perm filename PHONE[1,ALS] blob
sn#058557 filedate 1973-08-16 generic text, type T, neo UTF8
∂16-AUG-73 0050 network site BBN
-------
Date: 16-AUG-73 0352-EDT
From: CMU at BBN-TENEX
Re: PHONETIC CODE PROPOSAL FOR DATA BASE MEETING OF 20-AUG.
cc: UMICH, A610EH51 at CMU-10B
- - - -
ARPANET REPRESENTATION OF DETAILED PHONETIC CODING
15-Aug-73
T.T. BURNETT
E.H. HAYDEN
J.E. SHOUP-HUMMEL
AS PER THE REQUEST OF THE DATA BASE COMMITTEE THE FOL-
LOWING PHONETIC CODING IS SUGGESTED. IT REPRESENTS A CONSIDER-
ABLE MODIFICATION TO THE EARLIER PROPOSAL GIVEN AT THE DATA BASE MEET-
ING IN SANTA BARBARA. IN GENERAL, IT NOW IS INTEGRATED WITH THE
ARPABET AND THE FEATURES REQUESTED BY C-MU.
PLEASE READ THIS BEFORE THE NEXT MEETING OF THE DATA BASE
COMMITTEE AT PALO ALTO ON MONDAY, AUGUST 20, 1973.
(The meeting is at the STANFORD AI Project at 8:30 P.M.)
THE FOLLOWING GENERAL PRINCIPLES HAVE BEEN APPLIED TO THE
DEVELOPMENT OF THE SCHEME.
1. ALL PRIMARY PHONETIC SYMBOLS INCLUDE AN ALPHABETIC
CHARACTER.
2. WHEN THE ARPABET SYMBOL REPRESENTS A PRIMARY PHONETIC
SYMBOL OF THE THEORY THE ARPABET SYMBOL IS CHOSEN.
3. IF AN ARPABET SYMBOL IS NOT AVAILABLE, A TWO CHAR-
ACTER CODE CONSISTING OF A NON-ALPHABETIC SHIFT CHARAC-
TER FOLLOWED BY AN ALPHABETIC CHARACTER IS CHOSEN -
PREFERABLY ONE THAT IS PHONETICALLY OR
VISUALLY CLOSE TO THE NEEDED SYMBOL.
4. ALL FEATURES/PARAMETERS ARE REPRESENTED BY NON-ALPHA-
BETIC CHARACTER PAIRS TO ASSURE THAT THEY THEY ARE
MAXIMALLY DISTINCT FROM THE PRIMARY PHONETIC SYMBOLS
VISUALLY AND PROGRAMMATICALLY.
AN UNDERSTANDING OF THE FOLLOWING SPECIFIC PRINCIPLES WILL
ASSIST THE INTERPRETATION OF THE TABLES THAT FOLLOW. IT WILL BE
HELPFUL TO HAVE PAGES 45 AND 48 OF 'A PHYSIOLOGICAL THEORY OF
PHONETICS' (PETERSON AND SHOUP, JSHR, MARCH 1966) AVAILABLE AS YOU
STUDY THESE TABLES.
REGARDING THE PRIMARY PHONETIC SYMBOL MATRIX THESE RULES
APPLY.
1. @, &, ; AND % ARE SHIFT CHARACTERS.
2. THESE SHIFT CHARACTERS WHEN THEY OCCUR DENOTE THE FOLLOWING:
A. @ AND & BOTH DENOTE VOICED CONSONANTS OR ROUND-
ED VOWELS.
B. ; AND % BOTH DENOTE UNVOICED CONSONANTS OR UNROUNDED
VOWELS.
C. % IS USED ALMOST EXCLUSIVELY WITH VOWELS, EXCEPT
IN THE CASE OF USE BEFORE X, WHERE IT DENOTES
UNVOICING.
3. IF ; OR @ OCCUR AS SHIFT CHARACTERS BEFORE SOME ALIPHA-
BETIC CHARACTER THAT ALSO USES % OR & AS A SHIFT CHAR-
ACTER, THEN THE USE OF ; OR @ DENOTES THE FRONTMOST
AND/OR HIGHEST POSITION FOR THE AFFECTED PHONEME, WHETHER
VOWEL OR CONSONANT.
4. EXCEPT FOR ARPABET SYMBOLS INTEGRATED IN THE CODING, AL-
PHABETIC VOWEL CHARACTERS ARE USED TO ENCODE ONLY VOWEL
PHONEMES AND ALPHABETIC CONSONANTAL CHARACTERS TO ENCODE
CONSONANT PHONEMES. THE CHARACTER 'Y' IS USED IN THE
COMBINATION '@Y' TO INDICATE A ROUNDED VOWEL, /IY/.
