Re: Ancient writing systems

On May 14, 3:57 am, Neeraj Mathur wrote:
On 6 May, 22:03, "Douglas G. Kilday" wrote:
On May 2, 4:49 pm, António Marques wrote:
Paul J Kriha wrote:
Interesting. I know hardly any Latin. Is it generally accepted than
GAIVS had indeed 3 syllables (GA-I-VS)?

No, this conclusion is not generally accepted.

The matter of syllables notwithstanding, I find this a very muddy
conclusion. There's another phonemic element at play which is stress and
can readily account for having a particular realisation in place. E.g..,
/'maius/ vs /ga'ius/, /'kui/ vs /ku'i/, etc.

There is no evidence that any of these words ever had the stress on
the final syllable.

Or is someone suggesting the distinction was ['majus] vs ['gaius],
['kuj] vs ['kui]? The latter sound much weirder than admitting the
difference was the position of stress.

This is not the distinction. The distinction between the first two is
['majjus] (as Douglas also states) versus ['gajus] (where I shall
disagree with Douglas). The distinction between the second is
primarily in the consonants; I'll use Douglas' notation of the {w} to
indicate the the preceding consonant is labialised: the word <cui> is
[kuj], and <qui> is [k{w}i].

In these particular words, stress-position is not an issue.  The Greek
transcription Γάϊος shows that GAIVS was pronounced /'gaios/.

It would only show that if the Greek into which the word was being
transliterated had a clear way of distinguishing between the vowel [i]
and the glide [j]. It didn't, and so it doesn't. A very convincing
argument could be made that, rather, this was seen by the transcribers
as the best way to represent the glide, which was no longer a phoneme
of the target language. I don't know the evidence for the diaeresis,
but if it is in fact ancient, all it shows again is that the word did
not contain the diphthong spelled alpha-iota. (Do we have a diaeresis
in the record at a time when it is fairly likely that that diphthong
had not already begun to monophthongise into [e]?)

It is in printed texts of Polybius. Of course, it is highly unlikely
that P. used any accents or diaereses in his writing. They were added
centuries later when some scribes started including them in Greek
texts. But the marking on the Greek form of the praenomen must
reflect traditional pronunciation, and I see no reason to suppose that
it does not go back at least as far as Polybius.

The Greek transcription cannot rule out a pronunciation [gajus], and
I'm sure that prosody will back such a disyllabic interpretation (or
are we disallowing prosody in this discussion?).

Some passages in Martial and Catullus require a trissyllabic
<Ga:ius>. This is in agreement with the Oscan praenomen <Gaavis>, if
we assume that the Roman form came from a (Sabinizing?) dialect that
dropped the /w/, as also in <Gnaeus>, for which the expected form is
*Naevus.

MAIVS
on the other hand had a geminated semivowel /'majjos/, which can be
inferred from both "before" and "after": the protoform *magjos and the
Italian <maggiore>.  Dictionaries usually mark the first vowel as
long, <ma:jus> or <ma:ius>, but while this works for scansion, it is
technically incorrect: the failure to mark geminate -jj- is another
defect of classical Latin orthography.  One old textbook remedied the
situation by writing <mâius>, <Pompêius>, etc., but this diacritic
device never caught on.

I am in complete agreement with the opinions of the above sentences.

With CVI /'kuj/ and QVI /'kwi:/ we have arguably a reason for
retaining Q (as if any "reason" were needed; only conservatism
accounts for KALENDAE, KARTHAGO, KARISSIMI, etc.).

Why would we have seen fit to drop it in the first place?

All it did was mark the back allophone of /k/. Certainly you have
seen PEQVNIA and the like in inscriptions.

There's also the question, given /i/ and /u/ with [j] and [w]
allophones, in which of 4 possibilities is /ui/ to be pronounced? [wj]
and [ui] are odd. Can [uj] and [wi] contrast wihtout putting into
question the unity of /u/ and /i/? It's not like they're a proper
minimal pair.

This question seems to ignore the labiovelar consonants. The sequence
<VI> represents [uj] or [ujj] (these two not being distinguished,
admittedly) unless the preceding consonant is a labiovelar. If it is
the voiceless labiovelar, then we have missegmented, for <QV-> makes a
single unit in the Latin alphabet - that is, it is a digraph, as it is
in English (as evidenced in those archaic spellings where this unit
represents [k] rather than [k{w}], such as <QVOM>). The only
theoretical place of confusion is with the voiced labiovelar, where
the sequence <GVI> could represent either [gui] or [g{w}i]; as it is,
largely due to the distribution of g{w}, I don't think are any actual
places where this confusion could occur (unless there are some fourth
declension nouns I've forgotten which have a stem ending -ngu). The
other possibilities are the first and the last two in Douglas' list
below, which are predictable from syllabification and distribution
rules.

Latin VI has these pronunciations that I can think of:

/ui/ in FVIT, DVIM, etc.
/ui:/ in FVI, MONVI, etc.
/uj/ in CVI and HVIC
/ujj/ in HVIVS and CVIVS (note pre-class. QVOIIVS)
/wi/ in VITVLVS, NOVIT, etc.
/wi:/ in VITA, NOVI, etc.

If labiovelars are distinguished (which seems pointless for synchronic
description,

That's a rather cavalier attitude. It's certainly not pointless, and
your examples below (which my interjected comments have displaced)
should be included; you've already presented a strong case for
including labiovelars in the distinction of qui and cui. Besides, you
allowed diachronic arguments to establish your geminate -jj-; the
diachronic evidence working in both directions strongly argues in
favour of labiovelars for Latin pretty much up to the emergence of
Romance (which phrase I know is improper, but we both know that we
would both unpack it in more or less the same way).

Phonetic, but not phonemic, labiovelars. In Sardinian and Rumanian
where we have labial stops as reflexes we might be tempted to
postulate a post-classical phase in which the labiovelars became
phonemic, but I am not sure that it is necessary. If we use a
mechanism like

[k{w}] > [k{p}] > [{k}p] > [p]
[g{w}] > [g{b}] > [{g}b] > [b]

then all we are dealing with is assimilation and reduction. (I
suspect that the above is essentially the P-Italic mechanism, and that
it occurred later than commonly supposed in Oscan and Umbrian.)

and consistency would then demand a labiosibilant as
well)

I'm not sure that this holds true. As an illustrative example, when we
work with IE, we reconstruct labiovelars because they contrast with a
sequence of labial and velar; in Latin, the establishment is based on
a combination of arguments from orthography, prosody, and diachronic
sources and developments. I can't see how any of this applies to
sibilants. Is there an orthographic counterpart to the qui / cui
contrast that applies to sibilants, or an instance where <SV> fails to
'make position' in verse, or a pattern of treatment in the development
of Romance varieties where reflexes of SV cannot be understood as the
sum of the parts (which applies, for instance, to such forms as 'pis'
from 'quis' and so on)? I find your arguments that a synchronic
description of Latin should either exclude labiovelars or admit them
only with what you term a 'labiosibilant' rather uncharacteristically
hasty and fairly inconsistent. (This avoids the more general point
that rarely are arguments about contrasts in velar stops used to infer
symmetrical conclusions about sibilants.)

Hexametric examples of <SV> not making position:

Lucretius 1:7
ADVENTVMQVE TVVM TIBI SVAVIS DAEDALA TELLVS

Ennius ap. Gell. 12:4
INGENIVM CVI NVLLA MALVM SENTENTIA SVADET

This behavior is parallel to that of <QV> failing to make position.
(But -C#V- across a word-division makes position, as does -S#V-). And
you must know about the Old Latin labiodental <DV>, labialized to <B>
in Latin. There is nothing outlandish about a phonetic
labiosibilant. It acts like a phonetic labiovelar, but it is a
sibilant and not a velar.

we would also have:

/{w}i/ in NEQVIT, NINGVIT, etc.
/{w}i:/ in QVI, EQVI, etc.

I think these should clearly be added to the list.

To further analyse (and probably to make a claim which somebody will
find holes in, for which I apologise in advance):

1) Where <VI> is morpheme-initial, the only interpretation possible is
[wi(:)].

2) Where <VI> follows <Q> or <-NG>, the <V> is a part of a digraph
indicating a labiovelar consonant; thus the only possibility is
[{w}i(:)]. (The voiced case is the only one where real ambiguity could
potentially exist, though practically I don't believe there are any
examples where this statement does not hold.)

No, since Latin reduced inherited *gw to /w/, unless the nasal
preceded.

3) In any case where the <VI> appears between two consonants but rule
2 does not apply, the two consonants must not be in the same syllable
- that is to say, the diphthong [uj] cannot appear in a closed
syllable. This rule appears to have a single exception, the word
'huic' (which I note is the product of apocope from 'huice', in which
the rule would apply, as the second consonant belongs to a different
syllable and thus the diphthong is in an open syllable).

4) In any other case, <VI> represents either the diphthong [uj] or the
diphthong followed by a second glide, [ujj].

Does that sound rigorous and accurate?

I can find no fault with it.

.

Relevant Pages