can

rds

~ s

say

'e a

words

.m

luch

The

~ it

llgh­

fer-

the

are

) dia-

a rea­

:om

iding

ing

SOUND SYSTEMS IN CONFLICT: A HIERARCHY

OF DIFFICUL TY

2

In attempting to arrive at a reasonable hierarchy of difficulty, we must take

into account informatipn from what psychologists have developed as LEARNING THEORy.1

There are no doubt many aspects of learning theory from which we might benefit, but one

concept in particular seems promising: the notion of transfer-negative transfer, positive

transfer, and zero transfer. A student may_b..aY.~so.me habitual respo.nses which are con- .

tra_r.)T to the responses required for a new skill which he is trying to master (negative)_or

which are similar to the new responses (positive), or which have no relation to them (zero). ~.... --...- .. ~

Th\s notion of transfer is applicable throughout the structure of the language: the sound

sysb, the grammar, the vocabulary. Let us use illustrations of transfer based on the re-

lation of pron iation to spelling.

Suppos , for instance, that a student is trying to learn to pronounce Spanish by

using Spanish orthogra hy as a guide. He sees the word .Habana, spelled (as in English) h- is "silent" (Le. , represents no phonological reality) in Span­

ish orthography. The stude 's literacy habits have conditioned him to produce the initial

sound of have, hold, her, him hen he sees !!-. These are the conditions of negative trans­

fer-a familiar response to a f~niliar stimulus is carried over where a new response to

the stimulus is wanted. The effe~ of the old response is negative: he pronounces the Span­

ish word with an!!.-. On the other hand, to continue with orthographically conditioned trans­

fers, the existence of ch in both Spa!lish and English orthographies with approximately the

same sound values is a condition for, positive transfer the familiar ch of church carries

over ~o Chile, leche, lechuga with PO~ftiv_e effecy F inally, the sy~.TI might lead to zel~ transfer-but in fac t, s ince the s tudent, is familiar with n but not wi th ii, he often ignores

the tilde and hence encounter s negativ~ , transfer . An un~guable inst~ce of zero tr ansfer

for the reader of English does not exist \ n Spanish orthography; we must look instead to a

symbol system like those of Korean or Chinese to find true instances of zero transfer for him.

1. For a convenient summary of learning theory as relevant to linguistics, see J ames J. J enkins, " The Learning Theory Approach, " in P sycholinguisti.c s: A Survr:; of Theory and Research, ed. Charles E. Osgood (Ind. Univ. P ubs . in Antlll'opology and in­guistics, Memoir 10), pp. 20-35, Baltimore, ,1954.

\ 9

10 / SOUNDS OF ENGLISH AND SPANISH

The conditions of negative, positive, and zero transfer by themselves would

enable us to set up a reasonable hierarchy of difficulty. We could safely assume that in-. - - -- "-

stances where, c~?d~tions for positive transfer existed would lend themselves to n:ast~:y

more readily than instances where conditions for negative or zero transfer existed. It is --~ - .~

probable that we should have somewhat more difficulty determining whether the instances

of negative transfer were more difficult than those of zero transfer: does the student have

more trouble mastering gender concord in Spanish (el muchacho mejicano, but la mucha­

cha mejicana), an instance of zero transfer, or with por /para, where the phonetic similar­

ity of por and English for seems to set up an instance of negative transfer? Indeed, we

would have no little difficulty deciding exactly which instances involved negative transfer

and which ones zero: it is not at all clear, for example, whether ser/estar is difficult be­

cause of negative transfer from is to es, or because of zero transfer from lack of distinc­

tion between such verbs in English to presence of it in Spanish, or because of both factors

together.

It seems that we may get around the difficulties inherent in the question of

types of transfer by focusing our attention on the kinds of choices that exist at any given

point in the two languages. We have already seen that th~:c.QnunciaJien-of-cH-aR!Flag~ __ l!!~Y~

be characterized as a set of chOices, plus obligatory consequences, or, as we might say,

Qptional choices and obllgatory chOicig-l We can add to these a third set; zero choices":--.-----~ .......... ---

those which exist in one language but not at all in the other. An example is the phoneme -----~ "'-~ - -- -. -.---.-~

/z/ -the middle consonant of pleasure-which exists in English but not in Spanis~: We can

now set up the follOWIng three'-way correspondences between English and Spanish. (QE optional, Ob obligatory, ~ zero).

