III.AMBIGUITY.

B Y RICHARD ROBINSON.

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

THEBB is a certain early stage of human beliefs and assumptions

about ambiguity which may conveniently be called the naive
stage. A man who is in this stage habitually assumes that the
same word means the same thing every time it is used. He is
therefore not in the habit of Hintingnintiing various meanings of
the words he uses. He will say that a thing is ' good in one way
but not in another ' ; but he will not say that it is ' good in one
sense of good but not in another'. He willrijgtnngniBhkinds of
goodness, but not senses of the word good. He will be aware of
a few gross ambiguities, such as bank = seat and bank = the
turning of an airplane, and he will be capable of using them to
make puns; but they will appear to him as isolated freaks or
accidents, of no serious importance either in the search for truth
or in the effort to convey ideas. He has no fear of being deceived
by such a phenomenon, and uses all his words with unquestioning
confidence that each one of them means the same thing every
time it is used.
The early and middle dialogues of Plato provide an illustration
of these naive assumptions about ambiguity. The early dialogues
frequently represent Socrates as seeking for definitions of terms.
Now, before we seek to define a term we should make sure that
it has only one sense, or at least which of its senses we are trying
to define. But Socrates never does this in the Platonic dialogues.
In every case he puts the question and proceeds to look for an
answer with the most perfect confidence that the word means
the same tiling every time it is used. ' Since you call all these
things by one name, and say that although they are opposite to
one another each of them is figure, what is this " figure ", as you
call it, that belongs as much to the curved as to the straight ?'
(Meno, 7to). In rejecting proposed definitions his procedure is
often to produce a case of the definiendum that fails to come
under the definiens, or contrariwise. Such a refutation is perfectly satisfactory to him because be does not realise that, if a
term has two senses, a case of one of its senses naturally would
not fall under a definition of the other sense.

EICHABD ROBINSON : AMBIGUITY.

141

1 0

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

Only once in the early dialogues is there anything near a suggestion- that the definiendum is ambiguous. The idea that
perhaps virtue is not the same in all virtuous persons is made to
come from the somewhat stupid Meno; Socrates takes only a
page to refute this, and is evidently perfectly confident of its
untenability {Meno, 72-73). The essence of the Socratio search
for definitions is precisely the insistence that the word mutt mean
the same in all its uses, however various they at first Bight appear.
The theory of Ideas, as set forth in Plato's middle dialogues,
rests on the same naive assumption about language. ' We are
accustomed to assume that there is some one Idea related to
each collection of things to which we give the same name'
(Rp. 596A). TO each name there corresponds an Ideaand only
one Idea, however various the uses of the name may appear to be.
Not that these dialogues are perfectly unconscious of ambiguity.
In the Euthydemut Plato first depicts and then exposes at considerable length Borne fallacious arguments resting on the- ambiguity of the word fjuwBavtw. But he thereupon dismisses all
such matters as puerile. ' Even if a man knew many such
things, or all there are, he would not thereby know anything
more about reality; he would merely be able to play jokes on
men by means of the difference of words, tripping them up and
overturning them, like those who pull the chairs away from
persons about to sit down and rejoice to see them fall over backwards ' (Eutht. 278B). This profound contempt for the jugglers
of language, and this serene confidence that their art can have
nothing to do with the serious business of science, are dominant
in Plato's early dialogues; and they are characteristic of the
naive stage of human beliefs about ambiguity.
That these naive beliefs are false is easy to realise in this age
of many-volume dictionaries. The Oxford English Dictionary
assigns more than one sense to practically every word in the
language. Ambiguity is not a rare freak ; it permeates language
as much as water permeates an organism. Let us review some of
its forms.
First we may notice those ' singular' terms which ambiguously denote more than one particular or HingnUr One class
of these includes ' this ', ' now ', ' he '. Such words have a
different meaning practically every time they are used ; but they
are usually intelligible because there is a rule by which we can
tell in any case which particular is being denoted then. Another
and a stranger class of singular terms are the proper names,
which appear to denote different particulars ambiguously without
any rule to determine which is being denoted at any given time.

142

BICHABO ROBINSON:

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

The hearer just has to divine from the whole situation which of
several persons is now being meant by the word ' John '.
Passing from singular to general words, we may note first that
kind of ambiguity which Locke describes neatly, though he
perhaps regards it as explaining more of our speech than it does.
A general word to Locke is a man's arbitrary name for an arbitrary collection of simple ideas, and men vary slightly as to what
they put into the collection. The set of simple ideas which one
man collects under the word ' animal' is not quite the same as
his neighbour's ; and hence arises ambiguity between the two of
them. To Locke, perhaps, no word is ambiguous to a single
person, but every general word is ambiguous in the mouths of
more than one person.
Then comes what may be called ' sliding ambiguity', the
ambiguity of the term that covers a wide area and refers now to
a larger and now to a smaller part of it. Such a term embraces
a big complex of conceptions, put together under one word
because we feel them to be somehow connected, or because we
have not clearly distinguished them ; and it has varying uses in
which varying parts and selections of the complex are intended
and omitted. Our common moral terms are of this order.
' Virtue ' slides from monogamy to all moral desiderata ; and the
Greek ' arete' slides on to all the desiderata of character and
intellect and body. The word ' calon ' covers a complex including
most of what can be covered by our words beautiful and noble
and honourable. The account in the Symposium of the lover's
progress towards the absolutely ' calon' gives representative
samples of its field. In the Oorgias Socrates' more formal proofs
of his important positive theses are all based on the ambiguity
of this word or the corresponding ambiguity of its opposite
' aeschron '. For example, his proof that doing injustice is worse
(than suffering it) consists in inferring it from the three premises,
first, that doing injustice is more ' aeschron ' ; second, that the
more ' aeschron' is either the more painful or the worse or both ;
and, third, that doing injustice is not more painful. The first
premise, that doing injustice is more ' aeschron ', takes a part of
' aeschron ' roughly corresponding to our word shameful; but
the second, that the more ' aeschron ' is either the more painful
or the worse or both, consists in explicitly taking other parts of
the complex that can be covered by the word. The words
' good ' and ' agathon' are ambiguous to much the same degree
and in much the same way as ' calon'. They are always deceiving us, especially by sliding from the object of desire to the
morally good or contrariwise. The Qorgiaa (506D) provides a very

AMBIGUITY.

143

neat example of this particular shift in a short space. Having

been speaking of the ' agathon' in the sense of the object of
desire, Socrates savs : ' and the agathon is that by whose presence
we are agathoi ? '
Our last and most important WnH of ambiguity is ' relational
univocity'. The phrase ' relational univocity ' is used here to
summarise the following doctrine of Aristotle's :

There is no genus identical in healthy exercise and healthy

complexion and healthy roses. The word is used to indicate
now this and now that relation to health. Exercise causes
health ; complexion indicates health; roses possess health. What
particular relation to health is meant in each particular case
has to be learned from the context. By the ' context' is here
meant both the rest of the utterance, and the circumstances
in which it is uttered, and all the knowledge possessed by the
speaker and the hearer so far as it is relevant. The alcohol in
a thermometer isfixedin position at bottom and sides, but moves
upwards and downwards under the influence of temperature,
thus signifying something to us. Similarly, the word ' healthy '
is attached in our minds to some one notion of health, but moves
in other dimensions (the various- relations of health) as influenced
thereto by each context, thus signifying on each occasion some
relation of health to us. ' Relational univocity ' is a bond that
holds together the various meanings of an ambiguous word ; and
its mechanism is the varying force of the context and the fixed
force of the term in relation, together acting on the word as the
moving force of the pistons and the fixed force of the bearings
together act upon a crankshaft. It is immensely common;
practically every descriptive word has distinct meanings so
related.
This review of a few common kinds of ambiguity is sufficient
to show the falsity of the naive assumption that most words have
only one meaning. At any rate, when words are considered
singly and separately, and not in some sentence or other, the
truth is that ambiguity is the rule and not the exception. It is
therefore wrong to assume a single Form or universal corresponding to each word, and wrong to seek a definition of-a word without

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

The word ' being ' has many senses, bat they are all related to one
thing, to some one nature. The word is not ambiguous, but is like
the word ' healthy '. Everything ' healthy ' is related to health,
BOHTfttini|1 u uimoi ving hnaltihi sometimes as causing it, sometimes
as indicating it, sometimes as being capable of exemplifying it.
Metapk. T. 2, 1003a33.

