Within the framework of the project HerCoRe1 we are analyzing two historical works from 18th century “History of Rise and Decay of the Otoman Empire” and “Description of Moldavia” (both written by Dimitrie Cantemir) and investigate them with regard to the historiography of its time. We evaluate the usage of sources by the author and also the reliability of his references. We also seek to shed more light on the motivation behind the writing-process of these works by taking into account the political and cultural dynamics of the time and the position of Cantemir within the Ottoman elite. To determine missing or incorrectly translated parts of the work, the German and English translations are also compared with a copy of the Latin manuscripts. This comparative approach serves also to discuss the causes of the (un)conscious mistakes and omissions in the translations. We are performing this study by means of hermeneutic and IT approaches.
their request, he produced the two books which are the target of this
proposal:
- Descriptio antiqui et hodierni status Moldaviae, written in Latin, a
history of his country in which he describes not only pure historical
facts but also traditions, the language, as well as the political and
administration system. Local denominations and toponyms, as well
as names are written in Romanian with Latin script as his intention
was to demonstrate the Latin origin of his folk. The transcriptions
are not standardized and one retrieves for the same toponyms,
several name variations. Quotations as known today were very rare,
there is no bibliography. According to [
Given the fact that both works became standard reference for western authors until the middle of XIXth century, it is expected that their reception influenced also following historical material. There is no reprint/new edition of his works in German or English. There are, however, several reprints of the Romanian versions. Recent Romanian translations of Decriptio Moldaviae are done after the original Latin manuscript.
A lot of works were dedicated to the personality of Dimitrie Cantemir and its perception in different parts of Europe. A study of the reliability and consistency of the historical facts (as they are described in the latin copies) and their translations is practically impossible to be done only with traditional hermeneutic methods. One needs expertise at the same time in Latin, German, English, Romanian, Turkish, to enumerate just the main languages used in the two books, which additionally sum up to a quantity of about 1000 pages. Both German editions are printed in “Fracture” (“black letter”) script, which nowadays is very difficult to be read.
Already in the 1920s it was demonstrated (by using only a selections of texts), that the translations are not respecting the original all the time. E.g. information sources indicated by Cantemir were omitted, because they seemed too unreliable to the translator.
In the XXth century researchers claimed that some of the sources, persons and
facts quoted by Cantemir were not existing at all (e.g. [
• • • geographic distribution of material (originals in libraries in USA and Russia; translations and copies all across Europe; most part of the quoted sources in Turkey), the multilingual character of the materials to be investigated (Latin, German, Romanian, English, Turkish at least) and
no study about the reliability and consistency of the original and the translations could have been performed until now.
In the HerCoRe project we propose a mix of hermeneutic and IT-methods in order to:
• compare the Latin copies and the English and German translations, identify translation mistakes or gaps (made by purpose or not), search after the quoted works and identifiy related Ottoman sources, analyse Cantemir‘s writing and discourse style, assess the importance of the work in the Ottoman studies and compare them with other works contemporaneous to Cantemir or followup research about the Ottomans, develop electronic resources which may be of use for follow-up work about the Ottoman empire and the history of Balkans.
The hermeneutic investigation concentrates on the identification of sources quoted directly or indirectly by Dimitrie Cantemir, as well as the mentioned places, persons, events and dates.
The two works are very different with respect to the quotation style. While in the “Description of Moldavia” the quotation sources are almost missing, in the “History of rise and decay of Ottoman Empire” the author refers explicitely to different sources. However, there is no quotation style like in modern scientific works. Most references are real quotations or the author indicates the source of quotation through syntactic phrases followed by a reformulation of the semantic substance of a text section. Especially these cases are subject to the hermeneutic investigation.
By now we identified the main works quoted by Cantemir. These works are available only in paper form and are written in Ottoman Turkish (with Arabic alphabet) thus only a manual comparison can be performed. This systematic comparison led to a very unexpected result: we observed that linguistic expressions of certitude (e.g. “for sure”, “without any doubt”) are not an unambiguous indicator of the reliability of the quotation. E.g.: Cantemir is sure that all investigated sources mention 4 sons of Sultan Bayazid. However, all reliable sources of the time mention that the sultan had five sons.