There are eight possible Situations, not counting the theoretical ninth pOSSi­

bility of zero choice in both languages:

1.

2. 3. 4. 5. 6. 7. 8.

English choice

Op

Ob

~ Op

Ob

~

Op

Ob

Spanish choice

Op

Op

Op

Ob

Ob

Ob

~

~

In this method of comparison of sound systems, "optional choice" refers to

the possible selection among phonemes. For example, the English speaker may begin a

word with / p/ or with / b/. "Obligatory chOice" refers, for one thing, to the selection of

conditioned allophones. For example, when the English speaker has /p/ at the beginning

of a word, the structure of the language requires the aspirated allophone (ph] in that en­

vironment. Also, "obligatory choice" refers to limitations in distribution of phonemes.

For example, before / m/ at the beginning of a word, English has only lsi, never /z/. The

term "zero choice," which is meaningful only when two languages are being compared, re-

fe

th

J,lf

ni

Tt

su -fu]

a I

trc thi

lin

In

Mi

Th

of

cal

me

.P.!E cle

Spc

ogj phc the uttc mu ic , lex wis he eitl the ~ con in t pos WO! of c mir Ph.o

"J is

:es ,tVe

a­lar-

er )e­

.nc­

ors

m

~f.l:Y~

.y,

e

can

i-

o

a ~f

ng

'n-

The

I, re-

SOUND SYSTEMS IN CONFLICT I 11

fers to the existence of a certain sound in one language which has no counterpart at all in

the other. Let us see what sort of examples might exist for each type.2

1. English 2£, Spanish 2£. U,oth Janguag s allow certain consonants to ap­,Jlear at the beginning of a word before a vowel. There are words like me, knee, tea; mC

ni, tl; and others. We can symbolize this fact in a general way:

{English} C +­

Spanish

Iml Inl It I in env. -V

That is, initially before a vowel, English and SRanish share the possibility of choosing such consonants as 1m, n, t/. Although this description is obviously incomplete, sin?e the - full list of possible consonants is not specified, the mere fact that the two languages share

a specifiable list of pre-vocalic consonantal possibilities is a huge source of positive

transfer. One can barely imagine how much more difficult Spanish would be to teach if

this set of choices were not held in common .

2. English Ob, Spanish 2£. Examples for this comparison are scarce. If we

limit our coverage of English to a particular dialect, however, an example can be found.

In the dialect that is sometimes called southwest midland (Oklahoma, Arkansas, southern

Missouri, southern Kansas b northwest Texas), the vowels of pin and pen are identical.

That is, speakers of this dialect have no choice between /II and lei before In/. They can

of course choose other vowels, like those of pat, pot, bought, beat, but the only vowel they

can choose in the area of lei and /II is a vowel which is really neither one of these but more or less midway between. It is a well-known joke that they can distinguish between

pin and pen only by specifying a "stickin' pin" or a "writin' pin." For these speakers it is

clear that there is no choice between 1['1 and /II in the environment: -Q' Faced with a Spanish item like lento, the conditions of negative transfer exist for them: they will regu-

2. In the discussion of optional vs. obligatory choices on the level of phonol­ogy, we are reversing a familiar use of these terms. There is a sense in which nearly all phonological choices are obligatory: if one has in mind, so to speak, a string of words, then the distinctive phonological shape of each word is obligatory; that is, if one wishes to utter the word in English which has the meaning "one J>lus one" or "four minus two," he must say something which can be written phonetically [tvw I, or (in a more detailed phonet­ic writing) [thv-ul. In other words, the physical shape of the word two is established by lexical ~'ule-any other sequence of sounds will presumably be some other word. But if he wishes to say the word which means "dealing with monetary problems" (i.e., economic), he may say either [tk+mim+yk l or liyk+nam+ykJ-that is, the first syllable may rhyme either with Tech or with teak. This is free variation on a speCifiable level of analysis-the choice may be considered entirely optional. This usage of the terms optional vs. ob­ligatory is pOSSible-and quite proper-if, and only if, the phonology is viewed within tIle contexfof a complete set of rules of sentence formation (Le., a complete grammar). But in the present situation, where only phonology is under conSideration, and where the pur­pose is to specify the possibilities of combining sounds to produce words-where the words cannot be said to have been selected before the phonological rules-then the usage of optional and obligatory must be ~'eversed, because the matters that are optional are the minimally significant phonological elements, and the matters that are obligatory are the phonetic consequences of choosing one or another combination of these minimal elements.

