id	sid	tid	token	lemma	pos
elm-5391	1	1	representing	represent	VERB
elm-5391	1	2	affect	affect	VERB
elm-5391	1	3	information	information	NOUN
elm-5391	1	4	in	in	ADP
elm-5391	1	5	word	word	NOUN
elm-5391	1	6	embeddings	embedding	NOUN
elm-5391	1	7	yuhan	yuhan	PROPN
elm-5391	1	8	zhang	zhang	PROPN
elm-5391	1	9	,	,	PUNCT
elm-5391	1	10	wenqi	wenqi	PROPN
elm-5391	1	11	chen	chen	PROPN
elm-5391	1	12	,	,	PUNCT
elm-5391	1	13	ruihan	ruihan	PROPN
elm-5391	1	14	zhang	zhang	PROPN
elm-5391	1	15	&	&	CCONJ
elm-5391	1	16	xiajie	xiajie	PROPN
elm-5391	1	17	zhang	zhang	PROPN
elm-5391	1	18	*	*	PROPN
elm-5391	1	19	abstract	abstract	PROPN
elm-5391	1	20	.	.	PUNCT
elm-5391	2	1	a	a	DET
elm-5391	2	2	growing	grow	VERB
elm-5391	2	3	body	body	NOUN
elm-5391	2	4	of	of	ADP
elm-5391	2	5	research	research	NOUN
elm-5391	2	6	in	in	ADP
elm-5391	2	7	natural	natural	ADJ
elm-5391	2	8	language	language	NOUN
elm-5391	2	9	processing	processing	NOUN
elm-5391	2	10	(	(	PUNCT
elm-5391	2	11	nlp	nlp	NOUN
elm-5391	2	12	)	)	PUNCT
elm-5391	2	13	and	and	CCONJ
elm-5391	2	14	natural	natural	ADJ
elm-5391	2	15	language	language	NOUN
elm-5391	2	16	understanding	understanding	NOUN
elm-5391	2	17	(	(	PUNCT
elm-5391	2	18	nlu	nlu	NOUN
elm-5391	2	19	)	)	PUNCT
elm-5391	2	20	is	be	AUX
elm-5391	2	21	investigating	investigate	VERB
elm-5391	2	22	human	human	ADJ
elm-5391	2	23	-	-	PUNCT
elm-5391	2	24	like	like	ADJ
elm-5391	2	25	knowledge	knowledge	NOUN
elm-5391	2	26	learned	learn	VERB
elm-5391	2	27	or	or	CCONJ
elm-5391	2	28	encoded	encode	VERB
elm-5391	2	29	in	in	ADP
elm-5391	2	30	the	the	DET
elm-5391	2	31	word	word	NOUN
elm-5391	2	32	embeddings	embedding	NOUN
elm-5391	2	33	from	from	ADP
elm-5391	2	34	large	large	ADJ
elm-5391	2	35	language	language	NOUN
elm-5391	2	36	models	model	NOUN
elm-5391	2	37	.	.	PUNCT
elm-5391	3	1	this	this	PRON
elm-5391	3	2	is	be	AUX
elm-5391	3	3	a	a	DET
elm-5391	3	4	step	step	NOUN
elm-5391	3	5	towards	towards	ADP
elm-5391	3	6	understanding	understand	VERB
elm-5391	3	7	what	what	PRON
elm-5391	3	8	knowledge	knowledge	NOUN
elm-5391	3	9	language	language	NOUN
elm-5391	3	10	models	model	NOUN
elm-5391	3	11	capture	capture	VERB
elm-5391	3	12	that	that	PRON
elm-5391	3	13	resembles	resemble	VERB
elm-5391	3	14	human	human	ADJ
elm-5391	3	15	understanding	understanding	NOUN
elm-5391	3	16	of	of	ADP
elm-5391	3	17	language	language	NOUN
elm-5391	3	18	and	and	CCONJ
elm-5391	3	19	communication	communication	NOUN
elm-5391	3	20	.	.	PUNCT
elm-5391	4	1	here	here	ADV
elm-5391	4	2	,	,	PUNCT
elm-5391	4	3	we	we	PRON
elm-5391	4	4	investigated	investigate	VERB
elm-5391	4	5	whether	whether	SCONJ
elm-5391	4	6	and	and	CCONJ
elm-5391	4	7	how	how	SCONJ
elm-5391	4	8	the	the	PRON
elm-5391	4	9	affect	affect	VERB
elm-5391	4	10	meaning	meaning	NOUN
elm-5391	4	11	of	of	ADP
elm-5391	4	12	a	a	DET
elm-5391	4	13	word	word	NOUN
elm-5391	4	14	(	(	PUNCT
elm-5391	4	15	i.e.	i.e.	X
elm-5391	4	16	,	,	PUNCT
elm-5391	4	17	valence	valence	NOUN
elm-5391	4	18	,	,	PUNCT
elm-5391	4	19	arousal	arousal	NOUN
elm-5391	4	20	,	,	PUNCT
elm-5391	4	21	dominance	dominance	NOUN
elm-5391	4	22	)	)	PUNCT
elm-5391	4	23	is	be	AUX
elm-5391	4	24	encoded	encode	VERB
elm-5391	4	25	in	in	ADP
elm-5391	4	26	word	word	NOUN
elm-5391	4	27	embeddings	embedding	NOUN
elm-5391	4	28	pre	pre	VERB
elm-5391	4	29	-	-	VERB
elm-5391	4	30	trained	train	VERB
elm-5391	4	31	in	in	ADP
elm-5391	4	32	large	large	ADJ
elm-5391	4	33	neural	neural	ADJ
elm-5391	4	34	networks	network	NOUN
elm-5391	4	35	.	.	PUNCT
elm-5391	5	1	we	we	PRON
elm-5391	5	2	used	use	VERB
elm-5391	5	3	the	the	DET
elm-5391	5	4	human	human	NOUN
elm-5391	5	5	-	-	PUNCT
elm-5391	5	6	labeled	label	VERB
elm-5391	5	7	dataset	dataset	NOUN
elm-5391	5	8	(	(	PUNCT
elm-5391	5	9	mohammad	mohammad	PROPN
elm-5391	5	10	2018	2018	NUM
elm-5391	5	11	)	)	PUNCT
elm-5391	5	12	as	as	ADP
elm-5391	5	13	the	the	DET
elm-5391	5	14	ground	ground	NOUN
elm-5391	5	15	truth	truth	NOUN
elm-5391	5	16	and	and	CCONJ
elm-5391	5	17	performed	perform	VERB
elm-5391	5	18	various	various	ADJ
elm-5391	5	19	correlational	correlational	ADJ
elm-5391	5	20	and	and	CCONJ
elm-5391	5	21	classification	classification	NOUN
elm-5391	5	22	tests	test	NOUN
elm-5391	5	23	on	on	ADP
elm-5391	5	24	four	four	NUM
elm-5391	5	25	types	type	NOUN
elm-5391	5	26	of	of	ADP
elm-5391	5	27	word	word	NOUN
elm-5391	5	28	embeddings	embedding	NOUN
elm-5391	5	29	.	.	PUNCT
elm-5391	6	1	the	the	DET
elm-5391	6	2	embeddings	embedding	NOUN
elm-5391	6	3	varied	varied	ADJ
elm-5391	6	4	in	in	ADP
elm-5391	6	5	being	be	AUX
elm-5391	6	6	static	static	ADJ
elm-5391	6	7	or	or	CCONJ
elm-5391	6	8	contextualized	contextualize	VERB
elm-5391	6	9	,	,	PUNCT
elm-5391	6	10	and	and	CCONJ
elm-5391	6	11	how	how	SCONJ
elm-5391	6	12	much	much	ADJ
elm-5391	6	13	affect	affect	VERB
elm-5391	6	14	specific	specific	ADJ
elm-5391	6	15	information	information	NOUN
elm-5391	6	16	was	be	AUX
elm-5391	6	17	prioritized	prioritize	VERB
elm-5391	6	18	during	during	ADP
elm-5391	6	19	the	the	DET
elm-5391	6	20	pre	pre	NOUN
elm-5391	6	21	-	-	NOUN
elm-5391	6	22	training	training	ADJ
elm-5391	6	23	and	and	CCONJ
elm-5391	6	24	fine	fine	ADV
elm-5391	6	25	-	-	PUNCT
elm-5391	6	26	tuning	tuning	NOUN
elm-5391	6	27	phase	phase	NOUN
elm-5391	6	28	.	.	PUNCT
elm-5391	7	1	our	our	PRON
elm-5391	7	2	analyses	analysis	NOUN
elm-5391	7	3	show	show	VERB
elm-5391	7	4	that	that	SCONJ
elm-5391	7	5	word	word	NOUN
elm-5391	7	6	embedding	embed	VERB
elm-5391	7	7	from	from	ADP
elm-5391	7	8	the	the	DET
elm-5391	7	9	vanilla	vanilla	NOUN
elm-5391	7	10	bert	bert	PROPN
elm-5391	7	11	model	model	NOUN
elm-5391	7	12	(	(	PUNCT
elm-5391	7	13	devlin	devlin	PROPN
elm-5391	7	14	et	et	PROPN
elm-5391	7	15	al	al	PROPN
elm-5391	7	16	.	.	PROPN
elm-5391	7	17	2019	2019	NUM
elm-5391	7	18	)	)	PUNCT
elm-5391	7	19	did	do	AUX
elm-5391	7	20	not	not	PART
elm-5391	7	21	saliently	saliently	ADV
elm-5391	7	22	encode	encode	VERB
elm-5391	7	23	the	the	DET
elm-5391	7	24	affect	affect	ADJ
elm-5391	7	25	information	information	NOUN
elm-5391	7	26	of	of	ADP
elm-5391	7	27	english	english	ADJ
elm-5391	7	28	words	word	NOUN
elm-5391	7	29	.	.	PUNCT
elm-5391	8	1	only	only	ADV
elm-5391	8	2	when	when	SCONJ
elm-5391	8	3	the	the	DET
elm-5391	8	4	bert	bert	PROPN
elm-5391	8	5	model	model	NOUN
elm-5391	8	6	was	be	AUX
elm-5391	8	7	fine	fine	ADV
elm-5391	8	8	-	-	PUNCT
elm-5391	8	9	tuned	tune	VERB
elm-5391	8	10	on	on	ADP
elm-5391	8	11	emotion	emotion	NOUN
elm-5391	8	12	related	relate	VERB
elm-5391	8	13	tasks	task	NOUN
elm-5391	8	14	or	or	CCONJ
elm-5391	8	15	contained	contain	VERB
elm-5391	8	16	extra	extra	ADJ
elm-5391	8	17	contextualized	contextualized	ADJ
elm-5391	8	18	information	information	NOUN
elm-5391	8	19	from	from	ADP
elm-5391	8	20	emotion	emotion	NOUN
elm-5391	8	21	-	-	PUNCT
elm-5391	8	22	rich	rich	ADJ
elm-5391	8	23	contexts	contexts	NOUN
elm-5391	8	24	could	could	AUX
elm-5391	8	25	the	the	DET
elm-5391	8	26	corresponding	corresponding	ADJ
elm-5391	8	27	embedding	embed	VERB
elm-5391	8	28	encode	encode	ADJ
elm-5391	8	29	more	more	ADV
elm-5391	8	30	relevant	relevant	ADJ
elm-5391	8	31	affect	affect	VERB
elm-5391	8	32	information	information	NOUN
elm-5391	8	33	.	.	PUNCT
elm-5391	9	1	keywords	keyword	NOUN
elm-5391	9	2	.	.	PUNCT
elm-5391	10	1	language	language	NOUN
elm-5391	10	2	models	model	NOUN
elm-5391	10	3	;	;	PUNCT
elm-5391	10	4	word	word	NOUN
elm-5391	10	5	embeddings	embedding	NOUN
elm-5391	10	6	;	;	PUNCT
elm-5391	10	7	affect	affect	VERB
elm-5391	10	8	meaning	meaning	NOUN
elm-5391	10	9	;	;	PUNCT
elm-5391	10	10	lexical	lexical	ADJ
elm-5391	10	11	semantics	semantic	NOUN
elm-5391	10	12	1	1	NUM
elm-5391	10	13	.	.	PUNCT
elm-5391	10	14	introduction	introduction	NOUN
elm-5391	10	15	.	.	PUNCT
elm-5391	11	1	with	with	ADP
elm-5391	11	2	the	the	DET
elm-5391	11	3	success	success	NOUN
elm-5391	11	4	of	of	ADP
elm-5391	11	5	large	large	ADJ
elm-5391	11	6	neural	neural	ADJ
elm-5391	11	7	network	network	NOUN
elm-5391	11	8	models	model	NOUN
elm-5391	11	9	in	in	ADP
elm-5391	11	10	completing	complete	VERB
elm-5391	11	11	complicated	complicated	ADJ
elm-5391	11	12	language	language	NOUN
elm-5391	11	13	tasks	task	NOUN
elm-5391	11	14	,	,	PUNCT
elm-5391	11	15	evaluating	evaluate	VERB
elm-5391	11	16	the	the	DET
elm-5391	11	17	models	model	NOUN
elm-5391	11	18	’	’	PART
elm-5391	11	19	interpretability	interpretability	NOUN
elm-5391	11	20	and	and	CCONJ
elm-5391	11	21	intrinsic	intrinsic	ADJ
elm-5391	11	22	capabilities	capability	NOUN
elm-5391	11	23	has	have	AUX
elm-5391	11	24	become	become	VERB
elm-5391	11	25	a	a	DET
elm-5391	11	26	heated	heated	ADJ
elm-5391	11	27	research	research	NOUN
elm-5391	11	28	trend	trend	NOUN
elm-5391	11	29	(	(	PUNCT
elm-5391	11	30	e.g.	e.g.	ADV
elm-5391	11	31	,	,	PUNCT
elm-5391	11	32	manning	man	VERB
elm-5391	11	33	et	et	PROPN
elm-5391	11	34	al	al	PROPN
elm-5391	11	35	.	.	PROPN
elm-5391	11	36	2020	2020	NUM
elm-5391	11	37	,	,	PUNCT
elm-5391	11	38	mikolov	mikolov	PROPN
elm-5391	11	39	et	et	PROPN
elm-5391	11	40	al	al	PROPN
elm-5391	11	41	.	.	PROPN
elm-5391	11	42	2013	2013	NUM
elm-5391	11	43	)	)	PUNCT
elm-5391	11	44	.	.	PUNCT
elm-5391	12	1	the	the	DET
elm-5391	12	2	evaluation	evaluation	NOUN
elm-5391	12	3	work	work	NOUN
elm-5391	12	4	could	could	AUX
elm-5391	12	5	be	be	AUX
elm-5391	12	6	roughly	roughly	ADV
elm-5391	12	7	classified	classify	VERB
elm-5391	12	8	into	into	ADP
elm-5391	12	9	two	two	NUM
elm-5391	12	10	types	type	NOUN
elm-5391	12	11	:	:	PUNCT
elm-5391	12	12	one	one	NUM
elm-5391	12	13	that	that	PRON
elm-5391	12	14	relies	rely	VERB
elm-5391	12	15	on	on	ADP
elm-5391	12	16	the	the	DET
elm-5391	12	17	output	output	NOUN
elm-5391	12	18	of	of	ADP
elm-5391	12	19	the	the	DET
elm-5391	12	20	language	language	NOUN
elm-5391	12	21	models	model	NOUN
elm-5391	12	22	(	(	PUNCT
elm-5391	12	23	lms	lm	NOUN
elm-5391	12	24	)	)	PUNCT
elm-5391	12	25	to	to	PART
elm-5391	12	26	infer	infer	VERB
elm-5391	12	27	the	the	DET
elm-5391	12	28	model	model	NOUN
elm-5391	12	29	’s	’s	PART
elm-5391	12	30	linguistic	linguistic	ADJ
elm-5391	12	31	ability	ability	NOUN
elm-5391	12	32	and	and	CCONJ
elm-5391	12	33	the	the	DET
elm-5391	12	34	other	other	ADJ
elm-5391	12	35	that	that	PRON
elm-5391	12	36	looks	look	VERB
elm-5391	12	37	into	into	ADP
elm-5391	12	38	the	the	DET
elm-5391	12	39	components	component	NOUN
elm-5391	12	40	of	of	ADP
elm-5391	12	41	lms	lm	NOUN
elm-5391	12	42	(	(	PUNCT
elm-5391	12	43	e.g.	e.g.	ADV
elm-5391	12	44	,	,	PUNCT
elm-5391	12	45	word	word	NOUN
elm-5391	12	46	embeddings	embedding	NOUN
elm-5391	12	47	)	)	PUNCT
elm-5391	12	48	for	for	ADP
elm-5391	12	49	such	such	ADJ
elm-5391	12	50	inspiration	inspiration	NOUN
elm-5391	12	51	.	.	PUNCT
elm-5391	13	1	while	while	SCONJ
elm-5391	13	2	previous	previous	ADJ
elm-5391	13	3	evaluation	evaluation	NOUN
elm-5391	13	4	tasks	task	NOUN
elm-5391	13	5	have	have	AUX
elm-5391	13	6	focused	focus	VERB
elm-5391	13	7	on	on	ADP
elm-5391	13	8	testing	test	VERB
elm-5391	13	9	lms	lm	NOUN
elm-5391	13	10	’	'	PUNCT
elm-5391	13	11	explicit	explicit	ADJ
elm-5391	13	12	linguistic	linguistic	ADJ
elm-5391	13	13	knowledge	knowledge	NOUN
elm-5391	13	14	(	(	PUNCT
elm-5391	13	15	e.g.	e.g.	ADV
elm-5391	13	16	,	,	PUNCT
elm-5391	13	17	syntactic	syntactic	ADJ
elm-5391	13	18	knowledge	knowledge	NOUN
elm-5391	13	19	such	such	ADJ
elm-5391	13	20	as	as	ADP
elm-5391	13	21	islands	island	NOUN
elm-5391	13	22	,	,	PUNCT
elm-5391	13	23	semantic	semantic	ADJ
elm-5391	13	24	knowledge	knowledge	NOUN
elm-5391	13	25	such	such	ADJ
elm-5391	13	26	as	as	ADP
elm-5391	13	27	compositionality	compositionality	NOUN
elm-5391	13	28	,	,	PUNCT
elm-5391	13	29	word	word	NOUN
elm-5391	13	30	-	-	PUNCT
elm-5391	13	31	level	level	NOUN
elm-5391	13	32	knowledge	knowledge	NOUN
elm-5391	13	33	such	such	ADJ
elm-5391	13	34	as	as	ADP
elm-5391	13	35	polysemy	polysemy	NOUN
elm-5391	13	36	)	)	PUNCT
elm-5391	13	37	,	,	PUNCT
elm-5391	13	38	we	we	PRON
elm-5391	13	39	pick	pick	VERB
elm-5391	13	40	a	a	DET
elm-5391	13	41	piece	piece	NOUN
elm-5391	13	42	of	of	ADP
elm-5391	13	43	knowledge	knowledge	NOUN
elm-5391	13	44	that	that	PRON
elm-5391	13	45	is	be	AUX
elm-5391	13	46	less	less	ADV
elm-5391	13	47	studied	studied	ADJ
elm-5391	13	48	but	but	CCONJ
elm-5391	13	49	essential	essential	ADJ
elm-5391	13	50	to	to	ADP
elm-5391	13	51	intelligence	intelligence	NOUN
elm-5391	13	52	.	.	PUNCT
elm-5391	14	1	specifically	specifically	ADV
elm-5391	14	2	,	,	PUNCT
elm-5391	14	3	we	we	PRON
elm-5391	14	4	studied	study	VERB
elm-5391	14	5	whether	whether	SCONJ
elm-5391	15	1	and	and	CCONJ
elm-5391	15	2	how	how	SCONJ
elm-5391	15	3	word	word	NOUN
elm-5391	15	4	embeddings	embedding	NOUN
elm-5391	15	5	learned	learn	VERB
elm-5391	15	6	via	via	ADP
elm-5391	15	7	supervised	supervised	ADJ
elm-5391	15	8	methods	method	NOUN
elm-5391	15	9	in	in	ADP
elm-5391	15	10	large	large	ADJ
elm-5391	15	11	neural	neural	ADJ
elm-5391	15	12	networks	network	NOUN
elm-5391	15	13	encode	encode	VERB
elm-5391	15	14	the	the	DET
elm-5391	15	15	affect	affect	ADJ
elm-5391	15	16	information	information	NOUN
elm-5391	15	17	of	of	ADP
elm-5391	15	18	a	a	DET
elm-5391	15	19	word	word	NOUN
elm-5391	15	20	(	(	PUNCT
elm-5391	15	21	e.g.	e.g.	ADV
elm-5391	15	22	,	,	PUNCT
elm-5391	15	23	valence	valence	NOUN
elm-5391	15	24	,	,	PUNCT
elm-5391	15	25	arousal	arousal	NOUN
elm-5391	15	26	,	,	PUNCT
elm-5391	15	27	dominance	dominance	NOUN
elm-5391	15	28	)	)	PUNCT
elm-5391	15	29	.	.	PUNCT
elm-5391	16	1	our	our	PRON
elm-5391	16	2	work	work	NOUN
elm-5391	16	3	shows	show	VERB
elm-5391	16	4	that	that	SCONJ
elm-5391	16	5	even	even	ADV
elm-5391	16	6	though	though	SCONJ
elm-5391	16	7	contextualized	contextualize	VERB
elm-5391	16	8	word	word	NOUN
elm-5391	16	9	embeddings	embedding	NOUN
elm-5391	16	10	were	be	AUX
elm-5391	16	11	in	in	ADP
elm-5391	16	12	general	general	ADJ
elm-5391	16	13	better	well	ADV
elm-5391	16	14	at	at	ADP
elm-5391	16	15	capturing	capture	VERB
elm-5391	16	16	intricate	intricate	ADJ
elm-5391	16	17	affect	affect	NOUN
elm-5391	16	18	meanings	meaning	NOUN
elm-5391	16	19	compared	compare	VERB
elm-5391	16	20	to	to	ADP
elm-5391	16	21	static	static	ADJ
elm-5391	16	22	word	word	NOUN
elm-5391	16	23	embeddings	embedding	NOUN
elm-5391	16	24	,	,	PUNCT
elm-5391	16	25	especially	especially	ADV
elm-5391	16	26	after	after	ADP
elm-5391	16	27	being	be	AUX
elm-5391	16	28	fine	fine	ADV
elm-5391	16	29	-	-	PUNCT
elm-5391	16	30	tuned	tune	VERB
elm-5391	16	31	on	on	ADP
elm-5391	16	32	emotion	emotion	NOUN
elm-5391	16	33	related	relate	VERB
elm-5391	16	34	tasks	task	NOUN
elm-5391	16	35	,	,	PUNCT
elm-5391	16	36	word	word	NOUN
elm-5391	16	37	embeddings	embedding	NOUN
elm-5391	16	38	from	from	ADP
elm-5391	16	39	vanilla	vanilla	NOUN
elm-5391	16	40	bert	bert	PROPN
elm-5391	16	41	did	do	AUX
elm-5391	16	42	not	not	PART
elm-5391	16	43	attain	attain	VERB
elm-5391	16	44	salient	salient	NOUN
elm-5391	16	45	affect	affect	NOUN
elm-5391	16	46	knowledge	knowledge	NOUN
elm-5391	16	47	.	.	PUNCT
elm-5391	17	1	in	in	ADP
elm-5391	17	2	section	section	NOUN
elm-5391	17	3	2	2	NUM
elm-5391	17	4	,	,	PUNCT
elm-5391	17	5	we	we	PRON
elm-5391	17	6	detailed	detail	VERB
elm-5391	17	7	relevant	relevant	ADJ
elm-5391	17	8	work	work	NOUN
elm-5391	17	9	that	that	PRON
elm-5391	17	10	led	lead	VERB
elm-5391	17	11	us	we	PRON
elm-5391	17	12	to	to	ADP
elm-5391	17	13	our	our	PRON
elm-5391	17	14	investigation	investigation	NOUN
elm-5391	17	15	.	.	PUNCT
elm-5391	18	1	in	in	ADP
elm-5391	18	2	section	section	NOUN
elm-5391	18	3	3	3	NUM
elm-5391	18	4	,	,	PUNCT
elm-5391	18	5	we	we	PRON
elm-5391	18	6	laid	lay	VERB
elm-5391	18	7	out	out	ADP
elm-5391	18	8	the	the	DET
elm-5391	18	9	unsupervised	unsupervised	ADJ
elm-5391	18	10	and	and	CCONJ
elm-5391	18	11	supervised	supervised	ADJ
elm-5391	18	12	methodologies	methodology	NOUN
elm-5391	18	13	we	we	PRON
elm-5391	18	14	took	take	VERB
elm-5391	18	15	and	and	CCONJ
elm-5391	18	16	in	in	ADP
elm-5391	18	17	section	section	NOUN
elm-5391	18	18	4	4	NUM
elm-5391	18	19	,	,	PUNCT
elm-5391	18	20	we	we	PRON
elm-5391	18	21	spelled	spell	VERB
elm-5391	18	22	out	out	ADP
elm-5391	18	23	the	the	DET
elm-5391	18	24	findings	finding	NOUN
elm-5391	18	25	.	.	PUNCT
elm-5391	19	1	in	in	ADP
elm-5391	19	2	section	section	NOUN
elm-5391	19	3	5	5	NUM
elm-5391	19	4	,	,	PUNCT
elm-5391	19	5	we	we	PRON
elm-5391	19	6	discussed	discuss	VERB
elm-5391	19	7	implications	implication	NOUN
elm-5391	19	8	of	of	ADP
elm-5391	19	9	our	our	PRON
elm-5391	19	10	research	research	NOUN
elm-5391	19	11	and	and	CCONJ
elm-5391	19	12	possible	possible	ADJ
elm-5391	19	13	future	future	ADJ
elm-5391	19	14	directions	direction	NOUN
elm-5391	19	15	.	.	PUNCT
elm-5391	20	1	*	*	PUNCT
elm-5391	20	2	we	we	PRON
elm-5391	20	3	would	would	AUX
elm-5391	20	4	like	like	VERB
elm-5391	20	5	to	to	PART
elm-5391	20	6	thank	thank	VERB
elm-5391	20	7	mycal	mycal	ADJ
elm-5391	20	8	tucker	tucker	PROPN
elm-5391	20	9	,	,	PUNCT
elm-5391	20	10	roger	roger	PROPN
elm-5391	20	11	levy	levy	PROPN
elm-5391	20	12	,	,	PUNCT
elm-5391	20	13	and	and	CCONJ
elm-5391	20	14	the	the	DET
elm-5391	20	15	audience	audience	NOUN
elm-5391	20	16	at	at	ADP
elm-5391	20	17	elm	elm	PROPN
elm-5391	20	18	2022	2022	NUM
elm-5391	20	19	for	for	ADP
elm-5391	20	20	their	their	PRON
elm-5391	20	21	generous	generous	ADJ
elm-5391	20	22	feedback	feedback	NOUN
elm-5391	20	23	.	.	PUNCT
elm-5391	21	1	all	all	DET
elm-5391	21	2	mistakes	mistake	NOUN
elm-5391	21	3	are	be	AUX
elm-5391	21	4	ours	ours	PRON
elm-5391	21	5	.	.	PUNCT
elm-5391	22	1	authors	author	NOUN
elm-5391	22	2	:	:	PUNCT
elm-5391	22	3	yuhan	yuhan	PROPN
elm-5391	22	4	zhang	zhang	PROPN
elm-5391	22	5	,	,	PUNCT
elm-5391	22	6	harvard	harvard	PROPN
elm-5391	22	7	university	university	PROPN
elm-5391	22	8	(	(	PUNCT
elm-5391	22	9	yuz551@g.harvard.edu	yuz551@g.harvard.edu	NOUN
elm-5391	22	10	)	)	PUNCT
elm-5391	22	11	;	;	PUNCT
elm-5391	22	12	wenqi	wenqi	PROPN
elm-5391	22	13	chen	chen	PROPN
elm-5391	22	14	,	,	PUNCT
elm-5391	22	15	harvard	harvard	PROPN
elm-5391	22	16	university	university	PROPN
elm-5391	22	17	(	(	PUNCT
elm-5391	22	18	wenqichen@g.harvard.edu	wenqichen@g.harvard.edu	PROPN
elm-5391	22	19	)	)	PUNCT
elm-5391	22	20	;	;	PUNCT
elm-5391	22	21	ruihan	ruihan	PROPN
elm-5391	22	22	zhang	zhang	PROPN
elm-5391	22	23	,	,	PUNCT
elm-5391	22	24	massachusetts	massachusetts	PROPN
elm-5391	22	25	institute	institute	PROPN
elm-5391	22	26	of	of	ADP
elm-5391	22	27	technology	technology	PROPN
elm-5391	22	28	(	(	PUNCT
elm-5391	22	29	ruihanz@mit.edu	ruihanz@mit.edu	PROPN
elm-5391	22	30	)	)	PUNCT
elm-5391	22	31	;	;	PUNCT
elm-5391	22	32	xiajie	xiajie	PROPN
elm-5391	22	33	zhang	zhang	PROPN
elm-5391	22	34	,	,	PUNCT
elm-5391	22	35	massachusetts	massachusetts	PROPN
elm-5391	22	36	institute	institute	PROPN
elm-5391	22	37	of	of	ADP
elm-5391	22	38	technology	technology	PROPN
elm-5391	22	39	(	(	PUNCT
elm-5391	22	40	xiajie@mit.edu	xiajie@mit.edu	PROPN
elm-5391	22	41	)	)	PUNCT
elm-5391	22	42	.	.	PUNCT
elm-5391	23	1	proceedings	proceeding	NOUN
elm-5391	23	2	of	of	ADP
elm-5391	23	3	elm	elm	NOUN
elm-5391	23	4	2	2	NUM
elm-5391	23	5	:	:	PUNCT
elm-5391	23	6	310	310	NUM
elm-5391	23	7	-	-	SYM
elm-5391	23	8	321	321	NUM
elm-5391	23	9	,	,	PUNCT
elm-5391	23	10	2023	2023	NUM
elm-5391	23	11	c	c	X
elm-5391	23	12	©	©	PROPN
elm-5391	23	13	2023	2023	NUM
elm-5391	23	14	yuhan	yuhan	PROPN
elm-5391	23	15	zhang	zhang	PROPN
elm-5391	23	16	,	,	PUNCT
elm-5391	23	17	wenqi	wenqi	PROPN
elm-5391	23	18	chen	chen	PROPN
elm-5391	23	19	,	,	PUNCT
elm-5391	23	20	ruihan	ruihan	PROPN
elm-5391	23	21	zhang	zhang	PROPN
elm-5391	23	22	and	and	CCONJ
elm-5391	23	23	xiajie	xiajie	PROPN
elm-5391	23	24	zhang	zhang	PROPN
elm-5391	23	25	published	publish	VERB
elm-5391	23	26	by	by	ADP
elm-5391	23	27	the	the	DET
elm-5391	23	28	lsa	lsa	NOUN
elm-5391	23	29	with	with	ADP
elm-5391	23	30	permission	permission	NOUN
elm-5391	23	31	of	of	ADP
elm-5391	23	32	the	the	DET
elm-5391	23	33	author(s	author(s	NOUN
elm-5391	23	34	)	)	PUNCT
elm-5391	23	35	under	under	ADP
elm-5391	23	36	a	a	DET
elm-5391	23	37	cc	cc	NOUN
elm-5391	23	38	by	by	ADP
elm-5391	23	39	license	license	NOUN
elm-5391	23	40	.	.	PUNCT
elm-5391	24	1	310	310	NUM
elm-5391	24	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	24	3	https://www.elm-conference.net/	https://www.elm-conference.net/	ADP
elm-5391	24	4	2	2	NUM
elm-5391	24	5	.	.	PUNCT
elm-5391	24	6	related	relate	VERB
elm-5391	24	7	work	work	NOUN
elm-5391	24	8	.	.	PUNCT
elm-5391	25	1	three	three	NUM
elm-5391	25	2	major	major	ADJ
elm-5391	25	3	aspects	aspect	NOUN
elm-5391	25	4	in	in	ADP
elm-5391	25	5	the	the	DET
elm-5391	25	6	current	current	ADJ
elm-5391	25	7	nlp	nlp	NOUN
elm-5391	25	8	world	world	NOUN
elm-5391	25	9	are	be	AUX
elm-5391	25	10	related	relate	VERB
elm-5391	25	11	to	to	ADP
elm-5391	25	12	our	our	PRON
elm-5391	25	13	work	work	NOUN
elm-5391	25	14	:	:	PUNCT
elm-5391	25	15	the	the	DET
elm-5391	25	16	linguistic	linguistic	ADJ
elm-5391	25	17	knowledge	knowledge	NOUN
elm-5391	25	18	revealed	reveal	VERB
elm-5391	25	19	to	to	PART
elm-5391	25	20	be	be	AUX
elm-5391	25	21	grasped	grasp	VERB
elm-5391	25	22	by	by	ADP
elm-5391	25	23	language	language	NOUN
elm-5391	25	24	models	model	NOUN
elm-5391	25	25	,	,	PUNCT
elm-5391	25	26	the	the	DET
elm-5391	25	27	relationship	relationship	NOUN
elm-5391	25	28	between	between	ADP
elm-5391	25	29	what	what	DET
elm-5391	25	30	word	word	NOUN
elm-5391	25	31	embeddings	embedding	NOUN
elm-5391	25	32	have	have	AUX
elm-5391	25	33	achieved	achieve	VERB
elm-5391	25	34	to	to	PART
elm-5391	25	35	represent	represent	VERB
elm-5391	25	36	and	and	CCONJ
elm-5391	25	37	the	the	DET
elm-5391	25	38	actual	actual	ADJ
elm-5391	25	39	lexical	lexical	ADJ
elm-5391	25	40	semantics	semantic	NOUN
elm-5391	25	41	,	,	PUNCT
elm-5391	25	42	and	and	CCONJ
elm-5391	25	43	how	how	SCONJ
elm-5391	25	44	affect	affect	VERB
elm-5391	25	45	information	information	NOUN
elm-5391	25	46	is	be	AUX
elm-5391	25	47	studied	study	VERB
elm-5391	25	48	in	in	ADP
elm-5391	25	49	natural	natural	ADJ
elm-5391	25	50	language	language	NOUN
elm-5391	25	51	processing	processing	NOUN
elm-5391	25	52	.	.	PUNCT
elm-5391	26	1	2.1	2.1	NUM
elm-5391	26	2	.	.	PUNCT
elm-5391	27	1	what	what	PRON
elm-5391	27	2	language	language	NOUN
elm-5391	27	3	models	model	NOUN
elm-5391	27	4	can	can	AUX
elm-5391	27	5	learn	learn	VERB
elm-5391	27	6	.	.	PUNCT
elm-5391	28	1	a	a	DET
elm-5391	28	2	growing	grow	VERB
elm-5391	28	3	body	body	NOUN
elm-5391	28	4	of	of	ADP
elm-5391	28	5	research	research	NOUN
elm-5391	28	6	has	have	AUX
elm-5391	28	7	investigated	investigate	VERB
elm-5391	28	8	what	what	PRON
elm-5391	28	9	linguistic	linguistic	ADJ
elm-5391	28	10	knowledge	knowledge	NOUN
elm-5391	28	11	large	large	ADJ
elm-5391	28	12	artificial	artificial	ADJ
elm-5391	28	13	neural	neural	ADJ
elm-5391	28	14	networks	network	NOUN
elm-5391	28	15	can	can	AUX
elm-5391	28	16	learn	learn	VERB
elm-5391	28	17	while	while	SCONJ
elm-5391	28	18	they	they	PRON
elm-5391	28	19	are	be	AUX
elm-5391	28	20	trained	train	VERB
elm-5391	28	21	in	in	ADP
elm-5391	28	22	a	a	DET
elm-5391	28	23	supervised	supervised	ADJ
elm-5391	28	24	way	way	NOUN
elm-5391	28	25	to	to	PART
elm-5391	28	26	predict	predict	VERB
elm-5391	28	27	the	the	DET
elm-5391	28	28	next	next	ADJ
elm-5391	28	29	word	word	NOUN
elm-5391	28	30	.	.	PUNCT
elm-5391	29	1	in	in	ADP
elm-5391	29	2	the	the	DET
elm-5391	29	3	syntactic	syntactic	ADJ
elm-5391	29	4	category	category	NOUN
elm-5391	29	5	,	,	PUNCT
elm-5391	29	6	scholars	scholar	NOUN
elm-5391	29	7	have	have	AUX
elm-5391	29	8	shown	show	VERB
elm-5391	29	9	that	that	SCONJ
elm-5391	29	10	large	large	ADJ
elm-5391	29	11	lms	lm	NOUN
elm-5391	29	12	have	have	AUX
elm-5391	29	13	varying	vary	VERB
elm-5391	29	14	grammatical	grammatical	ADJ
elm-5391	29	15	knowledge	knowledge	NOUN
elm-5391	29	16	ranging	range	VERB
elm-5391	29	17	from	from	ADP
elm-5391	29	18	the	the	DET
elm-5391	29	19	local	local	ADJ
elm-5391	29	20	subject	subject	ADJ
elm-5391	29	21	-	-	PUNCT
elm-5391	29	22	verb	verb	NOUN
elm-5391	29	23	agreement	agreement	NOUN
elm-5391	29	24	to	to	ADP
elm-5391	29	25	long	long	ADJ
elm-5391	29	26	-	-	PUNCT
elm-5391	29	27	distance	distance	NOUN
elm-5391	29	28	filler	filler	NOUN
elm-5391	29	29	-	-	PUNCT
elm-5391	29	30	gap	gap	NOUN
elm-5391	29	31	dependencies	dependency	NOUN
elm-5391	29	32	(	(	PUNCT
elm-5391	29	33	e.g.	e.g.	ADV
elm-5391	29	34	,	,	PUNCT
elm-5391	29	35	hu	hu	PROPN
elm-5391	29	36	et	et	PROPN
elm-5391	29	37	al	al	PROPN
elm-5391	29	38	.	.	PROPN
elm-5391	29	39	2020	2020	NUM
elm-5391	29	40	,	,	PUNCT
elm-5391	29	41	linzen	linzen	VERB
elm-5391	29	42	et	et	PROPN
elm-5391	29	43	al	al	PROPN
elm-5391	29	44	.	.	PROPN
elm-5391	29	45	2016	2016	NUM
elm-5391	29	46	,	,	PUNCT
elm-5391	29	47	warstadt	warstadt	NOUN
elm-5391	29	48	et	et	PROPN
elm-5391	29	49	al	al	PROPN
elm-5391	29	50	.	.	PROPN
elm-5391	29	51	2020	2020	NUM
elm-5391	29	52	,	,	PUNCT
elm-5391	29	53	wilcox	wilcox	PROPN
elm-5391	29	54	et	et	PROPN
elm-5391	29	55	al	al	PROPN
elm-5391	29	56	.	.	PROPN
elm-5391	29	57	2018	2018	NUM
elm-5391	29	58	)	)	PUNCT
elm-5391	29	59	.	.	PUNCT
elm-5391	30	1	aside	aside	ADV
elm-5391	30	2	from	from	ADP
elm-5391	30	3	looking	look	VERB
elm-5391	30	4	at	at	ADP
elm-5391	30	5	predicted	predict	VERB
elm-5391	30	6	words	word	NOUN
elm-5391	30	7	to	to	PART
elm-5391	30	8	infer	infer	VERB
elm-5391	30	9	lms	lm	NOUN
elm-5391	30	10	’	'	PUNCT
elm-5391	30	11	linguistic	linguistic	ADJ
elm-5391	30	12	knowledge	knowledge	NOUN
elm-5391	30	13	,	,	PUNCT
elm-5391	30	14	manning	man	VERB
elm-5391	30	15	et	et	PROPN
elm-5391	30	16	al	al	PROPN
elm-5391	30	17	.	.	PROPN
elm-5391	31	1	(	(	PUNCT
elm-5391	31	2	2020	2020	NUM
elm-5391	31	3	)	)	PUNCT
elm-5391	31	4	show	show	VERB
elm-5391	31	5	that	that	SCONJ
elm-5391	31	6	a	a	DET
elm-5391	31	7	linear	linear	ADJ
elm-5391	31	8	transformation	transformation	NOUN
elm-5391	31	9	of	of	ADP
elm-5391	31	10	word	word	NOUN
elm-5391	31	11	embeddings	embedding	NOUN
elm-5391	31	12	from	from	ADP
elm-5391	31	13	bert	bert	PROPN
elm-5391	31	14	(	(	PUNCT
elm-5391	31	15	devlin	devlin	PROPN
elm-5391	31	16	et	et	PROPN
elm-5391	31	17	al	al	PROPN
elm-5391	31	18	.	.	PROPN
elm-5391	31	19	2019	2019	NUM
elm-5391	31	20	)	)	PUNCT
elm-5391	31	21	captures	capture	VERB
elm-5391	31	22	linguistic	linguistic	ADJ
elm-5391	31	23	hierarchical	hierarchical	ADJ
elm-5391	31	24	structures	structure	NOUN
elm-5391	31	25	.	.	PUNCT
elm-5391	32	1	there	there	PRON
elm-5391	32	2	is	be	VERB
elm-5391	32	3	also	also	ADV
elm-5391	32	4	increasing	increase	VERB
elm-5391	32	5	attention	attention	NOUN
elm-5391	32	6	to	to	ADP
elm-5391	32	7	understanding	understand	VERB
elm-5391	32	8	lms	lm	NOUN
elm-5391	32	9	’	'	PUNCT
elm-5391	32	10	abilities	ability	NOUN
elm-5391	32	11	to	to	PART
elm-5391	32	12	represent	represent	VERB
elm-5391	32	13	meanings	meaning	NOUN
elm-5391	32	14	.	.	PUNCT
elm-5391	33	1	in	in	ADP
elm-5391	33	2	semantics	semantic	NOUN
elm-5391	33	3	and	and	CCONJ
elm-5391	33	4	pragmatics	pragmatic	NOUN
elm-5391	33	5	,	,	PUNCT
elm-5391	33	6	promising	promising	ADJ
elm-5391	33	7	and	and	CCONJ
elm-5391	33	8	positive	positive	ADJ
elm-5391	33	9	results	result	NOUN
elm-5391	33	10	seem	seem	VERB
elm-5391	33	11	to	to	PART
elm-5391	33	12	support	support	VERB
elm-5391	33	13	lms	lm	NOUN
elm-5391	33	14	’	'	PUNCT
elm-5391	33	15	increasingly	increasingly	ADV
elm-5391	33	16	sophisticated	sophisticated	ADJ
elm-5391	33	17	abilities	ability	NOUN
elm-5391	33	18	such	such	ADJ
elm-5391	33	19	as	as	ADP
elm-5391	33	20	doing	do	VERB
elm-5391	33	21	natural	natural	ADJ
elm-5391	33	22	language	language	NOUN
elm-5391	33	23	inference	inference	NOUN
elm-5391	33	24	(	(	PUNCT
elm-5391	33	25	e.g.	e.g.	ADV
elm-5391	33	26	,	,	PUNCT
elm-5391	33	27	poliak	poliak	PROPN
elm-5391	33	28	et	et	PROPN
elm-5391	33	29	al	al	PROPN
elm-5391	33	30	.	.	PROPN
elm-5391	33	31	2018	2018	NUM
elm-5391	33	32	,	,	PUNCT
elm-5391	33	33	wang	wang	PROPN
elm-5391	33	34	et	et	PROPN
elm-5391	33	35	al	al	PROPN
elm-5391	33	36	.	.	PROPN
elm-5391	33	37	2018	2018	NUM
elm-5391	33	38	)	)	PUNCT
elm-5391	33	39	.	.	PUNCT
elm-5391	34	1	instead	instead	ADV
elm-5391	34	2	of	of	ADP
elm-5391	34	3	relying	rely	VERB
elm-5391	34	4	on	on	ADP
elm-5391	34	5	direct	direct	ADJ
elm-5391	34	6	output	output	NOUN
elm-5391	34	7	of	of	ADP
elm-5391	34	8	lms	lm	NOUN
elm-5391	34	9	,	,	PUNCT
elm-5391	34	10	finding	find	VERB
elm-5391	34	11	the	the	DET
elm-5391	34	12	relationship	relationship	NOUN
elm-5391	34	13	between	between	ADP
elm-5391	34	14	human	human	ADJ
elm-5391	34	15	understanding	understanding	NOUN
elm-5391	34	16	of	of	ADP
elm-5391	34	17	language	language	NOUN
elm-5391	34	18	and	and	CCONJ
elm-5391	34	19	representations	representation	NOUN
elm-5391	34	20	in	in	ADP
elm-5391	34	21	word	word	NOUN
elm-5391	34	22	embeddings	embedding	NOUN
elm-5391	34	23	have	have	AUX
elm-5391	34	24	also	also	ADV
elm-5391	34	25	been	be	AUX
elm-5391	34	26	fruitful	fruitful	ADJ
elm-5391	34	27	.	.	PUNCT
elm-5391	35	1	2.2	2.2	NUM
elm-5391	35	2	.	.	PUNCT
elm-5391	35	3	word	word	NOUN
elm-5391	35	4	embedding	embed	VERB
elm-5391	35	5	and	and	CCONJ
elm-5391	35	6	lexical	lexical	ADJ
elm-5391	35	7	semantics	semantic	NOUN
elm-5391	35	8	.	.	PUNCT
elm-5391	36	1	since	since	SCONJ
elm-5391	36	2	deep	deep	ADJ
elm-5391	36	3	contextual	contextual	ADJ
elm-5391	36	4	language	language	NOUN
elm-5391	36	5	models	model	NOUN
elm-5391	36	6	provide	provide	VERB
elm-5391	36	7	contextualized	contextualized	ADJ
elm-5391	36	8	word	word	NOUN
elm-5391	36	9	embeddings	embedding	NOUN
elm-5391	36	10	which	which	PRON
elm-5391	36	11	naturally	naturally	ADV
elm-5391	36	12	encode	encode	VERB
elm-5391	36	13	the	the	DET
elm-5391	36	14	distance	distance	NOUN
elm-5391	36	15	between	between	ADP
elm-5391	36	16	word	word	NOUN
elm-5391	36	17	tokens	token	NOUN
elm-5391	36	18	in	in	ADP
elm-5391	36	19	a	a	DET
elm-5391	36	20	vector	vector	NOUN
elm-5391	36	21	space	space	NOUN
elm-5391	36	22	,	,	PUNCT
elm-5391	36	23	this	this	DET
elm-5391	36	24	property	property	NOUN
elm-5391	36	25	can	can	AUX
elm-5391	36	26	be	be	AUX
elm-5391	36	27	utilized	utilize	VERB
elm-5391	36	28	to	to	PART
elm-5391	36	29	test	test	VERB
elm-5391	36	30	whether	whether	SCONJ
elm-5391	36	31	the	the	DET
elm-5391	36	32	trained	train	VERB
elm-5391	36	33	distance	distance	NOUN
elm-5391	36	34	in	in	ADP
elm-5391	36	35	word	word	NOUN
elm-5391	36	36	embeddings	embedding	NOUN
elm-5391	36	37	reflects	reflect	VERB
elm-5391	36	38	the	the	DET
elm-5391	36	39	natural	natural	ADJ
elm-5391	36	40	way	way	NOUN
elm-5391	36	41	words	word	NOUN
elm-5391	36	42	group	group	PROPN
elm-5391	36	43	together	together	ADV
elm-5391	36	44	according	accord	VERB
elm-5391	36	45	to	to	ADP
elm-5391	36	46	our	our	PRON
elm-5391	36	47	lexical	lexical	ADJ
elm-5391	36	48	semantic	semantic	ADJ
elm-5391	36	49	knowledge	knowledge	NOUN
elm-5391	36	50	.	.	PUNCT
elm-5391	37	1	existing	exist	VERB
elm-5391	37	2	studies	study	NOUN
elm-5391	37	3	have	have	AUX
elm-5391	37	4	shown	show	VERB
elm-5391	37	5	that	that	SCONJ
elm-5391	37	6	the	the	DET
elm-5391	37	7	pre	pre	ADJ
elm-5391	37	8	-	-	ADJ
elm-5391	37	9	trained	train	VERB
elm-5391	37	10	bert	bert	NOUN
elm-5391	37	11	model	model	NOUN
elm-5391	37	12	is	be	AUX
elm-5391	37	13	able	able	ADJ
elm-5391	37	14	to	to	PART
elm-5391	37	15	place	place	VERB
elm-5391	37	16	polysemous	polysemous	ADJ
elm-5391	37	17	words	word	NOUN
elm-5391	37	18	that	that	PRON
elm-5391	37	19	appear	appear	VERB
elm-5391	37	20	in	in	ADP
elm-5391	37	21	different	different	ADJ
elm-5391	37	22	contexts	context	NOUN
elm-5391	37	23	into	into	ADP
elm-5391	37	24	distinct	distinct	ADJ
elm-5391	37	25	regions	region	NOUN
elm-5391	37	26	of	of	ADP
elm-5391	37	27	the	the	DET
elm-5391	37	28	shared	share	VERB
elm-5391	37	29	vector	vector	NOUN
elm-5391	37	30	space	space	NOUN
elm-5391	37	31	(	(	PUNCT
elm-5391	37	32	wiedemann	wiedemann	PROPN
elm-5391	37	33	et	et	PROPN
elm-5391	37	34	al	al	PROPN
elm-5391	37	35	.	.	PROPN
elm-5391	37	36	2019	2019	NUM
elm-5391	37	37	)	)	PUNCT
elm-5391	37	38	and	and	CCONJ
elm-5391	37	39	the	the	DET
elm-5391	37	40	word	word	NOUN
elm-5391	37	41	sense	sense	NOUN
elm-5391	37	42	distances	distance	NOUN
elm-5391	37	43	are	be	AUX
elm-5391	37	44	correlated	correlate	VERB
elm-5391	37	45	with	with	ADP
elm-5391	37	46	human	human	ADJ
elm-5391	37	47	judgments	judgment	NOUN
elm-5391	37	48	(	(	PUNCT
elm-5391	37	49	nair	nair	PROPN
elm-5391	37	50	et	et	PROPN
elm-5391	37	51	al	al	PROPN
elm-5391	37	52	.	.	PROPN
elm-5391	37	53	2020	2020	NUM
elm-5391	37	54	)	)	PUNCT
elm-5391	37	55	.	.	PUNCT
elm-5391	38	1	there	there	PRON
elm-5391	38	2	is	be	VERB
elm-5391	38	3	also	also	ADV
elm-5391	38	4	the	the	DET
elm-5391	38	5	general	general	ADJ
elm-5391	38	6	claim	claim	NOUN
elm-5391	38	7	that	that	SCONJ
elm-5391	38	8	contextualized	contextualize	VERB
elm-5391	38	9	word	word	NOUN
elm-5391	38	10	embedding	embed	VERB
elm-5391	38	11	in	in	ADP
elm-5391	38	12	bert	bert	PROPN
elm-5391	38	13	is	be	AUX
elm-5391	38	14	good	good	ADJ
elm-5391	38	15	at	at	ADP
elm-5391	38	16	word	word	NOUN
elm-5391	38	17	sense	sense	NOUN
elm-5391	38	18	disambiguation	disambiguation	NOUN
elm-5391	38	19	(	(	PUNCT
elm-5391	38	20	loureiro	loureiro	PROPN
elm-5391	38	21	et	et	PROPN
elm-5391	38	22	al	al	PROPN
elm-5391	38	23	.	.	PROPN
elm-5391	38	24	2021	2021	NUM
elm-5391	38	25	)	)	PUNCT
elm-5391	38	26	.	.	PUNCT
elm-5391	39	1	but	but	CCONJ
elm-5391	39	2	in	in	ADP
elm-5391	39	3	addition	addition	NOUN
elm-5391	39	4	to	to	ADP
elm-5391	39	5	this	this	DET
elm-5391	39	6	line	line	NOUN
elm-5391	39	7	of	of	ADP
elm-5391	39	8	research	research	NOUN
elm-5391	39	9	on	on	ADP
elm-5391	39	10	word	word	NOUN
elm-5391	39	11	sense	sense	NOUN
elm-5391	39	12	disambiguation	disambiguation	NOUN
elm-5391	39	13	,	,	PUNCT
elm-5391	39	14	investigation	investigation	NOUN
elm-5391	39	15	into	into	ADP
elm-5391	39	16	other	other	ADJ
elm-5391	39	17	aspects	aspect	NOUN
elm-5391	39	18	of	of	ADP
elm-5391	39	19	lexical	lexical	ADJ
elm-5391	39	20	semantics	semantic	NOUN
elm-5391	39	21	is	be	AUX
elm-5391	39	22	limited	limited	ADJ
elm-5391	39	23	.	.	PUNCT
elm-5391	40	1	we	we	PRON
elm-5391	40	2	made	make	VERB
elm-5391	40	3	an	an	DET
elm-5391	40	4	attempt	attempt	NOUN
elm-5391	40	5	in	in	ADP
elm-5391	40	6	this	this	DET
elm-5391	40	7	work	work	NOUN
elm-5391	40	8	to	to	PART
elm-5391	40	9	bring	bring	VERB
elm-5391	40	10	together	together	ADV
elm-5391	40	11	studies	study	NOUN
elm-5391	40	12	of	of	ADP
elm-5391	40	13	word	word	NOUN
elm-5391	40	14	embedding	embed	VERB
elm-5391	40	15	representations	representation	NOUN
elm-5391	40	16	and	and	CCONJ
elm-5391	40	17	another	another	DET
elm-5391	40	18	aspect	aspect	NOUN
elm-5391	40	19	of	of	ADP
elm-5391	40	20	lexically	lexically	ADV
elm-5391	40	21	encoded	encode	VERB
elm-5391	40	22	meaning	meaning	NOUN
elm-5391	40	23	–	–	PUNCT
elm-5391	40	24	the	the	DET
elm-5391	40	25	affect	affect	VERB
elm-5391	40	26	information	information	NOUN
elm-5391	40	27	–	–	PUNCT
elm-5391	40	28	as	as	ADP
elm-5391	40	29	a	a	DET
elm-5391	40	30	novel	novel	ADJ
elm-5391	40	31	case	case	NOUN
elm-5391	40	32	study	study	NOUN
elm-5391	40	33	of	of	ADP
elm-5391	40	34	what	what	PRON
elm-5391	40	35	lms	lm	NOUN
elm-5391	40	36	can	can	AUX
elm-5391	40	37	learn	learn	VERB
elm-5391	40	38	from	from	ADP
elm-5391	40	39	supervised	supervised	ADJ
elm-5391	40	40	training	training	NOUN
elm-5391	40	41	.	.	PUNCT
elm-5391	41	1	2.3	2.3	NUM
elm-5391	41	2	.	.	PUNCT
elm-5391	41	3	affect	affect	VERB
elm-5391	41	4	information	information	NOUN
elm-5391	41	5	in	in	ADP
elm-5391	41	6	nlp	nlp	NOUN
elm-5391	41	7	.	.	PUNCT
elm-5391	42	1	affect	affect	VERB
elm-5391	42	2	information	information	NOUN
elm-5391	42	3	refers	refer	VERB
elm-5391	42	4	to	to	ADP
elm-5391	42	5	any	any	DET
elm-5391	42	6	explicit	explicit	ADJ
elm-5391	42	7	or	or	CCONJ
elm-5391	42	8	implicit	implicit	ADJ
elm-5391	42	9	emotion	emotion	NOUN
elm-5391	42	10	related	relate	VERB
elm-5391	42	11	information	information	NOUN
elm-5391	42	12	.	.	PUNCT
elm-5391	43	1	we	we	PRON
elm-5391	43	2	use	use	VERB
elm-5391	43	3	the	the	DET
elm-5391	43	4	word	word	NOUN
elm-5391	43	5	affect	affect	VERB
elm-5391	43	6	instead	instead	ADV
elm-5391	43	7	of	of	ADP
elm-5391	43	8	emotion	emotion	NOUN
elm-5391	43	9	because	because	SCONJ
elm-5391	43	10	we	we	PRON
elm-5391	43	11	do	do	AUX
elm-5391	43	12	not	not	PART
elm-5391	43	13	rely	rely	VERB
elm-5391	43	14	on	on	ADP
elm-5391	43	15	common	common	ADJ
elm-5391	43	16	emotion	emotion	NOUN
elm-5391	43	17	words	word	NOUN
elm-5391	43	18	such	such	ADJ
elm-5391	43	19	as	as	ADP
elm-5391	43	20	happy	happy	ADJ
elm-5391	43	21	and	and	CCONJ
elm-5391	43	22	sad	sad	ADJ
elm-5391	43	23	as	as	ADP
elm-5391	43	24	baselines	baseline	NOUN
elm-5391	43	25	of	of	ADP
elm-5391	43	26	comparison	comparison	NOUN
elm-5391	43	27	;	;	PUNCT
elm-5391	43	28	rather	rather	ADV
elm-5391	43	29	,	,	PUNCT
elm-5391	43	30	we	we	PRON
elm-5391	43	31	rely	rely	VERB
elm-5391	43	32	on	on	ADP
elm-5391	43	33	three	three	NUM
elm-5391	43	34	primary	primary	ADJ
elm-5391	43	35	independent	independent	ADJ
elm-5391	43	36	dimensions	dimension	NOUN
elm-5391	43	37	of	of	ADP
elm-5391	43	38	emotions	emotion	NOUN
elm-5391	43	39	as	as	ADP
elm-5391	43	40	scales	scale	NOUN
elm-5391	43	41	to	to	PART
elm-5391	43	42	quantify	quantify	VERB
elm-5391	43	43	the	the	DET
elm-5391	43	44	meaning	meaning	NOUN
elm-5391	43	45	of	of	ADP
elm-5391	43	46	emotion	emotion	NOUN
elm-5391	43	47	.	.	PUNCT
elm-5391	44	1	the	the	DET
elm-5391	44	2	three	three	NUM
elm-5391	44	3	dimensions	dimension	NOUN
elm-5391	44	4	are	be	AUX
elm-5391	44	5	valence	valence	NOUN
elm-5391	44	6	(	(	PUNCT
elm-5391	44	7	positiveness	positiveness	NOUN
elm-5391	44	8	-	-	PUNCT
elm-5391	44	9	negativeness	negativeness	NOUN
elm-5391	44	10	/	/	SYM
elm-5391	44	11	pleasantness	pleasantness	NOUN
elm-5391	44	12	-	-	NOUN
elm-5391	44	13	unpleasantness	unpleasantness	NOUN
elm-5391	44	14	)	)	PUNCT
elm-5391	44	15	,	,	PUNCT
elm-5391	44	16	arousal	arousal	NOUN
elm-5391	44	17	(	(	PUNCT
elm-5391	44	18	activepassive	activepassive	ADJ
elm-5391	44	19	,	,	PUNCT
elm-5391	44	20	some	some	DET
elm-5391	44	21	people	people	NOUN
elm-5391	44	22	also	also	ADV
elm-5391	44	23	take	take	VERB
elm-5391	44	24	it	it	PRON
elm-5391	44	25	to	to	PART
elm-5391	44	26	mean	mean	VERB
elm-5391	44	27	the	the	DET
elm-5391	44	28	intensity	intensity	NOUN
elm-5391	44	29	of	of	ADP
elm-5391	44	30	the	the	DET
elm-5391	44	31	emotion	emotion	NOUN
elm-5391	44	32	invoked	invoke	VERB
elm-5391	44	33	by	by	ADP
elm-5391	44	34	the	the	DET
elm-5391	44	35	word	word	NOUN
elm-5391	44	36	)	)	PUNCT
elm-5391	44	37	,	,	PUNCT
elm-5391	44	38	and	and	CCONJ
elm-5391	44	39	dominance	dominance	NOUN
elm-5391	44	40	(	(	PUNCT
elm-5391	44	41	dominant	dominant	ADJ
elm-5391	44	42	-	-	PUNCT
elm-5391	44	43	submissive	submissive	ADJ
elm-5391	44	44	,	,	PUNCT
elm-5391	44	45	or	or	CCONJ
elm-5391	44	46	the	the	DET
elm-5391	44	47	level	level	NOUN
elm-5391	44	48	of	of	ADP
elm-5391	44	49	control	control	NOUN
elm-5391	44	50	exerted	exert	VERB
elm-5391	44	51	by	by	ADP
elm-5391	44	52	the	the	DET
elm-5391	44	53	word	word	NOUN
elm-5391	44	54	)	)	PUNCT
elm-5391	44	55	(	(	PUNCT
elm-5391	44	56	osgood	osgood	PROPN
elm-5391	44	57	et	et	PROPN
elm-5391	44	58	al	al	PROPN
elm-5391	44	59	.	.	PROPN
elm-5391	44	60	1957	1957	NUM
elm-5391	44	61	,	,	PUNCT
elm-5391	44	62	russell	russell	PROPN
elm-5391	44	63	1980	1980	NUM
elm-5391	44	64	,	,	PUNCT
elm-5391	44	65	2003	2003	NUM
elm-5391	44	66	)	)	PUNCT
elm-5391	44	67	.	.	PUNCT
elm-5391	45	1	another	another	DET
elm-5391	45	2	reason	reason	NOUN
elm-5391	45	3	to	to	PART
elm-5391	45	4	choose	choose	VERB
elm-5391	45	5	this	this	DET
elm-5391	45	6	terminology	terminology	NOUN
elm-5391	45	7	is	be	AUX
elm-5391	45	8	due	due	ADJ
elm-5391	45	9	to	to	ADP
elm-5391	45	10	its	its	PRON
elm-5391	45	11	validity	validity	NOUN
elm-5391	45	12	and	and	CCONJ
elm-5391	45	13	consistency	consistency	NOUN
elm-5391	45	14	:	:	PUNCT
elm-5391	45	15	there	there	PRON
elm-5391	45	16	are	be	VERB
elm-5391	45	17	already	already	ADV
elm-5391	45	18	multiple	multiple	ADJ
elm-5391	45	19	human	human	ADV
elm-5391	45	20	-	-	PUNCT
elm-5391	45	21	labeled	label	VERB
elm-5391	45	22	datasets	dataset	NOUN
elm-5391	45	23	that	that	PRON
elm-5391	45	24	are	be	AUX
elm-5391	45	25	based	base	VERB
elm-5391	45	26	on	on	ADP
elm-5391	45	27	these	these	DET
elm-5391	45	28	scales	scale	NOUN
elm-5391	45	29	as	as	ADP
elm-5391	45	30	the	the	DET
elm-5391	45	31	quantitative	quantitative	ADJ
elm-5391	45	32	ground	ground	NOUN
elm-5391	45	33	truth	truth	NOUN
elm-5391	45	34	(	(	PUNCT
elm-5391	45	35	bradley	bradley	PROPN
elm-5391	45	36	&	&	CCONJ
elm-5391	45	37	lang	lang	PROPN
elm-5391	45	38	1999	1999	NUM
elm-5391	45	39	,	,	PUNCT
elm-5391	45	40	mohammad	mohammad	PROPN
elm-5391	45	41	2018	2018	NUM
elm-5391	45	42	,	,	PUNCT
elm-5391	45	43	warriner	warriner	NOUN
elm-5391	45	44	et	et	PROPN
elm-5391	45	45	al	al	PROPN
elm-5391	45	46	.	.	PROPN
elm-5391	45	47	2013	2013	NUM
elm-5391	45	48	)	)	PUNCT
elm-5391	45	49	.	.	PUNCT
elm-5391	46	1	in	in	ADP
elm-5391	46	2	the	the	DET
elm-5391	46	3	fields	field	NOUN
elm-5391	46	4	that	that	PRON
elm-5391	46	5	are	be	AUX
elm-5391	46	6	related	relate	VERB
elm-5391	46	7	to	to	ADP
elm-5391	46	8	emotion	emotion	NOUN
elm-5391	46	9	recognition	recognition	NOUN
elm-5391	46	10	,	,	PUNCT
elm-5391	46	11	sentiment	sentiment	NOUN
elm-5391	46	12	analysis	analysis	NOUN
elm-5391	46	13	,	,	PUNCT
elm-5391	46	14	and	and	CCONJ
elm-5391	46	15	affective	affective	ADJ
elm-5391	46	16	comproceedings	comproceeding	NOUN
elm-5391	46	17	of	of	ADP
elm-5391	46	18	elm	elm	NOUN
elm-5391	46	19	2	2	NUM
elm-5391	46	20	:	:	PUNCT
elm-5391	46	21	310	310	NUM
elm-5391	46	22	-	-	SYM
elm-5391	46	23	321	321	NUM
elm-5391	46	24	,	,	PUNCT
elm-5391	46	25	2023	2023	NUM
elm-5391	46	26	yuhan	yuhan	PROPN
elm-5391	46	27	zhang	zhang	PROPN
elm-5391	46	28	,	,	PUNCT
elm-5391	46	29	wenqi	wenqi	PROPN
elm-5391	46	30	chen	chen	PROPN
elm-5391	46	31	,	,	PUNCT
elm-5391	46	32	ruihan	ruihan	PROPN
elm-5391	46	33	zhang	zhang	PROPN
elm-5391	46	34	and	and	CCONJ
elm-5391	46	35	xiajie	xiajie	PROPN
elm-5391	46	36	zhang	zhang	PROPN
elm-5391	46	37	:	:	PUNCT
elm-5391	46	38	representing	represent	VERB
elm-5391	46	39	affect	affect	VERB
elm-5391	46	40	information	information	NOUN
elm-5391	46	41	in	in	ADP
elm-5391	46	42	word	word	NOUN
elm-5391	46	43	embeddings	embedding	NOUN
elm-5391	46	44	.	.	PUNCT
elm-5391	47	1	311	311	NUM
elm-5391	47	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	47	3	https://www.elm-conference.net/	https://www.elm-conference.net/	ADP
elm-5391	47	4	puting	put	VERB
elm-5391	47	5	,	,	PUNCT
elm-5391	47	6	transformer	transformer	NOUN
elm-5391	47	7	-	-	PUNCT
elm-5391	47	8	based	base	VERB
elm-5391	47	9	models	model	NOUN
elm-5391	47	10	and	and	CCONJ
elm-5391	47	11	more	more	ADV
elm-5391	47	12	fine	fine	ADV
elm-5391	47	13	-	-	PUNCT
elm-5391	47	14	grained	grain	VERB
elm-5391	47	15	annotation	annotation	NOUN
elm-5391	47	16	of	of	ADP
elm-5391	47	17	emotion	emotion	NOUN
elm-5391	47	18	categories	category	NOUN
elm-5391	47	19	have	have	AUX
elm-5391	47	20	gathered	gather	VERB
elm-5391	47	21	the	the	DET
elm-5391	47	22	attention	attention	NOUN
elm-5391	47	23	of	of	ADP
elm-5391	47	24	the	the	DET
elm-5391	47	25	nlp	nlp	PROPN
elm-5391	47	26	/	/	SYM
elm-5391	47	27	nlu	nlu	NOUN
elm-5391	47	28	community	community	NOUN
elm-5391	47	29	(	(	PUNCT
elm-5391	47	30	e.g.	e.g.	ADV
elm-5391	47	31	,	,	PUNCT
elm-5391	47	32	alswaidan	alswaidan	PROPN
elm-5391	47	33	&	&	CCONJ
elm-5391	47	34	menai	menai	PROPN
elm-5391	47	35	2020	2020	NUM
elm-5391	47	36	,	,	PUNCT
elm-5391	47	37	demszky	demszky	PROPN
elm-5391	47	38	et	et	PROPN
elm-5391	47	39	al	al	PROPN
elm-5391	47	40	.	.	PROPN
elm-5391	47	41	2020	2020	NUM
elm-5391	47	42	,	,	PUNCT
elm-5391	47	43	rashkin	rashkin	VERB
elm-5391	47	44	et	et	PROPN
elm-5391	47	45	al	al	PROPN
elm-5391	47	46	.	.	PROPN
elm-5391	47	47	2018	2018	NUM
elm-5391	47	48	,	,	PUNCT
elm-5391	47	49	suresh	suresh	PROPN
elm-5391	47	50	&	&	CCONJ
elm-5391	47	51	ong	ong	PROPN
elm-5391	47	52	2021	2021	NUM
elm-5391	47	53	)	)	PUNCT
elm-5391	47	54	.	.	PUNCT
elm-5391	48	1	zooming	zoom	VERB
elm-5391	48	2	into	into	ADP
elm-5391	48	3	existing	exist	VERB
elm-5391	48	4	studies	study	NOUN
elm-5391	48	5	relevant	relevant	ADJ
elm-5391	48	6	to	to	ADP
elm-5391	48	7	word	word	NOUN
elm-5391	48	8	embeddings	embedding	NOUN
elm-5391	48	9	and	and	CCONJ
elm-5391	48	10	affect	affect	VERB
elm-5391	48	11	information	information	NOUN
elm-5391	48	12	,	,	PUNCT
elm-5391	48	13	it	it	PRON
elm-5391	48	14	has	have	AUX
elm-5391	48	15	been	be	AUX
elm-5391	48	16	shown	show	VERB
elm-5391	48	17	that	that	SCONJ
elm-5391	48	18	multi	multi	ADJ
elm-5391	48	19	-	-	ADJ
elm-5391	48	20	dimensional	dimensional	ADJ
elm-5391	48	21	embeddings	embedding	NOUN
elm-5391	48	22	of	of	ADP
elm-5391	48	23	emotional	emotional	ADJ
elm-5391	48	24	words	word	NOUN
elm-5391	48	25	achieved	achieve	VERB
elm-5391	48	26	higher	high	ADJ
elm-5391	48	27	performance	performance	NOUN
elm-5391	48	28	in	in	ADP
elm-5391	48	29	sentiment	sentiment	NOUN
elm-5391	48	30	analysis	analysis	NOUN
elm-5391	48	31	tasks	task	NOUN
elm-5391	48	32	than	than	ADP
elm-5391	48	33	human	human	ADJ
elm-5391	48	34	annotated	annotate	VERB
elm-5391	48	35	emotion	emotion	NOUN
elm-5391	48	36	vectors	vector	NOUN
elm-5391	48	37	(	(	PUNCT
elm-5391	48	38	li	li	PROPN
elm-5391	48	39	et	et	PROPN
elm-5391	48	40	al	al	PROPN
elm-5391	48	41	.	.	PROPN
elm-5391	48	42	2017	2017	NUM
elm-5391	48	43	)	)	PUNCT
elm-5391	48	44	.	.	PUNCT
elm-5391	49	1	furthermore	furthermore	ADV
elm-5391	49	2	,	,	PUNCT
elm-5391	49	3	the	the	DET
elm-5391	49	4	multi	multi	ADJ
elm-5391	49	5	-	-	ADJ
elm-5391	49	6	dimensional	dimensional	ADJ
elm-5391	49	7	word	word	NOUN
elm-5391	49	8	embeddings	embedding	NOUN
elm-5391	49	9	for	for	ADP
elm-5391	49	10	emotional	emotional	ADJ
elm-5391	49	11	words	word	NOUN
elm-5391	49	12	are	be	AUX
elm-5391	49	13	often	often	ADV
elm-5391	49	14	added	add	VERB
elm-5391	49	15	onto	onto	ADP
elm-5391	49	16	base	base	ADJ
elm-5391	49	17	static	static	ADJ
elm-5391	49	18	word	word	NOUN
elm-5391	49	19	embeddings	embedding	NOUN
elm-5391	49	20	for	for	ADP
elm-5391	49	21	downstream	downstream	ADJ
elm-5391	49	22	sentiment	sentiment	NOUN
elm-5391	49	23	analysis	analysis	NOUN
elm-5391	49	24	tasks	task	NOUN
elm-5391	49	25	(	(	PUNCT
elm-5391	49	26	mao	mao	X
elm-5391	49	27	et	et	PROPN
elm-5391	49	28	al	al	PROPN
elm-5391	49	29	.	.	PROPN
elm-5391	49	30	2019	2019	NUM
elm-5391	49	31	)	)	PUNCT
elm-5391	49	32	.	.	PUNCT
elm-5391	50	1	while	while	SCONJ
elm-5391	50	2	existing	exist	VERB
elm-5391	50	3	research	research	NOUN
elm-5391	50	4	focused	focus	VERB
elm-5391	50	5	more	more	ADV
elm-5391	50	6	on	on	ADP
elm-5391	50	7	the	the	DET
elm-5391	50	8	application	application	NOUN
elm-5391	50	9	of	of	ADP
elm-5391	50	10	emotional	emotional	ADJ
elm-5391	50	11	word	word	NOUN
elm-5391	50	12	embeddings	embedding	NOUN
elm-5391	50	13	onto	onto	ADP
elm-5391	50	14	sentiment	sentiment	NOUN
elm-5391	50	15	analysis	analysis	NOUN
elm-5391	50	16	tasks	task	NOUN
elm-5391	50	17	,	,	PUNCT
elm-5391	50	18	seldom	seldom	ADV
elm-5391	50	19	has	have	AUX
elm-5391	50	20	explored	explore	VERB
elm-5391	50	21	whether	whether	SCONJ
elm-5391	50	22	state	state	NOUN
elm-5391	50	23	-	-	PUNCT
elm-5391	50	24	of	of	ADP
elm-5391	50	25	-	-	PUNCT
elm-5391	50	26	the	the	DET
elm-5391	50	27	-	-	PUNCT
elm-5391	50	28	art	art	NOUN
elm-5391	50	29	contextualized	contextualize	VERB
elm-5391	50	30	word	word	NOUN
elm-5391	50	31	embeddings	embedding	NOUN
elm-5391	50	32	have	have	AUX
elm-5391	50	33	learned	learn	VERB
elm-5391	50	34	the	the	DET
elm-5391	50	35	affect	affect	VERB
elm-5391	50	36	meaning	meaning	NOUN
elm-5391	50	37	of	of	ADP
elm-5391	50	38	all	all	DET
elm-5391	50	39	kinds	kind	NOUN
elm-5391	50	40	of	of	ADP
elm-5391	50	41	words	word	NOUN
elm-5391	50	42	.	.	PUNCT
elm-5391	51	1	our	our	PRON
elm-5391	51	2	study	study	NOUN
elm-5391	51	3	is	be	AUX
elm-5391	51	4	set	set	VERB
elm-5391	51	5	up	up	ADP
elm-5391	51	6	to	to	PART
elm-5391	51	7	fill	fill	VERB
elm-5391	51	8	the	the	DET
elm-5391	51	9	gap	gap	NOUN
elm-5391	51	10	.	.	PUNCT
elm-5391	52	1	3	3	X
elm-5391	52	2	.	.	X
elm-5391	52	3	methodology	methodology	NOUN
elm-5391	52	4	.	.	PUNCT
elm-5391	53	1	our	our	PRON
elm-5391	53	2	methods	method	NOUN
elm-5391	53	3	focus	focus	VERB
elm-5391	53	4	on	on	ADP
elm-5391	53	5	constructing	construct	VERB
elm-5391	53	6	the	the	DET
elm-5391	53	7	relationship	relationship	NOUN
elm-5391	53	8	between	between	ADP
elm-5391	53	9	word	word	NOUN
elm-5391	53	10	embeddings	embedding	NOUN
elm-5391	53	11	and	and	CCONJ
elm-5391	53	12	human	human	ADJ
elm-5391	53	13	cognition	cognition	NOUN
elm-5391	53	14	.	.	PUNCT
elm-5391	54	1	on	on	ADP
elm-5391	54	2	the	the	DET
elm-5391	54	3	human	human	ADJ
elm-5391	54	4	cognition	cognition	NOUN
elm-5391	54	5	side	side	NOUN
elm-5391	54	6	,	,	PUNCT
elm-5391	54	7	we	we	PRON
elm-5391	54	8	adapt	adapt	VERB
elm-5391	54	9	the	the	DET
elm-5391	54	10	dataset	dataset	NOUN
elm-5391	54	11	from	from	ADP
elm-5391	54	12	the	the	DET
elm-5391	54	13	affect	affect	NOUN
elm-5391	54	14	ratings	rating	NOUN
elm-5391	54	15	of	of	ADP
elm-5391	54	16	humans	human	NOUN
elm-5391	54	17	from	from	ADP
elm-5391	54	18	cognitive	cognitive	ADJ
elm-5391	54	19	studies	study	NOUN
elm-5391	54	20	.	.	PUNCT
elm-5391	55	1	on	on	ADP
elm-5391	55	2	the	the	DET
elm-5391	55	3	word	word	NOUN
elm-5391	55	4	embedding	embed	VERB
elm-5391	55	5	side	side	NOUN
elm-5391	55	6	,	,	PUNCT
elm-5391	55	7	we	we	PRON
elm-5391	55	8	adopted	adopt	VERB
elm-5391	55	9	both	both	CCONJ
elm-5391	55	10	contextualized	contextualized	ADJ
elm-5391	55	11	and	and	CCONJ
elm-5391	55	12	static	static	ADJ
elm-5391	55	13	word	word	NOUN
elm-5391	55	14	embeddings	embedding	NOUN
elm-5391	55	15	.	.	PUNCT
elm-5391	56	1	in	in	ADP
elm-5391	56	2	order	order	NOUN
elm-5391	56	3	to	to	PART
elm-5391	56	4	probe	probe	VERB
elm-5391	56	5	the	the	DET
elm-5391	56	6	relationships	relationship	NOUN
elm-5391	56	7	,	,	PUNCT
elm-5391	56	8	we	we	PRON
elm-5391	56	9	take	take	VERB
elm-5391	56	10	both	both	CCONJ
elm-5391	56	11	the	the	DET
elm-5391	56	12	unsupervised	unsupervised	ADJ
elm-5391	56	13	and	and	CCONJ
elm-5391	56	14	supervised	supervised	ADJ
elm-5391	56	15	methods	method	NOUN
elm-5391	56	16	to	to	PART
elm-5391	56	17	reflect	reflect	VERB
elm-5391	56	18	the	the	DET
elm-5391	56	19	relationships	relationship	NOUN
elm-5391	56	20	.	.	PUNCT
elm-5391	57	1	3.1	3.1	NUM
elm-5391	57	2	.	.	PUNCT
elm-5391	57	3	affect	affect	VERB
elm-5391	57	4	ratings	rating	NOUN
elm-5391	57	5	of	of	ADP
elm-5391	57	6	human	human	NOUN
elm-5391	57	7	as	as	ADP
elm-5391	57	8	the	the	DET
elm-5391	57	9	ground	ground	NOUN
elm-5391	57	10	truth	truth	NOUN
elm-5391	57	11	.	.	PUNCT
elm-5391	58	1	we	we	PRON
elm-5391	58	2	selected	select	VERB
elm-5391	58	3	the	the	DET
elm-5391	58	4	nrc	nrc	PROPN
elm-5391	58	5	vad	vad	PROPN
elm-5391	58	6	dataset	dataset	PROPN
elm-5391	58	7	(	(	PUNCT
elm-5391	58	8	national	national	ADJ
elm-5391	58	9	research	research	NOUN
elm-5391	58	10	councial	councial	ADJ
elm-5391	58	11	(	(	PUNCT
elm-5391	58	12	canada	canada	PROPN
elm-5391	58	13	)	)	PUNCT
elm-5391	58	14	valence	valence	NOUN
elm-5391	58	15	arousal	arousal	NOUN
elm-5391	58	16	dominance	dominance	NOUN
elm-5391	58	17	)	)	PUNCT
elm-5391	58	18	curated	curate	VERB
elm-5391	58	19	by	by	ADP
elm-5391	58	20	mohammad	mohammad	PROPN
elm-5391	58	21	(	(	PUNCT
elm-5391	58	22	2018	2018	NUM
elm-5391	58	23	)	)	PUNCT
elm-5391	58	24	as	as	ADP
elm-5391	58	25	the	the	DET
elm-5391	58	26	ground	ground	NOUN
elm-5391	58	27	truth	truth	NOUN
elm-5391	58	28	for	for	ADP
elm-5391	58	29	affect	affect	VERB
elm-5391	58	30	meanings	meaning	NOUN
elm-5391	58	31	of	of	ADP
elm-5391	58	32	english	english	PROPN
elm-5391	58	33	words.1	words.1	VERB
elm-5391	58	34	there	there	PRON
elm-5391	58	35	are	be	VERB
elm-5391	58	36	20,000	20,000	NUM
elm-5391	58	37	english	english	ADJ
elm-5391	58	38	words	word	NOUN
elm-5391	58	39	annotated	annotate	VERB
elm-5391	58	40	along	along	ADP
elm-5391	58	41	the	the	DET
elm-5391	58	42	three	three	NUM
elm-5391	58	43	independent	independent	ADJ
elm-5391	58	44	dimensions	dimension	NOUN
elm-5391	58	45	of	of	ADP
elm-5391	58	46	affect	affect	NOUN
elm-5391	58	47	.	.	PUNCT
elm-5391	59	1	in	in	ADP
elm-5391	59	2	each	each	DET
elm-5391	59	3	dimension	dimension	NOUN
elm-5391	59	4	,	,	PUNCT
elm-5391	59	5	the	the	DET
elm-5391	59	6	rating	rating	NOUN
elm-5391	59	7	starts	start	VERB
elm-5391	59	8	from	from	ADP
elm-5391	59	9	0	0	NUM
elm-5391	59	10	to	to	ADP
elm-5391	59	11	1	1	NUM
elm-5391	59	12	where	where	SCONJ
elm-5391	59	13	0	0	NUM
elm-5391	59	14	means	mean	VERB
elm-5391	59	15	the	the	DET
elm-5391	59	16	most	most	ADV
elm-5391	59	17	negative	negative	ADJ
elm-5391	59	18	,	,	PUNCT
elm-5391	59	19	passive	passive	ADJ
elm-5391	59	20	,	,	PUNCT
elm-5391	59	21	or	or	CCONJ
elm-5391	59	22	submissive	submissive	ADJ
elm-5391	59	23	in	in	ADP
elm-5391	59	24	the	the	DET
elm-5391	59	25	v	v	NOUN
elm-5391	59	26	,	,	PUNCT
elm-5391	59	27	a	a	PRON
elm-5391	59	28	,	,	PUNCT
elm-5391	59	29	or	or	CCONJ
elm-5391	59	30	d	d	ADP
elm-5391	59	31	dimension	dimension	NOUN
elm-5391	59	32	and	and	CCONJ
elm-5391	59	33	1	1	NUM
elm-5391	59	34	means	mean	VERB
elm-5391	59	35	the	the	DET
elm-5391	59	36	most	most	ADV
elm-5391	59	37	positive	positive	ADJ
elm-5391	59	38	,	,	PUNCT
elm-5391	59	39	active	active	ADJ
elm-5391	59	40	,	,	PUNCT
elm-5391	59	41	or	or	CCONJ
elm-5391	59	42	dominant	dominant	ADJ
elm-5391	59	43	in	in	ADP
elm-5391	59	44	their	their	PRON
elm-5391	59	45	respective	respective	ADJ
elm-5391	59	46	dimension	dimension	NOUN
elm-5391	59	47	.	.	PUNCT
elm-5391	60	1	this	this	PRON
elm-5391	60	2	is	be	AUX
elm-5391	60	3	the	the	DET
elm-5391	60	4	largest	large	ADJ
elm-5391	60	5	manually	manually	ADV
elm-5391	60	6	created	create	VERB
elm-5391	60	7	vad	vad	PROPN
elm-5391	60	8	corpus	corpus	NOUN
elm-5391	60	9	in	in	ADP
elm-5391	60	10	any	any	DET
elm-5391	60	11	language	language	NOUN
elm-5391	60	12	.	.	PUNCT
elm-5391	61	1	3.2	3.2	NUM
elm-5391	61	2	.	.	PUNCT
elm-5391	62	1	word	word	NOUN
elm-5391	62	2	embeddings	embedding	NOUN
elm-5391	62	3	under	under	ADP
elm-5391	62	4	study	study	NOUN
elm-5391	62	5	.	.	PUNCT
elm-5391	63	1	in	in	ADP
elm-5391	63	2	this	this	DET
elm-5391	63	3	paper	paper	NOUN
elm-5391	63	4	,	,	PUNCT
elm-5391	63	5	we	we	PRON
elm-5391	63	6	studied	study	VERB
elm-5391	63	7	four	four	NUM
elm-5391	63	8	kinds	kind	NOUN
elm-5391	63	9	of	of	ADP
elm-5391	63	10	word	word	NOUN
elm-5391	63	11	embeddings	embedding	NOUN
elm-5391	63	12	as	as	SCONJ
elm-5391	63	13	described	describe	VERB
elm-5391	63	14	below	below	ADV
elm-5391	63	15	.	.	PUNCT
elm-5391	64	1	we	we	PRON
elm-5391	64	2	took	take	VERB
elm-5391	64	3	from	from	ADP
elm-5391	64	4	each	each	DET
elm-5391	64	5	word	word	NOUN
elm-5391	64	6	embedding	embed	VERB
elm-5391	64	7	the	the	DET
elm-5391	64	8	same	same	ADJ
elm-5391	64	9	subset	subset	NOUN
elm-5391	64	10	of	of	ADP
elm-5391	64	11	words	word	NOUN
elm-5391	64	12	that	that	PRON
elm-5391	64	13	are	be	AUX
elm-5391	64	14	compatible	compatible	ADJ
elm-5391	64	15	with	with	ADP
elm-5391	64	16	the	the	DET
elm-5391	64	17	nrc	nrc	PROPN
elm-5391	64	18	vad	vad	PROPN
elm-5391	64	19	corpus	corpus	PROPN
elm-5391	64	20	.	.	PUNCT
elm-5391	65	1	the	the	DET
elm-5391	65	2	embeddings	embedding	NOUN
elm-5391	65	3	differ	differ	VERB
elm-5391	65	4	from	from	ADP
elm-5391	65	5	each	each	DET
elm-5391	65	6	other	other	ADJ
elm-5391	65	7	by	by	ADP
elm-5391	65	8	how	how	SCONJ
elm-5391	65	9	much	much	ADJ
elm-5391	65	10	and	and	CCONJ
elm-5391	65	11	what	what	DET
elm-5391	65	12	kinds	kind	NOUN
elm-5391	65	13	of	of	ADP
elm-5391	65	14	contextualized	contextualized	ADJ
elm-5391	65	15	information	information	NOUN
elm-5391	65	16	are	be	AUX
elm-5391	65	17	captured	capture	VERB
elm-5391	65	18	from	from	ADP
elm-5391	65	19	training	training	NOUN
elm-5391	65	20	.	.	PUNCT
elm-5391	66	1	we	we	PRON
elm-5391	66	2	took	take	VERB
