1

LEXICOLOGY COURSE
Year I


2
3






Contents






Introduction 5
The Structure of the English Lexicon 8
Varieties of English 18
The Linguistic Sign. Whats in a Word? 26
Denotation and Reference 39
Paradigmatic Relations 45





4
5


INTRODUCTION


Surprisingly enough, the term lexicology is not to be found in most
medium-sized dictionaries and in various handbooks and English
grammars. Only lexicography, explained as dictionary-making or the
writing and making of dictionaries, is usually mentioned. Apparently
the situation is now changing. Lexicology might be defined as the study
of the lexicon or lexis (specified as the vocabulary or total stock of
words of a language).
What is the most important, however, is that in lexicology the stock of
words or lexical items is not simply regarded as a list of isolated
elements. Lexicologists try to find out generalizations and regularities
and especially consider relations between elements. Lexicology is
therefore concerned with structures, not with a mere agglomeration of
words.
In order to illustrate this approach we will draw on some entries from
dictionaries and use as a starting point a remark made by Fillmore in his
article On the Organization of Semantic Information in the Lexicon.
Discussing the characteristics of the language of law, he points out:
The difference between jail and prison in many states is determined
by the duration of the incarceration less than one year, its a jail;
a year or more, its a prison.
So in the legal language of some parts of the United States the two
lexical items jail and prison could be said to be differentiated by a
distinctive feature (over / under one year). In other areas and in ordinary
language, the two words seem to be largely synonymous.
A relation of synonymy also seems to hold in the following
quotations from two dictionaries, where one item (namely jail) is
defined in terms of the other (namely prison). Let us first look at a
partial quotation (slightly modified) from the revised LDCE 1987:
Jail 1, also Gaol BrE n a place where criminals are kept as part of
their punishment; prison
6
Jail 2, also Gaol BrE v [T] to put in jail
Jailbird n infml a person who has spent a lot of time in prison
Jailbreak n an escape from prison
Jailer n esp. old use a person who is in charge of a prison or
prisoners
Synonymy and homonymy are counted among paradigmatic lexical
relations, since the lexical items so related stand in a relationship of
opposition or contrast to each other. Polysemy is a relationship internal
to a lexical entry or word. But we may also say that jail 1 and jail 2 are
derivationally connected and that jail 2 is a denominal verb derived
from the noun jail 1. We are then concerned with word-formation
processes. This could be justified by pointing out that the meaning of
the verb necessarily contains the meaning of the noun, and that the latter
is included in the paraphrase to put in jail. Since the derivation is not
formally expressed by a suffix, it is customary to speak in such cases of
zero-derivation. That this specific pattern is a productive word-
formative process can be seen from examples such as to bag, to bottle,
to box, to dock, and to kennel.
There are also the synonymous verbs to imprison and to incarcerate
to imprison. The latter is marked as formal (fml) in the LDCE and
literal or figurative in the COD. Only the COD gives a further zero-
derived verb to prison imprison, labelled poetical or dialect. Here, a
so-called zero-morpheme (usually symbolized as ) may be
postulated.
It should be obvious from these examples that the use of such
labels (e.g. fml), for indicating limitations of usage, is far from
identical in the two dictionaries. For example the LDCE makes many
more and finer distinctions with regard to regional varieties of English.
It is also clear that the use of an item in legal language may be very
different from the use of the same item when labelled infml. We will
have a closer look at the varieties of English later. However, it should
be stressed here that lexical variation and the restriction of items to
specific varieties must not be neglected in lexicology, as has so often
been the case in the past.
7
The relevance of syntagmatic lexical relations can also be seen from
the above examples. Here elements are not contrasted or substituted for
each other, but combined. This is the case with jail and bird which yield
the so-called compound jailbird. The second constituent has to be
understood metaphorically. This is not at all unusual, since animal
names are often transferred to human beings (cf. ass, goose, fox, rat).
Furthermore, bird in slang can denote a girl or woman, and in informal
language also a male person (e.g. Hes a dear old bird).
The next syntagm jailbreak can either be interpreted as a regular
compound, since the second constituent may stand for the whole
combination in the sense escape from prison. It could also be regarded
as a so-called pseudo-compound, with a zero-morpheme, analysable as
jail/break/ -parallel to the overt derivation prison/break/ing which also
means escape from prison.
There is also a syntagmatic relationship between jail and the suffix
er in jailer. The same systematic relationship is to be found in the
combination of the synonym prison and the identical suffix er,
resulting in prisoner. However, prisoner is a person kept in prison
(LDCE), not one in charge of a prison. From the point of view of the
system of the language, it could equally well be the other way round,
namely that the jailer was inside and the prisoner outside the prison.
Basically, the suffix only denotes someone who has something to do
with what is denoted by the derivative base.


8


THE STRUCTURE
OF THE ENGLISH LEXICON

In what sense can we say that the lexicon of English is structured, or
has structure? What does lexicon mean in the technical sense in which it
is used here? Why do we not simply speak of the vocabulary of
English? What does one understand by lexical structure?
Although it is true that, historically speaking, the vocabulary of a
language is an accumulation of words, we would claim that it is not
merely a list of items. It is certainly possible to make a number of
important generalizations. The lexicon is not simply an inventory of
unconnected, isolated elements, but it definitely has a structure. There
are various types of relations and connections between the elements and
we may establish(at least partial) regularities and recognize clear, if
partly incomplete, patterns.
If we consider language as a system of signs, or, following Saussure,
as a structure of interdependent elements, then we may distinguish at
least two types of structure in the lexicon: external (to the word) and
internal structure. Full words may substitute for each other, i.e. be in
opposition, or they may combine with each other. Such paradigmatic
and syntagmatic relations constitute external structure. We may also
look at the internal structure of lexical items, which are either
morphologically complex (such as compounds, or prefixal and suffixal
derivatives) or simple. In the last case, although there is no superficial,
morphological regularity visible, we may nevertheless find and
postulate underlying semantic structure.
Lexicon is defined in the LDCE as
a dictionary, esp. of an ancient language
and in the COD as
dictionary, esp, of Greek, Hebrew, Syriac, or Arabic; (fig) vocabulary
of a person, of a language, of a branch of knowledge, etc.
Obviously, the first definition is rather narrow and does not cover the
way we have used lexicon so far. The second part of the definition in
9
the COD comes close to it; the first part, containing the synonym
dictionary, is not of much use.
Two things should be pointed out here: both terms, lexicon and
dictionary, have been used as technical terms in the literature, and there
is no single correct use of lexicon or dictionary.
In the following, we will use lexicon in two senses that are not always
sharply distinguished: a) for a metalinguistic level, or a subcomponent
in a linguistic model (basically compatible with a variety of theories of
language); and b) in the sense of vocabulary as seen from a systematic,
synchronic point of view. We will deal with the two aspects in turn. Let
us first illustrate sense a) with the analogy of the following position of
lexicology in the classical structuralist model in Ullmanns Principles of
Semantics here reproduced as diagram. The justified criticism of the
model voiced in Jackson may be disregarded for our purposes. It is
certainly true that lexicology must include both the study of individual
words and their structure and of the overall structure of the vocabulary
as a whole, and that it cannot describe either from a purely formal point
of view, without considering semantic aspects and relations.











Morphology Semantics


In this three-dimensional linguistic model, lexicology, as the discipline
concerned with lexis or the lexicon, is represented as a level in between
phonology and syntax. All three levels, phonology, lexicology, and
Phonology
Lexicology
Syntax
Diachronic

Synchronic
10
syntax are subsystems of the entire language system. They may be
described at a particular point in time, i.e. synchronically, or in their
historical evolution, i.e. diachronically. Phonology, which is concerned
with the sound system of the language, has a formal, but no semantic
aspect. Lexicology has both a morphological and a semantic dimension,
and both may be regarded from a diachronic or a synchronic point of
view. The same holds for syntax, which is concerned with larger
linguistic units. Ullmann explicitly relates the distinction between
morphology and semantics to the duality of form and meaning and
to Saussures signifiant and signifie.
There is a controversy between the so-called lexicalists and the
transformationalists. The former argue for a very comprehensive
lexicon that contains as a list not only simple words and idiomatic
expressions but also compounds and derivatives. According to the
transformationalists, complex words are derived by a productive
mechanism of so-called transformations. Such complex lexical items
therefore need not be separately listed in the lexicon.



