This is part of archived UD v1 documentation. See http://universaldependencies.org/ for the current version.

home v2/v2 issue tracker

Multiword Expressions in UD v2

Since UD does not allow “words with spaces” (but see a partly new proposal under word segmentation), even completely fixed multiword expressions must be annotated with (dummy) dependency relations. To improve annotation consistency, we propose the following changes for v2:

Rename mwe to fixed and make clear that this should only be used for completely fixed expressions
Change the direction of arrows (right-to-left instead of left-to-right) for this relation as well as the other non-dependency relations name and foreign (see also semantic categories)

Rename mwe to fixed

It seems that the label mwe (multiword expression) has led to a lot of confusion. It was never intended for multiword expressions like “kick the bucket”, or Fr. “pomme de terre” (potato). It has always been restricted to the fixed expressions category of Sag et al., excluding any relations in scope of name or compound. The label fixed might reflect this fact better.

The proposed change is therefore to change the label mwe (multiword expression) to fixed, and making the guidelines more restrictive (namely, fixed is used only for completely fixed grammaticized expressions that behave like function words or short adverbials).

Change arrow direction

For non-dependency relations, it was (more or less) arbitrarily decided in v1 to draw arrows from left to right out of the first word. With hindsight, a more harmonious choice for most languages would have been to instead draw arrows from right to left out of the last word. We propose to make this change for fixed (currently mwe), foreign and flat (currently name). Examples:



  
    input
    brat
    info
  
I like dogs as well as cats
fixed(as-6, well-5)
fixed(as-6, as-4)
{
    "text": "I like dogs as well as cats ",
    "entities": [
        [ "embedded-1-T1", "token", [ [ 0, 1 ] ] ],
        [ "embedded-1-T2", "token", [ [ 2, 6 ] ] ],
        [ "embedded-1-T3", "token", [ [ 7, 11 ] ] ],
        [ "embedded-1-T4", "token", [ [ 12, 14 ] ] ],
        [ "embedded-1-T5", "token", [ [ 15, 19 ] ] ],
        [ "embedded-1-T6", "token", [ [ 20, 22 ] ] ],
        [ "embedded-1-T7", "token", [ [ 23, 27 ] ] ]
    ],
    "relations": [
        [ "embedded-1-R1", "fixed", [ [ "arg1", "embedded-1-T6" ], [ "arg2", "embedded-1-T5" ] ] ],
        [ "embedded-1-R2", "fixed", [ [ "arg1", "embedded-1-T6" ], [ "arg2", "embedded-1-T4" ] ] ]
    ],
    "attributes": [],
    "error": false,
    "sentlabels": [ "1" ]
}



  
    input
    brat
    info
  
He cried because of you
fixed(of, because)
{
    "text": "He cried because of you ",
    "entities": [
        [ "embedded-2-T1", "token", [ [ 0, 2 ] ] ],
        [ "embedded-2-T2", "token", [ [ 3, 8 ] ] ],
        [ "embedded-2-T3", "token", [ [ 9, 16 ] ] ],
        [ "embedded-2-T4", "token", [ [ 17, 19 ] ] ],
        [ "embedded-2-T5", "token", [ [ 20, 23 ] ] ]
    ],
    "relations": [
        [ "embedded-2-R1", "fixed", [ [ "arg1", "embedded-2-T4" ], [ "arg2", "embedded-2-T3" ] ] ]
    ],
    "attributes": [],
    "error": false,
    "sentlabels": [ "2" ]
}



