A COURSE IN

MODERN
LINGUISTICS

CHARLES F. HOCKE1T
PROPBIIOil OP LINOUII'I'ICI AND ANTBR.OPOLOOY
CORNBLL UNI'YBIWTY

• OXFORD & IBH PUBLISHING CO.

~ . , ~MI lllltriiHtp Clllalllo
A COURSE IN MODERN LINGUISTICS by Charles F. Hockett

1958 The Macmillan Compony

PubHshed by Mohan Prim/ani, Oxford ciiBH Pt~blishln6 Co.,

66 J1111path, New Delhi 110001 and printed at
Clranrt111 0 lfset Printers, Ne111 Delhi-110002 5-J .f
To S.O.H
 PREFACE
This book il intcaded Cor thole coUeae lludaata wilD tab aa iDtro-
ductory counc in linpiaticl. U othen &Del iDterat ar aatertaiDau:at iD
the WOI'k, the author will be delipted; but it Ia aot a 11papularbadoD,"
and the general reader muat in aU fairDCII be wuaed ~this. Simplicity
of praentatiou Jw been ICJUiht, but DOt a falle almplificetioD ~IUJiect
matter.
The duty ol the writer of a textbook Ia aat CD explan: hlaticn ar
indulge ia flights of faacy, but to praeat, Ia u arclerly a way u be caa,
the generally accepted f'ac:ta aDd priociples ol the lelcl. Tbia bas beea
my aim; the tenor of the book il co~UerYative. N~ Cor tbil
I must •pologize oa 10111e topic:l my eathusium ud iavolwmeat bPe
certainly led me to apeak IDOI"C emphatically dlaa oar currcat kaowledp:
warrants.
Terminolagical inaovatioal have been avoided u IIIUCb u pallible.
Complete avoida.ace bas been uaattaiDable, becawe it II -..tial to
dilcuu aU Upectl of the field in a COJIIilteDt laiDiaolalr1 aacl DO CXIID-
plcte and CODiilteat tenJiiaolocy bu cxilted.
Altbourh I have intcaded DD aclhercace to aay ...._.., "tchoor'
ofliqpiltic:l, the iafluaace ~ AmericaD liapllticl, aacl apedally that
~Leonard BloamfieJd, wUl be apparent oa every paae.
Liapistb II 100 rich a field for adequate coverap ~ aU tGpicl
in an elemeotary coune. The deciliaa ~what to ladude aad
what to Olllit, ~. n:1t1 pmperly with the lulructar. I have. there-
fore, tried to iaclude adequate elementary uatmeat ~aU topicl ._
two: the hiatary of liapisdc:l, .... che detailed NiftY~ dae .........
cl tbe wurld. Tbe amieeion relect1 my owa apinjoa daat .ather il a
¥ii
viii PREFACE

desirable topic in an elementary coun:e; the colleague who disagrees

has access elsewhere to several first-class discussions of each.
I owe a great debt to a number of my colleagues who offered me
advice on one or another portion of the book, or who read an earlier
version in its entirety. This earlier version was used for two successive
yean in our introductory cnurse at Comcll University, and the reac-
tions of the students have been in\•aluable to me. Of my colleagues,
I must especially mention Frederick B. Agard, Harold B. Allen, J
Milton Cowan (who taught the Cornell course during the trial runs),
Gordon H. Fairbanks, Murray Fowler, Robert A. Hall, Jr., Eric P.
Hamp, Sumner Ives, Norman A. McQuown, William G. Moulton,
W. Freeman Twaddell. Oscar Cargill and Norman E. Eliason were
_especially helpfu.l during later stages of the work. Any deficiencies
remaining in the book are due to my own obstinacy, not to any inad-
equacy in the scholars just named. I wish also to offer my sincere
thanks to the Rockefeller Foundation, for grants with which the
writing was begun; to the Center for Advanced Study in the Behav-
ioral Sciences, where the writing was completed; and, above all, to
Cornell University, which, with a magic seemingly unique, makCII
itself a congenial home for the scholar in linguistics.

CHARLES F. HocuTT
Ithaca, New York
 CONTENTS

1. INTRODUCTION 1
SIGNALliNG VIA SOUND: PHONOLOGY
2. PHONEMES 15
3. PHONEMIC NOTATION 27
4. ENGLISH INTONATION 33
5. ENGLISH ACCENT 47
6. ENGLISH JUNCTURE 54
7. PHONETICS 62
8. CONTOID ARTICULATIONS 69
9. VOCOID ARTICULATIONS; TIMING AND COORDINATION 77
10. PHONEMIC ARRANGEMENTS; REDUNDANCY 84
11. TYPES OF PHONEMIC SYSTEMS 92
12. PHONEMIC ANALYSIS 102
13. PHONEMES AND SOUND 112
PHONOLOGY AND GRAMMAR: LEVELS
OF PATTERNING
14. MORPHEMES 123
15. MORPHEMES AND PHONEMES 130
16. THE DESIGN OF A LANGUAGE 137
GRAMMATICAL SYSTEMS
17. IMMEDIATE CONSTITUENTS 147
18. FORM CLASSES AND CONSTRUCTIONS 157
ix
 CONTBNTI

19. WOitDI 166

2/J. MORPHOLOGY AND SYNTAX 177
21. SYNTACTICAL CONSTRUCTION-TYPES: ENDOCENTJUC 183
.?2. SYNTACTICAL CONSTRUCTION-TYPES: EXOCENTRIC . 191
23. SENTENCES AND CLAUSES 199
24. INPLECTION 209
25. KINDS OF SYNTACTICAL LINKAGE 214
26. PARTS OP SPEECH 221
Z7. GRAJoDIATICAL CATEGORIES 230
7.8. DEIUVATION 240
29. SURFACE AND DEEP GRAMMAR. 246
30. stJBSTlTUTES 253
31. THE GRAioDIATICAL CORE 261
MORPHOPHONEMIC SYSTEMS
32. MORPBOPBONEMICS 271
33. TYPES OP ALTERNATION 277
34. CANONICAL FORMS AND ECONOMY 284
35. SECONDARY EPPECTS OP PHONEMIC SHAPES 2~3

IDIOMS
36. miOM POIUIATION 303
31. TYP:a OF mrous 310
SYNCHRONIC DIALECTOLOGY
J8. miOLEC'r, DJALEcr, LANGUAGE 321
?JJ. OOIOION CORE AND OVERALL PATI'ERN 331
4J. AIIEJUCAN ENOLJSB STR.ESSED IYLLABICS 339
UNGUISTIC ONTOGENY
41. LINOUISTIC ONTOGENY 353
PHYLOGENY
42. PHYLOOBNETIC CHANGE 365
43. OLD AND IODDLE ENOLJSB 372
44. BINDS OP PHYLOOENE11C CHANG& 380
CONT&JfTI xi
45. DCIIANJIIIS OP PIM.OOBNETIC CHANGE 387
46. INNOVATION AND SURVIVAL 393
41. TID OONDmONS FOil BOIUt.OWINO 402
48. KINDS OP LOANS 408
49. ADAPTATION AND IMPAar 417
!LJ. ANALOGICAL CREATION 425
51. PORTliER VARD'l"DS OP ANALOGY 432
52. THE NA~ OP SOUND CHANGE 439
53. COALESCENCE AND SPlJT 446
54. TID OONSEQU.Dcr.s Of SOUND CHANGE 452
liNGUISTIC PREHISTORY
55. INTERNAL REOOMBTRUCTION 461
56. DIALBCT GEOGRAPHY 471
5/. TID OOUPARATIVB IIE1HoD 485
58. REOONS'I'IlUCTJNG PHONJUIICS 493
59. RBOONS'I'IlUCTJNQ :U:oa.PROPRONDOQI AND
GRAJOIAR 505
fJO. PORTliER RISUL11 OP TID OOIIPARATIVB IIBTHOD 512
67. GLOTI'OCRRONOLOGY 526
WRITING
62. WRITING 539
UTEIM..TURE
tJ3. Ll'I'DATURB 553
M~N'S PUCE IN NATURE
64. MAN'S PLACE IN NATURE 569
APPENDIX OP LANGUAGE NAIIII 587
BJBUOORAPHY 599
IND&X 607
A COURSE IN
MODERN LINGUISTICS
 1·

INTR.ODUGriON

1.1. This book ia about ,.,.,.,,, the m01t valuable lingle ~

of the human race.
Everyone, in every walk of life, ia conc:emed with language in a prac-
tical way, for we make uae of it in virtually everything we do. For the
DlOit part our uae of language ia so automatic and natural that we pay
no more attention to it than we do to our breathing or to the beating of
our hearts. But IOIDCtimea our attention ia drawn: we are struck by the
fact that othen do not speak quite as we do, or we obsezve a child
learning to talk, or we wonder whether one or another way of ..ying
drwntiDg something ia correct.
Beyond thia, many people have profCIIional need to know 10mething
aboul language-as oppOied to limply being able to use it. Here are
some examples:
(1) The speecb. correctioniat, since hia job ia to help people oven:ome
dif6culties or impediments in their use of language.
(2) The teacher of English composition, for a IOJDCWhat similar
reason.
(3) The foreign language ~·
(4) The literary artiat, who must know his medium and its capacities
just. u a painter must know pigments, brushes, and colon; the literary
critic for a similar reason.
(5) The psychologist, who knows that language ia one of the vital
factors differentiating human ~ from that of rats or apes.
(6) The andu"opologilt, botli becauae language is part of what he
calls "culture," and because in hia anthropological field work he ia
often co.i&onted by practical proble1111 of a linguistic sor.t.
1
 INTRODUCTION

