Xref: utzoo comp.ai:1141 sci.lang:1667 Path: utzoo!mnetor!uunet!husc6!hao!oddjob!gargoyle!ihnp4!homxb!houdi!marty1 From: marty1@houdi.UUCP (M.BRILLIANT) Newsgroups: comp.ai,sci.lang Subject: Re: Language Learning (anecdotes) Message-ID: <1433@houdi.UUCP> Date: 8 Dec 87 15:09:15 GMT References: <1966@uwmacc.UUCP> <12400009@iuvax> <1117@uhccux.UUCP> <2048@uwmacc.UUCP> Organization: AT&T Bell Laboratories, Holmdel Lines: 32 Summary: Word of warning about subject of argument In article <2048@uwmacc.UUCP>, edwards@uwmacc.UUCP (mark edwards) writes, in response to article <2999@bcsaic.UUCP> by rwojcik@bcsaic.UUCP (Rick Wojcik): > > ... It seems to me what you are saying is that the observed > evidence shows that no adult can lose their native accent. Which is > like saying "That man listens only to jazz because that is all I hear > him listening to". > > Just because there is no observed evidence does not prove your theory. > I say because any child can learn his native language any adult can > learn the childs native language also. My proof is that any adult can > do what ever any child can do. > I don't think you can learn much about natural processes by restricting the argument to terms of "can do" vs. "can't do." All good studies of natural systems are done statistically, with large subject pools and systematic controls, because all individuals are different. When you study the capabilities of psychological systems, important clues can come from subtle differences in learning time or response time. Since differences may be reversed in some individuals, because of their particular capabilities, large subject groups must be used. It is probably true that anything most children do easily, some adult can do with effort. But if you want to know how natural learning works, to figure out how to make artificial systems learn, that doesn't tell you much. M. B. Brilliant Marty AT&T-BL HO 3D-520 (201)-949-1858 Holmdel, NJ 07733 ihnp4!houdi!marty1