Chapter 1

Introduction

I.I
WHY STUDY WORDS?
Imagine a life wilhOut words! Trappist monks opt for it But most of us would not give up word'i for
anything. E\'ery day we uuer thousands and thousands of words. Communicating our joys. fears. opinions.
fantasies. wishes. requests. demands, feelings-and the occasional threat or insult-is a very important
aspect of being human. The air is always 1hick with our verbal emissions. There are so many things we
want to tell theworld. Some of them are imponant, some of [Link] are not 8Ul we talk anyway-even when
we know that what we are saying is totally unimportant. We love chitchat and find silent encounters
awkward, or even oppressive. A life without words wouJd be a horrendous privation.
It is a clicht tosay that words and language are probably humankind's most valuable single possession. It
is language that sets us apart from our [Link] relatives. the great primates.() would imagine that
many a chimp or gorilla would give an arm and a leg for a few words-but we will probably never know
because they cannot tell us.) Yet. surprisingly. most of us take words (and more generaJly language) for
gran1ed. We cannot discuss words with anything like the competence with which we can discuss fashion.
films or foo1ball.
We should not take words for granted. They are 100 important This book is intended 10 make explicit
some of the things that we know subconsciously about words. It is a linguistic introduction to the nature and
structure of English words. It addresses the question 'what sons of things do people need to know about
English words in order to use them in speech?' It is intended to increase the degree of sophistication with
which you think about words. It is designed to give you a theoretical grasp of English word-fonnation. the
sources of English vocabulary and the way in which we store and retrie\'e words from the mind.
I hope a desirable side effect of working through £11gUsh Words will be the enrichment of your
vocabulary. Thisbook will help10increase. ina very practical [Link] of the relationship
be1ween words. You will be equipped with the tools you need to work out the meanings of unfamiliar
words and to see in a new light the underlying structural pauerns in many familiar \\<Ords which you have
not previously Slopped 10 lhink aboul analy1ically.
For the student of language. word.i; are a very rewardingobject of study. An understanding of the nature
of words provides us with a key that opens the door 10 an understanding of important aspects of the nature
of language in general. Words give us a panoramic view of the entire field of lingui�tics because they
impinge on every aspect of language structure. This book stresses the ramifications of the fact tha1 words
are complex and multi-face1ed entities whose struc1ure and use interacts with the other modules of the
grammar
2 ENGLISH WORDS

such as PHONOLOGY. the study of how sounds are used to represent words in speech, SYJ\"TAX. the
study of sentence structure, and SEMANTICS. the study of meaning in language.
In order to use even a very simple word, such as frog, we need to access various types of infonnation
from the word-store which we all carry around with us in the MENTAL LEXlCON or DICTIONARY that
is mcked away in the mind. We need to know:

fl.I)
(i) itsshape. i.e. its PHONOLOGICAL REPRESEJ\t'fATION/frg/ which enables us to pronounce it. and
i1s ORTI-IOGRAPHIC REPRESBffATION frog. if we arc li1erate and know how to spell it (sec the
Key 10 symbols usedon page xix):
(ii) its grammatical propenies. e.g. it is a noun and it is countable-so you can have onc/rog and two/rogs;
its meanin .

Bui words tend not to wear their meaning on their sleeve. Nonnally, there is nothing aboul 1he form of
words that would enable anyone to work out Lhei.r meaning. Thus. the fact that frog refers to one of these
simply has to be listed in the lexicon and committed to memory by brute force. For the relationship between
a LINGUISTIC SIGN like this word and its meaning is ARBITRARY. 01her languages use differenl words
to refer to this small tailless amphibian. In French it is called (la) grenouille. In Malay 1hey caJI it katak
and in Swahili dmra. None of these words is more suited than the others to the job of refening 10 this
small reptile.
And of course. within a panicular language. any panicuJar pronunciation can be associated with any
meaning. So long as speakers accept that sound-meaning association. they ha,•e a kosher word. For
instance. conwmience originally meant 'suitability· or ·commodiousness· but in the middle of the nineteenlh
century a new meaning of ·toilet" was assigned 10 it and people began to talk of ·a public convenience·. 1n
the early 1960s the word acquired the additional new meaning of ·easy to use. designed for hassle-free use·
as in com·eniem:e food.
We are the masters. Words are our servants. \Ve can make them mean whatever we want them to
mean. Humpty Dumpty had all this worked out. The only thing missing from his analysis is the social
dimension. Any arbitrary meaning assigned to a word needs to be accepted by the speech community
which uses the language. Obviously. language would not be much use as a means of communication if
each individual language user assigned a priva1e meaning 10 each word which other users of the language
did not recognise. Apan from that. ii is instructive to listen in on the lesson on the nature of language
1hat Humpty Dumpty gave to Alice (seeoverleaf).
Let us now consider one [Link] example. All competent speakers of English know that you can add -s
to a noun to indicate that it refers to more than one entit·y. So. you say cat when referring to one and cats
if there is more than one. If youencountered in the blank in [ I.2a) an unfamiliar word like splet (which I
have [Link] made up), you would automatically know from the context that it must have the plural fonn
splets in this position since it is specified as plural [Link]. you would know1hat the pluralof spiel
must be splets (rather than sp/e1re11by analogy to children or sple,i by analogy to stimuU). You know that
the majority of nouns form their plural by adding the regular plural suffix or ending -s. You always add-
sunless express instructions are given to do otherwise. 'There is no need to memorise separa1ely the plural
form of most nouns. All we need i.s to know the rule that says 'add -s for plurar, So. without any
hesitation. you suffix -s to obtain the plural fonn spiels in[1.2b):
11.21
We pUIall thebi�• on lhe table.
b. We put all lhe bi2 splets on the table.

The study of word-fonnation and word-structure is called MORPHOLOGY. Morphological theory

provides a general theory of word-structure in all the languages of the world. Its task is to characterise
the kinds of Lhings Lhat speakers need 10 know about the structure of the words of their language in
order to be able lO lL� them to produce and to undersLand speech.
\Ve will see that in order to use language. speakers need to have two types of morphologicaJ knowledge.
First, they need to be able to analyse existing words (e.g. they must be able to tell that frogs contains frog
plus -s for plural). Usually. if we know the meanings of the elements that a word contains. it is possible
to determine the meaning of the entire word once we have worked out how the various elements relate
toeach other. For [Link]. if we examine a word like nutcracker we find that it is made up of two
words, namely the noun nut and the noun cracker. Funhennore, we see that the latter word, cracker is
divisible into the verb crack and anOlher meaningful element -er (roughly meaning 'an inslrument used
to do X"), which, however. is not a word in its own right. Numerous other words are fonned using this
panem of combining words(and smaJler meaningful elements) as seen in [1.3):

11.31
(tea)Noun--[suain-er])Noun
r1awn)Noun-[mow-er])Noon
[canJNoun--(open-erJ)Noun

Given the frame [[_]Noun-{_erJ) Noun, we can fill in different words with the
appropriate propenies and get another compound word (i.e. a word containing at least two words). Try
this frame out yourself. Find 1wo moresimilar examples of compound words formed using this pattern.
Second. speakers need to be able to work out the meanings of novel words consm1cted using the
word-building element's and standard word-construction rules of the language. Probably we all know and
use more words than are listed in dictionaries. We can construct and analyse the structw-e and meaning
of old wordsas well as new [Link], although many words must be listed in the dictionary and
memorised, listing every word in the dictionary is nOl necessary. If a word is formed following general
principles. it may be more efficient to reconstitute it from its constituent elements as the need arises
rather than permanently commit it tomemory. When people make upnew words usingexisting wordsand
wordfonning elements, we understand them with ease-providing we know what theelements:they use to
form those "'ords mean and providing the word-forming rules that they employ are familiar. This ability
is oneof the thingsexplored in morphological investigations.
1n an average week, we are likely to encounter a couple of unfamiliar words, We might reach for a
dic1ionary and look them [Link] them may belisted but others might be too newor tooephemeral to
have found their way into any dictionary. ln such an e\'em, we rely on our morphological knowledge to
tease out their meanings. If you heard someone describe their panner ai ·a great list maker and a ticker-
off. you would instantly know what son of person the [Link] was-although you almost cenainly have
never encountered the word ticke.r-offbefore. And it iscenainly not listed inanydictionary. The -er
endinghereha.i
4 ENGLISH WORDS

the meaning of ·someone who does whatever the verb means·. Given the verb [Link]. a ticker-off must be
a person who [Link]. if you know wha1es1ablished words like handful, cupful and spoonful
mean. you are also able to figure ou1 the meanings of novel words like fouutain-penful (as in a fou11tai11-
pe11/11/ of ink) or ho\'l!rcraftful (as in hm·ercmftful after lwvercraftful of English shoppers retume.d
from Calais loaded do,rn with cigareltes, [Link] and plank). Virtually any noun denming a container can
have -Jul
added to it in order to indica1e that it is 'full of something·.
To take another example. a number of words ending in -[Link], many of which have come into use in
recent yeart, refer to people who discriminme against, or hold negative views about. certain less
powerful subgroups [Link] who knowswhal racis1and sexist [Link] the right
context should have no difficulty in undenaanding the nature of discrimination perpe1rated by people
who are described using the novel words agei.s1. [Link] and speechist. Ageism [Link] groundsof
(old) age
-for ins:[Link], denying employment to people over the age of 60: sizism is discrimination (usually
against fat people) on grounds of size and speechism is discrimination against people with speech
impediments like stu1tering.
Did you notice how I exploited your taci1 knowledge of the fact that words ending in -isl and -ism
complement each other? You were glad 10 accept ageism, si.:ism and speechism because you know that
corresponding to an adjective ending in -ist there will nonnally be a noun ending in -[Link] is important.
It shows that you know that certain word-forming bi1s go together-and others do not. I suspect that you
would reject pulative words like *ageme111, *si:emeHI and *speecl1me111.(An asterisk is used
oonven1ionally to indicate that a form is disallowed.) In word-formation it is not a case of anything goes.
A challenging question which morphology addresses is, 'how do speakers know which non-occurring or
non-established words are permissible and which ones are not?' \Vhy are the wonis fo1111tai11pe11f11I,
ho1·ercraftful and speecl,ist allowed while *ageme111, *speechment and *si::.eme111 are not?
Morphological theory provides a general theory of wordfonnation applicable to any language but, as
mentioned earlier, this book focuses on word-formation in English. Its objective is to provide a description
of English words designed 10 make explicit the variou.� things speakers: know, albeit in an unconscious
manner, about English words. The emphasis will be on the description of English words rather than the
elaboration of morphological theory. So, data and facts about English words are brought to the fore and the
theore1ical and methodological issues are kepi in the background for 1he most part. The use of fonnal
notation has alsobeen kept to a minimum in order to keep the account simple.

