Interpreting the language-mixing in terms of codeswitching" The ...

8 downloads 74448 Views 586KB Size Report
to be found in an Arabic-Old French phrase-book written by Copts in the 13th ... The survey of the material contained in the phrase-book reveals that besides ...
jonlMh~ ELSEVIER

Journal of Pragmatics 32 (2000) 1273-1281 www.elsevier.nldocate/pragma

Interpreting the language-mixing in terms of codeswitching" The case of the Franco-Italian interface in the Middle Ages Cyril Aslanov Department of French Language and Literature, Faculty of Humanities, Institute of Arts and Letters, The Hebrew University of Jerusalem, Mount Scopus 91905 Jerusalem, Israel

Abstract This study concerns a strange coexistence of French and Italian within the same sentence to be found in an Arabic-Old French phrase-book written by Copts in the 13th century and in a manuscript of the Song of Roland copied in Venice at the beginning of the 14th century. The survey of the material contained in the phrase-book reveals that besides codeswitching which is motivated by extralinguistic factors, there are cases of codeswitching caused by the desire to avoid a French form and lo give preference to its Italian counterpart. The FrancoVenetian manuscript which is of a literary nature also shows many cases of codeswitching motivated by intralinguistic considerations (in this instance the desire to respect the prosody of the verse). These testimonies of medieval codeswitching from a bilingual Franco-Italian background show each in its own way how codeswitching can be reconstructed from texts written in the remote past. © 2000 Elsevier Science B.V. All rights reserved.

Keywords: Code switching; Coptic

I. Introduction Studies of codeswitching, a subject that has attracted the attention of many specialists of language contact have mainly been carried out on oral material. Therefore, it may sound highly paradoxical to try to study evidence of codeswitching in documents written seven hundred years ago. One of these is an Arabic-French phrasebook whose Old French was spelled in Coptic letters (Maspero, 1888). This Coptic writing is not merely a matter of interest for amateurs of curiosities. Unlike literary texts or various records written in Old French in Latin letters, this phrase-book may 0378-2166/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0378-2166(99)00100-9

1274

c. Aslanov /Journal of Pragmatics 32 (2000) 1273-1281

not be suspected of being influenced by any kind of French literacy. Beyond the garb of the Coptic letters, a genuine echo of the French spoken in the Holy Land at the end of the Crusaders' dominion in this country (Acre, the last stronghold of the Crusaders fell in 1291) can be found. The case of codeswitching I want to discuss concerns a strange intrasentential mixture of French and Italian. The first part of the study aims at showing that this mixture is a genuine case of codeswitching at the level of spoken language and not simply a case of confusion in the writing of the phrase-book due to the ignorance of the Copts. Once this point is confirmed, I would like to reconsider some striking features of the Old French represented here and to view them as additional cases of codeswitching between Italian and French. By themselves the items examined in this second part could be interpreted as simple cases of language-mixing. However, in the light of two unchallenged cases of codeswitching, the other cases of languagemixing in the phrase-book might be considered a reflection of a tendency to switch from one code to the other in oral speech that may be reconstructed through this curious conversation-book. Moreover, that the phrase-book is a testimony of intrasentential codeswitching is corroborated on a literary level by what are usually called Franco-Venetian texts, i.e. the texts of the Old French epic (e.g. Chanson de Roland), which were copied in Italy, mainly in Venice. I would like to compare the Arabic-French book with the Franco-Venetian cases of language mixture in order to reinterpret the latter as also showing examples of codeswitching.

