[tei-council] the "key" attribute

Lou Burnard lou.burnard at computing-services.oxford.ac.uk
Mon May 21 17:35:06 EDT 2007


Sebastian Rahtz wrote:
> Lou Burnard wrote:
>> How do we feel about pos tagging? At present the Guiidelines say you 
>> have to do things like
>> <w ana="#NOUN">bum</w>
>>
>> and point off stage to some definition for noun (an <interp> or a 
>> <fs> or something).
>>
>> But why shouldn't I do
>>
>> <w key="NOUN">bum</w>
>>
>> meaning "Everyone knows what a NOUN is, but if you insist on finding 
>> out use this magic string 'NOUN' and some mechanism external to this 
>> document will enlighten you"
> If <name target="#Napoleon">Boney</name> is right, I assume your 
> example should be <w target="#NOUN"> by
> the same argument.

Well, @ana is defined for this very purpose, so no, I wouldn't use @target.
>
> It is a slippery (and seductive) slope if we appear to let ID/IDREF 
> back in, which is what
> will happen if we allow key="NOUN", cos people will use it to point to 
> that <interp>,
> instead of "the universe".
>
But that's Wrong. If you want to point to the <interp> you need a 
pointer, not a key.

> I am fairly sure we should switch all the @keys on objects to @target 
> forthwith,
> and then consider separately whether to allow @key back in.
>
I'm happy with turning all the @keys on <personName> <placeName> ect 
into @target s (or @ana, or @ref) 

Actually @ref might be clearer. It's what we use for <g>, which is very 
similar.

I have no problem leaving @key in places where it's clear that a URI 
won't fit, i.e. where you might need to do some more processing to turn 
the @key value into what you want. But I don't see why <w key="noun"> 
might not fall into that category.






More information about the tei-council mailing list