treatedwhether

Treatedwhether is a term used in linguistics and natural language processing to describe a methodology for annotating and interpreting occurrences of the conjunction or particle "whether" in complex sentences. The concept centers on how "whether" marks alternative propositions, embedded questions, or hedged statements, and how such occurrences should be treated in parsing, translation, and model training.

The term was introduced in a 2023 paper by researchers at the Institute for Language Technology who proposed a standardized annotation protocol that classifies sentences into direct questions, embedded questions (whether-clauses), and conditional clauses where "whether" functions as a discourse marker. The protocol aims to improve cross-corpus comparability and model performance on tasks requiring semantic disambiguation.

Methodology: collect sentences containing "whether" from the target corpus; label each instance by syntactic role (direct vs embedded vs conditional) and by semantic force (question vs hedge); annotate scope and complementizer behavior; use the annotations to train parsers, QA systems, or sentiment models. Variants include strong and weak treatedwhether distinctions, depending on how rigidly labels are assigned.

Example sentences: "Do you know whether he will come?" (embedded question); "Whether it rains is uncertain." (whether as a proposition marker); "I asked whether, if necessary, we should delay." (complex embedding). In some contexts "whether" interacts with negation and polarity.

Criticism and status: Some linguists argue that the boundary between direct, embedded, and conditioned uses of "whether" is fluid, complicating annotation. Advocates emphasize transparent guidelines and cross-linguistic extension.