You are on page 1of 17

The Oxford Handbook of Modality and Mood

Jan Nuyts (ed.), Johan van der Auwera (ed.)

https://doi.org/10.1093/oxfordhb/9780199591435.001.0001
Published: 2014 Online ISBN: 9780191750199 Print ISBN: 9780199591435

CHAPTER

14 Modality and Mood in Oceanic 


Frantisek Lichtenberk

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


https://doi.org/10.1093/oxfordhb/9780199591435.013.15 Pages 330–361
Published: 09 July 2015

Abstract
The Oceanic languages form a subgroup within the Austronesian family. The chapter is a cross-
linguistic investigation of several kinds of modality: epistemic, deontic, dynamic, desiderative, and
timitive; and also of the realis and the irrealis moods. The modalities are expressed by various means:
a xes on verbs, particles associated with verbs, and verbs, which in some cases are not fully- edged
verbs. The timitive modality is a complex category: it simultaneously expresses epistemic modality and
apprehension about the possible state of a airs. While the validity of the realis–irrealis distinction has
been questioned in some of the general typological literature, the present chapter argues that the
distinction is useful in Oceanic, even though there are di erences of detail among the languages.

Keywords: Oceanic languages, modality, mood, deontic, desiderative, dynamic, epistemic, irrealis, realis,
timitive Oceanic languages, modality, mood, deontic, desiderative, dynamic, epistemic, irrealis, realis,
timitive
Subject: Semantics, Grammar, Syntax and Morphology, Linguistics
Series: Oxford Handbooks

14.1 Introduction1

THE Oceanic languages form a subgroup within Austronesian. There are 450-odd of them, spoken in Papua
New Guinea, (Island) Melanesia, Micronesia, and Polynesia. Given the large number of languages, it is not
surprising to nd great diversity in their modality and mood systems. Therefore, the discussion of them
here will, of necessity, not be very detailed. I will mainly discuss the dominant, recurrent types, and the
principal focus will be on the grammatical means of expressing modality and mood. These include in ection
on verbs and a variety of particles associated with verbs. Verbal expressions are included as well, focusing
on cases where the verb is not a fully- edged member of the verbal category and/or exhibits some
idiosyncratic properties. And in one case an apparently adverbial form will be included because it has
grammatical cognates in other languages. Grammatical descriptions of Oceanic languages vary greatly with
respect to the detail they provide on the modality and mood systems, and the part of speech of a mood- or
modality-expressing element is not always completely clear from the description. With respect to mood, the
focus here will be on the irrealis (and the realis), because the validity of the irrealis category has been
questioned in the typological literature.

We will see that in some cases one and the same element or construction can express more than one kind of
modality. However, because of a lack of historical records of any signi cant time depths, it is usually not
possible to make statements about the development of the polysemies.
This chapter deals with the following themes: epistemic modality (section 14.2), deontic modality (section
p. 331 14.3), dynamic modality (section 14.4), desiderative modality (section 14.5), timitive modality (section
14.6), and the realis–irrealis opposition (section 14.7). The nal section, 14.8, provides a brief summary.

14.2 Epistemic modality

In a number of languages epistemic modality is expressed by means of particles. In keeping with tradition in
Oceanic descriptive studies, the term “particle” is used here to designate non-a xal grammatical (non-
lexical) elements, which may, however, cliticize to a neighboring word. Verb-phrase internal epistemic
particles are found in, for example, Pukapukan and Standard Fijian. In Pukapukan, pā ‘perhaps, probably’ is
2
preverbal, while in Fijian, beka ‘perhaps, by chance’ is postverbal; see (1) and (2), respectively:

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


(1)

(2)

Toqabaqita has two clause-level epistemic particles. One of them is toqo ‘perhaps, possibly, I think, I guess’.
Toqo also functions as a verb-phrase level particle with a deontic force; see (19) in section 14.3. The other
particle is takona ‘most likely, certainly’ (also used to express surprise).

p. 332 (3)

(4)

Mokilese, according to Harrison (1976: 194), has “[a] number of … modal elements [that] defy
classi cation”, such as uhdahn ‘really, for sure’ and dapwa ‘perhaps’. He uses the term “adverbs” for them
in scare quotes. Uhdahn adds “a note of certainty” and dapwa a note of “doubt”. They are clause-level
elements. The certainty modal element is shown in (5):

(5)

Some languages have modal verbs. The verbs considered here may be auxiliaries or not fully- edged verbs.
However, it is not always clear why the elements in question are considered auxiliaries/verbs. Mokilese is
said to have a modal auxiliary verb pirin, which expresses intention or likelihood. In (6) it expresses “the
speaker’s opinion that the [event] in question [is] very likely to occur” (Harrison 1976: 174):

(6)

See also (5), where the auxiliary cooccurs with the modal element uhdahn ‘for sure’.

