=Paper=
{{Paper
|id=Vol-3315/paper08
|storemode=property
|title=Enhancements to the BOUN Treebank Reflecting the Agglutinative Nature of Turkish
|pdfUrl=https://ceur-ws.org/Vol-3315/paper08.pdf
|volume=Vol-3315
|authors=Büşra Marşan,Salih Furkan Akkurt,Muhammet Şen,Merve Gürbüz,Onur Güngör,Şaziye Betül Özateş,Suzan Üsküdarlı,Arzucan Özgür,Tunga Güngör,Balkız Öztürk
}}
==Enhancements to the BOUN Treebank Reflecting the Agglutinative Nature of Turkish==
https://ceur-ws.org/Vol-3315/paper08.pdf
Enhancements to the BOUN Treebank Reflecting the
Agglutinative Nature of Turkish
Büşra Marşan1 , Salih Furkan Akkurt2 , Muhammet Şen2 , Merve Gürbüz2 ,
Onur Güngör2 , Şaziye Betül Özateş2 , Suzan Üsküdarlı2 , Arzucan Özgür2 ,
Tunga Güngör2 and Balkız Öztürk1
1
Boğaziçi University, Linguistics
2
Boğaziçi University, Computer Engineering
Abstract
In this study, we aim to offer linguistically motivated solutions to resolve the issues of the lack of
representation of null morphemes, highly productive derivational processes, and syncretic morphemes of
Turkish in the BOUN Treebank without diverging from the Universal Dependencies framework. In order
to tackle these issues, new annotation conventions were introduced by splitting certain lemmas and
employing the MISC (miscellaneous) tab in the UD framework to denote derivation. Representational
capabilities of the re-annotated treebank were tested on a LSTM-based dependency parser and an updated
version of the BoAT Tool is introduced.
Keywords
Universal Dependencies, Turkish, morphological analysis, dependency annotation, dependency parsing
1. Introduction
Following the dependency grammar framework first proposed by Tesniére [1], dependency
trees illustrate how sentence elements relate to one another through head and dependent
relations. Universal Dependencies1 (UD) is an international cooperative treebank project based
on the dependency grammar framework and it aims to offer a standardized and comprehensive
dependency treebank collection covering 121 languages.
With the addition of new UD treebanks, Turkish does not qualify as a low resource language
anymore. With a total of 733,000 tokens, it is the 12th largest UD treebank in the UD repository.
Although the coverage of the treebanks plays an essential role in improving the performance
of natural language processing (NLP) systems [2], their ability to correctly and consistently
illustrate the morphosyntactic features of the target language should not be overlooked. As
Vincze et al. [3]’s study shows, the better a treebank’s ability to represent the morphology
The International Conference and Workshop on Agglutinative Language Technologies as a challenge of Natural
Language Processing (ALTNLP), June 7-8, Koper, Slovenia
$ busra.marsan@boun.edu.tr (B. Marşan); furkan.akkurt@boun.edu.tr (S. F. Akkurt); muhammet.sen@boun.edu.tr
(M. Şen); merve.gurbuz@boun.edu.tr (M. Gürbüz); onurgu@boun.edu.tr (O. Güngör); saziye.bilgin@boun.edu.tr
( B. Özateş); suzan.uskudarli@boun.edu.tr (S. Üsküdarlı); arzucan.ozgur@boun.edu.tr (A. Özgür);
gungort@boun.edu.tr (T. Güngör); balkiz.ozturk@boun.edu.tr (B. Öztürk)
© 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings CEUR Workshop Proceedings (CEUR-WS.org)
http://ceur-ws.org
ISSN 1613-0073
1
https://universaldependencies.org
and syntax of the target language, the better the performance of the NLP systems using that
treebank as a resource.
In this paper, we aim to abide by the linguistic framework set by Bedir et al. [4] and offer
an updated and comprehensive UD treebank for Turkish, the BOUN Treebank, along with an
improved UD annotation interface, the BoAT Tool, first introduced in Türk et al. [5].
The decisions made in the re-annotation process of the BOUN Treebank aim to offer solutions
to the issues posed by the morphologically rich and complex nature of Turkish: null morphemes
are frequently employed, agglutinative processes are heavily used to create new forms, and
numerous morphemes like copula and -ki are very syncretic. The main goal of this study is to
illustrate these phenomena without compromising the compliance with the UD framework.
