Language Sciences 27 (2005) 165–192

www.elsevier.com/locate/langsci

On defining modality again

Heiko Narrog
Graduate School of International Studies, Tohoku University, Aoba-ku, Kawauchi, Sendai 980-8576, Japan

Accepted 18 November 2003

Abstract

Perhaps no other grammatical category in linguistics has been defined and interpreted as
divergently as modality. Sometimes vastly different formal categories are subsumed under this
label. This paper focusses on the analysis of what are arguably the two major approaches to
the definition of modality in present-day linguistics, namely, the one that is based on ÔspeakersÕ
attitudes and opinionsÕ and the one that is based on a concept of realis/irrealis or factuality. It
is argued that the former is fundamentally flawed, as the definition fails to distinguish the for-
mal classes to which it does apply from those to which it do not. In contrast, modality can be
neatly and coherently defined in terms of Ôfactuality.Õ Speakers draw on the formal expressions
of this category as one among many categories as their resources to express their attitudes and
opinions.
Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: Assertion; Factuality; Modality; SpeakersÕ attitudes; Subjectivity

1. Introduction

The need for another discussion of how to define modality is obvious. There is
hardly any grammatical category which has been given more diverging definitions,
and under the label of which a wider range of phenomena has been studied. As Alan
King put it, ‘‘[u]ndeniably, [modality] is a subject of considerable complexity, fraught

E-mail address: narrog@mail.tains.tohoku.ac.jp

0388-0001/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.langsci.2003.11.007
166 H. Narrog / Language Sciences 27 (2005) 165–192

with possibly fuzzy, certainly slippery categories, [and is] poorly defined despite con-
siderable recent progress’’ (King, 1997, p. 362). The Japanese linguist Hiroshi Kudô
states the problem even more drastically: ‘‘modality [has become] the dustbin of
grammatical categories’’ (Kudô, 1989, p. 15).1
Some authors emphasize the relativity of definitions. So, for instance, van der
Auwera states that ‘‘[m]odality and its types can be defined and named in various
ways. There is no one correct way’’ (van der Auwera and Plungian, 1998, p. 80).
No doubt, this is a truism applicable to any concept in linguistics. On the other hand,
as a basic requirement, any definition should provide a basis to at least roughly de-
limit phenomena that belong to the area of investigation from those that do not.
Although this seems to be matter of course, I argue that in fact some definitions
of modality might not have lived up to the very concept of a definition. As a result,
vastly different categories have been subsumed under the single label, and it is essen-
tial to ask which of them have enough in common to justify the inclusion into a sin-
gle category. I suggest that by doing so it is possible to come to a definition that
fulfills its purposes.
I am aiming here at a cross-linguistic approach to modality that is not bound to
any specific linguistic theory. Three languages and their respective research traditions
will be mainly referred to, Japanese, English, and German. In a sense this selection is
arbitrary but it is certainly useful. English and German are closely related, but nev-
ertheless exhibit differences both in the language phenomena and in the approach to
their description and explanation. Taking Japanese into account significantly broad-
ens the perspective on the topic. Besides the considerable difference in the language
phenomena, Japanese might well be called the best-studied non-Indo-European lan-
guage in modern times. In the field of modality, the past fifteen years or so have seen
a flood of publications on the topic in Japan.
Below I will briefly discuss three preliminaries of this study, namely the level at
which modality should be defined (Section 1.1), the relationship between modality
and mood (Section 1.2), and the relationship between natural language modality
and modal logic (Section 1.3).

1.1. Modality as a semantic category

Modality traditionally has been viewed as a semantic, rather than syntactic or

morphological, category (cf. Bussmann, 1996, p. 307), and the reasons for this are
straightforward. Morphology and syntax differ significantly from language to lan-
guage. If, for instance, we defined modality as a category sharing the morphosyntac-
tic features of the English modal verbs, the definition would only suit the very same
modal verbs of English, and, viewed generously, to a lesser extent, modals in some
related languages. For a vast majority of other languages, this definition would be
pointless. A semantically based definition, on the other hand, offers ground for
cross-linguistic validity. It can be assumed that even speakers of structurally com-

1
Translations from Japanese and German are by the author if not stated otherwise.
 H. Narrog / Language Sciences 27 (2005) 165–192 167

pletely different languages share the same basic underlying cognitive processes that
enable and control linguistic categorization. The same basic experiences and needs,
that is ‘‘meanings’’, can be shared across people rather than the structures that are
used to convey them (cf. Heine, 1997, p. 3f., 14). This makes meaning more universal
than a particular syntactic or morphological structure, and more useful for our pur-
pose to define a cross-linguistically valid category. Furthermore, like many other
scholars, I also assume that modality is best defined on a semantic and not on a prag-
matic level.2 Semantics offers criteria to identify specific formal categories that share
semantic features, while pragmatics addresses the question about regularities in the
usage of such categories. Mechanisms in the pragmatic interpretation of meaning
usually operate across a conglomerate of formal and semantic categories, if they
are limited to any specific categories at all, and this makes it much harder to define
a category pragmatically.

1.2. Modality and mood

A distinction is usually made between modality and mood. While modality is

semantically defined, the term ÔmoodÕ refers to specific linguistic forms or paradigms
of forms, typically in verb inflection, whose primary function is to express modality
(cf. Declerck, 1992, p. 188; Frawley, 1992, p. 386; Palmer, 2001, p. 4). However,
modality and mood are not necessarily co-extensive. Thus, in many languages, not
mood but so-called Ômodal systemsÕ (e.g. modal verbs of English; cf. Palmer, 2001,
p. 4) are the typical means to express modality. Furthermore, and more crucially
for our discussion, not all functions of mood markers necessarily express modality,
and, on the other hand, markers of grammatical categories other than mood may
serve to express modal notions. This is a point about grammatical categories and
their forms of expression in general. Thus, for instance, with respect to modality,
the past tense form of the verb is used for irrealis marking in English conditionals,
and, on the other hand, the Spanish subjunctive marks notions that are, at least tra-
ditionally, not identified with modality, such as state-of-affairs that the speaker/wri-
ter assumes are already known by the hearer/reader (cf. Lunn, 1995, pp. 432–434).

1.3. Modality and modal logic

The scope of this study will not include formal logic, and no attempt at a formal-
ization will be made. Behind this stands the assumption that the formalization of a
concept of modality presupposes a non-formal understanding of the linguistic facts.
That is, I consider the non-formal (or pre-formal) understanding, at which I want to
arrive here, as more basic. It might also be added that only some of the modal
expressions of Japanese, which are a major subject here, can be reasonably defined

2
Particularly Dietrich (1992), Frawley (1992), and Kiefer (1997) are explicit on this point. Kiefer states
that, ‘‘[t]he act of asserting, asking, permitting, ordering or promising is alien to modality. There is no way
to ask a meaningful question about how such an act affects modality. This holds for illocution as well as
perlocution’’ (Kiefer, 1997, p. 247).
168 H. Narrog / Language Sciences 27 (2005) 165–192

in terms of possibility and necessity, the basic notions in modal logic. The same ap-
pears to be true for other languages of the world as well.3 Thus, it remains to be seen
whether the logical concept of modality is the best model for cross-linguisitic studies
of modality whose scope goes beyond Indo-European.4
In the following section of the paper, I will critically review what I consider the
major approaches to a definition of natural language modality in view of the pre-
mises discussed above. Given the enormous amount of literature in the field, it goes
without saying that only a selection can be offered, namely what I consider the most
important contributions to the discussion in semantically-oriented studies in Japa-
nese, German and English-language general linguistics.5 After pointing out the
advantages and deficiencies of each approach, I will present as a conclusion the def-
inition that I find the most feasible in the last Section 2.3.

2. Modality as defined in linguistics

Modality from a linguistic point of view is mainly defined in three different ways:
(1) modality as the expression of the attitude of the speaker, or the expression of sub-
jectivity and the speakerÕs opinions and emotions (e.g. Lyons, 1968, 1977; Palmer,
1986; Bybee et al., 1994; Nitta, 1989, 2000 for Japanese), (2) modality as something
including all linguistic expression outside the proposition (e.g. Fillmore, 1968; Ger-
stenkorn, 1976), and (3) modality as the expression of realis vs. irrealis or factuality
distinctions (e.g. Givón, 1995; Palmer, 1998, 2001; Dietrich, 1992 for German; Nar-
rog, 2002; Nomura, 2003 for Japanese). Furthermore, as will be seen below, elements
of these three approaches are sometimes combined, in particular (1) and (2) (e.g.
Sandhöfer-Sixel (1988) for German, Nakau (1994) for English, Masuoka (1991,
2000) for Japanese).
In Section 2.1. I will discuss approach (1). In Section 2.2 I discuss approach (2)
and its combination with approach (1), and in Section 2.3, a discussion of approach
(3), including a conclusion and my proposal for a definition, follows.

3
In this context, Palmer states the following: ‘‘[. . .] it is probable that the epistemi/deontic cum
possibility/necessity systems of modality are by no means universal, and it may be argued that the
logiciansÕ preoccupation with them is a reflection of the linguistic systems of only some of the languages of
the world, especially those of Europe.’’ (Palmer, 1986, p. 20). In the second edition of his book, Palmer
does not include possibility and necessity in basic modal categories (Palmer, 2001, p. 22).
4
I do not deny the usefulness of modal logic or the concepts of necessity and possibility as such. Thus, as
an anonymous reviewer has pointed out to me, McCawley (1993) provides a thoughtful application of
modal logic to natural language data. With respect to the concepts of necessity and possibility, there is the
example of Johan van der Auwera who bases cross-linguistic analyses of modality on the notions of
possibility and necessity (cf. van der Auwera and Plungian, 1998; van der Auwera, 2001). The purpose of
his research, however, has not been a cross-linguistically valid definition of modality. Rather, he uses the
terms of possibility and necessity with the practical purpose to delimit a field of research in which he can
carry out cross-linguistic comparisons.
5
A nice overview of views on modality in English up to the year 1987 is given by John (1987, pp. 1–7).
 H. Narrog / Language Sciences 27 (2005) 165–192 169

