id	sid	tid	token	lemma	pos
work_v44vkh6bzramngvcrficy2no44	1	1	Dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	1	2	Shaping	shape	VBG
work_v44vkh6bzramngvcrficy2no44	1	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	1	4	Reordering	Reordering	NNP
work_v44vkh6bzramngvcrficy2no44	1	5	Search	Search	NNP
work_v44vkh6bzramngvcrficy2no44	1	6	Space	Space	NNP
work_v44vkh6bzramngvcrficy2no44	1	7	of	of	IN
work_v44vkh6bzramngvcrficy2no44	1	8	Phrase	Phrase	NNP
work_v44vkh6bzramngvcrficy2no44	1	9	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	1	10	Based	Based	NNP
work_v44vkh6bzramngvcrficy2no44	1	11	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	1	12	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	1	13	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	1	14	Arianna	Arianna	NNP
work_v44vkh6bzramngvcrficy2no44	1	15	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	1	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	1	17	Marcello	Marcello	NNP
work_v44vkh6bzramngvcrficy2no44	1	18	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	1	19	Fondazione	Fondazione	NNP
work_v44vkh6bzramngvcrficy2no44	1	20	Bruno	Bruno	NNP
work_v44vkh6bzramngvcrficy2no44	1	21	Kessler	Kessler	NNP
work_v44vkh6bzramngvcrficy2no44	1	22	Trento	Trento	NNP
work_v44vkh6bzramngvcrficy2no44	1	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	1	24	Italy	Italy	NNP
work_v44vkh6bzramngvcrficy2no44	1	25	{	{	-LRB-
work_v44vkh6bzramngvcrficy2no44	1	26	bisazza,federico}@fbk.eu	bisazza,federico}@fbk.eu	CD
work_v44vkh6bzramngvcrficy2no44	1	27	Abstract	Abstract	NNP
work_v44vkh6bzramngvcrficy2no44	1	28	Defining	define	VBG
work_v44vkh6bzramngvcrficy2no44	1	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	1	30	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	1	31	search	search	NN
work_v44vkh6bzramngvcrficy2no44	1	32	space	space	NN
work_v44vkh6bzramngvcrficy2no44	1	33	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	1	34	a	a	DT
work_v44vkh6bzramngvcrficy2no44	1	35	cru-	cru-	RB
work_v44vkh6bzramngvcrficy2no44	1	36	cial	cial	JJ
work_v44vkh6bzramngvcrficy2no44	1	37	issue	issue	NN
work_v44vkh6bzramngvcrficy2no44	1	38	in	in	IN
work_v44vkh6bzramngvcrficy2no44	1	39	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	1	40	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	1	41	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	1	42	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	1	43	between	between	IN
work_v44vkh6bzramngvcrficy2no44	1	44	dis-	dis-	IN
work_v44vkh6bzramngvcrficy2no44	1	45	tant	tant	JJ
work_v44vkh6bzramngvcrficy2no44	1	46	languages	language	NNS
work_v44vkh6bzramngvcrficy2no44	1	47	.	.	.
work_v44vkh6bzramngvcrficy2no44	2	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	2	2	fact	fact	NN
work_v44vkh6bzramngvcrficy2no44	2	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	2	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	2	5	optimal	optimal	NN
work_v44vkh6bzramngvcrficy2no44	2	6	trade-	trade-	VBP
work_v44vkh6bzramngvcrficy2no44	2	7	off	off	RB
work_v44vkh6bzramngvcrficy2no44	2	8	between	between	IN
work_v44vkh6bzramngvcrficy2no44	2	9	accuracy	accuracy	NN
work_v44vkh6bzramngvcrficy2no44	2	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	2	11	complexity	complexity	NN
work_v44vkh6bzramngvcrficy2no44	2	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	2	13	de-	de-	RB
work_v44vkh6bzramngvcrficy2no44	2	14	coding	coding	JJ
work_v44vkh6bzramngvcrficy2no44	2	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	2	16	nowadays	nowadays	RB
work_v44vkh6bzramngvcrficy2no44	2	17	reached	reach	VBN
work_v44vkh6bzramngvcrficy2no44	2	18	by	by	IN
work_v44vkh6bzramngvcrficy2no44	2	19	harshly	harshly	RB
work_v44vkh6bzramngvcrficy2no44	2	20	lim-	lim-	RB
work_v44vkh6bzramngvcrficy2no44	2	21	iting	ite	VBG
work_v44vkh6bzramngvcrficy2no44	2	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	2	23	input	input	NN
work_v44vkh6bzramngvcrficy2no44	2	24	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	2	25	space	space	NN
work_v44vkh6bzramngvcrficy2no44	2	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	3	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	3	2	pro-	pro-	RB
work_v44vkh6bzramngvcrficy2no44	3	3	pose	pose	VBP
work_v44vkh6bzramngvcrficy2no44	3	4	a	a	DT
work_v44vkh6bzramngvcrficy2no44	3	5	method	method	NN
work_v44vkh6bzramngvcrficy2no44	3	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	3	7	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	3	8	shape	shape	VB
work_v44vkh6bzramngvcrficy2no44	3	9	such	such	JJ
work_v44vkh6bzramngvcrficy2no44	3	10	space	space	NN
work_v44vkh6bzramngvcrficy2no44	3	11	and	and	CC
work_v44vkh6bzramngvcrficy2no44	3	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	3	13	thus	thus	RB
work_v44vkh6bzramngvcrficy2no44	3	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	3	15	capture	capture	NN
work_v44vkh6bzramngvcrficy2no44	3	16	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	3	17	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	3	18	range	range	NN
work_v44vkh6bzramngvcrficy2no44	3	19	word	word	NN
work_v44vkh6bzramngvcrficy2no44	3	20	movements	movement	NNS
work_v44vkh6bzramngvcrficy2no44	3	21	without	without	IN
work_v44vkh6bzramngvcrficy2no44	3	22	hurting	hurt	VBG
work_v44vkh6bzramngvcrficy2no44	3	23	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	3	24	qual-	qual-	NNP
work_v44vkh6bzramngvcrficy2no44	3	25	ity	ity	NN
work_v44vkh6bzramngvcrficy2no44	3	26	nor	nor	CC
work_v44vkh6bzramngvcrficy2no44	3	27	decoding	decoding	JJ
work_v44vkh6bzramngvcrficy2no44	3	28	time	time	NN
work_v44vkh6bzramngvcrficy2no44	3	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	4	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	4	2	space	space	NN
work_v44vkh6bzramngvcrficy2no44	4	3	defined	define	VBN
work_v44vkh6bzramngvcrficy2no44	4	4	by	by	IN
work_v44vkh6bzramngvcrficy2no44	4	5	loose	loose	JJ
work_v44vkh6bzramngvcrficy2no44	4	6	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	4	7	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	4	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	4	9	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	4	10	pruned	prune	VBN
work_v44vkh6bzramngvcrficy2no44	4	11	through	through	IN
work_v44vkh6bzramngvcrficy2no44	4	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	4	13	binary	binary	JJ
work_v44vkh6bzramngvcrficy2no44	4	14	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	4	15	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	4	16	predicts	predict	VBZ
work_v44vkh6bzramngvcrficy2no44	4	17	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	4	18	a	a	DT
work_v44vkh6bzramngvcrficy2no44	4	19	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	4	20	input	input	NN
work_v44vkh6bzramngvcrficy2no44	4	21	word	word	NN
work_v44vkh6bzramngvcrficy2no44	4	22	should	should	MD
work_v44vkh6bzramngvcrficy2no44	4	23	be	be	VB
work_v44vkh6bzramngvcrficy2no44	4	24	trans-	trans-	JJ
work_v44vkh6bzramngvcrficy2no44	4	25	lated	late	VBN
work_v44vkh6bzramngvcrficy2no44	4	26	right	right	RB
work_v44vkh6bzramngvcrficy2no44	4	27	after	after	IN
work_v44vkh6bzramngvcrficy2no44	4	28	another	another	DT
work_v44vkh6bzramngvcrficy2no44	4	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	5	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	5	2	integration	integration	NN
work_v44vkh6bzramngvcrficy2no44	5	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	5	4	this	this	DT
work_v44vkh6bzramngvcrficy2no44	5	5	model	model	NN
work_v44vkh6bzramngvcrficy2no44	5	6	into	into	IN
work_v44vkh6bzramngvcrficy2no44	5	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	5	8	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	5	9	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	5	10	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	5	11	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	5	12	im-	im-	RB
work_v44vkh6bzramngvcrficy2no44	5	13	proves	prove	VBZ
work_v44vkh6bzramngvcrficy2no44	5	14	a	a	DT
work_v44vkh6bzramngvcrficy2no44	5	15	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	5	16	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	5	17	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	5	18	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	5	19	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	5	20	al-	al-	RB
work_v44vkh6bzramngvcrficy2no44	5	21	ready	ready	JJ
work_v44vkh6bzramngvcrficy2no44	5	22	including	include	VBG
work_v44vkh6bzramngvcrficy2no44	5	23	state	state	NN
work_v44vkh6bzramngvcrficy2no44	5	24	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	5	25	of	of	IN
work_v44vkh6bzramngvcrficy2no44	5	26	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	5	27	the	the	DT
work_v44vkh6bzramngvcrficy2no44	5	28	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	5	29	art	art	NN
work_v44vkh6bzramngvcrficy2no44	5	30	early	early	RB
work_v44vkh6bzramngvcrficy2no44	5	31	distor-	distor-	XX
work_v44vkh6bzramngvcrficy2no44	5	32	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	5	33	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	5	34	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	5	35	Moore	Moore	NNP
work_v44vkh6bzramngvcrficy2no44	5	36	and	and	CC
work_v44vkh6bzramngvcrficy2no44	5	37	Quirk	Quirk	NNP
work_v44vkh6bzramngvcrficy2no44	5	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	5	39	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	5	40	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	5	41	and	and	CC
work_v44vkh6bzramngvcrficy2no44	5	42	hierar-	hierar-	NNP
work_v44vkh6bzramngvcrficy2no44	5	43	chical	chical	NNP
work_v44vkh6bzramngvcrficy2no44	5	44	phrase	phrase	NNP
work_v44vkh6bzramngvcrficy2no44	5	45	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	5	46	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	5	47	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	5	48	Galley	Galley	NNP
work_v44vkh6bzramngvcrficy2no44	5	49	and	and	CC
work_v44vkh6bzramngvcrficy2no44	5	50	Manning	Manning	NNP
work_v44vkh6bzramngvcrficy2no44	5	51	,	,	,
work_v44vkh6bzramngvcrficy2no44	5	52	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	5	53	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	5	54	.	.	.
work_v44vkh6bzramngvcrficy2no44	6	1	Significant	significant	JJ
work_v44vkh6bzramngvcrficy2no44	6	2	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	6	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	6	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	6	5	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	6	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	6	7	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	6	8	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	6	9	achieved	achieve	VBN
work_v44vkh6bzramngvcrficy2no44	6	10	by	by	IN
work_v44vkh6bzramngvcrficy2no44	6	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	6	12	sys-	sys-	NN
work_v44vkh6bzramngvcrficy2no44	6	13	tem	tem	NN
work_v44vkh6bzramngvcrficy2no44	6	14	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	6	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	6	16	notably	notably	RB
work_v44vkh6bzramngvcrficy2no44	6	17	faster	fast	JJR
work_v44vkh6bzramngvcrficy2no44	6	18	than	than	IN
work_v44vkh6bzramngvcrficy2no44	6	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	6	20	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	6	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	6	22	while	while	IN
work_v44vkh6bzramngvcrficy2no44	6	23	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	6	24	and	and	CC
work_v44vkh6bzramngvcrficy2no44	6	25	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	6	26	remain	remain	VBP
work_v44vkh6bzramngvcrficy2no44	6	27	stable	stable	JJ
work_v44vkh6bzramngvcrficy2no44	6	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	6	29	or	or	CC
work_v44vkh6bzramngvcrficy2no44	6	30	even	even	RB
work_v44vkh6bzramngvcrficy2no44	6	31	increase	increase	VB
work_v44vkh6bzramngvcrficy2no44	6	32	,	,	,
work_v44vkh6bzramngvcrficy2no44	6	33	at	at	IN
work_v44vkh6bzramngvcrficy2no44	6	34	a	a	DT
work_v44vkh6bzramngvcrficy2no44	6	35	very	very	RB
work_v44vkh6bzramngvcrficy2no44	6	36	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	6	37	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	6	38	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	6	39	.	.	.
work_v44vkh6bzramngvcrficy2no44	7	1	1	1	CD
work_v44vkh6bzramngvcrficy2no44	7	2	Introduction	introduction	NN
work_v44vkh6bzramngvcrficy2no44	7	3	Word	word	NN
work_v44vkh6bzramngvcrficy2no44	7	4	order	order	NN
work_v44vkh6bzramngvcrficy2no44	7	5	differences	difference	NNS
work_v44vkh6bzramngvcrficy2no44	7	6	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	7	7	among	among	IN
work_v44vkh6bzramngvcrficy2no44	7	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	7	9	most	most	RBS
work_v44vkh6bzramngvcrficy2no44	7	10	impor-	impor-	JJ
work_v44vkh6bzramngvcrficy2no44	7	11	tant	tant	JJ
work_v44vkh6bzramngvcrficy2no44	7	12	factors	factor	NNS
work_v44vkh6bzramngvcrficy2no44	7	13	determining	determine	VBG
work_v44vkh6bzramngvcrficy2no44	7	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	7	15	performance	performance	NN
work_v44vkh6bzramngvcrficy2no44	7	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	7	17	statisti-	statisti-	NNP
work_v44vkh6bzramngvcrficy2no44	7	18	cal	cal	NNP
work_v44vkh6bzramngvcrficy2no44	7	19	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	7	20	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	7	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	7	22	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	7	23	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	7	24	on	on	IN
work_v44vkh6bzramngvcrficy2no44	7	25	a	a	DT
work_v44vkh6bzramngvcrficy2no44	7	26	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	7	27	language	language	NN
work_v44vkh6bzramngvcrficy2no44	7	28	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	7	29	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	7	30	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	7	31	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	7	32	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	7	33	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	7	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	7	35	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	7	36	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	7	37	.	.	.
work_v44vkh6bzramngvcrficy2no44	8	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	8	2	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	8	3	particularly	particularly	RB
work_v44vkh6bzramngvcrficy2no44	8	4	true	true	JJ
work_v44vkh6bzramngvcrficy2no44	8	5	in	in	IN
work_v44vkh6bzramngvcrficy2no44	8	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	8	7	framework	framework	NN
work_v44vkh6bzramngvcrficy2no44	8	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	8	9	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	8	10	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	8	11	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	8	12	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	8	13	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	8	14	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	8	15	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	8	16	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	8	17	Zens	Zens	NNP
work_v44vkh6bzramngvcrficy2no44	8	18	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	8	19	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	8	20	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	8	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	8	22	2002	2002	CD
work_v44vkh6bzramngvcrficy2no44	8	23	;	;	:
work_v44vkh6bzramngvcrficy2no44	8	24	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	8	25	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	8	26	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	8	27	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	8	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	8	29	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	8	30	;	;	:
work_v44vkh6bzramngvcrficy2no44	8	31	Och	Och	NNP
work_v44vkh6bzramngvcrficy2no44	8	32	and	and	CC
work_v44vkh6bzramngvcrficy2no44	8	33	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	8	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	8	35	2002	2002	CD
work_v44vkh6bzramngvcrficy2no44	8	36	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	8	37	,	,	,
work_v44vkh6bzramngvcrficy2no44	8	38	an	an	DT
work_v44vkh6bzramngvcrficy2no44	8	39	approach	approach	NN
work_v44vkh6bzramngvcrficy2no44	8	40	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	8	41	remains	remain	VBZ
work_v44vkh6bzramngvcrficy2no44	8	42	highly	highly	RB
work_v44vkh6bzramngvcrficy2no44	8	43	competitive	competitive	JJ
work_v44vkh6bzramngvcrficy2no44	8	44	despite	despite	IN
work_v44vkh6bzramngvcrficy2no44	8	45	the	the	DT
work_v44vkh6bzramngvcrficy2no44	8	46	recent	recent	JJ
work_v44vkh6bzramngvcrficy2no44	8	47	advances	advance	NNS
work_v44vkh6bzramngvcrficy2no44	8	48	of	of	IN
work_v44vkh6bzramngvcrficy2no44	8	49	the	the	DT
work_v44vkh6bzramngvcrficy2no44	8	50	tree	tree	NN
work_v44vkh6bzramngvcrficy2no44	8	51	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	8	52	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	8	53	approaches	approach	NNS
work_v44vkh6bzramngvcrficy2no44	8	54	.	.	.
work_v44vkh6bzramngvcrficy2no44	9	1	During	during	IN
work_v44vkh6bzramngvcrficy2no44	9	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	9	3	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	9	4	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	9	5	process	process	NN
work_v44vkh6bzramngvcrficy2no44	9	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	9	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	9	8	output	output	NN
work_v44vkh6bzramngvcrficy2no44	9	9	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	9	10	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	9	11	built	build	VBN
work_v44vkh6bzramngvcrficy2no44	9	12	from	from	IN
work_v44vkh6bzramngvcrficy2no44	9	13	left	leave	VBN
work_v44vkh6bzramngvcrficy2no44	9	14	to	to	IN
work_v44vkh6bzramngvcrficy2no44	9	15	right	right	NN
work_v44vkh6bzramngvcrficy2no44	9	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	9	17	while	while	IN
work_v44vkh6bzramngvcrficy2no44	9	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	9	19	input	input	NN
work_v44vkh6bzramngvcrficy2no44	9	20	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	9	21	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	9	22	can	can	MD
work_v44vkh6bzramngvcrficy2no44	9	23	be	be	VB
work_v44vkh6bzramngvcrficy2no44	9	24	covered	cover	VBN
work_v44vkh6bzramngvcrficy2no44	9	25	in	in	IN
work_v44vkh6bzramngvcrficy2no44	9	26	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	9	27	or-	or-	JJ
work_v44vkh6bzramngvcrficy2no44	9	28	ders	der	NNS
work_v44vkh6bzramngvcrficy2no44	9	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	10	1	Thus	thus	RB
work_v44vkh6bzramngvcrficy2no44	10	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	10	3	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	10	4	in	in	IN
work_v44vkh6bzramngvcrficy2no44	10	5	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	10	6	can	can	MD
work_v44vkh6bzramngvcrficy2no44	10	7	be	be	VB
work_v44vkh6bzramngvcrficy2no44	10	8	viewed	view	VBN
work_v44vkh6bzramngvcrficy2no44	10	9	as	as	IN
work_v44vkh6bzramngvcrficy2no44	10	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	10	11	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	10	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	10	13	choosing	choose	VBG
work_v44vkh6bzramngvcrficy2no44	10	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	10	15	input	input	NN
work_v44vkh6bzramngvcrficy2no44	10	16	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	10	17	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	10	18	leads	lead	VBZ
work_v44vkh6bzramngvcrficy2no44	10	19	to	to	IN
work_v44vkh6bzramngvcrficy2no44	10	20	the	the	DT
work_v44vkh6bzramngvcrficy2no44	10	21	highest	highest	RBS
work_v44vkh6bzramngvcrficy2no44	10	22	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	10	23	scoring	score	VBG
work_v44vkh6bzramngvcrficy2no44	10	24	output	output	NN
work_v44vkh6bzramngvcrficy2no44	10	25	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	10	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	11	1	Due	due	IN
work_v44vkh6bzramngvcrficy2no44	11	2	to	to	IN
work_v44vkh6bzramngvcrficy2no44	11	3	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	11	4	reasons	reason	NNS
work_v44vkh6bzramngvcrficy2no44	11	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	11	6	however	however	RB
work_v44vkh6bzramngvcrficy2no44	11	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	11	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	11	9	input	input	NN
work_v44vkh6bzramngvcrficy2no44	11	10	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	11	11	space	space	NN
work_v44vkh6bzramngvcrficy2no44	11	12	can	can	MD
work_v44vkh6bzramngvcrficy2no44	11	13	not	not	RB
work_v44vkh6bzramngvcrficy2no44	11	14	be	be	VB
work_v44vkh6bzramngvcrficy2no44	11	15	fully	fully	RB
work_v44vkh6bzramngvcrficy2no44	11	16	explored	explore	VBN
work_v44vkh6bzramngvcrficy2no44	11	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	11	18	and	and	CC
work_v44vkh6bzramngvcrficy2no44	11	19	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	11	20	therefore	therefore	RB
work_v44vkh6bzramngvcrficy2no44	11	21	lim-	lim-	RB
work_v44vkh6bzramngvcrficy2no44	11	22	ited	ite	VBN
work_v44vkh6bzramngvcrficy2no44	11	23	with	with	IN
work_v44vkh6bzramngvcrficy2no44	11	24	hard	hard	JJ
work_v44vkh6bzramngvcrficy2no44	11	25	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	11	26	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	11	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	12	1	Although	although	IN
work_v44vkh6bzramngvcrficy2no44	12	2	many	many	JJ
work_v44vkh6bzramngvcrficy2no44	12	3	solutions	solution	NNS
work_v44vkh6bzramngvcrficy2no44	12	4	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	12	5	been	be	VBN
work_v44vkh6bzramngvcrficy2no44	12	6	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	12	7	to	to	TO
work_v44vkh6bzramngvcrficy2no44	12	8	explicitly	explicitly	RB
work_v44vkh6bzramngvcrficy2no44	12	9	model	model	VB
work_v44vkh6bzramngvcrficy2no44	12	10	word	word	NN
work_v44vkh6bzramngvcrficy2no44	12	11	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	12	12	during	during	IN
work_v44vkh6bzramngvcrficy2no44	12	13	decoding	decoding	NN
work_v44vkh6bzramngvcrficy2no44	12	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	12	15	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	12	16	still	still	RB
work_v44vkh6bzramngvcrficy2no44	12	17	largely	largely	RB
work_v44vkh6bzramngvcrficy2no44	12	18	fails	fail	VBZ
work_v44vkh6bzramngvcrficy2no44	12	19	to	to	TO
work_v44vkh6bzramngvcrficy2no44	12	20	handle	handle	VB
work_v44vkh6bzramngvcrficy2no44	12	21	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	12	22	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	12	23	range	range	NN
work_v44vkh6bzramngvcrficy2no44	12	24	word	word	NN
work_v44vkh6bzramngvcrficy2no44	12	25	movements	movement	NNS
work_v44vkh6bzramngvcrficy2no44	12	26	in	in	IN
work_v44vkh6bzramngvcrficy2no44	12	27	language	language	NN
work_v44vkh6bzramngvcrficy2no44	12	28	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	12	29	with	with	IN
work_v44vkh6bzramngvcrficy2no44	12	30	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	12	31	syntac-	syntac-	NN
work_v44vkh6bzramngvcrficy2no44	12	32	tic	tic	JJ
work_v44vkh6bzramngvcrficy2no44	12	33	structures1	structures1	NN
work_v44vkh6bzramngvcrficy2no44	12	34	.	.	.
work_v44vkh6bzramngvcrficy2no44	13	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	13	2	believe	believe	VBP
work_v44vkh6bzramngvcrficy2no44	13	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	13	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	13	5	mostly	mostly	RB
work_v44vkh6bzramngvcrficy2no44	13	6	not	not	RB
work_v44vkh6bzramngvcrficy2no44	13	7	due	due	JJ
work_v44vkh6bzramngvcrficy2no44	13	8	to	to	IN
work_v44vkh6bzramngvcrficy2no44	13	9	deficiencies	deficiency	NNS
work_v44vkh6bzramngvcrficy2no44	13	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	13	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	13	12	existing	exist	VBG
work_v44vkh6bzramngvcrficy2no44	13	13	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	13	14	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	13	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	13	16	but	but	CC
work_v44vkh6bzramngvcrficy2no44	13	17	rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	13	18	to	to	IN
work_v44vkh6bzramngvcrficy2no44	13	19	a	a	DT
work_v44vkh6bzramngvcrficy2no44	13	20	very	very	RB
work_v44vkh6bzramngvcrficy2no44	13	21	coarse	coarse	JJ
work_v44vkh6bzramngvcrficy2no44	13	22	definition	definition	NN
work_v44vkh6bzramngvcrficy2no44	13	23	of	of	IN
work_v44vkh6bzramngvcrficy2no44	13	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	13	25	reorder-	reorder-	JJ
work_v44vkh6bzramngvcrficy2no44	13	26	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	13	27	search	search	NN
work_v44vkh6bzramngvcrficy2no44	13	28	space	space	NN
work_v44vkh6bzramngvcrficy2no44	13	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	14	1	Indeed	indeed	RB
work_v44vkh6bzramngvcrficy2no44	14	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	14	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	14	4	existing	exist	VBG
work_v44vkh6bzramngvcrficy2no44	14	5	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	14	6	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	14	7	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	14	8	rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	14	9	simple	simple	JJ
work_v44vkh6bzramngvcrficy2no44	14	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	14	11	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	14	12	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	14	13	on	on	IN
work_v44vkh6bzramngvcrficy2no44	14	14	word	word	NN
work_v44vkh6bzramngvcrficy2no44	14	15	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	14	16	to	to	IN
work_v44vkh6bzramngvcrficy2no44	14	17	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	14	18	word	word	NN
work_v44vkh6bzramngvcrficy2no44	14	19	distances	distance	NNS
work_v44vkh6bzramngvcrficy2no44	14	20	.	.	.
work_v44vkh6bzramngvcrficy2no44	15	1	Moreover	moreover	RB
work_v44vkh6bzramngvcrficy2no44	15	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	15	3	they	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	15	4	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	15	5	uni-	uni-	RB
work_v44vkh6bzramngvcrficy2no44	15	6	form	form	NN
work_v44vkh6bzramngvcrficy2no44	15	7	throughout	throughout	IN
work_v44vkh6bzramngvcrficy2no44	15	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	15	9	input	input	NN
work_v44vkh6bzramngvcrficy2no44	15	10	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	15	11	and	and	CC
work_v44vkh6bzramngvcrficy2no44	15	12	insensitive	insensitive	JJ
work_v44vkh6bzramngvcrficy2no44	15	13	to	to	IN
work_v44vkh6bzramngvcrficy2no44	15	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	15	15	actual	actual	JJ
work_v44vkh6bzramngvcrficy2no44	15	16	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	15	17	being	be	VBG
work_v44vkh6bzramngvcrficy2no44	15	18	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	15	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	16	1	Relaxing	relax	VBG
work_v44vkh6bzramngvcrficy2no44	16	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	16	3	kind	kind	NN
work_v44vkh6bzramngvcrficy2no44	16	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	16	5	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	16	6	means	mean	VBZ
work_v44vkh6bzramngvcrficy2no44	16	7	dramatically	dramatically	RB
work_v44vkh6bzramngvcrficy2no44	16	8	increasing	increase	VBG
work_v44vkh6bzramngvcrficy2no44	16	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	16	10	size	size	NN
work_v44vkh6bzramngvcrficy2no44	16	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	16	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	16	13	search	search	NN
work_v44vkh6bzramngvcrficy2no44	16	14	space	space	NN
work_v44vkh6bzramngvcrficy2no44	16	15	and	and	CC
work_v44vkh6bzramngvcrficy2no44	16	16	making	make	VBG
work_v44vkh6bzramngvcrficy2no44	16	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	16	18	reorder-	reorder-	JJ
work_v44vkh6bzramngvcrficy2no44	16	19	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	16	20	model	model	NN
work_v44vkh6bzramngvcrficy2no44	16	21	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	16	22	task	task	NN
work_v44vkh6bzramngvcrficy2no44	16	23	extremely	extremely	RB
work_v44vkh6bzramngvcrficy2no44	16	24	complex	complex	JJ
work_v44vkh6bzramngvcrficy2no44	16	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	17	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	17	2	a	a	DT
work_v44vkh6bzramngvcrficy2no44	17	3	result	result	NN
work_v44vkh6bzramngvcrficy2no44	17	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	17	5	even	even	RB
work_v44vkh6bzramngvcrficy2no44	17	6	in	in	IN
work_v44vkh6bzramngvcrficy2no44	17	7	language	language	NN
work_v44vkh6bzramngvcrficy2no44	17	8	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	17	9	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	17	10	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	17	11	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	17	12	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	17	13	reg-	reg-	RB
work_v44vkh6bzramngvcrficy2no44	17	14	ularly	ularly	RB
work_v44vkh6bzramngvcrficy2no44	17	15	observed	observe	VBN
work_v44vkh6bzramngvcrficy2no44	17	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	17	17	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	17	18	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	17	19	degrades	degrade	NNS
work_v44vkh6bzramngvcrficy2no44	17	20	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	17	21	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	17	22	word	word	NN
work_v44vkh6bzramngvcrficy2no44	17	23	movements	movement	NNS
work_v44vkh6bzramngvcrficy2no44	17	24	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	17	25	allowed	allow	VBN
work_v44vkh6bzramngvcrficy2no44	17	26	to	to	IN
work_v44vkh6bzramngvcrficy2no44	17	27	the	the	DT
work_v44vkh6bzramngvcrficy2no44	17	28	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	17	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	18	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	18	2	address	address	VBP
work_v44vkh6bzramngvcrficy2no44	18	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	18	4	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	18	5	by	by	IN
work_v44vkh6bzramngvcrficy2no44	18	6	training	train	VBG
work_v44vkh6bzramngvcrficy2no44	18	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	18	8	binary	binary	JJ
work_v44vkh6bzramngvcrficy2no44	18	9	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	18	10	to	to	TO
work_v44vkh6bzramngvcrficy2no44	18	11	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	18	12	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	18	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	18	14	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	18	15	input	input	NN
work_v44vkh6bzramngvcrficy2no44	18	16	position	position	NN
work_v44vkh6bzramngvcrficy2no44	18	17	should	should	MD
work_v44vkh6bzramngvcrficy2no44	18	18	be	be	VB
work_v44vkh6bzramngvcrficy2no44	18	19	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	18	20	right	right	RB
work_v44vkh6bzramngvcrficy2no44	18	21	after	after	IN
work_v44vkh6bzramngvcrficy2no44	18	22	another	another	DT
work_v44vkh6bzramngvcrficy2no44	18	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	18	24	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	18	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	18	26	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	18	27	at	at	IN
work_v44vkh6bzramngvcrficy2no44	18	28	those	those	DT
work_v44vkh6bzramngvcrficy2no44	18	29	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	18	30	and	and	CC
work_v44vkh6bzramngvcrficy2no44	18	31	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	18	32	contexts	contexts	NN
work_v44vkh6bzramngvcrficy2no44	18	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	19	1	When	when	WRB
work_v44vkh6bzramngvcrficy2no44	19	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	19	3	model	model	NN
work_v44vkh6bzramngvcrficy2no44	19	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	19	5	integrated	integrate	VBN
work_v44vkh6bzramngvcrficy2no44	19	6	into	into	IN
work_v44vkh6bzramngvcrficy2no44	19	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	19	8	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	19	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	19	10	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	19	11	predic-	predic-	JJ
work_v44vkh6bzramngvcrficy2no44	19	12	1For	1for	JJ
work_v44vkh6bzramngvcrficy2no44	19	13	empirical	empirical	JJ
work_v44vkh6bzramngvcrficy2no44	19	14	evidence	evidence	NN
work_v44vkh6bzramngvcrficy2no44	19	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	19	16	see	see	VB
work_v44vkh6bzramngvcrficy2no44	19	17	for	for	IN
work_v44vkh6bzramngvcrficy2no44	19	18	instance	instance	NN
work_v44vkh6bzramngvcrficy2no44	19	19	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	19	20	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	19	21	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	19	22	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	19	23	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	19	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	19	25	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	19	26	;	;	:
work_v44vkh6bzramngvcrficy2no44	19	27	Galley	Galley	NNP
work_v44vkh6bzramngvcrficy2no44	19	28	and	and	CC
work_v44vkh6bzramngvcrficy2no44	19	29	Manning	Manning	NNP
work_v44vkh6bzramngvcrficy2no44	19	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	19	31	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	19	32	;	;	:
work_v44vkh6bzramngvcrficy2no44	19	33	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	19	34	and	and	CC
work_v44vkh6bzramngvcrficy2no44	19	35	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	19	36	,	,	,
work_v44vkh6bzramngvcrficy2no44	19	37	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	19	38	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	19	39	.	.	.
work_v44vkh6bzramngvcrficy2no44	20	1	327	327	CD
work_v44vkh6bzramngvcrficy2no44	20	2	Transactions	transaction	NNS
work_v44vkh6bzramngvcrficy2no44	20	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	20	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	20	5	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	20	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	20	7	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	20	8	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	20	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	20	10	1	1	CD
work_v44vkh6bzramngvcrficy2no44	20	11	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	20	12	2013	2013	CD
work_v44vkh6bzramngvcrficy2no44	20	13	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	20	14	327–340	327–340	CD
work_v44vkh6bzramngvcrficy2no44	20	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	21	1	Action	Action	NNP
work_v44vkh6bzramngvcrficy2no44	21	2	Editor	Editor	NNP
work_v44vkh6bzramngvcrficy2no44	21	3	:	:	:
work_v44vkh6bzramngvcrficy2no44	21	4	Philipp	Philipp	NNP
work_v44vkh6bzramngvcrficy2no44	21	5	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	21	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	22	1	Submitted	submit	VBN
work_v44vkh6bzramngvcrficy2no44	22	2	1/2013	1/2013	CD
work_v44vkh6bzramngvcrficy2no44	22	3	;	;	:
work_v44vkh6bzramngvcrficy2no44	22	4	Revised	revise	VBN
work_v44vkh6bzramngvcrficy2no44	22	5	5/2013	5/2013	CD
work_v44vkh6bzramngvcrficy2no44	22	6	;	;	:
work_v44vkh6bzramngvcrficy2no44	22	7	Published	Published	NNP
work_v44vkh6bzramngvcrficy2no44	22	8	7/2013	7/2013	NNP
work_v44vkh6bzramngvcrficy2no44	22	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	23	1	c	c	NNP
work_v44vkh6bzramngvcrficy2no44	23	2	©	©	NNP
work_v44vkh6bzramngvcrficy2no44	23	3	2013	2013	CD
work_v44vkh6bzramngvcrficy2no44	23	4	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	23	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	23	6	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	23	7	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	23	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	24	1	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	24	2	can	can	MD
work_v44vkh6bzramngvcrficy2no44	24	3	be	be	VB
work_v44vkh6bzramngvcrficy2no44	24	4	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	24	5	not	not	RB
work_v44vkh6bzramngvcrficy2no44	24	6	only	only	RB
work_v44vkh6bzramngvcrficy2no44	24	7	as	as	IN
work_v44vkh6bzramngvcrficy2no44	24	8	an	an	DT
work_v44vkh6bzramngvcrficy2no44	24	9	additional	additional	JJ
work_v44vkh6bzramngvcrficy2no44	24	10	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	24	11	function	function	NN
work_v44vkh6bzramngvcrficy2no44	24	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	24	13	but	but	CC
work_v44vkh6bzramngvcrficy2no44	24	14	also	also	RB
work_v44vkh6bzramngvcrficy2no44	24	15	as	as	IN
work_v44vkh6bzramngvcrficy2no44	24	16	an	an	DT
work_v44vkh6bzramngvcrficy2no44	24	17	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	24	18	indication	indication	NN
work_v44vkh6bzramngvcrficy2no44	24	19	of	of	IN
work_v44vkh6bzramngvcrficy2no44	24	20	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	24	21	or	or	CC
work_v44vkh6bzramngvcrficy2no44	24	22	not	not	RB
work_v44vkh6bzramngvcrficy2no44	24	23	a	a	DT
work_v44vkh6bzramngvcrficy2no44	24	24	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	24	25	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	24	26	path	path	NN
work_v44vkh6bzramngvcrficy2no44	24	27	should	should	MD
work_v44vkh6bzramngvcrficy2no44	24	28	be	be	VB
work_v44vkh6bzramngvcrficy2no44	24	29	further	further	RB
work_v44vkh6bzramngvcrficy2no44	24	30	ex-	ex-	RB
work_v44vkh6bzramngvcrficy2no44	24	31	plored	plore	VBN
work_v44vkh6bzramngvcrficy2no44	24	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	25	1	More	more	RBR
work_v44vkh6bzramngvcrficy2no44	25	2	specifically	specifically	RB
work_v44vkh6bzramngvcrficy2no44	25	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	25	4	at	at	IN
work_v44vkh6bzramngvcrficy2no44	25	5	each	each	DT
work_v44vkh6bzramngvcrficy2no44	25	6	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	25	7	ex-	ex-	XX
work_v44vkh6bzramngvcrficy2no44	25	8	pansion	pansion	NN
work_v44vkh6bzramngvcrficy2no44	25	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	25	10	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	25	11	consider	consider	VBP
work_v44vkh6bzramngvcrficy2no44	25	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	25	13	set	set	NN
work_v44vkh6bzramngvcrficy2no44	25	14	of	of	IN
work_v44vkh6bzramngvcrficy2no44	25	15	input	input	NN
work_v44vkh6bzramngvcrficy2no44	25	16	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	25	17	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	25	18	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	25	19	reachable	reachable	JJ
work_v44vkh6bzramngvcrficy2no44	25	20	according	accord	VBG
work_v44vkh6bzramngvcrficy2no44	25	21	to	to	IN
work_v44vkh6bzramngvcrficy2no44	25	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	25	23	usual	usual	JJ
work_v44vkh6bzramngvcrficy2no44	25	24	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	25	25	con-	con-	NN
work_v44vkh6bzramngvcrficy2no44	25	26	straints	straint	NNS
work_v44vkh6bzramngvcrficy2no44	25	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	25	28	and	and	CC
work_v44vkh6bzramngvcrficy2no44	25	29	prune	prune	VB
work_v44vkh6bzramngvcrficy2no44	25	30	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	25	31	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	25	32	only	only	RB
work_v44vkh6bzramngvcrficy2no44	25	33	on	on	IN
work_v44vkh6bzramngvcrficy2no44	25	34	the	the	DT
work_v44vkh6bzramngvcrficy2no44	25	35	reorder-	reorder-	NNP
work_v44vkh6bzramngvcrficy2no44	25	36	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	25	37	model	model	NN
work_v44vkh6bzramngvcrficy2no44	25	38	score	score	NN
work_v44vkh6bzramngvcrficy2no44	25	39	.	.	.
work_v44vkh6bzramngvcrficy2no44	26	1	Then	then	RB
work_v44vkh6bzramngvcrficy2no44	26	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	26	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	26	4	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	26	5	can	can	MD
work_v44vkh6bzramngvcrficy2no44	26	6	be	be	VB
work_v44vkh6bzramngvcrficy2no44	26	7	ex-	ex-	RB
work_v44vkh6bzramngvcrficy2no44	26	8	panded	pande	VBN
work_v44vkh6bzramngvcrficy2no44	26	9	normally	normally	RB
work_v44vkh6bzramngvcrficy2no44	26	10	by	by	IN
work_v44vkh6bzramngvcrficy2no44	26	11	covering	cover	VBG
work_v44vkh6bzramngvcrficy2no44	26	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	26	13	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	26	14	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	26	15	pruned	pruned	JJ
work_v44vkh6bzramngvcrficy2no44	26	16	posi-	posi-	NN
work_v44vkh6bzramngvcrficy2no44	26	17	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	26	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	27	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	27	2	technique	technique	NN
work_v44vkh6bzramngvcrficy2no44	27	3	makes	make	VBZ
work_v44vkh6bzramngvcrficy2no44	27	4	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	27	5	possible	possible	JJ
work_v44vkh6bzramngvcrficy2no44	27	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	27	7	dynami-	dynami-	VB
work_v44vkh6bzramngvcrficy2no44	27	8	cally	cally	RB
work_v44vkh6bzramngvcrficy2no44	27	9	shape	shape	VB
work_v44vkh6bzramngvcrficy2no44	27	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	27	11	search	search	NN
work_v44vkh6bzramngvcrficy2no44	27	12	space	space	NN
work_v44vkh6bzramngvcrficy2no44	27	13	while	while	IN
work_v44vkh6bzramngvcrficy2no44	27	14	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	27	15	with	with	IN
work_v44vkh6bzramngvcrficy2no44	27	16	a	a	DT
work_v44vkh6bzramngvcrficy2no44	27	17	very	very	RB
work_v44vkh6bzramngvcrficy2no44	27	18	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	27	19	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	27	20	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	27	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	27	22	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	27	23	can	can	MD
work_v44vkh6bzramngvcrficy2no44	27	24	improve	improve	VB
work_v44vkh6bzramngvcrficy2no44	27	25	trans-	trans-	JJ
work_v44vkh6bzramngvcrficy2no44	27	26	lation	lation	NN
work_v44vkh6bzramngvcrficy2no44	27	27	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	27	28	and	and	CC
work_v44vkh6bzramngvcrficy2no44	27	29	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	27	30	at	at	IN
work_v44vkh6bzramngvcrficy2no44	27	31	the	the	DT
work_v44vkh6bzramngvcrficy2no44	27	32	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	27	33	time	time	NN
work_v44vkh6bzramngvcrficy2no44	27	34	.	.	.
work_v44vkh6bzramngvcrficy2no44	28	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	28	2	remainder	remainder	NN
work_v44vkh6bzramngvcrficy2no44	28	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	28	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	28	5	paper	paper	NN
work_v44vkh6bzramngvcrficy2no44	28	6	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	28	7	organized	organize	VBN
work_v44vkh6bzramngvcrficy2no44	28	8	as	as	IN
work_v44vkh6bzramngvcrficy2no44	28	9	fol-	fol-	NN
work_v44vkh6bzramngvcrficy2no44	28	10	lows	low	NNS
work_v44vkh6bzramngvcrficy2no44	28	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	29	1	After	after	IN
work_v44vkh6bzramngvcrficy2no44	29	2	an	an	DT
work_v44vkh6bzramngvcrficy2no44	29	3	overview	overview	NN
work_v44vkh6bzramngvcrficy2no44	29	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	29	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	29	6	relevant	relevant	JJ
work_v44vkh6bzramngvcrficy2no44	29	7	literature	literature	NN
work_v44vkh6bzramngvcrficy2no44	29	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	29	9	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	29	10	describe	describe	VBP
work_v44vkh6bzramngvcrficy2no44	29	11	in	in	IN
work_v44vkh6bzramngvcrficy2no44	29	12	detail	detail	NN
work_v44vkh6bzramngvcrficy2no44	29	13	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	29	14	word	word	NN
work_v44vkh6bzramngvcrficy2no44	29	15	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	29	16	model	model	NN
work_v44vkh6bzramngvcrficy2no44	29	17	.	.	.
work_v44vkh6bzramngvcrficy2no44	30	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	30	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	30	3	following	following	JJ
work_v44vkh6bzramngvcrficy2no44	30	4	section	section	NN
work_v44vkh6bzramngvcrficy2no44	30	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	30	6	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	30	7	introduce	introduce	VBP
work_v44vkh6bzramngvcrficy2no44	30	8	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	30	9	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	30	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	30	11	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	30	12	steps	step	NNS
work_v44vkh6bzramngvcrficy2no44	30	13	as	as	IN
work_v44vkh6bzramngvcrficy2no44	30	14	a	a	DT
work_v44vkh6bzramngvcrficy2no44	30	15	way	way	NN
work_v44vkh6bzramngvcrficy2no44	30	16	to	to	TO
work_v44vkh6bzramngvcrficy2no44	30	17	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	30	18	shape	shape	VB
work_v44vkh6bzramngvcrficy2no44	30	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	30	20	input	input	NN
work_v44vkh6bzramngvcrficy2no44	30	21	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	30	22	space	space	NN
work_v44vkh6bzramngvcrficy2no44	30	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	31	1	Finally	finally	RB
work_v44vkh6bzramngvcrficy2no44	31	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	31	3	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	31	4	present	present	VBP
work_v44vkh6bzramngvcrficy2no44	31	5	an	an	DT
work_v44vkh6bzramngvcrficy2no44	31	6	em-	em-	NN
work_v44vkh6bzramngvcrficy2no44	31	7	pirical	pirical	JJ
work_v44vkh6bzramngvcrficy2no44	31	8	analysis	analysis	NN
work_v44vkh6bzramngvcrficy2no44	31	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	31	10	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	31	11	approach	approach	NN
work_v44vkh6bzramngvcrficy2no44	31	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	31	13	including	include	VBG
work_v44vkh6bzramngvcrficy2no44	31	14	intrinsic	intrinsic	JJ
work_v44vkh6bzramngvcrficy2no44	31	15	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	31	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	31	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	31	18	model	model	NN
work_v44vkh6bzramngvcrficy2no44	31	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	31	20	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	31	21	experiments	experiment	NNS
work_v44vkh6bzramngvcrficy2no44	31	22	on	on	IN
work_v44vkh6bzramngvcrficy2no44	31	23	a	a	DT
work_v44vkh6bzramngvcrficy2no44	31	24	well	well	RB
work_v44vkh6bzramngvcrficy2no44	31	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	31	26	known	know	VBN
work_v44vkh6bzramngvcrficy2no44	31	27	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	31	28	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	31	29	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	31	30	news	news	NN
work_v44vkh6bzramngvcrficy2no44	31	31	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	31	32	task	task	NN
work_v44vkh6bzramngvcrficy2no44	31	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	32	1	2	2	CD
work_v44vkh6bzramngvcrficy2no44	32	2	Previous	previous	JJ
work_v44vkh6bzramngvcrficy2no44	32	3	Work	work	NN
work_v44vkh6bzramngvcrficy2no44	32	4	In	in	IN
work_v44vkh6bzramngvcrficy2no44	32	5	this	this	DT
work_v44vkh6bzramngvcrficy2no44	32	6	paper	paper	NN
work_v44vkh6bzramngvcrficy2no44	32	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	32	8	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	32	9	focus	focus	VBP
work_v44vkh6bzramngvcrficy2no44	32	10	on	on	IN
work_v44vkh6bzramngvcrficy2no44	32	11	methods	method	NNS
work_v44vkh6bzramngvcrficy2no44	32	12	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	32	13	guide	guide	VBP
work_v44vkh6bzramngvcrficy2no44	32	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	32	15	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	32	16	search	search	NN
work_v44vkh6bzramngvcrficy2no44	32	17	during	during	IN
work_v44vkh6bzramngvcrficy2no44	32	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	32	19	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	32	20	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	32	21	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	32	22	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	32	23	process	process	NN
work_v44vkh6bzramngvcrficy2no44	32	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	33	1	See	see	VB
work_v44vkh6bzramngvcrficy2no44	33	2	for	for	IN
work_v44vkh6bzramngvcrficy2no44	33	3	instance	instance	NN
work_v44vkh6bzramngvcrficy2no44	33	4	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	33	5	Costa	Costa	NNP
work_v44vkh6bzramngvcrficy2no44	33	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	33	7	jussà	jussà	NNS
work_v44vkh6bzramngvcrficy2no44	33	8	and	and	CC
work_v44vkh6bzramngvcrficy2no44	33	9	Fonol-	Fonol-	NNP
work_v44vkh6bzramngvcrficy2no44	33	10	losa	losa	NN
work_v44vkh6bzramngvcrficy2no44	33	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	33	12	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	33	13	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	33	14	for	for	IN
work_v44vkh6bzramngvcrficy2no44	33	15	a	a	DT
work_v44vkh6bzramngvcrficy2no44	33	16	review	review	NN
work_v44vkh6bzramngvcrficy2no44	33	17	of	of	IN
work_v44vkh6bzramngvcrficy2no44	33	18	pre-	pre-	NN
work_v44vkh6bzramngvcrficy2no44	33	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	33	20	post	post	JJ
work_v44vkh6bzramngvcrficy2no44	33	21	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	33	22	reordering	reordering	JJ
work_v44vkh6bzramngvcrficy2no44	33	23	approaches	approach	NNS
work_v44vkh6bzramngvcrficy2no44	33	24	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	33	25	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	33	26	not	not	RB
work_v44vkh6bzramngvcrficy2no44	33	27	treated	treat	VBN
work_v44vkh6bzramngvcrficy2no44	33	28	here	here	RB
work_v44vkh6bzramngvcrficy2no44	33	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	34	1	Assuming	assume	VBG
work_v44vkh6bzramngvcrficy2no44	34	2	a	a	DT
work_v44vkh6bzramngvcrficy2no44	34	3	one	one	CD
work_v44vkh6bzramngvcrficy2no44	34	4	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	34	5	to	to	IN
work_v44vkh6bzramngvcrficy2no44	34	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	34	7	one	one	CD
work_v44vkh6bzramngvcrficy2no44	34	8	correspondence	correspondence	NN
work_v44vkh6bzramngvcrficy2no44	34	9	between	between	IN
work_v44vkh6bzramngvcrficy2no44	34	10	source	source	NN
work_v44vkh6bzramngvcrficy2no44	34	11	and	and	CC
work_v44vkh6bzramngvcrficy2no44	34	12	target	target	NN
work_v44vkh6bzramngvcrficy2no44	34	13	phrases	phrase	NNS
work_v44vkh6bzramngvcrficy2no44	34	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	34	15	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	34	16	in	in	IN
work_v44vkh6bzramngvcrficy2no44	34	17	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	34	18	can	can	MD
work_v44vkh6bzramngvcrficy2no44	34	19	be	be	VB
work_v44vkh6bzramngvcrficy2no44	34	20	viewed	view	VBN
work_v44vkh6bzramngvcrficy2no44	34	21	as	as	IN
work_v44vkh6bzramngvcrficy2no44	34	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	34	23	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	34	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	34	25	searching	search	VBG
work_v44vkh6bzramngvcrficy2no44	34	26	through	through	IN
work_v44vkh6bzramngvcrficy2no44	34	27	a	a	DT
work_v44vkh6bzramngvcrficy2no44	34	28	set	set	NN
work_v44vkh6bzramngvcrficy2no44	34	29	of	of	IN
work_v44vkh6bzramngvcrficy2no44	34	30	permutations	permutation	NNS
work_v44vkh6bzramngvcrficy2no44	34	31	of	of	IN
work_v44vkh6bzramngvcrficy2no44	34	32	the	the	DT
work_v44vkh6bzramngvcrficy2no44	34	33	input	input	NN
work_v44vkh6bzramngvcrficy2no44	34	34	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	34	35	.	.	.
work_v44vkh6bzramngvcrficy2no44	35	1	Thus	thus	RB
work_v44vkh6bzramngvcrficy2no44	35	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	35	3	two	two	CD
work_v44vkh6bzramngvcrficy2no44	35	4	sub	sub	NN
work_v44vkh6bzramngvcrficy2no44	35	5	-	-	NNS
work_v44vkh6bzramngvcrficy2no44	35	6	problems	problem	NNS
work_v44vkh6bzramngvcrficy2no44	35	7	arise	arise	VBP
work_v44vkh6bzramngvcrficy2no44	35	8	:	:	:
work_v44vkh6bzramngvcrficy2no44	35	9	defining	define	VBG
work_v44vkh6bzramngvcrficy2no44	35	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	35	11	set	set	NN
work_v44vkh6bzramngvcrficy2no44	35	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	35	13	allowed	allow	VBN
work_v44vkh6bzramngvcrficy2no44	35	14	per-	per-	NN
work_v44vkh6bzramngvcrficy2no44	35	15	mutations	mutation	NNS
work_v44vkh6bzramngvcrficy2no44	35	16	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	35	17	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	35	18	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	35	19	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	35	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	35	21	scoring	score	VBG
work_v44vkh6bzramngvcrficy2no44	35	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	35	23	allowed	allow	VBN
work_v44vkh6bzramngvcrficy2no44	35	24	permutations	permutation	NNS
work_v44vkh6bzramngvcrficy2no44	35	25	according	accord	VBG
work_v44vkh6bzramngvcrficy2no44	35	26	to	to	IN
work_v44vkh6bzramngvcrficy2no44	35	27	some	some	DT
work_v44vkh6bzramngvcrficy2no44	35	28	likelihood	likelihood	NN
work_v44vkh6bzramngvcrficy2no44	35	29	criterion	criterion	NN
work_v44vkh6bzramngvcrficy2no44	35	30	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	35	31	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	35	32	model	model	NN
work_v44vkh6bzramngvcrficy2no44	35	33	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	35	34	.	.	.
work_v44vkh6bzramngvcrficy2no44	36	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	36	2	begin	begin	VBP
work_v44vkh6bzramngvcrficy2no44	36	3	with	with	IN
work_v44vkh6bzramngvcrficy2no44	36	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	36	5	lat-	lat-	NN
work_v44vkh6bzramngvcrficy2no44	36	6	ter	ter	NN
work_v44vkh6bzramngvcrficy2no44	36	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	36	8	returning	return	VBG
work_v44vkh6bzramngvcrficy2no44	36	9	to	to	IN
work_v44vkh6bzramngvcrficy2no44	36	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	36	11	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	36	12	later	later	RB
work_v44vkh6bzramngvcrficy2no44	36	13	in	in	IN
work_v44vkh6bzramngvcrficy2no44	36	14	this	this	DT
work_v44vkh6bzramngvcrficy2no44	36	15	section	section	NN
work_v44vkh6bzramngvcrficy2no44	36	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	37	1	2.1	2.1	CD
work_v44vkh6bzramngvcrficy2no44	37	2	Reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	37	3	modeling	modeling	NN
work_v44vkh6bzramngvcrficy2no44	37	4	In	in	IN
work_v44vkh6bzramngvcrficy2no44	37	5	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	37	6	original	original	JJ
work_v44vkh6bzramngvcrficy2no44	37	7	formulation	formulation	NN
work_v44vkh6bzramngvcrficy2no44	37	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	37	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	37	10	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	37	11	approach	approach	NN
work_v44vkh6bzramngvcrficy2no44	37	12	includes	include	VBZ
work_v44vkh6bzramngvcrficy2no44	37	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	37	14	basic	basic	JJ
work_v44vkh6bzramngvcrficy2no44	37	15	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	37	16	model	model	NN
work_v44vkh6bzramngvcrficy2no44	37	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	37	18	called	call	VBD
work_v44vkh6bzramngvcrficy2no44	37	19	distor-	distor-	FW
work_v44vkh6bzramngvcrficy2no44	37	20	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	37	21	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	37	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	37	23	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	37	24	exponentially	exponentially	RB
work_v44vkh6bzramngvcrficy2no44	37	25	penalizes	penalize	VBZ
work_v44vkh6bzramngvcrficy2no44	37	26	longer	long	JJR
work_v44vkh6bzramngvcrficy2no44	37	27	jumps	jump	VBZ
work_v44vkh6bzramngvcrficy2no44	37	28	among	among	IN
work_v44vkh6bzramngvcrficy2no44	37	29	consecutively	consecutively	RB
work_v44vkh6bzramngvcrficy2no44	37	30	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	37	31	phrases	phrase	NNS
work_v44vkh6bzramngvcrficy2no44	37	32	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	37	33	f̃i!1	f̃i!1	NNP
work_v44vkh6bzramngvcrficy2no44	37	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	37	35	f̃i	f̃i	NNP
work_v44vkh6bzramngvcrficy2no44	37	36	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	37	37	:	:	:
work_v44vkh6bzramngvcrficy2no44	37	38	d(f̃i!1	d(f̃i!1	NNP
work_v44vkh6bzramngvcrficy2no44	37	39	,	,	,
work_v44vkh6bzramngvcrficy2no44	37	40	f̃i	f̃i	NNP
work_v44vkh6bzramngvcrficy2no44	37	41	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	37	42	=	=	NFP
work_v44vkh6bzramngvcrficy2no44	37	43	e	e	NNP
work_v44vkh6bzramngvcrficy2no44	37	44	!	!	.
work_v44vkh6bzramngvcrficy2no44	37	45	|start(f̃i	|start(f̃i	CD
work_v44vkh6bzramngvcrficy2no44	37	46	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	37	47	!	!	.
work_v44vkh6bzramngvcrficy2no44	38	1	end(f̃i!1	end(f̃i!1	NNP
work_v44vkh6bzramngvcrficy2no44	38	2	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	38	3	!	!	.
work_v44vkh6bzramngvcrficy2no44	39	1	1|	1|	CD
work_v44vkh6bzramngvcrficy2no44	39	2	A	a	DT
work_v44vkh6bzramngvcrficy2no44	39	3	number	number	NN
work_v44vkh6bzramngvcrficy2no44	39	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	39	5	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	39	6	sophisticated	sophisticated	JJ
work_v44vkh6bzramngvcrficy2no44	39	7	solutions	solution	NNS
work_v44vkh6bzramngvcrficy2no44	39	8	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	39	9	been	be	VBN
work_v44vkh6bzramngvcrficy2no44	39	10	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	39	11	to	to	TO
work_v44vkh6bzramngvcrficy2no44	39	12	explicitly	explicitly	RB
work_v44vkh6bzramngvcrficy2no44	39	13	model	model	VB
work_v44vkh6bzramngvcrficy2no44	39	14	word	word	NN
work_v44vkh6bzramngvcrficy2no44	39	15	reorder-	reorder-	NNP
work_v44vkh6bzramngvcrficy2no44	39	16	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	39	17	during	during	IN
work_v44vkh6bzramngvcrficy2no44	39	18	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	39	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	40	1	These	these	DT
work_v44vkh6bzramngvcrficy2no44	40	2	can	can	MD
work_v44vkh6bzramngvcrficy2no44	40	3	mostly	mostly	RB
work_v44vkh6bzramngvcrficy2no44	40	4	be	be	VB
work_v44vkh6bzramngvcrficy2no44	40	5	grouped	group	VBN
work_v44vkh6bzramngvcrficy2no44	40	6	into	into	IN
work_v44vkh6bzramngvcrficy2no44	40	7	three	three	CD
work_v44vkh6bzramngvcrficy2no44	40	8	families	family	NNS
work_v44vkh6bzramngvcrficy2no44	40	9	:	:	:
work_v44vkh6bzramngvcrficy2no44	40	10	phrase	phrase	VB
work_v44vkh6bzramngvcrficy2no44	40	11	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	40	12	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	40	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	40	14	jump	jump	VB
work_v44vkh6bzramngvcrficy2no44	40	15	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	40	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	40	17	source	source	NN
work_v44vkh6bzramngvcrficy2no44	40	18	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	40	19	sequence	sequence	NN
work_v44vkh6bzramngvcrficy2no44	40	20	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	40	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	41	1	Phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	41	2	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	41	3	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	41	4	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	41	5	Tillmann	Tillmann	NNP
work_v44vkh6bzramngvcrficy2no44	41	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	7	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	41	8	;	;	:
work_v44vkh6bzramngvcrficy2no44	41	9	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	41	10	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	41	11	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	41	12	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	41	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	14	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	41	15	;	;	:
work_v44vkh6bzramngvcrficy2no44	41	16	Zens	Zens	NNP
work_v44vkh6bzramngvcrficy2no44	41	17	and	and	CC
work_v44vkh6bzramngvcrficy2no44	41	18	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	41	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	20	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	41	21	;	;	:
work_v44vkh6bzramngvcrficy2no44	41	22	Galley	Galley	NNP
work_v44vkh6bzramngvcrficy2no44	41	23	and	and	CC
work_v44vkh6bzramngvcrficy2no44	41	24	Manning	Manning	NNP
work_v44vkh6bzramngvcrficy2no44	41	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	26	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	41	27	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	41	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	29	also	also	RB
work_v44vkh6bzramngvcrficy2no44	41	30	known	know	VBN
work_v44vkh6bzramngvcrficy2no44	41	31	as	as	IN
work_v44vkh6bzramngvcrficy2no44	41	32	lexicalized	lexicalize	VBN
work_v44vkh6bzramngvcrficy2no44	41	33	reorder-	reorder-	JJ
work_v44vkh6bzramngvcrficy2no44	41	34	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	41	35	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	41	36	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	37	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	41	38	the	the	DT
work_v44vkh6bzramngvcrficy2no44	41	39	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	41	40	of	of	IN
work_v44vkh6bzramngvcrficy2no44	41	41	a	a	DT
work_v44vkh6bzramngvcrficy2no44	41	42	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	41	43	with	with	IN
work_v44vkh6bzramngvcrficy2no44	41	44	respect	respect	NN
work_v44vkh6bzramngvcrficy2no44	41	45	to	to	IN
work_v44vkh6bzramngvcrficy2no44	41	46	the	the	DT
work_v44vkh6bzramngvcrficy2no44	41	47	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	41	48	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	41	49	one	one	NN
work_v44vkh6bzramngvcrficy2no44	41	50	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	51	by	by	IN
work_v44vkh6bzramngvcrficy2no44	41	52	classifying	classify	VBG
work_v44vkh6bzramngvcrficy2no44	41	53	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	41	54	as	as	IN
work_v44vkh6bzramngvcrficy2no44	41	55	monotone	monotone	NN
work_v44vkh6bzramngvcrficy2no44	41	56	,	,	,
work_v44vkh6bzramngvcrficy2no44	41	57	swap	swap	NN
work_v44vkh6bzramngvcrficy2no44	41	58	or	or	CC
work_v44vkh6bzramngvcrficy2no44	41	59	discontinuous	discontinuous	JJ
work_v44vkh6bzramngvcrficy2no44	41	60	.	.	.
work_v44vkh6bzramngvcrficy2no44	42	1	These	these	DT
work_v44vkh6bzramngvcrficy2no44	42	2	mod-	mod-	NN
work_v44vkh6bzramngvcrficy2no44	42	3	els	el	NNS
work_v44vkh6bzramngvcrficy2no44	42	4	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	42	5	proven	prove	VBN
work_v44vkh6bzramngvcrficy2no44	42	6	very	very	RB
work_v44vkh6bzramngvcrficy2no44	42	7	useful	useful	JJ
work_v44vkh6bzramngvcrficy2no44	42	8	for	for	IN
work_v44vkh6bzramngvcrficy2no44	42	9	short	short	JJ
work_v44vkh6bzramngvcrficy2no44	42	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	42	11	medium-	medium-	JJ
work_v44vkh6bzramngvcrficy2no44	42	12	range	range	NN
work_v44vkh6bzramngvcrficy2no44	42	13	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	42	14	and	and	CC
work_v44vkh6bzramngvcrficy2no44	42	15	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	42	16	among	among	IN
work_v44vkh6bzramngvcrficy2no44	42	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	42	18	most	most	RBS
work_v44vkh6bzramngvcrficy2no44	42	19	widely	widely	RB
work_v44vkh6bzramngvcrficy2no44	42	20	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	42	21	in	in	IN
work_v44vkh6bzramngvcrficy2no44	42	22	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	42	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	43	1	However	however	RB
work_v44vkh6bzramngvcrficy2no44	43	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	43	3	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	43	4	coarse	coarse	JJ
work_v44vkh6bzramngvcrficy2no44	43	5	classification	classification	NN
work_v44vkh6bzramngvcrficy2no44	43	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	43	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	43	8	steps	step	NNS
work_v44vkh6bzramngvcrficy2no44	43	9	makes	make	VBZ
work_v44vkh6bzramngvcrficy2no44	43	10	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	43	11	unsuitable	unsuitable	JJ
work_v44vkh6bzramngvcrficy2no44	43	12	to	to	TO
work_v44vkh6bzramngvcrficy2no44	43	13	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	43	14	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	43	15	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	43	16	range	range	NN
work_v44vkh6bzramngvcrficy2no44	43	17	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	43	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	44	1	Jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	44	2	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	44	3	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	44	4	Al	Al	NNP
work_v44vkh6bzramngvcrficy2no44	44	5	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	44	6	Onaizan	Onaizan	NNP
work_v44vkh6bzramngvcrficy2no44	44	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	44	8	Papineni	Papineni	NNP
work_v44vkh6bzramngvcrficy2no44	44	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	44	10	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	44	11	;	;	:
work_v44vkh6bzramngvcrficy2no44	44	12	Green	Green	NNP
work_v44vkh6bzramngvcrficy2no44	44	13	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	44	14	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	44	15	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	44	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	44	17	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	44	18	;	;	:
work_v44vkh6bzramngvcrficy2no44	44	19	Yahyaei	Yahyaei	NNP
work_v44vkh6bzramngvcrficy2no44	44	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	44	21	Monz	Monz	NNP
work_v44vkh6bzramngvcrficy2no44	44	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	44	23	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	44	24	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	44	25	predict	predict	VBP
work_v44vkh6bzramngvcrficy2no44	44	26	the	the	DT
work_v44vkh6bzramngvcrficy2no44	44	27	direction	direction	NN
work_v44vkh6bzramngvcrficy2no44	44	28	and	and	CC
work_v44vkh6bzramngvcrficy2no44	44	29	length	length	NN
work_v44vkh6bzramngvcrficy2no44	44	30	of	of	IN
work_v44vkh6bzramngvcrficy2no44	44	31	a	a	DT
work_v44vkh6bzramngvcrficy2no44	44	32	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	44	33	to	to	TO
work_v44vkh6bzramngvcrficy2no44	44	34	perform	perform	VB
work_v44vkh6bzramngvcrficy2no44	44	35	after	after	IN
work_v44vkh6bzramngvcrficy2no44	44	36	a	a	DT
work_v44vkh6bzramngvcrficy2no44	44	37	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	44	38	input	input	NN
work_v44vkh6bzramngvcrficy2no44	44	39	word2	word2	NN
work_v44vkh6bzramngvcrficy2no44	44	40	.	.	.
work_v44vkh6bzramngvcrficy2no44	45	1	Both	both	DT
work_v44vkh6bzramngvcrficy2no44	45	2	these	these	DT
work_v44vkh6bzramngvcrficy2no44	45	3	works	work	NNS
work_v44vkh6bzramngvcrficy2no44	45	4	achieve	achieve	VBP
work_v44vkh6bzramngvcrficy2no44	45	5	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	45	6	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	45	7	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	45	8	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	45	9	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	45	10	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	45	11	within	within	IN
work_v44vkh6bzramngvcrficy2no44	45	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	45	13	rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	45	14	small	small	JJ
work_v44vkh6bzramngvcrficy2no44	45	15	DL	dl	NN
work_v44vkh6bzramngvcrficy2no44	45	16	:	:	:
work_v44vkh6bzramngvcrficy2no44	45	17	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	45	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	45	19	8	8	CD
work_v44vkh6bzramngvcrficy2no44	45	20	in	in	IN
work_v44vkh6bzramngvcrficy2no44	45	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	45	22	Al	Al	NNP
work_v44vkh6bzramngvcrficy2no44	45	23	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	45	24	Onaizan	Onaizan	NNP
work_v44vkh6bzramngvcrficy2no44	45	25	and	and	CC
work_v44vkh6bzramngvcrficy2no44	45	26	Papineni	Papineni	NNP
work_v44vkh6bzramngvcrficy2no44	45	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	45	28	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	45	29	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	45	30	and	and	CC
work_v44vkh6bzramngvcrficy2no44	45	31	5	5	CD
work_v44vkh6bzramngvcrficy2no44	45	32	in	in	IN
work_v44vkh6bzramngvcrficy2no44	45	33	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	45	34	Green	Green	NNP
work_v44vkh6bzramngvcrficy2no44	45	35	et	et	FW
work_v44vkh6bzramngvcrficy2no44	45	36	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	45	37	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	45	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	45	39	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	45	40	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	45	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	45	42	thus	thus	RB
work_v44vkh6bzramngvcrficy2no44	45	43	failing	fail	VBG
work_v44vkh6bzramngvcrficy2no44	45	44	to	to	TO
work_v44vkh6bzramngvcrficy2no44	45	45	capture	capture	VB
work_v44vkh6bzramngvcrficy2no44	45	46	the	the	DT
work_v44vkh6bzramngvcrficy2no44	45	47	rare	rare	JJ
work_v44vkh6bzramngvcrficy2no44	45	48	but	but	CC
work_v44vkh6bzramngvcrficy2no44	45	49	crucial	crucial	JJ
work_v44vkh6bzramngvcrficy2no44	45	50	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	45	51	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	45	52	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	45	53	were	be	VBD
work_v44vkh6bzramngvcrficy2no44	45	54	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	45	55	main	main	JJ
work_v44vkh6bzramngvcrficy2no44	45	56	motivation	motivation	NN
work_v44vkh6bzramngvcrficy2no44	45	57	.	.	.
work_v44vkh6bzramngvcrficy2no44	46	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	46	2	drawback	drawback	NN
work_v44vkh6bzramngvcrficy2no44	46	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	46	4	this	this	DT
work_v44vkh6bzramngvcrficy2no44	46	5	approach	approach	NN
work_v44vkh6bzramngvcrficy2no44	46	6	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	46	7	that	that	IN
work_v44vkh6bzramngvcrficy2no44	46	8	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	46	9	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	46	10	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	46	11	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	46	12	penalized	penalize	VBN
work_v44vkh6bzramngvcrficy2no44	46	13	because	because	IN
work_v44vkh6bzramngvcrficy2no44	46	14	of	of	IN
work_v44vkh6bzramngvcrficy2no44	46	15	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	46	16	low	low	JJ
work_v44vkh6bzramngvcrficy2no44	46	17	frequency	frequency	NN
work_v44vkh6bzramngvcrficy2no44	46	18	compared	compare	VBN
work_v44vkh6bzramngvcrficy2no44	46	19	to	to	IN
work_v44vkh6bzramngvcrficy2no44	46	20	short	short	JJ
work_v44vkh6bzramngvcrficy2no44	46	21	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	46	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	47	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	47	2	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	47	3	bias	bias	NN
work_v44vkh6bzramngvcrficy2no44	47	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	47	5	undesirable	undesirable	JJ
work_v44vkh6bzramngvcrficy2no44	47	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	47	7	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	47	8	that	that	IN
work_v44vkh6bzramngvcrficy2no44	47	9	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	47	10	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	47	11	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	47	12	in-	in-	RB
work_v44vkh6bzramngvcrficy2no44	47	13	terested	terested	JJ
work_v44vkh6bzramngvcrficy2no44	47	14	in	in	IN
work_v44vkh6bzramngvcrficy2no44	47	15	detecting	detect	VBG
work_v44vkh6bzramngvcrficy2no44	47	16	probable	probable	JJ
work_v44vkh6bzramngvcrficy2no44	47	17	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	47	18	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	47	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	48	1	Source	source	NN
work_v44vkh6bzramngvcrficy2no44	48	2	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	48	3	sequence	sequence	NN
work_v44vkh6bzramngvcrficy2no44	48	4	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	48	5	predict	predict	VBP
work_v44vkh6bzramngvcrficy2no44	48	6	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	48	7	input	input	NN
work_v44vkh6bzramngvcrficy2no44	48	8	word	word	NN
work_v44vkh6bzramngvcrficy2no44	48	9	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	48	10	likely	likely	JJ
work_v44vkh6bzramngvcrficy2no44	48	11	to	to	TO
work_v44vkh6bzramngvcrficy2no44	48	12	be	be	VB
work_v44vkh6bzramngvcrficy2no44	48	13	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	48	14	at	at	IN
work_v44vkh6bzramngvcrficy2no44	48	15	a	a	DT
work_v44vkh6bzramngvcrficy2no44	48	16	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	48	17	state	state	NN
work_v44vkh6bzramngvcrficy2no44	48	18	of	of	IN
work_v44vkh6bzramngvcrficy2no44	48	19	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	48	20	.	.	.
work_v44vkh6bzramngvcrficy2no44	49	1	For	for	IN
work_v44vkh6bzramngvcrficy2no44	49	2	instance	instance	NN
work_v44vkh6bzramngvcrficy2no44	49	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	49	4	reordered	reorder	VBN
work_v44vkh6bzramngvcrficy2no44	49	5	source	source	NN
work_v44vkh6bzramngvcrficy2no44	49	6	lan-	lan-	NN
work_v44vkh6bzramngvcrficy2no44	49	7	guage	guage	NN
work_v44vkh6bzramngvcrficy2no44	49	8	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	49	9	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	49	10	Feng	Feng	NNP
work_v44vkh6bzramngvcrficy2no44	49	11	et	et	FW
work_v44vkh6bzramngvcrficy2no44	49	12	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	49	13	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	49	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	49	15	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	49	16	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	49	17	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	49	18	smoothed	smooth	VBN
work_v44vkh6bzramngvcrficy2no44	49	19	n-	n-	PRP
work_v44vkh6bzramngvcrficy2no44	49	20	gram	gram	NN
work_v44vkh6bzramngvcrficy2no44	49	21	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	49	22	trained	train	VBN
work_v44vkh6bzramngvcrficy2no44	49	23	on	on	IN
work_v44vkh6bzramngvcrficy2no44	49	24	a	a	DT
work_v44vkh6bzramngvcrficy2no44	49	25	corpus	corpus	NN
work_v44vkh6bzramngvcrficy2no44	49	26	of	of	IN
work_v44vkh6bzramngvcrficy2no44	49	27	source	source	NN
work_v44vkh6bzramngvcrficy2no44	49	28	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	49	29	reordered	reorder	VBN
work_v44vkh6bzramngvcrficy2no44	49	30	to	to	TO
work_v44vkh6bzramngvcrficy2no44	49	31	match	match	VB
work_v44vkh6bzramngvcrficy2no44	49	32	the	the	DT
work_v44vkh6bzramngvcrficy2no44	49	33	target	target	NN
work_v44vkh6bzramngvcrficy2no44	49	34	word	word	NN
work_v44vkh6bzramngvcrficy2no44	49	35	order	order	NN
work_v44vkh6bzramngvcrficy2no44	49	36	.	.	.
work_v44vkh6bzramngvcrficy2no44	50	1	When	when	WRB
work_v44vkh6bzramngvcrficy2no44	50	2	inte-	inte-	NNP
work_v44vkh6bzramngvcrficy2no44	50	3	grated	grate	VBD
work_v44vkh6bzramngvcrficy2no44	50	4	into	into	IN
work_v44vkh6bzramngvcrficy2no44	50	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	50	6	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	50	7	system	system	NN
work_v44vkh6bzramngvcrficy2no44	50	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	50	9	they	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	50	10	assign	assign	VBP
work_v44vkh6bzramngvcrficy2no44	50	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	50	12	probabil-	probabil-	JJ
work_v44vkh6bzramngvcrficy2no44	50	13	ity	ity	NN
work_v44vkh6bzramngvcrficy2no44	50	14	to	to	IN
work_v44vkh6bzramngvcrficy2no44	50	15	each	each	DT
work_v44vkh6bzramngvcrficy2no44	50	16	newly	newly	RB
work_v44vkh6bzramngvcrficy2no44	50	17	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	50	18	word	word	NN
work_v44vkh6bzramngvcrficy2no44	50	19	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	50	20	the	the	DT
work_v44vkh6bzramngvcrficy2no44	50	21	n-1	n-1	NNP
work_v44vkh6bzramngvcrficy2no44	50	22	pre-	pre-	NN
work_v44vkh6bzramngvcrficy2no44	50	23	viously	viously	RB
work_v44vkh6bzramngvcrficy2no44	50	24	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	50	25	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	50	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	51	1	Finally	finally	RB
work_v44vkh6bzramngvcrficy2no44	51	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	51	3	source	source	NN
work_v44vkh6bzramngvcrficy2no44	51	4	word	word	NN
work_v44vkh6bzramngvcrficy2no44	51	5	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	51	6	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	51	7	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	51	8	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	51	9	Visweswariah	Visweswariah	NNP
work_v44vkh6bzramngvcrficy2no44	51	10	et	et	FW
work_v44vkh6bzramngvcrficy2no44	51	11	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	51	12	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	51	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	51	14	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	51	15	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	51	16	esti-	esti-	NNP
work_v44vkh6bzramngvcrficy2no44	51	17	mate	mate	NNP
work_v44vkh6bzramngvcrficy2no44	51	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	51	19	for	for	IN
work_v44vkh6bzramngvcrficy2no44	51	20	each	each	DT
work_v44vkh6bzramngvcrficy2no44	51	21	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	51	22	of	of	IN
work_v44vkh6bzramngvcrficy2no44	51	23	input	input	NN
work_v44vkh6bzramngvcrficy2no44	51	24	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	51	25	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	51	26	and	and	CC
work_v44vkh6bzramngvcrficy2no44	51	27	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	51	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	51	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	51	30	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	51	31	of	of	IN
work_v44vkh6bzramngvcrficy2no44	51	32	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	51	33	j	j	UH
work_v44vkh6bzramngvcrficy2no44	51	34	right	right	RB
work_v44vkh6bzramngvcrficy2no44	51	35	after	after	IN
work_v44vkh6bzramngvcrficy2no44	51	36	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	51	37	given	give	VBD
work_v44vkh6bzramngvcrficy2no44	51	38	various	various	JJ
work_v44vkh6bzramngvcrficy2no44	51	39	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	51	40	of	of	IN
work_v44vkh6bzramngvcrficy2no44	51	41	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	51	42	,	,	,
work_v44vkh6bzramngvcrficy2no44	51	43	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	51	44	and	and	CC
work_v44vkh6bzramngvcrficy2no44	51	45	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	51	46	respective	respective	JJ
work_v44vkh6bzramngvcrficy2no44	51	47	contexts	contexts	NN
work_v44vkh6bzramngvcrficy2no44	51	48	.	.	.
