The intelligibility level (percent) depends not only on the lexical distance percent but also (and to large extent) on the orthographic distance (for written intelligibility) which is the percent of the changed letters between the pairs of cognates or partial cognates with completely the same meaning in the total amount of the letters of the longest of two (partial) cognate words. Often there may be bigger amount of letters and signs than in the longest words if therr are so-called "insertions". The same method is used for the spoken (phonetic dustance) but insted the national alphabet letters there are sounds of the International phonetic alphabet (and there may be very close sounds and their change would be not like 1 ball but as 0,5 ball or less.
It is called in the some linguistic branches as "Normalized Levenshtein distance). So, the pairs from Swadesh list like Russian "ног" (legs in genitive case) and and Ukrainian " ніг" differs in spelling in 33% (big distance) and in pronuncuarion - basically in 66 % - very big phonetic distance which can lead in not complete understanding (if the context is confusive, incomprehensive or allows parralel translation).
Also, big role there may plays sintax and grammar distance.
The second issue is the definition of the word "common vocabulary". Actually there is not an unequivocal view between linguists (for example it was noticed by the Soviet linguists Arapov and Herz in 1970-s ) And there are
two these extremities in linguists.
- The non-cognates are only these words which have different roots till the root of the Proto-Indo- European or the other proto-languages. But, for example, all European languages would probably have in this case unusually big percent of cognates. Still there is need to look for complete false cognates like Russian "швирять" (to throw) and Ukrainian "жбурляти" (old form шпурляти which are occasionally similar to the Russian word). The representative of this approach was for example the Russian linguist Starostin, but he acknowledged non-cognacy in a words' meaning in the Swadesh list. Still, I have seen more "radical" presentatives of this view between the authors of the some linguistic publications whichvare accesible in the Internet . And Swadesh list calculation is at odds with this approach because the Ukrainian чоловік which is husband and male person cannot match the Russian человек in the demanded 100 % exact meaning of the English Swadesh list word "person".
- Unequivocally cognates (common vocabulary) - representatives (partial or complete) these linguists Istvan Fodor, Icidore Dyen, Kirschen Peust, Robert Lawrence Trask, Johanne Mattice-List and others.
According to the approach, cognates (relatives) is those which not only have the same root, but:
A) They are inherited from a Proto-language but not borrowed from any language including this proto - language or "third" Slavic or non-Slavic language. So, the pairs like "всегда - завжди", "враг - ворог", "сахар - цукор" are non-cognates.
B) They directly originated from one word-ancestor and language ancestor (so, the words like женщина - жінка, как -як are non-cognates.
C) 100 % same meaning: so, the cognates with the different meaning (богато -багато) or partial cognates with close meaning (люди - людина) are non-cognates)
D)Absence of so-called irregular sound changes in word roots (so он - він, медведь- ведмідь, толкать - штовхати are non-cognates.Irregular sound changes are relatively not frequent and they are the result of the word -formation but not the natural evolutionary sound change like o to i in вол - віл (an ox). For example, Istvan Fodor (Hungary) wrote about loanwords, indirect origin, and the situations when roots "drifted so far apart" in this publication http://www.lllf.uam.es/~clase/acceso_local/IF61Gloto.pdf especially on the pages 320 - 323. Still, there were some not perfect approaches used by I.Fodor in the classification and calculation of common vocabulary noticed by M.V.Arapov, M.M.Herz and I.Dyen.
E) Absence of the irregular sound changes in affixes. So,привичка - звичка (habit) are non-cognates.
Some principles of differentiation real cognates from these not complete cognates which are products of the word-formation are shown here:
But if the latter authors say that for the most of linguist the common root generally is enough, the Richard Lawrence Trask had the opposite view:
https://books.google.com.ua/books?id=EHeGzQ8wuLQC&pg=PA146&dq=handbook+of+comparative+linguistics+Robert+Lawrence+Trask&hl=ru&sa=X&ved=2ahUKEwiZ5uiS8Lj8AhVskosKHSqdCMEQ6AF6BAgEEAI#v=onepage&q=cognates&f=false / It's for example on the page 62 (item -"cognates") ("partial cognates"), page 234("oblique cognates"),page 248("partial cognates").
Mostly, the widespread views and principles in West linguistic concerning cognates-non-cognates are generalized in a book " Математические методы в исторической лингвистике" (the authors are М.В. Арапов, М.М. Херц), which was published in Moscow in 1974 and especially on pages 26 and 57-59. Here are just fragments of this book:
There is many linguists who are in the middle of two extremities like this Ukrainian linguist K.Tyshchenko (the examples of non-cognates ( called by him as "innovations" in the Swadesh list fragment are written and coloured by him here:
on the page 54, in the figure 20. Some principles of defining of words like innovations are shown on the pages 53 and 55 (there is, for example about the metatesis ( a kind of the irregular sound changes in a word's root of Czech word "hmla" (gaze) in the comparisons with the Russian "мгла" (the latter sounds like "mgla).
But the words which have just irregular suffixes and prefixes are not coloured by him as "non-cognates" ( like Belarusian "жывёла" and Russian "животное" (animal), or like Ukrainian "кістка" and Russian "кость" ("bone").