elm-5391	66	3	glove	glove	NOUN
elm-5391	66	4	as	as	ADP
elm-5391	66	5	a	a	DET
elm-5391	66	6	representative	representative	NOUN
elm-5391	66	7	of	of	ADP
elm-5391	66	8	existing	exist	VERB
elm-5391	66	9	static	static	ADJ
elm-5391	66	10	word	word	NOUN
elm-5391	66	11	embeddings	embedding	NOUN
elm-5391	66	12	in	in	ADP
elm-5391	66	13	the	the	DET
elm-5391	66	14	nlp	nlp	NOUN
elm-5391	66	15	community	community	NOUN
elm-5391	66	16	and	and	CCONJ
elm-5391	66	17	compared	compare	VERB
elm-5391	66	18	it	it	PRON
elm-5391	66	19	with	with	ADP
elm-5391	66	20	more	more	ADV
elm-5391	66	21	contextualized	contextualized	ADJ
elm-5391	66	22	ones	one	NOUN
elm-5391	66	23	.	.	PUNCT
elm-5391	67	1	we	we	PRON
elm-5391	67	2	also	also	ADV
elm-5391	67	3	varied	vary	VERB
elm-5391	67	4	the	the	DET
elm-5391	67	5	kinds	kind	NOUN
elm-5391	67	6	of	of	ADP
elm-5391	67	7	contextualized	contextualized	ADJ
elm-5391	67	8	information	information	NOUN
elm-5391	67	9	by	by	ADP
elm-5391	67	10	prioritizing	prioritize	VERB
elm-5391	67	11	emotion	emotion	NOUN
elm-5391	67	12	related	relate	VERB
elm-5391	67	13	and	and	CCONJ
elm-5391	67	14	sentiment	sentiment	NOUN
elm-5391	67	15	analysis	analysis	NOUN
elm-5391	67	16	related	relate	VERB
elm-5391	67	17	information	information	NOUN
elm-5391	67	18	.	.	PUNCT
elm-5391	68	1	•	•	NUM
elm-5391	68	2	pre	pre	ADJ
elm-5391	68	3	-	-	ADJ
elm-5391	68	4	trained	train	VERB
elm-5391	68	5	word	word	NOUN
elm-5391	68	6	embeddings	embedding	NOUN
elm-5391	68	7	using	use	VERB
elm-5391	68	8	the	the	DET
elm-5391	68	9	glove	glove	NOUN
elm-5391	68	10	algorithm	algorithm	NOUN
elm-5391	68	11	(	(	PUNCT
elm-5391	68	12	pennington	pennington	PROPN
elm-5391	68	13	et	et	PROPN
elm-5391	68	14	al	al	PROPN
elm-5391	68	15	.	.	PROPN
elm-5391	68	16	2014	2014	NUM
elm-5391	68	17	)	)	PUNCT
elm-5391	68	18	which	which	PRON
elm-5391	68	19	we	we	PRON
elm-5391	68	20	refer	refer	VERB
elm-5391	68	21	to	to	ADP
elm-5391	68	22	as	as	ADP
elm-5391	68	23	the	the	DET
elm-5391	68	24	glove	glove	NOUN
elm-5391	68	25	embedding	embed	VERB
elm-5391	68	26	in	in	ADP
elm-5391	68	27	this	this	DET
elm-5391	68	28	paper	paper	NOUN
elm-5391	68	29	.	.	PUNCT
elm-5391	69	1	the	the	DET
elm-5391	69	2	word	word	NOUN
elm-5391	69	3	embeddings	embedding	NOUN
elm-5391	69	4	were	be	AUX
elm-5391	69	5	downloaded	download	VERB
elm-5391	69	6	from	from	ADP
elm-5391	69	7	stanford	stanford	PROPN
elm-5391	69	8	nlp	nlp	PROPN
elm-5391	69	9	website2	website2	PROPN
elm-5391	69	10	which	which	PRON
elm-5391	69	11	were	be	AUX
elm-5391	69	12	trained	train	VERB
elm-5391	69	13	on	on	ADP
elm-5391	69	14	wikipedia	wikipedia	PROPN
elm-5391	69	15	and	and	CCONJ
elm-5391	69	16	gigaword	gigaword	PROPN
elm-5391	69	17	5	5	NUM
elm-5391	69	18	data	datum	NOUN
elm-5391	69	19	.	.	PUNCT
elm-5391	70	1	•	•	NUM
elm-5391	70	2	pre	pre	VERB
elm-5391	70	3	-	-	ADJ
elm-5391	70	4	trained	train	VERB
elm-5391	70	5	word	word	NOUN
elm-5391	70	6	embeddings	embedding	NOUN
elm-5391	70	7	that	that	PRON
elm-5391	70	8	were	be	AUX
elm-5391	70	9	retrieved	retrieve	VERB
elm-5391	70	10	from	from	ADP
elm-5391	70	11	the	the	DET
elm-5391	70	12	last	last	ADJ
elm-5391	70	13	hidden	hide	VERB
elm-5391	70	14	layer	layer	NOUN
elm-5391	70	15	of	of	ADP
elm-5391	70	16	the	the	DET
elm-5391	70	17	base	base	ADJ
elm-5391	70	18	bert	bert	PROPN
elm-5391	70	19	model	model	NOUN
elm-5391	70	20	(	(	PUNCT
elm-5391	70	21	devlin	devlin	PROPN
elm-5391	70	22	et	et	PROPN
elm-5391	70	23	al	al	PROPN
elm-5391	70	24	.	.	PROPN
elm-5391	70	25	2019	2019	NUM
elm-5391	70	26	)	)	PUNCT
elm-5391	70	27	after	after	SCONJ
elm-5391	70	28	we	we	PRON
elm-5391	70	29	directly	directly	ADV
elm-5391	70	30	ran	run	VERB
elm-5391	70	31	the	the	DET
elm-5391	70	32	model	model	NOUN
elm-5391	70	33	over	over	ADP
elm-5391	70	34	the	the	DET
elm-5391	70	35	nrc	nrc	PROPN
elm-5391	70	36	word	word	NOUN
elm-5391	70	37	list	list	NOUN
elm-5391	70	38	.	.	PUNCT
elm-5391	71	1	we	we	PRON
elm-5391	71	2	refer	refer	VERB
elm-5391	71	3	to	to	ADP
elm-5391	71	4	this	this	DET
elm-5391	71	5	type	type	NOUN
elm-5391	71	6	of	of	ADP
elm-5391	71	7	word	word	NOUN
elm-5391	71	8	embedding	embed	VERB
elm-5391	71	9	representation	representation	NOUN
elm-5391	71	10	as	as	ADP
elm-5391	71	11	base	base	NOUN
elm-5391	71	12	bert	bert	NOUN
elm-5391	71	13	throughout	throughout	ADP
elm-5391	71	14	the	the	DET
elm-5391	71	15	paper	paper	NOUN
elm-5391	71	16	.	.	PUNCT
elm-5391	72	1	•	•	NUM
elm-5391	72	2	pre	pre	ADJ
elm-5391	72	3	-	-	ADJ
elm-5391	72	4	trained	train	VERB
elm-5391	72	5	word	word	NOUN
elm-5391	72	6	embeddings	embedding	NOUN
elm-5391	72	7	retrieved	retrieve	VERB
elm-5391	72	8	from	from	ADP
elm-5391	72	9	the	the	DET
elm-5391	72	10	last	last	ADJ
elm-5391	72	11	hidden	hide	VERB
elm-5391	72	12	layer	layer	NOUN
elm-5391	72	13	of	of	ADP
elm-5391	72	14	the	the	DET
elm-5391	72	15	bert	bert	PROPN
elm-5391	72	16	model	model	NOUN
elm-5391	72	17	fined	fine	VERB
elm-5391	72	18	tuned	tune	VERB
elm-5391	72	19	on	on	ADP
elm-5391	72	20	the	the	DET
elm-5391	72	21	goemotion	goemotion	NOUN
elm-5391	72	22	dataset	dataset	NOUN
elm-5391	72	23	(	(	PUNCT
elm-5391	72	24	demszky	demszky	PROPN
elm-5391	72	25	et	et	PROPN
elm-5391	72	26	al	al	PROPN
elm-5391	72	27	.	.	PROPN
elm-5391	72	28	2020	2020	NUM
elm-5391	72	29	)	)	PUNCT
elm-5391	72	30	.	.	PUNCT
elm-5391	73	1	the	the	DET
elm-5391	73	2	goemotion	goemotion	NOUN
elm-5391	73	3	dataset	dataset	NOUN
elm-5391	73	4	contains	contain	VERB
elm-5391	73	5	58k	58k	ADP
elm-5391	73	6	english	english	ADJ
elm-5391	73	7	reddit	reddit	NOUN
elm-5391	73	8	comments	comment	NOUN
elm-5391	73	9	labeled	label	VERB
elm-5391	73	10	by	by	ADP
elm-5391	73	11	humans	human	NOUN
elm-5391	73	12	on	on	ADP
elm-5391	73	13	27	27	NUM
elm-5391	73	14	emotion	emotion	NOUN
elm-5391	73	15	categories	category	NOUN
elm-5391	73	16	.	.	PUNCT
elm-5391	74	1	we	we	PRON
elm-5391	74	2	refer	refer	VERB
elm-5391	74	3	to	to	ADP
elm-5391	74	4	this	this	DET
elm-5391	74	5	1data	1data	NUM
elm-5391	74	6	can	can	AUX
elm-5391	74	7	be	be	AUX
elm-5391	74	8	downloaded	download	VERB
elm-5391	74	9	from	from	ADP
elm-5391	74	10	http://saifmohammad.com/webpages/nrc-vad.html	http://saifmohammad.com/webpages/nrc-vad.html	NOUN
elm-5391	74	11	2https://nlp.stanford.edu/projects/glove/	2https://nlp.stanford.edu/projects/glove/	NUM
elm-5391	74	12	proceedings	proceeding	NOUN
elm-5391	74	13	of	of	ADP
elm-5391	74	14	elm	elm	NOUN
elm-5391	74	15	2	2	NUM
elm-5391	74	16	:	:	PUNCT
elm-5391	74	17	310	310	NUM
elm-5391	74	18	-	-	SYM
elm-5391	74	19	321	321	NUM
elm-5391	74	20	,	,	PUNCT
elm-5391	74	21	2023	2023	NUM
elm-5391	74	22	yuhan	yuhan	PROPN
elm-5391	74	23	zhang	zhang	PROPN
elm-5391	74	24	,	,	PUNCT
elm-5391	74	25	wenqi	wenqi	PROPN
elm-5391	74	26	chen	chen	PROPN
elm-5391	74	27	,	,	PUNCT
elm-5391	74	28	ruihan	ruihan	PROPN
elm-5391	74	29	zhang	zhang	PROPN
elm-5391	74	30	and	and	CCONJ
elm-5391	74	31	xiajie	xiajie	PROPN
elm-5391	74	32	zhang	zhang	PROPN
elm-5391	74	33	:	:	PUNCT
elm-5391	74	34	representing	represent	VERB
elm-5391	74	35	affect	affect	VERB
elm-5391	74	36	information	information	NOUN
elm-5391	74	37	in	in	ADP
elm-5391	74	38	word	word	NOUN
elm-5391	74	39	embeddings	embedding	NOUN
elm-5391	74	40	.	.	PUNCT
elm-5391	75	1	312	312	NUM
elm-5391	75	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	75	3	https://www.elm-conference.net/	https://www.elm-conference.net/	PRON
elm-5391	75	4	as	as	ADP
elm-5391	75	5	goemotion	goemotion	NOUN
elm-5391	75	6	bert	bert	NOUN
elm-5391	75	7	in	in	ADP
elm-5391	75	8	the	the	DET
elm-5391	75	9	paper	paper	NOUN
elm-5391	75	10	.	.	PUNCT
elm-5391	76	1	•	•	NUM
elm-5391	76	2	given	give	VERB
elm-5391	76	3	that	that	SCONJ
elm-5391	76	4	the	the	DET
elm-5391	76	5	bert	bert	NOUN
elm-5391	76	6	-	-	PUNCT
elm-5391	76	7	based	base	VERB
elm-5391	76	8	word	word	NOUN
elm-5391	76	9	embeddings	embedding	NOUN
elm-5391	76	10	are	be	AUX
elm-5391	76	11	supposed	suppose	VERB
elm-5391	76	12	to	to	PART
elm-5391	76	13	be	be	AUX
elm-5391	76	14	contextualized	contextualize	VERB
elm-5391	76	15	because	because	SCONJ
elm-5391	76	16	they	they	PRON
elm-5391	76	17	could	could	AUX
elm-5391	76	18	be	be	AUX
elm-5391	76	19	functions	function	NOUN
elm-5391	76	20	of	of	ADP
elm-5391	76	21	the	the	DET
elm-5391	76	22	entire	entire	ADJ
elm-5391	76	23	input	input	NOUN
elm-5391	76	24	sentence	sentence	NOUN
elm-5391	76	25	(	(	PUNCT
elm-5391	76	26	ethayarajh	ethayarajh	PROPN
elm-5391	76	27	2019	2019	NUM
elm-5391	76	28	)	)	PUNCT
elm-5391	76	29	,	,	PUNCT
elm-5391	76	30	we	we	PRON
elm-5391	76	31	utilized	utilize	VERB
elm-5391	76	32	this	this	DET
elm-5391	76	33	contextual	contextual	ADJ
elm-5391	76	34	sensitivity	sensitivity	NOUN
elm-5391	76	35	feature	feature	NOUN
elm-5391	76	36	and	and	CCONJ
elm-5391	76	37	output	output	NOUN
elm-5391	76	38	contextualized	contextualize	VERB
elm-5391	76	39	bert	bert	NOUN
elm-5391	76	40	word	word	NOUN
elm-5391	76	41	embeddings	embedding	NOUN
elm-5391	76	42	by	by	ADP
elm-5391	76	43	deriving	derive	VERB
elm-5391	76	44	them	they	PRON
elm-5391	76	45	from	from	ADP
elm-5391	76	46	running	run	VERB
elm-5391	76	47	imdb	imdb	NOUN
elm-5391	76	48	movie	movie	NOUN
elm-5391	76	49	reviews	review	NOUN
elm-5391	76	50	(	(	PUNCT
elm-5391	76	51	maas	maas	PROPN
elm-5391	76	52	et	et	PROPN
elm-5391	76	53	al	al	PROPN
elm-5391	76	54	.	.	PROPN
elm-5391	76	55	2011	2011	NUM
elm-5391	76	56	)	)	PUNCT
elm-5391	76	57	.	.	PUNCT
elm-5391	77	1	for	for	ADP
elm-5391	77	2	each	each	DET
elm-5391	77	3	nrc	nrc	NOUN
elm-5391	77	4	word	word	NOUN
elm-5391	77	5	,	,	PUNCT
elm-5391	77	6	we	we	PRON
elm-5391	77	7	first	first	ADV
elm-5391	77	8	extracted	extract	VERB
elm-5391	77	9	all	all	DET
elm-5391	77	10	its	its	PRON
elm-5391	77	11	embeddings	embedding	NOUN
elm-5391	77	12	by	by	ADP
elm-5391	77	13	retrieving	retrieve	VERB
elm-5391	77	14	the	the	DET
elm-5391	77	15	last	last	ADJ
elm-5391	77	16	hidden	hide	VERB
elm-5391	77	17	layer	layer	NOUN
elm-5391	77	18	of	of	ADP
elm-5391	77	19	the	the	DET
elm-5391	77	20	bert	bert	PROPN
elm-5391	77	21	model	model	NOUN
elm-5391	77	22	,	,	PUNCT
elm-5391	77	23	we	we	PRON
elm-5391	77	24	then	then	ADV
elm-5391	77	25	computed	compute	VERB
elm-5391	77	26	the	the	DET
elm-5391	77	27	first	first	ADJ
elm-5391	77	28	principal	principal	ADJ
elm-5391	77	29	component	component	NOUN
elm-5391	77	30	to	to	PART
elm-5391	77	31	derive	derive	VERB
elm-5391	77	32	the	the	DET
elm-5391	77	33	single	single	ADJ
elm-5391	77	34	word	word	NOUN
elm-5391	77	35	vector	vector	NOUN
elm-5391	77	36	for	for	ADP
elm-5391	77	37	each	each	DET
elm-5391	77	38	individual	individual	ADJ
elm-5391	77	39	word	word	NOUN
elm-5391	77	40	.	.	PUNCT
elm-5391	78	1	3.3	3.3	NUM
elm-5391	78	2	.	.	PUNCT
elm-5391	79	1	unsupervised	unsupervised	ADJ
elm-5391	79	2	probe	probe	NOUN
elm-5391	79	3	:	:	PUNCT
elm-5391	79	4	principal	principal	ADJ
elm-5391	79	5	component	component	NOUN
elm-5391	79	6	analysis	analysis	NOUN
elm-5391	79	7	.	.	PUNCT
elm-5391	80	1	firstly	firstly	ADV
elm-5391	80	2	,	,	PUNCT
elm-5391	80	3	in	in	ADP
elm-5391	80	4	order	order	NOUN
elm-5391	80	5	to	to	PART
elm-5391	80	6	investigate	investigate	VERB
elm-5391	80	7	whether	whether	SCONJ
elm-5391	80	8	the	the	DET
elm-5391	80	9	types	type	NOUN
elm-5391	80	10	of	of	ADP
elm-5391	80	11	word	word	NOUN
elm-5391	80	12	embeddings	embedding	NOUN
elm-5391	80	13	under	under	ADP
elm-5391	80	14	investigation	investigation	NOUN
elm-5391	80	15	encode	encode	ADJ
elm-5391	80	16	affect	affect	VERB
elm-5391	80	17	information	information	NOUN
elm-5391	80	18	,	,	PUNCT
elm-5391	80	19	we	we	PRON
elm-5391	80	20	conducted	conduct	VERB
elm-5391	80	21	dimensionality	dimensionality	NOUN
elm-5391	80	22	reduction	reduction	NOUN
elm-5391	80	23	via	via	ADP
elm-5391	80	24	principal	principal	ADJ
elm-5391	80	25	component	component	NOUN
elm-5391	80	26	analysis	analysis	NOUN
elm-5391	80	27	(	(	PUNCT
elm-5391	80	28	pca	pca	NOUN
elm-5391	80	29	from	from	ADP
elm-5391	80	30	the	the	DET
elm-5391	80	31	sklearn	sklearn	ADJ
elm-5391	80	32	package	package	NOUN
elm-5391	80	33	(	(	PUNCT
elm-5391	80	34	buitinck	buitinck	ADV
elm-5391	80	35	et	et	PROPN
elm-5391	80	36	al	al	PROPN
elm-5391	80	37	.	.	PROPN
elm-5391	80	38	2013	2013	NUM
elm-5391	80	39	,	,	PUNCT
elm-5391	80	40	pedregosa	pedregosa	PROPN
elm-5391	80	41	et	et	PROPN
elm-5391	80	42	al	al	PROPN
elm-5391	80	43	.	.	PROPN
elm-5391	80	44	2011	2011	NUM
elm-5391	80	45	)	)	PUNCT
elm-5391	80	46	)	)	PUNCT
elm-5391	80	47	over	over	ADP
elm-5391	80	48	each	each	PRON
elm-5391	80	49	of	of	ADP
elm-5391	80	50	the	the	DET
elm-5391	80	51	studied	study	VERB
elm-5391	80	52	word	word	NOUN
elm-5391	80	53	embeddings	embedding	NOUN
elm-5391	80	54	.	.	PUNCT
elm-5391	81	1	pca	pca	PROPN
elm-5391	81	2	is	be	AUX
elm-5391	81	3	a	a	DET
elm-5391	81	4	convenient	convenient	ADJ
elm-5391	81	5	tool	tool	NOUN
elm-5391	81	6	for	for	ADP
elm-5391	81	7	such	such	ADJ
elm-5391	81	8	visualization	visualization	NOUN
elm-5391	81	9	without	without	ADP
elm-5391	81	10	us	we	PRON
elm-5391	81	11	going	go	VERB
elm-5391	81	12	deep	deep	ADV
elm-5391	81	13	into	into	ADP
elm-5391	81	14	deciphering	decipher	VERB
elm-5391	81	15	the	the	DET
elm-5391	81	16	architecture	architecture	NOUN
elm-5391	81	17	of	of	ADP
elm-5391	81	18	multi	multi	ADJ
elm-5391	81	19	-	-	ADJ
elm-5391	81	20	dimensional	dimensional	ADJ
elm-5391	81	21	word	word	NOUN
elm-5391	81	22	embeddings	embedding	NOUN
elm-5391	81	23	.	.	PUNCT
elm-5391	82	1	specifically	specifically	ADV
elm-5391	82	2	,	,	PUNCT
elm-5391	82	3	we	we	PRON
elm-5391	82	4	sought	seek	VERB
elm-5391	82	5	for	for	ADP
elm-5391	82	6	the	the	DET
elm-5391	82	7	correlation	correlation	NOUN
elm-5391	82	8	between	between	ADP
elm-5391	82	9	each	each	DET
elm-5391	82	10	principal	principal	ADJ
elm-5391	82	11	component	component	NOUN
elm-5391	82	12	from	from	ADP
elm-5391	82	13	the	the	DET
elm-5391	82	14	embeddings	embedding	NOUN
elm-5391	82	15	and	and	CCONJ
elm-5391	82	16	the	the	DET
elm-5391	82	17	human	human	ADJ
elm-5391	82	18	ratings	rating	NOUN
elm-5391	82	19	of	of	ADP
elm-5391	82	20	each	each	DET
elm-5391	82	21	vad	vad	PROPN
elm-5391	82	22	dimension	dimension	NOUN
elm-5391	82	23	.	.	PUNCT
elm-5391	83	1	higher	high	ADJ
elm-5391	83	2	correlation	correlation	NOUN
elm-5391	83	3	indicates	indicate	VERB
elm-5391	83	4	that	that	SCONJ
elm-5391	83	5	the	the	DET
elm-5391	83	6	high	high	ADV
elm-5391	83	7	-	-	PUNCT
elm-5391	83	8	dimensional	dimensional	ADJ
elm-5391	83	9	information	information	NOUN
elm-5391	83	10	in	in	ADP
elm-5391	83	11	that	that	DET
elm-5391	83	12	word	word	NOUN
elm-5391	83	13	embedding	embed	VERB
elm-5391	83	14	is	be	AUX
elm-5391	83	15	more	more	ADV
elm-5391	83	16	likely	likely	ADJ
elm-5391	83	17	to	to	PART
elm-5391	83	18	saliently	saliently	ADV
elm-5391	83	19	encode	encode	VERB
elm-5391	83	20	affect	affect	NOUN
elm-5391	83	21	information	information	NOUN
elm-5391	83	22	.	.	PUNCT
elm-5391	84	1	this	this	DET
elm-5391	84	2	unsupervised	unsupervised	ADJ
elm-5391	84	3	approach	approach	NOUN
elm-5391	84	4	provides	provide	VERB
elm-5391	84	5	a	a	DET
elm-5391	84	6	rudimentary	rudimentary	ADJ
elm-5391	84	7	insight	insight	NOUN
elm-5391	84	8	into	into	ADP
elm-5391	84	9	whether	whether	SCONJ
elm-5391	84	10	affect	affect	NOUN
elm-5391	84	11	meanings	meaning	NOUN
elm-5391	84	12	are	be	AUX
elm-5391	84	13	well	well	ADV
elm-5391	84	14	captured	capture	VERB
elm-5391	84	15	by	by	ADP
elm-5391	84	16	the	the	DET
elm-5391	84	17	word	word	NOUN
elm-5391	84	18	embeddings	embedding	NOUN
elm-5391	84	19	being	be	AUX
elm-5391	84	20	investigated	investigate	VERB
elm-5391	84	21	.	.	PUNCT
elm-5391	85	1	3.4	3.4	NUM
elm-5391	85	2	.	.	PUNCT
elm-5391	85	3	unsupervised	unsupervised	ADJ
elm-5391	85	4	probe	probe	NOUN
elm-5391	85	5	:	:	PUNCT
elm-5391	85	6	cosine	cosine	NOUN
elm-5391	85	7	distance	distance	NOUN
elm-5391	85	8	for	for	ADP
elm-5391	85	9	semantic	semantic	ADJ
elm-5391	85	10	similarity	similarity	NOUN
elm-5391	85	11	.	.	PUNCT
elm-5391	86	1	it	it	PRON
elm-5391	86	2	has	have	AUX
elm-5391	86	3	been	be	AUX
elm-5391	86	4	shown	show	VERB
elm-5391	86	5	that	that	SCONJ
elm-5391	86	6	cosine	cosine	NOUN
elm-5391	86	7	distances	distance	NOUN
elm-5391	86	8	between	between	ADP
elm-5391	86	9	vector	vector	NOUN
elm-5391	86	10	representations	representation	NOUN
elm-5391	86	11	of	of	ADP
elm-5391	86	12	words	word	NOUN
elm-5391	86	13	indicate	indicate	VERB
elm-5391	86	14	the	the	DET
elm-5391	86	15	words	word	NOUN
elm-5391	86	16	’	'	PUNCT
elm-5391	86	17	semantic	semantic	ADJ
elm-5391	86	18	relatedness	relatedness	NOUN
elm-5391	86	19	(	(	PUNCT
elm-5391	86	20	dumais	dumais	NOUN
elm-5391	86	21	et	et	PROPN
elm-5391	86	22	al	al	PROPN
elm-5391	86	23	.	.	PROPN
elm-5391	86	24	1988	1988	NUM
elm-5391	86	25	,	,	PUNCT
elm-5391	86	26	mikolov	mikolov	PROPN
elm-5391	86	27	et	et	PROPN
elm-5391	86	28	al	al	PROPN
elm-5391	86	29	.	.	PROPN
elm-5391	86	30	2013	2013	NUM
elm-5391	86	31	,	,	PUNCT
elm-5391	86	32	bojanowski	bojanowski	NOUN
elm-5391	86	33	et	et	PROPN
elm-5391	86	34	al	al	PROPN
elm-5391	86	35	.	.	PROPN
elm-5391	86	36	2017	2017	NUM
elm-5391	86	37	)	)	PUNCT
elm-5391	86	38	.	.	PUNCT
elm-5391	87	1	the	the	DET
elm-5391	87	2	literature	literature	NOUN
elm-5391	87	3	we	we	PRON
elm-5391	87	4	learned	learn	VERB
elm-5391	87	5	from	from	ADP
elm-5391	87	6	is	be	AUX
elm-5391	87	7	nair	nair	PROPN
elm-5391	87	8	et	et	PROPN
elm-5391	87	9	al	al	PROPN
elm-5391	87	10	.	.	PROPN
elm-5391	88	1	(	(	PUNCT
elm-5391	88	2	2020	2020	NUM
elm-5391	88	3	)	)	PUNCT
elm-5391	88	4	.	.	PUNCT
elm-5391	89	1	they	they	PRON
elm-5391	89	2	investigated	investigate	VERB
elm-5391	89	3	whether	whether	SCONJ
elm-5391	89	4	contextualized	contextualize	VERB
elm-5391	89	5	word	word	NOUN
elm-5391	89	6	embeddings	embedding	NOUN
elm-5391	89	7	–	–	PUNCT
elm-5391	89	8	bert	bert	NOUN
elm-5391	89	9	embeddings	embedding	NOUN
elm-5391	89	10	in	in	ADP
elm-5391	89	11	this	this	DET
elm-5391	89	12	case	case	NOUN
elm-5391	89	13	–	–	PUNCT
elm-5391	89	14	capture	capture	VERB
elm-5391	89	15	human	human	ADJ
elm-5391	89	16	-	-	PUNCT
elm-5391	89	17	like	like	ADJ
elm-5391	89	18	distinctions	distinction	NOUN
elm-5391	89	19	between	between	ADP
elm-5391	89	20	english	english	ADJ
elm-5391	89	21	word	word	NOUN
elm-5391	89	22	senses	sense	NOUN
elm-5391	89	23	,	,	PUNCT
elm-5391	89	24	such	such	ADJ
elm-5391	89	25	as	as	ADP
elm-5391	89	26	polysemy	polysemy	NOUN
elm-5391	89	27	(	(	PUNCT
elm-5391	89	28	chicken	chicken	NOUN
elm-5391	89	29	as	as	ADP
elm-5391	89	30	animal	animal	NOUN
elm-5391	89	31	vs.	vs.	ADP
elm-5391	89	32	chicken	chicken	NOUN
elm-5391	89	33	as	as	ADP
elm-5391	89	34	meat	meat	NOUN
elm-5391	89	35	)	)	PUNCT
elm-5391	89	36	and	and	CCONJ
elm-5391	89	37	homonymy	homonymy	NOUN
elm-5391	89	38	(	(	PUNCT
elm-5391	89	39	bat	bat	NOUN
elm-5391	89	40	as	as	ADP
elm-5391	89	41	mammal	mammal	ADJ
elm-5391	89	42	vs.	vs.	ADP
elm-5391	89	43	bat	bat	NOUN
elm-5391	89	44	as	as	ADP
elm-5391	89	45	sports	sport	NOUN
elm-5391	89	46	equipment	equipment	NOUN
elm-5391	89	47	)	)	PUNCT
elm-5391	89	48	.	.	PUNCT
elm-5391	90	1	for	for	ADP
elm-5391	90	2	each	each	DET
elm-5391	90	3	pair	pair	NOUN
elm-5391	90	4	of	of	ADP
elm-5391	90	5	target	target	NOUN
elm-5391	90	6	word	word	NOUN
elm-5391	90	7	senses	sense	NOUN
elm-5391	90	8	,	,	PUNCT
elm-5391	90	9	they	they	PRON
elm-5391	90	10	measured	measure	VERB
elm-5391	90	11	(	(	PUNCT
elm-5391	90	12	1	1	X
elm-5391	90	13	)	)	PUNCT
elm-5391	90	14	the	the	DET
elm-5391	90	15	cosine	cosine	ADJ
elm-5391	90	16	distance	distance	NOUN
elm-5391	90	17	of	of	ADP
elm-5391	90	18	the	the	DET
elm-5391	90	19	two	two	NUM
elm-5391	90	20	embeddings	embedding	NOUN
elm-5391	90	21	and	and	CCONJ
elm-5391	90	22	(	(	PUNCT
elm-5391	90	23	2	2	X
elm-5391	90	24	)	)	PUNCT
elm-5391	90	25	the	the	DET
elm-5391	90	26	human	human	ADJ
elm-5391	90	27	judgment	judgment	NOUN
elm-5391	90	28	of	of	ADP
elm-5391	90	29	the	the	DET
elm-5391	90	30	relatedness	relatedness	NOUN
elm-5391	90	31	of	of	ADP
elm-5391	90	32	the	the	DET
elm-5391	90	33	word	word	NOUN
elm-5391	90	34	senses	sense	NOUN
elm-5391	90	35	in	in	ADP
elm-5391	90	36	a	a	DET
elm-5391	90	37	2	2	NUM
elm-5391	90	38	-	-	PUNCT
elm-5391	90	39	dimensional	dimensional	ADJ
elm-5391	90	40	spatial	spatial	ADJ
elm-5391	90	41	arrangement	arrangement	NOUN
elm-5391	90	42	task	task	NOUN
elm-5391	90	43	.	.	PUNCT
elm-5391	91	1	they	they	PRON
elm-5391	91	2	then	then	ADV
elm-5391	91	3	applied	apply	VERB
elm-5391	91	4	spearman	spearman	NOUN
elm-5391	91	5	rank	rank	NOUN
elm-5391	91	6	correlation	correlation	NOUN
elm-5391	91	7	analysis	analysis	NOUN
elm-5391	91	8	to	to	ADP
elm-5391	91	9	these	these	DET
elm-5391	91	10	two	two	NUM
elm-5391	91	11	measurements	measurement	NOUN
elm-5391	91	12	.	.	PUNCT
elm-5391	92	1	the	the	DET
elm-5391	92	2	results	result	NOUN
elm-5391	92	3	turn	turn	VERB
elm-5391	92	4	out	out	ADP
elm-5391	92	5	that	that	SCONJ
elm-5391	92	6	the	the	DET
elm-5391	92	7	distance	distance	NOUN
elm-5391	92	8	of	of	ADP
elm-5391	92	9	word	word	NOUN
elm-5391	92	10	senses	sense	NOUN
elm-5391	92	11	in	in	ADP
elm-5391	92	12	bert	bert	PROPN
elm-5391	92	13	’s	’s	PART
elm-5391	92	14	embedding	embed	VERB
elm-5391	92	15	space	space	NOUN
elm-5391	92	16	correlated	correlate	VERB
elm-5391	92	17	with	with	ADP
elm-5391	92	18	human	human	ADJ
elm-5391	92	19	judgments	judgment	NOUN
elm-5391	92	20	and	and	CCONJ
elm-5391	92	21	that	that	SCONJ
elm-5391	92	22	the	the	DET
elm-5391	92	23	correlation	correlation	NOUN
elm-5391	92	24	of	of	ADP
elm-5391	92	25	homonyms	homonym	NOUN
elm-5391	92	26	was	be	AUX
elm-5391	92	27	higher	high	ADJ
elm-5391	92	28	.	.	PUNCT
elm-5391	93	1	learning	learn	VERB
elm-5391	93	2	from	from	ADP
elm-5391	93	3	nair	nair	PROPN
elm-5391	93	4	et	et	PROPN
elm-5391	93	5	al	al	PROPN
elm-5391	93	6	.	.	PROPN
elm-5391	94	1	(	(	PUNCT
elm-5391	94	2	2020	2020	NUM
elm-5391	94	3	)	)	PUNCT
elm-5391	94	4	,	,	PUNCT
elm-5391	94	5	we	we	PRON
elm-5391	94	6	took	take	VERB
elm-5391	94	7	a	a	DET
elm-5391	94	8	small	small	ADJ
elm-5391	94	9	sample	sample	NOUN
elm-5391	94	10	of	of	ADP
elm-5391	94	11	affect	affect	VERB
elm-5391	94	12	words	word	NOUN
elm-5391	94	13	and	and	CCONJ
elm-5391	94	14	ran	run	VERB
elm-5391	94	15	correlation	correlation	NOUN
elm-5391	94	16	tests	test	NOUN
elm-5391	94	17	to	to	PART
elm-5391	94	18	see	see	VERB
elm-5391	94	19	the	the	DET
elm-5391	94	20	relationship	relationship	NOUN
elm-5391	94	21	between	between	ADP
elm-5391	94	22	the	the	DET
elm-5391	94	23	word	word	NOUN
elm-5391	94	24	embedding	embed	VERB
elm-5391	94	25	space	space	NOUN
elm-5391	94	26	and	and	CCONJ
elm-5391	94	27	the	the	DET
elm-5391	94	28	vector	vector	NOUN
elm-5391	94	29	space	space	NOUN
elm-5391	94	30	of	of	ADP
elm-5391	94	31	human	human	ADJ
elm-5391	94	32	judgments	judgment	NOUN
elm-5391	94	33	(	(	PUNCT
elm-5391	94	34	i.e.	i.e.	X
elm-5391	94	35	,	,	PUNCT
elm-5391	94	36	the	the	DET
elm-5391	94	37	vad	vad	PROPN
elm-5391	94	38	3	3	NUM
elm-5391	94	39	-	-	PUNCT
elm-5391	94	40	dimensional	dimensional	ADJ
elm-5391	94	41	space	space	NOUN
elm-5391	94	42	)	)	PUNCT
elm-5391	94	43	.	.	PUNCT
elm-5391	95	1	similar	similar	ADJ
elm-5391	95	2	to	to	ADP
elm-5391	95	3	the	the	DET
elm-5391	95	4	pca	pca	NOUN
elm-5391	95	5	results	result	NOUN
elm-5391	95	6	,	,	PUNCT
elm-5391	95	7	a	a	DET
elm-5391	95	8	stronger	strong	ADJ
elm-5391	95	9	correlation	correlation	NOUN
elm-5391	95	10	means	mean	VERB
elm-5391	95	11	a	a	DET
elm-5391	95	12	more	more	ADV
elm-5391	95	13	salient	salient	NOUN
elm-5391	95	14	encoding	encoding	NOUN
elm-5391	95	15	of	of	ADP
elm-5391	95	16	real	real	ADJ
elm-5391	95	17	-	-	PUNCT
elm-5391	95	18	world	world	NOUN
elm-5391	95	19	affect	affect	VERB
elm-5391	95	20	information	information	NOUN
elm-5391	95	21	in	in	ADP
elm-5391	95	22	the	the	DET
elm-5391	95	23	investigated	investigate	VERB
elm-5391	95	24	word	word	NOUN
elm-5391	95	25	embeddings	embedding	NOUN
elm-5391	95	26	.	.	PUNCT
elm-5391	96	1	more	more	ADJ
elm-5391	96	2	about	about	ADP
elm-5391	96	3	the	the	DET
elm-5391	96	4	sampling	sample	VERB
elm-5391	96	5	method	method	NOUN
elm-5391	96	6	:	:	PUNCT
elm-5391	96	7	shaver	shaver	PROPN
elm-5391	96	8	et	et	PROPN
elm-5391	96	9	al	al	PROPN
elm-5391	96	10	.	.	PROPN
elm-5391	97	1	(	(	PUNCT
elm-5391	97	2	1987	1987	NUM
elm-5391	97	3	)	)	PUNCT
elm-5391	97	4	defined	define	VERB
elm-5391	97	5	six	six	NUM
elm-5391	97	6	basic	basic	ADJ
elm-5391	97	7	emotion	emotion	NOUN
elm-5391	97	8	categories	category	NOUN
elm-5391	97	9	(	(	PUNCT
elm-5391	97	10	i.e.	i.e.	X
elm-5391	97	11	,	,	PUNCT
elm-5391	97	12	love	love	NOUN
elm-5391	97	13	,	,	PUNCT
elm-5391	97	14	surprise	surprise	NOUN
elm-5391	97	15	,	,	PUNCT
elm-5391	97	16	sadness	sadness	NOUN
elm-5391	97	17	,	,	PUNCT
elm-5391	97	18	anger	anger	NOUN
elm-5391	97	19	,	,	PUNCT
elm-5391	97	20	joy	joy	NOUN
elm-5391	97	21	,	,	PUNCT
elm-5391	97	22	fear	fear	NOUN
elm-5391	97	23	)	)	PUNCT
elm-5391	97	24	and	and	CCONJ
elm-5391	97	25	a	a	DET
elm-5391	97	26	list	list	NOUN
elm-5391	97	27	of	of	ADP
elm-5391	97	28	affect	affect	VERB
elm-5391	97	29	words	word	NOUN
elm-5391	97	30	in	in	ADP
elm-5391	97	31	each	each	DET
elm-5391	97	32	category	category	NOUN
elm-5391	97	33	,	,	PUNCT
elm-5391	97	34	totalling	total	VERB
elm-5391	97	35	132	132	NUM
elm-5391	97	36	affect	affect	NOUN
elm-5391	97	37	words	word	NOUN
elm-5391	97	38	.	.	PUNCT
elm-5391	98	1	we	we	PRON
elm-5391	98	2	took	take	VERB
elm-5391	98	3	80	80	NUM
elm-5391	98	4	affect	affect	NOUN
elm-5391	98	5	words	word	NOUN
elm-5391	98	6	from	from	ADP
elm-5391	98	7	shaver	shaver	PROPN
elm-5391	98	8	et	et	PROPN
elm-5391	98	9	al	al	PROPN
elm-5391	98	10	.	.	PROPN
elm-5391	99	1	(	(	PUNCT
elm-5391	99	2	1987	1987	NUM
elm-5391	99	3	)	)	PUNCT
elm-5391	99	4	(	(	PUNCT
elm-5391	99	5	e.g.	e.g.	ADV
elm-5391	99	6	,	,	PUNCT
elm-5391	99	7	disgust	disgust	NOUN
elm-5391	99	8	,	,	PUNCT
elm-5391	99	9	envy	envy	NOUN
elm-5391	99	10	,	,	PUNCT
elm-5391	99	11	enjoyment	enjoyment	NOUN
elm-5391	99	12	,	,	PUNCT
elm-5391	99	13	desire	desire	NOUN
elm-5391	99	14	)	)	PUNCT
elm-5391	99	15	and	and	CCONJ
elm-5391	99	16	calculated	calculate	VERB
elm-5391	99	17	the	the	DET
elm-5391	99	18	cosine	cosine	NOUN
elm-5391	99	19	similarity	similarity	NOUN
elm-5391	99	20	between	between	ADP
elm-5391	99	21	each	each	PRON
elm-5391	99	22	and	and	CCONJ
elm-5391	99	23	another	another	DET
elm-5391	99	24	word	word	NOUN
elm-5391	99	25	,	,	PUNCT
elm-5391	99	26	resulting	result	VERB
elm-5391	99	27	in	in	ADP
elm-5391	99	28	80	80	NUM
elm-5391	99	29	×	×	NOUN
elm-5391	99	30	79	79	NUM
elm-5391	99	31	similarity	similarity	NOUN
elm-5391	99	32	scores	score	NOUN
elm-5391	99	33	.	.	PUNCT
elm-5391	100	1	we	we	PRON
elm-5391	100	2	did	do	VERB
elm-5391	100	3	the	the	DET
elm-5391	100	4	same	same	ADJ
elm-5391	100	5	thing	thing	NOUN
elm-5391	100	6	by	by	ADP
elm-5391	100	7	iterating	iterate	VERB
elm-5391	100	8	over	over	ADP
elm-5391	100	9	the	the	DET
elm-5391	100	10	kind	kind	NOUN
elm-5391	100	11	of	of	ADP
elm-5391	100	12	vector	vector	NOUN
elm-5391	100	13	space	space	NOUN
elm-5391	100	14	,	,	PUNCT
elm-5391	100	15	from	from	ADP
elm-5391	100	16	the	the	DET
elm-5391	100	17	nrc	nrc	PROPN
elm-5391	100	18	vad	vad	NOUN
elm-5391	100	19	3	3	NUM
elm-5391	100	20	-	-	PUNCT
elm-5391	100	21	dimensional	dimensional	ADJ
elm-5391	100	22	human	human	ADJ
elm-5391	100	23	judgment	judgment	NOUN
elm-5391	100	24	space	space	NOUN
elm-5391	100	25	,	,	PUNCT
elm-5391	100	26	to	to	ADP
elm-5391	100	27	the	the	DET
elm-5391	100	28	four	four	NUM
elm-5391	100	29	kinds	kind	NOUN
elm-5391	100	30	of	of	ADP
elm-5391	100	31	word	word	NOUN
elm-5391	100	32	embeddings	embedding	NOUN
elm-5391	100	33	introduced	introduce	VERB
elm-5391	100	34	in	in	ADP
elm-5391	100	35	section	section	NOUN
elm-5391	100	36	3.2	3.2	NUM
elm-5391	100	37	,	,	PUNCT
elm-5391	100	38	resulting	result	VERB
elm-5391	100	39	in	in	ADP
elm-5391	100	40	80	80	NUM
elm-5391	100	41	×	×	NOUN
elm-5391	100	42	79	79	NUM
elm-5391	100	43	×	×	NOUN
elm-5391	100	44	5	5	NUM
elm-5391	100	45	pairwise	pairwise	NOUN
elm-5391	100	46	similarity	similarity	NOUN
elm-5391	100	47	scores	score	NOUN
