Dept. for Speech, Music and Hearing Quarterly Progress and Status Report A quantitative model of vowel production and the distinctive features of Swedish vowels Lindblom, B. and Sundberg, J. journal: volume: number: year: pages: STL-QPSR 10 1 1969 014-032 http://www.speech.kth.se/qpsr 11. SPEECH PRODUCTION A. A QUANTITATIVE MODEL O F VOWEL PRODUCTION AND THE DISTINCTIVE FEATURES O F SWEDISH VOWELS * B. Lindblom and J. Sundberg e In t h i s p a p e r we s h a l l do f o u r things. (1) We s h a l l p r e s e n t a d e s c r i p t i o n of a model of the s p e e c h o r g a n s with the aid of which we c a n g e n e r a t e a n acoustic w a v e f o r m a p p r o p r i a t e t o vowel sounds. T h i s model is a s o r t of a r t i f i c i a l mouth which h a s a tongue, a jaw, a larynx, and lips. ( 2 ) We s h a l l r e p o r t s o m e r e s u l t s f r o m a study of the a r t i c u l a t o r y and acoustic possibilities of this model. In o t h e r w o r d s , we t r y t o l e a r n something about the relation between a r t i c u l a t i o n and acoustic r e s u l t s by having the model g o through a "babbling" stage. ( 3 ) We then investigate how the a r t i c u l a t o r y m e c h a n i s m s of t h i s model m u s t b e utilized i n o r d e r t o produce the Swedish vowels and t r y t o a n s w e r questions concerning the phonetic dimensions of the Swedish vowel s y s t e m f r o m the point of view of the model. (4) Finally, we d i s c u s s briefly the problem of how the phonetic dimens i o n s a r e related t o t h e a b s t r a c t phonological p r o p e r t i e s that we a t - I 1 I , t r i b u t e to Swedish vowel s e g m e n t s to account f o r t h e i r s y s t e m a t i c behavior i n the phonological p a t t e r n of the language. Table 11-A-1 includes l i s t s of phonetic and distinctive f e a t u r e categ- , o r i e s that have been suggested by v a r i o u s a u t h o r s f o r the d e s c r i p t i o n of vowels. T h e s e l i s t s provide a s m a l l s a m p l e of the total n u m b e r of sug- gestions t h a t a n extensive s e a r c h through the phonetic l i t e r a t u r e would reveal. O u r intention i s not t o d i s c u s s o r c o m p a r e t h e s e f r a m e w o r k s a t g r e a t length but m e r e l y t o point out that a u t h o r s d i s a g r e e both a s r e g a r d s the definition of c a t e g o r i e s and t h e i r number. B y way of i l l u s t r a t i o n l e t u s mention only the t r e a t m e n t of the Swedish [ u ] sound. c a u s e s Malmberg This vowel to r e j e c t the Jakobsonian b i n a r y flat-plain distinc - tion i n f a v o r of a f e a t u r e of labialization that is t e r n a r y . T o account f o r [a] F a n t ( 2 ) introduces a s h a r p - p l a i n dimension along which Swedish [i] and [Y] a r e found a t the s h a r p (palatalized) end and [u] a t the plain (un~ a l a t a l i z e d )end. -2 Chomsky and Halle (3) e x p r e s s the hope that the b i n a r y --- somewhat expanded v e r s i o n of p a p e r read a t a conference devoted to the d e s c r i p t i o n of Swedish, A p r i l 1969 i n Stockholm. I TABLE 11-A-1 SOME CATEGORIES O F VOWEL SPECIFICATION THEORY O F CARDINAL VOWELS FRONT CLOSE BACK HALF -CLOSE JAKOBSON, FANT, HALLE ROUNDED GRAVE-ACUTE COMPACT-DIFFUSE CENTRAL HALF-OPEN FLAT-PLAIN OPEN MALMBERG (1 956) FANT (1 967) GRAVE-ACUTE GRAVE-ACUTE COMPACT -DIFFUSE COMPACT -DIFFUSE F L A T I-FLAT 11,- PLAIN FLAT-PLAIN SHARP-PLAIN LADEFOGED (1 967) FRONT HIGH CENTER MID BACK LOW CHOMSKY, HALLE (1 968) ROUNDED BACK HIGH ROUND LOW CORONAL COVERED WANG (1968) HIGH PALATAL MID VELAR STL-QPSR l / l 9 6 9 an6 Ey, u, 17. b] exhibited c e r t a i n s i m i l a r i t i e s provided that the tongue cont o u r s w e r e c o m p a r e d with r e f e r e n c e to the mandible. Another g r 3 u p cons i s t e d of [ a , o , @]. The [ u ] vowel deviated f r o m t h e s e two groups and T f o r m e d f o r m e d a category by itself. T h e s e r e s u l t s a p p e a r f r o m Fig. 11-A-1, which shows one family of tongue c s n t o u r s f o r which the tongue is r a i s e d i n the d i r e c t i o n of the h a r d palate, the contour f o r [ u ] f o r which the tongue midline i s r a i s e d towards the soft palate and a t h i r d family which displays a lowered contour which h a s a l s o been displaced backwards. Fig. 11-A-2 d e m o n s t r a t e s that the "rounded" and "front" g r o u p [ y, u, b e a r s the s a m e relation to the [ u l and t o u r s do. La J 81 f a m i l i e s a s the [i] type con- As we have s a i d t h e s e r e g u l a r i t i e s a p p e a r c l e a r l y only if c o m - p a r i ~ o n sa r e m a d e f r o m the point of view of the l o w e r jaw. Examining t h e s e groupings we cannot fail to r e c o l l e c t how the m u s c u l a r c o n t r o l of the tongue i s organized. As i s shown i n Fig. 11-A-3, which we have taken f r o m Ladefoged ( 6 ) , the tongue h a s t h r e e m a j o r e x t r i n s i c muscles: T h e- g m u s c l e i s attached to the mandible and c a n b e used t o push the tongue body f o r w a r d s and to d e p r e s s the tongue tip. It s e e m s reasonable to a s s u m e that c e r t a i n p a r t s of this m u s c l e should be r e c r u i t e d i n raising the tongue towards the h a r d palate; The styloglossus m u s c l e pulls the tongue backwards and upwards. I t is aided i n this action by the palatoglossus m u s c l e ; The hyoglossus m u s c l e a p p e a r s capable of pulling the body of the tongue backwards and downwards. Thus i t s action m a y b e invoked t o d e c r e a s e the c r o s s - s e c t i o n a l a r e a of the pharynx. Such a d e c r e a s e could a l s o b e brought about by the gl;.ssapharyngeus. T h e s e functional ariatomical f a c t s would make us expect t o find p r e c i s e l y those tongue shapes that we o b s e r v e i n F i g s . 