(unpublished work in progress)
Toledo and All That
© Anthony Pym 1998
[An earlier version of this chapter was published as "Twelfth-Century Toledo and Strategies of the Literalist Trojan horse", Target 6/1 (1994), 43-66.]
Unlike the Qur'an translation, which is rarely accorded more than anecdotal interest, mention of the 'School of Toledo' is obligatory in any history of translation. Toledo, we are told, was the place where Christendom gained what the Arabs knew in the fields of astronomy, astrology, mathematics, medicine and necromancy. Yet the references to Toledo are not strict in any geographical sense, since twelfth-century translations from Arabic were carried out in many parts of Hispania, often in the north. Some historians also stretch the label to include the thirteenth-century translations carried out for Alfonso X, as if the two centuries were really the one 'school.' Since I disagree with this extension, the thirteenth-century Alphonsine translations will be dealt with separately, in the following chapter, where I shall also explain the reasons for my quiet dissent. Our business here is mostly with the twelfth-century activities, with the politics of church patronage, and with the ways the Clunaic translation of the Qur'an, located more or less in the middle of the century, can be used to address a basic and fairly obvious question: To what extent can the translating associated with Toledo be described in terms of a 'school'?
In asking this question, I am not really interested in whether there was a 'School of Toledo' in any capital-letter sense of the term, which is a matter for historians of glory. My concern is more literal and lower-case: What relation could there have been between translation and a little actual schooling, in the sense of teaching and learning things Arabic? My working hypothesis is that any translating associated with pedagogical activities must have been in some way related to the church. The principles of the translating might thus be sought in this relation. Further, if there was tension in the relation, the principles might be formulated in terms of a negotiation, hopefully in the form of a much better negotiation than anything the abbot of Cluny was able to achieve.
Given this point of departure, it is convenient to divide the twelfth-century
work into two basic periods: everything that happened before the visit
of Petrus Venerabilis in 1142 (starting sometime before 1115, when we can
more or less assume Adelardus de Bada and Petrus Alphunsis had moved from
Hispania to London) and everything that happened after the visit, bringing
together the centralized church sponsorship that would continue through
to at least 1240. Prior to the visit, there is no obvious connection between
the translators and Toledo; after the visit, the connections are numerous
and institutionalized. This simple difference should be enough to justify
the two-period approach. Yet there are also grounds for seeking a third
stage somewhere within the long period of centralized church patronage.
This third stage should highlight the importance of the great international
translators that did actually work in Toledo and whose integration into
the church structure is sometimes ambiguous, as we shall soon see. We could
possibly date the third stage from 1157 to 1187, the years that the Italian
translator Girardus Cremonensis (Gerard of Cremona) would seem to have
spent at Toledo, although it could arguably be extended right through to
1220, when Michael Scotus left Toledo for Italy. The main point is that
the institutionalized work of the foreign translators was somehow carried
out within a period of official church sponsorship at Toledo. They were
the ones who established whatever basis there might be for associating
the town with a school of translators. My focus here will thus be on the
translators of this third period and their possible relations with the
What's in a School?
In the thirteenth century Gaufridus de Vinosalvo (Geoffrey of Vinsauf) referred to Toledo as a centre where the quadrivium was studied, placing it alongside the universities of Paris, Bologna and Salerno (Procter 1951: 8). Less definite references were circulated by Albertus Magnus and Roger Bacon in the same century. But no medieval author, as far as I can tell, ever referred to Toledo as a school, much less a school of translators. References to study of the quadrivium were certainly passed down through the centuries, and we shall soon see where they could well have had their origin. Yet the 'School of Toledo' as such, as a school of translators, is more properly the invention of two nineteenth-century historians.
In the nineteenth century Amable Jourdain (first published in 1819, better known through the revised edition of 1873) put together two translators' prefaces apparently addressed to archbishop Raimundus of Toledo, both apparently by the prolific translator Johannes Hispanensis (John of Seville, or of Spain). One of the prefaces mentioned close collaboration with 'Dominicus archidiaconus.' This enabled Jourdain to pin together an archbishop, an archdeacon, a translator, and translation as a repeated activity occurring in Toledo. The historian's task was suddenly simplified: 'We confess, with a joy that all men of letters will appreciate, that the discovery of this college [erroneously spelt collége] of translators has made up for the innumerable thorns that have covered our path.' Toledo was the prize that justified the scholar's voyage through dusty and difficult manuscripts. Jourdain had no doubt that archbishop Raimundus 'was the creator of a college of translators' (1873: 108, 119).
In 1874 Valentin Rose then made much of an Englishman's report of Girardus Cremonensis's work at Toledo. Rose concluded that Toledo was 'a formal school [Schule] for the transfer [Übertragung] of books and science from Arabic to Latin.' Rather than a school for training translators, this was supposed to be something like a proto-university: 'Girardus Cremonensis gave lectures to young men from all over the world. [...] Scholars went there not just to make books, but to use books to teach others.' [...] 'Girardus Cremonensis gave lectures (lectiones) before a circle of listeners (auditores), with whom he at the same time disputed in the Arabic way, among them Daniel the Englishman.' (1874: 327, 328-329, 332). We will meet Daniel in a minute. But note, in passing, Rose's insistence that these scholars disputed 'in the Arabic way' ('ganz auch nach arabischer Weise... disputiert').
Jourdain's 'college' referred to work done for the Clunaic archbishop Raimundus, presumably just after the abbot's visit. Rose's 'Schule' concerns what we have called the third stage, the years when the great international translators were working in what could have been a more independent way. We thus have two separate discoveries of two rather different kinds of schools, neither of which refers to anything that happened before 1142. Hence considerable confusion.
What might Jourdain have meant by a 'college' of translators? In the twelfth-century context, perhaps as now at Oxford and Cambridge, a college was essentially a boarding house for students. Such, at least, was the nature of the 'Collège des Dix-Huit', founded in Paris in 1180 by John of London. The idea was Islamic in origin. George Makdisi (1980) explains that Islamic colleges had functioned from at least the eighth century as charitable institutions that depended on benefactors and basically taught grammar and law. In the eleventh century, they developed into a system of teaching institutions with adjacent inns for out-of-town students. Did Jourdain know this? Is that what he meant? Whatever the case, the French historian may not have been entirely amiss when using the term 'college' for whatever was happening at Toledo.