2. Language mixing in the manuscript or codeswitching in the spoken language? One of the few sentences in the phrase-book, the main bulk of which consists of isolated words, is el N 1hl6n~,nl N einilijpapin, which translates the Arabic ma hu'a yaqra' mlih 'he does not read well'. The continuous string, el N i A i 6 n a n l N einilijpapin, is to be analysed as follows: cl N I ~16 n~, nl N ei ni lij pa pin. As usual in Coptic spelling, the diphthong el shall be interpreted as an/i/, representing the truncated form i < il, which still exists in colloquial French. The syllable N I ni represents the first element of the negation ne ... pas, the second element of which is n~, pa. As for the last element nl N pin, it is the notation of the adverb ben~bin, which is the unstressed form of bien 'well'. The use of rr p is a frequent feature of the Coptic spelling and it is explained by the fact that in Coptic the opposition between voiced and unvoiced consonants is not phonemic. The element h i 6 lij is the only part of the sentence that cannot be ascribed to French. It is impossible to agree with Maspero (Maspero, 1888: 507) when he reduces the whole sentence to a Modem French standard (il ne lit pas bien). Indeed, his reading consists in attributing the discrepancy between Modem French and the reading of the manuscript to the ignorance of the copyists and other transmitters rather than to assume that the language may differ from the speech of a French bourgeois during the Third Republic.

C. Aslanov / Journal of Pragmatics 32 (2000) 1273-1281

1275

The spelling, A t 6 lij, which is considered to represent a 3rd person singular of the indicative present (as shown by the Arabic original yaqra') cannot be French. Indeed, the element represented by the Coptic letter 6, which shall be interpreted as a notation for the voiced prepalatal affricate/d2/or the voiced postalveolar /2/ (as confirmed by many parallels in the phrase-book), is impossible to justify in terms of French morphophonemics (De la Chaussde, 1977: 149-151). The only Romance languages that have preserved a trace of the c o n s o n a n t / g / o f Latin legit, albeit as the affricate/d~/, are Italian (legge), Provencal (leg / lieg later replaced by legis) and Catalan (llig later replaced by llegeix). From these three languages, the most likely to be represented by ~l 6 lij is Italian despite the discrepancy between the t of h t6 and t h e / e / o f legge, although the ending {-e } of legge is not noted by the gloss. However, the omission of the f i n a l / e / h a s a parallel in another Italian gloss of the Arabic-French phrase-book: CO~,Nrr~,~ stanpag that stands for stai in pace 'be quiet'. The identification of ~ 16 lij as Italian legge is confirmed by two more glosses that must necessarily be attributed to Italian. The first is hi 6 c lije that translates the Arabic imperative iqra' and should therefore be interpreted as leggi. This time the ending {-i } is expressed by the Coptic spelling and only the vowel of the radical diverges from the Italian standard. Lastly, the gloss h e 6 e n l ~ E lejepiah contains both the ending {-i} and a correct vocalism. Thus the three forms A i r lij, ,~t6e lije and A e 6 e leje offer a shaded reflection of two homonymic forms, the 3rd person singular of the indicative present legge and the imperative 2nd person singular legge, A o6 lij representing the former and h t 6 c lije and A c 6 e leje representing the latter. Whereas ~ t 6 e lije is an isolated word, the gloss ~.c6ertJa.;~ lejepiah appears as a short sentence. It is of particular interest because it combines the Italian imperative leggi with the French adverb bien, this time in the stressed form. Therefore, the gloss cl N Inl6rt~,nl N einilijpapin corresponds to ; ~ c 6 e r t l ~ ? lejepiah insofar as it combines the same lexical elements (Italian imperative leggi; French adverbs bien/ben) with a different grammatical status assigned to the verbal nucleus: indicative on the one hand, imperative on the other. In both cases, the French and the Italian phrases constitute a 'discourse unity', where French is the matrix-language and Italian is the embedded one (Myers-Scotton, 1993: 1-2). Moreover, it seems that in the case of el N thJ6na, rri N einilijpapin , the spelling ~ i 6 lij reflects a hybridisation between It. legge and Fr. lit. This could explain the dropping of the {e} ending and perhaps the switch f r o m / e / t o / i / , although this latter switch may be due to the approximation inherent in the use of Coptic letters to record a language different from Coptic. This hybridisation could be considered as a process of adaptation of the embedded language according to the rules of the matrix-language. Such unification reflected through the process of morphophonemic adaptation is an important criterion that allows the two short sentences c i N i ~ l r n ~ r r t N einilijpapin and A c r ~ r r l ~ lejepiah to be considered as codeswitching. The fact that the same verb leggere appears three times in its Italian form in the framework of an Arabic-French phrase-book testifies that there is a special correlation between the switch to Italian and the lexical item 'to read'. This is obviously a

