The expressions fieldwork and informant1 work are sometimes used interchangeably. As a matter of fact, linguistic fieldwork is research in the field, i.e. in the speech community, whether or not data are obtained from informants, while informant work can be done everywhere in or outside his speech community.
There are many different settings in which a linguist may work with native speakers serving as informants on their language. The following is a list of practical considerations which may be heeded where possible and useful. They start from a set of presuppositions which may or may not be valid in particular cases:
- The linguist's task is the – complete or partial – description of a language, for which he collects and analyzes data. With a different research interest, e.g. a study of youth language or of the integration of loans, different presuppositions prevail.
- If the language is a minority language for which no full description is available, then the undertaking of describing it is by its very nature a conservative undertaking. The goal is to describe the language such as it has come down to the present speech community.
On these presuppositions, the following recommendations may be based:
- Priority attention is paid to the traditional form of the language.
- Just as in learning a foreign language, in doing fieldwork on a language it is best to get immersed in the speech community. Natural language data are best available in their natural context. Working with informants who are available at the researchers office desk is the second best solution.
- If the speech community is composed of several communities, e.g. villages or neighborhoods of towns, choose a community which leads a relatively traditional life with less contact with the dominant society.
- If one has a choice among a set of native speakers, older and conservative speakers have the preference. Especially with an endangered language, younger speakers are often semi-speakers.
- In many speech communities, working with informants of the opposite sex of the researcher – esp. of a younger researcher – is restricted or impossible. Then ways must be sought to overcome this barrier. Working in a mixed team is a good solution.
- It is in many respects easier to work with literate informants. Sometimes native school teachers are preferred. This has a set of disadvantages: Such persons usually have looser ties with the tradition of their speech communities, half of their personality being marked by their role in the dominant society. By their profession, they regard it as their duty and their capacity to distinguish between right and wrong in the language; i.e. they are averse to variation. In the extreme case, they teach the linguist linguistics.
- For whatever reason, the researcher may identify, over time, his favorite informant and limit his informant work to this person. This is counterproductive. No speaker dominates his language completely; different life experiences shape different linguistic competences (Lehmann 2007[L]). One speaker's data are biased; they must be counterbalanced by data from another speaker.
- An informant session with one native speaker is relatively sterile if compared with an informant session engaging two speakers. They can correct each other, one can dissipate the other one's doubts, together they can converge on a good characterization of the meaning and use of a linguistic form. By talking to each other, they inspire each other; their interaction generates a genuine added value above the result of separate work with each of them.
The observer's paradox
If the task is to obtain absolutely faithful data on natural speech events of the language or linguistic community, then the ideal option would seem to be to just observe such events without any interaction between the researcher and the speech community. This, however, is not feasible:
- As long as the researcher is present, even if only observing, he changes the situation, including importantly speech situations, by his mere presence. In terms of speech act theory, he is then a bystander. People will speak differently when a bystander is present, above all one who they know observes how they speak. As a consequence, this method precisely does not guarantee unbiased natural data.
- If the researcher is absent, and only a microphone or camera are there to record speech behavior, then either the people observed are aware of this; and then the observer's paradox strikes again; or else they do not know it, and then the procedure is incompatible with fieldwork ethics.
The only solution to this problem is to be constantly aware of it and to control the data obtained for a possible bias.
1 For readers finding a problem here, a solution may be, for the entire set of these pages: Find ‘informant’, replace by ‘consultant’. The same goes for ‘deaf’ vs. ‘differently hearing’, ‘he’ vs. ‘she’ etc.