(1982/2017). In A. Bunger (Ed.)., Special volume: Reissue of Innovations in Linguistic Education Volume 2.

IULC Working
Papers in Linguistics, Vol. 17, No. 3.
;':
PHONEMES AND FEATURES

by

Arnold M. Zwicky

Ohio State University

I. Introduction

In a general introduction to linguistics certain topics are both

indispensable and pedagogically difficult. I believe that the concepts
of the PHONEME and the FEATURE are indispensable to the phonology section
of such a course: the two constructs can be argued to be PSYCHOLOGICALLY
REAL, indeed, to be fundamental units in the mental organization of pho­
nological abilities; both constructs are ABSTRACT, not simply or directly
identifiable with actual physical events; these constructs figure promi­
nently in the statement of the REGULARITIES governing the phonological
side of any particular language; moreover, these regularities are LANGUAGE­
SPECIFIC; but they are phonetically NATURAL, explicable to a large extent
in terms of the conflicting needs of clarity and ease. The topics of
psychological reality, abstractness, regularity, language differences, and
naturalness are all important ones in an introductory linguistics course,
and all can be explored in a short treatment of phonemes and features.

Despite their centrality, the phoneme and feature concepts are

notoriously difficult for students to grasp. Undoubtedly their abstract
character has a lot to do with their difficulty; students in search of a
usable and memorizable explanation will not derive any practical benefit
from such definitions as the following, careful though they are:

A phoneme is a sound of a given language that native speakers

agree is just one segment, and which enables them to recognize
differences of meaning between words.
(Suzette Haden Elgin, What Is Linguistics?
2nd ed. (Prentice-Hall, 1979), p. 85)

DISTINCTIVE FEATURES. A set of universal, putatively innate,

phonetic and phonological properties by reference to which the
speech sounds of the world's languages are described ...
(Neil Smith and Deirdre Wilson, Modern Lin­
guistics (Indiana Univ. Press, 1979),
p. 275)

Instead, the student must be led to an appreciation of the concepts

through a series of examples. My own strategy is to lean very heavily on
material illustrating the psychological reality of phonemes and features,
so as to work against the students' tendency to see these constructs as
56 - Zwicky

something invented by linguists for their own arcane purposes, rather than
something relevant to ordinary human beings. In the two sections that
follow, I discuss briefly some pedagogical problems specific to phonemes/
features and then provide, for each, several exercises designed to illus­
trate psychological reality; sample answers to the exercises appear in an
appendix. This material was developed for a class of beginning graduate
students and advanced undergraduates, although some of it could be adapted
to classroom use in a lower-division 1 introduction to language• course
rather than an upper-division 'introduction to linguistics'.

I I. Phonemes

My focus here is on the classical phoneme and the principles governing

the distribution of allophones.

A special problem in introducing the phoneme concept has to do with

the sequencing of phonetics and phonology in an introductory course. If
phonetics leads, the student must first learn to attend to aspects of sound
that are normally outside of conscious reflection and then to disregard
these aspects in phonemic transcription. As a result, phonemics might seem
unreal. If the student first learns phonemic transcription for English,
then goes on to learn phonetic transcription and terminology, the student
may be encouraged to persist in English-based beliefs as to which sounds
are alike and different. I have found neither sequence entirely satisfac­
tory; the exercises below have been used with both orders of presentation.

Before these exercises are given out, the class has had a presentation
of contrast/complementary distribution/free variation in which the following
allophonic principles of American English (among others) were mentioned:

aspiration of voiceless stops at the beginning of a word or a

stressed syllable;

devoicing of liquids after voiceless stops;

affrication and retroflexion oft d before r;

h realized as� before y;

labialization of consonants before rounded vowels;

develarization of i before front vowels;

nasalization of vowel nuclei before nasals;

variation between plain and glottalized voiceless stops syllable­

finally, with? as a variant oft' in thi� position;

variation between aR and R when unstressed, for the resonants

r 1 n m;

i e u o as unstressed (free) variants of iY eY uw o w , with r as

a (free) variant of i word-finally.
 Zwicky - 57

With this background, it is possible to have the class analyze cases

of (a) phonemic hearing of other languages; (b) phonemic hearing of dialects
with different systems; (c) 'foreign accents' in reproducing phrases in
languages other than English; and (d) 'phonemic memory', manifested as a
faulty memory for actual pronunciations. All of these point to the psycho­
logical reality of the phoneme, as do the following: (e) the phenomenon
of categorial perception (which can be discussed in a later section of the
course on psycholinguistics); (f) the phonemic rather than phonetic nature
of alphabetic writing systems (which can be discussed in a later section
on writing systems); (g) slips of the tongue; (h) pig latins; and (i)
rhyming schemes. It is these last three cases I will illustrate here.
The exercises can of course be adapted if a different set of allophonic
processes has been introduced.

