You are on page 1of 24

Cows, Houses, Hooks: The Graeco-Semitic Letter Names as a Chapter in the History of the

Alphabet
Author(s): Andreas Willi
Source: The Classical Quarterly, New Series, Vol. 58, No. 2 (Dec., 2008), pp. 401-423
Published by: Cambridge University Press on behalf of The Classical Association
Stable URL: http://www.jstor.org/stable/27564173
Accessed: 21-04-2015 09:42 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Cambridge University Press and The Classical Association are collaborating with JSTOR to digitize, preserve and extend
access to The Classical Quarterly.

http://www.jstor.org

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

Classical

58.2 401-^23

Quarterly

(2008) Printed

in Great

Britain

401

doi:10.1017/S0009838808000517

COWS, HOUSES,
LETTER NAMES

HOOKS:
THE GRAECO-SEMITIC
IN THE HISTORY
AS A CHAPTER
OF THE ALPHABET
I. INTRODUCTION

It is well
and/or

known
Greek

encounter

first

can be traced back,


via the Latin
that all modern
European
alphabets
we
to a Semitic
ancestor.
This
ancestor
which
alphabet,
alphabet,
as the Proto-Canaanite
in the first half
have been created
script, may

of the second millennium


This

influence.1
dated

between

can
signs
and Old

Proto-Sinaitic
Proto-Canaanite
Phoenician

(from

eleventh

in the Sinai peninsula and under Egyptian

B.c., possibly

is the region
the eighteenth

where
and

the
the

century)

Proto-Sinaitic

thirteenth
as

be

regarded
Canaanite

so-called

inscriptions
(variously
found. The
B.C.) have been
or at least cognates
of
the
as
as
the
well
centuries)

centuries

predecessors

(seventeenth-eleventh
letters.

Apart from the formal similarities of the early alphabets, the Greek letter names
provide the most obvious proof of the Semitic ancestry of our script: not only do
most

of

number

them
can

also

closely
correspond
be etymologized

to

the Hebrew
on

easily

the basis

letter
of

but

names,

Semitic

a considerable

lexemes.

The date, place and circumstances of the adoption of the alphabet by the Greeks
have been the subject of much scholarly debate in recent years. Since the study of the
letter names promises little insight into these hot topics, it has been
Graeco-Semitic
pursued
up-to-date

This
frequently.
status
quaestionis

less

article

aims

as well

as

to
some

redress
new

the

balance

suggestions

by

presenting
the

regarding

an
early

history of the letter names in both Semitic and Greek.

of the alphabet', Eretz-Israel


'The origin and early evolution
8 (1967),
Cf. now F.M. Cross,
to the History
I (London,
at 8*-12*; D. Diringer,
A Key
The Alphabet:
of Mankind,
to Alphabet
Semitic Writing from Pictograph
(London,
19763), 140^1
19683), 160-3; G.R. Driver,
et al. (edd.), The
in J. Boardman
'The earliest
and 185-97; B. Isserlin,
alphabetic
writing',
East and the
III.l. The Prehistory
and the Middle
Ancient History,
of the Balkans,
Cambridge
b.c. (Cambridge,
at 794-802;
J.Naveh,
1982), 794-818,
Aegean World, Tenth to Eighth Centuries
8*-24*,

to West Semitic Epigraphy


An Introduction
and Palaeography
of the Alphabet.
Early History
in
and its Development
1982), 23-42; B. Sass, The Genesis of the Alphabet
(Jerusalem and Leiden,
B.c. (Wiesbaden,
the Second Millennium
On the
1988), esp. 135-68; B. Sass, Studia Alphabetica.
Semitic, South Semitic and Greek Alphabets
(Fribourg
of the Northwest
Origin and Early History
in V. Krings
and G?ttingen,
1991), 24-7; W. R?llig,
(ed.), La civilisation ph?nicienne
'L'alphabet',
at 193-6;
et punique. Manuel
de recherche (Leiden, New York and Cologne,
1995), 193-214,
zu den Griechen',
and A. Kyriatsoulis
in N. Dimoudis
W. R?llig,
und sein Weg
'Das Alphabet
der hellenischen
1996)
(Tagung Ohlstadt 3.-6. Oktober
(edd.), Die Geschichte
Sprache und Schrift
'The Egyptian
work of A.H. Gardiner,
1998), 359-86, at 360-2, after the pioneering
(Altenburg,
3 (1916),
Journal of Egyptian
1-16; W.F. Albright,
Archaeology
origin of the Semitic alphabet',
BAS OR 110 (1948), 6-22;
from Sinai and their decipherment',
'The early alphabetic
inscriptions
and
and their Decipherment
The Proto-Sinaitic
and W.F. Albright,
(Cambridge, MA
Inscriptions
London,

1966).

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

402

GRAECO-SEMITIC

LETTER

NAMES

II. THE EARLIEST ATTESTATIONS OF THE GRAECO-SEMITIC


LETTER NAMES
the

Paradoxically,
are not
names
which

during

found
there

earliest

attestations

in the Semitic
a single

is not

for

world

the

entire

series

they

originated.

where

piece

of

of

the

alphabetic

After
evidence

unquestionable2

letter

long centuries
for a Semitic

letter name, they all come to the fore within just a few decades: in their Hellenized
in Greek

forms

and

texts. This

of Athenian

bloom

to the

sudden

culture

increased

and

the most

among
overview

shows

important
that only

factors:

to the

B.C. on the one hand,


century
of language
and writing,
from
arising
the other. Thus,
the Platonic
dialogues,

aspects
as a whole,
on
in which
the phonetic

about

interrelated

in the fifth

in various

interest

the spread of literacy


in Greece
and in particular
the Cratylus
are central
to the discussion

to two

is due

appearance
literature

the

values

of

the various

letter

signs

of

are
legitimacy
etymological
speculation,
sources
for the Greek
letter names.
The
early
following
two names
of the classical
are absent
series
from Plato's

and ^et):3
writings (?JLV
aX(j)a (Crat. 393e etc., Eryx. 395c), ?rjra (Crat. 393e, 43le, Tht. 203b), y?pcpua (Crat.
427b), S?Xra (Crat. 403a etc.), et (Crat. 41 le etc., Tht. 207e, 208a), fr?ra (Crat. 418c
etc.),
412e),
403a),

393e etc.),
Kanna
l ra (Crat.
399b
(Tht. 207e),
etc.),
6rjra
(Crat.
vv (Crat. 414c
ov
403a
Tret (Crat.
414c
(Crat.
etc.),
Xd?8a
etc.),
(Crat.
etc.),
rav (Crat.
pw (Crat. 414c etc.),
(Crat. 402e, Tht. 203a etc., Eryx.
395c),
oiyfia

393eetc,
36b),fat
Even
'Letter

(Crat.

rjra

Tht. 208a, Phdr. 244c), v (Crat. 393d), <j>e?(Crat.427a), x^(Crat.


(Crat. 421b, 427a), d> (Crat. 420b, Phdr. 244d, Tht. 203a etc.).

before

Plato,

Tragedy'

a complete
(or T'papularlktj

enumeration
Oecopia

was
'Letter

given

in the
Epatar

Embassy'4)

by

414c, Tim.

lkt? rpaycoS?a
the Athenian
comic

poet Callias.5 In this play, presumably to be dated to the 430s, each of the 24 members
of the chorus represented one of the 24 letters of the classical (East Ionic) alphabet.6
2
On the uncertain
biblical attestations
of some letter names see ? III below.
3
On the letter names
in the Cratylus
cf. W. Stefanski,
'On the names of the letters used in
Plato's Kratylos\
Eos 80 (1992), 53-60; on the accentuation
of o?yp,a B. Einarson,
'Notes on the
of the Greek alphabet', CPh 62 (1967), 1-24 and 262-3, at 21-2, n. 38.
development
4
at Athen.
The transmission
7.276a and 10.448b supports
the title rpappLariKr)
hia,
rpay
the one at Athen.
10.453c the alternative
dew p?a. Since the play was clearly a
rpappariKrj
a
evidenced
the
number
of
chorus
comedy
(as
by
members),
parodistic rpappariKT]
rpaycohia
into rpap,pariKr)
de pia by a corrector.
may have been mistakenly
changed
5
Earlier doubts about the identity of the author with the comic poet Callias
(e.g. in U. von
review of A. Wilhelm,
Urkunden dramatischer
in Athen
Wilamowitz-MoellendorfF,
Auff?hrungen
are hardly justified:
at 631-2)
see
[Wien, 1906], G?ttingische
gelehrte Anzeigen
[1906], 611-34,
C.J. Ruijgh,
'Le Spectacle des lettres, com?die de Callias
avec un excursus
(Ath?n?e X 453c-455b),
sur les rapports
entre la m?lodie
du chant et les contours
du langage
m?lodiques
parl?',
ser. 4, 54 (2001), 257-335,
at 268-71,
after M. Pohlenz,
'Die Begr?ndung
der
Mnemosyne
abendl?ndischen
zu
durch die Stoa', Nachrichten
der Gesellschaft
der Wissenschaften
Sprachlehre
n.s. 3 (1939), 151-98, at 152-4.
G?ttingen
6
The East Ionic alphabet was officially
established
in Athens
in
under the archon Eucleides
403/2 B.c., but by that time it had been used in private inscriptions
for decades;
cf. L. Threatte,
I. Phonology
The Grammar of Attic Inscriptions,
(Berlin and New York,
1980), 33^4; L. Bodson,
et implications
culturelles
des adaptations
de l'alphabet attique pr?liminaires
'Aspects techniques
? la r?forme de 403/2', in C. Baurain, C. Bonnet
Lire
and V. Krings
grammata:
(edd.), Phoinikeia
et ?crire en M?diterran?e
in Aristophanes
S. Colvin,
and the
Dialect
(Namur,
1991), 591-611;
in Ancient
Politics
Greek Literature
A.J. D'Angour,
of Language
(Oxford,
1999), 92-100;
at
and the reform of the Athenian
43 (1999),
BICS
'Archinus, Eucleides
109-30,
alphabet',
112-14 and 122; Ruijgh
(n. 5), 269-71.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

403

WILLI

ANDREAS

The entrance of this peculiar chorus is described in one of the extant passages (Callias
test.

7 K.-A.,

transmitted

by Athen.

10.453c):

<ro

oY?ra, Oeov y?p ef,7


?X(f>a,> ?rjra, y?ppa,
?,t?t\ rjra, Br?r\ l ra, Kanna,
Xd?Ba, pv,
rav, <ro>
v,
vdy ?et, ro ou, 7T t, pw, to o?ypa,
1
T
L LSTO 00.
<TO>
<TO>
Tu)
TTCLpOV
^
(j)?L,
l/j

Further

and

dramatic,

comic,

especially

and

passages

the

confirm

fragments

impression that by the end of the fifth century B.C. a knowledge of the letter names
could

be

taken

Additional
Pindar,

for

attestations
Herodotus,

in various

Attic

among

granted
from

the classical

the majority
or early

in the Corpus Hippocraticum,


of the fourth
inscriptions
century.9

in classical
times
pupils
already
even before,
the corresponding

learned
letter

of

We

the canonical

shapes,

just

the Athenian

post-classical
in Xenophon,

as

may

thus

letter

names

it was

population.8
are found

period
Aeneas

the case

in

Tacticus

assume

and

that Greek

together
in later

or

with,
centuries

according to Dionysius of Halicarnassus (Dem. 52.2) and Quintilian (1.1.24).


Most of the letter names found in Plato and Callias are those with which we are
familiar.

The

vocalism

reflects the post-classical


El

in classical
The

found).10
is only

pu?ya

of modern

but

orthography;
replacement
the product

ksi,

etc.

pi

(for

iotacistic pronunciation
of
of

later

e?, ov,
late Greek

on,

such

spellings

v and
and

by

?ef,

rret

etc.)

of

course

simply

p] of original closed [?] (written as


e

Byzantine

as rrt with

i/jlXov,

simple
v

puiKp?v,
here

grammar;

/ are

also

ifjiX?v and
the names

i/jlXovand v ifjiXov'simple El Y helped to differentiate between the spellings E and AI


(both pronounced as [e]by that time), and Fand 01 (both pronounced as [?]),11 and
7

not least because


of this line is problematic,
The exact reconstruction
in deov
the allusion
is syntactically
awkward.
(cf. Plut. Mor.
384d-394c)
y?p er to the E of the sanctuary at Delphi
Pohlenz
?V?to,
(n. 5), 153, n. 2, suggests reading the first line as ?e
yap- aX<j>a, ?rjTa, y?ppa,
<t > eHnstead
(cf. also Ruijgh
(n. 5), 286-93).
8
we find
In Old and Middle
Comedy
?rjTa (Ar. Eccl. 684), Se?ra (Ar. Lys. 151), Qr?ra (Ar.
Eccl. 685), Kaima
fr. 394 K.-A.),
(Ar. Eccl. 686), Xa?Sa (Ar. Eccl. 920; Eupolis
pu (Ar. Thesm.
fr. 29 K.-A.),
in a fragment of the tragic poet Achaeus,
in which
the
781) and o?ypa (PI. Com.
name Aiovvo?is
described,
oVAtci, tcura, ov, vv, v, and o?v (Achae. fr. 33 Snell; on o?v cf. below
and ?VIII); cf. further Nicochares
fr. 5 K.-A.
avaX<f>a?rjTos.
9
Pind. Dith. 2, fr. 70b.3 Maehler
1.139 (o?v and oiypua: cf.
(o?v. cf. below and ? VIII); Hdt.
sense of '[river] delta'), 5.92 (Xa?Sa as a
below and ? VIII), 2.13.2 etc. (?Y?ra in the metaphorical
Ta?, x<^x0; Xen. Cyr. 7.1.5 and Oec. 19.9 (y?ppa), HG
name); Hp. VC 1 (p. 3.182 Littr?) (fa,
4.4.10 (o?ypa), Anab. 7.5.1 (Je'?Ta as a place name); Aen. Tact. 31.18 (?X(/>a, e?, l Ta, vv in the
of an encrypting
description
inscriptions:
procedure);
<xA<?a(IG 22.1425.95,22.1429.28),
?rjTa (IG
e? (IG 22.1496.223),
l ra (IG 22.1496.185),
22.1425.98,
22.2783.27),
p,v
y?ppua (IG 22.1496.187),
vv (IG 22.2783.21),
v (or ? according
Tau (IG 22.2783.20),
net (IG 22.2783.23),
to
(IG 22.2783.24),
J.Wackernagel,
review of K. Meisterhans,
Grammatik
der attischen Inschriften
[Berlin, 1885] and
M. Hecht,
Forschungen
Orthographisch-dialektische
auf Grund attischer Inschriften
[K?nigsberg,
16 [1886], 65-83, at 71, and LSJ, 1840, s.v. Y; but see also Einarson
Anzeiger
1885], Philologischer
and E. Schwyzer,
Grammatik
[n. 3], 263) (IG 22.2783.4),
(cf. K. Meisterhans
Xei (IG 22.1491.33)
der attischen Inschriften
[Berlin, 19003], 5-6 with n. 19).
10
Vox Graeca. A Guide to the Pronunciation
Cf. WS. Allen,
Greek (Cambridge,
of Classical
19873), 170.
11
Cf. LSJ, 2024, s.v. i/jiX?v, and LSJ, 1840, s.v. Y, where Theognost.
Can. 18, Schol. Ar. Plut.
896, and [Hdn.] Epim.
116,137 as well as Anon,
post Et. Gud. 679.9 and 678.55 and Chrysoloras
are cited as the earliest sources:
'but in Tr?oa X?^is ?no tt}? k? ovXXa?rjs apxop,?vr)
?t? tov ?
the name of the
i/jiXov yp?<f>eTai [...] ttXtjv tov Kai, ktX. Hdn. Epim. 62, ? ifj. is not yet merely
I. Allgemeiner
Flexion
Griechische
Teil, Lautlehre, Wortbildung,
Grammatik,
letter'; E. Schwyzer,
(Munich,
1939), 140, Allen
(n. 10), 172-3.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