11-A-1 and 11-A-2, provided that one of the t h r e e m u s c l e complexes dominates o v e r the o t h e r two i n the coritraction pattern. This finding provided s o m e physiological justification f o r o u r d e c i sion t o d e l i m i t the c l a s s of tongue s h a p e s i n t h e model to the [ i ] , [ u ] , and [ a ] contours and to s h a p e s that we obtained by interpolating l i n e a r ly between t h e s e b a s i c contours. After studying a l s o s o m e photographic F i g . 11-A-1. Midsagittal tongue contours for Swediah voD.velain relation to outline of mandible. Top left: [ i , e , e , m]. Top right: [u]. Below: [ a , o, $. Fig. 11-A-2. Midmagittal tongue contours for Swedimh vowels in relatiqn to outline of mandible. Top left: L y , u. b ] . Top rlght and below am in Fig. 11-A-1. STL-QPSR 1/1969 18. data of l a b i a l articulation we c h o s e a three-dimensional "horse-shoe" model f o r the r e p r e s e n t a t i o n of the lips. E a c h l i p contour is d e s c r i b e d mathematically by m e a n s of a power function (7). The p a r a m e t e r s of this analytical e x p r e s s i o n depend on the action of the "labial" m u s c l e s When t h e i r values have b e e n computed a s well a s that of the "mandible". a s a function of "labialization" and "jaw opening" i t is possible to d e r i v e t h e v e r t i c a l s e p a r a t i o n between the lips ( h i n F i g , 11-A-4) and the d i s t a n c e between the mouth c o r n e r s (w) and, by m e a n s of t h e s e n u m b e r s , the mouth opening a r e a (A). This m e a s u r e i s a d e t e r m i n a n t of the acoustic p r o p e r t i e s that a r e c o r r e l a t e d with the auditory quality of a vowel, that i s , the f o r m a n t frequencies. Another acoustically important m e a s u r e Where does the vocal t r a c t t e r m i n a t e ? is the length of the l i p section. In o u r model the vocal t r a c t t e r m i n a t e s a t the point w h e r e the l i p contours intersect. Both the opening a r e a and the length of the t r a c t a r e functions of rounding o r spreading on the one hand, and the position of the mandible, on the other. Since o u r X - r a y d a t a showed c e r t a i n v a r i a - tions of l a r y n x height this v a r i a b l e was a l s o introduced i n the model. T o s u m m a r i z e o u r model is m a d e up of the following independently controllable components: a mandible, a tongue, a q u a s i - m u s c u l a r and jaw-independent "labialization" p a r a m e t e r and l a r y n x height. It was now possible f o r u s to compute new "synthetic" a r t i c u l a t o r y configurations, o r X - r a y t r a c i n g s o r p r o f i l e s , by combining any given jaw position with a n a r b i t r a r y but physiologically f a i r l y r e a l i s t i c tongue shape and l a r y n x height. We could f u r t h e r m o r e introduce a perturbation of t h e b a s i c n e u t r a l jaw-dependent shape of the l i p s b y "rounding" o r "spreadingt1. Like the child who is about t o d i s c o v e r the potentialities of its s p e e c h o r g a n s we now wanted t o find out, of c o u r s e , what t h e s e new o r synthetic mouth positions sounded l i k e ? Thanks to the w o r k b y F a n t ( 8 ) , Heinz and Stevens (9), a h m a n ( l o )and o t h e r s we now know how t o utilize the information i n a n X - r a y profile of the vocal t r a c t i n o r d e r t o compute the acoustic r e s u l t , that i s , the f o r m a n t f r e q u e n c i e s , a s s o c i a t e d with a given a r t i c u l a t o r y vowel configuration. The p r o c e d u r e c o n s i s t s initially i n m e a s u r i n g t h e d i s t a n c e between tongue contour and the fixed and l e s s mobile s t r u c t u r e s a t a n u m b e r of points along the t r a c t . To make t h e s e m e a s u r e m e n t s , a coordinate s y s t e m of the type shown a t the bottom of Fig. 11-A-4 is used. T h e s e c r o s s d i s t a n c e s a r e then t r a n s l a t e d into e s t i m a t e s of the corresponding c r o s s - s e c t i o n a l a r e a s . tion gives a so-called a r e a function. This transforma- Given this a r e a function i t is Fig. II-A-4. In the upper left-hand p a r t the p a r a m e t e r s determining t h e mouth opening a r e a A a r e shown: h = vertical separation between lips; w = distance between mouth c o r n e r s ; p i s a number that specifies the curvature of the lip contours. T h e s e contours when projected on a frontal plane a r e a s sumed t o be given by In the upper right part the basic tongue shapes of the model a r e shown. A polar coordinate s y s t e m defined i n relation t o the mandible i s a l s o indicated. With the aid of this coordinate interpolated tongue shapes associated with [ i , u l and [ a ] w e r e computed. In the lower part of the figure a l a t e r a l X - r a y tracing c a n be seen. Superimposed on the profile i s a coordinate s y s t e m defined in relation to fixed s t r u c t u r e s such a s the maxilla. This s y s t e m was used in the determination of a r e functions. 19. STL-QPSR l / l 9 6 9 possible t o compute the f o r m a n t frequencies that a r e a s s o c i a t e d with the a r t i c u l a t o r y configuration i n question. O u r model i s related to the t h r e e - p a r a m e t e r models of F a n t (8)and I). of Stevens and House One f e a t u r e distinguishing o u r s f r o m previous models is the c o n s t r a i n t s that we introduce on the c l a s s of a r e a functions that the model g e n e r a t e s . physiological grounds. We have t r i e d t o motivate t h e s e c o n s t r a i n t s on In previous a t t e m p t s the mandible, f o r instance, w a s not a n independently controllable p a r a m e t e r . The babbling s t a g e of the model L e t u s proceed t o study the relation between articulation and acoustic r e s u l t f o r the model a s i t now stands. F i r s t examine what will happen when we fix the jaw and move only t h e tongue. [i] t o [ u ] t o [ a ] and then back t o [i] again. 