth: The book, Reading, Mass. etc., 1984, which despite of all its beauty is next to useless for the beginnner, there exist now, fortunately, a few very good introductions like Norbert Schwarz: Einführung in , Bonn etc., 1987.
18 Auw, innote14.
19
with täbles, which describe what is possible at a certain moment; as we pick from these tables what we want the machine to perform, they are usually known as menus. To do so, we need three pieces of information: (a) Where the text of the chartulary is stored (we will assume that somebody stored it as chartulary), (b) under which name the reference Iist of all words in that chartulary is stored – as opposed to an independent register of, e.g, persons – (we will assume, that somebody stored that Iist a� words) and (c) how to activate the program.
As computers are slightly di erent, this third step will vary; so for the time being we just assume, we know how to do that. As soon as we activated the program, we will read on the computer screen:
b
r
w c d a m G •
Pleaae aelect one of the following activities:
Selection of a new data BASE
Selection of a new REPERTORY
Selection of a new/additional WORD form
Redefinition of the CONTEXT for aearch and/or diaplay DISPLAY of the current list of quotations
SAVE list of quotations in a quotation library
MOVE the curaor within the quotation list
Leave Kleio
Leave thia menu
Pleaae enter the selected character and press ENTER/RETURN.
Understandably enough, the computer has to know, which source we are interested in; as he considers a source to be a data base, we type the Ietter b on our keyboard and touch the key that is labeled ENTER on some machines, RETURN on others and TRANSMIT others still. (To be a little bit more concise, we will speak about the ENTER/RETURN key from now on.) The machine reacts with
Please enter the name of a new data base:
to which we respond by entering chartulary and pressing ENTER/RETURN once more. Now we have decided about the source, and have to tell, which of the potentially numerous specialized registers we want to consult. Still having the initial menu in front of us, we enter r, hit ENTER/RETURN and are grceted by:
Please enter the name of a new repertory:
which we supply as words, hitting ENTER/RETURN again. Still in front of the initial menu, we decide, that we are nally getting to the colour red now and enter w, followed by ENTER/RETURN. This results in the computer pleading:
Pleaae specify a word form or a quotation list .
to which we react by typing rubr and ENTER/RETURN. Now the menu, that has accompanied us so far, changes. Our screen Iooks like:
Pleaae select one of the following activities: c Look for a COMPLETE word form
20

