Issue 258: P72 Quantification

Starting Date: 
2014-09-13
Working Group: 
3
Status: 
Done
Closing Date: 
2015-02-12
Background: 

Posted by Dan Matei on 13/9/2014 
Friends 

I need to apply P72_has_language to: 

* E49_Time_Appellation (e.g. "Vingt et unième siècle"); 
* E48_Place_Name (e.g. "Paris", "Parigi", "Aachen", "Aix-La-Chapelle", "Aquisgrana"); 
* E75_Conceptual_Object_Appellation (e.g. "Querelle des Bouffons"http://fr.wikipedia.org/wiki/Querelle_des_Bouffons). 

But these classes are not subclasses of E33_Linguistic_Object, so ... 
I'm sure others solved already the problem. How (please) ? 

Posted by Martin on 14/09/2014 

Hi Dan, 

We use to solve this with multiple instantiation (E49, E33). Note that most place names or not language specific. Few bigger places use to have language variants. 
See also FRBRoo 2.0, about name use practice. 

 

Posted by Dan Matei 15/09/2014 
Hi Martin

On 14 September 2014 21:40, martin wrote: 
We use to solve this with multiple instantiation (E49, E33). Note that most place names or not language specific. Few bigger places use to have language variants. See also FRBRoo 2.0, about name use practice.
@ics.forth.gr>

Yes, but why using acrobatic solutions ? There is a strong reason against making those classes subclasses of E33_Linguistic_Object ? 
 

Posted by Stephen Stead 15/09/2014 

Yes a very strong reason. Not all are, so making them sub-classes is a lie. In those cases where they are then multiple instantiation is the correct solution. 

 

Posted by Martin 15/09/2014 
Hi Dan,

On 15/9/2014 8:37 πμ, Dan Matei wrote: 
Hi Martin 

On 14 September 2014 21:40, martin 
wrote: 

We use to solve this with multiple instantiation (E49, E33). Note that most place names or not language specific. Few bigger places use to have language variants. 
See also FRBRoo 2.0, about name use practice. 

Yes, but why using acrobatic solutions ? There is a strong reason against making those classes subclasses of E33_Linguistic_Object ?
@ics.forth.gr>

Well, we regard that many (most) Appellations do not have a language. If we accept that, making linguistic object a superclass of appellation just because sometimes the property may appear, violates all principles of generalization. Even if you have thousands of bilingual placenames, there are a million more which do not have a language equivalent, and many which have survived different cultures. 

Multiple instantiation is not "acrobatic", but a very important feature of KR models like RDF, giving credit to the fact that there are incidental combinations of classes on particular instances. For instance, I can make a spoon-knife, its a real spoon, a real knife, but nothing more to say about it as a category. Avoiding to subclass any combination of classes that may appear in some reality, is one of the fundamental principles that has kept the CRM as small as it is. In terms of implementation, the overhead is negligible, and not "acrobatic ". 

If we want to be more precise, it is not the name which is translated, but the place which is renamed. A place "St. John" in Canada is not called "Sankt Johann" by Austrians, nor vice versa. The cases in which the placename in two languages is unique is even more rare. 

 

Posted by Dan Matei on 16/9/2014 

OK. I withdrow the term "acrobatic" :-) 

 

Posted by Dominic 16/09/2014 

Good discussion bringing out an important point about multiple instantiation. 

 

Posted by Vladimir on 19/09/2014

We use to solve this with multiple instantiation (E49, E33).

This is a good solution. 
We had many examples of multiple instantiation in BM, esp of Events. 
E.g. often an Acquisition is also Transfer of Custody, Part Addition (to the new collection), Part Removal (from the old collection), maybe even Move. 

Note that most place names or not language specific. Few bigger places use to have language variants.

But I don't think that's a criterion on whether something is a Linguistic Object! 
If it was, every unilingual book without translation would NOT be a Linguistic Object. 

The criterion is the scope note: Linguistic Object "identifiable expressions in *natural language* or languages". 
Let's consider the clases given by Dan, taking into account the class hierarchy 
http://personal.sirma.bg/vladimir/crm-graphical/#cidoc_class_hierarchy 
* E49_Time_Appellation: is not, eg "20140919" is not in natural language. This comes from its E50_Date subclass 
* E48_Place_Name: I think it is!! 

