Introduction To English Morphology

Textbooks in English Language
and Linguistics (TELL)

Edited by Magnus Huber and Joybrato Mukherjee
Volume 5
PETER LANG
Frankfurt am Main · Berlin · Bern · Bruxelles · New York · Oxford · Wien
Alexander Tokar
Introduction
to English Morphology
PETER LANG
Internationaler Verlag der Wissenschaften
Bibliographic Information published by the Deutsche
Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the
Deutsche Nationalbibliografie; detailed bibliographic data is
available in the internet at http://dnb.d-nb.de.
Cover Design:
© Olaf Gloeckler, Atelier Platen, Friedberg
ISSN 1862-510X
ISBN 978-3-631-61841-7 (Print)
ISBN 978-3-653-01564-5 (E-Book)
DOI 10.3726/978-3-653-01564-5
© Peter Lang GmbH

Internationaler Verlag der Wissenschaften
Frankfurt am Main 2012
All rights reserved.
All parts of this publication are protected by copyright. Any
utilisation outside the strict limits of the copyright law, without
the permission of the publisher, is forbidden and liable to
prosecution. This applies in particular to reproductions,
translations, microfilming, and storage and processing in
electronic retrieval systems.
www.peterlang.de
To Dieter Stein
Preface
This textbook grew out of the seminars and lectures on various aspects of
English theoretical linguistics (e.g. morphology, morphosyntax, lexicology,
phraseology, phonetics and phonology, etc.) which I taught at the Universities of
Düsseldorf and Gießen between April 2006 and February 2011.
Like the previous volumes in the TELL Series, this introduction to English
morphology is intended as a companion for students of English language and
linguistics throughout their studies. This means that the book contains a
discussion of both 1) very basic introductory issues requiring no or very little
prior background in linguistics (e.g. what is a morpheme?) and 2) fairly
controversial theoretical issues such as, for example, the question of whether a
fully-idiomatic word like understand can be segmented into the morphemes
{under} and {stand}, to which different linguists provide different answers. The
textbook can thus be used by students at both B.A. and M.A. levels.
The book consists of six chapters. Chapter 1 is an introductory chapter that
provides a definition of morphology as a branch of linguistics. Chapter 2 dwells
on the internal structure of English words. Chapter 3 discusses the formal and
semantic structure of English lexemes. Chapters 4 and 5 are concerned with
word-formation in English: special emphasis is laid on the methodological
question of how students of English morphology can determine whether a
particular word-formation mechanism is still productive in Present-day English.
Finally, Chapter 6 discusses grammatical categories in English such as tense,
aspect, voice, number, case, etc. Each of the chapters ends with exercises and
suggestions for further reading.
The present textbook departs from other recent introductions to morphology
in the following important respects:
x Inspired by the recent article Haspelmath (2011), it rejects the orthographic

criterion for distinguishing between free and bound forms.
x Following Mel'čuk (1968), it defines a linguistic sign not as a doublet

consisting of the signifier and the signified (as defined by Ferdinand de
Saussure), but as a conventionalized association between the signifier, the
signified, the syntactics, and the sociolinguistics.
x It classifies word-formation into lexeme-formation and lex-formation.
Like the authors of the previous TELL volumes, I have attempted to illustrate
my analyses with authentic language data, i.e. examples drawn from linguistic
VIII Preface
corpora such as the Corpus of Contemporary American English and British

National Corpus. In addition, I have extensively consulted the Oxford English
Dictionary and Paul McFedries's online-based collection of English neologisms
Word Spy (http://www.wordspy.com/).
I would like to thank Professor Magnus Huber for inviting me to write this
volume for the TELL Series and for providing extremely helpful comments on
an earlier version of this manuscript. I am also deeply indebted to Wiebke
Ostermann and Louise Nieroba for their scrupulous proof-reading of the final
version of the manuscript.
This textbook has undoubtedly benefited from my conversation with
Professor Vladimir Plungian in February 2010 in Moscow. Vladimir Plungian is
the author of Introduction to General Morphology (Plungian 2000), the textbook
which has sparked off my own interest to morphology and was a great source of
inspiration for me when I was writing this book.
Last but not least, I wish to thank Kate Butkus, Sergey Danilov, Patrick
Maiwald, Bridgit Nelezen, and Stefan Storm for insightful discussions and many
valuable comments on some theoretical issues relating to this volume.
I alone am responsible for any remaining errors and shortcomings.
This book is dedicated to Professor Dieter Stein. I would like to congratulate
him on his 65th birthday and wish him many more years of good health and
successful linguistic research.
Düsseldorf Alexander Tokar

November 2011
Table of Contents
1 Basic concepts 1
1.1 What is morphology? 1
1.2 Morphology and other branches of linguistics 3
1.2.1 Semantics 3
1.2.2 Phraseology 5
1.2.3 Phonetics and phonology 7
1.2.4 Syntax 9
1.2.5 Sociolinguistics 12
1.3 What is a word? 14
1.4 Exercises 21
1.5 Further reading 22
2 The internal structure of English words 25
2.1 What is a morpheme? 25
2.2 Morphemes as signs 27
2.2.1 One signifier Æ more than one signified 28
2.2.2 One signified Å more than one signifier 28
2.2.3 The syntactics of a sign 29
2.2.4 The sociolinguistics of a sign 30
2.2.5 The signified as the most important sign component 31
2.3 The distribution of morphs 33
2.4 The segmentation of words into morphemes 39
2.4.1 Anisomorphism. Full-idiomaticity 39
2.4.2 A purely semantic approach 40
2.4.3 Nida's purely formal approach 40
2.4.4 Nida's approach and the conception of differential meaning 43
2.4.5 Mel'čuk's theory of quasi-linguistic units 44
2.4.6 Anisomorphism. Partial idiomaticity 45
2.4.7 Anisomorphism. Additional meanings 47
2.5 The hierarchy of morphs and units alike 50
2.5.1 Affixes versus roots 51
2.5.2 Combining form as a distinct morpheme type? 53
2.5.3 One signifier Æ both a root and an affix 55
2.5.4 Typology of affixes 56
X
2.5.5 Typology of roots 59

2.6 Exercises 60
3 Analyzing English lexemes 63
3.1 What is a lexeme? 63
3.2 The structure of a lexeme 64
3.2.1 The lex of a lexeme 64
3.2.2 The typology of lexes 66
3.2.3 The signified of a lexeme 66
3.2.4 Three-component anisomorphic lexemes 67
3.2.5 Anisomorphic lexemes realized by phrases and sentences 68
3.2.6 How to distinguish between full-, semi-, and quasi-idioms? 70
3.3 Lexemes and vocables 72
3.3.1 Relations between members of the same vocable 72
3.4 Lexemes and lexeme families 74
3.5 Exercises 75
4 Word-formation: basic issues 79
4.1 Lexeme-formation versus lex-formation 79
4.2 Lexeme-formation 80
4.2.1 Purely semantic mechanisms 80
4.2.2 Purely formal mechanisms 81
4.2.3 Mechanisms involving formal and semantic modifications 83
4.2.4 Diachronic and synchronic perspectives 86
4.2.5 Why do speakers of English create new lexemes? 89
4.2.6 The establishment of new lexemes 92
4.2.7 The non-institutionalization of new lexemes 94
4.2.8 Productivity 99
4.3 Lex-formation 103
4.3.1 Lex-forming clipping 104
4.3.2 Lex-forming suppletion 104
4.3.3 Lex-forming abbreviation 105
4.3.4 Lex-forming borrowing 106
4.3.5 Lex-forming apophony 106
4.3.6 Lex-forming affixation 106
4.3.7 Lex-forming syntactics' change 107
4.3.8 Lex-forming orthographic modification 108
XI
4.4 Exercises 108

5 Lexeme-building mechanisms 111
5.1 Semantic change 111
5.1.1 Mechanisms of semantic change 112
5.1.2 Types of metonymies 112
5.1.3 Types of metaphors 114
5.1.4 Morphological conversion 115
5.2 Lexeme-manufacturing 127
5.3 Lexeme-building borrowing 130
5.4 Lexeme-building affixation 133
5.4.1 Affixes and their signifieds 133
5.4.2 Affixes and their syntactics 134
5.5 Lexeme-building apophony 143
5.6 Compounding 145
5.6.1 Compounding as an anisomorphic mechanism 146
5.6.2 The semantics of compounding 148
5.6.3 Endocentric and exocentric compounding 152
5.6.4 Compounding from a formal point of view 156
5.6.5 Compounds and phrases 159
5.7 Blending 163
5.8 Idiomatization of phrases and sentences 166
5.9 Back-formation 168
5.10 Exercises 171
6 Inflectional morphology 173
6.1 Grammatical category 173
6.2 Types of grammatical categories 174
XII
6.3 Wordform-building mechanisms 176

6.3.1 Inflectional affixation 176
6.3.2 Analytic formation 177
6.3.3 Grammatical apophony 177
6.3.4 Grammatical suppletion 180
6.3.5 Signifier-sharing 181
6.3.6 Allowordforms 181
6.4 Syntactic grammemes in English 183
6.4.1 Why do we need the passive voice? 183
6.4.2 Middle voice in English? 184
6.4.3 What can be passivized? 186
6.4.4 Get-passives 187
6.4.5 Types of cases 188
6.4.6 Cases in English 189
6.4.7 Functions of case 191
6.4.8 Semantic function of case? 192
6.5 Semantic grammemes in English 194
6.5.1 Typology of temporal meanings 194
6.5.2 No future tense in English 196
6.5.3 Idiomatic uses of temporal wordforms 198
6.5.4 Typology of aspectual meanings 200
6.5.5 Aspects in English 202
6.5.6 Moods in English 205
6.5.7 Person 209
6.5.8 Number 212
6.5.9 Degrees of comparison 214
6.5.10 Numerical qualification 216
6.6 Exercises 218
Key to exercises 221
References 225
Index 233
1 Basic concepts
It has become a tradition to begin monographs and textbooks on morphology

with a tribute to the German poet Johann Wolfgang von Goethe, who invented
the term Morphologie in 1790 to refer to "eine Anschauung von den Gestalten
und Wandlungen der Natur und Kunst" / 'a view about the forms and
transformations of nature and art' (Kluge 2002). In the English language the
word morphology has been used since 1828: according to the Oxford English
Dictionary (henceforth OED), it originally referred to "the branch of biology
that deals with the form of living organisms and their parts, and the relationships
between their structures".
In this introductory chapter we will learn in which respects our present-day
linguistic understanding of the term 'morphology' differs from that of Goethe
and that of the 19th century biologists. In addition, the chapter discusses the
relationship between morphology and other branches of linguistics as well as
one of the most important theoretical concepts in morphology: the concept of a
word. With regard to the latter, our focus will be on the question of whether
there is at least one operational criterion, or to be more precise, a formal feature
characteristic of a particular combination of sounds justifying the treatment of
that combination as a word. For example, why is it that the combination of the
sounds /kæt/ is usually regarded as a word? What is the difference between the
word cat and, say, black cat, which is usually regarded as a combination of no
less than two words: black and cat?
1.1 What is morphology?
In a linguistic context morphology is usually defined as the study of the internal

structure of words. To illustrate what this means, let us consider the word
waithood 'the stage in a young college graduate's life when activities such as
marrying and finding a place to live are postponed until a job is found or enough
money is saved'. According to Word Spy (i.e. a freely available online database
of English neologisms which can be accessed at http://www.wordspy.com/),
waithood is a relatively new word in English: its earliest citation provided by
Word Spy dates September 01, 2007.
From a formal point of view, waithood seems to be segmentable into the
components wait and -hood. The component wait is a verb which, apart from
occurring in waithood, also occurs in sentences like I cannot wait any longer.
And the component -hood is a noun-building element which, apart from
occurring in waithood, also occurs in nouns like adulthood, boyhood,
2 Chapter 1
parenthood, etc. Taking this into account, we can draw an important conclusion,
as far as the internal structure of waithood is concerned. The word under
analysis is a complex word, i.e. a word which can be analyzed as a combination
of no less than two formally identifiable components capable of occurring in
other morphologically relevant environments. (In the next section of this
chapter, we will learn which environments qualify as morphologically relevant.)
By contrast, the combination of the sounds /kæt/ representing the word cat
cannot be segmented into /k/ and /æt/, or /kæ/ and /t/, or /k/ and /æ/ and /t/.
Undeniably, these putative components occur elsewhere (e.g. cable, at, cap,
mat), but none of these environments seem to be morphologically relevant.
Accordingly, the word cat is a simple word (or a simplex), i.e. a word which
consists of no more than one formally indivisible component.
In addition to determining whether the word under analysis is a simple or
complex word, morphology also deals with the question of how new words like
the above mentioned waithood come into existence. Compare, for example, the
formation of waithood and that of the verb to wife 'to downplay a woman's
career accomplishments in favor of her abilities as wife and mother' (Word Spy).
As we established above, waithood is, from a formal point of view, segmentable
into the components wait and -hood. Accordingly, we can conclude that the
word under analysis was created via combining the components wait and -hood:
waithood = wait + -hood. By contrast, the creation of the formally simple verb
to wife seems to have involved a semantic modification of the already existing
noun wife. Thus the meaning 'to wife' is semantically more complex than the
meaning 'a wife': the former contains the meaning 'a wife' plus the additional
meaning 'to downplay a woman's career accomplishments in favor of her
abilities as mother'. Accordingly, we can conclude that the meaning 'a wife'
served as an input meaning for the meaning 'to wife'.
Finally, morphology studies those modifications (of existing words) that do
not give rise to new words but serve to express grammatical meanings such as,
for example, 'plurality', 'the past tense', 'the passive voice', etc. For instance, the
addition of -s to the noun book does not create a new word: both book and books
refer to representatives of the same class of objects (books). However, book and
books differ with regard to their grammatical meanings: while book is in the
singular number, books is in the plural number. The singular–plural
opposition exemplified by book and books forms the grammatical category
NUMBER. A similar case is the present tense–past tense opposition, which
forms the grammatical category TENSE: both work of I work and worked of I
worked can refer to the same action of working. However, the forms under
consideration differ with regard to their grammatical meanings: while work is in
the present tense, worked is in the past tense.
The present textbook contains a detailed discussion of each of the three
aspects of morphology named above. Chapters 2 and 3 deal with both the formal
Basic concepts 3
and the semantic structure of English words: among other things, we will
become acquainted with the types of components into which complex words like
waithood can be segmented. Chapters 4 and 5 deal with what has usually been
called word-formation, i.e. processes that produce new words like waithood
and to wife. Finally, Chapter 6 deals with grammatical categories in English
such as NUMBER, TENSE, VOICE, etc.
1.2 Morphology and other branches of linguistics
To be able to address the aforementioned issues, we will, first of all, need to

become acquainted with the most important theoretical concepts pertaining to
other branches of linguistics which morphology is closely connected to. These
include semantics, phraseology, phonetics and phonology, syntax, and
sociolinguistics.
1.2.1 Semantics
Semantics is the branch of linguistics which is concerned with meaning. To

demonstrate in what ways morphology is dependent on semantics, let us again
compare the internal structure of the words waithood and cat.
The reason why the occurrence of wait in sentences like I cannot wait any
longer and that of -hood in nouns like adulthood, boyhood, parenthood, etc. can
be considered morphologically relevant is that the overall meaning 'waithood' is
closely connected to the meanings 'to wait' and 'stage', which the components
wait and -hood express in other environments: e.g. while adulthood can be
defined as the stage in our lives that involves being an adult (or, alternatively,
the state of being an adult), waithood can be defined as a stage in the life of a
college graduate that involves waiting. This fact is the main justification for the
formal segmentation of waithood into the components wait and -hood.
In contrast, there is no similar justification for the segmentation of cat into
/k/ and /æt/, or /kæ/ and /t/, or /k/ and /æ/ and /t/. In neither cat nor words like
cable, at, cap, mat, etc., where these sounds occur as well, do they express
discernible meanings of their own. Thus the meaning 'cat' does not seem to be a
complex meaning segmentable into at least two independent meanings – e.g. the
meanings 'animal' and 'characteristic features of cats distinguishing them from
other animals' – which one could attribute to the putative components /k/ and
/æt/ or /kæ/ and /t/. What justifies this claim is that other words containing these
sounds (or sequences of these sounds) do not express related meanings. For
example, neither the meaning of cable, which contains the sound [k], nor the
meaning of at, which contains the sound sequence [æt], seem to have anything
4 Chapter 1
in common with the meaning 'cat'. The same is true of the word cap, which
contains the sound sequence [kæ], and the word mat, which contains the sound
[t]: neither the former nor the latter are semantically related to cat.
In summary, a morphological analysis of the internal structure of a word
depends on the semantic analysis of the meaning of that word. Accordingly, a
morphologist analyzing the internal structure of some word must have a very
clear idea of what that word means. But what precisely is meant by the meaning
of a word? What does it mean to mean something? How do we know that e.g.
the word waithood means 'a particular stage in the life of a college graduate that
involves waiting'?
In semantics meaning is usually defined as the concept associated with a
particular sound form. A concept is a mental description activated by that sound
form. For instance, those speakers of English who are familiar with the word
waithood have the concept WAITHOOD stored in their mental lexica. That is, as
soon as they hear the sound form /ˈweɪthʊd/, their minds 'picture' a college
graduate who postpones marrying until he or she finds a good job or saves
enough money.
The mental lexicon is often described as a kind of dictionary (which we have
in our brains) consisting of multiple sound form–concept correspondences like
the one exemplified by waithood. That is, the mental lexicon of the average
speaker of English contains the correspondences between e.g. the sound form
/kæt/ and the concept CAT (i.e. a description of an animal called cat) associated
with it; the sound form /ˈtiːtʃə(r)/ and the concept TEACHER (a description of a
person who teaches); the sound form /dɪˈmɒkrəsɪ/ and the concept DEMOCRACY
(a description of a democratic political system); etc.
The key task of semantics is thus the description of the concept constituting
the meaning of a particular linguistic expression. For this purpose, a semanticist
can resort to two strategies. One is the identification of so-called necessary and
sufficient conditions. These are the 'minimum requirements' whose fulfillment
suffices to qualify as a member of a particular conceptual category. For
example, it is often argued that the minimum requirement that is fulfilled by all
representatives of the conceptual category MOTHER is that of being a female
parent: any entity who is both female and a parent can be referred to as
someone's mother. Accordingly, the features [female] and [parent] can be said to
constitute the meaning of the word mother.
An alternative to this is the prototype approach. Its essence is the
identification of characteristics applying to the best representative of a given
conceptual category, its prototype. A prototypical mother, for instance, is a
person
who is and always has been female, and who gave birth to the child,
supplied her half of the child's genes, nurtured the child, is married to
Basic concepts 5
the father, is one generation older than the child, and is the child's
legal guardian. (Lakoff 1987: 83)
One of the most controversial issues in semantics is the question of whether

linguistic meaning can be equated with encyclopedic knowledge. That is, for
example, can we really describe the meaning of mother in terms of the feature
[supplied her half of the child's genes], given that a number of people – e.g.
small children who have not yet had a biology class – are not really aware of this
fact but nevertheless can easily distinguish between their own mothers and other
females?
According to Wierzbicka (1972: 46), mother means 'the human being inside
whose body (once) there was something that was becoming another person's
body'. (The word father then means 'the human being who once caused a woman
to have inside her body something that was becoming another person's body'.)
Undeniably, this is a much less encyclopedic definition than that which makes
use of the feature [supplied her half of the child's genes]. However, it is not non-
encyclopedic: the fact that a mother is a person who inside her body once used
to have the body of another person is also an encyclopedic fact that may not be
shared by all members of the English linguistic community.
1.2.2 Phraseology
Phraseology is the branch of semantics which is concerned with idiomatic

meanings (in particular, with combinations of more than one word such as e.g.
kick the bucket which have idiomatic meanings). According to Dobrovol'skij and
Piirainen (2005: 40), there are two types of idiomatic meanings: those that
involve semantic modification or opacity of the literal meaning of the form
under analysis. (The term 'literal meaning' can be defined negatively: a literal
meaning is neither an opaque nor a modified meaning.)
The difference between the two types of idiomatic meanings is as follows. In
the case of semantic modification, the average speaker (i.e. a layman lacking
any expertise in semantics and phraseology) can easily explain to him- or herself
why the idiomatic word in question does not mean what its components literally
stand for. By contrast, in the case of opacity, the idiomatic meaning is not
transparent, i.e. cannot be accounted for in terms of the components' literal
meanings.
As an illustration of a modified meaning, let us consider the meaning of
boyfriend in (1).
(1) She's decided to break up with her boyfriend (Corpus of Contemporary

American English, henceforth COCA)
6 Chapter 1
From a formal point of view, the word boyfriend seems to be segmentable into
the components boy and friend. However, the meaning 'boyfriend' cannot be
segmented into the meanings 'boy' and 'friend'. First of all, a boyfriend is not a
boy (i.e. a young male child, which boy literally means) but a male of almost
any age over puberty (Holder 2008: 104). Second, a boyfriend is not literally a
friend (i.e. any person whom a woman 'know[s] well and regard[s] with
affection and trust', as defined by WordNet) but a regular sexual partner in a
non-marital sexual relationship. The meaning 'boyfriend' is thus a fully-
idiomatic meaning that does not contain the meanings 'boy' and 'friend', inherent
in the components boy and friend in other environments.
At the same time, however, notice that the literal meanings 'boy' and 'friend'
partially motivate the idiomatic meaning 'boyfriend': boyfriends are often
perceived and explicitly referred to as friends in the literal meaning of this word;
e.g. in (2).
(2) My boyfriend is my best friend (http://tinyurl.com/6ys8pzz)
Similarly, the idiomatic meaning 'a male of almost any age over puberty' is
related to the meaning 'a young male child' in that both the former and the latter
share the semantic component [male]: both a boyfriend and a boy are males.
Because of these two facts, the average speaker of English can easily explain to
him- or herself what boyfriends have in common with friends who are boys.
As an illustration of an opaque meaning, let us consider the meaning of the
verb understand. Like boyfriend, this word seems to be formally segmentable
into two components: under and stand. However, as in the case of boyfriend, the
meaning 'to understand' does not contain the meanings 'under' and 'to stand':
understand does not mean 'to stand under somebody or something' but 'to grasp
the meaning or the reasonableness of something' (Merriam-Webster Online,
henceforth MWO). But while a boyfriend is typically a friend and a former male
child, understanding does not seem to have much in common with physical
standing: standing under somebody or something does not necessarily result in
the understanding of the person or the thing (under which you are standing).
Apart from fully-idiomatic words like boyfriend and understand, there are 1)
partially idiomatic words like blackboard, whose overall meanings contain one
of their components' literal meanings and 2) words like waithood whose overall
meanings do not only contain their components' literal meanings but also some
additional, unpredictable idiomatic meanings.
With regard to blackboard, it is obvious that its overall meaning contains the
meaning of the component board but not of the component black. A blackboard
is not a black board but a board for drawing or writing upon with chalk (MWO).
Blackboards typically have dark surfaces (and are in this respect different from
Basic concepts 7
whiteboards1), but they are not always black: green blackboards, for instance, do
occur as well.
With regard to waithood, observe that its meaning does not only contain the
meanings 'to wait' and 'stage', inherent in the components wait and -hood:
waithood does not mean 'any stage in the life of any person that has something
to do with waiting' but 'a particular waiting stage: the one that involves a young
college graduate who is waiting for financial security in his or her life and
therefore postpones marrying'. The meaning 'waithood' is thus narrower than
that of the mere sum of the meanings 'to wait' and 'stage', inherent in the
components wait and -hood.
Idiomatic words like boyfriend and understand pose a particular theoretical
challenge for a morphologist analyzing their internal structure. Can boyfriend
and understand be regarded as complex words, even if boyfriend does not mean
'a friend who is a boy' and understand does not mean 'to stand under somebody
or something'? This question will be addressed in Section 2.4 of the next
chapter.
1.2.3 Phonetics and phonology
Phonetics and phonology are related linguistic disciplines which are both
concerned with sound. Phonetics studies any physical aspect of sound. For
example, how do we produce and perceive sounds? (The branch of phonetics
which studies the production of sounds is called articulatory phonetics; the
branch of phonetics which deals with the perception of sounds is known as
acoustic phonetics.) Phonology, by contrast, is concerned with that aspect of
sound that performs a specific linguistic function (Trubetzkoy 1969: 11). For
example, the sound [k] of cable makes this word distinguishable from table.
Similarly, the sound [ð] of that makes this word distinguishable from chat.
The meaning-distinguishing function exemplified by [k] of cable and [ð]
of that is the most important but not the sole function of sound. According to
Trubetzkoy (1969: 16), sounds can also perform the expressive function. This
means that sounds we produce characterize us as members of particular
(regional, social, etc.) groups. For example, General American (i.e. the
standard accent of American English) differs from Received Pronunciation
(the standard accent of British English) with regard to the pronunciation of the
word metal. In General American this word is almost indistinguishable from
1
Many English speakers are reluctant to use blackboard as a synonym of whiteboard.
Accordingly, it can be argued that the former means 'a dark board used for writing or drawing
upon with chalk', i.e. the meaning 'dark' is part of the overall meaning 'blackboard'. In this
connection, it is important to emphasize that this fact does not undermine the analysis of
blackboard as a partially idiomatic word: the meaning 'dark' is not the literal meaning of the
component black.
8 Chapter 1
medal: both words have the pronunciation /ˈmɛdl/2. In contrast, in Received

Pronunciation metal is pronounced /ˈmɛtlˌ/ and medal is pronounced /ˈmɛdl/.
Accordingly, we can say that if a native speaker of English expresses the
meaning 'metal' with the sound form /ˈmɛdl/ rather than /ˈmɛtlˌ/, this person is
(most likely) an American who has a General American accent. The sound [d] of
medal can thus be said to perform the expressive function of characterizing
those English speakers who pronounce metal /ˈmɛdl/ as members of the group
'General American speakers'.
The sounds [k] and [t] of cable and table, which make these two words
distinguishable from each other, can be said to realize two different phonemes.
A phoneme is usually defined as the smallest meaning-distinguishing linguistic
unit. A sound that realizes a given phoneme is its phone. If one and the same
phoneme is realized by more than one sound, we speak of allophones. Compare,
for example, the articulation of the t-sounds in take and eighth. In take, it has an
alveolar articulation, i.e. one which involves the raising of the tip of the tongue
(its active articulator) towards the upper alveolar ridge (passive articulator).
In eighth, the articulation is dental, i.e. the tip of the tongue is raised not
towards the alveolar ridge but towards the upper teeth. The alveolar [t] of take
and the dental [t̪ ] of eighth can be regarded as allophones of the same phoneme
because these two sounds cannot perform the meaning-distinguishing function:
if some speaker of English pronounces take with the dental [t̪ ] rather than with
the alveolar [t], this will not give rise to a new word, i.e. /t̪ eɪk/ will also be
understood as a sound form representing the word take.
Allophones of a phoneme can either be in complementary distribution or
in free variation. The former involves two or more sounds that occur in
different environments, thereby complementing each other. For example, the
dental [t̪ ] occurs before the interdental sounds [θ] and [ð], but not before vowels;
in the latter environment the preference is usually given to the alveolar [t]. As
regards free variation, consider the word economic. As McMahon (2002: 58)
points out, one and the same speaker of English can pronounce this word as
/ˌɛkəˈnɒmɪk/ on one occasion and as /ˌiːkəˈnɒmɪk/ on another occasion. The
initial vowel sounds [ɛ] and [iː] can thus be regarded as allophones of the same
phoneme which occur in free variation: these sounds are interchangeable in the
same environment without a change in meaning.
Phones that realize two different phonemes – for example, [k] of cable and
[t] of table – are in contrastive distribution. In connection with contrastive
distribution, notice that two sounds that are in free variation in one environment
can be in contrastive distribution in another environment. For instance, the
sounds [ɛ] and [iː], which are in free variation in economic, perform the
meaning-distinguishing function in the words hell /hɛl/ and heal /hiːl/.
2
Most pronunciation transcriptions used in this book are taken from the OED.
Basic concepts 9
As we will see in Chapters 2 and 3, the categories of contrastive distribution,

complementary distribution, and free variation elaborated on above can also be
applied at the morphological level, describing the distribution of words and their
meaningful components.
1.2.4 Syntax
Syntax is a branch of linguistics which is concerned with units larger than the
word. These include:
1. the phrase
2. the clause
3. the sentence
A phrase is either an individual word or (more often) a combination of words

capable of performing a syntactic function. The most important syntactic
functions are the subject and the predicate. Consider, for example, (3).
(3) The French are fighting in Afghanistan (COCA)
This sentence can be segmented into two phrases: the French, functioning as
subject, and are fighting in Afghanistan, functioning as predicate.
Phrases like the French of (3) functioning as subjects have a number of
formal properties that make them distinguishable from phrases performing other
syntactic functions. According to Huddleston (2002a: 236-237), these properties
are as follows.
x The default position of the subject in English is before the main verb (i.e. a
verb like fight of (3) that has a lexical meaning). Cf. The French are fighting
in Afghanistan and *In Afghanistan are fighting the French.
x In interrogative sentences the subject follows the auxiliary verb (i.e. a verb
like are of (3) that helps to express a grammatical meaning). Cf. Are the
French fighting in Afghanistan? and *Are in Afghanistan the French
fighting?
x The subject agrees with the verb in person and number. Cf. The French are
fighting in Afghanistan and *The French am fighting in Afghanistan.
10 Chapter 1
x The subject position can be filled by a pronoun in the nominative, but not in
the accusative or the genitive case. Cf. They are fighting in Afghanistan and
*Them are fighting in Afghanistan or *Their are fighting in Afghanistan.
x The subject of a declarative clause agrees in person and number with the
subject pronoun of an appended interrogative tag. Cf. The French are
fighting in Afghanistan, aren't they? and *The French are fighting in
Afghanistan, isn't he?
x The subject is generally an obligatory element. Cf. The French are fighting in
Afghanistan and *Are fighting in Afghanistan.
There are no similar formal criteria justifying the isolation of the predicate.
However, the predicate can be isolated negatively. The predicate is that part of a
sentence like (3) which does not qualify as the subject. In (3), for example, the
phrase are fighting in Afghanistan qualifies as the predicate simply because it is
not part of the subject the French.
The predicate usually consists of the predicator and the predicative. The
predicator is the verb which heads the predicate phrase; e.g. are fighting of are
fighting in Afghanistan. The predicative is the rest of the predicate phrase; e.g. in
Afghanistan of are fighting in Afghanistan. (There may be predicates consisting
of the predicator only. For example, are fighting of The French are fighting.)
Predicatives can be classified into:
1. objects
2. complements
3. adjuncts
Like subject phrases, phrases fulfilling these three functions have several formal
properties that make them distinguishable from each other. Compare, for
example, the underlined predicatives in (4) and (5).
(4) He met the President

(5) He became the President
The predicative the President of (4) functions as object of the predicator met. By
contrast, the predicative the President of (5) functions as complement of the
predicator became. Objects are different from complements in that they can
serve as subjects of associated passive clauses. Thus there can only be The
President was met by him, but not *The President was become by him.
Another important difference is that in contrast to the object, the
complement can be expressed by a bare noun (i.e. a noun that is used without a
Basic concepts 11
determiner like the or a) and an adjective. That is, we can say He became
President and He became important, but not *He met President and *He met
important.
The adjunct is different from both the object and the complement in that it is
fairly independent of the predicator with which it combines in the predicate
phrase. The adjunct usually provides additional information modifying the
action denoted by the predicator – The French are fighting in Afghanistan – and
is therefore generally non-obligatory. For example, The French are fighting is a
perfectly grammatical sentence, even though it lacks the place adjunct in
Afghanistan. By contrast, sentences like *He met and *He became seem to be
incomplete without an object / a complement like the President.
The most important element of a phrase is its head; all other elements are its
dependents (which modify or complement the meaning of the head). The head
of a phrase determines its morphosyntactic properties. For example, the phrase
the French of (3) is a noun phrase (henceforth NP) because it is headed by the
unexpressed but understood noun soldiers: The French are fighting in
Afghanistan means 'The French soldiers are fighting in Afghanistan'. If this were
not the case, there would be no justification for the use of the plural are. The
predicate phrase are fighting in Afghanistan is a verb phrase (henceforth VP)
because it is headed by the verb are fighting. If this were not the case, this
phrase would not be able to function as predicate of (3). Finally, the adjunct
phrase in Afghanistan is a preposition phrase (henceforth PP) because it is
headed by the preposition in.
Having discussed both the formal properties and the internal structure of (at
least some) phrases, let us now proceed to another key concept in syntax, the
concept of a sentence. A sentence can be defined as a syntactic unit larger than a
phrase. The minimum requirement that a combination of two phrases must fulfill
in order to qualify as a sentence is that one of these phrases functions as subject
and the other as predicate. A combination of two phrases functioning as subject
and predicate forms a clause. Sentences consisting of no more than one clause
are traditionally called simple sentences. For example, the sentence The French
are fighting in Afghanistan is a simple sentence because it consists of only one
clause: as pointed out above, it can only be segmented into the NP the French
functioning as subject and the VP are fighting in Afghanistan functioning as
predicate.
Simple sentences must be distinguished from both complex and compound
sentences. Compare, for instance, the sentences (6), (7), and (8).
(6) This is a good idea (COCA)

(7) I don't know if this is a good idea (COCA)
(8) [t]his is a good idea and he needs to do it as soon as possible (COCA)
12 Chapter 1
Like The French are fighting in Afghanistan, (6) is an example of a simple

sentence. It consists of only one clause This is a good idea, in which the NP this
functions as subject and the VP is a good idea as predicate. Sentence (7) is an
example of a complex sentence. Here the clause this is a good idea is embedded
in a larger clause I don't know if this is a good idea, in which it functions as
object of the VP don't know if this is a good idea (Cf. If this is a good idea is not
known by me). Finally, (8) is an example of a compound sentence. It represents a
coordination of two syntactically independent clauses this is a good idea and he
needs to do it as soon as possible.
As will be shown in Chapter 6, some grammatical categories in English are
purely syntactic categories. For example, there is only a syntactic but not a
semantic difference between the active sentence He met the President and the
associated passive clause The President was met by him: both the former and the
latter can be used to refer to one and the same meeting event. Similarly, there is
mainly a syntactic difference between the nominative pronoun he and the
accusative pronoun him. Both can refer to one and the same male person, but
while the nominative he usually fills the subject position (e.g. He met the
President, not *Him met the President), the accusative him can usually be found
in the object position (e.g. The President met him, not *The President met he).
1.2.5 Sociolinguistics
Sociolinguistics is the branch of linguistics which is concerned with free

variation at all linguistic levels (i.e. not only at the level of sound, which we
discussed in 1.2.3). Free variation can be defined as the semantically irrelevant
occurrence of at least two different linguistic units (e.g. sounds, words) in
exactly the same environment. For example, as was pointed out in 1.2.3, the
word metal is pronounced with [t] in Received Pronunciation and with [d] in
General American. In a similar way, the word economic can be pronounced as
either /ˌɛkəˈnɒmɪk/ or /ˌiːkəˈnɒmɪk/.
Following Trubetzkoy (1969: 46), we will classify free variation into
general and pathological free variation. With regard to the latter, consider (9):
(9) My brother in his room (from Kaplan 1995: 13)
It is not difficult to see that this clause lacks the auxiliary be, connecting the
subject NP my brother with the predicative PP in his room: My brother in his
room = my brother is / was in his room. This is a well-known feature of African
American English, i.e. one of the non-standard varieties of American English.
As Kaplan (1995: 20) points out, the variant My brother is in his room, which
contains the auxiliary be, is not intrinsically superior to My brother in his room.
Basic concepts 13
Thus a number of languages (e.g. Standard Russian) do not use auxiliaries in

clauses like (9). However, given that in the English-speaking context people
who would say My brother is in his room are politically more powerful than
people who would say My brother in his room, the latter variant is regarded as a
deviation from the grammatical norm of the English language.
With regard to general free variation, consider (10).
(10) It was me who had changed (COCA)
In this sentence the complement position of the clause It was me is filled by the
accusative pronoun me. In contrast to the subject, which is usually not expressed
by the accusative forms me, him, her, us, them, the complement is often
expressed by both the nominative I, he, she, we, they and the accusative me, him,
her, us, them. Consider, for example, (11).
(11) It was I who requested the treatment (COCA)
In contrast to the non-standard My brother in his room, which can be regarded

as a 'pathological' deviation from the standard My brother is in his room, the use
of the accusative me in (10) does not represent a 'pathological' deviation from
the grammatical norm of the English language: It was me of (10) is not less
standard than It was I of (11). Indeed, as has been established by Biber et al.,
despite a traditional prescription based on the rules of Latin grammar

[i.e. that only a nominative pronoun must be used in a clause like It
was me who… of (10)], accusative forms are predominant in all
registers where the relevant forms are found. (Biber et al. 1999: 335)
Accordingly, the variation between It was me who… of (10) and It was I who…
of (11) is clearly an instance of general free variation.
In addition to the general–pathological distinction discussed above, all
instances of free variation can be classified into stylistically relevant and
stylistically irrelevant variation (Trubetzkoy 1969: 47-48). As an illustration
of the former, let us again compare the use of the nominative and the accusative
forms in sentences like (10) and (11). As just said, Sentence (10), containing the
accusative me, is not less standard than Sentence (11), containing the nominative
I. The only difference is that (11) is, from a stylistic point of view, more formal
than (10) (Payne and Huddleston 2002: 459). The characteristic 'more formal'
always applies to pronouns in the nominative case when they fill the
complement position in clauses like It is I who… of (11). That is, It is he who…
is more formal than It is him who…; It is she who… is more formal than It is her
who…; It is we who… is more formal than It is us who…; etc.
14 Chapter 1
By contrast, there does not seem to be a noticeable stylistic difference

between the two possible pronunciations of economic: neither /ˌɛkəˈnɑmɪk/ nor
/ˌiːkəˈnɒmɪk/ is more / less formal than the other. Similarly, there is no stylistic
difference between /ˈmɛtlˌ/ and /ˈmɛdl/. From the point of view of Received
Pronunciation, the American variant /ˈmɛdl/ can perhaps be regarded as a
'deviation' from the norm. However, it is neither more nor less formal than the
standard British pronunciation /ˈmɛtlˌ/. The variation between /ˌɛkəˈnɑmɪk/ and
/ˌiːkəˈnɒmɪk/ as well as between /ˈmɛtlˌ/ and /ˈmɛdl/ can thus be regarded as
instances of stylistically irrelevant free variation.
The categories of stylistically relevant and irrelevant free variation will be of
particular importance in Chapters 2 and 3.
1.3 What is a word?
Since morphology defines itself as the study of the internal structure of words,
any introduction to morphology must begin with the specification of the
wordhood criteria, i.e. formal characteristics of words distinguishing them
from other linguistic units. To begin with, let us consider the definition of a
word provided by Merriam-Webster Online. A word is
a speech sound or series of speech sounds that symbolizes and

communicates a meaning usually without being divisible into smaller
units capable of independent use. (MWO)
As we can see, this definition names two wordhood criteria: the meaning
criterion and the isolatability criterion. The former is fairly straightforward:
any combination of sounds that is associated with some meaning qualifies as a
word. For example, the combination of the sounds /kæt/ is a word because it is
associated with the meaning 'cat'; the combination of the sounds /dɪˈmɒkrəsɪ/ is
a word because it is associated with the meaning 'democracy'; the combination
of the sounds /ˈweɪthʊd/ is a word because it is associated with the meaning
'waithood'; etc.
Unfortunately, not all words (or to be more precise, combinations of sounds
which we usually regard as words) have meanings. Consider, for example, the
underlined forms in (12), (13), and (14).
(12) Do you think he lived a happy life? (COCA)

(13) He took a decision without asking anybody (COCA)
(14) We waited for him to unfold poems like pieces of origami (COCA)
Basic concepts 15
Objects like the NP a happy life of (12) are traditionally called cognate objects:
they are headed by nouns which are lexically cognate with the predicator verbs
with which they combine in the same predicate VP (e.g. to live and a life, to die
and a death, to sleep and a sleep, etc.). From a semantic point of view, cognate
objects are different from other object NPs in that their head nouns do not make
much contribution to the meanings of the VPs in which they occur. Indeed, He
lived a happy life means 'he lived happily'; He died a painful death means 'he
died painfully'; He slept a restless sleep means 'he slept restlessly'; etc.
Similarly, the predicator verb took of (13) does not make much contribution
to the meaning of (13): He took a decision without asking anybody means 'he
decided without asking anybody'. Verbs like the took of (13) are known as light
verbs. Apart from take, light verbs include make of make a decision 'to decide',
give of give a kiss 'to kiss', have of have a drink 'to drink', put of put the blame
on somebody 'to blame somebody', etc. All these verbs are 'light' in that they
carry only grammatical meanings such as e.g. 'the past tense' of took in (13), but
do not denote any actions or states, which is characteristic of other verbs as well
as of non-light uses of the same verbs. That is, for example, while gave of She
gave him a book is a 'normal' verb carrying the action meaning 'to give, to
transfer', gave of She gave him a kiss is light verb carrying only the grammatical
meaning 'the past tense'.
Finally, in contrast to the preposition for of I did it for him, which carries the
meaning 'in defense or support of; in favor of, on the side of' (OED), for of (14)
does not carry any meaning of its own but only serves to connect the predicator
waited with the object pronoun him. Verb + preposition combinations like wait
for somebody, account for something, depend on somebody / something, testify
to something, etc. are traditionally called prepositional verbs. These are
different from other verb + preposition combinations with regard to a number of
properties. One is the semantic emptiness of an accompanying preposition. Thus
there does not seem to be a particular reason why speakers of English wait for
somebody. In German, we wait on somebody (warten auf jemanden Æ wait on
somebody). In Russian, there is no preposition at all (ждать кого-то / zhdat
kogo-to Æ wait somebody). That for does not make any contribution to the
meaning of wait for somebody can also be corroborated by the existence of the
verb await, which expresses a similar meaning without any preposition; e.g.
(15).
(15) I awaited their critique (COCA)
To conclude: if there are words like life of (12), took of (13), and for of (14) that
express no (lexical) meanings, then the meaning criterion cannot be a reliable
wordhood criterion. (An alternative to this conclusion would be to regard the
forms under analysis as non-words. But this analysis seems to go against our
16 Chapter 1
feelings: in this case, we have to distinguish between e.g. the word take of He
took a book out of his pocket and the non-word take of He took a decision.) In
addition to this, if we want to regard the word as a formal category, we must try
to identify at least one formal property of words making them distinguishable
from (instances of) other formal categories.
Now, let us proceed to the isolatability criterion. This criterion defines a
word as a minimum free form, i.e. a form capable of independent use.
According to Bloomfield (1973[1934]: 178), 'independent use' means that the
form in question can form a one-word elliptical sentence, i.e. a sentence which
does not have the usual subject + predicate structure (e.g. He saw her) but
consists of only one word (e.g. He.). Consider, for example, the underlined
forms black of (16) and cat of (17).
(16) Pamela remains buried in her magazine, and Susan writes, Lauren.
Five. Karate-trained daughter. Miniature of her diminutive mother.
Fine, blond hair, pasted to nose and mouth. Scowls at turkey sandwich.
Could this small girl be a black belt...? Hmm. Black? Susan crosses it
out (COCA)
(17) Man I'm trying to sleep. Cat? Cat? I'll get the cat (COCA)
Both black and cat in the examples above form one-word elliptical sentences
and therefore qualify as words in accordance with the isolatability criterion. In
(16), the elliptical interrogative Black? can be expanded into Does this small girl
really possess a black belt?, whereas Cat? of (17) is expandable into Where is
the cat that is disturbing me at the moment?
In addition to adjectives and nouns, the isolatability requirement is also
fulfilled by pronouns, verbs (including auxiliaries), adverbs, conjunctions, and
interjections. Consider, for example, the underlined forms in (18), (19), (20),
(21), (22), and (23).
(18) And you know who that makes the ultimate winner? Us (COCA)
(19) Here are things she said he did: Kept secrets carefully. Knew who he
was. Believed in God. Gave. And gave. Studied hard. Thought. Took life
seriously (COCA)
(20) 'She was in Mrs. Potter's class,' said Peg Leg. Was? I thought, feeling a
stab of panic. I'd had Mrs. Linda Potter last year […] (COCA)
(21) He said, 'Would you like to know something? I went to Woodstock.' She
looked as though he'd started speaking Vietnamese. 'No kidding.
Recently? The movie The town? What? (COCA)
(22) VALENTINE: As you know, her brother died.
ORSINO: Yes. And?
VALENTINE: She plans to cry for the next seven years (COCA)
Basic concepts 17
(23) They want no oil. Wow! That leaves us with wind (COCA)
The elliptical accusative pronoun Us of (18) can be easily expanded into That
makes us the ultimate winner.
The elliptical verbs Gave and Thought of (19) can be expanded into He gave
and He thought.
The elliptical auxiliary Was? of (20) can be expanded into Is it really the
case that she was in Linda Potter's class? I think she is still in this class.
The elliptical adverb Recently? of (21) can be expanded into Is it really true
that you have recently gone to Woodstock?
The elliptical conjunction And? of (22) can be expanded into And how is her
brother's death related to what we are now talking about?
Finally, the elliptical interjection Wow! of (23) can be expanded into Wow,
that's great that they want no oil!
All these forms fulfill the isolatability criterion and, accordingly, qualify as
words.
Now, let us also discuss the status of the articles a and the. In contrast to the
word classes named above, the forms in question do not seem to be capable of
independent use in English. As Bloomfield (1973[1934]: 179) points out, "we
can imagine a hesitant speaker who says The… and is understood by his
hearers". However, such sentences rarely, if ever, occur in real life: the largest
balanced corpus of contemporary American English – the already mentioned
COCA, from which most of the examples used in this book are drawn – does not
contain a single elliptical sentence made up of a and the only. This fact may lead
us to the conclusion that the articles in English are non-free forms and hence
cannot be regarded as words. In this connection, Bloomfield observes that
the form the, though rarely spoken alone, plays much the same part in
our language as the forms this and that, which freely occur as
sentences; this parallelism leads us to class the as a word. (Bloomfield
1973[1934]: 179)
In other words, both the and this of e.g. the NPs the cat and this cat function as
determiners, marking the head noun cat as definite. Given this functional
similarity between this and the and given that this, in contrast to the, is capable
of independent use (e.g. (24)) and thus qualifies as a word, we are justified in
concluding that the is a word as well.
(24) Stevie was pulling on his shirt. His pale skin had turned bright pink, but
before he tugged down the shirt I saw a dark stripe on his chest. What
the hell is that? I said. This? He lifted his shirt (COCA)
18 Chapter 1
The same line of reasoning enables us to analogize the indefinite article a to the
word some: like a of the NP a cat, some of the NP some cat functions as
determiner, marking the head noun cat as indefinite. Given this functional
similarity between a and some and given that some, in contrast to a, is capable
of independent use and thus qualifies as a word, we are justified in concluding
that a is a word as well.
In addition to the meaning and isolatability criteria, many studies also
mention:
x the movement criterion

x the uninterruptability criterion
x the orthographic criterion
The movement criterion defines a word as a form that can move within a clause.
Unfortunately, in a language like English, which has a fixed word-order (recall
that the default position of the subject in English is before the main verb), this
criterion is of very little help. Consider, for example, (25)-(29).
(25) She gave him a kiss

(26) *Gave she him a kiss
(27) *Gave him she a kiss
(28) *Gave him a she kiss
(29) *Gave him a kiss she
The ungrammaticality of (26)-(29) illustrates that the subject she of (25)

cannot move within the sentence and thus does not qualify as a word. Likewise,
as (30)-(32) illustrate, the predicator verb give also does not fulfill the
movement requirement and thus cannot be regarded as a word.
(30) *She him gave a kiss

(31) *She him a gave kiss
(32) *She him a kiss gave
The only possible transformations of (25) are (33) and (34).
(33) A kiss, she gave him

(34) Him, she gave a kiss
Accordingly, we must conclude that (25) consists of the two non-words she and
gave and the two words him and a kiss. Needless to say, this is an extremely
counter-intuitive conclusion.
Basic concepts 19
The uninterruptability criterion defines a word as a string of sounds that does

not allow the insertion of a modifying element within its boundaries. This
criterion was proposed by Bloomfield (1973[1934]: 180) for distinguishing
between non-idiomatic adjective + noun NPs like a black board and idiomatic
adjective + noun complex words like blackboard. Of the two, only the former
allow the insertion of a modifier. That is, for example, we can say a black – that
is, bluish-black – board but not *a blackthatisbluishblackboard.
Like the movement criterion, the uninterruptability criterion has a serious
drawback: it provides no justification for distinguishing between words and
some fully-idiomatic VPs. Consider, for example, the VP shoot the breeze 'to
chat, talk idly' (OED). Just like blackboard, shoot the breeze is uninterruptable
in that the component breeze cannot be modified by adjectives like cool, warm,
gentle, soft, fresh, sudden, etc. which modify breeze in non-idiomatic contexts
(e.g. I felt the warm breeze in my face). We cannot say *shoot the cool breeze,
*shoot the warm breeze, *shoot the sudden breeze, etc. Accordingly, the VP
shoot the breeze fulfills the uninterruptability criterion and thus must be
regarded as a word.
Finally, the orthographic criterion defines a word as a string of letters that is
separated from other words by means of a blank space. For example, the non-
idiomatic NP a black cat can be segmented into three words – a, black, and cat –
simply because in this NP there are blank spaces separating a from black and
black from cat. (It does not really matter that of these components, only black
and cat fulfill the isolatability criterion discussed above.) The same applies to
the idiomatic VP shoot the breeze. The VP under consideration can be said to
consist of three words – shoot, the, and breeze – simply because we do not write
*shootthebreeze. Finally, recall the semantically empty preposition for of wait
for somebody. Again, we can argue that for qualifies as a word simply because
we write wait for somebody rather than *waitforsomebody.
The presence of a blank space seems to be the easiest way of determining
whether a particular string of letters represents a word or not. But it cannot be
regarded as a sufficient wordhood criterion. The major problem with the
orthographic criterion is not the frequently mentioned instability of the spelling
of some English words – an often cited example is flowerpot, which can be
spelled flowerpot, flower-pot, and flower pot – but the secondary and the
artificial character of the writing system in relation to spoken speech (see e.g.
Lieber and Štekauer 2009a: 7). The former appeared much later than the latter;
e.g. the orthographic forms flowerpot / flower-pot / flower pot appeared in the
English language later than the sound form /ˈflaʊəpɒt/. Hence it is extremely
doubtful that word boundaries can be determined only on the basis of
orthographic conventions. This position has been recently advocated by
Haspelmath (2011), who argues that
20 Chapter 1
[m]any orthographies, especially (but not only) those based on the

Greek, Latin and Cyrillic alphabets, use spaces between words.
However, there are also many orthographies that do not use spaces,
e.g. Chinese, Japanese, and Sanskrit. […] In the European languages,
too, word spacing is an innovation; until about a thousand years ago,
scriptio continua (continuous writing) was the norm in Western
writing […]. There is no doubt that the modern orthographic use of
spaces is to some extent guided by language structure, but not in such
a way that conventional spelling could be used to decide contentious
issues. (Haspelmath 2011: 36)
To conclude: of the five criteria discussed above, only the isolatability criterion
provides a relatively unproblematic formal basis for distinguishing between
words and instances of other formal categories (i.e. forms that are not capable of
independent use). In accordance with this criterion, a word is any combination
of sounds which can form a one-word elliptical sentence such as Black? of (16),
Cat? of (17), Us of (18), Gave and Thought of (19), Was? of (20), Recently? of
(21), And? of (22), and Wow! of (23). (The isolatability criterion does not
provide a basis for distinguishing between NPs like a black board and complex
words like blackboard: both the former and the latter can be analyzed as
combinations of two words, which are capable of independent use. We will
return to this issue in Section 5.6.)
The conclusion drawn above leads us to a very important practical question:
how can we actually establish whether some particular combination of sounds is
capable of occurring as a one-word sentence? There are two answers to this
question. If you are a native speaker of English, you can try to invent a context
(similar to those of the examples above) in which the form in question occurs as
an elliptical sentence. If you succeed in that, you will be justified in claiming
that the form under analysis does indeed qualify as a word. For example, if you,
like Bloomfield, can imagine a hesitant speaker who says The… or A… and is
understood by the hearers, you can claim that the articles in English are words.
Alternatively, this question can be answered by looking at actual language
use, i.e. consulting a balanced linguistic corpus like Corpus of Contemporary
American English / COCA (http://www.americancorpus.org/) or British
National Corpus / BYU-BNC (http://corpus.byu.edu/bnc/). Just enter e.g. . take .
(with spaces separating take from the full stops) to the search mask of the corpus
of your choice and click at 'Search'. Both COCA and BYU-BNC will then
search for the occurrences of take in which it is preceded and followed by a full
stop. One of such occurrences could be an elliptical sentence like .Gave. of (19).
If you enter . take ? or . take ! (also with spaces separating take from the
punctuation marks), the corpora will search for the occurrences of take in which
it is preceded by a full stop and followed by either a question mark or an
Basic concepts 21
exclamation mark. One of such occurrences could be a one-word interrogative

sentence like .Black? of (16) or a one-word exclamatory sentence like .Wow! of
(23). If you succeed in finding at least one elliptical sentence containing take,
you will be justified in regarding take as a word.
If neither COCA nor BYU-BNC yield elliptical sentences consisting of the
form in question, this does not necessarily mean that it cannot be regarded as a
word. It may be the case that the databases of both corpora simply do not
contain elliptical sentences consisting of the form in question. For instance,
neither COCA nor BYU-BNC contain elliptical sentences consisting of the form
baboon. This fact may lead us to the conclusion that baboon does not fulfill the
isolatability criterion and, accordingly, does not qualify as a word. However, at
an earlier point, we established that the isolatability criterion is fulfilled by a
semantically related word cat: Man I'm trying to sleep. Cat? Cat? I'll get the cat
of (17). We can imagine a similar context involving a person who is trying to
fall asleep and a baboon which is somehow disturbing that person. The former
gets angry and says Man I'm trying to sleep. Baboon? Baboon? I'll get the
baboon. This is sufficient to regard baboon as a word.
1.4 Exercises
1. Make sure you can explain each of the key terms printed in boldface (ideally,
using your own examples).
2. Which of the following statements are true?
a) A complex word consists of only one formally indivisible component.

b) A concept is a mental description that is activated by a particular sound form.
c) Opacity is one of the characteristics of literal meaning.
d) Acoustic phonetics is concerned with the production of sounds.
e) Sounds that realize two different phonemes are in contrastive distribution.
f) Phrases never perform syntactic functions.
g) Complement phrases can function as subjects of associated passive clauses.
h) Free variation can either be stylistically relevant or stylistically irrelevant.
i) Light verbs express no lexical meanings.
j) A word is a linguistic unit which is not capable of independent use.
3. State which of the following words are complex words. Explain your
analysis.
a) dog
b) untrue
22 Chapter 1
c) freedom
d) stage
e) democracy
f) to disambiguate
g) to satisfy
h) mortgage
i) defendant
j) chair
4. Name the syntactic function which the underlined phrases perform in the
sentences below. Explain your analysis.
a) He broke the record (COCA)

b) It is hot for September (COCA)
c) I kept quiet (COCA)
d) Yesterday I turned twenty-nine (COCA)
e) He stood in the doorway (COCA)
f) So I wrote a letter (COCA)
g) That data became even more important (COCA)
h) To err is human (COCA)
i) I helped him (COCA)
j) In case of emergency, you must do the following (COCA)
5. Using COCA, find out which of the following forms qualify as words in
accordance with the isolatability criterion.
a) she
b) truly
c) girls
d) has
e) to
f) stand
g) under
h) three
i) where
j) damn
1.5 Further reading
For an overview of how different linguists define the term 'morphology', see
Heringer (2009: 9-18).
Basic concepts 23
Löbner (2002) and Cruse (2004) are good introductory textbooks to

semantics: both contain very detailed discussions of such important concepts as
'meaning', 'conceptual category', 'semantic decomposition', 'necessary and
sufficient conditions', 'prototype', etc.
A recent introduction to English phraseology is Fiedler (2007). For an
overview of the contemporary research in phraseology, see Burger et al.'s (2007)
Handbook of Phraseology as well as the recent monographs by Dobrovol'skij
and Piirainen (2009; 2005).
There are many good introductions to English phonetics and phonology.
See, for example, Gut (2009), Cruttenden (2008), Yavaş (2006), McMahon
(2004). Trubetzkoy's (1939) Grundzüge der Phonologie is a classic monograph
dealing with such key issues as differences between phonetics and phonology,
identification of phonemes, distribution of sounds, etc. The English translation is
Trubetzkoy (1969).
Huddleston and Pullum (2005) is an excellent introduction to English syntax
discussing such issues as syntactic functions and their formal properties (e.g.
subject-hood criteria, differences between complements, objects, and adjuncts),
the internal structure of phrases in English, types of sentences in English, etc. A
more detailed discussion of the same issues can be found in their earlier work
The Cambridge Grammar of the English Language (Huddleston and Pullum
2002). A more recent introduction to English syntax is Kreyer (2010).
Many authors (e.g. Aronoff and Fudeman 2005: 33-44) introduce a number
of subcategories of words: for example, orthographic words, phonological
words, morphosyntactic words, etc. This approach is typical of authors who
conclude that words cannot be distinguished from non-words on the basis of one
criterion such as, for example, the isolatability criterion. Since this book takes a
different approach (namely, that most forms that language users regard as words
can be identified as words on the basis of the isolatability criterion), it will not
use any of these terms.
2 The internal structure of English words
This chapter focuses on the morphemic segmentation of English words as well

as on the typology and distribution of morphs in English. Section 2.1 provides a
very basic definition of a morpheme as the smallest meaning-carrying unit.
Section 2.2 analogizes the concept of a morpheme to that of a linguistic sign
and, on the basis of this, formulates a more precise definition of a morpheme as
a conventionalized association between a particular form and a particular
meaning. Section 2.3 is concerned with the distribution of morphs in English:
the main question here is when two different forms can be regarded as
allomorphic realizations of the same morpheme rather than as forms realizing
two different morphemes. Section 2.4 deals with the principles of morphemic
segmentation: how can we decide whether the word under analysis can be
segmented into (at least) two morphemes or is a monomorphemic word non-
segmentable into smaller meaning-carrying units? Finally, Section 2.5 discusses
the hierarchy of different morpheme types. In this section, we will learn, among
other things, in which respects roots are different from affixes and whether so-
called combining forms constitute a distinct morpheme type in English.
2.1 What is a morpheme?
A morpheme is traditionally defined as the smallest meaningful component of a

word. As an illustration of this definition, let us consider the meaning of untrue
in (35).
(35) Her lover had been untrue (MWO)
According to MWO, the untrue of (35) means 'not faithful'. It is evident that the
meaning of untrue represents the reverse of the meaning of true. Thus the
sentence Her lover had been true means the opposite of (35), namely, that the
subject her lover was faithful to his lover in that he did not have sexual relations
with a third person. Hence the word untrue can be segmented into two meaning-
carrying units: {un} and {true}3. The unit {true} carries the meaning 'faithful';
the unit {un} carries the meaning 'negation of the following adjective true'. In a
similar way, the word unable can be segmented into 1) the unit {able} carrying
the meaning 'able', i.e. 'having sufficient power, skill, or resources to accomplish
3
Linguists usually put morphemes in curly braces {}.
26 Chapter 2
an object' (MWO) and 2) the unit {un} carrying the meaning 'negation of the
following adjective able'.
Units like {un} and {true} of untrue and {un} and {able} of unable are
morphemes. They are the smallest meaningful components of the words under
analysis, i.e. components which cannot be further segmented into smaller
meaning-carrying units. That is, for example, the unit un- cannot be further
segmented into the morphemes {u} and {n} because the sounds [ʌ] and [n] do
not carry any discernible meanings of their own. Similarly, true cannot be
further segmented into the morphemes {tr} and {u} because the sounds [tr] und
[uː] do not mean anything.
In Section 1.1 we became acquainted with the distinction between simple
and complex words. Given what we have just learned about morphemes, we are
now in a position to give a more precise definition of the terms 'simple word'
and 'complex word'. Simple words are monomorphemic words, i.e. words like
cat that can be segmented only into one morpheme. Complex words, by
contrast, are polymorphemic words, i.e. words like untrue, unable, and
waithood which can be segmented into at least two morphemes.
Morphemes are traditionally classified into a number of categories. With
regard to their autonomy, morphemes are usually classified into free and bound
morphemes. Free morphemes fulfill the isolatability requirement which we
discussed in 1.3. For example, the simple word cat can be said to consist of the
free morpheme {cat} because, as we established in 1.3, the sound form /kæt/ can
form a one-word elliptical sentence. The same applies to the morpheme {true}
of untrue. By contrast, the morpheme {un} of untrue is a bound morpheme, i.e.
a morpheme which does not fulfill the isolatability requirement: we cannot form
an elliptical sentence consisting of un- only.
With regard to their function, morphemes are classified into lexical and
grammatical morphemes. Lexical morphemes are morphemes like {cat} of cat
and {un} of untrue which express optional lexical meanings, i.e. meanings
which are expressed only when language users specifically want them to be
expressed. Grammatical morphemes are morphemes like {s} of books and {ed}
of I worked which express obligatory grammatical meanings, i.e. meanings
which cannot be unexpressed. For example, in English nouns are always marked
with regard to the grammatical category NUMBER (i.e. nouns are either in the
singular or in the plural number) and verbs are always marked with regard to the
grammatical category TENSE (i.e. they are either in the present or in the past
tense).
With regard to their form, morphemes can be classified into continuous and
discontinuous morphemes. The latter are morphemes which are interrupted by
other morphemes. Consider, for example, the grammatical meaning 'the
progressive aspect' of I am working. It is evident that this meaning is inherent in
both the free auxiliary be and the bound form -ing. If we remove these forms,
The internal structure of English words 27
the clause under analysis will acquire a different grammatical meaning: I work is
in the non-progressive aspect. Accordingly, we can conclude that the
progressive meaning of I am working is expressed by the discontinuous
morpheme {am…ing}, which is interrupted by the lexical morpheme {work}.
The majority of morphemes in English and other languages are, however,
continuous morphemes, i.e. morphemes like {cat} of cat, {un} and {true} of
untrue, {wait} and {hood} of waithood, etc. which are made up of "sequences of
consecutive phonemes" (Harris 1945: 121).
Usually, morphemes carry only one meaning. That is, for instance, the
morpheme {cat} of cat carries the meaning 'cat'; the morphemes {un} and
{true} of untrue carry the meanings 'not' and 'faithful'; the morphemes {wait}
and {hood} of waithood carry the meanings 'to wait' and 'stage'; etc. Some
morphemes, however, simultaneously express more than one meaning.
Consider, for instance, the meanings inherent in the bound morpheme -s of runs
in e.g. He runs. It is evident that this morpheme carries several grammatical
meanings. These include:
x 'the present tense' (cf. He runs and He ran)

x 'the non-progressive aspect' (cf. He runs and He is running)
x 'the active voice' (cf. He runs and This business was run by him)
x 'the indicative mood' (cf. He runs and Run!)
x 'the third person' (cf. He runs and You run)
x 'the singular number' (cf. He runs and They run)
Morphemes like -s of runs which simultaneously express more than one

meaning are often called portmanteau morphemes (or mega-morphemes or
cumulative morphemes).
We will enlarge on the typology of morphemes in Sections 2.4 and 2.5.
2.2 Morphemes as signs
A morpheme is also very often defined as the smallest linguistic sign, i.e. a
conventionalized association between a particular form and a particular meaning
(also commonly referred to as the signifier and the signified). Analyzing the
previous example {true}, we can say that in the English language there exists a
conventionalized association between the sound form /truː/ and the concept of
being true. That is, when speakers of English hear the word true in sentences
like (35), they think of a person who is faithful to another person (in a sexual
sense). The morpheme {true} thus represents a linguistic sign made up of the
signifier /truː/ and the signified 'faithful', there being a conventionalized
association between the former and the latter.
28 Chapter 2
2.2.1 One signifier Æ more than one signified
Very often, however, one and the same signifier is associated with more than
one signified. In this case, we are dealing with either polysemy or homonymy.
The hallmark of polysemy is that two or more signifieds inherent in the same
signifier have something in common. A good example is the true of (36).
(36) Indicate whether each of the following statements is true or false.