2.1. Modality as the expression of subjectivity

Some authors have viewed and defined modality as the expression of the speakerÕs
attitude, or more generally as ‘‘subjective’’ expressions. In modern linguistics this
concept can be traced back to the Indoeuropeanists Hanns Oertel and Karl Brug-
mann (cf. Noreen, 1923, p. 307). The latter defined grammatical moods such as
the imperative and subjunctive as ‘‘a statement about a mental mood of the speaker,
a subjective state with respect to which the [verbalized] event constitutes the deter-
mining objective side element to which the state is related’’ (Brugmann, 1970
(1904), p. 578). BrugmannÕs concept initially was in competition with the prevailing
19th century belief arising from the logical tradition, which viewed modality as the
relationship between predicate and subject (cf. Harweg, 1994, p. 4).6 Later, Otto Jes-
persen, following Brugmann, speaks of moods as syntactic categories ‘‘express[ing]
certain attitudes of the mind of the speaker towards the contents of the sentence’’
(Jespersen, 1992 (1924), p. 313). In saying so, however, Brugmann and Jespersen
were far from intending anything like a bipartitioning of the sentence. Brugmann
made an attempt at characterizing the verbal moods in Indo-European languages,
and the same aim is at the core of JespersenÕs approach.
John Lyons also had considerable influence on other linguists when he defined
modality as the grammatical marking of the ‘‘ÔattitudeÕ of the speaker’’ (cf. Lyons,
1968, p. 308), or more explicitly, ‘‘[the speakerÕs] opinion or attitude towards the
proposition that the sentence expresses or the situation that the proposition de-
scribes’’ (Lyons, 1977, p. 452). In the style of Lyons, in many successive studies of
modality we can find definitions of modality as ‘‘the attitude of the speaker’’ (Cal-
bert, 1975, p. 51), as ‘‘expressing the speakerÕs point of view on the sentence contents
and the speakerÕs communicative attitude’’ (Nitta, 2000, p. 81), or as the ‘‘grammat-
icization of speakersÕ (subjective) attitudes and opinions (Bybee et al., 1994, p. 176)’’,
to name just a few.7
There are two major problems with this approach that need to be pointed out.
The first is how to define terms like ÔsubjectiveÕ or Ôexpression of the speakerÕs atti-
tudeÕ.8 It is evident that those authors that base their definitions on this term, do

6
This concept gradually lost influence on modality concepts in English and German linguistic traditions
but was still fairly influential in Japan until around the 1970s, in the guise of so-called chinjutsuron
(predication theory).
7
With work on Japanese linguistics, it is always difficult to pin down the sources of the concepts used.
The reason is that in Japanese (domestic) linguistics there is not a strong tradition of fatihfully citing the
sources of ideas when adapting or passing on an idea. Many of the decisive scholars in Japan cited here
(like Nitta, 1989; Masuoka, 1991, 2000) simply do not name the sources for their concepts. Nevertheless,
there can be little doubt that Jespersen–Lyonian concept of ‘‘the speakerÕs attitude’’, whether directly or
indirectly, has been most influential in Japanese modality studies. Kudô (2000, p. 184) confirms this
impression.
8
As seen above, some authors speak of ‘‘attitude towards the proposition/contents of the sentence/
situation’’, etc., but the additional introduction of such terms does not contribute to the clarification of the
definition, as long as they themselves are not exactly defined. Consequently, we will omit them here from
the ensuing discussion, in order not to add further complications.
170 H. Narrog / Language Sciences 27 (2005) 165–192

not necessarily agree on this issue. In fact, it is challenging to give a definition of

ÔsubjectivityÕ or speakersÕ attitudes in language that does not involve a fair amount
of vagueness.9 The second, related issue is where to draw the borderline between sub-
jective and objective (or attitudinal vs. non-attitudinal), that is, how to determine the
actual extension of the category. This entails the question of whether, once one has
included certain linguistic categories or forms as modal, really everything, or at least
most of what has been identified is used subjectively, and everything, or at least most,
of what has been excluded is used objectively. If, for instance, tense is subsumed
under modality, as is sometimes done by proponents of the ‘‘modality equals subjec-
tivity/speakerÕs attitude’’ stance, is it true that all temporal expressions are subjec-
tive? If voice is thought to be not part of modality, which to my knowledge is the
assumption of all scholars equating modality with subjectivity/speakerÕs attitude,
are all expressions of voice objective?
Below, I will focus on the second problem. I will give examples from four catego-
ries to illustrate the difficulties involved in demarking the boundaries of the expres-
sion of speakersÕ attitudes/subjective expressions.

2.1.1. Traditional modality

I will use here the ad hoc terms ‘‘traditional’’ and ‘‘typical’’ modality to refer to
notions like possibility, probability, necessity, obligation, etc. (i.e. deontic and episte-
mic modality, with the modal verbs at its core in languages like English and Ger-
man), that have been traditionally treated as the core of modality, regardless of
the concrete definition. Depending on the definition, of course, other notions could
turn out to be central to the concept of modality.
It has been pointed out repeatedly in the literature on modality that typical modal
expressions such as modal verbs have both subjective and objective uses (cf. Lyons,
1977, pp. 797–805;10 Öhlschläger (1989) for German; Noda, 1989, pp. 132–133;
Onoe, 1997; Kudô, 2000, pp. 185–187; for Japanese).11 Lyons has discussed this issue
in some detail, arguably with artificial examples, but other scholars like Coates
(Coates, 1983) were able to show subjective and non-subjective uses in actual
discourse.

9
The terms ‘‘subjective’’ and ‘‘[expressing] the speakerÕs attitude’’ will be regarded here as largely
synonymous. For one thing, most authors do not distinguish clearly between them. For another thing, we
may assume that ‘‘speakersÕ attitudes, opinions’’, etc. are intrinsically subjective.
10
Note that this is a potential contradiction to his definition of modality as ‘‘the speakerÕs opinion or
attitude’’ elsewhere in the same book. LyonsÕ approach to modality, like that of many other authors as
well, is basically eclectic, using terms like ‘‘speakerÕs attitudes’’, ‘‘factivity’’, or ‘‘subjective–objective’’ each
in such contexts where they appear to be useful.
11
Keisuke Onoe also points out the following:

Among the meanings expressed by modal forms there may be subjective ones, non-subjective ones, and
both at the same time (Onoe, 1997, p. 70f).

In other words, if one divides the modal system of a language by forms or by categories, many markers or
categories will have both subjective and non-subjective uses.
 H. Narrog / Language Sciences 27 (2005) 165–192 171

Let us assume now, as most scholars writing on modality do, that forms like
modal verbs are expressions of modality, and thus, in the view of those scholars
who identify modality with subjectivity, expressions of subjectivity. However, the
following examples of English must, according to Coates (1983), can be thought
of as representing non-subjective uses of expressions of modality (the notation of
the examples is abbreviated):

(1) Clay pots . . . must have some protection from severe weather (from written dis-
course; Coates, 1983, p. 35).
(2) ‘‘ Ô. . . have you noticed president, that, em, the boiled egges at Sunday breakfast
always hard?Õ and president said ÔAh, well, the simple truth is that, if youÕre
going to boil egges, communally, they must be hard . . . You have to crack the
head of an egg, when you take it out of the pan—otherwise it goes on cooking.Õ ’’
(from conversation; Coates, 1983, p. 42).

Note that (1) is an example for the so-called root or deontic use of must, and (2) is an
example of epistemic use. This is significant because some authors draw a borderline
between deontic and epistemic modality, the former being regarded as objective and
the latter as subjective (e.g. Helbig and Buscha, 1996; Lindemann, 1997). For some
of them, accordingly, deontic modality categorically does not belong to modality at
all (e.g. John, 1987; Sandhöfer-Sixel, 1988), whereby it remains unclear which cate-
gory, if any, it should belong to then. For others, it is only a ‘‘pseudo-form’’ of
modality (e.g. Nitta, 1989).
However, not only do there exist objective uses of epistemic modality, albeit infre-
quent, deontic modality also has fairly subjective uses as can be seen in the following
examples from Coates (1983).

(3) ‘‘You must tell me how to get to it.’’ (from conversation, Coates, 1983, p. 34)
(4) ‘‘. . . interesting ideas which surprised me rather . . .. . . I must admit’’ (from con-
versation, Coates, 1983, p. 34).

In (3), the speaker rhetorically imposes the obligation on the hearer to tell her or
him something. In (4), the speaker expresses that she or he feels forced to admit
something. However, arguably neither in (3) nor in (4) does must express any
kind of objective obligation, but rather the speakerÕs wish (in (3)) and evaluation
(in (4)).
In conclusion, typical expressions of ‘‘traditional’’ modality can have both objec-
tive and subjective uses, although the latter are certainly more common, particularly
in conversation. This demonstrates that it is hardly possible to classify the traditional
category of modality as a whole as either subjective or non-subjective. It is further-
more impossible to draw a clear borderline between subjective and non-subjective
uses within traditional modality. Rather, each use of a modal marker has its own
share of subjectivity or objectivity within its specific context. This makes traditional
modality a bad candidate to form the core of a category defined in terms such as
ÔsubjectivityÕ or ÔspeakersÕ attitudes.Õ
172 H. Narrog / Language Sciences 27 (2005) 165–192

2.1.2. Voice
I am not aware of any author propagating the equation of modality with subjec-
tivity or the speakerÕs attitude who would not exclude voice from the scope of modal-
ity (see, for example, Nitta, 1997, pp. 124, 136 and Masuoka (2000, 89) where the
exclusion is made explicit).
In contrast to this, there are uses of voice categories that are exclusively identifi-
able with the expression of speakersÕ attitudes, e.g. the ÔadversativeÕ passives (cf. Tsu-
jimura, 1996, p. 238) in Japanese.
The following passive sentence in Japanese is an example where a non-subjective
neutral passive is used.