1276

c. Aslanov / Journal of Pragmatics 32 (2000) 1273-1281

case of unmarked codeswitching or borrowing. Indeed the use of the Italian leggere instead of French lire seems to be expected and systematic, for it occurs three times in the framework of a relatively small corpus (228 items). Unlike the case of codeswitching which may have been motivated by strategic and extralinguistic implications (e.g. the sentence c o a N n ~ , ~ stanpag that sounds like a fixed formula), the cases of intrasentential codeswitching provided by i ne legge pas ben and leggi bien must have been brought about by the peculiar nature of the verb lire in Old French rather than by a desire to express something beyond the content of the message. A survey of the various reflections of Latin legere in the various Romance languages shows that due to the presence of a dorsovelar consonant/g/at the end of the radical stem leg-, this verb has been exposed to various phonetic accidents, so that it is considered irregular in the verbal systems of the various Romance languages. In French, in Castilian and in Portuguese, the dorsovelar element has been palatalised and finally dropped, this palatalisation having caused the diphthongation and then the triphthongation of the radical vowel/e/in French: Lat. legit > lieTet > lieyyet > lieyt > lit. (De la Chauss6e, 1977: 150; 1982:114-115). As mentioned above, Catalan and Provenqal have preserved the integrity of the radical stem leg- (leg / llig), but in the later evolution of these languages, these forms have been replaced by forms with a suffix which is the Romance reflection of the inchoative verbal suffix -iscere: leg-is instead of leg in Provenqal; lleg-eix instead of llig in Catalan (Badia Margarit, 1962: 319-322). As for Italian, it has conserved the integrity of the Latin radical leg- in spite of the processes of palatalisation and reduplication that have turned the Latin dorsovelar/g/ into a double prepalatal affricate in Italian/dd~/: legge de dd~e/. Lastly, it should be noticed that the difficulty involved in the conservation of the radical leg- has expressed itself in an extreme way in Rumanian, a language that uses the verb borrowed from the Slavic a citi instead of the continuation of Latin legere. And yet the preservation of the hereditary Romance form would not have caused more problems than in Italian. The replacement of the French form lit by the Italian form legge (albeit in the hybrid form lij < legge x lit, which may be reflected in the spelling hi6) and the substitution of the imperative/i (corresponding to Modem French lis) by leggi may be viewed as the manifestation of the same structural tendency that has brought Provenqal and Catalan to replace leg and llig by legis and llegeix and Rumanian to prefer a Slavic verb to express the idea of reading. In addition to the phonetic erosion that caused the reduction of legit and lege to the monosyllabic forms lit and/i in French, the homonymic clash with the verb lier may have influenced the Levantine speakers to use the Italian or the Italianized form instead of the genuine French form. Indeed the imperative li of lire and the imperative lie of lier are almost homonyms and the same is true about the indicatives lit and lie. Neither phonetic erosion nor the homonymic clash between some forms of lire and lier has been compensated in French, except in the specific context of the Holy Land, where the Crusaders and their descendants were at least bilingual, possessing

C. Aslanov / Journal of Pragmatics 32 (2000) 1273-1281

1277

a simultaneous knowledge of both French and Italian. This is sufficiently supported by the existence of the phrase-book itself, where we find strong Italian influences on the phonetics and the lexicon. The replacement of lit and li by legge/leggi to avoid phonetic erosion and the homonymic clash provides us with a case of codeswitching motivated solely by intralinguistic mobiles rather than by extralinguistic motivations. It may be described as a kind of polymorphism, whereby using Italian forms instead of their French counterparts would be more convenient due to the phonetic erosion that characterises the French words. The bilingual speaker could therefore have at his/her disposal two forms to express the action of reading, one French and one Italian, this latter being the most convenient when the message might be disturbed by a homonymic clash. By itself this constitutes a sufficient clue as to the genuine character of the codeswitching. Indeed the Coptic transmitters of this phrase-book do not seem to have had a clear knowledge of what they were copying, as attested by many errors and many cases of discrepancy between the Arabic of the original and its French rendering. On the contrary, replacement of lit and li by legge/leggi shows not only a knowledge of French, but even an awareness of the problem of homonymic clash, which is crucial in a language so deeply affected by phonetic erosion.