144

BICHABD ROBINSON :

Since we cannot introduce the realities themselves into our discussions, bat have to use words as symbols for them, we suppose that
what follows in the words will follow in the realities too, like people
reckoning with counter*. But it is not the same. For whereas
words and the quantity of sentenoes are Limited, realities are unlimited in number. It is therefore inevitable for the same sentence
and the one word to mean more than one thing. As, therefore, those
who are not good at using oounters get deceived by tbose who understand them, just so in the case of sentences those without experience
of the power of words reason fallaciously both when speaking themselves and when listening to others. (BE 1, 165a6 ff.)

Aristotle does not explain, here or elsewhere, so far as I know,

why the quantity of words and sentences is limited. The attempt
co do so would have led him to alter this account considerably.
The obvious question is why we do not make more words to fill
the gaps. The only answers that present themselves are, first,
that neologisms are inelegant, which seems thoroughly inadequate,
and, second, that oar memory cannot retain so many words.
But is our memory then so constructed that it can retain far
more tilings than words ? How does it come about that we have
in mind more things to be symbolised than we have words to
symbolise them ? Aristotle seems to be AMnming that we first
apprehend a reality and afterwards look for a word to symbolise
it. But surely our knowledge of insensible tilings very rarely
comes about in that way. Surely the process of apprehension,
in so far as it is not bodily sensation, usually occurs in and through
the presence to our imagination of appropriate words, not prior
thereto. It is only in getting the right words that we come to

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

mulring clear, h" the context or otherwise, which of its sensea is

to be defined.
The question arises why there is all this ambiguity in language.
Gould we deduce the necessity of ambiguity from die nature of
reality in general, or of oar t.ninlring in particular, or of symbols ?
This question no doubt requires criticising ; it may assume some
falsehood. Perhaps it is impossible to describe first the manner
of our apprehension of reality and thereafter the manner of our
symbolisation of it, because our apprehension is conditioned by
our symboUsation and exists only in and through it, so that
language is not subsequent to thought in the way the question
implies. Assuming, however, that the question is legitimate,
let us ask what suggestions can be made towards an answer.
Aristotle explains the prevalence of ambiguity by saying that,
whereas there is no limit to the things we may want to refer to,
there is a limit to the symbols at our disposal for referring to
them:

AMBIGUITY.

145

f 0 .

The Tyranny of Word* (New York, 1038), p. 169.

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

know the fact. It is true that our memory for words is limited ;
but our memory for abstract things is equally limited, because
only by means of words can we think of abstract things at all.
The reasons for the ambiguity of language lie elsewhere than
this passage of Aristotle's suggests. Some of them are obvious
and others are hidden. An obvious one is that people make
shifts in the meaning of a word through ignorance. They begin
to use a word which they have heard or seen, without consulting
the dictionary ; and the sense which they have divined from the
contexts turns out to be not quite what the authors had in mind.
Another obvious cause of ambiguity is that people make shifts
in the meaning of a word for a purpose, though not for an
analysed and deliberated purpose. They can think of no old
word for what they want to say ; but they see that in the present
context a certain word would, by the principle of relational
univocity, be forced to take on the new sense which they require.
There are also three less obvious and more important reasons
for the ubiquity of ambiguity. The first is the fact that we have
not yet completely discriminated the parts of reality. The
growth of our understanding is among other things an increase
in our discrimination, by which the big blooming buzzing confusion from which we start is gradually distinguished into its
components. Now our words must reflect the extent to which
we have carried this process, the extent to which we have detected
differences in things. They discriminate what we have discriminated, and confuse what we are still confusing. But, since
we shall never have done clearing up confusions, our words
will always be concealing differences, and there lies an eternal
possibility of ambiguity. Mr. Stuart Chase will therefore be disappointed in his hope that there will one day be a semantic
technique ' which will let us take a political speech . . . and tell
specifically what is wrong with it V because what is wrong with
any language always depends in part on what the facts are, and
we shall never have done ascertaining what the facts are. The
very same sentence in the same context may be ambiguous if
reality is this way but univocal if reality is that way. For
example, whether ' 257c' is an ambiguous reference to Plato's
writings depends on whether Stephanus, in bis edition of Plato,
recommenced with page one at the beginning of each volume.
It follows, as Alfred Sidgwick well shows, that every question,
however much it seems now to be a question of fact, may turn
out some day to be a question of definition, since definition or

146

KICHAED B0B1N80N I

TktVMoJ

Words in Seasoning (London, 1001), p. 228.

