I have a question about that pronunciation. "בוטן" is spelled with the vowel "segol" (the 3 dots) below the "tet". It is pronounced bo-ten. (Using English pronunciation of the "e" as in "bed".) When a plural ending is added to a word, I didn't think that the spelling or pronunciation of the beginning part of the word changed, unlike when a prefix is added to a word. It sounds to me like the pronunciation is bo-te-NIM, with the "te" being said very quickly. Can any of our Hebrew speaking site managers please weigh in on this question?
"When a plural ending is added to a word, I didn't think that the spelling or pronunciation of the beginning part of the word changed, unlike when a prefix is added to a word."
I can't answer your specific question about בוטנ/ים (I asked the original question!), but I do know that the pronunciation of the beginning part of a word can sometimes change a lot in Hebrew when a plural ending is added:
"book" ספר séfer > ספרים sfarim
"room" חדר khéder > חדרים khadarim
"shop" חנות khanut > חנויות khanuyot
"city" עיר ir > ערים arim
"pencil" עיפרון iparon > עפרונות efronot
Most of these changes are regular and make sense once you get used to the rules and learn enough Hebrew words that follow them, I've found.
There are historical reasons for this and other similar phenomena (see shwmae's excellent comment).
You can see the segol doing that often, it has developed in many places to replace a very short vowel which was apparently difficult to hold on to.
So yes, you should be aware that this exists. It can happen with any change - single vs. plural, conjugation, construct state etc.
As in Genesis 43:11:
וַיֹּ֨אמֶר אֲלֵהֶ֜ם יִשְׂרָאֵ֣ל אֲבִיהֶ֗ם אִם־כֵּ֣ן׀ אֵפוֹא֮ זֹ֣את עֲשׂוּ֒ קְח֞וּ מִזִּמְרַ֤ת הָאָ֙רֶץ֙ בִּכְלֵיכֶ֔ם וְהוֹרִ֥ידוּ לָאִ֖ישׁ מִנְחָ֑ה מְעַ֤ט צֳרִי֙ וּמְעַ֣ט דְּבַ֔שׁ נְכֹ֣את וָלֹ֔ט בָּטְנִ֖ים וּשְׁקֵדִֽים׃
And their father Israel said to them, “If it must be so, then do this: Take some of the best fruits of the land in your vessels and carry down a present for the man—a little balm and a little honey, spices and myrrh, pistachio nuts and almonds.
Well, it would be quite difficult to train the software in all eventualities of non-standard dialects, someone has to enter this manually! And where are the limits, in a Ow, eez ye-ooa san, is e? Wal, fewd dan y' de-ooty bawmz a mather should, eed now bettern to spawl a pore gel's flahrzn than ran awy atbaht pyin. Will ye-oo py me f'them?-style Cockney?
Well, I think the word bank collects the word forms it finds in the sentences of the whole course and mixes some of them into the actual choice in order to make it more difficult to choose the right ones. If you type freely, it is more accepting of typos which are unknown words for it, but you change it to a word form it knows, but which is not in its base for the actual sentence, it rather thinks that gave a wrong answer. Programmable machine logic is different from fuzzy human logic.