VIII. 2, Spri ng 1987 
Informal Logic 

Ellipsis: History and Prospects 

E.P . BRANDON University of The West Indies 

This paper is intended to bring the 
notion of ellipsis to the conscious at-
tention of philosophers. I expose the 
notion to the light because I believe it 
is a fruitful one, especially for informal 
logic. 

It is curious that in the thousands of 
years in which philosophers have had 
access to, and have used, the notion 
of ellipsis no one has, as far as I know, 
given it an account, explication, theory, 
criteria for recognition, or any of the 
other desiderata for terms of art in 
philosophical logic .[1] In fact those 
authors who make use of the term rare-
ly offer anything by way of explaining 
or justifying what they are doing. It 
would seem that the notion is employed 
as a piece of obvious common-sense 
that can be invoked when necessary, 
an unproblematic borrowing or adapta-
tion of a traditional notion in gramma-
tical or stylistic studies. Perhaps this 
location in another specialty may help 
to account for philosophical insou-
ciance, though in the twentieth century 
at least one would not expect a philo-
sopher to assume that he can borrow 
from traditional grammar in a logical 
discussion without even the slightest 
explanation or justification . 

I intend to sketch the broad outlines 
of the use of ellipsis both in traditional 
grammar and in recent philosophy, and 
then to turn to its role for us now. The 
survey of other people's usage will ne-
cessari Iy be somewhat lengthy, but it 
should illustrate the fecundity of the 
notion. 

The Greek etymon of 'ellipsis' (La-
tinized as detractio) was a part of the 
stock analytical vocabulary of rhetoric 
and was glossed as the omission of 
words that could be understood from 
the context. Thus Georgius Choerobos-
cus, a minor rhetorician, tells us that 
a Homeric phrase, kopton amphoterais 
(Od. 18, 28), literally 'striking with 
both', is elliptical for 'striking with 
both hands' (for this and other exam-
ples, see Lausberg, 1960, p . 346) . And 
ellipsis was often invoked by Greek and 
Latin writers to account for syntactic 
structures that appeared incomplete, 
such as subject-predicate sentences 
without a copula. 

Omissions that can be understood 
constitute a fairly mixed bunch, with 
imprecise boundaries. These difficul-
ties with the notion are clearly re-
vealed, for instance, in Quintilian's 
Institutes. He notes that other writers 
classify ellipsis with synecdoche-
part for whole or whole for part - (VIII. 
vi. 21) or with aposiopesis-the break-
ing off of a sentence in mid-stream-
(IX . iii. 60) but for him ellipsis is a more 
regular phenomenon 'where the word 
omitted may be clearly gathered from 
the context' .[2] He offers an example 
of the historic infinitive which was 
supposed to require a finite verb, and 
he also claims that obvious omissions 
of indelicate expressions can be sub-
sumed under the same concept. As 
a sub-species, he notes cases of syntac-



94 EP. Brandon 

tic yoking together such as the Cicero-
nian 'lust conquered shame, boldness 
fear, madness reason' .[3] 

Jumping two thousand years, we can 
find the same types of case and the 
same uncertainties about ellipsis in 
a recent generation of grammarians. 
Jespersen may serve as an example, 
both of a strict tradition in grammatical 
analysis and of a bridge to the concep-
tion of ellipsis used by philosophers. 

With respect to the first point, Jes-
persen is sharply critical of what he 
considered abuses of the idea of ellip-
sis to defend preconceived notions of 
grammatical structure against obvious-
ly normal sentences that fail to con-
form. Thus in a long footnote he attacks 
Sweet's account of sentences such 
as (1): 

(1) What you say is true. 

Sweet had claimed that the word 'what' 
was doing duty here for two words, 
one the object of 'say', the other the 
subject of 'is'. Jespersen objects that 
the subject of 'is true' cannot be 'what' 
(nor presumably any other word hidden 
within it)[4] but only the whole clause 
'what you say'. He rejects this 'kind of 
pseudo-grammatical analysis . .. one of 
the numerous uncalled-for fictions 
which have vitiated and complicated 
grammar without contributing to a real 
understanding of the facts of language' 
(1924, pp. 103-4). He does, however, 
concede a place for ellipsis, but only 
'where ... absolutely necessary, and 
where there can be no doubt as to what 
is understood' (p. 307). One of his 
examples of genuine ellipsis is (2) : 

(2) He is rich, but his brother is 
not / /. 

Elsewhere he allows that answers and 
retorts permit (indeed often all but re-
quire) words or phrases to stand alone 
that otherwise could not do so . Both 
these kinds of ellipsis continue to fea-
ture in popular guide books (e.g . 
Fowler, 1983) and in standard trans-
formational grammars .[5] 

Another traditional kind of example 
that deserves separate mention is pro-
vided by the semantics of some words 

or phrases, usually nominal. Thus Ull-
mann mentions the preceding cases of 
ellipsis but concentrates on examples 
such as 'a daily' meaning a daily news-
paper, or the doubly elliptical and appa-
rently aberrant 'un premiere de Lyon' 
which abbreviates 'un billet de pre-
miere c/asse de Lyon' (1962, pp. 222-3) . 
He also notes the strong tendency for 
such things to happen among special-
ized groups-lawyers will understand 
'action' as a legal action, students will 
interpret' a paper' as an examination 
paper, and soon (pp. 161-2). 