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

distinction is the cure for an inconvenient ambiguity. To the

man of more discrimination it will seem a question of words;
but when he says so to the man of less discrimination he will be
accused of refusing to answer a plain question. It is a great step
forward in the skilful use of language to have realised that no
question whatever can be guaranteed to be ' plain '.
The second important reason why ambiguity pervades all
language is that all language involves the use of descriptive words,
and, as Alfred Sidgwick says, 'it lies in the nature of language
that descriptive words shall, as such, be liable to cause ambiguity '.1 Let us examine the two premises of this argument in
turn. That all language involves the use of descriptive words is
equivalent to the fact that no sentence can consist purely of
proper names and demonstrative symbols like ' him'. At the
least there must be one descriptive word, as in ' John hit James '.
A descriptive word, which may also be called a connotative word,
can be denned, and can apply to more than one situation. With
these explanations probably no one will deny that' all language
involves the use of descriptive words'. The second premise is
that ' descriptive words as such are liable to cause ambiguity'.
Sidgwick justifies this as follows. A descriptive word is a classname. It collects things into groups where each member is
supposed to have the property connoted by the word. But now
every class has subclasses ; and those subclasses are ignored by
the name of the class. Therefore the name is indefinite; it
omits specifications which are always present in reality. It is
equally applicable to adjacent cases which nevertheless are
different. This indefiniteness, this ignoring of real differences, is
then an inevitable character of all language ; and it is a permanent possibility of ambiguity in a bad sense, since we are never
certain that no harm will come of our ignoring the sab-classes.
The third important reason for the prevalence of ambiguity is
that ambiguity is so often a good thing. To start at the bottom
of the scale, the pun is evidently a form of ambiguity that is
reckoned as (within limits 1) good. More powerful and therefore
more valuable than the pun is the double entendre. A superb
example of the latter is this of . M. Forstei's : ' The only thing
worth giving away is yourself '. Here ' give away ' carries the
meaning of revealing one's shortcomings, but also the meaning
of love, the perfect gift. The compaction of these two ideas into
a simple sentence explodes a whole rnftgarinft of emotions and
reactionsaffirmation of love,'joy in love, fear of blame, fear of
mockery, deprecation of mockery, all in a single phrase. You

AMBIGUITY.

147

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

do not achieve the same high tension by saying each of these

things successively.
Besides ambiguity of statement there is ambiguity of tone, of
the manner of the statement. And a special case of this is
ambiguity about the value the speaker puts on the fact stated.
Does he approve it or disapprove it ? Socrates deprecates his
own ignorance ; but there is at the same time a suggestion that
this ignorance is somehow better than the knowledge of the other
party. Such equivocation about one's valuations, when not
successively asserted but simultaneously implied, is something
we often mean by the word irony.
We may be unable to distinguish clearly all the seven types of
ambiguity described by William Empson.; but he proves convincingly enough that ambiguity plays an enormous part in the
effectiveness of poetry. Consider the phrase : ' Zuleika was not
strictly beautiful', falling where it does in Max Beerbohm's
description of her person. There is no doubt that he is very
successfully having it both ways. By disclaiming strict beauty
for his heroine, he wins a reputation for judicial gravity and
escapes from the class of romantic novelists. And yet he has at
the same time, without our knowing it, insinuated into us a
conviction that Zuleika was extremely beautiful. The whole
description is made by negations and flippancies ; but their
effect is to drag our critical faculties and make us passionate
devotees of Zuleika.
Empson draws the valid conclusion that we often misinterpret
a poet's word if we assume that it has only one of the dictionary
senses. One learned commentator says the poet meant this and
another says he meant that; but he meant both. And the full
force of the poem is to be gained not by following one interpreter
but by following them all at once, in that synthesis of contraries which gives the poem its intensity and magic. Hence the
nothingness of the literary criticism that condemns Shakespeare
and Keats because their images conflict.
This principle should be used in interpreting not merely words
and sentences, but also images and parables andfiguresof speech.
Every man who writes a parable is so far a poet; and he does not
mean his image to have one single precise meaning, with a mathematical one-one correspondence between image and archetype.
He seizes on every opportunity for significance that his image
presents, even if it was no part of his original purpose in conceiving it. I shall argue in my book on Plato that the figure of
the Divided Line in the Republic came to have a second meaning
to him as he wrote it. The figures of the Sun and the Line and

