Collection de la Maison de

l'Orient méditerranéen ancien.

Série philologique

Late and Vulgar Latin in Muslim Spain: the African connection

Prof. Roger Wright

Abstract
It seems probable that a considerable majority of those who came over from Africa to the Iberian Peninsula in the eighth
and ninth centuries were speakers of African Early Romance; the Western half of the North African seaboard had been an
important part of the Roman Empire, and although Berber languages were also spoken there it is likely that the common
language was Romance. And this African Romance would have been very similar to that spoken on the Northern side of
the Straits of Gibraltar; thus it is possible that it was one of the dialects that provided the input to the Ibero-Romance koine
which emerged in the Peninsula in the ninth century. Phonetic, morphosyntactic, lexical and semantic data thought to be
exclusive to African Latin do seem in the event to have contributed to the development of Ibero-Romance.

Citer ce document / Cite this document :

Wright Roger. Late and Vulgar Latin in Muslim Spain: the African connection. In: Latin vulgaire – latin tardif IX. Actes du
IXe colloque international sur le latin vulgaire et tardif, Lyon 2-6 septembre 2009. Lyon : Maison de l'Orient et de la
Méditerranée Jean Pouilloux, 2012. pp. 35-54. (Collection de la Maison de l'Orient méditerranéen ancien. Série
philologique, 49) ;

http://www.persee.fr/doc/mom_0184-1785_2012_act_49_1_3226

Document généré le 07/02/2017

 LATE AND VULGAR LATIN IN MUSLIM SPAIN:
THE AFRICAN CONNECTION

Roger Wright
University of Liverpool

abstract

It seems probable that a considerable majority of those who came over from
Africa to the Iberian Peninsula in the eighth and ninth centuries were speakers of
African Early Romance; the Western half of the North African seaboard had been
an important part of the Roman Empire, and although Berber languages were
also spoken there it is likely that the common language was Romance. And this
African Romance would have been very similar to that spoken on the Northern side
of the Straits of Gibraltar; thus it is possible that it was one of the dialects that
pro­vided the input to the Ibero-Romance koine which emerged in the Peninsula in
the ninth  century. Phonetic, morphosyntactic, lexical and semantic data thought
to be exclu­sive to African Latin do seem in the event to have contributed to the
development of Ibero-Romance.

This series of triennial international conferences has become a valuable and

enjoyable showcase for the presentation of research, ideas, information and data con­
cer­ning the topic that intrigues us all; the nature of Latin as a living language, as a
spoken language, as the first native language of millions of human beings from its
ear­liest recorded moments up to the Early Middle Ages. The combined knowledge
and experience of those who come to these meetings is unusually wide, because our
field of operations covers a thousand years and more, relying on data which extend
over wide areas of the Western Mediterranean and elsewhere, on archaeological evi­
dence such as inscriptions and defixiones, on textual evidence such as manuscript
errors and glosses, on the explicit comments of grammarians and other writers, on the
recon­structions made by the comparative-historical method, on the discoveries and
ana­lyses of historians working on Roman, Late Antique or Early Medieval history,
and on the theoretical evidence provided by modern advances in several areas of
linguistics, including pragmatics, historical linguistics, sociolinguistics, and indeed
his­torical sociolinguistics.

Latin vulgaire, latin tardif.indb 35 29/05/12 12:18

 36 r. wright

Our topic is Late and Vulgar Latin. “Late Latin” is usually taken to extend chro­
no­logically from the so-called “Classical” period to the time of the standardization
of Medieval Latin, whenever we think that happened, and was thus a living language
in use for many centuries; so-called “Vulgar Latin”, which is the name traditionally
applied to those linguistic features which we know existed but which were not
recom­mended by grammarians, existed throughout the whole Latin period. There is
a good case for re-naming “Vulgar” Latin, so-called in opposition to the traditionally
named “Classical” Latin, as simply “Latin”, a sophisticated and varying ensemble of
registers and styles that was spoken by everybody; if it still makes sense to establish
an opposition between this Latin and “Classical” Latin, “Classical” Latin is the
marked member of the opposition, being a recherché style adopted by a few people
in writing and perhaps nobody at all in speech. But this minority register was, as
we all know, the one that has supplied posterity with most of our direct data, and so
the data are not always easy to evaluate from our point of view as researchers into
spoken Latin. But such research can be done, as our conferences have shown, and
the results are unfailingly interesting and intriguing, which helps explain why our
mee­tings continue to be expanding in scope. When the late József Herman started
the pro­cess in 1985, there was a select gathering of some twenty invited specialists.
The num­bers have increased steadily through the years: at Oxford in 2006 there were
126  registered attenders, from 23 different countries, and 89 papers delivered, of
which 68 were submitted to the Actes, published with the usual title as Latin vulgaire,
latin tardif, VIII.1
The range of topics that were addressed there, and are addressed here in Lyon, is
impres­sive in itself. It is also indicative of a major difficulty which faces each of us
indivi­dually, and which can only be satisfactorily solved, if at all, collectively. To pre­
pare a definitive account, or even a survey, of Late and Vulgar Latin, a scho­lar would
need to have expertise in many different fields; in Latin language, in recon­struc­table
fea­tures of Early Romance and the several Romance languages, in at least the basic
aspects of historical linguistics, sociolinguistics, in textual and historical evi­dence of
many kinds from the Republican, Imperial, Late Antique and Early Medieval periods,
inclu­ding palaeography, calligraphy, and the nature of scribal trai­ning and customs,
and ideally in at least a basic knowledge of Greek and some awareness of the nature
of Etruscan, Oscan, Celtic, Germanic, Punic and even Basque. Nobody, of course, is
or could be an expert competent in all these fields at once, not even József Herman,
not even James Adams. All that most of us can hope to do is to become competent
in at least one or two of these interconnected scho­larly fields and then be aware, as
far as is practical, of progress made by others who are working in the related areas
of expertise. It is intrinsically important for us to meet and to talk and listen to each
other on occasions such as this, to be generous in sharing our knowledge with each
other, and to expect that such mutual illumination can lead to positive results.

1. Wright (éd.) 2008.

Latin vulgaire, latin tardif.indb 36 29/05/12 12:18

 late and vulgar latin in muslim spain 37

So I am going to consider here a question which exemplifies acutely the way in

which a confluence of expertises can be both necessary and valuable; a question
and a tentative answer. It comes directly from a consideration of one of the most
posi­tive facets of our meetings over the last twentyfour years, the fact that Latinists
and Romanists have been able to meet and interact and discover and share mutual
interests, and gradually become aware that some of those interests are actually clo­
sely related or even the same, as they, as we, look at the same data from different
perspectives. This becomes particularly noticeable in discussions of relative chrono­
logy. I can recall overhearing a conversation some time ago between a Latinist and
a Hispanist arguing over whether a particular linguistic phenomenon is “Late” or
“Early”, only for them to realize, after quite a long while, that they were in fact in
agreement; the eighth century is “late” to a Latinist and “early” to a Hispanist. These
terms are relative, naturally. Thus “Late” in the noun phrase “Late Latin” refers
roughly to the same period as “Early” in the noun phrase “Early Romance”, and those
of us who think that no such metalinguistic distinction could have been made at the
time, between Latin and Romance, similarly see the choice of the noun in these two
noun phrases as a difference of perspective on our part rather than as a dif­ference in
extra­linguistic reference. In general, as a result, the period between 600 and 840 A.D.
has not been well served by linguists of any perspective, since on the whole it has not
seemed to fall inside any prototypical scholarly sphere of reference. The implication
of the subject matter treated in Menéndez Pidal’s seminal Orígenes del español
(1972), for example, is that direct philological analysis only becomes relevant to the
study of Ibero-Romance in texts of the ninth, tenth and eleventh centuries, several
of which are stu­died in minute detail, and that earlier evidence is outside his brief.
Conversely, the recent book by James Adams, which is probably the most important
to have appeared in our field in the last three years,2 is explicitly entitled The
Regional Diver­si­fication of Latin, 200 BC - AD 600, with the implication that data
from later than 600 are not part of his brief; in turn this has had the consequence that
the post‑600 period seems even more inadequately explored now than it was before.
These books are not unrepresentative; Latinists in general tend to see the centuries
after 600 as not really their subject area, Romanists tend to see the years before the
Strasbourg Oaths as not really their subject area, and this has left a wide area of
uncer­tainty and loose ends from both directions which are not always straightforward
to tie together.
The gap has fortunately been filled, and these loose ends tied, to some extent, by
Michel Banniard, whose Viva Voce3 is a tour-de-force; and, it is good to see, recently
by a number of younger scholars, including most notably Marieke Van Acker,
whose excellent 2007 book is adventurously given a title in Late Latin, Ut quique
rustici et inlitterati hec audierint intellegant, subtitled in Romance Hagiographie

2. Adams 2007.
3. Banniard 1992.

Latin vulgaire, latin tardif.indb 37 29/05/12 12:18

 38 r. wright

et communication verticale au temps des mérovingiens (vii e-viii e siècles). And I

am going today to consider a facet of Early Ibero-Romance from that same period,
one which requires careful analysis from historical, linguistic and philological per­
spectives, both as regards Latin and as regards Romance. Not myself being an expert
in everything at once, this will take the form of the suggestion of a possibility rather
than the claim of a discovery.

The Late Latin, or, if you prefer, the Early Romance, of the Iberian Peninsula in
the seventh century was the most homogeneously monolingual of any area in the
for­mer Roman Empire. Basque was spoken there in a small area, but the influence
of Basque on Romance is minimal (but not non-existent4). Gothic had long since
stopped being spoken in the Peninsula, if it ever had been. In the seventh century,
Gaul saw influences from Germanic, Italy saw influences from Greek, Africa still
had the presence of Berber and Punic, and Britain had largely lost Latin as a spoken
language, but in essence the Peninsula was monolingually Latin-speaking. It used to
be thought that Spanish Latin was in some sense relatively archaic, but Adams esta­
blished beyond much doubt that the features which once led modern ana­lysts to that
conclusion are probably better explained in other ways, such as being symptoms of a
higher level of scribal training in the Peninsula than elsewhere. Isidore of Seville is
the best known of the scholars of the age, but there were many others. When Isidore
uses first-person verbs such as dicimus, or first-person adjectives such as nostra (as
in the phrase nostra lingua), he is undoubtedly referring to the Latin of his time
and place. I call it Latin here, rather than Romance, since that is how he referred
to it himself. The sources of inspiration and information in the linguistic sections
of Isidore’s work include North Africans, some of whom he may have known
personally; Roger Collins5 has proposed, plausibly in my view, that one important
sti­mulus to the revival of intellectual life in the Iberian seventh century was indeed
the arrival of scholars from North Africa. Literacy was not confined to an élite,
however; in earlier years it proved tempting for modern investigators to see literacy
in seventh-century Iberia as confined to the top of the social pyramid, but this view
is no longer tenable after the discovery and analysis of what have come to be called
the “Visigothic slates”, although there is nothing “visigothic” about them at all. These
have been found, and occasionally are still being found, in the rural provinces of Avila
and Salamanca, where slate was sometimes a more readily available writing surface
than the tree bark, wax, or even imported papyrus that would usually have been used
instead (which are biodegradable, and have thus not survived). These slates show that
people in rural areas with everyday concerns could not only read but also write, and
the idea of some kind of strict sociolinguistic apartheid between the educated and the

4. See Echenique Elizondo 2004.

5. Collins 2004.

Latin vulgaire, latin tardif.indb 38 29/05/12 12:18

 late and vulgar latin in muslim spain 39

rustic has disappeared with the slates’ appearance.6 In sum, seventh-century Iberia
was a sociolinguistically complex monolingual area of an essentially normal kind.
The African connection is intriguing, however. It has become customary, on at least
a subconscious level, for modern Romanists to see diatopic divisions of that age, if
there were such, in terms of subsequent political constructs; hence the terms Ibero-
Romance, Gallo-Romance, Italo-Romance, and what is usually called now “African
Latin” but which would probably be called “Afro-Romance” if a Romance lan­guage
were still spoken now along the southern Mediterranean coastline. The divisions
implied by these names are anachronistic when applied to the seventh century.
Dialect studies have shown that within a dialect continuum, isoglosses exist, often
as transition zones of greater or lesser width, but they feel no need either to bundle
together nor to follow political boundaries, unless they are the combined result of
exter­nal events such as large population movements. The Pyrenees, for example, look
to us like a natural boundary, but in fact there are no dialectal isoglosses running
West to East along the political border between Modern Spain and Modern France.
Many Romance features are common to dialects in both Gascony and in Navarra
and/or Aragón; many others are common to both Occitan and Catalan; in general
such isoglosses as there are in the mountains run North to South, not West to East.
Similarly, the evidence seems to suggest that the Straits of Gibraltar were almost cer­
tainly not the site of a bundle of Latin (Romance) isoglosses in the period between the
second century B.C. and the arrival of the Muslims in the late seventh century A.D.
It may seem natural for us now, in the light of subsequent historical events unfore­
seeable at the time, to perceive the Straits of Gibraltar as a huge cultural divide; but
it was not seen that way in those centuries. The shores of both sides of the Straits
had been ruled by the Carthaginians; the Western Mediterranean coastline of North
Africa was settled by Rome in the second century B.C. at the time of the defeat of the
Carthaginians, and that was itself an extension of the same process which had seen
the Eastern Mediterranean coastline of the Iberian Peninsula settled by Rome in the
years after the defeat of the Carthaginians there in 206 B.C. The subsequent spread
of Romanization and of Latin into Central and North-West regions of Iberia, and
Westwards and Southwards in Africa to reach the Atlantic and the Sahara, happened
at the same time as each other, and probably, from a linguistic point of view, in the
same sort of way; settlers, soldiers, merchants and others came from the Italian
peninsula, speaking varied dialects of Latin, but mutually comprehensible, and from
the interaction of the dialects of those settlers, in both Africa and Iberia, after two or
three generations, by the usual processes of koineization and accompanying inter­
dialects, an emigrant dialect is bound to have formed which was essentially similar in
both the Iberian Peninsula and the Western half of the African coastline.

6. Velázquez Soriano 2004.

Latin vulgaire, latin tardif.indb 39 29/05/12 12:18

 40 r. wright

This postulated koineization is a reconstruction based on the uniformitarian

principle, or in effect a guess, for which, as Adams points out,7 there is necessarily no
direct evidence, but which, as Adams also points out,8 is the kind of process which has
come to seem all but inevitable in colonial circumstances to the modern sociolinguists
who have studied such processes in more recent contexts. Peter Trudgill’s illuminating
book9 on the development of the English spoken in New Zealand is even called New-
Dialect Formation: the inevitability of Colonial Englishes, and if such a process was
inevitable in twentieth-century New Zealand it is at least plausible to envisage it for
second-century B.C. Iberia and Africa. (“Africa” here is used in its normal Latin
sense, to refer to the Western half of the Southern Mediterranean coastline; Cyrenaica
and Egypt were essentially Greek-speaking in the Empire, and even though there
were Latin-speakers there, such as Claudius Terentianus, they cannot be seen as
attesting a local Egyptian kind of Latin). Australian, South African and New Zealand
English are similar and mutually comprehensible without difficulty, although they
are not the same; it seems reasonable to propose that Iberian and African Latin were
too. The Straits of Gibraltar are no great barrier, just twelve miles wide at the closest
point, and were in practice a unifying factor rather than a separating one, as the sea
often is in times of peace.10 The Atlantic African coast west of the Straits, which
came to be known as Mauretania Tingitana, was an integral and flourishing part of
the empire (as can be exemplified at the coastal archaeological site at Thamusida11),
and was even administratively categorized by Diocletian as being part of Hispania.
In practice it seems to have been easier to trade and travel from Mauretania Tingitana
Northwards to the Iberian Peninsula than Eastwards to the African coastline East of
the Straits. There was, in short, no obvious cultural barrier along the Straits (in a
simi­lar way, Southern British and Northern Gallic Latin culture seem to have been
clo­sely related, regardless of the English Channel).
Hispania and Africa were culturally impressive, in similar ways, and continued to
be so after the Empire. Since Afro-Romance was destined not to survive, and has thus
never been studied by patriotic local Romanists in the same way as French, Italian
and Hispanic scholars have investigated the early history of Romance in their own
areas, the analysis of African Latin has been left to Latinists.12 In one way this under­
stan­dable reluctance of modern Maghrebí scholars to study the colonial inhe­ri­tance is
no bad thing, since it means that anachronistic nationalisms have not been imported
into the field; but it is also true that we cannot argue backwards from a sub­sequent
Afro-Romance to reconstruct features of Proto-Afro-Romance when investigating

7. Adams 2007, p. 407, n. 181.

8. Ibid., p. 698, 710-711.
9. Trudgill 2004.
10. See several studies in Nielsen, Schøsler (eds) 1996.
11. Mattingly, Hitchner 1995, p. 175.
12. As summarized by Lancel 1985.

Latin vulgaire, latin tardif.indb 40 29/05/12 12:18

 late and vulgar latin in muslim spain 41

the Latin of the area, so we are more reliant on evidence of the time, both explicit
and implicit, for African Latin than we are for the Latin of more Northerly regions.
But coincidentally, and fortunately, as Adams indicates, there is more such evidence
from Africa available to Latinists than there is from most other places. Augustine, for
example, who lived from 354 to 430, and who was of partly Berber ancestry, coming
from Hippo, near Carthage, acknowledged that he had himself an African way of
speaking, which his colleagues in Milan tended to criticize.13 There is no sign here
of the existence of a separate supra-regional educated sociolect; Augustine and his
neigh­bours all spoke in an African manner, apparently, noticeable to others when they
travelled elsewhere. The phrase apud nos, in Augustine, seems to mean “in Africa”,14
and Augustine seems to have preached, and expected to be understood, in contexts
which we might wish to characterise as being in some sense “Berber”.15 Africa also
had more than its obvious share of grammarians, some of whom comment on features
of their local speech (even Priscian, working in early sixth-century Constantinople,
was a native Latin-speaker from Mauretania); there were, in short, many educated
people there. But conversely, there are also useful corpora of what Adams refers to
as “subliterary” documentation from Africa, most notably the surviving ostraca from
Bu Njem in an essentially Punic-speaking area of inland Tunisia, of the 250s A.D.,
and the so-called “Albertini Tablets” of 493-496 A.D., found inland in the Maghrib,
both of which give invaluable linguistic data, precisely dated and placed.
The Albertini Tablets (named after their modern discoverer) were written at a time
when the area was politically ruled by the Vandals. Ethnically the Vandals were ori­
gi­nally Germanic, for the most part, but there is no definite evidence that they were
Germanic-speaking even during the two decades they had previously spent in the
Iberian Peninsula (409-429), and in the North African Vandal kingdom of the mid‑fifth
to the mid-sixth centuries the normal working language, spoken and written, was
Latin.16 They and their Afro-Roman subjects and neighbours could communicate, it
seems, without trouble. The Vandals even coined their own Roman sestertii.17 There
are a number of educated writers whose work comes from the Vandal kingdom,
although whether any of these writers were ethnically of Vandal descent is not clear.
And the Albertini Tablets, in the same way as the Visigothic slates, show that literacy
of a workaday kind was current even in areas that were not obviously significant
cultural centres. Indeed, linguistically the Vandal experience in Africa was probably
not all that different from the contemporary Visigothic experience in Iberia, in that
every­body could speak Latin, and probably still quite similar Latin in both places. It
seems unlikely that many Vandals learnt a Berber language.