THE STRUCTURE OF THE LEXICON


The second sense b) of lexicon, with stress on the systematic and
synchronic point of view, distinguishes our approach from other
descriptions of the vocabulary of English.
We will draw in the following on the approach to the vocabulary of
English in Ernst Leisis book Das heutige Englisch. There are two
relevant chapters namely 2, Der gemischte Wortschatz, on the mixed
vocabulary, and 5, Die Schichtung des Englischen, on the
stratification of English. Leisis treatment can be characterized as a
synchronic description on a historical basis. He distinguishes various
historical strata and deals with the consequences of language-contact. In
chapter 5 he discusses a purely synchronic star-shaped model of the
stratification of the English vocabulary, based on a diagram in the
11
Shorter Oxford English Dictionary (SOED). Common etymological
origin and period of adopting loanwords, as well as stylistic
stratification (in a wide sense of style) result in groups or categories of
words and thus permit generalizations and impose a sort of structure on
the vocabulary.
According to Leisi, contemporary English is a unique mixture of
Germanic and Romance elements and this mixing has resulted in the
international character of the vocabulary. Its great richness allows far
finer differentiation in comparison with other languages. For example
German Tier corresponds to either animal or beast in English. Wagen
can be rendered by a series of English translations, such as car, cart,
carriage, and chariot. As equivalents of nachdenken or uberlegen,
English has think, reflect, meditate, ponder, and cogitate. Clearly, the
structure of the vocabulary is different in both languages.
This is also true for the negative consequences of the mixing of
languages. Leisi treats these under the headings Dissoziation and
hard words. Dissoziation (dissociation) describes the phenomenon
that words are unrelated or not associated with each other, which is
typical of English as opposed to German. Thus for example in German
mundlich and Dreifuss are related( vergesellschaftet or konsoziiert)
to Mund , drei and Fuss. The English equivalents oral and tripod,
however, are completely unrelated to mouth, three and foot.Other
examples of the dissociation of the English vocabulary are appendix, as
opposed to Blinddarm, hippopotamus as opposed to Nilpferd or
Flusspferd, and syringe as opposed to Spritze.
Leisi uses the term hard words (following a tradition in English
lexicography) for words of Latin or Greek origin that pose problems for
those ignorant of classical languages. This may lead to more or less
funny mistakes and erroneous use, so-called malapropism. Thus, for
example epitaph and epiteth may be easily confused as well as illiterate
and illegitimate.
A brief look at the various historical strata may perhaps not be out
of place. Though cultural contact with the Romans, partly already on
the Continent, and also through the influence of Christianity, a very
early stratum of Latin-Greek words entered the language. This origin is
12
no longer felt by the normal speaker today in words such as dish,
cheese, mint, pound, devil. The same holds for some Scandinavian
words from about the 10
th
century that today belong to the central core
of the vocabulary. This means that their frequency is very high. The
stratum contains: they, them, their, sky, skin, skill, ill, die, cast and take.
A more radical change and profound influence on the English
vocabulary occurred in the wake of the Norman conquest in 1066. Until
the 15
th
century a great number of French words were adopted that
belonged especially to the areas of court, state, law, and church (another
possible generalization) such as: sovereign, country, minister,
parliament, noble, honour, justice, religion, service, virtue, vice, pity,
beauty and also preach, arch and chaste. The influx was strongest
before 1400 but continued up to the 17
th
century.
These early words are distinguished from later French loans, such as
champagne and machine, by phonological peculiarities of the latter,
namely the pronunciation [ ] instead of [t]. In Leisis opinion the
adoption of French words at various times prepared the ground for the
invasion of learned words of Latin and Greek origin during the
humanistic period.
We mentioned before that as a positive result of the mixing of
languages, the possibility of distinguishing fine nuances of meaning
arises. Leisi points out that the distinction between manly, masculine,
male, virile is not parallel to that between womanly, womanish, female,
feminine. It must be added that it is impossible to capture the exact
differences of meaning, unless we consider the combining potential of
these words, the so-called collocations. Thus, for example, masculine
can be combined with style, woman, and pronoun, but male cannot.
We have so far considered various historical strata in the English
vocabulary. In contrast to this analysis on the basis of etymological
criteria, we will now look at the synchronic stratification of English.
Although Leisi speaks of stratum (Schichtung) in this context, this
is perhaps not a very fortunate term. It would be better to speak of areas
or zones, since stratum implies the existence of various horizontal
levels on top of each other. For the structure of the vocabulary
discussed in the following, the metaphors of circles or zones, however,
13
seem more appropriate. We can establish a relatively clear-cut centre or
core. The peripheries, however, are rather vague and fuzzy. The zones
also overlap and penetrate mutually in various ways. This is related to
the problem of varieties of English, which we will discuss in greater
detail later.
The simple image of stratum may have some justification in the
explanation of sociolectal differences in the vocabulary. Sociologists
often do not only speak of classes, but also of strata and social
stratification. How difficult it is to reach general and permanent
agreement in this field can be seen from a concrete example mentioned
in Leisis book. In 1954, Alan S.C. Ross introduced a distinction
between two kinds of words in British English: U (for upper class) vs.
non-U (for non-upper class), which were said to be linguistic class-
indicators. For example he labelled ill, cycle, mirror, dentures as non-
U, while sick, bike, looking-glass, false teeth were claimed to be U.
The distinction between U and non-U soon became part of the public
consciousness in Britain and the whole business quickly developed into
a sort of party game. The linguistic situation was already changed by
the heated discussion itself and by the eagerness of many people to be
considered as U-speakers. There was a revival of the U vs. non-U
debate around 1979. What is more important, however, than discussing
further individual examples, is to realize that the simple image of two
distinct layers or strata in society and language is far too simplistic.
Reality is far more complex, as we will see when discussing varieties of
English. Furthermore, there is often an interrelationship between
various regional, social, and situational parameters which results in the
choice of a specific lexical item. A glance at the pair ill/sick alone will
show that there are crucial differences between British and American
English, but also within each geographical variety, depending on
specific collocations- cf. illness, sickness, be sick (on the boat), sick
leave, sick parade = AmE sick call.
In the following we will give a traditional, synchronic survey of the
structure of the English vocabulary based on a diagram given in the
introduction to the Shorter Oxford English Dictionary. This is also
discussed in Leisis book and should be related to the treatment of
14
varieties of English. Here is a slightly modified version of the original
diagram:





LITERARY
COMMON
COLLOQUIAL





The representation starts out from the assumption that the vocabulary of
English contains a large central area, which is common to all media,
styles, and social classes. This concept of common English, symbolized
by COMMON, is equivalent, with respect to the lexicon, to the
common core treated in the University Grammar of English, which is
present in all varieties of English. This central area, which contains
words like e.g. come, father, chair, good, bad, very is labelled
Common English by the editors of the SOED.
Above and below we have the neighbouring areas of Literary and
Colloquial. Connected with these in a radial manner are further areas
of the vocabulary, which are also related to each other in content,
according to the opinion of the editors of the Oxford Dictionary.
The literary vocabulary contains scientific, foreign and archaic
words such as Weltanschauung or blas. The transition from other
literary words such as firmament, similitude, whence, whither to the
outer areas is gradual and fluid. On the other side Colloquial English
contains dialectal and vulgar elements, as well as words from slang and
technical language. We should add that the term slang is often used for
special languages of specific groups, e.g. in army slang, public school
scientific
foreign
archaic
written
dialectal
vulgar slang
technical
spoken
medium
15
slang, navy slang, RAF slang. There are thus no clear boundaries
between slang in this sense and technical language.
It is true, in general, that all categories used in this diagram have no
sharp boundaries. Thus Colloquial, Slang, and Vulgar cannot be
precisely and unambiguously defined. In the diagram they are
consequently close neighbours. The areas of the vocabulary denoted by
them merge and interpenetrate. The radial lines connect the peripheral
and the central areas of the vocabulary. The further we get to the
periphery, the more we are removed from the common vocabulary of
Common English.
In the following we will give an example of the general treatment of
lexical structure, not restricted to the vocabulary of English. As a model
approach, we will discuss the following schema, adapted from Coseriu.
He basically proposes the following model for describing and analysing
lexical structure:

Lexical Structures


paradigmatic syntagmatic
(combinatorial)

primary secondary solidarities

word-field word-formation

word-class


For the moment, the following explanation of the diagram should
suffice. Paradigmatic structures comprise words which can be
substituted for each other in a specific slot in a sentence. They
necessarily belong to the same syntactic class. Syntagmatic structures,
which Coseriu labels solidarities (Solidaritaten), comprise the
relationship between words which are systematically and
16
conventionally combined in a sentence. Examples for the latter are
blond and hair or bark and dog.
Let us now have a closer look at the fundamental distinction between
paradigmatic and syntagmatic relationships in the lexicon. This
distinction basically derives from the linear nature of linguistic
substance, what Saussure called the caractere lineaire du signifiant.
This linearity goes back to the fact that speech sounds follow each other
in time. Successive linguistic elements that are combined were called
syntagme by Saussure. In English, both syntagma and sometimes
syntagm are used as equivalent terms. Elements that are in opposition or
contrast in the same position in a syntagma are said to be in a
paradigmatic relationship. One can also say that they can be substituted
for each other and form a paradigm. The different relationships and the
resulting two dimensions may be illustrated with the following simple
example adapted from Crystal:








This example only demonstrates syntagmatic and paradigmatic relations
between words, i.e. between full linguistic signs. As we will see
presently, the relations also obtain between smaller linguistic units, such
as morphemes and phonemes, and between higher units, such as clauses
and sentences.
Some further illustration may be perhaps useful. On the phonological
level, the English phonemes /i/, /e/, and /ae/ in the words pit, pet and pat
stand in paradigmatic relation, since they may be substituted for each
other. Each of these phonemes, however, is in syntagmatic relation with
the English phonemes /p/ and /t/ in all three words. On the lexical level
the distinction can be illustrated as follows:

17
pit /i/ the old man
pet /e/ the tall man
pat /ae/ the old woman


The adjectives old and tall as well as the nouns man and woman are in
paradigmatic relationship. The article, the adjective, and the noun,
however, are syntagmatically related in all three cases. The following
example has been taken over in modified form from Leech, who uses it
for a different purpose:


They cast stones at the police
They chucked stones at the cops


The relation between the italicized verbs and nouns marked by vertical
arrows is paradigmatic; the relation between the verbs and the nouns in
both sentences, however, is a syntagmatic one.