Pukapukan has a modal verb/‘predicate’ pēnei ‘maybe’, literally ‘like this’ (pe similative preposition, nei
‘near speaker’):

(7)

p. 333 Toqabaqita has a transitive verb quri ‘resemble, look like, be like’:

(8)

This verb can also be used as a higher verb to express epistemic possibility, in particular, inference: “it looks
like, it seems that”. It takes a clause that expresses the possible state of a airs as its direct object. The
complement clause is indexed on the verb by means of the object su x -a (see the su x in (8)). As its
subject the verb quri takes a nominalization of the verb riki ‘see, look at’; that is, “its looking-at/its
appearance is like X”. The evidence for the inference about a possible state of a airs is visual. In (9) the
subject of the complement clause, toqa loo ‘the people up there’, has been topicalized:

(9)

Being a verb, quri takes a subject-tense marker. In its epistemic function its subject marker is always third-
person singular. However, more often than not, in that function quri occurs without a subject marker:

(10)
While any verb can occur without a subject marker, this is not particularly common. With epistemic quri the
subject marker is absent quite commonly, which may be evidence that it is losing this particular verbal
p. 334 property.

14.3 Deontic modality

On the whole, dedicated forms for expressing deontic modality do not seem as common in Oceanic as those
expressing epistemic modality.

In some languages obligation is expressed by means of what are basically mood structures: imperatives,
hortatives, or more generally by constructions used to encode future states of a airs, including the irrealis.

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


In Manam, the irrealis functions in this way, as in (11). (For more on the Manam irrealis, see section 14.7.)

(11)

Similarly in Lolovoli:

(12)

Interestingly, Lolovoli has borrowed the deontic modal mas from Bislama, the Vanuatu variety of
Melanesian Pidgin. This is what Hyslop (2001) says about deontic modality in Lolovoli:

Before contact with English speakers there was no distinct way of expressing obligation or
necessity to perform an action: the irrealis mood was used to express obligation as well as simple
future time reference. In the modern language, however, the particle mas, from English “must”,
has been borrowed from Bislama with this function.

(Hyslop 2001: 256)

Verbs accompanied by mas are in the irrealis:

(13)

Bislama mas has been borrowed into other Vanuatu languages, such as Anejom̃ (considered a preverbal
“grammatical morpheme” there by Lynch and Tepahae 2001: 161) and South Efate (considered an auxiliary
p. 335 verb there by Thieberger 2006: 158).

Longgu has a preverbal obligative particle. In (14) it occurs in two phonologically-conditioned forms, go and
ge:

(14)

Māori has two obligative particles, both preverbal. One of them, me, is sometimes referred to as a “weak
imperative”, and is used “for a wide range of degrees of obligation, from advice to legal requirement”
(Bauer et al. 1997: 136). The sentence in (15) contains a plain imperative, and the one in (16) a weak
imperative with the obligative particle:

(15)

(16)

The other obligative particle, kia, is used “often when intention to comply with the obligation is involved”
(Bauer et al. 1997: 137):

(17)

However, according to Bauer et al., the two particles are sometimes just alternatives of each other.

Boumaa Fijian has a deontic element that Dixon calls a “semi-auxiliary verb”, dodonu ‘must, be necessary’.
What Dixon considers to be semi-auxiliaries “function syntactically as main verbs and take as subject an NP
p. 336 or complement clause—but semantically they qualify the predicate of their complement clause” (Dixon
1988: 279). The complement clause is introduced by the relator me ‘should’ (cf. Māori obligative me). The
subject of dodonu can be the same as the subject of the complement clause, or it can be impersonal, third-
person singular, as in (18a) and (18b), respectively:

(18)

Finally, Toqabaqita has a post-verbal particle toqo, used to express obligation. It co-occurs with the
assertive particle boqo:

(19)

However, toqo functions more commonly as an epistemic clause-level particle; see (3) in section 14.2. On the
whole, epistemic–deontic polysemy appears to be rare in Oceanic. More commonly one nds permission–
ability polysemy; see section 14.4.

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


14.4 Dynamic modality

Dynamic modality tends to be expressed in Oceanic by verbs. The verbal meanings involved include “be a
match for”, “be appropriate, be adequate”, and “be enough, be su cient”. Frequently, one and the same
form can express both participant-inherent and participant-external (or participant-imposed) dynamic
modalities (see Chapter 3), that is, dynamic possibility. In at least some cases the participant-inherent
meaning appears to be historically primary.

Toqabaqita has a transitive verb talaqa, which can be used with the non-modal meanings ‘(of a person) be a
p. 337 match for sb., (of something) be tting, appropriate for sb./st.’:

(20)

(21)

Talaqa is also used to express physical ability. The subject designates the agent. The object may be a nite
clause (the subject of the complement clause is coreferential with that of talaqa) or a nominalization. The
sentences in (22) and (23) are synonymous:

(22)

(23)

With nite complement clauses, the subject of talaqa, instead of designating the agent, may be impersonal,
third-person singular, regardless of the subject of the complement clause. Sentences (24) and (25) are
synonymous: in (24) the subject designates the agent; in (25) the subject is impersonal:

p. 338 (24)

(25)

Constructions with actor subjects express participant-inherent modality, as in (24), while the construction
with an impersonal subject may express participant-imposed modality, as in (26):

(26)

The construction with an impersonal subject can also be used to express deontic possibility, permission, as
in (27) and (28). In the impersonal type of construction it is not unusual for the verb talaqa to occur without
the (third-person singular) subject marker. This is the case in (27):

(27)

In the situation expressed in (27) it is the addressee that is asked for permission. On the other hand, in the
situation expressed in (28) the lack of permission is due to traditional cultural rules: in earlier times women
who were close to giving birth were not allowed to be near men.

