This is part of archived UD v1 documentation. See http://universaldependencies.org/ for the current version.
home en/dep issue tracker

mwe: multi-word expression

The multi-word expression (modifier) relation is used for certain fixed grammaticized expressions with function words that behave like a single function word. Multiword expressions are annotated in a flat, head-initial structure, in which all words in the expression modify the first one using the mwe label.

At present, this relation is used inside the following expressions:

as well

I like dogs as well
advmod(like, as)
mwe(as, well)

as well as

I like dogs as well as cats
mwe(as-4, well)
mwe(as-4, as-6)
cc(dogs, as-4)

such as

I like fluffy animals , such as dogs
case(dogs, such)
mwe(such, as)

due to (and other forms, such as d t and d/t)

He cried due to the fact that you hurt him
case(fact, due)
mwe(due, to)

because of (and other forms, such as b c of and b/c of)

He cried because of you
case(you, because)
mwe(because, of)

instead of

John went instead of Mary
mwe(instead, of)
case(Mary, instead)
John left early instead of staying for the whole thing
mwe(instead, of)
mark(staying, instead)

in case

I always back up my files in case my computer crashes
mwe(in, case)
mark(crashes, in)
I always back up my files just in case
mwe(in, case)
advmod(back, in)
advmod(in, just)

in case of

I always back up my files in case of a crash
mwe(in, case)
mwe(in, of)
case(crash, in)

of course

I like dogs , of course
advmod(like, of)
mwe(of, course)

so that

He cried so that you would feel bad
mark(feel, so)
mwe(so, that)

more than (when used synonymously with “over” in a quantity)

More than 90 percent
advmod(percent, More)
mwe(More, than)

less than (when used synonymously with “under” in a quantity)

Less than ten percent
advmod(percent, Less)
mwe(Less, than)

up to (when used in quantities)

Up to fifty percent
mwe(Up, to)
advmod(percent, Up)

according to

According to John
mwe(According, to)
case(John, According)

in order

He cried in order to make you feel bad
mark(feel, in)
mwe(in, order)
He cried in order that you might feel bad
mark(feel, in)
mwe(in, order)
He cried in order for you to have something to feel bad about
mark(have, in)
mwe(in, order)

rather than

I decided to get a dog rather than a cat
mwe(rather, than)
cc(rather, dog)

at least (when not used for quantities)

At least I like dogs
mwe(At, least)
advmod(like, At)

as if

It was as if he cried to make you feel bad
mwe(as, if)
mark(cried, as)

prior to

John left prior to the meeting
mwe(prior, to)
case(meeting, prior)

as to

As to whether I love dogs ...
mwe(As, to)
mark(love, As)
As to my love of dogs ...
mwe(As, to)
case(love, As)

kind of

I kind of like dogs
mwe(kind, of)
advmod(like, kind)

whether or not

He 's crying whether or not you feel bad about it
mwe(whether, or)
mwe(whether, not)
mark(feel, whether)

not to mention

This restaurant is pretty cheap with good food, not to mention their friendly staff
mwe(not, to)
mwe(not, mention)
cc(cheap, not)
conj(staff, cheap)

as opposed to

John decided to leave early , as opposed to Mary
mwe(as, opposed)
mwe(as, to-9)
case(Mary, as)

let alone

He could n't handle being hurt , let alone hurt by you
mwe(let, alone)
cc(hurt-6, alone)
conj(hurt-6, hurt-10)

so as to

John left early so as to miss the meeting
mwe(so, as)
mwe(so, to)
mark(so, miss)

in between

John left in between meetings
mwe(in, between)
case(in, meetings)

all but

John has all but left
mwe(all, but)
advmod(all, left)

that is

The dogs need to be housebroken -- that is , '' potty - trained ''
mwe(that, is)
advmod(trained, that)

how come

How come John left early ?
mwe(How, come)
mark(left, How)

had better (and ‘d better)

You had better apologize
mwe(had, better)
aux(had, apologize)

Not mwes

The following are not annotated as mwes, but are instead labeled according to their apparent internal structure.

out of, off of (All double prepositions denoting spatial relations are annotated with two cases on the nominal)