This paper is organized as follows. Previous attempts at creating dependency treebanks in
Turkish are laid out in Section 2. Annotation changes made in the current version of the BOUN
Treebank and their linguistic justification are discussed in Section 3. Statistics regarding the
changes made to the previous version are stated in Section 4. Improvements made to the BoAT
Annotation Tool are explained in Section 5. Finally, the performance of the parser trained using
the current version of the treebank is reviewed in Section 6.
2. Dependency Treebanks in Turkish
Shortly after the first dependency treebank for Turkish was presented by Atalay et al. [6],
Eryiğit and Pamay [7] offered a smaller dependency treebank consisting of 300 sentences as
part of the CoNLL 2007 Shared Task: MST Treebank. 13 years after its publication, Sulubacak et
al. [8] re-annotated this dataset, converted it to the UD framework, and published the updated
dataset as IMST-UD. Çöltekin’s The Grammar Book Treebank (GB) [9] which consists of 2,803
sentences extracted by a reference book on Turkish grammar by Göksel and Kerslake [10]
marks the very first effort in creating the first UD-style Turkish treebank. Another pioneer in
Turkish dependency treebanks is IWT-UD as it is the first Turkish dependency treebank that
covers informal texts. IWT-UD was introduced by Sulubacak and Eryiğit [11] three years after
a constituency-style treebank, IWT, was presented by Pamay et al. [12].
Tourism is one of the two domain-specific dependency treebanks in Turkish and consists of
hotel and restaurant reviews. The other one is ATIS treebank that covers the Turkish translation
of English ATIS (Airline Travel Information System) corpus [13].
Other Turkish UD-style treebanks include Kenet UD Treebank, Penn Treebank, and FrameNet
treebank. Penn Treebank consists of Turkish translations of English Penn Treebank [14] while
FrameNet consists of 2,700 sentences from the Turkish FrameNet database [15].
With 9,761 sentences and 121,214 tokens randomly selected from Turkish National Corpus
(TNC) [16], the BOUN Treebank is one of the largest UD-style treebanks in Turkish. Covering
five different registers (broadsheet national newspapers, biographical texts, essays, popular
culture articles, and instructional texts), it offers word order and sentence length variance in
addition to linguistically motivated dependency annotations (see Section 3).
3. Improving BOUN Treebank
The first step of the previous annotation process of the BOUN Treebank (see [5] for a detailed
discussion) was parsing the raw text to create CoNLL-U files using Kanerva et al.’s [17] pipeline
tool. During this parsing process, UPOS tags and certain morphological information were
automatically annotated. Then the dependency relations were manually annotated by two
native speakers of Turkish who are linguists. To ensure inter-annotator agreement, randomly
selected 1000 sentences were double annotated. Using Cohen’s Kappa measure, inter-annotator
agreement was calculated. Dependency label match score was 0.82, unlabelled attachment score
was 0.83, and labelled attachment score was 0.75.
3.1. The Re-annotation Process: Overcoming the Challenges
For the re-annotation process, a team of linguists detected shortcomings and problematic
annotations of the previous version of the BOUN Treebank [5]. Two major representation
challenges were detected: derivation and null copula. The following subsections discuss the
strategies employed to overcome these challenges.
Annotations were done by two linguists who are native speakers of Turkish. In order to
ensure inter-annotator agreement, 100 sentences were double annotated. Unlabelled attachment
and labelled attachment scores were calculated using Cohen’s Kappa measure. Their respective
values are 98.61 and 97.81.
3.1.1. Derivation
Having its focus on syntax, the UD framework falls short of representing derivational processes.
The official guide of UD argues that the final derivational suffix in a word is opaque in the sense
that it does not permit access to the morphemes that come before it. Hence in a construction as
the one shown in Figure 1, numerous derivational and inflectional morphemes before the last
derivational morpheme (-ki) are lost.
Figure 1: Derivational and inflectional analysis of “görüşmelerindeki" (“those who were/are at his/her
meetings”)
As one of the primary concerns of this study was finding ways to illustrate derivation
processes like those in Figure 1 without diverging from the UD framework, two strategies were
employed for different cases:
• For morphemes like -lI (“with”), df= function is introduced in the MISC tab.
• Lemmas containing -ki morpheme are splitted.
