Semantics

Stellenbosch Papers in Linguistics, Vol. 11, 1983, 115-136 doi: 10.
5774/11-0-106
115
TRANSLATIONAL SEMANTICS:
of Geoffrey Leech's
a discussion of the second edition
Semantics: the Study of Meaning
Nigel I,ove
Introduction
It LS widely accepted that a description of a natural language should

include an account of what it lS to know the meanings of its expresSlonS.
What will be referred to here as 'translational semantics' is
the idea that such an account can and ought to be provided in the form
of translations of (or a system for translating) the expressions of
the language under description into some chosen metalanguage.
It is
questionable whether (i) translational semantics is feasible, and if

it is, (ii) whether it is useful.
The first of these questions turns on the kind of translation held to
be necessary.
For many theorists, the chosen metalanguage is one
which must provide, inter alia, an exhaustively explicit enumeration

of the 'components' or 'features' which collectively comprise the meanings of units of form (e.g. 'words'), together with rules governing the
semantic interpretation of combinations of those units (e. g. 'sentences').
Construed in this way, the translationalist's enterprise requires that
the meanings of linguistic expressions be
deterrninate~
[or if they are
not, it will be unclear by what criteria a proposed feature-specification, for instance, is to be adjudged correct (i.e. exhaustive, explicit,
and in conformity with the facts of usage of the expression concerned).
It has been argued, for example by Sampson (1980), that the expressions
of natural languages are not in fact semantically determinate.
Indeed,
meanings could scarcely be other than intrinsically unstable, given the

way in which native speakers of a language learn them:
"we all spend
our time guessing what sets of criterial features would explain the
application of given words to given things in the speech we hear around
Stellenbosch Papers in Linguistics, Vol. 11, 1983, 115-136 doi: 10.5774/11-0-106
Love
us (and in the writing we read) while try

to each other's criteria
116
to conform our own usage
order to be understood" (Sampson 1980:48).
If this is so, the translationalist's demand for exhaustively explicit

and correct feature-specifications will not only be impossible to meet
in practice but incoherent in principle, for "it is not c ear what it
could mean in such a context to talk of a standard of correctness which
a given speaker has or has not achieved with respect to his use of a
G
given word" (Sampson 1980:48).

by considering "a
cal question of the kind which would have to be
glven some answer

of English:
(1980:46) illustrates his point
lational
N.L.] semantic description
the question whether possession of a handle is a criter
feature for the appl
tion of the word cup to an object".
This ques-
tion may be rephrased in various ways, depending on the descriptive

format with which the semanticist is working
e. g. 'sho.uld the
feature-specification for cup include the feature

'should a semantic description of
cup infer X has a handle"?',
analytically true?'
HANDl
ish contain a rule "from X is a
or
'is the sentence cups have handles
The quest
, in \oJhatever form, arises because
cup is one of a group of words, including mug, tumbler, beaker, vase

and others, whose meanings are similar but not identical, and it is the
task of the translational semanticist to determine precisely what set
of concepts lS summed up in the word cup, and hmV' tha t set differs from
the set summed up in tumbler, beaker, etc.
But is is far from clear
that such questions are answerable.

Less stringent demands than these may be n:ade on the chos en metalanguage.
A translational semantic account of a natural language might. for instance, be held to be accomplished
translating all of its expressions
This interpretation would circuT.vent
into another natural
on's objection, In as much as translation from one natural language

to another is clear
possible.
But even if, by allowing another natu-
ral language to stand as metalanguage to the language under descr

cs, there
we guarantee the feasibility of translational

the ques tion of its adequacy as an account of
This question has been voiced
problem is that, if to know the
know
tion,
semantic knowl edge' .
, e.g. Evans and HcDowell (1976:x). The

of an expresslon is merely to
translation into a meta1.anguage, the account of semantic know-
Love
117
ledge offered encompasses the possibility that someone might know all
the translational equivalences between the two languages who yet does
not, in any serious sense, know the meaning of any expression 1n either
of them.
Take, by way of illustration, a monoglot speaker of Afrikaans
who has before him Harrap's French-English, English-French dictionary

as his sale source of knowledge of both English and French.
According
to the traGslati0nal theory of meaning the fact that he has available

to him, for every word of English. its trarrslational equivD.1ent in French
(for purposes of the example the metalanguage in which the meanings of
English expressions are to be explicated) implies that he knows the meanings of all the words of English.
And it 1S certainly undeniable that
if asked what the English word dog means, he can reply 'chien', and if
asked what queen means, he can answer 'reine', and so on;
but if all he
knows about chien and reine turns out to be that they mean 'dog' and
'queen' respectively, it is unclear that it would be useful to ascribe
to him semantic knowledge of either English or French.
Objections to translational semantics along both of these lines are far
from novel.
One migbt therefore have expected a new, revised edition
of a popular textbook on the subject to have devoted considerable space

to an attempt to dispose of them.
The second edition of Geoffrey Leech's
Semantics (Leech 1981) is, in this respect, unsatiSfactory; 1) and the

remarks that follow are offered in the hope that they ,,,rill provide meat
that a third edition will see fit to get its teeth into.
What mean1ng 1S (not)
According to Leech (1981 :x), "Linguistics ... has brought to the subject
of semantics a certain degree of analytic rigour combined with a view of
the study of mean1ng as an integrated component of the total theory of
how language works".
The 'total theory of
Leech has in mind runs roughly as follows.
hOvl
language works' that
Language is envisaged prima-
rily as something one knows rather than something one does, and communication by means of language is a matter of implementing that knowledge.
Speaking and writing are, of course, activities, but acts of speaking or
writing or of understanding the speech and writing of others are not
Love
themselves the object of linguistic study.
118
ThGt object is, rather, the
(unconscious) knowledge that permits these activities to tJ.ke place:'.

"A .. , principle underlying many present-day approaches to semantics
see~ng
~s
the task of language study as the explication of the LINGUISTIC
COHPETENCE of the native speaker of a language;
that is, the provision
of rules and structures which specify the mental apparatus a person

must possess if he is
La
'know' a given language"
(p. S).
wl1at one is taken to have unconSC10US knowledge of 1S not 'language'

general but the one or more languages of which one
~s
~n
a native speaker.
Languages, according to the theory, are systems of correspondences

between two levels of structure, one of which is called 'form', and the
other of which is called 'meaning'.
The native speaker of a language
is held to know the set of forms comprised by that language, and the
set of meanings that correspond to them.
If confronted with the utte-
rance 'I would like a nice juicy apple' he understands it (so the theory
runs)
~n
virtue of knowing the meanings of the forms
~,
apple,
I,
juicy,
like, nice, would, and the meaning of the composite form consisting of
a sequence of those forms in the order
and apple.
2)
I,
would, like,
~,
nic~,
juicy
Conversely, if a native speaker of English wishes to
express a desire for a nice juicy apple, he knows that one way of doing
so is to utter that sequence of forms.
3)
Corrnnunication between speakers A and B of a language is seen as being

possible because A and B both have access to the system of correspondences between forms and
mean~ngs
communicate with B, A encodes his
which constitutes the language.

mean~ngs
~n
To
the appropriate forms.
To understand A, B matches up the forms which he hears A utter with the

corresponding meanings.
"mat exactly is a meaning?
This 1S not a question that bothers leech
unduly:
we might say that the whole point of setting up a theory of

semantics is to provide a 'definition' of meaning
that
is, a sys tematic account of the nature of meaning. To demand
a definition of meaning before we started discussing the subject would simply be to insist on treating certain other
concepts, e.g. stimulus and response, as in some sense more
Love
119
basic and more important. A physicist does not have to

define notions 'like 'time', 'heat', 'colour', 'atom',
before he starts investigating the.ir properties. Rather,
definitions, if they are needed, emerge from the study
itself (p. 4).
Nonetheless, Leech is prepared to tell us a certa

ceanings are noc.
The meaning of
amount about ,-<hnt
is not) for inscance, the sum
total of the true statements that can be made about
es.
The idea
that the meaning of a referring expression is the 'scientific' definition of what it refers to was put forward
Bloomfield (1955:139):
We can define the meaning of a

accurately when
this meaning has to do with some matter of which we possess
scientif~c knowledge.
\.]e can def
the names of minerals,
for example, in terms of
, as when
we say that the ordinary
ish word salt
is 'sodium chloride (NaCl)'. and we can
the names of
plants and animals by means of
cal terms of botany
and zoology, but \ve have no pr
e way of defining terms
like love or hate which concern situations that have not
been
ified, and these latter are in the
great majority.
Bloomfield's (1955:140) conclusion was that "the statement of meanl.ng

is therefore the weak point
study, and will remain so until
human knowledge advances very far beyond its present state".

field's conception of
practice, since
clear
Bloom-
makes its study impossible in
order to state the meanings of the words of a lan-
guage we need first to know
ing there is to be known about the
things referred to by those words.
Looked at from the point of view
of the speaker, the claim is that nobody knows the meaning of, for
example,
sal~,
unless he knows that salt is sodium chloride.
But
knowing that salt is sodium chloride has nothing to do, in most circumstances, with whether or not one can use the word to
successfully with other speakers of English.
Bloomfield's position
of a language knows the meanings of most of
implies that no s
the expressions he uses.
method of det
cOIT~unicate
Horeover, Leech points out that Bloomfield's
meanings leads to an infinite regression.
To
say that salt means 'sodium chloride' is to invite the question "Hhat
then do sodium and chlor
mean?".
The answers will consist of words,
Love
120
each of which will in tura require definition in terms of further words,

and so on.
In supposing that the meaning of a
could be
stic e.xpression
ained in terms of facts about the
ects it referred to,
Bloomfield was vainly attempting to locate
ng 'outside language',
whereas according to Leech
one ofthe keynotes of a modern linguistic approach to

semantics is that there is no escape from language: an
equation such as cent
hundredth of a dollar or salt =:
NaC1 is not a
w th something
outside
it is a corres
ence bet"Jeen linguistic
express
,supposedly having 'the same meaning'. '::'he
search for an
tion of ling~istic phenomena in terms
of what is not language is as vain as the search for an
ex
from a room which has no doors or windmvs, for the
word 'explanation' itself implies a statement in language.
Our
, then, is to be content with exploring what we
have
the room: to study relations within
... (p.
Some of the relations within language that Leech argues to be the proper
subject-matter of semantics are:
(i)
synonymy (the relation between the
meanings of 'I am an orphan' and 'I am a child and have

mother') ,
::10
father or
(ii) paraphrase (' the defects of the plan were obvious' and
'the demerits of the scheme were evident'), (iii) entailment ('the earth
goes round the sun' and 'the earth moves'), (iv) presupposition ('John's
son is called Marcus' and 'John has a son'), (v) inconsistency ('the
earth goes round the sun' and 'the earth is stationary').
'sameness of
Synonymy is
is taken to mean that two utterances
X and Yare synon)TffiOUS if they have the same truth-value.
4)
I f Y is syno-
nymous ivith X, then if X is true, Y is true, an(l vice-versa, and if X

is false. Y is false, and vice-versa.
Similar definitions can be stated
for the other semantic relations mentioned, according to Leech (p. 74).
The
tant point is that knowledge of these relations is taken to be

of the tacit knowledge of
s~eaker~
of
ish who assents to the truth of the earth
language.
A speaker of
es round the sun, for
example. is committed to assenting to the truth of tlle earth moves.