(7) The missionary, who may have to learn some exceedingly alien
language, for which there are no ready-made primen or dictionaries-
learning it not just for the management of everyday affain, but well
enough to deliver sermons and make Bible translations.
(8) The historian, because his sources of information are documents;
that is, written records of past speech. ·
{9) The philosopher, particularly in dealing with such topics as
logic, semantics, and so-called "logical syntax."
{10) The communications engineer, part of whose business is to
transmit messages in spoken form {telephone, radio) or in written form
(t~!~graph, teletype) from one place to another.
·.For aU these people, and for othen who could be added to the list,
knowledge of the workings of language is a means to some end. For a
small group of spec..ialists, knowing about language is an end in itselC.
These specialists call themselves linguists, and the organized body of
information about language which their investigations produce is
called linguistics. ,..1
The relation1liiP between linguistics and the various other fields in
which some knowledge of language is useful is much like that between,
say,. pure chemistry and chemical.engineering. Suppose that an indus-
trial plant has been using a natural dye to color certain products.
Something happens to threaten the source of the dye or to increase its
coat prohibitively. It then becomes the task of the chemical engineer to
find an effective substitute which requires only easily available and
i-clatively inexpensive raw materials. In hia efforts to solve this problem,
he calls on aU sorts of known facts of pure chemistry, many of which
were discovered with no such application in view.
Similarly, suppose that an American oil company wishes to develop
an oil-field in a region where the prevalent language is one not ordi-
narily taught in American schools. At least some of the company's per-
sonnel must learn the language. There will be no ready·made stock of
experienced teachers for the purpose, as there are for such languages as
French and German. Nor can one simply hire an inhabitant of the
region to serve as a teacher, since native control of a language does not
in itself' imply conscious understanding of how the language w()rb, or
ability to teach it-any more than having cancet' automatically makes
one a specialist in cancer diagnosis and therapy. But there are linguists
who are skilled at finding out how a language works, at preparing
SOUR.CES OF DIFFICULTY 3
teaching-materials in it, and at IUpervising the tutorial work of native
speaken. In all o( this, such linguists draw on the results of pure lin-
guistic research.
or coune, this proper relationship between "pure" and ""apelied"
does not always work out smoothly. Sometimes th~ faced with a prac-
tical language problem do not bOther to consult the "pJia:" linl'!isJB.
Sometimes they ask for help, but get none. This is occasionculy because
the particular linguist is not interested, but more often because the
organized body of information which linguists have so far gathered has
nothing to contribute to the problem at hand. When this happens, the
"applied" people sometimes forge ahead on their own and find a
workable solution. Many a key contribution to linguistics has come
about in just this way, from fields as diverse as classical philology and
electriC'.al engineering. Anything which anyone discovers about lan-
guage is grist for the linguist's mill. It is his job to work every :1ew
discovery into his systematic account of language, so that those who
come later will not waste their time exploring territory that has already
been clearly mapped.
The above cortsiderations reveal one reason why, in this book, we
shall deal with language in the frame of reference and the termin"lo7y
of linguistics, rather than in those of anth!!Pology, phili!!9phy, P-SY-
chology, foreign language teaching, or the like. Only in this way can ~t"
be sure of serving the interests of all those readers who are, or may latC'r
become, specialists in one or another of these fields. If we were to
present, say, a "psychologized"' linguistics, we might serve the psycho-
logically trained re;sder somewhat better (though this is not C'crtain),
but we would be doing a comparable disservice to the anthropologist,
the communications engineer, the foreign language teacher, and so on.
· Another and more fundamental reason is that language deserves
autonomous treatment, The objective study of human langt•age J1.es
not achieve its validity merely through actual or potential "pract1eal"
applications. Anything which plays as omnipresent and essential a
role iA human life as does language merits as careful study as possible.
The more we can undentand itl workings, the better we shall under-
stand ounelves and our place in the universe.
1.1. Soarca of Difticulty. Linguistics ~ an inherently di~t
subject. but there are several points which often make trouble for the
beginner. In part, these are merely matters of terminology; in part,
 4 INTRODUCTION

however, they have to do with the difFerence between the lay attitud.e
tuwards language and the orientation of the specialist.
(1) The linguist dist~es between lanpag1 and wr.J!i"'• whereas
the layman tends to confuse the two. Theliyman's terms "s~en
language" and "written language" suggest that speech and writing
anr- merely two difl'emif"manifestations of something fundamentally
the same. Often enough, the layman thinks that writing is somehow
more basic than speech. Almost the reverse is true.
Human beings have been speaking for a very long time, perhaps
milliODS of years. Compared to this, writing is a recent invention. &
late as a century or so ago, millions of people in civilized countries
could not read or write-literacy was a prerogative of the privileged
. c:laaes. Even today, there are large numbers of illiterates in some parts
of the world. Yet there is no human community anywhere which does
DOt have a fully developed language. Stories of peasants whose vocabu-
lary is limited to a few hundred words, or of savages who speak only in
gnmta, are pure myth.
Similarly, the child learns to speak his language at an earlier age
than he learns to read and write, and acquires the latter skills in the
framework supplied by the former. This in itself is one of the reasons
why we tend to misunderstand the relationship between language and
writing. When we begin to learn to speak, the problems involved can
hardly be discussed with us, since the discussion would require the very
lkill we have set out to achieve. But when we begin to leam to read and
write, our teachers can talk with us about the task. Thus we grow up
with a vocabulary for saying things about reading and writing, but
with none for dealing with language itself. Of course the relationship
between writing and language is close; it is only natural that we
should transfer the vocabulary fitted to the discussion of writing to our
remarks about language. ·For example, we constantly talk about spoken
words (which can be heard but not seen) as though they were compoeed
u letters {marks on paper which can be seen but DOt heard).
nc cban,ge of .orientation which is required in this connection is not
a ceq one to make.. OJd habits die hard. Long after one hu learned
the IUitable technical vocabulary for discuaing language directly,
...._. than via writing, ODC ia scill apt to tlip. Jt lhDuld aft"ord IOIDC
COIIIOiation to know that it took liopiatic ICbolanhip a goad 11118DJ
buDdndl of ycara to make jult this IIIUQe traalition.
 S011RCES OF DIFFICULTY 5
(2} Much of the time devoted by the layman to lanpaae il taken
'IJP by the problem of "correctness." Is it more "correct" to my il is 1
. than ifs me? To wluma than who to? What renders aia't incorrect? Are
"incorrect" forma to be avoided under all circumstances?
It may come as a shock to learn that the linguist is not particularly
interested in such questions. This statement mwt not be misunderstood.
It does not mean that the linguist is an advocate of incorrect forma, or
that he denies the reality of the distinction between correct and incor-
rect. As a wer of language, the linguist is bound by the conventicma of
his society jwt as everyone else is-and is allowed the same degrea and
'kinds of freedoms within those conventions. In wing language, he may
be a purist or not. But this has litde if any relationship to his special
concern, which is analyzing language.
AB an analyst of language, the linguist is bound to observe and record
"incorrect" forms as well as "correct" ones-if the language with
which he is working makes such a distinction. A particular linguist may
become interested in the whole phenomGD.on of correctaess, and may
study this in the same objective way in which he might examine Greek
verbs, or French phonetics, or the child's acquisition of speech. If he
does, he may soon discover that he needs help. The sociologist or
anthropologist, for example, is better prepared than he to explain the
special secondary values attached to certain patterns of behavior, be
they ways of speaking or points of table etiquette.
(3) The organization of affairs i.n our schools is such as to suggest a
very close tie between language and literature. A high school English
course is apt to devote some time to grammar and some time to Tenny-
son. The typical college French department offers instruction in that
language, and in French literature, as well aa-more rarely--a lew
courses in phonetics, philology, or the like.
The tie between language and literature is natundly cbe-the
literary artist works in the medium of language juat u tbe painter
works in the medium of colors and the compoeer in that oliOUnds.
Nevertheless, the study of the two mwt not be coaf"uacd. A painter and
a chemist are both interested in pigments. The painter's intcreat ·
focwaes on effective selection and placement of· different colora aod
textures on his canvas. The chemist's interest is in the ~ cca-
position of the pigments, whether used in one way ar aaother br tile
painter. Some physicists are specialiatl ill IIOUDd; C¥eD wileD they deal
6 INTRODUCTION

with the ldadl mIOUadt produced by mUiical instrument~, their iallftlt

il very difFerent from that c4 the comp01er, the perf'ormer, or the IDUii-
cal audience. Similarly, the raw-materials of the literary artist are d
c:oncem to the liapiat, but he il conc:erned with them whether they
are Uled far literary purposes or otherwile.
Unlike the pbenomenon of correctness, literature is apparently UDi-
venal. Some 1ort c4 literature is found in alm01t every known human
IIOCiety, and itl study is proportionately important for an improved
undentanding of human nature.
(4) A number of facton conspire to give ua a Calae notion of the rela-
tionship between .language, or grammar, and logic. If we carry this
notion with ua into our study of linguistics, we are apt to expect some
~ultl which are. not attainable, and to miu the point of some ol the
raultl actually attained.
One of these facton is the common asaumption that any usage which
il not "logical,. is therefore wrong. To aay he drm't is "iUogical," fOI'
example, since drm't is a contraction of do rwt, and we do not aay lw do.
CJuch a comment reflects the fact that, in historical origin, the disci-
plines of grammar and of logic were cl01e. More sophisticated reflec..
tiona of this are sometimes to be found in the opinions o1 contemporary
phU010phen. One of these recendy criticized linguists for their in-
aiatence that, in a sentence like Jolu& SfiW Bill, only Jolin is the subject.
The critic wanted to aaert that both Jolm and Bill are subjects, since
the sentence la)'llsomething about both.
Again, we olten feel, u we study some language other thaD our own.
that ita wa)'ll are mOlt peculiar. What acnae ia there in the French habit
u saying J1 ,..... d6 r.;., 'I want aome water' with the definite article
before.., 'water,' but J1., MUt: pas tl'~a 'I don't want any water' with-
out the article?
There are really· two dift"erent points at iaaue here. Oae il tbe ateat
to whic:h we can expect a .laquase to be "lopcal" ill the-m "c:oa-
. . .t and lellli.We,'' aad the extent. to which Janpqea diB'er in tbia
nprd. Tbe otbet il w..etber the 1inguiat, in analyzing aad clacribias
Dlle particulm'-laapqe, lbould work in ta'IDI of IOIDC pniCL1IItAived
llotiaa til abltract Josie or lhould accept what he fiDda.
T.he &aiWel' to the 6nt point il that , known ....... lhoWI
certaia CGIIIiltenc:ia aad many arbitrary inconailtenclea. We do n~
• the arbitrary featurel of CU' utive J.anBuap, because we are used
LANOUAOEB AND SPEECH COKKUNJTIBB 7
to them. Thole of IOJDe other languqe, ltudied when we are adul11,
1taDd out like 101e .thumbs. We are quite right in doubtiag the 1en1e ol
the French habit mentioned above: though regular, it is entirely arbi-
trary. But we ahould not be right were we to conclude that French is
"Ia~· Iogical" than English. h it not, in the Jut analyail, perfecdy
•w:bitrary that we should say I liHiftl Stmrl U~GIIr with Stmrl, but should
switch to tD9 in making the statement negative, I ti.M"t wat ""1 _,.?
The answer to the second point is that linguistic raearch can accom-
plilb. nothing unless it is strictly inductive. Phll010phical speculation
about what language oug/zt to be is sterile. In dea:ribing a languap
we must report actual usage, as determined by observation. In describ-
ing speech behavior in general, we must be most concerned with thOIC
f'eatures which have been empirically discovered in all the languages OD
which we have any information.
Thus if we observe that ccnain speakers of English say I do, I dtm"l,
Ill d01s, and 116 don't, we can only conclude that in their particular
variety of English don't functions as the contraction of d01s not as well
81 of do not. (This does not render 116 drm"l "correct": in standing aa
ccc:orrect" or "incorrect" is here beside the point.) When we auert that
Jolm, and only that, is the subject of the sentence Jolta SdW Bill, we arc
not contradicting (nor confirming) wnat a logician may want to say
about this sentence. The linguistic use of the term "Nbjcct"' baa rela-
tively little to do with the logician's use of the same term; the linpilt
111e1 this term, and otherR, to deacribe how llelltenc:ea are put tcJtlether,
rather th~ to describe what sentenc:a arc about and whether 01' not
they arc true.
From the linguist's point of view, the ''J.osical"" approach to lanpap
il too ll8ri'OW. Lanpagc is not used just to make 8IICI1ioal of fact. It
is ua«;d for lies as welt as truth, for DODICDIC u well • for - - , for
pcnuuion u 'well u for instruction, for entertainmeat u well· • far
bwllinea, for making war 81 well u for maldag love. l.altplp ia u
braMI aDd deep u the whole fabric of human exiltence; our appaiiCh to
it must be comparably catholic.
t.S. Laapa1a .md Speech Camanuaiti-. The liapdlt"1 .._. of
ltUdy ia not just English or just the politically impartaDt ......... of
the world, bat: <:very languap abollt which we have, ar caa obaUn,
information.
l'he muaber of Jaaauqe~lpoken in the world today·- .... Cll'
8 INTRODUCTION

