The distinction between an inclusive and an exclusive pronoun is a commonly attested feature of linguistic structure, yet from a Eurocentric point of view this distinction is particularly exotic. For speakers of English (or any other European language), both the inclusive and the exclusive pronouns are to be translated as we. The difference between the two depends on the intended meaning. An inclusive pronoun necessarily includes reference to the addressee. For example, the Mandarin inclusive pronoun zámen means ‘we, I and you’; others can optionally be included. An exclusive pronoun, like the Mandarin pronoun wŏmen, excludes the addressee from the reference, resulting in a meaning like ‘I and some others, but not you’.
This distinction between inclusive and exclusive is not found in any European language, nor in the languages in its wider surrounding. Because of its absence, this possibility of linguistic structure was not part of any classical linguistic analysis. Its first description dates back to the 16th century. The discovery was made by the Dominican friar Domingo de Santo Tomás, as described in his grammar of Quechua, the language of the Incas, first published in 1560. Today, the most widely spoken languages that have this distinction are found among Austronesian languages (in particular Malay and Javanese), among Dravidian languages (in particular Tamil and Telugu) and among northern varieties of Chinese.
Map 39A shows the distribution of inclusive/exclusive marking in independent pronouns. The basic distinction is between dots that are coloured red, which mark those languages that have such a distinction, and dots that are coloured blue, which mark those languages that do not have an inclusive/exclusive distinction. However, the actual typology for this map is finer grained: five different types of linguistic structure are distinguished. In the following description of these types, I will use the shorthand ‘we’ for a category that subsumes the inclusive and exclusive meanings (as in the English pronoun we).
| Go to map | ||
|---|---|---|
| Value | Representation | |
 |
No grammaticalised marking at all | 2 |
 |
'We' and 'I' identical | 10 |
 |
No inclusive/exclusive opposition | 120 |
 |
Only inclusive differentiated | 5 |
 |
Inclusive and exclusive differentiated | 63 |
| Total: | 200 | |
Some languages simply have no basic way of expressing any ‘we’-like concept. Of course, it is possible in all languages to express such meanings, but sometimes there is no specialized means for doing so. For example, in Pirahã (Mura; Amazonas, Brazil) the only way to express a ‘we’-like meaning is by a conjunction of the pronouns for ‘I’ and ‘you’.
(1) Pirahã (Everett 1986: 281)
ti
gíxai
pí-o
ahápií
1.pron
2.pron
also-obl
go
‘You and I will go.’
Some languages have a pronoun for expressing ‘we’, yet this pronoun is the same pronoun as is used for expressing ‘I’. For example, the following sentence from Qawasqar (Alacalufan; Chile) can mean either ‘I ran yesterday’ or ‘we ran yesterday’. There is no way to decide from this sentence alone which meaning is intended. The context must be used to disambiguate this sentence.
(2) Qawasqar (Clairis 1985: 201)
cecaw
qjeq’ja
qjenaq
afxat
1.pron
run
?
pst
‘We ran yesterday.’ or ‘I ran yesterday.’
In some languages, the pronoun for ‘I’ can be used for ‘we’ (as in Qawasqar), but this usage is extremely uncommon. For example, in Maricopa (Yuman; Arizona) there are no specialized plural pronouns. The existing pronouns "typically […] refer to singular entities. Eliciting overtly plural-marked pronouns is difficult and they appear to be used extremely infrequently" (Gordon 1986: 58). Languages of this type have no inclusive/exclusive distinction, so they are marked blue. However, they are marked light blue because they resemble the previous type with no marking at all (marked white).
This type is well known, as it is the type to which English and many other commonly known languages belong. In this type, there is one pronoun like English I and a different pronoun like English we, but there is no inclusive/exclusive opposition. Languages of this type are marked in dark blue.
