22
Nikolaus P. Himmelmann
Latin, Chinese, Arabic, Tagalog, Quechua, or Fijian, there exist not only
different types of grammars (pedagogical, historical, descriptive) but also
different descriptive grammars, each having its particular emphasis and
way of presenting the structure of the language system. This simply reflects
the fact that at least according to the current state of knowledge, there is not
just exactly one descriptive grammar which correctly and comprehensively
captures the system of a language. Instead, any given descriptive grammar
is a more or less successful attempt to capture the system of a language
(variety), rarely if ever comprehensive, and usually also including at least
some contested, if not clearly wrong, analyses.
As a consequence of this state of affairs, the following problem arises
with regard to the extended format for language documentations in Table 2.
Either one has to specify a particular type of descriptive grammar as the
one which is the most suitable one for the purposes of language documenta-
tions and thus is able to provide a reasonably precise definition of this part
of a documentation. Alternatively, one allows for a multitude of descriptive
grammars to be included in a documentation, thus declaring it a desirable
goal to include a number of different analyses of the language system as part
of the overall documentation of a language. The latter option clearly raises
the issue of practical feasibility, which leads us to the second problem men-
tioned above, i.e. the essentially pragmatic problem of research economy.
Practical feasibility also is an issue if just one analysis of the grammati-
cal system is assumed to be an essential part of a language documentation,
for the following reason. It is a well-known fact that it is possible to base
elaborate descriptive analyses exclusively on a corpus of texts (either texts
written by native speakers or transcripts of communicative events) – and
most good descriptive grammars are based to a large degree on a corpus of
(mostly narrative) texts. A large corpus of texts in fact provides for the pos-
sibility of writing a number of interestingly different descriptive grammars,
targeting different components of the language system and their interrela-
tion. Consequently, one could argue that even if one accepts the claim that a
comprehensive documentation should also document the language system,
there is no need to include a fully worked-out descriptive grammar in a
language documentation. The information needed to write such a grammar
is already contained in the corpus and the resources needed to extract this
information and to write it up in the conventional format of a descriptive
grammar are not properly part of the documentation efforts. In this view,
resources allocated to documentation should not be “wasted” on writing a
grammar but are better spent on enlarging the corpus of primary data, the
Chapter 1 – Language documentation: What is it and what is it good for?
23
quantity or quality of annotations, or on the “mobilization” of the data (mo-
bilization is further discussed in Chapter 15).
The major counterargument against this position would be the claim that
actually producing a descriptive grammar is a necessary part of a language
documentation because otherwise, essential aspects of the language system
would be left undocumented. The evaluation of this claim rests on the ques-
tion of whether there is some kind of important evidence for grammatical
structure which, as a matter of principle, cannot be extracted from a suffi-
ciently large and varied corpus of primary data as sketched in Section 3
above. As far as I am aware, there is especially one type of evidence of this
kind, i.e. negative evidence. Obviously, illicit structures cannot be attested
even in the largest and most comprehensive corpora.
15
However, the lack of explicit negative evidence in a corpus of texts does
not per se necessitate the inclusion of a descriptive grammar in a language
documentation. On the one hand, with regard to the usual way of obtaining
negative evidence (i.e. asking one or two speakers whether examples x, y, z
are “okay”), it is doubtful whether this really makes a difference in quality
compared to evidence provided by the fact that the structure in question is
not attested in a large corpus. Elicited evidence is only superior here if it is
very carefully elicited, paying adequate attention to the sample of speakers
interviewed, potential biases in presenting the material, and the like. On the
other hand, and more importantly, the basic documentation format of Table
1 does not only consist of a corpus of more or less natural communicative
events but also of documents recording metalinguistic knowledge. Metalin-
guistic knowledge includes negative evidence for grammatical structuring,
as already mentioned above.
Obviously, gathering negative evidence on grammatical matters presup-
poses that the researcher asks the right questions, which in turn presupposes
grammatical analysis. In this regard, it bears emphasizing that documenta-
tion does not exclude analysis. Quite the opposite: analysis is essential. What
the documentary approach implies, however, is that the analyses which are
carried out while compiling a documentation do not necessarily have to be
presented in the format of a descriptive grammar. Instead, analyses can (or
should) be included in a documentation through (scattered) annotations on
negative evidence, the inclusion of experiments generating important evi-
dence for problems of grammatical or semantic analysis, and so on (see
further Chapters 8 and 9).
The major reason for choosing a distributed grammatical annotation
format instead of the established descriptive grammar format is one of time