b Look for words BEGINNING with this text e Look for words ENDING with this text
1 LOAD a previously stored quotation list G Leave Kleio
* Leave this menu
Please enter the selected character and press ENTER/RETURN.
As we want to know about rubrum as weil as about rubra, rubro and the other forms, declension and the fantasy of a not necessarily perfect Latin scribe has provided, we enter b, followed by ENTER/RETURN. This action results in:
Please select one of the following activit ies :
Redefinition of the list of quotations by a new WORD form AND: restriction by an additional word fo
OR: expansion by an alternate word form
NOT: restriction by a negated word form
Leave Kleio
Leave this menu
Please enter the selected character and press ENTER/RETURN.
This is less immediately comprehensible as the previous menus; it hints at further possibilities (like selecting sentences, which contain any of twelve di erent words, but not any out of ve other ones). For the time being, we just select w, as indeed we just selected a new word form, and press ENTER/RETURN, which brings us back to the original menu, that is to
w
a
o
n
G
*
Please select one of the following activities:
Selection of a new data BASE
Selection of a new REPERTORY
Selection of a new/additional WORD form
Redefinition of the CONTEXT for search and/or display DISPLAY of the current list of quotations
SAVE list of quotations in a quotation library
MOVE the cursor within the quotation list
Leave Kleio
Leave this menu
Please enter the selected character and press ENTER/RETURN.
where we now select d (plus ENTER/RETURN) to Iook at the quotations dealing with red, which we just selected. As a result we are asked:
b
r
w
c
d
s
m
G
*
s
p m
Please select one of the following activities:
Display of the list of quotations on the SCREEN
Display of the list of quotations on the PRINTER MOVE the cursor within the quotation list
21
w
c
t
•
Selection of a new/additional WORD fo
Redefinition of the CORTEXT for search and/or display
Leave Kleio
Leave this menu
Please enter the selected charact e r and press ENTER/RETUR N .
Now, before we start wasting paper, it ght be useful, to see if the quotations are really worth to be printed, so we type s (ENTER/RETURN) and are asked:
How many quotations shall be processed? (Zero = last menu)
We decide that we just want to Iook at the rst three selected quotations, enter 3 (ENTER/RETURN) therefore and have them in front of us.
Most readers will probably feel sure by now that, by experimenting a bit further, they can convince the program to print the quotations dealing with red things, if these are interesting enough to merit the paper.
We went through this process in such detail to illustrate two points: on the one hand, what we have given as a description, would for most purposes be su cient already to work with the Computer. On the other hand, it is probably plain by now that, if so many di erent steps are encountered in looking for and displaying words, there is ultimately a li t to what we can do with such a logic: as, whosoever wrote the program, had to be able to foresee all possible junctions of the search process, where the user m.ight want to select one of two or more alternative courses. So menu oriented access is extremely easy to use; but inherently restricted in fexibility.
2.6.2 Command Languages
The other extreme would be, simply to device a command language, which allows users to ask for everything, that can be expressed in the rules of the syntax of this language. In the command language of the system we are discussing here, which is available as an alternative to the menu-oriented shell we discussed before, the same kind of quotations, which we just selected, would be printed, if we give the commands:
quaero nomen=charter s ; par s = : text=“rubr“
scribe
Now, such an approach, as the simple comparison of the length of this section with
that of the previous one shows, is much more concise: and at the same time much more powerful to express complex wishes for the computer. To master all the tricks of a command language, takes, however, usually much more than the thirty m.inutes most people need, to get the impression that they control a menu driven program.
3. Requirements of a „Historical Workstation“
So we have described what options are available in various elds of computer systems, where trade-o s between the various advantages and drawbacks of different ap proaches have to be made. On the back ground of these descriptions- and still having in nd our vague concept that, what research about the daily life of the Middle Agcs would need, is a kind of „super accessible edition“ -, we will rst try to de ne a best of possible worlds and nally see, how realistic it is to expect entering it in the not too dhtant future.
22

Computers are applied throughout historical research; many specialist elds pose their own speci c problems – be it source criticism, research upon legal terminology and others. In the introduction, we tried to describe that in our opinion the most speci c proble of research into the daily life is the need to use very many di erent sources with very many shifting foci of attention; this translates into the need to employ all the access methods we described above concurrently: a need that is emphasized, therefore, in the following sections. Still, the problems, that are central here, can be solved isolated: to produce a meaningful result, within the context of other historical disciplines, they should rather be solved in an attempt to address oneself to the speci c needs of the historical disciplines at large. In an attempt, that is, to study the possibility of creating an optimal tool for all elds of historical research.
3.1. A Popular De nition
Popularly speaking, a speci cally historical tool for computer supported research can best be described by the way it should be used by the historian.
We assume that, as a by-product of the continuing o ce automation throughout the university, it will be possible for every researcher who wants to do so, to have at his desk a email computer, used at least as a replacement for the typewriter placed there so far. Such a email, personally controlled computer is in many academic disciplines increasingly often tted out with specialized software tools that are geared towards the particular needs of a given discipline: in some engineering disciplines such „specialized needs“ toncentrate upon the possibilities to create three-dimensional drawings on the screen; in some philological disciplines the provision of lexica in machine readable form is considered as particularly useful. Such a personally controlled computer, together with the software components dedicated to the needs of a speci c discipline, is usually known as a workstation19.
What would for the historical disciplines be the speci c requirements of a Workstation?
What do we suppose a historian to do with it?
• A historical workstation will obviously need specialized text processing programs,
which provide for multiple apparatus, characters not frequent in present-day studies – like ö and similar symbols. So it should become a sophisticated replacement of the typewriter.
• The Workstation, however, should also allow to access source material in a more thor ough way than possible so far. This we can understand roughly as follows: the histo rian interested in furniture, who just heard, that a colleague has completed an edition of the wills of a given city, should be able to receive, together with the stately bound volume of the printed edition, a magnetic representation of the same source, which can be made accessible to the workstation: enabling the historian, in turn, to access systematically all phrases in these wills dealing with furniture.
• When, doing so, our potential user has selected a nurober of source references as important for his or her own research, the workstation shall enable him to transfer all the portions of the source relevant to the speci c topic into a collection of „excerpts“
te A good introduction into what a “workstation“ is in the publishing trade, easily compre hensible for a historian, is given by Jo W. Seybold: The Desktop-Publishing Phenomenon, in: Byte 12/5 (May 1987) 149-154.
23

