Skip to content

Multiword Expression

A conventional sequence of two or more lexemes whose meaning, syntax, selection, or use is partly idiosyncratic relative to ordinary composition.

Version
v1 · 2026-09-28 · History
Domain-specific #
10858
Domain group
Humanities
Origin domain
Linguistics & Semiotics
Subdomains
Lexicology, Phraseology → Linguistics & Semiotics
Aliases
MWE, Multi-word expression, Multiword unit, Lexicalized phrase

Core Idea

Multiword expressions occupy the boundary between lexicon and grammar. Several written or spoken words recur as one conventional unit even though they retain separate lexical forms.

The relevant irregularity is not always an opaque idiomatic meaning. Restricted word choice, unusual syntax, pragmatic function, or statistically entrenched association can also establish unithood, so analyses must state their diagnostic and language context.

Structural Signature

Sig role-phrases:

  • Lexeme sequence — Supplies two or more overt lexical components. It is form. Counterfactual: A single lexeme is not a multiword expression.
  • Conventional association — Makes the sequence recurrent within a speech community or register. It is lexicalization. Counterfactual: A one-off combination lacks unit status.
  • Idiosyncratic property — Marks what ordinary compositional rules fail to predict. It is diagnostic. Counterfactual: Fully free and predictable combination falls outside the class.
  • Internal variability — Specifies whether inflection, insertion, or word order can change. It is constraint. Counterfactual: Assuming all expressions are frozen misses productive patterns.
  • Context and register — Determine interpretation, frequency, and conventionality. It is licensing frame. Counterfactual: The same sequence can be literal in one context and idiomatic in another.
  • Recognition criterion — Operationalizes unithood for lexicography or language processing. It is evidence rule. Counterfactual: Frequency alone confuses common free phrases with lexical units.

What It Is Not

  • It is not every sequence of adjacent words.
  • It need not be completely fixed or semantically opaque.
  • It is not established by frequency alone.
  • Its tokenization and boundaries can vary across languages and analyses.
  • Closest near-miss. A frequent collocation can qualify through restricted lexical choice even when its overall meaning is transparent; a merely frequent free phrase does not.

Scope of Application

  • Lexicography. Records idioms, phrasemes, and collocations.
  • Syntax and semantics. Studies partial compositionality and internal constraints.
  • Corpus linguistics. Measures association and conventional recurrence.
  • Natural-language processing. Recognizes units that affect parsing and translation.

Clarity

State the language, lexeme boundaries, expression type, proposed idiosyncrasy, variability, corpus or speaker evidence, and whether unithood is lexical, constructional, pragmatic, or task-specific.

Manages Complexity

The category preserves both internal word structure and unit-level behavior, avoiding a false choice between ordinary phrases and indivisible words.

Abstract Reasoning

  1. Segment the candidate into lexemes.
  2. Establish recurrence and conventionality.
  3. Test semantic and syntactic predictability.
  4. Map permitted variation and contextual readings.
  5. Apply a stated unithood criterion for the intended analysis.

Knowledge Transfer

MWE diagnostics transfer across languages only after remapping morphology, spacing, wordhood, idiomaticity, and corpus conventions.

Examples

Canonical

'Kick the bucket' can conventionally mean die, a reading not compositionally recovered from kick and bucket, and its substitutions are restricted.

Mapped back: sequence → three lexemes; convention → idiomatic; idiosyncrasy → meaning; variability → restricted.

Applied / In Practice

'Read the new report' is ordinarily assembled from productive syntax and the usual meanings of its words, so frequency in a corpus alone does not make it an MWE.

Mapped back: sequence → multiword; convention → not shown; prediction → compositional.

Structural Tensions

T1 — Lexical Unity versus Internal Variation. An expression can act as a unit while allowing inflection or limited substitution.

Diagnostic: Which changes preserve its conventional reading and distribution?

T2 — Broad Coverage versus Diagnostic Precision. Inclusive inventories capture useful collocations but risk absorbing ordinary phrases.

Diagnostic: What independent evidence establishes unithood for this task?

Structural–Framed Character

Multiword Expression is structural as conventional multi-lexeme unithood and linguistically framed by language and diagnostic practice.

Structural Core vs. Domain Accent

The skeleton is components, conventional association, idiosyncratic property, and permitted variation. Linguistics supplies lexemes, constructions, registers, and compositionality.

This entry presupposes Lexicalization.

  • Approved root. No reviewed parent entails this lexical–grammatical unit.

  • Related — idiom, collocation, compound, construction, and phraseme. They name major subtypes or neighboring analyses.

Relationships to Other Abstractions

Local relationship map for Multiword ExpressionParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Multiword ExpressionDOMAINDomain-specific abstraction: Lexicalization — presupposesLexicalizationDOMAIN

Current abstraction Multiword Expression Domain-specific

Parents (1) — more general patterns this builds on

  • Multiword Expression presupposes Lexicalization Domain-specific

    A Multiword Expression presupposes Lexicalization because its multi-lexeme sequence is stored or conventionalized beyond ordinary free composition.

Hierarchy path (1) — routes to 1 parentless root

Neighborhood in Abstraction Space

Multiword Expression sits in a crowded region of the domain-specific corpus (35th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Language Structure & Grammar Formalisms (23 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With

  • Idiom. Tell: Usually emphasizes nonliteral or noncompositional meaning and is narrower.
  • Collocation. Tell: Emphasizes conventional co-occurrence and may remain transparent.
  • Compound word. Tell: May be analyzed as one lexeme despite multiple roots.
  • Free phrase. Tell: Is generated predictably by grammar and component meanings.

References

  • Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Multiword_expression (revision 1330265755).
  • Preserved source candidate: https://link.springer.com/chapter/10.1007/3-540-45715-1_1
  • Preserved source candidate: http://scidok.sulb.uni-saarland.de/volltexte/2010/3381/pdf/Diss_FINAL.pdf''Idiom
  • Preserved source candidate: http://www.springerlink.com/content/k7etlqv25lxj3j1w
  • Preserved source candidate: https://web.archive.org/web/20030825020644/http://mwe.stanford.edu/
  • Preserved source candidate: https://multiword.sourceforge.net/PHITE.php?sitesig=CONF
  • Preserved source candidate: https://aclweb.org/aclwiki/Multiword_Expressions
  • Preserved source candidate: http://neon.niederlandistik.fu-berlin.de/static/mh/Huening_Schluecker_2015_ms.pdf
  • Preserved source candidate: https://web.archive.org/web/20180127062943/http://neon.niederlandistik.fu-berlin.de/static/mh/Huening_Schluecker_2015_ms.pdf

The frozen Wikipedia revision is discovery provenance. The retained source set was reviewed for identity, formal or operational relation, and scope. The encyclopedia's structural synthesis is bounded to those claims; a thin authority surface is recorded as a nonblocking source-strengthening repair rather than concealed.