12 I SOUNDS OF ENGLISH AND SPANISH

larly produce the only vowel their dialect allows in the general phonetic area of I f,/ or

III, and it is not very similar to the correct vowel. 3

3. English~, Spanish QQ. This correspondence chara~erizes the classic dif­

ficulty the English speaker has with the erre of Spanish perro, or t11'\ jota of Spanish hijo.

In neither instance does the sound exist in English, although both sou,dS represent option­

al chOices of considerable frequency in Spanish. From the English SPjaker'S point of view,

they are new sounds.

4. English QQ, Spanish Ob. This correspondence characterizes one of the

more difficult problems of Spanish phonology for the English learner. Take, for example,

the pronunciation of items like dado and de do in isolation. The ~ at the beginning is pro­

nounced differently from the ~ in the middle. The initial ~ is much like the initial ~ of Eng­

lish den, doll, door. (It is not exactly the same, but the difference is irrelevant for this

purpose.) We will write it with the phonetic symbol [d]. The middle ~ of dado, dedo, on the

other hand, is conspicuously different-to the English ear-from the initial~. It sounds

more nearly like the initial th of then, there, !hose. We will write it with the phonetic sym­

bol [a]. Dado and dedo can now be written phonetically as [daao], [deao]. EQr the Spanish

speaker, the pronunciation of lat rather than Cd], in the middle of these words j§ Oblig~ .i2.r y. He will not ordinarily even be aware that he pronounces two quite different sounds

for the ~'s of dado and dedo. To use the technical terminology introduced earlier, (d] and

191 are allophones of a single phoneme I dl in Spanish. Among the consonants of Spanish";­

Idl exists as one possible optional choice, which may be symbolized:

Spanish C __

Ipl It! Ikl Ibl Idl Igl

inenv. -V

There is then a subSidiary rule about / dl (illustrated, incompletely, below):

Idl --Cd] in env.

[a] in env. V

That is, if Idl is preceded by silence (a break in utterance continuity symbolized in the

form,ula by #) or an Inl or Ill, it is pronounced as [d). If it is preceded by a vowel, it is

pronounced as [a]. The phonetic difference between Cd] and [a] is CONDITIONED by this

? . 3. Of the writers, H.P .S. has been plagued by this .£lis ability' since his first

exposure to Spanish. He can testify to its persistence. Even as a trained phonetician, he can avoid the obligatory vowel of his native dialect only with concentrated effort.

rule

ally,

Id/. ~ is pl

pOSE

In E

~ curr

this

will

choi,

ther

~, Qutt,

ed fc

[IJ]. :

orde

Hcul ([m and j

roof

It is

dif­

UQ. ion­

"iew,

lIe, )-

Eng­

s l the

sym­

lish -.... ga--ds and -sh,

the

it is this

rst , he

SOUND SYSTEMS IN CONFLICT / 13

rule-a rule which merely describes a set of conditions to which Spanish speakers habitu­ally, and unconsciously, conform. Because of this rule, [Ei] is for them simply a kind of

/d/. But for the EngLi.slLspeaker the cQnditions are different. For him ld] and laj are IN ~ONTRAST that is, they belong to differept phonemes Id/ and /Ei/. The fact of contrast

is proved by pairs such as dine/thine, dare/there, dough/though. [d] and [Ei] exist as two

possible choices among the consonants of English:

English C +-

/p/ It! /k/ /b/ /d/ /g/ /v/ /Ei/

in env. -V

In English, unlike Spanish, /d/ and I.!U are in contrast: they are both optional chOices, and -their distribution cannot be predicted. Predictability is at the heart of the matter: the oc-

currence of Spanish ld] and LEi] can be predicted by writing merely one symbol, /d/; given

this symbol in an environment, it is possible always and infallibly to predict whether it

will be pronounced [d] or (Ei]. The difference between them is obligatory.

