Ought and Must:

Some Philosophical Terapy

Alex Silk
a.silk,bham.ac.uk
Draf of September io1
Ought and Must they are contemptible auxiliaries.
George Eliot

Abstract
Ought is the philosophers normative term par excellence. But there has
been little inquiry on what ought, in contradistinction from other related
normative terms and necessity modals, actually means. I oer an analysis of
the meaning of ought, and of the distinction between ought and must, that
makes correct predictions concerning a wide range of semantic and pragmatic
phenomena. Ought, on this view, expresses a kind of conditional necessity,
necessity on the supposition that the applicability conditions of certain con-
siderations (norms, values, goals, etc.) are satisfed. Clarifying the distinction
between ought and must is not merely of linguistic interest; it can improve
theorizing on broader philosophical issues.

Tanks to MatthewChrisman, BrendanDill, JanDowell, Irene Heim, Ezra Keshet, DanLassiter,

Paul Portner, Bernhard Salow, Robert Shanklin, Bob Stalnaker, Eric Swanson, and audiences at SALT
ii, MIT, and Northwestern University for helpful discussion. Apreliminary version of this paper was
published in Anca Chereches (Ed.), Semantics and Linguistic Teory ::, o.

Mary Garth, in Middlemarch, Bk. i, Ch. 1. Shamelessly modifed from the original.
Contents
Introduction
Weak necessity is conditional necessity
i.1 Ought and must in context . . . . . . . . . . . . . . . . . . . . . . . i
i.i Te analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,
i. Counterfactual marking . . . . . . . . . . . . . . . . . . . . . . . . . 1i
i. Entailingness and performativity . . . . . . . . . . . . . . . . . . . . io
i., Negotiability and collective commitment . . . . . . . . . . . . . . . i
A little philosophical therapy
.1 Moral dilemmas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i,
.i Supererogation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . o
Introduction
Ought is the philosophers normative term par excellence. A primitive ought, it
has even been said, is the basic conceptual atomthat gives normative concepts their
special character (Ginn.vu iooo: ,8). But there has been little inquiry on what
ought itself means. Ought, when it is investigated at all, is ofen treated merely as
a representative of normative terms more generally.
Tis is a problem. Not all normative terms are created equal. Exhibit A: Intu-
itively, Ought and Should express that is necessary in some sense. Yet it
is well established that though ought and should (weak necessity modals) are
stronger than possibility modals like may or can, they are nevertheless weaker
than modals like must and have to (strong necessity modals).
1
Even holding
the reading of the modals fxed, Ought can be followed by Must but not vice
versa, as in (1); and Ought , but not must is consistent in a way that Must ,
but not ought is not, as in (i).
(1) a. I ought to (/should) help the poor. In fact, I must.
b. I must help the poor. -In fact, I ought to (/should).
(i) a. I ought to (/should) help the poor, but I dont have to.
b. -I must help the poor, but its not as if I ought to (/should).
What is not well established is howto capture this dierence in strength. Tis might
appear to be a narrowly linguistic problem. Leave it to the semanticist, one might
say. But such a judgment would be premature. Insensitivity to dierences among
necessity modals has obscured theorizing on broader philosophical issues.
Tis paper motivates and develops a novel account of the meaning of ought
and the distinction between weak and strong necessity modals (i). Despite a fairly
robust body of data on dierences between weak and strong necessity modals, there
has been nearly no sustained theoretical investigation of how to capture them. A
primary goal of the present paper is provide such an investigation. Te resulting
account integrates a wide range of linguistic data, some of which is entirely original
to the discussion here. It generalizes across favors of modality, elucidates a special
role that ought judgments play in conversation, deliberation, and planning, cap-
1
See, among many others, Siom. 1,o; Liicu 1,o, 1,1; Hov 1,i, 18; Wiv1uiimiv
1,i; H.vm. 1,,, 1; Lvos 1,,; Woisi1scui.iciv 1,,; Wiiii.ms 181b; Co.1is 18;
Jois & Pov 18o; Wiivznicx. 18,; McN.m.v. 1o, 1o; P.imiv 1o, ioo1; Bvnii i1 .i.
1; Gvoiisim. 1,; Mvuiii 1,; Zimmivm. 1o; HUuuiis1o & PUiiUm iooi; Wiviv
ioo, ioo,; Coviiv iooo; W.vsnv iooo; vo Fi1ii & I.1viuoU ioo8.
1
tures contrasting discourse properties of ought and must, and sheds light on how
weak necessity is expressed in other languages. Te project is not to argue that no
other theory can get the data right though ways in which my account improves
upon its rivals will be indicated as points of contrast are made salient. Rather it is to
motivate an account of the nature of weak necessity that is empirically adequate and
theoretically attractive. Finally, I argue that investigation of these linguistic mat-
ters has broader philosophical import (). Clarifying dierences among necessity
modals can help us diagnose problems with arguments concerning further philo-
sophical issues. I will focus on two such issues in normative theory: the possibility
of moral dilemmas and supererogatory acts. Tese diagnoses, I hope, will motivate
clearer answers and suggest new ways the dialectics may proceed.
(For expository purposes I will focus primarily on ought and must. Unless
otherwise noted I bracket dierences among weak necessity modals and dierences
among strong necessity modals.
i
Speakers who fnd should to be more natural
than ought may feel free to substitute should for ought in examples throughout
the paper.)
Weak necessity is conditional necessity
. Ought and must in context
It is well appreciated that a primary goal of conversation is to share information. We
express our beliefs about the world and invite others to share in these beliefs. But
we also express other aspects of our state of mind. We coordinate our expectations,
values, goals, and plans. Sometimes we assert propositions outright. We commit
to settling on the truth of our assertion for the remainder of the conversation.

But
sometimes we dont wish to impose such a strong restriction on the future course of
the conversation. I may want to propose that someone is obligated to do something
but be unsure about whether there might be competing norms at play that would
outweighor cancel her obligation. Or we may want to proceedas if some proposition
i
I address certain such dierences in i.. For further discussion, see Liicu 1,o: ii8, Liicu
ioo; L.xoii 1,ib; Wiv1uiimiv 1,i: ,,8o; BoUm. 1,,: i1io, ; L.vxi 1,o; Co.1is
18: ,,,, o,o, ,,8; P.imiv 1o: 11, 8, o,o, 1oo1o,, 1111o, 1,,, 18818, ioo1:
,, ,,; Coiiis 11; Bvnii i1 .i. 1: i1i18; Mvuiii 1,, 1o, 1,; HUuuiis1o &
PUiiUm iooi: 18o, 1oio8, 8, 1oo11ooi; Smi1u ioo; Liicu i1 .i. ioo; Ciosi & A.v1s
io1o; I.1viuoU & Zii,is1v. io1i.

As usual, I assume as an idealization a purely monotonic model of information gathering, i.e.,

that information is only added and, once added, is not subject to further discussion.
i
is true because it follows from our evidence while remaining open to the possibility
that our evidence is misleading. I suggest that it is the role of weak necessity modals
to allow us to carry out such goals, make such proposals, and express such states of
mind.
First consider the deontic case. Suppose I am considering whether to fght in
the Resistance or take care of my ailing mother. I mention that the value of family,
whichsupports my helping my mother over my fghting, is important, andyouagree.
But the issue is acknowledged to be complex, and it isnt settled in the conversation
whether there might be more important competing values. Sensitive to this, you
may fnd it more appropriate to express your advice that I help my mother by using
ought than by using must, as in ().

() Me: Family is very important.

You: I agree. You ought to (/
?
must, /
?
have to) take care of your mother.
But if we come to settle that family is of primary importance, as in (), it can become
more natural to use must and for us to accept that I have to help my mother.
() Me: Family is most important more important than country.
You: I agree. You must (/have to, /
?
ought to) take care of your mother.
Tis kind of case highlights a crucial dierence between weak necessity modals
like ought and strong necessity modals like must. In both () and () the value
of family is accepted in the common ground (what is mutually presupposed for the
purposes of the conversation; S1.i.xiv 1,, iooi).
,
How you can express your
advice that I help my mother depends on how the status of the value of family may
relate to other values in the context. In (), where it is common ground that the
value of family is to take precedence, using a strong necessity modal is preferred.
But in (), where this condition isnt settled, were you to use must you would imply
that you are foreclosing certain possibilities that I have lef open. Unless you have
the authority to do so, your using must is dispreferred. What is illuminating is that
you can felicitously express your advice that I help my mother using ought, advice
which I may accept, even if it isnt common ground that the above precondition for
my having a genuine obligation to help my mother is satisfed. By using ought you
can propose that I help my mother while remaining neutral on whether the value

I use : to indicate that using the marked itemis dispreferred. Tus : marks a weaker infelicity
than -.
,
Tis is contrary to suggestions by Aynat Rubinstein (io11: 11, 1o; io1ia: ,,oo) regarding
similar examples.

of family might be outweighed or defeated in some way. If I agree, we can plan

accordingly without having to presuppose that the value of family is more important
than other competing values we accept or may come to accept.
Similar points hold in the epistemic domain. Suppose we are working on a class
art project together. I ask you where the colored pencils are. Normally you put
them in the drawer with the crayons, but sometimes you accidentally put them on
the shelf. In this scenario it is more appropriate for you to use ought in responding
to my question, as in (,).
(,) Me: Where are the colored pencils:
You: Tey ought to (/
?
must) be in the drawer with the crayons.
Suppose, alternatively, that we are looking for the colored pencils together, and you
indicate that youhave just seensomething that leads youtoconclude that the colored
pencils are in fact in the drawer. Perhaps you noticed that they werent on the shelf,
and that is the only other place you think they could be. In this scenario it is more
natural for you to use must, as in (o).
(o) Me: Where are the colored pencils:
You: Tey must (/
?
ought to) be in the drawer with the crayons.
Te epistemic necessity of the colored pencils being in the drawer depends on the
fact that conditions are normal in the relevant respects, i.e., that this isnt one of the
atypical days when you accidentally put the colored pencils on the shelf. Using the
strong necessity modal must is preferred if, and only if, you know that conditions
are indeednormal inthis way. Parallel to the deontic case, what is illuminating is that
you can use ought even if you arent in a position to judge whether this precondition
for the truth of the epistemic necessity claim is satisfed. By using ought you can
propose that I proceed as if the colored pencils are in the drawer without having to
presuppose that your evidence is indefeasible.
I want to emphasize three points from these examples. First, ought does not
express necessity. In () we can accept your deontic ought claim without settling
that family is the most important relevant value, and thus without accepting that it
is deontically necessary that I help my mother. In (,) we can accept your epistemic
ought claim without settling that conditions are normal in the relevant respects,
and thus without accepting that it is epistemically necessary that the colored pencils
are in the drawer. Second, whether ought or must is preferred depends on certain
features of context specifcally, on whether such preconditions for the truth of
the necessity claim are accepted. If they are, must is preferred. But even if they are

