WILL, AND MUST A STUDY IN FAMILY RELATIONSHIPS
I propose to do here some philosophically interesting linguistics
and then show why I consider it philosophically interesting.
of English modal verbs is set up and modal predicates distinguished
from their arguments. These
may be epistemic or mandatory, intransitive or transitive, binding
or releasing. The major subsystems 1, 2, 3, 4, 4a are taken
up serially and the distinctions applied to them. Finally the minor performative subsystems 5,
6, 7, 8 are described briefly.
nature of these investigations is brought out and described. The philosophical implications of the following
features of this system are then suggested : (i) the absence of any
serious constraints on the Propositus that constitutes an Argument
of the Modal Predicate; (ii) the relevance of tense-considerations
in explicating modal sentences; (iii) the interrelations within each
subsystem and between subsystems that define how modal sentences can
be disputable, refutable, or infelicitous.
hoped that this account will induce some fresh digging into the problem.
of you are no doubt aware I am no philosopher—not by a professional
chalk anyway. Not unless occasionally worrying about problems
that customarily get called ‘philosophical’ entitles one to be taken
for a philosopher. Only politeness
then will make you hold back the question, What is this professional
linguist (for that’s what I am) doing here in a gathering of philosophers?
I can of course answer this question quite truthfully at a personal
level. Well, I can hear myself saying, I’m here because
some philosopher friends of mine have encouraged me to believe that,
since language has been worrying philosophers a good deal lately,
what a linguist has to say is likely to be of professional interest
to philosophers. So I’m here
in a sort of representative capacity.
sure that philosophers will find such an answer too personal. So let me see if I can’t get any philosophical
mileage out of this question. I
shall therefore offer my guess about the true relationship between
formal linguistics and the loosely defined body of activity known
as 20th-century linguistic philosophy.
Briefly put, while some linguistic philosophers have tried
to naturalize ideal languages, other linguistic philosophers have
tried to idealize natural languages. In so doing, they are to be credited with or
accused of doing linguistics informally—not all the time to be sure,
but some of the time. This
is especially true of the second group busy holding up ordinary language
and saying, Behold the language!
me to suggest that history is repeating itself here.
The domains of Natural Philosophy and Moral Philosophy have
now been largely annexed by Natural Science and Social Science respectively. As if in recognition of the pioneering work
done by philosophers in these fields, the sciences serve philosophy
by presenting it with certain brute facts.
Let Copernicus propose the heliocentric system, or Darwin biological
evolution, or Heisenberg the uncertainty principle, or behaviour scientists
various types of conditioning and philosophers sit up and take notice. Ordinary language analysis has already lost
some of its charm for philosophers, while linguists are moving into
the territory with enthusiasm. The
maps that linguists make will presumably continue to be of interest
to philosophers at least on two counts—to distil language is to distil
the native wisdom of generations of language users and this native
wisdom has been brought to bear upon all the principal areas and modes
of experience that a philosopher is likely to be concerned about,
and, secondly, the design of language itself apart from what language
has to ‘say’ is too important an aspect of characteristically human
activity for philosophers to ignore brute facts about language.
presented my credentials as it were, let me now put my cards on the
table. First, I shall present here an analysis of
the English modal verbs. In
so doing I shall naturally try to select those aspects of the system
that will be of interest to philosophers—I shall not, for example,
say anything about the substitution of shall for will
or about the relation between modal verbs and the expression of time
or about certain interesting facts about their behaviour in relation
to sentence accent. Further, I shall concentrate on the bold features of the modal
system, neglecting fuzzy edges, special cases, and the like. Even after selecting philosophically interesting
aspects, I shall deal in greater detail with only some of the subsystems,
content to present only a sketch of the remaining subsystems. It is in the second part of this study that
I shall try to suggest possible philosophical implications.
An English sentence often has two verbs—that is, in addition
to the usual main verb, there is a modal verb apparently sharing the
same subject and embodying some comment on the speaker’s part on the
validity of the rest of the sentence. The modal verbs that we shall be concerned
with are : will, may, can, must, ought, should, let, shall, would,
might, could, and, to some extent, need.
We shall leave out the modal-like uses of have (he has
to go), manage (he managed to go), be and be
combinations (he was to go, he was able to go, he is going to come),
and dare (he dare not show his face).
Explicating the modal is essentially explicating its relation
to the validity of the rest of the sentence.
From now on, we shall use the terms PROPOSITUS to avoid repeating
the awkward phrase ‘ the rest of the sentence ’ : the PROPOSITUS is
simply the event, state, process, action etc. reference to by the
main verb together with its accompaniments (subject, object, complements,
manner adverbials, and the like). To keep matters simple we shall keep the tense
variable constant—in all the examples that follow PAST time reference
is excluded, it is either PRESENT or FUTURE.
And, until towards the end, we shall stick to the affirmative
polarity. NOT introduces a
complication in that we have to decide whether it belongs to the PROPOSITUS
or to the MODAL in the sentence.
You must speak frankly.
You mustn’t speak frankly.
Here (2) cannot be paraphrased
(2a) It is not the case that you must speak frankly.
Rather, (2) should be
(2b) You must abstain from speaking frankly.
If what one wants to say
is (2a) and not (2b), one has to resort to (3)—
You needn’t speak frankly.
Need is, as it
were, a variant of must appropriate when not negates
the MODAL. When the PROPOSITUS is negated, mustn’t
is quite appropriate. So—
MODAL (PROPOSITUS) (example 1)
MODAL (NOT-PROPOSITUS) (example 2)
NOT-MODAL (PROPOSITUS) (example 3)
Coming now to the MODALS proper, we have
to establish certain distinctions to begin with.
First, a distinction between EPISTEMIC modals and MANDATORY
modals. Compare (1) with (4)—
You must speak frankly.
You must be crazy to speak so frankly.
And compare the pair (1)-(4)
with the pair (5)-(6) :
He must be careful.
He must be careless.
Clearly, (5) and (6) can
be said together without any contradiction, as in—
He must be careless. It’s a pity; for he must be careful you know.
