The diverse and complex nature of multiword units has led to the creation of various taxonomies and typologies in an attempt to make their study useful in various applications. However, each classification is created for a different purpose: information extraction, educational and lexicographic uses, NLP applications (parsing etc). As a consequence, a classification that has been created for a certain application/use can be used only or mainly for this application.

In any case, when it comes to multiword units there should be a distinction, between the various categories (noun, verb), also, it should be clear on what grounds the classification is created, and finally, which properties define it. In an effort to address these issues, we opted for an “integrative” method, which takes into account all the syntactic structures of our language data, a classification that is based on semantic, syntactic and morphological properties as well. This method allows users to choose what they really need to use in every occasion.

More specifically, the classification of Greek verbal MWEs has been defined in (Fotopoulou, 1993,Mini, 2009) totaling 6,000 entries, on the basis of the lexical and syntactic configurations involved. In this classification adopted for the verbal idiomatic expressions, fixedness can be limited to only certain constituents of the sentence; a combination of fixed and non-fixed constituents in Subject or Object position is permitted.

For example, in sentences below, fixedness relies on the relation between the verbs and the nouns that function as Objects (direct and indirect) (1) and as Subject (2), respectively:

(1)    δίνω/V τόπο/NP-acc στην οργή/PP-acc

to-give-way to anger (=to swallow one’s pride/anger)

(2)    ανάβουν/V τα λαμπάκια μου/NP-nom

are-switched-on my lights  (=to become very angry)

Moreover, the typology allows for a restricted alternation of fixed elements of the expression. For example, in the MWE of (3), two alternative lemmas can be used, viz. τάζω and υπόσχομαι ‘to promise’:

(3)    τάζω/υπόσχομαι/V τον ουρανό/NP-acc με τ’ άστρα/PP-acc

to offer/promise the sky with the stars

In the Polylex database, cases as the above are encoded as different entries in the syntactic tables in order to establish the range of lexical options.

This information is encoded in verbal MWEs, namely: (a) the syntactic properties of the verb that occurs in the expression (valency), and (b) fixed and non-fixed arguments, either in Subject or in Object position. Moreover, selectional restrictions applied to the arguments, constraints on word order, and other transformational properties like pronominalisation, passive, etc., are also added.

Below there are some examples of syntactic tables with classifications of verbal MWEs with fixed Subject and Complement, respectively:

Classification of verbal MWEs with fixed subject

Classification of verbal ΜWEs with fixed complement (s)