Also included in this type are languages that distinguish a dual without an inclusive/exclusive distinction, for instance Hmong Njua (Hmong-Mien; China and Vietnam; Harriehausen 1990: 124). In Hmong Njua, there are two pronouns for ‘we’, but the distinction is not inclusive/exclusive but dual/plural. The pronoun wb is used for dual reference (precisely two persons) and peb is used for plural reference (more than two persons).
This type consists of languages that have a special pronoun for the inclusive, but the marking of the exclusive is identical to ‘I’. Such a structure is exemplified in (3) by Canela-Krahô (Ge-Kaingang; Brazil). In (3a), the pronoun wa is used in the meaning ‘I’. This same sentence could also be used for the exclusive meaning ‘we (I and some others, but not you)’. However, to express the inclusive meaning ‘we, you included’, a different pronoun cu must be used, as shown in (3b).
(3) Canela-Krahô (Popjes and Popjes 1986: 175-176)
a.
wa
po
pupu
1.pron
deer
see
‘I see a deer.’
b.
ha
cu
jê
ne
po
nõ
cura
hey
incl
relative
and
deer
art
kill
‘Hey, relative, let’s go and kill a deer.’
Languages of this type have an inclusive/exclusive distinction which, however, is not fully differentiated lexically, so they are marked light red on the map. There is a clear asymmetry in the structure of the world’s languages in that this combination exclusive + ‘I’ does exist (though it is rare), but that the combination inclusive + ‘I’ does not exist.
The final type distinguished on this map consists of those languages with specialized pronouns for both inclusive and exclusive reference. This is found, for instance, in Chamorro (Austronesian; Guam). The pronoun for ‘I’ is hu, the inclusive pronoun is ta and the exclusive pronoun is in (Topping 1973: 106-108). This type is marked dark red on the map.
There are many languages included in this type that also mark dual number in their pronouns. The basic and most common way to mark duality is exemplified in (4) by the pronouns from Lavukaleve (Solomons East Papuan; Solomon Islands). A special dual pronoun exists both for the inclusive and for the exclusive, both marked by a suffix -l.
(4) Lavukaleve (Terrill 2003: 170)
ngai
‘I’
el
‘exclusive, exactly two’
e
‘exclusive, more than two’
mel
‘inclusive, exactly two’
me
‘inclusive, more than two’
Another strategy is to mark duality only in the inclusive (Plank 1996: 130-131). In such paradigms, the dual inclusive aligns structurally with the singular pronouns, yet strictly speaking it is of course not singular in reference. The term minimal inclusive is used to refer to such a dual inclusive. Paradigms with a dual only in the inclusive are known as minimal-augmented structures (Thomas 1955). This lexical structure is exemplified in (5) by the pronouns from Southern Sierra Miwok (Penutian; California).
(5) Southern Sierra Miwok (Broadbent 1964: 93)
kan·i
‘I’
mah·i
‘exclusive, two or more’
ʔoti·me
‘inclusive, exactly two’
ʔotic·i
‘inclusive, more than two’
The opposite distribution of duality – dual in the exclusive but not in the inclusive – exists among the world’s languages, but it is extremely rare (Cysouw 2003: 221-222). In the present sample it is attested in the pronouns from Yagua (Peba-Yaguan; Peru; Payne and Payne 1990: 369-370). Just as exotic is the division attested in Gooniyandi (Bunaban; Australia). Here the inclusive dual is expressed by the same pronoun as the exclusive ngidi, but is different from the inclusive plural yaadi (McGregor 1990: 167-173). There are more cases like this among the world’s language, but not many (Cysouw 2003: 93).
Finally, some languages with trial (‘exactly three’) or paucal (‘a few’) marking in the inclusive and the exclusive are also included in this type. Trial or paucal marking only occurs among Austronesian languages, but within this linguistic stock it is rather widespread. For example, in the present sample it is found in Paamese (Oceanic; Vanuatu; Crowley 1982: 80).