13. Adams 2007, p. 193.

14. Ibid., p. 291.
15. Banniard 1992, chapter ii.
16. Cp. Merrills (ed.) 2004.
17. Mattingly, Hitchner 1995, p. 203.

Latin vulgaire, latin tardif.indb 41 29/05/12 12:18

 42 r. wright

The Byzantines defeated the Vandals in the sixth century, and are said by the
Greek-writing historian Procopius to have transported the Vandal men eastwards into
the Byzantine realms; but many are likely to have stayed behind, at least outside
Carthage itself, and we are also told that many of the women remained. Their
children would still be mainly Latin-speaking, and the local Afro-Romans, whether
of part-Berber descent or not, would also still be Latin-speaking during the years
of Byzantine influence (which spread briefly to Iberia as well). The inhabitants of
North‑West Africa tend after this point to be referred to by modern historians as
Berbers, but most were at least partly of Afro-Roman origin there, as most of the
inha­bi­tants of so‑called Visigothic Spain were of Hispano-Roman descent, and in
addition a number in the seventh century were likely to have some Vandal ancestors
as well; all Latin-speaking. That is, the word “Berber” is now at times applied to
people who were not particularly Berber ethnically or culturally. Berber languages
sur­vived in addition, of course, particularly inland in the Maghrib, but Berber
languages are not all the same and were not necessarily mutually comprehensible,
and we can guess that most Berber-speakers were in practice bilingual, in both Berber
and Latin. The con­clusion is that Latin (Late Latin, or Afro-Latin if we want to name
it so, or Early Romance, or what we would have called Early Afro-Romance if there
happened to have survived a modern derived Romance in the area) was still the main
language in use along the south-western Mediterranean coastline all the way to the
Atlantic when the Muslims arrived in the late seventh century. This Afro-Latin would
have been easily understood in the Iberian Peninsula, and vice versa, even if they
had recognizably distinct accents, because there is still no reason for us to visualize
a bundle of Romance isoglosses running West to East along the Straits of Gibraltar.
The Straits still represented no barrier. The Visigoths established garrisons and
fortresses on both sides of the Straits, and communicated and traded with the African
hinter­land. One of the well-known accounts of the invasion of 711 assigns a leading
role to the Visigothic Count Julian of Ceuta, governor of a fortress on the southern
side of the Straits; whether or not the tale contains any truth, the idea of a Visigothic
count based south of the Straits must have seemed natural to those who propounded
the tale. Similarly, Iberia and Africa were still part of the same linguistic area in 700;
dia­topic variation would have existed, naturally, because it always does, but iso­
glosses didn’t bundle together or disrupt communication through the continuum. That
is, if the Muslims had not arrived, the inhabitants of the western part of the North
African coast would now be speaking a kind of Spanish. There is thus nothing at all
lin­guistically unlikely in the idea that Isidore, whose family had lived on the South
Coast of the Peninsula, knew and talked with scholarly visitors from Africa, including
grammarians, historians and theologians. He certainly knew Africans well enough
to refer to them collectively as Afros versipelles, “quick-witted” (Etymologiae IX,
2, 105).
The Muslim invasion of the Iberian Peninsula in 711 is usually seen subsequently
as a shattering geopolitical event. At the time, though, it seems to have been seen in
much the same light as all the other invasions and movements of people of the age

Latin vulgaire, latin tardif.indb 42 29/05/12 12:18

 late and vulgar latin in muslim spain 43

within the nominally Christian world, not all that different from the Visigothic arrival
in Hispania. Notably, the so-called Crónica mozárabe of 754, probably com­posed by
a Christian in Toledo, makes no mention of any religious, linguistic or even cultural
differences between the invaders and the local native population; its perspective has
been illuminatingly discussed by Ann Christys.18 These invaders preferred, when they
could, to use the already existing administrative mechanisms. There seems to have
been no need for interpreters. The leaders were Arabs and Arabic-speaking, but there
were only a few of them; the senior officers recruited from the Maghrib must have
learnt some Arabic, but conversely the Arabs must have learned enough Romance
to get by, and the great majority of the invaders were what are normally referred to
collec­tively by historians as “Berbers”. This name is used geographically, to refer
to local inhabitants from the Maghrib; but we should not deduce from this that they
neces­sarily spoke one or more Berber languages, or, at least, not once they were in
Spain. Successful attempts to find linguistic traces of medieval Berber languages in
the development of Ibero-Romance have been rare (outside a few examples of topo­
nymy); the only one I know of to be still accepted concerns the Portuguese word
tabúa,19 which was used to refer to a kind of rush imported from Africa, in which
the búa element is in origin an African Latin word buda, borrowed from Berber, and
the ta- is in origin a Berber definite article; but even this word is more likely to have
been coined in Africa and then taken whole to Lusitania, either before or after the
invasion, than invented later in situ. The so-called Berbers that participated in the
eighth-century population movements spoke a language which we could rationally
call any of Latin, Late Latin, Afro-Latin, Late Afro-Latin, Early Romance or Early
Afro-Romance, depending on our academic background and perspective. Whatever
we call it, in the early decades of Al-Andalus the great majority of the newcomers
spoke in effect the same language as the people who they met in the Peninsula,
whatever their ostensible religion.20
This conclusion is not especially adventurous, or even surprising, but it has rarely
been spelt out explicitly. It is true that more Arabic-speakers came into the Peninsula
in the years following the Crónica mozárabe, including the surviving Umayyads
fleeing from Syria, many of whom would not have known any Latin at all, but it
is also true that more Romance-speakers came into the Peninsula all the time from
Africa as well. Maybe by the tenth century these new arrivals from Africa would have
been predominantly Arabic-speaking; but by that century they were in a different
context, in which to a large extent in Africa, and to a lesser extent in Muslim Spain,
the use of Latin, and in particular the use of written Latin, was fading away.
Thus in the intervening two hundred years, in the 700s and 800s, there was a large
pro­portion of the population, perhaps as many as 10%, speaking the native language

18. Christys 2003.

19. Cp. Adams 2007, p. 522-528.
20. Cp. Wright, in press.

Latin vulgaire, latin tardif.indb 43 29/05/12 12:18

 44 r. wright

in an African way. Is it therefore plausible to investigate the suggestion that individual

fea­tures of African Latin could have been able in these two centuries to affect the
sub­sequent development of Hispanic Romance? That is the question I put before you
today; and my tentative answer is going to be “yes”.

To approach this question at all, we need to have some idea of what any distinctively
dialectal, or local, African Latin features might have been, separately identifiable on
the whole from those of Hispanic Latin, wherever the relevant isogloss might have
run. After extensive and careful analysis, Adams offers us,21 with varying degrees
of confidence, two phonetic features, one morpho-syntactic feature, one semantic
feature, and a few lexical features as belonging to African Latin. What we need to
inves­tigate, when answering this question, is not only whether the African features in
question seem to have consequences in the Iberian Peninsula, but whether it seems
reasonable to date the crucial period of their origin in the Peninsula to the years of the
Romance-speaking Berber presence, in particular the eighth and early ninth centuries,
and not before.
The first feature to be investigated here is a well-known one; it concerns the
confusion, particularly in word-initial position, of the written letters b and v, and the
pho­netic development which they seem to attest. As is generally known, in Castile the
two letters came to represent the same sound, since in Castile there is no voiced labio­
den­tal consonant (unlike elsewhere, including Portugal and Catalonia). Not only are
the two letters often used incorrectly in the earliest Ibero-Romance documentation,
their confusion is also one of the features of the contemporary manu­script of the
osten­sibly Latin works composed by Alvaro de Córdoba in the mid‑ninth  century;
the confusion of these two letters is common in intervocalic position and rare word-
initially, but González Muñoz’s study22 does indicate a number of word-initial
examples there, such as uine for binae and belatum for uelatum;23 there is no other
expla­nation of this phenomenon in the work of an otherwise aggressively anti­quarian
writer than the loss of the phonetic distinction originally represented by the two
letters.
So far, so good, but this confusion contrasts startlingly with what Isidore of Seville
had said, a mere two centuries earlier: birtus, boluntas, bita uel his similia quae Afri
scribendo uitiant omnino rejicienda sunt et non per b sed per u scribenda.24 This
com­ment is disappointingly imprecise as regards phonetics, but Isidore does see this
spel­ling mistake as a particularly African feature, referring to word-initial position.
On the other hand, the confusion between these letters is rarely found in seventh-
century inscriptions of the Peninsula, although it is not completely non‑existent

21. Adams 2007, p. 259-276, 516-576.

22. González Muñoz 1996, p. 63.
23. Gil (ed.) 1973, p. 180, l. 2; p. 189, l. 18.
24. Appendix III, 3 to Etymologiae 1, 27, in Migne (éd.) 1878, col. 732B.

