Introducing TrGLUE and SentiTurca: A Comprehensive Benchmark for Turkish General Language Understanding and Sentiment Analysis
Duygu Altinok
TL;DR
This work fills a critical gap in Turkish NLP by introducing TrGLUE, a GLUE-style NLU benchmark for Turkish, and SentiTurca, a diverse sentiment-evaluation suite including a Turkish hate-speech component. It leverages native Turkish corpora and a semi-automated, triaged annotation pipeline that combines lightweight classifiers and state-of-the-art LLMs with substantial human validation to ensure linguistic naturalness and reproducibility. The authors provide extensive corpus statistics, reveal Turkish-specific morphosyntactic properties (e.g., pro-drop, rich morphology, variable word order), and demonstrate how these factors shape model performance across classification, NLI, and QA-style tasks. By releasing fine-tuning/evaluation scripts and distributing data openly, the work offers a robust, scalable framework for future Turkish-NLU research and cross-lingual benchmark design.
Abstract
Evaluating the performance of various model architectures, such as transformers, large language models (LLMs), and other NLP systems, requires comprehensive benchmarks that measure performance across multiple dimensions. Among these, the evaluation of natural language understanding (NLU) is particularly critical as it serves as a fundamental criterion for assessing model capabilities. Thus, it is essential to establish benchmarks that enable thorough evaluation and analysis of NLU abilities from diverse perspectives. While the GLUE benchmark has set a standard for evaluating English NLU, similar benchmarks have been developed for other languages, such as CLUE for Chinese, FLUE for French, and JGLUE for Japanese. However, no comparable benchmark currently exists for the Turkish language. To address this gap, we introduce TrGLUE, a comprehensive benchmark encompassing a variety of NLU tasks for Turkish. In addition, we present SentiTurca, a specialized benchmark for sentiment analysis. To support researchers, we also provide fine-tuning and evaluation code for transformer-based models, facilitating the effective use of these benchmarks. TrGLUE comprises Turkish-native corpora curated to mirror the domains and task formulations of GLUE-style evaluations, with labels obtained through a semi-automated pipeline that combines strong LLM-based annotation, cross-model agreement checks, and subsequent human validation. This design prioritizes linguistic naturalness, minimizes direct translation artifacts, and yields a scalable, reproducible workflow. With TrGLUE, our goal is to establish a robust evaluation framework for Turkish NLU, empower researchers with valuable resources, and provide insights into generating high-quality semi-automated datasets.