1.2
OVERVIEW OF COMING CHAPTERS
At the very outsel we need to establish the nature of the subject we are going to be examining. So.
Chapter 2 discusses the na1ure of words. Then the next three chapters delve deep inside words and
inve."tigate their internal struc1ure. ht the process. traditional morphological concepts of structural
linguistics are introduced and extensively exemplified.
Morphology is not a stand-alone module. After the introductory chapters. in Chapter6 you are introduced
lo a theory where morphology is an integral pan of the LEXICON or DICTIONARY. This chapter focuses
on the interaction of phonology and morphology in word-fom1ation.
Chapter 7 explores the relationship between words in speech and in writing. \Vhat is the relationship
between saying words and writing them down? Is "'Tiling simply a mirror of speech-and an apparently
dislorting one in the caseof English?
 lJIITRODUCTION 5

The following chapter continues the discussion of the role of the lexicon. It attempts to answer questions
like ·what is the lexicon for?' '\Vhat items need to be li�led in the dictionary?" ·What is the difference
between idioms(like to llllil ont!'s colour.s lo 1/Jemast) :md syntactic phrase."i (like1011aila 11otict! to the
door)?" The next two chapters highlight the fac1 that the English word-store is va"it and infinitely
expandable. First. in Chapter 9 we consider the ways in which. using the internal resources of the
language. speakers are able to produce an indefinitely large number of words. In Chapter 10 attention shifts
to the expansion of English vocabulary through the importation of countless word"i from other languages.
The story of imported words is in many ways also the story of the contacts that speakers of English have
had with speakers of other languages O\'Cr the centuries.
Most of the space in this book is devoted to an examination of the structure of English words. Bm
the analysis of word-structure is seen not ru; an end in itself. but rather as a means to an end. And that
end is to understand what it means to k11ow a word. What sorts of information about words do you
need to have in order to use them in communication? So the final chapter is devoled lo the MENTAL
LEXICON. It addresses the question. ·how is il thal people are able to store a vast number of words in
the mind and to retrieve the right one so fa..(;l in communicalionT We will see that words are not piled
in a muddle in the mind. Rather. the menlal lexicon is very highly organised. This concluding chapter
will also pull together the [Link] developed in the earlier chapters.
I have already stressed the point that morphology is not a selfcontained module of language. Any
discussion of word-fom1ation muches on other area-. of linguislics, notably phonology and syntax. so I
have provided a key to the list of pronunciation symbols at the beginning of the book. I have also included
at the end a glossary of linguistic tenns (many of them from other branches of lingui.s1ics) which mighl be
unfamiliar. But still I may have missed out some terms. If you encounter any unfamiliar technical temtS that
are not explained in this book, I suggest that you consult a good dictionary of linguistics like Crystal
(1991). Sometimes it is useful to present data using phonetic notation. A key 10 the phonetic symbols used
is to be found on pp. xix-x.x.
After this introductory chapter. all chapters contain exercises. Several of the analytical exercises require
you to look up words and parts of \\-ords in a good dictionary like the Oxford English Dicrio11ary. Access
to such a dictionary is essential when you study this book. This is a practical way of learning about the
structure of English words (and may also be a useful way of enriching your vocabulary).
 Chapter 2
What is a word?

2.1
INTRODUCTION
Often we find it very difficuh to give a clear and systematic account of everyday things. ideas. actions
and evenL.; that surround us. We just take them for granted. We rarely need to state in an [Link] and
aniculaIe manner wha1 they are really like. For instance. we all know what a game is. Yet. as lhe
philosopher Wiugcnstein showed, we find it very difficuh to stateexplicitly what the simple word gamt'
means.
TI1e same is true of the tem1 ll'ord, Weuse ,�•onls all the time. \Ve intui1i,rely know what the wonls in
our language are. Neverthele.-;,s mos1 of us would be hard pushed 10 explain to anyone what kind of object
a word is. If a couple of Martian exploreri. (with a rudimentary undersianding of English) came off their
spacc•ship and stopped you in the street 10 enquire what eanhlings mcanl by the 1enn WORD whal would
you 1ell them? I suspect you might be somewhat vague and evasive. Ahhough you know very well whm
words arc. you might find it difficuh to express explicilly and succinctly what it is that you know about
them.
The purpose of this chapter is to try to find an answer 10 the queMion: what is a word? h is not only
Martian explorers curious about the way earthlings live who might want to know what words are. We too
have an interest in [Link] words because they play suchan important role in our lives. As we saw in
the last chap1er. it is impossible 10 imagine human socie1y wi1hoU1 language. And equally. i1is impossible 10
imagine a human language that has no words of any kind. It i.s imp0ssible to understand the nature of
language without gaining some understanding of the nature of [Link]. in thischapter we will clarify what
we mean when we use the term ·word". This clarification is:essential if our investigations are to make ;my
headway for. as you will see presently. we mean quite a few very different things when we talk of words.
A standard definition of the word is found in a paper written in 1926 by the American linguist Leonard
Bloomfield, one of the great�t linguis:ts of Lhc twentieth century. According to Bloomfield. ·a minimum
free fom1 is a word". By this he meant that the word is the smallest meaningful linguistic unit that can be
used on i1s own, It is a fom11ha1cannot be divided inlo any smaller units that can be used independently
10 convey meaning. For example child is a word. We cann01 divide it up into smaller units that can
convey meaning when they stand alone.
[Link] this with the word childish which can be analysed into chi/d. and •ish. While the child bit of
clrildisl, is meaningful when used on its own (and hence is a word). the san,e is not true of •isl,. Although
according to the Oxft,rd E11glis/r Dic:1io11ary (OED) .;.fl, means something like ·having the
(objectionable) qualitie!. or (as in mam1isl,, [Link],. dt!l·[Link], slreepish. apish etc.), there is no way
v.e can use it on its
 ENGLISH WORDS 7

own. If some shoU1ed 10 you in the s1ree1. ·Hey, are you •ishr you migh1 smile bemusedly and 1hink 10
yourself. ·1:,11·1 he weird!' In 1he next chapter we will 1ake up the question of what to do with pieces of
words 1ha1 cannot be used meaningfully on their own. But for the n1omen1 we will focus exclusively on
words.

2.2
WORDS ARE LIKE LIQUORICE ALLSORTS
\Vhen we 1alk of word:i. we do not always mean exactly the same thing. Like liquorice [Link]. words
come in all sorts of varieties. \Ve will stan our discussions by distinguishing the different senses in which
we u�e the 1enn ·word'.

2.2.1
\Vord,forms
Let us use the 1em1 WORD-FORM to describe the physical fonn which realises or represents a word in
speech or writing. Consider the "ords in the following extract from [Link]'s poem:

12.11
Half•past one.
The :-ICCCl•lamp spuncrcd

The:-lrcet•hlmp munercd.
'The:-trcc1•lamp said.·Regard1ha1 woman
Who hesirntes 1owal'ds you in the ligh1of the door
Which openson her like a grin..
('RJupsodv on a \\indy nigh!' m EliOI 1963)

In written English, words arc easy lo recognise. They arc preceded by a space and followed by a space.
Using 1his criterion. we can Sa)' tha1 1here are thiny--one words (i.e. [Link]) in the extract from
'Rhapsody'. We will call word-forms like 1hcse which we find in wri1ing ORTHOGRAPHIC WORDS. If
you loolr.. again nt the e:<tract. you migl11 wonder if some of the hyphenated orthographic words are
·really' indi"idual words. Many people would hyphenate half-pas, as Eliot docs but 1101 sm:el•lamp. They
would write s1ree1 lamp as 1wo separate words. with a space between them. What would you do?
The use of hyphens 10 indicate that some1hing is a complex word con1aining more than one [Link]
unit is variable. largely depending on how transparcnl the compound nature of a word is. Shakespeare wro1e
1oday as to•daJ' and10nwrro,,• a�IO•morrow:

12.21
a. To,.morTOw. Caesar.
I shall be furnished to infonn you righdy...
(A111t>n\' and Cleopatra, I�iv)
8 WHAT IS A WORD?

b. 0! thal ,...cnow had here

But 1cn 1hou$al1d or 1hose men in England
Thai do not work 1o-day.
/lrlfn•V IV iii

Hyphenating u,.day and to•morrow is less common now. probably because most speakers are [Link]
of the compound nalure of these words. Today comes from Old English r dtz ·to+day' and tomorrow is
from Middle English to mor(tt)U't! (i.e. Lo (the) morrow)-to• can be traced back uhirmuely 10 a fonn lha!
meant 'this' in lndo•Europcan. Note in passing lhat three major periods arc distinguished in 1he history of
lhe English language: Old English (conven1ionally abbrevia1ed as OE) was spoken c.450-1100: Middle
English (conventionally abbreviated � ME) was spoken c.1100-1500 and Modern English from 1500 10
the present.
Gencmlly, the use of the hyphen in such words 1ha1 arc no longer seen as compounds is in decline. The
hyphen tends to be mostI)' used in compounds that are regarded as fairly new words. Many [Link]
words1ha1are 1ransparen1ly compounded, [Link], are nonnally "'riuen wi1hou1 a hyphen. Of
course. judgements� 10 what is an established word vary greatly. There arc few finn rules here. For
insiance. in 1he OED both set11my and st'a.,,•ay are shown 10 be aeceplcd ways of writing the word
pronounced a� I si:wel/. Similarly, 1he compilers of the OED show variation in the way they enter both
hyphenatedflrst•rt,tt! and first rme written as 1wo words separated by a space.
Interestingly. hyphenation is also used creatively to indica1e that an idea that would nonnally be
expreSM!d by a phr<.1se is being treated a� a single word for communicative pu�scs because it has
crystallised in the writer's mind into a finn. single concept. Thus. for example. the expression simple to
sen•e is normally :, phrase. just like easy to comrol. But it can also be used a.'- a hyphenated word as in
.,implr-10-sen·e recipe dhrltes (M&S Maga:.ine 1992:9). Similarly. on page 48 of the some [Link]. the
writer of an advcnising femure uses the phrase fresh from the fann' as a hyphenated word in ·fresh-
from• the.fam1 eggs·. But for crcniive hyphcnaiion )'OU arc unlikely 10 find anything morestriking
than this:

12.31
011 Pi1caim !here i.s linlc evidence of the \!IJ0t·tt'<'•lun-e-11·t![Link]•J11rrtt1rder,1heQ11un·1-pic111rt"-ill-t!\·rry-room
son of a11i1udc.
Simon \Vinches1er in Tlir Gtwrdiai, magazine, 12June 1993: 27·(i1aJK",.added to highligh1 lhc compounds)