Latin vulgaire, latin tardif.indb 44 29/05/12 12:18

 late and vulgar latin in muslim spain 45

as has sometimes been said;25 perhaps we should conclude from this that the pho­
netic merger was just beginning to happen in seventh-century Hispania, present as
a minority variant in southern parts of the Peninsula, with the incipient isogloss, if
there already was one, crossing Hispania somewhere North of the Straits. Epigraphic
evidence from Africa itself, however, confirms beyond doubt that this pho­netic
merger was happening, and perhaps all but completed, there.26
Thus between Isidore and Alvaro the situation changed markedly. Perhaps what
hap­pened was this: in 710 Iberian Romance had a minority variant in which the
two con­sonants in word-initial position had merged, but also there were still many
speakers who preserved the distinction, and others who vacillated between the two,
with all able to understand both kinds of realization; this would be a normal state of
affairs to find during the course of a sound-change in progress. In such circumstances
the sud­den arrival of a significant number of speakers of essentially the same
language, who all used only the pronunciation which did not distinguish between
the two consonants, would have tipped the statistics of how many people used which
variant, eventually decisively, in favour of the Afro-Romance way of pronouncing
the words in question. The loss of the voiced labiodental fricative did not happen
in the areas later known as Portugal and Catalonia, and maybe this was because the
demographic effect of the new arrivals was less of a presence in those areas. There is
written evi­dence of the Romance spoken in Muslim Spain in the shape of Romance
texts written in the Arabic alphabet, and the evidence amassed and studied by Galmés
de Fuentes27 shows fairly unambiguously that the merger of the initial voiced labial
con­so­nants indeed happened in the Romance of central Al-Andalus but not in the
area subsequently known as Portugal.
If this is right, then the arrival of African Romance helped decide the outcome in a
case of existing phonetic variation in the Peninsula. If this scenario sounds familiar,
that may be because this is the process by which it is now generally thought that
bilin­gual Basque-Latin speakers decided between the more occluded [f-] and the less
occluded [h-] allophones of the word-initial Latin phoneme /f-/; their presence as
exclusively [h-]-users could weight the statistical balance in a decisive way. Similarly,
I am not claiming that Romance-speaking Berbers caused the Ibero-Romance merger,
but that they helped it decisively on its way.
This suggestion implies that there was contact between the Berber Romance-
speakers and the speakers of Cantabrian Romance at the decisive time, as well as
further south. And there was. Many of the Berbers were concentrated in the Northern
parts of Al-Andalus in the early decades of their presence, including the areas north of
the Duero which subsequently came into Castilian control in the late ninth century as

25. E.g. by Barbarino 1978, p. 82-83; cp. Lancel 1981, p. 280-281; Adams 2007, p. 628, 655-656,
puts the record straight.
26. Ibid., p. 626-666.
27. Galmés de Fuentes 1983, p. 86-88, 228-229.

Latin vulgaire, latin tardif.indb 45 29/05/12 12:18

 46 r. wright

the Christian states spread south of the Northern mountains; the settle­ment of Burgos
as the capital of Old Castile is datable to 884. As Ralph Penny28 and Donald Tuten29
have established, the resultant Old Castilian koinè was deve­loped by the speakers
during the ensuing years from a mix of several mutually com­pre­hen­sible Romance
dialects. The suggestion I am making is that one of those con­tri­bu­tory dialects could
well have been the originally African Romance spoken by the des­cendants of the
Berber communities that had come to the area during the previous 170 years, as
spoken by those (whether of indigenous, mixed or African descent) who remained
in the area after 884. These communities had already been living there for a long
time, intercommunicating with their neighbours to the north of the supposed religious
frontier. There is, in any event, no reason to suppose that the previous inha­bi­tants of
the Burgos area were all expelled or killed at the time of the settlement in the 880s,
even if several fled southwards. The lack of phonetic distinction among word-initial
voiced labials would, under this scenario, have been a variant in the speech of some
of the newcomers into Burgos from the North, but a constant in the speech of the
Romance-speakers who were in the area already; since the resultant koine in such
cases tends to prefer the simplest choice when faced with competition bet­ween
features of the contributing dialects, and one phoneme and one phonetic rea­lization is
unde­niably simpler than two, this conclusion is entirely feasible and con­sistent with
Penny’s convincing general scenario. Thus the feature continued to be used in the
same place despite the change of political ownership.
The second phonetic feature which might be relevant here is one which Africa pro­
ba­bly shared with Sardinia. In Sardinia and Africa speakers did not merge short /i/
with long /e:/, and short /u/ with long /o:/, as happened elsewhere, but simply lost
the long/short distinction in every vowel, as happened in the rest of the Romance
world only to /a/. This is probably the consequence of their no longer distinguishing
between short and long vowels at an early enough stage for the qualitative distinctions
between [e] and [i], and between [o] and [u], still not to have been eliminated. The
evi­dence for this as regards Sardinia lies in the subsequent evolution of the vowels of
Sardinian Romance, for there are no relevant Latin texts from Sardinia to help us; as
regards Africa, the evidence cannot be deduced from any subsequent Afro‑Romance
and is in contrast essentially based on metalinguistic comments of the time; but the
con­clusion is the same in each case. Augustine is helpful here. He expli­citly tells us
that he, and by implication other Africans, tends to pronounce cano with a vowel
which sounds to others like a long [a:].30 He also tells us that he, and Africans in
general, were unable to distinguish os with a long [o:] (“mouth”) from  os with a
short [o] (“bone”); and that Africans tended to pronounce ignoscere with a long [e:]
rather than a short [e] in the penultimate syllable, seeing this explicitly as a phonetic

28. Penny 2000, p. 82-93.

29. Tuten 2003.
30. In his De Musica II, 1, 1, as discussed by Adams 2007, p. 260-261.

Latin vulgaire, latin tardif.indb 46 29/05/12 12:18

 late and vulgar latin in muslim spain 47

error rather than a morphological confusion of conjugations.31 This inability applies

