Language and mental simulation
How grammatical and lexical choices shape the mental simulation of motion, space, and static scenes across Turkish and English.
Linguist · Psycholinguist · AI evaluation researcher
Language, cognition,
and AI.
I study how people represent meaning—and how language technologies can be evaluated with the same linguistic, cultural, and methodological care.
10+years in academic and applied research
TR · ENcross-linguistic research programme
21TurkBench evaluation subtasks
PhDlinguistics, Hacettepe University
In short
Meaning is both a human phenomenon and an evaluation problem.
I am a PhD-trained linguist and AI evaluation researcher in the Department of Linguistics at Istanbul University. My work connects morphosyntax, semantics, cognitive linguistics, and psycholinguistics with multilingual NLP and responsible AI.
From 15 January 2025 to 14 January 2026, I was a TÜBİTAK-funded Visiting Postdoctoral Fellow at the University of Birmingham, working across Psychology and Linguistics & Communication. The project examines how fictive-motion expressions in Turkish and English shape real-time language processing and spatial mental simulation.
Alongside experimental research, I work on Turkish LLM benchmarking, human and model-based quality evaluation, prompt and rubric design, annotation quality, and responsible AI. My applied experience spans academic benchmarking and remote AI/data roles; see the complete experience record ↗.
Research
Three connected lines of inquiry
Human experiments, linguistic theory, and model evaluation inform one another throughout my work.
Multilingual AI evaluation
How culturally and linguistically grounded benchmarks expose capability gaps that aggregate model scores often hide.
Responsible human-centred AI
How people understand bias, trust, accuracy, and responsibility—and how those views should inform evaluation practice.
Featured programme
2025–2026
2025–2026
Mental simulation, space and language
TÜBİTAK-funded research combining eye tracking, self-paced reading, and picture elicitation.
Recent outputs
Selected publications
TurkBench: A Benchmark for Evaluating Turkish Large Language Models
Toraman et al. · SIGTURK 2026 · Association for Computational Linguistics
On trust, bias, and responsibility
Topraksoy · AI & Society 41(4)
Evaluating the Quality of Benchmark Datasets for Low-Resource Languages
Umutlu et al. · GEM² 2025 · Association for Computational Linguistics
Fictive Motion and Visualisation: Dynamic Spatial Perception in Turkish
Topraksoy · Dilbilim Dergisi — Journal of Linguistics
Teaching
From linguistic theory to evidence-led analysis
My teaching moves from clear concepts to guided analysis of authentic language data, helping students formulate questions, select evidence, and defend interpretations.
Language and Gender
Discourse, identity, representation, and power
Linguistic Typology
Cross-linguistic patterns and variation
Applications in Linguistics
Research design and applied analysis
Linguistic Texts in English
Critical academic reading in linguistics
Cognitive Linguistics
Language, cognition, categorisation, space, and meaning
PhD in Linguistics
Hacettepe University · GPA 3.57/4.00 · Motion Predicates in Turkish: A Morpho-Syntactic Treatment
MA in Linguistics
Hacettepe University · GPA 3.64/4.00 · personal name signs in Turkish Sign Language (TİD)
BA in English Linguistics
Hacettepe University · GPA 3.85/4.00 · Top Student of the Faculty Award
Visiting Researcher
Syracuse University · two doctoral research visits supervised by Prof. Jaklin Kornfilt
Research · collaboration · speaking