The -tion (or -sion) suffix is commonly used to turn a verb into a noun. Some common examples include:
communicate => communication
relax => relaxation
educate => education
It's true that these nouns are not physical, tangible things, but the definition of noun as "person, place, or thing" is woefully incomplete. These nouns are abstract concepts, which are indeed nouns.
The answer is etymology.
Many English words are built from a stem and some prefixes and suffixes. (I think this is true of most languages.) There are many suffixes that preserve the core idea of the word, and give it a particular grammatical nature and a particular meaning. For example, from the stem defin-, you can build the verb define (to give something a definition), the noun definition (the act of giving something a definition, or the outcome of this act), the (uncommon) noun definer (someone or something that defines), etc.
Many of these stems, prefixes and suffixes existed in ancestor languages of English. Many of the resulting words existed in ancestor languages, and sometimes different words in the same family evolved differently.
The examples you give are quite illustrative. The verb define comes from the Latin verb definire, and the noun definition comes from the Latin noun definitio. The verb combine comes from the Latin verb combinare, and the noun combination comes from the Latin noun combinatio. Note how one pair has the infinitive verb ending in -are and the noun ending in -atio, and the other pair has the infinitive ending in -ire and the noun ending in -itio. What happened here is that the ending of the verb was eroded: Latin verbs had over a hundred different endings depending on the mood, tense and person, whereas English has 5. Many verb endings got simplified to -e (often in French before they reached English), which erased the previous vowel. So what happened isn't that defin- got the suffix -ition and combin- got the suffix -ation, but rather that defini- and combina- both lost their ending vowel.
Although many stems of this type end with the vowel -a or -i, Latin also had the vowel e, as in. delere, which gave the English words delete and deletion through a less direct route). Some stems had no final vowel, e.g. cedere (many forms of the verb, including the infinitive, have an additional -e- to ease pronunciation), giving cede in English, and the associated noun cessio (whence cession) where the suffix and the final consonant had merged together and diverged already in an ancestor language of Latin.
The suffix -tion to form a noun from a verb, meaning the action itself or the outcome of this action, was very common in Latin (shown in the form -tio above, but the -n was already present in some cases) and in French (also spelled -tion) but it is no longer productive in English.
English has many other suffixes, some of them productive, most of them not. Most of these suffixes come from Latin, often through French, even though English is primarily a Germanic language. The reason for this is historical: German was the language of ordinary people (who mostly use existing words to designates objects from everyday life), French and Latin were the languages of learned people (who invent new concepts and make up words for them).
Best Answer
The word "why" in your example is most likely an interrogative word.
RATIONALE: The matrix verb "know" can take an interrogative clause as complement. And the complement "Why" in the OP's example seems to be an interrogative word of a truncated interrogative clause. For example, let's use the following sentence(s) to show this:
That above sentence has the form of a declarative main clause; and the reason why Tom doesn't like that woman is "because that woman had kicked his dog". We can convert that above sentence into a corresponding interrogative main clause with the interrogative word "why" in situ:
Note that the interrogative word "why" corresponds to the reason "because that woman had kicked his dog".
We can front that interrogative word "why", which causes subject-auxiliary inversion when the interrogative clause is a main clause, and produce:
Obvious stuff, so far.
Now, let's have a person A talk to a person B to find out why Tom doesn't like that woman:
A1: "Why doesn't Tom like that woman?"
B1: "I don't know [why Tom doesn't like that woman]."
We can paraphrase the B1 response as: I don't know the answer to the question 'Why doesn't Tom like that woman?'
(Aside: Notice that the B1's subordinate interrogative clause "why Tom doesn't like that woman" doesn't have subject-auxiliary inversion that was seen in the corresponding main clause.)
Person B could have responded slightly differently, but with the same meaning:
A2: "Why doesn't Tom like that woman?"
B2: "I don't know [why he doesn't]."
or even as,
A3: "Why doesn't Tom like that woman?"
B3: "I don't know [why]."