to perception as well as production; in general, he tells us, Afrae aures de correptione
uocalium uel productione non iudicant (De doctrina christiana IV, 10, 24), “African
ears can’t tell the difference between short and long vowels”. The fifth-century
gramma­rian Consentius also tells us that Africans lengthened short stressed vowels,
quod ipsum vitium Afrorum speciale est.32 There is no reason not to believe these
remarks, and the reason seems indeed to have been the lengthening of the short
stressed vowels that Consentius referred to, rather than vice versa. In his book Adams
chose to investigate this phenomenon through the use of the letters e and i, rather than
o and u, but it is the same difference and the same conclusion to be drawn in each
case; in African epigraphy there are startlingly few cases of one of the letters e or i
being mista­kenly used for the other (whereas there are many confusions involving b
and v), and the reason seems to be simply that, unlike elsewhere, the corresponding
vowel sounds were not confused. Thus Sardinia and Africa, which are separated by
more sea­water than Iberia and Africa are33 but are still within easy reach of each
other, and shared close trading and cultural connec­tions both before and after the
Romans invaded them, came to form part of the same linguis­tic area as regards a rela­
ti­vely early failure to distinguish vowel lengths.
Thus it seems that the Afro-Romance speakers who came into the Peninsula are
likely to have produced long versions of originally short stressed vowels. These could
be identified most easily now by subsequent unexpected failures to diphthongize,
since diphthongization only affected the stressed occurrences of originally short /e/
and /o/. In general, of course, the handbooks tell us that Castilian diphthongized short
stressed /e/ and /o/ and failed to diphthongize long stressed /e:/ and /o:/, as well as
merging short /u/ with long /o:/ and short /i/ with long /e:/, thereby demonstrating that
the distinctions originally based on length survived in these words and those based on
aper­ture did not, unlike in the Afro-Sardinian space. But the Castilian evidence is less
clear-cut than this presentation of the evidence implies. There are, in fact, many cases
of Castilian words in which these stressed originally short vowels do not diphthongize
as would ostensibly be expected. Yakov Malkiel analysed these exhaustively in
an article in Language34 entitled “Old Spanish resistance to diphthongization, or
previous vowel lengthening?”. The latter explanation, previous vowel-lengthening,
is Malkiel’s preferred solution to explain the absence of diphthongization (that is,
the absence of [jé] or [wé]) in the many such cases as pechos < PECTUS, tengo
< TENEO, espejo < SPECULUM, sea > SEDEAT, madera < MATERIAM, reja <
REGULAM, teja < TEGULAM, ocho < OCTO, ojo < OCULUM, hoy < HODIE; in

31. De doctrina christiana II, 13, 19; Vainio 1999, p. 122.

32. Cp. Maltby, in this volume, “The De Barbarismis et Metaplasmis of Consentius as evidence for
Late and Vulgar Latin”, p. 727-737.
33. Despite what Adams implies (2007, p. 527).
34. Malkiel 1984a.

Latin vulgaire, latin tardif.indb 47 29/05/12 12:18

 48 r. wright

another study published in the same year35 he further considered proparoxytone cases
such as pérdida < PERDITAM, bóveda < VOLVITAM, Mérida < EMERITAM;36
Malkiel37 also considered a number of words in which a short  [u] refrained from
deve­loping to Spanish [o], although without giving them this explanation. The first
list includes words where the stressed vowel preceded a pala­tal consonant or semi-
vowel in Old Castilian, and other analysts have tended to see that as the reason
why the diphthongization failed to operate; but this supposedly dis­couraging effect
of palatals on diphthongization cannot be seen as a universal, or even as a general
Romance feature, since not far away over the mountains, in both Old Provençal and
Old French, the proximity of a palatal feature actually encouraged diph­thongization
to occur. As Malkiel says, “the very same environments which can con­sistently block
diph­thon­gization in Old Spanish appear to have provoked it, be it only on an optional
scale, in Old Provençal”; he finds a solution in making “generous allowance for the
early provincial differentiation of colloquial Latin”.38 Not all such undiphthongized
stressed vowels appear next to palatals: e.g. compare monte < MONTEM with fuente
< FONTEM. And the lengthening and/or closure of these stressed vowels must have
happened at a time which a Hispanist would call early, because although diph­thon­
gi­zation in these forms is indeed found in the mountains of Asturias and Aragón
(outside the area of influence of African Latin immigrant groups), in Castile there
are not attested diphthongized variants of the words on that list which contain fol­
lo­wing palatal features. Indeed, the vowel lengthening probably hap­pened earlier
than the time of the arrival of the palatal semivowel [j] into the original [-kt-], [-ks-]
and [-lt-] clusters of such words as PECTUS and OCTO. There are other words
without a following palatal in which there is indeed discoverable varia­tion between
diph­thongized and undiphthongized variants, such as conde/cuende < COMITEM
and omne/uemne < HOMINEM, in which the undiphthongized variant was due to
survive (conde, omne > later hombre).
Overall, Malkiel’s is an attractive analysis up to this point. Then having reached
this conclusion concerning early vowel lengthening, Malkiel wondered why the
lengthe­ning happened in North Central Spain but not elsewhere; suggesting that the
rea­son could lie in the fact that that area, like Sardinia, had been an early conquest
by the Romans. This is hardly true, however, and even if it were, it is unlikely to be
relevant. As Adams shows, analyses of Romance differentiation put forward solely
in terms of the relative date of initial colonization have been unsatisfactory. I would
ten­tatively suggest instead a later date for the lengthening of these vowels in the
Peninsula, the eighth century, when the area North of the Duero came to be inhabited
by many Afro-Romance speakers and their descendants, who were later to contribute

35. Malkiel 1984b.

36. Cp. also Malkiel 1962.
37. Malkiel 1983.
38. Malkiel 1984a, p. 80-81.

Latin vulgaire, latin tardif.indb 48 29/05/12 12:18

 late and vulgar latin in muslim spain 49

to the dialectal mix leading to the Burgos koinè we now call “Old Castilian”. The
indi­genous Hispano-Romance speakers in the eighth century may already have
had an incipient preference for not diphthongizing before palatals, but if so that
would have been decisively reinforced by the arrival of speakers who had already
lengthened these vowels, in all or most phonetic contexts, sufficiently for them not to
be candidates for diphthongization in the first place. Again, the presence of speakers
of Afro-Romance may here have helped to decide a case of native variation, and then
in due course contributed its halfpennyworth to the consequent interdialectal koinè.
The evidence from the Romance written in Arabic letters is extremely difficult to
assess in the case of most vowels and diphthongs, and it might be best not to try to
reach any conclusions from it as regards vowel lengthening; but that evidence does
not seem to be incompatible with these suggestions, at least.

The influence in the Peninsula of the only morphosyntactic feature that may have
some kind of specifically African connection is less clear. The only possible dis­tinc­
tive feature of African Latin morphology or syntax that Adams wishes to iden­tify is
the preference for using habeo and the infinitive with future meaning. It seems plau­
sible to hypothesize that this became widespread first in Africa, and was not widely
used in pre-invasion Hispania, but on the other hand it eventually spreads to the
whole Romance world, not just Hispania. Adams39 first exemplified the wide­spread
use of habeo as a future auxiliary in Africa through remarks made by the grammarian
Pompeius (when Pompeius is discursively commenting on other features than this),
who was probably but not definitely African. It may be significant that outside the
Iberian Peninsula, other auxiliaries with future meaning are also found in Romance,
such as debeo and uolo; there is no sign of that option in the Peninsula, where
analytic future endings all come from habeo. So it is significant and surprising to see
how little evidence of habeo there seems to be with clear future auxiliary meaning
in seventh-century texts from the Peninsula.40 That is, the generalized spread of this
use of habeo in the Peninsula does indeed coincide chronologically with the later
presence of the Afro-Romance speakers who liked to use it; but since Romance
futures formed with habeo are later found everywhere else too, there may have been
no direct causal connection there.
There is no further potential syntactic example of specially African Latin; early
Romance syntactic developments were in effect the same everywhere, which may
help explain why no Latin grammarian before Priscian was interested in syntax (and
Priscian himself was inspired to his own syntactic studies by making comparisons
with Greek).