not, we can still use ought. Tese points have not been noticed in the literature.
o
Tird, these features of ought and must are not specifc to any particular readings
of the modals. An adequate account of the weak/strong necessity distinction must
generalize across favors of modality. Capturing these points will be crucial for the
positive account that follows.
. e analysis
We can make progress in capturing the data from i.1 with a more nuanced un-
derstanding of the considerations that fgure in the interpretation of modals. Val-
ues, norms, goals, preferences, expectations, etc. ofen arent categorical; they arent
usually of the form ^o matter what, !. Rather they ofen come with what I will call
applicability conditions (ACs), or conditions under which they apply. If I want to
go for a run, my desire need not be that I go for a run, come what may. More plau-
sibly, it is that I go for a run given that its sunny, that I dont prefer to do anything
else more, and so on. Our preferences are ofen conditional, preferences for certain
circumstances. Similarly with moral norms. Suppose you promised Alice that you
would help her move. Anormagainst breaking your promise might be something to
the eect that you help Alice unless you made a more important conficting promise
to Bert, or keeping your promise would lead to some serious harm, orNorms can
thus be understood on the model of conditional imperatives, imperatives that en-
join an action or state of aairs given that certain circumstances obtain. Likewise
for goals, probabilities, and so on. Tis captures the intuitive idea that depending
on the circumstances i.e., depending on which applicability conditions are satis-
fed only certain norms, etc. may apply or be in force. Fixing terminology, lets
call the content of a conditional norm, preference, expectation, etc. a consideration.
Given a consideration If C, , let C be the considerations applicability condition, and
be the considerations premise, or what the consideration enjoins given C. (Cate-
o
Even the general point that the relative felicity of ought and must depends on certain features
of context has been largely overlooked in previous accounts. Te semantics for ought in Fii.v
ioo, io1o, L.ssi1iv io11, and Sw.so io11, for instance, have no obvious mechanism for cap-
turing the sort of context-dependence in these examples. vo Fi1ii & I.1viuoU (ioo8) briefy
mention a possible reference to context implicit the interpretation of ought claims (pp. 11o),
but the issue is not considered in any depth (see i. for discussion of their account). A notable
exception is recent work by Aynat Rubinstein (esp. her io1ib), though she has a dierent take on the
data (see i., for discussion, also my io1i, io1c). See also Woisi1scui.iciv 1,,: ch. , and Mc-
N.m.v. 1o: ch. for prescient early discussion of the context-dependence of ought and must.
See RUnis1ii io1ib: i.i for a similar critique of previous comparative and domain restriction
approaches to weak necessity, along with extensive broader critical discussion.
,
gorical considerations can be treated as conditional on the tautology.)
Tese points motivate an attractive interpretation of the standard premise se-
mantic framework for modals. Modals are treated as receiving their reading or in-
terpretation from a contextually supplied set of premises.
,
Since modals can occur
in intensional contexts, it is standard to index premise sets to a world of evaluation.
Which premise set is relevant for the evaluation of a given modal sentence can de-
pend on how things happen to be in the actual world, or on how things could be
but arent or could have been but werent. What Little Timmys parents command
might change from one world to the next; thus the meaning of a phrase like what
Little Timmys parents command that determines the intended reading for must
in (,) can be treated as a function that assigns to every possible world the set of
propositions describing the house rules in that world.
(,) In light of what Little Timmys parents command, Little Timmy must be in
bed by eight.
Similar remarks hold for the meanings of phrases like in view of the relevant cir-
cumstances, according to U.S. law, and so on. It is these functions that context
is said to supply for the interpretation of modals. Call these functions unsaturated
premise sets (written P). Call the value of an unsaturated premise set given a world
of evaluation a saturated premise set, or simply a premise set (written P
w
).
We can capture the role of applicability conditions in terms of variability in the
values of unsaturated premise sets at dierent worlds. Suppose we have a consid-
eration If C, which enjoins given that conditions C obtain. We can represent
the content of this consideration, intuitively, with an unsaturated premise set P that
assigns to every relevant C-world a premise set that includes . For example, the
normative import of your promise to help Alice move, mentioned above, would be
refected in Ps assigning a premise set that includes the proposition that you help
Alice to worlds in which you made this promise, you didnt make a more important
promise to Bert, etc. Te premises in a saturated premise set thus refect what is
enjoined by a body of considerations e.g., conditional norms, preferences, expec-
tations given the circumstances that obtain in the evaluation world.
8
,
Equivalently, a preorder (Liwis 181). See especially Kv.1ziv 1,,, 181, 11; also
v. Fv..ssi 1,, Liwis 1,, Vii1m. 1,o.
8
In my io1i: ,, and io1b I argue that thinking of premise semantics in this way can help
solve the problem of capturing weights and priorities among premises in premise sets. An alterna-
tive is to treat ACs as built into the specifcation of individual premises i.e., so that premises are
propositions expressed by material conditionals , where describes the AC of the action
or state of aairs described by (Siix io1i). However, this strategy founders in the case of the
o
Tese conceptual points about the premises relative to which modal claims are
interpreted can inform our semantics for weak and strong necessity modals. I give
strong necessity modals like must their usual semantics of necessity. Bracketing
some complications that wont be relevant here, Must says that the prejacent
proposition follows from P
w
where P
w
is the premise set that is the value of
the given unsaturated premise set P at the evaluation world w.

Formally:
.c1U.i s1voc
Must is true at w i P
w

Te truth of Must thus depends on the value of P at the world of evaluation. In
order for Must to be true at w, given an unsaturated premise set P, the circum-
stances in w must be such that the premise set P
w
verifes the necessity of .
Te same doesnt hold for ought. As we saw in i.1, Ought doesnt say that
is necessary. Rather, I suggest that what makes weak necessity modals weak is
that they express a kind of conditional necessity, necessity on the supposition that
certain conditions the applicability conditions of certain relevant premises are
satisfed.
1o
Roughly, Ought is true i Must would be true were it to turn out
that certain relevant norms, values, goals, etc. that bear on the necessity of are in
force and not defeated. Ought makes a claim about the necessity of at all clos-
est relevant -worlds, for some contextually supplied condition . More formally,
where s is a selection function that selects a set of closest -worlds to the evaluation
world w (this will be refned in i.):
coui1io.i wi.x
Ought is true at w i Must is true at all worlds w

s(w, ) i w

s(w, ) P
w

interpretation of possibility modals like may (Siix io1b).

For simplicity I assume that our premise sets are consistent, and I do not distinguish between
Kratzers modal bases and ordering sources. Dropping these assumptions, the simplifed talk in the
main text about what follows from P
w
can be understood as short for talk about what follows from
all maximally consistent subsets of F
w
G
w
that include F
w
as a subset, where F
w
is a modal base
that describes some set of relevant background facts and G
w
is an ordering source that represents the
content of a relevant ideal at the evaluationworld. (Tis still makes the limit assumption(Liwis 1,:
1io), the assumption that ordering consistent subsets of F
w
G
w
that include F
w
by set inclusion
yields a set of subsets that are -maximal. For semantics without the limit assumption, see Liwis
1,; Kv.1ziv 181, 11; Sw.so io11.)
1o
For precedents in descriptive linguistics, see P.imiv 1o: ,oo, 1i,, 1i8; ioo1: i, ,,.
,
In making an ought claim one need not commit to being in a world w such that P
w
verifes the necessity of .
A brief word about the implicit condition that fgures in the interpretation
of ought may be helpful (see also i.). Using ought allows us to coordinate on
somethings being necessary in the relevant sense without having to settle that all
the preconditions for this necessity are satisfed. For example, in the deontic case,
by using ought we can proceed as if an action is required without having to commit
that all the norms that render that action required apply and are not defeated. Given
this role for ought in an utterance of Ought , and given plausible conversational
maxims especially, that given a set of alternative propositions a speaker asserts
the proposition she thinks is most likely to remain accepted throughout the evolu-
tion of the conversation will be some condition such that there is no alternative
condition with respect to which Ought would be true that is (presumed to be) at
least as likely or desirable as , for contraries and . Tere may be independent
reasons for building this subjective element into the semantics for ought, but this
complication is not necessary for present purposes. We can derive it pragmatically.
Upon hearing an utterance of Ought a hearer might reason as follows:
(8) S said Ought . So S must think there is some relevant condition C given
whichMust would be true. If S thought there was another more probable
or desirable condition C

givenwhich Must would be true (for contraries

and ), she would think it more likely or desirable that Must is true
and accepted than Must . So S would have said Ought , assuming that
she is being cooperative and obeying the maxims. But S didnt say Ought
. So she must think C is not less likely or desirable than any relevant
alternative condition with respect to which Ought would be true.
11
Note that the truth conditions for ought and must given above apply for any
11
Tis account of howthe value of is derived in context may receive independent support from
work on implicit conditionals. For instance, in ScuUiiiv io11 likely the most in-depth investiga-
tion of implicit conditionals to date the implicit antecedent is treated as a condition that expresses
the most plausible alterations to current contextual knowledge which would make the content of the
[implicit conditional] (interpreted as a consequent in the paraphrase) relevant to the context of utter-
ance (8, also o,o; cf. K.sviv 1i). Insofar as weak necessity modals, on my view, are treated as
having the semantics of an implicit conditional (specifcally, a sentence-external implicit conditional,
i.e. one whose implicit condition is fully supplied by the extra-linguistic context), the treatment of
in the main text may be understood as an account of what relevance to the context amounts
to in the case of weak necessity modal claims. Tough see i. for discussion of several important
dierences between ought claims and (possibly implicit) would have to conditionals.
8
readings of the modals (deontic, epistemic, teleological, etc.) Our interpretation
of the standard premise semantic framework did not turn on features specifc to
particular types of premise sets. Te truth-conditional dierence between ought
and must refected in .c1U.i s1voc and coui1io.i wi.x is certainly not the
only dierence between weak and strong necessity modals, a point we will return
to in i. (see also n. i). However, as we will see in the remainder of this section,
this account of the weak/strong necessity distinction can go a long way in explaining
various properties of weak and strong necessity modals.
Treating weak necessity modals as expressing conditional necessity helps cap-
ture how they are well suited to play the sort of role in conversation, thought, and
planning discussed in i.1. In saying Must , you assume a body of relevant cir-
cumstances and considerations, and assert that follows fromwhat these consider-
ations enjoin given those circumstances. So, in order for Must to be accepted in
a context, and thus true at all worlds in the context set (the set of live possibilities,
the set of worlds compatible with the common ground), the context set must be re-
stricted to worlds in which the relevant applicability conditions are satisfed, i.e., to
worlds w such that P
w
entails . But sometimes placing such a strong constraint on
the development of the conversation would be inapt. Perhaps we want to coordi-
nate on the necessity of without having to restrict the future development of the
conversation in such a way that decisively settles on the truth of certain ACs, or that
presupposes that certain relevant considerations apply and are not defeated. We can
do so using a weak necessity modal. In uttering Ought one neednt assume that
the circumstances in the world are such that what the relevant considerations enjoin
given those circumstances entails ; one neednt commit to being in a world w such
that P
w
entails . Rather, one expresses ones commitment to the necessity of on
some implicit condition or other, and invites ones interlocutors to test whether this
same property holds of their states of mind. If it does, then the group can proceed as
if is necessary without having to presuppose that there arent any more important
competing considerations and while remaining open to new evidence.
Lets fll this in by reconsidering our examples ()(o) fromi.1. Start with ()
(), reproduced below.
() Me: Family is very important.
You: I agree. You ought to (/
?
must, /
?
have to) take care of your mother.
() Me: Family is most important more important than country.
You: I agree. You must (/have to, /
?
ought to) take care of your mother.
Te deontic necessity of my helping my mother depends on the value of family be-

