This document
contains a detailed specification of morphosyntactic properties
of German word forms.
This specification is based on the more general inventory of
descriptive labels for morphosyntax which has been proposed by the
EAGLES Work Group on Computational Lexicons, for all EU languages. It
is thus the German incarnation of the EAGLES standard morphosyntactic
classification proposal. We refer to it as ELM-DE (for:
EAGLES/Lexicon/Morphosyntax-Proposal, for German). Similar
specifications have been developed for French (ELM-FR,
[&make_named_href('',
"node323.html#vonRekowski95","[von Rekowski 95]")]) and Italian (ELM-IT, [&make_named_href('',
"node323.html#Monachini95","[Monachini 95]")]), as
well as, somewhat less detailed, for English (ELM-EN,
[&make_named_href('',
"node323.html#Teufel95","[Teufel 95]")]). The [&make_named_href('',
"node323.html#moca","[Monachini, Calzolari 93]")]
document is multilingual and application independent; the present
document, however, is monolingual and moderately oriented towards
lexical classification and corpus tagging.
The specifications are presented here in textual and tabular form, but
a formal encoding could be produced without any problems; a PROLOG
encoding of a type hierarchy for ELM-EN has indeed been
provided.
Along with the declarative description of the morphosyntactic classifications proposed by EAGLES, we also want to give hints for the practical use of the classification in concrete applications, such as lexicon building or corpus annotation. To this end, a number of delimitation tables have been included which allow the human user to more clearly grasp details of the distinctions we propose, often by contrasting word form, which are closely related but receive different analyses. This makes the present document usable as guidelines for practical classification and annotation work.
This document and the specifications it embodies owe much to the
EAGLES morphosyntax label inventory given in
[&make_named_href('',
"node323.html#moca","[Monachini, Calzolari 93]")], but also, and maybe even more, to the
application of lexical description in corpus tagging: the IMS tagset
(cf. [&make_named_href('',
"node323.html#Schiller94","[Schiller 94]")]) and the current Stuttgart-Tübingen-TagSet,
STTS (cf. [&make_named_href('',
"node323.html#STT","[Schiller, Teufel, Thielen 94/95]")]) have massively
influenced our work, both in terms of specifications and of the
presentation of the guidelines part.