Beruflich Dokumente
Kultur Dokumente
Abstract
Starting from Bybee, Perkins, & Pagliuca (1994) and complementing their
insights with observations that often originale in the non-English literature
on modality, this paper attempts to supply the grammaticalized expressions
ofmodaly with a semantic map. The term "modality" is taken to refer t o
just those (four) domains in which possibility contrasts with necessity, and
"semantic map' refers to a representation of cross-linguistically relevant
synchronic and diachronic connections between modal, premodal, and post-
modal meanings or uses. Special attention is given to meanings that are
vague between possibility and necessity, to developments from possibility to
necessity and vice versa, to postmodal meanings that can originale in either
possibility or necessity, and to demodalization of non-epistemic modality.
1. Introduction
Semantic maps are powerful tools in the analysis of cross-linguistic Varia-
tion. The point of this paper is to sketch some features of the semantic
map of modality, starting from the paths ofFered by Bybee, Perkins, &
Pagliuca (1994), and to use this to make predictions about the ways
languages express modality. In Section 2 we explain our use of the terms
"modality" and "semantic map" and we describe the data that were the
input to our proposal. Section 3 is about semantics and about semantic
change. First we provide a somewhat adjusted representation of the three
modal paths found in Bybee et al. (1994). Then we propose to connect
the three paths, thus approaching the semblance of a true map, and we
turn our attention to "demodalization",
>
i.e., the tracks that lead out of
modality. Section 4 exploits the map for the explanation of certain formal
features of the expression of modality.
2. Terminology andmethodology
2.1. Modality
2.1.1. Introduction
Modality and its types can be defined and named in various ways. There
is no one correct way. The only requirement is that one makes clear how
one uses one's terms. Our definitions and terms are fairly traditional. We
will first present our terminology, explain, and illustrate it. We then
compare our terms with some other terms and other uses of the same
terms, in particular with the terms and uses found in Bybee et al. (1994).
We also relate our relatively restricted over-all notion of modality to the
notions of volition and evidentiality.
With (4a) the Speaker presents John's arrival s uncertain. This concerns
possibility, for relative to some other judgments (e.g., the belief that John
is very unreliable) John's arrival is judged possible. Thus uncertainty is
epistemic possibility. With (4b) the Speaker represents John's arrival s
relatively certain, relatively probable. This concerns necessity, for relative
to some judgments (e.g., the belief that when John comes by bike, he
chains bis bike to the tree and the belief that the bike is in fact chained
to the tree right now) John's arrival is necessary. Certainty and a relatively
high degree of probability thus amount to epistemic necessity.
Within the set of participant-internal, participant-external, and episte-
mic modalities, theoretically three subgroupings are possible: (i) partici-
pant-internal and participant-external modality together make up
non-epistemic modality; (ii) participant-external and epistemic modality
together make up non-participant-internal modality; and (iii) participant-
internal and epistemic modality together make up non-participant-exter-
nal modality. Only the first subgrouping makes sense, the reason being
refinements, the deontic modality may or may not involve the speaker's
subjectivity, a dimension called "subjectification" or "grounding" (see,
e.g., Goossens 1997; Mortelmans 1997).
reason that the proposition may involve more than one participant
e.g., you and bus 66 in the examples in (3). On the basis of English alone,
"subject-internar would seem appropriate, but this notion does not
generalize easily, witness the dative experiencer in Kannada (7), which,
s Bhat (1991: 160-169) argues, is not a subject.
For the purpose of this paper, though, "participant" will have to do.
A widely used term is "root modality". Bybee et al. (1994: 178) use
"root pssibility" in much the sense of our "participant-external possi-
bility", but they have no use for "root necessity". An example of a
linguist who has both root pssibility and root necessity in much the
sense of our participant-external pssibility and necessity is Gamon
(1993). Most scholars, however, equate the notion "root" with "non-
epistemic" (e.g., Coates 1983; Palmer 1986: 103-104; De Haan 1997: 7).
This was also the original usage in Hofmann (1966 [1976: 93]), but
Hofmann does not give any motivation for the term, nor do, s far we
know, any later users,3 the usually tacit assumption being that the root
sense is more basic, whether diachronically (which is correct) or synchron-
ically (which is disputed) or both.
The view that the basic divide in modality is that between epistemic
and non-epistemic is fairly Standard. Not so the view that deontic mod-
ality is a special case of participant-external modality, but it agrees at
least partially with Bybee et al. (1994:193), and earlier also with Goossens
(1979: 88-89) and Bybee (1988: 256), who all make this point with
respect to deontic pssibility ("permission") and participant-external
("root") pssibility.4
(10) Ahmet
Ahmet come-EVi
'Ahmet must have come.' (Aksu- & Slobin 1986: 159)
The claim is not that the inferential reading is obligatory; in fact (10)
also has a quotative reading s well s a mirative one (expressing surprise;
see DeLancey 1997). The claim is only that the inferential reading
amounts to epistemic modality9 and more particularly epistemic necessity:
modality encompasses
volition evidentiality
Necessity
Deontic Epistemic necessity
necessity Quotative
Inferential evidentiality evidentiality
Evidentiality
should have a map of its own was a wish explicitly phrased by Bybee
(1985: 195-196). In fact, a year later, a map was given by Anderson in
his study on evidentiality (1986), but it seems fair to say that the map
occurs there only s an afterthought to his proposals on evidentiality,
without arguments and illustrations. Modality notions also appear on
some of the other maps, in particular, those offered for conditionality
(Traugott 1985: 299), and for the notion of getting (Matisoff 1991: 427).
