From Labels to Facets: Building a Taxonomically Enriched Turkish Learner Corpus

TL;DR

The paper tackles the challenge of richly annotating learner Turkish by moving beyond flat error labels to a multi-facet taxonomy. It introduces a Turkish-specific annotation extender that combines manual annotation with automated morphosyntactic inference via UDPipe, producing a Taxonomically Enriched Turkish Learner Corpus with high facet-level accuracy ($\textbf{Macro}=95.86\%$) and strong overall annotation reliability. The study demonstrates the feasibility of collaborative, taxonomy-driven enrichment for a morphologically rich, low-resource language and provides open-access resources (corpus, guidelines, extender code) to enable facet-aware analyses and GEC research. This approach enhances cross-resource comparability and supports targeted educational and NLP applications by enabling complex, multi-dimensional error queries.

Abstract

In terms of annotation structure, most learner corpora rely on holistic flat label inventories which, even when extensive, do not explicitly separate multiple linguistic dimensions. This makes linguistically deep annotation difficult and complicates fine-grained analyses aimed at understanding why and how learners produce specific errors. To address these limitations, this paper presents a semi-automated annotation methodology for learner corpora, built upon a recently proposed faceted taxonomy, and implemented through a novel annotation extension framework. The taxonomy provides a theoretically grounded, multi-dimensional categorization that captures the linguistic properties underlying each error instance, thereby enabling standardized, fine-grained, and interpretable enrichment beyond flat annotations. The annotation extension tool, implemented based on the proposed extension framework for Turkish, automatically extends existing flat annotations by inferring additional linguistic and metadata information as facets within the taxonomy to provide richer learner-specific context. It was systematically evaluated and yielded promising performance results, achieving a facet-level accuracy of 95.86%. The resulting taxonomically enriched corpus offers enhanced querying capabilities and supports detailed exploratory analyses across learner corpora, enabling researchers to investigate error patterns through complex linguistic and pedagogical dimensions. This work introduces the first collaboratively annotated and taxonomically enriched Turkish Learner Corpus, a manual annotation guideline with a refined tagset, and an annotation extender. As the first corpus designed in accordance with the recently introduced taxonomy, we expect our study to pave the way for subsequent enrichment efforts of existing error-annotated learner corpora.

Figures (10)

• Figure 1: Timeline of the five main stages of the collaborative annotation process.
• Figure 2: Knowledge representation for the faceted taxonomy from eryigit2025.
• Figure 3: Overview of Annotation Extension Workflow
• Figure 4: System Architecture of the Annotation Extender
• Figure 5: Output comparison of the original learner corpus annotation using Label Studio (left-side) and the taxonomically enriched annotation produced by the annotation extender. Attributes that were automatically added by Label Studio as part of its internal operating mechanism were excluded from the JSON representation in the Manual Annotation part.