Somewhat paradoxically, the plural of personal pronouns has been considered both as a core phenomenon of nominal number and as something that has nothing to do with it. On the one hand, several scholars have suggested that if a language has a number distinction anywhere, then it has it at least in pronouns, or, speaking in diachronic terms, that plurality has spread from plural personal pronouns to other nominals (cf. Forchheimer 1953). On the other hand, linguists have repeatedly argued that pronominal plurals are different from nominal plurals in that the two have a different reference structure. Thus, it is observed that ‘we’ is not the same as ‘I + I + … I’ (e.g. Benveniste 1966a, Barulin 1980; cf. also Corbett 2000: 83-84). This chapter provides an account of what types of formal correlations between plurals and singulars are actually observed in personal pronouns, and how the devices for expressing pronominal plurality relate to the devices for expressing nominal plurality. It is based on 260 languages, including the 200 languages of the WALS sample, and will thus contribute further empirical data to the discussion.
To ensure cross-linguistic comparability, I singled out one type of pronoun which can be identified in many languages: independent subject pronouns. Independent pronouns were chosen because many languages lack non-independent pronouns, and also because they have been most often discussed in comparisons of pronominal plurality with nominal plurality. However, a handful of languages are reported to have no independent personal pronouns, or at least no plural independent personal pronouns. Subject pronouns were chosen for the sake of consistency, because different case forms sometimes display different formal correlation patterns (see §5.2 below).
Both inclusives and third person pronouns have been excluded from consideration. Third-person pronouns are often deictic rather than personal and only very rarely (in about one tenth of the cases) show a morphological pluralization pattern identical to that of the first and second person pronouns (see also Chapter 43). Inclusives are not found in all languages (see Chapters 39 and 40) and often lack number distinctions altogether.
|No independent subject pronouns||2|
|Person-number stem with a pronominal plural affix||47|
|Person-number stem with a nominal plural affix||22|
|Person stem with a pronominal plural affix||23|
|Person stem with a nominal plural affix||19|
Type one is represented by two languages reported to have no plural independent subject pronouns or no independent subject pronouns at all: Acoma (Keresan; New Mexico) and Wari’ (Chapacura-Wanham; Brazil). Type two consists of languages which use the same stem for both plural and singular reference, i.e. have pronouns with number-indifferent stems, e.g. Pirahã (Mura; Brazil) ti ‘I, we’; gíxai ‘you., you.pl’.
The languages in type three have in common that both person and number are expressed by affixes, most often cumulatively, so that an affix may not express number without person also being expressed affixally. An example is Mundari (Munda; India) a-ñ ‘I’ ~ a-liŋ ‘we.du’ ~ a-le ‘we.pl’; a-m ‘you.sg’ ~ a-ben ‘you.du’ ~ a-pe ‘you.’. In most of these languages, the stem is common to all persons (like a- in Mundari) or conveys only person differences.
The most widespread type, type four, comprises languages combining person and number in an unanalyzable person-number stem, e.g. Dogon (Niger-Congo; Mali) mi ‘I’ ~ emme ‘we’; u ‘you.’ ~ e ‘you.’. The languages of the fifth type have a person-number-specific stem and additionally express plurality with a pronominal affix not used on nouns. An example is Amele (Madang; Papua New Guinea) ija ‘I’ ~ e-le ‘we.du’ ~ e-ge ‘we.pl’; hina ‘you.sg’ ~ a-le ‘you.du’ ~ a-ge ‘you.’. The languages in type six use a person-number-specific stem and additionally express plurality with a nominal plural affix, i.e. an affix also used on at least some nouns (e.g. Russian ja ‘I’ ~ m-y ‘we’; ty ‘you.sg’ ~ v-y ‘you.’; cf. slon ‘elephant’, slon-y ).
Type seven contains languages which have the same person stem in the singular and plural and express plurality with a pronominal affix not used on nouns. An example is Chuvash (Turkic; Russia) epĕ ‘I’ ~ ep-ir ‘we’; esĕ ‘you.’ ~ es-ir ‘you.’. Finally, type eight contains languages which have the same person stem in the singular and plural with a nominal affix to mark plurality (e.g. Mandarin wŏ ‘I’ ~ wŏmen ‘we’; nĭ ‘you.’ ~ nĭmen ‘you.pl’).