Get out of there !
nmod(Get, there)
case(there, out)
case(there, of)
Get off of that !
nmod(Get, that)
case(that, off)
case(that, of)

by far

Dogs are the best animal by far
nmod(animal, far)
case(far, by)

what about

What about John ?
nmod(What, John)
case(John, about)

at all

I don't like her at all
nmod(like, all)
case(all, at)

at most, at least (when used for quantities. To determine whether at least should be an mwe or not in borderline cases, substitute it with at most; if the sentence remains grammatical, it should receive its surface analysis)

at most 50 percent
nmod:npmod(percent, most)
case(most, at)
at least 50 percent
nmod:npmod(percent, least)
case(least, at)

at best, at worst

At best , they were guesses
nmod:npmod(guesses, best)
case(best, At)
At worst , they were lies
nmod:npmod(lies, worst)
case(worst, At)

what if

What if John left early ?
advcl(What, left)
mark(left, if)

so long

So long , Ham 's ... you will be missed
advmod(long, So)
vocative(long, Ham)
parataxis(long, missed)

Treebank Statistics (UD_English)

This relation is universal.

592 nodes (0%) are attached to their parents as mwe.

592 instances of mwe (100%) are left-to-right (parent precedes child). Average distance between parent and child is 1.09121621621622.

The following 25 pairs of parts of speech are connected with mwe: en-pos/ADV-en-pos/ADV (155; 26% instances), en-pos/ADP-en-pos/ADP (145; 24% instances), en-pos/ADJ-en-pos/ADP (79; 13% instances), en-pos/ADV-en-pos/ADP (57; 10% instances), en-pos/SCONJ-en-pos/SCONJ (32; 5% instances), en-pos/ADP-en-pos/NOUN (24; 4% instances), en-pos/VERB-en-pos/ADP (21; 4% instances), en-pos/ADP-en-pos/SCONJ (13; 2% instances), en-pos/SCONJ-en-pos/ADV (9; 2% instances), en-pos/SCONJ-en-pos/CONJ (9; 2% instances), en-pos/ADV-en-pos/VERB (8; 1% instances), en-pos/ADJ-en-pos/SCONJ (7; 1% instances), en-pos/ADV-en-pos/PART (7; 1% instances), en-pos/ADV-en-pos/SCONJ (6; 1% instances), en-pos/SCONJ-en-pos/NOUN (4; 1% instances), en-pos/SCONJ-en-pos/VERB (4; 1% instances), en-pos/ADP-en-pos/VERB (2; 0% instances), en-pos/SCONJ-en-pos/PART (2; 0% instances), en-pos/VERB-en-pos/ADV (2; 0% instances), en-pos/ADP-en-pos/ADV (1; 0% instances), en-pos/ADP-en-pos/CONJ (1; 0% instances), en-pos/ADP-en-pos/PUNCT (1; 0% instances), en-pos/AUX-en-pos/ADV (1; 0% instances), en-pos/SCONJ-en-pos/ADJ (1; 0% instances), en-pos/SCONJ-en-pos/ADP (1; 0% instances).

# visual-style 12	bgColor:blue
# visual-style 12	fgColor:white
# visual-style 11	bgColor:blue
# visual-style 11	fgColor:white
# visual-style 11 12 mwe	color:blue
1	To	to	PART	TO	_	2	mark	_	_
2	pander	pander	VERB	VB	VerbForm=Inf	0	root	_	_
3	to	to	ADP	IN	_	8	case	_	_
4	the	the	DET	DT	Definite=Def|PronType=Art	8	det	_	_
5	mythical	mythical	ADJ	JJ	Degree=Pos	8	amod	_	_
6	"	"	PUNCT	``	_	8	punct	_	SpaceAfter=No
7	Arab	arab	ADJ	JJ	Degree=Pos	8	amod	_	_
8	street	street	NOUN	NN	Number=Sing	2	nmod	_	SpaceAfter=No
9	"	"	PUNCT	''	_	8	punct	_	SpaceAfter=No
10	,	,	PUNCT	,	_	12	punct	_	_
11	of	of	ADV	RB	_	2	advmod	_	_
12	course	course	ADV	RB	_	11	mwe	_	SpaceAfter=No
13	.	.	PUNCT	.	_	12	punct	_	_

