Date: Tue, 2 Dec 1997 10:18:27 -0500 (EST) Message-Id: <199712021518.KAA19284@locke.ccil.org> Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: Re: irony markers X-To: sbelknap@UIC.EDU X-cc: lojban@cuvmb.cc.columbia.edu To: John Cowan Status: OR X-Mozilla-Status: 0011 Content-Length: 1394 X-From-Space-Date: Tue Dec 2 10:19:02 1997 X-From-Space-Address: LOJBAN@CUVMB.CC.COLUMBIA.EDU >lojbab: >>And the computer will fail to detect them. Lojban is designed assuming >>that at >>some not too distant time, someone will have a computer program that will >>try to understand spoken Lojban. Now whether it will ever be possible to get >>a computer to understand content is a real issue. But getting a computer to >>recognize unmarked irony is one step beyond understanding content that is not >>inhgerently deceptive. Isomorphiosm in the language is supposed to make the >>process easier. > >You are assuming that the computer will interact with human speakers of >lojban through ascii text. I am assuming that the computers will speak and >listen in the same way that people do; thus the computer will have access >not only to the text, but the metalinguistic context as well. No. I am assuming that most "unmarked irony" is indeed actually marked - by tone of voice and body language etc. Subtle things that we are not entirely aware of necessarily. That is why you need to mark more smileys and other such things in ASCII text than you need to do in human speech. I suspect that getting computers to understand context will be one level of difficulty. Getting them to recognize the subtleties of metalinguistics and especially those that arenot well conveyed through words and context alike (as for the ones that need marking in net.text) will be MUCH harder. lojbab