– i.e., the „own“ data base of the researcher -, which is contim�ously updated and
changed, as new material is collected from vazjous sources.
• At least during the second stage of this process – but more probably already at the rst -, the user of the workstation will probably encounter some portions of the material, which it is not possible to understand; or which ve suspected to merit, in the context of the speci c research, more attention than the original editor of tlie source has paid to it, or has been able to pay. An example for the rst situation would arise, if our researcher wants to compare prices in the newly encountered source with prices in material familiar to him or her: as relevant research may have progressed since the time of the edition, the currencies and units of measurements used should be automatically connected to the most recently available treatment of the meaning of such terminology. The second case can occur, if the researcher suspects, that additional insight into the source may, e.g., be got by comparing systematically the names of persans in the source with prosopogr phical information, the original editor did not consider relevant; or did indeed not have access to, as it just e erged recent result of the own e orts of the user of the Workstation.
• Finally, the Workstation should enable the historian to draw „services“, which are not so central, but useful: preparing statistical descriptions or analyses of suitable phenomena in the source; drawing, with the help of the machine, series of maps, showing the development of the spatial distribution of some distinct component of material culture in a series of documents spread across a !arger span of time and space.
The beginning and the end of this Iist is indeed in principle available right now: typesetting and/or wotd processing systems exist, as weil a !arge array of software for statistical or geographical/topographical displ y. While the integration of these various components into a workstation, that can be used by an average historian, is a proble of quite some di culty, it is ultimately a purely technical problem; and, though some of the problems mentioned have a speci c historical dimension, we can more or less rely upon them being solved in principle by the commercial developers of computer related tools – as the solutions of these problems can be sold to nanci ly much more potent parts of the public than the historical research community is.
Remain the problems of the accessible editions and of access to specialized knowledge: tagether with the question, what Ievel of knowledge may be demanded from a historian to make such an approach useful to him or her. lndeed, this may be the central question deciding about the feasibility of it . There have been examples of !arge specialized reference works20, turned into conventional data bases in relatively recent years, which afterwards have almost remained unused, because the results to be gained from accessing them did not seem worth the e ort to learn how to do so. We, therefore, suppose that all services of a workstation have in principle to be accessible by what we described before as a menu driven user shell. We assume, that it is a useful addition to the tools available
20 Almost pathetic: Ben Ross Schneider,jr.: The London Stage Project: its Status and Future, In: Joe Raben and Gregory Marks (Edd.): Data Bases in the Humanities and the Social Sciences, Amsterdam etc., 1980, 31-34.
24

