The causes of language death and extinction are numerous (Wurm 1992), and
may reflect deliberate human action, involving violence and coercion, or accidental
circumstances, through contact with neighbors, absorbtion into other linguistic groups or
natural causes. Generally, language loss is preceded by some sort of multilingualism,
whether societal, through coexistence of different language varieties in the same
geographic area, or individual, through individuals knowing more than one language.
Both circumstances can lead to language shift, especially of one language has more
speakers, a broader range of uses in the society, or greater economic power than the
others. Over successive generations, individuals can come to see the advantage in using
the larger, more powerful language, and so discontinue the use of their own languages.
Most of the Native North American languages face precisely this problem. The
impoverished means available to the approximately 100,000 speakers of Navajo on the
reservations simply cannot compete with the affordances available to English-educated
citizens of the metropolitan centers.
Similarly, pidginization and creolization of languages have been suggested to be
linked with language shift (Muhlhausler 1996). In this scenario, which is being played
out in many places in Oceania, a simplified version of a major language, also called a
pidgin language, serves initially to connect people into large economic trade networks.
Later, parts of the population are drawn off to metropolitan or industrial centers (such as
mines or plantations), where the pidgin becomes the primary language of shared
communication. The relocated people then often inter-marry, possibly settle in a
metropolitan area, and their offspring learn a creole based on the former pidgin language.
The creole is typically regarded as a low-status form of speech, and so speakers who wish
to advance economically in the metropolitan center may eventually give up the creole in
favor of the standard language that gave rise to the pidgin in the first place, thereby
completing the shift.
Historically, the circumstances of creole language formation have often been
extreme, as was the case in Surinam during its colonization, where the creoles Sranan and
Saramaccan are spoken today. In the period between 1650 and 1815, approximately
200,000 Africans were brought into Surinam, but due to harsh conditions and low life
expectancy (5-10 years), the population at the end of this period numbered only around
36,000 (Arends 1995; Postma 1990). Today, it is sometimes claimed that the conditions
for creole genesis no longer operate. Yet if large language contact in the context of
population dislocations, subjugation and high mortality are the necessary preconditions,
one need only look to many of the world’s trouble spots to wonder if this is in fact true.
Refugee and human rights crises gripping sub-Saharan Africa from East to West have all
of these hallmarks, as do similar situations in Central, Southern and Southeastern Asia,
and many of these are long-standing, lasting several generations.
Whatever the mechanisms of language shift involved, linguists are in agreement
that the past two centuries have been catastrophic for global linguistic diversity, and that
this next century is likely to prove even more so. According to some estimates, as many
as half of the world’s remaining languages may be extinct by the end of the present
century (Krauss 1992), unless serious efforts are made to reverse the trend. This
impending unparalleled mass extinction of human heritage has been called the
intellectual equivalent of an ecological catastrophe (Zepeda and Hill 1991). The notion of
linguistic ecology and its explicit parallels with biology is more than a metaphor. It is a
developing area of linguistic theory that contributes to understanding linguistic diversity
in historical, typological, and ethnological terms (Dixon, 1997; Muhlhausler 1996;
Nichols 1992; Dalby 2003), and in relation to local biodiversity (Maffi 2001).
1.4. Language statistics and linguistic fieldwork
Language statistics are collected in a number of ways, depending on the purpose,
resources available for their collection, and the nature of the entities collecting the
statistics. Large compilations of language statistics are therefore heterogeneous, in
comprising a body of statements gathered through different means. Unfortunately, a
major consequence of this is that the statistics so gathered are often not readily
comparable to one another, and it can be very difficult to know what sort of information
one really has.
A major source of language statistics, particularly on national and official
languages, comes from official censuses conducted in the countries where those
languages are spoken. The chief advantages of language statstics from censuses is that
they are large , and regularly administered, making it possible to view large-scale
compositions, global comparisons and longitudinal trends (Lieberson 1967). Nonetheless
they also have many problems (Fasold 1984). First, national censuses often do not ask
language questions at all. In such cases, it is sometimes possible to infer language
populations from other information, such as ethnicity or religious affiliation, where that is
known, but this is extremely hazardous as a general rule. Second, the nature of language
questions when they are asked is not always the same from census to census. Subtle
differences in the wording of language questions can lead to large differences in the
results obtained. Moreover, when language questions are asked, they may be asked in
ways that are not comparable from one year to the next, if the census is regularly revised.
A typical change may involve the number and organization of language categories:
languages may be added to or removed from census questions, leading to incomparable
results from year to year. Sometimes the language populations reported turn out to be
something else, such as ethnicities or religious groups. Finally, national governments
often have vested interests in the outcomes of language questions on a census. For
example, the establishment of educational or government services in particular languages
may hinge on a particular outcome, or parties in the government are intent on maintaining
the status and prestige of a national or official language. Census respondents, aware that
their governments are potentially observing their responses, may under-report minority
language use in such circumstances, leading to skewed results. Issues such as these have
impeded the recognition of Spanish in the US, as illegal immigrants and undocumented
workers from Latin America can lose their existing rights if their background and status
were revealed via the census.
A second source of language statistics comes from large-scale field surveys. This
generally involves a group of linguists, anthropologists, other researchers and/or aid