four ........ Precilely how IDUl)' we c:aDDOt •Y· ODe reuon ia Jack
of accun.te i.nformadoll on the 1aaguagea of certain regioaa, particulady
South America aad part1 of the Western Pacific. Another more fuada..
mental reuon ia that, ~ when our information is adequate, we can-
not alwayJ judge whether the speech of two groups lhouJd be counted
as separate languap or only as divergent dialects of a single languap:.
Each language defines a speecl& community: the whole set of people w~
communicate with each other, directly and indirectly, via the common
lanpage. The boundaries between speech communities arc not sharp.
'Ihen: are people, llilirrpals or polyglots, who have a practical command
of two or more languages and through whom members of different
speech communities can establish contact. Most polyglots belong pri-
aiarily to one speech commuqity, and have only partial control of any
other language, but there are occasional exceptions.
In many cases the boundaries of a speech community coincide with
political boundaries. Thus in aboriginal times the Menomini ~
wu spoken by all the members of the Menomini tribe, in what"ia now
DCJI"tbCm Wiacousin and Michigan, and by no other community. This
state of aft'ain held for many an American Indian tribe in earlier da.,..
aad is still to be encountered in 1nany parts of the world. But to tbia
aaaeralization, aJio, there are exceptions. Switzerland, a liagle political
unit, includes apeakera of four different languages: French, GenaaD.
Italian, and Lac:lin or Rhaeto-Romance. Contrariwise, English, a llin&le
Jansuage, is spoken not only in Britain and in many parts mthe Britilh
Commonwealth, but alao in the United States.
Some apeech communities of today are extremely large. English hu
leYeral hundred million native speakers, and millioDI with IODlC other
aadve language have learned English for business, profCIIioual, ar
political purpoaes. R.UIIiaa, French, Spaujlh, German. Wnesc, and a
few otherl also have vast numbcn of apcakcn. Some specialists My that
"Qtineae., is a pou.p of related 1anpages rather than a aingle lanpaae,
but if we break tbae up then at leut one of them, Maadarin a.;,..,
ICiJl belonp in the above list. In general, speech communitiea of such
~Mae propartioal bave come into exiltence only recently, 11 a I'CIUit ~
hiltGrical development~ in the put five hundred yean or 10.
At the appalite e&tleme l&lllda a 1aQ8Uaae Uke Cbldm•cba, •
Americua ,lndiaa Jaapqe which In the late 193()1a bid cmlr &wo
.,.._..left. Whea. JaDsuaF reachel such ........... It ........
LANGUAGE FAMILIES 9
-DO new generation will learn it, and when the old people die the
IanguagC is dead too. But no such prediction can be ventured if a
language has as many as even a few hundred speakers. New Guinea is
packed with lfi.llages of a few hundred inhabitants each, each village
or small group of villages with its own language, and all seemingly quite
viable. The same is true of vast regions of South America and Africa.
M01t languages of today have from a few hundred to a few tens of
thousands of speakers, and probably something of this sort has been tht"
rule throughout human history.
For a very small proportion of the languages of today, there are
written records which tell us something of what they were like in earlier
times. Thus we have documents in English from as early as the end of
the 7th century A.D. They do not look like English to us, but they are:
the language in which they were written has gradually changed, in the
intervening twelve hundred years, to become just the language we now
speak.
We also have written records attesting to the former existence of
languages which have now completely died out. Frou~ ancient Italy we
have numerous inscriptions, and a few documents, in several languages
besides Latin. Some ofthese, such as Oscan and Umbrian, were akin to
Latin, while others, like Etruscan, were not. All of them were swamped
by Latin as Rome rose to political supremacy-just as Chitimacha has
more recendy been swamped by English-and today only Latin
survives, in the form of the so-called Roman&e languages: French,
Portuguese, Spanish, Catalan, Italian, Rumanian, and a few otht"rs.
Some of these later forms have also become extinct. We l:now
that the last speaker of Dalmatian, a Romance language formerly
spoken in what is now Yugoslavia, was killed in a mine explosion in
1898.
l ..f. I.Dguage Familia. All languages are constantly undergoing
slight changes-in pronunciation, in grammar, in vocabulary-which,
in the course of a thousand years or so, have a tremendous cumulative
efFect. This is why the earliest written records of English are quite
unintellipble to us, as our speech would be to our linguistic ancestors of
a millenium ago if in some miraculous manner they could be exposed
to ~t. So long as the members of a speech community form a fairly
ticbt•knit group, any change tends to spread to all the speakers ol the
languase. But it the community is broken up, as by migration or by
 10 INTR.ODUCTION

invasion !rom oumide, then changes which begin in one subcommunity

uaually spread only within that subcommunity, and, as a result, the
speech habits of the different subcommunities dir"rge. If the divergence
becomes great enough, members of the different subcommunities no
J.cmaer understand each other, and we must then say that each sub-
community has become an independent spet'Ch community with its
own language. Whenever two or more languages have developed in thia
way from a single earlier language, we say that they are relaled. Like-
wise, we say that any group of clearly related languages constitutes a
/flllftily of languages. .
In the case of the Romance family, we are fortunate in having not
only direct documentary evidence for much of the history of each sepa•
rate language, but also ample written records of their common ancestor,
Latin. This is a· moat unusual state of affairs, recurring only for the
IDdo-Aryan languages of India (Bengali, Hindi, Bihari, Marathi, and
others) : Sanskrit, known to us through a vast body of literature, was a
standardized literary form of one dialect of the common ancestor. For
InUit other families of languages the earliest written records are much
more recent than the presumed common ancestor. Sometimes, by
methods which we shall describe later, positive pl"OOf of relationship
can be worked out even without the help of the direct evidence of
written records, but in many other casca we are forced to suspend
judpnent. Thus some languages, so far as we know, constitute families
all by themselves: for example, Basque. Similarly, it is clear that the
hunc:lred-and-fifty-odd aboriginal languages of America north of Mexico
can be grouped into about fifty families, within each of which the rela-
tionship ia incontestable, but proposals for the further grouping of these
fifty sets into a smaller number of larger families rest on more tenuous
evidence and have not yet met with general agreement.

NOTES

At the end of most sections in this book will be found a paragraph

01' 10 entitled "Notes." These notes always include a check-list of the
new terms which have been introduced; in some cues they contain, or
refer to, problems for the reader to work on; when relevant, lhey include
bibJiosraphical refcrenc:ea. The latter are given in the form of author
NOTES 11
and year of publication of the work ret'elred to; fuller bibliographical
information is presented alphabetically at the end of the book.
Insofar as possible, throughout the book, our examples will be drawn
from English. Where this is not possible, we draw most heavily on
German, French, Spanish, Latin, Chinc:se, and Menomini. These, and
all other languages mentioned anywhere in the discussion, are listed
alphabetically in the Appendix of Language-Names, togc-ther with
information as to where, when, and by whom each language is or was
spoken and the known or strongly suspected family affiliation of each.
The following terms receive somewhat sharpened meanings in §1:
lfJfllfuage (excludes writing), writing, linguist, linguistics. The following are
introduced as technical terms: spe~ch commu.nity, bilingual or polyglot,
divergence, related languages, famzly oi languages.
A brief but penetrating survey of the history of linguistics will be
found in Bloomfield 1933, chapter 1; also, his treatment of many topics
in chapters 18--27 incorporates historical discussion. More extensive is
Pedersen 1931. Hall 1951b surveyta developments in the United States
from 192t; to 1950. Carroll 1953 describes the present stat~ of develop-
ment of linguistics in the United States, in irs interrelationships with
other disciplines.
The best book-length worldwide survey of languages and language
families is Meillet and Cohen 1952. This treats a few areas lightly; for
Latin Americn, one can compare McQuown 1955. Other surveys of
value are Gray 1939 (especially for bibliography), Matthews 1951,
Hoijer 1946.
SIGNALLING
VIA SOUND:
PHONOLOGY
 2·

PHONEMES

2.1. Suppose you ask a grocer the price of eggs, and he says sixty
cents a dozen. How do you know that he has said this, rather than eighty
cents a dozen, or we luJw no eggs today, or something else?
The answer is obvious. The various things that someone might say
in a given situation (and a given language) sourul different. You tell
one utterance from another by ear, just as you recognize the faces of
your friends by sight. Of course, we are not infallible in either sort of
identification. You may mistake one friend for another if the light is
bad, and you may misunderstand what someone says if the surround-
ings are too noisy or if he mumbles his words. And just as you would
have particular trouble with a pair of identical twins, so it is with a
pair of English utterances like The sons raise meat and The sun's rays
mtet. In cases like these, context helps. You see Jean-or-Joan playing
tennis, and conclud~ that it is Jean because Joan doesn't like tennis.
You hear The sun's rays meet in the course of a lecture on optics, and
know perfectly well that this (rather than The stmS raise meal) is what
was said. Difficulties and exceptions of this sort are marginal; they do
not affect the general validity of our answer.
It follows that one subject which we must study if we wish to know
how language worb is sound-the IIOI1S of IOUDd used in speech, and
how they are produced and detected. This part of linguistics is called
plwrudogy or plumtimit:s.
Throughout the study or phonology, it must be remembered that
sounds and dift'erenees between them have one and only one function in
language: to luep tlltlr411Ces apart. This means that there is little to be
learned by examiaing the utterance~ of a language one by one, tryiag
15
 16 PHONEMES

10111ehow to describe the sound of each. It is much more to the point

to examine pairs of utterances to see how they dijfer in sound.
Now if we consider a pair such as sixty &mls a dozen and we luwe ,.,
eggs today, we find the difference so extensive that it is hard to pin any-
. thing down. At the other extreme, a pair like Tl~e sons raise meal a.nd
The sun's rays meet is of no help, si nee, by exception, this pair cannot be
kept apart by ear but only by context. Between these two extremes,
however, one can find pairs which might be called "almost-identical
:Wins." Here are some "almost-identical twins" that can be heard
around the Cornell University campus:

She's gone to Willard Straight.