18


VARIETIES OF ENGLISH



We now come to varieties of English and their relevance for the
lexicon. We have already seen that the vocabulary of English is far from
being homogeneous. This also holds for other linguistic levels. It will
immediately become clear that no language approaches the completely
homogeneous system postulated by many linguistic theories and
models.
We have noted that the vocabulary of English has a temporal
dimension visible in its stratification through various influences at
different times. English has changed so much that it is even relevant to
ask whether English before the Norman Conquest in 1066 can be said to
be the same language as modern language. Obviously the language of
Shakespeare compared to present-day English shows many more
correspondences and equivalents than Old English. Nevertheless, there
are very important differences that are often overlooked, due to the
frequent formal similarity of words. A precise analysis in many cases
shows considerable discrepancy of meaning between words used by
Shakespeare and formally identical contemporary words. A part of the
literary heritage, such words and their use in famous quotations
simultaneously belong to Early Modern English and the contemporary
language.
Besides the temporal dimension, the geographic dimension of a
language naturally also plays a very important role. The differences
between British English and American English immediately come to
mind. We will restrict ourselves to the level of the lexicon. In many
descriptions of the English language and in some school grammars we
find pairs of lists such as the following:

19
AmE: railroad, conductor, baggage, package, gas(oline), truck,
sidewalk, as opposed to BrE : railway, guard, luggage, parcel, petrol,
lorry, pavement.

It is rarely mentioned that a number of British words are perfectly
acceptable in many areas in America. On the other hand numerous
Americanisms have become quite familiar in Britain, due to an increase
in transatlantic travel and the influence of broadcast media. At
Heathrow Airport, for example, the signs read baggage not luggage. In
many cases the original American flavour has been lost completely,
especially for younger British speakers.
Before the Second World War some Americans insisted on referring to
the American language instead of the English language in America.
Today, nobody seriously claims that these are two different languages.
Nevertheless, there is a grain of salt in the well-known saying by
George Bernard Shaw that Britain and America are two nations divided
by a common language. We can conclude that the differences between
British and American English are on the same level as those between
other varieties of English, although the two might be regarded as the
most important "national standards.
Basically, the assumption that language is a homogeneous system,
which goes back to Saussures langue, echoed in Chomskys
competence, is always an abstraction. The idealization is particularly
evident in Chomskys claim that linguistic theory is concerned

with an ideal speaker-listener, in a completely homogeneous speech-
community, who knows its language perfectly.

As a rule, every speaker has at his or her disposal several variants of a
language, which becomes particularly apparent in phonology and the
choice of words. Especially in times of great mobility many members of
a speech community are subject to varying linguistic influences, which
are most clearly noticeable in the vocabulary.
Consequently, as a rule, every single speaker has his or her very
personal language. This language form of the single speaker is
20
nowadays generally called idolect. As opposed to this, dialect is the
language form of a particular group of speakers. The term dialect is
currently often used in a broad sense, i.e. both for a regional,
geographical variety, as well as for a social subclass of a speech
community. The neutral term variety is frequently preferred today since
it lacks the negative connotations, or shades of meaning, of dialect.
Sometimes a terminological distinction is made between dialect, used
for the regional varieties, and social dialect, or social-class dialect, for
the language of a socially determined group of speakers.
In Strang dialect is used for both a geographical and social
subsection of speakers of a language, and a scale of idolect-dialect-
language is set up. This is said to intersect with three other parameters
necessary for the description of a language variation: 1. medium, 2.
social role, labelled style(with inverted commas to mark the technical
sense), and 3. register. All these are obviously notational terms, and it is
not justified to speak of the correct definition of register. For Strang
both medium(denoting the distinction between writing and speech) and
style depend on the user of the language. The third type of variety,
register, according to Strang, does not depend on the user but on the use
of language.
The distinction between user vs. use of language has a long
tradition in British linguistics in the analysis of language variation. It is
sometimes equated with the opposition between dialectal varieties,
which are relatively permanent, and diatypic varieties, which depend
on situations. The former can be further subdivided into diachronic,
diatopic and diastratic varieties, according to the variation in time,
place, and social class. The diatypic varieties are often referred to as
registers. Halliday and others consider register as a complex notion
determined by the three variables: 1. field, 2. tenor, and 3. mode of
discourse. It is said that the configuration of all three variables taken
together determines situation-types and the context of situation. The
three situational features correspond to: 1. subject matter plus type of
social action, i.e. what is actually taking place(field), 2. role
relationships, i.e. who is taking part(tenor), and 3. Symbolic
organization, i.e. what part the language is playing(mode).
21
We shall now look at the explicit classification of varieties of
English as given by Quirk/Greenbaum in the UGE.

1. Region

For geographical or regional variation the term dialect is traditionally
used. As we have seen, this is a notational term that is not at all
ambiguous. It is often used for socially determined variants of a
language. Sometimes there is close interdependence between the two
variety classes 1 and 2. Thus, Cockney is really the dialect of a
particular social class living in a particular part of London. Regional
variation in language seems to be predominantly realized on the
phonological level. This is often referred to as accent, although the term
may also refer to variety class 6 Interference. We may speak of a
Welsh or Scottish accent, but also of a noticeable German or French
accent. A dialect, or rather a regional variety, may also be distinguished
by different lexical or grammatical features.
It has to be noted that there is regionally different familiarity with
specific variants of English. Thus, an Englishman will hear an
American Southener first as an American and may perhaps then classify
him as a Southener in addition. On the other hand, he may be able to
distinguish within Great Britain between Irish, Scots, Northern,
Midland, Welsh, South-Western, and may perhaps recognize several
London varieties. In North America, however, many people are able to
distinguish between Canadian, New England, Midland and Southern.
With regard to lexical variants, we have already seen a number of
examples of differences between British and American English.
Another regional variety, namely Scottish English, is perhaps less well
known, although words such as loch, kirk, bonnie, lassie, wee are
certainly not only familiar to British speakers. It may be difficult to
decide whether these are purely dialectal variants, or whether they have
a more specific descriptive meaning than their corresponding
stylistically neutral equivalents (lake, church etc.). Less well known is
e.g. outwith meaning outside, which may be found in formal written
Scottish texts.
22
Besides British and American English, the UGE distinguishes
several other national standards of English. Scots, Irish English, and
Canadian English are comparable to these two overwhelmingly
predominant standards. South African English, New Zealand English
and Australian English are in a different position in several respects.
Although they are similar in orthography and grammar to British
English, there are considerable differences in the lexicon and the
phonology.

2. Education and Social Standing

Within geographically different forms of English there is considerable
variation depending on education and social standing. This is sometimes
referred to as sociolect or social dialect. Although there is some
correlation between uneducated speech and regional dialect, the two are
certainly not coterminous. As an example of uneducated English the
UGE mentions the double negative. Educated English is often referred
to as Standard English which, however, is clearly a national term. There
is a partial correspondence between educated and literary English as
well as between uneducated and so-called substandard and slang.
With regard to the lexicon, the stratum of words of Greek and Latin
origin must be mentioned here. Hard words are a problem mainly for
the uneducated. The distinction between U-words (such as lavatory and
rich) and non-U-words (such as toilet and wealthy) is a matter of social
standing.

3. Subject Matter

The UGE points out that varieties according to subject matter are often
referred to as registers. We have already seen that this is a notational
term, and that it may be considered, for example by Halliday, as a
complex notion involving further variables. Both Strang and Halliday
define register as a variety according to use. More precisely, for
Halliday subject matter is only one element in the contextual variable
field. In Quirk class 3 is consistently replaced by field of discourse.
23
The use of particular lexical items is one of the most typical
characteristics of language variation according to subject matter or field
of discourse. The UGE mentions law, cookery, engineering, and
football in this connection. It also points out that there are certain
interrelations and dependencies between specific varieties. Thus, for
example, a well-formed legal sentence presupposes an educated variety
of English. Such connections, however, are not limited to relations
between the classes 2 and 3. There are also correlations with other
classes, as, for example, the following one.

4. Medium

This term is usually employed to distinguish the spoken from the
written variety of a language. Further subdivision is possible for the
distinctions made by Crystal and Davy, such as between simple
medium vs. complex medium. Thus, a dictation is spoken to be
written, while a radio news bulletin is written to be spoken. Many
legal expressions and specific grammatical peculiarities are only found
in written texts, such as documents and contracts.
Obviously, the choice of medium depends on the situation in which
language is used. The written medium normally presupposes the
physical absence of the addressee. There are also differences in the
devices used for the transmission of spoken and written language. In
print we have various typographical means, as well as headings and
paragraphs, for structuring the message. The spoken language, on the
other hand, makes use of devices such as intonation and speech tempo,
as well as extralinguistic communication signals, like gestures and eye
contact.
With regard to the lexicon, literary and poetical words are relevant
here, which occur practically only in the written language. Also, archaic
and extinct words are normally only found in writing and hardly in the
spoken medium. Clearly there is a close relationship here with variety
class 2. In French, the differences between the spoken and the written
medium are even more important. Here we have practically two
24
different languages co-existing side by side, the so-called code parle
and the code ecrit.