148

RICHARD ROBINSON :

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

the Cave are poetry as well as philosophy. It would be fantastic

to expect to reduce every element in them to a mathematically
precise and logically rigid science, and still more fantastic, though
.people are always doing it, to expect the Line, if so reduced, to
correspond precisely to the Cave.
Plato is both poet and philosopher; and in bringing him
forward we have been insinuating a paradoxical proposition which
must now be made explicit, namely that ambiguity is often
advantageous in philosophy and science as well as in poetry.
One argument in favour of this view, that ambiguity may be
valuable in philosophy as well as in poetry, would be the doctrine
that all language is both emotive and informative. Helen
Wodehoose has persuasively argued, in MIND for 1938, that
Richards' distinction between the emotive and the scientific
powers of language most be taken as indicating, not two species
of language that can exist apart, but two aspects inevitably
combined in every speech, though usually one is more prominent
than the other. If this be true, as it appears to be, all scientific
writing would have its poetic aspect; and therefore might
benefit from ambiguity in all the ways in which poetry does.
What is here to be maintained, however, is the view that
science receives the greatest benefits from ambiguity, not merely
in virtue of the poetic elements inseparable from any scientific
statement, but also in its strictly scientific aspect. This begins
to appear when we ask how we are to symbolise new things, for
which there is as yet no symbol. At first the obvious answer
seems to be that we should invent a new symbol; and in this
attitude of mind Housman praises the man who invented tile
word ' gas', an entirely new root to convey an entirely new
notion, conveniently brief, conveniently univocal, and submitting moderately well to inflection. We need not deny these
advantages ; and we need not say this sort of thing should never
be done. But we must ask how the inventor of this word got
other men to understand what he meant by it. Inevitably he
used a definition. And inevitably this definition was composed
of old words, whose meanings were already frmiliar. It seems,
therefore, that all the inventor did was to enable us to express in
a short word what had previously taken a long phrase. He
abbreviated language; a service undoubtedly, but not of the
first order of originality. Perhaps, however, he not merely
invented the word, but also was the first to bring together into
one complex that specific set of ideas which the word conveyB.
That would be another service, and a greater one. But still it is
not first-class originality ; because it is only a new combination

AMBIGUITY.

149

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

of old elements, and contains no new element. Is all oni thinking

always to be a reshuffling of the same elements, like the processes
of nature according to Democritus ? A reetringing of the old
beads, with never a new one added to the collection ? Bat we
never get an idea without the words for it; and, if we did, we
could not convey it to another without words. Here then is the
difficulty ; we hope to get a perfectly new idea from time to time ;
but we can only use the old words with which to get it, or a new
one that has to be defined in terms of the old ; and the old words
only mean the old things. So we are apparently condemned
always only to rearrange the old notions. Our original axioms
and postulates and definitions have been given to us once and
for all (though Heaven knows how); and all we can do is to
go on drawing their consequences. No doubt they are tremendously powerful, and there is an inexhaustible supply of
theorems to be proved. But if we hanker for something radically
different, for something outside the syBtem altogether, some
entirely new axiom or definition, we cannot be satisfied.
Irreducible originality is possible through the fact of ambiguity.
Great minds on the one hand, and the unconscious folk on the
other, hint their new idea, and we divine it from them, by means
of an old word in a new sense. The new sense is not given by
formal definition, for that could only be a new combination of
old ideas. It is given by ' relational univocity ', that is, by the
reaction of the context with the old sense. A strange context
forces the word to take a new sense. The word makes the least
shift from its old sense that the context allows, and makes it in
the direction dictated by the context. That is how the word
' wood ' came to convey the new idea of ' matter ' in Aristotle's
mind. Do not suppose that the beginning was those quasidefinitions of ' matter' that are to be found in his works ; on the
contrary, they are crystallisations or deposits of the idea already
formed, and could not be understood by one for whom the word
' wood ' still meant only the organic tissue.
In mathematics, where we aim at great rigidity in our symbols,
we sacrifice adaptability and copiousness for the sake of precision in statement and certainty in intuition and inference. We
limit ourselves to a prescribed field for good purposes. But we
cannot always limit ourselves. We wish to be able to deal, at
some time or other, with anything that may come along; and
no definite system of symbols, denned by a set of axioms, can be
adequate to deal with any situation that may arise in the future.
The more precisely the constitution of a country is defined, the
worse it meets unforeseen situations. For dealing with new