Part of his knowledge of English semantics is the knowledge that the
relation of entailment holds between those two sentences.
the semanticist is to det
that speakers have.
The task of
e and describe the knowledge of this sort
Love
This task requires that tvlO important distinctions be drawn.
121
The first
is between 'linguis tic knowledge I on the one hand, and 'factual' or

'real-world knowledge' on the other.
The speaker who kno\vs that I am an orphan is synonymous with I am a
child and have no father or mother knows it ln virtue of knowing che
meanlngs of the words concerned.
The truth of I am an orphan, given the
truth of I am a child and have no father or mother follows from the meaning of orphan.
One does not need to knmv any facts about particular orphans
before being in a position to assent to the synonymity of the two sentences.

Leech contrasts this case with what he calls 'factual synonymy'
(p. 75).
Charlotte lives ln Paris, for example, is factually synonymous with Charlotte lives in the capital of France.
What gives utterances of these two
sentences the same truth value is the fact that Paris happens to be the
capital of France.
But this is not a fact about the meanings of the Eng-
lish words used to express it so much as a fact about the sociopolitical

organisation of that corner of Europe,
To know that Paris is the capital
of France is to know something about the world.
But it is not similarly
the case that an orphan 'happens to be' a child with no father or mother.
To know that an orphan is a child with no father or mother is to know something about English.
A comparable distinction can be drawn for the other semantic relations
mentioned.
For example, the relation between it has been raining hard and
the ground is wet is the 'factual' counterpart of the relation between

the earth goes round the sun and the earth moves,
hard, the ground will (probably) be wet.
This 1S a matter of the connec-
If the earth goes round the sun, then
tion between physical phenomena.

the earth moves.
If it has been ra1nlng
This is a matter of the connection between the meaning
of moves and the mean1ng of goes.

relations of this latter type.
Semanticists are interes ted only in
They are interested in characterising the
knowledge someone has because he 1S a speaker of his language, not the

knowledge that he has because he 1S an intelligent inhabitant of the world.
This distinction can also be brought out with reference to two different kinds of nonsense.
The example Leech gives is the difference
between my uncle always sleeps standing on one toe and my uncle always
sleeps awake (p. 6).
are
both
The situations referred to by these sentences
impossible, but
they
are
impossible
for
different
reasons.
Love
122
The first runs counter to what we know about the postures Ln which
sleep is possible, the second to what we know about the meanings of
sleep and avake.
An analogy may be dravn here between the rules of a language

and the rules of a game. Events within a football match,
for instance, may be impossible (~\) because they violate
natural laws regarding physical strength of human beings,
the inability of footballs to defy ordinary laws of motion
(e.g. by moving in the air like boomerangs), etc. Thus a
football report that 'the centre-forvard scored a goal by
heading a ball from his own goal-line' would be disbelieved
as physically impossible, while 'the centre-forward scored
a goal by punching the ball into his own goal-mouth' would
be disbelieved on the grounds that if such a thing happened,
the game could not have been football (p. 7).
The second distinction is between 'conceptual' ('literal ') and other

kinds of meanLng.
This is rather more difficult to define, and Leech
contents himself with letting it emerge by contrast with the other kinds
that he distinguishes.
For example, 'connotative meaning' "is the com-
municative value an expression has by virtue of what it refers to, over

and above its purely conceptual content" (p. 12).
'Biped', 'posses-
sLng a womb', 'experienced in cookery' are all possible connotations of

woman.
But they are not, in Leech's terms, part of
meanLng of woman:
attributes.
th~
conceptual
one may be a woman without having any of these three
It emerges, then, that the conceptual meaning of a vord
has to do with the characteristics a referent of the word

order to be correctly so designated.
mus~
have in
In the case of woman, Leech sug-
gests, these characteristics are 'human', 'female' and 'adult'.

can 'literally' be a woman if it lacks any of these.
Nothing
The other quali-
ties that may be associated with voman, and may in some sense be said to
be part of the meaning of the word, are to be distinguished from these
as lying outside the domain of meaning that the semanticist is primarily
interes ted in.
Another type of meanLng that Leech distinguishes is 'social' ('stylistic'
in the first edition of the book) .
"SOCIAL HEANING is that which a
plece of language conveys about the social circumstances of its use" (p.
14).
These two sentences are stylistically very different:
(i) they
Love
chucked a s tone at the cops and then did a bunk wi th the
100 t
123
and
(ii) after casting a stone at the police they absconded with the money.
These are said to differ in stylistic meaning in that the information
conveyed about the social status of the speakers is different in the two
cases.
But they are nonetheless conceptually synonymous:
"their commOn
ground of conceptual meaning is evident in the difficulty anyone would

have in assentin6 to the truth of one of these sentences and denying the
truth of the other" (p.
15).
Although what is meant in principle by
conceptual meaning is fairly clear, Leech concedes that In some cases it
will be difficult to tell whether an observable difference in communicational effect between two sentences is to be explained as a difference
in conceptual meanlng or not.
pocket (Leech's
(1)
An example IS he stuck the key in his
and he put the key in his pocket (Leech's
(2)
He could argue that (1) and (2) are conceptually synonymous,

and that the difference between the two is a matter of style
(sentence (2) is neutral, while (1) is colloquial and casual).
On the other hand, we could maintain that the shift in style
is combined with a conceptual difference:
that stick in a
context such as (1) has a more precise denotation than (2)
and could be roughly defined as 'to put carelessly and quickly' .. , Often, in fact, the solution to a problem of demarcation is to conclude that quasi-synonyms differ on at least
two planes of meaning (p. 21).
Having delimited the focus of the semanticist's interest in meanlng, we

can now go on to ask exactly what doing semantics consists in.
Producing
a semantic description of, for example, English, involves (a) eliciting

and Cb) characterising the knowledge of meanlngs and relations between
meanlngs that a speaker has as a result of being a speaker of English.
Eliciting this knowledge depends on conducting empirical tests.
One
example that Leech gives is a test of knowledge of the relation of entailment (pp. 81-2).
Informants are presented with two sentences X and Yand
are instructed as follows:
Assuming X is true, judge whether Y is true or not.

