<LINK "dix-n*">

<TARGET "dix" DOCINFO AUTHOR "R. M. W. Dixon"TITLE "Demonstratives"SUBJECT "SL, Volume 27:1"KEYWORDS ""SIZE HEIGHT "220"WIDTH "150"VOFFSET "4">

Demonstratives
A cross-linguistic typology*

R. M. W. Dixon
Research Centre for Linguistic Typology
La Trobe University

A typology is presented of parameters of variation associated with three

main types of demonstratives: nominal (corresponding to demonstrative
pronouns and demonstrative adjectives in traditional terminology), local
adverbial and verbal. The paper surveys their basic characteristics; forms
(including relationship to articles and interrogatives); deictic, syntactic,
anaphoric and other functions; and types of reference — spatial, height and
stance, visibility, etc. There is a case study of demonstratives in Dyirbal, and
an array of questions which can usefully be pursued when investigating
demonstratives in a given language.

1. Introduction

It appears that all languages have one or more demonstratives. In this paper I
discuss the parameters in terms of which they vary — type, form, function and
reference — and examine their relationship to 3rd person pronouns, articles,
interrogatives, and other similar forms. The general discussion is followed by a
case study of demonstratives and related items in Dyirbal (§7). In §9 there is a
list of questions which should be addressed when investigating the properties of
demonstratives in a language.1
A demonstrative is here defined as a grammatical word (or, occasionally, a
clitic or affix) which can have pointing (or deictic) reference; for example This
is my favourite chair (pointing at an object) or Put it there! (pointing at a place).
It could be argued that 1st and 2nd person pronouns have, by their very nature,
implicit pointing reference. In view of this, the definition of demonstrative can
be modified to be “any item, other than 1st and 2nd person pronouns, which

Studies in Language 27:1 (2003), 61–112.

issn 0378–4177 / e-issn 1569–9978
© John Benjamins Publishing Company
<LINK "dix-r39">

62 R. M. W. Dixon

can have pointing (or deictic) reference”. Demonstratives always comprise a

closed grammatical system.
There are three well-attested types of demonstrative:
a. Nominal — can occur in an NP with a noun or pronoun (e.g. “[this stone] is
hot”) or, in most languages, can make up a complete NP (e.g. “[this] is hot”).
b. Local adverbial — occur either alone (e.g. “put it here”) or with a noun
taking local marking (e.g. “put it (on the table) there”)
c. Verbal “do it like this”, with an accompanying mimicking action — can
occur as the only verb in a predicate, or together with a lexical verb.
Types (a) and (b) are found in every language, with a single known exception
(Ju|’hoan — see §3.1). Type (c) is relatively rare; examples will be given from
languages on which I have done fieldwork — Boumaa Fijian (§3.3) and
Dyirbal (§7.3).
In some languages there is a series of manner adverbial demonstratives,
such as “(do) like this, (do) in this way/manner”. These are almost always
morphologically derived from nominal demonstratives, and generally function
as non-inflecting modifiers to verbs. Interestingly, in most of the languages in
which this occurs, local adverbial demonstratives are also morphologically
derived from nominal demonstratives; see §4. Individual languages can have
further specifications. For instance, in Telugu (Dravidian family, south India),
there is a series of plain manner adverbial demonstratives (“in this/that man-
ner”) and also a series of causal manner adverbial demonstratives (“for this/that
reason”); these involve different suffixes to the basic nominal demonstrative
roots (Krishnamurti 2001).
In some languages, nominal and/or local adverbial demonstratives have a
secondary temporal sense; for example, “this” or “here” may also relate to
“now”, and “that” or “there” to “then” (which may refer to past or to future,
depending on the language). In just a few languages there are separate temporal
forms. See §6.1.
There may well be further types of demonstratives in some languages. Note
that, in English, an angler may boast of the size of a catch either by holding his
hands wide apart and saying It was this big or It was so big, or else by holding up
a number of fingers are saying It was this many or It was so many. It appears that
so has deictic reference as an alternative to this in contexts like this. I do not
here enter into discussion of rather peripheral demonstrative items like so.
Terminology on this topic is confused and often misleading. It requires
preliminary discussion, in §2.
<LINK "dix-r73">
"dix-r7">
"dix-r43">

Demonstratives 63

2. Terminology

i. Deictic. This term is used by modern-day linguists in two entirely different

senses. The first — following Bühler (1934), and Lyons (1968, 1977) — is as a
cover term for personal pronouns, demonstratives, and time shifters such as
“now” and “tomorrow”. The second sense is for the pointing function of
demonstratives, when accompanied by a gesture indicating an object or place,
or mimicking an action (for example, Zandvoort 1975: 147). I shall here use
deictic just in the second sense.
ii. Demonstrative pronoun and demonstrative adjective. In most languages
nominal demonstratives can (1) make up a complete NP; and (2) occur in an
NP with a noun or a personal pronoun. Sometimes different — but related —
forms are used for (1) and (2) but generally the same forms are employed.
There is a tradition of using the term “demonstrative pronoun” for (1); in fact,
a nominal demonstrative is in most languages nothing like a personal pronoun,
in either form or function. And there is a tradition of using the term “demon-
strative adjective” for (2); in fact demonstrative nominals are almost always
totally different from adjectives, both in formal categories and in function.2 A
more satisfactory label sometimes used for (2) is “demonstrative determiner”.
As the discussion below shows, it is most appropriate to sidestep the tradition
and just use the label “nominal demonstrative”.
iii. Anaphora and cataphora. These terms are used for reference to something
earlier or later, respectively, in the discourse. Some linguists talk of “forwards
anaphora” in place of “cataphora” but it seems clearest to retain the term
“cataphora”.
“Anaphora” and “cataphora” are used to describe two quite different kinds
of phenomena. Firstly, a pronoun (or, sometimes, a deictic) can be used as a
substitute for an NP, as in the following examples from English:
(1) John failed the exam because he [anaphora] hadn’t studied.
(2) Because he [cataphora] hadn’t studied John failed the exam.

Secondly, the pronoun it, or a demonstrative, can be used to refer to something

that has been (anaphora) or will be (cataphora) described in the discourse. The
pronoun or demonstrative may refer just to a clause — such as the complement
clause in (4) — or to a stretch of discourse that may be several pages long — as
in (3) and (5). Again illustrating from English, and underlining both the
anaphoric/cataphoric element and the section of discourse that is referred to:
<LINK "dix-r40">
"dix-r43">

64 R. M. W. Dixon

(3) John hadn’t studied and failed the exam and Mary considered it/that/this
[anaphora] a terrible shame.
(4) It [cataphora] annoyed Mary that John hadn’t studied and failed the exam.
(5) These [cataphora] are the choices available: either study and pass the exam,
or become a politician.

John could be substituted for he in (1) and (2), producing an acceptable — if

clumsy — sentence. In contrast, the long underlined portions could not be
substituted for it or that or this or these in (3)–(5).
Different labels are needed for (1)–(2) and for (3)–(5). Lyons (1977: 667)
restricts “anaphora” to (1)–(2) and employs “textual deixis” for (3)–(5);
Levinson (1983: 83) adjusts the latter label to “discourse deixis”. As pointed out
above, use of the term “deixis” is so confused that it seems wisest not to use it
here. For lack of any better labels, I shall use:
– “substitution anaphora/cataphora” for examples such as (1)–(2); and
– “textual anaphora/cataphora” for examples such as (3)–(5).
Much modern work in linguistics is burdened by the yoke of traditional grammar
and eurocentrism; this applies particularly to work on demonstratives and related
items. I have heard it asserted that all languages have demonstratives “this” and
“that”, and that in all languages demonstratives have anaphoric as well as deictic
function. It will be shown below that neither of these a priori assertions is correct.
The first and most important step when describing a new language is to find
criteria for recognising demonstratives, and for distinguishing them from other
items with similar properties. There is a tendency to use the label “demonstra-
tive” for anything which could be translated by this and that in English. For an
analysis to be valid it must be based on internal criteria in the language under
study, rather than on the analysis of translation equivalents in another language.
One difficulty is the lack of an adequate set of accepted labels for the varied
classes and categories found across the languages of the world. There is a
tendency to use a common label to name something for which it is not entirely
appropriate — simply because there is not a fully appropriate label — and this
can then lead to confusion. There is a need for new labels to be coined, but this
should be done judiciously and parsimoniously.
In §3 there is discussion of the attested types of demonstratives, and then §4
deals with formal relationships between different types of demonstratives, and
between demonstratives and related items. The functions of demonstratives are
discussed in §5, and their referential possibilities in §6.
<LINK "dix-r22">
"dix-r65">
"dix-r58">

Demonstratives 65

3. Types

As mentioned above, demonstratives can be divided into three main types,

depending on whether they occur in an NP (nominal), as a local adverb to a
clause (adverbial), or in a predicate (verbal). We can discuss the types in turn.

3.1 Nominal demonstratives

With a single exception (Ju|’hoan — see the end of this sub-section), every
language that I know of has at least one nominal demonstrative. A nominal
demonstrative may always appear in an NP with a noun, and in some languages
with a pronoun. In most languages it may also make up a complete NP (the
“demonstrative pronoun” function) but this is sometimes not permitted. In
Ainu, for instance, nominal demonstratives may not be used without an
accompanying noun (Tamura 2000: 261–2; Refsing 1986: 97, 154–5); see also
(34) in §6.1 below.
In many languages where a nominal demonstrative may be a full NP, this
usage is relatively infrequent. In the Australian language Yidinj, for example,
less than 10% of the textual occurrences of the demonstratives yi]u “this” and
]u]u “that” are as a complete NP; about 80% of the remainder involve the
demonstrative occurring with a noun, as in (6), and about 20% with a pronoun,
as in (7) (examples are from texts given in Dixon 1977a: 537).
(6) ]anjdiA wanji:n]a-l [yi]u dungu]O
1pl do.what-non.past [this head
“What shall we do with this head? (which we have cut off from the dead
man, after burying the corpse, and which his returning spirit has smelled
the stench emanating from).”
(7) ]ayuA [njundu:banj yi]u]O badja-r-ala
1sg [2pl this leave-non.past-now
“I’m now leaving these-you.”

Note that it is impossible properly to render (7) by English translation; it is

literally “I’m now leaving this you-all” (note that number is not marked on
demonstratives in Yidinj).
Yidinj, which is a dependent-marking language, lacks 3rd person pronouns
per se, demonstratives taking some of the function of 3rd person forms in other
languages. We also find nominal demonstratives co-occurring with pronouns
in head-marking languages such as Jarawara (from the small Arawá family of
66 R. M. W. Dixon

southern Amazonia). Here the predicate commences with an object pronoun

(in a transitive clause) and then an obligatory subject pronoun. Unlike Yidinj,
Jarawara has a 3nsg pronoun, mee, but 3sg is always zero. The nominal demon-
strative haaha “this” can occur in a pre-predicate NP, in association with a
pronoun of any person.
Example (8) comes from a story about recent ancestors; the storyteller then
says (I use “[…]” to enclose an NP and “{…}” to enclose a predicate):
(8) [haaha]S {ee joto-hemete ee
[this:f {1nsg.inc follow-far.past:non.eyewitness:f 1nsg.inc
ama-ke}
ext-dec:f
“We these followed on (after the ancestors) (i.e. we are these who
followed on).”
In another text the narrator states that someone went to a place,
(9) [haaha]S {mee wina-haa}
[this:f {3nsg live-subordinate:f
“where these they lived (i.e. where this group lived).”
It can be seen that the demonstrative haaha may co-occur with the 1st inclusive
non-singular pronoun, ee, in (8) and with the 3rd non-singular pronoun, mee,
in (9). (Note that the demonstrative does not mark number; and that feminine
is the unmarked term in the gender system, all pronouns taking feminine
agreement.)
In fact the Jarawara 3nsg pronoun mee is a recent grammaticalisation from
the noun *madi “person”. To refer just to “people” in the modern language one
simply uses the pronoun mee; that is, (9) would be the rendering of “where
these people lived”.
Other languages, such as English, do not allow a demonstrative to co-occur
with a pronoun, just with a noun; and this is the most frequent context of use
for nominal demonstratives. In English this and that can only be used as a full
NP in limited circumstances. An NP this or that which has animate reference is
restricted to copula subject in an identity clause, e.g. That’s my wife. In other
contexts, a demonstrative requires a following noun, such as the dummy item
one, e.g. That one [animate] is beautiful, I like that one [animate]. Indeed,
although an NP this or that with inanimate reference may have any syntactic
function, a following one generally makes the sentence sound more felicitous.
(I’ll have that one is likely often to be preferred to I’ll have that when, say,
pointing to a chocolate bar.)
<LINK "dix-r27">
"dix-r22">

Demonstratives 67

Since in the majority of languages nominal demonstratives most often

occur with a noun (or pronoun) and only occasionally by themselves, one needs
to decide on what their status is when making up a complete NP. There are two
alternatives. Illustrating from English:
– Either (i) we could say that this in a sentence such as I like this is a headless
NP, with this as a modifier (as it is in this cake or this one) and the head
noun ellipsed.
– Or (ii) we could say that this is head of the object NP in I like this, which
would mean that it has a different function here from that which it has in
I like this cake/one, where it is modifier within the object NP.
This question must be answered separately for each language, on language-
internal criteria. And it may require different answers relating to the varying
functions of demonstratives. In English, for instance, in instances where the
nominal demonstratives are used for textual anaphora — as in (3) and (5) —
they must be regarded as heads of NPs, alternative (ii). But when they are used
for substitution anaphora, as in (10), or with deictic function, as in (11), a case
could be made out for the headless NP analysis, alternative (i).
(10) I read his first novel and that (one) was boring too
(11) Which cake would you like? I’ll have this (one) [pointing at it]

There are a number of kinds of link between nominal demonstratives and 3rd
person pronouns, and between nominal demonstratives and articles.
All languages have 1st and 2nd person pronouns but some lack a 3rd person
in the system; as already mentioned, nominal demonstratives may fill part of
this functional role. For example, nominal demonstratives may be more likely to
have anaphoric function in a language with no 3rd person pronouns than in a
language which does have these items. Within the Australian linguistic area, there
are a number of examples of a certain form functioning as a demonstrative in
one language and as 3rd person pronoun in another (Dixon 2002: 306, 335–6).3
There can also be obligatory (or almost obligatory) co-occurrence of
nominal demonstratives with 3rd person pronouns. Gragg (1976: 178–9)
describes how in the Wellegga dialect of Oromo (a Cushitic language spoken in
Ethiopia), when a nominal demonstrative does not have an accompanying
noun it is generally used with a 3rd person pronoun (perhaps functioning
something like one in English).
Turning now to articles, there can be a synchronic or diachronic connection
with nominal demonstratives. In standard German, the forms die (f and pl)/der
<LINK "dix-r50">
"dix-r64">
"dix-r55">

68 R. M. W. Dixon

(m)/das (n) have definite article function when unstressed and demonstrative
function when stressed. Old English had two nominal demonstratives, which
showed two numbers, three genders and five cases; from these have developed
the modern definite article the, and the nominal demonstratives this/these and
that/those (Sweet 1898: 112–5). The definite article in modern French has
developed out of the nominal demonstrative ille “that” in Latin; and the
nominal demonstratives in modern French come from the Latin nominal
demonstrative hic “this” with strengthening from deictic particle ecce (Pope
1934: 322–7).4
We also find languages where a nominal demonstrative can co-occur with
the definite article. Newman (2000: 143–9) recognises a definite article in
Hausa; this is “an enclitic that indicates that the NP to which it is attached is a
definite item previously referred to in the discourse or contextually inferable
therefrom”. It can co-occur in an NP with a nominal demonstrative, as in:
(12) wannàn yārò-n
this.m.sg boy-definite.article
“this boy (that we were referring to) (lit. this boy-the).”

