Keyboard shortcuts

Press or to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Annotating Nonlocal Dependencies

Nonlocal dependencies are dependencies from one clause into another. They do not exist as syntactic dependencies in (basic) UD, so for Superframes annotation, we add them when needed. We have already seen nonlocal dependencies in the previous chapter with attributive adjectives, adverbs, and relative clauses: these elements are attached to their head using the META frame (manifesting as a SCENE modifying relation), and then a nonlocal dependency (a “backlink”) is added from the modifying element back to its head.

There are a number of other constructions that systematically introduce nonlocal dependencies, which we also annotate. We give examples of each in the following.

Subject control:

Dependency graph for sentence: Kim promised Sandy to come. Token promised is labeled NORM-INIT and has an edge labeled SENDER to token Kim and an edge labeled EXPERIENCER to token Sandy and an edge labeled norm to token come. Token come is labeled LOCATION-INIT and has an edge labeled has-location to token Kim. Dependency graph for sentence: Kim used a hammer to smash the vase. Token used is labeled MEANS and has an edge labeled CAUSER to token Kim and an edge labeled means to token hammer and an edge labeled end to token smash. Token smash is labeled INTERNAL-STATE-INIT and has an edge labeled CAUSER to token Kim and an edge labeled has-state to token vase.

Object control:

Dependency graph for sentence: Kim persuaded Sandy to come. Token persuaded is labeled NORM-INIT and has an edge labeled SENDER to token Kim and an edge labeled EXPERIENCER to token Sandy and an edge labeled norm to token come. Token come is labeled LOCATION-INIT and has an edge labeled has-location to token Sandy.

Non-obligatory control:

Dependency graph for sentence: Kim left after trashing the room. Token left is labeled LOCATION-DEINIT and has an edge labeled has-location to token Kim and an edge labeled PRECEDENT to token trashing. Token trashing is labeled INTERNAL-STATE-INIT and has an edge labeled CAUSER to token Kim and an edge labeled has-state to token room. Dependency graph for sentence: Kim has come to stay. Token come is labeled LOCATION-INIT and has an edge labeled has-location to token Kim and an edge labeled END to token stay. Token stay is labeled LOCATION-CONTINUATION and has an edge labeled has-location to token Kim.

Non-obligatory control with parasitic gap:

Dependency graph for sentence: the question we raised without answering. Token question is labeled ENTITY and has an edge labeled SCENE to token raised. Token raised is labeled MESSAGE-INIT and has an edge labeled message to token question and an edge labeled SENDER to token we and an edge labeled ACCOMPANIER-NEG to token answering. Token answering is labeled MESSAGE-INIT and has an edge labeled topic to token question and an edge labeled SENDER to token we.

Raising:

Dependency graph for sentence: Kim seemed to fly. Token seemed is labeled MESSAGE and has an edge labeled topic to token Kim and an edge labeled message to token fly. Token fly is labeled LOCATION-DYN and has an edge labeled has-location to token Kim.

Secondary predicate (see also the section on secondary predicates below):

Dependency graph for sentence: Kim entered the room singing. Token entered is labeled LOCATION-INIT and has an edge labeled has-location to token Kim and an edge labeled location to token room and an edge labeled ACCOMPANIER to token singing. Token singing is labeled SENDING and has an edge labeled sender to token Kim.

Tough construction:

Dependency graph for sentence: Kim is hard to love. Token hard is labeled META-REALITY and has an edge labeled participant to token Kim and an edge labeled scene to token love. Token love is labeled EVALUATION and has an edge labeled evaluee to token Kim.

Relative clause:

Dependency graph for sentence: the song that I like. Token song is labeled ENTITY and has an edge labeled EXPERIENCER to token I and an edge labeled SCENE to token like. Token like is labeled EVALUATION and has an edge labeled evaluee to token song and an edge labeled evaluee to token that.

Reduced relative clause:

Dependency graph for sentence: the song I like. Token song is labeled ENTITY and has an edge labeled EXPERIENCER to token I and an edge labeled SCENE to token like. Token like is labeled EVALUATION and has an edge labeled evaluee to token song.