Then again, a 'college' is perhaps not quite the same thing as a 'school,' and there can be no doubt the the latter term has dominated more recent reflections. So exactly what is meant by a 'school'? When Myriam Salama-Carr (1990: 31) refers to the 'school of translators' that formed around Hunayn ibn Ishaq in ninth-century Baghdad, she carefully specifies that this Islamic 'school' involved no more than a 'team or group of translators,' which sounds more in keeping with the meagre evidence Jourdain put on display. True, the translational activity in Baghdad was associated with a 'House of Wisdom,' a college where teaching was in fact carried out. But Salama-Carr resists this extension of the term 'school.' In fact, the only evidence she finds of any actual teaching of translation is limited to passages where some members of translation teams are called 'students' or 'disciples' (32). Similar references can be found in northern Spain in the first of our periods, prior to anything that can be associated with Toledo: the Carinthian translator Hermannus Dalmata calls the Flemish translator Rudolfus Brugensis 'Hermanni secundi discipulus' (Haskins 1924: 56) and also refers to his fellow translator Robertus as 'ilustrus socius Rodbertus retinensis' (Jourdain 1819: 104). Yet no one has taken these links as indication of any school. Indeed, such references would generally concord more easily with a model of apprenticeship than with a teaching institution in any modern sense. If there is no more to a 'school' than apprenticeship, there would be little reason to attach particular importance to Toledo at all.
Yet Rose, as we have seen, used the term 'school' in the sense of a definite pedagogical activity. Following Rose's citations, Sudhoff (1914: 81) had few qualms about describing Toledo as being peopled by Girardus's 'students' and 'colleagues,' specifying their institutional relationship as a 'translators' school' (Übersetzerschule), presumably a place for the meeting and training of translators. D.M. Dunlop (1960: 55) cited Rose to insist that 'Toledo was a real teaching centre'; Georges Mounin (1965: 35) described Toledo as 'the first true school of translators' (scuola di traduttori). The references trickle down to nonspecialist texts of recent dates. Henri van Hoof (1986: 10) speaks of a 'true school where lectures were given.' And so on. All these opinions take their lead from Rose in that they associate the 'school' with the Toledo of Girardus Cremonensis. Yet the scholars have consistently confused or combined two distinct ideas. For some, Toledo was a place for the training of translators. For others, it was a school--or an incipient university--for scientific learning. No one has clarified the relation between these two ideas. Nor have the terms been distinguished from the 'school' as mere apprenticeship arrangement or the 'college' as an associated boarding house.
With or without clarification, there have been sporadic challenges to the notion of the 'school' as a whole. Charles Haskins (1924: 12) commented that 'of a formal school the sources tell us very little,' although he did admit that 'the succession of translators is clear for more than a century' and then referred unproblematically to the 'schools of Toledo' (156), in the plural. Clara Foz (1987: 29) similarly observes the lack of documental evidence for any kind of formal school, adding that even if taken in the more diffuse sense of a 'group of scholars working with the same doctrine, style or procedures,' the extreme differences between the translators and translations of the twelfth century should make the term quite inappropriate. More vigorously, Marie-Thérèse d'Alverny (1982: 144) seeks to undo 'the legend of the Toledo school of translators supposedly initiated under the patronage of archbishop Raymond.' According to her, Jourdain's dubious references to Raimundus have been picked up and associated with the later work of Girardus Cremonensis, confusing two quite different periods. The outcome would have been erroneously 'projected back onto the earlier period to justify viewing Toledo as the main center of translation during the twelfth century' (145). In a more detailed critique, d'Alverny (1957: 72) asks at least one very pertinent question: If Toledo had a 'school of translators'--in whatever sense--at the time of Petrus Venerabilis's visit to Castile, why could the abbot only have recruited Petrus Toletanus, just one mediocre Latinist, from that supposed center? Why did he insist that his own secretary check the Latin? And why were his main translators not from Toledo at all? Until these questions are answered, there should be no supposition of any 'school of Toledo' prior to at least 1142.
Yet this does not help us decide what kind of college or school might have existed after than date.
Despite all the vagaries that have followed them, Jourdain and Rose were not engaged full-time in the art of spreading fictions. Their insights can be supported by at least seven documents:
1. Johannes Hispanensis's preface to his translation of Costa ben Luca's De differentia spiritus et animæ is indeed addressed to archbishop Raimundus of Toledo, as cited by Jourdain. Unfortunately, the preface proves nothing by itself, since no collaborative work is mentioned and we know the same translator also worked outside of Toledo and for clients who had little to do with the Toledo cathedral (Jourdain 1819: 117; Haskins 1924: 13).
2. The preface to a translation of Avicenna's De anima (ed. van Riet 1972: 103*-104*) describes the procedure whereby one translator produced an oral Romance version and someone called 'Dominicus archidiaconus' wrote it down in Latin. Jourdain believed the preface was addressed to archbishop Raimundus and that the oral translator was Johannes Hispanensis. But d'Alverny (1964) has shown that Jourdain's conclusions depended on a miswritten inflection ('Johannes' instead of 'Johanni') and a correspondingly misplaced comma in Jourdain's transcription. Without bashing the philological details, suffice to say that the more likely addressee was archbishop Johannes, who succeeded Raimundus at Toledo in 1152, and the actual oral translator was someone called 'Avendauth Israelita.' The document thus does nothing to support Raimundus's claim to fame and should rightly undermine Jourdain's 'college' altogether. Yet it does indicate use of a particular collaborative translation procedure, as well as church financing of translations ('...ut vestro munere et nostro labore...,' says the translator).
3. A thirteenth-century manuscript edited as La Fazienda de Utra Mar (Lazar 1965) is presented as a copy of a mid-twelfth-century work commissioned by Raimundus ('Remont, arçobispo de Toledo') and carried out by 'Almeric, arçidiano de Antiochia' as a guide to places in the Holy Land. The text, not mentioned by Jourdain (who was chasing the tradition of Aristotle), is partly a series of translations from Hebrew biblical texts, supported by the Vulgate. It should thus count as indication of serious interest in translations in Toledo. Yet there are several mysteries involved. Why should this one text be in Castilian when the twelfth-century norm for the Toledan church was to translate into Latin? Why should we find no other mid-twelfth-century Castilian prose of such length? Why should Raimundus be interested in a guide to the Holy Land? If we feel uncomfortable with the answers available to these questions, the document would seem to stick out like a sore thumb when attached to the twelfth century church. Could this perhaps be a thirteenth-century text attributed to Raimundus' patronage, perhaps to gain authority once the reputation of Toledo had been established, or perhaps to justify a Biblical translation at a time when vernacular Bibles were being banned? More to the point, if there was a school of translators in Toledo, why would the archbishop there go looking for a translator in Antioch? Whatever else it might be, this document is not proof of a School of Toledo.