(5) [. . .] seisyo kenkyuu-kai-ga tetudoo-no ryoo-de hirak-are-ta.12

[. . .] bible study-group-NOM railway-GEN dorm-LOC open-PASS-PST
‘‘[The] bible study group was held at the railway dormitory’’
(Shiogaritôge, Miura Ayako)

The passive in this sentence, hirak-are-ta Ôwas heldÕ indeed can be regarded as an
expression of an objective description.
However, the use of the passive in the following sentences looks rather subjective:

(6) Kore-de ame-ni-demo hur-are-tara, sanzan-des-u-yo.

this-ESS rain-DAT-EXPL rain-PASS-COND awful-COP-NPST-EXC
‘‘If we got rained on, that would be awful.’’ (Hôrôki, Hayashi Fumiko)
(7) Ni-hiki-dake sime-ta-ga, ato-wa nige-rare-ta.
Two-CLS-LIM catch-PST-CTR, other-TOP flee-PASS-PST
‘‘I managed to catch two of them, but the others fled me (=literally Ô[I] was fled
by the othersÕ).’’ (Nobi, Ôoka Shôhei)

The passivized verb is intransitive. The sole purpose of passivizing such an intran-
sitive verb is to express negative affectedness of the subject, either as experienced by
the speaker-subject, or implied, in the case of a non-speaker subject.13 As such it re-

12
The examples here are taken from digitized novels. Thus page numbers are not provided. Example
sentences without explicit reference are made-up sentences. The transcription and morphological analysis
is based on Narrog (1998).
Abbreviations: (1) Japanese: NOM, nominative case; GEN, genitive case; DAT, dative case; ACC,
accusative case; ESS, essive case; LOC, locative; EXPL, exemplative; NOM, nominalizer; TOP,
topicalizer, PASS, passive; PST, past tense; NPST, non-past tense; COND, conditional; CTR, contrastive;
COP, copula; EXC, exclamative; AFF, affirmative; VER, verification; CLS, nominal classifier; LIM,
limitative. (2) Spanish: 2SG, second person singular; SUBJ, subjunctive; PRES, present tense.
13
Yoshida, in her study of the passive in conversation found that if the passive is used to convey ‘‘the
speakerÕs emotive and evaluative attitudes,’’ the passive subject is most commonly the first person. Second
and third person subjects are ‘‘possible when the speaker feels close enough to identify with the referent’’
(Yoshida, 1996, p. 127).
 H. Narrog / Language Sciences 27 (2005) 165–192 173

flects the speakerÕs subjective evaluation of the effects of an event. This would per-
fectly fit the above definitions of modality. Presumably, so would the use of the pas-
sive in the following examples:

(8) ‘‘Kinoo benkyoo-si-ta-no?’’ ‘‘Un, s-ase-rare-ta.’’

yesterday study-do-PST-NOM Yes do-CAUS-PASS-PST
‘‘Did you study yesterday?’’ ‘‘Yes, I had to’’ (=literally Ôwas let madeÕ).

The combination of causative and passive as well has the sole purpose of expressing
negative affectedness. The subject was made to do something against her/his wishes.
Here again, the speakerÕs point of view is clearly expressed, and consequently voice
in this context would have to be classified as modality, contrary to actual
classifications.
One might object that it should be possible to isolate adversative passives and
causative-passives from the rest of voice and include them in modality. However, this
is not a viable option. When the passivized verb in Japanese is transitive, it is impos-
sible to tell from the surface form whether this is the expression of some objective
state of affairs, or whether it is used as a means to express negative affectedness
(cf. Kuno, 1983, pp. 205–213 for a thorough discussion of various factors that influ-
ence the interpretation):

(9) ‘‘Boku-wa kinoo, sensei-ni sikar-are-ta.’’

I-TOP yesterday teacher-DAT scold-PASS-PST
‘‘I was scolded by the teacher yesterday.’’

2.1.3. Referent honorification

In the same manner, markers of referent honorification, a category that unani-
mously is classified as non-modal, can be used to exclusively express the speakerÕs
attitude. Unnecessitated use of honorifics, for instance, is a means to express criti-
cism and sarcasm. Take the following example from natural conversation, cited by
Okushi (Okushi, 1997, pp. 158f).

(10) Son: ‘‘Nee, mada tumetai otya aru?’’

VER still cold tea be
‘‘(I) say, is (there) still cold tea (in the refrigerator)?’’
Mother: ‘‘Gozaimasu-wa’’
be.HON-AFF
‘‘There is.’’

Using the highest available honorific form in a context where the plain form would
be most appropriate, the speaker expresses aversion towards her sonÕs attitude who
expects her to wait on him. Through the analysis of the use of honorifics in actual
conversation, Okushi comes to the conclusion that there is a distinct use that
expresses ‘‘not formality in the traditional sense’’ (Okushi, 1997, p. 162).
174 H. Narrog / Language Sciences 27 (2005) 165–192

Again, it is only reasonable to assume that through the use of honorifics the
speaker has expressed his or her attitude, and the honorific forms would have to
be classified as modality. Even if no irony were involved, referent honorifics presup-
pose the speakerÕs assessment of his or her own social position in relation to the
hearer in the speech situation. As such, it can be said to belong to modality if the
category is defined as the expression of the speakerÕs attitude or stance.

2.1.4. Aspect and other categories

A related and well-known example for a category of English that is usually iden-
tified with propositional content, but can be used highly subjectively is the English
Progressive as analyzed by Wright (1995). Wright interprets ÔexperientialÕ progres-
sives, such as in the following example as non-aspectual foregrounding strategies,
‘‘focussing on an experience, perception, observation from inside the speakerÕs con-
sciousness’’ (Wright, 1995, p. 156):

(11) YouÕre telling me you donÕt love me anymore.

Wright comments that this example ‘‘functions as the speakerÕs evaluation of her
interlocutorÕs utterance in terms of the real salience of the performance witnessed’’
(Wright, 1995, p. 157). According to her, this usage of the Progressive does not have
an aspectual or temporal function. Rather, it conveys ‘‘[the speakerÕs] epistemic stance
at a particular moment in the context of the utterance’’ (Wright, 1995, p. 157). With
respect to Japanese, Fujii (1992) shows that in actual text the completive (-te simau)
most frequently marks the speakerÕs emotions and evaluation, and Nishikawa (1995)
argues it is a strong expression of the speakerÕs attitude and as such has to be included
in modality. Shinzato shows that the unmarked -(r)u/ta verb endings are associated
with subjectivity in contrast to the distancing -te iru (Shinzato, 2003).
Other categories in Japanese that have been explicitly linked to the expression of
the speakerÕs attitude and modality include the speakerÕs viewpoint (-te kureru, -te
yaru, etc.; cf. Nitta, 1999) and comparison (-yoo na, etc.; Koyano, 1997). Taking into
account that categories like tense, negation, and addressee honorification are viewed
as part of modality by proponents of the ‘‘speakersÕ attitudes equal modality’’ con-
cept in any event, there is no other grammatical category of the verb in Japanese that
cannot, and has not been linked to speakersÕ attitudes and subjectivity, and conse-
quently, modality.
Apart from grammatical categories, even non-grammatical devices in language
can have a strongly subjective touch. Terms of address in Japanese are a good exam-
ple. The same person may be referred to as Tanaka-san (Ms Tanaka), Tanaka-sensei
(Prof Tanaka), Hiroko, anata (you (polite)), omae (you (non-polite)), etc., depending
on the situation. Likewise, in English, the same event may be referred to as steal,
heist, hook, lift, nab, etc.,14 not necessarily referring to any particular difference in

14
Similar examples corresponding to English steal and variations could be provided for Japanese
(nusumu, toru, pakuru, etc.) and German (stehlen, klauen, mitgehen lassen, etc.).
 H. Narrog / Language Sciences 27 (2005) 165–192 175

the described state-of-affairs, but rather expressing the speakerÕs stance and her/his
attitude towards the hearer.

2.1.5. Discussion
The examples I have provided above illustrate that for many grammatical catego-
ries and forms, a borderline between objective and subjective (or attitudinal vs. non-
attitudinal) simply does not exist. Rather, the ÔspeakerÕs attitudeÕ in the present
speech situation, and her/his subjectivity can be expressed pervasively throughout
all elements of the sentence. That begins with the choice of vocabulary, continues
with the perspectivization chosen to conceptualize a situation, as expressed in voice
and aspect, and continues to include such categories as tense and modality (in a tra-
ditional sense). For most grammatical categories, the degree of subjectivity with
which their formal expressions are used is dependent on usage and context, and thus
they do not render themselves to a taxonomy that draws a clear line between subjec-
tive and non-subjective linguistic forms.
Now, the consequence for those scholars who take their own definition of modal-
ity as the ‘‘expression of the speakerÕs attitude’’ seriously should be (1) including in
their study of modality a large variety of linguistic forms across many grammatical
categories, including those cited above, (2) to define as the core of the study formal
categories that predominantly, if not exclusively express the speakerÕs attitude, such
as modal particles (Modalpartikeln) in German (cf. Krivonosov, 1977,15 Feyrer,
1998), or the final particles (shûjoshi) in Japanese (cf. Noda, 2002), and not the forms
of traditional modality such as the modal verbs of English, (3) abandoning the cat-
egorization of these forms as one coherent grammatically definable category. The
problem with the concept of modality in works such as the above (e.g. Bybee
et al. (1994) or Nitta (1989, 2000), is that while modality is defined in terms such
as subjectivity or speakerÕs stance, this definition is not followed up when its formal
expressions are identified. Instead, the expressions of traditional modality are pri-
marily identified, and other categories highly expressive of the speakerÕs attitude
are arbitrarily excluded.
In fact, there is at least one research program in modality that takes its own claims
about subjectivity seriously, and this is the concept of discourse modality as champi-
oned by Senko Maynard. Maynard defines discourse modality as follows: ‘‘Dis-
course Modality refers to information that does not or only minimally conveys
objective propositional message content. Discourse Modality conveys the speakerÕs
subjective emotional, mental or psychological attitude toward the contents of the
message, the speech act itself or toward his or her interlocutor in discourse’’ (May-
nard, 1993, p. 38). Accordingly, linguistic expressions like the final particles in Jap-
anese, connectors like datte ÔbutÕ, and adverbs like yahari ÔstillÕ, are treated as typical
instances of forms to which this concept applies. In other words, discourse modality
is expressed throughout the sentence, and representative form classes expressing this

15
cf. Krivonosov (1977, p. 191f) ‘‘The modal particles are structural formants that the speaker introduces
into the sentence if (s)he deems it necessary to express her/his subjective attitude towards the contents or
his/her emotions.’’
176 H. Narrog / Language Sciences 27 (2005) 165–192

type of modality include markers of illocutionary force16 (Van Valin and LaPolla,
1997, p. 41f), and discourse markers as defined by Deborah Schiffrin (Schiffrin,
1987, p. 31), but not the markers of modality in the traditional sense.
MaynardÕs definition might not be without its problems.17 However, by distin-
guishing discourse modality from the grammatical category modality, and trying
to establish the former as the expression of the speakerÕs subjectivity at the level
of pragmatics, she has shown consequence and pointed in a promising direction.18
If, however, the projected object of research is the traditionally defined domain of
modality, expressing concepts like obligation, volition, conclusion and inference,
ÔspeakersÕ attitudes and opinionsÕ, or ÔsubjectivityÕ are certainly not suitable concepts
that would define this domain. As stated above, I would even suggest that there is no
category at all which can be reasonably defined as the expression of subjectivity.
Apart from the difficulty of how to define subjectivity in a manner that allows
researchers to distinguish subjective from non-subjective expressions, what one gets
when one looks for the expression of subjectivity in language is exactly the thing
looked for, namely expressions of subjectivity, and not a grammatical category. In
actual discourse, one finds expressions of subjectivity across the lexicon and across
grammatical categories. Arguably, subjectivity is not an intrinsic feature of certain
forms or form classes but a condition of use, some forms and categories being more
likely to be utilized as a resource for the expression of subjectivity or the speakerÕs
stance than others.19