Exercise for (g).

According to the American Heritage Dictionary (1976 college edition),

a SPOONERISM is 'an unintentional transposition of sounds in spoken lan­
guage, as Let me sew you to your sheet for Let me show you to your seat�
[After Wi 11 iam A. Spooner (1844-1930), English clergyman, noted for such
slips.]'. Consider the following spoonerism:

INTENDED TARGET ACTUAL UTTERANCE

A pink stems tink spems

And an example of a related type of speech error, involving misplacement:

B find wit fide wint

Suppose that the phonetic transcriptions for the intended targets are as
fo 11 ows:

A [ph rQk stsmz]

B [faynd wrt]

Now answer the following three questions:

l. If it is SOUNDS that are transposed or misplaced, what would be

the phonetic transcriptions for the errors tink spems and fide wint?
(Remember that each symbol in a phonetic transcription represents a single
sound.)

2. What are the correct phonetic transcriptions for tink spems and
fide wint?

3- Given your answers in l and 2, how would you revise the American
Heritage Dictionary definition of SPOONERISM? Why?
58 - Zwi cky

Exercise for (h).

According to Robbins Burling (Man's Many Voices, Holt, Rinehart and Winston,
1970, p. 135),

At some time during childhood, most American children learn to

use "pig la tin". Instead of he will give it to me, a child will
learn to say iyhey ilway ivgay itey uwtey iymay. This distorted
form, of course, is accomplished by a simple phonological trans­
formation. Children teach the pattern to one another by some
such instruction as this: "take the first sound of each word
and put it on the end and then add a." This rule is reasonably
accurate though a linguist might want to refine it •..

Many people have learned a form of pig latin in which not only the first
consonant of a word, but a whole word-initial consonant cluster, is moved
to the end of the word, so that stop is transformed to opstay; however,
we are going to consider a form of pig latin that follows exactly the
rule Burling cities, so that stop is transformed to topsay. Examine the
three following phrases with respect to this pig latin:

(A) stop play [stap ph leY]

0

s
(B) try Hugh [t-•�ay ':?JUW ]

(C) low cool [fo W kWhU WX]

and answer the following questions:

l. If it is a consonant SOUND that moves to the end of the word and

has [eY] attached to it, what would be the phonetic transcriptions of the
transformed versions of (A)-(C) in this pig latin?

2. What are the correct phonetic transcriptions for the transformed

versions of (A)-(C) in this pig latin?

3. In light of your answers to and 2, how would you revise Burl ing's
rough rule for this pig latin? Why?

Exercise for (i).

Consider ordinary RHYME in English. According to Karl Beckson and Arthur

Ganz, Literary Terms: A Dictionary (Farrar, Straus and Giroux, 1975), p.
210:
 Zwicky - 59

The most usual English rhyme is variously called t1true", "full",

"perfect", "complete", or RIME SUFFISANTE. In it, the final
accented vowels of the rhyming words and all succeeding sounds
are identical, while preceding sounds are different, as in bake­
rake, heaven-seven.

The following rendition of the beginning of a familiar nursery rhyme counts

as full rhyme:
(A) ;f;clc' spr�t'

So does the following rendition of the beginning of a poem/song from Edith

Sitwell 's Facade:

16Yzi am sflr

And so does the following rendition of the beginning of another Facade

piece:
krayd
, oa n6Yvi , ,
b3::u w go w st
(C)
,
av mrstr bel6Ykar
, ,
oi ak'th'el S6kr
alcgro nrgro kh,
'
However, if (A) ended in [f;clc'] it would not count as a full rhyme; nor
would (B) if it ended in [sili]; nor would (C) if it ended in [sekl].
'
l. Why are (A)-(C) acceptable rhymes in English, and these not?
2. How would you revise Beckson and Ganz's definition of full rhyme?
Why?

