302
Ulrike Mosel
introductory grammar chapter faces the problem of identifying the kind of
information the reader needs in order to understand the specialized investi-
gation in question. Since a specialized investigation focuses on a certain
type of linguistic phenomenon, the introductory grammar chapter can be
selective and concentrate on those aspects of grammar that are essential for
understanding the data presented in the main part of the book, but are not
its topic. Consequently, a book on tense and aspect may lack information
on the derivation of nominals because it is irrelevant, whereas a book on
morphology would not deal with derivation in the introductory grammar
chapter, because this constitutes an important part of the body of the book.
4
The main difference between these first two types is that in practice the
linguist starts writing the preliminary grammar as soon as he or she puts the
first analyses of paradigms and constructions into words, whereas the intro-
ductory grammar chapter is written after the research on the specialized area
in question has been concluded. Only then can the author identify which
grammatical information the reader will need to understand the investiga-
tion. Similarly, the summary or concise version of a reference grammar
only contains a selection of an already existing analysis of the language.
However, the selection criteria are different. Here the grammatical phenom-
ena to be described are not chosen with regard to the presentation of some
other research areas. Rather, the author selects what she or he considers as
essential features of the language from a more general perspective (see Quirk
et al. 1985: 37–91, Mosel and Hovdhaugen 1992: 49–65, for an example).
The fourth type of sketch grammar, the dictionary grammar “brings to-
gether elements of the language that are separated by the alphabetical order
of the headwords – typical examples are a list of numerals or an overview
of derivational means of expression” (Mugdan 1989: 732, translation
U. M.). Such informations can help to save space within the main body of
the dictionary. For example, if the language has a regular and productive
affix to derive diminutives from nouns, it is more economic to describe this
affix and its use in the grammar than listing all diminutives as headwords
or subentries (see Mugdan 1989 for a detailed discussion on dictionary
grammars).
The sketch grammar of a language documentation is a bit of everything.
Through most stages of the process of documenting the language, it is a
preliminary grammar that needs constant revising. Similar to introductory
grammar chapters and dictionary grammars, it is closely related to some
other major piece of linguistic work, in this case the corpus of annotated
recordings and a lexical database. And it also resembles the summary of a
Chapter 12 – Sketch grammar
303
reference grammar and the dictionary grammar in that it enhances the ac-
cessibility of the work for a more general readership.
2. The aims of a sketch grammar in language documentation
The LDSG is not a so-called “short grammar” as it does not aim at describ-
ing the language as a system but at facilitating the access of the documenta-
tion. Though not being an integral component of the main body of the
documentation, the LDSG is, as a kind of user’s guide, essential for the
utility value of the language documentation and therefore should be taken
seriously from the very beginning. In particular, the sketch grammar should
–
contain all the grammatical information that the reader needs to make use
of the lexical database and understand how in the corpus of annotated
recordings the translations relate to the transcriptions;
–
at any stage accurately reflect the author’s current knowledge of the
language;
–
be user-friendly.
2.1. Comprehensiveness
As a documentation should contain a fair amount of recordings that are
annotated by interlinear morphemic glosses, the amount of information
given in the sketch grammar partly depends on how much of this informa-
tion is provided by the annotations (for a detailed account of annotations
see Chapter 9).
For example, the Teop language has a very complex system of articles
that indicate noun class, number, specificity, and grammatical relations.
There are in principle two options to inform the reader about the use of
Teop articles. Firstly, all the grammatical properties of articles can be ex-
plained in the sketch grammar by giving the full paradigm, in which case it
is sufficient to gloss the articles in the text corpus simply as
ART
, e.g.
(1)
bona moon
ART
woman
so that
ART
only denotes the word/morpheme class of the glossed mor-
pheme, but not its particular grammatical features.
304
Ulrike Mosel
Secondly, one can indicate all grammatical features in the gloss, e.g.
(2)
bona
moon
ART
:
A
-
CLASS
.
SPEC
.
SG
.
OBJ
woman:
A
-
CLASS
‘specific singular object article agreeing with a noun belonging to the
A.class’
in which case the explanations in the grammar can be confined to explain-
ing the abbreviations. For three reasons we preferred the first solution in the
Teop project. Firstly, after we had done the analysis which was necessary
in any case, writing down the paradigm did not take much time; secondly,
the paradigm represents the grammatical features of all articles comprising
the category
ART
in a systematic way, and thirdly, long glosses as in (2) are
not userfriendly. They take so much space that only a few words fit into
one line and thus make it difficult to capture the form-meaning relations of
a sentence at a quick glance.
2.2. Accuracy
The sketch grammar of a language documentation cannot provide an abso-
lutely accurate description of linguistic facts because this would presuppose
a comprehensive grammatical analysis of the entire corpus and presumably
additional research on particular grammatical phenomena that are not fully
covered by the corpus (see Chapter 1). However, as the following example
illustrates, the sketch grammar should aim at accuracy in the sense of ex-
plicitly accounting for the assumptions that underlie the grammatical anno-
tations in the corpus.
When starting with the annotation of Teop recordings, we found hun-
dreds of examples where the particle
paa seemed to be a tense/aspect/mood
(
TAM
) marker locating the reported event in the past, e.g.
(3)
… me iaa
paa gigo
anaa Solomon
and
mum
TAM
pregnant me Solomons
‘… and mum got pregnant with me in the Solomon Islands.’
(Sii 2.114 –116 R)
Accordingly we glossed paa as
PAST
, ignoring a very few exceptions. Fur-
ther on, however, we discovered more and more examples like