Introduction

Language Death...
Language death is a crisis of global proportions. Some linguists estimate that in a century, half of the world's 6000 languages will have ceased to be spoken (Crystal 2000). These are startling projections, but what do they mean? How does a language die? Certainly, sometimes a language dies with a people, due to natural disaster or genocide, but more often we see a process that linguists have named "language shift," which is when the speakers of a language simply cease to speak their mother tongue, often in favor of a new national or global language (e.g. the many Native American languages that are spoken by only a few elders of the tribe).

…in Nepal
The trend of language shift is all too evident in the languages of Nepal. Nepal, with its mountain valleys and varied terrain, is one of the most linguistically diverse countries in the world. However, of the 108 languages in Nepal that we have population figures for (this is the Ethnologue count, considered by some to be high, see Watters 2002 for a defense of the catalogue), 91 have fewer than 100,000 speakers. (For comparison, the mostly widely spoken language in Nepal, Nepali, has almost 10 million speakers, while the mostly widely spoken language in the world, Mandarin Chinese, has 1 billion.) It is difficult to know just from population numbers the vitality of a language, but even the most optimistic linguists consider a language with less than 100,000 speakers to be endangered, unless there is outstanding social vitality in the community (Crystal 2000).
Unfortunately, the situation in Nepal is just as bleak sociolinguistically as numerically. Watters (2002) cites these factors which contribute to language shift – “voluntary or forced migration outside the homeland area, other language groups moving into the original homeland area, the language used in school, the desire of a nation to build an identity united under one language, urbanization, industrialization, and globalization.” All of these occur in Nepal, and apply especially to the more accessible areas like the villages I visited. There has been a mass migration of people towards the newly built roads (Katry 2003) as well as Nepali speaking Hindu castes moving into Rai lands (along with the roads and the tea industry, or as school-teachers); Nepal is often cited as an example of runaway urbanization and industrialization (Bhattarai 2003, Giri 1998, Dahal 1998, Omer 1994); and perhaps most importantly, all of the schools in the area, public and private, are taught in Nepali or English medium.
As David Watters, a linguist working with Tribhuvan University, puts it, “Nepal faces a huge task in documentation and potential revitalization” and in that spirit I set out to try to research Bantawa Rai.

Bantawa
Bantawa Rai is a Tibeto-Burman language spoken in the western Himalayan hills by the Bantawa Rai ethnic group. Population estimates vary between 20 and 60 thousand, but the more reasonable fall around 35,000 speakers. With this number it is one of the more widely spoken Rai languages, but still falls into the below-100,000 category of endangered languages
Bantawa has several dialects – at least two in my area of concern, Wana Bantawa (also called simply Bantawa) which is spoken by the Bantawa subcaste and Amcauke, which is spoken by the Amcauke subcaste.
Bantawa is also reportedly in use as a lingua franca among Rai minorities Himalayan India and Bhutan (N.K. Rai 1985 and Bradley 1996 qtd. in Ethnologue 2003).

Field Methods
The bulk of my fieldwork was eliciting the word list from the Bantawa people I spoke with. This served two purposes. Firstly, I learned some of the language, enough at least to tell when people were speaking Bantawa and when they were speaking Nepali, to compile a sketchy description of the language, and in general to have a basis to examine how the language was changing in response to the influence of Nepali and English. Secondly, in gathering the words, I got a good impression of how much of their own language the people I talked to know.
I also conducted informal interviews with many of the villagers, asking them about how much of their language they felt they knew, and why and whether they felt knowing Bantawa was important.
Within the small scope of my project, I tried to get as extensive as possible a spread in the domains of geography (limited by how far from Betini I could walk in half a day) and demography (I talked to soldiers, schoolteachers, tea farmers, laborers, housewives, shamans, college students and schoolchildren, of the Bantawa and Amcauke castes).
In eliciting the wordlist, I tried to avoid simple translation from English or Nepali into Bantawa (that is, I didn't ask “what's the word for chair” or “mechlaai ke bhaansa,”) in order to get more directly at the language and to avoid translation errors on my part or on theirs. Instead, I tried first to mime or point.
I also tried as much as possible to crosscheck words with different people, and people of different ages and families. This checking is represented in column 3 of the wordlist with a confidence score ranging from 3 (everyone agreed) to 1 (dubious). If my informants were sure of a word, but only a few people or only old people knew it, I recorded it as rare/being lost. One issue that arose during this crosschecking was dialectical differences – Bantawa has a number of different dialects, and there were at least two (Amcauke and Wana) in my area. Confusing the matter more were the issues surrounding intersubcaste marriage (The many Rai subcastes speak many different languages. However, since intersubcaste marriage is practiced, there are sometimes two or three languages spoken in one house, and the potential for horizontal language transfer (borrowing) in great.) as well as the matter of house dialects (It has been observed that as language shift progresses and the speakers of a language become few and isolated, a number of “house dialects” or miniature dialects appear (Crystal 2000).)
In addition to writing the wordlists I also made audio recordings of some of the most important words, in order to make sure that I transcribed them correctly, to help future work on phonology, and to make available to other researchers.
Technical
Recordings were made with a SONY TCM-40DV cassette recorder and an inexpensive SONY external microphone and processed with Felt-Tip Sound Studio 2.0.6, QuickTime Pro 6.3, and Adobe GoLive 6. They are presented on this website in the ima 4:1 compression.

Results/Analysis
Linguistic
Phonetic
The phoneme list for Bantawa agreed upon by most authors (Bantawa 1985, Werner 2003) is one that fits with the data I gathered - "b bh ch chh, d, dh, (g), h, (j, jh), k, kh, l, m, n, µ (engma), p, ph, r, s, t, th, w, y or ia" where the -h indicates asipiration. The g, j, and jh are rare and possibly occur only in borrowed words. The µ (engma) sounds the same as the ng in the English word sing, but in Bantawa (unlike English) can also be used initially (see µa, µaksi, µaluµ). One phonetic enigma is whether Bantawa has retroflex phonemes T and D - the latest linguistic work suggests not, but the speakers assured me that it did, suggesting that they are so influenced by Nepali (which has retroflex T and D) and Devanagari (the script used for Nepali and Hindi) that they are trying to differentiate fictitious phonemes. Another possibility is that these are phonemes or allophones in an undescribed dialect of Bantawa.
Lexical (see wordlists)
In general, lexical items (words), and especially nouns, are the first things to be lost to language shift. verbs, and to an even greater extent grammatical structures, are as a rule only lost after significant word loss. (Crystal 2000, Trask 1996).
Most of my informants had trouble generating Bantawa words that they knew easily in Nepali (or even English), and I estimate that almost no one I talked to had a vocabulary of more than 1000 words (and few that many). Many basic words are known by no one or very few in the area (for example "lake," colour names, and numbers above two.) Although this is not something I examined in depth, one would also expect the scope of the remaining words to be limited (containing relatively few terms in the realms of technology, business, and science, as the speakers find it easier to adopt a Nepali (e.g. mech [chair]) or English (e.g. presser cuker [sic]) word for a new object rather than coining a new Bantawa word or compound.
Grammar/Morphology
In my limited time, I focused on compiling the wordlists rather than on learning grammatical structures, but there are a few points that come up on the wordlist that need explaining. Bantawa has first and second person dual pronouns; that is they distinguish between "we two" (uµkachi) and "we many" (uµkanchi) as well as between "you one person" (khana) "you two" (khanachi) and "you more than two" (khananinchi).
The suffix -wa is often added to object words, also the suffixes -ma and -pa are often affixed to nouns to indicate, respectively female and male gender (see mangpa, wapa, wama).
Language shift is also evident at grammatical and morphological levels, indicating, as I mentioned above, a more advanced state of language shift. Speakers in my area always used the Nepali conditional bhane to make conditional statements (example) and the Nepali comparative indicator bhande to make comparative statements. These two basic grammatical structures have been lost. Examples at the morphological level are the words am and uµ, the possessive Bantawa words equivalent to your and my, which at some point were reanalysed with the Nepal possessive suffix -ko to form the present day alternate forms amko and uµko - redundant and half-Nepali.

Social
I observed Bantawa being used almost only in the home, and even there Nepali words and phrases were often interspersed. Outside the home in the village, even in Bantawa to Bantawa conversation, Nepali was often used, and when speakers of other languages were around, it was considered a little rude to speak Bantawa and thereby exclude them. Nepali is always used in school, at the bazaar, and in other more "formal" situations. With few exceptions, the youth (18 years and younger) were fluent in Nepali, and used it often. In general, their Bantawa skill was about on par with their English.
There seemed to be a general trend of a higher level of knowledge of Bantawa with age, but not all old people spoke Bantawa well. One domain that the language was preserved the most in was in worship, or maµ. To paraphrase what one elder told me, "without our language we have no maµ. Without our maµ we have no culture".
This preservation is also evident in the mangpa, the Bantawa shamans. Their primary task is to communicate with spirits in order to cure various afflictions, (an oversimplification, see Maskarinec 1995 for an excellent study of Nepali shamanism) and I was told that if the shamans did not speak in the Bantawa language, the spirits would not understand. This, as well as the shaman's culturally proprietary job (they were Bantawa shamans, other castes had other shamans) contributed to the maµpa being among the most linguistically knowledgable people in the village. I was repeatedly told that in their chants the shamans used "old" "pure" or "literary" words [an odd choice of words for an oral language] that were difficult even for Bantawa speakers to understand.