From newshub.ccs.yorku.ca!torn!utcsri!rutgers!jvnc.net!darwin.sura.net!mips!sdd.hp.com!usc!elroy.jpl.nasa.gov!ames!sun-barr!news2me.ebay.sun.com!exodus.Eng.Sun.COM!orfeo.Eng.Sun.COM!silber Thu Jul  9 16:19:54 EDT 1992
Article 6376 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!torn!utcsri!rutgers!jvnc.net!darwin.sura.net!mips!sdd.hp.com!usc!elroy.jpl.nasa.gov!ames!sun-barr!news2me.ebay.sun.com!exodus.Eng.Sun.COM!orfeo.Eng.Sun.COM!silber
>From: silber@orfeo.Eng.Sun.COM (Eric Silber)
Newsgroups: comp.ai.philosophy
Subject: gestalts, units, phonemes, morphemes, frame-emes
Message-ID: <l4k60aINNocc@exodus.Eng.Sun.COM>
Date: 25 Jun 92 19:01:30 GMT
Organization: Sun Microsystems, Mt. View, Ca.
Lines: 14
NNTP-Posting-Host: orfeo


 In terms of the current state of psycho-physiological-research with respect to
 human language processing, what general 'recognition characteristics' are
 prominent?  i.e., in processing speech, a linear phenomenon ( a time series (sound train) )
 is interpreted.  What types of tokens are identified? is there a precedence in the
 tokenizing, does the sequence of transformations used in the processing depend upon
 both endogenous and exogenous variables?  Presumably there are various processing
 scenarios, e.g., attempt to tokenize phonemes into the recognized set for the
 assumed language; parse / seize upon 'gestalts' etc. 
 This maps the process into a sequential  operation, obviously , there are parallel
 components, and success in some of the operations allows for pruning others etc . (???)
 ANYWAY, a linear phenomenon (speech) is fed through this complex 'understander',
 does it (the human understander ) have some standard, gross priorities for
 segmenting the input ???