(28)

p. 339 (For the timitive marker ada see section 14.6.)


Ability–possibility polysemy is also found in Manam, and there are some parallels between the
constructions in Manam and in Toqabaqita. (The two languages are only distantly related to each other.)
Manam has an intransitive verb boadu, which can be used in the sense of “have enough of something, be
su ciently provided with something”:

(29)

It can also be used with the sense of “be enough”:

(30)

(There is a related adjective boadu-boadu ‘powerful’ and a transitive verb boadu-ŋ ‘make something
su cient’, as for example in “provide a su cient amount of food”.) The intransitive verb can also express

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


physical ability and dynamic possibility. As is the case with the modality verb talaqa in Toqabaqita, Manam
boadu can occur in two types of constructions: (i) the subject of boadu designates the agent and is
coreferential with the subject of the verb expressing the action that can be performed; and (ii) the subject of
boadu is impersonal, third-person singular. Also as in Toqabaqita, the complement of boadu may be a nite
clause or a nominalization. For simplicity, only nite complement clauses will be exempli ed here. The
examples in (31) and (32) illustrate the two options. In (31) the subject of boadu designates the agent and it is
participant-inherent inability that is expressed, while in (32) the subject is impersonal and the event is
presented as participant-external impossibility:

p. 340 (31)

(32)

Similarly in (33) and (34). In (34) the subject of the verb “pierce” has been preposed to boadu (note the
plural subject marking on “pierce” and the singular impersonal marking on boadu).

(33)

(34)

There are additional complexities in the use of boadu to express ability and dynamic possibility; see
Lichtenberk (1983) for detail.

Boadu can also be used to express deontic possibility or permission. In that case it can only have an
impersonal subject, just as is the case in Toqabaqita.

(35)

With an impersonal subject, a sentence may be ambiguous between dynamic and deontic possibility:

(36)

p. 341 A similar kind of polysemy is found in Mangap-Mbula. There is a verb rao, which has the meaning ‘be
adequate, su cient’:

(37)

(Bugenhagen glosses rao as RAO in all the examples. Here the gloss ‘be adequate’ is used instead.)

And it is also used to express ability, as in (38), and permission, as in (39) and (40). The clause that
expresses the relevant state of a airs is optionally introduced by the non-factuality complementizer be:

(38)

Bugenhagen does not comment on the use of the subject markers with the verb, but in at least some cases
the subject marker is third-person singular, regardless of the subject of the other verb; see (39), where rao
signals permission:

(39)

In (40), where rao expresses lack of permission, it has a rst-person plural inclusive subject marker, just
like the verb expressing the event not permitted.
(40)

Bukawa is another language with ability–permission polysemy. Bukawa has the verb (gi/ndi)tôm ‘(to be)
p. 342 able, enough’, which “is commonly used to encode both abilitative and permissive moods by means of
complement clauses” (Eckermann 2007: 136). Example (41) shows the verb in the sense of “be t for, be
capable of” with a noun phrase as a non-clausal direct object:

(41)

In its modal uses, the verb may have a subject that designates the agent, co-referential with the subject of
the complement clause, or it may have an impersonal subject. In both (42) and (43) it is ability that is
expressed. In (42) the two verbs have coreferential subjects:

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


(42)

In (43) the modal verb has the impersonal third-person singular subject marker, while the verb of the
complement clause has the rst-person plural exclusive subject marker:

(43)

According to Eckermann, the same verb can also be used to express permission, although the two examples
he gives are not entirely clear. He says that “the form gitôm (bu) X (DO ) Y may have the sense “X can do Y/X
could have done Y”, but normally it encodes a question with the sense “can X do Y?” … Such questions are
often used when the speaker wants to request something in a polite manner …” (Eckermann 2007: 137). One
of the examples is given in (44). The subject is impersonal.

(44)

p. 343 It is possible that the intended meaning is “may we have some space/room?”.

Besides (gi/ndi)tôm ‘(to be) able, enough’, Bukawa also has the form gic/ndic (ŋa) waê ‘be proper/ tting’
(lit.: ‘hit its news/report/reputation’), which is used to express “an imperatival or cohortative mood”
(Eckermann 2007: 137):

(45)

Ability–permission polysemy appears quite common in Oceanic. The kind of ability involved here is
normally physical ability to do something, rather than mental ability. Besides the languages mentioned
above, this kind of polysemy is found in, for example, Mokilese (Harrison 1976: 174) and in Polynesian
languages. In Tuvaluan (Polynesian) only the verb of physical ability can also be used to express permission,
not the verb “know”, which “refers primarily to learned ability, but never permission” (Besnier 2000: 501).