Martin or Steve, can you give some examples of Place Names in unnatural language

The class name includes "Name", which suggests it is in natural language
The scope note "particular and common forms of E44 Place Appellation" is not helpful in making the distinction. 
Certainly its superclass E44 Place Appellation is not Linguistic Object, since it includes Coordinates etc 

* E75_Conceptual_Object_Appellation: "specific identifiers of intellectual products or standardized patterns."
The examples are not linguistic: ISBN 3-7913-1418-1, ISO2788-1986 (E)

e.g. "Querelle des Bouffons"

Dan, you should use E35_Title since P102 has title applies to E70_Thing, therefore also applies to E28_Conceptual_Object. 

But I fail to see the utility of E75_Conceptual_Object_Appellation: 
* for "specific identifiers" use E42 Indentifier 
* for names use E35_Title 

 

Posted by Stephen Stead 20/9/2014 

Martin or Steve, can you give some examples of Place Names in unnatural language?

Yes "K10" 
 

Posted by Christian Emil 20/09/2014 

Or K2, U2, R2D2 for that matter. On the other hand. As soon as a symbol is used to denote something and is used as a name (and us pronounced) , one may conclude that it has become a part of the vocabulary and thus is a part of a natural language. For a modern language user without a special interest for etymology and language history propria like K2 or Martin are names (words) in the natural language without internal meaning according to my in-house onomasticist Solveig. So the test is in the use and not in the form. 

 

Posted by Martin 20/09/2014 
Dear All, 
Of course it is debatable if anything we use in speech is regarded a Linguistic Object. I may however point you to the fact that E33 is defined as: 
"This class comprises identifiable expressions in natural language or languages." 

We exclude artificial languages of any kind. 

We understand (per default) that the direct properties of a class express the potential of its instances of having such properties. Of course, the question is not, as Vladimir remarked, if the translation exists, but if it has a natural language and if it has the potential to be translated. E33 has two properties: Has language and has translation. 

P72 is defined as "Linguistic Objects are composed in one or more human Languages. This property allows these languages to be documented." 

This poses an ISSUE: Following this, the quantification of P72 should be 1:n. 

Proper names are normally referred to in texts, but not translated, so we can argue that they do not belong to a particular language, but rather to the carrier. 

One could argue, that language equivalents of placenames are name use cases of groups, loosely bound to language, and not linguistic objects at all, once they are not expressions. For instance, German authorities may not use German placenames for the Balkan area anymore. The translation of Bei Jing is "Northern Capital", which would not be used in English. 

All Chinese and Japanese proper names can be translated, but the translation would not be the language equivalent. 

Another argument would be that a proper name is "translated", if it is phonetically/grammatically adapted to a particular language, for instance with a gender ending as "Στουτγαρδη" for Stuttgart in modern Greek, or Athens (plural!) for "Αθηναι". 

One could argue, that a proper name belongs to a natural language if it fits to its phonetic or symbolic (han characters). 

The Getty TGN refers to placenames used by the locals as being "vernacular", a nice solution I believe. 

I'd vote for the practical aspect, to denote a name as linguistic object of a language if translation into that language should take that into account. 

Opinions?? 

 

Posted by Øyevid Eide on 22/9/2014 
Two comments:

20. sep. 2014 kl. 18:58 skrev martin 
Dear All, 
[…] 
This poses an ISSUE: Following this, the quantification of P72 should be 1:n.
@ics.forth.gr>

This seems to be merely a typo. It says "many to many, necessary (0,n:0,n)" 

Proper names are normally referred to in texts, but not translated, so we can argue that they do not belong to a particular language, but rather to the carrier. […] 
Opinions??

I would say that a place name belongs to all languages in which it is used. When I say "München" in a Norwegian sentence it would make little sense to claim it is not a Norwegian word, especially when I pronounce it according to Norwegian rules (which happens to give a very similar pronunciation to High German (but different from Bavarian) but that is not relevant). What is the difference between saying München in Norwegian and saying harddisk in Norwegian? Both are (were) foreign words used in Norwegian (are foreign place names load words?). In some cases they have their original spelling (as München), in other cases the spelling is different (Tyskland for Deutschland). Same for pronunciation. 
However, it makes sense that place names are seen as linguistic objects only when complexities such as translation comes in, as you say. I would guess that documentation practice in museums is in line with this.

 

Posted by Martin 22/09/2014 
Dear Øyvind,

On 22/9/2014 12:00 πμ, Øyvind Eide wrote: 
Two comments: 
[…] 
This poses an ISSUE: Following this, the quantification of P72 should be 1:n. 
This seems to be merely a typo. It says "many to many, necessary (0,n:0,n)"

OK, we can correct that I believe without decision. Thank you! 
 