4. A Philosophia written by the Englishman Daniel de Merlai (presumably of Morley) (ed. Sudhoff 1917) does indeed mention Girardus Cremonensis and a certain Galippus talking on astronomy at Toledo. Galippus, a Mozarab, is also said to 'interpret' Ptolemy's Almagest in Romance while Girardus writes it down in Latin. Daniel mentions no actual translations, no formal school and no official church involvement. But the activities he describes could be associated with both a group learning process and the translation procedure outlined in the De anima preface. More tellingly, he explicitly mentions Toledo as a place where studies, 'almost all in the quadrivium,' are 'celebrated' (whatever that means). This mention of the quadrivium suggests we might have the source for the accounts circulated by Gaufridus de Vinosalvo and others in the following century.
5. Girardus Cremonensis's translation of the Almagest is dated 1175, thus allowing Daniel's account of Toledo to be located with some precision. Even if the date refers to a manuscript copy, we have grounds to suspect that Daniel was in Toledo in 1175 at the latest. Further, although the Almagest translation mentions no collaboration, Kunitzsch's analysis of the text (1974: 108-109) indicates Mozarab mediation between the Arabic and the Latin. Analyses of other translations by Girardus find evidence of similar oral mediation (Opelt 1959: 139, d'Alverny 1968: 134, Jacquart 1989: 117).
6. The Vita of Girardus Cremonensis (ed. Sudhoff 1914; trans. McVaugh 1974) was written after Girardus's death in 1187 and gives a list of his translations carried out in Toledo. The impressive number of works could justify Girardus's status as 'perhaps the greatest translator of all time' (van Hoof 1986: 10). But the key evidence is that the list was drawn up by the master's socii, mostly glossed as his 'students' or 'associates.' The document thus allows some kind of group activity to be associated with Girardus's translations. However, it mentions no collaboration in the translations, no formal school, and no church involvement.
7. The cartularies of Toledo (ed. Hernández 1985) indicate the presence of someone called 'Dominicus Colarensis archidiaconus' from 1162 to 1178, as well as Girardus Cremonensis named as 'magister Girardus' among the canons in 1157 (doc. 119) and as 'Girardus dictus magister' in 1174 and 1176 (doc. 165 and 174).
It seems quite reasonable to equate 'Dominicus Colarensis archidiaconus' of the cartularies with 'Dominicus archidiaconus' of the De anima preface (d'Alverny 1989: 196). This allows one of Jourdain's prime pieces of evidence to become contemporaneous with Girardus's time at Toledo, further indicating that Daniel de Merlai was indeed describing a translation procedure. To put it in a nutshell, Rose beats Jourdain, and archbishop Raimundus should have very little claim to fame.
A fairly coherent narrative can thus be construed by bringing together the positive elements in these six documents: Archbishop Raimundus's sponsoring of translations at Toledo was continued by his successor Johannes, who had his archdeacon Dominicus write Latin translations from oral Romance-language versions furnished by Jews or Mozarabs, a procedure that was gradually taken over after 1157 by Girardus and his collaborators, whose translating and debates were witnessed by Daniel in or before 1175.
This narrative would keep many historians happy. It could even make up for many thorns across complicated pathways. But translation history should also account for the numerous gaps in the documents. It should try to explain why Daniel mentions no translators, no formal school and no official church involvement, why there is no mention of collaboration or church involvement in Girardus's Almagest or Vita, why Girardus signs his name as 'dictus magister,' why the cartularies make no clear reference to any translation school, and why any archbishop of Toledo might be interested in translations from Arabic in the first place. Answers to these questions must begin from Daniel's Philosophia, which is the document with the most significant gaps.
Having returned to England after travels in France and Spain, Daniel tells the bishop of Norwich what he has seen and learnt. When referring to Toledo he twice mentions Galippus the Mozarab. One reference has Galippus speaking alone 'in the language of Toledo' and his words being written down in Latin: 'ut auditoris animus fortius cohereat, quod a Galippo mixtarabe in lingua Tholetana didici latine subscribitur.' Another describes Gallipus interpreting the Almagest for Girardus: 'Girardus Tholetanus, qui gallipo mixtabarbe interpretante Almagesti latinauit' (ed. Sudhoff 1917: 9, 39-40). The significant points are 'saying in the language of Toledo,' 'writing down in Latin' and the Almagest as the object of these processes.
The role played by Galippus has been interpreted in various ways. Rose (1874: 329, 332) talks of the 'lectures' (Vorlesungen) of both Girardus and Galippus. Haskins (1924: 15) agrees that 'both Gerard and Galippus lectured on astronomy in the hearing of Daniel of Morley.' For Thorndike (1923: II 88, 89) 'Daniel tells of Gerard of Toledo interpreting the Almagest in Latin with the aid of Galippus [...]. He speaks of setting down in Latin what he learned concerning the universe in the speech of Toledo from Galippus, the Mozarab.' According to Dunlop (1960: 57) Daniel heard Galippus 'lecture "in lingua Tholetana," presumably Romance.' Jacquart (1989: 110) believes that 'Daniel de Merlai, Girardus's disciple, indicates [signale] the participation of a Mozarab named Galib in Girardus's work on the Almagest.' D'Alverny (1982: 452), on the other hand, refers to a Galippus 'with whom Daniel used to chat.' As anyone who has taught in a university will realize, there are considerable formal differences between lecturing, interpreting, setting down in writing, participating, and chatting. Although these terms all refer to ways knowledge can be transferred, the power relations are quite different. Then again, as might be suspected by anyone who has participated in a reasonably active translation class, there is not such a wide formal difference between translating, chatting about a translation, and gaining knowledge from a source text. Peculiarly translational transfer makes authoritative power relations rather more difficult to maintain. So what kind of school is it to be?
Three aspects of Daniel's account arouse suspicion.
First, no Galippus is mentioned in any of the texts produced in Toledo. If Galippus existed, he must have had a subordinate role in whatever was going on, presumably because of his cultural status as a Mozarab. He was not likely to be regarded as a 'master.' Once away from the social structure of Toledo, Daniel was perhaps free to tell this tale out of... school.
Second, Daniel makes no reference to any official church patronage. This is strange. Since he is writing to his bishop, he might be expected to stress or even invent church interest in scholarship at Toledo. The fact that he does not do so suggests either there was no church involvement or there was some tension between Girardus's translation process and the Toledo cathedral.