These three rough groups-respon-
ses, some compressed structures, 
abbreviated phrases - seem to exhaust 
acceptable examples of ellipsis for 
much traditional grammar, and I 
shall lump them together under the 
label 'grammatical ellipsis'. They do 
not include the forced attempts to save 
preconceptions such as Jespersen at-
tacked, and they keep close to his ad-
monition to be in no doubt about what 
has been elided. They are also in large 
part based on syntactic considera-
tions.[6]I want to group these cases to-
gether because I intend to contrast 
them with other cases that Jespersen 
also mentions and that have, I believe, 
much more in common with the philo-
sophical appropriation of the notion 
that I shall sketch below. These other 
cases are semantically inspired and 
appeal to what I hope are intuitively 
'deeper' considerations than the gram-
matical ellipses. 

The first group are something of a 
bridge between grammatical ellipsis 
and the philosophers' kind-they arise 
from semantic reflection on syntax 
of the sort typical of transformationalist 
'deep' structure, and in fact Jesper-
sen's examples seem to belong among 
Allerton's (1975) 'obligatory deletions' . 
After noting some of the cases of simple 
grammatical ellipsis allowed in dia-
logue where only one of the two notions 
joined in what he calls a 'nexus' is 
expressly mentioned, Jespersen goes 
on to apply the same analysis to 'the 
great majority of cases in which we use 
either an infinitive or a nexus-substan-
tive' where 'there is no necessity ex-



pressly to indicate who or what is the 
subject of the nexus' (1924, p. 143). 
Thus he claims that in (3) : 

(3) I like to travel. 

an 'I ' must be understood as subject 
of the verb ' to travel'. After all, I'm 
not claiming to like the fact that people 
travel ; Ilike my travelling . In (4) : 

(4) To travel is easy nowadays . 

he says that the 'generic person' is 
understood: for one to travel is easy 
nowadays. Jespersen offers some 
evidence for the unexpressed existence 
of such items, but without entering into 
these grammatical arguments, I think 
we can see that fundamentally this sort 
of ellipsis is semantically motivated . 
But in these cases there need be no 
simple way in which the elided ele-
ments can be inserted into the surface 
structure, though (5) is perhaps not 
very ungrammatical : 

(5) *llike me to travel. 

Invoking elided subjects for subject-
less infinitive verbs might seem dan-
gerously close to the evasions J esper-
sen had condemned in others, but what 
he forbids in criticism of ellipsis he 
gives back in much more generous 
measure in his concluding remarks on 
' suppression' . Leaving aside detailed 
grammatical analysis for some general 
remarks about how we convey meaning 
in speech and writing, he says: 

In all speech activity there are three 
things to be distinguished, expression, 
suppression, and impression . Expres-
sion is what the speaker gives, suppres-
sion is what he does not give, though 
he might have given it, and impression 
is what the hearer receives ... . Sugges-
tion is impression through suppression. 
Only bores want to express everything, 
but even bores find it impossible to 
express everything. Not only is the wri-
ter's art rightly said to consist largely 
in knowing what to leave in the ink-
stand , but in the most everyday remarks 
we suppress a great many things which 
it would be pedantic to say expressly . .. . 
Compound nouns state two terms, 
but say nothing of the way the relation 
between them is to be understood .. .. 

Ellipsis 

As in the structure of compounds , so 
also in the structure of sentences much 
is left to the sympathetic imagination 
of the hearer, and what from the point 
of view of the trained thinker, or the 
pedantic schoolmaster, is only part of 
an utterance, is frequently the only 
thing said, and the only thing required 
to make the meaning clear to the 
hearer. (p . 309) 

95 

Here we have a classic expression of 
what I shall call 'semantic ellipsis ' , 
the kind that philosophers have appro-
priated . But it is noteworthy that like 
most appeals to the importance of con-
text it fails to specify how such omis -
sions are to be restored or how our 
grasp of such sketchy sentences is to 
be understood . Jespersen ' s remarks 
come after all the detai led work has 
been done; they remind us of the vast 
background that work has ignored , but 
they leave it as obvious and not to be 
analyzed . 

II 

From one perspective , the idea of 
ellipsis , or Jespersen's suppression , 
could be seen as imbuing virtually 
all philosophy or logic. As Altham re-
marks, many have thought that a cen-
tral task for logic is to make ' explicit 
what is implicit in informal usage' 
(1971, p. 77), and philosophers conti-
nually seek to make the hidden visible, 
to reveal apparent qualities as really 
relations, to uncover what we are im-
plicitly committed to, and in many 
other ways to fill in some of the com-
plexity we either leave to each other's 
sympathetic imaginations or are our-
selves unable normally to conceive. But 
while my enquiry could easily flow into 
this trackless expanse, I think it worth-
while to ask what exactly those philo-
sophers who have borrowed the termi-
nology of ellipsis have made of it. 

