Theoretical features model the syntactic and semantic properties of the classes described in Section 3. The features used, together with their mean and standard deviation values (computed on all 3,521 adjectives), are summarized in Table 3. A feature value va,i for an adjective lemma a and a feature i corresponds to the proportion of the occurrences in which i is observed for adjective a over all occurrences of a (see Equation (1); f stands for absolute frequency). 
Table 3
Theoretical features. The mean and SD values are computed on all clustered adjectives. Feature copular accounts for predicative constructions with the copula verbs ser, estar (‘be’). Feature predicative accounts for other predicative constructions, such as Example (4d).
| Feature | Textual correlate | Mean | SD |
|---|---|---|---|
| gradable | degree adverbs, degree suffixation | 0.04 | 0.08 |
| comparable | comparative constructions | 0.03 | 0.07 |
| copular | copular predicate syntactic tag | 0.06 | 0.10 |
| predicative | predicate syntactic tag | 0.03 | 0.06 |
| pre-nom | pre-nominal modifier syntactic tag | 0.04 | 0.08 |
| adjacent | first adjective in a series of two or more | 0.03 | 0.05 |
| Feature | Textual correlate | Mean | SD |
|---|---|---|---|
| gradable | degree adverbs, degree suffixation | 0.04 | 0.08 |
| comparable | comparative constructions | 0.03 | 0.07 |
| copular | copular predicate syntactic tag | 0.06 | 0.10 |
| predicative | predicate syntactic tag | 0.03 | 0.06 |
| pre-nom | pre-nominal modifier syntactic tag | 0.04 | 0.08 |
| adjacent | first adjective in a series of two or more | 0.03 | 0.05 |