ing more important in my situation than other potentially competing values. In

(), when my assertion is accepted this condition that the value of family is to take
precedence becomes common ground; the ACfor my helping my mother is satisfed
throughout the context set. So, the context set gets restricted to worlds w such that
P
w
includes the proposition that I help my mother, and your must claim is true
throughout the context set. Since using must is felicitous in this context, using the
weaker ought would violate a Gricean quantity maxim and so is dispreferred. (I
return to the precise sense in which ought is weaker in i..) In (), by contrast,
afer my assertion is accepted it still isnt common ground that the AC for my help-
ing my mother is satisfed hence the hesitancy one might feel in using must. But
since Ought can be accepted in a conversation even if not all worlds win the con-
text set are such that P
w
entails , your ought claimis still felicitous and acceptable.
By accepting your ought claim, we can provisionally proceed as if my helping my
mother is required without needing to settle that the value of family isnt outweighed
or defeated.
I have said that ought is preferred in (), but it is important to see that in cer-
tain contexts must may also be acceptable. A speaker may felicitously use must
in a context in which it isnt common ground whether certain relevant applicability
conditions are satisfed if she can be presumed to be a legitimate authority on them
and can thus make them common ground by her utterance. Tis is a familiar kind
of accommodation (Liwis 1,b, S1.i.xiv iooi). In (), if you can be presumed
a normative authority on the issue in question, I will accommodate in response to
your using must by accepting the assumption required for your must claim to be
true, namely, that we are in a world w such that P
w
includes the proposition that I
help my mother.
Tis is not an isolated example. Suppose Alice, a young teenager, wants to go to
Harlem and is considering with her mother, Martha, whether to take the A train or
the Ctrain. Te Atrain is quicker, but the Ctrain is safer. Martha takes the safety of
her child to be of paramount importance, though this value is not common ground
between them. Nevertheless Martha can felicitously say to Alice:
() You must take the C train, not the A train.
By uttering () Martha expresses her acceptance that the goal of traveling safely takes
priority over the goal of traveling quickly. Given her authority in the context, she can
legitimately expect Alice to share this acceptance and accommodate. Tis renders
it common ground that the applicability condition of the former but not the latter
goal is satisfed, and restricts the context set to worlds that verify the necessity of
1o
. But such special contexts notwithstanding contexts in which the speaker does
not have or does not wish to exercise this authority ought will be preferred.
Like with norms and goals, what is expected given certain evidence can be con-
ditional, conditional on things being normal in the relevant respects. In the case
of (,)(o), modifed below as (1o)(11), let w
^
be a world in which conditions are
normal and you put the colored pencils in the drawer, and w
^
be an abnormal world
in which you dont put the colored pencils in the drawer. (w
^
and w
^
may be un-
derstood as representatives of suitable equivalence classes of worlds.)
(1o) Me: Where are the colored pencils:
You: Tey ought to be in the drawer with the crayons.
(11) Me: Where are the colored pencils:
You: Tey must be in the drawer with the crayons.
Te import of the conditional expectation concerning the location of the colored
pencils can be refected in an unsaturated premise set Ps assigning w
^
a premise set
that includes the proposition Dthat the colored pencils are in the drawer, and its as-
signing w
^
a premise set that includes D. Relative to this unsaturated premise set,
D is only an epistemic necessity at worlds like w
^
in which conditions are normal.
So in order for D to be accepted as epistemically necessary, the context set must be
restricted to normal worlds like w
^
. As in the case of Martha and Alice in (), your
reply in (11) is felicitous only if you have the epistemic authority to restrict the con-
text set in this way and make it common ground by your utterance that conditions
are normal. If I am unsure whether you are in a position to judge that they are, I
may challenge your assumption and raise to salience certain live possibilities that
are incompatible with the epistemic necessity of D, as in (1i).
(1i) You: Te colored pencils must be in the drawer with the crayons.
Me: Really: I see that they arent on the shelf. But dont you sometimes
accidentally put them in the cabinet with the glue sticks:
You: No, I never put them there. (/Oh, I forgot about that.)
But if you use ought, my mentioning such possibilities, as in (1), may be beside
the point, for you are not committing to conditions being normal. You are making
an implicitly conditional claim about what would be epistemically necessary were
it to turn out that conditions are normal. If I check the drawer and fnd no colored
pencils, you can respond as in (1).
(1) You: Te colored pencils ought to be in the drawer with the crayons.
11
Me: -Really: I see that they arent on the shelf. But dont you sometimes
accidentally put them in the cabinet with the glue sticks:
You: I know; thats why I said ought!
(1) You: Te colored pencils ought to be in the drawer with the crayons.
Me: No, I checked and they arent there.
You: Oh, then Im not sure where they are. Tats where I normally put
them. I would have expected them to be there.
As in the deontic case, using must is preferred if one can reasonably treat the rele-
vant applicability conditions as settled, and ought is preferred if one cannot.
In these ways, the analyses in .c1U.i s1voc and coui1io.i wi.x gen-
eralize across favors of modality. Tis is an important feature. Previous accounts
are ofen developed by considering a limited range of readings, and extensions to
other readings, to the extent that they are discussed at all, are ofen strained.
1i
Te
analyses oered here, by contrast, provide a uniform explanation for the behavior
of weak and strong necessity modals in their various interpretations. As we will see
in the remainder of this section, this lends our account greater empirical coverage
and explanatory power.
. Counterfactual marking
Te proposed analyses capture a wide range of further linguistic phenomena in-
volving weak and strong necessity modals. One example which I fnd especially
compelling concerns how weak necessity is expressed cross-linguistically. In the
remainder of this section I will argue that the way in which the proposed account
can capture the cross-linguistic data also helps explain various seemingly unrelated
semantic and pragmatic properties of ought and must.
Past tense (preterite) forms of modals in English, would for will, could for
can, might for may are ofen used not to indicate past time reference, but to
express tentativeness or politeness and weaken the force of the modality, as in (1,)
(1o). Tey are also the forms that appear in the consequents of subjunctive condi-
tionals, as in (1,).
(1,) a. Alice will (/may, /cant) be at home now.
b. Alice would (/might, /couldnt) be at home now.
(1o) a. May (/Can) I comment on your proposal:
1i
E.g., Coviiv iooo, Cu.viow io11, RUnis1ii io1ib.
1i
b. Might (/Could) I comment on your proposal:
(1,) If you took the i:oo fight tomorrow, you would (/could, /might) get there
by :oo.
P.imiv ioo1 dubs these uses of past tense the modal past.
1
Strikingly, ought patterns with the past-marked modal forms. First, as we have
seen, ought weakens the necessity of must. Second, ought, unlike must, can ap-
pear in subjunctive conditional consequents, as in (18).
(18) a. If Alice came to the party tomorrow, Bert ought to leave.
b. -If Alice came to the party tomorrow, Bert must(ed) leave.
Tird, when used with nonpast time reference, ought, unlike must, patterns with
the past-marked forms in being non-entailing; Ought , unlike Must , is compat-
ible with . (We will return to this point in i..)
(1) I could give to Oxfam, but I wont.
(io) a. Alice ought to be here by now, but she isnt.
b. -Alice must be here by now, but she isnt.
Fourth, when used with the perfect, ought patterns with the past-marked forms in
implicating the negation of its prejacent.
(i1) I could have given to Oxfam. (Implicates: I didnt)
(ii) a. We ought to have given to Oxfam. (Implicates: we didnt)
b. -We must have given to Oxfam (but we didnt).
Must cannot even receive a deontic reading when used with past time reference.
Ought, but not must, can be used to communicate that a certain obligation held in
the past. Tis follows from the more general point that only past-marked forms can
take scope under the perfect (Couov.vui iooi). Tat ought cantake scope under
the perfect in (iia) is a ffh respect in which ought patterns with past-marked
modal forms.
In sum, although must does not have a past tense form, ought appears to func-
tion semantically as its modal past.
1
Tis is surprising. But it becomes less surpris-
1
See also, e.g., Joos 1o, Lvos 1,,, Co.1is 18, Pivxis 18, P.imiv 18,, 1o, Fiiis-
cum. 18, I.1viuoU iooo, HUuuiis1o & PUiiUm iooi. Terminology varies among authors.
1
Cf. Auivso 1,1: ,; P.imiv 1o: ,oo, 8o, 1i1i8; ioo1: i, ,,, 1i,, ioio.
Formally, should is the past tense formof shall. But should with this quasi-subjunctive meaning is
1
ing when we examine other languages. Te cross-linguistic norm is to mark the se-
mantic distinction between weak and strong necessity morphologically rather than
lexically. Weak necessity is typically expressed, not by using a dierent word like
ought, in contrast to must, in English but by using the modal past form of a
strong necessity modal, the formof a strong necessity modal that is used in counter-
factuals (e.g., a strong necessity modal with fake past-tense marking, i.e. past-tense
marking that receives a nonpast interpretation).
1,
Von Fintel and Iatridou (ioo,, ioo8) attempt to capture this cross-linguistic data
in their semantics for ought. Tey treat weak necessity modals as quantifying over
the best of the best worlds roughly, the relevant P
w
-compatible worlds that are
also compatible with certain further considerations, represented by a secondary or-
dering source (premise set that represents the content of some ideal; see n. ). (Tis
can be understood as formalizing the not uncommon informal claim that Ought is
related to must as best is related to only (Wiiii.ms 181b: 1i,) i.e., that Ought
implies that is best among various acceptable alternatives, whereas Must
implies that is the only acceptable alternative; see also, e.g., Siom. 1,o, Mc-
N.m.v. 1o.) It is speculated that the counterfactual marking is co-opted here
in a somewhat meta-linguistic kind of way: if we were in a context in which the sec-
ondary ordering source was promoted [to primary status], then it would be a strong
necessity that (ioo8: 1). Te tentativeness associated with the counterfactual
marking is attributed to the fact that the premises in the secondary ordering source
need not apply: Te choice of whether to really promote the secondary ordering
source is lef open (ioo8: 1).
I fnd this analysis unsatisfying. First, little is said about what makes a primary
ordering source primary and a secondary ordering source secondary apart from
the fact that the latter fgures only in the interpretation of weak necessity modals. No
story is given about how primary and secondary ordering sources are determined
independently of the truth conditions of the relevant ought and must sentences.
We should be able to say what the primary and secondary ordering sources are in
various contexts without simply reverse engineering themfromrelevant truth value
distinct fromshould as expressing weak necessity. I amnot distinguishing typographically between
word forms and lexical items; context should disambiguate.
1,
See esp. P.imiv ioo1: 1i,, 1o, 1,,, 1818o, 11o; vo Fi1ii & I.1viuoU ioo8; cf.
V. Liui & Vivs1v.i1i ioo8, M.11uiwso io1o (esp. ,). Interestingly, the marker for
weak necessity in Central Pomo is also a conditional marker (P.imiv ioo1: 18,). As von Fintel and
Iatridou note (1ion.ii), English ought fts this pattern historically; it was formerly the past subjunc-
tive of the verb owe (see also CUvmi 11: 1; BoUm. 1,,: ). It is unclear how comparative
probability approaches to ought (Fii.v ioo, io1o, L.ssi1iv io11) would capture these data.
1
judgments. Second, absent an account of what makes it the case about a speaker
that she is counterfactually promoting a secondary ordering source, the proposed
story about the role of the counterfactual marking seems ad hoc. Tird, one might
worry that the proposed explanation for the tentativeness associated with the coun-
terfactual morphology just redescribes what needs to be explained. Te tentative-
ness is explained by positing a parameter consisting of premises that need not
apply. Fourth, it is unclear how their explanation would generalize to explain the
tentativeness associated with counterfactual morphology on other kinds of lexical
items e.g., possibility modals like might, or desire verbs like wish, which, as von
Fintel and Iatridou themselves note, are expressed in many languages by placing
counterfactual morphology on the word for want (see also I.1viuoU iooo).
Treating weak necessity as conditional necessity suggests a more natural expla-
nation. We can capture the cross-linguistic data by taking our cue fromindependent
theories of the semantic and pragmatic role of counterfactual marking. It is gen-
erally agreed that counterfactual morphology signals that the worlds being talked
about (the topic worlds) neednt be candidates for actuality (S1.i.xiv 1,,,
vo Fi1ii 18, I.1viuoU iooo, Bi11iv io11). Tere are various ways of for-
malizing this signal and implementing it in the grammar.
1o
One option is to say that
counterfactual marking presupposes that the modals domain of quantifcation, or
the set of topic worlds more generally, isnt a subset of the context set. Tis could
be incorporated into our analysis by supplementing the truth conditions in coui-
1io.i wi.x with a pragmatic constraint that some of the closest worlds selected
by the selection function s, i.e. some of the worlds at which the necessity of the pre-
jacent is evaluated, be outside the context set formally, that s(w, ) c. An al-
ternative is to treat counterfactual markers as canceling a positive presupposition
associated with the indicative that the set of topic worlds is a subset of the context
set; on this view, counterfactual markers fail to signal that the set of topic worlds is
included in the context set, rather than positively signaling that some of the topic
worlds arent in the context set. Tis could be incorporated into our analysis by
failing to include a presupposition that s(w, ) c, i.e. by failing to require that all
the worlds at which the necessity of the prejacent is evaluated be within the con-
text set. Either way, we can incorporate the role of counterfactual marking into our
analysis of weak necessity modals. Ought is treated semantically as the modal past
of must, which is precisely what we would expect given how weak necessity is ex-
1o
For related general discussion in the literature on mood selection about whether the indicative
or subjunctive, if either, is the semantic default, see, e.g., vo Fi1ii 18 and Bi11iv io11 for the
former view, and Pov1iv 1,, Scuiixiv ioo,, Pov1iv & RUnis1ii io1i for the latter.
1,
pressed cross-linguistically. Tis helps explain why ought should pattern with the
past modal forms in the ways described above.
Refning the account in this way also gives precise expression to the informal in-
tuition that ought is weaker and more tentative than must. In uttering Ought
the speaker makes a claimabout the necessity of and marks her utterance as being
about a set of worlds that neednt be candidates for actuality. But, as Stalnaker notes,
normally a speaker is concerned only with possible worlds within the context set,
since this set is defned as the set of possible worlds among which the speaker wishes
to distinguish (1,,: o). So, using ought implicates that one isnt in a position
to make a claim about whether the prejacent is necessary throughout the set of live
possibilities. Interestingly, this suggests that the basis of the scale between ought
and must is not one of quantifcation but of epistemic strength.
1,
Since ought is
weaker than must in this way, Grices frst quantity maximMake your contribu-
tion as informative as is required (Gvici 18: io) can be exploited to generate a
familiar upper-bounding implicature, namely, fromthe use of ought that for all one
knows or, better, for all one is willing to presuppose in the conversation Must
is false.
18
Tis scalar implicature has the usual properties of implicatures. It is
cancelable and reinforceable, as in (1a) and (ia), respectively, and it is suspendable,
as in (i).
(1a) I ought to help the poor. In fact, I must.
(ia) I ought to help the poor, but I dont have to.
(i) I ought to help the poor. Maybe I have to.
In (1a), for example, the speaker frst conveys that the set of worlds being talked
about isnt included in the set of live possibilities, and then asserts that what holds in
the former worlds the deontic necessity of the proposition that I help the poor
also holds in the latter worlds. Te implicature data with ought can thus be treated
analogously to the implicature data with subjunctive conditionals.
(i) a. If you had the fu, you would have exactly the symptoms you have now.
(cf. Auivso 1,1: ,)
b. If you had the fu, you would have very dierent symptoms from the
symptoms you have now.
c. If you had the fu, you would be sick. Maybe you do have the fu; you
1,
Cf. Ziiciiiv iooo and V. Liui & Vivs1v.i1i ioo8 on counterfactual clauses; see also
I.1viuoU iooo and Couov.vui iooi.
18
See, e.g., Hov 1,i, 18; G.zu.v 1,; G.mU1 11.
1o
are pretty congested.
Likewise we can assimilate the tentativeness of ought to the tentativeness of past
forms more generally, as innon-counterfactual subjunctive conditionals (sometimes
called future less vivid conditionals) like (1,) above and (i,).
(i,) If youcame toour party tomorrowandImnot saying that youwill you
would have a great time.
In such examples the speaker may simply wish to admit the possibility that it could
become taken for granted that the marked clause is not true. Tis is what happens
with ought regarding the relevant applicability conditions (and, as we will see in
i., the modals prejacent as well).
In this section I have argued that Ought says that Must is true at certain
possibly counterfactual worlds. I would like to consider two objections to refning
the account from i.i in this way. First, it might seem to incorrectly predict that
sentences like (io) of the form Must ought are consistent: It would seem to
be possible for there to be an unsaturated premise set P such that follows from
P
w
and follows from P
w
, for all w