There is a noticeable difference between personal pronouns and nouns with respect to the preferred locus of number expression. Expressing nominal number by changing the nominal stem (as in English person ~ people) is exceptional (see Chapter 33). Wherever this pluralization strategy exists, it is available only for a small closed subset of nouns. In pronouns, by contrast, the expression of number by a special plural stem is the pattern which dominates in our sample. This provides a first basis of classification of pronominal number systems (reflected on the map by the color of the symbol) – the distribution of person and number marking between the stem and the affix. One can see that most languages express person in the stem (except some languages of type three), while number may be expressed either affixally (types three, seven and eight), or cumulatively with person in the stem (type four), or both (types 5 and 6). What I did not find is a language where number alone is expressed by the stem without person being also expressed in it. If one assumes that the stem belongs to the domain of lexical meaning while the affix belongs to the domain of grammatical meaning, it may be claimed that in most systems of independent personal pronouns the person is categorized as lexical meaning; pronominal number may be categorized either way.
Person and number expression may be intertwined in such a manner that an obviously regular correlation between singular and plural personal pronouns may hardly be describable in segmental terms of a stem and an affix (cf. Majtinskaja 1969). This pattern of non-segmental correlation occurs, for instance, in Yukaghir (Siberia) (met ‘I’ ~ mit ‘we’; tet ‘you.sg’ ~ tit ‘you.pl’) and in Ekari (Papua, Indonesia) (ani ‘I’ ~ inai ‘we’; aki ‘you.’ ~ ikai ‘you.pl’); further cases are Turkish, Evenki, Orok, and Murle. For simplicity, such cases had to be classified together with affixal number marking (type seven).
Expressing pronominal number affixally makes it structurally closer to nominal number. But even in those languages which express number in an affix, the formal identity of the plural marker used on nouns and the marker used on pronouns is rare. This provides a second basis of classification (reflected on the map by the shape of the symbol) – the languages are classified according to whether the number affix is used only on pronouns (type five and seven) or also on at least some other nominals (six and eight).
When we have a formal similarity between plural personal pronouns, it is not always clear whether this alone should be considered sufficient to isolate a plural marker. For instance, in the case of German wir ‘we’ and ihr ‘you.pl’, does -r qualify as a plural marker specific to personal pronouns proper? We would need independent evidence establishing the stem of the pronoun (for instance, that the stem is identical to the stem used by the singular or some other non-singular, e.g. dual, pronoun of the same person) or the plural marker (if it is used with other stems). However, neither of these is available for German (nor for Koyraboro Senni, Fur, Kongo, Grebo, Hindi, Hamtai, Mountain Arapesh, Usan, Una); for some languages, there is only indirect evidence (Finnish, Hungarian). Another complication is that in many languages, stems used in plural pronouns are obviously related to, although not identical with, the corresponding singular stems, this formal relatedness being structurally half-way between identical stems and different stems. Finally, a plural pronoun may apparently be composed of the singular stem plus a unique affix (this is especially clear in Yurok, Kutenai, Mapudungun, Makah). To make the data interpretation more robust, these languages are treated as showing no correlation between the singular and plural pronouns (type four).
The classification is complicated by the fact that in many languages the two persons behave differently with respect to the classification established above. An example is English, which uses person-number specific stems in the first person and a number-indifferent stem in the second person. In fact, about 20% of my sample have this kind of person split. Half of these split cases are languages where one of the two persons (two dozen for the second person against half a dozen for the first) contains a number affix while the other does not. In more than half of the split cases, one of the persons has a person-number-specific stem (the second person in two dozen against one dozen for the first person) while the other has the same stem both in singular and plural. Other cases are very rare. To keep the number of distinct types reasonable, the classification in this chapter always goes with the first person.
An independent personal pronoun may have different case forms. It is not infrequent that these case forms stand in different relations to each other. Compare the Modern Greek oblique forms emena ‘I’ ~ emas ‘we’, esena ‘you.’ ~ esas ‘you.pl’ with the nominative forms egho ‘I’ ~ emis ‘we’, esi ‘you.sg’ ~ esis ‘you.pl’. Apparently, the number correlation is more regular (even though specific to personal pronouns) in oblique forms. Similar splits appear in many languages, including some Turkic, Uralic and Mongolian languages. Conversely, the oblique forms are less regular than the subject forms in Tümpisa Shoshone (Uto-Aztecan; California). Finally, in Southeastern Pomo (California) first person pronouns are more regular in oblique forms, while second person pronouns are more regular in subject forms. For comparability reasons, the map reflects the correlation displayed in the subject pronouns, and disregards possible paradigm splits.
