• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Epistemic modality
quickinfo

In epistemic (or non-root) usage, modal verbs are used to evaluate the probability or certainty of an entire proposition as true, factual or valid. Thus, in (1a) moet must indicates an obligation in regard to a specific action, but in (1b) the speaker’s evaluation of the truth of the entire proposition.

1
a. Hy moet vandag hierdie takie afhandel.
he must.AUX.MOD today this matter.DIM finalise.INF
He must complete this task today.
b. Hy moet te siek wees om werk toe te kom.
he must.AUX.MOD too ill be.INF for.COMP work to PTCL.INF come.INF
He must be too ill to come to work.

This difference becomes very subtle in the case of sal/gaan, which are used in hypothetical situations, as in the following where sal may predict a certain situation (root sense) or express an evaluation of the degree of probability of a situation already existing.

2
Die verbruiker sal beslis die slagoffer wees van die prysverhogings.
the consumer will.AUX.MOD definitely the victim be.INF of the price.hikes
The consumer will definitely be the victim of the price hikes.

Epistemic differs from non-epistemic usage in that while the latter has the force of a speech act (here: ‘prediction’), epistemic usage is mainly intended to indicate the degree of probability that a certain proposition is true.

The proposition may be in any tense, but if not marked as a past tense, as in (3a), needs to be differentiated from the root sense to ensure an epistemic interpretation. In (3b) this is achieved through the explicit usage of a durative construction, indicating present relevance.

3
a. Sy kan 'n nuwe aria gesing het.
she can.AUX.MOD a new aria sing.PST.PTCP have.AUX
She could have been singing a new aria.
b. Sy kan 'n nuwe aria aan die sing wees.
she can.AUX.MOD a new aria on the sing.NMLZ be.INF
She may be singing a new aria.

The degree of factuality ascribed to the proposition in question is determined by the degree of probability associated with the modal used, but is also a function of its specific root sense. Thus in (4a) moet expresses ‘necessity’, which translates into ‘necessary truth’;  in (4b) sal expresses high probability, but here on the basis of predictable behaviour (perhaps as previously experienced by the speaker), and in (4c) kan as ‘possibility’ directly expresses epistemic ‘median probability’.

4
a. Hy moet siek wees, want hy is nie vandag hier nie.
he must.AUX.MOD ill be.INF because.CNJ he is not today here PTCL.NEG
It is highly likely that he is ill, because he isn’t here today.
b. Hy sal siek wees, want hy is nie vandag hier nie.
he will.AUX.MOD ill be.INF because.CNJ he is not today here PTCL.NEG
He is most probably ill, because he isn't here today.
c. Hy kan siek wees, want hy is nie vandag hier nie.
he can.AUX.MOD ill be.INF because.CNJ he is not today here PTCL.NEG
He may be ill, because he isn’t here today.

The use of the preterites of epistemic modals (kan can, moet must and sal will) is  idiosyncratic.  While kon could lowers the likelihood of a proposition being true (5a), moes must.PRT is not differentiated from moet must.PRS in epistemic usage (5b), and sou would has an evidential rather than an epistemic function (5c), e.g.

5
a. Sy kon 'n nuwe aria gesing het.
she can.AUX.MOD.PRT a new aria sing.PST.PTCP have.AUX
She might have sung a new aria.
b. Sy moes 'n nuwe aria gesing het.
she must.AUX.MOD.PRT a new aria sing.PST.PTCP have.AUX
She must have sung a new aria.
c. Sy sou 'n nuwe aria gesing het.
she shall.AUX.MOD.PRT a new aria sing.PST.PTCP have.AUX
She is said to have sung a new aria.

An epistemic interpretation is precluded when the modal falls within the scope of the auxiliary het have, has as in (6). (See Construction 2).

6
Sy het 'n nuwe aria moes sing.
she have.AUX a new aria must.AUX.MOD.PRT sing.INF
She had to sing a new aria.
readmore
[+]Epistemic meaning, probability and evaluation

Epistemic usage is a meta-function of modals in which modal values are employed by a speaker with the sole purpose of evaluating the probability or certainty of a proposition being true, factual or valid. Broekhuis et al. (2015:141-142) state that epistemic usage concerns the degree of certainty assigned to the truth of a proposition by an individual on the basis of his knowledge state, and that subjective assessment is an essential part of the meaning of epistemic modal verbs(Broekhuis et al. 2015:143). The difference between root meaning and epistemic usage is made clear by the following pairs, where (7a) expresses 'obligation', but (7b) in the context only expresses the speaker's strong suspicion that the referent is a good pianist.

7
a. Hy moet vandag hierdie takie afhandel.
he must.AUX.MOD today this matter.DIM finalise.INF
He must deal with this matter today.
b. “Jou broer moet baie goed klavier speel,” sê June.
your brother must.AUX.MOD very well piano play.INF say June
"Your brother must play the piano very well," says June.
C.Morton: Helena, 2013, 15

However, the difference becomes more subtle when modals such as sal and gaan, which refer semantically to hypothetical situations, are used, as in (8).