work_v44vkh6bzramngvcrficy2no44	52	1	Differently	differently	RB
work_v44vkh6bzramngvcrficy2no44	52	2	from	from	IN
work_v44vkh6bzramngvcrficy2no44	52	3	reordered	reorder	VBN
work_v44vkh6bzramngvcrficy2no44	52	4	source	source	NN
work_v44vkh6bzramngvcrficy2no44	52	5	LMs	lm	VBZ
work_v44vkh6bzramngvcrficy2no44	52	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	52	7	these	these	DT
work_v44vkh6bzramngvcrficy2no44	52	8	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	52	9	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	52	10	discrimina-	discrimina-	NNP
work_v44vkh6bzramngvcrficy2no44	52	11	tive	tive	JJ
work_v44vkh6bzramngvcrficy2no44	52	12	and	and	CC
work_v44vkh6bzramngvcrficy2no44	52	13	can	can	MD
work_v44vkh6bzramngvcrficy2no44	52	14	profit	profit	VB
work_v44vkh6bzramngvcrficy2no44	52	15	from	from	IN
work_v44vkh6bzramngvcrficy2no44	52	16	richer	rich	JJR
work_v44vkh6bzramngvcrficy2no44	52	17	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	52	18	sets	set	NNS
work_v44vkh6bzramngvcrficy2no44	52	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	53	1	At	at	IN
work_v44vkh6bzramngvcrficy2no44	53	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	53	3	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	53	4	time	time	NN
work_v44vkh6bzramngvcrficy2no44	53	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	53	6	they	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	53	7	do	do	VBP
work_v44vkh6bzramngvcrficy2no44	53	8	not	not	RB
work_v44vkh6bzramngvcrficy2no44	53	9	employ	employ	VB
work_v44vkh6bzramngvcrficy2no44	53	10	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	53	11	history-	history-	NNP
work_v44vkh6bzramngvcrficy2no44	53	12	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	53	13	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	53	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	53	15	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	53	16	allows	allow	VBZ
work_v44vkh6bzramngvcrficy2no44	53	17	for	for	IN
work_v44vkh6bzramngvcrficy2no44	53	18	more	more	JJR
work_v44vkh6bzramngvcrficy2no44	53	19	effective	effective	JJ
work_v44vkh6bzramngvcrficy2no44	53	20	hy-	hy-	JJ
work_v44vkh6bzramngvcrficy2no44	53	21	2In	2In	NNS
work_v44vkh6bzramngvcrficy2no44	53	22	this	this	DT
work_v44vkh6bzramngvcrficy2no44	53	23	paper	paper	NN
work_v44vkh6bzramngvcrficy2no44	53	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	53	25	input	input	NN
work_v44vkh6bzramngvcrficy2no44	53	26	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	53	27	or	or	CC
work_v44vkh6bzramngvcrficy2no44	53	28	source	source	NN
work_v44vkh6bzramngvcrficy2no44	53	29	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	53	30	word	word	NN
work_v44vkh6bzramngvcrficy2no44	53	31	denotes	denote	VBZ
work_v44vkh6bzramngvcrficy2no44	53	32	the	the	DT
work_v44vkh6bzramngvcrficy2no44	53	33	word	word	NN
work_v44vkh6bzramngvcrficy2no44	53	34	at	at	IN
work_v44vkh6bzramngvcrficy2no44	53	35	a	a	DT
work_v44vkh6bzramngvcrficy2no44	53	36	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	53	37	position	position	NN
work_v44vkh6bzramngvcrficy2no44	53	38	of	of	IN
work_v44vkh6bzramngvcrficy2no44	53	39	the	the	DT
work_v44vkh6bzramngvcrficy2no44	53	40	input	input	NN
work_v44vkh6bzramngvcrficy2no44	53	41	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	53	42	,	,	,
work_v44vkh6bzramngvcrficy2no44	53	43	rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	53	44	than	than	IN
work_v44vkh6bzramngvcrficy2no44	53	45	a	a	DT
work_v44vkh6bzramngvcrficy2no44	53	46	word	word	NN
work_v44vkh6bzramngvcrficy2no44	53	47	type	type	NN
work_v44vkh6bzramngvcrficy2no44	53	48	.	.	.
work_v44vkh6bzramngvcrficy2no44	54	1	328	328	CD
work_v44vkh6bzramngvcrficy2no44	54	2	pothesis	pothesis	NN
work_v44vkh6bzramngvcrficy2no44	54	3	recombination	recombination	NN
work_v44vkh6bzramngvcrficy2no44	54	4	.	.	.
work_v44vkh6bzramngvcrficy2no44	55	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	55	2	model	model	NN
work_v44vkh6bzramngvcrficy2no44	55	3	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	55	4	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	55	5	going	go	VBG
work_v44vkh6bzramngvcrficy2no44	55	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	55	7	present	present	VB
work_v44vkh6bzramngvcrficy2no44	55	8	belongs	belong	VBZ
work_v44vkh6bzramngvcrficy2no44	55	9	to	to	IN
work_v44vkh6bzramngvcrficy2no44	55	10	this	this	DT
work_v44vkh6bzramngvcrficy2no44	55	11	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	55	12	sub	sub	NN
work_v44vkh6bzramngvcrficy2no44	55	13	-	-	NN
work_v44vkh6bzramngvcrficy2no44	55	14	group	group	NN
work_v44vkh6bzramngvcrficy2no44	55	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	55	16	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	55	17	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	55	18	find	find	VBP
work_v44vkh6bzramngvcrficy2no44	55	19	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	55	20	suitable	suitable	JJ
work_v44vkh6bzramngvcrficy2no44	55	21	to	to	TO
work_v44vkh6bzramngvcrficy2no44	55	22	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	55	23	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	55	24	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	55	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	56	1	2.2	2.2	CD
work_v44vkh6bzramngvcrficy2no44	56	2	Reordering	Reordering	NNP
work_v44vkh6bzramngvcrficy2no44	56	3	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	56	4	The	the	DT
work_v44vkh6bzramngvcrficy2no44	56	5	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	56	6	constraint	constraint	NN
work_v44vkh6bzramngvcrficy2no44	56	7	originally	originally	RB
work_v44vkh6bzramngvcrficy2no44	56	8	included	include	VBN
work_v44vkh6bzramngvcrficy2no44	56	9	in	in	IN
work_v44vkh6bzramngvcrficy2no44	56	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	56	11	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	56	12	framework	framework	NN
work_v44vkh6bzramngvcrficy2no44	56	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	56	14	implemented	implement	VBD
work_v44vkh6bzramngvcrficy2no44	56	15	in	in	IN
work_v44vkh6bzramngvcrficy2no44	56	16	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	56	17	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	56	18	toolkit	toolkit	NNS
work_v44vkh6bzramngvcrficy2no44	56	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	56	20	Moses	Moses	NNP
work_v44vkh6bzramngvcrficy2no44	56	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	56	22	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	56	23	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	56	24	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	56	25	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	56	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	56	27	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	56	28	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	56	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	56	30	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	56	31	called	call	VBN
work_v44vkh6bzramngvcrficy2no44	56	32	dis-	dis-	RP
work_v44vkh6bzramngvcrficy2no44	56	33	tortion	tortion	NN
work_v44vkh6bzramngvcrficy2no44	56	34	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	56	35	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	56	36	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	56	37	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	56	38	.	.	.
work_v44vkh6bzramngvcrficy2no44	57	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	57	2	consists	consist	VBZ
work_v44vkh6bzramngvcrficy2no44	57	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	57	4	allowing	allow	VBG
work_v44vkh6bzramngvcrficy2no44	57	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	57	6	de-	de-	NN
work_v44vkh6bzramngvcrficy2no44	57	7	coder	coder	NN
work_v44vkh6bzramngvcrficy2no44	57	8	to	to	IN
work_v44vkh6bzramngvcrficy2no44	57	9	skip	skip	NNP
work_v44vkh6bzramngvcrficy2no44	57	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	57	11	or	or	CC
work_v44vkh6bzramngvcrficy2no44	57	12	jump	jump	VB
work_v44vkh6bzramngvcrficy2no44	57	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	57	14	at	at	IN
work_v44vkh6bzramngvcrficy2no44	57	15	most	most	JJS
work_v44vkh6bzramngvcrficy2no44	57	16	k	k	NN
work_v44vkh6bzramngvcrficy2no44	57	17	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	57	18	from	from	IN
work_v44vkh6bzramngvcrficy2no44	57	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	57	20	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	57	21	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	57	22	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	57	23	to	to	IN
work_v44vkh6bzramngvcrficy2no44	57	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	57	25	next	next	JJ
work_v44vkh6bzramngvcrficy2no44	57	26	one	one	CD
work_v44vkh6bzramngvcrficy2no44	57	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	58	1	More	more	RBR
work_v44vkh6bzramngvcrficy2no44	58	2	precisely	precisely	RB
work_v44vkh6bzramngvcrficy2no44	58	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	58	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	58	5	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	58	6	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	58	7	imposed	impose	VBN
work_v44vkh6bzramngvcrficy2no44	58	8	on	on	IN
work_v44vkh6bzramngvcrficy2no44	58	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	58	10	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	58	11	D	d	NN
work_v44vkh6bzramngvcrficy2no44	58	12	between	between	IN
work_v44vkh6bzramngvcrficy2no44	58	13	consec-	consec-	NNP
work_v44vkh6bzramngvcrficy2no44	58	14	utively	utively	RB
work_v44vkh6bzramngvcrficy2no44	58	15	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	58	16	phrases	phrase	NNS
work_v44vkh6bzramngvcrficy2no44	58	17	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	58	18	f̃i!1	f̃i!1	NNP
work_v44vkh6bzramngvcrficy2no44	58	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	58	20	f̃i	f̃i	NNP
work_v44vkh6bzramngvcrficy2no44	58	21	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	58	22	:	:	:
work_v44vkh6bzramngvcrficy2no44	58	23	D(f̃i!1	D(f̃i!1	NNP
work_v44vkh6bzramngvcrficy2no44	58	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	58	25	f̃i	f̃i	NNP
work_v44vkh6bzramngvcrficy2no44	58	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	58	27	=	=	NFP
work_v44vkh6bzramngvcrficy2no44	58	28	!	!	.
work_v44vkh6bzramngvcrficy2no44	58	29	!	!	.
work_v44vkh6bzramngvcrficy2no44	59	1	!	!	.
work_v44vkh6bzramngvcrficy2no44	59	2	start(f̃i	start(f̃i	NNP
work_v44vkh6bzramngvcrficy2no44	59	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	59	4	!	!	.
work_v44vkh6bzramngvcrficy2no44	60	1	end(f̃i!1	end(f̃i!1	NNP
work_v44vkh6bzramngvcrficy2no44	60	2	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	60	3	!	!	.
work_v44vkh6bzramngvcrficy2no44	61	1	1	1	LS
work_v44vkh6bzramngvcrficy2no44	61	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	61	3	!	!	.
work_v44vkh6bzramngvcrficy2no44	61	4	!	!	.
work_v44vkh6bzramngvcrficy2no44	61	5	"	"	''
work_v44vkh6bzramngvcrficy2no44	62	1	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	62	2	Limiting	limit	VBG
work_v44vkh6bzramngvcrficy2no44	62	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	62	4	input	input	NN
work_v44vkh6bzramngvcrficy2no44	62	5	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	62	6	space	space	NN
work_v44vkh6bzramngvcrficy2no44	62	7	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	62	8	necessary	necessary	JJ
work_v44vkh6bzramngvcrficy2no44	62	9	for	for	IN
work_v44vkh6bzramngvcrficy2no44	62	10	beam	beam	NN
work_v44vkh6bzramngvcrficy2no44	62	11	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	62	12	search	search	NN
work_v44vkh6bzramngvcrficy2no44	62	13	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	62	14	decoders	decoder	VBZ
work_v44vkh6bzramngvcrficy2no44	62	15	to	to	TO
work_v44vkh6bzramngvcrficy2no44	62	16	function	function	VB
work_v44vkh6bzramngvcrficy2no44	62	17	in	in	IN
work_v44vkh6bzramngvcrficy2no44	62	18	lin-	lin-	XX
work_v44vkh6bzramngvcrficy2no44	62	19	ear	ear	XX
work_v44vkh6bzramngvcrficy2no44	62	20	time	time	NN
work_v44vkh6bzramngvcrficy2no44	62	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	63	1	Reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	63	2	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	63	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	63	4	also	also	RB
work_v44vkh6bzramngvcrficy2no44	63	5	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	63	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	63	7	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	63	8	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	63	9	because	because	IN
work_v44vkh6bzramngvcrficy2no44	63	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	63	11	existing	exist	VBG
work_v44vkh6bzramngvcrficy2no44	63	12	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	63	13	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	63	14	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	63	15	not	not	RB
work_v44vkh6bzramngvcrficy2no44	63	16	discriminative	discriminative	JJ
work_v44vkh6bzramngvcrficy2no44	63	17	enough	enough	RB
work_v44vkh6bzramngvcrficy2no44	63	18	to	to	TO
work_v44vkh6bzramngvcrficy2no44	63	19	guide	guide	VB
work_v44vkh6bzramngvcrficy2no44	63	20	the	the	DT
work_v44vkh6bzramngvcrficy2no44	63	21	search	search	NN
work_v44vkh6bzramngvcrficy2no44	63	22	over	over	IN
work_v44vkh6bzramngvcrficy2no44	63	23	very	very	RB
work_v44vkh6bzramngvcrficy2no44	63	24	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	63	25	sets	set	NNS
work_v44vkh6bzramngvcrficy2no44	63	26	of	of	IN
work_v44vkh6bzramngvcrficy2no44	63	27	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	63	28	hypotheses	hypothesis	NNS
work_v44vkh6bzramngvcrficy2no44	63	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	64	1	Despite	despite	IN
work_v44vkh6bzramngvcrficy2no44	64	2	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	64	3	crucial	crucial	JJ
work_v44vkh6bzramngvcrficy2no44	64	4	effects	effect	NNS
work_v44vkh6bzramngvcrficy2no44	64	5	on	on	IN
work_v44vkh6bzramngvcrficy2no44	64	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	64	7	complexity	complexity	NN
work_v44vkh6bzramngvcrficy2no44	64	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	64	9	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	64	10	modeling	modeling	NN
work_v44vkh6bzramngvcrficy2no44	64	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	64	12	though	though	RB
work_v44vkh6bzramngvcrficy2no44	64	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	64	14	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	64	15	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	64	16	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	64	17	drawn	draw	VBN
work_v44vkh6bzramngvcrficy2no44	64	18	less	less	JJR
work_v44vkh6bzramngvcrficy2no44	64	19	attention	attention	NN
work_v44vkh6bzramngvcrficy2no44	64	20	in	in	IN
work_v44vkh6bzramngvcrficy2no44	64	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	64	22	literature	literature	NN
work_v44vkh6bzramngvcrficy2no44	64	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	65	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	65	2	ex-	ex-	JJ
work_v44vkh6bzramngvcrficy2no44	65	3	isting	iste	VBG
work_v44vkh6bzramngvcrficy2no44	65	4	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	65	5	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	65	6	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	65	7	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	65	8	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	65	9	on	on	IN
work_v44vkh6bzramngvcrficy2no44	65	10	word	word	NN
work_v44vkh6bzramngvcrficy2no44	65	11	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	65	12	to	to	IN
work_v44vkh6bzramngvcrficy2no44	65	13	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	65	14	word	word	NN
work_v44vkh6bzramngvcrficy2no44	65	15	distances	distance	NNS
work_v44vkh6bzramngvcrficy2no44	65	16	–	–	:
work_v44vkh6bzramngvcrficy2no44	65	17	IBM	IBM	NNP
work_v44vkh6bzramngvcrficy2no44	65	18	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	65	19	Berger	Berger	NNP
work_v44vkh6bzramngvcrficy2no44	65	20	et	et	FW
work_v44vkh6bzramngvcrficy2no44	65	21	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	65	22	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	65	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	65	24	1996	1996	CD
work_v44vkh6bzramngvcrficy2no44	65	25	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	65	26	and	and	CC
work_v44vkh6bzramngvcrficy2no44	65	27	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	65	28	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	65	29	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	65	30	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	65	31	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	65	32	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	65	33	,	,	,
work_v44vkh6bzramngvcrficy2no44	65	34	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	65	35	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	65	36	–	–	:
work_v44vkh6bzramngvcrficy2no44	65	37	or	or	CC
work_v44vkh6bzramngvcrficy2no44	65	38	on	on	IN
work_v44vkh6bzramngvcrficy2no44	65	39	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	65	40	pat-	pat-	NN
work_v44vkh6bzramngvcrficy2no44	65	41	terns	tern	NNS
work_v44vkh6bzramngvcrficy2no44	65	42	–	–	:
work_v44vkh6bzramngvcrficy2no44	65	43	ITG	ITG	NNP
work_v44vkh6bzramngvcrficy2no44	65	44	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	65	45	Wu	Wu	NNP
work_v44vkh6bzramngvcrficy2no44	65	46	,	,	,
work_v44vkh6bzramngvcrficy2no44	65	47	1997	1997	CD
work_v44vkh6bzramngvcrficy2no44	65	48	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	65	49	.	.	.
work_v44vkh6bzramngvcrficy2no44	66	1	Both	both	DT
work_v44vkh6bzramngvcrficy2no44	66	2	kinds	kind	NNS
work_v44vkh6bzramngvcrficy2no44	66	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	66	4	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	66	5	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	66	6	uniform	uniform	JJ
work_v44vkh6bzramngvcrficy2no44	66	7	throughout	throughout	IN
work_v44vkh6bzramngvcrficy2no44	66	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	66	9	input	input	NN
work_v44vkh6bzramngvcrficy2no44	66	10	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	66	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	66	12	and	and	CC
work_v44vkh6bzramngvcrficy2no44	66	13	in-	in-	RB
work_v44vkh6bzramngvcrficy2no44	66	14	sensitive	sensitive	JJ
work_v44vkh6bzramngvcrficy2no44	66	15	to	to	IN
work_v44vkh6bzramngvcrficy2no44	66	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	66	17	word	word	NN
work_v44vkh6bzramngvcrficy2no44	66	18	being	be	VBG
work_v44vkh6bzramngvcrficy2no44	66	19	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	66	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	66	21	to	to	IN
work_v44vkh6bzramngvcrficy2no44	66	22	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	66	23	con-	con-	NN
work_v44vkh6bzramngvcrficy2no44	66	24	text	text	NN
work_v44vkh6bzramngvcrficy2no44	66	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	67	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	67	2	results	result	VBZ
work_v44vkh6bzramngvcrficy2no44	67	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	67	4	a	a	DT
work_v44vkh6bzramngvcrficy2no44	67	5	very	very	RB
work_v44vkh6bzramngvcrficy2no44	67	6	coarse	coarse	JJ
work_v44vkh6bzramngvcrficy2no44	67	7	definition	definition	NN
work_v44vkh6bzramngvcrficy2no44	67	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	67	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	67	10	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	67	11	search	search	NN
work_v44vkh6bzramngvcrficy2no44	67	12	space	space	NN
work_v44vkh6bzramngvcrficy2no44	67	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	67	14	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	67	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	67	16	problematic	problematic	JJ
work_v44vkh6bzramngvcrficy2no44	67	17	in	in	IN
work_v44vkh6bzramngvcrficy2no44	67	18	lan-	lan-	NN
work_v44vkh6bzramngvcrficy2no44	67	19	guage	guage	NN
work_v44vkh6bzramngvcrficy2no44	67	20	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	67	21	with	with	IN
work_v44vkh6bzramngvcrficy2no44	67	22	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	67	23	syntactic	syntactic	JJ
work_v44vkh6bzramngvcrficy2no44	67	24	structures	structure	NNS
work_v44vkh6bzramngvcrficy2no44	67	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	68	1	To	to	TO
work_v44vkh6bzramngvcrficy2no44	68	2	address	address	VB
work_v44vkh6bzramngvcrficy2no44	68	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	68	4	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	68	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	68	6	Yahyaei	Yahyaei	NNP
work_v44vkh6bzramngvcrficy2no44	68	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	68	8	Monz	Monz	NNP
work_v44vkh6bzramngvcrficy2no44	68	9	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	68	10	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	68	11	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	68	12	present	present	VBP
work_v44vkh6bzramngvcrficy2no44	68	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	68	14	technique	technique	NN
work_v44vkh6bzramngvcrficy2no44	68	15	to	to	TO
work_v44vkh6bzramngvcrficy2no44	68	16	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	68	17	set	set	VB
work_v44vkh6bzramngvcrficy2no44	68	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	68	19	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	68	20	:	:	:
work_v44vkh6bzramngvcrficy2no44	68	21	they	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	68	22	train	train	VBP
work_v44vkh6bzramngvcrficy2no44	68	23	a	a	DT
work_v44vkh6bzramngvcrficy2no44	68	24	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	68	25	to	to	TO
work_v44vkh6bzramngvcrficy2no44	68	26	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	68	27	the	the	DT
work_v44vkh6bzramngvcrficy2no44	68	28	most	most	RBS
work_v44vkh6bzramngvcrficy2no44	68	29	prob-	prob-	JJ
work_v44vkh6bzramngvcrficy2no44	68	30	able	able	JJ
work_v44vkh6bzramngvcrficy2no44	68	31	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	68	32	length	length	NN
work_v44vkh6bzramngvcrficy2no44	68	33	after	after	IN
work_v44vkh6bzramngvcrficy2no44	68	34	each	each	DT
work_v44vkh6bzramngvcrficy2no44	68	35	input	input	NN
work_v44vkh6bzramngvcrficy2no44	68	36	word	word	NN
work_v44vkh6bzramngvcrficy2no44	68	37	,	,	,
work_v44vkh6bzramngvcrficy2no44	68	38	and	and	CC
work_v44vkh6bzramngvcrficy2no44	68	39	use	use	VB
work_v44vkh6bzramngvcrficy2no44	68	40	the	the	DT
work_v44vkh6bzramngvcrficy2no44	68	41	predicted	predict	VBN
work_v44vkh6bzramngvcrficy2no44	68	42	value	value	NN
work_v44vkh6bzramngvcrficy2no44	68	43	as	as	IN
work_v44vkh6bzramngvcrficy2no44	68	44	the	the	DT
work_v44vkh6bzramngvcrficy2no44	68	45	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	68	46	after	after	IN
work_v44vkh6bzramngvcrficy2no44	68	47	that	that	DT
work_v44vkh6bzramngvcrficy2no44	68	48	position	position	NN
work_v44vkh6bzramngvcrficy2no44	68	49	.	.	.
work_v44vkh6bzramngvcrficy2no44	69	1	Un-	Un-	NNP
work_v44vkh6bzramngvcrficy2no44	69	2	fortunately	fortunately	RB
work_v44vkh6bzramngvcrficy2no44	69	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	69	4	this	this	DT
work_v44vkh6bzramngvcrficy2no44	69	5	method	method	NN
work_v44vkh6bzramngvcrficy2no44	69	6	can	can	MD
work_v44vkh6bzramngvcrficy2no44	69	7	generate	generate	VB
work_v44vkh6bzramngvcrficy2no44	69	8	inconsistent	inconsistent	JJ
work_v44vkh6bzramngvcrficy2no44	69	9	constraints	constraint	NNS
work_v44vkh6bzramngvcrficy2no44	69	10	leading	lead	VBG
work_v44vkh6bzramngvcrficy2no44	69	11	to	to	IN
work_v44vkh6bzramngvcrficy2no44	69	12	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	69	13	dead	dead	JJ
work_v44vkh6bzramngvcrficy2no44	69	14	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	69	15	ends	end	NNS
work_v44vkh6bzramngvcrficy2no44	69	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	70	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	70	2	a	a	DT
work_v44vkh6bzramngvcrficy2no44	70	3	so-	so-	JJ
work_v44vkh6bzramngvcrficy2no44	70	4	lution	lution	NN
work_v44vkh6bzramngvcrficy2no44	70	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	70	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	70	7	dynamic	dynamic	JJ
work_v44vkh6bzramngvcrficy2no44	70	8	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	70	9	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	70	10	relaxed	relaxed	JJ
work_v44vkh6bzramngvcrficy2no44	70	11	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	70	12	needed	need	VBN
work_v44vkh6bzramngvcrficy2no44	70	13	to	to	TO
work_v44vkh6bzramngvcrficy2no44	70	14	reach	reach	VB
work_v44vkh6bzramngvcrficy2no44	70	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	70	16	first	first	JJ
work_v44vkh6bzramngvcrficy2no44	70	17	uncovered	uncovered	JJ
work_v44vkh6bzramngvcrficy2no44	70	18	position	position	NN
work_v44vkh6bzramngvcrficy2no44	70	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	71	1	Translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	71	2	im-	im-	NN
work_v44vkh6bzramngvcrficy2no44	71	3	provements	provement	NNS
work_v44vkh6bzramngvcrficy2no44	71	4	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	71	5	reported	report	VBN
work_v44vkh6bzramngvcrficy2no44	71	6	only	only	RB
work_v44vkh6bzramngvcrficy2no44	71	7	on	on	IN
work_v44vkh6bzramngvcrficy2no44	71	8	a	a	DT
work_v44vkh6bzramngvcrficy2no44	71	9	small	small	JJ
work_v44vkh6bzramngvcrficy2no44	71	10	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	71	11	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	71	12	task	task	NN
work_v44vkh6bzramngvcrficy2no44	71	13	with	with	IN
work_v44vkh6bzramngvcrficy2no44	71	14	short	short	JJ
work_v44vkh6bzramngvcrficy2no44	71	15	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	71	16	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	71	17	BTEC	BTEC	NNP
work_v44vkh6bzramngvcrficy2no44	71	18	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	71	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	71	20	over	over	IN
work_v44vkh6bzramngvcrficy2no44	71	21	a	a	DT
work_v44vkh6bzramngvcrficy2no44	71	22	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	71	23	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	71	24	in-	in-	NN
work_v44vkh6bzramngvcrficy2no44	71	25	cludes	clude	VBZ
work_v44vkh6bzramngvcrficy2no44	71	26	a	a	DT
work_v44vkh6bzramngvcrficy2no44	71	27	very	very	RB
work_v44vkh6bzramngvcrficy2no44	71	28	simple	simple	JJ
work_v44vkh6bzramngvcrficy2no44	71	29	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	71	30	model	model	NN
work_v44vkh6bzramngvcrficy2no44	71	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	72	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	72	2	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	72	3	work	work	NN
work_v44vkh6bzramngvcrficy2no44	72	4	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	72	5	develop	develop	VBP
work_v44vkh6bzramngvcrficy2no44	72	6	this	this	DT
work_v44vkh6bzramngvcrficy2no44	72	7	idea	idea	NN
work_v44vkh6bzramngvcrficy2no44	72	8	further	far	RBR
work_v44vkh6bzramngvcrficy2no44	72	9	and	and	CC
work_v44vkh6bzramngvcrficy2no44	72	10	use	use	VB
work_v44vkh6bzramngvcrficy2no44	72	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	72	12	reordering	reordering	JJ
work_v44vkh6bzramngvcrficy2no44	72	13	model	model	NN
work_v44vkh6bzramngvcrficy2no44	72	14	to	to	TO
work_v44vkh6bzramngvcrficy2no44	72	15	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	72	16	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	72	17	specific	specific	JJ
work_v44vkh6bzramngvcrficy2no44	72	18	input	input	NN
work_v44vkh6bzramngvcrficy2no44	72	19	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	72	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	72	21	rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	72	22	than	than	IN
work_v44vkh6bzramngvcrficy2no44	72	23	input	input	NN
work_v44vkh6bzramngvcrficy2no44	72	24	intervals	interval	NNS
work_v44vkh6bzramngvcrficy2no44	72	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	72	26	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	72	27	likely	likely	RB
work_v44vkh6bzramngvcrficy2no44	72	28	be	be	VB
work_v44vkh6bzramngvcrficy2no44	72	29	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	72	30	next	next	RB
work_v44vkh6bzramngvcrficy2no44	72	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	73	1	Moreover	moreover	RB
work_v44vkh6bzramngvcrficy2no44	73	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	73	3	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	73	4	solution	solution	NN
work_v44vkh6bzramngvcrficy2no44	73	5	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	73	6	not	not	RB
work_v44vkh6bzramngvcrficy2no44	73	7	affected	affect	VBN
work_v44vkh6bzramngvcrficy2no44	73	8	by	by	IN
work_v44vkh6bzramngvcrficy2no44	73	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	73	10	con-	con-	NNP
work_v44vkh6bzramngvcrficy2no44	73	11	straint	straint	NN
work_v44vkh6bzramngvcrficy2no44	73	12	inconsistency	inconsistency	NN
work_v44vkh6bzramngvcrficy2no44	73	13	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	73	14	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	73	15	see	see	VB
work_v44vkh6bzramngvcrficy2no44	73	16	Sect	Sect	NNP
work_v44vkh6bzramngvcrficy2no44	73	17	.	.	.
work_v44vkh6bzramngvcrficy2no44	74	1	4	4	LS
work_v44vkh6bzramngvcrficy2no44	74	2	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	74	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	75	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	75	2	another	another	DT
work_v44vkh6bzramngvcrficy2no44	75	3	related	related	JJ
work_v44vkh6bzramngvcrficy2no44	75	4	work	work	NN
work_v44vkh6bzramngvcrficy2no44	75	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	75	6	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	75	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	75	8	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	75	9	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	75	10	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	75	11	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	75	12	generate	generate	VB
work_v44vkh6bzramngvcrficy2no44	75	13	likely	likely	JJ
work_v44vkh6bzramngvcrficy2no44	75	14	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	75	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	75	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	75	17	input	input	NN
work_v44vkh6bzramngvcrficy2no44	75	18	sen-	sen-	NN
work_v44vkh6bzramngvcrficy2no44	75	19	tence	tence	NN
work_v44vkh6bzramngvcrficy2no44	75	20	by	by	IN
work_v44vkh6bzramngvcrficy2no44	75	21	means	mean	NNS
work_v44vkh6bzramngvcrficy2no44	75	22	of	of	IN
work_v44vkh6bzramngvcrficy2no44	75	23	language	language	NN
work_v44vkh6bzramngvcrficy2no44	75	24	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	75	25	specific	specific	JJ
work_v44vkh6bzramngvcrficy2no44	75	26	fuzzy	fuzzy	JJ
work_v44vkh6bzramngvcrficy2no44	75	27	rules	rule	NNS
work_v44vkh6bzramngvcrficy2no44	75	28	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	75	29	on	on	IN
work_v44vkh6bzramngvcrficy2no44	75	30	shallow	shallow	JJ
work_v44vkh6bzramngvcrficy2no44	75	31	syntax	syntax	NN
work_v44vkh6bzramngvcrficy2no44	75	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	76	1	Long	long	JJ
work_v44vkh6bzramngvcrficy2no44	76	2	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	76	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	76	4	then	then	RB
work_v44vkh6bzramngvcrficy2no44	76	5	sug-	sug-	RB
work_v44vkh6bzramngvcrficy2no44	76	6	gested	geste	VBN
work_v44vkh6bzramngvcrficy2no44	76	7	to	to	IN
work_v44vkh6bzramngvcrficy2no44	76	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	76	9	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	76	10	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	76	11	by	by	IN
work_v44vkh6bzramngvcrficy2no44	76	12	reducing	reduce	VBG
work_v44vkh6bzramngvcrficy2no44	76	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	76	14	distor-	distor-	NN
work_v44vkh6bzramngvcrficy2no44	76	15	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	76	16	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	76	17	for	for	IN
work_v44vkh6bzramngvcrficy2no44	76	18	specific	specific	JJ
work_v44vkh6bzramngvcrficy2no44	76	19	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	76	20	of	of	IN
work_v44vkh6bzramngvcrficy2no44	76	21	input	input	NN
work_v44vkh6bzramngvcrficy2no44	76	22	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	76	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	77	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	77	2	com-	com-	NN
work_v44vkh6bzramngvcrficy2no44	77	3	parison	parison	NN
work_v44vkh6bzramngvcrficy2no44	77	4	to	to	IN
work_v44vkh6bzramngvcrficy2no44	77	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	77	6	dynamic	dynamic	JJ
work_v44vkh6bzramngvcrficy2no44	77	7	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	77	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	77	9	that	that	DT
work_v44vkh6bzramngvcrficy2no44	77	10	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	77	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	77	12	much	much	RB
work_v44vkh6bzramngvcrficy2no44	77	13	finer	fine	JJR
work_v44vkh6bzramngvcrficy2no44	77	14	way	way	NN
work_v44vkh6bzramngvcrficy2no44	77	15	to	to	TO
work_v44vkh6bzramngvcrficy2no44	77	16	define	define	VB
work_v44vkh6bzramngvcrficy2no44	77	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	77	18	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	77	19	space	space	NN
work_v44vkh6bzramngvcrficy2no44	77	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	77	21	leading	lead	VBG
work_v44vkh6bzramngvcrficy2no44	77	22	to	to	IN
work_v44vkh6bzramngvcrficy2no44	77	23	consistent	consistent	JJ
work_v44vkh6bzramngvcrficy2no44	77	24	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	77	25	of	of	IN
work_v44vkh6bzramngvcrficy2no44	77	26	both	both	DT
work_v44vkh6bzramngvcrficy2no44	77	27	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	77	28	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	77	29	and	and	CC
work_v44vkh6bzramngvcrficy2no44	77	30	effi-	effi-	NN
work_v44vkh6bzramngvcrficy2no44	77	31	ciency	ciency	NN
work_v44vkh6bzramngvcrficy2no44	77	32	over	over	IN
work_v44vkh6bzramngvcrficy2no44	77	33	a	a	DT
work_v44vkh6bzramngvcrficy2no44	77	34	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	77	35	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	77	36	.	.	.
work_v44vkh6bzramngvcrficy2no44	78	1	However	however	RB
work_v44vkh6bzramngvcrficy2no44	78	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	78	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	78	4	need	need	NN
work_v44vkh6bzramngvcrficy2no44	78	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	78	6	specific	specific	JJ
work_v44vkh6bzramngvcrficy2no44	78	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	78	8	rules	rule	NNS
work_v44vkh6bzramngvcrficy2no44	78	9	makes	make	VBZ
work_v44vkh6bzramngvcrficy2no44	78	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	78	11	method	method	NN
work_v44vkh6bzramngvcrficy2no44	78	12	harder	hard	RBR
work_v44vkh6bzramngvcrficy2no44	78	13	to	to	TO
work_v44vkh6bzramngvcrficy2no44	78	14	apply	apply	VB
work_v44vkh6bzramngvcrficy2no44	78	15	to	to	IN
work_v44vkh6bzramngvcrficy2no44	78	16	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	78	17	language	language	NN
work_v44vkh6bzramngvcrficy2no44	78	18	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	78	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	79	1	3	3	LS
work_v44vkh6bzramngvcrficy2no44	79	2	The	the	DT
work_v44vkh6bzramngvcrficy2no44	79	3	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	79	4	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	79	5	model	model	NN
work_v44vkh6bzramngvcrficy2no44	79	6	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	79	7	model	model	VBP
work_v44vkh6bzramngvcrficy2no44	79	8	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	79	9	as	as	IN
work_v44vkh6bzramngvcrficy2no44	79	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	79	11	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	79	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	79	13	deciding	decide	VBG
work_v44vkh6bzramngvcrficy2no44	79	14	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	79	15	a	a	DT
work_v44vkh6bzramngvcrficy2no44	79	16	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	79	17	input	input	NN
work_v44vkh6bzramngvcrficy2no44	79	18	word	word	NN
work_v44vkh6bzramngvcrficy2no44	79	19	should	should	MD
work_v44vkh6bzramngvcrficy2no44	79	20	be	be	VB
work_v44vkh6bzramngvcrficy2no44	79	21	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	79	22	after	after	IN
work_v44vkh6bzramngvcrficy2no44	79	23	another	another	DT
work_v44vkh6bzramngvcrficy2no44	79	24	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	79	25	Word	word	NN
work_v44vkh6bzramngvcrficy2no44	79	26	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	79	27	after	after	IN
work_v44vkh6bzramngvcrficy2no44	79	28	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	79	29	Word	word	NN
work_v44vkh6bzramngvcrficy2no44	79	30	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	79	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	80	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	80	2	formulation	formulation	NN
work_v44vkh6bzramngvcrficy2no44	80	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	80	4	particularly	particularly	RB
work_v44vkh6bzramngvcrficy2no44	80	5	suitable	suitable	JJ
work_v44vkh6bzramngvcrficy2no44	80	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	80	7	help	help	VB
work_v44vkh6bzramngvcrficy2no44	80	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	80	9	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	80	10	decide	decide	VB
work_v44vkh6bzramngvcrficy2no44	80	11	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	80	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	80	13	reordering	reordering	JJ
work_v44vkh6bzramngvcrficy2no44	80	14	path	path	NN
work_v44vkh6bzramngvcrficy2no44	80	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	80	16	promising	promise	VBG
work_v44vkh6bzramngvcrficy2no44	80	17	enough	enough	RB
work_v44vkh6bzramngvcrficy2no44	80	18	to	to	TO
work_v44vkh6bzramngvcrficy2no44	80	19	be	be	VB
work_v44vkh6bzramngvcrficy2no44	80	20	further	further	RB
work_v44vkh6bzramngvcrficy2no44	80	21	explored	explore	VBN
work_v44vkh6bzramngvcrficy2no44	80	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	81	1	Moreover	moreover	RB
work_v44vkh6bzramngvcrficy2no44	81	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	81	3	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	81	4	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	81	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	81	6	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	81	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	81	8	choosing	choose	VBG
work_v44vkh6bzramngvcrficy2no44	81	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	81	10	next	next	JJ
work_v44vkh6bzramngvcrficy2no44	81	11	source	source	NN
work_v44vkh6bzramngvcrficy2no44	81	12	word	word	NN
work_v44vkh6bzramngvcrficy2no44	81	13	to	to	TO
work_v44vkh6bzramngvcrficy2no44	81	14	translate	translate	VB
work_v44vkh6bzramngvcrficy2no44	81	15	appears	appear	VBZ
work_v44vkh6bzramngvcrficy2no44	81	16	as	as	IN
work_v44vkh6bzramngvcrficy2no44	81	17	a	a	DT
work_v44vkh6bzramngvcrficy2no44	81	18	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	81	19	natural	natural	JJ
work_v44vkh6bzramngvcrficy2no44	81	20	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	81	21	than	than	IN
work_v44vkh6bzramngvcrficy2no44	81	22	guessing	guess	VBG
work_v44vkh6bzramngvcrficy2no44	81	23	how	how	WRB
work_v44vkh6bzramngvcrficy2no44	81	24	much	much	RB
work_v44vkh6bzramngvcrficy2no44	81	25	to	to	IN
work_v44vkh6bzramngvcrficy2no44	81	26	the	the	DT
work_v44vkh6bzramngvcrficy2no44	81	27	left	left	NN
work_v44vkh6bzramngvcrficy2no44	81	28	or	or	CC
work_v44vkh6bzramngvcrficy2no44	81	29	to	to	IN
work_v44vkh6bzramngvcrficy2no44	81	30	the	the	DT
work_v44vkh6bzramngvcrficy2no44	81	31	right	right	NN
work_v44vkh6bzramngvcrficy2no44	81	32	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	81	33	should	should	MD
work_v44vkh6bzramngvcrficy2no44	81	34	move	move	VB
work_v44vkh6bzramngvcrficy2no44	81	35	from	from	IN
work_v44vkh6bzramngvcrficy2no44	81	36	the	the	DT
work_v44vkh6bzramngvcrficy2no44	81	37	current	current	JJ
work_v44vkh6bzramngvcrficy2no44	81	38	source	source	NN
work_v44vkh6bzramngvcrficy2no44	81	39	position	position	NN
work_v44vkh6bzramngvcrficy2no44	81	40	.	.	.
work_v44vkh6bzramngvcrficy2no44	82	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	82	2	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	82	3	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	82	4	model	model	NN
work_v44vkh6bzramngvcrficy2no44	82	5	addresses	address	VBZ
work_v44vkh6bzramngvcrficy2no44	82	6	a	a	DT
work_v44vkh6bzramngvcrficy2no44	82	7	binary	binary	JJ
work_v44vkh6bzramngvcrficy2no44	82	8	decision	decision	NN
work_v44vkh6bzramngvcrficy2no44	82	9	task	task	NN
work_v44vkh6bzramngvcrficy2no44	82	10	through	through	IN
work_v44vkh6bzramngvcrficy2no44	82	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	82	12	following	follow	VBG
work_v44vkh6bzramngvcrficy2no44	82	13	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	82	14	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	82	15	entropy	entropy	JJ
work_v44vkh6bzramngvcrficy2no44	82	16	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	82	17	:	:	:
work_v44vkh6bzramngvcrficy2no44	82	18	P	p	NN
work_v44vkh6bzramngvcrficy2no44	82	19	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	82	20	Ri	Ri	NNP
work_v44vkh6bzramngvcrficy2no44	82	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	22	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	82	23	=	=	NFP
work_v44vkh6bzramngvcrficy2no44	82	24	Y	Y	NNP
work_v44vkh6bzramngvcrficy2no44	82	25	|f	|f	NNP
work_v44vkh6bzramngvcrficy2no44	82	26	J1	J1	NNP
work_v44vkh6bzramngvcrficy2no44	82	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	28	i	i	NNP
work_v44vkh6bzramngvcrficy2no44	82	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	30	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	82	31	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	82	32	=	=	NFP
work_v44vkh6bzramngvcrficy2no44	82	33	exp	exp	NN
work_v44vkh6bzramngvcrficy2no44	82	34	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	82	35	"	"	``
work_v44vkh6bzramngvcrficy2no44	82	36	m	m	NNP
work_v44vkh6bzramngvcrficy2no44	82	37	!	!	.
work_v44vkh6bzramngvcrficy2no44	82	38	mhm(f	mhm(f	.
work_v44vkh6bzramngvcrficy2no44	82	39	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	82	40	1	1	CD
work_v44vkh6bzramngvcrficy2no44	82	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	42	i	i	NNP
work_v44vkh6bzramngvcrficy2no44	82	43	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	44	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	82	45	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	46	Ri	Ri	NNP
work_v44vkh6bzramngvcrficy2no44	82	47	,	,	,
work_v44vkh6bzramngvcrficy2no44	82	48	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	82	49	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	82	50	Y	y	NN
work_v44vkh6bzramngvcrficy2no44	82	51	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	82	52	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	82	53	"	"	``
work_v44vkh6bzramngvcrficy2no44	82	54	Y	y	NN
work_v44vkh6bzramngvcrficy2no44	82	55	!	!	.
work_v44vkh6bzramngvcrficy2no44	83	1	exp	exp	NN
work_v44vkh6bzramngvcrficy2no44	83	2	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	83	3	"	"	``
work_v44vkh6bzramngvcrficy2no44	83	4	m	m	NNP
work_v44vkh6bzramngvcrficy2no44	83	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	83	6	mhm(f	mhm(f	.
work_v44vkh6bzramngvcrficy2no44	83	7	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	83	8	1	1	CD
work_v44vkh6bzramngvcrficy2no44	83	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	83	10	i	i	NNP
work_v44vkh6bzramngvcrficy2no44	83	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	83	12	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	83	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	83	14	Ri	Ri	NNP
work_v44vkh6bzramngvcrficy2no44	83	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	83	16	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	83	17	=	=	NN
work_v44vkh6bzramngvcrficy2no44	83	18	Y	Y	NNP
work_v44vkh6bzramngvcrficy2no44	83	19	")	")	''
work_v44vkh6bzramngvcrficy2no44	83	20	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	83	21	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	83	22	f	f	NNP
work_v44vkh6bzramngvcrficy2no44	83	23	J1	J1	NNP
work_v44vkh6bzramngvcrficy2no44	83	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	83	25	a	a	DT
work_v44vkh6bzramngvcrficy2no44	83	26	source	source	NN
work_v44vkh6bzramngvcrficy2no44	83	27	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	83	28	of	of	IN
work_v44vkh6bzramngvcrficy2no44	83	29	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	83	30	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	83	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	83	32	hm	hm	PRP
work_v44vkh6bzramngvcrficy2no44	83	33	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	83	34	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	83	35	functions	function	NNS
work_v44vkh6bzramngvcrficy2no44	83	36	and	and	CC
work_v44vkh6bzramngvcrficy2no44	83	37	!	!	.
work_v44vkh6bzramngvcrficy2no44	83	38	m	m	LS
work_v44vkh6bzramngvcrficy2no44	83	39	the	the	DT
work_v44vkh6bzramngvcrficy2no44	83	40	corresponding	correspond	VBG
work_v44vkh6bzramngvcrficy2no44	83	41	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	83	42	weights	weight	NNS
work_v44vkh6bzramngvcrficy2no44	83	43	.	.	.
work_v44vkh6bzramngvcrficy2no44	84	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	84	2	outcome	outcome	NN
work_v44vkh6bzramngvcrficy2no44	84	3	Y	Y	NNP
work_v44vkh6bzramngvcrficy2no44	84	4	can	can	MD
work_v44vkh6bzramngvcrficy2no44	84	5	be	be	VB
work_v44vkh6bzramngvcrficy2no44	84	6	either	either	CC
work_v44vkh6bzramngvcrficy2no44	84	7	1	1	CD
work_v44vkh6bzramngvcrficy2no44	84	8	or	or	CC
work_v44vkh6bzramngvcrficy2no44	84	9	0	0	CD
work_v44vkh6bzramngvcrficy2no44	84	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	84	11	with	with	IN
work_v44vkh6bzramngvcrficy2no44	84	12	Ri	Ri	NNP
work_v44vkh6bzramngvcrficy2no44	84	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	84	14	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	84	15	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	84	16	1	1	CD
work_v44vkh6bzramngvcrficy2no44	84	17	meaning	meaning	NN
work_v44vkh6bzramngvcrficy2no44	84	18	that	that	IN
work_v44vkh6bzramngvcrficy2no44	84	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	84	20	word	word	NN
work_v44vkh6bzramngvcrficy2no44	84	21	at	at	IN
work_v44vkh6bzramngvcrficy2no44	84	22	position	position	NN
work_v44vkh6bzramngvcrficy2no44	84	23	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	84	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	84	25	trans-	trans-	NNP
work_v44vkh6bzramngvcrficy2no44	84	26	lated	late	VBN
work_v44vkh6bzramngvcrficy2no44	84	27	right	right	RB
work_v44vkh6bzramngvcrficy2no44	84	28	after	after	IN
work_v44vkh6bzramngvcrficy2no44	84	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	84	30	word	word	NN
work_v44vkh6bzramngvcrficy2no44	84	31	at	at	IN
work_v44vkh6bzramngvcrficy2no44	84	32	position	position	NN
work_v44vkh6bzramngvcrficy2no44	84	33	i.	i.	NNP
work_v44vkh6bzramngvcrficy2no44	85	1	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	85	2	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	85	3	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	85	4	model	model	NN
work_v44vkh6bzramngvcrficy2no44	85	5	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	85	6	strongly	strongly	RB
work_v44vkh6bzramngvcrficy2no44	85	7	related	related	JJ
work_v44vkh6bzramngvcrficy2no44	85	8	to	to	IN
work_v44vkh6bzramngvcrficy2no44	85	9	that	that	DT
work_v44vkh6bzramngvcrficy2no44	85	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	85	11	Visweswariah	Visweswariah	NNP
work_v44vkh6bzramngvcrficy2no44	85	12	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	85	13	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	85	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	86	1	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	86	2	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	86	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	86	4	–	–	:
work_v44vkh6bzramngvcrficy2no44	86	5	hereby	hereby	NNP
work_v44vkh6bzramngvcrficy2no44	86	6	called	call	VBD
work_v44vkh6bzramngvcrficy2no44	86	7	Travelling	Travelling	NNP
work_v44vkh6bzramngvcrficy2no44	86	8	Salesman	Salesman	NNP
work_v44vkh6bzramngvcrficy2no44	86	9	Problem	Problem	NNP
work_v44vkh6bzramngvcrficy2no44	86	10	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	86	11	TSP	TSP	NNP
work_v44vkh6bzramngvcrficy2no44	86	12	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	86	13	model	model	NN
work_v44vkh6bzramngvcrficy2no44	86	14	–	–	:
work_v44vkh6bzramngvcrficy2no44	86	15	with	with	IN
work_v44vkh6bzramngvcrficy2no44	86	16	few	few	JJ
work_v44vkh6bzramngvcrficy2no44	86	17	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	86	18	differences	difference	NNS
work_v44vkh6bzramngvcrficy2no44	86	19	:	:	:
work_v44vkh6bzramngvcrficy2no44	86	20	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	86	21	i	i	NN
work_v44vkh6bzramngvcrficy2no44	86	22	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	86	23	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	86	24	do	do	VBP
work_v44vkh6bzramngvcrficy2no44	86	25	not	not	RB
work_v44vkh6bzramngvcrficy2no44	86	26	include	include	VB
work_v44vkh6bzramngvcrficy2no44	86	27	in	in	IN
work_v44vkh6bzramngvcrficy2no44	86	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	29	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	86	30	any	any	DT
work_v44vkh6bzramngvcrficy2no44	86	31	explicit	explicit	JJ
work_v44vkh6bzramngvcrficy2no44	86	32	indication	indication	NN
work_v44vkh6bzramngvcrficy2no44	86	33	of	of	IN
work_v44vkh6bzramngvcrficy2no44	86	34	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	35	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	86	36	length	length	NN
work_v44vkh6bzramngvcrficy2no44	86	37	,	,	,
work_v44vkh6bzramngvcrficy2no44	86	38	in	in	IN
work_v44vkh6bzramngvcrficy2no44	86	39	order	order	NN
work_v44vkh6bzramngvcrficy2no44	86	40	to	to	TO
work_v44vkh6bzramngvcrficy2no44	86	41	avoid	avoid	VB
work_v44vkh6bzramngvcrficy2no44	86	42	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	43	bias	bias	NN
work_v44vkh6bzramngvcrficy2no44	86	44	on	on	IN
work_v44vkh6bzramngvcrficy2no44	86	45	short	short	JJ
work_v44vkh6bzramngvcrficy2no44	86	46	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	86	47	;	;	:
work_v44vkh6bzramngvcrficy2no44	86	48	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	86	49	ii	ii	LS
work_v44vkh6bzramngvcrficy2no44	86	50	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	86	51	they	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	86	52	train	train	VBP
work_v44vkh6bzramngvcrficy2no44	86	53	a	a	DT
work_v44vkh6bzramngvcrficy2no44	86	54	linear	linear	JJ
work_v44vkh6bzramngvcrficy2no44	86	55	model	model	NN
work_v44vkh6bzramngvcrficy2no44	86	56	with	with	IN
work_v44vkh6bzramngvcrficy2no44	86	57	MIRA	MIRA	NNP
work_v44vkh6bzramngvcrficy2no44	86	58	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	86	59	Cram-	Cram-	NNP
work_v44vkh6bzramngvcrficy2no44	86	60	mer	mer	NNP
work_v44vkh6bzramngvcrficy2no44	86	61	and	and	CC
work_v44vkh6bzramngvcrficy2no44	86	62	Singer	Singer	NNP
work_v44vkh6bzramngvcrficy2no44	86	63	,	,	,
work_v44vkh6bzramngvcrficy2no44	86	64	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	86	65	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	86	66	by	by	IN
work_v44vkh6bzramngvcrficy2no44	86	67	minimizing	minimize	VBG
work_v44vkh6bzramngvcrficy2no44	86	68	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	69	number	number	NN
work_v44vkh6bzramngvcrficy2no44	86	70	329	329	CD
work_v44vkh6bzramngvcrficy2no44	86	71	of	of	IN
work_v44vkh6bzramngvcrficy2no44	86	72	input	input	NN
work_v44vkh6bzramngvcrficy2no44	86	73	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	86	74	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	86	75	get	get	VBP
work_v44vkh6bzramngvcrficy2no44	86	76	placed	place	VBN
work_v44vkh6bzramngvcrficy2no44	86	77	after	after	IN
work_v44vkh6bzramngvcrficy2no44	86	78	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	79	wrong	wrong	NN
work_v44vkh6bzramngvcrficy2no44	86	80	po-	po-	NN
work_v44vkh6bzramngvcrficy2no44	86	81	sition	sition	NN
work_v44vkh6bzramngvcrficy2no44	86	82	,	,	,
work_v44vkh6bzramngvcrficy2no44	86	83	while	while	IN
work_v44vkh6bzramngvcrficy2no44	86	84	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	86	85	use	use	VBP
work_v44vkh6bzramngvcrficy2no44	86	86	a	a	DT
work_v44vkh6bzramngvcrficy2no44	86	87	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	86	88	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	86	89	entropy	entropy	JJ
work_v44vkh6bzramngvcrficy2no44	86	90	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	86	91	trained	train	VBN
work_v44vkh6bzramngvcrficy2no44	86	92	by	by	IN
work_v44vkh6bzramngvcrficy2no44	86	93	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	86	94	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	86	95	likelihood	likelihood	NN
work_v44vkh6bzramngvcrficy2no44	86	96	;	;	:
work_v44vkh6bzramngvcrficy2no44	86	97	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	86	98	iii	iii	NNP
work_v44vkh6bzramngvcrficy2no44	86	99	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	86	100	they	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	86	101	use	use	VBP
work_v44vkh6bzramngvcrficy2no44	86	102	an	an	DT
work_v44vkh6bzramngvcrficy2no44	86	103	off	off	IN
work_v44vkh6bzramngvcrficy2no44	86	104	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	86	105	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	106	shelf	shelf	NN
work_v44vkh6bzramngvcrficy2no44	86	107	TSP	tsp	NN
work_v44vkh6bzramngvcrficy2no44	86	108	solver	solver	NN
work_v44vkh6bzramngvcrficy2no44	86	109	to	to	TO
work_v44vkh6bzramngvcrficy2no44	86	110	find	find	VB
work_v44vkh6bzramngvcrficy2no44	86	111	the	the	DT
work_v44vkh6bzramngvcrficy2no44	86	112	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	86	113	source	source	NN
work_v44vkh6bzramngvcrficy2no44	86	114	sen-	sen-	NN
work_v44vkh6bzramngvcrficy2no44	86	115	tence	tence	NN
work_v44vkh6bzramngvcrficy2no44	86	116	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	86	117	and	and	CC
work_v44vkh6bzramngvcrficy2no44	86	118	apply	apply	VB
work_v44vkh6bzramngvcrficy2no44	86	119	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	86	120	as	as	IN
work_v44vkh6bzramngvcrficy2no44	86	121	pre	pre	JJ
work_v44vkh6bzramngvcrficy2no44	86	122	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	86	123	processing	processing	NN
work_v44vkh6bzramngvcrficy2no44	86	124	to	to	IN
work_v44vkh6bzramngvcrficy2no44	86	125	training	training	NN
work_v44vkh6bzramngvcrficy2no44	86	126	and	and	CC
work_v44vkh6bzramngvcrficy2no44	86	127	test	test	NN
work_v44vkh6bzramngvcrficy2no44	86	128	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	86	129	.	.	.
work_v44vkh6bzramngvcrficy2no44	87	1	By	by	IN
work_v44vkh6bzramngvcrficy2no44	87	2	contrast	contrast	NN
work_v44vkh6bzramngvcrficy2no44	87	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	87	4	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	87	5	integrate	integrate	VBP
work_v44vkh6bzramngvcrficy2no44	87	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	87	7	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	87	8	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	87	9	entropy	entropy	JJ
work_v44vkh6bzramngvcrficy2no44	87	10	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	87	11	directly	directly	RB
work_v44vkh6bzramngvcrficy2no44	87	12	into	into	IN
work_v44vkh6bzramngvcrficy2no44	87	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	87	14	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	87	15	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	87	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	87	17	let	let	VB
work_v44vkh6bzramngvcrficy2no44	87	18	all	all	PDT
work_v44vkh6bzramngvcrficy2no44	87	19	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	87	20	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	87	21	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	87	22	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	87	23	phrase	phrase	VB
work_v44vkh6bzramngvcrficy2no44	87	24	orien-	orien-	NNP
work_v44vkh6bzramngvcrficy2no44	87	25	tation	tation	NN
work_v44vkh6bzramngvcrficy2no44	87	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	87	27	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	87	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	87	29	target	target	VB
work_v44vkh6bzramngvcrficy2no44	87	30	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	87	31	etc	etc	FW
work_v44vkh6bzramngvcrficy2no44	87	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	87	33	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	88	1	contribute	contribute	VB
work_v44vkh6bzramngvcrficy2no44	88	2	to	to	IN
work_v44vkh6bzramngvcrficy2no44	88	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	88	4	final	final	JJ
work_v44vkh6bzramngvcrficy2no44	88	5	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	88	6	decision	decision	NN
work_v44vkh6bzramngvcrficy2no44	88	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	89	1	3.1	3.1	CD
work_v44vkh6bzramngvcrficy2no44	89	2	Features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	89	3	Like	like	IN
work_v44vkh6bzramngvcrficy2no44	89	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	89	5	TSP	tsp	NN
work_v44vkh6bzramngvcrficy2no44	89	6	model	model	NN
work_v44vkh6bzramngvcrficy2no44	89	7	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	89	8	Visweswariah	Visweswariah	NNP
work_v44vkh6bzramngvcrficy2no44	89	9	et	et	FW
work_v44vkh6bzramngvcrficy2no44	89	10	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	89	11	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	89	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	89	13	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	89	14	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	89	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	89	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	89	17	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	89	18	model	model	NN
work_v44vkh6bzramngvcrficy2no44	89	19	builds	build	VBZ
work_v44vkh6bzramngvcrficy2no44	89	20	on	on	IN
work_v44vkh6bzramngvcrficy2no44	89	21	binary	binary	JJ
work_v44vkh6bzramngvcrficy2no44	89	22	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	89	23	similar	similar	JJ
work_v44vkh6bzramngvcrficy2no44	89	24	to	to	IN
work_v44vkh6bzramngvcrficy2no44	89	25	those	those	DT
work_v44vkh6bzramngvcrficy2no44	89	26	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	89	27	employed	employ	VBN
work_v44vkh6bzramngvcrficy2no44	89	28	for	for	IN
work_v44vkh6bzramngvcrficy2no44	89	29	dependency	dependency	NN
work_v44vkh6bzramngvcrficy2no44	89	30	parsing	parsing	NN
work_v44vkh6bzramngvcrficy2no44	89	31	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	89	32	McDonald	McDonald	NNP
work_v44vkh6bzramngvcrficy2no44	89	33	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	89	34	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	89	35	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	89	36	,	,	,
work_v44vkh6bzramngvcrficy2no44	89	37	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	89	38	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	89	39	:	:	:
work_v44vkh6bzramngvcrficy2no44	89	40	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	89	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	89	42	combinations	combination	NNS
work_v44vkh6bzramngvcrficy2no44	89	43	of	of	IN
work_v44vkh6bzramngvcrficy2no44	89	44	surface	surface	NN
work_v44vkh6bzramngvcrficy2no44	89	45	forms	form	NNS
work_v44vkh6bzramngvcrficy2no44	89	46	or	or	CC
work_v44vkh6bzramngvcrficy2no44	89	47	POS	POS	NNP
work_v44vkh6bzramngvcrficy2no44	89	48	tags	tag	NNS
work_v44vkh6bzramngvcrficy2no44	89	49	of	of	IN
work_v44vkh6bzramngvcrficy2no44	89	50	the	the	DT
work_v44vkh6bzramngvcrficy2no44	89	51	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	89	52	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	89	53	and	and	CC
work_v44vkh6bzramngvcrficy2no44	89	54	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	89	55	and	and	CC
work_v44vkh6bzramngvcrficy2no44	89	56	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	89	57	context	context	NN
work_v44vkh6bzramngvcrficy2no44	89	58	.	.	.
work_v44vkh6bzramngvcrficy2no44	90	1	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	90	2	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	90	3	templates	template	NNS
work_v44vkh6bzramngvcrficy2no44	90	4	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	90	5	presented	present	VBN
work_v44vkh6bzramngvcrficy2no44	90	6	in	in	IN
work_v44vkh6bzramngvcrficy2no44	90	7	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	90	8	1	1	CD
work_v44vkh6bzramngvcrficy2no44	90	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	91	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	91	2	main	main	JJ
work_v44vkh6bzramngvcrficy2no44	91	3	novelties	novelty	NNS
work_v44vkh6bzramngvcrficy2no44	91	4	with	with	IN
work_v44vkh6bzramngvcrficy2no44	91	5	respect	respect	NN
work_v44vkh6bzramngvcrficy2no44	91	6	to	to	IN
work_v44vkh6bzramngvcrficy2no44	91	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	91	8	TSP	tsp	NN
work_v44vkh6bzramngvcrficy2no44	91	9	model	model	NN
work_v44vkh6bzramngvcrficy2no44	91	10	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	91	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	91	12	mixed	mixed	JJ
work_v44vkh6bzramngvcrficy2no44	91	13	word	word	NN
work_v44vkh6bzramngvcrficy2no44	91	14	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	91	15	POS	pos	NN
work_v44vkh6bzramngvcrficy2no44	91	16	templates	template	NNS
work_v44vkh6bzramngvcrficy2no44	91	17	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	91	18	rows	row	VBZ
work_v44vkh6bzramngvcrficy2no44	91	19	16-	16-	NNP
work_v44vkh6bzramngvcrficy2no44	91	20	17	17	CD
work_v44vkh6bzramngvcrficy2no44	91	21	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	91	22	and	and	CC
work_v44vkh6bzramngvcrficy2no44	91	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	91	24	shallow	shallow	JJ
work_v44vkh6bzramngvcrficy2no44	91	25	syntax	syntax	NN
work_v44vkh6bzramngvcrficy2no44	91	26	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	91	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	92	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	92	2	particular	particular	JJ
work_v44vkh6bzramngvcrficy2no44	92	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	92	4	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	92	5	use	use	VBP
work_v44vkh6bzramngvcrficy2no44	92	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	92	7	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	92	8	types	type	NNS
work_v44vkh6bzramngvcrficy2no44	92	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	92	10	i	i	NNP
work_v44vkh6bzramngvcrficy2no44	92	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	92	12	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	92	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	92	14	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	92	15	context	context	NN
work_v44vkh6bzramngvcrficy2no44	92	16	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	92	17	18	18	CD
work_v44vkh6bzramngvcrficy2no44	92	18	-	-	SYM
work_v44vkh6bzramngvcrficy2no44	92	19	19	19	CD
work_v44vkh6bzramngvcrficy2no44	92	20	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	92	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	92	22	as	as	RB
work_v44vkh6bzramngvcrficy2no44	92	23	well	well	RB
work_v44vkh6bzramngvcrficy2no44	92	24	as	as	IN
work_v44vkh6bzramngvcrficy2no44	92	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	92	26	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	92	27	head	head	NN
work_v44vkh6bzramngvcrficy2no44	92	28	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	92	29	of	of	IN
work_v44vkh6bzramngvcrficy2no44	92	30	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	92	31	and	and	CC
work_v44vkh6bzramngvcrficy2no44	92	32	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	92	33	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	92	34	20	20	CD
work_v44vkh6bzramngvcrficy2no44	92	35	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	92	36	.	.	.
work_v44vkh6bzramngvcrficy2no44	93	1	Fi-	Fi-	NNP
work_v44vkh6bzramngvcrficy2no44	93	2	nally	nally	RB
work_v44vkh6bzramngvcrficy2no44	93	3	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	93	4	add	add	VBP
work_v44vkh6bzramngvcrficy2no44	93	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	93	6	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	93	7	to	to	TO
work_v44vkh6bzramngvcrficy2no44	93	8	indicate	indicate	VB
work_v44vkh6bzramngvcrficy2no44	93	9	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	93	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	93	11	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	93	12	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	93	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	93	14	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	93	15	belong	belong	VBP
work_v44vkh6bzramngvcrficy2no44	93	16	to	to	IN
work_v44vkh6bzramngvcrficy2no44	93	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	93	18	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	93	19	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	93	20	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	93	21	21	21	CD
work_v44vkh6bzramngvcrficy2no44	93	22	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	93	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	94	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	94	2	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	94	3	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	94	4	–	–	:
work_v44vkh6bzramngvcrficy2no44	94	5	forward	forward	RB
work_v44vkh6bzramngvcrficy2no44	94	6	/	/	SYM
work_v44vkh6bzramngvcrficy2no44	94	7	backward	backward	RB
work_v44vkh6bzramngvcrficy2no44	94	8	–	–	:
work_v44vkh6bzramngvcrficy2no44	94	9	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	94	10	included	include	VBN
work_v44vkh6bzramngvcrficy2no44	94	11	in	in	IN
work_v44vkh6bzramngvcrficy2no44	94	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	94	13	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	94	14	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	94	15	represent	represent	VBP
work_v44vkh6bzramngvcrficy2no44	94	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	94	17	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	94	18	comprised	comprise	VBN
work_v44vkh6bzramngvcrficy2no44	94	19	between	between	IN
work_v44vkh6bzramngvcrficy2no44	94	20	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	94	21	and	and	CC
work_v44vkh6bzramngvcrficy2no44	94	22	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	94	23	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	94	24	rows	row	VBZ
work_v44vkh6bzramngvcrficy2no44	94	25	6	6	CD
work_v44vkh6bzramngvcrficy2no44	94	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	94	27	7	7	CD
work_v44vkh6bzramngvcrficy2no44	94	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	94	29	14	14	CD
work_v44vkh6bzramngvcrficy2no44	94	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	94	31	15	15	CD
work_v44vkh6bzramngvcrficy2no44	94	32	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	94	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	95	1	No	no	DT
work_v44vkh6bzramngvcrficy2no44	95	2	explicit	explicit	JJ
work_v44vkh6bzramngvcrficy2no44	95	3	indication	indication	NN
work_v44vkh6bzramngvcrficy2no44	95	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	95	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	95	6	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	95	7	length	length	NN
work_v44vkh6bzramngvcrficy2no44	95	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	95	9	included	include	VBN
work_v44vkh6bzramngvcrficy2no44	95	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	95	11	any	any	DT
work_v44vkh6bzramngvcrficy2no44	95	12	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	95	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	96	1	3.2	3.2	CD
work_v44vkh6bzramngvcrficy2no44	96	2	Training	Training	NNP
work_v44vkh6bzramngvcrficy2no44	96	3	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	96	4	To	to	TO
work_v44vkh6bzramngvcrficy2no44	96	5	generate	generate	VB
work_v44vkh6bzramngvcrficy2no44	96	6	training	training	NN
work_v44vkh6bzramngvcrficy2no44	96	7	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	96	8	for	for	IN
work_v44vkh6bzramngvcrficy2no44	96	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	96	10	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	96	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	96	12	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	96	13	first	first	RB
work_v44vkh6bzramngvcrficy2no44	96	14	extract	extract	VBP
work_v44vkh6bzramngvcrficy2no44	96	15	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	96	16	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	96	17	from	from	IN
work_v44vkh6bzramngvcrficy2no44	96	18	a	a	DT
work_v44vkh6bzramngvcrficy2no44	96	19	word	word	NN
work_v44vkh6bzramngvcrficy2no44	96	20	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	96	21	aligned	align	VBN
work_v44vkh6bzramngvcrficy2no44	96	22	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	96	23	corpus	corpus	NN
work_v44vkh6bzramngvcrficy2no44	96	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	97	1	Given	give	VBN
work_v44vkh6bzramngvcrficy2no44	97	2	a	a	DT
work_v44vkh6bzramngvcrficy2no44	97	3	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	97	4	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	97	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	97	6	differ-	differ-	XX
work_v44vkh6bzramngvcrficy2no44	97	7	ent	ent	NNP
work_v44vkh6bzramngvcrficy2no44	97	8	heuristics	heuristic	NNS
work_v44vkh6bzramngvcrficy2no44	97	9	may	may	MD
work_v44vkh6bzramngvcrficy2no44	97	10	be	be	VB
work_v44vkh6bzramngvcrficy2no44	97	11	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	97	12	to	to	TO
work_v44vkh6bzramngvcrficy2no44	97	13	convert	convert	VB
work_v44vkh6bzramngvcrficy2no44	97	14	arbitrary	arbitrary	JJ
work_v44vkh6bzramngvcrficy2no44	97	15	word	word	NN
work_v44vkh6bzramngvcrficy2no44	97	16	alignments	alignment	NNS
work_v44vkh6bzramngvcrficy2no44	97	17	to	to	IN
work_v44vkh6bzramngvcrficy2no44	97	18	a	a	DT
work_v44vkh6bzramngvcrficy2no44	97	19	source	source	NN
work_v44vkh6bzramngvcrficy2no44	97	20	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	97	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	97	22	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	97	23	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	97	24	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	97	25	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	97	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	97	27	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	97	28	;	;	:
work_v44vkh6bzramngvcrficy2no44	97	29	Feng	Feng	NNP
work_v44vkh6bzramngvcrficy2no44	97	30	et	et	FW
work_v44vkh6bzramngvcrficy2no44	97	31	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	97	32	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	97	33	,	,	,
work_v44vkh6bzramngvcrficy2no44	97	34	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	97	35	;	;	:
work_v44vkh6bzramngvcrficy2no44	97	36	Visweswariah	Visweswariah	NNP
work_v44vkh6bzramngvcrficy2no44	97	37	et	et	FW
work_v44vkh6bzramngvcrficy2no44	97	38	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	97	39	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	97	40	,	,	,
work_v44vkh6bzramngvcrficy2no44	97	41	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	97	42	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	97	43	.	.	.
work_v44vkh6bzramngvcrficy2no44	98	1	Similarly	similarly	RB
work_v44vkh6bzramngvcrficy2no44	98	2	to	to	IN
work_v44vkh6bzramngvcrficy2no44	98	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	98	4	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	98	5	work	work	NN
work_v44vkh6bzramngvcrficy2no44	98	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	98	7	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	98	8	compute	compute	VBP
work_v44vkh6bzramngvcrficy2no44	98	9	for	for	IN
work_v44vkh6bzramngvcrficy2no44	98	10	each	each	DT
work_v44vkh6bzramngvcrficy2no44	98	11	source	source	NN
work_v44vkh6bzramngvcrficy2no44	98	12	word	word	NN
work_v44vkh6bzramngvcrficy2no44	98	13	fi	fi	IN
work_v44vkh6bzramngvcrficy2no44	98	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	15	mean	mean	NNP
work_v44vkh6bzramngvcrficy2no44	98	16	ai	ai	VBP
work_v44vkh6bzramngvcrficy2no44	98	17	of	of	IN
work_v44vkh6bzramngvcrficy2no44	98	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	19	target	target	NN
work_v44vkh6bzramngvcrficy2no44	98	20	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	98	21	aligned	align	VBN
work_v44vkh6bzramngvcrficy2no44	98	22	to	to	IN
work_v44vkh6bzramngvcrficy2no44	98	23	fi	fi	NN
work_v44vkh6bzramngvcrficy2no44	98	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	98	25	then	then	RB
work_v44vkh6bzramngvcrficy2no44	98	26	sort	sort	VB
work_v44vkh6bzramngvcrficy2no44	98	27	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	28	source	source	NN
work_v44vkh6bzramngvcrficy2no44	98	29	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	98	30	according	accord	VBG
work_v44vkh6bzramngvcrficy2no44	98	31	to	to	IN
work_v44vkh6bzramngvcrficy2no44	98	32	this	this	DT
work_v44vkh6bzramngvcrficy2no44	98	33	value.3	value.3	NN
work_v44vkh6bzramngvcrficy2no44	98	34	As	as	IN
work_v44vkh6bzramngvcrficy2no44	98	35	a	a	DT
work_v44vkh6bzramngvcrficy2no44	98	36	difference	difference	NN
work_v44vkh6bzramngvcrficy2no44	98	37	,	,	,
work_v44vkh6bzramngvcrficy2no44	98	38	though	though	RB
work_v44vkh6bzramngvcrficy2no44	98	39	,	,	,
work_v44vkh6bzramngvcrficy2no44	98	40	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	98	41	do	do	VBP
work_v44vkh6bzramngvcrficy2no44	98	42	not	not	RB
work_v44vkh6bzramngvcrficy2no44	98	43	discard	discard	VB
work_v44vkh6bzramngvcrficy2no44	98	44	unaligned	unaligned	JJ
work_v44vkh6bzramngvcrficy2no44	98	45	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	98	46	but	but	CC
work_v44vkh6bzramngvcrficy2no44	98	47	assign	assign	VB
work_v44vkh6bzramngvcrficy2no44	98	48	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	98	49	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	50	mean	mean	NNP
work_v44vkh6bzramngvcrficy2no44	98	51	3Using	3Using	NNP
work_v44vkh6bzramngvcrficy2no44	98	52	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	53	mean	mean	NN
work_v44vkh6bzramngvcrficy2no44	98	54	of	of	IN
work_v44vkh6bzramngvcrficy2no44	98	55	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	56	aligned	align	VBN
work_v44vkh6bzramngvcrficy2no44	98	57	indices	index	NNS
work_v44vkh6bzramngvcrficy2no44	98	58	makes	make	VBZ
work_v44vkh6bzramngvcrficy2no44	98	59	the	the	DT
work_v44vkh6bzramngvcrficy2no44	98	60	gener-	gener-	JJ
work_v44vkh6bzramngvcrficy2no44	98	61	ation	ation	NN
work_v44vkh6bzramngvcrficy2no44	98	62	of	of	IN
work_v44vkh6bzramngvcrficy2no44	98	63	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	98	64	permutations	permutation	NNS
work_v44vkh6bzramngvcrficy2no44	98	65	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	98	66	robust	robust	JJ
work_v44vkh6bzramngvcrficy2no44	98	67	to	to	TO
work_v44vkh6bzramngvcrficy2no44	98	68	alignment	alignment	VB
work_v44vkh6bzramngvcrficy2no44	98	69	er-	er-	JJ
work_v44vkh6bzramngvcrficy2no44	98	70	rors	ror	NNS
work_v44vkh6bzramngvcrficy2no44	98	71	.	.	.
work_v44vkh6bzramngvcrficy2no44	99	1	Admittedly	admittedly	RB
work_v44vkh6bzramngvcrficy2no44	99	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	99	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	99	4	heuristic	heuristic	JJ
work_v44vkh6bzramngvcrficy2no44	99	5	does	do	VBZ
work_v44vkh6bzramngvcrficy2no44	99	6	not	not	RB
work_v44vkh6bzramngvcrficy2no44	99	7	handle	handle	VB
work_v44vkh6bzramngvcrficy2no44	99	8	well	well	RB
work_v44vkh6bzramngvcrficy2no44	99	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	99	10	case	case	NN
work_v44vkh6bzramngvcrficy2no44	99	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	99	12	source	source	NN
work_v44vkh6bzramngvcrficy2no44	99	13	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	99	14	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	99	15	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	99	16	correctly	correctly	RB
work_v44vkh6bzramngvcrficy2no44	99	17	aligned	align	VBN
work_v44vkh6bzramngvcrficy2no44	99	18	to	to	IN
work_v44vkh6bzramngvcrficy2no44	99	19	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	99	20	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	99	21	consecutive	consecutive	JJ
work_v44vkh6bzramngvcrficy2no44	99	22	tar-	tar-	NN
work_v44vkh6bzramngvcrficy2no44	99	23	get	get	VB
work_v44vkh6bzramngvcrficy2no44	99	24	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	99	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	100	1	However	however	RB
work_v44vkh6bzramngvcrficy2no44	100	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	100	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	100	4	phenomenon	phenomenon	NN
work_v44vkh6bzramngvcrficy2no44	100	5	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	100	6	also	also	RB
work_v44vkh6bzramngvcrficy2no44	100	7	not	not	RB
work_v44vkh6bzramngvcrficy2no44	100	8	captured	capture	VBN
work_v44vkh6bzramngvcrficy2no44	100	9	by	by	IN
work_v44vkh6bzramngvcrficy2no44	100	10	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	100	11	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	100	12	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	100	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	100	14	who	who	WP
work_v44vkh6bzramngvcrficy2no44	100	15	only	only	RB
work_v44vkh6bzramngvcrficy2no44	100	16	learn	learn	VBP
work_v44vkh6bzramngvcrficy2no44	100	17	continuous	continuous	JJ
work_v44vkh6bzramngvcrficy2no44	100	18	phrases	phrase	NNS
work_v44vkh6bzramngvcrficy2no44	100	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	101	1	i!2	i!2	NNP
work_v44vkh6bzramngvcrficy2no44	101	2	i!1	i!1	NNP
work_v44vkh6bzramngvcrficy2no44	101	3	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	101	4	i+1	i+1	VBZ
work_v44vkh6bzramngvcrficy2no44	101	5	b	b	NNP
work_v44vkh6bzramngvcrficy2no44	101	6	j!1	j!1	NNP
work_v44vkh6bzramngvcrficy2no44	101	7	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	101	8	j+1	j+1	NNP
work_v44vkh6bzramngvcrficy2no44	101	9	1	1	CD
work_v44vkh6bzramngvcrficy2no44	101	10	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	11	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	12	2	2	NNP
work_v44vkh6bzramngvcrficy2no44	101	13	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	14	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	15	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	16	3	3	NNP
work_v44vkh6bzramngvcrficy2no44	101	17	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	18	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	19	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	20	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	21	4	4	CD
work_v44vkh6bzramngvcrficy2no44	101	22	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	23	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	24	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	25	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	26	5	5	NNP
work_v44vkh6bzramngvcrficy2no44	101	27	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	28	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	29	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	30	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	31	6	6	CD
work_v44vkh6bzramngvcrficy2no44	101	32	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	33	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	34	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	35	7	7	CD
work_v44vkh6bzramngvcrficy2no44	101	36	wall	wall	NNP
work_v44vkh6bzramngvcrficy2no44	101	37	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	38	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	39	8	8	CD
work_v44vkh6bzramngvcrficy2no44	101	40	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	41	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	42	9	9	CD
work_v44vkh6bzramngvcrficy2no44	101	43	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	44	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	45	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	46	10	10	CD
work_v44vkh6bzramngvcrficy2no44	101	47	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	48	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	49	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	50	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	51	11	11	CD
work_v44vkh6bzramngvcrficy2no44	101	52	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	53	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	54	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	55	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	56	12	12	CD
work_v44vkh6bzramngvcrficy2no44	101	57	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	58	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	59	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	60	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	61	13	13	CD
work_v44vkh6bzramngvcrficy2no44	101	62	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	63	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	64	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	65	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	66	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	67	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	68	14	14	CD
work_v44vkh6bzramngvcrficy2no44	101	69	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	70	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	71	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	72	15	15	CD
work_v44vkh6bzramngvcrficy2no44	101	73	pall	pall	NN
work_v44vkh6bzramngvcrficy2no44	101	74	p	p	NNP
work_v44vkh6bzramngvcrficy2no44	101	75	p	p	NNP
work_v44vkh6bzramngvcrficy2no44	101	76	16	16	CD
work_v44vkh6bzramngvcrficy2no44	101	77	w	w	NN
work_v44vkh6bzramngvcrficy2no44	101	78	p	p	NNP
work_v44vkh6bzramngvcrficy2no44	101	79	17	17	CD
work_v44vkh6bzramngvcrficy2no44	101	80	p	p	NN
work_v44vkh6bzramngvcrficy2no44	101	81	w	w	NNP
work_v44vkh6bzramngvcrficy2no44	101	82	18	18	CD
work_v44vkh6bzramngvcrficy2no44	101	83	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	84	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	85	19	19	CD
work_v44vkh6bzramngvcrficy2no44	101	86	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	87	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	88	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	89	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	90	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	91	c	c	NN
work_v44vkh6bzramngvcrficy2no44	101	92	20	20	CD
work_v44vkh6bzramngvcrficy2no44	101	93	h	h	NN
work_v44vkh6bzramngvcrficy2no44	101	94	h	h	NN
work_v44vkh6bzramngvcrficy2no44	101	95	21	21	CD
work_v44vkh6bzramngvcrficy2no44	101	96	belong	belong	NN
work_v44vkh6bzramngvcrficy2no44	101	97	to	to	IN
work_v44vkh6bzramngvcrficy2no44	101	98	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	101	99	chunk(i	chunk(i	NNP
work_v44vkh6bzramngvcrficy2no44	101	100	,	,	,
work_v44vkh6bzramngvcrficy2no44	101	101	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	101	102	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	101	103	?	?	.