Some remarks on the exercises. A substantial number of students will

give answers like the following to the final question in the exercises:
respectively, 'an unintentional transposition of sounds in spoken language,
with these sounds changed to fit their new places'; 'take the first sound
of each word and put it on the end and add a and then apply the allophonic
rules of English'; 'the final accented vowel of the rhyming words and all
succeeding sounds are identical, except for phonetic features due to
surrounding sounds'. A reference to the 'allophonic rules' of English is
precise but not always correct, due to the directionality of these pro­
cesses (e.g., there is a process devoicing 1 after p, but none voicing 1 0
60 - Zwicky

word-initially, but the latter process is what would be required for example
(A) in the pig latin exercise). A reference to contextually determined
changes in sounds is a great deal vaguer and fails to mention the language­
and dialect-particular character of these changes. All three answers treat
the contextual determination as fortuitously connected to the phenomena
at hand, indeed as an effect that wouldn't have to happen at all. Refer­
ring to phonemes gives a BETTER answer in each case. (This is one place
to introduce the lesson that some answers may be better than others, even
if they're all factually adequate, a lesson that some students--who object
to the importation of 'aesthetic' criteria into a 'scientific' enterprise-­
resist with passion.)
Titling the first of these exercises 'Sounds and Phonemes', or anything
with the word PHONEME in it, increases the percentage of 'right' answers,
but perhaps for the wrong reason.

Such exercises can be distributed over class discussions, homework,

and examinations. I usually save one for a review homework assignment or
an examination, where it can recall the student's mind to a type of reason­
ing previously used without asking for a mechanical replay of an earlier
answer.

Finally, I stress the importance of the 'why' in the final questions

of these exercises, if necessary assigning an actual point value to a brief
defense of the answer given. (This is one place to introduce the lesson
that a presentation of the evidence for some answer is usually more important
than the answer itself, again a lesson that some students--who object that
a linguistics course is not a course in thinking or writing--view with
distaste.)

111. Features
Here the stickiest point is the connection between the descriptors
of phonetics and the features of phonology. Most linguistics textbooks
develop separate vocabularies of descriptors and features, despite the
evident overlap between the two; some typographical distinction (initial
capitalization, italics, small caps) then has to bear the burden of
distinguishing, say, the feature 'Nasal' from the descriptor 'nasal'.
One text--Linguistics: An Introduction to Language and Communication, by
Adrian Akmajian, Richard A. Demers, and Robert M. Harnish (MIT Press,
1979)--a text with several admirable chapters, moves from phonetic de­
scriptors to phonological features within the space of a few pages in a
single chapter, thereby confusing all but the brightest students and
alienating all but the most passive.

One motivation for this double vocabulary is probably that descriptors

are believed to be phonetic, anatomic, physiological (or perhaps acoustic),
while features are believed to be phonological, mental, abstract. I see
 Zwicky - 61

no reason to characterize the distinction in these terms. Surely the de­

scriptors are abstract also: there is nothing anatomically in common to
the many physical gestures that result in stop consonants; the tongue-
root advancement associated with phonetically 'wide' or 'tense' vowels
results in some raising and fronting of the tongue body, but phonetically
wide vowels are not thereby classified also as high and front; the acoustic
activity during a voiceless stop consonant is indistinguishable from an
equally long pause; all the suprasegmental descriptors are inherently
relative; 'there is no agreed physical measurement corresponding to sylla­
bicity. But there is no doubt that segments can be described phonetically
as being syllabic (100 percent) or nonsyllabic (0 percent)', according to
Peter Ladefoged's Course in Phonetics (Harcourt Brace Jovanovich, 1975),
p. 267; and so on.
The question is then whether there should be two abstract categoriza­
tions or only one. As a PEDAGOGICAL question, the answer ought to be that
we would accept more than one abstract categorization only for the strongest
of (pedagogical) reasons. This is just the sort of situation in which
introductory texts do well to oversimplify; there are things it is better
to conceal for a while, lest the students sink into a quicksand of conceptual
and terminological refinements. Even as a THEORETICAL question, it seems
to me, the answer ought to be that we would accept more than one abstract
categorization only for the strongest of (theoretical) reasons. For
theoretical purposes, we need a vocabulary (applicable to all languages)
for naming natural classes of segments and natural relationships among
segments and for describing the phonetic distinctions between phonemes;
descriptors that serve none of these functions have no place in linguistic
phonetics, and if we are fortunate a single set of descriptors will suffice
for all of these functions.

My approach to descriptors and features in an introductory class is

therefore unified, with a single vocabulary for 'phonetic properties'.
Two types of exercises help the students gain some facility with this vo­
cabulary. The first type focusses on phonetic properties and NATURAL
CLASSES/RELATIONSHIPS, the second on phonetic properties and PHONEMIC
DISTINCTIONS.
In exercises of the first type, the student is provided with positive
instances of some phenomenon (and usually with negative instances as well)
and is asked to supply the appropriate generalization. The form of such
exercises is introduced in my initial discussion of phonetic properties,
as in the text below.