404

LETTER

GRAECO-SEMITIC

NAMES

the names o paKp?v 'small 0' and c5 pueya 'bigQ' ensured that 0 and Q could be kept
apart after distinctions of vowel length had been lost inGreek.12
we

As

on

concentrate

the early
of minor

a series

to note

interest

of

history

the

letter

names,

it is however

obtained

of

greater
that

divergences

apparently
regional,
synchronie,
to the Byzantine
in the classical
age. According
grammarian
of
used
Democritus
the names
yepupua and
puco instead

the

Eustathius,

puv,
y?pupua and
v kcli
v
in
treatise
his
'About well
ypapLpLarajv
presumably
evc/y
Ilepl
8vu(f>ojvojv
= Eust.
in Horn.
//. 3.1,
and ill-sounding
fr. 68B19 D.-K.
letters'
(Demoer.
sounding
s.v.
source
is an
Phot,
The
that yepupia
cf.
mentions
p. 370.15;
p, ).
pi 654,
explicitly
and the same may
be inferred
Ionic
form of the letter name,
for puw from a Delian
atomist

inscription of the third century B.c. (IGXI/2 205 Ab 25; cf. further ? IX).13Moreover,
establish
Ionic
when

an

inflected

the

innovation

letter

names,

than

rather

idiosyncratic
not Attic14)
at best,
such
usage;
(but certainly
the name
SeXra had acquired
the metaphorical

become

thus

to have

is said

Democritus

a usage

could

in an

to

attempt

widespread
early
have been promoted

meaning
'(river)
the descriptive
letter name

or also when

lexicalized,

fully

presumably
a more
following

delta'

and

aiypia

had

'hissing'

had replaced an earlier name of S (cf. ?VIII). As for themore common indeclinability
of

letter

the

names,

speculates
to
(rj hid

8vaKardXr)7TTa
and perceptions
pedagogy
be sure that the real reason
(or

half-lexical

only

pre-existing
A slightly
odv

and

same

letter,

the

for

status)

kXivovtcUj

this

tells

rather
Hilgard)
to learn for

easy

?!va pur) Tro??? yivopueva


us more
about
ancient

certainly
than about
else, and we can
anything
language
names was
that their non-Greek
indeclinable
origin
entrance
to allow
into a
too
felt
them
strongly

was

declensional

pattern.
case of
complicated
claims
that
Herodotus

'which
Trdvra

call

the Dorians
is

cf. also Athen.

times

(p. 184.3-19
letter names

of children's

tojvto
11.467a).

ypapLpca,
Quite

itwould be more precise if Herodotus


archaic

ov

naiocuv
but

more

oiypua.

reXevrojGL
oiypba;

on Dionysius
Thrax
was
to make
the

scholion

that

its purpose
tojv
apTtpuades
avro?s
vTai);
yiv

amusingly
children

and

at least until

the fifth

synchronie
all Persian

variation

concerns

the

letter

names

names
end in one and the
personal
1.139:
and
odv
the Ionians
(Hdt.
aiypia
to
"I ves oe
odv
KaXeovoi,
piiv
Aojptees

apart

from

the factual

error

in this

statement,15

did not just speak of a different letter name. In


century

B.c.,

the phoneme

/s/ was

represented

12Cf.
Can. 13, and
LSJ, 2029, s.v. Q, and LSJ, 1193, s.v. 0, with Hdn. Epim. 208-9, Theognost.
and o piKp?v; Allen
of
829.29 and 1828.49 as the earliest attestations
869.26,
p?ya
(n. 10), 173.
13
in Isid.
confirmed
Einarson
by the lexeme motacismus
(n. 3), 19, n. 10, sees the variant p
is the likely
Ionic y?ppa
formation
after iotacismus.
1.32.6, but this may be an analogical
Etym.
'Herkunft und Alter der deutschen
starting point for (Etruscan?-)Latin
g?ice: see E. Hermann,
zu G?ttingen
der Wissenschaften
der Gesellschaft
Nachrichten
(1929), 215-32,
Buchstabennamen',
at 225.
14With Democritus'
fr.
o?XTaTos and dr?TaTos contrast Ar. Eccl. 684 tov ?rJT(a) and Eupolis
Ta Xd?Sa; only in the special case of o?ypa, PI. Com. fr. 29 K.-A. may have to be read
394 K.-A.
as Twv oiyp?TOJv Evpm?hov
(n. 11), 141, also
Schwyzer
(but tcov o?ypa t?jv E. is possible).
a nominal
stem foiriraT-),
but this is formed after
points to the derivative foir-n aT-ias (as if from
OTtyjuaT-tas".
15
and A. Panaino
to I. Gershevitch,
'The Old Persian
(edd.),
lisp', in G. Gnoli
According
Iranian Studies
of Iranian Studies, I. Old and Middle
Proceedings
of the First European Conference
who was
at 133, Herodotus
may have been misled
by an informant
1990), 115-33,
(Rome,
familiar with the Elamite graphic rendering of Old Persian nouns ending in -a (< *-as). I owe this
referee for this journal.
reference to the anonymous
Eust.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS

405

WILLI

by two different letter signs: on the one hand by the descendant of Semitic sad? (in the
Doric regions of Crete and Corinth), and on the other by the descendants of sin (in
The

Ionia).

fact

that neither

nor

oiypua

odv has

an exact

lexical

the

among

equivalent

Semitic letter names is an additional difficulty to be discussed below (?VIII). For now
it will

to note

suffice

representing
the relevant
letter
common
marked

Athenian
by the letter
come
do not

which
Achaeus

of Eretria,
an

other

in fact have

that odv may

/s/ that was


the phoneme
continued
sad?ov
formally

letter

on

dialect

the well-known

Chalcedon, which was quoted by Neoptolemus


machus

test.

attested

85A8

= Athen.

D.-K.

letter

Greek

until

brand
for a race horse
(lit. 'a?v-bearer')
designation
oapb<j>6pas
texts
odv16 as by the retention
of the name
odv in two literary
a odvlsad?
one
a
the
from
is
of
region:
fragment
tragedian
who was active
in fifth-century
Athens
fr. 33.4 Snell),
the
(Achae.
in Ionian

epigram

the only Greek


letter name
for any
no matter
if
the fifth century,
sin. This
not so much
is suggested
by the
been

in use

as

name,

it is used

sophist

of Parion

of

Thrasymachus

(third century) (Thrasy


is even

odv
10.454f).
Incidentally,
in a dithyrambic
fragment

the

earliest

of Pindar's

where

the poet alludes to how earlier dithyrambists avoided the sound /s/ (Pind. Dith. 2, fr.
70b.3 Maehler).

Like odv in oapi^opas, the name Kon-na (^?tttto) for the letter p (rendering /k/
before a dark vowel: hence Latin Q) is also attested indirectly in classical times:
another

type

of

race

standing for f?pivdos


from

into disuse

horse

was

called

after

Koinrar?as

the brand-mark

the sixth

century

onward,

educated people in the Hellenistic

its name

still

seems

to have

/w/, but

formally

familiar

been

to

age, for in an iambus by Parmenon of Byzantium

drunkard
is said proverbially
B.C.) an uncultivated
(third century
fr. 1.2 Powell:
the p?nTra
ovoi
kOtttto
(Parmenon
yivojoKojv).
name
is Fav
last-attested
Graeco-Semitic
letter
the
Finally,
(rendering

possibly

p,

this letter fell

'Corinth' (Ar.Nub. 23,438, Ar. fr. 43).17Although

the ancestor

of Latin

F).

'to know

for

letter was

This

no

even

not

the

letter

needed

longer

to write the classical Attic and Ionic dialects, which had lost the sound /w/, but it
nevertheless

appears

in early Attic-Ionic

given the value '6' in the Milesian

descriptive
already

after

the Roman
scholar Varro
(first century
= fr.
Goetz-Schoell:
71, pp. 208-9
gramm.
name
'double
alternative
Siyapupua
ydpupua
as well
have existed
1.4.8; Trypho
(cf. Quint.

etc.).18 Whereas
fr. 270 Funaioli

may

abecedaries

E,

and

it was

system of alphabetic numbering


B.C.)

refers

appropriately

(a

1, ?

VAV),19 by that
to the letter
(referring
Pass.

2,

to it as Fav

11; Apoll.

(Varr.
time the

Dysc.

shape)
Pron.

76.32 Schneider).

16
As

L.H. Jeffery, The Local Scripts of Archaic Greece. A Study of the Origin of the Greek
b.c. (Oxford,
to the Fifth Centuries
the Eighth
19902), 33, n. 1,
from
Alphabet and its Development
und antiquarischen
Inhalts, II
arch?ologischen
B?ttiger, Kleine Schriften
rightly notes after CA.
1838), 162, the oapcfr?pai were imported from the o?v/'sad? region Zlkvcov
(Dresden and Leipzig,
the fact that it was not a
thus have simply reflected
the name may
(B?ttiger: ZvpaKovoai)',
as a mark.
that
served
oiypalsin
17
one might
think
Thus again Jeffery (n. 16), 33, n. 1, after B?ttiger
(n. 16), 162; alternatively
in Alcm.
fr. 1.59 PMGF
?ttttoi KoXa?a?oi
of the race of the (Scythian?)
(poXa?a?oi) mentioned
horse of
name KoXa?a?s
'The Kolaxaian
in Hdt. 4.5.2; G. Devereux,
(cf. the Scythian personal
Alkman's
Partheneion\
CQ n.s. 15 (1965), 176-84).
18
Jeffery (n. 16), 25.
19
and cf. Dion. Hal. AR 1.20.3.
of the Septuagint,
See also ? III on ouau in the manuscripts

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

406

LETTER

GRAECO-SEMITIC

NAMES

III. THE EARLIEST ATTESTATIONS


LETTER NAMES
The

non-Hellenized

source written

Semitic

to render

impossible

versions

in Greek.

of

the

In translating

the acrostichic

structure

OF THE SEMITIC

are

names

letter

also

the Hebrew

Bible

of various,

mainly

first

attested

into Greek

of

passages

poetical,

by

it proved

the original: to let the first verse of a poem that was acrostichic inHebrew start with A
in the Greek

the

translation,

second

with

B,

and

so on, would

themore fundamental aim of translating literally. However,


formal

have meant

of

the original,
of the Septuagint
various
manuscripts
as 'subtitles'
within
(in a fairly consistent
transcription)
are
as
Ps.
118 (119)
1-4. The
and Lam.
chosen
transcriptions
*
between
added
when
with
brackets,
obviously
corrupt):20
play
names

letter

abandoning

in order to highlight the


add

the Hebrew
at

the acrostichs
follows

(variants

aX<f>(aXe<j>),
?rjd, yipbX (ycpuaX,yipueX, gimel), oeXd (Se?e#, oaXed, deleth, *delech), r?
(he,

*y]7T, *heth),

ovav,

?atv

ju/nfx, vow,
(Xapueo),
oev
Oav.
(xoev),

oapux

Further

of

atio

attestations

evang?lica,

alphabet
irrefutable
(Euseb.
meaning

(Cat),
(samedi,

this

series

the Church
on

depends

the Hebrew
that

argument
evang.
Praep.
in a few cases

Father

the

10.5).
(e.g.

letter

r?6, ttj6,
*sanch),

are found
Eusebius
one,
names

ta>0 (ta>8,
atv,

*loth),

<?>r?,oaorj

xa(l>

(caph,

*coph),

kom/), prjs

(rtaSr/),

Xa?8
(p?]Xs)i

era. In his Praepar


in the Christian
only
out that the Greek
of Caesarea
points
and
in support
of
this he advances
the
are meaningful
Eusebius

Interestingly,
<f>f? or? pua 'mouth',

in Hebrew,
does
-

give

pr?s

but not

in Greek

actual

original
in others

the

<j>aXr) 'head'),

but

he seems to be following a secondary (Jewish school?) tradition which tried to build a


meaningful
rendered

text
correctly

connected with

from
by

the

letter-name

the word

the root

for

series
'house'

(or at
(Gr.

least

gen.

parts

oikov),

'lp 'cow' (cf. ? IV), but with

of

it). Thus,
the preceding

while

?r)6 is
not
dX<j> is

the root Ip 'to learn' (Gr.

as '*crook,
not interpreted
and the following
throw stick'
yt^eA
(yrjpLeX) is
pbddrjois),21
v
as baXr
or 'camel'
as
with
84X6
but
'fulfilment';
(?),
together
paraphrased
TrXrjpojois
the sequence
oiXd
and r? (e, et) taken as a demonstrative
pronoun,
y?pueX
f?
aX<j> ?rj6