Third, at the very beginning of his text, Daniel is by no means a neutral observer when he tells his bishop why he went to Toledo:
The idea of leaving the decadence of the known world to head off in search of uncorrupted learning is certainly a convention for this kind of text. Similar complaints can be found in Vitruvius, Galen, Firmicus, Boethius and through to Adelardus de Bada. For Daniel, Paris was a thorn in the path that led to Toledo. Perhaps like Amable Jourdain, the Englishman had a certain discursive interest in making Toledo look like the goal that justified the journey. It had to be a much better version of Paris, the school he had left behind. Daniel's Paris is founded on Roman tomes behind which authoritative silence hides ignorance. His Toledo is filled with oral interpretation and debate. Paris is seated; Toledo is standing up. The contrast could not be more extreme. But if this contrast is to work, Toledo must be more than just a few translators chatting among themselves. It must look like a school. It should be a school. And yet Daniel, frustratingly, makes no explicit reference to any school. He has Galippus merely 'say' and 'interpret,' actions that could apply equally well to either a lecture or a translation process. One must wonder if Daniel had a discursive interest in not interpreting the implications of these verbs. Although his overall textual strategy indicates that Toledo is to be read in terms of a school, an anti-Paris, his actual words are bluntly literal. He puts down exactly what he saw ('it walks like a duck...'). If any interpretations are to ensue they will be the responsibility of the reading bishop or future historians.
Literalism can be what happens when one has an interest in not interpreting.
Let's accept that what Daniel describes is both a learning process and a translation procedure. What do we gain or lose by such a hypothesis? If students write down a lecturer's words in another language, are they any less involved in translation than would be an Italian writing down a Mozarab's Romance in Latin? Aren't both processes translational and pedagogical? But there is at least one important difference. In the model of the formal school, a lecturing Galippus would be the master and those writing down his words would be students. And yet the omission of Galippus at Toledo would indicate quite the opposite evaluation. In the translation situation, the oral intermediary would appear to play a subordinate role to the master who writes the Latin. The hypothesis of a translation process that is perhaps also a school thus becomes a problem of the power relationship between oral and written discourse, between spoken and silent interpretation.
This analysis finds some justification in the wider twelfth-century
Intermediaries as Masters or Students?
Moving back from the Toledo of around 1175 to the Barcelona of around 1145, we find another Italian, Plato Tiburnensis (of Tivoli), translating with the help of the Jewish 'interpreter' Abraham Savasorda. As an interpreter, Savasorda was thought to be a mere intermediary, to the greater glory of the translator into Latin. But Steinschneider shows that the relationship was quite the reverse. Savasorda was a Jewish philosopher and mathematician, in at least one case interpreting his own texts so that Plato Tiburnensis could render them into Latin. According to Steinschneider (1867: 328) 'through a strange misunderstanding the thirteenth century saw the Jews as pupils [Schüler] of the Christians.' The situation was more like a translator working in collaboration with an author. And the greater glory should perhaps go to the Jewish philosopher who took the trouble to interpret.
Similar confusion surrounds the translator's preface to the Latin version of Avicenna's De anima, famous not just as the text that Jourdain interpreted wrongly but also as one of the few texts explicitly describing the use of Romance as an intermediary language (Jourdain 1819: 449-50; G. Menéndez Pidal 1951: 364). My version roughly follows Hilty's glosses (1954: xxxvii):
Here's the book you ordered [Hunc igitur librum vobis praecipientibus], which I have rendered orally [proferente] from Arabic [ex arabico translutum] in the ordinary way or speech [vulgariter] working word-for-word [singula verba] and which Dominicus archidiaconus has rendered into Latin word-for-word [singula in latinum convertente].
The passage is complicated by yet another mobile comma (van Riet 1972: 96*). But the interest of the text is not limited to the identity of the second person, who we now know is not Raimundus but his successor archbishop Johannes. Nor is it limited to the identity of the first person, perhaps also altered by the corrected comma. The really problematic aspect is the fact that the first person is certainly not Dominicus archidiaconus, who is named in the third person as the Latinist. And yet the text is in Latin. It must be concluded that the Jewish intermediary 'Avendauth' is able to write in Latin or has instructed his Latin scribe to write these words down. Either way, as d'Alverny emphasizes (1989: 195) it is the Christian Latinist and not the (perhaps Jewish) intermediary who is presented as the acolyte. Further, according to van Riet (1972: 98*, 99*n) the actual translation in this case bears no trace of oral mediation and its style differs from other translations attributed to the archdeacon. That is, the one preface that describes partitioned collaboration could have been a lie; 'Dominicus archidiaconus' may well have been no more than a name guaranteeing ideological acceptability. And the translations that do bear traces of collaboration--notably Girardus's--make no mention of it in their prefaces. Strange things were going on. The archdeacon nevertheless gained pride of place in most accounts of Toledo, usually under the name of Gundisalvi or variants thereon, mostly as Raimundus's second-in-command, official philosopher, and master of translational ceremonies. The master intermediary 'Avendauth' was relegated to mysterious shadows.
Since the inferiorization of oral intermediaries is indicated in interpretations of all three of our stages--Savasorda, Avendauth, Galippus--it might be projected as a general principle of whatever was going on. This would explain why there is no mention of intermediaries in Girardus's Almagest and Vita. But it does not explain why there might have been tension between the Latinist translators and the church.
How would Foreign Translators be Received in a Spanish School?
Almost everyone writing in Latin in the twelfth century was either a monk or a clerk trained by the church. All the Latinist translators in Spain were probably related to the church in one way or another. However, since little specific information is available with reference to Toledo, we are forced to develop hypotheses about this relation from several general trends in twelfth-century education.
Clerks had mostly been trained in monastery schools--yes, schools--where they would be taught to the age of fifteen. After this training, they could abandon the monastery without taking sacred orders. This left them free to work for public administrations. The fact that clerks were thus able to commercialize their written Latin created considerable tension within the church, some of whose scholarly business involved providing political services in its own written Latin. This tension increased further as clerks were progressively able to receive money from students, setting up external schools and laying down the economic foundations for university corporations. The result was the unruly world of relatively independent intellectuals idealized by Le Goff (1957).
Numerous cathedrals and large churches nevertheless capitalized on this process by setting up their own slightly more advanced schools, thus becoming 'collegiate churches,' probably with some kind of residential arrangements for outside students. Since the local church schools did not teach students beyond the age of fifteen, one of the key elements of the resulting system was exile, terra aliena, as part of a good education (Delhaye 1947). Students would move to the larger towns and from country to country; they needed places to stay; Islamic-like colleges were founded. Further, in order to alleviate the financial burdens involved, Pope Alexander III decreed at the Latran Council of 1179 that all cathedrals should have a master who taught without requiring students to pay. The move was good for traveling scholars, but perhaps not so good for financially-minded collegiate cathedrals.