1 We change its f o r m f r o m The acoustic r e s u l t s of t h e s e a r t i c u l a t o r y changes a r e shown i n Fig. 11-A-5 i n which the second f o r m a n t frequency is plotted against the f i r s t . Data pertaining t o t h r e e different d e g r e e s of jaw opening (5, 9, and 22 m m ) a r e shown. The u p p e r d a t a points of t h e figure, that i s , m a x i m a l F 2 v a l u e s t a r e a s s o c i a t e d with t h e [i] tongue contour. P e r t u r b i n g this tongue shape s o that i t i s t r a n s f o r m e d into a n [ u ] then a n [ a ] and finally back to the [i] shape again gives r i s e t o banana-like t r a c e s on t h e F -F c h a r t f o r a l l t h r e e jaw2 1 positions. T h e s e bananas i n f o r m u s that the change i n the f i r s t f o r m a n t frequency is insignificant i n c o m p a r i s o n with the change i n the second f o r m a n t when only the tongue of the model i s moved. F r o m s i m i l a r d i a g r a m s we c a n i n f e r t h a t if only t h e "mandible" is moved the change i n the f i r s t f o r m a n t is c o n s i d e r a b l e w h e r e a s t h a t i n the second f o r m a n t is relatively s m a l l . Now a s s u m e that like the child we combine mandible positions, tongue s h a p e s , and l i p s t a t e s i n a l l possible ways and l i s t e n t o t h e acoustic r e s u l t i n e a c h individual c a s e . Whatever we do with o u r a r t i c u l a t o r y com- ponents i t is c l e a r that the human s p e e c h o r g a n s a r e constrained i n s u c h a way s o a s to p e r m i t only c e r t a i n vowel qualities, o r combinations of C e r t a i n o t h e r F -F -F combinations c h a r 3 2 1 a c t e r i z e vowels that we could produce only with t h e aid of a t e r m i n a l f o r m a n t frequency values. analogue s y n t h e s i z e r - not with o u r mouths. F r o m the point of view of t h e human s p e e c h m e c h a n i s m such combinations would b e impossible. SECOND FORMANT FREQUENCY 20. STL-QPSR l / l 9 6 9 The s p a c e that c h a r a c t e r i z e s the acoustic possibilities of o u r model is shown i n Fig. 11-A-6 i n the f o r m of two two-dimensional projections. At the top we s e e a n a r e a that e n c l o s e s a l l possible combinations of Fg and F values. 1 binations. Below the corresponding field including a l l F2 and F1 c o m - Universal phonetic dimensions of vowels: An underlying principle Suppose t h a t we would want the vowel sounds that we s e l e c t i n o r d e r t o c o n s t r u c t a hypothetical vowel s y s t e m t o exhibit m a x i m a l efficiency i n o r a l communication and i n p r a c t i c a l use. T o attain this goal i t would ap- p e a r r e a s o n a b l e to s e e t o it that acoustically, o r pe rceptually, they would b e 2 s f a r a p a r t a s possible. This p r i n c i p l e of m a x i m a l c o n t r a s t (cf. Jakob s on (12)) would m a k e the r i s k of confusing vowels s m a l l e r - a risk that is inherent i n the inevitably noisy communication p r o c e s s . Adhering t o this principle we begin t o look f o r vowel qualities a t the e x t r e m e s of the acoustic s p a c e of o u r a r t i f i c i a l mouth. A three-vowel s y s t e m , f o r example, might f o r i n s t a n c e c o n s i s t of the vowels c o r r e s p o n d ing t o points A, B , and C i n Fig. 11-A-6. We find that t o a t t a i n A, B, and C t h e r e is only one way i n which t h e a r t i c u l a t o r y components of the model c a n b e positioned. F o r t h e s e ex- t r e m e c o r n e r s of the model s p a c e t h e r e is consequently a unique relation between articulation and acoustic result. An A-pattern c a n b e produced only i f the jaw opening is s m a l l ; - the l i p s a r e s p r e a d ; and t h e tongue i s t h r u s t forward and upward. A B - p a t t e r n is obtained only if the pharynx is constricted; the jaw opening is l a r g e ; and the l i p s a r e s p r e a d . A C - p a t t e r n will r e s u l t I )nl'y i f the vocal t r a c t is constricted a t a point approximately half-way between the lips and the glottis; the jaw opening is s m a l l ; and the l i p s a r e rounded. As you have probably guessed a l r e a d y , A. B, and C c o r r e s p o n d to the qualities [i,], [a 1, and [u]. . t U z 3 .O 9 kHz 0 - A Spread ---- - Rounded Rounded and -a-m-.- - I I - 0 /) C \ - \u-- - + # / ,' 0 -I4 I I I I \ I - 1 \ \ - I 1 \ - I I - I B I I I I - I rC / 0 4. ew-.- 0 s-n---,-,d- L*' I I I I I I I .I .2 .3 .4 .5 .6 .7 .6 0 FIRST FORMANT FREQUENCY Fig. 11-A-6. T w o - d i m e n s i o n a l p r o j e c t i o n s of t h e a c o u s t i c vowel s p a c e of t h e m o d e l . At t h e t o p a n a r e a e n c l o s i n g a l l p o s s i b l e c o m b i n a t i o n s of F and F1; below t h e c o r r e s p o n d i n g field 3 including a l l FZ and F c o m b i n a t i o n s . 1 . .8 kHz SECOND AND E 0 g g *N I I I SECOND FORMANT A b 0 , b I , b I k I A I -. b g I I I ) FREQUENCY A i - THIRD FORMANT FREQUENCIES a b A , I o I r P I E 1 2 I g N I I b $ g STL-QPSR 1/1969 22. It is a m a t t e r of common phonetic observation t h a t the languages of the r y vowels. world pick t h e i r vowel qualities f r o m among just this i n v ~ ~ l t o of T h i s f a c t i s brought out i n Fig. 11-A-9 which p r e s e n t s a s u r v e y of vowel systems. T h e s e d a t a have b e e n taken f r o m Hockett (I3). T h e vowels a r e plotted on a n i m a g i n a r y conventional f o r m a n t c h a r t r a t h e r than i n the t r a d i tional quadrilateral. The p a t t e r n s that a r e shown i n Fig. 11-A-9 f o r t h r e e through ten-vowel s y s t e m s a r e variations on the vowel p e r i p h e r y theme, O u r conclusion m u s t b e that the phonetic s t r u c t u r e of t h e s e s y s t e m s c a n b e explained a t l e a s t p a r t i a l l y by the principle of m a x i m a l contrast. I T a b l e 11-A-2 tabulates t h e a r t