Furthermore, Bybee et al. (1994) supply three paths, and paths or rather
the representations of them are maps, too, be it "mini-maps".10 The time
has come, we propose, to put these and other leads together and to
redraw the modality map. The strategy that we will follow is to Start
from the paths or "mini-maps" of Bybee et al. (1994) and to connect
and extend them.
In taking the Bybee paths to be our starting point we will assume that
they are essentially correct, even though there would be occasions to note
that a feature of a path is insufficiently argued for or insufficiently
illustrated. The Bybee paths are meant to portray the distribution and
development of those modal markers (i) that are verbal and are consid-
ered at least partially grammaticalized (to auxiliary, derivational, or
inflectional Status) or (ii) that may be assumed to develop into or out of
the Status described under (i). We adhere to this restriction too. Hence
we do not study modality markers that are nouns (necessity, Obligation),
adjectives (possible, inevitable), adverbs (perhaps, possibly), or mental
state predicates (think, be convinced). Clearly, for a fll study of modality,
these expression types will have to be included (see, e.g., Bader, Werlen,
& Wymann 1994; Nuyts 1994). We submit, however, that the basic
outline of the semantic map of modality that we will establish on the
basis of arguably grammaticalized verbal markers will be valid for the
other types of markers, too. This hypothesis is based on another hypoth-
esis, the one that associates grammaticalization with meaning change and
polyfunctionality, and it is these diachronic and synchronic links that
constitute the paths of the map.
An example of a Bybee et al. (1994) path is shown in Figure 1. It
represents a part of the path for possibility. Like some of the non-modal
semantic maps mentioned above, the mini-map of Figure l has an explicit
participan -exteraal
participant-intern epistemic possibility
possibility
(deontic possibility)
Figure 1. Possibilities
ability, like the modern French savoir illustrated in (5); it later extended
it to any ability, see Figure 2.
When in a figure a meaning develops in two directions, nothing is
implied s to the temporal order. In Figure l, for example, participant-
external possibility develops in two directions, those of deontic possibility
and epistemic possibility. We do not say anything about the temporal
priority of deontic vis-a-vis epistemic possibility. In the literature one
comes across the claim that deontic meanings precede epistemic ones
see the very title of Shepherd's (1982) "From deontic to epistemic: An
analysis of modals in the history of English, creoles, and language acquisi-
tion" (cf. also Bybee 1985: 168). However, when one comes to the details,
this claim becomes controversial. An instructive example concerns two
of the best-studied modal verbs, English may and can. If, s shown in
Table 4, even the may and can facts are possibly not that clear, one may
fear that the facts in languages that have attracted less attention are even
less clear.
The representation of a path is a map, and since the path shows
meanings, the map is semantic. The semantic map in Figure l does not
refer to any language-specific element or its semantics yet. The latter
Information could of course be entered. In Figure 3 we do this with
Table 4. On the diachrony of deontic and epistemic uses for may and can
may can
Figure 3. Some language-specific possibility markers: English may f and Dutch mgen
diagonal linesfor English may and its Dutch cognate mgen, respec-
tively. The meanings are exemplified in (11) and.(,12).
In this way semantic maps can teil us something about the semantics of
any relevant lexical element at any stage of a language.
Note that the format of Figure 3 does not say anything about the
frequency or markedness of any one meaning. It does not say, for
instance, that English may is predominantly epistemic, that the other
meanings are on the decline and that the participant-internal meaning is
archaic (see Coates 1983; Jacobsson 1994). Note also that Figure 3 is
not intended s a complete description of either English may or Dutch
mgen. Both have other uses, but they are out of the mini-map of
Figure 3.
2.3. Database
The data that we take recourse to have a double origin. First, we make
use of the well-sampled data collection oifered by Bybee et al. (1994)
themselves, thereby assuring the same degree of representativeness that
they aspire to. Second, to finetune the hypotheses we essentially use all
the data we had access to, and this particularly involves Germanic lan-
guages other than English.
Wicipant-intemal condition
possibility
participanl-external
possipility
complementaon
/i
bestrong, 'know'j
l 'arrive at', finishy participant-interaal
v :suffice'_ J possibility
participan :-external
'be permitted', condition
'dare' .
deontic possibility)
[ 'be', 'become', l epistemic possibility
'happen', 'stand',1
don't know',
'like'
complementaon
(1986), which also contains Information on the lexical sources, the central
domain will be enclosed within a square. In this way we can clearly
separate the central modal domain from the premodal and postmodal
domains.