5. NOTE THAT THREE NEW SYMBOLS HAVE BEEN ADDED TO THE PRI-
MARY PHONETIC SYMBOL MATRIX. THEY ARE:
;W VOICELESS W
;L VOICELESS L
;R VOICELESS R
REGARDING THE PHONETIC PARAMETERS THE FOLLOWING RULES APPLY:
1. THE SHIFT CHARACTERS ARE THE SAME AS ABOVE: @, &, ; AND %
2. WHEN POSSIBLE THE PRINCIPLES REGARDING THE SHIFT CHAR-
ACTERS ARE MAINTAINED FOR THE PRIMARY SYMBOLS AND THE
PARAMETERS.
3. FOR PARAMETERS THAT ARE CONCEPTUALLY RELATED AN ATTEMPT
HAS BEEN MADE TO USE THE SAME SECOND CHARACTER. THUS,
? REFERS TO LARYNGEAL ACTION, < TO AIR FLOW DIRECTION
AND ' TO PALATALIZATION.
4. DUE TO A LACK OF AVAILABLE SYMBOLS, ONLY CLASSES OF
VERTICAL PLACEMENT OF THE VOWELS AND CLASSES OF SOME HOR-
IZONTAL PLACES OF ARTICULATION CAN BE INDICATED. TO OBTAIN
OTHER POSITIONS SUPPLEMENTARY NOTATION MAY BE USED AS PER
THE PHONETIC THEORY.
THE MATRIX AND TABLES FOLLOW.
PRIMARY PHONETIC SYMBOL MATRIX
BI UN LD AL P1 P2 P3 P4 P5 PV V1 V2 UV UV UV
NA M @M N @N NX &N
------------------------------------------------------------------
ST P T ;J K ;K CL
B D @J G @G
--------------------------------------------------------------
FL @P DX @K
--------------------------------------------------------------
TR @B ;R &R
------------------------------------------------------------------
SI S SH ;S VC
Z ZH @Z
--------------------------------------------------------------
FR ;P F TH ;C ;X %X
&B V DH @C @X &X
WH
------------------------------------------------------------------
SO ;W ;L ;R CS
W L R Y
@W @V @R @L
------------------------------------------------------------------
IY ;I %O H3
@Y @U UW
----------------------------------------------
IH IX H2
@I
------------------------------------------
;E UH H1
@E
--------------------------------------
EH %E AX M2
&E &I &U
----------------------------------
AE ;A M1
------------------------------
%A AH ;U L3
@O &O
--------------------------
A %U L2
AO
----------------------
AA L1
&A
------------------
;H ;Q PH
@Q &G
---------------
HH Q GL
&H &Q
DIPHTHONGS, AFFRICATES, SYLLABICS, ETC.
AW DIPHTHONG
AY DIPHTHONG
OY DIPHTHONG
CH PALATAL-2 AFFRICATE, VOICELESS
ZH PALATAL-2 AFFRICATE, VOICED
EL SYLLABIC L
EM SYLLABIC M
EN SYLLABIC N
ER RETROFLEXED CENTRAL VOWEL
SECONDARY PHONETIC PARAMETERS
AIR MECHANISM LARYNGEAL ACTION
%! PULMONIC ;% VOICELESS
%? GLOTTIC ;! WHISPERED
%" VELIC @* BREATHY
%] GLOTTOPULMONIC @& VOICED
%[ VELOPULMONIC @# LARYNGEALIZED
;? VELOGLOTTIC @$ PULSATED
%. COMBINED @. CONSTRICTED
@! PHONOCONSTRICTED
@? STOPPED
AIR DIRECTION PHARYNX SHAPE
%< EGRESSIVE %& OPEN
%> INGRESSIVE %, PHARYNGEALIZED
AIR FLOW TONGUE SHAPE
%0 ZERO ;& NATURAL
%- NONFRICTIONAL ;' PALATALIZED
%* INTERMEDIATE ;" VELARIZED
%+ FRICTIONAL
APEX SHAPE
AIR PRESSURE ;) SIMPLE
;/ RETROFLEXED
&- LENIS
&* NORMAL LIP SHAPE
&+ FORTIS
@) SPREAD
%; UNROUNDED
AIR RELEASE &@ ROUNDED
;@ CONTINUED
;* PHONOASPIRATED SUPPLEMENTARY NOTATION
;+ ASPIRATED
;- UNASPIRATED @< ADVANCED
;$ UNEXPLODED @> RETRACTED
@[ ELEVATED
@] DEPRESSED
GEN. AIR PATH @( DENTALIZED
;0 ORAL ;: LONG
;# NASALIZED ;. LESS LONG
;, NASORAL
;! NON-ORAL # SPEAKER I. D.