(MWO)
In contrast to the true of (35), the true of (36) does not mean 'faithful' but 'not
false' (MWO). The fact that one and the same signifier true can express both the
meanings 'faithful' and 'not false' is clearly an instance of polysemy because the
meaning 'not false' can be easily derived from the meaning 'faithful': if a
statement is not false, its speaker can be said to be faithful to the truth. The sense
'not false' can thus be seen as a product of semantic narrowing of the sense
'faithful': 'faithful' > 'faithful to the truth, not false'.
Homonymy is different from polysemy in that two or more signifieds which
can be expressed by the same signifier do not have much in common. Compare,
for example, the meanings of case in (37) and (38).
(37) In this case, there's no question (COCA)

(38) Lifting a heavy case, Nick places it into the cab's trunk (COCA)
In contrast to the meanings 'faithful' and 'not false' of true, the two meanings of
case exemplified by (37) and (38) – 'a set of circumstances or conditions'
(MWO) and 'a box or receptacle for holding something' (MWO) – do not seem
to have anything in common. Accordingly, their co-existence in the signifier
case must be regarded as an instance of homonymy, not polysemy.
2.2.2 One signified Å more than one signifier
The reverse situation is possible as well: one and the same signified can be
expressed by more than one signifier. For example, the concept of a person who
has sex only with his or her regular sexual partner (i.e. spouse, girlfriend) can be
expressed by both true and faithful. Cf. (39) and (40).
(39) Her boyfriend had been true

(40) Her boyfriend had been faithful
Accordingly, the signifier faithful can be considered a synonym of the signifier

true.
A somewhat similar example is the word detail, which can be pronounced
(by one and the same speaker of General American) either /ˈdiːteɪl/ or /dɪˈteɪl/.
That is, the first syllable can either be stressed and end in the long vowel [i:] or
be unstressed and end in the short vowel [ɪ]. Similarly, doctrinal can be
pronounced both /ˈdɒktrɪnəl/ and /dɒkˈtraɪnəl/. That is, the second syllable can
either be unstressed and end in the short vowel [ɪ] or be stressed and end in the
diphthong [aɪ]. (According to MWO, the pronunciation /dɒkˈtraɪnəl/ is,
however, more typical of Received Pronunciation than of General American.)
As we have learned in the previous chapter, in phonetics and phonology
cases like detail and doctrinal are regarded as instances of free variation (or free
distribution) of sounds. Free variation is similar to synonymy in that in both
cases the same meaning is expressed by more than one sound form. However, in
the case of free variation, there remains a considerable phonetic similarity
between the forms in question: as pointed out above, /ˈdiːteɪl/ and /dɪˈteɪl/ differ
only with regard to the placement of stress and the length of the first vowel.
Similarly, /ˈdɒktrɪnəl/ and /dɒkˈtraɪnəl/ differ with regard to the placement of
stress and the quality of the second vowel. Conversely, in the case of
synonymous forms like true and faithful, there are no or very few phonetic
similarities. Compare, for example, /truː/ and /ˈfeɪθfʊl/. These words do not
share a single sound.
2.2.3 The syntactics of a sign
The conception of a linguistic sign as a conventionalized association between a

particular signifier and a particular signified was formulated by the Swiss
linguist Ferdinand de Saussure (1857-1913) in his posthumously published
Cours de Linguistique Générale / Course in General Linguistics (1973: 97-103).
An important revision of this conception was made by the Russian-Canadian
linguist Igor Mel'čuk (1979: 171; cf. 1968: 425), who, inspired by the works of
the American semiotician Charles Morris, defined the linguistic sign as a triplet
consisting of the following three elements:
1. the signifier
2. the signified
3. the syntactics
The latter component – the syntactics of the sign – is described by Mel'čuk

(1979: 171) as "the set of all data about the combinatorial properties" including
"part of speech, […] declension or conjugation type, phonological and / or
30 Chapter 2
morphological environments, selectional restrictions of all kinds, etc.".

Compare, for example, the adjective happy and the adverb happily. As regards
their signifieds, it may seem that the forms under analysis express different
meanings: while happy means 'happy', happily means 'in a happy manner'. But
compare the meanings of the clauses (41) and (42), containing the adjective
happy and the adverb happily.
(41) He lived a happy life

(42) He lived happily
As has been recognized by many authors (e.g. Pullum and Huddleston 2002:
529), adjectives like happy and morphologically related adverbs like happily
express essentially the same meaning. Indeed, both He lived a happy life and He
lived happily can be paraphrased by His life was happy. Accordingly, the
difference between happy and happily cannot be described in terms of their
signifieds. What distinguishes the former from the latter is that the adverb
happily cannot fill two syntactic positions that are typically filled by the
corresponding adjective happy. First of all, happily cannot function as modifier
of a head noun in an NP like a happy life. That is, we cannot say *a happily life.
Second, happily cannot function as complement in a sentence like His life was
happy: there can only be His life was happy but not *His life was happily.
Obviously, a linguistic theory that defines a sign as a doublet consisting of
the signifier and the signified cannot explain the difference between an adjective
like happy and an adverb like happily. As, for example, Pullum and Huddleston
(2002: 529) argue, "it is function that provides the primary basis for the
distinction between adjectives and adverbs". But what precisely is function? Is
function also a component of a linguistic sign? By contrast, a theory that defines
the sign as a triplet consisting not only of the signifier and the signified but also
of the syntactics can easily explain the difference: the adjective happy and the
adverb happily differ from each other with regard to their syntactics.
2.2.4 The sociolinguistics of a sign
There is no doubt that the inclusion of the syntactics represents a significant

improvement of Saussurean conception of a linguistic sign. However, as we will
see below, signs also have properties that cannot be attributed to either their
syntactics or their signifieds.
Compare, for example, the forms fever and pyrexia. Both of them express
the same meaning: 'abnormal elevation of body temperature' (MWO). Both of
them can head NPs that function as subjects or predicatives in predicate VPs.
E.g. (43) and (44).
(43) Jaden had a fever (COCA)

(44) A wound swab was taken as Julie still had a pyrexia (BYU-BNC)
Accordingly, these forms do not differ with regard to either their signifieds or
their syntactics. The difference between fever and pyrexia is that of register:
while the former is a stylistically neutral term that can be used in all kinds of
contexts, the latter is a highly professional term that occurs in highly specialized
contexts: COCA has only 6 occurrences of pyrexia in scientific articles
published in medical journals and 6276 occurrences of fever in all possible
contexts. The free variation between fever and pyrexia is thus an instance of
stylistically relevant free variation.
Recall also that the word doctrinal can be pronounced both /ˈdɒktrɪnəl/ and
/dɒkˈtraɪnəl/, the latter variant being more typical of Received Pronunciation
than of General American. Again, as in the case of fever and pyrexia, we
observe a sociolinguistic (not a syntactic or a semantic) variation between the
forms in question. That is, /ˈdɒktrɪnəl/ and /dɒkˈtraɪnəl/ do not differ with regard
to either their signifieds or their syntactics: both signifiers express the same
meaning 'doctrinal' and can fill the syntactic positions that are typically filled by
adjectives. The only difference is that the variant /dɒkˈtraɪnəl/ is more likely to
be used by a speaker of Received Pronunciation, whereas a speaker of General
American will most likely give preference to the variant /ˈdɒktrɪnəl/.
Taking all this into account, we can conclude that a linguistic sign represents
a conventionalized association between the following four elements:
1. the signifier
2. the signified
3. the syntactics
4. the sociolinguistics
The sociolinguistics of a sign is the component that contains the information as

to whether the signifier of the sign
is stylistically neutral, formal, colloquial, slang, etc.

is mainly used by people of a particular profession, social status, gender, etc.
is characteristic of a particular variety of English such as American English,
British English, Australian English, etc.
2.2.5 The signified as the most important sign component
The definition of a linguistic sign given above raises the question of which of
the four components of the sign – the signifier, the signified, the syntactics, the
32 Chapter 2
sociolinguistics – must be seen as its most important component. The answer to

this question has important implications for our analysis of the distribution of
signifiers like happy and happily that have different syntactics but identical
signifieds: if we decide that the syntactics of a sign is more important than its
signified, we will be justified in claiming that the signifiers happy and happily
are in contrastive distribution and, accordingly, can be regarded as instances of
two different signs. Similarly, if we decide that the sociolinguistics of a sign is
more important than its signified, we will be justified in claiming that fever and
pyrexia occur in contrastive distribution and thus realize two different signs. By
contrast, if we decide that the signified is more important than both the
syntactics and the sociolinguistics, we will have to analyze happy and happily as
instances of the same sign that occur in complementary distribution 4 and fever
and pyrexia as instances of the same sign that occur in stylistically relevant free
variation.
This book argues for the latter solution. It is the signified (not the syntactics
or the sociolinguistics) of a sign that must be seen as its most important
component. This is so because the semantic function is a much more important
motivation for the use of linguistic signs than both the syntactic and the
sociolinguistic function. That is, the primary reason why we use the signifiers
happily, pyrexia, /dɒkˈtraɪnəl/, etc. is not our wish to either fill particular
syntactic positions which the syntactics of these signifiers allow them to fill or
to mark these signifiers as stylistically neutral, formal, characteristic of a
particular variety of English, etc. We use these signifiers because, first and
foremost, we want to express the meanings that are conventionally associated
with these signifiers in English.
Accordingly, given the primacy of the signified over both the syntactics and
the sociolinguistics (and the signifier), we can conclude this section with the
following claims:
x Two different signifiers that are conventionally associated with two different
signifieds – e.g. the signifiers detail and doctrinal, which are associated with
the meanings 'detail' and 'doctrinal' – are instances of two different signs.
They are in contrastive distribution.
x Two identical signifiers that are conventionally associated with two different
signifieds – e.g. the signifier true, which is associated with the meanings
'faithful' and 'not false' – are likewise instances of two different signs. They
are also in contrastive distribution.
4
Recall that there can only be the NP a happy life and the clause His life was happy, but not
*a happily life and *His life was happily.
x Two different signifiers that are conventionally associated with one and the
same signified are instances of the same sign, regardless of the differences
between these signifiers concerning either their syntactics or their
sociolinguistics. Signifiers like happy and happily which differ from each
other with regard to their syntactics are in complementary distribution.
Signifiers like fever and pyrexia which differ from each other with regard to
their sociolinguistics are in free variation.
2.3 The distribution of morphs
In morphology the signifier of a morpheme is often referred to as its morph.

Similar to phones that realize different phonemes, morphs that realize different
morphemes (i.e. those that have different meanings) can be said to occur in
contrastive distribution. For example, the morphs /truː/ and /ˈeɪ.bl ̩/ of the
morphemes {true} and {able} are in contrastive distribution because they carry
two different meanings: 'true' and 'able'.
Like in phonetics and phonology, contrastive distribution of morphs that
realize two different morphemes must be distinguished from both free and
complementary distribution of morphs that realize the same morpheme. As
pointed out in the previous section, one and the same signified can sometimes be
expressed by more than one signifier. Consider again the two possible
pronunciations of detail: /ˈdiːteɪl/ and /dɪˈteɪl/. Given that one and the same
speaker of American English can pronounce detail on one occasion as /ˈdiːteɪl/
and on another occasion as /dɪˈteɪl/ and given that the word detail cannot be
segmented into smaller meaningful units5, we are justified in concluding that the
morpheme {detail} can be realized by the morphs /ˈdiːteɪl/ and /dɪˈteɪl/ occurring
in free distribution. Morphs that realize one and the same morpheme are called
allomorphs. E.g. the morphs /ˈdiːteɪl/ and /dɪˈteɪl/ are the two allomorphs of the
morpheme {detail}.
As regards complementary distribution, consider the words inadequate and
impossible. Since the meanings of these words are the opposites of the meanings
of adequate and possible, we can conclude that the negative meaning 'not' is
inherent in the morphs /ɪn/ and /ɪm/. What is important here is that unlike
/ˈdiːteɪl/ and /dɪˈteɪl/ of detail, /ɪn/ and /ɪm/ are not in free variation: we can say
neither */ɪmˈædɪkwət/ instead of /ɪnˈædɪkwət/ nor */ɪnˈpɒsɪb(ə)l/ instead of
/ɪmˈpɒsɪb(ə)l/. Similar to complementary distribution of allophones of the same
phoneme, complementary distribution of allomorphs of the same morpheme
involves a morpheme that is realized by at least two allomorphs which occur in
different environments, thereby complementing each other. In the case of the
5
In contrast to the morph de- of e.g. destabilize, the unit de- of detail does not carry the
reversative meaning 'do the opposite of' (MWO) and therefore cannot be considered a morph.
34 Chapter 2
English negative morpheme realized by /ɪn/ in inadequate and /ɪm/ in

impossible, the following regularity can be postulated:
x When followed by the bilabial sounds [p] and [b] (i.e. sounds that are
articulated with both the lower and the upper lip), the negative morpheme is
realized by the morph /ɪm/. E.g. impossible Æ /ɪmˈpɒsɪb(ə)l/, imbalance Æ
/ɪmˈbæləns/.
x When followed by the sounds [m], [l], and [r], it is realized by /ɪ/. E.g.
immoral Æ /ɪˈmɒrəl/, illegal Æ /ɪˈliːgəl/, irretrievable Æ /ɪrɪˈtriːvəb(ə)l/.
x In all other cases, it is realized by /ɪn/. E.g. inadequate Æ /ɪnˈædɪkwət/,

incompetent Æ /ɪnˈkɒmpɪtənt/, intolerable Æ /ɪnˈtɒlərəb(ə)l/, etc.
To summarize: the morphs /ɪm/, /ɪ/, and /ɪn/ express the same negative meaning
'not' and never occur in the same environment: there can be no */ɪnˈpɒsɪb(ə)l/,
*/ɪmˈædɪkwət/, */ɪnˈmɒrəl/, etc. Accordingly, they are allomorphs of the same
morpheme which occur in complementary distribution.
Finally, let us also consider the distribution of synonymous morphs which
are capable of occurring in the same environment. Compare, for example, the
negative morphs /ʌn/ of untrue and /ɪn/ of inadequate. From a diachronic point
of view (i.e. if we consider the history of the English language), these morphs
could perhaps be regarded as realizations of two different morphemes: /ʌn/ is a
native Germanic morph which has existed since the Old English period (until
~1100), whereas /ɪn/ came into English only in the 14th century with loans from
French such as incombustible, incomprehensible, ineffectual, etc. (For details,
see Marchand 1969: 168-170 or the corresponding entries in the OED). By
contrast, from a synchronic point of view (i.e. if we consider Present-day
English only), the morphs /ʌn/ and /ɪn/ can be regarded as allomorphs of the
same negative morpheme. This conclusion is supported by the fact that both /ʌn/
and /ɪn/ express the same negative meaning – untrue means 'not true' and
inadequate means 'not adequate' – and occur in the same environment (i.e.
before adjectives like true and adequate). It is of course true that in neither
untrue nor inadequate is there a free variation of the morphs /ʌn/ and /ɪn/. For
neither *intrue nor *unadequate exist in Present-day English.6 However, we do
find such adjective pairs as uncommunicative and incommunicative both
meaning 'not communicative', undistinguishable and indistinguishable both
6
The reason for this is that the syntactics of the morph in- allows it to combine only with
words of Latin or Romanic origin, whereas the syntactics of un- allows it to be used only with
native or completely naturalized words (OED). E.g. true is a word of Germanic origin and
therefore combines with un-, whereas adequate is a word of Latin origin and therefore takes
in-. The morphs un- of untrue and in- of inadequate can thus be regarded as allomorphs of the
same morpheme which occur in complementary distribution.
meaning 'not distinguishable', unmovable and immovable both meaning 'not

movable', etc. Hence, at least as far as these cases are concerned, we can speak
of the free variation of the allomorphs /ʌn/ and /ɪn/.
As Marchand (1969: 170) points out, the morph in- is stylistically different
from the morph un- in that it "forms learned, chiefly scientific words", whereas
un- is a stylistically neutral regular negative morph. The difference between
negative adjectives beginning with un- and in- is thus that in-adjectives sound
more scientific and learned than their counterparts beginning with un-.
Accordingly, the free variation of the morphs /ʌn/ and /ɪn/ in pairs like
uncommunicative and incommunicative is an instance of stylistically relevant
free variation.
Another interesting case involving synonymous morphs is represented by the
distribution of -ed of talked, walked, worked, etc. and ex- of ex-ambassador, ex-
husband, ex-president, etc. (Nida 1948: 425-426). If we again disregard the
diachronic history of these signifiers, namely, that -ed is a native Germanic
morph, whereas ex- has Latin origin (OED), we can arrive at the conclusion that
-ed and ex- are realizations of the same past time morpheme. Indeed, a sentence
like (45) indicates that the event in question took place before the moment of
utterance (i.e. in the past).
(45) He worked with President Bush (COCA)
Similarly, ex- of ex-president indicates a past state, namely, that the person
referred to as ex-president was president before the moment of utterance.
Now, let us also consider the distribution of the putative allomorphs -ed and
ex-. The former occurs only after verbs (talked, walked, worked), but never
before nouns (*ed-ambassador, *ed-husband, *ed-president). Conversely, the
form ex- occurs only before nouns (ex-ambassador, ex-husband, ex-president),
but never after verbs (*talkex, *walkex, *workex). Consequently, -ed and ex- are
allomorphs of the same morpheme that occur in complementary distribution.
However, this conclusion seems to be counter-intuitive. Among other things,
this impression emerges from the fact that this putative instance of
complementary distribution is not phonologically-conditioned. That is, for
example, in the case of impossible, there is a phonological explanation
accounting for the fact that the negative morpheme is not realized by the morph
/ɪn/. From the articulatory point of view, it is easier to say /ɪmˈpɒsɪb(ə)l/ than
*/ɪnˈpɒsɪb(ə)l/ because both [m] and [p] are bilabial sounds, whereas [n] is
characterized by an alveolar articulation, i.e. its active articulator is the tip of the
tongue, which is raised towards the alveolar ridge. The morph /ɪm/ is thus a
product of the place-of-articulation assimilation of the alveolar [n] to the
bilabial [p]. In contrast, no similar explanation can account for the impossibility
of forms like *ed-ambassador and *talkex.
36 Chapter 2
Nevertheless, the mere absence of phonological conditioning does not justify

the treatment of -ed and ex- as realizations of two different morphemes.
Consider, for instance, the plural form of the noun ox. In contrast to e.g. box,
which takes one of the regular (phonologically conditioned) plural morphs /ɪz/
(box Æ boxes), ox combines with the morph /ən/: ox Æ oxen. It is obvious that
this fact cannot be regarded as an instance of phonological conditioning: if
speakers of English can pronounce /bɒksɪz/, they must also be able to pronounce
/ɒksɪz/. (This is an instance of lexical conditioning, i.e. it is a peculiarity of the
noun ox that it forms the plural form with the morph /ən/ rather than with /ɪz/.
Similarly, it is a peculiarity of the noun mouse that its plural form is mice, not
*mouses.) However, in spite of the non-phonological conditioning of oxen, the
irregular morph /ən/ qualifies as an allomorph of the regular /ɪz/. Both express
the same plural meaning and occur in complementary distribution: /ən/ is used
with ox, whereas /ɪz/ attaches to the majority of other nouns ending in /s/.
Accordingly, returning to the forms -ed and ex-, we must either grant them the
status of allomorphs of the same past time morpheme which occur in non-
phonologically conditioned complementary distribution or think of another
explanation for why this cannot be the case.
With regard to the latter, it appears that our reluctance to regard -ed and ex-
as allomorphs of the same morpheme also stems from the fact that these forms
occur in tactically different environments (Nida 1948: 421), i.e. they combine
with words belonging to different parts of speech. As stated above, -ed follows
verbs, whereas ex- precedes nouns. (In contrast, all other allomorphs that have
been dealt with so far occur in tactically identical environments. E.g. /ʌn/ and
/ɪn/ precede adjectives; /ɪz/ and /ən/ follow nouns; etc.) However, the occurrence
in tactically different environments does not suffice to deny -ed and ex- the
status of allomorphs: the fact that they combine with words belonging to
different parts of speech can be seen as an instance of complementary
distribution. Being aware of this fact, Nida (1948: 425) proposes the following
solution: -ed and ex- would qualify as allomorphic realizations of the same past
time morpheme if English had another past time signifier possessing the
distributional properties of both -ed and ex-, i.e. a form being able to occur both
after verbs (e.g. *workeg meaning 'worked in the past') and before nouns (e.g.
*eg-president meaning 'former president'). Since English lacks such a form, -ed
and ex- are not allomorphs of the same morpheme but morphs realizing two
different morphemes (which have identical signifieds).
Unfortunately, Nida does not really explain why the existence of such a form
would suffice to regard -ed and ex- as allomorphs of the same morpheme. That
is why it is not clear why the absence of such a form must suffice to deny -ed
and ex- the status of allomorphs. Apart from this, the conclusion drawn by Nida
is clearly at odds with what we said about the distribution of signifiers having
the same signified: as we concluded in 2.2.5, two signifiers associated with the
same signified are instances of the same sign (in our case, allomorphs of the
same morpheme), regardless of their occurrence in tactically different
environments.
Taking this into consideration, this book proposes a different solution. When
analyzing the distribution of synonymous forms, which seem to express the
same meaning, one can always try to find a slight semantic difference between
them and thus be justified in regarding the forms in question as morphs realizing
two different morphemes. This is a promising approach because, as has been
pointed out by many authors, absolute synonymy (i.e. the existence of two
forms which express exactly the same meaning) does not occur very often (see
e.g. Cruse 2004: 154-155). As an illustration, let us, first of all, consider the
distribution of the synonymous forms ex- and former: both carry the past time
meaning and occur before nouns like ambassador, husband, and president. Are
ex- and former of pairs like ex-president–former president allomorphs of the
same morpheme which occur in free variation or morphs that have slightly
different meanings and thus realize two different morphemes?
Since ex- and former are forms of different origin – ex- is Latin, former is
Germanic (OED) – we can perhaps analogize the distribution of ex- and former
to that of the negative morphs un- and in- and thus conclude that they are
allomorphs of the same morpheme which occur in stylistically relevant free
distribution. However, in contrast to un- and in-, ex- and former do seem to
express slightly different meanings. As argued by the blogger David Goddard
(http://tinyurl.com/5uwxlxc), "for a lot of people, [ex] gives off a somewhat
negative feel, whereas [former] seems somehow more dignified, more
amicable". In this connection, it is interesting to note that Google yields many
more search results for ex-President Bush than it does for former President
Bush.7 In May 2011 there were approximately 273000 hits for former President
Bush and approximately 3370000 hits for ex-President Bush. One explanation
for this could be the just named semantic difference between ex- and former. If
ex- does indeed "give off a somewhat negative feel", as suggested by David
Goddard, then Internet users who use ex-President Bush instead of former
President Bush do not only refer to the fact that George Bush was a U.S.
president before the moment of utterance (i.e. which ex- literally means) but also
(perhaps subconsciously) express their rather negative attitude towards him. In
other words, ex-president does not only mean 'a person who was president
before the moment of utterance' but 'a person who was president before the
moment of utterance and whom the utterer of ex-president does not particularly
like'. (We can say that the morph ex- has undergone what some linguists call the
subjectivisation of meaning (see e.g. Stein and Wright 1995), i.e. the meaning
of ex- has become 'subjective' in that it now carries a rather subjective speaker's
7
I am very grateful to Bridgit Nelezen for drawing my attention to this fact.
38 Chapter 2
attitude to the referent of the noun preceded by ex-.) Since George Bush, Jr. was
a rather unpopular U.S. president, it is not surprising that ex-President Bush
occurs more often than former President Bush.
If we are right with this hypothesis, we can conclude that ex- and former are
morphs that express (slightly) different meanings and hence realize two different
morphemes. Similarly, given that -ed of worked does not carry that 'somewhat
negative feel' inherent in ex- of ex-president, we can conclude that -ed of worked
and ex- of ex-president express (slightly) different meanings and, accordingly,
are not allomorphs of the same morpheme but morphs realizing two different
morphemes.
In summary: we have become acquainted with six main types of the
distribution of morphs in English:
1. Contrastive distribution of two non-identical morphs expressing non-

identical meanings. The morphs in question are realizations of two different
morphemes. E.g. the morphs /truː/ and /ˈeɪ.bl ̩/ realizing the morphemes
{true} and {able}.
2. Contrastive distribution of two identical morphs expressing non-

identical meanings. The morphs in question are realizations of two different
morphemes. E.g. the morphs /truː/ and /truː/ realizing the morphemes {true1}
'faithful' and {true2} 'not false'.
3. Stylistically irrelevant free variation of two non-identical morphs

expressing the same meaning. The morphs in question are allomorphs of
the same morpheme. E.g. /ˈdiːteɪl/ and /dɪˈteɪl/ realizing the morpheme
{detail}.
4. Stylistically relevant free variation of two non-identical morphs

expressing the same meaning. The morphs in question are allomorphs of
the same morpheme. E.g. the morphs /ʌn/ and /ɪn/ realizing the same
negative morpheme in uncommunicative and incommunicative.
5. Phonologically conditioned complementary distribution of two non-

identical morphs expressing the same meaning. The morphs in question
are allomorphs of the same morpheme. E.g. the morphs /ɪn/, /ɪm/, and /ɪ/
realizing the same negative morpheme in inadequate, impossible, and
illiterate.
6. Non-phonologically conditioned complementary distribution of two non-

identical morphs expressing the same meaning. The morphs in question
are allomorphs of the same morpheme. E.g. the morphs /ən/ and /ɪz/ realizing
the same plural morpheme in oxen and boxes.
2.4 The segmentation of words into morphemes
As we established in Section 2.1, the word untrue can be segmented into the
morphemes {un} and {true} because the meaning 'untrue' can be segmented into
the meanings 'not' and 'faithful', which the components un- and true express in
other environments: un- carries the meaning 'not' in e.g. unable and true carries
the meaning 'faithful' when used without un-, i.e. in sentences like Her lover had
been true. The word untrue is thus an instance of isomorphism of formal and
semantic segmentation (Plungian 2000: 39). That is, the formal segmentation
of the signifier untrue into the morphemes {un} and {true} is paralleled by the
semantic segmentation of the signified 'not faithful' into the meanings 'not' and
'faithful', inherent in the morphemes {un} and {true}. The same applies to
unable. The formal segmentation of the signifier unable into the morphemes
{un} and {able} is paralleled by the semantic segmentation of the signified 'not
able' into the meanings 'not' and 'able', inherent in the morphemes {un} and
{able}. Isomorphic forms like untrue and unable present no theoretical
difficulties and will therefore be largely left out of consideration in the
remainder of this section.
2.4.1 Anisomorphism. Full-idiomaticity
One manifestation of anisomorphism of formal and semantic segmentation

(which is the reverse of isomorphism) is full-idiomaticity. As was pointed out in
1.2.2, fully-idiomatic words include semantically modified words like boyfriend,
whose idiomatic meanings are still explainable in terms of their components'
literal meanings (recall that boyfriends are typically friends and former young
male children), and opaque words like understand, whose idiomatic meanings
are no longer transparent.
Unlike the isomorphic words untrue and unable, the idiomatic words
boyfriend and understand raise the following questions. Can boyfriend be
segmented into the morphemes {boy} and {friend}, even if boyfriend does not
mean 'a friend who is a boy'? Similarly: can understand be segmented into the
morphemes {under} and {stand}, even if understand does not mean 'to stand
under somebody or something' but 'to grasp the meaning or the reasonableness
of something'? In the following we will become acquainted with three possible
theoretical solutions.
40 Chapter 2
2.4.2 A purely semantic approach
If we want to adhere to the definition of a morpheme as a meaning-carrying unit,

the answer to the questions raised above can only be 'no'. Boyfriend cannot be
segmented into the morphemes {boy} and {friend} because the putative
components boy and friend do not carry the meanings 'boy' and 'friend', which
they carry in other environments. Likewise, understand cannot be segmented
into the morphemes {under} and {stand} because the putative components
under and stand do not carry the meanings 'under' and 'to stand', which they
carry in other environments. Boyfriend and understand are monomorphemic
words consisting of the morphemes {boyfriend} and {understand}, which carry
the idiomatic meanings 'a male sexual partner of almost any age over puberty'
and 'to grasp the meaning or the reasonableness of something'.
This solution, however, does not seem to be very convincing. Especially in
the case of understand, the purely semantic approach is at odds with the fact that
the verb under analysis is headed by the component stand. In morphology, the
term 'head' refers to the most important structural component of a complex
word, which determines its word-class as well as the inflectional marking (i.e.
bound morphs expressing obligatory grammatical meanings such as e.g. 'the past
tense'). Thus understand is a verb because it is headed by the verb stand, not by
the preposition under. Also, the past tense of understand is not *understanded
but understood. This is because the past tense form of the head element stand is
also stood, not *standed. Both these facts do not support the analysis of
understand as a monomorphemic word. (For similar observations, see Aronoff
1976: 14-15.)
2.4.3 Nida's purely formal approach
An example of a purely formal approach is Nida (1974: 58-59), who argues that
a morpheme can be isolated if it satisfies one of the following conditions:
1. It occurs "in isolation".
2. It occurs "in multiple combinations in at least one of which the unit with
which it is combined occurs in isolation or in other combinations".
3. It occurs "in a single combination provided that the element with which it is
combined occurs in isolation or in other combinations with other non-unique
constituents".
'In isolation' means that the signifier in question can form a one-word elliptical
sentence and thus fulfills the isolatability criterion which we discussed in the
previous chapter. Consider, for example, the underlined forms in (46), (47), and
(48).
(46) Here, in a piece called 'The Problem With Boys,' Tom Chiarella
broods, 'There is something odd and forbidden about the word boy.
Typing it feels a little creepy, almost pornographic. Boy. A little word,
naked and weak, an iconic expression of smallness, of vulnerability....
(COCA)
(47) Maleus found his cap on the chair arm and traced a finger around its
rim. Friend? 'Oh, there was...' He shook his head (COCA)
(48) Some of our friends trotted up to see my beautiful babe, stuck their
heads through the curtains. They tossed their heads, chortled and
nibbled the back of her neck. 'Come on, little one. Stand! Stand!'
(COCA)
As we can see, the signifiers boy, friend, and stand can form elliptical sentences
and, accordingly, are isolatable as morphemes (or to be more precise, as morphs
realizing morphemes) in accordance with Nida's Condition 1: it does not matter
that the meaning 'boyfriend' does not contain the meanings 'boy' and 'friend' and
the meaning 'understand' does not contain the meaning 'to stand'. As was already
mentioned in 2.1, morphemes like {boy} and {friend} of boyfriend and {stand}
of understand whose morphs are capable of occurring in isolation are
traditionally called free morphemes.
Condition 2 is concerned with bound morphemes, i.e. morphemes whose
morphs, in contrast to those of free morphemes, never occur in isolation.
Consider, for example, the component under of understand. Despite the fact that
the preposition under is usually separated from other signifiers by means of a
blank space (e.g. under the table, not *underthetable), under does not seem to
be capable of occurring in isolation. Perhaps one could imagine a hesitant
speaker who says Under. as a response to a question like Where is my bag?
uttered by another speaker. However, neither COCA nor BYU-BNC contain
elliptical sentences made up of under only. The same applies to other
prepositions. For instance, as Haspelmath (2011: 40) points out, in the PP to
Lagos, the preposition to is a bound form: it cannot occur on its own without
something following it. Indeed, as in the case of under, neither COCA nor BYU-
BNC contain elliptical sentences consisting only of the signifier to.
What follows from this is that the component under- of understand does not
fulfill Bloomfield's isolatability criterion and therefore does not qualify as a
word. However, it does fulfill Nida's Condition 2: under- of understand occurs
in multiple combinations – e.g. understand, under the table, under that, etc. – in
42 Chapter 2
at least one of which (e.g. understand) the unit with which it is combined (i.e.
stand) occurs in isolation. In other words, under- of understand is isolatable as a
morpheme because in the word under analysis it combines with the unit stand,
which occurs in isolation. Again, it does not matter that the literal meaning
'under' is not part of the meaning 'to understand'.
A similar example of a bound morph is the negative morph un- of untrue.
Like under- of understand, un- of untrue is also not capable of forming elliptical
sentences consisting of un- only. (Its bound character is more obvious than that
of under- because in contrast to the latter, the former is usually not separated
from other morphs by means of a blank space.) As in the case of under- of
understand, un- of untrue is also isolatable as a morph realizing a morpheme in
accordance with Nida's Condition 2: it occurs in multiple combinations (e.g.
untrue, unable) in at least one of which (e.g. untrue) the unit with which it is
combined (i.e. true) occurs in isolation: both COCA and BYU-BNC contain
elliptical sentences made up of true only.
Besides bound morphs like under- of understand and un- of untrue which
occur in combination with free morphs, there are bound morphs which occur in
other combinations. Consider, for instance, the verb receive. Applying the latter
part of Condition 2, we can segment receive into the morphemes {re} and
{ceive}, even though the meaning 'receive' cannot be segmented into two
independent meanings attributable to the putative morphs re- and -ceive. The
unit re- qualifies as a morph of a morpheme because in addition to occurring in
receive, it also occurs in reduce, refer, and retain. Likewise, the unit -ceive
qualifies as a morph of a morpheme because in addition to occurring in receive,
it also occurs in the verbs conceive, deceive, and perceive.
Finally, Nida's Condition 3 helps us deal with unique morphemes, i.e.
morphemes whose morphs occur only in one particular environment. An often
cited example is cran- of cranberry. As argued by many authors (e.g. Aronoff
1976: 10; Taylor 2002: 273), cran- is a unique morph because it occurs only in
combination with the free morph berry in the word cranberry. Strictly speaking,
this is no longer true of the English language: the unit cran does occur as an
elliptical version of cranberry in cranapple juice, cranapple crunch, cranapple
pie, and the like. E.g. (49).
(49) Cranapple Juice – A drink produced by squeezing or crushing

cranapples (http://www.drinkswap.com/cranapple-juice.htm)
Anyway, the point of Nida's Condition 3 is that bound unique units can be
isolated as morphemes when they occur in combination with free morphs. Thus
if cran of cranberry still remained a unique bound unit, it would nevertheless
qualify as a morph realizing a morpheme because in the word cranberry it
combines with the unit berry, which, like other nouns, is capable of occurring in
isolation.
2.4.4 Nida's approach and the conception of differential meaning
The major problem with Nida's formal approach seems to be the fact that it
grants the morphemic status not only to opaque units like under- and stand of
understand but also to entirely meaningless re- and -ceive of receive. Indeed, if
re- and -ceive are isolatable as morphemes even though they do not mean
anything (but only occur in other combinations), a question arises as to whether
we can preserve the definition of a morpheme as a meaning-carrying unit. That
is, if units that do not carry any meanings can nevertheless be regarded as
(morphs realizing) morphemes, can a morpheme be still regarded and defined as
a meaning-carrying unit?
According to Ginzburg et al. (1979: 24), Nida's formal approach can be
reconciled with the classic conception of a morpheme as a meaning-carrying
unit if we assume that opaque components which can be isolated as morphs
from a formal point of view only possess the differential meaning. As an
illustration, let us again consider understand. As was stated in 1.2.2, its putative
components under- and stand are opaque because the idiomatic meaning 'to
understand' can hardly be accounted for in terms of the meanings 'under' and 'to
stand', which under- and stand express in other environments. However,
regardless of this fact, both these units can be said to carry differential
meanings. This means that the unit under- contributes to the overall meaning of
understand by making understand distinguishable from e.g. withstand. And the
unit stand contributes to the overall meaning of understand by making
understand distinguishable from e.g. undergo. In a similar way, the meaningless
units re- and -ceive of receive can be said to possess differential meanings
because they make receive distinguishable from words like conceive, deceive,
perceive, etc. and reduce, refer, retain, etc. which likewise contain either the
empty unit re- or the empty unit -ceive.
The conception of differential meaning is, however, also far from being
unproblematic. If we accept the view that 'differential meaning' is a sufficient
morpheme-hood condition (i.e. the minimum requirement that the form in
question must fulfill in order to qualify as a morpheme), then almost any
combination of sounds or even individual sounds become isolatable as
morphemes. For example, we can argue that the word car consists of the
morphemes {c} and {ar}: the unit c- qualifies as a morpheme because it makes
car distinguishable from bar. And the unit -ar is a morpheme because it makes
car distinguishable from cable. But in this case, we would no longer be able to
distinguish between morphemes (meaning-carrying units) and phonemes
44 Chapter 2
(meaning-distinguishing units), since any sound capable of distinguishing

meaning (e.g. /k/ of car), which has been traditionally regarded as (a phone
realizing) a phoneme, automatically becomes (a morph realizing) a morpheme.
2.4.5 Mel'čuk's theory of quasi-linguistic units
Another attempt to reconcile Nida's formal approach with the traditional

conception of a morpheme as a meaning-carrying unit is Mel'čuk's theory of
quasi-linguistic units (2001: 278-285). According to Mel'čuk, quasi-linguistic
units are forms that are isolatable as morphemes from a formal but not from a
semantic point of view. For example, as we established in 2.4.3, the words
boyfriend and understand can be segmented into the morphemes {boy} /
{friend} and {under} / {stand}: the components boy, friend, and stand are free
morphs, which occur in isolation, and the component under- is a bound morph
which occurs in multiple combinations with free units like stand. However,
boyfriend does not mean 'a friend who is a boy' and understand does not mean
'to stand under somebody or something'. Accordingly, the units boy / friend and
under- / stand are not morphs realizing morphemes but quasi-linguistic units (or
quasi-signs).
A very important aspect of Mel'čuk's approach is the classification of quasi-
linguistic units into morfoids and submorphs. Morfoids are units that occur in
semantically motivated idiomatic words like boyfriend, whereas opaque words
like understand and words like receive which are segmentable into meaningless
units occurring in other combinations are said to consist of submorphs.
As Mel'čuk acknowledges, sometimes it is rather difficult to decide in favor
of either the morfoid or the submorph solution, since one and the same idiomatic
form can be perceived as fully-motivated by one speaker and as opaque by a
different speaker. However, as will be shown below, this fact must be seen not
as a shortcoming but as one of the major advantages of this approach. Consider,
for example, the fully-idiomatic adjective bananas 'crazy' (MWO), which often
occurs in the fully-idiomatic VPs go bananas and drive bananas. Similar to our
previous examples boyfriend and understand, bananas seems to be formally
segmentable into the free morph banana and the bound morph -s. However, the
meaning 'bananas' does not contain the meanings 'banana' and 'plurality', which
the components banana and -s express in other environments. Accordingly, the
units banana and -s are not morphs realizing morphemes but quasi-linguistic
units.
As regards the question of whether the components banana and -s are
morfoids or submorphs, it appears that the fully-idiomatic meaning 'crazy' has
nothing in common with the meanings 'banana' and 'plurality'. Accordingly, one
may be tempted to conclude that banana and -s are opaque submorphs rather
than motivated morfoids. However, as argued by the author of the Web page
http://www.englishdaily626.com/slang.php?054,
when apes are given a bunch of bananas, they eat them with
tremendous enthusiasm, as though they've lost their minds.
This is most likely a popular (or folk) etymology of bananas, i.e. an

explanation invented by some particularly creative speaker of English who
wanted to find at least some connection between the idiomatic meaning 'crazy'
and the literal meaning 'more than one banana'. Nevertheless, this explanation
can be used as a justification for the treatment of the components banana and -s
as morfoids: for those speakers of English who do indeed attribute the idiomatic
meaning 'crazy' of bananas to the craziness of apes caused by eating more than
one banana, the word bananas does indeed consist of the two morfoids banana
and -s, whose literal meanings 'banana' and 'the plural number' partially motivate
the idiomatic meaning 'crazy'.
The major advantage of Mel'čukian approach is thus that it allows us to
distinguish between different groups of speakers (of English or any other
language) who may have different judgments as to whether the idiomatic
meaning under analysis is opaque or is somehow motivated by its components'
literal meanings. In the former case, the word under consideration becomes
segmentable into submorphs; in the latter case into morfoids.
Note that the theory of quasi-linguistic units does not claim that any word
can be segmented into at least two submorphs. Thus, given the segmentation of
receive into the submorphs re- and -ceive, why can we not segment e.g. car into
the submorphs c- and -ar? As stated in 2.4.4, both these units occur in other
combinations – c- also occurs in cable and -ar also occurs in bar – and therefore
seem to qualify as submorphs. However, the fact that c- occurs not only in car
but also in cable and -ar occurs not only in car but also in bar seems to be a
matter of coincidence. The members of the pairs car / cable and car / bar do not
have much in common. That is why the segmentation of car into the units c- and
-ar does not seem natural. In contrast, the segmentation of receive into the units
re- and -ceive does not seem unnatural because -ceive makes receive
distinguishable from a semantically related word reception. Compare also
deceive and deception, reduce and reduction, and examples alike.
2.4.6 Anisomorphism. Partial idiomaticity
Having established how to deal with fully-idiomatic forms like boyfriend and
understand, we can now proceed to the morphemic analysis of words which are
partially idiomatic. We will begin with blackboard.
46 Chapter 2
As we said in 1.2.2, the meaning 'blackboard' contains the meaning of the

component board but not of the component black. A blackboard is not a black
board but a board for drawing or writing upon with chalk. Accordingly, of its
two components, which are isolatable as morphemes from a formal point of
view, only board can be regarded as a morph realizing a morpheme: it can occur
in isolation and its meaning is part of the meaning 'blackboard'. By contrast,
black is a morfoid whose literal meaning only motivates the idiomatic meaning
'board for writing': blackboards have dark surfaces and the black color is the
most prominent representative of dark colors.
Consider also the morphemic structure of twilight. From a formal point of
view, twilight is segmentable into the components light and twi-: light is a free
morph, which can occur in isolation, whereas twi- is a non-unique bound unit,
which occurs in multiple combinations with forms occurring in isolation. For
example, in addition to occurring in twilight, twi- also occurs in e.g. twi-headed
(meaning 'having two heads'.) However, from a semantic point of view, the
morphemic status can only be granted to the component light. Thus, according
to MWO, twilight does not mean 'two lights' or 'double light' but
the light from the sky between full night and sunrise or between sunset
and full night produced by diffusion of sunlight through the
atmosphere and its dust. (MWO)
It follows that the component twi- is not a morph but a quasi-linguistic unit.
With regard to the question of whether twi- of twilight is a morfoid or a
submorph, it appears that this unit, in contrast to black of blackboard, is
semantically opaque : it is not clear what the sense 'twilight' has in common with
the meaning 'two, double', which twi- expresses in other environments. As just
pointed out, twilight does not mean 'two lights' or 'double light' but 'the light
from the sky between full night and sunrise'. Accordingly, the component twi- of
twilight is not a morfoid but a submorph.
In summary, semi-idiomatic words can consist of:
1. one normal morph and one motivated morfoid (e.g. blackboard)

2. one normal morph and one opaque submorph (e.g. twilight)
As in the case of fully-idiomatic words, two different speakers may have

different judgments regarding transparency / opacity of the idiomatic component
of the semi-idiomatic word under analysis. In this case, one and the same semi-
idiomatic word becomes segmentable into one normal morph and one motivated
morfoid (this applies to the speaker for whom the idiomatic component is
transparent) and one normal morph and one opaque submorph (this applies to
the speaker for whom the idiomatic component is opaque).
2.4.7 Anisomorphism. Additional meanings
Finally, we will consider anisomorphic words whose meanings contain not only
their components' literal meanings but also additional, unpredictable idiomatic
meanings. Let us begin with the word football.
From a formal point of view, football is easily segmentable into the
components foot and ball: both can occur in isolation and thus qualify as free
morphs in accordance with Nida's Condition 1. However, football does not mean
'a ball for a foot' (which its components literally stand for), but 'a particular sport
that involves a ball and the players' feet trying to kick it to the goal at the
opposite side of the field' (OED). In other words, the meaning 'football' contains
the meanings of the components foot and ball – these components provide a
(somewhat incomplete) explanation for how football is played – but in addition
to these meanings, it also contains the idiomatic meaning 'a particular sport
different from basketball, handball, etc.', this meaning being inherent in neither
foot nor ball. Obviously, words like handball, basketball, volleyball, etc. exhibit
a similar semantic structure. Their meanings contain their components' literal
meanings, which provide an incomplete explanation for how these games are
supposed to be played – e.g. handball is the game that involves a ball and the
players' hands throwing it to the goal at the opposite side of the field – but in
addition to these meanings, they also contain the idiomatic meaning 'a particular
sport different from other sports'.
Consider also the meaning of stealer in (50).
(50) But more than that, he feels guilty. Because now he is a thief, and not
just any thief, but a stealer of dreams and wishes (COCA)
As argued in some studies, the word stealer does not exist in English because
the meaning 'a person who steals' is expressed by the signifier thief. However, as
(50) demonstrates, this is not so. According to the OED, the noun stealer
appeared in the English language in 1508. It originally meant 'one who steals; a
thief' (and thus used to be a full-synonym of thief) but with the course of time
has undergone semantic narrowing and, as a result, come to mean 'one who
steals something specified': Example (50), in which stealer is contrasted with
thief, is a good illustration of the fact that in Present-day English the meaning
'stealer' is indeed narrower than the meaning 'thief'.
A similar case is the noun writer of (51).
(51) A prolific writer, he has published three short story collections, 12

picture books (including the highly acclaimed Zoom trilogy) and three
young adult novels (COCA)
48 Chapter 2
At first glance, it may seem that the meaning of the writer of (51) is made up of
the meanings of its components: the free unit write, which is capable of
occurring in isolation, and the non-unique bound unit -er, which occurs in
multiple combinations with units occurring in isolation; e.g. browser, preacher,
stealer, teacher. That is, writer means 'a person who writes, a performer of the
action of writing'. However, not any person who writes qualifies as a writer. For
instance, a person who is writing an e-mail message can hardly be referred to as
a writer. (And a person who has written a lot of e-mail messages definitely
cannot be referred to as a prolific writer.) A writer is a person who produces
particular products of writing: story collections, picture books, adult novels, and
the like. In other words, in addition to the meanings 'to write' and 'performer of
some action', inherent in the components write and -er, the meaning 'writer'
contains an additional idiomatic meaning 'particular products of writing', which
cannot be attributed to either write or -er.
Recall also the formation waithood, which we discussed in Chapter 1. As we
concluded in 1.2.2, the meaning 'waithood' does not only contain the meanings
'to wait' and 'stage', inherent in the components wait and -hood, but also the
idiomatic meaning 'a particular waiting stage involving a colleague graduate
who is waiting for a good job and / or financial security in his life and therefore
postpones marrying'.
Given these semantic structures of the words football, stealer, writer, and
waithood, we need to answer the question of where the idiomatic meanings 'a
particular sport' of football, 'specified objects of stealing' of stealer, 'particular
products of writing' of writer, and 'a particular waiting stage' of waithood
actually come from. We do not need to be concerned with the question of
whether these words can be segmented into the morphemes {foot} / {ball},
{steal} / {er}, {write} / {er}, and {wait} / {hood}. The answer to this question is
obvious: if the meanings 'football', 'stealer', 'writer', and 'waithood' contain the
meanings of these components, then all of them qualify as morphs realizing
normal morphemes.
To account for the presence of the additional idiomatic meanings in football,
stealer, writer, and waithood, we can resort to either the mega-morph or the
zero morph approach. In the former case, we attribute the idiomatic meaning
in question to one of the two overt components of the word under analysis. This
component then becomes a mega-morph (Mel'čuk 2001: Ch. 7), that is, a
morph which cumulatively expresses more than one signified. For instance, in
the case of stealer, we can conjecture that the idiomatic meaning 'specified
objects of stealing' is inherent in the morph -er, which cumulatively expresses
this meaning together with the literal meaning 'performer of some action'. We
are justified in arriving at this conclusion because in English there is also the
semantically and formally related noun writer, whose signified does not only
contain the meanings 'to write' and 'performer of some action' (i.e. the literal
meanings of the overt components write and -er) but also the idiomatic meaning
'particular products of writing'. Similarly, it appears that in football the idiomatic
meaning 'a particular sport' is inherent in the morph ball rather than in foot. This
is because in addition to football, there are the words handball, basketball,
volleyball, etc., whose signifieds exhibit the same idiomatic pattern as football.
An alternative to the mega-morph approach is the zero morph approach, i.e.
the attribution of an idiomatic meaning to a zero morpheme, i.e. a morpheme
that does not have an overt morph. For instance, instead of attributing the
idiomatic meaning 'specified objects of stealing' to the morph -er, we can
segment the word stealer into:
x the overt morph steal carrying the meaning 'to steal'

x the overt morph -er carrying the meaning 'performer of some action'
x a covert zero morph {ø} carrying the idiomatic meaning 'specified objects of
stealing'.
Similarly:
x writer = {write} + {er} + {ø}, in which {ø} is a covert zero morph carrying
the idiomatic meaning 'particular products of writing'.
x football = {foot} + {ball} + {ø}, in which {ø} is a covert zero morph

carrying the idiomatic meaning 'a particular sport'.
x waithood = {wait} + {hood} + {ø}, in which {ø} is a covert zero morph

carrying the idiomatic meaning 'a particular waiting stage'.
This textbook rejects both these approaches. The reason for this is that both the
mega-morph approach and the zero-morph approach are based on what Plungian
(2000: 39) calls the additive principle of morphology, i.e. a model describing
an ideal signifier–signified relation. In accordance with this model, if Signified
A can be segmented into independent meanings X and Y, then A's signifier must
likewise be segmentable into signifiers {x} and {y} carrying meanings X and Y.
The additive model works perfectly with isomorphic words like untrue. That
is, we have the impression that the signified 'untrue' is a semantically complex
signified segmentable into two independent meanings: 'not' and 'faithful'. Given
this and given that un- and true express the meanings 'not' and 'faithful' in other
environments, we can easily segment untrue into the morphs un- and true. By
contrast, anisomorphic words like stealer are perceived as deviations from the
additive model. Thus, while the signified 'stealer' is segmentable into three
independent meanings 'to steal', 'performer of some action', and 'specified
objects of stealing', the signifier stealer can be segmented only into two morphs:
50 Chapter 2
steal and -er. Similarly, while the signified 'waithood' is segmentable into three
independent meanings 'to wait', 'stage', and 'a particular waiting stage', the
signifier waithood is segmentable only into two morphs: wait and -hood. To
account for these deviations, we then need to attribute the idiomatic meanings
'specified objects of stealing' and 'a particular waiting stage' to either one of their
overt components, which then becomes a mega-morph, or to a covert zero
morph.
But is this really necessary? Do anisomorphic words like stealer and
waithood represent deviations from the 'normal' signifier–signified relation
exemplified by fully-isomorphic words like untrue? The answer to this question
is of course 'no'. As has been repeatedly pointed out by many authors (especially
those who study idiomatic meanings), fully-isomorphic complex words like
untrue whose signifieds are representable only in terms of their components'
signifieds are considerably outnumbered by anisomorphic words like stealer and
waithood whose meanings contain not only their components' literal meanings
but also additional, unpredictable idiomatic meanings. In other words, the
default case is represented by an anisomorphic complex word like stealer rather
than by an isomorphic complex word like untrue. (This will become especially
obvious in Chapters 4 and 5.)
Taking this into account, we can conclude this section with the claim that the
anisomorphic words football, stealer, writer, waithood, etc., whose signifieds
contain additional idiomatic meanings, must be segmented only into two normal
morphemes, i.e. the morphemes {foot} and {ball} of football, {steal} and {er}
of stealer, {write} and {er} of writer, and {wait} and {hood} of waithood.
Given the default character of the signifier–signified relation exemplified by
these words, there is no need to account for additional idiomatic meanings by
means of postulating mega-morphs or zero morphs.
2.5 The hierarchy of morphs and units alike
The last section of this chapter deals with the hierarchical differences between
different morphs (as well as morfoids and submorphs) into which complex
forms can be segmented.
As the starting point, let us consider the morphs un- and true of untrue. It
appears that one of these morphs – the morph true – is both formally and
semantically more independent than the morph un-. From a formal perspective,
true is a free morph, which can occur in isolation, whereas un- is a bound morph
which occurs in combination with free morphs like true and able. Similarly,
from a semantic perspective, the meaning 'true', inherent in the morph true,
seems to be a more independent meaning than the negative meaning inherent in
the morph un-: while the former can be expressed by true alone, the latter
requires a combination of un- with free morphs like true and able.
Traditionally, this contrast between morphs like un- and true of untrue has
been captured with the help of the affix–root dichotomy. That is, the morph true
is the root of the word untrue, whereas the morph un- has the status of an affix.
2.5.1 Affixes versus roots
The most important differences between roots and affixes are as follows.
x Affixes are bound units; roots can be both free and bound.
x Affixes are shorter than roots.
x Roots are obligatory elements; affixes are optional elements.
Now, let us briefly discuss each of these differences. The defining characteristic
of an affix is that it is a bound unit, which occurs only in combination with other
units. In contrast, a root can be represented by both a free and a bound element.
Prototypical roots are free units like true of untrue, foot and ball of football, etc.,
but bound roots can be found as well.
Consider, for instance, the noun history. From a formal point of view,
history can be segmented into the components histor- and -y. Both of them occur
in other combinations: the unit histor- also occurs in historical and the unit -y
also occurs in theory. (The word theory is thus likewise segmentable into the
components theor- and -y.) With regard to the semantic perspective, it is clear
that the meaning 'history' cannot be segmented into two independent meanings
attributable to the putative morphs histor- and -y. Like re- and -ceive of receive,
histor- and -y of history do not have discernible meanings of their own and thus
qualify as quasi-linguistic units of the submorphic type.
With regard to the hierarchy of the submorphs histor- and -y, it is intuitively
clear that the unit histor- is more like a root, whereas -y is more like an affix.
This impression arises, however, not because of the free–bound distinction
between them – both histor- and -y are bound submorphs, which never occur in
isolation – but because the submorph histor- is longer than the submorph -y:
whereas the latter is made up of only one sound /ɪ/, the former consists of six
sounds: /ˈhɪstər/.
The second often named difference between roots and affixes is thus that the
latter are typically shorter than the former. In English there are affixes that
consist of only one sound. For example, there are affixes like /ɪ/ of cloudy, rainy,
etc. (which is semantically unrelated to the /ɪ/ of history and theory) or /æ/ of
52 Chapter 2
asymmetric /æsɪˈmɛtrɪk/, asymbolic /æsɪmˈbɒlɪk/, etc. In contrast, the shortest

English root consists of two sounds: /aɪ/ of eye and I. Hence, concluding our
analysis of history, we can argue that this word consists of the bound root histor-
and the affix -y because the latter is much shorter than the former. 8
Third, while roots are obligatory elements, affixes are optional. That is, a
word must contain at least one root, but it may contain no affixes. For example,
the monomorphemic words table, boy, friend, dog, cat, etc. contain no affixes.
Accordingly, there can be no words consisting of affixes only. That is why
words like history cannot be segmented into two bound affixes: histor- and -y.
Given the general obligatoriness of roots, one of these units must be necessarily
regarded as a root.
Finally, it is sometimes argued that roots are different from affixes with
regard to their signifieds: roots usually express more concrete signifieds than
affixes. As an illustration of this, let us consider the word morphology. From a
formal point of view, morphology can be segmented into three units: morph, -o-,
and -logy. The unit morph is a free unit, which is capable of occurring in
isolation, and thus can be regarded as a root of morphology. The unit -o- is a
non-unique bound unit, which occurs in combination with other free units. E.g.
phonology. Since the unit -o- consists of only one sound, it can be regarded as
an affix. Finally, the unit -logy is a non-unique bound unit which occurs in
words like morphology, phonology, lexicology, etc., where it carries the meaning
'names of sciences, departments of study' (OED): whereas morphology is the
branch of linguistics that is concerned with morphs, phonology can be defined
as the study of phones; lexicology is the study of the lexis; and so on. (The word
morphology can thus be segmented into the morphs morph and -logy and the
meaningless submorph -o-, which serves as an interlinking element between
them.) The meaning 'study', expressed by the morph -logy, seems to be a more
independent and less abstract meaning than e.g. the meaning 'negation', inherent
in the morph un- of untrue. Whereas the former can be easily paraphrased as the
acquisition of some knowledge, the latter represents one of the most difficult
and hardly definable concepts. Given this fact, some authors conclude that
despite its bound status, the morph -logy is a root rather than an affix.
This textbook rejects the categorical analysis of -logy of morphology as a
root, especially if this analysis takes into account only the 'not so abstract' nature
of the signified inherent in -logy. It is of course true that in contrast to the formal
categories which we discussed in the previous chapter (i.e. the word, the phrase,
the clause, and the sentence), the morpheme is both a formal and a semantic
category. However, the distinction between the root and the affix seems to be a
formal rather than a semantic distinction: as stated above, the defining
characteristic of an affix is that it is a bound morph, which cannot occur in
8
In a similar way, we can argue that receive consists of the bound root -ceive and the affix re-
because the latter is shorter than the former.
isolation. Accordingly, roots must be distinguishable from affixes only on the

basis of formal, not semantic criteria. With regard to the former, there is no
special reason to regard -logy as a root. First of all, the morph -logy meaning
'names of sciences, departments of study' never occurs in isolation. Second, in
English there are affixes which, like -logy /lədʒɪ/, consist of five sounds: for
example, the adjective-building affix -aceous /ˈeɪʃəs/ of words like curvaceous,
rudaceous, etc. Finally, the analysis of -logy as an affix does not violate the
obligatoriness requirement: the word morphology has the root morph, so that
both the units -o- and -logy can be regarded as affixes.
2.5.2 Combining form as a distinct morpheme type?
Morphs like -logy of morphology that have the formal properties of affixes and
the semantic properties of roots are often called combining forms. According to
the OED, in Present-day English there are 2179 combining forms. These
include, for example:
x -babble of nouns like edubabble, neurobabble, designer-babble, etc. which

denote 'various types of confusing or pretentious jargon, esp. that
characteristic of a specified field or group' (OED).
x -core of nouns like emo-core, noisecore, sadcore, etc. which designate

'(usually more extreme or intense) subgenres of popular music, esp. punk,
grunge, techno, and heavy metal' (OED).
x e- of nouns like e-commerce, e-journal, e-signature, etc. which denote

'involvement in electronic media and telecommunications (esp. the use of
electronic data transfer over the Internet, etc.), usually to distinguish objects
or actions from their non-electronic counterparts' (OED).
x -erati of nouns like belligerati, chatterati, geekerati, etc. which designate

