by Johanna Nichols and David A. Peterson
Personal pronouns, like other closed-class forms, tend to contain nasals and other basic consonants. However, particular paradigmatic combinations of consonants in particular person-number categories are not especially frequent worldwide. There are only two such combinations that show up interestingly on a map: pronoun sets like English me and thee or Nanai (Tungusic; eastern Siberia) mi 'I, me' and si 'you', with m in the first person and a coronal obstruent in the second, which are common in northern Eurasia and rare elsewhere; and sets like Northeast Maidu (Penutian; California) ni 'I, me' and mi 'you' with n in the first person and m in the second, which are common in western America and rare elsewhere. Map 136A here shows the first type, henceforth "m-T" patterns, and Map 137A the second, henceforth "n-m" patterns.
The maps here are based on the first and second person singular forms only. They survey the first consonant in the pronoun root, whether it is word-initial as in Wintu (Penutian; California) ni 'I, me; we, us' or follows an initial vowel as in Nez Perce (Sahaptian; Idaho and Oregon) 'iin 'I, me'. (Whether written or not, the glottal stop that precedes an initial vowel is not considered the first consonant. The Nez Perce form just cited begins with such a glottal stop.) They survey three pronominal forms: independent pronoun words like English I, me, you; possessive affixes as in Hungarian ház-am 'my house', ház-ad 'your house'; and verbal affixes as in Rama (Chibchan; Nicaragua) n-taak-i 'I am going', m-taak-i 'you are going' (Grinevald 1988). (Of course, not all languages have all three series.) They consider the two most widely distributed allomorphs of each, e.g. English I and me for first person singular.
The consonants mapped are defined as follows:
m |
= [m] as well as glottalized, voiceless, lenis, and fortis labial nasals (in fact only plain [m] occurred in the language sample) |
n |
= dental or alveolar [n], palatal [n´] (and glottalized, voiceless, etc. variants, though none appeared in the sample) |
T |
= any apical obstruent: [t], [d], [s]; palatals such as [č], [š], etc. |
Here are some examples of pronouns that fit the two types. First person m and second person apical obstruent in independent forms:
(1) German
1sg
mich
(accusative)
mein
(possessive)
2sg
dich
(accusative)
dein
(possessive)
(2) Georgian
1sg
me
2sg
šen
(3) Nanai (Tungusic; eastern Siberia)
1sg
mi
2sg
si
in possessive affixes:
(4) Hungarian
1sg
ház-am
'my house'
2sg
ház-ad
'your house'
in verbal affixes:
(5) Turkish present tense (Lewis 1967: 109)
1sg
alıyor-um
'I am taking'
2sg
alıyor-sun
'you are taking'
First person n and second person m in independent forms:
(6) Wintu (Penutian; California)
1st person
ni
2nd person
mi
in possessive affixes:
(7) Pipil (Uto-Aztecan; El Salvador; Campbell 1985: 43)
1sg
nu-chi:l
'my chili pepper'
2sg
mu-chi:l
'your chili pepper'
(8) Karok (Hokan; California; Bright 1957: 560)
1sg
nani-tta:t
'my mother'
2sg
mi-tta:t
'your mother'
in verbal agreement formatives:
(9) Nisgha (Tsimshianic; British Columbia; Tarpent 1987: 461)
1sg
nə-
2sg
mə-
(transitive subject proclitics)
The following values are shown on the maps:
Go to map | ||
---|---|---|
Value | Representation | |
No M-T pronouns | 200 | |
M-T pronouns, paradigmatic | 27 | |
M-T pronouns, non-paradigmatic | 3 | |
Total: | 230 |
Go to map | ||
---|---|---|
Value | Representation | |
No N-M pronouns | 194 | |
N-M pronouns, paradigmatic | 25 | |
N-M pronouns, non-paradigmatic | 11 | |
Total: | 230 |
The maps distinguish between paradigmatic and non-paradigmatic distributions of the consonants. The consonants form a paradigm when they both occur in the same form class(es) of their respective pronouns. For instance, a consistent paradigm occurs in Mapudungun (isolate; Chile):
(10) Mapudungun (Smeets 1989)
1sg
2sg
Independent forms:
Nominative
iñche
eymi
Possessive
ñi
mi
Verbal suffixes:
Indicative
-(ü)n
-m-i
Conditional
-i-Ø
-m-i
The underlined consonants are in the same position in three different form series and occur in both forms of the set. A subparadigm occurs in Wichí (Mataco-Guaicuru; Argentina):
(11) Wichí (Viñas Urquiza 1974)
1sg
2sg
Independent Nominative
oƛam
am
Possessive
animate
olä
alä
inanimate
okä
akä
Verb subject
o-
le-
Verb object
-no
-am
Verb object
-o
-a
The first pair of verb-object suffixes (the next to last line here) shows the pattern; no other does. The maps in this chapter do not distinguish consistent paradigms from subparadigms.
A case of first person n and second person m which do not form a paradigm occurs in Asmat (Asmat-Kamoro; Papua, Indonesia):
(12) Asmat (Voorhoeve 1965b)
1sg
2sg
Independent Nominative
no-r
o-r
Independent Oblique
no
o
Verb subject
-i
-em
Verb object
-(e)n
-(e)n
Here there is first person n in the independent forms and second person m in the verb-subject agreement suffix. In fact, n occurs in both persons in the verb object forms. There is no place where n and m co-occur in the same form class.