Wh:11 we have�tablished is1ha1as a [Link] words have a spaceon eithersideof [Link]

there are cases where this simple rule of thumb is na. followed. There is a degree of [Link] in the
way in
which words are wrincn down: being, or 1101 being. separated by a space is in itself 1101 a sure sign of word
siatus. Some orthographic words which are uncon1roversially written n..�one unit contnin t\�O "'ords within
them. They are compound words likeflrs1rate. stah'll)', wheelbarrow and flW/>lJI. Furthennore. there arc
fom1s liketlury'rt>, l,adt1•, and /"m which arc joined 1ogc1her in writing yet which are not compound word..11;.
When you scmlch the �kin. you see immediately 1ha1 they're, had,,·,and /"mare really \·[Link] the pairs
of words they tJrl'. had t101 and / am. Our theory needs to say something about awkward customers like
these. Since the issues 1hey raise are complex. we will pos.1pone discu�sion of them until sections (4.3) and
(8.3). Finally. there are word., which are compounded (and maybe hyphenated as in 12.31) as a one-off 10
crystalli�c a pruticulnr meaning.
 ENGLISH WORDS 9

So far we have only considered orthographic words, i.e. recognisable physical written word-forms.
Obviously, words as physical objects exist not only in writing, bul alsoin speech. We will now briefly tum
10 word-forms in spoken language. We will refer to them as PHONOLOGICAL WORDS.
The challenge of word recognition arises in aneven more obvious way when we consider speech. Words
are not separated dislinctly from each other. We do not leave a pause between words that could be
equated to a space in writing. (If we did that, conversation would be painfully slow! Jusl try speaking to
one of your friends today leaving a two-second gap be1ween words. See how they react.) In normal speech
words come out in a torrent. They overlap. Just as droplets of water cannot be seen flowing down a
river, individual words do not stand out discretely in the flow of conversation. So they are much harder to
isolate than words in writing. None the less. we are able to isolate them. If you heard an utterance like:

12.41
The C:t.l [Link] in y<>ur bed.
/cka:1slept In: bedf
{NOie: ••• shows tha1lhc following syllable is stressed: phonemic lranscription is wriuen between slam lines.)

you would be able to recognise the six phonological words that have been written in PHONEMIC
TRANSCRJIYflON (which shows the PHONEMES, i.e. the sounds that are used to distinguish the
meanings of words) although what you hear is one continuous stream of sound. For purely practical
reasons, throughout the book. unless otherwise stated, phonemic transcriptions and references to
pronunciation will be based on RECEIVED PRONUNCIATION (RP), the prestige accent of standard
British English-the variety popularly known a.-. the Queen's English or BBC English.
An intriguing question that linguists and psychologists have tried to answer is: how do people
recognise words in speech? We will address this question in detail in section (11.2.1) below. For now
let us simply assume that phonological words can be identified. Our present task will simply be to
outline some of their key propenies. To do this it will be useful to distinguish between two types of
words: the so-called CONTENT WORDS and FUNCTION WORDS. Content words are the nouns.
verbs. adjectives and adverbs which contain most of the REFERE!\'TlAL (or COGNITIVE
MEANING) of a sentence. This roughly means that they name individuals and predicate of them
certain properties. They 1ell us, for instance, what happened or who did what to whom. and in what
circumstances. An example will make the point clear. In the old days, when people sent telegrams. it
wa.,; content words that were mainly (or exclusively) used. A proud parent could send a message like
Baby girl arrived yesterday which contained two nouns. a verb and an adverb. Obviously. this is not a
well-fom1ed, grammatical sentence. But its meaning would be clear enough.
Function words are the rest-prepositions, pronouns. conjunctions. articles and so on. They have a
predominantly grammatical role. A telegram comaining only the words Sheit and for us would convey little
idea of what the intended interpretation [Link] is not to say thal function words are superfluous. Without
them sentences are usually ungrammatical. A sentence like *Nelly we1111ow11 which lacks the preposition 10
is not permitted. We have to say Nelly welllto tow11.
In English, content words have this property: one of their syllables is more prominent than the
rest because it receives MAIN STRESS. This is seen in the words below where the syllable with main
stress is preceded by•":

1251
lllitial slrt:s:s Medial slrl'U Firwlstrl'n
JO WHAT IS A WORD'!

'acrobat
a·nnoying ca'hoot�
'lingfishc.r
de'molish gaber'dinc
"patriarchate
Otau·ccnan hullaba'loo

Main stresscan fall on only one syllable in a word. The location of main stress is part of the make-up of
a word and is not changed capriciously by individual s-peakers. You cannot decide to stress lwllabaloo on
the penultimate syllable on a Monday (lmlla'bl1loo), on the antepenultimate syllable on a Tuesday
(lm'llabaloo). on the initial syllable on a Wednesday ('hullabtiloo) and on the final syllable for the rest of
lhe week (/mllaba'loo).
Howe\/er. in some cases. if we wish to contras! two related words. we can shift stress from ils nom1al
position to a new position. This can be seen in ·,,,.,,dor and 1•,m'dee which normally arestressed on the first
and second syllable respectively. But if [Link] [Link] wants to contr�t these two words both words might
be stressed on the final syllable as I heard an estate agent do in a radio interview.

It ii,, ,·tn'dur. not the ,,,.,,•du who [Link] t.a.x.

This example illustrates well 1he point thal a word is allowed jusl one stress. Stre!.S can be shifted from
one syllable to another. but a word cannot have two main stresses. We could not have •·,·e11'dor and
•·i·eu'du where the two syllabics received equaJ stress. Stress has tu do with relative prominence. The
syllable that receives main stress is somewhat more prominent than lhe rest. some of which may be
unstressed or weaklystressed. By conltast. function words are nonnally unstressed. We can say Nelly
went 10 to,rn with no stresson 10 w1less we wish to highlight to for contraslive purposes. e.g. Nelly went
to town and ,iotfaraw1.1yfrom1mm).
It is easy to see how stress can function as a valuable clue in detem,ining whether twocontent words:
are a single compound word or two separale words. The nouns street and lamp are both stressed when
they occur in isolation, But if they appear in the compound 'street-lamp. only the first is stressed. The
stress on lamp issuppressed.
Stress is not the only phonological clue. In addition to stress. there are rules regulating the positions
in which varioussounds may occur in a word and the combinations of sounW that are pennissible. These
rules are called PHONOTACTIC RULES. They can help us to know whether we are at the beginning.
in the middle or al the end of a word. A phonological word must satisfy the requirements for words or
the spoken language. For in.!)tance. while any vowel can begin a word. and most consonants c-an
appear alone at the beginning or a word. the consonant l ) is subject to cenain restric1ions. ('Ibis
consonant is spelled 11g as in
long (sec the Key to symbols used on p. xix). In English words l J is nol allowed to occur initially although
it can occur in other positions. Thus. I ) is allowed internally and at the end of a word as in 11 I ) longing
and ll geJ longer. But you could not have an English word like 11gether. •t eej with ( J as its first sound.
However, in other languages this sound may be found word•initially as in the Chinese name Nga ( al and
lhe Zimbabwean name Nkomo I komoJ.
There are also phonotactic restrictions on the combination of consonanls in various positions in a word in
the spoken language. As everyone knows. English spelling is not always a perfect mirror of pronunciation.
So when considering words in the spoken language it is important to separate spelling from pronunciation
(cf. Chapter 7). You know that He is J:.11<x:k•k,ieed is pronounced /hi lz nk ni:d/ and not •/he ls knk kni:d/.
A particular combination of leuers can be associated with ,,ery different pronunciations in different words
or
 ENCUSII WORDS 11

in different positions in the same word. The spelling kn is [Link] /kn/ at the end of a word, as in I
belkn/. but al 1he beginning of a word as in /(}Jet and lmotk the /k/ is dropped and only 1hc II is sounded.
Similarly. otherslop-plus-na:-.al combinalions like tm /1111/ and d11 /dn/ arc allowed at lheend of a word
(e.g. bouom /btm/ and burdet1 lb :dn/) but lhcse consonanl clus1ers are not pennitted at the beginning of a
word. Pu1:uive words like •tunls/ (•tmiss) and •tdnel/ (•dne/1) .1re ju),,1 impermissible. In the spoken
13nguage we recogni� a.� English words only tho� forms thal have the right combimuion of sounds for
the position in 1hc word where they occur.
Moreover. even when a sound or combimuion of sounds is allowed. often a somewhat different
pronuncia1ion is used depending on the [Link] in which it occurs in a word. This can be seen in the
pronunciation of 1he / sound in srnndard Brilish English {RP) in different posi1ions in a word. Compare the
initial / wi1h the final / in1hc following:

12.11
Word•imtwl clt>ur
/JI \V,>rd•fintJIdctrk l'rt'•cmasommwl dllrk
labour lead lofc Ill Ill
lend let lick leaf' [Link] Mncll fulfil m,11,:!oOlt hell quilt
cool bull �poill colt wild
�prawl

TilC / sound is always made with the blade of the longuc aguin.),,t the teeth-ridge. with the sides lowered 10
allow air to escape. But 1here is a subtle difference. When / is in word-final posi1ion or when it is followed
by another consonan1 (as i1 it- in1he last 1wo columns). besides the aniculatory gestures mentioned above.
1hc back of the tongue is also simultaneously miscd towards the soft palate (or velum). This type of / is
called dark or vclariscd / (). But when / i.s at the beginning of n word. no velarisation takes place.111is lancr
1ype of I is called clear or noo-velariscd /(II), Thus. the kind of/ we hear gives an indica1ion of where in a
word it appears.
Do not foil10 note lhc use of square brackcls. They arc used 10 enclose ALLOPHONES. i.e. v,1rian1s of
a phoneme. Allophones arc diffcrenl sounds. e.g. I) and IJ.1ha1 <X'cur in diffcrcn1 conlcxts which all
represent 1he same phoneme /1/.
With regard to �pelling too. the si1uation is not chaotic. although admiuedly the rcla1ionship be1ween
lcucrs and phonemes is not always s1raigluforward. as kntt being pronounced /ni:/ demonstrates. We
recognise as English words only those onhographic words that conform 10 the spelling conventions of
English. If. for example, you �aw 1he word :yrogl�11you would treat ii as a foreign word. The le11er
combin:uion :1 r is not English. There i:,; no way a word in English could stan with those leners.
1

Let me summarise. Onesense in which we use the tenn ·word' is to refer lo WORD-FORMS. If we :trc
1hinking of the wrinen language. our word-fom1s :are ORTI·IOORAPHIC word�. TI,esc arc easily
recognised. They nom1ally have a space before and after dlCm, By contrJst. in nom,al spoken language our
word-forms arc PHONOLOGICAL words. These arc more difficult 10 identify because they are not
discrete entities tha1can be neally picked off one by one. None the less. phonological words can
beidentified on the basis of their phonological chnrncteristicssuch as stress and phonotaclic propenies.