As the focus of this paper is on modality itself, there will be no attempt
to make the description of the premodal and postmodal domains exhaus-
tive. The sense of incompleteness is bound to be strongest for the
premodal domain, the assumption being that the development from
premodal to modal to postmodal can be a grammaticalization chain,
with variegated lexical input and restricted grammaticalized Output. To
represent sets of premodal lexical elements that have been attested s
going in the same direction we will use dashed lines. Thus enclosing cbe
strong', 'know', 'arrive at', 'finish', and 'suffice' with dashed lines is not
meant to refer to any one marker that can have each of these meanings.
The point is rather to refer to all markers that have at least one of the
enclosed meanings. The description of the source expressions is further-
more incomplete in that it gives no Information about the grammatical
shape of the lexical predicate that co-occurs with the future modal marker.
As Bybee & Pagliuca (1985: 73) and Bybee et al. (1994: 284) show, often
the grammatical shape of the lexical predicate contributes a meaning too.
Their example concerns the fact that Obligation markers may develop
from a combination of a possession verb ('have') and a main verb in the
infinitive, s in English (2b). It is not really 'have' alone which is the
source, but also the prospective or incompleteness sense of the infinitive.
Figure 5 lists various sources for participant-internal possibility, such
s 'be strong' and 'know', obvious from Standard Average European
(e.g., may < 'be strong' and can < 'know'), which first led to inherent or
learnt participant-internal modality, hyponyms of participant-internal
possibility, which they later extended to (see Figure 2). We also list 'arrive
at' and fifinish' from Bybee et al. (1994: 190-191). Bybee et al. (1994:
188) also mention 'enough' (in Tok Pisin), possibly better glossed s
'suffice, be enough' (cf. Heine et al. 1993: 75 on the Nilotic language
Acholi) and 'get', which we will single out for special attention later
(Section 3.3). They also point out that deontic and epistemic possibility
can develop directly out of the lexicon without any intermediate stages.
c
Be', 'become', 'happen', 'befall', 'stand', don't know' are sources which
Bybee et al. (194: 206) list for epistemic possibility, and from Heine et al.
(1993: 139) one could add 'like', reported for Lango (Nilotic). For
deontic possibility Bybee et al. (1994: 191) list cbe permitted' s a source.
A semantically more interesting source, documented from our materials,
is cdare', s in Serbian/Croatian (13), but also in Czech and Sorbian.
Note that English he may be a genius or Dutch hij kan dan een genie zijn,
with the modal kunnen ccan' instead of mgen cmay', have concessive
interpretations too, but in the case of English may and Dutch kunnen
one can derive this Interpretation from their epistemic meanings. Dutch
mgen, however, does not have any epistemic meaning (any more).
->(epistemic possibility"
epistemic necessity
complementation
participant-external necessity
imperative
imperative are more complex, but this need not concern us here. It is not
clear whether Bybee et al.'s "Obligation" Stands for deontic necessity or
for its hyperonym participant-external necessity. Since the Obligation is
exemplified with modern English should and since the latter is probably
best described s expressing the more general participant-external neces-
sity, even though deontic necessity is both the original meaning s well
s the one that is still prominent in the modern language, we take
'Obligation" in the sense of "participant-external necessity".
Figure 10 differs from Figure 9 in that it includes deontic necessity s
well s a change from deontic necessity to participant-external necessity.
This change seems plausible not only for English should but also for
some of its cognates in other Germanic languages (see Ramat 1972:
196-200 on (older) Germanic; Coates 1983: 59 on Modern English;
Stevens 1995: 191-193 on the development in German). Interestingly,
the direction of this change is from hyponym to hyperonym, which is
the reverse of what we have seen with possibility.
Figures 9 and 10 say nothing about the source expressions yet, but
Bybee et al. (1994: 183, 196) list Owe', 'need', 'be good/proper/...', 'be',
'become', 'have', 'fall', 'want', 'see', and 'understand'. It is not explained,
however, which modal meaning these feed into. About some of these
sources we are ready to make some Claims. Thus it seems to us that
'need' feeds into a participant-internal necessity, lacking in Figures 9 and
10, but it is clearly needed for meanings such s (Ib). It furthermore
seems rather plausible to think that notions such s owing and duty (for
English should and Russian dolzen (Bulygina & Smelev 1991)), belonging
(for Dutch hren), and being good or proper skip the participant-internal
dimension and go directly to the deontic subtype of participant-external
necessity. The necessity that uses 'have', at least in English have to, is
also participant-external, but there is no reference to any authority or
moral code, so the participant-external necessity is general. Bybee et al.
(1994: 206) also mention one case of a direct source to epistemic necessity,
viz. a Tigre (Ethiopic) combination of 'become' and 'if, which we
suppose yields something that can be glossed s 'if it becomes'. The
epistemic necessity
^deontic necessitj^)
participai t-extemal
nece ssity
In Bybee et al.'s map (1994: 240) there are other paths going on from
the future and future is preceded by intention, but we will not represent
these Claims.