16
Illocutionary force: ‘‘[. . .] refers to whether an utterance is an assertion, a question, a command or an
expression of a wish.’’ (Van Valin and LaPolla, 1997, p. 41). Thus, grammatical markers that in discourse
impart interrogatory, imperative, assertive and related functions to an utterance may be regarded as
markers of illocutionary force.
17
‘‘Discourse Modality conveys the speakerÕs subjective emotional, mental or psychological attitude
toward the contents of the message, the speech act itself or toward his or her interlocutor in discourse’’
(Maynard, 1993, p. 38); emphasis by author.
There is one obvious problem with the definition. The ‘‘mental or psychological attitude’’ that is to
become the object of research on modality has to be expressed in the speech act itself. If modality is a
comment about the speech act, this would mean that modality is viewed as a meta-linguistic category,
which is, presumably, not the intention of Maynard.
18
The type of research that Maynard represents has been increasingly productive in pragmatics and
discourse studies over the past years (e.g. Schiffrin, 1987; Givón, 1995; Brinton, 1996), extending to a
variety of languages. On the other hand, the proposed term, ‘‘discourse modality’’ has not caught on yet.
19
Keisuke Onoe has has made the following interesting observation with respect to the question why
modality and ‘‘subjective expressions’’ tend to be mixed up by some linguists: ‘‘The original concept of
modality in linguistics was not one of attitude of expression or subjectivity in general, but the meaning that
resulted when an irreal state-of-affairs was expressed by specific predicative forms [for this purpose]. In a
sentence that reports state of affairs that exist in reality, it is possible to abstract away from the subjectivity
that arises from the act of utterance itself and treat them as given sentence contents, but in a sentence that
reports non-real things, it is not possible to abstract away from the purpose of utterance, the motivation
and [the speakerÕs] feelings. This is the reason why with respect to modality subjectivity often stands out.
However, one cannot claim that all modal expressions are subjective, and, as a matter of course, not all
subjective meanings have something to do with irreality’’ (Onoe and Tsuboi, 1999, p. 103f).
 H. Narrog / Language Sciences 27 (2005) 165–192 177

2.2. Modality being simultaneously identified with non-proposition and the

subjective elements of the sentence

Perhaps Charles Bally is the first modern scholar who preached a bipartitioning of
the sentence in subjective and objective elements (Bally, 1965 (1932), p. 36). Bally dis-
tinguished dictum and modus, and defines the former as the expression of represen-
tations that the speaker has received from the inner and the outer world, and the
latter as the expression of the ‘‘speaking self’’ or ‘‘thinking self’’. The modus bears
more importance than the dictum; without it there would be no sentence. Consider-
ing the concrete examples that Bally provides for the extension of these categories, it
is clear that BallyÕs concepts only partly overlap with more recent concepts of prop-
osition and modality.20 However, in distinguishing expressions of objective content
from expressions of the speakerÕs self, Bally influenced later research on modality
that favors a similar conception.
In an influential paper, Fillmore (1968) sketched modality as a constituent com-
plementary to the proposition in sentence structure, and thus comprising basically
everything outside the proposition. Fillmore defined the proposition as the ‘‘tenseless
set of relationships involving verbs and nouns.’’ All other elements of the sentence,
that is negation, tense, mood, and aspect, would accordingly belong to modality. He
represented the relationship between proposition and modality in the following for-
mula (Fillmore, 1968, p. 24):
sentence ! modality þ proposition
It must be mentioned, however, that Fillmore presented this concept merely in
order to clarify the concept of proposition on the way towards introducing his theory
of case grammar. The goal was not a definition or discussion of the concept of
modality itself. Accordingly, he argues that ‘‘[t]he exact nature of modality may
be ignored for our purposes,’’ and does not attempt to furnish his theory with any
more detail.
FillmoreÕs concept of modality as presented above was put forward rather casu-
ally. Still, other authors, focussing on modality per se, have followed this definition
closely. While Fillmore seems to have viewed the sentence as an addition of propo-
sition and modality, Alfred Gerstenkorn, for instance, adopting FillmoreÕs biparti-
tion of constituents of the sentence, located modality on a different level than
proposition, the layer of modality dominating the layer of proposition. This view
is represented in the formula below (Gerstenkorn, 1976, p. 13)
MOD (MAT),21
where MOD stands for modal and MAT for material, i.e. the propositional elements.
Provided that modality expresses some type of judgement about the propositional

20
Bally provides a few concrete examples for his categories, but does not explain systematically what
kind of linguistic expressions would fall under the categories dictum and modus, respectively.
21
It must be cautioned, however, that GerstenkornÕs modality and proposition do not completely overlap
with those of Fillmore. An explanation will follow below.
178 H. Narrog / Language Sciences 27 (2005) 165–192

contents, and not vice versa, we must conclude that this formula more accurately
renders the relationship between modality and proposition as sentence constituents.
Making the attempt at bi-partitioning the sentence into two components, one of
which may be named modality, in itself is a valid undertaking. Now, crucially, a
questionable interpretation of this provisional bi-partitioning of the sentence has
been a major source for some of the profound confusion witnessed in definitions
of modality in the past three decades: definitions for partitioning the sentence struc-
ture into proposition and non-proposition have been used to identify the grammat-
ical category of modality on the same level with other categories such as tense,
aspect, and negation. This results in a contradictory view of modality as a category
on a par with other traditional categories, which at the same time would somehow
comprise these categories. This view can be found in the work of some scholars
on European languages and has been highly influential in research on modality in
Japan.
Drawing on Fillmore, Gerstenkorn, and Lyons, Judith Sandhöfer-Sixel contrasts
modality and proposition in German, defining modality as ‘‘the expression of [the
speakerÕs] subjective attitude towards the proposition’’ (Sandhöfer-Sixel, 1988, p.
14). In his research on Japanese modality, Masuoka Takashi claims that ‘‘proposition
and modality are the two big elements that make up a sentence. I define them as the
element that expresses objective facts, and the element that expresses subjective
judgements and attitudes, respectively’’ (Masuoka, 1991, p. 6, also see Masuoka,
2000, p. 87). Both authors explicitly conflate the notion of elements of the sentence
outside the proposition with elements that express subjective judgements and
attitudes.
Now, interestingly, among theories that coincide with respect to the bipartition of
sentence into objective proposition and subjective modality, substantial differences
concerning the concrete extension of both categories can be found. Note that the his-
torical successors to Fillmore and Gerstenkorn, named above and below, reidentify
FillmoreÕs and GerstenkornÕs modality constituent as the grammatical category with
the name modality. In my view, this is a decisive misunderstanding. We have already
seen in Section 2.1 that identifying modality with subjectivity is highly problematic.
Now, for these authors, modality is not only identified with subjectivity, but also has
to exactly coincide with non-propositionality, and unlike in the case of MaynardÕs
Ôdiscourse modalityÕ, the product is supposed to be a coherent grammatical category,
such as voice or tense are grammatical categories. The outcome is a miscellany of
categories, varying widely by author. Fillmore, who unlike his more recent succes-
sors did not intend to define a new category, considered negation, tense, mood,
and aspect as examples of modal categories. Nakau, on the other hand, as a propo-
nent of Ôthe bipartite structureÕ of the sentence referring to the same language as Fill-
more, namely English, does not include negation, tense, and aspect in modality
(Nakau, 1994, pp. 53–60). With respect to Japanese, Masuoka lists up the following
types of modality: (1) modality of speech attitude, typically expressed by sentence-
final particles; (2) modality of politeness, expressed by politeness markers; (3) pattern
of expression modality (corresponding to sentence mood); (4) modality of truth
judgement (corresponding to epistemic modality); (5) modality of value judgement
 H. Narrog / Language Sciences 27 (2005) 165–192 179

(corresponding to deontic modality); (6) modality of tense; (7) modality of explana-

tion, often expressed by the sentence nominalizer no(-da); (8) modality of topic and
focus, expressed e.g. by the particle wa (Masuoka, 1991, pp. 47–59). This is a strik-
ingly wide area, but he does not include aspect or voice. GerstenkornÕs notion of
modality for German covers negation, strength of assertion (including expressions
of possibility and probability), emphasis, and grammatical mood (such as indicative
and subjunctive) (Gerstenkorn, 1976, pp. 14–35). Tense and aspect are regarded as
belonging to the ÔmaterialÕ expressions that make up the proposition.22 Sandhöfer-
Sixel considers only expressions of emotional evaluation and expressions of the
validityof the state-of-affairs, including negation, realis/irrealis, and epistemic possi-
bility as modal, and contrasts these categories with the proposition (Sandhöfer-Sixel,
1988, pp. 13–50). In Table 1, I have contrasted the scope of modality as perceived by
all these authors, adding one more for Japanese (Miyazaki (2002), largely corre-
sponding to Nitta (1989)).
Clearly, definitions that treat modality at the same time as a grammatical category
and as the sum of linguistic elements and expressions outside the proposition, and
furthermore equating Ôoutside the propositionÕ with notions like subjectivity or
ÔspeakersÕ attitudesÕ result in enormous differences in the perceived scope of the cat-
egory. To be sure, some differences might be attributed to the differences in linguistic
facts of the particular language the researcher is focussing on. More importantly,
however, the divergence in the perceived scope of modality as defined by these schol-
ars is an inevitable result of the problems of the definition itself.
The essential issue here is that there are no particular linguistic facts that would
provide a foundation for identifying a well-defined set of linguistic expressions Ôout-
side the proposition,Õ while at the same time expressing the speakerÕs attitude, and
then also constituting a grammatical category of its own. As I argued above, this def-
inition is probably the result of a misunderstanding, reidentifying the modality as a
grammatical category with the concept of modality as a sentence constituent.
The problems of the approach discussed here can be summed up as follows: How
can a grammatical category be defined in opposition to Ôproposition,Õ and where can
the line be drawn between them? Masuoka, for instance, while advocating a bipolar
partition of categories, concedes the following: ‘‘When adopting the bipartition [of
sentence] into proposition and modality the biggest problem is [to distinguish] which
elements concretely belong to the proposition, and where modality starts. [. . .] This
study does not have a concrete criterion at its disposal to divide the elements into
proposition and modality’’ (Masuoka, 1987, p. 9). Although no other author names
the problem with the same frankness as Masuoka, I am not aware of any study advo-
cating the Ôbipartite structure of the sentenceÕ, to which MasuokaÕs statement would

