Re: Celery

From: Richard Wordingham
Message: 15549
Date: 2002-09-18

--- In cybalist@..., "m_iacomi" <m_iacomi@...> wrote:
> --- In cybalist@..., "Richard Wordingham" wrote:

> > We may need a detailed knowledge of the phonetics of the Greek
> > dentals when the Dacian and Thracian words were recorded, and who
> > wrote them down. The Greek system was not stable!
> > (Note [] for sound, // for phoneme, <> for written form.)
> >
> > <s> Sigma: [s] (and this is stable!)
> > <z> Zeta: [dz] > [z] (I'd appreciate confirmation, as well as
> > dates)
> The phenomenon might have occurred in Dacian/Thracian (see 'zios')
> up to some moment (I-st, II-nd century?!). The phenomenon occurred
> surely in Daco-Romanian, after splitting from southern dialects
> (around X-th century A.D..)

I was talking about Greek sound changes!

> > Plausible Dacian and Thracian sounds:
> >
> > [s], [z], [dz], [dZ], [d], [g^], [ð], [ts], [tS], [Þ], [k^]
> As said, I have strong doubts for [ð] and [Þ]. The others seem
> to be fine.

My thoughts were partly influenced by Piotr's remark that Albanian
_might_ be descended from Dacian. Cyril's website gave PIE *d >
Albanian ð, but that must be wrong; it does not square with the
derivation of Albanian ditë 'day' from PIE *dei-.

Now, for Dacian, we appear to have:

PIE *t > Dacian <t>, probably [t]
PIE *d > Dacian <d>, probably [d]
PIE *dH > Dacian <d>, probably [d], but possibly [ð] as in Germanic.

PIE *k^ > Possibly any one of [s], [tS], [ts], [þ]
PIE *g^ > Possibly any one of [z], [dZ], [dz], [ð]
PIE *g^H > Possibly any one of [z], [dZ], [dz], [ð]

For Thracian, we appear to have:

PIE *t > Thracian <t> or <th>, probably [th] or [þ]
PIE *d > Thracian <d> or <t>, probably [t]
PIE *dH > Thracian <d>, probably [d], but possibly [ð] as in Germanic.

PIE *k^ > Thracian <s> or <z>
PIE *g^ > Thracian <z> or <d>? or <s>?
PIE *g^H > Thracian <z> or <s>

These all look consistent with [dz], [dZ], [ts] and [tS]. I would
plump for PIE *k^ > Thracian [ts], PIE *g^, *g^H > Thracian [dz], but
they might all be [ts].

IE *g^ > Thracian <d> is evidenced by *dentu- 'clan, tribe' in
Duridanov (at ) derived
from PIE g^en-. It looks odd.

PIE *g^ > Thracian <s> is evidenced by saldas for golden, derived from
PIE *g^Hult-. However, this derivative from the root is otherwise
only seen in Germanic, Baltic and Slavonic. There is also a
comparison of sem(e)la `land, earth' with Slavonic forms such as

In one or both there is also a change <di> > <z> giving an extra
source of affricates.

> >> I suppose Latin /z/ for Dacian /ð/ is plausible.
> I doubt about the presence of /ð/ in Dacian. Neither Russu, nor
> Poghirc now Wald mention something about it. What examples you have
> for it? The same stands for [þ].

I can't find any decent list of Dacian vocabulary! Plausible sources
of [ð] are PIE *dH, *g^H and *g^. The only way of identifying it
would be a hesitation between <d> and <z>, but there is no guarantee
that such hesitation would occur. Moreover, if Thracian *dentu- is
derived from PIE g^en-, such hesitation might not enable us to decide
between [ð] and [dz].

The lack of [þ], which in Dacian would be from PIE *k^, if adequately
demonstrated, makes Dacian [ð] from *g^H and *g^ unlikely. Such a
lack would tell us nothing about whether the reflex of PIE *dH was [d]
or [ð].

> > > IE *dhwer- > Thr. dero, dur 'a stockade' [*dh > d]
> > > IE *deiwo- > Thr. desa/disa 'deity, god' [*d > d] but also:
> >
> > Where does the <s> come from? Duridanov compares it to Greek
> > théos 'god', from PIE *dHe:s- (or is it *dHeh1s-?), the source of
> > Latin fa:num 'temple'.
> The examples were taken from the Thracian glossary available at

The statement there, 'Note: The Glossary is partly reconstructed from
hydronyms, place names and other traces of Dacian and Thracian
languages.' immediately makes me worry. What is Dacian doing in a
Thracian glossary? I also noticed a few other errors:

1. Latin *pibo: 'drink'; the Latin is bibo:.

2. It implies Latin deus 'god' and Greek Zeus are cognate. The PIE
origins are *deiwos and *dye:us respectively, the former presumably
*dei-w-o- and the second *di-e:u-s. Is *-e:u an otherwise known
ablaut form of the *-w- suffix?

> >> As said, only /di/ Lat. > /dzi/ > /zi/ Rom. or /de/ > /*die/ >
> >> /dze/ > /ze/. Otherwise, /d/.
> OK, I'll give now the precise rule, I mentioned above only some
> general cases when /d/ > /z/.
Thank you for the details. Perhaps I should code them up (for the
sound change program at ) to make sure I have
understood them. However, at present I am struggling with a practice
run deriving French from Latin.