work_v44vkh6bzramngvcrficy2no44	102	1	w	w	LS
work_v44vkh6bzramngvcrficy2no44	102	2	:	:	:
work_v44vkh6bzramngvcrficy2no44	102	3	word	word	NN
work_v44vkh6bzramngvcrficy2no44	102	4	identity	identity	NN
work_v44vkh6bzramngvcrficy2no44	102	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	102	6	p	p	NN
work_v44vkh6bzramngvcrficy2no44	102	7	:	:	:
work_v44vkh6bzramngvcrficy2no44	102	8	POS	POS	NNP
work_v44vkh6bzramngvcrficy2no44	102	9	tag	tag	NN
work_v44vkh6bzramngvcrficy2no44	102	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	102	11	c	c	NN
work_v44vkh6bzramngvcrficy2no44	102	12	:	:	:
work_v44vkh6bzramngvcrficy2no44	102	13	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	102	14	type	type	NN
work_v44vkh6bzramngvcrficy2no44	102	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	102	16	h	h	NN
work_v44vkh6bzramngvcrficy2no44	102	17	:	:	:
work_v44vkh6bzramngvcrficy2no44	102	18	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	102	19	head	head	NN
work_v44vkh6bzramngvcrficy2no44	102	20	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	102	21	1	1	CD
work_v44vkh6bzramngvcrficy2no44	102	22	:	:	:
work_v44vkh6bzramngvcrficy2no44	102	23	Feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	102	24	templates	template	NNS
work_v44vkh6bzramngvcrficy2no44	102	25	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	102	26	to	to	TO
work_v44vkh6bzramngvcrficy2no44	102	27	learn	learn	VB
work_v44vkh6bzramngvcrficy2no44	102	28	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	102	29	a	a	DT
work_v44vkh6bzramngvcrficy2no44	102	30	source	source	NN
work_v44vkh6bzramngvcrficy2no44	102	31	position	position	NN
work_v44vkh6bzramngvcrficy2no44	102	32	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	102	33	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	102	34	to	to	TO
work_v44vkh6bzramngvcrficy2no44	102	35	be	be	VB
work_v44vkh6bzramngvcrficy2no44	102	36	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	102	37	right	right	RB
work_v44vkh6bzramngvcrficy2no44	102	38	after	after	IN
work_v44vkh6bzramngvcrficy2no44	102	39	i.	i.	NN
work_v44vkh6bzramngvcrficy2no44	103	1	Positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	103	2	com-	com-	NN
work_v44vkh6bzramngvcrficy2no44	103	3	prised	prise	VBD
work_v44vkh6bzramngvcrficy2no44	103	4	between	between	IN
work_v44vkh6bzramngvcrficy2no44	103	5	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	103	6	and	and	CC
work_v44vkh6bzramngvcrficy2no44	103	7	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	103	8	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	103	9	denoted	denote	VBN
work_v44vkh6bzramngvcrficy2no44	103	10	by	by	IN
work_v44vkh6bzramngvcrficy2no44	103	11	b	b	NN
work_v44vkh6bzramngvcrficy2no44	103	12	and	and	CC
work_v44vkh6bzramngvcrficy2no44	103	13	generate	generate	VB
work_v44vkh6bzramngvcrficy2no44	103	14	two	two	CD
work_v44vkh6bzramngvcrficy2no44	103	15	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	103	16	templates	template	NNS
work_v44vkh6bzramngvcrficy2no44	103	17	:	:	:
work_v44vkh6bzramngvcrficy2no44	103	18	one	one	CD
work_v44vkh6bzramngvcrficy2no44	103	19	for	for	IN
work_v44vkh6bzramngvcrficy2no44	103	20	each	each	DT
work_v44vkh6bzramngvcrficy2no44	103	21	position	position	NN
work_v44vkh6bzramngvcrficy2no44	103	22	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	103	23	6	6	CD
work_v44vkh6bzramngvcrficy2no44	103	24	and	and	CC
work_v44vkh6bzramngvcrficy2no44	103	25	14	14	CD
work_v44vkh6bzramngvcrficy2no44	103	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	103	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	103	28	one	one	CD
work_v44vkh6bzramngvcrficy2no44	103	29	for	for	IN
work_v44vkh6bzramngvcrficy2no44	103	30	the	the	DT
work_v44vkh6bzramngvcrficy2no44	103	31	concatentation	concatentation	NN
work_v44vkh6bzramngvcrficy2no44	103	32	of	of	IN
work_v44vkh6bzramngvcrficy2no44	103	33	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	103	34	all	all	DT
work_v44vkh6bzramngvcrficy2no44	103	35	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	103	36	7	7	CD
work_v44vkh6bzramngvcrficy2no44	103	37	and	and	CC
work_v44vkh6bzramngvcrficy2no44	103	38	15	15	CD
work_v44vkh6bzramngvcrficy2no44	103	39	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	103	40	.	.	.
work_v44vkh6bzramngvcrficy2no44	104	1	of	of	IN
work_v44vkh6bzramngvcrficy2no44	104	2	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	104	3	neighbouring	neighbouring	NN
work_v44vkh6bzramngvcrficy2no44	104	4	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	104	5	’	'	''
work_v44vkh6bzramngvcrficy2no44	104	6	alignment	alignment	NN
work_v44vkh6bzramngvcrficy2no44	104	7	means	mean	VBZ
work_v44vkh6bzramngvcrficy2no44	104	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	104	9	so	so	IN
work_v44vkh6bzramngvcrficy2no44	104	10	that	that	IN
work_v44vkh6bzramngvcrficy2no44	104	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	104	12	complete	complete	JJ
work_v44vkh6bzramngvcrficy2no44	104	13	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	104	14	of	of	IN
work_v44vkh6bzramngvcrficy2no44	104	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	104	16	source	source	NN
work_v44vkh6bzramngvcrficy2no44	104	17	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	104	18	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	104	19	")	")	``
work_v44vkh6bzramngvcrficy2no44	104	20	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	104	21	obtained	obtain	VBN
work_v44vkh6bzramngvcrficy2no44	104	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	105	1	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	105	2	2(a	2(a	CD
work_v44vkh6bzramngvcrficy2no44	105	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	105	4	illustrates	illustrate	VBZ
work_v44vkh6bzramngvcrficy2no44	105	5	this	this	DT
work_v44vkh6bzramngvcrficy2no44	105	6	procedure	procedure	NN
work_v44vkh6bzramngvcrficy2no44	105	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	106	1	Given	give	VBN
work_v44vkh6bzramngvcrficy2no44	106	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	106	3	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	106	4	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	106	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	106	6	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	106	7	then	then	RB
work_v44vkh6bzramngvcrficy2no44	106	8	gener-	gener-	VBP
work_v44vkh6bzramngvcrficy2no44	106	9	ate	eat	VBD
work_v44vkh6bzramngvcrficy2no44	106	10	positive	positive	JJ
work_v44vkh6bzramngvcrficy2no44	106	11	and	and	CC
work_v44vkh6bzramngvcrficy2no44	106	12	negative	negative	JJ
work_v44vkh6bzramngvcrficy2no44	106	13	training	training	NN
work_v44vkh6bzramngvcrficy2no44	106	14	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	106	15	by	by	IN
work_v44vkh6bzramngvcrficy2no44	106	16	simu-	simu-	NN
work_v44vkh6bzramngvcrficy2no44	106	17	lating	lating	NN
work_v44vkh6bzramngvcrficy2no44	106	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	106	19	decoding	decoding	JJ
work_v44vkh6bzramngvcrficy2no44	106	20	process	process	NN
work_v44vkh6bzramngvcrficy2no44	106	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	107	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	107	2	traverse	traverse	VBP
work_v44vkh6bzramngvcrficy2no44	107	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	107	4	source	source	NN
work_v44vkh6bzramngvcrficy2no44	107	5	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	107	6	in	in	IN
work_v44vkh6bzramngvcrficy2no44	107	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	107	8	order	order	NN
work_v44vkh6bzramngvcrficy2no44	107	9	defined	define	VBN
work_v44vkh6bzramngvcrficy2no44	107	10	by	by	IN
work_v44vkh6bzramngvcrficy2no44	107	11	"	"	''
work_v44vkh6bzramngvcrficy2no44	107	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	107	13	keeping	keep	VBG
work_v44vkh6bzramngvcrficy2no44	107	14	track	track	NN
work_v44vkh6bzramngvcrficy2no44	107	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	107	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	107	17	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	107	18	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	107	19	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	107	20	already	already	RB
work_v44vkh6bzramngvcrficy2no44	107	21	been	be	VBN
work_v44vkh6bzramngvcrficy2no44	107	22	covered	cover	VBN
work_v44vkh6bzramngvcrficy2no44	107	23	and	and	CC
work_v44vkh6bzramngvcrficy2no44	107	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	107	25	for	for	IN
work_v44vkh6bzramngvcrficy2no44	107	26	each	each	DT
work_v44vkh6bzramngvcrficy2no44	107	27	t	t	NN
work_v44vkh6bzramngvcrficy2no44	107	28	:	:	:
work_v44vkh6bzramngvcrficy2no44	107	29	1	1	CD
work_v44vkh6bzramngvcrficy2no44	107	30	"	"	``
work_v44vkh6bzramngvcrficy2no44	107	31	t	t	NNP
work_v44vkh6bzramngvcrficy2no44	107	32	"	"	''
work_v44vkh6bzramngvcrficy2no44	107	33	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	107	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	107	35	generate	generate	NN
work_v44vkh6bzramngvcrficy2no44	107	36	:	:	:
work_v44vkh6bzramngvcrficy2no44	107	37	•	•	VB
work_v44vkh6bzramngvcrficy2no44	107	38	one	one	CD
work_v44vkh6bzramngvcrficy2no44	107	39	positive	positive	JJ
work_v44vkh6bzramngvcrficy2no44	107	40	sample	sample	NN
work_v44vkh6bzramngvcrficy2no44	107	41	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	107	42	R!t,!t+1	R!t,!t+1	NNP
work_v44vkh6bzramngvcrficy2no44	107	43	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	107	44	1	1	CD
work_v44vkh6bzramngvcrficy2no44	107	45	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	107	46	for	for	IN
work_v44vkh6bzramngvcrficy2no44	107	47	the	the	DT
work_v44vkh6bzramngvcrficy2no44	107	48	source	source	NN
work_v44vkh6bzramngvcrficy2no44	107	49	position	position	NN
work_v44vkh6bzramngvcrficy2no44	107	50	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	107	51	comes	come	VBZ
work_v44vkh6bzramngvcrficy2no44	107	52	right	right	RB
work_v44vkh6bzramngvcrficy2no44	107	53	after	after	IN
work_v44vkh6bzramngvcrficy2no44	107	54	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	107	55	,	,	,
work_v44vkh6bzramngvcrficy2no44	107	56	•	•	VB
work_v44vkh6bzramngvcrficy2no44	107	57	a	a	DT
work_v44vkh6bzramngvcrficy2no44	107	58	negative	negative	JJ
work_v44vkh6bzramngvcrficy2no44	107	59	sample	sample	NN
work_v44vkh6bzramngvcrficy2no44	107	60	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	107	61	R!t	R!t	NNP
work_v44vkh6bzramngvcrficy2no44	107	62	,	,	,
work_v44vkh6bzramngvcrficy2no44	107	63	u=0	u=0	NNS
work_v44vkh6bzramngvcrficy2no44	107	64	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	107	65	for	for	IN
work_v44vkh6bzramngvcrficy2no44	107	66	each	each	DT
work_v44vkh6bzramngvcrficy2no44	107	67	source	source	NN
work_v44vkh6bzramngvcrficy2no44	107	68	position	position	NN
work_v44vkh6bzramngvcrficy2no44	107	69	in	in	IN
work_v44vkh6bzramngvcrficy2no44	107	70	{	{	-LRB-
work_v44vkh6bzramngvcrficy2no44	107	71	u	u	NN
work_v44vkh6bzramngvcrficy2no44	107	72	:	:	:
work_v44vkh6bzramngvcrficy2no44	107	73	"	"	``
work_v44vkh6bzramngvcrficy2no44	107	74	t!#+1	t!#+1	NNP
work_v44vkh6bzramngvcrficy2no44	107	75	<	<	XX
work_v44vkh6bzramngvcrficy2no44	107	76	u	u	NNP
work_v44vkh6bzramngvcrficy2no44	107	77	<	<	XX
work_v44vkh6bzramngvcrficy2no44	107	78	"	"	``
work_v44vkh6bzramngvcrficy2no44	107	79	t+#+1	t+#+1	VBG
work_v44vkh6bzramngvcrficy2no44	107	80	#	#	$
work_v44vkh6bzramngvcrficy2no44	107	81	u	u	NNP
work_v44vkh6bzramngvcrficy2no44	107	82	$	$	$
work_v44vkh6bzramngvcrficy2no44	107	83	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	107	84	"	"	``
work_v44vkh6bzramngvcrficy2no44	107	85	t+1	t+1	NN
work_v44vkh6bzramngvcrficy2no44	107	86	}	}	-RRB-
work_v44vkh6bzramngvcrficy2no44	107	87	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	107	88	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	107	89	not	not	RB
work_v44vkh6bzramngvcrficy2no44	107	90	yet	yet	RB
work_v44vkh6bzramngvcrficy2no44	107	91	been	be	VBN
work_v44vkh6bzramngvcrficy2no44	107	92	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	107	93	.	.	.
work_v44vkh6bzramngvcrficy2no44	108	1	Here	here	RB
work_v44vkh6bzramngvcrficy2no44	108	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	108	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	108	4	sampling	sampling	NN
work_v44vkh6bzramngvcrficy2no44	108	5	window	window	NN
work_v44vkh6bzramngvcrficy2no44	108	6	#	#	NNP
work_v44vkh6bzramngvcrficy2no44	108	7	serves	serve	VBZ
work_v44vkh6bzramngvcrficy2no44	108	8	to	to	TO
work_v44vkh6bzramngvcrficy2no44	108	9	control	control	VB
work_v44vkh6bzramngvcrficy2no44	108	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	108	11	size	size	NN
work_v44vkh6bzramngvcrficy2no44	108	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	108	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	108	14	training	training	NN
work_v44vkh6bzramngvcrficy2no44	108	15	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	108	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	108	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	108	18	proportion	proportion	NN
work_v44vkh6bzramngvcrficy2no44	108	19	between	between	IN
work_v44vkh6bzramngvcrficy2no44	108	20	positive	positive	JJ
work_v44vkh6bzramngvcrficy2no44	108	21	and	and	CC
work_v44vkh6bzramngvcrficy2no44	108	22	negative	negative	JJ
work_v44vkh6bzramngvcrficy2no44	108	23	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	108	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	109	1	Its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	109	2	value	value	NN
work_v44vkh6bzramngvcrficy2no44	109	3	naturally	naturally	RB
work_v44vkh6bzramngvcrficy2no44	109	4	correlates	correlate	VBZ
work_v44vkh6bzramngvcrficy2no44	109	5	with	with	IN
work_v44vkh6bzramngvcrficy2no44	109	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	109	7	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	109	8	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	109	9	in	in	IN
work_v44vkh6bzramngvcrficy2no44	109	10	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	109	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	110	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	110	2	gener-	gener-	JJ
work_v44vkh6bzramngvcrficy2no44	110	3	ation	ation	NN
work_v44vkh6bzramngvcrficy2no44	110	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	110	5	training	training	NN
work_v44vkh6bzramngvcrficy2no44	110	6	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	110	7	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	110	8	illustrated	illustrate	VBN
work_v44vkh6bzramngvcrficy2no44	110	9	by	by	IN
work_v44vkh6bzramngvcrficy2no44	110	10	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	110	11	2(b	2(b	CD
work_v44vkh6bzramngvcrficy2no44	110	12	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	110	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	111	1	330	330	CD
work_v44vkh6bzramngvcrficy2no44	111	2	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	111	3	a	a	LS
work_v44vkh6bzramngvcrficy2no44	111	4	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	111	5	Converting	convert	VBG
work_v44vkh6bzramngvcrficy2no44	111	6	word	word	NN
work_v44vkh6bzramngvcrficy2no44	111	7	alignments	alignment	VBZ
work_v44vkh6bzramngvcrficy2no44	111	8	to	to	IN
work_v44vkh6bzramngvcrficy2no44	111	9	a	a	DT
work_v44vkh6bzramngvcrficy2no44	111	10	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	111	11	:	:	:
work_v44vkh6bzramngvcrficy2no44	111	12	source	source	NN
work_v44vkh6bzramngvcrficy2no44	111	13	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	111	14	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	111	15	sorted	sort	VBN
work_v44vkh6bzramngvcrficy2no44	111	16	by	by	IN
work_v44vkh6bzramngvcrficy2no44	111	17	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	111	18	target	target	NN
work_v44vkh6bzramngvcrficy2no44	111	19	alignments	alignment	NNS
work_v44vkh6bzramngvcrficy2no44	111	20	mean	mean	VBP
work_v44vkh6bzramngvcrficy2no44	111	21	a.	a.	NN
work_v44vkh6bzramngvcrficy2no44	112	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	112	2	unaligned	unaligned	JJ
work_v44vkh6bzramngvcrficy2no44	112	3	word	word	NN
work_v44vkh6bzramngvcrficy2no44	112	4	“	"	``
work_v44vkh6bzramngvcrficy2no44	112	5	D	d	NN
work_v44vkh6bzramngvcrficy2no44	112	6	”	"	''
work_v44vkh6bzramngvcrficy2no44	112	7	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	112	8	assigned	assign	VBN
work_v44vkh6bzramngvcrficy2no44	112	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	112	10	mean	mean	NN
work_v44vkh6bzramngvcrficy2no44	112	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	112	12	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	112	13	neighbouring	neighbouring	NN
work_v44vkh6bzramngvcrficy2no44	112	14	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	112	15	’	'	''
work_v44vkh6bzramngvcrficy2no44	112	16	a	a	DT
work_v44vkh6bzramngvcrficy2no44	112	17	values	value	NNS
work_v44vkh6bzramngvcrficy2no44	112	18	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	112	19	2	2	CD
work_v44vkh6bzramngvcrficy2no44	112	20	+	+	SYM
work_v44vkh6bzramngvcrficy2no44	112	21	5)/2	5)/2	CD
work_v44vkh6bzramngvcrficy2no44	112	22	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	112	23	3.5	3.5	CD
work_v44vkh6bzramngvcrficy2no44	112	24	:	:	:
work_v44vkh6bzramngvcrficy2no44	112	25	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	112	26	b	b	NN
work_v44vkh6bzramngvcrficy2no44	112	27	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	112	28	Generating	generate	VBG
work_v44vkh6bzramngvcrficy2no44	112	29	binary	binary	JJ
work_v44vkh6bzramngvcrficy2no44	112	30	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	112	31	by	by	IN
work_v44vkh6bzramngvcrficy2no44	112	32	simulating	simulate	VBG
work_v44vkh6bzramngvcrficy2no44	112	33	the	the	DT
work_v44vkh6bzramngvcrficy2no44	112	34	decoding	decoding	JJ
work_v44vkh6bzramngvcrficy2no44	112	35	process	process	NN
work_v44vkh6bzramngvcrficy2no44	112	36	:	:	:
work_v44vkh6bzramngvcrficy2no44	112	37	shaded	shaded	JJ
work_v44vkh6bzramngvcrficy2no44	112	38	rounds	round	NNS
work_v44vkh6bzramngvcrficy2no44	112	39	represent	represent	VBP
work_v44vkh6bzramngvcrficy2no44	112	40	cov-	cov-	JJ
work_v44vkh6bzramngvcrficy2no44	112	41	ered	ered	JJ
work_v44vkh6bzramngvcrficy2no44	112	42	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	112	43	,	,	,
work_v44vkh6bzramngvcrficy2no44	112	44	while	while	IN
work_v44vkh6bzramngvcrficy2no44	112	45	dashed	dash	VBN
work_v44vkh6bzramngvcrficy2no44	112	46	arrows	arrow	NNS
work_v44vkh6bzramngvcrficy2no44	112	47	represent	represent	VBP
work_v44vkh6bzramngvcrficy2no44	112	48	negative	negative	JJ
work_v44vkh6bzramngvcrficy2no44	112	49	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	112	50	:	:	:
work_v44vkh6bzramngvcrficy2no44	112	51	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	112	52	2	2	CD
work_v44vkh6bzramngvcrficy2no44	112	53	:	:	:
work_v44vkh6bzramngvcrficy2no44	112	54	The	the	DT
work_v44vkh6bzramngvcrficy2no44	112	55	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	112	56	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	112	57	training	training	NN
work_v44vkh6bzramngvcrficy2no44	112	58	data	data	NN
work_v44vkh6bzramngvcrficy2no44	112	59	generation	generation	NN
work_v44vkh6bzramngvcrficy2no44	112	60	process	process	NN
work_v44vkh6bzramngvcrficy2no44	112	61	.	.	.
work_v44vkh6bzramngvcrficy2no44	113	1	3.3	3.3	CD
work_v44vkh6bzramngvcrficy2no44	113	2	Integration	integration	NN
work_v44vkh6bzramngvcrficy2no44	113	3	into	into	IN
work_v44vkh6bzramngvcrficy2no44	113	4	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	113	5	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	113	6	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	113	7	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	113	8	Rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	113	9	than	than	IN
work_v44vkh6bzramngvcrficy2no44	113	10	using	use	VBG
work_v44vkh6bzramngvcrficy2no44	113	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	113	12	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	113	13	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	113	14	model	model	NN
work_v44vkh6bzramngvcrficy2no44	113	15	for	for	IN
work_v44vkh6bzramngvcrficy2no44	113	16	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	113	17	pre	pre	NN
work_v44vkh6bzramngvcrficy2no44	113	18	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	113	19	processing	processing	NN
work_v44vkh6bzramngvcrficy2no44	113	20	as	as	IN
work_v44vkh6bzramngvcrficy2no44	113	21	done	do	VBN
work_v44vkh6bzramngvcrficy2no44	113	22	by	by	IN
work_v44vkh6bzramngvcrficy2no44	113	23	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	113	24	Visweswariah	Visweswariah	NNP
work_v44vkh6bzramngvcrficy2no44	113	25	et	et	FW
work_v44vkh6bzramngvcrficy2no44	113	26	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	113	27	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	113	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	113	29	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	113	30	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	113	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	113	32	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	113	33	directly	directly	RB
work_v44vkh6bzramngvcrficy2no44	113	34	integrate	integrate	VBP
work_v44vkh6bzramngvcrficy2no44	113	35	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	113	36	into	into	IN
work_v44vkh6bzramngvcrficy2no44	113	37	the	the	DT
work_v44vkh6bzramngvcrficy2no44	113	38	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	113	39	de-	de-	XX
work_v44vkh6bzramngvcrficy2no44	113	40	coder	coder	VBP
work_v44vkh6bzramngvcrficy2no44	113	41	Moses	Moses	NNP
work_v44vkh6bzramngvcrficy2no44	113	42	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	113	43	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	113	44	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	113	45	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	113	46	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	113	47	,	,	,
work_v44vkh6bzramngvcrficy2no44	113	48	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	113	49	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	113	50	.	.	.
work_v44vkh6bzramngvcrficy2no44	114	1	Two	two	CD
work_v44vkh6bzramngvcrficy2no44	114	2	main	main	JJ
work_v44vkh6bzramngvcrficy2no44	114	3	computation	computation	NN
work_v44vkh6bzramngvcrficy2no44	114	4	phases	phase	NNS
work_v44vkh6bzramngvcrficy2no44	114	5	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	114	6	required	require	VBN
work_v44vkh6bzramngvcrficy2no44	114	7	by	by	IN
work_v44vkh6bzramngvcrficy2no44	114	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	114	9	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	114	10	model	model	NN
work_v44vkh6bzramngvcrficy2no44	114	11	:	:	:
work_v44vkh6bzramngvcrficy2no44	114	12	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	114	13	i	i	NN
work_v44vkh6bzramngvcrficy2no44	114	14	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	114	15	at	at	IN
work_v44vkh6bzramngvcrficy2no44	114	16	system	system	NN
work_v44vkh6bzramngvcrficy2no44	114	17	initialization	initialization	NN
work_v44vkh6bzramngvcrficy2no44	114	18	time	time	NN
work_v44vkh6bzramngvcrficy2no44	114	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	114	20	all	all	DT
work_v44vkh6bzramngvcrficy2no44	114	21	fea-	fea-	NNS
work_v44vkh6bzramngvcrficy2no44	114	22	ture	ture	NN
work_v44vkh6bzramngvcrficy2no44	114	23	weights	weight	NNS
work_v44vkh6bzramngvcrficy2no44	114	24	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	114	25	loaded	load	VBN
work_v44vkh6bzramngvcrficy2no44	114	26	into	into	IN
work_v44vkh6bzramngvcrficy2no44	114	27	memory	memory	NN
work_v44vkh6bzramngvcrficy2no44	114	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	114	29	and	and	CC
work_v44vkh6bzramngvcrficy2no44	114	30	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	114	31	ii	ii	NN
work_v44vkh6bzramngvcrficy2no44	114	32	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	114	33	before	before	IN
work_v44vkh6bzramngvcrficy2no44	114	34	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	114	35	each	each	DT
work_v44vkh6bzramngvcrficy2no44	114	36	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	114	37	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	114	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	114	39	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	114	40	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	114	41	extracted	extract	VBN
work_v44vkh6bzramngvcrficy2no44	114	42	from	from	IN
work_v44vkh6bzramngvcrficy2no44	114	43	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	114	44	and	and	CC
work_v44vkh6bzramngvcrficy2no44	114	45	model	model	NN
work_v44vkh6bzramngvcrficy2no44	114	46	probabilities	probability	NNS
work_v44vkh6bzramngvcrficy2no44	114	47	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	114	48	pre	pre	VBN
work_v44vkh6bzramngvcrficy2no44	114	49	-	-	VBN
work_v44vkh6bzramngvcrficy2no44	114	50	computed	compute	VBN
work_v44vkh6bzramngvcrficy2no44	114	51	for	for	IN
work_v44vkh6bzramngvcrficy2no44	114	52	each	each	DT
work_v44vkh6bzramngvcrficy2no44	114	53	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	114	54	of	of	IN
work_v44vkh6bzramngvcrficy2no44	114	55	source	source	NN
work_v44vkh6bzramngvcrficy2no44	114	56	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	114	57	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	114	58	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	114	59	,	,	,
work_v44vkh6bzramngvcrficy2no44	114	60	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	114	61	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	114	62	such	such	JJ
work_v44vkh6bzramngvcrficy2no44	114	63	that	that	IN
work_v44vkh6bzramngvcrficy2no44	114	64	|j	|j	NNP
work_v44vkh6bzramngvcrficy2no44	114	65	!	!	.
work_v44vkh6bzramngvcrficy2no44	115	1	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	115	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	116	1	1|	1|	CD
work_v44vkh6bzramngvcrficy2no44	116	2	"	"	``
work_v44vkh6bzramngvcrficy2no44	116	3	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	116	4	.	.	.
work_v44vkh6bzramngvcrficy2no44	117	1	Note	note	VB
work_v44vkh6bzramngvcrficy2no44	117	2	that	that	IN
work_v44vkh6bzramngvcrficy2no44	117	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	117	4	efficient	efficient	JJ
work_v44vkh6bzramngvcrficy2no44	117	5	solution	solution	NN
work_v44vkh6bzramngvcrficy2no44	117	6	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	117	7	possible	possible	JJ
work_v44vkh6bzramngvcrficy2no44	117	8	because	because	IN
work_v44vkh6bzramngvcrficy2no44	117	9	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	117	10	model	model	NN
work_v44vkh6bzramngvcrficy2no44	117	11	does	do	VBZ
work_v44vkh6bzramngvcrficy2no44	117	12	not	not	RB
work_v44vkh6bzramngvcrficy2no44	117	13	employ	employ	VB
work_v44vkh6bzramngvcrficy2no44	117	14	de-	de-	RB
work_v44vkh6bzramngvcrficy2no44	117	15	coding	code	VBG
work_v44vkh6bzramngvcrficy2no44	117	16	history	history	NN
work_v44vkh6bzramngvcrficy2no44	117	17	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	117	18	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	117	19	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	117	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	117	21	like	like	IN
work_v44vkh6bzramngvcrficy2no44	117	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	117	23	word	word	NN
work_v44vkh6bzramngvcrficy2no44	117	24	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	117	25	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	117	26	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	117	27	before	before	IN
work_v44vkh6bzramngvcrficy2no44	117	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	117	29	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	117	30	one	one	NN
work_v44vkh6bzramngvcrficy2no44	117	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	117	32	or	or	CC
work_v44vkh6bzramngvcrficy2no44	117	33	like	like	IN
work_v44vkh6bzramngvcrficy2no44	117	34	the	the	DT
work_v44vkh6bzramngvcrficy2no44	117	35	previous	previous	JJ
work_v44vkh6bzramngvcrficy2no44	117	36	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	117	37	legth	legth	NN
work_v44vkh6bzramngvcrficy2no44	117	38	.	.	.
work_v44vkh6bzramngvcrficy2no44	118	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	118	2	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	118	3	an	an	DT
work_v44vkh6bzramngvcrficy2no44	118	4	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	118	5	difference	difference	NN
work_v44vkh6bzramngvcrficy2no44	118	6	with	with	IN
work_v44vkh6bzramngvcrficy2no44	118	7	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	118	8	spect	spect	NN
work_v44vkh6bzramngvcrficy2no44	118	9	to	to	IN
work_v44vkh6bzramngvcrficy2no44	118	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	118	11	reordered	reorder	VBN
work_v44vkh6bzramngvcrficy2no44	118	12	source	source	NN
work_v44vkh6bzramngvcrficy2no44	118	13	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	118	14	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	118	15	by	by	IN
work_v44vkh6bzramngvcrficy2no44	118	16	Feng	Feng	NNP
work_v44vkh6bzramngvcrficy2no44	118	17	et	et	FW
work_v44vkh6bzramngvcrficy2no44	118	18	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	118	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	119	1	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	119	2	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	119	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	119	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	119	5	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	119	6	requires	require	VBZ
work_v44vkh6bzramngvcrficy2no44	119	7	inclusion	inclusion	NN
work_v44vkh6bzramngvcrficy2no44	119	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	119	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	119	10	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	119	11	n	n	JJ
work_v44vkh6bzramngvcrficy2no44	119	12	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	119	13	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	119	14	in	in	IN
work_v44vkh6bzramngvcrficy2no44	119	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	119	16	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	119	17	state	state	NN
work_v44vkh6bzramngvcrficy2no44	119	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	120	1	Fig	fig	NN
work_v44vkh6bzramngvcrficy2no44	120	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	121	1	1	1	CD
work_v44vkh6bzramngvcrficy2no44	121	2	illustrates	illustrate	VBZ
work_v44vkh6bzramngvcrficy2no44	121	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	4	scoring	score	VBG
work_v44vkh6bzramngvcrficy2no44	121	5	process	process	NN
work_v44vkh6bzramngvcrficy2no44	121	6	:	:	:
work_v44vkh6bzramngvcrficy2no44	121	7	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	121	8	a	a	DT
work_v44vkh6bzramngvcrficy2no44	121	9	par-	par-	NN
work_v44vkh6bzramngvcrficy2no44	121	10	tial	tial	JJ
work_v44vkh6bzramngvcrficy2no44	121	11	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	121	12	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	121	13	H	h	NN
work_v44vkh6bzramngvcrficy2no44	121	14	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	121	15	expanded	expand	VBN
work_v44vkh6bzramngvcrficy2no44	121	16	by	by	IN
work_v44vkh6bzramngvcrficy2no44	121	17	cover-	cover-	NNP
work_v44vkh6bzramngvcrficy2no44	121	18	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	121	19	a	a	DT
work_v44vkh6bzramngvcrficy2no44	121	20	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	121	21	source	source	NN
work_v44vkh6bzramngvcrficy2no44	121	22	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	121	23	f̃	f̃	NNP
work_v44vkh6bzramngvcrficy2no44	121	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	121	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	26	model	model	NN
work_v44vkh6bzramngvcrficy2no44	121	27	returns	return	VBZ
work_v44vkh6bzramngvcrficy2no44	121	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	29	log	log	NN
work_v44vkh6bzramngvcrficy2no44	121	30	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	121	31	probability	probability	NN
work_v44vkh6bzramngvcrficy2no44	121	32	of	of	IN
work_v44vkh6bzramngvcrficy2no44	121	33	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	121	34	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	35	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	121	36	of	of	IN
work_v44vkh6bzramngvcrficy2no44	121	37	f̃	f̃	NNP
work_v44vkh6bzramngvcrficy2no44	121	38	in	in	IN
work_v44vkh6bzramngvcrficy2no44	121	39	that	that	DT
work_v44vkh6bzramngvcrficy2no44	121	40	particular	particular	JJ
work_v44vkh6bzramngvcrficy2no44	121	41	order	order	NN
work_v44vkh6bzramngvcrficy2no44	121	42	,	,	,
work_v44vkh6bzramngvcrficy2no44	121	43	just	just	RB
work_v44vkh6bzramngvcrficy2no44	121	44	after	after	IN
work_v44vkh6bzramngvcrficy2no44	121	45	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	46	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	121	47	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	121	48	word	word	NN
work_v44vkh6bzramngvcrficy2no44	121	49	of	of	IN
work_v44vkh6bzramngvcrficy2no44	121	50	H.	H.	NNP
work_v44vkh6bzramngvcrficy2no44	121	51	In	in	IN
work_v44vkh6bzramngvcrficy2no44	121	52	details	detail	NNS
work_v44vkh6bzramngvcrficy2no44	121	53	,	,	,
work_v44vkh6bzramngvcrficy2no44	121	54	this	this	DT
work_v44vkh6bzramngvcrficy2no44	121	55	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	121	56	done	do	VBN
work_v44vkh6bzramngvcrficy2no44	121	57	by	by	IN
work_v44vkh6bzramngvcrficy2no44	121	58	converting	convert	VBG
work_v44vkh6bzramngvcrficy2no44	121	59	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	60	phrase-	phrase-	NN
work_v44vkh6bzramngvcrficy2no44	121	61	internal	internal	JJ
work_v44vkh6bzramngvcrficy2no44	121	62	word	word	NN
work_v44vkh6bzramngvcrficy2no44	121	63	alignment4	alignment4	NNP
work_v44vkh6bzramngvcrficy2no44	121	64	to	to	IN
work_v44vkh6bzramngvcrficy2no44	121	65	a	a	DT
work_v44vkh6bzramngvcrficy2no44	121	66	source	source	NN
work_v44vkh6bzramngvcrficy2no44	121	67	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	121	68	,	,	,
work_v44vkh6bzramngvcrficy2no44	121	69	in	in	IN
work_v44vkh6bzramngvcrficy2no44	121	70	just	just	RB
work_v44vkh6bzramngvcrficy2no44	121	71	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	72	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	121	73	way	way	NN
work_v44vkh6bzramngvcrficy2no44	121	74	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	121	75	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	121	76	done	do	VBN
work_v44vkh6bzramngvcrficy2no44	121	77	to	to	TO
work_v44vkh6bzramngvcrficy2no44	121	78	produce	produce	VB
work_v44vkh6bzramngvcrficy2no44	121	79	the	the	DT
work_v44vkh6bzramngvcrficy2no44	121	80	model	model	NN
work_v44vkh6bzramngvcrficy2no44	121	81	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	121	82	training	training	NN
work_v44vkh6bzramngvcrficy2no44	121	83	examples	example	NNS
work_v44vkh6bzramngvcrficy2no44	121	84	.	.	.
work_v44vkh6bzramngvcrficy2no44	122	1	Thus	thus	RB
work_v44vkh6bzramngvcrficy2no44	122	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	122	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	122	4	global	global	JJ
work_v44vkh6bzramngvcrficy2no44	122	5	score	score	NN
work_v44vkh6bzramngvcrficy2no44	122	6	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	122	7	inde-	inde-	VBN
work_v44vkh6bzramngvcrficy2no44	122	8	pendent	pendent	NN
work_v44vkh6bzramngvcrficy2no44	122	9	from	from	IN
work_v44vkh6bzramngvcrficy2no44	122	10	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	122	11	segmentation	segmentation	NN
work_v44vkh6bzramngvcrficy2no44	122	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	122	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	122	14	normalized	normalize	VBD
work_v44vkh6bzramngvcrficy2no44	122	15	across	across	IN
work_v44vkh6bzramngvcrficy2no44	122	16	outputs	output	NNS
work_v44vkh6bzramngvcrficy2no44	122	17	of	of	IN
work_v44vkh6bzramngvcrficy2no44	122	18	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	122	19	lengths	length	NNS
work_v44vkh6bzramngvcrficy2no44	122	20	:	:	:
work_v44vkh6bzramngvcrficy2no44	122	21	that	that	RB
work_v44vkh6bzramngvcrficy2no44	122	22	is	is	RB
work_v44vkh6bzramngvcrficy2no44	122	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	122	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	122	25	proba-	proba-	NN
work_v44vkh6bzramngvcrficy2no44	122	26	bility	bility	NN
work_v44vkh6bzramngvcrficy2no44	122	27	of	of	IN
work_v44vkh6bzramngvcrficy2no44	122	28	any	any	DT
work_v44vkh6bzramngvcrficy2no44	122	29	complete	complete	JJ
work_v44vkh6bzramngvcrficy2no44	122	30	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	122	31	decomposes	decompose	NNS
work_v44vkh6bzramngvcrficy2no44	122	32	into	into	IN
work_v44vkh6bzramngvcrficy2no44	122	33	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	122	34	factors	factor	NNS
work_v44vkh6bzramngvcrficy2no44	122	35	,	,	,
work_v44vkh6bzramngvcrficy2no44	122	36	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	122	37	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	122	38	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	122	39	the	the	DT
work_v44vkh6bzramngvcrficy2no44	122	40	length	length	NN
work_v44vkh6bzramngvcrficy2no44	122	41	of	of	IN
work_v44vkh6bzramngvcrficy2no44	122	42	the	the	DT
work_v44vkh6bzramngvcrficy2no44	122	43	input	input	NN
work_v44vkh6bzramngvcrficy2no44	122	44	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	122	45	.	.	.
work_v44vkh6bzramngvcrficy2no44	123	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	123	2	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	123	3	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	123	4	model	model	NN
work_v44vkh6bzramngvcrficy2no44	123	5	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	123	6	fully	fully	RB
work_v44vkh6bzramngvcrficy2no44	123	7	compatible	compatible	JJ
work_v44vkh6bzramngvcrficy2no44	123	8	with	with	IN
work_v44vkh6bzramngvcrficy2no44	123	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	123	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	123	11	complementary	complementary	JJ
work_v44vkh6bzramngvcrficy2no44	123	12	to	to	IN
work_v44vkh6bzramngvcrficy2no44	123	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	123	14	lexicalized	lexicalized	JJ
work_v44vkh6bzramngvcrficy2no44	123	15	reorder-	reorder-	NNP
work_v44vkh6bzramngvcrficy2no44	123	16	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	123	17	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	123	18	phrase	phrase	NNP
work_v44vkh6bzramngvcrficy2no44	123	19	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	123	20	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	123	21	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	123	22	included	include	VBN
work_v44vkh6bzramngvcrficy2no44	123	23	in	in	IN
work_v44vkh6bzramngvcrficy2no44	123	24	Moses	Moses	NNP
work_v44vkh6bzramngvcrficy2no44	123	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	124	1	Figure	figure	NN
work_v44vkh6bzramngvcrficy2no44	124	2	1	1	CD
work_v44vkh6bzramngvcrficy2no44	124	3	:	:	:
work_v44vkh6bzramngvcrficy2no44	124	4	Integrating	integrate	VBG
work_v44vkh6bzramngvcrficy2no44	124	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	124	6	binary	binary	JJ
work_v44vkh6bzramngvcrficy2no44	124	7	word	word	NN
work_v44vkh6bzramngvcrficy2no44	124	8	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	124	9	model	model	NN
work_v44vkh6bzramngvcrficy2no44	124	10	into	into	IN
work_v44vkh6bzramngvcrficy2no44	124	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	124	12	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	124	13	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	124	14	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	124	15	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	124	16	:	:	:
work_v44vkh6bzramngvcrficy2no44	124	17	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	124	18	a	a	DT
work_v44vkh6bzramngvcrficy2no44	124	19	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	124	20	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	124	21	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	124	22	covered	cover	VBN
work_v44vkh6bzramngvcrficy2no44	124	23	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	124	24	dashed	dash	VBN
work_v44vkh6bzramngvcrficy2no44	124	25	boxes	box	NNS
work_v44vkh6bzramngvcrficy2no44	124	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	124	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	124	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	124	29	model	model	NN
work_v44vkh6bzramngvcrficy2no44	124	30	returns	return	VBZ
work_v44vkh6bzramngvcrficy2no44	124	31	the	the	DT
work_v44vkh6bzramngvcrficy2no44	124	32	log-	log-	JJ
work_v44vkh6bzramngvcrficy2no44	124	33	probability	probability	NN
work_v44vkh6bzramngvcrficy2no44	124	34	of	of	IN
work_v44vkh6bzramngvcrficy2no44	124	35	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	124	36	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	124	37	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	124	38	in	in	IN
work_v44vkh6bzramngvcrficy2no44	124	39	the	the	DT
work_v44vkh6bzramngvcrficy2no44	124	40	order	order	NN
work_v44vkh6bzramngvcrficy2no44	124	41	defined	define	VBN
work_v44vkh6bzramngvcrficy2no44	124	42	by	by	IN
work_v44vkh6bzramngvcrficy2no44	124	43	the	the	DT
work_v44vkh6bzramngvcrficy2no44	124	44	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	124	45	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	124	46	internal	internal	JJ
work_v44vkh6bzramngvcrficy2no44	124	47	word	word	NN
work_v44vkh6bzramngvcrficy2no44	124	48	alignment	alignment	NN
work_v44vkh6bzramngvcrficy2no44	124	49	.	.	.
work_v44vkh6bzramngvcrficy2no44	125	1	4	4	LS
work_v44vkh6bzramngvcrficy2no44	125	2	Early	early	JJ
work_v44vkh6bzramngvcrficy2no44	125	3	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	125	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	125	5	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	125	6	steps	step	NNS
work_v44vkh6bzramngvcrficy2no44	125	7	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	125	8	now	now	RB
work_v44vkh6bzramngvcrficy2no44	125	9	explain	explain	VBP
work_v44vkh6bzramngvcrficy2no44	125	10	how	how	WRB
work_v44vkh6bzramngvcrficy2no44	125	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	125	12	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	125	13	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	125	14	model	model	NN
work_v44vkh6bzramngvcrficy2no44	125	15	can	can	MD
work_v44vkh6bzramngvcrficy2no44	125	16	be	be	VB
work_v44vkh6bzramngvcrficy2no44	125	17	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	125	18	to	to	TO
work_v44vkh6bzramngvcrficy2no44	125	19	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	125	20	refine	refine	VB
work_v44vkh6bzramngvcrficy2no44	125	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	125	22	input	input	NN
work_v44vkh6bzramngvcrficy2no44	125	23	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	125	24	space	space	NN
work_v44vkh6bzramngvcrficy2no44	125	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	126	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	126	2	method	method	NN
work_v44vkh6bzramngvcrficy2no44	126	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	126	4	not	not	RB
work_v44vkh6bzramngvcrficy2no44	126	5	dependent	dependent	JJ
work_v44vkh6bzramngvcrficy2no44	126	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	126	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	126	8	particu-	particu-	NN
work_v44vkh6bzramngvcrficy2no44	126	9	lar	lar	NN
work_v44vkh6bzramngvcrficy2no44	126	10	classifier	classifier	NNP
work_v44vkh6bzramngvcrficy2no44	126	11	described	describe	VBN
work_v44vkh6bzramngvcrficy2no44	126	12	in	in	IN
work_v44vkh6bzramngvcrficy2no44	126	13	this	this	DT
work_v44vkh6bzramngvcrficy2no44	126	14	paper	paper	NN
work_v44vkh6bzramngvcrficy2no44	126	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	126	16	but	but	CC
work_v44vkh6bzramngvcrficy2no44	126	17	can	can	MD
work_v44vkh6bzramngvcrficy2no44	126	18	in	in	IN
work_v44vkh6bzramngvcrficy2no44	126	19	prin-	prin-	JJ
work_v44vkh6bzramngvcrficy2no44	126	20	ciple	ciple	NN
work_v44vkh6bzramngvcrficy2no44	126	21	work	work	NN
work_v44vkh6bzramngvcrficy2no44	126	22	with	with	IN
work_v44vkh6bzramngvcrficy2no44	126	23	any	any	DT
work_v44vkh6bzramngvcrficy2no44	126	24	device	device	NN
work_v44vkh6bzramngvcrficy2no44	126	25	estimating	estimate	VBG
work_v44vkh6bzramngvcrficy2no44	126	26	the	the	DT
work_v44vkh6bzramngvcrficy2no44	126	27	probabil-	probabil-	JJ
work_v44vkh6bzramngvcrficy2no44	126	28	ity	ity	NN
work_v44vkh6bzramngvcrficy2no44	126	29	of	of	IN
work_v44vkh6bzramngvcrficy2no44	126	30	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	126	31	a	a	DT
work_v44vkh6bzramngvcrficy2no44	126	32	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	126	33	input	input	NN
work_v44vkh6bzramngvcrficy2no44	126	34	word	word	NN
work_v44vkh6bzramngvcrficy2no44	126	35	after	after	IN
work_v44vkh6bzramngvcrficy2no44	126	36	another	another	DT
work_v44vkh6bzramngvcrficy2no44	126	37	.	.	.
work_v44vkh6bzramngvcrficy2no44	127	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	127	2	method	method	NN
work_v44vkh6bzramngvcrficy2no44	127	3	consists	consist	VBZ
work_v44vkh6bzramngvcrficy2no44	127	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	127	5	querying	query	VBG
work_v44vkh6bzramngvcrficy2no44	127	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	127	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	127	8	model	model	NN
work_v44vkh6bzramngvcrficy2no44	127	9	at	at	IN
work_v44vkh6bzramngvcrficy2no44	127	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	127	11	time	time	NN
work_v44vkh6bzramngvcrficy2no44	127	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	127	13	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	127	14	expansion	expansion	NN
work_v44vkh6bzramngvcrficy2no44	127	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	127	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	127	17	fil-	fil-	RB
work_v44vkh6bzramngvcrficy2no44	127	18	tering	tere	VBG
work_v44vkh6bzramngvcrficy2no44	127	19	out	out	RP
work_v44vkh6bzramngvcrficy2no44	127	20	hypotheses	hypothesis	NNS
work_v44vkh6bzramngvcrficy2no44	127	21	solely	solely	RB
work_v44vkh6bzramngvcrficy2no44	127	22	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	127	23	on	on	IN
work_v44vkh6bzramngvcrficy2no44	127	24	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	127	25	reorder-	reorder-	JJ
work_v44vkh6bzramngvcrficy2no44	127	26	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	127	27	score	score	NN
work_v44vkh6bzramngvcrficy2no44	127	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	128	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	128	2	rationale	rationale	NN
work_v44vkh6bzramngvcrficy2no44	128	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	128	4	to	to	TO
work_v44vkh6bzramngvcrficy2no44	128	5	avoid	avoid	VB
work_v44vkh6bzramngvcrficy2no44	128	6	costly	costly	JJ
work_v44vkh6bzramngvcrficy2no44	128	7	hypoth-	hypoth-	PRP$
work_v44vkh6bzramngvcrficy2no44	128	8	esis	esis	NN
work_v44vkh6bzramngvcrficy2no44	128	9	expansions	expansion	NNS
work_v44vkh6bzramngvcrficy2no44	128	10	for	for	IN
work_v44vkh6bzramngvcrficy2no44	128	11	those	those	DT
work_v44vkh6bzramngvcrficy2no44	128	12	source	source	NN
work_v44vkh6bzramngvcrficy2no44	128	13	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	128	14	that	that	IN
work_v44vkh6bzramngvcrficy2no44	128	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	128	16	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	128	17	model	model	NN
work_v44vkh6bzramngvcrficy2no44	128	18	considers	consider	VBZ
work_v44vkh6bzramngvcrficy2no44	128	19	very	very	RB
work_v44vkh6bzramngvcrficy2no44	128	20	unlikely	unlikely	JJ
work_v44vkh6bzramngvcrficy2no44	128	21	to	to	TO
work_v44vkh6bzramngvcrficy2no44	128	22	be	be	VB
work_v44vkh6bzramngvcrficy2no44	128	23	cov-	cov-	RB
work_v44vkh6bzramngvcrficy2no44	128	24	ered	ere	VBN
work_v44vkh6bzramngvcrficy2no44	128	25	at	at	IN
work_v44vkh6bzramngvcrficy2no44	128	26	a	a	DT
work_v44vkh6bzramngvcrficy2no44	128	27	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	128	28	point	point	NN
work_v44vkh6bzramngvcrficy2no44	128	29	of	of	IN
work_v44vkh6bzramngvcrficy2no44	128	30	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	128	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	129	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	129	2	practice	practice	NN
work_v44vkh6bzramngvcrficy2no44	129	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	129	4	this	this	DT
work_v44vkh6bzramngvcrficy2no44	129	5	works	work	VBZ
work_v44vkh6bzramngvcrficy2no44	129	6	as	as	IN
work_v44vkh6bzramngvcrficy2no44	129	7	follows	follow	VBZ
work_v44vkh6bzramngvcrficy2no44	129	8	:	:	:
work_v44vkh6bzramngvcrficy2no44	129	9	•	•	VB
work_v44vkh6bzramngvcrficy2no44	129	10	at	at	IN
work_v44vkh6bzramngvcrficy2no44	129	11	each	each	DT
work_v44vkh6bzramngvcrficy2no44	129	12	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	129	13	expansion	expansion	NN
work_v44vkh6bzramngvcrficy2no44	129	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	129	15	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	129	16	first	first	RB
work_v44vkh6bzramngvcrficy2no44	129	17	enumer-	enumer-	NN
work_v44vkh6bzramngvcrficy2no44	129	18	ate	eat	VBD
work_v44vkh6bzramngvcrficy2no44	129	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	129	20	set	set	NN
work_v44vkh6bzramngvcrficy2no44	129	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	129	22	uncovered	uncovered	JJ
work_v44vkh6bzramngvcrficy2no44	129	23	input	input	NN
work_v44vkh6bzramngvcrficy2no44	129	24	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	129	25	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	129	26	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	129	27	reachable	reachable	JJ
work_v44vkh6bzramngvcrficy2no44	129	28	within	within	IN
work_v44vkh6bzramngvcrficy2no44	129	29	a	a	DT
work_v44vkh6bzramngvcrficy2no44	129	30	fixed	fix	VBN
work_v44vkh6bzramngvcrficy2no44	129	31	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	129	32	,	,	,
work_v44vkh6bzramngvcrficy2no44	129	33	and	and	CC
work_v44vkh6bzramngvcrficy2no44	129	34	query	query	NN
work_v44vkh6bzramngvcrficy2no44	129	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	129	36	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	129	37	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	129	38	model	model	NN
work_v44vkh6bzramngvcrficy2no44	129	39	for	for	IN
work_v44vkh6bzramngvcrficy2no44	129	40	each	each	DT
work_v44vkh6bzramngvcrficy2no44	129	41	of	of	IN
work_v44vkh6bzramngvcrficy2no44	129	42	them5	them5	NN
work_v44vkh6bzramngvcrficy2no44	129	43	;	;	:
work_v44vkh6bzramngvcrficy2no44	129	44	4Phrase	4phrase	CD
work_v44vkh6bzramngvcrficy2no44	129	45	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	129	46	internal	internal	JJ
work_v44vkh6bzramngvcrficy2no44	129	47	alignments	alignment	NNS
work_v44vkh6bzramngvcrficy2no44	129	48	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	129	49	provided	provide	VBN
work_v44vkh6bzramngvcrficy2no44	129	50	in	in	IN
work_v44vkh6bzramngvcrficy2no44	129	51	the	the	DT
work_v44vkh6bzramngvcrficy2no44	129	52	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	129	53	table	table	NN
work_v44vkh6bzramngvcrficy2no44	129	54	.	.	.
work_v44vkh6bzramngvcrficy2no44	130	1	5The	5the	CD
work_v44vkh6bzramngvcrficy2no44	130	2	score	score	NN
work_v44vkh6bzramngvcrficy2no44	130	3	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	130	4	to	to	TO
work_v44vkh6bzramngvcrficy2no44	130	5	prune	prune	VB
work_v44vkh6bzramngvcrficy2no44	130	6	a	a	DT
work_v44vkh6bzramngvcrficy2no44	130	7	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	130	8	word	word	NN
work_v44vkh6bzramngvcrficy2no44	130	9	range	range	NN
work_v44vkh6bzramngvcrficy2no44	130	10	f̃	f̃	NNP
work_v44vkh6bzramngvcrficy2no44	130	11	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	130	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	130	13	log	log	NN
work_v44vkh6bzramngvcrficy2no44	130	14	prob-	prob-	XX
work_v44vkh6bzramngvcrficy2no44	130	15	ability	ability	NN
work_v44vkh6bzramngvcrficy2no44	130	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	130	17	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	130	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	130	19	first	first	JJ
work_v44vkh6bzramngvcrficy2no44	130	20	aligned	align	VBN
work_v44vkh6bzramngvcrficy2no44	130	21	word	word	NN
work_v44vkh6bzramngvcrficy2no44	130	22	of	of	IN
work_v44vkh6bzramngvcrficy2no44	130	23	f̃	f̃	NNP
work_v44vkh6bzramngvcrficy2no44	130	24	right	right	RB
work_v44vkh6bzramngvcrficy2no44	130	25	after	after	IN
work_v44vkh6bzramngvcrficy2no44	130	26	the	the	DT
work_v44vkh6bzramngvcrficy2no44	130	27	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	130	28	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	130	29	word	word	NN
work_v44vkh6bzramngvcrficy2no44	130	30	of	of	IN
work_v44vkh6bzramngvcrficy2no44	130	31	the	the	DT
work_v44vkh6bzramngvcrficy2no44	130	32	current	current	JJ
work_v44vkh6bzramngvcrficy2no44	130	33	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	130	34	.	.	.
work_v44vkh6bzramngvcrficy2no44	131	1	See	see	VB
work_v44vkh6bzramngvcrficy2no44	131	2	also	also	RB
work_v44vkh6bzramngvcrficy2no44	131	3	Sect	Sect	NNP
work_v44vkh6bzramngvcrficy2no44	131	4	.	.	.
work_v44vkh6bzramngvcrficy2no44	132	1	3.3	3.3	CD
work_v44vkh6bzramngvcrficy2no44	132	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	133	1	331	331	CD
work_v44vkh6bzramngvcrficy2no44	133	2	•	•	NNP
work_v44vkh6bzramngvcrficy2no44	133	3	only	only	RB
work_v44vkh6bzramngvcrficy2no44	133	4	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	133	5	on	on	IN
work_v44vkh6bzramngvcrficy2no44	133	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	133	7	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	133	8	score	score	NN
work_v44vkh6bzramngvcrficy2no44	133	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	133	10	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	133	11	apply	apply	VBP
work_v44vkh6bzramngvcrficy2no44	133	12	his-	his-	RB
work_v44vkh6bzramngvcrficy2no44	133	13	togram	togram	NN
work_v44vkh6bzramngvcrficy2no44	133	14	and	and	CC
work_v44vkh6bzramngvcrficy2no44	133	15	threshold	threshold	NN
work_v44vkh6bzramngvcrficy2no44	133	16	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	133	17	to	to	IN
work_v44vkh6bzramngvcrficy2no44	133	18	this	this	DT
work_v44vkh6bzramngvcrficy2no44	133	19	set	set	NN
work_v44vkh6bzramngvcrficy2no44	133	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	133	21	proceed	proceed	VB
work_v44vkh6bzramngvcrficy2no44	133	22	to	to	TO
work_v44vkh6bzramngvcrficy2no44	133	23	expand	expand	VB
work_v44vkh6bzramngvcrficy2no44	133	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	133	25	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	133	26	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	133	27	pruned	pruned	JJ
work_v44vkh6bzramngvcrficy2no44	133	28	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	133	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	134	1	Furthermore	furthermore	RB
work_v44vkh6bzramngvcrficy2no44	134	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	3	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	134	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	134	5	possible	possible	JJ
work_v44vkh6bzramngvcrficy2no44	134	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	134	7	ensure	ensure	VB
work_v44vkh6bzramngvcrficy2no44	134	8	that	that	IN
work_v44vkh6bzramngvcrficy2no44	134	9	local	local	JJ
work_v44vkh6bzramngvcrficy2no44	134	10	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	134	11	orderings	ordering	NNS
work_v44vkh6bzramngvcrficy2no44	134	12	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	134	13	always	always	RB
work_v44vkh6bzramngvcrficy2no44	134	14	allowed	allow	VBN
work_v44vkh6bzramngvcrficy2no44	134	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	16	by	by	IN
work_v44vkh6bzramngvcrficy2no44	134	17	setting	set	VBG
work_v44vkh6bzramngvcrficy2no44	134	18	a	a	DT
work_v44vkh6bzramngvcrficy2no44	134	19	so	so	RB
work_v44vkh6bzramngvcrficy2no44	134	20	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	134	21	called	call	VBN
work_v44vkh6bzramngvcrficy2no44	134	22	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	134	23	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	134	24	prunable	prunable	JJ
work_v44vkh6bzramngvcrficy2no44	134	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	134	26	zone	zone	NN
work_v44vkh6bzramngvcrficy2no44	134	27	of	of	IN
work_v44vkh6bzramngvcrficy2no44	134	28	width	width	NN
work_v44vkh6bzramngvcrficy2no44	134	29	$	$	$
work_v44vkh6bzramngvcrficy2no44	134	30	around	around	IN
work_v44vkh6bzramngvcrficy2no44	134	31	the	the	DT
work_v44vkh6bzramngvcrficy2no44	134	32	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	134	33	cov-	cov-	RB
work_v44vkh6bzramngvcrficy2no44	134	34	ered	ered	JJ
work_v44vkh6bzramngvcrficy2no44	134	35	input	input	NN
work_v44vkh6bzramngvcrficy2no44	134	36	position.6	position.6	:
work_v44vkh6bzramngvcrficy2no44	134	37	According	accord	VBG
work_v44vkh6bzramngvcrficy2no44	134	38	to	to	IN
work_v44vkh6bzramngvcrficy2no44	134	39	how	how	WRB
work_v44vkh6bzramngvcrficy2no44	134	40	the	the	DT
work_v44vkh6bzramngvcrficy2no44	134	41	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	134	42	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	43	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	134	44	parameters	parameter	NNS
work_v44vkh6bzramngvcrficy2no44	134	45	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	46	and	and	CC
work_v44vkh6bzramngvcrficy2no44	134	47	$	$	$
work_v44vkh6bzramngvcrficy2no44	134	48	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	134	49	set	set	VBN
work_v44vkh6bzramngvcrficy2no44	134	50	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	51	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	134	52	can	can	MD
work_v44vkh6bzramngvcrficy2no44	134	53	actually	actually	RB
work_v44vkh6bzramngvcrficy2no44	134	54	aim	aim	VB
work_v44vkh6bzramngvcrficy2no44	134	55	at	at	IN
work_v44vkh6bzramngvcrficy2no44	134	56	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	134	57	tar-	tar-	NN
work_v44vkh6bzramngvcrficy2no44	134	58	gets	get	VBZ
work_v44vkh6bzramngvcrficy2no44	134	59	:	:	:
work_v44vkh6bzramngvcrficy2no44	134	60	with	with	IN
work_v44vkh6bzramngvcrficy2no44	134	61	a	a	DT
work_v44vkh6bzramngvcrficy2no44	134	62	low	low	JJ
work_v44vkh6bzramngvcrficy2no44	134	63	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	134	64	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	65	loose	loose	JJ
work_v44vkh6bzramngvcrficy2no44	134	66	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	134	67	parameters	parameter	NNS
work_v44vkh6bzramngvcrficy2no44	134	68	,	,	,
work_v44vkh6bzramngvcrficy2no44	134	69	and	and	CC
work_v44vkh6bzramngvcrficy2no44	134	70	$	$	$
work_v44vkh6bzramngvcrficy2no44	134	71	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	134	72	0	0	NFP
work_v44vkh6bzramngvcrficy2no44	134	73	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	134	74	can	can	MD
work_v44vkh6bzramngvcrficy2no44	134	75	try	try	VB
work_v44vkh6bzramngvcrficy2no44	134	76	to	to	TO
work_v44vkh6bzramngvcrficy2no44	134	77	speed	speed	VB
work_v44vkh6bzramngvcrficy2no44	134	78	up	up	RP
work_v44vkh6bzramngvcrficy2no44	134	79	search	search	NN
work_v44vkh6bzramngvcrficy2no44	134	80	without	without	IN
work_v44vkh6bzramngvcrficy2no44	134	81	sacrific-	sacrific-	NNP
work_v44vkh6bzramngvcrficy2no44	134	82	ing	ing	JJ
work_v44vkh6bzramngvcrficy2no44	134	83	much	much	JJ
work_v44vkh6bzramngvcrficy2no44	134	84	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	134	85	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	134	86	.	.	.
work_v44vkh6bzramngvcrficy2no44	135	1	With	with	IN
work_v44vkh6bzramngvcrficy2no44	135	2	a	a	DT
work_v44vkh6bzramngvcrficy2no44	135	3	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	135	4	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	135	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	135	6	strict	strict	JJ
work_v44vkh6bzramngvcrficy2no44	135	7	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	135	8	parameters	parameter	NNS
work_v44vkh6bzramngvcrficy2no44	135	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	135	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	135	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	135	12	medium	medium	JJ
work_v44vkh6bzramngvcrficy2no44	135	13	$	$	$
work_v44vkh6bzramngvcrficy2no44	135	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	135	15	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	135	16	ensure	ensure	VBP
work_v44vkh6bzramngvcrficy2no44	135	17	that	that	IN
work_v44vkh6bzramngvcrficy2no44	135	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	135	19	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	135	20	medium	medium	JJ
work_v44vkh6bzramngvcrficy2no44	135	21	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	135	22	range	range	NN
work_v44vkh6bzramngvcrficy2no44	135	23	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	135	24	space	space	NN
work_v44vkh6bzramngvcrficy2no44	135	25	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	135	26	ex-	ex-	RB
work_v44vkh6bzramngvcrficy2no44	135	27	plored	plore	VBN
work_v44vkh6bzramngvcrficy2no44	135	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	135	29	as	as	RB
work_v44vkh6bzramngvcrficy2no44	135	30	well	well	RB
work_v44vkh6bzramngvcrficy2no44	135	31	as	as	IN
work_v44vkh6bzramngvcrficy2no44	135	32	those	those	DT
work_v44vkh6bzramngvcrficy2no44	135	33	few	few	JJ
work_v44vkh6bzramngvcrficy2no44	135	34	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	135	35	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	135	36	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	135	37	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	135	38	promising	promise	VBG
work_v44vkh6bzramngvcrficy2no44	135	39	according	accord	VBG
work_v44vkh6bzramngvcrficy2no44	135	40	to	to	IN
work_v44vkh6bzramngvcrficy2no44	135	41	the	the	DT
work_v44vkh6bzramngvcrficy2no44	135	42	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	135	43	model	model	NN
work_v44vkh6bzramngvcrficy2no44	135	44	.	.	.
work_v44vkh6bzramngvcrficy2no44	136	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	136	2	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	136	3	experiments	experiment	NNS
work_v44vkh6bzramngvcrficy2no44	136	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	136	5	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	136	6	explore	explore	VBP
work_v44vkh6bzramngvcrficy2no44	136	7	this	this	DT
work_v44vkh6bzramngvcrficy2no44	136	8	second	second	JJ
work_v44vkh6bzramngvcrficy2no44	136	9	option	option	NN
work_v44vkh6bzramngvcrficy2no44	136	10	with	with	IN
work_v44vkh6bzramngvcrficy2no44	136	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	136	12	setting	set	VBG
work_v44vkh6bzramngvcrficy2no44	136	13	DL=18	dl=18	NN
work_v44vkh6bzramngvcrficy2no44	136	14	and	and	CC
work_v44vkh6bzramngvcrficy2no44	136	15	$	$	$
work_v44vkh6bzramngvcrficy2no44	136	16	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	136	17	5	5	CD
work_v44vkh6bzramngvcrficy2no44	136	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	137	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	137	2	underlying	underlie	VBG
work_v44vkh6bzramngvcrficy2no44	137	3	idea	idea	NN
work_v44vkh6bzramngvcrficy2no44	137	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	137	5	similar	similar	JJ
work_v44vkh6bzramngvcrficy2no44	137	6	to	to	IN
work_v44vkh6bzramngvcrficy2no44	137	7	that	that	DT
work_v44vkh6bzramngvcrficy2no44	137	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	137	9	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	137	10	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	137	11	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	137	12	by	by	IN
work_v44vkh6bzramngvcrficy2no44	137	13	Moore	Moore	NNP
work_v44vkh6bzramngvcrficy2no44	137	14	and	and	CC
work_v44vkh6bzramngvcrficy2no44	137	15	Quirk	Quirk	NNP
work_v44vkh6bzramngvcrficy2no44	137	16	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	137	17	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	137	18	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	137	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	137	20	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	137	21	consisted	consist	VBD
work_v44vkh6bzramngvcrficy2no44	137	22	in	in	IN
work_v44vkh6bzramngvcrficy2no44	137	23	discarding	discard	VBG
work_v44vkh6bzramngvcrficy2no44	137	24	possible	possible	JJ
work_v44vkh6bzramngvcrficy2no44	137	25	extensions	extension	NNS
work_v44vkh6bzramngvcrficy2no44	137	26	of	of	IN
work_v44vkh6bzramngvcrficy2no44	137	27	a	a	DT
work_v44vkh6bzramngvcrficy2no44	137	28	partial	partial	JJ
work_v44vkh6bzramngvcrficy2no44	137	29	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	137	30	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	137	31	on	on	IN
work_v44vkh6bzramngvcrficy2no44	137	32	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	137	33	estimated	estimate	VBN
work_v44vkh6bzramngvcrficy2no44	137	34	score	score	NN
work_v44vkh6bzramngvcrficy2no44	137	35	before	before	IN
work_v44vkh6bzramngvcrficy2no44	137	36	computing	compute	VBG
work_v44vkh6bzramngvcrficy2no44	137	37	the	the	DT
work_v44vkh6bzramngvcrficy2no44	137	38	exact	exact	JJ
work_v44vkh6bzramngvcrficy2no44	137	39	language	language	NN
work_v44vkh6bzramngvcrficy2no44	137	40	model	model	NN
work_v44vkh6bzramngvcrficy2no44	137	41	score	score	NN
work_v44vkh6bzramngvcrficy2no44	137	42	.	.	.
work_v44vkh6bzramngvcrficy2no44	138	1	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	138	2	technique	technique	NN
work_v44vkh6bzramngvcrficy2no44	138	3	too	too	RB
work_v44vkh6bzramngvcrficy2no44	138	4	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	138	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	138	6	effect	effect	NN
work_v44vkh6bzramngvcrficy2no44	138	7	of	of	IN
work_v44vkh6bzramngvcrficy2no44	138	8	introducing	introduce	VBG
work_v44vkh6bzramngvcrficy2no44	138	9	ad-	ad-	CC
work_v44vkh6bzramngvcrficy2no44	138	10	ditional	ditional	JJ
work_v44vkh6bzramngvcrficy2no44	138	11	points	point	NNS
work_v44vkh6bzramngvcrficy2no44	138	12	at	at	IN
work_v44vkh6bzramngvcrficy2no44	138	13	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	138	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	138	15	search	search	NN
work_v44vkh6bzramngvcrficy2no44	138	16	space	space	NN
work_v44vkh6bzramngvcrficy2no44	138	17	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	138	18	pruned	prune	VBN
work_v44vkh6bzramngvcrficy2no44	138	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	139	1	However	however	RB
work_v44vkh6bzramngvcrficy2no44	139	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	139	3	while	while	IN
work_v44vkh6bzramngvcrficy2no44	139	4	theirs	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	139	5	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	139	6	mainly	mainly	RB
work_v44vkh6bzramngvcrficy2no44	139	7	an	an	DT
work_v44vkh6bzramngvcrficy2no44	139	8	optimization	optimization	NN
work_v44vkh6bzramngvcrficy2no44	139	9	technique	technique	NN
work_v44vkh6bzramngvcrficy2no44	139	10	meant	mean	VBN
work_v44vkh6bzramngvcrficy2no44	139	11	to	to	TO
work_v44vkh6bzramngvcrficy2no44	139	12	avoid	avoid	VB
work_v44vkh6bzramngvcrficy2no44	139	13	useless	useless	JJ
work_v44vkh6bzramngvcrficy2no44	139	14	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	139	15	queries	query	NNS
work_v44vkh6bzramngvcrficy2no44	139	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	139	17	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	139	18	in-	in-	RB
work_v44vkh6bzramngvcrficy2no44	139	19	stead	stead	VBP
work_v44vkh6bzramngvcrficy2no44	139	20	aim	aim	VBP
work_v44vkh6bzramngvcrficy2no44	139	21	at	at	IN
work_v44vkh6bzramngvcrficy2no44	139	22	refining	refine	VBG
work_v44vkh6bzramngvcrficy2no44	139	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	139	24	search	search	NN
work_v44vkh6bzramngvcrficy2no44	139	25	space	space	NN
work_v44vkh6bzramngvcrficy2no44	139	26	by	by	IN
work_v44vkh6bzramngvcrficy2no44	139	27	exploiting	exploit	VBG
work_v44vkh6bzramngvcrficy2no44	139	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	139	29	fact	fact	NN
work_v44vkh6bzramngvcrficy2no44	139	30	that	that	IN
work_v44vkh6bzramngvcrficy2no44	139	31	some	some	DT
work_v44vkh6bzramngvcrficy2no44	139	32	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	139	33	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	139	34	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	139	35	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	139	36	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	139	37	than	than	IN
work_v44vkh6bzramngvcrficy2no44	139	38	others	other	NNS
work_v44vkh6bzramngvcrficy2no44	139	39	at	at	IN
work_v44vkh6bzramngvcrficy2no44	139	40	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	139	41	stages	stage	NNS
work_v44vkh6bzramngvcrficy2no44	139	42	of	of	IN
work_v44vkh6bzramngvcrficy2no44	139	43	the	the	DT
work_v44vkh6bzramngvcrficy2no44	139	44	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	139	45	pro-	pro-	NN
work_v44vkh6bzramngvcrficy2no44	139	46	cess	cess	NN
work_v44vkh6bzramngvcrficy2no44	139	47	.	.	.
work_v44vkh6bzramngvcrficy2no44	140	1	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	140	2	approach	approach	NN
work_v44vkh6bzramngvcrficy2no44	140	3	actually	actually	RB
work_v44vkh6bzramngvcrficy2no44	140	4	involves	involve	VBZ
work_v44vkh6bzramngvcrficy2no44	140	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	140	6	continuous	continuous	JJ
work_v44vkh6bzramngvcrficy2no44	140	7	alternation	alternation	NN
work_v44vkh6bzramngvcrficy2no44	140	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	140	9	two	two	CD
work_v44vkh6bzramngvcrficy2no44	140	10	processes	process	NNS
work_v44vkh6bzramngvcrficy2no44	140	11	:	:	:
work_v44vkh6bzramngvcrficy2no44	140	12	during	during	IN
work_v44vkh6bzramngvcrficy2no44	140	13	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	140	14	ex-	ex-	DT
work_v44vkh6bzramngvcrficy2no44	140	15	pansion	pansion	NN
work_v44vkh6bzramngvcrficy2no44	140	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	140	17	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	140	18	score	score	NN
work_v44vkh6bzramngvcrficy2no44	140	19	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	140	20	combined	combine	VBN
work_v44vkh6bzramngvcrficy2no44	140	21	with	with	IN
work_v44vkh6bzramngvcrficy2no44	140	22	all	all	DT
work_v44vkh6bzramngvcrficy2no44	140	23	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	140	24	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	140	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	140	26	while	while	IN
work_v44vkh6bzramngvcrficy2no44	140	27	during	during	IN
work_v44vkh6bzramngvcrficy2no44	140	28	early	early	RB
work_v44vkh6bzramngvcrficy2no44	140	29	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	140	30	some	some	DT
work_v44vkh6bzramngvcrficy2no44	140	31	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	140	32	ordering	ordering	NN
work_v44vkh6bzramngvcrficy2no44	140	33	decisions	decision	NNS
work_v44vkh6bzramngvcrficy2no44	140	34	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	140	35	taken	take	VBN
work_v44vkh6bzramngvcrficy2no44	140	36	only	only	RB
work_v44vkh6bzramngvcrficy2no44	140	37	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	140	38	on	on	IN
work_v44vkh6bzramngvcrficy2no44	140	39	the	the	DT
work_v44vkh6bzramngvcrficy2no44	140	40	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	140	41	ordering	ordering	NN
work_v44vkh6bzramngvcrficy2no44	140	42	score	score	NN
work_v44vkh6bzramngvcrficy2no44	140	43	.	.	.
work_v44vkh6bzramngvcrficy2no44	141	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	141	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	141	3	way	way	NN
work_v44vkh6bzramngvcrficy2no44	141	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	141	5	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	141	6	try	try	VBP
work_v44vkh6bzramngvcrficy2no44	141	7	to	to	TO
work_v44vkh6bzramngvcrficy2no44	141	8	combine	combine	VB
work_v44vkh6bzramngvcrficy2no44	141	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	141	10	benefits	benefit	NNS
work_v44vkh6bzramngvcrficy2no44	141	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	141	12	fully	fully	RB
work_v44vkh6bzramngvcrficy2no44	141	13	integrated	integrate	VBN
work_v44vkh6bzramngvcrficy2no44	141	14	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	141	15	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	141	16	with	with	IN
work_v44vkh6bzramngvcrficy2no44	141	17	those	those	DT
work_v44vkh6bzramngvcrficy2no44	141	18	of	of	IN
work_v44vkh6bzramngvcrficy2no44	141	19	monolingual	monolingual	JJ
work_v44vkh6bzramngvcrficy2no44	141	20	pre	pre	JJ
work_v44vkh6bzramngvcrficy2no44	141	21	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	141	22	ordering	ordering	JJ
work_v44vkh6bzramngvcrficy2no44	141	23	methods	method	NNS
work_v44vkh6bzramngvcrficy2no44	141	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	142	1	5	5	CD
work_v44vkh6bzramngvcrficy2no44	142	2	Evaluation	Evaluation	NNP
work_v44vkh6bzramngvcrficy2no44	142	3	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	142	4	test	test	VBP
work_v44vkh6bzramngvcrficy2no44	142	5	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	142	6	approach	approach	NN
work_v44vkh6bzramngvcrficy2no44	142	7	on	on	IN
work_v44vkh6bzramngvcrficy2no44	142	8	an	an	DT
work_v44vkh6bzramngvcrficy2no44	142	9	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	142	10	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	142	11	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	142	12	news	news	NN
work_v44vkh6bzramngvcrficy2no44	142	13	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	142	14	task	task	NN
work_v44vkh6bzramngvcrficy2no44	142	15	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	142	16	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	142	17	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	142	18	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	142	19	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	142	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	142	21	complex	complex	JJ
work_v44vkh6bzramngvcrficy2no44	142	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	143	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	143	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	143	3	language	language	NN
work_v44vkh6bzramngvcrficy2no44	143	4	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	143	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	143	6	long	long	RB
work_v44vkh6bzramngvcrficy2no44	143	7	reorder-	reorder-	NN
work_v44vkh6bzramngvcrficy2no44	143	8	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	143	9	errors	error	NNS
work_v44vkh6bzramngvcrficy2no44	143	10	mostly	mostly	RB
work_v44vkh6bzramngvcrficy2no44	143	11	concern	concern	NN
work_v44vkh6bzramngvcrficy2no44	143	12	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	143	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	143	14	as	as	IN
work_v44vkh6bzramngvcrficy2no44	143	15	all	all	DT
work_v44vkh6bzramngvcrficy2no44	143	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	143	17	Subject-	Subject-	NNP
work_v44vkh6bzramngvcrficy2no44	143	18	Verb	Verb	NNP
work_v44vkh6bzramngvcrficy2no44	143	19	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	143	20	Object	Object	NNP
work_v44vkh6bzramngvcrficy2no44	143	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	143	22	SVO	SVO	NNP
work_v44vkh6bzramngvcrficy2no44	143	23	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	143	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	143	25	VSO	VSO	NNP
work_v44vkh6bzramngvcrficy2no44	143	26	and	and	CC
work_v44vkh6bzramngvcrficy2no44	143	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	143	28	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	143	29	rarerly	rarerly	RB
work_v44vkh6bzramngvcrficy2no44	143	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	143	31	VOS	VOS	NNP
work_v44vkh6bzramngvcrficy2no44	143	32	6See	6see	NN
work_v44vkh6bzramngvcrficy2no44	143	33	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	143	34	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	143	35	2013	2013	CD
work_v44vkh6bzramngvcrficy2no44	143	36	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	143	37	for	for	IN
work_v44vkh6bzramngvcrficy2no44	143	38	technical	technical	JJ
work_v44vkh6bzramngvcrficy2no44	143	39	details	detail	NNS
work_v44vkh6bzramngvcrficy2no44	143	40	on	on	IN
work_v44vkh6bzramngvcrficy2no44	143	41	the	the	DT
work_v44vkh6bzramngvcrficy2no44	143	42	integration	integration	NN
work_v44vkh6bzramngvcrficy2no44	143	43	of	of	IN
work_v44vkh6bzramngvcrficy2no44	143	44	word	word	NN
work_v44vkh6bzramngvcrficy2no44	143	45	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	143	46	level	level	NN
work_v44vkh6bzramngvcrficy2no44	143	47	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	143	48	with	with	IN
work_v44vkh6bzramngvcrficy2no44	143	49	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	143	50	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	143	51	level	level	NN
work_v44vkh6bzramngvcrficy2no44	143	52	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	143	53	expansion	expansion	NN
work_v44vkh6bzramngvcrficy2no44	143	54	.	.	.
