Sememe¶
A theory-relative unit on the meaning plane used to encode the semantic contribution associated with a morpheme or lexical sense, with some traditions treating it as the whole meaning and others as a minimal component—so the inventory and grain must always be declared.
Core Idea¶
A sememe is a theory-relative unit on the meaning plane of language. It lets an analyst associate a bounded semantic unit with a linguistic expression, lexical sense, or entry in a semantic knowledge base. Its indispensable qualification is theory-relative: different established traditions place the boundary at different grains.
In Leonard Bloomfield's terminology, a morpheme is a minimum linguistic form and its meaning is a sememe. His later textbook preserves the parallel: a morpheme is a lexical form and the meaning of a morpheme is a sememe. Here the sememe is the whole meaning assigned to that morpheme, not necessarily a smaller semantic feature. In a componential European tradition represented in modern French lexicography, by contrast, a sememe can be the set of pertinent semantic features—semes—that constitutes a lexeme's sense.
Scope of Application¶
Sememe analysis applies in linguistic semantics, morphology, lexicology, componential analysis, lexicography, terminology work, and computational lexical knowledge bases. In a Bloomfieldian analysis it records the meaning paired with a morpheme. Wonderly's analysis of Kechua person morphemes retained that pairing while asking whether their sememes contain recurring semantic components, showing that analysts can examine internal structure without erasing the host-level unit. In componential semantics, a sememe supplies the sense-level bundle against which individual semes are identified. In HowNet-oriented natural-language processing, sememes annotate word senses and supply smaller semantic ingredients for representation learning, sense selection, and related tasks.
Clarity¶
The fastest diagnostic is to complete four fields: host, inventory, grain, and task. For “the plural sememe,” ask which plural morpheme or sense is the host, which semantic inventory contains the unit, whether the sememe is the whole grammatical meaning or one component, and what contrast or analysis it supports. If those fields cannot be filled, the term is decorative rather than explanatory.
Manages Complexity¶
Sememes compress open-ended glosses into a controlled semantic inventory. Rather than giving every lexical sense an unrelated prose description, an analyst can reuse units across entries, compare recurring contributions, expose shared components, and support structured retrieval or computation. A word-sense annotation can then be represented as a small set or structured configuration of sememes rather than an unbounded definition.
Abstract Reasoning¶
The structure licenses several deductions:
- If two studies use different sememe grains, raw unit counts are not comparable. A whole-sense inventory will usually assign fewer units per sense than a component inventory.
- If the inventory changes, an unchanged word sense may receive a different sememe analysis.
Knowledge Transfer¶
Literal transfer occurs within the language sciences. A morphologist can move from one language to another while preserving the role package—host form, meaning unit, inventory, grain, and contrast test—even though the actual sememes differ. A lexicographer can move from paper dictionaries to a structured lexical database while retaining sense-level bundles. A computational linguist can evaluate a sememe inventory on new representation or sense-disambiguation tasks while preserving the requirement that each unit belongs to a declared resource and annotation policy.
Relationships to Other Abstractions¶
Current abstraction Sememe Domain-specific
Parents (1) — more general patterns this builds on
-
Sememe presupposes Grain of Analysis Prime
Sememe strictly presupposes Grain of Analysis.
Hierarchy path (1) — routes to 1 parentless root
- Sememe → Grain of Analysis
Neighborhood in Abstraction Space¶
Sememe sits in a sparse region of the domain-specific corpus (95th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (1565 abstractions)
Nearest neighbors
- Lexical Definition — 0.78
- Lexical Converse Relation — 0.77
- Knowledge organization system — 0.77
- Intertextuality — 0.76
- Grammatical Relation — 0.76
Computed from structural-signature embeddings · 2026-09-08