lou.burnard at computing-services.oxford.ac.uk
Sat Jun 24 06:52:46 EDT 2006
Laurent.Romary at loria.fr wrote:
> Selon Lou's Laptop <lou.burnard at computing-services.oxford.ac.uk>:
>> Laurent Romary wrote:
>>> Le 21 juin 06 à 00:03, Lou's Laptop a écrit :
>>>> 1. CC defines both <person> and <personGrp> as potential members of
>>>> <listPerson>. A <personGrp> is a "multi-headed" person treated as a
>>>> single individual, for example "the audience" , "a howling mob" etc.
>>>> It seems to me to make sense to treat <personGrp> in exactly the same
>>>> way as <person> therefore, but it was not proposed in persw06. Anyone
>>>> object if I add it? It is definitely needed for spoken corpus work
>>>> and I would have thought useful in other situations too.
>>> When I look at the actual descriptions and meanings of personGrp and
>>> person, I just see the former as a specific case of the latter. Would
>>> not it be a good opportunity to simplify the picture by dropping
>>> personGrp and just keeping person with its recursive structure that it
>>> already has and just adding the couple of extra specificities hat
>>> personGrp had (namelly, "mixed" sex and size attribute (no jokes please)).
>> In what sense do you think that <person> has a recursive structure?
> Because you may describe a group either intensionaly (you would then use person
> as you used personGrp in the old days) or extensionaly, by allowing <person>
> (refering to a group) to contain recursively the <person>'s it is made of.
> The picture is clear in my mind, but I am not sure to be clear verbally...
> Indeed. I misread the content model of both element by wanting to see what I
> was expecting...
If you want to represent a group of people whose identities you know,
then you would use <listPerson>. "personGrp" is provided for the case
where you can't do that. For example, in a transcribed lecture, you
might have a <listPerson> to group all the attendees, containing a
<person> describing the lecturer, and maybe others describing the
particular students who intervened from the floor. You might also want a
<personGrp> to describe "the audience", to which you would attribute
(say) applause or booing, and concerning which you might be able to
supply some personal properties (e.g. sex or age or social status) as a
whole but certainly not names or persTraits.
I think it's misleading to pretend that a <persGrp> is the same as a
<person> although it does share some properties. Recursion isn't a panacea!
More information about the tei-council