In some languages, different non-singular numbers behave in different ways. Thus, in Nenets (Uralic; northern Russia) the pronominal dual suffix is specific to pronouns, while plural pronouns use a nominal plural suffix. Again, for comparability the classification only takes into account plurals in the strict sense, because not all languages have several non-singular numbers.
A language may have several pronominal forms in the same person-number slot. The classification of the language may then depend on which forms are chosen. Often one of the forms may be considered more peripheral or less pronominal. Many languages have a distinction of polite vs. unmarked in the 2nd singular, and some languages such as Basque apparently have the opposite distinction (unmarked vs. familiar). In all such cases, the marked form was disregarded. In some languages, nominals are widely used for personal reference. A nominal may expand to pronominal usage at the cost of losing its nominal meaning. Thus, in many Southeast Asian languages kin terms are used as subject pronouns even if the kin relation denoted by the term does not actually hold for the speaker and addressee (see Chapter 45). In some languages, nominals may have personal reference in their literal sense, as when the speaker refers to the addressee or to himself by means of a proper name. These "nominal" pronouns were disregarded.
However, some languages have several non-nominal pronominal forms none of which seems to be unmarked. This happens in some South-East Asian languages, e.g. Burmese, which has a rich set of pronominal forms to code different social dimensions. In these cases, the classification reflects the most regular correlation which is found among the pronominal pairs.
As with nouns (see Chapter 34), a plural marker on pronouns may be optional. Such optional markers are often the same as those used on nouns (Bagirmi, Aymara, Chalcatongo Mixtec, Gooniyandi, Wardaman, Uzbek, Turkish, Korean, Ladakhi, Nivkh, Itelmen); only Ju|'hoan has an optional plural marker specific to pronouns. Unmarked forms may be number-indifferent (Aymara, Chalcatongo Mixtec), but in most cases number is also expressed elsewhere - either in the stem (Ju|'hoan) or by an obligatory affix (Nivkh). The map disregards the distinction of optional vs. obligatory markers.
The classification is further complicated by the fact that a pronoun may contain two number affixes. Typically, double-marked pronouns use a nominal plural marker preceded by a marker specific to pronouns (e.g. Wardaman). This order is inverted only in Spanish, cf. no-s-otros ‘we’, vo-s-otros ‘you.pl’, where -s may be argued to be nominal plural. Often, the nominal marker is optional (e.g. Turkish, see §5.5). Such systems are classified as types six and eight, i.e. as displaying nominal number marking, although this is a disputable decision, especially in cases like Nivkh, whose plural pronouns consist of a number-indifferent stem, pronominal number marker and an optional nominal number marker.
The map shows the following areal patterns. Languages lacking plural independent subject pronouns (type one) are extremely rare; each of the two cases in the sample occurs in one of the Americas (an additional such case, not in the sample, is the African language Mbay, which lacks independent personal pronouns altogether; see Keegan 1997: 62-63). Affixal marking of both number and person (type three) is widespread only in North America; several cases are also attested in Amazonia; only isolated cases occur elsewhere (Ainu, Aleut, Mundari, Koromfe, Lango, Pitjantjatjara, Kambera); some languages show irregular pronominal morphology apparently derivative from this pattern. Pronouns which are indifferent to number reference (type two) are rare; they occur in the Americas and in South-East Asia. Pronouns which express both number and person in the stem without affixal marking (type four) dominate throughout the world, especially in the Middle East, the Pacific and Africa. Pronouns which use a number affix (types five through eight) are most typical of Asia and northern Australia; they are not infrequent in the Americas and also occur in central Africa. More specifically, pronouns consisting of a person-specific stem plus number marker cluster in eastern Asia and are rare elsewhere, while pronouns consisting of person-number-specific stem plus number marker specific to pronouns are frequent in northern Australia and the eastern Pacific.
Considering areas rather than types, the most homogeneous are Asia (presence of a number affix) and the Pacific (person-number stem with no affix). The eastern Pacific and northern Australia have person-number stems plus number affixes specific to pronouns. In Africa, again, person-number-specifying stems dominate, though central and northern Africa show greater variation. The Americas are diverse in their pronominal patterns; an isolatable number affix is typically absent. In North America there are quite a few languages which have both person and number expressed affixally.