The second step towards the desired integrated map of modality con-
sists of unifying the possibility and necessity maps, in particular, Figures
8 and 13. There are several reasons for doing so. One reason is that there
are postmodal meanings that are shared by the possibility and necessity
tracks. Bybee et al. (1994) make this claim for concession, complementa-
desire, movement
toward
' 4 if itbecomesv ^(epistemic necessity
perfect )
complementation
( Owe', (Jcontic necessity)
. 'belong',
vbe good/proper parucipai t-external
necc ssity (Imperative)
f 'have', .
l'be supposed' \
participant-interna
necessity
tion, and future. We can add condition s wel}.13 Bybee et al. (1994)
mention condition s a postmodal destination for possibility, a case
illustrated with Dutch mocht in (15). The same sentence, with the same
conditionality meaning, would in regional (Southern/Belgian) Dutch
employ the necessity modal moest, s in (21).
Other cases may be the Hungarian suffix -hat/-het (Kiefer 1981: 152-153)
or the verb dr in the Melanesian language Tinrin (Osumi 1995: 71).
alone', and finally 'he may not stay home alone', at which stage drfen
developed a positive deontic possibility sense (25b).
parucipant-extem
possibility
(deontic possibility)
(epistemic possibility condion
'participant-external
participant-intemal possibility possibility
^deontic possibiHtv)
(^deonticnecessity)
participant-internal necessity participant-external
necessity
And English also has relic uses of the German-style modal passives (see
Demske-Neumann 1994), which, we suppose, are also vague between
possibility and necessity.
/
So the English modal Infinitive may after all be best considered vague
toobut see Goossens (1992) for a path hypothesis.
Note that the vagueness may not be a reflection of a change from
possibility into necessity or vice versa, for this would take us back to
what we have seen with, e.g., Danish ma (22) or German drfen (25). It
may of course be difficult to distinguish between these two types of
vagueness. Consider the Latin "gerundive", more particularly, the predi-
catively used gerundive.14 In classical Latin, the predicative gerundive
typically expresses necessity, but possibility uses are also found.
The possibility meanings seem restricted to some verbs, to the extent that
one may even call them lexicalized (cf. Risch 1984: 54). Sometimes one
finds the possibility sense described in terms of the weakening of the
necessity sense (Ernout & Thomas 1972: 287; Risch 1984: 54, 61-62),
and sometimes in terms of the preservation of a more original possibility
sense, prior to its necessity sense (Michel 1978: 236). Either way, at least
in classical Latin the vagueness would seem to be of the type indicative
of semantic change and not of the type found with the German modal
infinitive. On the other hand, it is not to be excluded that in older Latin
the predicative gerundive was in fact truly vague between possibility and
necessity and that necessity, which was to dominate later, developed
through specialization.
The modal infinitives have their origin in a use of the 'be' verb. A
participant-external modality vague between possibility and necessity is
also documented for a construction that employs 'have', s in regional
and colloquial Basque (Alan King, personal communication):
The lexical verb fa 'get' still exists (37), and an intermediate meaning is
what we want to callpace Wagner's (1976) "ingressive""participant-
external actuality" (38), meaning that some state of affairs is actualized
because of participant-external circumstances, in particular, because the
participant in some way "received" the state of affairs.
3.4. Demodalization
This section is devoted to demodalization. Of course, we have already
seen various manifestations of it. We have seen how modal meanings
may yield postmodal ones, independently of whether the modal meaning
was possibility or necessity. This was the case for condition, concession,
and complementation, s coming from epistemic modality, and for future,
s coming from participant-external modality. There was one postmodal
meaning which was exclusive for necessity, more particularly participant-
'participant-exterm
possibility
Xoeontic possibility)
participant-extemal actuality
Qteontic necessit
fhavel participant-external
necessity
external necessity and that was imperative. In this section we will discuss
some other types.
The s- prefix furthermore is not the only perfectivizer. Attested are also
pere- Over-' in (colloquial) pere-mog Overcame (e.g., an illness)', pre-
voz- Over-up-' in pre-voz-mog Overcame', za-ne- 'INCH-NEG-' in za-ne-
mog cgot , and iz-ne- cout-NEG-' in iz-ne-mog 'became exhausted'.
Interestingly, Russian also has a modal for learned capacity, umet' 4know
to', like French savoir in (5). Umet' was originally imperfective and
without perfective counterpart, too, and it also underwent the perfectiviz-
ing development, with this twist that perfective s-umet' lacks the learned-
ness component (Satunovskij 1996: 215-216).
The fourth type of demodalization is Latin posse, s described in Lewis
& Short (1975: 1404, 1409). Posse ccan, may' is a verb which is a
univerbation of the adjective potis cable' and the verb esse cbe'. The
phrase potis esse was used in pre-classical times and seems to have had
only modal meanings. The later verb posse, however, also has the non-
modal meanings cto have influence, to avail', s in (44).
