On reflection I agree that we are talking about how you describe the
reference itself. And I'm quite happy to use <rs> to encode it.
My concern is [now] that your and Sebastian's replies suggest that the
use of the type attribute in e.g. <rs type="place"> is to be deprecated
("what type ought to mean"; "closer to correct"). A much more specific
point, I agree, but sometimes one needs the rapier-like responses
provided by the TEI list to get one's thoughts into order ...
This usage of <rs type> has been around since at least P3, and the
current Guidelines still give <rs type="person"> and <rs type="place">
as the two examples of <rs> usage. I don't see anything in the
Guidelines which suggests that for this element type, the type attribute
should mean "type of referring string" rather than "type of entity
referred to". P3 says it "indicates more specifically the [class of]
object referred to by the referencing string" (my addition). P5 doesn't
have specific words on this, but neither of us could marshal the P5
definition of the type attribute to our cause: it "characterizes the
element in some sense, using any convenient classification scheme or
One point that comes out of the discussion is that the use of your
favourite example in P5 13.2.3 ought to be changed, since it states
<rs key="NY1" type="place">modern Babylon</rs>
is equivalent to:
<placeName key="NY1">modern Babylon</placeName>
In message <[log in to unmask]>, Lou Burnard
<[log in to unmask]> writes
>I think you are misunderstanding the way the consensus is emerging.
>You *can* link your references to the things-they-refer-to, and
>standardising a way of doing that is a distinct step forward in what
>TEI offers. What we are talking about here is how you describe the
>In earlier versions of TEI (specifically the P4 version of the ND
>module), there was a vagueness about whether you could use e.g.
><placeName> and <persName> to contain ways of referring to places and
>people which were not actually names. The consensus now emerging seems
>to be, quite clearly, that you should not. You wouldn't say
><persName>her husband</persName>, so why should you say
><placeName>three miles from here</placeName>? In both cases, these
>should be <rs>s. And the great step forward is that we can in any case
>resolve the intended referent by means of @key and @ref, since these
>attributes are available on <rs> as well.
>In answer to my esteemed colleague's concern that this does "make it
>mildly harder to analyze the text", I can only reiterate that TEI
>policy, imffho, is always to prefer Truth above Expediency.
>Richard Light wrote:
>> In message <[log in to unmask]>, Sebastian Rahtz
>><[log in to unmask]> writes
>>> Lou Burnard wrote:
>>>> but for ones which don't, it seems much more useful to say
>>>> <rs type="/type of entity referred to/"> as Oyvind suggests than to
>>>>do what @type ought to mean, namely
>>>> <rs type="/type of reference/"> (with values such as "short",
>>>>"elliptical", "periphrastic", "cataphoric" etc.)
>>>> This is an old argument (which I lost last time we had it) ...
>>>>since we do now permit <name type="/type of thing named/"> it seems
>>>>only logical to extend the confusion to rs as well.
>>> ah, _that_ discussion.... how it all comes flooding back.
>>> so in practice we do all agree that Gabriel should use
>>> <rs ref=""> instead of <placeName ref="">, which will
>>> just make it mildly harder to analyze the text, but be
>>> closer to correct
>> Just to cheer everyone up on a grey Monday morning, I'm going to
>>disagree with all of this.
>> While the logic of the above responses may be impeccable, it does
>>beg the broader question of what we are trying to do with our TEI
>>markup. As I see it, in P5 we have developed the Names and Dates
>>module in two different directions. On the one hand, we have added
>>to our ability to focus on the "name-ness" of names, e.g. by
>>introducing <nym> and its friends. On the other, we have added a
>>whole battery of elements to describe people, places in their own
>>right (<person>, <place>, etc.).
>> The emerging consensus, as I understand it, is to say that we can't
>>validly link out from a reference to a person within a text to our
>>carefully constructed external biographical authority files, simply
>>because that reference is anaphorical. In other words, "name-ness"
>>trumps "referring-ness". This strikes me as a distressing step
>>backwards into a narrow linguistic interpretation of what TEI is all about.