160

BICHABD ROBINSON :

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

situations we must fall back upon the indefinite total of our symbols, which is relatively unorganised, does not clearly distinguish
axioms from theorems, and consists mostly of undefined and
ambiguous symbols.
This does not mean that first-class originality is non-existent
in mathematics, but that it occurs not in the march from axioms
to theorems within the system, but in the groping to the system
as such from and by means of the mass of ambiguity outside.
Nor does mathematics ever reach a stage whence it can proceed
without any more use of indefinite symbols. He who effectively
renounces indefinite symbols merely adds theorems to an established system. But the discoverers of non-Euclidean geometry,
and all such geniuses, did so by daring to wade again into tie
sea of ambiguity. Mathematicians sometimes say that their
science requires no words, no verbal symbols. However this
may be, their science, in its most creative moments, does require
the use of ambiguous and indefinite symbols, whether verbal or
otherwise. The exact part of mathematics will always be surrounded by an umbra of inexactitude.
There is a danger that this doctrine may encourage some of
our bad tendencies, may invite us to mistake obscurity for profundity and slipshodness for fertility. Part of our precaution
against this danger must be to remember that ambiguity is bad
as well as good, and to try to bring before our minds as fully and
distinctly as possible the difference between the two. We most
balance'the advantages of ambiguity against the undoubted need
for exact concepts, and try to define the use and abuse of each.
We must not let our hard-won habits of exactitude faint and lose
courage at the realisation that they are not the whole of virtue.
All that can be here undertaken towards this end is a brief
examination of what constitutes bad ambiguity.
There are two obvious sorts of bad ambiguity which we can
quickly note and pass on. The first occurs when the words a
person chooses convey to his hearer a meaning other than he
intends. It is possible, for example, that an American who
intended to tell an Englishman that there was on the highway
a commercial road vehicle driven by internal combustion should
choose words which would in fact give to the Englishman the
impression that there was on the sidewalk a railroad freightcar.
This is because the words ' pavement' and ' truck ' are used in
different ways by the Americans and the English. Ambiguity in
this sense, then, is what we also call misunderstanding, giving an
impression other than you intended because of a wrong opinion
about the effect of your words on your hearer. Special forms of

AMBIGUITY.

151

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

it occur when your words convey some but not all of what yon
intended, and when they convey all and more than you intended.
The second obvious sort of bad ambiguity is when my words
leave my hearer consciously doubting which of two things I
intended. I ask him whether he has seen John lately ; and he
finds himself thinking of two Johns and wondering which I mean.
Both of these obvious kinds of bad ambiguity involve more
than one person. Neither of them can arise between a man and
himself, except perhaps when he is reading something he wrote
a long time ago. In both of them the speaker may be perfectly
clear in his own mind about the facts he is symbolising. His
error may lie not at all in a misapprehension about the facts to be
symbolised, and wholly in his way of symbolising them. The
words he chooses do not have the effect he intends ; that is all.
He has made an error of expression ; for good expression includes
conveying to your chosen hearers exactly what you want to
convey and nothing else.
Let us turn to that more difficult form of bad ambiguity which
does not require more than one person, the error that a man
may lead himself into at the same time as he leads others. The
following phrase of Miss Stebbing's will give us a start. ' Ambiguity arises when (a person) is led on to extend to one referend
what is true only of another, without realizing that a transition
has been made.' Ambiguity is here regarded as connected with
a mental process that occupies time and is different in ite later
from its earlier parts. Part of the change involved in this
process is concealed from the thinker, to the detriment of his
inference. He knows that there has been a transition ; but he
does not see the whole nature of it, and if he did he would no
longer consider it valid. He thinks he is holding fast to some
identical referend throughout, whereas in truth he has only an
identical symbol which is changing its referend. We can think
only through our symbols ; and in this sort of erroneous ambiguity we fail to notice that the context of the symbol, changing
as our thought proceeds, has altered its reference so much that
the only identity between the earlier and the later part of our
cogitation is the word.
This ambiguity is an affair of inference, if inference is any
process of thinking in which we seem to follow an implication or
a necessitation. It occurs when both we suppose that the premises necessitate the conclusion, and the reason for our supposition is the identity of a word in successive parts of our thinking,
and also the supposition is mistaken because the word means
different things in the two parts.

