[Corpora-List] ParaConc and Grice

From: Santos Diana (Diana.Santos@sintef.no)
Date: Tue Sep 10 2002 - 14:08:33 MET DST

  • Next message: Kiril Simov: "[Corpora-List] Treebanks and Linguistic Theories 2002 - Programme"

    Dear colleagues,
    I realized (sigh!) that my previous mail on aligners could also be
    interpreted as a specific criticism of ParaConc.

    I had not that in mind, since I don't even know the last version of it.
    So let me try to counterbalance the (possible) harm done by pointing out two
    things:

    - the approach I suggested for alignment, with EasyAlign, could actually
    also be used to produce aligned texts for ParaConc

    - Michael Barlow has told me that parallel corpora providers can use
    ParaConc embedded in their CDs, and this is obviously a good (probably even
    better) alternative to having the resources on the Web. IMS-CWB samplers
    also exist that are run from a CD, but here again we have the irreducible OS
    difference, the former running in Windows and the latter in Unix.

    It is actually possible that the two systems (IMS-CWB and ParaConc) actually
    share a number of interesting features in dealing with parallel corpora,
    like the support for different languages and character sets and the display
    of several translations/languages in parallel (not only two). For what I
    know, they may be not so different in capabilities dealing with parallel
    corpora after all (except that IMS-CWB does have an aligner and not
    ParaConc, and this is why all this mail exchange started). I have no way of
    looking into this now, but maybe Sampo will, after all...

    And now some Gricean comments:
    Should one answer just the question that was posed, or try to help the
    person who is asking it, BEYOND what was actually asked?
    And list-Gricean: should one answer to the person who asked the question to
    the list, or should one answer also the potential readers with a "parallel"
    question?

    To be honest, I had someone telling me that alignment was a
    difficult/unsolved problem the other day, with the justification of the
    "recent discussion on the corpora list". That was why I became aware of the
    (potential) problem of posting answers to a particular question. And decided
    to post to the list.

    And, by the way, I know many other aligners... Not everyone who is working
    on aligment is interested in sharing their programs with other people,
    though. [And nobody should feel forced to do it, either!] So, I think it is
    quite good that there are commercial programs out there, where by definition
    people is interested in "sharing" the programs with us. I see absolutely
    nothing wrong in discussing commercial systems in this list. What I reacted
    to was the narrow subject: "aligners for a particular commercial system,
    without reference to alignment in general and parallel corpora systems in
    general".

    And I definitely promise that even if the present message is thoroughly
    misunderstood, I quit here and won't post more on these subjects to the
    corpora list.

    Diana



    This archive was generated by hypermail 2b29 : Tue Sep 10 2002 - 14:38:52 MET DST