This correspondence between English optional choices and Spanish Obligator*

choices is so important in its consequences that another example may clarify it still fur­

ther. Suppose we consider the possibilities of nasal consonants ([m] as in ham, [n] as in

hen, [IJ] as in hang) in the environment of following stop consonants ([p] as in~, (t] as in

putt, [k] as in puck, [b] as in tub, [d] as in dud, [g] as in dug). The phonetic symbols need­

ed for this discussion are all familiar letters of the alphabet in familiar values, except for

[IJ]· Note that the letters gg are used to spell both /IJ/ and /IJg/ in English: words like Sing­

er and banging have /IJ/, whereas words like finger and younger have /IJg/. Certain articulatory facts about these consonants must be briefly explained in

order to make the point clear. In terms of the place in the mouth at which the sound is ar­

ticulated, the nasal and stop consonants fall into three classes: those made at the lips

([m p b]); those made by the tip of the tongue at or just behind the upper teeth ([n t d]);

and those made toward the back of the mouth, with the to'ngue touching the back part of the

roof of the mouth (the VELUM) «(IJ kg]).

Lips Teeth Velum

m n IJ p t k

b d g

It is characteristic of Spanish that in a sequence of nasal consonant plus stop consonant,

•

14 / SOUNDS OF ENGLISH AND SPANISH

-=-

the point of articulation of BOTH consonants is fixed by the stop consonant. This can be formulated:

. {/p/} [m] in env. -/b/

Spanish N [n] in env. _ {/t/ } /d/

[IJ] in env. _ { /k/ } /g/

That is, a nasal (N) can be only [m] if the following consonant is [p] or [b], only [n] if the

following consonant is [t] or [d], only [IJ] if the following consonant is [k] or [g]. This re­

striction remains valid regardless of word boundaries and spelling: hombre, un beso; en­

dosar, un dfa; ingles, un gato. It is 0 tional whether a nasal be chosen at all; but if one is

chosen, it i.s. .o Q!if@.toJ;~..th _ it o'nt of articulatio be the same as top

C:~l!.~Qp.ant. In English, on the other hand, no such restriction exists: [mb] lumber, [nb] un-

bend, [IJb] kingbird; [md] lambda, [nd] under, llJd] kingdom ; [mg] Baumgardner, [ng] in-

grown, [IJg] finger. In English, not only is the choice of a nasal consonant optional. as i?

Spanish, but so is the choice of a particular nasal, regs,rdless of the [ollo..wing.stQp cons.2..­nantz which is not true in Spanisb,.

5. English Ob, Spanish Ob. It is here that we get maximum positive transfer.

Any English pattenL1:haUs Obligatory is neces sarily one to which the speaker Jiives no

thought-it is an area where he has no choice. If the same pattern is obligatory als o i~ Spanish, there should be no problem-indeed, there will not normally even be any aware­

ness that there might have been a problem. These instances are more frequent than we

realize: comparison between Japanese and Spanish, on the one hand, and between English

and Spanish, on the other, will reveal that the English speaker is not so bad off for Span­ish-like habits as we who are faced with the student's errors are prone to think. To take

a simple instance: given the consonantal sequence /s/ plus /w/, both languages require

that a vowel be chosen in the next position-swear, suerte. This is not a trivial observa­

tion: if the consonantal sequence is / p/ followed by / r / , English requires a. vowel, as in

~, but Spanish allows / y/ or / w/ , for example, prieto / pryeto/ , pruebo /prwebo/ .4

Thus the fact that the Spanish speaker has a different range of choice after / pr/ consti­

tutes a problem for\~he English speaker, even though the sequence / pr/ itself does not.

6. English~, Spanish Ob. This correspondence is the extreme of the scale.

In English, a given habit does not exist at all; in Spanish, it is obligatory and hence nor­

mally outside the speaker's conscious control-it is a habit which he internalized at an

early age and has given no thought to since. Zero may be viewed as a kind of negative ob­

ligatien: to say that a pattern is zero is about the same as saying that it is obligatory that the speaker not conform to the pattern . We have, as it were, an absolute negative restrfC:

4. The semivowels /y/ and /w/ are definitely different from ordinary vowels (even though they are spelled with the same letters), because they are not syllabic.

tiOI fou

wit phc

waJ

noL

bet'

a s i con

Tha

fere all I

gogi

redi

SOUl

in g'

that [ee ], wore prot

to e~

be d

.l::~ ter, ---. ~ Engl fully

in ite

of / t main does

whic. tran~

Son (

We c :

ARC]

.e

is :top

1 ,

er.

e-

sh

n-

ce

n

e.

