According to etymonline, definition (4) of "per", this root component formed words such as "repress," "express," etc. Except none of these words have "per" in them, only "pre." Incidentally, this is also the case in every other root definition on the same page, words forming in "pre," or even "pri" in "imprint", seem to have had the letter order, and in some cases the vowel flipped or swapped. Only in certain words, where the root meaning is different (definition (3)) do we find words like "perilous" or "expert" that do look the same (we also have another vowel swap with or "par" in "parlour").

Exactly what happened that it looks this way in virtually every single word, but the root element is different than the final result ("per" and not "pre")? What modified all these words and why?

3 Answers 3


Proto-Indo-European ablaut, mentioned in another answer, explains e/o/ē/ō/∅ alternation, but not "pri" in "imprint", which is also mentioned in this question.

This is caused by a much later change, Old Latin vowel reduction. Proto-Italic and Old Latin had accent on first syllable of word (it was diffferent in Classical Latin). During some period of Old Latin, short vowels in non-initial syllables were reduced. Long vowels were affected only in last syllable, but diphthongs were affected somehow more. Details are complicated and may be reconstructed differently by different linguists. You can read this, this and this page in Wikipedia.

Verb premō, premesi /premezi/ had /e/ in initial syllable, so it was not affected by vowel reduction. However last vowel was reduced:

premō, premesi /premezi/ → premō, premere

Verb derived with prefix in- was originally:

in- +‎ premō, premesi /premezi/ → impremō, impremesi /impremezi/

/e/ in second syllable was reduced into eventually /i/:

impremō, impremesi /impremezi/ → imprimō, imprimere

(Intervocalic s [z] changed to r [r] due to rhotacism.)

English word imprint comes, through Old French, from Latin imprimō, imprimere.

  • This table gives a better overview of the vowel developments than the three separate pages you link to; in particular, it explains why the first /e/ became /i/ while the second and third remained /e/. (We would actually expect the first /e/ to become sonus medius [ɨ ~ ʉ ~ ʏ ~ ə] here, but there doesn’t seem to be any evidence for spellings in †imprum-, so it appears it simply became /i/, perhaps through analogy because it’s a transparent compound.) Commented Jun 14, 2023 at 10:48
  • @Janus Bahs Jacquet: Your link is exactly the same as third link in my original answer. Sonus medius seems to have been spelled with <i> or <u> (table gives examples: docimentum, documentum, optimus, optumus, lacrima, lacruma), so it is possible that second vowel in <imprimō> was sonus medius for some time.
    – Arfrever
    Commented Jun 14, 2023 at 14:02
  • @Janus Bahs Jacquet: Regarding "third remained /e/": As you probably heard, Latin infinitive was originally locative of athematic s-stem noun. See this (first *-os suffix which is athematic, not second suffix which is thematic). One of reconstructed variants of locative of this suffix in PIE is *-esi ((é) means accented e-grade of preceding root). Later development was, in Proto-Italic, quoting this: "*s was also allophonically voiced to *z word-medially.".
    – Arfrever
    Commented Jun 14, 2023 at 14:17
  • (Besides *-es and *-esi, another dialectal variant of locative of this suffix seems to have been *-esen, which gave rise to Proto-Hellenic *-ehen, next Ancient Greek -ειν (which is spurious diphthong and was pronounced /eːn/, never /ei̯n/ in earlier times).)
    – Arfrever
    Commented Jun 14, 2023 at 14:27
  • You’re right, that is the same as your second link! I must have clicked the wrong thing before and missed it. Regarding sonus medius, the thing is that there don’t seem to be any examples at all of any verb containing premō where it’s written -prumo. If it had been sonus medius, we would expect that to be reflected in writing, just as it is elsewhere. This leads me to believe that in this group of words, the sound was always /i/, for whatever reason. Regarding the development of e’s, my point was just that you don’t specify in your answer that /i/ > /e/ at the end and /e/ > /i/ → Commented Jun 14, 2023 at 15:12

Quite simple. The Proto-Indo-European language had vowel ablaut. This means the root vowel could manifest in several ablaut grades: as -e-, as -o-, as long -e-, as long -o- or disappear. When the vowel would disappear, it is said the ablaut grade was "zero-grade".

This means, the initial root *per- could inflect in various word forms or derivations to become *pr-.

So, it is conjectured that the PIE root *pres- meaning "to press" is derived from the root *per- "first, forward" by adding the suffix -es. Alternatively, it could be a separate, unrelated root.


This is process called metathesis, and it is very common in language change. See the "English" section of that Wikipedia article for some examples of very familiar English words that have been shaped by metathesis.

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.