This change came at the same time as a technical reduction of the cathedrals' control over teaching practices. The established practice was that any clerk who wanted to teach externally first had to be granted and pay for a licencia docendi from the local chapter. But in 1170 Alexander III saw this measure as an unnecessary restriction on learning. He instructed all French bishops to grant the licencia free to anyone who wanted it and was able to teach (although he didn't say who was to determine the ability to teach). The measure became general law following the Council of 1179. Whether a cathedral in a frontier region like Toledo would go along with this change is a matter of some doubt. Yet possible tension over the granting of a licencia could provide an interesting explanation for Girardus's apparent change of status from a cathedral canon to a 'dictus magister.' The 'dictus' could of course be attributed to Girardus's proverbial modesty. But one might also see him as a foreign clerk who--at least to pay for the time spent translating--wanted to set up or work in an external school or college, perhaps largely for foreign students and scholars, but could not do so officially. His group translation activities would thus be described as a school in all but name, even by a visiting Englishman who had every interest in naming a school.
The situation at Toledo would also be complicated by the cathedral's virtual monopoly over administrative writing. The archbishops were--or were to become--nominal heads of the royal chancery (Renan 1852: 20; Proctor 1934), a position that involved considerable political power. Since the church monopoly would certainly have been challenged by an influx of advanced students and foreign clerks working outside the chapter's structure, protectionist measures might have been called for.
The potentially conflictual position of these foreigners must be understood within a fairly complex social mosaic. Translators working from Arabic in twelfth-century Spain fall into four cultural groups. The first is formed by the Arabic-speaking Mozarabs, Jews and Conversos who gave oral versions of the texts. The second, which includes Petrus Toletanus, Marcus Toletanus and perhaps Johannes Hispanensis, comprises Mozarab clerks employed to write in Latin. The third, represented only by Gundisalvi ('Dominicus archidiaconus') and Hugo Sanctallensis, comprises clerks of passably endemic Hispanic culture, although Gundisalvi is sometimes said to be a Converso and Menéndez Pelayo found him liberal enough to be guilty of paganism (1880 I: 31-32). The fourth group would be made up of a wide range of foreign intellectuals in terra aliena: three Englishmen, two Italians, a Carinthian, a Fleming, and the rest unknown. These foreigners would appear to have been the only ones with no fixed source of income: two of them dropped science in order to work for the abbot's gold. The traveling foreigners would thus the only translators with any financial reason to create trouble by commercializing their skills. They would have to accept whatever commissions were going, selling their services to administrators (breaking the church's monopoly) and perhaps accepting fees from paying students. On both these counts, their extramural activities would necessarily challenge the Toledo cathedral's control over written Latin. There were grounds for motivated conflict even before anyone bothered about empirical science challenging ecclesiastical authority. Even if no one in Toledo appreciated the extent of this latter long-term conflict--the archbishops did finance translations--there must have been at least a whiff of dissidence in the air.
These are all reasons why there might have been tension between the Toledo cathedral and a group of translators operating in a way similar to a school. These reasons would also explain why the documental evidence makes no explicit mention of any school of translators. The actual existence or nonexistence of any advanced school or college is virtually irrelevant here. Cases could be made for Jourdain's and Rose's respective terms, but the words would not necessarily tell us anything about how the institution affected translation practices. Nor is it particularly important whether or not there was any open conflict. What counts is that we can legitimately hypothesize the existence of a pedagogical translation practice, and that such a practice would have presented grounds for conflict. With or without any institutionalized school, the notion of tension itself provides an interesting way of explaining the strategies the translators used.
Literalism as a Misplaced Theory
The dominant but not exclusive translation procedure for scientific texts in the twelfth century was word-for-word literalism. The strategy has been described as 'slavish' (Kunitzsch 1974: 104) because it presents the translator as an inferiorized intermediary, subordinate to the source text. Yet one should not immediately suppose that all the translators adopted this position out of unthinking servility. There were quite subtle reasons for translating literally.
The first reason was the need to invent entire target-language vocabularies. Johannes Hispanensis, whose word-for-word renderings suggest greater competence in Arabic than in Latin, used Arabic verb-based generation to produce some quite exotic neologisms: Lemay (1968: 114) lists elementum as giving elementari, elementatum, elementator, elementans, and even elementabantur. Plato Tiburnensis and Girardus Cremonensis actually complained about the inferiority of Latin as a language of science, formulating a position that was significantly alien to the abbot of Cluny's mode of thought and which could have underlaid yet another divide between the church and the translators of science. Yet the translators needed solutions to some very practical problems: they frequently confronted passages they simply did not understand or could not express. The resulting errors (Opelt 1959, Lemay 1968, d'Alverny 1968, Jacquart 1989) indicate they were grappling with concepts and realia new to both them and their language.
As any practising translator will appreciate, a relatively opaque source allows for two basic strategies. On the one hand, basic literalism may be used in the hope that some later user will make more sense of it than can the translator. On the other, the passage in question may be questioned, omitted, or glossed on a higher level of generality. The weight given to textual authority means that the literalist strategy was predominant in the translation of Arabic texts, passing the task of interpretation on to the reader. This could lead to some peculiar solutions.
Although Aristotle knew there were elefanteV at both ends of the world, Girardus had no notion of the corresponding Arabic term, no conceptual connection with a Latin elephantus, and so transliterated the Arabic as rather opaque alcobati (Opelt 1959: 140). Similar examples concern apparently unknown proper names: Hesiod becomes a Syriac 'Chosmereos,' the Greek town Colophon becomes a transliterated 'Calfuniati,' and so on. A basic strategy was perhaps 'when in doubt, transliterate,' and this could be useful whether or not one believed in the inferiority of Latin. As much as the procedure appears inadequate to us today, some kind of transmission was achieved and later users did indeed clear up many of the difficulties, either through retranslation or through practical application. An eloquent example is the planta noctis (night plant) that Sudhoff found translated by Girardus and Alpagus as a cause of skin infections (1909: 352). The origins of the plant are to be found in the Arabic nabat, a defective transcription of the term banat, meaning 'daughter.' As was eventually stated in a sixteenth-century version--at a time when physicians were seeking information on the 'new' disease of syphilis--the real infections came from 'daughters of the night.' Jacquart (1989) gives a similar example concerning testicles, if anyone is interested. The important point, though, is that literalism was a partly successful way of transmitting knowledge that the translators did not understand. The logic sounds perverse, but it sometimes worked.
