himmelmann documentary and descriptive linguistics .Documentary and descriptive linguistics 163...
Embed Size (px)
Transcript of himmelmann documentary and descriptive linguistics .Documentary and descriptive linguistics 163...
Documentary and descriptive linguistics*
NIKOLAUS P. HIMMELMANN
Much of the work that is labeled "deseriptive" within linguistics comprisestwo activities, the collection of primary data and a (low-level) analysis ofthese data. These are indeed two separate activities as shown by the factthat the methods employed in each activity differ substantially. To date,the field concerned with the first aetivity called "doeumentary linguisties"here has received very little attention from linguists. It is proposed thatdocumentary linguistics be conceived of as a fairly independent field oflinguistic inquiry and practice that is no longer linked exclusively to thedescriptive framework. A format for language documentations (in contrastto language deseriptions) is presented (section 2), and various practicaland theoretical issues connected with this format are discussed. Theseinclude the rights of the individuals and communities contributing to alanguage doeumentation (section 3.1), the parameters for the selection ofthe data to be included in a doeumentation (section 3.2), and the assessmentof the quality of such data (section 3.3),
1. Distinguishing description and documentation
This article presents some reflections on the framework of descriptiveHnguistics. My concern is the apphcation of this framework for recordinglittle-known or previously unrecorded languages. Most of these languagesare endangered, and the present reflections have been occasioned in partby the recent surge of interest in endangered languages (cf., for example,Robins and Uhlenbeck 1991; Hale et al. 1992) and the concomitant callfor descriptive work on these languages.
The task of recording a little-known language comprises two activities,the first being the collection, transcription, and translation of primarydata and the second a low-level (i.e. descriptive) analysis of these data.
Linguistics 36 (1998). 16M95 0024-3949/98/0036-0161 Walter de Gruyter
162 A'. Hinimclinann
The two activities differ substantially with respect to the methodsemployed as well as to their immediate results. The collection of primarydata may, for example, result in a sample of some 50 utterances, in manyof which a segment lu occurs. The methods used in putting together thissample may include participant observation (jotting down overheardutterances), various forms of elicitation, or recording, transcribing, andtranslating a text. Methodological issues arise with respect to the reliabil-ity, naturalness, and representativeness of the data.
The second activity the analysis of these primary data leads tostatements such as that in language L an ergative case exists, formallyexpressed by a suffix /, which is part of a case paradigm and has suchand such further formal and semantic properties. The procedures usedin arriving at such statements involve distributional tests (commutation,substitution, etc.), the analysis of the semantic properties of the utterancescontaining Ui, etc. Methodological issues arise with respect to the defini-tion of the notions 'ergative', 'suffix, etc., and the kind of evidenceadduced for analyzing a certain segment as an ergative suffix.
Table 1 provides a synopsis of some of the conspicuous differencesbetween collecting primary data and providing a descriptive analysis ofthese data.
Despite the differences listed in Table 1, the two activities are alsoclosely interrelated and partially overlap for various epistemological,methodological, and practical reasons. The most important area of over-lap pertains to the transcription of primary data. Any transcriptionrequires some kind of orthographic representation. This representationwill be informed by at least a preliminary phonetic and phonological
, differ collecting cificl aiudvziii y linguistic
v^orpus 01 uttcriiiicesi nott
speaker and compiler on iparticular form or constriParticipant observation,
transcription and translatiprimary data
Descriptive statements,illustrated by one or twoexamples
Phonetic, phonological,morphosyntactic. andsemantic analyses(instrumental measurings.distributional tests, etc.)Definition of terms andlevels, justification(adequacy) of analysis
Documentary and descriptive linguistics 163
analysis. FurtheiTnore, any transcription involves decisions about seg-mentation (words, intonation units, turns, clauses, sentences/paragraphs).All of these decisions presuppose a certain amount of analysis on variouslevels. Some analysis is also involved in translation. In particular, aninterlinear translation obviously presupposes some kind of morphologicalanalysis.
Because of these interrelations and overlaps, a strong tendency existswithin descriptive linguistics to blur the differences between the twoactivities and to consider them part of a uniform project called "describ-ing a language." There are, however, various reasons to keep the twoactivities clearly separate or, more generally, to distinguish between thedocumentation and the description of a language. The major reasonpertains to the methodological differences between the two activities listedin Table 1. The methodological differences are mirrored by the substantialdifferences between the products of the two activities, a language docu-mentation and a language description, respectively, which will be high-lighted in section 2.1 below. Other reasons include the following threearguments.
First, no automatic, infallible procedure exists for deriving descriptivestatements from a corpus of primary data; that is, any collection ofprimary data allows for various kinds of analyses even within the frame-work of descriptive linguistics.^ Therefore, a data collection and its analy-sis are not just simply two different ways of presenting the sameinformation. This is obviously nothing new to linguistics but, rather,belongs to the few general assumptions shared by most, if not all, linguistssince the failure of the post-Bloomfieldian discovery procedures project.
Second, a comprehensive deseriptive analysis is not the only kind ofanalysis possible for a given set of primary data. A set of primary datamay be of interest to various other (sub-)disciplines, including socio-linguistics, anthropology, discourse analysis, oral history, etc. This, ofcourse, presupposes that the data set contains data and informationamenable to the research methodologies of these diseiplines. The chancethat this kind of data and information is found in a language descriptionis practically nil. Language descriptions are, in general, useful only togrammatically oriented and comparative linguists. Collections of primarydata have at least the potential of being of use to a larger group ofinterested parties. These include the speech community itself, which mightbe interested in a record of its linguistic practices and traditions.
Third, as long as collection and analysis are considered part of a single,uniform project, the collection activity is likely to be (relatively) neglected.Historically speaking at least, it has been the case that the collectioriactivity has never received the same attention within descriptive linguistics
164 N. Himmclmann
as the analytic activity. Descriptive theory has almost exclusively beenoccupied with the procedures for analyzing primary data and presentingthis analysis (in the format of a grammar and a lexicon). Methodologicalissues with respect to obtaining and presenting primary data have neverbeen dealt with in depth within descriptive linguistics.^ The presentation(publication) of the primary data has generally been considered a second-ary task. In recent decades, hardly any comprehensive collections ofprimary data have been published.^ A clear separation between documen-tation and description will ensure that the collection and presentation ofprimary data receive the theoretical and practical attention they deserve.
Much more fundamental objections could be raised against the ideathat language documentation and language description are part of asingle, uniform project. The essence of such objections pertains to thefact that any close link between these two activities has the consequenceof the descriptive concept of language determining the kind of dataconsidered relevaht in language documentation. Consequently, any objec-tions raised against the descriptive concept of language as a system ofunits and regularities will also apply to a language documentation donewithin the descriptive framework. As is well known, the descriptiveconcept of language has been criticized from various points of view, witha notable increase of criticism in recent times.'' Among the targets ofsuch criticism are its abstract and ahistoric conception of the speechcommunity as a homogeneous body, its neglect truly to confront thecomplexities of spoken language (rather than reducing spoken languageto "language as it may be written down")^ and the concept of a languageas an overall coherent system.
I will not, however, pursue this series of objections in detail. In myopinion, the arguments given above suffice in estabhshing the need fordistinguishing language documentation from language description.Further, indirect support for such a distinction will be found whilespelling out the details of a framework for language documentation inthe following sections. Needless to say, any discussion of a frameworkfor language documentation should be informed by the objections leveledagainst the descriptive concept of language.^ The following sections out-line such a framework. Throughout these sections, I will continue tocontrast language documentation with language description in order toprofile major characteristics of documentary linguistics.
2. Language documentation and documentary linguist