work_v44vkh6bzramngvcrficy2no44	144	1	constructions	construction	NNS
work_v44vkh6bzramngvcrficy2no44	144	2	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	144	3	attested	attest	VBN
work_v44vkh6bzramngvcrficy2no44	144	4	in	in	IN
work_v44vkh6bzramngvcrficy2no44	144	5	modern	modern	JJ
work_v44vkh6bzramngvcrficy2no44	144	6	written	write	VBN
work_v44vkh6bzramngvcrficy2no44	144	7	Ara-	Ara-	NNP
work_v44vkh6bzramngvcrficy2no44	144	8	bic	bic	NN
work_v44vkh6bzramngvcrficy2no44	144	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	145	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	145	2	issue	issue	NN
work_v44vkh6bzramngvcrficy2no44	145	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	145	4	well	well	RB
work_v44vkh6bzramngvcrficy2no44	145	5	known	know	VBN
work_v44vkh6bzramngvcrficy2no44	145	6	in	in	IN
work_v44vkh6bzramngvcrficy2no44	145	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	145	8	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	145	9	field	field	NN
work_v44vkh6bzramngvcrficy2no44	145	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	145	11	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	145	12	addressed	address	VBN
work_v44vkh6bzramngvcrficy2no44	145	13	by	by	IN
work_v44vkh6bzramngvcrficy2no44	145	14	several	several	JJ
work_v44vkh6bzramngvcrficy2no44	145	15	recent	recent	JJ
work_v44vkh6bzramngvcrficy2no44	145	16	works	work	NNS
work_v44vkh6bzramngvcrficy2no44	145	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	145	18	with	with	IN
work_v44vkh6bzramngvcrficy2no44	145	19	deep	deep	JJ
work_v44vkh6bzramngvcrficy2no44	145	20	or	or	CC
work_v44vkh6bzramngvcrficy2no44	145	21	shallow	shallow	JJ
work_v44vkh6bzramngvcrficy2no44	145	22	parsing	parsing	NN
work_v44vkh6bzramngvcrficy2no44	145	23	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	145	24	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	145	25	techniques	technique	NNS
work_v44vkh6bzramngvcrficy2no44	145	26	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	145	27	Green	Green	NNP
work_v44vkh6bzramngvcrficy2no44	145	28	et	et	FW
work_v44vkh6bzramngvcrficy2no44	145	29	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	145	30	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	145	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	145	32	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	145	33	;	;	:
work_v44vkh6bzramngvcrficy2no44	145	34	Carpuat	Carpuat	NNP
work_v44vkh6bzramngvcrficy2no44	145	35	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	145	36	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	145	37	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	145	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	145	39	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	145	40	;	;	:
work_v44vkh6bzramngvcrficy2no44	145	41	Andreas	Andreas	NNP
work_v44vkh6bzramngvcrficy2no44	145	42	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	145	43	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	145	44	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	145	45	,	,	,
work_v44vkh6bzramngvcrficy2no44	145	46	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	145	47	;	;	:
work_v44vkh6bzramngvcrficy2no44	145	48	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	145	49	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	145	50	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	145	51	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	145	52	,	,	,
work_v44vkh6bzramngvcrficy2no44	145	53	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	145	54	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	145	55	.	.	.
work_v44vkh6bzramngvcrficy2no44	146	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	146	2	question	question	VBP
work_v44vkh6bzramngvcrficy2no44	146	3	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	146	4	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	146	5	ap-	ap-	NNP
work_v44vkh6bzramngvcrficy2no44	146	6	proach	proach	NN
work_v44vkh6bzramngvcrficy2no44	146	7	–	–	:
work_v44vkh6bzramngvcrficy2no44	146	8	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	146	9	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	146	10	not	not	RB
work_v44vkh6bzramngvcrficy2no44	146	11	conceived	conceive	VBN
work_v44vkh6bzramngvcrficy2no44	146	12	to	to	TO
work_v44vkh6bzramngvcrficy2no44	146	13	solve	solve	VB
work_v44vkh6bzramngvcrficy2no44	146	14	this	this	DT
work_v44vkh6bzramngvcrficy2no44	146	15	spe-	spe-	RB
work_v44vkh6bzramngvcrficy2no44	146	16	cific	cific	JJ
work_v44vkh6bzramngvcrficy2no44	146	17	problem	problem	NN
work_v44vkh6bzramngvcrficy2no44	146	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	146	19	nor	nor	CC
work_v44vkh6bzramngvcrficy2no44	146	20	requires	require	VBZ
work_v44vkh6bzramngvcrficy2no44	146	21	manual	manual	JJ
work_v44vkh6bzramngvcrficy2no44	146	22	rules	rule	NNS
work_v44vkh6bzramngvcrficy2no44	146	23	to	to	TO
work_v44vkh6bzramngvcrficy2no44	146	24	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	146	25	verb	verb	JJ
work_v44vkh6bzramngvcrficy2no44	146	26	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	146	27	–	–	:
work_v44vkh6bzramngvcrficy2no44	146	28	will	will	MD
work_v44vkh6bzramngvcrficy2no44	146	29	succeed	succeed	VB
work_v44vkh6bzramngvcrficy2no44	146	30	in	in	IN
work_v44vkh6bzramngvcrficy2no44	146	31	improving	improve	VBG
work_v44vkh6bzramngvcrficy2no44	146	32	long	long	RB
work_v44vkh6bzramngvcrficy2no44	146	33	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	146	34	ordering	ordering	NN
work_v44vkh6bzramngvcrficy2no44	146	35	in	in	IN
work_v44vkh6bzramngvcrficy2no44	146	36	a	a	DT
work_v44vkh6bzramngvcrficy2no44	146	37	fully	fully	RB
work_v44vkh6bzramngvcrficy2no44	146	38	data	data	NN
work_v44vkh6bzramngvcrficy2no44	146	39	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	146	40	driven	drive	VBN
work_v44vkh6bzramngvcrficy2no44	146	41	way	way	NN
work_v44vkh6bzramngvcrficy2no44	146	42	.	.	.
work_v44vkh6bzramngvcrficy2no44	147	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	147	2	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	147	3	training	training	NN
work_v44vkh6bzramngvcrficy2no44	147	4	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	147	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	147	6	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	147	7	use	use	VBP
work_v44vkh6bzramngvcrficy2no44	147	8	all	all	PDT
work_v44vkh6bzramngvcrficy2no44	147	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	147	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	147	11	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	147	12	domain	domain	NN
work_v44vkh6bzramngvcrficy2no44	147	13	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	147	14	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	147	15	provided	provide	VBN
work_v44vkh6bzramngvcrficy2no44	147	16	for	for	IN
work_v44vkh6bzramngvcrficy2no44	147	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	147	18	NIST	NIST	NNP
work_v44vkh6bzramngvcrficy2no44	147	19	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	147	20	MT09	MT09	NNP
work_v44vkh6bzramngvcrficy2no44	147	21	evalua-	evalua-	NN
work_v44vkh6bzramngvcrficy2no44	147	22	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	147	23	for	for	IN
work_v44vkh6bzramngvcrficy2no44	147	24	a	a	DT
work_v44vkh6bzramngvcrficy2no44	147	25	total	total	NN
work_v44vkh6bzramngvcrficy2no44	147	26	of	of	IN
work_v44vkh6bzramngvcrficy2no44	147	27	986	986	CD
work_v44vkh6bzramngvcrficy2no44	147	28	K	K	NNP
work_v44vkh6bzramngvcrficy2no44	147	29	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	147	30	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	147	31	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	147	32	31	31	CD
work_v44vkh6bzramngvcrficy2no44	147	33	M	M	NNP
work_v44vkh6bzramngvcrficy2no44	147	34	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	147	35	words).7	words).7	NNP
work_v44vkh6bzramngvcrficy2no44	147	36	The	the	DT
work_v44vkh6bzramngvcrficy2no44	147	37	target	target	NN
work_v44vkh6bzramngvcrficy2no44	147	38	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	147	39	used	use	VBD
work_v44vkh6bzramngvcrficy2no44	147	40	to	to	TO
work_v44vkh6bzramngvcrficy2no44	147	41	run	run	VB
work_v44vkh6bzramngvcrficy2no44	147	42	the	the	DT
work_v44vkh6bzramngvcrficy2no44	147	43	main	main	JJ
work_v44vkh6bzramngvcrficy2no44	147	44	se-	se-	NN
work_v44vkh6bzramngvcrficy2no44	147	45	ries	rie	NNS
work_v44vkh6bzramngvcrficy2no44	147	46	of	of	IN
work_v44vkh6bzramngvcrficy2no44	147	47	experiments	experiment	NNS
work_v44vkh6bzramngvcrficy2no44	147	48	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	147	49	trained	train	VBN
work_v44vkh6bzramngvcrficy2no44	147	50	on	on	IN
work_v44vkh6bzramngvcrficy2no44	147	51	the	the	DT
work_v44vkh6bzramngvcrficy2no44	147	52	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	147	53	side	side	NN
work_v44vkh6bzramngvcrficy2no44	147	54	of	of	IN
work_v44vkh6bzramngvcrficy2no44	147	55	all	all	DT
work_v44vkh6bzramngvcrficy2no44	147	56	available	available	JJ
work_v44vkh6bzramngvcrficy2no44	147	57	NIST	NIST	NNP
work_v44vkh6bzramngvcrficy2no44	147	58	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	147	59	MT09	MT09	NNP
work_v44vkh6bzramngvcrficy2no44	147	60	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	147	61	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	147	62	,	,	,
work_v44vkh6bzramngvcrficy2no44	147	63	UN	UN	NNP
work_v44vkh6bzramngvcrficy2no44	147	64	included	include	VBD
work_v44vkh6bzramngvcrficy2no44	147	65	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	147	66	147	147	CD
work_v44vkh6bzramngvcrficy2no44	147	67	M	m	CD
work_v44vkh6bzramngvcrficy2no44	147	68	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	147	69	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	147	70	.	.	.
work_v44vkh6bzramngvcrficy2no44	148	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	148	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	148	3	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	148	4	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	148	5	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	148	6	experiments	experiment	NNS
work_v44vkh6bzramngvcrficy2no44	148	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	148	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	148	9	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	148	10	training	training	NN
work_v44vkh6bzramngvcrficy2no44	148	11	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	148	12	also	also	RB
work_v44vkh6bzramngvcrficy2no44	148	13	include	include	VBP
work_v44vkh6bzramngvcrficy2no44	148	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	148	15	sections	section	NNS
work_v44vkh6bzramngvcrficy2no44	148	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	148	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	148	18	En-	en-	XX
work_v44vkh6bzramngvcrficy2no44	148	19	glish	glish	NN
work_v44vkh6bzramngvcrficy2no44	148	20	Gigaword	Gigaword	NNP
work_v44vkh6bzramngvcrficy2no44	148	21	that	that	DT
work_v44vkh6bzramngvcrficy2no44	148	22	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	148	23	fit	fit	VBP
work_v44vkh6bzramngvcrficy2no44	148	24	to	to	IN
work_v44vkh6bzramngvcrficy2no44	148	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	148	26	development	development	NN
work_v44vkh6bzramngvcrficy2no44	148	27	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	148	28	in	in	IN
work_v44vkh6bzramngvcrficy2no44	148	29	terms	term	NNS
work_v44vkh6bzramngvcrficy2no44	148	30	of	of	IN
work_v44vkh6bzramngvcrficy2no44	148	31	perplexity	perplexity	NN
work_v44vkh6bzramngvcrficy2no44	148	32	:	:	:
work_v44vkh6bzramngvcrficy2no44	148	33	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	148	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	148	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	148	36	Agence	Agence	NNP
work_v44vkh6bzramngvcrficy2no44	148	37	France-	France-	NNP
work_v44vkh6bzramngvcrficy2no44	148	38	Presse	Presse	NNP
work_v44vkh6bzramngvcrficy2no44	148	39	,	,	,
work_v44vkh6bzramngvcrficy2no44	148	40	Xinhua	Xinhua	NNP
work_v44vkh6bzramngvcrficy2no44	148	41	News	News	NNP
work_v44vkh6bzramngvcrficy2no44	148	42	Agency	Agency	NNP
work_v44vkh6bzramngvcrficy2no44	148	43	and	and	CC
work_v44vkh6bzramngvcrficy2no44	148	44	Associated	Associated	NNP
work_v44vkh6bzramngvcrficy2no44	148	45	Press	Press	NNP
work_v44vkh6bzramngvcrficy2no44	148	46	Worldstream	Worldstream	NNP
work_v44vkh6bzramngvcrficy2no44	148	47	sections	section	NNS
work_v44vkh6bzramngvcrficy2no44	148	48	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	148	49	2130	2130	CD
work_v44vkh6bzramngvcrficy2no44	148	50	M	M	NNP
work_v44vkh6bzramngvcrficy2no44	148	51	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	148	52	in	in	IN
work_v44vkh6bzramngvcrficy2no44	148	53	total	total	NN
work_v44vkh6bzramngvcrficy2no44	148	54	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	148	55	.	.	.
work_v44vkh6bzramngvcrficy2no44	149	1	For	for	IN
work_v44vkh6bzramngvcrficy2no44	149	2	development	development	NN
work_v44vkh6bzramngvcrficy2no44	149	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	149	4	test	test	NN
work_v44vkh6bzramngvcrficy2no44	149	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	149	6	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	149	7	use	use	VBP
work_v44vkh6bzramngvcrficy2no44	149	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	149	9	newswire	newswire	JJ
work_v44vkh6bzramngvcrficy2no44	149	10	sections	section	NNS
work_v44vkh6bzramngvcrficy2no44	149	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	149	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	149	13	NIST	NIST	NNP
work_v44vkh6bzramngvcrficy2no44	149	14	benchmarks	benchmark	NNS
work_v44vkh6bzramngvcrficy2no44	149	15	:	:	:
work_v44vkh6bzramngvcrficy2no44	149	16	dev06-nw	dev06-nw	NNP
work_v44vkh6bzramngvcrficy2no44	149	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	149	18	eval08-	eval08-	NNP
work_v44vkh6bzramngvcrficy2no44	149	19	nw	nw	RB
work_v44vkh6bzramngvcrficy2no44	149	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	149	21	eval09-nw	eval09-nw	NNP
work_v44vkh6bzramngvcrficy2no44	149	22	consisting	consist	VBG
work_v44vkh6bzramngvcrficy2no44	149	23	of	of	IN
work_v44vkh6bzramngvcrficy2no44	149	24	1033	1033	CD
work_v44vkh6bzramngvcrficy2no44	149	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	149	26	813	813	CD
work_v44vkh6bzramngvcrficy2no44	149	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	149	28	586	586	CD
work_v44vkh6bzramngvcrficy2no44	149	29	sen-	sen-	NN
work_v44vkh6bzramngvcrficy2no44	149	30	tences	tence	NNS
work_v44vkh6bzramngvcrficy2no44	149	31	respectively	respectively	RB
work_v44vkh6bzramngvcrficy2no44	149	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	150	1	Each	each	DT
work_v44vkh6bzramngvcrficy2no44	150	2	set	set	NN
work_v44vkh6bzramngvcrficy2no44	150	3	includes	include	VBZ
work_v44vkh6bzramngvcrficy2no44	150	4	4	4	CD
work_v44vkh6bzramngvcrficy2no44	150	5	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	150	6	translations	translation	NNS
work_v44vkh6bzramngvcrficy2no44	150	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	150	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	150	9	average	average	JJ
work_v44vkh6bzramngvcrficy2no44	150	10	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	150	11	length	length	NN
work_v44vkh6bzramngvcrficy2no44	150	12	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	150	13	33	33	CD
work_v44vkh6bzramngvcrficy2no44	150	14	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	150	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	151	1	To	to	TO
work_v44vkh6bzramngvcrficy2no44	151	2	focus	focus	VB
work_v44vkh6bzramngvcrficy2no44	151	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	151	4	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	151	5	on	on	IN
work_v44vkh6bzramngvcrficy2no44	151	6	problematic	problematic	JJ
work_v44vkh6bzramngvcrficy2no44	151	7	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	151	8	ordering	ordering	NN
work_v44vkh6bzramngvcrficy2no44	151	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	151	10	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	151	11	also	also	RB
work_v44vkh6bzramngvcrficy2no44	151	12	consider	consider	VBP
work_v44vkh6bzramngvcrficy2no44	151	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	151	14	subset	subset	NN
work_v44vkh6bzramngvcrficy2no44	151	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	151	16	eval09-nw	eval09-nw	NN
work_v44vkh6bzramngvcrficy2no44	151	17	containing	contain	VBG
work_v44vkh6bzramngvcrficy2no44	151	18	only	only	JJ
work_v44vkh6bzramngvcrficy2no44	151	19	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	151	20	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	151	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	151	22	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	151	23	main	main	JJ
work_v44vkh6bzramngvcrficy2no44	151	24	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	151	25	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	151	26	placed	place	VBN
work_v44vkh6bzramngvcrficy2no44	151	27	before	before	IN
work_v44vkh6bzramngvcrficy2no44	151	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	151	29	subject	subject	NN
work_v44vkh6bzramngvcrficy2no44	151	30	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	151	31	vs-09	vs-09	NNP
work_v44vkh6bzramngvcrficy2no44	151	32	:	:	:
work_v44vkh6bzramngvcrficy2no44	151	33	299	299	CD
work_v44vkh6bzramngvcrficy2no44	151	34	sent	send	VBD
work_v44vkh6bzramngvcrficy2no44	151	35	.	.	.
work_v44vkh6bzramngvcrficy2no44	152	1	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	152	2	.8	.8	NFP
work_v44vkh6bzramngvcrficy2no44	152	3	As	as	IN
work_v44vkh6bzramngvcrficy2no44	152	4	pre	pre	JJ
work_v44vkh6bzramngvcrficy2no44	152	5	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	152	6	processing	process	VBG
work_v44vkh6bzramngvcrficy2no44	152	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	152	8	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	152	9	apply	apply	VBP
work_v44vkh6bzramngvcrficy2no44	152	10	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	152	11	tokeniza-	tokeniza-	NN
work_v44vkh6bzramngvcrficy2no44	152	12	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	152	13	to	to	IN
work_v44vkh6bzramngvcrficy2no44	152	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	152	15	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	152	16	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	152	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	152	18	while	while	IN
work_v44vkh6bzramngvcrficy2no44	152	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	152	20	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	152	21	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	152	22	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	152	23	segmented	segment	VBN
work_v44vkh6bzramngvcrficy2no44	152	24	with	with	IN
work_v44vkh6bzramngvcrficy2no44	152	25	AMIRA	AMIRA	NNP
work_v44vkh6bzramngvcrficy2no44	152	26	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	152	27	Diab	Diab	NNP
work_v44vkh6bzramngvcrficy2no44	152	28	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	152	29	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	152	30	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	152	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	152	32	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	152	33	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	152	34	accord-	accord-	NN
work_v44vkh6bzramngvcrficy2no44	152	35	ing	e	VBG
work_v44vkh6bzramngvcrficy2no44	152	36	to	to	IN
work_v44vkh6bzramngvcrficy2no44	152	37	the	the	DT
work_v44vkh6bzramngvcrficy2no44	152	38	ATB	ATB	NNP
work_v44vkh6bzramngvcrficy2no44	152	39	scheme9	scheme9	NN
work_v44vkh6bzramngvcrficy2no44	152	40	.	.	.
work_v44vkh6bzramngvcrficy2no44	153	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	153	2	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	153	3	tool	tool	NN
work_v44vkh6bzramngvcrficy2no44	153	4	also	also	RB
work_v44vkh6bzramngvcrficy2no44	153	5	pro-	pro-	VBP
work_v44vkh6bzramngvcrficy2no44	153	6	duces	duce	VBZ
work_v44vkh6bzramngvcrficy2no44	153	7	POS	pos	NN
work_v44vkh6bzramngvcrficy2no44	153	8	tagging	tag	VBG
work_v44vkh6bzramngvcrficy2no44	153	9	and	and	CC
work_v44vkh6bzramngvcrficy2no44	153	10	shallow	shallow	JJ
work_v44vkh6bzramngvcrficy2no44	153	11	syntax	syntax	NN
work_v44vkh6bzramngvcrficy2no44	153	12	annotation	annotation	NN
work_v44vkh6bzramngvcrficy2no44	153	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	154	1	7The	7The	``
work_v44vkh6bzramngvcrficy2no44	154	2	in	in	IN
work_v44vkh6bzramngvcrficy2no44	154	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	154	4	domain	domain	NN
work_v44vkh6bzramngvcrficy2no44	154	5	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	154	6	data	data	NN
work_v44vkh6bzramngvcrficy2no44	154	7	includes	include	VBZ
work_v44vkh6bzramngvcrficy2no44	154	8	all	all	PDT
work_v44vkh6bzramngvcrficy2no44	154	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	154	10	provided	provide	VBN
work_v44vkh6bzramngvcrficy2no44	154	11	cor-	cor-	NNP
work_v44vkh6bzramngvcrficy2no44	154	12	pora	pora	NNP
work_v44vkh6bzramngvcrficy2no44	154	13	except	except	IN
work_v44vkh6bzramngvcrficy2no44	154	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	154	15	UN	UN	NNP
work_v44vkh6bzramngvcrficy2no44	154	16	proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	154	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	154	18	and	and	CC
work_v44vkh6bzramngvcrficy2no44	154	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	154	20	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	154	21	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	154	22	newswire	newswire	JJ
work_v44vkh6bzramngvcrficy2no44	154	23	parts	part	NNS
work_v44vkh6bzramngvcrficy2no44	154	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	154	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	154	26	small	small	JJ
work_v44vkh6bzramngvcrficy2no44	154	27	GALE	GALE	NNP
work_v44vkh6bzramngvcrficy2no44	154	28	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	154	29	Y1-Q4	Y1-Q4	NNP
work_v44vkh6bzramngvcrficy2no44	154	30	corpus	corpus	NNP
work_v44vkh6bzramngvcrficy2no44	154	31	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	154	32	that	that	RB
work_v44vkh6bzramngvcrficy2no44	154	33	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	154	34	9	9	CD
work_v44vkh6bzramngvcrficy2no44	154	35	K	K	NNP
work_v44vkh6bzramngvcrficy2no44	154	36	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	154	37	of	of	IN
work_v44vkh6bzramngvcrficy2no44	154	38	audio	audio	JJ
work_v44vkh6bzramngvcrficy2no44	154	39	transcripts	transcript	NNS
work_v44vkh6bzramngvcrficy2no44	154	40	and	and	CC
work_v44vkh6bzramngvcrficy2no44	154	41	web	web	NN
work_v44vkh6bzramngvcrficy2no44	154	42	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	154	43	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	154	44	.	.	.
work_v44vkh6bzramngvcrficy2no44	155	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	155	2	reported	report	VBN
work_v44vkh6bzramngvcrficy2no44	155	3	by	by	IN
work_v44vkh6bzramngvcrficy2no44	155	4	Green	Green	NNP
work_v44vkh6bzramngvcrficy2no44	155	5	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	155	6	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	155	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	156	1	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	156	2	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	156	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	156	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	156	5	removal	removal	NN
work_v44vkh6bzramngvcrficy2no44	156	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	156	7	UN	UN	NNP
work_v44vkh6bzramngvcrficy2no44	156	8	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	156	9	does	do	VBZ
work_v44vkh6bzramngvcrficy2no44	156	10	not	not	RB
work_v44vkh6bzramngvcrficy2no44	156	11	affect	affect	VB
work_v44vkh6bzramngvcrficy2no44	156	12	baseline	baseline	JJ
work_v44vkh6bzramngvcrficy2no44	156	13	performances	performance	NNS
work_v44vkh6bzramngvcrficy2no44	156	14	on	on	IN
work_v44vkh6bzramngvcrficy2no44	156	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	156	16	news	news	NN
work_v44vkh6bzramngvcrficy2no44	156	17	benchmarks	benchmark	NNS
work_v44vkh6bzramngvcrficy2no44	156	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	157	1	8Automatically	8Automatically	NNP
work_v44vkh6bzramngvcrficy2no44	157	2	detected	detect	VBN
work_v44vkh6bzramngvcrficy2no44	157	3	by	by	IN
work_v44vkh6bzramngvcrficy2no44	157	4	means	mean	NNS
work_v44vkh6bzramngvcrficy2no44	157	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	157	6	shallow	shallow	JJ
work_v44vkh6bzramngvcrficy2no44	157	7	syntax	syntax	NN
work_v44vkh6bzramngvcrficy2no44	157	8	rules	rule	NNS
work_v44vkh6bzramngvcrficy2no44	157	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	158	1	9The	9the	CD
work_v44vkh6bzramngvcrficy2no44	158	2	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	158	3	Treebank	Treebank	NNP
work_v44vkh6bzramngvcrficy2no44	158	4	tokenization	tokenization	NN
work_v44vkh6bzramngvcrficy2no44	158	5	scheme	scheme	NN
work_v44vkh6bzramngvcrficy2no44	158	6	isolates	isolate	VBZ
work_v44vkh6bzramngvcrficy2no44	158	7	con-	con-	NNP
work_v44vkh6bzramngvcrficy2no44	158	8	junctions	junctions	NNPS
work_v44vkh6bzramngvcrficy2no44	158	9	w+	w+	NNP
work_v44vkh6bzramngvcrficy2no44	158	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	158	11	f+	f+	NNP
work_v44vkh6bzramngvcrficy2no44	158	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	158	13	prepositions	preposition	NNS
work_v44vkh6bzramngvcrficy2no44	158	14	l+	l+	$
work_v44vkh6bzramngvcrficy2no44	158	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	158	16	k+	k+	NNP
work_v44vkh6bzramngvcrficy2no44	158	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	158	18	b+	b+	NNP
work_v44vkh6bzramngvcrficy2no44	158	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	158	20	future	future	JJ
work_v44vkh6bzramngvcrficy2no44	158	21	marker	marker	NN
work_v44vkh6bzramngvcrficy2no44	158	22	s+	s+	CD
work_v44vkh6bzramngvcrficy2no44	158	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	158	24	pronominal	pronominal	JJ
work_v44vkh6bzramngvcrficy2no44	158	25	suffixes	suffix	NNS
work_v44vkh6bzramngvcrficy2no44	158	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	158	27	but	but	CC
work_v44vkh6bzramngvcrficy2no44	158	28	not	not	RB
work_v44vkh6bzramngvcrficy2no44	158	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	158	30	article	article	NN
work_v44vkh6bzramngvcrficy2no44	158	31	Al+	Al+	NNP
work_v44vkh6bzramngvcrficy2no44	158	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	159	1	332	332	CD
work_v44vkh6bzramngvcrficy2no44	159	2	5.1	5.1	CD
work_v44vkh6bzramngvcrficy2no44	159	3	Reordering	Reordering	NNP
work_v44vkh6bzramngvcrficy2no44	159	4	model	model	NN
work_v44vkh6bzramngvcrficy2no44	159	5	intrinsic	intrinsic	JJ
work_v44vkh6bzramngvcrficy2no44	159	6	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	159	7	Before	before	IN
work_v44vkh6bzramngvcrficy2no44	159	8	proceeding	proceed	VBG
work_v44vkh6bzramngvcrficy2no44	159	9	to	to	IN
work_v44vkh6bzramngvcrficy2no44	159	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	159	11	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	159	12	experiments	experiment	NNS
work_v44vkh6bzramngvcrficy2no44	159	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	159	14	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	159	15	evaluate	evaluate	VBP
work_v44vkh6bzramngvcrficy2no44	159	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	159	17	performance	performance	NN
work_v44vkh6bzramngvcrficy2no44	159	18	of	of	IN
work_v44vkh6bzramngvcrficy2no44	159	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	159	20	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	159	21	reorder-	reorder-	NNP
work_v44vkh6bzramngvcrficy2no44	159	22	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	159	23	model	model	NN
work_v44vkh6bzramngvcrficy2no44	159	24	in	in	IN
work_v44vkh6bzramngvcrficy2no44	159	25	isolation	isolation	NN
work_v44vkh6bzramngvcrficy2no44	159	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	160	1	All	all	PDT
work_v44vkh6bzramngvcrficy2no44	160	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	160	3	tested	test	VBN
work_v44vkh6bzramngvcrficy2no44	160	4	configura-	configura-	JJ
work_v44vkh6bzramngvcrficy2no44	160	5	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	160	6	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	160	7	trained	train	VBN
work_v44vkh6bzramngvcrficy2no44	160	8	with	with	IN
work_v44vkh6bzramngvcrficy2no44	160	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	160	10	freely	freely	RB
work_v44vkh6bzramngvcrficy2no44	160	11	available	available	JJ
work_v44vkh6bzramngvcrficy2no44	160	12	MegaM	MegaM	NNP
work_v44vkh6bzramngvcrficy2no44	160	13	Toolkit10	Toolkit10	NNP
work_v44vkh6bzramngvcrficy2no44	160	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	160	15	implementing	implement	VBG
work_v44vkh6bzramngvcrficy2no44	160	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	160	17	conjugate	conjugate	JJ
work_v44vkh6bzramngvcrficy2no44	160	18	gradient	gradient	NN
work_v44vkh6bzramngvcrficy2no44	160	19	method	method	NN
work_v44vkh6bzramngvcrficy2no44	160	20	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	160	21	Hestenes	Hestenes	NNP
work_v44vkh6bzramngvcrficy2no44	160	22	and	and	CC
work_v44vkh6bzramngvcrficy2no44	160	23	Stiefel	Stiefel	NNP
work_v44vkh6bzramngvcrficy2no44	160	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	160	25	1952	1952	CD
work_v44vkh6bzramngvcrficy2no44	160	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	160	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	160	28	in	in	IN
work_v44vkh6bzramngvcrficy2no44	160	29	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	160	30	100	100	CD
work_v44vkh6bzramngvcrficy2no44	160	31	iterations	iteration	NNS
work_v44vkh6bzramngvcrficy2no44	160	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	161	1	Training	training	NN
work_v44vkh6bzramngvcrficy2no44	161	2	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	161	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	161	4	generated	generate	VBN
work_v44vkh6bzramngvcrficy2no44	161	5	within	within	IN
work_v44vkh6bzramngvcrficy2no44	161	6	a	a	DT
work_v44vkh6bzramngvcrficy2no44	161	7	sampling	sample	VBG
work_v44vkh6bzramngvcrficy2no44	161	8	window	window	NN
work_v44vkh6bzramngvcrficy2no44	161	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	161	10	width	width	NN
work_v44vkh6bzramngvcrficy2no44	161	11	#	#	NN
work_v44vkh6bzramngvcrficy2no44	161	12	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	161	13	10	10	CD
work_v44vkh6bzramngvcrficy2no44	161	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	161	15	from	from	IN
work_v44vkh6bzramngvcrficy2no44	161	16	a	a	DT
work_v44vkh6bzramngvcrficy2no44	161	17	subset	subset	NN
work_v44vkh6bzramngvcrficy2no44	161	18	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	161	19	30	30	CD
work_v44vkh6bzramngvcrficy2no44	161	20	K	k	NN
work_v44vkh6bzramngvcrficy2no44	161	21	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	161	22	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	161	23	of	of	IN
work_v44vkh6bzramngvcrficy2no44	161	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	161	25	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	161	26	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	161	27	described	describe	VBN
work_v44vkh6bzramngvcrficy2no44	161	28	above	above	RB
work_v44vkh6bzramngvcrficy2no44	161	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	161	30	resulting	result	VBG
work_v44vkh6bzramngvcrficy2no44	161	31	in	in	IN
work_v44vkh6bzramngvcrficy2no44	161	32	8	8	CD
work_v44vkh6bzramngvcrficy2no44	161	33	M	m	NN
work_v44vkh6bzramngvcrficy2no44	161	34	training	training	NN
work_v44vkh6bzramngvcrficy2no44	161	35	word	word	NN
work_v44vkh6bzramngvcrficy2no44	161	36	pairs11	pairs11	NN
work_v44vkh6bzramngvcrficy2no44	161	37	.	.	.
work_v44vkh6bzramngvcrficy2no44	162	1	Test	test	NN
work_v44vkh6bzramngvcrficy2no44	162	2	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	162	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	162	4	generated	generate	VBN
work_v44vkh6bzramngvcrficy2no44	162	5	from	from	IN
work_v44vkh6bzramngvcrficy2no44	162	6	TIDES	TIDES	NNP
work_v44vkh6bzramngvcrficy2no44	162	7	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	162	8	MT04	MT04	NNP
work_v44vkh6bzramngvcrficy2no44	162	9	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	162	10	1324	1324	CD
work_v44vkh6bzramngvcrficy2no44	162	11	sen-	sen-	NN
work_v44vkh6bzramngvcrficy2no44	162	12	tences	tence	NNS
work_v44vkh6bzramngvcrficy2no44	162	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	162	14	370	370	CD
work_v44vkh6bzramngvcrficy2no44	162	15	K	k	NN
work_v44vkh6bzramngvcrficy2no44	162	16	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	162	17	with	with	IN
work_v44vkh6bzramngvcrficy2no44	162	18	#	#	$
work_v44vkh6bzramngvcrficy2no44	162	19	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	162	20	10	10	CD
work_v44vkh6bzramngvcrficy2no44	162	21	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	162	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	162	23	one	one	CD
work_v44vkh6bzramngvcrficy2no44	162	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	162	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	162	26	corpora	corpora	NN
work_v44vkh6bzramngvcrficy2no44	162	27	included	include	VBN
work_v44vkh6bzramngvcrficy2no44	162	28	in	in	IN
work_v44vkh6bzramngvcrficy2no44	162	29	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	162	30	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	162	31	training	training	NN
work_v44vkh6bzramngvcrficy2no44	162	32	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	162	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	163	1	Features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	163	2	with	with	IN
work_v44vkh6bzramngvcrficy2no44	163	3	less	less	JJR
work_v44vkh6bzramngvcrficy2no44	163	4	than	than	IN
work_v44vkh6bzramngvcrficy2no44	163	5	20	20	CD
work_v44vkh6bzramngvcrficy2no44	163	6	occurrences	occurrence	NNS
work_v44vkh6bzramngvcrficy2no44	163	7	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	163	8	ignored	ignore	VBN
work_v44vkh6bzramngvcrficy2no44	163	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	164	1	Classification	classification	NN
work_v44vkh6bzramngvcrficy2no44	164	2	accuracy	accuracy	NN
work_v44vkh6bzramngvcrficy2no44	164	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	165	1	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	165	2	3	3	CD
work_v44vkh6bzramngvcrficy2no44	165	3	presents	present	NNS
work_v44vkh6bzramngvcrficy2no44	165	4	preci-	preci-	JJ
work_v44vkh6bzramngvcrficy2no44	165	5	sion	sion	NN
work_v44vkh6bzramngvcrficy2no44	165	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	165	7	recall	recall	NN
work_v44vkh6bzramngvcrficy2no44	165	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	165	9	and	and	CC
work_v44vkh6bzramngvcrficy2no44	165	10	F	F	NNP
work_v44vkh6bzramngvcrficy2no44	165	11	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	165	12	score	score	NN
work_v44vkh6bzramngvcrficy2no44	165	13	achieved	achieve	VBN
work_v44vkh6bzramngvcrficy2no44	165	14	by	by	IN
work_v44vkh6bzramngvcrficy2no44	165	15	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	165	16	fea-	fea-	NN
work_v44vkh6bzramngvcrficy2no44	165	17	ture	ture	NN
work_v44vkh6bzramngvcrficy2no44	165	18	subsets	subset	NNS
work_v44vkh6bzramngvcrficy2no44	165	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	165	20	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	165	21	W	w	NN
work_v44vkh6bzramngvcrficy2no44	165	22	stands	stand	VBZ
work_v44vkh6bzramngvcrficy2no44	165	23	for	for	IN
work_v44vkh6bzramngvcrficy2no44	165	24	word	word	NN
work_v44vkh6bzramngvcrficy2no44	165	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	165	26	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	165	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	165	28	P	p	NN
work_v44vkh6bzramngvcrficy2no44	165	29	for	for	IN
work_v44vkh6bzramngvcrficy2no44	165	30	POS	POS	NNP
work_v44vkh6bzramngvcrficy2no44	165	31	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	165	32	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	165	33	and	and	CC
work_v44vkh6bzramngvcrficy2no44	165	34	C	C	NNP
work_v44vkh6bzramngvcrficy2no44	165	35	for	for	IN
work_v44vkh6bzramngvcrficy2no44	165	36	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	165	37	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	165	38	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	165	39	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	165	40	templates	template	NNS
work_v44vkh6bzramngvcrficy2no44	165	41	.	.	.
work_v44vkh6bzramngvcrficy2no44	166	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	166	2	can	can	MD
work_v44vkh6bzramngvcrficy2no44	166	3	see	see	VB
work_v44vkh6bzramngvcrficy2no44	166	4	that	that	IN
work_v44vkh6bzramngvcrficy2no44	166	5	all	all	DT
work_v44vkh6bzramngvcrficy2no44	166	6	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	166	7	types	type	NNS
work_v44vkh6bzramngvcrficy2no44	166	8	contribute	contribute	VBP
work_v44vkh6bzramngvcrficy2no44	166	9	to	to	TO
work_v44vkh6bzramngvcrficy2no44	166	10	im-	im-	RB
work_v44vkh6bzramngvcrficy2no44	166	11	prove	prove	VB
work_v44vkh6bzramngvcrficy2no44	166	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	166	13	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	166	14	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	166	15	performance	performance	NN
work_v44vkh6bzramngvcrficy2no44	166	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	167	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	167	2	word	word	NN
work_v44vkh6bzramngvcrficy2no44	167	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	167	4	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	167	5	model	model	NN
work_v44vkh6bzramngvcrficy2no44	167	6	achieves	achieve	VBZ
work_v44vkh6bzramngvcrficy2no44	167	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	167	8	highest	high	JJS
work_v44vkh6bzramngvcrficy2no44	167	9	precision	precision	NN
work_v44vkh6bzramngvcrficy2no44	167	10	but	but	CC
work_v44vkh6bzramngvcrficy2no44	167	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	167	12	very	very	RB
work_v44vkh6bzramngvcrficy2no44	167	13	low	low	JJ
work_v44vkh6bzramngvcrficy2no44	167	14	recall	recall	NN
work_v44vkh6bzramngvcrficy2no44	167	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	167	16	while	while	IN
work_v44vkh6bzramngvcrficy2no44	167	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	167	18	POS	POS	NNP
work_v44vkh6bzramngvcrficy2no44	167	19	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	167	20	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	167	21	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	167	22	much	much	RB
work_v44vkh6bzramngvcrficy2no44	167	23	more	more	JJR
work_v44vkh6bzramngvcrficy2no44	167	24	bal-	bal-	JJ
work_v44vkh6bzramngvcrficy2no44	167	25	anced	anced	JJ
work_v44vkh6bzramngvcrficy2no44	167	26	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	167	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	168	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	168	2	better	well	JJR
work_v44vkh6bzramngvcrficy2no44	168	3	performance	performance	NN
work_v44vkh6bzramngvcrficy2no44	168	4	overall	overall	NN
work_v44vkh6bzramngvcrficy2no44	168	5	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	168	6	ob-	ob-	RB
work_v44vkh6bzramngvcrficy2no44	168	7	tained	taine	VBN
work_v44vkh6bzramngvcrficy2no44	168	8	by	by	IN
work_v44vkh6bzramngvcrficy2no44	168	9	combining	combine	VBG
work_v44vkh6bzramngvcrficy2no44	168	10	word-	word-	NN
work_v44vkh6bzramngvcrficy2no44	168	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	168	12	POS-	pos-	JJ
work_v44vkh6bzramngvcrficy2no44	168	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	168	14	mixed	mixed	JJ
work_v44vkh6bzramngvcrficy2no44	168	15	word-	word-	NN
work_v44vkh6bzramngvcrficy2no44	168	16	POS	POS	NNP
work_v44vkh6bzramngvcrficy2no44	168	17	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	168	18	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	168	19	features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	168	20	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	168	21	62.6	62.6	CD
work_v44vkh6bzramngvcrficy2no44	168	22	%	%	NN
work_v44vkh6bzramngvcrficy2no44	168	23	F	F	NNP
work_v44vkh6bzramngvcrficy2no44	168	24	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	168	25	score	score	NN
work_v44vkh6bzramngvcrficy2no44	168	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	168	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	169	1	Finally	finally	RB
work_v44vkh6bzramngvcrficy2no44	169	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	169	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	169	4	addition	addition	NN
work_v44vkh6bzramngvcrficy2no44	169	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	169	6	chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	169	7	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	169	8	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	169	9	features	feature	VBZ
work_v44vkh6bzramngvcrficy2no44	169	10	yields	yield	NNS
work_v44vkh6bzramngvcrficy2no44	169	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	169	12	further	further	JJ
work_v44vkh6bzramngvcrficy2no44	169	13	im-	im-	XX
work_v44vkh6bzramngvcrficy2no44	169	14	provement	provement	NN
work_v44vkh6bzramngvcrficy2no44	169	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	169	16	about	about	RB
work_v44vkh6bzramngvcrficy2no44	169	17	1	1	CD
work_v44vkh6bzramngvcrficy2no44	169	18	point	point	NN
work_v44vkh6bzramngvcrficy2no44	169	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	169	20	reaching	reach	VBG
work_v44vkh6bzramngvcrficy2no44	169	21	63.8	63.8	CD
work_v44vkh6bzramngvcrficy2no44	169	22	%	%	NN
work_v44vkh6bzramngvcrficy2no44	169	23	F	F	NNP
work_v44vkh6bzramngvcrficy2no44	169	24	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	169	25	score	score	NN
work_v44vkh6bzramngvcrficy2no44	169	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	170	1	Given	give	VBN
work_v44vkh6bzramngvcrficy2no44	170	2	these	these	DT
work_v44vkh6bzramngvcrficy2no44	170	3	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	170	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	170	5	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	170	6	decide	decide	VBP
work_v44vkh6bzramngvcrficy2no44	170	7	to	to	TO
work_v44vkh6bzramngvcrficy2no44	170	8	use	use	VB
work_v44vkh6bzramngvcrficy2no44	170	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	170	10	W	w	NN
work_v44vkh6bzramngvcrficy2no44	170	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	170	12	P	p	NN
work_v44vkh6bzramngvcrficy2no44	170	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	170	14	C	c	NN
work_v44vkh6bzramngvcrficy2no44	170	15	model	model	NN
work_v44vkh6bzramngvcrficy2no44	170	16	for	for	IN
work_v44vkh6bzramngvcrficy2no44	170	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	170	18	rest	rest	NN
work_v44vkh6bzramngvcrficy2no44	170	19	of	of	IN
work_v44vkh6bzramngvcrficy2no44	170	20	the	the	DT
work_v44vkh6bzramngvcrficy2no44	170	21	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	170	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	171	1	Features	feature	NNS
work_v44vkh6bzramngvcrficy2no44	171	2	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	171	3	templates	template	NNS
work_v44vkh6bzramngvcrficy2no44	171	4	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	171	5	P	p	NN
work_v44vkh6bzramngvcrficy2no44	171	6	R	r	NN
work_v44vkh6bzramngvcrficy2no44	171	7	F	f	NN
work_v44vkh6bzramngvcrficy2no44	171	8	W	w	NN
work_v44vkh6bzramngvcrficy2no44	171	9	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	171	10	1	1	CD
work_v44vkh6bzramngvcrficy2no44	171	11	-	-	SYM
work_v44vkh6bzramngvcrficy2no44	171	12	7	7	CD
work_v44vkh6bzramngvcrficy2no44	171	13	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	171	14	73.1	73.1	CD
work_v44vkh6bzramngvcrficy2no44	171	15	16.4	16.4	CD
work_v44vkh6bzramngvcrficy2no44	171	16	26.8	26.8	CD
work_v44vkh6bzramngvcrficy2no44	171	17	P	p	NN
work_v44vkh6bzramngvcrficy2no44	171	18	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	171	19	8	8	CD
work_v44vkh6bzramngvcrficy2no44	171	20	-	-	SYM
work_v44vkh6bzramngvcrficy2no44	171	21	15	15	CD
work_v44vkh6bzramngvcrficy2no44	171	22	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	171	23	69.5	69.5	CD
work_v44vkh6bzramngvcrficy2no44	171	24	54.8	54.8	CD
work_v44vkh6bzramngvcrficy2no44	171	25	61.3	61.3	CD
work_v44vkh6bzramngvcrficy2no44	171	26	W	w	NN
work_v44vkh6bzramngvcrficy2no44	171	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	171	28	P	p	NN
work_v44vkh6bzramngvcrficy2no44	171	29	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	171	30	1	1	CD
work_v44vkh6bzramngvcrficy2no44	171	31	-	-	SYM
work_v44vkh6bzramngvcrficy2no44	171	32	17	17	CD
work_v44vkh6bzramngvcrficy2no44	171	33	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	171	34	70.2	70.2	CD
work_v44vkh6bzramngvcrficy2no44	171	35	56.5	56.5	CD
work_v44vkh6bzramngvcrficy2no44	171	36	62.6	62.6	CD
work_v44vkh6bzramngvcrficy2no44	171	37	W	W	NNP
work_v44vkh6bzramngvcrficy2no44	171	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	171	39	P	p	NN
work_v44vkh6bzramngvcrficy2no44	171	40	,	,	,
work_v44vkh6bzramngvcrficy2no44	171	41	C	C	NNP
work_v44vkh6bzramngvcrficy2no44	171	42	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	171	43	1	1	CD
work_v44vkh6bzramngvcrficy2no44	171	44	-	-	SYM
work_v44vkh6bzramngvcrficy2no44	171	45	21	21	CD
work_v44vkh6bzramngvcrficy2no44	171	46	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	171	47	70.6	70.6	CD
work_v44vkh6bzramngvcrficy2no44	171	48	58.1	58.1	CD
work_v44vkh6bzramngvcrficy2no44	171	49	63.8	63.8	CD
work_v44vkh6bzramngvcrficy2no44	171	50	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	171	51	3	3	CD
work_v44vkh6bzramngvcrficy2no44	171	52	:	:	:
work_v44vkh6bzramngvcrficy2no44	171	53	Classification	classification	NN
work_v44vkh6bzramngvcrficy2no44	171	54	accuracy	accuracy	NN
work_v44vkh6bzramngvcrficy2no44	171	55	of	of	IN
work_v44vkh6bzramngvcrficy2no44	171	56	the	the	DT
work_v44vkh6bzramngvcrficy2no44	171	57	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	171	58	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	171	59	model	model	NN
work_v44vkh6bzramngvcrficy2no44	171	60	on	on	IN
work_v44vkh6bzramngvcrficy2no44	171	61	TIDES	TIDES	NNP
work_v44vkh6bzramngvcrficy2no44	171	62	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	171	63	MT04	MT04	NNP
work_v44vkh6bzramngvcrficy2no44	171	64	,	,	,
work_v44vkh6bzramngvcrficy2no44	171	65	using	use	VBG
work_v44vkh6bzramngvcrficy2no44	171	66	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	171	67	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	171	68	subsets	subset	NNS
work_v44vkh6bzramngvcrficy2no44	171	69	.	.	.
work_v44vkh6bzramngvcrficy2no44	172	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	172	2	template	template	NN
work_v44vkh6bzramngvcrficy2no44	172	3	numbers	number	NNS
work_v44vkh6bzramngvcrficy2no44	172	4	refer	refer	VBP
work_v44vkh6bzramngvcrficy2no44	172	5	to	to	IN
work_v44vkh6bzramngvcrficy2no44	172	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	172	7	rows	row	NNS
work_v44vkh6bzramngvcrficy2no44	172	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	172	9	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	172	10	1	1	CD
work_v44vkh6bzramngvcrficy2no44	172	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	173	1	Ranking	rank	VBG
work_v44vkh6bzramngvcrficy2no44	173	2	accuracy	accuracy	NN
work_v44vkh6bzramngvcrficy2no44	173	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	174	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	174	2	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	174	3	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	174	4	aspect	aspect	NN
work_v44vkh6bzramngvcrficy2no44	174	5	to	to	TO
work_v44vkh6bzramngvcrficy2no44	174	6	evaluate	evaluate	VB
work_v44vkh6bzramngvcrficy2no44	174	7	for	for	IN
work_v44vkh6bzramngvcrficy2no44	174	8	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	174	9	application	application	NN
work_v44vkh6bzramngvcrficy2no44	174	10	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	174	11	how	how	WRB
work_v44vkh6bzramngvcrficy2no44	174	12	well	well	RB
work_v44vkh6bzramngvcrficy2no44	174	13	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	174	14	model	model	NN
work_v44vkh6bzramngvcrficy2no44	174	15	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	174	16	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	174	17	can	can	MD
work_v44vkh6bzramngvcrficy2no44	174	18	rank	rank	VB
work_v44vkh6bzramngvcrficy2no44	174	19	a	a	DT
work_v44vkh6bzramngvcrficy2no44	174	20	typical	typical	JJ
work_v44vkh6bzramngvcrficy2no44	174	21	set	set	NN
work_v44vkh6bzramngvcrficy2no44	174	22	of	of	IN
work_v44vkh6bzramngvcrficy2no44	174	23	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	174	24	options	option	NNS
work_v44vkh6bzramngvcrficy2no44	174	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	175	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	175	2	fact	fact	NN
work_v44vkh6bzramngvcrficy2no44	175	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	175	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	175	5	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	175	6	model	model	NN
work_v44vkh6bzramngvcrficy2no44	175	7	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	175	8	not	not	RB
work_v44vkh6bzramngvcrficy2no44	175	9	meant	mean	VBN
work_v44vkh6bzramngvcrficy2no44	175	10	to	to	TO
work_v44vkh6bzramngvcrficy2no44	175	11	be	be	VB
work_v44vkh6bzramngvcrficy2no44	175	12	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	175	13	as	as	IN
work_v44vkh6bzramngvcrficy2no44	175	14	10http://www.cs.utah.edu/˜hal/megam/	10http://www.cs.utah.edu/˜hal/megam/	CD
work_v44vkh6bzramngvcrficy2no44	175	15	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	175	16	Daumé	Daumé	NNP
work_v44vkh6bzramngvcrficy2no44	175	17	III	iii	CD
work_v44vkh6bzramngvcrficy2no44	175	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	175	19	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	175	20	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	175	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	176	1	11This	11this	CD
work_v44vkh6bzramngvcrficy2no44	176	2	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	176	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	176	4	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	176	5	number	number	NN
work_v44vkh6bzramngvcrficy2no44	176	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	176	7	samples	sample	NNS
work_v44vkh6bzramngvcrficy2no44	176	8	manageable	manageable	JJ
work_v44vkh6bzramngvcrficy2no44	176	9	by	by	IN
work_v44vkh6bzramngvcrficy2no44	176	10	MegaM.	megam.	NN
work_v44vkh6bzramngvcrficy2no44	177	1	However	however	RB
work_v44vkh6bzramngvcrficy2no44	177	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	177	3	even	even	RB
work_v44vkh6bzramngvcrficy2no44	177	4	scaling	scale	VBG
work_v44vkh6bzramngvcrficy2no44	177	5	from	from	IN
work_v44vkh6bzramngvcrficy2no44	177	6	4	4	CD
work_v44vkh6bzramngvcrficy2no44	177	7	M	M	NNP
work_v44vkh6bzramngvcrficy2no44	177	8	to	to	TO
work_v44vkh6bzramngvcrficy2no44	177	9	8	8	CD
work_v44vkh6bzramngvcrficy2no44	177	10	M	m	NN
work_v44vkh6bzramngvcrficy2no44	177	11	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	177	12	only	only	RB
work_v44vkh6bzramngvcrficy2no44	177	13	slightly	slightly	RB
work_v44vkh6bzramngvcrficy2no44	177	14	helpful	helpful	JJ
work_v44vkh6bzramngvcrficy2no44	177	15	in	in	IN
work_v44vkh6bzramngvcrficy2no44	177	16	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	177	17	experiments	experiment	NNS
work_v44vkh6bzramngvcrficy2no44	177	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	178	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	178	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	178	3	future	future	NN
work_v44vkh6bzramngvcrficy2no44	178	4	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	178	5	plan	plan	VBP
work_v44vkh6bzramngvcrficy2no44	178	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	178	7	test	test	VB
work_v44vkh6bzramngvcrficy2no44	178	8	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	178	9	learning	learning	NN
work_v44vkh6bzramngvcrficy2no44	178	10	approaches	approach	NNS
work_v44vkh6bzramngvcrficy2no44	178	11	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	178	12	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	178	13	better	well	RBR
work_v44vkh6bzramngvcrficy2no44	178	14	to	to	IN
work_v44vkh6bzramngvcrficy2no44	178	15	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	178	16	data	data	NN
work_v44vkh6bzramngvcrficy2no44	178	17	sets	set	NNS
work_v44vkh6bzramngvcrficy2no44	178	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	179	1	a	a	DT
work_v44vkh6bzramngvcrficy2no44	179	2	stand	stand	VB
work_v44vkh6bzramngvcrficy2no44	179	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	179	4	alone	alone	RB
work_v44vkh6bzramngvcrficy2no44	179	5	classifier	classifier	NN
work_v44vkh6bzramngvcrficy2no44	179	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	179	7	but	but	CC
work_v44vkh6bzramngvcrficy2no44	179	8	as	as	IN
work_v44vkh6bzramngvcrficy2no44	179	9	one	one	CD
work_v44vkh6bzramngvcrficy2no44	179	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	179	11	several	several	JJ
work_v44vkh6bzramngvcrficy2no44	179	12	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	179	13	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	179	14	functions	function	NNS
work_v44vkh6bzramngvcrficy2no44	179	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	180	1	Moreover	moreover	RB
work_v44vkh6bzramngvcrficy2no44	180	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	180	3	for	for	IN
work_v44vkh6bzramngvcrficy2no44	180	4	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	180	5	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	180	6	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	180	7	to	to	TO
work_v44vkh6bzramngvcrficy2no44	180	8	be	be	VB
work_v44vkh6bzramngvcrficy2no44	180	9	effective	effective	JJ
work_v44vkh6bzramngvcrficy2no44	180	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	180	11	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	180	12	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	180	13	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	180	14	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	180	15	that	that	IN
work_v44vkh6bzramngvcrficy2no44	180	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	180	17	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	180	18	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	180	19	option	option	NN
work_v44vkh6bzramngvcrficy2no44	180	20	be	be	VB
work_v44vkh6bzramngvcrficy2no44	180	21	ranked	rank	VBN
work_v44vkh6bzramngvcrficy2no44	180	22	in	in	IN
work_v44vkh6bzramngvcrficy2no44	180	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	180	24	top	top	NN
work_v44vkh6bzramngvcrficy2no44	180	25	n	n	CC
work_v44vkh6bzramngvcrficy2no44	180	26	among	among	IN
work_v44vkh6bzramngvcrficy2no44	180	27	those	those	DT
work_v44vkh6bzramngvcrficy2no44	180	28	available	available	JJ
work_v44vkh6bzramngvcrficy2no44	180	29	at	at	IN
work_v44vkh6bzramngvcrficy2no44	180	30	the	the	DT
work_v44vkh6bzramngvcrficy2no44	180	31	time	time	NN
work_v44vkh6bzramngvcrficy2no44	180	32	of	of	IN
work_v44vkh6bzramngvcrficy2no44	180	33	a	a	DT
work_v44vkh6bzramngvcrficy2no44	180	34	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	180	35	hypoth-	hypoth-	PRP$
work_v44vkh6bzramngvcrficy2no44	180	36	esis	esis	NN
work_v44vkh6bzramngvcrficy2no44	180	37	expansion	expansion	NN
work_v44vkh6bzramngvcrficy2no44	180	38	.	.	.
work_v44vkh6bzramngvcrficy2no44	181	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	181	2	order	order	NN
work_v44vkh6bzramngvcrficy2no44	181	3	to	to	TO
work_v44vkh6bzramngvcrficy2no44	181	4	measure	measure	VB
work_v44vkh6bzramngvcrficy2no44	181	5	this	this	DT
work_v44vkh6bzramngvcrficy2no44	181	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	181	7	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	181	8	simulate	simulate	VBP
work_v44vkh6bzramngvcrficy2no44	181	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	181	10	decoding	decoding	JJ
work_v44vkh6bzramngvcrficy2no44	181	11	process	process	NN
work_v44vkh6bzramngvcrficy2no44	181	12	by	by	IN
work_v44vkh6bzramngvcrficy2no44	181	13	traversing	traverse	VBG
work_v44vkh6bzramngvcrficy2no44	181	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	181	15	source	source	NN
work_v44vkh6bzramngvcrficy2no44	181	16	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	181	17	in	in	IN
work_v44vkh6bzramngvcrficy2no44	181	18	target	target	NN
work_v44vkh6bzramngvcrficy2no44	181	19	order	order	NN
work_v44vkh6bzramngvcrficy2no44	181	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	181	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	181	22	for	for	IN
work_v44vkh6bzramngvcrficy2no44	181	23	each	each	DT
work_v44vkh6bzramngvcrficy2no44	181	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	181	25	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	181	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	181	27	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	181	28	examine	examine	VBP
work_v44vkh6bzramngvcrficy2no44	181	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	181	30	ranking	ranking	NN
work_v44vkh6bzramngvcrficy2no44	181	31	of	of	IN
work_v44vkh6bzramngvcrficy2no44	181	32	all	all	DT
work_v44vkh6bzramngvcrficy2no44	181	33	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	181	34	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	181	35	may	may	MD
work_v44vkh6bzramngvcrficy2no44	181	36	be	be	VB
work_v44vkh6bzramngvcrficy2no44	181	37	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	181	38	next	next	RB
work_v44vkh6bzramngvcrficy2no44	181	39	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	181	40	i.	i.	NNP
work_v44vkh6bzramngvcrficy2no44	181	41	e.	e.	NNP
work_v44vkh6bzramngvcrficy2no44	181	42	the	the	DT
work_v44vkh6bzramngvcrficy2no44	181	43	uncovered	uncovered	JJ
work_v44vkh6bzramngvcrficy2no44	181	44	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	181	45	within	within	IN
work_v44vkh6bzramngvcrficy2no44	181	46	a	a	DT
work_v44vkh6bzramngvcrficy2no44	181	47	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	181	48	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	181	49	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	181	50	.	.	.
work_v44vkh6bzramngvcrficy2no44	182	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	182	2	check	check	VBP
work_v44vkh6bzramngvcrficy2no44	182	3	how	how	WRB
work_v44vkh6bzramngvcrficy2no44	182	4	often	often	RB
work_v44vkh6bzramngvcrficy2no44	182	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	182	6	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	182	7	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	182	8	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	182	9	ranked	rank	VBN
work_v44vkh6bzramngvcrficy2no44	182	10	first	first	RB
work_v44vkh6bzramngvcrficy2no44	182	11	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	182	12	Top-1	Top-1	NNP
work_v44vkh6bzramngvcrficy2no44	182	13	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	182	14	or	or	CC
work_v44vkh6bzramngvcrficy2no44	182	15	at	at	IN
work_v44vkh6bzramngvcrficy2no44	182	16	most	most	RBS
work_v44vkh6bzramngvcrficy2no44	182	17	third	third	JJ
work_v44vkh6bzramngvcrficy2no44	182	18	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	182	19	Top-3	Top-3	NNP
work_v44vkh6bzramngvcrficy2no44	182	20	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	182	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	183	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	183	2	also	also	RB
work_v44vkh6bzramngvcrficy2no44	183	3	com-	com-	VBP
work_v44vkh6bzramngvcrficy2no44	183	4	pute	pute	VBP
work_v44vkh6bzramngvcrficy2no44	183	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	183	6	latter	latter	JJ
work_v44vkh6bzramngvcrficy2no44	183	7	score	score	NN
work_v44vkh6bzramngvcrficy2no44	183	8	on	on	IN
work_v44vkh6bzramngvcrficy2no44	183	9	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	183	10	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	183	11	only	only	RB
work_v44vkh6bzramngvcrficy2no44	183	12	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	183	13	Top-	Top-	NNP
work_v44vkh6bzramngvcrficy2no44	183	14	3-long	3-long	CD
work_v44vkh6bzramngvcrficy2no44	183	15	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	183	16	:	:	:
work_v44vkh6bzramngvcrficy2no44	183	17	i.	i.	NNP
work_v44vkh6bzramngvcrficy2no44	183	18	e.	e.	NNP
work_v44vkh6bzramngvcrficy2no44	183	19	backward	backward	NNP
work_v44vkh6bzramngvcrficy2no44	183	20	jumps	jump	VBZ
work_v44vkh6bzramngvcrficy2no44	183	21	with	with	IN
work_v44vkh6bzramngvcrficy2no44	183	22	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	183	23	D>7	D>7	NNPS
work_v44vkh6bzramngvcrficy2no44	183	24	and	and	CC
work_v44vkh6bzramngvcrficy2no44	183	25	forward	forward	RB
work_v44vkh6bzramngvcrficy2no44	183	26	jumps	jump	VBZ
work_v44vkh6bzramngvcrficy2no44	183	27	with	with	IN
work_v44vkh6bzramngvcrficy2no44	183	28	D>6	D>6	NNP
work_v44vkh6bzramngvcrficy2no44	183	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	184	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	184	2	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	184	3	4	4	CD
work_v44vkh6bzramngvcrficy2no44	184	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	184	5	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	184	6	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	184	7	compared	compare	VBN
work_v44vkh6bzramngvcrficy2no44	184	8	with	with	IN
work_v44vkh6bzramngvcrficy2no44	184	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	184	10	ranking	ranking	NN
work_v44vkh6bzramngvcrficy2no44	184	11	produced	produce	VBN
work_v44vkh6bzramngvcrficy2no44	184	12	by	by	IN
work_v44vkh6bzramngvcrficy2no44	184	13	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	184	14	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	184	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	184	16	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	184	17	always	always	RB
work_v44vkh6bzramngvcrficy2no44	184	18	favors	favor	VBZ
work_v44vkh6bzramngvcrficy2no44	184	19	shorter	short	JJR
work_v44vkh6bzramngvcrficy2no44	184	20	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	184	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	185	1	Two	two	CD
work_v44vkh6bzramngvcrficy2no44	185	2	conditions	condition	NNS
work_v44vkh6bzramngvcrficy2no44	185	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	185	4	considered	consider	VBN
work_v44vkh6bzramngvcrficy2no44	185	5	:	:	:
work_v44vkh6bzramngvcrficy2no44	185	6	DL=10	DL=10	NNP
work_v44vkh6bzramngvcrficy2no44	185	7	corresponding	correspond	VBG
work_v44vkh6bzramngvcrficy2no44	185	8	to	to	IN
work_v44vkh6bzramngvcrficy2no44	185	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	185	10	sampling	sampling	NN
work_v44vkh6bzramngvcrficy2no44	185	11	window	window	NN
work_v44vkh6bzramngvcrficy2no44	185	12	#	#	NNP
work_v44vkh6bzramngvcrficy2no44	185	13	used	use	VBD
work_v44vkh6bzramngvcrficy2no44	185	14	to	to	TO
work_v44vkh6bzramngvcrficy2no44	185	15	produce	produce	VB
work_v44vkh6bzramngvcrficy2no44	185	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	185	17	training	training	NN
work_v44vkh6bzramngvcrficy2no44	185	18	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	185	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	185	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	185	21	DL=18	DL=18	NNP
work_v44vkh6bzramngvcrficy2no44	185	22	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	185	23	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	185	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	185	25	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	185	26	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	185	27	of	of	IN
work_v44vkh6bzramngvcrficy2no44	185	28	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	185	29	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	185	30	will	will	MD
work_v44vkh6bzramngvcrficy2no44	185	31	be	be	VB
work_v44vkh6bzramngvcrficy2no44	185	32	considered	consider	VBN
work_v44vkh6bzramngvcrficy2no44	185	33	in	in	IN
work_v44vkh6bzramngvcrficy2no44	185	34	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	185	35	early	early	RB
work_v44vkh6bzramngvcrficy2no44	185	36	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	185	37	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	185	38	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	185	39	experiment	experiment	NN
work_v44vkh6bzramngvcrficy2no44	185	40	.	.	.
work_v44vkh6bzramngvcrficy2no44	186	1	Model	Model	NNP
work_v44vkh6bzramngvcrficy2no44	186	2	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	186	3	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	186	4	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	186	5	err	err	NNP
work_v44vkh6bzramngvcrficy2no44	186	6	Top-1	Top-1	NNP
work_v44vkh6bzramngvcrficy2no44	186	7	Top-3	Top-3	NNP
work_v44vkh6bzramngvcrficy2no44	186	8	Top-3-long	Top-3-long	NNP
work_v44vkh6bzramngvcrficy2no44	186	9	back	back	RB
work_v44vkh6bzramngvcrficy2no44	186	10	forw	forw	NN
work_v44vkh6bzramngvcrficy2no44	186	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	187	1	Distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	187	2	10	10	CD
work_v44vkh6bzramngvcrficy2no44	187	3	2.4	2.4	CD
work_v44vkh6bzramngvcrficy2no44	187	4	61.8	61.8	CD
work_v44vkh6bzramngvcrficy2no44	187	5	79.6	79.6	CD
work_v44vkh6bzramngvcrficy2no44	187	6	50.7	50.7	CD
work_v44vkh6bzramngvcrficy2no44	187	7	66.0	66.0	CD
work_v44vkh6bzramngvcrficy2no44	187	8	18	18	CD
work_v44vkh6bzramngvcrficy2no44	187	9	0.8	0.8	CD
work_v44vkh6bzramngvcrficy2no44	187	10	62.0	62.0	CD
work_v44vkh6bzramngvcrficy2no44	187	11	80.0	80.0	CD
work_v44vkh6bzramngvcrficy2no44	187	12	18.9	18.9	CD
work_v44vkh6bzramngvcrficy2no44	187	13	52.3	52.3	CD
work_v44vkh6bzramngvcrficy2no44	187	14	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	187	15	10	10	CD
work_v44vkh6bzramngvcrficy2no44	187	16	2.4	2.4	CD
work_v44vkh6bzramngvcrficy2no44	187	17	71.2	71.2	CD
work_v44vkh6bzramngvcrficy2no44	187	18	91.2	91.2	CD
work_v44vkh6bzramngvcrficy2no44	187	19	76.4	76.4	CD
work_v44vkh6bzramngvcrficy2no44	187	20	69.3	69.3	CD
work_v44vkh6bzramngvcrficy2no44	187	21	18	18	CD
work_v44vkh6bzramngvcrficy2no44	187	22	0.8	0.8	CD
work_v44vkh6bzramngvcrficy2no44	187	23	71.2	71.2	CD
work_v44vkh6bzramngvcrficy2no44	187	24	91.8	91.8	CD
work_v44vkh6bzramngvcrficy2no44	187	25	68.0	68.0	CD
work_v44vkh6bzramngvcrficy2no44	187	26	51.8	51.8	CD
work_v44vkh6bzramngvcrficy2no44	187	27	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	187	28	4	4	CD
work_v44vkh6bzramngvcrficy2no44	187	29	:	:	:
work_v44vkh6bzramngvcrficy2no44	187	30	Word	word	NN
work_v44vkh6bzramngvcrficy2no44	187	31	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	187	32	to	to	IN
work_v44vkh6bzramngvcrficy2no44	187	33	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	187	34	word	word	NN
work_v44vkh6bzramngvcrficy2no44	187	35	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	187	36	ranking	rank	VBG
work_v44vkh6bzramngvcrficy2no44	187	37	accuracy	accuracy	NN
work_v44vkh6bzramngvcrficy2no44	187	38	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	187	39	%	%	NN
work_v44vkh6bzramngvcrficy2no44	187	40	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	187	41	of	of	IN
work_v44vkh6bzramngvcrficy2no44	187	42	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	187	43	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	187	44	and	and	CC
work_v44vkh6bzramngvcrficy2no44	187	45	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	187	46	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	187	47	model	model	NN
work_v44vkh6bzramngvcrficy2no44	187	48	,	,	,
work_v44vkh6bzramngvcrficy2no44	187	49	in	in	IN
work_v44vkh6bzramngvcrficy2no44	187	50	dif-	dif-	DT
work_v44vkh6bzramngvcrficy2no44	187	51	ferent	ferent	JJ
work_v44vkh6bzramngvcrficy2no44	187	52	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	187	53	conditions	condition	NNS
work_v44vkh6bzramngvcrficy2no44	187	54	.	.	.
work_v44vkh6bzramngvcrficy2no44	188	1	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	188	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	188	3	err	err	NNP
work_v44vkh6bzramngvcrficy2no44	188	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	188	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	188	6	percentage	percentage	NN
work_v44vkh6bzramngvcrficy2no44	188	7	of	of	IN
work_v44vkh6bzramngvcrficy2no44	188	8	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	188	9	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	188	10	beyond	beyond	IN
work_v44vkh6bzramngvcrficy2no44	188	11	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	188	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	189	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	189	2	test	test	NN
work_v44vkh6bzramngvcrficy2no44	189	3	set	set	VBD
work_v44vkh6bzramngvcrficy2no44	189	4	consists	consist	VBZ
work_v44vkh6bzramngvcrficy2no44	189	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	189	6	40	40	CD
work_v44vkh6bzramngvcrficy2no44	189	7	K	k	NN
work_v44vkh6bzramngvcrficy2no44	189	8	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	189	9	decisions	decision	NNS
work_v44vkh6bzramngvcrficy2no44	189	10	:	:	:
work_v44vkh6bzramngvcrficy2no44	189	11	one	one	CD
work_v44vkh6bzramngvcrficy2no44	189	12	for	for	IN
work_v44vkh6bzramngvcrficy2no44	189	13	each	each	DT
work_v44vkh6bzramngvcrficy2no44	189	14	source	source	NN
work_v44vkh6bzramngvcrficy2no44	189	15	word	word	NN
work_v44vkh6bzramngvcrficy2no44	189	16	in	in	IN
work_v44vkh6bzramngvcrficy2no44	189	17	TIDES	TIDES	NNP
work_v44vkh6bzramngvcrficy2no44	189	18	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	189	19	MT04	MT04	NNP
work_v44vkh6bzramngvcrficy2no44	189	20	.	.	.
work_v44vkh6bzramngvcrficy2no44	190	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	190	2	can	can	MD
work_v44vkh6bzramngvcrficy2no44	190	3	see	see	VB
work_v44vkh6bzramngvcrficy2no44	190	4	that	that	DT
work_v44vkh6bzramngvcrficy2no44	190	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	190	6	in	in	IN
work_v44vkh6bzramngvcrficy2no44	190	7	terms	term	NNS
work_v44vkh6bzramngvcrficy2no44	190	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	190	9	overall	overall	JJ
work_v44vkh6bzramngvcrficy2no44	190	10	accuracies	accuracy	NNS
work_v44vkh6bzramngvcrficy2no44	190	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	190	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	190	13	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	190	14	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	190	15	model	model	NN
work_v44vkh6bzramngvcrficy2no44	190	16	outperforms	outperform	VBZ
work_v44vkh6bzramngvcrficy2no44	190	17	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	190	18	distor-	distor-	IN
work_v44vkh6bzramngvcrficy2no44	190	19	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	190	20	by	by	IN
work_v44vkh6bzramngvcrficy2no44	190	21	a	a	DT
work_v44vkh6bzramngvcrficy2no44	190	22	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	190	23	margin	margin	NN
work_v44vkh6bzramngvcrficy2no44	190	24	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	190	25	about	about	RB
work_v44vkh6bzramngvcrficy2no44	190	26	10	10	CD
work_v44vkh6bzramngvcrficy2no44	190	27	%	%	NN
work_v44vkh6bzramngvcrficy2no44	190	28	absolute	absolute	JJ
work_v44vkh6bzramngvcrficy2no44	190	29	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	190	30	.	.	.
work_v44vkh6bzramngvcrficy2no44	191	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	191	2	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	191	3	an	an	DT
work_v44vkh6bzramngvcrficy2no44	191	4	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	191	5	result	result	NN
work_v44vkh6bzramngvcrficy2no44	191	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	191	7	considering	consider	VBG
work_v44vkh6bzramngvcrficy2no44	191	8	that	that	IN
work_v44vkh6bzramngvcrficy2no44	191	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	191	10	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	191	11	length	length	NN
work_v44vkh6bzramngvcrficy2no44	191	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	191	13	strongly	strongly	RB
work_v44vkh6bzramngvcrficy2no44	191	14	correlating	correlate	VBG
work_v44vkh6bzramngvcrficy2no44	191	15	with	with	IN
work_v44vkh6bzramngvcrficy2no44	191	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	191	17	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	191	18	likeli-	likeli-	NNP
work_v44vkh6bzramngvcrficy2no44	191	19	hood	hood	NNP
work_v44vkh6bzramngvcrficy2no44	191	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	191	21	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	191	22	not	not	RB
work_v44vkh6bzramngvcrficy2no44	191	23	directly	directly	RB
work_v44vkh6bzramngvcrficy2no44	191	24	known	know	VBN
work_v44vkh6bzramngvcrficy2no44	191	25	to	to	IN
work_v44vkh6bzramngvcrficy2no44	191	26	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	191	27	model	model	NN
work_v44vkh6bzramngvcrficy2no44	191	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	192	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	192	2	re-	re-	DT
work_v44vkh6bzramngvcrficy2no44	192	3	gards	gard	VBZ
work_v44vkh6bzramngvcrficy2no44	192	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	192	5	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	192	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	192	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	192	8	higher	high	JJR
work_v44vkh6bzramngvcrficy2no44	192	9	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	192	10	naturally	naturally	RB
work_v44vkh6bzramngvcrficy2no44	192	11	results	result	VBZ
work_v44vkh6bzramngvcrficy2no44	192	12	in	in	IN
work_v44vkh6bzramngvcrficy2no44	192	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	192	14	lower	low	JJR
work_v44vkh6bzramngvcrficy2no44	192	15	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	192	16	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	192	17	error	error	NN
work_v44vkh6bzramngvcrficy2no44	192	18	rate	rate	NN
work_v44vkh6bzramngvcrficy2no44	192	19	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	192	20	percentage	percentage	NN
work_v44vkh6bzramngvcrficy2no44	192	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	192	22	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	192	23	jumps	jumps	NNPS
work_v44vkh6bzramngvcrficy2no44	192	24	be-	be-	XX
work_v44vkh6bzramngvcrficy2no44	192	25	yond	yond	RB
work_v44vkh6bzramngvcrficy2no44	192	26	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	192	27	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	192	28	:	:	:
work_v44vkh6bzramngvcrficy2no44	192	29	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	192	30	0.8	0.8	CD
work_v44vkh6bzramngvcrficy2no44	192	31	%	%	NN
work_v44vkh6bzramngvcrficy2no44	192	32	instead	instead	RB
work_v44vkh6bzramngvcrficy2no44	192	33	of	of	IN
work_v44vkh6bzramngvcrficy2no44	192	34	2.4	2.4	CD
work_v44vkh6bzramngvcrficy2no44	192	35	%	%	NN
work_v44vkh6bzramngvcrficy2no44	192	36	.	.	.
work_v44vkh6bzramngvcrficy2no44	193	1	However	however	RB
work_v44vkh6bzramngvcrficy2no44	193	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	193	3	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	193	4	prediction	prediction	NN
work_v44vkh6bzramngvcrficy2no44	193	5	becomes	become	VBZ
work_v44vkh6bzramngvcrficy2no44	193	6	much	much	RB
work_v44vkh6bzramngvcrficy2no44	193	7	harder	hard	JJR
work_v44vkh6bzramngvcrficy2no44	193	8	:	:	:
work_v44vkh6bzramngvcrficy2no44	193	9	Top-3	Top-3	NNP
work_v44vkh6bzramngvcrficy2no44	193	10	accu-	accu-	NN
work_v44vkh6bzramngvcrficy2no44	193	11	racy	racy	NN
work_v44vkh6bzramngvcrficy2no44	193	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	193	13	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	193	14	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	193	15	by	by	IN
work_v44vkh6bzramngvcrficy2no44	193	16	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	193	17	drops	drop	NNS
work_v44vkh6bzramngvcrficy2no44	193	18	from	from	IN
work_v44vkh6bzramngvcrficy2no44	193	19	50.7	50.7	CD
work_v44vkh6bzramngvcrficy2no44	193	20	%	%	NN
work_v44vkh6bzramngvcrficy2no44	193	21	to	to	IN
work_v44vkh6bzramngvcrficy2no44	193	22	18.9	18.9	CD
work_v44vkh6bzramngvcrficy2no44	193	23	%	%	NN
work_v44vkh6bzramngvcrficy2no44	193	24	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	193	25	backward	backward	RB
work_v44vkh6bzramngvcrficy2no44	193	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	193	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	193	28	from	from	IN
work_v44vkh6bzramngvcrficy2no44	193	29	66.0	66.0	CD
work_v44vkh6bzramngvcrficy2no44	193	30	%	%	NN
work_v44vkh6bzramngvcrficy2no44	193	31	to	to	IN
work_v44vkh6bzramngvcrficy2no44	193	32	52.3	52.3	CD
work_v44vkh6bzramngvcrficy2no44	193	33	%	%	NN
work_v44vkh6bzramngvcrficy2no44	193	34	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	193	35	for-	for-	RB
work_v44vkh6bzramngvcrficy2no44	193	36	ward	ward	NNP
work_v44vkh6bzramngvcrficy2no44	193	37	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	193	38	.	.	.