14.5 Desiderative modality

Desiderative modality expresses desire, wish on the part of the referent of the subject to bring about a given
state of a airs. Here desiderative modality is considered distinct from volitive modality, which expresses
willingness on the part of the referent of the subject to bring about a given state of a airs. And
“desiderative” rather than “volitive” is the term used in Oceanic linguistics.

Polynesian and Fijian languages have a cognate preverbal desiderative adverb, particle or pre x. Examples
(46) and (47) show the elements in Samoan and in Boumaa Fijian, respectively. Mosel and Hovdhaugen
(1992) consider a to be an adverb in Samoan:

(46)

In Boumaa Fijian, via is a particle:

(47)

p. 344 In Pukapukan the cognate element is a fossilized pre x on a few verbs in “desiderative expressions of
physical processes” (Salisbury 2002: 105). Salisbury characterizes the meaning of wia- as ‘have a bodily
need’. All such verbs carry the su x -a. For example:
(48)

Similar use of a cognate desiderative pre x with a few verbs is found in, for example, Māori: hia-kai
‘hungry’ (kai ‘eat’), hia-moe ‘drowsy, sleepy’ (moe ‘sleep’) (Williams 1975).

In clauses with desiderative modality, the subject is normally human/animate. In Samoan and in Boumaa
Fijian the elements that express desiderative modality can also be used to express a kind of dynamic
modality, speci cally need of/necessity, as perceived (usually) by the speaker, that a given state of a airs
obtain. Here the subject need not be human/animate. In (49), from Samoan, the subject is inanimate; in (50)
from Boumaa Fijian, it is animate, non-human:

(49)

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


(50)

Compare (47) earlier, which expresses desiderative modality.

According to Blust (1980), these desiderative particles derive historically from the Proto-Malayo-
Polynesian verb *pian/*pia ‘want, desire, wish or long for’.

Another language with a desiderative marker, unrelated to the ones just discussed, is Toqabaqita. There the
marker/auxiliary-like element is related to the transitive verb “want”. The verb can take noun phrase non-
clausal objects (for example, “I want some food”) and object complement clauses. With clausal objects, the
subject of “want” and the subject of the complement clause may be di erent, as in (51), or identical, as in
p. 345 (52). In either case, the complement clause is nite, with its own subject/tense marker.

(51)

(52)

When the (notional) subjects of “want” and of the complement clause are identical, there is an alternative
construction available. Rather than the verb “want”, it is its detransitivized counterpart that is used. The
detransitivized form is a particle, an auxiliary-like element.

(53)

(There are several other transitive verbs that have detransitivized counterparts formed by means of –qi.)
While the construction with the transitive verb “want” is bi-clausal, the one with the detransitivized form
is monoclausal. (For evidence of the monoclausal structure, see Lichtenberk 2008b.)

Desiderative elements are found in other Oceanic languages. For example, Hoava has a preverbal optative
particle. The particle takes su xes that index the subject. The same su xes are also used in two types of
possessive noun phrases to index the possessor.

(54)

p. 346
14.6 Timitive modality
3
This type of modality is widespread in Oceanic. It was labeled “apprehensional-epistemic” in Lichtenberk
(1995), where detailed discussion of this modality type in Toqabaqita can be found. (The term
“apprehensive” is used in Hyslop 2001: 251.) Later, the term “timitive” was used for Toqabaqita
(Lichtenberk 2008a: 780), adopted from Palmer (2001: 131). The latter term will be used here in the general
discussion of the modality and in examples from Toqabaqita, but elsewhere the labels and glosses used in
the sources have been retained to show some of the range of the views on, and characterizations of, this
modality type.

Timitive modality is both epistemic and stance-expressing. It simultaneously expresses the speaker’s view
of the relevant state of a airs as (epistemically) possible, likely, and his or her stance: the speaker views the
possible state of a airs as undesirable or even as something (to be) feared. In other words, this is not a case
4
of polysemy: the notions of a state of a airs being possible and undesirable co-exist. Timitive modality is
typically expressed by means of a particle whose position varies depending on the language. In Toqabaqita
and Mangap-Mbula the particles are clause-initial; see (55) and (56), respectively:
(55)

(56)

(Bugenhagen glosses kokena as ‘kokena’ here, but see (59) below, where it is glossed ‘lest’.) In spite of the
single-sentence translation of (56), Bugenhagen makes it clear that the Mangap-Mbula example contains
two sentences, with kokena introducing the second one. In his grammar of the language, Bugenhagen (1995:
272) says that “kokena ‘lest’ … encode[s] states of a airs which are: 1) contemplated, and … 2) not desired”.
p. 347 He considers kokena to be a complementizer.