The fifth type, hinted at by Heine et al. (1993: 9), is totally different.
The marker sheds both the possibility and the participant-internal compo-
nent and becomes a consecutive marker. The case has been argued for
the Shona (Bantu) verb gona- C to 7 be able to' developing to -go-, a
(51) Auf den Planeten, die [...], ist die Art des Bodens eine
on the planets that is the manner of.the soil a
ganz andere: [...] und es gibt keine [...] Krater.
completely different and it gives no craters
Auf der Venus sollte es demnach keine Krater [...] geben.
on the Venus should it therefore no craters [...] give
On the planets that [...] the soil is constituted quite differently:
[...] and there are no craters. On Venus there should therefore be
no craters [...].'
s'
Dutch is interesting in this respect too. Outcti7moeten has been claimed
to allow a quotative meaning by De Haan (1997: 146, 153-154).
We find the latter proposal intuitively plausible and we are indeed inclined
to accept that moeten is different from English must in this respect,
although one can use the English It must be a good film in a context
where it is understood that the evidence relative to which the film is
certain to be good is hearsay evidence. In fact if Dutch moeten should
be attributed a separate quotative meaning, it seems highly plausible to
Interpret it s a conventionalization of the contextual reading just posited
for English. In that case we wold have a quotative that does not come
from participant-external modality but rather from epistemic modality.
interrogative
epistemic necessity
participant-internal
necessity
quotauve
participant-internal
necessity
premodal P articipant-intcmal
mcanings possibility
going to
possibility /postmodal meanings^
V coming from
participari ' possibility
prcmodal possi ility
mcanings epistemic possibility
going to whal (deontic possibility) postmodal meanings }
is vaguc .- coming from either .
bctwccn l possibility or necessity l
possibility
and ncccssity/ (jeontic necessit) epistemic necessity
participant-external /postmodal meanings^
prcmodal neceksity
mcanings coming from
going neccssity
to nccessity participant-intcmaP^/
(onc is future) . nccessity
respect to diachrony, the map shows that most if not all of the modal
notions are directly accessible from a premodal notion. In the horizontal
dimension there is fll unidirectionality, going from premodal notions to
participant-internal modality, participant-external modality (either the
general notion or the deontic subtype), and epistemic modality, with
demodalizations at (nearly) every modal stage and limited remodalization.
The latter scenario involves an element going from a modal meaning to
a postmodal one and taking this s the premodal source of a new modal
meaning. Meanings attested in both the premodal and the postmodal
sphere are future, actuality, and 'like'. At least in the case of the future
a remodalization cycle is attested, with the future s a postmodal develop-
ment out of participant-external necessity and subsequently serving s
premodal input to epistemic necessity. Methodologically, the demodaliza-
tion and cyclicity facts should remind us of the limits of our ability to
reconstruct the past from the present, a counsel also voiced by Anderson
(1986: 282).
In the vertical dimension, however, there is no unidirectionality, for
within the deontic realm markers have been found to switch from possi-
bility to necessity s well s from necessity to possibility.
Meaning l
>-(Meaning 6
-^Meaning 4
Meaning l -^(Meaning 2
->^Meaning 4,
> (Meaning 2
-^("Meaning 6
participant-mtern
possibility
participan -external
possi
epistemic possibility
In the very same paper, however, he mentions that the English modals
do not correlate the semantic distinction with any formal features. So,
to make (57) cross-linguistically more appropriate, we propose to change
'<' into '<', meaning 'exhibits an equal or lower degree of formal
grammaticalization':
English and German are not the only languages that can be mentioned
in support of (58). Especially the divide between participant-external and
epistemic modality seems to provide a good parameter. Thus, the Greek
possibility modal boro has fully inflected forms for participant-external
modality but an invariable form for epistemic modality (Palmer 1986:
19-20). Similarly, in French pouvoir 'can' allows inflected forms for both
participant-external and epistemic modality, but epistemic modality can
be expressed with an invariable, impersonal // se peut que 'it REFL may
that'.
Note that under the view in (58), semantic change is an enabling
condition for formal grammaticalization. This is in agrecment with Givon
(1975: 86), Heine, Claudi, & Hnnemeyer (1991a: 175), Heine (1993:
48, 58), and Haspelmath (1997b), among others, but not with Lehmann
(1982 [1995]), Croft (1990), of Bybee et al. (1994: 6-7, 19-21), who all
claim that the two processes go hand in hand. Probably both scenarios
are possible, but in the domain of modality semantic change seems to
come first.