It is useful to compare — for a given language — some of the properties of

nominal demonstratives, of 1st and 2nd person pronouns, of 3rd person
pronouns (if the language has these), and of the definite article (if the language
has one). Table 1 attempts this for English.5
It will be seen, from Table 1, that the nominal demonstratives in English
have a wide range of referential and grammatical properties — combining those
of 1st and 2nd person pronouns, 3rd person pronouns and the definite article,
except that nominal demonstratives are not attested in substitution cataphoric
function. English is, however, fairly unusual. The only properties common to
nominal demonstratives in all languages are 1, having deictic function (and this
is the defining criterion for an item to be a demonstrative) and 4, occurring in
an NP with a noun (this is the criterion for distinguishing this type from other
types of demonstratives). As is illustrated in §5.1, some languages have a single
nominal demonstrative, with no spatial contrast “near” versus “not near”
(property 2). As mentioned above, in some languages a nominal demonstrative
cannot make up an NP by itself (property 3). And in some languages nominal
demonstratives do not have any kind of anaphoric or cataphoric use (properties
5–8). The appendix to this paper provides examples of the different kinds of
anaphoric and cataphoric functions of nominal demonstratives in English.
 Demonstratives 69

Table 1.
Properties of demonstratives and related items in English

nominal 1st and 2nd 3rd person definite
demonstratives person pronouns article
pronouns

this/ that/ I, you, we he, she, it the

these those they

1. Can have deictic function – – – – – –

2. Has spatial reference –x

– x
– – – –

3. Can make up whole NP –x –x – – – –

4. Can occur in NP with noun – – –1 –2 – –

5. Substitution anaphora –x –x – – –
6. Substitution cataphora – – – – – n3
¸ /a
˝ n/a3
7. Textual anaphora –x –x – – – n3
˛ /a
8. Textual cataphora –x –x – – –
x

Although this is a property of nominal demonstratives in English, it is not shown by demonstratives
in all languages.
1.
This covers NPs such as you women.
2.
It is possible to have sentences such as They, the evil spirits, roamed around in the night, but this is
regarded as involving two NPs in apposition (they and the evil spirits) rather than a single NP.
3.
Not applicable; only items which make up a whole NP can have anaphoric or cataphoric function.

However, there is one way in which nominal demonstratives in English are

more restricted than those in other languages. Under property 4, they may
occur in an NP with a noun. As was shown in (6)–(9), other languages freely
allow a nominal demonstrative to co-occur with a noun or with a pronoun.
I mentioned at the beginning of this section that I know of one language
which appears not to have nominal (or local adverbial) demonstratives.
According to Dickens (n.d.: 30) the Khoisan language Ju|’hoan simply has two
verbs, hè “be here, be this one” and toÁà “be there, be that one”. “When qualify-
ing a noun they must, like any other verb, be preceded by a noun with a relative
suffix”. For example, Jùà hè lú¯á mí (person:who is.here told me), i.e. “this
person (or, the person who is here) told me”.

3.2 Local adverbial demonstratives

Whereas nominal demonstratives point to an object, local adverbial demon-
stratives point to a place. Leaving aside Ju|’hoan, languages are known with a
single nominal demonstrative (see §5.1), but no language is known which lacks
<LINK "dix-r22">

70 R. M. W. Dixon

two adverbial demonstratives, contrasting in terms of spatial function (like

English here and there) or in terms of visibility.
In some languages, local adverbial demonstratives must occur with a local
adposition or case affix, whereas in others they require no marking. There is a
tendency for there to be less adpositional or case marking on adverbial demon-
stratives than on other kinds of locational specification. Thus, in English we get:
(13) Local NP with noun head Adverbial demonstrative
He lives at the coast ¸ He lives there
˝
He lives in the mountains ˛
He went to the mountains He went there
He went from the mountains He went from there

That is, a locative preposition (at or in or on, etc.) and the allative preposition
(to) are not normally used before here and there. However, the ablative preposi-
tion (from) must be retained, to distinguish between allative and ablative
specifications, since these may occur with the same verbs (verbs of motion,
etc.). (The interrogative where behaves like there and here; see §4 below.)
We saw in §3.1 that a nominal demonstrative generally occurs with a noun
or pronoun in an NP, although in most languages it can also make up a full NP.
A local adverbial demonstrative is most often the sole locational specification in
its clause but it can, in most (or all) languages, co-occur with an NP bearing
local marking, as in John lives here in the mountains. It is probably best to
consider here and in the mountains as distinct, apposed constituents (which
must occur next to each other, but in either order), rather than as making up a
single constituent. For example, their order can be reversed, giving John lives in
the mountains here (with a slight difference of meaning). These remarks apply
to English; other languages may differ.
In Yidinj, demonstratives inflect for case in a similar manner to nouns and
adjectives. That is, there is just one demonstrative paradigm. We can use the
label “nominal demonstrative” for forms in core syntactic functions (S, A and
O) and for those marked by dative/instrumental, purposive and causal cases;
and “adverbial demonstrative” for those in locative, allative and ablative cases
(Dixon 1977a: 187–9). In (6) we had yi]u dungu “this head”, an object NP
consisting of nominal demonstrative yi]u “this” plus noun dungu “head”.
Similarly, in (14) — from the same text as (6)–(7) — we have yi]gu dugu:da “in
this house” or “here in the house”; this is plainly a single NP consisting of the
near demonstrative in locative inflection and the noun dugur “house” also in
locative inflection (Dixon 1977a: 537).
<LINK "dix-r70">
"dix-r43">

Demonstratives 71

(14) [djambu:l wagu:dja djadja]O gaba:l [yi]gu dugu:da]

[two man child gave.birth.to [this:loc house:loc
“Two male children were born in this house (or here in the house).”

That is, whereas here in the house and in the house here in English are recognised
as involving the apposition of two constituents, here and in the house, in Yidinj
yi]gu dugu:da (or dugu:da yi]gu, since word order is free for this language) is
regarded as one NP, just as yi]u dungu “this head” is in (6).
In English there is an obvious referential connection between this and here
and between that and there. This can lead to a temptation to define the adverbi-
al demonstratives in terms of the nominal ones, or vice versa. The temptation
should be resisted, since the two types of demonstrative, while similar in
reference, are not equivalent.
Lyons (1977: 646) says “roughly speaking … ‘this book’ means ‘the book
(which is) here’”. However, when the noun is changed the equation becomes
less viable. For example, this afternoon can scarcely be rephrased as the afternoon
(which is) here. And difficulties arise even when we confine ourselves to nouns
referring to concrete things. Compare:
(15) a. This hospital has a very fine reputation
b. The hospital here has a very fine reputation

These two statements certainly do not have the same meaning. Example (15a)
is focussing attention on the hospital in which the sentence is uttered, whereas
(15b) draws attention to the location of the hospital, that it is “here”. There
could be an implication in (15b) that the fine reputation is due, at least in part,
to the location of the hospital — that it is, say, part of a university system, or
situated in a rich suburb.
Wierzbicka looks in the opposite direction, taking “this” to be one of a
small number of universal semantic primitives, and defining “here” in terms of
it. She originally had “here” as “this place” (Wierzbicka 1980: 37), later modify-
ing it to “in this place” (Wierzbicka 1994: 469). This would have to be extended
to “in/at/to/etc. this place” (to cover, for instance, Mary is here [at work] and He
came here [to his aunt’s house]). But this direction of definition works no better
than Lyons’ “this man” equals “the man here”. Consider:
(16) a. John lives around here somewhere
b. John lives around this place somewhere

These two sentences have rather different meanings. Example (16a) indicates
that John lives somewhere in the vicinity of where the utterance is spoken. In
<LINK "dix-r22">

72 R. M. W. Dixon

contrast, (16b) identifies a specific place and relates the location of the residence
to that place.
Notice also that, as mentioned above, one can include both a locative
demonstrative and a locative phrase (as constituents in apposition) within a single
sentence: John lives here around this place somewhere or John lives around this place
somewhere here (with a slight difference in meaning). Indeed, one can also say,
in colloquial style, This hospital here has a very fine reputation, relating to (15).
Languages differ in the way different types of demonstrative are related.
Here and there have quite different forms from this and that in English and, in
keeping with this, this man cannot be substituted by the man here, nor here by
this place. But in Yidinj the nominal and adverbial demonstratives are different
sections of a single paradigm. We analysed yi]gu dugu:da in (14) as a single
constituent. Its internal structure could be taken to be “[demonstrative
house]-locative”, i.e. “in this house”, or “demonstrative-locative house-
locative”, i.e. “here in the house”. There is no way of deciding between these
alternative analyses. That is, one could not, in Yidinj, state a contrast similar to
that between (16a) and (16b). Wierzbicka’s suggestion would be appropriate for
Yidinj, although it is not correct for English.
A fair number of languages have a second type of adverbial demonstrative,
with a manner meaning, “like this, in this way/manner”; these are generally
non-inflecting, and serve to modify the verb of the clause. They are in most
cases derived from the nominal demonstratives.

3.3 Verbal demonstratives

A small number of languages lack manner adverbial demonstratives but have a
subclass of verbs with demonstrative meaning, involving deictic reference to an
action. Verbal demonstratives can be illustrated for the Boumaa dialect of Fijian
(Dixon 1988: 61, examples from text 4, lines 206, 86 and 24, pp. 308–27). A
further example is included in the case study of Dyirbal, in §7.3.
The Boumaa Fijian verb ’ene(ii) can be glossed “do like this”; it has three
basic functions. First, it may have deictic reference to an activity, either actual
or mimicked. One story tells how, in a battle, the Prince of Boumaa speared to
death many soldiers of the opposing army:
(17) [o ’ea]S {’eneii tuu gaa ’eneii}PREDICATE
[art 3sg {do.like.this aspect just do.like.this
“He did just like this.” [narrator mimes a spearing action]
 Demonstratives 73

The second function is anaphoric. Earlier in the same story a bridegroom is

slain on his marriage bed and the bride cries uncontrollably. The narrator
describes the intensity of her wailing and then tells how a relative was told to go
and investigate what had happened, since it was the first time that, on the first
night of a marriage:
(18) {saa tagi ti’o ’eneii-maa}PREDICATE [a wati-na]S
{aspect cry aspect do.like.this-that/there [art spouse-3sg.poss
“The wife had continually cried like that.”
Here the verbal demonstrative ’eneii accompanies the lexical verb tagi “cry”, the
whole predicate referring back to the earlier description of uncontrolled wailing.
The third function is to introduce direct speech, as in (from the same text):
(19) {aa ’eneii sara ’eneii-qee}PREDICATE
{past do.like.this immediately do.like.this-this/here
[o tabu]S ·direct speech followsÒ
[art attendant
‘The (Prince’s) attendant then spoke like this.” ·direct speech followsÒ
Looking now at the syntactic possibilities for the verbal demonstrative in
Boumaa Fijian, there are again three. Firstly, note that a predicate in Fijian
involves a verb which may be preceded and followed by a string of elements,
including bound pronouns, tense and aspect markers and quantifiers.
A verbal demonstrative may be the head of a predicate. One story describ-
ing traditional customs concludes with:
(20) {e ’eneii-mayaa}PREDICATE [a ’e-na iva’arau]S [i Taveuni]
{3sg.s do.like.this-that/there [art cl-3sg custom [on place
“That’s the way the custom (of greeting a visiting high chief) is carried
out on Taveuni (island) (lit. the custom is done like this on Taveuni).”
Alternatively, the verbal demonstrative can be a modifier to the head and is then
placed last in the predicate, as in (18), where the predicate head is tagi “cry”,
preceded by aspectual marker saa “happening now but not in the past” and
followed by aspectual element ti’o “continuous” and then the verbal demon-
strative ’eneii.
The third alternative, which is illustrated by (17) and (19) is for the
predicate to include two tokens of the verbal demonstrative, one as head and
the other as modifier.
Now Boumaa Fijian has a set of forms which function (without inflection)
both as nominal and as local adverbial demonstrative. There are three terms:
<LINK "dix-r22">

74 R. M. W. Dixon

“this/here (near speaker)”, “that/there (near addressee or mid distance)” and

“that/there (far)”. One of these may be added to a verbal demonstrative,
whether functioning as predicate head or as predicate modifier. Thus, we get
-mayaa “that/there (far)”, in (20); -maa, a variant of “that/there (far)” from
another dialect, in (18); and -qee a variant of “this/here (near speaker)” from
another dialect, in (19). (Note that speakers of Boumaa Fijian show more
dialect mixing in the use of demonstratives than in any other part of the
grammar; see Dixon 1988: 58–9.)