20

zur semitischen
see T. N?ldeke,
On details of the transcription
Beitr?ge
Sprachwissenschaft
'La date de la cr?ation de l'alphabet
grec et celle de
1904), 126-8, and C.J. Ruijgh,
(Strasburg,
at 577-9.
In the other biblical
Bibliotheca
54 (1997), 533-603,
Orientalis
l'?pop?e hom?rique',
1.2-8 [only alep
acrostichs
such 'subtitles' are not used (Ps. 9-10, 25, 34, 37, 111, 112, 145, Nah.
on these and similar texts see M. L?hr, 'Alphabetische
und
to kap], Prov. 31.10-31,
Sir. 51.13-30);
im Alten Testament',
Lieder
Wissenschaft
Zeitschrift
f?r die alttestamentliche
alphabetisierende
Psalmen
'Die alphabetische
Akrostichie
in der j?dischen
25 (1905),
173-98; P.A. Munch,
n.s.
15 (1936),
der Deutschen
703-10;
Morgenl?ndischen
Zeitschrift
Gesellschaft
dichtung',
R. Marcus,
in the Hellenistic
and Roman
periods', Journal of Near Eastern
'Alphabetic acrostichs
Journal of
'Isaiah 28 9-13 and the Ugaritic
Studies 6 (1947), 109-15; W.W. Hallo,
abecedaries',
at 328; A. Demsky,
'A Proto-Canaanite
11 (1958), 324-38,
Biblical Literature
dating
abecedary
Tel Aviv 4
for the history of the alphabet',
from the period of the Judges and its implications
texts: abecedaries',
in W.W. Hallo
and K.L.
'School
14-27, at 17, and A. Demsky,
(1977),
I. Canonical
the Biblical World
from
Compositions
of Scripture,
Younger
(edd.), The Context
1997), 362-5, at 364.
(Leiden, New York and Cologne,
21 Contrast
a 3321,
s.v. ?X<j>a- ?oos
'en effet,
the gloss Hsch.
Ke<f>aXr?. <&o?vLKes>;
"t?te de boeuf" ne peut se rapporter qu'? la forme de la lettre, puisque
l'appellatif
l'expression
comme
le savait Plutarque
"boeuf",
(Mor. 738a: ... to ?Xcfca irpoT?^ai
'alp- signifie simplement
8t? to <Po?vLKas ovTw KaXeiv tov ?ovvY (Ruijgh [n. 20], 539).

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS
Greek

yields

o?kov,

piddrjois

irXrjp

ois

407

WILLI

o?Xt?jv

avrr]

'the learning

of

the house,

this

is

the fulfilment of the tablets' vel sim.22


to

after Eusebius'
Only
names
the Semitic
of

parts
that

of

the

suggested
attested
in the Old
already
obliquely
to William
waw
'hook'
taw 'sign' would
and
not
Hallo,
to begin with,
but only names
for the corresponding
letters;
in the description
'hook'
of the Tabernacle
at Ex. 26.32
etc.,
letter

itself. According
nouns

waw

seems

to mean

taw means

and when

Greek

oeXra

names

are

in Job 31.35 and Ezek.


'sign, mark'
on the basis
of the meaningless
and O-Beine
for 'knock-knees'
and

secondary
developments
with German
X-Beine
also with

come
across
the first
finally
systematic
and spelling:
tradition
these appear
it has been
(fifth century
A.D.?).23 However,

Talmud

known

common

been

when

references
in various

time

the Palestinian

two

Testament
have

do we

in the Hebrew

9.4

and

letter

'bandy
for this assumption
in the
Ie saw qaw
neither
leqaw
nor
God's
the
conveying
plan'

for a river delta). Hallo


finds
support
Jes. 28.9-10
in his view,
saw
where,

passage
'an impenetrable
utterance
or
concealing
and cries of a party
'shouts
of drunkards',
but
rather
two
names
alternative
for the letters
sad? and qdp
'in the context

mysterious

represents

sarcastically

be
9.6, these would
names
(comparable
or
legs' respectively,

as being

pictured

administered

by

(equally
of a

meaningless)
spelling-lesson

the prophet'.24

IV.THE ORIGIN AND DATE OF THE SEMITIC LETTER NAMES


In view

of

these

several

meaningless
we may
Palestine,
obvious

terminus

alphabet

to Greece,

and
in particular
if Hallo
attestations,
letter names
in the first half of
existed
ask when
ante
no

the meaningful
is indicated
quern
later

than

around

Semitic

letter

is right
in assuming
the first millennium
names

of
by the transfer
800 B.c.25 It is much

that
B.c.

in

came

into being. An
the Northwest
Semitic
more

difficult

to say

22
And further ovav
'the living one lives in
ning'; xd(f> (xd?) 'opajs,
from it' + vovv 'aiWta,

+
(*oua) 'eV avTrj, in it' + ?at(*?ar?)
%f?, lives'
f?Q '? ? v, the living' -?
+ ??>6
-> 'a beautiful
it'; Tr?9 '/caA-rj, beautiful'
begin
'?pxrj, beginning'
+
nonetheless'
Xa?S 'p?de, learn' ?> 'learn nonetheless';
pr?p 'ef avTwv,
+
eternal'
oapx
'?orjdeLa, help' -> 'from it [comes] eternal help'; d?v (??v)
+
(or eye)' + </>ij 'oTopa, mouth'
'77-7777)77 ?cf)9aXp?s, fountain
justice' ?? 'a
o?8rj 'hiKaioovvr),
source (or: an eye) and a mouth
+ o?v
of justice'; kwcJ) 'kXtjols, calling' + pr?s
'Ke^aXrj, head'
(o?v) '?d?vTes, teeth' + dav 'or/peta, signs' -> 'the calling of the head and the signs of the teeth'.
23
G. Dalman,
Grammatik
des j?disch-pal?stinischen
Aram?isch
nach den Idiomen
des
und Prophetentargum
und der jerusalemischen
Talmud, des Onkelostargum
pal?stinischen
Targume
the following first attestations
in the Palestinian
Talmud
(Leipzig,
19052), 52, mentions
(cf. also
Einarson
(n. 3), 22, n. 41): 'alep (Sabb. 9b, Sanh.
18a), bet (Meg. 71d), gimel (Shek. 47b), dalet
(Maas. sh. 55b), he (Sabb. 9b, Pea 20b), waw (Meg. 7Ie, Sanh. 25b), zayin (Sabb. XII 5), h?t (Pea
sh. 55b), t?t (Maas. sh. 55b), y?d (Meg. 71d), kap (Meg. 71d), lamed (Sabb. 9b), m?m
20b, Maas.
(Meg. 71e), n?n (Meg. 71d), samek (Meg. 71e), 'ayin (Meg. l\c),pe
(Meg. 71d), sade~(Meg.
71d),
q?p (Maas. sh. IV 11), res (Maas. sh. 55b), sin (Meg. 71d), taw (Sanh. 18a). I owe this reference to
I.Willi-Plein
(Hamburg).
24
W.W. Hallo, Origins. The Ancient Near Eastern Background
Western
Insti
of Some Modern
tutions (Leiden, New York and Cologne,
1996), 38 (cf. Hallo
[n. 20], 336-8), who
rejects the
and objections
of A.F. Key,
'The magical
of Isaiah 6 9-13', Journal
interpretations
background
86 (1967), 198-204, at 203, and F.M. Cross and TO. Lambdin,
'AUgaritic
of Biblical Literature
and the origins of the Proto-Canaanite
BASOR
160 (1960), 21-6, at 24,
abecedary
alphabet',
n. 21; for the interpretation
as a school scene see also Driver
of Jes. 28.9-10
and
(n. 1), 89-90,168
242-3.
25
The exact date of the transfer is of course fiercely debated,
and a substantially
earlier date is
on
'Some Semitic epigraphical
considerations
among Semitists
preferred especially
(cf. J.Naveh,
the antiquity
of the Greek
American
Journal of Archaeology
11 [1973], 1-8; Naveh
alphabet',
'Semitic epigraphy and the antiquity
of the Greek alphabet', Kadmos
30
[n. 1], 175-86; J.Naveh,

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

408

LETTER

GRAECO-SEMITIC

NAMES

how many centuries one should go back from this date. Ultimately, the question
related to themuch more fundamental problem of the creation of the alphabet.
A

number

substantial

of

the Semitic

letter

names

to concrete

refers

things.

is

One

may cite here 'alp?alep) 'cow' for /'/, b?t 'house' for /b/, dalt (dalet) 'door' for Jdl,yod
'hand'

for Ikl, m?m


'water' for /m/,
?yl, kap
'palm (of the hand)'
'ayin 'eye' for /'/,
or less
for It/, sin 'tooth'
for ?II.26 In these cases,
the letter shapes
show more
an
one
can
with
similarities
In
the
referents.
recognizable
respective
early
'alp
easily
see the head
com
in an early kap a hand with
It is therefore
of a cow, and
fingers.
r?s

for

'head'

monly
The

that

assumed
'inventor'

of

the alphabet

the

was
would

alphabet

conceived

from

have

of words

acrophonically
a number
collected

the beginning.
all
which

started with a different phoneme, drawn a stylized and simplified picture of their
and

referents,

used

these

drawings

as

letters

for

the respective

initial

phonemes.27

This is possible, but not definitely provable. In theory at least, it could also be that
each

letter

shape

was

at first

correlated

more

or

less

arbitrarily

with

the

phoneme

designated by it and that itwas only at a later stage that someone thought of a word
which both started with that phoneme and had a referent (ideally) resembling the
letter

shape.28

In other

words,

people

would

first

have

had

the

shape

of

the

letter

'alp,

in Old Canaanite
and early Phoenician
[1991], 143-52; EM. Cross,
'Newly found inscriptions
238 [1980], 1-20, at 17; B. Isserlin,
'The antiquity
of the Greek
scripts', BASOR
alphabet',
22 [1983], 151-63, but also B.L. Ullman,
'How old is the Greek
Kadmos
American
alphabet?',
aux origines
Journal
38 [1934],
C.J. Ruijgh,
'D'Hom?re
359-81;
proto
of Archaeology
avec un
de la tradition
du langage hom?rique,
?pique. Analyse
myc?niennes
dialectologique
sur la cr?ation
excursus
in J.P. Crielaard
de l'alphabet
grec',
[ed.], Homeric
Questions
note however
and Ruijgh
the
1995], 1-96, at 26-47,
[Amsterdam,
[n. 20], 535-6 and 549-54);
on the early history of the
reservations
'The pitfalls of typology:
expressed by S.A. Kaufman,
Union College Annual
57 (1986),
of arguments
1-14, about the (in)validity
alphabet', Hebrew
a date just after 800 B.c. is certainly
the
based on the typology of various letter shapes. Nowadays
latest possibility
'The antiquity
of the Greek alphabet', American Journal
(whereas R. Carpenter,
in G. Pfohl,
720-700
37 (1933), 8-29, had still suggested
of Archaeology
B.c.); cf. the overviews
der griechischen
in G. Pfohl (ed.), Das Alphabet: Entstehung
und Entwicklung
Schrift
'Einleitung',
and H.-G. Buchholz
and A. Heubeck,
'Schrift', in F. Matz
(Darmstadt,
1968), ix-xl, at xv-xvii,
as well as P.K.
III. Kapitel X (G?ttingen,
Hom?rica,
1979), X75-X78,
(edd.), Archaeologia
The Antiquity
and the Early Phoenician
McCarter,
(Missoula,
Scripts
of the Greek Alphabet
linear alphabet and its
'The Canaanite
Mont.,
(n. 1), 176; A.R. Millard,
1975), 103-26; Driver
to the Greeks',
Kadmos
15 (1976),
greca
130-44, at 141-2; M. Guarducci,
passage
Lepigraf?a
des
'Zur
dalle
al tardo impero (Rome,
19-20; R. W?chter,
origini
1987),
Vorgeschichte
des
28 (1989),
'Die ?bernahme
Kadmos
19-78, at 69-76; R. W?chter,
griechischen
Alphabets',
der
durch die Griechen:
wie, wann, wo, durch wen und wozu? Eine aktuelle Abw?gung
Alphabets
in Nikolaos
Dimoudis
and Apost?los
und methodischen
Ans?tze',
Standpunkte,
Argumente
3.-6.
und Schrift
der hellenischen
(Tagung Ohlstadt
(edd.), Die Geschichte
Sprache
Kyriatsoulis
at 351-2;
and 426-7;
Oktober
345-58,
(n. 16), 12-21
1996)
(Altenburg,
1998),
Jeffery
sur l'introduction
'The "shadow
line". R?flexions
de l'alphabet en Gr?ce',
M.G. Amadasi Guzzo,
Lire et ?crire en M?di
in C. Baurain,
and V Krings
C. Bonnet
grammata.
(edd.), Phoinikeia
at 296-308;
h?:
'Tsad?im?
Sass (1991) (n. 1), 94-8; S.R. Slings,
terran?e (Namur,
1991), 293-311,
ser. 4, 51 (1998), 641-57,
at
two problems
in the early history of the Greek alphabet', Mnemosyne
656, and R?llig
(1998) (n. 1), 371.
26
On waw 'hook' and taw 'sign' cf. ? II above.
27
See e.g. Gardiner
(n. 1), 152-3 and 157-61;
(n. 1), 5-11; Albright
(1948) (n. 1), 7; Driver
et al. (edd.), Reading
the Past. Ancient Writing
J.F. Healey,
in J.T. Hooker
'The early alphabet',
to the Alphabet
(n. 20), 537?40.
(London,
1990), 197-258, at 211-12 and Ruijgh
from Cuneiform
28 For this or similar scenarios
see - after F. Lenormant,
de l'alphabet
Essai sur la propagation
I (Paris,
dans l'ancien monde,
and, more
already W
ambiguously,
18752), 94-7,
ph?nicien
Geschichte der hebr?ischen Sprache und Schrift. Eine philologisch-historische
Einleitung
Gesenius,
in die Sprachlehren
der hebr?ischen Sprache (Leipzig,
und W?rterb?cher
1815), 167 e.g. H. Bauer,
et
Der Ursprung des Alphabets
Byblia Grammata. Documents
(Leipzig,
1937), 17-23; M. Dunand,

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS

then assembled a list of words starting with


as

their

say, a father
At

name

letter

first

because

the

of

shape

409

WILLI

'alp, and finally decided in favour of 'alp


the

sign

more

looked

like a cow

than

like,

('ab).
this

sight,

second

looks

hypothesis

unnecessarily

complicated.

However,

it

may find some limited support in a number of letter names which either (a) do not
mean anything (thus, he' for Ihl, t?t for Itl) or (b) do mean something, but nothing that
could

does not particularly


shape: a qdp, for instance,
a b?t, whose
is not usually
character
pictographic
a surprisingly
if we posit
abstract
of a
only
drawing

in the
be easily
recognized
a 'monkey',
and even

letter

resemble

questioned,29
house
front

looks

like a 'house'

or a ground

to typological

parallels

plan.30
in other

Moreover,

the

traditions

where

secondary-name
letter names

hypothesis
are artificially

can point
in
created

accordance with the acrophonic principle. Thus, in the Old Irish Ogam alphabet the
were

trees (e.g. beith


lexemes
for <b>,
luis
after
'birch-tree'
designating
new
for
and
the
of
the
Slavonic
'alder-tree'
for
letters
(?)
<\>,fern
<i>31),
were
to by the words
text which
referred
of an acrostichic
have been
may
alphabets
at the same time as, or only
the
created
the Cyrillic
itself: here,
after,
shortly
alphabet
letters

named

'rowan-tree'

letter for /a/, for example, was arbitrarily called az? T,


'(I) know'
history

etc.32 That
as well

similar

is shown

which arementioned

not

revisions
only

by

could
the new

have

taken

meanings

the following letter for NI v?d?


place
for

by Eusebius (cf. ? III), but also, more

in the Semitic
the

conventional

alphabet
names

importantly, by the South

de l'?criture en Ph?nicie
recherches sur le d?veloppement
1945), 163-71; Hallo
(n. 20),
(Beirut,
335-8; I.J. Gelb, A Study of Writing
(n. 1),
(Chicago and London,
19632), 140-3, and Diringer
168-9.
29
Cf. e.g. Healey
(n. 1), 152-3 and 163;
(n. 27), 212, and Ruijgh
(n. 20), 542, but see also Driver
a
for t?t A.G. Lundin,
'O proischozdenii
alfavita', VDI2/160
17-28, at 25, hypothesizes
(1982),
'ear (of a needle)'.
meaning
30Additional
cases which are either formally or semantically
include garni (gimet)
problematic
to Driver
'camel' (thus Lundin
(n. 1), 155, 163-4 and 262,
(n. 29), 25, but 'throw stick' according
name *z?t 'olive tree' cf.
with reference to Akkadian
gamlu), zayin '?' (on an alleged alternative
?VIII), h?t 'fence, barrier (?)' (Dunand
[n. 28], 166), lamd (lamed) 'thorn (?)', nun 'fish' (cf. below),
'mouth' (but see Driver
samk (samek) 'fish (?), support W,pe
[n.
[n. 1], 153), and sad?(Dunand
se rattache sans doute ? la racine swd et pourrait
d?signer un engin de chasse
28], 168, 'Le mot
in M. Lidz
'Die Namen
der Alphabetbuchstaben',
o[u] de p?che'; differently M. Lidzbarski,
2 (1903-1907)
barski, Ephemeris f?r semitische Epigraphik,
1908], 125-39, at 126-7,
[Giessen,
and Driver
126-38; Dunand
(n. 28), 164-9; Diringer
(above),
[n. 1], 263); cf. further Lidzbarski
(n. 1), 161-71 and 262-6, and Sass (1988) (n. 1), 108-33.
(n. 1), 168-9; Driver
31
24 (1949), 19?43; D. McManus,
'Irish
See H. Meroney,
'Early Irish letter-names',
Speculum
and E. Seebold,
39 (1988),
letter-names
and their kennings',
?riu
127-68,
'Fujmrk, Beith
bei
die Systematik
der Zeichenaufz?hlung
und Alphabet.
?ber
Luis-Nion,
He-Lamedh,
Abgad
in F. Heidermanns,
H. Rix and E. Seebold
Buchstaben-Schriften',
(edd.), Sprachen und Schriften
zum 65. Geburtstag
des antiken Mittelmeerraums.
(Innsbruck,
Festschrift f?r J?rgen Untermann
a parallelism',
in
'Runic and Ogam
letter-names:
at 427-8; D. McManus,
1993), 411-44,
Celtic Studies
D. ? Corr?in, L. Breatnach
and K. McCone
(edd.), Sages, Saints and Storytellers.
stresses that the acrophonic
James Carney
inHonour
1988), 144-8,
(Maynooth,
of Professor
a new value
even when
sound change
was retained
the sign acquired
(cf.
through
principle
at 211, on
'Studies in the Ethiopie
syllabary', Africa 21 (1951), 207-17,
similarly E. Ullendorff,
<
also Greek
compare
y aman for y?d 'as 'od would have been useless';
Ethiopian
*r?Ta (/h?ta/
was
IhJ
was
East
Ionia
where
from
for
/h/
and
first
used
which
lost,
secondarily,
starting
h?t),
?or /<?/).
32
I
'Azbucna molitva',
in P. Dinekov
See K. Kuev,
Enciklopedija,
(ed.), Kirilo-Metodievska
Die
altkirchenslavische
and H. Birnbaum,
J. Schaeken
Schriftkultur.
1985), 50^;
(Sofia,
und Flexions
Glossar
Laute und Schriftzeichen,
Geschichte,
(mit Textproben,
Sprachdenkm?ler
runes and the Gothic
letters follow a
mustern)
1999), 76; the names for the Germanic
(Munich,
The runes and the origin of Wulfila's
'Runo-Gothica.
similar principle
script', Die
(cf. B. Mees,
55-79, at 56-63, with earlier literature).
Sprache 43 (2002/3),

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

410

LETTER

GRAECO-SEMITIC

NAMES

Semitic Ethiopian alphabet, an old side branch of theNorthwest Semitic alphabet. In


the former, the letter for /n/ is called nahas 'snake' instead of n?n 'fish', and while it
to speculate
which
of
be a secondary
creation.33

be useless

may

them must

V. FURTHER

the

two

names

it is clear

is older,

EARLY EVIDENCE: UGARIT


TZBET SARTAH

one

that

of

AND

a priori,
cannot
therefore
be dismissed
the
secondary-name
hypothesis
we
ante
the
terminus
far
back
the
for
remains
how
may
quern
safely push
question
not certain,
that we can
letter names.
It is possible,
of the Semitic
existence
though
at ancient
next
to
have yielded,
The excavations
reach the fourteenth
Ugarit
century.34

As

as the

long

other

many

cuneiform

texts

documents,

in a

written

cuneiform

locally-developed

alphabet. The Ugaritic alphabet represents letter values which correspond closely to
those of the Semitic linear alphabet, but its letters are designed with the typical
cuneiform
school

of

sequence
writer

has

one

and

cuneiform

Thus,

syllabary.

the Ugaritic
contains
letters,
cuneiform

next

to the

in the cuneiform

documents

a peculiar
abecedary
two thirds of which
about

them

of

alphabet
to each

next

added

cuneiform

of

Several

wedges.
exercises,

letter

letter
for

a cuneiform
/b/

the

alphabet

(KTU
have
sign

syllabogram

survived,

from

the

<be>

are
in this

5.14):

the
usual

is found,

next to the letter for Igl the syllabogram <ga> etc.35 It has been argued by Frank Cross
and

Lambdin

Thomas

authors

suggest

that

that
the vowels

these

are
correspondences
in the syllabograms

not

expressed

are

haphazard.
the same

The

two

vowels

that

occurred in the (first syllable of the) Semitic letter names: hence, <be> for /b/ because
the

letter

Overall,

name
these

was
vowel

b?t,

<ga>

for

correspondences

the letter name was garni, and


Igl because
are indeed
to make
enough
systematic

so on.36
coinci

dence unlikely, even if not every detail fits in exactly (one would not, for instance,
33 The wave-like
shape seems to favour nahas 'snake' (Driver [n. 1], 154 and 165; cf. however
letter names, only in 1548
like the other Ethiopian
Ruijgh
[n. 20], 541), but this name is attested,
of the New Testament
in a translation
suspects that
(n. 31), 211-14,
printed in Rome; Ullendorff
name series was invented in the sixteenth century by European missionaries
the entire Ethiopian
or scholars (cf. Sass [1991] [n. 1], 92, but differently N?ldeke
[n. 29],
[n. 20], 131-3, and Lundin
21). One may also ask why a name nahas 'snake', which fits the shape rather nicely, should have
nun 'fish'. On further name
Semitic tradition by a less plausible
been replaced in the Northwest
see Demsky
literature); Ruijgh
(n. 20),
(1997) (n. 20), 364 (hehin and pepin in Rabbinic
changes
542 (dolt renamed from dag 'fish'; cf. Cross and Lambdin
[n. 24], 25, Sass [1988] [n. 1], 113-14),
in Lidzbarski
and ?VIII on *tann -? sin; the excessive reliance on such changes
(n. 30), 126-38, is
(n. 1), 7-8.
already criticized by Gardiner
34
letter names
Cf. also Cross and Lambdin
(n. 1), 23*: 'The Ethiopie
(n. 24), 22 and Cross
and in turn these go back to the time
were taken over with the alphabet
from Old South Arabic,
toward
when
the Proto-Arabic
1,300 B.c.'
apart from the Proto-Canaanite
script branched
of the second millennium
B.c.); but see
(similarly Lundin
[n. 29], 21: not later than in the middle
letter names.
the Ethiopian
footnote on the uncertainties
the preceding
surrounding
35 The
series is a-a, b-be, g-ga, h-ha, d-di, h-ii, w~wa, z-zi, h-ku, t-ti, [...], [p-p]u,
complete
The
O. Loretz and J. Sanmart?n,
s-sa, q-qu, r-ra, t-sa, g-ha,
t-tu, i-i, u-u, s-zu (M. Dietrich,
Texts from Ugarit, Ras Ibn Hani and Other Places
(KTU: second, enlarged
Cuneiform Alphabetic
one may compare
the much
later transformation
edition)
1995], 493-4). Typologically
[M?nster,
to the
in the Old Hispanic
of alphabetic
letters into syllabograms
according
scripts where,
and
/ka/ (cf. garni, y?ppa),
<b> became
/be/ (cf. b?t, ?rjTa), <g> became
'alphabet' of Espanca,
und eine neue
'Neue ?berlegungen
<t> became
/ta/ (cf. taw, Ta?): on this see J. Untermann,
38 (1997), 49-66, at
der althispanischen
Schriften', Madrider Mitteilungen
Quelle zur Entstehung
58.
36

Cross

and Lambdin

(n. 24), 23-6;

cf. further Cross

(n. 1), 23*-24*,

and Driver

(n. 1), 264-6.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS

411

WILLI

expect the syllabogram <?> next to the letter for /h/ named he31). Thus, the tablet
does

to provide
the first evidence
it is still indirect
evidence,

appear

for

since

However,

the existence

one must

of

the Semitic
to hope

continue

letter
for

names.

additional

direct attestations to be discovered in epigraphic material from the second half of the
second

context

may
has

In 1976

in fact

be one

been

noticed

not

the earliest

in this
whose
relevance
potential
piece of evidence
so far, although
the item as such is not a new discovery.
was excavated
in Tzbet
in
non-cuneiform
Sartah
abecedary

known

on a postcard-size
which
of writing
exercises,

lines

followed by a fifth
written

first.

The

demonstrated
(and

small

It is written

Israel.
Four

B.c.

millennium

There

instead

line containing

writer
the

by

some

obviously
that he did

dated

do not

to the early twelfth


a meaningful

seem

a student.39
not

approximative

know

B.c.38

century

to have

the alphabet

was
fact

just drew

ostracon

are

content,

sequence, which may have been


His

limited
to write

how

that he

pseudo-signs),

is
competence
waw or zayin
left an empty
space

writing
a proper

where he should have placed m?m for /m/, no doubt because he had forgotten what the
shape

looked

The most
letter,

'alep,

like, and

that he

inverted

the

sequence

of

the

letters

zayin

and

h?t.40

striking thing, however, is the beginning of the abecedary line. The first
is clearly

recognizable.