i c u l a t o r y dimensions that uniquely c h a r a c t e r i z e the e x t r e m e vowel types. TABLE 11-A-2 THE ARTICULATORY DIMENSIONS O F PERIPHERAL VOWEL TYPES TONGUE SHAPE Palatal Close Palato-pharyngeal Pharyngezl Vclar u i o e a JAW Open a w E The points i n Fig. 11-A-8 labeled [ i , e, and tongue shape a s f o r [i]. E] I w e r e generated with a T h e r e f o r e we c l a s s i f y t h e m a s palatal. They d i f - f e r i n t e r m s of jaw-opening and a l l require s p r e a d lips. The [ a ] vowel required a tongue shape which was i n t e r m e d i a t e between [i] and [ a ] . We c l a s s i f y i t a s palato-pharyngeal. The lips has to be spread. The [ a ] vowel presupposed a n open jaw and a n r a W e c l a s s i f y i t a s pharyngeal. s h a p e tongue. The lips had to b e s p r e a d . The [ a ] vowel a l s o required s p r e a d lips. - It is pharyngeal i n t e r m s of the model tongue contour and open ( i n t e r m s of t h e jaw). VOWEL SYSTEMS ACCORDING TO HOCKETT 3-vowel r y r t e m : : i ARABIC ARUNTA a U (some dialects) (~ustralia) ' GERMAN DUTCH HUNGARIAN ( s o m e d i a l e c t s ) FRENCH ( s o m e dialects) ROMANIAN ( / and a instead of y LIFU (sand a instead of y VOTYAK (iand e instead of y SALISHAN 4-vowel rymtemr: I e I a 0 APACHEAN CAMPA NAHUATL a a } ITALIAN PORTUGUESE LOMA (west-Africa) KIOWA (Mexico) TETELCINGO (Mexico) (Peru) (Mexico) (AMAHUACA) U (Philippines) (Peru, Brazil) FINNISH HUNGARIAN ( s o m e d i a l e c t s ) CUITLATECO ( + , a , 3 instead of y . d ,a) KOREAN ( + , a , 3 instead of y , 0 some dialects) 5-vowel r y r t e m r : I JAPANESE CZECH GREEK YOKUTS KECHUA e a U I 0 e a +Y ++ U a (California) ( P e r u , Chile) { :%:FARAN(North-Caucasia) I POLISH U o3 ' e POTAWATOMI U CHIPEWYAN DARGWA MENOMINI PERSIAN UKRAINIAN YUCHI I TAKI-TAKI e a 0 u 0 a instead of 8 ) A instead of 8 ) HUICHOL ONEIDA I (some dialects; PORTUGUESE ( s o m e d i a l e c t s ; (North-America) U TRUKESE (Malayo-Polynesian) OTOMI ( C e n t r a l Mexico) ENGLISH ESTONIAN (y ,0 , + t THAI- (North-America) 0 6-vowel rymtemr: and 0 ) and d ) and 0 ) a +a 0 J , E instead of i . a ,s North- A m e r i c a ) [North-Caucasia) (North-America) MIXTECO SIERRA POPOLUCA ZOQVE Fig. IT-A-9. I U ' o3 +Y a +gj NORWEGIAN ( N o r t h W e a t e r n Norway) +ce Survey of vowel ryrtemr as e x e w f i e d By vartarre languages in the world. The data have been taken f r o m Hockett. ,=I The [o] vowel w a s obtained with a pharyngeal tongue s h a p e and rounded lips. T h e jaw was i n a half-open position. - - The [u] n e c e s s i t a t e d a v e l a r tongue shape, a c l o s e jaw and rounded lips. i n Sound P a t t e r n of English Chomsky and Halle ( 3 ) s k e t c h a t h e o r y of marking. According t o this theory any deviation that t h e ph(>nr)logical s y s t e m of a language exhibits f r o m the so-called "natural" p a t t e r n s i m plies a n i n c r e a s e of the complexity of the system. An "unnatural" o r un- expected phonological segment is m a r k e d with r e s p e c t t o t h e f e a t u r e f o r which it h a s the unexpected value. ated a n expense is i n c u r r e d - When the n a t u r a l n e s s condition is viol- a n expense that m u s t b e taken into account i n the phonological budget of the language. Among the u n i v e r s a l n a t u r a l laws regarding vowels t h a t Chomsky and Halle mention w e find the observations t h a t b a c k vowels a r e usually rounded w h e r e a s front vowels a r e not. As w e have s e e n t h e s e "laws" a r e m o s t s i m p l y explained by the principle of m a x i m a l c o n t r a s t which f a v o r s the u s e of e x t r e m e vowel types. By rounding the "back" vzwels and spreading t h e "front" s e r i e s we r e a c h t h e l i m i t s of the vowel space. Phonetic dimensions of Swedish vowels F r o m the point of view of o u r model what a r e the phonetic dimens i o n s of t h e Swedish vowels? would b e t o "go fishing". Well, one way of answering this question T h e "fish" that we would like t o c a t c h a r e the positions of the Swedish vowels i n the acoustic s p a c e and o u r "fishing nets" a r e the c u r v e s o r n o m o g r a m s that we c a n g e n e r a t e with o u r model. This method, which i s a s o r t of analysis-by-synthesis p r o c e d u r e , provides a n a n s w e r t o t h e question: How should t h e tongue, the l i p s , the mandible, and t h e l a r y n x of the model b e positioned i n o r d e r to g e n e r a t e the vowel sounds of Swedish? F o r o u r a r t i f i c i a l child t o l e a r n the c o r r e c t pronun- ciation of t h e s e vowels what a r t i c u l a t o r y p a t t e r n s should i t be a b l e to master? Fig. 11-A- 10 shows a "fishing net" and s o m e "submerged" Swedish vowels. The solid l i n e s r e p r e s e n t iso-lingual c u r v e s . indicate iso-mandibular conditions. The dashed l i n e s The m e s h e s of the "net" a r e r a t h e r open but i n principle we should be a b l e t o i n f e r the combination of a r t i c u l a t o r y p a r a m e t e r values that u n d e r l i e s any "vowel fish" f r o m the information provided by the "net" o r nomogram. 1 Tongue shapes: i, i - a , a J a w openings : 5, 7, 11, 15.22 Lips spread -isolinguals isomandibular s -- --- @ Formant frequencies of the subject 2.2 kHz kHz F i g . 