We do not know why these inconsistencies occur. One possible motivation is the context in which he wrote the two books: in exile in St. Petersburg, probably with few notes at hand, that he made in Istanbul. Whilst we cannot find the reason of the inconsistency, the hermeneutic analysis showed that a pure automatic annotation (searching for quotation marks) will not help in the case mentioned above, as the semantics of the quotation mark does not match the degree of reliability of the quoted information. This is something which cannot even be inferred by automatic methods; at least not at this stage, where documents in Ottoman Turkish are rarely digitised and no linguistic tools are available for this historical language variant.
A second part of the hermeneutic investigation concerns the collection of persons, places, and domain specific concepts which are mentioned by the author. An automatic identification is practically impossible, as names are not standardised (e.g. for the city of Iasi in Moldavia we identified at least 12 writing variants).
The third part of the hermeneutic investigation is concerned with the identification of missing paragraphs in the German and English translation. First result: all paragraphs written in the Latin original with Arab characters were systematically omitted. This leads to misunderstandings in the two translations. 3.
Digital methods can facilitate analysis on the reliability of translations but
also of the historical facts claimed by the author [
For the detection of linguistic vagueness we follow a multilingual
approach. We collected the above listed indicators in the three languages
involved in the project (Latin, German and Romanian). Based on [
The initial collections of linguistic indicators are enriched through synsets in the corresponding Wordnets.
The knowledge base backbone is ensured by a fuzzy ontology modelled in OWL2. We distinguish between fixed concepts and relations (like geographical elements: river, mountain, island) and notions for which several “contexts can be defined. E.g. a geographical notion like “Danube” is within one historical context a border of the administrative notion “Ottoman empire”, and in another one the border to the so called administrative notion “Roman empire”. The historical contexts are specified by further fuzzy data properties (e.g. time, placement).
4.
In “The History of the Growth and Decay of the Ottoman Empire (1734)”
Cantemir tells the story of a battle between the Moldavian Prince Ștefan and
the Ottoman Sultan Bayezid I. Cantemir does not give the exact date of the
battle, but one could think that this can be inferred from other details of the
text. Ștefan attacked the Ottoman camp in Rasboeni. According to the account
of Cantemir, in this first confrontation, Ștefan lost the battle and retreated to
his castle in Neamț. Then, after the inspiring speech of his mother in Neamț,
he drew his soldiers together and stroke back the Ottoman army twice in
succession. After the last defeat in Vaslui, Sultan Bayezid fled back to Edirne [
This account of Cantemirs includes several problems, which can mislead the reader and even a historian who does not have detailed knowledge about the Ottoman and Romanian (Wallachian and Moldavian) history.
Known and proven historical facts: • There were two sultans with the name Bazeyid in the history of Ottoman Empire Bayezid I and Bayezid II but only the first one had the additional name “Yildirim” i.e. the Thunderbold. Cantemir mentions exactly this appellative and not the numbering (I or II) so we can exclude any typo or damaged spot in the manuscript. We checked this information in all translations and the Latin facsimile. • The reign time of Bayezid I is known for sure (according to diplomatic text sources): 1389 – 1402. • The frontiers of the empire at that time leaned already towards the Danube River and the Ottoman Empire was yet neighbor to Wallachia and Moldavia, thus a military confrontation with both principalities is historically possible.
During this time Wallachia was ruled by several princes: Mircea I (1386-1395), Vlad I (1394-1397) and then again by Mircea I (1397-1418).
The Ottoman chronicles report on a battle in 1391 between the
Wallachian Prince Mircea and Bayezid I in a place called Arkaş (in
Romanian Rovine). According to the Ottoman historians, Bayezid
won the battle and Mircea recognized the Ottoman sovereignity
[
At the time of Bayezid Wallachia was ruled by Mircea I
(13861395), Vlad I (1394-1397) and then again by Mircea I (1397-1418).