These derivational morphemes imply that the host lemma and its modifier form a syntactic
unit together. As a result, the correct bracketing of the -lI adjective, its modifier, and the
constituent they modify together should look like (2) instead of (1). Representing the derivational
morphemes like -lI and -sIz allows keeping this crucial syntactic information.2
(1) [ [ [kahverengi] tüy-lü] kedi] (2) [ [kahverengi tüy]-lü kedi]
brown fur-attr cat brown fur-attr cat
“a cat with brown fur” “a cat with brown fur”
With the new df= function proposed in this study, “tüylü” is not decomposed as two lemmas:
“tüy” (“fur”) and “lü” (“with”). Instead, it is left intact and df=tüy (“derived from=fur") function
is annotated in the MISC tab.
Another challenge regarding the derivational processes is posed by -ki. There are two different
-ki morphemes in Turkish [18]. One is used as noun and the other derives adjectives from nouns.
Statistically, the vast majority of inflectional suffixes in Turkish occur after the derivational
ones. However, both types of the -ki morphemes diverge from this distribution as it can be
observed in Figure 1, where the inflectional suffix -ler precedes the derivational -ki. Hence
allowing -ki to “block access” [4]3 to the morphemes that were attached before it reduces the
capabilities of the annotation to represent a great deal of morphosyntactic information. With
the aim of offering a solution to this issue, a decision to split lemmas that contain either type of
-ki was made.4 Although splitting lemmas is not a common practice within the UD framework,
we believe that the theoretical motivation behind this decision justifies the divergence from the
framework. (For a detailed linguistic discussion of this issue, please refer to Bedir et al. [4])
3.1.2. Null Morpheme
Languages such as Turkish, Russian, Arabic, and Coptic have null morphemes, however, the
UD framework does not officially support such phenomena. As a result, independent strategies
have emerged to represent null morphemes in these languages. For example, Coptic avoids
null subject nodes by using fused forms[19], Marathi annotates the feature values of the null
morphemes, however, does not introduce any information (i.e. annotation) to indicate that
they are null morphemes[20]. Widely employing null morpheme for pluralization, Arabic
distinguishes between two types of annotation: Form-based and function-based. In their UD-
style dependency treebank annotation, Marton et al.[21] follow a function-based annotation
framework and annotate the feature values of null morphemes.
In Turkish, the copula can surface in three different forms [10]: i-, -y-, and ∅.After considering
UD guidelines and particularities of Turkish copula, it was decided to employ the MISC tab
2
Abbreviations: attr = attributive.
3
It is stated in the UD guidelines that “the lemma does not remove derivational morphology, so the lemma
of [en] “organizations" is “organization" not “organize" (nor “organ”).” See https://universaldependencies.org/u/
overview/morphology.html for the full picture.
4
Refer to the Appendix to compare previous and updated annotation schemes for both -ki morphemes.
again by introducing two new functions for the null copula: nullcop=3s (singular) and
nullcop=3p (plural). By following a function-based annotation schema, we were able to offer
more linguistically accurate annotations without diverging from the UD framework.
3.1.3. Copula
In Turkish, ol- copula has six distinct functions [4]: An intransitive verb meaning “to be
suitable/fit", a transitive verb meaning “to become", an auxiliary verb in embedded sentences,
an auxiliary verb following the participle, a light verb forming complex verbal constructions
(such as “sorun olmak" (“to be/become an issue")), and finally the existential predicate that
surfaces as “var" (“to exist") and “yok" (“not to exist"). Yet the previous annotation scheme of the
BOUN Treebank made no distinction between these different usages. To offer more accurate
representations, certain annotation changes were made regarding these functions.
3.2. Newly Introduced XPOS Tags and Dependency Relations
In an attempt to overcome the challenges thoroughly discussed in the previous subsection, a
set of new XPOS tags and dependency relations were introduced in the updated version of the
BOUN Treebank. A comprehensive list can be found in Table 1.
ol- (in ol- (as transitive
ol- (after ol- (in light verb -ki -ki
Lemma var yok embedded or intransitive
participle) constructions) (adjectivizer) (pronominal)
sentences) verb)
UPOS NOUN NOUN AUX AUX VERB VERB PART PRON
XPOS Exist Exist Ptcp Attr Partic
Deprel root root aux cop compound:lvc root dep:der
Table 1
New dependency relations and XPOS tags proposed for the updated BOUN Treebank.
4. Statistics
As part of the re-annotation process, 117,732 changes were made in the following tabs: UPOS,
XPOS, Deprel, MISC, and Features. The majority of annotation changes targeted UPOS and
XPOS tags (see Table 2). Since morphological information was automatically annotated in the
previous version of the treebank by the parsing tool [17], refinements by the annotators were
required in order to ensure accuracy.