work_v44vkh6bzramngvcrficy2no44	194	1	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	194	2	model	model	NN
work_v44vkh6bzramngvcrficy2no44	194	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	194	4	remarkably	remarkably	RB
work_v44vkh6bzramngvcrficy2no44	194	5	robust	robust	JJ
work_v44vkh6bzramngvcrficy2no44	194	6	to	to	IN
work_v44vkh6bzramngvcrficy2no44	194	7	this	this	DT
work_v44vkh6bzramngvcrficy2no44	194	8	effect	effect	NN
work_v44vkh6bzramngvcrficy2no44	194	9	on	on	IN
work_v44vkh6bzramngvcrficy2no44	194	10	backward	backward	JJ
work_v44vkh6bzramngvcrficy2no44	194	11	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	194	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	194	13	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	194	14	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	194	15	achieves	achieve	VBZ
work_v44vkh6bzramngvcrficy2no44	194	16	68.0	68.0	CD
work_v44vkh6bzramngvcrficy2no44	194	17	%	%	NN
work_v44vkh6bzramngvcrficy2no44	194	18	accu-	accu-	NN
work_v44vkh6bzramngvcrficy2no44	194	19	333	333	CD
work_v44vkh6bzramngvcrficy2no44	194	20	racy	racy	NN
work_v44vkh6bzramngvcrficy2no44	194	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	195	1	Due	due	IN
work_v44vkh6bzramngvcrficy2no44	195	2	to	to	IN
work_v44vkh6bzramngvcrficy2no44	195	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	195	4	syntactic	syntactic	JJ
work_v44vkh6bzramngvcrficy2no44	195	5	characteristics	characteristic	NNS
work_v44vkh6bzramngvcrficy2no44	195	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	195	7	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	195	8	and	and	CC
work_v44vkh6bzramngvcrficy2no44	195	9	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	195	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	195	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	195	12	typical	typical	JJ
work_v44vkh6bzramngvcrficy2no44	195	13	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	195	14	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	195	15	pattern	pattern	NN
work_v44vkh6bzramngvcrficy2no44	195	16	con-	con-	NN
work_v44vkh6bzramngvcrficy2no44	195	17	sists	sist	NNS
work_v44vkh6bzramngvcrficy2no44	195	18	in	in	IN
work_v44vkh6bzramngvcrficy2no44	195	19	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	195	20	i	i	NN
work_v44vkh6bzramngvcrficy2no44	195	21	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	195	22	skipping	skip	VBG
work_v44vkh6bzramngvcrficy2no44	195	23	a	a	DT
work_v44vkh6bzramngvcrficy2no44	195	24	clause	clause	NN
work_v44vkh6bzramngvcrficy2no44	195	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	195	26	initial	initial	JJ
work_v44vkh6bzramngvcrficy2no44	195	27	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	195	28	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	195	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	195	30	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	195	31	ii	ii	NNP
work_v44vkh6bzramngvcrficy2no44	195	32	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	195	33	covering	cover	VBG
work_v44vkh6bzramngvcrficy2no44	195	34	a	a	DT
work_v44vkh6bzramngvcrficy2no44	195	35	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	195	36	subject	subject	NN
work_v44vkh6bzramngvcrficy2no44	195	37	,	,	,
work_v44vkh6bzramngvcrficy2no44	195	38	then	then	RB
work_v44vkh6bzramngvcrficy2no44	195	39	finally	finally	RB
work_v44vkh6bzramngvcrficy2no44	195	40	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	195	41	iii	iii	NNP
work_v44vkh6bzramngvcrficy2no44	195	42	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	195	43	jumping	jump	VBG
work_v44vkh6bzramngvcrficy2no44	195	44	back	back	RB
work_v44vkh6bzramngvcrficy2no44	195	45	to	to	TO
work_v44vkh6bzramngvcrficy2no44	195	46	translate	translate	VB
work_v44vkh6bzramngvcrficy2no44	195	47	the	the	DT
work_v44vkh6bzramngvcrficy2no44	195	48	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	195	49	and	and	CC
work_v44vkh6bzramngvcrficy2no44	195	50	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	195	51	iv	iv	LS
work_v44vkh6bzramngvcrficy2no44	195	52	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	195	53	jumping	jump	VBG
work_v44vkh6bzramngvcrficy2no44	195	54	forward	forward	RB
work_v44vkh6bzramngvcrficy2no44	195	55	to	to	TO
work_v44vkh6bzramngvcrficy2no44	195	56	continue	continue	VB
work_v44vkh6bzramngvcrficy2no44	195	57	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	195	58	the	the	DT
work_v44vkh6bzramngvcrficy2no44	195	59	rest	rest	NN
work_v44vkh6bzramngvcrficy2no44	195	60	of	of	IN
work_v44vkh6bzramngvcrficy2no44	195	61	the	the	DT
work_v44vkh6bzramngvcrficy2no44	195	62	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	195	63	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	195	64	see	see	VB
work_v44vkh6bzramngvcrficy2no44	195	65	Fig	Fig	NNP
work_v44vkh6bzramngvcrficy2no44	195	66	.	.	.
work_v44vkh6bzramngvcrficy2no44	196	1	3	3	CD
work_v44vkh6bzramngvcrficy2no44	196	2	for	for	IN
work_v44vkh6bzramngvcrficy2no44	196	3	an	an	DT
work_v44vkh6bzramngvcrficy2no44	196	4	example).12	example).12	NN
work_v44vkh6bzramngvcrficy2no44	196	5	Deciding	decide	VBG
work_v44vkh6bzramngvcrficy2no44	196	6	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	196	7	to	to	TO
work_v44vkh6bzramngvcrficy2no44	196	8	jump	jump	VB
work_v44vkh6bzramngvcrficy2no44	196	9	back	back	RB
work_v44vkh6bzramngvcrficy2no44	196	10	to	to	TO
work_v44vkh6bzramngvcrficy2no44	196	11	cover	cover	VB
work_v44vkh6bzramngvcrficy2no44	196	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	196	13	verb	verb	FW
work_v44vkh6bzramngvcrficy2no44	196	14	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	196	15	iii	iii	NN
work_v44vkh6bzramngvcrficy2no44	196	16	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	196	17	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	196	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	196	19	hardest	hard	JJS
work_v44vkh6bzramngvcrficy2no44	196	20	part	part	NN
work_v44vkh6bzramngvcrficy2no44	196	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	196	22	this	this	DT
work_v44vkh6bzramngvcrficy2no44	196	23	process	process	NN
work_v44vkh6bzramngvcrficy2no44	196	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	196	25	and	and	CC
work_v44vkh6bzramngvcrficy2no44	196	26	that	that	DT
work_v44vkh6bzramngvcrficy2no44	196	27	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	196	28	precisely	precisely	RB
work_v44vkh6bzramngvcrficy2no44	196	29	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	196	30	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	196	31	model	model	NN
work_v44vkh6bzramngvcrficy2no44	196	32	seems	seem	VBZ
work_v44vkh6bzramngvcrficy2no44	196	33	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	196	34	helpful	helpful	JJ
work_v44vkh6bzramngvcrficy2no44	196	35	,	,	,
work_v44vkh6bzramngvcrficy2no44	196	36	while	while	IN
work_v44vkh6bzramngvcrficy2no44	196	37	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	196	38	always	always	RB
work_v44vkh6bzramngvcrficy2no44	196	39	prefers	prefer	VBZ
work_v44vkh6bzramngvcrficy2no44	196	40	to	to	TO
work_v44vkh6bzramngvcrficy2no44	196	41	proceed	proceed	VB
work_v44vkh6bzramngvcrficy2no44	196	42	monotonically	monotonically	RB
work_v44vkh6bzramngvcrficy2no44	196	43	achieving	achieve	VBG
work_v44vkh6bzramngvcrficy2no44	196	44	a	a	DT
work_v44vkh6bzramngvcrficy2no44	196	45	very	very	RB
work_v44vkh6bzramngvcrficy2no44	196	46	low	low	JJ
work_v44vkh6bzramngvcrficy2no44	196	47	ac-	ac-	JJ
work_v44vkh6bzramngvcrficy2no44	196	48	curacy	curacy	NN
work_v44vkh6bzramngvcrficy2no44	196	49	of	of	IN
work_v44vkh6bzramngvcrficy2no44	196	50	18.9	18.9	CD
work_v44vkh6bzramngvcrficy2no44	196	51	%	%	NN
work_v44vkh6bzramngvcrficy2no44	196	52	.	.	.
work_v44vkh6bzramngvcrficy2no44	197	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	197	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	197	3	case	case	NN
work_v44vkh6bzramngvcrficy2no44	197	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	197	5	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	197	6	forward	forward	NN
work_v44vkh6bzramngvcrficy2no44	197	7	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	197	8	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	197	9	iv	iv	NN
work_v44vkh6bzramngvcrficy2no44	197	10	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	197	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	197	12	instead	instead	RB
work_v44vkh6bzramngvcrficy2no44	197	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	197	14	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	197	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	197	16	advantaged	advantage	VBN
work_v44vkh6bzramngvcrficy2no44	197	17	as	as	IN
work_v44vkh6bzramngvcrficy2no44	197	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	197	19	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	197	20	choice	choice	NN
work_v44vkh6bzramngvcrficy2no44	197	21	typically	typically	RB
work_v44vkh6bzramngvcrficy2no44	197	22	corresponds	correspond	VBZ
work_v44vkh6bzramngvcrficy2no44	197	23	to	to	IN
work_v44vkh6bzramngvcrficy2no44	197	24	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	197	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	197	26	first	first	JJ
work_v44vkh6bzramngvcrficy2no44	197	27	uncovered	uncovered	JJ
work_v44vkh6bzramngvcrficy2no44	197	28	position	position	NN
work_v44vkh6bzramngvcrficy2no44	197	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	197	30	that	that	DT
work_v44vkh6bzramngvcrficy2no44	197	31	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	197	32	the	the	DT
work_v44vkh6bzramngvcrficy2no44	197	33	shortest	short	JJS
work_v44vkh6bzramngvcrficy2no44	197	34	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	197	35	avail-	avail-	NN
work_v44vkh6bzramngvcrficy2no44	197	36	able	able	JJ
work_v44vkh6bzramngvcrficy2no44	197	37	from	from	IN
work_v44vkh6bzramngvcrficy2no44	197	38	the	the	DT
work_v44vkh6bzramngvcrficy2no44	197	39	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	197	40	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	197	41	word	word	NN
work_v44vkh6bzramngvcrficy2no44	197	42	.	.	.
work_v44vkh6bzramngvcrficy2no44	198	1	Even	even	RB
work_v44vkh6bzramngvcrficy2no44	198	2	here	here	RB
work_v44vkh6bzramngvcrficy2no44	198	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	198	4	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	198	5	model	model	NN
work_v44vkh6bzramngvcrficy2no44	198	6	achieves	achieve	VBZ
work_v44vkh6bzramngvcrficy2no44	198	7	an	an	DT
work_v44vkh6bzramngvcrficy2no44	198	8	accuracy	accuracy	NN
work_v44vkh6bzramngvcrficy2no44	198	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	198	10	51.8	51.8	CD
work_v44vkh6bzramngvcrficy2no44	198	11	%	%	NN
work_v44vkh6bzramngvcrficy2no44	198	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	198	13	only	only	RB
work_v44vkh6bzramngvcrficy2no44	198	14	slightly	slightly	RB
work_v44vkh6bzramngvcrficy2no44	198	15	lower	low	JJR
work_v44vkh6bzramngvcrficy2no44	198	16	than	than	IN
work_v44vkh6bzramngvcrficy2no44	198	17	that	that	DT
work_v44vkh6bzramngvcrficy2no44	198	18	of	of	IN
work_v44vkh6bzramngvcrficy2no44	198	19	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	198	20	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	198	21	52.3	52.3	CD
work_v44vkh6bzramngvcrficy2no44	198	22	%	%	NN
work_v44vkh6bzramngvcrficy2no44	198	23	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	198	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	199	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	199	2	summary	summary	NN
work_v44vkh6bzramngvcrficy2no44	199	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	199	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	199	5	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	199	6	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	199	7	model	model	NN
work_v44vkh6bzramngvcrficy2no44	199	8	signifi-	signifi-	NNP
work_v44vkh6bzramngvcrficy2no44	199	9	cantly	cantly	RB
work_v44vkh6bzramngvcrficy2no44	199	10	outperforms	outperform	VBZ
work_v44vkh6bzramngvcrficy2no44	199	11	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	199	12	in	in	IN
work_v44vkh6bzramngvcrficy2no44	199	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	199	14	ranking	ranking	NN
work_v44vkh6bzramngvcrficy2no44	199	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	199	16	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	199	17	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	199	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	200	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	200	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	200	3	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	200	4	majority	majority	NN
work_v44vkh6bzramngvcrficy2no44	200	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	200	6	cases	case	NNS
work_v44vkh6bzramngvcrficy2no44	200	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	200	8	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	200	9	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	200	10	able	able	JJ
work_v44vkh6bzramngvcrficy2no44	200	11	to	to	TO
work_v44vkh6bzramngvcrficy2no44	200	12	rank	rank	VB
work_v44vkh6bzramngvcrficy2no44	200	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	200	14	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	200	15	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	200	16	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	200	17	in	in	IN
work_v44vkh6bzramngvcrficy2no44	200	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	200	19	top	top	JJ
work_v44vkh6bzramngvcrficy2no44	200	20	3	3	CD
work_v44vkh6bzramngvcrficy2no44	200	21	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	200	22	op-	op-	JJ
work_v44vkh6bzramngvcrficy2no44	200	23	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	200	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	200	25	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	200	26	suggests	suggest	VBZ
work_v44vkh6bzramngvcrficy2no44	200	27	that	that	IN
work_v44vkh6bzramngvcrficy2no44	200	28	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	200	29	can	can	MD
work_v44vkh6bzramngvcrficy2no44	200	30	be	be	VB
work_v44vkh6bzramngvcrficy2no44	200	31	effectively	effectively	RB
work_v44vkh6bzramngvcrficy2no44	200	32	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	200	33	for	for	IN
work_v44vkh6bzramngvcrficy2no44	200	34	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	200	35	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	200	36	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	200	37	.	.	.
work_v44vkh6bzramngvcrficy2no44	201	1	5.2	5.2	CD
work_v44vkh6bzramngvcrficy2no44	201	2	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	201	3	experimental	experimental	JJ
work_v44vkh6bzramngvcrficy2no44	201	4	setup	setup	NN
work_v44vkh6bzramngvcrficy2no44	201	5	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	201	6	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	201	7	systems	system	NNS
work_v44vkh6bzramngvcrficy2no44	201	8	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	201	9	built	build	VBN
work_v44vkh6bzramngvcrficy2no44	201	10	with	with	IN
work_v44vkh6bzramngvcrficy2no44	201	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	201	12	Moses	Moses	NNP
work_v44vkh6bzramngvcrficy2no44	201	13	toolkit	toolkit	NNS
work_v44vkh6bzramngvcrficy2no44	201	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	201	15	while	while	IN
work_v44vkh6bzramngvcrficy2no44	201	16	word	word	NN
work_v44vkh6bzramngvcrficy2no44	201	17	alignment	alignment	NN
work_v44vkh6bzramngvcrficy2no44	201	18	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	201	19	produced	produce	VBN
work_v44vkh6bzramngvcrficy2no44	201	20	by	by	IN
work_v44vkh6bzramngvcrficy2no44	201	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	201	22	Berke-	Berke-	NNP
work_v44vkh6bzramngvcrficy2no44	201	23	ley	ley	NN
work_v44vkh6bzramngvcrficy2no44	201	24	Aligner	Aligner	NNP
work_v44vkh6bzramngvcrficy2no44	201	25	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	201	26	Liang	Liang	NNP
work_v44vkh6bzramngvcrficy2no44	201	27	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	201	28	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	201	29	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	201	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	201	31	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	201	32	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	201	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	202	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	202	2	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	202	3	de-	de-	NN
work_v44vkh6bzramngvcrficy2no44	202	4	coder	coder	NN
work_v44vkh6bzramngvcrficy2no44	202	5	includes	include	VBZ
work_v44vkh6bzramngvcrficy2no44	202	6	a	a	DT
work_v44vkh6bzramngvcrficy2no44	202	7	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	202	8	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	202	9	model	model	NN
work_v44vkh6bzramngvcrficy2no44	202	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	202	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	202	12	lexi-	lexi-	XX
work_v44vkh6bzramngvcrficy2no44	202	13	calized	calize	VBN
work_v44vkh6bzramngvcrficy2no44	202	14	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	202	15	model	model	NN
work_v44vkh6bzramngvcrficy2no44	202	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	202	17	a	a	DT
work_v44vkh6bzramngvcrficy2no44	202	18	6-gram	6-gram	CD
work_v44vkh6bzramngvcrficy2no44	202	19	target	target	NN
work_v44vkh6bzramngvcrficy2no44	202	20	language	language	NN
work_v44vkh6bzramngvcrficy2no44	202	21	model	model	NN
work_v44vkh6bzramngvcrficy2no44	202	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	202	23	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	202	24	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	202	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	202	26	word	word	NN
work_v44vkh6bzramngvcrficy2no44	202	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	202	28	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	202	29	penalties	penalty	NNS
work_v44vkh6bzramngvcrficy2no44	202	30	.	.	.
work_v44vkh6bzramngvcrficy2no44	203	1	More	more	RBR
work_v44vkh6bzramngvcrficy2no44	203	2	specifically	specifically	RB
work_v44vkh6bzramngvcrficy2no44	203	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	203	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	203	5	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	203	6	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	203	7	model	model	NN
work_v44vkh6bzramngvcrficy2no44	203	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	203	9	a	a	DT
work_v44vkh6bzramngvcrficy2no44	203	10	hierarchical	hierarchical	JJ
work_v44vkh6bzramngvcrficy2no44	203	11	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	203	12	orientation	orientation	NN
work_v44vkh6bzramngvcrficy2no44	203	13	model	model	NN
work_v44vkh6bzramngvcrficy2no44	203	14	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	203	15	Tillmann	Tillmann	NNP
work_v44vkh6bzramngvcrficy2no44	203	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	203	17	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	203	18	;	;	:
work_v44vkh6bzramngvcrficy2no44	203	19	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	203	20	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	203	21	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	203	22	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	203	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	203	24	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	203	25	;	;	:
work_v44vkh6bzramngvcrficy2no44	203	26	Galley	Galley	NNP
work_v44vkh6bzramngvcrficy2no44	203	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	203	28	Manning	Manning	NNP
work_v44vkh6bzramngvcrficy2no44	203	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	203	30	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	203	31	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	203	32	trained	train	VBD
work_v44vkh6bzramngvcrficy2no44	203	33	on	on	IN
work_v44vkh6bzramngvcrficy2no44	203	34	all	all	PDT
work_v44vkh6bzramngvcrficy2no44	203	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	203	36	available	available	JJ
work_v44vkh6bzramngvcrficy2no44	203	37	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	203	38	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	203	39	.	.	.
work_v44vkh6bzramngvcrficy2no44	204	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	204	2	variant	variant	NN
work_v44vkh6bzramngvcrficy2no44	204	3	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	204	4	shown	show	VBN
work_v44vkh6bzramngvcrficy2no44	204	5	to	to	IN
work_v44vkh6bzramngvcrficy2no44	204	6	outperform	outperform	VB
work_v44vkh6bzramngvcrficy2no44	204	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	204	8	default	default	NN
work_v44vkh6bzramngvcrficy2no44	204	9	word-	word-	NN
work_v44vkh6bzramngvcrficy2no44	204	10	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	204	11	on	on	IN
work_v44vkh6bzramngvcrficy2no44	204	12	an	an	DT
work_v44vkh6bzramngvcrficy2no44	204	13	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	204	14	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	204	15	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	204	16	task	task	NN
work_v44vkh6bzramngvcrficy2no44	204	17	.	.	.
work_v44vkh6bzramngvcrficy2no44	205	1	To	to	TO
work_v44vkh6bzramngvcrficy2no44	205	2	make	make	VB
work_v44vkh6bzramngvcrficy2no44	205	3	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	205	4	base-	base-	NN
work_v44vkh6bzramngvcrficy2no44	205	5	line	line	NN
work_v44vkh6bzramngvcrficy2no44	205	6	even	even	RB
work_v44vkh6bzramngvcrficy2no44	205	7	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	205	8	competitive	competitive	JJ
work_v44vkh6bzramngvcrficy2no44	205	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	205	10	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	205	11	apply	apply	VBP
work_v44vkh6bzramngvcrficy2no44	205	12	early	early	RB
work_v44vkh6bzramngvcrficy2no44	205	13	distor-	distor-	JJ
work_v44vkh6bzramngvcrficy2no44	205	14	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	205	15	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	205	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	205	17	as	as	IN
work_v44vkh6bzramngvcrficy2no44	205	18	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	205	19	by	by	IN
work_v44vkh6bzramngvcrficy2no44	205	20	Moore	Moore	NNP
work_v44vkh6bzramngvcrficy2no44	205	21	and	and	CC
work_v44vkh6bzramngvcrficy2no44	205	22	Quirk	Quirk	NNP
work_v44vkh6bzramngvcrficy2no44	205	23	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	205	24	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	205	25	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	205	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	206	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	206	2	function	function	NN
work_v44vkh6bzramngvcrficy2no44	206	3	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	206	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	206	5	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	206	6	value	value	NN
work_v44vkh6bzramngvcrficy2no44	206	7	as	as	IN
work_v44vkh6bzramngvcrficy2no44	206	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	206	9	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	206	10	one	one	CD
work_v44vkh6bzramngvcrficy2no44	206	11	over	over	IN
work_v44vkh6bzramngvcrficy2no44	206	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	206	13	complete	complete	JJ
work_v44vkh6bzramngvcrficy2no44	206	14	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	206	15	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	206	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	206	17	but	but	CC
work_v44vkh6bzramngvcrficy2no44	206	18	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	206	19	antic-	antic-	VBD
work_v44vkh6bzramngvcrficy2no44	206	20	ipates	ipate	VBZ
work_v44vkh6bzramngvcrficy2no44	206	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	206	22	gradual	gradual	JJ
work_v44vkh6bzramngvcrficy2no44	206	23	accumulation	accumulation	NN
work_v44vkh6bzramngvcrficy2no44	206	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	206	25	the	the	DT
work_v44vkh6bzramngvcrficy2no44	206	26	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	206	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	206	28	mak-	mak-	NNP
work_v44vkh6bzramngvcrficy2no44	206	29	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	206	30	hypotheses	hypothesis	NNS
work_v44vkh6bzramngvcrficy2no44	206	31	of	of	IN
work_v44vkh6bzramngvcrficy2no44	206	32	the	the	DT
work_v44vkh6bzramngvcrficy2no44	206	33	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	206	34	length	length	NN
work_v44vkh6bzramngvcrficy2no44	206	35	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	206	36	compara-	compara-	JJ
work_v44vkh6bzramngvcrficy2no44	206	37	ble	ble	NN
work_v44vkh6bzramngvcrficy2no44	206	38	to	to	IN
work_v44vkh6bzramngvcrficy2no44	206	39	one	one	CD
work_v44vkh6bzramngvcrficy2no44	206	40	another	another	DT
work_v44vkh6bzramngvcrficy2no44	206	41	.	.	.
work_v44vkh6bzramngvcrficy2no44	207	1	Note	note	VB
work_v44vkh6bzramngvcrficy2no44	207	2	that	that	IN
work_v44vkh6bzramngvcrficy2no44	207	3	this	this	DT
work_v44vkh6bzramngvcrficy2no44	207	4	option	option	NN
work_v44vkh6bzramngvcrficy2no44	207	5	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	207	6	no	no	DT
work_v44vkh6bzramngvcrficy2no44	207	7	ef-	ef-	JJ
work_v44vkh6bzramngvcrficy2no44	207	8	12Clearly	12clearly	CD
work_v44vkh6bzramngvcrficy2no44	207	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	207	10	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	207	11	would	would	MD
work_v44vkh6bzramngvcrficy2no44	207	12	expect	expect	VB
work_v44vkh6bzramngvcrficy2no44	207	13	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	207	14	figures	figure	NNS
work_v44vkh6bzramngvcrficy2no44	207	15	from	from	IN
work_v44vkh6bzramngvcrficy2no44	207	16	testing	test	VBG
work_v44vkh6bzramngvcrficy2no44	207	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	207	18	model	model	NN
work_v44vkh6bzramngvcrficy2no44	207	19	on	on	IN
work_v44vkh6bzramngvcrficy2no44	207	20	another	another	DT
work_v44vkh6bzramngvcrficy2no44	207	21	language	language	NN
work_v44vkh6bzramngvcrficy2no44	207	22	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	207	23	like	like	IN
work_v44vkh6bzramngvcrficy2no44	207	24	German	German	NNP
work_v44vkh6bzramngvcrficy2no44	207	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	207	26	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	207	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	207	28	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	207	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	207	30	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	207	31	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	207	32	often	often	RB
work_v44vkh6bzramngvcrficy2no44	207	33	postponed	postpone	VBN
work_v44vkh6bzramngvcrficy2no44	207	34	in	in	IN
work_v44vkh6bzramngvcrficy2no44	207	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	207	36	source	source	NN
work_v44vkh6bzramngvcrficy2no44	207	37	with	with	IN
work_v44vkh6bzramngvcrficy2no44	207	38	respect	respect	NN
work_v44vkh6bzramngvcrficy2no44	207	39	to	to	IN
work_v44vkh6bzramngvcrficy2no44	207	40	the	the	DT
work_v44vkh6bzramngvcrficy2no44	207	41	target	target	NN
work_v44vkh6bzramngvcrficy2no44	207	42	.	.	.
work_v44vkh6bzramngvcrficy2no44	208	1	fect	fect	VB
work_v44vkh6bzramngvcrficy2no44	208	2	on	on	IN
work_v44vkh6bzramngvcrficy2no44	208	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	208	4	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	208	5	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	208	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	208	7	but	but	CC
work_v44vkh6bzramngvcrficy2no44	208	8	only	only	RB
work_v44vkh6bzramngvcrficy2no44	208	9	on	on	IN
work_v44vkh6bzramngvcrficy2no44	208	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	208	11	distor-	distor-	NN
work_v44vkh6bzramngvcrficy2no44	208	12	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	208	13	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	208	14	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	208	15	function	function	NN
work_v44vkh6bzramngvcrficy2no44	208	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	209	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	209	2	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	209	3	by	by	IN
work_v44vkh6bzramngvcrficy2no44	209	4	Johnson	Johnson	NNP
work_v44vkh6bzramngvcrficy2no44	209	5	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	209	6	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	209	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	210	1	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	210	2	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	210	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	210	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	210	5	statistically	statistically	RB
work_v44vkh6bzramngvcrficy2no44	210	6	improbable	improbable	JJ
work_v44vkh6bzramngvcrficy2no44	210	7	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	210	8	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	210	9	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	210	10	removed	remove	VBN
work_v44vkh6bzramngvcrficy2no44	210	11	from	from	IN
work_v44vkh6bzramngvcrficy2no44	210	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	210	13	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	210	14	model	model	NN
work_v44vkh6bzramngvcrficy2no44	210	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	211	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	211	2	lan-	lan-	NN
work_v44vkh6bzramngvcrficy2no44	211	3	guage	guage	NN
work_v44vkh6bzramngvcrficy2no44	211	4	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	211	5	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	211	6	estimated	estimate	VBN
work_v44vkh6bzramngvcrficy2no44	211	7	by	by	IN
work_v44vkh6bzramngvcrficy2no44	211	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	211	9	IRSTLM	IRSTLM	NNP
work_v44vkh6bzramngvcrficy2no44	211	10	toolkit	toolkit	NNS
work_v44vkh6bzramngvcrficy2no44	211	11	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	211	12	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	211	13	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	211	14	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	211	15	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	211	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	211	17	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	211	18	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	211	19	with	with	IN
work_v44vkh6bzramngvcrficy2no44	211	20	modified	modify	VBN
work_v44vkh6bzramngvcrficy2no44	211	21	Kneser	Kneser	NNP
work_v44vkh6bzramngvcrficy2no44	211	22	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	211	23	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	211	24	smoothing	smooth	VBG
work_v44vkh6bzramngvcrficy2no44	211	25	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	211	26	Chen	Chen	NNP
work_v44vkh6bzramngvcrficy2no44	211	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	211	28	Goodman	Goodman	NNP
work_v44vkh6bzramngvcrficy2no44	211	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	211	30	1999	1999	CD
work_v44vkh6bzramngvcrficy2no44	211	31	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	211	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	212	1	Feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	212	2	weights	weight	NNS
work_v44vkh6bzramngvcrficy2no44	212	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	212	4	optimized	optimize	VBN
work_v44vkh6bzramngvcrficy2no44	212	5	by	by	IN
work_v44vkh6bzramngvcrficy2no44	212	6	minimum	minimum	JJ
work_v44vkh6bzramngvcrficy2no44	212	7	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	212	8	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	212	9	error	error	NN
work_v44vkh6bzramngvcrficy2no44	212	10	training	training	NN
work_v44vkh6bzramngvcrficy2no44	212	11	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	212	12	Och	Och	NNP
work_v44vkh6bzramngvcrficy2no44	212	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	212	14	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	212	15	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	212	16	on	on	IN
work_v44vkh6bzramngvcrficy2no44	212	17	dev06-nw	dev06-nw	NNP
work_v44vkh6bzramngvcrficy2no44	212	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	213	1	To	to	TO
work_v44vkh6bzramngvcrficy2no44	213	2	reduce	reduce	VB
work_v44vkh6bzramngvcrficy2no44	213	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	213	4	effects	effect	NNS
work_v44vkh6bzramngvcrficy2no44	213	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	213	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	213	7	optimizer	optimizer	NN
work_v44vkh6bzramngvcrficy2no44	213	8	instability	instability	NN
work_v44vkh6bzramngvcrficy2no44	213	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	213	10	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	213	11	tune	tune	VBP
work_v44vkh6bzramngvcrficy2no44	213	12	each	each	DT
work_v44vkh6bzramngvcrficy2no44	213	13	configuration	configuration	NN
work_v44vkh6bzramngvcrficy2no44	213	14	four	four	CD
work_v44vkh6bzramngvcrficy2no44	213	15	times	time	NNS
work_v44vkh6bzramngvcrficy2no44	213	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	213	17	use	use	VB
work_v44vkh6bzramngvcrficy2no44	213	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	213	19	av-	av-	JJ
work_v44vkh6bzramngvcrficy2no44	213	20	erage	erage	NN
work_v44vkh6bzramngvcrficy2no44	213	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	213	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	213	23	resulting	result	VBG
work_v44vkh6bzramngvcrficy2no44	213	24	weight	weight	NN
work_v44vkh6bzramngvcrficy2no44	213	25	vectors	vector	NNS
work_v44vkh6bzramngvcrficy2no44	213	26	to	to	TO
work_v44vkh6bzramngvcrficy2no44	213	27	translate	translate	VB
work_v44vkh6bzramngvcrficy2no44	213	28	the	the	DT
work_v44vkh6bzramngvcrficy2no44	213	29	test	test	NN
work_v44vkh6bzramngvcrficy2no44	213	30	sets	set	NNS
work_v44vkh6bzramngvcrficy2no44	213	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	213	32	as	as	IN
work_v44vkh6bzramngvcrficy2no44	213	33	suggested	suggest	VBN
work_v44vkh6bzramngvcrficy2no44	213	34	by	by	IN
work_v44vkh6bzramngvcrficy2no44	213	35	Cettolo	Cettolo	NNP
work_v44vkh6bzramngvcrficy2no44	213	36	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	213	37	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	213	38	.	.	.
work_v44vkh6bzramngvcrficy2no44	214	1	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	214	2	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	214	3	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	214	4	.	.	.
work_v44vkh6bzramngvcrficy2no44	215	1	Finally	finally	RB
work_v44vkh6bzramngvcrficy2no44	215	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	215	3	eval08-nw	eval08-nw	NNP
work_v44vkh6bzramngvcrficy2no44	215	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	215	5	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	215	6	to	to	TO
work_v44vkh6bzramngvcrficy2no44	215	7	select	select	VB
work_v44vkh6bzramngvcrficy2no44	215	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	215	9	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	215	10	prun-	prun-	JJ
work_v44vkh6bzramngvcrficy2no44	215	11	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	215	12	parameters	parameter	NNS
work_v44vkh6bzramngvcrficy2no44	215	13	for	for	IN
work_v44vkh6bzramngvcrficy2no44	215	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	215	15	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	215	16	experiment	experiment	NN
work_v44vkh6bzramngvcrficy2no44	215	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	215	18	while	while	IN
work_v44vkh6bzramngvcrficy2no44	215	19	eval09-	eval09-	NNP
work_v44vkh6bzramngvcrficy2no44	215	20	nw	nw	NNP
work_v44vkh6bzramngvcrficy2no44	215	21	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	215	22	always	always	RB
work_v44vkh6bzramngvcrficy2no44	215	23	reserved	reserve	VBN
work_v44vkh6bzramngvcrficy2no44	215	24	as	as	IN
work_v44vkh6bzramngvcrficy2no44	215	25	blind	blind	JJ
work_v44vkh6bzramngvcrficy2no44	215	26	test	test	NN
work_v44vkh6bzramngvcrficy2no44	215	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	216	1	5.3	5.3	CD
work_v44vkh6bzramngvcrficy2no44	216	2	Evaluation	Evaluation	NNP
work_v44vkh6bzramngvcrficy2no44	216	3	metrics	metric	NNS
work_v44vkh6bzramngvcrficy2no44	216	4	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	216	5	evaluate	evaluate	VBP
work_v44vkh6bzramngvcrficy2no44	216	6	global	global	JJ
work_v44vkh6bzramngvcrficy2no44	216	7	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	216	8	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	216	9	with	with	IN
work_v44vkh6bzramngvcrficy2no44	216	10	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	216	11	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	216	12	Papineni	Papineni	NNP
work_v44vkh6bzramngvcrficy2no44	216	13	et	et	FW
work_v44vkh6bzramngvcrficy2no44	216	14	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	216	15	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	216	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	216	17	2002	2002	CD
work_v44vkh6bzramngvcrficy2no44	216	18	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	216	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	216	20	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	216	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	216	22	Banerjee	Banerjee	NNP
work_v44vkh6bzramngvcrficy2no44	216	23	and	and	CC
work_v44vkh6bzramngvcrficy2no44	216	24	Lavie	Lavie	NNP
work_v44vkh6bzramngvcrficy2no44	216	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	216	26	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	216	27	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	216	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	217	1	These	these	DT
work_v44vkh6bzramngvcrficy2no44	217	2	metrics	metric	NNS
work_v44vkh6bzramngvcrficy2no44	217	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	217	4	though	though	RB
work_v44vkh6bzramngvcrficy2no44	217	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	217	6	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	217	7	only	only	RB
work_v44vkh6bzramngvcrficy2no44	217	8	in-	in-	RB
work_v44vkh6bzramngvcrficy2no44	217	9	directly	directly	RB
work_v44vkh6bzramngvcrficy2no44	217	10	sensitive	sensitive	JJ
work_v44vkh6bzramngvcrficy2no44	217	11	to	to	IN
work_v44vkh6bzramngvcrficy2no44	217	12	word	word	NN
work_v44vkh6bzramngvcrficy2no44	217	13	order	order	NN
work_v44vkh6bzramngvcrficy2no44	217	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	217	15	and	and	CC
work_v44vkh6bzramngvcrficy2no44	217	16	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	217	17	un-	un-	RB
work_v44vkh6bzramngvcrficy2no44	217	18	likely	likely	JJ
work_v44vkh6bzramngvcrficy2no44	217	19	to	to	TO
work_v44vkh6bzramngvcrficy2no44	217	20	capture	capture	VB
work_v44vkh6bzramngvcrficy2no44	217	21	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	217	22	at	at	IN
work_v44vkh6bzramngvcrficy2no44	217	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	217	24	level	level	NN
work_v44vkh6bzramngvcrficy2no44	217	25	of	of	IN
work_v44vkh6bzramngvcrficy2no44	217	26	long-	long-	JJ
work_v44vkh6bzramngvcrficy2no44	217	27	range	range	NN
work_v44vkh6bzramngvcrficy2no44	217	28	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	217	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	218	1	For	for	IN
work_v44vkh6bzramngvcrficy2no44	218	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	218	3	reason	reason	NN
work_v44vkh6bzramngvcrficy2no44	218	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	218	5	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	218	6	also	also	RB
work_v44vkh6bzramngvcrficy2no44	218	7	com-	com-	VBP
work_v44vkh6bzramngvcrficy2no44	218	8	pute	pute	VBP
work_v44vkh6bzramngvcrficy2no44	218	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	218	10	Kendall	Kendall	NNP
work_v44vkh6bzramngvcrficy2no44	218	11	Reordering	Reordering	NNP
work_v44vkh6bzramngvcrficy2no44	218	12	Score	Score	NNP
work_v44vkh6bzramngvcrficy2no44	218	13	or	or	CC
work_v44vkh6bzramngvcrficy2no44	218	14	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	218	15	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	218	16	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	218	17	et	et	NNP
work_v44vkh6bzramngvcrficy2no44	218	18	al	al	NNP
work_v44vkh6bzramngvcrficy2no44	218	19	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	218	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	218	21	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	218	22	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	218	23	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	218	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	218	25	a	a	DT
work_v44vkh6bzramngvcrficy2no44	218	26	positive	positive	JJ
work_v44vkh6bzramngvcrficy2no44	218	27	score	score	NN
work_v44vkh6bzramngvcrficy2no44	218	28	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	218	29	on	on	IN
work_v44vkh6bzramngvcrficy2no44	218	30	the	the	DT
work_v44vkh6bzramngvcrficy2no44	218	31	Kendall	Kendall	NNP
work_v44vkh6bzramngvcrficy2no44	218	32	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	218	33	Tau	Tau	NNP
work_v44vkh6bzramngvcrficy2no44	218	34	distance	distance	NN
work_v44vkh6bzramngvcrficy2no44	218	35	between	between	IN
work_v44vkh6bzramngvcrficy2no44	218	36	the	the	DT
work_v44vkh6bzramngvcrficy2no44	218	37	source	source	NN
work_v44vkh6bzramngvcrficy2no44	218	38	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	218	39	output	output	NN
work_v44vkh6bzramngvcrficy2no44	218	40	permutation	permutation	NN
work_v44vkh6bzramngvcrficy2no44	218	41	%	%	NN
work_v44vkh6bzramngvcrficy2no44	218	42	and	and	CC
work_v44vkh6bzramngvcrficy2no44	218	43	the	the	DT
work_v44vkh6bzramngvcrficy2no44	218	44	source	source	NN
work_v44vkh6bzramngvcrficy2no44	218	45	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	218	46	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	218	47	permuta-	permuta-	NN
work_v44vkh6bzramngvcrficy2no44	218	48	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	218	49	"	"	''
work_v44vkh6bzramngvcrficy2no44	218	50	:	:	:
work_v44vkh6bzramngvcrficy2no44	218	51	KRS(%	KRS(%	NNP
work_v44vkh6bzramngvcrficy2no44	218	52	,	,	,
work_v44vkh6bzramngvcrficy2no44	218	53	")	")	''
work_v44vkh6bzramngvcrficy2no44	218	54	=	=	FW
work_v44vkh6bzramngvcrficy2no44	218	55	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	218	56	1	1	CD
work_v44vkh6bzramngvcrficy2no44	218	57	!	!	.
work_v44vkh6bzramngvcrficy2no44	219	1	#	#	$
work_v44vkh6bzramngvcrficy2no44	219	2	K(%	K(%	NNP
work_v44vkh6bzramngvcrficy2no44	219	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	219	4	")	")	''
work_v44vkh6bzramngvcrficy2no44	219	5	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	219	6	·	·	NFP
work_v44vkh6bzramngvcrficy2no44	219	7	BP	BP	NNP
work_v44vkh6bzramngvcrficy2no44	219	8	K(%	K(%	NNP
work_v44vkh6bzramngvcrficy2no44	219	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	219	10	")	")	''
work_v44vkh6bzramngvcrficy2no44	219	11	=	=	FW
work_v44vkh6bzramngvcrficy2no44	219	12	"	"	``
work_v44vkh6bzramngvcrficy2no44	219	13	n	n	NN
work_v44vkh6bzramngvcrficy2no44	219	14	i=1	i=1	NN
work_v44vkh6bzramngvcrficy2no44	219	15	"	"	''
work_v44vkh6bzramngvcrficy2no44	219	16	n	n	CC
work_v44vkh6bzramngvcrficy2no44	219	17	j=1	j=1	JJ
work_v44vkh6bzramngvcrficy2no44	219	18	d(i	d(i	NNP
work_v44vkh6bzramngvcrficy2no44	219	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	219	20	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	219	21	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	219	22	1	1	CD
work_v44vkh6bzramngvcrficy2no44	219	23	2	2	CD
work_v44vkh6bzramngvcrficy2no44	219	24	n(n	n(n	NN
work_v44vkh6bzramngvcrficy2no44	219	25	!	!	.
work_v44vkh6bzramngvcrficy2no44	220	1	1	1	LS
work_v44vkh6bzramngvcrficy2no44	220	2	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	220	3	d(i	d(i	NNP
work_v44vkh6bzramngvcrficy2no44	220	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	220	5	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	220	6	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	220	7	=	=	VBZ
work_v44vkh6bzramngvcrficy2no44	220	8	$	$	$
work_v44vkh6bzramngvcrficy2no44	220	9	1	1	CD
work_v44vkh6bzramngvcrficy2no44	220	10	if	if	IN
work_v44vkh6bzramngvcrficy2no44	220	11	%	%	NN
work_v44vkh6bzramngvcrficy2no44	220	12	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	220	13	<	<	XX
work_v44vkh6bzramngvcrficy2no44	220	14	%	%	NN
work_v44vkh6bzramngvcrficy2no44	220	15	j	j	NN
work_v44vkh6bzramngvcrficy2no44	220	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	220	17	"	"	``
work_v44vkh6bzramngvcrficy2no44	220	18	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	220	19	>	>	XX
work_v44vkh6bzramngvcrficy2no44	220	20	"	"	''
work_v44vkh6bzramngvcrficy2no44	220	21	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	220	22	0	0	CD
work_v44vkh6bzramngvcrficy2no44	220	23	otherwise	otherwise	RB
work_v44vkh6bzramngvcrficy2no44	220	24	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	220	25	BP	BP	NNP
work_v44vkh6bzramngvcrficy2no44	220	26	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	220	27	a	a	DT
work_v44vkh6bzramngvcrficy2no44	220	28	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	220	29	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	220	30	level	level	NN
work_v44vkh6bzramngvcrficy2no44	220	31	brevity	brevity	NN
work_v44vkh6bzramngvcrficy2no44	220	32	penalty	penalty	NN
work_v44vkh6bzramngvcrficy2no44	220	33	,	,	,
work_v44vkh6bzramngvcrficy2no44	220	34	similar	similar	JJ
work_v44vkh6bzramngvcrficy2no44	220	35	to	to	IN
work_v44vkh6bzramngvcrficy2no44	220	36	that	that	DT
work_v44vkh6bzramngvcrficy2no44	220	37	of	of	IN
work_v44vkh6bzramngvcrficy2no44	220	38	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	220	39	.	.	.
work_v44vkh6bzramngvcrficy2no44	221	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	221	2	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	221	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	221	4	robust	robust	JJ
work_v44vkh6bzramngvcrficy2no44	221	5	to	to	IN
work_v44vkh6bzramngvcrficy2no44	221	6	lexical	lexical	JJ
work_v44vkh6bzramngvcrficy2no44	221	7	choice	choice	NN
work_v44vkh6bzramngvcrficy2no44	221	8	because	because	IN
work_v44vkh6bzramngvcrficy2no44	221	9	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	221	10	performs	perform	VBZ
work_v44vkh6bzramngvcrficy2no44	221	11	no	no	DT
work_v44vkh6bzramngvcrficy2no44	221	12	comparison	comparison	NN
work_v44vkh6bzramngvcrficy2no44	221	13	between	between	IN
work_v44vkh6bzramngvcrficy2no44	221	14	output	output	NN
work_v44vkh6bzramngvcrficy2no44	221	15	and	and	CC
work_v44vkh6bzramngvcrficy2no44	221	16	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	221	17	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	221	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	221	19	but	but	CC
work_v44vkh6bzramngvcrficy2no44	221	20	only	only	RB
work_v44vkh6bzramngvcrficy2no44	221	21	between	between	IN
work_v44vkh6bzramngvcrficy2no44	221	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	221	23	positions	position	NNS
work_v44vkh6bzramngvcrficy2no44	221	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	221	25	their	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	221	26	translations	translation	NNS
work_v44vkh6bzramngvcrficy2no44	221	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	222	1	Besides	besides	RB
work_v44vkh6bzramngvcrficy2no44	222	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	222	3	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	222	4	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	222	5	shown	show	VBN
work_v44vkh6bzramngvcrficy2no44	222	6	to	to	IN
work_v44vkh6bzramngvcrficy2no44	222	7	corre-	corre-	NNP
work_v44vkh6bzramngvcrficy2no44	222	8	late	late	RB
work_v44vkh6bzramngvcrficy2no44	222	9	strongly	strongly	RB
work_v44vkh6bzramngvcrficy2no44	222	10	with	with	IN
work_v44vkh6bzramngvcrficy2no44	222	11	human	human	JJ
work_v44vkh6bzramngvcrficy2no44	222	12	judgements	judgement	NNS
work_v44vkh6bzramngvcrficy2no44	222	13	of	of	IN
work_v44vkh6bzramngvcrficy2no44	222	14	fluency	fluency	NN
work_v44vkh6bzramngvcrficy2no44	222	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	223	1	Our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	223	2	work	work	NN
work_v44vkh6bzramngvcrficy2no44	223	3	specifically	specifically	RB
work_v44vkh6bzramngvcrficy2no44	223	4	addresses	address	VBZ
work_v44vkh6bzramngvcrficy2no44	223	5	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	223	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	223	7	range	range	NN
work_v44vkh6bzramngvcrficy2no44	223	8	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	223	9	ordering	ordering	JJ
work_v44vkh6bzramngvcrficy2no44	223	10	phenomena	phenomenon	NNS
work_v44vkh6bzramngvcrficy2no44	223	11	in	in	IN
work_v44vkh6bzramngvcrficy2no44	223	12	language	language	NN
work_v44vkh6bzramngvcrficy2no44	223	13	pairs	pair	NNS
work_v44vkh6bzramngvcrficy2no44	223	14	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	223	15	these	these	DT
work_v44vkh6bzramngvcrficy2no44	223	16	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	223	17	quite	quite	RB
work_v44vkh6bzramngvcrficy2no44	223	18	rare	rare	JJ
work_v44vkh6bzramngvcrficy2no44	223	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	223	20	although	although	IN
work_v44vkh6bzramngvcrficy2no44	223	21	crucial	crucial	JJ
work_v44vkh6bzramngvcrficy2no44	223	22	for	for	IN
work_v44vkh6bzramngvcrficy2no44	223	23	preserving	preserve	VBG
work_v44vkh6bzramngvcrficy2no44	223	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	223	25	source	source	NN
work_v44vkh6bzramngvcrficy2no44	223	26	text	text	NN
work_v44vkh6bzramngvcrficy2no44	223	27	meaning	meaning	NN
work_v44vkh6bzramngvcrficy2no44	223	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	224	1	Hence	hence	RB
work_v44vkh6bzramngvcrficy2no44	224	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	224	3	an	an	DT
work_v44vkh6bzramngvcrficy2no44	224	4	improvement	improvement	NN
work_v44vkh6bzramngvcrficy2no44	224	5	at	at	IN
work_v44vkh6bzramngvcrficy2no44	224	6	this	this	DT
work_v44vkh6bzramngvcrficy2no44	224	7	level	level	NN
work_v44vkh6bzramngvcrficy2no44	224	8	may	may	MD
work_v44vkh6bzramngvcrficy2no44	224	9	not	not	RB
work_v44vkh6bzramngvcrficy2no44	224	10	be	be	VB
work_v44vkh6bzramngvcrficy2no44	224	11	detected	detect	VBN
work_v44vkh6bzramngvcrficy2no44	224	12	by	by	IN
work_v44vkh6bzramngvcrficy2no44	224	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	224	14	general	general	JJ
work_v44vkh6bzramngvcrficy2no44	224	15	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	224	16	purpose	purpose	NN
work_v44vkh6bzramngvcrficy2no44	224	17	metrics	metric	NNS
work_v44vkh6bzramngvcrficy2no44	224	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	225	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	225	2	then	then	RB
work_v44vkh6bzramngvcrficy2no44	225	3	develop	develop	VBP
work_v44vkh6bzramngvcrficy2no44	225	4	a	a	DT
work_v44vkh6bzramngvcrficy2no44	225	5	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	225	6	variant	variant	NN
work_v44vkh6bzramngvcrficy2no44	225	7	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	225	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	225	9	only	only	RB
work_v44vkh6bzramngvcrficy2no44	225	10	334	334	CD
work_v44vkh6bzramngvcrficy2no44	225	11	sensitive	sensitive	JJ
work_v44vkh6bzramngvcrficy2no44	225	12	to	to	IN
work_v44vkh6bzramngvcrficy2no44	225	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	225	14	positioning	positioning	NN
work_v44vkh6bzramngvcrficy2no44	225	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	225	16	specific	specific	JJ
work_v44vkh6bzramngvcrficy2no44	225	17	input	input	NN
work_v44vkh6bzramngvcrficy2no44	225	18	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	225	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	226	1	Assuming	assume	VBG
work_v44vkh6bzramngvcrficy2no44	226	2	that	that	IN
work_v44vkh6bzramngvcrficy2no44	226	3	each	each	DT
work_v44vkh6bzramngvcrficy2no44	226	4	input	input	JJ
work_v44vkh6bzramngvcrficy2no44	226	5	word	word	NN
work_v44vkh6bzramngvcrficy2no44	226	6	fi	fi	NN
work_v44vkh6bzramngvcrficy2no44	226	7	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	226	8	assigned	assign	VBN
work_v44vkh6bzramngvcrficy2no44	226	9	a	a	DT
work_v44vkh6bzramngvcrficy2no44	226	10	weight	weight	NN
work_v44vkh6bzramngvcrficy2no44	226	11	!	!	.
work_v44vkh6bzramngvcrficy2no44	226	12	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	226	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	226	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	226	15	formula	formula	NN
work_v44vkh6bzramngvcrficy2no44	226	16	above	above	RB
work_v44vkh6bzramngvcrficy2no44	226	17	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	226	18	modified	modify	VBN
work_v44vkh6bzramngvcrficy2no44	226	19	as	as	IN
work_v44vkh6bzramngvcrficy2no44	226	20	follows	follow	VBZ
work_v44vkh6bzramngvcrficy2no44	226	21	:	:	:
work_v44vkh6bzramngvcrficy2no44	226	22	d"(i	d"(i	NNP
work_v44vkh6bzramngvcrficy2no44	226	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	226	24	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	226	25	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	226	26	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	226	27	$	$	$
work_v44vkh6bzramngvcrficy2no44	226	28	!	!	.
work_v44vkh6bzramngvcrficy2no44	226	29	i+!j	i+!j	NNP
work_v44vkh6bzramngvcrficy2no44	226	30	if	if	IN
work_v44vkh6bzramngvcrficy2no44	226	31	%	%	NN
work_v44vkh6bzramngvcrficy2no44	226	32	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	226	33	<	<	XX
work_v44vkh6bzramngvcrficy2no44	226	34	%	%	NN
work_v44vkh6bzramngvcrficy2no44	226	35	j	j	NN
work_v44vkh6bzramngvcrficy2no44	226	36	and	and	CC
work_v44vkh6bzramngvcrficy2no44	226	37	"	"	``
work_v44vkh6bzramngvcrficy2no44	226	38	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	226	39	>	>	XX
work_v44vkh6bzramngvcrficy2no44	226	40	"	"	''
work_v44vkh6bzramngvcrficy2no44	226	41	j	j	NNP
work_v44vkh6bzramngvcrficy2no44	226	42	0	0	CD
work_v44vkh6bzramngvcrficy2no44	226	43	otherwise	otherwise	RB
work_v44vkh6bzramngvcrficy2no44	226	44	A	a	DT
work_v44vkh6bzramngvcrficy2no44	226	45	similar	similar	JJ
work_v44vkh6bzramngvcrficy2no44	226	46	element	element	NN
work_v44vkh6bzramngvcrficy2no44	226	47	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	226	48	weighted	weighted	JJ
work_v44vkh6bzramngvcrficy2no44	226	49	version	version	NN
work_v44vkh6bzramngvcrficy2no44	226	50	of	of	IN
work_v44vkh6bzramngvcrficy2no44	226	51	Kendall	Kendall	NNP
work_v44vkh6bzramngvcrficy2no44	226	52	Tau	Tau	NNP
work_v44vkh6bzramngvcrficy2no44	226	53	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	226	54	proposed	propose	VBN
work_v44vkh6bzramngvcrficy2no44	226	55	by	by	IN
work_v44vkh6bzramngvcrficy2no44	226	56	Kumar	Kumar	NNP
work_v44vkh6bzramngvcrficy2no44	226	57	and	and	CC
work_v44vkh6bzramngvcrficy2no44	226	58	Vassilvitskii	Vassilvitskii	NNP
work_v44vkh6bzramngvcrficy2no44	226	59	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	226	60	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	226	61	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	226	62	to	to	TO
work_v44vkh6bzramngvcrficy2no44	226	63	evaluate	evaluate	VB
work_v44vkh6bzramngvcrficy2no44	226	64	document	document	NN
work_v44vkh6bzramngvcrficy2no44	226	65	rankings	ranking	NNS
work_v44vkh6bzramngvcrficy2no44	226	66	in	in	IN
work_v44vkh6bzramngvcrficy2no44	226	67	information	information	NN
work_v44vkh6bzramngvcrficy2no44	226	68	retrieval	retrieval	NN
work_v44vkh6bzramngvcrficy2no44	226	69	.	.	.
work_v44vkh6bzramngvcrficy2no44	227	1	Because	because	IN
work_v44vkh6bzramngvcrficy2no44	227	2	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	227	3	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	227	4	errors	error	NNS
work_v44vkh6bzramngvcrficy2no44	227	5	in	in	IN
work_v44vkh6bzramngvcrficy2no44	227	6	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	227	7	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	227	8	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	227	9	mostly	mostly	RB
work_v44vkh6bzramngvcrficy2no44	227	10	affect	affect	VBP
work_v44vkh6bzramngvcrficy2no44	227	11	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	227	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	227	13	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	227	14	set	set	VBD
work_v44vkh6bzramngvcrficy2no44	227	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	227	16	weights	weight	NNS
work_v44vkh6bzramngvcrficy2no44	227	17	to	to	IN
work_v44vkh6bzramngvcrficy2no44	227	18	1	1	CD
work_v44vkh6bzramngvcrficy2no44	227	19	for	for	IN
work_v44vkh6bzramngvcrficy2no44	227	20	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	227	21	and	and	CC
work_v44vkh6bzramngvcrficy2no44	227	22	0	0	CD
work_v44vkh6bzramngvcrficy2no44	227	23	for	for	IN
work_v44vkh6bzramngvcrficy2no44	227	24	all	all	DT
work_v44vkh6bzramngvcrficy2no44	227	25	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	227	26	words	word	NNS
work_v44vkh6bzramngvcrficy2no44	227	27	to	to	TO
work_v44vkh6bzramngvcrficy2no44	227	28	only	only	RB
work_v44vkh6bzramngvcrficy2no44	227	29	capture	capture	VB
work_v44vkh6bzramngvcrficy2no44	227	30	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	227	31	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	227	32	ordering	order	VBG
work_v44vkh6bzramngvcrficy2no44	227	33	errors	error	NNS
work_v44vkh6bzramngvcrficy2no44	227	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	227	35	and	and	CC
work_v44vkh6bzramngvcrficy2no44	227	36	call	call	VB
work_v44vkh6bzramngvcrficy2no44	227	37	the	the	DT
work_v44vkh6bzramngvcrficy2no44	227	38	resulting	result	VBG
work_v44vkh6bzramngvcrficy2no44	227	39	metric	metric	JJ
work_v44vkh6bzramngvcrficy2no44	227	40	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	227	41	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	227	42	V.	V.	NNP
work_v44vkh6bzramngvcrficy2no44	228	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	228	2	source	source	NN
work_v44vkh6bzramngvcrficy2no44	228	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	228	4	reference	reference	NN
work_v44vkh6bzramngvcrficy2no44	228	5	word	word	NN
work_v44vkh6bzramngvcrficy2no44	228	6	alignments	alignment	NNS
work_v44vkh6bzramngvcrficy2no44	228	7	needed	need	VBN
work_v44vkh6bzramngvcrficy2no44	228	8	to	to	TO
work_v44vkh6bzramngvcrficy2no44	228	9	compute	compute	VB
work_v44vkh6bzramngvcrficy2no44	228	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	228	11	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	228	12	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	228	13	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	228	14	generated	generate	VBN
work_v44vkh6bzramngvcrficy2no44	228	15	by	by	IN
work_v44vkh6bzramngvcrficy2no44	228	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	228	17	Berkeley	Berkeley	NNP
work_v44vkh6bzramngvcrficy2no44	228	18	Aligner	Aligner	NNP
work_v44vkh6bzramngvcrficy2no44	228	19	previously	previously	RB
work_v44vkh6bzramngvcrficy2no44	228	20	trained	train	VBD
work_v44vkh6bzramngvcrficy2no44	228	21	on	on	IN
work_v44vkh6bzramngvcrficy2no44	228	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	228	23	training	training	NN
work_v44vkh6bzramngvcrficy2no44	228	24	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	228	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	229	1	Source	source	NN
work_v44vkh6bzramngvcrficy2no44	229	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	229	3	output	output	NN
work_v44vkh6bzramngvcrficy2no44	229	4	word	word	NN
work_v44vkh6bzramngvcrficy2no44	229	5	alignments	alignment	NNS
work_v44vkh6bzramngvcrficy2no44	229	6	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	229	7	instead	instead	RB
work_v44vkh6bzramngvcrficy2no44	229	8	ob-	ob-	RB
work_v44vkh6bzramngvcrficy2no44	229	9	tained	taine	VBN
work_v44vkh6bzramngvcrficy2no44	229	10	from	from	IN
work_v44vkh6bzramngvcrficy2no44	229	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	229	12	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	229	13	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	229	14	trace	trace	NN
work_v44vkh6bzramngvcrficy2no44	229	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	230	1	5.4	5.4	CD
work_v44vkh6bzramngvcrficy2no44	230	2	Results	result	NNS
work_v44vkh6bzramngvcrficy2no44	230	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	230	4	discussion	discussion	NN
work_v44vkh6bzramngvcrficy2no44	230	5	To	to	TO
work_v44vkh6bzramngvcrficy2no44	230	6	motivate	motivate	VB
work_v44vkh6bzramngvcrficy2no44	230	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	230	8	choice	choice	NN
work_v44vkh6bzramngvcrficy2no44	230	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	230	10	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	230	11	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	230	12	setup	setup	NN
work_v44vkh6bzramngvcrficy2no44	230	13	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	230	14	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	230	15	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	230	16	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	230	17	and	and	CC
work_v44vkh6bzramngvcrficy2no44	230	18	DL=8	DL=8	NNP
work_v44vkh6bzramngvcrficy2no44	230	19	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	230	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	230	21	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	230	22	first	first	RB
work_v44vkh6bzramngvcrficy2no44	230	23	compare	compare	VBP
work_v44vkh6bzramngvcrficy2no44	230	24	the	the	DT
work_v44vkh6bzramngvcrficy2no44	230	25	per-	per-	NN
work_v44vkh6bzramngvcrficy2no44	230	26	formance	formance	NN
work_v44vkh6bzramngvcrficy2no44	230	27	of	of	IN
work_v44vkh6bzramngvcrficy2no44	230	28	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	230	29	and	and	CC
work_v44vkh6bzramngvcrficy2no44	230	30	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	230	31	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	230	32	costs	cost	NNS
work_v44vkh6bzramngvcrficy2no44	230	33	un-	un-	XX
work_v44vkh6bzramngvcrficy2no44	230	34	der	der	XX
work_v44vkh6bzramngvcrficy2no44	230	35	various	various	JJ
work_v44vkh6bzramngvcrficy2no44	230	36	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	230	37	conditions	condition	NNS
work_v44vkh6bzramngvcrficy2no44	230	38	.	.	.
work_v44vkh6bzramngvcrficy2no44	231	1	!	!	.
work_v44vkh6bzramngvcrficy2no44	232	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	232	2	#	#	$
work_v44vkh6bzramngvcrficy2no44	232	3	$	$	$
work_v44vkh6bzramngvcrficy2no44	232	4	%	%	NN
work_v44vkh6bzramngvcrficy2no44	232	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	233	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	233	2	#	#	$
work_v44vkh6bzramngvcrficy2no44	233	3	&	&	CC
work_v44vkh6bzramngvcrficy2no44	233	4	'	'	''
work_v44vkh6bzramngvcrficy2no44	233	5	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	233	6	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	233	7	%	%	NN
work_v44vkh6bzramngvcrficy2no44	233	8	!	!	.
work_v44vkh6bzramngvcrficy2no44	234	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	234	2	#	#	NN
work_v44vkh6bzramngvcrficy2no44	234	3	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	234	4	$	$	$
work_v44vkh6bzramngvcrficy2no44	234	5	%	%	NN
work_v44vkh6bzramngvcrficy2no44	234	6	!	!	.
work_v44vkh6bzramngvcrficy2no44	235	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	235	2	#	#	NN
work_v44vkh6bzramngvcrficy2no44	235	3	&	&	CC
work_v44vkh6bzramngvcrficy2no44	235	4	%	%	NN
work_v44vkh6bzramngvcrficy2no44	235	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	236	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	236	2	#	#	$
work_v44vkh6bzramngvcrficy2no44	236	3	$	$	$
work_v44vkh6bzramngvcrficy2no44	236	4	%	%	NN
work_v44vkh6bzramngvcrficy2no44	236	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	237	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	237	2	#	#	NNP
work_v44vkh6bzramngvcrficy2no44	237	3	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	237	4	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	237	5	%	%	NN
work_v44vkh6bzramngvcrficy2no44	237	6	!	!	.
work_v44vkh6bzramngvcrficy2no44	238	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	238	2	#	#	NN
work_v44vkh6bzramngvcrficy2no44	238	3	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	238	4	$	$	$
work_v44vkh6bzramngvcrficy2no44	238	5	%	%	NN
work_v44vkh6bzramngvcrficy2no44	238	6	!	!	.
work_v44vkh6bzramngvcrficy2no44	239	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	239	2	#	#	VBG
work_v44vkh6bzramngvcrficy2no44	239	3	"	"	''
work_v44vkh6bzramngvcrficy2no44	239	4	$	$	$
work_v44vkh6bzramngvcrficy2no44	239	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	239	6	%	%	NN
work_v44vkh6bzramngvcrficy2no44	239	7	#	#	$
work_v44vkh6bzramngvcrficy2no44	239	8	"	"	``
work_v44vkh6bzramngvcrficy2no44	239	9	$	$	$
work_v44vkh6bzramngvcrficy2no44	239	10	!	!	.
work_v44vkh6bzramngvcrficy2no44	240	1	!	!	NFP
work_v44vkh6bzramngvcrficy2no44	240	2	#	#	$
work_v44vkh6bzramngvcrficy2no44	240	3	"	"	``
work_v44vkh6bzramngvcrficy2no44	240	4	$	$	$
work_v44vkh6bzramngvcrficy2no44	240	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	240	6	&	&	CC
work_v44vkh6bzramngvcrficy2no44	240	7	#	#	$
work_v44vkh6bzramngvcrficy2no44	240	8	"	"	``
work_v44vkh6bzramngvcrficy2no44	240	9	$	$	$
work_v44vkh6bzramngvcrficy2no44	240	10	!	!	.
work_v44vkh6bzramngvcrficy2no44	241	1	'	'	``
work_v44vkh6bzramngvcrficy2no44	241	2	#	#	$
work_v44vkh6bzramngvcrficy2no44	241	3	"	"	``
work_v44vkh6bzramngvcrficy2no44	241	4	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	5	&	&	CC
work_v44vkh6bzramngvcrficy2no44	241	6	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	241	7	#	#	NN
work_v44vkh6bzramngvcrficy2no44	241	8	"	"	``
work_v44vkh6bzramngvcrficy2no44	241	9	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	10	&	&	CC
work_v44vkh6bzramngvcrficy2no44	241	11	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	241	12	#	#	NNP
work_v44vkh6bzramngvcrficy2no44	241	13	"	"	``
work_v44vkh6bzramngvcrficy2no44	241	14	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	15	&	&	CC
work_v44vkh6bzramngvcrficy2no44	241	16	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	241	17	#	#	NN
work_v44vkh6bzramngvcrficy2no44	241	18	"	"	``
work_v44vkh6bzramngvcrficy2no44	241	19	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	20	&	&	CC
work_v44vkh6bzramngvcrficy2no44	241	21	+	+	CD
work_v44vkh6bzramngvcrficy2no44	241	22	#	#	NN
work_v44vkh6bzramngvcrficy2no44	241	23	"	"	``
work_v44vkh6bzramngvcrficy2no44	241	24	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	26	+	+	CC
work_v44vkh6bzramngvcrficy2no44	241	27	#	#	NN
work_v44vkh6bzramngvcrficy2no44	241	28	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	241	29	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	31	+	+	CC
work_v44vkh6bzramngvcrficy2no44	241	32	#	#	$
work_v44vkh6bzramngvcrficy2no44	241	33	%	%	NN
work_v44vkh6bzramngvcrficy2no44	241	34	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	35	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	36	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	37	#	#	NNP
work_v44vkh6bzramngvcrficy2no44	241	38	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	241	39	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	40	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	42	#	#	$
work_v44vkh6bzramngvcrficy2no44	241	43	%	%	NN
work_v44vkh6bzramngvcrficy2no44	241	44	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	45	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	46	"	"	''
work_v44vkh6bzramngvcrficy2no44	241	47	#	#	NN
work_v44vkh6bzramngvcrficy2no44	241	48	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	241	49	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	50	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	241	51	+	+	CC
work_v44vkh6bzramngvcrficy2no44	241	52	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	53	%	%	NN
work_v44vkh6bzramngvcrficy2no44	241	54	-"./%	-"./%	:
work_v44vkh6bzramngvcrficy2no44	241	55	-./01	-./01	:
work_v44vkh6bzramngvcrficy2no44	241	56	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	57	23.45./5	23.45./5	CD
work_v44vkh6bzramngvcrficy2no44	241	58	$	$	$
work_v44vkh6bzramngvcrficy2no44	241	59	Figure	figure	NN
work_v44vkh6bzramngvcrficy2no44	241	60	2	2	CD
work_v44vkh6bzramngvcrficy2no44	241	61	:	:	:
work_v44vkh6bzramngvcrficy2no44	241	62	Standard	standard	NN
work_v44vkh6bzramngvcrficy2no44	241	63	vs	vs	IN
work_v44vkh6bzramngvcrficy2no44	241	64	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	241	65	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	241	66	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	241	67	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	241	68	on	on	IN
work_v44vkh6bzramngvcrficy2no44	241	69	eval08-nw	eval08-nw	NNS
work_v44vkh6bzramngvcrficy2no44	241	70	under	under	IN
work_v44vkh6bzramngvcrficy2no44	241	71	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	241	72	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	241	73	limits	limit	NNS
work_v44vkh6bzramngvcrficy2no44	241	74	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	241	75	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	241	76	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	241	77	,	,	,
work_v44vkh6bzramngvcrficy2no44	241	78	using	use	VBG
work_v44vkh6bzramngvcrficy2no44	241	79	the	the	DT
work_v44vkh6bzramngvcrficy2no44	241	80	medium	medium	NN
work_v44vkh6bzramngvcrficy2no44	241	81	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	241	82	size	size	NN
work_v44vkh6bzramngvcrficy2no44	241	83	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	241	84	.	.	.
work_v44vkh6bzramngvcrficy2no44	242	1	Best	good	JJS
work_v44vkh6bzramngvcrficy2no44	242	2	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	242	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	242	4	on	on	IN
work_v44vkh6bzramngvcrficy2no44	242	5	top	top	JJ
work_v44vkh6bzramngvcrficy2no44	242	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	242	7	right	right	NN
work_v44vkh6bzramngvcrficy2no44	242	8	corner	corner	NN
work_v44vkh6bzramngvcrficy2no44	242	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	243	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	243	2	shown	show	VBN
work_v44vkh6bzramngvcrficy2no44	243	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	243	4	Fig	Fig	NNP
work_v44vkh6bzramngvcrficy2no44	243	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	244	1	2	2	LS
work_v44vkh6bzramngvcrficy2no44	244	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	244	3	most	most	JJS
work_v44vkh6bzramngvcrficy2no44	244	4	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	244	5	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	244	6	close	close	JJ
work_v44vkh6bzramngvcrficy2no44	244	7	to	to	IN
work_v44vkh6bzramngvcrficy2no44	244	8	each	each	DT
work_v44vkh6bzramngvcrficy2no44	244	9	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	244	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	244	11	terms	term	NNS
work_v44vkh6bzramngvcrficy2no44	244	12	of	of	IN
work_v44vkh6bzramngvcrficy2no44	244	13	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	244	14	and	and	CC
work_v44vkh6bzramngvcrficy2no44	244	15	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	244	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	244	17	but	but	CC
work_v44vkh6bzramngvcrficy2no44	244	18	early	early	RB
work_v44vkh6bzramngvcrficy2no44	244	19	distor-	distor-	JJ
work_v44vkh6bzramngvcrficy2no44	244	20	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	244	21	consistently	consistently	RB
work_v44vkh6bzramngvcrficy2no44	244	22	outperforms	outperform	VBZ
work_v44vkh6bzramngvcrficy2no44	244	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	244	24	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	244	25	one	one	CD
work_v44vkh6bzramngvcrficy2no44	244	26	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	244	27	sta-	sta-	VB
work_v44vkh6bzramngvcrficy2no44	244	28	tistically	tistically	RB
work_v44vkh6bzramngvcrficy2no44	244	29	significant	significant	JJ
work_v44vkh6bzramngvcrficy2no44	244	30	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	244	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	245	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	245	2	most	most	RBS
work_v44vkh6bzramngvcrficy2no44	245	3	striking	striking	JJ
work_v44vkh6bzramngvcrficy2no44	245	4	difference	difference	NN
work_v44vkh6bzramngvcrficy2no44	245	5	appears	appear	VBZ
work_v44vkh6bzramngvcrficy2no44	245	6	at	at	IN
work_v44vkh6bzramngvcrficy2no44	245	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	245	8	very	very	RB
work_v44vkh6bzramngvcrficy2no44	245	9	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	245	10	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	245	11	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	245	12	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	245	13	18	18	CD
work_v44vkh6bzramngvcrficy2no44	245	14	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	245	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	245	16	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	245	17	standard	standard	JJ
work_v44vkh6bzramngvcrficy2no44	245	18	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	245	19	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	245	20	drop	drop	VBP
work_v44vkh6bzramngvcrficy2no44	245	21	by	by	IN
work_v44vkh6bzramngvcrficy2no44	245	22	more	more	JJR
work_v44vkh6bzramngvcrficy2no44	245	23	than	than	IN
work_v44vkh6bzramngvcrficy2no44	245	24	1	1	CD
work_v44vkh6bzramngvcrficy2no44	245	25	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	245	26	point	point	NN
work_v44vkh6bzramngvcrficy2no44	245	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	245	28	almost	almost	RB
work_v44vkh6bzramngvcrficy2no44	245	29	7	7	CD
work_v44vkh6bzramngvcrficy2no44	245	30	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	245	31	points	point	NNS
work_v44vkh6bzramngvcrficy2no44	245	32	!	!	.
work_v44vkh6bzramngvcrficy2no44	246	1	Early	early	JJ
work_v44vkh6bzramngvcrficy2no44	246	2	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	246	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	246	4	much	much	RB
work_v44vkh6bzramngvcrficy2no44	246	5	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	246	6	robust	robust	JJ
work_v44vkh6bzramngvcrficy2no44	246	7	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	246	8	only	only	RB
work_v44vkh6bzramngvcrficy2no44	246	9	-1	-1	.
work_v44vkh6bzramngvcrficy2no44	246	10	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	246	11	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	246	12	going	go	VBG
work_v44vkh6bzramngvcrficy2no44	246	13	from	from	IN
work_v44vkh6bzramngvcrficy2no44	246	14	DL=8	DL=8	NNP
work_v44vkh6bzramngvcrficy2no44	246	15	to	to	IN
work_v44vkh6bzramngvcrficy2no44	246	16	DL=18	DL=18	NNP
work_v44vkh6bzramngvcrficy2no44	246	17	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	246	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	246	19	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	246	20	makes	make	VBZ
work_v44vkh6bzramngvcrficy2no44	246	21	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	246	22	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	246	23	system	system	NN
work_v44vkh6bzramngvcrficy2no44	246	24	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	246	25	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	246	26	at	at	IN
work_v44vkh6bzramngvcrficy2no44	246	27	the	the	DT
work_v44vkh6bzramngvcrficy2no44	246	28	level	level	NN
work_v44vkh6bzramngvcrficy2no44	246	29	of	of	IN
work_v44vkh6bzramngvcrficy2no44	246	30	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	246	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	247	1	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	247	2	5	5	CD
work_v44vkh6bzramngvcrficy2no44	247	3	presents	present	VBZ
work_v44vkh6bzramngvcrficy2no44	247	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	247	5	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	247	6	obtained	obtain	VBN
work_v44vkh6bzramngvcrficy2no44	247	7	by	by	IN
work_v44vkh6bzramngvcrficy2no44	247	8	integrat-	integrat-	NNP
work_v44vkh6bzramngvcrficy2no44	247	9	ing	e	VBG
work_v44vkh6bzramngvcrficy2no44	247	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	247	11	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	247	12	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	247	13	model	model	NN
work_v44vkh6bzramngvcrficy2no44	247	14	as	as	IN
work_v44vkh6bzramngvcrficy2no44	247	15	an	an	DT
work_v44vkh6bzramngvcrficy2no44	247	16	additional	additional	JJ
work_v44vkh6bzramngvcrficy2no44	247	17	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	247	18	function	function	NN
work_v44vkh6bzramngvcrficy2no44	247	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	247	20	and	and	CC
work_v44vkh6bzramngvcrficy2no44	247	21	by	by	IN
work_v44vkh6bzramngvcrficy2no44	247	22	applying	apply	VBG
work_v44vkh6bzramngvcrficy2no44	247	23	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	247	24	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	247	25	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	247	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	248	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	248	2	upper	upper	JJ
work_v44vkh6bzramngvcrficy2no44	248	3	part	part	NN
work_v44vkh6bzramngvcrficy2no44	248	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	248	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	248	6	table	table	NN
work_v44vkh6bzramngvcrficy2no44	248	7	refers	refer	VBZ
work_v44vkh6bzramngvcrficy2no44	248	8	to	to	IN
work_v44vkh6bzramngvcrficy2no44	248	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	248	10	medium	medium	JJ
work_v44vkh6bzramngvcrficy2no44	248	11	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	248	12	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	248	13	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	248	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	248	15	while	while	IN
work_v44vkh6bzramngvcrficy2no44	248	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	248	17	lower	low	JJR
work_v44vkh6bzramngvcrficy2no44	248	18	part	part	NN
work_v44vkh6bzramngvcrficy2no44	248	19	refers	refer	VBZ
work_v44vkh6bzramngvcrficy2no44	248	20	to	to	IN
work_v44vkh6bzramngvcrficy2no44	248	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	248	22	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	248	23	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	248	24	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	248	25	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	248	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	249	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	249	2	each	each	DT
work_v44vkh6bzramngvcrficy2no44	249	3	part	part	NN
work_v44vkh6bzramngvcrficy2no44	249	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	249	5	statis-	statis-	NNP
work_v44vkh6bzramngvcrficy2no44	249	6	tical	tical	JJ
work_v44vkh6bzramngvcrficy2no44	249	7	significance	significance	NN
work_v44vkh6bzramngvcrficy2no44	249	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	249	9	computed	compute	VBN
work_v44vkh6bzramngvcrficy2no44	249	10	against	against	IN
work_v44vkh6bzramngvcrficy2no44	249	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	249	12	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	249	13	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	249	14	B	b	NN
work_v44vkh6bzramngvcrficy2no44	249	15	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	249	16	by	by	IN
work_v44vkh6bzramngvcrficy2no44	249	17	approximate	approximate	JJ
work_v44vkh6bzramngvcrficy2no44	249	18	randomization	randomization	NN
work_v44vkh6bzramngvcrficy2no44	249	19	as	as	IN
work_v44vkh6bzramngvcrficy2no44	249	20	in	in	IN
work_v44vkh6bzramngvcrficy2no44	249	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	249	22	Riezler	Riezler	NNP
work_v44vkh6bzramngvcrficy2no44	249	23	and	and	CC
work_v44vkh6bzramngvcrficy2no44	249	24	Maxwell	Maxwell	NNP
work_v44vkh6bzramngvcrficy2no44	249	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	249	26	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	249	27	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	249	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	250	1	Run	Run	NNP
work_v44vkh6bzramngvcrficy2no44	250	2	times	time	NNS
work_v44vkh6bzramngvcrficy2no44	250	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	250	4	obtained	obtain	VBN
work_v44vkh6bzramngvcrficy2no44	250	5	by	by	IN
work_v44vkh6bzramngvcrficy2no44	250	6	an	an	DT
work_v44vkh6bzramngvcrficy2no44	250	7	Intel	Intel	NNP
work_v44vkh6bzramngvcrficy2no44	250	8	Xeon	Xeon	NNP
work_v44vkh6bzramngvcrficy2no44	250	9	X5650	X5650	NNP
work_v44vkh6bzramngvcrficy2no44	250	10	processor	processor	NN
work_v44vkh6bzramngvcrficy2no44	250	11	on	on	IN
work_v44vkh6bzramngvcrficy2no44	250	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	250	13	first	first	JJ
work_v44vkh6bzramngvcrficy2no44	250	14	500	500	CD
work_v44vkh6bzramngvcrficy2no44	250	15	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	250	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	250	17	eval08-nw	eval08-nw	NNS
work_v44vkh6bzramngvcrficy2no44	250	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	250	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	250	20	exclude	exclude	VB
work_v44vkh6bzramngvcrficy2no44	250	21	loading	loading	NN
work_v44vkh6bzramngvcrficy2no44	250	22	time	time	NN
work_v44vkh6bzramngvcrficy2no44	250	23	of	of	IN
work_v44vkh6bzramngvcrficy2no44	250	24	all	all	DT
work_v44vkh6bzramngvcrficy2no44	250	25	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	250	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	251	1	Medium	medium	JJ
work_v44vkh6bzramngvcrficy2no44	251	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	251	3	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	251	4	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	251	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	252	1	Integrating	integrate	VBG
work_v44vkh6bzramngvcrficy2no44	252	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	252	3	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	252	4	model	model	NN
work_v44vkh6bzramngvcrficy2no44	252	5	as	as	IN
work_v44vkh6bzramngvcrficy2no44	252	6	an	an	DT
work_v44vkh6bzramngvcrficy2no44	252	7	additional	additional	JJ
work_v44vkh6bzramngvcrficy2no44	252	8	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	252	9	function	function	NN
work_v44vkh6bzramngvcrficy2no44	252	10	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	252	11	in	in	IN
work_v44vkh6bzramngvcrficy2no44	252	12	small	small	JJ
work_v44vkh6bzramngvcrficy2no44	252	13	but	but	CC
work_v44vkh6bzramngvcrficy2no44	252	14	consistent	consistent	JJ
work_v44vkh6bzramngvcrficy2no44	252	15	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	252	16	in	in	IN
work_v44vkh6bzramngvcrficy2no44	252	17	all	all	DT
work_v44vkh6bzramngvcrficy2no44	252	18	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	252	19	condi-	condi-	NNP
work_v44vkh6bzramngvcrficy2no44	252	20	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	252	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	252	22	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	252	23	shows	show	VBZ
work_v44vkh6bzramngvcrficy2no44	252	24	that	that	IN
work_v44vkh6bzramngvcrficy2no44	252	25	this	this	DT
work_v44vkh6bzramngvcrficy2no44	252	26	type	type	NN
work_v44vkh6bzramngvcrficy2no44	252	27	of	of	IN
work_v44vkh6bzramngvcrficy2no44	252	28	model	model	NN
work_v44vkh6bzramngvcrficy2no44	252	29	conveys	convey	NNS
work_v44vkh6bzramngvcrficy2no44	252	30	information	information	NN
work_v44vkh6bzramngvcrficy2no44	252	31	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	252	32	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	252	33	missing	miss	VBG
work_v44vkh6bzramngvcrficy2no44	252	34	from	from	IN
work_v44vkh6bzramngvcrficy2no44	252	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	252	36	state	state	NN
work_v44vkh6bzramngvcrficy2no44	252	37	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	252	38	of	of	IN
work_v44vkh6bzramngvcrficy2no44	252	39	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	252	40	the	the	DT
work_v44vkh6bzramngvcrficy2no44	252	41	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	252	42	art	art	NN
work_v44vkh6bzramngvcrficy2no44	252	43	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	252	44	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	252	45	.	.	.