The simplest way of accounting
for these examples is to say that must is open to an EPISTEMIC
interpretation where grounds for believing the PROPOSITUS are being
evaluated and a MANDATORY interpretation where grounds for implementing
the PROPOSITUS are being evaluated.
(1), (5), and (5) as a part of (7) favour the MANDATORY interpretation;
(4), (6), and (6) as a part of (7) favour the EPISTEMIC interpretation.
The two groups of examples also illustrate another distinction—the
one between INTRANSITIVE modals and TRANSITIVE modals.
With the former, the PROPOSITUS is the subject of the modal. With the latter, the MODAL and the PROPOSITUS
genuinely share a common subject.
Thus, (4) and (6) in their normal interpretation show an INTRANSITIVE
must, (6a) being a suitable paraphrase of (6)—
That he is careless must be the case.
So, in (6) he is
only apparently or superficially the subject of must, somewhat as in (8). (Compare
(8) with (8a).)
He happened to visit my place.
What happened? It happened
that he visited my place.
So must of (6)
and (6a) is a one-place predicate with the PROPOSITUS as its
argument. The must of (5) is on the other hand
transitive—a two-place predicate.
must (PROP) (examples 6, 6a)
must (he, PROP) (example 5)
As we shall see later on EPISTEMIC modals can be
transitive too : and MANDATORY modals can be intransitive too.
The third distinction concerns the amount of value that we
place on the grounds being offered for believing (or implementing,
as the case may be) the PROPOSITUS.
The grounds may be made explicit as in the marked portion of
(4) or (9) or implicit as in (6).
You must be crazy to speak so frankly.
They must be newly-weds : they look so happy.
Now, in these two examples
the grounds are so strong to the speaker that in his eyes they BIND
one to believe the PROPOSITUS. But
suppose the grounds are not strong enough to effect such a BIND?
(As in 9a.)
(9a) They needn’t be newly-weds : they aren’t
At the same time there
is no ground for a negated PROP.
(As in 9b.)
(9b) They mustn’t be newly-weds : they look unhappy.
So that one may want to
retort to (9b) with (9c).
(9c) Well, they can be newly-weds : they look
happy enough. So, while
there is no BIND to believe their recent marriage there is a RELEASE
to that effect. While must is a BIND modal, can is
a RELEASE modal. The modal
negation of (9c) will bring us back to (9b)—so (9d)
is a more idiomatic version of (9b).
(9d) They can’t be newly-weds : they look so
Armed with these three distinctions, we can look at some subsystems
of English modals :
their interrelationship :
(PROP) = NOT-RELEASE (NOT-PROP)
in the equivalence of (9) and (9e).)
(9e) They can’t not be newly-weds : they look
Where there is a WILL there is a MAY
examples (10)-(13) :
Mary will leave if John comes.
Mary may leave if John comes.
Mary would leave if John come.
Mary might leave if John come.
Let us apply our three pairs (vi)-(viii).
These are all clearly EPISTEMIC—we are being asked
to believe in the PROPOSITUS—namely, Mary’s departure on John’s arrival. Again, these are all INTRANSITIVE; they are
readily paraphrasable as—
(10a) It will be the case
that Mary leaves if John comes.
(11a) It may be the case that Mary leaves if John
the idiomatic use of ‘ maybe ’ for ‘ perhaps ’ .)
(12a) It would be the case that Mary leave if John
(13a) It might be the case that Mary leave if John
Finally, while (10) and (12) have the force of a
BIND, the other two (11) and (13) merely connote RELEASE. Thus, there is nothing odd about (10b), (11b), (11c).
(10b) Mary will leave if John comes; and I believe
(11b) Mary may leave if John comes; but I don’t believe
(11c) Mary may leave if John comes; or she may not.
(may not her is may (NOT-PROP))
On the other hand (10c), (10d), (11d)
will be distinctly odd :
(10c) Mary will leave if John comes; but I don’t
(10d) Mary will leave if John comes; or she will
(11d) Mary may leave if John comes; and I believe
While (11c) can mean that both PROP and NON-PROP are
equally probable, (10d) is defensible only if we give it a
purely analytic interpretation as an application of the principle
of the excluded middle. (11d), again, is defensible only if ‘ it ’ is interpreted
not as ‘ Mary’s leaving if John comes ’ but as ‘ Mary may leave if
John comes ’.
Certain clarifications are needed at this point.
First, the presence of an if-clause seen in (10)-(13)
is not an essential feature. The
if-clause may be left suppressed.
(Compare the suppression of a since-clause in (6) and
its presence, in various guises, in (4), (9), (9a-d).
He will/may/would/might know the answer.
with an implied-‘if you ask/asked him’.)
(An early morning doorbell ring.) That will/may/would/might be the milkman.
with will under (14) and (15) also serve to bring out a second
fact. While will has
been saddled with the duty of providing a future tense for English
by traditional grammarians brought up on Latin and hence missing it
in English, will actually ranges over present as well as future.
(It is obvious that would, might, and ‘ if John came ’ are bereft here of their PAST time force.)
the condition introduced by if (or the concession introduced
by though) may be counterfactive as in (12), (13) with a strong
suggestion that John’s arrival is never going to come about or it
may be nonfactive as in (10), (11) where if can easily be replaced
by if and when.
subsystem to be considered now matches this except that the modals
there are TRANSITIVE.
Dispositions and capacities : One would
if one could
I have already hinted that EPISTEMIC modals can be TRANSITIVE. Consider (16)-(19) which match (10)-(13).
Mary will say yes if John proposes to her.
Mary can say yes if John proposes to her.
Mary would say yes if John proposes to her.
Mary could say yes if John proposes to her.
The comparison of the
two set should serve to bring out the two meanings of will.
will (PROP) (will-1 in example 10)
will (Mary, PROP) (will-2 in example 16)
Consider the following
which illustrate both the subsystems together.
If Mary will-2 not marry John, she will-1 not.
If what beggars will-2 beggars can, then beggars may—may, beggars
(10)-(13) BIND (or RELEASE, as the case may be) one to believe PROP,
(16)-(19) BIND (will-2) or RELEASE (can = be able to)
one to believe of someone that PROP is the case.
Thus (22) and (22a) are paraphrases of each other under
will-1 but not under will-2.