She's gone to Willard Sto.te.

Wj!IIJ!d ~ig}d is the student union building on the Cornell campua;

Willard State (Hospital) is a mental institution not many miles away.
The difference in meaning is thus very gn~at. But the difference in sou~
is mil!lmil: the first Of the two has an r-sound in the last wont which is
Jacking in the other. (This is the only difference in sound between the
two. There are· other differences in spelling, but they do not parallel
anything in our pronunciation, and nluitnot mislead us.)
Or one of a pair of almost-identical utterances may have one charac-
teristic feature where the other has a different one:
/
":;:,JuJt's a ni&e pin.
TluJt' s a nice bin.

Pairs like these yield important information about the way a lan-
auase mak.a use of dift'erences of sound; that is, about its phtmologit:fll
system. The second pair tells us, for example, that in English we some-
times k~ utterances apart solely by having a ~d at a certain
point in one of them, a~ at the same point in the other. This is
not very exciting information, but it is significant just the same. There
are many languages in which the difference between a /J-IOUDd and a
b-sound is not used in this same functional way.' The .functional use of
the dift'erence in English is therefore a characteristic feature of the
language-a feature in which English differ. from certain other
languages.'
We want to see in 1110re detaU just how the phonolop:al system~ of
ENGLISH INITIAL CONTRASTS 17
various languages dift'er from each other, but we obviously cannot
explore this topic merely on the basis of the isolated information that,
in English, p-sounds and b-sounds stand in contrast. We r.1ust explore
the whole of ths,.!Wcnelesie S)'ltem of EurJjah.
In the two sets of almost-identical utterances so far considered, the
dift'erences in sound are located in specific words: straight venus state
and pin versus bin. Now almost any word in English (or in any other
language) is capable, at least on rare occasions, of occurring as a whole
utterance. For example, to the question Do you wemt it with sodll M'
slrlligkt? one might merely answer Strlligkt. This suggests that a con-
venient way to begin our exploration of the phonologic system of
English is to limit ounelves, temporarily, to one-word utterances. For
the present we shall impose an even stricter limitation, considering only
utterances of one syllable.
2.2. English Initial Cuntrasts. The pair of words pin and bin demon-
strate as well as the longer utterances Tkllt's ll nice pin and Tkllt's a nice
bin that the contrast between a p-sound and a b-sound is functional in
English. But there arc various other words which are identical with
pin except at the outset: in, tin, dan, ehin, gin, kin, fin, thin, sin, shin, Min
(the nickname), Ly1111, win, spin, skin, grin. Any pair of words drawn
from this list bean testimony to the relevance in English of some differ-
ence of &Ound. Thus the pair pin and in shows that there is a functionally
relevant difference between having a p-sound at a certain place and
having no sound at all there. The pair din and skin demonstrates the
relevance of the dift'ercnce between a d-sound and an sk-sound. The
pair pin and spin att~ts the distinctiveness of the difference between
starting with a p-sound and starting with an s-sound followed imme-
diately by a p-sound. And so on.
Four of the words in the list arc special: in, since it begins with no
consonant ~nd at all; and sp~. s!!.n, e· each of which begins with a
combinatiOn of two consonant sounds. Setting these four aside, we are
left with a set of fourteen: pin, bin, tin, d;,a, ehin, gin, kitl, .{Ill, thin, sin,
skin, Min, L,,, will. ' · · •
Each of tbeae fOUI'UIC!ID words begins with a Jingle CODIODBDt 1011nd,
and each of the CODIODant sounds ia d•ercnt from each of the others.
Thus, io place ot a llqle two-way clliference of 100nd (as in }lin and
6ia) what we have is a ftlluJDrk of~ 4;/lftneU of SOIIIIII. Since each
of lihe wwcla cootraltl direcdy with each of the othen, there are DiDetr·
 18 PHONEMES

one contrasts in all. Thirteen of these contrasts involve the p-sound: p

1 different from b; p is different from t; p is different from d; and so on.
IWe lhalllater (§11) see that these differences are not all of the same.
order (p is mor• differml in sound from d than it is from b), but at the
moment this consideration is unimportant: the ninety-one contrasts
are functionally all on a par.
Now does this list of fourteen words cover all the contrasts which are
relevant for the fourteen consonant sounds involved? \\·e can find out
by considering some other family of partially similar words. Let us
examine the family which includes pie, a word which obviously begins
with the same p..ound found at the beginning of pin. Setting aside eye,
which has no initial consonant-sound, and words like pry, slyrfry, which
have two e~, we find the following: pie, buy, tie, die, guy, fie, vie, thigh,
thy, sigh, slry, my, nigh, lie, rye, Wye, high. Here we have seventeen differ-
ing in~tial consonant sounds, or 136 two-way contrasts, sixteen of them
involving the p..ound. Some of these are duplications of sounds or
contrasts revealed by the first list, but some of them are new: pie versus
guy, for example. And a few of the sounds and contrasts represented in
the first list do not recur here. What the two lists taken together show
can be determined by consolidating them in the following fashion:

pin bin tin din chin gin kin - fin - thin

pie buy til die - - guy fie vii thigh
sin shin Min Lynn,- win
1/ry sigh s"' my rriglr lit ryt Wyt high.

The coupled lista ahow twenty different consonanta, provided our

pairing-oft' is correct, but they do not ahow every imaginable two-way
contrast: the total number of two-way contrasta direcdy attested by the
forms in the c00101idated list is 169.
Tbia procedure can be continued, by adding another famD.y of par-
tially llimi1ar warda, and then another, and so on. Eventually, the addi-
tion of another new family faila to show anything not attested by those
already c:onaidered. One must naturally be careful not to stop too soon,
or 10mething may be missed. Table 2.1 presenta nine familia, which
among them both illustrate every different lingle consonant-sound with
w.hich English words begin aad afl'ard an example of every pallible
two-way contrast between thOle COIJIOD•nta. ~ column is a family;
 ENGLISH FINAL CONTRASTS

'I'ABLE 2.1
columu:

...
1 2 3 4 5 6 7 8 9
row:
1 paiD pie pooh• paw pinr pet pat pal
2 bane buy boo• baw 4 bine bet bat
3 Tainel tie too tine tat tea
.... Dane die do daw dine debt dcD
5 chain chew chaw chine Cbetl chat
6 Jane 1 Jew jaw jet
7 cane C'.OO caw kine cat
8 gain guy goo get gat
9 fain fie Coo• fine lat
tO vain vie viDe vet vat
11 thane thigh thaw
12 thy thine that
13 I<"IDe ligh aaw sign IICt aat
14 Zane1 zoo Syne1
15 Shane1 ahy shoe Shaw 1 ahinc::
16 main my moo maw mine met mat
17 nane 1 nigh gnaw nine net JD&t
18 lane lie loo l.w line let Leo•
19 rain rye Roo 1 raw ret rat wrea
20 waiD Wye woo wine wet wen
21 you yaw yrt ;en
22 hitrh who haw hat hen
I Suma'IDCI. • Given names or nicknaJilCII. 1 Scottish, but familiar through
poetry or atorie~. 4 lnterjeetioDI. 1 A variety of Buddhism. Other unfamiliar
words 1D the Table can be found in any college dictionary.

each row includes only words which begin with the same consonant
sound. As c-an be seen, there are twenty-two different consonant sounds, 1
and a total of 231 two-way contrasts. One could add other families'
endlCI'.S'y, but nothing new would turn up.
2.!. En.gluh Final Contnata. So far we have sortea out onJy the
distinctively different consonant sounds with which one..syllable Enfl(-
lish words begin. The pair pin and bin dift"er only at the beginning; but .
the pair pin and pan differ in the middle, and pi" and pip contraat only
at the end. As our next step, let us sort out thedift"erent single consonant
sounds with which one-syllable English worda end.
This we can do by the same procedure used for the word-iDitial posi-
tion, except that our word families will be dift'erently chOICD: e.g.,
Iii, Jlit, I*"- J1id, Jlil, pith, lisA, pin, p;,, /JUT, pill. It requires a lupr
N
Q
TABLE 2.2
cobulua:
t 2 3 4 s 6 7 8 9 10 t1 12 13 14 IS 16 17
row:
1 pip Jape rip cope pap lip c:op cap reap leap
2 lobe rib sib c:ob babe cab bub lib Rubel
3pit late coat pat sit c:ot bait cat mitt but wit fit root
4 laid load rid axk: pad Sid1 c:od bayed cad read lead bud rood
s pitcb rich coaclt patch catch reach leech witch Fitch
6 ridae c:odge cadge midge budge
7 pick lake Rick 1 coke pack lick cock bake reek leak Mick1 buck ¥lick
Bpig rig cog league Mig• bug wir fie
9 Lafe1 loaf Riff calf leaf roof
10 lave CCJYe sieve calve leave 1'00\le
u pitli lou:h path lith wreath myth with 1 nath
12 .athe loathe bathe wreathe with1
13 lace p.. Sial base ea.• Rccc:el leue mill bUI
14 Ia.,. )OWl COle leas buzz fizz ruse
15 piahl cosh cash leash wish filb ruche
l6 lop beige liege rouge
17 lame loam rim comb P111111 cam ream bum room
18 pia luc lea: CODe paD liD COD bane can Min• buD
leu wiD fin tulle
10 .... riDg ling Ming buDg wiDg
20 peer rear cere car men: weir fear
2tpill rill coal pal sill bale CaJC reel mill will fill rule
1 ......... 1 GiW:D aamcs or DickDamrs. 1 Iatcljcction. c A. in Cal T.rla ( - "Calilomia In~titute of TechaalatrY").
. . . . of .. kiiKI oflirplaDe. 1 The wofCI willl hu both proauaciatiDDI.
ENGLISH FINAL CONTRASTS 21
number of word families to illustrate all the poasibilitic:s here than we
needed for word-initial position: Table 2.2 includes IICYCllteen. And we
encounter certain difficulties which did not turn up in initial position.
Mainly, these difficulties are due to thr- fact that not aU of us speak
English in quite the same way. Table 2.2 includes a row pur, rear, sere,
car, mere, weir,fear. For the writer, and in general for speakers of Middle
Western American English, the vowel of peer is quite like that of pip,
ping, pill, and so on, so that the word is properly included in the first
column. But there are some Middle Westerners, and a great many
people elsewhere in the English-speaking world, for whom this does
not hold -indeed, in much of New England, parts of New York City,
parts of the South, and in British English, there is no terminal r-sound
at all in such words. It is unfortunate that we should have to encounter
difficulties of this sort so early in our discussion, but there is nothing
which can he done about it. We are forced to choose some one variety of
English, and since the ~ter is m011t familiar with the Middle Westem
variety he has chosen it.
The words of colulllllll 6, 8, 10, and 17 are likewise chosen in terms
of this one type of English, and require resorting Cor other types.
Some ~dle Westerners (and some others~ pronounce the four
words loge, blig"t, liege, and rouge (row 16) with exactly the same final
consonant sound that they use in ridge, codge, cadge, midge, budge (row 6).
Others, from all areas, use two different sounds. Row 16 must be
deleted for those who do not make this distinction.
Table 2.2 shows twenty-one dift'~t final single consonant sounds,
and afFords ·examples of 206 minimal two-way contrasts. One further
two-way contrast is attested by the pair suthe and siege, not on the table.
No minimal contraits can be found for the final CODIOD8Dt sounds of
·the pain
rouge versus ritlg1
rouge versus ring
rouge versus rMr.

