A lot of the time, Zamenhof seemed to think of Esperanto only as a proposed written language, with words designed to appear on paper instead of being spoken, and of the foreign words he was borrowing as strings of letters rather than sounds. Silent U, for instance, was often misidentified as a /w/ and Esperantised into a V – as in gvati = “to keep watch”, taken from French guetter (pronounced roughly “getEH”). Meanwhile the presence or absence of a circumflex accent was occasionally treated as no big deal – so “a German” is germano, not ĝermano; “a piston” is piŝto, not pisto; and “because” is ĉar even though the French word it's based on is car (with a “hard” C). In these cases the idea may be that another borrowing has already claimed the phonetically closer form, leaving only the slightly mangled version as a second best… but in the case of ĉar, the root that got there first was kara meaning “dear”, and in French that's cher, so you'd think it would make more sense to do it the other way around.
It's easier to sympathise with cases where Zamenhof was dealing with English words containing sounds that have no neat Esperanto equivalent. Under those circumstances it's hard to improve on the option of giving up and borrowing bird as birdo (pronounced “BEER‐doh”) and sun as suno (“SOO‐no”). But in other cases it's obvious that whoever borrowed the word had simply never heard it spoken: plejdo = “a plaid”, pronounced “PLAY‐doh”?
Or consider Esperanto's KZ as in ekzameno = “an exam”, with its tricky sequence of voiceless stop and voiced fricative that needs to be painstakingly distinguished from the KS in words like eksa = “former”. That could have been egzameno (cf. Polish egzamin), but GZ never occurs in Esperanto because that spelling never occurs in Russian! Instead when foreign words with an X are borrowed into Russian as words pronounced with /gz/, they indicate their origin by using a spelling with KZ (Cyrillic КЗ). Russians don't have a /k/‐sound in ekzamen any more than we have one in exam, but Zamenhof preferred to copy an unpronounceable irregular spelling even when that spelling was in a different alphabet!
And there are plenty of individual borrowings that demonstrate a policy of preserving the source‐language spellings at any cost:
- aĉeti (roughly “ah‐CHAY‐tee”) = “to buy”
- from (exclusively) French acheter (/aʃte/, roughly “ash‐TEH”)
- boato (roughly “bo‐AH‐toe”) = “a boat”
- from English boat; cf. German Boot (both pronounced just “BOHT”)
- ĉerko (roughly “CHAIR‐koe”) = “a coffin”
- from French cercueil (/sɛʁkœj/), ultimately sarcophagus
- fraŭlino (roughly “frow‐LEE‐no”) = “Miss”
- from (dated) German Fräulein (/'fʁɔʏlaɪn/, roughly “FROY‐line”)
- honto (roughly “HONE‐toe”) = “shame”
- from French honte (/ɔ̃t/, roughly “AWNGT”)
- pilko (roughly “PEEL‐co”) = “a ball”
- from Polish piłka (/'pʲiwka/, roughly “PEW‐ka”)
- pugno (roughly “POOG‐no”) = “a fist”
- from Italian pugno (/'puɲɲo/, roughly “POON‐nyo”); cf. Spanish puño
- saŭco (roughly “SOW‐tso”) = “sauce”
- from French sauce (/sos/, roughly “SOCE”), or the English
- soifo (roughly “so‐EE‐fo”) = “thirst”
- from French soif (/suaf/, roughly “SWAHF”)
- teamo (roughly “tay‐AH‐mo”) = “a team”
- from English team (/tiːm/ – yes, there's an “i” in team!)
Apart from anything else, where would Esperanto be if any of these languages significantly reformed their orthographies? After all, the original German and Russian texts in Zamenhof's writings are already hard to read today due to changes in the way the languages are written.