Consider the statement in (1) below. How can we replace the second
part of the statement (the part after the dots) so as to make it GENERAL,
not merely al ist of words that do one thing as opposed to a list of
words that do something else?
62 - Zwicky

(1) The English indefinite article is an rather than a ...

before the words ermine, easy, old, Australian, honor,
enormous, ivy, ounce, added, awesome, herb, approximate,
and early (but is a before useful, history, radio,
performer, European, dish, washer, fertile, and night).

The generalization has to do with the type of sound that begins the word
following the indefinite article: the first list consists entirely of
words beginning with 'vowel' sounds, the second consists entirely of words
beginning with 'consonant' sounds. (You should be able to give a convincing
argument from these examples that it is SOUNDS and not LETTERS that are
relevant.)
A slight complication is introduced here by the fact that ermine,
herb, and early are in the an list, while radio is in the a list. For
most American English speakers, the sound at the beginning of ermine is
articulated just like the sound at the beginning of radio; for these speakers,
ermine does not PHYSICALLY begin with a vowel followed by r. Yet the r at
the beginning of ermine, like the r in the middle of bird and the r at the
end of butter, counts as making a syllable, while the r at the beginning of
radio does not. Stirring has an ermine-type r, and two syllables, while
string has a radio-type r, and only one syllable. Now normally it is the
function of VOWELS to make syllables, so that ermine, stirring, butter, and
bird all have the consonant r 'acting like' a vowel; many English speakers
also have an 1 acting like a vowel in kettle, an n acting like a vowel in
kitten, and an m acting like a vowel in bottom. What all this adds up to
is that the phonetic classification VOWEL/CONSONANT is not quite what we
want in describing what's going on in (1). Instead, we want a distinction
between sounds that make syllables and those that do not--between SYLLABICS
and NONSYLLABICS. The generalization that completes the first part of (l)
correctly is

(l') before syllabics.

Further complete-the-generalization problems introduce such properties

as LABIAL, CORONAL, and SIBILANT, while the STOP/CONTINUANT and OBSTRUENT/
SONORANT distinctions are described and briefly justified without exempl ifi­
cation in a problem. (Properties like ALVEOLAR, LIQUID, FRICATIVE, APPROXI­
MANT, VOICELESS/VOICED, and NASAL, which distinguish English phonemes, have
already been introduced.) At this point the students are given a series of
exercises of the complete-the-generalization form, arranged roughly in
order of complexity. Some examples follow.
 Zwicky - 63

(8) At the beginning of a word before 1 or r, the only fricatives

permissible in English are
the ones in shred, slop, flicker, frazzled, slide, frog, thread
(so that ,',zlop, 1<vlicker, and ,�vrog are not possible words, nor
is thread if pronounced with the initial consonant of this
rather than the initial consonant of think).
(11) Some American English speakers have s e rather than � ...
in rash, has, gather, bath, raft, gas, castle, jazz (but have �
in fat, gap, stack, batch).

(13) Some American English speakers (largely in the Midwest and

South) pronounce s as r ...

in then, Kenney, pen, Bengals, gem, Mencken, Remington, and

temperature (while maintaining s in met, wedding, beggar, best,
gel, merry, kept, and mesh).

(18) English speakers have slightly labialized variants of word­

initial consonants ...

in toot, pooch, boat, known, cook, good, so, tall, fought, Shawn,
pull (but not in team, pet, bait, name, father, give, say, Cal,
fat, sham, pill, cut, birth).

(24) Especially before words beginning with consonants, many Americans

sometimes do not pronounce word-final
consonants in six, leads, past, gift, act, meant, mend, hold
(though they do pronounce the word-final consonants in branch,
Welsh, mask, filth, and lisp).

(31) Most speakers of English do not pronounce ...

a word-final bin limb and thumb or a word-final g in wing and

rung (though they do pronounce the word-final consonants in
limp, thump, wink, drunk, lend, bond, rant, branch, lab, and
rag).

(32) Some Southern Ohio and Indiana speakers replace ...

the vowel of met by the vowel of mate in special, measure,
pleasure, mesh, precious (but not in mess, fettle, retch,
methyl, pestle, wed) and the vowel of mitt by the vowel of meet
in commission, fish, partition, elision, derision (but not in
miss, fiddle, midge, nifty, whistle, sit).
64 - Zwicky

A few comments on this sort of exercise. The phenomena illustrated

include dialect variants (some of which can be referred to again in a
later discussion of historical change), casual speech variants (some of
which can be used in sociolinguistics and/or in historical change),
ordinary allophonic variants, phonologically conditioned morphophonemic
variants, and constraints on phoneme combinations. Consequently, no
framing brackets of any sort appear in the exercises.