If you think Y must be true, write-'YES'.
If you think ~ cannot be true, write 'NO'.
If you think Y mayor may not be true, write 'YES/NO'.
If you don't know which answer to give, write '?'
Love
124
For the case where X was someone killed the HaJrid chief ot p01ice Llst
night and Y was the Hadrid chief of police died last night, the responses
,olere as fa llows :
'NO' 0%,
'YES' 96%,
'YES/NO'
J~~,
'?' 1%.
The hypo-
thesis that speakers of English know that there is an entailment relation

between X and Y in this case was decisively confirmed.
Characterising the
~nowledge
established by tests of this kind is a mat-
ter of making it explicit by means of a formal notation.

meaning of 1.oloman is said to be 'adult human female'.
formally, [+HUMAN, -NALE, +ADULTJ
+ADULT] )
mean~ngs
The conceptual
Or, to put it more
(or, equivalently, [-+HU\1AN, +FEMALE,
These are said to be the 'features' or 'components' of the

a f the ,. . ords.
of features.
All words cons is t semant ically of corr.binations
It is sometimes claimed that there is a universal set of
semantic features:
the differences bet,. . een individual languages being a
matter of differences between the particular combinations of features

~hich
they represent as words.
One purpose of semantic feature analysis is to provide a formal characterisation of the various meaning-relations.
For example, there are
roses in his garden entails there are flowers in his garden.
In terms
compr~ses
of semantic features, this is because the specification for rose

all the features of flower, plus extra ones.
In general, X entails Y if
the semantic features of Yare a subset of the features of X.

feature specification for human is, presumably, [+HU t-'lAN ] ,
[+HUMAN -ADULT],
for boy [+HUMAN, -ADULT, -FEHALE].
for child
The features of
human are a subset of the features of child and boy.
The features of
chi Id are a subset of the features of boy.
Therefore we expect
a bOl 1n the garden to entail both there
a child
there
~s
a human ln the garden, there 1S a child
there 1S a human
so on.
~s
~n
the z,arden but not there
~s
~n
The
~n
ther~
the garden and
the garden to entail
a boy__ 1n the Earden, and
These arc predictions which might be subjected to the kind of
test with native-speaker informants illustrated above.
What ' . . e are
talking about here is primarily the meanings of individual words.

mean~ng
The
of larger linguistic units, such as the sentence, bears a com-
plex relation to the meanings of the words which compose them.
The ,vords
of which the dog bit the man consists are the same as those of Hhich the
man bit the dog consists, but the two sentences differ in meaning. A large
part of Leech's book is concerned with how sentence-meanings are to be
related to word-meanings.
Love
The role of context 1n linguistic
125
com~unication
Leech's initial failure to say just what a mean1ng 1S supposed to be is a

key to most of the problems of this approach to semantics.
It 1S taken
as axiomatic that mean1ngs are attributes of linguistic expressions

considered in the abstract, that they are 'things' that expressions
'have', and that knowing in advance wnich expresslons have which meanings is vital for communication.
The theory implicit here 1S that we
communicate in virtue of knowing meanlngs.
But consideration of one's
activity as a language-user fails to support this. Many speakers of

English, one may hazard, do not know the word outwith.
But many of
these ignoramuses c.vill, surprisingly, have no difficulty 1n understanding utterances like the following:
'Scottish candidates should report
direct to the commissioner in Glasgow.

should write in the first instance'.
Candidates from outwith Scotland

Clearly, anyone previously un-
acquainted with the word outwith who nonetheless understands this utterance does not do so in virtue of prior knowledge of its meaning.
On
the contrary, it is in virtue of understanding the utterance in which it

figures that he learns its meaning.
This latter account of the relation
between meaning and understanding seems to accord with hov.' in practice

we learn a new word.
We hear it uttered, we deduce from the context
what it must mean, and we test the correctness of that deduction by using
it with that meaning ourselves.
In so far as we thereby achieve our
communicational purpose (that is, so long as the response is not, for

instance, a blank stare), we count ourselves as having learned a new
word.
In practice it is rare for recourse to be had to a dictionary.
And yet, according to Leech, knowing the meaning of a word is a prerequisite for its successful use in communication.
Conversely, it is by no means the case that being a native speaker of a
language guarantees the ability to understand utterances in it.
Consider
the following;
When dummy went down, I realised that 6S would make if suits

broke well and the diamond honours \-Jere favourably placed.
I therefore decided to assume that twelve tricks were not
readily availo.ble, hoping thereby to beat all the pairs going
down in slam contracts and to make more tricks than those
who had lingered in game. I won the opening heart lead in
Love
126
dummy, dreVl trumps in two rounds and cashed the ace of clubs.
I then ruffed a heart in the closed hand, paving the way for
an elimination, and cashed the king of clubs, discarding:1
diamond from dummy. W11en the ten of clubs appeared from
East, I did not need to look any further for twelve tricks;
I ran the jack of clubs, throv.Ting another diamond from dummy,
and subsequently discarded a third diamond on the established
nine of clubs. I was then able to ruff two diamonds in dummy,
thereby collecting twelve tricks bv way of five spade tricks
and two ruffs, one heart, one diamond and three clubs (Markus
1982)
One lmaglnes that many speakers of English will fail to make much of this.
And yet the semantic theorist would be hard pressed to argue that this lS
because it is not English.
It is clearly English, and yet speakers of
English may have difficulty In understanding it.

