MIT Press

Skip to Main Content

Table 2:

Overview of the composition data sets evaluated in this paper. Reports the total number of phrases (Phrases), the number of unique words in the first (w1) and second (w2) position, as well as the number of unique words in either position (w1 & w2) for each language, phrase type pair.

Language	Phrase Type	Phrases	w1	w2	w1 & w2
German	Adj-Noun	119,434	7,494	16,557	24,031
	Compounds	32,246	5,661	4,899	8,079
	Adv-Adj	23,488	1,785	5,123	5,905
English	Adj-Noun	238,975	8,210	13,045	20,644
	Compounds	16,978	3,689	3,476	5,408
	Adv-Adj	23,148	820	3,086	3,817
Dutch	Adj-Noun	83,392	4,999	10,936	15,744
	Compounds	17,773	3,604	3,495	5,317
	Adv-Adj	4,540	476	1,050	1,335

Language	Phrase Type	Phrases	w1	w2	w1 & w2
German	Adj-Noun	119,434	7,494	16,557	24,031
	Compounds	32,246	5,661	4,899	8,079
	Adv-Adj	23,488	1,785	5,123	5,905
English	Adj-Noun	238,975	8,210	13,045	20,644
	Compounds	16,978	3,689	3,476	5,408
	Adv-Adj	23,148	820	3,086	3,817
Dutch	Adj-Noun	83,392	4,999	10,936	15,744
	Compounds	17,773	3,604	3,495	5,317
	Adv-Adj	4,540	476	1,050	1,335