# visual-style 8	bgColor:blue
# visual-style 8	fgColor:white
# visual-style 7	bgColor:blue
# visual-style 7	fgColor:white
# visual-style 7 8 mwe	color:blue
1	do	do	AUX	VBP	Mood=Ind|Tense=Pres|VerbForm=Fin	3	aux	_	_
2	you	you	PRON	PRP	Case=Nom|Person=2|PronType=Prs	3	nsubj	_	_
3	think	think	VERB	VB	VerbForm=Inf	0	root	_	_
4	they	they	PRON	PRP	Case=Nom|Number=Plur|Person=3|PronType=Prs	6	nsubj	_	_
5	are	be	VERB	VBP	Mood=Ind|Tense=Pres|VerbForm=Fin	6	cop	_	_
6	cool	cool	ADJ	JJ	Degree=Pos	3	ccomp	_	_
7	b/c	b/c	ADP	IN	_	12	case	_	_
8	of	of	ADP	IN	_	7	mwe	_	_
9	the	the	DET	DT	Definite=Def|PronType=Art	12	det	_	_
10	taco	taco	PROPN	NNP	Number=Sing	11	compound	_	_
11	bell	bell	PROPN	NNP	Number=Sing	12	compound	_	_
12	dog	dog	NOUN	NN	Number=Sing	3	nmod	_	SpaceAfter=No
13	?	?	PUNCT	.	_	3	punct	_	_

# visual-style 3	bgColor:blue
# visual-style 3	fgColor:white
# visual-style 2	bgColor:blue
# visual-style 2	fgColor:white
# visual-style 2 3 mwe	color:blue
1	Now	now	ADV	RB	_	5	advmod	_	_
2	more	more	ADJ	JJR	Degree=Cmp	4	advmod	_	_
3	than	than	ADP	IN	_	2	mwe	_	_
4	1100	1100	NUM	CD	NumType=Card	5	nummod	_	_
5	pictures	picture	NOUN	NNS	Number=Plur	0	root	_	SpaceAfter=No
6	.	.	PUNCT	.	_	5	punct	_	_


Treebank Statistics (UD_English-ESL)

This relation is universal.

357 nodes (0%) are attached to their parents as mwe.

357 instances of mwe (100%) are left-to-right (parent precedes child). Average distance between parent and child is 1.04761904761905.

The following 19 pairs of parts of speech are connected with mwe: en-pos/ADP-en-pos/ADP (96; 27% instances), en-pos/ADV-en-pos/ADV (68; 19% instances), en-pos/ADV-en-pos/ADP (55; 15% instances), en-pos/ADJ-en-pos/ADP (34; 10% instances), en-pos/SCONJ-en-pos/SCONJ (30; 8% instances), en-pos/ADP-en-pos/NOUN (28; 8% instances), en-pos/VERB-en-pos/ADP (25; 7% instances), en-pos/AUX-en-pos/ADV (4; 1% instances), en-pos/SCONJ-en-pos/PART (4; 1% instances), en-pos/SCONJ-en-pos/NOUN (3; 1% instances), en-pos/SCONJ-en-pos/ADP (2; 1% instances), en-pos/ADJ-en-pos/ADV (1; 0% instances), en-pos/ADP-en-pos/ADV (1; 0% instances), en-pos/ADV-en-pos/CONJ (1; 0% instances), en-pos/ADV-en-pos/DET (1; 0% instances), en-pos/DET-en-pos/VERB (1; 0% instances), en-pos/INTJ-en-pos/PRON (1; 0% instances), en-pos/PART-en-pos/PART (1; 0% instances), en-pos/PART-en-pos/VERB (1; 0% instances).

# visual-style 16	bgColor:blue
# visual-style 16	fgColor:white
# visual-style 15	bgColor:blue
# visual-style 15	fgColor:white
# visual-style 15 16 mwe	color:blue
1	_	_	ADP	IN	_	3	case	_	_
2	_	_	DET	DT	_	3	det	_	_
3	_	_	NOUN	NN	_	5	nmod	_	_
4	_	_	PRON	PRP	_	5	nsubj	_	_
5	_	_	VERB	VBD	_	0	root	_	_
6	_	_	PART	TO	_	7	mark	_	_
7	_	_	VERB	VB	_	5	xcomp	_	_
8	_	_	DET	DT	_	9	det	_	_
9	_	_	NOUN	NN	_	7	dobj	_	_
10	_	_	PUNCT	,	_	5	punct	_	_
11	_	_	CONJ	CC	_	5	cc	_	_
12	_	_	PRON	PRP	_	14	nsubj	_	_
13	_	_	VERB	VBD	_	14	cop	_	_
14	_	_	ADJ	JJ	_	5	conj	_	_
15	_	_	ADP	IN	_	17	case	_	_
16	_	_	ADP	IN	_	15	mwe	_	_
17	_	_	NOUN	NNS	_	14	nmod	_	_
18	_	_	ADP	IN	_	20	case	_	_
19	_	_	DET	DT	_	20	det	_	_
20	_	_	NOUN	NN	_	17	nmod	_	_
21	_	_	PUNCT	.	_	5	punct	_	_