5. Attitude

In this variety class, according to the UGE, linguistic form may be
determined by the speakers or writers attitude to the hearer or reader(
the addressee), to the subject matter, or the purpose of the message. The
UGE points out that this variety is often marked by the ambiguous label
style. We must add that Hallidays tenor is a more precise equivalent,
although it does not include attitude to the subject matter. The UGE
distinguishes five variants of attitude along the following scale:
rigid/FORMAL/neutral/INFORMAL/familiar(of which only the two in
capitals are explicitly marked).
With regard to the lexicon, finer distinctions are usually drawn in
many dictionaries. There are different kinds and different degrees of
formality and informality. From the point of view of etymology it is
noteworthy that formal words are normally of classical or Romance
origin, while informal words usually derive from Anglo-Saxon.
There is a problem here sometimes, parallel to the one mentioned in
connection with variety class1, namely whether we have to do with two
formally distinct variants of the same word, or with two denotationally
different lexical items. For example Leech mentions as synonyms with
stylistic overtones: steed (poetic), horse (general), nag(slang).
According to this, horse would be stylistically neutral or unmarked,
while both steed and nag are stylistic variants of horse. However, nag is
defined in the COD as small riding-horse or pony; (colloq.) horse.

6. Interference

This class (set off from the others by a broken line in the UGE) is in fact
rather different and, as the authors state, on a very different basis. It
refers to contact with a foreign language and includes varieties caused
by the traces left by a speakers native language when speaking English.
An example of grammatical interference given by the UGE is the
25
utterance I am here since Thursday made by a Frenchman. The
grammar points out that certain varieties of English spoken in India,
Pakistan, and some African countries, might be considered as almost
reaching the status of a national standard. This shows that there is
considerable overlap with the other variety classes. Also, pidgins and
creoles are relevant here.
On the level of the lexicon, interference is probably more important
than on the phonological level. In this connection loan-words and loan-
translations must be mentioned as the result of interference. The so-
called false friends also belong here, for example when Ger sensibel,
brav, pathetisch are identified with E sensible, brave, and pathetic.
Interference can further cause change of meaning, as for example in Ger
realisieren and kontrollieren, which have acquired the additional
meanings notice and command, influence under the influence of E
realize and control.




26


THE LINGUISTIC SIGN.
WHATS IN A WORD?


First and foremost we have to look at some important models of the
linguistic sign which derive from Saussures ideas about it. Closely
connected is the further question about the definition of the meaning of
the linguistic sign. In this context we may distinguish purely language-
immanent approaches from those that take into account the extra-
linguistic reality. If, following Saussure, we separate the language
system (his langue) from its realization and application (his parole) we
then must distinguish between denotation and reference.
But the linguistic sign not only has a meaning, it has also a form. If
we start from this aspect, we must distinguish between words, word-
forms, morphemes (defined here as the smallest signs of a language),
and lexemes. The distinction between type and token, is also
relevant here.

Models of the Sign

In modern linguistics, Saussure was the first scholar to consider
language as a structured system of signs. His binary model of the sign,
however, must be considered more closely. A second important concept
is Ogden/Richardss abstract from certain facts that have been
rediscovered in linguistics during the last twenty years or so. These are
the users of signs, the context of use, and the function of the linguistic
sign. All three have since been integrated into linguistics under the label
linguistic pragmatics. A model of the linguistic sign that does not
only encompass extra-linguistic reality, but also speaker, hearer, and the
functions of language, is the so-called Organon Model postulated in
1934 by Karl Buhler.


27
Saussures Approach

In Saussures view of language as a system of signs, sign and system
are mutually conditioning, since a sign only derives its value from
within the system on the basis of its relation to other signs. This
structural insight is particularly relevant for the theory of lexical fields.
Although Saussure in his posthumously published work of 1916
explicitly deals with the concrete process of communication, as an
exchange between a speaker and a hearer, he later abstracts from this
process and concentrates on the non-individual, social, abstract
language system which he called langue.
For him the linguistic sign itself has two sides: a given notion
(concept) that is associated in the brain with a certain phonic image
(image acoustique). Both are mutually conditioning and evoke, or call
each other up, mutually.
Saussure stresses repeatedly that the linguistic sign is a mental unit,
and does not link a thing and a name, but a concept and a phonic image.
This image is for him nothing material, physical, but a mental
impression of a sound. The connection of concept and image
acoustique, of concept and sound picture, for Saussure constitutes the
signe linguistique, the linguistic sign. The notions concept and
image acoustique are later replaced by him by the terms signifie and
signifiant, which have since become internationaly accepted technical
terms, due to their precision and unambiguousness.
Saussures psychological approach is essentially different from
another linguistic school, which is also usually included under the
comprehensive label structuralism, namely American linguistics
during the forties and fifties. The North American structuralists, who
considered as their authority Bloomfield and his book Language, tried
to ban subjectivism and intuition from the description of language. A
psychological view of language was therefore frowned upon in so-
called taxonomic Structuralism as the deadly sin of mentalism.
For this reason, most linguists regarded statements about the meaning of
linguistic elements that were based on introspection as highly suspect.
Semantic description, i.e. conclusions concerning the meaning side of
28
language, was either postponed to a later period, when semantics could
be captured with more objective means, or was identified with the
results of the natural sciences. Thus, Bloomfield himself claimed that
the meaning of the English word salt was sodium chloride (NaCl).
Saussure postulates two principles: 1. the arbitrariness of the sign,
and 2. the linear character of the signifiant. For him the relation between
the two sides of the linguistic sign is fundamentally arbitrary, non-
motivated, or conventional. There is no necessary connection between
the signifie and the signifiant. Even onomatopoeic words (sound
symbolism), like cuckoo, and interjections, like ouch, are for him
basically not motivated and are acquired conventions of a specific
language system. We may, however, object to this view. With some
onomatopoeic formations, e.g. crash, bang, ding-dong etc., there exists
nevertheless a non-arbitrary relation between signifiant and signifie.
Furthermore, as Saussure himself points out, there are degrees of
arbitrariness and transitions between arbitrary and mitivated formations
and vice versa. Thus, the principle of arbitrariness only holds for simple
linguistic signs.
According to Saussures second principle, of the linearity of the
signifiant, this is made up of a chain of temporally successive elements.
The principle is based in the last resort on the fact that one cannot
produce several sounds simultaneously. The linearity corresponds to the
syntagmatic dimension of language already treated, but on the phonetic
and phonological level.
To sum up, for Saussure the linguistic sign is binary or bilateral,
since it consists of two parts. Its meaning, its content, is defined as a
concept and therefore as a psychological or mental entity. Saussure
consequently uses a binary, mentalistic model of the sign abstracted
from the users and the functions of the sign. In this binary model the
extralinguistic object denoted by the linguistic sign is not included.
This, however, plays a role in the triadic model developed by Ogden
and Richards to which we will now turn.

29
Ogden/Richardss Semiotic Triangle

The model of the linguistic sign developed by Ogden and Richards is
represented in modified form:


Thought or Reference



Symbol Referent
(word) (thing)

Besides the term semiotic triangle, the labels triangle of
signification, and referential triangle are also used in the literature.
The last term is justified by the fact that the model includes the
referent. It has to be interpreted in the following way. There is no
direct relationship between the word or symbol and the extralinguistic
thing or referent denoted by it. This is symbolized by the broken line
connecting the two, which Ogden/Richards characterize as an imputed
relation, saying that the symbol stands for the referent. The
relationship between the two is indirect and mediated by a concept or
thought, which Ogden/Richards also label reference. Since in this
schema a psychological or mental entity, namely thought, plays a role,
we have a mentalistic model here too.
The term reference, identified with thought, is used in a different
way by Ogden/Richards than in many recent linguistic theories. In the
latter, reference is usually understood as a relational concept.
The term is either used for the relation between the full linguistic
sign and an extralinguistic referent, or the action of a speaker referring
to an extra-linguistic object by means o0f a linguistic sign. According to
Ogden/Richardss semiotic triangle, there is then, for example, no direct
relation between the significant (or symbol) dog and a certain class of
living beings, or a specific element of this class. They stress the point
that the meaning of a linguistic symbol (as a concept or thought) has to
be clearly distinguished from the extralinguistic object (or referent)
30
denoted by it. Words, as linguistic signs, are therefore indirectly related
to extralinguistic referents.
For this relationship (besides reference) some linguists, such as
Coseriu have used the term designation. Coseriu postulates that
designation, as a relation between linguistic signs and extralinguistic
referents, has to be clearly distinguished from signification, i.e. the
content of words or the relation between the contents of words. We will
return to this problem.
Saussures binary model of the sign is thus expanded in the semiotic
triangle into a triadic model with the corners symbol, thought, and
referent, where sign, concept, and significatum are used).
However, this model still excludes the speaker and the hearer. Thus,
both models, the Saussurean egg and the triangle, are non-pragmatic,
in that they abstract from the users of the sign. It was as late as 1934,
when Karl Buhler, in his organon theory of language, also included the
speaker and the hearer in his theory as well as the extralinguistic
referent.