The

following

letters

should

be

b?t

and

gimel

(garni). However, the editor Kochavi notes that the sign for b?t is problematic: 'it is
difficult to determine when the writer intended b?t and when lamed'.41 In fact, there is
hardly any difference at all between the open spirals of b?t and lamed later in the line.
in other

Moreover,
spirals

are usual

for

Proto-Canaanite
lamed,

but

clearly

and

early

distinguishable

Phoenician
from

inscriptions
b?t. In other

such
words,

open
if the

Tzbet Sartah letter did not occur in the second slot of the alphabet, one would read it
as

lamed.42

37 Cf. Hallo
(n. 24), 38, who also points to <pu> next to the letter for /p/ (for which Cross and
Lambdin
Semitic *puw 'mouth');
[n. 24], 25, and Driver
[n. 1], 153, 162, 261-2 and 264, postulate
<u> for /h/ is explained
'a form *hi?/*h? > h?' in Cross and Lambdin
(n. 24), 25-6, and
through
to a mix-up
of two syllabograms.
<ku> for /hi is ascribed
38
'An ostracon
of the period of the Judges from
First published
and dated by M. Kochavi,
'Izbet Sartah', Tel Aviv 4 (1977),
1-13; cf. also Demsky
(n. 25), 8-15; and
(1977) (n. 20); Cross
ten years later', in I. Finkelstein
with a summary A. Demsky,
'The 'Izbet Sartah ostracon
(ed.),
An Early
186-197.
Iron Age Site near Rosh Ha'ayin,
Israel
Tzbet Sartah.
1986),
(Oxford,
at 289, suggests
17 (1978), 287-95,
'Sull'alfabetario
di Tzbet Sartah', Oriens Antiquus
G. Garbini,
a later (eleventh-century)
date.
39
on the
'Some considerations
See Kochavi
(1977) (n. 20), 19-20; J.Naveh,
(n. 38), 6; Demsky
ostracon
Journal 28 (1978), 31-5, at 31-3; A. Lemaire,
from Tzbet Sartah', Israel Exploration
en ?pigraphie
266
et exercices
Journal Asiatique
d'?colier
nord-ouest
'Ab?c?daires
s?mitique',
at 222-25; Cross
'New light on the Tzbet Sartah
(n. 25), 8-9; and A. Dotan,
(1978), 221-35,
'short sentences or phrases containing
Tel Aviv 8 (1981), 160-72, who suspects
proper
ostracon',
names and names of objects
and food)' in the first four lines (but cf. Demsky
[n. 38],
(clothing
line as an
of the fifth
about
the interpretation
of Garbini
(n. 38), 291-2,
192). The doubts
are unjustified.
abecedary
40
Cf. in detail Kochavi
sequence Demsky
(1977) (n. 20),
(n. 38), 9 and 10, and on the h?t-zayin
to Demsky,
need not be a mistake
the equally
sequence
17-18; according
surprising pe-ayin
because of several parallels.
41
Kochavi
(n. 38), 10, on lamed as well as Cross (n. 25), 9: 'All our
(n. 38), 8; cf. also Kochavi
should have an angular,
that the b?t of the 12th century
other evidence
large, pointed
suggests
or slightly
Beth-shemesh
curved
and a short angular
Ostracon,
Bowl,
head,
leg (Lachish
'El-Khadr Arrowheads).'
42
is read as a failed dalet by Kochavi
Letter no. 11 of the second line, which
(n. 38), 5, looks
much more
like a normal b?t.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

412

GRAECO-SEMITIC

LETTER

NAMES

Similarly, the third letter poses a problem. 'The gimel of our ostracon is also
difficult to distinguish from another letter, namely pe. [...] The gimel of Tzbet Sartah
with its erect shaft and the small angle between the shaft and arm resembles more the
the

of

gimel

10th

onwards

century

than
material

early comparative
Unfortunately,
more
like a p?' on the somewhat
has to be ap?'.44
15, which

the gimel
for p?'

later Gezer

of

the

is rare,

calendar

12th-11th

centuries

but

no.

(tenth

letter
century)

B.C.E.'43
even

3 looks
than

no.

letter

Thus, for an unprejudiced reader the first three letters of the Tzbet Sartah ostracon
do not

read

sequence

as

. Is it
but as 'alep-lamed-pe
just a coincidence
to the name
of the first
letter (7/?)? It seems

'alep-b?t-gimel,

corresponds

exactly

that
at

this
as

least

likely that the letter series was dictated to an inexperienced pupil who first thought he
had

to write

the

down

letter

names,

but when

he noticed

his mistake,

the

teacher

and

his fellow pupils had already reached dalet and he had no time left to correct his initial
error.

Such

a scenario

not

would

take

only

into

account

that

'Zeichennamen

dienten

im Schulbetrieb, genau wie die Buchstabennamen


der
prim?r der Verst?ndigung
modernen Alphabete',45 but also explain why the student at Tzbet Sartah left an
empty space for m?m; no doubt he intended to fill this space later on, but did not have
time to ask or look up themissing shape immediately. In this way, the Tzbet Sartah
ostracon

may

second-millennium

provide
date

a welcome
of

confirmation

at least

some

VI. THE GREEK ADAPTATION


to Greece,

Returning
the Semitic

letter

we

names.

shall

action

apparently

were

taken.

of

the Semitic

the

Ugaritic
letter names.

evidence

for

OF THE SEMITIC LETTER NAMES

next

consider
how
the Greeks
and transformed
adapted
was
so
the acrophonic
it was out of
useful,
principle
names:
the Semitic
the letter for /b/, for instance,
not
could

Because

the question
to translate
or oikos
be called
simply
Scufxa
case of a letter whose
consonantal

Phoenician

of

The

in analogy
with
value was not
laryngeal

b?t
needed

consonant

in the particular
Only
was a similar
to render Greek

'house'.46

/7, which

was

represented

by

*'?h (Hebrew 'ayin), did not exist inGreek. We know that certain letters

to create new vowel-signs


'recycled'
was used quite
straightforwardly,
(aX(f>a)

for

the Greek

in accordance

thus, Phoenician
alphabet:
'alp
with
the acrophonic
principle,

for the vowel /a/ instead of the glottal stop /'/ (the latter having no phonemic status in
Greek), he (et) was used for lei instead of /h/ (because Greek /h/ was rendered by the
more strongly articulated h?t (f?Tal*r?Ta)47),yod (l ra) was used for III instead of lyl
43
Kochavi
(n. 38), 8.
44
Cf. the table in M. O'Connor,
Semitic
in P.T. Daniels
and W. Bright
scripts',
'Epigraphic
(edd.), The World's Writing Systems
(New York and Oxford,
1996), 88-107, at 91.
45
Y Gong, Die Namen der Keilschriftzeichen
1.
(M?nster,
2000),
46
Note
that such a procedure
is found in the Ethiopian
where p?' 'mouth' was
alphabet,
replaced by 'af 'without
regard to the loss of the otherwise
consistently
applied
acrophony'
(Ullendorff
[n. 31], 211).
47
Neither
the argument
therefore have been
by Einarson
(n. 3), 6, that the alphabet must
in the psilotic dialect of East Ionia, nor the one by Ruijgh
adopted
(n. 25), 29-30, Ruijgh
(n. 20),
535 and 567-8, and C.J. Ruijgh,
ser. 4,
'Sur la date de la cr?ation de l'alphabet grec', Mnemosyne
51 (1998), 658-87,
at 661-3,
that the use of h?t proves a second-millennium
date for the transfer
of the alphabet,
is watertight;
the latter is refuted by Slings (n. 25), 652-4, and the former would
not exclude the psilotic regions of Crete either, even if itwere reliable (on psilosis
in Central Crete
cf. A. Thumb
and E. Kieckers, Handbuch
der griechischen Dialekte,
/[Heidelberg,
19322], 155-6,
and Jeffery [n. 16], 28). On *7yTa as the original
form of the name see W
review of
Schulze,
P. Kretschmer,
Die griechischen
ihrer Sprache nach untersucht
Vaseninschriften
(G?tersloh,
1894),
at 256.
G?ttingische
gelehrte Anzeigen
(1896), 228-56,

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS

413

WILLI

(which no longer existed as a simple consonant inGreek), and a graphic variant of


waw (Fav, and then v) was used for ihl (later dialectal /u/).48As an additional (and
before

the

creation

secondary
'en was

ideal

laryngeal-sign
the Greek
ducing

as

of Q,

it was

because

the

vowel
for
sign,
only as superfluous
because
the Greek

last)

not

loi,
as

the

however,

'alp in repro
of
'en
translation

consonant
but also
system,
or
started with
the remaining
vowel
/o/; here too,
oc/)daX[x?s
?'f?/xa, appropriately
seems
to have operated
at least in the background.49
then, the acrophonic
principle
More
were
names
the Phoenician
Hellen
however,
commonly,
only
superficially
some
ized:
etc.
became
b?t
In
this
than
'alp
respects
(rather
?rjra,
transcription
aA^a,
'eye',

translation) raises minor


discussed in detail:
In letter

(1)

issues which may be mentioned

in passing, but need not be

a prop vowel
-a was
in a final consonant
ending
-tt etc.) was
consonant
in question
not
(-(/), -t,
<
<
<
in Greek
b?t, Karma
'alp, ?rjra
kap, poWa
(aA0a

names

the word-final
position

added

whenever
in this

admitted
<

q?p,

etc.).50

(2) Complex consonant groups were simplified through assimilations in y??x^ia <
garni and Xd?Sa < lamd (where Xafx?Sa is a secondary and late-attested phonetic
development).51
-v and
(3) Word-final
a motivating
without
sequence

alphabet
*
r?s s an, divided
The

(4)

Ionic

fluctuation

in vv <

(for -s) were


dropped
or as a sandhi
factor

>

(*n?n-samk
into w-san).52
an
between

yefjLfjia for ydfjifxa

<

e-vowel

garni

result
divided

*ni?ssamk,
and

(cf. ? II)

of

an

a-vowel

suggests

nun

and

<

their

pronunciation

into

n?-samk;

in ?VAra

a pronunciation

ros,

either
in the

*ws-san

< dalt
as

and

>

also

[ae] or as a

48

Cf. Ruijgh
(n. 25), 30-1, and Ruijgh
(n. 20), 569-73, and on vlF (which are still undifferen
see A. Heubeck,
in theW?rzburg
'Die W?rzburger
WJA n.s. 12
abecedary:
Alphabettafel',
'De la
(1989) (n. 25), 36-8, Jeffery (n. 16), 24-5 and 35, and C. Brixhe,
[1986], 7-20) eg. W?chter
au grec', in
? l'?criture: quelques
de l'alphabet
canan?en
aspects de l'adaptation
phonologie
C. Baurain, C. Bonnet and V. Krings
Lire et ?crire en M?diterran?e
grammata:
(edd.), Phoinikeia
(Namur,
1991), 313-56, at 345-50.
49
'The origin of the letter omicron',
Thus G.L. Cohen,
Kadmos
21 (1982),
122-4; Ruijgh
tiated

(n. 25), 31;Ruijgh (n. 20), 569;Ruijgh (n. 47), 665-6; R?llig (1995) (n. 1),202-3; andR?llig (1998)
(n. 1), 372. One might
(n. 1), 155 and 179, after Bauer (n. 28), 41, ' that loi is
object with Driver
a
at Ugarit
with
the 'ayin equivalent
written
and that 'the Semitic
showed
occasionally
for the o-sound'
'Zum semitisch-griechischen
preference
(cf. F. Praetorius,
Alphabet',
Zeitschrift
at 284; Gardiner
der Deutschen
62 [1908], 283-8,
Morgenl?ndischen
Gesellschaft
[n. 1], 11;
-a. Mit
E. Schwyzer,
'Griechische
und griechische
Buchstabennamen
auf
Interjektionen
Exkursen
?ber die Geschichte
der Buchstabennamen
und des Wortes
Alphabet',
Zeitschrift
f?r
at 180, n. 1, and Allen
58 [1931], 170-204,
vergleichende
Sprachforschung
[n. 10], 171, but also
this is no valid counterargument
Einarson
since Phoenician
'en yielded
[n. 3], 20, n. 28). However,
on the
the vowel sign for Id in the South
Iberian and Tartessan
also depend
scripts, which
Phoenician
and thus ensure that the pronunciation
of the Phoenician
version of the
alphabet
an e-vowel (cf. Untermann
letter name must
indeed have contained
[n. 35], 55).
50
Cf. N?ldeke
(n. 20), 135, Schwyzer
(n. 49), 177-84, Schwyzer
(n. 11), 140, and B.B. Powell,
Homer
and the Origin of the Greek Alphabet
to postulate
1991), 36; it is unnecessary
(Cambridge,
or
here the 'restitution' of a seemingly elided vowel (Einarson
[n. 3], 9), or a Semitic accusative
absolutive
'Griechisches
und semitisches
ending
(Ruijgh
[n. 20], 557-8; J. Tropper,
Alphabet:
Buchstabennamen
und Sibilantenentsprechungen',
der Deutschen
Zeitschrift
Morgenl?ndischen
150 [2000], 317-21, at 317-19).
Gesellschaft
51
Cf. Ruijgh
(n. 25), 27, n. 87, and Ruijgh (n. 20), 558; on Xdfi?Sa see Einarson
(n. 3), 3-4.
52
Thus Einarson
(n. 3), 2, followed by Powell (n. 50), 37, Ruijgh
(n. 25), 27, n. 88, and Ruijgh
loss.
(n. 20), 558, whereas Schwyzer
(n. 49), 179, and Tropper
(n. 50), 319, assume an unmotivated
On ??i)after vv cf. ? IX.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

414
weak

GRAECO-SEMITIC

LETTER

of the underlying
name
the Phoenician

Phoenician

sound

schwa
forms

of

nating
Somewhat

NAMES
rather

vowel,

two

than

alter

cf. Hebr.

(e.g. gamllgiml;
gimel).53
are normally
the Semitic
voiceless
rendered
inconsistently,
stops
by
<
<
Greek
voiceless
, rav <
b?t, SeXra < dalt, fjral*rjra
h?t, rr?.?<pe
stops (?rjra
<
in the case of
taw), but by a voiceless
aspirate
'alp and by a geminate
dX<f>a
<
<
Karma
case
ra < yod
in
the
of
and
voiceless
stop
kap
q?p;54 in l
forma

(5)

instead of the expected -fl Sa, the final syllable must have been influenced by the
preceding

rjra,

?rjra,

drjra.55

In the same domain, it is slightly puzzling that the emphatic stop of t?t was
identified with theGreek voiceless aspirate in Orjra for /th/ (contrast forma < q?p

(6)

about
Phoenician
emphatic
q-), but we do not know
enough
that this was
because
/t/ was
simply
already
represented
so that the remaining
tawlrav
of Phoenician
dental
could
plosive

with

for

service

the
remaining
differences.56

articulatory

to these

In addition

points,
for
relevant

more

and/or

dental

there
the

of

plosive

are a few

letter
of

history

VII. THE SIGNIFICANCE


Greek

anyway

the

was

traced

large-scale

adopted

from

corresponding
is reminiscent
of

to a proto-form
learned
the alphabet

back

Greeks

had

called

f pet,

intermediaries,

?xv, has not


regarding
will be presented
here,

third,

solution

OF 'PQ

it proves
what
is significant
because
trade
activities
of
the Phoenicians

The

letter, which
be

the

basin in the first half of the first millennium

Mediterranean
alphabet
Aramaeans.

name

letter
given

complex
of
these,

not

two

at least to highlight the issue (? IX).

itwill be worthwhile

The

are more
first

signs (? VIII), have been discussed

in the past, while


the problem
raised by
even though
no definitive
the same interest;

repeatedly
attracted

by
into
any

The

alphabet.