4. Forms

We can now discuss the ways in which languages have the same or different
forms for the various types and subtypes of demonstratives.
There is little data on verbal demonstratives. The Boumaa Fijian ’eneii —
illustrated in (17)–(20) — has a quite different form from the nominal/
adverbial demonstratives, set out in (33) below. The verbal demonstrative
yalama- in Dyirbal is similar in form to the adverbial demonstrative yalay
“here” and to a noun marker yala- “near speaker”, which is not in fact a
demonstrative; this is fully discussed in §7.
Table 2 presents a sample of the kinds of similarities and differences
encountered between (a) nominal demonstratives functioning as a full NP; (b)
nominal demonstratives functioning in an NP with a noun or pronoun; and (c)
local adverbial demonstratives. Each language has two or more items in each
set; just the “this” and “here” forms are given in the Table.
When a spatial element can be segmented out, it is underlined. For exam-
ple, in Japanese the syllable ko- indicates “this/here” (near speaker); it can be
substituted by so- “that/there” (near addressee), by a- “that/there” (not in
conversational space), or by do- for interrogatives. In Ponapean the final -t
indicates “this/here” and is replaced by -n for “that/there”. In Khmer nìh is
“here” and nùh is “there”. Mupun marks the spatial dimension by tone — low
tone on s6̀ “here” and high tone on s6́ “there”.
Looking now at the sets of rows in Table 2:
I. We here get a different form in each of the three columns. Note that in
French the forms are diachronically related but synchronically distinct. This
language has an unusual way of marking “this” and “that” with nominal
demonstratives. An element -ci “this” (related to ici “here”) or -là “that”
(identical to là “there”) attaches to the demonstrative in the first column and to
 Demonstratives 75

Table 2.
Forms of nominal and adverbial demonstratives, illustrated for ‘this’ and ‘here’
nominal demonstratives

as full NP in NP with noun/ local adverbial

pronoun demonstrative

I ko-re ko-no ko-ko Japanese – Coulmas (1982)

celui/celle + -ci ce(t)/cette … + -ci ici French

II i]gi ti Mangap-Mbula – Bugenhagen

(1994: 94)

III a-n-a a-n a-] Awa Pit (Barbacoan, Colombia)

– Curnow (1997: 87, 94, 108)

IV me-t -e-t ie-t Ponapean (Austronesian) –

Rehg (1981: 143–54)
mâ-n -ni èn Lango (Nilo-Saharan) –
Noonan (1992: 86, 108, 334)

V this here English

VI i-ni s-i-ni Indonesian (Austronesian) –

Sneddon (1996: 160, 189)
nìh ‘i-nìh Khmer (Austroasiatic) – Jacob
(1968: 142)

VI 16̀ + s6̀ s6̀ Mupun (Chadic) – Frajzyngier

I (1993: 84–9)

VI yai Boumaa Fijian (Austronesian)

II – Dixon (1988: 58)

the accompanying noun in the second column; for example celui-ci “this (m)”,
as a full NP, and cette plume-ci “this pen (f)”.6
II. In Mangap-Mbula, one form functions as an adverbial demonstrative and as
a nominal demonstrative used with a noun, with a separate form for a nominal
demonstrative used alone.
III. In Awa Pit the nominal demonstratives have basic forms a-n “this” and su-n
“that”. When functioning as a full NP the focus marker na is added, with
reductions an-na > ana and sun-na > suna. The adverbial demonstratives are a]
“here” and u] (where su] would be expected) “there”.
IV. In Ponapean and Lango, the nominal demonstrative has its full form when
making up a complete NP but when used with a noun it reduces to be a suffix to
the noun. In Ponapean the initial consonant is omitted, whereas in Lango the
<LINK "dix-r67">
"dix-r38">
"dix-r11">
"dix-r53">

76 R. M. W. Dixon

initial syllable má- is generally omitted, except that “this” (the form in the Table)
is irregular, with mân reducing to -ni. The adverbial demonstrative appears to
be paradigmatically related to the nominal in Ponapean, but not in Lango.
V. Many languages are like English in having one form for the nominal demon-
strative (in all contexts) and another for the adverbial.
VI. Here the adverbial demonstrative is derived from the nominal form. In
Indonesian an initial s- is added: i-ni “this”, s-i-ni “here”; i-tu “that”, s-i-tu
“there” (there is a third adverbial demonstrative — see §6.1). Khmer simply
forms a compound with the preposition ‘ae “at”, with reductions ‘ae-nìh >
‘i-nìh and ‘ae-nùh > ‘i-nùh.
VII. There is here a derivation in the opposite direction. In Mupun nominal
demonstratives are formed from adverbial demonstratives by adding an initial 16̀.
VIII. In the Boumaa dialect of Fijian a single form functions as nominal and as
adverbial demonstrative; the full set of forms is at (33) in §6.1.
For Koasati (Muskogean family), Kimball (1991: 486) identifies ten nominal
demonstratives; five of these are used unchanged as local adverbial demonstrat-
ives while two add -á:li (the remaining three appear not to have an adverbial
correspondent).
There is one other kind of conditioning for demonstrative forms. Whereas the
Boumaa dialect of Fijian has a single form for nominal and local adverbial
demonstratives, the Bau (or standard) dialect has two sets of forms — one is
used for nominal demonstratives and for local adverbials when no preposition
is employed, and the other after a local preposition “at”, “to” or “from”
(Churchward 1941: 28).7
In some languages nominal demonstratives can have an anaphoric (and
sometimes also a cataphoric) function but in other languages they lack this.
There are languages which employ a special form for anaphora. The forms in
the Harar dialect of Oromo include (Owens 1985: 27, 87–8):
(21) “this” “that” anaphoric
masculine xuni ¸ Ï xáaní
˝ suni Ì
feminine tuni ˛ Ó táaní

In Muna (Austronesian, spoken in Sulawesi) the root form of a nominal demon-

strative is used in anaphoric function, with an initial a- added for deictic function.
For example (van den Berg 1997: 199; the full system is set out in (39) below):
<LINK "dix-r3">
"dix-r31">
"dix-r10">
"dix-r14">
"dix-r21">

Demonstratives 77

(22) anaphoric deictic

nearby maitu a-maitu
far (neutral) watu a-watu
far (high) tatu a-tatu

Another possibility is for a language to use a reduced form of a nominal

demonstrative for anaphoric function. For example, in Longgu (Austronesian,
spoken in the Solomon Islands; Hill 1992: 96–7), the deictics nene “this” and
nina “that” are shortened to -ne and -na respectively when used anaphorically.
There is further discussion of the anaphoric and cataphoric use of demon-
stratives in §5.3.
Where there is a set of manner adverb demonstratives, these are typically
derived from the nominal forms. For example, in Mandarin Chinese corre-
sponding to nominal demonstratives zhè- “this” and nà- “that” there are
manner forms zhè-yàng “this way, like this” and nà-yàng(zi) “that way, like
that” (Chappell 2001). In Japanese, the manner adverbial demonstratives koo,
soo and aa correspond to the nominal demonstratives ko-re “this, near speaker”,
so-re “that, near addressee” and a-re “that, distant from speaker and addressee”
(Coulmas 1982). It appears that languages with manner adverbial demonstrat-
ives derived from nominal demonstratives typically have local adverbial
demonstratives also derived from nominals, e.g. zhèr ~ zhèli “here”, nàr ~ nàli
“there” in Mandarin Chinese; and koko “here, near speaker”, soko “there, near
addressee”, asoko “there, distant from speaker and addressee” in Japanese. (See
also Diessel 1999: 74–5.)
We can now look at formal similarities between demonstratives and other
items. In a number of languages, interrogatives fall into the same paradigm as
demonstratives. In Japanese there is a series of interrogative forms parallel to
the demonstratives: nominal dore “which”, local adverbial doko “where” and
manner adverbial doo “in what way, how”. In Tamil the paradigm includes
(Asher 1985: 150):
<LINK "dix-r19">
"dix-r50">
"dix-r22">

78 R. M. W. Dixon

(23) Demonstratives

Proximate Remote Interrogatives

nominal i-nta “this” a-nta “that” e-nta “which”

adverbial i-]ke “here” a-]ke “there” e-]ke “where”
temporal i-ppa “now” a-ppa “then” e-ppa “when”
quantity i-ttane “this many” a-ttane “that many” e-ttane “how
manner i-ppati “in this way” a-ppati “in that way” many”
e-ppati “how”

Similar paradigms are found in many languages from South Asia, both Indo-
Aryan and Dravidian.
Newman (2000: 147) provides a paradigm for Hausa which links together
nominal demonstratives (with four degrees of distance), three varieties of
interrogative (“who”, “which” and “which one”) plus two indefinites (“some/
other” and “so and so”), each in three genders. Other languages which have
interrogatives in the same paradigm as demonstratives include Bengali and
Vietnamese.
In English the nominal interrogatives (who, what, which) are quite different
in form from the nominal demonstratives (this, that) but adverbials do show
similarity in form: where, here, there. There is also a syntactic similarity — where
is like here and there — shown in (13) — in that it can omit a preceding allative
or locative preposition.
In some languages demonstratives have a quite different pattern of inflec-
tion from personal pronouns. In Yidinj, for instance, 1st and 2nd person
pronouns follow a nominative(SA)/accusative(O) system (there is no 3rd
person pronoun); inanimate demonstratives have an absolutive(SO)/erga-
tive(A) system (like nouns), while animate demonstratives have a distinct form
for each of the three core syntactic functions, S, A and O (Dixon 1977a: 187).
In other languages demonstratives show formal similarities to personal
pronouns. For example, in Hixkaryana (Carib, spoken in Brazil; my reanalysis
of Derbyshire 1985: 7) we get the following:
<LINK "dix-r30">
"dix-r4">

Demonstratives 79

(24) Animate

Pronouns Minimal Augmented Inanimate

1 (“I’) u-ro amna –

1 + 2 (“I and kṫw-ro kṫw-yamo –
you”) om-oro om-nyamo –
2 (“you”) n-oro nyamo-ro –
3 (“he, she, it”)

Demonstratives

near mosonṫ mox-amo onṫ

far mok-ro mok-yamo moro
distant mokṫ “ monṫ

This paradigm is rather irregular, although we can recognise a final -(o)ro in

five of the seven forms from the minimal animate column, and -nyamo or
-yamo or -amo in all but one of the forms from the animate augmented column.
The main point of similarity is categorial — that both pronouns and demon-
stratives have distinct forms for “minimal” (referring to one person, except for
the “I and you” form) and for augmented (one or more others in addition to
the minimal specification).
The Australian language Diyari has no demonstrative nominal roots per se.
However, the 3rd person pronouns (in sg, du and pl number) can optionally
take one of a set of deictic suffixes which indicate relative location with respect
to the speaker (Austin 1981: 60–1). That is, in Diyari demonstratives are based
on 3rd person pronouns.
Another kind of item to which a demonstrative may show formal similarity
is the definite article (in the minority of languages in which this occurs). We
noted above that in standard German the forms die (f and pl), der (m) and das
(n) function as definite articles when unstressed and as nominal demonstratives
when stressed. (Heine and Kuteva 2002: 109–11 illustrate how a demonstrative
may develop into a definite article.)

5. Functions

We first discuss the defining function for demonstratives, deictic reference

(§5.1); then syntactic function (§5.2), followed by anaphora and cataphora
80 R. M. W. Dixon

(§5.3), and finally other functions such as identification, new information and
discourse organisation (§5.4).

5.1 Deictic function

We can illustrate this for English, a language which has two nominal demon-
stratives, this and that, contrasting in terms of the relative spatial location of
their referents.
Consider the following situation. John and Mary are sitting at a table on
which are placed two bowls of strawberries, X and Y; bowl X is nearer to both
John and Mary with Y being further away. This can be shown diagrammatically:

Y
X Mary

John

Mary offers John a bowl of strawberries. She could offer X or Y, saying in either
case Would you like this one? Suppose that John prefers the other bowl; his reply
would vary, depending on whether he had been offered X or Y. Thus:
(25) a. Mary: Would you like this one? [pointing at X]
John: No, I’d rather have that one [pointing at Y]
b. Mary: Would you like this one? [pointing at Y]
John: No, I’d rather have this one [pointing at X]

In (25a) Mary offers John the nearer bowl and refers to it by this; he prefers the
farther one, and refers to it by that. In (25b) she offers him the farther bowl and
again refers to it by this. He again prefers the other one; however, he cannot
refer to it by that, since it is nearest to him of the two bowls, and must use this.
Note that the response in (25b) could be expanded to: No, I’d rather have this
one [pointing at X] than that one [pointing at Y].
Mary uses this in each of (25a) and (25b) since in English this is typically
employed to introduce new information. The spatial sense of this only comes
into play when there is an explicit spatial contrast between two objects, at
different distances from the speaker. In the second utterances of (25a) and
(25b), John is comparing X and Y and so uses this for the bowl that is nearest to
him (X) and that for the one which is further away (Y).
<LINK "dix-r21">
"dix-r8">
"dix-r46">
"dix-r15">

Demonstratives 81

Suppose that you go to the dentist complaining of a sore tooth. The dentist
taps a tooth and asks Is it this one? You point to a different tooth and say No, it’s
this one. In the case of a speaker’s teeth, there is no contrast of “relative distance
from speaker”; thus, only this would be used (not that) in such circumstances.
In summary, it can be seen that — in deictic use — this is the primary
nominal demonstrative in English. When only one object is being discussed,
this is used. When there are two objects which cannot be distinguished in terms
of relative distance from the speaker, this is used for each. When two objects
vary in relative distance then this is used for the one nearer to the speaker and
that for the one further off. Interleaving with this, in the case of English, is the
use of this to introduce new information (mentioned under (ii) in §5.4). It is
the “new information” sense which motivates Mary to use this for referring to
both X in (25a) and Y in (25b).
I would think it unlikely that this account of the use of this and that in
English would apply, point by point, for other languages which have two
nominal demonstratives distinguished in terms of distance. Each language
needs to be investigated, in its own terms, with cross-linguistic generalisations
then being put forward on an inductive basis.
Some languages have a single nominal demonstrative. In Supyire (Niger-
Congo type, spoken in Mali; Carlson 1994: 160–1) there is a single form
(marked for gender and number) which is, presumably, used to translate all
instance of this and of that in (25a/b). X and Y would be identified just by
pointing. In the English versions of (25a/b) the two bowls are identified by
pointing, with concomitant specification by the use of this and that.
Other languages with a single demonstrative include Dyirbal; see §7.1. A
single nominal demonstrative is also reported for colloquial Czech (Meyerstein
1972; Cummins 1998) and for some dialects of German. It appear that all of the
languages with just one nominal demonstrative do have two adverbial demon-
stratives, similar to here and there in English. (And see Diessel 1999: 50.)
The adverbial demonstratives in English, here and there, have parallel deictic
use to this and that. Referring again to the table at which John and Mary are
sitting, suppose now that X and Y are plates. Mary holds a cake in her hand and
enquires as to which plate she should put it on. There are again two scenarios:
(26) a. Mary: Shall I put it here? [pointing at X]
John: No, put it there. [pointing at Y]
b. Mary: Shall I put it here? [pointing at Y]
John: No, put it here. [pointing at X]
<LINK "dix-r29">
"dix-r22">
"dix-r17">

82 R. M. W. Dixon

Exactly the same discussion applies as for this and that. But it should not be
assumed that adverbial demonstratives always have the same deictic functions
as nominal demonstratives. This is something which must be investigated
separately for each language.

