Table 7 shows the distribution of adjectives in the gold standard into classes according to the three experts. These are the data used in the experiments presented in this section. The proportion of polysemous adjectives is quite high, over 17%, with qualitative-relational being the most frequent type of polysemy. Also note that 51% of the adjectives are qualitative; this will be the baseline for the machine learning experiments presented subsequently.
Gold standard classification: Distribution and examples.
| Class . | Label . | Example . | # . | % . |
|---|---|---|---|---|
| qualitative | Q | tenaç, ‘tenacious’ | 107 | 51.0 |
| event | E | informatiu, ‘informative’ | 37 | 17.6 |
| relational | R | cranià, ‘cranial’ | 30 | 14.3 |
| qualitative-relational | QR | familiar, ‘familiar’ | 23 | 11.0 |
| qualitative-event | QE | sabut, ‘known’ | 7 | 3.3 |
| event-relational | ER | comptable, ‘countable’ | 6 | 2.9 |
| Total | 210 | 100 |
| Class . | Label . | Example . | # . | % . |
|---|---|---|---|---|
| qualitative | Q | tenaç, ‘tenacious’ | 107 | 51.0 |
| event | E | informatiu, ‘informative’ | 37 | 17.6 |
| relational | R | cranià, ‘cranial’ | 30 | 14.3 |
| qualitative-relational | QR | familiar, ‘familiar’ | 23 | 11.0 |
| qualitative-event | QE | sabut, ‘known’ | 7 | 3.3 |
| event-relational | ER | comptable, ‘countable’ | 6 | 2.9 |
| Total | 210 | 100 |