  
    input
    brat
    info
  
Je préfère prendre un dessert plutôt qu' une entrée \n I prefer getting a dessert rather than an appetizer
fixed(qu', plutôt)
{
    "text": "Je préfère prendre un dessert plutôt qu' une entrée \n I prefer getting a dessert rather than an appetizer ",
    "entities": [
        [ "embedded-3-T1", "token", [ [ 0, 2 ] ] ],
        [ "embedded-3-T2", "token", [ [ 3, 10 ] ] ],
        [ "embedded-3-T3", "token", [ [ 11, 18 ] ] ],
        [ "embedded-3-T4", "token", [ [ 19, 21 ] ] ],
        [ "embedded-3-T5", "token", [ [ 22, 29 ] ] ],
        [ "embedded-3-T6", "token", [ [ 30, 36 ] ] ],
        [ "embedded-3-T7", "token", [ [ 37, 40 ] ] ],
        [ "embedded-3-T8", "token", [ [ 41, 44 ] ] ],
        [ "embedded-3-T9", "token", [ [ 45, 51 ] ] ],
        [ "embedded-3-T11", "token", [ [ 54, 55 ] ] ],
        [ "embedded-3-T12", "token", [ [ 56, 62 ] ] ],
        [ "embedded-3-T13", "token", [ [ 63, 70 ] ] ],
        [ "embedded-3-T14", "token", [ [ 71, 72 ] ] ],
        [ "embedded-3-T15", "token", [ [ 73, 80 ] ] ],
        [ "embedded-3-T16", "token", [ [ 81, 87 ] ] ],
        [ "embedded-3-T17", "token", [ [ 88, 92 ] ] ],
        [ "embedded-3-T18", "token", [ [ 93, 95 ] ] ],
        [ "embedded-3-T19", "token", [ [ 96, 105 ] ] ]
    ],
    "relations": [
        [ "embedded-3-R1", "fixed", [ [ "arg1", "embedded-3-T7" ], [ "arg2", "embedded-3-T6" ] ] ]
    ],
    "attributes": [],
    "error": false,
    "sentlabels": [ "3" ]
}



  
    input
    brat
    info
  
She said : ez esan lasai
parataxis(said, lasai)
foreign(lasai, esan)
foreign(lasai, ez)
{
    "text": "She said : ez esan lasai ",
    "entities": [
        [ "embedded-4-T1", "token", [ [ 0, 3 ] ] ],
        [ "embedded-4-T2", "token", [ [ 4, 8 ] ] ],
        [ "embedded-4-T3", "token", [ [ 9, 10 ] ] ],
        [ "embedded-4-T4", "token", [ [ 11, 13 ] ] ],
        [ "embedded-4-T5", "token", [ [ 14, 18 ] ] ],
        [ "embedded-4-T6", "token", [ [ 19, 24 ] ] ]
    ],
    "relations": [
        [ "embedded-4-R1", "parataxis", [ [ "arg1", "embedded-4-T2" ], [ "arg2", "embedded-4-T6" ] ] ],
        [ "embedded-4-R2", "foreign", [ [ "arg1", "embedded-4-T6" ], [ "arg2", "embedded-4-T5" ] ] ],
        [ "embedded-4-R3", "foreign", [ [ "arg1", "embedded-4-T6" ], [ "arg2", "embedded-4-T4" ] ] ]
    ],
    "attributes": [],
    "error": false,
    "sentlabels": [ "4" ]
}



  
    input
    brat
    info
  
Usain Bolt won the race
nsubj(won, Bolt)
flat(Bolt, Usain)
{
    "text": "Usain Bolt won the race ",
    "entities": [
        [ "embedded-5-T1", "token", [ [ 0, 5 ] ] ],
        [ "embedded-5-T2", "token", [ [ 6, 10 ] ] ],
        [ "embedded-5-T3", "token", [ [ 11, 14 ] ] ],
        [ "embedded-5-T4", "token", [ [ 15, 18 ] ] ],
        [ "embedded-5-T5", "token", [ [ 19, 23 ] ] ]
    ],
    "relations": [
        [ "embedded-5-R1", "nsubj", [ [ "arg1", "embedded-5-T3" ], [ "arg2", "embedded-5-T2" ] ] ],
        [ "embedded-5-R2", "flat", [ [ "arg1", "embedded-5-T2" ], [ "arg2", "embedded-5-T1" ] ] ]
    ],
    "attributes": [],
    "error": false,
    "sentlabels": [ "5" ]
}