The forms with a preffixes o-,qo-,y- and suffixes -y, -dy was collected.
For each form i the number of roots having it R(i) was calculated
For each form pair (i,j) the number of roots having both N(i,j) was calculated.
The correlation Corr(i,j)=N(i,j)/Sqrt(R(i)*R(j)) *100%
Table 1. "Correlation" between prefix-suffix-forms
Corr(%) | *** | ***dy | ***y | o*** | o***dy | o***y | qo*** | qo***dy | qo***y | y*** | y***dy | y***y |
*** | 100 | 11 | 11 | 21 | 8 | 8 | 19 | 7 | 6 | 7 | 7 | 11 |
***dy | 11 | 100 | 25 | 9 | 28 | 17 | 9 | 26 | 20 | 10 | 18 | 14 |
***y | 11 | 25 | 100 | 8 | 16 | 17 | 8 | 16 | 18 | 8 | 12 | 14 |
o*** | 21 | 9 | 8 | 100 | 13 | 13 | 34 | 11 | 9 | 27 | 12 | 8 |
o***dy | 8 | 28 | 16 | 13 | 100 | 32 | 14 | 45 | 35 | 11 | 32 | 23 |
o***y | 8 | 17 | 17 | 13 | 32 | 100 | 15 | 31 | 31 | 13 | 21 | 20 |
qo*** | 19 | 9 | 8 | 34 | 14 | 15 | 100 | 14 | 14 | 34 | 14 | 10 |
qo***dy | 7 | 26 | 16 | 11 | 45 | 31 | 14 | 100 | 39 | 12 | 39 | 27 |
qo***y | 6 | 20 | 18 | 9 | 35 | 31 | 14 | 39 | 100 | 11 | 28 | 24 |
y*** | 7 | 10 | 8 | 27 | 11 | 13 | 34 | 12 | 11 | 100 | 16 | 11 |
y***dy | 7 | 18 | 12 | 12 | 32 | 21 | 14 | 39 | 28 | 16 | 100 | 28 |
y***y | 11 | 14 | 14 | 8 | 23 | 20 | 10 | 27 | 24 | 11 | 28 | 100 |
The next table calculates the probability of existence of the second form if the first form exists.
It was calculated as P(j|i)=F(i,j)/H(i)
where H(i) - the number of words having the form i,
F(i,j) - the number of word, having the form i with existing form j.
For example, if we have words fory,kaly, kaly, kaldy then we have F(-y,-dy)=2,
H(-y)=3, H(-dy)=1, F(-dy,-y)=1
Table 2. Conditional probabilities of prefix-suffix-forms
P(%) | *** | ***dy | ***y | o*** | o***dy | o***y | qo*** | qo***dy | qo***y | y*** | y***dy | y***y |
*** | 100 | 39 | 52 | 60 | 10 | 17 | 48 | 4 | 7 | 53 | 3 | 8 |
***dy | 61 | 100 | 70 | 27 | 58 | 59 | 21 | 45 | 45 | 31 | 47 | 52 |
***y | 52 | 61 | 100 | 14 | 41 | 44 | 10 | 31 | 33 | 15 | 28 | 38 |
o*** | 81 | 26 | 41 | 100 | 38 | 48 | 74 | 25 | 30 | 60 | 11 | 34 |
o***dy | 39 | 89 | 81 | 45 | 100 | 86 | 21 | 82 | 79 | 23 | 68 | 60 |
o***y | 41 | 64 | 77 | 43 | 68 | 100 | 32 | 63 | 75 | 31 | 48 | 63 |
qo*** | 89 | 33 | 41 | 91 | 38 | 46 | 100 | 32 | 37 | 81 | 19 | 34 |
qo***dy | 25 | 95 | 92 | 28 | 96 | 94 | 17 | 100 | 91 | 18 | 88 | 82 |
qo***y | 37 | 74 | 91 | 39 | 76 | 92 | 33 | 81 | 100 | 35 | 71 | 83 |
y*** | 78 | 26 | 36 | 73 | 23 | 32 | 66 | 15 | 20 | 100 | 15 | 21 |
y***dy | 51 | 86 | 77 | 48 | 87 | 81 | 31 | 82 | 73 | 37 | 100 | 67 |
y***y | 44 | 66 | 77 | 31 | 69 | 76 | 25 | 60 | 69 | 31 | 62 | 100 |
Conclusions.
1.Words having prefixes o-, qo- and y- , very often have forms without
preffix or with any other prefix.
2.Words having suffixes dy-, and y- , very often have forms without
suffix or with other suffix.
3. Large amount of words do not accept some suffixes and preffixes, such words
should be detected.
Actually, some part of roots are "active" and can have many forms