elm-5391	100	48	.	.	PUNCT
elm-5391	101	1	then	then	ADV
elm-5391	101	2	,	,	PUNCT
elm-5391	101	3	we	we	PRON
elm-5391	101	4	did	do	VERB
elm-5391	101	5	spearman	spearman	NOUN
elm-5391	101	6	rank	rank	NOUN
elm-5391	101	7	correlation	correlation	NOUN
elm-5391	101	8	analysis	analysis	NOUN
elm-5391	101	9	between	between	ADP
elm-5391	101	10	one	one	NUM
elm-5391	101	11	and	and	CCONJ
elm-5391	101	12	another	another	PRON
elm-5391	101	13	embedding	embed	VERB
elm-5391	101	14	representations	representation	NOUN
elm-5391	101	15	.	.	PUNCT
elm-5391	102	1	we	we	PRON
elm-5391	102	2	adopted	adopt	VERB
elm-5391	102	3	a	a	DET
elm-5391	102	4	nonparametric	nonparametric	NOUN
elm-5391	102	5	correlation	correlation	NOUN
elm-5391	102	6	like	like	ADP
elm-5391	102	7	nair	nair	PROPN
elm-5391	102	8	et	et	PROPN
elm-5391	102	9	al	al	PROPN
elm-5391	102	10	.	.	PROPN
elm-5391	103	1	(	(	PUNCT
elm-5391	103	2	2020	2020	NUM
elm-5391	103	3	)	)	PUNCT
elm-5391	103	4	because	because	SCONJ
elm-5391	103	5	,	,	PUNCT
elm-5391	103	6	first	first	ADV
elm-5391	103	7	,	,	PUNCT
elm-5391	103	8	the	the	DET
elm-5391	103	9	numerical	numerical	ADJ
elm-5391	103	10	distribution	distribution	NOUN
elm-5391	103	11	of	of	ADP
elm-5391	103	12	a	a	DET
elm-5391	103	13	given	give	VERB
elm-5391	103	14	type	type	NOUN
elm-5391	103	15	proceedings	proceeding	NOUN
elm-5391	103	16	of	of	ADP
elm-5391	103	17	elm	elm	NOUN
elm-5391	103	18	2	2	NUM
elm-5391	103	19	:	:	PUNCT
elm-5391	103	20	310	310	NUM
elm-5391	103	21	-	-	SYM
elm-5391	103	22	321	321	NUM
elm-5391	103	23	,	,	PUNCT
elm-5391	103	24	2023	2023	NUM
elm-5391	103	25	yuhan	yuhan	PROPN
elm-5391	103	26	zhang	zhang	PROPN
elm-5391	103	27	,	,	PUNCT
elm-5391	103	28	wenqi	wenqi	PROPN
elm-5391	103	29	chen	chen	PROPN
elm-5391	103	30	,	,	PUNCT
elm-5391	103	31	ruihan	ruihan	PROPN
elm-5391	103	32	zhang	zhang	PROPN
elm-5391	103	33	and	and	CCONJ
elm-5391	103	34	xiajie	xiajie	PROPN
elm-5391	103	35	zhang	zhang	PROPN
elm-5391	103	36	:	:	PUNCT
elm-5391	103	37	representing	represent	VERB
elm-5391	103	38	affect	affect	VERB
elm-5391	103	39	information	information	NOUN
elm-5391	103	40	in	in	ADP
elm-5391	103	41	word	word	NOUN
elm-5391	103	42	embeddings	embedding	NOUN
elm-5391	103	43	.	.	PUNCT
elm-5391	104	1	313	313	NUM
elm-5391	104	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	104	3	https://www.elm-conference.net/	https://www.elm-conference.net/	PRON
elm-5391	104	4	of	of	ADP
elm-5391	104	5	word	word	NOUN
elm-5391	104	6	embedding	embed	VERB
elm-5391	104	7	does	do	AUX
elm-5391	104	8	not	not	PART
elm-5391	104	9	satisfy	satisfy	VERB
elm-5391	104	10	the	the	DET
elm-5391	104	11	normality	normality	NOUN
elm-5391	104	12	assumption	assumption	NOUN
elm-5391	104	13	and	and	CCONJ
elm-5391	104	14	,	,	PUNCT
elm-5391	104	15	second	second	ADJ
elm-5391	104	16	,	,	PUNCT
elm-5391	104	17	we	we	PRON
elm-5391	104	18	cared	care	VERB
elm-5391	104	19	more	more	ADJ
elm-5391	104	20	about	about	ADP
elm-5391	104	21	whether	whether	SCONJ
elm-5391	104	22	the	the	DET
elm-5391	104	23	two	two	NUM
elm-5391	104	24	embedding	embed	VERB
elm-5391	104	25	representations	representation	NOUN
elm-5391	104	26	share	share	VERB
elm-5391	104	27	a	a	DET
elm-5391	104	28	monotonic	monotonic	ADJ
elm-5391	104	29	relationship	relationship	NOUN
elm-5391	104	30	,	,	PUNCT
elm-5391	104	31	instead	instead	ADV
elm-5391	104	32	of	of	ADP
elm-5391	104	33	a	a	DET
elm-5391	104	34	strictly	strictly	ADV
elm-5391	104	35	linear	linear	ADJ
elm-5391	104	36	relationship	relationship	NOUN
elm-5391	104	37	.	.	PUNCT
elm-5391	105	1	3.5	3.5	NUM
elm-5391	105	2	.	.	PUNCT
elm-5391	106	1	supervised	supervise	VERB
elm-5391	106	2	probe	probe	NOUN
elm-5391	106	3	:	:	PUNCT
elm-5391	106	4	linear	linear	PROPN
elm-5391	106	5	classifier	classifier	NOUN
elm-5391	106	6	.	.	PUNCT
elm-5391	107	1	in	in	ADP
elm-5391	107	2	addition	addition	NOUN
elm-5391	107	3	,	,	PUNCT
elm-5391	107	4	we	we	PRON
elm-5391	107	5	experimented	experiment	VERB
elm-5391	107	6	with	with	ADP
elm-5391	107	7	supervised	supervised	ADJ
elm-5391	107	8	learning	learning	NOUN
elm-5391	107	9	methods	method	NOUN
elm-5391	107	10	to	to	PART
elm-5391	107	11	probe	probe	VERB
elm-5391	107	12	whether	whether	SCONJ
elm-5391	107	13	the	the	DET
elm-5391	107	14	word	word	NOUN
elm-5391	107	15	embeddings	embedding	NOUN
elm-5391	107	16	can	can	AUX
elm-5391	107	17	be	be	AUX
elm-5391	107	18	linearly	linearly	ADV
elm-5391	107	19	separable	separable	ADJ
elm-5391	107	20	into	into	ADP
elm-5391	107	21	the	the	DET
elm-5391	107	22	vad	vad	PROPN
elm-5391	107	23	dimensions	dimensions	PROPN
elm-5391	107	24	.	.	PUNCT
elm-5391	108	1	we	we	PRON
elm-5391	108	2	trained	train	VERB
elm-5391	108	3	a	a	DET
elm-5391	108	4	one	one	NUM
elm-5391	108	5	-	-	PUNCT
elm-5391	108	6	layer	layer	NOUN
elm-5391	108	7	neural	neural	ADJ
elm-5391	108	8	network	network	NOUN
elm-5391	108	9	(	(	PUNCT
elm-5391	108	10	i.e.	i.e.	X
elm-5391	108	11	,	,	PUNCT
elm-5391	108	12	a	a	DET
elm-5391	108	13	logistic	logistic	ADJ
elm-5391	108	14	regression	regression	NOUN
elm-5391	108	15	model	model	NOUN
elm-5391	108	16	)	)	PUNCT
elm-5391	108	17	to	to	PART
elm-5391	108	18	predict	predict	VERB
elm-5391	108	19	the	the	DET
elm-5391	108	20	binary	binary	PROPN
elm-5391	108	21	vad	vad	PROPN
elm-5391	108	22	labels3	labels3	PROPN
elm-5391	108	23	with	with	ADP
elm-5391	108	24	the	the	DET
elm-5391	108	25	embeddings	embedding	NOUN
elm-5391	108	26	as	as	ADP
elm-5391	108	27	inputs	input	NOUN
elm-5391	108	28	.	.	PUNCT
elm-5391	109	1	the	the	DET
elm-5391	109	2	linear	linear	ADJ
elm-5391	109	3	classifier	classifier	NOUN
elm-5391	109	4	was	be	AUX
elm-5391	109	5	designed	design	VERB
elm-5391	109	6	to	to	PART
elm-5391	109	7	be	be	AUX
elm-5391	109	8	as	as	ADV
elm-5391	109	9	simple	simple	ADJ
elm-5391	109	10	as	as	ADP
elm-5391	109	11	possible	possible	ADJ
elm-5391	109	12	to	to	PART
elm-5391	109	13	reveal	reveal	VERB
elm-5391	109	14	information	information	NOUN
elm-5391	109	15	encoded	encode	VERB
elm-5391	109	16	in	in	ADP
elm-5391	109	17	the	the	DET
elm-5391	109	18	embeddings	embedding	NOUN
elm-5391	109	19	.	.	PUNCT
elm-5391	110	1	if	if	SCONJ
elm-5391	110	2	the	the	DET
elm-5391	110	3	prediction	prediction	NOUN
elm-5391	110	4	results	result	VERB
elm-5391	110	5	match	match	VERB
elm-5391	110	6	well	well	ADV
elm-5391	110	7	with	with	ADP
elm-5391	110	8	the	the	DET
elm-5391	110	9	human	human	ADJ
elm-5391	110	10	criterion	criterion	NOUN
elm-5391	110	11	,	,	PUNCT
elm-5391	110	12	it	it	PRON
elm-5391	110	13	is	be	AUX
elm-5391	110	14	reasonable	reasonable	ADJ
elm-5391	110	15	to	to	PART
elm-5391	110	16	conclude	conclude	VERB
elm-5391	110	17	that	that	SCONJ
elm-5391	110	18	the	the	DET
elm-5391	110	19	word	word	NOUN
elm-5391	110	20	embeddings	embedding	NOUN
elm-5391	110	21	have	have	VERB
elm-5391	110	22	the	the	DET
elm-5391	110	23	capability	capability	NOUN
elm-5391	110	24	to	to	PART
elm-5391	110	25	represent	represent	VERB
elm-5391	110	26	the	the	DET
elm-5391	110	27	affective	affective	ADJ
elm-5391	110	28	information	information	NOUN
elm-5391	110	29	.	.	PUNCT
elm-5391	111	1	4	4	X
elm-5391	111	2	.	.	X
elm-5391	111	3	analyses	analysis	NOUN
elm-5391	111	4	&	&	CCONJ
elm-5391	111	5	results	result	NOUN
elm-5391	111	6	.	.	PUNCT
elm-5391	112	1	4.1	4.1	NUM
elm-5391	112	2	.	.	PUNCT
elm-5391	112	3	principal	principal	ADJ
elm-5391	112	4	component	component	NOUN
elm-5391	112	5	analysis	analysis	NOUN
elm-5391	112	6	.	.	PUNCT
elm-5391	113	1	figure	figure	NOUN
elm-5391	113	2	1	1	NUM
elm-5391	113	3	shows	show	VERB
elm-5391	113	4	the	the	DET
elm-5391	113	5	pca	pca	PROPN
elm-5391	113	6	result	result	NOUN
elm-5391	113	7	.	.	PUNCT
elm-5391	114	1	we	we	PRON
elm-5391	114	2	extracted	extract	VERB
elm-5391	114	3	5,586	5,586	NUM
elm-5391	114	4	words	word	NOUN
elm-5391	114	5	that	that	PRON
elm-5391	114	6	are	be	AUX
elm-5391	114	7	in	in	ADP
elm-5391	114	8	the	the	DET
elm-5391	114	9	vocabulary	vocabulary	NOUN
elm-5391	114	10	of	of	ADP
elm-5391	114	11	glove	glove	NOUN
elm-5391	114	12	,	,	PUNCT
elm-5391	114	13	base	base	NOUN
elm-5391	114	14	bert	bert	PROPN
elm-5391	114	15	,	,	PUNCT
elm-5391	114	16	goemotion	goemotion	NOUN
elm-5391	114	17	bert	bert	PROPN
elm-5391	114	18	,	,	PUNCT
elm-5391	114	19	and	and	CCONJ
elm-5391	114	20	contextualized	contextualized	ADJ
elm-5391	114	21	bert	bert	NOUN
elm-5391	114	22	embeddings	embedding	NOUN
elm-5391	114	23	,	,	PUNCT
elm-5391	114	24	ran	run	VERB
elm-5391	114	25	pca	pca	NOUN
elm-5391	114	26	over	over	ADP
elm-5391	114	27	these	these	DET
elm-5391	114	28	word	word	NOUN
elm-5391	114	29	embeddings	embedding	NOUN
elm-5391	114	30	,	,	PUNCT
elm-5391	114	31	and	and	CCONJ
elm-5391	114	32	represented	represent	VERB
elm-5391	114	33	the	the	DET
elm-5391	114	34	numerical	numerical	ADJ
elm-5391	114	35	value	value	NOUN
elm-5391	114	36	of	of	ADP
elm-5391	114	37	the	the	DET
elm-5391	114	38	first	first	ADJ
elm-5391	114	39	two	two	NUM
elm-5391	114	40	principal	principal	ADJ
elm-5391	114	41	components	component	NOUN
elm-5391	114	42	along	along	ADP
elm-5391	114	43	the	the	DET
elm-5391	114	44	two	two	NUM
elm-5391	114	45	axes	axis	NOUN
elm-5391	114	46	.	.	PUNCT
elm-5391	115	1	each	each	DET
elm-5391	115	2	horizontal	horizontal	ADJ
elm-5391	115	3	panel	panel	NOUN
elm-5391	115	4	represents	represent	VERB
elm-5391	115	5	one	one	NUM
elm-5391	115	6	dimension	dimension	NOUN
elm-5391	115	7	of	of	ADP
elm-5391	115	8	vad	vad	PROPN
elm-5391	115	9	.	.	PROPN
elm-5391	115	10	each	each	DET
elm-5391	115	11	column	column	NOUN
elm-5391	115	12	in	in	ADP
elm-5391	115	13	figure	figure	NOUN
elm-5391	115	14	1	1	NUM
elm-5391	115	15	represents	represent	VERB
elm-5391	115	16	a	a	DET
elm-5391	115	17	type	type	NOUN
elm-5391	115	18	of	of	ADP
elm-5391	115	19	tested	test	VERB
elm-5391	115	20	word	word	NOUN
elm-5391	115	21	embedding	embed	VERB
elm-5391	115	22	.	.	PUNCT
elm-5391	116	1	each	each	DET
elm-5391	116	2	point	point	NOUN
elm-5391	116	3	is	be	AUX
elm-5391	116	4	color	color	NOUN
elm-5391	116	5	coded	code	VERB
elm-5391	116	6	with	with	ADP
elm-5391	116	7	human	human	ADJ
elm-5391	116	8	judgment	judgment	NOUN
elm-5391	116	9	.	.	PUNCT
elm-5391	117	1	the	the	DET
elm-5391	117	2	darker	dark	ADJ
elm-5391	117	3	the	the	DET
elm-5391	117	4	color	color	NOUN
elm-5391	117	5	,	,	PUNCT
elm-5391	117	6	the	the	PRON
elm-5391	117	7	higher	high	ADJ
elm-5391	117	8	the	the	DET
elm-5391	117	9	rating	rating	NOUN
elm-5391	117	10	is	be	AUX
elm-5391	117	11	along	along	ADP
elm-5391	117	12	one	one	NUM
elm-5391	117	13	of	of	ADP
elm-5391	117	14	the	the	DET
elm-5391	117	15	vad	vad	PROPN
elm-5391	117	16	dimensions	dimensions	PROPN
elm-5391	117	17	.	.	PUNCT
elm-5391	118	1	table	table	NOUN
elm-5391	118	2	1	1	NUM
elm-5391	118	3	shows	show	VERB
elm-5391	118	4	the	the	DET
elm-5391	118	5	spearman	spearman	NOUN
elm-5391	118	6	correlation	correlation	NOUN
elm-5391	118	7	coefficient	coefficient	NOUN
elm-5391	118	8	between	between	ADP
elm-5391	118	9	the	the	DET
elm-5391	118	10	numerical	numerical	ADJ
elm-5391	118	11	values	value	NOUN
elm-5391	118	12	of	of	ADP
elm-5391	118	13	human	human	ADJ
elm-5391	118	14	judgments	judgment	NOUN
elm-5391	118	15	along	along	ADP
elm-5391	118	16	each	each	PRON
elm-5391	118	17	affect	affect	VERB
elm-5391	118	18	dimension	dimension	NOUN
elm-5391	118	19	and	and	CCONJ
elm-5391	118	20	the	the	DET
elm-5391	118	21	numerical	numerical	ADJ
elm-5391	118	22	values	value	NOUN
elm-5391	118	23	of	of	ADP
elm-5391	118	24	the	the	DET
elm-5391	118	25	corresponding	corresponding	ADJ
elm-5391	118	26	principal	principal	ADJ
elm-5391	118	27	component	component	NOUN
elm-5391	118	28	given	give	VERB
elm-5391	118	29	a	a	DET
elm-5391	118	30	type	type	NOUN
elm-5391	118	31	of	of	ADP
elm-5391	118	32	word	word	NOUN
elm-5391	118	33	embedding	embed	VERB
elm-5391	118	34	.	.	PUNCT
elm-5391	119	1	from	from	ADP
elm-5391	119	2	table	table	NOUN
elm-5391	119	3	1	1	NUM
elm-5391	119	4	,	,	PUNCT
elm-5391	119	5	we	we	PRON
elm-5391	119	6	learn	learn	VERB
elm-5391	119	7	that	that	SCONJ
elm-5391	119	8	every	every	DET
elm-5391	119	9	type	type	NOUN
elm-5391	119	10	of	of	ADP
elm-5391	119	11	word	word	NOUN
elm-5391	119	12	embedding	embed	VERB
elm-5391	119	13	can	can	AUX
elm-5391	119	14	capture	capture	VERB
elm-5391	119	15	some	some	DET
elm-5391	119	16	aspects	aspect	NOUN
elm-5391	119	17	of	of	ADP
elm-5391	119	18	the	the	DET
elm-5391	119	19	3	3	NUM
elm-5391	119	20	-	-	PUNCT
elm-5391	119	21	dimensional	dimensional	ADJ
elm-5391	119	22	affect	affect	NOUN
elm-5391	119	23	meaning	meaning	NOUN
elm-5391	119	24	based	base	VERB
elm-5391	119	25	on	on	ADP
elm-5391	119	26	the	the	DET
elm-5391	119	27	significance	significance	NOUN
elm-5391	119	28	level	level	NOUN
elm-5391	119	29	of	of	ADP
elm-5391	119	30	the	the	DET
elm-5391	119	31	correlation	correlation	NOUN
elm-5391	119	32	coefficients	coefficient	NOUN
elm-5391	119	33	.	.	PUNCT
elm-5391	120	1	yet	yet	ADV
elm-5391	120	2	judging	judge	VERB
elm-5391	120	3	from	from	ADP
elm-5391	120	4	the	the	DET
elm-5391	120	5	magnitude	magnitude	NOUN
elm-5391	120	6	of	of	ADP
elm-5391	120	7	correlation	correlation	NOUN
elm-5391	120	8	coefficients	coefficient	NOUN
elm-5391	120	9	for	for	ADP
elm-5391	120	10	both	both	DET
elm-5391	120	11	principal	principal	ADJ
elm-5391	120	12	components	component	NOUN
elm-5391	120	13	as	as	ADV
elm-5391	120	14	well	well	ADV
elm-5391	120	15	as	as	ADP
elm-5391	120	16	the	the	DET
elm-5391	120	17	number	number	NOUN
elm-5391	120	18	of	of	ADP
elm-5391	120	19	correlation	correlation	NOUN
elm-5391	120	20	coefficients	coefficient	NOUN
elm-5391	120	21	that	that	PRON
elm-5391	120	22	are	be	AUX
elm-5391	120	23	larger	large	ADJ
elm-5391	120	24	than	than	ADP
elm-5391	120	25	0.2	0.2	NUM
elm-5391	120	26	out	out	ADP
elm-5391	120	27	of	of	ADP
elm-5391	120	28	the	the	DET
elm-5391	120	29	three	three	NUM
elm-5391	120	30	vad	vad	NOUN
elm-5391	120	31	dimensions	dimension	NOUN
elm-5391	120	32	,	,	PUNCT
elm-5391	120	33	we	we	PRON
elm-5391	120	34	see	see	VERB
elm-5391	120	35	that	that	SCONJ
elm-5391	120	36	the	the	DET
elm-5391	120	37	goemotion	goemotion	NOUN
elm-5391	120	38	bert	bert	NOUN
elm-5391	120	39	embedding	embed	VERB
elm-5391	120	40	and	and	CCONJ
elm-5391	120	41	the	the	DET
elm-5391	120	42	contextualized	contextualized	ADJ
elm-5391	120	43	bert	bert	NOUN
elm-5391	120	44	embedding	embed	VERB
elm-5391	120	45	outperform	outperform	NOUN
elm-5391	120	46	the	the	DET
elm-5391	120	47	other	other	ADJ
elm-5391	120	48	two	two	NUM
elm-5391	120	49	.	.	PUNCT
elm-5391	121	1	dimension	dimension	NOUN
elm-5391	121	2	pca	pca	PROPN
elm-5391	121	3	glove	glove	PROPN
elm-5391	121	4	base	base	PROPN
elm-5391	121	5	bert	bert	PROPN
elm-5391	121	6	goemotion	goemotion	PROPN
elm-5391	121	7	bert	bert	PROPN
elm-5391	121	8	contextualized	contextualize	VERB
elm-5391	121	9	bert	bert	NOUN
elm-5391	121	10	valence	valence	NOUN
elm-5391	121	11	1st	1st	NOUN
elm-5391	121	12	0.117	0.117	NUM
elm-5391	121	13	(	(	PUNCT
elm-5391	121	14	.000	.000	NUM
elm-5391	121	15	)	)	PUNCT
elm-5391	121	16	0.015	0.015	NUM
elm-5391	121	17	(	(	PUNCT
elm-5391	121	18	.247	.247	NUM
elm-5391	121	19	)	)	PUNCT
elm-5391	121	20	0.084	0.084	NUM
elm-5391	121	21	(	(	PUNCT
elm-5391	121	22	.000	.000	NUM
elm-5391	121	23	)	)	PUNCT
elm-5391	121	24	-0.077	-0.077	PUNCT
elm-5391	121	25	(	(	PUNCT
elm-5391	121	26	.000	.000	NUM
elm-5391	121	27	)	)	PUNCT
elm-5391	121	28	2nd	2nd	ADJ
elm-5391	121	29	-0.006	-0.006	PROPN
elm-5391	121	30	(	(	PUNCT
elm-5391	121	31	.672	.672	NUM
elm-5391	121	32	)	)	PUNCT
elm-5391	121	33	0.133	0.133	NUM
elm-5391	121	34	(	(	PUNCT
elm-5391	121	35	.000	.000	NUM
elm-5391	121	36	)	)	PUNCT
elm-5391	121	37	0.227	0.227	NUM
elm-5391	121	38	(	(	PUNCT
elm-5391	121	39	.000	.000	NUM
elm-5391	121	40	)	)	PUNCT
elm-5391	121	41	0.291	0.291	NUM
elm-5391	121	42	(	(	PUNCT
elm-5391	121	43	.000	.000	NUM
elm-5391	121	44	)	)	PUNCT
elm-5391	121	45	arousal	arousal	NOUN
elm-5391	121	46	1st	1st	NOUN
elm-5391	121	47	-0.005	-0.005	PUNCT
elm-5391	121	48	(	(	PUNCT
elm-5391	121	49	.696	.696	NUM
elm-5391	121	50	)	)	PUNCT
elm-5391	121	51	0.011	0.011	NUM
elm-5391	121	52	(	(	PUNCT
elm-5391	121	53	.412	.412	NUM
elm-5391	121	54	)	)	PUNCT
elm-5391	121	55	0.285	0.285	NUM
elm-5391	121	56	(	(	PUNCT
elm-5391	121	57	.000	.000	NUM
elm-5391	121	58	)	)	PUNCT
elm-5391	121	59	0.251	0.251	NUM
elm-5391	121	60	(	(	PUNCT
elm-5391	121	61	.000	.000	NUM
elm-5391	121	62	)	)	PUNCT
elm-5391	121	63	2nd	2nd	NOUN
elm-5391	121	64	0.238	0.238	NUM
elm-5391	121	65	(	(	PUNCT
elm-5391	121	66	.000	.000	NUM
elm-5391	121	67	)	)	PUNCT
elm-5391	121	68	-0.094	-0.094	PUNCT
elm-5391	121	69	(	(	PUNCT
elm-5391	121	70	.000	.000	NUM
elm-5391	121	71	)	)	PUNCT
elm-5391	121	72	-0.042	-0.042	PUNCT
elm-5391	121	73	(	(	PUNCT
elm-5391	121	74	.002	.002	NUM
elm-5391	121	75	)	)	PUNCT
elm-5391	121	76	-0.007	-0.007	PUNCT
elm-5391	121	77	(	(	PUNCT
elm-5391	121	78	.599	.599	NUM
elm-5391	121	79	)	)	PUNCT
elm-5391	121	80	dominance	dominance	NOUN
elm-5391	121	81	1st	1st	NOUN
elm-5391	121	82	0.316	0.316	NUM
elm-5391	121	83	(	(	PUNCT
elm-5391	121	84	.000	.000	NUM
elm-5391	121	85	)	)	PUNCT
elm-5391	121	86	0.028	0.028	NUM
elm-5391	121	87	(	(	PUNCT
elm-5391	121	88	.037	.037	NUM
elm-5391	121	89	)	)	PUNCT
elm-5391	121	90	0.160	0.160	NUM
elm-5391	121	91	(	(	PUNCT
elm-5391	121	92	.000	.000	NUM
elm-5391	121	93	)	)	PUNCT
elm-5391	121	94	0.086	0.086	NUM
elm-5391	121	95	(	(	PUNCT
elm-5391	121	96	.000	.000	NUM
elm-5391	121	97	)	)	PUNCT
elm-5391	121	98	2nd	2nd	NOUN
elm-5391	121	99	0.182	0.182	NUM
elm-5391	121	100	(	(	PUNCT
elm-5391	121	101	.000	.000	NUM
elm-5391	121	102	)	)	PUNCT
elm-5391	121	103	0.111	0.111	NUM
elm-5391	121	104	(	(	PUNCT
elm-5391	121	105	.000	.000	NUM
elm-5391	121	106	)	)	PUNCT
elm-5391	121	107	0.283	0.283	NUM
elm-5391	121	108	(	(	PUNCT
elm-5391	121	109	.000	.000	NUM
elm-5391	121	110	)	)	PUNCT
elm-5391	121	111	0.380	0.380	NUM
elm-5391	121	112	(	(	PUNCT
elm-5391	121	113	.000	.000	NUM
elm-5391	121	114	)	)	PUNCT
elm-5391	121	115	table	table	NOUN
elm-5391	121	116	1	1	NUM
elm-5391	121	117	:	:	PUNCT
elm-5391	121	118	spearman	spearman	NOUN
elm-5391	121	119	correlation	correlation	NOUN
elm-5391	121	120	coefficients	coefficient	NOUN
elm-5391	121	121	(	(	PUNCT
elm-5391	121	122	p	p	NOUN
elm-5391	121	123	values	value	NOUN
elm-5391	121	124	in	in	ADP
elm-5391	121	125	the	the	DET
elm-5391	121	126	bracket	bracket	NOUN
elm-5391	121	127	)	)	PUNCT
elm-5391	121	128	between	between	ADP
elm-5391	121	129	human	human	ADJ
elm-5391	121	130	ratings	rating	NOUN
elm-5391	121	131	(	(	PUNCT
elm-5391	121	132	01	01	NUM
elm-5391	121	133	)	)	PUNCT
elm-5391	121	134	in	in	ADP
elm-5391	121	135	each	each	DET
elm-5391	121	136	affect	affect	VERB
elm-5391	121	137	dimension	dimension	NOUN
elm-5391	121	138	and	and	CCONJ
elm-5391	121	139	each	each	DET
elm-5391	121	140	corresponding	correspond	VERB
elm-5391	121	141	principal	principal	ADJ
elm-5391	121	142	component	component	NOUN
elm-5391	121	143	of	of	ADP
elm-5391	121	144	four	four	NUM
elm-5391	121	145	types	type	NOUN
elm-5391	121	146	of	of	ADP
elm-5391	121	147	word	word	NOUN
elm-5391	121	148	embeddings	embedding	NOUN
elm-5391	121	149	(	(	PUNCT
elm-5391	121	150	correlation	correlation	NOUN
elm-5391	121	151	coefficients	coefficient	NOUN
elm-5391	121	152	in	in	ADP
elm-5391	121	153	bold	bold	ADJ
elm-5391	121	154	when	when	SCONJ
elm-5391	121	155	larger	large	ADJ
elm-5391	121	156	than	than	ADP
elm-5391	121	157	0.2	0.2	NUM
elm-5391	121	158	,	,	PUNCT
elm-5391	121	159	also	also	ADV
elm-5391	121	160	see	see	VERB
elm-5391	121	161	each	each	DET
elm-5391	121	162	subplot	subplot	NOUN
elm-5391	121	163	in	in	ADP
elm-5391	121	164	figure	figure	NOUN
elm-5391	121	165	1	1	NUM
elm-5391	121	166	)	)	PUNCT
elm-5391	121	167	in	in	ADP
elm-5391	121	168	table	table	NOUN
elm-5391	121	169	2	2	NUM
elm-5391	121	170	,	,	PUNCT
elm-5391	121	171	we	we	PRON
elm-5391	121	172	present	present	VERB
elm-5391	121	173	the	the	DET
elm-5391	121	174	explained	explain	VERB
elm-5391	121	175	variance	variance	NOUN
elm-5391	121	176	ratios	ratio	NOUN
elm-5391	121	177	to	to	PART
elm-5391	121	178	show	show	VERB
elm-5391	121	179	what	what	PRON
elm-5391	121	180	percentage	percentage	NOUN
elm-5391	121	181	of	of	ADP
elm-5391	121	182	the	the	DET
elm-5391	121	183	variance	variance	NOUN
elm-5391	121	184	in	in	ADP
elm-5391	121	185	the	the	DET
elm-5391	121	186	whole	whole	ADJ
elm-5391	121	187	embedding	embed	VERB
elm-5391	121	188	space	space	NOUN
elm-5391	121	189	was	be	AUX
elm-5391	121	190	explained	explain	VERB
elm-5391	121	191	by	by	ADP
elm-5391	121	192	the	the	DET
elm-5391	121	193	first	first	ADJ
elm-5391	121	194	two	two	NUM
elm-5391	121	195	principal	principal	ADJ
elm-5391	121	196	components	component	NOUN
elm-5391	121	197	for	for	ADP
elm-5391	121	198	each	each	DET
elm-5391	121	199	type	type	NOUN
elm-5391	121	200	3we	3we	NOUN
elm-5391	121	201	transformed	transform	VERB
elm-5391	121	202	the	the	DET
elm-5391	121	203	numerical	numerical	ADJ
elm-5391	121	204	human	human	PROPN
elm-5391	121	205	rating	rating	NOUN
elm-5391	121	206	(	(	PUNCT
elm-5391	121	207	0	0	NUM
elm-5391	121	208	-	-	SYM
elm-5391	121	209	1	1	NUM
elm-5391	121	210	)	)	PUNCT
elm-5391	121	211	to	to	ADP
elm-5391	121	212	a	a	DET
elm-5391	121	213	binary	binary	ADJ
elm-5391	121	214	categorical	categorical	ADJ
elm-5391	121	215	variable	variable	NOUN
elm-5391	121	216	with	with	ADP
elm-5391	121	217	the	the	DET
elm-5391	121	218	threshold	threshold	NOUN
elm-5391	121	219	of	of	ADP
elm-5391	121	220	0.5	0.5	NUM
elm-5391	121	221	.	.	PUNCT
elm-5391	122	1	proceedings	proceeding	NOUN
elm-5391	122	2	of	of	ADP
elm-5391	122	3	elm	elm	NOUN
elm-5391	122	4	2	2	NUM
elm-5391	122	5	:	:	PUNCT
elm-5391	122	6	310	310	NUM
elm-5391	122	7	-	-	SYM
elm-5391	122	8	321	321	NUM
elm-5391	122	9	,	,	PUNCT
elm-5391	122	10	2023	2023	NUM
elm-5391	122	11	yuhan	yuhan	PROPN
elm-5391	122	12	zhang	zhang	PROPN
elm-5391	122	13	,	,	PUNCT
elm-5391	122	14	wenqi	wenqi	PROPN
elm-5391	122	15	chen	chen	PROPN
elm-5391	122	16	,	,	PUNCT
elm-5391	122	17	ruihan	ruihan	PROPN
elm-5391	122	18	zhang	zhang	PROPN
elm-5391	122	19	and	and	CCONJ
elm-5391	122	20	xiajie	xiajie	PROPN
elm-5391	122	21	zhang	zhang	PROPN
elm-5391	122	22	:	:	PUNCT
elm-5391	122	23	representing	represent	VERB
elm-5391	122	24	affect	affect	VERB
elm-5391	122	25	information	information	NOUN
elm-5391	122	26	in	in	ADP
elm-5391	122	27	word	word	NOUN
elm-5391	122	28	embeddings	embedding	NOUN
elm-5391	122	29	.	.	PUNCT
elm-5391	123	1	314	314	NUM
elm-5391	123	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	123	3	https://www.elm-conference.net/	https://www.elm-conference.net/	PRON
elm-5391	123	4	figure	figure	NOUN
elm-5391	123	5	1	1	NUM
elm-5391	123	6	:	:	SYM
elm-5391	123	7	2	2	NUM
elm-5391	123	8	-	-	PUNCT
elm-5391	123	9	components	component	NOUN
elm-5391	123	10	pca	pca	NOUN
elm-5391	123	11	representations	representation	NOUN
elm-5391	123	12	of	of	ADP
elm-5391	123	13	glove	glove	NOUN
elm-5391	123	14	,	,	PUNCT
elm-5391	123	15	base	base	NOUN
elm-5391	123	16	bert	bert	PROPN
elm-5391	123	17	,	,	PUNCT
elm-5391	123	18	goemotion	goemotion	NOUN
elm-5391	123	19	bert	bert	PROPN
elm-5391	123	20	,	,	PUNCT
elm-5391	123	21	and	and	CCONJ
elm-5391	123	22	contextualized	contextualized	ADJ
elm-5391	123	23	bert	bert	PROPN
elm-5391	123	24	&	&	CCONJ
elm-5391	123	25	human	human	ADJ
elm-5391	123	26	ratings	rating	NOUN
elm-5391	123	27	for	for	ADP
elm-5391	123	28	valence	valence	NOUN
elm-5391	123	29	,	,	PUNCT
elm-5391	123	30	arousal	arousal	NOUN
elm-5391	123	31	,	,	PUNCT
elm-5391	123	32	and	and	CCONJ
elm-5391	123	33	dominance	dominance	NOUN
elm-5391	123	34	(	(	PUNCT
elm-5391	123	35	human	human	ADJ
elm-5391	123	36	ratings	rating	NOUN
elm-5391	123	37	were	be	AUX
elm-5391	123	38	color	color	NOUN
elm-5391	123	39	-	-	PUNCT
elm-5391	123	40	coded	code	VERB
elm-5391	123	41	in	in	ADP
elm-5391	123	42	a	a	DET
elm-5391	123	43	spectrum	spectrum	NOUN
elm-5391	123	44	.	.	PUNCT
elm-5391	124	1	each	each	DET
elm-5391	124	2	dot	dot	NOUN
elm-5391	124	3	represents	represent	VERB
elm-5391	124	4	a	a	DET
elm-5391	124	5	word	word	NOUN
elm-5391	124	6	.	.	PUNCT
elm-5391	125	1	5,586	5,586	NUM
elm-5391	125	2	words	word	NOUN
elm-5391	125	3	are	be	AUX
elm-5391	125	4	represented	represent	VERB
elm-5391	125	5	.	.	PUNCT
elm-5391	126	1	the	the	DET
elm-5391	126	2	darker	dark	ADJ
elm-5391	126	3	the	the	DET
elm-5391	126	4	dot	dot	NOUN
elm-5391	126	5	,	,	PUNCT
elm-5391	126	6	the	the	PRON
elm-5391	126	7	more	more	ADV
elm-5391	126	8	prominent	prominent	ADJ
elm-5391	126	9	the	the	DET
elm-5391	126	10	human	human	ADJ
elm-5391	126	11	rating	rating	NOUN
elm-5391	126	12	in	in	ADP
elm-5391	126	13	the	the	DET
elm-5391	126	14	respective	respective	ADJ
elm-5391	126	15	dimension	dimension	NOUN
elm-5391	126	16	.	.	PUNCT
elm-5391	126	17	)	)	PUNCT
elm-5391	126	18	of	of	ADP
elm-5391	126	19	word	word	NOUN
elm-5391	126	20	embedding	embed	VERB
elm-5391	126	21	.	.	PUNCT
elm-5391	127	1	two	two	NUM
elm-5391	127	2	pieces	piece	NOUN
elm-5391	127	3	of	of	ADP
elm-5391	127	4	observations	observation	NOUN
elm-5391	127	5	are	be	AUX
elm-5391	127	6	worth	worth	ADJ
elm-5391	127	7	interpreting	interpret	VERB
elm-5391	127	8	.	.	PUNCT
elm-5391	128	1	first	first	ADV
elm-5391	128	2	,	,	PUNCT
elm-5391	128	3	the	the	DET
elm-5391	128	4	relatively	relatively	ADV
elm-5391	128	5	low	low	ADJ
elm-5391	128	6	ratio	ratio	NOUN
elm-5391	128	7	of	of	ADP
elm-5391	128	8	explained	explain	VERB
elm-5391	128	9	variance	variance	NOUN
elm-5391	128	10	across	across	ADP
elm-5391	128	11	the	the	DET
elm-5391	128	12	eight	eight	NUM
elm-5391	128	13	values	value	NOUN
elm-5391	128	14	indicates	indicate	VERB
elm-5391	128	15	that	that	SCONJ
elm-5391	128	16	the	the	DET
elm-5391	128	17	information	information	NOUN
elm-5391	128	18	represented	represent	VERB
elm-5391	128	19	in	in	ADP
elm-5391	128	20	the	the	DET
elm-5391	128	21	first	first	ADJ
elm-5391	128	22	two	two	NUM
elm-5391	128	23	principal	principal	ADJ
elm-5391	128	24	components	component	NOUN
elm-5391	128	25	might	might	AUX
elm-5391	128	26	not	not	PART
elm-5391	128	27	be	be	AUX
elm-5391	128	28	representative	representative	ADJ
elm-5391	128	29	to	to	PART
elm-5391	128	30	generalize	generalize	VERB
elm-5391	128	31	on	on	ADP
elm-5391	128	32	the	the	DET
elm-5391	128	33	capacity	capacity	NOUN
elm-5391	128	34	of	of	ADP
elm-5391	128	35	certain	certain	ADJ
elm-5391	128	36	word	word	NOUN
elm-5391	128	37	embedding	embed	VERB
elm-5391	128	38	to	to	PART
elm-5391	128	39	encode	encode	VERB
elm-5391	128	40	certain	certain	ADJ
elm-5391	128	41	affect	affect	NOUN
elm-5391	128	42	meaning	mean	VERB
elm-5391	128	43	.	.	PUNCT
elm-5391	129	1	there	there	PRON
elm-5391	129	2	might	might	AUX
elm-5391	129	3	be	be	AUX
elm-5391	129	4	some	some	PRON
elm-5391	129	5	affect	affect	VERB
elm-5391	129	6	meaning	meaning	NOUN
elm-5391	129	7	that	that	PRON
elm-5391	129	8	has	have	AUX
elm-5391	129	9	been	be	AUX
elm-5391	129	10	captured	capture	VERB
elm-5391	129	11	in	in	ADP
elm-5391	129	12	higher	high	ADJ
elm-5391	129	13	dimensions	dimension	NOUN
elm-5391	129	14	of	of	ADP
elm-5391	129	15	the	the	DET
elm-5391	129	16	vector	vector	NOUN
elm-5391	129	17	space	space	NOUN
elm-5391	129	18	but	but	CCONJ
elm-5391	129	19	not	not	PART
elm-5391	129	20	projected	project	VERB
elm-5391	129	21	in	in	ADP
elm-5391	129	22	the	the	DET
elm-5391	129	23	pca	pca	PROPN
elm-5391	129	24	measure	measure	NOUN
elm-5391	129	25	.	.	PUNCT
elm-5391	130	1	we	we	PRON
elm-5391	130	2	would	would	AUX
elm-5391	130	3	need	need	VERB
elm-5391	130	4	more	more	ADV
elm-5391	130	5	sensitive	sensitive	ADJ
elm-5391	130	6	probes	probe	NOUN
elm-5391	130	7	other	other	ADJ
elm-5391	130	8	than	than	ADP
elm-5391	130	9	pca	pca	NOUN
elm-5391	130	10	in	in	ADP
elm-5391	130	11	future	future	ADJ
elm-5391	130	12	investigations	investigation	NOUN
elm-5391	130	13	.	.	PUNCT
elm-5391	131	1	second	second	ADJ
elm-5391	131	2	,	,	PUNCT
elm-5391	131	3	assuming	assume	VERB
elm-5391	131	4	that	that	SCONJ
elm-5391	131	5	the	the	DET
elm-5391	131	6	pca	pca	NOUN