22
With respect to Japanese, the modality concept of Yoshio Nitta comes close to this theory. Nitta (Nitta
(1991)) divides the sentence in Japanese into a layer of Ôexpressed situationÕ (genpyô jitai) and Ôexpressed
attitudeÕ (genpyô taido). ÔExpressed situationÕ besides the core composition consists of aspect, voice, and
tense. The layer of Ôexpressed attitudeÕ includes modality and politeness. In this manner, his definition of
modality is somewhat narrower than that of Gerstenkorn. However, Nitta does not provide any grounds
for distinguishing these two layers (cf. Nitta, 1991, p. 17).
 180
Table 1
FillmoreÕs and GerstenkornÕs notion of the bipartite structure of sentence, and theories that define modality as a grammatical category based on such a
bipartite structure
Fillmore Gerstenkorn Nakau Sandhöfer-Sixel Masuoka Miyazaki (2002)
(1968) (1976) (1994) (1988) (1991, 2000)a

H. Narrog / Language Sciences 27 (2005) 165–192

Object of definition Sentence System Semantic category Semantic category Grammatical category Functional–semantic
constituent (‘‘Inhaltskategorie’’) category
Language(s) of English German English German Japanese Japanese
concern
Included in Tense, Negation, Grammatical mood, Negation, Speech attitude, Expression of sentence
modality types (order, wish,
statements, interrogatives,
etc.),
Aspect Strength of Negation and Emotional evaluation, Politeness, Expression of epistemic
assertion, quantifiers, and deontic modalities;
Mood Emphasis, Probability, Possibility, Validity, Pattern of expression, Explanation,
Negation Mood Conjunctions, Realis/irrealis, Epistemic modality, Sentence-final particles
Discourse markers, Epistemic possibility Deontic modality (1991),
Type of sentence, Tense (1991),
Topic and focus Explanation,
Topic and focus,
Negation (1991)
Categories explicitly ?? Tense, Negation, Deontic modality, Aspect, Tense,
excluded
Aspect Tense, Evidentiality Voice Aspect, voice, etc.
Aspect
a
Those categories in MasuokaÕs column marked with (1991) are only categorized as part of modality in Masuoka (1991) but not in Masuoka (2000).
 H. Narrog / Language Sciences 27 (2005) 165–192 181

not equally apply. Add the difficulty discussed in the previous section, of how to de-
fine a category in terms of subjectivity, and the problems multiply.
As a further complication, with a concept of modality as a grammatical category
including everything non-propositional, categories of different levels of description
are lumped together in one category. For example, categories like tense, negation
and traditional modality, which to a large extent can be dealt with at a semantic–syn-
tactic level, are combined with categories such as politeness and speech attitude (final
particles) that are primarily grounded at a discourse level in the spoken language,
referring, for instance, to the social relationship between speaker and hearer. Com-
bining things that are defined negatively, namely as not belonging to the proposition,
or vaguely defined positively in terms of ÔspeakersÕ attitudesÕ or ÔsubjectivityÕ, effec-
tively has created the ‘‘dustbin of grammatical categories’’ that modality has recently
become in some theories, a category conflated out of proportion.

2.3. Modality as the expression of factuality or realis/irrealis distinctions

A third, increasingly strong trend in the study of modality refers to the notions of
Ô(realis vs.) irrealisÕ (e.g. Mithun, 1999; Palmer, 2001), ÔfactualityÕ (also ÔfactivityÕ,
ÔrealityÕ, ÔactualityÕ (e.g. Chung and Timberlake, 1985; Papafragou, 2000; Narrog,
2002; Nomura, 2003), or ÔvalidityÕ (Kiefer, 1987, 1997; Dietrich, 1992) in its defini-
tions. As we will see below, the difference between these notions basically lies in
labeling only. In addition, some authors bring the concept of ÔassertionÕ into play
(Givón, 1995; Palmer, 2001), a concept which substantially differs, and which is
therefore discussed here in a separate Section 2.3.1).
Notions linking modality with ÔrealityÕ or ÔfactivityÕ or ÔvalidityÕ had already en-
tered the discussion earlier. John Lyons, for example, linked modality to non-factiv-
ity (Lyons, 1977, p. 794f), and Palmer in the first edition of his book on mood and
modality, linked it to ÔfactualityÕ (1986, p. 17f),23 both simultaneously defining
modality as related to the speakerÕs attitude in an eclectic manner. Abraham defined
modality as the ‘‘attitude of the speaker to the validity of the state of affairs to which
the utterance refers’’ (Abraham, 1988, p. 489). In Russian linguistics, Vinogradov in
the late 1940s founded a tradition linking modality to the relationship between sen-
tence and reality (Kristophson, 1994, p. 40). In Japan, this tradition influenced schol-
ars who analyzed Japanese on the basis of Russian linguistic theory (cf. Suzuki,
1972.24).
Now, while notions such as realis/irrealis have been alluded to for decades, a trend
can be perceived in which they are no longer treated as ancillary to the ÔspeakerÕs atti-
tudeÕ or a similar notion.

23
Palmer in the first edition of his book correctly criticizes the term ÔfactivityÕ and instead proposes the
term ÔfactualityÕ (Palmer, 1986, p. 17f).
24
Shigeyuki Suzuki defines modality as follows: ‘‘Modality (the mood of the sentence), is the linguistic
expression of the the speakerÕs reality and her or his attitude towards the hearer, with relation to the
material contents of the sentence’’ (Suzuki, 1972, p. 44).
182 H. Narrog / Language Sciences 27 (2005) 165–192

Thus, Kiefer, while acknowledging the importance of the speaker, states that
‘‘[t]he essence of modality consists in the relativization of the validity of the sentence
meanings to a set of possible worlds (Kiefer, 1997, p. 243),’’ and Papafragou gives
the following definition: ‘‘Modal expressions allow us to talk (and modal concepts
allow us to think) about states of affairs which are not present in the current situation
and may never occur in the actual world (Papafragou, 2000, p. 3).’’
Now, one of my claims in this paper, a claim not yet made explicitly as far as I
know, is that we do not need ÔspeakersÕ attitudesÕ at all to define modality. Rather,
it is advantageous to completely exclude them, because, as I argued in the previous
sections, this notion, if taken seriously, makes it impossible to delimit grammatical
form classes that belong to the category from those that do not. Choosing the no-
tions of factuality/actuality/validity, etc. has three obvious advantages. The first is
that the rather vague notion of speakerÕs attitude is abandoned in favor of a notion
that is more suitable to actually delimit the category. Further, the result is a seman-
tically defined grammatical category on the same level as tense, aspect, negation, etc.
Third, if the object of definition is meant to be a category with traditional modal no-
tions such as necessity, obligation, possibility, and probability at its core. I suggest
that this is the only feasible definition. Below, we will discuss this type of definition
and the benefits I claim for it in more detail.
ÔIrrealisÕ according to Mithun, ‘‘portrays situations as purely within the realm of
thought, knowable only through imagination’’, while ÔrealisÕ ‘‘portrays situations as
actualized, as having occurred or actually occurring, knowable through direct per-
ception’’ (Mithun, 1999, p. 173). Factuality (i.e. nonfactuality in this case) stands
for expressions that ‘‘do not make straightforward statements of facts or categorical
assertions’’ (Palmer, 1998, p. 230). Validity, according to Dietrich (1992, p. 27) refers
to the fact that the state-of-affairs portrayed in a sentence may either be valid, that is
exist in the time referred to, or modalized, that is, be open to question whether it
exists or not. In the case of all three concepts, a binary opposition exists, between
non-modalized (realis, factual, valid) and modalized (irrealis, nonfactual, open valid-
ity).25 This opposition can be exemplified by the following examples.

(12) Mary is at home now.

(13) Mary may be at home now.

The situation ÔMary is at home nowÕ in (13) is not modally marked. It is at the
same time portrayed as an actualized situation, i.e. realis, as factual, and as valid
for the time referred to. In contrast, in (14) the situation is portrayed as purely within
the realm of thought, as indeterminate with respect to its factuality, or as open with
respect to its actual existence.

25
A slightly different view is presented by Talmy Givón. He distinguishes the four modalities of
presupposition, realis assertion, irrealis assertion, and NEG-assertion (Givón, 1995, p. 114). These can be
seen as four types of factual status. When, like here, a two-way distinction between factual and
undetermined factuality is made, presuppositions and negated propositions (or state-of-affairs), the latter
representing negated factuality, both are regarded as sub-types of factual status.
 H. Narrog / Language Sciences 27 (2005) 165–192 183

The next question that needs to be addressed now is how the labels brought into
play in this section differ, and which are the most appropriate ones. The answer to
the first part of the question is trivial. There is no apparent difference in the meaning
of these labels, wherever an author makes his or her terms more explicit. In principle,
they all boil down to one thing, namely whether a state of affairs is ÔactualÕ or Ôexis-
tentÕ at a specific point or interval of time. Thus, Mithun, using the labels ÔrealisÕ/
ÔirrealisÕ speaks of ‘‘situations as actualized, as having occurred, or actually occur-
ring,’’ Dietrich, whose definition is based on the term ÔvalidityÕ (Geltung), speaks
of ‘‘existence or non-existence in time’’ as ‘‘the semantic commonality of all sen-
tences understood as modalized’’ (Dietrich, 1992, p. 27; emphasis by Dietrich), Papa-
fragou uses the terms ‘‘not present and may never occur’’ (see above), and Chung
and Timberlake speak of an event as being either actual or ‘‘less than completely ac-
tual’’ (Chung and Timberlake, 1985, p. 241). Frawley, in his definition of modality
makes a hierarchical distinction between ÔfactualityÕ as a hypernym and ÔactualityÕ,
ÔvalidityÕ, and ÔbelievabilityÕ as hyponyms (Frawley, 1992, p. 385),26 but this seems
to be an ad hoc distinction, not followed through systematically by the author. Note
that the terms validity as used by Dietrich and Kiefer are also ad hoc creations and
not identical to the term validity in modal logic (cf. Cann, 1993, p. 203ff).
The term ÔirrealisÕ/ÔrealisÕ, however, even if conceived of as having the same exten-
sion as the other terms, is different, in that it is a distinctly technical term. Palmer
sees this as an advantage because terms like ÔfactualÕ and Ônon-factualÕ, or ÔrealÕ
and ÔunrealÕ as familiar, non-technical terms have some connotations that might
be carried over to their use as technical terms (Palmer, 2001, p. 1). However, I sug-
gest that the use of realis/irrealis has the following disadvantages. (1) These are terms
that have been frequently used for a specific type of typically bipartite modal systems
in specific geographical areas, e.g. North America (cf. Mithun, 1995). Defining the
whole of modality in these terms is quasi equivalent to defining the whole of modal-
ity in the terms ÔindicativeÕ and ÔsubjunctiveÕ. Of course, everything can be used in the
way it is defined, but using such narrow terms for much broader categories seems an
unfortunate choice. (2) The terms are primarily identified with epistemic modality.
(3) The term ÔrealÕ is generally narrower than Ôfactual,Õ since in philosophy it is de-
fined as follows: ‘‘what is actually given as a fact27 is real; [the term] reality denotes
the (feature of) existence of things real’’ (Awata and Kozai (1979, p. 70)). In other
words, the term ÔrealÕ denotes only a part of things factual, namely those that are
actually existent, and given in the present.28 Things presented by the speaker as
un-modalized or real/factual, however, may also be things of the past or future.

