Epistemic modality: degrees of likelihood
 |
| That must be her brother |
You may find epistemic modality referred to as propositional modality because it refers to the speaker / writer's view of the truthfulness of a proposition. We don't use that term in this guide but it is an alternative.
The example above is just one way in which a speaker / writer can
express the belief that a proposition has a high likelihood of being
true. The truth or otherwise of propositions is what epistemic
modality is about (from the Greek epistēmē (knowledge)).
In this example, we have used a modal auxiliary verb (must)
to express a high degree of likelihood and we can substitute others
such as may, could or might to suggest lower
levels of likelihood.
There is, however, a range of other ways to express the same
viewpoint. Here are some examples. Your task, should you
choose to accept it, is to divide them into 4 categories (there are
two examples of each).
Click here when you have an answer.
- That is definitely her brother
- I'm quite sure that's her brother
- I think that's her brother
- That might be her brother
- I doubt that's her brother
- That's probably her brother
- I'm almost certain that's her brother
- That has to be her brother
- Modal auxiliary verbs
- That might be her
brother
and
That has to be her brother
Both of these use modal auxiliary verbs, the first expressing low likelihood and the second expressing high likelihood.
Intonation and sentence stress will play a significant role here. Generally, the greater the stress on the modal auxiliary, the greater its effect:
stressing might lowers the likelihood
stressing has increases the impression of certainty. - Modal adjuncts (and disjuncts)
- That is definitely her
brother
and
That's probably her brother
The adverbs in these sentences are acting as adjuncts to express the speaker's attitude to the likelihood of the proposition being true.
Other adverbials, such as, without a doubt, by all accounts, to my mind etc. are possible adverbials of one kind or another performing the same epistemic function. - Attributive clauses
- I'm quite sure that's
her brother
and
I'm almost certain that's her brother
Here, we have the expression of the attribute the speaker writer is applying personally. Structurally, it is an adjective linked to the pronoun by a copular verb. In both cases the adjective is modified adverbially.
Other copular verbs are possible as in, e.g.:
I feel sure that's her brother
I'm growing more certain that's her brother
and so on. - Verbs of mental and verbal processes
- I think that's her
brother
and
I doubt that's her brother
Both these verbs refer to cognition but verbs normally associated with speaking are also commonly used as in, e.g.,
I say that's her brother
I suggest that's her brother.
Less commonly, and often informally, we can use other types of process verbs such as
I bet that's her brother.
 |
Where's the evidence? |
| The car must have come this way |
All four of the ways of expressing the speaker's view of a proposition's truth or likelihood can be used to express a range of types of epistemic modality. Following Palmer (2001) there are these:
- Speculative
- in which the speaker indicates that he or she is simply
guessing about the likelihood of a proposition being true.
For example:
Mary might be in the garden
Mary is possibly in the garden
I'm guessing that Mary is in the garden
I imagine Mary's in the garden
I doubt if he's at home
She might not want to come
In all these cases, we can paraphrase what is said as:
This is a possible but by no means certain conclusion.
This is sometimes referred to a judgemental modality because it concerns the speaker / writer's judgement alone. - Deductive
- in this case the speaker indicates that there is evidence to
back up what is said and the proposition has, therefore, more
chance of being true. For example:
Mary must be in the garden
Mary's certainly in the garden
I'm quite sure that Mary's in the garden
I conclude that Mary's in the garden
That can't be true
Mary won't be at home at this time
In all these cases, we can paraphrase what is said as:
This is the only possible conclusion from the evidence we have available.
This is sometimes referred to as evidential modality when what is said or written includes some notion of an assessment of the evidence as in, for example:
It's very cold in there so the heating can't be on. - Assumptive
- in this case the speaker is indicating not that he or she is
working on clues and evidence but that the proposition will be
true because of what is generally known about the situation and
the people involved. For example:
Mary will be in the garden at this time
Mary's always in the garden at this time
We can assume that Mary's in the garden
I assume Mary's in the garden
Mary won't be at work on Sunday
They won't be having dinner at this time of day
In all these cases, we can paraphrase what is said as:
This is a reasonable conclusion to draw from what is generally known.