5.2 Syntactic function

In §3.1 we noted that a nominal demonstrative can always occur in an NP with a
noun (and, in some languages, also in an NP with a pronoun). In the majority of
languages it can make up an NP on its own, but this is not a universal property.
In most languages, an NP including a nominal demonstrative can appear in
any core or peripheral function in a clause. But some languages do have restric-
tions. In §7.1 we describe how in Dyirbal an NP including the nominal demon-
strative may only occur in S or O function (the pivot functions in this syntacti-
cally ergative language). If a speaker needs to point out an object that is in
underlying A function, then the antipassive derivation must be applied,
bringing that argument into derived S function. It appears that a similar restriction
applied in Warrgamay, Dyirbal’s southerly neighbour (Dixon 1981:44–5) and also
in the Philippines language Northern Subanen (Daguman 2001).
§4 discussed the different forms nominal demonstratives can take, depending
on whether they make up a full NP or occur with a noun or pronoun. More rarely,
nominal demonstratives can have different forms depending on the syntactic
function of the NP in which they occur. For example, Hayward (1990: 273–4)
lists the forms of the six nominal demonstratives in Zayse (a Cushitic language
from Ethiopia) used for (a) functioning in an NP with a noun; (b) as a full NP
in subject function; (c) ditto in object function; (d) ditto in copula complement
function. These include (further terms are mentioned in §6.2):

(27) in NP as full NP
with noun
subject object copula complement

near speaker ha há(j) hayá háytte

near addressee yi yií yá yítte
far away so só(j) soyá sóytte

The syntactic functions of local adverbial and of verbal demonstratives should

also be examined, in terms of the clause types they may occur in. There appear
to be no restrictions for most languages, but there may well be occasional
 Demonstratives 83

instances where an adverbial or a verbal demonstrative is not permitted in

certain types of subordinate clause. These will only be uncovered if a linguist
working on the language is alert to the possibility.

5.3 Anaphora and cataphora

In a number of languages (for example, Ainu, Dyirbal, Jarawara), nominal
demonstratives do not have any anaphoric or cataphoric use. In languages
where they do, it is interesting to relate anaphoric and cataphoric functions to
the distance specifications of the demonstratives.
Table 3 provides some examples from the literature. (The majority of
sources do not specify what type of anaphora is involved; in most cases it is
textual anaphora but may sometimes be substitution anaphora.) In both
Quechua and Fox the “near” demonstrative is used for cataphora. For ana-
phora, Fox uses the “non-near” term from a two-term system, and Quechua the
“mid” term from a three-term system (the “far” term apparently has neither
anaphoric nor cataphoric use). In Cayuga only the “non-near” term is used for
textual reference, all of the available examples being anaphoric. Tamil again

Table 3.
Anaphoric and cataphoric functions of nominal demonstratives

Nominal demon- Used for anaphora Used for cataphora
stratives

near, mid, far mid near Quechua – Weber

(1989: 38)
near, non-near non-near near Fox (Algonquian) –
Dahlstrom (mss)
near, non-near non-near – Cayuga (Iroquian) –
Sasse (1999)
near, non-near both near Tamil – Asher
(1985: 80–1)
near speaker, near all far Georgian (Kart-
addresee, far velian) – Imedadze
and Tuite (1992: 105)
near, mid, far near: recent ana- – Basque – Saltarelli
phora (1988: 213)
far: distant anaphora
one only demonstrative demonstrative plus Supyire (Niger-Con-
3rd person pronoun go) – Carlson
(1994: 190–4)
<LINK "dix-r24">
"dix-r32">

84 R. M. W. Dixon

uses the “near” term for (infrequent) cataphora, and both demonstratives for
anaphora. Georgian appears to go against the pattern in using the “far” demon-
strative for cataphora (and all three for anaphora). In Basque, just the “near”
and “far” terms have anaphoric reference, the “near” demonstrative for
something recently mentioned in the text and the “far” term for something
mentioned further in the past.
As mentioned before, Supyire has just one nominal demonstrative. This can
be used by itself for anaphora, and it is combined with a 3rd person pronoun
for cataphora. We saw, in (12), that in Hausa a nominal demonstrative used
with the definite article can have anaphoric function.
Verbal and local adverbial demonstratives may also have anaphoric
function. The anaphoric use of the verbal demonstrative ‘eneii “do like this”, in
Boumaa Fijian, was illustrated in (18).
The anaphoric function of adverbial demonstratives in English is illustrated in:
(28) John moved to Melbourne in 1959 and lived here [if the speaker is situated
in Melbourne] / there [if the speaker is situated elsewhere] for the next
ten years

In (28) the demonstrative has both anaphoric and deictic effect. That is, which
of here and there is used in this instance of anaphora depends on the location of
Melbourne with respect to the speaker.
Very little work has been done on the anaphoric (and cataphoric) functions
of local adverbial demonstratives; this should be a priority for future research.

5.4 Other functions

i. Identification. There are several ways in which nominal demonstratives can be
used for identification or recognition (see Himmelmann 1996 and Lindström
2000).
For example:
(29) It was that sort of gluggy rice which the Japanese go in for.

Here that identifies what sort of rice it was; note that in this particular use the NP
must include a relative clause providing a description of what sort of rice it was.
Some languages have a special demonstrative for identification.8 For
example, ce in French can be used as subject of the copula être “be” (or of
pouvoir être “can be” or devoir être “must be”). In an example from Flaubert
(Ferrar 1967: 215):
<LINK "dix-r48">

Demonstratives 85

(30) Un beuglement formidable s’éleva. C’était un taureau.

“A frightful bellowing arose. It was a bull.”

ii. New information. The nominal demonstrative this in English is typically used
as a mark of new information. Indeed, this use has in recent years expanded in
colloquial speech. For instance, a child may come home from school and say to
its parent:
(31) There’s this new girl at school today and she talks really funny

iii. Discourse organisation. There are two local adverbial demonstratives in

Jarawara, ahi “here, visible” and fahi “here/there, not visible”. These are used in
deictic function, and also to mark the development of a discourse.
Fahi is employed to mark the climax of some particular segment of discourse.
In one recorded story, the narrator spends 16 clauses telling how his tape-recorder
wouldn’t work — he took the back off, fixed the insides, put it back together
again, and then (including fahi on the final clause of this segment of discourse):
(32) karafatoO jaro
tape-recorder (f) start.up
o-ka-na-ma-hara-ke fahi
1sg.a-applic-aux-back-immediate.past:eyewitness:f-dec:f climax
“I switched it back on.”

Other instances of the “climax” use of fahi include: (a) he got hold of the snuff;
he rocked the snuff back and forth in his hand; and then he sniffed the snuff
fahi; and (b) he found the woman; he grabbed her; he put her down on the
ground; and then he copulated with her fahi.
We also find ahi as a discourse marker (although much less often than fahi).
It can mark something that is a “lead up” to a final climax. One or more ahi
“lead up” clauses are generally followed by a “climax” clause marked by fahi.
Mithun (1987) provides an insightful discussion of the discourse function
of demonstratives in the Iroquoian language Tuscarora.

6. Reference

The major parameters of reference for demonstratives are: spatial — sometimes

extended to temporal — location (§6.1), height and stance (§6.2), and visibility
(§6.3). Other, more abstract, senses are mentioned in §6.4. In §6.5 there is brief
discussion of person, gender/noun class, classifiers and number.
<LINK "dix-r44">
"dix-r43">
"dix-r22">

86 R. M. W. Dixon

All of these kinds of referential information may also be coded at other

places in a clause. There may be verbal affixes “up” or “down”, or “coming” or
“going”. Gender, noun class and number may be marked on other nominal and
verbal elements. We are here concerned only with these categories as they are
encoded in demonstratives — that is, in forms which can have deictic function.

6.1 Spatial reference

The majority of languages have two demonstratives, relating to “near speaker”
and “not near speaker”. There are also a fair number of languages with a three-
term system. This can be one of two types:
a. Near speaker; near addressee; not near speaker or addressee — found in
Japanese, Basque, Quechua, Swahili, and Thai, among many other languages.
b. Near speaker; mid distance to speaker; far from speaker — found in Georgian,
Lango, Ponapean and Hixkaryana, among many other languages.
(Note that a two-term system is “±[near speaker]” rather than “near ±[speak-
er]”, pace Lyons 1968:278 who, comparing the system in English with a three-
term system of type (b), suggests “English neutralizes the distinction
between the ‘second’ and ‘third’ person in this category of ‘proximity’”.)
Some languages have a three-term system which is a blend of (a) and (b). For
example, the nominal/verbal demonstratives in Boumaa Fijian are (Dixon
1988: 58–9):
(33) yai “this/here”, near speaker
yaa “that/there”, mid distance from speaker, often near addressee
mayaa “that/there”, far distance from speaker and addressee
If two people are talking together, each will use yaa for “near you”. But in a
narrative yaa is employed simply for “mid distance”. (Similar remarks appear
to apply for Hopi — see Malotki 1982.)
An important point to note is that the distances involved for “mid” and
“far” terms are relative. One story that I recorded in Boumaa Fijian took place
entirely within the north-eastern part of the island of Taveuni. The “far” term
was used for a place about ten kilometres away from where the narrator was
sitting. In another story, the action flowed across from Taveuni to the nearby
island of Vanua Levu. The “mid distance” term was used of a place on the near
side of Vanua Levu, about thirty kilometres from where the narrator was sitting,
with the “far” term used to refer to a place further away in Vanua Levu, about
fifty kilometres from the place of narration.
<LINK "dix-r54">
"dix-r65">
"dix-r62">

Demonstratives 87

Ainu has a demonstrative system of particular interest. Quoting from

Tamura (2000: 261): “there are the following three kinds of noun-modifying
spatial demonstratives.
– tan, emphatic form tapan. Expresses something that is present, visible, is
the new topic of conversation, or that is located where the conversation is
taking place: ‘this, here’.
– taa. Something that is in the immediate vicinity of the speaker: ‘this, here’.
– toan. Separated from oneself: ‘that, there’.”
As illustrations of use, Tamura gives:
(34) tan/tapan kampi “this letter” (e.g. that one has in possession, or is
writing now)
taan kampi “this letter” (e.g. on the desk, right next to the speaker)
toan kampi “that letter” (e.g. a letter in a letter rack over there)
Some languages have different deictic gestures for relating to varying distances
and visibility. In the Tucano and Arawak languages of the Vaupes river basin
(spanning the border between Brazil and Colombia), for instance, we find: (i)
pointing with the lips for “visible and near”; (ii) pointing with the lips plus a
backwards tilt of the head for “visible and not near”; (iii) pointing with the
index finger for “not visible” (if the direction in which the object lies is known).
In most languages, nominal and local adverbial demonstratives make the
same spatial distinctions. However, there are rare exceptions. We noted in §5.1
that the few languages with a single nominal demonstrative appear all to have
a spatial contrast for adverbial demonstratives. Other languages exhibiting a
difference between the types include:
i. As illustrated in Table 2, Indonesian has just two nominal demonstratives.
There is, however, a three-term system for the adverbials. That is (Sneddon
1996: 160, 189):
(35) Nominal Local adverbial
ini “this”, near speaker sini “here”, near speaker
itu “that”, not near speaker situ “there”, mid distance from speaker
sana “there”, far from speaker
ii. Yagua (spoken in Peru; Payne and Payne 1990: 374–5, 257, 271) has two
nominal demonstratives, jiyi- “this” (near) and ru- “that” (far); these are
suffixed with an appropriate classifier. This language has adverbial demon-
stratives specifying four degrees of distance from the speaker: jiyu “here”, diíy
“there (nearby)”, jásiy “there (mid distant)” and múúy “there” (far distant).
<LINK "dix-r2">
"dix-r66">
"dix-r49">
"dix-r13">

88 R. M. W. Dixon

In these and a variety of other instances, adverbial demonstratives make more

distinctions than do the nominal variety. I know of two examples in the
opposite direction. Tariana (Arawak family, Brazil; Aikhenvald forthcoming)
has four spatially distinguished terms in the nominal system but only two for
adverbials; and Lillooet (Salish family; van Eijk 1997: 168–77) has three nominal
demonstratives but only two local adverbials (each nominal and adverbial form
being combined with a visible/invisible specification).
In some languages the spatial reference of nominal demonstratives can, by
analogy, be extended to time. For example, Iraqw (a Cushitic language spoken
in Tanzania), has four nominal demonstratives: “near speaker”, “near address-
ee”, “mid distant” and “far”. The two latter are used anaphorically, “mid
distant” in present tense and “far” in past tense (Mous 1993: 90–1). For Sarcee
(an Athapaskan language spoken in Alberta, Canada), Cook (1984: 73–5)
describes a system of three nominal demonstratives; the “near” term can refer
to “someone who is physically present near the speaker (and hearer) or to
somebody being talked about at the narrative moment”, and the “far” term
“refers to someone who is away from the speaker (and hearer) or to someone
mentioned in a story on an earlier occasion”. In some languages, local adverbial
demonstratives may have a secondary sense referring to time. In others, there
may be distinct forms for temporal adverbial demonstratives; this applies for
Tamil, as shown in (23). (Anderson and Keenan 1985 provide further examples
of spatial demonstratives also being used for temporal reference, in Australian
and Austronesian languages.)
I know of only two languages with verbal demonstratives — Boumaa Fijian
and Dyirbal — and in each instance there is just one verb “do it like this” (with
transitive and intransitive variants in Dyirbal). Further work may well reveal a
language with a contrast between two verbal demonstratives, but I suspect that
most languages which show this category will have a single item.
A topic which has not been systematically studied on a cross-linguistic basis
is the iconicity or sound symbolism typically found in demonstratives which
distinguish degrees of distance. As a first approximation, the “near” term is
more likely than “non near” terms to involve a front high vowel and/or laminal
or dental consonant(s). Compare for instance, this and that in English, -ci and
-là in French, yi]u and ]u]u in Yidinj, níh and núh in Khmer, ii and aa in
Telugu. This is a topic which will require careful and systematic study.9
In Indo-Pakistan Sign Language and in American Sign Language (the only
sign languages for which I have this kind of data) there are two deictic terms, as
in the majority of spoken languages. Interestingly, in Indo-Pakistani Sign
<LINK "dix-r5">
"dix-r74">
"dix-r37">
"dix-r45">
"dix-r28">
"dix-r72">
"dix-r29">

Demonstratives 89

Language a sign which involves pointing downwards in front of the signer’s feet
may have spatial reference “here” or temporal reference “now” or “today”
(Zeshan 1999: 60).