s(w, ) and some that is false at w.

1
(io) -I mustnt lie to Alice, but I should.
As the reinforcability data in (ia) shows, it is felicitous and consistent to predicate
the necessity of of worlds at least some of which are outside the context set (by
accepting Ought ) while denying that is necessary (by denying Must ). But our
treatment of the pragmatics of ought claims confrms that it should be anomalous
to predicate the necessity of of certain worlds outside the context set (by accepting
Ought ) while accepting that is necessary (by accepting Must ). Given a
set of relevant considerations that bear on the necessity of , the worlds of which
is predicated to be necessary in an utterance of Ought are, roughly, those worlds
compatible with the applicability conditions that the speaker most endorses in some
relevant sense (i.i). If a speaker successfully asserts Must , then some of these
worlds will be in the context set. So, if Must is felicitous and true, Ought is
false. One cannot accept both Must and Ought .
Second, one might worry that the account predicts that ought is semantically
equivalent to would have to. But, as vo Fi1ii & I.1viuoU (ioo8: 1i811) ob-
serve, ought cannot be replaced by would have to, as refected in (i,)(i).
1
A simple model: Let V = {w, w

}, P
w
= {}, P
w
= {}, w , w

, and s(w, ) = {w

}.
1,
(i,) a. I ought to help the poor. In fact, I must.
b.
?
I would have to help the poor. In fact, I must.
(i8) a. I ought to help the poor, but I dont have to.
b.
?
I would have to help the poor, but I dont have to.
(i) a. (If Fred wanted to get to the island) he would have to use this boat.
b. He ought to use this boat.
(= (,) and (oi) in vo Fi1ii & I.1viuoU ioo8)
Languages that mark the weak/strong necessity distinction morphologically use the
same string to express weak necessity and counterfactual necessity. Given compo-
sitionality, it would be preferable, other things equal, to give these strings the same
semantic interpretation. Te account in this paper, unlike the one in vo Fi1ii
& I.1viuoU ioo8, has this advantage. Expressions of weak necessity are given pre-
cisely the semantics we would expect given their lexical and morphological features
across languages. Even so, we can still capture the dierences between ought and
would have to sentences.
First, speakers can accept Ought without having to make explicit and agree
upon a specifc condition given which would be necessary. Tis locates a contrast
in the anaphoric properties of ought and would have to. Whereas ought is ac-
ceptable even if no particular condition is salient, as in (oa), would have to is not,
as in (ob). Te condition that fgures in a would have to claim must be readily
retrievable by the addressee; it must be salient in the linguistic or extra-linguistic
context, as in (1) and (i), respectively.
(o) [Context: We are strangers standing in a hotel lobby. I notice you fumbling
with your bags.]
a. Here, I should help you.
b.
?
Here, I would have to help you.
(1) [Context: We are trying out electric guitars in a music store. We come
across a very expensive vintage Gibson Les Paul.]
a. If I really wanted to buy this guitar, I would have to check with my
wife.
b. I could buy this guitar. But I would have to check with my wife.
(i) [Context: Same as in (1)]
I would have to check with my wife.
Second, uttering Ought expresses that there is some relevant condition that is
18
most likely or desirable given which is necessary. Te same doesnt hold for would
have to: one can accept that is necessary given , for some possibly counterfactual
condition , without thinking that is the most likely or desirable condition that
bears on whether is necessary. Tis can lead to contrasts of the sort refected in
()(); the (a)-examples with would have to are true, but the (b)-examples with
ought are false.
() a. If I was a mobster, which Im not, I would have to kill you.
b. I ought to kill you.
() [Context: I dont knowwhether Alice will come to my wedding next month.
As a matter of fact, unbeknownst to me, she wont end up coming.]
a. If Alice came to my wedding next month, I would have to send her a
thank you card.
b. I ought to send Alice a thank you card.
Tis predicts that if we make explicit a condition that is mutually endorsed in the
conversation, the eects of a would have to claim should be closer to those of an
ought claim. Tis prediction appears to be borne out.
(,) a. Alice lef an hour ago. If there wasnt any trac and everything was
normal, she would have to be at her oce by now. In fact, I checked
and there wasnt any trac and everything was normal. So she must
be at her oce by now.
b. Alice lef an hour ago. She ought to be at her oce by now. In fact, I
checked, and there wasnt any trac and everything was normal. So
she must be at her oce by now.
(o) [Context (see ()): Alice wants to go to Harlemand is considering with her
mother, Martha, whether to take the Aor the Ctrain. Te Atrainis quicker,
but the C train is safer. Martha says:]
a. If safety was most important, you would have to take the C train. In
fact, safety is more important, as we can agree. So you have to take the
C train.
b. You ought to take the C train. In fact, safety is most important, as we
can agree. So you have to take the C train.
So, although ought claims are claims about what is necessary in certain possibly
counterfactual worlds, they dier from would have to conditionals.
1
. Entailingness and performativity
In i. we saw that there are good reasons for treating ought semantically as the
modal past of must, and I suggested how we might do so by incorporating gen-
eral insights about the role of counterfactual marking. In this section I will argue
that refning our account of weak necessity as conditional necessity in this way can
also help explain several seemingly unrelated puzzles concerning entailingness and
performativity with ought and must.
First, though many authors have claimed that Ought on its epistemic reading
expresses that is probable,
io
the data strongly suggests that ought is unlike must
in this respect. Must commits the speaker to a high unconditional credence in ,
whereas Ought does not commit the speaker to any unconditional credence in
(cf. Sw.so io1ia). Reconsider (io).
(io) a. Alice ought to be here by now, but she isnt.
b. -Alice must be here by now, but she isnt.
Epistemic Must but is inconsistent in a way that epistemic Ought , but
is not.
i1
Surprisingly, this appears to hold for deontic readings as well.
ii
(,) a. You ought to help your mother, but you wont (/but I knowyou wont).
b. -You must help your mother, but you wont (/but I know you wont).
Of course obligations can go unfulflled. What is interesting is that speakers appear
to assume otherwise, at least for the purposes of conversation, when expressing obli-
gations with must.
Our discussion in i. of the pragmatic import of counterfactual markers sug-
gests one natural way of capturing this data. Since must does not have a counter-
io
E.g., Siom. 1,o; Hov 1,i; Wiv1uiimiv 1,i; Fii.v io1o; L.ssi1iv io11.
i1
Te felicity of epistemic Ought , but suggests that the condition must be live at least
in the sense that one does not have evidence from ones direct experience that it is false. As long
as one would have expected that was satisfed were it not for ones indirect evidence, one can still
utter Ought . Support for this comes from the fact that it is preferred for denials of the relevant
ACs to be headed by epistemic must, which typically signals that its prejacent is the conclusion of
an indirect inference (see, e.g., K.v11Ui 1,i: 11-1,; Co.1is 18: 18, 1, oo,, ,, 1,,; ioo1:
,; vo Fi1ii & Giiiiis io1o).
(i) She ought to be here by now, but she isnt. Something must have (/
?
has) gone wrong!
ii
See esp. Wiviv ioo: 1i1, and ioo,; Ni.ioo,; Pov1iv ioo: 1o, 181o. See also,
among many others, Limmo 1oi, Liicu 1,1, Wiv1uiimiv 1,i, H.vm. 1,, Lvos 1,,,
Wiiii.ms 181b, Co.1is 18, P.imiv 1o, ioo1, Swii1siv 1o, Mvuiii 1o, HUuuiis1o
& PUiiUm iooi, Hovviv & Tv.Uco11 ioo, Ciosi & A.v1s io1o, Sw.so io1ia.
io
factual element to its meaning, the context set must include P
w
, for all worlds w in
the context set. (Again, it is immaterial for my purposes whether this is because of a
positive presupposition associated with the indicative, or because there is a default
presupposition that a modals domain of quantifcation is a subset of the context
set.) So, if Must is accepted, cannot be satisfed throughout the context set;
hence the inconsistency of Must but on any reading. However, there is no
similar restriction on the value of the unsaturated premise set P at worlds outside
the context set; the value of P at non-live possibilities neednt be compatible with
the common ground. Since ought can take us to worlds outside the context set in
assessing the necessity of its prejacent, as long as all worlds w