Non-finite reduced relative clause:

Dependency graph for sentence: the song liked by Kim. Token song is labeled ENTITY and has an edge labeled SCENE to token liked. Token liked is labeled EVALUATION and has an edge labeled evaluee to token song and an edge labeled EXPERIENCER to token Kim. Dependency graph for sentence: students living on campus. Token students is labeled ENTITY and has an edge labeled SCENE to token living. Token living is labeled LOCATION-HABIT and has an edge labeled has-location to token students and an edge labeled location to token campus.

Relative clause with complex extracted element:

Dependency graph for sentence: eine Gestalt , deren Magerkeit durch den Trainingsanzug noch betont wurde. Token Gestalt is labeled ENTITY and has an edge labeled SCENE to token betont. Token Magerkeit is labeled PROPERTY and has an edge labeled has-property to token Gestalt. Token noch is labeled ACCOMPANIMENT and has an edge labeled accompanier to token betont. Token betont is labeled MESSAGE and has an edge labeled topic to token Magerkeit and an edge labeled SENDER to token Trainingsanzug and an edge labeled SCENE to token noch.

Relative clause with extraction across clause boundaries:

Dependency graph for sentence: Atmosphäre , mit der sie sich zu umgeben wusste. Token Atmosphäre is labeled ENTITY and has an edge labeled SCENE to token wusste. Token umgeben is labeled SITUATION-INIT and has an edge labeled situator to token Atmosphäre and an edge labeled situator to token der and an edge labeled situee to token sich. Token wusste is labeled META-POSSIBILITY and has an edge labeled participant to token sie and an edge labeled scene to token umgeben.

Coordination:

Dependency graph for sentence: ein sogenannter Televisor oder Hörsehschirm. Token sogenannter is labeled IDENTIFIER-INIT and has an edge labeled identified to token Televisor and an edge labeled identified to token Hörsehschirm. Token Televisor is labeled ENTITY and has an edge labeled SCENE to token sogenannter. Token Hörsehschirm is labeled ENTITY and has an edge labeled SCENE to token sogenannter.

Secondary Predicates

Secondary predicates are modifiers that syntactically attach to a (primary) predicate, but semantically predicate over one of the primary predicate’s arguments, or even something more deeply embedded. The semantic relation between the primary and the secondary predicate can be one of simple accompaniment (these secondary predicates are also called depictives), effect (these secondary predicates are also called resultatives), or something else.

Dependency graph for sentence: Kim entered the room singing. Token entered is labeled LOCATION-INIT and has an edge labeled has-location to token Kim and an edge labeled location to token room and an edge labeled ACCOMPANIER to token singing. Token singing is labeled SENDING and has an edge labeled sender to token Kim. Dependency graph for sentence: You ’re talking me silly. Token talking is labeled SENDING and has an edge labeled sender to token You and an edge labeled EFFECT-INIT to token me and an edge labeled EFFECT-INIT to token silly. Token silly is labeled INTERNAL-STATE and has an edge labeled has-state to token me. Dependency graph for sentence: Stop drawing sheep , whether small or big. Token Stop is labeled META-DEINIT and has an edge labeled scene to token drawing and an edge labeled CONDITION to token small and an edge labeled CONDITION to token big. Token drawing is labeled MESSAGE-INIT and has an edge labeled topic to token sheep. Token small is labeled PROPERTY and has an edge labeled has-property to token sheep. Token big is labeled PROPERTY and has an edge labeled has-property to token sheep. Dependency graph for sentence: Some people , though short , reach amazing heights. Token short is labeled PROPERTY and has an edge labeled has-property to token people. Token reach is labeled META-INIT and has an edge labeled participant to token people and an edge labeled CONCEDED to token short and an edge labeled scene to token heights. Token amazing is labeled PROPERTY and has an edge labeled has-property to token heights. Token heights is labeled PROPERTY and has an edge labeled has-property to token people.

Note that in the resultative example You’re talking me silly, the primary predicate talking does not assign a role to its syntactic object me, only the secondary predicate silly does. But the annotation scheme still forces us to label the dependency from talking to me. We solve this by labeling it EFFECT-INIT too, just like the dependency to the secondary predicate.