In Moldavia there was just one Ruler called Ștefan (Ștefan I
13941399)
There was a Moldavian ruler Ștefan III (1457-1512) who defeated
the Ottomans in 1475 after a loss in Rasboieni, and who defeated
also the Hungarian King Matthias (known as Matthias Corvinus
[1443-1490]). Moreover, this ruler had a son called Bogdan who
made Moldavia tributary to the Ottomans. These facts are
confirmed by Ottoman chronicles [
Class Ruler and hasName ‘Bayezid’ and ‘has additionalName Yildirim’ and hasBattlesIn some (Class Principality and belogsTo some (Historical Region and (liesIn value NorthDanube) -> Moldavia and Wallachia Class Ruler and was RullerOf exactly Moldavia and had BattleWith value ‘Bayezid Yildirim’ -> this will show all rulers which fulfil the criteria.
At a closer look there is a strong mismatch between Cantemir and all other established chronicles, but a historian would not know here how to interpret the text: • Is it referring to a battle against Moldavia or Wallachia? • Which Ruler opposed Bayezid Yildirim? • Where took the battle place?
An historian using only traditional methods (source inspection, reflection, re-evaluation of text) will face here a bunch of unsure and contradictory information, very difficult to resolve. An historian with less background knowledge about the Romanian history will be tempted to interpret wrongly the text section, either choosing the wrong rulers or the wrong place.
The HerCoRe System aims at helping historians in their interpretation, and suggests different reading paths. From the ontology and additional annotations the following inferences are possible: •
Class Ruler and wasRulerof value ‘OttomanEmpire’ and has Name Bayezid and has Additional Name Yildirim -> Bayezid Yildirim 1389-1402
Continuing this queries to the ontology, or invoking a complex inference chain the system will propose following solutions: • The paragraph is about a battle in Wallachia in a place called Rovine and against a ruler Mircea I. -> contradicts Cantemir text: (it is not a Moldavian king but a Wallachian). The user may infer also that the place called ‘Rasboe’ by Cantemir might be the place known as ‘Rovine’ • The paragraph is about a battle in Moldavia against the Ruler Ștefan I, and Cantemir mistakes the information about Ștefan I for Ștefan III • The paragraph is about the battle in Răsboieni against the Moldavian prince Ștefan III and the mismatch here is about the Sultan (Bayezid I or Bayezid II). Thus, the battle place here called by
Cantemir ‘Rasboe” would be ‘Răsboieni’.
All this information will have attached a score indicating a degree of truth. The latter one e.g. will have a lower score when introducing an additional scoring parameter: the metadata. The metadata will say, that the mentioned paragraph is within a chapter about the sultan Bayezid I, so it is less probable that Cantemir mistakes the name of the Sultan.
HerCoRe System does not aim at proposing a final solution. This decision is left entirely to the user/researcher, who is the hermeneutic subject and also can store the inference paths presented above as motivation for his choice. 5.
In this article we intend to show how hermeneutic and IT methods can be combined in order to investigate the reliability of historical texts (original and their translations). We show that a deep analysis can be performed only by the combination of the two approaches. Current research focus on the semi-automatic annotation and development of the ontology. Further work concerns the implementation of the reasoner and the visualisation of results.
In our project we need the cooperation between computer scientists, linguists (Latin, German and Romanian) as well as researchers in turcology. We cooperate also very close with the editor and translator into Romanian of the two works. The cooperation already revealed to date interesting aspects: - The simple availability of raw texts in digital form does not help. E.g. the
German translation of the History of Ottoman Empire is available from the German Text Archive. However, the text was digitized for visualisation purposes. The usage of the underlying text versions leads to an unsorted mixture of paragraphs written by the Cantemir, his side notes and the comments of the translator. These parts, marked correspondingly in the TEI version are melted in the .TXT version. We had to invest additional work on separating these distinct parts.
Mark-up in the editions have to be considered, as they can enrich the text. However, first one has to know the semantics of the mark-up. The ontological formalisation of the notions mentioned in the text was a great help for the humanist researchers leading to a better reflexion of the used notions.
Many of the computational linguistics approaches had to be revised given the particularities of the historical text.