Given the inclusion of authority="naf" in the example's attributes and 
Karen's reference to the "'real' title of the work" (emphasis on 
"work"), there are at least three representations of the title that need 
to be considered. For example, consider an Arabic edition of the tales 
of Scheherezade. There would likely be an Arabic title for the edition, 
and (for an English-language users) a transliteration of that title, and 
the uniform title from the NAF for the work, "Arabian nights." All of 
these are appropriate forms of the title for a description for an 
English-language user population, and each is based on a different set 
of rules. Note also that the authority="naf'" attribute would refer 
meaningfully only to "Arabian nights," since NAF authorities do not 
currently authorize any vernacular forms, do not follow any 
transliteration rules uniformly for authorized headings, and can contain 
multiple variant, transliterated, and vernacular alternate forms for the 
one authorized heading.

The same would apply to names. Maybe within the larger "name" or "title" 
element there need to be nested types of each of these 
(titleTranscribed, titleTranliterated, titleAuthorized) to each of which 
the more specific attributes (transliteration rules, authority source) 
could be assigned. Assigning authority="naf" at the top just makes the 
whole thing very confusing.


Karen Coyle wrote:
> Jon Stroop wrote:
>> Language and script are definitely separate beasts.  Karen, are there 
>> situations where the combination of @lang and @script (we're using 
>> ISO 15924 codes for scripts here) don't cover what you are pointing 
>> out?  Probably so, and I've just been lucky thus far.
> It's not a question of language v. script, but titles that use more 
> than one script. You can have a title in Latin script with one word 
> using the Greek script. (Or one letter - lamda calculus, as an 
> example). Japanese is often written combining characters from the 3 
> different Japanese scripts, and it appears that they now throw in 
> Western words using Latin script.
> Using Unicode (which I hope we all are!), it's generally considered 
> unnecessary to mark scripts because Unicode handles them all, even 
> mixed together, smoothly. So if you think about what you are wishing 
> to say about your titles, it's not the script that they are in, but 
> whether you wish to display the original title, regardless of script, 
> or a transliterated substitute. If, in your library, you transliterate 
> to Latin script, then you may be able to identify that through the 
> transliteration standard or scheme you use.
>> Regarding the transliteration, I agree with Dan, the snippet I sent 
>> around is probably too Latin-centric.  But, as far as one version 
>> being privileged over the other, shouldn't one version (whether 
>> transcribed from the resource or taken from the preferred form in an 
>> authority file; probably in the native script) be privileged?  Maybe 
>> not, I really don't know, but would like to hear.
> You may want to privilege the original; you may want to privilege the 
> transliteration, depending on the needs of your users. In any case, 
> you need to know which is the title from the piece. In an ideal world, 
> the title proper, as determined by the cataloging rules, would always 
> represent the piece, and any other titles would be identified and 
> characterized. It hasn't been done that way in our cataloging past 
> because we didn't have a way to represent the titles except through 
> transliteration. I think this past practice has confused the issue, 
> because we allowed transliterated titles to be entered as if they were 
> the "real" title of the work, and we didn't code them as 
> transliterations. (Hmmm, I wonder how RDA handles this...?)
> kc
>> Also, to fuel the fire against my original suggestion a bit more, 
>> it's not good design for attributes to provide metadata about other 
>> attributes, so having (@transliteration or @romanization) and 
>> @transliterationStandard on the same element isn't the right way to 
>> go either.
>> On the other hand, presumably version elements would be repeatable, 
>> and could easily be misused.  What about something more specific:
>> <name type="personal" authority="naf" lang="ara">
>>   <namePart>
>>       <value script="Arab">طوسي، نصير الدين محمد بن محمد</value>
>>       <transliteration script="Latn" standard="?">Ṭūsī, Naṣīr al-Dīn 
>> Muḥammad ibn Muḥammad</transliteration>
>>   </namePart>
>>   <role>
>>      <roleTerm authority="marcrelator" type="code">aut</roleTerm>
>>   </role>
>> </name>
>> "value" may not be the right name for that tag, but I don't like 
>> "transcribed" or "preferred" either.  Ditto for title and note.
>> -Jon
>> Jon Stroop
>> Metadata Analyst
>> C-17-D2 Firestone Library
>> Princeton University
>> Princeton, NJ 08544
>> Email: [log in to unmask]
>> Phone: (609)258-0059
>> Fax: (609)258-0441
>> Karen Coyle wrote:
>>> I'm not sure the "scripts" solution works. You can have a title that 
>>> is a mixture of scripts. It seems to me that the distinction is 
>>> between transcribed (from the piece) and transliterated.
>>> From Unicode page on supported scripts:
>>> "In many cases, a single script may serve to write tens or even 
>>> hundreds of languages (e.g., the Latin script). In other cases only 
>>> one language employs a particular script (e.g., Hangul, which is 
>>> used only for the Korean language). The writing systems for some 
>>> languages may also make use of more than one script; for example, 
>>> Japanese traditionally makes use of the Han (Kanji), Hiragana, and 
>>> Katakana scripts, and modern Japanese usage commonly mixes in the 
>>> Latin script as well."
>>> kc
>>> Dot Porter wrote:
>>>> I second Dan's suggestion - using an element with @script attribute
>>>> for each original/transliteration of a name (vel sim.). Not only does
>>>> this not privilege one 'version' over the other, it also allows for
>>>> more than two 'versions', multiple possible transliterations into
>>>> scripts other than the roman script.
>>>> Dot
>>>> On Fri, Aug 14, 2009 at 8:13 PM, Dan Matei<[log in to unmask]> wrote:
>>>>> -----Original Message-----
>>>>> From: Jon Stroop <[log in to unmask]>
>>>>> Date: Fri, 14 Aug 2009 09:46:42 -0400
>>>>> I would prefer:
>>>>> <name type="personal" authority="naf" script="Arab" lang="ara">
>>>>>    <namePart>
>>>>>        <version script="Arabic">طوسي، نصير الدين محمد بن 
>>>>> محمد</version>
>>>>>        <version script="Latin" transliterationStandard="?">Ṭūsī, 
>>>>> Naṣīr al-Dīn Muḥammad ibn Muḥammad</version>
>>>>>    </namePart>
>>>>>    <role>
>>>>>       <roleTerm authority="marcrelator" type="code">aut</roleTerm>
>>>>>    </role>
>>>>> </name>
>>>>> In this way: a) no (even more) privileged position for the Latin 
>>>>> script, b) the transliteration standard is revealed.
>>>>>> -Jon
>>>>> Dan Matei
>>>>> ----------------------------------------------------------------------- 
>>>>> Dan Matei, director
>>>>> CIMEC - Institutul de Memorie Culturala [Institute for Cultural 
>>>>> Memory]
>>>>> Piata Presei Libere nr. 1, CP 33-90
>>>>> 013701 București [Bucharest], Romania,
>>>>> tel. (+4)021 317 90 72; fax (+4)021 317 90 64

Stephen Hearn
Authority Control Coordinator/Head, Database Management Section
Technical Services, University Libraries, University of Minnesota
160 Wilson Library
309 19th Avenue South
Minneapolis, MN  55455
Ph: 612-625-2328 / Fax: 612-625-3428