As noted above , philosophers seem 
not to treat ellipsis as a central concept 
to be explained or justified, but simply 
as something they can invoke when it 
suits th em. There is apparently at least 
one tradition of recent philosophical 
writing in which ellipsis occurs as an 



96 E.P. Brandon 

opening gambit: the philosophy of 
probability. Ever since Keynes at 
least writers have claimed that in 
som~ cases our use of the word 'prob-
able' is elliptical . Thus Keynes him-
self: 

When in ordinary speech we name some 
opinion as probable without further 
qualification, the phrase is generally 
elliptical . We mean that it is probable 
when certain considerations, implicitly 
or explicitly present to our minds at the 
moment, are taken into account. We use 
the word for the sake of shortness, 
just as we speak of a place as being 
three miles distant, when we mean 
three miles distant from where we are 
then situated, or from some starting-
point to which we tacitly refer. (1921, 
p.7) 

A very similar doctrine was expounded 
by Kneale (1949, p. 10), only to be re-
jected by Toulmin (1956). More recent-
ly Ayer has also said that these ordinary 
judgments of probability, or what he 
calls 'credibility', are not to be under-
stood in the Keynes-Kneale manner-
'they are not judgements about the 
evidence on which they are based' 
(1972, p. 58)-but he continues the 
tradition in his own way by remarking 
that 'judgements of probability, in the 
statistical sense, turn out to be ellip-
tical' (p. 54) . 

Ayer's own account of what ordinary 
non-statistical probability judgments do 
mean seems not to arrive at any precise 
conclusion (pp. 58-61), but his posi-
tion seems close to Mackie's view of 
what he calls 'simple probability' 
(the 'good reasons' end of informal 
probability): 

One particularly tricky point is that 
simple probability is relative but not 
relational. The simple probability that 
I ought to assign to a proposition de-
pends upon what information I have and 
what I have not: it is relative to my state 
of mixed knowledge and ignorance. 
In saying 'Probably P' or 'II is x% 
probable that P' or 'There is an n to 1 
chance that P' I am speaking from a cer-
tain point of view; but it is P itself that 
I am speaking about, not the relation 
between P and the knowledge-ignor-
ance mixture that fixes this point of 

view .. .. The simple probability state-
ment takes one's present state of 
knowledge and ignorance as setting 
the framework, as fixing the universe 
of possibilities under consideration. 
(1973, p. 173) 

We shall return later to the signifi-
cance of Mackie's distinction between 
relative and relational . 

Keyne's analogy with distance in-
dicates another branch of philosophy 
in which we can find explicit mention 
of ellipsis. Swinburne, for instance, 
argues that claims about distance 
assume a frame of reference, so that 
most talk of distance is elliptical in a 
deeper way than Keynes suggests. 
While Swinburne says that such a 
frame of reference need not be expli-
citly stated in most ordinary cases, 
because it is patently obvious which 
frame is being used, he invokes the 
notion of ellipsis when he considers 
a case in which it is not clear which 
frame of reference is in question- 'here 
indeed 'distance' is elliptical and we 
need explicit speCification of the frame 
in which it is to be measured before we 
can say how far apart are the places' 
(1968, p. 80). It seems that Swinburne 
would say that (6): 

(6) A is 10 metres from B. 

is not elliptical when we all know which 
frame of reference is left unstated, but 
becomes elliptical when we do not. 
While this contrast is important, it does 
not seem helpful to make it in this way . 
Sentences such as (6) are elliptical 
(so Keynes' example is doubly ellip-
tical) but in ordinary uses unambiguous 
and determinate, since we all know 
which frame is being suppressed . But 
when we do not know this, their ellip-
sis makes them ambiguous or indeter-
minate. Often, especially in criticism, 
it is only in the second sort of case that 
one wishes to insist on the ellipsis, 
but, as Swinburne's argument shows, 
important philosophical points can arise 
in the first sort of case too. 

The awkwardness due to people's 
lack of anything definite 'in mind sur-
faces in two interestingly similar 
analyses of 'ought', Sloman (1970) and 



Wertheimer (1972) . Sloman unhesita-
tingly invokes ellipsis and discovers 
in (7): 

(7) It ought to be the case that p . 

the two elided elements restored in (8) : 

(8) Considering the possibilities in 
Z, p ought to be the case relative to the 
basis, B. 

Since he claims that sentences like (7) 
involve a choice among possibilities 
made on some basis, he asks how all 
this connects with what is going on in 
people ' s minds . Apart from 'considered 
attempts to say something true or 
false', his answer is that what people 
have in mind may be more or less un-
clear, but instructively he concludes 
that it is 'less important here to give a 
strictly accurate account of what people 
actually do say than to suggest what 
they might say if only they knew how' 
(1970, p. 394) . While Wertheimer es -
chews the term ' ellipsis', I do not think 
there is any difference of substance 
between his kind of analysis of the 
verbal auxiliaries in terms of an under-
stood ' System' and that offered by 
Sloman; and he too has to say how his 
understood System connects with ordi-
nary people.[7] And similarly I would 
suggest that the kind of analysis of 
causal language that appeals to the 
notion of a causal field [8] can rea-
sonably be construed along the lines of 
Sloman's outright use of ellipsis . 

But to continue our sampling of 
philosophical uses of ellipsis, let us 
note a chapter in Platts ' discussion of 
a Davidsonian theory of meaning, a 
chapter on adjectives (1979, ch. VII) . 
Ellipsis is continually invoked, but with 
a sense of its lack of the appropriate 
rigour (it does not appear in the index) : 

Consider the sentences: 

(9) Theo is large. (9) 

and 

(10) Rudy is attractive. 

Until we know, or form beliefs about, 
the appropriate filling, we have no com-
plete understanding of these sentences. 
We know (up to a point) the meanings 

Ellipsis 

of the words they contain, and (perhaps) 
their syntactic structure ; but we have 
no idea which conditions would make 
the sentences true or false . This (vague) 
thought is what I mean by saying that 
(9) and (10) are elliptical. It also ex-
plains why 'thing' is a dummy filling, 
since such filling effects no change in 
the deficiency. (pp. 167-8) 

97 

Platts has given us to believe that 
Theo is a flea, and he later argues that 
if we try to understand (9) as involving 
a comparison between Theo and other 
fleas we sti II have an ell i ps i s of the res-
pect(s) in which Theo is larger than the 
others (d . p . 182ff). He also suggests 
that Rudy may be attractive to look at, 
to have dinner with, as a ballet-dancer , 
or what-have-you . I refrain from pur-
suing the fascinating issues Platts has 
raised, but it is worth entering a doubt 
about (10)-it is unspecific, but should 
we regard it as elliptical? 

Platts ' example confronts us with a 
problem that has bedevilled the notion 
of ellipsis since the Greeks: the rela-
tion of a useful notion of ellipsis to lack 
of specificity. We always leave some 
things unsaid, and often such details 
can easily be restored by a sympa-
thetic hearer; should we then say that 
such details are elided? This problem 
has been viewed as a fundamental 
objection to the kind of elliptical 
analyses we have been examining. Ac-
cording to Hamblin ' s history of logical 
inattention to fallacy, Boethius noted 
that some grammatical contradictions 
were only apparent, because of the sup-
pression of different relata, as for 
instance: 

(11) Socrates is on the right-hand 
side / /. 

and 

(12) Socrates is not on the right-hand 
side / /. 

Hamblin later suggests a formal treat-
ment of this and other examples in 
terms of ellipsis, but rejects it in these 
words: 

The odious feature of our formal theory, 
considered as a remedial device, is its 
insistence that what we have called 



98 E.P. Brandon 

'incomplete' predicates are deficient 
and need to be completed .. .. That there 
are certain respects in which it is un· 
specific is an essential characteristic 
of any utterance whatsoever . (1970, 
p . 212) 

But this criticism is beside the point, 
for genuinely elliptical examples such 
as (11) and (12) , if not for some of the 
other cases Hamblin considers. We 
need a distinction between truth the 
whole truth, and the lack of suffi~ient 
detail to start enquiring about truth . 
Thus in the case of (11), we cannot 
begin to judge whether what it says is 
true until we have fixed the perspective 
(e.g., looking north from where Pericles 
is standing), but once that has been 
fixed its truth or falsity needs no more 
detai I, although of course there is an 
unlimited amount of such detail that it 
omits-is Socrates standing or sitting, 
asleep or awake, .. . . As Platts says, and 
as Swinburne's example shows , there 
is a certain amount of vagueness about 
this contrast, but by no means enough 
to put it out of action . 

This brief survey of what some 
philosophers have made of the notion 
of ellipsis shows, I hope, several things : 
their concept is an extension of the one 
found in traditional grammar and 
rhetoric; it tends to focus on what 
must be restored for there to be a deter-
minately true or false proposition 
asserted; its users are not always clear 
about how this indeterminacy relates 
to a general lack of specificity; and 
there is a question how far the theory 
of ellipsis is intended to be a true des-
cription of linguistic behaviour and how 
far a normative replacement for it. 
But the range, and importance, of the 
examples also suggests that it would 
be salutary to clarify these issues and 
evaluate the contribution the notion 
of ellipsis can make to philosophical 
and logical enquiry . 

III 

have suggested that the core idea 
of semantic ellipsis in philosophy is 
that in some, but by no means all, 

sentences certain items are left un-
specified which are required in order 
for the sentence to be evaluated for 
truth or falsity . Ellipsis gives certain 
sentences a truth-value indeterminacy. 
But we must immediately distinguish 
this source of indeterminacy from the 
worki ngs of proper names, token-re-
flexives , or non-designating referring 
expressions. 