For teaching purposes, this is a useful three-way distinction to make because it allows us to present and practise the forms and realisations discretely without getting confused about what is intended in terms of the speaker's implications and what the hearer can infer.
 |
Evidentiality |
English, in common with a very wide range of languages has no
special grammatical form which allows the speaker to signal that a
statement is dependent on hearsay or personal assumption concerning
its truth or likelihood.
Some languages, Japanese, for example, do have an evidential form of
the verb, often signalled by suffixes or other verb-form changes.
Arabic has a range of particle additions to the syntax which can
signal various levels of certainty.
One estimate, Aikhenvald (2004), suggests that around 25% of all
languages have evidential forms.
In English, therefore, evidentiality is often signalled through
the use of disjuncts such as:
I hear that
he's lost his job
I see that you have a new
car
I think that its going to
rain
As I hear it, the road will
be closed
As far as I understand, the
figures will be ready tomorrow
As far as I know, that's the
situation
They say he's pretty wealthy
It is said to be expensive
It seems the train will be
on time
it seems to me that the work
is poorly done
It looks like / appears it's
going to rain
Allegedly, she's being
investigated
Reportedly, that's a good
solution
Obviously / Clearly / Apparently /
Seemingly / Patently etc. that's wrong
Clearly, there are semantic issues rather than grammatical ones,
concerning the strength of the likelihood which users of the
language suggest in their choice of disjunct.
Stressing a part of the disjunct usually lowers the estimation of
certainty so, for example:
They say he's pretty
wealthy
emphasises the conclusion that he probably is
wealthy, but
They
say he's pretty wealthy
reduces the certainty.
When it is present, emphasising the pronoun as in
As far as
she knows, it's OK
removes the onus from the speaker concerning the truth of a
proposition.
This is a teachable area which does not depend on mastery of modal auxiliary verbs with their connected structural and semantic problems for learners but one which can significantly enhance the range and precision even low-level learners of English can use.
 |
Overlapping categories |
One other sort of modality overlaps epistemic modality and this is what is known as alethic modality (from the Greek word for truth, alēthia).
Usually, alethic modality expresses that which is necessarily
true by the way the world works. For example:
A square must have four sides
Parallel lines cannot meet
Lead will melt at 327.5°C
Parliament meets in Westminster
and so on.
Epistemic modality is also concerned with the truth or otherwise
of a proposition so it is easy to see where the overlap occurs.
The difference, however, lies in the view taken of a proposition:
- View 1: the world of the observer
- In this case, it is the language user's world in which the
truth or otherwise of a proposition is set. For example:
John must be at home by now
You can't be serious
That's probably not true
This is certain to be the right train
etc.
All these statements apply not to all possible worlds but only to the world of the participants in the exchange of meanings. In none of these cases can it be said that we are referring to truths outside that sphere. - View 2: all possible worlds
- Leaving aside the genuine possibility of other universes in
which the laws of physics and mathematics are different,
statements from this standpoint refer to universal truths, not
truths contingent on the setting. So, for example:
Early humans must have developed bipedalism
Humans cannot breathe in a vacuum
Vertebrates must have backbones
17 is a prime number
and so on.
None of these statements is dependent at all on the environment of the participants in an exchange and none is contingent.
The present simple form of the verb is often used to express alethic modality because, in English, it carries the sense of timelessness.
Some languages, English not being one, reserve a different form of modality to distinguish between alethic and epistemic modality.
Alethic modality is considered slightly more fully in the guide to deontic modality, linked below, because there is also an overlap between obligation and necessity.
 |
Seeing the wood for the trees |
We can take the four categories we looked at above one by one to do a bit of analysis.
There's a good teaching reason for doing this, of course, because it
focuses you and your learners on one style of epistemic modality at
a time and allows people to 'see the wood for the trees'.