11-A-10. C u r v e s i l l u s t r a t i n g t h e effect o n F1 and F of i s o l i n g u a l 2 ( d a s h e d ) and i s o m a n d i b u l a r ( s o l i d ) conditions. T h e a d v a n t a g e s a n d l i m i t a t i o n s of using c u r v e s of t h i s type t o d e t e r m i n e a r t i c u l a t o r y p a r a m e t e r v a l u e s a s s o c i a t e d with vowel f o r m a n t f r e q u e n c y d a t a , a r e d i s c u s s e d i n t h e text. All this sounds fine but we soon r e a l i z e that, i n c o n t r a s t with the conditions f o r the e x t r e m e vowel types, the relation between articulation and acoustic r e s u l t is ambiguous. It i s possible t o c a t c h a given "fish" using different types of nets, f o r instance, rounded o r s p r e a d "nets". It might b e suggested that we introduce c e r t a i n principles i n o r d e r t o r e d u c e t h e ambiguity. One possibility would b e t o s a y , f o r instance, that a vowel f o r m a n t p a t t e r n that c a n b e produced i n m o r e than one way should b e generated using the a r t i c u l a t o r y co nfiguration that constitutes the m i n i m a l deviation f r o m the b a s i c s p e e c h p o s t u r e o r articulation b a s e of the language i n question. However, we a r e a s yet unable t o apply this principle i n a quantitative and Moreover, we m u s t not r a s h l y a s s u m e that t h e child meaningful m a n n e r . who l e a r n s t o s p e a k n e v e r m a k e s u s e of the possibility of compensatory articulation o r that the vowel acquisition problem is alway solved by a l l If s u c h assumptions a r e i n c o r r e c t t h e r e c h i l d r e n i n a n identical m a n n e r . is l e s s need f o r a disambiguation principle. Returning t o F i g s . 11-A-1 and 11-A-2 we o b s e r v e that t h e [ i ] shows a maximally palatal tongue contour. The vowels [ e , than [ i j . E , ie] T h i s i s a l s o t r u e of [ Y ] i n Fig. 11-A-2. s e e m to have a somewhat m o r e constricted pharynx It i s of i n t e r e s t t o note t h a t the p e r i p h e r a l vowel types of the model that w e r e labeled [ e , E, E) w e r e found just along a continuum of i n c r e a s i n g jaw-opening and pharyngealization. exhibit varying d e g r e e s of pharyngealization. a r e both l e s s palatal than Ky]. In t h e s e t r a c i n g s [ a , o, [ u ] i s velar. [b] Q] T and [u] Whether they should b e c l a s s i f i e d a s p a l a t o - v e l a r o r pzlato-pharyngeal cannot b e d e t e r m i n e d with c e r t a i n t y f r o m t h e s e data. I t would b e perfectly f e a s i b l e to u s e , f o r i n s t a n c e , a l e a s t s q u a r e s c r i t e r i o n to study the m i x t u r e s of the b a s i c tongue shapes [ i , a , u] that provide the c l o s e s t fit to the e m p i r i c a l contours. Systematic differences between the model predictions and the d a t a that r e m a i n even when the d e viations have been minimized could f u r n i s h clues to a revision and refinement of the model. We have initiated this p a r t of the p r o j e c t only recently. So f a r we have nothing to report. Fig. 11-A-11 shows l i p and mandible data. The position of the mouth c o r n e r along a d o r s a l - v e n t r a l dimension a s a function of jaw -opening. The solid line shows the jaw-dependent n e u t r a l location of the mouth c o r n e r a s predicted by the model. T h e rounded c l a s s of vowels is above this line and the s p r e a d group below it. 5 10 JAW F i g . 11-A- 11. 15 20 OPENING T h e position of t h e mouth c o r n e r along a d o r s a l - v e n t r a l d i m e n s i o n a s a function of jaw opening. S a m e Swedish s u b j e c t a s i n F i g . 11-A-1 and 11-A-2. The solid line shows the jaw-dependent n e u t r a l location of the mouth c o r n e r a s p r e d i c t e d b y t h e model. 25. STL-QPSR 1/1969 T h e s e phonetic d a t a a r e s u m m a r i z e d i n Fig. 11-A- 12 i n the f o r m of a vowel c i r c l e in which the d e g r e e of opening i n c r e a s e s towards t h e p e r i phery and the shape of the tongue is indicated i n t e r m s of the angle b e tween the vowel i n question and one of the b a s i c dimensions: pharyngeal o r velar. palatal, Rounded vowels a r e shown within c i r c l e s . In Table 11-A-3 we have attempted to t r a n s f o r m the phonetic i n f o r m a tion of F i g , 11-A-12 into a b i n a r y format. We e n t e r a + a t the i n t e r s e c t i o n of a given column and row of the m a t r i x i f o u r d a t a have shown that a vowel s e g m e n t p o s s e s s e s a c e r t a i n attribute; a - if i t does not. T o exemplify: a l l vowels enclosed by t h e radii labeled palatal and pharyngeal r e c e i v e a plus i n t h e rows corresponding to both of t h e s e f e a t u r e s . PALATAL i y TABLE 11-A-3 u u e B i + r PHARYNGEAL - VELAR - CLOSE + OPEN - LABIAL (ROUNDED) ' , - I - - + + - + , + + + - + + + + + - - - - - - - - + + + + + -lj -- - - + - - - .- - --. - - + + + - - - + + - - + -. a + + - a a E 0 - - t - i - - - - + + + _I ' - - - the phonetic and phonological dimensions -of vowels On the relation between The child who a c q u i r e s the c o r r e c t pronunciation of say, the Swedish vowels /i, y, u/ and /u/ d e m o n s t r a t e s that he h a s l e a r n e d c e r t a i n m o t o r s k i l l s ; he h a s l e a r n e d how t o coordinate the m u s c u l a r activity underlying c e r t a i n positions and movemerrts of h i s l i p s , tongue and jaw etc. In s o m e d i a l e c t s a c o r r e c t pron~lnciationof t h e s e vowels i m p l i e s diphthongizatiorl. We o b s e r v e [ i j ] and cyj] on the one hand and [up] and [wf3] on the other. C o r r e c t pronunciation i m p l i e s a l s o that a l l o t h e r vowel s e g m e n t s a r e monophthongs. Consequently a n informant who p r e s e n t s t h e s e f a c t s i n h i s s p e e c h h a s d i s c o v e r e d not only that the Swedish vowels /i, y, w, u/ presuppose c e r t a i n a r t i c u l a t o r y events but a l s o t h a t they d i f f e r f r o m the remaining vowels; they constitute a c l a s s , arid the vowels of t h i s c l a s s a r e subject to diphthongization. T h i s p r o p e r t y is a s t r u c t u r a l c h a r a c t e r i s t i c , a r e g u l a r i t y i n the sound p a t t e r n of the language: a n a b s t r a c t phonological property. PALATA OPEN Fig. 11-A-12. The phonetic data presented in F i g s . 11-A- 1 , 11-A-2,and 11-A-11summarized. Tongue shape i s indicatdd a s a function of angle. Degree of mandibular opening i s shown a long the radii. Rounded vowels are encircled. 