Mary will not marry John.
(22a) John will not be married to Mary.
between will-2 (willingness, disposition) and can (ability,
capacity) is of course a fundamental one that runs through the world
of man and certainly through the world of cognition.
What one will-2 say is what one thinks or believes. What one can say is what one knows.
(22b) Believe of something that PROP is the case.
This look won’t open.
This pen won’t write.
This key will open any lock.
Note : This look/this
pen/this key is what one believes of that PROP is the case.
What one can’t say one mustn’t speak.
must is MANDATORY modal as in (2).)
will be one possible paraphrase of the Wittgenstein dictum (Tractatus
Is ought a must? And
can a may?
We have already seen how must is either intransitive
and epistemic or transitive and mandatory.
Both are, however, BIND modals.
The can of (9c) is the epistemic modal of RELEASE.
We are actually dealing here with the third and the fourth
subsystems. Let us take up
the epistemic subsystem first. (We will suppress the grounds for brevity’s sake.)
They must be newly-weds.
They ought to be/should be newly-weds.
(9c) They can be newly-weds.
immediately arise. What is
the difference between the will-1/may pair of subsystem
1 and the present must/can pair of
subsystem 3 ? (This
can is of course different from the ‘ be-able-to ’ can
in the will-2/can pair of subsystem 2.)
Both pairs are epistemic and intransitive.
Compare (25) with (26), and (27) with (28).
That will be four shillings.
That must be three shillings.
He may be annoying sometimes.
He can be annoying sometimes.
The must-can pair is certainly more contentious in tone than
the may-will pair. While the shopkeeper will use (25) (with an
implied ‘ as a matter of course ’), the customer must use (26) ! The
grounds of validity being offered in each case are different in character—knowledge
about relevant circumstances (KRC) in one case and belief about relevant
circumstances (BRC) in the other case.
The schemata for (25)-(28) are respectively :--
will (PROP) = BIND (BRC, anyone, BELIEVE-PROP)
must (PROP) = BIND (KRC, anyone, BELIEVE-PROP)
may (PROP) = RELEASE (BRC, anyone, BELIEVE-PROP)
can (PROP) = RELEASE (KRC, anyone, BELIEVE-PROP)
BRC is associated with
if-clauses, KRC with since-clauses, (27) may be continued
appropriately with as far as I know; while (28) may
be continued appropriately with such is his nature or he is unable
to help it.
The second question concerns the place of ought and
should in the must-can system, for it seems
clear that they belong there and call for KRC.
Both, one may add, are BIND rather than RELEASE modals. But there is a crucial difference between must
and ought/should—one who concedes (19) will readily concede
(24), but one who concedes (24) needn’t concede (9) at all. So (24) presents a weaker version of (9), it is what one believes
but doesn’t know for sure. Let
us symbolize this added element found in ought/should but not
in must as THINK. Going
back to may and can, one may point out an additional
difference between the two—may has a THINK element in it but
can hasn’t, which means that conceding (27) implies conceding
(28) but not the other way round.
The three EPISTEMIC modal verb systems can now be calibrated
with each other and with the EPISTEMIC modal adjectives which philosophers
feel more at home with.
(xv) EPISTEMIC (i.e. BELIEVE-PROP) modal verbs
Subsystem 1 Intransitive
(BRC, anyone, BELIEVE-PROP)
Subsystem 2 Transitive (BRC,
anyone, Agent, BELIEVE-IMPLEMENT PROP)
Subsystem 3 Intransitive
(KRC, anyone, BELIEVE-PROP)
(B) THINK (BIND)
almost certain, more than probable
(C) THINK (RELEASE)
Probable, more than possible
Note : Gaps in the system are indicated by . . In respect of less
careful usage, however, . . may be interpreted as ditto signs. Thus, ought/should are used not only
for B-3 but also for C-3;
similar observations hold good for will-1, may and will-2. Note further that subsystems 1,2 aren’t speaker’s
endorsement’, but subsystems 3, 4 are ‘---speaker’s endorsement’ (the
speaker taking the stance ‘as
for as I know ’.
The MANDATORY modal systems in English are not entirely parallel. They merge C and D.
(xvi) (A) must : essential, more than
(B) ought/should : desirable/advisable,
more than permissible
Note : Note that the subsystem
of modal verbs are ‘+ speaker’s endorsement’ and the subsystem of
modal adjectives and participles are ‘- speaker’s endorsement’, The
‘have to’ construction which is like A must is, however, ‘-
The appropriate schemata will be—
must (Agent, PROP) = BIND (KRC, Agent, IMPLEMENT-PROP)
ought/should (Agent, PROP) = THINK (BIND (KRC, Agent, IMPLEMENT-PROP)
may/can (Agent, PROP) = RELEASE (KRC, Agent, IMPLEMENT-PROP)
Relevant Circumstances may be the various demands on the Agent (of
law, custom, fashion, morality, common sense, and the like) or the
interests of the agent. (When
should has this latter sense, it can be paraphrased by had
better.) The source of the demands made on the Agent
may be someone other than the Agent—very often the Speaker who wants
to declare them (you must, he may) or the Addressee who is
being asked about them (may I ?, If I may should he ?) or some
third party (I must, must you ?, he can).
Note that KRC of (xvii)-(xix) is known relevant circumstances
in subsystem 4 but knowledge of relevant circumstances in subsystem
significant that must, ought, should, can, and may have
each an EPISTEMIC (see (xvi) interpretation.
(Originally they were all transitive and, with the exception
of the last two, mandatory.) Ordinary
language frequently paraphrases the EPISTEMIC in terms of the MANDATORY.
This must be true = I must say that this is true.
This ought to/should be true = I ought to/should say that this is
This may/can be true = I may/can/dare say that this is true.
It also permits mutual
embedding as in—
I must say that this can be true.
the contentious EPISTEMIC can is embedded in
PROP of the MANDATORY must.)
I can say that this may be true.
It is possible/is probable/may be the case that you must speak frankly.
(Here the MANDATORY must is embedded in the PROP of
He may certainly/probably/possibly be careless.
