In fact, the class distribution varies with the suffix (see Table 13): Some types, such as -or and the participle, show a clear predominance of the event class (see dark-gray shaded cells); other types, such as -ble, -iu, or -nt, are more spread in their distribution (see light-gray shaded cells). Thus, the suffix seems to influence the resulting readings, with some active-like suffixes building a much more transparent relation to the event (creador ‘creating’, exportador ‘exporting’, recomanat ‘recommended’), and some passive-like or stative suffixes being more prone to creating a stative meaning (contingent ‘contingent’, formidable ‘formidable | terrific’, significatiu ‘significant’). The aspectual class of the deriving verb (Vendler 1957) also plays a role: For instance, although the meaning of abundant (‘abundant’) is related to that of the verb abundar (‘abound’), it clearly has a more stative (property-like) meaning than many of the other event adjectives, due to the fact that the deriving verb is stative. Correspondingly, abundant is classified as qualitative by the Attribute Bagging algorithm. This variation in the morphology–semantics interface is also mirrored in the feature value distributions, as will be shown subsequently.

Table 13 

Contingency table of the most frequent deverbal suffixes (rows) and classification (columns). Dark-gray cells highlight suffixes that mostly create event-related adjectives, and light-gray cells indicate suffixes with a more spread class distribution.

.	Q .	E .	R .	QR .	QE .	ER .	Total .
-ble			0	0	1	1	11
-iu		1	1	2	0		11
-nt			0	0	0	1	11
-or	1		0	0	0	0	11
participle	2		0	0	5	0	15
Total	16	36	3	2	7	6	70

.	Q .	E .	R .	QR .	QE .	ER .	Total .
-ble			0	0	1	1	11
-iu		1	1	2	0		11
-nt			0	0	0	1	11
-or	1		0	0	0	0	11
participle	2		0	0	5	0	15
Total	16	36	3	2	7	6	70

View Large