Received: with ECARTIS (v1.0.0; list lojban-beginners); Tue, 06 Sep 2005 06:56:13 -0700 (PDT) Received: from nobody by chain.digitalkingdom.org with local (Exim 4.52) id 1ECdvp-0008Io-9x for lojban-beginners-real@lojban.org; Tue, 06 Sep 2005 06:56:13 -0700 Received: from wproxy.gmail.com ([64.233.184.207]) by chain.digitalkingdom.org with esmtp (Exim 4.52) id 1ECdvm-0008Ih-Vy for lojban-beginners@chain.digitalkingdom.org; Tue, 06 Sep 2005 06:56:13 -0700 Received: by wproxy.gmail.com with SMTP id 71so660808wri for ; Tue, 06 Sep 2005 06:56:07 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:references; b=RW/OwZ0BC471cpcOYvUvwmBIymJ8qrIkOILbcBdnNsLFFX/fIgXaCeTVf2tQtllqrdwBpx75fYLy3RjqScjRIDEtPtYthYaHCnrPTMId/4pSoSHAGuOS5gUjt2RjUPmA9kSWuuFLnOdcvUohgcWQTT/wK48FkmiqZ5xDDMaquLc= Received: by 10.54.36.9 with SMTP id j9mr4338363wrj; Tue, 06 Sep 2005 06:56:07 -0700 (PDT) Received: by 10.54.140.6 with HTTP; Tue, 6 Sep 2005 06:56:07 -0700 (PDT) Message-ID: Date: Tue, 6 Sep 2005 09:56:07 -0400 From: Matt Arnold To: lojban-beginners@chain.digitalkingdom.org Subject: [lojban-beginners] Re: Lojban is poor for machine translation In-Reply-To: Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_12990_3426824.1126014967428" References: X-Spam-Score: -2.4 (--) X-archive-position: 1983 X-Approved-By: matt.mattarn@gmail.com X-ecartis-version: Ecartis v1.0.0 Sender: lojban-beginners-bounce@chain.digitalkingdom.org Errors-to: lojban-beginners-bounce@chain.digitalkingdom.org X-original-sender: matt.mattarn@gmail.com Precedence: bulk Reply-to: lojban-beginners@chain.digitalkingdom.org X-list: lojban-beginners Content-Length: 4606 ------=_Part_12990_3426824.1126014967428 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline He points out the problem of a computer knowing what the speaker "really= =20 means" but he seems to want complete automation of the process of inference= .=20 This is a serious problem no matter what languages are involved. What I see= =20 as the usefulness of Lojban is in the ability for a speaker to tell the=20 translating computer what they really mean so that the computer can say tha= t=20 in the target language. I have sometimes imagined two people glancing at a display (which is=20 handheld or clipped to their collars) while speaking different languages to= =20 each other. Their conversation is shown on the display, but in a series of= =20 multiple-choice selections between different possible meanings of what they= =20 have said, and different attitudes. This way, the computer can select which= =20 idiom in the target language best expresses the intended meaning and=20 attitude in which it is expressed. It can't do this until it knows the=20 intended meaning and attitude, and until mind-reading technology improves,= =20 it can't know that until the speaker tells it. So it needs a=20 human-comprehensible system in which to represent meaning. Lojban seems to= =20 be good at that. -epkat On 9/6/05, Naomi K wrote: >=20 > "Loglans are poor MT interlinguas" -=20 > http://www.rickharrison.com/language/mtil.html >=20 > Most of you would probably know of this article, which argues that the=20 > syntax of Lojban is overly complex; sumti place structures are too rigid = and=20 > counter-intuitive; that an artificial language would be more difficult to= =20 > translate into natural languages as opposed to natural language to natura= l=20 > language translation; that required pauses are unnatural and impractical;= =20 > and that the compound words are difficult for ML to translate.=20 >=20 > What is your view of this argument? >=20 > mi'e.nei,omis. > ------=_Part_12990_3426824.1126014967428 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline    He points out the problem of a computer knowing what the speaker "really means" but he seems to want complete automation o= f the process of inference. This is a serious problem no matter what languages are involved. What I see as the usefulness of Lojban is in the ability for a speaker to tell the translating computer what they really mean so that the computer can say that in the target language.

   I have sometimes imagined two people glancing at a display (which is handheld or clipped to their collars) while speaking different languages to each other. Their conversation is shown on the display, but in a series of multiple-choice selections between different possible meanings of what they have said, and different attitudes. This way, the computer can select which idiom in the target language best expresses the intended meaning and attitude in which it is expressed. It can't do this until it knows the intended meaning and attitude, and until mind-reading technology improves, it can't know that until the speaker tells it. So it needs a human-comprehensible system in which to represent meaning. Lojban seems to be good at that.

-epkat


On 9/6/05, Naomi K <ali= en.juxtaposition@gmail.com> wrote:
"Loglans are poor MT interlinguas" - http://www.rickharrison.com/language/mtil.ht= ml

Most of you would probably know of this article, which argues that the syntax of Lojban is overly complex; sumti place structures are too rigid and counter-intuitive; that an artificial language would be more difficult to translate into natural languages as opposed to natural language to natural language translation; that required pauses are unnatural and impractical; and that the compound words are difficult for ML to translate.

What is your view of this argument?

mi'e.nei,omis.

------=_Part_12990_3426824.1126014967428--