We can paraphrase the B2 and B3 responses as: I don't know the answer to the question 'Why doesn't he?' and I don't know the answer to the question 'Why?', respectively. We know that the embedded questions in B2 and B3 are shortened forms of the question "Why doesn't Tom like that woman?"
Thus, the "why" in the OP's original example could be considered to be an interrogative word. That is, the word "why" represents a shortened form of an embedded question somewhat similar to "Why don't I like that woman?"
but there's something I don't like about that woman.where #2 could be paraphrase as: I don't know the answer to the question 'Why don't I like that woman?'
Notice that if the shortened form is replaced with the longer interrogative clause (as is done in #2), then there is a duplication of the string "I don't like (about) that woman" in the sentence in such a way that the result is awkward and unwieldy, and the result might even be ungrammatical (which is why the last half of #2 was struck-through). This could be a reason why, in #1 (the OP's original example), the interrogative clause was truncated to merely the word "why".
RATIONALE: Why the word "why" in the OP's example is probably not a noun.
The dictionary on my iMac, New Oxford American Dictionary, has this as the "noun" part of the definition of the word "why":
That sounds somewhat reasonable (that is, for our purposes it might be good enough). In that dictionary's example, you can see that:
The noun "why" inflects for number: singular "why" vs plural "whys". And you can see that their example uses the determiner "the" in their noun phrase (NP), e.g. "the whys of these procedures". In general, a word like the article "the" will usually only be found in an NP (though there are exceptions).
Also, notice that in the expression "the whys of these procedures" that the head noun "whys" has a dependent of-phrase "of these procedures". An of-phrase is often found as a post-head dependent in NPs.
Now let's look at the OP's original example:
I would consider the word "why" to be a noun if it had been in something like:
where the NP "the why of it" would have a determiner "the" and a post-head dependent of-phrase. But the OP's original example doesn't have its word "why" in a phrase that has a structure that is commonly used by noun phrases. And the interrogative word rationale appears to be much more convincing, syntactically and semantically.
RATIONALE: Why the word "why" in the OP's example is probably not a relative word.
The dictionary on my iMac, New Oxford American Dictionary, has this as the "relative adverb" part of the definition of the word "why":
The first part of its definition appears okay, as it corresponds to what is in the 2002 CGEL. The dictionary's example has an integrated relative clause and its antecedent:
The relative clause is:
Notice how the gap inside the relative is co-indexed with the relative word "why" and with the antecedent "reason" (which is a noun). This is all unremarkable, as this is typical for an integrated relative clause.
Notice how in this example that the noun "reason" is a variable that has the value "that the virus changes", and so, the relative clause can be seen as corresponding to something like the following main clause:
To create that above main clause, the gap in the relative clause had been replace with the previously missing reason.
Here's some related info in CGEL as to this relative word "why". On page 1051:
Also, there's this in CGEL on the bottom of page 1059:
As to fused relatives, there's also this footnote in CGEL on page 1077:
From those above CGEL excerpts, that basically means that if the word "why" is a relative word, then it will almost always be in an integrated relative clause that uses the noun "reason" as its antecedent. The few times that a relative "why" isn't part of a normal integrated relative, then, that means that it is a fused relative (which means that there is no separate antecedent for that relative) and will be part of a pseudo-cleft.
So, now let's look at the second part of that dictionary's "relative adverb" definition:
That definition's example seems to either be using a fused relative in a pseudo-cleft construction (when compared against the info in CGEL) or an embedded interrogative clause. The important part of its example is:
A corresponding clause could be created by replacing the word "that" with its antecedent,
That clause has an apparent possible resemblance to a pseudo-cleft and could possibly be interpreted as one. Though, it is more likely that it would be interpreted instead as using an interrogative "why", where the paraphrase could then be,
But in the OP's original example, in the text ("I don't know why") which involves the word "why", it does not use the noun "reason", nor does its construction have the form or shape of a pseudo-cleft. And so, the interrogative word rationale appears to be much more convincing.
NOTE: CGEL is the 2002 reference grammar by Huddleston and Pullum et al., The Cambridge Grammar of the English Language.