multiwordexpressies

Multiwordexpressies, often abbreviated MWE, refer to lexical units composed of two or more words that function as a single semantic and syntactic unit in a language. They are distinguished from freely compositional phrases because their overall meaning is not always derivable from the meanings of the individual parts.

MWEs encompass a range of phenomena, including idioms like kick the bucket (to die), fixed expressions such as by and large, collocations like strong tea, phrasal verbs such as look up, and light verb constructions like take a walk. They can also include proper name expressions or specialized technical phrases. The boundaries between subtypes are not always strict, and a single expression may be described in multiple ways depending on linguistic perspective.

Linguistic properties of MWEs include non compositional or partially non compositional semantics, fixed or semi fixed word order, and limited syntactic variation. Some MWEs are highly productive in some languages, while others are highly idiosyncratic and resistant to standard parsing. Detecting and analyzing MWEs presents difficulties because their meanings may depend on context, they may change form with inflection or derivation, and they often resist straightforward dictionary based lookup.

In computational linguistics MWEs pose challenges for natural language processing tasks such as tokenization, disambiguation, parsing, machine translation, and information retrieval. Effective handling typically relies on specialized resources such as MWE dictionaries or lexicons, statistical or hybrid detection methods, and dedicated evaluation benchmarks. Applications include improving translation quality, enhancing search and indexing, and supporting lexicography and language teaching.