to a researcher, if, encountering the name of a person about which he or she holds no previous knowledge, it is possible to consult within five or ten minutes a prosopographical data base,_ where this person may be mentioned with some marginal probability, getting a result- or rather information, if it is sensible to Iook further for information dealing with
-such a person – within another ve to ten minutes; if and only if the knowledge how to do so can be used for similar consultations and can be acquired initially within an hour or less. The greatest proble with existing historical data bases, in our opinion, is that they do not favour their casual consultation. There are many historians, who would occasionally like to consult a data base, holding some specialized historical information, as, e.g., a data base of pictorial sources of medieval Austria: there are extremely few, however, who are ready to spend a day or two in learning how to do so, as they know that, while such a consultation may be extremely important for a question occurring at a specific point of research, such a question is improbable to turn up in short intervals.
3.2. Libraries of Sources …
So, the first condition for a useful workstation, that endeavors to make accessible sourceinformation,ispurelyquantitative: thereexistsprobablynosinglehistoricalsource, which is so central that even the most optimal access to it would be a su cient reason for historians to make themselves familiar with a new working environment just for the sake of this source.
We envisage, therefore, a workstation, which contains a general purpose data base system, which is able to access a !arge nurober of data bases: these data bases being available on individual magnetic media- to make things more concrete, we can envisage the now available oppy disks, though the technology involved would ultimately have to be somewhat di erent21.
Now, this data base system has to be completely hybrid with regard to the various access mechanisms, which we described in a preceding section, and has to support all access types equally weil. That is a necessity, which has extremely important consequences for the person designing the system (which shall not be discussed any further here1 however1 because we think that the reasons for this decision – as weil as why it is a very central one
– should be clear by now).
Even the best data base system is useless, if there are no data: so we have to make
provisions, making it probable that historians are willing to contribute machine readable sources for the workstation – or indeed, discover that after having edited a source in a basically traditional way, the source has become during that process available as a machine readable data base, and can now be used with the workstation as weil.
We propose the following behaviour of the data base system to be used by the work station: as background syste we propose to e ploy one, which administers all data as a structured network of semantic terms (providing structural and se antic access parallely1
21 The so called „Iaser disks“1 more correctly: „optical digital storage media“1 which right now are becoming standard hardware . On considerations regarding historical research see: Immo Gathmann: Datenbanken und lnformationsverwaltungssysteme. Probleme ihrer Implementierung auf nur einmal beschreibbaren Speichermedien, in: Historital Social Research 41 (January 1987) 76-87.
25

therefore). Access by content shall be possible; every individual historian has the tools and the responsibility, however, to create those selections of elds and of terminology, which shall be accessible by content – as there exist historical disciplines, like the study of the daily Jife and the material culture, which have to access their sources with so rapidly varying emphasis that it is not sensible to expect any historian to provide all possibly needed access paths through the sources in question. By similar reasons, the user of the workstation shall have the possibility to re ne structural and semantic means of access; while it is not possible to decide beforehand, which parts of the vocabulary of a source may be central to a given historian, we ca assume that there exist generally definable entities – like persons, material objects, pieces of land -, which are of some importance to virtually every historian, and can be described by categories – like name, location, age -, which are of equally.general interest. This is not the case for any access method related to meaning, as the speci c relationships between categories of explanatory frameworks and the vocabulary of the source are the very aim of the studies to be undertaken; particularly in a discipline like the study of the daily life. For these purposes, a workstation shall therefore provide tools for the administration of thesauri, and general tools related to the philological background of the sourc , , e.g, tools for lemmatization. The sources them selves shall be made available in a non-lem atized form and be always accessible without recourse to any thesaurus, even if some achine readable sources thesauri are available optionally.
We assume that a system of structured input can be most easily made exible enough to be �u ciently powerful for providing all kinds of access modes simultaneously. As the development of the last ten years seems to show, however, that the easier availability of systems supporting structured input directs an increasing number of historians also to the analysis of sources, which in fact would be served better by systems using conventions of pre-editing; and, furthermore, as an increasing number of historical sources is bound to be available prepared for electronic typesetting, we propose that the Workstation shall contain a parser that can convert pre-edited input, supplied by a user, into structured data to be processed later by the input modules of the data base acomponent of the Workstation. This modularity should result in much more exibility than an attempt to integrale hoth kinds of parsing mechanism into the data base components directly.
3.3. … and of Background Knowledge
While we did assume so far that a workstation shall be able to access many in dividual machine readable sources as possible, but will be used to access each of them only relatively rarely – or at least by very few historians more often than occassionally -, there exists „historical knowledge“, which is generally enough to be applicable to any sources, as information on chronology, currency systems (and indeed prosopography, if more narrowly de ned prosopographical information is extracted from the context of the entire source). Such knowledge can – and indeed has been – administered by dedicated data bases. We assume that a historical work station shall contain information of that kind enveloped in specific programs (roughly: so called „expert systems“ 2 2 ) , which allow the
22 Cf. Ricbard Ennals: Arti cial Intelligence. Applications to Logical Reasoning and Historical Research, Chichester etc., 1986.
26

