Let us we divide all the letters in 3 large hypergroups
1. פמב                    labial
2. לזסנטשדרצת        all, that need fore-tongue
3. כחעקג                all, that need back-tongue and throat
 
Now let us examine statistic of the hebrew roots in our database regarding all the combinations of 1.2.3. without repeating (such as 1.2.1.)
The difference in quantities of the letters in each group (3,10,5) should not influent on the statistics because we take only combinations where all groups are present.
 
It is convenient to take one letter from each group as  a notation: 1=P, 2=T, 3=K. It also would be useful
considering our speed/conveniency-test for each triad. In fact each triad combination of letters P,T,K represends caracteristic movements of all according roots. For example pronouncing KPT you make the similar movement of pronouncing גִּבֵּשׁ , גָּמַר  חָמֵשׁ  and so on.
 
Here is the statistic:
............................................................
312 KPT
[כחעקג][פמב][לזסנטשדרצת]
57
 
123 PTK
[פמב][לזסנטשדרצת][כחעקג]
54
 
132 PKT
[פמב][כחעקג][לזסנטשדרצת]
45
 
213 TPK
[לזסנטשדרצת][פמב][כחעקג]
38
 
321 KTP
[כחעקג][לזסנטשדרצת][פמב]
23
 
231 TKP
[לזסנטשדרצת][כחעקג][פמב]
12
 
Now we can compare it with pair statistics:
...........................................................................
 
32 KT
[כחעקג][לזסנטשדרצת]
587
 
23 TK
[לזסנטשדרצת][כחעקג]
447
 
12 PT
[פמב][לזסנטשדרצת]
420
 
21 TP
[לזסנטשדרצת][פמב]
280
 
13 PK
[פמב][כחעקג]
152
 
31 KP
[כחעקג][פמב]
141
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

This "root"-statistic stickingly differs from that of "hole-text"-statistic.
Here is the triplet statistik for humash:

SORTED 3

PKT=3055

TPK=2944

PTK=2675

KTP=2496

TKP=2252

KPT=1916



SORTED 2 TWO 2

PT=17692

TP=16401

KT=15914

TK=13164

PK=6537

KP=5553

*This results were also tested with "random words" method, so the influence of words conjunction on the statistics was expluded.
Other TaNaKh texts give very close statistics.

Here we can see, that most "inconvenient pair is KP. If you try to speek very quick KP and than PK you will notice, that PK is much easier to pronounce, due to the "opening-mouth" movement.
But difference in PK>KP statistic can not explain the difference of triplets with PK, KP pairs : PKT=3055/KPT=1916 ~= 3/2 but K=6537/KP=5553 ~= 6/5.

As we see, words-statistics of phonetic contour reflect conveniency of pronounciation.