elm-5391	131	7	analysis	analysis	NOUN
elm-5391	131	8	is	be	AUX
elm-5391	131	9	a	a	DET
elm-5391	131	10	great	great	ADJ
elm-5391	131	11	approximant	approximant	NOUN
elm-5391	131	12	for	for	ADP
elm-5391	131	13	affect	affect	PROPN
elm-5391	131	14	encoding	encode	VERB
elm-5391	131	15	,	,	PUNCT
elm-5391	131	16	the	the	DET
elm-5391	131	17	fact	fact	NOUN
elm-5391	131	18	that	that	SCONJ
elm-5391	131	19	the	the	DET
elm-5391	131	20	first	first	ADJ
elm-5391	131	21	component	component	NOUN
elm-5391	131	22	of	of	ADP
elm-5391	131	23	base	base	NOUN
elm-5391	131	24	bert	bert	NOUN
elm-5391	131	25	accounts	account	VERB
elm-5391	131	26	for	for	ADP
elm-5391	131	27	29.4	29.4	NUM
elm-5391	131	28	%	%	NOUN
elm-5391	131	29	of	of	ADP
elm-5391	131	30	the	the	DET
elm-5391	131	31	total	total	ADJ
elm-5391	131	32	variance	variance	NOUN
elm-5391	131	33	and	and	CCONJ
elm-5391	131	34	yet	yet	ADV
elm-5391	131	35	does	do	AUX
elm-5391	131	36	not	not	PART
elm-5391	131	37	correlate	correlate	VERB
elm-5391	131	38	with	with	ADP
elm-5391	131	39	vad	vad	PROPN
elm-5391	131	40	as	as	ADV
elm-5391	131	41	much	much	ADV
elm-5391	131	42	as	as	SCONJ
elm-5391	131	43	the	the	DET
elm-5391	131	44	other	other	ADJ
elm-5391	131	45	word	word	NOUN
elm-5391	131	46	embeddings	embedding	NOUN
elm-5391	131	47	shows	show	VERB
elm-5391	131	48	that	that	SCONJ
elm-5391	131	49	base	base	NOUN
elm-5391	131	50	bert	bert	PROPN
elm-5391	131	51	might	might	AUX
elm-5391	131	52	not	not	PART
elm-5391	131	53	be	be	AUX
elm-5391	131	54	optimized	optimize	VERB
elm-5391	131	55	to	to	PART
elm-5391	131	56	distinguish	distinguish	VERB
elm-5391	131	57	intricate	intricate	ADJ
elm-5391	131	58	affect	affect	NOUN
elm-5391	131	59	meaning	meaning	NOUN
elm-5391	131	60	of	of	ADP
elm-5391	131	61	words	word	NOUN
elm-5391	131	62	.	.	PUNCT
elm-5391	132	1	this	this	PRON
elm-5391	132	2	offers	offer	VERB
elm-5391	132	3	a	a	DET
elm-5391	132	4	nice	nice	ADJ
elm-5391	132	5	window	window	NOUN
elm-5391	132	6	to	to	PART
elm-5391	132	7	peek	peek	VERB
elm-5391	132	8	into	into	ADP
elm-5391	132	9	what	what	PRON
elm-5391	132	10	linguistic	linguistic	ADJ
elm-5391	132	11	information	information	NOUN
elm-5391	132	12	is	be	AUX
elm-5391	132	13	and	and	CCONJ
elm-5391	132	14	is	be	AUX
elm-5391	132	15	n’t	not	PART
elm-5391	132	16	weighted	weight	VERB
elm-5391	132	17	the	the	DET
elm-5391	132	18	most	most	ADJ
elm-5391	132	19	for	for	ADP
elm-5391	132	20	bert	bert	PROPN
elm-5391	132	21	.	.	PUNCT
elm-5391	133	1	explained	explain	VERB
elm-5391	133	2	ratio	ratio	NOUN
elm-5391	133	3	glove	glove	PROPN
elm-5391	133	4	base	base	PROPN
elm-5391	133	5	bert	bert	PROPN
elm-5391	133	6	goemotion	goemotion	PROPN
elm-5391	133	7	bert	bert	PROPN
elm-5391	133	8	contextualized	contextualize	VERB
elm-5391	133	9	bert	bert	PROPN
elm-5391	133	10	first	first	ADV
elm-5391	133	11	0.040	0.040	NUM
elm-5391	133	12	0.294	0.294	NUM
elm-5391	133	13	0.118	0.118	NUM
elm-5391	133	14	0.021	0.021	NUM
elm-5391	133	15	second	second	NOUN
elm-5391	133	16	0.030	0.030	NUM
elm-5391	133	17	0.040	0.040	NUM
elm-5391	133	18	0.070	0.070	NUM
elm-5391	133	19	0.016	0.016	NUM
elm-5391	133	20	table	table	NOUN
elm-5391	133	21	2	2	NUM
elm-5391	133	22	:	:	PUNCT
elm-5391	133	23	explained	explain	VERB
elm-5391	133	24	variance	variance	NOUN
elm-5391	133	25	ratio	ratio	NOUN
elm-5391	133	26	of	of	ADP
elm-5391	133	27	the	the	DET
elm-5391	133	28	two	two	NUM
elm-5391	133	29	principal	principal	ADJ
elm-5391	133	30	components	component	NOUN
elm-5391	133	31	proceedings	proceeding	NOUN
elm-5391	133	32	of	of	ADP
elm-5391	133	33	elm	elm	NOUN
elm-5391	133	34	2	2	NUM
elm-5391	133	35	:	:	PUNCT
elm-5391	133	36	310	310	NUM
elm-5391	133	37	-	-	SYM
elm-5391	133	38	321	321	NUM
elm-5391	133	39	,	,	PUNCT
elm-5391	133	40	2023	2023	NUM
elm-5391	133	41	yuhan	yuhan	PROPN
elm-5391	133	42	zhang	zhang	PROPN
elm-5391	133	43	,	,	PUNCT
elm-5391	133	44	wenqi	wenqi	PROPN
elm-5391	133	45	chen	chen	PROPN
elm-5391	133	46	,	,	PUNCT
elm-5391	133	47	ruihan	ruihan	PROPN
elm-5391	133	48	zhang	zhang	PROPN
elm-5391	133	49	and	and	CCONJ
elm-5391	133	50	xiajie	xiajie	PROPN
elm-5391	133	51	zhang	zhang	PROPN
elm-5391	133	52	:	:	PUNCT
elm-5391	133	53	representing	represent	VERB
elm-5391	133	54	affect	affect	VERB
elm-5391	133	55	information	information	NOUN
elm-5391	133	56	in	in	ADP
elm-5391	133	57	word	word	NOUN
elm-5391	133	58	embeddings	embedding	NOUN
elm-5391	133	59	.	.	PUNCT
elm-5391	134	1	315	315	NUM
elm-5391	134	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	134	3	https://www.elm-conference.net/	https://www.elm-conference.net/	ADP
elm-5391	134	4	4.2	4.2	NUM
elm-5391	134	5	.	.	PUNCT
elm-5391	134	6	cosine	cosine	NOUN
elm-5391	134	7	similarity	similarity	NOUN
elm-5391	134	8	.	.	PUNCT
elm-5391	135	1	table	table	NOUN
elm-5391	135	2	3	3	NUM
elm-5391	135	3	captures	capture	VERB
elm-5391	135	4	the	the	DET
elm-5391	135	5	spearman	spearman	NOUN
elm-5391	135	6	correlation	correlation	NOUN
elm-5391	135	7	coefficients	coefficient	NOUN
elm-5391	135	8	based	base	VERB
elm-5391	135	9	on	on	ADP
elm-5391	135	10	pairwise	pairwise	PROPN
elm-5391	135	11	cosine	cosine	NOUN
elm-5391	135	12	similarity	similarity	NOUN
elm-5391	135	13	of	of	ADP
elm-5391	135	14	80	80	NUM
elm-5391	135	15	words	word	NOUN
elm-5391	135	16	whose	whose	DET
elm-5391	135	17	embeddings	embedding	NOUN
elm-5391	135	18	were	be	AUX
elm-5391	135	19	from	from	ADP
elm-5391	135	20	each	each	DET
elm-5391	135	21	type	type	NOUN
elm-5391	135	22	of	of	ADP
elm-5391	135	23	word	word	NOUN
elm-5391	135	24	embeddings	embedding	NOUN
elm-5391	135	25	.	.	PUNCT
elm-5391	136	1	the	the	DET
elm-5391	136	2	main	main	ADJ
elm-5391	136	3	comparison	comparison	NOUN
elm-5391	136	4	is	be	AUX
elm-5391	136	5	in	in	ADP
elm-5391	136	6	the	the	DET
elm-5391	136	7	first	first	ADJ
elm-5391	136	8	column	column	NOUN
elm-5391	136	9	where	where	SCONJ
elm-5391	136	10	human	human	ADJ
elm-5391	136	11	judgments	judgment	NOUN
elm-5391	136	12	on	on	ADP
elm-5391	136	13	the	the	DET
elm-5391	136	14	semantic	semantic	ADJ
elm-5391	136	15	similarity	similarity	NOUN
elm-5391	136	16	of	of	ADP
elm-5391	136	17	the	the	DET
elm-5391	136	18	80×	80×	NOUN
elm-5391	136	19	79	79	NUM
elm-5391	136	20	pairs	pair	NOUN
elm-5391	136	21	of	of	ADP
elm-5391	136	22	words	word	NOUN
elm-5391	136	23	are	be	AUX
elm-5391	136	24	compared	compare	VERB
elm-5391	136	25	with	with	ADP
elm-5391	136	26	the	the	DET
elm-5391	136	27	cosine	cosine	ADJ
elm-5391	136	28	distance	distance	NOUN
elm-5391	136	29	of	of	ADP
elm-5391	136	30	those	those	DET
elm-5391	136	31	word	word	NOUN
elm-5391	136	32	pairs	pair	NOUN
elm-5391	136	33	based	base	VERB
elm-5391	136	34	on	on	ADP
elm-5391	136	35	different	different	ADJ
elm-5391	136	36	types	type	NOUN
elm-5391	136	37	of	of	ADP
elm-5391	136	38	word	word	NOUN
elm-5391	136	39	embedding	embed	VERB
elm-5391	136	40	.	.	PUNCT
elm-5391	137	1	the	the	DET
elm-5391	137	2	result	result	NOUN
elm-5391	137	3	shows	show	VERB
elm-5391	137	4	that	that	SCONJ
elm-5391	137	5	all	all	DET
elm-5391	137	6	the	the	DET
elm-5391	137	7	four	four	NUM
elm-5391	137	8	types	type	NOUN
elm-5391	137	9	of	of	ADP
elm-5391	137	10	word	word	NOUN
elm-5391	137	11	embeddings	embedding	NOUN
elm-5391	137	12	show	show	VERB
elm-5391	137	13	a	a	DET
elm-5391	137	14	significant	significant	ADJ
elm-5391	137	15	correlation	correlation	NOUN
elm-5391	137	16	with	with	ADP
elm-5391	137	17	human	human	ADJ
elm-5391	137	18	affect	affect	NOUN
elm-5391	137	19	judgment	judgment	NOUN
elm-5391	137	20	.	.	PUNCT
elm-5391	138	1	the	the	DET
elm-5391	138	2	base	base	ADJ
elm-5391	138	3	bert	bert	NOUN
elm-5391	138	4	embedding	embed	VERB
elm-5391	138	5	has	have	VERB
elm-5391	138	6	the	the	DET
elm-5391	138	7	weakest	weak	ADJ
elm-5391	138	8	correlation	correlation	NOUN
elm-5391	138	9	of	of	ADP
elm-5391	138	10	all	all	PRON
elm-5391	138	11	,	,	PUNCT
elm-5391	138	12	which	which	PRON
elm-5391	138	13	also	also	ADV
elm-5391	138	14	echoes	echo	VERB
elm-5391	138	15	with	with	ADP
elm-5391	138	16	figure	figure	NOUN
elm-5391	138	17	1	1	NUM
elm-5391	138	18	.	.	PUNCT
elm-5391	139	1	then	then	ADV
elm-5391	139	2	,	,	PUNCT
elm-5391	139	3	in	in	ADP
elm-5391	139	4	the	the	DET
elm-5391	139	5	order	order	NOUN
elm-5391	139	6	of	of	ADP
elm-5391	139	7	more	more	ADJ
elm-5391	139	8	relatedness	relatedness	NOUN
elm-5391	139	9	is	be	AUX
elm-5391	139	10	the	the	DET
elm-5391	139	11	goemotion	goemotion	NOUN
elm-5391	139	12	bert	bert	PROPN
elm-5391	139	13	embedding	embed	VERB
elm-5391	139	14	,	,	PUNCT
elm-5391	139	15	glove	glove	NOUN
elm-5391	139	16	,	,	PUNCT
elm-5391	139	17	and	and	CCONJ
elm-5391	139	18	the	the	DET
elm-5391	139	19	contextualized	contextualized	ADJ
elm-5391	139	20	bert	bert	NOUN
elm-5391	139	21	.	.	PUNCT
elm-5391	140	1	it	it	PRON
elm-5391	140	2	is	be	AUX
elm-5391	140	3	worthwhile	worthwhile	ADJ
elm-5391	140	4	to	to	PART
elm-5391	140	5	deliberate	deliberate	VERB
elm-5391	140	6	on	on	ADP
elm-5391	140	7	why	why	SCONJ
elm-5391	140	8	the	the	DET
elm-5391	140	9	static	static	ADJ
elm-5391	140	10	word	word	NOUN
elm-5391	140	11	embeddings	embedding	NOUN
elm-5391	140	12	like	like	SCONJ
elm-5391	140	13	glove	glove	NOUN
elm-5391	140	14	outperforms	outperform	VERB
elm-5391	140	15	the	the	DET
elm-5391	140	16	base	base	ADJ
elm-5391	140	17	bert	bert	PROPN
elm-5391	140	18	.	.	PUNCT
elm-5391	141	1	also	also	ADV
elm-5391	141	2	,	,	PUNCT
elm-5391	141	3	it	it	PRON
elm-5391	141	4	is	be	AUX
elm-5391	141	5	interesting	interesting	ADJ
elm-5391	141	6	that	that	SCONJ
elm-5391	141	7	even	even	ADV
elm-5391	141	8	though	though	SCONJ
elm-5391	141	9	the	the	DET
elm-5391	141	10	goemotion	goemotion	NOUN
elm-5391	141	11	bert	bert	PROPN
elm-5391	141	12	is	be	AUX
elm-5391	141	13	fined	fine	VERB
elm-5391	141	14	tuned	tune	VERB
elm-5391	141	15	on	on	ADP
elm-5391	141	16	emotion	emotion	NOUN
elm-5391	141	17	specific	specific	ADJ
elm-5391	141	18	task	task	NOUN
elm-5391	141	19	,	,	PUNCT
elm-5391	141	20	the	the	DET
elm-5391	141	21	correlation	correlation	NOUN
elm-5391	141	22	is	be	AUX
elm-5391	141	23	still	still	ADV
elm-5391	141	24	worse	bad	ADJ
elm-5391	141	25	than	than	ADP
elm-5391	141	26	bert	bert	NOUN
elm-5391	141	27	-	-	PUNCT
elm-5391	141	28	based	base	VERB
elm-5391	141	29	embeddings	embedding	NOUN
elm-5391	141	30	with	with	ADP
elm-5391	141	31	movie	movie	NOUN
elm-5391	141	32	specific	specific	ADJ
elm-5391	141	33	contextualized	contextualized	ADJ
elm-5391	141	34	information	information	NOUN
elm-5391	141	35	.	.	PUNCT
elm-5391	142	1	corr	corr	NOUN
elm-5391	142	2	(	(	PUNCT
elm-5391	142	3	p	p	NOUN
elm-5391	142	4	)	)	PUNCT
elm-5391	142	5	vad	vad	PROPN
elm-5391	142	6	glove	glove	NOUN
elm-5391	142	7	base	base	PROPN
elm-5391	142	8	goemotion	goemotion	NOUN
elm-5391	142	9	contextualized	contextualize	VERB
elm-5391	142	10	vad	vad	PROPN
elm-5391	142	11	1.000	1.000	NUM
elm-5391	142	12	(	(	PUNCT
elm-5391	142	13	.000	.000	NUM
elm-5391	142	14	)	)	PUNCT
elm-5391	142	15	glove	glove	NOUN
elm-5391	142	16	0.272	0.272	NUM
elm-5391	142	17	(	(	PUNCT
elm-5391	142	18	.000	.000	NUM
elm-5391	142	19	)	)	PUNCT
elm-5391	142	20	1.000	1.000	NUM
elm-5391	142	21	(	(	PUNCT
elm-5391	142	22	.000	.000	NUM
elm-5391	142	23	)	)	PUNCT
elm-5391	142	24	base	base	NOUN
elm-5391	142	25	0.116	0.116	NUM
elm-5391	142	26	(	(	PUNCT
elm-5391	142	27	.000	.000	NUM
elm-5391	142	28	)	)	PUNCT
elm-5391	142	29	0.148	0.148	NUM
elm-5391	142	30	(	(	PUNCT
elm-5391	142	31	.000	.000	NUM
elm-5391	142	32	)	)	PUNCT
elm-5391	142	33	1.000	1.000	NUM
elm-5391	142	34	(	(	PUNCT
elm-5391	142	35	.000	.000	NUM
elm-5391	142	36	)	)	PUNCT
elm-5391	142	37	goemotion	goemotion	NOUN
elm-5391	142	38	0.252	0.252	NUM
elm-5391	142	39	(	(	PUNCT
elm-5391	142	40	.000	.000	NUM
elm-5391	142	41	)	)	PUNCT
elm-5391	142	42	0.172	0.172	NUM
elm-5391	142	43	(	(	PUNCT
elm-5391	142	44	.000	.000	NUM
elm-5391	142	45	)	)	PUNCT
elm-5391	142	46	0.013	0.013	NUM
elm-5391	142	47	(	(	PUNCT
elm-5391	142	48	.471	.471	NUM
elm-5391	142	49	)	)	PUNCT
elm-5391	142	50	1.000	1.000	NUM
elm-5391	142	51	(	(	PUNCT
elm-5391	142	52	.000	.000	NUM
elm-5391	142	53	)	)	PUNCT
elm-5391	142	54	contextualized	contextualize	VERB
elm-5391	142	55	0.314	0.314	NUM
elm-5391	142	56	(	(	PUNCT
elm-5391	142	57	.000	.000	NUM
elm-5391	142	58	)	)	PUNCT
elm-5391	142	59	0.710	0.710	NUM
elm-5391	142	60	(	(	PUNCT
elm-5391	142	61	.000	.000	NUM
elm-5391	142	62	)	)	PUNCT
elm-5391	142	63	0.240	0.240	NUM
elm-5391	142	64	(	(	PUNCT
elm-5391	142	65	.000	.000	NUM
elm-5391	142	66	)	)	PUNCT
elm-5391	142	67	0.204	0.204	NUM
elm-5391	142	68	(	(	PUNCT
elm-5391	142	69	.000	.000	NUM
elm-5391	142	70	)	)	PUNCT
elm-5391	142	71	1.000	1.000	NUM
elm-5391	142	72	(	(	PUNCT
elm-5391	142	73	.000	.000	NUM
elm-5391	142	74	)	)	PUNCT
elm-5391	142	75	table	table	NOUN
elm-5391	142	76	3	3	NUM
elm-5391	142	77	:	:	PUNCT
elm-5391	142	78	spearman	spearman	NOUN
elm-5391	142	79	correlation	correlation	NOUN
elm-5391	142	80	(	(	PUNCT
elm-5391	142	81	p	p	NOUN
elm-5391	142	82	value	value	NOUN
elm-5391	142	83	)	)	PUNCT
elm-5391	142	84	of	of	ADP
elm-5391	142	85	pairwise	pairwise	NOUN
elm-5391	142	86	cosine	cosine	NOUN
elm-5391	142	87	similarities	similarity	NOUN
elm-5391	142	88	between	between	ADP
elm-5391	142	89	each	each	PRON
elm-5391	142	90	and	and	CCONJ
elm-5391	142	91	the	the	DET
elm-5391	142	92	rest	rest	NOUN
elm-5391	142	93	of	of	ADP
elm-5391	142	94	word	word	NOUN
elm-5391	142	95	embedding	embed	VERB
elm-5391	142	96	types	type	NOUN
elm-5391	142	97	(	(	PUNCT
elm-5391	142	98	coefficients	coefficient	NOUN
elm-5391	142	99	larger	large	ADJ
elm-5391	142	100	than	than	ADP
elm-5391	142	101	0.2	0.2	NUM
elm-5391	142	102	and	and	CCONJ
elm-5391	142	103	less	less	ADJ
elm-5391	142	104	than	than	ADP
elm-5391	142	105	1	1	NUM
elm-5391	142	106	are	be	AUX
elm-5391	142	107	in	in	ADP
elm-5391	142	108	bold	bold	ADJ
elm-5391	142	109	.	.	PUNCT
elm-5391	142	110	)	)	PUNCT
elm-5391	143	1	4.3	4.3	NUM
elm-5391	143	2	.	.	PUNCT
elm-5391	144	1	linear	linear	PROPN
elm-5391	144	2	classifier	classifier	NOUN
elm-5391	144	3	.	.	PUNCT
elm-5391	145	1	in	in	ADP
elm-5391	145	2	this	this	DET
elm-5391	145	3	section	section	NOUN
elm-5391	145	4	,	,	PUNCT
elm-5391	145	5	we	we	PRON
elm-5391	145	6	report	report	VERB
elm-5391	145	7	the	the	DET
elm-5391	145	8	classification	classification	NOUN
elm-5391	145	9	results	result	NOUN
elm-5391	145	10	from	from	ADP
elm-5391	145	11	the	the	DET
elm-5391	145	12	linear	linear	PROPN
elm-5391	145	13	classifier	classifier	NOUN
elm-5391	145	14	probe	probe	NOUN
elm-5391	145	15	.	.	PUNCT
elm-5391	146	1	the	the	DET
elm-5391	146	2	training	training	NOUN
elm-5391	146	3	covariate	covariate	NOUN
elm-5391	146	4	matrix	matrix	NOUN
elm-5391	146	5	of	of	ADP
elm-5391	146	6	the	the	DET
elm-5391	146	7	classification	classification	NOUN
elm-5391	146	8	model	model	NOUN
elm-5391	146	9	was	be	AUX
elm-5391	146	10	each	each	DET
elm-5391	146	11	type	type	NOUN
elm-5391	146	12	of	of	ADP
elm-5391	146	13	word	word	NOUN
elm-5391	146	14	embeddings	embedding	NOUN
elm-5391	146	15	.	.	PUNCT
elm-5391	147	1	the	the	DET
elm-5391	147	2	response	response	NOUN
elm-5391	147	3	variables	variable	NOUN
elm-5391	147	4	were	be	AUX
elm-5391	147	5	the	the	DET
elm-5391	147	6	binary	binary	ADJ
elm-5391	147	7	categorical	categorical	ADJ
elm-5391	147	8	label	label	NOUN
elm-5391	147	9	representing	represent	VERB
elm-5391	147	10	lower	low	ADJ
elm-5391	147	11	or	or	CCONJ
elm-5391	147	12	higher	high	ADJ
elm-5391	147	13	range	range	NOUN
elm-5391	147	14	of	of	ADP
elm-5391	147	15	one	one	NUM
elm-5391	147	16	of	of	ADP
elm-5391	147	17	the	the	DET
elm-5391	147	18	vad	vad	PROPN
elm-5391	147	19	dimensions	dimensions	PROPN
elm-5391	147	20	.	.	PUNCT
elm-5391	148	1	the	the	DET
elm-5391	148	2	conversion	conversion	NOUN
elm-5391	148	3	was	be	AUX
elm-5391	148	4	based	base	VERB
elm-5391	148	5	on	on	ADP
elm-5391	148	6	a	a	DET
elm-5391	148	7	0.5	0.5	NUM
elm-5391	148	8	threshold	threshold	NOUN
elm-5391	148	9	.	.	PUNCT
elm-5391	149	1	each	each	DET
elm-5391	149	2	type	type	NOUN
elm-5391	149	3	of	of	ADP
elm-5391	149	4	word	word	NOUN
elm-5391	149	5	embedding	embed	VERB
elm-5391	149	6	with	with	ADP
elm-5391	149	7	each	each	DET
elm-5391	149	8	affect	affect	NOUN
elm-5391	149	9	dimension	dimension	NOUN
elm-5391	149	10	constituted	constitute	VERB
elm-5391	149	11	a	a	DET
elm-5391	149	12	linear	linear	ADJ
elm-5391	149	13	classifier	classifier	NOUN
elm-5391	149	14	and	and	CCONJ
elm-5391	149	15	together	together	ADV
elm-5391	149	16	we	we	PRON
elm-5391	149	17	constructed	construct	VERB
elm-5391	149	18	12	12	NUM
elm-5391	149	19	classifiers	classifier	NOUN
elm-5391	149	20	.	.	PUNCT
elm-5391	150	1	we	we	PRON
elm-5391	150	2	used	use	VERB
elm-5391	150	3	the	the	DET
elm-5391	150	4	nrc	nrc	PROPN
elm-5391	150	5	vad	vad	NOUN
elm-5391	150	6	vocabulary	vocabulary	PROPN
elm-5391	150	7	as	as	ADP
elm-5391	150	8	the	the	DET
elm-5391	150	9	training	training	NOUN
elm-5391	150	10	and	and	CCONJ
elm-5391	150	11	the	the	DET
elm-5391	150	12	validation	validation	NOUN
elm-5391	150	13	data	datum	NOUN
elm-5391	150	14	(	(	PUNCT
elm-5391	150	15	with	with	ADP
elm-5391	150	16	a	a	DET
elm-5391	150	17	70/30	70/30	NUM
elm-5391	150	18	split	split	NOUN
elm-5391	150	19	)	)	PUNCT
elm-5391	150	20	.	.	PUNCT
elm-5391	151	1	the	the	DET
elm-5391	151	2	test	test	NOUN
elm-5391	151	3	data	datum	NOUN
elm-5391	151	4	comprised	comprise	VERB
elm-5391	151	5	of	of	ADP
elm-5391	151	6	130	130	NUM
elm-5391	151	7	word	word	NOUN
elm-5391	151	8	with	with	ADP
elm-5391	151	9	strong	strong	ADJ
elm-5391	151	10	affect	affect	VERB
elm-5391	151	11	information	information	NOUN
elm-5391	151	12	randomly	randomly	ADV
elm-5391	151	13	extracted	extract	VERB
elm-5391	151	14	from	from	ADP
elm-5391	151	15	shaver	shaver	PROPN
elm-5391	151	16	et	et	PROPN
elm-5391	151	17	al	al	PROPN
elm-5391	151	18	.	.	PROPN
elm-5391	151	19	(	(	PUNCT
elm-5391	151	20	1987	1987	NUM
elm-5391	151	21	)	)	PUNCT
elm-5391	151	22	.	.	PUNCT
elm-5391	152	1	we	we	PRON
elm-5391	152	2	compared	compare	VERB
elm-5391	152	3	the	the	DET
elm-5391	152	4	performance	performance	NOUN
elm-5391	152	5	of	of	ADP
elm-5391	152	6	glove	glove	NOUN
elm-5391	152	7	,	,	PUNCT
elm-5391	152	8	base	base	NOUN
elm-5391	152	9	bert	bert	PROPN
elm-5391	152	10	,	,	PUNCT
elm-5391	152	11	goemotion	goemotion	NOUN
elm-5391	152	12	bert	bert	PROPN
elm-5391	152	13	,	,	PUNCT
elm-5391	152	14	and	and	CCONJ
elm-5391	152	15	contextualized	contextualize	VERB
elm-5391	152	16	bert	bert	NOUN
elm-5391	152	17	embeddings	embedding	NOUN
elm-5391	152	18	on	on	ADP
elm-5391	152	19	each	each	PRON
elm-5391	152	20	of	of	ADP
elm-5391	152	21	the	the	DET
elm-5391	152	22	three	three	NUM
elm-5391	152	23	affect	affect	NOUN
elm-5391	152	24	dimensions	dimension	NOUN
elm-5391	152	25	and	and	CCONJ
elm-5391	152	26	present	present	VERB
elm-5391	152	27	the	the	DET
elm-5391	152	28	result	result	NOUN
elm-5391	152	29	in	in	ADP
elm-5391	152	30	table	table	NOUN
elm-5391	152	31	4	4	NUM
elm-5391	152	32	.	.	PUNCT
elm-5391	153	1	it	it	PRON
elm-5391	153	2	is	be	AUX
elm-5391	153	3	clear	clear	ADJ
elm-5391	153	4	that	that	SCONJ
elm-5391	153	5	the	the	DET
elm-5391	153	6	contexualized	contexualize	VERB
elm-5391	153	7	bert	bert	NOUN
elm-5391	153	8	embedding	embed	VERB
elm-5391	153	9	shows	show	VERB
elm-5391	153	10	superior	superior	ADJ
elm-5391	153	11	performance	performance	NOUN
elm-5391	153	12	compared	compare	VERB
elm-5391	153	13	to	to	ADP
elm-5391	153	14	the	the	DET
elm-5391	153	15	others	other	NOUN
elm-5391	153	16	.	.	PUNCT
elm-5391	154	1	to	to	PART
elm-5391	154	2	be	be	AUX
elm-5391	154	3	noted	note	VERB
elm-5391	154	4	,	,	PUNCT
elm-5391	154	5	our	our	PRON
elm-5391	154	6	validation	validation	NOUN
elm-5391	154	7	set	set	NOUN
elm-5391	154	8	included	include	VERB
elm-5391	154	9	many	many	ADJ
elm-5391	154	10	words	word	NOUN
elm-5391	154	11	whose	whose	DET
elm-5391	154	12	vad	vad	NOUN
elm-5391	154	13	score	score	NOUN
elm-5391	154	14	might	might	AUX
elm-5391	154	15	not	not	PART
elm-5391	154	16	be	be	AUX
elm-5391	154	17	as	as	ADP
elm-5391	154	18	salient	salient	NOUN
elm-5391	154	19	and	and	CCONJ
elm-5391	154	20	intuitive	intuitive	ADJ
elm-5391	154	21	as	as	ADP
elm-5391	154	22	the	the	DET
elm-5391	154	23	ones	one	NOUN
elm-5391	154	24	in	in	ADP
elm-5391	154	25	the	the	DET
elm-5391	154	26	test	test	NOUN
elm-5391	154	27	set	set	NOUN
elm-5391	154	28	.	.	PUNCT
elm-5391	155	1	this	this	PRON
elm-5391	155	2	might	might	AUX
elm-5391	155	3	explain	explain	VERB
elm-5391	155	4	that	that	SCONJ
elm-5391	155	5	the	the	DET
elm-5391	155	6	validation	validation	NOUN
elm-5391	155	7	set	set	NOUN
elm-5391	155	8	received	receive	VERB
elm-5391	155	9	lower	low	ADJ
elm-5391	155	10	accuracy	accuracy	NOUN
elm-5391	155	11	score	score	NOUN
elm-5391	155	12	than	than	ADP
elm-5391	155	13	the	the	DET
elm-5391	155	14	test	test	NOUN
elm-5391	155	15	sample	sample	NOUN
elm-5391	155	16	.	.	PUNCT
elm-5391	156	1	besides	besides	SCONJ
elm-5391	156	2	,	,	PUNCT
elm-5391	156	3	we	we	PRON
elm-5391	156	4	observed	observe	VERB
elm-5391	156	5	that	that	SCONJ
elm-5391	156	6	all	all	PRON
elm-5391	156	7	of	of	ADP
elm-5391	156	8	the	the	DET
elm-5391	156	9	model	model	NOUN
elm-5391	156	10	embeddings	embedding	NOUN
elm-5391	156	11	are	be	AUX
elm-5391	156	12	best	good	ADJ
elm-5391	156	13	at	at	ADP
elm-5391	156	14	predicting	predict	VERB
elm-5391	156	15	the	the	DET
elm-5391	156	16	valence	valence	NOUN
elm-5391	156	17	dimension	dimension	NOUN
elm-5391	156	18	compared	compare	VERB
elm-5391	156	19	with	with	ADP
elm-5391	156	20	arousal	arousal	NOUN
elm-5391	156	21	and	and	CCONJ
elm-5391	156	22	dominance	dominance	NOUN
elm-5391	156	23	.	.	PUNCT
elm-5391	157	1	this	this	PRON
elm-5391	157	2	matches	match	VERB
elm-5391	157	3	our	our	PRON
elm-5391	157	4	prior	prior	ADJ
elm-5391	157	5	knowledge	knowledge	NOUN
elm-5391	157	6	because	because	SCONJ
elm-5391	157	7	valence	valence	NOUN
elm-5391	157	8	represents	represent	VERB
elm-5391	157	9	the	the	DET
elm-5391	157	10	positive	positive	ADJ
elm-5391	157	11	or	or	CCONJ
elm-5391	157	12	negative	negative	ADJ
elm-5391	157	13	of	of	ADP
elm-5391	157	14	the	the	DET
elm-5391	157	15	words	word	NOUN
elm-5391	157	16	and	and	CCONJ
elm-5391	157	17	carries	carry	VERB
elm-5391	157	18	the	the	DET
elm-5391	157	19	most	most	ADV
elm-5391	157	20	straightforward	straightforward	ADJ
elm-5391	157	21	meaning	meaning	NOUN
elm-5391	157	22	for	for	SCONJ
elm-5391	157	23	the	the	DET
elm-5391	157	24	context	context	NOUN
elm-5391	157	25	to	to	PART
elm-5391	157	26	capture	capture	VERB
elm-5391	157	27	.	.	PUNCT
elm-5391	158	1	but	but	CCONJ
elm-5391	158	2	arousal	arousal	NOUN
elm-5391	158	3	and	and	CCONJ
elm-5391	158	4	dominance	dominance	NOUN
elm-5391	158	5	will	will	AUX
elm-5391	158	6	be	be	AUX
elm-5391	158	7	more	more	ADV
elm-5391	158	8	complicated	complicated	ADJ
elm-5391	158	9	and	and	CCONJ
elm-5391	158	10	subtle	subtle	ADJ
elm-5391	158	11	to	to	PART
elm-5391	158	12	capture	capture	VERB
elm-5391	158	13	depending	depend	VERB
elm-5391	158	14	on	on	ADP
elm-5391	158	15	its	its	PRON
elm-5391	158	16	distributional	distributional	ADJ
elm-5391	158	17	semantics	semantic	NOUN
elm-5391	158	18	.	.	PUNCT
elm-5391	159	1	interestingly	interestingly	ADV
elm-5391	159	2	,	,	PUNCT
elm-5391	159	3	goemotion	goemotion	NOUN
elm-5391	159	4	bert	bert	PROPN
elm-5391	159	5	does	do	AUX
elm-5391	159	6	not	not	PART
elm-5391	159	7	perform	perform	VERB
elm-5391	159	8	very	very	ADV
elm-5391	159	9	well	well	ADV
elm-5391	159	10	in	in	ADP
elm-5391	159	11	the	the	DET
elm-5391	159	12	dimensions	dimension	NOUN
elm-5391	159	13	of	of	ADP
elm-5391	159	14	arousal	arousal	NOUN
elm-5391	159	15	and	and	CCONJ
elm-5391	159	16	dominance	dominance	NOUN
elm-5391	159	17	.	.	PUNCT
elm-5391	160	1	this	this	DET
elm-5391	160	2	phenomenon	phenomenon	NOUN
elm-5391	160	3	leaves	leave	VERB
elm-5391	160	4	ample	ample	ADJ
elm-5391	160	5	space	space	NOUN
elm-5391	160	6	to	to	PART
elm-5391	160	7	investigate	investigate	VERB
elm-5391	160	8	further	far	ADV
elm-5391	160	9	how	how	SCONJ
elm-5391	160	10	word	word	NOUN
elm-5391	160	11	embeddings	embedding	NOUN
elm-5391	160	12	are	be	AUX
elm-5391	160	13	trained	train	VERB
elm-5391	160	14	to	to	PART
elm-5391	160	15	capture	capture	VERB
elm-5391	160	16	certain	certain	ADJ
elm-5391	160	17	meaning	meaning	NOUN
elm-5391	160	18	but	but	CCONJ
elm-5391	160	19	not	not	PART
elm-5391	160	20	the	the	DET
elm-5391	160	21	others	other	NOUN
elm-5391	160	22	.	.	PUNCT
elm-5391	161	1	proceedings	proceeding	NOUN
elm-5391	161	2	of	of	ADP
elm-5391	161	3	elm	elm	NOUN
elm-5391	161	4	2	2	NUM
elm-5391	161	5	:	:	PUNCT
elm-5391	161	6	310	310	NUM
elm-5391	161	7	-	-	SYM
elm-5391	161	8	321	321	NUM
elm-5391	161	9	,	,	PUNCT
elm-5391	161	10	2023	2023	NUM
elm-5391	161	11	yuhan	yuhan	PROPN
elm-5391	161	12	zhang	zhang	PROPN
elm-5391	161	13	,	,	PUNCT
elm-5391	161	14	wenqi	wenqi	PROPN
elm-5391	161	15	chen	chen	PROPN
elm-5391	161	16	,	,	PUNCT
elm-5391	161	17	ruihan	ruihan	PROPN
elm-5391	161	18	zhang	zhang	PROPN
elm-5391	161	19	and	and	CCONJ
elm-5391	161	20	xiajie	xiajie	PROPN
elm-5391	161	21	zhang	zhang	PROPN
elm-5391	161	22	:	:	PUNCT
elm-5391	161	23	representing	represent	VERB
elm-5391	161	24	affect	affect	VERB
elm-5391	161	25	information	information	NOUN
elm-5391	161	26	in	in	ADP
elm-5391	161	27	word	word	NOUN
elm-5391	161	28	embeddings	embedding	NOUN
elm-5391	161	29	.	.	PUNCT
elm-5391	162	1	316	316	NUM
elm-5391	162	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	162	3	https://www.elm-conference.net/	https://www.elm-conference.net/	ADP
elm-5391	162	4	embedding	embed	VERB
elm-5391	162	5	types	type	NOUN
elm-5391	162	6	validation	validation	NOUN
elm-5391	162	7	accuracy	accuracy	NOUN
elm-5391	162	8	affect	affect	VERB
elm-5391	162	9	word	word	NOUN
elm-5391	162	10	sample	sample	NOUN
elm-5391	162	11	accuracy	accuracy	NOUN
elm-5391	162	12	valence	valence	NOUN
elm-5391	162	13	arousal	arousal	NOUN
elm-5391	162	14	dominance	dominance	NOUN
elm-5391	162	15	valence	valence	NOUN
elm-5391	162	16	arousal	arousal	NOUN
elm-5391	162	17	dominance	dominance	NOUN
elm-5391	162	18	glove	glove	NOUN
elm-5391	162	19	0.75	0.75	NUM
elm-5391	162	20	0.70	0.70	NUM
elm-5391	162	21	0.73	0.73	NUM
elm-5391	162	22	0.84	0.84	NUM
elm-5391	162	23	0.74	0.74	NUM
elm-5391	162	24	0.75	0.75	NUM
elm-5391	162	25	base	base	NOUN
elm-5391	162	26	bert	bert	NOUN
elm-5391	162	27	0.76	0.76	NUM
elm-5391	162	28	0.73	0.73	NUM
elm-5391	162	29	0.74	0.74	NUM
elm-5391	162	30	0.93	0.93	NUM
elm-5391	162	31	0.85	0.85	NUM
elm-5391	162	32	0.82	0.82	NUM
elm-5391	162	33	goemotion	goemotion	NOUN
elm-5391	162	34	bert	bert	NOUN
elm-5391	162	35	0.68	0.68	NUM
elm-5391	162	36	0.68	0.68	NUM
elm-5391	162	37	0.70	0.70	NUM
elm-5391	162	38	0.92	0.92	NUM
elm-5391	162	39	0.76	0.76	NUM
elm-5391	162	40	0.76	0.76	NUM
elm-5391	162	41	contextualized	contextualized	ADJ
elm-5391	162	42	bert	bert	NOUN
elm-5391	162	43	0.85	0.85	NUM
elm-5391	162	44	0.77	0.77	NUM
elm-5391	162	45	0.85	0.85	NUM
elm-5391	162	46	0.95	0.95	NUM
elm-5391	162	47	0.88	0.88	NUM
elm-5391	162	48	0.90	0.90	NUM
elm-5391	162	49	table	table	NOUN
elm-5391	162	50	4	4	NUM
elm-5391	162	51	:	:	PUNCT
elm-5391	162	52	performance	performance	NOUN
elm-5391	162	53	of	of	ADP
elm-5391	162	54	different	different	ADJ
elm-5391	162	55	word	word	NOUN
elm-5391	162	56	embeddings	embedding	NOUN
elm-5391	162	57	in	in	ADP
elm-5391	162	58	predicting	predict	VERB
elm-5391	162	59	vad	vad	PROPN
elm-5391	162	60	labels	label	NOUN
elm-5391	162	61	.	.	PUNCT
elm-5391	163	1	5	5	X
elm-5391	163	2	.	.	X
elm-5391	163	3	discussion	discussion	NOUN
elm-5391	163	4	&	&	CCONJ
elm-5391	163	5	conclusion	conclusion	NOUN
elm-5391	163	6	.	.	PUNCT
elm-5391	164	1	in	in	ADP
elm-5391	164	2	this	this	DET
elm-5391	164	3	work	work	NOUN
elm-5391	164	4	,	,	PUNCT
elm-5391	164	5	we	we	PRON
elm-5391	164	6	investigated	investigate	VERB
elm-5391	164	7	the	the	DET
elm-5391	164	8	capability	capability	NOUN
elm-5391	164	9	of	of	ADP
elm-5391	164	10	word	word	NOUN
elm-5391	164	11	embeddings	embedding	NOUN