6.2 Height and stance

Some languages spoken in hilly country code, onto their demonstratives, reference
not only to distance but also to height. In the Arakul dialect of Lak (North-east
Caucasian family) there are three nominal demonstratives (Khaidakov 1966:12):
(36) aha close to speaker
hava farther from speaker, but on the same level
ho higher or lower than the speaker

Lahu (Tibeto-Burman, spoken in northern Thailand) distinguishes between

higher and lower in its system of five spatial demonstratives (Matisoff 1973: 51):
(37) chó here
ô there
cô way over there
nô up there
mô down there

“Uphill” and “downhill” demonstratives are also found in Hua (Papuan type,
spoken in the highlands of New Guinea; Haiman 1980: 258) and in Southeast
Tepehuan (Uto-Aztecan family, spoken around moutainous ridges in the state
of Durango, Mexico; Willett 1991: 187–8). Example (27) in §5.2 listed three of
the demonstratives for Zayse; the remaining terms are “equally near speaker
and addressee”, “at lower level than speaker” and “at higher level than speaker”
(Hayward 1990: 273). As shown at (22) in §4, Muna has a “far (high)” term; the
“far (neutral)” term refers to something on the same level as the speaker or
lower than them (the full system is given in (39)).
As described in §7, Dyirbal has a number of spatial elements which may
optionally be added to a demonstrative or noun marker; these specify three
degrees of distance in terms of uphill and downhill, two degrees of distance in
terms of upriver and downriver, plus “across river” and “long way out”.
Other languages indulge in further kinds of spatial specification. For
instance, Boas (1911: 41) mentions an Eskimo system with reference not only to
“near me”, “near thee”, “near him”, “above me” and “below me”, but also to
“behind me”, “in front of me”, “to the right of me” and “to the left of me”.
<LINK "dix-r47">
"dix-r9">
"dix-r57">
"dix-r1">
"dix-r5">
"dix-r66">

90 R. M. W. Dixon

In languages of the Waikurúan family (spoken in the Chaco area of Brazil

and Argentina) demonstratives must specify the stance or motion of the
referent — whether “standing”, “sitting”, “lying”, “coming” or “going” (Ceria
and Sandalo 1995: 181). Similar specification is made in some languages of the
Siouan family, in the set of classifiers which is used only with demonstratives
(Rankin 1976. forthcoming); further examples are mentioned in Aikhenvald
(2000: 176–83).

6.3 Visibility
The criterion used here for a demonstrative is that it must be able to have
deictic (or pointing) function. A number of languages have a grammatical
system with one or more terms referring to visible objects and also one or more
terms referring to something which is not visible. It is natural to ask how a term
referring to something which is not visible can be used deictically, and qualify
as a demonstrative. There are a number of possible responses. The “non-visible”
term could be used to refer to something which was visible but has now moved
out of sight (over a hill, say); one can point in the direction in which it has
gone. Or the “non-visible” term can refer to something which is audible —
round a corner, perhaps, or behind a wall. If we can hear something we can
certainly point in the direction from which the noise is coming. In other
instances, the pointing may be metaphorical, with a “non-visible” term relating
to something which is remembered from the past. It is likely that the labels
“visible” and “non-visible” have varying implications in different grammars.
There is need for detailed study of the meaning of “non-visible” terms, in
languages in which they occur, and then comparison of them.
Boas (1911: 41) drew attention to “visible/invisible” as an obligatory
distinction for demonstratives in Kwakiutl (a Wakashan language). Here it
combines with three degrees of spatial distance, yielding a six-term system:
(38) visible, near me visible, near thee visible, near him
invisible, near me invisible, near thee invisible, near him

In the Salish language Lillooet there is a similar system with “visible/invisible”

applying for each of three distance terms: “near”, “at some distance” and “way
over there” (van Eijk 1997: 168–9).
Other languages simply have one “non-visible” term within the demonstra-
tive system. For example, Shoshone (Uto-Aztecan; Miller 1996: 709) has a four
term system: “near”, “not quite so near”, “far, but in sight” and “not in sight,
<LINK "dix-r73">
"dix-r52">
"dix-r1">
"dix-r2">
"dix-r67">
"dix-r56">

Demonstratives 91

usually far”. Bengali has a three-term system consisting of “near and visible”,
“non-near and visible” and “invisible” (Onishi 1997: 25–6). Palikur (Arawak
family, Brazil; Aikhenvald and Green 1998: 437) has a five term system: “in
speaker’s hand”, “near to both speaker and addressee”, “near to only one of
speaker and addressee”, “far from both but visible” and “far from both and not
visible”. There is a three-term system including a “not visible” term in Dyirbal,
discussed in §7. Jarawara has a two-term system with a contrast between “here,
visible” and “here/there, not visible”.
Some languages have complex demonstrative systems, involving a mixture
of the parameters discussed here, and more besides (there are further examples
in Anderson and Keenan 1985: 286–95).10 There is a seven-term system in
Muna; expanding on the terms given in (22) above (van den Berg
1997: 199–201):
(39) aini near speaker
aitu near addressee
amaitu away from speaker and addressee, but nearby
awatu far away, lower than or level with point of speaking or
orientation
atatu far away, higher than point of speaking or orientation
anagha not visible (may be audible), unspecified for time
awaghaitu not visible, was in view but no longer is

6.4 Other senses

In some languages, demonstratives can take on other kinds of meaning; for
example, indicating an emotional attitude, or personal interest, or familiarity.
Quirk and Greenbaum (1973: 107) suggest that in English this/these may be
used to “connote interest and familiarity”, whereas that/those may imply “a
corresponding emotive rejection”. Zandvoort (1975: 148) is more perceptive in
stating that: “the demonstrative pronouns, especially in their deictic function,
are often used with emotional connotation. The kind of feeling implied
(affection, vexation, contempt, disgust, etc.) depends on the situation.” That is,
both this and that may carry either a positive or a negative overtone. Examples
include: I can’t stand that/this mother-in-law of mine (negative), These/those
modern poets publish a lot (could be positive or negative), This headache is killing
me (negative). Example (40), from a newspaper article, uses that with distinctly
positive connotation:
<LINK "dix-r63">
"dix-r48">
"dix-r2">
"dix-r21">

92 R. M. W. Dixon

(40) Mums and Dads loved his on-screen persona: that nice boy who used to be
in The Andy Griffith Show was now that nice young man with the sensible,
short red hair and freckles.

Chinese has two nominal demonstratives, zhe “this” and na “that”. Zhe is used
for “something close to the speaker, either literally in space or time or in their
thoughts and present interests”. Na is used “when the speaker is referring to
something in space or time which is at a distance relative to themself, or when
they are thinking of objects or situations removed from their present interests
and of small importance to themself” (Li 1996: 22).
Mithun (1999: 132–6) provides a fine survey of complex class systems in
some North American languages, involving such features as “familiarity to
speaker”, and “whether stationary or moving and, if moving, whether towards
the speaker”. There are also surveys of a number of complex systems in Ander-
son and Keenan (1985: 280–99) and Diessel (1999: 35–55).

6.5 Person, gender/noun class, classifiers and number

A number of grammatical categories may be realised at several places in the
clause, including in demonstratives.
There are two ways in which demonstratives can relate to person. Firstly,
they can specify whether an object referred to is near the speaker or near the
addressee. Secondly, a nominal demonstrative may actually refer to speech-act
participants, occurring in an NP with a 1st and 2nd person pronoun. This was
illustrated in §3.1. Example (7) is from Yidinj where there are three demon-
stratives, yi]u “this”, ]u]u “that (far)” and yu]u “that (“very far but visible” in
one dialect, “non-visible” in another dialect); in (7), yi]u “this” co-occurs with
njundu:banj “you all”. Examples (8)–(9) illustrate the nominal demonstrative
haaha “this” in Jarawara co-occurring with a 1st person non-singular pronoun
“these we” and with the 3rd person non-singular pronoun “these they”. A
further example, from Dyirbal, is provided at (51) in §7.1.
Grammarians often do not specify whether or not a demonstrative may co-
occur with a 1st or 2nd (or 3rd) person pronoun. It is important to investigate
this matter.
In languages with a gender or noun class system, this is frequently marked
on nominal demonstratives, as illustrated above in (8)–(9) from Jarawara; in
(12) from Hausa, and in (21) from Oromo. In Shilh (Berber branch of the
Afroasiatic family; Stumme 1899: 92–3), a nominal demonstrative will be
<LINK "dix-r43">
"dix-r1">
"dix-r54">

Demonstratives 93

specified for gender when making up a full NP, but not when modifying a noun
within an NP. And, just as a nominal demonstrative may agree with its (actual
or ellipsed) head noun in gender or noun class, it will often also agree with it in
number, as in Hausa and Supyire.
If a language has classifiers which are used in a multiplicity of contexts,
occurrence with demonstratives is often one of them. There may be a different
set of classifiers used with demonstratives than in other contexts, or they may
have different forms, different orderings, or different degrees of obligatoriness
— see Aikhenvald 2000: 206–41, and especially her Table 9.3.
There can be a dependency between the spatial/visibility system and the
number system. In Panare (a Carib language from Venezuela; Payne and Payne
1999: 97) there are parallel systems of demonstratives for animate and inani-
mate reference. The animate set is:
(41)
singular plural

near and visible mëj mëjchanton

far and visible muku(j)
invisible kën kamonton/kamënton

That is, a number distinction is made for the “near and visible” and “invisible”
demonstratives, but is neutralised for “far and visible”.
One question that has not so far been mentioned is (formal and functional)
markedness within demonstrative systems. For example, which term from a
spatially-determined system will be used in neutral circumstances, if spatial
location is not relevant? Lyons (1977: 647) suggests that “generally speaking, in
English ‘this’ is marked and ‘that’ is unmarked”. This may apply for anaphoric
usage; however, our exploration in §5.1 of the deictic reference of this and that
— in examples such as (25a,b) — would point to this being the unmarked term
in deictic use.
The question of markedness is a difficult one, which may have different
solutions depending on whether one is dealing with deictic or with anaphoric
functions. In some languages there may be no markedness within the demon-
strative system(s), but in others there certainly is. There appears to be a tenden-
cy for the “that” form from a two-term system to be functionally unmarked (it
is in Mandarin Chinese and in Telugu, for instance); however, in Tariana the
“this” nominal demonstrative (from a four-term system) is functionally
unmarked. The question of markedness requires careful examination, in each
individual language.
<LINK "dix-r22">

94 R. M. W. Dixon

7. Case study — Dyirbal

Dyirbal — spoken in north-east Queensland, Australia — has one nominal

demonstrative (described in §7.1), a system of local adverbial demonstratives
(§7.2) and also intransitive and transitive verbal demonstratives (§7.3). The
information here is based on Dixon (1972), with further examples from texts.11

7.1 The nominal demonstrative and related items

Dyirbal has no definite article and no 3sg pronoun. (There is a 3du pronoun,
balagarra, and a 3pl, balama]gan, but these are undoubtedly recent creations,
based on the unmarked noun marker bala plus nominal suffixes -garra “group
of two”, and -ma]gan “group of three or more”. Note that each part can inflect,
as in genitive ba]ugarra]u, in (55b) and (56a) below.)
Each non-local NP may include a “noun marker”, which agrees with the
head noun in case and noun class. There are three possibilities for the first
element in a noun marker:
(42) ba(la)- referent is visible and not near speaker (i.e. “there”)
ya(la)- referent is visible and near speaker (i.e. “here”)
]a(la)- referent is not visible (but may be audible, or remembered from
the past)

The use of noun markers is illustrated in (43), which includes a ]a(la)- marker
in the O NP and a ba(la)- marker in the A NP.
(43) []ala mu]ga]O [ba-]gu-l yara-]gu]A ]amba-n
[non.vis:abs:n [noise(n) there-erg-m man(m)-erg hear-non.fut
“Man there heard noise.”