3. Other cases of codeswitching Once we have demonstrated that the material of the phrase-book reflects the dynamics of genuine codeswitching, let us consider other cases of code-mixing which in their turn may be interpreted as instances of codeswitching in the light of the unchallenged case of codeswitching discussed above. From a phonetic or at least from an orthographic point of view, contact with Italian may explain noting the diphthong/oe/resulting from the closed/e/in the free stressed syllable by e / e / ( e . g . Mec mes for mois /moes/probably under the influence of Italian mese) or the notation of/s/before a consonant (e.g. Be u e c'r p e fenestre forfenetre probably under the influence of Italianfinestra). Due to the phonetic character of Coptic spelling, this interference cannot be considered as reflecting standard Latin spelling in French. Therefore, mes and fenestre may be viewed as compromises between French mois and Italian mese and between Frenchfenetre (by the 13th century fenestre had already evolved to fenetre) and Italian finestra. Such a compromise may be explained once again in terms of codeswitching. At least in the case of Me c mes, the preference showed to the Italianized form could be considered an attempt at avoiding a homonymic clash between m o i s / m o e s / > / m o e / ' m o n t h ' and moi /moe/ ' me'. Lastly, let us remark that like in ~16 lij = legge and CO2tNn2t~ stanpag = stai in pace/ a compromise between the matrix-language and the embedded language seems to be the reason for dropping the Italian ending {-e }. It appears that there is a certain degree of coherence, which confirms that this is a genuine case of codeswitching from French to Italian in spoken language rather than an erroneous confusion between the two languages in the writing or copying of the

1278

C. Aslanov / Journal of Pragmatics 32 (2000) 1273-1281

phrase-book. Although the alternation between A I r l i j , A l O e lije and A e O e leje discloses a gradation in the adaptation of Italian to French, that some other Italian words appear in a solely Italian form hints at the possibility that the two languages have preserved their respective identities. Another kind of Italian influence may explain the form HaNK2kNI(.D~ mankanirh, a reinterpretation of the French word mangoniaus with a switch f r o m / o / t o /a/(mangoniaus > manganiaus) under the pressure of the Italian word manganello. The process of hybridisation is even more striking if we consider that the ending -lu~E reflects a typically French phonetic process: mangonels > mangonews > mangoneaws > mangoniaws (a Picard evolution) > mangoniows > mangonio(s). Like in the case of AI6 lij that appears as a crossing between Fr. lit and It. legge, M,~.N K2kN l(Jl)~ mankaniOh combines the Italian radical mangan- (instead of French mangon-) with the allophonic treatment o f / e l / + / s / , which is typically French. Other examples of codeswitching between French as the matrix-language and Italian as the embedded language are 6eI, Aa gerla (a kind of basket) and c a g a 6 6 1 N = saraceno/saracino, this last being superficially adapted to French morphology by the truncation of the ending -o. However, the reduplication of the 6 leaves little doubt about the Italian character of this gloss because the Italian phonemes/ts/,/dz/, /6/,/d~/,/~/,/1'/,/fi/are always duplicated between two vowels. Had this gloss represented a French or Provenqal word, we would have found a double/r/: sarrazin. The very same form appears in a Franco-Venetian manuscript of the Chanson de Roland, whereas the Oxford version (written in an Anglonorman blend of French) shows the French form Sarrazin, the manuscript V4 (ms. 225 of the Biblioteca Marciana) has the form Saracins, which is a combination of the Italian radical Saracin/sararin/with the ending {-s}. The notation of the Italian unvoiced palatal affricate/6/by the same letter that serves to represent its voiced counterpart/d~/(cf. 6 e p A a gerla) is a normal feature of the spelling convention of the phrase-book. As already mentioned above, this confusion between the voiced and unvoiced consonants is a reflection of the Coptic phonemic system in Coptic writing, even if the language written in it is no longer Coptic but French (or Italian). In the case of gerla, the switch to Italian may be conditioned by the fact that this item corresponds to a realia. The genuine French word for the tool designated by gerla would have been hotte 'pannier'. However, as it happens in the overseas varieties of a given language, the vocabulary of the metropolis tends to be partly replaced by borrowings from languages in contact, especially for objects from everyday life. In the cases discussed above, we have found at least three motivations for codeswitching from French to Italian: the desire to avoid a homonymic clash caused by phonetic erosion (lit > legge; mois > mes / mese), the lack of a precise word in the lexical stock of genuinely French origin and the use of a standard formula in the case of CO2tNI-I2t(JLQ stanpag = stai in pace. t Let us return to the first of these three instances, that of the codeswitching motivated by intralinguistic reasons as it appears through the literary evidence of the Franco-Venetian manuscripts of the Chanson de Roland.