When wishes are the horses, performatives will ride them
When the Known Relevant Circumstances take the shape of the
Author of the Speech Act, the Speech Act takes on a performative force. This is seen in four minor subsystems—one EPISTEMIC and three MANDATORY.
The fifth subsystem has let as the only member.
Let John come, and Mary will/may leave.
is a paraphrase of (10), (11”) respectively.)
John come. Or/for Mary will/may leave
John doesn’t come. Mary will/may
Let A be equal to B, and B will b equal to A.
(Let yourself/anyone) scratch a Russian, and you/he
find a Tartar. (One could
yourself/anyone) scratch a Russian, but you/he
find a Tartar.)
(Let yourself) do it’ or it will/may be the worse for you
(Here the first two words, are usually left understood.)
second example, but can be left understood or replaced by yet. The example can be paraphrased as follows :
if you/anyone scratch a Russian.
You/he won’t find a Tartar. Consider also :
happen/Whatever happened/Happen what may, he won’t budge. (Also :
schema covers the let-sentences :
let (PROP) = BIND (SPEAKER, Agent, SUPPOSE-PROP
BIND is replaced by RELEASE the condition with let…and is converted
into the concession with let…but.)
subsystem has may and let.
(May) God bless/damn you !
(Let) Devil take the hindmost !
may ( PROP) = BIND (SPEAKER, BRC, IMPLEMENT-PROP)
let (PROP) = RELEASE (SPEAKER, BRC, IMPLEMENT-PROP)
can also be a BIND modal as in (41), a line from Rabindranath
Into that Heaven, O Lord, let my country awake !
Speaker to invoke the world (Believed Relevant Circumstances, to be
precise) to implement the PROPOSITUS is certainly a brave performative
act that harmonizes poorly with the modern world (whence the archaism
the subsystem has shall as the only member.
They (shall) pay now/later.
Either they (shall) go or I (shall) go.
schema will be—
shall (PROP) = BIND (SPEAKER, SPEAKER, IMPLEMENT-
is here, so to say, staking his honour on that the PROPOSITUS be carried
out. There is an epistemic suggestion also—‘what
I undertake to implement will come about’.
three subsystems above are intransitive.
The eighth one is transitive.
If the source of the demands on the Agent is someone other
than the Agent, the appropriate modal is shall (I shall,
shall I ?, thou shall not, they shall). If it is the Speaker, the appropriate modal
is will. If the Agent
is someone other than the Addressee, the appropriate modal is let. The distinction between BIND and RELEASE is
somewhat blurred in this subsystem.
Some of the permissible types are illustrated below.
Let me join you, shall I?
Let us have our own way, will you?
Let’s/Let me and you draw lots, shall we?
I will/shall write to you later.
(You will) leave this place, will you?
(You will) give us a chance, won’t you (please)?
(You will) leave before it is too late.
(You will) pay later if you like/please.
You shall report to me tomorrow.
(You will) get well soon, won’t you?
He will/shall report to me tomorrow. (Said by the boss)
Let him pay later if he likes. (Said by the boss)
Let him pay later if he likes. (Someone reporting the boss)
He shall report to our boss tomorrow.
Out of these, all except the last two have a performative
force—in (44), (45) the Speaker seeks direction from the addressee;
in (46) he makes a proposal to the Addressee which may amount to a
direction; in (47) he makes a promise (in this case, to the Addressee);
in (48) to (52) he seeks to direct the Addressee respectively by way
of command, request, advice, permission, and demand; the advice in
(50) can be mock-advice—that is, a disguised threat; in (53) the Speaker
makes a mock-request which really conveys a wish; in
(54) he issues a command or a demand; in (55) he conveys a
permission, request, or advice. The
last two, (56) and (57), are of course only relays of performative
acts—the quotation marks are left understood, as it were.
Note that (47) lacks the solemn urgency of (43); and that (53)
lacks the solemn urgency of (39)—for obvious reasons.
modal subsystem of must, ought/should, and may/can described
earlier is mildly performative in that these convey the Speaker’s
endorsement of the BIND or the RELEASE over and above his report of
it. Compare (58) which conveys endorsement with
(59) which does not.
The girls may not wear miniskirts in this school (which is at it should
parenthetical addition is a MANDATORY embedding the earlier MANDATORY
The girls are not permitted to wear minishirts in this school. (This may be
with : ‘ which is idiotic ’; (58) cannot be so continued.)
You have to speak frankly at times. (This lacks the appeal of (1) to the
completes in broad outline the description and analysis of the English
modal verb system.
You may have observed that I have so far jealously tried to
keep all my eggs in the basket of linguistics.
Not only will a good deal of this stuff be familiar to linguists—familiar
insights contributed by generations of linguists from Jespersen the
Dane to Antinucci and Parisi the young Italian students.
But whatever innovations, renovations, and departures I have
made and whatever insights I have taken over from logicians (for example,
the interlocking logic of require/blind and permit/release)
and philosophers (for example, the distinction between alethic/epistemic
and deontic/mandatory and the notion of the performative aspect of
utterances), I have done so without ceasing to look for the sort of
arguments that a linguist will look for.
Though I have naturally refrained from presenting such arguments
at length in this study, the flavour of such arguments may have become
apparent by now. A linguist will look for formal correlates
of distinctions of use or meaning : the mandatory must never
loses its accent; epistemic may not is NOT-PROP and mandatory
may not is NOT-MODAL; will-1 is passive-transparent
while will-2 isn’t (examples 22, 22a).
of course, these formal correlates may be fairly subtle sometimes—for
example, the arguments for establishing the deeper intransitivity
of some modals.
It may be noted in passing that information about earlier stages
of language (say, about Old English and about the buried relationship
between may/might and might/mighty) or about cognate
languages (say, about modals in German) is, strictly speaking, unnecessary
and inconclusive for analytic linguistics, thought there is no denying
its suggestive value. This self-denying ordinance which linguists
have imposed on themselves lately need not upset one if it is realized
that information of this kind is not available either to the child
learning his very first language—become as little children if ye will
enter the heaven of linguistic analysis.