39. Adams 1991.

40. Adams 2007, p. 727-728.

Latin vulgaire, latin tardif.indb 49 29/05/12 12:18

 50 r. wright

The possible semantic example concerns the Latin word rostrum, originally used
to refer to a bird’s beak, which extended its meaning in Africa to apply to a human or
ani­mal mouth. Modern Spanish rostro in fact means “face”, usually of human beings,
but in Old Spanish it meant both “mouth” (still) and “face” (already). In Latin, this
mea­ning of rostrum may have been exclusive to Africa, in Adams’ view,41 although
the evidence is suggestive rather than conclusive; in which case, the spread of the
seman­tic change in the Peninsula would have coincided chronologically with the
Afro-Romance presence, and that may be significant. For it had earlier definitely
meant “beak” to Isidore. In his Etymologiae (XIX, 1, 13), the phrase rostratae naues
means ships whose prow is in the shape of a bird’s beak, rostrum: rostratae naues
uocatae ab eo quod in fronte rostra aerea habeant propter scopulos, ne feriantur et
conlidantur, where the rostra of bronze stick out in front of the boats to protect them
against reefs. There is a probable earlier source for this comment in Pliny’s Natural
History 18, 171,42 but Isidore is unlikely to have used words with meanings which
they did not still have; and this can only refer to the shape of a bird’s beak rather
than a mammal’s mouth, so it seems sure that the semantic development post-dates
Isidore and pre-dates the lion’s mouth (or face) referred to in the Poema de Mio Cid
line 2299, when the lion “Ante myo çid la cabeça p~mio y el rostro finco”.

The lexical example to be referred to here is the form centenarium. This word is
attested almost exclusively in Africa, including three times in the Albertini Tablets;
the one use found elsewhere, in the East, seems to have a different meaning and may
well have been separately coined. This word, as used in Africa, has hitherto been
related by Latinists to centurions, but Adams argues convincingly43 that it should
be segmented morphologically as centen + -arium, interpretable as a grain-store
for keeping rye (centenum) in, on the analogy of other words with similar mea­ning
ending in -arium. Centenum existed in seventh-century Iberia, being used by Isidore,
but not, so far as we can see, centenarium; other synonyms existed instead.
Centenero exists in modern Spanish as an adjective coined off centeno (“rye”), but
it is unattested (according to Corominas) until c.1400, on the Romance side of one of
the earliest extensive Latin-Romance glossaries; it now has the meaning of “(a piece
of land) suitable for growing rye”. It would not be convincing to relate this word to the
African centenarium of a thousand years earlier, since it is a different part of speech
with a different meaning and easily coinable on productive patterns of derivational
mor­phology some time before 1400. But what neither Corominas nor the Academia’s
dic­tionary mention is that there is a village called Centenero in the province of
Huesca, in Aragón, and (as can be confirmed both with the available pictures on
the vil­lage website and on Google Earth) a derivation of the toponym from a word

41. Ibid., p. 543-544.

42. Cp. Oroz Reta (ed.) 1982, II, p. 432.
43. Adams 2007, p. 550-554.

Latin vulgaire, latin tardif.indb 50 29/05/12 12:18

 late and vulgar latin in muslim spain 51

meaning “grain-store” is fully plausible. Centenero lies North-North-West of Huesca

itself, due south of Jaca, apparently now having a population of just 9 to frequent
its ancient church and two ancient hermitages. It is attested in documentation of the
eleventh century. The Academia’s diachronic corpus (CORDE) offers documentation
from 1036, 1046 and 1065, which has been published.44 These show that it had a
church and an abbot at that time. The village’s website refers to an even earlier
document, of 14 April 1035, in which King Sancho Garcés III donated the villas of
Salamaña and Centenero.45 This must have happened almost immediately after the
vil­lage had been captured from Muslim Zaragoza, and the documentation seems to
imply that the village had been in existence during its time under Muslim rule rather
than that it was being newly founded in the 1030s. If so, the name was probably
given to it by descendants of the Afro-Roman Berbers, themselves descendants of
ances­tors similar to those who wrote the Albertini Tablets. There is another village
in Huesca province called Pardina Centenero;46 this probably has the same original
refe­rence to a grain-store, with the root in Latin PARATA (or PARIETINA, according
to Benito Moliner); and in addition another called Centenera (this feminine form
also occurs in Salamanca province). As Emilio Nieto Ballester47 established in the
pre­vious conference, the toponymy of Huesca province seems particularly revealing
of Old Romance features of the area, and this looks like a good example. There is
an Aldeacentenera in the province of Cáceres, to the East of Trujillo, a one-word
compound, the first part of which, aldea “village”, is taken from Arabic, and thus the
whole toponym seems likely to have been coined within Muslim Spain by bilinguals.

These five examples (or six if we include Portuguese tabúa) are not many. But they
do include all the main features identified by Adams as being probably distinctively
Afro-Latin in the seventh century. They are enough to suggest the presence after the
inva­sion of an Afro-Romance element in the general dialect mix that was to lead to
Old Spanish. And they also exemplify well the general theme of this presentation:
by extending forwards in time the discoveries of Latinists working before 600, as
here exemplified by Adams, and by extending backwards the discoveries made
by Romance specialists working on data from after 800, as exemplified in this
case by Malkiel and Menéndez Pidal, and by exploiting the conclusions of recent
historians (such as Christys, in this case), in order to form coherent theories of the
socio­linguistic context, and allying all this with careful philological analysis in the
combination which I call “sociophilology”,48 we might be able to illuminate some of
the darker corners of Late Vulgar Latin. The nature of Late and Vulgar Latin in the

44. Ibarra y Rodríguez 1904.

45. Durán Gudiol 1965, doc. 15.
46. Ariño Rico 1980.
47. Nieto Ballester 2008.
48. Wright 2003.

Latin vulgaire, latin tardif.indb 51 29/05/12 12:18

 52 r. wright

early centuries of Muslim Spain will always be less than clear, but the darkness is not
neces­sarily impenetrable; in particular, it seems reasonable to conclude that fea­tures
which seem to have changed markedly there between the early eighth and late ninth
centuries may have been at least partly catalysed, if not caused, by the undoubted
presence there of many Early Romance-speakers who came originally from Africa.

Bibliography

Adams J.N. 1991, “Some neglected evidence for Latin habeo with infinitive: the order of the
constituents”, TPhS 89, p. 131-196.

— 2007, The Regional Diversification of Latin, 200 BC - AD 600, Cambridge.

Ariño Rico L. 1980, Repertorio de nombres geográficos: Huesca, Saragossa.

Banniard M. 1992, Viva Voce  : communication écrite et communication orale du iv e au

ix e  siècle en Occident latin, Coll. des études augustiniennes. Série Moyen âge et temps
modernes 25, Paris.

