A Study of Imagery Translation Strategies in English and American Poetry Aided by Natural Language Processing Technology

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Some of the images of deep meaning and their deep meaning

Imagery	Artificial interpretation	Frequency	Artificial interpretation	Frequency
Moon	Think of	16	Think of one’s home	4
Rain	Tears	2	Strong	2
Floating clouds	The weather is unpredictable	2	Wanderer	2
Tears	Sadness	4	Grief	2
Green Mountain	Live in seclusion	6	Hometown	2
Wind	Miserable	2	Fast	2
Ape	Sorrowful	5	Sad	2
Smoke	Decline	2	Dense	2
Spring wind	Good time	3	Love	2
Sunset	Farewell	2	Decline	2

Model comparison experiment results

Model	English-German			English-German
	TED	News	Europarl	IWSLT
Snmt	24.14	27.03	30.81	40.78
IMPR	25.09	23.25	30.23	--
HAN	25.58	26.35	30.78	--
SAN	25.61	25.62	30.27	--
QCN	26.47	23.42	30.39	--
Flat	25.73	24.95	31.55	--
CAEnc	25.21	27.47	31.21	41.01
This method	26.13	28.04	32.15	41.92

High-frequency words of British and American modern poetry text (Top 30)

Sequence	Word	Frequency	Sequence	Word	Frequency
1	the	2477	16	with	289
2	of	1371	17	are	287
3	and	1355	18	Not	287
4	a	1084	19	but	277
5	to	1077	20	on	258
6	in	720	21	one	251
7	I	684	22	They	244
8	it	613	23	at	230
9	is	608	24	you	221
10	that	587	25	we	220
11	was	410	26	his	219
12	for	321	27	have	216
13	as	310	28	all	195
14	he	307	29	an	188
15	be	289	30	or	187

Top 15 high frequency imagery and corresponding imagery

Metonymy type	Image type	Frequency	Source domain-Target domain	Frequency	Metonymy examples
Part-whole	25	46	Location-Person	10	Pretty eyebrows	Belle
			Part-Ship	11	Ship	Set sail
			Component-Instrument	5	String	Musical instrument
Category metaphor	38	42	Scene-View	7	Spring scenery	Falling flower
			Place-Refers to a place in general	5	Place	Paradise
			Person-Person	5	Nobleman	Distant friend
Characteristic metaphor	26	29	Color-Object	8	Painting	Jadeite
			Characteristic-Group	5	Guards of honor	Army
			Characteristics-People	3	Fragrant	Belle
Causal metaphor	2	28	Component-Reason	16	White hair	Aged
Production metaphor	5	11	Materials-Tools	7	Canvas	Weapons
			Birds-Sound	2	Crying bird	Twitter
Possessive metaphor	5	21	Clothing-People	9	Dress	Government officials
			Tools-People	8	Chief minister’s seal	Bureaucrat
			Transportation-People	6	Enemy troops	Nobility
Container metaphor	4	9	Utensils-Beverages	7	Cup	Liquor
			Tools-Things	2	New cooking	Food
Place metaphor	27	40	Landscape-Nonspecific location	14	Hometown	Fields and gardens
			Landscape-Specific location	9	Capital	Area
			Location-People	12	High buildings	Concubines
Time metaphor	9	11	Tools-Time	5	Clock	Dusk
			Poultry-Time	5	Crow	Spring

The theme words of the British and American poetry text (Top 30)

Sequence	Word	Frequency	Topicality	Sequence	Word	Frequency	Topicality
1	the	2476.6	277.321	16	school	21	40.283
2	is	607.6	104.698	17	or	187.6	39.290
3	we	219.8	104.045	18	grade	18.9	37.907
4	English	51.1	98.177	19	white	25.9	32.884
5	a	1084.3	88.043	20	out	102.2	32.490
6	one	250.6	84.459	21	through	40.6	32.471
7	are	287	82.874	22	social	16.1	31.975
8	water	33.6	63.303	23	don’t	20	31.146
9	our	81.2	49.092	24	its	77	29.792
10	students	24.5	48.525	25	snow	14.7	29.457
11	old	38.5	45.868	26	sun	14.7	29.122
12	like	93.8	45.738	27	black	18.2	27.998
13	sea	23.8	44.718	28	years	37.8	26.154
14	lake	21.7	43.124	29	down	63	25.277
15	up	96.6	42.824	30	Englishman	14	24.883

Experimental results on the sentence level model

Model	BLEU	Δ
Snmt	24.14
Snmt+Lenc	20.27	-3.87
CAEnc +Lenc	26.13	1.99