A special variant of trial marking are pronouns that only distinguish a trial in the inclusive, but not in the exclusive. This is analysed as an extension of the minimal-augmented pattern in (5), adding a category in between ‘minimal’ and ‘augmented’ known as unit-augmented (McKay 1978). This structure is attested almost exclusively in northern Australia (cf. Cysouw 2003: 232-236). In the present sample this type is attested in Mangarrayi (Merlan 1982: 102).
In general, the inclusive/exclusive distinction is rather uncommon in Africa and Eurasia. There is no distinction in any language in Europe and its wider surroundings. The nearest cases are a few languages in the Caucasus. The distinction is also relatively uncommon in Africa; only a few sub-Saharan languages show the distinction. In Asia, the Dravidian and the Munda languages have an inclusive/exclusive distinction, although the Dravidian language Kannada in the present sample has lost the distinction under the influence of the neighbouring Indo-Aryan languages. In northeast Asia there might be an area with an inclusive/exclusive distinction, as exemplified here by Evenki, Ainu and Nivkh. Originally part of this cluster, Khalkha (Mongolian) has lost the distinction, while the northern dialects of Chinese have acquired the distinction. There is an interesting little cluster of languages that do not differentiate ‘I’ from ‘we’ in Southeast Asia. It is rather common in this area for languages not to mark number at all, or to mark it only optionally (cf. Chapter 34).
Off the Asian mainland, the inclusive/exclusive distinction is regularly attested. This is mainly due to the Austronesian languages and the non-Pama-Nyungan languages of northern Australia. In both groups, almost all languages have the distinction. In contrast, it is rather uncommon (and indeed absent in the present sample) among the non-Austronesian (‘Papuan’) languages of New Guinea. Among the Pama-Nyungan languages of Australia the distinction is roughly evenly divided (see next section).
In the Americas, there are about as many languages with an inclusive/exclusive distinction as without it. There seems to be no clear areal division between the two types here. Interestingly, the minor types (lightly coloured) are relatively common throughout the Americas.
There are no clear worldwide patterns in the distribution of the inclusive/exclusive opposition. The few patterns that are attested appear to be on a smaller scale. To exemplify the smaller scale, I carried out an in-depth investigation of the areal distribution of the inclusive/exclusive opposition among the Pama-Nyungan languages from Australia, as presented in Map 39B.
| Go to map | ||
|---|---|---|
| Value | Representation | |
|
No inclusive/exclusive opposition | 31 |
|
Inclusive and exclusive differentiated | 40 |
| Total: | 71 | |
The Pama-Nyungan stock covers Australia almost completely. Only in northern Australia are there languages which are not part of this stock, commonly called non-Pama-Nyungan (these languages are not included in this map). The Pama-Nyungan languages originally did not have an inclusive/exclusive opposition, although currently many Pama-Nyungan languages have developed it (Dixon 1980: 334-336).
Concerning the distribution of the inclusive/exclusive distinction among the Pama-Nyungan languages, Dixon (1980: 335) claimed that “there is no regularity to the distribution – languages of both types are found in every quarter of the continent.” However, as can be seen from the distribution in Map 39B, there are clear areas with languages that have an inclusive/exclusive distinction and areas without. First, and most prominent, the whole area that borders on the non-Pama-Nyungan languages has developed an inclusive/exclusive distinction. In Western Australia, this area extends roughly to the 22nd parallel of latitude, as claimed by O’Grady et al. (1966: 104-105). At some time in the past, this area extended through central Australia downwards into southern Australia; currently, however, this connection has been interrupted because the Central and Southern Arandic languages have recently lost the inclusive/exclusive distinction (Koch 1997). Southeastern Australia appears to be a separate area with an inclusive/exclusive distinction. However, the situation is not completely clear, partly because there is not that much known about these languages.
The areal patterns are stronger than the genetic bonds among the Pama-Nyungan languages. For example, most languages of the Paman family in Cape York have an inclusive/exclusive distinction, except for the two southernmost languages Ngawun and Mbabaram; these border on the Maric and Galgadungic families, which do not have an inclusive/exclusive distinction.