ob­

:hat -ric-

leIs

SOUND SYSTEMS IN CONFLICT / 15

tion in the one instance, an absolute positive restriction in the other. An example is to be

found in the middle consonant of Spanish words like haba, leva, avance. Although spelled

with ~ or ~, this sound is different from anything represented by ~ or ~ in English. The

phonetic symbol we will use for it is [1:3]. It is articulated by bringing the lower lip up to­

ward the upper lip, as if for ~, but without touching, so that the air produces a friction

noise, as if for ~. In Spanish, the difference between [b] and [1:3] is closely parallel'to that

between [d] and [a]. TIle two sounds are a llophones of a single phoneme1 predictable from

a single symbol in the following way (this formulation of the rule is illustrative only, not complete):

Spanish /b/ -+- {[b] in env. [ 1:1 J - } [1:3] in env. V -

That is, [1:3] normally occurs after vowels, [b] elsewhere. The situation of [b]-[I:3] is dif­

ferent from that of [d]-[a] in only one Significant respect: [1:3] does not exist in English at

all (a zero category), but [a] does (an optional category). But this is a big difference peda­

gogically. IllJ:he instance of [a], the English speaker must transfer a familiar sound and

redistribute it.l.\litb 'r;espect to other sounds ; in the instance of ll:3 t he must learn a new

sound as well a.s....a new dj stribution.

7. English.9£, Spanish~. This particular correspondence is a frequent one

in going from English to Spanish pronunciation. English has several vowels, for instance,

that are entirely lacking in Spanish. The vowel of American English grass, symbolized by

[re], does not exist in Spanish. Partly because of negative transfer from the spelling ~,

words like gracias are often pronounced with this vowel in early stages of learning. The

problem is merely to reduce the range of choice that the English speaker is accustomed

to exercising.

8. English Ob, Spanish~. An English Obligatory pattern of pronunciation can

be difficult to get rid of. For instance, it j§ ghligatory in most English dialects that items

with t or d between syllables, where the first syllable is stressed butter, shudder s lat­

~ ~,~), have an allophone of /t/ (or / d ) that is rather like the S anish r of ara

pero. It IS a VOIce 'oogue-tip FLAP. Faced with Spanish words like foto, beta, pita, the

English speaker of most dialects will produce the Obligatory English flap rather than the

fully articulated /t/ of Spanish. Another example also involves allophones of English /t/:

in items like mountain, button, latent, the English speaker of most dialects has a variety

of / t/ for which instead of dropping the tongue tip as he usually does to release a /t!, he

maintains the tongue tip in the same position for the following /n/. Such an articulation

does not exist under any conditions in any dialect of Spanish. Words like quitari, meten,

which always have a normally released /t! and a full vowel, are subject to this kind of

transfer. , Having at least an idea, now, of the eight kinds of differences that a compari­

son can reveal when it is based on the different possibilities of choice in the two languages,

we can attempt to rearrange the comparisons in an order which will constitute a HIER­

ARCHY OF DIFFICULTY. We must know which kinds of differences will be most difficult

16 ! SOUNDS OJ:<' E.N(jLl~H AND i::i!"Al'U;:,n

to master and which will be easiest, in order to grade our teaching materials, arrange

them into an effective sequence, and determine how much drill is needed on each point.

The hierarchy suggested below is by no means final; further experience with it may well

result in readjustments in the relative position of one category of difficulty or another.

Difficulty Comparison Examples (froin preceding discussion) Type

Magnitude Order Eng. Span.

1 tl Ob 6 [13] .-

I 2 tl Op 3 erre, jota

3 Op Ob 4 [d]/[ 6]

4 Ob Op 2 i/e before n

II 5 Ob tl 8 flap It I between vowels

6 Op ~ 7 [ad

7 Op Op 1 List of prevocalic consonants III

8 Ob Ob 5 sw-plus vowel

Given such a hierarchy, we must examine several other criteria that will en­

ter into the grading and sequencing of materials designed to eliminate these difficulties.

The most important of these is FIINCTWNA J I QAP-that is, the extent to

}Vhich a given sound is used in Spanis to distinguish one word from another , the quantity

of distinctive information that it carries. The Spanish ii. belongs in Group I in the hier­

archy qf djW c IJJ q,d~ in English, Optional in Spanish). But its functional load is almost

zero. There are about a dozen words in which fi carries the burden of contrast with the cl:ster [ny] (spelled -ni -): ufi6n (big toenail) ;s. uni6n and the like. 5 An American can

speak Spanish for a long time without ever needing this contrast. For!! he can substitute

[ny], modifying a cluster he controls from his English habits only to the extent of being

careful not to make the syllable division between [n] and [y]-that is, he must say [u.nyon]

rather than [un.y6n]. The ~ would, therefore, in spite of its relatively high rank in the hi­

erarchy of difficulty, be placed very late (indeed, almost last) in a reasonable pedagogical

hierarchy.