the difficulty?
bulary.
I t is not,
tfuat is the source of
for the most part, the obscurity of the voca-
Apart from ruff, dummy, slam contract and a few others, feH of
the words in the passage are in themselves obscure to anyone with at

least a slight acquaintance with the practice of card-playing.
And, apart
from the occasional oddity (Hhen dummy went dm.;rn, those who had lingered
In game, where common nouns are used without an article, as though they
were proper names), the syntax is quite straightforward too.
Nor does the
problem arise from the fact that the words are used metaphorically or in
some other non-literal way:
what we have here lS a perfectly sober and
prosaically literal account of a game of bridge.
The conclusion must be
that those who find it incomprehensible do so because of their lack of

familiarity, not with the English language, but Hith bridge.
This conclusion has an important bearing on the distinction between
'knowledge of language' and 'knO\vledge of the world' on which, we are
told, a viable practice of semantics depends.
Lack of the relevant 'real-
world' knowledge may hinder understanding of the linguistic expressions

used to describe that piece of the world.
But the theorist's position
must presumably be that although we understand the language of the passage

as such, because we are native speakers of the language of which it is a
sample, ,,,here He fall down is in our grasp of the situation being described.
But this seems to be a distinction without a difference.
This point lS confirmed if we look more closely at Leech's illustration
of the distinction between knowledge of language and knowledge of the
Love
world.
127
My uncle always sleeps standing on one toe is said to be factual-
ly, rather than linguistically (or contingently, rather than necessarily)

absurd.
It just happens to be the case that human beings cannot sleep in
that posture.
But if the world had been ordered differently they might
have been able to, and if they were, there \.J'Ould be nothing absurd about
such a sentence.
Whereas no conceivable reordering of the world could
eliminate the absurdity of my uncle always sleeps a\vake, for if my uncle

lS asleep then, by definition, he is not aHake, and if he is mvake, then
he lS not asleep.
But there is no very obvious reason why these explana-
tions of the two kinds of absurdity should not be transposed.
Instead of
saying that it is the nature of standing on one toe that makes talk of
sleeping in that position absurd, why should we not locate the absurdity
In the conjunction of the meanings of the expressions sleeps and standing
on one toe?
Conversely, it might be argued that what makes sleeping
awake inconceivable is not the incompatibility of the meanings of sleep

and awake, but the incompatibility of the states of consciousness referred
to by those words.
Our understanding of sleep and wakefulness lS such
that we have no (non-absurd) use for sentences about creatures sleeping

awake.
But it might have been the case that there were animals which
exhibit a state of consciousness that bears some of the characteristics

of both.
If so, talk of them sleeping awake might make perfectly good
sense. The non-existence of such creatures, if it is a fact, is as much

a fact about 'the world' as is the non-existence of uncles who sleep
standing on one toe.
To put Leech's point another way, it follows from the conceptual meanlng
of sleep that to say of someone that he sleeps awake is contradictory.
But to put the point In this way is to raise the question of how the
'conceptual meaning' of a word is determined.
One possible answer is to
say that the conceptual meaning of, for instance, sleep, is determined
by such judgements on the part of native speakers of English as that to
talk of sleeping awake is not just absurd but contradictory.
But (granted
that native speakers would make such a judgement) how do they come to do
so?
The only answer implied by theorists' like Leech is:
because they
have assimilated the form-meaning connection in the case of the word

sleep.
How can we break out of this circle?
Love
It
1S
128
relevant at this point to observe that semanticists' statements
as to the conceptual meanings of linguistic expressions are not for the

most part arrived at in [Hacticc lly empiric;)i
illVl'stih;lLlUI1,
they particularly well supported by observation.
l10r
;lrc
Leech's book is full
of confident assertions as to the meanings and relations between the

meanings of English words and sentences.
\fuat, for example, is the
basis for the claim that the semantic features Ivhich make up the conceptual meaning of woman are [+HUNAN, +ADULT, +FEru\LE]?
It is, of course,
clear that this judgement does not arUE 1n any very obvious wa.y from
mere consideration of how the word 1S 1n fact used.
'My father is a bit
of an old woman' is not, in the appropriate circumstances, an anomalous

utterance, nor does it imply a sex-change operaLion.
In the conversation
Found anything?
Yes, can't say more.
Just one thing, Chief - man or woman?
Woman.
(Williams 1967:320)
the word woman

year-old
child.
refers
to
the recently
exhumed
corpse of a
'The "lOman undenvent oophorectomy last
ten-
year'
seems an unexceptionably literal way of expressing oneself, and yet there

is room for wondering whether it should not in fatt be held to be a contradiction, given that woman includes the feature specification C+FE}~EJ
And so on.
The point here is not that cases of this sort cannot be
explained away with reference to deviations of aile kind or another [rom

the literal meaning [+HUMAN, +ADULT, +FEMALEJ.
The point is that it is
not immediately clear where the idea comes from in the ficst place that
cases where the meaning of the word "vaman can be unequivocally accounted
for Hith reference to the bundle of features [+HLl'1AN, +ADULT, +FEMALEJ
are the norm in terms of which others are to be explained as deviations.
One likely source is the dictionary.
And, sure enough, the Oxford Eng-
lish Dictionary (OED) gives as the first meaning of woman 'adult female
human being'.
How do the compilers of the OeD come by that definition?
By a procedure identical, in one respect,
to that used by anyone who
had not previously encoulltered the "vord out,vith \.]ho nonetheless understood
the quoted utterance containing it.
Namely, by deducing from the context
the most plausible interpretation to be assigned to it.
Rut there is
Love
another respect in which the procedures are different.
129
For the diction-
ary-compi ler deduces his defini tion of a given 1.Jord from a very large
number of examples.
meaning of
The conclusion that the first, primary or literal
is 'adult female human being' 1S a sort of summary of
what is common to the majority of instances of its use in the data on

which the dictionary is based.
A dictionary definition is thus a retro-
spective analysis of whac speakers have meant by a given word in the

past.
As such it has a number of uses.
It may, for instance, be an
important p1ece of information for someone learning the language as a

foreigner.
But rarely does it play any role in the use of a word, on
a particular occasion, by a native speaker.
For that purpose a diction-
ary definition is usually neither necessary or sufficient.
I cannot
recall ever having looked up the definition of definition in a dictionary,

and indeed have only the haziest idea of what its definition might be.
But this is no hindrance to my use of the word in, for example, the last
sentence but one, or to my understanding of it when used by others.
Conversely, no English dictionary known to me offers the information
that there may be occasions when it is appropriate to refer to the body
of a dead child as a 'woman'.
If the dictionary definition of a word normally plays no role in its

use by native speakers, it follows that it likewise plays no role in a
theoretical explanation of what is involved in the use of a word by a
native speaker.
But Leech's theory has it that prior knowledge of the
'conceptual meaning' (that is, the dictionary definition) of words 1S

what makes communication by use of language possible.
In the course of
discussing the difference between conceptual and connotative meaning,

