Re: Ancient Greek Phonology

I must apologize for trying to add a few details to the other replies
without having any AG materials handy. That is, every line in this post
should have been marked 'IIRC'. Still worse, some details were obtained
not from books or other publications, but from personal communication with
specialists in various fields of linguistics, who pointed to borrowings
from AG into other langs - but regrettably I cannot cite a single example
by heart. So my generalizations, not thoroughly traditional at any rate,
still must be checked. And, finally, I am not an AG connoisseur at all;
rather, I am 'phonetically conscious' like Oskar, and I used to ask many
people a lot of questions, trying to make things clear to myself.

On Tue, 19 Sep 2000 01:53:54 GMT, Oskar Gudlaugsson <hr_oskar@...>
wrote:

> Also, how come /y/ 'y
>psilon' has been rendered as either 'y' or 'u' in English (and probably
>Latin)?
In English it's a late scholarly innovation (which BTW I don't like). For
Latin, see below.

>There has to be, as always, a simple answer to this all, making my question
>silly.
Unfortunately, things aren't that easy.

First, a few things that seem to be generally accepted.

The oldest known stage of the AG monophthong system was as follows:

short:                            long:

i (iota)    u (upsilon)       i: (iota)     u: (upsilon)
e (epsilon) o (omicron)
                              E: (eta)      O: (omega)
       a (alpha)                       a: (alpha)

It was more or less uniform across dialects (as a system, while different
dialects could have different vowels in same words).

Note that epsilon and omicron denoted *narrower* vowels than the respective
long ones, contrary to what became typical at later stages.

Then the system of long vowels was transformed by a nearly universal
change: the monophthongization of [ei] and [ou], as well as the
contraction of certain vowel combinations, produced long [e:] and [o:].
In some local orthographies these new sounds were denoted with epsilon
and omicron (that is, the use of letters was initially based on vowel
quality rather than quantity), but more widespread was the spelling
system where they were written the same way as the former diphthongs
(that is, epsilon+iota and omicron+upsilon).

Then there was a strong tendency for [e:] and [o:] to become still
narrower, approaching [i:] and [u:]. However, this change produced
different results in different dialects.

It was somewhere at this stage that massive borrowing of AG words into
Latin began. This older layer of loanwords uniformly has <u> for upsilon.
It seems that as late as in the times of Plautus rendering this sound as
[u(:)] was the norm. The former AG diphthongs [ei] and [ou] were rendered
as [i:] and [u:], but these could result from the internal development
in Latin.

Up to this moment, everything is quite symmetrical, right?

Now, where the trouble begins. In some AG dialects, including Attic (note
that the dialects that first came in contact with Latin didn't belong to
this group), [u(:)] underwent *some* change. Perhaps, to keep [u:] apart
from the narrowing [o:] > [u:]. As the Attic dialect (and Koine) spread
over Oikoumene, becoming the exemplary form of AG, Romans introduced
a special letter (<y>)in their alphabet for this Attic vowel. However,
in later forms of spoken Latin this 'unnative' sound was lost and merged
with [i(:)].

It seems that some ancient sources do describe this vowel as something
intermediate between [i] and [u]. But next is the point where the most
popular interpretation cannot be treated as 'generally accepted' any more.

The matter is that the evidence for describing the sound of upsilon as
'narrow, front, rounded' is not lacking... but belongs to *much* later
period (from ca. 12th century on, I'm not sure about exact datings).
There is no problem with this later system. By that time, long vowels
(including most diphthongs) shortened, producing a rather natural
inventory of monophthongs:

i (iota, eta and <ei>)    ü (upsilon and <oi>)  u (<ou>)
e (epsilon and <ai>)                            o (omicron and omega)
                          a (alpha)

(If it were one of my conlangs, I'd find a way to add [ö] in the middle...
but tolerable as is ;) )

However, as you correctly noted, for the period before vowel shortening
the traditional interpretation yields something rather unnatural for the
short vowels (it's better not to touch the long ones here, since there is
another load of problems with diphthong contraction):

i (iota)           ü (upsilon)
e (or E?)(epsilon)              o (omicron)
                   a (alpha)

As stated above, omicron formerly denoted a rather narrow vowel. Its long
counterpart, <ou>, had been narrowed to [u:]. And some loans in neighboring
languages do have [u] for omicron. On the other hand, short vowels were
probably 'lax' compared to the long ones. This allows (IMO) for a minor
reinterpretation:

i/I (iota)     (?) (upsilon)  U (omicron)
e/E (epsilon)
                a (alpha)

- which means nearly the same thing but looks considerably better :)

But the hardest thing is the quality of upsilon. The problem is that AG
(and Medieval Greek) had few contacts with languages having a sound like
[ü]. And when they did, the result was sometimes weird. For example,
there are some early medieval inscriptions in Turkic languages using the
Greek alphabet. Surprisingly, they use upsilon (reportedly) for 'narrow,
BACK, UNROUNDED' (the sound of the Turkish dotless i), and invent special
ways to render their [ü]. Coptic (starting from 2nd century) mixes atonic
epsilon and upsilon in loanwords, while in native words unstressed epsilon
seems to stand for something like schwa. Finally, Arabs (from 7th century
on) often render upsilon as [u].

The only thing that can be IMO inferred from the above is that upsilon
*wasn't* any [ü]...

I think, several interpretations can be put forward to explain all this.
But what comes first to *my* mind is something displaced towards 'central'
and (partly?) delabialized.

Sorry for a longish message offering no simple answer to your questions,
but this is the way I see it...


Basilius