Therefore the final consonants are defined in terms of a network of oaly

207 two-way contrasts.
Before passing on to an examination of the vowel sounds ol mcmo-
syllabw 'ct us first note that to a large extent the cWf'ereot po.ible
22 PHONEMES

lingle initial conaonants and the different possible single final con-
IODants can be paired ofF. Thus:
pain begins as pip ends; and likewiae
bane and lobe Taine and pit
DaM and laid rhain and piteh
Ja'M and ridge caM and pick
gain and pig fain and Lafe
vain and lave tha'M and pith
thy and lathe sane and l~J~Jr
Za'M and lays Shane and pish
main and lame na'M and pin
la'M and pill rain and /JI".
These leave three initial consonant sounds (those in wain, you, Haine)
apparently unmatched by anything final, and two final ones (those in
loge, ping) apparently unmatched by anything initial. Occasionally one
hears someone pronounce the French name Jean'M somewhat in the
French way (though in an English context), with an initial consonant
sound like the final one of rouge. This is so infrequent that we might
almost leave it out of account altogether. Even rarer is the pronuncia-
tion of the authoress's name Ngaio Marsh with an initial consonant
sound like the final consonant sound of sing.
2.4:. English Medial Vowel Contrasts. The procedure which we
have now usCd for the initial and for the final consonant sounds of
monosyllables can also be used for the medial vowd sounds (demon-
strated by the difference between pin and pan). On this score, dialect
variation becomes really great. Table 2.3 shows, with ten families of
words, those contrasts which are made regularly by most speakers .of
Middle Western English. Any such speaker, chasm at random. may
lhow a few additional distinctions, but there is little agreement as to
these additional ones from one speaker to another, and we omit them
for the sake of greater simplicity at this stage of our discussion. The
Table shows fourteen contrasting vowel sounds, and minimal two-way
contrasts for every pair of the fourteen, giving an interlocking network
of 91 differences of sound.
Many varieties of English show a much larger number of contrasting
vowel sounds. A speaker of any of these other varieties will inevitably
be distutbed hy.T!_blc 2.3, and particularly by the way in which the
SUMMARY; PHONEMES AND SPEECH SOUNDS 23
TABLE 2.3
column:
1 2 3 4 5 6 7 8 9 10
row:
1 bee ye thee beat keen beak heel keyed Deal'
2 bay yea 4 they bait cane bake hail bail
3 by thy bite kine bike Hile 7 bile
4 boy coin Hoyle' boil
5 boo you boot coon cooed Boole'
6 bow 1 though boat cone whole code bowl
7 bow 1 yow thou bout howl cowed
8 baw yaw bought hall cawed ball bore
9 bah bot con bock cod bar
10 baa• bat can back Hal• cad
11 yeah 1 bet ken beck hell Ked' bell bear
12 the 1 but buck hull cud burr
13 bit kin hill kid biD beer
14 book could bull boor
1 As for shooting an arrow. t As from the waist. • The bleat of a sheep; some
people pronounce this the same as bah. t As in yea, team! 1 As in Oh yetlil! Some
people pronounce this so as to put it in the preceding row. 1 One way of pro-
nouncing the word in isolation; the other way makes it identical with 1/ue. 7 Sur-
names. •Nickname. 8 A trade name.

words in column 10 are aligned horizontally with those in the other

columns. In the case of bore, bar, bear, beer, and boor, the reader will
simply have to accept, on faith, the writer's assertion that in his speech
the vowel sounds are as indicated-bar has the vowel sound of bah, bot,
con, bock, cod; bear that of bet, ken; and so on. In the case of burr, the
writer "Cannot insist that the vowel is exactly the same as that of but, buek,
hull, cud, for,· to a trained ear, it is not. But it is nearer to that of
but, buck, and so on, than it is to any of the other vowel sounds repre-
sented in the table, and the writer is unable to find any direct contrast.
in any environment, between the vowel sound of burr and that of but or
buck. Since contrast is much inore important than exact identity, the
alignment is made as shown on the Table. There are certain problem~
connected with this that we shall not be able to deal with until much
later (§40).
2.5. Summary; Phonemes and Speech Souncb. From fl2.2-4 we
see that a monosyllabic utterance of Middle Western English may
begin with any of the single consonant sounds sorted out in Table 2.1~
or with certain combinations of two or three of them (spin, .strilw), or
24 PHONEMES

with no COJ'IIOD•nt II011DCf at all (ill, mttl). It then continues with one or
another of the vowel sounds sotted out in Table 2.3. It then ends with
one or another of the single consonant sounds sorted out in Table 2.2,
or with certain combinations of two, three, or four of them (stand, le:tl,
61impsetf), or with no final consonant at all. Of course, not all thr
theoretically possible combinations actn;J Jly occur-if they did, then
there would be no holes in our tables, and one family of words would
auffice for each table. But any monosyllable which d01s oc.cur accord•
with the description just given. We have sorted out all the differences
of co111onant and vowel sound which function to keep monosyllabic
utterances apart.
Our utterances, of course, are rarely monosyllabic. Therefore we
cannot be sure that we have covered every functionally relevant dif-
ference of sound in English. Before we continue our analysis of English
phonology, however, let us for a moment assume that the description
given above is complete, so that we can pose some crucial questions.
Just what is the P-sound (as we have been calling it) at the beginning
of the word pin? There are two different senses in which this question
can be understood, leading to two different answers.
The first (and less important) answer is to describe the p of pin in
terms of what it sounds like, or in terms of how it is produced by a
speaker. As we shall see in §§7-9, the p of pin can be approximately
described as a "voiceless bilabial stop." This description is cast in the
terminology of what is called articulatory phonetics. When we say that
pin begins as tip ends, we are identifying the first consonant of the
former and the last consonant of the latter on the basis of such descrip-
tion. Any two sounds, in the same language or in different languages,
which fit the same description in terms of articulatory phonetics, are
uid to be instances or recurrences of "the same" speeeh sound. Thil
renders the term "speech sound" relati~·e, since our articulatory de-
ICription may be either loose or precise.
The second (and more important) answer turns on the consideration
mentioned in 12.1: that the sole function of II011DCf in language is to keep
utterances apart. The phonological l}'ltem of a language is therefore
not so much a "act of sounds" as it is a network of di.f~rmees belwlm
ltlflfltls. In this frame of reference, the clements of a pbonologicall}'ltem
cannot be defined positively in terms of what they "are," but only
negatively in term1 of what they are not, what they r:onlrt1sl with. While
SUMMARY; PHONEMES AND SPEECII SOUNDS 25

it is true, and not irrelevant, that English p is a "voiceless bilabial stop,''

it is much more important that p is different from certain other elements
namely, those sorted out in Tables 2.1, 2.2, and 2.3.
Since the element p in English is something different from the various
other elements, we are trained to pronounce it, by and large, in such a
way that it cannot easily be mistaken for any of the other elements;
and, as hearers, we are trained to catch even the subtlest clue of pro-
nunciation which identifies what we hear as a p rather than as any of
the others. But this does not mean that we must necessarily pronounce
p in exactly the same way every time, and we do not. So long as what
we say sounds sufficiently unlike any of the other elements which might
occur in the same context, yielding a different utterance, our hearers
will interpret what they hear as a p.
This affords us more leeway than we realize. It is only natural that
we should be unaware of our actual variation in the pronunciation of
an element like p, since throughout our experience with English, from
earliest childhood, we have been trained to ignore certain variations
in pronunciation, and to pay attention only to key differences. Accord-
ingly, it is difficult to demonstrate to a speaker of English that his
pronunciation of an element like p does vary quite widely. But it is
easy enough to demonstrate comparable irrelevant ranges of variation
in other languages. For example, if we listen to a Menom.ini Indian
saying several times his word with the meaning 'he looks at him,' we
hear in the middle of the word now a sound something like ouc p and
now a sound more like our 6. We hear the difference because English
trains us to hear it. But the Menomini does not hear it, because in his
language this particular difference of sound never functions to keep
utterances apan-and therefore he has no reason to need to hear it.
Conversely, a speaker of Hi.udi, hearing us say pin a number of times,
will report that we fluctuate between two initial conspnants, the dif-
ference between which U: functional in his language, but not in English.
In this functional frame of reference, when we are dealing with the
phonological system of a single language we do not ordinarily use the
term "speech sound." For example, we do not call Englishp a "speech
sound" in this context, because whether all occurrences of English /1
are "the same" speech sound, or some are one speech sound and some
are another, depends on who is listening and on how we c:lcacribc what
is heard. Instead, we say that English p is a plwnem~.
26 PHONEMES

The phonemes of a language, then, are the elements which stand in

contrast with each other in the phonological system of the language.
English p is a phoneme, and so are the various other clements sorted
out in Tables 2.1, 2.2, and 2.3. Throughout all our subsequent discus-
lion, it must constantly be remembered that a phoneme in a giveo
language is defined only in tlm1s of its t#lfertmres from th. otlwr pho,.,.s of
IM satM lt.mgutJ.ge.

