Xref: utzoo comp.sys.next:1145 comp.sys.mac:24715 comp.cog-eng:817
Path: utzoo!utgpu!watmath!clyde!att!osu-cis!killer!ames!ucsd!nosc!humu!uhccux!lee
From: lee@uhccux.uhcc.hawaii.edu (Greg Lee)
Newsgroups: comp.sys.next,comp.sys.mac,comp.cog-eng
Subject: Re: Sorry, no PDP (or any other) speech recognition yet
Message-ID: <2934@uhccux.uhcc.hawaii.edu>
Date: 5 Jan 89 11:16:56 GMT
References: <18726@agate.BERKELEY.EDU>
Organization: University of Hawaii
Lines: 22

From article <18726@agate.BERKELEY.EDU>, by csm@violet.berkeley.edu:
" In article <679@cogsci.ucsd.EDU> norman@cogsci.UUCP (Donald A Norman-UCSD Cog Sci Dept) writes:
" >... "new display" and "nudist play" ...
" 
" In southern CA people often ask:
" 	"Djagotoessee?"

Similarly, "Judas get up?" = Did you just get up?.

It may be that the spectra of "new display" and "nudist play" can appear
essentially identical, but for many, the "d" part of "nudist" is
articulated differently (very short) and the diphthong of "new" is
appreciably longer than that of "nu".

These kinds of phonetic details are at least partially law-like and
predictable.  A substantial amount of work has been done to understand
them in the last century (but much earlier in India).  It's
disappointing that some (most?) contemporary work in speech recognition
seems innocent of this body of scholarship.  I guess the idea is that
computation can substitute for understanding.  Maybe.  I have my doubts.

		Greg, lee@uhccux.uhcc.hawaii.edu