Jump to sentence processing, morpho-syntax, treebanking, semantics, computing stack.


cats of '18 summer & bit espresso bar, photo by me

sentence processing

My primary focus at the moment is on people’s degree of fallibility when it comes to comprehending number agreement. I am currently working on to reconcile conflicting findings on whether attraction effects are modulated by case ambiguity in sentence comprehension. My goal is to eventually tie this to the work focusing on exploring how people access memory and why we need to recontruct them using either abstract or form-related cues. Broadly, I’m interested in online addressable memory-building, its interaction with syntactic dependencies in typologically different languages, and the nature of the mechanisms and representations this interaction involves.

With Pavel Logacev, I focused on whether people’s failure to notice ungrammatical sentences was due to more cognitive-general issues such as, response bias, shallow processing, or task-effects. Through series of behavioral experiments, we learned that even though that they affect to what degree people fail to assess grammaticality of a sentence, these cognitive-general issues cannot be sole explanation of failure to detect ungrammaticality.

output Türk, U. (2022). Agreement Attraction in Turkish. [localpdf]. [repo]. [summary].

Türk, U., Logacev, P. (resubmitted). Agreement Attraction in Turkish: The case of genitive attractors. Doi: 10.31234/osf.io/5rmvu.

Türk, U., Logacev, P. (in prep). Response Bias in Turkish Agreement Attraction]. [repo] (_available upon a request_)

morpho-syntax

I was lucky to join Pavel Caha at Masaryk University to work on various topics including Turkish case syncretism, augmentatives, and suspended affixation.

One of the most exciting work I did was on Turkish adjectivals. I formalized the degree-sensitivity of Turkish Evaluative morphology, specifically “diminutives”. Adjectival diminutives, in Turkish, only attaches to negative-ordered adjectives. I also showed that Turkish -CIK is actually a complex formative unlike previously believed. While -K is an inherent part of the adjective, -CI is the amplifier that attaches to defective adjective “root” that is stripped from -K.

I also tried to formalize differential object marking in Turkish. Using nanosyntactic algorithm, I modeled the nominal case paradigm of Turkish.

Lastly, I argued that candidates in morphology can be re-ranked via phonological information. Words with root suppletion generally is not licensed in suspended affixation. However, preceding only a vowel-harmonic conjunction, suppletion was reverted and suspended affixation was licensed.

output Türk, U., Caha, P. (2021). Nanosyntactic Analysis of Turkish Case System. In S. Gundogdu, S. Taghipour, & A. Peters (eds.), Proceedings of the 6th Workshop on Turkic and languages in contact with Turkic, LSA, Doi: 10.3765/ptu.v6i1.5051.

Türk, U. (2020). Tackling the Augmentative Puzzle in Turkish. In B. Palaz & L. Stromdahl (eds.), Proceedings of the 5th Workshop on Turkic and languages in contact with Turkic, LSA, Doi: 10.3765/ptu.v5i1.4771.

treebanking

As a part of two projects (funded by TUBITAK and European Commission), I partake in creation and re-annotation of multiple Universal Dependencies Treebanks in multiple languages. My current aim is to create a set of treebanks for un(der)represented languages in Asia Minor.

With Arzucan Ozgur, Tunga Gungor, and Balkiz Ozturk, we created a new guidelines for annotating Turkish data within Universal Dependencies framework. Following these guidelines, we re-annotated two already existing treebanks, and created a new treebank.

As a part of my efforts to document minority languages, we also created the first Laz treebank, using data from published linguistic papers and theses. I am currently working on Ladino and Cappadocian Greek (with Konstantinos Sampanis) treebanks. I am extremely open to collaborate and willing to work on any minority language documentation/treebanking effort! Do not hesitate to reach out!

output Türk, U., Atmaca, F., Özateş, Ş.B. et al. (2022). Resources for Turkish dependency parsing: introducing the BOUN Treebank and the BoAT annotation tool. Language Resources & Evaluation 56, 259–307. Doi: 10.1007/s10579-021-09558-0. [resources]

Türk, U., Bayar, K., Özercan, A. D., Öztürk, G. Y., Özateş, S. B. (2019). First Steps towards Universal Dependencies for Laz. In M. de Marneffe, M. de Lhoneux, J. Nivre & S. Schuster (eds.), Proceedings of the Fourth Workshop on Universal Dependencies (UDW 2020), 189-194, Acl Anthology: 2020.udw-1.21.

Türk, U., Atmaca, F., Özateş, S. B., Bedir, S. T., Köksal, A., Öztürk B., Güngör, T., Özgür, A. (2019). Turkish Treebanking: Unifying and Constructing Efforts. In A. Friedrich, D. Zeyrek & J. Hoek (eds.), Proceedings of the 13th Linguistic Annotation Workshop, 166-177, Acl Anthology: W19-4019.

Türk, U., Atmaca, F., Özateş, S. B., Bedir, S. T., Köksal, A., Öztürk B., Güngör, T., Özgür, A. (2019). Improving the Annotations in the Turkish Universal Dependency Treebank. In A. Rademaker & F. Tyers (eds.), Proceedings of the Third Workshop on Universal Dependencies (UDW, SyntaxFest 2019), 108-117, Acl Anthology: W19-8013.

semantics

With Omer Demirok, we investigated how hypothetical comparions are formed in Turkish. -mışcasına is licensed with gradable adjectives but not with non-gradable ones. We took this to suggest that its semantics involves comparison of degrees. We pursued a previously uncharted route: HCM constructions in Turkish compare degrees, not eventualities. Our analysis is sensitive to the semantic difference between OPEN and CLOSE scale adjectives that are independently justified.

I am currently interested in how pragmatic reality is structured in natural language such as personal experiences, interpretation of indefinite nouns, and justified beliefs. To investigate these topics, I use data from de re/de dicto and acquitance inference literature.

output Türk, U., Demirok, Ö. (2021). Hypothetical Comparison in Turkish. In S. Gundogdu, S. Taghipour, & A. Peters (eds.), Proceedings of the 6th Workshop on Turkic and languages in contact with Turkic, LSA, Doi: 10.3765/ptu.v6i1.5054.

Computing stack