s(w, ) are outside

the context set, Ought can be true at a live possibility w even if all the worlds in
the context set are -worlds; hence the consistency of Ought , but . So, even if
Sw.so (io1ia: iii) is right that pure premise semantics on its own cannot
capture the above data, the analyses of ought and must in i.i, conjoined with
independently motivated pragmatic constraints on mood and the interpretation of
modals, suggest a natural explanation.
i
Deontic ought and must are ofen thought to dier in their conventional force.
At least in root clauses, deontic must, unlike deontic ought, is ofen thought to con-
ventionally performa directive speech act, or be used to try to get someone (possibly
oneself) to do something, perhaps in addition to performing an assertion.
i
Pre-
vious attempts to capture this performative property of deontic must have relied
i
Tis is of course not the only possible explanation for the data in (io) and (,). One option is to
say that must is only interpreted with respect to a modal base (or the union of several modal bases),
not an ordering source (see n. ). (As far as the interpretation of must goes, one might say, all laws
are natural laws (cf. Pi.ci1 1oi: o).) Since modal bases are realistic they consist of propo-
sitions that are true at the world of evaluation if Must is true, then must also be true. But
there are good reasons for introducing ordering sources into the semantics, both conceptual (in dis-
tinguishing facts fromnorms) and empirical (in handling inconsistent premise sets and gradability).
Te proposal in the main text has more independent motivation. Second, Sw.so io1ia captures
the data by including in the semantic entry for must but not ought a constraint that requires high
credence in the modals prejacent. It would be better, I take it, if we could explain the data in terms
of independent features of the semantics of ought and must, as I have argued we can. Tird, one
might explain the data by appealing to the dierent performative properties of weak and strong ne-
cessity modals (Ni. ioo,; Sw.so ioo8: 1io1io; Pov1iv ioo: 1o, 181o). I argue
below that the analysis in the main text has the advantage of deriving these performative properties,
without needing to take them as basic.
i
See esp. Ni. ioo,; Pov1iv ioo,: oo,, ioo: 1o1o,, 181o; Sw.so ioo8:
1io1io. See also Biuvi 1,,, Liicu 1,o, 1,1, Lvos 1,,, Liwis 1,a, Wiiis 1,,
Co.1is 18, P.imiv 1o, Bvnii i1 .i. 1, v. Roov iooo, HUuuiis1o & PUiiUm iooi,
NUv1s i1 .i. ioo,; see also the references in n. i,.
i1
on ad hoc stipulations e.g., by positing an independent dynamic component to
musts lexical semantics (see esp. Pov1iv ioo, also Ni. ioo,, Pov1iv ioo,,
Sw.so ioo8). Tough a dynamic implementation might be desirable on other
grounds, our explanations for (io) and (,) suggest the following alternative static
explanation. Tis explanation has the advantage of deriving the contrasting speech
act properties of deontic ought and must from their truth conditions and general
considerations about their use.
Since accepting Must is incompatible with denying , if the truth of is
assumed to be under the control of some relevant subject, updating with Must
will commit the subject of the obligation to seeing to it that . So, it is no surprise
that must should be thought to be conventionally directive. But since accepting
Ought is compatible with denying , updating with Ought neednt commit
anyone to seeing to it that . So, we correctly predict that even if deontic ought
can be used to perform a directive speech act, it doesnt do so as a matter of its
conventional meaning. Further, given our discussion of the conversational role of
ought claims (i.1i.i), it should come as no surprise that deontic ought claims
should ofen perform more moderate speech acts of recommending or advising.
In uttering Ought one can express ones preference that be true, but without
imposing the truth of on the common ground.
i,
I have said that Must but is inconsistent on any reading. Tis claim may
need to be qualifed. Tough there is a robust body of evidence that speakers fnd
deontic examples of Must but like (,a) to be anomalous (see n. ii), some
speakers report that they can fnd such examples to be consistent in certain contexts.
(I am yet to fnd this judgment expressed in published work, though I have heard it
i,
It is not counterintuitive that the drastic decline in frequency of must is due in no small part
to the above features of its meaning and use (cf. Mvuiii 1,: 1,,, 1oo, 18o, 11,, iooioi,
io,io,, 1o: ,181; Liicu ioo: i,; Smi1u ioo: i,io, i,, i,ioo, ioio; Liicu
i1 .i. ioo: 8,o, 1111,; Ciosi & A.v1s io1o: 1,,1,8). In my io1a I apply these explana-
tions of the contrasting discourse properties of deontic modals to the Miners Puzzle, and argue that
they elucidate previously unnoticed contrasts among deontic ought, must, and may conditionals.
I said above that if Must is accepted, then cannot be true throughout the context set. Tis
would seem to leave open whether might be true at some worlds in the context set, and would
thus seem to predict the consistency of (i).
(i)
?
You must go to confession, but maybe you wont (/but you might not).
Even if (i) is consistent for this reason, one might try to capture its anomalousness in terms of a
general norm of cooperative conversation that interlocutors do what they can to make the actual
world be among the best worlds (cf. Pov1iv ioo,: ,8). (i) would be anomalous to the extent that
it is anomalous to commit someone to help see to it that while expressly admitting the possibility
of . Tis is not full-blown inconsistency, but that seems like the correct prediction for this case.
ii
voiced in personal conversations.) Consider (8).
(8) I must go to confession; Im a Catholic. But Im not going to. I havent
practiced for years.
I myself fnd it hard to hear an utterance of (8) as being consistent and sincere.
But what is important is that even speakers who can hear it as such agree that it
would sound better if a strong necessity modal like have to or be required was
used instead of must, as in ().
() I have to (/Im required to) go to confession; Im a Catholic. But Im not
going to. I havent practiced for years.
Intuitively, in () it is consistent for the speaker to go on to deny the modals preja-
cent because she isnt endorsing the norms that would make it true that she has the
obligation in question. She is simply reporting what is required by a given system
of norms, norms which she does not treat as binding on her actions.
Tis suggests that one can hear an utterance of (8) as consistent and sincere to
the extent that one accepts objective uses of must, in the sense of Lvos 1,,.
Following Lyonss terminology, a modal is used subjectively if it presents the speaker
as endorsing the considerations with respect to which the modal claim would be
true; a modal is used objectively if it does not. Be required is typically used objec-
tively. Have to and (have) got to are more fexible, with have to tending more
toward the objective side of the spectrum and (have) got to more toward the sub-
jective side. Tough must is nearly always used subjectively, objective uses may be
possible for some speakers.
io
Te distinction between objective and subjective uses can help in the following
way. Te explanation given above for the contrast between the (a)-examples and the
(b)-examples in (io) and (,) turned on a dierence between weak and strong ne-
cessity modals namely, that strong necessity modals presuppose that the modals
domain of quantifcation is a subset of the context set. Tis explanation is, I take
it, still correct. What the data in () highlights is that this presupposition can be
locally satisfed (satisfed in the original context as modifed by a clause or part of
a clause) specifcally, in the accommodated context that represents the body of
considerations and background facts that the modal claim is about. Since in objec-
io
Tese generalizations are supported by a robust body of data in descriptive linguistics (cf. n. i).
In addition to Lvos 1,,, see Liicu 1,o, 1,1, ioo, H.iiiu.v 1,o, L.xoii 1,ia,b, L.vxi
1,o, Co.1is 18, P.imiv 1o, ioo1, Swii1siv 1o, Mvuiii 1,, 1o, Vivs1v.i1i ioo1,
HUuuiis1o & PUiiUm iooi, Smi1u ioo. See also Sw.so io1ib for discussion.
i
tive uses the speaker simply reports what follows from (/is compatible with) these
considerations, without presenting herself as endorsing them, this accommodated
context neednt be compatible with the global context set. If it isnt, the prejacent of
the strong necessity modal can be true throughout the context set. Objective uses
like those in (8)() pose no more of a challenge than other ordinary examples
like (o) in which a strong necessity modal appears in an intensional context.
(o) Alice thinks that she must go to confession tomorrow. But well talk to her
tonight, and shell see that shes wrong. So she wont end up going and can
come out with us instead.
It is only in subjective uses that the set of best worlds must be included in the global
context set.
So, the promised qualifcation is this: for subjective uses of a strong necessity
modal M, M but is inconsistent. To the extent to which one fnds objective
uses of must acceptable, to that same extent one will fnd Must but to be
consistent and one will fnd deontic Must to lack imperatival force. So, even if
deontic must permits objective uses and is only canonically, rather than conven-
tionally, performative, we can capture what distinguishes the non-canonical, non-
performative uses. What is distinctive about weak necessity modals is that even when
they are used subjectively they are non-entailing and lack directive force.
. Negotiability and collective commitment
We have seen two dimensions along which necessity modals can dier: frst, they
candier instrength (weak vs. strong); second, they candier inthe degree to which
they are used to express the speakers endorsement of the considerations that would
verify them (subjective vs. objective). Te primary emphasis in this paper has been
on elucidating the frst dimension, the distinction between weak and strong neces-
sity. But the discussion in this section of the distinction between subjective and
objective uses can help us respond to another recent account of the weak/strong ne-
cessity distinction, Aynat Rubinsteins (see esp. her io1ib). Rubinsteins discussion
is rich. For reasons of space I must reserve a more thorough discussion for elsewhere
(see also my io1i, io1c).
On Rubinsteins account, what makes weak necessity modals weak is that they
are interpreted with respect to priorities (norms, goals, desires) that are presumed
to be negotiable. (Rubinstein does not examine epistemic modals.) In short:
[S]trong necessity modals are only sensitive to prioritizing premises
i
that the conversational participants are presupposed to be collectively
committed to [I]f any participant in the conversation were given the
chance to defend these priorities, it is assumed in the context of the con-
versation that they would do so. Weak necessity modals take into ac-
count all these premises plus some more. For these additional premises,
lack of collective commitment is presupposed[A] speaker uses a weak
modal when he or she believes (perhaps mistakenly) that the secondary
priorities it depends on are still up for discussion. (io1ib: ,1,i)
It is hard to know what is predicted by Rubinsteins account until more is said about
what exactly is meant by a commitment to defend certain priorities, how the con-
siderations that intuitively fgure in interpretation are to be represented in specifc
examples, and how this relates to their status as negotiable or non-negotiable. Even
so, our previous discussion already gives us reasons to think that the weak/strong
necessity distinction is not best captured in terms of whether there is collective com-
mitment to a body of priorities.
Rubinstein is explicit that commitment to a priority amounts to an endorsement
of that priority as desirable (io1ib: ,8; see also Pov1iv & RUnis1ii io1i). But
the weak/strong necessity distinction cuts across the distinction between modals
that express (collective) commitment in this sense and those that dont. First, there
are uses of weak necessity modals that involve collective commitment to a prior-
ity. In () the interlocutors are both expressly committed to the value of family,
though it is ought that is used (and, indeed, preferred). Second, there are uses of
strong necessity modals in which the speaker expressly rejects commitment to the
premises that verify the necessity claim, as in () above. Tis lack of commitment
or endorsement can even be common ground, as in (1).
(1) [Context: Were teenage siblings. Its 1o:o p.m., and we plan on staying out
and going to a party. We know our parents are already asleep.]
You: When is curfew, again: We need to make sure we tell Mom we got
back before then if she asks.
Me: We have to be home by 11. Isnt Mom stupid: Its amazing that she
thinks her rules matter at all. Cmon, lets go.
Te account in this paper the concrete interpretation of how considerations
are represented inpremise semantics, the truth conditions of weak and strong neces-
sity modals, and the associated pragmatic norms better captures Rubinsteins in-
tuitions about negotiability andcollective commitment. Anutterance of STRONG
i,
expresses collective commitment in the same ordinary sense as any other context-
sensitive utterance: Doing so assumes a value for the context-dependent item and
asserts the proposition that results when this value is assigned. Uttering Te dog
is sleeping assumes that some dog d is the contextually salient dog and proposes
to restrict the common ground to worlds in which d is sleeping; likewise, uttering
STRONG assumes an unsaturated premise set P that represents the content of
the relevant considerations and proposes to restrict the common ground to worlds
w in which follows from P
w
. Strong modals express a commitment that the value
of the relevant contextual parameter is a certain way e.g., that morality, the house
rules, your goals are representable by such-and-such unsaturated premise set and
that we are in a world in which what is enjoined by the considerations thus repre-
sented entails the prejacent. Tis is compatible with not endorsing the premises in
P(w), for any w in the context set. In (1) I can think that the house rules require us
to be home by 11 while rejecting that those rules ought to play any role in guiding
our plans.
i,
Te putative negotiability and lack of collective commitment associated with
weak modals is elucidated by treating weak necessity as conditional necessity. It
isnt the considerations that fgure in the interpretation of weak modals that need
be negotiable. What is negotiable is whether we are in a world that determines a
saturated premise set that verifes the necessity of the prejacent. But this lack of
commitment that the prejacent is in fact necessary is compatible with endorsing the
considerations represented by the given unsaturated premise set indeed, ought
is typically used subjectively (i.). In () we can endorse the value of family even
if we prefer not to settle how it might relate to other potentially competing values.
In sum, strong modals imply collective commitment in the sense that they as-
sume that the relevant considerations are thus-and-so, and make a claimabout what
is in fact necessary in light of those considerations. Tis is compatible with failing to
endorse the premises that verify the necessity of the prejacent. Weak modals imply
negotiability in the sense that they only say that the prejacent is necessary on some
supposition. Tis is compatible with endorsing the premises that would verify the
necessity of the prejacent. Te relation between weak and strong necessity modals
and common ground assumptions neednt be stipulated or treated as basic in an ac-
count of the weak/strong necessity distinction. It follows from, and can be explained
by, the modals truth conditions and general pragmatic principles.
i,
Rubinstein cannot accept this alternative way of understanding (collective) commitment given
her broader applications of the notion to explaining patterns of verbal mood marking. Tat the
operative notion is commitment to the content of the premise set as reasonable, desirable, etc. is
crucial to the explanations of mood selection in Pov1iv & RUnis1ii io1i.
io
A little philosophical therapy
Tus concludes our whirlwind tour of various linguistic advantages of understand-
ing weak necessity as conditional necessity. Ought is the pet termof many a philoso-
pher. It is thus surprising how little theoretical work exists investigating the mean-
ing of ought in contradistinction to other necessity modals. I have argued that
the common semantic core of weak necessity modals across their various readings
(deontic, epistemic, etc.) is that they express conditional necessity, necessity on the
supposition that the applicability conditions of certain norms, values, goals, etc. are
satisfed. Tis analysis carves out an important role for expressions of weak ne-
cessity in conversation, deliberation, and planning: they allow us to entertain and
plan for hypothetical extensions of the current context, and coordinate our norms,
values, goals, and expectations while remaining open to new evidence about how
these considerations apply. Te account also captures various linguistic data that
has heretofore resisted systematic explanation e.g., concerning the relative felic-
ity of weak and strong necessity modals in context, the relation between weak and
strong necessity modals and standing contextual assumptions, the morphosyntactic
properties of expressions of weak necessity cross-linguistically, and the contrasting
performative properties of weak and strong necessity modals. Tis wide range of
semantic and pragmatic phenomena that are unifed under and explained by our
analysis lend it a robust base of support. I welcome the development of alternative
ways of systematizing the data with which the present account may be compared.
As interesting as our linguistic ruminations have been no presupposition fail-
ure I hope! the distinction between ought and must is not merely of interest to
the philosopher of language or linguist. Greater sensitivity to dierences among
necessity modals can improve theorizing on broader philosophical issues. In clos-
ing will briefy discuss two such issues in normative theory: moral dilemmas and
supererogation (see my io1 for additional discussion and examples).
. Moral dilemmas
Suppose you promised Alice that you would help her move and promised Bert that
you would help him move, but you discover that you cannot help them both. Sup-
pose also that each promise is uniquely important and that there is no higher-order
principle you can use to resolve whom to help. On the face of it, you are in a moral
dilemma. But some have argued that genuine moral dilemmas are impossible.
On the one hand, it is not uncommon for arguments against the existence of
moral dilemmas to be motivated by the thought that it is inconsistent for contrary
i,
acts to be simultaneously required. It would seeminconsistent for you to be required
to help Alice and required to help Bert when you cannot help them both. Morality,
the thought goes, could never land us in such a genuine contradiction. As Alan
Donagan puts it:
i8
Rationalist theories cannot allowmoral dilemmasEachprinciple and
each derivative proposition of a rationalist theory asserts, of some rule
or precept that it assumes all human beings can observe in all situations
to whichthat rule or precept applies, that practical reasonrequires them
all to observe it. If, therefore, any such theory were to assert that prac-
tical reason requires any human being in any situation to observe a set
of precepts that cannot all be observed in it, it would contradict itself;
for it would assert that set of precepts not to be what it also asserts or
assumes them to be. (Do.c. 1: 1,)
On the other hand, there is a long tradition in ethics and logic that is compelled by
the thought that it is at least coherent to think that Ought and Ought might
both be true, for contraries and .
i
Even if, fortuitously, there are no moral
dilemmas, it is plausible that the possibility of moral dilemmas should not be ruled
out simply by the meaning of ought. Tis intuitionis bolsteredwhenwe turnto non-
moral dilemmas. Evenif morality could not place us in a situation in which we could
not satisfy all its demands, our desires, for example, certainly could. Suppose you
want to keep both of your promises to Alice and Bert but regard them as uniquely
important. You have landed yourself in a dilemma concerning what to do in light of
your desires. We needa semantics that allows dilemmas tobe consistently expressed.
Attending to dierences between weak and strong necessity modals can help us
capture both of these intuitions. Even if we cannot be required to perform each of
several mutually incompatible actions, ought is required. Many fnd it plausi-
ble that the dilemma in (i) expressed with ought is consistent in a way that the
dilemma in () expressed with must is not (assuming the modals are interpreted
subjectively (i.) and with respect to a constant context, i.e. a constant unsaturated
premise set).
o
i8
See also Avis1o1ii 18, AqUi.s 18, Miii 18, K.1 1o, Ross 1o, D.viuso 1o,
Si.vii 18o, H.vi 181, Coii 18i, Do.c.18, Fiium.18o, Tuomso1o, Pii1vosxi
1, Bvix 1, McCoiii 1o, Zimmivm. 1o. (Tough see Hiii 1o for an interpreta-
tion of Kant on which dilemmas are possible.)
i
See, among many others, Limmo 1oi, v. Fv..ssi 1,, Wiiii.ms 1,, B.vc. M.v-
cUs 18o, Foo1 18, Gow.s 18,, D.cv 1, H.vm. 1, Hov1v 1, ioo, Gonii ioo,
Sw.so io11.
o
See, e.g., Limmo 1oi, Gow.s 18,, H.vm. 1, Sw.so io11; cf. Foo1 18.
i8
(i) I ought to help Alice and I ought to help Bert, but I cant help them both.
() -I must help Alice and I must help Bert, but I cant help them both.
Te truth conditions in .c1U.i s1voc and coui1io.i wi.x capture this by
validating agglomeration for must but not for ought: Must and Must entail
Must , but Ought and Ought dont entail Ought . Treating must
as an ordinary necessity modal validates agglomeration. If P
w
and P
w
,
thenP
w
. If and are contraries, thenMust andMust canbothbe
true only if P
w
is inconsistent, i.e., only if P
w
= . Insofar as quantifers presuppose
that their domains are non-empty, this correctly predicts that dilemmas expressed
with must are either infelicitous or inconsistent. By contrast, coui1io.i wi.x
allows dilemmas to be consistently expressed with ought. Ought and Ought
can both be true, for contraries and , if (and only if) there are conditions C
and C