26. STL-QPSR 1/1969 The d i s c o v e r y of this p r o p e r t y i s made on a l e v e l that is m o r e "abstract" than that on which the d e t a i l s of the motor control underlying [ i j , yj, up] and [up1 a r e discovered. In the l i n g u i s t ' s d e s c r i p t i o n the c h i l d ' s d i s - c o v e r y of the s u p e r o r d i n a t e c l a s s p r o p e r t y i s reflected i n the phonological function of the distinctive f e a t u r e s and the child' s d i s c o v e r y of m o t o r Hence commands auditory i r n p r e s s i o n s etc. i n t h e i r phonetic function. the need to distinguish between phonological and phonetic f e a t u r e s . The only phonetic data n e c e s s a r y to the formulation of the phonological r u l e One might operative i n the above example is the f e a t u r e diphthongization. thus b e tempted to suggest that i n the lexicon s e t up f o r the dialect i n question within the f r a m e w o r k of a generative g r a m m a r the vowel s e g m e n t s /i, y, u, u/ b e specified a s /t diphthongizable(. It would i n principle b e possible t o give many m o r e examples of s i m i l a r " c l a s s e s " of s e g m e n t s in the sound p a t t e r n s of v a r i o u s languages. F o r e v e r y r e g u l a r i t y of this type, f o r e v e r y c l a s s of s e g m e n t s that behaved i n a s y s t e m a t i c m a n n e r i n relation to a given p h o ~ o l o g i c a lp r o c e s s we would b e able to e x t r a c t a n "abstract" phonological p r o p e r t y o r feature. I L e t u s now take the phonological f e a t u r e of diphthongization that w e have just c o m e a c r o s s and a s k f o r a m o r e detailed phonetic i n t e r p r e t a t i o n of it. What a r e the phonetic f e a t u r e s that distinguish /i, a l l o t h e r vowels and thus c h a r a c t e r i z e t h e m a s a c l a s s ? 11-A-3 i t a p p e a r s that the phonological c l a s s of y, u, u/ f r o m F r o m Table /+diphthongizable/ seg- m e n t s i s d e s c r i b e d by a single phonetic feature, viz., c l o s e n e s s . /i, y, u, u/ a r e the only [t c l o s e ] vowels. F u r t h e r m o r e we s e e that /u/ - and /u/ whose diphthongization d i f f e r s f r o m that of /i/ and /Y/ s e r v e [ i j ] , Cyj] along with [ u p ] and p r o p e r t y of being [ + velar]. [LIB] not +t[,uj] and "[uj]) - (we ob- s h a r e the Consequently, t h e s e observations suggest that r u l e s of t h e following type be incorporated in the g r a m m a r : /t diphthongizable/--, [ t clos e l (1) w h e r e s l a n t p a r e n t h e s i s b r a c k e t s indicate phonological and s q u a r e phonetic l e v e l s of r e p r e s e n t a t i o n respectively. of the f e a t u r e of /+ This r u l e s t a t e s that a l l mentions diphthongizable/ a t the l e x i c a l and phonological l e v e l s should be replaced by [+ c l o s e ] a t a lower phonetic level. If the relation between phonological and phonetic dimensions w e r e a l ways a s s i m p l e and s t r a i g h t - f o r w a r d a s i n the above example r u l e s such a s (1) would not b e n e c e s s a r y in the g r a m m a r . In the lexicon s e g m e n t s 1 STL-QPSR 1/1969 27. like /i, y, u, u/ might just a s well b e specified a s [ + c l o s e ] f r o m the s t a r t and among the phonological pules of the d i a l e c t i n question we would find the following diphthongization rule: - 7 tvoc -cons -cons +close +tense cr palatal +close $velar apalatal y labial $velar y labial etc, i L w h e r e a , $, and y a r e v a r i a b l e s assuming t h e values + or -. This r u l e should b e i n t e r p r e t e d i n a c c o r d a n c e with c u r r e n t conventions of phono- . logical notation (3, 14) The f e a t u r e s of vocalic, consonantal and t e n s e a r e used h e r e a s in previous formulations of distinctive f e a t u r e theor)( 15*3) the remaining f e a t u r e s being defined within o u r p r e s e n t phonetic framework. .c T h i s rule will g e n e r a t e the sequences [ij], Lyj3, [up] and [ a $ ] provided 4 that we a s s i g n the following m a t r i c e s t o the segments: + + + + - - - - + - labial - + + - + + + + + + + + velar + + + + + palatal + + VOC close 35 t + - + + - + + Note that the introdr~ct i o r , of t h i s p r o c e s s m;ilces t h l , . . xces of f e a t u r e s u n n e c e s s a r y i n the d e s r r i p t i o t ~of S \ i - c ~ r l i s l l (cf. ~ a l ~ ~ ~ l b ~ ~ ~ ~ The e x c e s s i v e d e g r e e of roc~ndirig ( i n s c , ~ ~ t ~ c lobserved in~) f o r [u](and [ u ] ) c a n be attributed to t h e l a b i a l lide LB]." Accr)rding to o u r P L I . ~ ~ , , ~ c l a s s i f i c a t i o n /u/ i s distinct f r o m i/ i n t e r m s of velarizntic,n 2nd rounding and f r o m /y/ i n t e r m s of velarization only, Table 11-A-3. As pointed out classifying /u/ a s velarized r a t h e r than pha ryngeali zed i s somewhat a r b i t r a r y a s indicated by the p r e s e n t phonetic evidence. It i s c l e a r , however, that [ i j and Cy] a r e phonetically both m o r e palatalized (= s h a r p ) than [ u ] . T h e s e f a c t s a r e thus compatible with the dimensions of F a n t ( 2 ) . f 28. STL-QPSR 1/1969 Thus ( 2 ) s t a t e s that a glide is i n s e r t e d a f t e r any long and c l o s e vowel and that this glide a s s i m i l a t e s a l l the remaining f e a t u r e s of the preceding vowel. What we have done is t o a s c r i b e a phonological function t o t h e phonetic f e a t u r e of c l o s e n e s s . This function is distinct f r o m i t s phonetic tole and m o r e a b s t r a c t i n that i t c o n s i s t s i n t r i g g e r i n g the diphthongization r u l e thus delimiting a c e r t a i n vowel c l a s s . Thus i n the g r a m m a r t h e r e s e e m s t o b e no need t o distinguish between the phonological f e a t u r e of diphthongization and t h e phonetic f e a t u r e of c l o