152

RICHARD ROBINSON:

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

This account of bad ambiguity seems at first to deal satisfactorily with such philosophical arguments as we consider erroneous
owing to ambiguity. The second part of Plato's Parmemdes,
for example, consists of eight' hypotheses ', of which four consist
in affirming every conceivable predicate of a certain subject, and
the other four consist in denying every conceivable predicate
of a certain subject. The four negative hypotheses appear to
obtain their conclusion by means of one of the ambiguities of the
word is. One use of this multivocal term is to mean identity,
as when I say that Alfred Sidgwick is not Henry Sidgwick.
Another is to mean attribution, as when I say that Alfred
Sidgwick is not tall. Now the Parmenides seems to obtain its
negative conclusions, that X is neither A nor B nor G nor anything else, which are denials of attribution, by shifting from
premises which were denials of identity. The proposition that
' X is not A ' is taken in the premise in the acceptable sense that
Xness and Aness are not identical; but in the conclusion it is
taken in the unacceptable sense that X does not have the property
A. That is an example of the way in which our account of
erroneous ambiguity seems to cover these philosophical mistakes.
But, on second thoughts, even if it covers them, it surely does
not fit them. For surely this ambiguity in transition presupposes
an ambiguity without transition, an ambiguity in statement
as opposed to inference. If a man can start with the relation
of identity but end with that of attribution, and not realise
that he has made a shift, surely that must be because, at the very
time when he is thinking of identity without transition, he does
not distinguish it from attribution. He means ambiguously
both in the single statement as well as in the inference.
Let us try, therefore, to define this more fundamental kind of
bad equivocity which occurs in the statement even apart from
transition. It will not do merely to follow the dictionary and
say that it consists in a statement's having more than one meaning.
Every statement has more than one meaning, or may turn out to
have. Can we then Bay that it consists in failing to distinguish
these separate meanings, that, for example, the argument in the
negative hypotheses of the Parmenides misleads us because we
fail to distinguish non-identity from non-attribution ? No;
because, though it may be true that all bad equivocity involves
a failure to distinguish, the converse is false. The proposition
that all men are mortal is not bad for failing to distinguish
between Chinese and Japanese. On the contrary, it is essential
to every statement to overlook certain distinctions.
Narrowing the proposed definition still further, let us say that

AMBIGUITY.

163

Tkt Application of Logic (London, 1910), p. 129.

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

ambiguity is failing to Hiatingni}i the two senses although they

ore incompatible. Every statement, we shall now say, has more
than one meaning ; but this is a defect only when these meanings
conflict among themselves. The formula, however, seems still
too broad; for we have found that the inconsistency of the
mn.ningH may be intended and enjoyed, as in ' Zuleika was not
strictly beautiful'. Let us narrow it to this : ' meaning two
inconsistent assertions without seeing that they are inconsistent'.
But the Parmenidet shows that this will not do either. There the
ambiguity is that the sentence means both the non-identity of A
with B and the non-attributableness 'of B to A. But these two
are not incompatible; they could both hold at once between
a pair of conceptions. This example compels us to strike all
reference to inconsistency from our definition.
Shall we just say that bad ambiguity is ' dangerous indistinction ', the case where our inevitable failure to distinguish all the
senses brings us to grief, differing in no definable way from other
indistinctions except that it does bring us to grief 1 That seems
safe bat hardly dmtiing An unkind critic might say that it was
abandoning the quest. We were attempting to separate from the
general mass of ambiguity, much of which is good, a special sort
which is always bad ; and now we are in effect defining this bad
ambiguity as just bad ambiguity. Yet this seems to be the line
taken by Alfred Sidgwick in his brilliant discussions of ambiguity.
He says that ambiguity is ' important vagueness, and . . . vagueness consists in the absence of a needed distinction between
subclasses of X '.* In what sense can the vagueness be ' important ' except that it leads you into error ; and why was the
distinction ' needed' except to avoid an error ? The whole
trend of Sidgwick's discussion is to establish with very great
probability that you can give no definition of bad ambiguity
except just this circularity that it is the ambiguity which is bad.
The reason is this. Our ambiguous statements and our adequate
statements are not the records of distinct species of thought. It
is, or can be, the same process of human thinking in the statements that we consider excellent and in those we call fallaciously
ambiguous. In our best as well as in our worst cogitations we
are failing to discriminate some of the elements of reality. Sometimes this indiscrimination leads to a mistake, and sometimes it
does not. But the conditions under which it leads to a mistake
are not to be found in the mere indiscrimination itself ; they lie
in nothing less than our whole funded knowledge, our whole