Leech says that "It can be assW1Ied, as a principle witrrOut which comJ?nA.-
nication through that language would not be possible that on the whole
[speakers of a language
... " (p. 13, emphasis added).
N.L.]
share the same conceptual framework
How communication through a language 1S
possible is an empirical question which has yet to be answered.
By
merely presupposing an answer, Leechian semantics merely makes it more

difficult to see that it is a question that even needs to be asked.
However, the Leechian answer is one which the theorist finds it hard in
practice to uphold consistently.
Communication between speakers of a
Love
130
language supposedly depends on their being masters of onE' and the same
set of correspondences between forms and meanings.
Now this may provide
an acceptable account of the working of certain very simple semiotic

systems.
For example, traffic lights.
It is quite plausible to suggest
that what makes communication possible bet,veen the operator of a set of

traffic lights and the road-using recipients of his messages is shared
knowledge of the form-meaning correspondences associated with the various
permissible combinations of lights.
But this is because the set of
messages communicable by the system 1S finite and fixed.
One cannot use
the system to send messages other than those laid down 1n advance bv the
Highway Code.
The combination of red and green lights together, for
example, is simply uninterpretable.
In contrast
the semantic system [Of a natural language

N.L.l 1S
continually being extended and revised.
In a language like
English, new concepts are introduced in large numbers day
by day and week by week, and in very little time (owing to
modern mass connnunications) becomes familiar to many people.
These new concepts are eventually not felt to be strange,
but are fully assimilated into the language and so become
part of standard mental equipment (p. 30).
The trouble with Leech's explanation of linguistic communication 1S that

it fails to offer any account of how this process can come about.
If a
speaker's understanding of a new word depends on his already knowing its

meaning, it is hard to see how it is ever possible to communicate by
uSlng a new word, or an old word with a new meaning.
What is lacking here is due recognition of the role of context 1n linguistic communication.
context.
Not that Leech has nothing at all to say about
He is aware that the meaning of an expression is somehow con-
nected with the context in which it is used, and illustrates this point
with reference to the different meanings of the phrase
p~t
on, 1n
shall I put X on?, depending on whether X is, for example, the portable
radio, the sweater, or the lump of '(mod (p. 67).
One might suppose
that the conclusion must be that the meaning of put on (granted that it
makes sense to talk of a unitary expression put on, with a significance
that varies according to context, rather than of (at least) three distinct expressions with different meanings, that happen to share [he same
Love
131
form) is a summary of what lS common to its meanlng ln the different

situations in which it occurs.
Leech rej ects:
But just this is the conclusion that
"Instead of seeing total meanlng as an aggregate derived
from context, we see the contextual meanings as dependent on a previously established set of potential meanings" (p. 68).
In other words, a
meaning for put on lS somehow first established in the abstract, and

its meaning ln a particular context (e.g. in some given utterance of
put the sweater on ) is derived in the light of this abstract meaning.
But, outside dictionaries and other works of linguistics, the expresSlons of a language simply never occur 'in the abstract'.
Language
never manifests itself but in a context.

The notion that there are abstract, decontextualised expresslons, and
that it makes sense to talk about their meanings, lS of course crucial
to this approach to semantics.
It is implicit in the idea that the
semanticist "is interested in studying the communication system itself,

rather than what use or misuse lS made of it" (p. 2).
However, the
contrary conviction that there is no 'communication system' as such over

and above the use or misuse that is made of it is reinforced by closer
consideration of some of the statements Leech makes about English.
Part of a speaker's semantic knowledge is held to be knowledge of when
the relation called 'paraphrase' holds between different sentences.
Paraphrase, we are told, is "roughly, 'sameness of meaning'" (p. 4).
Leech gives, as an example of a pair of English sentences so related,
the defects of the plan were obvious and the demerits of the scheme
were evident.
Now the first point that calls for comment here is that defect does not
necessarily mean 'demerit'.
One cannot readily substitute demerit for
defect in a sentence like the defects of his teeth were obvious.

versely, a demerit is not ipso facto a defect.
Con-
A scheme for disposing
of stray dogs by gassing them may have its demerits, but it is not on
that account necessarily defective.
successful.)
(On the contrary, it may be highly
Similarly, a plan is not always a scheme.
One cannot
paraphrase the defects of the plan were obvious by the demerits of the
scheme were evident where the plan in question happens to be a plan of
the sort otherwise referred to as a large-scale street map.
Comparable
Love
132
points apply to other statements as to the meaning-relations holding

between certain sentences that are scattered through Leech's book.
Doctors can read, we .1re informed (p. 261), is synonymolls \vi th cioctors
are literate.
utteranc~
But not in the case where the
'doctors can
read' is followed by 'the others must stand in the corner with their
hands on their heads'.
They chucked a stone at the cops and then did a
bunk with the loot is supposedly
synon)~ous
with after casting a stone
at the police, they absconded with the money (p. 15).
But not if the
'loot' in question consists of jewellery and share certificates.
In all
such cases, the meaning-relation held to be exemplified only holds provided certain assumptions are made about the context of utterance.
What
saying that the demerits of the scheme were evident is a paraphrase of

the defects of the plan were obvious boils down to is merely that it is
possible to envisage circumstances in which both might be uttered with
reference to the same situation, and with roughly the same communicational
effect.
But if this is all that being a paraphrase amounts to, we could
equally well count as paraphrases of the defects of the plan were obvious
such forms of words as it was clear why it couldn't be done or the proposal would, without question, have led to a complete cock-up or nobody
thought the idea had much chance of working or when he'd finished outlining it, they all said "don't be daft".
But if these are not para-
phrases of the defects of the plan were obvious, Leech has failed to
tell us why not.
And if we are In doubt hOI" to tell \"hether X and Yare
paraphrases, it is quite unclear \oJhat it means to insist that knowledge

of when the paraphrase-relation holds
lS
part of the native speaker's
knowledge of the semantics of his language.

