Making sense
- From: Wolf Kirchmeir <wolfekir@xxxxxxxxxxxx>
- Date: Fri, 04 Nov 2005 10:50:15 -0500
"Making sense" is a problem I first encountered when marking students' essays. Two IMO relevant observations :
a) Most people know when they are not making sense at least some of the time. Students usually express this in words such as "I know what I want to say but I don't know how to say it."
b) We often say things that make sense to us but not to other people.
a) is an odd observation. How come we can tell that our language doesn't express what we want to express? In conceptualese/mentalese, a commonly held explanation is that the concepts expressed by the language don't match the concepts we have in mind. This little more than a rephrasing of the student's plaint, so it's not much of an explanation. But it does imply that concepts are independent of language in some way. Hence, mentalists posit pre-linguistic symbols (PLS). The behaviour of animals that lack language yet appear to be able to solve problems seems to support this notion. But there is a downside: we cannot know what these PLSs are, since the only evidence we have for them is a) language; and b) problem solving. Yet language does not necessarily match the PLSs, as noted above; and when we attribute problem solving to a chimp, say, we in fact describe its behaviour in language.
Thus, assuming PLS (in whatever form) does not get us anywhere. In particular, it doesn't help the AI enterprise, since we perforce must express in language any PLSs we desire to build into an AI system. If we could do that, we would presumably not get the mismatch between thought and expression that we all experience from time to time. Yet we need that, since it is a very powerful error-correction process: it corrects errors that are not knowable until they occur. (All existing error-correction processes I know of can correct only those types of errors of we already know about.)
An alternative explanation starts with noticing that we rehearse more or less consciously what we are about to write down: we talk to ourselves before we talk to others in such settings as essay writing. That is, when a student claims that he hasn't made the sense he wants, that could mean no more than that the internally produced monologue does not match the externally produced writing. The student rehearsed the expression internally ("Talked to himself"), then put the words down on paper, and discovered that there was a mismatch between written expression and the recalled internal monologue.
I propose that the mismatch occurs because the internally generated language is incomplete in the same sense that casual conversation is incomplete. That is, it consists of sentence fragments, sentences that change syntactic structure before completion, syntactically or semantically incorrect phrasing, and so on. (Listen to a tape of actual conversation to understand what I am referring to.) The written expression however conforms to the usage rules for formal English: it consists of complete sentences, in syntax and usage as correct as the student can make them. This completeness exposes illogic, inconsistent meanings, muddled semantics, and so on in a way that the conversational internal monologue does not. Hence the effect, "This doesn't say what I mean - it doesn't make sense."
b) raises several questions, which I won't deal with at this time. mostly because I realise my internal monologue isn't as clear as it should be. :-) However, what I have written above makes sense to me.
.
- Follow-Ups:
- Re: Making sense
- From: Randy
- Re: Making sense
- From: Randy
- Re: Making sense
- From: makc . the . great
- Re: Making sense
- From: Allan C Cybulskie
- Re: Making sense
- From: Just Playing
- Re: Making sense
- From: JGCASEY
- Re: Making sense
- Prev by Date: Re: Jeff Hawkins Q&A
- Next by Date: Re: Jeff Hawkins Q&A
- Previous by thread: Same Old, same Old
- Next by thread: Re: Making sense
- Index(es):
Relevant Pages
|