'elite or prominent groups of people who are associated with what is
specified by the stem word' (OED).
x -tainment of nouns like docutainment, edutainment, irritainment, etc. which

denote 'genres of broadcasting, journalism, etc., in which entertainment is
combined with aspects of the genre, etc., indicated by the first element'
(OED).
As in the case of -logy of morphology, the answer to the question of whether a

particular combining form can be regarded as a root or an affix must depend
54 Chapter 2
only on the fulfillment of the formal criteria which were introduced in 2.5.1.
That is, for example, despite the fact that e- of e-commerce, e-journal, e-
signature, etc. is associated with a 'not so abstract' meaning 'electronic', the
combining form under analysis must be regarded as an affix: like /ɪ/ of cloudy
and /æ/ of asymmetric, /iː/ of e-commerce consists of only one sound. By
contrast, the combining form -tainment /ˈteɪnm(ə)nt/ of e.g. docutainment
consists of eight sounds and therefore must be regarded as a bound root.
In the case of combining forms like -babble of designer-babble, -core of
sadcore, and -erati of geekerati, etc. which cannot be analyzed as either roots or
affixes only because of the length of their signifiers, the preference must be
given to the affix solution. That is, all these combining forms can be regarded as
affixes unless one can show that these forms are capable of occurring in
isolation.
Note that the existence of the noun babble, which, according to the OED,
can express the meanings:
x 'inarticulate or imperfect speech, such as that of infants; prattle'

x 'idle, foolish, or unseasonable talk; prating'
x 'confused murmur, as of a stream'
does not suffice to regard the combining form -babble of e.g. designer-babble as
a root. There is no doubt that the latter is semantically related to the former:
according to the OED, the combining form -babble actually goes back to the
noun babble. However, -babble of designer-babble does not refer to an
inarticulate or improper speech, an idle or a foolish talk, or a confused murmur.
-Babble of designer-babble refers to a particular jargon used by designers.
Accordingly, given the semantic non-identity between the noun babble and the
combining form -babble, the existence of the former cannot serve as a
justification for the root analysis of the latter.
A different case is represented by the combining form must- of e.g. a must-
read, a must-see, a must-have, etc., which 'denote things that are essential,
obligatory, or highly recommended' (OED). The combining form must- of must-
read seems to be semantically identical with the verb must of e.g. You must read
this book. That is, a must-read is a book, an article, etc. which other people think
you must read. Accordingly, must of must-read is not an affix but a root. (Given
the non-bound character of must in words like must-read, must-see, must-have,
etc., it is doubtful that this morph can be regarded as a combining form.)
Returning to the question raised in the title of this part of Section 2.5, we can
say that so-called combining forms do not constitute a distinct morpheme type in
English. Combining forms are bound morphs that are semantically different
from prototypical affixes in that they express less abstract meanings. However,
they typically possess all other properties characteristic of affixes: they are
bound morphs, which do not occur in isolation, and their signifiers are not much
longer than those of prototypical affixes. Accordingly, in the overwhelming
majority of cases combining roots can be analyzed as affixes.
2.5.3 One signifier Æ both a root and an affix
As illustrated by the combining form -babble of e.g. designer-babble and the

noun babble, one and the same signifier can occur both as an affix and a root. A
similar example involving a prototypical affix is out- of verbs like outdo,
outnumber, outsell, etc. Despite the fact that the signifier out can occur in
isolation, out- of these verbs cannot be regarded as a root. Let us compare the
meanings of out in (52) and in the previously mentioned verbs outdo,
outnumber, and outsell.
(52) You will be the first student who will NOT be in my violin class. Out!
(COCA)
The out of (52) is an adverb whose meaning is described by MWO as 'in a

direction away from the inside or center' / 'outside'. By contrast, the out- of
outdo, outnumber, and outsell carries the meaning 'in a manner that exceeds or
surpasses and sometimes overpowers or defeats' (MWO): i.e. outnumber means
'to exceed in number'; outsell means 'to exceed in selling'; outdo means 'to
surpass in doing or performing'.
It is important to observe that the latter meaning is expressed by out- only
when it is followed by verbs like do, number, and sell: it never carries this
meaning when used in isolation, i.e. in contexts like that of (52). Accordingly,
this is clearly a bound affixal meaning, so that out- of outdo, outnumber, and
outsell must be regarded as an affix, not a root. (Note that out- of outnumber
cannot be regarded as a morfoid. In the case of outnumber and verbs alike, there
is no idiomaticity: in all these verbs out- carries the non-idiomatic meaning 'in a
manner that exceeds or surpasses', which it does not express when used in
isolation.)
Now, let us also consider the status of out in the verb outsource of (53).
(53) The company outsources many of its jobs to less developed countries
(MWO)
As defined by the OED, outsource means 'to obtain (goods, a service, etc.) by
contract from an outside source; to contract (work) out'. (The overall meaning of
outsource thus contains the meanings of both of its components out and source
plus the idiomatic meaning 'to obtain from', which is not inherent in either out or
56 Chapter 2
source.) Given this definition, we can conclude that in contrast to out- of e.g.
outnumber, out of outsource carries the meaning 'outside', which it also carries
when used in isolation. Accordingly, in contrast to out- of outdo, outnumber,
and outsell, out of outsource is not an affix but a root.
In summary, the signifier out can occur both as a root and an affix. When
used in isolation and in words like outsource where it carries the meaning
'outside', out must be regarded as a root. In contrast, when occurring in verbs
like outdo, outnumber, and outsell where it carries the meaning 'in a manner that
exceeds or surpasses', out- must be regarded as an affix.
2.5.4 Typology of affixes
With respect to their position in relation to the root, English affixes are
traditionally classified into the following four categories:
1. prefixes
2. suffixes
3. interfixes
4. infixes
Prefixes are affixes that precede the root: e.g. un- of untrue, out- of outnumber,
etc. Suffixes are affixes that follow the root: e.g. -er of teacher, -y of cloudy, etc.
Interfixes are meaningless affixes that serve as interlinking elements between
either two roots or one root and one meaningful suffix: e.g. -o- of morphology.
Infixes are suffixes that 'break' the root. For example, the slang word
edumacation 'an education that someone received at a crappy school, or a lack
of education all together' (Urban Dictionary) contains the infix -ma- 'poor
quality', which is inserted into the root educate.
With regard to their function, affixes are traditionally classified into
inflectional and derivational affixes. The most important differences are as
follows.
x Inflectional affixes express obligatory grammatical meanings; derivational

affixes express optional lexical meanings.
x Inflectional affixes are more productive than derivational affixes.
x Forms differing only with regard to the presence / absence of some

inflectional affix always express the same lexical meaning; forms differing
only with regard to the presence / absence of some derivational affix always
express different lexical meanings.
x Forms differing only with regard to the presence / absence of some

inflectional affix are always members of the same word class; forms differing
only with regard to the presence / absence of some derivational affix are very
often members of different word classes.
x Derivational affixes expressing temporal concepts can be iterated;

inflectional affixes expressing temporal concepts cannot be iterated.
As in 2.5.1, let us elaborate on each of these differences. As was already

mentioned in 2.1, lexical meanings are optional meanings which are expressed
only when language users specifically want them to be expressed. Grammatical
meanings, by contrast, are obligatory meanings which cannot be unexpressed.
To illustrate this difference, let us again compare the past tense suffix -ed of e.g.
I worked and the past time prefix ex- of e.g. ex-president. As we established in
2.3, these affixes express almost identical meanings and, accordingly, can be
regarded as allomorphs of the same morpheme, unless one can show that ex- has
an additional negative meaning, which is not inherent in -ed. However, while the
past tense meaning inherent in -ed is an obligatory grammatical meaning, the
past time meaning inherent in ex- is an optional lexical meaning. Thus in stark
contrast to verbs, which in English are almost always marked with regard to
tense, English nouns are essentially tenseless. That is, while the absence of -ed
typically locates the verb in question (which forms the past tense with -ed) in the
present (e.g. talk, walk, work), the absence of ex- does not necessarily mean that
the noun in question denotes a non-past characteristic. For example, the noun
president can be used in connection with a person who is no longer president,
even when it is not preceded by the past time marker ex-. E.g. (54).
(54) In 2005, President Clinton established CGI to turn ideas into action
and to help our world move beyond the current state of globalization
[…] (http://tinyurl.com/26jko5z)
Note also that ex- is a non-obligatory past time marker for nouns: instead of
saying ex-president, we can say former president. By contrast, -ed is an
obligatory past tense marker for verbs like talk, walk, work: no other morph can
be used instead of -ed in combination with these verbs. Given these differences,
we are justified in concluding that -ed is an inflectional suffix and ex- is a
derivational prefix.
In addition to the obligatoriness–optionality distinction, inflectional affixes
are usually more productive than derivational affixes. For example, the past
tense marker -ed can combine with almost any English verb: e.g. talked, worked,
destroyed, deduced, turned, etc. In contrast, the past time marker ex- combines
with a relatively small number of nouns. According to the OED, it occurs with
58 Chapter 2
nouns denoting 'titles of office or dignity' (e.g. ex-president) or 'designating

persons with respect to their calling, station, character, etc.' (e.g. ex-boyfriend).
However, it does not seem to occur with object nouns. For instance, looking at
what has remained of his or her house after an earthquake or a tornado, a
speaker of English will most likely not say This is my ex-house meaning 'this
used to be my house in the past'. Likewise, the formations ex-chair, ex-room, ex-
table, etc. do not seem to be possible in similar contexts.
English inflectional affixes never change the lexical meaning of their input
signifiers (i.e. signifiers to which they are attached). That is, both the singular
book and the plural books have the same lexical meaning 'book'; both the present
tense work and the past tense worked have the same lexical meaning 'work'; both
the non-progressive read and the progressive am reading have the same lexical
meaning 'read'; etc. By contrast, the addition of a derivational affix always
results in a change of lexical meaning. For example, while teach means 'to
teach', teacher means 'performer of the action of teaching'; while cloud means
'cloud', cloudy means 'full of clouds'; while number means 'number', outnumber
means 'to exceed in number'; etc.
English inflectional affixes never change the word class of their input
signifiers. For example, book and books are both nouns; work and worked are
both verbs; pretty and prettier are both adjectives; etc. By contrast, derivational
affixes often change the word class: e.g. teach is a verb, but teacher is a noun;
cloud is a noun, but cloudy is an adjective; number is a noun, but outnumber is a
verb, etc.
Derivational affixes expressing temporal concepts can be iterated. For
instance, in order to express the meaning 'former boyfriend prior to another
former boyfriend', we can say ex ex-boyfriend. By contrast, we cannot say
*workeded in order to express a similar anterior meaning 'had worked' (i.e. an
action that took place before some other action in the past).
Very often it is also argued that inflectional affixes are semantically more
regular than derivational affixes. This means that anisomorphic complex words,
whose signifieds are not (or not entirely) representable in terms of their
components' signifieds, can be found only among derivational formations like
stealer and writer, but not among inflected forms like boys, worked, prettier,
etc. Indeed, it seems that the obligatoriness of grammatical meanings must a
priori disallow their idiomaticity. That is, if the meaning 'the past tense' is an
obligatory grammatical meaning, then any verb that contains the past tense
suffix -ed must express the past tense meaning: namely, that the action
expressed by the root of the verb took place before the moment of utterance. (By
contrast, the optionality of lexical meanings inherent in derivational affixes must
sanction their idiomaticity.) However, as we will see below, this is not always
the case. Consider, for example, the meanings inherent in the underlined
inflectional affixes occurring in (55) and (56).
(55) CONAN: All right, Don, thanks very much for the call. And Myron, we
just have a couple of minutes left with you. I wanted to ask you to tell
the story of your father taking you to see the Brooklyn Dodgers in 1947.
Mr-UHLBERG: I'll never forget that. That was April 15th, 1947. Jackie
Robinson, the first African-American player to play in the white major
leagues […] (COCA)
(56) I don't want to blame Christopher for all the things I can't do anymore.
He is always trying to interest me in inactive activities -- reading,
collage, wine-tasting (COCA)
I wanted to ask you […] of (55) is grammatically in the past tense: cf. I wanted
to ask you and I want to ask you. However, the meaning expressed by the former
in (55) is clearly 'the present tense': I wanted to ask you means 'I want to ask you
now'. If this were not the case, Mr. Uhlberg would not start telling the story of
his father taking him to see the Brooklyn Dodgers in 1947.
He is always trying to interest me in inactive activities […] of (56) is
grammatically in the progressive aspect: cf. He is always trying to interest me in
inactive activities and He always tries to interest me in inactive activities.
However, in (56) the former does not refer to an ongoing action coinciding with
the moment of utterance (e.g. I am now reading a book) but expresses a
habitual situation recurring on a more or less regular basis. That is, He is
always trying to interest me in inactive activities means that 'he tries to interest
me in inactive activities whenever he has the opportunity to do so'.
These examples illustrate that idiomaticity is characteristic not only of
complex words containing derivational affixes like -er of stealer and worker but
also of complex forms containing inflectional affixes like -ed of wanted and -ing
of is trying: the former can be used in connection with non-past events and the
latter can refer to non-progressive events.
2.5.5 Typology of roots
Like affixes, roots can also be classified into those which express optional
lexical meanings and those which express obligatory grammatical meanings.
The latter are known as analytic forms. Consider, for instance, the comparative
degree of the adjective beautiful: more beautiful. In contrast to e.g. prettier, the
grammatical meaning 'the comparative degree' is expressed here not by the
inflectional affix -er but by the free form more, which is capable of occurring in
isolation, thus qualifying as a root.
A well-known peculiarity of the English language is that a number of
grammatical meanings are expressed both inflectionally and analytically. Recall
the discontinuous progressive morph am…ing of I am reading, which we
60 Chapter 2
discussed in 2.1. This morph represents a combination of the analytic form am

and the inflectional affix -ing. Similarly, the discontinuous passive morph
was…ed of e.g. was replaced by is a combination of the analytic form was and
the inflectional affix -ed.
2.6 Exercises
a) The signifier of a morpheme is called the morph.

b) One and the same signifier is never associated with more than one signified.
c) The signifier is the most important component of a linguistic sign.
d) Allomorphs of the same morpheme can be in contrastive distribution.
e) A morfoid is an opaque component of an idiomatic word.
f) A portmanteau morph is a morph that cumulatively expresses more than one
meaning.
g) Semi-idiomatic words can be segmented either into one normal morph and
one morfoid or into one normal morph and one submorph.
h) Roots are typically shorter than affixes.
i) Combining forms are always free morphs.
j) Inflectional affixes express optional lexical meanings.
3. Analyze the distribution of the following morphs. Explain your analysis.
a) table and sample

b) /d/ of /steɪd/ in e.g. he stayed and /ɪd/ of /ækˈsɛptɪd/ in e.g. he accepted
c) cat and mat
d) in- of intolerable and ir- of irresponsible
e) in- of inability and non- of non-ability
f) /iˈkɑnəmi/ and /əˈkɑnəmi/ of economy
g) /breɪv/ of brave and /ˈbrɛv/ of brevity
h) -er of teacher and -eer of auctioneer
i) -s of boys and -ren of children
j) -ed of he has worked and -en of he has taken
4. Analyze the morphemic structure of the following words:
a) greenhouse
b) table
c) barman
d) foretell
e) within
f) permit
g) great
h) after-party
i) redo
j) birdbrain
5. Name all possible characteristics of the following units:
a) -ist of biologist
b) has of he has worked hard
c) mat
d) -en of taken
e) -less of careless
f) -en of oxen
g) -gress of progress
h) fore- of foresee
i) red of redneck
j) cat
2.7 Further reading
My understanding of how words must be segmented into morphemes has been

largely influenced by Plungian (2000: Ch. 2, part 1) and Mel'čuk (2001).
Unfortunately, I can recommend these works only to those students of English
morphology who can understand Russian. Those who can understand French are
referred to Mel'čuk (1997), which is the original French version of Mel'čuk
(2001). A concise English introduction to the main concepts of Mel'čukian
morphology such as e.g. the concept of a mega-morph can be found in Mel'čuk
(2006: Ch. 7); see also an earlier work Mel'čuk (1982).
Classic works dealing with the distribution of morphs are Harris (1942),
Hockett (1947), and Nida (1948). If you wish to refresh your knowledge of
phonetic processes like place-of-articulation assimilation, see Cruttenden (2008:
especially part III). For differences between phonological and lexical
conditioning of allomorphs occurring in complementary distribution, see
Haspelmath and Sims (2010: 25-26). For an introduction to various semantic
relations including absolute synonymy, see Cruse (2004: especially Ch. 8). A
62 Chapter 2
classic article dealing with the differences between polysemy and homonymy is
Jakobson (1972).
Section 2.5.1, which presented the differences between roots and affixes,
was largely based on Mel'čuk (2001: 69-79). Since the book is in Russian, I
recommend a similar discussion in Haspelmath and Sims (2010: 19-22). A
similar analysis of forms that exist both as roots and affixes can be found in Plag
(2003: Sec. 4.1). For a discussion of the differences between inflectional and
derivational affixes, see Haspelmath and Sims (2010: Ch. 5), Bauer (2003),
Plungian (2000: Ch. 1, part 2), Jakobson (1959).
Finally, it must be noted that not all morphologists accept the views that a
morpheme is a meaning-carrying unit and that words must be segmented into
morphemes. If you want to become acquainted with these views, see Aronoff
(1976) and Anderson (1992).
3 Analyzing English lexemes
In this chapter we will be concerned with the formal and semantic structure of
English lexemes. Section 3.1 provides a definition of a lexeme. Section 3.2 deals
with the distribution of English lexemes as well as their typology from both a
formal and a semantic point of view. As regards the distribution of lexemes, the
section will argue that like morphs that realize two different morphemes, lexes
that realize two different lexemes occur in contrastive distribution, while
allolexes of the same lexeme, like allomorphs of the same morpheme, can either
be in complementary distribution or in free variation. Section 3.3 introduces the
concept of a vocable and discusses all possible semantic relations that can hold
between two or more lexemes which form the same vocable. Finally, Section 3.4
briefly touches on the concept of a lexeme family.
3.1 What is a lexeme?
A lexeme can be defined as a conventionalized association between any free

form and some particular lexical meaning. For example, the conventionalized
association between the signifier cat and the lexical meaning 'cat' forms the
lexeme CAT; the conventionalized association between the signifier untrue and
the lexical meaning 'not faithful' forms the lexeme UNTRUE; the conventionalized
association between the signifier waithood and the lexical meaning 'a particular
waiting stage in the life of a college graduate' forms the lexeme WAITHOOD; etc.
Consider also the idiomatic VP kicked the bucket of (57) and the idiomatic
clause (58).
(57) There's the guy who kicked the bucket after a shot (COCA)
(58) A bird in the hand is worth two in the bush (COCA)
The VP kicked the bucket is a fully-idiomatic phrase whose meaning contains

only the grammatical past tense meaning of the inflectional affix -ed but not the
literal lexical meanings of the components kick, the, and bucket: kicked the
bucket does not mean 'kicked some particular bucket' but 'died'. Similarly, (58)
is a fully-idiomatic clause which does not mean 'a bird in some particular hand
is worth two in some particular bush' but 'it's better to have a small real
advantage than the possibility of a greater one' (The Phrase Finder). Idiomatic
clauses and sentences like (58) are traditionally called proverbs. According to
Dobrovol'skij and Piirainen (2005: 51), the defining characteristic of a proverb
is that its meaning contains the semantic component [recommendation (that is
64 Chapter 3
supposed to be of a universal character) how to behave in a particular situation],

i.e. e.g. (58) recommends that we always choose a small real advantage rather
than the possibility of a greater one.
Both the fully-idiomatic VP kick the bucket and the fully-idiomatic proverb
A bird in the hand is worth two in the bush are similar to the words cat, untrue,
and waithood in that like the latter, the former are also stored in the mental
lexica of many English speakers. That is, those English speakers who know that
the underlined VP of (57) and Sentence (58) do not mean what their overt
components literally stand for can be said to have conventionalized associations
(stored in their mental lexica) between the signifiers kick the bucket / A bird in
the hand is worth two in the bush and the signifieds 'to die' / 'it's better to have a
small real advantage than the possibility of a greater one'. Accordingly, it makes
sense to regard idiomatic phrases and sentences as lexemes as well. (The term
'lexeme' should thus not be equated with the term 'word', even though the
overwhelming majority of lexemes are words.)
3.2 The structure of a lexeme
Like other signs, lexemes have four structural components:
1. the signifier
2. the signified
3. the syntactics
4. the sociolinguistics
For example, the lexeme CAT consists of the signifier cat, the signified 'cat', the
syntactics 'the signifier cat is a noun and thus can head NPs', and the
sociolinguistics 'cat is a stylistically neutral signifier and thus can occur in all
possible contexts'. Similarly, the fully-idiomatic lexeme KICK THE BUCKET
consists of the signifier kick the bucket, the signified 'to die', the syntactics 'kick
the bucket is a VP and thus can fill the predicate position', and the
sociolinguistics 'kick the bucket is a very informal way of expressing the concept
of dying'.
3.2.1 The lex of a lexeme
Just as the signifier of a morpheme is referred to as its morph, the signifier of a

lexeme can be referred to as its lex. For example, the signifier cat is the lex of
the lexeme CAT. The signifier untrue is the lex of the lexeme UNTRUE. The
signifier kick the bucket is the lex of the lexeme KICK THE BUCKET.
Analyzing English lexemes 65
As in the case of phonemes and morphemes, one and the same lexeme can
be realized by more than one lex. Compare, for example, the lexes kick the
bucket and die. Both of them express the meaning 'to die' and can therefore be
regarded as allolexes of the same lexeme. Since kick the bucket is an informal
way of expressing the meaning 'to die', whereas die is stylistically neutral, we
can say that the allolexes kick the bucket and die occur in stylistically relevant
free variation. Similarly, the lexes fever and pyrexia, which are associated with
the same signified 'abnormal elevation of body temperature', but differ with
regard to their sociolinguistics, can be considered allolexes of the same lexeme
which occur in stylistically relevant free variation.
In addition to free variation, allolexes realizing the same lexeme can also be
in complementary distribution. Recall the pair happy–happily, which we
discussed in 2.2.3. As we established, the adjective happy and the adverb
happily express the same lexical meaning but occur in different syntactic
environments: there can only be a happy life and His life was happy, but not *a
happily life and *His life was happily. Accordingly, happy and happily can be
regarded as allolexes of the same lexeme which occur in complementary
distribution.
A special type of complementary distribution of allolexes realizing the same
lexeme is represented by the distribution of wordforms of the same lexeme.
Wordforms are allolexes that express different grammatical meanings. For
example, book and books, pretty and prettier, worked of he worked and has
worked of he has worked, etc. Wordforms which realize the same lexeme
typically occur in identical syntactic environments: e.g. this book and these
books; She is pretty and She is prettier than his ex-girlfriend; He worked hard
and He has worked hard; etc. In other words, both the singular and the plural
wordform of a noun can head NPs; both the positive and the comparative
wordform of an adjective can function as complement of the auxiliary be; both
the simple past and the present perfect wordform of a verb can function as
predicator in a predicate VP; etc. However, despite these facts, this textbook
suggests that the distribution of wordforms of the same lexeme should be
regarded as an instance of complementary distribution rather than of free
variation. Thus, there can only be these schools and She is prettier than his ex-
girlfriend, but not *this schools and *She is pretty than his ex-girlfriend.
Similarly, the simple past and the present perfect wordform of a verb are usually
not interchangeable: both denote events that took place in the past, but the
present perfect wordform is used to refer to a past event that is somehow
relevant in the present: He has worked hard (in the past) and, because of this, he
is now the CEO of Microsoft. Accordingly, wordforms of the same lexeme that
have different grammatical meanings are usually not used in entirely identical
environments and can therefore be said to occur in complementary distribution.
66 Chapter 3
3.2.2 The typology of lexes
As we learned in Section 3.1, the lex of a lexeme can be represented by a word,

a phrase, and a clause / a sentence. As we already know, words that represent
lexes of lexemes can be classified into simple and complex words. The latter are
words like untrue, boyfriend, understand, receive, history, etc. which can be
segmented into at least two morphs or quasi-linguistic units. The former are
words like true, boy, friend, stand, etc. which consist of no more than one
morph.
Complex words can be further classified into compounds and affixed or
derived words. Compounds are complex words like boyfriend that contain at
least two roots. Affixed or derived words are complex words like untrue that
contain at least one root and one derivational affix.
3.2.3 The signified of a lexeme
From a semantic point of view, lexemes realized by complex lexes can be

classified into isomorphic and anisomorphic lexemes. Isomorphic lexemes are
realized by complex lexes like untrue and unable whose signifieds contain only
their components' signifieds. Anisomorphic lexemes are realized by complex
lexes like understand, blackboard, football, etc. whose signifieds are not (or not
entirely) representable in terms of their components' signifieds.
According to Mel'čuk (2001: 447-460), anisomorphic lexemes fall into three
categories:
1. full-idioms
2. semi-idioms
3. quasi-idioms
A full-idiom is a lexeme realized by a complex lex like understand whose

signified contains none of its components' signifieds. For example, the lexeme
UNDERSTAND is a full-idiom because the signified 'to understand' does not
contain the signifieds 'under' and 'to stand', inherent in the submorphs under-
and stand when they occur in other environments.
A semi-idiom (or a partial idiom) is a lexeme realized by a complex lex like
blackboard whose signified contains at least one of its components' signifieds.
For example, the lexeme BLACKBOARD is a semi-idiom because the signified
'blackboard' contains only the signified 'board', inherent in the morph board, but
not the signified 'black', inherent in the morfoid black.
Finally, a quasi-idiom is a lexeme realized by a complex lex like football
whose signified contains not only its overt components' signifieds but also some
additional unpredictable signified. For example, the lexeme FOOTBALL is a

quasi-idiom because the signified 'football' contains not only the signifieds 'foot'
and 'ball', inherent in the morphs foot and ball, but also the idiomatic signified 'a
particular sport different from handball, basketball, volleyball, etc.'.
3.2.4 Three-component anisomorphic lexemes
All examples of anisomorphic lexemes that have been dealt with so far are
lexemes realized by two-component complex lexes: UNDERSTAND and
BOYFRIEND are fully-idiomatic lexemes realized by two-component complex
lexes understand and boyfriend; BLACKBOARD and TWILIGHT are semi-idiomatic
lexemes realized by two-component complex lexes blackboard and twilight;
FOOTBALL, STEALER, WRITER, WAITHOOD are quasi-idiomatic lexemes realized
by two-component complex lexes football, stealer, writer, waithood. In the
following we will become acquainted with anisomorphic lexemes which are
realized by complex lexes consisting of more than two components.
Compare, for example, the signifieds associated with the lexes forget-me-
not, mother-in-law, and nationalism. The lexeme FORGET-ME-NOT is an obvious
full-idiom whose signified '[…] a plant which flourishes in damp or wet soil,
having bright blue flowers with a yellow eye' (OED) does not contain the
signifieds 'to forget', 'me', 'not', inherent in the morfoids forget, me, not when
they occur in other environments. The quasi-linguistic units forget, me, not can
be regarded as morfoids rather than submorphs because the literal meanings
inherent in these components seem to partially motivate the idiomatic meaning
'forget-me-not'. According to the OED,
in the 15th century the flower was supposed to have the virtue of
ensuring that those wearing it should never be forgotten by their
lovers. (OED)
Those speakers of English who do not remember this true etymology of forget-
me-not can easily invent their own popular etymologies establishing a
connection between the idiomatic meaning 'forget-me-not' and the literal
meanings inherent in the components forget, me, not. For example, the name
forget-me-not can be attributed to "the nauseous taste that it leaves in the mouth"
(OED).
The lexeme MOTHER-IN-LAW is a semi-idiom whose signified 'the mother of
one's spouse' (OED) contains the signified of the morph mother but not of the
morfoids in and law. As in the case of forget-me-not, the literal meanings of the
quasi-linguistic units in and law seem to partially motivate the idiomatic
meaning 'the mother of one's spouse': since sons-in-law are married to their
68 Chapter 3
mothers-in-law's daughters, there seems to exist a legal relation holding between

a husband and his wife's mother. Accordingly, in and law are morfoids, not
submorphs.
Finally, the lexeme NATIONALISM is a quasi-idiom whose signified does not
only contain the meanings of the root nation and the suffixes -al and -ism. If
nationalism were an isomorphic word, it would mean something like 'a national
ideology or an ideology relating to a particular nation'. Indeed, the suffix -ism
often denotes ideologies (e.g. socialism, capitalism) and the suffix -al carries the
meaning 'of, relating to, or characterized by' (MWO); e.g. cultural, suicidal.
However, nationalism is of course much more than just an ideology relating to a
particular nation. As defined by MWO, nationalism is
a sense of national consciousness exalting one nation above all others

and placing primary emphasis on promotion of its culture and interests
as opposed to those of other nations or supranational groups. (MWO)
The signified 'nationalism' thus contains the idiomatic meaning 'belief in the
supremacy of one's own nation and its national interests', which is not inherent
in either the root nation or the suffixes -al and -ism.
3.2.5 Anisomorphic lexemes realized by phrases and sentences
First of all, it must be noted that isomorphic phrases and sentences, whose
signifieds contain only their components' signifieds – e.g. the NP a black cat
meaning 'some black cat' and the clause I saw a black cat meaning 'I saw some
black cat' – are not stored in our mental lexica and thus cannot be regarded as
lexes realizing lexemes. The reason for the former is the finiteness of our mental
lexica: it is extremely uneconomical to store phrases and sentences whose
signifieds can be easily arrived at by adding their components' signifieds: e.g. 'a
black cat' = 'indefiniteness' + 'black' + 'cat' or 'I saw a black cat' = 'I' + 'saw' +
'indefiniteness' + 'black' + 'cat'.
By contrast, idiomatic phrases and (at least some) idiomatic sentences are
usually stored in our mental lexica. This means that, for example, a speaker of
English who knows that the VP kick the bucket means 'to die' and the proverb A
bird in the hand is worth two in the bush means 'it's better to have a small real
advantage than the possibility of a greater one' has a conventionalized
association between these signifiers and these signifieds stored in his or her
mental lexicon. Accordingly, as we concluded in Section 3.1, both the fully-
idiomatic VP kick the bucket and the fully-idiomatic proverb A bird in the hand
is worth two in the bush must be regarded as lexemes.
In addition to fully-idiomatic phrasal and sentential lexemes like those

discussed above, anisomorphic lexemes realized by phrasal and sentential lexes
can also have semi- and quasi-idiomatic signifieds. Compare, for example, the
meanings of the underlined VPs in (59) and (60).
(59) Answer the door,' she said and gazed back out at the grotto (COCA)
(60) Back in Greensboro, Duke married and started a family (COCA)
The VP answer the door of (59) is a semi-idiom whose signified contains the
signifieds of the components the and door, but not of the component answer:
answer the door means '[after hearing the doorbell or a knock] to go to the door
to see who is there' (The Free Dictionary). In other words, answering the door
involves going to a particular door (the one that was knocked at). Accordingly,
both the signifieds 'door' and 'definiteness' can be considered part of the
signified 'to answer the door'. By contrast, answering the door may not
necessarily involve answering to the person who knocked at a particular door or
rang its doorbell: opening the door without literally answering to that person
would also qualify as an act of answering the door. Accordingly, the meaning 'to
answer' is not part of the meaning 'to answer the door'.
The VP started a family of (60) is a quasi-idiom whose signified contains the
signifieds of its overt components start, -ed, a, and family – that is, the signifieds
'to start', 'the past tense', 'indefiniteness', and 'family' – but in addition to these
signifieds, it also contains the idiomatic signified 'to conceive the first child with
one's spouse (thereby starting to have a real family)' (Mel'čuk 1995: 184). In
other words, what Duke did back in Greensboro was marry someone and then
conceive his first child with that very person whom he married. In this way,
Duke started to have a real family, i.e. a family that includes a father, a mother,
and at least one child. Marrying someone without giving birth to at least one
child does not suffice to start a family.
Finally, compare the semantic structure of the idiomatic pick-up lines (61)
and (62), taken from http://linesthataregood.com/.
(61) Didn't I see you on the cover of Vogue?

(62) Would you like to have morning coffee with me?
A pick-up line is defined by the Urban Dictionary as "a conversation opener

with the intent of engaging an unfamiliar person for sex, romance, or dating".
Notice that pick-up lines are typically represented by idiomatic clauses and
sentences like (61) and (62). Indeed, (61) is a sentential semi-idiom whose
signified contains the meanings of the components I, you, on, the, cover, of,
Vogue but not the meanings of the components did, not, and see. That is, the
speaker of (61) does not really request the addressee to confirm that the former
70 Chapter 3
did indeed see the latter on the cover of Vogue but simply expresses a
compliment: 'You are so beautiful. I think you deserve to be on the cover of
Vogue'. By contrast, (62) is not a semi- but a quasi-idiom that does not mean
'Would you like to have morning coffee with me?' but 'Would you like to go to
bed with me and afterwards have morning coffee with me?'. The signified of
(62) thus contains the additional idiomatic meaning 'invitation to go to bed with
the speaker of (62)', which is inherent in none of (62)'s overt components.
To conclude, those speakers of English who actively use (61) and (62) with
the intent of initiating a sexual relationship with an unfamiliar person can be
said to have conventionalized correspondences between (61) and (62) and the
just named semi- / quasi-idiomatic signifieds (which can be expressed by them)
stored in their mental lexica. Accordingly, sentences (61) and (62) can be
regarded as sentential lexes realizing corresponding anisomorphic lexemes.
3.2.6 How to distinguish between full-, semi-, and quasi-idioms?
Given what we have learned in the previous parts of this section, this may seem
a superfluous question. As we said in 3.2.3, a full-idiom is an isomorphic
lexeme realized by a lex whose signified does not contain either of its
components' signifieds. A semi-idiom is an anisomorphic lexeme realized by a
lex whose signified contains one of its components' signifieds. Finally, a quasi-
idiom is an anisomorphic lexeme realized by a lex whose signified contains not
only its overt components' signifieds but also some additional, unpredictable
signifieds. These definitions can be used as guidelines for deciding whether the
lexeme under analysis has a full-, semi-, or quasi-idiomatic signified.
But let us again consider the fully-idiomatic lexeme BOYFRIEND. As was
pointed out in 1.2.2, the meaning 'friend', inherent in the morfoid friend,
partially motivates the idiomatic meaning 'boyfriend': boyfriends are usually
perceived (by their girlfriends) as friends, i.e. as people whom they know well
and regard with affection and trust. Given this fact, why can we not analyze
BOYFRIEND as a semi-idiomatic lexeme whose signified contains the signified of
the component friend but not of the component boy? More generally, how can
we distinguish between components' signifieds which are part of lexemes'
idiomatic signifieds and components' signifieds which only motivate lexemes'
idiomatic signifieds, without, however, being part of those signifieds?
To answer this question, let us recall what we said about blackboards in
1.2.2. Blackboards typically have dark surfaces (and are in this respect different
from whiteboards), but they are not always black: green blackboards, for
instance, do occur as well. In other words, the lex blackboard can be used to
refer to a non-black blackboard. This fact justifies the treatment of the lexeme
BLACKBOARD as a semi-idiom whose signified contains the signified of the
component board but not of the component black. Hence of these two
components, only board can be regarded as a morph realizing a morpheme,
whereas black is a morfoid whose signified only motivates the idiomatic
signified 'blackboard', without, at the same time, being part of that idiomatic
signified.
In a similar way, we can prove that neither the signified 'boy', inherent in the
morfoid boy, nor the signified 'friend', inherent in the morfoid friend, are part of
the idiomatic signified 'boyfriend'. As regards the signified 'boy', consider the
use of boyfriend in (63).
(63) I am 23 and my boyfriend is 40? (http://tinyurl.com/5tc222w)
As (63) demonstrates, the lex boyfriend can be used to refer to a 40-year-old

man, who obviously no longer qualifies as a young male child (which the
signifier boy literally means). As stated earlier, a boyfriend is a male of almost
any age over puberty. Accordingly, we are justified in concluding that the
signified 'boy', inherent in the morfoid boy, is not part of the idiomatic signified
'boyfriend'.
Similarly, as (64) demonstrates, there are boyfriends who are not friends.
(64) my boyfriend is not a friend at all. he has done me so wrong i cant even
count […] all the mistakes he has made (http://tinyurl.com/6zxfmmj)
That is, a male person with whom a woman has romantic or sexual relations
(and whom she regards as her boyfriend) may not be regarded by her as a friend
in the literal meaning of this word. Nevertheless, she will use the term boyfriend
as long as the man in question remains her regular sexual / romantic partner in a
non-marital sexual relationship. This is because the signified 'friend', inherent in
the morfoid friend, is also not part of the idiomatic signified 'boyfriend'. Hence
we can conclude that the lexeme under analysis is indeed a fully-idiomatic
lexeme whose signified does not contain the signifieds 'boy' and 'friend',
inherent in the morfoids boy and friend.
In summary, in order to distinguish between components' signifieds which
are part of lexemes' idiomatic signifieds and components' signifieds which only
motivate lexemes' idiomatic signifieds, without, however, being part of those
signifieds, we need to establish whether the lex under analysis can be used to
refer to a person or a thing that does not possess the characteristics denoted by
its components. If we discover that, for example, there are blackboards that are
not black and boyfriends who are neither boys nor friends, we will be justified in
claiming that the signified of the lexeme under analysis does not contain one or
both of its components' signifieds.
72 Chapter 3
3.3 Lexemes and vocables
As we established in 2.2.1, one and the same signifier can be associated with
more than one signified. Recall that the signifier true can mean 'faithful' (e.g.
Her lover had been true) and 'not false' (e.g. Indicate whether each of the
following statements is true or false). Accordingly, we must distinguish between
two different TRUE lexemes: TRUE1 'faithful' and TRUE2 'not false'. At the same
time recall that these two signifieds inherent in true are polysemes: as we said,
being not false can be explained as being faithful to the truth, so that the
signified 'not false' can be regarded as a product of semantic narrowing of the
signified 'faithful'. Given this fact, the two different TRUE lexemes can be united
into the same vocable (Mel'čuk and Zholkovsky 1988: 27).
A vocable is a set of polysemous lexemes. For example, the vocable TRUE is
the set of polysemous lexemes like TRUE1 'faithful' and TRUE2 'not false'. In
addition to these signifieds, the signifier true can also mean 'properly so called'
(e.g. true love), 'legitimate, rightful' (e.g. our true and lawful king), 'narrow,
strict' (e.g. in the truest sense), etc. (MWO). Along with TRUE1 and TRUE2, these
polysemous lexemes can be said to constitute the vocable TRUE.
Uniting polysemous lexemes into vocables is a usual lexicographic practice.
Thus a dictionary entry for true (i.e. what you find in a dictionary like the OED
or MWO when looking up true) is typically not just one signified 'faithful' but a
set of polysemous signifiers like 'faithful', 'not false', 'properly so called',
'narrow, strict', etc.
Unlike polysemous lexemes, homonymous lexemes such as, for example,
CASE 'a set of circumstances or conditions' and CASE 'a box or receptacle for
holding something' do not form vocables or other similar units. This is because
the signifieds of homonymous lexemes do not have much in common. That is, if
the signifieds 'a set of circumstances or conditions' and 'a box or receptacle for
holding something' have nothing in common, what is the point of uniting these
signifieds into vocables or units alike?
3.3.1 Relations between members of the same vocable
As illustrated by the lexemes TRUE1 'faithful' and TRUE2 'not false', two lexemes
that form the same vocable often exhibit a quasi-idiomatic relation, i.e. the
signified of one of the two lexemes can be analyzed as a quasi-idiom in relation
to the signified of the other lexeme. Thus the signified of the lexeme TRUE2 can
be analyzed as a quasi-idiom in relation to the signified of the lexeme TRUE1 :
while TRUE1 is associated with the signified 'faithful', TRUE2 is associated with the
signified 'faithful to the truth'. Compare also the meanings of had in (65) and
(66).
(65) I always had a good time when Crystal and I got together (COCA)
(66) He has not had many women, has drifted in solitude for much of his life
[…] (COCA)
In (65), had carries the meaning 'to experience something' (Ganshina and
Vasilevskaya 1964: 126): to have a good time means 'to experience a good time'.
The 'experience'-sense of have is historically posterior to the lexeme HAVE
meaning 'to possess' (which perhaps can be regarded as a more basic meaning of
have): according to the OED, the 'experience'-have has been known since circa
1000, whereas the 'possess'-have has been documented since circa 888.
However, in Present-day English the 'experience'-uses of have as those in (65)
are not perceived as idiomatic. As an illustration, consider the meanings of have
in (67).
(67) You work to have the American dream, to have a nice house, a car, two
dogs and a cat (COCA)
Some native speakers of English fail to recognize that the have of have the
American dream is semantically different from the have of have a nice house.
While the latter have means 'to possess', the former have means 'to experience':
to have a nice house means 'to possess a nice house', but to have the American
dream means 'to experience the American dream'. That this difference is,
however, a hardly recognizable difference corroborates the analysis of the
'experience'-sense of have as one of the literal meanings of the signifier have.
What follows from this is that one and the same vocable can have more than one
literal lexeme. That is, for example, the vocable HAVE can be said to consist of
at least two literal lexemes: HAVE1 'to possess' and HAVE2 'to experience'.
In contrast to the literal haves of (65) and (67), the had of (66) is a lex
realizing an idiomatic HAVE lexeme. The had of (66) carries the quasi-idiomatic
meaning 'to experience sexual relations with somebody': He has not had many
women […] means 'he has not experienced sexual relations with many women'.
That is, the meaning of the had of (66) contains the literal meaning 'to
experience' plus the idiomatic meaning 'sexual relations'. Accordingly, we can
conclude that within the vocable HAVE, there is the idiomatic lexeme HAVE3,
whose signified 'to experience sexual relations' represents a quasi-idiom in
relation to the signified of one of the literal lexemes HAVE2 'to experience'.
Two polysemous lexemes that form the same vocable can also exhibit a
fully- or a semi-idiomatic relation. As an illustration of the former, compare the
meanings of mouse in (68) and (69).
(68) He taught me probably as much as anybody I coached because it was a

struggle […] and we got along like cat and mouse (COCA)
74 Chapter 3
(69) All computers have similar attributes such as a mouse, CD drive, key
placements, visual characteristics, and operating systems (COCA)
In (68), mouse carries its literal meaning 'any of numerous small rodents […]
with pointed snout, rather small ears, elongated body, and slender tail' (MWO).
By contrast, the mouse of (69) denotes a computer mouse, i.e. a device that
'controls movement of the cursor and selection of functions on a computer
display' (MWO). It is evident that the semantic relation between the two MOUSE
lexemes is different from that holding between the lexemes TRUE1 'faithful' and
TRUE2 'not false' and between the lexemes HAVE2 'to experience' and HAVE3 'to
experience sexual relations with somebody'. The lexeme MOUSE2 'a computer
device' can be analyzed as a full-idiom in relation to the lexeme MOUSE1 'a small
rodent': the signified 'a computer device' does not contain the signified 'a small
rodent'.
As an illustration of a semi-idiomatic relation holding between two
polysemous lexemes, compare the meanings of the verb to better in (70) and
(71).
(70) They are trying to better the lives of working people (MWO)
(71) His financial situation has bettered (COCA)
While the better of (70) means 'to make better', the bettered of (71) expresses
the meaning 'to become better'. Accordingly, we can argue that the two TO
BETTER lexemes are in a semi-idiomatic relation to each other. Both of their
signifieds contain the signified 'better' but differ with regard to the verbal
signified preceding the signified 'better'. As just said, while the better of (70)
means 'to make better', the bettered of (71) means 'to become better'.
3.4 Lexemes and lexeme families
Finally, we will introduce another important concept in morphology, the concept

of a lexeme family. According to Haspelmath and Sims (2010: 17), lexeme
families are formed by lexemes whose lexes have the same root but different
derivational affixes. For example, all lexemes realized by the lexes true, untrue,
truth, truthful, etc. (i.e. those which share the root true) can be said to form a
lexeme family. In a similar way, we can say that a lexeme family is formed by
all lexemes realized by the lexes history, historic, historical, historically,
historicalness, historian, etc., i.e. lexes which all share the bound root histor-.
Lexeme families are similar to vocables in that both consist of semantically
and formally related lexemes. However, lexemes that form a vocable always
share the same signifier, whereas members of the same lexeme family have
different signifiers: they share the same root but have different derivational
affixes.
3.5 Exercises
a) A proverb is usually an idiomatic sentence that contains a recommendation

as to how one should behave in a particular situation.
b) All linguistic signs can be regarded as lexemes.
c) Allolexes of the same lexeme are always in contrastive distribution.
d) Wordforms are allolexes of the same lexeme which differ with regard to their
grammatical meanings.
e) The mental lexicon of the average speaker of English is infinite.
f) A complex word consisting of two roots is an affixed word.
g) A lexeme realized by a complex lex whose signified contains only its overt
components' signifieds is an isomorphic lexeme.
h) A quasi-idiom is a lexeme whose signified does not contain its components'
signifieds.
i) A vocable cannot consist of more than one literal lexeme.
j) Two different lexemes whose signifiers share the same root but have
different derivational affixes form a lexeme family.
3. Analyze the distribution of the following lexes.
a) cat and dog

b) they and their
c) painful and painfully
d) United States of America and U.S.A.
e) book and books
f) true of Her lover had been true and true of in the truest sense
g) sad and sadly
h) true of Her lover had been true and faithful of Her lover had been faithful
i) laboratory and lab
j) I and my
4. State whether the members of the following pairs of lexes are different
wordforms of the same lexeme, lexes realizing different lexemes of the same
76 Chapter 3
vocable, lexes realizing different lexemes of the same lexeme family, or

lexes realizing different lexemes that do not form either lexemes or vocables.
a) table and sample

b) part and partial
c) bank 'financial institution' and bank 'the ground near the river'
d) teacher and teacher's
e) bachelor 'lowest university degree' and bachelor 'an unmarried man'
f) response and irresponsibility
g) a wife and to wife
h) to wait and waithood
i) is doing and has done
j) professor and university
5. Classify the following words into both the formal and semantic categories
which were introduced in this chapter.
a) greenhouse
b) table
c) barman
d) foretell
e) skinhead
f) diary
g) browser
h) after-party
i) redo
j) birdbrain
3.6 Further reading
Terms like 'lexeme', 'mental lexicon', 'complex word', 'compound', etc. are
standard morphological terms that are introduced in any textbook on
morphology. See, for instance, Lieber (2010: Chs. 1 and 2), Haspelmath and
Sims (2010: Ch. 2), Katamba and Stonham (2006: Chs. 2 and 3).
The semantic classification of lexemes into full-idioms, semi-idioms, and
quasi-idioms that was proposed in this chapter was largely based on the article
Mel'čuk (1995) and the already mentioned book Mel'čuk (2001: Ch. 9), whose
French original is Mel'čuk (1997: Ch. 9).
Idiomatic VPs like kick the bucket, answer the door, start a family, etc. have
traditionally been the focus of attention of phraseology. For an overview of the
most important phraseological issues, see Burger et al. (2007), Dobrovol'skij and
Piirainen (2005; 2009) and references therein.
Idiomatic sentences like the two pick-up lines discussed in 3.2.5 have
traditionally been dealt with in pragmatics, where they are regarded not as semi-
idioms or quasi-idioms but as indirect speech acts, in which 'what is said' is
different from 'what is meant'. A classic article dealing with this topic is Searle
(1975). For an overview of the most important pragmatics-related issues, see
Horn and Ward (2004).
4 Word-formation: basic issues
In Chapter 3 we classified English lexemes into a number of semantic and

formal categories. For example, we said that UNTRUE is an isomorphic lexeme
whose signified contains only the signifieds of the root true and the affix un-,
whereas FOOTBALL is a quasi-idiomatic lexeme whose signified contains not
only the signifieds of its roots foot and ball but also the idiomatic signified 'a
particular sport different from other sports'. Proceeding from this analysis, we
can conclude that the lexeme UNTRUE represents a product of isomorphic
affixation of the root true by means of the prefix un-, whereas the lexeme
FOOTBALL came into existence via the compounding of the roots foot and ball
accompanied by the quasi-idiomatization of the meanings 'foot' and 'ball'.
Beginning with this chapter, we will be concerned with the nature of processes
like affixation and compounding with the help of which speakers of English
create new words, thereby enlarging the vocabulary of their language. The first
section of this chapter proposes a classification of word-formation into lexeme-
formation and lex-formation. Section 4.2 dwells on different lexeme-formation
mechanisms as well as on such issues as differences between diachronic and
synchronic approaches to lexeme-formation, institutionalization of newly
formed lexemes, productivity of various lexeme-formation mechanisms, etc.
Finally, Section 4.3 discusses lex-formation mechanisms, which give rise to
additional allolexes of already existing lexemes rather than to lexes realizing
new lexemes.
4.1 Lexeme-formation versus lex-formation
Word-formation (understood as a formation of a new combination of sounds

capable of independent use in accordance with the isolatability criterion) can be
classified into lexeme-formation and lex-formation. The former gives rise to
output lexes which realize new lexemes; the latter produces allolexes of already
existing lexemes.
As an illustration of this difference, let us compare the lexes untrue and
happily. From a formal point of view, both can be regarded as instances of
affixation: the output lex untrue is a product of affixation of the input lex true
by means of the prefix un- and the output lex happily is a product of affixation
of the input lex happy by means of the suffix -ly. However, while the output lex
untrue realizes a different lexeme than the corresponding input lex true – the
signified 'untrue' is the reverse of the signified 'true' – the output lex happily is
an allolex of the input lex happy: as we argued in 2.2.3, happy and happily differ
80 Chapter 4
with regard to their syntactic functioning but express essentially the same
meaning. Accordingly, the formation of untrue is an instance of lexeme-
formation, whereas the formation of happily is an instance of lex-formation.
Apart from classifying word-formation into lexeme-formation and lex-
formation, we are now also in a position to conclude that the typology of affixes
which was presented in 2.5.4 was not complete. Affixes fall not only into
derivational and inflectional affixes. In addition to these two traditionally
recognized categories, there are also lex-forming or lex-building affixes like -ly
of happily. Lex-forming affixes are similar to inflectional affixes in that both
form allolexes of already existing lexemes. However, while the addition of an
inflectional affix gives rise to an output allolex that has a different grammatical
meaning than a corresponding input lex (cf. e.g. book and books), the addition of
a lex-forming affix never changes the grammatical meaning of a corresponding
input lex. For example, as Pullum and Huddleston (2002: 532) observe, both
adjectives and adverbs can be marked with regard to the grammatical category
DEGREES OF COMPARISON or simply GRADE. This means that there can be the
grammatical contrast between both the adjectival positive and comparative
wordforms happy and happier and the adverbial positive and comparative
wordforms happily and more happily. Accordingly, we are justified in
concluding that the output adverbial lex happily does not differ from its input
adjectival lex happy with regard to the grammatical category DEGREES OF
COMPARISON : both express the grammatical meaning 'positive degree of
comparison'. This justifies our analysis of the suffix -ly as a lex-forming suffix.
4.2 Lexeme-formation
In this section, we will be concerned with the most important issues pertaining
to lexeme-formation. We will begin with a classification of lexeme-building
mechanisms into the following three categories:
1. purely semantic mechanisms

2. purely formal mechanisms
3. mechanisms that involve both semantic and formal modifications
4.2.1 Purely semantic mechanisms
Purely semantic mechanisms are those that involve only semantic modifications
of the signified of an already existing input lexeme. To put it in a simpler way:
while the signifier remains the same, the signified undergoes semantic change.
For example, the lexeme HAVE3 'to experience sexual relations' (e.g. He did not
Word-formation: basic issues 81
have many women) can be said to have come into existence via quasi-
idiomatization of the lexeme HAVE2 'to experience' (e.g. He did not have a good
time). Similarly, the lexeme MOUSE2 'a computer device' can be said to have
come into existence via full-idiomatization of the lexeme MOUSE1 'a small
rodent'.
An instance of semantic change that has received a lot of attention in English
theoretical linguistics is so-called morphological conversion. Recall the verb to
wife 'to downplay a woman's career accomplishments in favor of her abilities as
wife and mother', which we discussed in 1.1. Like HAVE3 'to experience sexual
relations', TO WIFE came into existence via quasi-idiomatization of the lexeme A
WIFE: the signified 'to wife' contains the signified 'a wife', inherent in the
component wife, plus the additional idiomatic meaning 'to downplay a woman's
career accomplishments in favor of her abilities as mother'. However, in contrast
to HAVE3, the output lexeme TO WIFE is realized by a lex which is a member of a
different word class than the lex which realizes the input lexeme A WIFE: while a
wife is a noun, to wife is a verb. Similarly, the verb to Thomas 'to accuse a
person of sexual harassment' (Word Spy) came into existence via full-
idiomatization of the proper noun Clarence Thomas: according to Word Spy,
this converted verb "originates from the sexual harassment accusations aimed at
Clarence Thomas during his Supreme Court confirmation hearings". As in the
case of to wife, the semantic change undergone by the input lexeme THOMAS
gave rise to the output lexeme TO THOMAS, whose lex is a member of a different
word class than the lex of the corresponding input lexeme: while Thomas is a
proper noun, to Thomas is a verb.
4.2.2 Purely formal mechanisms
Purely formal mechanisms include:
1. lexeme manufacturing
2. isomorphic borrowing
3. isomorphic affixation
4. apophony
Lexeme-manufacturing can be described as an arbitrary formation of a

previously non-existent signifier for expressing a previously unexpressed
signified. According to Ferdinand de Saussure (1973: 100-103), arbitrariness is
one of the major characteristics of the signifier–signified relation exhibited by
linguistic signs. That is, in a number of cases there is no special reason why a
particular signified is expressed by a particular signifier. For example, there is
no special reason why the signified 'table' is expressed in English by the signifier
82 Chapter 4
table, but not by e.g. the signifier Tisch, which expresses the same signified in
German, or by the signifier стол / stol, which expresses the same signified in
Russian. Arbitrary formation is thus the creation of an unmotivated signifier like
table that does not provide any clue as to why that very signifier was chosen by
its creator for expressing a particular signified. Consider, for example, the
famous proprietary name Viagra 'the drug sildenafil citrate, given orally in the
treatment of male impotence' (OED). According to the OED, Viagra is
probably an arbitrary formation. The first element vi- is perhaps

influenced by English words such as virile […], virility […], etc., or
their etymon classical Latin vir […]. According to the makers of the
drug, the last two syllables (and first vowel) were not suggested by
Niagara […], as sometimes proposed. A phonetic similarity to
Sanskrit vy-agra 'excited, agitated' and vyāghra 'tiger' has also been
noted, but these words seem unlikely to have influenced the
formation. (OED)
A similar example of an arbitrary formation is Gonk 'the proprietary name of an

egg-shaped doll' (OED). While vi- of Viagra could have been influenced by vi-
of virile and virility, Gonk seems to be an entirely unmotivated signifier that
does not provide any clue as to why it was created as the proprietary name of an
egg-shaped doll.
Isomorphic borrowing is the importation of a foreign language lexeme
which is not accompanied by a semantic modification of the signified of that
lexeme (which is characteristic of anisomorphic borrowing). For example,
speakers of English borrowed the lexeme KINDERGARTEN from German without
changing its signified: at the moment of borrowing (1852), the German lexeme
had a semi-idiomatic signified 'a school for the instruction of young children
according to a method devised by Friedrich Froebel […]' (OED). This is the
signified which the borrowed German lex kindergarten came to be associated
with in English. Similarly, the German Sprachgefühl, which has been recorded
in English since 1902, came to be associated with essentially the same quasi-
idiomatic signified 'the intuitive feeling of a speaker for the essential character
of a language; linguistic instinct' (OED), which this signifier expressed in
German.
Isomorphic affixation gives rise to new lexemes like UNTRUE and UNABLE
whose signifieds are fully-representable in terms of their components' signifieds.
That is, 'untrue' = 'not' + 'true' and 'unable' = 'not' + 'able'. More recent instances
of isomorphic affixation by means of the derivational prefix un- are unsackable
'not sackable' (OED) and uncreolized 'of a language or dialect: not creolized;
that has not undergone creolization' (OED). According to the OED, both have
been used in English since 1980.
Finally, apophony can be defined as any isomorphic modification of the lex

of an input lexeme that cannot be regarded as an instance of isomorphic
affixation. Consider, for instance, the pair to increase–an increase. At first
glance, it may seem that AN INCREASE is a product of quasi-idiomatization of TO
INCREASE : the signified of the former lexeme can be described as 'the action of
increasing' (OED). However, in contrast to the members of the conversion pairs
a wife–to wife and Thomas–to Thomas, the signifiers to increase and an increase
are not identical. While the verb /ɪnˈkriːs/ is stressed on the last syllable, the
noun /ˈɪnkriːs/ is stressed on the first syllable. Accordingly, the difference in
meaning between these two lexemes can be attributed to the shifting of the
stress: /ɪnˈkriːs/ Æ /ˈɪnkriːs/. This formal modification cannot qualify as an
instance of affixation because stress is not an affix but a non-segmental
property of words, i.e. one that cannot be represented in terms of sounds.
Consequently, /ɪnˈkriːs/ Æ /ˈɪnkriːs/ can be regarded as an instance of apophony.
A more recent example is buildering 'the practice of climbing up large, urban
buildings […] as a recreational activity' (OED), which differs from its input lex
bouldering 'practice climbing on large boulders' (OED) only with regard to its
first vowel: cf. /ˈbɪld(ə)rɪŋ/ and /ˈbəʊldərɪŋ/. As in the case of to increase Æ an
increase, the vowel change [əʊ] Æ [ɪ] exemplified by the pair bouldering–
buildering does not qualify as an instance of affixation: neither [əʊ] of
bouldering nor [ɪ] of buildering can be regarded as either morphs or quasi-
linguistic units of the lexes under analysis. Hence bouldering Æ buildering can
also be seen as an instance of apophony.
4.2.3 Mechanisms involving formal and semantic modifications
These mechanisms include:
1. compounding
2. blending
3. idiomatization of phrases and sentences
4. anisomorphic affixation
5. back-formation
6. anisomorphic borrowing
Compounding is an anisomorphic lexeme-building mechanism that produces

compound lexemes (i.e. lexemes whose lexes are segmentable into at least two
roots) whose signifieds are not (or not entirely) representable in terms of their
components' signifieds. That is, those compounds that are neither semi- nor full-
idioms can be analyzed as quasi-idioms, whose signifieds contain some
additional, unpredictable signifieds, which are not inherent in their roots'
84 Chapter 4
signifieds. Consider, for example, the signifieds of the compounds piano-tuner,

brake cable, and spring festival. As argued by Haspelmath and Sims (2010:
191), these are non-idiomatic compounds whose signifieds are made up of their
components' signifieds. However, spring festival does not mean 'any festival that
has something to do with spring' but 'a festival that takes place in spring'. Given
the components spring and festival, it could have acquired a meaning like 'a
festival devoted to the celebration of spring that takes place in winter or any
other season of the year'. The compound spring festival is thus an obvious quasi-
idiom whose signified does not only contain the signifieds 'spring' and 'festival',
inherent in the roots spring and festival, but also the idiomatic meaning 'takes
place in'. Similarly, brake cable is a quasi-idiom whose signified does not only
contain the signifieds 'brake' and 'cable', inherent in the roots brake and cable,
but also the idiomatic meaning 'a component part of': a brake cable is not any
cable that has something to do with a brake but a cable that is a component part
of a brake. Finally, a piano-tuner is not a person who tunes some piano but a
person who tunes (more than one) particular pianos: those that other people ask
him or her to tune. Again, we see that the signified 'piano-tuner' is not entirely
representable in terms of the signifieds 'piano', 'tune', and 'performer of some
action', inherent in the roots piano and tune and the derivational suffix -er.
Blending is an anisomorphic lexeme-building mechanism that produces
blended lexemes, i.e. lexemes whose lexes contain the shortened lexes of (at
least some of) their input lexemes. For example, Brangelina Å Brad Pitt and
Angelina Jolie, Chermany Å China and Germany, etc. From a semantic point of
view, the signified of a blend is, like the signified of a compound, never fully-
representable in terms of its components' signifieds. Thus Brangelina does not
mean 'Brad Pitt' + 'Angelina Jolie' but 'a celebrity supercouple consisting of
Brad Pitt and Angelina Jolie' (Wikipedia). Similarly, Chermany does not mean
'China' + 'Germany' but 'China and Germany taken together, particularly as an
economic entity or market' (Word Spy). Both the blends Brangelina and
Chermany are thus quasi-idioms whose signifieds contain additional signifieds
'supercouple' / 'economic entity', which are not inherent in the signifieds of these
blends' shortened components.
Idiomatization of phrases and sentences is an anisomorphic lexeme-
building mechanism that produces lexemes which are realized by phrasal and
sentential lexes like those discussed in 3.2.5. For example, the semi-idiomatic
lexeme TO ANSWER THE DOOR is realized by the VP answer the door; the fully-
idiomatic lexeme A BIRD IN THE HAND IS WORTH TWO IN THE BUSH is realized by
the clause A bird in the hand is worth two in the bush; etc.
Anisomorphic affixation is the reverse of isomorphic affixation, which we
discussed in the previous part of this section. Consider, for example, the lexeme
UNGOOGLEABLE 'a person for whom no information appears in an Internet search
engine, particularly Google' (Word Spy). In contrast to the signifieds of the
isomorphic lexemes UNTRUE and UNABLE, which came into existence via
affixation of the input lexes true and able by means of the prefix un-, the
signified 'ungoogleable ' does not only contain the signifieds 'not', 'Google', and
'able', which are inherent in the components un-, Google, and able, but also the
idiomatic signified 'information about a person that is not available on the
Internet'. A similar example is endism 'the belief that something of significant
scope and duration, particularly something negative, is coming to an end' (Word
Spy). Like the signified 'ungoogleable', the signified 'endism' does not only
contain the signifieds 'end' and 'belief, ideology', inherent in the root end and the
suffix -ism, but also the idiomatic meaning 'something of significant scope and
duration coming to'. That is, endism is not the belief in the end, but the belief
that something of significant scope and duration will come to an end. Both
ungoogleable and endism are thus quasi-idioms, whose signifieds are not
entirely representable in terms of their components' signifieds.
Back-formation is the removal of a derivational affix (or a part of an input
lex which is perceived as a derivational affix) from the lex of an input lexeme.
For instance, the output verb to tase 'to use a Taser on (a person); to subdue or
incapacitate using a Taser' (OED / 1991, i.e. the date of the earliest citation
documented by the dictionary) came into existence via the removal of -er from
the input noun Taser 'a weapon which fires barbs attached by wires to batteries,
and causes temporary paralysis' (OED / 1972). Similarly, the output noun skeeve
'an obnoxious or contemptible person; a person regarded as disgusting,
unpleasant, etc.' (OED / 1987) came into existence via the removal of -y from
the input adjective skeevy 'disgusting, distasteful, or dirty; discomforting; sleazy'
(OED / 1976). As regards the semantic perspective, observe that the output
signifieds 'to tase' and 'skeeve' can be analyzed as quasi-idioms in relation to
their input signifieds 'Taser' and 'skeevy': to tase means 'to use a Taser on
somebody' and skeeve means 'a skeevy person'. Accordingly, we are justified in
regarding back-formation as an anisomorphic lexeme-building mechanism.
Finally, anisomorphic borrowing is the importation of the signifier of a
foreign language lexeme accompanied by a semantic modification of its
signified. Consider, for instance, the lexeme WIKI 'a type of web page designed
so that its content can be edited by anyone who accesses it, using a simplified
markup language' (OED / 1995). According to the OED, the lex wiki was
borrowed into English from Hawaiian. However, whereas the input Hawaiian
lexeme is associated with the signified 'quick', the corresponding output lexeme
in English came to be associated with the signified 'a type of Web page'.
Accordingly, we can conclude that the borrowing of the lex wiki from Hawaiian
into English was accompanied by the full-idiomatization of its input signified
'quick'.
86 Chapter 4
4.2.4 Diachronic and synchronic perspectives
The study of lexeme-formation can be approached from two different theoretical

perspectives:
1. lexeme-formation as a diachronic history of a particular lexeme

2. lexeme-formation as a synchronic relation holding between two lexemes
As an illustration of the difference between the two approaches, let us again

consider the word boyfriend. Thanks to the OED (the most important tool for the
diachronic study of lexeme-formation in English), we know that the fully-
idiomatic lexeme BOYFRIEND2 'a male of almost any age over puberty with
whom a woman has a non-marital sexual relationship' is not a product of
anisomorphic compounding of the roots boy and friend but a product of full-
idiomatization of the lexeme BOYFRIEND1 'a friend who is a boy, a boyhood
friend' (OED). According to the OED, the quasi-idiomatic BOYFRIEND1 has been
used in English since 1822, whereas the fully-idiomatic BOYFRIEND2 appeared
only in 1906. This is the diachronic history of BOYFRIEND2, which allows us to
conclude that the lexeme under analysis is a product of a purely semantic
lexeme-building mechanism, not compounding.
However, from a synchronic perspective, a (slightly) different analysis is
called for. According to the OED, the original lexeme BOYFRIEND1 does not
occur very often in Present-day English. (In this respect, BOYFRIEND1 is different
from GIRLFRIEND1 'a female friend; especially a woman's close female friend'
(OED), which still coexists with GIRLFRIEND2 'a female of almost any age over
puberty with whom a person has a non-marital sexual relationship'.) This means
that for the majority of English speakers, the vocable BOYFRIEND consists of
only one fully-idiomatic lexeme BOYFRIEND 'a male of almost any age over
puberty with whom a woman has a non-marital sexual relationship', whose lex is
easily segmentable into the components boy and friend capable of occurring in
isolation. Accordingly, laymen who do not know the diachronic history of this
lexeme can 'falsely' assume that BOYFRIEND is a product of anisomorphic
compounding of the roots boy and friend.
Another interesting case is represented by the pair king / queen–royal.
Semantically, the signified 'royal' can be analyzed as a quasi-idiom in relation to
the signifieds 'king' and 'queen': according to the OED, the lex royal is
associated with the signified 'befitting or appropriate to a monarch, especially in
quality, size, or ostentation'. That is, the signified 'royal' contains the signifieds
'king' / 'queen' plus the signified 'befitting to'. However, despite this obvious
semantic connection, the signifiers king / queen and royal do not seem to have
much in common. Cf. /kɪŋ/ and /ˈrɔɪəl/. The reason for this is that king and queen
are lexes of Germanic origin, which have been used in English since the Old
English period, whereas royal was borrowed into English from Middle French
around 1400 (OED). This is the diachronic history of the lexeme ROYAL, which
accounts for the fact that it is realized by a lex which has virtually nothing in
common with the lexes of the semantically related lexemes KING and QUEEN.
However, as in the case of BOYFRIEND2, we can imagine that a considerable
number of laymen no longer remember that royal is a lex of French origin which
was borrowed into English in 1400. For them the output lexeme ROYAL is a
product of suppletion of the input lexemes KING and QUEEN. Suppletion can be
defined as the use of a formally unrelated signifier (i.e. one which has a different
root) for expressing a related signified. As Apresjan (1974: 172) points out,
Present-day English abounds in suppletive pairs: in addition to king / queen–
royal, there are also the pairs town–urban, law–legal, father–paternal, noun–
nominal, tense–temporal, etc. Suppletive pairs can also often be found among
geographical terms. For example, an inhabitant of Manchester is usually called a
Mancunian and the language of the Netherlands is usually called Dutch (even
though the affixed lexes Manchesterian and Netherlandish do exist as well). As
in the case of king / queen–royal, these suppletive pairs came into existence
because the lexes which realize corresponding adjectival lexemes were
borrowed from other languages: urban, legal, nominal, temporal, and
Mancunian were borrowed from Latin, paternal was borrowed from French, and
Dutch was borrowed from Middle Dutch (OED).
Finally, let us consider the fully-idiomatic verb babysit. From a formal point
of view, this verb can be segmented into the morfoids baby and sit, whose
signifieds 'baby' and 'to sit' partially motivate the signified 'to babysit', without,
at the same time, being part of that signified. Babysit does not mean 'to sit with
or near someone's baby' but 'to care for children usually during a short absence
of the parents' (MWO): we can imagine a hyperactive babysitter who never sits
while doing the job of a babysitter. Evidently, this person will nevertheless
qualify as a babysitter as long as he or she takes care of the baby whom he or
she is supposed to babysit: it does not really matter whether a babysitter sits or
stands with or near someone's baby while doing babysitting. Accordingly, we
can conclude that the signified 'to sit', inherent in the morfoid sit, is not part of
the idiomatic signified 'to babysit'. Similarly, the signified 'baby', inherent in the
morfoid baby, is also not part of the signified 'to babysit': a babysitter can care
for children of all ages, not for babies only.
Thanks to the OED, we know that the verb babysit appeared later than the
noun babysitter: while the latter has been used in English since 1937, the earliest
citation of the former provided by the OED dates 1947. Accordingly, we can
conclude that the verbal lexeme TO BABYSIT is, from a diachronic point of view,
a product of back-formation of the nominal lexeme BABYSITTER: the output lex
babysit came into existence via the removal of -er from the input lex babysitter.
88 Chapter 4
But what is the synchronic relation holding between the lexemes TO BABYSIT
and BABYSITTER? Do laymen who do not know that the former appeared after
the latter nevertheless analyze the verbal lexeme as a back-derivative of the
corresponding nominal lexeme? As argued by Marchand (1969: 394), this is
indeed the case. That is, from a synchronic point of view, TO BABYSIT must also
be regarded as a back-derivative of BABYSITTER because the signified 'to babysit'
is more complex than the signified 'babysitter': whereas babysitter means
'babysitter', to babysit means 'to do the job of a babysitter'. A similar example
analyzed by Marchand is the verbal lexeme PROOFREAD, which, like BABYSIT,
qualifies as an instance of back-formation from a diachronic point of view:
according to the OED, TO PROOFREAD appeared later than PROOFREADER. As
Marchand argues, the former can be considered a back-derivative of the latter
not only from a diachronic but also from a synchronic perspective because the
signified 'to proofread' is more complex than the signified 'proofreader': whereas
proofreader means 'proofreader', to proofread means 'to do the job of a
proofreader'.
This textbook rejects these analyses. From a synchronic point of view, TO
BABYSIT is a fully-idiomatic lexeme whose signified does not contain the
signifieds of the components baby and sit. That is, speakers of Present-day
English know that to babysit does not mean 'to sit with or near someone's baby'
but 'to take care of someone's child during the temporary absence of that child's
parents'. Accordingly, a babysitter can be defined as a performer of the action of
babysitting. The nominal lexeme BABYSITTER is thus semantically more
complex than the verbal lexeme BABYSIT and can therefore be regarded as its
derivative from a synchronic perspective. (In other words, we can claim that the
pair to babysit–babysitter is synchronically not different from the pair to blog–
blogger, in which the latter came into existence via affixation of the former by
means of the suffix -er.) Similarly, the verbal lexeme PROOFREAD is, from a
synchronic point of view, a quasi-idiom whose signified does not only contain
the signifieds of the components proof and read but also the idiomatic signified
'to find mistakes and make corrections': to proofread does not mean 'to read
proofs' but 'to read proofs in order to find mistakes in them and make the
necessary corrections'. Accordingly, a proofreader can be defined as a performer
of the action of proofreading. The lexeme PROOFREADER is thus semantically
more complex than the lexeme TO PROOFREAD and can therefore be regarded as
its derivative from a synchronic point of view.
This does not mean to deny that Marchand's analyses are correct from a
diachronic perspective, i.e. as in the case of to tase and skeeve, immediately
after their creation the output verbs babysit and proofread came to be associated
with the quasi-idiomatic signifieds 'to do the job of a babysitter' and 'to do the
job of a proofreader'. However, this is no longer true of present-day English
speakers who do not remember the true etymologies of these verbs.
To conclude, the diachronic history of the lexeme under analysis (its true
etymology) may be at odds with present-day speakers' intuitions (its folk
etymologies). Accordingly, a student of lexeme-formation must always specify
whether he or she approaches lexeme-formation from a synchronic or a
diachronic perspective. (The present textbook does not argue for either the
former of the latter approach: both can provide important insights into the
history and nature of morphological processes that produce new lexemes.)
4.2.5 Why do speakers of English create new lexemes?
The most important reasons are as follows.
1. lexical gaps
2. taboo
3. language users' desire for expressivity
With regard to lexical gaps, consider the lexeme FAKE-ATION 'a vacation where a
significant amount of time is spent reading email and performing other work-
related tasks' (Word Spy). According to Word Spy, FAKE-ATION is a relatively
new lexeme in English: its earliest citation provided by the database dates
February 16, 2009. From a formal point of view, the lex fake-ation can be
analyzed as a blend of the components fake and vacation: fake-ation = fake
vacation. From a semantic point of view, the lexeme FAKE-ATION can be
analyzed as a quasi-idiom whose signified does not only contain the signifieds
'fake' and 'vacation', inherent in the blended components fake and vacation, but
also the idiomatic signified 'defining characteristics of such a fake vacation
distinguishing it from other fake vacations', i.e. a fake-ation is a particular kind
of fake vacation.
It is evident that the blend fake-ation appeared in English in the late 2000s
(but not in, say, the 1950s) because the phenomenon of a fake-ation is a new
phenomenon, brought about by the development of the Internet during the last
two decades. Apart from FAKE-ATION, there are many other lexemes whose
creation in the 1990s-2000s can be attributed to the need to fill the lexical gap,
resulting from the development of the Internet. For example, the quasi-idiom
BROWSER 'a computer program enabling Internet users to browse Internet pages',
the semi-idiom SEARCH ENGINE 'a service like Google or Bing that allows
Internet users to search for information on the Internet', the full-idiom FIREWALL
'a software protecting your computer from hackers' attacks and viruses', etc.
In addition to lexical gaps, new lexemes often come into existence because
of taboo. In linguistics, the term 'taboo' is used in connection with unpleasant
topics such as, for example, death, illnesses and disabilities, poverty, urination
90 Chapter 4
and defecation, sexuality, etc. These topics are said to be marked by taboo,
which means that we do not talk about these topics as freely as we talk about
taboo-free topics such as, for example, watching TV, eating, sleeping, washing
hands, etc. However, on some occasions we cannot avoid talking about things
like having sexual relationships with other people, going to the toilet, being
seriously ill, dying, and so on. For instance, many people have sexual partners
whom they are not officially married to. To describe the relationship holding
between these people without explicitly referring to sex, speakers of English
usually use the fully-idiomatic lexemes BOYFRIEND2 and GIRLFRIEND2.
Lexemes like BOYFRIEND2 and GIRLFRIEND2 are usually called euphemisms.
A euphemism is a lexeme whose lex 'indirectly' expresses a taboo-marked
concept. For example, by calling somebody boyfriend, we do not directly name
the most important but taboo-marked characteristic of the relationship holding
between that person and his girlfriend, i.e. that the two are sexual partners.
Instead, we use a peripheral but taboo-free characteristic 'friendship', which is
sometimes (but not always) typical of the boyfriend–girlfriend relationship.
The need to create a euphemism can thus be considered another important
motivation for lexeme-creation. However, it must be observed that euphemistic
expressions are not necessarily lexes which realize new lexemes. Thus the
famous American euphemisms bathroom and restroom (when used to refer to
toilets rather than to rooms where people can take a bath / have a rest) can be
said to be associated with the signified 'toilet' and can therefore be regarded as
allolexes of the already existing TOILET lexeme rather than as lexes realizing
new lexemes. Hence the need to create a euphemism is usually a motivation for
lex-formation rather than for lexeme-formation.
Another reason accounting for the creation of new lexemes, which is far
more important than both lexical gaps and taboo, is language users' expressivity
or creativity. As was pointed out by Zipf (1949: 19), "the main motivation for
speaking is to achieve success". Achieving success by means of speaking does
not really mean to be understood by other people. Much more important than
this is what Keller and Kirschbaum (2003: 12) call the wish to produce an
impression on other people ("der Wunsch zu imponieren oder positiv
aufzufallen"), the wish to show off. To illustrate this point, let us consider the
lexeme FLUNAMI 'an overwhelming number of flu cases in the same area at the
same time' (Word Spy). From a formal point of view, the lex flunami can be
analyzed as a blend of flu and tsunami. From a semantic point of view, the
lexeme FLUNAMI can be analyzed as a semi-idiom whose signified contains the
signified of the component flu but not of the shortened component -nami (which
is a morfoid whose signified 'tsunami' partially motivates the idiomatic meaning
'flunami'). Just like FAKE-ATION, FLUNAMI is a relatively new lexeme in English:
its earliest citation provided by Word Spy dates January 04, 2006. However, in
contrast to the phenomenon of a fake vacation that involves reading non-private
emails while on vacation, the phenomenon of a flunami is definitely not a new