The material to be replaced has forms cited in ordinary English spel­

ling. This is deliberate. These exercises give students additional
practice in phonemic transcription, at which they are probably shaky.

It may be necessary to give some explicit advice about solving special­

ly designed problems like these: (a) these problems are so designed that
there IS a general solution (a right answer covers all the cases, and there
are no 'exceptions'); (b) all the information needed to get a solution is
available in the problem statement (so that if there is no way to tell
what pitch level particular words are spoken on, say, then this factor
cannot be relevant to the answer); (c) such problems are ordinarily designed
to have strikingly simple answers (so that if your proposed answer has
several clauses in it, or rivals the problem statement itself in length,
there is probably a better answer); (d) if there is negative evidence
given, it is important (the devisers of such problems don't throw in whole
categories of facts just for fun); (e) if your current hypothesis begins
to look unpromising, try another, remembering that sometimes you might
want to go back to an earlier idea.

After students have had a reasonable amount of experience with exercises

like those above, it is possible to expand the range of exercises to include
types that must be presented in transcription: data in languages other than
English (indeed, standard phonemics problems can usually be recast in the
format of (1) above), data from the acquisition of English by young children,
and data from historical change.

I turn now to exercises focussed on phonetic properties as phoneme di­

scriminators. First, a paragraph of introductory text.

The properties that define natural classes--for instance, vo1c1ng,

nasality, continuancy, and point of articulation for consonants and
height, frontness, and rounding for vowels--often act as independent
elements of linguistic structure, so that individual sounds or phonemes
must be viewed as 'broken down' into an assemblage of these properties.
The English phoneme /p/ would then be seen as an assemblage of the
properties VOICELESS, LABIAL, and STOP, therefore as distinguished from
/b/ and /m/ as being voiceless rather than voiced, from /ti and /k/ by
being labial rather than alveolar or velar, from /f/ by being a stop
rather than a continuant, and from other English phonemes by differences
in two or more of these properties.
 Zwicky - 65

Exercise A.

Below is a list of slips of the tongue (from the collection in Victoria

Fromkin's Speech Errors as Linguistic Evidence (Mouton, 1973)). Using
appropriate phonetic terminology, describe what has happened in each of
these errors. Do not merely say, "The speaker said m instead of b and said
d instead of n," and the like, but look for some REASON why these particular
errors should have been made. Hint: there is a sense in which all these
errors are of the same type. Further hint: these errors are similar, in
a way, to the common type of error known as the SPOONERISM (my queer dean
FOR my dear queen; you have hissed my mystery lectures FOR you have missed
my history lectures; stretch and piss FOR stress and pitch).

INTENDED TARGET ACTUAL UTTERANCE

a. Cedars of Lebanon Cedars of Lemadon

b. Terry and Ju 1 ia Derry and Chulia /culya/
c. big and fat pig and vat
d. clear blue sky glear plue sky
e. pedestrian tebestrian
f. scatterbrain spattergrain

Exercise B.

According to Thrall, Hibbard, and Holman, A Handbook to Literature

(Odyssey Press, revised ed., 1970), a pun is
A play on words based on the similarity of sound between
two words with different meanings. An example is Thomas Hood's:
"They went and told the sexton and the sexton tolled the bell."
Their example is an instance of what I will call a PERFECT PUN, a play
on words based on the phonological identity (or HOMOPHONY) of two words with
different meanings. Below are some examples of perfect puns from John S.
Crosbie's Dictionary of Puns (Harmony Books, 1977):
(l) bound

The zoo's kangaroo lacks zip: He is frequently discovered

out of bounds.

(2) clap

VD is nothing to clap about.

66 - Zwicky

(3) prone
We are all prone to die.
(4) worn

"Is anything worn under your kilt?"

"No, it's a 11 in working order."

Very often, however, puns are less than perfect. Sometimes the difference
between a syllabic and a nonsyllabic consonant is disregarded, as in these
examples from Crosbie:

(5) mower

If you can't afford a power lawnmower, then mower

power to you.

(6) wire

As the tightrope walker asked himself, "Wire we here?"

And sometimes the difference has to do with where words are divided:
(7) stripper
He was afraid to go out with the burlesque queen because
he didn't know how to stripper.
(8) tall

It is better to have loved a short girl than never to have

loved a ta11 .

