A multiword expression (MWE), also called

phraseme A phraseme, also called a set phrase, idiomatic phrase, multi-word expression (in computational linguistics), or idiom, is a multi-word or multi-morphemic utterance whose components include at least one that is selectionally constrained or restr ...

, is a

lexeme A lexeme () is a unit of lexical meaning that underlies a set of words that are related through inflection. It is a basic abstract unit of meaning, a unit of morphological analysis in linguistics that roughly corresponds to a set of forms take ...

-like unit made up of a sequence of two or more lexemes that has properties that are not predictable from the properties of the individual lexemes or their normal mode of combination. MWEs differ from

lexemes A lexeme () is a unit of lexical meaning that underlies a set of words that are related through inflection. It is a basic abstract unit of meaning, a unit of morphological analysis in linguistics that roughly corresponds to a set of forms taken ...

in that the latter are required by many sources to have meaning that cannot be derived from the meaning of separate components. While MWEs must have some properties that cannot be derived from the same property of the components, the property in question does not need to be meaning. For a shorter definition, MWEs can be described as "idiosyncratic interpretations that cross word boundaries (or spaces)". A multiword expression can be a compound, a fragment of a sentence, or a sentence. The group of lexemes which makup up a MWE can be continuous or discontinuous. It is not always possible to mark a MWE with a

part of speech In grammar, a part of speech or part-of-speech (abbreviated as POS or PoS, also known as word class or grammatical category) is a category of words (or, more generally, of lexical items) that have similar grammatical properties. Words that are as ...

. A MWE may be more or less frozen. Example#1 in English: to

kick the bucket To kick the bucket is an English idiom considered a euphemistic, informal, or slang term meaning "to die". Its origin remains unclear, though there have been several theories. Origin theories A common theory is that the idiom refers to hanging, ei ...

, which means ''to die'' rather than ''to hit a bucket with one's foot''. In this example, that is an endocentric compound, the

may be determined as being a ''

verb A verb () is a word ( part of speech) that in syntax generally conveys an action (''bring'', ''read'', ''walk'', ''run'', ''learn''), an occurrence (''happen'', ''become''), or a state of being (''be'', ''exist'', ''stand''). In the usual descr ...

''. The MWE is frozen, in the sense that no variation is possible. Example#2 in English: ''to throw to the lions''. The pattern restricts the usage. The expression is half-frozen because a certain degree of variation is possible but everything is not possible. It is not possible for instance to say ''to the three lions''. Like the previous example, the

is a

. Example#3 in French: ''la moutarde monte au nez''. This MWE is more frozen than the other examples. Let us add that a tense variation is allowed for the

but we cannot determine what is the

for the whole expression because it is a sentence.

Machine Translation (MT) of Multiword Expressions

According to Sag ''et al.'' (2002) Multiword Expressions are, apart from Disambiguation, one of the two key problems for Natural Language Processing (NLP) and especially for machine translation (MT). The number of MWEs in a speaker's lexicon is estimated to be of the same order of magnitude as the number of single words. Specialized domain vocabulary overwhelmingly consists of MWEs, hence, the proportion of MWEs will rise as a system adds vocabulary for new domains, because each domain adds more MWEs than simplex words.

Problems

The greatest problem for translating MWEs might be the idiomaticity problem, as many MWEs have an idiomatic sense, to a higher or a lesser degree. For example, it is hard to predict for a system that an expression like ''kick the bucket'' has a meaning that is totally unrelated to the meaning of ''kick'', ''the'' and ''bucket'' while appearing to conform to the grammar of English Vps. Idioms cannot be translated literally, because in many cases the idiom does not exist in an equivalent form in the target language. Attention has to be paid to syntactic and/or semantic (non)equivalence. Also, not every MWE of the source language has a MWE in the target language as well. For example, the German MWE ''ins Auge fassen'' can only be translated by the English one-word term ''envisage''.

Approaches

The most promising approach to the challenge of translating MWEs is example based MT, because in this case each MWE can be listed as an example with its translation equivalent in the target language. For rule based MT it would be to difficult to define rules to translate MWEs, due to the magnitude of different kinds of MWEs. Nevertheless, an example based MT system has to apply different rules for the translation of continuous and discontinuous MWEs as it is harder to identify a discontinuous MWE in a sentence where words are inserted between the different components of one MWE.

References

*Dimitra Anastasiou
Treatment Experiments in Machine Translation''
(2010), Saarbrücken. *Ivan A. Sag, Timothy Baldwin, Francis Bond, Ann Copestake and Dan Flickinger
''Multiword Expressions: A Pain in the Neck for NLP''
(2002) in: LECTURE NOTES IN COMPUTER SCIENCE, Vol. 2276, pp. 1-15.

External links

Multiword Expression Project
at

Stanford University Stanford University, officially Leland Stanford Junior University, is a private research university in Stanford, California. The campus occupies , among the largest in the United States, and enrolls over 17,000 students. Stanford is conside ...

Workshops on Multiword Expressions

Multiword Expressions in ACL Wiki
* * * * {{Cite book , veditors = Parmentier Y, Waszczuk J , title = Representation and parsing of multiword expressions: Current trends , place = Berlin , publisher = Language Science Press , date = 2019 , format = pdf , url = http://langsci-press.org/catalog/book/202 , doi = 10.5281/zenodo.2579017 , doi-access=free , isbn = 978-3-96110-145-0 , last1 = Parmentier , first1 = Yannick , last2 = Waszczuk , first2 = Jakub Lexicology

Machine Translation (MT) of Multiword Expressions

Problems

Approaches

See also

References

External links