Overview
ISO 24615-1:2014 - Language resource management: Syntactic annotation framework (SynAF) - Part 1: Syntactic model defines a high-level metamodel and reference data categories for representing syntactic annotation of linguistic data. Its primary goal is to support interoperability between language resources (e.g., treebanks) and language processing components (parsers, taggers) by providing a common conceptual model for both constituency and dependency information.
Keywords: ISO 24615-1:2014, SynAF, syntactic annotation, syntactic model, language resource management, interoperability, constituency, dependency, data categories.
Key topics and technical requirements
- SynAF metamodel (UML-based): Describes classes and relationships used to model syntactic structure. Key classes include SyntacticNode, T_Node (terminal nodes / word forms and empty elements), NT_Node (non-terminal/phrasal nodes), SyntacticEdge (relations between nodes), and Annotation.
- Constituency and dependency coverage: The model explicitly supports hierarchical constituency (phrases, clauses, sentences) and dependency relations (head–modifier, subject–predicate), allowing both representations and their interrelation.
- Multi-layered annotation strategy: Encourages layered annotations so constituency and dependency information can coexist and reference morpho-syntactic annotations (integration with MAF).
- Spans and discontinuous constituents: Terminal nodes are defined over spans (including multiple spans) to account for discontinuous or non-contiguous elements.
- Data categories and DCS: Annex A lists normative data categories; implementers are expected to define a Data Category Selection (DCS) using the specified categories (available via ISOCat).
- Interoperability requirements: The standard is designed to be complementary to ISO 24611 (MAF) and compatible with other language resource frameworks; it mandates use of the provided metamodel and data category registry for consistent annotation exchange.
Practical applications
- Building and sharing treebanks that combine constituency and dependency annotations.
- Designing annotation schemes and tools (annotation editors, conversion utilities) that must interoperate with other language resources.
- Standardizing parser output formats to be compatible with corpora and downstream NLP components.
- Integrating morpho-syntactic (MAF) and syntactic annotations in multi-layer corpora for research and production NLP pipelines.
Who should use this standard
- NLP engineers and parser developers
- Corpus linguists and annotation teams
- Language technology vendors and research groups
- Tool and format designers working on syntactic annotation, treebanking, or resource conversion
Related standards
Using ISO 24615-1:2014 helps ensure consistent syntactic annotation, improves resource reusability, and simplifies integration across language-processing components.