In other languages the timitive marker occurs in a position other than initial. In Manam it occurs post-
verbally. It serves to express caution, a warning about a possible state of a airs, and is called a “caveat”
marker:

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


(57)

Since timitive constructions are used to express possible, potential situations that are undesirable or feared,
in at least some languages they may occur in clauses that function as complements of verbs of fearing. This
is so in Toqabaqita, as in (58), and in Mangap-Mbula, as in (59):

(58)

(59)

In some languages the use of the timitive construction is restricted. For example, in Pukapukan the
“warning” particle is used “with a restricted range of verbs: those in the semantic domain of causing
misfortune or injury to oneself or others” (Salisbury 2002: 278):

(60)

And in Lolovoli “the apprehensive mood particle … is only used with second person subjects and … it
p. 348 includes a de nite warning component that there will be certain negative consequences if the action
described by the verb takes place” (Hyslop 2001: 251):

(61)

Sometimes an undesirable state of a airs may be prevented from taking place, and the timitive modality
construction may be used to express negative purpose: the otherwise possible, likely, or certain state of
a airs is to be avoided. This is so in (62) from Toqabaqita:

(62)

It appears that a timitive modality marker may develop into a simple epistemic modality marker to signal
possibility without any connotation of the possible state of a airs being undesirable. This is apparently
happening in Boumaa Fijian. Boumaa has a “relator” dee, which “introduces a type of complement where
there is an element of uncertainty involved…. Typically, the dee clause refers to some unpleasant possibility,
and the other clause to something that can be done either to avoid that possibility or to compensate for it”
(Dixon 1988: 260):

(63)

Nevertheless, dee need not signal that the possible state of a airs is undesirable: “However, a dee clause
p. 349 need not refer to anything unpleasant. The other clause may describe something which, if accomplished,
might yield a welcome result (referred to by the dee clause)” (Dixon 1988: 261; original emphasis):

(64)

Such neutral epistemic downtoning force is also found in Standard Fijian. There dē is typically used in a
timitive function, but not necessarily so:

(65)
14.7 Realis and irrealis

Quite a few grammars of Oceanic languages make reference to a realis–irrealis opposition. The validity of
the contrast has been questioned in some general typological literature because of the diversity of functions
that both the realis and the irrealis can have; witness Bybee et al.’s (1994: 238) suggestion that “this binary
distinction is not cross-linguistically valid”. More recently de Haan (2012: 128) concluded in a similar vein
that “there is no category that is marked uniformly as either realis or irrealis” (see also Chapters 5 and 9).
(However, see Palmer 2001 for a di erent view.) The purpose of this section is to discuss some of the
interpretations given to the two categories in Oceanic languages and some of the ranges of their use. To
anticipate, while it is true that there is some variability in how the contrast operates in di erent languages,
there is su cient overlap and thus the validity of the distinction is supported, at least in the case of Oceanic.
As our starting point we can take Elliott’s (2000: 67) characterization of the realis and the irrealis

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


categories: “the grammatical category of reality status can be described as the grammaticalized expression
of location in either the real or some unreal world”. Languages can depart in various ways from this
prototype. Elliott’s expression “reality status” will be used here as a convenient cover term, with the realis
p. 350 and the irrealis as subtypes.

One language with a realis–irrealis contrast is Manam. Manam has two sets of pre xes attached to verbs
that simultaneously index the subject and express either the realis or the irrealis. Every nite verb must
carry one of the pre xes; that is, expression of the reality status is obligatory. The irrealis is “basically used
to express envisioned, imagined events, i.e., events that will (will not) take place in the future in relation to
the time of the speech act or some other event, including commands, exhortations, and warnings, as well as
counterfactual events, i.e., envisioned events whose contraries in fact took, or are taking, place”
(Lichtenberk 1983: 183). However, as we will see, there is one type of exception to this generalization. The
realis is used for states of a airs that are (not) taking or have (not) taken place in relation to the time of
reference, including habitual states of a airs; but see (69) and (70) below and the associated discussion.
Example (66) illustrates the realis, and example (67) the irrealis:

(66)

As (67) shows, the irrealis occurs in the apodosis of conditional sentences and, depending on the type of
conditional, also in the protasis. If the speaker assumes that the condition is likely to be ful lled or if it is
already ful lled, the verb of the protasis is marked for realis:

(67)

If there is less certainty that the state of a airs of the protasis will obtain and in the case of
counterfactuality, the verb of the protasis is marked for irrealis:

(68)

Ross (1988: 361, 369) suggests that the rst-person singular irrealis marker n-/m- (and ŋ-) is historically
p. 351 related to an erstwhile future marker *na; see also the further discussion of *na later in this section.

The irrealis is also used to express habitual or at least multiply occurring sequences of events. (To express a
habitual, timeless occurrence of a single state of a airs, the verb is reduplicated and the realis aspect is
used; see (66).) The following passage describes the gardening cycle, and all the verbs are in the irrealis:

(69)

The sequence of events described in the next passage was performed several times in the (distant,
mythological) past:

(70)

The realis and the irrealis are used in the same ways described above in negative declarative and
interrogative sentences:

(71)

(Mása marks inde nite irrealis: it simply says that the given state of a airs will/will not obtain in the future;
it says nothing about how likely that state of a airs is. In its absence the implication is that the state of
p. 352 a airs is relatively likely to take place.)

The irrealis is used in positive imperatives and exhortations:

(72)

However, the irrealis is not used in prohibitives/negative imperatives. There it is the realis that must be
used. Negative imperatives use a special prohibitive marker, not the general negator tágo; cf. (71) above.