+ MORPHEME BOUNDARY
, NON-TERMINAL JUNCTURE
LINGUAL AIR PATH - SILENCE
. DECLAR. TERMINATION
@- NON-LATERAL / WORD BOUNDARY
@* UNILATERAL ? INTERR. TERMINATION
@+ BILATERAL
&0 NON-SPEECH
;0 NULL PHONEME
&/ LAUGHTER
@/ COUGHING
PROSODEMIC PARAMETERS
1 STRESS, PRIMARY %1 PITCH LEVEL 1
2 " SECONDARY %2 " " 2
3 " TERTIARY %3 " " 3
4 " ETC. %4 " " 4
5 " " %5 " " 5
6 " " %6 " " 6
7 " " %7 " " 7
8 " " %8 " " 8
9 " " %9 " " 9
: TONE GROUP BOUNDARY
PRIMARY PHONETIC PARAMETERS
HORIZ. PLACE OF ARTIC. VERT. PLACE OF ARTIC.
&) BILABIAL @6 CLOSED
%) LABIODENTAL @5 VERY CLOSE
%/ LINGUALABIAL @2 HIGH-2
%% INTERDENTAL &1 MID-1
%( DENTAL &2 LOW-2
&# ALVEOLAR &! PHARANGEAL
&' PALATAL-3 &? GLOTTAL
&, PALATOVELAR
&" VELAR-1
&$ UVULAR
MANNER OF ARTICULATION
&# NASAL
&< PLOSIVE
&[ EJECTIVE
&] CLICK
&> IMPLOSIVE
;1 FLAP
;2 TRILL-2
;3 TRILL-3
;4 TRILL-4
;; TRILL-M
&; SIBILANT
&% FRICATIVE
&@ SONORANT
@@ VOWEL
CMU "PHONETIC MODIFIERS" PROPOSAL
15-AUG-73
LINDA SHOCKEY
LEE ERMAN
THIS IS AN EXPLANTION OF THE MODIFIERS USED AT CMU TO 1) PROVIDE A MEANS
FOR DOING A REASONABLY NARROW TRANSCRIPTION OF ENGLISH SOUNDS
AND 2)TO APPROXIMATE IPA NOTATION, USING THE ARPABET CHARACTERS AS A
BASE.
THIS SCHEME PROVIDES AN ALTERNATE NOTATION TO THAT SUGGESTED
BY SCRL, BUT IS NOT (CURRENTLY) AS COMPREHENSIVE AS THEIRS.
WHEN ONE TRIES TO DO MORE THAN A VERY BROAD TRANSCRIPTION OF
ENGLISH OR ATTEMPTS TO TRANSCRIBE NON-ENGLISH LANGUAGES IN THE APA
PHONETIC SYMBOLS CURRENTLY IN USE, ONE FINDS THAT THESE SYMBOLS ARE
INADEQUATE TO EXPRESS MANY SIGNIFICANT DIFFERENCES BETWEEN
SPEECH SOUNDS. IN ORDER TO MAKE THE CURRENT SETUP MORE FLEXIBLE A SET
OF MODIFIERS WAS CONSTRUCTED WHICH CAN BE APPENDED TO THE ARPA
PHONETIC CHARACTERS IN ORDER TO INDICATE VARIATIONS IN PRONUNCIATION.
THE MODIFIERS WERE CHOSEN FROM THOSE LISTED IN THE PETERSON-SHOUP-
ASCII PHONETIC ENCODING SYSTEM ON THE BASIS OF REPRESENTING FREQUENTLY-
OBSERVED PHENOMENA. THEY ARE AS FOLLOWS:
DIACRITIC
TWO CHARACTER ONE CHARACTER (LC=LOWER CASE)
RETROFLEXED RF LC R
VOICELESS O LC O
PALATIZED P LC P
UNEXPLODED CH C
EXPLODED
LARYNGEALIZED X LC X
BREATHY BH B
VELARIZED L LC L
ADVANCED AD A
(FRONTED FOR VOWELS)
RETRACTED RD R
(BACKED FOR VOWELS)
ELEVATED ED E
(RAISED FOR VOWELS)
DEPRESSED DD D
(LOWERED FOR VOWELS)
FRICTIONAL F LC F
NASALIZED N LC N
ROUNDED W LC W
VOICED V LC V
DENTALIZED T LC T
CENTRALIZED CD LC C
SEGUE ← ←
ASPIRATED H LC H
UNROUNDED M LC M
TRILLED RL L
LONG : :
DYNAMIC
(TRANSITION)
UNREALIZED
RELEASED + +
DYNAMIC DY Y
THESE MODIFIERS ARE TO APPEAR BETWEEN "UPARROWS" (↑) AFTER THE
SOUND THEY ARE TO MODIFY. FOR EXAMPLE, A RETROFLEX "T" WILL
BE REPRESENTED AS T↑RF↑, AN UNROUNDED UH AS UH↑M↑, ETC.