8
Die verbruiker sal beslis die slagoffer wees van die prysverhogings.
the consumer will.AUX.MOD definitely the victim be.INF of the price.hikes
The consumer will definitely be the victim of the price hikes.

Two interpretations are possible: (i) a prediction that price-hikes are under way, and (ii) the degree of probability in an existing situation. Only the second is epistemic. The difference is more subtle in the next pair of examples, where gaan in (9a) suggests 'intention', while gaan in (9b) expresses fear of a 'possibility'.

9
a. Ek is seker mense gaan jou wil gebruik.
I be.PRS sure people go.AUX.MOD you.2SG want.to.AUX.MOD abuse.INF
I am sure people are going to want to abuse you.
I. Roggeband: Noot, 2013, 145
b. en daar is selfs vrese dat salarisse nie betaal gaan kan word nie
and there be.PRS even fears that.COMP salaries not pay.PST.PTCP go.AUX.MOD can.AUX.MOD be.AUX.PASS.PRS PTCL.NEG
and there are even fears that salaries will not be able to be paid
Rapport,  2014/3/6

The difference between non-epistemic and epistemic usage hinges on whether the modal (i) has the force of a speech act, i.e. bringing about change in the situation by motivating the addressee (‘obligation’) or  promising/undertaking to do something or (ii) is mainly intended as an indication of (degree of) probability (typically but not exclusively referring to completed actions/situations).

Epistemic evaluation may apply to propositions in any tense, whether it be the past, the present or the future. In constructions including a perfect, proposition time is clearly tense-marked by the perfect tense geslaap het (see Construction 1), which makes an epistemic interpretation likely:

10
a. Sy moet geslaap het, want sy skrik haar boeglam.
she must.AUX.MOD sleep.PST.PTCP have.AUX because.CNJ she get.fright.PRS her.REFL lame
She must have slept, because she had the fright of her life.
J.Vermeulen: Geraamtes, 2000, 93

An epistemic interpretation is unlikely with an infinitival complement, as the direct control of the modal over the infinitive forces a root reading, as in:

11
Doringrosie moet nou slaap.
Sleeping Beauty must.AUX.MOD now sleep.INF
Sleeping Beauty must sleep now.

However, an epistemic reading of the modal is made possible by a syntactic separation between the modal and a continuous construction as complement, as in (12).

12
Doringrosie moet nou aan die slaap wees.
Sleeping Beauty must.AUX.MOD now on the sleep.NMLZ be.INF
Sleeping Beauty must be asleep/sleeping now.
[+]Root and epistemic meaning

There is a trading relationship between root and epistemic meaning, in that the level of certainty the speaker attributes to his/her proposition relates to or is derivable from the root meaning of the modal. For instance, the ‘obligation’ expressed by moet must and the ‘prediction’ expressed by sal shall both translate into  ‘high probability / likelihood’ or ‘certainty’. Moet derives 'certainty' from the binding force of ‘obligation’ and sal from the confidence presupposed by prediction.

13
a. Hy moet siek wees, want hy is nie vandag hier nie.
he must.AUX.MOD ill be.INF because.CNJ he is not today here PTCL.NEG
It is highly likely that he is ill, as he isn’t here today.
b. Hy het nie kom werk nie; hy sal weer siek wees.
he have.AUX not come.LINK work.INF PTCL.NEG he shall.AUX.MOD again ill be.INF
He didn’t come to work; he is sure to be ill again.

Kan can and mag may both express ‘possibility’, which is implied in the inherent potential of ‘capability’, and the relaxation of control in ‘permission’, respectively. In epistemic use, ascribing ‘possibility’ to an occurrence is tantamount to a belief that its likelihood is average or even low. Examples are:

14
a. Maar dit kan nie Marlene wees nie?
but.CNJ it can.AUX.MOD not Marlene be.INF PTCL.NEG
But it can’t be Marlene?
VivA-KPO
b. Sy mag nou weg wees, haar stem het agtergebly.
she may.AUX.MOD now away be.INF her voice have.AUX behind.stay.PST.PTCP
She may be gone now, her voice remained behind.
J.Miles: Op 'n dag, 2016, 230

The ‘expectation’ underlying behoort (te) be supposed  (to) is transferred to the speaker’s evaluation of an embedded proposition as ‘likely’.

15
Hulle behoort al by hierdie tyd al soveel ondervinding te gehad het
they be.supposed.AUX.MOD already by this time already so.much experience PTCL.INF have.PST.PTCP have.AUX
By this time they must have gained so much experience.
VivA-KPO

Hoef nie need not is not as binding as moet nie must not; consequently the speaker is not bound to the truth of the proposition to the same extent as epistemic moet:

16
a. Hy hoef nie haar pa te wees nie.
he need.AUX.MOD not her dad PTCL.INF be.INF PTCL.NEG
He need not be her father.