2.2.2
Words as vocabulary ilems
We need 10 distinguish between word� in1hc sense of word-fom1 as opposed to words as vocnbulary
ilcms. Let us revisit the cx:unples in (2.2.1J on pp. 11-12. If we urc considering wordfonns. we can sec
that the
I? WHAT IS A WORD?

hyphenated word-fonn strut-lamp occurs three times, So if we were couming differem word-forms, we
would count street-lamp three times. However. if we were counting distinct words, in the sense of
distinct VOCABULARY ITEMS we would only count it once.
The di.s1inction be1wecn word-fom1s and vocabulary items is important. Very often, when we talk about
words what we have in mind is not word-fom1s, but something more [Link]-what we will refer to here as
LEXEMES (i.e. vocabulary items). Anyone compiling a dictionary lists words in this sense. So. although
the word-fom1s ineach of 1he columns in 12.8I below are different. we do not find each one of them given a
sep:irate entry in an English dictionary. The first word in each column is listed under a heading of iu, own.
1ne rest may be mentioned under 1ha1 heading. if theydo not follow a regular pattern of the languagc-e.g.
write, wri11e11 (past participle}. wrore (past tense). But if they do follow the general pattern (e.g. [Link],
wmhi11g. u·ashed; smile. smiling. smiled) they will be left out of the dictionary ahogether. Instead, the
grammar will be expected to provide a general statement to the effect that verbs take an -in>: suffix. which
marks progressive aspect. andan -td suffix that marks both 1he pas! tense and the pas1 paniciplc. andso on.

1281
WASH TAKE BRING WRITE
wai.h mkc bring write
y.,a_.;;hes 1alcs brings writes
y.•a.,hing talmg bringing wniing
wa..;;hed 100k brought wro«
y.,a,ihed [Link] brought wnucn

In 12.81 each lexeme (i.e. vocabulary item) 1hat would be en1ered in n dictionary is shown in capital
letters and all the different word-forms belonging to it are shown in lower-case letters.
The examples in (2.81 are all verbs. But. of course. lexemes can be nouns. adjectives or adverbs as
well. In l2.9) you will find examples from these other word classes.

12.9]

Noun Adjt'Crfr, Adl't'rb

,. MATCH KIND SOON
mtHCh lind soon
matrhes

b. GOOSE kinder
BAD """'"
WELL
goose b,d well
•eesc worse belier

In (2.9) we have three pairs of lexemes: the nouns, match and goost>; the adjectives kind and bad: and
adverbs soon and 11•e/J. In each case the word-fonns belonging 10 each lexeme in (2.9a) follow a general
pallem for words of their type and need not be !isled in the dictionary. But all the ones in [2.9b] are
irregular and must be lisled in the dictionary.
TI1e lexeme is an abstract entity that is found in the dictionary and that has a certain meaning. Word-
fonns are 1he concrete objecL.;; that we put down on paper ([Link] words) or uller (phonological
words) when we use language. The [Link] between a lexeme and the word-forms belonging to it is
one
 ENGLISH WORDS 13

12.101

of REALISATION or [Link] or MANIFESTATION. If we take the lexeme wr;1e which is

entered in the dictionary. for example. we can see that it may be realised by any one of the word-fom1s
write. wr;tes. writing. wrote and wrilfe11 which belong to it. These are the actual fonns that are used in
speech or appear on paper. When you see the orthographic words wri11e11 and wrore on the page. you know
that ahhough they are spelt differently they are manifestations of the same vocabulary item WRJTE.
The distinction between word-fonns and lexemes which I have just made is not abstruse. It is a
distinction that we are intuitively aware of from an early age. It is the distinction on which word-play in
puns and in intentional ambiguity in e"eryday life depends. At a certain period in our childhood we were
fascinated by words. We loved jokes---even awful ones like [2.10]
The humour, of course, lies in recognising that the word-fonn shrimp can belong to twoseparate lexemes
whose very different and unrelated meanings are none the less pertinenl here. It can mean either ·an edible,
long. slender crustacean· or 'a tiny person' (in colloquial English). Also. the word .un·e has 1wo possible
interpretations. It can mean 'to wait upon a person at table' or ·todish up food'. ·[Link], word-play exploits the
lexical ambiguity arising from the fact that the same word~form represents two distinct lexemes with very
distinct meanings.
ln real-life communication. where potential ambiguity occurs we generally manage to come 10 just one
interpretation without too much difficulty by selecting the most appropriate and R£LEVANT interpretation
in thesituation. Suppose a 20-stone super heavyweight boxer wenl to Joe's Vegetarian Restaurant and asked
the waiterfor a niceshrimp curry and the waiter said in reply. ·we don't serveshrimps'. it would be
obvious that it was .shrimps in the sense of crustaceans thal was intended. If, on the other hand, a little
man, barely 5 feet tall and weighing a mere7 stone, went to a fish restaurant and saw almost everyone at
the tables around him tucking imo a plateful of succulent shrimps. and though1 that he would quite
fancy some himself, he would be rightly offended if the waiter said 'We do not serve shrimps.' It is
obvious in this situation that shrimps are on the menu and are dished up for consumption. What is not
done is serve up food to people deemed to be puny.
Puns are not restricted to jokes. Many advertisements like that for Standens rely on puns for their effecl.
Given thecontext, it is obvious that sound is meant 10 be read in more Lhan onesense here.
Serious literature also uses this device. For instance, the First World War poet Siegfried Sassoon
gives the title ·Base details' to the poem in which he parodies cowardly generals who stay away at the
base. at a safe distance from the action. and gladly speed young soldiers to their dealh at lhe front. The
word-form base in the title represenl� two distinct lexemes here whose meanings are both relevant: (i)
Base derails are details of what is happening at the base tNounl (meaning •military encampment'). and (ii)
Base dewils are particulars of something that is base(AdJCcti'"-cl (meaning ·reprehensibly cowardly. mean
etc.').
The 1enn HOMONYM is used to denote word-forms belonging to distinct lexemes that are written
and pronounced in the same way. There are separate dictionary entries for such words. Shr;mp and
base are examples of homonyms. But perhaps they are not so obvious. Beller examples of homonyms
are shown in [2.11).

12.1 I]
a. ba1:bat t"NouaJ 'a small flying mammar
bi11!NWD! 'a wooden implemenl for hilling a ball in cricke1·
b. bar: bara-.-·,th(: profcs�ion of barrister·
bar ,s-i·a,·[Link] across as1aveused10 mark metrical accent in music·
bar (\'eib) ·10 obs1n.tc1·
14 WHAT IS A WOKO?

C. fair: fairtAd,t..1,1-c1'bcaul1ful. auracth•c·

air 'hohdav·

By contras!. word-forms may have the same pronunciation but different spellings and meanings. Such
forms are called HOMOPHONES. See [Link] example from a joke book:

12.121
Why�thepony rough?
Bec:a,ue he'sa li11fe hoarse.
(Youngand Young 1981:57)

The joke is a pun on /h:s/, the pronuncialion of the two lexemes represented in writing by horse and
hoarse. Otherexamples of homophones include rail - tale. sail - sale. wearlter - wherl,u, see - sea. reod
- reed, reel - real. see,i - sa,[Link] - kJ1eud.
Conversely. it is also possible to havesevcrnl closely rela1ed meanings 1ha1 are realised by thesame
word• form. The name for this is POLYSEMY. Oflen you find severaJ senses fo,tcd under a single heading
in a dictionary. For instance, under the entry for the noun/ore,-, the OED listsover ten senses. I have
reproduced lhe first six below:

12.131
I. Physical ),,1rcngth. Rarely 1n ,,,. (• [Link]-1818.)
2. Streng1h, impetus.1 iolcoce. or 1nten!lit)' of effect ME.
3. Power or might �p. [Link] p0wer ME. b. ln early use, the.1,trcnglh (of a dcfcn.•,he work etc.). Sub�.•
lhe fighungstrength of a ship. 1577.
4. A bod)' of anncd men. anam1y. In11I. the1roops or .w,klicr.. composing the lighting:!-lrength of a lingdom or a
commander ME. b. A body of police:often ab...ol. th�/11n""r=-pol1cemcncollcctavcly. 185I.
5. Physical s1rcnglh or [Link] on an object:C".Sp. violence or physical coercion. ME.
6. Mental or moral strength. Now only, powerof effc<:tnc oct1on1or of mercoming rc.,istancc. ME.

The line that separates polysemy from homonymy is somewhat blurred because ii is not :.thogethcr clear
how far meanings need to diverge before we should 1reat words represcntmg them as belonging to distinct
lexemes. In 12.13). it is not entirely clear that the i;ixth sense of the nounj(Jrct' is not sufficienlly removed
from the other meanings lo merit an enlry of its own. The other meanings all show a reasonably slrong
family resemblance. But mental or moral strength shows a somewhat weaker rela1ionship.
In 1hc OED, there is a scpamle entry for the lexeme force. the verb. It is considered a different lexeme
because it has a differcnl meaning and belongs to a different word-class. being a verb and not a noun.
Belonging to different word-classes is an important [Link] in detennining whether separate
diclionary entries arc needed.
In real•life communication. the lack of a one•t0•one match between lexemes and word•fom1s does not
necessarily cause ambiguity. In contexl. the relcv�mt meaning is normally easy to detennine. Bui there are
cases where it 1s not. For instance. the homonymy of bat in (2.14] can causesemantic confusion:

12,141
I sawa bm under the u-cc.
 F..NCLISII WORDS 15

11 could be a bat with which you play cricket or a small. flying mammal. lbis isa case of LEXICAL
AMBIGUITY. We have in this sen1ence a word-fom11ha1 rcprcscms more than one lexeme wi1h a
meaning 1hat is quite plausible. It is not pos.,ib1e 10determine the right in1erprctation of thesentence
without looking at the wider context in which ii appcar1>.
We have es1ablished that the rcla1ion$hip bc1wcen a word-fonu m,d 1he meaning that it represents is a
complex one. This is cxploi1ed not only in liter:uure and word-play as we saw above but also in the
language of advcnising. For insiancc. a recent British Gas new�l)aper advenisemcnt for gas heating said:

12.151
You "'II worm to our credit 11·�free.

This advcni!>Cmenl exploits the lexical ambiguity that is due10 the fac1that ll'arm (ro)can mean ·become
enthusiaslic' or ·experience a rise in tempenuurc'. Next 1ime you look .it nn advcnisemc111. see whether
it cxploi1s any of 1hc relationships between lexemes and word-forms that we have examined,

2.2.3
Grammatical words
Finally. let us consider the word from a grammatical perspective. Words playa key role in syntax. So. some
of lheir propcnics arc a.��igned laking into account syntactic foctors. Often words are required 10 ha\'c
ccnain propenics if they serve cenain symac1ic purposes. Thus. although in12.16a) we have the same sense
of the same lexeme (play} realised by lhe same word-form {played), we know that this word docsat least two
quite different grammatical jobs in the�n1ence of which ii i�n pan:

12.161
[Link] pl<t)Cd the flute. b. She IOOk the flute.
She ha.�pla)'c<l1hc flute. She has [Link] the flu1c.

