Here the plural and dual are represented not by a modification of the singular but by a new word; as different from nga as nos is from ego. The tu, of course, is non-radical, the Gudang form being ngai.
Nga, expressive of the first person, is as common as ngi, equivalent to the second. Thus, nga-nya, nga-toa, nga-i, nga-pe = I, me, in the Western Australian, New South Wales, Parnkalla, and Encounter Bay dialects.
Now, the difference between the first and second persons being expressed by different modifications (nga, ngi) of the same root (ng), rather than by separate words, suggests the inquiry as to the original power of that root. It has already been said that, in many languages, the pronoun of the third person is, in origin, a demonstrative. In the Kowrarega it seems as if even the basis of the first and second was the root of the demonstrative also; since, by looking lower down in the list, we find that i-na = this, che-na = that, and nga-du (nga in Gudang) = who. Ina and chena also means here and there, respectively.
The dual form albei reappears in the Yak-kumban dialect of the River Darling where allewa = we two. Arri = us, is also the first syllable in the Western Australian form ar-lingul = we; or, rather it is ar-lingul in a simpler and less compounded form. In a short specimen of Mr. Eyre's from the head of the Great Australian Bight, the form in a appears in the singular number, ajjo = I and me. The root tana = they, is not illustrated without going as far as the Western Australian of Mr. Eyre. Here, however, we find it in the compound word par-tanna = many. Its original power is probably others; and it is most likely a widely diffused Australian root.
The pronouns in question are compound rather than simple; i.e. instead of nga = me, and ngi = thee, we have nga-tu and ngi-du. What is the import and explanation of this? It may safely be said, that the termination in the Australian is NOT a termination like the Latin met in ego- met, inasmuch as this last is constant throughout the three persons (ego-met, tute-met, se-met), whereas, the former varies with the pronoun to which it is appended (nga-tu, and ngi-du). I hazard the conjecture that the two forms correspond with the adverbs here and there; so that nga-tu = I here, and ngi-du = thou there, and nu-du = he there. In respect to the juxtaposition of the simple forms (ngai, ngi, and nue) of the Gudang with the compound ones (nga-tu, ngi-du, and nu-du) of the Kowrarega, it can be shown that the same occurs in the Parnkalla of Port Lincoln; where Mr. Eyre gives the double form ngai and nga-ppo each = I or me.
Now, this analysis of the Kowrarega personals has exhibited the evolution of one sort of pronoun out of another, with the addition of certain words expressive of number, the result being no true inflexion but an agglutination or combination of separate words. It has also shown how the separate elements of such combinations may appear in different forms and with different powers in different dialects of the same language, and different languages of the same class, even where, in the primary and normal signification, they may be wanting in others. The first of these facts is a contribution to the laws of language in general; the second shows that a great amount of apparent difference may be exhibited on the surface of a language which disappears as the analysis proceeds.
In rude languages the Numerals vary with the dialect more than most other words. We can understand this by imagining what the case would be in English if one of our dialects counted things by the brace, another by the pair, and a third by the couple. Nevertheless, if we bear in mind the Greek forms Thalassa and Thalatta, we may fairly suppose that the Kowrarega word for two, or quassur, is the same word with the Head of Australian Bight kootera, the Parnkalla kuttara, and the Western Australian kardura, having the same meaning.
The difference, then, between the numerals of the Australian languages--and it is undoubtedly great--is no proof of any fundamental difference of structure or origin. It is just what occurs in the languages of Africa, and, in a still greater degree, in those of America.
The extent to which the numeration is carried, is a matter of more importance. Possibly a numeration limited to the first three, four, or five numbers is the effect of intellectual inferiority. It is certainly a cause that continues it. As a measure of ethnological affinity it is unimportant. In America we have, within a limited range of languages, vigesimal systems like the Mexican, and systems limited to the three first units like the Caribb. The difference between a vigesimal and decimal system arises simply from the practice of counting by the fingers and toes collectively, or the fingers alone, being prevalent; whereas the decimal system as opposed to the quinary is referrible to the numeration being extended to both hands, instead of limited to one. Numerations not extending as far as five are generally independent of the fingers in toto. Then as to the names of particular numbers. Two nations may each take the name of the number two from some natural dualism; but they may not take it from the name. For instance, one American Indian may take it from a pair of skates, another from a pair of shoes. If so, the word for two will differ in the two languages, even when the names for skate and shoe agree. All this is supported by real facts, and is no hypothetical illustration; so that the inference from it is, that, in languages where a numeral system is in the process of formation, difference in the names of the numbers is comparatively unimportant.
The extent to which the numerals vary, the extent to which they agree, and the extent to which this variation and agreement are anything but coincident with geographical proximity or distance, may be seen in the following table: