From Dependency to CCG to Incremental CCG: Approaches to Flexible Word Order in Turkish

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing · Depth: Expert, quick

Summary

Researchers developed a Turkish Combinatory Categorial Grammar (CCG) lexicon and parser, building on Kuzgun et al. (2023). This lexicon was automatically induced from a dependency treebank, and the parser achieved a robust coverage of 92.5% by utilizing standard and extended operations adapted for Turkish syntax. They also introduced the first partially incremental, left-to-right CCG parser for Turkish, designed to integrate words immediately into the evolving linguistic representation. An example experiment demonstrated that CCG parsers can model psycholinguistic evidence for extra processing costs associated with arguments in noncanonical positions, specifically through the frequency of order-reversing operations. These findings suggest that CCG provides a cognitively plausible framework for modeling real-time language processing in flexible word order languages such as Turkish.

Key takeaway

For NLP Engineers developing parsers for morphologically rich languages with flexible word order, consider Combinatory Categorial Grammar. Your parsing efforts can benefit from automatically inducing CCG lexicons from existing dependency treebanks, potentially achieving high coverage like the 92.5% demonstrated for Turkish. Furthermore, exploring incremental, left-to-right CCG parsing can facilitate real-time word integration, offering a cognitively plausible approach for modeling language processing and associated psycholinguistic costs.

Key insights

Combinatory Categorial Grammar provides a cognitively plausible framework for real-time processing in flexible word order languages.

Principles

• CCG is well-suited for head-final, flexible word order languages.
• Lexicalized formalisms can model psycholinguistic processing costs.
• Automatic induction from dependency treebanks is viable.

Method

Automatically induce CCG categories from a dependency treebank, then develop a parser using standard and extended operations. Introduce an incremental, left-to-right parsing approach.

In practice

• Develop CCG parsers for flexible word order languages.
• Model psycholinguistic processing costs via operation frequency.
• Integrate words immediately into evolving representations.

Topics

• Combinatory Categorial Grammar
• Turkish Language Processing
• Flexible Word Order
• Incremental Parsing
• Psycholinguistics

Best for: AI Scientist, NLP Engineer, Research Scientist

Related on AIssential

• Grammar as an Injectable: A Trojan Horse to NLP | Towards Data Science — CCG offers a non-statistical, algebraic framework for NLP, treating words as functions to ensure grammatical consistency and enhance interpretability.
• Optimal Turkish Subword Strategies II: What WordPiece Learns from Turkish Morphology — Morphology-aware subword splits significantly improve Turkish NLP model performance across various tasks.
• AfriSUD: A Dependency Treebank Collection for Evaluating Models on African Languages — AfriSUD provides the first large-scale, native-speaker verified dependency treebanks for nine diverse African languages, revealing a significant syntax gap in current NLP models.
• Prefix Parsing is Just Parsing — Prefix parsing can be efficiently reduced to ordinary parsing via a grammar transformation.
• EPIC: Efficient and Parallel Inference under CFG Constraints for Diffusion Language Models — EPIC significantly accelerates CFG-constrained parallel decoding for diffusion LMs by optimizing validation and token commitment.
• Beyond Surface Forms: A Comprehensive, Mechanism-Oriented Taxonomy of Indirect Linguistic Encoding for LLM-Based Coded Language Detection — A mechanism-oriented taxonomy of indirect linguistic encoding significantly enhances LLM-based coded language detection.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.