I found the following etymology of the word "ambassador" on Wiktionary.

From Middle English ambassadore, from Anglo-Norman ambassadeur, ambassateur, from Old Italian ambassatore, ambassadore, from Old Occitan ambaisador (“ambassador”), derivative of ambaissa (“service, mission, errand”), from Medieval Latin ambasiator, from Gothic 𐌰𐌽𐌳𐌱𐌰𐌷𐍄𐌹 (andbahti, “service, function”), from Proto-Germanic *ambahtiją (“service, office”), derivative of Proto-Germanic *ambahtaz (“servant”), from Gaulish ambaxtos ("servant"; also the source of Latin ambactus (“vassal, servant, dependent”)), from Proto-Celtic *ambaxtos (“servant”), from Proto-Indo-European *h₂m̥bʰi-h₂eǵ- (“drive around”), from *h₂m̥bʰi- (“around”) + *h₂eǵ- (“to drive”).

I'm wondering how Gothic 𐌰𐌽𐌳𐌱𐌰𐌷𐍄𐌹 (andbahti) became Medieval Latin ambasiator phonetically, especially how h in andbahti became s in ambasiator. I have read some books but still couldn't find a specific reason for this sound change.

Here is my personal explanation of this sound change:

This is a sound change of partial assimilation. h changed into s retaining its own fricative feature and gaining the alveolar feature from t. Then epenthetic vowels (between s and t) were added.

Still, I am not sure if my the explanation is correct. If yes, are there any other sound changes similar to this one?

Yet, I am also not sure whether h actually changed into s historically.

Edit: Thanks for the answers below, really appreciate them!

2 Answers 2


I think a likly path to the "s" is through "kt" (as in ambactus) which then palatalized before j. A variety of spellings are apparently found in this word and related words such as ambascia: single s, double ss, x, sc, c.

It's a bit hard for me to find similar examples of the outcome of Latin -cti- in Romance languages, but perhaps the -ss- in French cuisson, which is supposed to be from Latin coctio, is analogous.

The t in -ator is originally part of the Latin agent noun ending, not part of the stem found in 𐌰𐌽𐌳𐌱𐌰𐌷𐍄𐌹/ambaxtos.

  • An even more likely path, I think, is -(h)tj-, with no /k/. Since /kt/ > /xt/ (optionally) already in Gaulish, and /xt/ > /ht/ in various Germanic languages (where the -ja- derivative was created), the form borrowed form was likely /ambahtja/, with typical Romance loss of /h/, so /ambatja/ or perhaps /ambatːja/. Dec 29, 2021 at 8:23
  • @JanusBahsJacquet: The thing that makes me question a derivation from intervocalic /tj/ is that I think the usual reflex of that in French is voiced /z/, as in raison < rationem. Jan 9, 2022 at 8:40
  • That’s for inherited /tʲ/. The development /tʲ > tsʲ > dzʲ/ took place in Common Western Romance; the loss of the palatal quality (with compensatory insertion of /j/ in the previous syllable before voiced palatalised consonants) happened later, on the way to OF. Assuming that *ambahtja was borrowed from Germanic some time between WR and OF (say some time around AD 600–1000), the input at the relevant time would have an unvoiced /t(ː)sʲ/ and the expected OF outcome would be /ts > s/. The forms with /ʃ/ are seemingly through Occitan, which I don’t know enough about to comment on. Jan 9, 2022 at 10:23

The shift of /k/ to /h/ is regular in Germanic (assuming that the borrowing from Celtic to Germanic is very old: i.e. pre-Grimm). But I do not see why the Romance forms should derive from Gothic, rather than directly from Celtic. The development of -kt- to Italian -sc- and then to French -ss- does not seem problematic.

  • Where does /kt/ give /ʃ/ in Italian? I would expect it to yield /tː/ in general āctum > atto), or /ts/ before /j/ (āctiōnem > azione), but not /ʃ/. The Occitan-ish area would be a more likely place for such an outcome, but I don’t think you can skip the Germanic stage quite so easily – as far as I know, the -ja- derivative isn’t known in Celtic, but was created in Germanic, and it must be that which sired the Romance forms. The /kt ~ pt/ > /xt/ development also took place in Celtic (it’s attested in both forms in Gaulish), though, so the loan doesn’t have to be early. Dec 29, 2021 at 0:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.