And sometimes the difference has to do with whether there is a consonant or

nothing at a11.
(9) cling

Wrestling is the sport of clings.

( l O) sequel
A wolf is a man who treats all women as sequels.

And sometimes--especially when the original expression is a well-known ex­

pression--the difference is very great:
 Zwicky - 67

(11 ) bovine

There once was a tolerant cow who stood for absolutely

anything her favorite bull tried to get away with. She
mooed, 11 Too err is human, to forgive, bovine. 11
(12) Persian

One man's Mede is another man's Persian.

Usually, however, in imperfect puns the difference is quite small, as in:

(13) clothe

Sign by gate to nudist colony: "Come in. We Are Never

Clothed."

Consider the examples below (also from Crosbie): for each, identify the
punning word in the example (clothed in (13)) and the word it puns on
(closed in (13)); then identify the distinct phonemes that are matched in
the pun (here, o and z), and say what phonetic properties distinguish these
phonemes (here, a difference in point of articulation, interdental versus
al veo1 a r).
(14) crab
Once there was a girl
Who kept fishing for a pearl,
But her chances were drab for it-­
Until she made a crab for it.

(15) fever
01 iver Wendell Holmes, Sr., was a physician as well as an
author and lecturer. He is said to have remarked of his
medical career that he was grateful for small fevers.

(l 6) money
Sign outside an amusement park: "Children under 14 must be
accompanied by money and daddy."

(17) radish
Health food can give you a radish complexion.
68 - Zwicky

(18) choker

Mrs. reported to Mr.: "It says here that a man on the next
block throttled his mother-in-law yesterday," "Hmmm," mused
Mr., "sounds to me like he was a practical choker."
(19) bleach

When the blonde he married faded into brunette, he sued for

bleach of promise.
(20) curl
Labia majora: the curly gates.
(21) sicken

"Aha!" cried Sherlock Holmes, "the plot sickens!"

(22) bottle
When it came to drinking, comedian W. C. Fields was a veteran
who suffered from bottle fatigue.
(23) sylph
Why is it that many a woman with a sylphlike figure insists
on keeping it to her sylph?
(24) clash

The late poet J. Ogden Nash

Always made of his English a hash.
When asked where it led
He flippantly said,
"It gives it a great touch of clash."
(25) sty
For many a farmer the price of pork has created a gold mine in
the sty.
Sometimes imperfect puns involve differences in two or even three places, as
in the following examples. Analyze these as you did (14)-(25), treating
each corresponding pair of distinct phonemes separately.

(26) rabbit
Fast, speedy (as in Rabbit Transit).
 Zwicky - 69

(27) crass
. . . It is a p1at itude
That only a halter
Can alter
The middlecrass assitude.
(28) breeze
In Chicago, every prospect breezes.

(29) morse 1

What foods these morsels be!

(30) mutton
Lamb stew is much ado about mutton.

Exercise C.

Most familiar verse in English uses FULL RHYME: the peak of the last
accented syllable of a line, plus everything that follows that peak, is
identical to the peak of the last accented syllable of a matching line,
plus everything that follows it--
(1) Lizzie Borden took an axe
And gave her mother forty whacks.
(American verse of unknown authorship)

(2) I'm just a poor wayfaring stranger,

A-trav'ling through this world of woe;
But there's no sickness, toil nor danger
In that bright world to which I go.
('Wayfaring Stranger', #97 in Alan Lomax,
Folk Song U.S.A., New American Library,
1975)
But some verse--especially traditional English ballads, nursery rhymes,
blues lyrics, and the lyrics of rock music--frequently uses HALF RHYME, in
which the matched parts are not entirely identical. In many such cases, a
consonant counts as rhyming with a cluster including that consonant--

(3) [n-nd]

Well lookin' for a woman

an' a well oh man
is just lookin'for a needle
that is lost in the sand
(Dylan, 'Just Allow Me One More Chance')
70 - Zwicky

(4) [d-nd]

She left one too many a boy behind

He committed suicide
(Dylan, 'Gypsy Lou')

or a word ending in a vowel counts as rhyming with one ending in that vowel
plus some consonant--
(5) [o-od]
stood a wondering which way to go,
lit a cigarette on a parking meter
And walked on down the road.
(Dylan, 'Talkin'World War 111 Blues')

In other cases, distinct consonants count as rhyming, or distinct vowels

count as rhyming. In each of the examples below you are to pick out the
distinct phonemes that are counted as rhyming in the italicized word
(remember that material BEFORE the peak of the last accented syllable will
of course be different, as in the full rhymes axe-whacks and stranger-danger
and the half rhymes man-sand, behind-suicide, and go-road), and you are to
say what phonetic properties distinguish those matched but different phonemes.