A less important additional criterion is liJQTENTIAL MISHEARING. Spanish

initial [t=]-the variety of It I that appears before vowels-provides an example. This

sound is very difficult, Group I (tl in English, Obligatory in Spanish), in our hierarchy

above. But failure to produce it correctly (with the tongue tip against the back side of the

uppe~ teeth, without a puff of air) will rarely cause misunderstanding. However, the Amer-

5. The word uni6n is to be transcribed as [uny6nl. In the view of other ana­lysts, it is better transcribed as [uni6nl, which would alter the example. The comparison to be made with English still stands, because in the English pattern the syllable division would obligatorily be [un. yon],

lca

hea ly j

ped

tiOI

tell

gua

Ibl ThE

but

cau

que

ly n

genl

eith

the)

SOUl

the

We 1

iari1

!II

en-

IS.

I -:ity

e

an

ute

Ig

y6n]

hi­

~ical

sh

the

~mer-

La­tson ion

SOUND SYSTEMS IN CONFLICT I 17

ican who is listening rather than speaking-receiving rather than producing-will often

hear a Spanish initial [C'] as being a [d). One good way for him to learn to hear it correct­

ly is for him to produce it correctly. We would therefore place the [t=] fairly high in a

pedagogically oriented sequence even though when evaluated as to its effect on the produc­

tion of Spanish, it will only add American accent to the student's pronunciation-not unin­telligibility at any point.

The final additional criterion is PATTERN CONGRUITY. The sounds of a lan­

guage pattern lhemselves in groups or sets. In Spanish, Ib/, Id/, and Igl constitute a set.

Ibl and Idl are high in difficulty, in functional load, and in potentiality for mishearing.

There is no doubt they must appear early in a pedagogical sequence. Igl is also difficult,

but it is considerably lower in functional load and has less potential for mishearing. Be­

cause it patterns like Ibl and I d/, we feel it would be incongruous to place it out of se­

quence with them even though it does not constitute a problem of the same order.

These, then, are the criteria which havl( determjned the sequence of our pres-

..entation:

1. Hierarchy of difficulty

2. Functional load

3. Potential mishearing

4. Pattern congruity

Matching these criteria against one another is no easy task, and there is clear­

ly no single "right" or "best" sequence of presentation. Our own procedure has been, in

general, to put those things first that were most important in the tas~of communication,

either because mishandling of them could easily result in misunderstanding or because

they carried a heavy functional load and would therefore be especially obvious and frequent

sources of accent. In order to get similar problems together, however, we have violated

the mixed criteria of importance. Our preferred pedagogical sequence is:

1. Basic intonation features and patterns

(including stress, pitch, juncture, and rhythm)

2. Weak stressed vowels

3. Strong stressed vowels and diphthongs

4. Voiced stop-spirants

5. Vibrants and liquids

6. Voiceless stops

7. Spirants

8. Nasals and palatals

9. Semivowels

10. Consonant clusters

11. Other intonation features and patterns.

The above order is not identical with the order of presentation in this volume.

We have arbitrarily followed intonation with consonants, on the assumption that the famil­

iarity of teachers with the problems of the vowels might justify our leaving the vowels un-

til after the matters that showed the more interesting details of dialectal variation had been described. The above order is, however, identical with that adopted for our text on pronunciation. 6

6. J. Donald Bowen and Robert P. Stockwell, Patterns of Spanish Pronuncia­tion (Chicago: University of Chicago Press, 1960).

\

•

ing ho' tion of

nals U --... iUy di: -­native --- -It is Cl

in ana]

low thE

these c smalle tonatio

PITCH ---which -:

of Silel

parts a

pauses

vOwel c bounda:

ciable 1

We hav4

--aSSocia these e: consan' ditianal abOve tl not Wril that We strong I

A15015 795295

THE~OUNDS OF

©J965

Robert P. Stockw!l1 ~ ......

and J. Donald Bowen

pc 1.f13!J

1~7' Ct~

ill THE UNIVERSITY OF CHICAGO PRESS

~ CHICAGO AND LONDON