elm-5391	164	12	to	to	PART
elm-5391	164	13	capture	capture	VERB
elm-5391	164	14	affect	affect	VERB
elm-5391	164	15	meaning	meaning	NOUN
elm-5391	164	16	.	.	PUNCT
elm-5391	165	1	by	by	ADP
elm-5391	165	2	combining	combine	VERB
elm-5391	165	3	results	result	NOUN
elm-5391	165	4	from	from	ADP
elm-5391	165	5	unsupervised	unsupervised	ADJ
elm-5391	165	6	and	and	CCONJ
elm-5391	165	7	supervised	supervised	ADJ
elm-5391	165	8	learning	learning	NOUN
elm-5391	165	9	,	,	PUNCT
elm-5391	165	10	we	we	PRON
elm-5391	165	11	found	find	VERB
elm-5391	165	12	that	that	SCONJ
elm-5391	165	13	our	our	PRON
elm-5391	165	14	contextualized	contextualized	ADJ
elm-5391	165	15	bert	bert	NOUN
elm-5391	165	16	embedding	embed	VERB
elm-5391	165	17	,	,	PUNCT
elm-5391	165	18	where	where	SCONJ
elm-5391	165	19	the	the	DET
elm-5391	165	20	representation	representation	NOUN
elm-5391	165	21	of	of	ADP
elm-5391	165	22	each	each	DET
elm-5391	165	23	word	word	NOUN
elm-5391	165	24	is	be	AUX
elm-5391	165	25	the	the	DET
elm-5391	165	26	first	first	ADJ
elm-5391	165	27	principal	principal	ADJ
elm-5391	165	28	component	component	NOUN
elm-5391	165	29	of	of	ADP
elm-5391	165	30	all	all	DET
elm-5391	165	31	embeddings	embedding	NOUN
elm-5391	165	32	of	of	ADP
elm-5391	165	33	the	the	DET
elm-5391	165	34	same	same	ADJ
elm-5391	165	35	word	word	NOUN
elm-5391	165	36	in	in	ADP
elm-5391	165	37	its	its	PRON
elm-5391	165	38	multiple	multiple	ADJ
elm-5391	165	39	occurrences	occurrence	NOUN
elm-5391	165	40	in	in	ADP
elm-5391	165	41	the	the	DET
elm-5391	165	42	imdb	imdb	PROPN
elm-5391	165	43	dataset	dataset	PROPN
elm-5391	165	44	.	.	PUNCT
elm-5391	166	1	in	in	ADP
elm-5391	166	2	comparison	comparison	NOUN
elm-5391	166	3	,	,	PUNCT
elm-5391	166	4	the	the	DET
elm-5391	166	5	base	base	NOUN
elm-5391	166	6	bert	bert	PROPN
elm-5391	166	7	does	do	AUX
elm-5391	166	8	not	not	PART
elm-5391	166	9	show	show	VERB
elm-5391	166	10	prominent	prominent	ADJ
elm-5391	166	11	encoding	encoding	NOUN
elm-5391	166	12	of	of	ADP
elm-5391	166	13	affect	affect	VERB
elm-5391	166	14	meaning	meaning	NOUN
elm-5391	166	15	and	and	CCONJ
elm-5391	166	16	sometimes	sometimes	ADV
elm-5391	166	17	is	be	AUX
elm-5391	166	18	even	even	ADV
elm-5391	166	19	inferior	inferior	ADJ
elm-5391	166	20	to	to	ADP
elm-5391	166	21	the	the	DET
elm-5391	166	22	static	static	ADJ
elm-5391	166	23	glove	glove	NOUN
elm-5391	166	24	.	.	PUNCT
elm-5391	167	1	the	the	DET
elm-5391	167	2	goemotion	goemotion	NOUN
elm-5391	167	3	bert	bert	NOUN
elm-5391	167	4	embedding	embed	VERB
elm-5391	167	5	also	also	ADV
elm-5391	167	6	does	do	AUX
elm-5391	167	7	not	not	PART
elm-5391	167	8	perform	perform	VERB
elm-5391	167	9	as	as	ADV
elm-5391	167	10	well	well	ADV
elm-5391	167	11	as	as	ADP
elm-5391	167	12	the	the	DET
elm-5391	167	13	contextualized	contextualized	ADJ
elm-5391	167	14	bert	bert	NOUN
elm-5391	167	15	embedding	embed	VERB
elm-5391	167	16	.	.	PUNCT
elm-5391	168	1	out	out	ADP
elm-5391	168	2	of	of	ADP
elm-5391	168	3	the	the	DET
elm-5391	168	4	vad	vad	PROPN
elm-5391	168	5	dimensions	dimensions	PROPN
elm-5391	168	6	,	,	PUNCT
elm-5391	168	7	it	it	PRON
elm-5391	168	8	is	be	AUX
elm-5391	168	9	hard	hard	ADJ
elm-5391	168	10	to	to	PART
elm-5391	168	11	know	know	VERB
elm-5391	168	12	which	which	DET
elm-5391	168	13	dimension	dimension	NOUN
elm-5391	168	14	is	be	AUX
elm-5391	168	15	relatively	relatively	ADV
elm-5391	168	16	well	well	ADV
elm-5391	168	17	represented	represent	VERB
elm-5391	168	18	or	or	CCONJ
elm-5391	168	19	easy	easy	ADJ
elm-5391	168	20	to	to	PART
elm-5391	168	21	capture	capture	VERB
elm-5391	168	22	:	:	PUNCT
elm-5391	168	23	based	base	VERB
elm-5391	168	24	on	on	ADP
elm-5391	168	25	the	the	DET
elm-5391	168	26	pca	pca	NOUN
elm-5391	168	27	results	result	NOUN
elm-5391	168	28	,	,	PUNCT
elm-5391	168	29	all	all	DET
elm-5391	168	30	vad	vad	PROPN
elm-5391	168	31	dimensions	dimension	NOUN
elm-5391	168	32	could	could	AUX
elm-5391	168	33	attain	attain	VERB
elm-5391	168	34	a	a	DET
elm-5391	168	35	correlation	correlation	NOUN
elm-5391	168	36	higher	high	ADJ
elm-5391	168	37	than	than	ADP
elm-5391	168	38	0.2	0.2	NUM
elm-5391	168	39	in	in	ADP
elm-5391	168	40	some	some	DET
elm-5391	168	41	embedding	embed	VERB
elm-5391	168	42	representations	representation	NOUN
elm-5391	168	43	;	;	PUNCT
elm-5391	168	44	based	base	VERB
elm-5391	168	45	on	on	ADP
elm-5391	168	46	the	the	DET
elm-5391	168	47	linear	linear	ADJ
elm-5391	168	48	classifier	classifier	NOUN
elm-5391	168	49	results	result	NOUN
elm-5391	168	50	,	,	PUNCT
elm-5391	168	51	valence	valence	NOUN
elm-5391	168	52	seems	seem	VERB
elm-5391	168	53	to	to	PART
elm-5391	168	54	be	be	AUX
elm-5391	168	55	better	well	ADV
elm-5391	168	56	represented	represent	VERB
elm-5391	168	57	than	than	ADP
elm-5391	168	58	the	the	DET
elm-5391	168	59	other	other	ADJ
elm-5391	168	60	two	two	NUM
elm-5391	168	61	.	.	PUNCT
elm-5391	169	1	the	the	DET
elm-5391	169	2	discrepant	discrepant	ADJ
elm-5391	169	3	pattern	pattern	NOUN
elm-5391	169	4	might	might	AUX
elm-5391	169	5	result	result	VERB
elm-5391	169	6	from	from	ADP
elm-5391	169	7	the	the	DET
elm-5391	169	8	different	different	ADJ
elm-5391	169	9	sample	sample	NOUN
elm-5391	169	10	sizes	size	NOUN
elm-5391	169	11	used	use	VERB
elm-5391	169	12	in	in	ADP
elm-5391	169	13	these	these	DET
elm-5391	169	14	tasks	task	NOUN
elm-5391	169	15	(	(	PUNCT
elm-5391	169	16	i.e.	i.e.	X
elm-5391	169	17	,	,	PUNCT
elm-5391	169	18	5,586	5,586	NUM
elm-5391	169	19	words	word	NOUN
elm-5391	169	20	for	for	ADP
elm-5391	169	21	pca	pca	NOUN
elm-5391	169	22	and	and	CCONJ
elm-5391	169	23	130	130	NUM
elm-5391	169	24	words	word	NOUN
elm-5391	169	25	as	as	ADP
elm-5391	169	26	the	the	DET
elm-5391	169	27	test	test	NOUN
elm-5391	169	28	set	set	VERB
elm-5391	169	29	for	for	ADP
elm-5391	169	30	the	the	DET
elm-5391	169	31	linear	linear	PROPN
elm-5391	169	32	classifiers	classifier	NOUN
elm-5391	169	33	)	)	PUNCT
elm-5391	169	34	and	and	CCONJ
elm-5391	169	35	more	more	ADV
elm-5391	169	36	controlled	control	VERB
elm-5391	169	37	experiments	experiment	NOUN
elm-5391	169	38	are	be	AUX
elm-5391	169	39	needed	need	VERB
elm-5391	169	40	to	to	PART
elm-5391	169	41	pin	pin	VERB
elm-5391	169	42	down	down	ADP
elm-5391	169	43	the	the	DET
elm-5391	169	44	fact	fact	NOUN
elm-5391	169	45	.	.	PUNCT
elm-5391	170	1	5.1	5.1	NUM
elm-5391	170	2	.	.	PUNCT
elm-5391	170	3	implications	implication	NOUN
elm-5391	170	4	.	.	PUNCT
elm-5391	171	1	here	here	ADV
elm-5391	171	2	we	we	PRON
elm-5391	171	3	provide	provide	VERB
elm-5391	171	4	a	a	DET
elm-5391	171	5	novel	novel	ADJ
elm-5391	171	6	aspect	aspect	NOUN
elm-5391	171	7	of	of	ADP
elm-5391	171	8	testing	testing	NOUN
elm-5391	171	9	word	word	NOUN
elm-5391	171	10	embedding	embed	VERB
elm-5391	171	11	capabilities	capability	NOUN
elm-5391	171	12	that	that	PRON
elm-5391	171	13	involve	involve	VERB
elm-5391	171	14	intricate	intricate	ADJ
elm-5391	171	15	emotion	emotion	NOUN
elm-5391	171	16	and	and	CCONJ
elm-5391	171	17	affect	affect	VERB
elm-5391	171	18	related	related	ADJ
elm-5391	171	19	meaning	meaning	NOUN
elm-5391	171	20	.	.	PUNCT
elm-5391	172	1	based	base	VERB
elm-5391	172	2	on	on	ADP
elm-5391	172	3	what	what	PRON
elm-5391	172	4	lexical	lexical	ADJ
elm-5391	172	5	and	and	CCONJ
elm-5391	172	6	linguistic	linguistic	ADJ
elm-5391	172	7	knowledge	knowledge	NOUN
elm-5391	172	8	word	word	NOUN
elm-5391	172	9	embeddings	embedding	NOUN
elm-5391	172	10	can	can	AUX
elm-5391	172	11	and	and	CCONJ
elm-5391	172	12	can	can	AUX
elm-5391	172	13	not	not	PART
elm-5391	172	14	capture	capture	VERB
elm-5391	172	15	,	,	PUNCT
elm-5391	172	16	especially	especially	ADV
elm-5391	172	17	for	for	ADP
elm-5391	172	18	the	the	DET
elm-5391	172	19	renowned	renowned	ADJ
elm-5391	172	20	bert	bert	NOUN
elm-5391	172	21	embedding	embed	VERB
elm-5391	172	22	(	(	PUNCT
elm-5391	172	23	ettinger	ettinger	PROPN
elm-5391	172	24	2020	2020	NUM
elm-5391	172	25	,	,	PUNCT
elm-5391	172	26	klafka	klafka	VERB
elm-5391	172	27	&	&	CCONJ
elm-5391	172	28	ettinger	ettinger	PROPN
elm-5391	172	29	2020	2020	NUM
elm-5391	172	30	,	,	PUNCT
elm-5391	172	31	manning	man	VERB
elm-5391	172	32	et	et	PROPN
elm-5391	172	33	al	al	PROPN
elm-5391	172	34	.	.	PROPN
elm-5391	172	35	2020	2020	NUM
elm-5391	172	36	,	,	PUNCT
elm-5391	172	37	pandia	pandia	PROPN
elm-5391	172	38	et	et	NOUN
elm-5391	172	39	al	al	PROPN
elm-5391	172	40	.	.	PROPN
elm-5391	172	41	2021	2021	NUM
elm-5391	172	42	)	)	PUNCT
elm-5391	172	43	,	,	PUNCT
elm-5391	172	44	we	we	PRON
elm-5391	172	45	add	add	VERB
elm-5391	172	46	one	one	NUM
elm-5391	172	47	additional	additional	ADJ
elm-5391	172	48	piece	piece	NOUN
elm-5391	172	49	of	of	ADP
elm-5391	172	50	information	information	NOUN
elm-5391	172	51	.	.	PUNCT
elm-5391	173	1	the	the	DET
elm-5391	173	2	relatively	relatively	ADV
elm-5391	173	3	poor	poor	ADJ
elm-5391	173	4	encoding	encoding	NOUN
elm-5391	173	5	of	of	ADP
elm-5391	173	6	affect	affect	VERB
elm-5391	173	7	information	information	NOUN
elm-5391	173	8	in	in	ADP
elm-5391	173	9	bert	bert	PROPN
elm-5391	173	10	might	might	AUX
elm-5391	173	11	suggest	suggest	VERB
elm-5391	173	12	these	these	DET
elm-5391	173	13	lines	line	NOUN
elm-5391	173	14	of	of	ADP
elm-5391	173	15	implications	implication	NOUN
elm-5391	173	16	:	:	PUNCT
elm-5391	173	17	(	(	PUNCT
elm-5391	173	18	1	1	X
elm-5391	173	19	)	)	PUNCT
elm-5391	173	20	the	the	PRON
elm-5391	173	21	affect	affect	VERB
elm-5391	173	22	information	information	NOUN
elm-5391	173	23	of	of	ADP
elm-5391	173	24	english	english	ADJ
elm-5391	173	25	lexicon	lexicon	NOUN
elm-5391	173	26	is	be	AUX
elm-5391	173	27	by	by	ADP
elm-5391	173	28	nature	nature	NOUN
elm-5391	173	29	implicit	implicit	ADJ
elm-5391	173	30	and	and	CCONJ
elm-5391	173	31	hard	hard	ADJ
elm-5391	173	32	to	to	PART
elm-5391	173	33	capture	capture	VERB
elm-5391	173	34	from	from	ADP
elm-5391	173	35	the	the	DET
elm-5391	173	36	distributional	distributional	ADJ
elm-5391	173	37	features	feature	NOUN
elm-5391	173	38	of	of	ADP
elm-5391	173	39	words	word	NOUN
elm-5391	173	40	in	in	ADP
elm-5391	173	41	huge	huge	ADJ
elm-5391	173	42	corpora	corpus	NOUN
elm-5391	173	43	;	;	PUNCT
elm-5391	173	44	(	(	PUNCT
elm-5391	173	45	2	2	X
elm-5391	173	46	)	)	PUNCT
elm-5391	173	47	for	for	ADP
elm-5391	173	48	sentiment	sentiment	NOUN
elm-5391	173	49	related	relate	VERB
elm-5391	173	50	nlp	nlp	NOUN
elm-5391	173	51	tasks	task	NOUN
elm-5391	173	52	or	or	CCONJ
elm-5391	173	53	affect	affect	VERB
elm-5391	173	54	computing	computing	NOUN
elm-5391	173	55	tasks	task	NOUN
elm-5391	173	56	(	(	PUNCT
elm-5391	173	57	maas	maas	PROPN
elm-5391	173	58	et	et	PROPN
elm-5391	173	59	al	al	PROPN
elm-5391	173	60	.	.	PROPN
elm-5391	173	61	2011	2011	NUM
elm-5391	173	62	,	,	PUNCT
elm-5391	173	63	socher	socher	PROPN
elm-5391	173	64	et	et	PROPN
elm-5391	173	65	al	al	PROPN
elm-5391	173	66	.	.	PROPN
elm-5391	173	67	2013	2013	NUM
elm-5391	173	68	,	,	PUNCT
elm-5391	173	69	zhang	zhang	PROPN
elm-5391	173	70	et	et	PROPN
elm-5391	173	71	al	al	PROPN
elm-5391	173	72	.	.	PROPN
elm-5391	173	73	2015	2015	NUM
elm-5391	173	74	)	)	PUNCT
elm-5391	173	75	,	,	PUNCT
elm-5391	173	76	we	we	PRON
elm-5391	173	77	might	might	AUX
elm-5391	173	78	consider	consider	VERB
elm-5391	173	79	using	use	VERB
elm-5391	173	80	affect	affect	VERB
elm-5391	173	81	enriched	enrich	VERB
elm-5391	173	82	embeddings	embedding	NOUN
elm-5391	173	83	or	or	CCONJ
elm-5391	173	84	leveraging	leverage	VERB
elm-5391	173	85	additional	additional	ADJ
elm-5391	173	86	emotional	emotional	ADJ
elm-5391	173	87	capabilities	capability	NOUN
elm-5391	173	88	of	of	ADP
elm-5391	173	89	lms	lm	NOUN
elm-5391	173	90	from	from	ADP
elm-5391	173	91	transfer	transfer	NOUN
elm-5391	173	92	learning	learning	NOUN
elm-5391	173	93	.	.	PUNCT
elm-5391	174	1	5.2	5.2	NUM
elm-5391	174	2	.	.	PUNCT
elm-5391	174	3	limitations	limitation	NOUN
elm-5391	174	4	&	&	CCONJ
elm-5391	174	5	future	future	ADJ
elm-5391	174	6	work	work	NOUN
elm-5391	174	7	.	.	PUNCT
elm-5391	175	1	as	as	ADV
elm-5391	175	2	far	far	ADV
elm-5391	175	3	as	as	SCONJ
elm-5391	175	4	we	we	PRON
elm-5391	175	5	know	know	VERB
elm-5391	175	6	,	,	PUNCT
elm-5391	175	7	this	this	PRON
elm-5391	175	8	is	be	AUX
elm-5391	175	9	the	the	DET
elm-5391	175	10	first	first	ADJ
elm-5391	175	11	attempt	attempt	NOUN
elm-5391	175	12	in	in	ADP
elm-5391	175	13	the	the	DET
elm-5391	175	14	nlp	nlp	NOUN
elm-5391	175	15	community	community	NOUN
elm-5391	175	16	to	to	PART
elm-5391	175	17	investigate	investigate	VERB
elm-5391	175	18	the	the	DET
elm-5391	175	19	capability	capability	NOUN
elm-5391	175	20	of	of	ADP
elm-5391	175	21	word	word	NOUN
elm-5391	175	22	embeddings	embedding	NOUN
elm-5391	175	23	to	to	PART
elm-5391	175	24	encode	encode	VERB
elm-5391	175	25	affect	affect	VERB
elm-5391	175	26	information	information	NOUN
elm-5391	175	27	in	in	ADP
elm-5391	175	28	words	word	NOUN
elm-5391	175	29	.	.	PUNCT
elm-5391	176	1	we	we	PRON
elm-5391	176	2	lay	lay	VERB
elm-5391	176	3	three	three	NUM
elm-5391	176	4	lines	line	NOUN
elm-5391	176	5	of	of	ADP
elm-5391	176	6	research	research	NOUN
elm-5391	176	7	for	for	ADP
elm-5391	176	8	future	future	ADJ
elm-5391	176	9	investigations	investigation	NOUN
elm-5391	176	10	.	.	PUNCT
elm-5391	177	1	first	first	ADV
elm-5391	177	2	,	,	PUNCT
elm-5391	177	3	from	from	ADP
elm-5391	177	4	an	an	DET
elm-5391	177	5	algorithmic	algorithmic	ADJ
elm-5391	177	6	point	point	NOUN
elm-5391	177	7	of	of	ADP
elm-5391	177	8	view	view	NOUN
elm-5391	177	9	,	,	PUNCT
elm-5391	177	10	we	we	PRON
elm-5391	177	11	would	would	AUX
elm-5391	177	12	like	like	VERB
elm-5391	177	13	to	to	PART
elm-5391	177	14	ask	ask	VERB
elm-5391	177	15	what	what	PRON
elm-5391	177	16	specific	specific	ADJ
elm-5391	177	17	training	training	NOUN
elm-5391	177	18	mechanisms	mechanism	NOUN
elm-5391	177	19	enable	enable	VERB
elm-5391	177	20	word	word	NOUN
elm-5391	177	21	embeddings	embedding	NOUN
elm-5391	177	22	to	to	PART
elm-5391	177	23	capture	capture	VERB
elm-5391	177	24	certain	certain	ADJ
elm-5391	177	25	information	information	NOUN
elm-5391	177	26	,	,	PUNCT
elm-5391	177	27	especially	especially	ADV
elm-5391	177	28	those	those	DET
elm-5391	177	29	intricate	intricate	ADJ
elm-5391	177	30	meanings	meaning	NOUN
elm-5391	177	31	that	that	PRON
elm-5391	177	32	do	do	AUX
elm-5391	177	33	n’t	not	PART
elm-5391	177	34	rely	rely	VERB
elm-5391	177	35	on	on	ADP
elm-5391	177	36	co	co	NOUN
elm-5391	177	37	-	-	NOUN
elm-5391	177	38	occurrence	occurrence	ADJ
elm-5391	177	39	and	and	CCONJ
elm-5391	177	40	distributional	distributional	ADJ
elm-5391	177	41	semantics	semantic	NOUN
elm-5391	177	42	.	.	PUNCT
elm-5391	178	1	second	second	ADJ
elm-5391	178	2	,	,	PUNCT
elm-5391	178	3	talking	talk	VERB
elm-5391	178	4	about	about	ADP
elm-5391	178	5	meaning	meaning	NOUN
elm-5391	178	6	and	and	CCONJ
elm-5391	178	7	semantics	semantic	NOUN
elm-5391	178	8	in	in	ADP
elm-5391	178	9	a	a	DET
elm-5391	178	10	broader	broad	ADJ
elm-5391	178	11	sense	sense	NOUN
elm-5391	178	12	,	,	PUNCT
elm-5391	178	13	there	there	PRON
elm-5391	178	14	are	be	VERB
elm-5391	178	15	still	still	ADV
elm-5391	178	16	tons	ton	NOUN
elm-5391	178	17	of	of	ADP
elm-5391	178	18	unknowns	unknown	NOUN
elm-5391	178	19	that	that	PRON
elm-5391	178	20	fall	fall	VERB
elm-5391	178	21	outside	outside	ADV
elm-5391	178	22	of	of	ADP
elm-5391	178	23	the	the	DET
elm-5391	178	24	realm	realm	NOUN
elm-5391	178	25	of	of	ADP
elm-5391	178	26	formal	formal	ADJ
elm-5391	178	27	semantics	semantic	NOUN
elm-5391	178	28	or	or	CCONJ
elm-5391	178	29	compositionality	compositionality	NOUN
elm-5391	178	30	.	.	PUNCT
elm-5391	179	1	more	more	ADV
elm-5391	179	2	rigorous	rigorous	ADJ
elm-5391	179	3	identifications	identification	NOUN
elm-5391	179	4	and	and	CCONJ
elm-5391	179	5	descriptions	description	NOUN
elm-5391	179	6	of	of	ADP
elm-5391	179	7	different	different	ADJ
elm-5391	179	8	kinds	kind	NOUN
elm-5391	179	9	of	of	ADP
elm-5391	179	10	meaning	meaning	NOUN
elm-5391	179	11	are	be	AUX
elm-5391	179	12	needed	need	VERB
elm-5391	179	13	for	for	ADP
elm-5391	179	14	studying	study	VERB
elm-5391	179	15	the	the	DET
elm-5391	179	16	knowledge	knowledge	NOUN
elm-5391	179	17	of	of	ADP
elm-5391	179	18	meaning	meaning	NOUN
elm-5391	179	19	in	in	ADP
elm-5391	179	20	lms	lm	NOUN
elm-5391	179	21	.	.	PUNCT
elm-5391	180	1	third	third	ADJ
elm-5391	180	2	,	,	PUNCT
elm-5391	180	3	to	to	ADP
elm-5391	180	4	what	what	DET
elm-5391	180	5	extent	extent	NOUN
elm-5391	180	6	can	can	AUX
elm-5391	180	7	language	language	VERB
elm-5391	180	8	carry	carry	VERB
elm-5391	180	9	affective	affective	ADJ
elm-5391	180	10	information	information	NOUN
elm-5391	180	11	?	?	PUNCT
elm-5391	181	1	this	this	PRON
elm-5391	181	2	is	be	AUX
elm-5391	181	3	an	an	DET
elm-5391	181	4	essential	essential	ADJ
elm-5391	181	5	question	question	NOUN
elm-5391	181	6	in	in	ADP
elm-5391	181	7	affective	affective	ADJ
elm-5391	181	8	computing	computing	NOUN
elm-5391	181	9	(	(	PUNCT
elm-5391	181	10	picard	picard	PROPN
elm-5391	181	11	2000	2000	NUM
elm-5391	181	12	,	,	PUNCT
elm-5391	181	13	2003	2003	NUM
elm-5391	181	14	,	,	PUNCT
elm-5391	181	15	tao	tao	PROPN
elm-5391	181	16	&	&	CCONJ
elm-5391	181	17	tan	tan	PROPN
elm-5391	181	18	2005	2005	NUM
elm-5391	181	19	)	)	PUNCT
elm-5391	181	20	.	.	PUNCT
elm-5391	182	1	a	a	DET
elm-5391	182	2	better	well	ADJ
elm-5391	182	3	underproceedings	underproceeding	NOUN
elm-5391	182	4	of	of	ADP
elm-5391	182	5	elm	elm	NOUN
elm-5391	182	6	2	2	NUM
elm-5391	182	7	:	:	PUNCT
elm-5391	182	8	310	310	NUM
elm-5391	182	9	-	-	SYM
elm-5391	182	10	321	321	NUM
elm-5391	182	11	,	,	PUNCT
elm-5391	182	12	2023	2023	NUM
elm-5391	182	13	yuhan	yuhan	PROPN
elm-5391	182	14	zhang	zhang	PROPN
elm-5391	182	15	,	,	PUNCT
elm-5391	182	16	wenqi	wenqi	PROPN
elm-5391	182	17	chen	chen	PROPN
elm-5391	182	18	,	,	PUNCT
elm-5391	182	19	ruihan	ruihan	PROPN
elm-5391	182	20	zhang	zhang	PROPN
elm-5391	182	21	and	and	CCONJ
elm-5391	182	22	xiajie	xiajie	PROPN
elm-5391	182	23	zhang	zhang	PROPN
elm-5391	182	24	:	:	PUNCT
elm-5391	182	25	representing	represent	VERB
elm-5391	182	26	affect	affect	VERB
elm-5391	182	27	information	information	NOUN
elm-5391	182	28	in	in	ADP
elm-5391	182	29	word	word	NOUN
elm-5391	182	30	embeddings	embedding	NOUN
elm-5391	182	31	.	.	PUNCT
elm-5391	183	1	317	317	NUM
elm-5391	183	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	183	3	https://www.elm-conference.net/	https://www.elm-conference.net/	ADP
elm-5391	183	4	standing	stand	VERB
elm-5391	183	5	from	from	ADP
elm-5391	183	6	both	both	CCONJ
elm-5391	183	7	the	the	DET
elm-5391	183	8	algorithmic	algorithmic	ADJ
elm-5391	183	9	level	level	NOUN
elm-5391	183	10	and	and	CCONJ
elm-5391	183	11	the	the	DET
elm-5391	183	12	linguistic	linguistic	ADJ
elm-5391	183	13	level	level	NOUN
elm-5391	183	14	of	of	ADP
elm-5391	183	15	affect	affect	NOUN
elm-5391	183	16	encoding	encoding	NOUN
elm-5391	183	17	and	and	CCONJ
elm-5391	183	18	recognition	recognition	NOUN
elm-5391	183	19	would	would	AUX
elm-5391	183	20	provide	provide	VERB
elm-5391	183	21	valuable	valuable	ADJ
elm-5391	183	22	insights	insight	NOUN
elm-5391	183	23	into	into	ADP
elm-5391	183	24	this	this	DET
elm-5391	183	25	question	question	NOUN
elm-5391	183	26	.	.	PUNCT
elm-5391	184	1	references	reference	NOUN
elm-5391	184	2	alswaidan	alswaidan	PROPN
elm-5391	184	3	,	,	PUNCT
elm-5391	184	4	nourah	nourah	PROPN
elm-5391	184	5	&	&	CCONJ
elm-5391	184	6	mohamed	mohamed	PROPN
elm-5391	184	7	el	el	PROPN
elm-5391	184	8	bachir	bachir	PROPN
elm-5391	184	9	menai	menai	PROPN
elm-5391	184	10	.	.	PUNCT
elm-5391	185	1	2020	2020	NUM
elm-5391	185	2	.	.	PUNCT
elm-5391	186	1	a	a	DET
elm-5391	186	2	survey	survey	NOUN
elm-5391	186	3	of	of	ADP
elm-5391	186	4	state	state	NOUN
elm-5391	186	5	-	-	PUNCT
elm-5391	186	6	of	of	ADP
elm-5391	186	7	-	-	PUNCT
elm-5391	186	8	the	the	DET
elm-5391	186	9	-	-	PUNCT
elm-5391	186	10	art	art	NOUN
elm-5391	186	11	approaches	approach	NOUN
elm-5391	186	12	for	for	ADP
elm-5391	186	13	emotion	emotion	NOUN
elm-5391	186	14	recognition	recognition	NOUN
elm-5391	186	15	in	in	ADP
elm-5391	186	16	text	text	NOUN
elm-5391	186	17	.	.	PUNCT
elm-5391	187	1	knowledge	knowledge	NOUN
elm-5391	187	2	and	and	CCONJ
elm-5391	187	3	information	information	NOUN
elm-5391	187	4	systems	system	NOUN
elm-5391	187	5	62(8	62(8	NUM
elm-5391	187	6	)	)	PUNCT
elm-5391	187	7	.	.	PUNCT
elm-5391	188	1	2937–2987	2937–2987	NUM
elm-5391	188	2	.	.	PUNCT
elm-5391	189	1	bojanowski	bojanowski	NOUN
elm-5391	189	2	,	,	PUNCT
elm-5391	189	3	piotr	piotr	PROPN
elm-5391	189	4	,	,	PUNCT
elm-5391	189	5	edouard	edouard	NOUN
elm-5391	189	6	grave	grave	NOUN
elm-5391	189	7	,	,	PUNCT
elm-5391	189	8	armand	armand	PROPN
elm-5391	189	9	joulin	joulin	PROPN
elm-5391	189	10	&	&	CCONJ
elm-5391	189	11	tomas	tomas	PROPN
elm-5391	189	12	mikolov	mikolov	PROPN
elm-5391	189	13	.	.	PUNCT
elm-5391	190	1	2017	2017	NUM
elm-5391	190	2	.	.	PUNCT
elm-5391	191	1	enriching	enrich	VERB
elm-5391	191	2	word	word	NOUN
elm-5391	191	3	vectors	vector	NOUN
elm-5391	191	4	with	with	ADP
elm-5391	191	5	subword	subword	PROPN
elm-5391	191	6	information	information	NOUN
elm-5391	191	7	.	.	PUNCT
elm-5391	192	1	acl	acl	PROPN
elm-5391	192	2	transactions	transaction	NOUN
elm-5391	192	3	of	of	ADP
elm-5391	192	4	the	the	DET
elm-5391	192	5	association	association	NOUN
elm-5391	192	6	for	for	ADP
elm-5391	192	7	computational	computational	ADJ
elm-5391	192	8	linguistics	linguistic	NOUN
elm-5391	192	9	5	5	NUM
elm-5391	192	10	.	.	PUNCT
elm-5391	193	1	135–146	135–146	NUM
elm-5391	193	2	.	.	PUNCT
elm-5391	194	1	bradley	bradley	PROPN
elm-5391	194	2	,	,	PUNCT
elm-5391	194	3	margaret	margaret	PROPN
elm-5391	194	4	m	m	PROPN
elm-5391	194	5	&	&	CCONJ
elm-5391	194	6	peter	peter	PROPN
elm-5391	194	7	j	j	PROPN
elm-5391	194	8	lang	lang	PROPN
elm-5391	194	9	.	.	PROPN
elm-5391	194	10	1999	1999	NUM
elm-5391	194	11	.	.	PUNCT
elm-5391	195	1	affective	affective	NOUN
elm-5391	195	2	norms	norm	NOUN
elm-5391	195	3	for	for	ADP
elm-5391	195	4	english	english	ADJ
elm-5391	195	5	words	word	NOUN
elm-5391	195	6	(	(	PUNCT
elm-5391	195	7	anew	anew	ADV
elm-5391	195	8	):	):	PUNCT
elm-5391	195	9	instruction	instruction	NOUN
elm-5391	195	10	manual	manual	NOUN
elm-5391	195	11	and	and	CCONJ
elm-5391	195	12	affective	affective	ADJ
elm-5391	195	13	ratings	rating	NOUN
elm-5391	195	14	.	.	PUNCT
elm-5391	196	1	tech	tech	NOUN
elm-5391	196	2	.	.	PUNCT
elm-5391	197	1	rep	rep	PROPN
elm-5391	197	2	.	.	PROPN
elm-5391	198	1	the	the	DET
elm-5391	198	2	center	center	NOUN
elm-5391	198	3	for	for	ADP
elm-5391	198	4	research	research	NOUN
elm-5391	198	5	in	in	ADP
elm-5391	198	6	psychophysiology	psychophysiology	NOUN
elm-5391	198	7	.	.	PUNCT
elm-5391	199	1	buitinck	buitinck	PROPN
elm-5391	199	2	,	,	PUNCT
elm-5391	199	3	lars	lars	PROPN
elm-5391	199	4	,	,	PUNCT
elm-5391	199	5	gilles	gilles	PROPN
elm-5391	199	6	louppe	louppe	PROPN
elm-5391	199	7	,	,	PUNCT
elm-5391	199	8	mathieu	mathieu	PROPN
elm-5391	199	9	blondel	blondel	PROPN
elm-5391	199	10	,	,	PUNCT
elm-5391	199	11	fabian	fabian	PROPN
elm-5391	199	12	pedregosa	pedregosa	PROPN
elm-5391	199	13	,	,	PUNCT
elm-5391	199	14	andreas	andreas	PROPN
elm-5391	199	15	mueller	mueller	PROPN
elm-5391	199	16	,	,	PUNCT
elm-5391	199	17	olivier	olivier	PROPN
elm-5391	199	18	grisel	grisel	PROPN
elm-5391	199	19	,	,	PUNCT
elm-5391	199	20	vlad	vlad	VERB
elm-5391	199	21	niculae	niculae	ADJ
elm-5391	199	22	,	,	PUNCT
elm-5391	199	23	peter	peter	PROPN
elm-5391	199	24	prettenhofer	prettenhofer	NOUN
elm-5391	199	25	,	,	PUNCT
elm-5391	199	26	alexandre	alexandre	NOUN
elm-5391	199	27	gramfort	gramfort	PROPN
elm-5391	199	28	,	,	PUNCT
elm-5391	199	29	jaques	jaque	NOUN
elm-5391	199	30	grobler	grobler	NOUN
elm-5391	199	31	,	,	PUNCT
elm-5391	199	32	robert	robert	PROPN
elm-5391	199	33	layton	layton	PROPN
elm-5391	199	34	,	,	PUNCT
elm-5391	199	35	jake	jake	PROPN
elm-5391	199	36	vanderplas	vanderplas	PROPN
elm-5391	199	37	,	,	PUNCT
elm-5391	199	38	arnaud	arnaud	PROPN
elm-5391	199	39	joly	joly	PROPN
elm-5391	199	40	,	,	PUNCT
elm-5391	199	41	brian	brian	PROPN
elm-5391	199	42	holt	holt	PROPN
elm-5391	199	43	&	&	CCONJ
elm-5391	199	44	gaël	gaël	PROPN
elm-5391	199	45	varoquaux	varoquaux	PROPN
elm-5391	199	46	.	.	PUNCT
elm-5391	200	1	2013	2013	NUM
elm-5391	200	2	.	.	PUNCT
elm-5391	201	1	api	api	NOUN
elm-5391	201	2	design	design	NOUN
elm-5391	201	3	for	for	ADP
elm-5391	201	4	machine	machine	NOUN
elm-5391	201	5	learning	learn	VERB
elm-5391	201	6	software	software	NOUN
elm-5391	201	7	:	:	PUNCT
elm-5391	201	8	experiences	experience	NOUN
elm-5391	201	9	from	from	ADP
elm-5391	201	10	the	the	DET
elm-5391	201	11	scikit	scikit	NOUN
elm-5391	201	12	-	-	PUNCT
elm-5391	201	13	learn	learn	NOUN
elm-5391	201	14	project	project	NOUN
elm-5391	201	15	.	.	PUNCT
elm-5391	202	1	in	in	ADP
elm-5391	202	2	ecml	ecml	PROPN
elm-5391	202	3	pkdd	pkdd	NOUN
elm-5391	202	4	workshop	workshop	NOUN
elm-5391	202	5	:	:	PUNCT
elm-5391	202	6	languages	language	NOUN
elm-5391	202	7	for	for	ADP
elm-5391	202	8	data	data	NOUN
elm-5391	202	9	mining	mining	NOUN
elm-5391	202	10	and	and	CCONJ
elm-5391	202	11	machine	machine	NOUN
elm-5391	202	12	learning	learning	NOUN
elm-5391	202	13	,	,	PUNCT
elm-5391	202	14	108–122	108–122	NUM
elm-5391	202	15	.	.	PUNCT
elm-5391	202	16	demszky	demszky	ADJ
elm-5391	202	17	,	,	PUNCT
elm-5391	202	18	dorottya	dorottya	ADJ
elm-5391	202	19	,	,	PUNCT
elm-5391	202	20	dana	dana	PROPN
elm-5391	202	21	movshovitz	movshovitz	PROPN
elm-5391	202	22	-	-	PUNCT
elm-5391	202	23	attias	attia	NOUN
elm-5391	202	24	,	,	PUNCT
elm-5391	202	25	jeongwoo	jeongwoo	PROPN
elm-5391	202	26	ko	ko	PROPN
elm-5391	202	27	,	,	PUNCT
elm-5391	202	28	alan	alan	PROPN
elm-5391	202	29	cowen	cowen	PROPN
elm-5391	202	30	,	,	PUNCT
elm-5391	202	31	gaurav	gaurav	PROPN
elm-5391	202	32	nemade	nemade	PROPN
elm-5391	202	33	&	&	CCONJ
elm-5391	202	34	sujith	sujith	PROPN
elm-5391	202	35	ravi	ravi	PROPN
elm-5391	202	36	.	.	PUNCT
elm-5391	203	1	2020	2020	NUM
elm-5391	203	2	.	.	PUNCT
elm-5391	204	1	goemotions	goemotion	NOUN
elm-5391	204	2	:	:	PUNCT
elm-5391	204	3	a	a	DET
elm-5391	204	4	dataset	dataset	NOUN
elm-5391	204	5	of	of	ADP
elm-5391	204	6	fine	fine	ADV
elm-5391	204	7	-	-	PUNCT
elm-5391	204	8	grained	grain	VERB
elm-5391	204	9	emotions	emotion	NOUN
elm-5391	204	10	.	.	PUNCT
elm-5391	205	1	in	in	ADP
elm-5391	205	2	proceedings	proceeding	NOUN
elm-5391	205	3	of	of	ADP
elm-5391	205	4	the	the	DET
elm-5391	205	5	58th	58th	ADJ
elm-5391	205	6	annual	annual	ADJ
elm-5391	205	7	meeting	meeting	NOUN
elm-5391	205	8	of	of	ADP
elm-5391	205	9	the	the	DET
elm-5391	205	10	association	association	NOUN
elm-5391	205	11	for	for	ADP
elm-5391	205	12	computational	computational	ADJ
elm-5391	205	13	linguistics	linguistic	NOUN
elm-5391	205	14	(	(	PUNCT
elm-5391	205	15	acl	acl	PROPN
elm-5391	205	16	)	)	PUNCT
elm-5391	205	17	,	,	PUNCT
elm-5391	205	18	4040	4040	NUM
elm-5391	205	19	–	–	PUNCT
elm-5391	205	20	4054	4054	NUM
elm-5391	205	21	.	.	PUNCT
elm-5391	206	1	online	online	ADJ
elm-5391	206	2	:	:	PUNCT
elm-5391	206	3	association	association	NOUN
elm-5391	206	4	for	for	ADP
elm-5391	206	5	computational	computational	ADJ
elm-5391	206	6	linguistics	linguistic	NOUN
elm-5391	206	7	.	.	PUNCT
elm-5391	207	1	10.18653	10.18653	NUM
elm-5391	207	2	/	/	SYM
elm-5391	207	3	v1/2020.acl	v1/2020.acl	PROPN
elm-5391	207	4	-	-	PUNCT
elm-5391	207	5	main.372	main.372	PROPN
elm-5391	207	6	.	.	PUNCT
elm-5391	208	1	https://aclanthology.org/2020.acl-main.372	https://aclanthology.org/2020.acl-main.372	PROPN
elm-5391	208	2	.	.	PUNCT
elm-5391	209	1	devlin	devlin	PROPN
elm-5391	209	2	,	,	PUNCT
elm-5391	209	3	jacob	jacob	PROPN
elm-5391	209	4	,	,	PUNCT
elm-5391	209	5	ming	ming	PROPN
elm-5391	209	6	-	-	PUNCT
elm-5391	209	7	wei	wei	PROPN
elm-5391	209	8	chang	chang	PROPN
elm-5391	209	9	,	,	PUNCT
elm-5391	209	10	kenton	kenton	PROPN
elm-5391	209	11	lee	lee	PROPN
elm-5391	209	12	&	&	CCONJ
elm-5391	209	13	kristina	kristina	PROPN
elm-5391	209	14	toutanova	toutanova	PROPN
elm-5391	209	15	.	.	PROPN