26
‘‘Modality concerns the factual status of information; it signals the relative actuality, validity, or
believability of the content of an expression’’ (Frawley, 1992, p. 385).
27
‘‘Facts’’, according to Awata and Kozai (1979, p. 90) are ‘‘entia or events that have been found to exist
in time and space.’’
28
Take the example of a concrete expression: When a speaker says ‘‘Tomorrow a meeting is going to be
held’’, the state of affairs expressed is not yet realized, but still the speaker expresses it as a fact. This would
be a case where something is presented as factual which is not real. Also, events of the past are presented as
facts, although they are not actually given in reality, and thus, strictly speaking, not real.
184 H. Narrog / Language Sciences 27 (2005) 165–192

If the definition makes clear that its subject matter is language and not the real
world, this connotation can be avoided, and terms like ÔfactualityÕ become accept-
able.29 The point is that in linguistics we are talking about things depicted as fac-
tual/actual/valid in language, not necessarily things factual in the ‘‘real world’’.
I have established now that each of the terms ÔfactualÕ, ÔactualÕ, ÔrealÕ, and ÔvalidÕ
can be used and have been used in the same way when referring to the definition of
modality, unless, of course, we give them a new technical definition.
I will choose here the term ÔfactualÕ, because it has neither misleading, narrow, or
technical associations like ÔvalidÕ or ÔirrealisÕ, nor does it bear unnecessary connota-
tions from everyday language like ÔrealÕ, and I will define modality as follows:
Modality is a linguistic category referring to the factual status of a state of
affairs. The expression of a state of affairs is modalized if it is marked for being
undetermined with respect to its factual status, i.e. is neither positively nor neg-
atively factual.
Note that while some definitions make explicit reference to the ÔpropositionÕ as the
scope of operation for modality (e.g. Palmer, 2001, p. 1), others avoid this by refer-
ring to more general terms like ÔexpressionÕ (Frawley, 1992), ÔeventÕ (Chung and Tim-
berlake, 1985), or Ôstate of affairsÕ (Sachverhalt; Dietrich, 1992). There are two good
reasons for doing so. First, the term ÔpropositionÕ by itself does not contribute much
to the clarification of the concept because while on the one hand it gives a more exact
and technical impression than the other terms named here, its actual range of appli-
cation in natural language data can be stretched and is actually being stretched just
as much as that of the term modality.30 Second, it is advisable to leave room for dif-
ferent scopes of modality. Different types of modality may operate at different levels.
Decisively, there are types of modality in the sense defined here, particularly non-epi-
stemic modality, that operate not on the proposition, but within the proposition
(depending, of course, on the definition of proposition; cf. Kiefer, 1987). We will
immediately return to this problem in the next section.
Now I want to demonstrate that the definition based on the notion of factuality
achieves what a definition based on the notion of speakersÕ attitudes does not,
namely delimiting modal markers from expressions of other categories. Recall the
examples from the discussion in Section 2.1, which contained grammatical markers
that expressed the speakerÕs evaluation and attitudes, but belonged to different cat-
egories, such as voice, aspect, and referent honorification.

(14) Ni-hiki-dake sime-ta-ga ato-wa nige-rare-ta (Nobi, Ôoka Shôhei)

Two-CLS-LIM catch-PST-CTR other-TOP flee-PASS-PST
‘‘I managed to catch two of them, but the others fled me (=literally Ô[I] was fled
by the othersÕ).’’

29
I agree that with Palmer that the term ÔrealityÕ is less fortunate. Much more than ÔfactualityÕ, ÔactualityÕ,
or ÔvalidityÕ it indeed carries a heavy baggage of connotations from everyday use.
30
Cf., for example, the widely varying definitions in Fillmore (1968, p. 23), Sandhöfer-Sixel (1988, p. 52),
Givón (1995, p. 112), and Löbner (2002, p. 24).
 H. Narrog / Language Sciences 27 (2005) 165–192 185

(15) Son: ‘‘Nee, mada tumetai otya aru?’’

VER still cold tea be
(I) say, is (there) still cold tea (in the refrigerator)?
Mother: ‘‘Gozaimasu-wa’’
be.HON-AFF
‘‘There is.’’
(16) YouÕre telling me you donÕt love me anymore.

Although the speakerÕs attitude is marked on the verb, this fact has no influ-
ence whatsoever on the modal properties of the sentence if defined in terms of
realis/irrealis, factuality, or validity. In each case the sentence in question is realis,
factual, and valid. That is, unlike in the case of the definition by ÔspeakerÕs atti-
tudeÕ, it is possible to demarcate clearly what belongs to this category and what
does not.
In this context, it is also important to note that ÔfactualityÕ is to be understood
strictly at a semantic, and not a pragmatic level. Consider, for example, the following
sentence:

(17) Why must you lie to these kids?

At a pragmatic level, the factuality of the state of affairs (Ôyou are lying to these kidsÕ)
is implied. We might even say that in a rhetorical sentence like this the speaker pre-
supposes the factuality of the state-of-affairs. At a semantic level, however, the fac-
tuality of the state-of-affairs is non-determined, that is, non-factual in our
terminology. It is this semantic level that defines the category.

2.3.1. The problem of assertion

The last problem left with respect to the definition is how to deal with the concept
of ÔassertionÕ. Palmer gives the following reason for its necessity. There is a particular
use of the Spanish subjunctive that cannot be explained as non-factual (or, indeter-
mined factuality, in our terms).

(18) Me alegra que sepas la verdad.

me it pleases that know-2SG-PRES-SUBJ the truth
ÔIÕm glad that you know the truthÕ

The Spanish subjunctive can mark state-of-affairs that the speaker/writer assumes
are already known by the hearer/reader (cf. Lunn, 1995, p. 432–434). The state-of-
affairs marked by the subjunctive in the above example, a presupposition, according
to Palmer is clearly factual, even ‘‘super realis’’ (Palmer, 2001, p. 4). Now, according
to Palmer, (based on the analysis by Lunn), the difference between subjunctive and
indicative in Spanish is that the former is linked to non-assertion while the later is
linked to assertion. Modality, therefore, would have to be defined in terms of
assertion.
186 H. Narrog / Language Sciences 27 (2005) 165–192

This conclusion, however, is questionable. While this particular use of the Spanish
subjunctive is probably correctly analyzed in terms of ÔassertionÕ, this does not nec-
essarily mean the same term is the appropriate concept for the definition of modality
as a cross-linguistic category. Remember that we stated in the beginning that modal-
ity and modal forms need not be co-extensive in every case. Likewise, according to
some analyses, the modal verb can in English, in its ability sense, is non-modal (e.g.
Palmer, 1986, p. 12; Kiefer, 1997, p. 248). This does not mean that we define modal-
ity cross-linguistically in a manner that is co-extensive to the meaning of can.
Besides presupposition, however, Palmer, also recognizes negatives as non-asser-
tive (Palmer, 1979, p. 185f, Palmer, 2001, p. 11f). According to my definition, and
many other definitions, the former would be negatively factual, that is not undeter-
mined with respect to factuality, and therefore not modal.31
The crucial problem of a definition of modality additionally through the concept
of ÔassertionÕ from my point of view is that by introducing a term on the speech act
level (cf. Lyons, 1995, p. 251; Grewendorf and Zaefferer, 1991, p. 270 as the decisive
factor in the definition of the category, a second level is introduced on top of the
semantic level of non-factuality (or realis/irrealis). This by itself is of course problem-
atic. If the formal categories identified as modal by the pragmatic definition coincide
with those identified, the additional definition is superfluous. If they do not coincide,
I suggest that the latter is the case, that is, non-factuality and assertion do not coin-
cide. Consider the following example featuring a modal expression operating within
the proposition.

(19) She has the time and the energy. She can do it.

In the second sentence, the state-of-affairs Ôshe does itÕ is marked as non-factual by
the modal verb can. However, the whole sentence ‘‘she can do it’’ is being asserted. I
thus suggest that the logical structure, in essence, corresponds to (20) but not to (21)
(PROP standing for proposition, MOD for modality and ASS for assertion):

(20) [ASS [MOD can [PROP she do it]]]

(21) [MOD/ASS can [PROP she do it]]

Thus, while the modal can operate on the proposition Ôshe do itÕ, and renders it as
non-factual, the assertion takes the enlarged proposition Ôshe can do itÕ as a whole
into its scope. The fact that assertion does not operate on the same level as can,
but includes it, can also be demonstrated by the fact that the assertion moves further
outside if another grammatical element, such as negation, is added:

(22) She canÕt do it.