(73)

Elliott (2000) mentions some other languages, not Oceanic, where negative imperatives use the realis and
says that “a satisfactory explanation for this phenomenon has not been found” (Elliott 2000: 77). In a more
recent and more extensive treatment of the phenomenon, van der Auwera and Devos (2012: 180) ask the

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


following question: given that in comparison to (positive) imperatives, prohibitives can be seen as more
irrealis because they include the notions of imperative and negation, “does irrealis marking in the
imperative imply irrealis marking in the prohibitive?” Their investigation of 24 relevant languages (out of a
total sample of 179 languages) shows that the answer is negative. In fact, they argue that there is really no
reason why there has to be a link between realis/irrealis marking in imperatives and in prohibitives because
the two types of constructions may arise independently from di erent sources.

There is another possibility that ought to be considered. Positive imperatives are normally about future
states of a airs, and so the irrealis might be expected there. On the other hand, prohibitions are used not
only to avoid a future state of a airs but also to get the addressee to stop what he or she is already doing, or
what they sometimes or often do, that is to stop a real event, as in (74) from Manam:

(74)

It is conceivable that in languages that use the realis in prohibitions the construction was originally used,
p. 353 solely or mainly, to stop the addressee from doing what he or she was already doing. There the realis
might be the expected mood. Subsequently the use of the construction was broadened to include future
situations that were to be prevented from occurring. Unfortunately, the etymology of the dedicated
prohibitive marker in Manam (see (73) and (74)) is not known. It would be interesting to see if there is any
correlation between the use of the realis in prohibitions and the existence of prohibitive markers. This
possibility is, of course, consistent with van der Auwera and Devos’ point about the prohibitive in a given
language possibly having a diachrony unrelated to that of the imperative: the existence of a dedicated
prohibitive marker is evidence of a separate diachronic process.

According to Bugenhagen (1993: 16), in Sursurunga, another Oceanic language with a realis–irrealis
contrast, the irrealis contrast is used “[i]n positive commands, but not in prohibitions”. Unfortunately, no
5
further detail is provided about Sursurunga negative imperatives.

An obligatory realis–irrealis opposition is also found in Araki: “any verbal predicate is basically coded as
being Realis or Irrealis, in such a way that all sentences seem to be de ned by this two-fold opposition”
(François 2002: 105). As in Manam, in Araki too the elements that express the reality status also index the
subject. The Araki elements are clitics rather than a xes. What is interesting about the Araki system is that
the realis is formally marked, while the irrealis is unmarked. All the realis subject clitics contain the element
m, which, according to François, re ects an erstwhile realis morpheme; see mo and nam in (75):

(75)

The irrealis subject clitics are characterized by the absence of m; see na in (76) and compare nam in (75):

(76)

Araki is spoken in Vanuatu, and as will be discussed later in this section, there are a number of Vanuatu
languages in which the erstwhile realis morpheme m is implicated in variations in verb forms that have to
p. 354 do with reality status.

In Araki, “[i]rrealis refers to events which do not belong in the real world, but are only conceived as possible
facts in the imagination” (François 2002: 107), while “[a] Realis verb form indicates that the action referred
to is ‘real’, that is, already belongs to the world” (François 2002: 106). The irrealis is used in both positive
and negative imperatives. There is, however, a link between reality status and negation, having to do with
the scope of negation. Furthermore, François points out a link between negation and virtuality/non-
referentiality. Both in (77) and (78) realis/irrealis marking in the relative clause is sensitive to the polarity of
the main clause. In (77) the main clause is positive, the relative clause has a referential interpretation and it
is marked for realis:

(77)

In (78) the main clause is negative, the relative clause is in the scope of the negation, it has a non-
6
referential interpretation, and so is marked for irrealis:

(78)

Lolovoli, a fairly close relative of Araki, also has a realis–irrealis contrast. The realis is marked by the

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


particle mo, which follows the subject markers and cliticizes to those that are monosyllabic. Lolovoli mo is
historically linked to the m element found in the Araki realis subject forms. However, while Araki has no
overt marking for the irrealis, Lolovoli does. The marker is ni, and it too cliticizes to the monosyllabic
subject markers. As far as the functions of the two reality status markers are concerned, Lolovoli exhibits an
interesting feature. The realis is used not only for present and past states of a airs, but also for habitual
states of a airs, and even for future states of a airs, “if the event is in the near future, and the speaker
believes that it is certain to occur” (Hyslop 2001: 236). Future states of a airs can also be expressed by
means of the irrealis mood, “if the speaker expresses doubt about the occurrence of a future event … or if
p. 355 s/he is asking a question about whether or not an event will take place in the future” (Hyslop 2001: 241).
Either the realis or the irrealis can be used to express a future state of a airs when the speaker is certain it
7
will take place, but if he or she is not certain, only the irrealis can be used. The irrealis is also used for states
of a airs in the distant future. This is presumably because one is less certain about distant future than about
immediate or near future. Example (79) contains the realis because “the speaker is stating the plans of the
addressee, which she assumes will occur, and is seeking con rmation of these events from the addressee”
(Hyslop 2001: 239):

(79)

The events expressed in (80) are not certain, and so it is the irrealis that is used:

(80)

(In addition to the irrealis marker ni, the modality is also marked by the irrealis form of the subject marker
vi.)