Buhlers Organon Model

Buhler proposed his model of the sign in the book Sprachtheorie,
published in 1934. His theory, following Plato, which views language
and its signs as a tool (= organon), is represented in somewhat
simplified form in the following diagram:










Representation
s
e
n
d
e
r
a
d
d
r
e
s
s
e
e
a
p
p
e
a
l
Expression
objects and relations
sign
31
This diagram has to be understood in the following way. The sign in the
centre links a sender (normally the speaker) with an addressee
(normally the hearer) and the represented objects and relations. The
connecting lines between the sign and the three elements just mentioned
symbolize the three most important functions of the complex sign, i.e.
language, namely: EXPRESSION (also called emotive function),
REPRESENTATION (also called referential function) and APPEAL(
also called conative or vocative function). Language, as well as the
individual linguistic sign, is an organon, i.e. an instrument or tool for
its users, hence the name of Buhlers linguistic model. The linguistic
sign, as an instrument, is an expression of the sender who uses it, i.e. the
speaker in the spoken and the writer in the written medium. At the same
time it serves for the representation of objects, states of affairs, and
relations, i.e. the extralinguistic referents. Thirdly the linguistic sign and
language in general have the function of appeal to the addressee, i.e. the
hearer or reader.
As an expression of the speaker or writer, i.e. due to its dependence
on the sender, the linguistic sign is a Symptom. Because of its
correlation with an extra-linguistic referent, it is at the same time a
Symbol. In view of its relation to an addressee, whose behaviour it is
meant to direct and control, the sign, finally, is a Signal. These three
types of sign, as well as functions of the complex linguistic sign (and
therefore language in general) may be summarized as follows:

1. EXPRESSION (speaker, writer) Symptom
2. REPRESENTATION (referent) Symbol
3. APPEAL (hearer, reader) Signal


Buhler is interested in the concrete speech situation and therefore in the
problem of deixis. By this term one usually understands some kind of
pointing, anchored in the speech situation, by means of gestures and
words, to the people involved, in particular the speaker, as well as to the
place and time of speaking. Deictic expressions, sometimes called
indexical signs, relate utterances to the spatio-temporal co-ordinates
32
of the act of speaking, namely to speaker, hearer, and their location.
Traditionally, a distinction is made between person deixis, place deixis,
and time deixis.
In Buhlers Sprachtheorie the important second chapter deals with
the field of pointing and the pointer words. The origin of the field of
pointing and of the orientational system is the speaker at a particular
place and a particular time. This centre point can be represented by the
pointer words, such as here, there, function primarily as Signals,
whereas naming words, such as table, tree, hat, are symbols for
extralinguistic objects. Pointer words are at the same time Symbols,
since e.g. here denotes a location, today may denote any day, and I, you
denote the sender and addressee.



THE MEANING OF SIGNS
AND KINDS OF MEANING


That meaning is a notational term (like several others we have already
discussed) becomes particularly apparent if we look at
Ogden/Richardss book entitled The Meaning of Meaning where a list
of 22 definitions of meaning is given. There is certainly no single,
correct explanation of meaning. Lyons distinguishes three kinds of
meaning, or semantic information, namely descriptive meaning,
social meaning, and expressive meaning. For him, these are
correlated with the descriptive, social, and expressive functions of
language. In most chapters of his book, however, he is solely concerned
with descriptive meaning.
Leech identifies meaning in the widest sense with communicative
value. This comprehensive notion may be split up into three groups: 1.
sense, 2. associative meaning and 3. thematic meaning. The second
category may itself be further divided into a number of subgroups (a-e),
as shown in the following diagram :

33
MEANING = 1. conceptual m.(sense) a. connotative m.
COMMUNICATIVE 2. associative m. b. stylistic m.
VALUE 3. thematic m. c. affective m.
d. reflected m.
e. collocative m.


According to Leech sense, or conceptual meaning, is the most important
element in linguistic communication. This is probably the reason why
he restricts himself, like Lyons, almost exclusively to this type of
meaning. We will return to the other kinds later, as well as to the three
levels of word meaning Leech introduced in the second edition of his
book. A crucial point in any discussion of meaning is the relationship
between language and the extralinguistic world. This is closely
connected with the question of the fuzzy nature of meaning. Both
cannot be regarded independently of the general problem of which
approach to semantics is chosen. In this context Lyons convincingly
introduces a useful distinction between reference and denotation.
Following a discussion of this distinction, we must have a look at the
various kinds of nondescriptive meaning. Finally, the opposition
between denotation and connotation leads to a consideration of the
notion of markedness.

Language and Reality

We have already seen that Coseriu stresses the importance of the
distinction between signification and designation. Designation, for him,
denotes the relationship between the full linguistic sign (combining
signifiant and signifie) and the extralinguistic object or referent. Only
signification, however, is purely linguistic and linguistically structured,
and it alone is therefore relevant for structural semantics. Nevertheless,
he does not deny the importance of designation for communication and
actual speech.
According to Coseriu the distinction between signification and
designation is necessary for two reasons. Firstly, it is possible to
34
denote the same extralinguistic object with different linguistic means.
Secondly, linguistic and extralinguistic relations must not be confused,
since knowledge of the language is not to be identified with the
knowledge of the world and of objects.
Coseriu, Lyons and other linguists often quote famous examples
from philosophers such as Frege and Husserl which demonstrate the
need to distinguish between the meaning of a linguistic expression and
its relationship to an extralinguistic object in relity. These are often
treated in the context of a discussion of what to understand by
reference. Thus, according to the logician Freges classic example, the
complex words morning star and evening star both refer to the same
extralinguistic referent, namely the planet Venus. They obviously have a
different meaning. Similarly, as Husserl noted, the expressions the
victor at Jena and the loser at Waterloo clearly differ in meaning (or
sense in Lyons terminology), but have the same reference, or, as
Coseriu would call it, designation. Although victor and loser are not
only different in meaning, but opposites, they may be used for referring
to an identical referent, to whom, for instance, the proper name
Napoleon could also have been applied.
However, in this case and strictly speaking also in the case of the
morning star and the evening star we do not have a single word for
referring to an object, but a whole expression, a so-called definite
description. Only in this case does Lyons use the term reference and
speak of an extralinguistic referent. The relationship between a single
word, or lexeme, and a class of objects (or a single element from this
class) is called denotation by Lyons, which he calls the class, or set, or
category of objects denoted by a lexeme denotatum. We will return to
these distinctions presently. What is important at the moment is that
both reference and denotation, as used by Lyons, fall together in
Coserius designation, but are strictly separated by him from
signification.
Coserius concept of meaning (signification, Bedeutung) excludes
extralinguistic objects and relations, and is therefore restricted to
language itself. His position can thus be characterized as a language-
intrinsic or language-immanent approach to structural semantics. In his
35
theory, lexical items are opposed to each other, and this opposition or
contrasts yields specific distinctive features or semantic components.
The methods or techniques of phonology, as a functional and language-
immanent discipline, are carried over to lexical and semantic structures
and applied in semantic analysis.
It must be pointed out that words are not simply names for pre-
existing extralinguistic categories. Different languages do not capture
reality in the same way. This becomes particularly evident in language
comparison, as carried out by contrastive linguistics. Thus, e.g. Ger
Student and Lehrer do not denote the same class of extralinguistic
objects as student and teacher, since in German we also have Studentin
and Lehrerin, which must be used if the referent is female. The classes,
or categories, denoted by the first two pairs of words, are therefore not
identical. Student and Studentin taken together are equivalent to the
category student in English.
The distinction that must be made in German, but is neutralized in
English, is also relevant in English when a different word class is at
stake, namely pronouns. In the singular, the linguistic distinction of
gender between he and she largely coincides with the respective
biological categories of sex. On the other hand, they may refer to
persons of any sex and also to things. The meaning of they therefore
contains less than that of he or she. In more precise terms, adopted
from the logic of classes, the intension, i.e. the set of properties which
determines the applicability of the term, is smaller. Conversely, the
extension of they is larger, since the class of things to which it may be
applied is larger, i.e. it may be used for denoting more referents.
We may say that two lexical items in one language (e.g. German)
converge in a single lexical item in another (e.g. English), and thus, that
there is convergence. Looking at the same situation from the point of
view of the other language, e.g. English, or, more concretely from the
perspective of a native speaker of that language, we have divergence,
i.e. the speaker has to make a choice between two (or more) items in the
target language when he wants to render a single item of his own
language.
36
This non-coincidence of categories, even with very concrete
everyday words, can be illustrated with the following examples. The
German category Schnecke is split up in English into snail and slug.
The category Affe corresponds to the two categories ape and monkey in
English. The situation can be represented in the following way, where
the extension of the English terms taken together is equivalent to the
range or class of referents of the German lexeme:

S c h n e c k e A f f e
snail slug ape monkey

If a German speaker wants to translate Strasse into English, he has to
make a choice between street and road. In this case we have divergence
of the lexical items in the target language English. But beware of false
friends such as Apartament/apartment, Blitz/blitz, Butterbohnen/butter
beans etc.
In the case of snail and slug there are no equivalent items in
German, at least in the simple vocabulary, the so called primary
vocabulary. However, as soon as we turn to complex lexemes, the so-
called secondary vocabulary, the differences can be rendered in German
as well, by using the compounds Weinbergschnecke and Nachtschnecke.
Syntactic paraphrases could also be used, such as e.g. Schnecke
mit/ohne Haus. This possibility is due to the interrelation of syntagmatic
modification and paradigmatic lexical structuring in languages. The
same holds for Menschenaffe/ape and Affchen/monkey.
Cases of divergence in German, for native speakers of English,
are eat vs. essen/fressen and drink vs. trinken/saufen. The English
learner of German must discover under which conditions each word is
used. For the German learner of English, however, no problem arises,
since there is convergence in eat and drink.
To show that the situation may be even more complicated, let us
look at the German equivalents for paperweight and paperknife. Here,
the German lexical items Briefbeschwerer and Briefoffner may be
considered as denoting practically the same class of extralinguistic
objects. Although the denotation (Coserius designation) of the English
37
and German linguistic signs is almost identical, the meaning of the
complex words (e.g. knife for paper(s), Offner fur Briefe) is clearly
different.
If we confine ourselves to a single language, and return to the
example student discussed above, we can see that the same referent may
be designated, or referred to, by means of a variety of different words.
The choice of words with a different meaning depends on the speaker
and his intentions. Thus, instead of student, the following linguistic
signs could be used: beginner, candidate, but also son, uncle, husband,
or daughter, aunt, wife. This is another case of what has been called
multiple designation above. Consequently, the meaning of linguistic
signs and their use for referring to extralinguistic objects or relations, in
specific situations, must be distinguished in principle.
In the previous discussion we have not been concerned with
individual extralinguistics objects in he real world, but mainly with
whole categories or classes or referents in reality. A class can be
considered as a collection of individuals. In the graphic representation
of such relations so-called Venn diagrams are often used. Class
membership may be given either by means of an extensional definition,
i.e. by listing its members, or by an intensional definition, i.e. by giving
its identifiable, essential properties.
For our purposes the most important concept is that of the inclusion
of classes, or class-inclusion. Since linguistic signs have both meaning
and denotation, two different types of inclusion are possible that must
not be confused. Either the meaning (or sense in Leechs use) of one
item may be included in that of another, or the class of extralinguistic
referents denoted by a sign may be included in the class of referents
denoted by another sign. Thus, we may distinguish between meaning
(or sense) inclusion and referential (or denotational) inclusion (derived
from the noun referent). This is represented in the following diagram:


HORSE animal mosquito

ANIMAL horse elephant
38

Meaning (sense)inclusion referential (denotational)inclusion
intension extension


As pointed out by Leech there is an inverse relationship between
meaning inclusion and referential inclusion. This is equivalent to
the traditional relationship between the intension and the extension of a
term. Thus, the meaning of animal is included in the meaning of horse,
but the class of referents of horse is included in the class of referents of
animal. Similarly, the meaning of animal is included in the meaning of
poodle, but the class of referents of poodle is included in the category,
class or set of dogs, and is therefore a subclass or subset.
This brings us back to the vexing problem of the fuzzy nature of
word meaning.



39


DENOTATION AND REFERENCE


In an article entitled Denotational Structure Labov describes
experiments he conducted to discover the conditions for the denotative
use of cup, bowl, glass and other container terms. Labov starts out by
saying that linguistic analysis is the study of linguistic categories, and
claims that the character of category boundaries has remained largely
unexamined within linguistics. The studies of container terms he reports
on deal with denotata that form continuous series of increasing width or
height. Informants were first asked to name the objects without any
particular context. In a second experiment they had to imagine them
filled with coffee, food, soup, or flowers. Besides diameter, height, and
shape, Labov also introduced material (e.g. china or glass) as a variable.
In the following diagram b) and c) show extreme cases of the varying
ratio of height to width. a) shows the typical proportion for the use of
cup, i.e. it can be considered the prototype (in more recent
terminology). Shape d) is the most favoured form for mug.










Labov shows that the boundaries between cup and bowl are vague (or
fuzzy in modern terms), and that context plays an important role, so
that in spite of increasing width, an object like b) may still be called cup
if it is filled with coffee. Presence and absence of a handle also has a
decisive influence. Labov finds two general properties of denotational
40
boundaries: (1) vagueness and (2) interdependence. By the second
term he means that variables (such as height, width, or function) are not
independent of each other.
Labov distinguishes two kinds of semantic judgements, which he
calls categorical and weighted judgements. Both are necessary in
denotational structure. The former concern yes/no-decisions, or
necessary conditions, such as the condition female for mother. The
latter concern gradable properties, such that if a variable X is present,
the use of a particular word is more or less likely.
Labov also reports on experimental studies with drawings of
bottle-shaped containers, where several dimensions such as neck
opening, width, and overall height were varied. The results were that the
object most likely to be called bottle, the prototypical bottle, appeared
to be a glass object with a narrow opening and a neck one-third the
width of the bottle. Labov also discusses properties of a prototypical
mama and a prototypical cat in child language acquisition.
Both the experimental studies on container terms reported by Labov
and the (originally purely psychological) research in Prototype Theory
or Prototype Semantics are concerned with the culture-dependent
categorization of reality, of the extralinguistic world, in natural
languages. While Labov is primarily interested in the fuzziness of
category boundaries, Prototype Theory is concerned with their
prototypical kernel. Language, or more specifically the lexicon of each
natural language, is a means for interpreting our environment, for
classifying or categorizing our experience. By such language-specific
categorization we impose a certain structure on extralinguistic reality. In
the view of the recent discipline of Cognitive Linguistics, each language
is equivalent to a particular conceptual system by means of which we
perceive, structure, organize, and understand the infinite flow of
impressions from the world we live in. Rosch has pointed out that
prototypes relieve the human brain of laborious cognitive processes by
providing an efficient processing mechanism of matching to a
prototype.
The notion of prototypic category or prototype derives from
research in psychology, from which it has been argued that we
41
recognize members of a category by matching them with a prototype.
It has recently been adopted into linguistics, e.g. in the work of Fillmore
and Leech.
When talking about categories and categorization, we are
concerned with classes of extralinguistic objects, not with single,
individual referents. This important difference is most clearly brought
out in the distinction made in Lyons between denotation and
reference. In order to clarify this distinction, we must make use of the
concept of lexeme. By this concept, coined in analogy to phoneme and
morpheme, Lyons understands an abstract linguistic unit (spelt in
capitals) with different variants (e.g. WRITE as against wrote, written).
For Lyons the relation of denotation holds between a lexeme and a
whole class of extralinguistic objects, as represented in the following
schema:

Denotation Denotatum
RELATION: lexeme class of objects
DOG, COW, ANIMAL, MAMMAL


He characterizes the denotatum of e.g. cow as a particular class of
animal, and adds that the individual animals also are its denotata. Lyons
further points out that the denotation of a lexeme is independent of the
concrete circumstances of an utterance. On the other hand, expressions
such as the cow, Johns cow, or those three cows over there may be
used to establish a relationship of reference with specific individuals as
referents. In this case the reference of these expressions containing cow
is partly determined by the denotation of the lexeme COW in the system
of the English language.
His use of the term reference can be summarized and illustrated in
the following way:


Reference Referent

42
Expression
RELATION : specific object
definite
description
the man over there, the cow, Napoleon,
the queen, the king of France.


It should be noted that Lyons distinguishes various kinds of reference,
including successful reference. This, obviously, is a pragmatic
concept, no longer purely relational and language-immanent, but based
on the concept of speech act. Lyons does not make use of it, but he
abstracts from the speaker who refers by means of a referring
expression to a referent, thus establishing the relationship of reference.
Closer inspection of Lyonss concept of reference reveals, however,
that he is not really in disagreement with Searle. It is only on a single
page in his monumental work that Lyons states explicitly that he too
considers reference as an act performed by a speaker for a hearer by
means of a linguistic expression (successfully or not). He believes that
it is the speaker who refers, namely in the act of referring.
The clearest example of such a pragmatic use of the term reference
is to be found in Searles Speech Acts.
According to Searle definite reference may be performed by
means of the following linguistic categories:
1. Proper names, e.g. Socrates, Russia;
2. Complex noun phrases in the singular often with a definite article
a category that is normally termed definite descriptions in
philosophy, e.g. the man who called, Johns brother;
3. Pronouns, e.g. this, that, I, he, she, it;
4. Titles, e.g. the Prime Minister, the Pope.
The concept of reference as used by Searle demonstrates again how
important it is to distinguish between reference or more clearly
referring as (establishing) a relation to a single extralinguistic object,
and denotation, as the relationship between a class of objects and a
lexeme.
43