NOTES

New terms: phonology or plunvmi&s, phonological syst1m of a language,

pho,.,., 1/.Jetl&h 10'U111i. Articulatory phonetics is mentioned in passing, but
will not come up for detailed discussion until §7.
Hockett 1955 is a detailed treatment of the substance of §§2-13;
further bibliography can be found therein.
 3·

PHONEMIC NOTATION

3.1. Linguistics is one of many fields in which it is necessary to devise

certain sorts of special notation.
A football coach charts plays on a blackboard with O's, X's, and
arrows. Radio circuits are represented on paper by diagrams which are
half pictorial and half arbitrary. Chemists have a special notation for
representing elements, compounds, and reactions.
Let us for a moment consider the special notation of chemistry. At
the basis of this symbolism is the assignment of a capital letter, or in
some cases a capitallcttcr followed by a lower case letter, to eaoo ele-
ment. Thus "'H" represents hydrogen, "C" carbon, and "Cl" chlorine.
In structural formulas, each of these symbols represents not just the
element, but a single atom thereof; a dash represents a valence bond.
With these conventions, the structural formula for methane or manh

H
I
H-c-H
I '
H
actually forms a sort of picture of a molecule of that compound. Of
course, the individual symbols "H" and "C" don't "look like" atoiDI
of hydrogen and carbon-nothing "looks like" atoms, since atoms are
too small to reflect light, and thus too small to see with any physically
poaible microscope. But at the size-level of the whole structural far-
mula, the gt'Ol!letrical arrangement of the constituent symbols is pre-
sumed to be roughly parallel to the geometrical arrangement of the
con~tituent atoms in a molecule of manh ps. It is to this geometrical
2.7
 28 PHONEMIC NOTATION

parallelism that we refer when we call the formula a sort of .. picture"

of what it represents.
Now in discussing the phonologic system of any one language-and,
indeed, often in discussing other aspects of the language-it is necessary·
to use a system of notation which will reflect the phonemic system of the
language and show the phonemic composition of any utterance or part
of an utterance. The way in which a phonmaic notation for a language is
developed is very muC'h like the way in which chemists have developed
their notation for structural formulas.
To start with, it is necessary to assign a different visible mark to each
phoneme of the language. We could use any symbols whatsoever for
this purpose; logically, all that would matter would be that we had
exacdy enough of them, all different in appearance. But it is helpful if,
insofar as possible, we choose marks which already have some sort of
association with the phonemes to be represented. The chemist's sym-
bols have been chosen with an eye to such mnemonic help: "H" is the
first letter of the written word "hydrogen," "Na" for sodium is the first
two letters of Latin "natrium," and so on. In English writing, the letter
"p" is often associated with the English p.phoneme (pin, pull, Mp,
apple, bump) and rarely with anything else. Thus the sensible choice for
this phoneme in our phonemic notation for English is the letter "p."
Next, it is necessary to decide upon some proper geometrical ar-
rangement for the symbols which represent the constituent phonemes
of any utterance we want to depict. In speech, the phonemes of an
utterance occur largely one after another in time. The surface of a sheet
of paper or a blackboard has no time. on it, but we can follow the habit
of ordinary Western writing, placing the symbol for a phoneme which
occurs earlier to the left of that for a phoneme which occurs later, and
going on to another line when we reach the edge of the writing surface.
When these two steps have been taken, our phonemic notation for the
particular language is established. We have the necessary apparatus
for a graphic representation of IJII}thing th. speakers of tile language '"" SOJ.
Furthermore, our phonemic notation for the particular language ill
pictorial to about the same degree as ~re the chemist's structural for··
mulu. There is nothing similar between the visual shape of the mark
up" and the auditory shape of the English phoneme which we choose to
represent by that mark iri our notation. But then: is a resemblance
between the geometrical arrangement of the CODitituent symbola in the
ENGLISH PHONEMIC NOTATION 29
phonemic formula displayed here:
pin
and the geometrical arrangement of the three phonemes we produce
when we say the English word pin. In one case, the arrangement is in a
single spatial dimension, and in the other it is in the single dimension
of time; but, given our arbitrary equating of"earlier than" with "to the
left of," the linear ordering is the same.
3.2. Special Problems in Phonemic Notations. The linguist faces
one problem of notation which does not disturb the chemist. The latter
needs only one notation for all possible structural formulas, for thc1·e
is only one set of elements in the universe. The linguist needs as many
di.flerent phonemic notations as there are languages to be dealt with.
It would be wasteful to make up a whole new set of visible marks for
each new language. But if some of the same marks are to be used from
one language to another, their assignment should not be random.
What we do is to pay some attention to the similarities between
phonemes of different languages as speech sounds: "p" in English for a
certain phoneme, and "p" in our notation for French, Menomini, or
Chinese for phonemes in those languages which are at least something
like English p to the ear. Complete consistency cannot be achieved,
since the phonemes of different languages are defined by different
interlocking sets of contrast. But the most extremely misleading assign-
ments of marks to phonemes can be avoided.
Formulas in phonemic notation often have to be displayed directly
in the middle of a passage of written English. To set the notation ofF
clearly from context, we shall henceforth enclose every phonemic
formula between slant lines:
the English phoneme /p/
the English word /pin/ is a monosyllable
Menomini /wa·pamew/
and ao forth. The context will always indicate what language is being
discussed. But it must be remembered that "English /p/" and, say,
"Menomini /p/" refer to entirely different phonemes, despite the use of
the mark ''p" for both. .
S.3. Eaglilb Phonemic Notation. We shall now assign marks to aU
the English phonemes uncovered by our investigation in 12. The re-
 30 PHONEMIC NOTATION

maining symbols which are needed will be added as we unravel the JUt
of the English phonological system.
First the symbols for the initial consonant phonemes illustrated by .
Table 2.1:

/p/ pain, pie /6/ thy, lhitll

/b/ banet buy /s/ sane, sigh
/t/ Taine, tie /z/ Zane, zoo
/d/ Dane, die /1/ SIUJne, shy
/~/ ciUJin, chew /m/ main, my
/J/ Jane, Jew /n/ nane, nigh
/k/ cane, coo /1/ lane, lie
/g/ gain, guy /r/ rain, rye
/f/ fain, fie /w/ wain, Wy1
/v/ vain, vie /j/ you,yaw
/6/ thane, thigh /h/ Haint, high.
Nineteen of these will be used also, of course, for nineteen of the final
consonant phonemes illustrated by Table 2.2: /p/ pip, /b/ lobt, /t/ pit,
/d/ laid,/~/ pitcll, /J/ ridge, /k/ pick, /g/ pig, /f/ Lafe, /v/ lave, /6/ pith,
/6/ lathr, /s/ lace, /z/ lays, /1/ pish, /m/ lame, /n/ pin, /1/ pill, /r/ /Jier.
In addition, we need for final-position consonants the following two:

/f./ loge, rt1ug1

/Q/ ping, ring.
This totals twenty-four marks for consonant phonemes, and this ia
all that is needed even for utterances longer than a single syllable.
So far as the graphic shapes of the marks are concerned, sixteen are
identical with English letters often used, in ordinary English spelling,
with the values assigned them; one, /j/, is a familiar letter used
in an unfamiliar way (though this usage is familiar in the spelling-
systems of German and some other languages); the remaining seven,
/~J 118&" Q/, are strange, but will variously recur in our phonemic
notations for certain other languages.
For the vowels sorted out in Table 2.3 we shall UIC the followinc
symbols:
 ENGLISH PHONEMIC NOTATION 31
/ij/ hr.r, ye, bt:at /bij, jij, bijt/
/ej/ hay, bait /bej, bejt/
/aj/ by, bite /baj, IJajt/
/oj/ hoy, coin /boj, kojn/
/uw/ boo, boot /buw, buwt/
/ow/ though, boat /oow, bowt/
/aw/ llzou, hou·l /oaw, haw)/
/a/ haw, hall, bore /ba, hal, bar/
/a/ bah, hot, bar /ba, bat, bar/
/z/ baa, but, cad /bre, bret, kred/
/e/ yeah, bet, be11r /ye, bet, ber/
fa/ the, but, burr faa, bdt, barI
/i/ hit, beer /bit, bir/
/u/ hook, boor /buk, bur/.

To the right we have placed the complete phonemic formula for each
of the sample words.
Here even some of the symbols which are quite like English letters in
shape have been assigned values for which English writing affords little
mnemonic help. Our reason for this is the other mnemonic principle
explained in §3.2: some of the same marks are to be used in our pho-
nemic notations for other languages, and it is easier to maintain some
degree of consistency in this, from one language to another, if the
marks listed above are assigned the indicated English values.
In seven cases, we have used a combination of two marks instead of
a single mark: /ij ej aj oj uw ow aw/. Linguists who have worked
with English phonology are in disagreement as to the status of these
seven vowel-like elements. Some believe that a word like bay contains
a vowel which can be identified with that in bd (our /e/), followed by
something which can be identified with the initial phoneme of ytJU, yes
(our /j/). For them, once the symbols "e" and "j" have been assigned
the values just indicated, then the notation /bej/ for the word bay, or
/bejt/ for the word bait, is a necessary consequence. A similar line of
reasoning justifies the other compound notations with "w" and "j" as
the second mark.
Other specialists in English phonology disagree, feeling that every-
thing after the /b/ in bay is just a single phoneme, entirely on a par
32 PHONEMIC NOTATION

with everything between the initial and the final CODIOIWlt of betzl, bit,
b•t, btzt, and 10 on.
We do not have to take sides in this argument for our present pur-'
poees. Fortunately, we can use one and the same notation whichever
opinion eventually prcwes correct. For those who prefer to interpret the
vowel of bt.IJ as a sequence of two phonemes, the notation /ej/ can be
interpreted in that way. For those who prefer to interpret the vowel of
b11J as a single phoneme, the notation /ej/ can be interpreted as a
compound mark for a unitary element, like the chemist's "He" for
helium (in contrast to "H" for hydrogen). Table 2.2 shows that the
initial consonants /j/ and /w/ are not matched by any final consonants;
therefore we run no danger of ambiguity in monosyllables in using the
marks "j" and "w" also as constituents of our compound symbols; and
it will turn out that the notation remains unambiguous for utterances
longer than a single syllable.

NOTES

Problems. Gleason, Workbook (see bibliography, Gleason 1955b), on

pages 15-18, gives a list of English moDOByllables for practice in
transcription.
 4·

ENGLISH INTONATION

4.1. In our analysis of English one-syllable utterances in §2, we pre.

tended that two such utterances are either kept apart by some differ-
ence in constituent vowels or consonants, or else are not kept apart at
all. This was an oversimplification. Consider the following dialog:
JACK: Where're you going?
BILL: Home.
JACK: Home?
BILL: Yes.
The second and third lines of this dialog are not identical to the ear,
ct.nd certainly not the same in meaning. Yet the constituent vowels and
consonants of the two are the same: /howm/. The difference in sound
which keeps them apart lies in the changing pitc.h of the voice. Dis-
tinctively different features of English speech melody constitute
intoNJtion.
Although our conventional punctuation marks permit ua to note
some intonational contrasts in writing (Home. versus Homt?), they do
not provide for this either consistently or fully. The same mark may be
used in spite of intonational differences, and, conversely, different
tnarks may have to be used even though the intonation is the same, •
it commonly would be in each of the following two utteranccl:

JACK: Where're you going?