There is another kind of empirical evidence that the linguist
uses—namely, observations on the mode of use (e.g. the solemn urgency
of may in (39) and its absence in the comparable use of the
Imperative in (53); the sort of continuations that are admissible
and the sort that aren’t). This kind of evidence could also include logical
judgements of compatibility, incompatibility, and the like offered
by language users (e.g. our discussion about the must/ought contrast). Some of the constraints can be traced to the
logic of communication as such (e.g. one has no occasion to put a
question about one’s own wishes or to inform the addressee of the
Now whether it is the
behaviour of forms within the linguistic system or the behaviour of
speakers and addressee within a linguistic transaction that a linugist6
is examining, in either case he is playing the anthropologist.
If by the term ‘ category ’ we understand how members of a
society sharing a culture customarily attribute similar characteristics
to a class of things, respond similarly to these, and discriminate
them from other categories, then it is the linguist’s job to identify
linguistic categories by observing linguistic transactions.
His interest is not in the linguistic transactions as such;
his interest in them is for sake of the shared linguistic intuitions
or categories of the users of that language that are revealed through
Finally, while the starting point of a linguist’s investigation
is quite often an intuited family resemblance, the object of his investigation
can more properly be called family relationships.
He has no interest in a single sense of a single term as such
– say, the ability sense of can or the mandatory sense of ought--
but rather in the whole fabric of relationships from which the term
derives its whole raison d’être.
Thus, rather than look at the moral ought in splendid
isolation, a linguist relates this particular use first to the whole
mandatory range of ought as seen in –
You ought to try these biscuits : they’re delicious.
The red ought to go here and the blue there.
We ought to bump off the old man one of these days.
then to the other members
of the subsystem must, should, can, may; then to the other epistemic use of ought; and finally to the
matching modal adjectives desirable and more than permissible. In working out these systems, a linguist is
always ready to come across untidy gaps and accretions, fossil items,
blurred edges and the like. Being
an anthropologist, he will almost be disappointed if he doesn’t !
After his analytic job is over, a linguist may want not only
to analyze the next language in view but also, within the frame-work
of his discipline, to compare languages as wholes or in respect of
some chosen detail – say, modal components of the verbal system.
Such comparisons may be either historical – that is, undertaken
to investigate relationships of descent and influence – or correlative
– that is, undertaken to find what trait in one language translates
(in the broadest sense of that term) what trait in another.
For example, he will point out to the matching ambiguity of
(64) in English and (64a) in Marathi.
It ought to rain today : it has been very sultry for some time.
(64a) āj pāus paḍāvā
: phār ukaḍta āhe
One doesn’t know if the
ground (sultriness) is being offered as a symptom of the rain in the
EPISTEMIC mode or as a justification for desiring rain in the MANDATORY
Philosophically, correlative comparisons are going to be the
more interesting of the two in that they are expected to reveal either
language universals or deep-seated cleavages between languages.
Having thus let you have a peep into the linguist’s kitchen
which may be of special interest to conceptual analysts, I shall now
proceed to keep my promise to indicate possible philosophical relevance
of the foregoing analysis of English modals verbs, which, as modal
verb systems in ordinary language go, probably present an unusually
perspicuous example of this species. Marathi and French, which I know something
of, are probably pretty messy and Hindi is rather sparse on this given
point. In selecting English,
it is as if a botany teacher happily picked up a flower of unusual
“normality” as a classroom specimen.
Everything is grist to the Embedding mill
A negative feature of this system is that we have had no occasion
to state and salient constraints on the sort of predicates and arguments
that can go into the PROPOSITUS when a modal of a certain type is
dominating it. We have already seen that modals can embed other modals
rather freely with results that are not always calculated to make
a logician happy. We have
also seen mutual paraphrase relations (as in (20) to (31)) between
Subsystems 3 and 4. The nearest
to such a constraint that one can think of is the recommendation that
the PROPOSITUS of a transitive modal have an identifiable Agent. The identifiable agent need not be overt.
With the newer techniques a ten-storey building can be built in as
many weeks. (ability can)
And, of course, since
this is only a recommendation, the Agent need not be there at all,
as in (64) or (64a) above in their MANDTORY interpretation.
In spite of this lack, (64) is as firmly tied to the desirability
ought as it is distinct from (6), (67), and (68).
It ought to rain today : it has been very sultry for some time.
May it rain today !
continuation in (64) will be incongruous here.)
I wish that it rain today.
I will be happy if it rains today.
Clearly there is a shadow subsystem by the side of the fourth
subsystem – both are MANDATORY but while the fourth is transitive,
the shadow (let us call it Subsystem 4a) is intransitive.
So the schema appropriate to the intransitive MANDATORY ought
as seen in the desirability interpretation of (64) is –
( xxiv )ought ( PROP ) = THINK ( BIND ( KRC, HAPPEN-PROP
Note that BIND, RELEASE
in Subsystem 4a will be not three-place but two-place predicates and
that IMPLEMENT (attempt and accomplish HAPPEN) will be replaced by
HAPPEN. Another example of this ought is (69),
which is paraphrase of (70).
Your parents ought to be helped.
Someone ought to help your parents.
(71) will be an appropriate
paraphrase of (69) and (70) :
It is desirable that your parents be helped.
On the other hand, (72)
will paraphrase as (73) –
He ought to help your parents.
It is incumbent on him that he help your parents.
Going back to (69) and
(70), consider (74) –
(74) Your parents deserve to be helped.
(74) implies (69)-(71),
but not the other way round. The
reason probably is that (74) says all that (69)-(71) have to say (though
in a different format) and something more; (74) can be paraphrased
(75) Your parents being what they are, they ought
to be helped.
Note, incidentally that
while (72a) will be all right.
(72b) will be distinctly odd.
(72a) Your parents deserve to be helped by him.
(72b) He deserves to help your parents.
In short, Subsystem 4 has a version Subsystem 4a in
which even the constraint calling for an identifiable Agent is removed. If one considers how philosophers tend to draw
a sharp line between truth-claims and judgements of rightness, goodness,
beauty, expediency, legality, and the like this casualness about the
distinction between EPISTEMIC and MANDATORY modals and the freedom
to choose any PROPOSITUS s surprising if not scandalous.