such that Must is true at all closest relevant C-worlds w and Must is
true at all closest relevant C

-worlds w

. If C and C

are incomparable in likelihood

or desirability and there is no other relevant condition C

that is at least as likely or

desirable as C or C

, then both Ought and Ought are assertable as well.

1
So, if circumstances conspire aright, Ought and Ought can both be true,
evenrelative tothe same unsaturatedpremise set. Tis raises the questionof whether
circumstances could so conspire, or whether, for a given favor of modality, such an
unsaturated premise set could ever be called for. But this is a question for the world
or for substantive normative theory, not for semantics. In the case of, say, desire-
based dilemmas, the answer seems to be a clear Yes. But perhaps the correct moral
view excludes this possibility. Tere may be independent arguments that there are
no genuine moral dilemmas. But such arguments will need to be just that: inde-
pendent that is, independent of linguistic intuitions like the intuition that () is
inconsistent. Dilemmas are coherently expressible. Tey neednt entail a contra-
diction (D.viuso 1o: ).
1
Sw.so io11 builds this sensitivity to incomparabilities into his semantics for ought. Swan-
sons semantics or at least an adaptation of it with the limit assumption treats Ought as saying
that follows from some maximally consistent subset of the given premise set. (Must is given an
ordinary semantics of necessity. Compare these semantics for ought and must with the confict
account and disjunctive account, respectively, of all-things-considered ought claims in Hov1v
ioo, io1i.) Tis semantics also predicts that dilemmas are coherently expressible with ought. Te
analysis in this paper has the advantage of capturing a wider range of data.
i
. Supererogation
Supererogatory acts are acts that go beyond the call of duty. Tey are permitted but
not required, and better than what is minimally required. Tink: throwing yourself
on a grenade to protect your friends, giving a substantial portion of your income to
the poor, and so on.
Some ethicists claim that there are no supererogatory acts, and that acts that are
intuitively supererogatory are in fact required (anti-supererogationists). Others
grant that there are supererogatory acts, but maintain that they are still binding in
some sense and thus deserving of some criticism if not performed (qualifed su-
pererogationists).
i
Te following is a not uncommon line of argument: It would
be much better if I gave more money to the poor. I really ought to do so. So, I must
have conclusive reason, and hence an obligation, to give more to the poor. So, my
not giving more to the poor must be wrong and hence subject to criticism. (And
so, the anti-supererogationist would add, my giving more to the poor must not
be supererogatory afer all.) Tis kind of argument generalizes, leading to the so-
called paradox of supererogation, or the good-ought tie-up. Here is Joseph Raz
(though he ultimately rejects this line of thought):
If doing a supererogatory act is praiseworthy there must be reasons for
doing it, and the reasons must outweigh any conficting reasons for not
doing it. But if there are conclusive reasons for performing the act then
not to perform it is to act against the balance of reasons. If reason
requires that the act be done then surely one ought to do it, and the
ought is based on all the reasons which apply to the case; it is a con-
clusive ought. But this entails that failing to perform the act is failing
to do what one ought (conclusively) to do, so why isnt it blameworthy
not to perform a supererogatory act: (R.z 1,,: 1o)
Tere are several things to be distinguished in this type of argument: what is
good, what is blameworthy, what one ought to do, and what one is required to do.
Suppose that it would be supererogatory for you to give more to the poor and that
you ought to do so. (For the sake of argument we can suppose that supererogatory
acts are acts that one ought to perform, as opposed to simply being acts that it would
i
Tis terminology follows Hivu 18i. For classic examples of anti-supererogationism, see
Moovi 1o, Niw1,, Fiium.18o, K.c.18. For classic examples of qualifed supereroga-
tionism, see R.wis 1,1, Ricu.vus 1,1, R.z 1,,. See Uvmso 1,8 for the seminal work that
prompted contemporary interest in supererogation. For extensive treatments of supererogation, see
especially Hivu 18i and Miiiim. 11.
o
be better to perform.) Tat you ought to give more to the poor, however, doesnt
imply that you must, are required, or have conclusive reason to do so. Must, be
required and have conclusive reason express strong necessity. While it is plausible
that failing to do what one must is blameworthy, it is less obvious that one may
always be blamed for failing to do what is good. Te anti-supererogationist gets
traction with the intuition that failing to do what one must is wrong and subject
to criticism. It is only if we confate must with ought something facilitated by
the pervasive talk in ethics of obligation

that this intuition will seem to speak

against supererogationism.
Of course, this does not showthat supererogationismis correct, or that some acts
are supererogatory. One might accept on independent grounds a demanding moral
theory according to which one must do what is evaluatively best. But many fnd
such a view implausible, and it would be preferable if an alternative was available.
By clarifying the distinction between ought and must we can see how it is at least
coherent for a moral view to (a) distinguish what one ought to do and what would
be best fromwhat one must do and what is minimally required, and (b) attach blame
or criticism to failing to do the latter.
Te particular account of weak necessity developed in this paper is well placed
to capture two additional intuitions that many have had about supererogation. First,
there are long traditions according to which it is only a select few, perhaps those who
have some sort of higher calling, who ought to performsupererogatory acts.

Sec-
ond, many agents who perform supererogatory acts regard them as their duty, or as
things they must do.
,
By treating premises as in force only given the satisfaction of
certain applicability conditions, and treating ought as expressing necessity condi-
tional on certain such applicability conditions being satisfed, we can capture both
of these intuitions.
First, suppose there are conditional norms to the eect that one s if one desires
greater merit, or has a higher calling, or has been given a special dispensation of di-
vine grace, etc., where is some intuitively supererogatory action. If Alice does not
satisfy some such applicability condition, Alice must is false. But if it is plausible

Cf. HUuuiis1o & PUiiUm iooi: io,: Deontic necessity is commonly glossed as obligation,
but the noun obligation covers the range of should as well as must.