phenomenon: influenza epidemics had surely occurred long before the mid-
2000s. Accordingly, the creation of FLUNAMI cannot be attributed to the urgent
need to fill the lexical gap, arising from the emergence of a new phenomenon.
Also, it cannot be attributed to the need to create a euphemism: influenza is not a
taboo-marked illness. The creation of FLUNAMI can only be due to the above
named desire for creativity as a means of achieving success. Thus the creator of
flunami demonstrates his or her creativity in that he or she invents a semi-
idiomatic blend one of whose components (-nami) is a morfoid whose signified
partially motivates the idiomatic meaning of the blend. The same can be said
about the creator of fake-ation. In addition to inventing a new lex for expressing
a supposedly new phenomenon, he or she demonstrates his or her ability to
creatively use the available lexeme-building resources of the English language,
thereby producing an impression on other people.
Finally, it must be mentioned that the creators of lexemes like FAKE-ATION
and FLUNAMI achieve success because of the hypostatization potential of
words. A well-known fact is that "the existence of a particular word creates the
impression that there is a corresponding thing or entity to which the word refers
to" (Schmid 2008: 5). This effect is usually called hypostatization. Indeed, the
existence of the blend fake-ation creates the impression that there are masses of
people who spend their vacations answering non-private emails, even though we
do not know for sure whether this is actually the case. Undoubtedly, there are
many people who check their emails while on vacation, but it is not clear
whether because of this, they regard their vacations as fake vacations. The term
fake-ation does, however, create the impression that fake-ations constitute a
serious problem which negatively affects the lives of many people.
Precisely because of the hypostatization potential of words, new lexemes
may be created in an attempt to draw other people's attention to phenomena like
fake-ation which the creators of these lexemes find particularly important.
Consider, for example, the following lexemes.
x PRECARIAT 'people whose lives are precarious because they have little or no
job security' (Word Spy)
x ON-CALL-OGIST 'a doctor who is frequently on call, particularly one who

earns a living by filling in for other doctors' (Word Spy)
x BIKELASH 'a strong, negative reaction towards cyclists, particularly by police

officers or drivers' (Word Spy)
Again, as in the case of the lexeme FAKE-ATION, the mere existence of these
lexemes creates the impression that there are many people whose lives are
92 Chapter 4
precarious because of the absence of job security; there are many doctors who
earn a living by filling in for other doctors; there are many police officers and
drivers who are aggressive towards cyclists. The creators of these lexemes thus
wanted to achieve success by drawing other people's attention to (in their view)
important social problems.
4.2.6 The establishment of new lexemes
As our starting point, let us compare the synchronic status of the lexemes
BOYFRIEND2 and FAKE-ATION. The obvious difference is that the former is an
established lexeme (i.e. one which is perceived to be the norm of the English
language), whereas the latter is a neologism, i.e. a lexeme that has only recently
been created and, accordingly, has not yet become a standard English lexeme.
According to Schmid (2008: 3), the establishment of a lexeme – i.e. its
development from one speaker's coinage to a standard word recognized and
used by the majority of members of the same linguistic community – has three
stages:
1. creation
2. consolidation
3. establishing
Each of these stages has three perspectives:
1. structural perspectives
2. socio-pragmatic perspectives
3. cognitive perspectives
Let us discuss each of these stages and perspectives using the neologism FAKE-
ATION as an illustrative example. We will begin with the structural perspective.
At the moment when some speaker of English produced the blend fake-
ation, a nonce-formation was created. What defines the creation stage from a
structural perspective is an instability of both the signifier and the signified of a
nonce-formation. For example, as regards fake-ation, it is not entirely clear how
the word must be spelled. Word Spy mentions four distinct orthographic
variants: fake-ation, fakeation, fakation, and facation. The same is true of its
signified: according to Word Spy, "fake-ation has as many meanings as it does
spellings". In addition to the signified discussed above, these include:
x 'calling in sick when you're healthy'

x 'a miserable or problem-filled vacation'
x 'a pretend vacation where you stay at home but take steps to make it appear
as [though] you went away (e.g. applying tanning cream)'
Observe that, according to Word Spy, the signified 'a pretend vacation' is the
earliest recorded sense of fake-ation, which has been documented since 2004.
The next stage – the consolidation – is characterized by the beginning of the
stabilization of both the signifier and the signified of a former nonce-formation.
Thus it appears that despite the still existing variation between the orthographic
forms fake-ation, fakeation, fakation, facation and the four different signifieds
mentioned above, both the signifier and the signified of the lexeme FAKE-ATION
have begun to stabilize. We are justified in arriving at this conclusion because
Word Spy is not the only Internet source where the lex of this neologism is
spelled fake-ation and defined as a vacation for people who read non-private
emails. Consider, for example, (72).
(72) Florida hotel offers 'fake-ation' package to working travelers. People

who find themselves taking 'fake-ations' – a new term coined to
describe trips taken by those who work during their vacations - may
want to take advantage of a new package offered by a Florida resort,
reports the Los Angeles Times. (http://tinyurl.com/67agrll)
The final stage – the establishment – which FAKE-ATION apparently has not yet
reached, will result in the full-lexicalization of a former nonce-formation. That
is, there will be a conventionalized association between a particular form and a
particular meaning: the signifier fake-ation will mean approximately the same
thing for the majority of English speakers. (It is also possible that at the end of
the consolidation process there will be the vocable FAKE-ATION consisting of
several lexemes expressing polysemous signifieds rather than a single FAKE-
ATION lexeme.)
Now, let us proceed to the socio-pragmatic perspective. As said above, at the
moment when some speaker of English produced the blend fake-ation, a nonce-
formation was formed, which with the course of time began to spread in the
English linguistic community. As a consequence of this, fake-ation is at present
no longer a nonce-formation, i.e. there are many speakers who use this blend or
at least understand what it means. (If this were not the case, it would not appear
in neologisms databases like Word Spy). However, despite its spreading, fake-
ation has not yet achieved the status of an institutionalized lexeme, i.e. one
which is perceived as a standard English lexeme. This can be corroborated by
the fact that recent uses of fake-ation in contexts like that of (72) are still
accompanied by meta-linguistic comments concerning both its signified and its
socio-pragmatic status, i.e. 'fake-ations' – a new term coined to describe trips
taken by those who work during their vacations. Lexes which realize
94 Chapter 4
institutionalized and fully-lexicalized lexemes like BOYFRIEND2 are usually not

accompanied by comments of this type.
Finally, from the cognitive perspective, the creation of the new lexeme
FAKE-ATION began with the creation of a new concept of a fake-ation. This does
not mean that speakers of English had been unaware of the phenomenon of a
vacation that involves spending a significant amount of time answering non-
private emails before the blend fake-action appeared in the late 2000s. However,
this phenomenon was not experienced by them as a "manifestation […] of
recurrent and familiar events or personal habits" (Schmid 2008: 8). In other
words, despite the existence of the phenomenon of a fake-ation, speakers of
English did not have the concept of a fake-ation prior to the creation of the blend
fake-ation around the year 2009. Accordingly, the coinage of the blend fake-
ation by a particular speaker of English gave rise to a pseudo-concept, i.e. a
concept that existed only in the mind of that particular speaker. Then, as the
blend began to spread in the English community, the concept of a fake-ation
began to hypostatize. That is, more and more people began to experience the
phenomenon of such a vacation as a manifestation of a recurrent and a familiar
event. This is supported by (72), which mentions that Florida hotel has recently
begun offering their guests the so-called Perfect 'Fake-ation' package. According
to the hotel Web site,
The Shores Resort & Spa has created a new package for career-
conscious travelers who want to stay connected to work while on
vacation – the 'Fake-ation' Package. According to TripAdvisor®, 59
percent of travelers are connected to work than more ever before
during leisure travel, with 62 percent checking work email, and 13
percent calling into the office while on vacation. Starting at just $119
per night, the new Fake-ation package caters to this growing group of
business / pleasure travelers, with WiFi throughout the resort –
including the pool area – a business center available 24-hours daily,
coffee on demand, the Wall Street Journal upon check-in, and a
private office available for conference calls. Call 866.934.7467 to
book your 'Fake-ation' today! (http://tinyurl.com/ddbgqs).
Obviously, the existence of such a package demonstrates that the concept FAKE-
ATIONis already a fairly hypostatized concept.
4.2.7 The non-institutionalization of new lexemes
Not all nonce-formations develop into fully-established lexemes. The reasons

for this are usually of an extra-linguistic character. That is, for example, if we
suddenly stop checking emails while on vacation, the neologism FAKE-ATION

will most likely not become an institutionalized lexeme. Also, it seems that
PRECARIAT has more chances to fully-institutionalize than BIKELASH: many
people do not have job securities and are therefore indeed in a rather precarious
situation. By contrast, the lexeme BIKELASH seems to denote a less significant
social problem: there are many people who are neither cyclists nor police
officers and hence do not really know (or do not care much about the fact) that
the latter are aggressive towards the former.
In addition to extra-linguistic reasons like those named above, the non-
establishment can be due to (linguistic) blocking. As an illustration, let us
consider the non-institutionalized formations studier and liver. According to the
OED, the signifier studier denoting 'a person who is addicted to or engaged in
study; a student' appeared in English around 1380. By 1593 the lexeme STUDIER1
'a person engaged in study / student' had given rise to the quasi-idiom STUDIER2
'one who studies a specified subject'. Finally, around 1597 there emerged the
full-idiom STUDIER3 'one who strives after or pursues (an object or end)'. Of
these lexemes, none can be regarded as an established lexeme of Present-day
English. According to the OED, STUDIER1 is obsolete (i.e. non-existent in
Present-day English), whereas STUDIER2 and STUDIER3 are marked as rare.
A slightly different situation is characteristic of liver. According to the OED,
the non-idiomatic lexeme LIVER1 'a person who lives or is alive, a living creature'
was created in 1382. This lexeme still exists in Present-day English. However,
according to the OED, it can be found only in the South and the South-West of
England. Apart from LIVER1, there are also the quasi-idiomatic lexemes LIVER2
'an inhabitant, a dweller' (i.e. a person who lives at a particular place) and LIVER3
'a person who lives a life of pleasure or activity'. Of these lexemes, only LIVER3
can be regarded as a more or less established lexeme of Present-day English,
whereas LIVER2 'an inhabitant, a dweller' occurs only in the U.S.A.
In summary, what both the vocables STUDIER and LIVER seem to share is
that their non-idiomatic lexemes STUDIER1 'a person who studies' and LIVER1 'a
person who lives', which existed in English in the past, did not become fully-
institutionalized lexemes of Present-day English. That is, we cannot use the
affixed forms studier and liver to refer to performers of the actions of studying
and living. In this respect, studier and liver are different from forms like baker,
teacher, preacher, worker, writer, etc.
The non-institutionalization of the lexemes STUDIER1 and LIVER1 is often
attributed to blocking. As for studier, it is argued that its establishment was
prevented by the synonymic lexeme STUDENT. A very similar case is an often
cited non-institutionalized formation stealer. According to the OED, the lexeme
STEALER1 'one who steals; a thief' appeared in English around 1508. However,
just like STUDIER1, it did not manage to develop into a fully-established lexeme.
As we pointed out in 2.4.7, in Present-day English there is only the quasi-
96 Chapter 4
idiomatic lexeme STEALER2 'one who steals something specified'. A possible

explanation for the non-institutionalization of STEALER1 is the existence of the
fully-synonymic lexeme THIEF. (If we accept these analyses, we can regard the
non-institutionalization of STUDIER1 and STEALER1 as instances of synonymic
blocking.)
In contrast, the non-institutionalization of the lexeme LIVER1 is usually
attributed to homonymic and semantic blocking. As for the former, the
signifier liver is associated with the totally unrelated homonymic meaning 'an
inner organ of the human body'. According to the OED, the lexeme LIVER 'an
inner organ' has existed since the Old English period, whereas LIVER 'a person
who lives' did not appear until the end of the 14th century. It is possible that the
former could have blocked the institutionalization of the latter. As argued by
Ullmann (1957: 122), a homonymic clash like that of LIVER 'an inner organ' and
LIVER 'a person who lives' is a "pathological situation [which] arises whenever
two or more incompatible senses [...] develop around the same name".
At the same time, however, as Plag (2003: 64) argues, this homonymic clash
could not have been the sole reason for the non-institutionalization of LIVER1.
Indeed, all languages tolerate homonyms. For example, as was observed in
2.2.1, in English there are the homonymic lexemes CASE 'a set of circumstances
or conditions' and CASE 'a box or receptacle for holding something'. Similarly,
there are the homonymic lexemes BANK 'financial institution' and BANK 'the
ground near the river'. If homonymic clashes were indeed such an intolerable
pathology, the just named pairs of homonymous lexemes (both of whose
members are fully-institutionalized lexemes in Present-day English) would most
likely not exist.
According to Plag, the non-institutionalization of LIVER1 can be attributed to
the fact that the derivational suffix -er typically denotes an agent, i.e. a
deliberate initiator of the action specified by a preceding verb. That is, a
preacher is the agent of the action of preaching (a person who deliberately
preaches); a teacher is the agent of the action of teaching (a person who
deliberately teaches); a writer is the agent of writing (a person who deliberately
writes); etc. The agentive meaning is also evident in formations like New Yorker
'an inhabitant of New York', Londoner 'an inhabitant of London', Berliner 'an
inhabitant of Berlin', etc. A New Yorker is the agent of the action of living in
New York: he or she initiated this action by either moving to New York or (if he
or she was born in New York) by not moving to a different place. By contrast,
living (in the sense 'just being alive') is not an agentive activity. We are not the
agents but the experiencers of the action of living. We live as long as we are
alive. This is an inherent property of all living creatures rather than a
deliberately initiated activity.
That this could indeed have been the reason for the non-institutionalization
of LIVER1 is corroborated by the existence of the lexeme LIVER3 'a person who
lives a life of pleasure or activity'. (According to the OED, this lexeme is neither
obsolete nor rare and can thus be regarded as more institutionalized than
LIVER1.) Thus a person who lives a life of pleasure or activity is obviously the
agent of the action of living a life of pleasure or activity: he lives such a life
because he decided to do so.
A number of recent studies regard blocking not (only) as a diachronic
phenomenon accounting for the non-institutionalization of a particular lexeme
(which, however, existed at some point in the past) but rather as a synchronic
explanation for the non-occurrence of new formations like STEALER1 and
STUDIER1. In other words, why do present-day English speakers (who do not
remember that STEALER1 and STUDIER1 existed in the past but could not develop
into fully-institutionalized lexemes and therefore disappeared from the English
language) not create these lexemes anew? Or to be more precise, why do
present-day speakers of English not create additional allolexes of the lexemes
THIEF and STUDENT?
The synchronic answer to this question given by these studies is, however,
identical with what was said in connection with the diachronic perspective
discussed above: the formation of the allolexes stealer and studier is still
blocked by the established and frequently used lexes thief and student. As e.g.
Rainer (2005: 337) argues, the degree of synchronic semantic blocking largely
depends on the frequency of the blocking lex: frequently used lexes have a
stronger blocking potential than low frequency lexes; e.g. thief and student are
still capable of blocking the formation of the allolexes stealer and studier
because present-day English speakers fairly often hear the words thief and
student.
This textbook cannot accept this analysis. As we have seen at earlier points,
English (as well as any other language) abounds in lexemes whose lexes express
fully- or at least near-synonymic signifieds. For example, true and faithful are
fully-synonymic allolexes of the same TRUE / FAITHFUL lexeme; FEVER and
PYREXIA are fully-synonymic allolexes of the same FEVER / PYREXIA lexeme;
BATHROOM and RESTROOM are fully-synonymic allolexes of the same TOILET
lexeme; etc. Moreover, as we will see in the next section of this chapter, English
has several mechanisms that are exclusively used for lex-formation (i.e. the
formation of fully-synonymic allolexes of already existing lexes). One of such
mechanisms is clipping, i.e. the shortening of an input lex. For example, girlf is
a fully-synonymic clipped allolex of the input lex girlfriend. According to the
OED, girlf has been recorded in (chiefly British) English since 1991. Note that
the input lex thief, which supposedly blocks the formation of the synonymous
allolex stealer, has a lower frequency of use than girlfriend, which has recently
served as an input lex for the synonymous allolex girlf: BYU-BNC has 1332
occurrences of girlfriend, but only 730 occurrences of thief. Why is it then that
the more frequently used lex girlfriend did not block the formation of the fully-
98 Chapter 4
synonymous allolex girlf, whereas the less frequently used lex thief still blocks
the formation of the fully-synonymous allolex stealer?
The answer to this question seems to be that the clipping girlfriend Æ girlf is
an instance of deliberate lex-formation, i.e. the clipper of girlfriend wanted to
have a shorter allolex for expressing the same signified 'girlfriend'. The creator
of girlf knew from the very beginning that the clipped output lex would become
a fully-synonymic allolex of the corresponding non-clipped input lex. Also, he
or she knew that the two allolexes would occur in stylistically relevant free
variation: i.e. that the shorter girlf would be a less formal variant of the longer
girlfriend. (This is a general characteristic of clipping: clipped output lexes are
usually less formal than corresponding non-clipped input lexes.) By contrast, if
someone ever coins stealer and studier, this will most likely not be because of
the wish to create additional allolexes of the established lexes thief and student
but because of the wish to form affixed forms by means of the derivational
suffix -er. The consequence of this, however, will be the formation of two
additional allolexes of uncertain status. This means that it will not be clear
whether thief and student must retain their status of the primary signifiers of the
lexemes THIEF and STUDENT or instead be superseded by the new lexes stealer
and studier. The latter lexes will thus 'encroach' on the dominant status of the
former, thereby considerably enhancing the possibility of blocking (or to be
more precise, the possibility of the non-institutionalization) of stealer and
studier.
At the same time, note that there is absolutely no reason why thief and
student must block the formation of stylistically different allolexes stealer and
studier. That is, if some speaker of English decides to use these affixed forms as
less formal allolexes of thief and student, there will be no synonymic blocking:
if the lexes stealer and studier coexisted with thief and student in the past, it is
very possible that this situation will recur in the future.
Finally, let us briefly return to the complementary distribution of the
allomorphs un- and in-, which we discussed in Chapter 2. As was observed in
2.3, in- combines only with words of Latin or Romanic origin, whereas un- is
used with native or completely naturalized words. This is the reason why there
can only be untrue and inadequate but not *intrue and *unadequate. For some
authors, examples like this are instances of morphological blocking.
This book argues for a different terminological solution. First of all, notice
that like synonymic blocking, so-called morphological blocking exemplified by
the impossibility of *intrue and *inadequate seems to be more relevant for lex-
formation rather than for lexeme-formation. That is, if it were not for these
combinatory restrictions, it would now be possible to form additional allolexes
of the lexemes UNTRUE and INADEQUATE. Second, as was stated in 2.3, the
impossibility of the formations *intrue and *unadequate is due to the different
syntactics of the prefixes un- and in-: there can be no *intrue because the
syntactics of in- does not allow us to combine this prefix with the Germanic
input lex true. And there can be no *unadequate because the syntactics of un-
does not allow us to combine this prefix with the Latinate input lex adequate.
This phenomenon could perhaps be better referred to as syntactics' blocking
rather than morphological blocking. However, observe that in contrast to the
examples discussed above, in the case of syntactics' blocking, there is no
prevention of the institutionalization of either a new lexeme or a new lex of the
same lexeme: neither the former nor the latter is actually formed as a result of
existing syntactics' blocking. That is, according to the OED, the lex *intrue has
never existed in English.
Taking this into account, this book suggests that the term 'blocking' should
be reserved only for cases of semantic blocking in which the institutionalization
of an already formed lexeme or a lex of the same lexeme is prevented by some
inherent semantic properties of the signified of the lexeme in question, whereas
the non-existence of forms like *intrue and *unadequate should be regarded as
consequences of the inherent properties of the syntactics of the prefixes un- and
in- rather than as instances of blocking.
4.2.8 Productivity
The term 'productivity' refers to the ability of a particular lex- or lexeme-

building mechanism to produce new signifiers, be it lexes that realize new
lexemes or additional allolexes that realize already existing lexemes. In the case
of affixation, we usually speak of the productivity of an affix (e.g. the
productivity of the derivational suffix -er), whereas in the case of processes like
compounding, conversion, etc., we speak of the productivity of a particular
formation pattern (e.g. the productivity of the noun Æ verb conversion).
Since productive morphological processes are those that give rise to new
signifiers, any study of productivity must begin with the identification and
documentation of recently coined signifiers. With regard to lexeme-formation
processes, we can rely on databases of English neologisms such as, for example,
the already mentioned Word Spy (http://www.Word Spy.com/).
Word Spy is a continually updated, online-based database of neologisms that
appeared in the English language during the last three decades. The entries of
the database are arranged both chronologically and thematically. However, if
you click at 'Posts by date' and then at e.g. '2011', Word Spy will list all
neologisms that were added to the database in the year 2011, not the neologisms
that were necessarily created in 2011. For example, the lexeme COPYFIGHTER 'a
person who opposes copyright laws and practices that he or she perceives to be
unfair' was posted to the Word Spy Web site in 2011. However, its earliest
citation provided by Word Spy dates January 22, 2003. Similarly, there is an
100 Chapter 4
entry for the lexeme AFTER-PARTY 'a social gathering that occurs after a party,
especially after a rave' which was added to the database in 2002. However, its
earliest citation dates March 17, 1980.
If you click at 'Posts by category', Word Spy will list a number of (mainly
thematic) categories into which recently created lexemes have been classified by
the author of the Web site. These include, for example, 'Books and magazines',
'Corporate culture', 'Economics', 'Health and fitness', 'Money', 'Medicine',
'Psychology', 'Race', etc. If you click at e.g. 'Money', Word Spy will list all
recently created lexemes which, according to the author of the Web site, have
something to do with money.
Note that some of the categories are not thematic but genuinely linguistic
categories. For example, there is the category 'Verbed nouns', listing verbal
lexemes which have recently come into existence via morphological conversion
of corresponding input nominal lexemes. For example:
x TO BACKGROUND 'to surreptitiously perform a task in the background while

one's attention is supposed to be on another task' (Word Spy)
x TO OFFICE 'to perform office-related tasks, such as photocopying and faxing'

(Word Spy)
x TO PIE 'to hit a person, particularly a political or business leader, in the face
with a pie' (Word Spy)
Also, there is the category 'Idioms', which lists recently coined idiomatic
phrases. For example:
x BIRDS OF A FEATHER MEETING 'a meeting held at a computer-related trade

show or conference in which people who work in the same technology area
at different companies exchange information and experiences' (Word Spy)
x NAILING JELLY TO A TREE 'tackling a particularly messy, and probably

impossible, problem' (Word Spy)
x TO PUT WOOD BEHIND THE ARROW 'to provide a product or company with
money and other resources' (Word Spy)
If you want to study the productivity of a derivational affix, go to the Advanced

search page at http://wordspy.com/search.asp, enter '*suffix' or 'prefix*' (both
without quotation marks) to the search mask, choose the option 'Within title',
and then click at 'Search'. That is, for example, if you want to learn whether
Word Spy contains neologisms whose lexes came into existence via affixation
by means of the verb-forming suffix -ize (e.g. centralize), enter '*ize' (without
quotation marks). If you want to learn whether Word Spy contains neologisms
whose lexes came into existence via affixation by means of the prefix un-, enter
'un*' (again without quotation marks). Word Spy will then yield all words
ending in -ize and beginning with un-. Some of these words could be verbs
containing the suffix -ize and negative adjectives containing the prefix un-. As
regards the former, Word Spy contains, for example, the following verbs:
x TO BAGONIZE 'to wait anxiously for your suitcase to appear on the airport
baggage carousel' (Word Spy)
x TO DOLLARIZE 'for a country to abandon its national currency in favor of the

U.S. dollar' (Word Spy)
x TO VELOCITIZE 'to cause a person to become used to a fast speed' (Word Spy)
As regards the latter, Word Spy contains the already mentioned negative
adjective ungoogleable. Accordingly, we are justified in concluding that both
the suffix -ize and the prefix un- are productive derivational affixes in Present-
day English.
An alternative to Word Spy is the OED, or to be more precise, the online
version of the OED available at http://www.oed.com/. Note that you or your
institution must subscribe to the electronic version of the OED in order to be
able to access it online.
The two major advantages of the OED is that it enables us to (easily)
determine the productivity of virtually any lexeme-building mechanism as well
as to compare the diachronic productivity of a particular mechanism. For
example, if you want to find out (using Word Spy) whether speakers of English
have recently created new lexemes by means of apophony, you have to
manually analyze each of the Word Spy entries: at present, it is impossible to
order Word Spy to specifically search for instances of apophony (as well as for
instances of such lexeme-building mechanisms as lexeme-manufacturing, back-
formation, borrowing, etc.). By contrast, using http://www.oed.com/, you can
search the OED for instances of all these mechanisms. Just go to the OED
Advanced search page at http://www.oed.com/advancedsearch, enter 'alteration'
(without quotation marks) to the search mask of the first row from above,
choose the option 'Etymology' instead of 'Full text' (near to the search mask),
enter '1990-' to the 'Date of entry' below the search mask, and then click at
'Search'. The OED will then find all recent words (i.e. those which appeared
after 1990) whose Etymology-section contains the word alteration, i.e. how the
OED calls phonetic alterations similar to those of to increase Æ an increase and
102 Chapter 4
bouldering Æ buildering, which we discussed in 4.2.2. Most of the results will

be forms that came into existence via apophony.
Recent instances of apophony provided by the OED include, for example:
x adultescent 'an adult who has retained the interests, behavior, or lifestyle of
adolescence' (OED / 1996) Å adolescent
x Google 'a proprietary name for an Internet search engine launched in

September 1998' (OED / 1999) Å googol 'a fanciful name […] for ten raised
to the hundredth power' (OED / 1940)
x Lollywood 'The Pakistani film industry, based in Lahore […]' (OED / 1995)
Å Bollywood 'the Indian film industry, based in Bombay' (OED / 1976)
Given these examples, we can claim that apophony is still a productive lexeme-
building mechanism in Present-day English.
If you are interested in the diachronic productivity of apophony (i.e. the
question whether this mechanism was more / less productive during a particular
period in the history of the English language), repeat the same search without
entering '1990-' to 'Date of entry'. The OED will then find all instances of
apophony that have ever been created in English. Then click at 'Timeline' (near
'View as: List') and the OED will yield a graph consisting of 20 bars. Each bar
corresponds to the interval of 50 years. The higher the bar is, the larger the
number of words that came into existence via apophony during that particular
interval of time.
Finally, it is often suggested that the synchronic productivity of a lexeme-
building mechanism (especially, the productivity of a particular derivational
affix) can be determined with the help of a synchronic corpus like COCA. The
point here is that neologisms can often be found among low frequency forms, in
particular, among so-called hapax legomena, i.e. forms which occur no more
than once in a given corpus. The reason for this is obvious: (lexes which realize)
newly created lexemes which have not yet reached the status of fully-established
lexemes are usually not used very often.
For example, if you want to determine the synchronic productivity of the
noun-building suffix -hood (of e.g. childhood and waithood) using COCA, you
have to enter '*hood.[nn*]' (without quotation marks) to the search mask of the
corpus and click at 'Search'. COCA will then yield all nouns (occurring in this
corpus) that end in -hood. Some of them will be suffixed nouns like childhood
and waithood.
Then you have to write down all hapax legomena yielded by COCA and
check whether these forms are indeed lexes that realize previously unattested
lexemes. (This is necessary because a hapax legomenon is not necessarily a
neologism. It may be just a low frequency word that appeared long ago.) For this
purpose, establish whether the hapax legomena found by COCA occur in
dictionaries like the OED or MWO. If you succeed in finding at least one
unattested hapax (i.e. a form for which neither the OED nor MWO have an
entry), you will be justified in claiming that the derivational affix under analysis
is still a productive noun-building affix.
At present, COCA does indeed contain a number of hapax legomena that are
not listed in either the OED or MWO. These include, for example, refugee-hood,
rookiehood, otherhood, pointhood, etc. Consider, for instance, the meaning of
refugee-hood in (73).
(73) The fact that Palestinian cultural and political communities have not
yet coincided in time and space also points to the current link between
Palestinian identity and refugee-hood (COCA)
It appears that the meaning of the refugee-hood of (73) is very similar to that of
childhood and waithood. Like the latter, refugee-hood also seems to denote the
state of being a refugee. Accordingly, we can conclude that -hood is still a
productive suffix that attaches to nouns like refugee, thereby forming nouns
denoting a state: refugee-hood = the state of being a refugee.
If you want to establish the synchronic productivity of a prefix like un-, just
enter 'un*.[j*]' (without quotation marks) and repeat the procedure discussed
above. (COCA will then search for adjectives beginning with un-. If you enter
only 'un*', COCA will search for lexes of all word classes beginning with un-.)
4.3 Lex-formation
As we argued in 4.1, lex-formation is different from lexeme-formation in that it

produces allolexes of already existing lexemes rather than lexes realizing new
lexemes. Another important difference is that lex-formation is never caused by
the need to fill a lexical gap: this is a priori impossible, given that lex-formation
never gives rise to new lexemes. The primary motivation for lex-formation is the
language user's desire to show off, the wish to produce an impression on other
language users by replacing an old 'dull' lex by a new 'fresh' allolex.
Below we will discuss the most important mechanisms that speakers of
English have at their disposal for creating allolexes of already existing lexemes.
1. lex-forming clipping
2. lex-forming suppletion
3. lex-forming abbreviation
104 Chapter 4
4. lex-forming borrowing
5. lex-forming apophony
6. lex-forming affixation
7. lex-forming syntactics' change
8. lex-forming orthographic modification
4.3.1 Lex-forming clipping
Lex-forming clipping is the shortening of an input lex. According to Marchand

(1969: 442-445), clipped allolexes can be classified into:
1. back-clippings
2. fore-clippings
3. mid-clippings
Back-clippings are clipped output allolexes in which the back part of their non-
clipped input lexes is retained. For example, girlf (Å girlfriend), mobe (Å
mobile), refi (Å refinancing), etc.
Fore-clippings are clipped output allolexes in which the fore part of their
non-clipped input lexes is retained. For example, brane (Å membrane), droid
(Å android), fro (Å Afro), etc.
Finally, mid-clippings are clipped output allolexes in which the middle part
of their non-clipped input lexes is retained. For example, flu (Å influenza),
fridge (Årefrigerator), jams (Åpyjamas), etc. (The last example jams can be
regarded as both a fore-clipping and a mid-clipping.)
Lex-forming clipping is often a means of creating less formal first names.
For instance, Alex is a less formal back-clipping of Alexander; Tina is a less-
formal fore-clipping of Christina; Liz is a less formal mid-clipping of Elizabeth;
etc.
4.3.2 Lex-forming suppletion
In addition to clipping, informal first names can be formed with the help of
suppletion. For example, in Russian Sasha is a less formal suppletive allolex of
the first names Alexander and Alexandra (which are of Greek origin). In the
English linguistic community, the signifiers Sasha and Alexander are usually
regarded as two different names (see e.g. Hanks et al. 2006) and hence must be
seen as lexes realizing two different lexemes; e.g. the British comedian Sacha9
9
Sacha is a spelling variant of Sasha.
Baron Cohen cannot be addressed as Alexander Baron Cohen. At the same time,
observe that the American skater Alexandra Pauline Cohen is usually referred to
as Sasha Cohen. (The reason for this is, however, the Russian origin of her
mother, who emigrated to the U.S.A. from the Soviet Union.10). Accordingly, at
least in this case, we can analyze the signifier Sasha as a less formal suppletive
allolex of Alexandra.
Genuinely English instances of lex-forming suppletion include, for example,
Bill Å William, Bob Å Robert, Dick Å Richard, Ted Å Edward, etc.
4.3.3 Lex-forming abbreviation
Lex-forming abbreviation can be illustrated by the shortening of input idiomatic

phrases like British Broadcasting Corporation and Oxford English Dictionary to
shorter output allolexes like BBC and OED. That is, both the former and the
latter are quasi-idioms, whose signifieds are not entirely representable in terms
of their components' signifieds: both British Broadcasting Corporation and BBC
denote a particular corporation that broadcasts in Britain and both Oxford
English Dictionary and OED denote a particular English dictionary that has
something to do with Oxford.
Lex-forming abbreviation must be distinguished from (lexeme-building)
blending. Compare, for example, the just named abbreviations BBC / OED and
the blend BRICs 'the countries of Brazil, Russia, India, and China viewed as a
group of emerging economies with large potential markets' (Word Spy). At first
glance, it may seem that BRICs came into existence in essentially the same way
as both BBC / OED. That is, BBC = British Broadcasting Corporation and OED
= Oxford English Dictionary. Similarly, BRICs = Brazil, Russia, India, China.
Nevertheless, despite this formal similarity, BBC / OED and BRICs must be
analyzed as instances of two different processes: lex-forming abbreviation and
lexeme-building blending. The difference between the former and the latter is
that while the shortening of British Broadcasting Corporation and Oxford
English Dictionary to BBC and OED gave rise to allolexes of already existing
lexemes – as we said above, BBC and OED mean the same thing as British
Broadcasting Corporation and Oxford English Dictionary – the shortening of
Brazil, Russia, India, China to BRICs gave rise to a new quasi-idiomatic lexeme
whose signified is not entirely representable in terms of its components'
signifieds: BRICs does not mean 'Brazil' + 'Russia' + 'India' + 'China' but 'these
countries viewed as an economic unit'. Accordingly, BRICs is an instance of an
anisomorphic lexeme-building blending, whereas BBC and OED are products of
lex-building abbreviation.
10
See http://tinyurl.com/ygfwam8.
106 Chapter 4
Abbreviations are traditionally classified into alphabetisms and acronyms.

The former are abbreviations that are pronounced letter by letter. For instance,
BBC /biːbiːˈsiː/. The latter are abbreviations that are pronounced as normal
words. For example, SARS (Åsevere acute respiratory syndrome) is pronounced
/sɑːz/. Very often one and the same abbreviation can be pronounced as both an
alphabetism and as an acronym. For example, FAQ (Å frequently asked
questions) is pronounced both /ɛfeɪˈkjuː/ and /fak/.
4.3.4 Lex-forming borrowing
Lex-forming borrowing is the process of borrowing a foreign language signifier

with the purpose of creating a stylistically different allolex of a lex of an already
existing lexeme. For example, the borrowing of the Latin pyrexia, which took
place around 1777, gave rise to a stylistically different allolex of the already
existing lex fever, which has been recorded in English since circa 1000 (OED).
Similarly, the borrowing of the German signifier hausfrau in 1798 gave rise to
an additional allolex of the already existing lexeme HOUSEWIFE 'a woman
(usually, a married woman) who manages or directs the affairs of her household'
(OED), which has been realized by the native English lex housewife since circa
1225.
4.3.5 Lex-forming apophony
Lex-forming apophony has recently given rise to the following allolexes:
x feck (Å fuck)
x Merkin (Å American)
x shedload 'a large amount or number' (Å shitload)
As in the case of all other lex-forming mechanisms discussed above, these

output signifiers have the same meaning as their corresponding input signifiers:
feck means 'fuck', Merkin means 'American', shedload means 'shitload'.
4.3.6 Lex-forming affixation
In Section 4.1 we got acquainted with one instance of lex-forming affixation: the
formation of adverbs from adjectives by means of the lex-forming suffix -ly.
E.g. happy Æ happily.
Apart from this, lex-forming affixation also includes so-called expletive

infixation. Recall the infix -ma- of edumacation, which we discussed in 2.5.4.
Since the output signifier edumacation has a different meaning than the input
signifier education – edumacation means 'poor education' – we are justified in
concluding that the infix -ma- is a usual lexeme-building affix. Now, consider
the infix -my- of jurismydiction in (74).
(74) Agent Smith: Lieutenant.

Lieutenant: Oh shit.
Agent Smith: Lieutenant, you were given specific orders.
Lieutenant: Hey, I'm just doing my job. You give me that juris-my-
diction crap, you can cram it up your ass (The Matrix)
This exchange takes place in the beginning of the famous science fiction film
The Matrix. Obviously, the infix -my- inserted by the Lieutenant to the input
signifier jurisdiction does not create a lex realizing a different lexeme: the
output signifier juris-my-diction has exactly the same meaning as the input
signifier jurisdiction and, accordingly, can be considered its allolex. What -my-
does here is express the emotional state of the Lieutenant, who is not quite
happy with Agent Smith's interference. The infix -my-, in contrast to the infix -
ma- of edumacation, is thus a lex-forming affix.
4.3.7 Lex-forming syntactics' change
Lex-forming syntactics' change can be illustrated by means of the syntactic

contrast between the early of (75) and the early of (76).
(75) John Andrew's own family – Gala and their two girls – had departed
early in May […] (COCA)
(76) Glancing back at the party on the lawn, he saw a bespectacled woman
twirling shirtless on the sidewalk and regretted his early departure
(COCA)
While the early of (75) has the syntactics of an adverb (cf. They lived happily),
the early of (76) has the syntactics of an adjective (cf. his happy life). This
syntactic contrast is, however, not accompanied by a difference in meaning:
both the former and the latter carry the meaning 'early'. Accordingly, the early of
(75) and the early of (76) can be regarded as allolexes of the same lexeme which
occur in complementary distribution.
Compare also the signifieds and the syntactics of the sells of (77) and the
sells of (78).
108 Chapter 4
(77) The book sells for $ 19.95 (COCA)

(78) Amazon sells the book for $ 19.95
As regards their signifieds, there does not seem to be a difference: both the sells
of (77) and the sells of (78) mean 'sells'. However, while the sells of (77) is a
transitive verb, which is accompanied by the object NP the book (cf. The book
is sold by Amazon for $ 19.95), the sells of (78) is an intransitive verb, which is
followed by the complement PP for $ 19.95. Recall that complements (but not
objects) can be expressed by adjectives. For example, The book sells well.
In summary, the sells of (77) and the sells of (78) have the same meaning
but different syntactics. Accordingly, these are not lexes realizing two different
lexemes but allolexes of the same lexeme which, like the early of (75) and the
early of (76), occur in complementary distribution.
Thanks to the OED, we know that it is the adverbial early which had
appeared before the adjectival early and it is the transitive sell which had
appeared before the intransitive sell. Accordingly, we can argue that adverbial
early Æ adjectival early and transitive sell Æ intransitive sell are instances of
lex-forming syntactics' change.
4.3.8 Lex-forming orthographic modification
Lex-forming orthographic modification is the creation of an orthographically

different output allolex which has the same pronunciation as a corresponding
input lex. For example, there is only an orthographic difference between the
input lex gangster and the output allolex gangsta: both have the pronunciation
/ˈgaŋstə/. Similarly, there is only an orthographic but not a phonetic difference
between through and thru, U.S.A. and USA, you and u (as often used on the
Internet), etc.
4.4 Exercises
a) The terms 'lexeme-formation' and 'lex-formation' are synonyms.

b) Lexical gap is one of the reasons for lexeme-formation.
c) Nonce-formations always develop into fully-institutionalized lexemes.
d) Established lexemes have hypostatized signifieds.
e) Neologisms can often be found among hapax legomena.

f) Morphological conversion is an instance of semantic change.
g) Compounding always produces allolexes of already existing lexemes.
h) New lexemes can be borrowed from other languages.
i) There are both lex-forming and lexeme-building apophonies.
j) Alphabetisms are pronounced letter by letter.
3. State which mechanisms gave rise to the following neologisms:
a) cheapuccino 'an inexpensive, low-quality cappuccino […]' (Word Spy)

b) pumpkineer 'a person who grows giant pumpkins, particularly ones meant to
be entered in pumpkin-weighing contests' (Word Spy)
c) to Alt-Tab 'to switch from one running computer program to another' (Word
Spy)
d) to pope 'to convert to Catholicism' (Word Spy)
e) 9/11 'September 11, 2001, the date of the terrorist attacks on the World Trade
Center in New York and the Pentagon in Washington' (Word Spy)
f) thumbo 'an error made while using the thumbs to type […]' (Word Spy)
g) globality 'a worldwide, interconnected economy that ignores national
boundaries' (Word Spy)
h) mission from God 'a crucially important task that must not fail […]' (Word
Spy)
i) soft power 'power based on intangible or indirect influences such as culture,
values, and ideology' (Word Spy)
j) to elder 'to share wisdom and knowledge with people who are younger than
oneself' (Word Spy)
4. State which lex-forming mechanisms gave rise to the following signifiers.
a) feck-all (= fuck-all) 'nothing at all' (OED)

b) NATO
c) ad
d) painfully
e) Jeff
f) almost of his almost victory
g) Biddy (Å Bridgit)
h) math
i) UK
j) bobby 'a slang nickname for a policeman' (OED)
110 Chapter 4
5. Analyze the synchronic productivity of the following derivational suffixes

using Word Spy, i.e. determine whether the database lists lexemes that have
been recently created with the help of the following affixes.
a) -er (e.g. teacher)

b) de- (e.g. to decentralize)
c) -th (e.g. length)
d) -ling (e.g. wolfling)
e) -fold (e.g. twofold)
f) neo- (e.g. neoconservative)
g) -ation (e.g. formation)
h) multi- (e.g. multilateral)
i) -eer (e.g. auctioneer)
j) in- (e.g. inadequate)
4.5 Further reading
The starting point of this chapter was the classification of word-formation into
lexeme- and lex-formation. This may seem a terminological innovation, but do
observe that similar ideas were already expressed by Marchand (1969), who
argued that mechanisms like clipping should be treated differently from
compounding and derivation.
The idea that the established term 'word-formation' should be replaced by
'lexeme-formation' is also expressed by e.g. Lipka (2002: 92). Proceeding from
this claim, Chapter 3 of his textbook on English Lexicology (Lipka 2002)
provides a fairly similar classification of lexeme-building mechanisms into both
formal and semantic mechanisms.
Section 4.2.6 The establishment of new lexemes was based on the article
Schmid (2008). For a discussion of the concepts 'institutionalization' and
'lexicalization', see also the article Hohenhaus (2005).
Productivity of various lexeme-building mechanisms has been the topic of
many recent studies. For example, Mühleisen (2010) deals with the productivity
of the suffix -ee (e.g. employee). For an overview of how the term 'productivity'
has been dealt with in lexical morphology, see the article Bauer (2005).
5 Lexeme-building mechanisms
In this chapter we will enlarge on all lexeme-building mechanisms that were

briefly introduced in 4.2. Each section of this chapter begins with a discussion of
the most important theoretical issues relevant for a particular lexeme-building
mechanism (e.g. the directionality of conversion, differences between idiomatic
adjective + noun compounds and idiomatic adjective + noun phrases) and then
proceeds to the analysis of that mechanism's synchronic productivity, i.e. the
question of whether the mechanism under analysis has recently been used for
creating new lexemes. Special emphasis is laid on the methodological issue of
how (i.e. with the help of which tools) students of English morphology
themselves can determine whether a particular mechanism is still productive in
Present-day English.
5.1 Semantic change
As was pointed out in 4.2.1, purely semantic mechanisms can be subsumed

under the term 'semantic change'. For example, the lexeme STEALER1 'one who
steals; a thief' can be said to have undergone semantic change in that it gave rise
to the quasi-idiom STEALER2 'one who steals something specified'. That is,
without any overt changes, the signifier stealer came to be associated with a
different signified, thereby giving rise to a new lexeme. A similar example is
boyfriend. As we established in 4.2.4, the fully-idiomatic lexeme BOYFRIEND2 'a
male of almost any age over puberty with whom a woman has a non-marital
sexual relationship' is not a product of compounding of the input roots boy and
friend but of semantic change of the already existing lexeme BOYFRIEND1 'a
friend who is a boy, a boyhood friend'.
What is shared by the output lexemes STEALER2 and BOYFRIEND2 is that their
input lexemes STEALER1 and BOYFRIEND1 are not institutionalized lexemes of
Present-day English. That is, the signifier stealer is now associated only with the
quasi-idiomatic signified 'one who steals something specified' and the signifier
boyfriend is now chiefly associated with the fully-idiomatic signified 'a male
sexual partner'. This, however, is not the default outcome of semantic change. In
the majority of cases, input and output lexemes coexist within the same vocable.
For example, the output lexeme TRUE2 'not false' coexists with the input lexeme
TRUE1 'faithful'; the output lexeme HAVE3 'to experience sexual relations' coexists
with the input lexeme HAVE2 'to experience'; the output lexeme TO WIFE 'to
downplay a woman's career accomplishments in favor of her abilities as wife
and mother' coexists with the input lexeme A WIFE 'a female spouse'; etc.
112 Chapter 5
5.1.1 Mechanisms of semantic change
The two mechanisms of semantic change are metonymy and metaphor. The
former produces output lexemes whose signifieds are more or less objectively
connected to those of their input lexemes. For example, the output lexeme
BOYFRIEND2 is a product of a metonymic change: as was observed in 1.2.2,
boyfriends are often perceived by their girlfriends as friends. Similarly, the
lexeme STEALER2 can be regarded as a product of a metonymic change of
STEALER1: both the former and the latter share the meaning 'one who steals'.
By contrast, in the case of a metaphoric change, no such connection can be
established. Consider, for instance, the lexemes FIREWALL1 'a wall that prevents
the spread of fire in a building' and FIREWALL2 'a computer program that protects
computers from viruses and especially hackers' attacks'. On the one hand, both
FIREWALL1 and FIREWALL2 are protection devices: FIREWALLS1 protect houses
from fire and FIREWALLS2 protect computers from outside intruders. On the other
hand, FIREWALLS2 are not walls that prevent the spread of fire. First of all, a
computer firewall is not 'one of the sides of a room or building connecting floor
and ceiling or foundation and roof' (MWO), which the signifier wall literally
means. Second, computer firewalls have nothing to do with fire, for neither
hackers' attacks nor viruses set computers on fire. Accordingly, the output
lexeme FIREWALL2 can be regarded as a product of a metaphoric semantic
change of the input lexeme FIREWALL1: there exists a perceived similarity
between the senses 'a wall that prevents the spread of fire' and 'a computer
program that prevents hackers' attacks' rather than an objective connection
between building firewalls and their counterparts in the digital world.
A very similar example is BOOKMARK2 'a menu entry or icon on a computer
that is usually created by the user and that serves as a shortcut to a previously
viewed location' (MWO). As in the case of FIREWALL2, BOOKMARKS2 never
qualify as bookmarks in the literal meaning of this word. That is, BOOKMARKS2
are never 'markers for finding a place in a book' (MWO). At the same time, there
is an obvious similarity between the two lexemes: both BOOKMARKS2 and
BOOKMARKS1 can be seen as devices that enable us to easily find previously seen
Web pages / places in a book that we have found particularly interesting.
Accordingly, the output lexeme BOOKMARK2 can also be regarded as a product of
a metaphoric semantic change of the input lexeme BOOKMARK1.
5.1.2 Types of metonymies
Following Kövecses and Radden (1998), metonymies can be classified into:
1. part-for-whole metonymies
Lexeme-building mechanisms 113
2. whole-for-part metonymies
3. part-for-part metonymies
As an illustration of a part-for-whole metonymy, let us consider the meaning of

speaks in (79).
(79) Mary speaks Spanish (from Kövecses and Radden 1998: 52)
As Kövecses and Radden point out, the speaks of (79) does not only refer to the
ability 'to utter words or articulate sounds with the ordinary voice' (MWO),
which the signifier speak literally means, but to the entire linguistic proficiency,
which includes not only the ability to speak but also the ability to comprehend
other people's speech. Accordingly, the speaks of (79) can be seen as a product
of a part-for-whole metonymic change: the signified of the input lexeme SPEAK1
'to utter words or articulate sounds with the ordinary voice' is part of the
signified of the output lexeme SPEAK2 'to utter words as well as to comprehend
words uttered by other speakers'. Part-for-whole metonymies are sometimes
referred to as instances of semantic widening. That is, the signifier speak can be
said to have undergone semantic widening in that it can now be used not only in
connection with speaking but also in connection with the entire linguistic
proficiency, of which speaking is a part.
The reverse of part-for-whole metonymies are whole-for-part metonymies.
These produce output lexemes whose signifieds are narrower than those of
corresponding input lexemes. In the previous chapters, we already discussed a
number of output lexemes whose signifieds can be seen as products of whole-
for-part metonymies. For example, the output lexeme HAVE3 'to experience
sexual relations' (e.g. He had many women) has a narrower signified than the
input lexeme HAVE2 'to experience' (e.g. He had a good time): experiencing
sexual relations is an instance of experiencing. Similarly, the output lexeme
TRUE2 'not false / faithful to the truth' has a narrower signified than the input
lexeme TRUE1 'faithful': being faithful to the truth is an instance of being faithful.
Both have and true can thus be said to have undergone semantic narrowing.
(This term was already introduced in 2.2.1.)
Finally, part-for-part metonymies are metonymies that cannot be regarded as
either part-for-whole or whole-for-part metonymies. Consider, for instance, the
semi-idiomatic phrase health insurance. It is a well-known fact that a health
insurance is not really a health but illness insurance (Holder 2008: 211), i.e. an
insurance that covers medical costs when the insured falls ill. We do not really
need a health insurance as long as we are healthy. Similarly, the UK Department
of Health is in reality an illness department, i.e. a department that is concerned
with illnesses rather than with health. Given these uses of health, we can argue
that in English (and in many other languages) there exists the fully-idiomatic
114 Chapter 5
lexeme HEALTH2 'illnesses / pertaining to illnesses'. The semantic development

undergone by health cannot be regarded as either a part-for-whole or a whole-
for-part metonymy. An illness is neither a part nor an instance of health and
health is neither a part nor an instance of an illness. Instead, we can regard the
signifieds 'health' and 'illnesses' as different instances of the concept THE
GENERAL CONDITION OR STATE OF THE HUMAN BODY. That is, our bodies can
either be healthy or ill. Accordingly, in the case of the semi-idioms health
insurance and department of health, one of the instances of the concept THE
GENERAL STATE OF THE HUMAN BODY – HEALTH – stands for another instance of
the same concept: ILLNESS. The metonymy HEALTH-FOR-ILLNESSES is thus a
part-for-part metonymy. (We use this metonymy because of the taboo of fear or
to be more precise, because of our being superstitious (Moskvin 2010: 102-106;
Holder 2008: 211). That is, we are still afraid of a number of illnesses and
therefore replace the taboo-marked subject ILLNESS by its converse HEALTH.)
5.1.3 Types of metaphors
Instances of a metaphoric semantic change can be classified into those that

involve an experiential similarity or an experiential co-occurrence of the
signifieds of their input and output lexemes (Lakoff and Johnson 1980: 154-
155). With regard to the former, consider again the lexeme pairs FIREWALL1–
FIREWALL2 and BOOKMARK1–BOOKMARK2. As was noted in 5.1.1, computer
firewalls are similar to building firewalls in that both the former and the latter
serve as protection devices: computer firewalls protect computers from outside
intruders and building firewalls protect buildings from the spread of fire.
Likewise, Internet bookmarks are similar to traditional book bookmarks in that
both enable us to easily find the information that we find particularly interesting
and important.
As far as experiential co-occurrence is concerned, consider the metaphoric
use of big in (80).
(80) Tomorrow is a big day for me (from Lakoff and Johnson 1999: 50)
The big of (80) obviously does not mean 'big' but 'of great importance or
significance' (MWO): a big day is a very important day. In contrast to the
lexemes FIREWALL2 and BOOKMARK2, the output lexeme BIG2 'of great
importance' does not exhibit an experiential similarity to the lexeme BIG1 'big':
important days are never literally bigger than unimportant days. What accounts
for the metaphoric use of big is the existence of things that are simultaneously
big and important. As Lakoff and Johnson (1999: 50) conjecture, as children, we
often discover that big things such as e.g. our parents are not only big but also
important. This experiential co-occurrence of objects that simultaneously

possess the characteristics of being big and being important could have given
rise to the metaphoric lexeme BIG2 'of great importance'.
5.1.4 Morphological conversion
Morphological conversion is a type of semantic change that produces output

lexemes whose lexes are members of different word classes than the lexes of
their input lexemes. For example, as we established in 4.2.1, while the input
lexemes A WIFE and THOMAS are realized by the nominal lexes a wife and
Thomas, the output lexemes TO WIFE and TO THOMAS are realized by the verbal
lexes to wife and to Thomas. The explanation for this is that the signifieds of the
converted output lexemes are headed by verbal semantic components. That is,
the quasi-idiomatic signified of the output lexeme TO WIFE can be represented as
the VP to downplay a woman's career accomplishments in favor of her abilities
as wife and mother, whose head is the verb downplay. Similarly, the fully-
idiomatic signified of the lexeme TO THOMAS is representable as the VP to
accuse a person of sexual harassment, whose head is the verb accuse. This
accounts for the fact that both the output lexes to wife and to Thomas are verbs,
whereas the input lexes a wife and Thomas, which are headed by the nominal
meanings 'a wife' and 'Thomas', are nouns. The syntactics' change undergone by
the input lexes a wife and Thomas can thus be regarded as a peripheral
consequence of the semantic change undergone by the input signifieds 'a wife'
and 'Thomas'.
Converted lexemes are usually products of a metonymic semantic change.
For example, the verbal lexemes TO WIFE and TO THOMAS are products of a part-
for-whole metonymic modification of the corresponding nominal lexemes A
WIFE and THOMAS. Thus the nominal concept of a wife is part of the verbal
concept of downplaying a woman's career accomplishments in favor of her
abilities as wife and mother. Similarly, the concept of the person called Clarence
Thomas was originally part of the concept of accusing a person of sexual
harassment. At the same time, notice that metaphoric conversion does exist as
well. Consider, for example, the converted verb to ape 'to imitate, mimic'
(OED). In contrast to both the output signifieds 'to wife' and 'to Thomas', the
output signified 'to ape' is a product of metaphorization of the input signified 'an
ape': there is only a perceived similarity between an ape and a person who apes.
As regards the semantic outcome of conversion, most converted lexemes can
be analyzed as quasi-idioms in relation to their corresponding input lexemes, i.e.
the signifieds of the former contain the signifieds inherent in the latter plus some
additional idiomatic signifieds. For example, the output lexeme TO WIFE is a
quasi-idiom in relation to the input lexeme A WIFE: the signified 'to wife'
116 Chapter 5
contains the signified 'a wife', inherent in the component wife, plus the idiomatic
signified 'to downplay a woman's career accomplishments in favor of her
abilities as mother'. The same applies to the converted lexemes TO BACKGROUND
'to surreptitiously perform a task in the background while one's attention is
supposed to be on another task', TO OFFICE 'to perform office-related tasks, such
as photocopying and faxing', and TO PIE 'to hit a person, particularly a political
or business leader, in the face with a pie', which were mentioned in 4.2.8.
Fully-idiomatic converted lexemes are usually instances of a metaphoric
semantic change like TO APE as well as metonymies like TO THOMAS whose
input lexemes are the names of people, places, institutions, etc. who / which
played some role in what later came to be referred to by means of their
converted output lexemes. For example, the converted lexeme TO DELL 'to beat a
competitor by eliminating the middleman and selling directly to customers'
(Word Spy) is, like TO THOMAS, a full-idiom: its signified does not contain the
signified of the input lexeme DELL. The latter only motivates the former:
according to Word Spy, TO DELL
comes from Dell Computers, which used direct sales to become one of
the top computer manufacturers in the world, with current sales of
US$18 billion. (Word Spy)
From a formal point of view, converted lexemes are usually classified with
regard to the direction of conversion. For example:
x The output verbal lexemes TO APE, TO BACKGROUND, TO DELL, TO OFFICE, TO