If you compare the sentences in \2.161abo,·e. you will sec 1h:u in I2. I6al 1he verb play is realised by the
word-form played regardlessof whether i1 simply indicates that the ac11on happened in 1hc pasl as in1he
firsl [Link] or lhat an action was (recently) completed as in the SCt.'Ond example. Contrjst thi.s with
the situation in 12.16bJ where these 1wo grammatical meanings are signalled by 1wo different form.'i.
T1Jok indicates 1hat lhe action happened in lhe past while taken (after lu,s/luuf} indicates that the
action is complc1e. In She played the flute and Sh(' took rheflwe lhe wordsplayt1d and took arcdescribed
grammatically as the 'past tense fornti of the verbs p/c,y and toke'. By contrast, in She has playtd the
flute and She ha.t take,i the flute we dcscribep/ayt>dand wkeu as 1he ·pa.�l p.1niciple' of play and wke.
Lingui,1.._: use the tenn SYNCRE'l'ISM to describe situations such as that [Link] by played where
the same word-fonn of a lexeme is used to realise two (or more) dislincl grJmmalicaJ words th:ll are
rcpresemed separately in the gr:,mm::uical rcprcscn1a1ions of word$ belonging to some other comparable
lexemes. The phenomenon of syncrctism is one good reason for distinguishing between word�fonns and
gramma1ical words. h enables us to show that words belonging 10 the same lexeme and having the same
form in!opecch and writing can still differ,
A funher example should make the ideus of grammatical words and syncrctism even clearer. Consider
1hc verbs in the following semenccs:
16 WHAT JS A WORD1

ll.171
a. You hit me. (=you hit meM>mc time in lhe pa."1)
o,
(=you hit me habitually)
b. Yoo CUI it. (=you cut i1some time in the pa..�1)
°'
{ =you cut ii habitually)

As the paraphrases show. the word-fom1 hil belonging to the lexeme Mr can repre.�ent either the
present tense or the pasl tense form of the verb. In other words, there is syncretism. We have two different
grammatical words hit IHcrb. +�n1J and 1,;, IH'crb. +�I but a single word-fom1. The same analysis also
applies to cw. It can represent either the present or past tenseof the verbcur.
Syocretism is not limited to verbs. It can apply to other word cla..,;;ses (e.g. nouns) as well:

ll.181
(a) ll\e wolf Lilied a �heep and one deer.
(b) The wolr killed 1wosheepand three deer.

In these two sentences, although the word-fonn .sheep belongs to the same lexeme and is unchanged in
form, we know that its grammatical value is not the same. In (2.18a] it realises the word with the
grammatical propenies of noun and singular, but in l2.18b] it represents a plural form. Likewise. the same
word-fomt deer represents a singular noun in 12.18aJ and a plural noun in[2.18b].
What can we say about the word as an entity that functions as a grammatical unit in the syntax of a
language? As mentioned already. the (grammatical) word is nonnally defined a.'i the MINlMAL FREE
FORM that is used in the grammar of a language. Let us now put some flesh on [Link] terse and somewhat
cryp1ic sta1ement.
By free fom1 we mean an entity that can stand on its own and act as a freeagent; it is an element whose
position in a sentence is not totally dictated by other items. In order toexplain what ·freedom' means in this
context, we need to take on board lwo ancillary ideas: POSITIONAL MOBILITY and STABlLITY.
Although words are not the smallesl grammatical uniL'> used lo construct sentences (see the discussion of
morphemes in the next chapter), at the level of semence organisation the rules of sentence forma1ion treal
words as unanalysable units. Often it is possible to change the order in which words appear in a sentence
and still produce a well-fanned sentence. Words enjoy considerable positional mobility. However, the
elements inside a word do not enjoy such mobility. While syntactic rules can transpon words to new places
in a sentence. they cannot shift in thesame wayelements that are found inside words. Moving words around
in 1he following produces grammatical sentences with basically the same meaning, but ·with somewhat
differem emphasis:

ll.191
•. This old induslriafo,I re\liiilcd Lanc:1s,ttt, fonunately,
b. Fortunately. thisokl indusufalis1 revi�ited Ulncas1cr.
C. Lancaster.1his old industriali�I rc\li�itcd, fonuna1cly,
d. Fortunatel)·· Lanca�tcr was revisited by lhis old
industrialist.
 ENGLISH WORDS 17

Evidently, the position of words in a senLence is noL rigidly fixed. They can. and often do, get moved
around if the communicative needs of the speaker or writer require it. However. the interior of a word is a
no-go area for syntactic rules. They are strictly barred from manipulating elemenL'- found inside a word. As
far as syntax is concerned. words are indivisible units that cannot be split and whose internal units are
inaccessible (cf. Bauer 1988. Matthews 1991, Lyons I968. Di SciuJloand Williams 1987).
The word a.._ a grammatical unit shows stability (or LNTERNAL COHESION). The order of elements
inside a word is rigidly fixed. If theelements of a sentence areshifted. certain meaningful units(in thiscase
re-visit-ed and fort1111-ate-l-y) all move en bloc, and their order always remains unchanged. The internal
structure of the word cannot be tampered with. We are not allowed to perform operations that would yield
words like •ed-l'isit-re, •ate-fortune-I_\' etc. We will return to this point on p. 33 below.
The definition of the word includes the term 'minimal· for a good reason. This is intended lo separate
words from phrases like thi.v old i11d11striali:St. Like word._, phrases can occur in isolation and they can be
moved from one position to anoLher (as we have seen in 12.19J). Bui the expression this old i11dus1rialis1
is
not a minimal form since it contains smaller forms capable of occurring independently namely, this,old and
industrialist. Furthermore. the sequence this old [Link] does not ha\'e lhe kind of internal cohesion
found in words. h can be interrupted by other words e.g. this weal1hy old indus1rialis1; this \'t'f')'
weahlry, old, bene\'Olem [Link].
The assurnpLion tha1 the grammatical word is ·a minimum free form' works well as a rule of thumb.
Bui ii encounlers difficulties when confronted by a COMPOUND WORD like wheelbarrow which contains
the words wheel and barrow which can stand alone. In such cases it is clear that the word is not the smallest
meaningful unit thaL can be used on its own. It is for this reason !hat the definition of the word as the unit
on which purely synlactic operaLions can be performed is preferable. ln the case of compounds this
definition works. The interior of a compound is a syntactic no-go area. Syntactic rules are no1 allowed to
apply separately to words that make up a compound. Thus. for example although the nouns wheel and
barrow can be modified by theadjective big (/big barrow/, /big wheel/).and although we can talk of /big
wheelbarrow/. in whjch case big modifies the entire compound, there is no possibility of saying ll'heel /big
barrow], with the adjective only modifying the second element of thecompound word.

2.3
SUMMARY
In lhis chapter we have established that normally, the tem1 ·word' is used ambiguously. To a\'oid the
ambiguity, we need to djs1inguish between three different types of word: (i) a word-fonn (i.e. a partjcular
physical manifestation of one or more lexemes in speech or writing); (ii) a vocabulary item (i.e. lexeme):
and (iii) a unit of grammalical strucLure lhai hascertain morphological and syntactic properties.
We will revisit the distinction between lexemes. grammatical words and word-forms mainly in Chapters
7 and 11. In Chapter 7 our main concern will be [Link] realisation of words in speech and in writing. In
Chapter 11 we will show that this distinction is not an artefact of the linguist's analysis. Ralher, it is a
distinction that is well supported by studies in the way in which we store words in the mind and retrieve
them for use in communication in real life.
In the coming chapters. in cases where the relevant sense of the term ·word' is clear from the context I
wiJI not spell out whether it is the word as a vocabulary item, gmmmmical word. phonological or
orthographic form Lhat is beingdealt with. But where it is not clear.1will indicatethe sense in which I am
using this term. We are now ina position to consider in detail the internal structure of words. That is the
task of the next chapter.
18 WHAT IS A WORD?

EXERCISES

I. Comment on the problems you encounter in de1cm1ining the number or words in 1he following nu�cry
rhyme. Relate your answer to the different senses in which the tenn ·word" is used.

The grand old Duke of York

He had ten thousand men.
He marched them up to the 1op of the hill.
1ben he marched them down again.
When 1hey were up. they were up.
And when they were down. they were down.
And when they were ooly haff way up
1ney were neither up nor down.

2. Find and analyse al least three examples of advcrtisemenls lhnt exploit the homonymy. polyscmy or
homophony or words.
3. Which ones of the italicised word•fom1s in the following sen1ences belong tu the same lexeme? What
difficulties. ir [Link] you comeacross in de1em1ining whether [Link] belong to the same lexeme?

a. She saw him saw through that plank of wood.

b. 8;// will pay the bill.
c. I saw Farmer near your farm again this morning.
d. Jan looked pale when�he walked toward.., the pail.
e. I run sick of yourclaiming to be .fitk all the time.
f. I was looking a1 the: book when she book�d the ticket

4. Using at lcasl two fresh examples. show how syncrctism can be u.�ed 10 support the distinction bc1ween
[Link] and gramnm1ical words.
5. This is the beginning [Link]·s poem 'MuSCe des Beaux Arts·.

Abou1 suffering they were never

wrong. The Old Masters...

These linescan be pamphrased as "The Old Masters were never wrong about suffering.·
Referring 10 lhc definition of the word given in 1his chup1er. explain why it is correct to regard
.mfftrillg a.�a word but incorrect to [Link] abou, sujferi,,g [Link] as a word.
 Chapter 3
Close encounters of a morphemic kind

3.1
THE QUF..ST FOR VF..RBAL ATOMS
We saw in the last chapter 1h01 the word i.s the smallest meaningful unit of language that can function
indcpcndenlly in the grammar, A word can be used on its own. without appending it 10 some other
unit. Thus. inthe word childish we can isolatechild and use it on its own because it is a word in its own
right. But we cannot use -isl,as a sland-alone unit, for -is/J is not a word.
While recognising that word1- are the smallest meaningful units which function independently in the
grammar. we also need 10 recognise th:u words can be decomposed into smaller units that are alw
meaningful. Our task in this chapter is to explore the internal structure of words in order 10 gain some
understanding of the basic units which are used to fonn words.