adequately
be pressed

notwithstanding

which

adaptations

the

(?VII) and the sibilant/affricate

concerning p

Greek,

to

phonetics

affirm

the

Phoenicians
letter

Hebrew
a head
*ra's,

and

not,

for

is called

r?s

'head'

on

could

be

the ancient
pco.5S Thus,
are fully vindicated.59

instance,

which

widely

the

from

(cf. the shape of the


r?s can
and Aramaic

top of a neck). Hebrew


this in turn yielded
Phoenician
their letter should
the Aramaeans,
sources

the

B.c.: that the Greek

and

from

suspected

throughout

agree

on

r?s.57

If the

therefore

be

Phoenician

53
and M.G.
Amadasi
After N?ldeke
Guzzo,
(n. 20), 135, cf. now J. Friedrich, W R?llig
auch e fur a, wohl weil
Grammatik
(Rome, 19993), 39 ('Gelegentlich
begegnet
Ph?nizisch-punische
form
the Cyprian
hin ausgesprochen
das a hier nach
(n. 20), 558, also mentions
wurde'); Ruijgh
hdXros for S?Xros 'tablet' in ICS 217.26 (ta-la-to-ne).
54 For
to explain
this cf. Einarson
(n. 3), 1 and 19, n. 5, and
attempts
(highly speculative)
that to some degree the
(n. 20), 558, but see also McCarter
(n. 25), 91, n. 69: 'it is probable
Ruijgh
Greeks
alphabet; and one need not seek motives where
played fast and loose with the Phoenician
none exists'.
55
Cf. Wackernagel
(n. 49), 181, and Ruijgh
(n. 3), 3,
(n. 20), 558; Einarson
(n. 9), 71, Schwyzer
>
an assimilation
rather
to the following
*y?tkap), but this would
kap (i.e. *y?~d-kap
postulates
b?t.
yield 'fyok and one would also expect Y alb for 'alp because of the following
56
Cf. McCarter
(n. 20), 560.
(n. 25), 95, n. 77, Ruijgh
(n. 25), 28 with n. 91, and Ruijgh
57
Guzzo
See Friedrich, R?llig and Amadasi
(n. 53), 12-13, 36 and 43.
58 Cf. N?ldeke
Guzzo
(n. 25), 100, n. 87, Amadasi
(n. 20), 135-6, Gelb
(n. 28), 176, McCarter
source is postulated
by
(n. 20), 545; an Aramaic
(n. 25), 296 (with literature in n. 9), and Ruijgh
Kilo 41 (1963), 38-57,
des griechischen
Schrift und Anfange
S. Segert, 'Altaram?ische
Alphabets',
das Alphabet
at 48-52, Driver
'Haben Aram?er
den Griechen
and E.A. Knauf,
(n. 1), 266-7,

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS

415

WILLI

VIII. THE SIBILANT

SIGNS

The case of the names for the Phoenician sibilant letters is less straightforward. The
Northwest Semitic/Phoenician alphabet had four letters for sibilant sounds: zayin for
voiced

Izl as no.

sharp

or affricated
one

only

sibilant

7 in the alphabet
samk for voiceless
Is/ as no.
sequence,
15, sad? for
I si as no. 18, and sin for I si as no. 21. Greek,
on the other hand,
had
was
Isl. The
sound which
as
phoneme
represented
by Z and realized

either [dz] or (at least in classical times) [zd]need not have been monophonemic,60
it certainly

a voiced

included

sibilant

element;

when

hence,

the

but

of

adaptors

the

alphabet decided to employ Phoenician spare signs to render the biphonematic groups
[dz] and [ks] by a single letter, the choice of the zayin sign for the former group was
obvious.

relatively

It would

the name ?rjra. However,

follow

that

the name

zayin

should

also

be

continued

by

it has been suggested repeatedly in the past that this is not

name
name
rise to the Greek
odv for the letter
that
gave
zayin
in
and
to
sad?.
since Greek
and ?ef
corresponds
Moreover,
shape
position
oiyfia
sin and samk respectively
continue
Phoenician
a complete
in both
shape and position,
confusion
would
have arisen
from the transfer
of the name
zayin: for the name aiypua
would
also have been
transformed
from the name
the name ?Jjra from sad?, and
samk,
so,

but

that

the

|ef

from

the name

All

of

sin.61

this is neither historically


must

development

have

been much

nor

simpler,

linguistically
as Roger

to write their one sibilant phoneme Isl the Greeks


which

was

sharp

sad?;

pronounced
note
that

as [s], not
Phoenician

[s], at least
names

in some
starting

In reality, the

plausible.

Woodard

has

shown.62

In order

initially used either the palatal sin,


of
with

the Phoenician
the

sad?

dialects,
are

sound

transcribed with the Greek Isl sign in historical texts (e.g. Sid?n ~ UiS

or the
indeed

v).63As for the

no counterevidence
Welt des Orients
18 (1987), 45-8, although
to the p
vermittelt?',
argument
exists (so that Segert [above] accepts
that the names of the letters were mediated
by the Phoeni
is said to reflect a Phoenician
sound change ? >
cians). Given Hebrew y?d, the name Iwra, which
? in *y?d > yod (Cross [n. 25], 14;Naveh
[n. 1], 183; Ruijgh
[n. 20], 543), seems less telling. Why
'clearly Aramaic'
(n. 1), 266, remains unclear, not least
by Driver
dX(f)a and ?rjra are called
because aA</>a is phonetically
closer to Phoenician
'alp than to Aramaic
'alep.
59
Cf. Hdt. 5.58; Critias fr. 88B2.10 D.-K.;
fr. 501 Rose; Diod.
Ephorus FGrH 70F105; Arist.
Sic. 3.67.1; Plut. Mor.
Plin. HN
Dion.
738f; Lucan.
3.220-4;
7.192; Tac. Ann.
11.14; Nonn.
for dissenting
voices
FGrH 9F3) see A. Willi,
1F20; Anaximander
(Hecataeus FGrH
der Alphabetschrift
MH
nach Griechenland',
62 (2005),
'KdSfjio? av49r?Ke. Zur Vermittlung
also the term (?yoivLKrjta {(fxuviKiKa, <f>oiv?Kia) ypdpuxara,
whose
162-71, at 169, n. 28. Note
were disputed
to Schol. Dion.
Thr. p. 184.20-185.2
implications
already in antiquity
according
= FGrH
(Hdt. 5.58.2; Soph. fr. 514 Radt; SIG3 38.37; Chron. Lind.
532, B15; cf. M.
Hilgard
'L'adozione
dell'alfabeto
nel
mondo
PdP
at 83-4, and
31
Burzachechi,
greco',
[1976], 82-102,
Heubeck
and the Cretan
'scribe' and rroiviKa^ev
in SEG
'to write'
iroiviKaor?s
[n. 25], X108),
27.631 (L.H. Jeffery and A. Morpurgo
and TroiviKa^ev. BM 1969. 4-2. 1, a
'rroiviKaar?s
Davies,
new archaic
9 [1970], 118-54, esp. at 132-3 and 152-3; G.P.
from Crete', Kadmos
inscription
Edwards
and R.B. Edwards,
'The meaning
and etymology
of irotviKaords',
Kadmos
16 [1977],
4.259-66;

131-40).
60
On the pronunciation
status of Greek Z see Allen
and the phonemic
(n. 10), 56-9.
61
Thus Jeffery (n. 16), 25-8, as well as Driver
(n. 1), 268, Brixhe
(n. 48), 332-3, and Powell
>
(n. 50), 46-8 (with *o ?jlk *oe?j,Ka > ^aefxya > *aeyfjLa > oiyp,a\), after I. Taylor, The History
of
the Alphabet,
and N?ldeke
but
//(London,
1883), 97-102,
(n. 20), 134; somewhat
differently,
<
7-8
Einarson
and
and
318
(n. 3),
11,
equally implausibly,
(n. 50),
Tropper
*sinna\).
(aiyixa
62
to Homer.
R.D. Woodard,
Greek Writing from Knossos
A Linguistic
Interpretation
of the
and the Continuity
Greek Literacy
Origin of the Greek Alphabet
of Ancient
(New York and
Oxford,
1997), 147-88.
63
See Friedrich,
and Amadasi
Guzzo
(n. 53), 11-12, n. 4 and 26, on the sad?
R?llig
and Woodard
(n. 62), 184, on the pronunciation
transcription,
[s] of sin in Cyprian Phoenician;
cf. also Friedrich, R?llig and Amadasi
Guzzo
(n. 53), 25, on eg. 'asr for 'asr in Sidon.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

416

LETTER

GRAECO-SEMITIC

letter

McCarter

names,

and

others

had

NAMES
on

suspected,

already

the basis

of

the evidence

from Ugarit where the cuneiform letter for /t/ is equated with the Akkadian
syllabogram <sa> (cf. ?V), that the letter sin was originally called san:64 themeaning
of Phoenician *san (< *t_anri) 'bow' would fit the letter shape just as well as the
at first adopted
to this theory,
of sin 'tooth'. According
the
then, the Greeks
meaning
letter shapes
and names
of both
sad? and sin (*san)
for their phoneme
Isl. However,
because
consonant
of *san as [san] may
have been
the pronunciation
of the initial
more
to that of a normal
in
Greek
sad? disappeared.
similar
Isl, the name
Similarly,
each

local

script

of Greece

one

only

of

the

two

letter

shapes

survived.

This

situation

was further obscured when, probably only in the fifth century (cf. ? II), an additional
name

came

to be used

for

the

letter

that had

a transformed
Phoenician
being
common
noun meaning
'hissing,
hissing
from

'to hiss').65 The only slightly paradoxical


(1.139)
where

is right,
the shape

the name
of

odv

sad? was

survived

continued

arisen

from

samk,

Greek

sound

(i.e.

sin (*san), namely


Far
oiyfia.
formed
is a regularly

olypua

sibilant)'

at?co

(cf. onomatopoetic

thing in all this is the fact that, ifHerodotus

not exclusively66
in regions
longer
though
name
over in the
the
(M), whereas
o?y?xa took

Ionian sin (*san) region (?); but even this 'contradiction' ismitigated by the transfer
of the sad? shape into the sin (*san) slot in odv regions:67 hence, of the three
constituents

position,

name

(in o?y?xa
to
Turning

Phoenician
(*r/ra)

*zch

and

drjra

regions)

name

this
l,r)ra,
Hebrew
(~
than

and

name

is more

zayin;

the descendant

either

only

shape,
'innovated'.

was

cf. Phoen.
of

the

to be
likely
~
'en Hebr.

an alternative

(in odv

shape

or

regions)

the

an

of
analogical
reshaping
after
the
'ayin)
following
rjra
*z?t 'olive tree'.68
letter name

Finally, ?et has nothing to do with any of the Phoenician sibilant names, but follows
the general pattern by which all the newly-created letters for Greek consonants (or
consonant
the

end

of

not
groups)
the alphabet,

e-vowel

in the names

analogy
Phoenician

with

found
after

in Phoenician
the

last Semitic

were

named:
letter

rav

these
<

letters

taw, and

were

added

at

the

long closed
in
been chosen

have
<?5e?for /ph/, x^t for /kh/ and j/fet for /ps/ must
was
7ret < p?' for /p/ where
in the
vowel
the same
present
already
to become
can see here
of a process
that was
the beginnings
name.69 We

64
/ is rendered by Akkadian
name of Ugaritic
McCarter
(n. 25), 100-1, n. 88: 'The abbreviated
name of the letter, repre
sa. This indicates
of the old Canaanite
that the correct vocalization
*tann- and not *tjnn- (cf.
sented by the "composite
bow" pictogram,
was, at least in one dialect,
one could
[n. 24], 26, and now Sass [1988] [n. 1], 132). Alternatively,
already Cross and Lambdin
oav a Greek
after *ow, the latter being eventually
innovation
consider
given up in favour of
atyfjia (Ruijgh [n. 20], 559).
65
Cf. Schwyzer
188-9; Schwyzer
(n. 25), 99, n. 85;
(n. 11), 140-1; McCarter
(n. 49),
des mots (Paris, 1968-80),
de la langue grecque. Histoire
P. Chantraine,
Dictionnaire
?tymologique
s.w. aiypua and o?toj, Ruijgh
2.1002-3,
(n. 20), 559; implausibly Gelb
(n. 25), 27, n. 87; Ruijgh
< *simk
'shoulder').
(n. 28), 141 (o?yfxa
66
and
Burzachechi
regions (e.g. Laconia
(n. 59), 95-6, rightly stresses that there are also Doric
the sin (*san) shape survived and that the ancient sources (Hdt. 1.139; Athen.
where
Messenia)
san quella consonante
che gli Ioni
che i Dori
chiamavano
'dicono
11.467a)
semplicemente
san il segno M e gli Ioni sigma il segno 2?.
chiamavano
sigma, non che iDori chiamassero
67
This is pointed out by Brixhe (n. 48), 330, even though he derives the name odv from samk;
cf. also Jeffery (n. 16), 131 and 404 no. 16, 261 and 410 no. 19, and Slings (n. 25), 650.
68
Cf. N?ldeke
(n. 10), 170;
(n. 1), 159; Allen
(n. 25), 94, n. 74; Driver
(n. 20), 134; McCarter
Powell
(n. 20), 543; the *z?t theory is preferred by
(n. 25), 27, n. 87; Ruijgh
(n. 50), 37; Ruijgh
Guzzo
and Amadasi
Lidzbarski
R?llig
(n. 11), 140, n. 4, and Friedrich,
(n. 30), 132, Schwyzer
(n. 53), 44 and 133.
69
'Abbreviated writing', Kadmos
Cf. Hermann
(n. 10), 170, and R. W?chter,
(n. 13), 225, Allen
30 (1991), 49-80, at 52.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

predominant
was

in later alphabets derived from the Greek;

names'
such as Latin
syllable
at the end alongside
added

'minimal
not

417

WILLI

ANDREAS

b?, ce, de were


</>et etc.

was

in many

that

of these, only

only reason why


?et
was
borrowed
from

The

used.70

its shape

Phoenician samk; but its phonetic value was about as like or unlike that of samk as the
value

of

Irjra

was

like or unlike

that of

zayin

(cf. above).71

IX. THE NAME OF M


With

the

'minimal

syllable

have

touched
already
the Greek
letter-name

upon

names'
an

of

issue

the

that

Tret, x^t etc.


type Gr.
is of potential
relevance

to be

adaptations

discussed

The

here.

or Lat.