This seems eminently plausible for the premodals, too. Premodals are
either normal lexical items or they are elements already grammaticalized,
cline is one of subcategoriality, but one may wonder whether the differ-
ence between category and subcategory is all that essential. For German,
in any case, it has been claimed that a transfer to the class of preterite-
presents constitutes grammaticalizationsee Stevens (1995: 184-185),
who then also considers the development of the 'like' sense of mgen s
degrammaticalization. Also, there is the point that Hopper & Traugott
(1993: 108-114) make about intermediate stages between lexical verbs
and auxiliaries,19 illustrated by Indo-Aryan "vector verbs", which com-
bine with lexical verbs in a compound-like fashion but still retain their
lexical meaning. If vector verbs can be considered a verbal subcategory,
then we have a case of a formal grammaticalization chain from major to
minor verbal subcategory to auxiliary. This is like the scenario for most
of the Germanic preterite-presents (see, e.g., Plank 1984 on English)
except that a few Germanic preterite-presents made a U-turn at the minor
verbal subcategory s tage.
Finally, with Ramat (1992) we believe that degrammaticalization really
does exist. Powerfully simple examples are the English nouns if and ism.
And to go by the example of the English noun must, which still has a
modal sense, the degrammaticalization it undergoes in its conversion
from auxiliary to noun does not even presuppose postmodality.
Notes
The initial versions of this paper were written during the time Plungian was a guest
scholar at the University of Antwerp (January-July 1997). Thanks are due to the
Research Council of the University of Antwerp for fmancing this stay. Thanks are
also due to the Science Foundation Flanders (Krediet aan Navorsers, Onderzoeks-
gemeenschap 'Cognitieve Taalkunde). We are furthermore grateful to D. N. S. Bhat,
Ferdinand de Haan, Patrick Dendale, Louis Goossens, Martin Haspelmath, Tanja
Mortelmans, and Jan Nuyts for comments and conversations. The paper is dedicated
to the pathfinder of note 10.
Correspondence addresses: (van der Auwera) Departement Germaanse Taal- en
Letterkunde, * Universiteit Antwerpen (UIA), Universiteitsplein, l, B-2610 Wilrijk,
Belgium; e-mail: auwera@uia.ua.ac.be; (Plungian) Institute of Linguistics, Rossijskaja
Akademija Nauk, B. Kislovsky per. 1/12, Moskva 103009, Russia; e-mail:
plungian@iling.msk.su
1. We thereby decline to adopt Coates' notion of an "objective epistemic" use, s in (i).
(i) Certainly if there is endeavour to x, there must be attention to x'. (Coates 1983: 42).
For reasons of logic, the speaker's certainty s expressed with must happens to be
absolute, but it remains the certainty of the Speaker. One can of course accept the use
in (i) s a subtype, but then we would single it out with a cline of degrees of certainty
rather than with degrees of Speaker involvement. The difference may seem terminologi-
cal only, but it is not. Our choice allows us to avoid having to admit a use of English
supposed to s a counterexample to a claim about the ynidirectionality of any change
involving epistemic modality (see Section 3.1).
2. The notions imperative, prohibitive, and Optative furthermore appeal to volition, which
falls outside of modality, in the sense of modality we embrace (see below).
3. In Coates (1983), for instance, a very influential study of English modality, the term
"root modality" is central, but there is again no motivation or even reference to earlier
uses. Or take Bybee & Pagliuca (1985: 77), criticizing Coates for not being faithful to
the original usage, but not actually referring to any source either. A semi-explicit
attempt to motivate the term is undertaken by Sweetser (1990: 152) when she admits
of "leaning towards an analysis of epistemic modal meaning s rooted in sociophysical
(root) modality".
4. The reason why Bybee et al. (1994) do not make this claim with respect to necessity
hangs together with the fact that they do not employ any notion of participant-external
or "root" necessity.
5. In this respect it may also be mentioned that Bybee et al. (1994: 320) employ a notion
of modality that is very wide, s it encompasses also the notion of attempt Otry') (see
also Givon 1990: 533) and what they call "andative" ('be going t do something') and
"venitive" (cbe coming to do something').
6. Chung & Timberlake (1985) do not actually use the term "modality" but "mood" and
furthermore include all nuances of illocutionary type.
7. Note that of the linguists who include volition and exclude evidentiality, two, viz.
Longacre (1976) and Mel'cuk (1994), do not even accept any notion of epistemic
modality.
8. Abbreviations: l, 2, 3 first, second, third person, ACC accusative, ART article, COMP
complementizer, CONS consecutive, DAT dative, EVI evidential, F feminine, FUT future,
GEN genitive, GRD gerundive, INCH inchoative, INFInfinitive, M masculine, NEG negative,
NOM nominative, PL plural, POSS possessive, PF perfect, PFV perfective, PRS present, PST
past, REFL reflexive, SG singular, VNverbal noun.
9. A terminological variant of our decision to include inferential evidentiality under
epistemic modality is to include epistemic modality under inferential evidentiality (see
Dendale 1994 and Tasmowski & Dendale 1994). Compare also Nuyts (1996), who
accepts evidentiality s modality, but the only type of evidentiality that he considers is
inferential evidentiality.
10. Supplying modality with "paths" with both synchronic and diachronic relevance is not
new. At least for English, Goossens has been doing that since 1979 (e.g., Goossens
1979,1983,1985a, 1985b, 1987).