J.2
CLOSE MORPHOLOGICAL ENCOUNTF..RS: ZOOMING IN ON MORPHEMES
[Link] 'morphology' meant the study of biological fonns. But ninc1eenth-centUI)' students of language
borrowed the 1cm1 and applied it to thestudy of word•structure. In linguistics MORPHOLOGY is the study
of the fom1ation and internal organisation of words.
Let us begin our morphological analysis by considering half a [Link] words (not altogether randomly
chosen):

13.1I
[Link] mend boil safe leaf14-ord clcpham

Obviously all the words in [Link] have a meaning, but lack internal structure. We cannot identify any
smaller unils that are themselves meaningful which occur inside them. If a Martian stopped you in a street
near the local zoo and enquired wha1 /Jham in tleplumr or ho in hope means. you would think she was
asking a mos1 bizarre question that did 1101 merit an answer. Or you might condescendingly explain that. of
course. ineach case1he whole word means�mething, but i1s partscannot be said to mean any1hing on their
own, Though somewhat puzzled. the Manian migh1 accept your explanation.
Bui. being the persistent type. let us suppose she enquired further whether !he word� in (3.21 were also
indivisible into smaller meaningful unils:
20 CLOSE ENCOUNTE:RS OP A MORPHEMIC KIND

13.2]
childish hopeless sooner mendedelephants re-boil unsafe ex-wife

You would have to give a different answer. You would need to tell your interrogator, who by now would
be getting increasingly bewildered, that the words in (3.2] can be divided into smaller units of meaning as
shown in (3.3):

13.31
ch1ld-ish bope-lrs.s soon-a mcnd-ed clepham-.i-u-boil ,m-safe n-w1fe

The part of the word that is nol italicised can funclion as an independent word in the grammar. indeed.
each of the nonitaljcised chunks is a word (i.e. vocabulary i1em) that is listed as such in the dictionary. By
contrast. the italicised bits. though meaningful (and their meanings can be indicated as shown in (3.41).
cannot function on their own in the grammar.

13.41
-ish 'havingthe (objectionable) qualities of"
-less •withoul X' child-is.h= "having lhe qua1ities of a child"
�r ·morex· hopeless= 'wi1hout hope.
--ed ·past' sooner.= ·more soon·
·S "plural" mended= 'mend in lhe p�t'
re ·again" elephants= ·more than oneelephanf
"notX" re-boil= 'boil again'
unsagfe= ·not safe"
ex-wife= 'former wire·

\Vhat we have done lo the words in £3.4} can be done to thousands of other words in English. Theycan
be decomposed intosmaller units of meaning (e.g. re- ·again') or grammatical function (e.g. -ed 'pas1').
The tem1 MORPHEME is used to refer to the smallest unit thal has meaning or serves a grammatical
function in a language. Morphemes are the aloms with which words are bujh_ h is not possible to find sub-
morphemic units that are themselves meaningful or have a grammatical function. Thm;, given -less or 1111-.
it would make no sense lo lr)' lo assign some identifiable meaning to any part of these forms. Of course. it
is possible to isolate the individual sounds fl-1-s/ or / -n/. but those sounds in themselves do not mean
anything.
We have now established that words are made up of morpheme.s. But how do we recognise a morpheme
when we see one'! Our definition of the morpheme as the smallest unit of meaning (or grammatical
function) will be the guiding principle. Any chunk of a word with a particular meanjng will be said to
represent a morpheme. That is how we proceeded in [3.3] and (3.4) above.
Morphemes tend to have a fairly stable meaning which they bring to any word in which they appear.
If we take re- and un-. for example. they mean 'again' and ·not" [Link]-notjusl in the words we
have listed above, but also in thousands of other words. Usually morphemes are used again and again
lo form different words. Thus re- meaning ·re-do whatever the verb means' can be attached before
most verbs to
yield a new word with a predictable meaning ([Link]-run, re-lake, re-build etc.). In like manner. 1111-
meaning 'not x· (where X stands for whatever the adjective means) can be attached to various adjectives
(e.g.1111-real. 1111-c/ea,r, 1111-happy etc.) to yield a new word with a predictable negative meaning.
 ENGLISH WORDS 21

The segmentalion of word-; into morphemes is not a trivial and arcane pastime indulged in by linguists 10
while away the time on a wet Bank Holiday aflemoon. It is something that is important for all users of
language. During yow lifetime. you will probably encounter hundreds of thousands of differenl words.
Many of these words will be new to you. For no mailer how extensive your vocabulary is. you will
inevitably come across words that are unfamiliar. It is impossible for anyone 10 know all the words Lhat are
found in English.
So. what do you do when faced with an unfamiliar word? Reach for a good dictionary? Perhaps. But this
is 1101 always feasible. Nor is it always necessary. Very often you just figure oul what the strange word
means using the context. together with yow knowledge of the meaning of the morphemes which the word
contains. You nom1ally do lhis subconsciom:ly. Whal we are doing here is making explicit your lacit
knowledge of word-slructure.
lnmgine lhis scenario. In 1992. a newspaper report on the war in the Bosnian republic slates that what
we are witnessing is 1hc Lelxmo11isa1io11of Bosnia. Suppose you have not encountered the word
leba11011isatio,i before. Would you understand what the writer is saying? Probably you would-without
looking it up in any dictionary. How would you do il? The answer is simple, By using your knowledge of
the world-in particular history (Ba/ka,,isatio11) -and your knowledge of current affairs (the civil war in
Lebanon) plus your knowledge of the principles of ,vord-formation you are able to work oul the meaning of
Lebano11isatio11.
Let us focus on principles of word-fom1ation. You know that -ize/-he is used when talking about nations
10 mean 'to make x·.e.g. from America we get America11ise. from Kort!a we get Kort!a11ise, from Keuya
we get Ke11rtmise etc. By attaching -(a11jise we tum a noun into a verb. So, given the noun Leba11011 we
can
form lhe verb Lelxmo,rise. Nexl. from the verb leba11onise. we can create a new noun by adding -a1io11
(which forms nouns of action).
If you know that various warlords created warring fiefdoms that destroyed the Lebanese slale during the
civil war that raged in Lebanon in the 1970s and 1980s. you will know thal the Croats, Muslims and Serbs
engaged in the Bosnian connict risk doing thesame to the Bosnian state in the 1990s. Leba,w11isarion i.s the
act of 'turning a country into anolher Lebanon·. Thus. our knowledge of word-structure contributes to ow
understanding of the meaning of unfamiliar words.
\Ve have demonstrated that words can be decomposed into morphemes. Now we are going to see that
words have INTERNAL STRUCTURE. A simple way of showing lhis is to analyse words like1mca11ny
and 1111/,appy. From these words we can derive 1111ca,mier and 1111/Jappier. lf you analyse 1mhappier. you
will see that exlracting lhe correcl meaning ·more lnot happy)' (i.e. sadder) ralher than the incorrect one 'not
Lmore happyf (i.e. not happier) depends on the way we group1ogether the morphemes. ln the first
analysis where anhappier is interpreted as sadder. the meaning ·not' con"eyed by 1111- is bracketed
together with happy LunhappyJ as one unit and this is intensified by the -er suffix. In the alternative second
analysis. Jwpp)' and -
er are bracketed logether as a unit lhappierJ (i.e. more happy) which then is negated by tun-] to give 'not
more happy'. which is incorrect. When someone is unhappier. ii does not mean they are simply less happy,
it mean.s r.1Lher that they are not happy at all. They are sad. This shows that morphemes in a word with
several morphemes may be grouped togelher in different ways for semantic purposes. The way in which
1his is done has semantic consequences. Conceivably, morphemes could be thrown together higgledy-
piggledy to form a word. So long as you had the right morphemes. a well-fom1ed word would popout. But
1hat is definitely not the case. Words have internal structural groupings. as we have seen.
Furthermore. the sequencing of morphemes in a word may be subjecl to restrictions. Take a word like
tmgowm,ability which contains four morphemes. namely w1-. gm·t>m, abil, ity. Everyone who knows this
22 CLOSlil:.NCOUNTl:.RSOF A [Link].\11C [Link]

word knows that these four morphemes mus1 appear in 1hc order in [3.5aJ. Any other order is striclly
forbidden:

13.51
un-gMern-'1bil-it)'
b. •go\cm-ab1l-un•it)'
c. •1ty-un-nb1l-govcm
d •abil-un-1t)'·(;0\cm
•un-g0\crn-i1y-abil c1c.

Clearly, knowing a word means not just knowing the morphemei- it contains. but abo the rigid order in
which they llfC allowed to appear, We will rcrnm to this poin1in section (4.4).
To sum up the discussion so far. words [Link] built using morphemes. If we know how morphemes arc used
10 fom1 words. we do not need 10 be unduly flustered when we come across a strange word. Usually ii is
possible to work ou1the meaning of a strange word if i1 contains familiur morphemes.

J.J
MORPHEMES AND THEIR DISGUISES
The identification of morphemes b not altogether straightforward.11lis b because there is no simple one-to•
one curre�pondencc bc1wecn morphemes :md 1he�pecch sounds that represent chem. In this section we will
attemp1 to unravel the complexities of Lhe rclatiom,hip between morphemes and the ac1ual forms (sound.s
of groups of sounds) by which they are manifested in speech.

