T(h) in Libyan and Canarian place-names

              Almogaren XX/ 1 / 1989 Hallein 1990 32-41
Lionel Galand
'T(h)" inLibyan and Canarian place-names*
1. The problem of "T(h)-" in place-names
1.1. Many place-names in ancient North Africa have an
initial "T-" or "Th-" and, since the accepted opinion is that the
language then spoken was the old form of Berber, one is temp­ted
to compare such names with Berber feminine nouns, most
of which bear the mark "t-". The same can be said of Canarian
place-names, and initial 'T-" might be and indeed has been taken
as evidence of a connection between Canarian and Berber (Vycichl
1987: 52).
There is no reason to stop at this point and, considering
the existence of place-names with initial "A-" in ancient North
Africa andin the Canary Islands, it may seem legitimate to draw
a parallel between those names and Berber masculine nouns
in "a-". And again, the comparison has actually been made (Pey­ras
1988: 190 ff.).
1.2. I do not intend to object to the principle of such pro­cedures,
which certainly are a means of exploring the mysteri­ous
fields of Libyan (i.e. the language or languages of ancient
North Africa) and Canarian. But I think it necessary to keep in
mind the whole situation in which the phenomena appear, so
that the information they are supposed to yield may be proper­ly
evaluated. The object of the present paper is therefore to sketch
the problem of gender in Berber nouns, in order to ascertain to
what extent Berber data can help us analyse Libyan and Cana­rian
names.
1.3. Two preliminary remarks must be made:
32
a) Place-names are perhaps the most conservative part of
'the linguistic material, so that, even if Libyan really is old Ber­ber,
a place-name in ancient North Africa may be pre-Libyan
{pre-Berber), since prehistorians know of the presence of po­pulations
prior to Berbers in the area. Why, then, should every
place-name be referred to Berber?
b) The gramatical gender of ancient African place-names
(assuming that the language or languages to which they belong
did make use of genders, and disregarding the probable exi­stence
of dialectal differences) remains unknown in most ca­ses,
if not all. What can be concluded from the treatment those
names received in Latin, for instance? If that treatment means
anything at all, then we have to discard a name like T(h)uburbo
from the feminine group, since the full form is T(h)uburbo Maius
(neuter !), not" *T(h)uburbo Maior". Canarian place-names rai­se
similar problems. There is a risk of circularity in arguing at
the same time that a name in "T-" is feminine because it looks
like Berber feminine nouns, and that Libyan or Canarian are
akin to Berber because they display the same marks of gender.
2. Nouns in Berber; an outline
2.1. Berber nouns, including modern place-names, can be
classified in three groups: Group I is rather heterogeneous and
would bebest defined as consisting of those nouns which be­lang
neither to Group II nor to Group III. Some are relicts of
an ancient vocabulary: fad "thirst", laz "hunger", mddn "people".
Others refer to close kinship: baba "my father", gwma "my bro­ther",
etc. A place-name like Mrrakws "Marrakech" could be
assigned to Group I.
Group II is represented by a large number of nouns with
an initial consonant "1-", which is the Arabic article. They have
been borrowed f rom Arabic ( or through Arabic) and they re­tain
the article once for all, w hatever function they fulf il: lqayd
33
"chief ( of a tribe )", lflus "money", lxdmt (f em.) "work", etc. Be­fore
certain consonants, the article "1-" is subject to phonetic
assimilation, as it is in Arabic: ·1suq > ssuq "market", a word to
be found in place-names too.
Group III includes the bulk of nouns, those which are
considered as representative of Berber morphology. But they
need not be genuine Berberwords: manywere borrowed from
another language and received a Berber form. Masculine nouns
of Group III have an initial vovel: adrar "mountain", agadir (pro­bablyfrom
Phoenician origin) ''wall" or ''fortified collective grana­ry",
itri "star", etc. Feminine nouns in the singular have "t-" be­fore
the vowel and also, in most cases, a fi_nal "-t": tafukt "sun ",
tilist "fleece" etc.
Comparison with other Hamito-Semitic languages (Ara­bicfor
instance) shows that in an earlier stage final "-t" was the
proper mark of gender (Cohen 1988: 20), a markwhich is used
to point to a range of various specifications, and not only sex.
In Berber, initial "t-" probably appeared before the nominal stem
as apart of "ta", feminine of ''wa" (and without explicit masculi­ne
mark: "a", a pronoun which often became the nucleus of
demonstratives, but was also used as a kind of article and final­ly
stuck to the noun, thus forming with it a single word: ·a-drar
> adrar, ·ta-fukt > tafukt, etc. The hypothesis was elaborated
by Vycichl ( 1957), and there are good reasons to accept it, so
that we can consider initial "t-" as diachronically second to fi­nal
"-t".
There are facts, though, which do not fit in and have tobe
accounted for. Final "-t" is missing in the plural and sometimes
also in the singular: tigwmma "houses", tilasiwin "fleeces", and
tamgra "harvest", tizi "pass (through mountains)", etc. To ex­plain
the plural forms, Basset (1952: 23) suggested that Ber­ber,
in aformer stage, did not oppose genders in the plural (1),
but, even if we accept his opinion, we are still faced with the
problem of those nouns that have no final "-t" in the singular.
34
Basset observed that "-t" does not occur when the stem ends in
· "-a" or "-i", but this is not a rule, as is shown by taydit "bitch",
and I had rather put things the otherway round: feminine nouns
without final "-t" end in "-a" or "-i". Furthermore, with rare ex­ceptions,
those nouns are isolated, I mean that they are not opposed
to masculine forms, whereas taydit ( not *taydi) is the feminine
of aydi "dog". The situation might lead to the conclusion that "­t"
disappeared under certain phonetic conditions, say after a
vowel, unless an opposition had tobe maintained with a mas­culine
form. But of course this is rather guesswork.
Whatever lacunae subsist, I think that any comparison
between Libyan or Canarian and Berber should allow for the
present situation in Berber and its historical background. Nay,
comparing a "t" with a "t" would be meaningless if we failed to
consider the frame in which they appear.
3. Libyan and Canarian data: a discussion
3.1. Data (2) from ancient North Africa (ANA) and from
the Canary Islands (CI) will be discussed at the same time for
the sake of brevity, but that does not imply the recognition of
any privileged relation between those areas. Now, what do we
observe in the onomastic material?
-Place-names in "T(h)---t(vowel)", which might correspond
to modern nouns in "t---t": ANA Thinissut, Tigit, Thagaste,
Thelepte, Theueste; Tupusuctu, etc.; - CI Taburiente, Tacote,
Tagomate, Tagojaite, Tagarajita, etc.
- Place-names in "T(h)-", without a final "-t": ANA Taca­pe,
Thabraca, Tipasa, Thugga, T(h)uburbo-Maius, etc.;-CITa­ganana,
Tagoja, Tamanca, etc.
At first sight, then, the situation looks very much the same
as in Berber. But questions arise as soon as one proceeds with
further investigations.
3.2. Are those place-names feminine? - Neither Libyan
35
nor Canarian sources reveal the gender of place-names (as to
other sources, see § 1.3.b ). Personal names can be of some help,
on condition that we know whether a man or a woman is na­med.
lt then appears that forms with initial "T(h)-" are not ne­cessarily
feminine, as is shown by the names of Tacfarinas, the
famous rebel, or Taufia, a warrior (but also a place-name; Wöl­fe}
1965: 720). Even names in "T---t" may apply to men: Chabot
( 1940) registered fourteen such names, out of which one is f e­minine
(Tsblt, according to Fevrier 1956: 270), two (Tkdt, Tstt)
refer to men, one (Trlt) may be the name of a man; there is no
clue for the rest of them (3). Could one be satisfied with the
objection that a feminine noun or name may be transferred to
a man ( cf. French "La Fontaine")?
3.3. Are final "ta/e/i" rightly identified with final "-t"? Cha­bot
(1940) records four personal names in T---t', where '"" (H
in Chabot) probably represents a vowel. But again nothing pro­ves
that those were names of women. At any rate the presence
of the final vowel should be accounted for.
3.4. How is it that names with final "-t" are not so frequent
as might be expected for remote periods, when "-t", the genui­ne
mark of thefeminine gender (§ 2.l), should be - or so it seems
- much better exemplified? Some facts can help us to an ans­wer:
a) Names in "-t" do exist in ancient North Africa, mainly
in the West. Ptolemy the Geographer mentions many places or
rivers the names of which had a "-th" ending: Iagath, Akrath,
Thikath, Dorath, Assarath (river), etc. In a few case s, the en­ding
still exists today as a feminine mark: the rivers Molokhath
(Ptolemy) and Masath (Pliny) are called Mlwit ( < ·Mtuyt? <
·Mlukt ?) and Masst by the Berbers.
b) There are reasons to think that many names with final
"-a" might be allomorphs of names in "-at". Greek and Latin
transcriptions of Libyan names in"-( a)n" usually drop the last
consonant and have the ending "-a": a well-known example is
36
Massinissa, spelled Msnsn in Libyan andin Punic. But the pro-i
blem of "-t" is not merely one of transcription. Some names
apparently have two forms, one with "-t", the other without "-t",
sometimes in the same author. Pliny mentions the town Sala
and the river "eiusdem nominis" ("of the same name" / V, 5),
then calls that same river Salat (V, 9); Ptolemy says Sala (IV, 1,
2, p. 576: the town; IV, 1, 2, p. 579: the river, genitive) and Sala­ta
(IV, 1, 2, 576: the river, genitive ); there are other variants
(4). The name of the river Masath (Pliny, V, 9), today Masst in
Berber (see above ), takes the form Massa in Ptolemy (IV, 6, 2,
p. 731: genitive). The river Darad(os) (Ptolemy, IV, 6, 2, p. 731:
genitive) or Darat (Pliny, V, 9) is now named Dra. A few Cana­rian
examples might belong here: Taco / Tacote (Diaz Alay6n
1987: 145) and perhaps Tagoja / Tagojaite (147, 148).
Oscillations between 0 and t occur elsewhere. Tuareg verbs
of the type dal;;}t "tobe green" do not retain the consonant "t" in
all syllabic structures: dalätmät "you (fern.) are green" / dalim
(but dalätäm is not excluded) "you (masc.) are green". In the
Arabic dialects of North Africa, a feminine noun like mdina
"town" may become mdin;;}t in certain constructions (Man;;ais
1977: 166). There is a vague resemblance between those phe­nomena,
but each of them must be studied in the system to which
it belongs and no general statement can be made for the mo­ment.
3.5. When was the initial element "(t)a/i-" incorporated
in Berber nouns and when did it lose its previous function and
value (§ 2.1)? - Rössler (1958: 107), Chaker (1984: 269) and
Vycichl (1987: 52) are of opinion that the change had notyet
ocurred in Libyan times. If they are right as I believe they are,
then we must assume that Libyan names in "T{h)-" are phrases
with two constituents, a determinative and a noun. Shilha, in
Southern Morocco, has place-names (microtoponyms) with the
demonstrative ann "that"; the article is not uncommon in French
as like "Le Havre" (Galand 1980). So the case of Libyan names
37
would not be unique. Another possibilitywould be to take ''T(h)a/
i-" as the nucleus of a construct like Shilha "ta mlluln" Iiterally
"that (fern.) being-white" - "the one which is white" ( cf. the ancient
place-name Thamallula !). One could also quote Tuareg na­mes
of the type "In-" (masc.) or "Tin-" (fern.) "one of -" (Ga­land:
1986). If we suppose such names existed in Libyan, we must
also adrnit thatfor some reason theyfavoured the feminine gender:
forms in "A-/1-", though not unknown, are comparatively rare,
as Chaker (1984: 269) observed for personal names. Moreo­ver,
there are cases when one may be reluctant to split a name
into two words: such forms as Thala, Thugga impress as loo­king
very much like single words, not phrases, and a f ew pairs
can be found that remind of the present opposition of gen der
in Berber: J. Desanges kindly drew many attention to two of
them: Alele(n) (Pliny, V, 35: accusative) / Talalati (Itin. Ant.,
76, 7) and Balla (Pliny, V, 37) / Tabalati (Itin. Ant., 75, 3). On
the other hand, if the present system already existed in Libyan,
should not the number of Libyan names in "A-" or "1-", "Ta-" or
'Ti-" be much larger than it is?
