Table 4.
Lexical categorization scores for the lexical-distributional model (L-D) and baseline model in Simulation 2, averaged across all ten corpora.
| L-D | baseline | |
|---|---|---|
| F-score | 0.799/0.854 | 0.523 |
| Variation of information | 1.263/0.921 | 1.853 |
Note: The first number evaluates performance by treating each cluster as separate, regardless of phonological form, and the second number treats all clusters with identical phonological forms as constituting a single lexical item. The mean number of lexical items recovered is not shown, as the target number of lexical items differed across the ten corpora.