BJLL: I'm going home.

A better' way of marking intonation is therefore necesaary. ldcaUy,

the marb should provide for all significant intonational contraats, but,
33
34 ENGLISH INTONATION

Iince the precise analysis of English intonation is still a matter of some

doubt, in some instanct"s they may not.
Until recently, intonation was more or less tacitly ignored by most
linguistic scholars on the assumption that it did not vary significantly
from language to language, or that anything so "natural" hardly war-
ranted serious consideration. It is perhaps true that certain features of
speech melody are to be found in all languages (e.g., rise of pitch and
volume under the stimulus of pain or anger), but such universal fea-
tures, if they exist, are not part of intonation as we now use that term.
Recent research suggests that every language has a system of basic
speech melodies which is as unique to the language as is its set of
vowel and consonant phonemes. It may be that the normal effect of
an intonation is sometimes concealed, or overridden, by the superposi-
tion of nonlinguistic features of speech melody under the stimulus of
strong emotion, but this is on a par with the fact that normal articula-
tion of vowels and consonants is sometimes distorted by excitement,
depression, or drugs.
4.2. PLI and TC.. Let us begin by examining in more detail the
ways in which Bill might have said home in answer to Jack's question,
and the ways in which Jack might have intoned his rejoinder. If Bill
is offering a perfectly matter-of-fact reply, without any implication
that Jack really ought to know the answer without asking, he will usu-
ally start with a relatively high pitch, and let it fall rapidly to a very
low pitch at the end of the word. The relatively high initial pitch will
be represented by a / 1/ before thf' wnrrl hnmt, and thr: rr:latively low
level reached at the end by a / 1/ written after the word, thus:
(1) 1home 1 !
(The arrow at the end will be discussed later.) Instead of this, Bill may
start at about the same pitch but let it fall only slightly: in this case we
write / 1/ instead of/'/:
(2)
·-··!
This carries a different meaning in the present context: 'I dOD't par-
ticularly rlltlftl to go home, but there's nothing else left to do.'
If Bill wants to imply 'Of course I•m going home; what else would
you expect at this hour?'-then he may do one of th~ things. One is
to start at an especially high pitch, higher than that ll)'lllbolized by
PLS AND TCS 3S
/ 1/-we ahall UJe the mark /•/-but to glide down just as far by the cad
of the word as in way (1):

(3) ·-·'!
Or else he may start at a very low pitch (/ 1/), and let the pitch rise,
either alighdy-
(4)
-or more aharply-
(5)
The symbol / 1/ at the end of (5) obviously represents a level of pitch
higher than fl/ but lower than / 1/.In (4), the terminal upwards-point-
ing arrow means that the rise in pitch ia to a level higher than /'/ bUt
lower than / 1/ ; in (5), the same terminal mark mCilDI that the rise in
pitch ia to a level higher than / 1/ but lower than / 1/. Thus the extent to
which the pitch rises in these two ia indicated joindy by the last au~
acri.pt numeral and the arrow after it: the arrow mCilDI 'rise to a point
higher than such-and-such,• and the last superscript numeral defines
the 'such-and-such.• That these two rises are potentially different in
function can be shown in another context. Suppose that Bill's answer ia
delivered in way (4). Jack's comment might then be delivered in way
(5), with the meaning 'You don't mean 1/ral, do you? I hadn't ex-
pected itl' But Jack cannot achieve this meaning by using way (4),
which ICCIDI never to imply any feeling of interrogation.
The intonational phonemes /'/, / 1/, / 1/, and 1•1 are called /Jilel
Uw/1 (Pu); I!/ and /1 I are tmlliul ttmtour1 (TO.). Our examples so
far do not illustrate PL / 1/ very well. The contrasts among/'/, / 1/, and
/ 1/ can be more clearly shown as follows. If Jack asks k• you.goitw ,._
IIDW? Bill may aDSWCr

(6)
which either carries the same overtones described for (4) above, or else
implies that Bill is about to continue with some comment on his answer.
If Jack quietly calla Bill! in order to get Bill's attention, Bill may indi-
cate that he is listening by saying

(7) •Yrf
36 ENGLISH INTONATION

Some people answer the telephone this way, or say 1u with this intona-
tion to a stranger who comes to the door; in the.e contexts it strikes
many of us 81 bnuque and a bit impatient. Lastly, suppose that BiU hu
asked Jack to do something, hoping for an affirmative answer, but that
·he does not quite hear Jack's response. He may aak 'Did you say yes?'
by saying
(8) 1Yes 1 f
In all three ofthe.e ways ofsaying1es (6, 7, and 8), there is slight rise in
pitch 81 the word is spoken, but the pitch starts lowest in {6), somewhat
higher in (7), and still higher in {8).
So far we have Wuatrated two TCs. There are three in all, but before
introducing the third iet ua clarify the distinction between the two
already introduced. There are two common ways for Jack to call quiedy
to BW to attract his attention-assuming that both are in the same
room, but that BW is occupied with his own work or thoughts. One is
like the ICCODd way (2) of saying lttnn6:
(9)

and the other is

(10)

The IJOlBtion in (10) indicates that the pitch of the voice starts at level
/ 1/, ~all the way to / 2/, and then rises somewhat, though not far
enough to reach Jl/ again. Notice, thus, that the TC /f / •IUJd1S in-
VQives a terminal rise in pitch, even if' the voice has. first dipped down
&om a higher level.
An exaggerated variety of (9) is our usual way of calling someone
from a distance; the overall inCieue in volume, pitch, and duration is
probably part of the "natural" framework common to all languages.
Either of the following is a sort of concessive a~~ent, implying 'that
may be true, but 1 have doubts 81 to its relevance':

(11)
(12)

To the writer, the second of these tends to imply more serioua reaerva-
tiona than the fint.
IIACR.OSBGIIENTS 37
Thus the key characteristic ol It/, diatinguiahing it from / !/ and
from the third TC (to be discussed in a moment), is a terminal rist in
pitch, which ends at a level somewhat higher than the PL written
direcdy before the mark /f /, but usually not 10 high as the next higher
PL.
The TC / l/ ia distinguished from /f I basically by the absence of
this terminal rise. Its pOiitive characteriatic:l are a fading-away of the
force of articulation, often with a drawling of the last few vowelJ and
consonants. When the immediately preceding PL ia /'/,the fade ia usu-
ally accompanied by a fall of the pitch to a point IIODlewhat Hlt1111
PL /1/.
The third TC, /I/, is marked by the ablence of the positive featurel
for either If I or I!/. This TC most often occun where the speaker
goes right on talking, so that isolated examples, direcdy comparable to
those that have been given for the other TC., are harder to find and to
illustrate. However, colllider someone who is about to answer a com-
plicated question, and who must think through what he is going to say
before saying it. He may begin in either of the following ways:

(13)
(14)

(13) is cut off suddenly without forewarning; (14) fades away from the
outset. Different impressions are conveyed to the audience. Hearing
(13), we sense that the speaker has realized onJy after beginning to
speak that he must take time to think hia answer out. Hearing (14), we
sense that the speaker realized this necessity from the outset, and that
hia Will is uttered, in part, to let us know that the necessary cogitation is
under way.
4.3. Mac:nJ~egmentl. Utterances longer than a single syllable differ,
as t? intonation, in one or more of three ways from those of a single
syllable.
First: a longer utterance lllilY contain two successive intonatiou:

(15)

Second: a single intonation may be stretched through two or IDOI'e

suCCCIIive syllables, a PL at the beginning, a PL at the end, and a TC
38 ENGLISH INTONATION

at the end; compare the following:

(16) 1No 1 !
(17) 1Newr 1 !
(18) 1 TnTih/e 1 !
(19) 1Dictionary 1 !
(20) 11 want to go 1!
(21) 11 want to go tlrlre 1 !

(22) 11 want to go along with you 1l

(23) *Ele~~t~tor operator 1!

Note that in all of these (16-23), the most prominent syllable is the very
first one. (21), thus, answers the question WHO wants ID go there?-not
the question· Where do you want to go? The most prominent syllable of an
intonation, be it the first or not, and the PL which accompanies it, are
said to be at the center of the intonation.
Third: a single intonation may include one or more syllables before
ita center. The PLat the center of an intonation is always the next to
the last PL in the intonation, the last one being that which occurs at
the end along with the TC. If there are syllables before the center, then
the pitch on which they are spoken is also distinctive, and we place a
mark for a PL at the beginning:
(24) 11 WfJIII to go 1there1!
(25) IJ want to •go there1!
(26) •r~want tD go lllerel!
(27) 1.d11 1eli1Hllor operator'!
(28) 1He's fJII 1elel1tllllr operator 1!

