Re: language abuse and machine translation

From: Party of Citizens (citizens@vcn.bc.ca)
Date: Sat Mar 08 2003 - 12:38:20 MST

  • Next message: Alfio Puglisi: "Re: language abuse and machine translation"

    On Sat, 8 Mar 2003, Alfio Puglisi wrote:

    > On Fri, 7 Mar 2003, spike66 wrote:
    >
    > >In this forum I have decried the practice of
    > >nonstandard language use for the reason that
    > >soon now machines will be advanced enough to
    > >do realtime translations using speech recognition
    > >and a simple sort of table-lookup style
    > >substitution. The overuse of obscure allusions
    > >and verbing, among other language bad habits,
    > >will delay practical machine translation, as
    > >well as interfere with machines' ability to
    > >understand humans.
    > >
    > >[...]
    > >
    > >We need newspeak. Before it's too late.

    How about robospeak instead of newspeak (aka moronspeak when expressed
    according to Whitehouse dialect)? And yes, we need robospeak before it is
    too late and the Queen's English drifts into irreversible moronspeak.

    Sample moronspeak:

    "All chickens are for us or against us. Some chickens cross the road.
    There are good doer chickens and bad doer chickens. Therefore the chicken
    crossed the road."

    Now this argument has three premisses and a conclusion. Logicians will
    immediately recognize it as translatable into the symbolic language of
    predicate calculus. But try translating that into French (and back to
    English again) and you will understand why Powell is having problems at the
    UN these days.

    > I'm not going to dumb down my language for the programmers' inability to
    > understand it (I'm a programmer). Screw the stupid machines that we
    > have now, and invent new, richer, different forms of speech.

    You could use the YES/NO/AND/OR connectors of combinatorial logic which
    mean exactly the same in Adult Normative Standard English (ANSE), as the
    only connectors for linking descriptors. A descriptor would be any symbol,
    word, phrase, clause, sentence etc. which can be UNDERSTOOD by
    ANSE-speakers even if it is not conventionally used in ANSE as long as it
    can be assigned a T or F truth value. The criterion for the phrasing of
    decriptors is purely semantic, ie it must communicate (to that ANSE target
    population). Its lexicon consists of the descriptors as above. Its syntax
    is the syntax of combinatorial logic, to which all other forms of
    arithmetic, logic and mathematics can be reduced. Can it be developed to
    the stage at which it meets the semantic criterion of accounting for
    everything we MEAN to say when we use ANSE?

    POC

     Mix them,
    > like it's already happening between English and local languages, and use
    > them freely.
    >
    > Computers, for now, are merely a tool. A tool's job is to help his owner,
    > not to restrict his freedom. A program must meet some requirements before
    > granted the "human-level intelligence" medal. Excellent speech recognition
    > and understanding are a given. And I would expect it to be much better
    > than us at moving between languages and inventing new ones.
    >
    > Ciao,
    > Alfio
    >
    >



    This archive was generated by hypermail 2.1.5 : Sat Mar 08 2003 - 12:43:45 MST