Any use of sentence (13): 

(13) Ronald Reagan is walking. 

requires us to supplement it with 
knowledge of which Ronald Reagan 
is in question and with knowledge of 
which extent of space-time is in ques-
tion, but these ways in which the con-
tributions of the context are mediated 
are well known and can be taken for 
granted here. Once they are given, 
(13) provides us with all we need to 
determine its truth on an occasion of 
utterance (leaving aside odd border-
line forms of locomotion), though of 
course there is much that it leaves un-
said about the walking. But (14) : 

(14) Ronald Reagan needs to walk 
more / / . 

still cannot be evaluated after we have 
deciphered the clues offered by proper 
names and occasion of utterance un-
less we know at least for what end or 
purpose he needs to walk more. For his 
health , it may not be true, but to stand 
a chance of winning the White House 
staff walking race it may well be (see 
my 1980 for a fuller discussion of the 
verb 'need') . 

It is worth noting here that one can-
not resolve the ellipsis in (14) by appeal 
to some existentially quantified item-
A needs X for some Y or other-since 
virtually all such claims would then be 
true . But we do not treat such claims in 
this way. Still, there are items that in-
volve elided existential quantifications 
(e.g. 'is married') but I would suggest 
that they not be regarded as cases of 
semantic ellipsis; rather I would be 
inclined to put the implicit quantifi-
cation into the meaning of the predi-
cate. Similarly agent deletions with 
passives (e .g . 'John was murdered', 



given that Bill murdered John) and 
analogous transformational deletions 
are not to be regarded as semantic 
ellipses. But where the kind of quanti-
ficiation is left unspecified, I think we 
can usefully invoke ellipsis (see my 
1982 for a discussion of 'truth is worth 
pursuing' in such terms). These cases, 
like example (10), reveal the imprecise 
boundaries of the conception of ellip-
sis I have adumbrated, but I think they 
suggest a need for collaborative effort 
rather than for a rejection of the whole 
approach. 

Besides looking directly at how much 
must be restored for anything true or 
false to be in question, one can also 
follow Boethius, Sloman, and Werthei-
mer in using what appear to be logical 
irregularities as symptoms of ellipsis. 
Thus the fact that it might be true both 
that Mary can go to the office party 
(her car is back on the road) and that 
she can't (she has promised to baby-
sit) suggests very strongly that there is 
more logical structure than meets the 
eye. In the spirit of the Anderson-
Mackie causal fields mentioned earlier, 
one might suggest here that in using 
'can' we elliptically invoke various 
'obstacle fields' -as far as physical 
obstacles go, there is nothing to pre-
vent Mary attending the party, but as 
far as moral obstacles go, there is one, 
since she has agreed to baby-sit. 
Boethius and I have offered examples 
of apparent contradictions; Sloman has 
a section looking at apparent failures 
of simple patterns of valid inference. 
The moral is the same: logical regular-
ity can be restored by postulating 
el ided structure, by specifying what 
Mackie might have called the 'logical 
cues' governing our normal use of lan-
guage (1973, p. 7). 

If an ellipsis is suspected, how is it 
to be restored? The rough question one 
is trying to answer is fairly clear: 
what else, over and above the workings 
of proper names and token-reflexives, 
do we need to know to begin to ascer-
tain the truth value of the sentence in 
question? But is there any special way 
to answer it? The question of the me-
thods of logical investigation seems not 

Ellipsis 99 

to have been much discussed. I am fair-
ly sure we have no special roads to the 
truth here; I have elsewhere (1978) 
suggested an analogy with a Lakatosian 
research programme, but it was just 
a hint. So, unhelpfully, one can only 
say that the way to find the best charac-
terization of talk of needs or spatial 
position is yet another process of con-
jecture and refutation, even if some of 
the conjectures are pretty obvious. Talk 
of left and right obviously has some-
thing to do with a perspective; it is 
not so clear whether it is the speaker's 
(which would generate a kind of token-
reflexivity) or a more variable frame of 
reference. Another way in which this 
answer is unhelpful, though true, is 
that the analyses it describes will de-
pend on a vast amount of background 
knowledge about what is in fact re-
quired to determine truth or falsehood. 
We can see this in the contrast between 
Keynes' view of talk of distance and 
that advocated by Swinburne: in the 
I ight of relativity theory, we now see 
the need to specify a frame of reference 
to measure distance, so a hitherto un-
suspected ellipsis is revealed in our 
use of language.[10) 

If this characterization of ellipsis in 
philosophy is on the right lines, it may 
help to explain the disinclination to 
focus on the notion that I mentioned at 
the beginning. In cases of ellipsis peo-
ple omit crucial components of a deter-
minate meaning. Their words, the 
language they make publicly available, 
does not contain the whole story . 
Proper names and token-reflexivity 
are so pervasive that we do not usually 
worry overmuch about how accurately 
our theories of these phenomena cap-
ture the reality of language use and 
thought. In describing real language 
use we can hardly avoid noticing the 
phenomena I grouped together as 
grammatical ellipsis, but these are 
equally of little or no philosophical con-
cern. The philosophically interesting 
cases are precisely those where the 
specification of elided elements is more 
speculative. But talk of ellipsis offers 
a fairly precise way of specifying the 
contributions usually sloughed off 



100 E.P. Brandon 

into context. It all but forces one to face 
the possibility, acknowledged, as we 
have seen, by both Sloman and Wer-
theimer, that many people have not 
in fact got any determinate meaning in 
mind . As Swinburne claims, we may all 
know which frame of reference is being 
elided in our normal talk of places and 
distances; but what are we agreeing to 
omit in talk &hout equality or needs? 