11. In allowing metaphor or metonymy s well s specialization and generalization, seman-
tic maps come close to semantic networks in the sense of Cognitive Grammar
(Langacker 1988: 50-52).
12. In Germanic languages, 'dare' verbs developed into modality too, with the extra com-
plication that there is a formal closeness to verbs meaning 'need'e.g., Gothic gadaur-
san 'dare' andpaurban 'need'. The developments differ from one Germanic language to
the next, and according to Birkmann (1987: 371) there is no satisfying explanation yet.
SeealsoDufiley(1994).
13. Yet another postmodal meaning reachable via necessity and possibility, but whose
placement and connections oix the map are not clear to us, is preference. In Dutch,
preference can be expressed both via necessity with zou beter/best 'should better/best'
References
Aksu-Ko9, Ayhan A. & Dan I. Slobin (1986). A psychological account of the development
and use of evidentials in Turkish. In Chafe & Nichols (eds.) 1986, 159-167.
Anderson, Lloyd B. (1982). The Terfect' s a universal and s a language-particular cate-
gory. In Paul J. Hopper (ed.), Tense-Aspect: Between Semantics and Pragmatics, 227-274.
Amsterdam: Benjamins.
(1986). Evidentials, paths of change, and mental maps: Typologically regulr asymmet-
ries. In Chafe & Nichols (eds.) 1986, 273-312.
Bader, Thomas, Iwar Werlen, & Adrian Wymann (1994). Towards a Typology of Modality:
The Encoding of Modal Attitudes in English, Turkish, Korean, Japanese and Tagalog.
(Arbeitspapiere, 32.) Bern: Institut fr Sprachwissenschaft, Universitt Bern.
Bech, Gunnar (1951). Grundzge der semantischen Entwicklungsgeschichte der hoch-
deutschen Modalverba. Det Kongelige Danske Videnskabernes Selskab, Historisk-lologi-
ske Meddelelser 32-6: 1-28.
Bhat, D. N. S. (1991). Grammatical Relations: The Evidence against their Necessity and
Universality. London: Routledge.
Birkmann, Thomas (1987). Prteritoprsentia: Morphologische Entwicklungen einer Sonder-
klasse in den altgermanischen Sprachen. Tbingen: Niemeyer.
Bjrkstam, Harald (1919). De modala hjlpverben isvenskan. L Tor, lr, mon, m, matte och
vill. Lund: Hkon Ohlsson.
Blanchon, Jean A. (1988). BE + infinitif et la possibilite. In Sylvianne Remi-Giraud (ed.),
L'infinitif, 167-178. Lyon: Presses Universitaires de Lyon.
Bolinger, Dwight (1980). Wanna and the gradience of auxiliaries. In Gunter Brettschneider
& Christian Lehmann (eds.), Wege zur Universalienforschung: Sprachwissenschaftliche
Beitrge zum 60. Geburtstag von Hansjakob Seiler, 292-299. Tbingen: Narr.
Bulygina, Tat'jana V. & Aleksej D. Smelev (1990). "Vozmoznosti" estestvennogo jazyka i
modaPnaja logika. [Natural language "possibilities" and modal logic.] In Vjaceslav Vs.
Ivanov (ed.), Jazyk logiki i logika jazyka: K 60-letiju professora V. A. Uspenskogo,
135-167. Moskva: Nauka.
(1991). Koncept dolga v pole dolzenstvovanija. [The concept of duty in the domain of
Obligation.] In Nina D. Arutjunova (ed.), Logiceskij analiz jazyka: Kul'turnye koncepty,
14-21. Moskva: Nauka.
Burridge, Kate (1995). From modal auxiliary to lexical verb: The curious case of
Pennsylvania German wotte. Paper read at the 1995 International Conference of
Historical Linguistics.
Bybee, Joan (1985). Morphology: A Study of the Relation between Meaning and Form.
Amsterdam: Benjamins.
(1988). Semantic substance vs. contrast in the development of grammatical meaning.
Berkeley LinguisticsSociety 14: 247-279.
(1991). Back to the future. In Traugott & Heine (eds.) 1991, 17-58.
Bybee, Joan & Suzanne Fleischman (1995). Modality in grammar and discourse: An intro-
ductory essay. In Joan Bybee & Suzanne Fleischman (eds.), Modality in Grammar and
Discourse, 1-14. Amsterdam: Benjamins.
Bybee, Joan & William Pagliuca (1985). Cross-linguistic comparison and the development
of grammatical meaning. In Jacek Fisiak (ed.), Historical Semantics and Historical Word
Formation, 59-83. Berlin: Mouton.
Bybee, Joan, Revere Perkins, & William Pagliuca (1994). The Evolution of Grammar: Tense,
Aspect and Modality in the Languages ofthe World. Chicago: University of Chicago Press.
Chafe, Wallace & Johanna Nichols (eds.) (1986). Evidentiality: The Linguistic Coding of
Epistemology. Norwood: Ablex.