One possible unstated criterion
of X
IS
that the words of a would-be paraphrase
should match those of X fairly closely.
But hOI, closely?
Demerit
may, sometimes, for some speakers, serve as a rough equivalent of defect.

At other times, or for other speakers, it will not.
lS
~lat
emerges here
that paraphrase, pace Leech, is not a relation that holds bet\oJeen
forms of words considered in the abstract, irrespective of particular

speakers and particular occasions of speech.
This is a point that Leech would perhaps claim to have circumvented by
stressing the importance of his semantic testing procedures.
Ultimately,
Love
133
he would say that whether or not the demerits of the scheme were evident
lS a paraphrase of the defects of the plan were obvious is a matter to
be decided by eliciting the judgements of native informants.
look again at the example of such a test already mentioned.
Let us
Ninety-six
per cent of respondents considered that, if someone killed the Madrid

chief of police last night is true, then the Madrid chief of police died
last night must also be true.
Thus, so Leech would argue, the entailment
relation between these two sentences has been established as a matter of

objective fact.
Furthermore, the informants' readiness to provide an
unequivocal answer to the question indicates that it does make sense to

treat decontextualised sentences as having meaning in the abstract.
Let us consider these claims.
First of all, it must be pointed out that
ninety-six per cent of Leech's informants were simply wrong, at least on

one interpretation of the semantic relation between kill and die.
They
overlooked the possibility that the police chief was killed by the administration of a slow-acting poison, ln which case he might not have died
on the night he was killed.
Secondly, it is simply not true that the
informants in such an exercise are being presented with decontextualised

language, for there is no such thing as decontextualised language.
The
context here is the lanugage-game of being called upon to answer questions about the meaning-relations betwee particular forms of words.
And
like any other language-game, this one has its own particular rules.
One of the unstated rules here is that one must take the two sentences
as referring to one and the same event.
For, of course, it is nonsense
to suppose that any sentence, as such, entails any other.
Everything
depends on a tacit contextualisation whereby they are taken as having a
particular communicational force.

A slow-acting pOlson may cause death some time after its ingestion.
But
there is room for disagreement among speakers of English as to whether

it is legitimate to count the victim of such a poison as having been
killed at the moment of administration, or at some later time
instance, the time of death.
for
The possibility of disagreement on such a
matter points to a further problem with Leechian semantics.
We are
invited to believe in the existence of a monolithic, determinate structure of correspondences between forms and meanings, called 'English',
Love
134
which, because it 1S the corrunon possess1on of 'speakers of English',

allows them to corrrrnunicate with one another.
And yet, on consideration,
it seems unlikely that all those people ,ve would want to count as
speakers of English have exactly the same knowledge of exactly the same
correspondences between forms and meanings.
escapes Leech's attention entirely.
This is not a point that
For instance, he observes that
"it is too simple to suggest that all speakers of a particular language

speak exactly 'the same language'" (p. 13).
ElseHhere (pp. 111, 113)
when discussing 'taxonomic oppositions' of meaning betHeen Hords, he says

that
"it must be emphasised here, as before, that the semantic

oppositions and their interrelations need not reflect categories of scientific thought: He are concerned with the
'folk taxonomy' or everyday classification of things that
is re~lected in the ordinary use of language. There are
many instances where folk taxonomies involve what a scientist would consider a misclassification: in the past
history of English, for example, the word Horm has been
applied to both worms and snakes, and the word fish has
been applied to whales. One of the difficulties-of trying
to arrive at a f01k taxonomy for present-day English is
the interference to varying degrees of technical taxonomies
(e.g. the biological taxonomy in terms of classes, orders,
genera, speC1es, etc.) in the ordinary non-specialist use
of language . .. "
But pointing out that there is such a difficulty is not a substitute

for overcoming it.
And unless it can in principle be overcome, Leech's
theory is in jeopardy.
If you think a ,.;hale is a kind of fish, HhereClS
I think it is a mammal, or if Leech thinks a dog is' "an animal of the

canine species
Ir
(p_ 84), while everyone else knows, or thinks the knows,
that living dogs belong to a number of different species, then in each

case the semantic knoHledge of the t,vo parties is different.
But suc-
cessful communication, we arc told, depends on their semantic knm.;ledge

being the same.
It is a mistake to treat 'a language' as an independently identifiable
entity that somehm.; exists in the
ab~tract
cative acts in which it is manifest.

express ions, as such, 'have'.
over and above the commun1-
Heanings are not things ,.;hich
If they are things at all, they are things
Love
135
that utterances have in virtue of their use at particular times ln particular places by particular speakers.
language is not
Communication by means of
matter of making use of a prearranged set of corres-
pondences between forms and meanings, for there is no such prearrangement.