PIE, TO THOMAS, and TO WIFE, whose input lexemes are the nominal lexemes
AN APE, A BACKGROUND, DELL, AN OFFICE, A PIE, THOMAS, and A WIFE, are
instances of noun Æ verb conversion.
x The output verbal lexeme TO ELDER 'to share wisdom and knowledge with
people who are younger than oneself' (Word Spy), whose input lexeme is the
adjectival lexeme ELDER, is an instance of adjective Æ verb conversion.
x The output verbal lexeme TO DOWN 'to bring, put, throw, or knock down'
(OED), whose input lexeme is the adverbial lexeme DOWN, is an instance of
adverb Æ verb conversion.
x The output nominal lexeme A SHOUT-OUT 'a mention, acknowledgement, or

greeting, especially one made over the radio or during a live performance; a
namecheck' (OED), whose input lexeme is the verbal lexeme TO SHOUT OUT,
is an instance of verb Æ noun conversion.
x The output nominal lexeme AN EX 'one who formerly occupied the position
or office denoted by the context; specially a former husband or wife' (OED),
which goes back to the prefix ex- of ex-husband and ex-wife, is an instance of
affix Æ noun conversion.
Of particular theoretical interest are adjective Æ noun conversion and noun Æ

adjective conversion. As regards the former, let us consider the nominal lexeme
A GOLD 'a medal awarded as the first prize in a competition: a gold medal'
(MWO). At first glance, it seems that this lexeme is a product of metonymy-
based quasi-idiomatization of the input adjectival lexeme GOLD. That is, the
output nominal signified 'a gold medal' contains the adjectival signified 'gold',
inherent in the input adjective gold, plus the idiomatic signified 'medal'. The
same seems to be true of the nominal lexeme A FINAL 'the last examination in a
course' (MWO), which is often realized by the plural lex finals: its signified
contains the adjectival signified 'final', inherent in the input adjective final, plus
the idiomatic signified 'exam in a course'. However, in spite of these facts,
neither A GOLD nor A FINAL can be regarded as products of adjective Æ noun
conversion. In the case of these lexemes, we are dealing not with a lexeme-
building conversion but with a lex-forming clipping of the semi-idiomatic NP a
gold medal and the quasi-idiomatic NP a final examination. That is, a gold Å a
gold medal and a final Å a final examination. This analysis is true not only
from a diachronic but also from a synchronic point of view. That is, present-day
English speakers still analyze a gold and a final as shorter versions of the input
NPs a gold medal and a final examination. Accordingly, cases like a final Å
final and a gold Å gold must be regarded as pseudo-conversion.
As regards noun Æ adjective conversion, consider so-called stone wall-
combinations. These are NPs in which the modifier position is filled by a noun:
e.g. a phone call, a police officer, a car accident, a movie star, etc. In these NPs,
the modifiers phone, police, car, and movie fill the syntactic slot that is typically
filled by adjectives. That is, adjectives typically function as modifiers of head
nouns in NPs. For example, a recent call, a kind officer, a horrible accident, a
popular star. Given this fact, a question arises as to whether nouns like phone,
police, car, and movie that fill the modifier position in NPs are not really nouns
but adjectives. If so, we will be justified in concluding that in English adjectival
lexemes can come into existence via noun Æ adjective conversion.
To answer this question, let us recall in which respects adjectives are
syntactically different from members of other word classes. As we observed in
2.2.3, adjectives function as either modifiers of head nouns in NPs (e.g. a
horrible accident) or as complements of predicator verbs in clauses like The
accident was horrible. Accordingly, to determine whether phone, police, car,
and movie of the NPs a phone call, a police officer, a car accident, and a movie
118 Chapter 5
star are indeed adjectives, we must try to place them in complement positions.
Consider, for example, (81), (82), (83), and (84).
(81) This call is phone* ('this call is being made with the help of a
telephone')
(82) This officer is police* ('this officer is a policeman')
(83) This accident is car* ('this accident involves a car')
(84) This star is movie* ('this person has become famous because of the
movies in which he or she played the main roles')
Given the obvious ungrammaticality of these invented sentences, we can

conclude that the modifiers phone, police, car, and movie of the NPs a phone
call, a police officer, a car accident, and a movie star are not converted
adjectives but nouns functioning as modifiers of the head nouns call, officer,
accident, and star.
However, it is important to emphasize that this does not apply to all stone
wall-combinations. Consider, for example, the NP a stone wall (which gave the
name to the stone wall-problem). According to MWO, in Present-day English
there is the adjectival lexeme STONE2 meaning 'of, relating to, or made of stone'.
That is, we can say This wall is stone referring to a wall made of stone. Usually,
the adjective stone is used metaphorically. For example, Cindy Lauper's 1992
song The World is Stone deals with the cruelty and the indifference of the
contemporary world. This metaphorical use is, however, possible because of the
literal adjectival meaning 'made of stone'. Accordingly, the output adjective
stone of the NP a stone wall can be regarded as a product of noun Æ adjective
conversion of the input noun stone. We are justified in claiming this not only
because stone can fill both the modifier and the complement position but also
because the output adjectival lexeme is a quasi-idiom in relation to the input
nominal lexeme: the adjectival signified 'of, relating to, or made of stone' does
not only contain the signified 'stone', inherent in the component stone, but also
the idiomatic signified 'of, relating to, or made of'. The adjective stone is thus
clearly a lex which realizes a new lexeme.
Finally, let us discuss the practical problem of determining the direction of
conversion. As an illustrative example, let us consider the converted verbal
lexeme TO KNIFE 'to use a knife to; to cut, strike, or stab with a knife' (OED).
How do we actually know that it is the nominal lexeme A KNIFE which served as
an input lexeme for TO KNIFE, not vice versa? As we will see below, this
question can be approached from both a diachronic and a synchronic
perspective.
As regards the former, we know (thanks to the OED) that the verb to knife
appeared in the English language much later than the noun a knife. While the
latter has been recorded since 1100, the former has been used only since circa
1885 (OED). However, this fact alone does not suffice to conclude that it was
the nominal lexeme A KNIFE which served as an input lexeme for the verbal
lexeme TO KNIFE. The main reason why the latter can be regarded as an instance
of conversion of the former is that the verbal meaning 'to use a knife to; to cut,
strike, or stab with a knife' is semantically more complex than the nominal
meaning 'a knife'. The verbal meaning is a quasi-idiom in relation to the nominal
meaning: the signified 'to knife' contains the signified 'a knife', inherent in the
component knife, plus the idiomatic signified 'to cut, strike, or stab'.
Accordingly, given the fact that to knife appeared later than a knife and given
that the signified 'to knife' is a quasi-idiom in relation to the signified 'a knife',
we are justified in claiming that it is the nominal lexeme A KNIFE which served
as an input lexeme for the verbal lexeme TO KNIFE, not vice versa.
Similarly, thanks to the OED, we know that the verb to ape appeared much
later (1634) than the noun an ape (700). However, as in the case of to knife, this
fact alone does not suffice to regard the verbal lexeme TO APE as a product of
conversion of the nominal lexeme AN APE. The main reason why we can do so is
that the output verbal signified 'to mimic' is a product of metaphorization of the
input nominal signified 'an ape': as stated above, there is a perceived similarity
between an ape (animal) and a person who apes.
Sometimes the diachronic analysis is not as straightforward as in the case of
TO KNIFE and TO APE. Consider, for example, the verbal vocable TO DOCTOR.
According to the OED, this vocable consists of the following six lexemes.
x TO DOCTOR1 'to confer the degree or title of Doctor upon; to make a Doctor'
(OED / 1599)
x TO DOCTOR2 'to treat, as a doctor or physician; to administer medicine or

medical treatment' (OED / 1740)
x TO DOCTOR3 'to repair, patch up, set to rights' (OED / 1829)
x TO DOCTOR4 'to castrate an animal'(OED / 1902)
x TO DOCTOR5 'to treat so as to alter the appearance, flavor, or character of; to

disguise, falsify, tamper with, adulterate, sophisticate' (OED / 1777)
x TO DOCTOR6 'to take medicine, undergo medical treatment' (OED / 1865)
Each of these lexemes appeared after the noun doctor, which has been
documented in English since the 14th century. The OED lists a number of
nominal signifieds that doctor could express at that time. For example:
120 Chapter 5
x 'a teacher, instructor; one who gives instruction in some branch of

knowledge, or inculcates opinions or principles' (OED / 1387)
x 'one who, by reason of his skill in any branch of knowledge, is competent to

teach it, or whose attainments entitle him to express an authoritative opinion;
an eminently learned man' (OED / 1400)
x 'a person who, in any faculty or branch of learning, has attained to the highest
degree conferred by a University; a title originally implying competency to
teach such subject or subjects, but now merely regarded as a certificate of the
highest proficiency therein' (OED / 1377)
x 'a doctor of medicine; in popular current use, applied to any medical

practitioner. Also: (amongst indigenous peoples) a traditional healer or
diviner, esp. one dealing with afflictions thought to be caused by spirit
possession or witchcraft' (OED / 1377)
However, despite the fact that the noun doctor appeared before the verb to
doctor, we cannot claim that each of the six verbal TO DOCTOR lexemes are
instances of noun Æ verb conversion. Let us begin with the lexeme TO DOCTOR1
'to confer the degree or title of Doctor upon; to make a Doctor' (OED / 1599).
This verbal lexeme is a quasi-idiom in relation to the nominal lexeme DOCTOR1 'a
person who has earned the degree of a doctor' (OED / 1377): the signified of TO
DOCTOR1 contains the signified 'a doctor', inherent in the component doctor, plus
the idiomatic signified 'to confer this degree upon somebody'. Accordingly,
given the quasi-idiomatic character of the relation between the two lexemes
under consideration and given that the verbal lexeme appeared after the
corresponding nominal lexeme, we are justified in concluding that TO DOCTOR1 is
an instance of noun Æ verb conversion of DOCTOR1.
A similar analysis applies to the lexeme TO DOCTOR2 'to treat, as a doctor or
physician; to administer medicine or medical treatment' (OED / 1740). This
lexeme is a quasi-idiom in relation to the nominal lexeme DOCTOR2 'a medical
practitioner' (OED / 1377). The signified of TO DOCTOR2 contains the signified 'a
doctor', inherent in the component doctor, plus the idiomatic signified 'to do the
job of'. Accordingly, given this fact and given that the verbal lexeme appeared
later than the corresponding nominal lexeme, we are justified in concluding that
TO DOCTOR2 is also a product of noun Æ verb conversion of the nominal lexeme
DOCTOR2.
But what about the lexeme TO DOCTOR3 'to repair, patch up, set to rights'
(OED / 1829)? In contrast to the lexemes discussed above, it appears that TO
DOCTOR3 is not an instance of noun Æ verb conversion of any of the nominal
DOCTOR lexemes but a product of part-for-whole metonymization of the verbal
lexeme TO DOCTOR2 'to treat, as a doctor or physician; to administer medicine or

medical treatment' (OED / 1740). Thus the signified 'to treat as a doctor or
physician' seems to be an instance of the signified 'to repair'. That is, giving a
medical treatment to somebody often results in the repairing of that person's
health.
Similarly, the lexemes TO DOCTOR4 'to castrate an animal' (OED / 1902) and
especially TO DOCTOR6 'to take medicine' (OED / 1865) seem to be products of a
metonymic semantic change of the already existing verbal lexeme TO DOCTOR2:
castrating an animal can be seen as an instance of giving medical treatment to
that animal and taking medicine can be seen as giving medical treatment to
oneself. Hence the lexemes TO DOCTOR4 and TO DOCTOR6 can be regarded as
products of whole-for-part metonymization of the lexeme TO DOCTOR2.
Finally, let us consider the lexeme TO DOCTOR5 'to falsify' (OED / 1777). The
etymology of this lexeme is not entirely clear. On the one hand, it may have
been the case that the signifier to doctor acquired this meaning because of some
people's belief that doctors (be it holders of doctoral degrees or practitioners of
medicine) often falsify things. In this case, TO DOCTOR5 can be regarded as an
instance of noun Æ verb conversion of any of the nominal DOCTOR lexemes. On
the other hand, however, it can be conjectured that any instance of doctoring (be
it conferring the degree of a doctor upon somebody or giving medical treatment
to somebody) involves falsifications. In this case, TO DOCTOR5 can be regarded as
a product of metonymy-based semantic change of one of the previously formed
TO DOCTOR lexemes.
In summary, of the six verbal TO DOCTOR lexemes, only two can be
considered products of morphological conversion of two different nominal
DOCTOR lexemes, whereas the other four lexemes are instances of a metonymic
semantic change (not involving a change of word class) of the previously
formed verbal TO DOCTOR lexemes.
As a rule, the diachronic history of the members of a pair like A KNIFE–TO
KNIFE confirms the intuitions of present-day speakers who do not have the
diachronic memory (i.e. do not remember the true etymologies of the lexemes in
question). Indeed, in Present-day English we still have the noun–verb pair A
KNIFE–TO KNIFE, in which the verbal lexeme can be analyzed as a quasi-idiom in
relation to the corresponding nominal lexeme: the signified 'to knife' contains
the signified 'a knife', inherent in the component knife, plus the idiomatic
signified 'to use a knife in a particular way: to cut, strike, or stab with a knife'.
Accordingly, even without looking up the diachronic history of the lexemes A
KNIFE and TO KNIFE in a dictionary like the OED, we can claim that the verbal
lexeme TO KNIFE is, from a synchronic point of a view, a product of conversion
of the nominal lexeme A KNIFE. Likewise, we can claim that the verbal lexeme
TO APE is, from a synchronic point of a view, a product of conversion of the
122 Chapter 5
nominal lexeme AN APE: the verbal signified 'to ape' can still be analyzed as an
instance of metaphorization of the nominal signified 'an ape'.
However, as will be shown below, in a number of cases the diachronic
history is at odds with present-day speakers' intuitions. Consider, for example,
the lexemes AN ANSWER and TO ANSWER. From a synchronic point of view, we
seem to be dealing with a conversion pair one of whose members can be
analyzed as a quasi-idiom in relation to the other member: an answer seems to
mean 'the product of the process of answering'. However, the diachronic history
of the lexemes under analysis does not corroborate this conjecture. According to
the OED, the nominal lexeme AN ANSWER originally (i.e. in the Old English
period) was realized by the lex andswaru, whereas the verbal lexeme TO
ANSWER was originally realized by the lex andswarian. In other words, the
lexemes AN ANSWER and TO ANSWER originally had different lexes: andswaru
and andswarian had the same root andswar but different derivational affixes: -u
and -ian. (With the course of time, andswaru and andswarian 'got rid' of these
suffixes, so that in Present-day English both the nominal lexeme AN ANSWER and
the verbal lexeme TO ANSWER are realized by the same lex answer.) A very
similar case is the lexeme pair LOVE–TO LOVE, whose members were originally
realized by the non-identical lexes lufu (love) and lufian (to love). Again, as in
the case of AN ANSWER and TO ANSWER, the pair LOVE–TO LOVE can be regarded
as a conversion pair only from a synchronic but not from a diachronic point of
view. (From a diachronic perspective, the verbal lexes andswarian and lufian
can perhaps be regarded as output lexes which came into existence via affixation
of the input nominal lexes andswaru and lufu by means of the derivational suffix
-ian.)
A number of studies provide additional synchronic criteria for determining
the direction in conversion. For example, Ginzburg et al. (1976: 133-134)
mention the synonymic sets criterion. As they argue, to determine the direction
of conversion in the pairs A CHAT–TO CHAT and A SHOW–TO SHOW, we need to
consider the near-synonymic pairs A CONVERSATION–TO CONVERSE and AN
EXHIBITION–TO EXHIBIT, in which the lexes realizing the nominal lexemes can be
synchronically analyzed as products of affixation of the lexes realizing the
corresponding verbal lexemes: a conversation = to converse + -ation and an
exhibition = to exhibit + -ion.11 Accordingly, given the derived character of the
nominal lexemes A CONVERSATION and AN EXHIBITION and given a semantic
similarity between the signifieds 'conversation' / 'chat' and 'exhibition' / 'show', it
can be conjectured that in the conversion pairs A CHAT–TO CHAT and A SHOW–TO
SHOW, it is the nominal lexemes A CHAT and A SHOW that are products of
conversion of the corresponding verbal lexemes TO CHAT and TO SHOW, not vice
versa.
11
Note that this analysis is not true from a diachronic point of view: According to the OED,
both conversation and exhibition were borrowed into English from Old French.
Another often mentioned criterion is the frequency criterion (Ginzburg et

al. 1976: 135-136; Plag 2003: 111). As is well-known, output lexemes that came
into existence via derivation and conversion are often characterized by a lower
frequency of use than corresponding input lexemes. The reason for this is that
the signifieds of output lexemes (especially those that came into existence via
quasi-idiomatization) are more complex than those of corresponding input
lexemes. As a result of this semantic complexity, the former cannot be used as
freely as the latter: the use of semantically more complex output lexemes is
constrained by additional meanings which they have acquired as a result of their
quasi-idiomatization. What follows from this is that the direction of conversion
can be determined by comparing the frequencies of use of the members of a
conversion pair. For example, if you use COCA or BYU-BNC, this can be done
in the following way. Just enter the combination [chat].[v*] to the search mask
and both corpora will yield all verbal occurrences of the verb to chat, including
the wordforms chatting, chats, chatted. If you enter [chat].[n*], both corpora
will yield all nominal occurrences of the noun chat, including the plural
wordform chats. At present (i.e. as of November 28, 2011), COCA finds 3071
occurrences of the nominal lexeme A CHAT and 5201 occurrences of the verbal
lexeme TO CHAT. The lower frequency of a chat testifies to the analysis of this
noun as a product of verb Æ noun conversion of the verb to chat. As regards a
show and to show, COCA yields 225869 occurrences of the verb to show but
only 96566 occurrences of the noun a show. Again, this can be seen as a
corroboration of the analysis of the noun a show as a product of conversion of
the verb to show.
Both of these criteria are problematic. As regards the synonymic sets
criterion, its major shortcoming acknowledged by Ginzburg et al. is its
somewhat limited applicability: it is not always possible to find a synonymic
pair like to converse–a conversation one of whose members can be analyzed as
a product of affixation of the other member.
Similarly, the frequency criterion is of very little help when we have to deal
with polysemous lexemes. As an illustrative example, let us again consider the
verb to doctor. Using COCA, we can perhaps prove that this verb came into
existence via noun Æ verb conversion of the noun doctor: while a search for
[doctor].[v*] yields only 56 results, a similar search for [doctor].[n*] yields
75565 results. However, the only thing these numbers tell us is that some of the
nominal DOCTOR lexemes must have given rise to some of the six verbal TO
DOCTOR lexemes. This, however, is an incomplete analysis: as we have
established above, of the six TO DOCTOR lexemes, no more than two can be
regarded as instances of conversion.
But even if it were possible to compare the frequencies of different TO
DOCTOR lexemes with those of the corresponding nominal lexemes, the
frequency criterion would still be of fairly marginal significance. Thus the only
124 Chapter 5
reason why the noun chat can be regarded as a product of conversion of the verb
to chat is that the signified of the nominal lexeme represents a quasi-idiom in
relation to the signified of the corresponding verbal lexeme: a chat is the product
of chatting. If you do not analyze the signified 'a chat' as being semantically
more complex than the signified 'to chat', you cannot claim that the former came
into existence via conversion of the latter even if you can show that a chat has a
lower frequency of use than to chat.
To conclude: the most important criterion for determining the direction of
conversion from both diachronic and synchronic perspectives is the semantic
criterion. In the case of converted lexemes that came into existence via
metonymy-based quasi-idiomatization, the signified of a converted output
lexeme is semantically more complex than the signified of a corresponding input
lexeme. Similarly, in the case of converted lexemes which came into existence
via metaphor-based full-idiomatization, the signified of an output lexeme can be
analyzed as a product of metaphorization of the signified of a corresponding
input lexeme. No other criterion justifies the treatment of the lexeme under
analysis as an instance of morphological conversion of some other lexeme in
question.
5.1.5 Productivity
Semantic change is the most productive lexeme-building mechanism.12 People

are constantly changing the meanings of already existing lexemes in an attempt
to produce an impression on other people (i.e. being motivated by the wish to
achieve success). Compared to other lexeme-building mechanisms, semantic
modification requires the least effort on the part of a language user, i.e. he or she
only has to modify the signified without modifying or inventing the signifier.
The claim that semantic change is more productive than other mechanisms
can be corroborated by the structure of English vocables. Most of them consist
of multiple polysemous lexemes. Recall that, for example, the vocable TRUE
consists of the polysemous lexemes TRUE1 'faithful', TRUE2 'not false', TRUE3
'properly so called', TRUE4 'legitimate, rightful', TRUE5 'narrow, strict', etc. One-
lexeme vocables can be found as well, but they are relatively rare. The reason
for this is the above mentioned fact that language users, motivated by the wish
to produce an impression on other people, are constantly changing the signifieds
of already existing lexemes, thereby creating new lexemes.
As regards morphological conversion, it is the noun Æ verb pattern,
exemplified by AN APE Æ TO APE, A BACKGROUND Æ TO BACKGROUND, A KNIFE
Æ TO KNIFE, etc., which is more productive than all other patterns named in
12
I owe the understanding of this to Dieter Stein.
5.1.4. As, for example, Ginzburg et al. (1976: 138) point out, "the possibility for
the verbs to be formed from nouns through conversion seems to be illimitable".
The explanation for this is the fact that in Present-day English there are only two
derivational suffixes that can be used for deriving verbs from nouns. These are
the suffix -ify of e.g. personify and the suffix -ize of e.g. actualize. Both suffixes
are still productive nowadays; however, as Marchand (1969: 364) notes, they
have "restricted range of derivative force": "-ify is learned while -ize is chiefly
technical". Indeed, of 55 verbal lexemes that appeared in the English language
between 1990 and 2011, we find only one instance of -ify affixation – TO
MATTIFY 'of a cosmetic: to produce a matt appearance; to reduce the shiny
appearance of the skin, esp. on the face' (OED / 1997)13 – and only four
instances of -ize affixation:
x TO CLINTONIZE 'to modify in accordance with or as a result of the policies of

President Clinton' (OED / 1992)
x TO FERBERIZE 'to use the Ferber method or a similar technique to train (a

child) to fall asleep independently' (OED / 1990)
x TO MODULIZE 'to render modular; to modularize' (OED / 1992)
x TO TALIBANIZE 'to treat in a manner associated with the Taliban […]' (OED /
1997)
By contrast, 29 verbal lexemes (of the 55 verbal lexemes which appeared during
the last two decades) are products of noun Æ verb conversion. Consider, for
example, the following lexemes.
x TO BITCH-SLAP 'to deliver a stinging slap to (a person), especially in order to

humiliate one regarded as inferior' (OED / 1991)
x TO BOTOX 'to treat with Botox' (OED / 1994)
x TO DOTCOM 'to be overwhelmed or driven out of existence by pressure from

companies doing business over the Internet' (OED / 1996)
x TO GOOGLE 'to use the Google search engine to find information on the
Internet' (OED / 1999)
13
Note that to mattify is a product of affixation of the input adjective matt 'of a surface, finish,
etc.: without lustre, dull; unpolished' (OED) by means of the suffix -ify.
126 Chapter 5
x TO HOT-DESK 'to practice hot-desking; to work at any of a number of different

desks or workstations on a temporary, ad hoc, or part-time basis, rather than
routinely occupying one desk permanently; to allocate workspace in this way'
(OED / 1994)
x TO NINJA 'to act or move in a manner similar to a ninja' (OED / 1992)
x TO PEPPER-SPRAY 'to douse with pepper spray' (OED / 1993)
x TO TEXT MESSAGE 'to send or enter as a text message; to send a text message
to (a person)' (OED / 1994)
x TO TWOC 'to steal (a car), especially for the purpose of joy-riding' (OED /
1992)
Noun Æ verb conversion can thus be regarded as the most productive verb-
forming mechanism in Present-day English.
Unfortunately, the Etymology-sections of most OED entries do not contain
the word conversion. For example, the Etymology-section of the entry for TO
BITCH-SLAP, which, as stated above, came into existence via noun Æ verb
conversion of the nominal lexeme A BITCH SLAP, contains the following
information: 'Etymology: < bitch slap n.' (OED). This information enables us to
recognize that the nominal lexeme A BITCH SLAP served as an input lexeme for
the verbal lexeme TO BITCH SLAP. However, since the direction sign < also
occurs in the Etymology-sections of the lexemes that came into existence via
other mechanisms (e.g. the Etymology-section of the affixed verb to modulize
contains the information 'Etymology: < module n. + -ize suffix'), this sign is not
very helpful for making the OED searchable for instances of conversion. That is,
there does not seem to be a way of making the OED specifically search for
converted lexemes. This is why the best means to study the productivity of
various conversion patterns is to ask the OED to yield all lexemes belonging to a
particular word class that were formed during a particular period of time. For
instance, if you want to find out whether verb Æ noun conversion has recently
created more / less nominal lexemes than a particular noun-building affix, go to
the OED Advanced Search page http://www.oed.com/advancedsearch, enter
'1990-' (without quotation marks) to 'Date of entry', and choose the option 'Noun'
in 'Part of speech'. The OED will then yield all nominal lexemes which appeared
in the English language between 1990 and 2011. Then carefully read the
Etymology-sections of all found lexemes and assign them to the categories
'affixed nominal lexemes' and 'converted nominal lexemes'.
If you are interested in other recent instances of semantic change (i.e. those
not involving a change of word class), go again to the above named OED
Advanced Search page http://www.oed.com/advancedsearch, enter e.g. '2010'

(without quotation marks) to the first row of the search mask from above, and
choose the option 'Quotation date' (near the first row of the search mask to
which you will enter '2010'). The OED will then yield all entries containing the
year 2010. One of such entries could be a vocable one of whose lexemes has
recently undergone semantic change, thereby giving rise to a new lexeme which
has been registered in English since the year 2010. Of course, you can refine
your search in a variety of ways. For example, if you choose the option 'Noun' in
'Part of speech', the OED will yield nominal lexemes. If you choose the option
'Colloquial' in 'Usage', the OED will yield informal lexemes.
5.2 Lexeme-manufacturing
Lexeme-manufacturing or arbitrary formation is the formation of a new lexeme

by means of creating an unmotivated signifier for expressing a previously
unexpressed signified. For example, as we pointed out in 4.2.2, the unmotivated
proprietary names Viagra and Gonk were created to express the previously
unexpressed signifieds 'a new impotence drug' and 'an egg-shaped doll'.
The most important theoretical question raised by this lexeme-building
mechanism is whether any combination of sounds can become a lex realizing a
new lexeme. That is, for example, does it make a difference whether the creator
of an arbitrary lex chooses the signifier plafe or lpafe? Or more generally: which
of these signifiers has more chances to become a lex of a new lexeme?
The answer to these questions is provided by the well-known sonority
sequencing generalization (McMahon 2002: 107; Yavaş 2006: 127-151). In
phonetics and phonology, the term 'sonority' refers to the loudness of a sound
(compared to other sounds.) The two main factors determining the sonority of
the sound are the degree of opening of the vocal tract and voicing.
As regards the former, a greater opening of the vocal tract results in a greater
sonority. Thus vowels are acoustically much louder than consonants because
they are articulated in such a way that the airflow coming from the lungs can
freely leave the mouth cavity without encountering any obstruction. In contrast,
the articulation of a consonant involves a major obstruction to the airflow. For
example, in the case of the plosive [p], there is a complete obstruction formed
by the upper and the lower lip, which are firmly pressed against each other: the
airflow can leave the mouth cavity only when this obstruction is removed.
(When the obstruction is removed, we hear a kind of explosion. Hence the term
'plosive'.)
The term 'voicing' refers to the vibration of the vocal folds (sometimes also
referred to as vocal cords). When they are vibrating, a voiced sound is
128 Chapter 5
produced. Vowels are always voiced, whereas consonants can be both voiced
and voiceless. For example, while [b] is voiced, [p] is voiceless.
The sonority sequencing generalization is concerned with the question of
which combinations of sounds qualify as permissible syllables. A syllable is a
phonetic unit that often consists of the following three structural components:
1. the onset
2. the nucleus
3. the coda
The nucleus is the most sonorous element of a syllable. Since vowels are more
sonorous than consonants, the nucleus position is typically filled by vowels. For
example, in the putative word plafe only the diphthong /eɪ/ can be the nucleus.
The nucleus is the only obligatory element of a syllable. Thus there are syllables
consisting of nuclei only. For example, /ʌɪ/ of I.
The onset is a consonant or a consonant cluster occurring before the nucleus.
For example, the cluster [pl] is the onset of the syllable /pleɪf/.
Finally, the coda is a consonant or a consonant cluster which occurs after the
nucleus. For example, the consonant [f] is the coda of the syllable /pleɪf/.
According to the sonority sequencing generalization, onsets that consist of
more than one consonant are characterized by an increasing sonority. That is, the
following sound has a higher sonority than the preceding one. For example, the
double onset [pl] is a permissible onset because the following sound [l] is more
sonorous than [p]. Both [p] and [l] involve a complete obstruction to the airflow.
However, [l] is a lateral sound. This means that despite the obstruction formed
by the raising of the tip of the tongue towards the alveolar ridge, the airflow
coming from the lungs can escape from the mouth cavity along the sides of the
tongue. By contrast, as stated above, in the case of the plosive [p], the airflow
can leave the mouth cavity only when the obstruction formed by the lower and
the upper lip is removed. In addition to this, the lateral [l] is a voiced sound,
whereas the plosive [p] is a voiceless one.
That [pl] is indeed a permissible double onset in English is corroborated by
the existence of numerous signifiers beginning with this cluster. For example,
plate /pleɪt/, plan /plan/, plum /plʌm/.
However, there are no signifiers beginning with [lp]. E.g. *lpate, *lpan,
*lpum, etc. This consonant cluster is a non-permissible double onset because it
does not fulfill the sonority sequencing generalization: the lateral [l] has a
greater sonority than the plosive [p].
With regard to codas consisting of more than one consonant, the sonority
sequencing generalization states that such codas must be characterized by a
decreasing sonority. That is, the following consonant must have a lower sonority
than the preceding one. This explains, for example, why English has the
signifier help /hɛlp/, but not *hepl /hɛpl/. The cluster [lp] is a permissible double
coda because the following plosive [p] is less sonorous than the preceding lateral
[l]. By contrast, [pl] is a non-permissible double coda because it is characterized
by an increasing sonority.
Taking all this into account, we can now conclude that only plafe, but not
lpafe, can become an arbitrary lex of a new lexeme: in plafe the cluster [pl] is a
permissible double onset, the diphthong /eɪ/ is a permissible nucleus, and the
consonant [f] is a permissible single coda; e.g. chief /tʃiːf/, roof /ruːf/, etc.
5.2.1 Productivity
Lexeme-manufacturing requires a considerable effort on the part of a language

user: he or she has to create a totally unmotivated signifier that has not been
used before. Precisely because of this fact, lexeme-manufacturing is the least
productive lexeme-building mechanism: as e.g. Algeo (1998: 66) points out,
arbitrary formations like plafe "are extremely rare, if they exist at all". Indeed,
according to the OED, of 587 new lexemes which appeared in the English
language between 1990 and 2011, only three can be regarded as instances of
lexeme-manufacturing. Apart from the already mentioned proprietary name
Viagra, these include:
x NEVIRAPINE 'a drug structurally related to the benzodiazepines which inhibits

the reverse transcriptase of HIV-1' (OED / 1991)
x ZORB 'a large, transparent, inflatable PVC ball used in the sport of zorbing,
containing an inner capsule into which a participant is secured and then
rolled along the ground, down hills, etc.' (OED / 1996)
Note that as in the case of Viagra, the OED is not entirely sure whether
nevirapine and Zorb are indeed arbitrary formations. Thus nevirapine is,
according to the OED, "apparently an arbitrary formation, probably including
the elements vir- (in virus n.) and -pine (compare benzodiazepine n.)". Similarly,
Zorb is, according to the OED, "apparently an arbitrary formation", there being a
possibility that its formation was influenced by orb 'each of the concentric
hollow spheres formerly believed to surround the earth and carry the planets
[…]' (OED / circa 1449).
Earlier examples of arbitrary formations include:
x GOOP 'a stupid or fatuous person' (OED / 1900)

130 Chapter 5
x TO GROK 'to understand intuitively or by empathy; to establish rapport with'

(OED / 1961)
x MITHRIL 'In the fiction of J. R. R. Tolkien […]: a rare silver-colored precious

metal of great hardness and beauty' (OED / 1944)
x NERF 'a type of foam rubber used esp. in the manufacture of children's toys
and sports equipment' (OED / 1970)
x SUKEBIND 'name given by Stella Gibbons […] to an imaginary plant

associated with superstition and fertility, hence used allusively with reference
to intense rustic passions' (OED / 1932)
To study arbitrary formations in English, go to the above named OED Advanced

Search page http://www.oed.com/advancedsearch, enter 'arbitrary formation'
(without quotation marks) to the first row of the search mask from above, and
choose the option 'Etymology' (near the search mask to which you will enter
'arbitrary formation'). The OED will then yield all lexemes that have ever been
coined via lexeme-manufacturing. Of course, you can refine your search by
entering e.g. '1900-' to 'Date of entry'. In this case, the OED will yield all
arbitrary formations that appeared in the English language between the years
1900 and 2011.
5.3 Lexeme-building borrowing
Lexeme-building borrowing is the importation of a foreign language signifier

that gives rise to a new lexeme. As we established in 4.2.2 and 4.2.3, there are
two kinds of lexeme-building borrowing:
x the borrowing of both the signifier and the signified of a foreign language
lexeme: e.g. the borrowing of kindergarten from German.
x the borrowing of a foreign language signifier accompanied by a semantic

modification of its (foreign language) input signified: e.g. the borrowing of
wiki from Hawaiian.
Sometimes it is argued that loan-translations (also known as semantic calques)

must be regarded as a distinct type of borrowing. Consider, for example, the
lexeme SUPERMAN 'an ideal superior man conceived by Nietzsche as being
evolved from the normal human type; loosely, a man of extraordinary power
[…]' (OED / 1903). According to the OED, the signifier superman is a product
of affixation of man by means of the derivational prefix super-. However, what

is interesting here is that this instance of affixation was inspired by the German
signifier Übermensch, popularized by Friedrich Nietzsche's famous work Thus
spoke Zarathustra. In other words, speakers of English did not borrow the
German Übermensch into English. What they did was translate the German
components über and Mensch as super- and man, thereby giving rise to
superman. The English superman is thus a loan-translation of the German
Übermensch.
Undeniably, the creation of the English lexeme SUPERMAN was inspired by
the German lexeme ÜBERMENSCH. However, the formation of the English
signifier superman did not involve the borrowing of any German signifier: as
said above, the lex under analysis is a product of affixation of man by means of
the prefix super-. Accordingly, the lexeme SUPERMAN cannot be regarded as an
instance of borrowing; the lexeme under analysis is an instance of prefixation,
just as the lexeme UNSACKABLE, whose creation in 1980 was not inspired by any
foreign language signifier.
5.3.1 Productivity
Throughout its history, English has extensively borrowed from other languages.
Below are just some of the most famous examples.
Latin loanwords in English:

altar, balsam, calculator, debit, editor, to fabricate, to generate, habitual,
icon, janitor, kinesis, laboratory, magnet, to narrate, to obliterate, pagan, to
quote, radiation, salient, tact, ubiquity, vacuum, xenia, zeal, etc.
Greek loanwords in English:

aerodrome, bronchotomy, calligraphy, deixis, ellipse, glaucoma, hapax
legomenon, iris, laconic, macron, narcosis, oligarch, pathos, strophe, telic,
xenon, etc.
French loanwords in English:

to abandon, bachelor, to commence, to damage, economy, fable, garage,
harmony, to identify, jewel, kilometer, lechery, machine, naïve, obedience, to
paint, quality, rage, to sacrifice, uncle, vacation, zenith, etc.
Italian loanwords in English:

accordion, balcony, cantata, fascism, gambit, impresario, latte macchiato,
mafia, novella, opera, paparazzo, quartet, ravioli, salami, tarantella,
umbrella, vendetta, zucchini, etc.
132 Chapter 5
Spanish loanwords in English:

albino, banana, cargo, demarcation, El Dorado, fiesta, gringo, habanera,
iguana, Latino, macho, Negro, paella, Quixote, sombrero, tequila, vanilla,
etc.
German loanwords in English:

ablaut, bratwurst, delicatessen, ersatz, festschrift, gestalt, kindergarten,
leitmotiv, Nazi, Ossi, panzer, quartz, Realpolitik, Schadenfreude, Vaseline,
Zeitgeist, etc.
Japanese loanwords in English:

aikido, bushido, geisha, haiku, judo, kamikaze, manga, Nikkei, samurai,
Tamagotchi, yakuza, etc.
At present, borrowing still remains a fairly productive lexeme-building

mechanism in English: of 29 new lexemes which, according to the OED,
appeared in the English language between 2000 and 2011, seven were borrowed
from other languages. These include:
x GALACTICO 'a skilled and celebrated footballer, especially one bought by a

team for a very large fee; a football superstar' (OED / borrowed from Spanish
in 2003)
x GOJI 'the edible bright red berry of either of two species of wolfberry, Lycium
barbarum and L. chinense, widely cultivated in China and supposed to
contain high levels of certain vitamins' (OED / borrowed from Chinese in
2002)
x HAWALADAR 'a person who acts as an agent in a hawala transaction; a hawala

dealer' (OED / borrowed from Persian in 2000)
x NOROVIRUS 'a genus of caliciviruses comprising the Norwalk virus, which

causes epidemic gastroenteritis in humans, and closely related caliciviruses'
(OED / borrowed from scientific Latin in 2002)
x PARKOUR 'the discipline or activity of moving rapidly and freely over or

around the obstacles presented by an (esp. urban) environment by running,
jumping, climbing' (OED / borrowed from French in 2002)
x REGGAETON 'a form of dance music of Puerto Rican origin incorporating

Latin rhythms and sounds alongside elements of dancehall reggae, hip-hop,
and rap music' (OED / borrowed from Spanish in 2001)
x SUDOKU 'a type of logic puzzle, the object of which is to fill a grid of nine
squares by nine squares (subdivided into nine regions of three by three
squares) with the numbers one to nine, in such a way that every number
appears only once in each horizontal line, vertical line, and three-by-three
subdivision' (OED / borrowed from Japanese in 2000)
The OED is an excellent tool for studying not only the synchronic productivity
but also the history of borrowing in English. Thus if you go to the Advanced
Search page http://www.oed.com/advancedsearch, you can ask the OED to find
all words that were borrowed from a particular language. Just enter e.g. 'French'
to 'Language of origin' and the OED will yield all words that have ever been
borrowed from French. If you enter e.g. '1900-', the OED will yield all French
words which appeared in the English language between 1900 and 2011.
5.4 Lexeme-building affixation
Affixation is a major formal lexeme-building mechanism. It is different from

lexeme-manufacturing and borrowing in that it involves a formal modification
of the lex of an already existing lexeme. For instance, the lex of the existing
lexeme TRUE was modified by means of the prefix un-, which gave rise to the
new lexeme UNTRUE 'not faithful'.
To begin with, let us briefly recapitulate what we have already learned about
affixes and affixation in the previous chapters.
x An affix is a bound morph or a bound quasi-linguistic unit.
x Affixes are shorter than roots.
x The syntactics of an affix determines its combinatory possibilities.
x Affixes can be classified into prefixes, suffixes, interfixes, and infixes.
x Affixation can be classified into isomorphic and anisomorphic affixation.
5.4.1 Affixes and their signifieds
Affixes can express a number of signifieds. For example:
x 'performer of some action': e.g. -er of blogger, -ist of tourist

x 'nationality': e.g. -an of Belizean, -ese of Surinamese
134 Chapter 5
x 'gender': e.g. -ess of princess, -ine of heroine

x 'collectivity': e.g. -dom of Arabdom, -ry of computery
x 'diminutiveness': e.g. -ie of mushie 'mushroom', -let of applet
x 'location in space and time': e.g. inter- of intercommunal, post- of post-
minimalist
x 'pejoration': e.g. mis- of miscommunicate, mal- of malware
x 'reversal': e.g. un- of unsubscribe, de- of demotivate
x 'negation': e.g. un- of untrue, in- of inadequate
As in the case of roots, one and the same affix can be associated with more than
one signified. Compare, for instance, the meanings inherent in the suffix -er in
blogger and Londoner. In the former case, the suffix under analysis denotes a
habitual performer of an action: a blogger is a person who blogs (on a more or
less habitual basis). By contrast, -er of Londoner denotes an inhabitant or a
native of a particular place: a Londoner is an inhabitant of London. Similarly, a
New Yorker is an inhabitant or a native of New York and a Berliner is an
inhabitant or a native of Berlin. Clearly, the signifieds 'performer of some action'
and 'inhabitant of a particular place' are related signifieds: an inhabitant of
London can be defined as a performer of the action of living in London.
Nonetheless, the signifieds under analysis are not identical signifieds, so that the
suffixes -er of blogger and -er of Londoner must be regarded as realizations of
two different morphemes. (Analogous to uniting polysemous lexemes like TRUE1
'faithful' and TRUE2 'not false' into lexeme vocables, we can perhaps unite
polysemous affixes like -er of blogger and -er of Londoner into affix vocables.)
5.4.2 Affixes and their syntactics
The syntactics of an affix contains two distinct kinds of information:
x what we can do with the affix: i.e. whether we can use the affix for new
derivational formations, whether we can attach the affix to particular input
signifiers, whether the affix can occupy a particular position in an output
signifier.
x what affixation will result in: i.e. whether the addition of the affix will
trigger a change of word class or whether it will impose phonetic changes on
the signifier of an input lexeme.
In the following let us discuss each of these aspects. As far as the first category
is concerned, we already know that the syntactics determines the combinatory
properties of an affix. For example, it is the syntactics that makes the negative
prefix un- attachable to Germanic and fully naturalized input lexes like true, but
not to Latinate lexes like adequate.
A similar property concerns the ordering of an affix. As Bauer and
Huddleston (2002: 1669) point out, the suffix -hood can be preceded but not
followed by another affix: there can be a signifier like magicianhood but not
*childhoodic. The latter is ungrammatical because the syntactics of -hood does
not allow it to be followed by another affix.
The syntactics of an affix also determines whether it can be attached to input
lexes that are members of a particular word class. As a first approximation, it
may seem that this property of an affix is largely determined by its signified.
Indeed, if it were not for the signified 'performer of some action', we would not
be able to use the suffix -er for deriving deverbal nouns like teacher, preacher,
worker, etc. denoting performers of the actions expressed by corresponding
input verbs.
But consider the suffix -eer of nouns like auctioneer 'one who conducts sales
by auction' (OED / 1708) and sonneteer 'a composer of sonnets' (OED / 1667).
According to the OED, the suffix -eer carries the meaning 'one who is concerned
with, one who deals in'. This suffix is still productive in Present-day English.
According to Word Spy, the most recent neologisms involving this suffix are:
x PUMPKINEER 'a person who grows giant pumpkins, particularly ones meant to
be entered in pumpkin-weighing contests' (Word Spy / 1988)
x FUSIONEER 'a person who investigates nuclear fusion or attempts to build a

nuclear fusion reactor' (Word Spy / 1988).
According to the OED, the most recent -eer formations are:
x FREE MARKETEER 'a proponent of, or believer in, the free market' (OED /
1963)
x SUMMITEER 'one who takes part in summit meetings' (OED / 1957)
x WEAPONEER 'one who has charge of a weapon of war prior to its deployment'
(OED / 1945)
x ROBOTEER 'a person who designs or constructs robots; an expert in robotics'

(OED / 1930)
x RACKETEER 'a person (especially a member of a gang or crime syndicate)

who earns money through a dishonest or illegal business, typically involving
extortion, intimidation, or violence' (OED / 1924)
136 Chapter 5
x FICTIONEER 'a writer or inventor of fiction' (OED / 1923)
x SLOGANEER 'one who devises or who uses slogans' (OED / 1922)
Given these formations, we can conclude that -eer can only attach to nominal
input lexes. That is:
x pumpkineer Å pumpkin + -eer

x fusioneer Å fusion + -eer
x free marketeer Å free market + -eer
x summiteer Å summit + -eer
x weaponeer Å weapon + -eer
x roboteer Å robot + -eer
x racketeer Å racket + -eer
x fictioneer Å fiction + -eer
x sloganeer Å slogan + -eer
Evidently, this fact cannot only be attributed to the signified of the suffix -eer
'one who is concerned with'. We can easily imagine a person who is concerned
with teaching (without, at the same time, being a teacher) or a person who is
concerned with blogging (without, at the same time, being a blogger). However,
it is extremely unlikely that an English speaker will ever call such people
*teacheer and *bloggeer. The reason for this is that the syntactics of -eer does
not allow attaching it to non-nominal input lexes.
The last aspect pertaining to combinatorial properties of an affix is its
productivity, i.e. the ability to attach to new input lexes (i.e. to those lexes it has
not been previously attached to), thus creating new lexemes. One conditio sine
qua non (i.e. a necessary condition) of the productivity of an affix is that of
being a living affix (Gizburg et at. 1979: 123), i.e. being recognizable as a
component part of complex lexes in which it occurs. As an illustration of this
point, let us consider the signifiers dead /dɛd/ and death /dɛθ/. Since these
semantically related signifiers differ only with regard to the final consonant
sound, it can be conjectured that /d/ of dead is an adjective-forming suffix
carrying the meaning 'quality' – dead means 'the quality of being dead' – and /θ/
of death is a noun-building suffix carrying the nominal meaning 'state': death is
the state of being dead.14 This analysis is true from a diachronic point of view.
According to the OED, /d/ of dead is indeed a suffix that goes back to Germanic
daudo-z, whereas /θ/ of death goes back to Germanic dauþu-z. However, today
there do not seem to be speakers of English who do indeed segment the word
dead into the root /dɛ/, carrying the meaning 'dead', and the suffix /d/, carrying
14
The analysis of /θ/ of death as a suffix meaning 'state' seems to be supported by words like
truth, where /θ/ is indeed a suffix carrying the meaning 'state': truth is the state of being true.
the meaning 'quality'. Similarly, the word death does not seem to be
synchronically segmentable into the root /dɛ/ and the suffix /θ/, carrying the
meaning 'state'. Both dead and death are monomorphemic words whose morphs
dead and death carry the meanings 'dead' and 'death'. /d/ of dead and /θ/ of death
are thus dead affixes and, accordingly, cannot be used for new derivational
formations. That is, there cannot be new formations like *gooth meaning 'the
state of being good' (from the adjectival input lex good, where /d/ is an affix
carrying the adjectival meaning 'quality') and *reth meaning 'the state of being
red' (from the adjectival input lex red, where /d/ is an affix carrying the
adjectival meaning 'quality'). Since /d/ of dead is a dead affix, which is not
recognized as a component part of the signifier dead, it cannot give rise to the
re-analysis of the originally monomorphemic words good and red as
combinations of the adjectival roots /gʊ/ and /rɛ/, carrying the meanings 'good'
and 'red', and the suffix /d/, carrying the meaning 'quality'.
However, note that not all living affixes are productive affixes. Consider, for
instance, the affix /θ/ of forms like strength, length, depth, etc. Despite the
obvious idiomaticity of the signifieds of these lexemes – e.g. strength does not
really mean 'the state of being strong' but 'a scale with regard to which a person
can be described as strong or weak' – /θ/ of strength, length, depth, etc. is
undoubtedly a recognizable component of these signifiers (and can thus be
regarded as at least a morfoid). However, despite its living character, the affix
/θ/ is currently not used for new derivational formations. According to the OED,
the last formations involving this suffix are illth 'ill-being' (1862) and sidth
'length; depth' (1855). The living suffix /θ/ is currently an unproductive
derivational suffix, whose syntactics does not allow it to be used for new
derivational formations. Hence we can argue for the impossibility or at least the
extreme unlikelihood of formations like *goodth, redth*, badth*.
A similar example of a synchronically living but unproductive affix is the
suffix -ant of words like defendant and accountant. According to the OED, the
last formations involving this suffix took places in the 1970s and the 1960s.
These include:
x REASSORTANT 'of a virus: having a genome consisting of parts derived from

the genomes of two (or more) different viruses, as a result of reassortment
during mixed infection of the same host or cells' (OED / 1979)
x TRANSCONJUGANT 'a plasmid or a bacterial cell which has received genetic

material by conjugation with another bacterium' (OED / 1974)
x RECORDANT 'a person who records a trademark with the customs authorities
in order to help prevent the importation of goods that infringe the mark'
(OED / 1969)
138 Chapter 5
x PROPPANT 'a granular material which is pumped with a fluid medium under
pressure into rock, so that fractures formed in this process are held open
when the pressure is released, allowing oil or gas to flow more freely' (OED /
1966)
x TRANSDUCTANT 'a cell into which genetic material has been transduced'
(OED / 1963)
x DOPANT 'the substance used in doping a semiconductor' (OED / 1963)
x INCAPACITANT 'a substance that can be used to incapacitate a person for a

time without wounding or killing him' (OED / 1961)
x ANOVULANT 'a drug or other agent that suppresses ovulation' (OED / 1960)
In addition to these lexemes, the OED also regards the 1983 formation
EXFOLIANT 'a cosmetic product designed to remove dead cells from the surface
of the skin' as a product of affixation of the verb exfoliate by means of the suffix
-ant. At the same time, the OED acknowledges that this form could have been
borrowed from French: cf. the English exfoliant and the French exfoliante.
Anyway, since the early 1980s the suffix -ant has not been used for new
derivational formations and can thus be regarded as a non-productive affix.
Now, let us proceed to consequences of affixation, i.e. the questions of
whether the addition of a particular affix will result in a change of word class
and whether it will trigger phonetic changes of the lex of an input lexeme.
What is meant by the former question is that while e.g. the addition of the
suffix -er to verbal input lexes like read produces nominal output lexemes
(READER, BLOGGER), the addition of the negative prefix un- to adjectival input
lexes like true produces adjectival output lexemes (UNTRUE, UNSACKABLE). A
change of word class is usually brought about by affixation (cf. TEACH and
TEACHER, CLOUD and CLOUDY, BELIZE and BELIZEAN, etc.), whereas prefixation
usually produces output lexemes whose lexes are members of the same word
class as lexes realizing their input lexemes: cf. TRUE and UNTRUE, COMMUNAL
and INTERCOMMUNAL, COMMUNICATE and MISCOMMUNICATE. At the same time,
note that there are word class non-changing suffixes like -eer of auctioneer: both
the input lexeme AUCTION and the output lexeme AUCTIONEER are nouns. And
there are word class changing prefixes: for example, while the input lex friend is
a noun, the output prefixed lex to befriend 'to act as a friend to, to help, favor; to
assist, promote, further' (OED) is a verb.
At first glance, it may be tempting to attribute this important aspect of the
syntactics of an affix to its signified. Indeed, if -er means 'performer of some
action', then the addition of this suffix to verbal input lexes like to read must
give rise to nominal output lexes like reader. Similarly, if -y means 'full', then
the addition of this suffix to a nominal input lex like cloud must give rise to an
adjectival output lex like cloudy. However, the main reason why the lex reader
is a noun and the lex cloudy is an adjective is that both these output lexes (or to
be more precise, their signifieds) are headed by the suffixes -er and -y. That is,
the output signified 'teacher' can be represented as the NP performer of the
action of teaching, whose head is the nominal meaning component 'performer',
inherent in the affix -er. Similarly, the output signified 'cloudy' can be
represented as the adjective phrase full of clouds, whose head is the adjectival
meaning component 'full', inherent in the suffix -y. The nominal meaning
'performer' and the adjectival meaning 'full' are thus the head meanings of the
output lexemes READER and CLOUDY.
By contrast, the head meaning of the output lexeme MISCOMMUNICATE is not
the meaning 'pejoration', inherent in mis-, but the meaning 'to communicate',
inherent in communicate: miscommunicate means 'to communicate badly'. This
is the reason why the addition of the prefix mis- to the input lex communicate
gives rise to the output lex miscommunicate, which is a member of the same
word class as the input lex communicate.
The signified 'auctioneer' is similar to the signifieds 'teacher' and 'cloudy' in
that it is also headed by the suffix -eer. That is, the head meaning of the
signified 'auctioneer' is the nominal meaning 'person', inherent in the affix -eer.
However, since in auctioneer this meaning combines with the nominal meaning
'auction' – auctioneer means 'a person who conducts auctions' – the suffix -eer
seems similar to the prefix mis- in that it has also produced the output lexeme
which is a member of the same word class as the corresponding input lexeme.
The suffix -eer of auctioneer is, however, different from the prefix mis- of
miscommunicate in that the former heads the output lexeme AUCTIONEER.
In summary, the syntactics of an affix also contains the information as to
whether the affix can function as head of a new lexeme. This information
determines whether the meaning inherent in the affix will be able to function as
head meaning of the signified of an output lexeme, thereby (often, but not
always) giving rise to an output lex which is a member of a different word class
than a corresponding input lex.
As regards the ability to trigger phonetic changes, let us compare the
following three formations:
x CLOUDY /ˈklaʊdɪ/; cf. the input lex /klaʊd/

x BAZILLIONAIRE /bəˌzɪljəˈnɛː/; cf. the input lex /bəˈzɪljən/
x ADDITIONALITY /əˌdɪʃəˈnalᵻti/; cf. the input lex /əˈdɪʃnˌ(ə)l/
These examples serve to illustrate that affixes can be classified into the
following three categories:
140 Chapter 5
1. stress-neutral affixes
2. stress-attracting affixes
3. stress-shifting affixes
The adjective-building suffix -y of cloudy is an example of a stress-neutral affix.

As one can notice, the output signifier cloudy is stressed on the first syllable, just
as the input signifier cloud.
The noun-building suffix -aire of bazillionaire 'a person of enormous wealth'
(OED / 1987) is an example of a stress-attracting affix: the primary stress of
bazillionaire falls on the suffix -aire, not on the second syllable as in the input
signifier /bəˈzɪljən/.
Finally, the suffix -ity of additionality is an example of a stress-shifting
suffix: its addition to the input signifier additional has shifted the primary stress
from the second to the fourth syllable: cf. /əˌdɪʃəˈnalᵻti/ and /əˈdɪʃnˌ(ə)l/.
Stress-neutral affixes also include the majority of prefixes as well as the
suffixes -al -er, -ist, -ship, etc. Compare, for example:
x /sɪˈmɛtrɪk/ and /æsɪˈmɛtrɪk/ (symmetric and asymmetric)

x /kəˈmjuːnᵻkeɪt/ and /ˌmɪskəˈmjuːnᵻkeɪt/ (communicate and miscommunicate)
x /ˈsɛntrəlaɪz/ and /diːˈsɛntrəlaɪz/ (centralize and decentralize)
x /əˈkjuːz/ and /əˈkjuːzəl/ (accuse and accusal)
x /tiːtʃ/ and /ˈtiːtʃə(r)/ (teach and teacher)
x /dɪˈnæmɪks/ and /daɪˈnæmɪsɪst/ (dynamics and dynamicist)
x /ˈdəʊnə(r)/ and /ˈdəʊnəʃɪp/ (donor and donorship)
Stress-attracting affixes also include the suffixes -ation, -eer, -ese, -esque, -ette,
etc. For example:
x /ˈɛskəleɪt/ and /ˌɛskəˈleɪʃən/ (escalate and escalation)

x /ˈɔːkʃən/ and /ɔːkʃəˈnɪə(r)/ (auction and auctioneer)
x /s(j)ʊərɪˈnæm/ and /s(j)ʊərɪnæˈmiːz/ (Suriname and Surinamese)
x /kəmˈpjuːtə/ and /ˌkəmpjuːtəˈrɛsk/ (computer and computeresque)
x /ˈpɛdəgɒg/ and /ˌpɛdəgɒˈgɛt/ (pedagogue and pedagoguette)
Finally, stress-shifting affixes also include -ic, -ial, -ual, etc. For example:
x /ˈsɛljᵿlʌɪt/ and /ˌsɛljᵿˈlɪtɪk/ (cellulite and cellulitic)

x /kəmˈpəʊnənt/ and /kɒmpəˈnɛnʃɪəl/ (component and componential)
x /ˈkɒnflɪkt/ and /kənˈflɪktjuːəl/ (conflict and conflictual)
Like synchronic productivity, the status of an affix as a stress-neutral, stress-

attracting, or stress-shifting affix is not predictable from either its signifier or its
signified. Also, there does not seem to exist a purely phonetic explanation (such
as e.g. a place-of-articulation assimilation) for the fact that, for example, the
addition of the suffix -aire to bazillion must necessarily shift the stress from the
second syllable /zɪ/ to the fourth syllable /nɛː/.15 Nouns in English are usually
stressed on the penultimate syllable (i.e. the pre-final syllable) provided that it
is heavy, i.e. a syllable which ends in a consonant, a diphthong, or a long vowel.
When the penultimate syllable is light (i.e. a syllable ending in a short vowel),
the primary stress moves to the antepenult (i.e. the syllable preceding the
penult). For example, algebra /ˈældʒɪbrə/ is stressed on the antepenult /æl/
because the penult /dʒɪ/ is a light syllable: it ends in the short vowel /ɪ/. Now, let
us consider the placement of the primary stress in /bəˌzɪljəˈnɛː/. Given that the
penultimate syllable /jə/ is light, one would expect bazillionaire to retain the
primary stress on the second syllable /zɪl/. This, however, does not happen
because, as said above, the suffix -aire is a stress-attracting suffix, i.e. its
syntactics orders English speakers to place the primary stress on -aire, whenever
this suffix is used for derivational formations.
5.4.3 Productivity
Affixation is a very productive lexeme-building mechanism in Present-day

English. Thus, of 587 new signifiers which, according to the OED, appeared in
the English language between 1990 and 2011, 170 can be regarded as instances
of affixation. Below are just some of the examples:
x ANORAKY (Å anorak + -y) 'boring, overly studious, unfashionable, or

socially inept; specially displaying obsessive or fastidious concern with the
details of a hobby or special interest […]' (OED / 1992)
x BLAIRISM (Å Blair + -ism) 'the policies and principles advocated by Tony

Blair' (OED / 1994)
x CABBAGED (Å cabbage + -ed) 'incapacitated by drugs or alcohol (or their

after-effects); extremely intoxicated' (OED / 1991)
15
Note that this does not apply to all phonetic changes triggered by the addition of a
derivational affix. For example, as Halle (2005) argues, so-called velar softening exemplified
by the change of the velar consonant [k] of e.g. /aɪˈkɒnɪk/ into the alveolar [s] of e.g.
/aɪkəˈnɪsɪtɪ/ is an instance of place-of-articulation assimilation of the velar [k] to the front
vowel [ɪ] rather than an inherent property of the syntactics of the noun-building affix -ity.
142 Chapter 5
x DOWNSHIFTER (Å to downshift + -er) 'a person who adopts a less pressured

and demanding career or lifestyle, especially one who accepts a reduced
income in pursuit of personal fulfillment' (OED / 1990)
x EEYORISH (Å Eeyore + -ish) 'deeply pessimistic; gloomy' (OED / 1992)
x GENOMICIST (Ågenomic + -ist) 'a scientist who works in the field of

genomics' (OED / 1995)
x LADETTE (Å lad + -ette) 'a young woman characterized by her enjoyment of

social drinking, sport, or other activities typically considered to be male-
oriented […]' (OED / 1995)
x TO OVERSUPINATE (Å over- + to supinate) 'to run or walk so that the weight

falls upon the outer sides of the feet to a greater extent than is necessary,
desirable, etc.; to supinate excessively' (OED / 1990)
x TO REWILD (Åre-+ to wild) 'to return (land) to a wilder and more natural
state' (OED / 1990)
To study the productivity of a particular suffix (e.g. -ish of Eeyorish), go to the

OED Advanced Search page http://www.oed.com/advancedsearch and enter
'*ish' (without quotation marks) to 'Restrict to entry letter or range'. The OED
will then yield all lexemes whose lexes end in -ish. Some of them will be
adjectives like Eeyorish which contain the suffix -ish. You can refine your
search by entering e.g. '1990-' to 'Date of entry' and choosing the option
'Adjective' in 'Part of speech'. The OED will then yield all recent adjectival
lexemes (i.e. which appeared between 1990 and 2011) whose lexes end in -ish.
To study the productivity of a particular prefix (e.g. re- of rewild), go to the
above mentioned OED Advanced Search page and enter 're*' (without quotation
marks) to 'Restrict to entry letter or range'. The OED will then yield all lexemes
whose lexes begin with re-. Some of them will be verbs like to rewild which
contain the prefix re-. You can refine your search by entering e.g. '1990-' to
'Date of entry' and choosing the option 'Verb' in 'Part of speech'. The OED will
then yield all recent verbal lexemes (i.e. which appeared between 1990 and
2011) whose lexes begin with re-.
As was pointed out in 4.2.8, the synchronic productivity of a derivational
affix can also be studied with the help of Word Spy. Just go to the Word Spy
Advanced Search page http://www.wordspy.com/search.asp, enter '*ish' or 're*'
(both without quotation marks) to the search mask, choose the option 'Within
title', and then click at 'Search'. Word Spy will yield all words ending in -ish and
beginning with re-. Some of them could be adjectives like Eeyorish and verbs
like to rewild. Unfortunately, it is not possible to instruct Word Spy to search for
words that are members of a particular word class.
5.5 Lexeme-building apophony
Lexeme-building apophony is any modification of the lex of an input lexeme

that does not qualify as an instance of segmental affixation. One type of
apophony (which has been mentioned in 4.2.2) is the stress shift. Consider, for
example, the following formations:
x to increase /ɪnˈkriːs/ Æ an increase /ˈɪnkriːs/

x to insult /ɪnˈsʌlt/ Æ an insult /ˈɪnsʌlt/
x to permit /pəˈmɪt/ Æ a permit /ˈpəːmɪt/
In all these cases, the stress shift is accompanied by quasi-idiomatization of the

input signifieds, i.e. an increase is the process of increasing; an insult is the act
of insulting; a permit is a document that permits something (e.g. a residence
permit is a document that grants permission to reside in some country).
In addition to the stress shift, apophonies also include:
x consonant change
x vowel change
x consonant change accompanied by vowel change
x stress shift accompanied by vowel change
As an illustration of the first category, let us again consider the pair dead–death,
which we discussed in the previous section. As we established, neither [d] of
dead nor [θ] of death can be regarded as an affix from a synchronic point of
view. Given this fact and given that the signified 'death' can be analyzed as a
quasi-idiom in relation to the signified 'dead' – death is the state or condition of
being of dead – we are justified in concluding that the lexeme DEATH is a
product of apophony of the lexeme DEAD: the former came into existence via the
change of the final consonant [d] of the input lex dead to [θ] in the output lex
death. Similar examples include:
x to defend /dɪˈfɛnd/ Æ defense /dɪˈfɛns/

x to speak /spiːk/ Æ speech /spiːtʃ/
x to believe /bɪˈliːv/ Æ belief /bɪˈliːf/
As an illustration of the second category, let us consider the pair full /fʊl/–to fill
/fɪl/. From a semantic point of view, the signified of the verb to fill can be
144 Chapter 5
analyzed as a quasi-idiom in relation to the signified of the adjective full: to fill

can be said to mean 'to make full'. From a formal point of view, the lexes of the
two lexemes differ only with regard to the vowel sounds [ʊ] of full and [ɪ] of fill.
Neither the former nor the latter can be regarded as an affix carrying a
discernible meaning of its own: neither full nor fill seem to be synchronically
segmentable into the root /fl/, carrying the meaning 'full', and the infixes [ʊ] / [ɪ],
carrying the meaning 'quality of being […]' / 'process of making […]'. Both
lexes are monomorphemic words, which cannot be segmented into smaller units
(be it morphs or quasi-linguistic units). Accordingly, we can conclude that the
lexeme TO FILL is a product of apophony of the lexeme FULL: the former came
into existence via the change of the root vowel [ʊ] of the input lex full to [ɪ] in
the output lex fill. Similar examples include:
x blood /blʌd/ Æ to bleed /bliːd/

x separate /ˈsɛpərət/ Æ to separate /ˈsɛpəreɪt/
x legitimate /lɪˈdʒɪtɪmət/ Æ to legitimate /lɪˈdʒɪtɪmeɪt/
The third category is essentially the blend of the first two categories. Consider,
for instance, the pair life /lʌɪf/–to live /lɪv/. As in the previous examples, there
exists a quasi-idiomatic relation holding between the signifieds of the lexemes in
question: to live can be said to mean 'to be alive or to possess life'. As regards
the formal side, the difference concerns the last two sounds /ʌɪf/ of life and /ɪv/
of live. Similar to previous examples, we cannot regard these sounds as affixes.
That is, the noun life cannot be segmented into the root /l/ and the suffix /ʌɪf/
and the verb live cannot be segmented into the root /l/ and the suffix /ɪv/. (Recall
that roots are usually longer than affixes.) Given this fact, we are justified in
concluding that the lexeme TO LIVE is a product of apophony of the lexeme LIFE:
the former came into existence via the change of the root vowel [ʌɪ] to [ɪ] as
well as via the change of the final consonant [f] to [v].
Finally, the last category is the blend of the categories 'stress shift' and
'vowel change'. Examples include such pairs as:
x conduct /ˈkɒndəkt/ Æ to conduct /kənˈdʌkt/

x contest /ˈkɒntɛst/ Æ to contest /kənˈtɛst/
x fragment /ˈfrægmənt/ Æ to fragment /frægˈmɛnt/
Here the change of the stress is accompanied by the change of the quality of one
or more vowels. Strictly speaking, the latter represents a consequence of the
former: it is a well-known fact that vowels in unstressed syllables often undergo
reduction and change to a schwa (i.e. the sound /ə/). That is why this last type of
apophony represented by pairs like conduct–to conduct can also be regarded as
an instance of the category 'stress shift' rather than as an independent category.
5.5.1 Productivity
In Present-day English apophony is a much less productive lexeme-building

mechanism than affixation, borrowing, and semantic change: of 587 new
signifiers which, according to the OED, appeared in the English language
between 1990 and 2011, no more than three can be regarded as instances of
lexeme-building apophony. These are the lexes that were already named in
4.2.8:
x adultescent /ˌadʌlˈtɛsnt/ Å adolescent /ædəʊˈlɛsənt/

x Google /ˈguːgl/ Ågoogol /ˈguːgɒl/
x Lollywood /ˈlɒlɪwʊd/ Å Bollywood /ˈbɒlɪwʊd/
Earlier examples of lexeme-building apophony include:
x def (Å death) 'excellent, outstanding; fashionable, cool' (OED / 1981)
x herstory (Å history) '[…] history emphasizing the role of women or told

from a woman's point of view; also, a piece of historical writing by or about
women' (OED / 1970)
x to wazz (Åto whizz 'to make a sound as of a body rushing through the air') 'to
urinate' (OED / 1984)
To study the productivity of apophony, go to the OED Advanced Search page

http://www.oed.com/advancedsearch, enter 'alteration' (without quotation
marks) to the first row of the search mask from above, and choose the option
'Etymology' instead of 'Full-text' (near the search mask). The OED will then
yield all entries whose Etymology-sections contain the word alteration (i.e. how
the OED refers to phonetic changes which this book regards as instances of
apophony). Some of these entries will be devoted to lexemes whose lexes came
into existence via apophony.
5.6 Compounding
Compounding is the creation of a new lexeme by means of combining at least

two input roots into the lex of a new complex lexeme, a compound lexeme. For
example, PLAYGROUND Å play + ground.
Note that in English a considerable number of compound lexemes, whose
lexes qualify as compounds from a synchronic point of view (i.e. they are
synchronically segmentable into at least two roots), do not qualify as instances
146 Chapter 5
of compounding from a diachronic perspective. For example, as we have learned

in 4.2.4, the compound nominal lexeme BOYFRIEND2 'a male sexual partner',
whose lex is synchronically segmentable into the roots boy and friend, came into
existence via metonymy-based full-idiomatization of the lexeme BOYFRIEND1 'a
boyhood friend'. Similarly, the compound verbal lexeme BABYSIT came into
existence via back-formation of the nominal lexeme BABYSITTER.
Some authors regard compound lexemes like BOYFRIEND2 and BABYSIT as
pseudo-compounds. The distinction between a pseudo-compound like BABYSIT,
whose lex is a product of another lexeme-building mechanism, and a genuine
compound like PLAYGROUND, whose lex is indeed a product of compounding, is
of particular theoretical importance for those students of English morphology
who investigate the mechanism of compounding (e.g. differences between
compounding and other lexeme-building mechanisms). Indeed, if you are
interested in the peculiarities of this process, you must ignore pseudo-
compounds like BOYFRIEND2 and BABYSIT: these do not qualify as instances of
compounding and, accordingly, must be left out of consideration.
Since the goal of this section is to describe the mechanism of compounding,
we will accept the distinction between a pseudo-compound and a genuine
compound. To find out whether the compound under analysis is indeed a
product of compounding, we will rely on the OED. For example, the
Etymology-section of the OED entry for PLAYGROUND contains the following
information: 'Etymology < play n. + ground n'. Relying on this information, we
can conclude that playground is indeed an instance of compounding, i.e. a
compound lex which was created via combining the two input roots play and
ground into the new compound lex playground.
5.6.1 Compounding as an anisomorphic mechanism
As we argued in 4.2.3, compounding is an anisomorphic lexeme-building

mechanism, i.e. a mechanism that produces output lexemes whose signifieds are
not (or not entirely) representable in terms of their components' signifieds. Thus
of 81 recent instances of pure compounding (of 587 new signifiers which
appeared in the English language between 1990 and 2011), there are no
isomorphic compounds, i.e. compound lexemes whose signifieds are made up of
their components' signifieds only. The overwhelming majority of the recently
formed compounds – 50 compounds – exhibit quasi-idiomatic signifieds. For
example:
x DRUM AND BASS 'a style of popular dance music originating in Britain in the
early 1990s, variously thought of as derived from or identical to jungle, and
characterized primarily by a fast drum track and a heavy, usually slower, bass
track, but often also featuring synthesized or sampled strings, piano, or other
instrumentation' (OED / 1992)
x STAR 69 'a call return service which automatically dials the number from
which the last incoming call was made, activated by dialing the 'star' (*), 6,
and 9 buttons on a touch-tone phone' (OED / 1990)
x INFORMATION FATIGUE 'apathy, indifference, or mental exhaustion arising

from exposure to too much information, especially (in later use) stress
induced by the attempt to assimilate excessive amounts of information from
the media, the Internet, or at work' (OED / 1991)
x NO-MATES 'a person (usually a man) regarded as lonely or having no friends'

