id	sid	tid	token	lemma	pos
9019s18114q	1	1	the	the	DET
9019s18114q	1	2	traditional	traditional	ADJ
9019s18114q	1	3	method	method	NOUN
9019s18114q	1	4	for	for	ADP
9019s18114q	1	5	documenting	document	VERB
9019s18114q	1	6	a	a	DET
9019s18114q	1	7	language	language	NOUN
9019s18114q	1	8	involves	involve	VERB
9019s18114q	1	9	the	the	DET
9019s18114q	1	10	collection	collection	NOUN
9019s18114q	1	11	of	of	ADP
9019s18114q	1	12	audio	audio	NOUN
9019s18114q	1	13	or	or	CCONJ
9019s18114q	1	14	video	video	NOUN
9019s18114q	1	15	sources	source	NOUN
9019s18114q	1	16	,	,	PUNCT
9019s18114q	1	17	which	which	PRON
9019s18114q	1	18	are	be	AUX
9019s18114q	1	19	then	then	ADV
9019s18114q	1	20	annotated	annotate	VERB
9019s18114q	1	21	at	at	ADP
9019s18114q	1	22	multiple	multiple	ADJ
9019s18114q	1	23	granularity	granularity	NOUN
9019s18114q	1	24	levels	level	NOUN
9019s18114q	1	25	by	by	ADP
9019s18114q	1	26	trained	train	VERB
9019s18114q	1	27	linguists	linguist	NOUN
9019s18114q	1	28	.	.	PUNCT
9019s18114q	2	1	this	this	PRON
9019s18114q	2	2	is	be	AUX
9019s18114q	2	3	a	a	DET
9019s18114q	2	4	painstaking	painstaking	NOUN
9019s18114q	2	5	and	and	CCONJ
9019s18114q	2	6	time	time	NOUN
9019s18114q	2	7	-	-	PUNCT
9019s18114q	2	8	consuming	consume	VERB
9019s18114q	2	9	process	process	NOUN
9019s18114q	2	10	,	,	PUNCT
9019s18114q	2	11	which	which	PRON
9019s18114q	2	12	could	could	AUX
9019s18114q	2	13	benefit	benefit	VERB
9019s18114q	2	14	from	from	ADP
9019s18114q	2	15	machine	machine	NOUN
9019s18114q	2	16	learning	learning	NOUN
9019s18114q	2	17	techniques	technique	NOUN
9019s18114q	2	18	at	at	ADP
9019s18114q	2	19	almost	almost	ADV
9019s18114q	2	20	all	all	DET
9019s18114q	2	21	stages	stage	NOUN
9019s18114q	2	22	.	.	PUNCT
9019s18114q	3	1	however	however	ADV
9019s18114q	3	2	,	,	PUNCT
9019s18114q	3	3	most	most	ADJ
9019s18114q	3	4	existing	exist	VERB
9019s18114q	3	5	machine	machine	NOUN
9019s18114q	3	6	learning	learning	NOUN
9019s18114q	3	7	methods	method	NOUN
9019s18114q	3	8	are	be	AUX
9019s18114q	3	9	being	be	AUX
9019s18114q	3	10	developed	develop	VERB
9019s18114q	3	11	for	for	ADP
9019s18114q	3	12	high	high	ADJ
9019s18114q	3	13	-	-	PUNCT
9019s18114q	3	14	resource	resource	NOUN
9019s18114q	3	15	languages	language	NOUN
9019s18114q	3	16	and	and	CCONJ
9019s18114q	3	17	rely	rely	VERB
9019s18114q	3	18	on	on	ADP
9019s18114q	3	19	abundant	abundant	ADJ
9019s18114q	3	20	data	datum	NOUN
9019s18114q	3	21	,	,	PUNCT
9019s18114q	3	22	rendering	render	VERB
9019s18114q	3	23	them	they	PRON
9019s18114q	3	24	unsuitable	unsuitable	ADJ
9019s18114q	3	25	for	for	ADP
9019s18114q	3	26	such	such	ADJ
9019s18114q	3	27	applications	application	NOUN
9019s18114q	3	28	.	.	PUNCT
9019s18114q	4	1	at	at	ADP