Other Kinds of Meaning

Lyons distinguishes descriptive meaning from social and expressive
meaning, while Leech separates conceptual meaning (or sense) from
various types of associative meaning and from thematic meaning.
At this point it should also be mentioned that within descriptive
meaning Lyons makes a further distinction between the central or
focal, denotation of a lexeme and its total denotation. Languages
may differ with respect to the denotational boundaries of words, but still
be in agreement concerning the focal denotation of equivalent words.
Thus, certain focal types of colour, shape, texture, biological and social
function have to be recognized. For this, the notions of biological and
cultural salience are introduced by Lyons.
The different kinds of meaning are clearly related to different
language functions. The position adopted by Lyons can be summarized
as follows:

language functions :
1. descriptive

2. social
interpersonal (emotive, expressive )
3. expressive

According to Lyons, whose main concern in the book is with
descriptive meaning, the distinction between the expressive and social
functions of language, and the respective expressive and social
meanings is far from clear-cut. Both may be included under a single
term, such as interpersonal, emotive, expressive etc., all of which
have been used in publications.
Besides distinguishing between denotation and reference, Lyons
within descriptive meaning further recognizes the existence of sense.
Already in his Introduction to Theoretical Linguistics he had defined
the sense of a word as:
44
its place in a system of relationships which it contracts with other
words in the vocabulary.
This, obviously, is another relational definition of meaning (cf. the
discussion of sense-relations). Generally speaking, Lyonss approach
as a whole can be characterized as relational semantics. In his later two-
volume work on semantics, a clear, explicit definition of sense is not
given. He remarks at one point that sameness or difference of sense
depend on descriptive meaning and that expressions with the same
sense must be regarded as synonymous. Clearly, his concept of sense is
here again primarily seen from a relational point of view. This is also
apparent in the following quotation:
Sense is here defined to hold between the words or expressions of a
single language independently of the relationship, if any, which
holds between those words or expressions and their referents or
denotata.
It follows that sense is a purely language-immanent relationship. On the
other hand, both reference and denotation are relations between
linguistic units and the extralinguistic world. However, single lexemes,
as well as larger expressions, have sense and denotation. But only
expressions have reference.
With regard to lexemes it has been customary in linguistics to
distinguish between denotation and connotation. Lyonss social and
expressive meaning, as well as Leechs associative meaning can all be
regarded as various aspects of the same phenomenon that may be
summarized under the term connotation. Thematic meaning is different
in that it concerns the ordering of elements larger than lexemes, and is
thus mainly relevant for sentence semantics. This is illustrated by the
following example, where the (a) and (b) sentences are identical in
denotative or conceptual meaning, but differ in thematic meaning:

(a) My father owns the Rolls Royce.
(b) The Rolls Royce belongs to my father.



45


PARADIGMATIC RELATIONS



We distinguished on the level of the lexicon between paradigmatic
and syntagmatic lexical relations. These were defined as relations
between lexical items which are not necessarily semantic relations. Not
all possible semantic relations between elements of the lexicon,
however, are also sense-relations (in the sense introduced by Lyons in
1968). We will try to clarify these distinctions by drawing on the
concept of the binary sign as defined by Saussure. Basically, there are
four possibilities of paradigmatic relations between pairs of linguistic
signs:
1. Their signifiant, the phonological or graphemic form, is identical,
the signifie, the meaning, is different and unrelated. In this case, the
two signs, or lexical items, are traditionally called homonyms and
the relation between them is labelled homonymy. If the meaning is
partly identical, one also speaks of polysemy, or multiple meaning.
2. The reverse is given when the signifiant is different, but the
signifies are almost identical. Such a relation has for a long time
been called synonymy. Modern linguists believe that there is no
perfect or total synonymy.
3. Antonymy, in the wider sense, is a paradigmatic relation between
signs whose content is different but undoubtedly related. The
meanings of such signs are opposed in various ways.
4. In the case of hyponymy the relationship between the contents is of
a different kind, not one of oppositeness. In this case, the signs are
related in a hierarchic order. The subordinate lexical item is called
hyponym, the other one hyperonym, or (less confusingly)
superordinate (term), or archilexeme. Such hierarchic relations lead
to the postulation of one type of lexcal field.
Only the last three types of paradigmatic relations, synonymy,
antonymy and hyponymy, are semantic relations between full signs (or
46
more precisely between lexical units). Lyons has treated these under the
label sense-relations. The first type of paradigmatic relation,
homonymy, is not a meaning-relation or semantic relation between
signs, lexical items, or lexical units. In the case of homonymy there is
no relation whatsoever between the contents. On the other hand,
polysemy too is not a paradigmatic sense-relation between lexical items,
because we have a single lexeme and we can have no relations between
one and the same lexeme. However, we may well discuss the
relationship between different lexical units within a lexeme. The more
serious question, on the other hand, which has troubled linguists for a
long time, is how we can distinguish between homonymy and
polysemy. We will now turn to this subject.

Homonymy versus Polysemy

Criteria for the distinction which have been used by various linguists
are thoroughly examined in Lipka.
Some linguists have also changed their mind over the years. Thus,
Lyons claimed that
the distinction between homonymy and multiple meaning is, in the
last resort, indeterminate and arbitrary.
In Lyons he argues that it is preferable
to maximize polysemy. This will have the effect of producing a
lexicon with far fewer entries than are to be found in our standard
dictionaries.
One of his reasons for this change of position is the greater frequency of
polysemy, compared to homonymy. More importantly, he argues that
polysemy is the result of metaphorical creativity of natural languages,
and that this creativity is essential for the functioning of languages as
efficient semiotic systems.
In order to distinguish one word with several meanings (i.e.
polysemy or multiple meaning) from two different words with unrelated
meanings (i.e. homonymy) basically three types of criteria have been
used in research : 1. etymology, 2. formal identity or distinctness, and 3.
close semantic relatedness.
47
We will take up the criterion of etymology first. For ease of
reference, it is useful to adopt a convention normally followed by
dictionaries, which often distinguish words which are spelled
identically, but belong to a different word class, by raised numbers. In
order not to confuse these with the numbers of footnotes, it is better,
however, to write them below the word as subscripts. Ear1 organ of
hearing and ear2 head of corn are distinguished as homonyms,
because they were formally distinct in Old English and thus have a
different etymology. Consequently, they are treated as two separate
words or lexemes in present-day English dictionaries. Following this
argument, we would have to treat flower part of plant and flour
powder made by crushing grain as a single, polysemous word with
two meanings. The reason for this is that they are etymologically
identical, since both go back to the same Middle English word flour. In
spite of the different spelling, both are also identically pronounced.
Lyons points out that also port1 harbour and port2 fortified wine
which are certainly considered as two words by most speakers would
have to be treated as polysemous, since ultimately they both derive from
Latin portus. We can add a number of pairs of words with the same
origin, which can demonstrate that etymology cannot be a decisive
criterion. Thus, glamour and grammar, person and parson, shirt and
skirt have a common source.
The second possible criterion is formal identity or distinctness of a
linguistic sign. This must be further differentiated, although this is often
not done. We must distinguish between the spoken and the written
medium and also consider morphological and syntactic aspects. Hansen
et al. speak of complete homonymy only in the case of spoken, written,
and grammatical identity of two signs. Traditionally, a further
distinction is made between homophony and homography, as illustrated
in the following diagram:
homonymy vs. polysemy
(=multiple meaning)


homophony homography
48
flower : flour read : read
led : lead [I:] [e]
[e] [e]


Flower and flour are homophonous signs, as well as led (past participle)
and lead (noun), or red colour and read (past participle). On the other
hand, the infinitive read and the past tense and past participle of the
same verb are only spelt the same way but pronounced differently. For
this reason they are examples of homography. It must be mentioned,
however, that the distinction illustrated in the diagram is often not made
and that the terms homonym and homonymous are frequently used in
their wider sense, not only for complete homonymy.
The following is meant as an illustration of a clear case of
complete homonymy. The identical form bat clearly has two different
meanings and can therefore be assigned to two separate lexemes as in
the following diagram:

(c) a specially shaped stick for hitting the
ball in
cricket etc. = BAT1

bat


2. a flying mouselike animal = BAT2

But even with monosyllabic words (as opposed to polysyllabic ones),
different morphological and syntactic properties will lead to the
postulation of separate lexemes. We can clearly distinguish between
can1, can2, and can3 because we have a modal auxiliary in one case, a
noun in the second, and a transitive verb with the meaning put into a
can in the third case. Can3 can be interpreted as a zero-derivation from
the noun can2. Thus, different word class and meaning must lead to the
distinction of homonymous lexemes.
49
To return to polysemy, let us look at the third criterion, close
semantic relatedness. Hansen et al speaks of a relativ engen
Zusammenhang zwischen den Sememen. We opt for polysemy in two
cases : 1. Semantic inclusion or hyponymy, and 2. Semantic transfer,
i.e. metaphor and metonymy. We will here adopt this delimitation of
polysemy and will use the notation introduced above, distinguishing
lexical units and lexemes by means of subscripts. Thus, the lexeme
MAN1 can be said to contain the lexical units, man1 human being
(Ger Mensch), man2 adult male human being (Ger Mann), but not
MAN2 to furnish with men (Ger bemannen).
Tests of another kind for semantic relatedness are discussed in
Katovsky. He claims that we must decide in favour of homonymy in
three cases:
1. If two meanings of the same form belong to different lexical fields,
they have to be treated as homonyms. An example is Fr voler1 fly,
which belongs to the field of movement verbs, and voler2 steal,
which does not.
2. Membership in different word-formation families is a second
objective criterion. Thus, man1 (Ger Mensch) is related to the
compound mankind. On the other hand man2 (Ger Mann) is related
to manly, manhood, to man, and to unman.
3. The third criterion is the coordination test. Thus, from the
unacceptability of *He saw many fair girls and games we must
conclude that there are two homonymous adjectives fair.
We have seen in the discussion above that it is often not easy to
distinguish clearly between homonymy and polysemy. They are not
absolute opposites, and there are various degrees of formal and
semantic unity.