historian to consult them, instead of printed information dealing with the same subject. Why would it be potentially more sensible to consult an „expert system“ on chronological problems \nstead of simply a book? One might quote increased comfort, greater exibility and so on: for all practical purposes, however, we do not think that all these arguments Iogether will convince any historian, who does not simply like computers to begin with. The real importance for such integrated expert systems, in our opinion, is to be found in two special aspects: on the one band, the kind of knowledge we are talking about now, is precisely that one, where the „addidivity“ of historical research, about which we talked in the beginning, is most important. If one historian discovers that a certain counting unit means something di erent from what so far has been assumed, this discovery has a more practical e ect upon alt other historians than even a very detailed study about a long chain of events in some community, as it can immediately be applied in any context, where this counting unit occurs. Printed compendia, however, are very cumbersome, when it comes to take notice of such singular discoveries. This, indeed, is one of the areas, where an electronic tool, by virtue of being updated more easily, is simply more accurate than a
traditional one.
The other eld of application, and this is the most important one, lies in the imme diate application of such „background“ knowledge to a given historical source: e.g., by telling with one command, that prices within the source shall be interpreted according to a given table of identi ed currency units – or, in a very advanced version, by automatically comparing every person appearing in a source with a large prosopograp cal data base, to ease the identifying tasks of the editor of the source.
3.4. The User Interface
Now, in some of the preceding paragraphs, we emphasized that tools like the one we describe here, are useful only, if the historical user can learn how to control them in a minimal amount of time; and apply, what has been learnt once, to very many di erent sources. In the preceding sections we explained in detail, what we consider an easy, menu oriented way to approach a source: we tried to explain also, however, that menu oriented systems, easy to use as they are, have certain inbuilt restrictions regarding exibility. This, therefore, would precisely be the approach not to take, if the resulting workstation shall be handling as many different classes of sources as possible. lndeed, out of our very rst arguments it follows, that while a „genealogical work station“ or a „workstation for an economic historian“ might be conceptualized very much as an extremely specialized tool, the distinguish.ing property of a „workstation for research on the daily life“ would have to be, that it is as unspecialized possible – as what we gain here is not (as in genealogy) the possibility to identify persons, or ( in economic history) the one to concentrate speci cally on the development of prices, but the possibility to apply as many di erent tools to as many di erent types of sources as casually as possible. – Which in due turn makes such a Workstation, of course, much more useful to the historical disciplines in general than a
more narrowly de ned one could ever become.
This particular circle can, in our opinion, be squared by distinguishing clearly between two situations: on the one band, we have the historian concentrating upon a certain source, e.g., an account book; on the other the historian, who, using many sources at a time, wants to use that very same account book, as part of his general base of sources, casually. The
27