s e n e s s s i n c e + c l o s e d o e s the job alone a t both levels. As pointed out above f u r t h e r examples of the c i r c u m s t a n c e that phonological p r o c e s s e s a p p e a r to o p e r a t e on n a t u r a l c l a s s e s could b e given. Such examples have m a d e investigators g e n e r a l i z e and conclude that i t "is a n e x t r e m e l y important and by no m e a n s obvious fact that the distinctive f e a t u r e s of the c l a s s i f i c a t o r y phonemic* m a t r i x define c a t e g o r i e s that c o r r e s p o n d closely t o those d e t e r m i n e d by the rows of the phonetic matrices". (16) The exclusion of r u l e s s u c h a s (1) f r o m a theory of g r a m m a r would b e justified provided that assumptions like the following w e r e c = r r e c t : A1 Phonological f e a t u r e s a r e i n a 1-1 relation with phonetic features. A2 The relation between phonological and phonetic f e a t u r e s i s n e v e r language-specific. In t h e s e s t a t e m e n t s phonological f e a t u r e r e f e r s t o a p r o p e r t y that a s e g m e n t o r a c l a s s of s e g m e n t s h a s on account of i t s s y s t e m a t i c behavior with r e s p e c t t o s o m e p r o c e s s of phonology i n the language, e. g. , diphthongization, raising, vowel harmony etc. Phonetic f e a t u r e r e f e r s to a dimension that a future u n i v e r s a l physiological and quantitative theory of s p e e c h production and recognition might reveal. (1) above. A1 is i l l u s t r a t e d by r u l e A2 would b e refuted if i t could b e shown that a phonological p r o p e r t y s u c h a s f o r i n s t a n c e diphthongizable has different phonetic i n t e r pretations i n different languages. A1 p r e s u p p o s e s that /diphthongization/ h a s a phonetic i n t e r p r e t a t i o n in t e r m s of a single dimension. A2 pre- supposes i n addition that t h i s single phonetic p r o p e r t y is always and univ e r s a l l y f o r example c l o s e n e s s . 3% = phonological a s t h i s t e r m is used above (BL/JS) 2 9. STL-QPSR l / l 9 6 9 A r e t h e s e assumptions c o r r e c t ? C u r r e n t accounts of phonological t h e o r y (3' 17) have devoted l i t t l e attention t o s u c h questions. It is c l e a r though that they will b e r a i s e d a s the modeling of phonetic "competence" p r o g r e s s e s and the notion of phonetic f e a t u r e is elucidated, ( ~ f . ' ~ ' h o w e v e r ) . C o n s i d e r a hypothetical Swedish dialect f o r which the following histor i c a l f a c t s have been observed: At a c e r t a i n s t a g e u 4 u (3) and l a t e r k ---+ q i n the context of a l l "soft" vowels (4) The c l a s s of "soft" vowels includes [i, e , E, 8, 8, _a, y]. In a synchronic d e s c r i p t i o n of the d i a l e c t a f t e r r u l e (4) has been incorporated we e x t r a c t a phonological p r o p e r t y x c h a r a c t e r i s t i c of the "soft" vowel c l a s s . p r o p e r t y c o n s i s t s i n a "softening" of the [k]. front of the "hard" vowels /u, u , o, a/. This T h i s does not o c c u r i n Going back to Table 11-A-3 what i s the phonetic i n t e r p r e t a t i o n of the phonological f e a t u r e of [ k ] softening? We find that r+ palatal 7 Consequently we have found a n example f o r which the 1-1 relation d o e s not hold. A1 is violated. A c e r t a i n phonological p r o c e s s , v e l a r softening, is t r i g g e r e d by two phonetic f e a t u r e s . Now a s s u m e that A1 is a u n i v e r s a l law that c a n n e v e r b e violated i n human language. If t h i s i s t r u - t h e r e is only one condition under which - both ( 3 ) and ( 4 ) c a n occur: the o r d e r of t h e s e p r o c e s s e s m u s t b e r e versed. F o r r e v e r s i n g the o r d e r m e a n s that we c a n change (5) into /+ and A1 holds. k-softening/ [ + palatal] (6) T h e r e a r e many Swedish d i a l e c t s that have undergone changes s u c h a s (3) and (4). Provided that A1 is accepted i t c a n b e p r e - dicted that no Swedish d i a l e c t could b e found f o r which the o r d e r of the h i s t o r i c a l events is (3)-(4). The o r d e r (4)-(3) is found e m p i r i c a l l y would provide support i n favor of A1.+ - + - - Wess6n ( I 8 ) c l a i m s that (4) was initiated a s e a r l y a s 1225-1375 in c e r t a i n Swedish d i a l e c t s w h e r e a s t h e f i r s t t r a c e s of (3) a p p e a r i n the period 1375-1526. The exact r e l a t i v e chronology of (3) and (4) f o r individual d i a l e c t s i s not indicated. We have not yet been able t o t e s t the p r e d i c tion based on A l , Our hypothetical examples have been intended t o r a i s e doubts regarding the validity of Al. Moreover, they indicate how t h i s assumption could b e t e s t e d against h i s t o r i c a l facts. Before concluding l e t u s examine one m o r e example involving A1 i n o r d e r t o d e m o n s t r a t e how useful a tool this a s sumption might b e provided that i t w e r e t r u e i n the s e a r c h f o r the phonetic dimensions of segments. In English the c l a s s of s e g m e n t s consisting of [ s z 3 E characterized on two levels. 2) c a n b e Phonologically, nouns ending i n t h r e e con- sonants r e q u i r e that the p l u r a l b e f o r m e d by adding [&z). Phonetically, t h i s c l a s s a r e e x h a u s t i v e l y d e s c r i b e d a s n o n g r a v e a n d s t r i d e n t (19) . Phonologically t h e consonants s h a r e the p r o p e r t y of requiring [ I z realization of the p l u r a l morpheme. p r e t e d i n t e r m s of two f e a t u r e s . Now I If s o the phonetic d e s c r i p t i o n just One way of doing this would be t o c l a s s i f y [ f ] and [ v ) a s [ - s t r i d e n t ] instead of [+strident]. f r o m [ 8 ] and a Phonetically t h i s p r o p e r t y is i n t e r - Consequently A1 d o e s not hold. suppose again that A1 m u s t hold. given m u s t b e modified. 