UD for Turkish-German 
In general the Turkish-German code-switching treebank follows the guidelines of individual Turkish and German treebanks.
Language IDs and Intra-word Code-switching
-
All sentences in the treebank include at least one intrasentential code-switching point. The language information of tokens are represented in the MISC column with the
LangIDfeature. The values of this feature are:TRfor Turkish,DEfor German,MIXEDfor intra-word code-switching (see below),LANG3for non-Turkish, non-German languages (mostly English in the treebank), andOTHERfor punctuation. -
Some tokens employ intra-word code-switching, e.g. Terminim, where Termin ‘appointment’ is in German and -im ‘my’ is a Turkish possessive suffix. Code-switching points are marked on the surface forms as the values of the
CSPointfeature in the MISC column, e.g.,CSPoint=Termin§im.§indicates the switch point.
Tokenization and Word Segmentation
- In general, words are delimited by white spaces or punctuation. Punctuation may appear in some abbreviations or numeric expressions.
- German compounds are written as one word, and they are kept as is in the treebank.
- Contractions of German prepositions and definite articles are segmented, e.g., zum zu → dem ‘to the’.
- Turkish copular suffix and a group of productive derivational suffixes are segmented, e.g., -lI deriving nouns and adjectives from a noun (N) with the meaning of ‘with N’ (mavi elbise “blue dress” → mavi elbiseli ‘(the one) with a blue dress’.
Segmentation, code-switching points, and their interaction are described and exemplified in (Çetinoğlu and Çöltekin, 2016)
Morphology
Tags
- Turkish uses all 17 universal POS categories, German uses all except
SYM. - The Turkish and German negation particles, değil and nicht respectively are
PART. German infinitive marker zu ‘to’ is alsoPART. - The Turkish question particle mI is tagged as
AUX. - The Turkish copular suffix -(y)-, which is treated as a syntactic word, and its clitic counterpart i is marked as
AUX. - The German
AUXare dürfen, haben, können, mögen, müssen, sein, sollen, tun, werden, wollen.
Features
- The treebank uses the features of individual Turkish and German treebanks.
- Features introduced with this treebank are
LangIDandCSPointas described in Section Language IDs and Intra-word Code-switching
Syntax
-
The treebank uses the standard UD relations and the subtype relations of individual Turkish and German treebanks.
-
Subtype relations introduced with this treebank are appos:trans and parataxis:trans. Both are introduced to handle translation pairs , that is, a word, phrase or clause in one language and its repetition as a translation in the other language. Further description and examples can be found in (Çetinoğlu and Çöltekin, 2019)
Treebanks
There is one Turkish-German UD treebank:
References
-
Özlem Çetinoğlu and Çağrı Çöltekin (2016). Part-of-speech annotation of a Turkish-German code-switching corpus. In Proceedings of the 10th Linguistic Annotation Workshop (LAW-X). Berlin, Germany. https://www.aclweb.org/anthology/W16-1714.pdf
-
Özlem Çetinoğlu and Çağrı Çöltekin (2019). “Challenges of Annotating a Code-Switching Treebank”. In Proceedings of the 18th International Workshop on Treebanks and Linguistic Theories (TLT, SyntaxFest 2019). Paris, France.