LINGUIST List 2.15

Wednesday, 23 Jan 1991

Disc: IPA

Editor for this issue: <>

Directory

1. John Bro, Ascii IPA: Klattese
2. Derk Ederveen, Re: IPA in ASCII
3. , Re: IPA in ASCII
4. Osamu Fujimura, Re: Answers and Queries
5. Paul Saka, IPA on ASCII
6. , IPA in Wordperfect

======================================================================
In response to Allan Wechsler's request, here are 2 postings and 3 
versions of an Ascii IPA which I picked up on Usenet Sci.Lang a few 
months ago. Hope they're of use.
 John Bro bougiepine.circa.ufl.edu
=======================================================================
From: zacharskcs.umn.edu (Ronald Zacharski)
Subject: Re: ASCII Version of IPA??? Date: 10 Aug 90 18:46:47 GMT

Klatt, Dennis H. Review of text-to-speech conversion for English. 
 J.Acoust.Soc.Am 82(3):737-793.

describes 2 transcription systems. One is case insensitive and requires
one or two characters per symbol. Klatt states this system is
"nearly identical to ARPAbet" The other system is a case sensitive one
character system.

Here's the chart from that article.
 system
2 ch 1 ch example 2 ch 1 ch example
---------------------- --------------------
 IY i beet CH C Chet 
 IH I bit JH J jet 
 EY e bait M m met 
 EH E bet N n net 
 AE  bat NX G sang 
 AA a pot F f fed 
 AO c bought V v vet 
 AH ^ but TH t thin 
 OW o boat DH D this 
 UH U book S s set 
 UW u boot Z z zero 
 RR r Bert SH S shed 
 AY A bite ZH Z azure 
 OY O boy W w wet 
 AW W bout YX y yet 
 YU Y Butte R r red 
 AX x about L l let 
 IX | nieces HX h head 
 P p pet EN N button 
 B b bet EL L bottle 
 T t tet _ _ silence "phoneme" 
 D d debt
 K k kit
 G g get

Ron Zacharski
University of Minnesota
================================================================================
From: zwickypterodactyl.itstd.sri.com
Subject: IPA and ASCII
Organization: SRI Intl, Menlo Park, CA 94025

Actual IPA-ASCII mapping standards don't exist, partly because ASCII
doesn't have enough bits in it, unless you use digraphs for what
are single characters in IPA. However, there is an existing standard
for the ASCII transliteration of *phonemic* transcription of
*English*, using one character per phoneme. 

a = the V in cot I = the V in kit 
c = the V in caught o = the V in coat
 = the V in cat u = the V in coot
e = the V in Kate U = the V in put 
E = the V in pet ^ = the V in cut 
i = the V in Pete x = schwa 

b = the 1st Cons. in bat N = the Last Cons. in king 
d = the 1st Cons. in date p = the 1st Cons. in pat 
C = the 1st Cons. in chat r = the 1st Cons. in rat 
D = the 1st Cons. in that s = the 1st Cons. in sat 
f = the 1st Cons. in fat S = the 1st Cons. in shape 
g = the 1st Cons. in gate t = the 1st Cons. in tat 
h = the 1st Cons. in hat T = the 1st Cons. in thick 
j = the 1st Cons. in jilt v = the 1st Cons. in vat 
k = the 1st Cons. in cat w = the 1st Cons. in wet 
l = the 1st Cons. in late y = the 1st Cons. in yak 
m = the 1st Cons. in mat z = the 1st Cons. in Zach 
n = the 1st Cons. in Nate Z = the med. Cons. in leisure 

DUs, DIs Iz taipd in kltiz, alDo nat t Enihwer nir Dx spid Dt ai
yUzd tu bi aibxl tu aten, nd kwait pasIbli wiT Ercrz In Dx
trnskripSxn (xbut hwIC ai du nat wiS tu bi Infcrmd). nd yEs, ai du
rili tck laik DIs, haipxrkcrEkt Do It mei bi.

I'm not a phonetician, just working with the phonetic alphabet.

> 1. Is there an accepted convention for typing phonetics on an ASCII
> keyboard? All the strategies I can think of require multigraphs. One
> could choose to resolve ambiguities in two ways. (i) You could use some
> delimiter to separate graphs only in ambiguous cases. Example [n'g] in
> "ungrateful" vs [ng] in "singer". (ii) You could *always* delimit
> segments, as in [s.i&.ng.r].
Not that I know.

> 2. Is there an accepted assignment of ASCII codes to IPA characters?
> That is, if I take the trouble to create an IPA font, to what codes
> should I assign "eng" and "esh"?
Not that I know.

However, there are several Computer-readable Phonetic Alphabets (CPA).
The one I work with is COST-CPA, described in proposal 209TD(87)138 by
Peter Molbaek Hansen, in: report EUR 12023 EN, 1989: Man-Machine
Communication bu Means of Speech Signals, European Research Project COS 209.

In this alphabet, most IPA-characters are assigned to ASCII-characters,
enterable on most keyboards en readable on most terminal screens.
If interested, I can send you more information by mail.

groeten / salutojn,
Derk Ederveen
-------------
Kath. Universiteit, Nijmegen / PTT Research NT-TWS, Leidschendam - - NL
ederveenlett.kun.nl D_Ederveenpttrnl.nl tel. +31/0 70 3323202
kunlt1::ederveen dnlts::ederveen ederveenhlsdnl5.bitnet
** esperanto(Lingvo) :- neuxtrala(Lingvo), internacia(Lingvo), dua(Lingvo). **

In answer to Allan Wechsler's question, the first 1990 issue
of the Journal of the International Phonetic Association
has an article on the proposed ASCII coding for IPA symbols--
issue 20.1 pp22-26. Since I just found this reference today,
I don't know what it is good for yet, but there *is* now a
set.
Geoff Nathan<ga3662siucvmb.bitnet>

As for IPA in ascii, Richard Sproat might have some useful information.
He is at AT&T Bell Labs, and his address is rwsresearch.att.com.
Osamu Fujimura

Altho I do not have a general solution for IPA characters, I would like
to propose the following convention for represennig common English
phonemes:
 /&/ central vowel
 // low front
 /$/ palatal fricative
These symbols are mnemonic: the ampersand is a stylized or modified "e",
just as the inverted "e" is a modified "e"; the at-sign is a modified
"a", just as the ash-symbol is a modified "a"; and the dollar-sign is a
modified "s", just as the esh-sign is a modified s. And for keyboards
that have the cent-sign, modified "c" can stand for the palatal affricate.

Does anyone know of a source (hopefully cheap) for IPA symbols that can be
called up by Wordperfect 5.0 and can be used with HP laser printers? If so,
please let me know as soon as possible, as I need to produce a camera-ready
text over the next few months. Thanks. Alan Slotkin, English Department, Box
5053, Tennessee Technological University, Cookeville, TN 38505.