corpora of multiword expressions - version 1.2 (2020)
shared task on semi-supervised identification of verbal multiword expressions - edition 1.2 (2020)
Inherently adpositional verbs (IAVs)
Inherently adpositional verb (IAV) is a special optional and experimental category (corresponding to the IPrepV category in the first pilot annotations), and to what is also sometimes called in English prepositional verbs. It consists of a verb or VMWE and an idiomatic selected preposition or postposition that is either always required or, if absent, changes the meaning of the verb of VMWE significantly. Language teams who decide to annotate IAV should do so after annotating other categories (step 4 of the annotation process), since overlapping can be quite frequent with other categories, as detailed below. Language teams are not required to use this category.
Our definition of inherently adpositional verbs is a generalization (applying to many languages) of the annotation guidelines of the English STREUSLE corpus, which define guidelines for annotating prepositional verbs.
IAVs are verb+adposition combinations in which:
- the dependents of the adposition are not lexicalized
разчитам на някого/нещо to rely on somebody/something is annotated as IAV because the object is not lexicalised,
but in the ID вземам на мушка някого/нещо take on target to critisise heavily somebody/something cannot be annotated as IAV because мушка is also lexicalized in the ID
to stand for something is annotated as IAV because the object is not lexicalized,
but in the ID to take something for granted, to take for cannot be annotated as IAV because granted is also lexicalized in the ID
- entender de algo understand of somethingto know about something is annotated as IAV because the object is not lexicalised, whereas entender algo would not be any type of VMWE.
pristati na kaj to land on (something) to agree (with something)is annotated as IAV because the object is not lexicalized,
but in the ID ostati na trdnih tleh to remain on solid ground to remain realistic ostati na to remain on cannot be annotated as IAV because trdnih tleh solid ground is also lexicalized in the ID
- разчитам на някого/нещо to rely on somebody/something is annotated as IAV because the object is not lexicalised,
- the adposition is integral, that is, "it cannot be omitted without markedly altering the meaning of the verb"
считам за to take for → *считам can never occur without the preposition за
разчитам на to rely on → разчитам can occur without the preposition, but it will never have a sense of to depend/rely on
to rely on → *to rely can never occur without the preposition on
to count on → to count can occur without the preposition, but it will never have a sense of to depend/rely on
entender de understand of somethingto know about something → entender to understandcan occur without the preposition, but it will never have a sense of to be an expert about something
contar con count withto rely on → contar to countcan occur without the preposition, but it will never have a sense of to rely on.
temeljiti na to be based on → *temeljiti can never occur without the preposition na
biti za to be for to agree with or support (something or someone)→ biti to be can occur without the preposition, but it will never have a sense of to agree with or to support
- считам за to take for → *считам can never occur without the preposition за
Note that idiomatic adpositional valency, in which the adposition opens a slot for a complement, should not be mistaken for verb-particle constructions. Tests distinguishing particles from prepositions can be used to disambiguate these categories.
to wake up somebody cannot be annotated as IAV because up is a particle, and not a preposition.
Particles can occur after the object: to wake somebody up but prepositions cannot *to come a new restaurant across
Not only single verbs but also VMWEs may be inherently adpositional. This is why IAV annotation needs to be the last step, after all other VMWEs in a sentence have been identified and categorized. In case of overlap between another category and IAV, the whole VMWE annotation needs to be repeated with the addition of the lexicalized adposition, and the whole is annotated as an IAV.
to put up with bears 2 annotations:
1. to put up is annotated as VPC
2. the whole sequence to put up with is annotated as IAV
atenerse a abide.self to to abide by bears 2 annotations:
1. atenerse is annotated as IRV
2. the whole sequence atenerse a is annotated as IAV
ubadati se z to deal RCLI with to deal with bears 2 annotations:
1. ubadati se to deal RCLI is annotated as IRV, since the verb without the RCLI does not exist
2. the whole sequenceubadati se z to deal RCLI withis annotated as IAV, since the verb also does not exist without the preposition
In response to a declarative sentence with the verb+adposition combination, is there a natural way to query the circumstances of the verbal event using the verb, but not the adposition?
- it is not an IAV
- I care about the environment.
- Why do you care?
→ to care about is not annotated as IAV
- me preocupo por su salud. me worry.I for his/her health I'm worried about his/her health
- ¿Por qué te preocupas?why you worry.you? Why are you worried?
→ preocuparse por is not annotated as IAV
- Lahko se zanesem na pomoč svojih prijateljev. I can rely on my friends' help
- Se lahko zaneseš, da ti bo kdo pomagal? Can you rely that someone will help you?Can you rely on that someone will help you?
→ zanesti se to rely on is not annotated as IAV
- annotate as an IAV
- I came across a nice restaurant downtown.
- #When did you come?
to come across is annotated as IAV
- Ana entiende de música clásica. Ana understands of music classic Ana knows about classical music
- #¿Desde cuándo entiende? Since when understands.she?Since when does she know?
entender de is annotated as IAV
Gre za enakovrednost. It goes about equality. It is about equality.
- #Kaj gre? #What goes?
gre za is annotated as IAV