9019s18114q	4	2	the	the	DET
9019s18114q	4	3	same	same	ADJ
9019s18114q	4	4	time	time	NOUN
9019s18114q	4	5	,	,	PUNCT
9019s18114q	4	6	for	for	ADP
9019s18114q	4	7	many	many	ADJ
9019s18114q	4	8	low	low	ADJ
9019s18114q	4	9	-	-	PUNCT
9019s18114q	4	10	resource	resource	NOUN
9019s18114q	4	11	and	and	CCONJ
9019s18114q	4	12	endangered	endangered	PROPN
9019s18114q	4	13	languages	language	NOUN
9019s18114q	4	14	speech	speech	NOUN
9019s18114q	4	15	data	datum	NOUN
9019s18114q	4	16	is	be	AUX
9019s18114q	4	17	easier	easy	ADJ
9019s18114q	4	18	to	to	PART
9019s18114q	4	19	obtain	obtain	VERB
9019s18114q	4	20	than	than	ADP
9019s18114q	4	21	textual	textual	ADJ
9019s18114q	4	22	data	datum	NOUN
9019s18114q	4	23	,	,	PUNCT
9019s18114q	4	24	particularly	particularly	ADV
9019s18114q	4	25	since	since	SCONJ
9019s18114q	4	26	most	most	ADJ
9019s18114q	4	27	of	of	ADP
9019s18114q	4	28	the	the	DET
9019s18114q	4	29	world	world	NOUN
9019s18114q	4	30	's	's	PART
9019s18114q	4	31	languages	language	NOUN
9019s18114q	4	32	are	be	AUX
9019s18114q	4	33	unwritten	unwritten	VERB
9019s18114q	4	34	.	.	PUNCT
9019s18114q	5	1	nevertheless	nevertheless	ADV
9019s18114q	5	2	,	,	PUNCT
9019s18114q	5	3	it	it	PRON
9019s18114q	5	4	is	be	AUX
9019s18114q	5	5	relatively	relatively	ADV
9019s18114q	5	6	easy	easy	ADJ
9019s18114q	5	7	to	to	PART
9019s18114q	5	8	provide	provide	VERB
9019s18114q	5	9	written	write	VERB
9019s18114q	5	10	or	or	CCONJ
9019s18114q	5	11	spoken	speak	VERB
9019s18114q	5	12	translations	translation	NOUN
9019s18114q	5	13	for	for	ADP
9019s18114q	5	14	audio	audio	ADJ
9019s18114q	5	15	sources	source	NOUN
9019s18114q	5	16	,	,	PUNCT
9019s18114q	5	17	as	as	SCONJ
9019s18114q	5	18	speakers	speaker	NOUN
9019s18114q	5	19	of	of	ADP
9019s18114q	5	20	a	a	DET
9019s18114q	5	21	minority	minority	NOUN
9019s18114q	5	22	language	language	NOUN
9019s18114q	5	23	are	be	AUX
9019s18114q	5	24	often	often	ADV
9019s18114q	5	25	bilingual	bilingual	ADJ
9019s18114q	5	26	and	and	CCONJ
9019s18114q	5	27	literate	literate	ADJ
9019s18114q	5	28	in	in	ADP
9019s18114q	5	29	a	a	DET
9019s18114q	5	30	high	high	ADJ
9019s18114q	5	31	-	-	PUNCT
9019s18114q	5	32	resource	resource	NOUN
9019s18114q	5	33	language	language	NOUN
9019s18114q	5	34	.	.	PUNCT
9019s18114q	6	1	this	this	DET
9019s18114q	6	2	work	work	NOUN
9019s18114q	6	3	is	be	AUX
9019s18114q	6	4	aimed	aim	VERB
9019s18114q	6	5	at	at	ADP
9019s18114q	6	6	solving	solve	VERB
9019s18114q	6	7	certain	certain	ADJ
9019s18114q	6	8	problems	problem	NOUN
9019s18114q	6	9	that	that	PRON
9019s18114q	6	10	arise	arise	VERB
9019s18114q	6	11	in	in	ADP
9019s18114q	6	12	the	the	DET
9019s18114q	6	13	documentation	documentation	NOUN
9019s18114q	6	14	process	process	NOUN
9019s18114q	6	15	of	of	ADP