(6) The things that sit and wait for you

To stumble in the dark
Will take the cobwebs from your eyes
And plant them in your heart.
(Byrd, 'The Elephant at the Door')

(7) Going where the orange sun has never died,

And your swirling, marble eyes shine laughing,
Burning blue the light.
(Lamm, 'Fancy Co 1 ours')

(8) Farewell to Greer County where blizzards arise,

Where the sun never sinks and the flea never dies,
And the wind never ceases but always remains
Till it starves us all out on our government claims.
('Starving to Death on a Government Claim',
#70 in Lomax)
(9) [This is a full rhyme in some dialects.]

Some of us were willing, while others they were not.

For to work on jams on Sunday they did not think they'd ought.
('The Jam on Gerry's Rocks', #50 in Lomax)
 Zwi cky - 71

(10) Tying faith between our teeth

Sleeping in that old abandoned beach house
Getting wasted in the heat
(Springstee;,;-,Backstreets')
(11) Well the technical manual 1 s busy
She's not going to fix it up too easy.
(Mitchell, 'Electricit7}

(12) Old Reilly stole a stall ion

But they caught him and brought him back
And they laid him down on the jail house ground
With an iron chain around his neck.
(Dylan, 'Seven Curses'T

( 13) Git out the way, ol' Dan Tucker,

You too late to git yo' supper.
('Old Dan Tucker', #27 in Lomax)

( 14) Oh, yes, I am wise

but it's wisdom born of pain,
Yes, I paid the price
but look how much I gained.
(Reddy, I I am Woman')

(15) I '11 remember Frank Lloyd Wright.

All of the nights we'd harmonize till dawn.
I never laughed so long.
(Simon, 'SoLong, Frank Lloyd Wright')

(16) My experience was limited and underfed,

You were talking while I hid
To the one who was the father of your kid.
(Dylan, 'Love is Just a Four Letter Word')
(17) Like dust in the wind you're gone forever
You're wind-blown leaves you're a change in the weather
(Taylor, 'Something's Wrong')

(18) Love my wife, love my baby,

Love my biscuits soppedii, gravy
('Blackeyed Susie', #29 in Lomax)
72 - Zwicky

All the types of exercises I have illustrated are consistent with a

number of different ways of treating phonemes and features. They are neutral
with respect to the question of whether 'phonemic representation' is to be
treated as essentially identical to 'morphophonemic underlying form' in an
introductory linguistics course (not my ordinary practice) and with respect
to the question of whether distinctive features are binary or not (the
system being developed in the material above LOOKS nonbinary but can be
fashioned into a binary system with little trouble). They can be used with
various formalisms, or in a setting where students are instructed to give
answers in ordinary but precise English, using the technical terms of lin­
guistics where appropriate (my own preference, especially since this
approach allows me to finesse the issues of redundant and unspecified fea­
tures, two technical matters that generate a surprising amount of anxiety
in students who want to get everything right). On the minus side, they
present special difficulties to the non-native speaker of English, and must
be revised depending upon the dialect make-up of the class. But then it is
hard to think of a way of introducing phonology that is free of both of
these drawbacks.

Appendix: Sample Answers

I I. Phonemes

Exercise for (g):

1. [trqk sph smz]

[fa.yd wrnt]

2. [t h rqk spcmz]
[fayd wrnt]

3, An unintentional transposition of phonemes in spoken language.

If we say that it is SOUNDS that are transposed, then we pre­
dict incorrect sequences of sounds in actual pronunciations;
but if we say that it is PHONEMES that are transposed, then the
correct allophones of these phonemes are automatically predicted.

Exercise for (h):

1. [tapseY leYpheY]
0

[rayt�e
•
Y qu wgeY]
0

[o WreY UWXkwheY]
 Zwicky - 73

2. [thapseY leYpeY]
[rayteY yu wheY)
[o wleY u w �keY)

3, Take the first phoneme of each word and put it on the end and
then add /e/. If we say that the SOUNDS are involved, then
we predict incorrect sounds both at the beginnings of the pig
latin words and before their final [eY); but if we say that
PHONEMES are involved, then the correct allophones are auto­
matically predicted in both places.