Chung, Sandra & Alan Timberlake (1985). Tense, aspect and mood. In Timothy Shopen
(ed.), Language Typology and Syntactic Description, Volume 3: Grammatical Categories
and the Lexicon, 202-258. Cambridge: Cambridge University Press.
Coates, Jennifer (1983). The Semantics ofthe Modal Auxiliaries. London: Croom Helm.
Croft, William (1990). Typology and Universals. Cambridge: Cambridge University Press.
Croft, William, Hava Bat-Zeev Shyldkrot, & Suzanne Kemmer (1987). Diachronie semantic
processes in the middle voice. In Anna Giacalone Ramat, Onofrio Carruba, & Giuliano
Bernini (eds.), Papers from the 7th International Conference on Historical Linguistics,
179-192. Amsterdam: Benjamins.
DavidsenrNielsen, Niels (1990). Tense and Mood in English: A Comparison with Danish.
Berlin: Mouton de Gruyter.
de Haan, Ferdinand (1997). The Interaction of Modality and Negation: A Typological Study.
New York: Garland.
Rosengren, Inger (ed.) (1981). Sprache und Pragmatik: Lunder Symposium 1980. Lund:
Gleerup.
Satunovskij, Il'ja B. (1996). Semantika predlozenija i nereferentnye slova. [Sentence
Semantics and Non-referential Words.] Moskva: Skola "Jazyki russkoj kul'tury".
Shepherd, Susan C. (1982). From deontic to epistemic: An analysis of modals in the history
of English, creoles, and language acquisition. In Anders Ahlqvist (ed.), Papersfrom the
5th International Conference on Historical Linguisfics, 316-324. Amsterdam: Benjamins.
Stevens, Christopher M. (1995). On the grammaticalization of German knnen, drfen,
sollen, mgen, mssen, and wollen. Americal Journal of Germanic Linguistics and
Literatures 7: 179-206.
Sundmann, Marketta (1983). Svenska modalverbett kontinuum frn hjlpverb till
huvudverb [Swedish modal verbsa continuum from auxiliary to main verb]. In Struktur
och Variation: Festskrift till Bengt Loman, 321-334. Abo: Abo Akademi.
Swan, Michael (1980). PracticalEnglish U sage. Oxford: Oxford University Press.
Sweetser, Eve (1990). From Etymology to Pragmatics:Metaphorical and Cultural Aspects of
Semantic Structure. Cambridge: Cambridge University Press.
Tasmowski, Liliane & Patrick Dendale (1994). PouvoirE\ Un marqueur d'evidentialite.
Langue francaise 102: 4155.
Thurgood, Graham (1986). The nature and origins of the Akha evidentials System. In Chafe
&Nichols (eds.) 1986, 214-222.
Traugott, Elizabeth Closs (1972). A History of English Syntax: A Transformational Approach
to the History of English Sentence Structure. New York: Holt, Rinehart & Winston.
(1985). Conditional markers. In John Haiman (ed.), Iconicity in Syntax, 289-308.
Amsterdam: Benjamins.
(1989). On the rise of epistemic meaning in English: An example of subjectification in
semantic change. Language 65: 31-55.
Traugott, Elizabeth Closs & Bernd Heine (1991). Introduction. In Traugott & Heine (eds.)
1991,1-35.
Traugott, Elizabeth Closs & Bernd Heine (eds.) (1991). Approaches to Grammaticalization,
Volume 2: Focus on Types of Grammatical Markers. Amsterdam: Benjamins.
Valentin, Paul (1984). Zur Geschichte der Modalisation im Deutschen. In Sieglinde
Hartmann & Claude Lecouteuz (eds.), Deutsch-franzsische Germanistik: Melanges pour
Emile Georges Zink, 185-195. Gppingen: Kmmerle,
van der Auwera, Johan (1996). Modality: The three-layered scalar square. Journal of
Semantics 13: 181-195.
(forthcoming a). Negating dynamic and deontic modality in Hindi-Urdu and Bangla. In
Anvita Abbi (ed.), Proceedings of the 18th South Asian Language Analysis Roundtable.
New Delhi.
(forthcoming b). On the typology of negative modals. In Jack Hoeksema et al. (eds.),
Proceedings ofthe Groningen Conference on Negation.
Wagner, Johannes (1976). Eine kontrastive Analyse von Modalverben des Deutschen und
Schwedischen. International Review of Applied Linguistics in Language Teaching 14:49-66.
Warner, Anthony R. (1993). English Auxiliaries: Structure and History. Cambridge:
Cambridge University Press.
Wierzbicka, Anna (1992). Lexical universals and universals of grammar. In Michel Kefer &
Johan van der Auwera (eds.), Meaning and Grammar: Cross-linguistic Perspectives,
383-415. Berlin: Mouton de Gruyter.
Willett, Thomas (1988). A cross-linguistic survey ofthe grammaticalization of evidentiality.
Studies in Language 12: 51-97.
Wunderlich, Dieter (1981). Modalverben im Diskurs und im System. In Rosengren (ed.)
1981,11-53.