Issue 671: Example for propositional objects

Starting Date: 
Working Group: 
Current Proposal: 

Post by Martin Doerr (6 January 2024)

Dear All,

I suggest to create an example using Bekker numbers. They constitute excellent examples of identifiers for propositional content.


Post by George Bruseker (10 January 2024)

Dear Martin,

As a scholar of ancient philosophy, I do love Bekker numbers, but I am curious why they would be an example of propositional object. They are a reference to a particular chunk of text in the original Greek as setup in the Bekker edition. As such, I think as a scholar using ancient texts, I use it to locate the original Information Object upon which an interpretation (formulation of the proposition(s) that we think thinker X was making) is based. The exact propositional content of that information object is usually the subject of debate rather than the object of reference. Did Aristotle mean X or Y in passage 99a of the Posterior Analytics, is the usual topic of conversation. If we knew the exact propositional content, we'd be golden, but usually that is the very topic we want to endlessly swirl around and the Bekker number is the pointer for people who can read ancient Greek in order to be able to find the original passage, read it, translate it and cogitate on what was really meant there (the propositions encoded).

But perhaps you have another use in mind?



Post by Achille Felicetti (10 January 2024)

Dear George,

What I read in Martin's email is that Bekker numbers are examples of identifiers for propositional content, not propositional objects themselves.

Which, it seems to me, is not so far from your thoughts :-)



Post by George Bruseker (10 January 2024)

Dear Achille,

Yeah, clearly the Bekker numbers themselves are not propositional objects but identifiers. But I don't think that they are identifiers for propositional objects. They are identifiers for chunks of text in an edition, an information object which has a series of symbols and may encode one or more propositions. The Bekker number literally points to a section of a text as an edition, so the principle of identity is symbols not propositions. There is no correlation between a Bekker number and a proposition or set thereof. The Bekker number can point you to a passage in the 'original' Bekker edition and it is used for cross correlation when you want to see where your translation might match to the Bekker (so that you can read the original Greek and see if you think that is what so and so is saying).  So it is generally a modelling situation of E73s composed of (p106) E73. If it were an identifier for a proposition then the principle of identity would be to break down the propositions (noetic content) that are encoded by the words and then give them identifiers (the propositions not the encoded symbols) and then put them together by P148.

Or at least this is how I know Bekker numbers to be used in actual practice. In a work on ancient philosophy if you are going to refer to Aristotle or some other author and their passages, you don't quote Penguin English, you quote the Bekker. And then, you say what it means (the propositions).



Post by Martin Doerr (10 January 2024)


Dear George,

Yes, I am very much aware of what you are describing and completely agree. I am right now looking for the original text. The text itself in Bekker's edition constitutes a Symbolic Object with propositional meaning, an Expression in the sense of FRBR.
The search for precision is one aspect of what we do.

The other aspect is accepting a certain fuzziness. The class E89 Propositional Object was introduced to capture the sense of FRBR Work, which, in one interpretation, constitutes an abstraction of meaning from the symbolic form, in particular from translations.

As "knowledge engineer" I just neutrally observe, that sufficient people support the idea of some sort of preservation of meaning across translations, and others vehemently oppose. In the christian theological background, authorized translations are regarded as "the Word of God", i.e., transferring an even identical and in any case comprehensible meaning, which, within this tradition, must not be questioned. Medieval theological and philosophical tradition was widely using Aristotle in Latin translation without questioning the essential transfer of meaning by the Latin text.

We need also not forget that early Latin (and Arabic) translators were much closer to the common senses of the ancient Greek world. As such, our ability today approximating the Greek original meaning from its linguistic expression only may not necessarily be superior to consulting also relevant translations.

As such, my position about the preservation of meaning across translations is an observational one.

I assume you agree, that undeniably scholars around the world cite such texts in translated form, and refer via Bekker identifiers in their citations, often without referring to the translator at all (regarded as "editor" and not "author" as I just read in a scholarly text !), expressing that they mean the intended meaning of the corresponding original, approximated by the translation provided.

Since the CRM project is not about absolute precision, but about "minimal ontological commitment" in the sense of Thomas Gruber, for the purpose of information integration, rather than resolution, I maintain that we need to model two different senses:

A) the actual intended meaning, which is over thousands of years more and more approximated by scholarly commentaries, and

B) the minimal common or approximate meaning, as rendered by several good translations.

I would model A) as instance of Information Object, as it gives priority to the original wording, implicitly Propositional Object as intended by the author, as you correctly stress in your message below,

and B) as E89 Propositional Object only, as E89 is about meaning possibly abstract from symbolic form.

The latter sense should be expressed in the example. I propose to talk about the

approximate meaning of Met.Г 4.3,1005b 19-20,

and add a comment with translations in 3 languages and the original. I currently have a German and two English ones (below) at hand:

“the same thing cannot at the same time belong and also not belong to the same thing and in the same respect”
"It is impossible for the same attribute at once to belong and not to belong [20] to the same thing and in the same relation; 
(Met.Г 4.3,1005b19-20)

Thus stated, users can make up their own mind about the common meaning in this example, isn't it?

Would that find your agreement?



Post by George Bruseker (13 January 2024)

Dear Martin et al.,

If what you mean by using the practice of referencing Bekker numbers as an example for propositional object is to create a complex example that spans multiple classes and properties and illustrates the interrelation between information object and propositional object and potentially symbolic object, that can be a nice illustration of that complex of classes and properties in CIDOC CRM which many people struggle to understand. 

That said, the example should then probably be drawn from a real world scenario. The Bekker numbers are identifiers for an information object not a propositional object. So one could talk about the imagined meaning of the passage according to one author but this would not be a super convincing example. As I mentioned people tend to be fundamentally contesting the meaning of the passage (it is not the case like in positive science that we are slowly over time converging to the one agreed meaning of Plato or Aristotle that everyone believes is the case), arguing that one or the other thing has a completely different propositional content.

So if we wanted a good example on this, we should try to look for a passage with a well known debate around it where important scholar A contends that this one information object with Bekker number X has meaning Y and then scholar A contends that the same information object with Bekker number X has meaning Q. Y and Q can be our examples of propositional objects.

I obviously don't have an example in my mind at the moment, but such arguments are pretty much the heart of interpreting ancient texts, so  it shouldn't be hard to find one.

Would that find your agreement?


Post by George Bruseker (13 January 2024)

Other fun examples from philosophy could be the Dao De Jing for which there are radically different translations that assume different propositional content. Wittgenstein scholarship also has, I believe, extensive archival numbering system and a developed commentary practice around the number information objects ascribing different propositional contents to them. Could go on a hunt for some of those.