„c ual user“ wants to be botbered as little as possible by speci cities of the workstation; the „accounts book specialist“ wants to use as mucb of tbe workstation’s power as possible. In our opinion, the solution consists in building the workstation basically upon a powerful command language, and put a menu driven system on top of it, which translates tbe menu items cbosen into the command language. The caaual user will never transcend below this uppermost of the system;‘ the 3pecialist engr.ged in getting th maximum out of a source, will have the possibility to do just that, by „looking behind“ the menus. – And, if a mechanism is provided to in uence the menus presented to the next „casual user“, who is accessing the same source, the speci ist, preparing a particular source for his own purposes, will, as a side e ect, indeed prepare the material for such a more casual user as well; that is, undertake the classical editorial tasks.
4. Possibilitie’s for Realization
So much on the concepts; it would be pointless to describe them in such detail, if they would not be related to some on-going work. lndeed, that is the case:
• A data b e managment system that includes the principles discussed here, is currently being developed, under the name �Hw , at the Max-Planck-lnstitut für Geschichte, building upon the experiences gained there with an earlier system known aa CLIO since 1978.
• The description of the workstation presented, forms part of a project, currently being formulated at the same institute, negotiations with a number of other institutions about possible forms of cooperation being at this stage the main focus.
• The software needed to convert pre-edited input into structured one, has been planned pretty much in detail and is supposed to be developed within the framework of the subgroup on standardization of the Association for History and Computing23.
• Prototypes of Public Domain Data Bases, which can be accessed by the data base man agement system � , dealing both with sources proper and with the background knowledge necessary to handle them, are currently being prepared by a number of re search institutions24• They are supposed to become available during the spring of 1988
23 Further information on the Association and its standardization subgroup, which is preparing a comprehensive publication on problems regarding the exchange of machine readable historical data, ia available from the author.
24 stitut für Mittelalterliche Geachichte, Universität Freiburg/Breisgau (PD Dr. Dieter Geue· nich); Forschungsinstitut für Historische Grundwissenschaften, Universität Graz (Dr. Ingo Kro pac); Hietorical Institute, Hebrew University Jerusalem (Dr. Michael Toch); Institut fiir mit telalterliche Realienkunde, Krems/Donau (Dr. Gerhard Jaritz); Stadtarchiv Regensburg (Dr. Heinrich Wanderwitz / Bettina Callies); project-group „Migration und horizontale Mobilität im Mittelalter“ (Dr. Peter Teibenbacher, Graz)
28