In Lolovoli there is a correlation between negation and reality status. Unlike in positive clauses, in negative
clauses only the irrealis can be used to express future states of a airs:

(81)

p. 356 The realis cannot be used in a negative clause even though it would be used in the corresponding positive
clause. In such cases there is no marking of the reality status:

(82)

In Bierebo, another Vanuatu language, the realis is used for “present and past time reference, and habitual
events/situations, both a rmative and negative…. Irrealis is used to express future events/situations
(negative and a rmative), including future-projecting modal contexts such as imperatives, hortatives, and
past-time counterfactual events/situations (negative and a rmative)” (Budd 2010: 516; footnote omitted).

There are languages, such as Manam and Araki, where all nite verbs make a realis–irrealis distinction, but
there are also languages with only partial marking. For Bukawa, Eckermann (2007) distinguishes four
classes of verbs on the basis of how they mark reality status. As is often the case in Oceanic, in Bukawa the
elements in question both mark reality status and index the subject, although in this language there is
heterogeneity with respect to the marking of the reality status, depending on verb class. In class I the
realis–irrealis distinction is made only in the rst-person singular. In classes II and III the distinction is
made in all the singular numbers, but in di erent ways. That is, none of these three classes makes the
distinction in the plural. In class IV the distinction holds in all grammatical persons, singular and plural
(with a clusivity contrast in the plural). In this class the distinction is marked primarily by di erent forms of
the verb stems, but in a few grammatical persons there are also subject-reality status pre xes.

(83)

(84)

The rst-person singular realis pre x ga and the rst-person singular irrealis pre x wa are also used, as
sole indicators of reality status in the other classes (with ga replaced by ka in one class). As we will see, the
realis–irrealis contrast is marked by verb-stem alternations in some Vanuatu languages, but the
p. 357 alternations in Bukawa and in the Vanuatu languages are historically independent of each other.

Another language with only a partial realis–irrealis contrast is Tamambo, a Vanuatu language. There the

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


fully- edged contrast is found only in the third-person singular. The realis is marked by mo, which in the
absence of other speci cation also signals a third-person singular subject. This mo is historically related to
the Araki realis marker m (recently discussed above). Mo can also be used in combination with the plural
subject markers, “but has slipped from use as a realis marker after other [i.e., non-third singular; F. L.]
subject pronouns with the majority of speakers. Occasionally it is used as a marker of realis by older
speakers (age 50+)” (Jauncey 2011: 297). The irrealis is marked by a, which also signals that the subject is
third-person singular. In persons other than third-person singular, there is no (obligatory) realis–irrealis
contrast. (The plural forms occasionally receive realis marking by older speakers.)

Example (85) shows the realis–irrealis contrast with third-person singular subjects:

(85)

And (86) shows absence of the contrast with rst-person singular subjects:

(86)

The irrealis marker is also used for past habitual events, but here too only with third-person singular
subjects. (Elsewhere, habituality may be inferred from the context, and for timeless habituals a tense–
aspect marker may be used.) In (87) the third-person singular irrealis marker a is used twice to mark
habitualness in the past (HAB ) and purpose/future in the past (IRR ):

(87)

In the history of the languages of Vanuatu there was a realis marker ma. This marker survives in some of the
p. 358 languages; see the examples from Araki, Lolovoli, and Tamambo given previously. In some others it has
disappeared as such but has given rise to mutations in initial consonants in verb roots which signal
di erences in reality status (see Lynch 1975 and Crowley 1991 for discussion). Examples (88) and (89)
illustrate the mutation in Namakir. The verb “walk” has an initial t in the irrealis (imperative) form, and an
initial “mutated” d in the realis form:

(88)

(89)

Verb-initial consonant mutation, apparently distinguishing realis and irrealis, has also arisen,
independently of that found in the Vanuatu languages, in some languages spoken in Papua New Guinea.
There, however, it is the irrealis form that exhibits the mutation. Ross (1988) argues that the mutated
consonants result from sequences of *n, from the future marker *na, and verb-stem initial consonants, as
in Mangga Buang:

(90)

This development is independent from, and di erent from, that in which the future marker *na and the
following subject-indexing clitic fused, as in Manam (see (68) and the immediately following paragraph)
and its close relatives.

As seen from the previous discussion, there are some di erences in how the realis–irrealis contrast
operates in various Oceanic languages. In some languages, but not in others, the irrealis is used to express
habitual and/or multiply occurring events or sequences of events, regardless of their temporal status,
although there are speci c conditions in some of the languages. On the other hand, the realis can be used to
express future states of a airs in at least one language. In some languages, but not in others, there are links
with negation. Nevertheless, when it comes to Oceanic, realis and irrealis do appear to be useful concepts.
The former has to do primarily with states of a airs in “the real [world]”, and the latter with states of
p. 359 a airs in “some unreal world” (Elliott 2000: 67), but for each language it is necessary to specify more
closely the ranges of use of the two reality status categories. This does not seem signi cantly di erent from
taking, for example, the perfect as a cross-linguistically valid concept, even though it is necessary to specify
its semantic range for individual languages.

