The remaining factor is a historical one whose various outcomes include the just-mentioned decrease in language diversity with increasing political complexity. World regions have repeatedly been swept by “language steamrollers,” in which one group enjoying some advantage of population numbers, food base, or technology exploits that advantage to expand at the expense of neighboring groups, imposing its own language on the region and replacing previous local languages by driving out or killing their speakers or converting them to speaking the invader’s language. The most familiar steamrollers are those associated with expansions of powerful states over stateless peoples. Recent examples have included European expansions replacing native languages of the Americas, the British conquest of Australia replacing Aboriginal Australian languages, and the Russian expansion over the Ural Mountains to the Pacific Ocean replacing native Siberian languages. In the past, as well, there have been historically documented state-driven steamrollers. The Roman Empire’s expansion over the Mediterranean basin and most of Western Europe extinguished Etruscan, Continental Celtic languages, and many other languages. The expansion of the Inca Empire and its predecessors similarly spread the Quechua and Aymara languages over the Andes.
Less familiar to non-linguists are the steamrollers driven by expansions of pre-literate farmers over the lands of hunter-gatherers, and inferred from linguistic and archaeological rather than historical evidence. Well understood ones include the expansions of Bantu and Austronesian farmers, which largely replaced the former languages of hunter-gatherers in subequatorial Africa and Island Southeast Asia respectively. There were also steamrollers in which hunter-gatherers overran other hunter-gatherers, driven by improved technology: e.g., the expansion of the Inuit 1,000 years ago eastwards across the Canadian Arctic, based on technological advances such as dog sleds and kayaks.
A consequence of these several types of historical expansions is that some world regions containing few geographic barriers have repeatedly been overrun by linguistic steamrollers. The immediate result is very low linguistic diversity, because an invading language sweeps away pre-existing linguistic diversity. With time, the invading language differentiates into local dialects and then into separate languages, but all of them still closely related to each other. An early stage in this process is illustrated by the Inuit expansion of 1,000 years ago; all eastern Inuit people from Alaska to Greenland still speak mutually intelligible dialects of a single language. The Roman and Bantu expansions of 2,000 years ago represent a slightly later stage: the various Italic languages (such as French, Spanish, and Romanian) are very similar but no longer mutually intelligible, as is also true of the hundreds of closely related Bantu languages. At a still later stage, the Austronesian expansion that began around 6,000 years ago has by now generated a thousand languages falling into eight branches, but still sufficiently similar that there is no doubt about their relationship.
Contrasting with those easily overrun areas that Johanna Nichols terms “language spread zones” are what she terms “residual zones” or refugia: mountainous and other areas that are difficult for states and other outsiders to overrun, where languages survive and diffentiate for long times, and hence where unique language groups survive. Famous examples are the Caucasus Mountains, with 3 unique language families plus a few recently invaded languages belonging to three other widespread families; northern Australia, to which 26 of Aboriginal Australia’s 27 language families are confined; Indian California, with about 80 languages variously classified into somewhere between 6 and 22 families; and, of course, New Guinea, with its 1,000 languages classified in dozens of families.
We thus have several more reasons why New Guinea leads the world in number of languages and of language families. In addition to the ecological reasons previously mentioned—little seasonal variation, sedentary populations, a productive environment supporting high human population densities, and great ecological diversity supporting many co-existing human groups with different subsistence strategies—we now have some socio-economic and historical factors as well. Those include the facts that traditional New Guinea never developed state government, so there was never a state steamroller to homogenize linguistic diversity; and that, as a result of New Guinea’s highly dissected mountainous terrain, the steamroller probably caused by the spread of Highlands farming (that associated with the so-called Trans–New Guinea language phylum) was unable to eliminate dozens of older New Guinea language phyla.
Traditional multilingualism
Those are the reasons why the modern world inherited 7,000 languages from the traditional world until yesterday, and why language communities of hunter-gatherers and small-scale farmers without state government contained many fewer speakers than do modern state societies. What about bilingualism and multilingualism? Are traditional societies more, less, or equally often bilingual compared to modern state societies?
The distinction between bilingualism (or multilingualism) and monolingualism proves even more difficult to define and more arbitrary than is the distinction between a language and a dialect. Should you count yourself as bilingual only if you can converse fluently in a second language besides your mother tongue? Should you count languages in which you can converse clumsily? What about languages that you can read but not speak—e.g., Latin and classical Greek for those of us who learned those languages at school? And what about languages that you can’t speak, but that you can understand when spoken by others? American-born children of immigrant parents often can understand but not speak their parents’ language, and New Guineans often distinguish between languages that they can both speak and understand, and languages that they say that they can only “hear” but not speak. Partly because of this lack of agreement on a definition of bilingualism, we lack comparative data on the frequency of bilingualism around the world.
Nevertheless, we don’t have to throw up our hands in despair and ignore the subject, because there is much anecdotal information about bilingualism. Most native-born Americans with English-speaking parents are effectively monolingual for obvious reasons: in the United States there is little need, and for most Americans little regular opportunity, to speak a second language; most immigrants to the U.S. learn English; and most English-speaking Americans marry English-speaking spouses. Most European countries have only a single official national language, and most native-born Europeans with native-born parents learn only that national language as pre-school children. However, because European countries are all much smaller in area and (today) much less self-sufficient economically, politically, and culturally than is the United States, most educated Europeans now learn additional languages in school by formal instruction and often achieve fluency. Shop assistants in many Scandinavian department stores wear pins on their jackets showing the flags of the various languages in which they are competent to help foreign customers. Nevertheless, this widespread multilingualism in Europe is a recent phenomenon that has resulted from mass higher education, post–World War II economic and political integration, and the spread of English-language mass media. Formerly, monolingualism was widespread in European nation-states, as in other state societies. The reasons are clear: state speech communities are huge, often millions of speakers; state societies favor the state’s own language for use in government, education, commerce, the army, and entertainment; and (as I’ll discuss below) states have potent intentional and unintentional means of spreading their state language at the expense of other languages.