Or maybe, turning the tables around, the philosopher’s insistence
on the sharp line between Subsystems 1-3 and Subsystems 4, 4Ia
is surprising. Even more
surprising is his insistence on separating, say, (1), (2), (5), (69),
(74) – all tame ‘moral’ judgements from, say, (64 in MANDATORY version)
and the group (61)-(63). What separates ‘moral’ judgements from other
judegements – including ‘immoral’ judgements like (63) and judgements
of truth—is, it should be apparent by now, not their ‘logical grammar’
but something else.
Actually, subsystems 4 and 4a supply philosophers with
a set of versatile tools that they have not fully exploited.
Let us go back to –
(23) What one can’t say one mustn’t speak.
Compare this with—
(23a) Wovon man nicht sprechen kann, daruber muss
(23b) Whereof one cannot
speak, thereof one must be silent.
(The accepted translation of 23a.)
What is the status of this injunction of silence?
Is it a moral injunction in the narrow sense? Or an injunction of some other order? Consider again the following :
(76) This sight is beautiful.
(76a) How beautiful this sight is !
(76b) I will say/I think that this sight is beautiful.
(76c) One can/ought to/must say that this sight is
An aesthetic judgement
is exemplified by (76c) and the corresponding interpretation
of (76) but not by (76a), (76b), and the corresponding
interpretations of (76). The
following two are aesthetic judgements of a somewhat different kind
(77) This sight deserves to be called beautiful.
(77a) This sight being what it is, one ought to say
that it is beautiful.
(62) The red ought to go here and the blue there.
A word about Moods and Logical Operators.
May one set up the following equivalences?
(BIND (BRC, anyone, BELIEVE-‘p implies q’)=(p truth-functionally
implies q) =
(it will be that if p, q)
(BIND (KRC, anyone, BELIEVE- ‘p implies q’)=(p entails q’=)
(it must be that
if p, q)
Moods and tenses
English sentences with modal verbs are sometimes claimed to
be tense-neutral. This is not quite true. English has only two tenses – past and nonpast.
There is no future tense in English.
In a modal sentence, past tense (like negation) can be inserted
at two places – in the MODAL and in the PROPOSITUS.
This is broadly true of the major Subsystems (i.e. 1, 2, 3,
4, 4a) : the actual facts of usage are rather messy.
(78) He will/would go/have gone for a walk.
(79) He won’t/wouldn’t confess his crime.
(79a) He can’t/couldn’t confess his crime.
(80) He can/could be/have been annoying sometimes.
(81) He doesn’t/didn’t need to go/have gone there.
I think/thought you ought to help/have helped your parents.
The performative force of the four minor Subsystems (i.e. 5,
6, 7, 8) preludes the use of past with the modal; the PROPOSITUS too
is always nonpast.
While, some languages undoubtedly have a future tense in their
formal system, can we say the same thing about their semantic system? Does the modal handling of the future in English typify a language
universal or does it typify a major cleavage among human languages?
Whatever answer linguists offer to this question is going to
be of interest to philosophers. Of course the modal substitute for future need
not be will; it can be go; the substitute need not be
even modal. In French je
donerai is literally “I have to give” (i.e. “I will give”).
There is no denying of course that English modals have tense
connotations. Consider –
(83) John may be obnoxious/tall.
(84) John can be obnoxious.
(‘tall’ will be odd here)
Englishmen may be obnoxious.
will be odd here)
(86) Englishmen can be obnoxious/tall.
(You will) remain seated.
the Addressee may or may not be seated at the time of saying.)
(88) The train may/will be in the station.
(The train may or may not be in the station at the time of
(89) The train must/can be in the station.
(The reference is to the certainty/possibility of the train
being in the station at the time of saying.)
(90) (You will) open your mouth/sit down.
(It will be odd to say this to someone whose mouth is already
open/who is already seated.)
(91) The train may/will have left the station.
(This is compatible equally with ‘by now’ and with ‘by that
(92) The train may/will leave the station.
(It will be odd to say this if the train is already out of
Vulnerability of Modals
Are English modals vulnerable?
More exactly, does English permit the Addressee to confront
a Statement containing a Modal with its contradictory by way of disputing
it? And further, is one permitted to refute a Modal Statement? Presumably, disputability is a weaker demand
Let us take disputability first.
Or rather let us take the linguistic provision for disputing
a statement. Usually this linguistic provision takes the
form of negation – but not always.
Thus (93) can be disputed not by saying (94) which is unavailable
in English (and hence starred) but by (95) which doesn’t look like
the counterpart of (93) with opposite polarity.
(93) The train didn’t leave until the Chief Minister
(94) *The train left until the Chief Minister
The train left before the Chief Minister boarded it.
Our earlier distinction between MODAL and PROPOSITUS negation
is relevant at this point.
The Subsystems may now be surveyed from this point of view. (Symbols : MN modal negation, PN propositus
negation, I incompatible and so can justify a ‘no, but’ retort, I*
just enough incompatible to justify or rather, I** just enough incompatible
to justify but, I*** just enough incompatible to justify but
or though, C compatible, A collectively exhaustive.)
(xxv) Subsystem 1
will-1, won’t are PN, I
will-1, may not
are MN, I, A
may, won’t are MN, I, A
may, may not are PN, C, I**
will-1, may are C, I*
won’t may not are C,I*
(This is analogously applicable to would, might.)
(xxvi) Subsystem 2
will-2, won’t are PN, I
will-2, can’t are I***
can, can’t are MN, I, A
can, won’t are I***
(will-2 more common with the negative.)
(xxvii) Subsystem 3
must/ought/should, mustn’t/oughtn’t/shouldn’t are PN,
must, needn’t are MN, I, A
can, can’t are MN, I, A.
must/ought/should, can’t are PN, I.
can, mustn’t/oughtn’t/shouldn’t are PN, I.
ought/should, needn’t are MN, I***, A
can, needn’t are PN, I**.
must, ought, should, can are C, I*
(xxviii) Epistemic modal adjectives
certain, uncertain are I, A
more than probable, improbable are I, A
probable, more than improbable are I, A
possible, impossible are I, A
other relationships can be worked out.
