60. Diffusion and Parallelism in Language and Culture
A phenomenon which language shows more conspicuously than culture, or which is more readily demonstrated in it, is parallel or convergent development, the repeated, independent growth of a trait (§ [89], [100]).
Thus sex gender is an old part of Indo-European structure. In English, by the way, it has wholly disappeared, so far as formal expression goes, from noun, adjective, and demonstrative and interrogative pronoun. It lingers only in the personal pronoun of the third person singular—he, she, it. A grammar of living English that was genuinely practical and unbound by tradition would never mention gender except in discussing these three little words. That our grammars specify man as a masculine and woman as a feminine noun is due merely to the fact that in Latin the corresponding words vir and femina possess endings which are recognized as generally masculine and feminine, and that an associated adjective ends respectively in masculine -us or feminine -a. These are distinctions of form of which English possesses no equivalents. The survival of distinction between he, she, and it, while this and the and which have become alike irrespective of the sex of the person or thing they denote, is therefore historically significant. It points back to the past and to surviving Indo-European languages.
Besides, Indo-European, Semitic and Hamitic express sex by grammatical forms, although like French and Spanish and Italian, they know only two genders, the neuter being unrepresented. These three are the only large language stocks in which sex gender finds expression. Ural-Altaic, Chinese, Japanese, Dravidian, Malayo-Polynesian, Bantu, and in general the language families of Asia, Africa, and America do without, although a number of languages make other gender classifications, as of animate and inanimate, personal and impersonal, superior and inferior, intelligent and unintelligent. Sex gender however reappears in Hottentot of South Africa, and in the Chinook and Coast Salish and Pomo languages of the Pacific coast of North America.
How is this distribution to be accounted for? Indo-European, Semitic, and Hamitic occupy contiguous territory, in fact surround the Mediterranean over a tract approximately co-extensive with the Caucasian area. Could they in the remote past have influenced one another? That is, could grammatical sex gender have been invented, so to speak, by one of them, and borrowed by the others, as we know that cultural inventions are constantly diffused? Few philologists would grant this as likely: there are too few authenticated cases of formal elements or concepts having been disseminated between unrelated languages. Is it then possible that our three stocks are at bottom related? Sex gender in that case would be part of their common inheritance. For Semitic and Hamitic a number of specialists have accepted a common origin on other grounds. But for Semitic and Indo-European, philologists, who are professionally exacting, are in the main quite dubious. Positive evidence seems yet to be lacking. Still, the territorial continuity of the three speech groups showing the trait is difficult to accept as mere coincidence. In a parallel case in the realm of culture history, a common source would be accepted as highly probable. Even Hottentot has been considered a remote Semitic-Hamitic offshoot, largely, it is true, because of the very fact that it expresses gender. Philologists, accordingly, may consider the case still open; but it is at least conceivable that the phenomenon goes back to a single origin in these four Old World stocks.
Yet no stretch will account for sex gender in the three American languages as due to contact influence or diffusion, nor relate these tongues to the Old World ones. Clearly here is a case of independent origin or parallel “invention.” Chinook and Coast Salish, indeed, are in contiguity, and one may therefore have taken up the trait in imitation of the other. But Pomo lies well to the south and its affiliations run still farther south. Here sex gender is obviously an independent, secondary, and rather recent growth in the grammar.
In short, it remains doubtful whether sex gender originated three or four or five or six times among these seven language stocks; but it evidently originated repeatedly.
Other traits crop out the world over in much the same manner. A dual, for instance, is found in Indo-European, Malayo-Polynesian, Eskimo, and a number of other American languages. The distinction between inclusive and exclusive we—you and I as opposed to he and I—is made in Malayo-Polynesian, Hottentot, Iroquois, Uto-Aztecan.
A true nominative case-ending, such as Latin and the other varieties of Indo-European evince, is an exceedingly specialized formation; yet is found in the Maidu language of California. Articles, in regard to which Indo-European varies, Latin for instance being without, while its Romance daughter tongues have developed them, recur in Semitic, in Polynesian, and in several groups of American languages, such as Siouan and Hokan. The growth in Romance is significant because of its historicity, and because it was surely not due to imitation of an unrelated language. That is, French developed its articles independently and secondarily; a fact that makes it probable that many languages in other parts of the world, whose history we do not know, developed theirs in a parallel manner, as a product of wholly internal causes—“invented” them, in short, although wholly unconsciously.