There was also some theoretical background for literalism at Toledo. Jerome, of course, had argued that translation should be sense for sense except in the case of the scriptures, which should rendered word for word 'since there is mystery in the very order.' Closer to our twelfth-century translators, Boethius extended literalist strategies ('verbum verbo expressum comparatumque') into the realm of authoritative secular texts, where the translator's goal was 'non luculentae orationis lepos sed incorrupta veritas' (Isagogen Porphyrii, cited in Haskins 1924: 223). Johannes Scotus adopted essentially the same strategy, arguing that he should be considered 'the translator of [the] work, not its expositor' (cit. Copeland 1991: 52). A little later than our Toledan activities, the Pisian translator Burgundio also declared his strategy to be 'de verbo ad verbum' because he had no right to add his own opinions to authoritative texts (Haskins 1924: 151). Literalism was a strategy that had been taken over from the church and should at least have looked respectable in the eyes of the church. After all, it socially restricted knowledge of science in the same way as it had restrained critical knowledge of sacred texts. It also corresponded to real respect for non-religious authors. As Haskins comments (1924: 151-152), 'the texts which these scholars rendered were authorities in a sense that the modern world has lost, and their words were not to be trifled with.'
Yet there were also several reasons why this strategy should perhaps not have been applied at Toledo.
The principles of medieval literalism had been formulated on the basis of reasonably cognate language pairs, notably Greek and Latin, where word-for-word renderings could at least fall back on the accrued results of previous translations. This principle could not be applied so blindly when working from Arabic, where sentence structure and morphological generation patterns were markedly different from those of Latin. The oral use of an intermediary Romance language would also complicate the principle, since it is difficult to translate 'word for word' when a Romance word stands between Arabic and Latin. And literalism would make even less sense when the Arabic word is itself a translation from Syriac, which was in turn a translation from Greek. Further, there is evidence that many translations, including Girardus's Almagest, were made from several Arabic manuscripts. When all of this is coupled with the scriptio defectiva of the early Arabic manuscripts, with neither vowels nor diacritics, significant transformation was bound to result. The extended series of variants would necessarily make 'word for word' principles difficult to apply.
In view of these factors there should at least be some suspicion that the translators in Spain were using a principle that did not really suit their circumstances. If they still applied literalism, they must have had a certain interest in doing so, just as Daniel de Merlai had a certain interest in describing literally what he saw, passing the task of interpretation on to later readers.
This hypothesis is partly confirmed by cases where literalism was either not applied or applied in a self-interested way.
Some of the translators who were working directly for French patrons (Hugo Sanctallensis and of course Robertus Ketenensis when doing the Qur'an) were not particularly willing to accept a position of linguistic inferiority with respect to Arabic. Their Latin could be quite refined, with relatively little regard for their Islamic sources. Similarly, Hermannus Dalmata, who had been trained at Chartes, translated in a clear and elegant Latin that made few concessions to literalism. Indeed, he is reported as being frustrated at the prolixity of his Arabic authors and actually reduced Abu Ma'shar by about a third (Lemay 1968: 113). One could say these translators allowed secondary strategies--glosses and omissions--to win out over primary literalism. Yet even here we find considerable ambiguity. Remember that Robertus Ketenensis, having producing an extremely free version of the Qur'an, claimed not to have tried to 'cover so vile and unworthy a matter in gold.' This statement ostensibly recuperates the politics of literalism, according Latin no priority over Arabic and, more importantly, dissociating the translator's own Latinist subjectivity from the content of the foreign text. Nonliteralism could hide behind a literalist mask.
A variant on this strategy was to combine literalist techniques with secondary concessions to political correctness. For example, in Johannes Hispanensis's Abu Ma'shar we find omission of the Virgin Mary described as a 'coquette' (ganija), omission of the sun as the giver of life, and a general attenuation of sexual references and astral fatalism (Lemay 1968: 118-121). In Girardus's Aristotle there are references to the perfection of the Trinity, to hell, and to God as creator, where the Greek text certainly did not make such references (Opelt 1959: 145). Diplomatic christianization--possibly also due to earlier mediation through Syriac--meant that basic literalism could be modified in the interests of a higher authority.
Of course, literalism could also be made to work against higher authorities, particularly to the extent that it led to recognizably translational language. There is some evidence that it provided a convenient way of presenting dissident or non-authorized opinions. Thorndike argues that many of the texts presented as translations may well have been pseudotranslations: 'The number is suspiciously large of works of which the lost originals were supposedly by Greek or Arabian authors but which are extant only in later Latin "translations"' (1923: II 26-27). Adelardus de Bada claimed to have disguised many of his personal opinions in a way quite compatible with pseudotranslation: 'For I am aware what misfortunes pursue the professors of truth among the common crowd. Therefore it is the cause of the Arabs that I plead, not my own' (cit. Thorndike 1923: II 25). Such strategies would help explain the maintenance of literalism as a general measure of self-protection, if not of subtle subversion.
In view of these strategies, it would be wrong to associate literalism with a simple notion of ethical fidelity. Nor should literalism be assumed to exclude domesticating procedures, as if there were only two basic ways of translating. On the contrary, because literalism was generally the strategy deemed appropriate for authoritative texts, the relative opacity of the resulting translations frequently required recourse to secondary domesticating techniques. In twelfth-century Spain we thus find practices of omission, questioning, and generalization that are quite compatible with the primacy of literalism.
Examples of such techniques would be the accompanying notes in Girardus's version of the Verba filiourum (Clagett 1964: I 231-233). The material falls into three categories: translation variants or double readings, variants found in different Arabic manuscripts, and additions designed to complete or explain the text. When in doubt, Gerardus was prepared to apply literalism twice (giving variants) or to provide marginal glosses. The effect would have been to question the authoritative status of the actual translation (the reader is given two or more terms from which to choose) and to generalize on a level that at least the translator understood. Jacquart (1989: 111-12) finds similar examples in Girardus's Liber ad Almansorum, where the glosses are written in the margins or in a separate column, although they are incorporated into the text in some manuscripts after the fourteenth century. Added to this must be the many in-text glosses and expansions already present in the Arabic texts that had come from Greek.
Relatively extreme literalism thus leads to a divided discourse, marking out a thin line between literalist authority and the voice of variants and glosses, which could significantly expand the source texts. Indeed, literalism could be seen a necessary precondition for this division of discursive labour. The result was a double discourse: one voice was the source as authority; the other was the translator as guide.
This division in turn allowed for a tendency towards scholia and commentaries as such, some of which can also be understood the result of retranslation combined with radical omission. Clagett provides the prime example in his 1953 analysis Adelardus de Bada's work on Euclid's Elements (from the first of our periods). The first Latin version of the text indicates translation directly from an Arabic manuscript; the second includes didactic commentaries and omits much of the proofs; the third puts the proofs back in. The fact that a translator could work on the same text in three different ways once again shows the limits of literalism. If the literalist version was opaque, a later version could adopt an explanatory discourse to clarify the first, without negating its authority. We thus have evidence of a translator who was also prepared to explain.