(OED / 1993)
x YEAR 2000 'designating or relating to computer problems arising from the

inability of certain software and firmware to deal correctly with dates of 1
January 2000 or later, owing to the numerical representation of calendar
years by the last two digits only' (OED / 1993)
x TO EGO-SURF 'to search on the Internet for mentions of one's own name or the
name of one's business, website, etc.' (OED / 1995)
x EURO NOTE 'any of various banknotes denominated in the euro currency'

(OED / 1995)
x DOTCOM 'an Internet address for a commercial site expressed in terms of the
formulaic suffix .com; a web site with such an address' (OED / 1994)
x ALWAYS-ON (adjective) 'relating to or designating a continuously accessible

connection to the Internet, especially in contrast to one that requires
activation by dial-up or other means' (OED / 1996)
21 compounds have semi-idiomatic signifieds. For example:
x HOT DESK 'a shared office desk or workstation, occupied on a temporary, ad

hoc basis or part-time basis, and not allocated permanently to an individual'
(OED / 1990)
x GENDERQUAKE 'a radical alteration in the relationship between the sexes,

especially one resulting from deliberate changes in women's economic or
political activity' (OED / 1993)
148 Chapter 5
x DADROCK 'rock music that appeals to an older generation, or is heavily

influenced by that of an earlier era, especially the 1960s' (OED / 1994)
x MEATSPACE 'the physical world, as opposed to cyberspace or a virtual

environment' (OED / 1995)
x BLUE STATE 'a state (projected to be) won by the Democratic candidate in a
presidential election. More generally: a Democratic state; a state which tends
to vote Democrat' (OED / 2000)
Finally, only 10 compounds can be analyzed as full-idioms. For example:
x TO DRAG AND DROP 'to move or copy (an image, icon, text, etc.) from one part
of a display screen to another using a mouse or similar device using a drag-
and-drop facility' (OED / 1990)
x CARPET MUNCHER 'a lesbian' (OED / 1992)
x DOWN-LOW (adjective) 'secret, quiet, low-profile; (in later use) specially of or

relating to men who secretly engage in homosexual activity' (OED / 1991)
These numbers tell us that quasi-idiomatization is the default semantic outcome

of compounding. That is, the majority of lexemes that come into existence via
compounding have signifieds that contain their components' signifieds plus
some additional idiomatic signifieds. The reason for this is that the creation of a
quasi-idiomatic compound requires a lesser effort on the part of a language user
than does the creation of a semi- or fully-idiomatic compound. The latter –
especially metaphor-based full-idioms like CARPET-MUNCHER and DOWN-LOW –
require a considerable creativity and therefore do not occur as often as quasi-
idiomatic compounds.
5.6.2 The semantics of compounding
Since compounding is an anisomorphic lexeme-building mechanism, which

always produces idiomatic lexemes, any study of compounding must try to
answer the question of whether there are any regularities governing the creation
of quasi-, semi-, and fully-idiomatic compound signifieds.
To begin with, it must be observed that everything we have learned about
semantic change in Section 5.1 is true of anisomorphic compounding. This
means that the compounding of the lexes of two or more input lexemes into the
lex of a new compound lexeme can be accompanied by either metonymization
or metaphorization of one or both of its components' signifieds. For instance, the

semi-idiom DADROCK 'rock music that appeals to an older generation' is a
product of part-for-whole metonymization of the signified 'dad': our fathers
(alongside with mothers, grandmothers, grandfathers, etc.) are representatives of
an older generation. By contrast, the semi-idiom GENDERQUAKE 'a radical
alteration in the relationship between the sexes' came into existence via
metaphorization of the signified 'quake': a radical alteration in the relationship
between the sexes was metaphorically analogized to a physical quake.
Similarly, fully-idiomatic compounds may come into existence via:
1. metaphorization of all of their components' signifieds

2. metonymization of all of their components' signifieds
3. combination of these two strategies
For example, the compound carpet muncher 'a lesbian' came into existence via
metaphorization of the input signifieds 'carpet' and 'muncher': lesbian sex
(especially cunnilingus) must have been metaphorized as munching a carpet.
This analysis is corroborated by an earlier formation RUG MUNCHER 'a lesbian.
Also (occasionally): a man who performs cunnilingus' (OED / 1981): in this
formation female genitalia are metaphorized as a rug on which a man
performing cunnilingus munches.
As an illustration of a metonymy-based fully-idiomatic compound, consider
the lexeme GREEN ACCOUNTING 'a system in which economic measurements take
into account the effects of production and consumption on the environment'
(Word Spy / 1989). This compound came into existence via metonymization of
the input signifieds 'green' and 'accounting': the green color is a well-known
symbol of environmental issues, so that the input signified 'green' can be said to
metonymically stand for the output signified 'environment'; similarly, economic
measuring can be seen as an instance of accounting, so that the input signified
'accounting' can be said to metonymically stand for the output signified
'economic measurements'.
Finally, consider the fully-idiomatic lexeme GREY NOMAD 'a retired person
who travels extensively, particular in a recreational vehicle' (Word Spy). This
compound came into existence via metonymization of the input signified 'grey'
and metaphorization of the input signified 'nomad'. As regards the former, old
people often have grey hair, so that the input signified 'grey' can be said to
metonymically stand for the idiomatic output signified 'old or retired people'. As
for the latter, the input signified 'nomad' cannot be regarded as a metonym for
the output signified 'one who travels extensively, particular in a recreational
vehicle': a person who extensively travels in a recreational vehicle is not literally
a nomad, i.e. 'a member of a people that travels from place to place to find fresh
pasture for its animals, and has no permanent home' (OED). The former can
150 Chapter 5
only be metaphorically analogized to the latter: that is, the activity of

extensively travelling in a recreational vehicle can be perceived of as being
similar to the travelling activities of nomads.
As regards quasi-idiomatic compounds, it appears that they fall into two
main categories:
1. quasi-idiomatic compounds of the information fatigue-type

2. quasi-idiomatic compounds of the drum and bass-type
Quasi-idiomatic compounds of the information fatigue-type are quasi-idioms

like INFORMATION FATIGUE whose additional idiomatic meanings make these
compounds' signifieds narrower than the mere sum of their components'
signifieds. That is, information fatigue is a particular kind of fatigue that has
something to do with information: fatigue that arises from exposure to too much
information; a euro note is a particular banknote that has something to do with
the euro currency: a banknote which is denominated in the euro currency; ego-
surfing is a particular instance of surfing the Internet that has something to do
with oneself: surfing the Internet for mentions of one's own name; etc. With
regard to the quasi-idiomatic compounds that were discussed in the previous
chapters, this pattern is characteristic of e.g. the compounds spring festival and
brake cable: as we argued, a spring festival is a particular festival that has
something to do with spring and a brake cable is a particular cable that has
something to do with a brake.
Quasi-idiomatic compounds of the drum and bass-type are quasi-idioms like
DRUM AND BASS whose components name some (more or less) important
characteristics of what the lexes realizing these lexemes are used to refer to.
Thus drum and bass is a style of dance music which is characterized by a fast
drum track and a slower bass track. Similarly, star 69 is a service that is
activated by dialing '*69'; a No-Mates is a lonely man who has no mates; a
dotcom is an Internet address that is expressed in terms of the formulaic suffix
'.com'; etc. With regard to the previously discussed quasi-idiomatic compounds,
this pattern can be found in FOOTBALL, BASKETBALL, HANDBALL, etc. As we
established in 2.4.7, the lexes of these lexemes are segmentable into the
components foot / ball, basket / ball, hand / ball, volley / ball, etc., whose
signifieds provide an (incomplete) explanation for how these games are
supposed to be played: a football is a game that involves a ball and the players'
feet; basketball is a sport that involves a ball and a basket; handball is a sport
that involves a ball and the players' hands; etc.
Quasi-idiomatic compounds of the drum and bass-type are sometimes called
bahuvrihi compounds, from the Sanskrit bahuvrihi 'having much rice' (OED).
The term 'bahuvrihi' has usually been used in connection with compounds like
SKINHEAD 'a person whose hair is worn very short or shaved off entirely' (OED)
and CUTTHROAT 'one who cuts throats; a ruffian who murders or does deeds of
violence […]' (OED) which denote individuals possessing the characteristics
named by these compounds' components. That is, a skinhead is a person whose
hair is worn very short, so that other people can see the skin of his head. And a
cutthroat is a murderer who murders by cutting other people's throats. In other
words, the components skin and head name an important characteristic of a
skinhead and the components cut and throat name an important characteristic of
a cutthroat. It is not difficult to notice that the bahuvrihis SKINHEAD and
CUTTHROAT exhibit the very same semantic pattern as the quasi-idiomatic
compounds DRUM AND BASS, STAR 69, NO-MATES, DOTCOM, FOOTBALL, etc.
In addition to classifying quasi-idiomatic compounds into these two rather
general categories, one may also try to discover the most frequently recurrent
patterns of quasi-idiomatization in compounding. This issue has been recently
addressed by Jackendoff (2009: 123-124), who, without using the term 'quasi-
idiom', has proposed "a list of the (most prominent) basic functions for English
compounds". What Jackendoff calls 'a basic function' is, however, nothing more
than just an idiomatic meaning that is a part of a quasi-idiomatic compound's
signified which cannot be attributed to either of its components' signifieds.
These include, for example:
x the meaning 'both' in compounds like POLITICIAN-TYCOON: a politician-

tycoon is a person who has the properties of both a politician and a tycoon.
x the meaning 'serves as' in compounds like BUFFER STATE : a buffer state is a
state that serves as a buffer.
x the meaning 'located at / in / on / near' in compounds like WINDOW SEAT: a

window seat is a seat that is located near a window.
x the meaning 'takes place at specified time' in compounds like SPRING

FESTIVAL: a spring festival is a festival that takes place in spring.
x the meaning 'caused by' in compounds like KNIFE WOUND: a knife wound is a
wound that was caused by a knife.
x the meaning 'consists of' in compounds like BRASS INSTRUMENT: a brass

instrument is an instrument that consists of brass.
x the meaning 'part of' in compounds like BRAKE CABLE: a brake cable is a
cable that is a part of a brake.
152 Chapter 5
x the meaning 'made by' in compounds like FOOTPRINT: a footprint is a print

made by a foot.
x the meanings 'protect something' and 'protect from something' in compounds

like CHASTITY BELT and SUN HAT: a chastity belt is a belt that protects chastity
and a sun hat is a hat that protects from the sun.
Despite the fact that many quasi-idiomatic compounds do indeed come to

signify these 'basic' meanings, the semantic outcome of compounding is to a
very large extent unpredictable and unexplainable. That is, we cannot really
explain why a particular quasi-idiomatic compound came to be associated with a
particular idiomatic meaning. A good example illustrating this point is the quasi-
idiom HEADACHE PILL. As a first approximation, it can be observed that its
signified is rather similar to the signified of the quasi-idiom SUN HAT, mentioned
above: a sun hat is a hat that is supposed to protect from the sun and a headache
pill is a pill that is supposed to act as a pain reliever, i.e. as a pill that reduces the
headache. At the same time, notice that the idiomatic meaning 'to reduce' is not
the only logically possible idiomatic meaning which the compounding of the
components headache and pill could have given rise to. Apart from the
idiomatic meaning 'to reduce', headache pill could also have acquired the
idiomatic meaning 'to cause': that is, a headache pill is a pill that causes the
headache as a side-effect of the treatment of another condition. This quasi-
idiomatic interpretation of headache pill is no less plausible than that of a pill
that reduces the headache (Cf. Bauer 1979: 45-46, who discusses the semantic
structure of the quasi-idiomatic compounds SLEEPING PILL, SEA-SICKNESS PILL,
and ANTIHISTAMINE PILL).
5.6.3 Endocentric and exocentric compounding
Compounds are often classified into endocentric and exocentric compounds (or
simply endocentrics and exocentrics). The distinction between these two
categories relies heavily on the notion of the head of a compound, which, as we
will see below, has both a semantic and a formal dimension. The unfortunate
consequence of this is that one and the same compound can simultaneously
qualify as an exocentric compound from a semantic point of view and as an
endocentric compound from a formal point of view.
As regards the semantic dimension, recall the converted verb to wife and the
affixed noun reader, which we discussed in the previous sections of this chapter.
As we argued, complex signifieds like 'to wife' and 'a reader' can be represented
as syntactic phrases (i.e. VPs, NPs, etc.) which are headed by a particular
element. For example, the signified 'to wife' can be represented as the VP to
downplay a woman's career accomplishments in favor of her abilities as wife

and mother, whose head is the verb downplay. This fact explains why the quasi-
idiomatization of the input nominal signified 'a wife' resulted in the word class
change of the nominal input lex a wife. Similarly, the signified 'a reader' can be
represented as the NP a performer of the action of reading, whose head is the
noun performer. This fact explains why the affixation of the nominal input lex
read by means of the suffix -er gave rise to an output lex which is a member of a
different word class than the corresponding verbal input lex read. Comparing
the two examples, one can notice that while in the case of reader, the head
meaning 'performer' is part of the signified of the suffix -er, the head meaning 'to
downplay' of to wife is an idiomatic meaning which is not part of the signified of
the component wife. Complex lexemes like READER whose head meanings are
inherent in one of their components' signifieds are endocentrics, while lexemes
like TO WIFE whose head meanings are idiomatic meanings that are not part of
their components' signifieds are exocentrics.
Returning to compounds, it appears that the endocentric–exocentric
distinction neatly corresponds to the distinction between the two types of quasi-
idiomatic compounds that was elaborated on in the previous part of this section.
That is, while quasi-idiomatic compounds of the information fatigue-type
always qualify as endocentric compounds – for example, the quasi-idiom
INFORMATION FATIGUE is headed by the meaning 'fatigue', inherent in the
component fatigue; the quasi-idiom EURO NOTE is headed by the meaning 'note',
inherent in the component note; the quasi-idiom TO EGO-SURF is headed by the
meaning 'to surf', inherent in the component surf; etc. – quasi-idiomatic
compounds of the drum and bass-type are always exocentric compounds: e.g.
the quasi-idiom DRUM AND BASS is headed by the idiomatic meaning 'music
style', which is not inherent in either drum or bass; the quasi-idiom STAR 69 is
headed by the idiomatic meaning 'service', which is not inherent in either star or
69; the quasi-idiom NO-MATES is headed by the meaning 'person', which is not
inherent in either No or Mates; etc.
At the same time, observe that the endocentric–exocentric distinction is
broader than that between quasi-idiomatic compounds of the information
fatigue- and drum and bass-types. In contrast to the latter, the former distinction
can also be applied to semi- and fully-idiomatic compounds. Consider, for
example, the semi-idiomatic compound lexeme DADROCK. Despite the fact that
this semi-idiom does not qualify as a quasi-idiomatic compound of the
information fatigue-type, it does qualify as an endocentric compound: its
signified can be represented as the NP rock that appeals to an older generation,
whose head element is the noun rock. Since the head meaning 'rock' is inherent
in the component rock, we are justified in claiming that DADROCK is an
endocentric compound. By contrast, the fully-idiomatic compound lexeme
CARPET MUNCHER is clearly an exocentric compound. Its head meaning 'lesbian'
154 Chapter 5
is inherent in neither carpet nor muncher: like quasi-idiomatic compounds of the

drum and bass-type, the components carpet and muncher only name a feature
that can be metaphorically conceived of as an important characteristic of a
lesbian.
In addition to endocentric and exocentric compounds, many authors also
introduce the category 'copulative compound' as a distinct compound type.
Copulative compounds are said to be different from endocentric compounds in
that both of their components are equally important from a semantic point of
view. An often cited example is the compound FIGHTER-BOMBER 'an aircraft that
combines the functions of a fighter and a bomber' (OED), whose components
fighter and bomber denote equally important characteristics of fighter-bomber
aircrafts: these can be used as both fighters and bombers. Similarly, the already
mentioned compound POLITICIAN-TYCOON can also be regarded as a copulative
compound because its lex can be used to refer to people who are both tycoons
and politicians.
This textbook rejects the category of a copulative compound. What other
studies regard as copulative compounds are compound lexemes which are in no
essential respect different from exocentric compounds of the drum and bass-
type. That is, the signified 'fighter-bomber' can be represented as the NP an
aircraft that can be used as both a fighter aircraft and a bomber aircraft. This
NP is headed by the meaning 'aircraft', which is not inherent in either fighter or
bomber: in the compound under analysis, the components fighter and bomber
only name the two equally important characteristics of a fighter-bomber aircraft
distinguishing it from other military aircrafts. Similarly, the signified 'politician-
tycoon' can be represented as the NP a person who is both a politician and a
tycoon. This NP is headed by the meaning 'person', which is not inherent in
either politician or tycoon: in the compound under analysis, the components
politician and tycoon only name the two equally important characteristics of
politician-tycoons distinguishing them from other people. Given these semantic
structures, both fighter-bomber and politician-tycoon can be analyzed as
exocentric compounds: they are headed by meanings which are not inherent in
either of their components' signifieds.
Now, let us proceed to the endocentric–exocentric distinction from a formal
point of view. Recall that analyzing the morphemic structure of the semantically
opaque verb understand, we said that the formal head of the lex of a complex
lexeme is that component which determines its inflectional marking. Thus even
without considering the signified 'to understand', we can conclude that this verb
is headed by the component stand: there can only be he understands but not *he
undersstand; likewise, there can only be he understood, not *he understanded,
let alone *he underedstand. Similarly, even without considering the signified
'fighter-bomber', we can easily analyze this compound as a formally endocentric
compound headed by the component bomber: there can only be two fighter-
bombers but not *two fighters-bomber or *two fighters-bombers. Similarly,

there can only be two politician-tycoons but not *two politicians-tycoon or *two
politicians-tycoons.
A well-known characteristic of English compounds is their formal right-
headedness, i.e. the righthand member receives the inflectional marking and thus
functions as the formal head of a compound. This characteristic enables us to
distinguish compounds from other idiomatic combinations involving two or
more roots. For example, the fully-idiomatic VP kick the bucket 'to die' cannot
be regarded as a compound because it is headed by the lefthand member kick but
not by the righthand member bucket: we can only say He kicked the bucket last
night but not *He kick the bucketed last night. Similarly, so-called phrasal
verbs such as, for example, the fully-idiomatic come across 'to find by chance'
cannot be regarded as compounds because they are formally headed by their
lefthand components: we can only say He came across her diary but not *He
come acrossed her diary.
The general right-headedness of compounds in English allows us to discard
the endocentric–exocentric distinction, as far as the formal perspective is
concerned. Indeed, if the righthand member always functions as the formal head
of a compound, then all compounds can be regarded as endocentric compounds.
Even if there were left-headed compounds (e.g. *two fightersbomber), we would
still be justified in regarding them as endocentric compounds, i.e. if fighter-
bomber were headed by the lefthand component fighter rather than by the
righthand component bomber, it would nevertheless qualify as an endocentric
compound because in this putative compound the head position would still be
filled by one of its overt components, the lefthand member fighter.
Finally, let us consider the possibility of formal double-headedness. As the
term suggests, a double-headed compound is a compound that has two formal
heads. As an illustration, let us consider the fully-idiomatic verbal compound TO
DRAG AND DROP 'to move or copy (an image, icon, text, etc.) from one part of a
display screen to another using a mouse or similar device'. According to the
OED, the past tense form of this verb can be both dragged and dropped and
drag and dropped. In the latter case, we are dealing with an endocentric
compound which is headed by the righthand member drop. By contrast, dragged
and dropped can be regarded as a formally copulative compound, i.e. a
compound which is headed by both drag and drop. Alternatively, we can argue
that the lexeme TO DRAG AND DROP is realized by the following two allolexes: 1)
the endocentric compound lex drag and drop, whose past tense form is drag and
dropped and 2) the copulative phrase drag and drop, whose past tense form is
dragged and dropped.
In summary, from a semantic point of view, English compounds can be
classified into endocentric and exocentric compounds. The former are
compounds whose head meanings are inherent in one of their components'
156 Chapter 5
signifieds; the latter are compounds whose head meanings are idiomatic
meanings non-inherent in their components' signifieds. Endocentric compounds
include all quasi-idiomatic compounds of the information fatigue-type and semi-
idiomatic compounds like dadrock. Exocentric compounds include all quasi-
idiomatic compounds of the drum and bass-type (including bahuvrihis like
skinhead) and fully-idiomatic compounds like carpet muncher. There are no
copulative compounds: compounds like fighter-bomber and politician-tycoon
which are traditionally classed as copulative compounds belong in the category
of exocentric compounds.
From a formal point of view, English compounds can be classified into
endocentric and copulative compounds. The former are compounds like fighter-
bomber which are formally headed by their righthand components; the latter are
compounds like to drag and drop which are formally headed by both their
components. The overwhelming majority of English compounds are formally
endocentric. (Forms like to drag and drop can also be analyzed as copulative
phrases rather than as copulative compounds). There are no formally exocentric
compounds.
Since the term 'head' has both a formal and a semantic dimension, one and
the same compound can simultaneously qualify as e.g. an exocentric compound
from a semantic point of view and as an endocentric compound from a formal
point of view. E.g. fighter-bomber is a semantically exocentric, but formally
endocentric compound.
5.6.4 Compounding from a formal point of view
As regards the formal side of compounding, the most important question is

which input lexes can be combined into output lexes realizing new compound
lexemes. Considering the compounds that were discussed in the previous parts
of this section, we can already make several generalizations. First of all, nominal
compounds can represent combinations of:
1. two input nouns: e.g. FIGHTER-BOMBER

2. one input adjective and one input noun: e.g. BLUE STATE
3. one input verb and one input noun: e.g. PLAYGROUND
As for verbal compounds, consider the following recent formations:
x TO CROWD-SURF 'to engage in crowd-surfing' (OED / 1991)

x TO DUMBSIZE 'to dismiss (staff) in excessive numbers or without regard to

organizational function, with the result that work can no longer be carried out
effectively' (OED / 1993)
x TO MWAH-MWAH 'to kiss in exaggerated fashion, especially on the cheek; to

give air kisses' (OED / 1993)
x TO COPYPASTE '[to] copy[…] the contents of a document or a program to be

added to another document' (Urban Dictionary / 2004)
TO CROWD-SURF is, like the previously mentioned TO EGO-SURF, a noun + verb

compound: its input lexes are the noun crowd and the verb to surf.
TO DUMBSIZE is an adjective + verb compound: its input lexes are the
adjective dumb and the verb to size.
TO MWAH-MWAH is a reduplicative interjection + interjection compound:
its input lex is the interjection mwah 'representing the sound of a kiss,
deliberately exaggerated to convey superficiality or pretentiousness' (OED).
Finally, TO COPYPASTE is a verb + verb compound: its input lexes are the
verbs to copy and to paste.
Adjectival compound lexemes usually exhibit either 1) the adverb +
adjective pattern, exemplified by the already mentioned adjectival compounds
ALWAYS-ON 'always online' and DOWN-LOW 'secret, quiet' – both are products of
combining the input adverbial lexes always and down and the adjectival lexes on
and low – or 2) the adjective + adjective pattern. With regard to the latter,
consider the adjectival compound BI-CURIOUS 'of a heterosexual person:
interested in experiencing an (especially first) sexual encounter or relationship
with a person of the same sex' (OED / 1990). Its input lexes are the adjectival bi
(which is a colloquial abbreviation of bisexual) and the adjectival curious.
Finally, let us briefly dwell on the following special types of compounds:
x reduplicative compounds
x thing-compounds
x neo-classical compounds
Reduplicative compounds are compounds like the above mentioned TO MWAH-

MWAH whose lexes are formed via reduplication of one and the same input lex:
to mwah-mwah = mwah + mwah. Other examples include:
x TO NYAH-NYAH 'to behave in a childishly supercilious or derisive manner

towards someone; to taunt someone' (OED / 1986)
x PUM-PUM 'the female external genitals, the vagina […]' (OED / 1983)
158 Chapter 5
x the interjection BOOM-BOOM 'used (as a following tag or as a response) to

draw attention to a joke or pun, especially one the speaker or writer regards
as weak, obvious, or labored' (OED / 1972)
x the adjective GO-GO 'fashionable, 'swinging', 'fabulous', unrestrained; (of

funds on the stock exchange) speculative' (OED / 1962)
x the adverb NOW-NOW 'in the immediate future, in a moment; very soon' (OED
/ 1948)
Sometimes reduplicative compounding is accompanied by apophony. For

example:
x the adjective EASY-PEASY 'in childish or children's use (especially as

interjection): extremely easy, very simple' (OED / 1976)
x the noun RUMPY-PUMPY 'sexual intercourse' (OED / 1968)
x the interjection OKEY-DOKEY 'OK' (OED / 1932)
x the adjective FLIPPY-FLOPPY 'inclined to flop, having a tendency to flop about'

(OED / 1858).
Thing-compounds are noun + noun compounds like bodybuilding thing in (85)

and jet thing in (86).
(85) I don't like that bodybuilding thing where they got no neck at all, and
then every vein pops out (COCA)
(86) And it is really fantastic to have your own jet, and anybody who says it
isn't is lying to you. That jet thing is really good (COCA)
A notable peculiarity of thing-compounds is that they never become fully-

established lexemes and are therefore highly context-dependent (Lieber 2009:
365). Thus it is the following part of the sentence where they got no neck at all,
and then every vein pops out that enables us to understand that thing of
bodybuilding thing stands for the particular outcome of bodybuilding. And it is
the preceding sentence and it is really fantastic to have your own jet, and
anybody who says it isn't is lying to you that enables us to understand that thing
of jet thing stands for the state of possessing a jet.
Finally, neo-classical compounds are compounds that come into existence
via combining roots of Greek and Latin origin (either with each other or with
native roots). Consider, for example, the lexeme SARCOPENIA 'loss of skeletal
muscle mass as a result of ageing' (OED / 1991). According to the OED, the lex
of this lexeme is a product of combining the following roots of Greek origin:
sarco- (σαρκ-, σάρξ meaning 'flesh') and -penia (Greek πενία meaning 'poverty,
need'). However, despite the Greek origin of both sarco- and -penia, the lexeme
sarcopenia is not an instance of borrowing: the compounding of the forms
sarco- and -penia took place in English, not in Greek. Neoclassical compounds
like sarcopenia usually consist of bound roots that never occur in isolation.
(Given the general obligatoriness of roots, we cannot regard sarcopenia as a
combination of two affixes. Either sarco- or -penia must be regarded as a root.
The lex sarcopenia can thus be analyzed as either a compound segmentable into
the two bound roots sarco- and -penia or as an affixed word segmentable into
either the bound root sarco- and the suffix -penia or into the prefix sarco- and
the bound root -penia. Both analyses are equally plausible here.)
5.6.5 Compounds and phrases
One of the most controversial theoretical issues regarding compounding in

English is the question of how a compound can be distinguished from a phrase.
In 5.6.3 we argued that fully-idiomatic VPs like kick the bucket and fully-
idiomatic phrasal verbs like come across do not qualify as compounds because
they are headed by their lefthand members kick and come. Compounds, by
contrast, are right-headed: whereas the past tense forms of kick the bucket and
come across are kicked the bucket and came across, the past tense form of the
compound ego-surf is ego-surfed, not *egoed-surf. Unfortunately, the right-
headedness criterion is helpful only for distinguishing verbal compounds from
idiomatic VPs and phrasal verbs. However, this criterion does not allow us to
distinguish nominal compounds from idiomatic NPs. As an illustration, recall
the compounds HOT DESK /ˈhɒt dɛsk/ 'a shared office desk or workstation,
occupied on a temporary, ad hoc basis or part-time basis, and not allocated
permanently to an individual' (OED / 1990) and BLUE STATE /ˈblu steɪt/ 'a state
(projected to be) won by the Democratic candidate in a presidential election'
(OED / 2000), which were mentioned in 5.6.1. How do we know that the lexes
realizing these lexemes are indeed semi-idiomatic adjective + noun compounds
rather than semi-idiomatic adjective + noun NPs? To answer this question, we
cannot resort to the right-headedness criterion because both adjective + noun
compounds and adjective + noun NPs are right-headed: in both cases the plural
marking occurs on the righthand head noun; e.g. blackboards and black boards.
Of little help also are the spelling criterion and the meaning criterion that are
sometimes mentioned in connection with the compound–phrase distinction. As
regards spelling, recall that orthographic systems are artificial systems, which
appeared much later than spoken speech. Accordingly, the separation of hot and
160 Chapter 5
blue from desk and state by means of a blank space cannot be seen as a
corroboration of the phrasal status of hot desk and blue state. As regards the
meaning criterion that states that adjective + noun compounds are sometimes
more idiomatic than corresponding homonymous phrases – cf. e.g. the semi-
idiomatic signified of the compound a blackboard and the non-idiomatic
signified of the homonymous NP a black board – observe that semi-idiomatic
NPs do occur as well. For example, just like the compound blackboard, the
adjective + noun NP black coffee has a semi-idiomatic signified that contains the
signified of the component coffee but not the signified of the component black:
black coffee does not mean 'black coffee' but 'coffee without milk or cream'
(Mel'čuk 1995: 182). Does it follow from this that black coffee is a compound as
well?
The only reason why the lexes realizing the semi-idiomatic lexemes HOT
DESK and BLUE STATE can be regarded as compounds is their leftward stress:
/ˈhɒt dɛsk/ and /ˈblu steɪt/. The point here is that while the primary stress of an
NP falls on the righthand element (e.g. black cóffee, not *bláck coffee),
compounds are stressed on their lefthand members. Accordingly, since in both
hot desk and blue state, the stress falls on the lefthand members hot and blue, we
can claim that both these combinations represent semi-idiomatic compounds
rather than semi-idiomatic phrases. The same can be said about the lexes
realizing the following lexemes:
x NEW JILL /ˈnjuː ˌdʒɪl/ 'new jack swing as performed by women' (OED / 1990)
x RIOT GIRL /ˈrʌɪət gəːl/ 'a member or follower of any of several loosely
affiliated, mainly American, feminist rock or punk groups of the early 1990s'
(OED / 1991)
x UK GARAGE /ˌjuːˈkeɪ ˌgarɑː(d)ʒ/ 'a form of garage music […] originating in

the United Kingdom, retaining the emphasis on vocals but characterized by a
syncopated rhythm track influenced by drum and bass' (OED / 1992)
Since in new jill, riot girl, and UK garage, the primary stress falls on the
lefthand members new, riot, and UK, we are justified in regarding these forms as
compounds.
Now, consider the stress pattern of the lexes realizing the following five
lexemes:
x MARTIAL ARTISTRY /ˌmɑːʃl ˈɑːtᵻstri/ 'achievement or skill in a martial art'

(OED / 1990)
x ETHNIC CLEANSING /ˌɛθnɪk ˈklɛnzɪŋ/ 'the purging, by mass expulsion or

killing, of one ethnic or religious group by another, especially from an area
of former cohabitation' (OED / 1991)
x NEW LAD /ˌnjuː ˈlad/ 'a (type of) young man who embraces sexist attitudes
and the traditional male role as a reaction against the perceived effeminacy of
the 'new man'' (OED / 1991)
x HOME ZONE /ˌhəʊm ˈzəʊn/ 'a residential area in which a variety of traffic-
calming measures are employed to create a safer environment for pedestrians'
(OED / 1992)
x FREE RUNNING /ˌfriː ˈrʌnɪŋ/ 'the discipline or activity of moving rapidly and
freely over or around the obstacles presented by an (especially urban)
environment by running, jumping, climbing' (OED / 2003)
In all these forms the primary stress falls on their righthand members artistry,
cleansing, lad, zone, and running. Accordingly, the lexes martial artistry, ethnic
cleansing, new lad, home zone, and free running are all phrases, not compounds.
Many authors are rather skeptical about the applicability of the stress
criterion (e.g. Giegerich 2009: 184-185) for distinguishing compounds from
phrases. The main reason for this is that the difference in stress often
necessitates a different treatment of semantically related combinations. For
example, while the combination apple pie /ˌapl ˈpʌɪ/ qualifies as a phrase
because in British English it is stressed on the righthand element pie, the
combination apple cake /ˈapl ˌkeɪk / qualifies as a compound because its
primary stress falls on the lefthand element apple. For many authors, this
analysis is rather counter-intuitive, given the semantic similarities between the
combinations apple pie and apple cake.
In spite of this and other similar cases, this textbook argues for the stress
criterion. The distinction between a compound and a phrase is a formal rather
than a semantic distinction. This means that if we do not want to discard this
distinction (i.e. to regard compounds and phrases as manifestations of the same
formal category), we must ignore the fact that apple pie and apple cake express
similar meanings. The only thing that must be taken into account is the formal
difference between these combinations. That is, since apple pie is stressed on the
righthand element pie, it must be regarded as a phrase and since apple cake is
stressed on the lefthand element apple, it must be regarded as a compound.
Moreover, the combination apple pie, which in American English is stressed on
the lefthand element apple /ˈæpəl ˌpaɪ/, must be treated differently than the same
combination in British English: while the British apple pie is a phrase, the
American apple pie is a compound, even though both express the same meaning.
162 Chapter 5
Finally, let us briefly discuss the status of double-stressed combinations.

Consider again the above mentioned combination new lad /ˌnjuː ˈlad/. While in
British English this combination is characterized by the rightward primary stress
and thus qualifies as a phrase, in American English both new and lad receive the
primary stress: /ˈn(j)u ˈlæd/. What is the status of new lad in American English?
Is it a compound or a phrase? Or do we need a separate category for double-
stressed combinations like new lad?
According to Bloomfield (1973[1934]: 180), double-stressed combinations
must be regarded as phrases rather than as compounds. One rationale for this is
that combinations like new lad are no different from phrases like ethnic
cleansing in that one of their primary stresses also falls on their righthand
members. (The only difference is that combinations like new lad have two
primary stresses.) Another rationale for the phrasal solution is that words are
often said to be able to carry only one primary stress. That is, a combination of
two free morphs like new and lad of new lad both of which receive the primary
stress cannot be regarded as a compound simply because a compound is a word
and hence can only have one primary stress.
To conclude: the compound–phrase distinction is a formal, not a semantic
distinction. Accordingly, this distinction must be specified in terms of formal
properties of noun + noun and adjective + noun compounds distinguishing them
from corresponding phrasal noun + noun and adjective + noun combinations
which lack these properties. Of the available criteria, the difference in stress
seems to be the most promising criterion: combinations like hot desk and blue
state which stress their lefthand elements are compounds, whereas combinations
like ethnic cleansing and free running in which the primary stress falls on their
righthand elements as well as double-stressed combinations like new lad in
American English are phrases.
5.6.6 Productivity
Compounding is a very productive lexeme-building mechanism. As was pointed

out in 5.6.1, of 587 new signifiers which, according to the OED, appeared in the
English language between 1990 and 2011, 81 are instances of pure
compounding (i.e. not pseudo-compounds like TO BABYSIT).
As in the case of conversion, the Etymology-sections of the OED entries do
not contain the word compound. This means that the OED is not searchable for
compounds the way it is searchable for e.g. apophonies and borrowings. The
best means to study the synchronic productivity of compounding using the OED
is thus to make the OED search for words belonging to a particular word class
(e.g. verbs, nouns, adjectives) that were formed during a particular period of
time (e.g. after 1990) and then manually classify the compounds found into
instances of particular categories (e.g. noun + noun compounds, exocentric

compounds, neo-classical compounds, etc.). For example, if you want to find out
how many nominal compounds appeared in English after 1990, go to the OED
Advanced Search page http://www.oed.com/advancedsearch and choose the
options '1990-' in 'Date of entry' and 'Noun' in 'Part of speech'. The OED will
then yield all nominal lexemes which appeared in the English language after
1990. Then carefully read the Etymology-section of each entry. If it contains
information like 'Etymology < drum n.1 + and conj.1 + bass n' for drum and
bass or 'Etymology < no adj. + the plural of mate n.2' for No-Mates, you can be
justified in regarding the signifiers under analysis as products of noun-building
compounding.
The OED is searchable for compound lexemes whose lexes were formed via
reduplication of their input lexemes. Go to the above named OED Advanced
Search page, enter 'reduplication' to the first row of the search mask from above,
and choose the option 'Etymology' (near the search mask to which you will enter
'reduplication'). The OED will then yield all entries whose Etymology-sections
contain the word reduplication. Some of them will be instances of reduplicative
compounding like to mwah-mwah and to nyah-nyah.
5.7 Blending
Blending is similar to compounding in that it also produces new lexemes by

combining the lexes of two or more input lexemes into the lex of a new complex
lexeme. However, in contrast to compounding, blending is accompanied by the
shortening of at least one of the input lexes. For example, the compounding of
the input lex flu and tsunami into the blend flunami was accompanied by the
shortening of one of its input lexes: tsunami Æ nami. Similarly, the
compounding of the input lexes flexibility and security into the blend flexicurity
'labor practices that give companies the flexibility to fire workers as needed and
offer fired workers the security of government-backed benefits and retraining
opportunities' (Word Spy / 1997) was accompanied by the shortening of both of
its input lexes: flexibility Æ flexi and security Æ curity.
Sometimes it may be rather difficult to determine which of the input lexes
that gave rise to a new blended lexeme was subjected to shortening. Consider,
for example, the blend cheapuccino 'an inexpensive, low-quality cappuccino,
particularly one from a vending machine […]' (Word Spy / 2002). Has the lex of
this lexeme been created via the shortening of cheap into chea and cappuccino
into puccino (i.e. chea + puccino) or only via the shortening of cappuccino into
uccino (i.e. cheap + uccino)? Both analyses seem equally plausible here, so that
the exact diachronic history of this blend cannot be established. However, the
question of whether cheapuccino = chea + puccino or cheap + uccino does not
164 Chapter 5
seem to be an important theoretical question. The only thing that matters here is
that the lex under analysis is indeed a product of blending of the input lexes
cheap and cappuccino.
Like compounding, blending is an anisomorphic lexeme building-
mechanism, which always produces output lexemes whose signifieds are not /
not entirely representable in terms of their components' signifieds. As in the case
of compounding, quasi-idiomatization is the default semantic outcome of
blending. Thus of 20 blended signifiers which, according to the OED, appeared
in the English language between 1990 and 2011, 17 can be analyzed as quasi-
idioms. For example:
x EMOTICON (Å emotion + icon) 'a representation of a facial expression formed

by a short sequence of keyboard characters (usually to be viewed sideways)
and used in electronic mail, etc., to convey the sender's feelings or intended
tone' (OED / 1990)
x CLINTONOMICS (Å Bill Clinton + economics) 'the economic policies of

President Clinton' (OED / 1992)
x RACINO (Å racetrack + casino) 'a building complex or grounds having a

racetrack and gambling facilities traditionally associated with casinos, such
as slot machines, blackjack, roulette, etc.' (OED / 1995)
While the blended lexemes EMOTICON and CLINTONOMICS are quasi-idioms of

the information fatigue-type – that is, an emoticon is a particular icon that has
something to do with emotions and Clintonomics are particular economic
policies that have something to do with Clinton – the blend RACINO exhibits the
drum and bass-pattern: its blended components name two important
characteristics of building complexes that came to be referred to as racinos: they
contain both racetracks and casinos. The blends EMOTICON and CLINTONOMICS
are thus semantically endocentric blends, whereas RACINO is a semantically
exocentric blend.
Examples of semi-idiomatic blends include:
x SCREENAGER (Å screen + teenager) 'a young person (typically in his or her

teens or twenties) who is at ease with and adept at using new technology and
media, especially computers' (OED / 1994)
x MACHINIMA (Å machine + cinema or machine + cinema) 'the practice or

technique of producing animated films using the graphics engine from a
video game' (OED / 2000)
Both these blends involve a metonymic modification of their first components

screen and machine: screen of screenager stands for a new technology which
typically involves screens (e.g. computer screens) and machine of machinima
stands for computers with the help of which such films are created.
The only example of a fully-idiomatic blend yielded by the OED is CRUNK
(Å crazy + drunk or crazy + drunk) 'exciting or fun; (of a person) extremely
energized or excited, especially as a result of listening to (usually hip-hop or
rap) music' (OED / 1995).
5.7.1 Productivity
Blending is often described as a minor lexeme-building mechanism in English.

Indeed, while e.g. compounding has recently given rise to 81 new lexemes,
blending has created only 20 new lexemes. At the same time, note that the Word
Spy database lists 189 lexemes whose lexes have recently come into existence
via blending. For example:
x MANUFACTROVERSY (Åmanufactured and controversy) 'a contrived or non-

existent controversy, manufactured by political ideologues or interest groups
who use deception and specious arguments to make their case' (Word Spy /
2008)
x CELEBUTARD (Åcelebutante and retard or celebutard and retard) 'a celebrity

who is or is perceived to be unintelligent' (Word Spy / 2006)
x RENOVICTION (Årenovation and eviction or renovation and eviction) 'the

mass eviction of an apartment building's tenants because the building's owner
plans a large renovation' (Word Spy / 2008)
x CHURNALISM (Åchurn and journalism) 'journalism that churns out articles

based on wire stories and press releases, rather than original reporting' (Word
Spy / 2001)
x SPIME (Åspace and time) 'a theoretical object that can be tracked precisely in
space and time over the lifetime of the object' (Word Spy / 2004)
x LUSER (Å loser and user) 'a person who doesn't have the faintest idea what
they're doing and who, more importantly, refuses to do anything about it'
(Word Spy / 1990)
166 Chapter 5
x BLOGEBRITY (Åblog and celebrity) 'a famous or popular blogger' (Word Spy
/ 2005)
x WOMENOMICS (Å women and economics) 'the theory that women play a

primary role in economic growth' (Word Spy / 1995)
x WARMEDY (Å warm and comedy ) 'a comedy that features warm-hearted,

family-oriented content' (Word Spy)
According to the author of the Word Spy database, "blends are the engines that
drive much of neology". In other words, blending is currently one of the most
productive lexeme-building mechanisms in English.
To study the productivity of blending using the OED, go to the OED
Advanced Search page http://www.oed.com/advancedsearch, enter 'blend' to the
search mask of the first row from above, and choose the option 'Etymology'
(near the search mask to which you will enter 'blend'). The OED will then yield
all entries whose Etymology-sections contain the word blend. Some of them will
be instances of blending like the above mentioned emoticon, racino, screenager,
etc.
To study the productivity of blending using Word Spy, go to the Advanced
Search page http://wordspy.com/search.asp, enter 'blend' to the search mask, and
choose the option 'Anywhere'. Word Spy will then find all items whose
descriptions contain the word blend. Since blended signifiers are almost always
described by the database as instances of blending, the majority of the results
yielded by Word Spy will indeed be recently created blended lexemes like
manufactroversy, spime, luser, etc.
5.8 Idiomatization of phrases and sentences
Like compounding and blending, phrasal idiomatization is an anisomorphic

lexeme-building mechanism that produces new lexemes via combining the lexes
of more than one input lexeme into the lex of a new complex lexeme. In fact, as
we established in 5.6.5, idiomatic phrases (especially idiomatic NPs such as, for
example, new lad) are sometimes hardly distinguishable from idiomatic
compounds.
As we said in 3.2.5, idiomatic phrases and sentences exhibit the same
degrees of idiomaticity as idiomatic words. That is, there are not only quasi-
idiomatic words like football but also quasi-idiomatic phrases like the VP start a
family and quasi-idiomatic sentences like the pick-up line Would you like to
have morning coffee with me?, whose signifieds contain not only their
components' signifieds but also some unpredictable idiomatic signifieds.
Similarly, there are not only semi-idiomatic words like blackboard but also
semi-idiomatic phrases like the VP answer the door and the pick-up line Didn't I
see you on the cover of Vogue?, whose signifieds contain only some of their
components' signifieds. Finally, there are not only fully-idiomatic words like
boyfriend but also fully-idiomatic phrases like the VP kick the bucket and the
proverb A bird in the hand is worth two in the bush, whose signifieds contain
none of their components' signifieds.
5.8.1 Productivity
Like compounding and blending, phrasal idiomatization is a fairly productive

lexeme-building mechanism in English. Thus the Idioms-section of the Word
Spy database lists 18 recently formed idiomatic phrases. Below are some of the
examples:
x TO STARVE THE BEAST 'to cut taxes with the intent of using the reduced
revenue as an excuse to drastically reduce the size and number of services
offered by a government' (Word Spy / 1981)
x TO PUT SKIN IN THE GAME 'to take an active interest in a company or

undertaking by making a significant investment or financial commitment'
(Word Spy / 1993)
x TO PAINT THE TAPE 'to increase the price of the stock by using unscrupulous
methods (such as breaking up a large stock purchase into multiple small
purchases to give the illusion of a buying frenzy)' (Word Spy / 2000)
x DOG THAT CAUGHT THE CAR 'a person who has reached their goal but doesn't
know what to do next' (Word Spy / 1985)
x DOG WATCHING TV 'a person who is viewing or working with something

without understanding what it is or what it does' (Word Spy / 1997)
Fully-idiomatic phrases and sentences have traditionally been the focus of

attention of phraseology rather than morphology. The explanation for this is that
morphology, as a branch of linguistics, is supposed to be concerned with words,
not with phrases and sentences. Similarly, lexicographers tend to regard phrasal
idiomatization as a special lexeme-building mechanism distinct from
conversion, affixation, compounding, and the like. As a consequence, a number
of established idiomatic phrases can only be found in specialized idiom
dictionaries, but not in 'word' dictionaries like the OED and the MWO.
168 Chapter 5
One specialized dictionary that can be recommended for the study of

idiomatic phrases is Ayto et al's Oxford Dictionary of English Idioms (2009),
which contains more than 6000 entries. An entry in this dictionary is usually a
noun or a verb that occurs in multiple idiomatic phrases: e.g. the entry for the
noun door lists such phrases as:
x at death's door
x beat a path to someone's door
x blow the doors off
x by the back door
Like the OED, this dictionary can be used online. However, to gain access to the
dictionary, you need to subscribe to Oxford Reference Online Premium services.
Apart from this, there are a number of freely-available databases collecting
English idiomatic phrases. For instance, the already mentioned Phrase Finder
(http://www.phrases.org.uk/) is a good resource for studying the etymologies of
idiomatic phrases.
5.9 Back-formation
Back-formation is the removal of a derivational affix (or a part of an input lex

perceived as a derivational affix) from the lex of an input lexeme accompanied
by quasi-idiomatization of the signified of an input lexeme. For example, the
removal of -er from the input lex Taser was accompanied by the quasi-
idiomatization of the input signified 'Taser': while a Taser means 'a Taser', to
tase came to mean 'to incapacitate using a Taser'. Likewise, the removal of -y
from the input lex skeevy was accompanied by the quasi-idiomatization of the
input signified 'skeevy': while skeevy means 'skeevy', a skeeve came to mean 'a
skeevy person'. Similar examples include:
x TO CARJACK (Åcarjacking) 'to steal or to commandeer an occupied car by

threatening the driver with violence' (OED / 1991)
x TACK (Åtacky) 'that which is 'tacky' or cheap and shabby; shoddy or gaudy
material; rubbish, junk' (OED / 1986)
x A DITZ (Åditzy) 'a person who is 'ditzy', scatterbrained, or cute' (OED / 1984)
x TO DIVISIONALIZE (Ådivisionalization) 'to organize (a company, etc.) on a

divisional basis' (OED / 1982)
x A SHONK (shonky) 'one engaged in irregular or illegal business activities; a

'shark'' (OED / 1981)
x TO EXFILTRATE (Åexfiltration)' to withdraw (troops, spies, etc.) from a

dangerous position' (OED / 1980)
x TO CHEMOTAX (Åchemotaxis) 'to exhibit chemotaxis; to move in response to

certain chemical substances' (OED / 1979)
x TO INCENT (Åincentive ) 'to provide (a person) with an incentive; to

encourage, incite, inspire' (OED / 1977)
x A GIGAFLOP (Ågigaflops) 'a unit of computing speed equal to 1000

megaflops' (OED / 1976)
As these examples illustrate, back-formation typically gives rise to verbal and

nominal lexemes. As for adjectives, it appears that back-formation is no longer
used for producing new adjectival lexemes: according to the OED, the last
adjectival back-derivative is INTERTEXTUAL (Å intertextuality) 'denoting literary
criticism which considers a text in the light of its relation to other texts; also
used of texts so considered' (OED / 1973). Since the early 1970s not a single
adjectival lexeme has been formed by means of back-formation.
As can be inferred from what was said in 4.2.4, the most important
theoretical question raised by back-formation is when English speakers
'incorrectly' reanalyze instances of back-formation as input lexemes. That is,
while to tase still means 'to incapacitate using a Taser' and, accordingly, can be
regarded as an instance of back-formation not only from a diachronic but also
from a synchronic point of view, to babysit and to proofread no longer mean 'to
do the job of a babysitter' and 'to do the job of a proofreader': as we argued in
4.2.4, the semantic relation holding between 'to babysit' / 'to proofread' and 'a
babysitter' / 'a proofreader' is in Present-day English no different from the
semantic relation holding between the signifies 'to blog' and 'a blogger'.
Accordingly, from a synchronic point of view, the signifieds 'to babysit' / 'to
proofread' must be regarded as input signifieds for 'a babysitter' / 'a proofreader'.
In which respects is TO TASE different from TO BABYSIT and TO PROOFREAD?
The first obvious difference is that while TO TASE is a relatively recent
formation – it has been recorded in English since 1991 – the earliest citation of
TO BABYSIT dates 1947 and the earliest citation of TO PROOFREAD dates 1845.
Given these dates, we can conjecture that speakers of English have already
forgotten the true etymologies of the lexemes TO BABYSIT and TO PROOFREAD,
but they have not forgotten the true etymology of TO TASE.
170 Chapter 5
Another important difference is that in contrast to the lex to tase, the lexes to
babysit and to proofread do not fulfill the additional naming requirement.
(Dobrovol'skij and Piirainen 2005: 18). That is, to babysit is a primary lex (i.e.
the most basic signifier) expressing the signified 'to take care of a baby during
the temporary absence of the parents'. Likewise, to proofread is the most basic
signifier expressing the signified 'to read a proof, identify mistakes in it, and
make the necessary corrections'. By contrast, to tase seems to be an additional
expression for the signified 'to incapacitate a person using a Taser'. Usually this
signified is expressed by the euphemistic VP to temporarily incapacitate a
person. The point here is that in the case of primary expressions like to babysit
and to proofread, language users often 'get rid' of the diachronic memory: there
is absolutely no need to remember that e.g. to babysit and to proofread came
into existence via back-formation of the corresponding nouns babysitter and
proofreader. In contrast, in the case of additional expressions like to tase, the
true etymology usually provides a motivating link explaining why a particular
signifier can be used as an additional signifier expressing a particular signified.
That is, for example, the fact that to tase came into existence via back-formation
of Taser provides a synchronic motivation for the possibility of the use of that
signifier for expressing the signified 'to incapacitate a person using an
electroshock weapon like a Taser'. If it were not for this motivating link between
the signifier to tase and the signified 'to incapacitate somebody using a Taser',
the lexeme TO TASE would be synchronically opaque. (For a more detailed
discussion of the additional naming requirement, see Tokar 2009: 10-11).
5.9.1 Productivity
At present, back-formation does not seem to be a productive lexeme-building

mechanism in English. According to the OED, the most recent instances of
back-formation – the verbal lexemes TO TASE and TO CARJACK – appeared in
1991. Since then there have been no new back-formations.
To study the history of this lexeme-building mechanism, go to the OED
Advanced Search page at http://www.oed.com/advancedsearch, enter 'back-
formation' to the search mask of the first row from above, and choose the option
'Etymology' (near the search mask to which you will enter 'back-formation').
The OED will then yield all entries whose Etymology-sections contain the word
back-formation. Some of them will be signifiers like to tase and to carjack
which came into existence via back-formation. You can refine your search by
choosing the option 'Noun' and entering e.g. '1900-' to 'Date of entry'. The OED
will then yield all nominal signifiers which came into existence via back-
formation between 1900 and 2011.
5.10 Exercises
a) Metonymy and metaphor are two different mechanisms of semantic change.

b) Morphological conversion is no longer a productive lexeme-building
mechanism in English.
c) Arbitrary formation is the most productive lexeme-building mechanism in
Present-day English.
d) English has extensively borrowed from other languages.
e) The addition of a derivational suffix often produces an output lexeme whose
lex is a member of a different word class than the lex of a corresponding
input lexeme.
f) Compounds in English are left-headed.
g) The head meaning of an exocentric compound is not inherent in its
components' signifieds.
h) Full-idiomatization is the default semantic outcome of compounding.
i) Adjective + noun compounds have the leftward stress and are in this respect
different from phrases.
j) Blending is now a very productive lexeme-building mechanism in English.
3. Using the OED, establish how many verbs which appeared in English in
1989 are products of morphological conversion.
4. Using the OED, establish which of the following languages – German or

Spanish – played a greater role in the enlargement of the English vocabulary
in the 20th century.
5. Using the OED, establish whether lexeme-manufacturing was a productive

lexeme-building mechanism in English between the years 1300 and 1350.
6. Using the OED, establish how many verbs which appeared in English
between the years 1900 and 1950 can be regarded as instances of apophony.
7. Using the OED, establish how many nouns which appeared in English in
1990 are products of affixation by means of the suffix -er.
8. Using the OED, establish whether blending gave rise to more new lexemes
between the years 1600-1649 or between the years 1900-1949.
172 Chapter 5
9. Using the OED, establish how many nominal lexemes were formed in
English via back-formation between the years 1700-1749.
5.11 Further reading
A classic monograph dealing with lexeme-formation in English is Marchand

(1969). A more recent discussion of the most important theoretical issues
pertaining to English word-formation can be found in the recent Handbook of
Word-Formation (Štekauer and Lieber 2005) and The Oxford Handbook of
Compounding (Lieber and Štekauer 2009b). A recent study of the directionality
of conversion in English is Balteiro (2007). A recent study of the semantics of
compounding in English is Benczes (2006).
Due to space limitations, this chapter does not discuss lexeme-formation in
Old and Middle English. The reader is referred to the articles Kastovsky (1992)
and Burnley (1992) from The Cambridge History of the English Language
(Volumes I and II). The former discusses lexeme-formation in Old English; the
latter deals with the Middle English period.
6 Inflectional morphology
Having discussed both the lex- and lexeme-building mechanisms, we can finally
proceed to wordform-building mechanisms, i.e. mechanisms like inflectional
affixation producing output allolexes which express different grammatical
meanings than corresponding input lexes (e.g. talked and talk, books and book,
prettier and pretty). The chapter has the following structure. Section 6.1
provides a more precise definition of the term 'grammatical category', which was
already introduced in 1.1. Section 6.2 classifies grammatical categories into
syntactic and semantic grammemes. Section 6.3 introduces all wordform-
building mechanisms with the help of which speakers of English create
wordforms like talked, books, prettier, etc. Finally, Sections 6.4 and 6.5 dwell
on the most important theoretical issues pertaining to both syntactic and
semantic grammemes in English.
6.1 Grammatical category
A grammatical category is the set of mutually exclusive grammatical meanings

such as, for example, 'the singular number' and 'the plural number' (forming the
grammatical category NUMBER) or 'the present tense' and 'the past tense'
(forming the grammatical category TENSE).
To explain what is meant by the 'mutual exclusiveness' of grammatical
meanings , let us recall what we said about the contrast between the past tense
meaning inherent in the inflectional suffix -ed of e.g. talked, walked, worked and
the past time meaning inherent in the derivational prefix ex- of e.g. ex-
ambassador, ex-boyfriend, ex-president: while the latter is an optional lexical
meaning that is expressed only when we want to specifically refer to people who
are former ambassadors, boyfriends, and presidents, the former is an obligatory
grammatical meaning, i.e. a meaning which cannot be unexpressed. This
characterization of grammatical meanings cannot but raise the following
objection: if the past tense meaning inherent in -ed were indeed an obligatory
meaning, than all verbal lexemes would have only past tense wordforms (talked,
walked, worked). Similarly, if the meaning 'the plural number' were an
obligatory nominal meaning, than all nominal lexemes would consist of only
plural wordforms like books, chairs, tables. But as we know, verbal lexemes
have present tense wordforms (talks, walks, works) and nominal lexemes have
singular wordforms (book, chair, table).
As Plungian (2000: 107) points out, obligatoriness is not a characteristic of a
particular grammatical meaning such as 'the past tense' or 'the plural number' but
174 Chapter 6
of a set of mutually exclusive grammatical meanings of which that particular

meaning is a member. For example, the meaning 'the past tense' is a member of
the set of the mutually exclusive meanings 'the present tense' / 'the past tense';
the meaning 'the plural number' is a member of the set of the mutually exclusive
meanings 'the singular number' / 'the plural number'; the meaning 'the
comparative degree of comparison' is a member of the set of the mutually
exclusive meanings 'the positive degree of comparison' / 'the comparative degree
of comparison' / 'the superlative degree of comparison'; etc.
The mutual exclusiveness of grammatical meanings means that in no case
can two meanings of the set in question be simultaneously expressed by one and
the same wordform: e.g. there can be no verbal wordforms that express both the
meanings 'the present tense' and 'the past tense' and there can be no nominal
wordforms that express both the meanings 'the singular number' and 'the plural
number'. But one of the meanings of the set must always be expressed by a
given wordform. That is, a wordform of a verbal lexeme must express either the
meaning 'the present tense' or 'the past tense'; a wordform of a nominal lexeme
must express either the meaning 'the singular number' or 'the plural number'; a
wordform of an adjectival lexeme must express either the meaning 'the positive
degree of comparison' or 'the comparative degree of comparison' or 'the
superlative degree of comparison'; etc.
Meanings that are members of such mutually exclusive sets of grammatical
meanings form a grammatical category (or a grammeme). For example, as said
above, the mutually exclusive verbal meanings 'the present tense' and 'the past
tense' form the grammatical category TENSE; the mutually exclusive nominal
meanings 'the singular number' and 'the plural number' form the grammatical
category NUMBER; the mutually exclusive adjectival meanings 'the positive
degree of comparison', 'the comparative degree of comparison', and 'the
superlative degree of comparison' form the grammatical category DEGREES OF
COMPARISON or simply GRADE; etc.
Just as a signifier of a morpheme is named a morph and a signifier of a
lexeme a lex, a signifier that expresses a particular grammatical meaning and
thus realizes a particular grammeme can be called a gram (or a grammatical
marker). For example, -ed of e.g. He worked is a past tense gram; -s of books is
a plural gram; -er of prettier is a comparative gram; etc.
6.2 Types of grammatical categories
Following Plungian (2000: Sec. 2.1), we will classify English grammatical

categories into syntactic and semantic grammemes. The difference between
them is that while syntactic grammemes produce output wordforms that differ
from their input wordforms only with regard to their syntactic functioning but
Inflectional morphology 175
not with regard to their referential meanings, semantic grammemes produce

output wordforms that are referentially different from their corresponding input
wordforms. As an illustration of a syntactic grammeme, let us compare (87) and
(88).
(87) He met the President

(88) The President was met by him
The obvious grammatical difference between the two clauses is that while (87)
is in the active voice, (88) is in the passive voice. However, despite this
grammatical difference, both clauses can be used to refer to one and the same
meeting situation: some male person referred to in (87) as he and in (88) as him
met another person (either male or female) referred to in both (87) and (88) as
the President. The difference between the two sentences is thus not that of
semantics but that of syntax: the NP the President, which in (87) functions as
object of the predicate VP met the President, is in (88) promoted to the subject.
The grammatical category VOICE is thus a syntactic grammeme that gives rise to
output wordforms like was met of (88) that have the same referential meanings
as corresponding input wordforms like met of (87).
As an illustration of a semantic grammeme, let us consider the referential
meanings of the singular wordform book and the plural wordform books.
Evidently, the two wordforms cannot refer to the same object: while the singular
book typically refers to a single representative of the class of books, the plural
books typically refers to more than one representative of the same class of
objects. Accordingly, NUMBER is a semantic grammeme that gives rise to output
nominal wordforms like books that have different referential meanings than
corresponding input wordforms like book.
Semantic grammemes considerably outnumber syntactic grammemes: in
English, apart from the grammatical category VOICE, only CASE can also be
regarded as a syntactic grammeme. There is no referential difference between
e.g. the nominative wordform he and the accusative wordform him: one and the
same male person can be referred to as both he and him. By contrast, there is a
referential difference between:
x an event that took place before the moment of utterance (e.g. He met the
President) and an event that is taking place at the moment of utterance (e.g.
He is meeting the President).
x an event that is taking place at the moment of utterance (e.g. He is meeting

the President) and a habitual situation recurring on a regular basis (e.g. He
meets the President).
176 Chapter 6
x a real event that indeed took place (e.g. He met the President) and an
imaginary event (e.g. If he had met the President, he would have …).
x a speaker who utters a particular utterance (I) and the addressee of the same
utterance (you).
x some representative of a particular class of objects (e.g. a good book) and the
best representative of the same class (e.g. the best book).
Accordingly, we are justified in regarding the grammatical categories TENSE,

ASPECT, MOOD, PERSON, and DEGREES OF COMPARISON as semantic grammemes,
which, like NUMBER, also produce output wordforms whose referential meanings
are not identical with those of their input wordforms.
6.3 Wordform-building mechanisms
1. inflectional affixation
2. analytic formation
3. grammatical apophony
4. grammatical suppletion
5. signifier-sharing
6.3.1 Inflectional affixation
This wordform-building mechanism produces the majority of wordforms in

English. These include:
x nominal plural wordforms: e.g. books Å book + -s

x nominal genitive wordforms: e.g. father's Å father + -'s
x pronominal genitive wordforms: e.g. his Å he + -s
x pronominal accusative wordforms: e.g. him Å he + -m
x adjectival comparative wordforms: prettier Å pretty + -er
x adjectival superlative wordforms: prettiest Å pretty + -est
x verbal third person present tense wordforms: e.g. (he) reads Å read + -s
x verbal past tense wordforms: e.g. (he) worked Å work + -ed
x ordinal numerals: e.g. seventh Å seven + -th
In addition to these, inflectional affixation also produces participial forms.