Now an NP can involve just a noun, or noun plus noun marker, or just a noun
marker. All possibilities occur in the following textual extract:
(44) a. [Burrbula-nja]O [ba-]gu-l Jumbulu-gu]A balay
[name(m)-acc [there-erg-m name(m)-erg there:loc
jaymba-n yuramu-ga
find-non.fut place-loc
“Jumbulu found Burrbula there at Yuramu.”
b. ba-]gu-lA ]anba-n
there-erg-m ask-non.fut
“He (Jumbulu) asked (Burrbula).” ·direct speech followsÒ
<LINK "dix-r22">

Demonstratives 95

In (44a) the O NP consists just of the proper name Burrbula, with an accusative
case suffix (the narrator could have included the noun marker bayi in this NP,
but did not). The A NP in (44a) has the proper name of a man, in ergative case,
and a noun marker agreeing with this noun in case and in masculine noun
class. The coordinated clause, (44b), simply has a verb plus an A NP consisting
just of noun marker ba]gul, referring back to Jumbulu in (44a). The O argu-
ment is not stated in (44b). Dyirbal has an S/O syntactic pivot and if no O
argument is stated, this is inferred to be the same as the O argument in the
previous clause (if it was transitive) or to the S argument (if the preceding
clause was intransitive).
There are two ways of analysing an NP such as ba]gul in (44b). One is to
say that the noun marker is here in a function of substitution anaphora.
Another is to say that the A NP in (44b) has its head ellipsed, and it is this
ellipsis which has anaphoric function. It is difficult to chose between these
alternatives.
Now ba(la)- is the unmarked term of the system shown at (42), used when
specification of location and visibility are not important; it is ten times as
common as ya(la)- and ]a(la)- forms in texts. And a noun is generally cited
with a ba(la)- noun marker, in absolutive form, to show its noun class, e.g. bala
mu]ga (n) “noise”, bayi yara (m) “man”.
The full paradigm for ba(la)- noun markers is (Dixon 1972: 44):
(45)
Noun class Absolutive Ergative (A function) Dative Genitive
(S and O functions) and instrumental

masculine bayi ba-]gu-l ba-gu-l ba-]u-l

feminine bala-n, ba-n ba-]gu-n ba-gu-n ba-]u-n
edible bala-m, ba-m ba-]gu-m ba-gu-m –
neuter bala ba-]gu ba-gu ba-]u

The ]a(la)- paradigm is exactly the same, with initial ]- in place of b-. However,
the ya(la)- paradigm differs:
96 R. M. W. Dixon

(46)
Noun class Absolutive Ergative (A function) Dative Genitive
(S and O functions) and instrumental

masculine (yayi) ya-]gu-l ya-gu-l ya-]u-l

feminine (yala-n) ya-]gu-n ya-gu-n ya-]u-n
edible (yala-m) ya-]gu-m ya-gu-m –
neuter (yala) ya-]gu ya-gu ya-]u

Now the ba(la)-, ]a(la)- and ya(la)- noun markers refer to distance and
visibility but they do not qualify as demonstratives under the criterion followed
here — they are not used in deictic or pointing function.
There is a single nominal demonstrative in Dyirbal, which occurs only in
absolutive form. That is, it can only be used in an NP that is in S or O function.
The forms are:

(47) Noun class Absolutive (S and O functions)

masculine giyi
feminine ginya-n
edible ginya-m
neuter ginya

Now the reason for enclosing the absolutive ya(la)- forms in (46) in parentheses
can be explained. These are not used, the demonstrative forms from (47) being
employed instead. That is, the forms yayi/yalan/yalam/yala do not occur as
complete noun markers. However, they are found with certain affixes which
attach to the absolutive forms of nouns markers (but not of demonstratives),
e.g. yalan-ambila “with the feminine thing here”.
Dyirbal is a language with strong syntactic ergativity. As demonstrated in
(44a,b), two clauses can only be coordinated if they share an argument that is
in S or O function in each; mention of the argument can then be omitted from
the second clause. In keeping with this, if something is to be referred to deicti-
cally, then it must be expressed by an argument in S or O function.
In one story, a woman sees her son returning (from the dead) and utters
(48), which has the nominal demonstrative giyi in S function:
(48) []aygu giyi-n-dayi daman]S
[1sg:poss this:abs:m-catalyst-short.way.uphill son(m)
<LINK "dix-r22">

Demonstratives 97

]urba-yarra-nyu
return-start-non.fut
“This my son a short way uphill is starting to return.”

In (49) the narrator is describing some men looking for lost goods. The verb
guniy- “search” is transitive and the NP referring to the men is in underlying A
function. In order to use a demonstrative to point at the men, the clause must
be recast as an antipassive, with the “men” NP taking on derived S function.
(49) [giyi-m-bawal yara]S guni-marri-nyu
[this:abs:m-catalyst-way.out man(m) search-antipass-non.fut
“These men out there were searching,”

As is often the case, the underlying O NP (which would take dative or instru-
mental case in an antipassive construction) is omitted from (49).
Dyirbal has a number of other syntactic devices for putting a non-pivot
argument into S or O function, and these will be used if the argument is to be
expressed through a demonstrative. An instrumentive derivation, marked by
suffix -m(b)al- on the verb, takes an argument that was in underlying instru-
mental case and places it in derived O function. And a verb like “give” has three
possible case frames; in two the “gift” is in O function (and can be referred to
with a demonstrative) and in the other the “recipient” is in O function (and can
be referred to with a demonstrative). (Details are in Dixon 1972: 95–6, 300.)
As already stated, ba(la)- is the unmarked noun marker, and non-abso-
lutive forms of ba(la)- are ten times as common in texts as non-absolutive
forms of ya(la)- and ]a(la)-.
But when we compare absolutive forms of ba(la)- with the demonstrative
(only used in absolutive case), then ba(la)- forms are only twice as common as
ginya- forms.
Demonstratives are like noun markers in most morphological and syntactic
respects. Noun class is marked by a final suffix, zero in the case of the neuter class
(there is discussion of the semantics of the noun classes in Dixon 1972: 306–11).
The absolutive masculine form is irregular: bayi where noun marker bala-l
would be expected, and giyi where demonstrative ginya-l would be expected.
(An explanation for these irregularities is suggested in Dixon 2002: 466.)
The nominal demonstrative can make up a complete NP, or it can be used
with a noun or pronoun, just like noun markers. In examination of (44a,b) we saw
that a noun marker could be said to have substitution anaphora function. There
are no examples of the demonstrative functioning in this way. And neither noun
markers nor the demonstrative have textual anaphora, or any kind of cataphora.
<LINK "dix-r22">

98 R. M. W. Dixon

It was mentioned in §3.1 that in most languages demonstratives occur in an

NP with a noun or pronoun much more frequently than they make up an NP
on their own. We saw that in Yidinj, Dyirbal’s northerly neighbour, less than
10% of the textual occurrences of demonstratives are as full NPs. Dyirbal has
considerably more ellipsis than Yidinj (see Dixon 1977b); as a consequence,
around half of the textual occurrences of noun markers and of the demonstra-
tive in Dyirbal are in an NP together with a noun or pronoun, and about half
make up a full NP on their own.
The nominal demonstrative can occur in an NP with the ba(la)- noun
marker (here being used in its “unmarked”, non-spatial sense). One legend tells
of how a dreamtime person turned into the cassowary:
(50) [ginya-n-dawala ba-n]S nyalba-nyu
[this-f-medium.way.upstream there-f change-non.fut
“This one a short way upstream metamorphosed.”
Like noun markers, the demonstrative may appear in an NP with an interroga-
tive, or with a 1st or 2nd person pronoun, as in:
(51) []aja-riga giyi]S yinyja ba]ganda-ny
[1sg-certainly this:m still be.sick-present
“I’m certainly still sick.” (lit. “this I …”)
The demonstrative, or any noun marker, may optionally be followed by choices
from two sets of spatial indicators.12 Firstly, a member of the set:
(52)
-bayji short distance downhill -dayi short distance uphill
-bayja medium distance downhill -daya medium distance uphill
-bayju long distance downhill -dayu long distance uphill

-balbala medium distance downriver -dawala medium distance upriver

-balbulu long distance downriver -dawulu long distance upriver
-guya across the river
-bawal long way (in any direction)

This can optionally be followed by a member of the set:

(53) -gala up (vertically)
-gali down (vertically)
-galu out in front (with reference to the way either actor or speaker is
facing)
<LINK "dix-r22">

Demonstratives 99

The examples given above have included a number of forms from set (52); we
had -dayi “short distance uphill” in (48), -bawal “long way” in (49), and
-dawala “medium distance upstream” in (50). In the textual examples collected,
if demonstrative and noun marker appear together in an NP, a spatial specifica-
tion from set (52) or set (53) can be added just to the demonstrative — as in
(50) — or to both elements, e.g. ba-n-bayji ginya-n-bayji “there-f-
short.way.downhill this-f-short.way.downhill”
There is no formal similarity between noun markers and the nominal
demonstrative, on the one hand, and the main nominal interrogatives (wanya
“who”, minya “what”) on the other. However, there is a further set of nominal
demonstratives with root wunyja- “where” which has exactly the same para-
digm as noun markers. That is, wunyja- is simply substituted for ba(la)- in (45);
there is again an irregular masculine absolutive form wunyjiny “where is the
masculine thing”, rather than the expected wunyja-l. See Dixon (1972: 49).
In summary, Dyirbal has a set of noun markers (a type of determiner)
relating to location and visibility. However, these do not qualify as demonstrat-
ives since they are not used deictically. The only nominal form than can be used
with a pointing gesture is the demonstrative ginya-, and this may only be used
in a pivot function (S or O).

7.2 Local adverbial demonstratives

The paradigm of adverbial demonstratives in Dyirbal is set out in (54), together
with adverbial interrogatives, which have similar form. Note that, unlike the
nominal demonstratives, there is no marking for noun class.
(54)
Demonstratives Interrogative

“here” “there” “remembered” “where”

“to a place” yalu balu ]alu wunyjarru

(allative of place)
“in a direction” yali bali – wunyjarri
(allative of direction)
“from” (ablative) ya]um ba]um ]a]um wunyja]um
“at” (locative) yalay balay ]alay wunyjay
100 R. M. W. Dixon

All of the forms in (54), excepting the interrogatives, may be followed by a

suffix from (52) and/or one from (53), just like noun markers and the nominal
demonstrative.
An adverbial demonstrative is most often the only adverbial form in its clause.
However, it may occur with a noun, with which it agrees in local case, as in:
(55) a. balagaraS balay-balbulu nyinay-gu
3du there:loc-long.way.downstream stay-purposive
mija-]ga …
camp-loc 
“They two (woman travelled) to stay there at the camp …”
b. [daman-jarran ba]ugarra]u]O balay warrayma-n
[son-dual 3du:genitive there:loc find-non.fut
“Their two sons were encountered there.”

In (55a) balay-balbulu and mija]ga make up a single discontinuous local NP;

the adverbial demonstrative balay in (55b) refers back to balay-balbulu mija]ga.
It could be said that balay has a substitution anaphoric function in (55b).
Later in the same story we find:
(56) a. balaO ba-]gu-lA mijaO bura-n ba]ugarra]uO
there:n there-erg-m camp(n) see-non.fut 3du:genitive
“He saw the camp of the two of them.”
b. balay bayiS mijurri-nyu
there:loc there:m wait-non.fut
“He waited there.”

The O NP in (56a) consists of the first, third and fifth words of the clause (an
example of the extreme freedom of word order in Dyirbal), bala mija
ba]ugarra]u “the camp of the two of them”. The adverbial demonstrative balay
in (56b) refers back to the O argument of the preceding clause. This could be
regarded as an example of textual anaphora.
It was shown in §7.1 that Dyirbal has a single nominal demonstrative,
ginja-, used for pointing at objects. In translation of (25a/b), ginya- would be
used to render both this and that in English. Noun markers, based on ba(la)-,
ya(la)- and ]a(la)-, are not used deictically and do not qualify as demon-
stratives. However, the adverbial forms based on ba(la)-, ya(la)- and ]a(la)- —
as set out in (54) — are used deictically and must be regarded as demonstrat-
ives. In translating (26a/b) into Dyirbal, yalay would be likely to be used for
here (“put it here!”, with pointing) and balay for there (“put it there!”, with
<LINK "dix-r22">

Demonstratives 101

pointing). The ]a(la)- adverbs refer to something remembered from the past,
and can also have a deictic function. For instance, I was once on an expedition
with some elderly Dyirbal people to try to find a pigeon nest and eggs which
had, in legend, been turned into stone. The old people had seen these stones
many years before. As we set off they pointed up the track and said ]alu “to the
place remembered from the past”.

