UD for Turkish-English 
Tokenization and Word Segmentation
- Words are generally delimited by whitespace, with some language-mixing exceptions such as mixed words with Turkish morphology on English stems (e.g., “drop-bylayacağım”, “Soggyleşmiş”).
- English phrasal verbs may be written as a single token when integrated with Turkish morphology (e.g., “turn-offluyor”, “drop-bylayacağım”).
- English multi-word expressions like “on the same page” are tokenized as separate words when appearing in code-switched contexts.
- Abbreviations and acronyms (like “KK” for “Kuzey Kampüs” - North Campus) are treated as single tokens.
- Morphological integration points are annotated in the MISC column with
CSID=MIXEDandCSPoint=attributes (e.g., “turn-off§luyor”).
Morphology
Tags
- Participles are assigned the ADJ tag when used as modifiers (both Turkish and English).
- English verbs in light verb constructions with Turkish verbs like “etmek” (to do/make) and “olmak” (to be/become) are tagged as NOUN (e.g., “navigate etmek”, “tempt eyle”).
- English nouns (“distinction”, “clothing”) retain their NOUN tag when integrated into Turkish syntax.
- English particles in phrasal verbs are annotated as part of the verb when morphologically integrated (e.g., “drop-by” in “drop-bylayacağım”).
- The DET-PRON distinction follows the same principles as monolingually annotated treebanks, with English determiners like “my”, “a”, “the” tagged as DET and English pronouns like “I”, “it”, “you” tagged as PRON.
- Discourse markers from both languages (e.g., English “Guys”, Turkish “yani”) are tagged according to their original part of speech but connected with the discourse relation.
Features
- Nouns have inherent Number features, with English nouns maintaining singular/plural distinctions and receiving Turkish case marking (e.g., “machinelerden” with plural and ablative case).
- English verbs with Turkish morphology have the full range of Turkish verbal features including Aspect, Mood, Number, Person, and Tense.
- English verbs typically receive Turkish morphological markers, showing a Lang feature in the MISC column (e.g., “Lang=en” for English words, “Lang=tr” for Turkish words).
- The morphological integration follows Turkish agglutinative patterns, with English roots receiving Turkish suffixes according to Turkish morphotactic rules (e.g., “aspectlere” with dative case).
-
Mixed verbs may have complex feature combinations reflecting their hybrid nature, particularly in tense and aspect marking (e.g., “drop-bylayacağım” has Aspect=Perf Number=Sing Person=1 Polarity=Pos Tense=Fut). - Possession is marked on both Turkish and mixed nouns (e.g., “itemıyla” with instrumental case and singular possession).
- Code-switching points within words are annotated with CSID=MIXED and CSPoint features to indicate morpheme boundaries.
- Loan translation patterns are marked with LoanTrans in the MISC column (e.g., “toplantı alabiliriz” marked as LoanTrans=en:get_a_meeting).
Syntax
- Turkish-English UD is based primarily on universal dependency guidelines, while considering Turkish syntactic structures as the base for mixed constructions.
- The main predicate of a clause is the head of the clause, regardless of its language of origin.
- In mixed verb phrases with light verb constructions like “navigate etmek” (to navigate), “bill etmek” (to bill), the English element is treated as a compound and the Turkish light verb is the head.
- English phrasal verbs with Turkish morphology (like “drop-bylayacağım”) are treated as single verbal units with the entire construction serving as the predicate.
- Mixed sentences may maintain separate clausal structures in each language connected by parataxis (e.g., “Bir şey diyeceğim bence he doesn’t deserve this”).
- Discourse connectors from both languages (e.g., “bence”, “yani”) are attached with the discourse relation.
Core Arguments, Oblique Arguments and Adjuncts
- Nominal subject (nsubj) - Typically in nominative case with no overt marking; can be from either language (e.g., “he” in “he doesn’t deserve this”)
- Direct object (obj) - Usually marked with accusative case for definite objects (e.g., “Bunu yapmak”), but can be unmarked for indefinite ones; English objects in Turkish matrix clauses follow Turkish case marking
- Oblique modifiers (obl) - Other case-marked nominal phrases; Turkish case markers are applied to English nouns (e.g., “metroyla” with instrumental case)
- Adjuncts - Include adverbial modifiers (advmod) and other non-core elements from both languages (e.g., “perfectly” in “perfectly fine”)
Non-verbal Clauses
- English-origin copular constructions are analyzed with the Turkish copular pattern
- In predicative constructions, the predicate is the head and the copula is attached as a cop dependency
- Negative copular constructions use the Turkish “değil” (e.g., “acceptable değil”)
- Question particles like “mi” are attached as auxiliaries to the predicate
Relations Overview
- Basic dependencies follow the standard Universal Dependencies scheme
- Fixed expressions in either language (e.g., “of course”) use the fixed relation internally
Treebanks
There is 1 Turkish English UD treebank: