Detection of rhymes
Training set: "vertical collocations"
High precision, low recall (completely unsupervised):
|
CZECH |
ENGLISH |
FRENCH |
PRECISION: |
94.6 % |
96.0 % |
99.5 % |
RECALL: |
14.9 % |
14.7 % |
2.7 % |
Rhyme-type / rhyme-token ratio
Does the size of rhyme repertory corellate with the extent of inflection ?
10 random samples w/o replacement; n = 1000; 10 times repeated:
|
CZECH |
ENGLISH |
FRENCH |
MEAN: |
0.9784 - 0.9826 |
0.9199 - 0.9295 |
0.9562 - 0.9644 |
ST.DEV.: |
0.0028 - 0.0062 |
0.0043 - 0.0099 |
0.0037 - 0.0080 |
10,000 random samples w/ replacement; n = 10,000:
|
CZECH |
ENGLISH |
FRENCH |
MEAN: |
0.8885 |
0.6881 |
0.8052 |
ST.DEV.: |
0.0040 |
0.0029 |
0.0033 |