(22) for instance, must have the structure of (23) and not of (24):

31
Of course, it has to be questioned whether negated propositions/state-of-affairs are really non-asserted.
Givón, for instance, who also recognizes special factual status to presupposition and negatives, as I argue
below, correctly, regards the latter as asserted (Givón, 1995, p. 114).
 H. Narrog / Language Sciences 27 (2005) 165–192 187

(23) [ASS [NEG not [MOD can [PROP she do it]]

(24) [NEG not [MOD/ASS can [PROP she do it]]

That is, the speaker asserts a negative fact ((23)) but cannot negate his or her own
assertion ((24)). In this way, the assertion will always move to the outermost layer of
the sentence (or clause). It follows that we have modal expressions like can that have
nothing to do with assertion or non-assertion, and are nevertheless modal, if modal-
ity is defined in terms of factuality. A semantic definition of modality based on terms
such as ÔfactualityÕ, ÔvalidityÕ, or Ôrealis/irrealisÕ, encompasses modality at the propo-
sitional level while a definition based on the term ÔassertionÕ would not.32 In conclu-
sion, the term ÔassertionÕ should be excluded from the definition of modality, if this
definition is already based on other terms like factuality and realis/irrealis which
refer to concepts that operate at a different level.

2.3.2. The scope of modality

In this last section I will give a brief outline of the scope of modality that results
from the definition presented here, in contrast to other definitions. Modality will
have different semantic sub-categories of which I distinguish epistemic modality, evi-
dentiality and agent-oriented modality, the latter including traditional deontic and
dynamic modality. The only category that perhaps needs a justification is evidential-
ity, a category which lacked general recognition as a modal category until recently,
perhaps due to the fact that it does not have salient formal expression in English.
However, it unambiguously falls under the scope of modality defined by factuality.
Consider the following example with the evidential modal sollen (Ôis said toÕ, Ôalleg-
edlyÕ) from German:

(25) Er soll bei einem Unfall umgekommen sein.

ÔHe allegedly died in an accidentÕ

32
This is not at all to deny the widely recognized fact that modality can operate at two or more layers of
the proposition (cf. Cinque, 1999 for a formal grammar and Arista, 2003 for DikÕs Functional Grammar),
a fact that would not change if modality is defined by factuality. We can clearly see modality operating on
two or more layers when there are two modal expressions in one sentence, like in the following example
and the logical structure given to it:

(i) We certainly should acknowledge his efforts.

(ii) [MOD2 certainly[MOD1 should[PROP we acknowledge his efforts]]]

An anonymous reviewer has kindly made me aware of the fact that depending on the way proposition is
defined, the layer of MOD1 can be viewed as a larger proposition by itself, and the layer of MOD2 as a yet
larger proposition. The point I want to make here, however, is that a factuality-based definition is not
bound to a specific layer of proposition, and thus has no problem with multiple layers of proposition and
modality. In contrast, an assertion-based definition will only pick out the speech-act level outmost layer
(unless the concept of assertion itself would be drastically changed). Therefore, inner-propositional types
of modality, such as represented by non-epistemic can, will fall through the grid if the definition is based on
the concept of assertion.
188 H. Narrog / Language Sciences 27 (2005) 165–192

Table 2
The scope of the definition of modality put forward in this paper
Object of definition Semantically defined grammatical category
Languages of concern Cross-linguistic applicability
Included in modality Epistemic modality, evidentiality, agent-oriented
(deontic, dynamic modality)
Categories explicitly excluded Tense, aspect, voice, negation, honorifics, politeness,
illocutionary force (discourse markers), information structure
(topic and focus)a
a
I suggest sentence type should also be dealt with as a distinct category although the inclusion of certain
sentence types, such as yes/no questions would not contradict my definition.

The factuality of the proposition (Ôdied in an accidentÕ) is undetermined, just as it

is in the case of epistemic modality. The difference is that while in epistemic modality
the proposition is undetermined with respect to its factuality relative to the world of
knowledge and beliefs of the speaker, with evidentiality it is undetermined relative to
sources of information other than the speaker.
I will present here the scope of modality as defined in my approach in contrast
with those presented in Table 1 (the full contents of Table 1 will not be repeated
here). See Table 2.
The decisive point here would be that this definition allows us to confine the study
of modality to a small set of sub-categories which largely correspond to those which
actually have been studied under this label, under exclusion of other categories like
tense, aspect, and voice which might well express the speakerÕs attitude, but which
otherwise obviously form a semantic and grammatical domain of their own.
With respect to its formal expression, the salient grammatical means of modal
marking in English and German would be (1) modal auxiliaries (e. g. can, must,
may for English and their cognates in German), (2) semi-modals (e.g. have to, be
to, seem to, want to for English and brauchen, scheinen, sein zu, haben zu for Ger-
man), (3) adverbs (e.g. perhaps, possibly for English and wohl, vielleicht for German),
and (4) verbal mood, which is highly reduced in Modern English but still very much
alive in German, at least in the written language. The formal marking in Japanese,
on the other hand, is strikingly different. Due to its agglutinative character, we do
not find a class of auxiliary verbs, but rather suffixes on verbs (e.g. -daroo for sup-
position, -beki for weak obligation) and periphrastic constructions (e.g. -ka-mo sire-
nai for possibility and -nakereba naranai for obligation). On the other hand, like
English and German, Japanese does have modal adverbs (e.g. tabun ÔprobablyÕ),
and verbal mood (although highly reduced as well).
It thus becomes clear that modality will have different encoding across languages,
and even within one language it will have exponents from different formal classes.
But this is merely the same as with any other semantically defined grammatical cat-
egory such as tense, aspect, negation, etc. Negation, similarly, can be expressed
through adverbs (never), determiners (no), pronouns (nobody), and clitics (nÕt), that
is at least three form classes.33

33
I am indebted for this point to an anonymous reviewer of Language Sciences.
 H. Narrog / Language Sciences 27 (2005) 165–192 189

3. Conclusion

In this paper I have argued that the most reasonable cross-linguistically oriented
definition of modality is one through the concept of factuality.
Although it is certainly true that there cannot be a single correct definition of a
grammatical category like modality, it is possible to demonstrate that some defini-
tions fulfill their functions better than others. In particular, the definition of modality
through the concept of the expression of the ÔspeakerÕs attitude,Õ as I have argued,
does not fulfill the basic function of a definition reliably, as it does not serve to dis-
tinguish formal categories that belong to this category from those that do not.
Especially in conversation, the speakerÕs attitude is expressed pervasively through-
out the sentence, starting with the choice of lexical items, but particularly through a
wide array of grammatical categories, not to mention other non-segmental features
like prosody and intonation and paralinguistic means. Among the three languages
mainly referred to in this paper, Japanese has particularly rich grammatical resources
for the expression of subjectivity. Speakers draw on modal markers, that is, markers
of non-factuality, as one among many resources to express their attitude, just as they
draw on markers of aspect, voice, referent honorifics, politeness and other categories.
There are many studies that show how markers of modal, aspectual, honorific and
other categories each by themselves express subjectivity. However, a study based
on the same (preferably conversational) data that would compare markers of differ-
ent categories and show which category more typically or frequently serves the
expression of the speakerÕs attitude to my knowledge has still to be done.
The negative definition of modality as everything non-propositional was origi-
nally not intended to identify a grammatical category, but as a means to very roughly
identify a constituent of the sentence. Reinterpreting this concept as a grammatical
category has led, as I have argued in Section 2.2 to a fair deal of the confusion that
has made modality the ‘‘dustbin’’ of grammatical categories. In contrast, the concept
of factuality is reasonably suited to delimit a coherent category with those notions at
its core that have traditionally been identified with the term ÔmodalityÕ in grammar,
and which most scholars purport to be defining.

References

Abraham, W., 21988. Terminologie zur neueren Linguistik, 2. Auflage. Niemeyer Tübingen.
Arista, J.M., 2003. Terms, clauses and constructions in functional grammar. Language Sciences 25, 515–
530.
Awata, K., Kozai, Y., 1979. Iwanami tetsugaku shôjiten [IwanamiÕs Small Dictionary of Philosophy]
Iwanami, Tôkyô.
Bally, Ch., 1965 (1932). Linguistique Générale et Linguistique Française. Fourth ed. Francke, Bern.
Brinton, L., 1996. Pragmatic Markers in English. Mouton de Gruyter, Berlin.
Brugmann, K., 1970 (1904). Kurze Vergleichende Grammatik der Indogermanischen Sprachen. Walter de
Gruyter, Berlin.
Bussmann, H., 1996. Routledge Dictionary of Language and Linguistics. Routledge, London.
Bybee, J., Perkins, R., Pagliuca, W., 1994. The Evolution of Grammar. The University of Chicago Press,
Chicago.
190 H. Narrog / Language Sciences 27 (2005) 165–192