# visual-style 2	bgColor:blue
# visual-style 2	fgColor:white
# visual-style 1	bgColor:blue
# visual-style 1	fgColor:white
# visual-style 1 2 mwe	color:blue
1	_	_	ADV	RB	_	16	advmod	_	_
2	_	_	ADV	RB	_	1	mwe	_	_
3	_	_	PUNCT	,	_	16	punct	_	_
4	_	_	SCONJ	IN	_	6	mark	_	_
5	_	_	PRON	NN	_	6	nsubj	_	_
6	_	_	VERB	VBZ	_	16	advcl	_	_
7	_	_	ADJ	JJ	_	8	amod	_	_
8	_	_	NOUN	NN	_	6	dobj	_	_
9	_	_	CONJ	CC	_	8	cc	_	_
10	_	_	NOUN	NN	_	8	conj	_	_
11	_	_	PUNCT	,	_	16	punct	_	_
12	_	_	DET	DT	_	13	det	_	_
13	_	_	NOUN	NNS	_	16	nsubj	_	_
14	_	_	AUX	MD	_	16	aux	_	_
15	_	_	VERB	VB	_	16	cop	_	_
16	_	_	ADJ	JJR	_	0	root	_	_
17	_	_	PUNCT	.	_	16	punct	_	_

# visual-style 18	bgColor:blue
# visual-style 18	fgColor:white
# visual-style 16	bgColor:blue
# visual-style 16	fgColor:white
# visual-style 16 18 mwe	color:blue
1	_	_	ADP	IN	_	3	case	_	_
2	_	_	DET	DT	_	3	det	_	_
3	_	_	NOUN	NN	_	7	nmod	_	_
4	_	_	PUNCT	,	_	7	punct	_	_
5	_	_	PRON	PRP	_	7	nsubj	_	_
6	_	_	AUX	MD	_	7	aux	_	_
7	_	_	VERB	VB	_	0	root	_	_
8	_	_	NOUN	NNS	_	7	dobj	_	_
9	_	_	SCONJ	IN	_	11	mark	_	_
10	_	_	PRON	PRP	_	11	nsubj	_	_
11	_	_	VERB	VBP	_	7	advcl	_	_
12	_	_	DET	DT	_	13	det	_	_
13	_	_	NOUN	NN	_	11	dobj	_	_
14	_	_	ADP	IN	_	15	case	_	_
15	_	_	NOUN	NN	_	13	nmod	_	_
16	_	_	ADV	RB	_	13	cc	_	_
17	_	_	ADV	RB	_	16	mwe	_	_
18	_	_	ADP	IN	_	16	mwe	_	_
19	_	_	NOUN	NN	_	20	compound	_	_
20	_	_	NOUN	NNS	_	13	conj	_	_
21	_	_	PUNCT	.	_	7	punct	_	_


Treebank Statistics (UD_English-LinES)

This relation is universal.

469 nodes (1%) are attached to their parents as mwe.

468 instances of mwe (100%) are left-to-right (parent precedes child). Average distance between parent and child is 1.12366737739872.