work_v44vkh6bzramngvcrficy2no44	253	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	253	2	regards	regard	VBZ
work_v44vkh6bzramngvcrficy2no44	253	3	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	253	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	253	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	253	6	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	253	7	model	model	NN
work_v44vkh6bzramngvcrficy2no44	253	8	makes	make	VBZ
work_v44vkh6bzramngvcrficy2no44	253	9	decoding	decoding	JJ
work_v44vkh6bzramngvcrficy2no44	253	10	time	time	NN
work_v44vkh6bzramngvcrficy2no44	253	11	increase	increase	VB
work_v44vkh6bzramngvcrficy2no44	253	12	by	by	IN
work_v44vkh6bzramngvcrficy2no44	253	13	8	8	CD
work_v44vkh6bzramngvcrficy2no44	253	14	%	%	NN
work_v44vkh6bzramngvcrficy2no44	253	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	254	1	Among	among	IN
work_v44vkh6bzramngvcrficy2no44	254	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	254	3	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	254	4	settings	setting	NNS
work_v44vkh6bzramngvcrficy2no44	254	5	considered	consider	VBN
work_v44vkh6bzramngvcrficy2no44	254	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	254	7	DL=8	DL=8	NNP
work_v44vkh6bzramngvcrficy2no44	254	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	254	9	con-	con-	NN
work_v44vkh6bzramngvcrficy2no44	254	10	firmed	firm	VBN
work_v44vkh6bzramngvcrficy2no44	254	11	as	as	IN
work_v44vkh6bzramngvcrficy2no44	254	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	254	13	optimal	optimal	JJ
work_v44vkh6bzramngvcrficy2no44	254	14	one	one	CD
work_v44vkh6bzramngvcrficy2no44	254	15	–	–	:
work_v44vkh6bzramngvcrficy2no44	254	16	with	with	IN
work_v44vkh6bzramngvcrficy2no44	254	17	or	or	CC
work_v44vkh6bzramngvcrficy2no44	254	18	without	without	IN
work_v44vkh6bzramngvcrficy2no44	254	19	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	254	20	model	model	NN
work_v44vkh6bzramngvcrficy2no44	254	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	255	1	Raising	raise	VBG
work_v44vkh6bzramngvcrficy2no44	255	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	255	3	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	255	4	to	to	IN
work_v44vkh6bzramngvcrficy2no44	255	5	18	18	CD
work_v44vkh6bzramngvcrficy2no44	255	6	with	with	IN
work_v44vkh6bzramngvcrficy2no44	255	7	no	no	DT
work_v44vkh6bzramngvcrficy2no44	255	8	special	special	JJ
work_v44vkh6bzramngvcrficy2no44	255	9	prun-	prun-	NN
work_v44vkh6bzramngvcrficy2no44	255	10	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	255	11	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	255	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	255	13	negative	negative	JJ
work_v44vkh6bzramngvcrficy2no44	255	14	impact	impact	NN
work_v44vkh6bzramngvcrficy2no44	255	15	on	on	IN
work_v44vkh6bzramngvcrficy2no44	255	16	both	both	DT
work_v44vkh6bzramngvcrficy2no44	255	17	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	255	18	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	255	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	255	20	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	255	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	256	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	256	2	effect	effect	NN
work_v44vkh6bzramngvcrficy2no44	256	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	256	4	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	256	5	visible	visible	JJ
work_v44vkh6bzramngvcrficy2no44	256	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	256	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	256	8	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	256	9	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	256	10	:	:	:
work_v44vkh6bzramngvcrficy2no44	256	11	that	that	RB
work_v44vkh6bzramngvcrficy2no44	256	12	is	is	RB
work_v44vkh6bzramngvcrficy2no44	256	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	256	14	from	from	IN
work_v44vkh6bzramngvcrficy2no44	256	15	84.7	84.7	CD
work_v44vkh6bzramngvcrficy2no44	256	16	to	to	IN
work_v44vkh6bzramngvcrficy2no44	256	17	83.9	83.9	CD
work_v44vkh6bzramngvcrficy2no44	256	18	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	256	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	256	20	from	from	IN
work_v44vkh6bzramngvcrficy2no44	256	21	86.2	86.2	CD
work_v44vkh6bzramngvcrficy2no44	256	22	to	to	IN
work_v44vkh6bzramngvcrficy2no44	256	23	85.8	85.8	CD
work_v44vkh6bzramngvcrficy2no44	256	24	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	256	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	256	26	V	V	NNP
work_v44vkh6bzramngvcrficy2no44	256	27	on	on	IN
work_v44vkh6bzramngvcrficy2no44	256	28	eval09-nw	eval09-nw	NNP
work_v44vkh6bzramngvcrficy2no44	256	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	257	1	Run	Run	NNP
work_v44vkh6bzramngvcrficy2no44	257	2	times	time	NNS
work_v44vkh6bzramngvcrficy2no44	257	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	257	4	almost	almost	RB
work_v44vkh6bzramngvcrficy2no44	257	5	doubled	double	VBN
work_v44vkh6bzramngvcrficy2no44	257	6	:	:	:
work_v44vkh6bzramngvcrficy2no44	257	7	from	from	IN
work_v44vkh6bzramngvcrficy2no44	257	8	87	87	CD
work_v44vkh6bzramngvcrficy2no44	257	9	to	to	IN
work_v44vkh6bzramngvcrficy2no44	257	10	164	164	CD
work_v44vkh6bzramngvcrficy2no44	257	11	and	and	CC
work_v44vkh6bzramngvcrficy2no44	257	12	from	from	IN
work_v44vkh6bzramngvcrficy2no44	257	13	94	94	CD
work_v44vkh6bzramngvcrficy2no44	257	14	to	to	TO
work_v44vkh6bzramngvcrficy2no44	257	15	178	178	CD
work_v44vkh6bzramngvcrficy2no44	257	16	ms	m	NNS
work_v44vkh6bzramngvcrficy2no44	257	17	/	/	SYM
work_v44vkh6bzramngvcrficy2no44	257	18	word	word	NN
work_v44vkh6bzramngvcrficy2no44	257	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	257	20	that	that	DT
work_v44vkh6bzramngvcrficy2no44	257	21	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	257	22	a	a	DT
work_v44vkh6bzramngvcrficy2no44	257	23	89	89	CD
work_v44vkh6bzramngvcrficy2no44	257	24	%	%	NN
work_v44vkh6bzramngvcrficy2no44	257	25	increase	increase	NN
work_v44vkh6bzramngvcrficy2no44	257	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	258	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	258	2	then	then	RB
work_v44vkh6bzramngvcrficy2no44	258	3	proceed	proceed	VBP
work_v44vkh6bzramngvcrficy2no44	258	4	to	to	IN
work_v44vkh6bzramngvcrficy2no44	258	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	258	6	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	258	7	experiment	experiment	NN
work_v44vkh6bzramngvcrficy2no44	258	8	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	258	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	258	10	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	258	11	space	space	NN
work_v44vkh6bzramngvcrficy2no44	258	12	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	258	13	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	258	14	pruned	prune	VBN
work_v44vkh6bzramngvcrficy2no44	258	15	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	258	16	on	on	IN
work_v44vkh6bzramngvcrficy2no44	258	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	258	18	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	258	19	model	model	NN
work_v44vkh6bzramngvcrficy2no44	258	20	score	score	NN
work_v44vkh6bzramngvcrficy2no44	258	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	259	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	259	2	explained	explain	VBN
work_v44vkh6bzramngvcrficy2no44	259	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	259	4	Sect	Sect	NNP
work_v44vkh6bzramngvcrficy2no44	259	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	260	1	4	4	LS
work_v44vkh6bzramngvcrficy2no44	260	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	260	3	a	a	DT
work_v44vkh6bzramngvcrficy2no44	260	4	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	260	5	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	260	6	prunable	prunable	JJ
work_v44vkh6bzramngvcrficy2no44	260	7	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	260	8	zone	zone	NN
work_v44vkh6bzramngvcrficy2no44	260	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	260	10	width	width	NN
work_v44vkh6bzramngvcrficy2no44	260	11	$	$	$
work_v44vkh6bzramngvcrficy2no44	260	12	=	=	SYM
work_v44vkh6bzramngvcrficy2no44	260	13	5	5	CD
work_v44vkh6bzramngvcrficy2no44	260	14	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	260	15	set	set	VBN
work_v44vkh6bzramngvcrficy2no44	260	16	around	around	IN
work_v44vkh6bzramngvcrficy2no44	260	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	260	18	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	260	19	covered	covered	JJ
work_v44vkh6bzramngvcrficy2no44	260	20	position	position	NN
work_v44vkh6bzramngvcrficy2no44	260	21	.	.	.
work_v44vkh6bzramngvcrficy2no44	261	1	To	to	TO
work_v44vkh6bzramngvcrficy2no44	261	2	set	set	VB
work_v44vkh6bzramngvcrficy2no44	261	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	261	4	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	261	5	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	261	6	pa-	pa-	XX
work_v44vkh6bzramngvcrficy2no44	261	7	rameters	rameter	NNS
work_v44vkh6bzramngvcrficy2no44	261	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	9	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	261	10	perform	perform	VBP
work_v44vkh6bzramngvcrficy2no44	261	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	261	12	grid	grid	NN
work_v44vkh6bzramngvcrficy2no44	261	13	search	search	NN
work_v44vkh6bzramngvcrficy2no44	261	14	over	over	IN
work_v44vkh6bzramngvcrficy2no44	261	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	261	16	values	value	NNS
work_v44vkh6bzramngvcrficy2no44	261	17	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	261	18	1	1	CD
work_v44vkh6bzramngvcrficy2no44	261	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	20	2	2	CD
work_v44vkh6bzramngvcrficy2no44	261	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	22	3	3	CD
work_v44vkh6bzramngvcrficy2no44	261	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	24	4	4	CD
work_v44vkh6bzramngvcrficy2no44	261	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	26	5	5	CD
work_v44vkh6bzramngvcrficy2no44	261	27	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	261	28	for	for	IN
work_v44vkh6bzramngvcrficy2no44	261	29	histogram	histogram	NNP
work_v44vkh6bzramngvcrficy2no44	261	30	and	and	CC
work_v44vkh6bzramngvcrficy2no44	261	31	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	261	32	0.5	0.5	CD
work_v44vkh6bzramngvcrficy2no44	261	33	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	34	0.25	0.25	CD
work_v44vkh6bzramngvcrficy2no44	261	35	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	36	0.1	0.1	CD
work_v44vkh6bzramngvcrficy2no44	261	37	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	261	38	for	for	IN
work_v44vkh6bzramngvcrficy2no44	261	39	relative	relative	JJ
work_v44vkh6bzramngvcrficy2no44	261	40	threshold	threshold	NN
work_v44vkh6bzramngvcrficy2no44	261	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	42	and	and	CC
work_v44vkh6bzramngvcrficy2no44	261	43	select	select	VB
work_v44vkh6bzramngvcrficy2no44	261	44	the	the	DT
work_v44vkh6bzramngvcrficy2no44	261	45	values	value	NNS
work_v44vkh6bzramngvcrficy2no44	261	46	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	261	47	achieve	achieve	VBP
work_v44vkh6bzramngvcrficy2no44	261	48	the	the	DT
work_v44vkh6bzramngvcrficy2no44	261	49	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	261	50	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	261	51	and	and	CC
work_v44vkh6bzramngvcrficy2no44	261	52	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	261	53	on	on	IN
work_v44vkh6bzramngvcrficy2no44	261	54	eval08-nw	eval08-nw	NNP
work_v44vkh6bzramngvcrficy2no44	261	55	,	,	,
work_v44vkh6bzramngvcrficy2no44	261	56	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	261	57	3	3	CD
work_v44vkh6bzramngvcrficy2no44	261	58	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	261	59	his-	his-	RB
work_v44vkh6bzramngvcrficy2no44	261	60	togram	togram	NN
work_v44vkh6bzramngvcrficy2no44	261	61	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	261	62	and	and	CC
work_v44vkh6bzramngvcrficy2no44	261	63	0.1	0.1	CD
work_v44vkh6bzramngvcrficy2no44	261	64	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	261	65	threshold	threshold	NN
work_v44vkh6bzramngvcrficy2no44	261	66	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	261	67	.	.	.
work_v44vkh6bzramngvcrficy2no44	262	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	262	2	resulting	result	VBG
work_v44vkh6bzramngvcrficy2no44	262	3	configu-	configu-	NN
work_v44vkh6bzramngvcrficy2no44	262	4	ration	ration	NN
work_v44vkh6bzramngvcrficy2no44	262	5	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	262	6	then	then	RB
work_v44vkh6bzramngvcrficy2no44	262	7	re	re	VBG
work_v44vkh6bzramngvcrficy2no44	262	8	-	-	VBN
work_v44vkh6bzramngvcrficy2no44	262	9	optimized	optimize	VBN
work_v44vkh6bzramngvcrficy2no44	262	10	by	by	IN
work_v44vkh6bzramngvcrficy2no44	262	11	MERT	MERT	NNP
work_v44vkh6bzramngvcrficy2no44	262	12	on	on	IN
work_v44vkh6bzramngvcrficy2no44	262	13	dev06-nw	dev06-nw	NNP
work_v44vkh6bzramngvcrficy2no44	262	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	263	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	263	2	setting	setting	NN
work_v44vkh6bzramngvcrficy2no44	263	3	implies	imply	VBZ
work_v44vkh6bzramngvcrficy2no44	263	4	that	that	IN
work_v44vkh6bzramngvcrficy2no44	263	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	263	6	at	at	IN
work_v44vkh6bzramngvcrficy2no44	263	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	263	8	given	give	VBN
work_v44vkh6bzramngvcrficy2no44	263	9	point	point	NN
work_v44vkh6bzramngvcrficy2no44	263	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	263	11	decod-	decod-	RB
work_v44vkh6bzramngvcrficy2no44	263	12	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	263	13	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	263	14	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	263	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	263	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	263	17	last	last	JJ
work_v44vkh6bzramngvcrficy2no44	263	18	covered	covered	JJ
work_v44vkh6bzramngvcrficy2no44	263	19	position	position	NN
work_v44vkh6bzramngvcrficy2no44	263	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	263	21	a	a	DT
work_v44vkh6bzramngvcrficy2no44	263	22	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	263	23	word	word	NN
work_v44vkh6bzramngvcrficy2no44	263	24	can	can	MD
work_v44vkh6bzramngvcrficy2no44	263	25	be	be	VB
work_v44vkh6bzramngvcrficy2no44	263	26	translated	translate	VBN
work_v44vkh6bzramngvcrficy2no44	263	27	only	only	RB
work_v44vkh6bzramngvcrficy2no44	263	28	if	if	IN
work_v44vkh6bzramngvcrficy2no44	263	29	:	:	:
work_v44vkh6bzramngvcrficy2no44	263	30	•	•	VB
work_v44vkh6bzramngvcrficy2no44	263	31	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	263	32	lies	lie	VBZ
work_v44vkh6bzramngvcrficy2no44	263	33	within	within	IN
work_v44vkh6bzramngvcrficy2no44	263	34	a	a	DT
work_v44vkh6bzramngvcrficy2no44	263	35	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	263	36	of	of	IN
work_v44vkh6bzramngvcrficy2no44	263	37	5	5	CD
work_v44vkh6bzramngvcrficy2no44	263	38	from	from	IN
work_v44vkh6bzramngvcrficy2no44	263	39	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	263	40	,	,	,
work_v44vkh6bzramngvcrficy2no44	263	41	or	or	CC
work_v44vkh6bzramngvcrficy2no44	263	42	•	•	VB
work_v44vkh6bzramngvcrficy2no44	263	43	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	263	44	lies	lie	VBZ
work_v44vkh6bzramngvcrficy2no44	263	45	within	within	IN
work_v44vkh6bzramngvcrficy2no44	263	46	a	a	DT
work_v44vkh6bzramngvcrficy2no44	263	47	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	263	48	of	of	IN
work_v44vkh6bzramngvcrficy2no44	263	49	18	18	CD
work_v44vkh6bzramngvcrficy2no44	263	50	from	from	IN
work_v44vkh6bzramngvcrficy2no44	263	51	i	i	PRP
work_v44vkh6bzramngvcrficy2no44	263	52	and	and	CC
work_v44vkh6bzramngvcrficy2no44	263	53	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	263	54	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	263	55	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	263	56	score	score	NN
work_v44vkh6bzramngvcrficy2no44	263	57	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	263	58	among	among	IN
work_v44vkh6bzramngvcrficy2no44	263	59	the	the	DT
work_v44vkh6bzramngvcrficy2no44	263	60	top	top	JJ
work_v44vkh6bzramngvcrficy2no44	263	61	3	3	CD
work_v44vkh6bzramngvcrficy2no44	263	62	and	and	CC
work_v44vkh6bzramngvcrficy2no44	263	63	at	at	IN
work_v44vkh6bzramngvcrficy2no44	263	64	least	least	JJS
work_v44vkh6bzramngvcrficy2no44	263	65	equal	equal	JJ
work_v44vkh6bzramngvcrficy2no44	263	66	to	to	IN
work_v44vkh6bzramngvcrficy2no44	263	67	1/10	1/10	CD
work_v44vkh6bzramngvcrficy2no44	263	68	of	of	IN
work_v44vkh6bzramngvcrficy2no44	263	69	the	the	DT
work_v44vkh6bzramngvcrficy2no44	263	70	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	263	71	score	score	NN
work_v44vkh6bzramngvcrficy2no44	263	72	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	263	73	in	in	IN
work_v44vkh6bzramngvcrficy2no44	263	74	linear	linear	JJ
work_v44vkh6bzramngvcrficy2no44	263	75	space	space	NN
work_v44vkh6bzramngvcrficy2no44	263	76	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	263	77	.	.	.
work_v44vkh6bzramngvcrficy2no44	264	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	264	2	shown	show	VBN
work_v44vkh6bzramngvcrficy2no44	264	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	264	4	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	264	5	5	5	CD
work_v44vkh6bzramngvcrficy2no44	264	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	264	7	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	264	8	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	264	9	achieves	achieve	VBZ
work_v44vkh6bzramngvcrficy2no44	264	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	264	11	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	264	12	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	264	13	overall	overall	RB
work_v44vkh6bzramngvcrficy2no44	264	14	:	:	:
work_v44vkh6bzramngvcrficy2no44	264	15	despite	despite	IN
work_v44vkh6bzramngvcrficy2no44	264	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	264	17	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	264	18	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	264	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	264	20	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	264	21	report	report	VBP
work_v44vkh6bzramngvcrficy2no44	264	22	335	335	CD
work_v44vkh6bzramngvcrficy2no44	264	23	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	264	24	Reo.models	Reo.models	NNPS
work_v44vkh6bzramngvcrficy2no44	264	25	eval08-nw	eval08-nw	NNP
work_v44vkh6bzramngvcrficy2no44	264	26	eval09-nw	eval09-nw	:
work_v44vkh6bzramngvcrficy2no44	264	27	vs-09	vs-09	NNP
work_v44vkh6bzramngvcrficy2no44	264	28	ms/	ms/	NNP
work_v44vkh6bzramngvcrficy2no44	264	29	bleu	bleu	NNP
work_v44vkh6bzramngvcrficy2no44	264	30	met	meet	VBD
work_v44vkh6bzramngvcrficy2no44	264	31	krs	krs	NNP
work_v44vkh6bzramngvcrficy2no44	264	32	krs	krs	NNP
work_v44vkh6bzramngvcrficy2no44	264	33	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	264	34	V	V	NNP
work_v44vkh6bzramngvcrficy2no44	264	35	bleu	bleu	NN
work_v44vkh6bzramngvcrficy2no44	264	36	met	meet	VBD
work_v44vkh6bzramngvcrficy2no44	264	37	krs	krs	NNP
work_v44vkh6bzramngvcrficy2no44	264	38	krs	krs	NNP
work_v44vkh6bzramngvcrficy2no44	264	39	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	264	40	V	V	NNP
work_v44vkh6bzramngvcrficy2no44	264	41	krs	krs	JJ
work_v44vkh6bzramngvcrficy2no44	264	42	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	264	43	V	v	NN
work_v44vkh6bzramngvcrficy2no44	264	44	word	word	NN
work_v44vkh6bzramngvcrficy2no44	264	45	Using	use	VBG
work_v44vkh6bzramngvcrficy2no44	264	46	the	the	DT
work_v44vkh6bzramngvcrficy2no44	264	47	medium	medium	NN
work_v44vkh6bzramngvcrficy2no44	264	48	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	264	49	size	size	NN
work_v44vkh6bzramngvcrficy2no44	264	50	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	264	51	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	264	52	147	147	CD
work_v44vkh6bzramngvcrficy2no44	264	53	M	M	NNP
work_v44vkh6bzramngvcrficy2no44	264	54	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	264	55	tokens	token	NNS
work_v44vkh6bzramngvcrficy2no44	264	56	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	264	57	:	:	:
work_v44vkh6bzramngvcrficy2no44	264	58	5	5	CD
work_v44vkh6bzramngvcrficy2no44	264	59	hier.lexreo	hier.lexreo	NNS
work_v44vkh6bzramngvcrficy2no44	264	60	,	,	,
work_v44vkh6bzramngvcrficy2no44	264	61	early	early	RB
work_v44vkh6bzramngvcrficy2no44	264	62	disto	disto	NN
work_v44vkh6bzramngvcrficy2no44	264	63	44.7	44.7	CD
work_v44vkh6bzramngvcrficy2no44	264	64	35.1	35.1	CD
work_v44vkh6bzramngvcrficy2no44	264	65	!	!	.
work_v44vkh6bzramngvcrficy2no44	265	1	83.0	83.0	CD
work_v44vkh6bzramngvcrficy2no44	265	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	266	1	84.7	84.7	CD
work_v44vkh6bzramngvcrficy2no44	266	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	267	1	50.3	50.3	CD
work_v44vkh6bzramngvcrficy2no44	267	2	"	"	``
work_v44vkh6bzramngvcrficy2no44	267	3	38.1	38.1	CD
work_v44vkh6bzramngvcrficy2no44	267	4	84.6	84.6	CD
work_v44vkh6bzramngvcrficy2no44	267	5	85.9	85.9	CD
work_v44vkh6bzramngvcrficy2no44	267	6	84.7	84.7	CD
work_v44vkh6bzramngvcrficy2no44	267	7	59	59	CD
work_v44vkh6bzramngvcrficy2no44	267	8	+	+	SYM
work_v44vkh6bzramngvcrficy2no44	267	9	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	267	10	model	model	NN
work_v44vkh6bzramngvcrficy2no44	267	11	44.8	44.8	CD
work_v44vkh6bzramngvcrficy2no44	267	12	35.1	35.1	CD
work_v44vkh6bzramngvcrficy2no44	267	13	83.7	83.7	CD
work_v44vkh6bzramngvcrficy2no44	267	14	85.4	85.4	CD
work_v44vkh6bzramngvcrficy2no44	267	15	51.0	51.0	CD
work_v44vkh6bzramngvcrficy2no44	267	16	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	17	38.3	38.3	CD
work_v44vkh6bzramngvcrficy2no44	267	18	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	19	85.1	85.1	CD
work_v44vkh6bzramngvcrficy2no44	267	20	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	21	86.6	86.6	CD
work_v44vkh6bzramngvcrficy2no44	267	22	$	$	$
work_v44vkh6bzramngvcrficy2no44	267	23	85.5	85.5	CD
work_v44vkh6bzramngvcrficy2no44	267	24	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	25	64	64	CD
work_v44vkh6bzramngvcrficy2no44	267	26	8	8	CD
work_v44vkh6bzramngvcrficy2no44	267	27	hier.lexreo	hier.lexreo	NNS
work_v44vkh6bzramngvcrficy2no44	267	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	267	29	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	267	30	disto[B	disto[b	CD
work_v44vkh6bzramngvcrficy2no44	267	31	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	267	32	44.8	44.8	CD
work_v44vkh6bzramngvcrficy2no44	267	33	35.2	35.2	CD
work_v44vkh6bzramngvcrficy2no44	267	34	83.4	83.4	CD
work_v44vkh6bzramngvcrficy2no44	267	35	85.6	85.6	CD
work_v44vkh6bzramngvcrficy2no44	267	36	50.6	50.6	CD
work_v44vkh6bzramngvcrficy2no44	267	37	38.1	38.1	CD
work_v44vkh6bzramngvcrficy2no44	267	38	84.7	84.7	CD
work_v44vkh6bzramngvcrficy2no44	267	39	86.2	86.2	CD
work_v44vkh6bzramngvcrficy2no44	267	40	84.8	84.8	CD
work_v44vkh6bzramngvcrficy2no44	267	41	87	87	CD
work_v44vkh6bzramngvcrficy2no44	267	42	+	+	CC
work_v44vkh6bzramngvcrficy2no44	267	43	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	267	44	model	model	NN
work_v44vkh6bzramngvcrficy2no44	267	45	45.0	45.0	CD
work_v44vkh6bzramngvcrficy2no44	267	46	35.2	35.2	CD
work_v44vkh6bzramngvcrficy2no44	267	47	83.7	83.7	CD
work_v44vkh6bzramngvcrficy2no44	267	48	$	$	$
work_v44vkh6bzramngvcrficy2no44	267	49	85.9	85.9	CD
work_v44vkh6bzramngvcrficy2no44	267	50	51.1	51.1	CD
work_v44vkh6bzramngvcrficy2no44	267	51	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	52	38.3	38.3	CD
work_v44vkh6bzramngvcrficy2no44	267	53	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	54	85.1	85.1	CD
work_v44vkh6bzramngvcrficy2no44	267	55	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	56	86.8	86.8	CD
work_v44vkh6bzramngvcrficy2no44	267	57	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	58	85.8	85.8	CD
work_v44vkh6bzramngvcrficy2no44	267	59	#	#	$
work_v44vkh6bzramngvcrficy2no44	267	60	94	94	CD
work_v44vkh6bzramngvcrficy2no44	267	61	18	18	CD
work_v44vkh6bzramngvcrficy2no44	267	62	hier.lexreo	hier.lexreo	NNS
work_v44vkh6bzramngvcrficy2no44	267	63	,	,	,
work_v44vkh6bzramngvcrficy2no44	267	64	early	early	RB
work_v44vkh6bzramngvcrficy2no44	267	65	disto	disto	NN
work_v44vkh6bzramngvcrficy2no44	267	66	44.7	44.7	CD
work_v44vkh6bzramngvcrficy2no44	267	67	34.9	34.9	CD
work_v44vkh6bzramngvcrficy2no44	267	68	!	!	.
work_v44vkh6bzramngvcrficy2no44	268	1	82.4	82.4	CD
work_v44vkh6bzramngvcrficy2no44	268	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	269	1	84.9	84.9	CD
work_v44vkh6bzramngvcrficy2no44	269	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	270	1	50.3	50.3	CD
work_v44vkh6bzramngvcrficy2no44	270	2	38.0	38.0	CD
work_v44vkh6bzramngvcrficy2no44	270	3	"	"	``
work_v44vkh6bzramngvcrficy2no44	270	4	83.9	83.9	CD
work_v44vkh6bzramngvcrficy2no44	270	5	!	!	.
work_v44vkh6bzramngvcrficy2no44	271	1	85.8	85.8	CD
work_v44vkh6bzramngvcrficy2no44	271	2	"	"	``
work_v44vkh6bzramngvcrficy2no44	271	3	84.3	84.3	CD
work_v44vkh6bzramngvcrficy2no44	271	4	"	"	``
work_v44vkh6bzramngvcrficy2no44	271	5	164	164	CD
work_v44vkh6bzramngvcrficy2no44	271	6	+	+	SYM
work_v44vkh6bzramngvcrficy2no44	271	7	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	271	8	model	model	NN
work_v44vkh6bzramngvcrficy2no44	271	9	44.8	44.8	CD
work_v44vkh6bzramngvcrficy2no44	271	10	35.2	35.2	CD
work_v44vkh6bzramngvcrficy2no44	271	11	82.7	82.7	CD
work_v44vkh6bzramngvcrficy2no44	271	12	!	!	.
work_v44vkh6bzramngvcrficy2no44	272	1	85.5	85.5	CD
work_v44vkh6bzramngvcrficy2no44	272	2	51.0	51.0	CD
work_v44vkh6bzramngvcrficy2no44	272	3	$	$	$
work_v44vkh6bzramngvcrficy2no44	272	4	38.3	38.3	CD
work_v44vkh6bzramngvcrficy2no44	272	5	#	#	$
work_v44vkh6bzramngvcrficy2no44	272	6	84.2	84.2	CD
work_v44vkh6bzramngvcrficy2no44	272	7	"	"	``
work_v44vkh6bzramngvcrficy2no44	272	8	86.2	86.2	CD
work_v44vkh6bzramngvcrficy2no44	272	9	85.2	85.2	CD
work_v44vkh6bzramngvcrficy2no44	272	10	178	178	CD
work_v44vkh6bzramngvcrficy2no44	272	11	+	+	CC
work_v44vkh6bzramngvcrficy2no44	272	12	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	272	13	reo.pruning($=5	reo.pruning($=5	NN
work_v44vkh6bzramngvcrficy2no44	272	14	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	272	15	45.0	45.0	CD
work_v44vkh6bzramngvcrficy2no44	272	16	35.3	35.3	CD
work_v44vkh6bzramngvcrficy2no44	272	17	83.7	83.7	CD
work_v44vkh6bzramngvcrficy2no44	272	18	$	$	$
work_v44vkh6bzramngvcrficy2no44	272	19	86.3	86.3	CD
work_v44vkh6bzramngvcrficy2no44	272	20	#	#	$
work_v44vkh6bzramngvcrficy2no44	272	21	50.9	50.9	CD
work_v44vkh6bzramngvcrficy2no44	272	22	38.3	38.3	CD
work_v44vkh6bzramngvcrficy2no44	272	23	#	#	$
work_v44vkh6bzramngvcrficy2no44	272	24	84.9	84.9	CD
work_v44vkh6bzramngvcrficy2no44	272	25	87.0	87.0	CD
work_v44vkh6bzramngvcrficy2no44	272	26	#	#	$
work_v44vkh6bzramngvcrficy2no44	272	27	86.2	86.2	CD
work_v44vkh6bzramngvcrficy2no44	272	28	#	#	$
work_v44vkh6bzramngvcrficy2no44	272	29	68	68	CD
work_v44vkh6bzramngvcrficy2no44	272	30	Using	use	VBG
work_v44vkh6bzramngvcrficy2no44	272	31	the	the	DT
work_v44vkh6bzramngvcrficy2no44	272	32	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	272	33	interpolated	interpolate	VBN
work_v44vkh6bzramngvcrficy2no44	272	34	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	272	35	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	272	36	2130	2130	CD
work_v44vkh6bzramngvcrficy2no44	272	37	M	M	NNP
work_v44vkh6bzramngvcrficy2no44	272	38	English	English	NNP
work_v44vkh6bzramngvcrficy2no44	272	39	tokens	token	NNS
work_v44vkh6bzramngvcrficy2no44	272	40	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	272	41	and	and	CC
work_v44vkh6bzramngvcrficy2no44	272	42	double	double	JJ
work_v44vkh6bzramngvcrficy2no44	272	43	beam	beam	NN
work_v44vkh6bzramngvcrficy2no44	272	44	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	272	45	size	size	NN
work_v44vkh6bzramngvcrficy2no44	272	46	:	:	:
work_v44vkh6bzramngvcrficy2no44	272	47	8	8	CD
work_v44vkh6bzramngvcrficy2no44	272	48	hier.lexreo	hier.lexreo	NNS
work_v44vkh6bzramngvcrficy2no44	272	49	,	,	,
work_v44vkh6bzramngvcrficy2no44	272	50	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	272	51	disto[B	disto[b	CD
work_v44vkh6bzramngvcrficy2no44	272	52	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	272	53	46.3	46.3	CD
work_v44vkh6bzramngvcrficy2no44	272	54	35.0	35.0	CD
work_v44vkh6bzramngvcrficy2no44	272	55	83.2	83.2	CD
work_v44vkh6bzramngvcrficy2no44	272	56	85.0	85.0	CD
work_v44vkh6bzramngvcrficy2no44	272	57	51.6	51.6	CD
work_v44vkh6bzramngvcrficy2no44	272	58	38.3	38.3	CD
work_v44vkh6bzramngvcrficy2no44	272	59	84.5	84.5	CD
work_v44vkh6bzramngvcrficy2no44	272	60	85.8	85.8	CD
work_v44vkh6bzramngvcrficy2no44	272	61	84.5	84.5	CD
work_v44vkh6bzramngvcrficy2no44	272	62	2579	2579	CD
work_v44vkh6bzramngvcrficy2no44	272	63	18	18	CD
work_v44vkh6bzramngvcrficy2no44	272	64	hier.lexreo	hier.lexreo	NNS
work_v44vkh6bzramngvcrficy2no44	272	65	,	,	,
work_v44vkh6bzramngvcrficy2no44	272	66	early	early	RB
work_v44vkh6bzramngvcrficy2no44	272	67	disto	disto	NN
work_v44vkh6bzramngvcrficy2no44	272	68	45.9	45.9	CD
work_v44vkh6bzramngvcrficy2no44	272	69	"	"	``
work_v44vkh6bzramngvcrficy2no44	272	70	34.9	34.9	CD
work_v44vkh6bzramngvcrficy2no44	272	71	!	!	.
work_v44vkh6bzramngvcrficy2no44	273	1	81.7	81.7	CD
work_v44vkh6bzramngvcrficy2no44	273	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	274	1	84.1	84.1	CD
work_v44vkh6bzramngvcrficy2no44	274	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	275	1	51.4	51.4	CD
work_v44vkh6bzramngvcrficy2no44	275	2	38.1	38.1	CD
work_v44vkh6bzramngvcrficy2no44	275	3	!	!	.
work_v44vkh6bzramngvcrficy2no44	276	1	83.0	83.0	CD
work_v44vkh6bzramngvcrficy2no44	276	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	277	1	84.6	84.6	CD
work_v44vkh6bzramngvcrficy2no44	277	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	278	1	83.1	83.1	CD
work_v44vkh6bzramngvcrficy2no44	278	2	!	!	.
work_v44vkh6bzramngvcrficy2no44	279	1	5462	5462	CD
work_v44vkh6bzramngvcrficy2no44	279	2	+	+	SYM
work_v44vkh6bzramngvcrficy2no44	279	3	WaW+reo.pruning($=5	WaW+reo.pruning($=5	NNP
work_v44vkh6bzramngvcrficy2no44	279	4	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	279	5	46.3	46.3	CD
work_v44vkh6bzramngvcrficy2no44	279	6	35.2	35.2	CD
work_v44vkh6bzramngvcrficy2no44	279	7	83.4	83.4	CD
work_v44vkh6bzramngvcrficy2no44	279	8	85.7	85.7	CD
work_v44vkh6bzramngvcrficy2no44	279	9	#	#	$
work_v44vkh6bzramngvcrficy2no44	279	10	52.8	52.8	CD
work_v44vkh6bzramngvcrficy2no44	279	11	#	#	$
work_v44vkh6bzramngvcrficy2no44	279	12	38.6	38.6	CD
work_v44vkh6bzramngvcrficy2no44	279	13	#	#	$
work_v44vkh6bzramngvcrficy2no44	279	14	84.6	84.6	CD
work_v44vkh6bzramngvcrficy2no44	279	15	86.6	86.6	CD
work_v44vkh6bzramngvcrficy2no44	279	16	#	#	$
work_v44vkh6bzramngvcrficy2no44	279	17	85.5	85.5	CD
work_v44vkh6bzramngvcrficy2no44	279	18	#	#	$
work_v44vkh6bzramngvcrficy2no44	279	19	1588	1588	CD
work_v44vkh6bzramngvcrficy2no44	279	20	Table	table	NN
work_v44vkh6bzramngvcrficy2no44	279	21	5	5	CD
work_v44vkh6bzramngvcrficy2no44	279	22	:	:	:
work_v44vkh6bzramngvcrficy2no44	279	23	Effects	effect	NNS
work_v44vkh6bzramngvcrficy2no44	279	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	279	25	WaW	waw	NN
work_v44vkh6bzramngvcrficy2no44	279	26	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	279	27	modeling	modeling	NN
work_v44vkh6bzramngvcrficy2no44	279	28	and	and	CC
work_v44vkh6bzramngvcrficy2no44	279	29	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	279	30	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	279	31	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	279	32	on	on	IN
work_v44vkh6bzramngvcrficy2no44	279	33	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	279	34	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	279	35	,	,	,
work_v44vkh6bzramngvcrficy2no44	279	36	measured	measure	VBN
work_v44vkh6bzramngvcrficy2no44	279	37	with	with	IN
work_v44vkh6bzramngvcrficy2no44	279	38	%	%	NN
work_v44vkh6bzramngvcrficy2no44	279	39	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	279	40	,	,	,
work_v44vkh6bzramngvcrficy2no44	279	41	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	279	42	,	,	,
work_v44vkh6bzramngvcrficy2no44	279	43	and	and	CC
work_v44vkh6bzramngvcrficy2no44	279	44	Kendall	Kendall	NNP
work_v44vkh6bzramngvcrficy2no44	279	45	Reordering	Reordering	NNP
work_v44vkh6bzramngvcrficy2no44	279	46	Score	Score	NNP
work_v44vkh6bzramngvcrficy2no44	279	47	:	:	:
work_v44vkh6bzramngvcrficy2no44	279	48	regular	regular	JJ
work_v44vkh6bzramngvcrficy2no44	279	49	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	279	50	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	279	51	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	279	52	and	and	CC
work_v44vkh6bzramngvcrficy2no44	279	53	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	279	54	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	279	55	specific	specific	JJ
work_v44vkh6bzramngvcrficy2no44	279	56	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	279	57	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	279	58	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	279	59	V	V	NNP
work_v44vkh6bzramngvcrficy2no44	279	60	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	279	61	.	.	.
work_v44vkh6bzramngvcrficy2no44	280	1	Statistically	statistically	RB
work_v44vkh6bzramngvcrficy2no44	280	2	significant	significant	JJ
work_v44vkh6bzramngvcrficy2no44	280	3	differences	difference	NNS
work_v44vkh6bzramngvcrficy2no44	280	4	with	with	IN
work_v44vkh6bzramngvcrficy2no44	280	5	respect	respect	NN
work_v44vkh6bzramngvcrficy2no44	280	6	to	to	IN
work_v44vkh6bzramngvcrficy2no44	280	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	280	8	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	280	9	[	[	-LRB-
work_v44vkh6bzramngvcrficy2no44	280	10	B	b	NN
work_v44vkh6bzramngvcrficy2no44	280	11	]	]	-RRB-
work_v44vkh6bzramngvcrficy2no44	280	12	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	280	13	marked	mark	VBN
work_v44vkh6bzramngvcrficy2no44	280	14	with	with	IN
work_v44vkh6bzramngvcrficy2no44	280	15	#	#	$
work_v44vkh6bzramngvcrficy2no44	280	16	!	!	.
work_v44vkh6bzramngvcrficy2no44	281	1	at	at	IN
work_v44vkh6bzramngvcrficy2no44	281	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	281	3	p	p	NN
work_v44vkh6bzramngvcrficy2no44	281	4	"	"	``
work_v44vkh6bzramngvcrficy2no44	281	5	.05	.05	CD
work_v44vkh6bzramngvcrficy2no44	281	6	level	level	NN
work_v44vkh6bzramngvcrficy2no44	281	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	281	8	$	$	$
work_v44vkh6bzramngvcrficy2no44	281	9	"	"	''
work_v44vkh6bzramngvcrficy2no44	281	10	at	at	IN
work_v44vkh6bzramngvcrficy2no44	281	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	281	12	p	p	NN
work_v44vkh6bzramngvcrficy2no44	281	13	"	"	``
work_v44vkh6bzramngvcrficy2no44	281	14	.10	.10	CD
work_v44vkh6bzramngvcrficy2no44	281	15	level	level	NN
work_v44vkh6bzramngvcrficy2no44	281	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	282	1	Decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	282	2	time	time	NN
work_v44vkh6bzramngvcrficy2no44	282	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	282	4	measured	measure	VBN
work_v44vkh6bzramngvcrficy2no44	282	5	in	in	IN
work_v44vkh6bzramngvcrficy2no44	282	6	milliseconds	millisecond	NNS
work_v44vkh6bzramngvcrficy2no44	282	7	per	per	IN
work_v44vkh6bzramngvcrficy2no44	282	8	input	input	NN
work_v44vkh6bzramngvcrficy2no44	282	9	word	word	NN
work_v44vkh6bzramngvcrficy2no44	282	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	283	1	no	no	DT
work_v44vkh6bzramngvcrficy2no44	283	2	loss	loss	NN
work_v44vkh6bzramngvcrficy2no44	283	3	in	in	IN
work_v44vkh6bzramngvcrficy2no44	283	4	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	283	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	283	6	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	283	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	283	8	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	283	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	283	10	but	but	CC
work_v44vkh6bzramngvcrficy2no44	283	11	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	283	12	actually	actually	RB
work_v44vkh6bzramngvcrficy2no44	283	13	see	see	VBP
work_v44vkh6bzramngvcrficy2no44	283	14	several	several	JJ
work_v44vkh6bzramngvcrficy2no44	283	15	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	283	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	284	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	284	2	particular	particular	JJ
work_v44vkh6bzramngvcrficy2no44	284	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	284	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	284	5	gains	gain	NNS
work_v44vkh6bzramngvcrficy2no44	284	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	284	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	284	8	blind	blind	JJ
work_v44vkh6bzramngvcrficy2no44	284	9	test	test	NN
work_v44vkh6bzramngvcrficy2no44	284	10	eval09-nw	eval09-nw	NNP
work_v44vkh6bzramngvcrficy2no44	284	11	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	284	12	+0.3	+0.3	NNP
work_v44vkh6bzramngvcrficy2no44	284	13	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	284	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	284	15	+0.2	+0.2	NNP
work_v44vkh6bzramngvcrficy2no44	284	16	ME-	ME-	NNP
work_v44vkh6bzramngvcrficy2no44	284	17	TEOR	TEOR	NNP
work_v44vkh6bzramngvcrficy2no44	284	18	and	and	CC
work_v44vkh6bzramngvcrficy2no44	284	19	+0.2	+0.2	NNP
work_v44vkh6bzramngvcrficy2no44	284	20	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	284	21	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	284	22	only	only	RB
work_v44vkh6bzramngvcrficy2no44	284	23	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	284	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	284	25	significant	significant	JJ
work_v44vkh6bzramngvcrficy2no44	284	26	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	284	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	285	1	While	while	IN
work_v44vkh6bzramngvcrficy2no44	285	2	these	these	DT
work_v44vkh6bzramngvcrficy2no44	285	3	gains	gain	NNS
work_v44vkh6bzramngvcrficy2no44	285	4	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	285	5	admittedly	admittedly	RB
work_v44vkh6bzramngvcrficy2no44	285	6	small	small	JJ
work_v44vkh6bzramngvcrficy2no44	285	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	285	8	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	285	9	recall	recall	VBP
work_v44vkh6bzramngvcrficy2no44	285	10	that	that	IN
work_v44vkh6bzramngvcrficy2no44	285	11	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	285	12	techniques	technique	NNS
work_v44vkh6bzramngvcrficy2no44	285	13	affect	affect	VBP
work_v44vkh6bzramngvcrficy2no44	285	14	rare	rare	JJ
work_v44vkh6bzramngvcrficy2no44	285	15	and	and	CC
work_v44vkh6bzramngvcrficy2no44	285	16	isolated	isolated	JJ
work_v44vkh6bzramngvcrficy2no44	285	17	events	event	NNS
work_v44vkh6bzramngvcrficy2no44	285	18	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	285	19	can	can	MD
work_v44vkh6bzramngvcrficy2no44	285	20	hardly	hardly	RB
work_v44vkh6bzramngvcrficy2no44	285	21	emerge	emerge	VB
work_v44vkh6bzramngvcrficy2no44	285	22	from	from	IN
work_v44vkh6bzramngvcrficy2no44	285	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	285	24	general	general	JJ
work_v44vkh6bzramngvcrficy2no44	285	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	285	26	purpose	purpose	NN
work_v44vkh6bzramngvcrficy2no44	285	27	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	285	28	metrics	metric	NNS
work_v44vkh6bzramngvcrficy2no44	285	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	286	1	Moreover	moreover	RB
work_v44vkh6bzramngvcrficy2no44	286	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	286	3	to	to	IN
work_v44vkh6bzramngvcrficy2no44	286	4	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	286	5	knowledge	knowledge	NN
work_v44vkh6bzramngvcrficy2no44	286	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	286	7	this	this	DT
work_v44vkh6bzramngvcrficy2no44	286	8	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	286	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	286	10	first	first	JJ
work_v44vkh6bzramngvcrficy2no44	286	11	time	time	NN
work_v44vkh6bzramngvcrficy2no44	286	12	that	that	IN
work_v44vkh6bzramngvcrficy2no44	286	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	286	14	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	286	15	system	system	NN
work_v44vkh6bzramngvcrficy2no44	286	16	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	286	17	shown	show	VBN
work_v44vkh6bzramngvcrficy2no44	286	18	to	to	TO
work_v44vkh6bzramngvcrficy2no44	286	19	maintain	maintain	VB
work_v44vkh6bzramngvcrficy2no44	286	20	a	a	DT
work_v44vkh6bzramngvcrficy2no44	286	21	good	good	JJ
work_v44vkh6bzramngvcrficy2no44	286	22	performance	performance	NN
work_v44vkh6bzramngvcrficy2no44	286	23	on	on	IN
work_v44vkh6bzramngvcrficy2no44	286	24	this	this	DT
work_v44vkh6bzramngvcrficy2no44	286	25	language	language	NN
work_v44vkh6bzramngvcrficy2no44	286	26	pair	pair	NN
work_v44vkh6bzramngvcrficy2no44	286	27	while	while	IN
work_v44vkh6bzramngvcrficy2no44	286	28	admitting	admit	VBG
work_v44vkh6bzramngvcrficy2no44	286	29	very	very	RB
work_v44vkh6bzramngvcrficy2no44	286	30	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	286	31	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	286	32	range	range	NN
work_v44vkh6bzramngvcrficy2no44	286	33	reorderings	reordering	NNS
work_v44vkh6bzramngvcrficy2no44	286	34	.	.	.
work_v44vkh6bzramngvcrficy2no44	287	1	Finally	finally	RB
work_v44vkh6bzramngvcrficy2no44	287	2	and	and	CC
work_v44vkh6bzramngvcrficy2no44	287	3	more	more	RBR
work_v44vkh6bzramngvcrficy2no44	287	4	importantly	importantly	RB
work_v44vkh6bzramngvcrficy2no44	287	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	287	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	287	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	287	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	287	9	verbs	verbs	NNP
work_v44vkh6bzramngvcrficy2no44	287	10	improves	improve	VBZ
work_v44vkh6bzramngvcrficy2no44	287	11	significantly	significantly	RB
work_v44vkh6bzramngvcrficy2no44	287	12	on	on	IN
work_v44vkh6bzramngvcrficy2no44	287	13	both	both	DT
work_v44vkh6bzramngvcrficy2no44	287	14	generic	generic	JJ
work_v44vkh6bzramngvcrficy2no44	287	15	tests	test	NNS
work_v44vkh6bzramngvcrficy2no44	287	16	and	and	CC
work_v44vkh6bzramngvcrficy2no44	287	17	on	on	IN
work_v44vkh6bzramngvcrficy2no44	287	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	287	19	VS-	vs-	JJ
work_v44vkh6bzramngvcrficy2no44	287	20	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	287	21	subset	subset	NN
work_v44vkh6bzramngvcrficy2no44	287	22	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	287	23	vs-09	vs-09	NNP
work_v44vkh6bzramngvcrficy2no44	287	24	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	287	25	:	:	:
work_v44vkh6bzramngvcrficy2no44	287	26	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	287	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	287	28	in	in	IN
work_v44vkh6bzramngvcrficy2no44	287	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	287	30	latter	latter	JJ
work_v44vkh6bzramngvcrficy2no44	287	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	287	32	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	287	33	achieve	achieve	VBP
work_v44vkh6bzramngvcrficy2no44	287	34	a	a	DT
work_v44vkh6bzramngvcrficy2no44	287	35	notable	notable	JJ
work_v44vkh6bzramngvcrficy2no44	287	36	gain	gain	NN
work_v44vkh6bzramngvcrficy2no44	287	37	of	of	IN
work_v44vkh6bzramngvcrficy2no44	287	38	1.4	1.4	CD
work_v44vkh6bzramngvcrficy2no44	287	39	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	287	40	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	287	41	V.	V.	NNP
work_v44vkh6bzramngvcrficy2no44	288	1	Efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	288	2	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	288	3	also	also	RB
work_v44vkh6bzramngvcrficy2no44	288	4	largely	largely	RB
work_v44vkh6bzramngvcrficy2no44	288	5	improved	improve	VBN
work_v44vkh6bzramngvcrficy2no44	288	6	by	by	IN
work_v44vkh6bzramngvcrficy2no44	288	7	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	288	8	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	288	9	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	288	10	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	288	11	technique	technique	NN
work_v44vkh6bzramngvcrficy2no44	288	12	:	:	:
work_v44vkh6bzramngvcrficy2no44	288	13	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	288	14	time	time	NN
work_v44vkh6bzramngvcrficy2no44	288	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	288	16	re-	re-	RB
work_v44vkh6bzramngvcrficy2no44	288	17	duced	duce	VBN
work_v44vkh6bzramngvcrficy2no44	288	18	to	to	IN
work_v44vkh6bzramngvcrficy2no44	288	19	68	68	CD
work_v44vkh6bzramngvcrficy2no44	288	20	ms	m	NNS
work_v44vkh6bzramngvcrficy2no44	288	21	/	/	SYM
work_v44vkh6bzramngvcrficy2no44	288	22	word	word	NN
work_v44vkh6bzramngvcrficy2no44	288	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	288	24	corresponding	correspond	VBG
work_v44vkh6bzramngvcrficy2no44	288	25	to	to	IN
work_v44vkh6bzramngvcrficy2no44	288	26	a	a	DT
work_v44vkh6bzramngvcrficy2no44	288	27	22	22	CD
work_v44vkh6bzramngvcrficy2no44	288	28	%	%	NN
work_v44vkh6bzramngvcrficy2no44	288	29	speed	speed	NN
work_v44vkh6bzramngvcrficy2no44	288	30	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	288	31	up	up	NN
work_v44vkh6bzramngvcrficy2no44	288	32	over	over	IN
work_v44vkh6bzramngvcrficy2no44	288	33	the	the	DT
work_v44vkh6bzramngvcrficy2no44	288	34	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	288	35	.	.	.
work_v44vkh6bzramngvcrficy2no44	289	1	Large	large	JJ
work_v44vkh6bzramngvcrficy2no44	289	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	289	3	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	289	4	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	289	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	290	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	290	2	also	also	RB
work_v44vkh6bzramngvcrficy2no44	290	3	investigate	investigate	VBP
work_v44vkh6bzramngvcrficy2no44	290	4	whether	whether	IN
work_v44vkh6bzramngvcrficy2no44	290	5	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	290	6	methods	method	NNS
work_v44vkh6bzramngvcrficy2no44	290	7	can	can	MD
work_v44vkh6bzramngvcrficy2no44	290	8	be	be	VB
work_v44vkh6bzramngvcrficy2no44	290	9	useful	useful	JJ
work_v44vkh6bzramngvcrficy2no44	290	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	290	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	290	12	scenario	scenario	NN
work_v44vkh6bzramngvcrficy2no44	290	13	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	290	14	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	290	15	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	290	16	less	less	RBR
work_v44vkh6bzramngvcrficy2no44	290	17	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	290	18	and	and	CC
work_v44vkh6bzramngvcrficy2no44	290	19	more	more	JJR
work_v44vkh6bzramngvcrficy2no44	290	20	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	290	21	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	290	22	available	available	JJ
work_v44vkh6bzramngvcrficy2no44	290	23	for	for	IN
work_v44vkh6bzramngvcrficy2no44	290	24	training	training	NN
work_v44vkh6bzramngvcrficy2no44	290	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	291	1	To	to	IN
work_v44vkh6bzramngvcrficy2no44	291	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	291	3	end	end	NN
work_v44vkh6bzramngvcrficy2no44	291	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	291	5	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	291	6	build	build	VBP
work_v44vkh6bzramngvcrficy2no44	291	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	291	8	very	very	RB
work_v44vkh6bzramngvcrficy2no44	291	9	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	291	10	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	291	11	by	by	IN
work_v44vkh6bzramngvcrficy2no44	291	12	interpolating	interpolate	VBG
work_v44vkh6bzramngvcrficy2no44	291	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	291	14	main	main	JJ
work_v44vkh6bzramngvcrficy2no44	291	15	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	291	16	with	with	IN
work_v44vkh6bzramngvcrficy2no44	291	17	three	three	CD
work_v44vkh6bzramngvcrficy2no44	291	18	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	291	19	LMs	lm	NNS
work_v44vkh6bzramngvcrficy2no44	291	20	trained	train	VBN
work_v44vkh6bzramngvcrficy2no44	291	21	on	on	IN
work_v44vkh6bzramngvcrficy2no44	291	22	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	291	23	Gigaword	Gigaword	NNP
work_v44vkh6bzramngvcrficy2no44	291	24	sec-	sec-	NN
work_v44vkh6bzramngvcrficy2no44	291	25	tions	tion	NNS
work_v44vkh6bzramngvcrficy2no44	291	26	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	291	27	see	see	VB
work_v44vkh6bzramngvcrficy2no44	291	28	Sect	Sect	NNP
work_v44vkh6bzramngvcrficy2no44	291	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	292	1	5	5	LS
work_v44vkh6bzramngvcrficy2no44	292	2	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	292	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	293	1	Moreover	moreover	RB
work_v44vkh6bzramngvcrficy2no44	293	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	293	3	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	293	4	relax	relax	VBP
work_v44vkh6bzramngvcrficy2no44	293	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	293	6	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	293	7	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	293	8	beam	beam	NN
work_v44vkh6bzramngvcrficy2no44	293	9	size	size	NN
work_v44vkh6bzramngvcrficy2no44	293	10	from	from	IN
work_v44vkh6bzramngvcrficy2no44	293	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	293	12	default	default	NN
work_v44vkh6bzramngvcrficy2no44	293	13	value	value	NN
work_v44vkh6bzramngvcrficy2no44	293	14	of	of	IN
work_v44vkh6bzramngvcrficy2no44	293	15	200	200	CD
work_v44vkh6bzramngvcrficy2no44	293	16	to	to	TO
work_v44vkh6bzramngvcrficy2no44	293	17	400	400	CD
work_v44vkh6bzramngvcrficy2no44	293	18	hy-	hy-	JJ
work_v44vkh6bzramngvcrficy2no44	293	19	potheses	pothese	NNS
work_v44vkh6bzramngvcrficy2no44	293	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	293	21	to	to	TO
work_v44vkh6bzramngvcrficy2no44	293	22	reduce	reduce	VB
work_v44vkh6bzramngvcrficy2no44	293	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	293	24	risk	risk	NN
work_v44vkh6bzramngvcrficy2no44	293	25	of	of	IN
work_v44vkh6bzramngvcrficy2no44	293	26	search	search	NN
work_v44vkh6bzramngvcrficy2no44	293	27	errors	error	NNS
work_v44vkh6bzramngvcrficy2no44	293	28	and	and	CC
work_v44vkh6bzramngvcrficy2no44	293	29	ob-	ob-	NN
work_v44vkh6bzramngvcrficy2no44	293	30	tain	tain	NN
work_v44vkh6bzramngvcrficy2no44	293	31	the	the	DT
work_v44vkh6bzramngvcrficy2no44	293	32	best	good	JJS
work_v44vkh6bzramngvcrficy2no44	293	33	possible	possible	JJ
work_v44vkh6bzramngvcrficy2no44	293	34	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	293	35	performance	performance	NN
work_v44vkh6bzramngvcrficy2no44	293	36	.	.	.
work_v44vkh6bzramngvcrficy2no44	294	1	By	by	IN
work_v44vkh6bzramngvcrficy2no44	294	2	comparing	compare	VBG
work_v44vkh6bzramngvcrficy2no44	294	3	the	the	DT
work_v44vkh6bzramngvcrficy2no44	294	4	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	294	5	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	294	6	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	294	7	with	with	IN
work_v44vkh6bzramngvcrficy2no44	294	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	294	9	medium-	medium-	NN
work_v44vkh6bzramngvcrficy2no44	294	10	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	294	11	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	294	12	in	in	IN
work_v44vkh6bzramngvcrficy2no44	294	13	Table	Table	NNP
work_v44vkh6bzramngvcrficy2no44	294	14	5	5	CD
work_v44vkh6bzramngvcrficy2no44	294	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	294	16	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	294	17	note	note	VBP
work_v44vkh6bzramngvcrficy2no44	294	18	that	that	IN
work_v44vkh6bzramngvcrficy2no44	294	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	294	20	addition	addition	NN
work_v44vkh6bzramngvcrficy2no44	294	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	294	22	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	294	23	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	294	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	294	25	especially	especially	RB
work_v44vkh6bzramngvcrficy2no44	294	26	beneficial	beneficial	JJ
work_v44vkh6bzramngvcrficy2no44	294	27	for	for	IN
work_v44vkh6bzramngvcrficy2no44	294	28	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	294	29	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	294	30	+1.5	+1.5	NNP
work_v44vkh6bzramngvcrficy2no44	294	31	on	on	IN
work_v44vkh6bzramngvcrficy2no44	294	32	eval08-nw	eval08-nw	NNP
work_v44vkh6bzramngvcrficy2no44	294	33	and	and	CC
work_v44vkh6bzramngvcrficy2no44	294	34	+1.0	+1.0	ADD
work_v44vkh6bzramngvcrficy2no44	294	35	on	on	IN
work_v44vkh6bzramngvcrficy2no44	294	36	eval09-nw	eval09-nw	NNP
work_v44vkh6bzramngvcrficy2no44	294	37	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	294	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	294	39	but	but	CC
work_v44vkh6bzramngvcrficy2no44	294	40	not	not	RB
work_v44vkh6bzramngvcrficy2no44	294	41	as	as	RB
work_v44vkh6bzramngvcrficy2no44	294	42	much	much	JJ
work_v44vkh6bzramngvcrficy2no44	294	43	for	for	IN
work_v44vkh6bzramngvcrficy2no44	294	44	the	the	DT
work_v44vkh6bzramngvcrficy2no44	294	45	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	294	46	metrics	metric	NNS
work_v44vkh6bzramngvcrficy2no44	294	47	,	,	,
work_v44vkh6bzramngvcrficy2no44	294	48	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	294	49	challenges	challenge	VBZ
work_v44vkh6bzramngvcrficy2no44	294	50	the	the	DT
work_v44vkh6bzramngvcrficy2no44	294	51	commonly	commonly	RB
work_v44vkh6bzramngvcrficy2no44	294	52	held	hold	VBN
work_v44vkh6bzramngvcrficy2no44	294	53	idea	idea	NN
work_v44vkh6bzramngvcrficy2no44	294	54	that	that	IN
work_v44vkh6bzramngvcrficy2no44	294	55	more	more	JJR
work_v44vkh6bzramngvcrficy2no44	294	56	data	datum	NNS
work_v44vkh6bzramngvcrficy2no44	294	57	always	always	RB
work_v44vkh6bzramngvcrficy2no44	294	58	improves	improve	VBZ
work_v44vkh6bzramngvcrficy2no44	294	59	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	294	60	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	294	61	.	.	.
work_v44vkh6bzramngvcrficy2no44	295	1	Here	here	RB
work_v44vkh6bzramngvcrficy2no44	295	2	too	too	RB
work_v44vkh6bzramngvcrficy2no44	295	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	295	4	relaxing	relax	VBG
work_v44vkh6bzramngvcrficy2no44	295	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	295	6	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	295	7	without	without	IN
work_v44vkh6bzramngvcrficy2no44	295	8	special	special	JJ
work_v44vkh6bzramngvcrficy2no44	295	9	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	295	10	hurts	hurt	NNS
work_v44vkh6bzramngvcrficy2no44	295	11	not	not	RB
work_v44vkh6bzramngvcrficy2no44	295	12	only	only	RB
work_v44vkh6bzramngvcrficy2no44	295	13	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	295	14	but	but	CC
work_v44vkh6bzramngvcrficy2no44	295	15	also	also	RB
work_v44vkh6bzramngvcrficy2no44	295	16	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	295	17	qual-	qual-	NNP
work_v44vkh6bzramngvcrficy2no44	295	18	ity	ity	NN
work_v44vkh6bzramngvcrficy2no44	295	19	:	:	:
work_v44vkh6bzramngvcrficy2no44	295	20	all	all	PDT
work_v44vkh6bzramngvcrficy2no44	295	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	295	22	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	295	23	decrease	decrease	VBP
work_v44vkh6bzramngvcrficy2no44	295	24	considerably	considerably	RB
work_v44vkh6bzramngvcrficy2no44	295	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	295	26	showing	show	VBG
work_v44vkh6bzramngvcrficy2no44	295	27	that	that	IN
work_v44vkh6bzramngvcrficy2no44	295	28	even	even	RB
work_v44vkh6bzramngvcrficy2no44	295	29	the	the	DT
work_v44vkh6bzramngvcrficy2no44	295	30	stronger	strong	JJR
work_v44vkh6bzramngvcrficy2no44	295	31	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	295	32	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	295	33	not	not	RB
work_v44vkh6bzramngvcrficy2no44	295	34	sufficient	sufficient	JJ
work_v44vkh6bzramngvcrficy2no44	295	35	to	to	TO
work_v44vkh6bzramngvcrficy2no44	295	36	guide	guide	VB
work_v44vkh6bzramngvcrficy2no44	295	37	search	search	VB
work_v44vkh6bzramngvcrficy2no44	295	38	through	through	IN
work_v44vkh6bzramngvcrficy2no44	295	39	a	a	DT
work_v44vkh6bzramngvcrficy2no44	295	40	very	very	RB
work_v44vkh6bzramngvcrficy2no44	295	41	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	295	42	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	295	43	search	search	NN
work_v44vkh6bzramngvcrficy2no44	295	44	space	space	NN
work_v44vkh6bzramngvcrficy2no44	295	45	.	.	.
work_v44vkh6bzramngvcrficy2no44	296	1	As	as	IN
work_v44vkh6bzramngvcrficy2no44	296	2	for	for	IN
work_v44vkh6bzramngvcrficy2no44	296	3	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	296	4	enhanced	enhanced	JJ
work_v44vkh6bzramngvcrficy2no44	296	5	system	system	NN
work_v44vkh6bzramngvcrficy2no44	296	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	296	7	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	296	8	achieves	achieve	VBZ
work_v44vkh6bzramngvcrficy2no44	296	9	simi-	simi-	NN
work_v44vkh6bzramngvcrficy2no44	296	10	lar	lar	NN
work_v44vkh6bzramngvcrficy2no44	296	11	gains	gain	NNS
work_v44vkh6bzramngvcrficy2no44	296	12	as	as	IN
work_v44vkh6bzramngvcrficy2no44	296	13	in	in	IN
work_v44vkh6bzramngvcrficy2no44	296	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	296	15	medium	medium	JJ
work_v44vkh6bzramngvcrficy2no44	296	16	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	296	17	scale	scale	NN
work_v44vkh6bzramngvcrficy2no44	296	18	scenario	scenario	NN
work_v44vkh6bzramngvcrficy2no44	296	19	:	:	:
work_v44vkh6bzramngvcrficy2no44	296	20	that	that	RB
work_v44vkh6bzramngvcrficy2no44	296	21	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	296	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	296	23	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	296	24	and	and	CC
work_v44vkh6bzramngvcrficy2no44	296	25	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	296	26	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	296	27	preserved	preserve	VBN
work_v44vkh6bzramngvcrficy2no44	296	28	or	or	CC
work_v44vkh6bzramngvcrficy2no44	296	29	slightly	slightly	RB
work_v44vkh6bzramngvcrficy2no44	296	30	im-	im-	VBZ
work_v44vkh6bzramngvcrficy2no44	296	31	proved	prove	VBN
work_v44vkh6bzramngvcrficy2no44	296	32	despite	despite	IN
work_v44vkh6bzramngvcrficy2no44	296	33	the	the	DT
work_v44vkh6bzramngvcrficy2no44	296	34	very	very	RB
work_v44vkh6bzramngvcrficy2no44	296	35	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	296	36	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	296	37	,	,	,
work_v44vkh6bzramngvcrficy2no44	296	38	while	while	IN
work_v44vkh6bzramngvcrficy2no44	296	39	all	all	PDT
work_v44vkh6bzramngvcrficy2no44	296	40	the	the	DT
work_v44vkh6bzramngvcrficy2no44	296	41	re-	re-	JJ
work_v44vkh6bzramngvcrficy2no44	296	42	ordering	ordering	NN
work_v44vkh6bzramngvcrficy2no44	296	43	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	296	44	increase	increase	VB
work_v44vkh6bzramngvcrficy2no44	296	45	.	.	.
work_v44vkh6bzramngvcrficy2no44	297	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	297	2	particular	particular	JJ
work_v44vkh6bzramngvcrficy2no44	297	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	297	4	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	297	5	report	report	VBP
work_v44vkh6bzramngvcrficy2no44	297	6	sta-	sta-	VBP
work_v44vkh6bzramngvcrficy2no44	297	7	tistically	tistically	RB
work_v44vkh6bzramngvcrficy2no44	297	8	significant	significant	JJ
work_v44vkh6bzramngvcrficy2no44	297	9	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	297	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	297	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	297	12	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	297	13	of	of	IN
work_v44vkh6bzramngvcrficy2no44	297	14	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	297	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	297	16	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	297	17	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	297	18	where	where	WRB
work_v44vkh6bzramngvcrficy2no44	297	19	the	the	DT
work_v44vkh6bzramngvcrficy2no44	297	20	impact	impact	NN
work_v44vkh6bzramngvcrficy2no44	297	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	297	22	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	297	23	method	method	NN
work_v44vkh6bzramngvcrficy2no44	297	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	297	25	expected	expect	VBN
work_v44vkh6bzramngvcrficy2no44	297	26	to	to	TO
work_v44vkh6bzramngvcrficy2no44	297	27	concentrate	concentrate	VB
work_v44vkh6bzramngvcrficy2no44	297	28	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	297	29	+0.7	+0.7	NNP
work_v44vkh6bzramngvcrficy2no44	297	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	297	31	+0.8	+0.8	NNP
work_v44vkh6bzramngvcrficy2no44	297	32	and	and	CC
work_v44vkh6bzramngvcrficy2no44	297	33	+1.0	+1.0	NNP
work_v44vkh6bzramngvcrficy2no44	297	34	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	297	35	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	297	36	V	V	NNP
work_v44vkh6bzramngvcrficy2no44	297	37	on	on	IN
work_v44vkh6bzramngvcrficy2no44	297	38	eval08-nw	eval08-nw	NNP
work_v44vkh6bzramngvcrficy2no44	297	39	,	,	,
work_v44vkh6bzramngvcrficy2no44	297	40	eval09-nw	eval09-nw	NNP
work_v44vkh6bzramngvcrficy2no44	297	41	and	and	CC
work_v44vkh6bzramngvcrficy2no44	297	42	vs-09	vs-09	NNP
work_v44vkh6bzramngvcrficy2no44	297	43	,	,	,
work_v44vkh6bzramngvcrficy2no44	297	44	respectively	respectively	RB
work_v44vkh6bzramngvcrficy2no44	297	45	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	297	46	.	.	.
work_v44vkh6bzramngvcrficy2no44	298	1	These	these	DT
work_v44vkh6bzramngvcrficy2no44	298	2	results	result	NNS
work_v44vkh6bzramngvcrficy2no44	298	3	confirm	confirm	VBP
work_v44vkh6bzramngvcrficy2no44	298	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	298	5	usefulness	usefulness	NN
work_v44vkh6bzramngvcrficy2no44	298	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	298	7	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	298	8	method	method	NN
work_v44vkh6bzramngvcrficy2no44	298	9	not	not	RB
work_v44vkh6bzramngvcrficy2no44	298	10	only	only	RB
work_v44vkh6bzramngvcrficy2no44	298	11	as	as	IN
work_v44vkh6bzramngvcrficy2no44	298	12	an	an	DT
work_v44vkh6bzramngvcrficy2no44	298	13	optimization	optimization	NN
work_v44vkh6bzramngvcrficy2no44	298	14	technique	technique	NN
work_v44vkh6bzramngvcrficy2no44	298	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	298	16	but	but	CC
work_v44vkh6bzramngvcrficy2no44	298	17	also	also	RB
work_v44vkh6bzramngvcrficy2no44	298	18	as	as	IN
work_v44vkh6bzramngvcrficy2no44	298	19	a	a	DT
work_v44vkh6bzramngvcrficy2no44	298	20	way	way	NN
work_v44vkh6bzramngvcrficy2no44	298	21	to	to	TO
work_v44vkh6bzramngvcrficy2no44	298	22	improve	improve	VB
work_v44vkh6bzramngvcrficy2no44	298	23	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	298	24	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	298	25	on	on	IN
work_v44vkh6bzramngvcrficy2no44	298	26	top	top	NN
work_v44vkh6bzramngvcrficy2no44	298	27	of	of	IN
work_v44vkh6bzramngvcrficy2no44	298	28	a	a	DT
work_v44vkh6bzramngvcrficy2no44	298	29	very	very	RB
work_v44vkh6bzramngvcrficy2no44	298	30	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	298	31	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	298	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	299	1	336	336	CD
work_v44vkh6bzramngvcrficy2no44	299	2	SRC	SRC	NNP
work_v44vkh6bzramngvcrficy2no44	299	3	!	!	.
work_v44vkh6bzramngvcrficy2no44	300	1	!	!	.
work_v44vkh6bzramngvcrficy2no44	301	1	!	!	.
work_v44vkh6bzramngvcrficy2no44	302	1	"	"	``
work_v44vkh6bzramngvcrficy2no44	302	2	#	#	$
work_v44vkh6bzramngvcrficy2no44	302	3	$	$	$
work_v44vkh6bzramngvcrficy2no44	302	4	!	!	.
work_v44vkh6bzramngvcrficy2no44	302	5	%	%	NNP
work_v44vkh6bzramngvcrficy2no44	302	6	&	&	CC
work_v44vkh6bzramngvcrficy2no44	302	7	'	'	''
work_v44vkh6bzramngvcrficy2no44	302	8	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	302	9	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	302	10	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	302	11	+	+	SYM
work_v44vkh6bzramngvcrficy2no44	302	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	302	13	-	-	:
work_v44vkh6bzramngvcrficy2no44	302	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	303	1	/0$%	/0$%	NNP
work_v44vkh6bzramngvcrficy2no44	303	2	&	&	CC
work_v44vkh6bzramngvcrficy2no44	303	3	'	'	NNP
work_v44vkh6bzramngvcrficy2no44	303	4	&	&	CC
work_v44vkh6bzramngvcrficy2no44	303	5	-1	-1	.
work_v44vkh6bzramngvcrficy2no44	303	6	!	!	.
work_v44vkh6bzramngvcrficy2no44	304	1	2	2	CD
work_v44vkh6bzramngvcrficy2no44	304	2	+1	+1	SYM
work_v44vkh6bzramngvcrficy2no44	304	3	+03	+03	CD
work_v44vkh6bzramngvcrficy2no44	304	4	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	304	5	+045(67!8	+045(67!8	NNP
work_v44vkh6bzramngvcrficy2no44	304	6	+9#+77!5	+9#+77!5	NNP
work_v44vkh6bzramngvcrficy2no44	304	7	:	:	:
work_v44vkh6bzramngvcrficy2no44	304	8	65	65	CD
work_v44vkh6bzramngvcrficy2no44	304	9	&	&	CC
work_v44vkh6bzramngvcrficy2no44	304	10	-3*;24<5	-3*;24<5	NNP
work_v44vkh6bzramngvcrficy2no44	304	11	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	304	12	&	&	CC
work_v44vkh6bzramngvcrficy2no44	304	13	-73	-73	NNP
work_v44vkh6bzramngvcrficy2no44	304	14	!	!	.
work_v44vkh6bzramngvcrficy2no44	305	1	045	045	CD
work_v44vkh6bzramngvcrficy2no44	305	2	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	305	3	&	&	CC
work_v44vkh6bzramngvcrficy2no44	305	4	-=	-=	NNP
work_v44vkh6bzramngvcrficy2no44	305	5	>	>	XX
work_v44vkh6bzramngvcrficy2no44	305	6	?	?	.
