In parshat Noach, after encountering the root סגר, we encounter the root סכר. In Bereishit 8:
| ב וַיִּסָּכְרוּ מַעְיְנֹת תְּהוֹם, וַאֲרֻבֹּת הַשָּׁמָיִם; וַיִּכָּלֵא הַגֶּשֶׁם, מִן-הַשָּׁמָיִם. | 2 the fountains also of the deep and the windows of heaven were stopped, and the rain from heaven was restrained. |
And Ibn Ezra says that it is similar to סגר, though of course it is not the same root. Because in Hebrew, certain letters switch, but others do not:
ח, ב]
ויסכרו -
כמו ויסגרו, ואין הכ"ף מתחלפת בגימ"ל, רק הם שתי מלות.
והטעם אחד, כי לא יתחלפו חוץ מאותיות יהו"א וסמ"ך בשי"ן.
Thus, ehevi switch, and samech and sin switch, but no others. He neglected to mention one exception to this general rule, which is that in the hitpael and the like, the tav instead manifests as a tet, due to a phonological rule where it becomes the more emphatic t sound in the context of other emphatic sounds such as the tzadi. Though it cannot simply randomly switch in most contexts, in terms of meaning.
So it is kemo vayisagru, where the similar letters in these roots share similar meanings (see Rav Shamshon Refael Hirsch for a well-developed system of this), but of course they are not the same word.
In other instance, Ibn Ezra speaks of a gimel / kaf alternation with similar meaning, which prompts the sefer Michlol (is this Radak) to declare that Ibn Ezra didn't write a certain peirush, but that an erring student did. And this gives Avi Ezer, supercommentator of Ibn Ezra, license to claim this whenever he feels that Ibn Ezra is out of bounds in arguing against Chazal. (See here.)
In Mishlei 9:3:
| ג שָׁלְחָה נַעֲרֹתֶיהָ תִקְרָא-- עַל-גַּפֵּי, מְרֹמֵי קָרֶת. | 3 She hath sent forth her maidens, she calleth, upon the highest places of the city: |
And Ibn Ezra writes:
ט, גשלחה נערותיה תקרא על גפי מרומי קרת -כמו גבי, כי אותיות בומ"ף מתחלפות.
Note he does not only say kemo, like before, but that the letters bumaf switch off with each other, something he explicitly distances himself from in parshas Noach.
Avi Ezer writes on parshat Bereishit what is pictured to the right, saying to see Michlol, nikud gimel, where he writes that what the Rav wrote on Mishlei "al gapei" is like al gabei, a talmid toeh wrote it, see there.
I don't know which Michlol he refers to. I was not able to find it in Michlol of Radak, but maybe I just did not know how to look.
At any rate, on parshas Noach, when Ibn Ezra says what he says about ויסכרו, Avi Ezer takes notice. And he says:"However, in sefer Mishlei the Rav {=Ibn Ezra} wrote upon al gapei: kemo al gabei. And in parshat Miketz, on נצטדק, that the tet is in place of the tav of hitpael. Thus it is clear that the Rav retracted from this, and maintains like the opinion of those medakdekim that letters which come from the same place switch with one another as well. And it is forced to say in all of these that they are two words. And see the beginning of parshat Kedoshim, on the pasuk lo telech rachil. Or else, choose for yourself the words of the Michlol which I brought in Bereishit 1."
What he means by choosing the words of the Michlol is to say that a talmid toeh wrote the words on gapei.
In terms of the tet / tav in hitpael, see Mechokekei Yehuda who cited what I think is a commentary of Ibn Ezra (shita rishona) that includes the tet / tav switchoff as well as tav / heh.
Radak has a similar take on the gimel / kaf switchoff or meaning, in this instance. 
He writes on Vayiskeru, in parshas Noach what appears to the right -- that with a kaf it is like with a gimel, and gives a parallel in Tehillim 63, where it means to shut up.
He writes on Vayiskeru, in parshas Noach what appears to the right -- that with a kaf it is like with a gimel, and gives a parallel in Tehillim 63, where it means to shut up.
And on gapei, in sefer Shorashim, Radak considers the possibility that it is kemo al gabei, since the bet and the peh are from the same place of articulation.
It certainly does seem that Ibn Ezra reversed himself between the two comments, especially when we consider the full quote in Mishlei. While one might be able to argue that he is saying that roots with letters from the same place of articulation bear related meanings, while not being the same word, I am not convinced that his language can sustain such a meaning.