154

BICHABD EOBINSON:

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

present purpose, our whole present situation, and the whole

nature of reality so far as it is relevant.
This is the main conclusion of the matter. The search for the
nature of the defect in ' ambiguous' arguments ends with the
discovery that it is a defect in all our thinking as such; that
what we assumed to be a special and isolated phenomenon was
really only the painful consequence of an ever-present defect.
The ignorant man, when his toothache is still, may say and believe
that his tooth is all right now ; but the informed man knows that
his tooth is aa much decayed as it was before; only, owing to
the total configuration, undesired consequences have for a time
ceased to occur. Occasionally, however, bad ambiguity includes
more particular defects, special and eradicable errors; and we
may end by examining two of these.
In the first place, when we blame a sentence for ambiguity
we occasionally mean that it implies ' an identity which is not
there ', or, possibly, ' a resemblance which is not there '. Thus
Hegel's aufgthoben suggests that there is something common to
cancellation and conservation, whereas there is nothing common
to them, or at least nothing of the sort he suggests. And
Parmenideo' juggling of the verb ' to b e ' may be taken as the
false suggestion that identity and attribution are species of some
genus narrower than the category of relation itself. A first
objection to this proposal would run thus : ' Suppose a man to
assert that sharks and dogs are identical in being cold-blooded.
Now there is no such identity. Yet we should never say on this
account that his statement was ambiguous. We should simply
call it mistaken.' This difficulty can be overcome by distinguishing explicit assertion from implication or assumption. We
shall say that ambiguity is not the explicit and self-conscious
assertion, but the unquestioned and unnoticed assumption, of a
certain identity which happens not to exist. It is the assumption
of a non-existent genus, the taking of two things as fellow species
when they are not. This definition seems to be a fairly satisfying
description of some ambiguities.
The other special form of bad ambiguity is confusion. The
confusion of differents is something other than the failure to
distinguish differents. It is not just taking two things as one in
some way in which they are not one. It may be taking a pair
of Xs not as one pab of Xs but as one X, as when I take Henry
and Alfred Sidgwick not for a pair of 8idgwicks but for one
Sidgwick.- Or it may be taking a pair of species not as one genus
but as one species; I confuse identity and attribution if I take
them, not as one pair of relations either of which might hold

AMBIGUITY.

155

Downloaded from http://mind.oxfordjournals.org/ at Monash University on June 22, 2015

between a given subject and predicate, but as one single species

of relation possessing in some mysterious way the nature both
of identity and of attribution. It is possible, apparently, to be
thinking of two distinct things as one thing, to think of a couple
as a single having the qualities of each of tile couple; and that
is confusion. The strangest thing about it is that it seems in
some sense to presuppose distinction. To confuse identity with
attribution is not just never to have heard of their difference.
On the contrary, it seems to entail that you somehow appreciate
the real character of identity and the real character of attribution.
Otherwise you are no more confusing them than when you talk
about relations between subjects and predicates in general. But
these their real specific characters are just what make them
distinct 1 So that in confusing you are also fUgt.ingnifining them.
How this can be is not clear; but it really occurs and really
receives the title of ambiguity.
We set out in this last section to seek a kind of bad ambiguity
that was found only in some language and not in all. Such are
confusion and the signification of an identity which is not there.
But our more important conclusion has been that the cases
which we commonly call ambiguous differ from other cases of
language only in that we feel that our symbols have failed. The
root cause of the failure, so far as it lies in the symbols themselves,
is equally present in them all. We can never finally determine
whether our arguments are ambiguous or not; the problem
always remains open, and only an approximative frame of mind is
rational. Whether an argument is ambiguous depends on what
the facts are; and our only way of knowing what the facts are
is by argument and words; and the best we can do with that
vicious circle is to make it a spiraL We can never make it a
straight line.