A trait found in a large proportion of the American languages is the so-called incorporation of the object pronoun (§ [51]). The objective pronoun, or an element representing it, is prefixed or suffixed to the verb, made a part of it. The process is familiar enough to us from Indo-European so far as the subject is concerned: in Latin ama-s, ama-t, ama-nt, the suffixes express “you, he, they” and pronouns comparable to the English ones—independent words—are usually omitted. The -s in he love-s is the sole survival of the process in modern English. None of the older Indo-European tongues however showed an inclination to affix similar elements for the objects, although there are some approaches in a few recent languages of the family: Spanish diga-me, “tell me,” and mata-le, “kill him,” for instance. Semitic on the other hand, and Basque, do “incorporate” objective elements, whereas most Asiatic and some American languages do not. Many other instances of parallel or convergent traits could be cited.
This greater frequency of parallel developments in language than in culture is perhaps in part due to easier demonstrability in the field of speech. But in the main the higher frequency seems real. Two reasons for the difference suggest themselves.
First, the number of possibilities is small in language, so far as structure is concerned. The categories or concepts used for classifying and for the indication of relations are rigorously limited, and so are the means of expression. The distinctions expressed by gender, for instance, may refer to sex, animateness, personality, worth, shape, position, or possibly one or two other qualities; but there they end. If a language recognizes gender at all, it must have gender of one of these few types. Consequently there is some probability of several unconnected languages sooner or later happening upon the same type of gender. Similarly, for the kinds of number, and of case, and so on, that are denotable. These larger categories, like gender and number and case, are not numerous. Then, the means of expressing such relational and classificatory concepts are limited. There is position or relative order of words; compounding of them; accretions of elements to stems, namely prefixes, infixes, and suffixes; reduplication, the repetition of part or the whole of words; internal changes by shift of vowel or accent within words; and therewith the types of grammatical means are about exhausted. The number of possible choices is so small that the law of accidental probability must cause many languages to hit upon the same devices.
A second reason for the greater frequency of parallelism in language is that structural traits appear to resist diffusion by imitation to a considerable degree. Words are borrowed, sometimes freely, almost always to some degree, between contiguous languages; sounds considerably less; grammar least of all. That is, linguistic content lends itself to diffusion readily, linguistic form with difficulty.
At bottom, the same holds of culture. Specific elements of culture or groups of such elements diffuse very widely at times and may be said to be always tending to diffuse: the wheel, for instance, smelting of metals, the crown as a symbol of royalty, the swastika, Buddhism. The relations of elements among themselves, on the other hand, change by internal growth rather than external imitation. Of this sort are the relations of the classes and members of societies, the fervor with which religion is felt, the esteem accorded to learning or wealth or tradition, the inclination toward this or that avenue of subsistence or economic development. By conquest or peaceful pressure or penetration one people may shatter the political structure or social fabric of another, may undermine its conservatism, may swerve its economic habits. But it is difficult to find cases of one people adopting such tendencies or schemes of cultural organization in mere imitation of the example of another, as it will adopt specific culture content—the wheel or crown or Buddhism, for instance—from outside, often readily. The result is that culture relations or forms develop spontaneously or from within rather than as a result of direct taking over. Also, the types of culture forms being limited in number, the same type is frequently evolved independently. Thus monarchical and democratic societies, feudal or caste-divided ones, priest-ridden and relatively irreligious ones, expansive and mercantile or self-sufficient and agricultural nations, evolve over and over again. On the whole, comparative culture history more often deals with the specific contents of civilization, perhaps because events like the spread of an invention can be traced more definitely and exactly than the rather complex evolutions of say two feudal systems can be compared. The result is that diffusions seem to outweigh parallels; as is set forth in several of the chapters that follow this one (§ [105], [111], [127]).
In comparative linguistics, on the other hand, interest inclines to the side of form rather than content; hence the parallelisms or convergences are conspicuous. If as much attention were generally given to words as to grammar, and if they could be traced in their prehistoric or unrecorded wanderings as reliably as many culture traits have been, it is probable that diffusion would loom larger as a principle shaping human speech. There are words that have traveled almost as far as the objects they denote: tobacco and maize, for example. And the absorption of words of Latin origin into English was as extensive as the absorption for over a thousand years of Latin, Christian, and Mediterranean culture by the English people—went on as its accompaniment and result.