7.3 Verbal demonstratives

There are two dimensions of variation for verbs in Dyirbal — conjugation and
transitivity. Each verb belongs to either the y or the l conjugation; and each verb
is either intransitive or transitive. There is a correlation but no coincidence
between the parameters: most intransitive verbs belong to the y conjugation and
most transitive verbs to the l class.
There are two verbal demonstratives, transitive yalama-l (in the l conjuga-
tion) and intransitive yalama-y (in the y conjugation) “do like this”. Interest-
ingly, there are corresponding interrogative verbs, transitive wiyama-l and
intransitive wiyama-y; these mean “do what” when used alone and “do how”
when used with a lexical verb (see Dixon 1972: 55–6).
Three functions were described, in §3.3, for the verbal demonstrative in
Boumaa Fijian. The verbal demonstrative in Dyirbal appears to lack an ana-
phoric function, but can be used deictically, and to introduce direct speech. Its
deictic use is illustrated in (57)–(59).
In one story a dreamtime person has been strangling children. To put a stop
to this, the other people resolve to cut off his arms. They lull him to sleep, and
then stretch his arms out, preparatory to amputating them. The narrator says:
(57) [ginya-ginya garrgal]O [yuba-n yalama-n]PREDICATE
[this:n-this:n arm(n) [put.down-non.fut do.like.this-non.fut
“These arms were put down (stretched out) like this.” (narrator mimes
what was done)

Note that in (57) there is a nominal demonstrative in the O NP (reduplicated

to indicate plurality, that there were two arms) and a verbal demonstrative in
the predicate.
In a legend telling of the origin of water, the blue-tongue lizard is hiding away
the only water there is in the world. The other animals (who were dreamtime
people) ask the lizard where the water is but he denies having any. He chews spear
grass, for the moisture which it contains, and tells the others to do likewise:
<LINK "dix-r22">

102 R. M. W. Dixon

(58) balaO baja! ]ajaA [yalama-n baja-n]PREDICATE

there:n chew:imp 1sg [do.like.this-non.fut chew-non.fut
“Chew it [the spear grass]!, I’m chewing (it) like this.”
In (57) the verbal demonstrative referred to a gesture mimicking an action, but
in (58) is refers directly to the action of chewing spear grass.
These two examples have involved the transitive verbal demonstrative,
yalama-l. The intransitive version, yalama-y, is used in:
(59) bayiS yalama-nyu gu]garri yanu
there:m do.like.this-non.fut north go:non.fut
“He went this way to the north.” (pointing)
Here yalaman “did like this” occurs in the same predicate as yanu “went” (the
two words are separated in the clause, a further example of the free word order
in Dyirbal). Note that a verbal demonstrative may also be used by itself in a
predicate, to refer to a mimed or concomitant action. Further examples of the
deictic use of yalama- can be found in Dixon (1972: 56).
The verbal demonstratives are also used to introduce or conclude direct
speech (they can either precede or follow the direct speech). In this function
they should be used with a verb of speaking, such as intransitive wurrba-y “say”,
transitive buwa-y “tell” or transitive ]anba-l “ask”. At the beginning of one text,
the main storyteller instructs his companion on how he should assist him in the
narration. He uses a 3rd person imperative with a discontinuous verbal com-
plex, yalama … ]aril-gani:
(60) “wunyja wunyja bala?”, yalama yu]gugan-duA
“where:n where:n there:n do.like.this:imp another-erg
]aril-gani
reply-keep.on:imp
“‘Where’s that?’, another one should keep on replying like that.”
In practice, the verb of speaking is generally omitted in texts, with the verbal
demonstrative taking on the whole role of marking direct speech, as in:
(61) mu]a-gabun-duA yalama-n ·direct speechÒ
many-another-erg do.like.this-non.fut
“Another lot (of people) said:” ·direct speechÒ
Note that in (61) the direct speech effectively fills the O functional slot for the
transitive verb yalama-l. Interestingly, in elicitation, consultants would always
insist that an actual verb of speaking should be included with yalama- in
sentences like (61).
 Demonstratives 103

A verbal demonstrative may be used to refer to other kinds of noise. In one

legend, two women are up a tree, hiding from a man who is running along
below with an erection. The younger woman tries to to stop herself laughing
but eventually can’t help herself and bursts out in a peel of laughter:
(62) [ba-n midi]S mugu banda-banda-nyu,
[there-f small couldn’t.help.it redup-burst.out-non.fut
yalama-nyu [i, i, i]
do.like.this-non.fut ·noise of laughingÒ
“The small (woman) couldn’t help bursting out, (laughing) like this: ‘i, i, i’.”

In conclusion, the functions of demonstratives and related items in Dyirbal can

be summarised:

(63) Nominal Adverbial Verbal

Demonstratives ginya- forms ba(la)-, ya(la)- yalama- forms

and ]a(la)-
forms

Not demonstrat- ba(la)-, ya(la)-

ives and ]a(la)-
noun markers

We see that forms based on yala- are found in all three columns. The adverbial
and verbal yala- forms do have deictic function and are demonstratives, but the
nominal ya(la)- forms (like the nominal ba(la)- and ]a(la)- forms) cannot be
used for pointing at an object, and do not qualify as demonstratives.

8. Conclusions

We identified three main types of demonstrative:

– Nominal demonstratives (such as “this” and “that”) may always modify a
head noun (and sometimes also a pronoun) in an NP. In some — but not
all — languages they may also make up a full NP.
– Local adverbial demonstratives (such as “here” and “there”) modify a verb,
sometimes occurring together with a NP in local function.
– Verbal demonstratives (“do it like this”) may modify a verb, or function as
the only verbal element in a clause.
104 R. M. W. Dixon

A form is identified as a demonstrative if it can have deictic (or pointing)

reference. Some, but not all, demonstratives may also have anaphoric and/or
cataphoric function.
Further functions may include identification, presentation of new informa-
tion, and marking discourse organisation. There can be constraints on the
syntactic function of a nominal demonstrative (for example, in Dyirbal,
Warrgamay and Northern Subanen it may only occur in S or O function).
One language (Ju|’hoan) is reported just to have two verbal demonstratives.
All others have at least one nominal and at least two local adverbal demonstrat-
ives. Most demonstrative systems have two terms, but some have three or more.
A three term system may be in terms of relative distance (near, mid, far) or it
may relate to the participants (near speaker, near addressee, near neither).
There may also be reference to height and/or stance and/or visibility.
Markedness in demonstrative systems is in need of further detailed studies.
We saw that, in English, this appears to be the unmarked term in the system of
nominal demonstratives for deictic reference, but that is the unmarked term for
anaphoric reference.

9. Questions to be addressed when investigating demonstratives in a

language

The following are among the main questions that should be confronted when
examining which items in a language should be considered demonstratives, and
what their properties are. These will be of use to field workers, starting work on
a previously undescribed language, and also to scholars writing grammars of
better-known languages.13
A.
Does the language have (a) nominal; (b) local adverbial; (c) verbal types of
demonstrative? Are there any other types of demonstratives?
– What word classes do the various types of demonstratives belong to?
– What are the similarities in form between the types of demonstrative?
B.
Does the language have (i) definite article; (ii) 3rd person pronouns; (iii)
other related items?
– Are there formal and functional similarities between these items and
demonstratives?
– Are there formal and functional similarities between interrogatives and
demonstratives?
 Demonstratives 105

C.
What are the functions of each of the types of demonstrative?

1. Describe the deictic function.
2. Describe the syntactic functions.
– For nominal demonstratives, can they occur (i) in an NP with a noun;
(ii) in an NP with a 1st or 2nd person pronoun; (iii) in an NP with a
3rd person pronoun; (iv) in an NP with the definite article; (v) in an
NP with an indefinite word; (vi) making up a full NP?
– Can an NP including a nominal demonstrative occur in any syntactic
function in a clause, or are there restrictions?
– Can local adverbial and verbal demonstratives occur in all clause types,
or are there restrictions?
3. For each of the types of demonstrative, can it be used for (i) substitution
anaphora/cataphora, (ii) textual anaphora/cataphora? How do they contrast
with 3rd person pronouns, and with NPs including the definite article, in
these functions?
4. Can a clause include more than one type of demonstrative? Give details.
5. Any other functions? See the (far from exhaustive) list in §5.4.
D.
Describe the parameters of reference for each type of demonstrative.
– How many nominal demonstratives are there? If more than one, are they
distinguished in terms of distance, height, stance, visibility, etc? Provide full
details.
– If there are one or more “non-visible” terms, describe their reference (e.g.
something that was visible but has just gone out of sight, something far and
non-visible, something audible, something remembered). Does “visibility”
relate to the speaker or to the addressee (or both)?
– Are there further senses to demonstratives, such as temporal reference,
emotional attitude, familiarity, personal interest?
– Describe how categories such as gender/noun class, classifiers, number and
case are coded in demonstratives (and any differences from the ways they
are coded on other types of words).
– Is it possible to recognise one term in a demonstrative system as (formally
and/or functionally) “marked” with respect to the others?
<DEST "dix-r21">
<LINK "dix-r23">
"dix-r24">
"dix-n*">

106 R. M. W. Dixon

Notes

*
I owe a debt to all participants in a Workshop on Demonstratives and Related Items, held
in the Research Centre for Linguistic Typology at La Trobe University between February and
June 2001, and also to Alexandra Aikhenvald, D. N. S. Bhat, Paul Geraghty and Kevin Tuite
for ideas and for responses to my questions.
1. The following abbreviations are used in this paper: a, transitive subject function; abs,
absolutive case; acc, accusative case; antipass, antipassive; applic, applicative; art, article;
aux, auxiliary; cl, classifier; dec, declarative; du, dual; erg, ergative; ext, extended in time;
f, feminine; fut, future; imp, imperative; inc, inclusive; loc, locative; m, masculine; o,
transitive object function; n, neuter; nsg, non-singular; pl, plural; poss, possessor; s,
intransitive subject function; sg, singular; vis, visible.
2. Diessel (1999: 57) employs the term “pronominal demonstrative” in place of the tradition-
al “demonstrative pronoun”; this is equally unsuitable. He uses the term “adnominal
demonstrative” in place of “demonstrative adjective”, which is an improvement.
3. Emeneau (1961) describes how in the Dravidian language Brahui, the original “near”
demonstrative base has developed into the base of an enclitic pronoun.
4. Diessel (1999: 115–55) provides a discussion of the diachronic development of demon-
stratives.
5. Table 62 in Diessel (1999: 103) presents results similar to those given here in Table 1 for
English. Diessel’s examples are from English but he appears to assume, gratuitously, that
these results apply cross-linguistically. As shown in Table 3 below, this is not the case.
6. French also has special forms voici and voilà which can be used in place of ceci est “this is”
and cela est “that is”, respectively (ceci and cela are alternatives to celui-ci and celui-là); they
have a slightly different grammar and a distinctive stylistic effect (Ferrar 1967: 214).
7. For example, “it is here (by me)” can be rendered by e tiko e kee (“it be at here”) or by e
tiko qoo (“it be here”); the first clause involves kee “here” which is used only after preposi-
tions (such as e “at”) and the second involves qoo “this, here”, which is used both as a
nominal demonstrative and as a local adverbal when there is no preposition included. The
two clauses appear to have the same meaning (Paul Geraghty, p.c.).
8. Diessel (1999: 78–88) recognises a special category of “demonstrative identifiers” — . after
translatable as “here/there it is” — with examples from a range of languages.
9. Diessel (1999: 151–2) provides a useful list of references to the literature on sound
symbolism in demonstratives.
10. It does seem that there is some sort of inverse correlation between size of demonstrative
system and size of language community. The most complex systems are found in languages
with a relatively small number of speakers, while languages spoken by a very large number of
people tend to have just a two-term system. For example, within the Dravidian language
family, demonstrative systems with three or four members are found in the small tribal
languages whereas each of the four languages with tens of millions of speakers (Telugu,
Tamil, Malayalam and Kannada) show just two terms. At an earlier stage, English had a
three-term system (this/here, that/there and yon/yonder) but the third term was lost as English
<LINK "dix-r40">
<DEST "dix-r36">
"dix-r21">
"dix-r22">
"dix-r12">
"dix-r71">
"dix-r1">
"dix-r2">
"dix-r3">
"dix-r4">
"dix-r5">
"dix-r6">
"dix-r7">

Demonstratives 107

developed into a world language. Spanish and Portuguese are said to have three or four
demonstratives but in daily usage there is just a two-term contrast (see, for example,
Jungbluth 2001). (I am grateful to D.N.S. Bhat for discussion of this matter, and for drawing
attention to the Dravidian data.) (See also Diessel 1999: 160–1 and references provided there
to the relevant literature.)
11. The discussion of noun markers and demonstratives in Dixon (1972:44–7) is not so clear
as it should have been, and has misled some scholars. There I placed the ginya- forms in the
“nominative” (now called “absolutive”) column of the yala- paradigm. It is better (as here)
to recognise separate paradigms for ginya- forms (which are demonstratives) and for yala-
forms (which are not), with the ginya- forms suppressing the absolutive yala- forms in most
— but not all — contexts.
Fuller details, including the retention or omission of -la-, will be found in Dixon
(1972: 254–62).
12. Diessel (1999: 43–4) implies that these specifications are demonstratives. They are not;
the two sets of suffixes can be added to demonstratives, to noun markers, and also to nouns
(Dixon 1972: 232).
13. An earlier set of questions concerning demonstratives is in Comrie and Smith
(1977: 44–6). Wilkins (1999) and Levinson (1999) are useful documents but appear to rely
on elicitation. The study of demonstratives, as of any other aspect of a language, can only
properly be conducted through analysis of texts and by participant observation, with at most
minor augmentation through direct elicitation.

References

Aikhenvald, Alexandra Y. 2000. Classifiers: a typology of noun classification devices. Oxford:

Oxford University Press.
Aikhenvald, Alexandra Y. Forthcoming. The Tariana language of north-west Amazonia.
Cambridge: Cambridge University Press.
Aikhenvald, Alexandra Y. and Green, Diana. 1998. “Palikur and the typology of classifiers.”
Anthropological Linguistics 40: 429–80.
Anderson, Stephen R. and Keenan, Edward L. 1985. “Deixis.” In Shopen, Timothy (ed.),
Language typology and syntactic description, Vol 3, 259–308. Cambridge: Cambridge
University Press.
Asher, R. E. 1985. Tamil. London: Croom Helm.
Austin, Peter. 1981. A grammar of Diyari, South Australia. Cambridge: Cambridge University
Press.
Boas, Franz. 1911. “Introduction.” In: Boas, Franz (ed.), Handbook of American Indian
languages, Vol 1, 1–83. Washington: Smithsonian Institution.
Bugenhagen, Robert D. 1994. “The exponents of semantic primitives in Mangap-Mbula.” In:
Goddard and Wierzbicka 87–108.
Bühler, Karl. 1934. Sprachtheorie. Jena: Fischer. [1990. Theory of language: the representation-
al function of language, translated by Donald F. Goodwin. Amsterdam: John Benjamins.]
<DEST "dix-r27">
"dix-r8">
"dix-r9">
"dix-r10">
"dix-r11">
"dix-r12">
"dix-r13">
"dix-r14">
"dix-r15">
"dix-r16">
"dix-r17">
"dix-r18">
"dix-r19">
"dix-r20">
"dix-r21">
"dix-r22">
"dix-r23">
"dix-r24">
"dix-r25">
"dix-r26">

108 R. M. W. Dixon

Carlson, Robert. 1994. A grammar of Supyire. Berlin: Mouton de Gruyter.

Ceria, Verónica G. and Sandalo, Filomena. 1995. “A preliminary reconstruction of Proto-
Waikurúan with special reference to pronominals and demonstratives.” Anthropological
Linguistics 37: 169–91.
Chappell, Hilary. 2001. “Mandarin demonstratives.” RCLT internal document.
Churchward, C. Maxwell. 1941. A new Fijian grammar. Sydney: Australasian Medical
Publishing Company.
Comrie, Bernard and Smith, Norval. 1977. “Lingua descriptive series: questionnaire.” Lingua
42: 1–72.
Cook, Eng-Do. 1984. A Sarcee grammar. Vancouver: University of British Columbia Press.
Coulmas, Florian. 1982. “Some remarks on Japanese deictics.” In: Weissenborn and Klein
209–21.
Cummins, George M, III. 1998. “Definiteness in Czech.” Studies in language 22: 567–96.
Curnow, Timothy J. 1997. A grammar of Awa Pit (Cuaiquer), an indigenous language of
south-western Colombia. PhD thesis, ANU.
Daguman, Josephine S. 2001. “Demonstratives in Northern Subanen.” RCLT internal
document.
Dahlstrom, Amy. Mss. “Grammar of Fox.”
Derbyshire, Desmond C. 1985. Hixkaryana and linguistic typology. Dallas: Summer Institute
of Linguistics.
Dickens, Patrick. n.d. “Ju|’hoan grammar.” Typescript.
Diessel, Holger. 1999. Demonstratives: form, function and grammaticalization. Amsterdam:
John Benjamins.
Dixon, R. M. W. 1972. The Dyirbal language of North Queensland. Cambridge: Cambridge
University Press.
Dixon, R. M. W. 1977a. A grammar of Yidi\. Cambridge: Cambridge University Press.
Dixon, R. M. W. 1977b. “The syntactic development of Australian languages.” In: Li, Charles
N. (ed.), Mechanisms of syntactic change 365–415. Austin: University of Texas Press.
Dixon, R. M. W. 1981. “Wargamay.” In: Dixon, R. M. W. and Blake, Barry J. (eds.), Handbook
of Australian languages, Vol 2, 1–144. Canberra: ANU Press and Amsterdam: John
Benjamins.
Dixon, R. M. W. 1988. A grammar of Boumaa Fijian. Chicago: University of Chicago Press.
Dixon, R. M. W. 2002. Australian languages: their nature and development. Cambridge:
Cambridge University Press.
Emeneau, M. B. 1961. “Brahui demonstrative pronouns.” Journal of the Asiatic Society
(Bengal) 3: 1–5. Reprinted as pp 17–22 of M. B. Emeneau. 1994. Dravidian studies:
selected papers. Delhi: Motilal Banarsidass.
Ferrar, H. 1967. A French reference grammar. Second edition. Oxford: Oxford University
Press.
Frajzyngier, Zygmunt. 1993. A grammar of Mupun. Berlin: Dietrich Reimer.
Goddard, Cliff and Wierzbicka, Anna (eds.), 1994. Semantic and lexical universals: theory and
empirical findings. Amsterdam: John Benjamins.
Gragg, Gene. 1976.”Oromo of Wellegga.” In: Bender, M. Lionel (ed.), The non-Semitic
languages of Ethiopia 166–95. East Lancing: African Studies Center, Michigan State
University.
<DEST "dix-r50">
"dix-r28">
"dix-r29">
"dix-r30">
"dix-r31">
"dix-r32">
"dix-r33">
"dix-r34">
"dix-r35">
"dix-r36">
"dix-r37">
"dix-r38">
"dix-r39">
"dix-r40">
"dix-r41">
"dix-r42">
"dix-r43">
"dix-r44">
"dix-r45">
"dix-r46">
"dix-r47">
"dix-r48">
"dix-r49">

Demonstratives 109

Haiman, John. 1980. Hua: a Papuan language of the Eastern Highlands of New Guinea.
Amsterdam: John Benjamins.
Hayward, Richard J. 1990. “Notes on the Zayse language.” In: Hayward, Richard J. (ed.),
Omotic language studies 210–355. London: School of Oriental and African Studies.
Heine, Bernd and Kuteva, Tania. 2002. World lexicon of grammaticalization. Cambridge:
Cambridge University Press.
Hill, Deborah. 1992. Longgu grammar. PhD thesis, ANU.
Himmelmann, Nikolaus P. 1996. “Demonstratives in narrative discourse: a taxonomy of
universal uses.” In: Fox, Barbara (ed.), Studies in anaphora 205–54. Amsterdam: John
Benjamins.
Imedadze, Natela and Tuite, Kevin. 1992. “The acquisition of Georgian.” In: Slobin, Dan I.
(ed.), The crosslinguistic study of language acquisition, Vol 3, 39–109. Hillsdale, N. J.:
Lawrence Erlbaum.
Jacob, Judith M. 1968. Introduction to Cambodian. London: Oxford University Press.
Jespersen, Otto. 1933. Essentials of English grammar. London: Allen and Unwin.
Jungbluth, Konstanze. 2001. “Binary and ternary deictic systems in speech and writing.
Evidence from the use of demonstratives in Spanish.” Philologie im Netz 15: 1–24.
Khaidakov, S. M. 1966. “The dialect divisions of Lak.” Studia Caucasica 2: 9–18.
Kimball, Geoffrey D. 1991. Koasati grammar. Lincoln: University of Nebraska Press.
Krishnamurti, Bh. 2001. “Telugu demonstratives.” RCLT internal document.
Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.
Levinson, Stephen C. 1999. “Deictics and demonstratives: field guide for 1998”. In: Wilkins,
David (ed.), “Manual” for the 1999 field season 29–40. Nijmegen: Language and
Cognition Group, Max Planck Institute.
Li Xinyin. 1996. “Chinese: grammar summary.” RCLT internal document.
Lindström, Eva. 2000. “Some uses of demonstratives in spoken Swedish.” In: Botley, Simon
and McEnery, Anthony N. (eds.), Corpus-based and computational approaches to
discourse anaphora 107–28. Amsterdam: John Benjamins.
Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University
Press.
Lyons, John. 1977. Semantics, 2 vols. Cambridge: Cambridge University Press.
Malotki, Eckehart. 1982. “Hopi person deixis.” In: Weissenborn and Klein 223–52.
Matisoff, James A. 1973. The grammar of Lahu. Berkeley and Los Angeles: University of
California Press.
Meyerstein, Zlata P. 1972. “Czech deictics: pronouns or articles?” Linguistics 91: 17–30.
Miller, Wick R. 1996. “Sketch of Shoshone, a Uto-Aztecan language.” In: Goddard, Ives
(ed.), Handbook of North American Indians, Vol 17, Languages 693–720. Washington:
Smithsonian Institution.
Mithun, Marianne. 1987. “The grammatical nature and discourse power of demonstratives.”
Berkeley Linguistics Society Proceedings 13: 184–94.
Mithun, Marianne. 1999. The languages of native North America. Cambridge: Cambridge
University Press.
Mous, Marten. 1993. A grammar of Iraqw. Hamburg: Helmut Buske.
Newman, Paul. 2000. The Hausa language: an encyclopedic reference grammar. New Haven:
Yale University Press.
<DEST "dix-r70">
"dix-r51">
"dix-r52">
"dix-r53">
"dix-r54">
"dix-r55">
"dix-r56">
"dix-r57">
"dix-r58">
"dix-r59">
"dix-r60">
"dix-r61">
"dix-r62">
"dix-r63">
"dix-r64">
"dix-r65">
"dix-r66">
"dix-r67">
"dix-r68">
"dix-r69">

110 R. M. W. Dixon

Noonan, Michael. 1992. A grammar of Lango. Berlin: Mouton de Gruyter.

Onishi, Masayuki. 1997. “Bengali: grammar summary.” RCLT internal document.
Owens, Jonathan. 1985. A grammar of Harar Oromo (Northeastern Ethiopia). Hamburg:
Helmut Buske.
Payne, Thomas E. and Payne, Doris L. 1990. “Yagua.” In: Derbyshire, Desmond C. and
Pullum, Geoffrey K. (eds.), Handbook of Amazonian Languages, Vol 2, 249–474. Berlin:
Mouton de Gruyter.
Payne, Thomas E. and Payne, Doris L. 1999. “Panare: grammar summary.” RCLT internal
document.
Pope, M. K. 1934. From Latin to Modern French. Manchester: Manchester University Press.
Quirk, Randolph and Greenbaum, Sidney. 1973. A university grammar of English. London:
Longman.
Quirk, Randolph; Greenbaum, Sidney; Leech, Geoffrey; and Svartvik, Jan. 1985. A compre-
hensive grammar of the English language. London: Longman.
Rankin, Robert L. 1976. “From verb to auxiliary to noun classifier and definite article:
grammaticalization of the Siouan verbs ‘sit’, ‘stand’, ‘lie’.” In: Brown, R. L., Houlihan,
K., Hutchinson, L. and Macleish, A. (eds.), Proceedings of the 1976 Mid-America
linguistics conference 273–83. Minneapolis: University of Minnesota.
Rankin, Robert L. To appear. “The history and development of Siouan positionals, with
special attention to polygrammaticalization in Dhegiha.” In Aikhenvald, Alexandra Y.
(ed.), special issue on “Nominal classification.” of Sprachtypologie und Universalien-
forschung.
Refsing, Kirsten. 1986. The Ainu language: the morphology and syntax of the Shizunai dialect.
Aarhus: Aarhus University Press.
Rehg, Kenneth L. 1981. Ponapean reference grammar. Honolulu: University of Hawaii Press.
Saltarelli, Mario. 1988. Basque. London: Croom Helm.
Sasse, Hans-Jürgen. 1999. “Cayuga: grammar summary.” RCLT internal document.
Sneddon, James N. 1996. Indonesian reference grammar. Sydney: Allen and Unwin.
Stumme, Hans. 1899. Handbuch des Schilhischen von Tazerwalt. Leipzig: J. C. Hinrichs.
Sweet, Henry. 1898. A short historical English grammar. London: Oxford University Press.
Tamura, Suzuko. 2000. The Ainu language. Tokyo: Sanseido. [English translation of article
“Ainu-go” (The Ainu language) from Gengogaku-dai-jiten (The Sanseido Encyclopaedia
of Linguistics), Vol 1, Part 1, 1988.]
van Eijk, Jan. 1997. The Lillooet language: phonology, morphology, syntax. Vancouver: UBC
Press.
van den Berg. René. 1997. “Spatial deixis in Muna (Sulawesi).” In Senft, Gunter (ed.),
Referring to space: studies in Austronesian and Papuan languages 197–220. Oxford:
Clarendon Press.
Weber, David J. 1989. A grammar of Huallaga (Huánuco) Quechua. Berkeley and Los
Angeles: University of California Press.
Weissenborn, Jürgen and Klein, Wolfgang (eds.), 1982. Here and there: cross-linguistic studies
in deixis and demonstration. Amsterdam: John Benjamins.
Wierzbicka, Anna. 1980. Lingua mentalis: the semantics of natural language. Sydney:
Academic Press.
<DEST "dix-r74">
"dix-r71">
"dix-r72">
"dix-r73">

Demonstratives 111

Wierzbicka, Anna. 1994. “Semantic primitives across languages: a critical review.” In:
Goddard and Wierzbicka, 445–90.
Wilkins, David. 1999. “1999 demonstratives questionnaire: ‘THIS’ and ‘THAT’ in compara-
tive perspective.” In: Wilkins, David (ed.), “Manual” for the 1999 field season 1–24.
Nijmegen: Language and Cognition Group, Max Planck Institute.
Willett, Thomas L. 1991. A reference grammar of Southeastern Tepehuan. Dallas: Summer
Institute of Linguistics and University of Texas at Arlington.
Zandvoort, R. W. 1975. A handbook of English grammar. London: Longmans.
Zeshan, Ulrike. 1999. “Indo-Pakistan Sign Language: grammar summary.” RCLT internal
document.

Appendix — Anaphoric and cataphoric functions of demonstratives

in English

Jespersen (1933: 158, apparently referring to textual anaphora/cataphora, states: “this often
refers to what is following, and that to what precedes”. Quirk and Greenbaum (1973:302–3),
again apparently referring to textual anaphora/cataphora, state that this can be either
anaphoric or cataphoric, while that is only anaphoric. This is repeated in Quirk, Greenbaum,
Leech and Svartvik (1985: 375–6) with an added note that “in very limited contexts, e.g. in
expressions of indignation, that can be used cataphorically.”
In fact, as shown in Table 1 (in §3.1), both this and that can be used for substitution
anaphora, textual anaphora and textual cataphora; I have not found any examples of
substitution cataphora. In some sentences either of the nominal demonstratives is acceptable,
in others only one. I use the following convention: “this/that” indicates that they are equally
acceptable, and “this/*that” indicates that only this is acceptable.
(a) Substitution anaphora
THIS/THAT
(64) He asked for tonic water, insisting that this/that (drink) was the best remedy against
insomnia
(65) Cheddar or Brie or Stilton, that/this is the full set of cheeses available
THAT/*THIS — example (10) in §3.1, and
(66) The chocolates Mary sent are not so nice as those/*these she sent last year
(b) Textual anaphora
THIS/THAT — example (3) in §2, and
(67) John drank and drove, and that/this annoyed Mary
THIS/*THAT
(68) She thought over what he had said. “Why is he telling me all this/*that?”, she
wondered
THAT/*THIS
(69) I like Bach, Brahms and Debussy, in that/*this order
</TARGET "dix">

112 R. M. W. Dixon

(70) A fool and his money are soon parted, have you ever heard that/*this saying?
(c) Textual cataphora
THIS/THAT
(71) What do you think of this/that! My aunt has just died and left me all her money
(/nothing)!
THIS/*THAT — example (5) in §2 and
(72) Have you heard this/*that saying: a fool and his money are soon parted?
It appears that when referring to a preceding/following list or saying quoted verbatim, this/
these must be used for cataphora, as in (5) and (72) and that/those for anaphora, as in
(69–70). (This is consistent with Jespersen’s observation.)
Further, detailed work is needed on why both nominal demonstratives are acceptable in
some contexts and only one in others.

Author’s address
R. M. W. Dixon
Research Centre for Linguistic Typology
La Trobe University
Victoria, 3086
Australia