These include:
x so-called participle I, which, in combination with the auxiliary be, forms

verbal progressive wordforms: e.g. (he) was reading Å was+ read + -ing
x so-called participle II, which, in combination with the auxiliaries be and

have, forms passive and perfect wordforms: e.g. (it) was created Å was +
create + -ed and (he) has created Å has + create + -ed
6.3.2 Analytic formation
Analytic formation is the addition of an analytic form such as, for example, the
adjectival and the adverbial comparative and superlative grams more and most.
For instance:
x more beautiful / more beautifully Å more + beautiful / beautifully

x most beautiful / most beautifully Å most + beautiful / beautifully
Apart from this, analytic formation can also produce wordforms together with
inflectional affixation. This is true of the just mentioned:
x progressive wordforms: e.g. (He) was reading Å was + read + -ing

x passive wordforms: e.g. (It) was created Å was + create + -ed
x perfect wordforms: e.g. (He) has created Å has + create + -ed
6.3.3 Grammatical apophony
Similar to lexical apophony, grammatical apophony can be defined as any

modification of an input lex which does not qualify as an instance of inflectional
affixation. For example, the plural wordform mice /maɪs/ cannot be segmented
into the root /ms/, expressing the meaning 'mouse', and the infix /ʌɪ/, expressing
the meaning 'plurality'. Accordingly, mice /maɪs/ can be regarded as a product of
apophony of the input singular wordform mouse /maʊs/. Mice came into
existence via vowel change: /aʊ/ of mouse Æ /aɪ/ in mice. Similarly, the past
tense wordform met /met/ cannot be segmented into the root /mt/, expressing the
meaning 'meet', and the infix /e/, expressing the meaning 'the past tense', and
hence can also be regarded as a product of apophony: /iː/ of meet Æ /e/ in met.
In English, grammatical apophony usually gives rise to so-called irregular
wordforms. For example, irregular plural wordforms of nominal lexemes are
178 Chapter 6
those that do not take the regular plural affix -s. Apart from mice, these
include, for example:
x teeth (*tooths)
x geese (*gooses)
x feet (*foots)
x lice (*louses)
x men (*mans)
Note that vowel change illustrated by mouse Æ mice, tooth Æ teeth, goose Æ
geese, foot Æ feet, etc. is not the sole mechanism yielding irregular plural
wordforms in English. For example, the irregular plural wordform oxen is a
product of inflectional affixation of the input singular wordform ox by means of
the irregular plural affix -en.
In addition to producing the irregular plural wordforms named above,
English uses grammatical apophony for creating the following wordforms:
x irregular third person present tense wordforms

x irregular past tense wordforms
x irregular participle II forms
As an illustration of the first category, let us compare the present tense

wordforms have /hæv/ and has /hæz/. As analyzed by Palmer et al. (2002: 1599),
/hæz/ is a product of inflectional affixation of the input wordform /hæv/ by
means of the regular third person suffix -s. However, in contrast to e.g. save
/seɪv/, whose present tense third person wordform is /seɪvz/, the third person
wordform of have is not */hævz/ but /hæz/. In other words, /hæz/ Å /hæv/ + /z/.
That is, the affixation of the input wordform have by means of the regular suffix
-s is accompanied by the reduction of the final consonant [v]. Obviously, the
latter cannot be regarded as an instance of affixation (for have cannot be
segmented into the root /hæ/ and the suffix /v/) and hence constitutes an instance
of apophony. A similar example is the wordform does /dʌz/. Like has, does can
be analyzed as a product of affixation of the input wordform do /duː/ by means
of the regular suffix -s. However, in contrast to has, the formation of does is
accompanied not by a consonant reduction but by a vowel change. That is, [uː]
of do changes into [ʌ] in does. Again, as in the case of has, this modification
cannot be regarded as an instance of affixation (for do is not segmentable into
the root /d/ and the suffix /uː/) and, accordingly, constitutes an instance of
apophony.
As far as past tense and participle II wordforms are concerned, we can find
irregular forms that came into existence via both pure apophonies and
apophonies accompanying inflectional affixation. With regard to the latter,
consider the wordform had /hæd/. Similar to has, had can be analyzed as a
product of affixation of have by means of the regular past tense suffix -ed
accompanied by the reduction of the final consonant [v]. That is, /hæd/ = /hæv/
+ /d/. Examples of pure apophonies include:
x break Æ broke ([eɪ] > [əʊ])

x come Æ came ([ʌ] > [eɪ])
x drink Æ drank ([ɪ] > [æ])
x eat Æ ate ([iː] > [e])
x find Æ found ([aɪ] > [aʊ])
x give Æ gave ([ɪ] > [eɪ])
x hold Æ held ([oʊ] > [e])
x take Æ took ([eɪ] > [ʊ])
x wake Æ woke ([eɪ] > [əʊ])
As in the case of irregular wordforms of nominal lexemes, irregular wordforms

of verbal lexemes are not necessarily products of grammatical apophony.
Consider, for example, the participle II given /ˈgɪv.ən/. While the past tense
wordform gave /geɪv/ is a product of vowel change of the input present tense
wordform give /gɪv/, given came into existence via affixation of give by means
of the irregular inflectional suffix -en. Similarly, the irregular participle II taken
/ˈteɪ.kən/ is not a product of apophony but of affixation of the input wordform
take /teɪk/ by means of the irregular inflectional suffix -en.
Finally, observe that grammatical apophony participates in the production of
the following pronominal wordforms:
x the genitive wordform your

x the genitive wordform their
x the accusative wordform them
Your is a product of vowel change of you: the vowel [uː] of you changes into
[ɔː(r)] in your. Their is a product of affixation of they by means of the suffix -ir
accompanied by the reduction of the vowel [ɪ]: /ðeə(r)/ Å /ðeɪ/ + /ə(r)/. Them is a
product of affixation of they by means of the suffix -m accompanied by the
reduction of the vowel [ɪ]: /ðem/ Å /ðeɪ/ + /m/.
In contrast to the irregular nominal and verbal wordforms discussed above,
these wordforms cannot be regarded as irregular pronominal wordforms: as will
be shown below, the formation of genitive and accusative pronouns does not
have a regular mechanism comparable to the affixation of nouns by means of -s
and the affixation of verbs by means of -ed.
180 Chapter 6
6.3.4 Grammatical suppletion
Grammatical suppletion is the creation of a wordform by using a signifier that

has a different root than a corresponding input wordform. In English this
mechanism produces:
x pronominal genitive wordforms: e.g. my Å I

x pronominal accusative wordforms: e.g. us Å we
x pronominal person wordforms: e.g. you Å I
x pronominal plural wordforms: e.g. we Å I
x adjectival comparative wordforms: e.g. better Å good
x adjectival superlative wordforms: e.g. worst Å bad
x verbal person wordforms: e.g. am Å be
x past tense verbal wordforms: e.g. went Å go
x ordinal numerals: e.g. first Å one
With the exception of pronominal wordforms, all other instances of grammatical

suppletion are irregular wordforms. That is, the suppletive better and worst are
irregular comparative and superlative adjectival wordforms (cf. prettier Å
pretty + -er and coldest Å cold + -est). Similarly, the suppletive am is an
irregular verbal present tense wordform (cf. go of I go) and the suppletive went
is an irregular verbal past tense wordform (cf. I worked Å work + -ed).
As for the pronominal suppletive wordforms my and us, we cannot say that
they are irregular pronominal wordforms because pronouns do not seem to have
regular genitive and accusative affixes. It is true that the addition of the suffix -s
produces the genitive wordforms his (Å he + -s), its (Å it + -s), whose (Å who
+ -s) and the addition of the suffix -m produces the accusative wordforms him
(Å he + -m), them (Å they + -m), and whom (Å who + -m). At the same time,
suppletion produces not only my and us but also me (Å I), our (Å we), and her
(Å she). In addition, as we established in 6.3.3, your is a product of apophony
of you, and their and them are products of affixation of the nominative
wordform they by means of the suffixes -ir and -m accompanied by the
reduction of the vowel /ɪ/. It is not clear which of these mechanisms must be
regarded as the regular genitive- and accusative-building mechanism.
As regards personal pronouns differing with regard to person, we can argue
that suppletion is the default person-building mechanism. Thus the first person I,
the second person you, and the third person he, she, it are wordforms that have
different roots and, accordingly, can be analyzed as products of suppletion.
Consequently, if we analyze the first person I as the input wordform, then both
the second person you and the third person he, she, it can be analyzed as
suppletive output wordforms.
6.3.5 Signifier-sharing
Finally, signifier-sharing is the creation of a wordform by not changing an input

signifier. For example, the signifier deer /dɪə(r)/ is both the singular and the
plural wordform of the nominal lexeme DEER. Similarly, the signifier put /pʊt/ is
both the present tense and the past tense wordform of the verbal lexeme TO PUT.
In addition to nominal plural wordforms and verbal past tense wordforms,
signifier-sharing also produces:
x nominal plural genitive wordforms: e.g. fathers' (Å fathers)

x the pronominal accusative wordform you: cf. I gave you … and You are …
x the pronominal genitive wordform her: cf. Her book and I gave her …
x verbal imperative wordforms: cf. Read it! and I read it
x verbal subjunctive wordforms: cf. It's time that he read it and He read it
Apart from these wordforms, signifier-sharing is characteristic of verbal

wordforms differing with regard to the categories PERSON and NUMBER. That is,
for example, we say he goes, she goes, and it goes, but I go, you go, and they go.
Consider also the signifier-sharing between past tense wordforms like
worked and found of he worked and he found and participles II like worked and
found of he has worked and he has found. Notice that this instance of signifier-
sharing does not qualify as an instance of wordform-building signifier-sharing.
What we call participle II is a form of a verbal lexeme like TO WORK and TO FIND
that is used for producing perfect and passive wordforms. That is, a perfect
wordform of TO WORK is has worked (i.e. a combination of the auxiliary have
and the participle II worked) and a passive wordform of TO FIND is was found
(i.e. a combination of the auxiliary be and the participle II found). In other
words, the past tense wordform worked is not identical with the perfect
wordform has worked and the past tense wordform found is not identical with
the passive wordform was found.
6.3.6 Allowordforms
Some lexemes have more than one wordform for expressing the same
grammatical meaning. In this case, we are dealing with allowordforms. For
example, the affixed form noisier and the analytic formation more noisy are the
comparative allowordforms of the adjectival lexeme NOISY. (The majority of
allowordforms can be found among adjectival comparative and superlative
mono- and polysyllabic wordforms, which can often be formed both
inflectionally and analytically.) Similarly, formulae and formulas are the two
plural allowordforms of the nominal lexeme FORMULA: the formal difference
182 Chapter 6
between these allowordforms is that while formulas is a product of affixation of

the input wordform formula by means of the regular plural suffix -s, formulae is
an instance of apophony of formula: the final [ə] of /ˈfɔːmjʊlə/ changes into [iː]
in /ˈfɔːmjʊliː/. Finally, webcast and webcasted can be regarded as the two past
tense allowordforms of the verbal lexeme TO WEBCAST 'to broadcast live over
the Internet; to make viewable in real time by users of a web site' (OED / 1995):
the difference between them is that while webcasted is an instance of affixation
of the input wordform webcast by means of the regular past tense suffix -ed,
webcast is a product of signifier-sharing.
6.3.7 Productivity
As in the case of lexeme-building mechanisms, a question can be raised as to

which of the above mentioned wordform-building mechanisms are more / less
productive in Present-day English. To answer this question, let us again consider
the distinction between regular and irregular wordforms, which was introduced
in 6.3.3. We said that regular nominal plural wordforms are those that contain
the regular plural suffix -s and regular verbal past tense forms are those that
contain the regular past tense suffix -ed. But why is it actually the case? How do
we know that -s is indeed the regular plural suffix for nominal lexemes and -ed
is the regular past tense suffix for verbal lexemes?
The answer to this question seems to be the fact that -s and -ed are the
suffixes that a speaker of English is most likely to use for forming plural and
past tense wordforms of recently coined nominal and verbal lexemes. That is,
for example, he or she will most likely say fake-ations, pumpkineers, (he) wifed,
(she) Thomased, etc. It is extremely unlikely that apophony, suppletion, or
signifier-sharing will be used instead. The distinction between regular and
irregular wordform-building mechanisms is thus a distinction between very
productive mechanisms like -s plural affixation of nouns and -ed past tense
affixation of verbs and rather unproductive mechanisms like apophony,
suppletion, and signifier-sharing.
To study the productivity of various wordform-building mechanisms, go to
the OED Advanced Search page http://www.oed.com/advancedsearch, enter
'1990-' to 'Date of entry', and choose e.g. the option 'Verb'. The OED will then
yield all verbal signifiers which appeared in the English language after 1990.
Then carefully read each entry. If the signifier under analysis has irregular
wordforms, this information will be provided in the Inflections-section of the
entry. For example, the Inflections-section of the above mentioned verb TO
WEBCAST contains the following information: 'Past tense webcast, webcasted'.
This means that this lexeme has two allowordforms: the regular webcasted and
the irregular webcast, which is formally identical with the corresponding present
tense wordform webcast. Given the existence of the latter, we are justified in
claiming that signifier-sharing is (at least sometimes) still used for creating
verbal past tense wordforms. If you are interested in the productivity of
mechanisms producing nominal wordforms, search the OED for all nominal
signifiers which appeared in English after 1990 and establish (in the same way
as in the case of verbs) whether at least one of them has an irregular plural form.
6.4 Syntactic grammemes in English
In this section we will discuss the syntactic grammemes VOICE and CASE. Our
focus will be on the most important theoretical questions pertaining to these
categories. For example, why does English need the passive voice if passive and
active wordforms express the same referential meaning? Does English have the
middle voice? Are get-passives allowordforms of be-passives? How many cases
do we have in English and what functions do case wordforms perform in the
language?
6.4.1 Why do we need the passive voice?
The answer to this question, which arises because of the non-semantic nature of
the voice grammeme, is that different elements of a clause introduce different
kinds of information. Among other things, there is the distinction between the
topic and the comment of a clause. The topic is the center of attention of the
clause, what it is about (Finegan 2004: 264). By contrast, the comment is that
component of the same clause that provides some information about the topic.
For example, while the active clause He met the President is about him who met
the President, the corresponding passive clause The President was met by him is
about the President who was met by him. In other words, in the active clause he
is the topic and met the President is the comment, whereas in the passive clause
the President is the topic and was met by him is the comment.
Following Mel'čuk (2006: Ch. 3), we can say that passivization is a means of
changing the communicative rank of different pieces of information (usually,
those denoting two different participants of the same event). By syntactically
promoting the object NP the President to the subject of the passive clause The
President was met by him, we also promote its communicative rank: the NP the
President, which in the active clause is only a part of the comment met the
President, becomes the topic of the passive clause The President was met by
him. Similarly, by syntactically demoting the subject pronoun he to the
complement of the preposition by of the PP by him, we also demote its
184 Chapter 6
communicative rank: he, the topic of the active clause He met the President,
becomes a part of the comment was met by him.
6.4.2 Middle voice in English?
Middle voice is a voice that is intermediate between at least two different voices.
Consider, for example, the clause The book sells for $ 19.95, which we already
discussed in 4.3.7. In this clause the subject position is filled by the NP The
book, which denotes the object of selling. That is, it is not the book that sells for
$ 19.95. It is some people (bookstores, etc.) who sell it for $ 19.95. Given this
fact, we can argue that the clause under analysis is a product of both syntactic
and communicative promotion of the object NP the book. In this respect, the
clause The book sells for $ 19.95 does not seem to be different from the passive
clause The President was met by him. However, in sharp contrast to the latter,
the former involves the suppression of the performer of selling. That is, we
cannot add a by-phrase specifying the person (the bookstore, the organization,
etc.) who sells the book for $ 19.95; e.g. (89).
(89) *The book sells for $ 19.95 by Amazon
Given the ungrammaticality of (89), we can conjecture that The book sells for $
19.95 is an instance of the middle voice. The middle voice in English seems to
be intermediate between the passive voice and the suppressive voice16. Like the
former, it promotes the objects of input active clauses to subjects of output
middle clauses. And like the latter, it suppresses the subjects of input active
clauses.
However, at least two facts do not support the inclusion of the middle voice
into the voice grammeme in English. First of all, from a formal point of view,
middle clauses like The book sells for $ 19.95 do not seem to differ from active
clauses like Amazon sells the book for $ 19.95. In contrast to passive wordforms,
which are formed with the help of the suffix -ed and the analytic form be,
middle wordforms do not have a distinct grammatical marker of their own. This
is the reason why many authors analyze middle clauses as clauses intermediate
16
Consider, for example, the Russian clause Собака кусается / Sobaka kusaetsya 'The dog
bites'. Whereas the English translation is an instance of the absolute transitive use – i.e. there
is an understood but unexpressed object people, which can be added to the clause: The dog
bites people – the Russian clause is in the partial suppressive voice (Mel'čuk: 205-206). The
understood object people cannot be added to the clause: *Собака кусается людей / Sobaka
kusaetsya lydej / The dog bites people. The addition of the object people is suppressed by the
suffix ся / sya: we can only say Собака кусает людей / Sobaka kusaet lydej, i.e. use kusaet
'bites' without -sya.
between active and passive clauses rather than between passive and suppressive
clauses.
Secondly, as Huddleston (2002b: 308) points out, middle clauses are usually
"concerned with whether and how […] the subject-referent undergoes the
process expressed in the verb". Because of this semantic property, middle
clauses are usually 1) negative clauses like (90), 2) clauses containing modal
verbs (mainly will) like (91), and 3) clauses containing adverbs (mainly well and
easily) like (92):
(90) The homes did not sell (COCA)

(91) You put her name on anything and it will sell (COCA)
(92) If the map sells well, it therefore immediately goes out of print (COCA)
Finally, as Huddleston (2002b: 308) adds, a middle clause "expresses a general

state, not a particular event". Consider, for example, (93) and (94).
(93) Martinez doesn't intimidate easily (COCA)

(94) With a full side zip, the bag ventilates well on balmy nights (COCA)
Clause (93) does not refer to a particular event in which somebody or something
did not manage to intimidate Martinez but to a general characteristic of the
person called Martinez. Similarly, (94) does not refer to a particular event
involving somebody ventilating the bag on balmy nights but to a general
characteristic of the bag.
Given these restrictions, it is clear that a number of active clauses cannot be
changed into middle clauses. For example, we cannot change the active clause
He met the President into the middle clause *The President met meaning 'he met
the President'. Likewise, we cannot change the active clause They played
football into the middle clause *Football played meaning 'they played football'.
Both He met the President and They played football denote particular events, not
general characteristics.
To conclude: the so-called middle voice exemplified by clauses like The
book sells for $ 19.95 is not a member of the set of the mutually exclusive
meanings 'active voice' / 'passive voice' / 'middle voice'. The latter is not an
instance of an obligatory grammatical category but a product of a particular lex-
forming mechanism: lex-forming syntactics' change, which we discussed in
4.3.7. That is, some verbal lexemes that were originally realized by transitive
lexes only with the course of time have acquired intransitive allolexes. For
example, the transitive sell of clauses like Amazon sells the book for $ 19.95
gave rise to the intransitive sell, which can now be used in middle clauses like
The book sells for $ 19.95. Similarly, the transitive lex intimidate of clauses like
He intimidated her gave rise to the intransitive intimidate of middle clauses like
186 Chapter 6
Martinez does not intimidate easily. The English middle voice is a lexical, not a
grammatical phenomenon.
6.4.3 What can be passivized?
The answer to this question was already (implicitly) given in 1.2.4. Passivizable
are transitive clauses like He met the President, in which the predicative position
is filled by an object: He met the President Æ The President was met by him.
Transitive clauses fall into monotransitive and ditransitive clauses. The
former are clauses like He met the President that contain only one object (the
NP the President). The latter are two-object clauses. Consider, for example,
(95).
(95) I don't think that people gave the President the right information
(COCA)
The clause people gave the President the right information is a ditransitive
clause because it contains two objects: the direct object the right information
and the indirect object the President. Direct objects differ from indirect objects
with regard to their position: the latter can only precede the former. Consider,
for instance, (96).
(96) *People gave the right information the President
This clause is ungrammatical because the direct object the right information is
placed before the indirect object the President.
Finally, consider the PP to the President of (97).
(97) People gave to the President the right information
This PP expresses the same meaning as the indirect object NP the President of
(95). Nevertheless, to the President is not indirect object but complement. We
are justified in arriving at this conclusion because in contrast to indirect objects,
the PP to the President can be placed either before or after the direct object the
right information. Thus (98) is as grammatical as (97).
(98) People gave the right information to the President
Another justification for the complement analysis of the PP to the President is

that it cannot function as subject of an associated passive clause. That is, we can
say The President was given the right information (by people) but not *To the
President was given the right information (by people). The latter variant is
ungrammatical because in this passive clause the subject position is filled by the
complement phrase to the President, which, in contrast to the object phrase the
President, cannot fill this position.
6.4.4 Get-passives
In English a 'passive' meaning can also be expressed by a combination of the

verb get and a participle II. Consider, for example, the underlined combination
in (99).
(99) What do people in this city think of George Bush? Here's somebody
who didn't campaign here. He got shellacked by Al Gore in this state
(COCA)
A clause like He got shellacked by Al Gore in this state is similar to the be-
passives discussed above in that in this clause the subject position is filled by the
object of shellacking (he / George Bush), while the performer of this action (Al
Gore) functions only as complement in the PP by Al Gore. Given this similarity,
a question arises as to whether the combination got shellacked can be regarded
as an allowordform of be shellacked.
To answer this question, we need to establish whether get-passives are as
obligatory as be-passives and whether clauses containing get-passives are
semantically identical with corresponding clauses containing be-passives. In
other words, does it make a difference whether George Bush got or was
shellacked by Al Gore?
As far as the first question is concerned, consider (100), (101), and (102).
(100) His colleagues loved him

(101) He was loved by his colleagues
(102) *He got loved by his colleagues
The grammaticality of (101) and the ungrammaticality of (102) illustrate that

get-passives are possible only with verbs denoting dynamic events (e.g.
shellacking), but not with stative verbs like love (Ward et al. 2002: 1442). Be-
passives, by contrast, are possible with both dynamic and stative verbs. That is,
we can say both He was shellacked and He was loved. Accordingly, get-passives
are less obligatory than be-passives.
As for the semantics of get-passives, consider (103), (104), (105), and (106).
(103) I opened the door (neutral situation)

188 Chapter 6
(104) The door was opened by me

(105) *The door got opened by me
(106) im totally shocked by how quick […] my house door got opened by
the guy. He must have a lot of experience cause he made it seem like
nothing (http://tinyurl.com/4yww57t, a review of an emergency
locksmith service)
Opening a door is a dynamic event and hence must be capable of undergoing

passivization by means of get. However, (103) does not seem to be passivizable
as (105) unless the event of my opening the door was an unusual event. Cf.
(106), where the speaker is surprised by how quickly the guy managed to open
his or her house door.
As has been pointed out in a number of studies, 'unusual' contexts favoring
the use of get-passives are those 1) where the subject is felt to be responsible for
what happened to him and 2) where the subject either benefited or suffered from
what happened to him. For example, (99) is an instance of get-passive because,
on the one hand, the subject he is felt to have suffered from what happened to
him in this state and, on the other hand, because he himself is guilty of that: he
did not campaign in this state and therefore was shellacked by Al Gore. As for
(106), the speaker (and indirectly his or her house door) has also benefited from
the fact that the guy has opened it so quickly: the speaker could quickly get into
his or her house.
In summary, it appears that get-passives are quasi-idioms in relation to
corresponding be-passives. The former often contain additional meanings that
are not inherent in the latter. That is, for example, (99) means 'he was shellacked
in this state by Al Gore and he himself is guilty of that', whereas He was
shellacked by Al Gore in this state means only 'he was shellacked in this state by
Al Gore (i.e. it may be the case that he is not to blame for this)'. Accordingly,
given the semantic non-identity between clauses containing get- and be-passives
and given the non-obligatoriness of get-passives, we can conclude this part of
the section with the claim that got shellacked is not an allowordform of be
shellacked. Like middle clauses, get-passives are a lexical, not a grammatical
phenomenon.
6.4.5 Types of cases
Instances of the case grammeme such as, for example, the nominative case, the
genitive case, the accusative case, etc. are sometimes classified into straight and
oblique cases. The former includes only the nominative case: a nominative
wordform like I is metaphorically the straight wordform because it is regarded
as the basic or standard form of a noun or a pronoun. By contrast, all other case
wordforms (e.g. the genitive my and the accusative me) have usually been
metaphorized as forms that 'fall away' from their standard nominative forms
(Blake 1994: 19). Accordingly, cases like the genitive and the accusative are
oblique cases.
Taking this into account, we can regard the 'straight' nominative wordform
(e.g. I) as an input wordform and all oblique forms (e.g. the accusative me and
the genitive my) as output wordforms.
6.4.6 Cases in English
English personal pronouns I, he, we, they as well as the interrogative pronoun
who have distinct nominative, genitive, and accusative wordforms:
x I, my, me
x he, his, him
x we, our, us
x they, their, them
x who, whose, whom
You and it have identical nominative and accusative wordforms but distinct
genitive wordforms:
x you, your, you

x it, its, it
Finally, she has identical genitive and accusative wordforms but a distinct
nominative wordform: she, her, her.
In contrast to pronouns, nouns have only distinct nominative and genitive
wordforms: e.g. teacher and teacher's. But there does not exist a formally
marked contrast between the nominative and the accusative case. For example,
as illustrated by (107) and (108), one and the same wordform teacher can fill
both the subject and the object position.
(107) The teacher is able to present from the front, and is better positioned
to observe pupils' response (COCA)
(108) I visited a social studies class and gave the teacher a map of the
United States when the period was over (COCA)
Wordforms like teacher of (107) and (108) are sometimes said to be in the plain
or the common case, i.e. the case that combines both nominative and accusative
properties.
190 Chapter 6
Unlike Present-day German and Old English, Present-day English does not
have the dative case. As stated above, the only distinctions that exist are that
between the nominative, the genitive, and the accusative case (for personal
pronouns and the interrogative pronoun who) and that between the plain case
and the genitive case (for nouns). In spite of this fact, some authors often use the
term 'dative shift' describing the alternation between ditransitive clauses like
People gave the President the right information and semantically identical
monotransitive clauses containing to-complements like People gave to the
President the right information (or People gave the right information to the
President). This unfortunate term may create the wrong impression that English
has the prepositional dative case marked by the preposition to. In the following
we will show that this analysis is not correct.
To begin with, observe that clauses like People gave to the President the
right information are regarded as instances of the dative shift because in
languages that have distinct dative forms, the dative case marks the indirect
object function. For example, we can argue that in the German clause Man hat
dem Präsidenten die richtige Information gegeben 'people gave the President the
right information', the NP dem Präsidenten functions as indirect object not
because it precedes another object die richtige Information but because dem
Präsidenten is in the dative case. Similarly, in the sentence Ich habe ihm
geholfen 'I helped him', the pronoun ihm 'him' functions as indirect object
because it is in the dative case. However, as far as the English language is
concerned, the only criterion for distinguishing between direct and indirect
objects is the position of the two objects in relation to each other. When a clause
contains two objects, the indirect object precedes the direct object: thus we can
only say People gave the President the right information, but not *People gave
the right information the President. (Objects that occur in monotransitive
clauses are always direct objects: e.g. even though him of the English I helped
him is semantically identical with the German indirect object ihm of Ich habe
ihm geholfen, the English him is the direct object of the clause I helped him.) As
for monotransitive clauses containing to-phrases, we already pointed out that
PPs headed by the preposition to such as to the President are not indirect objects
but complements. On the one hand, this is so because they can occur both before
and after direct objects: we can say both People gave to the President the right
information and People gave the right information to the President. On the other
hand, this is so because to-phrases cannot serve as subjects of associated passive
clauses: we cannot say *To the president was given the right information.
Accordingly, since to-phrases cannot function as indirect objects (which in
languages that have the dative case are expressed by NPs in the dative), the
preposition to cannot be a dative gram. Hence English does not have the dative
case.
6.4.7 Functions of case
Since case is a syntactic grammeme, it is not surprising that wordforms differing

with regard to case often mark different syntactic functions. For example, as was
already pointed out in 1.2.4, the subject position can be filled by a pronoun in
the nominative (e.g. He is a good guy), but not by a pronoun in the accusative
(e.g. *Him is a good guy). In contrast, the object position can be filled by a
pronoun in the accusative (e.g. I helped him), but not by a pronoun in the
nominative (e.g. *I helped he).
With regard to genitive wordforms, consider (109), (110), (111), (112),
(113), (114).
(109) Mike's eyes widened and he asked Leo, 'Isn't the ground starting to
tilt? (COCA)
(110) His being out of control was her fault (COCA)
(111) Our favorite find? Frozen biscuits. They taste nearly as good as
grandma's, minus all the work (COCA)
(112) He was a friend of father's (COCA)
(113) Much as they like the 'sameness' of routine to end the day, kids will
often happily accommodate two rituals, especially if one is mom's
and one is dad's (COCA)
(114) It was a girl's voice, behind him: strong and bright and sure (COCA)
These examples illustrate what Payne and Huddleston (2002: 467) call six types
of genitive constructions in English. The genitive wordform Mike's of (109)
functions as determiner of the head noun eyes. Mike's eyes are particular
representatives of the class of objects called eyes: those that belong to Mike.
The genitive pronoun his of (110) functions as subject of the embedded non-
finite clause His being out of control (which in turn functions as subject of the
larger clause His being out of control was her fault, into which it is embedded):
His being out of control means 'he was out of control'.
The genitive wordform grandma's of (111) is a quasi-idiom whose signified
contains not only the signifieds 'grandmother' and 'the genitive case' but also the
idiomatic meaning 'biscuits' recoverable from the preceding clause Frozen
biscuits. Precisely because of its quasi-idiomatic signified, grandma's of (111)
functions as a fused determiner-head phrase, i.e. a phrase in which both the
determiner and the head are expressed by one and the same element.
The genitive wordform father's of (112) functions as post-head modifier of
the head noun friend. As the term makes clear, a post-head modifier is a
modifier occurring after the head of a phrase, not before it.
The genitive wordforms mom's and dad's of (113) function as complements
of the auxiliary verb is.
192 Chapter 6
Finally, girl's of (114) exemplifies a genitive construction that is often called

classifying or descriptive genitive. In contrast to Mike's of Mike's eyes, girl's of
a girl's voice does not identify the voice as the voice of a particular girl but
describes it as a voice that is generally characteristic of a young woman.
Classifying genitives like girl's of a girl's voice function as pre-head modifiers
of head nouns in NPs.
As we established in 1.2.5, sometimes one and the same syntactic position
can be filled by more than one case wordform. For example, we can say both It
is I who did it and It is me who did it, It is he who did it and It is him who did it,
etc. As we said, nominative and accusative pronouns filling the complement
position in sentences like It is I / me who… differ with regard to their
sociolinguistics. The nominative wordforms make these sentences sound more
formal than the corresponding accusative wordforms. Given this difference, we
can say that case in English has one more important function: the sociolinguistic
(or the stylistic) function.
Indeed, the complement position in sentences like It is I / me who… is not
the only syntactic position where a speaker of English can use more than one
case wordform. Consider, for example, the variation in the object position
between the nominative who of (115) and the accusative whom of (116).
(115) Who did you vote for? (COCA)

(116) Whom did you vote for […]? (COCA)
In this position, it is the accusative whom that makes a clause like (116) sound
more formal than a clause like (115), in which the object position is filled by the
nominative who.
6.4.8 Semantic function of case?
Given what we have said about the syntactic nature of the case grammeme, this
title may seem a contradictio in adjecto: CASE cannot have a semantic function
because it is a syntactic grammeme. Nevertheless, Quirk et al. (1985: 321-322)
speak of 'genitive meanings' which can be expressed by genitive wordforms in
English. For example, the meaning 'possession' inherent in Mike's house: 'Mike's
house' = 'the house that Mike possesses'. The possessive meaning is often seen
as the default genitive meaning, so that the English genitive case is sometimes
referred to as the possessive case.
Apart from this meaning, the English genitive is said to express subjective
and objective meanings. For instance, the signified of the NP Mike's decision
can be represented as the clause Mike made a decision, where Mike is the
subject. By contrast, the signified of the NP Mike's arrest can be represented as
the clause The police arrested Mike, where Mike is the object. (Instead of using
the syntactic terms 'subject' and 'object', it is perhaps better to describe these
meanings in terms of semantic roles: Mike is the agent of making a decision
and Mike is the patient of being arrested by the police.) Also, the genitive case
in English can express:
x partitive meanings: e.g. Mike's eyes = 'eyes that are part of Mike'
x temporal meanings: e.g. New Year's Eve = 'evening before the New Year'
x measure meanings: e.g. an hour's discussion = 'a discussion that lasted one
hour'
This book argues against the treatment of these meanings as genitive meanings.
These meanings are not inherent in the genitive gram -'s itself but in NPs like
Mike's house, Mike's decision, Mike's arrest, Mike's eyes, New Year's Eve, and
an hour's discussion. That is, all these NPs are quasi-idioms whose signifieds
contain additional idiomatic meanings that are not inherent in either of their
components' signifieds. For example, the NP Mike's house is a quasi-idiom
whose signified contains not only the signifieds 'Mike' and 'house' but also the
idiomatic meaning 'possession', which is not inherent in either Mike or house or
the genitive gram -'s. Similarly, the NP Mike's decision is a quasi-idiom whose
signified does not only contain the signifieds 'Mike' and 'decision' but also the
idiomatic meaning 'made by', which is not inherent in either Mike or decision or
the genitive gram -'s.
The existence of multiple 'genitive meanings' corroborates the quasi-
idiomatic analysis of all these NPs. If there were such thing as an obligatory
genitive meaning, then all genitive wordforms would carry that meaning in all
(or at least the majority of the) contexts where they occur. However, as the NPs
under consideration demonstrate, phrases containing genitive wordforms can
express more than one genitive meaning: 'possession', 'agent', 'patient', 'being a
part of', 'temporal location', 'measure', etc. Each of these meanings is not
inherent in the genitive suffix -'s but is a product of quasi-idiomatization of NPs
containing genitive wordforms.
To conclude: like the nominative and the accusative, the English genitive
case is a syntactic case, which does not carry any (genitive) meaning of its own.
Accordingly, we can reiterate the characterization of CASE as a syntactic
grammeme that produces output wordforms like Mike's that have the same
referential meaning as corresponding input wordforms like Mike. (Since this
book is concerned with English morphology, we can ignore languages that have
semantic cases. For example, in Finnish there is the inessive case. This case
does not only serve to mark a particular syntactic function of a noun but also to
express the meaning 'location within the referent' (Lyons 1968: 299). For
example, the noun talo-ssa, where -ssa is an inessive gram, means 'inside the
194 Chapter 6
house'. English lacks semantic cases, so that we are fully justified in regarding
the case grammeme in English as a syntactic grammeme).
6.5 Semantic grammemes in English
Having discussed the syntactic grammemes VOICE and CASE, we can proceed to
the semantic grammemes TENSE, ASPECT, MOOD, PERSON, NUMBER, DEGREES OF
COMPARISON, and NUMERICAL QUALIFICATION.
6.5.1 Typology of temporal meanings
Tense is the grammatical category that is concerned with the relation holding
between an event expressed by a predicate VP and its tense locus, i.e. the
reference point in relation to which the event is located in time (Frawley 1992:
340). Tenses that are concerned with the relation holding between events and the
moment of utterance (i.e. the primary tense locus) are simple tenses. These
include the past simple tense and the present simple tense.
If the event under analysis precedes the moment of utterance, it is in the past
tense. If the event coincides with the moment of utterance, it is in the present
tense. As an illustration of this difference, compare the temporal meanings of
ask in (117) and (118).
(117) Between the beer course and the pizza, I asked you to describe the
perfect woman (COCA)
(118) So I ask you to keep quiet about this thing […] (COCA)
I asked you to describe the perfect woman of (117) is in the past tense: the event
described by the clause took place before the speaker uttered (117), i.e. he or she
first said Please describe the perfect woman and then by uttering (117) reminded
the hearer(s) about this request. By contrast, I ask you to keep quiet about this
thing of (118) is in the present tense: the event described by the clause coincides
with the moment of utterance, i.e. at the moment when the speaker uttered (118),
he or she asked the hearer(s) to keep quiet about some thing.
Tenses which have more than one tense locus are perfect tenses. Compare,
for example, (119) and (120).
(119) I saw the movie

(120) I had seen the movie
Both (119) and (120) denote events which were accomplished before the
moment of utterance. That is, the speaker first saw the movie and then said I saw
the movie / I had seen the movie. However, while (119) is located in the past
only in relation to the moment of utterance, (120) also expresses anteriority in
relation to an additional tense locus: either some other event or some specified
time in the past; e.g. (121) and (122).
(121) I had seen the movie before my mother came

(122) I had seen the movie by 10 o'clock yesterday
The past perfect tense, exemplified by (121) and (122), can be contrasted with
the present perfect tense. Consider, for example, (123).
(123) I have seen the movie
Like (119) and (120), this clause denotes an event that took place before the
moment of utterance: the speaker first saw the movie and then said I have seen
the movie. The difference between (123) and both (119) and (120) is that the
past event expressed by (123) is somehow relevant in the present. Consider, for
example, (124).
(124) I have seen the movie, and can confirm that it might be the most
staggering filmic achievement of this or any other century
(http://tinyurl.com/66n68ja)
Here the use of the present perfect tense can be explained in the following way.
It is relevant in the present that the speaker saw the movie in the past because
owing to this, he or she now knows which movie can be considered the most
staggering filmic achievement of this or any other century. The difference
between the past perfect tense and the present perfect tense is thus that while the
secondary tense locus of the past perfect tense is in the past, the secondary tense
locus of the present perfect tense is in the present. (Hence the term 'present
perfect', even though this tense is also associated with past events.)
In addition to the simple tense–perfect tense distinction, there is also the
distinction between absolute and relative tenses. Absolute tenses are tenses
whose tense locus is the moment of utterance. Both simple tenses and perfect
tenses qualify as absolute tenses because both the former and the latter
characterize the relation between an event and the moment of utterance. The
only difference is that in the case of simple tenses, the moment of utterance is
the sole tense locus relative to which the event is located in time, whereas
perfect tenses have two tense loci: the moment of utterance and some other
event / some specified time. By contrast, relative tenses do not depend on the
196 Chapter 6
moment of utterance as their tense locus. Consider, for example, the temporal
meanings inherent in the underlined non-finite clauses in (125), (126), (127),
(128).
(125) I'm glad to have the opportunity to apologize to you, Ms. Shipman, in
person (COCA)
(126) I'm glad to have had the opportunity to meet you, Ranger Strong […]
(COCA)
(127) I enjoyed writing songs (COCA)
(128) Many other playwrights have written better, but no one has enjoyed
having written more (COCA)
Non-finite clauses are clauses that contain non-finite forms such e.g. the
infinitive have of (125) and the gerund writing of (127). Among other things,
these are different from finite forms (e.g. enjoyed of (127)) with regard to their
temporal meanings. The tenses of non-finite clauses are always relative tenses
whose tense locus is not the moment of utterance but the event expressed by the
finite clause. For example, the infinitival clause in (125) is an instance of the
indefinite infinitive, i.e. a relative tense expressing simultaneousness. That is,
the event denoted by the non-finite clause to have the opportunity to apologize
to you, Ms. Shipman, in person is simultaneous with the event denoted by the
finite clause I'm glad. In other words, the speaker of (125) is glad that he or she
at the moment of utterance has the opportunity to apologize to Ms. Shipman.
The same can be said about the indefinite gerund exemplified by the gerundial
clause in (127). The event denoted by the non-finite clause writing songs was
simultaneous with the event denoted by the finite clause I enjoyed.
Indefinite infinitives and indefinite gerunds must be distinguished from
perfect infinitives and perfect gerunds. For example, the infinitival clause in
(126) is a perfect infinitive which expresses that the event denoted by the non-
finite clause to have had the opportunity to meet you, Ranger Strong is anterior
to the event denoted by the finite clause I'm glad. Likewise, the gerundial clause
in (128) is a perfect gerund which expresses that the event denoted by the non-
finite clause having written more is anterior to the event denoted by the finite
clause no one has enjoyed.
6.5.2 No future tense in English
It is often argued that the English language lacks the future tense. The reasons
for this provided by the authors who share this view can be summarized in the
following way:
x No inflectional affixes are used for forming future tense wordforms. Cf. e.g.
(he) worked and (he) will work.
x There is an overlap between the present and the future tense, i.e. one and the
same wordform can be used for expressing both present and future meanings.
x The meaning 'the future tense' is a modal rather than a temporal meaning.
As for the absence of inflectional grams, it is indeed the case that in contrast to
e.g. Lithuanian, which has separate inflectional encodings for the past tense, the
present tense, and the future tense (Frawley 1992: 359), English has only the
past tense regular affix -ed and the present tense (third person singular) suffix -s.
However, it is doubtful that this fact alone suffices to conclude that English
lacks the future tense: as we have learned in 6.3, wordform-building in English
relies not only on inflectional affixation but also on analytic formation (e.g.
more beautiful). Accordingly, a future tense form like will work can be regarded
as an analytic wordform consisting of the input present tense wordform work
and the analytic auxiliary verb will.
With regard to the overlap between the present tense and the future tense,
consider (129) and (130).
(129) The next solar eclipse is on July 11 2011 (http://tinyurl.com/45ytebe)

(130) Hockey World Cup: England meets Germany in semis tomorrow
(http://tinyurl.com/3bbs6tm)
These examples serve to illustrate that present tense wordforms may have a
future reference. As pointed out by Huddleston (2002b: 132), this is true of
scientifically calculable cyclic events such as e.g. a solar eclipse of (129) and
scheduled events such as e.g. a hockey game of (130). But again this fact is not
sufficient to claim that English does not have the future tense. Thus in addition
to the present–future overlap, there are also the present–past and past–future
overlaps. Consider, for example, (131) and (132).
(131) […] the Bible says that having respect for God is the beginning of
wisdom (COCA)
(132) I figured it was best to wait until he was ready (COCA)
In (131), the present tense wordform says has a past tense reference: the Bible
says that having respect for God is the beginning of wisdom means 'the author of
the Bible said (long before the moment of utterance of (131)) that having respect
for God is the beginning of wisdom'. By using the present tense wordform says,
the speaker of (131) emphasizes the present-day importance of what the author
198 Chapter 6
of the Bible said long ago. In (132), the past tense wordform was ready refers to
a future event viewed from the past, i.e. the person referred to as he will be
ready after the speaker of (132) figured it was best to wait. Obviously, these
overlaps provide no justification for the claim that the English language lacks
either the present or the past tense.
Finally, let us consider the claim that the future tense meaning is not a
temporal but a modal meaning. The point here is that no matter how certain you
might be about your future plans, a future event is always an event that may or
may not take place. Consider, for example, (133).
(133) I will travel to Mexico soon (COCA)
Even if the speaker has already booked the flight to Mexico and has already
made a hotel reservation, (133) remains a prediction: it cannot be excluded that
he or she will die in a car accident while driving to the airport and thus will not
be able to travel to Mexico.
This leads us to the conclusion that the future tense is different from both the
past tense and the present tense in that while the latter denote real events that
either coincide with or precede the tense locus, the former denotes unreal
events that may or may not take place in the future. Accordingly, the future
tense must be analyzed not as a member of the tense grammeme but as an
instance of so-called epistemic modality, i.e. a type of modality which is
concerned with "the degree of certainty the speaker has that what s/he is saying
is true" (Haan 2006: 29). That is, the future will of (133) is a modal verb which
expresses more certainty than e.g. the epistemic may of I may travel to Mexico
soon 'it's possible that I will travel to Mexico soon. But I am not certain about
this'.
6.5.3 Idiomatic uses of temporal wordforms
As we argued in 2.5.4, wordforms containing inflectional affixes such as the

past tense suffix -ed can express idiomatic meanings. For example, the use of
past tense wordforms in politeness-related contexts (Huddleston 2002b: 138)
such as e.g. I wanted to ask you to tell the story of your father taking you to see
the Brooklyn Dodgers in 1947 meaning 'I want to ask you to tell this story now'
is a case in point. Here -ed does not characterize the event as a past event that
was valid before the moment of utterance (i.e. 'some time ago I wanted to ask
you to tell this story, but at present I no longer want to hear it') but as a past
event that continues to the present. That is, a person who says I wanted to ask
you (most likely) began to experience this wish prior to the moment of utterance.
However, at the moment of utterance he still wants the addressee to tell the
story. Past tense wordforms like wanted of politeness-related contexts like I

wanted to ask you can thus be analyzed as quasi-idioms in relation to 'normal'
past tense wordforms such as e.g. wanted of (134) denoting past tense events
which do not continue to the present:
(134) A year ago I wanted to end my life. I am now 50 and looking forward
to the next 50 (COCA)
Evidently, the speaker of (134) does not want to end his or her life anymore. By
contrast, wanted of I wanted to ask you contains not only the meaning 'the past
tense', inherent in the past tense suffix -ed, but also the idiomatic meaning
'continuation to the present'.
Consider also the temporal meaning of the present tense wordform becomes
in (135).
(135) August 3 – Bolivia becomes the first South American country to

declare the right of indigenous people to govern themselves
(http://en.wikipedia.org/wiki/2009)
(135) is taken from Wikipedia's 2009 historical chronicle. It thus describes a

past event that took place in August 2009. The past tense use of the present tense
wordform becomes of (135) can thus be analyzed as a full-idiom in relation to
becomes of e.g. It now becomes apparent that …, where it denotes an event
coinciding with the moment of utterance.
Finally, consider (136).
(136) I have lived in New York
This clause is ambiguous between the following two readings: 'I lived in New
York in the past and this is somehow relevant in the present' and 'I moved to
New York in the past and still live there'. The former is known as the non-
continuative present perfect; the latter as the continuative present perfect.
According to Huddleston (2002b: 141), "the non-continuative reading of the
perfect is much more frequent, and can be regarded as the default one". Indeed,
it is more likely that a speaker of English will interpret (136) non-continuatively,
i.e. as an event that does not continue to the present. A continuative
interpretation will occur only in contexts like I have lived in New York for ten
years, where the continuative reading is reinforced by the duration adjunct for
ten years. Taking this into account, we can analyze the continuative present
perfect as an idiomatic use of the present perfect tense. The continuative perfect
is a semi-idiom in relation to the non-continuative perfect. While the latter can
be segmented into the features [beginning in the past], [accomplishment in the
200 Chapter 6
past], [present relevance], the former is segmentable into the features [beginning
in the past] and [continuation in the present].
6.5.4 Typology of aspectual meanings
Aspect is the grammatical category which is concerned with aspectual meanings

(sometimes referred to as Aktionsarten). Following Maslov (1973; 1978),
Frawley (1992: Ch. 7), Plungian (2000: 292-308), and Huddleston (2002b: 118-
125), we will classify aspectual meanings into those relating to:
1. qualitative aspectuality
2. quantitative aspectuality
3. phase aspectuality
With regard to qualitative aspectuality, which is concerned with the inherent

properties of an event expressed by a predicate VP, events can be classified into
dynamic events and static events. These are different from each other with
regard to the following grammatical properties:
x The present simple combines with static events but not with dynamic events:
i.e. e.g. we can say He knows Mike meaning 'He knows Mike at the moment
of utterance' but not *He reads a book meaning 'He is reading a book at the
moment of utterance'.
x Only dynamic events can be progressivized: cf. e.g. He is reading a book and
*He is knowing Mike.
x Only dynamic events can occur in a wh-cleft: cf. e.g. What he did next was
read a book and *What he did next was know Mike.
Dynamic events can be further classified into achievements (e.g. He found a

book) and processes (e.g. He read a book). These are different from each other
with regard to the following properties:
x Verbs like begin, finish, continue, etc. combine with processes but not with
achievements: cf. e.g. He began to read a book and *He began to find a
book.
x Only processes can be progressivized: cf. e.g. He is reading a book and *He
is finding a book.
x Duration adjuncts like the PP for an hour combine freely with processes but
not with achievements: cf. e.g. He read a book for an hour and *He found a
book for an hour.
Finally, processes can be further classified into telic events (e.g. We went to the
beach) and atelic events (e.g. He teaches history at Yale University). The
differences are as follows:
x Telic events are begun in order to be finished: e.g. we go to the beach in

order to be at the beach. As soon as we are at the beach, we no longer
continue going to the beach. Atelic events, by contrast, do not have such
terminal points that terminate the realization of an event as soon as they are
achieved: e.g. one can teach history at Yale as long as one wishes to do so.
x We can say It took us an hour to go to the beach but not *It took him a year
to teach history at Yale University.
With regard to quantitative aspectuality, which is concerned with the question of

how many times a given event was realized, events can be classified into:
1. single or semelfactive events: e.g. Mike knocked (once) at the door
2. multiple or iterative events: e.g. Mike knocked at the door for a minute
3. permanent events that always take place: e.g. CNN broadcasts 24 hours a
day
Multiple events can be further classified into the following types:
a) multiplicative events, consisting of a set of semelfactive events which are

repeated instantaneously: e.g. Mike knocked at the door for a minute
b) distributive events, applying to members of the same group one after

another: e.g. Mike knocked at several doors
c) habitual events, recurring on a more or less regular basis; e.g. He eats one
meal a day
Finally, phase aspectuality is concerned with the question of which phase of an

event was realized at some specified time. According to Plungian (2000: 297),
event phases or stages include:
202 Chapter 6
1. preparation / prospective phase: e.g. I am going to read a book
2. beginning / inchoative phase: e.g. I begin to read a book
3. cessation / terminal phase: e.g. I finish reading a book
4. middle phase (i.e. the phase between the beginning and the cessation of an
event): e.g. I am reading a book
5. result / resultative phase: e.g. I finish reading a book and go to the movie to
see the film based on the book
The stages 'beginning', 'cessation', and 'the middle phase' constitute the internal
stages of an event, while the preparatory and the resultative phases are the
external phases. This is so because it can be argued that, for example, the
opening of a book, which is usually a part of the preparatory phase of the
reading event (we cannot start reading a book without opening it), is not really a
part of the reading event itself: at the moment when we are opening a book, we
are not reading it.
6.5.5 Aspects in English
English has a number of aspectualizers that are used for expressing the above
named aspectual meanings. But only one of them – a combination of the
auxiliary be and a participle I like working – can be regarded as a marker of an
obligatory grammatical meaning. This textbook thus accepts the traditional view
that English has only one aspect: the progressive aspect.
The progressive is the aspect which expresses the meaning 'the realization of
the middle phase of an event'. For example, the progressive clause He was
reading a book expresses the meaning 'the realization of the middle phase of the
reading event', i.e. at the moment when the speaker's attention was attracted by
this event, the subject he was in the middle of reading a book. He was neither
beginning nor finishing reading.
In 6.5.4 we have learned that the progressive aspect does not combine with
states (e.g. *He is knowing Mike) and achievements (e.g. *He was finding a
book). With regard to the latter, the explanation is that achievements are
instantaneous events, i.e. events that lack not only the middle phase but also all
other internal phases discussed above. That is, it is impossible *to be finding a
book because it is not possible either *to begin to find a book or *to finish
finding a book. You either find it or not. With regard to states, consider (137)
and (138).
(137) ???He began to know Mike

(138) ???He finished knowing Mike
Both (137) and (138) are hardly possible in English because both the inchoative
and the terminal phase of a static event are perceived as external stages of the
event. That is, the inchoative meaning inherent in (137) and the terminal
meaning inherent in (138) are usually expressed by sentences like He got
acquainted with Mike and He forgot Mike, i.e. sentences which do not mention
the knowing event itself. The impossibility of the progressivization of the state
He knows Mike can thus be attributed to the fact that the knowing event consists
of only one middle phase, i.e. there is no need to progressivize He knows Mike
because, by virtue of being a state, this event already conveys the meaning 'the
middle phase'.
Sometimes it is argued that English has the habitual aspect expressed by
used to of e.g. (139) and the prospective aspect expressed by be going to of e.g.
(140).
(139) I used to walk down the street in New York City […] (COCA)
(140) I was going to get there (COCA)
Indeed, (139) expresses that the action of walking down some street in New
York City was repeated by the speaker on a fairly regular basis, and (140)
expresses that the speaker planned to get to some place. However, both used to
and be going to are much less obligatory than the progressive marker be +
participle I. With regard to used to, observe that a repetition of an action in the
past can also be expressed with the help of would. For example, (141).
(141) […] every day he would go out in the backyard and pass the tales on
to whatever animals happened to be lounging about (COCA)
And in the case of a present tense event, a habitual meaning is usually expressed
with the help of the present simple tense. That is, a clause like He reads a book
means 'He habitually repeats the action of reading a book'. By contrast, a clause
like He used to read a book can only mean 'This series of events took place in
the past (not in the present)'.
Similarly, as Comrie (1976: 64) points out, English has several prospective
aspectualizers. For example, be about to of (142) and be on the point of of (143).
(142) I remember the day I was about to teach my first course in political
psychology (COCA)
(143) I was on the point of quitting for years (COCA)
204 Chapter 6
Also, it is worth mentioning that be going to "is characteristic of relatively

informal style" (Huddleston 2002b: 211), whereas all other grams that have been
discussed so far are stylistically neutral.
Given the non-obligatory character of both used to and be going to, we are
justified in concluding that these idiomatic combinations are not grams of
obligatory aspectual meanings but lexical aspectualizers for expressing optional
habitual and prospective meanings.
Finally, let us consider the possibility of regarding the present simple non-
progressive as a marker of the habitual aspect in English. This conjecture may
arise because, as said above, the present simple tense exemplified by He reads a
book expresses the habitual meaning 'He regularly reads some book' and, as we
established in 6.5.4, habituality is an instance of quantitative aspectuality. To
answer the question of whether the English present simple tense is indeed an
aspectual rather than a temporal category, we first need to consider the question:
why is it actually the case that a clause like He reads a book can only be
associated with an aspectual habitual meaning rather than with the temporal
meaning 'an event that coincides with the moment of utterance'?
The answer to this question relates to what we have learned about the stages
of an event in 6.5.4. A dynamic event like reading a book has three internal
stages: the beginning, the cessation, and the middle stage. It is obvious that the
moment of utterance cannot coincide with all three stages of this event: it takes
much more time to read a book than it does to say He reads a book. The moment
of utterance can coincide only with one of these stages. If it coincides with the
middle stage, we use the progressive wordform is reading. Accordingly, we can
say that He reads a book expresses an aspectual habitual meaning because due
to its dynamic aspectuality, it cannot express the temporal present tense meaning
'an event taking place at the moment of utterance'.
At the same time, note that some dynamic events do have a non-habitual
interpretation when used in the present simple tense. This is true of e.g. running
commentaries such as (144):
(144) Now on the board Semyonov, number 30, back to Fetisov, to

Kasatonov, shoots, he scores!!! (http://tinyurl.com/63l6pls, Canada–
USSR, Canada Cup 1987, Final, Game 1)
At the moment when the commentator said shoots, he scores!!!, Kasatonov (i.e.
a Soviet ice hockey player) was indeed shooting and scoring a goal. In other
words, the shooting and scoring events coincided with the moment of utterance.
Recall also the use of the present simple wordform say in I ask you to keep
quiet about this thing. As stated in 6.5.1, by uttering this sentence the speaker
did indeed ask the addressee(s) to keep quiet about some thing. Again, we can
say that the event of asking other people coincided with the moment of
utterance. Both these facts support the traditional analysis of the present simple
tense as a temporal rather than an aspectual category.
6.5.6 Moods in English
Similar to tense and aspect, which can be defined as grammatical categories

concerned with the expression of temporal and aspectual meanings, mood can be
defined as the grammatical category concerned with the expression of modal
meanings.
Traditionally, modality is classified into:
1. root modality, which is associated with the meaning 'ability': e.g. Sarah can
speak three languages 'Sarah has the ability to speak three languages'.
2. deontic modality, which is associated with the meaning 'obligation': e.g.

Sarah must speak three languages 'Sarah is required to have the ability to
speak three languages'.
3. epistemic modality, which is associated with the meanings 'certainty' and

'possibility': e.g. Sarah will speak three languages 'It is possible that in the
future Sarah will have the ability to speak three languages'.
As in the case of aspectual meanings, modal meanings can be expressed in a

number of ways. In addition to modal verbs like the above named can, must, and
will, modal (especially epistemic) meanings are also often expressed with the
help of modal adverbs like probably and maybe of (145) and (146) and so-
called modal tags like I believe and I guess of (147) and (148).
(145) Barry is probably at home watching our games (COCA)

(146) Maybe the danger lies within ourselves (COCA)
(147) I believe I developed better negotiating skills in the business world
than a lot of people who've been around here for years […] (COCA)
(148) I guess he's been busy (COCA)
However, none of these modalizers can be regarded as obligatory grammatical

markers in English. Thus instead of saying Sarah must speak three languages,
we can say Sarah is required to speak three languages or It is necessary that
Sarah speak three languages. And instead of saying Sarah may speak three
languages meaning 'it is possible that Sarah speaks three languages but I am not
certain about it', we can say I guess Sarah speaks three languages or Sarah
probably speaks three languages.
206 Chapter 6
The only modal grams in English are those modalizers that characterize an
event as an instance of one of the following traditionally recognized moods:
1. the indicative mood

2. the imperative mood
3. the subjunctive mood
The indicative mood is described by Khaimovich and Rogovskaya (1967: 142)

as "a fact mood [which] serves to present an action as a fact of reality". For
example, a clause like He is reading a book denotes an event that, according to
the speaker, coincides with the moment of utterance. Similarly, a clause like He
was reading a book denotes an event that, according to the speaker, preceded the
moment of utterance.
A notable feature of the indicative mood in English is that there are no
grams that serve to express indicative meanings only. Consider, for example,
(149).
(149) He studies engineering and business management at Chongqing

University […] (COCA)
This sentence can only be interpreted indicatively, i.e. as a sentence denoting an

event that is still valid at the moment of utterance. That is, at the moment of
utterance the subject he was still a registered student of engineering and business
management at Chongqing University. What justifies this analysis is that the
predicate VP is headed by the wordform studies, which is a product of affixation
of study by means the suffix -s. However, as we have learned in 2.1, this suffix
is a portmanteau morph, which cumulatively expresses more than one
grammatical meaning. That is, in addition to the meaning 'the indicative mood',
it also expresses the meanings 'the present tense', 'the non-progressive aspect',
'the active voice', 'the third person', etc. The same is true of all other indicative
wordforms such as e.g. (He) studied, (He) has studied, (It) was studied, etc. In
all of them, the indicative meaning is expressed cumulatively, i.e. together with
other grammatical meanings such as 'tense', 'aspect', 'voice', 'number', etc.
In contrast to the indicative mood, the imperative mood has several
genuinely imperative formal characteristics. One is that the majority of
imperative clauses have covert second person subjects. For example, (150) can
be said to mean 'You go to the bedroom' and the signified of (151) can be
analyzed as 'You replace the vacuum bag once a month'.
(150) Go to the bedroom (COCA)

(151) Replace the vacuum bag once a month (COCA)
In other words, it is the addressee(s) of (150) and (151) that are expected to
perform the actions of going to some bedroom and replacing the vacuum bag
once a month.
The second formal characteristic is that the VPs of imperative clauses are
headed by verbs in the plain form. (Huddleston 2002b: 83). This is the citation
form of a verb (go, replace, follow). Usually it is identical with the present tense
wordform. However, in contrast to the latter, the former does not receive the
affix -s when used with third person subjects like he and she. Compare, for
example, (152) and (153).
(152) Everybody follows that (COCA)

(153) Come on. Everybody follow me (COCA)
Clause (152) is in the indicative mood: Everybody follows that = 'According to

the speaker, at the moment of utterance everybody followed that'. By contrast,
(153) is in the imperative mood: Everybody follow me = 'The speaker asks the
addresses to follow him'. We are justified in arriving at this conclusion because
the predicator follow is in the plain form: it does not receive the third person
suffix -s, which it receives in an indicative clause like (152).
Semantically, the imperative mood is associated with the meanings 'order',
'command', 'request', 'instruction' and the like. In addition to these core
imperative meanings, imperative clauses can also be used for expressing wishes
and conditions. Consider, for example, (154) and (155).
(154) Have a happy Thanksgiving (COCA)

(155) In spite of repeated warnings you insist on posting in capitals! Do
this again and you will receive a formal warning
(http://tinyurl.com/5uz4pee)
The imperative clause (154) means 'I wish you a happy Thanksgiving' and the
imperative clause in (155) means 'If you do this again, you will receive a formal
warning'.
Finally, let us turn our attention to the subjunctive mood. If the indicative is
a fact mood, the subjunctive is a non-fact mood, i.e. a mood that represents an
event as "something imaginary, desirable, problematic, contrary to reality"
(Khaimovich and Rogovskaya 1967: 150-151). The subjunctive mood is often
classified into subjunctive I and subjunctive II. (Subjunctive I is sometimes
called present subjunctive and subjunctive II past subjunctive. (Quirk et al.
1985: 155). Huddleston (2002: 88) uses the term 'subjunctive mood' only in
connection with subjunctive I, whereas subjunctive II is referred to by him as the
irrealis mood.) Compare, for example, (156) and (157).
208 Chapter 6
(156) I would suggest that he start helping the people of the United States
that [have] worked and paid in taxes all their lives (COCA)
(157) I wish you were here (COCA)
The clause he start helping the people of the United States […] of (156) is an
instance of subjunctive I, i.e. a mood which represents events as problematic but
not as entirely contradicting reality (Ganshina and Vasilevskaya 1964: 203).
Thus it is possible that the subject he will indeed start helping the people of the
United States. By contrast, the clause you were here of (157) is an instance of
subjunctive II, i.e. a mood which represents events as contrary to reality: at the
moment of utterance the subject you of (157) was not near the speaker.
Instances of subjunctive I are similar to imperative clauses in that the
predicator verb is also in the plain form: we say I suggest that he start … instead
of I suggest that he starts … This is their main formal characteristic. (In spite of
this formal similarity, instances of subjunctive I are easily distinguishable from
imperatives: in contrast to the latter, the former contain overt subjects like he (of
156).)
With regard to subjunctive II, we need to distinguish between the present
subjunctive II and the past subjunctive II. Compare, for example, sentences
(157) / (158) and (159).
(158) I wish the news were better (COCA)

(159) I wish the news had been better (COCA)
The present subjunctive II is a relative tense that conveys simultaneousness: i.e.

e.g. in (158) the speaker wishes that the news were better at the moment of
utterance. The defining formal characteristic of the present subjunctive II is that
with the exception of be, all other present subjunctive II wordforms are identical
with past tense wordforms. For example, (160) and (161).
(160) I wish he dumped that bill that we enacted this year that's going to
make it a lot worse on you taxpayers to meet your responsibilities
(COCA)
(161) I wish he gave us time to probe (COCA)
With regard to be, there is a stylistically relevant free variation between the
allowordforms were and was, the latter being more informal than the former
(Huddleston 2002b: 86). For example, (162) and (163)
(162) I wish I were a better man (COCA)

(163) I wish I was still there (COCA)
The past subjunctive II, exemplified by the clause I wish the news had been
better, is a relative tense that conveys anteriority. That is, the speaker wishes
that the news had been better at some point in the past prior to the moment of
utterance. From a formal point of view, past subjunctive II wordforms like had
been better of (159) are identical with perfect wordforms of indicative past
perfect clauses like The news had been better.
This textbook departs from the above named authors in that it regards
subjunctive I wordforms like (he) start of (156) as allowordforms of indicative
mood wordforms like (he) starts of e.g. He starts fighting Titan. The main
justification for this analysis is the non-obligatoriness of subjunctive I in
Present-day English. Already in 1964, Ganshina and Vasilevskaya noted that "in
Modern English subjunctive I is rapidly falling into disuse" (1964: 204).
Similarly, in a more recent reference grammar of English, Quirk et al. (1985:
155) point out that "the subjunctive in modern English is generally an optional
and stylistically somewhat marked variant of other constructions […]". Finally,
according to Huddleston and Pullum,
subjunctives occur as main clauses in a few more or less fixed

expressions, as in God bless you, Long live the Emperor, etc. Their
most common use is as subordinate clauses of the kind [It's essential
that he keep us informed]. Structurally these differ only in the verb
inflection from subordinate clauses with a primary verb-form – and
many speakers would here use a present tense in preference to the
slightly more formal subjunctive: It's essential that he keeps us
informed. (Huddleston and Pullum 2005: 32)
To sum it up: subjunctive I wordforms occur in a relatively small number of

constructions (mainly subordinate clauses like that of (156)), in which they are
often replaceable by indicative wordforms: i.e. instead of saying I would suggest
that he start helping the people of the United States, a speaker of English can
also say I would suggest that he starts helping the people of the United States.
Given this fact, it is clear that the traditional analysis of subjunctive I as a
distinct member of the mood grammeme in English is untenable. As stated
above, the plain form (he) start of (156) is not a subjunctive I wordform but a
stylistically more formal allowordform of the indicative (he) starts.
6.5.7 Person
PERSON is the grammatical category which is concerned with the relation

between the participants of a speech event (i.e. an event that involves the
210 Chapter 6
uttering of at least one clause) and the participants of an event denoted by that
clause. As an illustration, let us consider (164) and (165).
(164) I trust you completely (COCA)

(165) She is a beautiful child (COCA)
The two participants of any speech event are the speaker and the addressee. For
example, the two participants of the speech event (164) are the person who
uttered the clause I trust you completely (the speaker) and the person(s)
addressed by him or her (his or her addressee(s)). Similarly, the two participants
of the speech event (165) are the person who uttered She is a beautiful child and
the person(s) addressed by him or her. (Normally, the speaker and the addressee
are different people. However, in the case of self-speech, the speaker is
simultaneously the addressee.)
Now, as regards the relation between the participants of the speech events
(164) and (165) and the participants of the events denoted by these clauses, one
can easily notice that in (164) the speaker and the addressee are simultaneously
the two participants of the event denoted by this clause. That is, the speaker is
the person who completely trusts the addressee(s). By contrast, the participant of
the event expressed by (165) is a person who is neither the speaker nor the
addressee. Neither the former nor the latter is a beautiful child.
The category of person can thus be defined as the set of the following
mutually exclusive grammatical meanings:
1. 'the first person', i.e. the speaker

2. 'the second person', i.e. the addressee
3. 'the third person', i.e. a person or an object who is neither the speaker nor
the addressee
The first person gram in English is the personal pronoun I; the second person
gram is the personal pronoun you; the third person grams are the personal
pronouns he, she, and it (differing with regard to gender). Each of these grams
has plural wordforms: I Æ we, you Æ you, he / she / it Æ they. In this
connection, it is important to observe that while the plural you can be
semantically analyzed as 'more than one addressee' (e.g. the speaker of (164)
completely trusts more than one person), the plural we does not mean 'more than
one speaker' but 'a group of people which includes the speaker'. If, for example,
we change (164) into We completely trust you, this sentence will acquire the
following meaning: 'the speaker believes that he or she is a member of some
group each of whose members, including the speaker, trusts the addressee(s)'.
We is thus a quasi-idiom in relation to I: the signified of the former does not
only contain the meanings 'the first person' and 'the plural number' but also the
idiomatic meaning 'a group including [the speaker and at least one other
individual]'. As for they, the idiomatic meaning 'group' is not always part of its
signified. For example, if we change (164) in I trust them completely, this will
not necessarily imply that the speaker conceptualizes the people referred to as
them as members of some group. These can only be people whom he or she
happens to know. Consequently, the signified 'they' can be analyzed in a similar
way as the signified of the plural you: 'they' = 'more than one individual
(excluding both the speaker and the addressee)'.
According to Huddleston and Pullum (2005: 100-101), personal pronouns
can fill the same syntactic positions as nouns and, accordingly, must be regarded
as a subclass of nouns rather than as an independent syntactic category.
Consider, for example, (166).
(166) He gave it me! (COCA)
With the exception of the verb gave, which functions as the predicator of the
predicate VP gave it me, all other syntactic positions in this clause are filled by
pronouns: the nominative he functions as the subject; the accusative it is the
indirect object and the accusative me is the direct object. (Note that even though
the direct object me of (166) is semantically very similar to the indirect object
the President of People gave the President the right information, me of (166) is
nevertheless the direct object: we are justified in claiming this because me is
placed after another object.17) As for the complement position, recall clauses like
it is I / me who … in which the complement position can be filled by personal
pronouns in either the nominative or the accusative case.
The two major differences between nouns and personal pronouns are as
follows.
x In contrast to nouns, personal pronouns cannot be modified by determiners.