b?, ce etc., we
for a last problem
of
name

Hebrew

for

the

letter from which the Greek M derives ism?m, and the Phoenician name would be
expected to be identical; the zigzag shape of the letter nicely reflects the meaning of
'water'.

m?m,

modelled
II),

there

It would

be

to explain

easy

name

the Greek

pbv, not

f/xc?

or

like, as

the

after the following vv < n?n (cf. ?VI).72 However, as we have seen before (?
was

(possibly

Ionic)

?x

by-form

. Since

?jl?o could

be

into

changed

?xv

without difficulty given the following vv, and since there is no reason why jjlvshould
have

been

changed

into

puw,

it may

be

inferred

that

puw is the older

of

variant

the

Greek name. But why should the Greeks have replaced Phoenician m?(m) by puoP.Of
course

one might

that

argue

the Phoenicians

had

already

changed

the name

of

the

sign into *mo,73 and it has in fact been suggested that such a Phoenician *mo had
developed out of a hypothetical *maw that would be similar to taw and waw;14 but
since

the

latter

two names

are

represented

by Greek

rav

and

Fav,

not

"\r

and

^Foj,

70
see A.E. Gordon,
The
On the Latin letter names and their possible Etruscan
predecessors
'Le
Letter Names
and London,
1973); J. Bo??aert,
(Berkeley, Los Angeles
of the Latin Alphabet
nom des lettres de l'alphabet
'De litterarum
34 (1975),
152-60; F.V. Mares,
latin', Latomus
WSt n.s. 11 (1977), 219-24; W.D. Lebek,
'Eine Eselei aus Ostia und die
latinarum nominibus',
ZPE
42 (1981),
after W
'Die lateinischen
lateinischen
Schulze,
Buchstabennamen',
59-65,
zu Berlin (1904), 760-85;
der Akademie
der Wissenschaften
Buchstabennamen',
Sitzungsberichte
zur Geschichte
und
des etruskischen,
lateinischen
Hermann
(n. 13); M. Hammarstr?m,
Beitr?ge
'Die antiken Buchstaben
1920), 15-34; M. Hammarstr?m,
(Helsinki,
griechischen
Alphabets
zur Geschichte
namen. Zugleich
1 (1930),
Arctos
der griechischen
ein Beitrag
Lauttheorien',
'The Etruscan
3-40; B.L. Ullmann,
origin of the Roman
alphabet and the names of the letters',
CPh 22 (1927), 372-7; Einarson
(n. 3), 16, notes that
(n. 3), 15-17 and 23-4, nn. 56-63. Einarson
or influenced
the
from the Greek
that were borrowed
'of all the alphabets
by it, and of which
and Cyrillic
letter names are known
Armenian,
Roman,
only
Georgian,
Coptic, Gothic,
and
literate Copts were
the Greek
because
retained
names, doubtless
largely bilingual
Coptic
found one set of names convenient'
(cf. also Schwyzer
[n. 49], 193-9, on the spread of the 'ABC
principle').
71
Cf. W?chter
n. 83.
72 Thus it is

(1989)

(n. 25), 42, on

/ks/ as a new phonetic

value,

and McCarter

(n. 25), 98,

to posit, with Driver


unnecessary
(n. 1), 158 and 262, a singular *muw next to the
Semitic plurale tantum m?m. After
according
?jlvand vv, even $v may have been used occasionally:
of Luc. Jud. Voc. 9 and AP 9.385.14
to LSJ, 1188, s.v. fet, ?v is a falsa
lectio in manuscripts
(cf.
when ?v and f el were no
Wackernagel
[n. 9], 17; but f? may have intruded into the manuscripts
as [ks?]).
both being pronounced
longer distinguishable,
73
There is no support for the name *mom postulated
by Powell (n. 50), 37.1 am also unable to
instead of vv (as posited
of a letter name v
find any attestation
by Schwyzer
[n. 49], 188; cf.
W?chter
[n. 69], 51-2).
74
-au- > -?- see Friedrich,
and Amadasi
Guzzo
For later Phoenician
(n. 53), 45.
R?llig
some suggestive
forms
McCarter
(n. 25), 97, n. 82, points out that 'the Semitic languages provide
m? (< *mawi?), Old South Arabic
Akkadian
for comparison:
(Sabaean) mw, and so on' (cf. also
*t? < taw; Einarson
Cross and Lambdin
[n. 3], 2).
[n. 24], 25, on Ugaritic

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

418

LETTER

GRAECO-SEMITIC

NAMES

such speculations do not lead very far and itmay be preferable to consider other
possibilities.
One (admittedly hypothetical) alternative is suggested by the phenomenon of
'abbreviated writing', which has been discussed in detail by Rudolf W?chter.75 In early
Greek

vowel

inscriptions

Semitic

but

alphabets,

is regular

that

letter names.

in the respective

omitted

sometimes

in a way

vowels are most

discernible. The missing


consonant

are

letters
still

not
enough

as

systematically
to make

in the

a pattern

often those which also follow the initial


of

One

the most

common

instances

is that of

0, not least because the verbal form dve6r]K '(s)he dedicated' is so frequent in the
the spelling ANE0KE
is then found. In this and
early texts: instead of ANE0HKE,
cases

similar
d-v

-dr)-K

vv, ?et,

stonecutter

the

). Several

Tret, pw,

of

the

the word

into

open

to such

already

corresponded
like drjra, must
have been

i/*et. Others,

(fre?, ^et,

divided

apparently
letter names

(i.e.

syllables
open

shortened

syllables:

accordingly

(i.e. into 6rj etc.) when the stonecutter spelled to himself in a low voice what he had to
In fact,

write.
source

for

such

shortened

Latin

the

letter

of

versions
names

the

names

letter
etc.

de

ce,

b?,

(cf.

have

may

been

another
the

when

Thus,

? VIII).76

stonecutter had incised the 0 and muttered Orjto himself, he could easily 'forget' that
he needed

to complete
the opposite:

letter

another
let us

Now

imagine

the syllable.
someone
wants

to learn

how

to read

on

and write

the basis of words like ANE0KE. His or her teacher points to A and reads a, points
toNE and reads ve, points to 0 and reads Or],points toKE and reads *e. That pupils
at elementary
minimal
open

syllables

that

the

letter

single

was

to recognize
and
vowel-less
orthographically
trained

indeed
The

in our imaginary school

of ANE0KE
impression

were
in antiquity
is well
attested.77

schools

scene would
a

really

such

separate

'syllabogram'

'syllable'

therefore have created


the

representing

the

entire

sequence /th?/-Of course, once the Greek alphabet was established there would have
been

of 0

instances

enough

+ vowel

to correct

such

an erroneous

values
the

and

case:

system.

names
after

It may

of

all,

therefore

when

accidentally,

the

letters

the vowels

some

for

the first

this would

time,

were

consistently
that
be hypothesized
learned

Greeks

'omitted'

by

observing

in

learned the shapes,

not necessarily
in the Phoenician

the unexpected

to read

But

conclusion.

the initial stages of theWestern alphabet history, when theGreeks

have

been

spelling
created

name

pew was
a Phoenician

teacher

slowly spelling out aloud a word inwhich the sign m?m stood for a syllable fmql.
Such a scenario, while ultimately unprovable, would gain further plausibility if it
could be shown that a single m?m did in fact represent a syllable /m?/ very frequently
(and possibly more frequently than /m?/, /m?/, /m?/ etc.) in the first Phoenician texts
with which the Greeks became familiar. It is often assumed that the transfer of the
alphabet into Greece took place in a trading context; one might therefore think of
lists of goods,

sale deeds

and

the

like. However,

there

is not

a single

piece

of

evidence

75
IF 30 (1912), 1-47,
W?chter
(n. 69), after a hint in F. Solmsen,
'ZiX-qv?s Edrvpos
Tirvpos,
at 20, n. 1, on Z for /si/; for a similar phenomenon
in early Latin cf. Terentius
Scaurus, De ortho
Keil, and Hammarstr?m
graphia 7.14.15-7.15.7
(1920) (n. 70), 31-3.
76 Because
of the Latin names, Ullman
(n. 70), 374-5, even suggests that short names like *?et
for j87JTawere regularly used; but see the reservations
in Hammarstr?m
(1930) (n. 70), 5-14.
77
See W?chter
Zur Geschichte
'BA-BE-BH-BI-BO-BY-BQ...
(n. 69), 73, and R. W?chter,
ZPE
des elementaren
bei den Griechen,
Etruskern
und Venetern',
Schreibunterrichts
146 (2004),
Histoire
after H.-I. Marrou,
de l'?ducation
dans l'antiquit?
61-74,
19656), 231, and
(Paris,
R. Cribiore,
in Graeco-Roman
and Students
Teachers,
Writing,
(Atlanta,
Egypt
1996), 40-2;
cf. also, again, the transformation
of the alphabet
into a syllabic script in ancient
Iberia (n. 35
above).

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

to

this

support

records,

idea
not

would

even

have

Phoenician

at

Greece

to honour

decided

of

the

their

time

in the same

gods

can

The

for

that we

such

should

as a corollary

of the

observed

Greeks

how

and
orientalizing
'import'),
manner.
This would
explain

they
not

Semitic script that have been found in

as votive
texts
most
votive
easily
interpreted
gifts, but also why
our
In
context
the
Greek
material.
among
early
present
epigraphic

are

be

predominant

therefore intriguing that the best-attested


inscriptions from Sicily
together with
documentation

of

'Phoenician/Punic

at Motya.79

stelae

it is

sequence in the early Phoenician (Punic)


Sardinia the region for which the early

abroad'

is richest

is the name

of

the

important

(blhmn), appearing on nearly three dozen formulaic

Phoenician god Baal Hammon


votive

B.c.

(as another
fashionable

only why the few objects inscribed inNorthwest


Greece

into Greece

century

used

been

elsewhere78

inscribed votive gifts in the temples that were newly

same

the

ninth

have

might
argued

of the alphabet

tradesmen deposited

over

all

built

wave

cultural

orientalizing

I have

Hence,

perished.

rather view the introduction

which

ostraca,

though

419

WILLI

ANDREAS

hmn

= Hammon

the kind

of

the naming
pico, we have now considered
history
for one, with which
this survey may
except

be

Obviously,

would

be precisely

word inwhich M represented the syllable /m?/ (Gr. ?x ).


X. A LOST NAME
Whatever
all

the

story

the commonly

behind
used

Greek

Greek

FOR A LOST LETTER

completed most appropriately. In theMilesian


not

of

letters

numbering system (cf. ? II), Q for IqI is

even

at the end of
its usual
the sequence
of
though
position
taw = rav (i.e. after Y, 0, X, W) indicates
additional
letters following
Phoenician
that
at a late stage,
in East
it was
innovated
the rest of the alphabet
had
Ionia, when
the Greek
Since Q has the numerical
world.80
value
'800',
already
spread
throughout
as f or
and really
the following,
'900'. It is often written
last, letter has the value
~\
the

last

letter,

but its oldest shape is similar to a T with additional vertical hastae at each end of the
horizontal

hasta

(T):

this

is how

it appears

in a

seventh-century

from

abecedary

Samos, which again places it after Q in the alphabetic sequence (and which also
provides one of the first attestations of ?2).81Like Q, the letter in question must be an
Ionian

Cyzicus)
Carian
their

since

invention

Mesambria),
and
as well

neighbours

so far

in East
only
in the Ionian
as

it has
Ionian

been

found,

with

few

exceptions

Pontic

(Attica,

cities

colony
the Phrygian
scripts,
in Asia Minor.83

Teos,
Halicarnassus,
Erythrae,
(Ephesus,
a similar
in
Since
letter
Massalia.82
exists

the

some

of

the Greeks

may

have

adopted

it from

78
Willi
(n. 59).
79
Le iscrizioni (Rome,
See especially M.G. Amadasi
Scavi a Mozia:
Guzzo,
1986), and also
M.G. Guzzo Amadasi,
Le iscrizioni fenicie
(Rome, 1967), with
epuniche d?lie colonie in occidente
to b'l hmn from Malta
and Sardinia;
in Motya
the only other words
further dedications
containing m?m and occurring with some frequency
(though much more rarely than b'l hmn) are
mlkt for a type of sacrifice and mtnt for 'gift'.
80 Cf.
[...] a chance de refl?ter l'ordre
Ruijgh
(n. 25), 44: 'L'ordre des lettres additionelles
de leur addition'; W?chter
(1989) (n. 25), 48.
chronologique
81
B.c.
Jeffery (n. 16), 428 and 471 no. la, with pi. 79, with a date of 660-650
82
en grec et en phrygien', Bulletin de la
'Palatalisations
Jeffery (n. 16), 38-9; cf. also C Brixhe,
and G. Genzardi,
'Una singolare
Soci?t? de Linguistique
de Paris 11 (1982), 209^17, at 216-22,
dell'Accademia
Nazionale
dei Lincei, ser. 8, 42 (1987), 303-9.
lettera greca: il sampf, Rendiconti
83
For Carian
cf. Schwyzer
(n. 82), 305-6;
Jeffery
(n. 16), 39; for
(n. 11), 149; Genzardi
an Athena
von Assesos',
Anatolica
30 (1998),
'Eine Weihung
Epigraphica
Phrygian R. W?chter,
Some guidelines
for
1-8, at 3, n. 10; Brixhe
(n. 82), 229-35; C. Brixhe,
'History of the alphabet:

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

420

LETTER

GRAECO-SEMITIC

NAMES

The value of this letter was a sharp [s(s)].Where it occurs it competes with the
spelling ZTin words which would show a geminate TT inAttic Greek (e.g. O?Xaooa).
its use

However,

is never

entirely

'it has

consistent:

[...] been

tentatively

restored

in the

poetry of Hipponax of Chios, to explain the apparent Atticism OaXar[rr)s] of the


papyrus; but in the earliest Chian inscription [...] (frvXaoo v is spelt with double
sigma' and 'though it is attested inMilesian colonies, inMiletos itself there is as yet no
we
or double
instead
find
and
e.g.
single
sigma,
[...]
Tet^toa^s
Kr?pvoG raC }4
The name
this final
in modern
letter is known
is oapurr? or o?p,m.
times
by which
The origin
of this name
is not entirely
not ancient.
Because
clear, but it is certainly
example;

o?v could stand for ws ?v Tike' in Byzantine Greek, itmay have been coined in
times as the letter shape ^ was felt to resemble that of FI (i.e. odv rrl Tike