9019s18114q	6	16	an	an	DET
9019s18114q	6	17	endangered	endangered	ADJ
9019s18114q	6	18	language	language	NOUN
9019s18114q	6	19	,	,	PUNCT
9019s18114q	6	20	due	due	ADP
9019s18114q	6	21	to	to	ADP
9019s18114q	6	22	the	the	DET
9019s18114q	6	23	minimal	minimal	ADJ
9019s18114q	6	24	annotated	annotate	VERB
9019s18114q	6	25	resources	resource	NOUN
9019s18114q	6	26	that	that	PRON
9019s18114q	6	27	are	be	AUX
9019s18114q	6	28	available	available	ADJ
9019s18114q	6	29	at	at	ADP
9019s18114q	6	30	this	this	DET
9019s18114q	6	31	stage	stage	NOUN
9019s18114q	6	32	.	.	PUNCT
9019s18114q	7	1	this	this	DET
9019s18114q	7	2	dissertation	dissertation	NOUN
9019s18114q	7	3	mainly	mainly	ADV
9019s18114q	7	4	focuses	focus	VERB
9019s18114q	7	5	on	on	ADP
9019s18114q	7	6	spoken	spoken	PROPN
9019s18114q	7	7	corpora	corpora	PROPN
9019s18114q	7	8	of	of	ADP
9019s18114q	7	9	endangered	endangered	PROPN
9019s18114q	7	10	and	and	CCONJ
9019s18114q	7	11	low	low	ADJ
9019s18114q	7	12	-	-	PUNCT
9019s18114q	7	13	resource	resource	NOUN
9019s18114q	7	14	languages	language	NOUN
9019s18114q	7	15	with	with	ADP
9019s18114q	7	16	limited	limited	ADJ
9019s18114q	7	17	translation	translation	NOUN
9019s18114q	7	18	annotations	annotation	NOUN
9019s18114q	7	19	,	,	PUNCT
9019s18114q	7	20	tackling	tackle	VERB
9019s18114q	7	21	problems	problem	NOUN
9019s18114q	7	22	that	that	PRON
9019s18114q	7	23	cover	cover	VERB
9019s18114q	7	24	every	every	DET
9019s18114q	7	25	layer	layer	NOUN
9019s18114q	7	26	of	of	ADP
9019s18114q	7	27	linguistic	linguistic	ADJ
9019s18114q	7	28	annotation	annotation	NOUN
9019s18114q	7	29	:	:	PUNCT
9019s18114q	7	30	speech	speech	NOUN
9019s18114q	7	31	-	-	PUNCT
9019s18114q	7	32	to	to	ADP
9019s18114q	7	33	-	-	PUNCT
9019s18114q	7	34	translation	translation	NOUN
9019s18114q	7	35	alignment	alignment	NOUN
9019s18114q	7	36	:	:	PUNCT
9019s18114q	7	37	we	we	PRON
9019s18114q	7	38	present	present	VERB
9019s18114q	7	39	an	an	DET
9019s18114q	7	40	unsupervised	unsupervised	ADJ
9019s18114q	7	41	method	method	NOUN
9019s18114q	7	42	for	for	ADP
9019s18114q	7	43	discovering	discover	VERB
9019s18114q	7	44	word	word	NOUN
9019s18114q	7	45	or	or	CCONJ
9019s18114q	7	46	phrase	phrase	NOUN
9019s18114q	7	47	boundaries	boundary	NOUN
9019s18114q	7	48	in	in	ADP
9019s18114q	7	49	the	the	DET
9019s18114q	7	50	audio	audio	ADJ
9019s18114q	7	51	signal	signal	NOUN
9019s18114q	7	52	and	and	CCONJ
9019s18114q	7	53	aligning	align	VERB
9019s18114q	7	54	the	the	DET
9019s18114q	7	55	discovered	discover	VERB
9019s18114q	7	56	segments	segment	NOUN
9019s18114q	7	57	with	with	ADP
9019s18114q	7	58	translation	translation	NOUN
9019s18114q	7	59	words	word	NOUN
9019s18114q	7	60	.	.	PUNCT
9019s18114q	8	1	speech	speech	NOUN