Or were these translators also prepared to teach? If the literalist discourse marks out a thin line between translation and pedagogical elaboration, its maintenance could be crucial to any school that was not supposed to be a school. It allowed one to teach without being responsible for the actual knowledge transmitted. Such strategies are common enough among ideological infiltrators. But they are likely to be overlooked by translation theories that see literalism as the simple opposite of freedom.
Systematic work on manuscript traditions has done much to show how a primary literalism connected with later reworkings and retranslations. The translational series linking Adelardus de Bada's versions of Euclid can be continued to Gerardus Cremonensis's version of the same text, apparently done from a revised Arabic translation or possibly from two Arabic manuscripts. The dominant translations were nevertheless Adelardus's, which spawned a later series of reworkings and commentaries in Latin (Murdoch 1968). In a comparable piece of research, Lemay (1968) has studied Latin translations of Abu Macshar's Introductorium in Astronomium, working from some forty Latin manuscripts and five manuscripts of the Arabic version. He identifies a 1133 translation by Johannes Hispanensis (here 'Jean de Séville'), the 1140 adaptive summarizing version we have mentioned by Hermannus Dalmata ('de Carinthie'), and a significant 1171 revision of Johannes Hispanensis's version, attributed to Gerardus Cremonensis. Although the study of manuscript traditions brings us closer to individual moments of literalist translation--we can more or less reconstruct the source and target texts involved--the real message is that such moments could not have been regarded as definitive. Since the literalist version was bound to be relatively opaque, it presupposed embedding in a series of future retranslations and pedagogic elaborations. The problems involving secondary strategies could be put in margins, left for later translators, or even left for the more specialized interpreters at the colleges to the north. This meant that the kind of literalism employed by individual translators did not imply a definitive translation. It instead required reworkings.
A peculiar example of the relation between literalism and commentary is to be found in the work of Hermannus Alemannus, carried out in the mid-thirteenth century but perhaps not entirely out of place here. Hermannus describes how he set out to translate Aristotle's Poetics and Rhetoric but found the task too difficult 'propter disconventientiam modi metricandi in græco cum modo metricandi in arabico, et propter vocabulorum obscuritatem et plures alias causas' (Jourdain 1819: 141). Since literalist precepts contradicted the different language structures and prohibited omission, Hermannus sought a secondary strategy. He translated Averroes' abridgement of the Poetics and Alfarabi's glosses on the Rhetoric, which would then ideally lead to an understanding of Aristotle himself. In this case the pedagogical strategies precede the primary strategy of literalism. But this was only because the primary strategy was there first, as an ideal to which Hermannus felt inadequate.
The relative absence of any ideal of definitive translations becomes apparent in the way manuscript traditions were used. Although Björnbo believed that the first or best translations held sway over the rest, Haskins has shown that this is not so. Gerardus Cremonensis's 1175 version of Ptolemy's Almagest, although translated into Latin from Arabic, enjoyed far greater medieval circulation than an earlier Latin version translated directly from the Greek. The general result was that later scholars had recourse to a plurality of translations and commentaries, giving rise to the category of alia translatio. Lemay (1968: 104) similarly finds evidence that Hermannus's non-literalist version of Abu Macshar was later used to correct and orient a reading of Johannes's literalist version; Murdoch (1968: 67) sees alie translationes of Euclid as 'more the rule than the exception'; and Haskins (1924: 108) describes a later Latin manuscript of Ptolemy's Almagest compiled from at least three earlier Latin translations, two of which were from Arabic.
We thus find that literalism could produce not just recourse to pedagogical strategies but also a historical plurality of translations, none of which need become definitive to the point of replacing the rest. The later user of translations could bring together the products of both literalist and secondary strategies, actively making them comment upon and correct each other.
We can take this model further. Would not much the same result be achieved when the actual translators came together to work in teams? Bear in mind that the team working on the Qur'anic documents pooled a range of complementary competences: Latinist scholars of science, a Mozarab reader of Arabic, an Islamic 'native informant,' and a watchdog Latin stylist. Other teams were similarly based on complementary expertise: the Jew Savasorda and the Italian Plato Tiburnensis, the more mysterious Jew Avendauth alongside the Christian Gundisalvi, the Mozarab Galippus with the Italian Girardus. While working together, the translators would orally exchange and combine different translations to produce one final product. If you like, they would deploy alie translationes before rather than after the ideal of a complete translation job. This means that the corrective comparison of written alie translationes would become, in the space of the translation team, something like the oral confrontations of an incipient disputation. The actual practice of group translation, oral then written, could well have looked like a process of debate, intellectual discussion, pedagogical transmission, or even a school.
These various connections between translation and teaching suggest that it would be very wrong to seek any 'school' in the existence of a shared doctrine or style of translation. When the translators worked together, even across extreme power differentials, they could presumably settle on some kind of presentable product. But when they worked apart, as also happened, they would naturally produce very different translations, since they had very different backgrounds and competences. Thus, even extreme diversity in translation methods by no means rules out properly pedagogical interpretations of the term 'school.' Thanks to their differences-not despite them-the translators had every interest in coming together within some kind of institutionalized structure.
Strategies for Translating within the Church
Most of the translators working on scientific texts during our first period formed a small network among themselves. There is little evidence of any attachment to local church structures. Robertus Ketenensis and Hermannus Dalmata were friends and in contact with Plato Tiburnensis, Rudolfus Brugensis and Johannes Hispanensis. Yet they were geographically dispersed and had few visible means of support. This first-period network cannot properly be associated with Toledo. Nevertheless, as we have seen, a certain controlled institutionalization can be dated roughly from the visit of Petrus Venerabilis in 1142. The abbot's gold gave the translators a new status.
Remember that Petrus met archbishop Raimundus of Toledo when he was passing through Salamanca. This meeting quite possibly inspired the archbishop to sponsor a translation himself--the Costa ben Luca version by Johannes Hispanensis--since his personal interests until then had apparently had more to do with money and power (González Palencia 1937). The stage was set for a long collaboration between church and translators, for a school if one likes, or at least a period of institutionalized cooperation.
If we accept that different translation strategies can coexist within a schooling institution, no matter what the specific nature of the institution, we should not be surprised to find the strategists forming some substantial common ground. But how would this have been possible in the Toledan situation?
As we saw in the previous chapter, the abbot's strategy was to use translation as a weapon, an idea that was quite possibly handed on to the archbishops of Toledo. The translators of science were also interested in intellectual conquest, but of a different kind. This can be appreciated from our friend Daniel de Merlai. Here he is giving further justification for his visit to Toledo, trying to cover over quite a few gaps:
Let no one be shocked if, with reference to the creation of the world, I should invoke the testimony of pagan philosophers rather than the church fathers. Although not numbered among the faithful, some of the gentiles' words are full of faith and should be incorporated into our teaching. Sine our freedom from the servitude of sin has been symbolized in the Israelites' liberation from their Egyptian masters, we too have been ordered by the Lord to take the Egyptians' gold and silver vessels to enrich the Hebrews. Let us then borrow from them and, with God's help and command, rob the pagan philosophers of their wisdom and eloquence. Let us take from the unfaithful so as to enrich ourselves faithfully with the spoils.
