In the grammar section of Pharr and Wright’s Homeric Greek section 601, the following is stated:

The loss of one or more consonants in a word usually occasions the lengthening of the preceding vowel. This is called compensative lengthening. When it takes place, α, ι, υ = ᾱ, ῑ, ῡ; ε = ει; ο = ου.

I know that in later Greek, the pronunciation of ᾳ, ῃ and ῳ was as though there was no iota subscript (§ 506), but ‘in Homeric times the iota was probably sounded to some extent’ (ibīdem). Similarly, the dipthongs are all listed (§ 504) as having both vowels pronounced.

For these reasons, I would assume that compensative lengthening should have yielded the following:

  • α → ᾱ
  • ε → *η
  • ι → ῑ
  • ο → *ω
  • υ → ῡ

With my very sparse knowledge of Greek (I am still very much a beginner student), I can think of only three reasons for having ε → ει and ο → ου:

  1. Spelling conventions, similar to Old Latin EI for Ī, e.g. DEIVIDVNDA (Ward, Ralph L.: _Evidence for the Pronunciation of Latin (Continued)’, The Classical World, Vol. 55, No. 9 (Jun., 1962), pp. 273–275). ɔ: Though spelt ει and ου, they were in actuality pronounced η /eː/ and ω /ɔː/ rather than /ɛː/ and /uː/.
  2. These vowels actually changed to /ei̯/ and /ou̯/ when compensatively lengthened.
  3. These vowels actually changed to /ɛː/ and /uː/ when compensatively lengthened.

Why did ε and ο get compensatively lengthened to ει and ου and how were they pronounced? If we here have an example a change in vowel quality, what prompted this?

1 Answer 1


In earlier Greek, there were four different types of mid vowels:

  • The short monophthongs ε ο /e o/, inherited from PIE *e *o
  • The long monophthongs ει ου /eː oː/, created through contraction and compensatory lengthening
  • The long diphthongs ει ου /ej ow/, inherited from PIE *ei *ou
  • The open monophthongs η ω /ɛː ɔː/, inherited from PIE *ē *ō

Note that the long monophthongs and long diphthongs were written the same (in standardized Classical orthography), but pronounced differently! We know there was a phonemic difference at some point because the outcome varies in some dialects: "severer" Doric and Aeolic, for example, merged the long monophthongs with the open monophthongs, giving forms like φιλήτω for Attic φιλείτω (from a contraction ε + ε).

In Attic, though, the long monophthongs and long diphthongs merged, just like in Latin. Nowadays, the descendant of a long monophthong is sometimes called a "spurious" diphthong, and the descendant of a long diphthong "genuine". But phonologically there seems to have been no difference in Attic, which is why there's no difference in standard orthography.

P.S. Some epichoric scripts also divided up these four categories of mid vowels differently, providing additional evidence for a phonemic contrast. Corinthian inscriptions, for example, use one letter for short monophthongs and open monophthongs, another letter for long monophthongs, and sometimes a digraph for long diphthongs (though they were rapidly disappearing).

  • This has me wondering, were there any differences in pronunciation between ᾱ/ῑ/ῡ and α/ι/υ? I would assume they too could exist as compensatory lengthened vowels.
    – Canned Man
    Commented Jan 19, 2022 at 15:21
  • 1
    There is not good evidence of a difference in quality between ᾱ/ῑ/ῡ and α/ι/υ in Attic and Koine Greek generally and synchronically, but the ᾱ of proto-Greek did sometimes develop differently and diacronically from short α as seen by the results in Ionic and Attic, indicating a difference in quality at that time and in those dialects at the beginning of the change and during it. It is conjectured that long vowels in Indo-European were not originally phonemic, but developed through various processes of "compensatory lengthening." By the time of Greek, they were phonemic and unpredictable. Commented Jan 21, 2022 at 17:52
  • 1
    @Vegawatcher There's also some evidence for a separate /æ:/ phoneme at some points, separate from both /a:/ and /ɛ:/, which I find really interesting. But yeah, I don't know of any evidence for a quality difference in those long vowels.
    – Draconis
    Commented Jan 21, 2022 at 17:55
  • By separate, do you mean existing as dialectical variants; or as a distinct phonemes carrying meaning, in other words that both /aː/, /ɛː/ and /æː/ existed simultaneously within the same area?
    – Canned Man
    Commented Jan 31, 2022 at 13:24
  • @CannedMan The latter! For example we see ΚΑΣΙΓΝΗΤ𐌇 (for κασιγνητη "sister") in some Central Ionian inscriptions, with the "open" eta letter used to write inherited ē and the "closed" eta letter used for the result of ā > ē. I haven't found a word with both of those and also a long alpha in it, since the inscriptional corpus is not huge, but the accusative plural would be expected to be ΚΑΣΙΓΝΗΤΑΣ with a long alpha in the ending (long ā created by compensatory lengthening is not affected by ā > ē).
    – Draconis
    Commented Jan 31, 2022 at 16:49

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.