J.J.I
Allomorphs: nlOrJ>h families
Any [Link] form 1ha1 represents a morpheme is called a MORPH. The forms -ish, -less. -er. -ed. -s. re-,
w1- and ex- in 13.4) on p. 31 arc all morphs. Morphologic:11 analysis begins wi1h the idcn1ifica1ion of
morphs, i.e. fonns that curry some meaning or arc nssocia1ed with�ome gmmmatical function.
In"spota,:us there is jwa one morph but in all lhc words in13.41 there are two.
II is imponant not to confuse morphs wi1h SYLLABLES. When we talk of morphs we have in mind
sounds that can be refuted 10 a panicular meaning or grammatical func1ion (e.g. plural or pa.!ot tcnM:),
However. when we talk of syllables all we have in mind [Link] chunks inio which words can bedivided for1he
purposes of pronunciation.
TI1is is not an [Link] distinc1ion. We nrc 1101 being pedantic. h is a dis1inc1ion 1ha1 muuers10 ordinal)'
people becuu�human languagci-arc organised insucha way thatthe construction of uniis that arc
mcaningful is normally in principle scpara1c from the construction of strings that urc pronounceable. Thus.
for rhythmical effec1. nursery rhymes often use nonsense syllables like 'Deed/�. deed/e' in ·J)e�dle
deedle dumpling my sou10h11' which do not represent anything meaningful.
Altcma1ively. a sound rcprcscn1ing a morpheme may 1101 be a i,,yllablc in [Link] right.e.g. by itself. lhe
- s which represents the plural morpheme is not a [Link] word ems has1wo morphemes. cm and •,f,
but it is ull ju,:1 one syllable. 111c single i,.ylfoble ca1s realise$ two morphemes. The convcn.e situation.
where
 [Link] WORDS 2J

several syllables realise a single morpheme, is equally possible. Thus, the trisyllabic and quadrisyllabic
word�forms elepham and a.s{Xlragu.f both realise just a single morpheme.
The nature [Link] relalionship between sounds and morphemes is intriguing. At first sight, it mighl look
reasonable to assume that morphemes are made up of PHONEMES. We mighl be tempted IO think that car.
the English morpheme with 1he meaning is made up of the phonemes /k:r.t/. But we h:.ive several kinds of
evidence showing that lhis is not the case.
First, if morphemes were nwde up of phonemes. a given morpheme would be uniquely associated with a
gi,·en phonological representation. In reality. the same morpheme can be realised by different morphs (i.e.
sounds or written fom1s). Morphs which realise [Link] same morpheme are referred to as ALLOMORPHS of
that morpheme.
1be INDEFINITE AR"rlCLE is a good example of a morpheme with more than one allomorph. It is
realised by the lwo forms a and an. The sound at the beginning of the following word detennines the
allomorph that is selected. If the word followini; the indefinite article begins with a consonant. the
allomorph a is selected, but if it begins with a \'owel the allomorph m, is used instead:

13.61
a dic1iom1ry b. an i�l:md
a boat an evening
a pineapple an opinion
a leg an eye
a big (ma1) an old (mat)
a dull (song) an exciting(finish)

Hence the incorrectness of the scmence marked wilh an asterisk in 13.7):

13.71
a. I �pent anCHming with them.
•1�pcm"evening with1hcm.
b. I :-pent tht' C\ cning withlhcm.

Allomorphs of1he same morpheme are said to be in COMPLEMENTARY DISTRIBUTION. This means
that they do not occur in identical contexts and therefore they cannot be used to distinguish meanings. In
other words, it is impossible to have two otherwise identical utterances that differ in their meanings
depending on the allomorph of a morpheme that is selected. So. because a and 011 both realise the same
indefinite article morpheme, it is impossible 10 have two sentences like those in (3.7a) above which are
identical in all ways. except in thechoice of a or an. but mean different things.
Complementary distribution presupposes lhe more basic notion of DISTRIBUTION. Distribution is 10do
with establishing facls about the occurrence of allomorphs of a particular morpheme. h is concerned with
establishing the con1exts in which the morpheme which we are investigaring occurs and [Link] [Link] by
which it is reaJised in those different contexts. In other word-.. by distribution we mean the total set of
distinct linguistic contexts in which a given form appears, perhaps in different guises. For instance, the
indefinite article has thedis1ribu1ion: a before consonants (e.g. a tree) and an before vowels (e.g. au eagle).
As mentioned already, such functionally related forms which all represent the same morpheme in
different environments are called allomorphs of that morpheme. Another way of pulling it is
24 CLOSE ENCOUNTERS OP A MORPH�11C KL'\ID

[Link] allomorphs are forms that are phonologically distinguishable which. none1he less. are not functionally
distinct. In other words, although they are physically distinct morphs with different pronunciations,
allomorphs do share the same function in the language.
An analogy might help to clarify this point. Let us compare allomorphs to workers who share the same
job. Imagine a jobshare situation where Mrs Jones teaches maths to form 2DY on Monday afternoons. Mr
Kato on Thursday mornings and Ms Smith on Tuesdays and [Link]. these teachers are clifferem
individuals. But they all share the role of ·maths teacher' for the class and each teacher only performs that
role on particular days. Likewise. all allomorphs share the same function but one allomorph cannot occupy
a position tha! is already occupied by another allomorph of the same morpheme. To summarise. we say that
allomorphs of a morpheme are in complementary distribution. This means [Link] they cannot substitute for
each 01her. Hence, we cannot replace one allomorph of a morpheme by another allomorph of that
morpheme and change meaning.
For our next example of allomorphs we wilJ tum to the plural morpheme. The idea of ·more than one' is
expressed by the plural morpheme using a variety of atlomorphs including the following:

13.81
Singular Plural
rad-iu1> radi-i
cactus cac1-i
b. dal-um dat-a
stm1-um str.1.1--a
[Link]-is analys-cs
ax-is ax-�
d. skirt skirt-s
rood road-s
branch branch.-e"l

Goingby theorthography, wecan identify theallomorphs -i. -a, -es and -s. Thelast is by far the
commonest: see section (7.3).
Try and say the batch of words in l3.8dJ aloud. You will observe that the pronunciation of the plural
allomorph in these words is variable. It is [s] in skirts, [zJ in roods and [lz] (or for some speakeP.j: [ez1) in
branches. What is interesting about these word.,;; is that the selection of the allomorph that represents the
plural is determined by the last sound in the noun to which the plural morpheme is appended. We will return
to this in more depth insection (5.2).
We have already .seen. that because allomorphs cannot substitute for each other, we never have two
sentences with different meanings which solely differ in that one sentence has allomorph X in a slot where
another sentence has allomorph Y. Compare the two sentences in{3.9J:

[3.9]
They have two b. They ha\'c twodogs
ca1s (el h:rv tu: fcl ha.-v tu: dg-.:]
k:tt-sl •1e1h:tv tu:dl:!;-s)
..Id ha-v tu: ka-t-zJ
 ENCLISH WORDS 2.S

We cannot find two otherwise identical sentences which differ in meaning simply becau� the wonl c:ms
is pronounced as [ka:t-sJ and •[k:ct-zJ respectively. Likewise. it is not possible 10 have two otherwise
identical sentences with different meanings where the word dog.f is pronounced as fdgz] and •ldgsJ. In
other words. Lhe difference between the allomorphs Js) and (zJ of the plural morpheme cannot be used 10
distinguish meanings.

3.3.2
Conlrasl
Different morphemes CONTRAST meanings but different allomorphs do not. If a difference in meaning is
attributable to the fact that one minimal meaningful unit has been replaced by another, we identify the
morphs involved as [Link] of distinct morphemes. So. in (3.7]on p. 36 the indefinite article realised
by a or an is a distinct morpheme from Lhe definite: article realised by tl1t! since a semantic difference is
detectable when" or011 isreplaced wilh tht!.
A further example of contra.-;! is given in [3.10]:

13.101
a, I unlocked the door. b. She is untidy.
I re-loded lhc door

The two sentences in [3.I0aJ mean very different Lhings. Since they are identical except for Lhe fact that
where one has u,1- the other ha.s re-. the difference in meaning between Lhese 1wo sentences is due to the
difference in 1neaning between the morphemes realised by re- (meaning 'do again') and 1111- (meaning
're\'Crse the action').
Now. contras! the w1- of unlocked with lhe 1111- of 11111idy. In both cases we have 1he same morph ,m-
(which is spelt and pronounced in exactly the same way). But it is obvious that ,,,,. represents different
morphemes in these two word-forms. Ln / 1111/ocked the door the morph 1111- found in unlocked realises a
reversive morpheme which is attached to verbs-it reverses the action of locking. But in untidy it realises
a negative morpheme allached to adjectives- untidy means 'not tidy·.(If a person is umidy, it does not
mean that at some earlier point they were tidy and someone has reversed or undone their tidiness.)
If morphemes were m:.tde up of phonemes a simple correlation of morphs with morphemes is what we
would find. But, in fact. it is quite common for the same phonological form {i.e. morph) to represent more
than one morpheme. It is from thecontext that we can tell which morpheme it represents. This is the second
piece of evidence against the assumption that morphemes arecomposed of phoneme.,;.
The complex relationship between morphemes and [Link] allomorphs that represent them gives u.s a
window through which we can glimpse one of the most fascinating aspects of language: the relationship
be1ween FORM and FUNCTION. In linguistics we explore lhe fom1of various elements of language
structure. e.g. words and sentences. because it is important to know how they arc constructed.
However. form is not
everything. We are also interested in knowing what linguistic elements are used for. what function they
serve.
Just consider for a moment this non-linguistic analogy. Imagine a friend returns from a foreign vacation
with two beautiful oman1ental glass containers with a globular shape and gives one to you as a present and
keeps the other for herself. She does not tell you what your present is used for. She uses hers as a vessel for
containing wineat thetable-she gotthe ideaof buying these containers when she wasserved winein a similar
container in a fancy restaurant. You do not know this. You look at your present and decide lo put it on the
26 CLOSE f,.NCOUN'TUS OP A MOH.l'tlF..MIC KIND

1ablc as a container for cut fresh flowers. She calls hers n flagon. for 1ha1 is wha1 she is using ii as. You
call yours a va$C.
Herc are thequestions now: arc these objects •flagons' or ·va.�c�·1 Which oneof you is riglu1 I am not
being evasive if I say that txuh of you :lte right. For. ahhough the two objf..-CIS are identical as fur as their
fonn. lheir physical properties. is concerned. they arc very differcnl with regard to the functions that they
serve in your two households.
TI1erc arc numerous linguis1ic parallels. What is physically the same linguistic fonn can be used 10
represent dislinct morphcmc1o. In order for fonn1o 10 be regarded as atlomorphs belonging to 1hc s:amc
morpheme, it is not �ufficient for 1hem 10 have the same fom1 -to be pronounced or wriuen in the same
way. They must also have the same gramm:aticul or semantic function. The significance of this point was
hinted at in 1he discussion of ,,,,. in 1mloclu�d and 11mid,v when we .showed that the �mne morph can
[Link]:.clll differcn1 morphemes. It l!hould become even more obvious when you consider1he form •ff in
the following:

ll,J11
,. thmk - lhinl..cr dme-drivcr
writc-wri1er sing - singer
[Link] - [Link] 1oCII - seller
b. cool - cooker :-train- Mrainer
receive - rccc1\cr compute - compulcr
propel - pro1,cller cra..c - cra)<,cr
,. l..ondon - Londoner nor1h - n0r1hcmcr
Iceland - lcehmder ca:.t - c�tcmer
New Yort.- New Yorker Highland:.- HIJ;,hlandcr

The .same fonn. -u, represe111s 1hrec different meanings und hence has 10 be assigned to 1hrce di:.linct
morphemes. In (3.11al it forms an :1gen1ive noun from a verb, with the meaning 'someone who does X'
(i.e.
whalcvcr the verb means). In !3.J lbl the same -er fonn.!> an instrumental noun from a verb. wilh the
meaning 'something used 10 x·
(i.e. to do whatever the verb means). Finally. in (3.1lc} thesame -er form is
attached 10 n noun referring lo a pince 10 mean ·an inhabitant or.
Ocarly, the same form does serve differenl func1ions here. So. it realise.,;; different morphemes. This is
further evidence that should quickly disabuse us of1he assumption that morphemes are made up of morphs.
Not only can a single morpheme have several aHomorphs (a.,;; in1hc case of the plural morpheme). the
same
morph (e.g. -er) can represent different morphemes. There is no simple one-to-one matching of
morpheme.-. with morphs.

