Sally Thomason gave a plenary talk at the 26th annual University of Wisconsin-Milwaukee Linguistics Symposium: Language Death, Endangerment, Documentation, and Revitalization, in October 2011. The title of her talk was "How to Avoid Pitfalls in documenting endangered languages."
Some of the issues that arise in documenting endangered languages are the same ones that come up with any primary documentation project -- making initial contacts with the community, getting permission to conduct fieldwork, organizing a field session, preparing carefully for each session, selecting techniques for collecting lexical and grammatical data, trying hard not to offend anyone (e.g. by introducing taboo words and concepts), and so forth. This talk will focus on some of the special circumstances that make fieldwork on an endangered language especially challenging. Among these circumstances are the need to start by approaching community leaders rather than potential language consultants; the need for extra flexibility in organizing the work, given the (probable) advanced age and precarious state of health of the consultants; the need to use a variety of methods in collecting data from elderly speakers who might not have used their language regularly for decades; the need to accept most of the variant forms provided by different speakers, even when the speakers disagree among themselves about the acceptability of some variants; and the need to ensure that the results of the research are usable by the community as well as by linguists. In addition, in some cases -- both with research on nonstandard dialects and with research on languages that haven't (yet) been standardized -- processes of standardization complicate speakers' assessment of certain forms that they and other community members produce, and the linguist must deal with the effects of varying grammaticality judgments that arise from this cause. Most of my examples will be drawn from my own fieldwork in two very different situations: documentation of word-formation patterns in endangered nonstandard dialects of the language formerly known as Serbo-Croatian; and, more extensively, primary documentation of the Salish-Pend d'Oreille (Montana Salish) language that is spoken on the Flathead Reservation in northwestern Montana. The youngest speakers I worked with in the former Yugoslavia were in their 60s, had never attended school, and were not all literate; the youngest speakers I work with in Montana now are in their mid-70s, are fully fluent in English as well as in their native language, and are fully literate in English. It has been two or more generations since any children were raised mainly speaking Salish-Pend d'Oreille, so that the language as spoken by the few remaining traditional native speakers will be gone within the next 20-30 years.