The following 32 pairs of parts of speech are connected with mwe: en-pos/ADV-en-pos/ADV (124; 26% instances), en-pos/ADP-en-pos/ADP (66; 14% instances), en-pos/ADV-en-pos/ADP (43; 9% instances), en-pos/PRON-en-pos/PRON (40; 9% instances), en-pos/ADV-en-pos/NOUN (38; 8% instances), en-pos/ADP-en-pos/SCONJ (31; 7% instances), en-pos/SCONJ-en-pos/SCONJ (21; 4% instances), en-pos/ADJ-en-pos/ADP (14; 3% instances), en-pos/ADP-en-pos/NOUN (14; 3% instances), en-pos/CONJ-en-pos/ADP (10; 2% instances), en-pos/CONJ-en-pos/ADV (10; 2% instances), en-pos/DET-en-pos/ADJ (9; 2% instances), en-pos/NUM-en-pos/ADV (6; 1% instances), en-pos/NUM-en-pos/CONJ (6; 1% instances), en-pos/PROPN-en-pos/PROPN (6; 1% instances), en-pos/ADV-en-pos/ADJ (5; 1% instances), en-pos/ADV-en-pos/CONJ (4; 1% instances), en-pos/PART-en-pos/ADP (3; 1% instances), en-pos/PART-en-pos/NOUN (3; 1% instances), en-pos/DET-en-pos/ADP (2; 0% instances), en-pos/DET-en-pos/NOUN (2; 0% instances), en-pos/PRON-en-pos/ADJ (2; 0% instances), en-pos/ADJ-en-pos/ADJ (1; 0% instances), en-pos/ADJ-en-pos/NOUN (1; 0% instances), en-pos/DET-en-pos/ADV (1; 0% instances), en-pos/DET-en-pos/CONJ (1; 0% instances), en-pos/PRON-en-pos/ADP (1; 0% instances), en-pos/SCONJ-en-pos/ADP (1; 0% instances), en-pos/SCONJ-en-pos/ADV (1; 0% instances), en-pos/SCONJ-en-pos/CONJ (1; 0% instances), en-pos/SCONJ-en-pos/PART (1; 0% instances), en-pos/VERB-en-pos/ADP (1; 0% instances).

# visual-style 7	bgColor:blue
# visual-style 7	fgColor:white
# visual-style 6	bgColor:blue
# visual-style 6	fgColor:white
# visual-style 6 7 mwe	color:blue
1	All	_	ADV	_	_	0	root	_	_
2	right	_	ADJ	_	_	1	mwe	_	_
3	,	_	PUNCT	Comma	_	1	punct	_	_
4	said	_	VERB	PAST	_	1	parataxis	_	_
5	Quinn	_	PROPN	SG	_	4	nsubj	_	_
6	at	_	ADV	_	_	4	advmod	_	_
7	last	_	ADV	_	_	6	mwe	_	_
8	.	_	PUNCT	Period	_	1	punct	_	_

# visual-style 7	bgColor:blue
# visual-style 7	fgColor:white
# visual-style 6	bgColor:blue
# visual-style 6	fgColor:white
# visual-style 6 7 mwe	color:blue
1	Or	_	CONJ	_	_	5	cc	_	_
2	you	_	PRON	PERS-P2	_	5	nsubj	_	_
3	might	_	AUX	PAST-AUX	_	5	aux	_	_
4	well	_	ADV	_	_	5	advmod	_	_
5	fall	_	VERB	INF	_	0	root	_	_
6	out	_	ADP	_	_	10	case	_	_
7	of	_	ADP	_	_	6	mwe	_	_
8	the	_	DET	DEF	_	10	det	_	_
9	wrong	_	ADJ	POS	_	10	amod	_	_
10	fireplace	_	NOUN	SG	_	5	nmod	_	_
11	–	_	PUNCT	Dash	_	5	punct	_	_

# visual-style 4	bgColor:blue
# visual-style 4	fgColor:white
# visual-style 2	bgColor:blue
# visual-style 2	fgColor:white
# visual-style 2 4 mwe	color:blue
1	Surely	_	ADV	_	_	6	advmod	_	_
2	no	_	ADV	NEG	_	6	advmod	_	_
3	more	_	ADV	CMP-SG	_	2	mwe	_	_
4	than	_	ADP	_	_	2	mwe	_	_
5	a	_	DET	IND-SG	_	6	det	_	_
6	minute	_	NOUN	SG	_	0	root	_	_
7	or	_	CONJ	_	_	6	cc	_	_
8	two	_	NUM	CARD-PL	_	6	conj	_	_
9	.	_	PUNCT	Period	_	6	punct	_	_


mwe in other languages: [bg] [cs] [de] [el] [en] [es] [eu] [fa] [fi] [fr] [ga] [he] [hu] [it] [ja] [ko] [sv] [u]