Contents of the corpus
----------------------
The corpus consists of extracts from a number of texts:
    danish: 
	A:	Retsplejeloven (Chap 1)
        B:	Karnorvs lovsamling (Chap 2)
        C:	Gomard: Civilprocessen (Chap 4)
        D:	Dansk dokumentsamling (Chap 5)
    spanish:
	A:	Pietro-Castro: Tratado de derecho procesal civil (Chap 6) 
        B:	Ley de Enjuiciamiento Civil (Chap 7)
        C:      Spansk dokumentsamling (Chap 8)
    english:
	A:	The Rules of the Supreme Court
            	Supreme Court act
            	County Court Rules (Chap 9)
        B:	Engelsk dokumentsamling (Chap 10)
        C:	O'Hara & Hill: Civil Litigation (Chap 11)

We have tried to build a corpus of similar texts in as much as for
each language there is the law text itself, comments on the law, textbooks
and different kinds of documents from 'real life'. It is supposed to be 
a representative text corpus in the area of civil law.
We have chosen a specific area of civil law i.e. Civil procedure, courts of
first instance.

The corpus was contributed to the ECI project by:

          ****************************************
          *  Steffen Leo Hansen                  *
          *  Dept.of Computational Linguistics   *
          *  Dalgas Have 15,  DK - 2000  F.      *
          * ==================================== *
          *  e-mail: slh/id@cbs.dk               *
          *  phone: + 45 38 15 31 27             *
          *  fax: + 45  38 15 38 20              *
          ****************************************

Structure of the ECI corpus
---------------------------

The Civil Law corpus has been split into three subcorpora
corresponding to the three languages which it contains viz. Danish
(mda12), English (men12) and Spanish (msp12).

Each subcorpus has been split into a number of files corresponding to
the divisions given above, i.e. each .eci file contains all of the
extracts from one of the book chapters given above. These divisions
are marked in the corpus with:

	<div0 type=subcorpus>

Each file contains a number of extracts taken from the book. These are
marked by:

	<div1 type=extract n=NNNNNN> 

The NNNNNN are identification numbers given to the extracts by the
CBS. They are not neccessarily unique, ie more than one extract can
have the same identifier as some long sections have been split into
more than one extract.

Each extract consists of a bibliographic header, the text extract
itself, and a bibliographic trailer (which mostly repeats the
information in the header). The texts are marked with:

	<div2 type=text>

The text of the extracts is divided into paragraphs with

	<p type=cbs>. 

The type 'cbs' indicates that these paragraphs were marked up in the
original corpus from the Copenhagen Business School. <p> was used to
mark the start of those texts which did not begin with a CBS
paragraph. It may also sometimes contain notes marked with:

	<div3 type=note>

The structure of a text extract is thus:

	<div1 type=extract n=NNNNNN>
	   <bibl default=no>
	   ... header bibliographic entry ...
	   </bibl>
	   <div2 type=text>
	      <p type=cbs>
	      ... text ...
	      </p>
	      ...
	      <div3 type=note>
	         <p>
		 ... text of note ...
	         </p>
	      </div3>
	      ...
	   </div2>
	   <div2 type=trailer>
	      <bibl default=no n=NNNNNN>
	      ... trailer bibliographic entry ...
	      </bibl>
	   </div2>
	</div1>

where there may be multiple <p> and <div3 type=note> within the <div2 type=text>.

Corrections/Modifications to the corpus
---------------------------------------

The corpus originally contained many occurances of the character ctl-U
(\025).  These have all been quietly converted to the character §
(paragraph marker), as they usually appeared where this character
would have been appropriate. And in fact \025 is the code for §
in the IBM Code Page 850 character set.

Each line in the original corpus ended with the "<" character. These have been
removed.

A number of obvious mistakes have been corrected with either <corr> .. </corr>
or <omit> markup, with the original text as an attribute.