Lexical specifications : application to German Petra Steiner and Lothar Lemnitzer - MS

The application of the MULTEXT encoding scheme to German attempts to keep as close as possible to the conventions. However, some deviations were unavoidable. This concerns:

We will try to justify the changes, or mark them as language-specific.

Contents

• 1. Nouns (N)
• 1.1. Lexicon
• 1.2. Corpus
• 1.3. Combinations
• 2. Verbs (V)
• 2.1. Lexicon
• 2.2. Corpus
• 2.3. Combinations
• 3. Adjectives (A)
• 3.1. Lexicon
• 3.2. Corpus
• 3.3. Combinations
• 4. Pronouns (P)
• 4.1. Lexicon
• 4.2. Corpus
• 4.3. Combinations
• 5. Determiners (D)
• 5.1. Lexicon
• 5.2. Corpus
• 5.3. Combinations
• 6. Articles (T)
• 6.1. Lexicon
• 6.2. Corpus
• 6.3. Combinations
• 7. Adverbs (R)
• 7.1. Lexicon
• 7.2. Corpus
• 7.3. Combinations
• 8. Adpositions (S)
• 8.1. Lexicon
• 8.2. Corpus
• 8.3. Combinations
• 9. Conjunctions (C)
• 9.1. Lexicon
• 9.2. Corpus
• 9.3. Combinations
• 10. Numerals (M)
• 11. Interjection (I)
• 11.1. Lexicon
• 11.2. Corpus
• 11.3. Combinations
• 12. Add on Classes
• 12.1. Particle (Q)
• 12.1.1. Lexicon
• 12.1.2. Corpus
• 12.1.3. Combinations
• 12.2. Punctuation (F)
• 12.2.1. Corpus
• 12.3. Abbreviations (Y)
• 12.3.1. Lexicon
• 12.3.2. Corpus
• 12.4. Others (X)
• 12.5. Unique Membership class (U)
• 12.4.1. Corpus

1. German Nouns (N)

1.1. Lexicon

============ =========== =========== ====
Attribute      Value       Example   Code
============ =========== =========== ====
Type           common       Buch       c
               proper       Peter      p
------------ ----------- ----------- ----
Gender         masculine    Mann       m
               feminine     Frau       f
               neuter       Kind       n
------------ ----------- ----------- ----
Number         singular     Mann       s
               plural       Frauen     p
------------ ----------- ----------- ----
Case           nominative   Kind       n
               genitive     Kindes     g
               dative       Kinde      d
               accusative   Kind       a
============ =========== =========== ====

1.2. Corpus

1.3. Combinations

Compare with :

• English nouns
• French nouns
• Spanish nouns
• Italian nouns

2. German Verbs (V)

2.1. Lexicon

============ =========== =========== ====
Attribute      Value       Example   Code
============ =========== =========== ====
Type           main         gehen      m
               modal        sollen     o    this value has been added
               auxiliary    haben      a
------------ ----------- ----------- ----
Mood           indicative   geht       i
               subjunctive  gehe       s
               imperative   geht       m
               infinitive   gehen      n
               inf. with inc.               this value has been added
               particle     wegzugehen u
               participle   gehend     p
------------ ----------- ----------- ----
Tense          present      geht       p
               imperfect    ging       i
------------ ----------- ----------- ----
Person         first        bin        1
               second       bist       2
               third        ist        3
------------ ----------- ----------- ----
Number         singular     geht       s
               plural       gehen      p
------------ ----------- ----------- ----
Gender          ///         ///        -
------------ ----------- ----------- ----
Clitic         no           hat        n
               yes          hats       y
============ =========== =========== ====

Notes:

Gender :There is no distinction in gender for the third person singular.

2.2. Corpus

2.3. Combinations

Note:

Adding participles here would imply the addition of adjective features. Morphologically at least, participles behave like adjectives. A good - but not very elegant - solution would therefore be to handle participles as a type of adjective. Secondly there are ambigous cases, e.g.

where gerührt is a participle, but might be tagged either as a verb or an adjective, depending on the context. This would be violating the assumption for applicativeness. The best solution at hand is to treat these forms as ambiguous with respect to membership in a word class (adjective and verb, respectively). However, this is a mixture of morphosyntactic with distributional criteria, and therefore unsatisfactory.