Barbarino J.L. 1978, The Evolution of the Latin /b/-/v/ Merger: a Quantitative and Com­pa­
ra­tive Analysis of the B-V Alternation in Latin Inscriptions, North Carolina Studies in the
Romance Languages and Literatures 203, Chapel Hill.

Benito Moliner M., Pueblos del Alto Aragón: el origen de sus nombres. Web site: <www.
aragob.es/edycul/patrimo/etno/pueblos/portada.htm>.

Christys A. 2003, “The transformation of Hispania after 711”, in H.-W. Goetz, J. Jarnut,
A. Pohl (eds), Regna and gentes: the Relationship between Late Antique and Early Medie­val
Peoples and Kingdoms in the Transformation of the Roman World, The Trans­for­ma­tion of
the Roman World 13, Leiden, p. 219-241.

Collins R. 2004, Visigothic Spain, 409-711, A History of Spain, Oxford.

CORDE = Corpus diacrónico del español. Web site: < www.rae.es>.

Corominas J., Pascual J.A. 1980-1991, Diccionario crítico etimológico castellano e

hispánico, Madrid, 6 vol.

Durán Gudiol A. 1965, Colección diplomática de la Catedral de Huesca, I, Saragossa.

Echenique Elizondo M.T. 2004, “La lengua vasca en la historia lingüística hispánica”, in
R. Cano (ed.), Historia de la lengua española, Ariel lingüística, Barcelona, p. 59-80.

Galmés de Fuentes Á. 1983, Dialectología mozárabe, Biblioteca románica hispánica. 3,

Manuales 58, Madrid.

Gil J. (ed.) 1973, Corpus Scriptorum Muzarabicorum, Manuales y anejos de «Emerita» 28,
Madrid.

Goetz H.-W., Jarnut J., Pohl A. (eds) 2003, Regna and gentes: the Relationship between
Late Antique and Early Medieval Peoples and Kingdoms in the Transformation of the
Roman World, The Transformation of the Roman World 13, Leiden.

Latin vulgaire, latin tardif.indb 52 29/05/12 12:18

 late and vulgar latin in muslim spain 53

González Muñoz F. 1996, Latinidad mozárabe: estudios sobre el latín de Álbaro de Córdoba,
Monografías 32, Corunna.

Ibarra y Rodríguez E. 1904, Colección de documentos para el estudio de la historia de

Aragón, I, Saragossa.

Lancel S. 1981, “La fin et la survie de la latinité en Afrique du Nord. État des questions”,
REL 59, p. 269-297.

— 1985, “Y a-t-il une Africitas ?”, REL 63, p. 161-182.

Malkiel Y. 1962, “Towards a unified system of classification of Latin-Spanish vowel corres­

pon­dences”, Romance Philology 16, p. 153-169.

— 1983, “Alternatives to the classic dichotomy family tree/wave theory? The Romance
evidence”, in I. Rauch, G.F. Carr (eds), Language Change, Bloomington, p. 192-256.

— 1984a, “Old Spanish resistance to diphthongization, or previous vowel lengthening?”,

Language 60, p. 70-114.

— 1984b, “Spanish diphthongization and accentual structure (in diachronic perspective)”,

Diachronica 1, p. 217-241.

Mattingly D.J., Hitchner R.B. 1995, “Roman Africa: an archaeological review”, JRS 85,
p. 165-213.

Menéndez Pidal R. 1972, Orígenes del español: estado lingüístico de la península ibérica
hasta el siglo xi, Madrid.

Merrills A.H. (ed.) 2004, Vandals, Romans and Berbers: New Perspectives on Late Antique
North Africa, Aldershot.

Migne J.-P. (éd.) 1878, Patrologia Latina, LXXXII, Paris.

Nielsen H.-F., Schøsler L. (eds) 1996, The Origins and Development of Emigrant
Languages: proceedings from the Second Rasmus Rask Colloquium, Odense University,
November 1994, North-Western, European Language Evolution Suppl. 17, Odense.

Nieto Ballester E. 2008, “Ongotituero (Linás de Broto), Bernatuara (Torla), Tripilituara

(Hoz de Jaca): sustantivos latino-romances derivados en -toriu y en -toria en la toponimia
de Huesca”, in R.  Wright  (éd.), Latin vulgaire, latin tardif. VIII, Actes du VIII e  colloque
inter­­national sur le latin vul­gaire et tardif, Oxford, 6-9 septembre 2006, Hildesheim, p. 263-
270.

Oroz Reta J. (ed.) 1982, San Isidoro de Sevilla. Etimologías, Biblioteca de autores cristianos
433, Madrid.

Penny R. 2000, Variation and Change in Spanish, Cambridge.

Poema de Mío Cid, 2a ed., Burgos, 1988, 2 vol.

Schmitt C. 2003, “Die verlorene Romanität in Afrika: Afrolatein/Afroromanisch”, in

G. Ernst et al. (Hrsg.), Romanische Sprachgeschichte: ein internationales Hand­buch zur
Geschichte der romanischen Sprachen, I, Handbücher zur Sprach- und Kommunikations­
wissenschaft 23, 1, Berlin, p. 668-675.

Trudgill P. 2004, New-Dialect Formation: the Inevitability of Colonial Englishes, Edinburgh.

Latin vulgaire, latin tardif.indb 53 29/05/12 12:18

 54 r. wright

Tuten D.N. 2003, Koineization in Medieval Spanish, Contributions to the Sociology of

Language 88, Berlin.

Vainio R. 1999, Latinitas and Barbarisms According to the Roman Grammarians: Attitudes
towards Language in the Light of Grammatical Examples, Turku.

Van Acker M. 2007, Ut quique rustici et inlitterati hec audierint intellegant : hagiographie
et communication verticale au temps des mérovingiens (vii e‑viii e  siècles), Corpus
Christianorum. Lingua Patrum 4, Turnhout.

— 2008, “Dans les méandres de la communication verticale mérovingienne : connaissances

pas­sives et perte d’informations”, in R. Wright (éd.), Latin vulgaire, latin tardif. VIII, Actes
du VIII e colloque international sur le latin vul­gaire et tardif, Oxford, 6-9 septembre 2006,
Hildesheim, p. 463-471.

Velázquez Soriano I. 2004, Las pizarras visigodas: entre el latín y su disgregación: la

lengua hablada en Hispania, siglos vi-viii, Col. Beltenebros 8, Madrid.

Wright R. 2003, A Sociophilological Study of Late Latin, Utrecht Studies in Medie­val

Literacy 10, Turnhout.

— (éd.) 2008, Latin vulgaire, latin tardif. VIII, Actes du VIII e colloque international sur le
latin vulgaire et tardif, Oxford, 6-9 septembre 2006, Hildesheim.

— in press, “Linguistic and ethnic identities in the Iberian Peninsula (400‑1000 A.D.)”, in

W. Pohl et al. (Hrsg.), Ethnische Identitäten im Frühmittel­alterlichen Europa, Vienna.

Latin vulgaire, latin tardif.indb 54 29/05/12 12:18