It also helps you plan and analyse the language before you plunge
in.
Mixing up the four main ways to express likelihood in the classroom
is not a good way to present and practise the area because languages
vary in what's possible and what's natural. It can confuse and
bewilder and that's unhelpful.
 |
Modal auxiliary verbs |
| What's that?! |
The issue here is the strength of the modal auxiliary verbs. Some coursebooks and many teachers are tempted to invent some kind of cline between 0% and 100% likelihood of something being true. The verbs and other expressions are then placed on the cline where the teacher or course-book writer believes they ought to go. There are problems with this approach because:
- it is almost impossible to get two native speakers to agree exactly where on the cline each verb should go
- as we saw above, intonation and stress play a crucial role in expressing meaning. Stress a verb like must and you increase the certainty, stress a verb like might and you decrease the level of certainty
- the real difference is not in the level of certainty itself but in the level of certainty of whether something is true or not true.
A better approach is to consider three levels of likelihood and divide the expressions into positive and negative propositions. Here they are, expressed without using modal auxiliary verbs:
- High likelihood:
- That's definitely a unicorn
- That's certainly not a unicorn
- Mid likelihood:
- That's probably a unicorn
- That's probably not a unicorn
- Low likelihood:
- That's possibly a unicorn
- That's possibly not a unicorn
There is some utility in having a template scale such as:
on which you can hang the modal auxiliary verbs and other modal expressions but it needs to remain mobile so that you can move things to show how stress and intonation can affect how strong the expressions are and whether it increases or decreases the level of certainty communicated.
This is not the point to consider in any detail all the possible modal auxiliary verbs used in this sense. There are guides elsewhere on the site for that. Go to the overall map of modality, linked below, for more and navigate to what you want from there.
Now pause for a moment and consider which modal auxiliary verbs are used in each case. There are six choices:
- high likelihood something is true
- high likelihood something is not true
- mid likelihood something is true
- mid likelihood something is not true
- low likelihood something is true
- low likelihood something is not true
Click here when you have a list.
| Level | Type | Example | Complications and issues |
| High likelihood | Positive | That
must be the postman That has to be her brother |
Some will assert that, especially in spoken language, has to carries the stronger sense because it can be stressed on both syllables. |
| Negative | That
can't be the right answer That couldn't be a unicorn |
Both can't and
couldn't express certainty that a proposition is false.
However:
|
|
| Mid likelihood | Positive | He
should be here soon They ought to arrive about noon |
ought to is slightly rarer
in terms of expressing epistemic modality but both verbs
have more or less the same strength. Because it has
two syllables, ought to can, like have to, be more emphatically
stressed. Stressing either auxiliary verb here reduces the level of certainty communicated. |
| Negative | This
shouldn't take long This oughtn't to be a difficult job |
As with the positive, both verbs
carry the same sense of mid-level likelihood but, again,
ought to is somewhat rarer. Again, stressing either auxiliary verb here reduces the level of certainty communicated. |
|
| Low likelihood | Positive | There
might be a problem There could be a mistake here She may have a solution |
might, could and may
express pretty much the same likelihood although some will
aver that might is the weaker of the three. Stressing the modal auxiliary verb reduces the level of implied likelihood. |
| Negative | There
mightn't be an issue There may not be a problem |
Both mightn't and may
not express low-level unlikelihood. However:
|
 |
Sources of potential confusion with modal auxiliaries |
- These auxiliaries are also used to express degrees of duty
and obligation (deontic modality) as well as ability (dynamic
modality) and, if the use is not made
clear by very explicit contexts which include the speaker /
writer's intentions, there is a serious danger of confusing learners.