In such cases, it is very plausible to 
think that what many people mean is 
simply incomplete; they have no clear 
idea either of what they are saying or 
of what would be relevant criticism or 
support thereof. An accurate account of 
what is meant is almost beyond us. This 
is somewhat similar to Mackie's posi-
tion on some probability statements. 
People are ascribing a predicate to an 
opinion; it is relative in that its ascrip-
tion can only be tested by reference 
to some body of evidence, as Keynes 
suggested, but the claim itself is not a 
relational claim, linking, either expli-
citly or implicitly, this opinion to the 
background of mixed knowledge and 
ignorance. What is consciously enter-
tained is indeterminate, perhaps con-
fused, an attempt to think a relational 
property as an intrinsic one. 

Altham's alternatives, 'either attri-
butive plurality assertions are inherent-
ly unclear, and have no determinate 
logic, or the predicates are implicitly 
ordered' (1971, p. 76), may present 
too extreme a contrast, but it captures 
the perplexity; have been trying to sug-
gest. If the contrast is appropriate, 
the only option for a philosopher or 
logician would seem to be to work with 
his expl icit theory -one cannot do 
much with inherent unclarity and in-
determinacy-so there would be little 
encouragement to spend much time on 
the failure of ordinary language to come 
up to scratch. But that is precisely 
what the notion of ellipsis does; it re-
turns you to the ordinary language 
with the bland assurance that all will 
be well when a few holes have been 
filled in. 

A somewhat similar situation can be 
found in the failure of logicians to take 
the theorizing of fallacies seriously, 

a failure well argued by Hamblin 
(1970) and Johnson and Blair (1980). 
It is no doubt hard enough to get logic 
right without worrying overmuch about 
what exactly is going on when people 
get it wrong. Even closer to my topic is 
the similarly off-hand treatment meted 
out to enthymemes-suppressions of 
whole sentences rather than their 
components-and implicit assumptions 
generally (for which see Ennis, 1982). 

Not only does ellipsis return our at-
tention to what is, philosophically, a 
somewhat inadequate natural lan-
guage, but as I have suggested, it does 
so in a very simple way. It supposes 
that an adequate semantic story can 
be told simply by inserting a few ele-
ments into what is already a gramma-
tical sentence. But very many philo-
sophical analyses can hardly be regi-
mented in this way . Russell's theory 
of definite descriptions, or c.J .F . 
Williams' theory of truth (1976), which 
treat natural language sentences as 
existentially quantified conjunctions, 
change the whole shape of the natural 
language they analyse. When they 
invoke elements not visible on the 
ordinary language surface there need 
not be any obvious gap in which to 
insert them. Of course, over the cen-
turies logicians have contorted natural 
languages in various ways, so it might 
be possible to provide an analysis in 
terms of ellipsis, but such an analysis 
might well not have the attractions 
of those we have been considering. 

I have indicated some reasons the 
notion of ellipsis might seem uncon-
genial or inadequate to philosophers. 
Can we add anything in defence? I 
think at least two points can be made. 
The first is perhaps the less important, 
philosophically, though it might be 
more attractive to linguists and those 
philosophers currently trying to offer 
fragments of a theory of meaning for 
a natural language. Linguistic theory, 
at least since early Chomsky, has been 
plagued by its apparent power. Any 
number of very differently organized 
rule systems could, in principle, gene-
rate the observed sentences of a lan-
guage. Of course, it is logically possible 



to have unlimited numbers of alterna-
tive theories that account for our data 
in any area; but for all its philosophical 
interest, this is hardly a live issue in 
most of the sciences . But linguistics 
has seemed less able to cope than these 
other areas. One reaction has been the 
rejection of the contrast of 'deep' and 
'surface' structures (found, for in-
stance, in Matthews, 1970, and Hud-
son, 1976) which would seem to require 
linguistic analyses to keep very close 
to what is actually said. In an associated 
semantic theory, it would not be sur-
prising to find ellipses playing a major 
role: 

In arguing that deletions have no place 
in syntax, I shall of course be implying 
that . understood' elements are to be 
identified in the semantics, which 
seems the right place for them to be. 
(Hudson, 1976, p . 126) 

But while ellipsis might still be a fruit-
ful concept in linguistics, I think that, 
for philosophy and informal logic, a 
stronger defence can be given. What-
ever the technical virtuosity it may per-
mit, I do not believe that perfecting a 
theory of comparative adjectives is the 
most important end of philosophical in-
quiry (ef. Mackie, 1977). Rather, the 
most useful point of much philosophical 
logic is critical and normative. In this 
light, to talk of ellipsis is, as Sloman 
suggests, more a critical move than an 
attempt to elucidate what people have 
in mind. It says, 'If you want to say 
something determinately true or false, 
you must be understood as adding .... ' 
Ellipsis can provide a way of under-
standing valid and invalid inferences 
more perspicuously; it can account for 
many of the nuances usually attributed 
to an unanalyzed context or shunted off 
into pragmatics; it can illuminate the 
ways ideology and unclarity persist 
(ef. my 1984b); and the varied examples 
we have glanced at should indicate the 
range over which it can make these 
sorts of contribution. 

For these critical purposes we do not 
need the final truth, the full story . 
Indeed, its baroque complexity may 
well make it positively unhelpful in 

Ellipsis 101 

these areas . Aristotle's logic, or Euler's 
diagrams, may not capture any part of 
any natural language in its complexity; 
but one could hardly deny their contri-
butions to clearer thought. (Nor, of 
course, can one deny that it is possible 
to use them to pervert and hinder 
thought.) I have elsewhere (1984a) 
advocated such a brico/eur approach 
to philosophical analysis generally, 
in which one uses what tools are to 
hand and able to do the job (as I have 
been using traditional grammatical 
categories in this discussion) . Ellipsis 
seems to me a peculiarly simple and 
powerful tool to elucidate context de-
pendencies, equivocations, and con-
fusions of thought, whether or not 
it will form part of our final grand 
theory of meaning . 

Whatever the motivations of philo-
sophers of language, logicians, and es-
pecially informal logiCians, are usually 
concerned with the implicit logic of ex-
pressions not for its own sake but rather 
as a means of exposing and correcting 
incoherence and invalidity . As a central 
part of the ordinary language philo-
sopher's arsenal, ellipsis might too 
obviously have falsified that illusory 
neutrality some such philosophers 
aspired to (ef . Graham, 1977, ch. II), 
but its affinity for such critical uses 
should make it very welcome in the in-
formal logician ' s camp where values 
are somewhat more conspicuous. I 
hope this survey of its uses , attempt to 
clear up some of its problems, and in-
dication of a few of its ramifications, 
will inspire further work both on and 
with the notion, and also in assessing 
its pedagogical utility in the promotion 
of critical thinking. 

Notes 

[1) I cannot claim to have searched ex-
haustively for such an account, but 
I could try the patience of my 
readers with a long list of likely 
works in which such silence reigns. 
Thus, in the Encyclopedia of philo-
sophy (Edwards , 1967), the term 



D2 E.P. Brandon 

appears neither in the general 
index nor in the catalogue of lo-
gical terminology; the Kneales do 
not treat of it, though they casually 
remark that /statements of relative 
necessity or possibility are often 
made elliptically and may for this 
reason be misunderstood as state-
ments of absolute necessity or 
possibility' (Kneale and Kneale, 
1962/ p. 93); Alston/s (1964) intro-
duction to the philosophy of lan-
guage discusses various ways in 
which words may be vague but 
does not mention the indeter-
minacy of phrases or sentences 
due to ellipsis; and so one could go 
on . 

[2) /Cum subtractum verbum aliquod 
satis ex ceteris intelligitur' (1921/ 
IX. iii. 58; trans. H.E. Butler) . 

[3) / Vicit pudorem libido, timorem 
audacia, rationem amentia' / a 
quotation given by Quintilian from 
Cicero's Pro Cluent. vi . 15. Trans-
lation by H.E. Butler. 

[4) Some arguments go on and on-
how wou Id Jespersen react to H ud-
son/s recent remark 'here what 
seems to act as both relative pro-
noun and antecedent-like a com-
pressed version of that which' 
(1976/ p. 135)? Hudson does not, 
I think, endorse quite this analysis, 
and his surrounding discussion is 
explicitly tentative, but it testifies 
to the attractiveness of Sweet's 
analysis. 

[5) There they, and various other 
phenomena to be mentioned later, 
are usually handled by deletion 
rules, though such grammarians 
are much more aware of the fright-
ening complexity of such rules 
when explicitly stated. For dele-
tions in transformational grammar 
see, for instance, Akmajian and 
Heny (1975/ esp. ch. 7), Allerton 
(1975)/ and Hudson (1976/ esp. 
ch. 3.9) . A wide-ranging discus-
sion can be found in Matthews 
(1981) . 

[6) The third type of case, abbreviated 
phrases, is semantically moti-
vated/ but ellipsis here has usually 
been seen as part of a historical 
account of the genesis of new 
meanings for a word or phrase. 
The elided element is usually 
obvious and is of ten , but not al-
ways/ marked by syntactic oddities 
(such as the clash of genders in the 
French example). 

[7) What he does say is: 
For a speaker to mean some Sys-
tem/ he need not actively think 
about the System. Nor need he be 
able to say much about the Sys-
tem or even know what a System 
is. To say that he has a certain 
System in mind is only to say that 
he is inclined to accept certain 
kinds of reasons in support or 
criticism of his utterance, and to 
regard other kinds of reasons as 
irrelevant. If he is unable to do 
such things even in a rough-and-
ready way, then the truth of what 
he says cannot be assessed, for 
there is no way to determine what, 
if any thing, he means . (1972/ 
p.95) 

[8) 'We are concerned not with rela-
tions between A and B in general 
but with relations within certain 
limits or in a certain 'field'; ... 
a cause is always a cause within 
a field' (Anderson, 1938/ p. 130). 
This notion has been used by 
Mackie in several places in dis-
cussions of causality and respon-
sibility. 