Calbert, J.P., 1975. Towards the semantics of modality. In: Calbert, J.P., Vater, H., Aspekte der
Modalität, Gunter Narr, Tübingen, pp. 1–70.
Cann, R., 1993. Formal Semantics: an Introduction. Cambridge University Press, Cambridge.
Chung, S., Timberlake, A., 1985. Tense, aspect and mood. In: Shopen, T. (Ed.), Language Typology and
Syntactic Description, vol. 3. Cambridge, pp. 202–225.
Cinque, G., 1999. Adverbs and Functional Heads. Oxford University Press, Oxford.
Coates, J., 1983. The Semantics of the Modal Auxiliaries. Croom Helm, London.
Declerck, R., 1992. Some theoretical issues in the description of tense and mood. Cognitive Linguistics 3/2,
187–218.
Dietrich, R., 1992. Modalität im Deutschen. Westdeutscher Verlag, Opladen.
Feyrer, C., 1998. Modalität im Kontrast: Ein Beitrag zur übersetzungsorientierten Modalpartikelfors-
chung anhand des Deutschen und des Französischen. Peter Lang Verlag, Frankfurt.
Fillmore, Ch., 1968. The case for case. In: Bach, E., Harms, R.T. (Eds.), Universals in Linguistic Theory.
Holt, Rinehart and Winston Inc., New York, pp. 1–88.
Frawley, W., 1992. Linguistic Semantics. Lawrence Erlbaum Associates, Hillsdale.
Fujii, Y., 1992. shite shimau no imi [The meaning of shite shimau]. In: Gengogaku kenkyûkai (Ed.),
Kotoba no kagaku 5 (Language Sciences 5). Mugi shobô, Tôkyô, pp. 17–40.
Gerstenkorn, A., 1976. Das Modal-System im heutigen Deutsch (Münchner Germanistische Beiträge
Band 16) Wilhelm Fink Verlag, München.
Givón, T., 1995. Functionalism and Grammar. John Benjamins, Amsterdam.
Grewendorf, G., Zaefferer, D., 1991. Semantic foundations of speech acts. In: von Stechow, A.,
Wunderlich, D., Semantik—Semantics (Handbücher für Sprach- und Kommunikationswissenschaft,
vol. 6). de Gruyter, Berlin, pp. 270–285.
Harweg, R., 1994. Modus und Modalität. In: Jachnow, H.H., Modalität und Modus. Allgemeine Fragen
und Realisierung im Slavischen. Harrassowitz, Wiesbaden, pp. 1–23.
Helbig, G., Buscha, J., 1996. Deutsche Grammatik. Ein Handbuch für den Ausländerunterricht,
seventeenth ed. Langenscheidt, Leipzig.
Jespersen, O., 1992 (1924). The Philosophy of Grammar. University of Chicago Press, Chicago.
John, A.P., 1987. The Semantics of Modality—a Study Based on Malayalam and English. Dravidian
Linguistics Association, Trivandrum.
Kiefer, F., 1987. On defining modality. Folia Linguistica 21/1, 67–94.
Kiefer, F., 1997. Presidential address—modality and pragmatics. Folia Linguistica 31/3–4, 241–253.
King, A.R., 1997. Review of Bybee. In: Fleischman, S. (Ed.), Modality in grammar and discourse
(Typological Studies in Language 32). Sprachtypologie und Universalienforschung 50/2, 362–364.
Koyano, T., 1997. Giji modariti no fukushi ni tsuite—marude o rei to shite—[About the adverbs of
pseudo-modality—taking marude as an example]. In: Satô, K., Kindaigo no kenkyû [Studies of
Modern Japanese], Meiji shoin, Tôkyô, pp. 287–318.
Kristophson, J., 1994. Zur Geschichte der Modalitätsforschung in Russland. In: Jachnow, H.H. et al.
(Eds.), Modalität und Modus. Allgemeine Fragen und Realisierung im Slavischen, Harrassowitz,
Wiesbaden, pp. 37–51.
Krivonosov, A.T. 1997. Deutsche Modalpartikeln im System der unflektierten Wortklassen. In: Weydt,
H., Aspekte der Modalpartikeln, Niemeyer, Tübingen, pp. 176–216.
Kudô, H., 1989. Gendai nihongo no bun no johôsei ¿ joshô [Modality of the Modern Japanese sentence ¿
an introduction]. Tôkyô Gaikokugo Daigaku Ronshû 39, 13–33.
Kudô, H., 2000. Fukushi to bun no chinjutsuteki na taipu [Adverbs and predicative sentence types]. In:
Moriyama, T., Nitta, Y., Kudô, H. (Eds.), Modariti [Modality] (Nihongo no bunpô 3). Iwanami,
Tôkyô, pp. 164–234.
Kuno, S., 1983. Shin Nihon bunpô kenkyû [New Research on Japanese Grammar], Taishûkan shoten,
Tôkyô.
Lindemann, B., 1997. Zur objektiven Verwendung von müssen und sollen und ihren norwegischen
Entsprechungen. Germanistische Linguistik 136, 135–150.
Lunn, P.V., 1995. The evaluative function of the Spanish subjunctive. In: Bybee, J., Fleischman, S. (Eds.),
Modality in Grammar and Discourse. John Benjamins, Amsterdam, pp. 429–449.
 H. Narrog / Language Sciences 27 (2005) 165–192 191

Lyons, J., 1968. Introduction to Theoretical Linguistics. Cambridge University Press, Cambridge.
Lyons, J., 1977. Semantics, vol. 2. Cambridge University Press, Cambridge.
Lyons, J., 1995. Linguistic Semantics—an Introduction. Cambridge University Press, Cambridge.
Masuoka, T., 1987. Meidai no bunpô [The grammar of proposition], Kuroshio, Tôkyô.
Masuoka, T., 1991. Modariti no bunpô [The grammar of proposition], Kuroshio, Tôkyô.
Masuoka, T., 2000. Meidai to modariti no kyôkai o motomete [Seeking for the boundary between
proposition and modality]. In: Masuoka, T. Nihongo bunpô no shosô [Various aspects of japanese
grammar]. Kuroshio, Tôkyô, pp. 87–98.
Maynard, S.K., 1993. Discourse modality. Subjectivity, Emotion and Voice in the Japanese Language.
Benjamins, Amsterdam.
McCawley, J.D., 1993. Everything That Linguists Have Always Wanted to Know about Logic, second ed.
University of Chicago Press, Chicago.
Mithun, M., 1995. On the relativity of irreality. In: Bybee, J., Fleischman, S. (Eds.), Modality in Grammar
and Discourse. John Benjamins, Amsterdam, pp. 367–388.
Mithun, M., 1999. The Languages of Native North America. Cambridge University Press, Cambridge.
Miyazaki, K., 2002. Modariti no gainen [The concept of modality]. In: Miyazaki, K., et al. (Eds.) Modariti
[Modality] (Shin Nihongo bunpô sensho 4), Kuroshio, Tôkyô, pp. 1–15.
Nakau, M., 1994. Ninchi imiron no genri [Principles of cognitive semantics]. Taishûkan, Tôkyô.
Narrog, H., 1998. Nihongo dôshi no katsuyô taikei [The inflection system of Japanese verbs]. Nihongo
kagaku 4 [Japanese Linguistics 4], 7–30.
Narrog, H., 2002. Imironteki kategorii to shite no modariti [Modality as a semantic category] . In: Ôhori,
T. (Ed.), Ninchi gengogaku II: kategoriika [Cognitive linguistics II: categorization] (Shiriizu gengo
kagaku 3), Tôkyô University Press, Tôkyô, pp. 217–251.
Nishikawa, M., 1995. Nihongo no hyôkateki jodôshi—te kureru, te simau, yagaru—[The valuative
auxiliaries of Japanese—te kureru, te simau, yagaru–]. In: Paper Presented at the 110th Meeting of the
Linguistic Society of Japan, 1995/6/11.
Nitta, H., 1999. Sprecherstandpunkt und Sprechereinstellung—yaru/kureru im Japanischen und deutsche
Dative. In: Nitta, H. et al. (Eds.), Kontrastive Studien zur Beschreibung des Japanischen und des
Deutschen. iudicium, München, pp. 145–157.
Nitta, Y., 1989. Bun no kôzô [The structure of the sentence]. In: Kitahara, Y. (Ed.), Kôza nihongo to
nihongo kyôiku, vol. 4. Nihongo no bunpô, buntai 1 [Course in Japanese and Japanese Language
Education, vol. 4. Japanese Grammar and Style]. Meiji shoin, Tôkyô, pp. 25–52.
Nitta, Y., 1991. Nihongo no modariti to ninshô [Japanese Modality and Person]. Hitsuji shobô, Tôkyô.
Nitta, Y., 1997. Nihongo bunpô kenkyû josetsu [Preliminaries to the study of Japanese], Kuroshio, Tôkyô.
Nitta, Y., 2000. Ninshiki no modariti to sono shûhen [Epistemic modality and its periphery]. In:
Moriyama, T., Nitta, Y., Kudô, H. (Eds.), Modariti [Modality] (Nihongo no bunpô 3). Iwanami,
Tôkyô, pp. 81–159.
Noda, H., 1989. Shinsei modariti o motanai bun. In: Nitta, Y. Masuoka, T. (Eds.), Nihongo no modariti
[Japanese Modality]. Kuroshio, Tôkyô, pp. 131–157.
Noda, H., 2002. Shûjoshi no kinô [The functions of the final particles]. In: Miyazaki, K. (Ed.), Modariti
[Modality]. Kuroshio, Tôkyô, pp. 261–288.
Nomura, T., 2003. Modariti keishiki no bunrui [On the classification of Japanese modal forms].
Kokugogaku [National Language Studies] 54/1, 17–31.
Noreen, A., 1923. Einführung in die wissenschaftliche Betrachtung der Sprache. Beiträge zur Methode und
Terminologie der Grammatik. (Translation of parts of the 9 volume work ‘‘Vårt Språk’’ by the same
author, published 1903–1919). Niemeyer, Saale.
Öhlschläger, G., 1989. Zur Syntax und Semantik der Modalverben des Deutschen. Niemeyer, Tübingen.
Okushi, Y., 1997. Patterns of Honorific Use in the Everyday Speech of Four Japanese Women. Ph.D.
Dissertation, University of Pennsylvania.
Onoe, K., 1997. Kokugogaku to ninchi gengogaku no taiwa II—modariti wo megutte [Dialogue between
national language sciences and cognitive linguistics II—about modality]. Gekkan gengo [Monthly
Language] 26/13, 70–83.
192 H. Narrog / Language Sciences 27 (2005) 165–192

Onoe, K., Tsuboi, E., 1999. Bun no kôzô to shukanteki imi [The structure of the sentence and subjective
meaning]. Gekkan gengo[Monthly Language] 28 (1), 95–105.
Palmer, F.R., 1979. Non-assertion and modality. In: Allerton, D.J., Carney, E., Holdcroft, D. (Eds.),
Function and Context in Linguistic Analysis—a Festschrift for William Haas. Cambridge University
Press, Cambridge, pp. 185–195.
Palmer, F.R., 1986. Mood and Modality. Cambridge University Press, Cambridge.
Palmer, F.R., 1998. Mood and modality: basic principles. In: Brown, K., Miller, J. (Eds.), Concise
Encyclopedia of Grammatical Categories. Elsevier, Oxford, pp. 229–235.
Palmer, F.R., 22001. Mood and Modality, second edition. Cambridge University Press, Cambridge.
Papafragou, A., 2000. Modality: Issues in the Semantics–pragmatics Interface. Elsevier, Oxford.
Sandhöfer-Sixel, J., 1988. Modalität und gesprochene Sprache—Ausdrucksformen subjektiver Bewertung
in einem lokalen Substandard des Westmitteldeutschen (Mainzer Studien zur Sprach- und Volksfors-
chung 15). Franz Steiner Verlag, Stuttgart.
Schiffrin, D., 1987. Discourse Markers. Cambridge University Press, Cambridge.
Shinzato, R., 2003. Experiencing self versus observing self: the semantics of stative extensions in Japanese.
Language Sciences 25 (2), 211–238.
Tsujimura, N., 1996. An Introduction to Japanese Linguistics. Blackwell, Oxford.
van der Auwera, J., Plungian, V.A., 1998. ModalityÕs semantic map. Linguistic Typology 2 (1), 125–139.
VanValin Jr., R.D., LaPolla, R.J., 1997. Syntax. Structure, Meaning and Function. Cambridge University
Press, Cambridge.
Yoshida, E., 1996. A discourse study of passives in Japanese Conversation. Ph.D. Dissertation, University
of California, Los Angeles.