Field UPOS XPOS Features Deprel MISC
Changes 11,396 63,829 27,098 23,32 4,973
Table 2
Changes made in the re-annotation process.
The changes in UPOS values reflect the linguistics-based decisions made in the re-annotation
process. Previous version of BOUN Treebank made no distinction between two -ki morphemes
in Turkish. As a result, almost all -ki instances were labeled as CConj (clausal conjunction).
After deciding to make a distinction between adjectivizer -ki and pronominal -ki, the UPOS tag
of the former was changed to Part.
Field UPOS XPOS
Change Adj ->Noun CConj ->Part Noun ->Propn Verb ->Ptcp Verb ->Vnoun ANum ->Indef
Count 1,595 1,025 968 2,459 1,664 1,622
Table 3
Most frequent changes targeting the UPOS and XPOS tags
Due to the shortcomings of automatic morphological tagging, some proper nouns were
labeled as nouns. Re-annotation process targeted them as well: UPOS tags of 986 proper nouns
were changed. In addition, XPOS tags of verbal nouns and participle forms were updated (see
Table 3).
5. The BoAT Tool
The BoAT tool offered in Türk et al. [5] is a desktop application for manually annotating
sentences parsed by a dependency parser. In the scope of the current work, in addition to the
reannotation of the BOUN Treebank, we enhanced the tool with additional functionalities. We
will publish the tool as open source on GitLab accompanied with a user manual.
5.1. Changes
BoAT is a desktop application, written in Python and based on Qt. The Qt version has been
incremented from 5 to 6. This resulted in a more modern-looking user interface (UI). With
ample feedback from the annotators who had used the tool for the BOUN Treebank, some
requested features and improvements have been surfaced.
Clutter: Annotation table’s columns were being shown or hidden by checkboxes above the
table. These were removed and a textbox beside the other buttons has been added to replace
them. This textbox serves exactly the same purpose while taking less space.
Autocompletion: Annotators use the tool for long hours at a time. Thus after a while, mistakes
tend to occur. Another requested feature, autocompletion of the table fields, aims to prevent
such mistakes. Many fields of the table have predetermined sets of values they can take. By
not allowing values outside these sets and having a shorthand writing system whereby an
annotator enters only the start of a value and it gets filled automatically, we implemented this
much requested feature.
Saving as CoNLL-U : Another change was regarding saving of the CoNLL-U documents. The
initial tool saved every edit automatically, yet treebanks with thousands of sentences tend to
take time to save. This mishap seemed to slow our annotators. We added a save button instead
of the autosave feature. Currently, a user uses the save button to edit the actual CoNLL-U file.
Shortcuts: The initial tool already had shortcut support. Going with the focus-oriented
approach, all the new tasks have keyboard shortcuts associated with them as well. This alleviates
the need to use a mouse while annotating via keyboard.
Dependency graphs: The vertical dependency graph in the initial version was replaced by the
horizontal dependency graph of spaCy [22], displaCy, due to the spatial concerns.
Validation: The validation script for annotations has been upgraded to the latest version
written by the UD framework, which has much more detailed explanations for why a specific
annotation is invalid.
6. Parser Performance
Within the scope of this study, an NLP task was conducted. BiLSTM-based biaffine dependency
parser proposed by Dozat and Manning [23] was trained using the updated BOUN Treebank.
The train set contained 7.803 sentences, development set contained 982 sentences and test set
contained 979 sentences. Considering the average arc length and average token count (see
Table 4 for details) of each set, BOUN Treebank offers a well-balanced data set.
Train Development Test Entire Data
Average
2.91 2.88 2.82 2.90
Arc Length
Average
12.83 12.42 12.36 12.74
Token Count
Number
7,803 982 979 9,761
of Sentences
Table 4
Sizes and specifications of train, development and test data sets
The previous version of the BOUN Treebank [5] yielded 77.36 unlabeled attachment score
(UAS) and 70.37 labeled attachment score (LAS). After being trained on the new dataset, UAS is
increased by 0.59 points to reach 77.96 while LAS showed a 0.10 decrease by dropping to 70.26
points. After the re-annotation process, several faulty or disputed dependency relations were
fixed in the data, hence a rise in UAS was observed.
