1. Analysieren Sie die Mikrostruktur der Einträge
so genau wie möglich. Unterscheiden Sie dabei zwischen form-
und inhaltsbasierten Elementen!

2. Kodieren Sie einen dieser Wörterbuchartikel in XML, indem Sie die fol-
gende DTD verwenden:
<!ELEMENT dictionary (entry)+>
<!ELEMENT entry(hom+)>
<!ATTLIST entry id ID #REQUIRED>
<!ELEMENT hom (form, sense+)>
<!ATTLIST hom id CDATA #REQUIRED>
<!ELEMENT form (orth, (pron | grammar)*)>
<!ELEMENT grammar (pos | (gender, infl?))+>
<!ELEMENT sense (trans | phrase | xr)+>
<!ATTLIST sense id CDATA #REQUIRED>
<!ELEMENT trans (usg?, arg*, tr, context*)+>
<!ELEMENT tr (equiv, usg?)>
<!ELEMENT context (source, target)>
<!ELEMENT phrase (orth, trans+)>
<!ELEMENT xr (orth, sense-id?)>
<!ATTLIST xr idref CDATA #REQUIRED>
<!ELEMENT pos (n|v|a|adv)>
<!ELEMENT gender (#PCDATA)>
<!ELEMENT orth (#PCDATA)>
<!ELEMENT pron (#PCDATA)>
<!ELEMENT usg (#PCDATA)>
<!ELEMENT arg (#PCDATA)>
<!ELEMENT infl (#PCDATA)>
<!ELEMENT source (#PCDATA)>
<!ELEMENT target (#PCDATA)>
<!ELEMENT equiv (#PCDATA)>
<!ELEMENT sense-id (#PCDATA)>

Hinweise zur Arbeit (die vollständige und dokumentierte DTD finden Sie
unten):
• Segmentieren und analysieren Sie den Artikel.
• Übernehmen Sie die Artikelteile, die sich mit der vorhandenen DTD
annotieren lassen.
• Wenn ein Artikelteil nicht in die vorhandene DTD passt, dann vermer-
ken Sie dies bitte.


<!-- DTD for bilingual dictionary articles-->
<!-- Most of the element names used are conform with the TEIdict2.dtd -->

<!ELEMENT dictionary (entry)+>
<!ELEMENT entry   (hom+)>
<!ATTLIST entry id ID #REQUIRED>

<!-- The entry has a form section and one or more homonym readings; an example for a homonymous lexeme is "Pony": a kind of horse, gender=neuter; a kind of haircut, gender=masculine -->

<!ELEMENT hom   (form, sense+)>
<!ATTLIST hom id CDATA #REQUIRED>

<!-- Every homograph reading consists of a section about the form of the headword and at least one senses (or "readings") -->

<!ELEMENT form   (orth, (pron | grammar)*)>

<!-- the form section must contain the orthographic representation of the lemma, followed by, in no particular order, information about the pronunciation of the headword and grammatical information. Both these types of information are not obligatory. If you specify them in the head, they should hold for all senses -->

<!ELEMENT grammar (pos | (gender, infl))+>

<!-- the grammatical information in our article is restricted to part of speech (for all noun-nouns) and gender plus inflection (for the nouns) -->

<!ELEMENT sense   (trans | phrase | xr)+>
<!ATTLIST sense id CDATA #REQUIRED>

<!-- a sense conists of groups of target language equivalents, phrases with the headword as a part, and cross references, in no particular order -->

<!ELEMENT trans  (usg?, arg*, tr, context*)+>

<!-- a target language equivalent can be accompanied by a usage note and a syntactic argument and a context marker  - in both the source and the target language - as further conditions (or constraints) on the use of this equivalent. Al these additional data can be seen as conditions on the translation -->

<!ELEMENT tr (equiv, usg?)>

<!-- The translation equivalent itself can be accompanied by a usage note -->

<!ELEMENT  context (source, target)>

<!-- Information about context restrictions are given in both source and target language -->

<!ELEMENT phrase (orth, trans+)>

<!-- The orthographical form of the source language phrase is given, together with at least one target language equivalent -->

<!ELEMENT xr (orth, sense-id?)>
<!ATTLIST xr idref CDATA #REQUIRED>

<!-- an xr points to a particular sense on another entry specified by its orthographic form -->

<!ELEMENT pos (n|v|a|adv)>
<!ELEMENT gender (#PCDATA)>
<!ELEMENT orth (#PCDATA)>
<!ELEMENT pron (#PCDATA)>
<!ELEMENT usg (#PCDATA)>
<!ELEMENT arg (#PCDATA)>
<!ELEMENT infl (#PCDATA)>
<!ELEMENT source (#PCDATA)>
<!ELEMENT target (#PCDATA)>
<!ELEMENT equiv (#PCDATA)>
<!ELEMENT sense-id (#PCDATA)>

3. Bei der Analyse von Wörterbuchdefinitionen sind Bedeutungsbeschrei-
bungen für Nomen oft schwierig zu parsen. Vossen et al. (1989) haben das
sog. Kontrollvokabular des LDOCE, das ca. 2 000 Wörter umfasst, die
in den Definitionen zur Bedeutungsexplikation verwendet werden, unter-
sucht und aufbereitet. Die sog. Kernels (Headwords) in den Definitionen
werden unterschiedlich, je nach ihrem semantischen Gehalt, klassifiziert.
Mitunter gibt es Ambiguitäten (vgl. die of-Phrasen-Modifikation inner-
halb der Definitionen).

Anhand der folgenden Beispiele sollen Gruppen analog typischer Struk-
turen gebildet und informell erklärt werden:

• cocktail: „a mixed alcoholic drink“
• bourgeois: „a member of the middle class“
• air mattress: „airbed“
• author: „the writer of a book, newspaper article, play, poem, etc.“
• ambiguity: „the condition of being ambiguous“
• destroyer: „a person who destroys“
• abbey: „the group of people living in such a building“
• detonation: „the noise of an explosion“
• discouragement: „something that discourages“
• car: „automobile