1as [a] i n t e r m s of gravity. c l a s s of [ + s t r i d e n t ] segments. [s A1 i s met. z They will s t i l l b e distinct 3 E f ] now constitute the Changing the value of s t r i d e n c y f o r [ f ] and [ v ] docs not lead to a phonetic d e s c r i p t i o n that i s too much a t v a r i a n c e with any known facts. s i t y the Engliah [ f ] and [ti] everything e l s e being equal. On the c o n t r a r y , i n t e r m s of physical inten- souncle a r e c l e a r l y w e a k e r than [ s z 4. Z They a r e thus a l s o l e s s strident. E] Other s i m i l a r phonetic a r g u m e n t s have been put forward i n favor of considering [ f ] and [ v ] a s [ - s t r i d e n t ] ( 2 0 ) The important point about the above i l l u s t r a t i o n is the demonstration that phonological f a c t s could be used to a n s w e r the question: What a r e the t r u e phonetic dimensions of vowel and consonant s e g m e n t s ? They m a y b e used i n this way however only on the condition that A1 is valid. To s u m m a r i z e : I s i t valid? in this section we have s a i d that (i) phonetic dimensions have to b e distinguished f r o m phonological dimensions. We have asked (ii) how t h e s e dimensions a r e related: a ) i s the relation 1- 1 ? b) is i t language-dependent? We have r a i s e d the question (iii) whether r u l e s that i n t e r p r e t phonological f e a t u r e s phonetically m u s t b e incorporated into a g e n e r a l linguistic t h e o r y and in individual g r a m m a r s . Such r u l e s have not bee11 considered s o f a r i n phonological the0 ry. The a n s w e r t o the last question depends on t h e a n s w e r s t o (ii a ) and (ii b). F o r instance, if the relation between phonological and phonetic f e a t u r e s is one-to-many o r if t h i s relation i s language-dependent r u l e s of the above kind m u s t b e incorporated into linguistic description. r u l e s , then, d e s c r i b e a relation that the child h a s t o l e a r n . Such We f e e l a t p r e s e n t i t is not possible t o defend o r refute convincingly hypotheses about t h i s relation s i n c e the notions of phonological f e a t u r e and phonetic f e a t u r e need t o b e f u r t h e r clarified. Consequently f u t u r e rev e a r c h will have t o d e t e r m i n e whether "the distinctive f e a t u r e of the c l a s s i f i c a t o r y phonemic m a t r i x define c a t e g o r i e s that c o r r e s p o n d c l o s e l y t o those d e t e r m i n e d by the rows of the phonetic . m a t r i c e s " (16) Among the p r o b l e m s that this r e s e a r c h m u s t tackle is t h e construction of a u n i v e r s a l and physiologically c o r r e c t model of speech production and recognition. Only with s u c h a model would i t b e possible t o explicate t h e notion of phonetic f e a t u r e , References: (1) Malmberg, B. : "Distinctive f e a t u r e s of Swedish vowels; s o m e ins t r u m e n t a l and s t r u c t u r a l data", i n F o r Roman Jakobson (' s-Gravenhage 1956), pp. 3 16-321. I (2) F a n t , G. : "The n a t u r e of distinctive features", i n To Honor Roman ~ a k o b s o n(' s - ~ r a v e n h a ~1967), e pp. 632-642. ( 3 ) Chomsky, N. and Halle, M. : Sound P a t t e r n of English ( ~ e w York 1968). (4) Ladefog ed, P. : "Linguistic Phonetics1', Working P a p e r s in Phonetics 6, UCLA, June 1967. ( 5 ) Wang, W. S-Y.: "Vowel f e a t u r e s , paired v a r i a b l e s , and the English vowel shift", Language 44 (1 968), pp. 695-708. (6) Ladefoged, P. : "Physiological c h a r a c t e r i z a t i o n of speech", pp. 2- 9 i n Working P a p e r s in Phonetics, UCLA, June 1964. (7) Lindblom, B. : "Studies of labial articulation", Z. f. Phondik, Sprachwissenschaft und Kommunikationsforschung 21, Heft 1/2 ( 1 968), pp. 171-172. (8) F a n t , G. : Acoustic Theory of Speech Production (' s -Gravenhage 1960). (9) Heinz, 3. M. and Stevens, K. N. : "On the relations between l a t e r a l cineradiographs", p a p e r A44 i n P r o c . of 5e Cong r k s International d ' Acoustique, Likge 1965. 32. STL-QPSR 1/1969 (10) o h m a n , S. E. G. : " N u m e r i c a l m o d e l of coarticulation", J. Acoust. Soc.Am. 41 (1967), pp. 310-320. (1 1) Stevens, K. N. and House, A. S. : "Development of a quantitative d e s c ription of vowel articulation", J. Acoust. Soc. Am. 17 ( 1955), pp. 484-493. (1 2) Jakobson, R. : " K i n d e r s p r a c h e , Aphasie und a l l g e m e i n e Lautgesetze", pp. 328 -401 i n Selected Writings I (' s - G r a v e n h a g e 1962). , 1955). ( 13) Hockett, C. F. : Manual of Phon9logy ( ~ l o o m i n ~ t o nInd. (14) Halle, M. : "A d e s c r i p t i v e convention f o r t r e a t i n g a s s i m i l a r i n g and dissimilation", M a s s a c h u s e t t s Institute of Technology, Q P R No. 66, J u l y 15, 1962, pp. 295-297. (15) . . Jakobson, R., F a n t , G., and Halle, M.: P r e l i m i n a r i e s t o S ~ e e c h Analvsis he ~ i s t i n c t i v e~ e a t u r e sand T h e i r ~- o r r e- l.a..L- -e s ( c a m b r i d g e , M a s s . 1967), 7th printing. . - - - - - (16) Chomsky, N. and M i l l e r , G. A. : "Introduction t o t h e f o r m a l a n a l y s i s of n a t u r a l languages", pp. 269-321 i n Handbook of Mathematical Psychology I1 by R. D. Luce, R. R. Bush, and E. G a l a n t e r ( E d s . ) (New York 1963). York 1968). (17) P o s t a l , P. M. : A s p e c t s of Phonological T h e o r y ( ~ e w (18) Wessi.n, E. : Svensk s p r 8 k h i s t o r i a . (Stockholm 1965). I . L j u d l a r a och o r d b a j n i n g s l a r a (1 9) Halle, M. : "On t h e b a s e s of phonology", pp. 324-333 i n The S t r u c t u r e by J.A. F o d o r and J. J, Katz ( ~ n g l e w o o dCliffs, ; (20) S t e v e n s , K. N. : "Acoustic c o r r e l a t e s of place of a r t i c u l a t i o n f o r s t o p and f r i c a t i v e consonants ", M a s s a c h u s e t t s Institute of Technology, Q P R No. 89, A p r i l 15, 1968, pp. 199-205.