(THE SEGUE MODIFIER (←) IS USED WHEN ONE WISHES TO INDICATE THAT
ONE CANNOT CLEARLY DISTINGUISH THE BOUNDARY BETWEEN TWO SOUNDS.
THE SEGUE MARKER IS PLACED AFTER THE FIRST OF TWO SOUNDS FOR
WHICH THE BOUNDARIES ARE INDETERMINATE.)
USING THESE MODIFIERS, ONE CAN RECORD A LARGE NUMBER OF THE
FREQUENT PHONETIC MODIFICATIONS OF ENGLISH SOUNDS FOUND
IN NATURAL SPEECH. THERE IS STILL A PROBLEM WITH TRANSCRIBING A
NON-ENGLISH LANGUAGE, THERE BEING NO SYMBOLS FOR MANY OF THE
SOUNDS ENCOUNTERED, SUCH AS THE FINAL SOUND IN THE GERMAN
WORD "BACH" OR THE INTERVOCALIC SOUND IN SPANISH "TORRO." ONE MIGHT LIKE TO
BE ABLE TO CONSTRUCT UNAMBIGUOUSLY PARSABLE ONE- AND TWO-LETTER
CODES FOR AT LEAST EVERY SYMBOL IN AN IPA CHART. IN THE PRESENT
IMPLEMENTATION, HOWEVER, THE MODIFIER SYSTEM DOES DOUBLE DUTY
BY PROVIDING A WAY TO APPROXIMATELY DESCRIBE THE PRONUNCIATION OF A
GIVEN NON-ENGLISH SOUND BY REFERING TO AN ENGLISH
SOUND. FOR EXAMPLE, THE FINAL SOUND IN "BACH" IS LIKE A K, BUT
FRICATIVE; ITS DESCRIPTION IS THEREFORE K↑F↑. ALTHOUGH SOME OF THE
SYMBOL-PLUS-MODIFIER CLUSTERS THUS DERIVED ARE INCONVENIENT TO USE, THEY
ARE REASONABLY MNEMONIC.
THIS SYSTEM IS OF COURSE MERELY A NOTATIONAL
VARIANT OF A SUBSET OF THE SYSTEM PROPOSED BY SCRL. ITS MAIN DIFFERENCE IN
PHILOSOPHY IS THAT THE PRIMARY PHONETIC SYMBOLS ARE THE
ARPABET CHARACTERS RATHER THAN SOME LARGER PHONETIC ALPHABET.
IN ADDITION, THE MODIFIERS ARE DELINEATED FROM THE PRIMARY
PHONETIC SYMBOLS BY CONTEXT (I.E. THEY APPEAR WITHING "↑"S);
THUS, THERE ARE TWO SEPARATE NOTATIONS, EACH OF WHICH CAN
MAKE USE OF THE SAME ALPHABETIC SYMBOLS IN ORDER TO MAKE
THE NOTATIONS MORE "NATURAL".
WE HAVE FOUND THAT IT IS BOTH EASY TO USE AND EASY TO TEACH
TO PHONETICALLY UNSOPHISTICATED STUDENT WORKERS.
IT ALSO OFFERS POSSIBILITIES FOR BEING USED IN PUBLICATIONS, SINCE
THE READER IS NOT SEVERELY TAXED IN TRANSLATING FROM CMU
MODIFIERS TO STANDARD PHONETIC TERMINOLOGY.
FINALLY, IT IS POSSIBLE TO ADD TO THE SET OF MODIFIERS
AS ADDITIONS ARE DEEMED DESIRABLE; BECAUSE THERE IS NO
CONCERN ABOUT OVERLAP WITH OTHER KINDS OF DESCRIPTORS,
THERE IS STILL A FAIRLY WIDE RANGE OF (HOPEFULLY MNEUMONIC)
SYMBOLS TO CHOOSE FROM.
-------