E.g. we can say the guy but not *the he.
x In contrast to nouns, personal pronouns have only grammatical meanings.
With regard to the latter, Plungian (2000: 255) argues that personal pronouns are
"fully grammaticalized lexemes". Indeed, the pronoun I carries only the
grammatical meanings 'the first person', 'the singular number', and 'the
nominative case'. Similarly, the pronoun me carries only the grammatical
17
In clauses like (166) where the direct and the indirect object position is expressed by
personal pronouns, the order of objects can be reversed. That is, the clause He gave me it is as
grammatical as He gave it me. Nevertheless, the syntactic status of the personal pronouns me
and it must be determined only on the basis of their position in relation to each other. That is,
in He gave me it, me is the indirect object and it is the direct object. By contrast, in He gave it
me, it is the indirect object and me is the direct object.
212 Chapter 6
meanings 'the first person', 'the singular number', and 'the accusative case'.
Perhaps only the quasi-idiomatic we can be analyzed as a carrier of the lexical
meaning 'a group including [the speaker and at least one other individual]'.
Finally, it must be observed that in addition to personal pronouns, the
grammatical category of person has another distinct marker: the agreement
between the subject and the predicator verb in the present simple tense. This is
especially obvious in the case of the verb be, which has three distinct person
wordforms: am for the first person (singular), are for the second person
(singular), and is for the third person (singular). Also in the past simple tense,
we find the distinction between the wordform was, which is used with first
person and third person singular subjects (i.e. I was and she was), and the
wordform were, which is used with second person singular subjects (i.e. you
were). With the exception of modal verbs like can, may, must, etc., all other
verbs distinguish only between present tense first person / second person
singular wordforms like read (I read and you read) and present tense third
person singular wordforms like reads (e.g. She reads).
6.5.8 Number
In 6.2 the category of number was used as an illustrative example of a semantic

grammeme. As we said, there is a very clear referential difference between one
representative of a particular class of objects (e.g. a book) and more than one
representative of the same class of objects (e.g. two books). Given this
difference, we can define NUMBER as a grammeme concerned with the
expression of the referential contrast 'oneness–more-than-oneness'.
With regard to number, English nominal lexemes are traditionally classified
into the following three categories:
1. lexemes that can be realized by both singular and plural wordforms: e.g.
BOOK Å book and books; DEER Å deer and deer, where the output plural
wordform deer is identical with the singular input wordform deer; etc.
2. singularia tantum, i.e. lexemes that can be realized by singular wordforms

only: e.g. PERSEVERANCE Å perseverance, not *perseverances; LINGUISTICS
Å linguistics, not *linguisticses; etc.
3. pluralia tantum, i.e. lexemes that can be realized by plural wordforms only:
e.g. TROUSERS Å trousers, not *trouser; POLICE Å police were investigating,
but not *police was investigating; etc.
Note that -s of linguistics is not a plural morph but a quasi-linguistic unit. This
analysis is supported by the following facts:
x When the NP linguistics fills the subject position, it can only be followed by
singular verbal wordforms: i.e. e.g. we can say Linguistics is the study of
language, but not *Linguistics are the study of language.
x Linguistics is not associated with the meaning 'more-than-oneness': there is

only one department of study which we call linguistics.
The existence of singularia and pluralia tantum may raise the question of
whether NUMBER fulfills the obligatoriness requirement which we discussed in
6.1. That is, if there are nominal lexemes that lack either singular or plural
wordforms, can we still analyze the singular–plural contrast as an obligatory
grammatical contrast? The answer to this question is 'yes'. Both singularia and
pluralia tantum fulfill the obligatoriness requirement because both of them are
associated with one of the two meanings forming the grammatical category
NUMBER: singularia tantum express the meaning 'singularity' and pluralia tantum
express the meaning 'plurality'. There are no nouns expressing neither the
meaning 'singularity' nor the meaning 'plurality'.
In addition to this, observe that the existence of both singularia and pluralia
tantum is motivated by inherent properties of the concepts which they denote.
For example, both perseverance and linguistics are uncountable nouns, i.e.
nouns that cannot be modified by cardinal numerals like one, two, three, etc.
(Huddleston and Pullum 2005: 86). That is, we cannot say *two perseverances
and *two linguisticses. The obvious explanation for this is that there is only one
type of behavior that we conceptualize as perseverance – i.e. a behavior
characterized by 'continued effort to do or achieve something despite
difficulties, failure, or opposition' (MWO) – and there is only one department of
study which we conceptualize as linguistics. Similarly, the plural nature of both
trousers and police can be attributed to the 'plural' properties of the objects
'trousers' and 'police' which they denote. The latter is not really an object but an
institution consisting of numerous police officers. And trousers is an example of
what Quirk et al. (1985: 300) call summation plurals, i.e. nouns denoting 'tools,
instruments, and articles consisting of two equal parts which are joined together'.
Similar pluralia tantum include glasses, binoculars, scissors, tongs, tweezers,
etc.
Many nominal vocables consist of both countable and uncountable lexemes.
For example, the vocable COFFEE consists of the uncountable COFFEE1 (e.g. a
cup of coffee) and the countable COFFEE2 (e.g. two coffees). The latter is a
product of quasi-idiomatization of the former: i.e. while COFFEE1 means 'coffee,
i.e. a beverage made by percolation, infusion, or decoction from the roasted and
214 Chapter 6
ground seeds of a coffee plant' (MWO), COFFEE2 means 'a cup of coffee'. In
other words, the signified of COFFEE2 contains not only the signified 'coffee',
inherent in the component coffee, but also the idiomatic meaning 'a cup of'. As
Quirk et al. (1985: 298-299) observe, quasi-idiomatic countable lexemes like
COFFEE2 are typically products of semantic change of corresponding input
uncountable lexemes. Apart from COFFEE2, this is true of e.g. BEER2 'a glass of
beer' (e.g. two beers) and PLEASURE2 'an instance of pleasure' (e.g. the pleasures
of life). The reverse situation is, however, possible as well. That is, a countable
input lexeme can give rise to an uncountable output lexeme. For example,
Radden (2006) discusses the uncountable uses of house in sentences like (167).
(167) How much house can you afford? (http://tinyurl.com/4qecz)
In this sentence house does not refer to a house as a countable object but denotes
the substance 'house', just as the uncountable coffee of e.g. a cup of coffee
denotes the substance 'coffee'. The semantic change HOUSE1 'a countable object'
Æ HOUSE2 'house as substance' is a relatively recent instance of semantic change
in English, so that sentences like (167) may still be perceived as ungrammatical
by some English speakers.
6.5.9 Degrees of comparison
As defined in Section 6.1, DEGREES OF COMPARISON or simply GRADE is the set

of the mutually exclusive meanings 'the positive degree of comparison', 'the
comparative degree of comparison', and 'the superlative degree of comparison'.
From a formal standpoint, a positive adjectival wordform like pretty and a
positive adverbial wordform like soon can be seen as input wordforms that are
used for producing both comparative and superlative adjectival and adverbial
wordforms: the comparative prettier / sooner and the superlative prettiest /
soonest can be analyzed as products of inflectional affixation of the input
positive wordforms pretty / soon by means of the suffixes -er and -est. Likewise,
the analytic comparatives more beautiful / more beautifully and the analytic
superlatives most beautiful / most beautifully are products of combining the
input positive wordforms beautiful / beautifully with the analytic grams more
and most.
By contrast, from a semantic standpoint, a positive wordform can often be
considered an output wordform in relation to a corresponding input comparative
wordform. Consider, for example, the signified of the positive wordform pretty
in an NP like a pretty woman. Evidently, the characterization of a woman as a
pretty woman is based on a comparison of that woman with other women who
are less pretty than her. As was argued by Sapir (1944: 95), "all comparatives
are primary in relation to their corresponding absolutes ('positives')". Thus,

according to Sapir, large means 'larger than of average size' and small means
'smaller than of average size'. Similarly, we can argue that a pretty woman is
prettier than the average woman; a big city is bigger than the average city; a
difficult task is more difficult than the average task; etc. What this means is that
the formally less complex positive wordforms large, small, pretty, big, difficult
are semantically more complex than the formally more complex comparative
wordforms larger, smaller, prettier, bigger, more difficult: the signifieds of the
former can be analyzed as quasi-idioms in relation to the signifieds of the latter.
For example, the signified 'large' contains the signified 'larger than', inherent in
the input comparative wordform larger, plus the idiomatic meaning 'average'.
(The latter can be regarded as an idiomatic meaning because, according to Sapir,
the signifieds of some positive wordforms do not contain this meaning. For
instance, good is analyzed by Sapir not as 'better than the average quality' but as
'better than indifferent'. Similarly, much is said to mean not 'more than the
average amount' but 'more than a fair amount'.)
Wierzbicka (1972: 71-92) revises Sapir's claim that all comparative
wordforms are semantically primary in relation to their corresponding positive
wordforms. According to her, in the case of e.g. sick and healthy, the positive
meanings cannot be analyzed as 'sicker / healthier than the average person': in
contrast to a pretty woman, the characterization of a person as sick or healthy
does not seem to be based on a comparison of that person with other people who
are less sick and less healthy than that person. Hence, as Wierzbicka concludes,
sick and healthy can be regarded as positives that are semantically less complex
than the corresponding comparatives sicker and healthier: a sicker person is a
person who is more sick than some other sick person and a healthier person is a
person who is more healthy than some other healthy person.
Another important revision made by Wierzbicka concerns the meaning
'average', which, according to Sapir, is part of the signifieds of positive
wordforms like large, small, pretty, big, and difficult. As she points out,
it is clear what average means when used in respect of the elements of

a finite set. When used, however, in respect of an open, infinite set, it
becomes a problem in itself. (Wierzbicka 1972: 73; italics mine)
Indeed, what precisely is meant by the average size, the average woman, the
average city, the average task, etc.? Given this obvious difficulty, Wierzbicka
proposes the following formulaic solution:
x X1 is small. = X1 is a small X. = X1 is smaller than one would expect an X to

be. (Wierzbicka 1972: 73)
216 Chapter 6
In other words, we have a certain expectation about the size of representatives of

a particular class of objects. One representative of this class turns out to be
smaller than we thought a representative of this class would be. Similarly, a
pretty woman is not prettier than the average woman but a woman prettier than
one would expect a woman to be.
As for the superlative degree of comparison, the semantic analysis of a
superlative wordform depends on whether it realizes the relative or the absolute
superlative (Sapir 1944: 113). For example, the superlative wordform prettiest
of e.g. the prettiest of the three women is an instance of the relative superlative:
it means 'prettier than two other women (of the three women the speaker has in
mind; it is possible that some other woman is prettier than her.)'. By contrast,
prettiest of the prettiest woman is an instance of the absolute superlative: it
means 'prettier than all other women'.
In 6.5.8 we established that not all nominal lexemes have both singular and
plural wordforms. Similar to this, there are adjectival and adverbial lexemes that
lack the comparative and the superlative wordforms. For example, an American
citizen cannot be *a more American citizen than other American citizens and he
or she could not *more always live in New York. The adjective American is thus
a non-gradable adjective and always is a non-gradable adverb.
As in the case of uncountable nouns, linguistic non-gradability is motivated
by the extra-linguistic non-gradability of concepts denoted by non-gradable
adjectives and adverbs. That is, for example, a person can either be an American
or a non-American citizen: there are no more or less American citizens.
Likewise, there can only be the distinction between always and not always
living in New York, but not between more and less always living there. Also, as
in the case of nominal vocables like BEER, COFFEE, PLEASURE, HOUSE, etc.,
which consist of both countable and uncountable lexemes, there are many
adjectival vocables containing both gradable and non-gradable lexemes. For
example, an American citizen cannot be *more American citizen than other
American citizens, but he or she can be more American (i.e. have more
American qualities) than other Americans. Similarly, there can be no *more
public transport, but there can be a more public discussion.
6.5.10 Numerical qualification
Finally, let us discuss the grammatical category NUMERICAL QUALIFICATION.

Following Khaimovich and Rogovskaya (1967: 93), this grammeme can be
defined as the set of the mutually exclusive grammatical meanings 'numerical
quantity' and 'numerical order'. The former meaning is expressed by cardinal
numerals like one, two, three, four, etc. The latter meaning is expressed by
ordinal numerals like first, second, third, fourth, etc.
Cardinal numerals respond to the question how many? (OED). For example,
(168).
(168) Look again . How many Enigma days?' 'Three, five, two, seven... and
the only sample taken was by me on the day we arrived!' (COCA)
Ordinal numerals, by contrast, "designate the place […] occupied by an item in

an ordered sequence" (MWO). For example, in the ordered sequence SIX
CHAPTERS OF THE PRESENT TEXTBOOK ON ENGLISH MORPHOLOGY, the present
chapter on inflectional morphology occupies the sixth place.
Sometimes cardinal numerals are used for the expression of ordinal
meanings. For example, instead of saying the sixth chapter, we can also say
chapter six. In the latter variant, the cardinal numeral six does not answer the
question how many chapters? but designates the sixth place occupied by some
chapter in the ordered sequence CHAPTERS OF SOME BOOK. This is clearly an
idiomatic use of cardinal numerals.
As was mentioned in 6.3.1, ordinal numerals are formed in English with the
help of the suffix -th. This suffix attaches to input cardinal wordforms: e.g.
seventh Å seven + -th. The suffix -th is a very productive inflectional suffix that
attaches to the majority of input cardinals:
x fourth Å four + -th

x ninth Å nine + -th
x forty-sixth Å forty-six + -th
The only exceptions are cardinal numerals one, two, and three, which form
ordinal wordforms with the help of suppletion. That is, first is the suppletive
ordinal wordform of the input cardinal one; second is the suppletive ordinal
wordform of the input cardinal two; third is the suppletive ordinal wordform of
the input cardinal three. In addition, suppletion produces ordinal wordforms of
compound input cardinals headed by one, two, and three. That is, for example,
twenty-first is the suppletive ordinal wordform of the compound input cardinal
twenty-one; one hundred and fifty-second is the suppletive ordinal wordform of
the input cardinal one hundred and fifty-two; one thousand and ninety-third is
the suppletive ordinal wordform of the input cardinal one thousand and ninety-
three; etc.
The regular ordinal-forming suffix -th has two allomorphs: /θ/ and /əθ/. The
latter occurs in ordinal wordforms whose input cardinal wordforms end in -ty.
For example:
x /ˈsɪk.sti.əθ/ Å /ˈsɪk.sti/ + /əθ/

x /ˈeɪ.ti.əθ/ Å /ˈeɪ.ti/ + /əθ/
218 Chapter 6
x /ˈnaɪn.ti.əθ/ Å /ˈnaɪn.ti/ + /əθ/
In all other cases, the suffix has the realization /θ/. For example:
x /sɪksθ/ Å /sɪks/ + /θ/

x /nʌɪnθ/ Å/nʌɪn/ + /θ/
x /eɪˈtiːnθ/ Å /eɪˈtiːn/ + /θ/
Note also that in the case of fifth /fɪfθ/ and twelfth /twɛlfθ/, the addition of -th is
accompanied by apophony. That is, [aɪv] of the input cardinal five changes in
[ɪf] in fifth. Similarly, the final consonant [v] of the input cardinal twelve
changes in [f] in twelfth.
6.6 Exercises
a) Grammatical meanings that are members of the same grammeme are

mutually exclusive.
b) Semantic grammemes in English are considerably outnumbered by syntactic
grammemes.
c) Grammatical apophony usually produces irregular wordforms.
d) Get-passives are as obligatory as be-passives.
e) English has the prepositional dative case.
f) Simple tenses have two tense loci: the moment of utterance and some other
event in the past.
g) The progressive aspect in English is associated with the meaning 'the middle
phase of an event'.
h) The subjunctive mood is a non-fact mood which represents an action denoted
by a clause as problematic.
i) The personal pronoun you expresses the meaning 'the third person'.
j) A nominal vocable can consist of both countable and uncountable lexemes.
3. State which wordform-building mechanisms produced the following

wordforms.
a) the genitive third person its

b) the passive past tense was discussed
c) the past tense began

d) the genitive plural boys'
e) the nominative plural boys
f) the first person plural nominative we
g) the ordinal numeral thirty-fourth
h) the imperative wordform do
i) the comparative wordform more British
j) the second person present tense plural are
4. Comment on the temporal meanings of the underlined events. State whether

they are in the present or in the past tense and whether they are instances of
1) simple or perfect tenses, and 2) absolute or relative tenses.
a) I came here to see you (COCA)

b) When he came home a few months ago, he was a changed man (COCA)
c) Today I have had an extraordinary experience (COCA)
d) She will not be so eager when she notices me (COCA)
e) They go and they flip the pancakes (COCA)
f) If he had been kept in an institution, he certainly wouldn't have done so well
(COCA)
g) Through these kin relationships, participants reported having received
emotional and concrete support (COCA)
h) We did sleep together, we had done so many times over the years, but not in
a sexual way (COCA)
i) […] it's time we did something about crime (COCA)
j) I'm sorry to have come here uninvited […] (COCA)
5. Prove that:
a) the clause They kissed flauntingly (COCA) denotes a dynamic event;

b) the clause […] Clinton was far from a passive observer (COCA) denotes a
state;
c) the clause Another car crashed into ours (COCA) denotes an achievement;
d) the clause He was lying in an awkward position (COCA) denotes a process;
e) the clause She came to me (COCA) denotes a telic event;
f) the clause She has stayed in Grogan's Mill for the past two decades (COCA)
denotes an atelic event;
g) the clause Sleep well (COCA) is in the imperative mood;
h) the underlined clause in Her parole requires that she stay in Peru until her
sentence ends in November 2015 (COCA) is an instance of subjunctive I;
i) the clause She is a feature at every party (COCA) is in the indicative mood;
220 Chapter 6
j) the underlined clause If she were a different woman, would you think the
same of her? (COCA) is an instance of subjunctive II.
6. Explain the ungrammaticality of the following sentences, i.e. what is wrong

with them.
a) *The suspect gave the wrong answer the committee

b) *Him was frightened by that
c) *His results were gooder than mine
d) *He watches TV meaning 'he is doing this at the moment of utterance'
e) *The she is a pretty woman
f) *Awake was stayed by me
g) *Another car continued crashing into ours
h) *He am a good guy
i) *Martinez doesn't intimidate easily by other people
j) *Barack Obama is being the U.S. president
6.7 Further reading
With the exception of the first two sections, this chapter was largely based on
Huddleston and Pullum's The Cambridge Grammar of the English Language
(2002), a must-read for any student of English theoretical linguistics. Earlier
major reference grammars of English are Biber et al. (1999) and Quirk et al.
(1985).
An excellent introduction to grammatical semantics (i.e. the branch of
semantics which is concerned with grammatical meanings like those discussed
in this chapter) is Frawley (1992). A more concise introduction can be found in
Cruse (2004: 275-311). A more advanced reading is Mel'čuk (2006).
Due to space limitations, this chapter does not provide a discussion of how
wordform-building mechanisms in English have changed with the course of
time. The reader is therefore referred to the chapters Hogg (1992; especially
122-164) and Lass (1992; especially 91-147) in The Cambridge History of the
English Language (Volumes I and II). The former discusses wordform-building
in Old English; the latter deals with the Middle English period.
Key to exercises
Chapter 1
2. Statements b, d, e, h, i are correct.
3. Words b, c, f, i are complex words.
4. a) predicate, b) subject, c) complement, d) adjunct, e) adjunct, f) object, g)

predicate, h) subject, i) object, j) adjunct.
5. Forms a, b, c, d, f, h, i, j are words.
Chapter 2
2. Statements a, f, g are correct.
3. a) contrastive distribution, b) complementary distribution, c) contrastive

distribution, d) complementary distribution, e) free variation, f) free
variation, g) complementary distribution, h) contrastive distribution, i)
complementary distribution, j) complementary distribution.
4. a) green is a morfoid, house is a morph, b) table is a monomorphemic word,

c) both bar and man are morphs, d) both fore- and tell are morphs, e) both
with and in are submorphs, f) both per- and -mit are submorphs, g) great is a
monomorphemic word, h) both after and party are morphs, i) both re- and do
are morphs, j) bird is a morfoid, brain is a morph.
5. a) derivational suffix, b) analytic free form / root, c) free form / root, d)

inflectional suffix, e) derivational suffix, f) inflectional suffix, g) bound
submorph (cf. progress and congress), h) derivational prefix, i) free morfoid /
root, j) free form / root.
Chapter 3
2. Statements a, d, g, j are correct.

222 Key to exercises
3. a) contrastive distribution, b) complementary distribution, c) complementary

distribution, d) free variation, e) complementary distribution, f) contrastive
distribution, g) complementary distribution, h) free variation, i) free
variation, j) complementary distribution.
4. a) lexes of two different lexemes, b) lexes of two different lexemes which

form a lexeme family, c) lexes of two different lexemes, d) wordforms of the
same lexeme, e) lexes of two different lexemes which form a vocable, f)
lexes of two different lexemes which form a lexeme family, g) lexes of two
different lexemes which form a vocable, h) lexes of two different lexemes
which form a lexeme family, i) wordforms of the same lexeme, j) lexes of
two different lexemes.
5. a) compound / semi-idiom, b) simplex, c) compound / quasi-idiom, d)

prefixed word / isomorphic word, e) compound / quasi-idiom, f) simplex, g)
suffixed word / quasi-idiom, h) compound / quasi-idiom, i) prefixed word /
isomorphic word, j) compound / semi-idiom.
Chapter 4
2. Statements b, d, e, f, h, i, j are correct.
3. a) blending, b) anisomorphic affixation, c) morphological conversion, d)

morphological conversion, e) compounding, f) blending, g) anisomorphic
affixation, h) full-idiomatization of the NP mission from God, i)
compounding / semi-idiomatization of the NP soft power, j) morphological
conversion.
4. a) apophony, b) abbreviation, c) clipping, d) affixation, e) clipping, f)

syntactics' change, g) suppletion, h) clipping, i) abbreviation, j) suppletion.
5. Suffixes a, b, d, f, g, h, i are productive.
Chapter 5
2. Statements a, d, e, g, i, j are correct.

3. 5
4. German
5. no
6. 11
Key to exercises 223
7. 3
8. between 1900-1949
9. 10
Chapter 6
2. Statements a, c, g, h, j are correct.
3. a) affixation, b) analytic formation + affixation, c) apophony, d) signifier-

sharing with boys, e) affixation, f) suppletion, g) affixation, h) signifier-
sharing with the plain form / present tense do, i) analytic formation, j)
suppletion.
4. a) indefinite infinitive / relative tense, b) past simple / absolute tense, c)

present perfect / absolute tense, d) present simple / absolute tense, e) present
simple / absolute tense, f) perfect subjunctive II / relative tense, g) perfect
gerund / relative tense, h) past perfect / absolute tense, i) present subjunctive
II / relative tense, j) perfect infinitive / relative tense.
5. a) progressivization possible, b) progressivization impossible, c) cannot

combine with e.g. begin, d) is in the progressive aspect, e) it was begun in
order to be finished, f) contains a duration adjunct, g) the verb is in the plain
form / no overt subject, h) the verb is in the plain form / an overt subject, i)
the predicator verb agrees with the subject in person and number, j) the
predicator position is filled by were.
6. a) the direct object precedes the indirect object, b) the subject is expressed by
a personal pronoun in the accusative case, c) good does not have an
inflectional comparative wordform, d) present simple tense does not combine
freely with processes like watching TV, e) personal pronouns do not take
determiners, f) complements cannot serve as subjects of associated passive
clauses, g) the clause denotes an achievement, h) no person agreement
between the subject and the predicator, i) middle clauses suppress the by-
phrase, j) the clause denotes a state.
References
Algeo, J. (1998): "Vocabulary", The Cambridge History of the English

Language, Volume IV 1776-1997, ed. S. Romaine. Cambridge: Cambridge
University Press. 57-91.
Anderson, S. (1992): A-Morphous Morphology. Cambridge: Cambridge
University Press.
Apresjan, Y. (1974): Leksičeskaja Semantika. Sinonimičeskie Sredstva Jazyka
[Lexical Semantics. Synonymic Means of a Language]. Moscow: Nauka.
Aronoff, M. (1976): Word Formation in Generative Grammar. Cambridge, MA:
MIT Press.
Aronoff, M. & K. Fudeman (2005): What is Morphology? Malden: Blackwell.
Ayto, J., J. Siefring & J. Speake (eds.) (2009): Oxford Dictionary of English
Idioms. Third edition. Oxford: Oxford University Press.
Balteiro, I. (2007): The Directionality of Conversion in English. A Dia-
Synchronic Study. Bern: Peter Lang.
Bauer, L. (1979): "On the need for pragmatics in the study of nominal
compounding", Journal of Pragmatics 3, 45-50.
Bauer, L. (2003): Introducing Linguistic Morphology. Second edition.
Edinburgh: Edinburgh University Press.
Bauer, L. (2005): "Productivity: theories", Handbook of Word-Formation, ed. P.
Štekauer & R. Lieber. Dordrecht: Springer. 315-332.
Bauer, L. & R. Huddleston (2002): "Lexical word-formation", The Cambridge
Grammar of the English Language, R. Huddleston & G. Pullum.
Cambridge: Cambridge University Press. 1621-1721.
Benczes, R. (2006): Creative Compounding in English. The Semantics of
Metaphorical and Metonymical Noun-Noun Combinations. Amsterdam:
John Benjamins.
Biber, D., S. Johansson, G. Leech, S. Conrad & E. Finegan (1999): Longman
Grammar of Spoken and Written English. Harlow: Longman.
Blake, B. (1994): Case. Cambridge: Cambridge University Press.
Bloomfield, L. (1973[1934]): Language. Reprint. London: George Allen &
Unwin LTD.
Burger, H., D. Dobrovol'skij, P. Kühn & N. Norrick (eds). (2007): Phraseologie.
Ein Internationales Handbuch der Zeitgenössischen Forschung. Berlin:
Walter de Gruyter.
Burnley, D. (1992): "Lexis and semantics", The Cambridge History of the
English Language, Volume II 1066-1476, ed. N. Blake. Cambridge:
Cambridge University Press. 409-499.
Comrie, B. (1976): Aspect. Cambridge: Cambridge University Press.
226 References
Cruse, A. (2004): Meaning in Language. An Introduction to Semantics and

Pragmatics. Second edition. Oxford: Oxford University Press.
Cruttenden, A. (2008): Gimson's Pronunciation of English. Seventh edition.
London: Hodder Education.
Davies, M. (2004-): BYU-BNC: The British National Corpus. Available online
at <http://corpus.byu.edu/bnc>, accessed 29 November 2011.
Davies, M. (2008-): The Corpus of Contemporary American English (COCA).
Available online at <http://www.americancorpus.org>, accessed 29
November 2011.
Dobrovol'skij, D. & E. Piirainen (2005): Figurative Language. Cross-Cultural
and Cross-Linguistic Perspectives. Bingley: Emerald.
Dobrovol'skij, D. & E. Piirainen (2009): Zur Theorie der Phraseologie.
Kognitive und Kulturelle Aspekte. Tübingen: Stauffenburg.
Fiedler, S. (2007): English Phraseology. A Coursebook. Tübingen: Gunter Narr.
Finegan, E. (2004): Language. Its Structure and Use. Fourth edition. Boston:
Thomson Wadsworth.
Frawley, W. (1992): Linguistic Semantics. Hilsdale: Lawrence Erlbaum.
The Free Dictionary. Available at <http://www.thefreedictionary.com/>,
accessed 15 October 2011.
Ganshina, M. & N. Vasilevskaya (1964): English Grammar. Ninth revised
edition. Moscow: Vysšaja škola.
Giegerich, H. (2009): "Compounding and lexicalism", The Oxford Handbook of
Compounding, ed. R. Lieber & P. Štekauer. Oxford: Oxford University
Press. 178-200.
Ginzburg, R, S. Khidekel, G. Knyazeva & A. Sankin (1979): A Course in
Modern English Lexicology. Second edition. Moscow: Vysšaja škola.
Gut, U. (2009): Introduction to English Phonetics and Phonology. Frankfurt am
Main: Peter Lang.
Haan, F. (2006): "Typological approaches to modality", The Expression of
Modality, ed. W. Frawley. Berlin: Mouton de Gruyter. 27-69.
Halle, M. (2005): "Palatalization/velar softening: What it is and what it tells us
about the nature of language", Linguistic Inquiry 36/1, 23-41.
Hanks P, K. Hardcastle & F. Hodges (2006): A Dictionary of First Names.
Second edition. Oxford: Oxford University Press.
Harris, Z. (1942): "Morpheme alternants in linguistic analysis", Language 18/3,
169-180.
Harris, Z. (1945): "Discontinuous morphemes", Language 21/3, 121-127.
Haspelmath, M. (2011): "The indeterminacy of word segmentation and the
nature of morphology and syntax", Folia Linguistica 45/1, 31-80.
Haspelmath, M. & A. Sims (2010): Understanding Morphology. Second edition.
London: Hodder Education.
Heringer, H.-J. (2009): Morphologie. Paderborn: Wilhelm Fink.
References 227
Hockett, C. (1947): "Problems of morphemic analysis", Language 23/4, 321-

343.
Hogg, R. (1992): "Phonology and morphology", The Cambridge History of the
English Language, Volume I The Beginnings to 1066, ed. R. Hogg.
Hohenhaus, P. (2005): "Lexicalization and institutionalization", Handbook of
Word-Formation, ed. P. Štekauer & R. Lieber. Dordrecht: Springer. 353-
373.
Holder, R. (2008): Oxford Dictionary of Euphemisms. Fourth edition. Oxford:
Oxford University Press.
Horn, L. & G. Ward (eds.) (2004): The Handbook of Pragmatics. Malden:
Blackwell Publishing.
Huddleston, R. (2002a): "The verb", The Cambridge Grammar of the English
Language, R. Huddleston & G. Pullum. Cambridge: Cambridge University
Press. 71-212.
Huddleston, R. (2002b): "The clause: complements", The Cambridge Grammar
of the English Language, R. Huddleston & G. Pullum. Cambridge:
Huddleston, R. & G. Pullum (2002): The Cambridge Grammar of the English
Language. Cambridge: Cambridge University Press.
Huddleston, R. & G. Pullum (2005): A Student's Introduction to English
Grammar. Cambridge: Cambridge University Press.
Jackendoff, R. (2009): "Compounding in the parallel architecture and conceptual
semantics", The Oxford Handbook of Compounding, ed. R. Lieber & P.
Štekauer. Oxford: Oxford University Press. 105-128.
Jakobson, R. (1959): "Boas's view of grammatical meaning", Selected Writings.
Volume II, R. Jakobson (1971). Berlin/New York: Mouton de Gruyter. 489-
496.
Jakobson, R. (1972): "Some questions of meaning", On Language, R. Jakobson
& K. Pomorska (1990), ed. L. Waugh & M. Monville-Burston. Cambridge,
MA: Harvard University Press. 315-324.
Kaplan, J. (1995): English Grammar. Principles and Facts. Englewood Cliffs:
New Jersey.
Kastovsky, D. (1992): "Semantics and vocabulary", The Cambridge History of
the English Language, Volume I The Beginnings to 1066, ed. R. Hogg.
Katamba, F. & J. Stonham (2006): Morphology. Second edition. Basingstoke:
Palgrave Macmillan.
Keller, R. & I. Kirschbaum (2003): Bedeutungswandel. Eine Einführung. Berlin:
Walter de Gruyter.
Khaimovich, B. & B. Rogovskaya (1967): A Course in English Grammar.
Moscow: Vysšaja škola.
228 References
Kluge, F. (2002): Etymologisches Wörterbuch der Deutschen Sprache. 24th

edition. Berlin: Walter de Gruyter.
Kövecses, Z. & G. Radden (1998): "Metonymy: developing a cognitive
linguistic view", Cognitive Linguistics 9/1, 37–77.
Kreyer, R. (2010): Introduction to English Syntax. Frankfurt am Main: Peter
Lang.
Lakoff, G. (1987): Women, Fire, and Dangerous Things. What Categories
Reveal about the Mind. Chicago: The University of Chicago Press.
Lakoff, G. & M. Johnson (1980): Metaphors We Live by. Chicago: The
University of Chicago Press.
Lakoff, G. & M. Johnson (1999): Philosophy in the Flesh. The Embodied Mind
and its Challenge to Western Thought. New York: Basic Books.
Lass, R. (1992): "Phonology and morphology", The Cambridge History of the
English Language, Volume II 1066-1476, ed. N. Blake. Cambridge:
Lieber, R. (2009): "IE, Germanic: English", The Oxford Handbook of
Compounding, ed. R. Lieber & P. Štekauer. Oxford: Oxford University
Press. 357-369.
Lieber, R. (2010): Introducing Morphology. Cambridge: Cambridge University
Press.
Lieber, R. & P. Štekauer (2009a): "Introduction: status and definition of
compounding", The Oxford Handbook of Compounding, ed. R. Lieber & P.
Štekauer. Oxford: Oxford University Press. 3-18.
Lieber, R. & P. Štekauer (eds.) (2009b): The Oxford Handbook of
Compounding. Oxford: Oxford University Press.
Lipka, L. (2002): English Lexicology. Lexical Structure, Word Semantics and
Word-Formation. Third edition. Tübingen: Gunter Narr.
Löbner, S. (2002): Understanding Semantics. London: Arnold.
Lyons, J. (1968): Introduction to Theoretical Linguistics. Cambridge:
Cambridge University Press.
Marchand, H. (1969). The Categories and Types of Present-Day English Word-
Formation. Second completely revised and enlarged edition. Munich: Beck.
Maslov, Y. (1973): "Universal'nye semantičeskie komponenty v soderžanii
grammatičeskoj kategorii soveršennogo/nesoveršennogo vida [Universal
components in the content of the grammatical category
perfective/imperfective aspect]", Izbrannye Trudy. Aspektologija. Obščee
Jazykoznanie [Selected Works. Aspectology. General Linguistics], Y.
Maslov (2004). Moscow: Jaziki Slavjanskoj Kultury. 396-410.
Maslov, Y. (1978): "K osnovanijam sopostavitel'noj aspektologii [To the bases
of comparative aspectology]", Izbrannye Trudy. Aspektologija. Obščee
Jazykoznanie [Selected Works. Aspectology. General Linguistics], Y.
Maslov (2004). Moscow: Jaziki Slavjanskoj Kultury. 305-364.
References 229
McMahon, A. (2002): An Introduction to English Phonology. Edinburgh:

Edinburgh University Press.
Mel'čuk, I. (1968): "The structure of linguistic signs and formal-semantic
relations between them", The Russian Language in the Meaning-Text
Perspective, I. Mel'čuk (1995). Moscow/Vienna: Wiener Slawistischer
Almanach. 425-441.
Mel'čuk, I. (1979): "Syntactic, or lexical, zero", The Russian Language in the
Meaning-Text Perspective, I. Mel'čuk (1995). Moscow/Vienna: Wiener
Slawistischer Almanach. 169-205.
Mel'čuk, I. (1982): Towards a Language of Linguistics. A System of Formal
Notions for Theoretical Morphology. Munich: Fink.
Mel'čuk, I. (1995): "Phrasemes in language and phraseology in linguistics",
Idioms: Structural and Psychological Perspectives, ed. M. Everaert.
Hilsdale: Lawrence Erlbaum Associates. 167-223.
Mel'čuk, I. (1997): Cours de Morphologie Générale. Cinquième Partie: Signes
Morphologiques. Montréal: Les Pr. de l'Univ. de Montréal.
Mel'čuk, I. (2001). Kurs Obščej Morfologii. Morfologičeskie Znaki [Course in
General Morphology. Morphological Signs]. Moscow/Vienna: Wiener
Slawistischer Almanach.
Mel'čuk, I. (2006): Aspects of the Theory of Morphology. Berlin: Mouton de
Gruyter.
Mel'čuk, I. & A. Zholkovsky (1988): "An explanatory combinatorial dictionary
of Modern Russian", The Russian Language in the Meaning-Text
Perspective, I. Mel'čuk (1995). Moscow/Vienna: Wiener Slawistischer
Almanach. 17-54.
Merriam-Webster Online. Available at <http://www.merriam-webster.com/>,
accessed 29 November 2011.
Moskvin, V. (2010): Èvfemizmy v Leksičeskoj Sisteme Sovremennogo Russkogo
Jazyka [Euphemisms in the Lexical System of Present-Day Russian]. Fourth
edition. Moscow: Lenand.
Mühleisen, S. (2010): Heterogeneity in Word-Formation Patterns. Amsterdam:
John Benjamins.
Nida, E. (1948): "The identification of morphemes", Language 24/4, 414-441.
Nida, E. (1974): Morphology. The Descriptive Analysis of Words. Second
edition. Ann Arbor: The University of Michigan Press.
Oxford English Dictionary. Available at <http://oed.com/>, accessed 29
November 2011.
Palmer, F., R. Huddleston & G. Pullum (2002): "Inflectional morphology and
related matters", The Cambridge Grammar of the English Language, R.
Huddleston & G. Pullum. Cambridge: Cambridge University Press. 1565-
1619.
230 References
Payne, J. & R. Huddleston (2002): "Nouns and noun phrases", The Cambridge
The Phrase Finder. Available at <http://www.phrases.org.uk/>, accessed 29
November 2011.
Plag, I. (2003). Word-Formation in English. Cambridge: Cambridge University
Press.
Plungian, V. (2000): Obščaja Morfologija [General Morphology]. Moscow:
EditorialURSS.
Pullum, G. & R. Huddleston (2002): "Adjectives and adverbs", The Cambridge
Quirk, R., S. Greenbaum, G. Leech & J. Svartvik (1985): A Comprehensive
Grammar of the English Language. London/New York: Longman.
Radden, G. (2006): "Metonymic blending and the construction of meaning",
Presentation held at the Second International Workshop on Metaphor and
Discourse, University of Jaume I de Castelló, February 2-3 2006.
Rainer, F. (2005): "Constrains on productivity", Handbook of Word-Formation,
ed. P. Štekauer & R. Lieber. Dordrecht: Springer. 335-352.
Sapir, E. (1944): "Grading, a study in semantics", Philosophy of Science 11/2,
93-116.
Saussure, F. (1973): Cours de Linguistique Générale. Paris: Payot.
Schmid, H.-J. (2008): "New words in the mind: Concept-formation and
entrenchment of neologisms", Anglia - Zeitschrift für englische Philologie
126/1, 1-36.
Searle, J. (1975): "Indirect speech acts", Speech Acts. Syntax and Semantics.
Volume 3, ed. P. Cole & J. Morgan. New York: Academic Press. 59-82.
Stein, D. & S. Wright (eds.) (1995). Subjectivity and Subjectivisation.
Cambridge: Cambridge University Press.
Štekauer, P. & R. Lieber (eds.) (2005). Handbook of Word-Formation.
Dordrecht: Springer.
Taylor, J. (2002): Cognitive Grammar. Oxford: Oxford University Press.
Tokar, A. (2009): Metaphors of the Web 2.0. With Special Emphasis on Social
Networks and Folksonomies. Frankfurt am Main: Peter Lang.
Trubetzkoy, N. (1939): Grundzüge der Phonologie. Prag: Travaux du Cercle
Linguistique de Prague.
Trubetzkoy, N. (1969): Principles of Phonology. Berkeley: University of
California Press.
Ullmann, S. (1957): The Principles of Semantics. Oxford: Basil Blackwell.
Urban Dictionary. Available at <http://www.urbandictionary.com/>, accessed
29 November 2011.
References 231
Ward, G., B. Birner & R. Huddleston (2002): "Information packaging", The

Cambridge Grammar of the English Language, R. Huddleston & G. Pullum.
Wierzbicka, A. (1972): Semantic Primitives. Frankfurt am Main: Athenäum.
Wikipedia. Available at <http://en.wikipedia.org/wiki/Main_Page>, accessed 29
November 2011.
WordNet. Available at <http://wordnetweb.princeton.edu/perl/webwn>,
accessed 29 November 2011.
Word Spy. Available at <http://www.wordspy.com/>, accessed 29 November
2011.
Yavaş, M. (2006): Applied English Phonology. Malden: Blackwell.
Zipf, G. (1949): Human Behavior and the Principle of the Least Effort. New
York: Hafner.
Index
A analytic formation 176, 177, 181,

abbreviation 103, 105, 106, 157 197
absolute synonymy 37, 61 anisomorphic borrowing 82, 83, 85
absolute tense 195 anisomorphism 39, 45, 47
absolute transitive use 184 antepenult 141
accusative 10, 12, 13, 17, 175, 176, anterior meaning 58
179-193, 211, 212 apophony 81, 83, 101, 102, 104,
achievement 200-202, 219 106, 109, 143-145, 158, 162, 171,
acoustic phonetics 7 176-182, 218
acronym 106 arbitrariness 81
active voice 27, 175, 185, 206 arbitrary formation 81, 82, 127,
additional naming requirement 170 129, 130
additive principle of morphology articulatory phonetics 7
49 aspect 26, 27, 59, 202-206, 218
addressee 69, 176, 198, 204, 207, aspectualizer 202-204
210 atelic event 201
adjunct 10, 11, 23, 199, 201 auxiliary verb 9, 191, 197
affix 51-60, 63, 66, 79, 80, 83, 85,
99, 100, 102, 103, 107, 117, 126, B
133-144, 168, 178, 197, 207 back-clipping 104
affixation 79, 81-84, 88, 99, 100, back-formation 83, 85, 87, 88, 101,
104, 106, 107, 122-125, 130, 131, 146, 168-170, 172
133, 134, 138, 141, 143, 145, 153, bahuvrihi 150, 151, 156
167, 171, 173, 176-180, 182, 197, basic function 151
206, 214 blending 83, 84, 105, 163-167, 171,
affix vocable 134 172
African American English 12 blocking 95-99
agent 96, 97, 132, 138, 193 borrowing 82, 85, 101, 104, 106,
Aktionsart 200 130-133, 145, 159
allolex 63, 65, 75, 79, 80, 90, 97- bound morph 40, 42, 44, 50, 52, 54,
99, 103-109, 155, 173, 185 55
allomorph 33-39, 57, 61, 63, 98, bound morpheme 26, 27, 41
217
allophone 8, 33 C
allowordform 181-183, 208, 209 cardinal numeral 213, 216, 217
alphabetism 106 clipping 97, 98, 103, 104, 110, 117
alveolar articulation 8, 35 coda 128, 129
analytic form 59, 60, 177, 184 cognate object 15
234 Index
coinage 92, 94 derivational prefix 57, 82, 130, 173

combining form 25, 53, 54 derived word 66
comment 183 descriptive genitive 192
communicative rank 183 diachronic history 35, 86, 87, 89,
complement 10, 11, 13, 23, 30, 65, 121, 122, 163
108, 117, 118, 183, 186, 187, 190- diachronic point of view 34, 87, 88,
192, 211 122, 136
complementary distribution 8, 9, diachronic productivity 101
32-38, 61, 63, 65, 98, 107, 108 dictionary entry 72
complex sentence 12 differential meaning 43
complex word 2, 3, 7, 19-21, 26, direct object 186, 190, 211
40, 50, 59, 66, 75, 76 direction of conversion 116, 118,
compound sentence 11, 12 122-124
compounding 79, 83, 86, 99, 109- discontinuous morpheme 26
111, 145, 146, 148, 151, 152, 156, distributive event 201
158, 159, 162-167, 171, 172 ditransitive clause 186, 190
compound of the drum and bass- dynamic event 187, 200, 204
type 150, 154, 156
compound of the information E
fatigue-type 150, 153, 156 elliptical sentence 20, 21, 26
conceptual category 4, 23 encyclopedic knowledge 5
consolidation 92, 93 endocentric blend 164
consonant change 143 endocentric compound 152-155
consonant change accompanied by epistemic modality 198, 205
vowel change 143 established lexeme 95
continuative present perfect 199 establishment 92, 93, 95, 110
continuous morpheme 27 euphemism 90, 91
contrastive distribution 8, 9, 21, 32, exocentric blend 164
33, 60, 63, 75 exocentric compound 152-156, 163
copulative compound 154, 155 experiencer 96
creativity 90, 148 experiential co-occurrence 114, 115
cumulative morpheme 27 experiential similarity 114
expletive infixation 107
D expressive function 7
dative shift 190 expressivity 89, 90
degrees of comparison 80, 174, external phase 202
176, 194, 214, 216
dental sound 8 F
deontic modality 205 finite form 196
derivational affix 56, 58, 62 finiteness 68
derivational formation 58, 134, first person 180, 210, 212
137, 138, 141 fixed word-order 18
Index 235
fore-clipping 104 homonymy 28, 62

formal double-headedness 155 hypostatization 91
free distribution 29, 33, 37
free form 59, 63 I
free morpheme 26, 41 idiomatic phrase 64, 68, 100, 105,
free variation 8, 9, 12-14, 29, 31- 160, 166-168
38, 63, 65, 98, 208 idiomatic VP 19, 63, 64, 68, 76,
frequency criterion 123 155
full-idiom 66, 67, 70, 74, 76, 83, idiomaticity 39, 55, 58, 59, 137,
89, 95, 113, 116, 148, 199 166
full-idiomatization 81, 85, 86, 124, idiomatization of phrases and
146 sentences 83, 84, 166
full-lexicalization 93 imperative 181, 206-208, 219
future tense 197, 198 inchoative phase 202
indefinite gerund 196
G indefinite infinitive 196
General American 7, 12, 29, 31 indefiniteness 68, 69
genitive 10, 176, 179-181, 188-193, indicative mood 27, 206, 207, 209,
218, 219 219
genitive meaning 192, 193 indirect object 186, 190, 211
genuine compound 146 inessive case 193
gerund 196 infinitive 196
gram 174, 177, 190, 193, 197, 204, infix 56, 107, 133, 144, 177
206, 210, 214 inflected form 58
grammatical apophony 177 inflectional affix 60
grammatical category 2, 26, 80, inflectional marking 40, 154, 155
173-175, 185, 194, 200, 205, 209, input lex 79, 80, 83, 87, 97-99, 108,
212, 213, 216 137, 139, 153, 163, 168
grammatical meaning 2, 9, 15, 26, input meaning 2
27, 57-59, 65, 75, 80, 173, 174, input signifier 106
181, 202, 206, 210, 211, 216, 218, instantaneous event 202
220 interfix 56, 133
grammatical morpheme 26 internal stage 202, 204
grammeme 174, 175, 183, 184, interrogative sentence 9
188, 191-193, 198, 209, 212, 216, interrogative tag 10
218 intransitive verb 108
irrealis mood 207
H irregular wordform 179, 180
habitual action 59 isolatability criterion 14, 16, 17, 19,
habitual event 201 20-23, 41, 79
hapax legomenon 102, 103, 109 isomorphic affixation 82, 84
homonymic clash 96 isomorphic borrowing 81, 82
236 Index
isomorphic lexeme 66 metaphor 112, 114, 124, 148, 171

isomorphism 39 metaphorization 115, 119, 122, 124,
iterative event 201 149
metonymization 120, 121, 148, 149
L metonymy 112-114, 117, 121, 124,
lateral sound 128 146, 149
lex 64-75, 79-87, 89, 90, 93, 97- mid-clipping 104
100, 102-110, 155-118, 122, 127, middle phase 202-204, 218
129, 131, 133, 135-139, 143-146, minimum free form 16
148, 153-157, 159-161, 163, 165, modal adverb 205
166, 168, 170, 171, 173, 174, 177, modal meaning 205
182, 185 modal tag 205
lexeme 63-76, 79-103, 105-127, modalizer 205, 206
129-134, 136-139, 141-146, 148- monomorphemic word 26, 40, 52,
150, 153-160, 162-174, 177, 181, 137, 144
182, 185, 211-214, 216, 218 monotransitive clause 186, 190
lexeme family 63, 74, 76 mood 176, 194, 205-207, 209, 218,
lexeme-building mechanism 80, 219
101, 111, 124, 146, 166, 173, 182 morfoid 44-46, 50, 55, 60, 66, 67,
lexeme-formation 79, 80 70, 71, 87, 90, 137
lexeme-manufacturing 81, 127, 129 morph 25, 33-38, 40-44, 46-52, 54,
lex-formation 79, 80, 103 57, 59-61, 63, 64, 66, 67, 71, 83,
lexical conditioning 36, 61 133, 137, 144, 174, 206, 213
lexical gap 89, 90 morpheme 25-27, 33-44, 46, 48-50,
lexical meaning 9, 21, 56-58, 63, 52-54, 57, 60- 65, 71, 134, 162,
65, 173, 212 174
lexical morpheme 26 morphological conversion 81, 100,
light verb 15 109, 115, 121, 124, 171
linguistic sign 25, 27, 29-31, 60 movement criterion 18, 19
literal meaning 5, 6, 39, 45, 47, 48, multiplicative event 201
50, 63, 67, 73 mutually exclusive meanings 173,
living affix 136 174, 185, 210, 214, 216, 218
loan-translation 130, 131
N
M necessary and sufficient conditions
main verb 9, 18 4
meaning-distinguishing function 7, neo-classical compound 163
8 neologism 92, 93, 95, 103
mega-morph 48 nominative 10, 12, 13, 175, 180,
mega-morpheme 27 188-193, 211, 219
mental lexicon 4, 64, 68, 70, 75, 76 nonce-formation 92, 93
meta-linguistic comment 93
Index 237
non-continuative present perfect past simple tense 194, 212

199 pathological free variation 12
non-finite form 196 penultimate syllable 141
non-gradable adjective 216 perceived similarity 112, 115, 119
non-institutionalization 94-98 perfect gerund 196
non-progressive aspect 27, 206 perfect infinitive 196
non-segmental property 83 perfect tense 194, 195, 219
nucleus 128, 129 permanent event 201
numerical order 216 permissible syllable 128
numerical qualification 194, 216 phase aspectuality 200, 201
numerical quantity 216 phoneme 8, 21, 23, 27, 33, 43, 44,
65
O phonological conditioning 35
objective meaning 192 phrasal verb 155, 159
obligatoriness 52-53, 57, 58, 159, phrase 9-11, 52, 63, 66, 113, 139,
173, 188, 209, 213 155, 159, 161, 162, 184, 187, 191
obligatory grammatical meaning place-of-articulation assimilation
26, 40, 56-59 35, 61, 141
oblique case 188 plain form 207-209
ongoing action 59 plosive sound 127-129
onset 128, 129 plural number 2, 26, 44, 45, 173,
opacity 5, 46 174, 210
opening of the vocal tract 127 pluralia tantum 212, 213
optional lexical meaning 26, 56, 59, politeness-related context 198
60 polymorphemic word 26
ordinal numeral 176, 180, 216, 217 polysemy 28, 62
orthographic criterion 18, 19 popular / folk etymology 45
orthographic modification 104, 108 predicate 9-12, 15, 16, 30, 64, 65,
output lex 79, 98, 139, 143, 144, 175, 194, 200, 206, 211
153 predicative 10, 12, 186
predicator 10, 11, 15, 18, 65, 117,
P 207, 208, 211, 212
part-for-whole metonymy 112 prefix 56, 98, 99, 133, 138, 140
part-for-part metonymy 113, 114 prepositional verb 15
partial idiom 66 present perfect tense 195
partial idiomaticity 45 present simple tense 194, 204
participle I 177, 202, 203 present tense 2, 27, 58, 59, 173-
participle II 177-179, 181, 187 183, 194, 197-199, 203, 204, 206,
partitive meaning 193 207, 209, 212, 219
passive articulator 8 primary lex 170
passive voice 183-185 progressive aspect 202
past perfect tense 195 prospective aspect 203
238 Index
prospective phase 202 semantic change 80, 81, 109, 111,

prototype 4, 23 112, 114-116, 121, 124, 126, 145,
proverb 63, 64, 68, 75, 167 148, 171, 214
pseudo-compound 146 semantic grammeme 173, 174, 176,
pseudo-concept 94 194
purely formal mechanism 80, 81 semantic modification 2, 5, 80, 82,
purely semantic mechanism 80, 83, 85, 124, 130
111 semantic narrowing 28, 47, 72, 113
semantic role 193
Q semantic widening 113
qualitative aspectuality 200 semantically more complex 2, 88,
quantitative aspectuality 200, 201, 119, 123, 124, 215
204 semantically more regular 58
quasi-idiom 66, 68, 69, 70, 72, 73, semelfactive event 201
75, 84, 86, 88, 89, 95, 111, 115, semi-idiom 66, 67, 69-71, 76, 77,
118-124, 143, 144, 151-153, 191, 89, 90, 114, 149, 153, 199
193, 210 semi-idiomatic relation 73, 74
quasi-idiomatic relation 72, 144 signifier-sharing 181-183
quasi-idiomatic signified 70, 82, simple sentence 11
146, 191 simple tense 194, 195
quasi-idiomatization 79, 81, 83, simple word 2, 26
117, 123, 124, 143, 148, 151, 153, simultaneousness 196, 208
164, 168, 193, 213 singular number 2, 27, 173, 174,
quasi-linguistic unit 44-46, 51, 66, 211
67, 83, 133, 144, 213 singularia tantum 212, 213
six types of genitive constructions
R 191
Received Pronunciation 7, 12, 14, sociolinguistics 3, 30-33, 64, 65,
29, 31 192
referential meaning 175, 176 socio-pragmatic perspective 92
register 31 sonority sequencing generalization
regular past tense suffix 182 127, 128
relative tense 195, 196, 219 speech event 209, 210
resultative phase 202 static event 200
stative verb 187
S stone wall-combination 117, 118
schwa 144 stress shift 144
second person 180, 210, 212 stress shift accompanied by vowel
self-speech 210 change 143
semantic blocking 99 stress-attracting affix 140
semantic calque 130 stress-neutral affix 140
semantic case 193 stress-shifting affix 140
Index 239
structural perspective 92 tactically identical environment 36

stylistically irrelevant variation 13 telic event 201
stylistically relevant variation 13, temporal meaning 193, 194, 196,
14, 21, 31, 32, 35, 37, 65, 98, 208 219
subjectivisation of meaning 37 tense locus 194, 195, 198
subjunctive 181, 206-209, 218-220 terminal phase 202, 203
subjunctive I 207-209 third person 178, 180, 210, 212
subjunctive II 207-209 topic 183, 184
submorph 44-46, 50-52, 60, 66-68 transitive verb 108
suffix 56-58, 68, 79, 80, 84, 85, 88, true etymology 67, 89, 169, 170
96, 98-103, 106, 110, 122, 125,
126, 133-142, 144, 147, 150, 153, U
159, 171, 173, 178-180, 182, 184, uncountable noun 213, 216
193, 197-199, 206, 207, 214, 217, uninterruptability criterion 18, 19
218 unique unit 42, 46, 48, 52
summation plural 213 unreal event 198
suppletion 87, 103-105, 176, 180,
182, 217 V
suppression 184 velar softening 141
suppressive voice 184 vocable 63, 72-76, 86, 93, 95, 111,
syllable 29, 83, 128, 140, 141 119, 124, 127, 134, 213, 216, 218
synchronic point of view 34, 88, vocal cords 127
117, 122, 143, 145, 169 voiced sound 127, 128
synchronic relation 86, 88 voiceless sound 128
synonym 7, 29, 47 voicing 127
synonymic blocking 96 vowel change 83, 143, 144, 177-
synonymic sets criterion 122, 123 179
syntactic function 9, 22, 193
syntactic grammeme 174-175, 183, W
192, 194, 218 whole-for-part metonymy 113
syntactics 29, 30-34, 64, 98, 99, wordform 65, 75, 80, 123, 173-184,
104, 107, 108, 115, 133-141, 185 188-193, 197-199, 204, 206-210,
syntactics' blocking 99 212-220
word-formation 3, 79, 80, 110, 172
T wordhood criteria 14
taboo 89, 90, 114
tactically different environment 36, Z
37 zero morph 48-50
Textbooks in English Language and Linguistics (TELL)
Edited by Magnus Huber and Joybrato Mukherjee
Band 1 Ulrike Gut: Introduction to English Phonetics and Phonology. 2009.

Band 2 Jürgen Esser: Introduction to English Text-linguistics. 2009.
Band 3 Rolf Kreyer: Introduction to English Syntax. 2010.
Band 4 Alexander Bergs: Synchronic English Linguistics. 2012.
Band 5 Alexander Tokar: Introduction to English Morphology. 2012.
www.peterlang.de

Introduction To English Morphology

Uploaded by

Copyright:

Available Formats

Introduction To English Morphology

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Introduction To English Morphology

Uploaded by

Copyright:

Available Formats

Textbooks in English Language

and Linguistics (TELL)

© Peter Lang GmbH

x Inspired by the recent article Haspelmath (2011), it rejects the orthographic

x Following Mel'čuk (1968), it defines a linguistic sign not as a doublet

x It classifies word-formation into lexeme-formation and lex-formation.

corpora such as the Corpus of Contemporary American English and British

Düsseldorf Alexander Tokar

2.5.5 Typology of roots 59

4.4 Exercises 108

6.3 Wordform-building mechanisms 176

It has become a tradition to begin monographs and textbooks on morphology

1.1 What is morphology?

In a linguistic context morphology is usually defined as the study of the internal

1.2 Morphology and other branches of linguistics

To be able to address the aforementioned issues, we will, first of all, need to

Semantics is the branch of linguistics which is concerned with meaning. To

One of the most controversial issues in semantics is the question of whether

Phraseology is the branch of semantics which is concerned with idiomatic

(1) She's decided to break up with her boyfriend (Corpus of Contemporary

(2) My boyfriend is my best friend (http://tinyurl.com/6ys8pzz)

1.2.3 Phonetics and phonology

medal: both words have the pronunciation /ˈmɛdl/2. In contrast, in Received

As we will see in Chapters 2 and 3, the categories of contrastive distribution,

A phrase is either an individual word or (more often) a combination of words

(3) The French are fighting in Afghanistan (COCA)

(4) He met the President

(6) This is a good idea (COCA)

Like The French are fighting in Afghanistan, (6) is an example of a simple

Sociolinguistics is the branch of linguistics which is concerned with free

(9) My brother in his room (from Kaplan 1995: 13)

Thus a number of languages (e.g. Standard Russian) do not use auxiliaries in

(10) It was me who had changed (COCA)

(11) It was I who requested the treatment (COCA)

In contrast to the non-standard My brother in his room, which can be regarded

despite a traditional prescription based on the rules of Latin grammar

By contrast, there does not seem to be a noticeable stylistic difference

1.3 What is a word?

a speech sound or series of speech sounds that symbolizes and

(12) Do you think he lived a happy life? (COCA)

(15) I awaited their critique (COCA)

x the movement criterion

(25) She gave him a kiss

The ungrammaticality of (26)-(29) illustrates that the subject she of (25)

(30) *She him gave a kiss

The only possible transformations of (25) are (33) and (34).

(33) A kiss, she gave him

The uninterruptability criterion defines a word as a string of sounds that does

[m]any orthographies, especially (but not only) those based on the

exclamation mark. One of such occurrences could be a one-word interrogative

2. Which of the following statements are true?

a) A complex word consists of only one formally indivisible component.

a) He broke the record (COCA)

1.5 Further reading

Löbner (2002) and Cruse (2004) are good introductory textbooks to

This chapter focuses on the morphemic segmentation of English words as well

2.1 What is a morpheme?

A morpheme is traditionally defined as the smallest meaningful component of a

(35) Her lover had been untrue (MWO)

is stylistically neutral, formal, colloquial, slang, etc.