J.4
FREEDOM ANO IIONOAGE
When we classify rno11>hemes in terms of where they are allowed 10 appear. we find thal they fall in10
two major groupings, Some morphemes are capable of occurring on their own as words. while 01her
morphemes nre only allowed lo occur in combination wi1h some other morpheme(s) but they c:mnot be
used by [Link] as indcpendcnl words.
 [Link] WORDS 27

Those morphemes that are allowed 10 occur on their own in sentences as words are called 1-"'REE
MORPHEMES while 1hosc morphemes 1hat musl occur in the company of some 01her morphemes are
called BOUND MORPHEMES. ln (3.12) the bound morphemes are italicised.

13.12]
p<>I pes1(i)-cide
modem [Link]-modem-is,
child child-ish
pack pre-pacJ...-ed
lau•h laui::h-ill

The free morphemes in [3.12] can all be manipulated by syniactic rules; they can stand on their own as
words. By contrast. it is impossible to use the forms -cide. po.-.1-, -i.-.t, -[Link], pre-. -ed or -ing, independently.
So far,all the examplesof freemorphemes that functionas roolli that wehaveencountered havebeencontent
words (see p. 14). However, nOI all free morphemes are content words. Some are employed to indicate
grammatical functions and logical [Link] rather than to com·ey lexical or cognitive meaning in a
sentence. Hence such words are [Link] FUNCTION WORDS. They include wordssuch as the following:

13.131
article!>: a/[Link]
[Link]\'C.1,: [Link]. 1ha1, thrsr andthosr
pronouns: e.g./, yo11,"'"'·[Link])', yo11r. his, hrrs, M"lw etc.
prepo�i1ions: [Link], imo, 011 to, m. 01, elC.
conjunctio1h: e.g.a11d, or. b111, b«mue i(etl.'.

In ordinary language use such words are extremely common. But on their own they would not convey a
lot of information. If you received a telegram like Bw ii my 011 10 the itt you might suspect that the sender
ei1her had a strange sense of humour or was not menially sound.

3.5
SOUND SYMBOLISM: PHONAESTHEMES AND ONOMATOPOEIA
In the va.�t majority of words, the relationship between sound and meaning is arbitrary (seep. 2). lnere is no
reason why a particular sound. or group of sounds, should be used to represent a particular word, with a
particular meaning. If someone asked you what [b) in bed or lstr] in strange mean!, you would think they
were asking a veryoddqueslion. As a rule, sounds qua sounds do not mean anything.
However, the general principle that says that lhe link between sound and meaning in words is arbitrary is
occa�ionally dented. This happens in two sets of [Link]. First. certain individual sounds, or groups
of sounds, which do not represent a specific enough meaning to be called morphs nevenheless appear10 be
vaguely associated with some kind of meaning. Such sounds are Called PHONAESTHEMES.
As our first example of a phonacstheme. let us take the RP vowel I ] (which is historically
descended from (U]. the vowel that is still used in word.�likedull and /1111in the north of England). This
phonaestheme is found in words associated with various kinds of dullness or indistinctness.e.g. dull.
th11d. 1/umder. d11sk,
28 CLOSE. ENCOUNTERS OF A MORPHEMIC KIND

blum, mud. slush, sludge. slump etc. Obviously. the vowel I J per se does not mean 'dull'. If it did, dim
which contains the vowel [11 would not bea virtual synonym for dull.
Many words which mean 'to talk indistinctly' contain oneor more occurrences of the labialconsonant [mJ,
which is made with the lips fimdy closed. preventing clear articula1ion. Thal way. lhe very act of
pronouncing the word iconically mimics a key aspect of its meaning. Youcan see this if you watch yourself
in a mirror saying words like 1111mrble, mw11111r, mutter. muted, grumble etc. It is probably not an
accident that these words alsocontain [Link] phonaestheme L ]. Similarly. thesound L mp] (spelled -ump)
as in chimp, dump, bump, lump and Jwmp is often found at the end of words which are associated with
heaviness and clumsiness although no one would wish to suggest that -ump in itself represents the ideas of
heaviness and
clumsiness. Interestingly. here again we have the vowel l ] followed by the labial consonants [mp].
Observe also that whereas [ I tends to have associations of heaviness or dullness. the high front vowels
li:] and nJ frequently occur as phonaesthemes in words associated with smallness, as in wee. 1ee11_,·-
wee11y, lea11, meagre. mini, 1/,i11 and liu/e. (The fact that big has the opposite meaning just goes to
show that
phonaesthemes only represent a tendency.)
Second. and more importantly, in addition to phonaesthemes. there are onomatopoeic word'\ in which a
direct association is made between the sounds of a word+forrn and the meaning that it represents. In cases
of ONOMATOPOEIA, the sounds (qua sounds and not as morphs) symbolise or renect some aspect of the
meaning of the word that they represent. So. if speakers of any language want an onomatopoeic word for
the noise a cal makes, they will not choose a noise like bimbobam [Link], in the land of the
Ning
Nang Nong.
The words for sounds made by various animals e.g. neigh. mioow. moo etc. are the most obvious
examples of onomatopoeia. Bui there are others such as roor. crack. clang. bang. splash.. n..,ish, whoosh.
buz.:., hiss,cheep, bleep, gurgle. plop and plod. In the case of onomatopoeic words. the relationship between
sound and meaning is to some exlent ICONIC. The sounds mimic an aspect of the meaning of the linguistic
sign much in the same way thal this iconic sign for a restaurant represents, more or less directly, the
meaning ·restaurant'. This symbol is still conventional to some degree. To people who eal with chopsticks.
it might nol be immediately obvious why this sign represents a re..staurant (rather than a cutlery shop), but
once it is pointed oul the link can beseen quite ea.'\ily.
Onomatopoeic words are iconic in so far as theydirectly reflect some aspect of the meaning of what they
stand for. So, conventiona!Jy in English cows go ·moo' and horses go ·neigh' and bees go ·buzz'. lliat is
why Spike Milligan·s nonsense poem ·on the Ning Nang Nong' is bizarre.
To be onomatopoeic. lhe sound must imitate to some degree an a'\pect of the noise made by the bird or
animal. But exactly whal is imitated will vary from language to language. An English cock will say
cockodoodledoo, a Russian cock kuk11rik1t and in Uganda it may say kookolilookoo.(These differences are
not attributable lo dialectical variation among the males of the Go/hl.f domesticus species.) Onomatopoeic
words are not purely and simply fom1ed by mimicking prech,ely the meanings that they convey. To some
extent onomatopoeic word'\ are also moulded by linguistic convention. That is why in different places in lhe
world different onomatopoeic words may be used for the same animal or bird noise.

3.6
VERBAL BLUEPRINTS
Linguistic theOI)' incorporates the hypothesis that there are universal principles of grnmmar that regulate the
amount of variation in linguistic structure across languages. In the last section we saw the marginal role
played by sound symbolism in word-fonnation. This does not obscure the fac1 that normally languages form
30 CLOSE ENCOUN1'ERS Of' A MORl'llhMIC KIND

words by using sounds in n non-imitative way. There is an overriding 1endcncy for the relationship
between sounds and meanings to be arbitrary. Nonnally there is no reason why a particular morpheme is
realised by any particular sounds. The choice of the allomorph or allomorphs that reprcsen1 a panicular
morpheme is amilrary.
Obviously, as everyone knows, nil languages do not have 1hc same words. Since vinually any arbitrnry
ma1ch of sound and meaning can produce a word. it is not surprising lhat words vary greatly in their
structure across languages. But this does not mean that chaos reigns. The ways in which morphs are used 10
form words is regukucd by general principles. So. the amount of crosslinguis1ic variation in word-formalion
foils within certain broad parameters. It is as if there is a menu of blueprints for word-formalion from which
all languages make their selection.�:

13,141
(i) ISOLATING (or anuly11c) ltmguagcil
(11) AGGLUTINA1'1NG languages
(iii) lNFLECTING (or !lyntheuc) languages
(iv)POLYSYNTH&TIC languages

No language mukes all its choices from just one pan of the menu. To \'arying degrees all languages make
mixed choices. l'hc idea of 1hi.s menu is to indicale the predominant word-formation tendencies. if they
exis1. In the subsection.�below we shall consider in turn examples of the differcnl morphological types.

3.6.1
Tin)' words(isolating lant,:uagcs)
In an archetypical isoliuini; language the word is virtually indist"inguishable from lhe morpheme. for e\'cry
word conloins just one morpheme. Every morpheme is a free morpheme. TI1erc are no bound morphemes.
Vietnamese comes close 10 this ideal:

1).151
Vietnamt)I.C
a. T6i qua· b6ng vi hn
I lick pa.,1 cla.):-.. ball and he punch
·[Link] ball and he punched me:
b. Ch Ung 1{J,1 mua ii g 0
Pl. buy p31'1 rice
·we bou I thenee.'

Typically, the words arc short and coniain [Link] one morpheme each. Almo�I every concept is expressed
by a separate word. Look again. for example. at lhc lreatmcnl of pa.,t tense in verbs (e.g. punched, !,ought)
:md 1he plurality of u·e (plural plu.$ first person),
 l:.NCLISII WOkOS 29

On the Ning Nang Nong

On 1hc Ning Nang Nong
Where 1he cows go Bong!
And 1he Monkeys all say Boo!
There's a Nong Nang Ning
Where the trees go Ping!
And 1he 1ea pols Jibbcr Jabber Joo.
On 1he Nong Ning Nang
All the mice go Clang!
And you jus1 can'1 catch 'cm when they do!
So it's Ning Nang Nong!
Cows go Bong!
Nong Nang Ning!
Trees go Ping!
Nong Ning Nang!
The mice go Clang!
Wha1 a noisy place 10 belong,
Is 1hc Ning Nang ing Nang Nongl
SPIKE MILLICAN