C. Aslanov / Jourr,,al of Pragmatics 32 (2000) 1273-1281

1279

3. Codeswitching motivated by prosodic considerations In the above mentioned manuscripts of the Chanson de Roland, we find striking parallels of the codeswitching documented in the Arabic-French phrase-book. Like there, Italian is embedded in a French matrix. Although Italian sometimes appears as more or less adapted to the matrix, the two languages preserve their respective identities. This literary case of codeswitching is important to help us to understand French-Italian codeswitching in the phrase-book. Moreover it confirms the intralinguistic motivations of codeswitching. Indeed, many Italian words embedded in a French matrix present the advantage of containing one syllable more or one syllable less than their French counterparts. Medieval copyists often adapted the text they copied according to their own linguistic background. Such is the case of the Oxford manuscript (Bodleian Library, Digby 23) of the Chanson de Roland, which is written in the Anglonorman dialect, although the epic itself had probably been composed in another area of the French-speaking world. By the same token, the Franco-Venetian manuscripts of Roland's Song reflect the Italian environment of the Venetian copyists to whom we owe them. I would like to analyse several examples taken from ms. 225 of the Biblioteca Marciana in Venice in order to show that in the rewriting process involved in the act of copying, codeswitching from French to Italian is a part of the prosodic reinterpretation of the verses. It should be remembered that in this Old French epic, each verse ought to contain 10 syllables (sometimes 11, if we consider that before the caesura a supplementary unstressed syllable was allowed). This compulsory syllable count functioned as an important motivation in choosing an Italian form instead of its French counterpart. Of course, other motivations may have been at work, like the misreading of the original French word and its replacement by the Italian equivalent. However, there is an obvious connection between unintentional misreading on the one hand, and voluntary replacement of a French form by an Italian form which is more suitable from a prosodic point of view on the other. Cases of misreading may have caused the collapse of the prosodic framework which was then repaired by the choice of Italian words embedded in the French matrix. An interesting example of a prosody-motivated codeswitching is the alternation between forms with a prosthetic/e/and those without it. Forms with the prosthetic vowel are genuinely French, whereas those which are deprived of it may be viewed as phonetic adaptations of the French according to the Italian standard. Let us compare, for example, verse 2319 of the Franco-Venetian manuscript: (1) Perdh avomes//Espagne la vaillant with verse 2038 of the same: (2) Laseit sa reine,//vers en Spagna se'n fu. Verse 2319 is short, containing only five words, whereas 2038 is saturated with no less than nine words. The aphaeresis of the adverbial pronoun en in P'n and the use

