8
Nikolaus P. Himmelmann
The theoretically grounded sampling procedure will be determined to a
significant degree by the purposes and goals of the particular project. The
rather broad and unspecific goal of a lasting, multipurpose record of a lan-
guage envisioned here implies that, as much as possible, a sufficiently large
number of examples for every type of communicative event found in a
given speech community is collected. This in turn raises the highly complex
issue of how the typology of communicative events in a given speech com-
munity can be uncovered. Within sociolinguistics, the framework known as
the ethnography of communication provides a starting point for dealing
with this issue. Chapter 5 provides a brief introduction to major concepts
relevant here. Chapter 8 lists a range of important topics and parameters.
Besides observable linguistic behavior, is there anything else that needs
to be documented in order to provide for a lasting, multipurpose record of a
language? Or can all relevant information be extracted from a comprehen-
sive corpus of recordings of communicative events? One aspect of “a lan-
guage” that is not, or at least not easily, accessible by analyzing observable
linguistic behavior is the tacit knowledge speakers have about their lan-
guage. This is also known as metalinguistic knowledge
and refers to the
ability of native speakers to provide interpretations and systematizations for
linguistic units and events. For example, speakers know that a given word
is a taboo word, that speech event X usually has to be followed by speech
event Y, or that putting a given sequence of elements in a different order is
awkward or simply impossible. Similarly, metalinguistic knowledge as
understood here also includes all kinds of linguistically based taxonomies,
such as kinship systems, folk taxonomies for plants, animals, musical in-
struments and styles, and other artifacts, expressions for numbers and
measures, but also morphological paradigms.
The documentation of metalinguistic knowledge, while not involving
principled theoretical or ethical problems, is also not a straightforward task
because much of it is not directly accessible. To be sure, in some instances
there are conventional speech events involving the display of metalinguistic
knowledge, such as reciting a genealogy or lengthy mythological narratives
which sketch a cognitive map of the landscape. In many societies, there are
also a number of well established and much discussed topics where speakers
engage in metalinguistic discussions regarding the differences between dif-
ferent varieties (in village X they say “da” but we say “de”; young people
cannot pronounce our peculiar /k/-sound correctly anymore, etc.). Further-
more, transcripts prepared by native speakers without direct interference by
a linguist often provide interesting evidence regarding morpheme, word,
Chapter 1 – Language documentation: What is it and what is it good for?
9
and sentence boundaries (see Chapters 3 and 10 for further discussion). But
very often documenting metalinguistic knowledge will involve the use of a
broad array of elicitation strategies, guided by current theories about different
kinds of metalinguistic knowledge and their structure. One very important
type of elicited evidence are monolingual definitions of word meanings
provided by native speakers. See Chapters 3 and 6 for further discussion
and exemplification.
The documentation of metalinguistic knowledge as understood here in-
cludes much of the basic information that is needed for writing descriptive
grammars and dictionaries. In particular, it includes all kinds of elicited
data regarding the grammaticality or acceptability of phonological or mor-
phosyntactic structures and the meaning, use, and relatedness of lexical
items. However, it should be clearly understood that documentation here
means that the elicitation process itself is documented in its entirety, in-
cluding the questions asked or the stimuli presented by the researcher as
well as the reaction by the native speaker(s). That is, documentation per-
tains to the level of primary data which provide evidence for metalinguistic
knowledge, i.e. what native speakers can actually articulate regarding their
linguistic practices or their recordable reactions in experiments designed to
probe metalinguistic knowledge.
3
A grammatical rule as stated in a grammar
or an entry in a published dictionary are not primary data in this sense, even
though some linguists may believe that they are part of a native speakers’
(unconscious) metalinguistic knowledge. In this view, grammatical rules
and dictionary entries are analytical formats for metalinguistic knowledge.
Whether and to what extent these have a place in a language documentation
is an issue we will take up in Section 4.2.
It is also worth noting that the documentation of observable linguistic
behavior and metalinguistic knowledge are similar
in that they basically
consist of records of communicative events. In the case of observable lin-
guistic behavior, the communicative event involves the interaction of native
speakers among themselves, while in the case of metalinguistic knowledge
it involves the interaction between native speakers and documenters. There
is a superficial difference with regard to the preferred documentation for-
mat in that it is now standard practice to make (video) recordings of ob-
servable linguistic behavior, while for the elicitation of metalinguistic knowl-
edge it is still more common simply to take written notes. In principle,
(video-)recording would also be the better (i.e. more reliable and compre-
hensive) documentation format for elicited metalinguistic knowledge, but
there may often be practical reasons to stay with paper and pencil (among