elm-5391	209	16	2019	2019	NUM
elm-5391	209	17	.	.	PUNCT
elm-5391	210	1	bert	bert	PROPN
elm-5391	210	2	:	:	PUNCT
elm-5391	210	3	pre	pre	ADJ
elm-5391	210	4	-	-	NOUN
elm-5391	210	5	training	training	NOUN
elm-5391	210	6	of	of	ADP
elm-5391	210	7	deep	deep	ADJ
elm-5391	210	8	bidirectional	bidirectional	ADJ
elm-5391	210	9	transformers	transformer	NOUN
elm-5391	210	10	for	for	ADP
elm-5391	210	11	language	language	NOUN
elm-5391	210	12	understanding	understanding	NOUN
elm-5391	210	13	.	.	PUNCT
elm-5391	211	1	in	in	ADP
elm-5391	211	2	proceedings	proceeding	NOUN
elm-5391	211	3	of	of	ADP
elm-5391	211	4	the	the	DET
elm-5391	211	5	2019	2019	NUM
elm-5391	211	6	conference	conference	NOUN
elm-5391	211	7	of	of	ADP
elm-5391	211	8	the	the	DET
elm-5391	211	9	north	north	ADJ
elm-5391	211	10	american	american	ADJ
elm-5391	211	11	chapter	chapter	NOUN
elm-5391	211	12	of	of	ADP
elm-5391	211	13	the	the	DET
elm-5391	211	14	association	association	NOUN
elm-5391	211	15	for	for	ADP
elm-5391	211	16	computational	computational	ADJ
elm-5391	211	17	linguistics	linguistic	NOUN
elm-5391	211	18	:	:	PUNCT
elm-5391	211	19	human	human	ADJ
elm-5391	211	20	language	language	NOUN
elm-5391	211	21	technologies	technology	NOUN
elm-5391	211	22	,	,	PUNCT
elm-5391	211	23	volume	volume	NOUN
elm-5391	211	24	1	1	NUM
elm-5391	211	25	(	(	PUNCT
elm-5391	211	26	long	long	ADJ
elm-5391	211	27	and	and	CCONJ
elm-5391	211	28	short	short	ADJ
elm-5391	211	29	papers	paper	NOUN
elm-5391	211	30	)	)	PUNCT
elm-5391	211	31	,	,	PUNCT
elm-5391	211	32	4171–4186	4171–4186	NUM
elm-5391	211	33	.	.	PUNCT
elm-5391	211	34	minneapolis	minneapolis	PROPN
elm-5391	211	35	,	,	PUNCT
elm-5391	211	36	minnesota	minnesota	PROPN
elm-5391	211	37	:	:	PUNCT
elm-5391	211	38	association	association	NOUN
elm-5391	211	39	for	for	ADP
elm-5391	211	40	computational	computational	ADJ
elm-5391	211	41	linguistics	linguistic	NOUN
elm-5391	211	42	.	.	PUNCT
elm-5391	212	1	10.18653	10.18653	NUM
elm-5391	212	2	/	/	SYM
elm-5391	212	3	v1	v1	NOUN
elm-5391	212	4	/	/	SYM
elm-5391	212	5	n19	n19	NOUN
elm-5391	212	6	-	-	PUNCT
elm-5391	212	7	1423	1423	NUM
elm-5391	212	8	.	.	PUNCT
elm-5391	213	1	https	https	NOUN
elm-5391	213	2	:	:	PUNCT
elm-5391	213	3	//aclanthology.org	//aclanthology.org	ADJ
elm-5391	213	4	/	/	SYM
elm-5391	213	5	n19	n19	NOUN
elm-5391	213	6	-	-	PUNCT
elm-5391	213	7	1423	1423	NUM
elm-5391	213	8	.	.	PUNCT
elm-5391	214	1	dumais	dumais	PROPN
elm-5391	214	2	,	,	PUNCT
elm-5391	214	3	susan	susan	PROPN
elm-5391	214	4	t	t	PROPN
elm-5391	214	5	,	,	PUNCT
elm-5391	214	6	george	george	PROPN
elm-5391	214	7	w	w	PROPN
elm-5391	214	8	furnas	furnas	PROPN
elm-5391	214	9	,	,	PUNCT
elm-5391	214	10	thomas	thomas	PROPN
elm-5391	214	11	k	k	PROPN
elm-5391	214	12	landauer	landauer	PROPN
elm-5391	214	13	,	,	PUNCT
elm-5391	214	14	scott	scott	PROPN
elm-5391	214	15	deerwester	deerwester	PROPN
elm-5391	214	16	&	&	CCONJ
elm-5391	214	17	richard	richard	PROPN
elm-5391	214	18	harshman	harshman	PROPN
elm-5391	214	19	.	.	PUNCT
elm-5391	214	20	1988	1988	NUM
elm-5391	214	21	.	.	PUNCT
elm-5391	215	1	using	use	VERB
elm-5391	215	2	latent	latent	NOUN
elm-5391	215	3	semantic	semantic	ADJ
elm-5391	215	4	analysis	analysis	NOUN
elm-5391	215	5	to	to	PART
elm-5391	215	6	improve	improve	VERB
elm-5391	215	7	access	access	NOUN
elm-5391	215	8	to	to	ADP
elm-5391	215	9	textual	textual	ADJ
elm-5391	215	10	information	information	NOUN
elm-5391	215	11	.	.	PUNCT
elm-5391	216	1	in	in	ADP
elm-5391	216	2	proceedings	proceeding	NOUN
elm-5391	216	3	of	of	ADP
elm-5391	216	4	the	the	DET
elm-5391	216	5	sigchi	sigchi	PROPN
elm-5391	216	6	conference	conference	NOUN
elm-5391	216	7	on	on	ADP
elm-5391	216	8	human	human	ADJ
elm-5391	216	9	factors	factor	NOUN
elm-5391	216	10	in	in	ADP
elm-5391	216	11	computing	computing	NOUN
elm-5391	216	12	systems	system	NOUN
elm-5391	216	13	,	,	PUNCT
elm-5391	216	14	281–285	281–285	NUM
elm-5391	216	15	.	.	PUNCT
elm-5391	216	16	ethayarajh	ethayarajh	PROPN
elm-5391	216	17	,	,	PUNCT
elm-5391	216	18	kawin	kawin	NOUN
elm-5391	216	19	.	.	PUNCT
elm-5391	216	20	2019	2019	NUM
elm-5391	216	21	.	.	PUNCT
elm-5391	217	1	how	how	SCONJ
elm-5391	217	2	contextual	contextual	ADJ
elm-5391	217	3	are	be	AUX
elm-5391	217	4	contextualized	contextualize	VERB
elm-5391	217	5	word	word	NOUN
elm-5391	217	6	representations	representation	NOUN
elm-5391	217	7	?	?	PUNCT
elm-5391	217	8	comparing	compare	VERB
elm-5391	217	9	the	the	DET
elm-5391	217	10	geometry	geometry	NOUN
elm-5391	217	11	of	of	ADP
elm-5391	217	12	bert	bert	PROPN
elm-5391	217	13	,	,	PUNCT
elm-5391	217	14	elmo	elmo	PROPN
elm-5391	217	15	,	,	PUNCT
elm-5391	217	16	and	and	CCONJ
elm-5391	217	17	gpt-2	gpt-2	PUNCT
elm-5391	217	18	embeddings	embedding	NOUN
elm-5391	217	19	.	.	PUNCT
elm-5391	218	1	in	in	ADP
elm-5391	218	2	proceedings	proceeding	NOUN
elm-5391	218	3	of	of	ADP
elm-5391	218	4	the	the	DET
elm-5391	218	5	2019	2019	NUM
elm-5391	218	6	conference	conference	NOUN
elm-5391	218	7	on	on	ADP
elm-5391	218	8	empirical	empirical	ADJ
elm-5391	218	9	methods	method	NOUN
elm-5391	218	10	in	in	ADP
elm-5391	218	11	natural	natural	ADJ
elm-5391	218	12	language	language	NOUN
elm-5391	218	13	processing	processing	NOUN
elm-5391	218	14	and	and	CCONJ
elm-5391	218	15	the	the	DET
elm-5391	218	16	9th	9th	ADJ
elm-5391	218	17	international	international	ADJ
elm-5391	218	18	joint	joint	ADJ
elm-5391	218	19	conference	conference	NOUN
elm-5391	218	20	on	on	ADP
elm-5391	218	21	natural	natural	ADJ
elm-5391	218	22	language	language	NOUN
elm-5391	218	23	processing	processing	NOUN
elm-5391	218	24	(	(	PUNCT
elm-5391	218	25	emnlp	emnlp	NOUN
elm-5391	218	26	-	-	PUNCT
elm-5391	218	27	ijcnlp	ijcnlp	NOUN
elm-5391	218	28	)	)	PUNCT
elm-5391	218	29	,	,	PUNCT
elm-5391	218	30	55–65	55–65	NUM
elm-5391	218	31	.	.	PUNCT
elm-5391	219	1	hong	hong	PROPN
elm-5391	219	2	kong	kong	PROPN
elm-5391	219	3	,	,	PUNCT
elm-5391	219	4	china	china	PROPN
elm-5391	219	5	:	:	PUNCT
elm-5391	219	6	association	association	NOUN
elm-5391	219	7	for	for	ADP
elm-5391	219	8	computational	computational	ADJ
elm-5391	219	9	linguistics	linguistic	NOUN
elm-5391	219	10	.	.	PUNCT
elm-5391	220	1	10.18653	10.18653	NUM
elm-5391	220	2	/	/	SYM
elm-5391	220	3	v1	v1	NOUN
elm-5391	220	4	/	/	SYM
elm-5391	220	5	d19	d19	PROPN
elm-5391	220	6	-	-	PUNCT
elm-5391	220	7	1006	1006	NUM
elm-5391	220	8	.	.	PUNCT
elm-5391	221	1	https	https	NOUN
elm-5391	221	2	:	:	PUNCT
elm-5391	221	3	//aclanthology.org	//aclanthology.org	ADJ
elm-5391	221	4	/	/	SYM
elm-5391	221	5	d19	d19	PROPN
elm-5391	221	6	-	-	PUNCT
elm-5391	221	7	1006	1006	NUM
elm-5391	221	8	.	.	PUNCT
elm-5391	222	1	ettinger	ettinger	PROPN
elm-5391	222	2	,	,	PUNCT
elm-5391	222	3	allyson	allyson	PROPN
elm-5391	222	4	.	.	PUNCT
elm-5391	222	5	2020	2020	NUM
elm-5391	222	6	.	.	PUNCT
elm-5391	223	1	what	what	PRON
elm-5391	223	2	bert	bert	NOUN
elm-5391	223	3	is	be	AUX
elm-5391	223	4	not	not	PART
elm-5391	223	5	:	:	PUNCT
elm-5391	223	6	lessons	lesson	NOUN
elm-5391	223	7	from	from	ADP
elm-5391	223	8	a	a	DET
elm-5391	223	9	new	new	ADJ
elm-5391	223	10	suite	suite	NOUN
elm-5391	223	11	of	of	ADP
elm-5391	223	12	psycholinguistic	psycholinguistic	ADJ
elm-5391	223	13	diagnostics	diagnostic	NOUN
elm-5391	223	14	for	for	ADP
elm-5391	223	15	language	language	NOUN
elm-5391	223	16	models	model	NOUN
elm-5391	223	17	.	.	PUNCT
elm-5391	224	1	transactions	transaction	NOUN
elm-5391	224	2	of	of	ADP
elm-5391	224	3	the	the	DET
elm-5391	224	4	association	association	NOUN
elm-5391	224	5	for	for	ADP
elm-5391	224	6	computational	computational	ADJ
elm-5391	224	7	linguistics	linguistic	NOUN
elm-5391	224	8	(	(	PUNCT
elm-5391	224	9	acl	acl	PROPN
elm-5391	224	10	)	)	PUNCT
elm-5391	224	11	8	8	NUM
elm-5391	224	12	.	.	PUNCT
elm-5391	224	13	34–48	34–48	NUM
elm-5391	224	14	.	.	PUNCT
elm-5391	225	1	hu	hu	PROPN
elm-5391	225	2	,	,	PUNCT
elm-5391	225	3	jennifer	jennifer	PROPN
elm-5391	225	4	,	,	PUNCT
elm-5391	225	5	jon	jon	PROPN
elm-5391	225	6	gauthier	gauthier	PROPN
elm-5391	225	7	,	,	PUNCT
elm-5391	225	8	peng	peng	PROPN
elm-5391	225	9	qian	qian	PROPN
elm-5391	225	10	,	,	PUNCT
elm-5391	225	11	ethan	ethan	PROPN
elm-5391	225	12	wilcox	wilcox	PROPN
elm-5391	225	13	&	&	CCONJ
elm-5391	225	14	roger	roger	PROPN
elm-5391	225	15	levy	levy	PROPN
elm-5391	225	16	.	.	PUNCT
elm-5391	226	1	2020	2020	NUM
elm-5391	226	2	.	.	PUNCT
elm-5391	227	1	a	a	DET
elm-5391	227	2	systematic	systematic	ADJ
elm-5391	227	3	assessment	assessment	NOUN
elm-5391	227	4	of	of	ADP
elm-5391	227	5	syntactic	syntactic	ADJ
elm-5391	227	6	generalization	generalization	NOUN
elm-5391	227	7	in	in	ADP
elm-5391	227	8	neural	neural	ADJ
elm-5391	227	9	language	language	NOUN
elm-5391	227	10	models	model	NOUN
elm-5391	227	11	.	.	PUNCT
elm-5391	228	1	in	in	ADP
elm-5391	228	2	proceedings	proceeding	NOUN
elm-5391	228	3	of	of	ADP
elm-5391	228	4	the	the	DET
elm-5391	228	5	58th	58th	ADJ
elm-5391	228	6	annual	annual	ADJ
elm-5391	228	7	meeting	meeting	NOUN
elm-5391	228	8	of	of	ADP
elm-5391	228	9	the	the	DET
elm-5391	228	10	association	association	NOUN
elm-5391	228	11	for	for	ADP
elm-5391	228	12	computational	computational	ADJ
elm-5391	228	13	linguistics	linguistic	NOUN
elm-5391	228	14	(	(	PUNCT
elm-5391	228	15	acl	acl	PROPN
elm-5391	228	16	)	)	PUNCT
elm-5391	228	17	,	,	PUNCT
elm-5391	228	18	1725–1744	1725–1744	NUM
elm-5391	228	19	.	.	PUNCT
elm-5391	229	1	onproceedings	onproceeding	NOUN
elm-5391	229	2	of	of	ADP
elm-5391	229	3	elm	elm	NOUN
elm-5391	229	4	2	2	NUM
elm-5391	229	5	:	:	PUNCT
elm-5391	229	6	310	310	NUM
elm-5391	229	7	-	-	SYM
elm-5391	229	8	321	321	NUM
elm-5391	229	9	,	,	PUNCT
elm-5391	229	10	2023	2023	NUM
elm-5391	229	11	yuhan	yuhan	PROPN
elm-5391	229	12	zhang	zhang	PROPN
elm-5391	229	13	,	,	PUNCT
elm-5391	229	14	wenqi	wenqi	PROPN
elm-5391	229	15	chen	chen	PROPN
elm-5391	229	16	,	,	PUNCT
elm-5391	229	17	ruihan	ruihan	PROPN
elm-5391	229	18	zhang	zhang	PROPN
elm-5391	229	19	and	and	CCONJ
elm-5391	229	20	xiajie	xiajie	PROPN
elm-5391	229	21	zhang	zhang	PROPN
elm-5391	229	22	:	:	PUNCT
elm-5391	229	23	representing	represent	VERB
elm-5391	229	24	affect	affect	VERB
elm-5391	229	25	information	information	NOUN
elm-5391	229	26	in	in	ADP
elm-5391	229	27	word	word	NOUN
elm-5391	229	28	embeddings	embedding	NOUN
elm-5391	229	29	.	.	PUNCT
elm-5391	230	1	318	318	NUM
elm-5391	230	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	230	3	https://www.elm-conference.net/	https://www.elm-conference.net/	DET
elm-5391	230	4	line	line	NOUN
elm-5391	230	5	:	:	PUNCT
elm-5391	230	6	association	association	NOUN
elm-5391	230	7	for	for	ADP
elm-5391	230	8	computational	computational	ADJ
elm-5391	230	9	linguistics	linguistic	NOUN
elm-5391	230	10	.	.	PUNCT
elm-5391	231	1	10.18653	10.18653	NUM
elm-5391	231	2	/	/	SYM
elm-5391	231	3	v1/2020.acl	v1/2020.acl	PROPN
elm-5391	231	4	-	-	PUNCT
elm-5391	231	5	main.158	main.158	PROPN
elm-5391	231	6	.	.	PUNCT
elm-5391	231	7	https	https	NOUN
elm-5391	231	8	:	:	PUNCT
elm-5391	231	9	//aclanthology.org/2020.acl	//aclanthology.org/2020.acl	PROPN
elm-5391	231	10	-	-	PUNCT
elm-5391	231	11	main.158	main.158	PROPN
elm-5391	231	12	.	.	PUNCT
elm-5391	232	1	klafka	klafka	VERB
elm-5391	232	2	,	,	PUNCT
elm-5391	232	3	josef	josef	PROPN
elm-5391	232	4	&	&	CCONJ
elm-5391	232	5	allyson	allyson	PROPN
elm-5391	232	6	ettinger	ettinger	PROPN
elm-5391	232	7	.	.	PUNCT
elm-5391	233	1	2020	2020	NUM
elm-5391	233	2	.	.	PUNCT
elm-5391	234	1	spying	spy	VERB
elm-5391	234	2	on	on	ADP
elm-5391	234	3	your	your	PRON
elm-5391	234	4	neighbors	neighbor	NOUN
elm-5391	234	5	:	:	PUNCT
elm-5391	234	6	fine	fine	ADJ
elm-5391	234	7	-	-	PUNCT
elm-5391	234	8	grained	grain	VERB
elm-5391	234	9	probing	probing	NOUN
elm-5391	234	10	of	of	ADP
elm-5391	234	11	contextual	contextual	ADJ
elm-5391	234	12	embeddings	embedding	NOUN
elm-5391	234	13	for	for	ADP
elm-5391	234	14	information	information	NOUN
elm-5391	234	15	about	about	ADP
elm-5391	234	16	surrounding	surround	VERB
elm-5391	234	17	words	word	NOUN
elm-5391	234	18	.	.	PUNCT
elm-5391	235	1	arxiv	arxiv	PROPN
elm-5391	235	2	preprint	preprint	NOUN
elm-5391	235	3	arxiv:2005.01810	arxiv:2005.01810	NOUN
elm-5391	235	4	.	.	PUNCT
elm-5391	236	1	li	li	PROPN
elm-5391	236	2	,	,	PUNCT
elm-5391	236	3	minglei	minglei	PROPN
elm-5391	236	4	,	,	PUNCT
elm-5391	236	5	qin	qin	PROPN
elm-5391	236	6	lu	lu	PROPN
elm-5391	236	7	,	,	PUNCT
elm-5391	236	8	yunfei	yunfei	PROPN
elm-5391	236	9	long	long	PROPN
elm-5391	236	10	&	&	CCONJ
elm-5391	236	11	lin	lin	PROPN
elm-5391	236	12	gui	gui	PROPN
elm-5391	236	13	.	.	PROPN
elm-5391	236	14	2017	2017	NUM
elm-5391	236	15	.	.	PUNCT
elm-5391	237	1	inferring	infer	VERB
elm-5391	237	2	affective	affective	ADJ
elm-5391	237	3	meanings	meaning	NOUN
elm-5391	237	4	of	of	ADP
elm-5391	237	5	words	word	NOUN
elm-5391	237	6	from	from	ADP
elm-5391	237	7	word	word	NOUN
elm-5391	237	8	embedding	embed	VERB
elm-5391	237	9	.	.	PUNCT
elm-5391	238	1	ieee	ieee	NOUN
elm-5391	238	2	transactions	transaction	NOUN
elm-5391	238	3	on	on	ADP
elm-5391	238	4	affective	affective	ADJ
elm-5391	238	5	computing	computing	NOUN
elm-5391	238	6	8(4	8(4	NUM
elm-5391	238	7	)	)	PUNCT
elm-5391	238	8	.	.	PUNCT
elm-5391	239	1	443–456	443–456	NUM
elm-5391	239	2	.	.	PUNCT
elm-5391	240	1	10.1109	10.1109	NUM
elm-5391	240	2	/	/	SYM
elm-5391	240	3	taffc.2017.2723012	taffc.2017.2723012	PROPN
elm-5391	240	4	.	.	PUNCT
elm-5391	241	1	linzen	linzen	PROPN
elm-5391	241	2	,	,	PUNCT
elm-5391	241	3	tal	tal	X
elm-5391	241	4	,	,	PUNCT
elm-5391	241	5	emmanuel	emmanuel	PROPN
elm-5391	241	6	dupoux	dupoux	PROPN
elm-5391	241	7	&	&	CCONJ
elm-5391	241	8	yoav	yoav	PROPN
elm-5391	241	9	goldberg	goldberg	PROPN
elm-5391	241	10	.	.	PROPN
elm-5391	241	11	2016	2016	NUM
elm-5391	241	12	.	.	PUNCT
elm-5391	242	1	assessing	assess	VERB
elm-5391	242	2	the	the	DET
elm-5391	242	3	ability	ability	NOUN
elm-5391	242	4	of	of	ADP
elm-5391	242	5	lstms	lstms	NOUN
elm-5391	242	6	to	to	PART
elm-5391	242	7	learn	learn	VERB
elm-5391	242	8	syntax	syntax	NOUN
elm-5391	242	9	-	-	PUNCT
elm-5391	242	10	sensitive	sensitive	ADJ
elm-5391	242	11	dependencies	dependency	NOUN
elm-5391	242	12	.	.	PUNCT
elm-5391	243	1	transactions	transaction	NOUN
elm-5391	243	2	of	of	ADP
elm-5391	243	3	the	the	DET
elm-5391	243	4	association	association	NOUN
elm-5391	243	5	for	for	ADP
elm-5391	243	6	computational	computational	ADJ
elm-5391	243	7	linguistics	linguistic	NOUN
elm-5391	243	8	4	4	NUM
elm-5391	243	9	.	.	PUNCT
elm-5391	244	1	521–535	521–535	NUM
elm-5391	244	2	.	.	PUNCT
elm-5391	245	1	loureiro	loureiro	PROPN
elm-5391	245	2	,	,	PUNCT
elm-5391	245	3	daniel	daniel	PROPN
elm-5391	245	4	,	,	PUNCT
elm-5391	245	5	kiamehr	kiamehr	NOUN
elm-5391	245	6	rezaee	rezaee	PROPN
elm-5391	245	7	,	,	PUNCT
elm-5391	245	8	mohammad	mohammad	PROPN
elm-5391	245	9	taher	taher	PROPN
elm-5391	245	10	pilehvar	pilehvar	PROPN
elm-5391	245	11	&	&	CCONJ
elm-5391	245	12	jose	jose	PROPN
elm-5391	245	13	camacho	camacho	PROPN
elm-5391	245	14	-	-	PUNCT
elm-5391	245	15	collados	collado	NOUN
elm-5391	245	16	.	.	PUNCT
elm-5391	245	17	2021	2021	NUM
elm-5391	245	18	.	.	PUNCT
elm-5391	246	1	analysis	analysis	NOUN
elm-5391	246	2	and	and	CCONJ
elm-5391	246	3	evaluation	evaluation	NOUN
elm-5391	246	4	of	of	ADP
elm-5391	246	5	language	language	NOUN
elm-5391	246	6	models	model	NOUN
elm-5391	246	7	for	for	ADP
elm-5391	246	8	word	word	NOUN
elm-5391	246	9	sense	sense	NOUN
elm-5391	246	10	disambiguation	disambiguation	NOUN
elm-5391	246	11	.	.	PUNCT
elm-5391	247	1	computational	computational	ADJ
elm-5391	247	2	linguistics	linguistic	NOUN
elm-5391	247	3	47(2	47(2	NUM
elm-5391	247	4	)	)	PUNCT
elm-5391	247	5	.	.	PUNCT
elm-5391	248	1	387–443	387–443	NUM
elm-5391	248	2	.	.	PUNCT
elm-5391	249	1	maas	maas	PROPN
elm-5391	249	2	,	,	PUNCT
elm-5391	249	3	andrew	andrew	PROPN
elm-5391	249	4	l.	l.	PROPN
elm-5391	249	5	,	,	PUNCT
elm-5391	249	6	raymond	raymond	PROPN
elm-5391	249	7	e.	e.	PROPN
elm-5391	249	8	daly	daly	PROPN
elm-5391	249	9	,	,	PUNCT
elm-5391	249	10	peter	peter	PROPN
elm-5391	249	11	t.	t.	PROPN
elm-5391	249	12	pham	pham	PROPN
elm-5391	249	13	,	,	PUNCT
elm-5391	249	14	dan	dan	PROPN
elm-5391	249	15	huang	huang	PROPN
elm-5391	249	16	,	,	PUNCT
elm-5391	249	17	andrew	andrew	PROPN
elm-5391	249	18	y.	y.	PROPN
elm-5391	249	19	ng	ng	PROPN
elm-5391	249	20	&	&	CCONJ
elm-5391	249	21	christopher	christopher	PROPN
elm-5391	249	22	potts	potts	PROPN
elm-5391	249	23	.	.	PUNCT
elm-5391	250	1	2011	2011	NUM
elm-5391	250	2	.	.	PUNCT
elm-5391	251	1	learning	learn	VERB
elm-5391	251	2	word	word	NOUN
elm-5391	251	3	vectors	vector	NOUN
elm-5391	251	4	for	for	ADP
elm-5391	251	5	sentiment	sentiment	NOUN
elm-5391	251	6	analysis	analysis	NOUN
elm-5391	251	7	.	.	PUNCT
elm-5391	252	1	in	in	ADP
elm-5391	252	2	proceedings	proceeding	NOUN
elm-5391	252	3	of	of	ADP
elm-5391	252	4	the	the	DET
elm-5391	252	5	49th	49th	ADJ
elm-5391	252	6	annual	annual	ADJ
elm-5391	252	7	meeting	meeting	NOUN
elm-5391	252	8	of	of	ADP
elm-5391	252	9	the	the	DET
elm-5391	252	10	association	association	NOUN
elm-5391	252	11	for	for	ADP
elm-5391	252	12	computational	computational	ADJ
elm-5391	252	13	linguistics	linguistic	NOUN
elm-5391	252	14	:	:	PUNCT
elm-5391	252	15	human	human	ADJ
elm-5391	252	16	language	language	NOUN
elm-5391	252	17	technologies	technology	NOUN
elm-5391	252	18	(	(	PUNCT
elm-5391	252	19	acl	acl	PROPN
elm-5391	252	20	)	)	PUNCT
elm-5391	252	21	,	,	PUNCT
elm-5391	252	22	142–150	142–150	NUM
elm-5391	252	23	.	.	PUNCT
elm-5391	253	1	portland	portland	PROPN
elm-5391	253	2	,	,	PUNCT
elm-5391	253	3	oregon	oregon	PROPN
elm-5391	253	4	,	,	PUNCT
elm-5391	253	5	usa	usa	PROPN
elm-5391	253	6	:	:	PUNCT
elm-5391	253	7	association	association	NOUN
elm-5391	253	8	for	for	ADP
elm-5391	253	9	computational	computational	ADJ
elm-5391	253	10	linguistics	linguistic	NOUN
elm-5391	253	11	.	.	PUNCT
elm-5391	254	1	http	http	ADJ
elm-5391	254	2	:	:	PUNCT
elm-5391	254	3	//www.aclweb.org	//www.aclweb.org	PUNCT
elm-5391	254	4	/	/	SYM
elm-5391	254	5	anthology	anthology	NOUN
elm-5391	254	6	/	/	SYM
elm-5391	254	7	p11	p11	NOUN
elm-5391	254	8	-	-	PUNCT
elm-5391	254	9	1015	1015	NUM
elm-5391	254	10	.	.	PUNCT
elm-5391	255	1	manning	man	VERB
elm-5391	255	2	,	,	PUNCT
elm-5391	255	3	christopher	christopher	PROPN
elm-5391	255	4	d	d	PROPN
elm-5391	255	5	,	,	PUNCT
elm-5391	255	6	kevin	kevin	PROPN
elm-5391	255	7	clark	clark	PROPN
elm-5391	255	8	,	,	PUNCT
elm-5391	255	9	john	john	PROPN
elm-5391	255	10	hewitt	hewitt	PROPN
elm-5391	255	11	,	,	PUNCT
elm-5391	255	12	urvashi	urvashi	PROPN
elm-5391	255	13	khandelwal	khandelwal	PROPN
elm-5391	255	14	&	&	CCONJ
elm-5391	255	15	omer	omer	PROPN
elm-5391	255	16	levy	levy	PROPN
elm-5391	255	17	.	.	PUNCT
elm-5391	256	1	2020	2020	NUM
elm-5391	256	2	.	.	PUNCT
elm-5391	257	1	emergent	emergent	ADJ
elm-5391	257	2	linguistic	linguistic	ADJ
elm-5391	257	3	structure	structure	NOUN
elm-5391	257	4	in	in	ADP
elm-5391	257	5	artificial	artificial	ADJ
elm-5391	257	6	neural	neural	ADJ
elm-5391	257	7	networks	network	NOUN
elm-5391	257	8	trained	train	VERB
elm-5391	257	9	by	by	ADP
elm-5391	257	10	selfsupervision	selfsupervision	NOUN
elm-5391	257	11	.	.	PUNCT
elm-5391	258	1	proceedings	proceeding	NOUN
elm-5391	258	2	of	of	ADP
elm-5391	258	3	the	the	DET
elm-5391	258	4	national	national	PROPN
elm-5391	258	5	academy	academy	PROPN
elm-5391	258	6	of	of	ADP
elm-5391	258	7	sciences	science	NOUN
elm-5391	258	8	117(48	117(48	NUM
elm-5391	258	9	)	)	PUNCT
elm-5391	258	10	.	.	PUNCT
elm-5391	259	1	30046–30054	30046–30054	NUM
elm-5391	259	2	.	.	PUNCT
elm-5391	259	3	https://doi.org/10.1073/pnas.1907367117	https://doi.org/10.1073/pnas.1907367117	PROPN
elm-5391	259	4	.	.	PUNCT
elm-5391	260	1	mao	mao	PROPN
elm-5391	260	2	,	,	PUNCT
elm-5391	260	3	xingliang	xingliang	PROPN
elm-5391	260	4	,	,	PUNCT
elm-5391	260	5	shuai	shuai	PROPN
elm-5391	260	6	chang	chang	PROPN
elm-5391	260	7	,	,	PUNCT
elm-5391	260	8	jinjing	jinje	VERB
elm-5391	260	9	shi	shi	PROPN
elm-5391	260	10	,	,	PUNCT
elm-5391	260	11	fangfang	fangfang	PROPN
elm-5391	260	12	li	li	PROPN
elm-5391	260	13	&	&	CCONJ
elm-5391	260	14	ronghua	ronghua	PROPN
elm-5391	260	15	shi	shi	PROPN
elm-5391	260	16	.	.	PROPN
elm-5391	260	17	2019	2019	NUM
elm-5391	260	18	.	.	PUNCT
elm-5391	261	1	sentiment	sentiment	NOUN
elm-5391	261	2	-	-	PUNCT
elm-5391	261	3	aware	aware	ADJ
elm-5391	261	4	word	word	NOUN
elm-5391	261	5	embedding	embed	VERB
elm-5391	261	6	for	for	ADP
elm-5391	261	7	emotion	emotion	NOUN
elm-5391	261	8	classification	classification	NOUN
elm-5391	261	9	.	.	PUNCT
elm-5391	262	1	applied	apply	VERB
elm-5391	262	2	sciences	science	NOUN
elm-5391	262	3	9(7	9(7	NOUN
elm-5391	262	4	)	)	PUNCT
elm-5391	262	5	.	.	PUNCT
elm-5391	263	1	10.3390	10.3390	NUM
elm-5391	263	2	/	/	SYM
elm-5391	263	3	app9071334	app9071334	NOUN
elm-5391	263	4	.	.	PUNCT
elm-5391	264	1	https://www.mdpi.com/2076-3417/9/7/1334	https://www.mdpi.com/2076-3417/9/7/1334	PROPN
elm-5391	264	2	.	.	PUNCT
elm-5391	265	1	mikolov	mikolov	PROPN
elm-5391	265	2	,	,	PUNCT
elm-5391	265	3	tomas	tomas	PROPN
elm-5391	265	4	,	,	PUNCT
elm-5391	265	5	ilya	ilya	PROPN
elm-5391	265	6	sutskever	sutskever	PROPN
elm-5391	265	7	,	,	PUNCT
elm-5391	265	8	kai	kai	PROPN
elm-5391	265	9	chen	chen	PROPN
elm-5391	265	10	,	,	PUNCT
elm-5391	265	11	greg	greg	PROPN
elm-5391	265	12	s	s	PROPN
elm-5391	265	13	corrado	corrado	PROPN
elm-5391	265	14	&	&	CCONJ
elm-5391	265	15	jeff	jeff	PROPN
elm-5391	265	16	dean	dean	PROPN
elm-5391	265	17	.	.	PUNCT
elm-5391	266	1	2013	2013	NUM
elm-5391	266	2	.	.	PUNCT
elm-5391	267	1	distributed	distribute	VERB
elm-5391	267	2	representations	representation	NOUN
elm-5391	267	3	of	of	ADP
elm-5391	267	4	words	word	NOUN
elm-5391	267	5	and	and	CCONJ
elm-5391	267	6	phrases	phrase	NOUN
elm-5391	267	7	and	and	CCONJ
elm-5391	267	8	their	their	PRON
elm-5391	267	9	compositionality	compositionality	NOUN
elm-5391	267	10	.	.	PUNCT
elm-5391	268	1	in	in	ADP
elm-5391	268	2	advances	advance	NOUN
elm-5391	268	3	in	in	ADP
elm-5391	268	4	neural	neural	ADJ
elm-5391	268	5	information	information	NOUN
elm-5391	268	6	processing	processing	NOUN
elm-5391	268	7	systems	system	NOUN
elm-5391	268	8	(	(	PUNCT
elm-5391	268	9	neurips	neurip	NOUN
elm-5391	268	10	)	)	PUNCT
elm-5391	268	11	,	,	PUNCT
elm-5391	268	12	3111–3119	3111–3119	NUM
elm-5391	268	13	.	.	PUNCT
elm-5391	269	1	mohammad	mohammad	PROPN
elm-5391	269	2	,	,	PUNCT
elm-5391	269	3	saif	saif	PROPN
elm-5391	269	4	.	.	PUNCT
elm-5391	269	5	2018	2018	NUM
elm-5391	269	6	.	.	PUNCT
elm-5391	270	1	obtaining	obtain	VERB
elm-5391	270	2	reliable	reliable	ADJ
elm-5391	270	3	human	human	ADJ
elm-5391	270	4	ratings	rating	NOUN
elm-5391	270	5	of	of	ADP
elm-5391	270	6	valence	valence	NOUN
elm-5391	270	7	,	,	PUNCT
elm-5391	270	8	arousal	arousal	NOUN
elm-5391	270	9	,	,	PUNCT
elm-5391	270	10	and	and	CCONJ
elm-5391	270	11	dominance	dominance	NOUN
elm-5391	270	12	for	for	ADP
elm-5391	270	13	20,000	20,000	NUM
elm-5391	270	14	english	english	ADJ
elm-5391	270	15	words	word	NOUN
elm-5391	270	16	.	.	PUNCT
elm-5391	271	1	in	in	ADP
elm-5391	271	2	proceedings	proceeding	NOUN
elm-5391	271	3	of	of	ADP
elm-5391	271	4	the	the	DET
elm-5391	271	5	56th	56th	ADJ
elm-5391	271	6	annual	annual	ADJ
elm-5391	271	7	meeting	meeting	NOUN
elm-5391	271	8	of	of	ADP
elm-5391	271	9	the	the	DET
elm-5391	271	10	association	association	NOUN
elm-5391	271	11	for	for	ADP
elm-5391	271	12	computational	computational	ADJ
elm-5391	271	13	linguistics	linguistic	NOUN
elm-5391	271	14	(	(	PUNCT
elm-5391	271	15	volume	volume	NOUN
elm-5391	271	16	1	1	NUM
elm-5391	271	17	:	:	PUNCT
elm-5391	271	18	long	long	ADJ
elm-5391	271	19	papers	paper	NOUN
elm-5391	271	20	)	)	PUNCT
elm-5391	271	21	,	,	PUNCT
elm-5391	271	22	174–184	174–184	X
elm-5391	271	23	.	.	PUNCT
elm-5391	271	24	melbourne	melbourne	PROPN
elm-5391	271	25	,	,	PUNCT
elm-5391	271	26	australia	australia	PROPN
elm-5391	271	27	:	:	PUNCT
elm-5391	271	28	association	association	NOUN
elm-5391	271	29	for	for	ADP
elm-5391	271	30	computational	computational	ADJ
elm-5391	271	31	linguistics	linguistic	NOUN
elm-5391	271	32	.	.	PUNCT
elm-5391	272	1	10.18653	10.18653	NUM
elm-5391	272	2	/	/	SYM
elm-5391	272	3	v1	v1	NOUN
elm-5391	272	4	/	/	SYM
elm-5391	272	5	p18	p18	NOUN
elm-5391	272	6	-	-	PUNCT
elm-5391	272	7	1017	1017	NUM
elm-5391	272	8	.	.	PUNCT
elm-5391	273	1	https://aclanthology.org/	https://aclanthology.org/	VERB
elm-5391	273	2	p18	p18	NOUN
elm-5391	273	3	-	-	PUNCT
elm-5391	273	4	1017	1017	NUM
elm-5391	273	5	.	.	PUNCT
elm-5391	274	1	nair	nair	PROPN
elm-5391	274	2	,	,	PUNCT
elm-5391	274	3	sathvik	sathvik	PROPN
elm-5391	274	4	,	,	PUNCT
elm-5391	274	5	mahesh	mahesh	PROPN
elm-5391	274	6	srinivasan	srinivasan	PROPN
elm-5391	274	7	&	&	CCONJ
elm-5391	274	8	stephan	stephan	PROPN
elm-5391	274	9	c.	c.	PROPN
elm-5391	274	10	meylan	meylan	PROPN
elm-5391	274	11	.	.	PUNCT
elm-5391	275	1	2020	2020	NUM
elm-5391	275	2	.	.	PUNCT
elm-5391	276	1	contextualized	contextualize	VERB
elm-5391	276	2	word	word	NOUN
elm-5391	276	3	embeddings	embedding	NOUN
elm-5391	276	4	encode	encode	ADJ
elm-5391	276	5	aspects	aspect	NOUN
elm-5391	276	6	of	of	ADP
elm-5391	276	7	human	human	NOUN
elm-5391	276	8	-	-	PUNCT
elm-5391	276	9	like	like	ADJ
elm-5391	276	10	word	word	NOUN
elm-5391	276	11	sense	sense	NOUN
elm-5391	276	12	knowledge	knowledge	NOUN
elm-5391	276	13	.	.	PUNCT
elm-5391	277	1	proceedings	proceeding	NOUN
elm-5391	277	2	of	of	ADP
elm-5391	277	3	the	the	DET
elm-5391	277	4	cognitive	cognitive	ADJ
elm-5391	277	5	aspects	aspect	NOUN
elm-5391	277	6	of	of	ADP
elm-5391	277	7	the	the	DET
elm-5391	277	8	lexicon	lexicon	ADJ
elm-5391	277	9	workshop	workshop	NOUN
elm-5391	277	10	at	at	ADP
elm-5391	277	11	the	the	DET
elm-5391	277	12	28th	28th	ADJ
elm-5391	277	13	international	international	ADJ
elm-5391	277	14	conference	conference	NOUN
elm-5391	277	15	on	on	ADP
elm-5391	277	16	computational	computational	ADJ
elm-5391	277	17	linguistics	linguistic	NOUN
elm-5391	277	18	(	(	PUNCT
elm-5391	277	19	coling	cole	VERB
elm-5391	277	20	)	)	PUNCT
elm-5391	277	21	https://arxiv.org/abs/2010.13057	https://arxiv.org/abs/2010.13057	NOUN
elm-5391	277	22	.	.	PUNCT
elm-5391	278	1	osgood	osgood	PROPN
elm-5391	278	2	,	,	PUNCT
elm-5391	278	3	charles	charles	PROPN
elm-5391	278	4	egerton	egerton	PROPN
elm-5391	278	5	,	,	PUNCT
elm-5391	278	6	george	george	PROPN
elm-5391	278	7	j	j	PROPN
elm-5391	278	8	suci	suci	PROPN
elm-5391	278	9	&	&	CCONJ
elm-5391	278	10	percy	percy	PROPN
elm-5391	278	11	h	h	PROPN
elm-5391	278	12	tannenbaum	tannenbaum	PROPN
elm-5391	278	13	.	.	PUNCT
elm-5391	279	1	1957	1957	NUM
elm-5391	279	2	.	.	PUNCT
elm-5391	280	1	the	the	DET
elm-5391	280	2	measurement	measurement	NOUN
elm-5391	280	3	of	of	ADP
elm-5391	280	4	meaning	mean	VERB
elm-5391	280	5	47	47	NUM
elm-5391	280	6	.	.	PUNCT
elm-5391	281	1	university	university	PROPN
elm-5391	281	2	of	of	ADP
elm-5391	281	3	illinois	illinois	PROPN
elm-5391	281	4	press	press	PROPN
elm-5391	281	5	.	.	PUNCT
elm-5391	282	1	pandia	pandia	PROPN
elm-5391	282	2	,	,	PUNCT
elm-5391	282	3	lalchand	lalchand	PROPN
elm-5391	282	4	,	,	PUNCT
elm-5391	282	5	yan	yan	PROPN
elm-5391	282	6	cong	cong	PROPN
elm-5391	282	7	&	&	CCONJ
elm-5391	282	8	allyson	allyson	PROPN
elm-5391	282	9	ettinger	ettinger	PROPN
elm-5391	282	10	.	.	PROPN
elm-5391	282	11	2021	2021	NUM
elm-5391	282	12	.	.	PUNCT
elm-5391	283	1	pragmatic	pragmatic	ADJ
elm-5391	283	2	competence	competence	NOUN
elm-5391	283	3	of	of	ADP
elm-5391	283	4	pre	pre	ADJ
elm-5391	283	5	-	-	ADJ
elm-5391	283	6	trained	train	VERB
elm-5391	283	7	language	language	NOUN
elm-5391	283	8	models	model	NOUN
elm-5391	283	9	through	through	ADP
elm-5391	283	10	the	the	DET
elm-5391	283	11	lens	lens	NOUN
elm-5391	283	12	of	of	ADP
elm-5391	283	13	discourse	discourse	NOUN
elm-5391	283	14	connectives	connective	NOUN
elm-5391	283	15	.	.	PUNCT
elm-5391	284	1	arxiv	arxiv	NOUN
elm-5391	284	2	preprint:2109.12951	preprint:2109.12951	ADJ
elm-5391	284	3	.	.	PUNCT
elm-5391	285	1	proceedings	proceeding	NOUN