Theft, conquest and military metaphors were nothing new in translation theory. What is striking, though, is that Daniel justifies the transfer of knowledge from Arabic by citing precisely the authoritative source most likely to be upset by Islamic science. Since the Bible itself could legitimate appropriation, it could also justify incorporation of a non-Biblical world-view. And the Spanish Crusades could also be a conquest of knowledge. One wonders how far Daniel's bishop would have gone along with the argument. Rather than resolve a problem, Daniel at least indicates the problem that had to be resolved. The abbot of Cluny might have wanted to use translated science against Islam, but the translators wanted to use it as a source of wealth in itself. What the abbot saw as the creation of a weapon for future use, the translators saw as direct appropriation.
These two strategies should probably involve quite different principles, especially since the translation of science was eventually to undermine the authority of the church. And yet what we find in Spain are various modes of compatibility between conflicting interests.
For the sake of prestige, bishop Michael was no doubt pleased to see his name on a translation of service to astronomers. For the sake of storing weapons, archbishop Raimundus and his followers no doubt thought they were simply applying the abbot's strategy. And for the sake of money, the translators were no doubt prepared to accommodate these interests, diplomatically christianizing their translations and finding whatever biblical analogies were convenient. For all these reasons, science could pass from Arabic to Latin, presenting gifts of conquest to the authorities it would later contest.
There was also a properly historical dimension to this process. Foreign translators were in Hispania before the church took an official interest in translations. Period one came before period two. The church sought to use the translators to its own ends, partly institutionalizing them at Toledo where they could at least be seen and superficially controlled. Period two thus provided the intercultural space for stage three, the translative activity around Girardus Cremonensis. But in bringing together voices that knew and were interested in Islamic science, the church set up a mode of group translation that was also a pedagogical activity based on oral discussion and debate. Tension resulted. And the lessons written down in this extended translational practice would eventually extend beyond ecclesiastical control.
Six Principles for Translators without a Theory
This three-stage process conforms to the now classical model of a translation culture starting on the periphery of an expanding target-culture system and working its way towards the centre. The model applies well in this case because Islamic learning, which was in decline in the twelfth century, played virtually no active role in the reception process. The Jewish and Mozarab intermediaries would also seem to have left the target system untroubled once they collected their fees. But the classical model cannot explain the causality of this process. It cannot describe the complex determinism of different social groups using strategic action to achieve different but superficially compatible ends. These latter aspects can be approached through the model of negotiation that we have already applied to Petrus Venerabilis and his 'Arabs.' Although the previous negotiation was spectacularly unsuccessful, it could have set the stage for a more effective exchange.
Given the passive role of the source culture and the Arabic-speaking intermediaries, the general translation situation should this time be reduced to a bilateral negotiation between Latinists. On one side we have the foreign clerks who accepted the authority of their source texts and had some idea about what they were translating. On the other stand the interests of the church, which was concerned with extending its own authority, represented by an abbot, archbishops, bishops, and perhaps one archdeacon. The exchange we are interested in took place between these two parties.
Once again, the abstract negotiation process requires principles that are to some degree shared by both parties. It is convenient to start from the principles that are most shared and then proceed to those that involve most tension. If the negotiation is to work, the priority of the first principles will override the tensions of the latter, establishing a more or less coherent hierarchy.
We have already analyzed most of the principles involved:
1. The translation of authoritative texts should be literal. The church believed in literalism because of the sacred status of its own authoritative texts. The translators also found literalism convenient because it meant they could not be held directly responsible for what was said in the texts.
2. Secondary elaboration may be used. The abbot of Cluny added a didactic secondary discourse to the translation of the Qur'an. The translators on the side of science were also using secondary strategies, mostly to compensate for the effects of literalism. In both cases the result was a divided discourse legitimizing marked additions to source texts.
3. Translators should work in teams. The church was interested in teams because it could put the styli in the hands of its own Latinists. The scientific translators also benefited from teams, first as a distribution of linguistic competence and second as a place for intellectual debate.
4. Oral intermediaries may be inferiorized. The church had little interest in mentioning the role of Jewish and Mozarab intermediaries. But the Latinist translators were not particularly interested in showing subservience to oral intermediaries either. The Jews and Mozarabs could mostly collect their money and disappear from written history.
5. Translation was legitimate conquest. For the abbot of Cluny, translations from Arabic could forge weapons. For the translators, they were a conquest in the sense of direct appropriation. There was thus room for agreement, but also room for potential disagreement about what some translators saw as the inferior status of the target language and culture. The fact that this contradiction did not surface as a conflict-as it later would-owes much to the way common accord on the principle of literalism blocked any developed debate about the use of eloquent Latin in translations.
6. Non-Christian texts could be authoritative. The church could appreciate Islamic science as at least a counter-authority to Islam. The translators, on the other hand, generally recognized the texts as both authoritative and directly useful. They made them superficially compatible with the authority of the church, and the church went along with a process that it could hardly have stopped anyway.
As formulated, these principles are no more than hypotheses tentatively applicable to the process formed by all three of our stages. But they do constitute a fairly coherent whole that would have enabled conflicting interests to be reconciled. The translations of science could become a Trojan horse, proposed and accepted.
Some would say the principles look like a translation theory. And yet twelfth-century Spain offers no formal theory of translation. Our principles might help explain this absence.
Some theoretical words could have been cited at Toledo: Boethius against oratory, Eriugena against auctorial responsibility, and perhaps John of Salisbury in favour of additions. But more developed theories like Leonardo Bruni's or Etienne Dolet's could not have been formulated there. When Dolet argued that the translator should completely understand the matter to be translated, that the source language should be mastered, and that liberty was preferable to word-for-word servitude (1540: 14-16), he would appear to be stating the obvious until we compare his position with the principles of twelfth-century collaboration. His theory involved an individualism that is too easily presupposed today. Why should we seek the secrets of translation in neurophysiology rather than the interculturality of past teams? More fatally, Dolet's liberty implicitly made the individual translator responsible for what was translated. Although no doubt a theorist in tune with the practice of his time (Norton 1984: 31), Dolet was not a great negotiator. The church held him personally responsible for his translations, condemning him to the stake just four years after he wrote down his theory.
Trojan horses can subvert from within. But a gift too loudly proclaimed can also be burned.