work_v44vkh6bzramngvcrficy2no44	306	1	@A	@A	NNP
work_v44vkh6bzramngvcrficy2no44	306	2	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	306	3	0B	0B	NNP
work_v44vkh6bzramngvcrficy2no44	306	4	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	306	5	+	+	CC
work_v44vkh6bzramngvcrficy2no44	306	6	CD	cd	NN
work_v44vkh6bzramngvcrficy2no44	306	7	EF(23	ef(23	NN
work_v44vkh6bzramngvcrficy2no44	306	8	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	306	9	verb	verb	JJ
work_v44vkh6bzramngvcrficy2no44	306	10	subj	subj	NN
work_v44vkh6bzramngvcrficy2no44	306	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	307	1	obj	obj	UH
work_v44vkh6bzramngvcrficy2no44	307	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	308	1	compl	compl	NNP
work_v44vkh6bzramngvcrficy2no44	308	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	309	1	ywASl	ywASl	NFP
work_v44vkh6bzramngvcrficy2no44	309	2	sfyr	sfyr	NNS
work_v44vkh6bzramngvcrficy2no44	309	3	Almmlkp	Almmlkp	NNP
work_v44vkh6bzramngvcrficy2no44	309	4	AlErbyp	AlErbyp	NNP
work_v44vkh6bzramngvcrficy2no44	309	5	AlsEwdyp	AlsEwdyp	NNP
work_v44vkh6bzramngvcrficy2no44	309	6	ldY	ldY	NNP
work_v44vkh6bzramngvcrficy2no44	309	7	lbnAn	lbnAn	-LRB-
work_v44vkh6bzramngvcrficy2no44	309	8	EbdAlEzyz	EbdAlEzyz	NNP
work_v44vkh6bzramngvcrficy2no44	309	9	xwjp	xwjp	FW
work_v44vkh6bzramngvcrficy2no44	309	10	tHrk	tHrk	NNP
work_v44vkh6bzramngvcrficy2no44	309	11	-h	-h	:
work_v44vkh6bzramngvcrficy2no44	309	12	fy	fy	NNP
work_v44vkh6bzramngvcrficy2no44	309	13	AtjAh	AtjAh	NNP
work_v44vkh6bzramngvcrficy2no44	309	14	...	...	NFP
work_v44vkh6bzramngvcrficy2no44	309	15	continues	continue	VBZ
work_v44vkh6bzramngvcrficy2no44	309	16	ambassador	ambassador	NN
work_v44vkh6bzramngvcrficy2no44	309	17	Kingdom	Kingdom	NNP
work_v44vkh6bzramngvcrficy2no44	309	18	Arabian	Arabian	NNP
work_v44vkh6bzramngvcrficy2no44	309	19	Saudi	Saudi	NNP
work_v44vkh6bzramngvcrficy2no44	309	20	to	to	IN
work_v44vkh6bzramngvcrficy2no44	309	21	Lebanon	Lebanon	NNP
work_v44vkh6bzramngvcrficy2no44	309	22	Abdulaziz	Abdulaziz	NNP
work_v44vkh6bzramngvcrficy2no44	309	23	Khawja	Khawja	NNP
work_v44vkh6bzramngvcrficy2no44	309	24	move	move	VB
work_v44vkh6bzramngvcrficy2no44	309	25	his	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	309	26	in	in	IN
work_v44vkh6bzramngvcrficy2no44	309	27	direction	direction	NN
work_v44vkh6bzramngvcrficy2no44	309	28	REF	ref	NN
work_v44vkh6bzramngvcrficy2no44	309	29	The	the	DT
work_v44vkh6bzramngvcrficy2no44	309	30	Kingdom	Kingdom	NNP
work_v44vkh6bzramngvcrficy2no44	309	31	of	of	IN
work_v44vkh6bzramngvcrficy2no44	309	32	Saudi	Saudi	NNP
work_v44vkh6bzramngvcrficy2no44	309	33	Arabia	Arabia	NNP
work_v44vkh6bzramngvcrficy2no44	309	34	’s	's	POS
work_v44vkh6bzramngvcrficy2no44	309	35	ambassador	ambassador	NN
work_v44vkh6bzramngvcrficy2no44	309	36	to	to	IN
work_v44vkh6bzramngvcrficy2no44	309	37	Lebanon	Lebanon	NNP
work_v44vkh6bzramngvcrficy2no44	309	38	Abdulaziz	Abdulaziz	NNP
work_v44vkh6bzramngvcrficy2no44	309	39	Khawja	Khawja	NNP
work_v44vkh6bzramngvcrficy2no44	309	40	continues	continue	VBZ
work_v44vkh6bzramngvcrficy2no44	309	41	his	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	309	42	moves	move	NNS
work_v44vkh6bzramngvcrficy2no44	309	43	towards	towards	IN
work_v44vkh6bzramngvcrficy2no44	309	44	...	...	:
work_v44vkh6bzramngvcrficy2no44	309	45	BASE	BASE	NNP
work_v44vkh6bzramngvcrficy2no44	309	46	continue	continue	VBP
work_v44vkh6bzramngvcrficy2no44	309	47	to	to	IN
work_v44vkh6bzramngvcrficy2no44	309	48	Saudi	saudi	JJ
work_v44vkh6bzramngvcrficy2no44	309	49	Arabian	arabian	JJ
work_v44vkh6bzramngvcrficy2no44	309	50	ambassador	ambassador	NN
work_v44vkh6bzramngvcrficy2no44	309	51	to	to	IN
work_v44vkh6bzramngvcrficy2no44	309	52	Lebanon	Lebanon	NNP
work_v44vkh6bzramngvcrficy2no44	309	53	,	,	,
work_v44vkh6bzramngvcrficy2no44	309	54	Abdulaziz	Abdulaziz	NNP
work_v44vkh6bzramngvcrficy2no44	309	55	Khwja	Khwja	NNP
work_v44vkh6bzramngvcrficy2no44	309	56	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	309	57	move	move	NN
work_v44vkh6bzramngvcrficy2no44	309	58	in	in	IN
work_v44vkh6bzramngvcrficy2no44	309	59	the	the	DT
work_v44vkh6bzramngvcrficy2no44	309	60	direction	direction	NN
work_v44vkh6bzramngvcrficy2no44	309	61	of	of	IN
work_v44vkh6bzramngvcrficy2no44	309	62	...	...	:
work_v44vkh6bzramngvcrficy2no44	309	63	NEW	NEW	NNP
work_v44vkh6bzramngvcrficy2no44	309	64	The	the	DT
work_v44vkh6bzramngvcrficy2no44	309	65	Kingdom	Kingdom	NNP
work_v44vkh6bzramngvcrficy2no44	309	66	of	of	IN
work_v44vkh6bzramngvcrficy2no44	309	67	Saudi	Saudi	NNP
work_v44vkh6bzramngvcrficy2no44	309	68	Arabia	Arabia	NNP
work_v44vkh6bzramngvcrficy2no44	309	69	’s	's	POS
work_v44vkh6bzramngvcrficy2no44	309	70	ambassador	ambassador	NN
work_v44vkh6bzramngvcrficy2no44	309	71	to	to	IN
work_v44vkh6bzramngvcrficy2no44	309	72	Lebanon	Lebanon	NNP
work_v44vkh6bzramngvcrficy2no44	309	73	,	,	,
work_v44vkh6bzramngvcrficy2no44	309	74	Abdulaziz	Abdulaziz	NNP
work_v44vkh6bzramngvcrficy2no44	309	75	Khwja	Khwja	NNP
work_v44vkh6bzramngvcrficy2no44	309	76	continue	continue	VBP
work_v44vkh6bzramngvcrficy2no44	309	77	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	309	78	move	move	NN
work_v44vkh6bzramngvcrficy2no44	309	79	in	in	IN
work_v44vkh6bzramngvcrficy2no44	309	80	the	the	DT
work_v44vkh6bzramngvcrficy2no44	309	81	direction	direction	NN
work_v44vkh6bzramngvcrficy2no44	309	82	of	of	IN
work_v44vkh6bzramngvcrficy2no44	309	83	...	...	:
work_v44vkh6bzramngvcrficy2no44	309	84	SRC	SRC	NNP
work_v44vkh6bzramngvcrficy2no44	309	85	;	;	:
work_v44vkh6bzramngvcrficy2no44	309	86	#	#	$
work_v44vkh6bzramngvcrficy2no44	309	87	7*$G	7*$g	CD
work_v44vkh6bzramngvcrficy2no44	309	88	'	'	''
work_v44vkh6bzramngvcrficy2no44	309	89	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	309	90	H	H	NNP
work_v44vkh6bzramngvcrficy2no44	309	91	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	309	92	+0	+0	NNP
work_v44vkh6bzramngvcrficy2no44	309	93	&	&	CC
work_v44vkh6bzramngvcrficy2no44	309	94	B5	B5	NNP
work_v44vkh6bzramngvcrficy2no44	309	95	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	309	96	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	309	97	I	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	309	98	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	309	99	E4	E4	NNP
work_v44vkh6bzramngvcrficy2no44	309	100	J	J	NNP
work_v44vkh6bzramngvcrficy2no44	309	101	<	<	XX
work_v44vkh6bzramngvcrficy2no44	309	102	K	K	NNP
work_v44vkh6bzramngvcrficy2no44	309	103	65	65	CD
work_v44vkh6bzramngvcrficy2no44	309	104	#	#	NN
work_v44vkh6bzramngvcrficy2no44	309	105	+	+	CC
work_v44vkh6bzramngvcrficy2no44	309	106	L	L	NNP
work_v44vkh6bzramngvcrficy2no44	309	107	M#NL	M#NL	NNPS
work_v44vkh6bzramngvcrficy2no44	309	108	&	&	CC
work_v44vkh6bzramngvcrficy2no44	309	109	-O01	-O01	NNP
work_v44vkh6bzramngvcrficy2no44	309	110	P5	p5	NN
work_v44vkh6bzramngvcrficy2no44	309	111	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	309	112	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	309	113	QR#7*<5	QR#7*<5	NNP
work_v44vkh6bzramngvcrficy2no44	309	114	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	309	115	S	S	NNP
work_v44vkh6bzramngvcrficy2no44	309	116	!	!	.
work_v44vkh6bzramngvcrficy2no44	310	1	&	&	CC
work_v44vkh6bzramngvcrficy2no44	310	2	7=@A	7=@A	NNP
work_v44vkh6bzramngvcrficy2no44	310	3	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	310	4	TU*V3W	TU*V3W	NNP
work_v44vkh6bzramngvcrficy2no44	310	5	XY	XY	NNP
work_v44vkh6bzramngvcrficy2no44	310	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	311	1	#	#	NN
work_v44vkh6bzramngvcrficy2no44	311	2	8	8	CD
work_v44vkh6bzramngvcrficy2no44	311	3	;	;	:
work_v44vkh6bzramngvcrficy2no44	311	4	#	#	$
work_v44vkh6bzramngvcrficy2no44	311	5	?	?	.
work_v44vkh6bzramngvcrficy2no44	311	6	7	7	CD
work_v44vkh6bzramngvcrficy2no44	311	7	*	*	NFP
work_v44vkh6bzramngvcrficy2no44	311	8	+	+	CC
work_v44vkh6bzramngvcrficy2no44	311	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	311	10	adv	adv	NN
work_v44vkh6bzramngvcrficy2no44	311	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	312	1	verb	verb	NNP
work_v44vkh6bzramngvcrficy2no44	312	2	obj	obj	NN
work_v44vkh6bzramngvcrficy2no44	312	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	313	1	subj	subj	NNP
work_v44vkh6bzramngvcrficy2no44	313	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	314	1	compl	compl	NNP
work_v44vkh6bzramngvcrficy2no44	314	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	315	1	fymA	fymA	NNP
work_v44vkh6bzramngvcrficy2no44	315	2	dEA	dEA	NNP
work_v44vkh6bzramngvcrficy2no44	315	3	-hm	-hm	-RRB-
work_v44vkh6bzramngvcrficy2no44	315	4	r}ys	r}ys	NNP
work_v44vkh6bzramngvcrficy2no44	315	5	Almktb	Almktb	NNP
work_v44vkh6bzramngvcrficy2no44	315	6	AlsyAsy	AlsyAsy	NNP
work_v44vkh6bzramngvcrficy2no44	315	7	l-	l-	NN
work_v44vkh6bzramngvcrficy2no44	315	8	Hrkp	Hrkp	NNP
work_v44vkh6bzramngvcrficy2no44	315	9	HmAs	HmAs	NNP
work_v44vkh6bzramngvcrficy2no44	315	10	xAld	xAld	NNP
work_v44vkh6bzramngvcrficy2no44	315	11	m$El	m$El	NNP
work_v44vkh6bzramngvcrficy2no44	315	12	AlY	AlY	NNP
work_v44vkh6bzramngvcrficy2no44	315	13	AltzAm	altzam	CC
work_v44vkh6bzramngvcrficy2no44	315	14	AlHyAd	AlHyAd	NNP
work_v44vkh6bzramngvcrficy2no44	315	15	meanwhile	meanwhile	RB
work_v44vkh6bzramngvcrficy2no44	315	16	called	call	VBD
work_v44vkh6bzramngvcrficy2no44	315	17	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	315	18	head	head	NN
work_v44vkh6bzramngvcrficy2no44	315	19	bureau	bureau	NNP
work_v44vkh6bzramngvcrficy2no44	315	20	political	political	JJ
work_v44vkh6bzramngvcrficy2no44	315	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	315	22	movement	movement	NN
work_v44vkh6bzramngvcrficy2no44	315	23	Hamas	Hamas	NNP
work_v44vkh6bzramngvcrficy2no44	315	24	Khaled	Khaled	NNP
work_v44vkh6bzramngvcrficy2no44	315	25	Mashal	Mashal	NNP
work_v44vkh6bzramngvcrficy2no44	315	26	to	to	IN
work_v44vkh6bzramngvcrficy2no44	315	27	necessity	necessity	NN
work_v44vkh6bzramngvcrficy2no44	315	28	neutrality	neutrality	NN
work_v44vkh6bzramngvcrficy2no44	315	29	REF	REF	NNP
work_v44vkh6bzramngvcrficy2no44	315	30	Meanwhile	meanwhile	RB
work_v44vkh6bzramngvcrficy2no44	315	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	32	the	the	DT
work_v44vkh6bzramngvcrficy2no44	315	33	Head	Head	NNP
work_v44vkh6bzramngvcrficy2no44	315	34	of	of	IN
work_v44vkh6bzramngvcrficy2no44	315	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	315	36	Political	Political	NNP
work_v44vkh6bzramngvcrficy2no44	315	37	Bureau	Bureau	NNP
work_v44vkh6bzramngvcrficy2no44	315	38	of	of	IN
work_v44vkh6bzramngvcrficy2no44	315	39	the	the	DT
work_v44vkh6bzramngvcrficy2no44	315	40	Hamas	Hamas	NNP
work_v44vkh6bzramngvcrficy2no44	315	41	movement	movement	NN
work_v44vkh6bzramngvcrficy2no44	315	42	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	43	Khaled	Khaled	NNP
work_v44vkh6bzramngvcrficy2no44	315	44	Mashal	Mashal	NNP
work_v44vkh6bzramngvcrficy2no44	315	45	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	46	called	call	VBD
work_v44vkh6bzramngvcrficy2no44	315	47	upon	upon	IN
work_v44vkh6bzramngvcrficy2no44	315	48	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	315	49	to	to	TO
work_v44vkh6bzramngvcrficy2no44	315	50	remain	remain	VB
work_v44vkh6bzramngvcrficy2no44	315	51	neutral	neutral	JJ
work_v44vkh6bzramngvcrficy2no44	315	52	BASE	BASE	NNP
work_v44vkh6bzramngvcrficy2no44	315	53	The	the	DT
work_v44vkh6bzramngvcrficy2no44	315	54	called	call	VBN
work_v44vkh6bzramngvcrficy2no44	315	55	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	315	56	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	57	head	head	NN
work_v44vkh6bzramngvcrficy2no44	315	58	of	of	IN
work_v44vkh6bzramngvcrficy2no44	315	59	Hamas	Hamas	NNP
work_v44vkh6bzramngvcrficy2no44	315	60	’	’	POS
work_v44vkh6bzramngvcrficy2no44	315	61	political	political	JJ
work_v44vkh6bzramngvcrficy2no44	315	62	bureau	bureau	NN
work_v44vkh6bzramngvcrficy2no44	315	63	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	64	Khalid	Khalid	NNP
work_v44vkh6bzramngvcrficy2no44	315	65	Mashal	Mashal	NNP
work_v44vkh6bzramngvcrficy2no44	315	66	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	67	to	to	TO
work_v44vkh6bzramngvcrficy2no44	315	68	remain	remain	VB
work_v44vkh6bzramngvcrficy2no44	315	69	neutral	neutral	JJ
work_v44vkh6bzramngvcrficy2no44	315	70	NEW	NEW	NNP
work_v44vkh6bzramngvcrficy2no44	315	71	The	the	DT
work_v44vkh6bzramngvcrficy2no44	315	72	head	head	NN
work_v44vkh6bzramngvcrficy2no44	315	73	of	of	IN
work_v44vkh6bzramngvcrficy2no44	315	74	Hamas	Hamas	NNP
work_v44vkh6bzramngvcrficy2no44	315	75	’	’	POS
work_v44vkh6bzramngvcrficy2no44	315	76	political	political	JJ
work_v44vkh6bzramngvcrficy2no44	315	77	bureau	bureau	NN
work_v44vkh6bzramngvcrficy2no44	315	78	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	79	Khalid	Khalid	NNP
work_v44vkh6bzramngvcrficy2no44	315	80	Mashal	Mashal	NNP
work_v44vkh6bzramngvcrficy2no44	315	81	,	,	,
work_v44vkh6bzramngvcrficy2no44	315	82	called	call	VBD
work_v44vkh6bzramngvcrficy2no44	315	83	on	on	IN
work_v44vkh6bzramngvcrficy2no44	315	84	them	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	315	85	to	to	TO
work_v44vkh6bzramngvcrficy2no44	315	86	remain	remain	VB
work_v44vkh6bzramngvcrficy2no44	315	87	neutral	neutral	JJ
work_v44vkh6bzramngvcrficy2no44	315	88	Figure	figure	NN
work_v44vkh6bzramngvcrficy2no44	315	89	3	3	CD
work_v44vkh6bzramngvcrficy2no44	315	90	:	:	:
work_v44vkh6bzramngvcrficy2no44	315	91	Long	long	JJ
work_v44vkh6bzramngvcrficy2no44	315	92	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	315	93	examples	example	NNS
work_v44vkh6bzramngvcrficy2no44	315	94	showing	show	VBG
work_v44vkh6bzramngvcrficy2no44	315	95	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	315	96	over	over	IN
work_v44vkh6bzramngvcrficy2no44	315	97	the	the	DT
work_v44vkh6bzramngvcrficy2no44	315	98	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	315	99	system	system	NN
work_v44vkh6bzramngvcrficy2no44	315	100	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	315	101	BASE	BASE	NNP
work_v44vkh6bzramngvcrficy2no44	315	102	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	315	103	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	315	104	the	the	DT
work_v44vkh6bzramngvcrficy2no44	315	105	DL	DL	NNP
work_v44vkh6bzramngvcrficy2no44	315	106	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	315	107	raised	raise	VBN
work_v44vkh6bzramngvcrficy2no44	315	108	to	to	IN
work_v44vkh6bzramngvcrficy2no44	315	109	18	18	CD
work_v44vkh6bzramngvcrficy2no44	315	110	and	and	CC
work_v44vkh6bzramngvcrficy2no44	315	111	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	315	112	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	315	113	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	315	114	on	on	IN
work_v44vkh6bzramngvcrficy2no44	315	115	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	315	116	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	315	117	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	315	118	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	315	119	enabled	enable	VBN
work_v44vkh6bzramngvcrficy2no44	315	120	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	315	121	NEW	NEW	NNP
work_v44vkh6bzramngvcrficy2no44	315	122	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	315	123	.	.	.
work_v44vkh6bzramngvcrficy2no44	316	1	Long	long	JJ
work_v44vkh6bzramngvcrficy2no44	316	2	jumps	jump	VBZ
work_v44vkh6bzramngvcrficy2no44	316	3	statistics	statistic	NNS
work_v44vkh6bzramngvcrficy2no44	316	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	316	5	examples	example	NNS
work_v44vkh6bzramngvcrficy2no44	316	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	317	1	To	to	TO
work_v44vkh6bzramngvcrficy2no44	317	2	better	well	RBR
work_v44vkh6bzramngvcrficy2no44	317	3	understand	understand	VB
work_v44vkh6bzramngvcrficy2no44	317	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	317	5	behavior	behavior	NN
work_v44vkh6bzramngvcrficy2no44	317	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	317	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	317	8	early	early	RB
work_v44vkh6bzramngvcrficy2no44	317	9	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	317	10	pruning	prune	VBG
work_v44vkh6bzramngvcrficy2no44	317	11	system	system	NN
work_v44vkh6bzramngvcrficy2no44	317	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	317	13	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	317	14	extract	extract	VBP
work_v44vkh6bzramngvcrficy2no44	317	15	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	317	16	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	317	17	to	to	IN
work_v44vkh6bzramngvcrficy2no44	317	18	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	317	19	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	317	20	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	317	21	statistics	statistic	NNS
work_v44vkh6bzramngvcrficy2no44	317	22	from	from	IN
work_v44vkh6bzramngvcrficy2no44	317	23	the	the	DT
work_v44vkh6bzramngvcrficy2no44	317	24	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	317	25	log	log	NN
work_v44vkh6bzramngvcrficy2no44	317	26	file	file	NN
work_v44vkh6bzramngvcrficy2no44	317	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	318	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	318	2	find	find	VBP
work_v44vkh6bzramngvcrficy2no44	318	3	that	that	IN
work_v44vkh6bzramngvcrficy2no44	318	4	132	132	CD
work_v44vkh6bzramngvcrficy2no44	318	5	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	318	6	beyond	beyond	IN
work_v44vkh6bzramngvcrficy2no44	318	7	the	the	DT
work_v44vkh6bzramngvcrficy2no44	318	8	non	non	JJ
work_v44vkh6bzramngvcrficy2no44	318	9	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	318	10	prunable	prunable	JJ
work_v44vkh6bzramngvcrficy2no44	318	11	zone	zone	NN
work_v44vkh6bzramngvcrficy2no44	318	12	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	318	13	D>5	D>5	NNP
work_v44vkh6bzramngvcrficy2no44	318	14	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	318	15	were	be	VBD
work_v44vkh6bzramngvcrficy2no44	318	16	performed	perform	VBN
work_v44vkh6bzramngvcrficy2no44	318	17	to	to	TO
work_v44vkh6bzramngvcrficy2no44	318	18	trans-	trans-	VB
work_v44vkh6bzramngvcrficy2no44	318	19	late	late	RB
work_v44vkh6bzramngvcrficy2no44	318	20	the	the	DT
work_v44vkh6bzramngvcrficy2no44	318	21	586	586	CD
work_v44vkh6bzramngvcrficy2no44	318	22	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	318	23	of	of	IN
work_v44vkh6bzramngvcrficy2no44	318	24	eval09-nw	eval09-nw	NN
work_v44vkh6bzramngvcrficy2no44	318	25	;	;	:
work_v44vkh6bzramngvcrficy2no44	318	26	38	38	CD
work_v44vkh6bzramngvcrficy2no44	318	27	out	out	IN
work_v44vkh6bzramngvcrficy2no44	318	28	of	of	IN
work_v44vkh6bzramngvcrficy2no44	318	29	these	these	DT
work_v44vkh6bzramngvcrficy2no44	318	30	were	be	VBD
work_v44vkh6bzramngvcrficy2no44	318	31	longer	long	JJR
work_v44vkh6bzramngvcrficy2no44	318	32	than	than	IN
work_v44vkh6bzramngvcrficy2no44	318	33	8	8	CD
work_v44vkh6bzramngvcrficy2no44	318	34	and	and	CC
work_v44vkh6bzramngvcrficy2no44	318	35	mostly	mostly	RB
work_v44vkh6bzramngvcrficy2no44	318	36	concentrated	concentrate	VBD
work_v44vkh6bzramngvcrficy2no44	318	37	on	on	IN
work_v44vkh6bzramngvcrficy2no44	318	38	the	the	DT
work_v44vkh6bzramngvcrficy2no44	318	39	VS-	vs-	JJ
work_v44vkh6bzramngvcrficy2no44	318	40	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	318	41	subset	subset	NN
work_v44vkh6bzramngvcrficy2no44	318	42	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	318	43	27	27	CD
work_v44vkh6bzramngvcrficy2no44	318	44	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	318	45	D>8	D>8	NNS
work_v44vkh6bzramngvcrficy2no44	318	46	performed	perform	VBN
work_v44vkh6bzramngvcrficy2no44	318	47	in	in	IN
work_v44vkh6bzramngvcrficy2no44	318	48	vs-09).13	vs-09).13	NNP
work_v44vkh6bzramngvcrficy2no44	318	49	This	this	DT
work_v44vkh6bzramngvcrficy2no44	318	50	and	and	CC
work_v44vkh6bzramngvcrficy2no44	318	51	the	the	DT
work_v44vkh6bzramngvcrficy2no44	318	52	higher	high	JJR
work_v44vkh6bzramngvcrficy2no44	318	53	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	318	54	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	318	55	sug-	sug-	RB
work_v44vkh6bzramngvcrficy2no44	318	56	gest	gest	VB
work_v44vkh6bzramngvcrficy2no44	318	57	that	that	IN
work_v44vkh6bzramngvcrficy2no44	318	58	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	318	59	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	318	60	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	318	61	mainly	mainly	RB
work_v44vkh6bzramngvcrficy2no44	318	62	carried	carry	VBN
work_v44vkh6bzramngvcrficy2no44	318	63	out	out	RP
work_v44vkh6bzramngvcrficy2no44	318	64	to	to	IN
work_v44vkh6bzramngvcrficy2no44	318	65	cor-	cor-	RB
work_v44vkh6bzramngvcrficy2no44	318	66	rectly	rectly	RB
work_v44vkh6bzramngvcrficy2no44	318	67	reorder	reorder	VB
work_v44vkh6bzramngvcrficy2no44	318	68	clause	clause	NN
work_v44vkh6bzramngvcrficy2no44	318	69	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	318	70	inital	inital	JJ
work_v44vkh6bzramngvcrficy2no44	318	71	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	318	72	over	over	IN
work_v44vkh6bzramngvcrficy2no44	318	73	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	318	74	subjects	subject	NNS
work_v44vkh6bzramngvcrficy2no44	318	75	.	.	.
work_v44vkh6bzramngvcrficy2no44	319	1	Fig	fig	NN
work_v44vkh6bzramngvcrficy2no44	319	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	320	1	3	3	CD
work_v44vkh6bzramngvcrficy2no44	320	2	shows	show	VBZ
work_v44vkh6bzramngvcrficy2no44	320	3	two	two	CD
work_v44vkh6bzramngvcrficy2no44	320	4	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	320	5	sentences	sentence	NNS
work_v44vkh6bzramngvcrficy2no44	320	6	taken	take	VBN
work_v44vkh6bzramngvcrficy2no44	320	7	from	from	IN
work_v44vkh6bzramngvcrficy2no44	320	8	eval09-nw	eval09-nw	NNP
work_v44vkh6bzramngvcrficy2no44	320	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	320	10	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	320	11	were	be	VBD
work_v44vkh6bzramngvcrficy2no44	320	12	erroneuously	erroneuously	RB
work_v44vkh6bzramngvcrficy2no44	320	13	reordered	reorder	VBN
work_v44vkh6bzramngvcrficy2no44	320	14	by	by	IN
work_v44vkh6bzramngvcrficy2no44	320	15	the	the	DT
work_v44vkh6bzramngvcrficy2no44	320	16	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	320	17	system	system	NN
work_v44vkh6bzramngvcrficy2no44	320	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	321	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	321	2	system	system	NN
work_v44vkh6bzramngvcrficy2no44	321	3	including	include	VBG
work_v44vkh6bzramngvcrficy2no44	321	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	321	5	WaW	WaW	NNP
work_v44vkh6bzramngvcrficy2no44	321	6	model	model	NN
work_v44vkh6bzramngvcrficy2no44	321	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	321	8	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	321	9	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	321	10	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	321	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	321	12	instead	instead	RB
work_v44vkh6bzramngvcrficy2no44	321	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	321	14	pro-	pro-	NN
work_v44vkh6bzramngvcrficy2no44	321	15	duced	duce	VBD
work_v44vkh6bzramngvcrficy2no44	321	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	321	17	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	321	18	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	321	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	322	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	322	2	first	first	JJ
work_v44vkh6bzramngvcrficy2no44	322	3	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	322	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	322	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	322	6	typical	typical	JJ
work_v44vkh6bzramngvcrficy2no44	322	7	example	example	NN
work_v44vkh6bzramngvcrficy2no44	322	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	322	9	VSO	VSO	NNP
work_v44vkh6bzramngvcrficy2no44	322	10	order	order	NN
work_v44vkh6bzramngvcrficy2no44	322	11	with	with	IN
work_v44vkh6bzramngvcrficy2no44	322	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	322	13	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	322	14	subject	subject	NN
work_v44vkh6bzramngvcrficy2no44	322	15	:	:	:
work_v44vkh6bzramngvcrficy2no44	322	16	while	while	IN
work_v44vkh6bzramngvcrficy2no44	322	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	322	18	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	322	19	system	system	NN
work_v44vkh6bzramngvcrficy2no44	322	20	left	leave	VBD
work_v44vkh6bzramngvcrficy2no44	322	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	322	22	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	322	23	in	in	IN
work_v44vkh6bzramngvcrficy2no44	322	24	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	322	25	Ara-	Ara-	NNP
work_v44vkh6bzramngvcrficy2no44	322	26	bic	bic	JJ
work_v44vkh6bzramngvcrficy2no44	322	27	position	position	NN
work_v44vkh6bzramngvcrficy2no44	322	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	322	29	producing	produce	VBG
work_v44vkh6bzramngvcrficy2no44	322	30	an	an	DT
work_v44vkh6bzramngvcrficy2no44	322	31	incomprehensible	incomprehensible	JJ
work_v44vkh6bzramngvcrficy2no44	322	32	trans-	trans-	JJ
work_v44vkh6bzramngvcrficy2no44	322	33	lation	lation	NN
work_v44vkh6bzramngvcrficy2no44	322	34	,	,	,
work_v44vkh6bzramngvcrficy2no44	322	35	the	the	DT
work_v44vkh6bzramngvcrficy2no44	322	36	new	new	JJ
work_v44vkh6bzramngvcrficy2no44	322	37	system	system	NN
work_v44vkh6bzramngvcrficy2no44	322	38	placed	place	VBD
work_v44vkh6bzramngvcrficy2no44	322	39	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	322	40	rightly	rightly	RB
work_v44vkh6bzramngvcrficy2no44	322	41	between	between	IN
work_v44vkh6bzramngvcrficy2no44	322	42	the	the	DT
work_v44vkh6bzramngvcrficy2no44	322	43	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	322	44	subject	subject	NN
work_v44vkh6bzramngvcrficy2no44	322	45	and	and	CC
work_v44vkh6bzramngvcrficy2no44	322	46	object	object	NN
work_v44vkh6bzramngvcrficy2no44	322	47	.	.	.
work_v44vkh6bzramngvcrficy2no44	323	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	323	2	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	323	3	involved	involve	VBD
work_v44vkh6bzramngvcrficy2no44	323	4	two	two	CD
work_v44vkh6bzramngvcrficy2no44	323	5	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	323	6	jumps	jump	NNS
work_v44vkh6bzramngvcrficy2no44	323	7	:	:	:
work_v44vkh6bzramngvcrficy2no44	323	8	one	one	CD
work_v44vkh6bzramngvcrficy2no44	323	9	with	with	IN
work_v44vkh6bzramngvcrficy2no44	323	10	D=9	D=9	NNP
work_v44vkh6bzramngvcrficy2no44	323	11	backward	backward	JJ
work_v44vkh6bzramngvcrficy2no44	323	12	and	and	CC
work_v44vkh6bzramngvcrficy2no44	323	13	one	one	CD
work_v44vkh6bzramngvcrficy2no44	323	14	with	with	IN
work_v44vkh6bzramngvcrficy2no44	323	15	D=8	D=8	NNP
work_v44vkh6bzramngvcrficy2no44	323	16	forward	forward	RB
work_v44vkh6bzramngvcrficy2no44	323	17	.	.	.
work_v44vkh6bzramngvcrficy2no44	324	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	324	2	second	second	JJ
work_v44vkh6bzramngvcrficy2no44	324	3	sentence	sentence	NN
work_v44vkh6bzramngvcrficy2no44	324	4	displays	display	VBZ
work_v44vkh6bzramngvcrficy2no44	324	5	another	another	DT
work_v44vkh6bzramngvcrficy2no44	324	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	324	7	less	less	RBR
work_v44vkh6bzramngvcrficy2no44	324	8	com-	com-	NN
work_v44vkh6bzramngvcrficy2no44	324	9	mon	mon	NN
work_v44vkh6bzramngvcrficy2no44	324	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	324	11	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	324	12	construction	construction	NN
work_v44vkh6bzramngvcrficy2no44	324	13	:	:	:
work_v44vkh6bzramngvcrficy2no44	324	14	namely	namely	RB
work_v44vkh6bzramngvcrficy2no44	324	15	VOS	VOS	NNP
work_v44vkh6bzramngvcrficy2no44	324	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	324	17	with	with	IN
work_v44vkh6bzramngvcrficy2no44	324	18	a	a	DT
work_v44vkh6bzramngvcrficy2no44	324	19	per-	per-	JJ
work_v44vkh6bzramngvcrficy2no44	324	20	sonal	sonal	JJ
work_v44vkh6bzramngvcrficy2no44	324	21	pronoun	pronoun	NN
work_v44vkh6bzramngvcrficy2no44	324	22	object	object	NN
work_v44vkh6bzramngvcrficy2no44	324	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	325	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	325	2	this	this	DT
work_v44vkh6bzramngvcrficy2no44	325	3	case	case	NN
work_v44vkh6bzramngvcrficy2no44	325	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	325	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	325	6	backward	backward	JJ
work_v44vkh6bzramngvcrficy2no44	325	7	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	325	8	with	with	IN
work_v44vkh6bzramngvcrficy2no44	325	9	D=10	d=10	NN
work_v44vkh6bzramngvcrficy2no44	325	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	325	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	325	12	forward	forward	JJ
work_v44vkh6bzramngvcrficy2no44	325	13	jump	jump	NN
work_v44vkh6bzramngvcrficy2no44	325	14	with	with	IN
work_v44vkh6bzramngvcrficy2no44	325	15	D=8	D=8	NNP
work_v44vkh6bzramngvcrficy2no44	325	16	were	be	VBD
work_v44vkh6bzramngvcrficy2no44	325	17	nec-	nec-	RB
work_v44vkh6bzramngvcrficy2no44	325	18	essary	essary	JJ
work_v44vkh6bzramngvcrficy2no44	325	19	to	to	TO
work_v44vkh6bzramngvcrficy2no44	325	20	achieve	achieve	VB
work_v44vkh6bzramngvcrficy2no44	325	21	the	the	DT
work_v44vkh6bzramngvcrficy2no44	325	22	correct	correct	JJ
work_v44vkh6bzramngvcrficy2no44	325	23	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	325	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	326	1	13Statistics	13statistics	CD
work_v44vkh6bzramngvcrficy2no44	326	2	computed	compute	VBD
work_v44vkh6bzramngvcrficy2no44	326	3	on	on	IN
work_v44vkh6bzramngvcrficy2no44	326	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	326	5	medium	medium	JJ
work_v44vkh6bzramngvcrficy2no44	326	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	326	7	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	326	8	system	system	NN
work_v44vkh6bzramngvcrficy2no44	326	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	327	1	6	6	CD
work_v44vkh6bzramngvcrficy2no44	327	2	Conclusions	Conclusions	NNPS
work_v44vkh6bzramngvcrficy2no44	327	3	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	327	4	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	327	5	trained	train	VBN
work_v44vkh6bzramngvcrficy2no44	327	6	a	a	DT
work_v44vkh6bzramngvcrficy2no44	327	7	discriminative	discriminative	JJ
work_v44vkh6bzramngvcrficy2no44	327	8	model	model	NN
work_v44vkh6bzramngvcrficy2no44	327	9	to	to	TO
work_v44vkh6bzramngvcrficy2no44	327	10	predict	predict	VB
work_v44vkh6bzramngvcrficy2no44	327	11	likely	likely	JJ
work_v44vkh6bzramngvcrficy2no44	327	12	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	327	13	steps	step	NNS
work_v44vkh6bzramngvcrficy2no44	327	14	in	in	IN
work_v44vkh6bzramngvcrficy2no44	327	15	a	a	DT
work_v44vkh6bzramngvcrficy2no44	327	16	way	way	NN
work_v44vkh6bzramngvcrficy2no44	327	17	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	327	18	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	327	19	complemen-	complemen-	NN
work_v44vkh6bzramngvcrficy2no44	327	20	tary	tary	NN
work_v44vkh6bzramngvcrficy2no44	327	21	to	to	IN
work_v44vkh6bzramngvcrficy2no44	327	22	state	state	NN
work_v44vkh6bzramngvcrficy2no44	327	23	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	327	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	327	25	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	327	26	the	the	DT
work_v44vkh6bzramngvcrficy2no44	327	27	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	327	28	art	art	NN
work_v44vkh6bzramngvcrficy2no44	327	29	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	327	30	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	327	31	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	327	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	328	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	328	2	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	328	3	effectively	effectively	RB
work_v44vkh6bzramngvcrficy2no44	328	4	integrated	integrate	VBN
work_v44vkh6bzramngvcrficy2no44	328	5	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	328	6	into	into	IN
work_v44vkh6bzramngvcrficy2no44	328	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	328	8	PSMT	PSMT	NNP
work_v44vkh6bzramngvcrficy2no44	328	9	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	328	10	as	as	IN
work_v44vkh6bzramngvcrficy2no44	328	11	additional	additional	JJ
work_v44vkh6bzramngvcrficy2no44	328	12	feature	feature	NN
work_v44vkh6bzramngvcrficy2no44	328	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	328	14	ensuring	ensure	VBG
work_v44vkh6bzramngvcrficy2no44	328	15	that	that	IN
work_v44vkh6bzramngvcrficy2no44	328	16	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	328	17	total	total	JJ
work_v44vkh6bzramngvcrficy2no44	328	18	score	score	NN
work_v44vkh6bzramngvcrficy2no44	328	19	over	over	IN
work_v44vkh6bzramngvcrficy2no44	328	20	a	a	DT
work_v44vkh6bzramngvcrficy2no44	328	21	complete	complete	JJ
work_v44vkh6bzramngvcrficy2no44	328	22	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	328	23	hypothesis	hypothesis	NN
work_v44vkh6bzramngvcrficy2no44	328	24	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	328	25	consistent	consistent	JJ
work_v44vkh6bzramngvcrficy2no44	328	26	across	across	IN
work_v44vkh6bzramngvcrficy2no44	328	27	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	328	28	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	328	29	segmentations	segmentation	NNS
work_v44vkh6bzramngvcrficy2no44	328	30	.	.	.
work_v44vkh6bzramngvcrficy2no44	329	1	Lastly	lastly	RB
work_v44vkh6bzramngvcrficy2no44	329	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	329	3	we	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	329	4	have	have	VBP
work_v44vkh6bzramngvcrficy2no44	329	5	pro-	pro-	NN
work_v44vkh6bzramngvcrficy2no44	329	6	posed	pose	VBN
work_v44vkh6bzramngvcrficy2no44	329	7	early	early	JJ
work_v44vkh6bzramngvcrficy2no44	329	8	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	329	9	pruning	pruning	NN
work_v44vkh6bzramngvcrficy2no44	329	10	as	as	IN
work_v44vkh6bzramngvcrficy2no44	329	11	a	a	DT
work_v44vkh6bzramngvcrficy2no44	329	12	novel	novel	JJ
work_v44vkh6bzramngvcrficy2no44	329	13	method	method	NN
work_v44vkh6bzramngvcrficy2no44	329	14	to	to	TO
work_v44vkh6bzramngvcrficy2no44	329	15	dynamically	dynamically	RB
work_v44vkh6bzramngvcrficy2no44	329	16	shape	shape	VB
work_v44vkh6bzramngvcrficy2no44	329	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	329	18	input	input	NN
work_v44vkh6bzramngvcrficy2no44	329	19	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	329	20	space	space	NN
work_v44vkh6bzramngvcrficy2no44	329	21	and	and	CC
work_v44vkh6bzramngvcrficy2no44	329	22	capture	capture	VB
work_v44vkh6bzramngvcrficy2no44	329	23	long	long	JJ
work_v44vkh6bzramngvcrficy2no44	329	24	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	329	25	range	range	NN
work_v44vkh6bzramngvcrficy2no44	329	26	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	329	27	phenomena	phenomenon	NNS
work_v44vkh6bzramngvcrficy2no44	329	28	that	that	WDT
work_v44vkh6bzramngvcrficy2no44	329	29	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	329	30	often	often	RB
work_v44vkh6bzramngvcrficy2no44	329	31	critical	critical	JJ
work_v44vkh6bzramngvcrficy2no44	329	32	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	329	33	translating	translate	VBG
work_v44vkh6bzramngvcrficy2no44	329	34	between	between	IN
work_v44vkh6bzramngvcrficy2no44	329	35	languages	language	NNS
work_v44vkh6bzramngvcrficy2no44	329	36	with	with	IN
work_v44vkh6bzramngvcrficy2no44	329	37	different	different	JJ
work_v44vkh6bzramngvcrficy2no44	329	38	syntactic	syntactic	JJ
work_v44vkh6bzramngvcrficy2no44	329	39	structures	structure	NNS
work_v44vkh6bzramngvcrficy2no44	329	40	.	.	.
work_v44vkh6bzramngvcrficy2no44	330	1	Evaluated	evaluate	VBN
work_v44vkh6bzramngvcrficy2no44	330	2	on	on	IN
work_v44vkh6bzramngvcrficy2no44	330	3	a	a	DT
work_v44vkh6bzramngvcrficy2no44	330	4	popular	popular	JJ
work_v44vkh6bzramngvcrficy2no44	330	5	Arabic	arabic	JJ
work_v44vkh6bzramngvcrficy2no44	330	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	330	7	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	330	8	news	news	NN
work_v44vkh6bzramngvcrficy2no44	330	9	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	330	10	task	task	NN
work_v44vkh6bzramngvcrficy2no44	330	11	against	against	IN
work_v44vkh6bzramngvcrficy2no44	330	12	a	a	DT
work_v44vkh6bzramngvcrficy2no44	330	13	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	330	14	baseline	baseline	NN
work_v44vkh6bzramngvcrficy2no44	330	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	330	16	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	330	17	ap-	ap-	CD
work_v44vkh6bzramngvcrficy2no44	330	18	proach	proach	NN
work_v44vkh6bzramngvcrficy2no44	330	19	leads	lead	VBZ
work_v44vkh6bzramngvcrficy2no44	330	20	to	to	IN
work_v44vkh6bzramngvcrficy2no44	330	21	similar	similar	JJ
work_v44vkh6bzramngvcrficy2no44	330	22	or	or	CC
work_v44vkh6bzramngvcrficy2no44	330	23	even	even	RB
work_v44vkh6bzramngvcrficy2no44	330	24	higher	high	JJR
work_v44vkh6bzramngvcrficy2no44	330	25	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	330	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	330	27	ME-	ME-	NNP
work_v44vkh6bzramngvcrficy2no44	330	28	TEOR	TEOR	NNP
work_v44vkh6bzramngvcrficy2no44	330	29	and	and	CC
work_v44vkh6bzramngvcrficy2no44	330	30	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	330	31	scores	score	NNS
work_v44vkh6bzramngvcrficy2no44	330	32	at	at	IN
work_v44vkh6bzramngvcrficy2no44	330	33	a	a	DT
work_v44vkh6bzramngvcrficy2no44	330	34	very	very	RB
work_v44vkh6bzramngvcrficy2no44	330	35	high	high	JJ
work_v44vkh6bzramngvcrficy2no44	330	36	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	330	37	limit	limit	NN
work_v44vkh6bzramngvcrficy2no44	330	38	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	330	39	18	18	CD
work_v44vkh6bzramngvcrficy2no44	330	40	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	330	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	330	42	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	330	43	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	330	44	by	by	IN
work_v44vkh6bzramngvcrficy2no44	330	45	itself	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	330	46	an	an	DT
work_v44vkh6bzramngvcrficy2no44	330	47	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	330	48	achievement	achievement	NN
work_v44vkh6bzramngvcrficy2no44	330	49	.	.	.
work_v44vkh6bzramngvcrficy2no44	331	1	At	at	IN
work_v44vkh6bzramngvcrficy2no44	331	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	331	3	same	same	JJ
work_v44vkh6bzramngvcrficy2no44	331	4	time	time	NN
work_v44vkh6bzramngvcrficy2no44	331	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	331	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	331	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	331	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	331	9	verbs	verb	NNS
work_v44vkh6bzramngvcrficy2no44	331	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	331	11	measured	measure	VBN
work_v44vkh6bzramngvcrficy2no44	331	12	with	with	IN
work_v44vkh6bzramngvcrficy2no44	331	13	a	a	DT
work_v44vkh6bzramngvcrficy2no44	331	14	novel	novel	JJ
work_v44vkh6bzramngvcrficy2no44	331	15	version	version	NN
work_v44vkh6bzramngvcrficy2no44	331	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	331	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	331	18	KRS	KRS	NNP
work_v44vkh6bzramngvcrficy2no44	331	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	331	20	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	331	21	consistently	consistently	RB
work_v44vkh6bzramngvcrficy2no44	331	22	im-	im-	XX
work_v44vkh6bzramngvcrficy2no44	331	23	proved	prove	VBN
work_v44vkh6bzramngvcrficy2no44	331	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	331	25	while	while	IN
work_v44vkh6bzramngvcrficy2no44	331	26	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	331	27	gets	get	VBZ
work_v44vkh6bzramngvcrficy2no44	331	28	significantly	significantly	RB
work_v44vkh6bzramngvcrficy2no44	331	29	faster	fast	RBR
work_v44vkh6bzramngvcrficy2no44	331	30	.	.	.
work_v44vkh6bzramngvcrficy2no44	332	1	The	the	DT
work_v44vkh6bzramngvcrficy2no44	332	2	improvements	improvement	NNS
work_v44vkh6bzramngvcrficy2no44	332	3	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	332	4	also	also	RB
work_v44vkh6bzramngvcrficy2no44	332	5	confirmed	confirm	VBN
work_v44vkh6bzramngvcrficy2no44	332	6	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	332	7	a	a	DT
work_v44vkh6bzramngvcrficy2no44	332	8	very	very	RB
work_v44vkh6bzramngvcrficy2no44	332	9	large	large	JJ
work_v44vkh6bzramngvcrficy2no44	332	10	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	332	11	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	332	12	used	use	VBN
work_v44vkh6bzramngvcrficy2no44	332	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	332	14	the	the	DT
work_v44vkh6bzramngvcrficy2no44	332	15	decoder	decoder	NN
work_v44vkh6bzramngvcrficy2no44	332	16	’s	’s	POS
work_v44vkh6bzramngvcrficy2no44	332	17	beam	beam	NN
work_v44vkh6bzramngvcrficy2no44	332	18	size	size	NN
work_v44vkh6bzramngvcrficy2no44	332	19	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	332	20	dou-	dou-	RB
work_v44vkh6bzramngvcrficy2no44	332	21	bled	bleed	VBN
work_v44vkh6bzramngvcrficy2no44	332	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	332	23	which	which	WDT
work_v44vkh6bzramngvcrficy2no44	332	24	shows	show	VBZ
work_v44vkh6bzramngvcrficy2no44	332	25	that	that	IN
work_v44vkh6bzramngvcrficy2no44	332	26	our	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	332	27	method	method	NN
work_v44vkh6bzramngvcrficy2no44	332	28	reduces	reduce	VBZ
work_v44vkh6bzramngvcrficy2no44	332	29	not	not	RB
work_v44vkh6bzramngvcrficy2no44	332	30	only	only	RB
work_v44vkh6bzramngvcrficy2no44	332	31	search	search	NN
work_v44vkh6bzramngvcrficy2no44	332	32	errors	error	NNS
work_v44vkh6bzramngvcrficy2no44	332	33	but	but	CC
work_v44vkh6bzramngvcrficy2no44	332	34	also	also	RB
work_v44vkh6bzramngvcrficy2no44	332	35	model	model	NN
work_v44vkh6bzramngvcrficy2no44	332	36	errors	error	NNS
work_v44vkh6bzramngvcrficy2no44	332	37	even	even	RB
work_v44vkh6bzramngvcrficy2no44	332	38	when	when	WRB
work_v44vkh6bzramngvcrficy2no44	332	39	base-	base-	NN
work_v44vkh6bzramngvcrficy2no44	332	40	line	line	NN
work_v44vkh6bzramngvcrficy2no44	332	41	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	332	42	are	be	VBP
work_v44vkh6bzramngvcrficy2no44	332	43	very	very	RB
work_v44vkh6bzramngvcrficy2no44	332	44	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	332	45	.	.	.
work_v44vkh6bzramngvcrficy2no44	333	1	Word	word	NN
work_v44vkh6bzramngvcrficy2no44	333	2	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	333	3	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	333	4	probably	probably	RB
work_v44vkh6bzramngvcrficy2no44	333	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	333	6	most	most	RBS
work_v44vkh6bzramngvcrficy2no44	333	7	difficult	difficult	JJ
work_v44vkh6bzramngvcrficy2no44	333	8	as-	as-	NNS
work_v44vkh6bzramngvcrficy2no44	333	9	pect	pect	NN
work_v44vkh6bzramngvcrficy2no44	333	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	333	11	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	333	12	and	and	CC
work_v44vkh6bzramngvcrficy2no44	333	13	an	an	DT
work_v44vkh6bzramngvcrficy2no44	333	14	important	important	JJ
work_v44vkh6bzramngvcrficy2no44	333	15	factor	factor	NN
work_v44vkh6bzramngvcrficy2no44	333	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	333	17	both	both	CC
work_v44vkh6bzramngvcrficy2no44	333	18	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	333	19	qual-	qual-	NN
work_v44vkh6bzramngvcrficy2no44	333	20	ity	ity	NN
work_v44vkh6bzramngvcrficy2no44	333	21	and	and	CC
work_v44vkh6bzramngvcrficy2no44	333	22	efficiency	efficiency	NN
work_v44vkh6bzramngvcrficy2no44	333	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	334	1	Given	give	VBN
work_v44vkh6bzramngvcrficy2no44	334	2	its	-PRON-	PRP$
work_v44vkh6bzramngvcrficy2no44	334	3	strong	strong	JJ
work_v44vkh6bzramngvcrficy2no44	334	4	interaction	interaction	NN
work_v44vkh6bzramngvcrficy2no44	334	5	with	with	IN
work_v44vkh6bzramngvcrficy2no44	334	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	334	7	other	other	JJ
work_v44vkh6bzramngvcrficy2no44	334	8	aspects	aspect	NNS
work_v44vkh6bzramngvcrficy2no44	334	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	334	10	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	334	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	334	12	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	334	13	appears	appear	VBZ
work_v44vkh6bzramngvcrficy2no44	334	14	natural	natural	JJ
work_v44vkh6bzramngvcrficy2no44	334	15	to	to	TO
work_v44vkh6bzramngvcrficy2no44	334	16	solve	solve	VB
work_v44vkh6bzramngvcrficy2no44	334	17	337	337	CD
work_v44vkh6bzramngvcrficy2no44	334	18	word	word	NN
work_v44vkh6bzramngvcrficy2no44	334	19	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	334	20	during	during	IN
work_v44vkh6bzramngvcrficy2no44	334	21	decoding	decoding	NN
work_v44vkh6bzramngvcrficy2no44	334	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	334	23	rather	rather	RB
work_v44vkh6bzramngvcrficy2no44	334	24	than	than	IN
work_v44vkh6bzramngvcrficy2no44	334	25	before	before	IN
work_v44vkh6bzramngvcrficy2no44	334	26	or	or	CC
work_v44vkh6bzramngvcrficy2no44	334	27	after	after	IN
work_v44vkh6bzramngvcrficy2no44	334	28	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	334	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	335	1	To	to	IN
work_v44vkh6bzramngvcrficy2no44	335	2	date	date	NN
work_v44vkh6bzramngvcrficy2no44	335	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	335	4	however	however	RB
work_v44vkh6bzramngvcrficy2no44	335	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	335	6	this	this	DT
work_v44vkh6bzramngvcrficy2no44	335	7	objective	objective	NN
work_v44vkh6bzramngvcrficy2no44	335	8	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	335	9	only	only	RB
work_v44vkh6bzramngvcrficy2no44	335	10	partially	partially	RB
work_v44vkh6bzramngvcrficy2no44	335	11	achieved	achieve	VBN
work_v44vkh6bzramngvcrficy2no44	335	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	336	1	We	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	336	2	believe	believe	VBP
work_v44vkh6bzramngvcrficy2no44	336	3	there	there	EX
work_v44vkh6bzramngvcrficy2no44	336	4	is	be	VBZ
work_v44vkh6bzramngvcrficy2no44	336	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	336	6	promising	promising	JJ
work_v44vkh6bzramngvcrficy2no44	336	7	way	way	NN
work_v44vkh6bzramngvcrficy2no44	336	8	to	to	TO
work_v44vkh6bzramngvcrficy2no44	336	9	go	go	VB
work_v44vkh6bzramngvcrficy2no44	336	10	between	between	IN
work_v44vkh6bzramngvcrficy2no44	336	11	fully	fully	RB
work_v44vkh6bzramngvcrficy2no44	336	12	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	336	13	integrated	integrate	VBN
work_v44vkh6bzramngvcrficy2no44	336	14	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	336	15	mod-	mod-	NN
work_v44vkh6bzramngvcrficy2no44	336	16	els	els	NNP
work_v44vkh6bzramngvcrficy2no44	336	17	and	and	CC
work_v44vkh6bzramngvcrficy2no44	336	18	monolingual	monolingual	JJ
work_v44vkh6bzramngvcrficy2no44	336	19	pre	pre	JJ
work_v44vkh6bzramngvcrficy2no44	336	20	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	336	21	ordering	ordering	JJ
work_v44vkh6bzramngvcrficy2no44	336	22	methods	method	NNS
work_v44vkh6bzramngvcrficy2no44	336	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	337	1	This	this	DT
work_v44vkh6bzramngvcrficy2no44	337	2	work	work	NN
work_v44vkh6bzramngvcrficy2no44	337	3	has	have	VBZ
work_v44vkh6bzramngvcrficy2no44	337	4	started	start	VBN
work_v44vkh6bzramngvcrficy2no44	337	5	to	to	TO
work_v44vkh6bzramngvcrficy2no44	337	6	explore	explore	VB
work_v44vkh6bzramngvcrficy2no44	337	7	it	-PRON-	PRP
work_v44vkh6bzramngvcrficy2no44	337	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	338	1	Aknowledgments	aknowledgment	NNS
work_v44vkh6bzramngvcrficy2no44	338	2	This	this	DT
work_v44vkh6bzramngvcrficy2no44	338	3	work	work	NN
work_v44vkh6bzramngvcrficy2no44	338	4	was	be	VBD
work_v44vkh6bzramngvcrficy2no44	338	5	partially	partially	RB
work_v44vkh6bzramngvcrficy2no44	338	6	funded	fund	VBN
work_v44vkh6bzramngvcrficy2no44	338	7	by	by	IN
work_v44vkh6bzramngvcrficy2no44	338	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	338	9	European	European	NNP
work_v44vkh6bzramngvcrficy2no44	338	10	Union	Union	NNP
work_v44vkh6bzramngvcrficy2no44	338	11	FP7	FP7	NNP
work_v44vkh6bzramngvcrficy2no44	338	12	grant	grant	NN
work_v44vkh6bzramngvcrficy2no44	338	13	agreement	agreement	NN
work_v44vkh6bzramngvcrficy2no44	338	14	287658	287658	CD
work_v44vkh6bzramngvcrficy2no44	338	15	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	338	16	EU	EU	NNP
work_v44vkh6bzramngvcrficy2no44	338	17	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	338	18	BRIDGE	BRIDGE	NNP
work_v44vkh6bzramngvcrficy2no44	338	19	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	338	20	.	.	.
work_v44vkh6bzramngvcrficy2no44	339	1	References	reference	NNS
work_v44vkh6bzramngvcrficy2no44	339	2	Yaser	Yaser	NNP
work_v44vkh6bzramngvcrficy2no44	339	3	Al	Al	NNP
work_v44vkh6bzramngvcrficy2no44	339	4	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	339	5	Onaizan	Onaizan	NNP
work_v44vkh6bzramngvcrficy2no44	339	6	and	and	CC
work_v44vkh6bzramngvcrficy2no44	339	7	Kishore	Kishore	NNP
work_v44vkh6bzramngvcrficy2no44	339	8	Papineni	Papineni	NNP
work_v44vkh6bzramngvcrficy2no44	339	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	340	1	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	340	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	341	1	Distor-	Distor-	NNP
work_v44vkh6bzramngvcrficy2no44	341	2	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	341	3	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	341	4	for	for	IN
work_v44vkh6bzramngvcrficy2no44	341	5	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	341	6	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	341	7	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	341	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	342	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	342	2	Pro-	Pro-	NNP
work_v44vkh6bzramngvcrficy2no44	342	3	ceedings	ceeding	NNS
work_v44vkh6bzramngvcrficy2no44	342	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	342	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	342	6	21st	21st	JJ
work_v44vkh6bzramngvcrficy2no44	342	7	International	International	NNP
work_v44vkh6bzramngvcrficy2no44	342	8	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	342	9	on	on	IN
work_v44vkh6bzramngvcrficy2no44	342	10	Com-	Com-	NNP
work_v44vkh6bzramngvcrficy2no44	342	11	putational	putational	JJ
work_v44vkh6bzramngvcrficy2no44	342	12	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	342	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	342	14	44th	44th	JJ
work_v44vkh6bzramngvcrficy2no44	342	15	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	342	16	Meeting	Meeting	NNP
work_v44vkh6bzramngvcrficy2no44	342	17	of	of	IN
work_v44vkh6bzramngvcrficy2no44	342	18	the	the	DT
work_v44vkh6bzramngvcrficy2no44	342	19	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	342	20	for	for	IN
work_v44vkh6bzramngvcrficy2no44	342	21	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	342	22	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	342	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	342	24	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	342	25	529–536	529–536	CD
work_v44vkh6bzramngvcrficy2no44	342	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	342	27	Sydney	Sydney	NNP
work_v44vkh6bzramngvcrficy2no44	342	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	342	29	Australia	Australia	NNP
work_v44vkh6bzramngvcrficy2no44	342	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	342	31	July	July	NNP
work_v44vkh6bzramngvcrficy2no44	342	32	.	.	.
work_v44vkh6bzramngvcrficy2no44	343	1	Jacob	Jacob	NNP
work_v44vkh6bzramngvcrficy2no44	343	2	Andreas	Andreas	NNP
work_v44vkh6bzramngvcrficy2no44	343	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	343	4	Nizar	Nizar	NNP
work_v44vkh6bzramngvcrficy2no44	343	5	Habash	Habash	NNP
work_v44vkh6bzramngvcrficy2no44	343	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	343	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	343	8	Owen	Owen	NNP
work_v44vkh6bzramngvcrficy2no44	343	9	Rambow	Rambow	NNP
work_v44vkh6bzramngvcrficy2no44	343	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	344	1	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	344	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	345	1	Fuzzy	fuzzy	JJ
work_v44vkh6bzramngvcrficy2no44	345	2	syntactic	syntactic	JJ
work_v44vkh6bzramngvcrficy2no44	345	3	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	345	4	for	for	IN
work_v44vkh6bzramngvcrficy2no44	345	5	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	345	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	345	7	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	345	8	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	345	9	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	345	10	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	345	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	346	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	346	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	346	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	346	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	346	5	Sixth	sixth	JJ
work_v44vkh6bzramngvcrficy2no44	346	6	Work-	work-	NN
work_v44vkh6bzramngvcrficy2no44	346	7	shop	shop	NN
work_v44vkh6bzramngvcrficy2no44	346	8	on	on	IN
work_v44vkh6bzramngvcrficy2no44	346	9	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	346	10	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	346	11	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	346	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	346	13	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	346	14	227	227	CD
work_v44vkh6bzramngvcrficy2no44	346	15	–	–	:
work_v44vkh6bzramngvcrficy2no44	346	16	236	236	CD
work_v44vkh6bzramngvcrficy2no44	346	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	346	18	Edinburgh	Edinburgh	NNP
work_v44vkh6bzramngvcrficy2no44	346	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	346	20	Scotland	Scotland	NNP
work_v44vkh6bzramngvcrficy2no44	346	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	346	22	July	July	NNP
work_v44vkh6bzramngvcrficy2no44	346	23	.	.	.
work_v44vkh6bzramngvcrficy2no44	347	1	Satanjeev	Satanjeev	NNP
work_v44vkh6bzramngvcrficy2no44	347	2	Banerjee	Banerjee	NNP
work_v44vkh6bzramngvcrficy2no44	347	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	347	4	Alon	Alon	NNP
work_v44vkh6bzramngvcrficy2no44	347	5	Lavie	Lavie	NNP
work_v44vkh6bzramngvcrficy2no44	347	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	348	1	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	348	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	349	1	METEOR	METEOR	NNP
work_v44vkh6bzramngvcrficy2no44	349	2	:	:	:
work_v44vkh6bzramngvcrficy2no44	349	3	An	an	DT
work_v44vkh6bzramngvcrficy2no44	349	4	automatic	automatic	JJ
work_v44vkh6bzramngvcrficy2no44	349	5	metric	metric	NN
work_v44vkh6bzramngvcrficy2no44	349	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	349	7	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	349	8	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	349	9	with	with	IN
work_v44vkh6bzramngvcrficy2no44	349	10	improved	improved	JJ
work_v44vkh6bzramngvcrficy2no44	349	11	correlation	correlation	NN
work_v44vkh6bzramngvcrficy2no44	349	12	with	with	IN
work_v44vkh6bzramngvcrficy2no44	349	13	human	human	JJ
work_v44vkh6bzramngvcrficy2no44	349	14	judgments	judgment	NNS
work_v44vkh6bzramngvcrficy2no44	349	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	350	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	350	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	350	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	350	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	350	5	ACL	ACL	NNP
work_v44vkh6bzramngvcrficy2no44	350	6	Workshop	Workshop	NNP
work_v44vkh6bzramngvcrficy2no44	350	7	on	on	IN
work_v44vkh6bzramngvcrficy2no44	350	8	Intrinsic	Intrinsic	NNP
work_v44vkh6bzramngvcrficy2no44	350	9	and	and	CC
work_v44vkh6bzramngvcrficy2no44	350	10	Extrinsic	Extrinsic	NNP
work_v44vkh6bzramngvcrficy2no44	350	11	Evalu-	Evalu-	NNP
work_v44vkh6bzramngvcrficy2no44	350	12	ation	ation	NN
work_v44vkh6bzramngvcrficy2no44	350	13	Measures	measure	NNS
work_v44vkh6bzramngvcrficy2no44	350	14	for	for	IN
work_v44vkh6bzramngvcrficy2no44	350	15	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	350	16	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	350	17	and/or	and/or	CC
work_v44vkh6bzramngvcrficy2no44	350	18	Sum-	Sum-	NNP
work_v44vkh6bzramngvcrficy2no44	350	19	marization	marization	NN
work_v44vkh6bzramngvcrficy2no44	350	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	350	21	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	350	22	65–72	65–72	CD
work_v44vkh6bzramngvcrficy2no44	350	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	350	24	Ann	Ann	NNP
work_v44vkh6bzramngvcrficy2no44	350	25	Arbor	Arbor	NNP
work_v44vkh6bzramngvcrficy2no44	350	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	350	27	Michigan	Michigan	NNP
work_v44vkh6bzramngvcrficy2no44	350	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	350	29	June	June	NNP
work_v44vkh6bzramngvcrficy2no44	350	30	.	.	.
work_v44vkh6bzramngvcrficy2no44	351	1	A.	A.	NNP
work_v44vkh6bzramngvcrficy2no44	351	2	L.	L.	NNP
work_v44vkh6bzramngvcrficy2no44	351	3	Berger	Berger	NNP
work_v44vkh6bzramngvcrficy2no44	351	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	351	5	P.	P.	NNP
work_v44vkh6bzramngvcrficy2no44	351	6	F.	F.	NNP
work_v44vkh6bzramngvcrficy2no44	351	7	Brown	Brown	NNP
work_v44vkh6bzramngvcrficy2no44	351	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	351	9	S.	S.	NNP
work_v44vkh6bzramngvcrficy2no44	351	10	A.	a.	NN
work_v44vkh6bzramngvcrficy2no44	351	11	Della	Della	NNP
work_v44vkh6bzramngvcrficy2no44	351	12	Pietra	Pietra	NNP
work_v44vkh6bzramngvcrficy2no44	351	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	351	14	V.	V.	NNP
work_v44vkh6bzramngvcrficy2no44	351	15	J.	J.	NNP
work_v44vkh6bzramngvcrficy2no44	351	16	Della	Della	NNP
work_v44vkh6bzramngvcrficy2no44	351	17	Pietra	Pietra	NNP
work_v44vkh6bzramngvcrficy2no44	351	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	351	19	J.	J.	NNP
work_v44vkh6bzramngvcrficy2no44	351	20	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	351	21	Gillett	Gillett	NNP
work_v44vkh6bzramngvcrficy2no44	351	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	351	23	A.	A.	NNP
work_v44vkh6bzramngvcrficy2no44	351	24	S.	S.	NNP
work_v44vkh6bzramngvcrficy2no44	351	25	Kehler	Kehler	NNP
work_v44vkh6bzramngvcrficy2no44	351	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	351	27	and	and	CC
work_v44vkh6bzramngvcrficy2no44	351	28	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	351	29	L.	L.	NNP
work_v44vkh6bzramngvcrficy2no44	351	30	Mercer	Mercer	NNP
work_v44vkh6bzramngvcrficy2no44	351	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	352	1	1996	1996	CD
work_v44vkh6bzramngvcrficy2no44	352	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	353	1	Language	language	NN
work_v44vkh6bzramngvcrficy2no44	353	2	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	353	3	apparatus	apparatus	NN
work_v44vkh6bzramngvcrficy2no44	353	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	353	5	method	method	NN
work_v44vkh6bzramngvcrficy2no44	353	6	of	of	IN
work_v44vkh6bzramngvcrficy2no44	353	7	using	use	VBG
work_v44vkh6bzramngvcrficy2no44	353	8	context	context	NN
work_v44vkh6bzramngvcrficy2no44	353	9	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	353	10	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	353	11	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	353	12	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	353	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	354	1	United	United	NNP
work_v44vkh6bzramngvcrficy2no44	354	2	States	States	NNP
work_v44vkh6bzramngvcrficy2no44	354	3	Patent	Patent	NNP
work_v44vkh6bzramngvcrficy2no44	354	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	354	5	No	no	UH
work_v44vkh6bzramngvcrficy2no44	354	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	355	1	5510981	5510981	CD
work_v44vkh6bzramngvcrficy2no44	355	2	,	,	,
work_v44vkh6bzramngvcrficy2no44	355	3	Apr.	April	NNP
work_v44vkh6bzramngvcrficy2no44	356	1	Alexandra	Alexandra	NNP
work_v44vkh6bzramngvcrficy2no44	356	2	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	356	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	356	4	Phil	Phil	NNP
work_v44vkh6bzramngvcrficy2no44	356	5	Blunsom	Blunsom	NNP
work_v44vkh6bzramngvcrficy2no44	356	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	356	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	356	8	Miles	Miles	NNP
work_v44vkh6bzramngvcrficy2no44	356	9	Osborne	Osborne	NNP
work_v44vkh6bzramngvcrficy2no44	356	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	357	1	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	357	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	358	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	358	2	quantitative	quantitative	JJ
work_v44vkh6bzramngvcrficy2no44	358	3	analysis	analysis	NN
work_v44vkh6bzramngvcrficy2no44	358	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	358	5	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	358	6	phenom-	phenom-	NNP
work_v44vkh6bzramngvcrficy2no44	358	7	ena	ena	NNP
work_v44vkh6bzramngvcrficy2no44	358	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	359	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	359	2	StatMT	StatMT	NNP
work_v44vkh6bzramngvcrficy2no44	359	3	’	’	POS
work_v44vkh6bzramngvcrficy2no44	359	4	09	09	CD
work_v44vkh6bzramngvcrficy2no44	359	5	:	:	:
work_v44vkh6bzramngvcrficy2no44	359	6	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	359	7	of	of	IN
work_v44vkh6bzramngvcrficy2no44	359	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	359	9	Fourth	Fourth	NNP
work_v44vkh6bzramngvcrficy2no44	359	10	Work-	Work-	NNP
work_v44vkh6bzramngvcrficy2no44	359	11	shop	shop	NN
work_v44vkh6bzramngvcrficy2no44	359	12	on	on	IN
work_v44vkh6bzramngvcrficy2no44	359	13	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	359	14	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	359	15	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	359	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	359	17	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	359	18	197	197	CD
work_v44vkh6bzramngvcrficy2no44	359	19	–	–	:
work_v44vkh6bzramngvcrficy2no44	359	20	205	205	CD
work_v44vkh6bzramngvcrficy2no44	359	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	359	22	Morristown	Morristown	NNP
work_v44vkh6bzramngvcrficy2no44	359	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	359	24	NJ	NJ	NNP
work_v44vkh6bzramngvcrficy2no44	359	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	359	26	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	359	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	360	1	Alexandra	Alexandra	NNP
work_v44vkh6bzramngvcrficy2no44	360	2	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	360	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	360	4	Miles	Miles	NNP
work_v44vkh6bzramngvcrficy2no44	360	5	Osborne	Osborne	NNP
work_v44vkh6bzramngvcrficy2no44	360	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	360	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	360	8	Phil	Phil	NNP
work_v44vkh6bzramngvcrficy2no44	360	9	Blunsom	Blunsom	NNP
work_v44vkh6bzramngvcrficy2no44	360	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	361	1	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	361	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	362	1	Metrics	metric	NNS
work_v44vkh6bzramngvcrficy2no44	362	2	for	for	IN
work_v44vkh6bzramngvcrficy2no44	362	3	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	362	4	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	362	5	:	:	:
work_v44vkh6bzramngvcrficy2no44	362	6	evaluating	evaluate	VBG
work_v44vkh6bzramngvcrficy2no44	362	7	reorder-	reorder-	NNP
work_v44vkh6bzramngvcrficy2no44	362	8	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	362	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	363	1	Machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	363	2	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	363	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	363	4	24(1):15–26	24(1):15–26	CD
work_v44vkh6bzramngvcrficy2no44	363	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	364	1	Arianna	Arianna	NNP
work_v44vkh6bzramngvcrficy2no44	364	2	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	364	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	364	4	Marcello	Marcello	NNP
work_v44vkh6bzramngvcrficy2no44	364	5	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	364	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	365	1	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	365	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	366	1	Modi-	Modi-	NNP
work_v44vkh6bzramngvcrficy2no44	366	2	fied	fie	VBD
work_v44vkh6bzramngvcrficy2no44	366	3	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	366	4	matrices	matrix	NNS
work_v44vkh6bzramngvcrficy2no44	366	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	366	6	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	366	7	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	366	8	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	366	9	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	366	10	ma-	ma-	JJ
work_v44vkh6bzramngvcrficy2no44	366	11	chine	chine	JJ
work_v44vkh6bzramngvcrficy2no44	366	12	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	366	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	367	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	367	2	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	367	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	367	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	367	5	50th	50th	JJ
work_v44vkh6bzramngvcrficy2no44	367	6	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	367	7	Meeting	Meeting	NNP
work_v44vkh6bzramngvcrficy2no44	367	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	367	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	367	10	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	367	11	for	for	IN
work_v44vkh6bzramngvcrficy2no44	367	12	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	367	13	Linguis-	Linguis-	NNP
work_v44vkh6bzramngvcrficy2no44	367	14	tics	tic	NNS
work_v44vkh6bzramngvcrficy2no44	367	15	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	367	16	Volume	volume	NN
work_v44vkh6bzramngvcrficy2no44	367	17	1	1	CD
work_v44vkh6bzramngvcrficy2no44	367	18	:	:	:
work_v44vkh6bzramngvcrficy2no44	367	19	Long	Long	NNP
work_v44vkh6bzramngvcrficy2no44	367	20	Papers	Papers	NNPS
work_v44vkh6bzramngvcrficy2no44	367	21	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	367	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	367	23	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	367	24	478–487	478–487	CD
work_v44vkh6bzramngvcrficy2no44	367	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	367	26	Jeju	Jeju	NNP
work_v44vkh6bzramngvcrficy2no44	367	27	Is-	Is-	NNP
work_v44vkh6bzramngvcrficy2no44	367	28	land	land	NN
work_v44vkh6bzramngvcrficy2no44	367	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	367	30	Korea	Korea	NNP
work_v44vkh6bzramngvcrficy2no44	367	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	367	32	July	July	NNP
work_v44vkh6bzramngvcrficy2no44	367	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	368	1	Arianna	Arianna	NNP
work_v44vkh6bzramngvcrficy2no44	368	2	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	368	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	368	4	Daniele	Daniele	NNP
work_v44vkh6bzramngvcrficy2no44	368	5	Pighin	Pighin	NNP
work_v44vkh6bzramngvcrficy2no44	368	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	368	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	368	8	Marcello	Marcello	NNP
work_v44vkh6bzramngvcrficy2no44	368	9	Fed-	Fed-	NNP
work_v44vkh6bzramngvcrficy2no44	368	10	erico	erico	NN
work_v44vkh6bzramngvcrficy2no44	368	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	369	1	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	369	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	370	1	Chunk	chunk	NN
work_v44vkh6bzramngvcrficy2no44	370	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	370	3	lattices	lattice	NNS
work_v44vkh6bzramngvcrficy2no44	370	4	for	for	IN
work_v44vkh6bzramngvcrficy2no44	370	5	verb	verb	JJ
work_v44vkh6bzramngvcrficy2no44	370	6	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	370	7	in	in	IN
work_v44vkh6bzramngvcrficy2no44	370	8	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	370	9	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	370	10	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	370	11	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	370	12	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	370	13	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	370	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	371	1	Ma-	ma-	DT
work_v44vkh6bzramngvcrficy2no44	371	2	chine	chine	NN
work_v44vkh6bzramngvcrficy2no44	371	3	Translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	371	4	,	,	,
work_v44vkh6bzramngvcrficy2no44	371	5	Special	Special	NNP
work_v44vkh6bzramngvcrficy2no44	371	6	Issue	Issue	NNP
work_v44vkh6bzramngvcrficy2no44	371	7	on	on	IN
work_v44vkh6bzramngvcrficy2no44	371	8	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	371	9	for	for	IN
work_v44vkh6bzramngvcrficy2no44	371	10	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	371	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	371	12	26(1	26(1	CD
work_v44vkh6bzramngvcrficy2no44	371	13	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	371	14	2):85–103	2):85–103	CD
work_v44vkh6bzramngvcrficy2no44	371	15	.	.	.
work_v44vkh6bzramngvcrficy2no44	372	1	Arianna	Arianna	NNP
work_v44vkh6bzramngvcrficy2no44	372	2	Bisazza	Bisazza	NNP
work_v44vkh6bzramngvcrficy2no44	372	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	373	1	2013	2013	CD
work_v44vkh6bzramngvcrficy2no44	373	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	374	1	Linguistically	linguistically	RB
work_v44vkh6bzramngvcrficy2no44	374	2	Motivated	motivated	JJ
work_v44vkh6bzramngvcrficy2no44	374	3	Re-	Re-	NNP
work_v44vkh6bzramngvcrficy2no44	374	4	ordering	order	VBG
work_v44vkh6bzramngvcrficy2no44	374	5	Modeling	model	VBG
work_v44vkh6bzramngvcrficy2no44	374	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	374	7	Phrase	Phrase	NNP
work_v44vkh6bzramngvcrficy2no44	374	8	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	374	9	Based	Based	NNP
work_v44vkh6bzramngvcrficy2no44	374	10	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	374	11	Ma-	Ma-	NNP
work_v44vkh6bzramngvcrficy2no44	374	12	chine	chine	NN
work_v44vkh6bzramngvcrficy2no44	374	13	Translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	374	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	375	1	Ph.D.	Ph.D.	NNP
work_v44vkh6bzramngvcrficy2no44	375	2	thesis	thesis	NNP
work_v44vkh6bzramngvcrficy2no44	375	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	375	4	University	University	NNP
work_v44vkh6bzramngvcrficy2no44	375	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	375	6	Trento	Trento	NNP
work_v44vkh6bzramngvcrficy2no44	375	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	376	1	http://eprints-phd.biblio.unitn.it/1019/.	http://eprints-phd.biblio.unitn.it/1019/.	NNP
work_v44vkh6bzramngvcrficy2no44	377	1	Marine	Marine	NNP
work_v44vkh6bzramngvcrficy2no44	377	2	Carpuat	Carpuat	NNP
work_v44vkh6bzramngvcrficy2no44	377	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	377	4	Yuval	Yuval	NNP
work_v44vkh6bzramngvcrficy2no44	377	5	Marton	Marton	NNP
work_v44vkh6bzramngvcrficy2no44	377	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	377	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	377	8	Nizar	Nizar	NNP
work_v44vkh6bzramngvcrficy2no44	377	9	Habash	Habash	NNP
work_v44vkh6bzramngvcrficy2no44	377	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	378	1	2012	2012	CD
work_v44vkh6bzramngvcrficy2no44	378	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	379	1	Improved	Improved	NNP
work_v44vkh6bzramngvcrficy2no44	379	2	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	379	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	379	4	to	to	IN
work_v44vkh6bzramngvcrficy2no44	379	5	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	379	6	English	english	JJ
work_v44vkh6bzramngvcrficy2no44	379	7	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	379	8	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	379	9	trans-	trans-	JJ
work_v44vkh6bzramngvcrficy2no44	379	10	lation	lation	NN
work_v44vkh6bzramngvcrficy2no44	379	11	by	by	IN
work_v44vkh6bzramngvcrficy2no44	379	12	reordering	reorder	VBG
work_v44vkh6bzramngvcrficy2no44	379	13	post	post	JJ
work_v44vkh6bzramngvcrficy2no44	379	14	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	379	15	verbal	verbal	JJ
work_v44vkh6bzramngvcrficy2no44	379	16	subjects	subject	NNS
work_v44vkh6bzramngvcrficy2no44	379	17	for	for	IN
work_v44vkh6bzramngvcrficy2no44	379	18	word	word	NN
work_v44vkh6bzramngvcrficy2no44	379	19	alignment	alignment	NN
work_v44vkh6bzramngvcrficy2no44	379	20	.	.	.