14.8 Conclusion

Several kinds of modality have been discussed here: epistemic, deontic, dynamic, desiderative, and timitive;

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


as well as the mood categories of realis and irrealis. The main grammatical or semi-grammatical means of
expression include auxiliaries, not-fully edged verbs, particles, and a xes.

In some cases the grammatical elements are related, historically and/or synchronically, to lexical elements.
Thus the desiderative elements of Boumaa Fijian, Pukapukan, and Māori (as well as Standard Fijian and
other Polynesian languages) continue an earlier verb “want, desire, wish or long for”. And in Toqabaqita,
the desiderative element is a particle/auxiliary-like element synchronically related to the verb “want”. In
several languages expressions of dynamic modality involve verbs with meanings such as “be a match for”,
“be appropriate, be adequate”, and “be enough, be su cient”. Commonly, the same elements can also
express possibility and in some of those languages also permission. In the possibility and permission
functions those verbs take impersonal subjects.

While ability–permission(–possibility) polysemy is not unusual in Oceanic, epistemic–deontic polysemy


does appear to be rare. One also nds desiderative–necessity polysemy.

Timitive modality, which simultaneously expresses epistemic possibility and apprehension, fear, is
common in Oceanic. In Fijian (Boumaa and Standard) the apprehension/fear meaning may be absent, in
which case the elements function just as markers of epistemic possibility. However, this function is less
common than the dual function.

A brief survey of the marking of reality status has revealed some of the variation in the uses of the realis and
the irrealis. Nevertheless, in spite of the variation, realis and irrealis are valid categories for the languages
discussed and, presumably, for other Oceanic languages. Instead of abandoning the realis–irrealis contrast,
what is needed are detailed speci cations of how it operates in individual languages.
Abbreviations

1,2 ,3
rst, second, third person

ACC

accusative

AD

adnominal (su x)

AG

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


agent

AND

andative

APPRE

apprehensive
p. 360

ART

article

ASP

aspect

ASRT

assertive

AUX

auxiliary

BF

bu er

CAUS

causative

CAV

caveat

CLF

classi er

COM

comitative

COMP

complementizer

CON

coordinating conjunction

DEIC

deictic

DEM

demonstrative

DETR

detransitivizer
DU

dual

ERG

ergative

EXCL

exclusive

FUT

future

GENP

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


general preposition

GENR

general tense–aspect–mood particle

GIV

given information

HAB

habitual

IN

inessive

INCL

inclusive

INIRR

inde nite irrealis

IRR

irrealis

LIM

limiter

LOC

locative

LOCPRO

locative proform

NEG

negative

NEGV

negative verb

NF

non-asserted factuality

NFUT

non-future

NMLZ

nominalizer

NOM

nominative
p. 361
NR

non-realis modal adverb

NSG

non-singular

NUM

numeral marker

OBJ

object

OBLIG

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


obligative

OPT

optative

PERS

personal (su x)

PFV

perfective

PL

plural

POSS

possessive

PREP

preposition

PRF

perfect

PRO

anaphoric proform

PROFORE

pronominal foregrounder

PROH

prohibitive

PROP

proprietive

PURP

purpose

RDP

reduplication

REAL

realis

REF

referent

RESPRO

resumptive proform

SBJ
subject

SEQ

sequential

SG

singular

SIM

similative

SPEC

speci c

Downloaded from https://academic.oup.com/edited-volume/40412/chapter/347345689 by Australian National University user on 16 March 2023


SUB

subordinating conjunction

TA

tense–aspect marker

TAM

tense–aspect–mood marker

TIM

timitive

TNS

tense

TR

transitive

VENT

ventive

Notes
1 I am grateful to the editors of the volume for their valuable comments on an earlier version of this chapter.
2 The glossing conventions employed here are those of the sources, except that in a few cases the original glosses have
been modified to conform to the Leipzig Glossing Rules and/or for the sake of uniformity. In a few cases glosses have been
added.
3 This modality type is also found in the three varieties of Melanesian Pidgin—Tok Pisin (Papua New Guinea), Solomons Pijin
(Solomon Islands), and Bislama (Vanuatu).
4 Cf. von Wrightʼs (1951: 3) concept of mixed modalities, that is, combinations of modalities.
5 Bugenhagenʼs study looks at the realis and the irrealis in several Oceanic languages in Papua New Guinea (including
Manam), with a special focus on the irrealis.
6 However, cf. Mithun (1995) for the use of the irrealis in Caddo (a Caddoan language spoken in Oklahoma). Mithun argues
that in negative sentences the irrealis is used because it has negation in its scope.
7 Similarly in Central Pomo (a Pomoan language spoken in northern California), future states of a airs can be expressed
either by the irrealis or by the realis forms. As Mithun (1995: 378) says, “Futures can vary in their probability of occurrence.
Speakers could exploit the Irrealis/Realis distinction to mark their expectation of occurrence”. And Chafe (1995: 358)
suggests an expectation that “predicted future events will take place” as a possible explanation for the use of the realis to
express future states of a airs.

You might also like