Byzantine

even seems
to be found only
in the early seventeenth
pf );85 the first attestation
century
in the writings
of Joseph
The ancient
is unknown.87
Claims
name,
however,
Scaliger.86
ss
was
or
that
and
'the name
with
doubtless
*ss?ri
'a
that
the
is
*ss?n
began
sign

carefully distorted to distinguish it from M have no foundation whatsoever.88 The


latter claim implicitly acknowledges that the sign does not look like a sad? (Doric
o?v), and the former forgets that the sharp sibilant is not used word-initially inGreek
so

that

the

acrophonic

principle

cannot

be

invoked;

moreover,

if the

'o?puri

sign

continued

the Semitic sad?, it should also be placed in the position of sad? in the
alphabet sequence (just as F and p continued to be placed in the positions of waw and

in A.-F. Christidis
the
Greek. From
avoiding
(ed.), A History
oversimplification',
of Ancient
to Late Antiquity
Beginnings
(Cambridge,
2007), 277-87, at 281.
84
see K. Latte,
'De nonnullis
(n. 16), 39; for the Hipponax
argument
Jeffery
papyris
97 (1948), 37-57, at 46. That the spelling of Sappho's name as Warr^
Oxyrrhynchiis',
Philologus
an erroneous
transliteration
of Ta7r</>a> (G. Zuntz,
'On the etymology
of the name
represents
8 [1951], 12-35, at 16-22) and thus indirectly attests T also for Aeolic Lesbos,
also
Sappho', MH
remains a possibility
'Alc?e 384 LP, Voigt', RPh 62 [1988], 291-8, at 294
(cf. now G. Liberman,
with further bibliography
in n. 9).
85
Thus e.g. Schwyzer
(n. 11), 149; cf. Schulze
(n. 70), 769: '?brigens w?re es w?nschenswert,
wenn die Grammatiker,
die das Zahlzeichen
f?r 900 anstandslos
o?prn oder gar o?v zu nennen
endlich
einmal mit
einem
brauchbaren
sei es auch nur aus der
fortfahren,
Zeugnisse
Zeit herausr?ckten.'
byzantinischen
86
See Einarson
de literarum Ionicarum
(n. 3), 13 and 22, n. 51: 'In the "Digressio
origine" on
in chronologica
to his Thesaurus
of his Animadversiones
Eusebii,
appended
1606 [...]) he devotes the better part of a folio page (p. 108) to the eVtcjTijitov
(Leyden,
an
tt?). He calls it this because
spelled E?v Til, San pi, San Pi, E?v
"^ looks like
=
Clouds 23, explaining
antisigma
san) and a 77-,and because a scholiast on Aristophanes
(sigma
says that these horses have the imprint of a sigma, ro y?p a, Kal ro k xapctacro/xevov
oapcf)6pas,
the scholiast writes
E?v
k because he is confusing
the
says Scaliger;
eXeyov. This is nonsense,
etymology with that of kotttt arias. He is garbling his ancient source, who must have said ro y?p
rrl eXeyov. From Scaliger
the name reached G.J. Vossius
?, /cat ro ttl xaPa(JCJ?pevov E?v
(De
arte grammatica
[Amsterdam,
1635] 1. 23, p. 91), and from him it reached the school grammars:
...
a Nicolao
nunc autem
cf. Institutiones
olim quidem
Clenardo
linguae Graecae
scriptae
...
Gerardi Vossii. Editio altera ... (Leyden,
48: "...vocatur
p.
expurgatae
studioatqueopera
1642),
'
that, against Einarson,
sanpi, quia conflata est ex inverso a?v, hoc est o?yp,a, & incluso ir." Note
sense
if
and
emendation
make
better
he
Scaliger's argumentation
already knew the name o?p,rri;
a more
otherwise
correction
of the scholion would
be ro y?p a [/cat ro k]
straightforward
vov E?v
intruded from an additional
reference on
eXeyov (with /cat to k having
Xapaoo?p,
pp.

102-13

temporum...
a?vTTL (also

/C07T77aTia?).
87
Cf. Jeffery (n. 16), 39, and Brixhe (n. 48), 335.
88
Einarson
(n. 3), 13; cf. similarly Ruijgh
(n. 25),
Ruijgh
(n. 47), 675.

32-3; Ruijgh

(n. 20),

536, 544, and

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

563-5;

q?p

even

respectively,

when

no

were

they

421

WILLI

ANDREAS

longer

outside

employed

the

numbering

system).89
to discard

Fortunately,

has
-

letter
'o?poTL
misunderstood
to realize

order

as unfounded
does not mean
hypothesis
There
is a hitherto
nameless.
unnoticed

the

to

*ss?n

remain

ancient

source

this, all we

have

which

tells
is once

to do

us

what
explicitly
to
consider
again

that
-

name

its true
the position

the
since

was.

of

In

'o?puri

after Q. The 'classical' East Ionic alphabet of the fifth century, which was officially
adopted by Athens in 403/2 B.c., had 24 letters (cf. ? II): of the 22 Phoenician letters, F
(waw), <?(q?p) and M (sade) had been given up (except that the former twomarginally
lived on as numeral signs), but Y, 0, X, W, Q had been added at the end. In the
above-mentioned

cities

the written

etc.,

Erythrae

Ephesus,

(non-numerical)

alphabet

therefore had 25 letters before 'o?arrC fell out of use around the middle of the fifth
B.c.:

century

Q was

no.

letter

and

24,

was

'o?parC

letter

no.

we

25. Now

read

the

= fr.
46,
following statement in a fragment of the Roman scholar Varro (fr. 3 Funaioli

p. 201.4-9

Goetz-Schoell):

est littera, quam uocant


Ion scribit, quinta uicesima
agma, cuius forma nulla est et uox
est Graecis
et Latinis, ut his uerbis: aggulus, aggens, agguilla,
communis
iggerunt. in eius modi
et Accius
noster bina G scribunt, alii N et G, quod in hoc ueritatem
Graeci
uidere facile non est.
similiter agceps, agcora.
ut

is called 'agma', which has no shape, but a phonetic


As Ion writes,
there is a 25th letter, which
value that is the same in Greek
and Latin, as in the following words: aggulus, aggens, agguilla,
and our Accius write a geminate
GG, while others
iggerunt. In words of this type, the Greeks
to recognize
it is difficult
in the former;
write NG, because
the real sound
similarly agceps,
agcora.
reason

The

why

this

never

has

fragment

been

with

connected

the

is

'oapLrrC problem

clearly that Varro's words at first sight point in a completely different direction. Of
course

Varro

was

right

in observing

across

the work

of

that words

rather than NT

commonly written with iT


a certain

Ion who

containing

mentioned

a velar

nasal

[rj] were

In his studies he then came

in Greek.

a twenty-fifth

letter

to as

referred

=
aypua, which would be pronounced as [anma] (cf. (fr?eypia
[phthenma] next to
no
Varro
could
be
familiar
with
real
letter after Q, he
Since
any
longer
c?ddyyopLai90).
inferred that his source must be thinking of the graphemic peculiarity in spelling the
to occur

[n], which
happened
this conclusion
authority,

has

the

should

sound

default

twenty-fifth

assumption
letter, he

convention
spelling
cuius forma
nulla

in the name

never

been

be

means
probably
a letter.92 Thus,
with

est onwards

are

aypca

questioned
an
that when

of Varro's
[arjma]. Because
lives on today,91
although
a
author
of
ancient
speaks

and

not mix
he says and does
that Varro's
it is a priori
unlikely
on Ion.
still based
what

up
words

an

odd
from

89

Cf. already Schulze


(n. 82), 305; Woodard
(n. 62), 178-9; Slings (n.
(n. 70), 769; Genzardi
also that Herodotus
could hardly have regarded the name o?v as Doric
(cf. ? II)
25), 644-6. Note
if it had still been in use in his own East Ionic dialect.
90
et du grec ancien (Paris, 1972), 146, with
Cf. M. Lejeune, Phon?tique
historique du myc?nien
from ocf>iyyoj, iXrjXeypat from iXeyxw,
further examples
etc.); it is less clear whether
(eo^Lypai
as [rjm], even when
a group TM was always pronounced
from *gm, not *ngm
it originated
[above], 77-8).
Varron grammairien
latin (Paris, 1954), 128;
e.g. Schwyzer
(n. 49), 181, n. 3; J. Collart,
Laut- und Formenlehre
Lateinische
(Munich,
1977), 15;
Lejeune
(n. 90), 146, n. 4; M. Leumann,
in sein Werk (Heidelberg,
Terentius Varro. Einf?hrung
Allen
(n. 10), 35-6; B. Cardauns, Marcus
2001), 38-9.
92With
as for TT = [rj] one would otherwise
the same justification
have to call El = [?] or O Y
=
[?] a 'letter'.
(Lejeune
91
Cf.

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

422

LETTER

GRAECO-SEMITIC

NAMES

reasons
from the numbering
of Ion's letter, there are further
good
name
con
of 'o?pari,
and that Varro's
ay pua is the ancient
ingenious
on a misunderstanding.
is based
in any way,
struction
Ion is not
Since
introduced
Varro must
that his readers
will
which
be assuming
know
Ion he is talking
about.
Ion must
and scholar
Ion of Chios,
who
be the most
famous
Ion:93 the poet
Hence,
Moreover,

to believe

apart

that

lived in themiddle of the fifth century B.c., competed with Sophocles and Euripides in
the

tragic

at Athens,

agones

lyric

composed

and

history of his native island. This island, Chios,


East

Ionic

area

in which

the

letter

was

'o?pari

still

elegiac

and

poems

wrote

a prose

lies precisely in the relatively small


in use

in Ion's

So when

lifetime.

Ion

felt that he had to tell the rest of theGreek world about Chios, it is natural that he also
in this

mentioned
familiar

with:

an

context

as natural

in fact

that the Danes have an additional letter 0


The ultimate
proof,
the epigraphic
'o?puri
or anchor.
The
hook

The Greek word

however,
looks
lexeme

most

orthographic
peculiarity
as if an encyclopaedia
comes

Greeks
culture

were
pointed

not
out

in their alphabet.
itself. As we have
seen,
aypia
or also
like an inverted
upwards,
noun
neuter
in -ju,a.
formed
Greek

the name

from

like an arrow
?ypia

other

of Danish

pointed

is a regularly

for 'anchor' is ?yKvpa;

it belongs,

like the adjective ?yKvXos

to a root ay/c- < *ank


words,
an
and
a?cati
thus
has
'curved/
Skt.
(cf.
underlying
meaning
a
same
in
The
characterize
derivative
basic
would
crooked
-pua from
meaning
object'.
a formation
rather:
this root. Such
ayaa
(or
*ank-mo)
yields
[arjma]
by
*?yK-ua
or also
the word
'curved/crooked
sound
Thus,
object',
regular
change.95
oiy/xa
is not an invented
but a purely
'anchor-like
nonce-word,
hook',
descriptive
object,
'crooked,

curved'

a large number
to curve'),94
'to bend,
and

of

other

Greek

name for what the alphabetic sign after Q looked like in the sixth and fifth centuries
B.c. As such, and with the revised meaning 'name of the East Ionic letter T/ f /^
not 'nasalized
its own
separate
g', it deserves
("sampi"Y,
its older
the Graeco-Semitic
Greek
alongside
siblings,

Worcester College, Oxford

entry
letter

in our

dictionaries

of

names.96

ANDREASWILLI
andreas.

willi@classics.

ox .ac.uk

93

von Wilamowitz-Moellendorff,
Einarson
Thus
(n. 3), 19, n. 11, after U
correctly
at 279, n. 2, and against Schwyzer
62 (1927), 278-98,
'Lesefr?chte', Hermes
(n. 49), 181, n. 3.
94
s.v. ?y/c-.
Chantraine
(n. 65), 1.10-11,
95
cf. e.g. Sety-jita from *deik-m?,
For the voicing assimilation
etc.;
rrX?y-p,a from *plek-m?,
Lejeune
(n. 90), 77.
96
'nasalized g' is given, wrongly
The entry in LSJ, 11, s.v. ?ypa, where the meaning
groups the
'to break' (root *uag-). The view of
lexeme together with aypa
'fragment', a derivative of ?yvvpu
L. Lupas, Phonologie
and Paris,
du grec attique
1972), 21, that ay pa is a mere
(The Hague
as [agma], is both phonetically
and historically
untenable.
anagram of y?pp,a, pronounced

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions

ANDREAS

APPENDIX.

423

WILLI

THE NORTHWEST SEMITIC AND GREEK


THEIR NAMES

LETTERS AND

ak?pa

'alep ('alp)
b?t

aXcp
?T]6

'house'

?fyca

gimel (garni)
dalet (dalt)

yt|xX
beld

'throw stick', 'camel' (?)


'door'

y?\i[ia (y?\i\ia)
bekxa

he'

el (?tyik?v)

?atv

10

y?d

1008

'hand'

kap
lamed (lamd)

Xa<p

12

'palm (of a hand)'


'thorn' (?)

rie
-rne

Xa?6

j=av (biyaynia)

('hook'?)
'?' ('olive tree'?)
'fence, barrier' (?)

zayin (z?n, *z?t?)


h?t

dz, zd

?fjxa
rrca (*rrca)

h,?

Offia
xcfotJia
30

'water'
nun (nahas?)
samek (samk)

20

22

60

<x*HX 'support' (?)


eye
<pn

pe_
sad?

oa&T]

q?p

ucoqp

r?s (ros)
sin (*san?)

Hfl (M??>)
50

'fish' ('snake')

'ayin ('en)

CT?

o?v
oorara

k(q)

'monkey'
'head'

('mark, sign'?)

23

?(?)

25
26

ps

27
28

jet
ov (? hixq?v)

'mouth'

'tooth' ('bow')
Qav

ka(n)?oa

ss (s?)

100

Q?)

200
300

o?y\ia (*o?v?)
xav

400

v (v \|)lA.?v)

600

X?i

700

l|)?l

800

d) (cb^?ya)
?y^a (o?yuzi)

This content downloaded from 163.1.255.60 on Tue, 21 Apr 2015 09:42:41 UTC
All use subject to JSTOR Terms and Conditions