Compare with :

• English verbs
• French verbs
• Spanish verbs
• Italian verbs

3. German Adjectives (A)

3.1. Lexicon

============ =========== =========== ====
Attribute      Value       Example   Code
========== ============= =========== ====
Type        qualificat.   gut          f
            ordinal       zweites      o
            cardinal      zwei         c    this value has been added
            possessive    mein         s
            part1         lachende     1    this value has been added
            part2         gesungene    2    this value has been added
------------------------ ------------ -----
Degree      positive      gut          p
            comparative   besser       c
            superlative   beste        s
------------------------ ------------ -----
Gender      masculine     guter        m
            feminine      gute         f
            neuter        gutes        n
------------------------ ------------ -----
Number      singular      guter        s
            plura         gute         p
------------------------ ------------ -----
Case        nominative    guter        n
            genitive      guten        g
            dative        guten        d
            accusative    guten        a
========== ============= =========== ====
          

Notes:

and

to the possessive Determiner or Pronouns because 'mein' originally is a possessive pronoun, we recognize that the present definition of these categories does not allow the addition of the value 'predicative'. This treatment is therefore a compromise.

3.2. Corpus

3.3. Combinations

Compare with :

• English adjectives
• French adjectives
• Spanish adjectives
• Italian adjectives

4. German Pronouns (P)

4.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example   Code
========== ============= ============ =====
Type        personal       ich          p
            demonstrat.    dieser       d
            indefinite     kein         i
            interrog.      wer          t
            relative       der          r
            reflexive      sich         x
---------- ------------- ------------ -----
Person      first          ich          1
            second         du           2
            third          es           3
---------- ------------- ------------ -----
Gender      masculine      dieser       m
            feminine       diese        f
            neutre         dieses       n
---------- ------------- ------------ -----
Number      singular       dieser       s
            plural         diese        p
---------- ------------- ------------ -----
Case        nominative     dieser       n
            genitive       dieses       g
            dative         diesem       d
            accusative     diesen       a
---------- ------------- ------------ -----
Possessor        -           -          -
========== ============= ============ =====

Notes

Possessive : In German there are no possessives in pronominal use, so we do not need the attribute possessor here.

4.2. Corpus

4.3. Combinations

Compare with :

• English pronouns
• French pronouns
• Spanish pronouns
• Italian pronouns

5. German Determiners (D)

5.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example    Code
========== ============= ============ =====
Type        demonstrat.    dieser       d
            indefinite     kein         i
            possessive     mein         s
            interrog       welche       t
---------- ------------ ------------- ----- 
Person      first          mein         1
            second         dein         2
            third          sein         3
---------- ------------ ------------- ----- 
Gender      masculine      dieser       m
            feminine       diese        f
            neutre         dieses       n
---------- ------------ ------------- ----- 
Number      singular       dieser       s
            plural         diese        p
---------- ------------ ------------- ----- 
Case        nominative     dieser       n
            genitive       dieses       g
            dative         diesem       d
            accusative     diesen       a
---------- ------------ ------------- ----- 
Possessor        -            -         -
========== ============= ============ =====

Note

5.2. Corpus

5.3. Combinations

Note:

We deliberately treat the Articles as an extra class, but related to the Determiners. The syntactic behaviour of Articles and Determiners in NPs is different, as can be seen by the adjective in the following examples:

This difference supports our attitude towards separating articles and determiners generally.

Compare with :

• English articles/determiners
• French articles/determiners
• Spanish determiners
• Italian determiners

6. German Articles (T)

6.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example    Code
========== ============= ============ =====
Type        definite         der        d
            indefinite       ein        i
---------- ------------- ------------ -----
Gender      masculine        der        m
            feminine         die        f
            neuter           da         n
---------- ------------- ------------ -----
Number      singular         ein        s
            plural           die        p
---------- ------------- ------------ -----
Case        nominative       der        n
            genitive         dessen     g
            dative           dem        d
            accusative       den        a
========== ============= ============ =====

6.2. Corpus

6.3. Combinations

Compare with :

• English articles/determiners
• French determiners
• Spanish articles
• Italian articles

7. German Adverbs (R)

7.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example    Code
========== ============= ============ =====
Type         general       frischweg   g
             degree        sogar       d    this value has been added
             interrog      worum       i    this value has been added
             conjunction   mithin      c    this value has been added
             modal         scheinbar   m    this value has been added
             pronom        so          p    this value has been added
             temporal      heute       t    this value has been added
             place         hier        l    this value has been added
---------- ------------- ------------ -----
Degree       positive      hoch        p
             comparative   höher       c
             superlative   höchst      s
========== ============= ============ =====