See Hivu 18i: ch. 1; Miiiim. 11: ch. for discussion.

,
For instance, concerning the righteous gentiles he interviewed from Le Chambon who pro-
tected Jews from Nazis, Philip Hallie relates that they invariably pulled back from me but looked
frmly into my eyes and said: How can you call us good: We were doing what had to be done
(H.iiii 1,: io, cited in Hovc. & Timmos io1o: on.i). See also, e.g., Uvmso 1,8:
1o1o; Eisinivc 1oo.
1
that she does, one may truly say that Alice ought to . In this sense -ing may still
count as supererogatory for her. Further, the act of -ing is supererogatory inso-
far as agents in general fail to satisfy the relevant applicability conditions and the
generic claim One must is false.
Second, given that agents are typically in a position to settle on whether they sat-
isfy the above sorts of applicability conditions whether they desire greater merit,
etc. it is unsurprising that those who perform supererogatory acts may regard
them as things they must do. In saying I must the agent expresses her assump-
tion that the supererogatory norms in question apply to her. Even if the agent is
correct about this, the generic claim One must is false and thus there is still a
sense in which her act is supererogatory. Tere is also a sense in which her act may
be supererogatory for her. For even if she is among the select few, the fact that she
is may itself be the result of some supererogatory act(s). Her wanting to go beyond
the call, say, is good but not required of her. Even if she must , it isnt the case that
she must be such that she must . Her being such that -ing is required for her is
itself supererogatory. Tough she is just doing her duty, that it is her duty renders
her deserving of praise.
As the discussion in this section highlights, we must be sensitive to the partic-
ularities of the terms we use when engaging in normative inquiry. Tis suggests
a more general methodological lesson. Te ethicist ofen comes to the table with
various implicit views about how normative language works, some correct but oth-
ers not. Tese assumptions can sometimes brought to bear, perhaps inadvertently,
in arguments on substantive normative issues. By locating these assumptions, the
philosopher of language can free up the ethicist in her investigations of normativity
and normative thought and talk. Our study of the distinction between ought and
must is illustrative of howtheoretical inquiry into normative language and conver-
sation can clarify and improve normative evaluation in practice.
References
Auivso, Ai. Ross. 1,1. Anote on subjunctive and counterfactual condition-
als. Analysis rr, ,8.
Auivso, Jou M. 1,1. Some proposals concerning the modal verb in English.
In A.J. Ai1xi, AcUs McI1osu, &Hivm. PIisso (Eds.), Edinburgh stud-
ies in English and Scots, pp. o1io. London: Longman.
AqUi.s, Tuom.s. 18. Summa theologica. Westminster, MD: Christian Classics.
i
Avis1o1ii. 18. Te ^ichomachean Ethics. Ross, W.D. (Trans.), Oxford: Oxford
University Press, rev. edn.
B.vc. M.vcUs, RU1u. 18o. Moral dilemmas and consistency. journal of Philos-
ophy ,8, 1i11o.
Biuvi, Fv.x. 1,,. Meditative-polemic should in modern English that-clauses.
Stockholm: Almqvist and Wiksell.
Bi11iv, M.vi.. io11. Time and modality without tenses or modals. In Moix.
R.1uiv1 & Ri.1i MUs. (Eds.), Tense across languages, pp. 1,188. Tbin-
gen: Niemeyer.
BoUm., Lowiii. 1,,. On contrasting the semantics of the modal auxiliaries of
German and English. Lingua ,, 1.
Bvix, D.viu O. 1. Moral confict and its structure. Te Philosophical Review
ro, i1,i,.
Bvnii, Jo., Rivivi Pivxis, & Wiiii.m P.ciiUc.. 1. Te evolution of gram-
mar. Tense, aspect, and modality in the languages of the world. Chicago: University
of Chicago Press.
Cu.viow, N.1i. io11. What we know and what to do. Synthese pp. 1.
Ciosi, Jo.i & B.s A.v1s. io1o. Current change in the modal system of En-
glish: A case study of must, have to and have got to. In UvsUi. Lixiv, JUui1u
HUniv, &Roniv1 M.iiu.mmiv (Eds.), Te history of English verbal and nominal
constructions, pp. 1o,181. Amsterdam: John Benjamins.
Co.1is, Jiiiiv. 18. Te semantics of the modal auxiliaries. London: Croom
Helm.
Coiiis, Pi1iv. 11. Te modals of obligationandnecessity inAustralianEnglish.
In K.vi Ai,miv & Bic1 Ai1inivc (Eds.), English corpus linguistics. Studies
in honour of jan Svartvik, pp. 1,1o,. London: Longman.
Couov.vui, Ciio. iooi. Temporal interpretation of modals: Modals for the
present and for the past. In D.viu I. Bi.viv, LUis D. C.siii.s M.v1iz,
Bv.uv Z. Ci.vx, & S1ii. K.Uim. (Eds.), Te construction of meaning, pp.
,88. Stanford: CSLI Publications.

Coii, E.vi. 18i. Against moral dilemmas. Te Philosophical Review r, 8,,.

Coviiv, Bviuci1. iooo. What should should mean: MS, CNRS/Universit Paris 8.
CUvmi, Giovci O. 11. A grammar of the English language. Syntax, vol. . New
York: D.C. Health.
D.cv, Jo.1u.. 1. Moral reasons. Cambridge: Blackwell.
D.viuso, Do.iu. 1o. How is weakness of the will possible: In Essays on
actions and events, pp. i1i. ioo1. New York: Clarendon Press, ind edn.
Do.c., Ai.. 18. Consistency in rationalist moral systems. journal of Philos-
ophy 8r, i1o.
Do.c., Ai.. 1. Moral dilemmas, genuine and spurious: A comparative
anatomy. Ethics ro,, ,i1.
Eisinivc, P.Ui D. 1oo. From the forbidden to the supererogatory: Te ba-
sic ethical categories in Kants Tugendlehre. American Philosophical Quarterly
, i,,io.
F.ccuii11i, Roniv1., M.iviu KvUc, & Fv.x P.imiv (Eds.). ioo. Modality
in contemporary English. New York: Mouton de Gruyter.
Fiium., Fviu. 18o. Doing the best we can. an essay in informal deontic logic. Dor-
drecht: Reidel.
Fii.v, S1ivui. ioo. Oughts and ends. Philosophical Studies r,, 1,o.
Fii.v, S1ivui. io1o. What ought probably means, and why you cant detach it.
Synthese r,,, o,8.
vo Fi1ii, K.i. 18. Te presupposition of subjunctive conditionals. In Uii
S.Uivi.u & Ovi PivcUs (Eds.), MIT working papers in linguistics. Te inter-
pretive tract, vol. i,. Cambridge: MIT Press.
vo Fi1ii, K.i & A1uov S. Giiiiis. io1o. Must stay strong! ^atural
Language Semantics r8, ,18.
voFi1ii, K.i &S.nii I.1viuoU. ioo,. What to do if you want to go to Harlem:
Anankastic conditionals and related matters. MS, MIT.

vo Fi1ii, K.i & S.nii I.1viuoU. ioo8. How to say ought in foreign: Te
composition of weak necessity modals. In J.cqUiiii GUivo & J.cqUiiii
Lic.vmi (Eds.), Time and modality, pp. 11,11. Springer.
Fiiiscum., SUz.i. 18. Temporal distance: A basic linguistic metaphor.
Studies in Language r, 1,o.
Foo1, Puiiivv.. 18. Moral realism and moral dilemma. Te journal of Philosophy
8o, ,8.
v. Fv..ssi, B.s C. 1,. Values and the hearts command. Te journal of Phi-
losophy ,o, ,1.
G.mU1, L.T.F. 11. Logic, language, and meaning, Volume r. Introduction to logic.
Chicago: University of Chicago Press.
G.zu.v, Giv.iu. 1,. Pragmatics. Implicatures, presupposition, and logical form.
New York: Academic Press.
Ginn.vu, Aii.. iooo. Reply to critics. Philosophy and Phenomenological Research
,:, ,i,.
Gonii, LoU. ioo. Normative conficts and the logic of ought. ^ous ,, ,o8.
Gow.s, Cuvis1ovuiv W. (Ed.). 18,. Moral dilemmas. New York: Oxford Uni-
versity Press.
Gvici, H.P. 18. Studies in the ways of words. Cambridge: Harvard University Press.
Gvoiisim., M.v,oiii. 1,. Can, may, must, and should: A Relevance theoretic
account. journal of Linguistics r, ,,.
H.iiiu.v, M.A.K. 1,o. Functional diversity in language as seen from a consider-
ation of modality and mood in English. Foundations of Language o, iio1.
H.iiii, Puiiiv. 1,. Lest innocent blood be shed. New York: Harper.
H.vi, R.M. 181. Moral thinking. Its levels, method, and point. Oxford: Clarendon
Press.
H.vm., Giiniv1. 1,. Review of Roger Wertheimer, Te signifcance of sense.
Meaning, modality, and morality. Te Philosophical Review 8:, i,i.
,
H.vm., Giiniv1. 1,,. Reasons. Crtica. Revista Hispanoamericana de Filosofa
,, 1,.
H.vm., Giiniv1. 1. Stringency of rights and ought. Philosophy and Phe-
nomenological Research ,, 18118,.
Hivu, D.viu. 18i. Supererogation. Its status in ethical theory. Cambridge: Cam-
bridge University Press.
Hiii, Jv., Tuom.s E. 1o. Moral dilemmas, gaps, and residues: A Kantian per-
spective. In M.so (1o), pp. 1o,18.
Hovviv, P.Ui J. & Eiiz.ni1u Cioss Tv.Uco11. ioo. Grammaticalization. Cam-
bridge: Cambridge University Press.
Hovc., Tivvv & M.vx Timmos. io1o. Untying a knot from the inside out:
Refections on the paradox of supererogation. Social Philosophy and Policy :,,
io.
Hov, L.Uvici R. 1,i. On the semantic properties of logical operators in English.
Ph.d. thesis, University of California, Los Angeles.
Hov, L.Uvici R. 18. A natural history of negation. Chicago: University of
Chicago Press.
Hov1v, Jou F. 1. Moral dilemmas and nonmonotonic logic. journal of Philo-
sophical Logic :, ,o,.
Hov1v, Jou F. ioo. Reasoning with moral conficts. ^ous ,, ,,,oo,.
Hov1v, Jou F. io1i. Reasons as defaults. New York: Oxford University Press.
HUuuiis1o, Rouiv & Gioiiviv K. PUiiUm(Eds.). iooi. Te Cambridge gram-
mar of the English language. Cambr: Cambridge University Press.
I.1viuoU, S.nii. iooo. Te grammatical ingredients of counterfactuality. Lin-
guistic Inquiry r, i1i,o.
I.1viuoU, S.nii & Hiuui Zii,is1v.. io1i. Negation, polarity, and deontic
modals. MS, MIT and University of Amsterdam.
Jois, Auviw J.I. & Icm.v Pov. 18o. Ought and must. Synthese oo, 8.
o
Joos, M.v1i. 1o. Te English verb. Formand meanings. Madison: Te University
of Wisconsin Press.
K.c., Suiiiiv. 18. Te limits of morality. Oxford: Oxford University Press.
K.1, Imm.Uii. 1o. Te metaphysics of morals. M.vv Gvicov (Ed.), Cam-
bridge: Cambridge University Press.
K.v11Ui, L.Uvi. 1,i. Possible and must. In J. Kimn.ii (Ed.), Syntax and se-
mantics, vol. 1, pp. 1io. New York: Academic Press.
K.sviv, W.i1iv. 1i. Presuppositions, composition, and simple subjunctives.
journal of Semantics , o,1.
Kv.1ziv, Aciiix.. 1,,. What must and can must and can mean. Linguistics
and Philosophy r, ,,,.
Kv.1ziv, Aciiix.. 181. Te notional category of modality. In H.s J0vci
Eixmiviv & H.is Riisiv (Eds.), Vords, worlds, and contexts. ^ew approaches
in word semantics, pp. 8,. Berlin: de Gruyter.
Kv.1ziv, Aciiix.. 11. Modality/Conditionals. In AvimvoS1icuow&Di-
i1iv WUuiviicu (Eds.), Semantics. An international handbook of contemporary
research, pp. oo,o. New York: de Gruyter.
L.xoii, Roni. 1,ia. Language in context. Language ,8, o,i,.
L.xoii, Roni. 1,ib. Te pragmatics of modality. In Proceedings of the Chicago
Linguistics Society 8, pp. iiio. Chicago.
L.vxi, Do. 1,o. Some notes on English modals. In J.mis McC.wiiv (Ed.),
Syntax and semantics ,. ^otes from the linguistic underground, pp. 8,8. Lon-
don: Academic Press.
L.ssi1iv, D.iii. io11. Measurement and modality. Te scalar basis of modal se-
mantics. Ph.D. thesis, New York University.
Liicu, Gioiiviv. ioo. Modality on the move: Te English modal auxiliaries
1o11i. In F.ccuii11i i1 .i. (ioo), pp. iiio.
Liicu, Gioiiviv, M.vi.i HUu1, Cuvis1i. M.iv, & Nicuoi.s Smi1u.
ioo. Change in contemporary English. A grammatical study. Cambridge: Cam-
bridge University Press.
,
Liicu, Gioiiviv N. 1,o. Towards a semantics description of English. Bloomington:
Indiana University Press.
Liicu, Gioiiviv N. 1,1. Meaning and the English verb. London: Longman.
Limmo, E.J. 1oi. Moral dilemmas. Philosophical Review ,r, 11,8.
Liwis, D.viu. 1,. Counterfactuals. Cambridge: Harvard University Press.
Liwis, D.viu. 1,a. A problem about permission. In E. S..vii, R. Hiivii,
I. NiiiiUo1o, & M.P. Hi1ixx. (Eds.), Essays in honour of jaakko Hintikka, pp.
1o1,,. Dordrecht: Reidel.
Liwis, D.viu. 1,b. Scorekeeping in a language game. journal of Philosophical
Logic 8, ,.
Liwis, D.viu. 181. Ordering semantics and premise semantics for counterfactu-
als. journal of Philosophical Logic ro, i1,i.
Lvos, Jou. 1,,. Semantics, vol. i. Cambridge: Cambridge University Press.
M.so, H.E. (Ed.). 1o. Moral dilemmas and moral theory. New York: Oxford
University Press.
M.11uiwso, Lis.. io1o. Cross-linguistic variation in modality systems: Te role
of mood. Semantics and Pragmatics , 1,.
McCoiii, Tivv.ci. 1o. Moral residue and dilemmas. In M.so (1o),
pp. o,.
McN.m.v., P.Ui. 1o. Te deontic quadecagon. Ph.d. thesis, University of Mas-
sachusetts.
McN.m.v., P.Ui. 1o. Must I do what I ought: (Or will the least I can do do:).
In M.vx A. Bvow & Josi C.vmo (Eds.), Deontic logic, agency, and normative
systems, pp. 1,1,. Berlin: Springer-Verlag.
Miiiim., Gvicovv. 11. Beyond the call of duty. Supererogation, obligation, and
oence. Albany: SUNY Press.
Miii, Jou S1U.v1. 18. Utilitarianism. Rociv Cvisv (Ed.), New York: Oxford
University Press.
8
Moovi, G.E. 1o. Principia ethica. New York: Cambridge University Press.
Mvuiii, Jou. 1,. Change and continuity in the functions of the American En-
glish modals. Linguistics , 1,,i11.
Mvuiii, Jou. 1o. Te development of the strong obligation systemin American
English. American Speech ,r, 88.
Mvuiii, Jou. 1,. Should and ought: Te rise of individually-oriented modality
in American English. English Language and Linguistics r, i.
Niw, Cuvis1ovuiv. 1,. Saints, heroes, and utilitarians. Philosophy ,, 1,18.
Ni., Diiiv. ioo,. Two puzzles about deontic necessity. In JoG.,iwsxi, V.ii-
1ii H.cqU.vu, Biv.vu Nicxii, & Si1u Y.ici (Eds.), MIT working papers
in linguistics. ^ew work on modality, vol. ,1, pp. 11,8.
NUv1s, J., Pii1iv Bvioo, &J.ixi Diivivii. ioo,. On deontic modality, di-
rectivity, and mood: A case study of Dutch mogen and moeten. In Antwerp papers
in linguistics, vol. 11o. Center for grammar, cognition, and typology, University
of Antwerp.
P.imiv, F.R. 18,. Te English verb. Longman: Longman, ind edn.
P.imiv, F.R. 1o. Modality and the English modals. New York: Longman, ind edn.
P.imiv, F.R. ioo1. Mood and modality. Cambridge: Cambridge University Press,
ind edn.
Pivxis, Micu.ii R. 18. Modal expressions in English. Norwood, NJ: Ablex.
Pi.ci1, Ji.. 1oi. Te moral judgment of the child. New York: Collier Books.
Pii1vosxi, P.Ui M. 1. Prima facie obligations, ceteris paribus laws in moral
theory. Ethics ro, 8,1,.
Pov1iv, P.Ui. 1,. Te semantics of mood, complementation, and conversa-
tional force. ^atural Language Semantics ,, 1o,i1i.
Pov1iv, P.Ui. ioo,. Imperatives and modals. ^atural Language Semantics r,,
,18.
Pov1iv, P.Ui. ioo. Modality. Oxford: Oxford University Press.