In all of these (24-28) 1 the PL at the center is / 1I (and the center ia the
syllable on which / 1/ is written).
The stretch of material spoken with a single intcmation is called a
fltl#rOSiptlftl, We may freely speak either ol the center of an intonation
or of the center of a macrosegment. Everything from the center to the
end, including the center, ia the IINid; anything which precedes the
center is a pnultml. By definition, then, every macrosegment includes a
center and a head (though the two may be coterminous, u 1-14, 16,
24), but only some include a pendant (24-28). Every macrosegment
ends with a TC, which therefore automatically marb the boundary
between suc:cessive macroecgmentl in a single utterance (u in 15, the
MACROSEGMENTS 39
only example so far of more than one macrosegment). Every macro-
aegment includes at least two PLs, one at the end and one at the center;
if there is a pendant, there is an additional PL at the beginning, and
there may be a fourth somewhere between the beginning and the
center. Examples of the latter possibility will come later.
The commonest and most colorless intonation for short statementa
is 11 11!/. By this abstracted notation, we mean that the pendant, if
any, is spoken on PL 121, that the head has PL 111, and that the last PL
is 111 and the TC I!f. To the examples already given (1, 16-23, 24-28)
we may add one here:
(29) 2My nmne is 1Bill 1!
This is also the most colorless intonation for short questions built around
a "question word" such as who, where, what:
(30)
On the other hand, questions of the sort that allow a yes-or-no answer
have a different most-neutral intonation, 12 11 f I:
(31) 2/s ynw name 1Bill1 f
A aeries of words in the proper order for a statement can be made into a
question merely by using the indicated intonation:
(32) 1His name is 1Bill1 f
The following is a perfectly normal question:
(33) 1 What do you 1do 21 1with a stiff 1neck 1!
Notice what happens if we change the intonations as follows:

(34) 2 What do you do with a 1sti.ffl! 'neck1 f

The discovery of questions which could be distorted in this manner used
to be a sort of parlor-game.
Again, notice the following two normal questions:

(35) 1What are we luming for 1diftfllr 1j 1Mother1 f

(36) 1 What are we hmli"' for 1dift1Ur 1! •6eerf

Placing the intonations of (36) on the words of (35) give~ a grue10mcly

40 :ENGLISH INTONATION

cannibalistic eft'ect:
(37) 1 W/ud •• we ltaTJirrg frw 1tli'11111r1l 1Mother1 f
Another illustration of thia difference appean in the following pair:
(38) 1 W/ult flf'liYDU 'reatli"'111MGleardq 1 f
(39) 1Witat twlyDU 'reatlirrg 1l 1Ma\-tudq1 f
(38) ia addressed to someone named Macauley; (39) asks about an
author of that name. In (36), (37), and (39), the second intonation,
given as /( 1) 11l /, can be replaced by /(1) 11l I without altering the
efFects.
To the following distorted utterance-
(40) ·11U ltas a 1f~ebly 1 l 1growirrg doum on his lthrOGtll
-the most natural response seems to be "What's a feebly?" That is, the
intonational pattern seems to mark f~eblifu a noun. Replace feebly by
wart or mol1 and the sentence makes BeDSe. Or keep the words, but
change the intonation as follows:
(41) 1Heltas a 1/eebly growirrg 1doum1 l!on Iris 1thr0t111l
The last example includes a macrosegm.ent (the first) in which there
are four PU. instead of just three or two-two before the center, instead
of one. Another example of this is the BecODd one below. One can say,
quite colorleuly,
(42) •rw bun lwre jiw 'minutes 1l
but one can also emphasize the length of time slightly by riaiDg to
PL /'/ on the word jiw~
(43) •rw 6un lwre 'ji111 'mimllls1l
This last is distinct not only from (42) but also from two othen. If lOme-
one asks "Did you say si:t minutes?", you may reply ·
(44) 1 No1 f •r, b1111 ,., 'jiH mimdls l 1

And instead of (43) one can put even more emphuis on the length of
time, perhaps in protest or complaint, by saying

( 45) 11'w 6un lw11 ~·1 .,_,s l 1

PL / ' / 41
In (44), the word. five not only carries PL / 1/, but it ia a1lo IIIOI'e promi-
nent than anything before or after it; the clovtnwa.MI glide of pitch
begins withfive. In (43), on the other hand, the eenter ia the mira- ol
minulls, which is more prominent than anything uound it, even though
it ia spoken at the same pitch as the word fiDe immediately bef'ore it.
Though the center ol an intonation is by definition the 11101t promi-
nent syllable in the macro:.~gment, it need not carry the highest pitch.
Here is an example in which the center is lower in pitch than most of
the rest. Suppose a child starts to eat something that is inteodcd for
some other use. You may admonish him by saying
(46) 1 You lflon't 'eat tlult1 f
This should be distinguished sharply from the following two, which are
also admonitions, but with a different force: the child is supposed to eat
something, but preferably not "that'':
(47) 1You 1don'l eat 11hat1 f
{48) 1 You lflon'l eat 1tbat1 f
4.4. PL /'/.The highest of the four Pu occun somewhat lea fre-
quently than the other three, in a smaller variety of. intonations. The
following statements subsume most occurrences ol /'/ and will be
made as though there were no exceptions, though in fact there may be a
few. Any intonation which involves /'/ anywhere has /'/at the center.
Furthermore, no intonation with /'/ at the center occun unlca it
parallels an intonation identical throughout save for the presence ol
/'/wherever the fint has/'/. All intonations with / 1/ at the center
seem to be th\18 paralleled. Thus one can say {24), and this is matched
by an utterance which adds some sort of special or contrastive emphalia
to thne:
(49) 11 Wtmllo go 'tllere 1!
Similarly, question (31) is matched ~ a somewhat ~ IIUI'priled
query:
(50) •Is ,_. ,.,.. 'Bill'f
Compare also the following two-one straightforward, the CJCbao
surprised:
(51) •Is ~,. ,.,.. Billlf
(52) 1Is ~ flllllll Bill'f
 42 ENGLISH INTONATION

4.5. Adclitioul Eumpla. The four PLa and three TO. which have
been described CODBtitutc the stock ol intonational phonemes ol Engllsh.
The examples which follow illustrate further combinations ol these cle-
ments into whole intonations.
Note the following four ways ol intoning the word-sequence it'1 1m
o'el«k I Wlml to go luJtM:
(53) 1ft's 11m o' 1e!ock1 11I Watlt to go 1htnru1!
(54) 1ft's 1tm o' 1elock1 f 11 want to go IIUJtrre1!
(55) 1 ft's 11m o' 1clock1 f 11 want to go IIUJm81!

(56) 1ft's 1tm o' 1c/ock 1! 11 want to go •JUJmel!

The last ol these ROUnds most like "two sentences": this effect is indica-
tive ol the k.j.nd of meaning the intonations / 1 11!/ and / 1 1 11 !/ carry
for us. The other three give the impression ol a closer linkage between
the two parts.
There are many strings ol words which are delivered now as a single
macroscgment, now as two or more. In part, this depends on the tempo
of speech. For example, in normal rapid speech most ol us could say
either ol the following two, though in slightly more careful speech we
would much more often say the second:
(57) 1He luJs an 1office in tluJt bui/di~r~ 1 !
(58) 1He lull an 1office 1llin tluJt 1bui/ding1!
A more extensive breakdown would not be natural save under special
conditions. If we were dictating to someone who did not know short-
hand very well, we might say
(59) He !uu1 l Pat~ 1 l 1office 1 llin 1 l 1tluJt 1 l 1buildi~r~ 1 !
1

And in real exasperation we might even say

(60) 1He1 f 1/uu1 f 1an 1 f 1qffiee1 f 1in1 f 1tlrat1 f l

16uildilrf 1

An interesting intonational habit is observable in the reading ol

dialog, where a IICiltence like "Are you goU.,J" said Jan~ is olten rendered
as
(61)
Here the use ol PL / 1/ throughout the ~Jwker-illllieatitm is really not iD
itiC1f m.eaninsful, for if the direct quotation had ended with a different
ADDITIONAL EXAMPLES 43
PL, it would be this that would be carried through the speaker-indica-
tion. Compare the following four:
(62) 2Pm 1goingt! lsaid lBilllj
(63) 21 1guess so 2! 2replied 1Bill1 j
(64) 1Fire 3 f 1said 1Bi//8 j

(65) 1Do you 'mean it'j 'cried 'BerthD'f

A bit out of patience:

(66) 11 1told you so 1 j
{67) 21 •told you so 1 j
The first of the following is perhaps commoner than the second:
(68) 8So 2/ong 1 !
(69) •so 2long 2 j
Very businesslike:
(70) 1 Tell me about your 1/rinui'l

You're bound to anyway, so let's get it over with:

(71) 3 Tell me about your "frieruJI!

Choose between thr. alternatives:

Answer yes ·or no:

(73) 1 Do you want 1coffee 1 j 1or 'mi/k1 f

Philosophical assertion versus biological report:

(74) •The 'man in the 1street1l 'is 'my 1brother 1l

(75) 1The man in the 1street1l 'is my 1brother 1l

A threat, versus mere advice:

(76) 1You'd 1better do it 1f

{77) 'You'd better 1do it1l

44 ENGLISH INTONATION

Prediction verified., versus prediction wrong . but/ relinquished

reluctantly:
(78) •&e2 j 21 'thought so 1!
{79) i 21 1thought so 1j
1 Well 2

Regretful or doubtful:
(80) 2 He's 1gone1 !
{81} 1 We'll 1try 2!
(82) •rm 1going2 !

(83) 1 Nobody 1came1 !

Exasperated:
(84)
Tired, possibly disgusted:
(85) 1 want to go 2home 1!
1
(86) 1 want to go 1home 1!
2

Naturally, what did you expect! (d. example 4):

(87) it 1 j
1He 1bought

{88) He bought it 1 j
1 1

The first of these is more peremptory and tired than the second:

(89} 1She'll ex 2plain it to yor, 1 j

(90) 1She'll "' 2plain it to you•t
Of the following three, the first is neutral, the second emphatic, and
the third helpless:

{91) 1 What a lman 1!

(92) 1 What a lman 1!

(93) 1 What a lman1 !

Of the following two, the first signals aloofness on the part of the
speaker, while the second is friendly:

(94) 'Good lmorning 1! 1class 1!

(95) 'Good 'morning 1! lclass1f
 INTONATIONLESS SPEECH 45
The contrast in the next pair is somewhat similar:

(96) 1 You better not 1do thal1! 1Mac1!

(97) 1 You better not 1do that 2 ! 1Mac1 f
Just before being interrupted, or with what follows conveyed by
gestures rather than speech:
(98) 1/n 1/«111 'my 'friend'!
Miscellany:

(99) 1] 'nelltr 1116ard it before' l

(100) 10ne1 1 1two 1 1 1three 11 'fourl!

(101) •one•! 1two1 ! 1three1 ! 1/our\~

(102) 10ne 2j 1two 2j 1tlzreeZj 1four 1~
(103) 1That's 1all2!
(104) 2Well 2i 2boys will be 1IHI]s1!
(105) 1Pm 2gonna 1tell1 !

(106) 11 •won't 'tell 1!

(107) 1Right behind th6 1eight ball 1!

(108) 1 TM 'man's llm 2 j

(109) 1J•kunted 1! 11 didn't ~h 1 j

(110) 1M iss 1Jones 1 j 2this is Mister 1Smith 1!

(111) 1 Tke lrat 2 !

(112) 1 Time to get 'up' I 2son 1 !

{113) 1 Well 'look who's 4here1 l

(114) 1 That's 1o2kay 2!

(115) 1 Tke twortPI 1pig 1 ! 1is a 2noun 1!

4.6. Intonationlea Speech. Before our discussion of English intona-

tion closes, it must be added that there are certain types of speech which
lhow either a sharp reduction or a total loss of intqpational contrasts.
Speech in an especially high and narrow register, as under certain
types of emotional pressure, may compress the ordinary distances be-
tween the four PLs so as to render them difficult, perhaps impossible, to
keep apart. Very low-voiced conspiratorial speech in a low narrow
register can do the same. At the same time, the distinctions among the
three TCs may become blurred, but it is usually stili possible to tell
where the TCs occur.