Zero-Derivation and Word Metaphors

From our conclusions we can see that the word metaphors must be
interpreted as different lexical units belonging to the same lexeme.
Zero-derivation, however, produces new lexemes with a different
(though related) meaning and superficially identical, but grammatically
50
distinct form, belonging to another word class. This can be illustrated
with the following examples:
(a) tick1 parasitic insect tick2 = lexical units
annoying person
(b) tick3 sound of a clock tick4 / , verb = homonyms
or watch make ticking sounds
(c) tick5 a mark in a list tick6 / , verb = homonyms
mark with a tick
The literal and figurative nouns tick in (a) are two lexical units within
the same polysemous lexeme. The verbs in (b) and (c) are derived by
zero-derivation from the respective nouns and therefore homonyms.
There are three nouns in the diagram are also homonymous lexemes,
since they are semantically unrelated.



SYNONYMY, HYPERONYMY, INCOMPATIBILITY,
ANTONYMY


Synonymy

Lyons points out that neither total synonymy nor complete
synonymy exist and then concentrates on cognitive synonymy. This
is defined precisely in terms of bilateral implication or equivalence
which may be illustrated with the following example:
x
(S1) The Arabian sheikh was wealthy, and
(S2) The Arabian sheikh was rich
y
S1 S2 and S2 S1, therefore S1 S2 = equivalence

This can be explained as follows. Two sentences S1 and S2 must
mutually imply each other ( is the logical symbol for imply,
implication). If this is the case, then S1 and S2 are equivalent (
51
stands for is equivalent to). In addition, the two sentences must have
an identical syntactic structure. If, furthermore, one differs from the
other only in that where one has a lexical item x, the other has y (in the
same position), then x and y are synonymous.
On closer inspection we realize that synonymy does not hold
between lexical items (as claimed by Lyons) but between lexical
units. If, for example, we distinguish between rich1 wealthy and rich2
fertile, then, obviously, the latter is not synonymous with wealthy. Let
us look at some additional examples. We may distinguish e.g. the
lexical units pig1 animal, and pig2 policeman. The lexical unit pig2
is never synonymous with the lexeme swine.
The examples demonstrate not only that synonymy is a matter of
lexical units, but that cognitively more or less identical words usually
differ in their connotations. It is perhaps worth pointing out that Lyons
proposes to extend the notion of synonymy beyond equivalent lexical
items to groups of lexemes. One could then call drake and its
paraphrase male duck synonyms. This proposal results from the
recognition of an important principle of the interrelation between
paradigmatic lexical structuring and syntagmatic modification. Several
other linguists even use the terms synonym and synonymy for
relationships between sentences.
According to Lyons synonymy is not a structural relationship,
especially if compared to hyponymy. In other words, it is fairly
irrelevant for the structure of the lexicon of a language. If all instances
of synonymy were eliminated from the vocabulary, this would not
effect the structural relationship between the remaining lexical items.
The so-called impoverished lexicon would no longer provide stylistic
variety and choice, but everything could be said with the smaller
synonymy-free vocabulary.

Hyponymy and Incompatibility

Lyons characterizes hyponymy and incompatibility as the most
fundamental paradigmatic relations of sense. Both are largely
interdependent and extremely important for the structure of the lexicon.
52
As opposed to synonymy, they cannot simply be eliminated without
serious consequences.
Let us look at hyponymy first, which is the relation of lexical
subordination or superordination. Lyons gives a definition in terms of
unilateral implication, i.e. logical implication of lexemes in identical
syntactic position. The subordinate term (or hyponym) necessarily
implies the superordinate one, but not vice versa. The latter (which may
sometimes be missing in a natural language, because there is a lexical
gap in the system) is called hyperonym by some linguists. We will avoid
this term in the following, because of the danger of confusion with
hyponym.
Following Coserius and Kastovskys usage we will prefer the
unambiguous term archilexeme.
Hyponyms which are on the same level of the hierarchy are
labelled co-hyponyms. The following simple diagram is meant to
illustrate both the nature of the sense-relation and the terminology used
by a number of linguists.

Flower superordinate (term)
tulip violet rose (=archilexeme)

HYPONYMY
(co-) hyponyms ( )


Lyons defines the unilateral (or assymetrical) implication between e.g.
tulip and flower as proper hyponymy. He also points out that the
principle of hyponymy enables us to be more general or more specific
according to circumstances. On the same page, he points out that
synonymy may be interpreted as a special case of hyponymy, namely as
symmetrical hyponymy. If x is a hyponym of y and if y is also a
hyponym of x, then x and y are synonyms. Thus, logical implication can
be made the underlying basis for both lexical sense-relations.
Again, it must be pointed out that hyponymy is a relation between
lexical units, not lexemes.
53
Incompatibility is defined by Lyons on the basis of the
contradictoriness between two sentences. If S1 explicitly or implicitly
denies another sentence S2 and the two sentences only differ in that one
has the lexical item x and the other has y (in the same syntactic position)
then x and y are incompatible. This is illustrated by: Mary was wearing
a red hat and Mary was wearing a green (blue, white ...) hat. Colour-
terms are obviously incompatible lexical items. Within the same
hierarchic lexical field, all those co-hyponyms are incompatible which
belong to the same dimension. Such dimensions (which group together
a number of specific semantic features) are, for example, COLOUR,
SEX, AGE, SHAPE, PURPOSE etc. Incompatible lexemes must
therefore have important semantic features in common. Simple
difference of meaning is not the same as incompatibility.
Again, we must stress that this sense-relation is a matter of lexical
units. Senses which belong to different dimensions are not affected by
incompatibility. Examples are blue and its senses depressed or
indecent and black and its senses illegal or sinister.
To conclude this section on two closely connected sense-relations,
it seems appropriate to draw attention to their functional aspects. Lyons
points out that the principle of hyponymy enables us to be more
general or more specific according to circumstances.

Complementarity, Antonymy, and Converseness
As already pointed out, Lyons calls the traditional lexical relation of
antonymy oppositeness of meaning and distinguishes three different
types of oppositeness. He restricts the term antonymy to just one of
those three types. We will review them in turn. In doing this, we should
keep in mind that these sense-relations too are defined on the basis of
logical relationships. The most recent (corpus-based) study of binary
opposition in English is Mettinger.
Complementarity can be illustrated by pairs of words like male
and female, or single and married. It is characteristic of
complementaries that the denial of the one implies the assertion of the
other and vice versa. The definition of this sense-relation is therefore
based on implication combined with negation.
54
Thus, e.g. John is not married implies John is single, and also
John is married implies John is not single. There is no third possibility,
which is also captured by the traditional Latin expression tertium non
datur.
Lyons emphasizes that the relation only holds for the normal
usage of such pairs and that dichotomization is a very important
principle in the semantic structure of language. Thus there may well be
other possibilities besides the complementaries e.g. male and female,
namely hermaphrodite. In our terminology we could say that the two
complementaries do not cover the entire dimension of SEX.
In this connection he also states that gradability is a linguistic
characteristic. In normal usage such terms are neither qualifiable, nor
gradable. In an article published in 1980, however, Cruse sets up a
further subclass, namely gradable complementaries. Examples are the
following pairs: clean-dirty, true-false, pure-impure. In Cruse
opposites are then discussed in very great detail in the three chapters
9-11.
Antonymy (in the narrow, restricted of Lyons) is the second
subclass of oppositeness. It is distinguished from complementarity by
being based on different relationships. The assertion containing one
member does imply the negation of the other, but not vive versa. Thus,
John is good implies John is not bad. But John is not good does not
necessarily imply John is bad. The negation of one term does not
necessarily imply the assertion of the other. Using a further example,
The water is not hot does not necessarily imply The water is cold.
However, from The water is cold the negation The water is not hot does
follow. Furthermore, The water is hot logically implies the negation The
water is not cold.
An important linguistic difference from complementaries (like
single/married) is that antonyms (in the narrow sense) are always fully
gradable. This is true, of course, for good/bad, hot/cold and many
adjectives. Additional criteria for the determination and
subclassification of antonymy have been used in various bublications by
Cruse, Ljung, and Geckeler.
55
Converseness is the third subclass of oppositeness of meaning
distinguished by Lyons. This term introduced by Lyons converse
denotes an equivalent mirror-image relation, or function, in which the
order of the arguments is reversed.
The criterion for the sense-relation of converseness is the
possibility of permuting noun phrases, functioning as arguments, in
sentences which remain otherwise equivalent. In other words, the
sentences imply each other and thus have the same meaning. Thus, e.g.
John bought the car from Bill implies Bill sold the car to John and vice
versa. This example with the three-place predicates buy and sell is, of
course, more complex than sentences with two-place predicates. We
will represent it schematically in the following way:

NP1 bought NP3 from NP2
NP2 sold NP3 to NP1

Mirror-image sentences are in several ways similar to the relation
between active and passive sentences.