[9) I have altered the numbering of 
examples to fit this paper's se-
quence. 

[10) One could of course treat unso-
phisticated talk about distance ra-
ther in the way I suggested for 
'is married ' , i.e., put the everyday 
frame of reference into the mean-
ing of the predicate. Descriptively 
the choice between these ap-
proaches is a matter of the best 
overall theory of the language 
and its speakers / but for critical 



purposes the elliptical analysis 
seems to me much the better . 

References 

Akmajian, A . & Heny, F .W. An intro-
duction to the principles of trans-
formation syntax. Cambridge, 
Mass .: MIT Press, 1975 . 

Allerton, D .J . Deletion and proform 
reduction . Journal of Linguistics, 
1975,11,213-237. 

Alston, W .P. Philosophy of language. 
Englewood Cliffs: Prentice-Hall, 
1964 . 

Altham, J .E .J . The logic of plurality . 
London: Methuen, 1971. 

Anderson, J . The problem of causality . 
In Studies in empirical philosophy . 
Sydney : Angus and Robertson, 
1962 (orig . 1938). 

Ayer, A .J . Probability and evidence . 
London : Macmillan, 1972 . 

Brandon, E .P. Hintikka on akolouthein . 
Phronesis, 1978, 23, 173-178 . 

Brandon, E . P. 0 reason not the need. 
Education for Development, 1980, 6, 
18-25 . 

Brandon, E . P. Quantifiers and the pur-
suit of truth . Educational Philosophy 
& Theory, 1982, 14,50-58. 

Brandon, E .P. The philosophy in philo-
sophy of education . Teaching Philo-
sophy, 1984, , 1"-15 . (a) 

Brandon, E .P. Ellipsis and ideology . 
Paper to be presented at the Con-
ference on Thinking, Harvard Uni-
versity, Aug. 1984 . (b) 

Edwards, P. (Ed .) The encyclopedia 
of philosophy . New York : Crowell 
Collier and Macmillan, Inc., 1967 . 

Ennis, R.H . Identifying implicit as-
sumptions. Synthese, 1982, 51, 
61-86 . 

Fowler, H .W . A dictionary of modern 
English usage. Revised by Sir Ernest 
Gowers . Oxford : Oxford University 
Press, 1983 . 

Graham, K. J.L. Austin: a critique of 
ordinary language philosophy. New 
Jersey: Humanities, 1977 . 

Hamblin, c.L. Fallacies . London : 
Meth uen , 1970. 

Hudson, R.A . Arguments for a non-
transformational grammar . Chicago: 
University of Chicago Press, 1976. 

Ellipsis 103 

Jespersen, O . The philosophy of gram-
mar. London : George Allen & Unwin, 
1924. 

Johnson, R.H . & Blair, J . A . The recent 
development of informal logic . In 
J.A. Blair & R.H. Johnson (Eds .), 
Informal logic . Inverness, California : 
Edgepress, 1980 . 

Keynes, J .M. A treatise on probability . 
London: Macmillan, 1921 . 

Kneale, W . Probability and induction. 
Oxford : Clarendon Press , 1949 . 

Kneale, W . & Kneale, M . The develop-
ment of logic . Oxford : Clarendon 
Press, 1962 . 

Lausberg, H . Handbuch der literaris-
chen Rhetorik . Munchen: Max 
Hueber Verlag, 1960. 

Mackie, J.L. Truth, probability, and 
paradox. Oxford: Clarendon Press, 
1973. 

Mackie, J .L. Review of I. Hacking, 
Why does language matter to philo-
sophy? Philosophy, 1977 , 52, 
359-362. 

Matthews, P.H. The adequacy of gram-
mars . Aristotelian Society Supple-
mentary Volume, 1970,44,175-190. 

Matthews, P.H. Syntax. Cambridge: 
Cambridge University Press, 1981. 

Platts, M. Ways of meaning . London: 
Routledge, 1979 . 

Quintilian . The institutio oratoria of 
Quintilian, trans . H .E . Butler . 
London : Heinemann, 1921 . 

Sloman . A. 'Ought' and 'better'. 
Mind, 1970, 79, 385-394 . 

Swinburne, R. Space and time . London: 
Macmillan, 1968 . 

Toulmin, S. Probability. In A . Flew 
(Ed .), Essays in conceptual analysis. 
London : Macmillan, 1956. 

Ullmann, S. Semantics : an introduc-
tion to the science of meaning . 
Oxford : Basil Blackwell, 1962 . 

Wertheimer, R. The significance of 
sense. Ithaca : Cornell University 
Press, 1972 . 

Williams, c.J .F. What is truth? 
Cambridge : Cambridge University 
Press, 1976 . 

Professor E.P. Brandon, School of 
Education, University of the West 
Indies, Mona, Kingston 7, Jamaica 0