As to Canarian, it raises similar difficulties. Vycichl ( 1987:
52), who regards Canarian as a Berber language, thinks that
the "article" (t)a/i-, still independent in Libyan, is treated as a
mere element of the word in Canarian. But the number of Canarian
nouns and names in (t)a/i-, though rather important, does not
seem to reach the expected level. Pairs like Tanausu / Atanau­su
(with a + ta !) induce tobe cautious when describing the
functions of "a" and "ta" (names quoted by Diaz Alay6n 1987:
118). Finally, attention must be paid to forms that look Berber
and are not, as Canarian 'Talisca" which, according to Diaz Alay6n,
is of Lusitanian origin.
4. Mine is by no means a negative attitude. The problem
of the relation between Libyan and Berber, or Canarian and
Berber, is extreemly complex. Even connections that seem evi­dent,
such as might be inf erred from the existence of names in
38
"T(h)---t", raise difficult questions which cannot be answered
1 without weighing pros and cons. But on the whole, doubt is more
rewarding than quickly accepted hypotheses.
Notes:
* An abridged version of this paper was presented to the Comission d'histoire et
d'archeologie de l'Afrique du Nord (Comite des travaux historiques et scientifi­ques),
Paris, on the 20th of February, 1989.
(1) On the occasion of the Journee d'etudes linguistique berbere, Paris, March
11th, 1989, W. Vycichl argued (if I correctlyunderstood), that afew Berber nouns
Iike tigattn "goats" combine the marks "t" (fern.) and "n" (plur.).
(2) African names are easilyfound in Chabot (1940), Müller's edition of Ptole­my
(1901), Desanges' edition and rich comentary of Pliny (1980), Lassere (1977).
Many Canarian names are listed in the very good book by Diaz Alay6n (1987)
and, of course, in Wölfel (1%5). There are many other sources. - The question
cf the phonetic value of "T-" and "Th-" in Greek and Latin transcriptions is not
dealt with here; the difference may be purely graphic or actually phonetic ( cf.
the variants d t in Berber); I assume that in any case a dental consonant is meant.
(3) In respect of those difficulties, it is interesting to note that Fevrier (1956:
271) wrote: "Je propose clone l'hypothese 'de travail' suivante. En libyque beau­coup
de noms propresfeminins n'auraient pas uneforme feminine."
(4) Variants with "th" or even with "g": see Desanges (1980: 96 and n. 7), who
speaks of an extension of the stem.
Bibliographical references:
BASSET, A. 1952 (repr. 1969): La langue herbere.Oxford Univ. Press
CHABOT, J.-B.1940: Recueil des inscriptions libyques. Paris
CHAKER, S. 1984: Textes en linguistique berbere. Paris
COHEN, D. 1988: Le chamito-sernitique. fn: J. Perrot, Les langues dans Je monde
ancien et moderne. 3e partie: Les langueschamito-semitiques, textes reunis
par D. Cohen. Paris, 9-29
DESANGES, J.1980: Pline l'Ancien, Histoire naturelle: Livre V 1-46, texteetabli,
tradui t et commente. Paris (Collection des Universites de France)
DIAZ ALA YON, Carmen 1987: Materiales toponimicos de La Palma. Sta. Cruz
de La Palma
FEVRIER, J.G. 1956: Que savons-nousdu libyque? Revueaf ricaine 100, 263-
273
39
GALAND, L. 1980: L'opposition defini-indefini en toponymie: exemples her­beres.
Melanges d'onomastiqe, linguistique et philologique offerts ä M.
Raymond Sindou T, 21-24
LASSERRE, J.-M. 1977: Ubique populus. Paris
MAR<;AIS, Ph.1977: Esquisse grammaticale de l'arabe maghrebin. Paris
MÜLLER, C. 1901: Claudii Ptolemaei Geographia, I, 2. Parisiis
PEYRAS, J. & TROUSSET, P. 1988: Le lac Tritonis et les noms anciens du
Chott el Jerid. Antiquites af ricaines 24, p. 149-204
PLI NY THE ELDER: see DESANGES
PTOLEMY: see MÜLLER
RÖSSLER, 0. 1958: Die Sprache Numidiens. Sybaris. Festschrift Hans Krahe.
Wiesbaden, 94-120.
VYCTCHL, W.1957: L'article defini du herbere. Memorial Andre Basset (1895-
1956). Paris, 139-146
VYCICHL, W. 1987: Les Herberes des Jles Canaries: elements historiques et
linguistiques. Etudes et documents berberes, 2, p. 42-62
WÖLFEL, D.J. 1965: Monumenta Linguae Canariae. Graz
"T(h)" in libyschen und kanarischen Ortsnamen
(Zusammenfassung von Helmut Stumfohl)
Professor Galand berührt eine grundsätzliche Frage, die
für die Beziehungen zwischen dem Altkanarischen und dem
Libyschen und damit für die Frage des Verhältnisses zum Ber­berischen
wesentlich ist: es ist die Frage, ob die zahlreichen
Ortsnamen mit/oder initialem und/oder finalem "t-" bzw. "-t"
( oder "th" in der Anlautposition) als weiblich aufzufassen sind.
Die altlibyschen Ortsnamen dieses Typs wurden zuerst durch
M. Gustave Mercier gesammelt (Journal asiatique 25, 1924, pp.189-
320). Eine ähnliche Parallele könnten die modernen berberi­schen
Ortsnamen darstellen, die mit "a-" beginnen. Wenn Li­bysch
eine Vorstufe des Berberischen ist, dann gilt für die Antike,
daß es auch vorberberische Ortsnamen unseres Typs geben kann.
Berberische Substantive lassen sich, besonders vom An­laut
her, in drei Klassen einteilen: 1) in heterogene, die keines
dieser Elemente enthalten; 2) in die mit "1-" beginnenden, worin
sich der arabische Artikel verbirgt, der fest geworden ist; und
40
3) die zahlreichste Gruppe, solche mit "a-"und mit "t-"bzw. "th-"
beginnende und/ oder mit "-t" endende (im Singular): t-afuk-t
"Sonne". Für die hamito-semitischen Sprachen gilt, daß "t-" im
Anlaut früh als weibliches Formans aufgefaßt wurde; im Ber­berischen
scheint diese Entwicklung später eingetreten zu sein:
ein ursprüngliches Pronomen "t-", "th-"wird über eine demon­strative
Stufe zum Artikel, der endlich fest mit dem Wort ver­schmolzen
wird wie auch der männliche in "a-". Dies führt zur
Frage, ob die antiken libyschen Ortsnamen so zu verstehen sind,
wobei die Frage unberührt bleibt, wie das grammatische Ge­schlecht
eines Ortsnamen überhaupt zu verstehen sei. Wir haben
etwa Thinissut, Tigit, Tacote etc. Der Name des Rebellen
Tacfarinas sollte uns warnen; Tkdt (nicht vokalisiert) ist ein
Männername.
Dazu kommt ein statistisches Element: Sollte vor allen
Dingen die Gruppe auf "-t" weiblich sein, dann wäre eine grö­ßere
Häufigkeit zu erwarten. Dazu kommen die Doppelformen,
wie Sala ( eine Stadt) und der dazu gehörige Fluß Salat (nach
Plinius). Die mit "t-" bzw. "th-" anlautenden Namen könnten auch
durch Zusammenrückung einer Fügung entstanden sein. Ein
weiteres Element des Zweifels ergibt sich aus einem abgelei­teten
statistischen Argument: Wörter mit sehr kurzem Wort­körper
wie Thala, Thugga zerlegt man ungern weiter: Sie ma­chen
einen ursprünglichen Eindruck. Ferner: Wenn Wörter in
"i-", "ta-ti" beginnend weiblich sind, sollten sie nicht schon im
Libyschen häufiger sein? Aber die antike Fügung "Thuburbo
maius" zeigt, daß man diesen Ortsnamen als Neutrum auffaß­te.
Sollte die betreffende Vorsilbe "t-" bzw. "th-" im Altkanari­schen
weiblich bezogen sein, warum ist die Zahl dieser Wörter
nicht größer? Dazu kommen Paare, wie Tanausu / A tanausu.
Galands scharfsinnige Überlegungen zeigen wieder einmal, daß
die Beziehungen zwischen dem Altkanarischen und dem Libysch­Berberischen
nicht einlinig, sondern komplexer Natur sind, sodaß
jede direkte Abgleichung fragwürdig bleibt.
41