BELIEVE-PROP can be replaced by CONCEIVE-PROP yielding another
set of modal adjectives, inevitable, contingent, barely conceivable,
almost inconceivable, conceivable, inconceivable.
This other set may hold the key to the analytic-synthetic problem
(xxix) Subsystem 4, 4a
must/ought/should, mustn’t/oughtn’t/shouldn’t are PN,
must, needn’t are MN, I, A
may/can, may not/can’t are MN, I, A
must/ought/should, may not/can’t are PN, I
may/can, mustn’t/oughtn’t/shouldn’t are PN, I, A
(The other relationships can be worked out.)
(xxx) Subsystem 5
let, don’t let are MN,I, A
let/Imperative, let … not/don’t are PN, I
(xxxi) Subsystem 6
may/let, may/let … not are PN, I
let, let … not are PN, I
may, let … not are I, A
(Note that let has two senses : RELEASE and also BIND
Subsystems 5 and 6.)
(xxxii) Subsystem 7
shall, shan’t are PN, I
(xxxiii) Subsystem 8
will/Imperative, won’t/don’t are PN, I
let, let … not are
let, don’t let are MN, I
shall, shan’t are PN, I
(The other relationships can be worked out.)
For the epistemic subsystems 1 to 3 English operates with a
five-valued scale which perhaps could be symbolized as +1, 0, -1 and
two intermediate values. For the mandatory subsystems 4 and 4a,
a four-valued scale operates with +1, 0, -1 and a value intermediate
between +1 and 0. Subsystems
5 and 6 operate with a three-valued scale +1, 0, -1.
subsystems 7 and 8 operate with a two-valued scale +1, -1,
or perhaps a three-valued one with an intermediate value.
All this is disgustingly messy no doubt to formal logicians.
Also, this seems to render hopes for a logic universal to all ordinary
languages rather murky.
How do we fare if we pass on from questions of disputability
to those of refutability?
In subsystem, I, will and won’t are
refutable if the PROPOSITUS has a built-in time-bar or condition;
may and may not are not refutable; would and
might and their negative counterparts raise special problems
because of the explicit or implicit counter-factive condition.
In Subsystem 2, won’t and can’t are refutable,
will and can are refutable if the PROPOSITUS has a built-in
time-bar; incidentally only can is provable out of the four.
The other four would, wouldn’t, could, couldn’t
raise special problems.
In Subsystem 3, must and mustn’t/can’t are refutable
if the PROPOSITUS has a time-bar or some condition; the rest are not
Question of refutability in Subsystem 4 and 4a are often
tied up with similar questions regarding Subsystem 2.
Take the familiar dictum – ought implies can,
for example. The dictum can be meaningfully discussed I
non-ethical domains also : thus, one can argue that
(76d) One can say one ought
to find this sight beautiful only if one can say one can find it beautiful.
(ought (to find) from Subsystem 4; can (find)
from Subsystem 2; and of course can (say) from subsystem
Note, incidentally, that while (76d) seems to be a reasonable
claim, (76e) doesn’t which puts the can condition inside
the propositus of which ought is predicated.
(76e) One ought to find this sight beautiful only
if one can find it beautiful.
(76f) Perhaps one can say –
not-can P implies not-ought P
ought-P presupposes can P
Consider Gombay 1964: ‘You ought to feed
the cat’ commits one to ‘You can feed the cat’, but ‘you ought to
feed the cat’ does not support ‘you can feed the cat’.
The relation of mandatory ought with the mandatory must
and should on the one hand and with the mandatory shall
(Subsystem 7) and will (Subsystem 8) on the other needs to
be investigated in the context of ethics.
We have already identified the element of THINK that marks
ought and should off from must. However, some ethical theories seem to use
ought as a disguised must or a disguised shall.
Either the disguise (and the caution implied by THINK) be given up
or the ethical ought be used with its face value.
One way of showing that ought implies can is
to show that ought implies will-Imperative.
The logic of Imperatives can be presented in some such terms.
(Cf. Gordon & LakoFF 1971.)
(xxxiv) For A say to B ‘(You will) do X’ is felicitous
if and only if
A believes that B can do X or refrain from doing X (i.e. that B is
a potential Agent for doing X).
A wants that B do X.
A expects that B will do X if the Imperative is addressed to B and
A does not expect that B will do X anyway.
The following Imperatives are infelicitous in that they violate
one or the other of these felicity conditions.
This normally induces the Addressee to look for other interpretations.
Open your mouth. (To
someone whose mouth is open; cf. Keep your mouth open.)
See this clearly. (cf. Look
at this carefully.)
Be tall. (cf. Be healthy.)
Be funny. (cf. Don’t be funny.)
Get well soon. (cf. its
interpretation as a disguised wish).
Leave this place and don’t leave this place.
Go ahead and ruin your health. (cf.
its interpretation as a reproach disguised by irony.)
Go home, will you? – Where do you think I am going?
Go home, will you? -- Fat
Go home, will you? – But I have already decided to go home.
Questions of refutability in Subsystem 2 are connected with
similar questions in Subsystem 3 – in the context of the philosophy
of mind one can discuss, for example, whether will-2 (disposition)
implies must epistemic (need).
The notion of refutability when applied to performatives (as
the modals of Subsystems 5 to 8 are) takes on the form of the notion
of being liable to be exposed as infelicitous – defelicitizability,
if you can tolerate such a monster of a term.
The other side of the coin of disputing and refuting is justifying
a disputed claim and successfully defending it.
We have dwelt at length on vulnerability the better to throw
light on justifying and defending BELIEVE-PROP or IMPLEMENT-PROP or
HAPPEN-PROP. Justifications are of course quite different
from explanations. We justify
our acts including presumably our acts of belief.
We explain or account for facts including presumably our acts. Our acts include our speech acts – explicating
an expression is part of the explanation for the speech act concerned.
We have sought so far to explicate expressions containing modal
verbs. Have we in the process explained why such expressions
are used? If our explications
are correct, we have gone a long way towards explaining their use
– but not all the way.
Some aspects of these explanations, to be sure, are going to
belong to historical linguistics – showing the links between present-day
English modal verbs and their Old English ancestors, for example –
and thus going to be of no great interest to philosophers.