9019s18114q	8	2	transcription	transcription	NOUN
9019s18114q	8	3	:	:	PUNCT
9019s18114q	8	4	we	we	PRON
9019s18114q	8	5	developed	develop	VERB
9019s18114q	8	6	two	two	NUM
9019s18114q	8	7	novel	novel	ADJ
9019s18114q	8	8	neural	neural	ADJ
9019s18114q	8	9	methods	method	NOUN
9019s18114q	8	10	for	for	ADP
9019s18114q	8	11	creating	create	VERB
9019s18114q	8	12	a	a	DET
9019s18114q	8	13	phoneme	phoneme	NOUN
9019s18114q	8	14	or	or	CCONJ
9019s18114q	8	15	grapheme	grapheme	NOUN
9019s18114q	8	16	level	level	NOUN
9019s18114q	8	17	transcription	transcription	NOUN
9019s18114q	8	18	of	of	ADP
9019s18114q	8	19	the	the	DET
9019s18114q	8	20	audio	audio	NOUN
9019s18114q	8	21	,	,	PUNCT
9019s18114q	8	22	also	also	ADV
9019s18114q	8	23	utilizing	utilize	VERB
9019s18114q	8	24	any	any	DET
9019s18114q	8	25	available	available	ADJ
9019s18114q	8	26	translations	translation	NOUN
9019s18114q	8	27	.	.	PUNCT
9019s18114q	9	1	speech	speech	NOUN
9019s18114q	9	2	translation	translation	NOUN
9019s18114q	9	3	:	:	PUNCT
9019s18114q	9	4	our	our	PRON
9019s18114q	9	5	novel	novel	ADJ
9019s18114q	9	6	multitask	multitask	ADJ
9019s18114q	9	7	neural	neural	ADJ
9019s18114q	9	8	model	model	NOUN
9019s18114q	9	9	jointly	jointly	ADV
9019s18114q	9	10	produces	produce	VERB
9019s18114q	9	11	a	a	DET
9019s18114q	9	12	transcription	transcription	NOUN
9019s18114q	9	13	and	and	CCONJ
9019s18114q	9	14	a	a	DET
9019s18114q	9	15	(	(	PUNCT
9019s18114q	9	16	free	free	ADJ
9019s18114q	9	17	)	)	PUNCT
9019s18114q	9	18	translation	translation	NOUN
9019s18114q	9	19	of	of	ADP
9019s18114q	9	20	an	an	DET
9019s18114q	9	21	audio	audio	ADJ
9019s18114q	9	22	segment	segment	NOUN
9019s18114q	9	23	.	.	PUNCT
9019s18114q	10	1	morphological	morphological	ADJ
9019s18114q	10	2	analysis	analysis	NOUN
9019s18114q	10	3	:	:	PUNCT
9019s18114q	10	4	producing	produce	VERB
9019s18114q	10	5	a	a	DET
9019s18114q	10	6	layer	layer	NOUN
9019s18114q	10	7	of	of	ADP
9019s18114q	10	8	annotation	annotation	NOUN
9019s18114q	10	9	that	that	PRON
9019s18114q	10	10	provides	provide	VERB
9019s18114q	10	11	word	word	NOUN
9019s18114q	10	12	-	-	PUNCT
9019s18114q	10	13	level	level	NOUN
9019s18114q	10	14	or	or	CCONJ
9019s18114q	10	15	morpheme	morpheme	NOUN
9019s18114q	10	16	-	-	PUNCT
9019s18114q	10	17	level	level	NOUN
9019s18114q	10	18	information	information	NOUN
9019s18114q	10	19	.	.	PUNCT
9019s18114q	11	1	in	in	ADP
9019s18114q	11	2	this	this	DET
9019s18114q	11	3	work	work	NOUN
9019s18114q	11	4	we	we	PRON
9019s18114q	11	5	focus	focus	VERB
9019s18114q	11	6	on	on	ADP
9019s18114q	11	7	grammatical	grammatical	ADJ
9019s18114q	11	8	(	(	PUNCT
9019s18114q	11	9	part	part	NOUN
9019s18114q	11	10	-	-	PUNCT
9019s18114q	11	11	of	of	ADP