Proper names are normally referred to in texts, but not translated, so we can argue that they do not belong to a particular language, but rather to the carrier. 
[…] 
I would say that a place name belongs to all languages in which it is used. When I say "München" in a Norwegian sentence it would make little sense to claim it is not a Norwegian word, especially when I pronounce it according to Norwegian rules (which happens to give a very similar pronunciation to High German (but different from Bavarian) but that is not relevant). What is the difference between saying München in Norwegian and saying harddisk in Norwegian? Both are (were) foreign words used in Norwegian (are foreign place names load words?). In some cases they have their original spelling (as München), in other cases the spelling is different (Tyskland for Deutschland). Same for pronunciation.

I see two problems with your argument: First, "harddisk" is a proper noun, not a proper name. 
My argument was about proper names. I assume linguists do not regard foreign proper names as "loan words"? Any linguist here? 

Second, a smaller placename may appear in all languages in the same form, or by standard phonetic transliteration, such as "Bei Jing", formerly transliterated as "Peking". 
That would make the set of languages open ended. Therefore I'd propose, similar to your argument, that only placenames used in foreign countries that deviate from the vernacular form are regarded as language specific. 
That poses again a principle of current practice against a principle of substance. But in a way, it is consistent with translation practice I think: Only loan words historically introduced into a language would be used in a proper translation. Words without equivalent would be circumscribed in words of the language. Possibly the original term would be cited, but not regarded as part of the language. In case of proper names, the meaning is the thing named. 

"Deutsch" and "Tysk" I believe is more than a transliteration. I think it has a common etimological root,whereas "Alemania" pertains to the local German tribe at the south-western borders.

However, it makes sense that place names are seen as linguistic objects only when complexities such as translation comes in, as you say. I would guess that documentation practice in museums is in line with this.

I think so! 
 

Posted by Christian Emil 22/09/2014 

This has become a very long email correspndence. Languages: I have mostly knowledge about European languages. If proper names are linguistic objects in such languages, we need to include them as instances of E33. In many European languages a proper name can be inflected: in number, definite/indefinite and cases. Thus a name behave almost as any other noun in many languages. A proper noun can be used as a proper name and a proper noun can be used as a proper name (as place names for example). Participles and adjectives can be used as proper noun and proper names can be used as the first part of a compound etc. etc. So it should be relatively clear that a proper name can be a "linguistic object". 

A linguistic object in CRM may have more than 1 language, Scope note of E72 " Linguistic Objects are composed in one or more human Languages. This property allows these languages to be documented." What about a linguistic object consisting of a single word e.g. inscription) which can be a word in more than one language? The author may have done this intentionally or we as readers cannot tell which of the languages it was meant to be. 

Loanwords (written languages): pub can be written 'pøbb' in Norwegian. Genre is written 'sjanger'. If I write 'pub' in Norwegian (as most do) is that a English word or a Norwegian? A Norwegian name is Eirin (written version of the Norwegian pronunciation of the English Irene (in a popular TV series). Another variant is Irene with a Norwegian pronunciation letter by letter. (similar to the German). I Eirin Norwegian or English and what about Irene? 

Proper names: A Finnish proper names lexicon is divided into three parts: Finnish names, Swedish Names, Saami Names (based on their origin). Still a Finnish text with a proper name of Swedish origin in it will be considered as a Finnish text. Another example: The name in the written form 'Peter' can be used at least in German (?), Norwegian and English. 

In general the possible vocabulary of a language does not define the language. The word inventory of a living language is open. If a word or a name (originating in another language) is used in a language and felt as a natural part of that language it is a part of the language even though academies and language councils may object. 

Returning to P72: What is meant by " This property describes the E56 Language of an E33 Linguistic Object. Linguistic Objects are composed in one or more human Languages. This property allows these languages to be documented."? Do we mean "the words felt as natural part of the vocabulary by the speakers/writers of the language", "official orthography", "is a loan from a language", "is a loan but has been adapted to the pronunciation or spelling of the languages" and so on? 

Finally the ISO 639, the International Standard for language codes, badly needs revision. So what is a language?

Outcome: 

In the 32nd joined meeting of the CIDOC CRM SIG and ISO/TC46/SC4/WG9 and the 25th FRBR - CIDOC CRM Harmonization meeting, the sig changed the quantification. The issue is closed.

Oxford, February 2015