Table 3
Tagging accuracy on the respective development sets (WSJ Section 24 for English and Penn Chinese Treebank articles 301–325 for Chinese) for binary classes 
, 
, and 
, for various Markov orders.
| Tagging Task | Markov order | ||
|---|---|---|---|
| 0 | 1 | 2 | |
| English | |||
 (no multi-word constituent begin) | 96.7 | 96.9 | 96.9 |
 (no multi-word constituent end) | 97.3 | 97.3 | 97.3 |
 (no span-1 unary constituent) | 98.3 | 98.3 | 98.3 |
| Chinese | |||
 (no multi-word constituent begin) | 94.8 | 95.4 | 95.2 |
 (no multi-word constituent end) | 96.2 | 96.4 | 96.6 |
 (no span-1 unary constituent) | 95.9 | 96.2 | 96.3 |
| Tagging Task | Markov order | ||
|---|---|---|---|
| 0 | 1 | 2 | |
| English | |||
| (no multi-word constituent begin) | 96.7 | 96.9 | 96.9 |
| (no multi-word constituent end) | 97.3 | 97.3 | 97.3 |
| (no span-1 unary constituent) | 98.3 | 98.3 | 98.3 |
| Chinese | |||
| (no multi-word constituent begin) | 94.8 | 95.4 | 95.2 |
| (no multi-word constituent end) | 96.2 | 96.4 | 96.6 |
| (no span-1 unary constituent) | 95.9 | 96.2 | 96.3 |