For example,
You could be a bit more helpful
expresses the speaker's certainty (epistemic modality) that the hearer is able to help but that the hearer is unwilling to help (dynamic modality) and, at the same time, implies that it is the hearer's duty to help (deontic modality). However,
You could reach it if you stood on the chair
expresses purely dynamic modality (ability), and
You could be mistaken
expresses epistemic modality (likelihood). - The auxiliary verbs may and might are
also sources of potential ambiguity, because the verbs
express both epistemic modality as in, e.g.:
It may rain
I might be delayed
and deontic modality as in, e.g.:
You may leave
She said I might ask questions at the end
so there can be some ambiguity.
Does, for example:
She may come with us
mean:
I permit her to come with us
or
There is a possibility that she will come with us?
In interrogative forms this is even more obvious so questions such as:
May he come?
and
Might she ask a question?
are often avoided and native speakers will prefer to say:
Do you think he will come?
or
Is she allowed to ask a question?
in order to be clear. - In the past, there are three issues:
- although the levels of likelihood are unchanged, the
form of e.g.,
That couldn't have been her brother
looks like a present perfect and learners may be tempted consider it conceptually like that. It is, however, in sense, akin to a past simple because the situations referred to are finished past ones. For more, see the guide to modality and aspect, linked in the list of related guides at the end. - although both have to and must are
used to express high levels of likelihood in the present, in
the past, there is a conceptual difference:
He had to be her brother
can imply that the speaker / writer made the deduction in the past or is making it now
He must have been her brother
implies the speaker / writer is only now making the deduction. - Some of these verbs are rarely used in the past to
express epistemic modality and are reserved almost
exclusively for deontic modality. Compare:
We ought to have arrived before 6 but the traffic was impossible (mid-level epistemic modality but shading into a sense of duty or obligation)
We oughtn't to have arrived before 6 (a sense of mid-level negative obligation)
He should be here by 6 (mid-level epistemic modality)
He should have been here by 6 (carrying a greater sense of obligation rather than a likelihood)
- although the levels of likelihood are unchanged, the
form of e.g.,
 |
Modal adjuncts and disjuncts |
| in my humble opinion |
As we saw above, modal adjuncts are often simple
adverbs or adverb phrases such as:
He's definitely her brother
That's probably her brother
That's very likely her brother
etc.
They come in two main flavours:
Amplifiers or emphasisers: definitely, certainly
etc.
Downtoners or softeners: possibly, conceivably,
apparently etc.
and a few are limiters which constrain the nature of the utterance:
Currently,
this must be the case
That was always true formerly
Prepositional phrases, such as, without any doubt, in all likelihood, at the moment, in principle etc. perform the same functions.
As we also saw above under evidentiality, disjuncts, especially of attitude
or content, are
frequently used. Style disjuncts, too may signal truthfulness
or doubt as in, e.g.:
In my opinion, that's her
brother
(style disjunct restricting the way the speaker wants to be
understood)
Truthfully, I have no idea
where I put it
(style disjunct of manner expressing the way the speaker wants
to be understood and implying it is true)
Arguably, this is a problem
Obviously, that's her brother
(both attitude or content disjuncts referring to the
truthfulness of the proposition)
There is a guide on this site for more on adverbials, linked in the list at the end.
 |
Personal opinion |
| I suspect she's lying |
In addition to the attitudinal or content disjuncts noted above, there are two further ways to express epistemic modality to consider. In the list above, they appear as:
- Attributive clauses
- The examples above were:
I'm quite sure that's her brother
and
I'm almost certain that's her brother
Attributes, whether modified adverbially or not, can be expressed in the whole range:
High perceived likelihood:
I am convinced that's (not) her brother
I feel absolutely sure that's her brother
She appears certain that that's her brother
Mid-level perceived likelihood:
I am partially convinced that's her brother
I am a bit doubtful about whether it's her brother or not
Low-level perceived likelihood:
I am very doubtful that's her brother
I feel quite sure that's not her brother
etc.
The most common copular verbs are appear and seem (for other people's views), feel and be (for personal views) but others such as grow, come and become are possible as in, e.g.:
I have come to the view that he doesn't like me
I have grown to understand the truth of the matter
She has become convinced of the validity
etc.