7.2. Corpus

7.3. Combinations

Compare with :

• English adverbs
• French adverbs
• Spanish adverbs
• Italian adverbs

8. German Adpositions (S)

8.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example    Code
========== ============= ============ =====
Type           pre           an         p
               post          wegen      t   this value has been added
               circum        von - an   c   this value has been added
               part1         von        a   this value has been added
               part2         an         z   this value has been added
---------- ------------- ------------ -----
Formation      clitic        ans        c
               simple        an         s
========== ============= ============ =====

In German, most prepositions precede the NP. However, there is a good deal of them which follow the NP ("entlang") or enclose it ("von" NP än"). This behaviour is unpredictable and must therefore be marked lexically. This is done by increasing the values of the "Type" attribute. This extension can be considered as language specific, as long as there is no evidence from other languages which supports this distinction. For practical reasons of text segmentation, the two parts of a circumposition have to be distinguished by different tags.

The attribute "formation" allows us to deal with clitic contraction of preposition and article of the following NP ("zum", äns")

8.2. Corpus

8.3. Combinations

Compare with :

• English adpositions
• French adpositions
• Spanish adpositions
• Italian adpositions

9. German Conjunctions (C)

9.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example    Code
========== ============= ============ =====
Type       coordinat.       oder        c
           subordinat.      als         s
           compar           als         v   this value has been added
           infinitive       um          i   this value has been added
           part1            entweder    a   this value has been added
           part2            oder        z   this value has been added
========== ============= ============ =====

Comparative conjunctions are restricted to constitutents as arguments. Subordinate conjunctions introducing an infinite clause lead to an infinite verb form in sentence final position. The infinitive particle may be incorporated, as in "wegzugehen". These morphosyntactic features lead to an extension of the feature-Type values, which may be considered as language specific.

There are also a few complex conjunctions which appear at the beginning of both phrases which are conjoined by it ("entweder - oder"). This feature has to be marked lexically, again by increasing the values of the "Type" attribute. For practical reason of text segmentation, the two parts of a complex conjunction have to be distinguished by different tags.

9.2. Corpus

9.3. Combinations

Compare with :

• English conjunctions
• French conjunctions
• Spanish conjunctions
• Italian conjunctions

10. German Numerals (M)

We do not support the class "Numerals". Possible elements of this class will be treated as nouns, adjectives, or adverbs.

Compare with :

• English numerals
• French numerals
• Spanish numerals
• Italian numerals

11. German Interjection (I)

11.1. Lexicon

11.2.Corpus

11.3.Combinations

Compare with :

• English interjections
• French interjections
• Spanish interjections
• Italian interjections

12. German Add on Classes

12.1. Particle (Q)

12.1.1. Lexicon

========== ============= ============ =====
Attribute      Value       Example    Code
========== ============= ============ =====
Type        infinitive        zu        i
            superlative       am        s
            verbal pref.      hinzu     v
========== ============= ============ =====

12.1.2. Corpus

Superlative particles precede the superlative form of adjectives, if used predicatively (er ist am grö"sten), and adverbs (applies also to Dutch).

Infinitive particles are found in each of the Germanic languages to be described (Danish, Dutch, English, German). This should be accounted for somewhere in the generic set of classes.

Verbal prefixes are a particular German (and Dutch) phenomenon and can be considered to be language specific.

12.1.3. Combinations

12.2. German Punctuation (F)

12.2.1. Corpus

The use of Tags for punctuation sign is obvious for the stochastic modelling of utterances. However, it is not clear whether they should be treated lexically. Therefore, we can describe our corpus tags, but not a lexical treatment of these units.

Compare with :

• English Punctuation
• French Punctuation
• Spanish Punctuation
• Italian Punctuation

12.3. German Abbreviations (Y)

12.3.1. Lexicon

12.3.2. Corpus

Lexical abbreviations do not have a particular corresponding Corpus Tag, but are tagged according to the morphosyntactic behaviour of the form written out in full.

12.3.3. Combinations

12.4. German Others (X)

Other particular corpus tags can correspond to the common class of Residual.

12.4.1. Corpus

Compare with :

• English residuals
• French residuals
• Spanish residuals
• Italian residuals

12.5. German Uniq Membership Class (U)

None.

Compare with :

• English unique membership class
• French unique membership class
• Spanish unique membership class
• Italian unique membership class