But there is no doubt that the remaining, functional aspects
of these explanations are gong to be of interest to philosophers.
Have I kept my promise to do in the first part linguistics
that is philosophically interesting without constituting philosophizing
about language? And have I have kept my promise to do in the
second part philosophizing that is linguistically grounded without
being linguistics plain and simple?
And, further, whether I have observed the rules of the game
or not, have I succeeded in saying something that will induce you
to do your own digging?
All that naturally is your privilege to decide.
Anderson, J.M. 1971. Some proposals concerning the modal verb
in English. In : McIntosh, Angus; P’alsson, H. (ed.). Edinburgh
Studies in English. London: Longmans.
Antinucci, Francesco; Parisi, Domenico.
1971. On English modal verbs. Papers from the 7th
Regional Meeting, Chicago Linguistic Society.
Chicago: Chicago Linguistic Society.
Binnick, Robert I. 1971-2.
Will and be going to. Papers
from the 7th Regional Meeting Chicago Linguistic Society.
Chicago : Chicago Linguistic Society, 1971. Pp. 40-52; Papers from the 8th Regional Meeting…1972.
Boyd, J.; Thorne, J.P. 1969. The Deep grammar of modal verbs. Journal of linguistics 5. 57-74
Bruton, J.G. 1965. Exercises
on the English verb for intermediate students.
London : Cambridge University Press.
Caton, Charles, E. 1966. On
the general structure of the epistemic qualifications of things said
in English. Foundations of language 2. 37-66.
Caton, Charles E. 1969. Epistemic qualifiers and English grammar. In
: Todd, William (ed.) studies in philosophical linguistics. Evanston,
Illinois : Great Expectations, 1969.
Chaplin, Paul G. 1973. Quasi-modals. Journal of
linguistics 9. 1-9.
Diver, William. 1964. The Modal system of the English verb. Word 20. 322-59.
Ehrman, Madeline Elizabeth.
1965. The Meanings of the modals in present-day American
English. The Hague : Mouton. Summary in : Linguistics no.28, 43-72.
Gombay, André, 1964. Types
of implication. The Journal of Philosophy 61:
Gordon, David; Lakoff, George.
1971. Conversational Postulates. Papers from the
7th Regional Meeting Chicago Linguistic Society.
Chicago : Chicago Linguistic Society, 1971.
Grady, Michael. 1970. Syntax
and semantics of the English verb phrase.
The Hague : Mouton, 1970.
Hakutani, Yoshinobu. 1972.
English modal auxiliaries. Linguistics.
Hakutani, Yoshinobu; Hargis Charles H. 1972. The Syntax of
modal constructions in English. Lingua
Halliday, M.A.K. 1969. On
fitness and modality in the English verb.
London: Dept. of General Linguistics, University of London.
Hall, Barpara. 1964. The
Auxiliary in English sentences with “ing”.
Cambridge, Mass : MIT. Duplicated (July).
Halliday, M.A.K. 1970. Functional diversity in language as seen from a consideration of
modality and mood in English. Foundations
of Language 6. 322-61.
Huang, Shuan-fan. 1969. On
the syntax and semantics of English modals.
Working papers in linguistics, Ohio State University 3. 159-81.
Jackendoff, Ray S. 1968. Speculations on presentences and determiners. Bloomington, Indiana
: Indian University Linguistics Club.
Duplicated. Also :
International journal of Dravidian linguistics 1 : 1.
Jackendoff, Ray S. 1972. Semantic interpretation in generative grammar. Cambridge, Mass : The MIT Press.
Jenkins, Lyle. 1972. Will-deletion. Papers in
the 8th Regional Meeting Chicago Linguistic society. Chicago : Chicago Linguistic society. Pp. 173-82.
Jespersen, Otto. 1932. Modern English grammar on historical principles, Part IV. London : Allen and Unwin.
Jespersen, Otto. 1933. Essentials of English grammar.
London : Allen and Unwin.
Joos, Martin. 1964.
The English verb : Form and meaning.
Madiśon, Wisconsin : The University of Wisconsin Press.
Robin. 1972 a. Language
in context. Language 48. 907-27.
Robin. 1972 b. The Pragmatics of modality. Duplicated.
Yvan. 1965. Can and may in
present-day English. Bruxelles,
Belgium : Press Universities de Bruxelles.
GeoFFrey N. 1971. Meaning and the English verb. London : Longman. Ch. 4 The Expression of Future time. Ch.5 The Modal auxiliaries
Ch.5 The Theoretical and hypothetical meaning.
Frank R. 1965. A Linguistic
study of the English verb. London
Perlmutter David. 1971. Deep and surface structure constraints in syntax. New York : Holt, Rinehart and Winston. (MIT, Ph.D. diss., 1968.)
Reich, Peter A. 1971. The English auxiliaries : A Relational
network description. Canadian
journal of linguistics 16. 18-50.
John Robert. 1967. Auxiliaries as main verbs. Cambridge, Massachusetts : MIT. Duplicated.
Also in : Todd, William (ed.).
studies in philosophical linguistics.
Evanston, Illinois : Great expectations, 1969. Pp. 77-102.
Sonnenschein, E. A. 1921-4.
A New English grammar. 3 parts.
Oxford : Clarendon Press.
Thorne, J.P. 1966. English
imperative sentences. Journal
of linguistics 2. 69-78.
Ultan, Russel. 1972. The Nature of future tenses. Working
papers on languages universals, Stanford University.
Vendler, Zeno. 1967. Facts and events. In his
: Linguistics in philosophy, Ithac,
N.Y. : Cornell University Press.
Portions of this paper were presented before the Poona Philosophy
Union, S.P. College, Poona on 19 December, 1973 and the Linguistic
Club of Poona, Deccan College, Poona on 9 January, 1974.
Centre of Advanced Study
in Linguistics, Deccan
College Ashok R. Kelkar
Poona 411006, India.
Indian Philosophical Quarterly
welcomes papers in all areas of Philosophy, History’ of Philosophy
and Philosophy of Indian Origin.
It is interested in persistent, resolute inquiries into basic
questions regardless of writers’ affiliations.