elm-5391	285	2	of	of	ADP
elm-5391	285	3	elm	elm	NOUN
elm-5391	285	4	2	2	NUM
elm-5391	285	5	:	:	PUNCT
elm-5391	285	6	310	310	NUM
elm-5391	285	7	-	-	SYM
elm-5391	285	8	321	321	NUM
elm-5391	285	9	,	,	PUNCT
elm-5391	285	10	2023	2023	NUM
elm-5391	285	11	yuhan	yuhan	PROPN
elm-5391	285	12	zhang	zhang	PROPN
elm-5391	285	13	,	,	PUNCT
elm-5391	285	14	wenqi	wenqi	PROPN
elm-5391	285	15	chen	chen	PROPN
elm-5391	285	16	,	,	PUNCT
elm-5391	285	17	ruihan	ruihan	PROPN
elm-5391	285	18	zhang	zhang	PROPN
elm-5391	285	19	and	and	CCONJ
elm-5391	285	20	xiajie	xiajie	PROPN
elm-5391	285	21	zhang	zhang	PROPN
elm-5391	285	22	:	:	PUNCT
elm-5391	285	23	representing	represent	VERB
elm-5391	285	24	affect	affect	VERB
elm-5391	285	25	information	information	NOUN
elm-5391	285	26	in	in	ADP
elm-5391	285	27	word	word	NOUN
elm-5391	285	28	embeddings	embedding	NOUN
elm-5391	285	29	.	.	PUNCT
elm-5391	286	1	319	319	NUM
elm-5391	286	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	286	3	https://www.elm-conference.net/	https://www.elm-conference.net/	ADP
elm-5391	286	4	pedregosa	pedregosa	PROPN
elm-5391	286	5	,	,	PUNCT
elm-5391	286	6	f.	f.	PROPN
elm-5391	286	7	,	,	PUNCT
elm-5391	286	8	g.	g.	PROPN
elm-5391	286	9	varoquaux	varoquaux	PROPN
elm-5391	286	10	,	,	PUNCT
elm-5391	286	11	a.	a.	NOUN
elm-5391	286	12	gramfort	gramfort	NOUN
elm-5391	286	13	,	,	PUNCT
elm-5391	286	14	v.	v.	PROPN
elm-5391	286	15	michel	michel	PROPN
elm-5391	286	16	,	,	PUNCT
elm-5391	286	17	b.	b.	PROPN
elm-5391	286	18	thirion	thirion	PROPN
elm-5391	286	19	,	,	PUNCT
elm-5391	286	20	o.	o.	PROPN
elm-5391	286	21	grisel	grisel	PROPN
elm-5391	286	22	,	,	PUNCT
elm-5391	286	23	m.	m.	NOUN
elm-5391	286	24	blondel	blondel	PROPN
elm-5391	286	25	,	,	PUNCT
elm-5391	286	26	p.	p.	NOUN
elm-5391	286	27	prettenhofer	prettenhofer	NOUN
elm-5391	286	28	,	,	PUNCT
elm-5391	286	29	r.	r.	PROPN
elm-5391	286	30	weiss	weiss	PROPN
elm-5391	286	31	,	,	PUNCT
elm-5391	286	32	v.	v.	ADP
elm-5391	286	33	dubourg	dubourg	PROPN
elm-5391	286	34	,	,	PUNCT
elm-5391	286	35	j.	j.	PROPN
elm-5391	286	36	vanderplas	vanderplas	PROPN
elm-5391	286	37	,	,	PUNCT
elm-5391	286	38	a.	a.	NOUN
elm-5391	286	39	passos	passos	PROPN
elm-5391	286	40	,	,	PUNCT
elm-5391	286	41	d.	d.	PROPN
elm-5391	286	42	cournapeau	cournapeau	PROPN
elm-5391	286	43	,	,	PUNCT
elm-5391	286	44	m.	m.	NOUN
elm-5391	286	45	brucher	brucher	ADJ
elm-5391	286	46	,	,	PUNCT
elm-5391	286	47	m.	m.	NOUN
elm-5391	286	48	perrot	perrot	PROPN
elm-5391	286	49	&	&	CCONJ
elm-5391	286	50	e.	e.	PROPN
elm-5391	286	51	duchesnay	duchesnay	PROPN
elm-5391	286	52	.	.	PUNCT
elm-5391	287	1	2011	2011	NUM
elm-5391	287	2	.	.	PUNCT
elm-5391	288	1	scikit	scikit	NOUN
elm-5391	288	2	-	-	PUNCT
elm-5391	288	3	learn	learn	VERB
elm-5391	288	4	:	:	PUNCT
elm-5391	288	5	machine	machine	NOUN
elm-5391	288	6	learning	learning	NOUN
elm-5391	288	7	in	in	ADP
elm-5391	288	8	python	python	PROPN
elm-5391	288	9	.	.	PUNCT
elm-5391	289	1	journal	journal	PROPN
elm-5391	289	2	of	of	ADP
elm-5391	289	3	machine	machine	NOUN
elm-5391	289	4	learning	learn	VERB
elm-5391	289	5	research	research	NOUN
elm-5391	289	6	12	12	NUM
elm-5391	289	7	.	.	PUNCT
elm-5391	290	1	2825–2830	2825–2830	NUM
elm-5391	290	2	.	.	PUNCT
elm-5391	291	1	pennington	pennington	PROPN
elm-5391	291	2	,	,	PUNCT
elm-5391	291	3	jeffrey	jeffrey	PROPN
elm-5391	291	4	,	,	PUNCT
elm-5391	291	5	richard	richard	PROPN
elm-5391	291	6	socher	socher	PROPN
elm-5391	291	7	&	&	CCONJ
elm-5391	291	8	christopher	christopher	PROPN
elm-5391	291	9	d.	d.	PROPN
elm-5391	291	10	manning	manning	PROPN
elm-5391	291	11	.	.	PUNCT
elm-5391	292	1	2014	2014	NUM
elm-5391	292	2	.	.	PUNCT
elm-5391	293	1	glove	glove	NOUN
elm-5391	293	2	:	:	PUNCT
elm-5391	293	3	global	global	ADJ
elm-5391	293	4	vectors	vector	NOUN
elm-5391	293	5	for	for	ADP
elm-5391	293	6	word	word	NOUN
elm-5391	293	7	representation	representation	NOUN
elm-5391	293	8	.	.	PUNCT
elm-5391	294	1	in	in	ADP
elm-5391	294	2	empirical	empirical	ADJ
elm-5391	294	3	methods	method	NOUN
elm-5391	294	4	in	in	ADP
elm-5391	294	5	natural	natural	ADJ
elm-5391	294	6	language	language	NOUN
elm-5391	294	7	processing	processing	NOUN
elm-5391	294	8	(	(	PUNCT
elm-5391	294	9	emnlp	emnlp	ADJ
elm-5391	294	10	)	)	PUNCT
elm-5391	294	11	,	,	PUNCT
elm-5391	294	12	1532–1543	1532–1543	NUM
elm-5391	294	13	.	.	PUNCT
elm-5391	294	14	http://www.aclweb.org/anthology/d14-1162	http://www.aclweb.org/anthology/d14-1162	SYM
elm-5391	294	15	.	.	PUNCT
elm-5391	294	16	picard	picard	PROPN
elm-5391	294	17	,	,	PUNCT
elm-5391	294	18	rosalind	rosalind	PROPN
elm-5391	294	19	w.	w.	PROPN
elm-5391	294	20	2000	2000	NUM
elm-5391	294	21	.	.	PUNCT
elm-5391	295	1	affective	affective	ADJ
elm-5391	295	2	computing	computing	NOUN
elm-5391	295	3	.	.	PUNCT
elm-5391	296	1	mit	mit	PROPN
elm-5391	296	2	press	press	NOUN
elm-5391	296	3	.	.	PUNCT
elm-5391	297	1	picard	picard	NOUN
elm-5391	297	2	,	,	PUNCT
elm-5391	297	3	rosalind	rosalind	PROPN
elm-5391	297	4	w.	w.	PROPN
elm-5391	297	5	2003	2003	NUM
elm-5391	297	6	.	.	PUNCT
elm-5391	298	1	affective	affective	ADJ
elm-5391	298	2	computing	computing	NOUN
elm-5391	298	3	:	:	PUNCT
elm-5391	298	4	challenges	challenge	NOUN
elm-5391	298	5	.	.	PUNCT
elm-5391	299	1	international	international	ADJ
elm-5391	299	2	journal	journal	PROPN
elm-5391	299	3	of	of	ADP
elm-5391	299	4	humancomputer	humancomputer	PROPN
elm-5391	299	5	studies	study	NOUN
elm-5391	299	6	59(1	59(1	ADJ
elm-5391	299	7	-	-	SYM
elm-5391	299	8	2	2	NUM
elm-5391	299	9	)	)	PUNCT
elm-5391	299	10	.	.	PUNCT
elm-5391	300	1	55–64	55–64	NUM
elm-5391	300	2	.	.	PUNCT
elm-5391	301	1	poliak	poliak	PROPN
elm-5391	301	2	,	,	PUNCT
elm-5391	301	3	adam	adam	PROPN
elm-5391	301	4	,	,	PUNCT
elm-5391	301	5	aparajita	aparajita	PROPN
elm-5391	301	6	haldar	haldar	NOUN
elm-5391	301	7	,	,	PUNCT
elm-5391	301	8	rachel	rachel	PROPN
elm-5391	301	9	rudinger	rudinger	PROPN
elm-5391	301	10	,	,	PUNCT
elm-5391	301	11	j.	j.	PROPN
elm-5391	301	12	edward	edward	PROPN
elm-5391	301	13	hu	hu	PROPN
elm-5391	301	14	,	,	PUNCT
elm-5391	301	15	ellie	ellie	PROPN
elm-5391	301	16	pavlick	pavlick	PROPN
elm-5391	301	17	,	,	PUNCT
elm-5391	301	18	aaron	aaron	PROPN
elm-5391	301	19	steven	steven	PROPN
elm-5391	301	20	white	white	PROPN
elm-5391	301	21	&	&	CCONJ
elm-5391	301	22	benjamin	benjamin	PROPN
elm-5391	301	23	van	van	PROPN
elm-5391	301	24	durme	durme	PROPN
elm-5391	301	25	.	.	PUNCT
elm-5391	302	1	2018	2018	NUM
elm-5391	302	2	.	.	PUNCT
elm-5391	303	1	collecting	collect	VERB
elm-5391	303	2	diverse	diverse	ADJ
elm-5391	303	3	natural	natural	ADJ
elm-5391	303	4	language	language	NOUN
elm-5391	303	5	inference	inference	NOUN
elm-5391	303	6	problems	problem	NOUN
elm-5391	303	7	for	for	ADP
elm-5391	303	8	sentence	sentence	NOUN
elm-5391	303	9	representation	representation	NOUN
elm-5391	303	10	evaluation	evaluation	NOUN
elm-5391	303	11	.	.	PUNCT
elm-5391	304	1	in	in	ADP
elm-5391	304	2	proceedings	proceeding	NOUN
elm-5391	304	3	of	of	ADP
elm-5391	304	4	the	the	DET
elm-5391	304	5	2018	2018	NUM
elm-5391	304	6	emnlp	emnlp	ADJ
elm-5391	304	7	workshop	workshop	NOUN
elm-5391	304	8	blackboxnlp	blackboxnlp	NOUN
elm-5391	304	9	:	:	PUNCT
elm-5391	304	10	analyzing	analyze	VERB
elm-5391	304	11	and	and	CCONJ
elm-5391	304	12	interpreting	interpret	VERB
elm-5391	304	13	neural	neural	ADJ
elm-5391	304	14	networks	network	NOUN
elm-5391	304	15	for	for	ADP
elm-5391	304	16	nlp	nlp	NOUN
elm-5391	304	17	,	,	PUNCT
elm-5391	304	18	337–340	337–340	NUM
elm-5391	304	19	.	.	PUNCT
elm-5391	304	20	brussels	brussels	PROPN
elm-5391	304	21	,	,	PUNCT
elm-5391	304	22	belgium	belgium	PROPN
elm-5391	304	23	:	:	PUNCT
elm-5391	304	24	association	association	NOUN
elm-5391	304	25	for	for	ADP
elm-5391	304	26	computational	computational	ADJ
elm-5391	304	27	linguistics	linguistic	NOUN
elm-5391	304	28	.	.	PUNCT
elm-5391	305	1	10.18653	10.18653	NUM
elm-5391	305	2	/	/	SYM
elm-5391	305	3	v1	v1	NOUN
elm-5391	305	4	/	/	SYM
elm-5391	305	5	w18	w18	NOUN
elm-5391	305	6	-	-	PUNCT
elm-5391	305	7	5441	5441	NUM
elm-5391	305	8	.	.	PUNCT
elm-5391	306	1	https	https	NOUN
elm-5391	306	2	:	:	PUNCT
elm-5391	306	3	//aclanthology.org	//aclanthology.org	ADJ
elm-5391	306	4	/	/	SYM
elm-5391	306	5	w18	w18	NOUN
elm-5391	306	6	-	-	PUNCT
elm-5391	306	7	5441	5441	NUM
elm-5391	306	8	.	.	PUNCT
elm-5391	307	1	rashkin	rashkin	PROPN
elm-5391	307	2	,	,	PUNCT
elm-5391	307	3	hannah	hannah	PROPN
elm-5391	307	4	,	,	PUNCT
elm-5391	307	5	eric	eric	PROPN
elm-5391	307	6	michael	michael	PROPN
elm-5391	307	7	smith	smith	PROPN
elm-5391	307	8	,	,	PUNCT
elm-5391	307	9	margaret	margaret	PROPN
elm-5391	307	10	li	li	PROPN
elm-5391	307	11	&	&	CCONJ
elm-5391	307	12	y	y	PROPN
elm-5391	307	13	-	-	PUNCT
elm-5391	307	14	lan	lan	PROPN
elm-5391	307	15	boureau	boureau	NOUN
elm-5391	307	16	.	.	PUNCT
elm-5391	308	1	2018	2018	NUM
elm-5391	308	2	.	.	PUNCT
elm-5391	309	1	towards	towards	ADP
elm-5391	309	2	empathetic	empathetic	ADJ
elm-5391	309	3	open	open	ADJ
elm-5391	309	4	-	-	PUNCT
elm-5391	309	5	domain	domain	NOUN
elm-5391	309	6	conversation	conversation	NOUN
elm-5391	309	7	models	model	NOUN
elm-5391	309	8	:	:	PUNCT
elm-5391	309	9	a	a	DET
elm-5391	309	10	new	new	ADJ
elm-5391	309	11	benchmark	benchmark	NOUN
elm-5391	309	12	and	and	CCONJ
elm-5391	309	13	dataset	dataset	NOUN
elm-5391	309	14	.	.	PUNCT
elm-5391	310	1	arxiv	arxiv	PROPN
elm-5391	310	2	preprint	preprint	VERB
elm-5391	310	3	arxiv:1811.00207	arxiv:1811.00207	NOUN
elm-5391	310	4	.	.	PUNCT
elm-5391	311	1	russell	russell	PROPN
elm-5391	311	2	,	,	PUNCT
elm-5391	311	3	james	james	PROPN
elm-5391	311	4	a.	a.	PROPN
elm-5391	311	5	1980	1980	NUM
elm-5391	311	6	.	.	PUNCT
elm-5391	312	1	a	a	DET
elm-5391	312	2	circumplex	circumplex	NOUN
elm-5391	312	3	model	model	NOUN
elm-5391	312	4	of	of	ADP
elm-5391	312	5	affect	affect	PROPN
elm-5391	312	6	.	.	PUNCT
elm-5391	313	1	journal	journal	NOUN
elm-5391	313	2	of	of	ADP
elm-5391	313	3	personality	personality	NOUN
elm-5391	313	4	and	and	CCONJ
elm-5391	313	5	social	social	ADJ
elm-5391	313	6	psychology	psychology	NOUN
elm-5391	313	7	39(6	39(6	NUM
elm-5391	313	8	)	)	PUNCT
elm-5391	313	9	.	.	PUNCT
elm-5391	314	1	1161	1161	NUM
elm-5391	314	2	.	.	PUNCT
elm-5391	315	1	russell	russell	PROPN
elm-5391	315	2	,	,	PUNCT
elm-5391	315	3	james	james	PROPN
elm-5391	315	4	a.	a.	PROPN
elm-5391	315	5	2003	2003	NUM
elm-5391	315	6	.	.	PUNCT
elm-5391	316	1	core	core	NOUN
elm-5391	316	2	affect	affect	VERB
elm-5391	316	3	and	and	CCONJ
elm-5391	316	4	the	the	DET
elm-5391	316	5	psychological	psychological	ADJ
elm-5391	316	6	construction	construction	NOUN
elm-5391	316	7	of	of	ADP
elm-5391	316	8	emotion	emotion	NOUN
elm-5391	316	9	.	.	PUNCT
elm-5391	317	1	psychological	psychological	ADJ
elm-5391	317	2	review	review	NOUN
elm-5391	317	3	110(1	110(1	NUM
elm-5391	317	4	)	)	PUNCT
elm-5391	317	5	.	.	PUNCT
elm-5391	318	1	145	145	NUM
elm-5391	318	2	.	.	X
elm-5391	318	3	shaver	shaver	PROPN
elm-5391	318	4	,	,	PUNCT
elm-5391	318	5	phillip	phillip	PROPN
elm-5391	318	6	,	,	PUNCT
elm-5391	318	7	judith	judith	PROPN
elm-5391	318	8	schwartz	schwartz	PROPN
elm-5391	318	9	,	,	PUNCT
elm-5391	318	10	donald	donald	PROPN
elm-5391	318	11	kirson	kirson	PROPN
elm-5391	318	12	&	&	CCONJ
elm-5391	318	13	cary	cary	PROPN
elm-5391	318	14	o’connor	o’connor	ADV
elm-5391	318	15	.	.	PUNCT
elm-5391	319	1	1987	1987	NUM
elm-5391	319	2	.	.	PUNCT
elm-5391	320	1	emotion	emotion	NOUN
elm-5391	320	2	knowledge	knowledge	NOUN
elm-5391	320	3	:	:	PUNCT
elm-5391	320	4	further	further	ADJ
elm-5391	320	5	exploration	exploration	NOUN
elm-5391	320	6	of	of	ADP
elm-5391	320	7	a	a	DET
elm-5391	320	8	prototype	prototype	NOUN
elm-5391	320	9	approach	approach	NOUN
elm-5391	320	10	.	.	PUNCT
elm-5391	321	1	journal	journal	NOUN
elm-5391	321	2	of	of	ADP
elm-5391	321	3	personality	personality	NOUN
elm-5391	321	4	and	and	CCONJ
elm-5391	321	5	social	social	ADJ
elm-5391	321	6	psychology	psychology	NOUN
elm-5391	321	7	52(6	52(6	NOUN
elm-5391	321	8	)	)	PUNCT
elm-5391	321	9	.	.	PUNCT
elm-5391	322	1	1061	1061	NUM
elm-5391	322	2	.	.	PUNCT
elm-5391	323	1	socher	socher	PROPN
elm-5391	323	2	,	,	PUNCT
elm-5391	323	3	richard	richard	PROPN
elm-5391	323	4	,	,	PUNCT
elm-5391	323	5	alex	alex	PROPN
elm-5391	323	6	perelygin	perelygin	PROPN
elm-5391	323	7	,	,	PUNCT
elm-5391	323	8	jean	jean	PROPN
elm-5391	323	9	wu	wu	PROPN
elm-5391	323	10	,	,	PUNCT
elm-5391	323	11	jason	jason	PROPN
elm-5391	323	12	chuang	chuang	PROPN
elm-5391	323	13	,	,	PUNCT
elm-5391	323	14	christopher	christopher	PROPN
elm-5391	323	15	d	d	PROPN
elm-5391	323	16	manning	manning	PROPN
elm-5391	323	17	,	,	PUNCT
elm-5391	323	18	andrew	andrew	PROPN
elm-5391	323	19	y	y	PROPN
elm-5391	323	20	ng	ng	PROPN
elm-5391	323	21	&	&	CCONJ
elm-5391	323	22	christopher	christopher	PROPN
elm-5391	323	23	potts	potts	PROPN
elm-5391	323	24	.	.	PUNCT
elm-5391	324	1	2013	2013	NUM
elm-5391	324	2	.	.	PUNCT
elm-5391	325	1	recursive	recursive	ADJ
elm-5391	325	2	deep	deep	ADJ
elm-5391	325	3	models	model	NOUN
elm-5391	325	4	for	for	ADP
elm-5391	325	5	semantic	semantic	ADJ
elm-5391	325	6	compositionality	compositionality	NOUN
elm-5391	325	7	over	over	ADP
elm-5391	325	8	a	a	DET
elm-5391	325	9	sentiment	sentiment	NOUN
elm-5391	325	10	treebank	treebank	VERB
elm-5391	325	11	.	.	PUNCT
elm-5391	326	1	in	in	ADP
elm-5391	326	2	proceedings	proceeding	NOUN
elm-5391	326	3	of	of	ADP
elm-5391	326	4	the	the	DET
elm-5391	326	5	2013	2013	NUM
elm-5391	326	6	conference	conference	NOUN
elm-5391	326	7	on	on	ADP
elm-5391	326	8	empirical	empirical	ADJ
elm-5391	326	9	methods	method	NOUN
elm-5391	326	10	in	in	ADP
elm-5391	326	11	natural	natural	ADJ
elm-5391	326	12	language	language	NOUN
elm-5391	326	13	processing	processing	NOUN
elm-5391	326	14	,	,	PUNCT
elm-5391	326	15	1631–1642	1631–1642	NUM
elm-5391	326	16	.	.	PUNCT
elm-5391	326	17	suresh	suresh	PROPN
elm-5391	326	18	,	,	PUNCT
elm-5391	326	19	varsha	varsha	PROPN
elm-5391	326	20	&	&	CCONJ
elm-5391	326	21	desmond	desmond	PROPN
elm-5391	326	22	c	c	PROPN
elm-5391	326	23	ong	ong	PROPN
elm-5391	326	24	.	.	PROPN
elm-5391	326	25	2021	2021	NUM
elm-5391	326	26	.	.	PUNCT
elm-5391	327	1	using	use	VERB
elm-5391	327	2	knowledge	knowledge	NOUN
elm-5391	327	3	-	-	PUNCT
elm-5391	327	4	embedded	embed	VERB
elm-5391	327	5	attention	attention	NOUN
elm-5391	327	6	to	to	PART
elm-5391	327	7	augment	augment	VERB
elm-5391	327	8	pretrained	pretraine	VERB
elm-5391	327	9	language	language	NOUN
elm-5391	327	10	models	model	NOUN
elm-5391	327	11	for	for	ADP
elm-5391	327	12	fine	fine	ADV
elm-5391	327	13	-	-	PUNCT
elm-5391	327	14	grained	grain	VERB
elm-5391	327	15	emotion	emotion	NOUN
elm-5391	327	16	recognition	recognition	NOUN
elm-5391	327	17	.	.	PUNCT
elm-5391	328	1	in	in	ADP
elm-5391	328	2	2021	2021	NUM
elm-5391	328	3	9th	9th	ADJ
elm-5391	328	4	international	international	ADJ
elm-5391	328	5	conference	conference	NOUN
elm-5391	328	6	on	on	ADP
elm-5391	328	7	affective	affective	ADJ
elm-5391	328	8	computing	computing	NOUN
elm-5391	328	9	and	and	CCONJ
elm-5391	328	10	intelligent	intelligent	ADJ
elm-5391	328	11	interaction	interaction	NOUN
elm-5391	328	12	(	(	PUNCT
elm-5391	328	13	acii	acii	ADJ
elm-5391	328	14	)	)	PUNCT
elm-5391	328	15	,	,	PUNCT
elm-5391	328	16	1–8	1–8	X
elm-5391	328	17	.	.	PUNCT
elm-5391	328	18	ieee	ieee	PROPN
elm-5391	328	19	.	.	PUNCT
elm-5391	329	1	tao	tao	PROPN
elm-5391	329	2	,	,	PUNCT
elm-5391	329	3	jianhua	jianhua	PROPN
elm-5391	329	4	&	&	CCONJ
elm-5391	329	5	tieniu	tieniu	PROPN
elm-5391	329	6	tan	tan	PROPN
elm-5391	329	7	.	.	PUNCT
elm-5391	330	1	2005	2005	NUM
elm-5391	330	2	.	.	PUNCT
elm-5391	331	1	affective	affective	ADJ
elm-5391	331	2	computing	computing	NOUN
elm-5391	331	3	:	:	PUNCT
elm-5391	331	4	a	a	DET
elm-5391	331	5	review	review	NOUN
elm-5391	331	6	.	.	PUNCT
elm-5391	332	1	in	in	ADP
elm-5391	332	2	international	international	ADJ
elm-5391	332	3	conference	conference	NOUN
elm-5391	332	4	on	on	ADP
elm-5391	332	5	affective	affective	ADJ
elm-5391	332	6	computing	computing	NOUN
elm-5391	332	7	and	and	CCONJ
elm-5391	332	8	intelligent	intelligent	ADJ
elm-5391	332	9	interaction	interaction	NOUN
elm-5391	332	10	,	,	PUNCT
elm-5391	332	11	981–995	981–995	NUM
elm-5391	332	12	.	.	PUNCT
elm-5391	333	1	springer	springer	NOUN
elm-5391	333	2	.	.	PUNCT
elm-5391	334	1	wang	wang	PROPN
elm-5391	334	2	,	,	PUNCT
elm-5391	334	3	alex	alex	PROPN
elm-5391	334	4	,	,	PUNCT
elm-5391	334	5	amanpreet	amanpreet	VERB
elm-5391	334	6	singh	singh	PROPN
elm-5391	334	7	,	,	PUNCT
elm-5391	334	8	julian	julian	PROPN
elm-5391	334	9	michael	michael	PROPN
elm-5391	334	10	,	,	PUNCT
elm-5391	334	11	felix	felix	PROPN
elm-5391	334	12	hill	hill	PROPN
elm-5391	334	13	,	,	PUNCT
elm-5391	334	14	omer	omer	PROPN
elm-5391	334	15	levy	levy	PROPN
elm-5391	334	16	&	&	CCONJ
elm-5391	334	17	samuel	samuel	PROPN
elm-5391	334	18	bowman	bowman	PROPN
elm-5391	334	19	.	.	PROPN
elm-5391	334	20	2018	2018	NUM
elm-5391	334	21	.	.	PUNCT
elm-5391	335	1	glue	glue	NOUN
elm-5391	335	2	:	:	PUNCT
elm-5391	335	3	a	a	DET
elm-5391	335	4	multi	multi	ADJ
elm-5391	335	5	-	-	ADJ
elm-5391	335	6	task	task	ADJ
elm-5391	335	7	benchmark	benchmark	NOUN
elm-5391	335	8	and	and	CCONJ
elm-5391	335	9	analysis	analysis	NOUN
elm-5391	335	10	platform	platform	NOUN
elm-5391	335	11	for	for	ADP
elm-5391	335	12	natural	natural	ADJ
elm-5391	335	13	language	language	NOUN
elm-5391	335	14	understanding	understanding	NOUN
elm-5391	335	15	.	.	PUNCT
elm-5391	336	1	in	in	ADP
elm-5391	336	2	proceedings	proceeding	NOUN
elm-5391	336	3	of	of	ADP
elm-5391	336	4	the	the	DET
elm-5391	336	5	2018	2018	NUM
elm-5391	336	6	emnlp	emnlp	ADJ
elm-5391	336	7	workshop	workshop	NOUN
elm-5391	336	8	blackboxnlp	blackboxnlp	NOUN
elm-5391	336	9	:	:	PUNCT
elm-5391	336	10	analyzing	analyze	VERB
elm-5391	336	11	and	and	CCONJ
elm-5391	336	12	interpreting	interpret	VERB
elm-5391	336	13	neural	neural	ADJ
elm-5391	336	14	networks	network	NOUN
elm-5391	336	15	for	for	ADP
elm-5391	336	16	nlp	nlp	NOUN
elm-5391	336	17	,	,	PUNCT
elm-5391	336	18	353–355	353–355	NUM
elm-5391	336	19	.	.	PUNCT
elm-5391	336	20	brussels	brussels	PROPN
elm-5391	336	21	,	,	PUNCT
elm-5391	336	22	belgium	belgium	PROPN
elm-5391	336	23	:	:	PUNCT
elm-5391	336	24	association	association	NOUN
elm-5391	336	25	for	for	ADP
elm-5391	336	26	computational	computational	ADJ
elm-5391	336	27	linguistics	linguistic	NOUN
elm-5391	336	28	.	.	PUNCT
elm-5391	337	1	10.18653	10.18653	NUM
elm-5391	337	2	/	/	SYM
elm-5391	337	3	v1	v1	NOUN
elm-5391	337	4	/	/	SYM
elm-5391	337	5	w18	w18	NOUN
elm-5391	337	6	-	-	PUNCT
elm-5391	337	7	5446	5446	NUM
elm-5391	337	8	.	.	PUNCT
elm-5391	338	1	https://aclanthology.org/w18-5446	https://aclanthology.org/w18-5446	NOUN
elm-5391	338	2	.	.	NOUN
elm-5391	338	3	warriner	warriner	NOUN
elm-5391	338	4	,	,	PUNCT
elm-5391	338	5	amy	amy	PROPN
elm-5391	338	6	beth	beth	PROPN
elm-5391	338	7	,	,	PUNCT
elm-5391	338	8	victor	victor	PROPN
elm-5391	338	9	kuperman	kuperman	PROPN
elm-5391	338	10	&	&	CCONJ
elm-5391	338	11	marc	marc	PROPN
elm-5391	338	12	brysbaert	brysbaert	PROPN
elm-5391	338	13	.	.	PUNCT
elm-5391	339	1	2013	2013	NUM
elm-5391	339	2	.	.	PUNCT
elm-5391	340	1	norms	norm	NOUN
elm-5391	340	2	of	of	ADP
elm-5391	340	3	valence	valence	NOUN
elm-5391	340	4	,	,	PUNCT
elm-5391	340	5	arousal	arousal	NOUN
elm-5391	340	6	,	,	PUNCT
elm-5391	340	7	and	and	CCONJ
elm-5391	340	8	dominance	dominance	NOUN
elm-5391	340	9	for	for	ADP
elm-5391	340	10	13,915	13,915	NUM
elm-5391	340	11	english	english	PROPN
elm-5391	340	12	lemmas	lemmas	PROPN
elm-5391	340	13	.	.	PUNCT
elm-5391	341	1	behavior	behavior	NOUN
elm-5391	341	2	research	research	NOUN
elm-5391	341	3	methods	method	NOUN
elm-5391	341	4	45(4	45(4	NOUN
elm-5391	341	5	)	)	PUNCT
elm-5391	341	6	.	.	PUNCT
elm-5391	342	1	1191–1207	1191–1207	NUM
elm-5391	342	2	.	.	PUNCT
elm-5391	343	1	proceedings	proceeding	NOUN
elm-5391	343	2	of	of	ADP
elm-5391	343	3	elm	elm	NOUN
elm-5391	343	4	2	2	NUM
elm-5391	343	5	:	:	PUNCT
elm-5391	343	6	310	310	NUM
elm-5391	343	7	-	-	SYM
elm-5391	343	8	321	321	NUM
elm-5391	343	9	,	,	PUNCT
elm-5391	343	10	2023	2023	NUM
elm-5391	343	11	yuhan	yuhan	PROPN
elm-5391	343	12	zhang	zhang	PROPN
elm-5391	343	13	,	,	PUNCT
elm-5391	343	14	wenqi	wenqi	PROPN
elm-5391	343	15	chen	chen	PROPN
elm-5391	343	16	,	,	PUNCT
elm-5391	343	17	ruihan	ruihan	PROPN
elm-5391	343	18	zhang	zhang	PROPN
elm-5391	343	19	and	and	CCONJ
elm-5391	343	20	xiajie	xiajie	PROPN
elm-5391	343	21	zhang	zhang	PROPN
elm-5391	343	22	:	:	PUNCT
elm-5391	343	23	representing	represent	VERB
elm-5391	343	24	affect	affect	VERB
elm-5391	343	25	information	information	NOUN
elm-5391	343	26	in	in	ADP
elm-5391	343	27	word	word	NOUN
elm-5391	343	28	embeddings	embedding	NOUN
elm-5391	343	29	.	.	PUNCT
elm-5391	344	1	320	320	NUM
elm-5391	344	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	344	3	https://www.elm-conference.net/	https://www.elm-conference.net/	X
elm-5391	344	4	warstadt	warstadt	NOUN
elm-5391	344	5	,	,	PUNCT
elm-5391	344	6	alex	alex	PROPN
elm-5391	344	7	,	,	PUNCT
elm-5391	344	8	alicia	alicia	PROPN
elm-5391	344	9	parrish	parrish	PROPN
elm-5391	344	10	,	,	PUNCT
elm-5391	344	11	haokun	haokun	PROPN
elm-5391	344	12	liu	liu	PROPN
elm-5391	344	13	,	,	PUNCT
elm-5391	344	14	anhad	anhad	VERB
elm-5391	344	15	mohananey	mohananey	NOUN
elm-5391	344	16	,	,	PUNCT
elm-5391	344	17	wei	wei	PROPN
elm-5391	344	18	peng	peng	PROPN
elm-5391	344	19	,	,	PUNCT
elm-5391	344	20	sheng	sheng	PROPN
elm-5391	344	21	-	-	PUNCT
elm-5391	344	22	fu	fu	PROPN
elm-5391	344	23	wang	wang	PROPN
elm-5391	344	24	&	&	CCONJ
elm-5391	344	25	samuel	samuel	PROPN
elm-5391	344	26	r	r	PROPN
elm-5391	344	27	bowman	bowman	PROPN
elm-5391	344	28	.	.	PUNCT
elm-5391	344	29	2020	2020	NUM
elm-5391	344	30	.	.	PUNCT
elm-5391	345	1	blimp	blimp	PROPN
elm-5391	345	2	:	:	PUNCT
elm-5391	345	3	the	the	DET
elm-5391	345	4	benchmark	benchmark	NOUN
elm-5391	345	5	of	of	ADP
elm-5391	345	6	linguistic	linguistic	ADJ
elm-5391	345	7	minimal	minimal	ADJ
elm-5391	345	8	pairs	pair	NOUN
elm-5391	345	9	for	for	ADP
elm-5391	345	10	english	english	PROPN
elm-5391	345	11	.	.	PUNCT
elm-5391	346	1	transactions	transaction	NOUN
elm-5391	346	2	of	of	ADP
elm-5391	346	3	the	the	DET
elm-5391	346	4	association	association	NOUN
elm-5391	346	5	for	for	ADP
elm-5391	346	6	computational	computational	ADJ
elm-5391	346	7	linguistics	linguistic	NOUN
elm-5391	346	8	8	8	NUM
elm-5391	346	9	.	.	PUNCT
elm-5391	347	1	377–392	377–392	NUM
elm-5391	347	2	.	.	NOUN
elm-5391	347	3	wiedemann	wiedemann	PROPN
elm-5391	347	4	,	,	PUNCT
elm-5391	347	5	gregor	gregor	PROPN
elm-5391	347	6	,	,	PUNCT
elm-5391	347	7	steffen	steffen	PROPN
elm-5391	347	8	remus	remus	PROPN
elm-5391	347	9	,	,	PUNCT
elm-5391	347	10	avi	avi	PROPN
elm-5391	347	11	chawla	chawla	PROPN
elm-5391	347	12	&	&	CCONJ
elm-5391	347	13	chris	chris	PROPN
elm-5391	347	14	biemann	biemann	PROPN
elm-5391	347	15	.	.	PUNCT
elm-5391	348	1	2019	2019	NUM
elm-5391	348	2	.	.	PUNCT
elm-5391	349	1	does	do	AUX
elm-5391	349	2	bert	bert	PROPN
elm-5391	349	3	make	make	VERB
elm-5391	349	4	any	any	DET
elm-5391	349	5	sense	sense	NOUN
elm-5391	349	6	?	?	PUNCT
elm-5391	350	1	interpretable	interpretable	ADJ
elm-5391	350	2	word	word	NOUN
elm-5391	350	3	sense	sense	NOUN
elm-5391	350	4	disambiguation	disambiguation	NOUN
elm-5391	350	5	with	with	ADP
elm-5391	350	6	contextualized	contextualized	ADJ
elm-5391	350	7	embeddings	embedding	NOUN
elm-5391	350	8	.	.	PUNCT
elm-5391	351	1	conference	conference	NOUN
elm-5391	351	2	on	on	ADP
elm-5391	351	3	natural	natural	ADJ
elm-5391	351	4	language	language	NOUN
elm-5391	351	5	processing	processing	NOUN
elm-5391	351	6	http://arxiv.org/abs/1909.10430	http://arxiv.org/abs/1909.10430	NOUN
elm-5391	351	7	.	.	PUNCT
elm-5391	352	1	wilcox	wilcox	PROPN
elm-5391	352	2	,	,	PUNCT
elm-5391	352	3	ethan	ethan	PROPN
elm-5391	352	4	,	,	PUNCT
elm-5391	352	5	roger	roger	PROPN
elm-5391	352	6	levy	levy	PROPN
elm-5391	352	7	,	,	PUNCT
elm-5391	352	8	takashi	takashi	PROPN
elm-5391	352	9	morita	morita	PROPN
elm-5391	352	10	&	&	CCONJ
elm-5391	352	11	richard	richard	PROPN
elm-5391	352	12	futrell	futrell	PROPN
elm-5391	352	13	.	.	PUNCT
elm-5391	353	1	2018	2018	NUM
elm-5391	353	2	.	.	PUNCT
elm-5391	354	1	what	what	PRON
elm-5391	354	2	do	do	AUX
elm-5391	354	3	rnn	rnn	VERB
elm-5391	354	4	language	language	NOUN
elm-5391	354	5	models	model	NOUN
elm-5391	354	6	learn	learn	VERB
elm-5391	354	7	about	about	ADP
elm-5391	354	8	filler	filler	NOUN
elm-5391	354	9	–	–	PUNCT
elm-5391	354	10	gap	gap	NOUN
elm-5391	354	11	dependencies	dependency	NOUN
elm-5391	354	12	?	?	PUNCT
elm-5391	355	1	in	in	ADP
elm-5391	355	2	proceedings	proceeding	NOUN
elm-5391	355	3	of	of	ADP
elm-5391	355	4	the	the	DET
elm-5391	355	5	2018	2018	NUM
elm-5391	355	6	emnlp	emnlp	ADJ
elm-5391	355	7	workshop	workshop	NOUN
elm-5391	355	8	blackboxnlp	blackboxnlp	NOUN
elm-5391	355	9	:	:	PUNCT
elm-5391	355	10	analyzing	analyze	VERB
elm-5391	355	11	and	and	CCONJ
elm-5391	355	12	interpreting	interpret	VERB
elm-5391	355	13	neural	neural	ADJ
elm-5391	355	14	networks	network	NOUN
elm-5391	355	15	for	for	ADP
elm-5391	355	16	nlp	nlp	NOUN
elm-5391	355	17	,	,	PUNCT
elm-5391	355	18	211–221	211–221	NUM
elm-5391	355	19	.	.	PUNCT
elm-5391	356	1	brussels	brussels	PROPN
elm-5391	356	2	,	,	PUNCT
elm-5391	356	3	belgium	belgium	PROPN
elm-5391	356	4	:	:	PUNCT
elm-5391	356	5	association	association	NOUN
elm-5391	356	6	for	for	ADP
elm-5391	356	7	computational	computational	ADJ
elm-5391	356	8	linguistics	linguistic	NOUN
elm-5391	356	9	.	.	PUNCT
elm-5391	357	1	10.18653	10.18653	NUM
elm-5391	357	2	/	/	SYM
elm-5391	357	3	v1	v1	NOUN
elm-5391	357	4	/	/	SYM
elm-5391	357	5	w18	w18	NOUN
elm-5391	357	6	-	-	PUNCT
elm-5391	357	7	5423	5423	NUM
elm-5391	357	8	.	.	PUNCT
elm-5391	358	1	https://aclanthology.org/w18-5423	https://aclanthology.org/w18-5423	PROPN
elm-5391	358	2	.	.	PUNCT
elm-5391	359	1	zhang	zhang	PROPN
elm-5391	359	2	,	,	PUNCT
elm-5391	359	3	xiang	xiang	PROPN
elm-5391	359	4	,	,	PUNCT
elm-5391	359	5	junbo	junbo	PROPN
elm-5391	359	6	zhao	zhao	PROPN
elm-5391	359	7	&	&	CCONJ
elm-5391	359	8	yann	yann	PROPN
elm-5391	359	9	lecun	lecun	PROPN
elm-5391	359	10	.	.	PUNCT
elm-5391	360	1	2015	2015	NUM
elm-5391	360	2	.	.	PUNCT
elm-5391	361	1	character	character	NOUN
elm-5391	361	2	-	-	PUNCT
elm-5391	361	3	level	level	NOUN
elm-5391	361	4	convolutional	convolutional	ADJ
elm-5391	361	5	networks	network	NOUN
elm-5391	361	6	for	for	ADP
elm-5391	361	7	text	text	NOUN
elm-5391	361	8	classification	classification	NOUN
elm-5391	361	9	.	.	PUNCT
elm-5391	362	1	advances	advance	NOUN
elm-5391	362	2	in	in	ADP
elm-5391	362	3	neural	neural	ADJ
elm-5391	362	4	information	information	NOUN
elm-5391	362	5	processing	processing	NOUN
elm-5391	362	6	systems	system	NOUN
elm-5391	362	7	(	(	PUNCT
elm-5391	362	8	neurips	neurip	NOUN
elm-5391	362	9	)	)	PUNCT
elm-5391	362	10	28	28	NUM
elm-5391	362	11	.	.	PUNCT
elm-5391	363	1	proceedings	proceeding	NOUN
elm-5391	363	2	of	of	ADP
elm-5391	363	3	elm	elm	NOUN
elm-5391	363	4	2	2	NUM
elm-5391	363	5	:	:	PUNCT
elm-5391	363	6	310	310	NUM
elm-5391	363	7	-	-	SYM
elm-5391	363	8	321	321	NUM
elm-5391	363	9	,	,	PUNCT
elm-5391	363	10	2023	2023	NUM
elm-5391	363	11	yuhan	yuhan	PROPN
elm-5391	363	12	zhang	zhang	PROPN
elm-5391	363	13	,	,	PUNCT
elm-5391	363	14	wenqi	wenqi	PROPN
elm-5391	363	15	chen	chen	PROPN
elm-5391	363	16	,	,	PUNCT
elm-5391	363	17	ruihan	ruihan	PROPN
elm-5391	363	18	zhang	zhang	PROPN
elm-5391	363	19	and	and	CCONJ
elm-5391	363	20	xiajie	xiajie	PROPN
elm-5391	363	21	zhang	zhang	PROPN
elm-5391	363	22	:	:	PUNCT
elm-5391	363	23	representing	represent	VERB
elm-5391	363	24	affect	affect	VERB
elm-5391	363	25	information	information	NOUN
elm-5391	363	26	in	in	ADP
elm-5391	363	27	word	word	NOUN
elm-5391	363	28	embeddings	embedding	NOUN
elm-5391	363	29	.	.	PUNCT
elm-5391	364	1	321	321	NUM
elm-5391	364	2	https://doi.org/10.3765/elm	https://doi.org/10.3765/elm	NOUN
elm-5391	364	3	https://www.elm-conference.net/	https://www.elm-conference.net/	PRON