The use of the modal preterites kon, moes and sou   – the only modal preterites in common usage apart from wou – differs in an idiosyncratic way from their present tense usage. The change from kan, in (17a), to kon, in (17b) and (17c), is epistemically significant in that it lowers the likelihood of the embedded proposition.

17
a. Doringrosie kan in 'n diepe slaap wees.
Sleeping Beauty can.AUX.MOD in a deep sleep be.INF
Sleeping Beauty may be sleeping soundly.
b. Doringrosie kon in 'n diepe slaap wees.
Sleeping Beauty can.AUX.MOD.PRT in a deep sleep be.INF
Sleeping Beauty might be sleeping soundly.
c. Dit was gevaarlik vir 'n bees om in die riet in te gaan, want hy kon somaar wegsink en verdwyn.
it was dangerous for a head.of.cattle for.COMP in the reed in.POSTP PTCP.INF go.INF because.CNJ he can.AUX.MOD.PRT easily away.sink.INF and disappear.INF
It was dangerous for a head of cattle to go into the reeds, as it could just sink down and disappear.
J.D.Kestell: Christiaan de Wet, 1920, 4

The degree of likelihood of an embedded proposition  is, however, not affected by an alternation between moet must and its preterite moes; moes is an optional (and highly favoured) choice with propositions in the past tense and probably the result of a kind of retrograde sequence of tenses.

18
Doringrosie moet/moes baie diep geslaap het.
Sleeping Beauty must.AUX.MOD / must.AUX.MOD.PRT very deeply sleep.PST.PTCP have.AUX
Sleeping Beauty must have slept soundly.

Preterite moes is however not compatible with a present tense proposition:

19
* Doringrosie moes in 'n diepe slaap wees.
Sleeping Beauty must.AUX.MOD.PRT in a deep sleep be.INF
To mean: ‘It must have been the case that Sleeping Beauty is sleeping soundly now.’

When preterite sou is used instead of sal, the likelihood of  the proposition being true is no longer assessed by the speaker as in the previous instances, but by an external source. This is also true of Dutch, in that in the case of the preterite zou/zouden (in contrast to the preterites of moeten and kunnen) the speaker is the default value of the source in the present but not in the past tense according to Broekhuis et al. (2015:147). Sou therefore has an evidential rather than an epistemic function. Note that sou is then also compatible with a present tense proposition:

20
a. Doringrosie sal aan die slaap gemaak gewees het.
Sleeping Beauty will.AUX.MOD on the sleep.NMLZ make.PST.PTCP be.PST.PTCP have.AUX
Sleeping Beauty is likely to have been put to sleep.
b. Doringrosie sou aan die slaap gemaak gewees het.
Sleeping Beauty will.AUX.MOD.PRT on the sleep.NMLZ make.PST.PTCP be.PST.PTCP have.AUX
Sleeping Beauty is likely to have been put to sleep.
[+]The het + modal preterite + infinitive construction

An epistemic interpretation is precluded when the auxiliary het have, has precedes the modal. Wurmbrand (2001:184) points out that in German sentences with a modal in the scope of the perfective auxiliary haben have, as in (21), only a  root interpretation is possible.

21
Sue hat zu Hause arbeiten müssen.
Sue have.AUX at home work.INF must.INF
Sue was obliged to work at home.

This is also true for Dutch (22a) and Afrikaans (22b) (see Construction 2):

22
a. Sue heeft thuis moeten werken.
Sue have.AUX at.home must.INF work.INF
Sue had to work from home.
b. Sue het tuis moes werk.
Sue have.AUX at.home must.AUX.MOD.PRT work.INF
Sue had to work from home.

In sum, a change from kan to kon is an epistemic change (i.e. a decrease in the probability of what is expressed in the embedded proposition); a change from moet to moes when the embedded proposition is past tense, is a kind of retrograde sequence of tenses not affecting the present relevance of the epistemic evaluation, and a change from sal to sou brings about an evidential rather than an epistemic interpretation. In constructions where the auxiliary has scope over the modal (see Construction 2), the modal does not allow an epistemic interpretation.

References
  • Broekhuis, Hans, Corver, Norbert & Vos, Riet2015Syntax of Dutch. Verbs and verb phrasesComprehensive grammar resourcesAmsterdam University Press
  • Broekhuis, Hans, Corver, Norbert & Vos, Riet2015Syntax of Dutch. Verbs and verb phrasesComprehensive grammar resourcesAmsterdam University Press
  • Broekhuis, Hans, Corver, Norbert & Vos, Riet2015Syntax of Dutch. Verbs and verb phrasesComprehensive grammar resourcesAmsterdam University Press
  • Wurmbrand, Susanne2001Infinitives. Restructuring and clause structure.Mouton de Gruyter
printreport errorcite