Referentboard

Referentboard is a theoretical construct in linguistics and natural language processing that describes a persistent representation of entities, or referents, encountered in discourse. The referentboard functions as a memory-like structure used to track which entities have been introduced, which are currently salient, and how they can be referred to by subsequent expressions.

This representation may take the form of a data structure (for example, a graph or a set of keyed records) or a mental model in cognitive theories. Each referent entry typically includes identity information (such as a canonical name or description), type or category, properties or attributes, discourse status (new, established, salient), and links to linguistic expressions that refer to it (pronouns, definite descriptions, proper names).

Operations on a referentboard include adding a newly introduced referent, updating its properties as new information becomes available, and linking it to coreferent expressions. It supports tasks such as anaphora resolution and coreference by providing a candidate pool for proforms based on compatibility constraints (gender, number, animacy, persistence).

Referentboard concepts appear in discussions of discourse representation, memory-based models of reference, and computational approaches to pronoun resolution. Related ideas include discourse representation theory, Centering theory, and reference tracking in natural language processing.

Applications include dialogue systems, annotation tools for linguistics, and studies of how humans establish and maintain referential continuity.

Notes: The term is not universally standardized and implementations vary; in practice, similar ideas are described under discourse representation, referential grounding, or coreference resolution frameworks.