work_v44vkh6bzramngvcrficy2no44	380	1	Machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	380	2	Translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	380	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	380	4	Special	Special	NNP
work_v44vkh6bzramngvcrficy2no44	380	5	Issue	Issue	NNP
work_v44vkh6bzramngvcrficy2no44	380	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	380	7	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	380	8	for	for	IN
work_v44vkh6bzramngvcrficy2no44	380	9	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	380	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	380	11	26(1	26(1	CD
work_v44vkh6bzramngvcrficy2no44	380	12	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	380	13	2):105–120	2):105–120	CD
work_v44vkh6bzramngvcrficy2no44	380	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	381	1	Mauro	Mauro	NNP
work_v44vkh6bzramngvcrficy2no44	381	2	Cettolo	Cettolo	NNP
work_v44vkh6bzramngvcrficy2no44	381	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	381	4	Nicola	Nicola	NNP
work_v44vkh6bzramngvcrficy2no44	381	5	Bertoldi	Bertoldi	NNP
work_v44vkh6bzramngvcrficy2no44	381	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	381	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	381	8	Marcello	Marcello	NNP
work_v44vkh6bzramngvcrficy2no44	381	9	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	381	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	382	1	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	382	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	383	1	Methods	method	NNS
work_v44vkh6bzramngvcrficy2no44	383	2	for	for	IN
work_v44vkh6bzramngvcrficy2no44	383	3	smoothing	smooth	VBG
work_v44vkh6bzramngvcrficy2no44	383	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	383	5	optimizer	optimizer	NN
work_v44vkh6bzramngvcrficy2no44	383	6	instability	instability	NN
work_v44vkh6bzramngvcrficy2no44	383	7	in	in	IN
work_v44vkh6bzramngvcrficy2no44	383	8	SMT	SMT	NNP
work_v44vkh6bzramngvcrficy2no44	383	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	384	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	384	2	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	384	3	Summit	Summit	NNP
work_v44vkh6bzramngvcrficy2no44	384	4	XIII	XIII	NNP
work_v44vkh6bzramngvcrficy2no44	384	5	:	:	:
work_v44vkh6bzramngvcrficy2no44	384	6	the	the	DT
work_v44vkh6bzramngvcrficy2no44	384	7	Thirteenth	Thirteenth	NNP
work_v44vkh6bzramngvcrficy2no44	384	8	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	384	9	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	384	10	Summit	Summit	NNP
work_v44vkh6bzramngvcrficy2no44	384	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	384	12	pages	page	NNS
work_v44vkh6bzramngvcrficy2no44	384	13	32–39	32–39	RB
work_v44vkh6bzramngvcrficy2no44	384	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	384	15	Xiamen	Xiamen	NNP
work_v44vkh6bzramngvcrficy2no44	384	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	384	17	China	China	NNP
work_v44vkh6bzramngvcrficy2no44	384	18	.	.	.
work_v44vkh6bzramngvcrficy2no44	385	1	Stanley	Stanley	NNP
work_v44vkh6bzramngvcrficy2no44	385	2	F.	F.	NNP
work_v44vkh6bzramngvcrficy2no44	385	3	Chen	Chen	NNP
work_v44vkh6bzramngvcrficy2no44	385	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	385	5	Joshua	Joshua	NNP
work_v44vkh6bzramngvcrficy2no44	385	6	Goodman	Goodman	NNP
work_v44vkh6bzramngvcrficy2no44	385	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	386	1	1999	1999	CD
work_v44vkh6bzramngvcrficy2no44	386	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	387	1	An	an	DT
work_v44vkh6bzramngvcrficy2no44	387	2	empiri-	empiri-	JJ
work_v44vkh6bzramngvcrficy2no44	387	3	cal	cal	JJ
work_v44vkh6bzramngvcrficy2no44	387	4	study	study	NN
work_v44vkh6bzramngvcrficy2no44	387	5	of	of	IN
work_v44vkh6bzramngvcrficy2no44	387	6	smoothing	smooth	VBG
work_v44vkh6bzramngvcrficy2no44	387	7	techniques	technique	NNS
work_v44vkh6bzramngvcrficy2no44	387	8	for	for	IN
work_v44vkh6bzramngvcrficy2no44	387	9	language	language	NN
work_v44vkh6bzramngvcrficy2no44	387	10	model-	model-	NNP
work_v44vkh6bzramngvcrficy2no44	387	11	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	387	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	388	1	Computer	Computer	NNP
work_v44vkh6bzramngvcrficy2no44	388	2	Speech	Speech	NNP
work_v44vkh6bzramngvcrficy2no44	388	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	388	4	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	388	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	388	6	4(13):359–393	4(13):359–393	CD
work_v44vkh6bzramngvcrficy2no44	388	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	389	1	Marta	Marta	NNP
work_v44vkh6bzramngvcrficy2no44	389	2	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	389	3	Costa	Costa	NNP
work_v44vkh6bzramngvcrficy2no44	389	4	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	389	5	jussà	jussà	NNP
work_v44vkh6bzramngvcrficy2no44	389	6	and	and	CC
work_v44vkh6bzramngvcrficy2no44	389	7	José	José	NNP
work_v44vkh6bzramngvcrficy2no44	389	8	A.	A.	NNP
work_v44vkh6bzramngvcrficy2no44	389	9	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	389	10	Fonollosa	Fonollosa	NNP
work_v44vkh6bzramngvcrficy2no44	389	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	390	1	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	390	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	391	1	State	state	NN
work_v44vkh6bzramngvcrficy2no44	391	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	391	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	391	4	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	391	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	391	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	391	7	art	art	NN
work_v44vkh6bzramngvcrficy2no44	391	8	word	word	NN
work_v44vkh6bzramngvcrficy2no44	391	9	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	391	10	approaches	approach	NNS
work_v44vkh6bzramngvcrficy2no44	391	11	in	in	IN
work_v44vkh6bzramngvcrficy2no44	391	12	statisti-	statisti-	NNP
work_v44vkh6bzramngvcrficy2no44	391	13	cal	cal	NN
work_v44vkh6bzramngvcrficy2no44	391	14	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	391	15	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	391	16	:	:	:
work_v44vkh6bzramngvcrficy2no44	391	17	A	a	DT
work_v44vkh6bzramngvcrficy2no44	391	18	survey	survey	NN
work_v44vkh6bzramngvcrficy2no44	391	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	392	1	IEICE	ieice	NN
work_v44vkh6bzramngvcrficy2no44	392	2	TRANSAC-	TRANSAC-	NNS
work_v44vkh6bzramngvcrficy2no44	392	3	TIONS	tion	VBZ
work_v44vkh6bzramngvcrficy2no44	392	4	on	on	IN
work_v44vkh6bzramngvcrficy2no44	392	5	Information	Information	NNP
work_v44vkh6bzramngvcrficy2no44	392	6	and	and	CC
work_v44vkh6bzramngvcrficy2no44	392	7	Systems	Systems	NNPS
work_v44vkh6bzramngvcrficy2no44	392	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	392	9	E92-D(11):2179	E92-D(11):2179	NNP
work_v44vkh6bzramngvcrficy2no44	392	10	–	–	:
work_v44vkh6bzramngvcrficy2no44	392	11	2185	2185	CD
work_v44vkh6bzramngvcrficy2no44	392	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	393	1	Koby	Koby	NNP
work_v44vkh6bzramngvcrficy2no44	393	2	Crammer	Crammer	NNP
work_v44vkh6bzramngvcrficy2no44	393	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	393	4	Yoram	Yoram	NNP
work_v44vkh6bzramngvcrficy2no44	393	5	Singer	Singer	NNP
work_v44vkh6bzramngvcrficy2no44	393	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	394	1	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	394	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	395	1	Ultraconser-	Ultraconser-	NNP
work_v44vkh6bzramngvcrficy2no44	395	2	vative	vative	JJ
work_v44vkh6bzramngvcrficy2no44	395	3	online	online	JJ
work_v44vkh6bzramngvcrficy2no44	395	4	algorithms	algorithm	NNS
work_v44vkh6bzramngvcrficy2no44	395	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	395	6	multiclass	multiclass	NN
work_v44vkh6bzramngvcrficy2no44	395	7	problems	problem	NNS
work_v44vkh6bzramngvcrficy2no44	395	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	396	1	J.	J.	NNP
work_v44vkh6bzramngvcrficy2no44	396	2	Mach	Mach	NNP
work_v44vkh6bzramngvcrficy2no44	396	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	397	1	Learn	learn	VB
work_v44vkh6bzramngvcrficy2no44	397	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	398	1	Res	Res	NNP
work_v44vkh6bzramngvcrficy2no44	398	2	.	.	NNP
work_v44vkh6bzramngvcrficy2no44	398	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	398	4	3:951–991	3:951–991	CD
work_v44vkh6bzramngvcrficy2no44	398	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	398	6	March	March	NNP
work_v44vkh6bzramngvcrficy2no44	398	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	399	1	Hal	Hal	NNP
work_v44vkh6bzramngvcrficy2no44	399	2	Daumé	Daumé	NNP
work_v44vkh6bzramngvcrficy2no44	399	3	III	iii	CD
work_v44vkh6bzramngvcrficy2no44	399	4	.	.	.
work_v44vkh6bzramngvcrficy2no44	400	1	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	400	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	401	1	Notes	note	NNS
work_v44vkh6bzramngvcrficy2no44	401	2	on	on	IN
work_v44vkh6bzramngvcrficy2no44	401	3	CG	CG	NNP
work_v44vkh6bzramngvcrficy2no44	401	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	401	5	LM	LM	NNP
work_v44vkh6bzramngvcrficy2no44	401	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	401	7	BFGS	BFGS	NNP
work_v44vkh6bzramngvcrficy2no44	401	8	op-	op-	XX
work_v44vkh6bzramngvcrficy2no44	401	9	timization	timization	NN
work_v44vkh6bzramngvcrficy2no44	401	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	401	11	logistic	logistic	JJ
work_v44vkh6bzramngvcrficy2no44	401	12	regression	regression	NN
work_v44vkh6bzramngvcrficy2no44	401	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	402	1	Paper	paper	NN
work_v44vkh6bzramngvcrficy2no44	402	2	available	available	JJ
work_v44vkh6bzramngvcrficy2no44	402	3	at	at	IN
work_v44vkh6bzramngvcrficy2no44	402	4	http://pub.hal3.name	http://pub.hal3.name	CD
work_v44vkh6bzramngvcrficy2no44	402	5	,	,	,
work_v44vkh6bzramngvcrficy2no44	402	6	implementation	implementation	NN
work_v44vkh6bzramngvcrficy2no44	402	7	avail-	avail-	XX
work_v44vkh6bzramngvcrficy2no44	402	8	able	able	JJ
work_v44vkh6bzramngvcrficy2no44	402	9	at	at	IN
work_v44vkh6bzramngvcrficy2no44	402	10	http://hal3.name/megam	http://hal3.name/megam	ADD
work_v44vkh6bzramngvcrficy2no44	402	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	403	1	Mona	Mona	NNP
work_v44vkh6bzramngvcrficy2no44	403	2	Diab	Diab	NNP
work_v44vkh6bzramngvcrficy2no44	403	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	403	4	Kadri	Kadri	NNP
work_v44vkh6bzramngvcrficy2no44	403	5	Hacioglu	Hacioglu	NNP
work_v44vkh6bzramngvcrficy2no44	403	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	403	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	403	8	Daniel	Daniel	NNP
work_v44vkh6bzramngvcrficy2no44	403	9	Jurafsky	Jurafsky	NNP
work_v44vkh6bzramngvcrficy2no44	403	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	404	1	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	404	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	405	1	Automatic	automatic	JJ
work_v44vkh6bzramngvcrficy2no44	405	2	Tagging	Tagging	NNP
work_v44vkh6bzramngvcrficy2no44	405	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	405	4	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	405	5	Text	Text	NNP
work_v44vkh6bzramngvcrficy2no44	405	6	:	:	:
work_v44vkh6bzramngvcrficy2no44	405	7	From	from	IN
work_v44vkh6bzramngvcrficy2no44	405	8	Raw	Raw	NNP
work_v44vkh6bzramngvcrficy2no44	405	9	Text	Text	NNP
work_v44vkh6bzramngvcrficy2no44	405	10	to	to	IN
work_v44vkh6bzramngvcrficy2no44	405	11	Base	Base	NNP
work_v44vkh6bzramngvcrficy2no44	405	12	Phrase	Phrase	NNP
work_v44vkh6bzramngvcrficy2no44	405	13	Chunks	Chunks	NNPS
work_v44vkh6bzramngvcrficy2no44	405	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	406	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	406	2	Daniel	Daniel	NNP
work_v44vkh6bzramngvcrficy2no44	406	3	Marcu	Marcu	NNP
work_v44vkh6bzramngvcrficy2no44	406	4	Susan	Susan	NNP
work_v44vkh6bzramngvcrficy2no44	406	5	Dumais	Dumais	NNP
work_v44vkh6bzramngvcrficy2no44	406	6	and	and	CC
work_v44vkh6bzramngvcrficy2no44	406	7	Salim	Salim	NNP
work_v44vkh6bzramngvcrficy2no44	406	8	Roukos	Roukos	NNP
work_v44vkh6bzramngvcrficy2no44	406	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	406	10	editors	editor	NNS
work_v44vkh6bzramngvcrficy2no44	406	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	406	12	HLT	HLT	NNP
work_v44vkh6bzramngvcrficy2no44	406	13	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	406	14	NAACL	NAACL	NNP
work_v44vkh6bzramngvcrficy2no44	406	15	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	406	16	:	:	:
work_v44vkh6bzramngvcrficy2no44	406	17	Short	Short	NNP
work_v44vkh6bzramngvcrficy2no44	406	18	Papers	Papers	NNPS
work_v44vkh6bzramngvcrficy2no44	406	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	406	20	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	406	21	149–152	149–152	CD
work_v44vkh6bzramngvcrficy2no44	406	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	406	23	Boston	Boston	NNP
work_v44vkh6bzramngvcrficy2no44	406	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	406	25	Massachusetts	Massachusetts	NNP
work_v44vkh6bzramngvcrficy2no44	406	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	406	27	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	406	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	407	1	Marcello	Marcello	NNP
work_v44vkh6bzramngvcrficy2no44	407	2	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	407	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	407	4	Nicola	Nicola	NNP
work_v44vkh6bzramngvcrficy2no44	407	5	Bertoldi	Bertoldi	NNP
work_v44vkh6bzramngvcrficy2no44	407	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	407	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	407	8	Mauro	Mauro	NNP
work_v44vkh6bzramngvcrficy2no44	407	9	Cettolo	Cettolo	NNP
work_v44vkh6bzramngvcrficy2no44	407	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	408	1	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	408	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	409	1	IRSTLM	IRSTLM	NNP
work_v44vkh6bzramngvcrficy2no44	409	2	:	:	:
work_v44vkh6bzramngvcrficy2no44	409	3	an	an	DT
work_v44vkh6bzramngvcrficy2no44	409	4	Open	open	JJ
work_v44vkh6bzramngvcrficy2no44	409	5	Source	source	NN
work_v44vkh6bzramngvcrficy2no44	409	6	Toolkit	Toolkit	NNP
work_v44vkh6bzramngvcrficy2no44	409	7	for	for	IN
work_v44vkh6bzramngvcrficy2no44	409	8	Handling	handle	VBG
work_v44vkh6bzramngvcrficy2no44	409	9	Large	large	JJ
work_v44vkh6bzramngvcrficy2no44	409	10	Scale	Scale	NNP
work_v44vkh6bzramngvcrficy2no44	409	11	Language	language	NN
work_v44vkh6bzramngvcrficy2no44	409	12	Models	Models	NNPS
work_v44vkh6bzramngvcrficy2no44	409	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	410	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	410	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	410	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	410	4	In-	In-	NNP
work_v44vkh6bzramngvcrficy2no44	410	5	terspeech	terspeech	NN
work_v44vkh6bzramngvcrficy2no44	410	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	410	7	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	410	8	1618–1621	1618–1621	CD
work_v44vkh6bzramngvcrficy2no44	410	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	410	10	Melbourne	Melbourne	NNP
work_v44vkh6bzramngvcrficy2no44	410	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	410	12	Australia	Australia	NNP
work_v44vkh6bzramngvcrficy2no44	410	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	411	1	Minwei	Minwei	NNP
work_v44vkh6bzramngvcrficy2no44	411	2	Feng	Feng	NNP
work_v44vkh6bzramngvcrficy2no44	411	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	411	4	Arne	Arne	NNP
work_v44vkh6bzramngvcrficy2no44	411	5	Mauser	Mauser	NNP
work_v44vkh6bzramngvcrficy2no44	411	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	411	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	411	8	Hermann	Hermann	NNP
work_v44vkh6bzramngvcrficy2no44	411	9	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	411	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	412	1	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	412	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	413	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	413	2	source	source	NN
work_v44vkh6bzramngvcrficy2no44	413	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	413	4	side	side	NN
work_v44vkh6bzramngvcrficy2no44	413	5	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	413	6	sequence	sequence	NN
work_v44vkh6bzramngvcrficy2no44	413	7	model	model	NN
work_v44vkh6bzramngvcrficy2no44	413	8	for	for	IN
work_v44vkh6bzramngvcrficy2no44	413	9	statisti-	statisti-	NNP
work_v44vkh6bzramngvcrficy2no44	413	10	cal	cal	NNP
work_v44vkh6bzramngvcrficy2no44	413	11	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	413	12	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	413	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	414	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	414	2	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	414	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	414	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	414	5	Associa-	Associa-	NNP
work_v44vkh6bzramngvcrficy2no44	414	6	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	414	7	for	for	IN
work_v44vkh6bzramngvcrficy2no44	414	8	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	414	9	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	414	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	414	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	414	12	Americas	Americas	NNPS
work_v44vkh6bzramngvcrficy2no44	414	13	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	414	14	AMTA	AMTA	NNP
work_v44vkh6bzramngvcrficy2no44	414	15	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	414	16	,	,	,
work_v44vkh6bzramngvcrficy2no44	414	17	Denver	Denver	NNP
work_v44vkh6bzramngvcrficy2no44	414	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	414	19	Colorado	Colorado	NNP
work_v44vkh6bzramngvcrficy2no44	414	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	414	21	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	414	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	415	1	Michel	Michel	NNP
work_v44vkh6bzramngvcrficy2no44	415	2	Galley	Galley	NNP
work_v44vkh6bzramngvcrficy2no44	415	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	415	4	Christopher	Christopher	NNP
work_v44vkh6bzramngvcrficy2no44	415	5	D.	D.	NNP
work_v44vkh6bzramngvcrficy2no44	415	6	Manning	Manning	NNP
work_v44vkh6bzramngvcrficy2no44	415	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	416	1	2008	2008	CD
work_v44vkh6bzramngvcrficy2no44	416	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	417	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	417	2	simple	simple	JJ
work_v44vkh6bzramngvcrficy2no44	417	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	417	4	effective	effective	JJ
work_v44vkh6bzramngvcrficy2no44	417	5	hierarchical	hierarchical	JJ
work_v44vkh6bzramngvcrficy2no44	417	6	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	417	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	417	8	model	model	NN
work_v44vkh6bzramngvcrficy2no44	417	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	418	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	418	2	EMNLP	EMNLP	NNP
work_v44vkh6bzramngvcrficy2no44	418	3	’	'	''
work_v44vkh6bzramngvcrficy2no44	418	4	08	08	CD
work_v44vkh6bzramngvcrficy2no44	418	5	:	:	:
work_v44vkh6bzramngvcrficy2no44	418	6	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	418	7	of	of	IN
work_v44vkh6bzramngvcrficy2no44	418	8	the	the	DT
work_v44vkh6bzramngvcrficy2no44	418	9	Confer-	Confer-	NNP
work_v44vkh6bzramngvcrficy2no44	418	10	ence	ence	NN
work_v44vkh6bzramngvcrficy2no44	418	11	on	on	IN
work_v44vkh6bzramngvcrficy2no44	418	12	Empirical	Empirical	NNP
work_v44vkh6bzramngvcrficy2no44	418	13	Methods	Methods	NNPS
work_v44vkh6bzramngvcrficy2no44	418	14	in	in	IN
work_v44vkh6bzramngvcrficy2no44	418	15	Natural	Natural	NNP
work_v44vkh6bzramngvcrficy2no44	418	16	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	418	17	Pro-	Pro-	NNP
work_v44vkh6bzramngvcrficy2no44	418	18	cessing	cessing	NN
work_v44vkh6bzramngvcrficy2no44	418	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	418	20	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	418	21	848–856	848–856	CD
work_v44vkh6bzramngvcrficy2no44	418	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	418	23	Morristown	Morristown	NNP
work_v44vkh6bzramngvcrficy2no44	418	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	418	25	NJ	NJ	NNP
work_v44vkh6bzramngvcrficy2no44	418	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	418	27	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	418	28	.	.	.
work_v44vkh6bzramngvcrficy2no44	419	1	Spence	Spence	NNP
work_v44vkh6bzramngvcrficy2no44	419	2	Green	Green	NNP
work_v44vkh6bzramngvcrficy2no44	419	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	419	4	Conal	Conal	NNP
work_v44vkh6bzramngvcrficy2no44	419	5	Sathi	Sathi	NNP
work_v44vkh6bzramngvcrficy2no44	419	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	419	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	419	8	Christopher	Christopher	NNP
work_v44vkh6bzramngvcrficy2no44	419	9	D.	D.	NNP
work_v44vkh6bzramngvcrficy2no44	419	10	Man-	Man-	NNP
work_v44vkh6bzramngvcrficy2no44	419	11	ning	ning	NN
work_v44vkh6bzramngvcrficy2no44	419	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	420	1	2009	2009	CD
work_v44vkh6bzramngvcrficy2no44	420	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	421	1	NP	NP	NNP
work_v44vkh6bzramngvcrficy2no44	421	2	subject	subject	JJ
work_v44vkh6bzramngvcrficy2no44	421	3	detection	detection	NN
work_v44vkh6bzramngvcrficy2no44	421	4	in	in	IN
work_v44vkh6bzramngvcrficy2no44	421	5	verb	verb	NN
work_v44vkh6bzramngvcrficy2no44	421	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	421	7	initial	initial	JJ
work_v44vkh6bzramngvcrficy2no44	421	8	Ara-	Ara-	NNP
work_v44vkh6bzramngvcrficy2no44	421	9	bic	bic	JJ
work_v44vkh6bzramngvcrficy2no44	421	10	clauses	clause	NNS
work_v44vkh6bzramngvcrficy2no44	421	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	422	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	422	2	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	422	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	422	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	422	5	Third	Third	NNP
work_v44vkh6bzramngvcrficy2no44	422	6	Workshop	Workshop	NNP
work_v44vkh6bzramngvcrficy2no44	422	7	338	338	CD
work_v44vkh6bzramngvcrficy2no44	422	8	on	on	IN
work_v44vkh6bzramngvcrficy2no44	422	9	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	422	10	Approaches	approach	NNS
work_v44vkh6bzramngvcrficy2no44	422	11	to	to	IN
work_v44vkh6bzramngvcrficy2no44	422	12	Arabic	Arabic	NNP
work_v44vkh6bzramngvcrficy2no44	422	13	Script	Script	NNP
work_v44vkh6bzramngvcrficy2no44	422	14	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	422	15	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	422	16	Languages	Languages	NNP
work_v44vkh6bzramngvcrficy2no44	422	17	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	422	18	CAASL3	CAASL3	NNP
work_v44vkh6bzramngvcrficy2no44	422	19	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	422	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	422	21	Ottawa	Ottawa	NNP
work_v44vkh6bzramngvcrficy2no44	422	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	422	23	Canada	Canada	NNP
work_v44vkh6bzramngvcrficy2no44	422	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	423	1	Spence	Spence	NNP
work_v44vkh6bzramngvcrficy2no44	423	2	Green	Green	NNP
work_v44vkh6bzramngvcrficy2no44	423	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	423	4	Michel	Michel	NNP
work_v44vkh6bzramngvcrficy2no44	423	5	Galley	Galley	NNP
work_v44vkh6bzramngvcrficy2no44	423	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	423	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	423	8	Christopher	Christopher	NNP
work_v44vkh6bzramngvcrficy2no44	423	9	D.	D.	NNP
work_v44vkh6bzramngvcrficy2no44	423	10	Man-	Man-	NNP
work_v44vkh6bzramngvcrficy2no44	423	11	ning	ning	NN
work_v44vkh6bzramngvcrficy2no44	423	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	424	1	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	424	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	425	1	Improved	improved	JJ
work_v44vkh6bzramngvcrficy2no44	425	2	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	425	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	425	4	distortion	distortion	NN
work_v44vkh6bzramngvcrficy2no44	425	5	cost	cost	NN
work_v44vkh6bzramngvcrficy2no44	425	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	425	7	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	425	8	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	425	9	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	425	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	426	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	426	2	Human	Human	NNP
work_v44vkh6bzramngvcrficy2no44	426	3	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	426	4	Technologies	Technologies	NNPS
work_v44vkh6bzramngvcrficy2no44	426	5	:	:	:
work_v44vkh6bzramngvcrficy2no44	426	6	The	the	DT
work_v44vkh6bzramngvcrficy2no44	426	7	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	426	8	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	426	9	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	426	10	of	of	IN
work_v44vkh6bzramngvcrficy2no44	426	11	the	the	DT
work_v44vkh6bzramngvcrficy2no44	426	12	North	north	JJ
work_v44vkh6bzramngvcrficy2no44	426	13	American	American	NNP
work_v44vkh6bzramngvcrficy2no44	426	14	Chapter	Chapter	NNP
work_v44vkh6bzramngvcrficy2no44	426	15	of	of	IN
work_v44vkh6bzramngvcrficy2no44	426	16	the	the	DT
work_v44vkh6bzramngvcrficy2no44	426	17	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	426	18	for	for	IN
work_v44vkh6bzramngvcrficy2no44	426	19	Com-	Com-	NNP
work_v44vkh6bzramngvcrficy2no44	426	20	putational	putational	JJ
work_v44vkh6bzramngvcrficy2no44	426	21	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	426	22	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	426	23	NAACL	NAACL	NNP
work_v44vkh6bzramngvcrficy2no44	426	24	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	426	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	426	26	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	426	27	867–875	867–875	CD
work_v44vkh6bzramngvcrficy2no44	426	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	426	29	Los	Los	NNP
work_v44vkh6bzramngvcrficy2no44	426	30	Angeles	Angeles	NNP
work_v44vkh6bzramngvcrficy2no44	426	31	,	,	,
work_v44vkh6bzramngvcrficy2no44	426	32	California	California	NNP
work_v44vkh6bzramngvcrficy2no44	426	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	427	1	Magnus	Magnus	NNP
work_v44vkh6bzramngvcrficy2no44	427	2	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	427	3	Hestenes	Hestenes	NNPS
work_v44vkh6bzramngvcrficy2no44	427	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	427	5	Eduard	Eduard	NNP
work_v44vkh6bzramngvcrficy2no44	427	6	Stiefel	Stiefel	NNP
work_v44vkh6bzramngvcrficy2no44	427	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	428	1	1952	1952	CD
work_v44vkh6bzramngvcrficy2no44	428	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	429	1	Meth-	meth-	JJ
work_v44vkh6bzramngvcrficy2no44	429	2	ods	od	NNS
work_v44vkh6bzramngvcrficy2no44	429	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	429	4	conjugate	conjugate	JJ
work_v44vkh6bzramngvcrficy2no44	429	5	gradients	gradient	NNS
work_v44vkh6bzramngvcrficy2no44	429	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	429	7	solving	solve	VBG
work_v44vkh6bzramngvcrficy2no44	429	8	linear	linear	JJ
work_v44vkh6bzramngvcrficy2no44	429	9	systems	system	NNS
work_v44vkh6bzramngvcrficy2no44	429	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	430	1	Journal	Journal	NNP
work_v44vkh6bzramngvcrficy2no44	430	2	of	of	IN
work_v44vkh6bzramngvcrficy2no44	430	3	Research	Research	NNP
work_v44vkh6bzramngvcrficy2no44	430	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	430	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	430	6	National	National	NNP
work_v44vkh6bzramngvcrficy2no44	430	7	Bureau	Bureau	NNP
work_v44vkh6bzramngvcrficy2no44	430	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	430	9	Stan-	Stan-	NNP
work_v44vkh6bzramngvcrficy2no44	430	10	dards	dards	NNP
work_v44vkh6bzramngvcrficy2no44	430	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	430	12	49(6):409–436	49(6):409–436	CD
work_v44vkh6bzramngvcrficy2no44	430	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	431	1	H.	H.	NNP
work_v44vkh6bzramngvcrficy2no44	431	2	Johnson	Johnson	NNP
work_v44vkh6bzramngvcrficy2no44	431	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	431	4	J.	J.	NNP
work_v44vkh6bzramngvcrficy2no44	431	5	Martin	Martin	NNP
work_v44vkh6bzramngvcrficy2no44	431	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	431	7	G.	G.	NNP
work_v44vkh6bzramngvcrficy2no44	431	8	Foster	Foster	NNP
work_v44vkh6bzramngvcrficy2no44	431	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	431	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	431	11	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	431	12	Kuhn	Kuhn	NNP
work_v44vkh6bzramngvcrficy2no44	431	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	432	1	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	432	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	433	1	Im-	im-	JJ
work_v44vkh6bzramngvcrficy2no44	433	2	proving	proving	NN
work_v44vkh6bzramngvcrficy2no44	433	3	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	433	4	quality	quality	NN
work_v44vkh6bzramngvcrficy2no44	433	5	by	by	IN
work_v44vkh6bzramngvcrficy2no44	433	6	discarding	discard	VBG
work_v44vkh6bzramngvcrficy2no44	433	7	most	most	JJS
work_v44vkh6bzramngvcrficy2no44	433	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	433	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	433	10	phrasetable	phrasetable	JJ
work_v44vkh6bzramngvcrficy2no44	433	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	434	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	434	2	In	in	IN
work_v44vkh6bzramngvcrficy2no44	434	3	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	434	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	434	5	EMNLP	EMNLP	NNP
work_v44vkh6bzramngvcrficy2no44	434	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	434	7	CoNLL	CoNLL	NNP
work_v44vkh6bzramngvcrficy2no44	434	8	07	07	CD
work_v44vkh6bzramngvcrficy2no44	434	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	434	10	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	434	11	967–975	967–975	CD
work_v44vkh6bzramngvcrficy2no44	434	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	435	1	Philipp	Philipp	NNP
work_v44vkh6bzramngvcrficy2no44	435	2	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	435	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	435	4	Franz	Franz	NNP
work_v44vkh6bzramngvcrficy2no44	435	5	Josef	Josef	NNP
work_v44vkh6bzramngvcrficy2no44	435	6	Och	Och	NNP
work_v44vkh6bzramngvcrficy2no44	435	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	435	8	and	and	CC
work_v44vkh6bzramngvcrficy2no44	435	9	Daniel	Daniel	NNP
work_v44vkh6bzramngvcrficy2no44	435	10	Marcu	Marcu	NNP
work_v44vkh6bzramngvcrficy2no44	435	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	436	1	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	436	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	437	1	Statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	437	2	phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	437	3	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	437	4	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	437	5	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	437	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	438	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	438	2	Proceed-	Proceed-	NNP
work_v44vkh6bzramngvcrficy2no44	438	3	ings	ing	NNS
work_v44vkh6bzramngvcrficy2no44	438	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	438	5	HLT	HLT	NNP
work_v44vkh6bzramngvcrficy2no44	438	6	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	438	7	NAACL	NAACL	NNP
work_v44vkh6bzramngvcrficy2no44	438	8	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	438	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	438	10	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	438	11	127–133	127–133	CD
work_v44vkh6bzramngvcrficy2no44	438	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	438	13	Edmonton	Edmonton	NNP
work_v44vkh6bzramngvcrficy2no44	438	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	438	15	Canada	Canada	NNP
work_v44vkh6bzramngvcrficy2no44	438	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	439	1	Philipp	Philipp	NNP
work_v44vkh6bzramngvcrficy2no44	439	2	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	439	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	439	4	Amittai	Amittai	NNP
work_v44vkh6bzramngvcrficy2no44	439	5	Axelrod	Axelrod	NNP
work_v44vkh6bzramngvcrficy2no44	439	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	439	7	Alexandra	Alexandra	NNP
work_v44vkh6bzramngvcrficy2no44	439	8	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	439	9	Mayne	Mayne	NNP
work_v44vkh6bzramngvcrficy2no44	439	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	439	11	Chris	Chris	NNP
work_v44vkh6bzramngvcrficy2no44	439	12	Callison	Callison	NNP
work_v44vkh6bzramngvcrficy2no44	439	13	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	439	14	Burch	Burch	NNP
work_v44vkh6bzramngvcrficy2no44	439	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	439	16	Miles	Miles	NNP
work_v44vkh6bzramngvcrficy2no44	439	17	Osborne	Osborne	NNP
work_v44vkh6bzramngvcrficy2no44	439	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	439	19	and	and	CC
work_v44vkh6bzramngvcrficy2no44	439	20	David	David	NNP
work_v44vkh6bzramngvcrficy2no44	439	21	Talbot	Talbot	NNP
work_v44vkh6bzramngvcrficy2no44	439	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	440	1	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	440	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	441	1	Edinburgh	Edinburgh	NNP
work_v44vkh6bzramngvcrficy2no44	441	2	system	system	NN
work_v44vkh6bzramngvcrficy2no44	441	3	description	description	NN
work_v44vkh6bzramngvcrficy2no44	441	4	for	for	IN
work_v44vkh6bzramngvcrficy2no44	441	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	441	6	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	441	7	IWSLT	IWSLT	NNP
work_v44vkh6bzramngvcrficy2no44	441	8	speech	speech	NN
work_v44vkh6bzramngvcrficy2no44	441	9	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	441	10	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	441	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	442	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	442	2	Proc	Proc	NNP
work_v44vkh6bzramngvcrficy2no44	442	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	443	1	of	of	IN
work_v44vkh6bzramngvcrficy2no44	443	2	the	the	DT
work_v44vkh6bzramngvcrficy2no44	443	3	International	International	NNP
work_v44vkh6bzramngvcrficy2no44	443	4	Workshop	Workshop	NNP
work_v44vkh6bzramngvcrficy2no44	443	5	on	on	IN
work_v44vkh6bzramngvcrficy2no44	443	6	Spoken	Spoken	NNP
work_v44vkh6bzramngvcrficy2no44	443	7	Lan-	Lan-	NNP
work_v44vkh6bzramngvcrficy2no44	443	8	guage	guage	NN
work_v44vkh6bzramngvcrficy2no44	443	9	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	443	10	,	,	,
work_v44vkh6bzramngvcrficy2no44	443	11	October	October	NNP
work_v44vkh6bzramngvcrficy2no44	443	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	444	1	P.	P.	NNP
work_v44vkh6bzramngvcrficy2no44	444	2	Koehn	Koehn	NNP
work_v44vkh6bzramngvcrficy2no44	444	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	4	H.	H.	NNP
work_v44vkh6bzramngvcrficy2no44	444	5	Hoang	Hoang	NNP
work_v44vkh6bzramngvcrficy2no44	444	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	7	A.	A.	NNP
work_v44vkh6bzramngvcrficy2no44	444	8	Birch	Birch	NNP
work_v44vkh6bzramngvcrficy2no44	444	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	10	C.	C.	NNP
work_v44vkh6bzramngvcrficy2no44	444	11	Callison	Callison	NNP
work_v44vkh6bzramngvcrficy2no44	444	12	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	444	13	Burch	Burch	NNP
work_v44vkh6bzramngvcrficy2no44	444	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	15	M.	M.	NNP
work_v44vkh6bzramngvcrficy2no44	444	16	Federico	Federico	NNP
work_v44vkh6bzramngvcrficy2no44	444	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	18	N.	N.	NNP
work_v44vkh6bzramngvcrficy2no44	444	19	Bertoldi	Bertoldi	NNP
work_v44vkh6bzramngvcrficy2no44	444	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	21	B.	B.	NNP
work_v44vkh6bzramngvcrficy2no44	444	22	Cowan	Cowan	NNP
work_v44vkh6bzramngvcrficy2no44	444	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	24	W.	W.	NNP
work_v44vkh6bzramngvcrficy2no44	444	25	Shen	Shen	NNP
work_v44vkh6bzramngvcrficy2no44	444	26	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	27	C.	C.	NNP
work_v44vkh6bzramngvcrficy2no44	444	28	Moran	Moran	NNP
work_v44vkh6bzramngvcrficy2no44	444	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	30	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	444	31	Zens	Zens	NNP
work_v44vkh6bzramngvcrficy2no44	444	32	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	33	C.	C.	NNP
work_v44vkh6bzramngvcrficy2no44	444	34	Dyer	Dyer	NNP
work_v44vkh6bzramngvcrficy2no44	444	35	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	36	O.	O.	NNP
work_v44vkh6bzramngvcrficy2no44	444	37	Bojar	Bojar	NNP
work_v44vkh6bzramngvcrficy2no44	444	38	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	39	A.	a.	NN
work_v44vkh6bzramngvcrficy2no44	444	40	Constantin	Constantin	NNP
work_v44vkh6bzramngvcrficy2no44	444	41	,	,	,
work_v44vkh6bzramngvcrficy2no44	444	42	and	and	CC
work_v44vkh6bzramngvcrficy2no44	444	43	E.	E.	NNP
work_v44vkh6bzramngvcrficy2no44	444	44	Herbst	Herbst	NNP
work_v44vkh6bzramngvcrficy2no44	444	45	.	.	.
work_v44vkh6bzramngvcrficy2no44	445	1	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	445	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	446	1	Moses	Moses	NNP
work_v44vkh6bzramngvcrficy2no44	446	2	:	:	:
work_v44vkh6bzramngvcrficy2no44	446	3	Open	Open	NNP
work_v44vkh6bzramngvcrficy2no44	446	4	Source	source	NN
work_v44vkh6bzramngvcrficy2no44	446	5	Toolkit	Toolkit	NNP
work_v44vkh6bzramngvcrficy2no44	446	6	for	for	IN
work_v44vkh6bzramngvcrficy2no44	446	7	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	446	8	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	446	9	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	446	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	447	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	447	2	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	447	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	447	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	447	5	45th	45th	JJ
work_v44vkh6bzramngvcrficy2no44	447	6	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	447	7	Meeting	Meeting	NNP
work_v44vkh6bzramngvcrficy2no44	447	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	447	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	447	10	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	447	11	for	for	IN
work_v44vkh6bzramngvcrficy2no44	447	12	Com-	Com-	NNP
work_v44vkh6bzramngvcrficy2no44	447	13	putational	putational	JJ
work_v44vkh6bzramngvcrficy2no44	447	14	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	447	15	Companion	Companion	NNP
work_v44vkh6bzramngvcrficy2no44	447	16	Volume	Volume	NNP
work_v44vkh6bzramngvcrficy2no44	447	17	Proceed-	Proceed-	NNP
work_v44vkh6bzramngvcrficy2no44	447	18	ings	ing	NNS
work_v44vkh6bzramngvcrficy2no44	447	19	of	of	IN
work_v44vkh6bzramngvcrficy2no44	447	20	the	the	DT
work_v44vkh6bzramngvcrficy2no44	447	21	Demo	Demo	NNP
work_v44vkh6bzramngvcrficy2no44	447	22	and	and	CC
work_v44vkh6bzramngvcrficy2no44	447	23	Poster	Poster	NNP
work_v44vkh6bzramngvcrficy2no44	447	24	Sessions	Sessions	NNPS
work_v44vkh6bzramngvcrficy2no44	447	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	447	26	pages	page	NNS
work_v44vkh6bzramngvcrficy2no44	447	27	177–180	177–180	CD
work_v44vkh6bzramngvcrficy2no44	447	28	,	,	,
work_v44vkh6bzramngvcrficy2no44	447	29	Prague	Prague	NNP
work_v44vkh6bzramngvcrficy2no44	447	30	,	,	,
work_v44vkh6bzramngvcrficy2no44	447	31	Czech	Czech	NNP
work_v44vkh6bzramngvcrficy2no44	447	32	Republic	Republic	NNP
work_v44vkh6bzramngvcrficy2no44	447	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	448	1	Ravi	Ravi	NNP
work_v44vkh6bzramngvcrficy2no44	448	2	Kumar	Kumar	NNP
work_v44vkh6bzramngvcrficy2no44	448	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	448	4	Sergei	Sergei	NNP
work_v44vkh6bzramngvcrficy2no44	448	5	Vassilvitskii	Vassilvitskii	NNP
work_v44vkh6bzramngvcrficy2no44	448	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	449	1	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	449	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	450	1	General-	General-	NNP
work_v44vkh6bzramngvcrficy2no44	450	2	ized	ize	VBD
work_v44vkh6bzramngvcrficy2no44	450	3	distances	distance	NNS
work_v44vkh6bzramngvcrficy2no44	450	4	between	between	IN
work_v44vkh6bzramngvcrficy2no44	450	5	rankings	ranking	NNS
work_v44vkh6bzramngvcrficy2no44	450	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	451	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	451	2	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	451	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	451	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	451	5	19th	19th	JJ
work_v44vkh6bzramngvcrficy2no44	451	6	international	international	JJ
work_v44vkh6bzramngvcrficy2no44	451	7	conference	conference	NN
work_v44vkh6bzramngvcrficy2no44	451	8	on	on	IN
work_v44vkh6bzramngvcrficy2no44	451	9	World	World	NNP
work_v44vkh6bzramngvcrficy2no44	451	10	Wide	Wide	NNP
work_v44vkh6bzramngvcrficy2no44	451	11	Web	Web	NNP
work_v44vkh6bzramngvcrficy2no44	451	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	451	13	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	451	14	571–580	571–580	CD
work_v44vkh6bzramngvcrficy2no44	451	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	451	16	New	New	NNP
work_v44vkh6bzramngvcrficy2no44	451	17	York	York	NNP
work_v44vkh6bzramngvcrficy2no44	451	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	451	19	NY	NY	NNP
work_v44vkh6bzramngvcrficy2no44	451	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	451	21	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	451	22	.	.	.
work_v44vkh6bzramngvcrficy2no44	452	1	ACM	ACM	NNP
work_v44vkh6bzramngvcrficy2no44	452	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	453	1	Percy	Percy	NNP
work_v44vkh6bzramngvcrficy2no44	453	2	Liang	Liang	NNP
work_v44vkh6bzramngvcrficy2no44	453	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	453	4	Ben	Ben	NNP
work_v44vkh6bzramngvcrficy2no44	453	5	Taskar	Taskar	NNP
work_v44vkh6bzramngvcrficy2no44	453	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	453	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	453	8	Dan	Dan	NNP
work_v44vkh6bzramngvcrficy2no44	453	9	Klein	Klein	NNP
work_v44vkh6bzramngvcrficy2no44	453	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	454	1	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	454	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	455	1	Align-	Align-	NNP
work_v44vkh6bzramngvcrficy2no44	455	2	ment	ment	JJ
work_v44vkh6bzramngvcrficy2no44	455	3	by	by	IN
work_v44vkh6bzramngvcrficy2no44	455	4	agreement	agreement	NN
work_v44vkh6bzramngvcrficy2no44	455	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	456	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	456	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	456	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	456	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	456	5	Human	Human	NNP
work_v44vkh6bzramngvcrficy2no44	456	6	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	456	7	Technology	Technology	NNP
work_v44vkh6bzramngvcrficy2no44	456	8	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	456	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	456	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	456	11	NAACL	naacl	NN
work_v44vkh6bzramngvcrficy2no44	456	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	456	13	Main	Main	NNP
work_v44vkh6bzramngvcrficy2no44	456	14	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	456	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	456	16	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	456	17	104–111	104–111	CD
work_v44vkh6bzramngvcrficy2no44	456	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	456	19	New	New	NNP
work_v44vkh6bzramngvcrficy2no44	456	20	York	York	NNP
work_v44vkh6bzramngvcrficy2no44	456	21	City	City	NNP
work_v44vkh6bzramngvcrficy2no44	456	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	456	23	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	456	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	456	25	June	June	NNP
work_v44vkh6bzramngvcrficy2no44	456	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	457	1	Ryan	Ryan	NNP
work_v44vkh6bzramngvcrficy2no44	457	2	McDonald	McDonald	NNP
work_v44vkh6bzramngvcrficy2no44	457	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	457	4	Fernando	Fernando	NNP
work_v44vkh6bzramngvcrficy2no44	457	5	Pereira	Pereira	NNP
work_v44vkh6bzramngvcrficy2no44	457	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	457	7	Kiril	Kiril	NNP
work_v44vkh6bzramngvcrficy2no44	457	8	Ribarov	Ribarov	NNP
work_v44vkh6bzramngvcrficy2no44	457	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	457	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	457	11	Jan	Jan	NNP
work_v44vkh6bzramngvcrficy2no44	457	12	Hajič.	Hajič.	NNP
work_v44vkh6bzramngvcrficy2no44	458	1	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	458	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	459	1	Non	non	JJ
work_v44vkh6bzramngvcrficy2no44	459	2	-	-	JJ
work_v44vkh6bzramngvcrficy2no44	459	3	projective	projective	JJ
work_v44vkh6bzramngvcrficy2no44	459	4	dependency	dependency	NN
work_v44vkh6bzramngvcrficy2no44	459	5	parsing	parsing	NN
work_v44vkh6bzramngvcrficy2no44	459	6	using	use	VBG
work_v44vkh6bzramngvcrficy2no44	459	7	spanning	span	VBG
work_v44vkh6bzramngvcrficy2no44	459	8	tree	tree	NN
work_v44vkh6bzramngvcrficy2no44	459	9	algorithms	algorithm	NNS
work_v44vkh6bzramngvcrficy2no44	459	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	460	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	460	2	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	460	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	460	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	460	5	conference	conference	NN
work_v44vkh6bzramngvcrficy2no44	460	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	460	7	Human	Human	NNP
work_v44vkh6bzramngvcrficy2no44	460	8	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	460	9	Technology	Technology	NNP
work_v44vkh6bzramngvcrficy2no44	460	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	460	11	Em-	Em-	NNP
work_v44vkh6bzramngvcrficy2no44	460	12	pirical	pirical	JJ
work_v44vkh6bzramngvcrficy2no44	460	13	Methods	method	NNS
work_v44vkh6bzramngvcrficy2no44	460	14	in	in	IN
work_v44vkh6bzramngvcrficy2no44	460	15	Natural	Natural	NNP
work_v44vkh6bzramngvcrficy2no44	460	16	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	460	17	Processing	Processing	NNP
work_v44vkh6bzramngvcrficy2no44	460	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	460	19	HLT	HLT	NNP
work_v44vkh6bzramngvcrficy2no44	460	20	’	'	''
work_v44vkh6bzramngvcrficy2no44	460	21	05	05	CD
work_v44vkh6bzramngvcrficy2no44	460	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	460	23	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	460	24	523–530	523–530	CD
work_v44vkh6bzramngvcrficy2no44	460	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	460	26	Stroudsburg	Stroudsburg	NNP
work_v44vkh6bzramngvcrficy2no44	460	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	460	28	PA	PA	NNP
work_v44vkh6bzramngvcrficy2no44	460	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	460	30	USA	USA	NNP
work_v44vkh6bzramngvcrficy2no44	460	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	461	1	Robert	Robert	NNP
work_v44vkh6bzramngvcrficy2no44	461	2	C.	C.	NNP
work_v44vkh6bzramngvcrficy2no44	461	3	Moore	Moore	NNP
work_v44vkh6bzramngvcrficy2no44	461	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	461	5	Chris	Chris	NNP
work_v44vkh6bzramngvcrficy2no44	461	6	Quirk	Quirk	NNP
work_v44vkh6bzramngvcrficy2no44	461	7	.	.	.
work_v44vkh6bzramngvcrficy2no44	462	1	2007	2007	CD
work_v44vkh6bzramngvcrficy2no44	462	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	463	1	Faster	fast	RBR
work_v44vkh6bzramngvcrficy2no44	463	2	beam-	beam-	NN
work_v44vkh6bzramngvcrficy2no44	463	3	search	search	NN
work_v44vkh6bzramngvcrficy2no44	463	4	decoding	decode	VBG
work_v44vkh6bzramngvcrficy2no44	463	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	463	6	phrasal	phrasal	JJ
work_v44vkh6bzramngvcrficy2no44	463	7	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	463	8	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	463	9	transla-	transla-	XX
work_v44vkh6bzramngvcrficy2no44	463	10	tion	tion	NN
work_v44vkh6bzramngvcrficy2no44	463	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	464	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	464	2	In	in	IN
work_v44vkh6bzramngvcrficy2no44	464	3	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	464	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	464	5	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	464	6	Summit	Summit	NNP
work_v44vkh6bzramngvcrficy2no44	464	7	XI	XI	NNP
work_v44vkh6bzramngvcrficy2no44	464	8	,	,	,
work_v44vkh6bzramngvcrficy2no44	464	9	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	464	10	321	321	CD
work_v44vkh6bzramngvcrficy2no44	464	11	–	–	:
work_v44vkh6bzramngvcrficy2no44	464	12	327	327	CD
work_v44vkh6bzramngvcrficy2no44	464	13	,	,	,
work_v44vkh6bzramngvcrficy2no44	464	14	Copenhagen	Copenhagen	NNP
work_v44vkh6bzramngvcrficy2no44	464	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	464	16	Denmark	Denmark	NNP
work_v44vkh6bzramngvcrficy2no44	464	17	.	.	.
work_v44vkh6bzramngvcrficy2no44	465	1	F.	F.	NNP
work_v44vkh6bzramngvcrficy2no44	465	2	Och	Och	NNP
work_v44vkh6bzramngvcrficy2no44	465	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	465	4	H.	H.	NNP
work_v44vkh6bzramngvcrficy2no44	465	5	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	465	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	466	1	2002	2002	CD
work_v44vkh6bzramngvcrficy2no44	466	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	467	1	Discriminative	discriminative	JJ
work_v44vkh6bzramngvcrficy2no44	467	2	training	training	NN
work_v44vkh6bzramngvcrficy2no44	467	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	467	4	maximum	maximum	JJ
work_v44vkh6bzramngvcrficy2no44	467	5	entropy	entropy	JJ
work_v44vkh6bzramngvcrficy2no44	467	6	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	467	7	for	for	IN
work_v44vkh6bzramngvcrficy2no44	467	8	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	467	9	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	467	10	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	467	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	468	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	468	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	468	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	468	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	468	5	40th	40th	JJ
work_v44vkh6bzramngvcrficy2no44	468	6	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	468	7	Meet-	Meet-	NNP
work_v44vkh6bzramngvcrficy2no44	468	8	ing	ing	NN
work_v44vkh6bzramngvcrficy2no44	468	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	468	10	the	the	DT
work_v44vkh6bzramngvcrficy2no44	468	11	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	468	12	for	for	IN
work_v44vkh6bzramngvcrficy2no44	468	13	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	468	14	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	468	15	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	468	16	ACL	ACL	NNP
work_v44vkh6bzramngvcrficy2no44	468	17	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	468	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	468	19	pages	page	NNS
work_v44vkh6bzramngvcrficy2no44	468	20	295–302	295–302	CD
work_v44vkh6bzramngvcrficy2no44	468	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	468	22	Philadelhpia	Philadelhpia	NNP
work_v44vkh6bzramngvcrficy2no44	468	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	468	24	PA	PA	NNP
work_v44vkh6bzramngvcrficy2no44	468	25	.	.	.
work_v44vkh6bzramngvcrficy2no44	468	26	Franz	Franz	NNP
work_v44vkh6bzramngvcrficy2no44	468	27	Josef	Josef	NNP
work_v44vkh6bzramngvcrficy2no44	468	28	Och	Och	NNP
work_v44vkh6bzramngvcrficy2no44	468	29	.	.	.
work_v44vkh6bzramngvcrficy2no44	469	1	2003	2003	CD
work_v44vkh6bzramngvcrficy2no44	469	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	470	1	Minimum	Minimum	NNP
work_v44vkh6bzramngvcrficy2no44	470	2	Error	Error	NNP
work_v44vkh6bzramngvcrficy2no44	470	3	Rate	rate	NN
work_v44vkh6bzramngvcrficy2no44	470	4	Training	Training	NNP
work_v44vkh6bzramngvcrficy2no44	470	5	in	in	IN
work_v44vkh6bzramngvcrficy2no44	470	6	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	470	7	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	470	8	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	470	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	471	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	471	2	Erhard	Erhard	NNP
work_v44vkh6bzramngvcrficy2no44	471	3	Hinrichs	Hinrichs	NNP
work_v44vkh6bzramngvcrficy2no44	471	4	and	and	CC
work_v44vkh6bzramngvcrficy2no44	471	5	Dan	Dan	NNP
work_v44vkh6bzramngvcrficy2no44	471	6	Roth	Roth	NNP
work_v44vkh6bzramngvcrficy2no44	471	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	471	8	editors	editor	NNS
work_v44vkh6bzramngvcrficy2no44	471	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	471	10	Proceedings	Proceedings	NNPS
work_v44vkh6bzramngvcrficy2no44	471	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	471	12	the	the	DT
work_v44vkh6bzramngvcrficy2no44	471	13	41st	41st	JJ
work_v44vkh6bzramngvcrficy2no44	471	14	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	471	15	Meeting	Meeting	NNP
work_v44vkh6bzramngvcrficy2no44	471	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	471	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	471	18	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	471	19	for	for	IN
work_v44vkh6bzramngvcrficy2no44	471	20	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	471	21	Linguis-	Linguis-	NNP
work_v44vkh6bzramngvcrficy2no44	471	22	tics	tic	NNS
work_v44vkh6bzramngvcrficy2no44	471	23	,	,	,
work_v44vkh6bzramngvcrficy2no44	471	24	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	471	25	160–167	160–167	CD
work_v44vkh6bzramngvcrficy2no44	471	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	472	1	Kishore	Kishore	NNP
work_v44vkh6bzramngvcrficy2no44	472	2	Papineni	Papineni	NNP
work_v44vkh6bzramngvcrficy2no44	472	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	472	4	Salim	Salim	NNP
work_v44vkh6bzramngvcrficy2no44	472	5	Roukos	Roukos	NNP
work_v44vkh6bzramngvcrficy2no44	472	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	472	7	Todd	Todd	NNP
work_v44vkh6bzramngvcrficy2no44	472	8	Ward	Ward	NNP
work_v44vkh6bzramngvcrficy2no44	472	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	472	10	and	and	CC
work_v44vkh6bzramngvcrficy2no44	472	11	Wei-	Wei-	NNP
work_v44vkh6bzramngvcrficy2no44	472	12	Jing	Jing	NNP
work_v44vkh6bzramngvcrficy2no44	472	13	Zhu	Zhu	NNP
work_v44vkh6bzramngvcrficy2no44	472	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	473	1	2002	2002	CD
work_v44vkh6bzramngvcrficy2no44	473	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	474	1	BLEU	BLEU	NNP
work_v44vkh6bzramngvcrficy2no44	474	2	:	:	:
work_v44vkh6bzramngvcrficy2no44	474	3	a	a	DT
work_v44vkh6bzramngvcrficy2no44	474	4	method	method	NN
work_v44vkh6bzramngvcrficy2no44	474	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	474	6	automatic	automatic	JJ
work_v44vkh6bzramngvcrficy2no44	474	7	eval-	eval-	NN
work_v44vkh6bzramngvcrficy2no44	474	8	uation	uation	NN
work_v44vkh6bzramngvcrficy2no44	474	9	of	of	IN
work_v44vkh6bzramngvcrficy2no44	474	10	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	474	11	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	474	12	.	.	.
work_v44vkh6bzramngvcrficy2no44	475	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	475	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	475	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	475	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	475	5	40th	40th	JJ
work_v44vkh6bzramngvcrficy2no44	475	6	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	475	7	Meeting	Meeting	NNP
work_v44vkh6bzramngvcrficy2no44	475	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	475	9	the	the	DT
work_v44vkh6bzramngvcrficy2no44	475	10	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	475	11	of	of	IN
work_v44vkh6bzramngvcrficy2no44	475	12	Compu-	Compu-	NNP
work_v44vkh6bzramngvcrficy2no44	475	13	tational	tational	NN
work_v44vkh6bzramngvcrficy2no44	475	14	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	475	15	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	475	16	ACL	ACL	NNP
work_v44vkh6bzramngvcrficy2no44	475	17	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	475	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	475	19	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	475	20	311–318	311–318	CD
work_v44vkh6bzramngvcrficy2no44	475	21	,	,	,
work_v44vkh6bzramngvcrficy2no44	475	22	Philadel-	Philadel-	NNP
work_v44vkh6bzramngvcrficy2no44	475	23	phia	phia	NN
work_v44vkh6bzramngvcrficy2no44	475	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	475	25	PA	PA	NNP
work_v44vkh6bzramngvcrficy2no44	475	26	.	.	.
work_v44vkh6bzramngvcrficy2no44	475	27	Stefan	Stefan	NNP
work_v44vkh6bzramngvcrficy2no44	475	28	Riezler	Riezler	NNP
work_v44vkh6bzramngvcrficy2no44	475	29	and	and	CC
work_v44vkh6bzramngvcrficy2no44	475	30	John	John	NNP
work_v44vkh6bzramngvcrficy2no44	475	31	T.	T.	NNP
work_v44vkh6bzramngvcrficy2no44	475	32	Maxwell	Maxwell	NNP
work_v44vkh6bzramngvcrficy2no44	475	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	476	1	2005	2005	CD
work_v44vkh6bzramngvcrficy2no44	476	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	477	1	On	on	IN
work_v44vkh6bzramngvcrficy2no44	477	2	some	some	DT
work_v44vkh6bzramngvcrficy2no44	477	3	pitfalls	pitfall	NNS
work_v44vkh6bzramngvcrficy2no44	477	4	in	in	IN
work_v44vkh6bzramngvcrficy2no44	477	5	automatic	automatic	JJ
work_v44vkh6bzramngvcrficy2no44	477	6	evaluation	evaluation	NN
work_v44vkh6bzramngvcrficy2no44	477	7	and	and	CC
work_v44vkh6bzramngvcrficy2no44	477	8	significance	significance	NN
work_v44vkh6bzramngvcrficy2no44	477	9	test-	test-	NN
work_v44vkh6bzramngvcrficy2no44	477	10	ing	ing	NNP
work_v44vkh6bzramngvcrficy2no44	477	11	for	for	IN
work_v44vkh6bzramngvcrficy2no44	477	12	MT	MT	NNP
work_v44vkh6bzramngvcrficy2no44	477	13	.	.	.
work_v44vkh6bzramngvcrficy2no44	478	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	478	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	478	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	478	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	478	5	ACL	ACL	NNP
work_v44vkh6bzramngvcrficy2no44	478	6	Workshop	Workshop	NNP
work_v44vkh6bzramngvcrficy2no44	478	7	on	on	IN
work_v44vkh6bzramngvcrficy2no44	478	8	Intrinsic	Intrinsic	NNP
work_v44vkh6bzramngvcrficy2no44	478	9	and	and	CC
work_v44vkh6bzramngvcrficy2no44	478	10	Extrinsic	Extrinsic	NNP
work_v44vkh6bzramngvcrficy2no44	478	11	Evaluation	Evaluation	NNP
work_v44vkh6bzramngvcrficy2no44	478	12	Measures	Measures	NNPS
work_v44vkh6bzramngvcrficy2no44	478	13	for	for	IN
work_v44vkh6bzramngvcrficy2no44	478	14	Ma-	ma-	JJ
work_v44vkh6bzramngvcrficy2no44	478	15	chine	chine	NN
work_v44vkh6bzramngvcrficy2no44	478	16	Translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	478	17	and/or	and/or	CC
work_v44vkh6bzramngvcrficy2no44	478	18	Summarization	Summarization	NNP
work_v44vkh6bzramngvcrficy2no44	478	19	,	,	,
work_v44vkh6bzramngvcrficy2no44	478	20	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	478	21	57	57	CD
work_v44vkh6bzramngvcrficy2no44	478	22	–	–	:
work_v44vkh6bzramngvcrficy2no44	478	23	64	64	CD
work_v44vkh6bzramngvcrficy2no44	478	24	,	,	,
work_v44vkh6bzramngvcrficy2no44	478	25	Ann	Ann	NNP
work_v44vkh6bzramngvcrficy2no44	478	26	Arbor	Arbor	NNP
work_v44vkh6bzramngvcrficy2no44	478	27	,	,	,
work_v44vkh6bzramngvcrficy2no44	478	28	Michigan	Michigan	NNP
work_v44vkh6bzramngvcrficy2no44	478	29	,	,	,
work_v44vkh6bzramngvcrficy2no44	478	30	June	June	NNP
work_v44vkh6bzramngvcrficy2no44	478	31	.	.	.
work_v44vkh6bzramngvcrficy2no44	479	1	Christoph	Christoph	NNP
work_v44vkh6bzramngvcrficy2no44	479	2	Tillmann	Tillmann	NNP
work_v44vkh6bzramngvcrficy2no44	479	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	480	1	2004	2004	CD
work_v44vkh6bzramngvcrficy2no44	480	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	481	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	481	2	Unigram	Unigram	NNP
work_v44vkh6bzramngvcrficy2no44	481	3	Orientation	Orientation	NNP
work_v44vkh6bzramngvcrficy2no44	481	4	Model	Model	NNP
work_v44vkh6bzramngvcrficy2no44	481	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	481	6	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	481	7	Machine	Machine	NNP
work_v44vkh6bzramngvcrficy2no44	481	8	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	481	9	.	.	.
work_v44vkh6bzramngvcrficy2no44	482	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	482	2	Pro-	Pro-	NNP
work_v44vkh6bzramngvcrficy2no44	482	3	ceedings	ceeding	NNS
work_v44vkh6bzramngvcrficy2no44	482	4	of	of	IN
work_v44vkh6bzramngvcrficy2no44	482	5	the	the	DT
work_v44vkh6bzramngvcrficy2no44	482	6	Joint	Joint	NNP
work_v44vkh6bzramngvcrficy2no44	482	7	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	482	8	on	on	IN
work_v44vkh6bzramngvcrficy2no44	482	9	Human	Human	NNP
work_v44vkh6bzramngvcrficy2no44	482	10	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	482	11	Technologies	Technologies	NNPS
work_v44vkh6bzramngvcrficy2no44	482	12	and	and	CC
work_v44vkh6bzramngvcrficy2no44	482	13	the	the	DT
work_v44vkh6bzramngvcrficy2no44	482	14	Annual	Annual	NNP
work_v44vkh6bzramngvcrficy2no44	482	15	Meeting	Meeting	NNP
work_v44vkh6bzramngvcrficy2no44	482	16	of	of	IN
work_v44vkh6bzramngvcrficy2no44	482	17	the	the	DT
work_v44vkh6bzramngvcrficy2no44	482	18	North	north	JJ
work_v44vkh6bzramngvcrficy2no44	482	19	American	American	NNP
work_v44vkh6bzramngvcrficy2no44	482	20	Chapter	Chapter	NNP
work_v44vkh6bzramngvcrficy2no44	482	21	of	of	IN
work_v44vkh6bzramngvcrficy2no44	482	22	the	the	DT
work_v44vkh6bzramngvcrficy2no44	482	23	Association	Association	NNP
work_v44vkh6bzramngvcrficy2no44	482	24	of	of	IN
work_v44vkh6bzramngvcrficy2no44	482	25	Computa-	Computa-	NNP
work_v44vkh6bzramngvcrficy2no44	482	26	tional	tional	JJ
work_v44vkh6bzramngvcrficy2no44	482	27	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	482	28	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	482	29	HLT	HLT	NNP
work_v44vkh6bzramngvcrficy2no44	482	30	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	482	31	NAACL	NAACL	NNP
work_v44vkh6bzramngvcrficy2no44	482	32	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	482	33	.	.	.
work_v44vkh6bzramngvcrficy2no44	483	1	Karthik	Karthik	NNP
work_v44vkh6bzramngvcrficy2no44	483	2	Visweswariah	Visweswariah	NNP
work_v44vkh6bzramngvcrficy2no44	483	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	483	4	Rajakrishnan	Rajakrishnan	NNP
work_v44vkh6bzramngvcrficy2no44	483	5	Rajkumar	Rajkumar	NNP
work_v44vkh6bzramngvcrficy2no44	483	6	,	,	,
work_v44vkh6bzramngvcrficy2no44	483	7	Ankur	Ankur	NNP
work_v44vkh6bzramngvcrficy2no44	483	8	Gandhe	Gandhe	NNP
work_v44vkh6bzramngvcrficy2no44	483	9	,	,	,
work_v44vkh6bzramngvcrficy2no44	483	10	Ananthakrishnan	Ananthakrishnan	NNP
work_v44vkh6bzramngvcrficy2no44	483	11	Ramanathan	Ramanathan	NNP
work_v44vkh6bzramngvcrficy2no44	483	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	483	13	and	and	CC
work_v44vkh6bzramngvcrficy2no44	483	14	Jiri	Jiri	NNP
work_v44vkh6bzramngvcrficy2no44	483	15	Navratil	Navratil	NNP
work_v44vkh6bzramngvcrficy2no44	483	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	484	1	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	484	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	485	1	A	a	DT
work_v44vkh6bzramngvcrficy2no44	485	2	word	word	NN
work_v44vkh6bzramngvcrficy2no44	485	3	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	485	4	model	model	NN
work_v44vkh6bzramngvcrficy2no44	485	5	for	for	IN
work_v44vkh6bzramngvcrficy2no44	485	6	im-	im-	JJ
work_v44vkh6bzramngvcrficy2no44	485	7	proved	proved	JJ
work_v44vkh6bzramngvcrficy2no44	485	8	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	485	9	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	485	10	.	.	.
work_v44vkh6bzramngvcrficy2no44	486	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	486	2	Proceedings	Proceedings	NNP
work_v44vkh6bzramngvcrficy2no44	486	3	of	of	IN
work_v44vkh6bzramngvcrficy2no44	486	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	486	5	2011	2011	CD
work_v44vkh6bzramngvcrficy2no44	486	6	Conference	Conference	NNP
work_v44vkh6bzramngvcrficy2no44	486	7	on	on	IN
work_v44vkh6bzramngvcrficy2no44	486	8	Empirical	Empirical	NNP
work_v44vkh6bzramngvcrficy2no44	486	9	Methods	Methods	NNPS
work_v44vkh6bzramngvcrficy2no44	486	10	in	in	IN
work_v44vkh6bzramngvcrficy2no44	486	11	Natu-	Natu-	NNP
work_v44vkh6bzramngvcrficy2no44	486	12	ral	ral	JJ
work_v44vkh6bzramngvcrficy2no44	486	13	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	486	14	Processing	Processing	NNP
work_v44vkh6bzramngvcrficy2no44	486	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	486	16	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	486	17	486–496	486–496	CD
work_v44vkh6bzramngvcrficy2no44	486	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	486	19	Edinburgh	Edinburgh	NNP
work_v44vkh6bzramngvcrficy2no44	486	20	,	,	,
work_v44vkh6bzramngvcrficy2no44	486	21	Scotland	Scotland	NNP
work_v44vkh6bzramngvcrficy2no44	486	22	,	,	,
work_v44vkh6bzramngvcrficy2no44	486	23	UK	UK	NNP
work_v44vkh6bzramngvcrficy2no44	486	24	.	.	.
work_v44vkh6bzramngvcrficy2no44	486	25	,	,	,
work_v44vkh6bzramngvcrficy2no44	486	26	July	July	NNP
work_v44vkh6bzramngvcrficy2no44	486	27	.	.	.
work_v44vkh6bzramngvcrficy2no44	487	1	Dekai	Dekai	NNP
work_v44vkh6bzramngvcrficy2no44	487	2	Wu	Wu	NNP
work_v44vkh6bzramngvcrficy2no44	487	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	488	1	1997	1997	CD
work_v44vkh6bzramngvcrficy2no44	488	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	489	1	Stochastic	stochastic	JJ
work_v44vkh6bzramngvcrficy2no44	489	2	inversion	inversion	NN
work_v44vkh6bzramngvcrficy2no44	489	3	transduction	transduction	NN
work_v44vkh6bzramngvcrficy2no44	489	4	grammars	grammar	NNS
work_v44vkh6bzramngvcrficy2no44	489	5	and	and	CC
work_v44vkh6bzramngvcrficy2no44	489	6	bilingual	bilingual	JJ
work_v44vkh6bzramngvcrficy2no44	489	7	parsing	parsing	NN
work_v44vkh6bzramngvcrficy2no44	489	8	of	of	IN
work_v44vkh6bzramngvcrficy2no44	489	9	parallel	parallel	JJ
work_v44vkh6bzramngvcrficy2no44	489	10	corpora	corpora	NN
work_v44vkh6bzramngvcrficy2no44	489	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	490	1	Computational	Computational	NNP
work_v44vkh6bzramngvcrficy2no44	490	2	Linguistics	Linguistics	NNP
work_v44vkh6bzramngvcrficy2no44	490	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	490	4	23(3):377–403	23(3):377–403	CD
work_v44vkh6bzramngvcrficy2no44	490	5	.	.	.
work_v44vkh6bzramngvcrficy2no44	491	1	Sirvan	Sirvan	NNP
work_v44vkh6bzramngvcrficy2no44	491	2	Yahyaei	Yahyaei	NNP
work_v44vkh6bzramngvcrficy2no44	491	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	491	4	Christof	Christof	NNP
work_v44vkh6bzramngvcrficy2no44	491	5	Monz	Monz	NNP
work_v44vkh6bzramngvcrficy2no44	491	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	492	1	2010	2010	CD
work_v44vkh6bzramngvcrficy2no44	492	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	493	1	Dynamic	dynamic	JJ
work_v44vkh6bzramngvcrficy2no44	493	2	dis-	dis-	VBP
work_v44vkh6bzramngvcrficy2no44	493	3	tortion	tortion	NN
work_v44vkh6bzramngvcrficy2no44	493	4	in	in	IN
work_v44vkh6bzramngvcrficy2no44	493	5	a	a	DT
work_v44vkh6bzramngvcrficy2no44	493	6	discriminative	discriminative	JJ
work_v44vkh6bzramngvcrficy2no44	493	7	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	493	8	model	model	NN
work_v44vkh6bzramngvcrficy2no44	493	9	for	for	IN
work_v44vkh6bzramngvcrficy2no44	493	10	sta-	sta-	JJ
work_v44vkh6bzramngvcrficy2no44	493	11	tistical	tistical	JJ
work_v44vkh6bzramngvcrficy2no44	493	12	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	493	13	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	493	14	.	.	.
work_v44vkh6bzramngvcrficy2no44	494	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	494	2	International	International	NNP
work_v44vkh6bzramngvcrficy2no44	494	3	Work-	Work-	NNP
work_v44vkh6bzramngvcrficy2no44	494	4	shop	shop	NN
work_v44vkh6bzramngvcrficy2no44	494	5	on	on	IN
work_v44vkh6bzramngvcrficy2no44	494	6	Spoken	Spoken	NNP
work_v44vkh6bzramngvcrficy2no44	494	7	Language	Language	NNP
work_v44vkh6bzramngvcrficy2no44	494	8	Translation	Translation	NNP
work_v44vkh6bzramngvcrficy2no44	494	9	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	494	10	IWSLT	IWSLT	NNP
work_v44vkh6bzramngvcrficy2no44	494	11	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	494	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	494	13	Paris	Paris	NNP
work_v44vkh6bzramngvcrficy2no44	494	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	494	15	France	France	NNP
work_v44vkh6bzramngvcrficy2no44	494	16	.	.	.
work_v44vkh6bzramngvcrficy2no44	495	1	Richard	Richard	NNP
work_v44vkh6bzramngvcrficy2no44	495	2	Zens	Zens	NNP
work_v44vkh6bzramngvcrficy2no44	495	3	and	and	CC
work_v44vkh6bzramngvcrficy2no44	495	4	Hermann	Hermann	NNP
work_v44vkh6bzramngvcrficy2no44	495	5	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	495	6	.	.	.
work_v44vkh6bzramngvcrficy2no44	496	1	2006	2006	CD
work_v44vkh6bzramngvcrficy2no44	496	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	497	1	Discriminative	discriminative	JJ
work_v44vkh6bzramngvcrficy2no44	497	2	reordering	reordering	NN
work_v44vkh6bzramngvcrficy2no44	497	3	models	model	NNS
work_v44vkh6bzramngvcrficy2no44	497	4	for	for	IN
work_v44vkh6bzramngvcrficy2no44	497	5	statistical	statistical	JJ
work_v44vkh6bzramngvcrficy2no44	497	6	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	497	7	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	497	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	498	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	498	2	Proceedings	proceeding	NNS
work_v44vkh6bzramngvcrficy2no44	498	3	on	on	IN
work_v44vkh6bzramngvcrficy2no44	498	4	the	the	DT
work_v44vkh6bzramngvcrficy2no44	498	5	Workshop	Workshop	NNP
work_v44vkh6bzramngvcrficy2no44	498	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	498	7	Statistical	Statistical	NNP
work_v44vkh6bzramngvcrficy2no44	498	8	Ma-	Ma-	NNP
work_v44vkh6bzramngvcrficy2no44	498	9	chine	chine	NN
work_v44vkh6bzramngvcrficy2no44	498	10	Translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	498	11	,	,	,
work_v44vkh6bzramngvcrficy2no44	498	12	pages	page	VBZ
work_v44vkh6bzramngvcrficy2no44	498	13	55–63	55–63	CD
work_v44vkh6bzramngvcrficy2no44	498	14	,	,	,
work_v44vkh6bzramngvcrficy2no44	498	15	New	New	NNP
work_v44vkh6bzramngvcrficy2no44	498	16	York	York	NNP
work_v44vkh6bzramngvcrficy2no44	498	17	City	City	NNP
work_v44vkh6bzramngvcrficy2no44	498	18	,	,	,
work_v44vkh6bzramngvcrficy2no44	498	19	June	June	NNP
work_v44vkh6bzramngvcrficy2no44	498	20	.	.	.
work_v44vkh6bzramngvcrficy2no44	499	1	R.	R.	NNP
work_v44vkh6bzramngvcrficy2no44	499	2	Zens	Zens	NNP
work_v44vkh6bzramngvcrficy2no44	499	3	,	,	,
work_v44vkh6bzramngvcrficy2no44	499	4	F.	F.	NNP
work_v44vkh6bzramngvcrficy2no44	499	5	J.	J.	NNP
work_v44vkh6bzramngvcrficy2no44	499	6	Och	Och	NNP
work_v44vkh6bzramngvcrficy2no44	499	7	,	,	,
work_v44vkh6bzramngvcrficy2no44	499	8	and	and	CC
work_v44vkh6bzramngvcrficy2no44	499	9	H.	H.	NNP
work_v44vkh6bzramngvcrficy2no44	499	10	Ney	Ney	NNP
work_v44vkh6bzramngvcrficy2no44	499	11	.	.	.
work_v44vkh6bzramngvcrficy2no44	500	1	2002	2002	CD
work_v44vkh6bzramngvcrficy2no44	500	2	.	.	.
work_v44vkh6bzramngvcrficy2no44	501	1	Phrase	phrase	NN
work_v44vkh6bzramngvcrficy2no44	501	2	-	-	HYPH
work_v44vkh6bzramngvcrficy2no44	501	3	based	base	VBN
work_v44vkh6bzramngvcrficy2no44	501	4	sta-	sta-	NN
work_v44vkh6bzramngvcrficy2no44	501	5	tistical	tistical	JJ
work_v44vkh6bzramngvcrficy2no44	501	6	machine	machine	NN
work_v44vkh6bzramngvcrficy2no44	501	7	translation	translation	NN
work_v44vkh6bzramngvcrficy2no44	501	8	.	.	.
work_v44vkh6bzramngvcrficy2no44	502	1	In	in	IN
work_v44vkh6bzramngvcrficy2no44	502	2	25th	25th	JJ
work_v44vkh6bzramngvcrficy2no44	502	3	German	german	JJ
work_v44vkh6bzramngvcrficy2no44	502	4	Confer-	Confer-	NNP
work_v44vkh6bzramngvcrficy2no44	502	5	ence	ence	NN
work_v44vkh6bzramngvcrficy2no44	502	6	on	on	IN
work_v44vkh6bzramngvcrficy2no44	502	7	Artificial	Artificial	NNP
work_v44vkh6bzramngvcrficy2no44	502	8	Intelligence	Intelligence	NNP
work_v44vkh6bzramngvcrficy2no44	502	9	(	(	-LRB-
work_v44vkh6bzramngvcrficy2no44	502	10	KI2002	KI2002	NNP
work_v44vkh6bzramngvcrficy2no44	502	11	)	)	-RRB-
work_v44vkh6bzramngvcrficy2no44	502	12	,	,	,
work_v44vkh6bzramngvcrficy2no44	502	13	pages	page	NNS
work_v44vkh6bzramngvcrficy2no44	502	14	18–32	18–32	CD
work_v44vkh6bzramngvcrficy2no44	502	15	,	,	,
work_v44vkh6bzramngvcrficy2no44	502	16	Aachen	Aachen	NNP
work_v44vkh6bzramngvcrficy2no44	502	17	,	,	,
work_v44vkh6bzramngvcrficy2no44	502	18	Germany	Germany	NNP
work_v44vkh6bzramngvcrficy2no44	502	19	.	.	.
work_v44vkh6bzramngvcrficy2no44	503	1	Springer	Springer	NNP
work_v44vkh6bzramngvcrficy2no44	503	2	Verlag	Verlag	NNP
work_v44vkh6bzramngvcrficy2no44	503	3	.	.	.
work_v44vkh6bzramngvcrficy2no44	504	1	339	339	CD
work_v44vkh6bzramngvcrficy2no44	504	2	340	340	CD