In order to better account for the particularities of Turkish morphosyntax, four new depen-
dency labels were introduced: dep:der for adjectivizer -ki, obl:tmod for obliques that offer
temporal information regarding the predicate, advmod:emph for dA clitics, and compound:lvc
for light verb constructions with ol- copula. Moreover, new use cases for cop dependency label
were offered to represent different functions of the ol- copula. These changes in the annotation
framework added four new dependency types, thus the total class number was increased.
A sum of 1,032 dep:der, 894 obl:tmod, 1,860 advmod:emph, and 1,545 compound:lvc
tags were added. Newly introduced tags and classes introduced added complexity, hence they
might be the reason behind the slight decrease in the LAS results.
The main annotation changes made as part of this study were focused on morphology yet
BiLSTM-based biaffine dependency parser proposed by Dozat and Manning [23] doesn’t refer
to the morphological information. In fact, it almost completely ignores the morphology features
while parsing. Hence the significance of the improvements offered by this study can be better
gauged by a morphology-aware parser or a morphological analysis based downstream task.
7. Conclusion and Further Research
The UD framework highly emphasizes syntactic relations and aims to offer a universal founda-
tion to represent typologically different languages in a uniform way. While doing so, certain
particularities of these languages tend to get lost in the annotation process in an attempt to
abide by the UD convention. The aim of this study is to offer linguistically sound solutions
to illustrate syntactically relevant morphological features of Turkish such as null morpheme
realizations and derivational processes without diverging significantly from the UD framework.
In order to test the morphological capabilities of the re-annotated BOUN Treebank, a parser
that refers to morphological information can be implemented in further research.
Acknowledgments
This work was supported by Boğaziçi University Research Fund Grant Number 16909. TUBAGE-
BIP Award of the Turkish Science Academy (to A.O.) is gratefully acknowledged.
References
[1] L. Tesnière, Eléments de syntaxe structurale, paris: Klinck sieck, German translation:(1980)
Grundzüge der strukturalen Syntax Stuttgart: Klett-Cotta (1959).
[2] K. Foth, A. Köhn, N. Beuck, W. Menzel, Because size does matter: The hamburg dependency
treebank (2014).
[3] V. Vincze, K. I. Simkó, Z. Szántó, R. Farkas, Universal dependencies and morphology for
hungarian-and on the price of universality, Association for Computational Linguistics,
2017.
[4] T. Bedir, K. Şahin, O. Güngör, S. Uskudarli, A. Özgür, T. Güngör, B. Ö. Başaran, Over-
coming the challenges in morphological annotation of turkish in universal dependencies
framework, in: Proceedings of The Joint 15th Linguistic Annotation Workshop (LAW) and
3rd Designing Meaning Representations (DMR) Workshop, 2021, pp. 112–122.
[5] U. Türk, F. Atmaca, Ş. B. Özateş, G. Berk, S. T. Bedir, A. Köksal, B. Ö. Başaran, T. Güngör,
A. Özgür, Resources for turkish dependency parsing: Introducing the boun treebank and
the boat annotation tool, Language Resources and Evaluation (2021) 1–49.
[6] N. B. Atalay, K. Oflazer, B. Say, The annotation process in the turkish treebank, in:
Proceedings of 4th International Workshop on Linguistically Interpreted Corpora (LINC-
03) at EACL 2003, 2003.
[7] G. Eryiğit, T. PAMAY, Itu validation set for metu-sabancı turkish treebank, Türkiye Bilişim
Vakfı Bilgisayar Bilimleri ve Mühendisliği Dergisi 7 (2007) 31–37.
[8] U. Sulubacak, M. Gökırmak, F. Tyers, Ç. Çöltekin, J. Nivre, G. Eryiğit, et al., Universal
dependencies for turkish, in: Proceedings of COLING 2016, the 26th International Confer-
ence on Computational Linguistics: Technical Papers, The Association for Computational
Linguistics, 2016.
[9] C. Cöltekin, A grammar-book treebank of turkish, in: Proceedings of the 14th workshop
on Treebanks and Linguistic Theories (TLT 14), 2015, pp. 35–49.
[10] A. Göksel, C. Kerslake, Turkish: A comprehensive grammar, Routledge, 2004.
[11] U. Sulubacak, G. Eryiğit, Implementing universal dependency, morphology, and multiword
expression annotation standards for turkish language processing, Turkish Journal of
Electrical Engineering & Computer Sciences 26 (2018) 1662–1672.
