THE DIACRITICAL DOTS

AND THE DEVELOPMENT OF THE

ARABIC ALPHABET
By E. J. R E V E L L

The alphabet from which the Arabic, Hebrew/Aramaic and

Syriac scripts were developed was deficient not only in its failure
to represent vowels, but also in the fact that its letters were fewer
than the consonantal phonemes of any of these languages. These
deficiencies were eventually remedied not by the development of
new letters, but by the use of small signs - dots, strokes, or

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

miniature letters - placed above, below, or (rarely) between the
letters of the text. These signs are generally treated in separate
groups: vowel signs, diacritics (i), marking the strong/doubled
pronunciation of a consonant or its opposite, and diacritics (ii),
distinguishing the different consonants represented by a letter.
It is equally reasonable, however, to consider all these signs as
marking different features of the sound unit represented by the
letter. Diacritics (ii) distinguish the basic consonantal phoneme;
diacritics (i) give information on possible variations in its
quantity or quality, and vowel signs show it as "at rest" or "in
motion" with a particular vowel quality.1 The signs clearly all
derive ultimately from phonological analysis of spoken language,
and the little information available on their origin seems to
support the view that they should be seen as a single group rather
than as three. This paper attempts to show that all these signs are,
indeed, the product of the same set of ideas about speech sounds,
and that diacritics (ii) - henceforward the "diacritical d o t s " -
provide very interesting information on those ideas.
The letters marked by diacritical dots in the standard Arabic
alphabet can be divided, broadly, into two groups: I, those in
which the letter is marked by dots above or below to distinguish
each consonant it represents; II, those in which the letter is
marked by a dot when it represents one of a pair of consonants,
but is unmarked when it represents the other. The marking of
1
The •view that the Semitic alphabets were, in fact, syllabaries, is strongly
argued by I. J. Gelb in A Study of Writing (London, 1952), pp. 147-52. On
pp. 1 5 2-3 he lists other scholars who have held the same opinion.

178

r
4
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
fim, bd' and kbd' seems to unite both groups, but for present
purposes this problem can be ignored. The pairy/w x kbd' (Jbd')
is provisionally included in group I, and the pair bd' x kbd' in
group II. The pairs sin x sbin and fa' x qdf as marked in the
standard alphabet do not fit these two groups, but in early
manuscripts and inscriptions both pairs are marked by the
group I method: sin usually with three dots in horizontal line
below and sbin with the same arrangement of dots above ;fd' with
one dot below and kdf with one above (as found also in Western
texts).1 Consequently the groups can be listed as follows:
Group I: bd',yd'xnm, to*, tbd'; jimxkbd' (Jbd'); sinxsbin;
and fa* xqdf.
Group YL'.bd' x kbd';rd' x %dy;sddx dad; td' x ^d>;^n6i'ain xgbain.

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

The available manuscripts and inscriptions show that the use of
(most of) these diacritical dots was already firmly established in
the first Islamic century,, and remained remarkably stable. The
arrangement of the dots in horizontal or vertical line, or tri-
angular group, might vary, but their number, and their position
above or below the letter, remained generally the same. The chief
exception occurs with fd'xqdf, for which several different
systems of diacritics were used, but other cases of variation are
also found. It is assumed here that these variations derive from
originally independent local usage, which sometimes became
confused with, and was eventually supplanted by, the standard
convention.2 Such local usage was presumably motivated by
particular considerations, phonological or otherwise, but it is not
the intention here to attempt to bring these to light. The usage
described for groups I and II above is demonstrably early, and in
most cases was accepted as standard, so this investigation will be
1
Examples can be found in N. Abbott, The Rise of the North Arabic Script
(Chicago, 1939), pp. 38-9, and a fuller discussion in A. Grohmann, Arabiscbe
Papyruskunde ( = Handbucb der Orientalistik, erste Abteilung, Erganzungsband
n, erster Halbband), Leiden, 1966, pp. 95-6.
2
Variant usage is described in Grohmann, Arabiscbe Papyruskunde,
pp. 9; £ J. von Karabacek discusses the points onfd' and qdf, and attempts to
determine the period and area of the different systems, in Sitytmgsbericbte der
Akademie der Wissenscbaften in Wien, pbil.-bist. KJasse, CLXXXIV, 3 (1917), 25-8.
He is probably not right, but Grohmann's remark that all systems noted by
Karabacek appear in Egypt (From the World of Arabic Papyri, Cairo, 1952,
p. 85) does not rule out the possibility of different local origins for the
different systems. All three systems of Hebrew pointing were, after all, also
•vused in Egypt.

»
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
concentrated on it, leaving local variations for consideration
elsewhere.
The standard Arabic alphabet also makes use of other signs
which can be considered as diacritics on the same level as those
discussed above. These can all be classed in a group HI, which
would include ham^a, the dots on td' marbiita, and the mark in
final kdf. The motivation for these signs is obvious: the sign
itself indicates the sound required. Thus the mark which dis-
tinguishes final kdf from final lam is a miniature kdf, as is clear
in early manuscripts. The bam^a sign, used on the letters id7,
wdw andjjw' when they represent this consonant, is derived from
'ain, to which its sound was likened.1 The two dots which mark
bd* as td' marbiita are those of td' itself. Other such signs are widely
used in literary texts, where the use of unusual words and poetic
language increased the problems of the reader and made the

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

accurate distinction of the consonants more than ever a necessity.
For instance, a letter representing bd', 'ain or sad (which is un-
marked in the standard system) is shown to represent the sound
in question by a miniature form of the same letter, placed above
or below it. Thus the letter itself is used to characterize the sound
which it most commonly represents.2 These signs are, however, a
supplement, not an alternative, to the standard system, and
represent the latest layer of diacritics to be applied to the Arabic
alphabet.3 The introduction of these letter-form signs as conso-
1
E.g. by 'Uthman al-Dlnl, "The place oibam^a in a word can be tested
by 'ain. Wherever 'ain comes (in pronunciation), there is the place olbatmyt."
Kitab al-Naqt, ed. O. Pretzl in Ortbographie tmdPmktierungdes Koran (Istanbul,
1932), p. 150,1. 13.
1
There is an unusually complete example of such usage in N. Abbott,
Studies in Arabic Literary Papyri, in (Chicago, 1972), text 6 (p. 149). Here, as
elsewhere, the unmarked member of some group II pairs is given a sublinear
dot, as described by A. Grohmann, in Allgemeine Einfubrung in die arabiscben
Papyri (Wien, 1924), p. 72. Such signs were necessary because scribes did not
mark the standard diacritical dots consistently. A special sign (mubmal)
indicating that a letter was correctly left undotted eventually came into
standard use. The function of these "differentiating signs" (as Grohmann
calls them) is the same as that of the standard dots: to indicate the particular
consonant represented by the letter. The confusion in the marking of
slnjsbtn noted there is perhaps to be explained by the fact that the signs used
(a stroke, or one or three dots, usually above the letter) are (in some cases at
least) intended to identify the letter (which in an undotted text could be
easily confused with others) but not the particular consonant represented.
3 Thus B. Moritz remarks that td' marbiita appears to have been the last
letter to receive its dots - in the second century of the Hegira. 'Encyclopaedia
of Islam, 1st edition, art. "Arabia, d. Arabic Writing".
180
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
nantal diacritics, and also as signs for vowels and tashdid as in the
standard system is attributed to al-Khalil.1 We have here, then,
a clear case of a uniform approach to all three types of vowel and
diacritical signs which seems to support the suggestion, made
above, that they were not considered as marking phenomena
fundamentally different in nature. This is perhaps not surprising
in al-Khalil's situation, however, and our interest is mainly with
the earlier period.
The consonantal diacritics of group LI represent nothing more
than the easiest way of distinguishing two sounds represented by
the same letter. The letter is marked when it represents one of
the sounds, but not when it represents the other. No informa-
tion about the sounds themselves is given. Economy of effort
obviously requires that (in contrast to the group LTI method
where the sound marked by a diacritic must be that most charac-

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

teristically represented by the letter) in group II the less common
of the two sounds is represented by the marked form of the letter.
This is a simple and effective method, but in the difficult literary
texts it was found necessary to supplement it with signs from
group i n , particularly since scribes often failed to mark the
standard diacritical dots.
Once the group has been identified, the motivation for the use
of consonantal diacritics in group I also seems quite clear. It is
based on the physical production of the sounds represented,
following the view that progression from the back of the mouth
to the front was progression downwards. A sound which is
articulated towards the back of the mouth is "high" (and so
marked with a supralinear dot) in contrast with a sound articu-
lated towards the front of the mouth, which is " low " and marked
with a sublinear dot. This view is best known from the descrip-
tion of the vowel system of Hebrew by Saadya Ga'on and from
its use in the Masora. It is also, however, the basis of the East
Syrian system of vowel signs, and of the system taken over by
the Arabs, as is shown by the vowel names nasb "raising" and
khafd "lowering". 2 In group I, as in group HI, then, the dia-
critics are based on the consonantal sounds they represent.
The application of this view to the consonants of Arabic was
1
See N. Abbott, Studies in Arabic Literary Papyri, ni, 7.
2
Arab grammarians offered other explanations. See E. W. Lane, An
Arabic-English Lexicon, Book 1, part 8 (London, 1893), p. 2799, col. 2, but it
seems more likely that the names were taken from Syriac qtqafa "raised"
and rebasa "depressed" (or similar terms) along with the signs.

13 181 SS20H
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
probably based on an arrangement of the alphabet such as that of
al-Khalil. He lists the Arabic consonants in linear order according
to their position of articulation (makhraj), starting at the back of
the mouth and working forwards.1 An arrangement of this sort
would allow the distinction of any two consonants on the basis
of their makhraj. The type of presentation more typical of Arabic
grammarians (and also used by al-Khalil), which lists "articu-
lation groups" of several consonants produced (in different ways)
from the same makhraj could only serve as the basis for the
distinction of members of different articulation groups. However,
an arrangement of this sort could have been the basis of the
rather small number of distinctions marked by the group I
diacritical dots (see note 2 below). Al-Khalil also shows the basis
for the use of sublinear dots to mark yd'. For him, yd' is not a
palatal consonant, but is one of the hawd'iya, which are considered

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

not to have any makhraj within the mouth, and so to be really
"extra-buccal"., Hence yd' is placed at the end of al-Khalil's list,
as even further forward than the labials. The diacritical dots were
clearly based on the same view. The dental/alveolar nun, td' and
thd' are marked with supralinear dots as "high" or "back"
sounds in contrast to the "bilabial and beyond" bd' andjw'.
Similarly pharyngeal khd' Qhd') contracts with palataljirn, palatal
shin with dental sin, and palatal qdf with bilabial fd'. In this
respect, then, the view of the Arabic consonants presented by
al-Khalil was the same as that which had been used, some time
before, as the basis for the group I use of diacritical dots.
One question must still be considered. Why are some pairs of
consonants distinguished by the group I method when the
simpler method of group II would have sufficed? The problem is,
indeed, emphasized by the fact that sin and shin, which were
originally distinguished by the group I method, are, in the
standard alphabet, members of group II (with a supralinear sign
composed of three dots, instead of one). It seems probable that
the group I method was used only to distinguish consonants
widely different in makhraj.2 Thus the makhraj of palatal jim is
1
This list is given in the introduction to the Kitdb al-'Ain (ed. A. Darwish,
Cairo, 1968, pp. 53, 65). A form of this introduction is translated by J. A.
Haywood in Arabic Lexicography (Leiden, i960), pp. 28-37. Al-Khalil's
views on phonology are discussed by S. Wild in Das Kitdb al-'Ain und die
arabiscbe Lexicographic (Wiesbaden, 1965), pp. 29-35.
2
It is, then, possible that only sounds occurring in different articulation
groups were distinguished, but no known list of articulation groups fits this

182
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
distant, and quite distinct, from that of pharyngeal hd' and khd'.1
•Similarly the dental/alveolar makhraj of nun, to* and thd' is
distinct from the "bilabial and beyond" position of bd' and jd'f
and so also are the positions of qdf and jd\ and of sin and shin.
In contrast, the pairs of sounds distinguished by the group II
method are articulated in similar or identical positions. Thus the
points at which hd' and khd' are articulated are very dose, and the
same is true for 'ain and gbain. The positions oitd' and £<?', and of
sad and dad are less close - even quite far apart according to
some - but they are pronounced in the centre of the mouth,
where obviously distinct articulators are lacking, so the failure to
treat the members of these pairs as distinguished by a major
difference in makhraj would not, in any case, be surprising.
A similar explanation can cover rd' and %ay. These sounds are,
indeed, far apart in al-Khalil's list. For Sibawaih, however, the

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

articulators of rd' and %dy are very dose, 3 and in the standard
Hebrew listing the two sounds occur in the same articulation
group (see below). If this view is correct, then, it would seem
that group I, based on major differences in makhraj, represents the
first attempt to develop an alphabet suitable for Arabic from the
sixteen or so letters available. Group II, making finer distinc-
tions, represents a second stage in this work, and group DI
represents the finishing touches.
As has been pointed out, the ideas on which al-Khalil based
his arrangement of the alphabet are of Indian origin, and are
exemplified in the order of the Sanskrit alphabet.4 This must also
situation. The closest is the Hebrew list given below, but there sin and shin
are placed in the same group.
1
This fits the assumption, made above, that initiallyj7m was distinguished
from bd'\kbd', and that the distinction of bd' from khd' was introduced later.
2
Al-KhalU assigns nun, to" and tbd' to three different articulation groups,
but in the Hebrew and Syriac lists given below they belong to the same
group. It is assumed that originally the letter was marked with a supralinear
dot when it represented nun, td' or tbd', and with a sublinear dot when it
represented bd' otjd', and that further differentiation came later.
s Zdy is articulated with the tip of the tongue, rd' with the part a little
behind the tip. The other articulator "above the incisors" is the same for
both. See Wild, Das Kitdb al-Ain, pp. 92-3. The descriptions attributed to
al-Khalil are given for comparison, and suggest that for him also the makbraj
of rd' was in fact close to that of %dy.
4
See Wild, Das Kitdb al-'Ain, pp. 37-40. He concentrates on al-Khalil's
articulation groups. J. A. Haywood, Arabic Lexicography, p. 37, points out
the dependence of al-Khalil's arrangement of the alphabet on Indian sources,
and that, in the present context, is probably equally significant

183 13-2
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
be true of the ideas on which the diacritical dots are based. Arabic
tradition states that the diacritical points were borrowed from the
Syrians. This cannot be true of the signs themselves, since the
requirements of Arabic were quite different from those of Syriac.
If the tradition is based on anything more than the mere fact that
the Syrians had an older tradition of written literature than the
Arabs, it must refer to the theory on which the use of the points
was based. There are, in fact, some features of the Syrian
pointing - and also that of Hebrew - which might derive from
this same Indian influence. It is possible, then, that the Arabic
tradition reflects the fact that Indian ideas were transmitted to
the Arabs by Syrians or Jews.
The only two consonants which are distinguished by diacritics
in the Syriac alphabet as normally presented are resh, marked with
a dot above the letter, and ddlath, marked with a dot below the

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

same letter. In addition, however, the six letters bgdkp t can re-
present either stops or fricatives. The stop is represented by a dot
above the letter, the fricative by a dot below. The signs with the
corresponding function in some Hebrew MSS with Palestinian
pointing can be interpreted as pointing upwards {dagesh, marking
the stopped pronunciation), or downwards (rafe, marking the
fricative), although both signs are placed above the letter. The
corresponding signs in the Tiberian pointing do not visually
suggest the idea of "above" or "below", but in some contexts
in the Masora dagesh is a synonym of milh'el "above" and rafe
of milhra' "below". 1 The terms milh'el and milhra' are most
familiar to Hebrew scholars as designations of accent or stress
position, where milhra' "below" designates stress on the final or
"left-hand" syllable, while milk"el "above" designates stress on
the penultimate - or, in a two-syllable word,; the initial or "right-
hand" syllable. One case is known in a Tiberian Masoretic text
in which, in connection with these terms, dots are used above the
syllables in question to mark the position of the stress.2 Dots are
similarly used to distinguish sin from shin. A dot above the left-
hand side of the letter (corresponding to the left-hand accent
position, milhra' "below") marks sin, and a dot above the
right-hand side (and so milh'el "above") marks shin. As far as
I know, the terms milhra' and milh'el are not used in connection
with the dots on sin and shin, but this interpretation of the graphic
1
See S. Morag, "Some Aspects of the Methodology and Terminology of
the Early Massoretes " in Lesbonenu xxxvrn (i 973-4), 68-9.
* See ibid., 63, note 35. e?>
184 J
I
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
symbolism is supported by the fact that, in at least one MS with
Palestinian pointing, the sign marking sin is placed below the
letter, that marking shin above.1
In addition to these features of the pointing systems, which, at
least in the case of the Syriac items, were in use in the early
seventh century, there is a certain amount of relevant information
in grammatical writings. The consonants of Hebrew are tradi-
tionally classified in five groups, according to position of articu-
lation, and described, starting with those produced farthest back
in the mouth, as letters of the-throat' h h', palate gy k q, tongue
dtln/, teeth %ssrs and lips bwmp. This list goes back to the
eighth century or before.2 Similarly the Syrian grammarian Elias
of Soba (c. A.D. iooo) lists the consonants - also in five groups -
"in order according to the natural pronunciation of the conso-
nants" ' h hc gkq \wy I \^ss Ir : dttn :b m p.3 It is inter-

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

esting to note that this grouping is not fully consistent with the
more detailed description of the articulation of the consonants
which precedes it, and which is undoubtedly based on Arabic
models. This listing shows that the Syrian grammarians felt it
1
See Revell, Hebrew Texts with Palestinian Vocalisation (Toronto, 1970),
p. 87 (TS 16: 96). The sign marking sin is placed below the letter in various
other MSS, but (at least in the small fragments available for study) the
method used in these must be classed with group II (shin has no diacritic) or
III (samek and shin are used as diacritics), and in these groups the position of
the sign is not significant.
2
The earliest known source for this list is the Sefer Yesira, which is
generally dated to the eighth century, though placed earlier by some. Allony
also dates this list to the eighth century, on the assumption that the phono-
logical knowledge on which it is based, which must have originated in
India, could only have come to the Jews through the Arabs (see Sinai LXXIV,
1974, 49 f.). However, it is possible that the knowledge could have come
through other channels, and it seems unlikely that the Sefer Yesira contains
the first formulation of the list. The work is concerned with cosmology and
cosmogony, not language, and the groups are arranged not according to their
positions of articulation, but to suit the author's purpose, with their initial
letters in alphabetical order. (See the opinion of Saadya Gaon in his comment-
ary on Sefer Yesira iv. 3 in M. Lambert, ed., Commentaire sur le Sefer Yesira...
par le Gaon Saadya ( = Bibliotbique de I'Scole pratique des bautes itudes, Sciences
pbilologiques et bistoriques, fasc. 8j, Paris, 1891), text p. 75. Consequently it
would seem that the list must have been produced by the eighth century, and
could have been produced earlier.
3
See A Treatise on Syriac Grammar by Mari Elia of Sobba, ed. R. J. H.
Gottheil (London, 1887), text p. 7. Despite his assertion, the order of the
letters within the groups clearly owes something to the alphabetical
order.

185

\
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
not only important to describe the physical production of the
sounds of their language, but also to arrange them in linear order
according to the point of articulation, just as was done by
al-Khalil. The order in which the sounds are given shows a
number of differences from that of al-Khalil, however, so it
seems quite possible that Elias is here presenting an independent
Syrian tradition, not one derived from the Arabs. Both Syrian
and Hebrew lists are divided into five groups, and in the case of
the Hebrew list not only the number, but even the names of these
articulation groups recall the five major groups of Indian con-
sonants: velar, palatal, "lingual", dental, labial.1 Allony has
pointed out that this list is based on ideas originating in India,
but considers that the Jews acquired them from the Arabs. This is,
of course, possible, but by al-Khalil's time there was already a
long history of communication between the Near East and India

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

so it is also possible that the Hebrew list, and also the Syriac, was
based on Indian thought which did not come by way of Arabic
scholars.
According to Elias of Soba, the Syrians pronounced resh
further back in the mouth than dalet. This would justify its being
marked as "higher" by a supralinear diacritical dot, in contrast
to the "lower" dalet which had a sublinear dot. This appears to
be a coincidence, however, as the earliest Syriac writing uses the
dot only with resb, and not with dalet, following the group II
1
These terms-save for "lingual"-are those used by W. S.Allen in
"Phonetics in Ancient India (London, 1953). The "velar" series is said by the
Indians to be pronounced at the "root of the tongue" (Allen, p. 51) and the
Hebrew "throat letters" "at the far end of the tongue" in the Sefcr Yesira
(sof) and Saadya's (somewhat expanded) translation ('aqsa) (ed. Lambert,
p. 74). The usual Indian term for the third series is obscure, giving rise to
terms like "cerebral" and "domal" in European grammars. Allen calls it
"retroflex" because it is characterized by retroflection of the tongue rather
than by a place of articulation (Allen, p. 5 2). The same consideration (presum-
ably) has given rise to the term "lingual" in some European grammars.
Wild suggests that the term inbiraf applied by al-Khalil to rd', lam and nun is
simply a translation of this Indian terminology (Das Kitdb al-'Ain, p. 32, n. 27
and p. 40). It would not be surprising if other Near Eastern scholars, also
copying the Indian tradition, should also base the name for their third group
of consonants on this tongue movement, for the tongue is, after all, an
articulator, and the traditional Indian name does not seem to indicate a place
of articulation. Wild sees al-Khalil's articulation groups as derived from the
same Indian tradition, and cites his listing of eight groups as indicative of this.
However, these eight groups exclude the four bawa'tya, while the eight
Indian groups include the semivowels. In some respects, then, the Hebrew
listing seems closer to the Indian than does al-Khalil's.
186
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
method already found in the Nabataean alphabet.1 The available
information certainly does not suggest that the six stops bgdkpt
were pronounced farther back, or "higher" than the corres-
ponding fricatives, so the position of the dots distinguishing stops
from fricatives cannot depend on position of articulation.2 The
traditional listings do not, however, include both stops and
fricatives, and the interest in their different positions of articu-
lation was certainly a product of Arabic influence. However, in
the Sanskrit alphabet stops are listed before other varieties of
consonant (aspirates, nasals), and so would be "higher", and the
Syrian and Hebrew traditions may have imitated this. The
marking of sin and shin in Hebrew could possibly have been
copied from Arabic, but if this were so one would expect the
Jewish scholars to copy the Arabic signs as well as the idea. It
seems more likely that both Jews and Arabs derived the idea

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

from a common source - the knowledge of the Sanskrit alphabet
and its phonological basis which had been brought to the Near
East from India.
It is possible that the view common to Jewish and Syrian
grammarians that the back vowels were "high" in contrast to
"low" front vowels also reflects Indian influence. This idea seems
to be symbolized graphically by the use, in Syriac, of a supra-
linear dot to distinguish a word characterized by a vowel pro-
duced towards the back of the mouth from a homograph charac-
terized by a vowel produced more towards the front, which was
marked by a sublinear dot.3 However, according to Jacob of
Edessa, who gives the earliest description of the basis for this
classification of vowels, the back vowels are "hard" or "thick"
('be' otpte'), while the front vowels are "thin" or "clear" (qatin
or nqed), and the same terms are applied to consonants, where
1
The dot below dalatb was still optional in the fifth century. See J. B.
Segal, TbeDiacritical Point andthe Accents in Syriac (London, 1953), p. u , n. 3.
* For Bar Hebraeus the two pronunciations of gdkt were evidently
(much) the same as Arabic j gb, ddb, k kh, t and tb, and the stopped pro-
nunciation of b and^> was bilabial (as Arabic b), while the fricative pronuncia-
tion was labiodental (as Arabic/). See his description in Le livre des splendeurs
(ed. A. Moberg, Lund, 1922), pp. 193-4 (translated, ibid., Buck der Strablen,
part 2, Leipzig, 1907, pp. 7-8). The situation, at least for d, k and /, was
evidently the same for Saadya. See M. J. De'renbourg in Journal Asiatique,
6eme s&ie, xvi (1870), 518.
* But note that it was not necessary to mark o/u vowels in this way, so
that %eqafa (/) was the farthest back of the vowels to be so marked. See
J. B. Segal, The Diacritical Point, pp. 20 f.

187

\
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
they cannot refer to position of articulation.1 Consequently there
is no reason to suspect Indian influence. The listing of the
vowels in the order in which they are produced, "descending"
from the back to the front of the mouth appears only in treatises
by Saadya Gaon and later writers, and in certain Masoretic lists.2
These lists distinguish homographs in the same way as described
for the dots in Syriac. A word characterized by a back vowel is
described as "high" {milh'el) in contrast to a homograph
charactemed by a vowel produced more towards the front, which
is "low" (mi/Ura'). The distinctions made in this way are quite
narrow, e.g. holem is distinguished from qames, and qames from
patah, and it does not seem possible that they could have been
based on any phenomenon other than the position of origin of
the vowels in the mouth, envisaged in much the same way as
described by Saadya. This view of the vowel system was not

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

typical of the early Jewish and Syrian grammarians,3 and is
therefore likely to have derived from foreign influence - ulti-
mately Indian.
1
For the terms applied to vowels, see G. Phillips, ed., A Letter by Mar
Jacob Bishop ofEdessa on Syriac Orthography (London, 1869), text p. 14 and for
consonants see A. Merx, Historia artis grammaticae apudSyros (Leipzig, 1889),
text p. 78. g and d are "thick", k and t "medium", and q and / "clear".
The terms presumably refer to some quality such as "sonorousness" which
was recognized as present ing, dand back vowels to a much greater extent
than in q, t and front vowels. However, it is significant for this paper that
both vowels and consonants were classified on the same basis.
2
See the edition of Saadya's treatise on the vowels by S. Skoss in J.Q.K.
XLII (1951-2), 283 ff., and for the "descending" (hatta or habata) of the
vowels in order from back to front, pp. 302 ff. A Masoretic list of the sort
described is given in S. Frensdorff, Okhlah we-Okblab (Hanover, 1864), list
no. 5. The order of vowels on which these lists from the "Tiberian" Masora
is (from "high" to "low") uoiacei. Lists of vowels ascribed to "the
Tiberians" begin with ; o r « and end with 0 or u, although the order of
individual vowels varies. See Allony, Beth Mikra m i , 2 (1973), 251 (lists
ascribed to the Tiberians) and Lesbonenu xxrx (1968), 140-2 (treatise ascribed
to Moshe ben Asher). Since the question of Arabic influence on Hebrew has
been raised, note that the vowel listing showing Arabic influence is the tri-
partite one: rum ( = raf) 0, u, matta ( = kbafd) e, i and lebassiv ( = nasb) j , a, c.
See e.g. Die Diqduqe ha-Te'amim, ed. S. Baer and H. L. Strack (Leipzig, 1879),
no. 36, and cf. Journal Asiatique, 6eme se'rie, xvrc (1870), 364.
3
For Jacob of Edessa's order see A. Merx, Historia artis grammaticae, text
pp. 80 ff., where noun bases are listed with characteristic vowels in the order
a, e, i, I, 0, u, u (with Bar Hebraeus' names, petdhd, rebdsd, hebdsd arlkd, rebdsd
karyd, 'esdsd arlkd, 'esdsd karyd). The MS is broken at both ends of this list,
but it is highly probable (considering other lists of Syriac vowels) that 3
(&qafa) stood at the beginning. Thus this is probably based on the Greek
188
i
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
The method of distinguishing homographs used in the
Masoretic lists described must have originated before the intro-
duction of vowel signs, and so must go back at least to the sixth
century. It seems likely, then, that some knowledge of Indian
phonological science - at least the knowledge of the traditional
order of the Sanskrit alphabet and its phonological basis - reached
the Near East in the sixth century or before. This was combined
with the rather limited views on phonology already held there,
and gave rise to the idea that sounds produced towards the back
of the mouth were "high" in contrast to those produced towards
the front. In Hebrew and in Syriac, this concept was applied
mainly to the distinction of vowel sounds, because in these
languages the consonantal phonemes were adequately repre-
sented in the script, so that homographs generally had to be
distinguished on the basis of their characteristic vowels.1 In

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

Arabic, however, the number of consonants was nearly twice as
great as the number of different letters in the script (designed for
Aramaic or Syriac) which was being used to write it, so that the
immediate requirement for the distinction of homographs was
some method of identifying further consonants. The theory of
"high" and "low" sounds was applied to this purpose, and the
group I method evolved. In this first stage, apparently, only
broad distinctions in position of articulation were indicated, so
that a number of cases remained in which two or even three
consonants were represented by the same letter. The use of the
group II method to distinguish these evidently represents a
second stage in the development of the Arabic alphabet. The
differentiation of nun, td' and thd', and of bd' andjw', by different
numbers of dots was probably also introduced at this stage. The
extra dots seem to have been added according to the order of the
abjad (or the Hebrew/Syrian alphabet). The whole process was
already complete in the first Islamic century, when most if not
all of the diacritical dots are found in use.
In contrast to the Syrian/Jewish theory which gave rise to the
group I method of distinguishing consonants, al-Khalil describes
sounds produced towards the front of the mouth as "higher"
order with i and a corresponding to alpha, i to ita, t to iota, and u and u to
upsilon. Other lists give the Syriac vowels in various orders, but typically
begin s a e or ) e a.
1
The basic concern of early efforts to ensure correct reading of written
texts was to distinguish homographs. The attempt to indicate pronunciation
accurately represents a later stage. Cf. Morag, Lesbonenu xxxvin, 63 f.

189

-. z.
 DIACRITICAL DOTS AND THE ARABIC ALPHABET
than those produced towards the back.1 It is probable, then, that
his ideas derive from a different source, perhaps closer to the
Indian origin of the material. The group I method of distin-
guishing consonants must have been the work of Christians or
Jews who began to use Arabic in literary texts, very likely in Iraq
or Syria.2 Once the theory had served its purpose, it was likely
forgotten, and never passed on to adherents of Islam. The
diacritical dots based on this theory remain in use, however, a
testimony to the accurate analysis of speech sounds by the
scholars who began the development of the letters used for
Aramaic or Syriac into the Arabic alphabet.
1
Arfa' (Kitabal-'Ain, ed. Darwish, p. 64). This might derive from the view
(held by the Indians, Allen, Phonetics, p. 21, but obviously likely to be arrived
at by anyone considering the problem) that the airstream which produces
speech sounds rises through the body. It is, however, noteworthy that

Downloaded from http://jss.oxfordjournals.org/ at Duke University on October 1, 2012

Saadya also held this view (see Skoss, J•£)•&. XLH (1951-2), 292-3).
2
For the probable literary use of Arabic in pre-Islamic times, see N.
Abbott, Studies in Arabic Literary Papyri, n (Chicago, 1967), 5.

190

r'
c