Pov1iv, P.Ui & Av.1 RUnis1ii. io1i. Mood and contextual commitment.
In Ac. Cuivicuis (Ed.), Proceedings of SALT ::, vol. ii, pp. o18,. Ithaca:
CLC Publications.
R.wis, Jou. 1,1. A theory of justice. Cambridge: Harvard University Press.
R.z, Josivu. 1,,. Permissions and supererogation. American Philosophical Quar-
terly r:, 1o11o8.
Ricu.vus, D.viu A.J. 1,1. Atheory of reasons for action. Oxford: Clarendon Press.
v. Roov, Roniv1. iooo. Permission to change. journal of Semantics r,, 111,.
Ross, W.D. 1o. Te right and the good. Oxford: Oxford University Press.
RUnis1ii, Av.1. io11. Projective signals of weak necessity modals. In Gvi-
covv Kiivs1i.u (Ed.), ESSLLI :orr. Vorkshop on projective content, pp. 111,,.
Ljubljana.
RUnis1ii, Av.1. io1ia. Figuring out what we ought to do. Poster for the Penn
Linguistics Colloquium (PLC o).
RUnis1ii, Av.1. io1ib. Roots of modality. Ph.d. thesis, University of Mas-
sachusetts.
Scuiixiv, Puiiivvi. ioo,. Te lazy Frenchmans approach to the subjunctive
(Speculations on reference to worlds and semantic defaults in the analysis of
mood). In Tw. Giiv1s, Ivo v. Giixi, & H.ixi J.cons (Eds.), Romance
languages and linguistic theory :oo. Selected papers from Going Romance r,, pp.
ioo. Nijmegen: John Benjamins.
ScuUiiiv, D.viu. io11. Te syntax and semantics of implicit conditionals. Filling in
the antecedent. Ph.d. thesis, University of California, Los Angeles.
Si.vii, Jou R. 18o. Prima-facie obligations. In Z.x v. S1v..1i (Ed.), Philo-
sophical subjects. Essays presented to P.F. Strawson, pp. i8i,. Oxford: Oxford
University Press.
Siix, Aiix. io1i. Modality, weights, and inconsistent premise sets. In Ac.
Cuivicuis (Ed.), Proceedings of SALT ::, pp. o. Ithaca: CLC Publications.
o
Siix, Aiix. io1a. Deontic conditionals: Weak and strong. MS, University of
Michigan.
Siix, Aiix. io1b. Modality, weights, and inconsistent premise sets. MS, University
of Michigan.
Siix, Aiix. io1c. Mood, necessity, and collective commitment. MS, University of
Michigan.
Siix, Aiix. io1. What normative terms mean and why it matters for ethical the-
ory. In M.vx Timmos (Ed.), Oxford studies in normative ethics. Oxford: Oxford
University Press.
Siom., A.vo. 1,o. Ought and better. Mind ,, 8,.
Smi1u, Nicuoi.s. ioo. Changes in modals and semi-modals of strong obligation
and epistemic necessity in recent British English. In F.ccuii11i i1 .i. (ioo),
pp. i1ioo.
S1.i.xiv, Roniv1. 1,. Pragmatic presuppositions. In S1.i.xiv (1), pp.
,oi.
S1.i.xiv, Roniv1. 1,,. Indicative conditionals. In S1.i.xiv (1), pp.
o,,.
S1.i.xiv, Roniv1. 1. Context and content. Essays on intentionality in speech
and thought. Oxford: Oxford University Press.
S1.i.xiv, Roniv1. iooi. Common ground. Linguistics and Philosophy :,,
,o1,i1.
Sw.so, Evic. ioo8. Modality in language. Philosophy Compass , 111io,.
Sw.so, Evic. io11. On the treatment of incomparability in ordering semantics
and premise semantics. journal of Philosophical Logic ,o, o,1.
Sw.so, Evic. io1ia. Te application of constraint semantics to the language of
subjective uncertainty. journal of Philosophical Logic pp. 1i8.
Sw.so, Evic. io1ib. Imperative force in the English modal system. Handout,
Philosophy-Linguistics Workshop, University of Michigan.
1
Swii1siv, E. 1o. From etymology to pragmatics. Cambridge: Cambridge Univer-
sity Press.
Tuomso, JUui1uJ.vvis. 1o. Te realmof rights. Cambridge: Harvard University
Press.
Uvmso, J.O. 1,8. Saints and heroes. In A.I. Miiui (Ed.), Essays in moral phi-
losophy, pp. 18i1o. Seattle: University of Washington Press.
V. Liui, A & Ji.-Cuvis1ovui Vivs1v.i1i. ioo8. Te nature and origins
of counterfactuality in simple clauses: Cross-linguistic evidence. journal of Prag-
matics ,o, 18o,18,.
Vii1m., Fv.x. 1,o. Prejudices, presuppositions, and the theory of condition-
als. In Jivoi Gvoiiui,x & M.v1i S1oxuoi (Eds.), Amsterdam papers in
formal grammar, vol. 1, pp. i8i81. Central Interfaculteit, University of Ams-
terdam.
Vivs1v.i1i, Ji.-Cuvis1ovui. ioo1. Subjective and objective modality: Inter-
personal and ideational functions in the English modal auxiliary system. journal
of Pragmatics , 1,o,1,i8.
W.vsnv, A.. iooo. (De)coding modality. Te case of must, may, maste, and kan.
Lund: Lund University.
Wiiis, Govuo. 1,. Learning and using the auxiliary verb in English. In V. Lii
(Ed.), Language development, pp. i,oi,o. London: Croom Helm.
Wiviv, Tom. ioo. Deducing the future and distinguishing the past. Temporal inter-
pretation in modal sentences in English. Ph.d. thesis, Rutgers University.
Wiviv, Tom. ioo,. You do what you gotta do, or why must implies will. In
Ei1uvmi. Giovc.i. & Jo.1u. Howiii (Eds.), Proceedings of SALT r,, pp.
io8. Ithaca: CLC Publications.
Wiv1uiimiv, Rociv. 1,i. Te signifcance of sense. Meaning, modality, and moral-
ity. Ithaca: Cornell University Press.
Wiivznicx., A.. 18,. Te semantics of modality. Folia Linguistica :r, i,.
Wiiii.ms, Biv.vu. 1,. Ethical consistency. In Problems of the self, pp. 1oo18o.
Cambridge: Cambridge University Press.
i
Wiiii.ms, Biv.vu. 181a. Moral luck. Cambridge: Cambridge University Press.
Wiiii.ms, Biv.vu. 181b. Practical necessity. InWiiii.ms (181a), pp. 1i11.
Woisi1scui.iciv, Evicu Fviiuvicu. 1,,. A semantic theory of the English auxil-
iary system. Ph.d. thesis, MIT.
Ziiciiiv, Dinv.. iooo. Hypothetical modality. Grammaticalization in an L: dialect.
Amsterdam: Benjamins.
Zimmivm., Micu.ii J. 1o. Te concept of moral obligation. Cambridge: Cam-
bridge University Press.