It is well known that Italian -/.tj/- developed into -/t.tsj/- after stressed syllables (gratiam -> grazia). There are, however, several rare words that end in post-stressed -/tj/- like "òstia", "àstio", "modèstia", "lìtio" and so on. However, I have noticed that all of these feature an s in the coda of the preceding syllable, leaving me to think they were spared from the sound shift that turned -ti- to -zi- because of the phonotactical impossibility (in early Romance, I suppose) of the -/s.ts/- cluster, at least in that position. What I find puzzling is how, despite all this, there is a single exception in the word sàrtia (which by the way refers to a steel cable used to keep a mast stationary), which despite not having the s there sounds perfectly "right" to my (North) Italian ears.

What gives? Shouldn't it sound weird or wrong?

  • Interesting question. Wiktionary says it’s a loan from Byzantine Greek (ἐξάρτια), which could be an explanation. Romance palatalisations are a minefield, but I think Spanish jarcia and Catalan xarxa /'ʃaɾʃə/ (instead of *jarza and *xarça) also indicate that this is not an inherited word that developed all the way from Latin, but one that was borrowed later on and only took part in some of the subsequent sound changes, evidently postdating the /tj/ > /tsj/ one. (I believe grazia is also an early reborrowing from Latin – otherwise, it ought to have become *grazza.) Jun 4, 2023 at 21:54
  • (Of course litio, which you cite as well, is another example of a word with an unexpected /tj/ cluster, but in this case it’s known for certain that the word is a recent borrowing: the name itself was only coined as a Latinate neologism by Swedish chemist Jöns Jakob Berzelius around 1817. Sartia appears to be a good deal older than that, at least.) Jun 4, 2023 at 22:23

1 Answer 1


The regular outcome of -sti- + vowel in Italian words inherited from Latin is [ʃ(ː)], as in uscio from Latin ōstium.

The use of -sti- [stj] + vowel is part of the Italian academic pronunciation of Latin (I described the rule in more detail here: How to pronounce the sequence "ti" when reading Latin). So we can infer that words like "ostia" and "modestia" didn't develop regularly as inherited words, but rather are learned forms. When not after S, Latin -ti- before a vowel is generally pronounced by Italians as [t͡s(ː)j] or [t͡s(ː)i] and respelled in Italian learned forms as -zi-, but there seems to be an exception for cases where the Latin spelling has th. Thus litio, simpatia, etc. In inherited words, -ti- did not become [t͡s(ː)j]: it could variously end up as [t͡s(ː)], [d͡ʒ], or [t͡ʃ(ː)].

None of the rules for the alteration of -ti- to something else are synchronic constraints in Italian phonology (or at least, they aren't inviolable constraints): /tj/ exists unchanged in forms like partiamo. In that sense, I don't think it's too surprising from a synchronic perspective that /ˈsartja/ is inoffensive to your ears. But I don't know the diachronic explanation of how sartia developed its current pronunciation.

  • Thanks for the [ʃ(ː)] thing, I had no idea. I had noticed a great deal of words like "simpatia" and "partiamo", but didn't include them because I felt that the stressed /i/ (not even a /j/!) in the first, and the stressed /a/ in the second somewhat relax phonotactical constraints: in the sense that Italian speakers might be more willing to perform unusual phonetics in a stressed syllable than in an unstressed one. Cheers!
    – Enrico
    Jun 18, 2023 at 15:21

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.