1280

c. Aslanov / Journal of Pragmatics 32 (2000) 1273-1281

of the Italian form Spagna instead of the French Espagne are examples of the same struggle to reduce syllable count to the required minimum of 11 (4 + 1 / 6) syllables. The use of Spagna instead of Espagne is all the more striking in that according to Italian phonetic rules, words beginning with the nexus SP must be preceded by a prosthetic v o w e l / i / w h e n e v e r the word that comes before it ends with a consonant. Thus, in Spagna / en Spagna should actually have been pronounced in ispagna. However it is not, for if it had been the case, the pronunciation ispagna would have added a superfluous syllable to the decasyllabic verse. The use of the short Italian form Spagna without the p r o s t h e t i c / e / o f the French Espagne and without the prost h e t i c / i / o f the Italian allophone ispagna shows that this switch is obviously motivated by the tendency to diminish the number of syllables in the verse, albeit in contradiction with Italian sandhi rules. Whereas the alternation between Espagne and Spagna shows a clear boundary between the French matrix and the embedded Italian, other cases of codeswitching reflect an adaptation of the Italian according to the French morphophonemic standards. Such is the case of speron in verse 2037: (3) Li civals b r o q e / / d e ses speron agfa. Here the use of speron 'spur' instead of esperon unveils the same kind of crossing as in the case of lit x legge > lij; mois x mese > mes; mangoniaus x manganello > manganiaus. Both the French esperon and the Italian sperone would have given a trisyllabic word that might have added a supplementary syllable to the required 10 (11). By using the form speron, which combines the Italian lack of the prosthetic vowel with the French lack of a final {-e}, the copyist/rewriter to whom we owe the Franco-Venetian manuscript has found a compromise allowing him to reduce the three syllables to only two. The two cases of codeswitching studied above are part of a more general historical situation that brought French and Italian in contact during the Middle Ages. Let us mention among other aspects of this contact the presence of Norman knights and rulers in Sicily, of Frenchmen in the Angevine-dominated Naples or more generally the fact that French literature, especially the epic genre, was in great vogue in Italy. Thus the Coptic phrase-book and the Franco-Venetian manuscript may be viewed as two complementary pieces of evidence of French-Italian bilingualism during the Middle Ages: the Arabic-French glossary confirms that Italian-French codeswitching really existed in the spoken language, whereas the Franco-Venetian manuscript, whose Latin writing is more transparent, helps us to distinguish between the embedding of an Italian word in the French matrix and the adaptation of an Italian word according to the morphophonemic rules of French (what we have called hybridisation). From a theoretical point of view, two dimensions may be of general interest to specialists of codeswitching: the situation of polymorphism resulting from FrenchItalian bilingualism, on the one hand, and intralinguistic motivation for switching from French to Italian, on the other. In the case of two cognate languages like Italian and French, the faculty of switching from one language to another permitted the

C. Aslanov / Journal of Pragmatics 32 (2000) 1273-1281

1281

bilingual speaker to choose the most suitable of two etymologically related forms. In the Coptic phrase-book, the preference accorded to leggi/legge instead of li or lit was motivated by the desire to avoid a disturbing homonymy with the verb lier. In the case of the Franco-Venetian manuscript the mobile was based on prosody. However, despite this difference between the two cases, their common denominator is that codeswitching is sometimes motivated by intralinguistic rather than by sociolinguistic reasons. It may enhance the intelligibility of the message, as in the phrase-book, or allow a better fit into a given prosodic pattern, as in the Franco-Venetian manuscripts. It would be interesting to look into whether this specific kind of codeswitching is limited to contact between two cognate languages.

References Badia Margarit, Antonio M., 1962. Gramatica Catalana. Madrid: Gredos. De la Chauss6e, Franqois, 1977. Initiation h la morphologie historique de l'ancien franqais. Paris: Klincksieck. De la Chauss6e, Franqois, 1982. Initiation ~t la phon6tique historique de l'ancien franqais. Paris: Klincksieck. Maspero, Gaston, 1888. Le vocabulaire franqais d'un Copte du XIII~me si~cle. Romania 17: 481-512. Myers-Scotton, Carol, 1993. Social motivations for codeswitching. Oxford: Clarendon Press. Segre, Cesare, ed., 1971. La Chanson de Roland. Milano-Napoli: Riccardo Ricciardi. Cyril Aslanov (born 1964) wrote his thesis on the epithets in the Septuagint. He teaches French and Proven~:al linguistics at the Hebrew University of Jerusalem. Recently his research has been on the linguistic background of the crusaders with special reference to the problematic of language contacts. His other publications include articles on Jewish mediaeval glosses, on the art of translation and on the language of the Septuagint. His book-length study Pour comprendre la Bible: La lefon d'Andr~ Chouraqui has been published in 1999 by I~ditions du Rocher.

Suggest Documents