Rather, communication is a continuously creative form of social interaction in which the precise role of language will vary according to
circumstarrces.
Suppose I want to offer you a cigarette.
One way of doing this is to
turn to face you, holding out a packet of cigarettes, and to utter, with
an approprate intonation pattern, the words 'would you like a cigarette,
X?', where X is your name.
Another and no less effective way would be
to make exactly the same bodily gestures, but to omit the verbal utterance entirely.
Or, provided the intonation was correct, I could
probably get away with accompanying the bodily behaviour with a piece of
verbal gibberish, without detriment to my communicational success.
But
now suppose that, although within earshot of one another, we are ln a

crowded room, engaged in conversation with different third parties, and,
without looking in your direction, producing a packet of cigarettes, or
ln any other non-verbal way indicating my intention, I say 'would you
like a cigarette, X?'.
There is a strong possibility that your response
would be 'what was that?' or 'were you talking to me?'.
That is, the
case where, for getting my message across, I rely exclusively on the

putative shared code of correspondences between forms and meanings is
the very one where we are most likely to encounter a communicational
hitch.
The moral of the story is that language itself may be neither
necessary or sufficient for communication, and that any attempt to

explicate linguistic meaning that fails to take account of the complex
ways in which, in acts of communication, language is integrated with
other aspects of behaviour, is likely to be inadequate.
observes (p. xi):
As Leech truly
"theoretical semantics can easily lose contact with
practical problems of communication, and so can suffer from a somewhat

distorted, etiolated view of the subject it is meant to be studying".
Love
136
NOTES
1.
E.g. the space allocated to a discussion of criticisms of componential analysis in the first (1974) edition (pp. 122-4) is expanded
by three hundred per cent in the second (pp. 117-122), but six
pages is not, relative to the length of the book, mucll of an advance
on two.
2.
Whether this way of putting it corresponds to how knowledge of this

sort might be represented in e.g. a generative grammar of English
is not at issue here.
3.
Of course, non-native speakers, if they are at all proficient, know

these things too.
4.
It is unclear how much importance Leech attaches to the distinction
between (a) sentences and utterances and (b) sentences and propositions.
For example, 'my uncle
~lways
sleeps standing on one toe' is
introduced as an utterance (p. 6), but is then almost immediately

referred to as a sentence which for an English-speaker would be
"unbelievable because of what he knm,s about the "orld we live in".
Similarly, in discussing tests of speakers' knowledge of logical
relations between propositions, he says:
to ask them 'if sentence
(p. 81).
"il might be worth "hile
is true, does sentence Y have to be true?'"
Leech's failure to respecl these distinctions is an impor-
tant source of the difficulties encountered by his semantic theory;

and since he apparently thinks that in practice there is no need to
invoke them systematically, in what follm.'s it is assumed, for e:.;pository convenience, lhat the argument wilL be unaffected if various
English forms of words, cited by Leech as either sentences or utterances or propositions, are referred to hereafter as 'sentences'
throughout.
Love
137
REFERENCES
Bloomf
1935
L.
Language.
London:
Allen & Unwin.
Evans, G. and J. McDowel

1976
Truth and
meanin~:
essa.ls l.n semantics.
Oxford:
Clarendon Press.
Leech, G.
1981
Semantics:
the study of mean1.ng.
Harmondsworth:
Second Edition.
Penguin Books.
Markus, R.
1982
'Bridge'.
The Guardian Weekl.l, 5 December 1982, p. 23.
Sampson, G.
1980
Making sense.
Oxford:
Oxford University Press.
Williams, E.
1967
Beyond belief.
London:
Hamish Hamilton.

Semantics

Uploaded by

Copyright:

Available Formats

Semantics

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Semantics

Uploaded by

Copyright:

Available Formats

Stellenbosch Papers in Linguistics, Vol. 11, 1983, 115-136 doi: 10.

a discussion of the second edition

Semantics: the Study of Meaning

It LS widely accepted that a description of a natural language should

What will be referred to here as 'translational semantics' is

questionable whether (i) translational semantics is feasible, and if

For many theorists, the chosen metalanguage is one

which must provide, inter alia, an exhaustively explicit enumeration

[or if they are

meanings could scarcely be other than intrinsically unstable, given the

"we all spend

Stellenbosch Papers in Linguistics, Vol. 11, 1983, 115-136 doi: 10.5774/11-0-106

us (and in the writing we read) while try

to conform our own usage

order to be understood" (Sampson 1980:48).

If this is so, the translationalist's demand for exhaustively explicit

given word" (Sampson 1980:48).

cal question of the kind which would have to be

glven some answer

(1980:46) illustrates his point

N.L.] semantic description

the question whether possession of a handle is a criter

feature for the appl

tion of the word cup to an object".

tion may be rephrased in various ways, depending on the descriptive

feature-specification for cup include the feature

ish contain a rule "from X is a

'is the sentence cups have handles

, in \oJhatever form, arises because

cup is one of a group of words, including mug, tumbler, beaker, vase

But is is far from clear

that such questions are answerable.

translating all of its expressions

This interpretation would circuT.vent

into another natural

on's objection, In as much as translation from one natural language

But even if, by allowing another natu-

ral language to stand as metalanguage to the language under descr

we guarantee the feasibility of translational

semantic knowl edge' .

, e.g. Evans and HcDowell (1976:x). The

translation into a meta1.anguage, the account of semantic know-

Stellenbosch Papers in Linguistics, Vol. 11, 1983, 115-136 doi: 10.5774/11-0-106

Take, by way of illustration, a monoglot speaker of Afrikaans

who has before him Harrap's French-English, English-French dictionary

to the traGslati0nal theory of meaning the fact that he has available

And it 1S certainly undeniable that

One migbt therefore have expected a new, revised edition

of a popular textbook on the subject to have devoted considerable space

The second edition of Geoffrey Leech's

Semantics (Leech 1981) is, in this respect, unsatiSfactory; 1) and the

What mean1ng 1S (not)

The 'total theory of

Leech has in mind runs roughly as follows.

language works' that

Language is envisaged prima-

Stellenbosch Papers in Linguistics, Vol. 11, 1983, 115-136 doi: 10.5774/11-0-106

themselves the object of linguistic study.

ThGt object is, rather, the

(unconscious) knowledge that permits these activities to tJ.ke place:'.