to test, how the distribution d the casual use of machine readable source material can best be organized.
Address all communications to:
Manfred Thaller Max-Planck-lnstitut für Geschichte Hermann Föge Weg 11
D-3400 Göttingen
29
MEDIUM AEVUM QUOTIDIANUM
newsletter 10
Krems 1987
Herausgeber: Medium Aevum Quotidianum. Gesellschaft zur Erforschung der mate riellen Kultur des Mittelalters. Körnermarkt 13, A-3500 Krems, Österreich. – Für den Inhalt verantwortlich: Univ.Pro{. Dr. Harry Kühne!. -Druck: HTU-Wirtschaftsbetrieb Ges.m.b.H., K sg se 16, 1040 Wien.
INHALTSVERZEICHNIS / CONTENTS
Vorwort ……………………………………………………………….4
Manfred Thaller, The Daily Life o{ the Middle Ages, Editions of Sources and Data
Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Ingo H. Kropac und Peter Becker, Die Prosopographische Datenbank zur Geschichte
der südöstlichen Reichsgebiete bis 1250 (PDB): Konzepte und Kurzdokumentation . . . 30 Brigitte Rath, Auf der Suche nach Alltagen – der ‚Frauenalltag‘ im MittelaHer oder
„Leicht hatten es die Frauen nicht“ . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Rezensionen und Berichte . . . . . .. . . . .. . . . .. . .. . . … . . . …….. . .. . .. . . . . . . . . . . . . . 57
3
Vorwort
Dae vorliegende Heft 10 von Medium Aevum Quotidianum- Newsletter widmet sich einem Aspekt, der in der internationalen historischen Diskussion seit geraumer Zeit einen entscheidenden Schwerpunkt darstellt: der Anwendung computerunterstützter Methoden.
Es braucht nicht gesondert betont zu werden, daß in der Alltagsgeschichte, und nicht allein in jener des Mittelalters, diesbezüglich ein bedeutender Nachholbedarf zu konstatie ren ist. Doch auch in unserem Fachbereich mehren sich Einsicht und Zeichen, daß EDV mehr bieten kann als Textverarbeitungsprogramme zur schnellen und billigen Produktion von Publikationen. Ein entscheidender Vorstoß in diese Richtung ergibt sich beileibe nicht nur für den deutschsprachigen Raum auf Grund der Initiative des Max-Planck-lnstituts für Geschichte in Göttingen und dessen EDV-Spezialisten Manfred Thaller. Wir sind des halb besonders froh, gerade ihn für einen sehr grundsätzlichen Beitrag in Medium Aevum Quotidianum-Newsletter gewonnen zu haben. Auf Grund der internationalen Relevanz des Aufsatzes haben wir uns entschlossen, diesen in englischer Sprache wiederzugeben.
Der zweite umfassendere Beitrag des Heftes, verfaßt von Ingo Kropac und Peter Becker (beide Graz), widmet sich einem Anwendungsbeispiel, dessen inhaltlicher Schwerpunkt zwar auf dem Gebiet der mittelalterlichen Prosopagraphie zu suchen ist, welches jedoch nicht nur wegen der angewandten Methoden auch für die Alltagsgeschichte des Mittelalters von Wichtigkeit erscheint.
Beide Beiträge sollen den Au akt darstellen für eine in unserer Reihe fortzusetzende Diskussion zur Anwendung computerunterstützter Methoden.
Der bereits seit längerem angekündigte Band mit den Ergebnissen der von Medium Aevum Quotidianum im Oktober 1985 mitveranstalteten Tagung „Migration und hori zontale Mobilität vom Mittelalter bis zum Ende des Ancien Regime“ be ndet sich im Druck (Campus-Verlag, Frankfurt am Main- New York) und wird noch heuer (für unsere Mitglieder als MAQ-Newsletter 11/12) zur Auslieferung gelangen.
Ein besonderer Hinweis muß auf den bereits im vorigen Heft erwähnten Kongreß gegeben werden, den Medium Aevum Quotidianum in Zusammenarbeit mit dem Institut für mittelalterliche Realienkunde der Österreichischen Akademie der Wissenschaften vom 27. bis 30. September 1988 in Krems veranstalten wird. Er befaßt sich mit dem Thema „Mensch und Objekt im Spätmittelalter und in der frühen Neuzeit. Leben -Alltag – Kultur“ und läßt im Rahmen der Diskussion von international anerkanntesten Fachleuten vor allem im theoretischen und methodischen Bereich neue Ansätze und Wege erwarten. Voreinladungen an unsere Mitglieder wurden bereits vor einiger Zeit versandt.
Wie gewohnt wird sich einer der im Jahre 1988 erscheinenden Newsletter den Zusam menfassungen der genannten Kongreß gebotenen Referate widmen. Darüber hinaus wird, wie bereits zuvor angekündigt, Teil II der Auswahl-Bibliographie zur materiellen
4
Kultur des Mittelalters erscheinen. Im Planungsstadium be nden sich einige He e, die sich mit der Erforschung von Alltag und materieller Kultur des Mittelalters in einzelnen Ländern auseinandersetzen werden und einen Schwerpunkt insbesondere auf Forschungs und Literaturberichte legen sollen. Das erste Heft dieser Reihe wird vermutlich Dänemark gewidmet sein.
Gerhard Jaritz, Schriftleiter
5

Medieval and Early Modern Material Culture Online

The Daily Life of the Middle Ages, Editions of Sources and Data Processing

The Daily Life of the Middle Ages, Editions of Sources and Data Processing

' . WSA_translate_string( 'Abstract' ) . '

' . WSA_translate_string( 'Abstract (englisch)' ) . '

' . WSA_translate_string( 'Abstract' ) . '

' . WSA_translate_string( 'Abstract (deutsch)' ) . '