@oskda One has a choice: take the spelled version of the English word, take the spoken version of the English word or take some easier to pronounce version of the spoken version of the English word.
As for it being "unacceptable" to base all the vocabulary on English: a billion people would already have a large part of this vocabulary, while in current lojban there is no obvious rule (it is random) for where any particular word came from so even those who are Chinese will probably not make the connection to a lojban word taken from Chinese as, after mangling to fit within lojban's phonetic/phonemic constraints (such as the removal of tones) and with the ever-present possibility of any particular word being a "false friend" there is no way to guess that the word means the same as a similar word in your language or is just another coincidental word.
If all knew that all words were taken from English (perhaps after phonemic and therefore spelling transformation) then it would be easy for the billion to guess what a large amount of the words mean when first seeing them (exceptions: "their", "they're" and "there" in many common dialects).
As a fun example of such a spelling transformation (though not entirely phonemic or alphabetic) consider
https://en.wikipedia.org/wiki/SoundSpel :
1/ You don't understand that if you try to construct an INTERNATIONAL and NEUTRAL artificial language you can't get all (or near all) words from English. You must get them from a bigger number of very important languages. And even in some cases it is better to invent some words as no one on them is ideal, or to mix words from different languages to form other so in this way it is more common to them (not only to one).
2/ If it is true that today English is spoken, usually bad spoken by lots of people (more for foreigners , but not only for them but too for natural speakers as English is very wide and they spoke it in different ways), more true it is that it ISN'T an ideal language (in a grammar nor in semantic way).
3/ If you get all words (or near all) in Lojban from English, people would end ruining it trying to use it as English or a mix of English and Lojban, and that isn't the Lojban purpose.
4/ "false friend" words exist between English and other western languages (for example between Spanish and English there is a lot of false friends words).
4/ Even if you try to use English for near all words, you find it is far from ideal, as there are words too similar in English, or simply they don't fit in a logical simple but true powerful language.
Simply Lojban is a NEW language, nor a mutilated English.
Sometimes it is better to make a new building from ground.
For me, if I would learn Lojban, it would be easier if all English words would be used, as I know a lot of them (at least in written form), but that would serve bad to Lojban spirit.
SoundSpel may be a good idea to transform English (but for sure it won't be successful [1]), but not for a new simply logical artificial neutral and really powerful language.
Of course nor SoundSpel nor Lojban (or other artificial languages) will have real success: we will continue to use natural languages as
Lingua franca and sometime (it may be a lot of decades, even some centuries from today) English will be replaced for other natural language with the same problems today English offer, or even worse (imagine for example Chineses like Lingua franca, with its Han characters
).
[1]: SoundSpel idea isn't unknown for me: in Spanish there is like a 95% bidirectional correlation between spoken and written language, but for that small differences (for example B and V are pronounced exactly the same from some centuries ago, so people sometimes write incorrectly one in place of other) and there has been proposal to modify written language so that correlation be 100%.