9019s18114q	11	12	-	-	PUNCT
9019s18114q	11	13	speech	speech	NOUN
9019s18114q	11	14	)	)	PUNCT
9019s18114q	11	15	tagging	tag	VERB
9019s18114q	11	16	on	on	ADP
9019s18114q	11	17	an	an	DET
9019s18114q	11	18	endangered	endangered	ADJ
9019s18114q	11	19	language	language	NOUN
9019s18114q	11	20	.	.	PUNCT
9019s18114q	12	1	building	build	VERB
9019s18114q	12	2	off	off	ADP
9019s18114q	12	3	limited	limited	ADJ
9019s18114q	12	4	or	or	CCONJ
9019s18114q	12	5	no	no	PRON
9019s18114q	12	6	annotations	annotation	NOUN
9019s18114q	12	7	,	,	PUNCT
9019s18114q	12	8	our	our	PRON
9019s18114q	12	9	methods	method	NOUN
9019s18114q	12	10	are	be	AUX
9019s18114q	12	11	capable	capable	ADJ
9019s18114q	12	12	of	of	ADP
9019s18114q	12	13	producing	produce	VERB
9019s18114q	12	14	helpful	helpful	ADJ
9019s18114q	12	15	suggestions	suggestion	NOUN
9019s18114q	12	16	for	for	ADP
9019s18114q	12	17	word	word	NOUN
9019s18114q	12	18	or	or	CCONJ
9019s18114q	12	19	phrase	phrase	NOUN
9019s18114q	12	20	boundaries	boundary	NOUN
9019s18114q	12	21	,	,	PUNCT
9019s18114q	12	22	as	as	ADV
9019s18114q	12	23	well	well	ADV
9019s18114q	12	24	as	as	ADP
9019s18114q	12	25	transcriptions	transcription	NOUN
9019s18114q	12	26	,	,	PUNCT
9019s18114q	12	27	translations	translation	NOUN
9019s18114q	12	28	,	,	PUNCT
9019s18114q	12	29	or	or	CCONJ
9019s18114q	12	30	grammatical	grammatical	ADJ
9019s18114q	12	31	tags	tag	NOUN
9019s18114q	12	32	.	.	PUNCT
9019s18114q	13	1	as	as	ADP
9019s18114q	13	2	a	a	DET
9019s18114q	13	3	result	result	NOUN
9019s18114q	13	4	,	,	PUNCT
9019s18114q	13	5	our	our	PRON
9019s18114q	13	6	work	work	NOUN
9019s18114q	13	7	provides	provide	VERB
9019s18114q	13	8	the	the	DET
9019s18114q	13	9	machine	machine	NOUN
9019s18114q	13	10	learning	learn	VERB
9019s18114q	13	11	methods	method	NOUN
9019s18114q	13	12	that	that	PRON
9019s18114q	13	13	could	could	AUX
9019s18114q	13	14	form	form	VERB
9019s18114q	13	15	the	the	DET
9019s18114q	13	16	backbone	backbone	NOUN
9019s18114q	13	17	of	of	ADP
9019s18114q	13	18	a	a	DET
9019s18114q	13	19	modern	modern	ADJ
9019s18114q	13	20	linguistic	linguistic	ADJ
9019s18114q	13	21	annotation	annotation	NOUN
9019s18114q	13	22	toolkit	toolkit	NOUN
9019s18114q	13	23	,	,	PUNCT
9019s18114q	13	24	one	one	NUM
9019s18114q	13	25	that	that	PRON
9019s18114q	13	26	could	could	AUX
9019s18114q	13	27	have	have	VERB
9019s18114q	13	28	the	the	DET
9019s18114q	13	29	potential	potential	NOUN
9019s18114q	13	30	to	to	PART
9019s18114q	13	31	significantly	significantly	ADV
9019s18114q	13	32	accelerate	accelerate	VERB
9019s18114q	13	33	the	the	DET
9019s18114q	13	34	language	language	NOUN
9019s18114q	13	35	documentation	documentation	NOUN
9019s18114q	13	36	process	process	NOUN
9019s18114q	13	37	.	.	PUNCT