[12] T. Pamay, U. Sulubacak, D. Torunoğlu-Selamet, G. Eryiğit, The annotation process of the
itu web treebank, in: Proceedings of the 9th Linguistic Annotation Workshop, 2015, pp.
95–101.
[13] C. T. Hemphill, J. J. Godfrey, G. R. Doddington, The atis spoken language systems pilot
corpus, in: Speech and Natural Language: Proceedings of a Workshop Held at Hidden
Valley, Pennsylvania, June 24-27, 1990, 1990.
[14] A. Taylor, M. Marcus, B. Santorini, The penn treebank: an overview, Treebanks (2003)
5–22.
[15] B. Marşan, N. Kara, M. Özçelik, B. N. Arıcan, N. Cesur, A. Kuzgun, E. Sanıyar, O. Kuyrukçu,
O. T. Yıldız, Building the turkish framenet, in: Proceedings of the 11th Global Wordnet
Conference, 2021, pp. 118–125.
[16] Y. Aksan, M. Aksan, A. Koltuksuz, T. Sezer, Ü. Mersinli, U. U. Demirhan, H. Yılmazer,
G. Atasoy, S. Öz, İ. Yıldız, et al., Construction of the turkish national corpus (tnc), in:
Proceedings of the Eighth International Conference on Language Resources and Evaluation
(LREC’12), 2012, pp. 3223–3227.
[17] J. Kanerva, F. Ginter, N. Miekka, A. Leino, T. Salakoski, Turku neural parser pipeline: An
end-to-end system for the conll 2018 shared task, in: Proceedings of the CoNLL 2018
Shared Task: Multilingual parsing from raw text to universal dependencies, 2018, pp.
133–142.
[18] J. Hankamer, Why there are two–ki’s in turkish, Current research in Turkish linguistics,
Eastern Mediterranean University (2004).
[19] A. Zeldes, M. Abrams, The coptic universal dependency treebank, in: Proceedings of the
Second Workshop on Universal Dependencies (UDW 2018), 2018, pp. 192–201.
[20] V. Ravishankar, A universal dependencies treebank for marathi, in: Proceedings of the
16th International Workshop on Treebanks and Linguistic Theories, 2017, pp. 190–200.
[21] Y. Marton, N. Habash, O. Rambow, Improving arabic dependency parsing with form-based
and functional morphological features, in: Proceedings of the 49th annual meeting of
the association for computational linguistics: Human language technologies, 2011, pp.
1586–1596.
[22] Explosion, spaCy, 2022. URL: https://spacy.io/, [Online; last accessed 27 March 2022].
[23] T. Dozat, P. Qi, C. D. Manning, Stanford’s graph-based neural dependency parser at the
conll 2017 shared task, in: Proceedings of the CoNLL 2017 shared task: Multilingual
parsing from raw text to universal dependencies, 2017, pp. 20–30.
A. A comparison of previous and updated annotation schemes
Token Form Lemma UPOS XPOS Features Head Deprel MISC
1 başındaki başındaki ADJ Adj 2 amod
3 şapkası şapka NOUN Case=Nom|Number=Sing|Number[psor]= 0 root
Sing|Person=3|Person[psor]=3
Table 5
Previous annotation scheme for “başındaki şapkası" (“the hat on his/her head")
Token Form Lemma UPOS XPOS Features Head Deprel MISC
1-2 başındaki başındaki
1 başında baş NOUN Case=Loc|Number=Sing|Number[psor]= 3 nmod
Sing|Person=3|Person[psor]=3
2 ki ki PART Attr 1 dep:der
3 şapkası şapka NOUN Case=Nom|Number=Sing|Number[psor]= 0 root
Sing|Person=3|Person[psor]=3
Table 6
Updated annotation scheme for “başındaki şapkası" (“the hat on his/her head")
Token Form Lemma UPOS XPOS Features Head Deprel MISC
1 seninki seninki NOUN Case=Nom|Number=Sing 0 root
Table 7
Previous annotation scheme for “seninki" (“that of yours")
Token Form Lemma UPOS XPOS Features Head Deprel MISC
1-2 seninki seninki
1 senin sen PRON PERS Case=Gen|Number=Sing|Person=2|PronType=Prs 2 nmod:poss
2 ki ki PRON Partic Case=Nom|Number=Sing 0 root
Table 8
Updated annotation scheme for “seninki" (“that of yours")