Wikipedia says about stress in Latinate English words:

In words of three or more syllables, stress falls either on the penult or the antepenult (third from the end), according to these criteria:

  • If the penult contains a short vowel in an open syllable, the stress falls on the antepenult: e.g. stá.mi.na, hy.pó.the.sis.
  • If the penult contains a long vowel; a diphthong; a closed syllable (with any length of vowel); or is followed by z, the stress falls on the penult.

[...] The fact that decorum is stressed on the penult, and exodus on the antepenult, is a fact about each of these words that must be memorized separately (unless one is already familiar with the Classical quantities, and in the former case, additionally with the fact that decus -ŏris n. with short -o- syllable became in late Latin decus/decor -ōris m. with long -o- syllable: 'Dómine, diléxi decórem domus tuæ').

However, English speakers often pronounce words like this with penult stress instead:

  • stigmata (coexists with antepenult-stressed variant)
  • schemata (coexists with antepenult-stressed variant)
  • Uranus (coexists with antepenult-stressed variant)
  • uroboros/ouroboros (Wikipedia: /jʊərɵˈbɒrəs/, /ɔːˈrɒbɔrəs/; Collins: /uːˈrɒbəˌrɒs; ˌuːrəˈbɒrəs/)

Is there any linguistic explanation of this? Native English words don't seem to me to have a general preference for the penultimate syllable, but I may be wrong — please tell me if there's evidence for this.

It does seem to me that for many English speakers, penult stress is the default for pronouncing foreign words. For example, in the Anglicized pronunciation of words from Japanese, Japanese pitch-accent seems to be totally ignored, and penult stress is common (but not universal; see "samurai," "tycoon," and some pronunciations of "haiku" for counterexamples). Has there been any linguistic research that confirms this idea? If so, do we have any idea of the conditions under which this tendency applies to the pronunciation of a word (does it depend on apparent language of origin, phonological structure, analogy to other specific English words with the same spelling) and how long it has been active historically?

  • I don’t really see how the examples you give are inconsistent with the general rule (except perhaps ouroboros, for which all the transcriptions you lost look odd to me; I’d pronounce it /jʊˈrɒbərəs/ or /jʊrəbɔːrəs/, with either /jʊ/ or /ɔː/ in the first syllable). Stigmata, schemata and Uranus are paenults when the vowel is long /ɑː/ (/eɪ/ for Uranus), and antepaenults when it’s short /ə/. May 1, 2020 at 6:55
  • You could say that native (= inherited) words do have a general preference for paenultimate stress, but only because inherited words generally have initial stress and are most commonly no more than two syllables. This doesn’t go for the plethora of French borrowings which are perceived as every bit as native as the actually inherited words, though, so it’s probably irrelevant here. May 1, 2020 at 6:57
  • @JanusBahsJacquet: the "short vowel" part of the rule is supposed to be based on vowel length in Latin. In English (if this kind of thing is treated as part of English phonology at all), there seems to be disagreement about whether vowel length causes penult stress or penult stress causes vowel length for open syllables. May 1, 2020 at 7:16
  • Oh, Latin vowel length. That makes more sense. I would surmise that the ‘correctly’ stressed variants are the original ones in English, with the ‘wrong’ ones being later reading (?) pronunciations which started in people less familiar with Latinate borrowings – but that’s only guesswork. May 1, 2020 at 7:19
  • @JanusBahsJacquet: It does seem plausible that the pronunciations with stress placement that is not expected based on that rule were based on spelling. In that case, I think there is an interesting linguistic question of whether there are any systematic tendencies to how English speakers guess at the pronunciation of these words from the spelling pattern. I found a source, Greg Brooks' Dictionary of the British English Spelling System, that indicates that the spelling pattern "a + C + a/i/o" tends to be associated with penult stress. May 1, 2020 at 7:24

2 Answers 2


To summarize, I don't know. But the question seems a little fuzzy, to me. Are you concerned with predicting how such words will be borrowed into English given their source forms, or is it about how English speakers (not necessarily the borrowers) will say them? And are you just looking at the spelling of the English word?

In the SPE analysis, "decorum" could be assigned an underlying form with a tense penult, and "exodus" could be assigned an underlying form with a lax penult, and that would predict the correct stresses, on the penult and antepenult, respectively. The fact that they are both spelled with "o" in the penult is neither here nor there. I don't know how they were borrowed.

"Tycoon" and "samurai" aren't much like the other examples you're concerned with that have their last stress on the antepenult or penult, because these words have stress on their ultimate syllables, just as one would expect, since the ultimate syllables have tense vowels.

  • I don't care only about the spelling, but I imagine the spelling has a large influence on how speakers pronounce words like this. In terms of lax-tense vowels, what would be the relevant difference between "tycoon" (always stressed on the last syllable) and "haiku" (often stressed on the first)? Oct 7, 2015 at 1:36
  • I'm especially interested in the historical aspect: when did these forms get established? I'm also interested in knowing if there are any factors that help predict the stress pattern with which these words will be borrowed. Oct 7, 2015 at 1:39
  • Both "tycoon" and "haiku" are stressed on the last syllable and also the first syllable, but "haiku" has the regular retraction of primary stress to the preceding stress when the primary stress would fall on the last syllable. There are many words like this with final secondary stress. "Tycoon" is the exception, and there are quite a few words like this which are spelled with "oo" and retain final stress -- "festoon", "cartoon", ...
    – Greg Lee
    Oct 7, 2015 at 1:46
  • I don't know anything about the history end of this.
    – Greg Lee
    Oct 7, 2015 at 1:50

Consider the examples you have given more closely. Surely, the key is that in the variants with antepenultimate stress, the penultimate vowel sound is reduced to a near-schwa or schwa sound. Conversely, in the penultimate stress variants, the vowel will likely be Anglicized in a different way. Note that in either pronunciation of "Uranus", the "a" sounds nothing like a Latin a.

In this way, the general principle of the rule you cite is maintained.

When we initially hear foreign words, we will naturally try to approximate the pronunciation. There is no good reason why variant pronunciations of foreign-origin words shouldn't exist, especially when English pronunciation itself is so varied - no doubt influencing the pronunciation of the foreign word.

  • But, the rule I cite predicts that penult stress will not be used in words like this, so I don't see what general principle is maintained. In terms of imitating pronunciation, nobody nowadays can hear native speakers of ancient Greek or Latin. Oct 7, 2015 at 1:42
  • So many things could have happened as "Uranus" made its way into English. Here's an implausible one: it was borrowed from Swedish /ɵˈrɑːnɵs/ (first other pronunciation I found with penultimate stress). Mark, a more ordinary analysis would be that the penult is reduced because it is unstressed, rather than that it is unstressed because it is reduced.
    – Greg Lee
    Oct 7, 2015 at 5:12
  • @GregLee I should have been more clear. I wasn't trying to say which came first - the stress or the reduction - but just that what was logical and important was that they came hand-in-hand. Otherwise, the words would sound very "foreign" indeed.
    – Mark D
    Oct 7, 2015 at 16:35

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.