Finally, the occasional token of one or the other paradigm can turn out to be truly artifactual, as in the following from Grebo (Kru, Niger-Congo; Liberia):
(13) Grebo (Innes 1966)
1sg
2sg
Emphatic
mɔ 1
mɔ 2/3
Object
mo 1/2
mo 2/3
Possessive
na 1/2
na 3
(numbers indicate tones)
The consonant-vowel forms of the pronouns are the same in all form classes, and only the tones differentiate them. One can find a first person n and second person m in this set, so this technically counts as a (non-paradigmatic) occurrence of the n-m consonantism, but this is rather clearly due to a fluke rather than to a principled linguistic analysis of the material.
It is the paradigmatic sets that are most interesting, and in both maps they are by far the most common of the examples of the two consonant pairs. Non-paradigms (like that of Asmat, above) and flukes (like the forms of Grebo just above) are rare, and occur only outside of the areas where the paradigms are relatively common.
The pronoun systems in question form two large areal clusters: an m-T one centered in northern Eurasia (Map 136A) and an n-m one extending from western coastal North America to western South America (Map 137A). There are a number of other maps in this atlas that have similar areal distributions: the n-m concentration on the Pacific Rim and especially the American Pacific Rim recalls the distribution of multiple possessive classification, bound inalienables, numeral classifiers, verb-initial word order, and double marking (see Chapters 59, 58, 55, 81, 23, respectively); the m-T cluster in northern Eurasia recalls other Greater Silk Road distributions such as dependent marking, low synthesis, absence of possessive classification, and case-number coexponence (see Chapters 23, 22, 59, 21, respectively).
Mapping just m in either first person singular (Map 136B) or second person singular (Map 137B) yields somewhat more diffuse clusters but reveals the same large areas clearly.
Go to map | ||
---|---|---|
Value | Representation | |
No m in first person singular | 177 | |
m in first person singular | 53 | |
Total: | 230 |
Go to map | ||
---|---|---|
Value | Representation | |
No m in second person singular | 152 | |
m in second person singular | 78 | |
Total: | 230 |
First person m (Map 136B) is ubiquitous in the Greater Silk Road area and nearly pan-Eurasian, missing in the Eurasian islands, Southeast Asia, and the Asian Pacific Rim; it is well attested in sub-Saharan Africa and New Guinea, sparsely attested in the Americas, and absolutely absent from Australia. Second person m (Map 137B) has a solid Pacific Rim distribution with occasional tokens inland in the Americas, in the Eurasian islands, and in Africa; it is absolutely absent from the Greater Silk Road area and Australia.
What explains these two large unique clusters of languages with the distinctive pronoun consonantisms? The two main explanations that have been offered in the literature are:
Sound-symbolic universals: nasals and other basic consonants are the most elementary consonants, hence the most easily learned by children, and such consonants figure frequently in pronouns and other basic words. While this generalization is correct for broad phonological categories like "nasal", it cannot explain the distributions of the two-consonant paradigms or even of just m mapped here, for their distribution is not universal. No explanation based on universals can account for the fact that each of the two paradigms is fairly common in one large area and vanishingly rare elsewhere. The near-absence of each paradigm in every large area but one indicates that they are not due to any underlying tendency but have particular historical origins.
Genealogical relatedness of the languages: each cluster is made up of the descendants of a single protolanguage. While this explanation may seem plausible, there is no other evidence for the relatedness of all and only the conforming languages, and the pronoun systems are not sufficient evidence to prove relatedness. The conforming languages in each area belong to several different well-established families:
Language(s) |
Family |
---|---|
Eurasia: |
|
French, German, Modern Greek, Hindi, Ossetic, Persian, Russian, Spanish, Waigali |
|
(isolate) |
|
Elsewhere: |
|
Utian |
|
Adelbert Range |
Language(s) |
Family |
---|---|
Americas: |
|
(isolate) |
|
(isolate) |
|
Mixe |
|
Elsewhere: |
|
On the other hand, in each cluster there are a few language families that are likely to be deeply related. Indo-European and Uralic are likely to be related, though the relatedness is so distant that we may never have sufficiently firm evidence to prove it by conventional means (e.g. Ringe 1998, Oswalt 1991). Still deeper relatedness of several of the other m-T families of Eurasia to these has been proposed but never demonstrated. In any event the first person m is secondary in Turkic and Tungusic (where it developed from a b). In North America, a set of language families which have the n-m system is thought to be possibly deeply related: Utian, Wintun, Maidu (Northeast), Sahaptian, and possibly also Chinookan may belong to a proposed megafamily called Penutian. But not all of the putative Penutian daughter families exhibit the system, and the putative Penutian set represents less than half of the tokens of n-m systems. Furthermore, one putative Penutian language, Southern Sierra Miwok, exhibits the m-T system. Thus deep genetic relatedness is no explanation for the pronoun patterns.
Both the m-T and the n-m paradigms are found in areas where spreads are known to have been centered -- the Greater Silk Road and the Pacific Rim – and both appear to be the products of geographical spread rather than just universals or just inheritance. Such spreads of pronominal consonantism patterns evidently do not occur often: we have only these two clear cases in the whole world. The origins of both are old (older than any of the individual families exhibiting the systems, for instance): the Eurasian m-T pattern is at least pre-Neolithic (Nichols 2001), and the American n-m one is late glacial at the latest (Nichols and Peterson 1996). Since these ages are greater than the time depth to which linguistics can generally trace genealogical relations, the details of their origins are lost in time.
We also looked for other interesting consonant patterns in pronominal roots, e.g. those involving velar nasals and velar obstruents k, g, etc. However, no such patterns showed conspicuous large areal distributions; all are more or less evenly distributed worldwide. So far, then, the m-T and n-m paradigms mapped here are unto themselves.
For more discussion see Nichols and Peterson 1996 and references therein; for subsequent discussion, see Nichols 2001.