Exercise for (i):

l. In (A)-(C) the matched sounds are allophones of the same phoneme

or phoneme combination: /t/, /i/, and /ar/, respectively. But
[k'] and [t'] are allophones of different phonemes, /k/ and /t/;
and stressed [r] and [{] are allophones of different phonemes,
/r/ and /i/; and [r]
' and [l) are allophones of different phonemes,
/r/ and /1/.

2. All succeeding phonemes are identical. If we required that suc­

ceeding SOUNDS be identical, then different sounds in free varia­
tion with one another wouldn't count as rhyming, any more than
different sounds that are allophones of different phonemes do;
they are all different sounds. But this is incorrect. If we
require that succeeding PHONEMES be identical, then we predict
(correctly) that different sounds in free variation count as the
'same sound' for the purposes of rhyme.

111. Features

(8) voiceless.

(l l) before fricatives.

(13) before nasal consonants.

(18) before rounded vowels.

(24) alveolar consonants.

(31) a word-final peripheral [or noncoronal] voiced stop after

a nasal.

(32) ... nonlow front lax vowels by their tense counterparts before
posterior [or nonanterior, or more specifically, alveopalatal]
fricatives.
74 - Zwicky

Exercise A: In each case a single phonetic property has been trans­

posed between one phoneme and another: in example a,
nasality appears with the earlier bilabial consonant in
Lebanon instead of the later alveolar one; in examples
b and d, voicing appears with a word-initial consonant
in an earlier word instead of a later one, and in example
c, with a word-initial consonant on a later word instead
of an earlier one; and in examples e and f, the points of
articulation for two consonants in a word have been
exchanged. In every case all other phonetic properties
of the consonants affected remain unchanged.

Exercise B:

(14) crab punning on grab; k and g; voicing (voiceless versus voiced).

(15) fevers punning on favors; i and e; height (high versus mid).

(16) money punning on mummy; n and m; point of articulation (bilabial

versus alveolar).

(17) radish punning on reddish; re and c; height (low versus mid).

(18) choker punning on joker; c and J; voicing (voiceless versus voiced).

(19) bleach punning on breach; 1 and r; point of articulation (alveolar

versus postalveolar), tongue configuration (lateral versus retro­
flex).

(20) curly punning on pearly; k and p; point of articulation (velar

versus bilabial).

(21) sickens punning on thickens;s and 0; point of articulation

(alveolar versus (inter)dental).

(22) bottle punning on battle; a and re; frontness (back versus front).

(23) sylph punning on self; r and c; height (high versus mid).

(24) clash punning on class;� ands; point of articulation ((alveo)

palatal versus alveolar).

(25) sty punning on sky; t and k; point of articulation (alveolar

versus velar).

(26) rabbit punning on rapid; b and p, t and d; voicing (voiced versus

voiceless), voicing (voiceless versus voiced)--cf. Exercise A.
 Zwicky - 75

(27) middlecrass assitude punning on middleclass attitude; r and 1,

s and t; point of articulation (but see (19) above), manner of
articulation (fricative versus stop).
(28) breezes punning on pleases; band p, r and l; vo1c1ng (voiced
versus voiceless), point of articulation (but see (19) above).

(29) foods ... morsels punning on fools ... mortals; d and 1, s and
t; manner of articulation (stop versus liquid), manner of
articulation (fricative versus stop).
(30) mutton punning on nothing; m and n, t and e, n and Q; point of
articulation (bilabial versus alveolar), point and manner of
articulation (alveolar stop versus (inter)dental fricative),
point of articulation (alveolar versus velar).

Exercise C:

(6) k and t, velar versus alveolar.

(7) a and t, voiced versus voiceless.

(8) n and m, alveolar versus bilabial.

(9) a and �, unrounded versus rounded.
(10) 8 and t, (inter)dental fricative versus alveolar stop.
(11) I and i, lax versus tense.

(12) re and e, low versus mid.

(13) k and p, velar versus bilabial.

(14) z ands, voiced versus voiceless.

(15) n and Q, alveolar versus velar.

(16) e and r, mid versus high.
(17) v and o, labiodental versus (inter)dental.
(18) b and v, bilabial stop versus labiodental fricative.
76 - Zwicky

FOOTNOTE

*The material presented here has benefited enormously from the comments
and criticisms of Linguistics 601 students at Ohio State from 1972 on, and
especially from the advice of my teaching assistants in this course. This
paper was completed at the Center for Advanced Study in the Behavioral
Sciences. I am grateful for financial support provided by the Spencer
Foundation and for sabbatical leave from the Ohio State University.