These attributive expressions can be combined with modal auxiliaries as in, e.g.,
I am sure that must be her brother
I am doubtful that could be her brother
to provide quite subtle shades of viewpoint expression. - Verbs of mental processes
- The examples above were:
I think that's her brother
and
I doubt that's her brother
A range of verbs are possible both in the positive and the negative. For example:
I (don't) think that's going to happen
I (don't) doubt that's the truth
I suspect that's her brother
I (don't) believe that's her brother
An added complication here is that English speakers often opt for what is known as transferred negation (see below).
 |
Hedging |
Epistemic modality is widely used in formal writing, especially
writing in English for Academic Purposes because expression which
seems overly assertive and dogmatic is generally avoided.
For example:
Instead of:
It is clear that ...
writers often prefer something like:
It may be successfully argued that ...
Academic writing makes full use of modal auxiliary verbs, modal
adjectives, modal nouns, adverbs and prepositional phrases in an
effort to sound reasonable and tentative rather than assertive and
over-confident.
Other examples are:
The discrepancy could be the result of ...
It might be argued that ...
It seems to be the case that ...
This tends to be the result of ...
It may, quite arguably, be the case that ...
It seems, as far as we can discover, to be the result of ...
The guide on this site to the use of hedging and cautious
language in EAP is linked in the list of related guides at the end.
It contains many more examples (including the six above) of how
epistemic modality expression is toned down to remain appropriately
tentative.
 |
A peculiarity of English |
When negating clauses containing mental process verbs, most languages will put the negation where it
belongs, i.e., with the verb that is being negated.
For example:
I think that he hasn't done it yet
I expect he won't come
I imagine she has no money
I think she's not very friendly
etc.
In all these cases, we have a verb signalling belief followed by a
subordinate clause which is negated in some way.
English, bizarrely, often chooses to negate the main clause and
leave the subordinate clause positive. This is called transferred negation. It only occurs with verbs
signalling belief or assumption (i.e., epistemic modality) including, but not limited to
believe, suppose, guess, fancy,
imagine, reckon, expect
For the three sentences above, for example, English speakers would
naturally select:
I don't think he's done it yet
I don't expect he will come
I don't imagine she has any money
I don't think she's very friendly
The illogicality of transferred negation is revealed when a
questions tag is included so we get, for example:
I don't think he's done it yet, has he?
instead of the expected, do I?
To make matters worse, not all such verbs allow this transfer.
For example, assume and presume do not work this
way:
I assume you haven't paid [assumption = you haven't
paid]
is not the same as
I don't assume you've paid [no assumption made]
The verb hope is also odd because it will not allow
the transfer:
I don't hope it rains [= I do not hope for rain but I may
expect rain]
is not the equivalent of
I hope it doesn't rain [= I am hoping for no rain]
See the guide to negation in English for more, linked from the list below. There is also a PDF file with a list of verbs which do and do not allow transferred negation, available here.
| Related guides: | |
| non-modal-verb modality | covers some of the above and extends it to include epistemic and other forms of modality |
| types of modality | for an overview of the area |
| dynamic modality | which follows a similar format to this one |
| deontic modality | which follows a similar format to this one |
| multiple modalities | for a guide which considers how different types of modality may be combined in single utterances |
| modality and aspect | for more on the use of modal auxiliary forms in complex tenses |
| the modality map | for more choices |
| negation | to find out more about transferred negation |
| ambiguity | for a guide which considers the polysemous nature of many modal auxiliary verbs and the ambiguity which can arise |
| hedging in EAP | in academic writing epistemic modality is widely used to write cautiously |
| adverbials | for more on adjuncts and disjuncts (and much else) |
References:
Aikhenvald, AY, 2004, Evidentiality, Oxford: Oxford
University Press
Palmer, FR, 2001, Mood and Modality, 2nd edition,
Cambridge: Cambridge University Press